diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..c22958f969d9099843503e9a71a4e4d765b0b44d --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e26ef621a6f96c9ccd801372a6cd50cc75821cbae86df7af3e1f2606f12c7e65 +size 8066685992 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..46fd354455b820442f291ee0398a379ca314d044 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2b0001ad209d57997615dc6d599f56431095d8de041fef41e957783713a6dace +size 8066667688 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..483c8031ad987b762d74dface79dfb655b85286b --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ba3297ec1f26b63937fb039139826a18cacf5ec9cca2f970cd973cab620b575f +size 8066672104 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..787cf946e95e2fff60660c2009780ea884368348 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:171c4ccaf28af4d0b5bc310e5cae47ae0baa35de536fdbe595bfc1a03b8bef9d +size 8066671208 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_50000/global_step50000/mp_rank_00_model_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_50000/global_step50000/mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..a509fe8dea47a4c66b4bbf0c8388eb7dcf05d7cc --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_50000/global_step50000/mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:031244614fb75859c3799bf9a2e44171af0ad9597422d557c09b4201f56aec24 +size 5378167768 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_50000/latest b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_50000/latest new file mode 100644 index 0000000000000000000000000000000000000000..59a558c0ee32a055bd8fc16bf0d99a5950c59ac1 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_50000/latest @@ -0,0 +1 @@ +global_step50000 \ No newline at end of file diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_50000/zero_to_fp32.py b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_50000/zero_to_fp32.py new file mode 100644 index 0000000000000000000000000000000000000000..f40f1f0acbbff30e8e8d29906b56aae4e8b8a138 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_50000/zero_to_fp32.py @@ -0,0 +1,790 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: +# python zero_to_fp32.py . output_dir/ +# or +# python zero_to_fp32.py . output_dir/ --safe_serialization + +import argparse +import torch +import glob +import math +import os +import re +import gc +import json +import numpy as np +from tqdm import tqdm +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS, AUTOEP_LAYERS_KEY, + AUTOEP_LAYERS_KEY_LEGACY, AUTOEP_ZERO3_EXPERT_STATE_FORMAT_KEY, + AUTOEP_ZERO3_PARTITIONED_EXPERT_STATE_FORMAT) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def _has_autoep_zero3_partitioned_metadata(state_dict): + autoep_layers = state_dict.get(AUTOEP_LAYERS_KEY) + if autoep_layers is None: + autoep_layers = state_dict.get(AUTOEP_LAYERS_KEY_LEGACY) + if not isinstance(autoep_layers, list): + return False + return any( + isinstance(entry, dict) + and entry.get(AUTOEP_ZERO3_EXPERT_STATE_FORMAT_KEY) == AUTOEP_ZERO3_PARTITIONED_EXPERT_STATE_FORMAT + for entry in autoep_layers) + + +def _raise_if_autoep_zero3_partitioned_state(state_dict): + if _has_autoep_zero3_partitioned_metadata(state_dict): + raise NotImplementedError("zero_to_fp32 does not support AutoEP ZeRO-3 partition-native checkpoints. " + "AutoEP expert parameters are partitioned over expert replica groups, so " + "global data-parallel consolidation would produce incomplete expert tensors. " + "Use ds_to_universal.py for expert-aware conversion.") + + +def _raise_if_autoep_zero3_partitioned_checkpoint(model_files): + for file in model_files: + state_dict = torch.load(file, map_location=device, weights_only=False) + _raise_if_autoep_zero3_partitioned_state(state_dict) + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device, weights_only=False) + _raise_if_autoep_zero3_partitioned_state(state_dict) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + total_files = len(files) + state_dicts = [] + for f in tqdm(files, desc='Loading checkpoint shards'): + state_dict = torch.load(f, map_location=device, mmap=True, weights_only=False) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if ZERO_STAGE not in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + # parse_model_states rejects AutoEP ZeRO-3 partition-native checkpoints + # before the expensive optimizer-shard load below. + model_files = get_model_state_files(ds_checkpoint_dir) + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +class GatheredTensor: + """ + A pseudo tensor that collects partitioned weights. + It is more memory efficient when there are multiple groups. + """ + + def __init__(self, flat_groups, flat_groups_offset, offset, partitioned_numel, shape): + self.flat_groups = flat_groups + self.flat_groups_offset = flat_groups_offset + self.offset = offset + self.partitioned_numel = partitioned_numel + self.shape = shape + self.dtype = self.flat_groups[0][0].dtype + + def contiguous(self): + """ + Merge partitioned weights from flat_groups into a single tensor. + """ + end_idx = self.offset + self.partitioned_numel + world_size = len(self.flat_groups) + pad_flat_param_chunks = [] + + for rank_i in range(world_size): + # for each rank, we need to collect weights from related group/groups + flat_groups_at_rank_i = self.flat_groups[rank_i] + start_group_id = None + end_group_id = None + for group_id in range(len(self.flat_groups_offset)): + if self.flat_groups_offset[group_id] <= self.offset < self.flat_groups_offset[group_id + 1]: + start_group_id = group_id + if self.flat_groups_offset[group_id] < end_idx <= self.flat_groups_offset[group_id + 1]: + end_group_id = group_id + break + # collect weights from related group/groups + for group_id in range(start_group_id, end_group_id + 1): + flat_tensor = flat_groups_at_rank_i[group_id] + start_offset = self.offset - self.flat_groups_offset[group_id] + end_offset = min(end_idx, self.flat_groups_offset[group_id + 1]) - self.flat_groups_offset[group_id] + pad_flat_param_chunks.append(flat_tensor[start_offset:end_offset]) + + # collect weights from all ranks + pad_flat_param = torch.cat(pad_flat_param_chunks, dim=0) + param = pad_flat_param[:self.shape.numel()].view(self.shape).contiguous() + return param + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = sum([flat_group.numel() for flat_group in fp32_flat_groups[0]]) * world_size + + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + flat_groups_offset = [0] + list(np.cumsum([flat_tensor.numel() for flat_tensor in fp32_flat_groups[0]])) + for name, shape in tqdm(param_shapes.items(), desc='Gathering sharded weights'): + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # memory efficient tensor + tensor = GatheredTensor(fp32_flat_groups, flat_groups_offset, offset, partitioned_numel, shape) + state_dict[name] = tensor + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def to_torch_tensor(state_dict, return_empty_tensor=False): + """ + Convert state_dict of GatheredTensor to torch tensor + """ + torch_state_dict = {} + converted_tensors = {} + for name, tensor in state_dict.items(): + tensor_id = id(tensor) + if tensor_id in converted_tensors: # shared tensors + shared_tensor = torch_state_dict[converted_tensors[tensor_id]] + torch_state_dict[name] = shared_tensor + else: + converted_tensors[tensor_id] = name + if return_empty_tensor: + torch_state_dict[name] = torch.empty(tensor.shape, dtype=tensor.dtype) + else: + torch_state_dict[name] = tensor.contiguous() + return torch_state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, + tag=None, + exclude_frozen_parameters=False, + lazy_mode=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + - ``lazy_mode``: get state_dict in lazy mode. It returns a dict of pesduo tensor instead of torch tensor, which is more memory efficient. + Convert the pesduo tensor to torch tensor by ``.contiguous()`` + + Returns: + - pytorch ``state_dict`` + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + Note: the above usage may not work if your application doesn't have sufficient free CPU memory. + You may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. Or you can load state_dict in lazy mode :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, lazy_mode=True) # not on cpu + for name, lazy_tensor in state_dict.item(): + tensor = lazy_tensor.contiguous() # to cpu + print(name, tensor) + # del tensor to release memory if it no longer in use + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + state_dict = _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + if lazy_mode: + return state_dict + else: + return to_torch_tensor(state_dict) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, + output_dir, + max_shard_size="5GB", + safe_serialization=False, + tag=None, + exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_dir``: directory to the pytorch fp32 state_dict output files + - ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB + - ``safe_serialization``: whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`). + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + # Dependency pre-check + if safe_serialization: + try: + from safetensors.torch import save_file + except ImportError: + print('If you want to use `safe_serialization`, please `pip install safetensors`') + raise + if max_shard_size is not None: + try: + from huggingface_hub import split_torch_state_dict_into_shards + except ImportError: + print('If you want to use `max_shard_size`, please `pip install huggingface_hub`') + raise + + # Convert zero checkpoint to state_dict + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, + tag, + exclude_frozen_parameters, + lazy_mode=True) + + # Shard the model if it is too big. + weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin" + if max_shard_size is not None: + filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors") + # an memory-efficient approach for sharding + empty_state_dict = to_torch_tensor(state_dict, return_empty_tensor=True) + state_dict_split = split_torch_state_dict_into_shards(empty_state_dict, + filename_pattern=filename_pattern, + max_shard_size=max_shard_size) + else: + from collections import namedtuple + StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"]) + state_dict_split = StateDictSplit(is_sharded=False, + filename_to_tensors={weights_name: list(state_dict.keys())}) + + # Save the model by shard + os.makedirs(output_dir, exist_ok=True) + filename_to_tensors = state_dict_split.filename_to_tensors.items() + for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"): + shard_state_dict = {tensor_name: state_dict[tensor_name] for tensor_name in tensors} + shard_state_dict = to_torch_tensor(shard_state_dict) + output_path = os.path.join(output_dir, shard_file) + if safe_serialization: + save_file(shard_state_dict, output_path, metadata={"format": "pt"}) + else: + torch.save(shard_state_dict, output_path) + # release the memory of current shard + for tensor_name in list(shard_state_dict.keys()): + del state_dict[tensor_name] + del shard_state_dict[tensor_name] + del shard_state_dict + gc.collect() + + # Save index if sharded + if state_dict_split.is_sharded: + index = { + "metadata": state_dict_split.metadata, + "weight_map": state_dict_split.tensor_to_filename, + } + save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json" + save_index_file = os.path.join(output_dir, save_index_file) + with open(save_index_file, "w", encoding="utf-8") as f: + content = json.dumps(index, indent=2, sort_keys=True) + "\n" + f.write(content) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info("Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info("Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument("output_dir", + type=str, + help="directory to the pytorch fp32 state_dict output files" + "(e.g. path/checkpoint-12-output/)") + parser.add_argument( + "--max_shard_size", + type=str, + default="5GB", + help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size" + "lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`" + "We default it to 5GB in order for models to be able to run easily on free-tier google colab instances" + "without CPU OOM issues.") + parser.add_argument( + "--safe_serialization", + default=False, + action='store_true', + help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_dir, + max_shard_size=args.max_shard_size, + safe_serialization=args.safe_serialization, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..c22958f969d9099843503e9a71a4e4d765b0b44d --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e26ef621a6f96c9ccd801372a6cd50cc75821cbae86df7af3e1f2606f12c7e65 +size 8066685992 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..46fd354455b820442f291ee0398a379ca314d044 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2b0001ad209d57997615dc6d599f56431095d8de041fef41e957783713a6dace +size 8066667688 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..483c8031ad987b762d74dface79dfb655b85286b --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ba3297ec1f26b63937fb039139826a18cacf5ec9cca2f970cd973cab620b575f +size 8066672104 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..787cf946e95e2fff60660c2009780ea884368348 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:171c4ccaf28af4d0b5bc310e5cae47ae0baa35de536fdbe595bfc1a03b8bef9d +size 8066671208 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_final/global_step50000/mp_rank_00_model_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_final/global_step50000/mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..a509fe8dea47a4c66b4bbf0c8388eb7dcf05d7cc --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_final/global_step50000/mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:031244614fb75859c3799bf9a2e44171af0ad9597422d557c09b4201f56aec24 +size 5378167768 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_final/latest b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_final/latest new file mode 100644 index 0000000000000000000000000000000000000000..59a558c0ee32a055bd8fc16bf0d99a5950c59ac1 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_final/latest @@ -0,0 +1 @@ +global_step50000 \ No newline at end of file diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_final/zero_to_fp32.py b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_final/zero_to_fp32.py new file mode 100644 index 0000000000000000000000000000000000000000..f40f1f0acbbff30e8e8d29906b56aae4e8b8a138 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_final/zero_to_fp32.py @@ -0,0 +1,790 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: +# python zero_to_fp32.py . output_dir/ +# or +# python zero_to_fp32.py . output_dir/ --safe_serialization + +import argparse +import torch +import glob +import math +import os +import re +import gc +import json +import numpy as np +from tqdm import tqdm +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS, AUTOEP_LAYERS_KEY, + AUTOEP_LAYERS_KEY_LEGACY, AUTOEP_ZERO3_EXPERT_STATE_FORMAT_KEY, + AUTOEP_ZERO3_PARTITIONED_EXPERT_STATE_FORMAT) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def _has_autoep_zero3_partitioned_metadata(state_dict): + autoep_layers = state_dict.get(AUTOEP_LAYERS_KEY) + if autoep_layers is None: + autoep_layers = state_dict.get(AUTOEP_LAYERS_KEY_LEGACY) + if not isinstance(autoep_layers, list): + return False + return any( + isinstance(entry, dict) + and entry.get(AUTOEP_ZERO3_EXPERT_STATE_FORMAT_KEY) == AUTOEP_ZERO3_PARTITIONED_EXPERT_STATE_FORMAT + for entry in autoep_layers) + + +def _raise_if_autoep_zero3_partitioned_state(state_dict): + if _has_autoep_zero3_partitioned_metadata(state_dict): + raise NotImplementedError("zero_to_fp32 does not support AutoEP ZeRO-3 partition-native checkpoints. " + "AutoEP expert parameters are partitioned over expert replica groups, so " + "global data-parallel consolidation would produce incomplete expert tensors. " + "Use ds_to_universal.py for expert-aware conversion.") + + +def _raise_if_autoep_zero3_partitioned_checkpoint(model_files): + for file in model_files: + state_dict = torch.load(file, map_location=device, weights_only=False) + _raise_if_autoep_zero3_partitioned_state(state_dict) + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device, weights_only=False) + _raise_if_autoep_zero3_partitioned_state(state_dict) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + total_files = len(files) + state_dicts = [] + for f in tqdm(files, desc='Loading checkpoint shards'): + state_dict = torch.load(f, map_location=device, mmap=True, weights_only=False) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if ZERO_STAGE not in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + # parse_model_states rejects AutoEP ZeRO-3 partition-native checkpoints + # before the expensive optimizer-shard load below. + model_files = get_model_state_files(ds_checkpoint_dir) + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +class GatheredTensor: + """ + A pseudo tensor that collects partitioned weights. + It is more memory efficient when there are multiple groups. + """ + + def __init__(self, flat_groups, flat_groups_offset, offset, partitioned_numel, shape): + self.flat_groups = flat_groups + self.flat_groups_offset = flat_groups_offset + self.offset = offset + self.partitioned_numel = partitioned_numel + self.shape = shape + self.dtype = self.flat_groups[0][0].dtype + + def contiguous(self): + """ + Merge partitioned weights from flat_groups into a single tensor. + """ + end_idx = self.offset + self.partitioned_numel + world_size = len(self.flat_groups) + pad_flat_param_chunks = [] + + for rank_i in range(world_size): + # for each rank, we need to collect weights from related group/groups + flat_groups_at_rank_i = self.flat_groups[rank_i] + start_group_id = None + end_group_id = None + for group_id in range(len(self.flat_groups_offset)): + if self.flat_groups_offset[group_id] <= self.offset < self.flat_groups_offset[group_id + 1]: + start_group_id = group_id + if self.flat_groups_offset[group_id] < end_idx <= self.flat_groups_offset[group_id + 1]: + end_group_id = group_id + break + # collect weights from related group/groups + for group_id in range(start_group_id, end_group_id + 1): + flat_tensor = flat_groups_at_rank_i[group_id] + start_offset = self.offset - self.flat_groups_offset[group_id] + end_offset = min(end_idx, self.flat_groups_offset[group_id + 1]) - self.flat_groups_offset[group_id] + pad_flat_param_chunks.append(flat_tensor[start_offset:end_offset]) + + # collect weights from all ranks + pad_flat_param = torch.cat(pad_flat_param_chunks, dim=0) + param = pad_flat_param[:self.shape.numel()].view(self.shape).contiguous() + return param + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = sum([flat_group.numel() for flat_group in fp32_flat_groups[0]]) * world_size + + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + flat_groups_offset = [0] + list(np.cumsum([flat_tensor.numel() for flat_tensor in fp32_flat_groups[0]])) + for name, shape in tqdm(param_shapes.items(), desc='Gathering sharded weights'): + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # memory efficient tensor + tensor = GatheredTensor(fp32_flat_groups, flat_groups_offset, offset, partitioned_numel, shape) + state_dict[name] = tensor + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def to_torch_tensor(state_dict, return_empty_tensor=False): + """ + Convert state_dict of GatheredTensor to torch tensor + """ + torch_state_dict = {} + converted_tensors = {} + for name, tensor in state_dict.items(): + tensor_id = id(tensor) + if tensor_id in converted_tensors: # shared tensors + shared_tensor = torch_state_dict[converted_tensors[tensor_id]] + torch_state_dict[name] = shared_tensor + else: + converted_tensors[tensor_id] = name + if return_empty_tensor: + torch_state_dict[name] = torch.empty(tensor.shape, dtype=tensor.dtype) + else: + torch_state_dict[name] = tensor.contiguous() + return torch_state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, + tag=None, + exclude_frozen_parameters=False, + lazy_mode=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + - ``lazy_mode``: get state_dict in lazy mode. It returns a dict of pesduo tensor instead of torch tensor, which is more memory efficient. + Convert the pesduo tensor to torch tensor by ``.contiguous()`` + + Returns: + - pytorch ``state_dict`` + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + Note: the above usage may not work if your application doesn't have sufficient free CPU memory. + You may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. Or you can load state_dict in lazy mode :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, lazy_mode=True) # not on cpu + for name, lazy_tensor in state_dict.item(): + tensor = lazy_tensor.contiguous() # to cpu + print(name, tensor) + # del tensor to release memory if it no longer in use + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + state_dict = _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + if lazy_mode: + return state_dict + else: + return to_torch_tensor(state_dict) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, + output_dir, + max_shard_size="5GB", + safe_serialization=False, + tag=None, + exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_dir``: directory to the pytorch fp32 state_dict output files + - ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB + - ``safe_serialization``: whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`). + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + # Dependency pre-check + if safe_serialization: + try: + from safetensors.torch import save_file + except ImportError: + print('If you want to use `safe_serialization`, please `pip install safetensors`') + raise + if max_shard_size is not None: + try: + from huggingface_hub import split_torch_state_dict_into_shards + except ImportError: + print('If you want to use `max_shard_size`, please `pip install huggingface_hub`') + raise + + # Convert zero checkpoint to state_dict + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, + tag, + exclude_frozen_parameters, + lazy_mode=True) + + # Shard the model if it is too big. + weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin" + if max_shard_size is not None: + filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors") + # an memory-efficient approach for sharding + empty_state_dict = to_torch_tensor(state_dict, return_empty_tensor=True) + state_dict_split = split_torch_state_dict_into_shards(empty_state_dict, + filename_pattern=filename_pattern, + max_shard_size=max_shard_size) + else: + from collections import namedtuple + StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"]) + state_dict_split = StateDictSplit(is_sharded=False, + filename_to_tensors={weights_name: list(state_dict.keys())}) + + # Save the model by shard + os.makedirs(output_dir, exist_ok=True) + filename_to_tensors = state_dict_split.filename_to_tensors.items() + for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"): + shard_state_dict = {tensor_name: state_dict[tensor_name] for tensor_name in tensors} + shard_state_dict = to_torch_tensor(shard_state_dict) + output_path = os.path.join(output_dir, shard_file) + if safe_serialization: + save_file(shard_state_dict, output_path, metadata={"format": "pt"}) + else: + torch.save(shard_state_dict, output_path) + # release the memory of current shard + for tensor_name in list(shard_state_dict.keys()): + del state_dict[tensor_name] + del shard_state_dict[tensor_name] + del shard_state_dict + gc.collect() + + # Save index if sharded + if state_dict_split.is_sharded: + index = { + "metadata": state_dict_split.metadata, + "weight_map": state_dict_split.tensor_to_filename, + } + save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json" + save_index_file = os.path.join(output_dir, save_index_file) + with open(save_index_file, "w", encoding="utf-8") as f: + content = json.dumps(index, indent=2, sort_keys=True) + "\n" + f.write(content) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info("Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info("Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument("output_dir", + type=str, + help="directory to the pytorch fp32 state_dict output files" + "(e.g. path/checkpoint-12-output/)") + parser.add_argument( + "--max_shard_size", + type=str, + default="5GB", + help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size" + "lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`" + "We default it to 5GB in order for models to be able to run easily on free-tier google colab instances" + "without CPU OOM issues.") + parser.add_argument( + "--safe_serialization", + default=False, + action='store_true', + help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_dir, + max_shard_size=args.max_shard_size, + safe_serialization=args.safe_serialization, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/train.log b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/train.log new file mode 100644 index 0000000000000000000000000000000000000000..f7573aff51f3c7acae9f221c83314dad2903184c --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/train.log @@ -0,0 +1,5026 @@ +2026-09-14 13:22:24 | INFO | train rank=0 | Starting training project=codebase_v8_droid_joint_50ksteps_bs128_qwen3vl2b_end_padding_mask save_dir=/data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask distributed=True deepspeed=True world_size=4 device=cuda:0 +2026-09-14 13:22:47 | INFO | train rank=0 | No pretrained checkpoint provided; training from scratch +2026-09-14 13:22:47 | INFO | train rank=0 | Initializing Droid LeRobot Dataset: path=/data/NTU_slab/draven/data/DROID fps=15.00 use_relabelled_instruction=False +2026-09-14 13:22:47 | INFO | train rank=0 | Batch config: total_batch_size=128 world_size=4 grad_acc_steps=1 per_device_batch_size=32 +2026-09-14 13:22:47 | INFO | train rank=0 | Dataset image_resize_size=320 +2026-09-14 13:22:47 | INFO | train rank=0 | Droid uses parquet-shard streaming with a per-rank shuffle buffer: batch_size=32 buffer_size=1024 num_workers=8 drop_last=True +2026-09-14 13:38:53 | INFO | train rank=0 | Starting training project=codebase_v8_droid_joint_50ksteps_bs128_qwen3vl2b_end_padding_mask save_dir=/data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask distributed=True deepspeed=True world_size=4 device=cuda:0 +2026-09-14 13:39:14 | INFO | train rank=0 | No pretrained checkpoint provided; training from scratch +2026-09-14 13:39:14 | INFO | train rank=0 | Initializing Droid LeRobot Dataset: path=/data/NTU_slab/draven/data/DROID fps=15.00 use_relabelled_instruction=False +2026-09-14 13:39:14 | INFO | train rank=0 | Batch config: total_batch_size=128 world_size=4 grad_acc_steps=1 per_device_batch_size=32 +2026-09-14 13:39:14 | INFO | train rank=0 | Dataset image_resize_size=320 +2026-09-14 13:39:14 | INFO | train rank=0 | Droid uses parquet-shard streaming with a per-rank shuffle buffer: batch_size=32 buffer_size=1024 num_workers=8 drop_last=True +2026-09-14 13:43:07 | INFO | train rank=0 | Starting training project=codebase_v8_droid_joint_50ksteps_bs128_qwen3vl2b_end_padding_mask save_dir=/data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask distributed=True deepspeed=True world_size=4 device=cuda:0 +2026-09-14 13:43:30 | INFO | train rank=0 | No pretrained checkpoint provided; training from scratch +2026-09-14 13:43:30 | INFO | train rank=0 | Initializing Droid LeRobot Dataset: path=/data/NTU_slab/draven/data/DROID fps=15.00 use_relabelled_instruction=False +2026-09-14 13:43:30 | INFO | train rank=0 | Batch config: total_batch_size=128 world_size=4 grad_acc_steps=1 per_device_batch_size=32 +2026-09-14 13:43:30 | INFO | train rank=0 | Dataset image_resize_size=320 +2026-09-14 13:43:30 | INFO | train rank=0 | Droid uses parquet-shard streaming with a per-rank shuffle buffer: batch_size=32 buffer_size=1024 num_workers=8 drop_last=True +2026-09-14 13:43:36 | INFO | train rank=0 | Droid streaming layout: parquet_shards=50 active_workers=8 worker_buffer_min=128 worker_buffer_max=128 max_active_shards=2 +2026-09-14 13:43:36 | INFO | train rank=0 | Resolved Droid joint action stats from /data/NTU_slab/draven/data/DROID/meta/joint_delta_stats_h8.json: q01=[-0.21484030783176422, -0.3382192850112915, -0.21358108520507812, -0.3538644015789032, -0.3344809114933014, -0.3184388279914856, -0.3800385296344757] q99=[0.21101456880569458, 0.39177048206329346, 0.21355487406253815, 0.3544503450393677, 0.3296765685081482, 0.3758541941642761, 0.38154837489128113] +2026-09-14 13:43:36 | INFO | train rank=0 | Trainable params: 2,688,879,624 / 2,688,879,624 +2026-09-14 13:46:27 | INFO | train rank=0 | Starting training start_step=0 max_steps=50000 remaining_steps=50000 log_interval=10 save_interval=50000 +2026-09-14 13:48:29 | INFO | train rank=0 | step=10/50000 progress=0.0% loss=1.188354 lr=3.600000e-07 lr_backbone=3.600000e-07 elapsed=00:02:01 avg_step=12.18s recent_step=12.18s eta=169:12:01 +2026-09-14 13:48:51 | INFO | train rank=0 | step=20/50000 progress=0.0% loss=1.203709 lr=7.600000e-07 lr_backbone=7.600000e-07 elapsed=00:02:24 avg_step=7.23s recent_step=2.28s eta=100:26:03 +2026-09-14 13:49:18 | INFO | train rank=0 | step=30/50000 progress=0.1% loss=1.231849 lr=1.160000e-06 lr_backbone=1.160000e-06 elapsed=00:02:51 avg_step=5.71s recent_step=2.67s eta=79:18:12 +2026-09-14 13:49:41 | INFO | train rank=0 | step=40/50000 progress=0.1% loss=1.213999 lr=1.560000e-06 lr_backbone=1.560000e-06 elapsed=00:03:14 avg_step=4.85s recent_step=2.27s eta=67:19:37 +2026-09-14 13:50:08 | INFO | train rank=0 | step=50/50000 progress=0.1% loss=1.201466 lr=1.960000e-06 lr_backbone=1.960000e-06 elapsed=00:03:40 avg_step=4.42s recent_step=2.68s eta=61:17:42 +2026-09-14 13:50:29 | INFO | train rank=0 | step=60/50000 progress=0.1% loss=1.179175 lr=2.360000e-06 lr_backbone=2.360000e-06 elapsed=00:04:02 avg_step=4.03s recent_step=2.12s eta=55:57:43 +2026-09-14 13:50:48 | INFO | train rank=0 | step=70/50000 progress=0.1% loss=1.157839 lr=2.760000e-06 lr_backbone=2.760000e-06 elapsed=00:04:21 avg_step=3.74s recent_step=1.95s eta=51:48:45 +2026-09-14 13:51:09 | INFO | train rank=0 | step=80/50000 progress=0.2% loss=1.135914 lr=3.160000e-06 lr_backbone=3.160000e-06 elapsed=00:04:42 avg_step=3.53s recent_step=2.09s eta=48:56:47 +2026-09-14 13:51:37 | INFO | train rank=0 | step=90/50000 progress=0.2% loss=1.130946 lr=3.560000e-06 lr_backbone=3.560000e-06 elapsed=00:05:09 avg_step=3.44s recent_step=2.76s eta=47:45:11 +2026-09-14 13:52:02 | INFO | train rank=0 | step=100/50000 progress=0.2% loss=1.124174 lr=3.960000e-06 lr_backbone=3.960000e-06 elapsed=00:05:35 avg_step=3.35s recent_step=2.55s eta=46:30:06 +2026-09-14 13:52:25 | INFO | train rank=0 | step=110/50000 progress=0.2% loss=1.109153 lr=4.360000e-06 lr_backbone=4.360000e-06 elapsed=00:05:57 avg_step=3.25s recent_step=2.24s eta=45:05:26 +2026-09-14 13:52:51 | INFO | train rank=0 | step=120/50000 progress=0.2% loss=1.100060 lr=4.760000e-06 lr_backbone=4.760000e-06 elapsed=00:06:24 avg_step=3.20s recent_step=2.65s eta=44:22:56 +2026-09-14 13:53:23 | INFO | train rank=0 | step=130/50000 progress=0.3% loss=0.979861 lr=5.160000e-06 lr_backbone=5.160000e-06 elapsed=00:06:55 avg_step=3.20s recent_step=3.15s eta=44:18:59 +2026-09-14 13:53:45 | INFO | train rank=0 | step=140/50000 progress=0.3% loss=1.054357 lr=5.560000e-06 lr_backbone=5.560000e-06 elapsed=00:07:18 avg_step=3.13s recent_step=2.29s eta=43:24:12 +2026-09-14 13:54:05 | INFO | train rank=0 | step=150/50000 progress=0.3% loss=1.005468 lr=5.960000e-06 lr_backbone=5.960000e-06 elapsed=00:07:37 avg_step=3.05s recent_step=1.92s eta=42:16:43 +2026-09-14 13:54:27 | INFO | train rank=0 | step=160/50000 progress=0.3% loss=1.036110 lr=6.360000e-06 lr_backbone=6.360000e-06 elapsed=00:08:00 avg_step=3.00s recent_step=2.28s eta=41:36:03 +2026-09-14 13:54:55 | INFO | train rank=0 | step=170/50000 progress=0.3% loss=0.988663 lr=6.760000e-06 lr_backbone=6.760000e-06 elapsed=00:08:28 avg_step=2.99s recent_step=2.74s eta=41:22:47 +2026-09-14 13:55:21 | INFO | train rank=0 | step=180/50000 progress=0.4% loss=0.913105 lr=7.160000e-06 lr_backbone=7.160000e-06 elapsed=00:08:53 avg_step=2.97s recent_step=2.57s eta=41:02:43 +2026-09-14 13:55:50 | INFO | train rank=0 | step=190/50000 progress=0.4% loss=0.900910 lr=7.560000e-06 lr_backbone=7.560000e-06 elapsed=00:09:22 avg_step=2.96s recent_step=2.89s eta=40:59:02 +2026-09-14 13:56:09 | INFO | train rank=0 | step=200/50000 progress=0.4% loss=0.893876 lr=7.960000e-06 lr_backbone=7.960000e-06 elapsed=00:09:42 avg_step=2.91s recent_step=1.96s eta=40:16:55 +2026-09-14 13:56:32 | INFO | train rank=0 | step=210/50000 progress=0.4% loss=0.855501 lr=8.360000e-06 lr_backbone=8.360000e-06 elapsed=00:10:04 avg_step=2.88s recent_step=2.26s eta=39:50:38 +2026-09-14 13:56:55 | INFO | train rank=0 | step=220/50000 progress=0.4% loss=0.747119 lr=8.760000e-06 lr_backbone=8.760000e-06 elapsed=00:10:28 avg_step=2.86s recent_step=2.32s eta=39:29:00 +2026-09-14 13:57:17 | INFO | train rank=0 | step=230/50000 progress=0.5% loss=0.694270 lr=9.160000e-06 lr_backbone=9.160000e-06 elapsed=00:10:49 avg_step=2.83s recent_step=2.17s eta=39:03:49 +2026-09-14 13:57:37 | INFO | train rank=0 | step=240/50000 progress=0.5% loss=0.677010 lr=9.560000e-06 lr_backbone=9.560000e-06 elapsed=00:11:10 avg_step=2.79s recent_step=2.06s eta=38:36:43 +2026-09-14 13:58:06 | INFO | train rank=0 | step=250/50000 progress=0.5% loss=0.561498 lr=9.960000e-06 lr_backbone=9.960000e-06 elapsed=00:11:38 avg_step=2.80s recent_step=2.84s eta=38:37:53 +2026-09-14 13:58:31 | INFO | train rank=0 | step=260/50000 progress=0.5% loss=0.482870 lr=1.036000e-05 lr_backbone=1.036000e-05 elapsed=00:12:03 avg_step=2.78s recent_step=2.50s eta=38:28:04 +2026-09-14 13:58:52 | INFO | train rank=0 | step=270/50000 progress=0.5% loss=0.495736 lr=1.076000e-05 lr_backbone=1.076000e-05 elapsed=00:12:25 avg_step=2.76s recent_step=2.13s eta=38:07:34 +2026-09-14 13:59:16 | INFO | train rank=0 | step=280/50000 progress=0.6% loss=0.412582 lr=1.116000e-05 lr_backbone=1.116000e-05 elapsed=00:12:49 avg_step=2.75s recent_step=2.45s eta=37:57:50 +2026-09-14 13:59:42 | INFO | train rank=0 | step=290/50000 progress=0.6% loss=0.387799 lr=1.156000e-05 lr_backbone=1.156000e-05 elapsed=00:13:15 avg_step=2.74s recent_step=2.58s eta=37:52:34 +2026-09-14 14:00:08 | INFO | train rank=0 | step=300/50000 progress=0.6% loss=0.361962 lr=1.196000e-05 lr_backbone=1.196000e-05 elapsed=00:13:40 avg_step=2.74s recent_step=2.53s eta=37:46:20 +2026-09-14 14:00:30 | INFO | train rank=0 | step=310/50000 progress=0.6% loss=0.372816 lr=1.236000e-05 lr_backbone=1.236000e-05 elapsed=00:14:03 avg_step=2.72s recent_step=2.29s eta=37:33:54 +2026-09-14 14:00:55 | INFO | train rank=0 | step=320/50000 progress=0.6% loss=0.346866 lr=1.276000e-05 lr_backbone=1.276000e-05 elapsed=00:14:28 avg_step=2.71s recent_step=2.49s eta=37:27:21 +2026-09-14 14:01:26 | INFO | train rank=0 | step=330/50000 progress=0.7% loss=0.321073 lr=1.316000e-05 lr_backbone=1.316000e-05 elapsed=00:14:58 avg_step=2.72s recent_step=3.03s eta=37:34:52 +2026-09-14 14:01:49 | INFO | train rank=0 | step=340/50000 progress=0.7% loss=0.322917 lr=1.356000e-05 lr_backbone=1.356000e-05 elapsed=00:15:22 avg_step=2.71s recent_step=2.34s eta=37:25:08 +2026-09-14 14:02:15 | INFO | train rank=0 | step=350/50000 progress=0.7% loss=0.263151 lr=1.396000e-05 lr_backbone=1.396000e-05 elapsed=00:15:48 avg_step=2.71s recent_step=2.62s eta=37:22:36 +2026-09-14 14:02:39 | INFO | train rank=0 | step=360/50000 progress=0.7% loss=0.295909 lr=1.436000e-05 lr_backbone=1.436000e-05 elapsed=00:16:12 avg_step=2.70s recent_step=2.40s eta=37:14:58 +2026-09-14 14:03:01 | INFO | train rank=0 | step=370/50000 progress=0.7% loss=0.310321 lr=1.476000e-05 lr_backbone=1.476000e-05 elapsed=00:16:34 avg_step=2.69s recent_step=2.22s eta=37:03:48 +2026-09-14 14:03:26 | INFO | train rank=0 | step=380/50000 progress=0.8% loss=0.269754 lr=1.516000e-05 lr_backbone=1.516000e-05 elapsed=00:16:59 avg_step=2.68s recent_step=2.44s eta=36:57:59 +2026-09-14 14:03:53 | INFO | train rank=0 | step=390/50000 progress=0.8% loss=0.310154 lr=1.556000e-05 lr_backbone=1.556000e-05 elapsed=00:17:26 avg_step=2.68s recent_step=2.72s eta=36:58:18 +2026-09-14 14:04:17 | INFO | train rank=0 | step=400/50000 progress=0.8% loss=0.320251 lr=1.596000e-05 lr_backbone=1.596000e-05 elapsed=00:17:50 avg_step=2.68s recent_step=2.39s eta=36:51:52 +2026-09-14 14:04:43 | INFO | train rank=0 | step=410/50000 progress=0.8% loss=0.279843 lr=1.636000e-05 lr_backbone=1.636000e-05 elapsed=00:18:16 avg_step=2.67s recent_step=2.61s eta=36:50:06 +2026-09-14 14:05:06 | INFO | train rank=0 | step=420/50000 progress=0.8% loss=0.285131 lr=1.676000e-05 lr_backbone=1.676000e-05 elapsed=00:18:39 avg_step=2.66s recent_step=2.27s eta=36:41:46 +2026-09-14 14:05:30 | INFO | train rank=0 | step=430/50000 progress=0.9% loss=0.349291 lr=1.716000e-05 lr_backbone=1.716000e-05 elapsed=00:19:03 avg_step=2.66s recent_step=2.45s eta=36:37:16 +2026-09-14 14:05:53 | INFO | train rank=0 | step=440/50000 progress=0.9% loss=0.229608 lr=1.756000e-05 lr_backbone=1.756000e-05 elapsed=00:19:26 avg_step=2.65s recent_step=2.31s eta=36:30:18 +2026-09-14 14:06:18 | INFO | train rank=0 | step=450/50000 progress=0.9% loss=0.249500 lr=1.796000e-05 lr_backbone=1.796000e-05 elapsed=00:19:50 avg_step=2.65s recent_step=2.41s eta=36:25:28 +2026-09-14 14:06:45 | INFO | train rank=0 | step=460/50000 progress=0.9% loss=0.243035 lr=1.836000e-05 lr_backbone=1.836000e-05 elapsed=00:20:17 avg_step=2.65s recent_step=2.71s eta=36:26:09 +2026-09-14 14:07:06 | INFO | train rank=0 | step=470/50000 progress=0.9% loss=0.211088 lr=1.876000e-05 lr_backbone=1.876000e-05 elapsed=00:20:39 avg_step=2.64s recent_step=2.14s eta=36:16:51 +2026-09-14 14:07:30 | INFO | train rank=0 | step=480/50000 progress=1.0% loss=0.243504 lr=1.916000e-05 lr_backbone=1.916000e-05 elapsed=00:21:03 avg_step=2.63s recent_step=2.41s eta=36:12:29 +2026-09-14 14:07:55 | INFO | train rank=0 | step=490/50000 progress=1.0% loss=0.288641 lr=1.956000e-05 lr_backbone=1.956000e-05 elapsed=00:21:27 avg_step=2.63s recent_step=2.45s eta=36:08:56 +2026-09-14 14:08:17 | INFO | train rank=0 | step=500/50000 progress=1.0% loss=0.216981 lr=1.996000e-05 lr_backbone=1.996000e-05 elapsed=00:21:50 avg_step=2.62s recent_step=2.20s eta=36:01:30 +2026-09-14 14:08:38 | INFO | train rank=0 | step=510/50000 progress=1.0% loss=0.198885 lr=2.036000e-05 lr_backbone=2.036000e-05 elapsed=00:22:10 avg_step=2.61s recent_step=2.08s eta=35:52:23 +2026-09-14 14:09:03 | INFO | train rank=0 | step=520/50000 progress=1.0% loss=0.254130 lr=2.076000e-05 lr_backbone=2.076000e-05 elapsed=00:22:35 avg_step=2.61s recent_step=2.50s eta=35:50:10 +2026-09-14 14:09:27 | INFO | train rank=0 | step=530/50000 progress=1.1% loss=0.254473 lr=2.116000e-05 lr_backbone=2.116000e-05 elapsed=00:23:00 avg_step=2.60s recent_step=2.47s eta=35:47:33 +2026-09-14 14:09:49 | INFO | train rank=0 | step=540/50000 progress=1.1% loss=0.276839 lr=2.156000e-05 lr_backbone=2.156000e-05 elapsed=00:23:22 avg_step=2.60s recent_step=2.22s eta=35:41:14 +2026-09-14 14:10:15 | INFO | train rank=0 | step=550/50000 progress=1.1% loss=0.252633 lr=2.196000e-05 lr_backbone=2.196000e-05 elapsed=00:23:48 avg_step=2.60s recent_step=2.55s eta=35:40:03 +2026-09-14 14:10:36 | INFO | train rank=0 | step=560/50000 progress=1.1% loss=0.246619 lr=2.236000e-05 lr_backbone=2.236000e-05 elapsed=00:24:09 avg_step=2.59s recent_step=2.14s eta=35:32:58 +2026-09-14 14:11:02 | INFO | train rank=0 | step=570/50000 progress=1.1% loss=0.216940 lr=2.276000e-05 lr_backbone=2.276000e-05 elapsed=00:24:34 avg_step=2.59s recent_step=2.54s eta=35:31:49 +2026-09-14 14:11:23 | INFO | train rank=0 | step=580/50000 progress=1.2% loss=0.230885 lr=2.316000e-05 lr_backbone=2.316000e-05 elapsed=00:24:55 avg_step=2.58s recent_step=2.10s eta=35:24:27 +2026-09-14 14:11:47 | INFO | train rank=0 | step=590/50000 progress=1.2% loss=0.238247 lr=2.356000e-05 lr_backbone=2.356000e-05 elapsed=00:25:20 avg_step=2.58s recent_step=2.45s eta=35:22:13 +2026-09-14 14:12:06 | INFO | train rank=0 | step=600/50000 progress=1.2% loss=0.259885 lr=2.396000e-05 lr_backbone=2.396000e-05 elapsed=00:25:39 avg_step=2.57s recent_step=1.91s eta=35:12:39 +2026-09-14 14:12:32 | INFO | train rank=0 | step=610/50000 progress=1.2% loss=0.217432 lr=2.436000e-05 lr_backbone=2.436000e-05 elapsed=00:26:05 avg_step=2.57s recent_step=2.55s eta=35:11:58 +2026-09-14 14:12:53 | INFO | train rank=0 | step=620/50000 progress=1.2% loss=0.234061 lr=2.476000e-05 lr_backbone=2.476000e-05 elapsed=00:26:26 avg_step=2.56s recent_step=2.15s eta=35:06:00 +2026-09-14 14:13:14 | INFO | train rank=0 | step=630/50000 progress=1.3% loss=0.237063 lr=2.516000e-05 lr_backbone=2.516000e-05 elapsed=00:26:47 avg_step=2.55s recent_step=2.09s eta=34:59:30 +2026-09-14 14:13:39 | INFO | train rank=0 | step=640/50000 progress=1.3% loss=0.250796 lr=2.556000e-05 lr_backbone=2.556000e-05 elapsed=00:27:11 avg_step=2.55s recent_step=2.44s eta=34:57:38 +2026-09-14 14:14:04 | INFO | train rank=0 | step=650/50000 progress=1.3% loss=0.202480 lr=2.596000e-05 lr_backbone=2.596000e-05 elapsed=00:27:36 avg_step=2.55s recent_step=2.50s eta=34:56:36 +2026-09-14 14:14:23 | INFO | train rank=0 | step=660/50000 progress=1.3% loss=0.253517 lr=2.636000e-05 lr_backbone=2.636000e-05 elapsed=00:27:56 avg_step=2.54s recent_step=1.93s eta=34:48:29 +2026-09-14 14:14:45 | INFO | train rank=0 | step=670/50000 progress=1.3% loss=0.248501 lr=2.676000e-05 lr_backbone=2.676000e-05 elapsed=00:28:18 avg_step=2.54s recent_step=2.23s eta=34:44:14 +2026-09-14 14:15:11 | INFO | train rank=0 | step=680/50000 progress=1.4% loss=0.301181 lr=2.716000e-05 lr_backbone=2.716000e-05 elapsed=00:28:44 avg_step=2.54s recent_step=2.57s eta=34:44:14 +2026-09-14 14:15:40 | INFO | train rank=0 | step=690/50000 progress=1.4% loss=0.185126 lr=2.756000e-05 lr_backbone=2.756000e-05 elapsed=00:29:13 avg_step=2.54s recent_step=2.94s eta=34:48:40 +2026-09-14 14:16:02 | INFO | train rank=0 | step=700/50000 progress=1.4% loss=0.218977 lr=2.796000e-05 lr_backbone=2.796000e-05 elapsed=00:29:34 avg_step=2.54s recent_step=2.13s eta=34:43:28 +2026-09-14 14:16:26 | INFO | train rank=0 | step=710/50000 progress=1.4% loss=0.202615 lr=2.836000e-05 lr_backbone=2.836000e-05 elapsed=00:29:59 avg_step=2.53s recent_step=2.46s eta=34:42:13 +2026-09-14 14:16:51 | INFO | train rank=0 | step=720/50000 progress=1.4% loss=0.232880 lr=2.876000e-05 lr_backbone=2.876000e-05 elapsed=00:30:24 avg_step=2.53s recent_step=2.48s eta=34:41:08 +2026-09-14 14:17:24 | INFO | train rank=0 | step=730/50000 progress=1.5% loss=0.212456 lr=2.916000e-05 lr_backbone=2.916000e-05 elapsed=00:30:57 avg_step=2.54s recent_step=3.30s eta=34:49:17 +2026-09-14 14:17:45 | INFO | train rank=0 | step=740/50000 progress=1.5% loss=0.216195 lr=2.956000e-05 lr_backbone=2.956000e-05 elapsed=00:31:18 avg_step=2.54s recent_step=2.08s eta=34:43:43 +2026-09-14 14:18:10 | INFO | train rank=0 | step=750/50000 progress=1.5% loss=0.260956 lr=2.996000e-05 lr_backbone=2.996000e-05 elapsed=00:31:43 avg_step=2.54s recent_step=2.55s eta=34:43:26 +2026-09-14 14:18:31 | INFO | train rank=0 | step=760/50000 progress=1.5% loss=0.248588 lr=3.036000e-05 lr_backbone=3.036000e-05 elapsed=00:32:04 avg_step=2.53s recent_step=2.07s eta=34:37:59 +2026-09-14 14:18:58 | INFO | train rank=0 | step=770/50000 progress=1.5% loss=0.211887 lr=3.076000e-05 lr_backbone=3.076000e-05 elapsed=00:32:31 avg_step=2.53s recent_step=2.72s eta=34:39:34 +2026-09-14 14:19:23 | INFO | train rank=0 | step=780/50000 progress=1.6% loss=0.234526 lr=3.116000e-05 lr_backbone=3.116000e-05 elapsed=00:32:55 avg_step=2.53s recent_step=2.44s eta=34:38:07 +2026-09-14 14:19:45 | INFO | train rank=0 | step=790/50000 progress=1.6% loss=0.234879 lr=3.156000e-05 lr_backbone=3.156000e-05 elapsed=00:33:18 avg_step=2.53s recent_step=2.28s eta=34:35:06 +2026-09-14 14:20:09 | INFO | train rank=0 | step=800/50000 progress=1.6% loss=0.198000 lr=3.196000e-05 lr_backbone=3.196000e-05 elapsed=00:33:41 avg_step=2.53s recent_step=2.32s eta=34:32:32 +2026-09-14 14:20:34 | INFO | train rank=0 | step=810/50000 progress=1.6% loss=0.240658 lr=3.236000e-05 lr_backbone=3.236000e-05 elapsed=00:34:07 avg_step=2.53s recent_step=2.52s eta=34:31:59 +2026-09-14 14:20:56 | INFO | train rank=0 | step=820/50000 progress=1.6% loss=0.246084 lr=3.276000e-05 lr_backbone=3.276000e-05 elapsed=00:34:29 avg_step=2.52s recent_step=2.20s eta=34:28:18 +2026-09-14 14:21:16 | INFO | train rank=0 | step=830/50000 progress=1.7% loss=0.181437 lr=3.316000e-05 lr_backbone=3.316000e-05 elapsed=00:34:49 avg_step=2.52s recent_step=1.99s eta=34:22:39 +2026-09-14 14:21:39 | INFO | train rank=0 | step=840/50000 progress=1.7% loss=0.205334 lr=3.356000e-05 lr_backbone=3.356000e-05 elapsed=00:35:12 avg_step=2.51s recent_step=2.32s eta=34:20:21 +2026-09-14 14:22:02 | INFO | train rank=0 | step=850/50000 progress=1.7% loss=0.216565 lr=3.396000e-05 lr_backbone=3.396000e-05 elapsed=00:35:35 avg_step=2.51s recent_step=2.34s eta=34:18:14 +2026-09-14 14:22:22 | INFO | train rank=0 | step=860/50000 progress=1.7% loss=0.199583 lr=3.436000e-05 lr_backbone=3.436000e-05 elapsed=00:35:55 avg_step=2.51s recent_step=2.00s eta=34:12:58 +2026-09-14 14:22:48 | INFO | train rank=0 | step=870/50000 progress=1.7% loss=0.182488 lr=3.476000e-05 lr_backbone=3.476000e-05 elapsed=00:36:21 avg_step=2.51s recent_step=2.58s eta=34:13:16 +2026-09-14 14:23:10 | INFO | train rank=0 | step=880/50000 progress=1.8% loss=0.181963 lr=3.516000e-05 lr_backbone=3.516000e-05 elapsed=00:36:43 avg_step=2.50s recent_step=2.22s eta=34:10:11 +2026-09-14 14:23:42 | INFO | train rank=0 | step=890/50000 progress=1.8% loss=0.211274 lr=3.556000e-05 lr_backbone=3.556000e-05 elapsed=00:37:15 avg_step=2.51s recent_step=3.14s eta=34:15:37 +2026-09-14 14:24:09 | INFO | train rank=0 | step=900/50000 progress=1.8% loss=0.218281 lr=3.596000e-05 lr_backbone=3.596000e-05 elapsed=00:37:42 avg_step=2.51s recent_step=2.71s eta=34:17:03 +2026-09-14 14:24:34 | INFO | train rank=0 | step=910/50000 progress=1.8% loss=0.238470 lr=3.636000e-05 lr_backbone=3.636000e-05 elapsed=00:38:07 avg_step=2.51s recent_step=2.52s eta=34:16:40 +2026-09-14 14:24:58 | INFO | train rank=0 | step=920/50000 progress=1.8% loss=0.187841 lr=3.676000e-05 lr_backbone=3.676000e-05 elapsed=00:38:30 avg_step=2.51s recent_step=2.34s eta=34:14:42 +2026-09-14 14:25:23 | INFO | train rank=0 | step=930/50000 progress=1.9% loss=0.274339 lr=3.716000e-05 lr_backbone=3.716000e-05 elapsed=00:38:55 avg_step=2.51s recent_step=2.50s eta=34:14:09 +2026-09-14 14:25:47 | INFO | train rank=0 | step=940/50000 progress=1.9% loss=0.201572 lr=3.756000e-05 lr_backbone=3.756000e-05 elapsed=00:39:19 avg_step=2.51s recent_step=2.40s eta=34:12:47 +2026-09-14 14:26:09 | INFO | train rank=0 | step=950/50000 progress=1.9% loss=0.183388 lr=3.796000e-05 lr_backbone=3.796000e-05 elapsed=00:39:41 avg_step=2.51s recent_step=2.21s eta=34:09:45 +2026-09-14 14:26:31 | INFO | train rank=0 | step=960/50000 progress=1.9% loss=0.273142 lr=3.836000e-05 lr_backbone=3.836000e-05 elapsed=00:40:04 avg_step=2.50s recent_step=2.22s eta=34:06:52 +2026-09-14 14:27:01 | INFO | train rank=0 | step=970/50000 progress=1.9% loss=0.252222 lr=3.876000e-05 lr_backbone=3.876000e-05 elapsed=00:40:33 avg_step=2.51s recent_step=2.97s eta=34:10:21 +2026-09-14 14:27:25 | INFO | train rank=0 | step=980/50000 progress=2.0% loss=0.209958 lr=3.916000e-05 lr_backbone=3.916000e-05 elapsed=00:40:58 avg_step=2.51s recent_step=2.46s eta=34:09:33 +2026-09-14 14:27:46 | INFO | train rank=0 | step=990/50000 progress=2.0% loss=0.190704 lr=3.956000e-05 lr_backbone=3.956000e-05 elapsed=00:41:19 avg_step=2.50s recent_step=2.08s eta=34:05:35 +2026-09-14 14:28:11 | INFO | train rank=0 | step=1000/50000 progress=2.0% loss=0.207850 lr=3.996000e-05 lr_backbone=3.996000e-05 elapsed=00:41:43 avg_step=2.50s recent_step=2.47s eta=34:04:55 +2026-09-14 14:28:41 | INFO | train rank=0 | step=1010/50000 progress=2.0% loss=0.250460 lr=4.036000e-05 lr_backbone=4.036000e-05 elapsed=00:42:14 avg_step=2.51s recent_step=3.06s eta=34:08:58 +2026-09-14 14:29:06 | INFO | train rank=0 | step=1020/50000 progress=2.0% loss=0.173495 lr=4.076000e-05 lr_backbone=4.076000e-05 elapsed=00:42:39 avg_step=2.51s recent_step=2.48s eta=34:08:20 +2026-09-14 14:29:25 | INFO | train rank=0 | step=1030/50000 progress=2.1% loss=0.250598 lr=4.116000e-05 lr_backbone=4.116000e-05 elapsed=00:42:58 avg_step=2.50s recent_step=1.90s eta=34:03:05 +2026-09-14 14:29:51 | INFO | train rank=0 | step=1040/50000 progress=2.1% loss=0.233624 lr=4.156000e-05 lr_backbone=4.156000e-05 elapsed=00:43:23 avg_step=2.50s recent_step=2.55s eta=34:03:01 +2026-09-14 14:30:15 | INFO | train rank=0 | step=1050/50000 progress=2.1% loss=0.244548 lr=4.196000e-05 lr_backbone=4.196000e-05 elapsed=00:43:48 avg_step=2.50s recent_step=2.45s eta=34:02:12 +2026-09-14 14:30:42 | INFO | train rank=0 | step=1060/50000 progress=2.1% loss=0.230803 lr=4.236000e-05 lr_backbone=4.236000e-05 elapsed=00:44:15 avg_step=2.51s recent_step=2.71s eta=34:03:24 +2026-09-14 14:31:10 | INFO | train rank=0 | step=1070/50000 progress=2.1% loss=0.206632 lr=4.276000e-05 lr_backbone=4.276000e-05 elapsed=00:44:43 avg_step=2.51s recent_step=2.78s eta=34:05:03 +2026-09-14 14:31:46 | INFO | train rank=0 | step=1080/50000 progress=2.2% loss=0.192495 lr=4.316000e-05 lr_backbone=4.316000e-05 elapsed=00:45:18 avg_step=2.52s recent_step=3.57s eta=34:12:40 +2026-09-14 14:32:24 | INFO | train rank=0 | step=1090/50000 progress=2.2% loss=0.236403 lr=4.356000e-05 lr_backbone=4.356000e-05 elapsed=00:45:56 avg_step=2.53s recent_step=3.78s eta=34:21:43 +2026-09-14 14:32:54 | INFO | train rank=0 | step=1100/50000 progress=2.2% loss=0.212736 lr=4.396000e-05 lr_backbone=4.396000e-05 elapsed=00:46:27 avg_step=2.53s recent_step=3.07s eta=34:25:18 +2026-09-14 14:33:32 | INFO | train rank=0 | step=1110/50000 progress=2.2% loss=0.209365 lr=4.436000e-05 lr_backbone=4.436000e-05 elapsed=00:47:05 avg_step=2.55s recent_step=3.82s eta=34:34:18 +2026-09-14 14:33:57 | INFO | train rank=0 | step=1120/50000 progress=2.2% loss=0.188593 lr=4.476000e-05 lr_backbone=4.476000e-05 elapsed=00:47:30 avg_step=2.55s recent_step=2.47s eta=34:33:21 +2026-09-14 14:34:46 | INFO | train rank=0 | step=1130/50000 progress=2.3% loss=0.171594 lr=4.516000e-05 lr_backbone=4.516000e-05 elapsed=00:48:18 avg_step=2.57s recent_step=4.84s eta=34:49:28 +2026-09-14 14:35:14 | INFO | train rank=0 | step=1140/50000 progress=2.3% loss=0.195473 lr=4.556000e-05 lr_backbone=4.556000e-05 elapsed=00:48:47 avg_step=2.57s recent_step=2.82s eta=34:50:54 +2026-09-14 14:35:38 | INFO | train rank=0 | step=1150/50000 progress=2.3% loss=0.238377 lr=4.596000e-05 lr_backbone=4.596000e-05 elapsed=00:49:11 avg_step=2.57s recent_step=2.43s eta=34:49:28 +2026-09-14 14:36:17 | INFO | train rank=0 | step=1160/50000 progress=2.3% loss=0.218618 lr=4.636000e-05 lr_backbone=4.636000e-05 elapsed=00:49:50 avg_step=2.58s recent_step=3.90s eta=34:58:26 +2026-09-14 14:37:02 | INFO | train rank=0 | step=1170/50000 progress=2.3% loss=0.240016 lr=4.676000e-05 lr_backbone=4.676000e-05 elapsed=00:50:35 avg_step=2.59s recent_step=4.47s eta=35:11:10 +2026-09-14 14:37:31 | INFO | train rank=0 | step=1180/50000 progress=2.4% loss=0.212350 lr=4.716000e-05 lr_backbone=4.716000e-05 elapsed=00:51:04 avg_step=2.60s recent_step=2.92s eta=35:12:58 +2026-09-14 14:38:09 | INFO | train rank=0 | step=1190/50000 progress=2.4% loss=0.199527 lr=4.756000e-05 lr_backbone=4.756000e-05 elapsed=00:51:41 avg_step=2.61s recent_step=3.76s eta=35:20:27 +2026-09-14 14:38:46 | INFO | train rank=0 | step=1200/50000 progress=2.4% loss=0.227495 lr=4.796000e-05 lr_backbone=4.796000e-05 elapsed=00:52:19 avg_step=2.62s recent_step=3.72s eta=35:27:34 +2026-09-14 14:39:35 | INFO | train rank=0 | step=1210/50000 progress=2.4% loss=0.214673 lr=4.836000e-05 lr_backbone=4.836000e-05 elapsed=00:53:08 avg_step=2.63s recent_step=4.91s eta=35:42:34 +2026-09-14 14:40:04 | INFO | train rank=0 | step=1220/50000 progress=2.4% loss=0.203900 lr=4.876000e-05 lr_backbone=4.876000e-05 elapsed=00:53:37 avg_step=2.64s recent_step=2.96s eta=35:44:16 +2026-09-14 14:40:34 | INFO | train rank=0 | step=1230/50000 progress=2.5% loss=0.210962 lr=4.916000e-05 lr_backbone=4.916000e-05 elapsed=00:54:07 avg_step=2.64s recent_step=2.94s eta=35:45:48 +2026-09-14 14:41:05 | INFO | train rank=0 | step=1240/50000 progress=2.5% loss=0.171759 lr=4.956000e-05 lr_backbone=4.956000e-05 elapsed=00:54:38 avg_step=2.64s recent_step=3.10s eta=35:48:23 +2026-09-14 14:41:51 | INFO | train rank=0 | step=1250/50000 progress=2.5% loss=0.215163 lr=4.996000e-05 lr_backbone=4.996000e-05 elapsed=00:55:24 avg_step=2.66s recent_step=4.60s eta=36:00:40 +2026-09-14 14:42:27 | INFO | train rank=0 | step=1260/50000 progress=2.5% loss=0.208120 lr=5.036000e-05 lr_backbone=5.036000e-05 elapsed=00:56:00 avg_step=2.67s recent_step=3.61s eta=36:06:21 +2026-09-14 14:43:06 | INFO | train rank=0 | step=1270/50000 progress=2.5% loss=0.198639 lr=5.076000e-05 lr_backbone=5.076000e-05 elapsed=00:56:39 avg_step=2.68s recent_step=3.93s eta=36:13:58 +2026-09-14 14:43:35 | INFO | train rank=0 | step=1280/50000 progress=2.6% loss=0.246651 lr=5.116000e-05 lr_backbone=5.116000e-05 elapsed=00:57:08 avg_step=2.68s recent_step=2.93s eta=36:15:06 +2026-09-14 14:44:18 | INFO | train rank=0 | step=1290/50000 progress=2.6% loss=0.215197 lr=5.156000e-05 lr_backbone=5.156000e-05 elapsed=00:57:51 avg_step=2.69s recent_step=4.29s eta=36:24:50 +2026-09-14 14:44:55 | INFO | train rank=0 | step=1300/50000 progress=2.6% loss=0.242495 lr=5.196000e-05 lr_backbone=5.196000e-05 elapsed=00:58:28 avg_step=2.70s recent_step=3.68s eta=36:30:33 +2026-09-14 14:45:27 | INFO | train rank=0 | step=1310/50000 progress=2.6% loss=0.173183 lr=5.236000e-05 lr_backbone=5.236000e-05 elapsed=00:59:00 avg_step=2.70s recent_step=3.22s eta=36:33:20 +2026-09-14 14:46:03 | INFO | train rank=0 | step=1320/50000 progress=2.6% loss=0.194598 lr=5.276000e-05 lr_backbone=5.276000e-05 elapsed=00:59:36 avg_step=2.71s recent_step=3.56s eta=36:38:09 +2026-09-14 14:46:52 | INFO | train rank=0 | step=1330/50000 progress=2.7% loss=0.210164 lr=5.316000e-05 lr_backbone=5.316000e-05 elapsed=01:00:24 avg_step=2.73s recent_step=4.87s eta=36:50:51 +2026-09-14 14:47:27 | INFO | train rank=0 | step=1340/50000 progress=2.7% loss=0.208300 lr=5.356000e-05 lr_backbone=5.356000e-05 elapsed=01:00:59 avg_step=2.73s recent_step=3.49s eta=36:55:01 +2026-09-14 14:47:50 | INFO | train rank=0 | step=1350/50000 progress=2.7% loss=0.185374 lr=5.396000e-05 lr_backbone=5.396000e-05 elapsed=01:01:23 avg_step=2.73s recent_step=2.37s eta=36:52:24 +2026-09-14 14:48:31 | INFO | train rank=0 | step=1360/50000 progress=2.7% loss=0.179171 lr=5.436000e-05 lr_backbone=5.436000e-05 elapsed=01:02:04 avg_step=2.74s recent_step=4.09s eta=37:00:04 +2026-09-14 14:49:14 | INFO | train rank=0 | step=1370/50000 progress=2.7% loss=0.248833 lr=5.476000e-05 lr_backbone=5.476000e-05 elapsed=01:02:47 avg_step=2.75s recent_step=4.33s eta=37:08:59 +2026-09-14 14:49:51 | INFO | train rank=0 | step=1380/50000 progress=2.8% loss=0.199956 lr=5.516000e-05 lr_backbone=5.516000e-05 elapsed=01:03:24 avg_step=2.76s recent_step=3.65s eta=37:13:48 +2026-09-14 14:50:23 | INFO | train rank=0 | step=1390/50000 progress=2.8% loss=0.234787 lr=5.556000e-05 lr_backbone=5.556000e-05 elapsed=01:03:56 avg_step=2.76s recent_step=3.19s eta=37:15:53 +2026-09-14 14:50:49 | INFO | train rank=0 | step=1400/50000 progress=2.8% loss=0.247749 lr=5.596000e-05 lr_backbone=5.596000e-05 elapsed=01:04:22 avg_step=2.76s recent_step=2.63s eta=37:14:39 +2026-09-14 14:51:40 | INFO | train rank=0 | step=1410/50000 progress=2.8% loss=0.187840 lr=5.636000e-05 lr_backbone=5.636000e-05 elapsed=01:05:13 avg_step=2.78s recent_step=5.10s eta=37:27:40 +2026-09-14 14:52:16 | INFO | train rank=0 | step=1420/50000 progress=2.8% loss=0.170178 lr=5.676000e-05 lr_backbone=5.676000e-05 elapsed=01:05:49 avg_step=2.78s recent_step=3.59s eta=37:31:51 +2026-09-14 14:52:46 | INFO | train rank=0 | step=1430/50000 progress=2.9% loss=0.206715 lr=5.716000e-05 lr_backbone=5.716000e-05 elapsed=01:06:19 avg_step=2.78s recent_step=2.98s eta=37:32:30 +2026-09-14 14:53:13 | INFO | train rank=0 | step=1440/50000 progress=2.9% loss=0.188895 lr=5.756000e-05 lr_backbone=5.756000e-05 elapsed=01:06:46 avg_step=2.78s recent_step=2.75s eta=37:31:52 +2026-09-14 14:54:04 | INFO | train rank=0 | step=1450/50000 progress=2.9% loss=0.168412 lr=5.796000e-05 lr_backbone=5.796000e-05 elapsed=01:07:36 avg_step=2.80s recent_step=5.02s eta=37:43:53 +2026-09-14 14:54:35 | INFO | train rank=0 | step=1460/50000 progress=2.9% loss=0.257845 lr=5.836000e-05 lr_backbone=5.836000e-05 elapsed=01:08:08 avg_step=2.80s recent_step=3.15s eta=37:45:21 +2026-09-14 14:55:10 | INFO | train rank=0 | step=1470/50000 progress=2.9% loss=0.202990 lr=5.876000e-05 lr_backbone=5.876000e-05 elapsed=01:08:43 avg_step=2.80s recent_step=3.48s eta=37:48:35 +2026-09-14 14:55:44 | INFO | train rank=0 | step=1480/50000 progress=3.0% loss=0.190512 lr=5.916000e-05 lr_backbone=5.916000e-05 elapsed=01:09:17 avg_step=2.81s recent_step=3.46s eta=37:51:44 +2026-09-14 14:56:27 | INFO | train rank=0 | step=1490/50000 progress=3.0% loss=0.257725 lr=5.956000e-05 lr_backbone=5.956000e-05 elapsed=01:10:00 avg_step=2.82s recent_step=4.30s eta=37:59:22 +2026-09-14 14:56:59 | INFO | train rank=0 | step=1500/50000 progress=3.0% loss=0.181588 lr=5.996000e-05 lr_backbone=5.996000e-05 elapsed=01:10:32 avg_step=2.82s recent_step=3.14s eta=38:00:38 +2026-09-14 14:57:26 | INFO | train rank=0 | step=1510/50000 progress=3.0% loss=0.159207 lr=6.036000e-05 lr_backbone=6.036000e-05 elapsed=01:10:59 avg_step=2.82s recent_step=2.76s eta=37:59:50 +2026-09-14 14:57:59 | INFO | train rank=0 | step=1520/50000 progress=3.0% loss=0.183425 lr=6.076000e-05 lr_backbone=6.076000e-05 elapsed=01:11:32 avg_step=2.82s recent_step=3.30s eta=38:01:54 +2026-09-14 14:58:55 | INFO | train rank=0 | step=1530/50000 progress=3.1% loss=0.214946 lr=6.116000e-05 lr_backbone=6.116000e-05 elapsed=01:12:28 avg_step=2.84s recent_step=5.54s eta=38:15:45 +2026-09-14 14:59:22 | INFO | train rank=0 | step=1540/50000 progress=3.1% loss=0.208245 lr=6.156000e-05 lr_backbone=6.156000e-05 elapsed=01:12:54 avg_step=2.84s recent_step=2.68s eta=38:14:27 +2026-09-14 14:59:53 | INFO | train rank=0 | step=1550/50000 progress=3.1% loss=0.197759 lr=6.196000e-05 lr_backbone=6.196000e-05 elapsed=01:13:26 avg_step=2.84s recent_step=3.14s eta=38:15:33 +2026-09-14 15:00:29 | INFO | train rank=0 | step=1560/50000 progress=3.1% loss=0.168896 lr=6.236000e-05 lr_backbone=6.236000e-05 elapsed=01:14:02 avg_step=2.85s recent_step=3.63s eta=38:19:09 +2026-09-14 15:01:08 | INFO | train rank=0 | step=1570/50000 progress=3.1% loss=0.183736 lr=6.276000e-05 lr_backbone=6.276000e-05 elapsed=01:14:40 avg_step=2.85s recent_step=3.82s eta=38:23:40 +2026-09-14 15:01:31 | INFO | train rank=0 | step=1580/50000 progress=3.2% loss=0.231807 lr=6.316000e-05 lr_backbone=6.316000e-05 elapsed=01:15:04 avg_step=2.85s recent_step=2.38s eta=38:20:47 +2026-09-14 15:02:00 | INFO | train rank=0 | step=1590/50000 progress=3.2% loss=0.192490 lr=6.356000e-05 lr_backbone=6.356000e-05 elapsed=01:15:32 avg_step=2.85s recent_step=2.82s eta=38:20:07 +2026-09-14 15:02:30 | INFO | train rank=0 | step=1600/50000 progress=3.2% loss=0.206506 lr=6.396000e-05 lr_backbone=6.396000e-05 elapsed=01:16:03 avg_step=2.85s recent_step=3.05s eta=38:20:41 +2026-09-14 15:03:12 | INFO | train rank=0 | step=1610/50000 progress=3.2% loss=0.186650 lr=6.436000e-05 lr_backbone=6.436000e-05 elapsed=01:16:45 avg_step=2.86s recent_step=4.19s eta=38:26:55 +2026-09-14 15:03:42 | INFO | train rank=0 | step=1620/50000 progress=3.2% loss=0.212829 lr=6.476000e-05 lr_backbone=6.476000e-05 elapsed=01:17:15 avg_step=2.86s recent_step=2.99s eta=38:27:05 +2026-09-14 15:04:18 | INFO | train rank=0 | step=1630/50000 progress=3.3% loss=0.174173 lr=6.516000e-05 lr_backbone=6.516000e-05 elapsed=01:17:51 avg_step=2.87s recent_step=3.65s eta=38:30:32 +2026-09-14 15:04:56 | INFO | train rank=0 | step=1640/50000 progress=3.3% loss=0.221811 lr=6.556000e-05 lr_backbone=6.556000e-05 elapsed=01:18:29 avg_step=2.87s recent_step=3.75s eta=38:34:24 +2026-09-14 15:05:44 | INFO | train rank=0 | step=1650/50000 progress=3.3% loss=0.201863 lr=6.596000e-05 lr_backbone=6.596000e-05 elapsed=01:19:17 avg_step=2.88s recent_step=4.79s eta=38:43:18 +2026-09-14 15:06:21 | INFO | train rank=0 | step=1660/50000 progress=3.3% loss=0.210416 lr=6.636000e-05 lr_backbone=6.636000e-05 elapsed=01:19:54 avg_step=2.89s recent_step=3.76s eta=38:47:05 +2026-09-14 15:06:51 | INFO | train rank=0 | step=1670/50000 progress=3.3% loss=0.205674 lr=6.676000e-05 lr_backbone=6.676000e-05 elapsed=01:20:24 avg_step=2.89s recent_step=2.98s eta=38:47:01 +2026-09-14 15:07:18 | INFO | train rank=0 | step=1680/50000 progress=3.4% loss=0.188903 lr=6.716000e-05 lr_backbone=6.716000e-05 elapsed=01:20:51 avg_step=2.89s recent_step=2.66s eta=38:45:26 +2026-09-14 15:08:04 | INFO | train rank=0 | step=1690/50000 progress=3.4% loss=0.205763 lr=6.756000e-05 lr_backbone=6.756000e-05 elapsed=01:21:37 avg_step=2.90s recent_step=4.61s eta=38:53:09 +2026-09-14 15:08:40 | INFO | train rank=0 | step=1700/50000 progress=3.4% loss=0.179389 lr=6.796000e-05 lr_backbone=6.796000e-05 elapsed=01:22:13 avg_step=2.90s recent_step=3.59s eta=38:55:57 +2026-09-14 15:09:13 | INFO | train rank=0 | step=1710/50000 progress=3.4% loss=0.241615 lr=6.836000e-05 lr_backbone=6.836000e-05 elapsed=01:22:46 avg_step=2.90s recent_step=3.34s eta=38:57:33 +2026-09-14 15:09:46 | INFO | train rank=0 | step=1720/50000 progress=3.4% loss=0.210736 lr=6.876000e-05 lr_backbone=6.876000e-05 elapsed=01:23:19 avg_step=2.91s recent_step=3.26s eta=38:58:45 +2026-09-14 15:10:31 | INFO | train rank=0 | step=1730/50000 progress=3.5% loss=0.175705 lr=6.916000e-05 lr_backbone=6.916000e-05 elapsed=01:24:03 avg_step=2.92s recent_step=4.47s eta=39:05:33 +2026-09-14 15:11:03 | INFO | train rank=0 | step=1740/50000 progress=3.5% loss=0.234680 lr=6.956000e-05 lr_backbone=6.956000e-05 elapsed=01:24:35 avg_step=2.92s recent_step=3.20s eta=39:06:24 +2026-09-14 15:11:34 | INFO | train rank=0 | step=1750/50000 progress=3.5% loss=0.168123 lr=6.996000e-05 lr_backbone=6.996000e-05 elapsed=01:25:07 avg_step=2.92s recent_step=3.18s eta=39:07:07 +2026-09-14 15:12:07 | INFO | train rank=0 | step=1760/50000 progress=3.5% loss=0.226906 lr=7.036000e-05 lr_backbone=7.036000e-05 elapsed=01:25:40 avg_step=2.92s recent_step=3.26s eta=39:08:10 +2026-09-14 15:12:48 | INFO | train rank=0 | step=1770/50000 progress=3.5% loss=0.187281 lr=7.076000e-05 lr_backbone=7.076000e-05 elapsed=01:26:21 avg_step=2.93s recent_step=4.15s eta=39:13:15 +2026-09-14 15:13:28 | INFO | train rank=0 | step=1780/50000 progress=3.6% loss=0.216388 lr=7.116000e-05 lr_backbone=7.116000e-05 elapsed=01:27:01 avg_step=2.93s recent_step=3.99s eta=39:17:34 +2026-09-14 15:14:00 | INFO | train rank=0 | step=1790/50000 progress=3.6% loss=0.173665 lr=7.156000e-05 lr_backbone=7.156000e-05 elapsed=01:27:33 avg_step=2.93s recent_step=3.14s eta=39:18:01 +2026-09-14 15:14:29 | INFO | train rank=0 | step=1800/50000 progress=3.6% loss=0.189063 lr=7.196000e-05 lr_backbone=7.196000e-05 elapsed=01:28:02 avg_step=2.93s recent_step=2.91s eta=39:17:25 +2026-09-14 15:15:16 | INFO | train rank=0 | step=1810/50000 progress=3.6% loss=0.154938 lr=7.236000e-05 lr_backbone=7.236000e-05 elapsed=01:28:48 avg_step=2.94s recent_step=4.67s eta=39:24:38 +2026-09-14 15:15:45 | INFO | train rank=0 | step=1820/50000 progress=3.6% loss=0.180716 lr=7.276000e-05 lr_backbone=7.276000e-05 elapsed=01:29:18 avg_step=2.94s recent_step=2.91s eta=39:24:00 +2026-09-14 15:16:17 | INFO | train rank=0 | step=1830/50000 progress=3.7% loss=0.215216 lr=7.316000e-05 lr_backbone=7.316000e-05 elapsed=01:29:50 avg_step=2.95s recent_step=3.23s eta=39:24:46 +2026-09-14 15:17:02 | INFO | train rank=0 | step=1840/50000 progress=3.7% loss=0.206129 lr=7.356000e-05 lr_backbone=7.356000e-05 elapsed=01:30:35 avg_step=2.95s recent_step=4.52s eta=39:31:10 +2026-09-14 15:17:47 | INFO | train rank=0 | step=1850/50000 progress=3.7% loss=0.225148 lr=7.396000e-05 lr_backbone=7.396000e-05 elapsed=01:31:20 avg_step=2.96s recent_step=4.49s eta=39:37:19 +2026-09-14 15:18:19 | INFO | train rank=0 | step=1860/50000 progress=3.7% loss=0.201946 lr=7.436000e-05 lr_backbone=7.436000e-05 elapsed=01:31:52 avg_step=2.96s recent_step=3.20s eta=39:37:50 +2026-09-14 15:18:51 | INFO | train rank=0 | step=1870/50000 progress=3.7% loss=0.167820 lr=7.476000e-05 lr_backbone=7.476000e-05 elapsed=01:32:24 avg_step=2.97s recent_step=3.23s eta=39:38:30 +2026-09-14 15:19:36 | INFO | train rank=0 | step=1880/50000 progress=3.8% loss=0.174379 lr=7.516000e-05 lr_backbone=7.516000e-05 elapsed=01:33:09 avg_step=2.97s recent_step=4.43s eta=39:44:16 +2026-09-14 15:20:06 | INFO | train rank=0 | step=1890/50000 progress=3.8% loss=0.208917 lr=7.556000e-05 lr_backbone=7.556000e-05 elapsed=01:33:39 avg_step=2.97s recent_step=3.03s eta=39:44:01 +2026-09-14 15:20:38 | INFO | train rank=0 | step=1900/50000 progress=3.8% loss=0.212809 lr=7.596000e-05 lr_backbone=7.596000e-05 elapsed=01:34:11 avg_step=2.97s recent_step=3.20s eta=39:44:28 +2026-09-14 15:21:23 | INFO | train rank=0 | step=1910/50000 progress=3.8% loss=0.268322 lr=7.636000e-05 lr_backbone=7.636000e-05 elapsed=01:34:55 avg_step=2.98s recent_step=4.45s eta=39:50:11 +2026-09-14 15:21:59 | INFO | train rank=0 | step=1920/50000 progress=3.8% loss=0.183153 lr=7.676000e-05 lr_backbone=7.676000e-05 elapsed=01:35:32 avg_step=2.99s recent_step=3.68s eta=39:52:36 +2026-09-14 15:22:37 | INFO | train rank=0 | step=1930/50000 progress=3.9% loss=0.185425 lr=7.716000e-05 lr_backbone=7.716000e-05 elapsed=01:36:10 avg_step=2.99s recent_step=3.79s eta=39:55:26 +2026-09-14 15:23:09 | INFO | train rank=0 | step=1940/50000 progress=3.9% loss=0.217405 lr=7.756000e-05 lr_backbone=7.756000e-05 elapsed=01:36:42 avg_step=2.99s recent_step=3.19s eta=39:55:45 +2026-09-14 15:23:49 | INFO | train rank=0 | step=1950/50000 progress=3.9% loss=0.194802 lr=7.796000e-05 lr_backbone=7.796000e-05 elapsed=01:37:22 avg_step=3.00s recent_step=3.98s eta=39:59:18 +2026-09-14 15:24:17 | INFO | train rank=0 | step=1960/50000 progress=3.9% loss=0.187924 lr=7.836000e-05 lr_backbone=7.836000e-05 elapsed=01:37:49 avg_step=2.99s recent_step=2.77s eta=39:57:52 +2026-09-14 15:24:53 | INFO | train rank=0 | step=1970/50000 progress=3.9% loss=0.148058 lr=7.876000e-05 lr_backbone=7.876000e-05 elapsed=01:38:26 avg_step=3.00s recent_step=3.69s eta=40:00:10 +2026-09-14 15:25:31 | INFO | train rank=0 | step=1980/50000 progress=4.0% loss=0.185675 lr=7.916000e-05 lr_backbone=7.916000e-05 elapsed=01:39:04 avg_step=3.00s recent_step=3.79s eta=40:02:52 +2026-09-14 15:26:14 | INFO | train rank=0 | step=1990/50000 progress=4.0% loss=0.214090 lr=7.956000e-05 lr_backbone=7.956000e-05 elapsed=01:39:47 avg_step=3.01s recent_step=4.31s eta=40:07:36 +2026-09-14 15:26:57 | INFO | train rank=0 | step=2000/50000 progress=4.0% loss=0.217328 lr=7.996000e-05 lr_backbone=7.996000e-05 elapsed=01:40:29 avg_step=3.01s recent_step=4.23s eta=40:11:59 +2026-09-14 15:27:31 | INFO | train rank=0 | step=2010/50000 progress=4.0% loss=0.159565 lr=8.036000e-05 lr_backbone=8.036000e-05 elapsed=01:41:04 avg_step=3.02s recent_step=3.46s eta=40:13:16 +2026-09-14 15:28:00 | INFO | train rank=0 | step=2020/50000 progress=4.0% loss=0.225973 lr=8.076000e-05 lr_backbone=8.076000e-05 elapsed=01:41:33 avg_step=3.02s recent_step=2.86s eta=40:12:08 +2026-09-14 15:28:51 | INFO | train rank=0 | step=2030/50000 progress=4.1% loss=0.210077 lr=8.116000e-05 lr_backbone=8.116000e-05 elapsed=01:42:23 avg_step=3.03s recent_step=5.06s eta=40:19:41 +2026-09-14 15:29:31 | INFO | train rank=0 | step=2040/50000 progress=4.1% loss=0.192273 lr=8.156000e-05 lr_backbone=8.156000e-05 elapsed=01:43:04 avg_step=3.03s recent_step=4.03s eta=40:23:06 +2026-09-14 15:30:05 | INFO | train rank=0 | step=2050/50000 progress=4.1% loss=0.188228 lr=8.196000e-05 lr_backbone=8.196000e-05 elapsed=01:43:37 avg_step=3.03s recent_step=3.39s eta=40:24:00 +2026-09-14 15:30:37 | INFO | train rank=0 | step=2060/50000 progress=4.1% loss=0.183183 lr=8.236000e-05 lr_backbone=8.236000e-05 elapsed=01:44:10 avg_step=3.03s recent_step=3.28s eta=40:24:27 +2026-09-14 15:31:07 | INFO | train rank=0 | step=2070/50000 progress=4.1% loss=0.191144 lr=8.276000e-05 lr_backbone=8.276000e-05 elapsed=01:44:40 avg_step=3.03s recent_step=2.98s eta=40:23:44 +2026-09-14 15:31:49 | INFO | train rank=0 | step=2080/50000 progress=4.2% loss=0.223883 lr=8.316000e-05 lr_backbone=8.316000e-05 elapsed=01:45:22 avg_step=3.04s recent_step=4.15s eta=40:27:31 +2026-09-14 15:32:30 | INFO | train rank=0 | step=2090/50000 progress=4.2% loss=0.169193 lr=8.356000e-05 lr_backbone=8.356000e-05 elapsed=01:46:03 avg_step=3.04s recent_step=4.13s eta=40:31:11 +2026-09-14 15:33:08 | INFO | train rank=0 | step=2100/50000 progress=4.2% loss=0.175927 lr=8.396000e-05 lr_backbone=8.396000e-05 elapsed=01:46:41 avg_step=3.05s recent_step=3.82s eta=40:33:38 +2026-09-14 15:33:44 | INFO | train rank=0 | step=2110/50000 progress=4.2% loss=0.215951 lr=8.436000e-05 lr_backbone=8.436000e-05 elapsed=01:47:17 avg_step=3.05s recent_step=3.59s eta=40:35:10 +2026-09-14 15:34:21 | INFO | train rank=0 | step=2120/50000 progress=4.2% loss=0.171465 lr=8.476000e-05 lr_backbone=8.476000e-05 elapsed=01:47:54 avg_step=3.05s recent_step=3.70s eta=40:37:06 +2026-09-14 15:35:03 | INFO | train rank=0 | step=2130/50000 progress=4.3% loss=0.217909 lr=8.516000e-05 lr_backbone=8.516000e-05 elapsed=01:48:36 avg_step=3.06s recent_step=4.21s eta=40:40:56 +2026-09-14 15:35:44 | INFO | train rank=0 | step=2140/50000 progress=4.3% loss=0.190514 lr=8.556000e-05 lr_backbone=8.556000e-05 elapsed=01:49:17 avg_step=3.06s recent_step=4.09s eta=40:44:16 +2026-09-14 15:36:18 | INFO | train rank=0 | step=2150/50000 progress=4.3% loss=0.173098 lr=8.596000e-05 lr_backbone=8.596000e-05 elapsed=01:49:50 avg_step=3.07s recent_step=3.33s eta=40:44:45 +2026-09-14 15:36:48 | INFO | train rank=0 | step=2160/50000 progress=4.3% loss=0.227122 lr=8.636000e-05 lr_backbone=8.636000e-05 elapsed=01:50:21 avg_step=3.07s recent_step=3.09s eta=40:44:19 +2026-09-14 15:37:22 | INFO | train rank=0 | step=2170/50000 progress=4.3% loss=0.210285 lr=8.676000e-05 lr_backbone=8.676000e-05 elapsed=01:50:55 avg_step=3.07s recent_step=3.35s eta=40:44:51 +2026-09-14 15:37:52 | INFO | train rank=0 | step=2180/50000 progress=4.4% loss=0.164421 lr=8.716000e-05 lr_backbone=8.716000e-05 elapsed=01:51:25 avg_step=3.07s recent_step=3.05s eta=40:44:16 +2026-09-14 15:38:26 | INFO | train rank=0 | step=2190/50000 progress=4.4% loss=0.181026 lr=8.756000e-05 lr_backbone=8.756000e-05 elapsed=01:51:58 avg_step=3.07s recent_step=3.33s eta=40:44:42 +2026-09-14 15:38:58 | INFO | train rank=0 | step=2200/50000 progress=4.4% loss=0.205076 lr=8.796000e-05 lr_backbone=8.796000e-05 elapsed=01:52:31 avg_step=3.07s recent_step=3.25s eta=40:44:50 +2026-09-14 15:39:38 | INFO | train rank=0 | step=2210/50000 progress=4.4% loss=0.150432 lr=8.836000e-05 lr_backbone=8.836000e-05 elapsed=01:53:10 avg_step=3.07s recent_step=3.95s eta=40:47:30 +2026-09-14 15:40:13 | INFO | train rank=0 | step=2220/50000 progress=4.4% loss=0.165410 lr=8.876000e-05 lr_backbone=8.876000e-05 elapsed=01:53:46 avg_step=3.07s recent_step=3.55s eta=40:48:43 +2026-09-14 15:40:45 | INFO | train rank=0 | step=2230/50000 progress=4.5% loss=0.197579 lr=8.916000e-05 lr_backbone=8.916000e-05 elapsed=01:54:17 avg_step=3.08s recent_step=3.15s eta=40:48:28 +2026-09-14 15:41:24 | INFO | train rank=0 | step=2240/50000 progress=4.5% loss=0.187279 lr=8.956000e-05 lr_backbone=8.956000e-05 elapsed=01:54:57 avg_step=3.08s recent_step=3.95s eta=40:51:03 +2026-09-14 15:42:21 | INFO | train rank=0 | step=2250/50000 progress=4.5% loss=0.193863 lr=8.996000e-05 lr_backbone=8.996000e-05 elapsed=01:55:54 avg_step=3.09s recent_step=5.70s eta=40:59:49 +2026-09-14 15:42:59 | INFO | train rank=0 | step=2260/50000 progress=4.5% loss=0.207448 lr=9.036000e-05 lr_backbone=9.036000e-05 elapsed=01:56:32 avg_step=3.09s recent_step=3.83s eta=41:01:53 +2026-09-14 15:43:28 | INFO | train rank=0 | step=2270/50000 progress=4.5% loss=0.206498 lr=9.076000e-05 lr_backbone=9.076000e-05 elapsed=01:57:01 avg_step=3.09s recent_step=2.88s eta=41:00:37 +2026-09-14 15:43:57 | INFO | train rank=0 | step=2280/50000 progress=4.6% loss=0.186338 lr=9.116000e-05 lr_backbone=9.116000e-05 elapsed=01:57:30 avg_step=3.09s recent_step=2.91s eta=40:59:29 +2026-09-14 15:44:47 | INFO | train rank=0 | step=2290/50000 progress=4.6% loss=0.206388 lr=9.156000e-05 lr_backbone=9.156000e-05 elapsed=01:58:20 avg_step=3.10s recent_step=4.99s eta=41:05:34 +2026-09-14 15:45:26 | INFO | train rank=0 | step=2300/50000 progress=4.6% loss=0.209582 lr=9.196000e-05 lr_backbone=9.196000e-05 elapsed=01:58:59 avg_step=3.10s recent_step=3.90s eta=41:07:48 +2026-09-14 15:45:58 | INFO | train rank=0 | step=2310/50000 progress=4.6% loss=0.209055 lr=9.236000e-05 lr_backbone=9.236000e-05 elapsed=01:59:30 avg_step=3.10s recent_step=3.13s eta=41:07:22 +2026-09-14 15:46:31 | INFO | train rank=0 | step=2320/50000 progress=4.6% loss=0.183329 lr=9.276000e-05 lr_backbone=9.276000e-05 elapsed=02:00:04 avg_step=3.11s recent_step=3.39s eta=41:07:48 +2026-09-14 15:47:29 | INFO | train rank=0 | step=2330/50000 progress=4.7% loss=0.201896 lr=9.316000e-05 lr_backbone=9.316000e-05 elapsed=02:01:02 avg_step=3.12s recent_step=5.73s eta=41:16:15 +2026-09-14 15:48:06 | INFO | train rank=0 | step=2340/50000 progress=4.7% loss=0.182859 lr=9.356000e-05 lr_backbone=9.356000e-05 elapsed=02:01:38 avg_step=3.12s recent_step=3.69s eta=41:17:41 +2026-09-14 15:48:31 | INFO | train rank=0 | step=2350/50000 progress=4.7% loss=0.234885 lr=9.396000e-05 lr_backbone=9.396000e-05 elapsed=02:02:03 avg_step=3.12s recent_step=2.50s eta=41:15:05 +2026-09-14 15:49:10 | INFO | train rank=0 | step=2360/50000 progress=4.7% loss=0.162984 lr=9.436000e-05 lr_backbone=9.436000e-05 elapsed=02:02:43 avg_step=3.12s recent_step=3.91s eta=41:17:14 +2026-09-14 15:50:01 | INFO | train rank=0 | step=2370/50000 progress=4.7% loss=0.221780 lr=9.476000e-05 lr_backbone=9.476000e-05 elapsed=02:03:33 avg_step=3.13s recent_step=5.07s eta=41:23:16 +2026-09-14 15:50:26 | INFO | train rank=0 | step=2380/50000 progress=4.8% loss=0.185142 lr=9.516000e-05 lr_backbone=9.516000e-05 elapsed=02:03:58 avg_step=3.13s recent_step=2.51s eta=41:20:40 +2026-09-14 15:50:53 | INFO | train rank=0 | step=2390/50000 progress=4.8% loss=0.195765 lr=9.556000e-05 lr_backbone=9.556000e-05 elapsed=02:04:26 avg_step=3.12s recent_step=2.72s eta=41:18:48 +2026-09-14 15:51:29 | INFO | train rank=0 | step=2400/50000 progress=4.8% loss=0.178820 lr=9.596000e-05 lr_backbone=9.596000e-05 elapsed=02:05:02 avg_step=3.13s recent_step=3.59s eta=41:19:50 +2026-09-14 15:52:16 | INFO | train rank=0 | step=2410/50000 progress=4.8% loss=0.184098 lr=9.636000e-05 lr_backbone=9.636000e-05 elapsed=02:05:48 avg_step=3.13s recent_step=4.68s eta=41:24:25 +2026-09-14 15:52:48 | INFO | train rank=0 | step=2420/50000 progress=4.8% loss=0.207389 lr=9.676000e-05 lr_backbone=9.676000e-05 elapsed=02:06:21 avg_step=3.13s recent_step=3.28s eta=41:24:22 +2026-09-14 15:53:25 | INFO | train rank=0 | step=2430/50000 progress=4.9% loss=0.184414 lr=9.716000e-05 lr_backbone=9.716000e-05 elapsed=02:06:58 avg_step=3.14s recent_step=3.71s eta=41:25:43 +2026-09-14 15:54:02 | INFO | train rank=0 | step=2440/50000 progress=4.9% loss=0.148037 lr=9.756000e-05 lr_backbone=9.756000e-05 elapsed=02:07:35 avg_step=3.14s recent_step=3.67s eta=41:26:56 +2026-09-14 15:54:48 | INFO | train rank=0 | step=2450/50000 progress=4.9% loss=0.164472 lr=9.796000e-05 lr_backbone=9.796000e-05 elapsed=02:08:20 avg_step=3.14s recent_step=4.55s eta=41:31:00 +2026-09-14 15:55:19 | INFO | train rank=0 | step=2460/50000 progress=4.9% loss=0.176404 lr=9.836000e-05 lr_backbone=9.836000e-05 elapsed=02:08:52 avg_step=3.14s recent_step=3.11s eta=41:30:22 +2026-09-14 15:55:58 | INFO | train rank=0 | step=2470/50000 progress=4.9% loss=0.195001 lr=9.876000e-05 lr_backbone=9.876000e-05 elapsed=02:09:31 avg_step=3.15s recent_step=3.92s eta=41:32:20 +2026-09-14 15:56:33 | INFO | train rank=0 | step=2480/50000 progress=5.0% loss=0.172931 lr=9.916000e-05 lr_backbone=9.916000e-05 elapsed=02:10:06 avg_step=3.15s recent_step=3.50s eta=41:32:56 +2026-09-14 15:57:08 | INFO | train rank=0 | step=2490/50000 progress=5.0% loss=0.157087 lr=9.956000e-05 lr_backbone=9.956000e-05 elapsed=02:10:41 avg_step=3.15s recent_step=3.51s eta=41:33:34 +2026-09-14 15:57:42 | INFO | train rank=0 | step=2500/50000 progress=5.0% loss=0.147246 lr=9.996000e-05 lr_backbone=9.996000e-05 elapsed=02:11:15 avg_step=3.15s recent_step=3.38s eta=41:33:46 +2026-09-14 15:58:06 | INFO | train rank=0 | step=2510/50000 progress=5.0% loss=0.204980 lr=9.999999e-05 lr_backbone=9.999999e-05 elapsed=02:11:39 avg_step=3.15s recent_step=2.40s eta=41:30:53 +2026-09-14 15:58:36 | INFO | train rank=0 | step=2520/50000 progress=5.0% loss=0.150093 lr=9.999996e-05 lr_backbone=9.999996e-05 elapsed=02:12:09 avg_step=3.15s recent_step=3.02s eta=41:29:57 +2026-09-14 15:59:14 | INFO | train rank=0 | step=2530/50000 progress=5.1% loss=0.207979 lr=9.999990e-05 lr_backbone=9.999990e-05 elapsed=02:12:46 avg_step=3.15s recent_step=3.77s eta=41:31:22 +2026-09-14 15:59:38 | INFO | train rank=0 | step=2540/50000 progress=5.1% loss=0.192124 lr=9.999983e-05 lr_backbone=9.999983e-05 elapsed=02:13:11 avg_step=3.15s recent_step=2.41s eta=41:28:33 +2026-09-14 16:00:11 | INFO | train rank=0 | step=2550/50000 progress=5.1% loss=0.175227 lr=9.999973e-05 lr_backbone=9.999973e-05 elapsed=02:13:44 avg_step=3.15s recent_step=3.36s eta=41:28:41 +2026-09-14 16:00:37 | INFO | train rank=0 | step=2560/50000 progress=5.1% loss=0.195697 lr=9.999961e-05 lr_backbone=9.999961e-05 elapsed=02:14:09 avg_step=3.14s recent_step=2.53s eta=41:26:14 +2026-09-14 16:01:28 | INFO | train rank=0 | step=2570/50000 progress=5.1% loss=0.163905 lr=9.999946e-05 lr_backbone=9.999946e-05 elapsed=02:15:01 avg_step=3.15s recent_step=5.17s eta=41:31:56 +2026-09-14 16:02:03 | INFO | train rank=0 | step=2580/50000 progress=5.2% loss=0.197191 lr=9.999930e-05 lr_backbone=9.999930e-05 elapsed=02:15:35 avg_step=3.15s recent_step=3.44s eta=41:32:17 +2026-09-14 16:02:40 | INFO | train rank=0 | step=2590/50000 progress=5.2% loss=0.166576 lr=9.999911e-05 lr_backbone=9.999911e-05 elapsed=02:16:13 avg_step=3.16s recent_step=3.71s eta=41:33:29 +2026-09-14 16:03:09 | INFO | train rank=0 | step=2600/50000 progress=5.2% loss=0.170598 lr=9.999891e-05 lr_backbone=9.999891e-05 elapsed=02:16:42 avg_step=3.15s recent_step=2.97s eta=41:32:22 +2026-09-14 16:04:00 | INFO | train rank=0 | step=2610/50000 progress=5.2% loss=0.131465 lr=9.999868e-05 lr_backbone=9.999868e-05 elapsed=02:17:33 avg_step=3.16s recent_step=5.08s eta=41:37:40 +2026-09-14 16:04:38 | INFO | train rank=0 | step=2620/50000 progress=5.2% loss=0.179206 lr=9.999843e-05 lr_backbone=9.999843e-05 elapsed=02:18:11 avg_step=3.16s recent_step=3.77s eta=41:38:58 +2026-09-14 16:05:11 | INFO | train rank=0 | step=2630/50000 progress=5.3% loss=0.180645 lr=9.999815e-05 lr_backbone=9.999815e-05 elapsed=02:18:44 avg_step=3.17s recent_step=3.34s eta=41:38:58 +2026-09-14 16:05:47 | INFO | train rank=0 | step=2640/50000 progress=5.3% loss=0.181013 lr=9.999786e-05 lr_backbone=9.999786e-05 elapsed=02:19:20 avg_step=3.17s recent_step=3.58s eta=41:39:41 +2026-09-14 16:06:38 | INFO | train rank=0 | step=2650/50000 progress=5.3% loss=0.157706 lr=9.999754e-05 lr_backbone=9.999754e-05 elapsed=02:20:10 avg_step=3.17s recent_step=5.05s eta=41:44:45 +2026-09-14 16:07:12 | INFO | train rank=0 | step=2660/50000 progress=5.3% loss=0.175861 lr=9.999720e-05 lr_backbone=9.999720e-05 elapsed=02:20:45 avg_step=3.17s recent_step=3.44s eta=41:45:00 +2026-09-14 16:07:46 | INFO | train rank=0 | step=2670/50000 progress=5.3% loss=0.156901 lr=9.999684e-05 lr_backbone=9.999684e-05 elapsed=02:21:19 avg_step=3.18s recent_step=3.39s eta=41:45:06 +2026-09-14 16:08:10 | INFO | train rank=0 | step=2680/50000 progress=5.4% loss=0.178048 lr=9.999646e-05 lr_backbone=9.999646e-05 elapsed=02:21:43 avg_step=3.17s recent_step=2.41s eta=41:42:20 +2026-09-14 16:08:58 | INFO | train rank=0 | step=2690/50000 progress=5.4% loss=0.215350 lr=9.999605e-05 lr_backbone=9.999605e-05 elapsed=02:22:31 avg_step=3.18s recent_step=4.82s eta=41:46:37 +2026-09-14 16:09:26 | INFO | train rank=0 | step=2700/50000 progress=5.4% loss=0.156707 lr=9.999563e-05 lr_backbone=9.999563e-05 elapsed=02:22:59 avg_step=3.18s recent_step=2.75s eta=41:44:51 +2026-09-14 16:09:56 | INFO | train rank=0 | step=2710/50000 progress=5.4% loss=0.193579 lr=9.999518e-05 lr_backbone=9.999518e-05 elapsed=02:23:29 avg_step=3.18s recent_step=3.07s eta=41:44:01 +2026-09-14 16:10:36 | INFO | train rank=0 | step=2720/50000 progress=5.4% loss=0.169381 lr=9.999471e-05 lr_backbone=9.999471e-05 elapsed=02:24:08 avg_step=3.18s recent_step=3.92s eta=41:45:38 +2026-09-14 16:11:18 | INFO | train rank=0 | step=2730/50000 progress=5.5% loss=0.164640 lr=9.999422e-05 lr_backbone=9.999422e-05 elapsed=02:24:51 avg_step=3.18s recent_step=4.26s eta=41:48:13 +2026-09-14 16:11:48 | INFO | train rank=0 | step=2740/50000 progress=5.5% loss=0.198879 lr=9.999370e-05 lr_backbone=9.999370e-05 elapsed=02:25:21 avg_step=3.18s recent_step=3.01s eta=41:47:12 +2026-09-14 16:12:24 | INFO | train rank=0 | step=2750/50000 progress=5.5% loss=0.188543 lr=9.999317e-05 lr_backbone=9.999317e-05 elapsed=02:25:57 avg_step=3.18s recent_step=3.59s eta=41:47:49 +2026-09-14 16:12:54 | INFO | train rank=0 | step=2760/50000 progress=5.5% loss=0.154667 lr=9.999261e-05 lr_backbone=9.999261e-05 elapsed=02:26:27 avg_step=3.18s recent_step=2.99s eta=41:46:44 +2026-09-14 16:13:38 | INFO | train rank=0 | step=2770/50000 progress=5.5% loss=0.206217 lr=9.999203e-05 lr_backbone=9.999203e-05 elapsed=02:27:11 avg_step=3.19s recent_step=4.43s eta=41:49:46 +2026-09-14 16:14:13 | INFO | train rank=0 | step=2780/50000 progress=5.6% loss=0.176173 lr=9.999143e-05 lr_backbone=9.999143e-05 elapsed=02:27:46 avg_step=3.19s recent_step=3.48s eta=41:50:03 +2026-09-14 16:15:01 | INFO | train rank=0 | step=2790/50000 progress=5.6% loss=0.156680 lr=9.999080e-05 lr_backbone=9.999080e-05 elapsed=02:28:34 avg_step=3.20s recent_step=4.78s eta=41:54:00 +2026-09-14 16:15:31 | INFO | train rank=0 | step=2800/50000 progress=5.6% loss=0.149898 lr=9.999016e-05 lr_backbone=9.999016e-05 elapsed=02:29:03 avg_step=3.19s recent_step=2.96s eta=41:52:48 +2026-09-14 16:15:59 | INFO | train rank=0 | step=2810/50000 progress=5.6% loss=0.186801 lr=9.998949e-05 lr_backbone=9.998949e-05 elapsed=02:29:32 avg_step=3.19s recent_step=2.85s eta=41:51:19 +2026-09-14 16:16:34 | INFO | train rank=0 | step=2820/50000 progress=5.6% loss=0.157132 lr=9.998880e-05 lr_backbone=9.998880e-05 elapsed=02:30:07 avg_step=3.19s recent_step=3.51s eta=41:51:40 +2026-09-14 16:17:21 | INFO | train rank=0 | step=2830/50000 progress=5.7% loss=0.156063 lr=9.998809e-05 lr_backbone=9.998809e-05 elapsed=02:30:54 avg_step=3.20s recent_step=4.67s eta=41:55:13 +2026-09-14 16:17:54 | INFO | train rank=0 | step=2840/50000 progress=5.7% loss=0.169302 lr=9.998736e-05 lr_backbone=9.998736e-05 elapsed=02:31:27 avg_step=3.20s recent_step=3.34s eta=41:55:05 +2026-09-14 16:18:31 | INFO | train rank=0 | step=2850/50000 progress=5.7% loss=0.184927 lr=9.998660e-05 lr_backbone=9.998660e-05 elapsed=02:32:04 avg_step=3.20s recent_step=3.65s eta=41:55:47 +2026-09-14 16:19:02 | INFO | train rank=0 | step=2860/50000 progress=5.7% loss=0.151568 lr=9.998583e-05 lr_backbone=9.998583e-05 elapsed=02:32:35 avg_step=3.20s recent_step=3.10s eta=41:54:58 +2026-09-14 16:19:52 | INFO | train rank=0 | step=2870/50000 progress=5.7% loss=0.169035 lr=9.998503e-05 lr_backbone=9.998503e-05 elapsed=02:33:25 avg_step=3.21s recent_step=5.05s eta=41:59:30 +2026-09-14 16:20:27 | INFO | train rank=0 | step=2880/50000 progress=5.8% loss=0.210686 lr=9.998421e-05 lr_backbone=9.998421e-05 elapsed=02:34:00 avg_step=3.21s recent_step=3.51s eta=41:59:48 +2026-09-14 16:20:55 | INFO | train rank=0 | step=2890/50000 progress=5.8% loss=0.181953 lr=9.998337e-05 lr_backbone=9.998337e-05 elapsed=02:34:28 avg_step=3.21s recent_step=2.76s eta=41:58:02 +2026-09-14 16:21:25 | INFO | train rank=0 | step=2900/50000 progress=5.8% loss=0.162130 lr=9.998250e-05 lr_backbone=9.998250e-05 elapsed=02:34:58 avg_step=3.21s recent_step=2.98s eta=41:56:54 +2026-09-14 16:22:19 | INFO | train rank=0 | step=2910/50000 progress=5.8% loss=0.165440 lr=9.998162e-05 lr_backbone=9.998162e-05 elapsed=02:35:52 avg_step=3.21s recent_step=5.42s eta=42:02:21 +2026-09-14 16:22:48 | INFO | train rank=0 | step=2920/50000 progress=5.8% loss=0.165981 lr=9.998071e-05 lr_backbone=9.998071e-05 elapsed=02:36:21 avg_step=3.21s recent_step=2.93s eta=42:01:02 +2026-09-14 16:23:24 | INFO | train rank=0 | step=2930/50000 progress=5.9% loss=0.129956 lr=9.997978e-05 lr_backbone=9.997978e-05 elapsed=02:36:57 avg_step=3.21s recent_step=3.60s eta=42:01:33 +2026-09-14 16:23:52 | INFO | train rank=0 | step=2940/50000 progress=5.9% loss=0.179281 lr=9.997883e-05 lr_backbone=9.997883e-05 elapsed=02:37:25 avg_step=3.21s recent_step=2.79s eta=41:59:52 +2026-09-14 16:24:42 | INFO | train rank=0 | step=2950/50000 progress=5.9% loss=0.156419 lr=9.997786e-05 lr_backbone=9.997786e-05 elapsed=02:38:15 avg_step=3.22s recent_step=5.02s eta=42:04:08 +2026-09-14 16:25:09 | INFO | train rank=0 | step=2960/50000 progress=5.9% loss=0.146595 lr=9.997686e-05 lr_backbone=9.997686e-05 elapsed=02:38:42 avg_step=3.22s recent_step=2.66s eta=42:02:07 +2026-09-14 16:25:37 | INFO | train rank=0 | step=2970/50000 progress=5.9% loss=0.184392 lr=9.997584e-05 lr_backbone=9.997584e-05 elapsed=02:39:09 avg_step=3.22s recent_step=2.76s eta=42:00:23 +2026-09-14 16:26:12 | INFO | train rank=0 | step=2980/50000 progress=6.0% loss=0.170969 lr=9.997481e-05 lr_backbone=9.997481e-05 elapsed=02:39:45 avg_step=3.22s recent_step=3.58s eta=42:00:48 +2026-09-14 16:27:11 | INFO | train rank=0 | step=2990/50000 progress=6.0% loss=0.187241 lr=9.997375e-05 lr_backbone=9.997375e-05 elapsed=02:40:44 avg_step=3.23s recent_step=5.83s eta=42:07:07 +2026-09-14 16:27:45 | INFO | train rank=0 | step=3000/50000 progress=6.0% loss=0.191884 lr=9.997266e-05 lr_backbone=9.997266e-05 elapsed=02:41:18 avg_step=3.23s recent_step=3.46s eta=42:07:12 +2026-09-14 16:28:15 | INFO | train rank=0 | step=3010/50000 progress=6.0% loss=0.176003 lr=9.997156e-05 lr_backbone=9.997156e-05 elapsed=02:41:48 avg_step=3.23s recent_step=2.98s eta=42:06:02 +2026-09-14 16:28:50 | INFO | train rank=0 | step=3020/50000 progress=6.0% loss=0.185769 lr=9.997043e-05 lr_backbone=9.997043e-05 elapsed=02:42:23 avg_step=3.23s recent_step=3.53s eta=42:06:16 +2026-09-14 16:29:43 | INFO | train rank=0 | step=3030/50000 progress=6.1% loss=0.137707 lr=9.996928e-05 lr_backbone=9.996928e-05 elapsed=02:43:15 avg_step=3.23s recent_step=5.21s eta=42:10:51 +2026-09-14 16:30:07 | INFO | train rank=0 | step=3040/50000 progress=6.1% loss=0.153364 lr=9.996811e-05 lr_backbone=9.996811e-05 elapsed=02:43:40 avg_step=3.23s recent_step=2.42s eta=42:08:13 +2026-09-14 16:30:47 | INFO | train rank=0 | step=3050/50000 progress=6.1% loss=0.144232 lr=9.996692e-05 lr_backbone=9.996692e-05 elapsed=02:44:20 avg_step=3.23s recent_step=4.06s eta=42:09:48 +2026-09-14 16:31:22 | INFO | train rank=0 | step=3060/50000 progress=6.1% loss=0.197993 lr=9.996571e-05 lr_backbone=9.996571e-05 elapsed=02:44:55 avg_step=3.23s recent_step=3.45s eta=42:09:49 +2026-09-14 16:32:10 | INFO | train rank=0 | step=3070/50000 progress=6.1% loss=0.175215 lr=9.996447e-05 lr_backbone=9.996447e-05 elapsed=02:45:43 avg_step=3.24s recent_step=4.80s eta=42:13:16 +2026-09-14 16:32:50 | INFO | train rank=0 | step=3080/50000 progress=6.2% loss=0.156169 lr=9.996322e-05 lr_backbone=9.996322e-05 elapsed=02:46:23 avg_step=3.24s recent_step=4.06s eta=42:14:48 +2026-09-14 16:33:21 | INFO | train rank=0 | step=3090/50000 progress=6.2% loss=0.190266 lr=9.996194e-05 lr_backbone=9.996194e-05 elapsed=02:46:54 avg_step=3.24s recent_step=3.06s eta=42:13:48 +2026-09-14 16:33:57 | INFO | train rank=0 | step=3100/50000 progress=6.2% loss=0.141790 lr=9.996064e-05 lr_backbone=9.996064e-05 elapsed=02:47:29 avg_step=3.24s recent_step=3.56s eta=42:14:04 +2026-09-14 16:34:34 | INFO | train rank=0 | step=3110/50000 progress=6.2% loss=0.144762 lr=9.995931e-05 lr_backbone=9.995931e-05 elapsed=02:48:07 avg_step=3.24s recent_step=3.73s eta=42:14:45 +2026-09-14 16:35:06 | INFO | train rank=0 | step=3120/50000 progress=6.2% loss=0.171571 lr=9.995797e-05 lr_backbone=9.995797e-05 elapsed=02:48:39 avg_step=3.24s recent_step=3.22s eta=42:14:08 +2026-09-14 16:35:39 | INFO | train rank=0 | step=3130/50000 progress=6.3% loss=0.174771 lr=9.995660e-05 lr_backbone=9.995660e-05 elapsed=02:49:12 avg_step=3.24s recent_step=3.32s eta=42:13:47 +2026-09-14 16:36:12 | INFO | train rank=0 | step=3140/50000 progress=6.3% loss=0.134012 lr=9.995521e-05 lr_backbone=9.995521e-05 elapsed=02:49:45 avg_step=3.24s recent_step=3.33s eta=42:13:27 +2026-09-14 16:36:57 | INFO | train rank=0 | step=3150/50000 progress=6.3% loss=0.142534 lr=9.995380e-05 lr_backbone=9.995380e-05 elapsed=02:50:29 avg_step=3.25s recent_step=4.41s eta=42:15:49 +2026-09-14 16:37:32 | INFO | train rank=0 | step=3160/50000 progress=6.3% loss=0.206569 lr=9.995237e-05 lr_backbone=9.995237e-05 elapsed=02:51:05 avg_step=3.25s recent_step=3.52s eta=42:15:57 +2026-09-14 16:37:59 | INFO | train rank=0 | step=3170/50000 progress=6.3% loss=0.193985 lr=9.995092e-05 lr_backbone=9.995092e-05 elapsed=02:51:31 avg_step=3.25s recent_step=2.67s eta=42:14:00 +2026-09-14 16:38:41 | INFO | train rank=0 | step=3180/50000 progress=6.4% loss=0.145247 lr=9.994944e-05 lr_backbone=9.994944e-05 elapsed=02:52:13 avg_step=3.25s recent_step=4.21s eta=42:15:50 +2026-09-14 16:39:18 | INFO | train rank=0 | step=3190/50000 progress=6.4% loss=0.162886 lr=9.994794e-05 lr_backbone=9.994794e-05 elapsed=02:52:50 avg_step=3.25s recent_step=3.68s eta=42:16:21 +2026-09-14 16:39:49 | INFO | train rank=0 | step=3200/50000 progress=6.4% loss=0.185553 lr=9.994642e-05 lr_backbone=9.994642e-05 elapsed=02:53:22 avg_step=3.25s recent_step=3.17s eta=42:15:36 +2026-09-14 16:40:26 | INFO | train rank=0 | step=3210/50000 progress=6.4% loss=0.149421 lr=9.994488e-05 lr_backbone=9.994488e-05 elapsed=02:53:59 avg_step=3.25s recent_step=3.66s eta=42:16:03 +2026-09-14 16:41:06 | INFO | train rank=0 | step=3220/50000 progress=6.4% loss=0.130615 lr=9.994332e-05 lr_backbone=9.994332e-05 elapsed=02:54:39 avg_step=3.25s recent_step=4.06s eta=42:17:27 +2026-09-14 16:41:46 | INFO | train rank=0 | step=3230/50000 progress=6.5% loss=0.172909 lr=9.994173e-05 lr_backbone=9.994173e-05 elapsed=02:55:19 avg_step=3.26s recent_step=3.98s eta=42:18:39 +2026-09-14 16:42:16 | INFO | train rank=0 | step=3240/50000 progress=6.5% loss=0.135078 lr=9.994013e-05 lr_backbone=9.994013e-05 elapsed=02:55:49 avg_step=3.26s recent_step=2.98s eta=42:17:27 +2026-09-14 16:42:53 | INFO | train rank=0 | step=3250/50000 progress=6.5% loss=0.157361 lr=9.993850e-05 lr_backbone=9.993850e-05 elapsed=02:56:26 avg_step=3.26s recent_step=3.74s eta=42:18:04 +2026-09-14 16:43:40 | INFO | train rank=0 | step=3260/50000 progress=6.5% loss=0.165024 lr=9.993685e-05 lr_backbone=9.993685e-05 elapsed=02:57:13 avg_step=3.26s recent_step=4.64s eta=42:20:50 +2026-09-14 16:44:12 | INFO | train rank=0 | step=3270/50000 progress=6.5% loss=0.124456 lr=9.993518e-05 lr_backbone=9.993518e-05 elapsed=02:57:45 avg_step=3.26s recent_step=3.21s eta=42:20:10 +2026-09-14 16:44:41 | INFO | train rank=0 | step=3280/50000 progress=6.6% loss=0.153597 lr=9.993348e-05 lr_backbone=9.993348e-05 elapsed=02:58:13 avg_step=3.26s recent_step=2.88s eta=42:18:43 +2026-09-14 16:45:19 | INFO | train rank=0 | step=3290/50000 progress=6.6% loss=0.127248 lr=9.993176e-05 lr_backbone=9.993176e-05 elapsed=02:58:52 avg_step=3.26s recent_step=3.85s eta=42:19:34 +2026-09-14 16:46:05 | INFO | train rank=0 | step=3300/50000 progress=6.6% loss=0.151669 lr=9.993003e-05 lr_backbone=9.993003e-05 elapsed=02:59:38 avg_step=3.27s recent_step=4.61s eta=42:22:13 +2026-09-14 16:46:52 | INFO | train rank=0 | step=3310/50000 progress=6.6% loss=0.147920 lr=9.992827e-05 lr_backbone=9.992827e-05 elapsed=03:00:24 avg_step=3.27s recent_step=4.63s eta=42:24:52 +2026-09-14 16:47:25 | INFO | train rank=0 | step=3320/50000 progress=6.6% loss=0.178272 lr=9.992649e-05 lr_backbone=9.992649e-05 elapsed=03:00:57 avg_step=3.27s recent_step=3.31s eta=42:24:24 +2026-09-14 16:48:00 | INFO | train rank=0 | step=3330/50000 progress=6.7% loss=0.158465 lr=9.992468e-05 lr_backbone=9.992468e-05 elapsed=03:01:32 avg_step=3.27s recent_step=3.50s eta=42:24:23 +2026-09-14 16:48:37 | INFO | train rank=0 | step=3340/50000 progress=6.7% loss=0.149665 lr=9.992286e-05 lr_backbone=9.992286e-05 elapsed=03:02:10 avg_step=3.27s recent_step=3.78s eta=42:25:02 +2026-09-14 16:49:22 | INFO | train rank=0 | step=3350/50000 progress=6.7% loss=0.126910 lr=9.992101e-05 lr_backbone=9.992101e-05 elapsed=03:02:54 avg_step=3.28s recent_step=4.42s eta=42:27:08 +2026-09-14 16:49:55 | INFO | train rank=0 | step=3360/50000 progress=6.7% loss=0.149931 lr=9.991914e-05 lr_backbone=9.991914e-05 elapsed=03:03:28 avg_step=3.28s recent_step=3.35s eta=42:26:46 +2026-09-14 16:50:25 | INFO | train rank=0 | step=3370/50000 progress=6.7% loss=0.154214 lr=9.991725e-05 lr_backbone=9.991725e-05 elapsed=03:03:57 avg_step=3.28s recent_step=2.94s eta=42:25:27 +2026-09-14 16:50:57 | INFO | train rank=0 | step=3380/50000 progress=6.8% loss=0.159309 lr=9.991534e-05 lr_backbone=9.991534e-05 elapsed=03:04:30 avg_step=3.28s recent_step=3.25s eta=42:24:51 +2026-09-14 16:51:45 | INFO | train rank=0 | step=3390/50000 progress=6.8% loss=0.126541 lr=9.991340e-05 lr_backbone=9.991340e-05 elapsed=03:05:18 avg_step=3.28s recent_step=4.81s eta=42:27:50 +2026-09-14 16:52:16 | INFO | train rank=0 | step=3400/50000 progress=6.8% loss=0.154806 lr=9.991145e-05 lr_backbone=9.991145e-05 elapsed=03:05:49 avg_step=3.28s recent_step=3.07s eta=42:26:48 +2026-09-14 16:52:45 | INFO | train rank=0 | step=3410/50000 progress=6.8% loss=0.154891 lr=9.990947e-05 lr_backbone=9.990947e-05 elapsed=03:06:18 avg_step=3.28s recent_step=2.90s eta=42:25:24 +2026-09-14 16:53:14 | INFO | train rank=0 | step=3420/50000 progress=6.8% loss=0.119007 lr=9.990747e-05 lr_backbone=9.990747e-05 elapsed=03:06:47 avg_step=3.28s recent_step=2.89s eta=42:23:58 +2026-09-14 16:53:51 | INFO | train rank=0 | step=3430/50000 progress=6.9% loss=0.144981 lr=9.990545e-05 lr_backbone=9.990545e-05 elapsed=03:07:23 avg_step=3.28s recent_step=3.69s eta=42:24:21 +2026-09-14 16:54:28 | INFO | train rank=0 | step=3440/50000 progress=6.9% loss=0.175367 lr=9.990340e-05 lr_backbone=9.990340e-05 elapsed=03:08:01 avg_step=3.28s recent_step=3.76s eta=42:24:54 +2026-09-14 16:54:58 | INFO | train rank=0 | step=3450/50000 progress=6.9% loss=0.136398 lr=9.990134e-05 lr_backbone=9.990134e-05 elapsed=03:08:30 avg_step=3.28s recent_step=2.94s eta=42:23:35 +2026-09-14 16:55:49 | INFO | train rank=0 | step=3460/50000 progress=6.9% loss=0.162746 lr=9.989925e-05 lr_backbone=9.989925e-05 elapsed=03:09:21 avg_step=3.28s recent_step=5.10s eta=42:27:07 +2026-09-14 16:56:22 | INFO | train rank=0 | step=3470/50000 progress=6.9% loss=0.147795 lr=9.989714e-05 lr_backbone=9.989714e-05 elapsed=03:09:55 avg_step=3.28s recent_step=3.32s eta=42:26:38 +2026-09-14 16:56:55 | INFO | train rank=0 | step=3480/50000 progress=7.0% loss=0.186115 lr=9.989501e-05 lr_backbone=9.989501e-05 elapsed=03:10:27 avg_step=3.28s recent_step=3.28s eta=42:26:05 +2026-09-14 16:57:24 | INFO | train rank=0 | step=3490/50000 progress=7.0% loss=0.154655 lr=9.989286e-05 lr_backbone=9.989286e-05 elapsed=03:10:57 avg_step=3.28s recent_step=2.91s eta=42:24:43 +2026-09-14 16:58:18 | INFO | train rank=0 | step=3500/50000 progress=7.0% loss=0.136255 lr=9.989068e-05 lr_backbone=9.989068e-05 elapsed=03:11:51 avg_step=3.29s recent_step=5.46s eta=42:28:59 +2026-09-14 16:58:51 | INFO | train rank=0 | step=3510/50000 progress=7.0% loss=0.126314 lr=9.988848e-05 lr_backbone=9.988848e-05 elapsed=03:12:24 avg_step=3.29s recent_step=3.28s eta=42:28:25 +2026-09-14 16:59:16 | INFO | train rank=0 | step=3520/50000 progress=7.0% loss=0.125137 lr=9.988627e-05 lr_backbone=9.988627e-05 elapsed=03:12:49 avg_step=3.29s recent_step=2.48s eta=42:26:05 +2026-09-14 16:59:50 | INFO | train rank=0 | step=3530/50000 progress=7.1% loss=0.158515 lr=9.988403e-05 lr_backbone=9.988403e-05 elapsed=03:13:23 avg_step=3.29s recent_step=3.42s eta=42:25:49 +2026-09-14 17:00:45 | INFO | train rank=0 | step=3540/50000 progress=7.1% loss=0.129946 lr=9.988176e-05 lr_backbone=9.988176e-05 elapsed=03:14:18 avg_step=3.29s recent_step=5.54s eta=42:30:13 +2026-09-14 17:01:12 | INFO | train rank=0 | step=3550/50000 progress=7.1% loss=0.145898 lr=9.987948e-05 lr_backbone=9.987948e-05 elapsed=03:14:45 avg_step=3.29s recent_step=2.70s eta=42:28:22 +2026-09-14 17:01:45 | INFO | train rank=0 | step=3560/50000 progress=7.1% loss=0.149247 lr=9.987718e-05 lr_backbone=9.987718e-05 elapsed=03:15:17 avg_step=3.29s recent_step=3.21s eta=42:27:38 +2026-09-14 17:02:13 | INFO | train rank=0 | step=3570/50000 progress=7.1% loss=0.186359 lr=9.987485e-05 lr_backbone=9.987485e-05 elapsed=03:15:46 avg_step=3.29s recent_step=2.87s eta=42:26:11 +2026-09-14 17:03:03 | INFO | train rank=0 | step=3580/50000 progress=7.2% loss=0.146829 lr=9.987250e-05 lr_backbone=9.987250e-05 elapsed=03:16:36 avg_step=3.30s recent_step=5.01s eta=42:29:22 +2026-09-14 17:03:36 | INFO | train rank=0 | step=3590/50000 progress=7.2% loss=0.173560 lr=9.987013e-05 lr_backbone=9.987013e-05 elapsed=03:17:09 avg_step=3.30s recent_step=3.27s eta=42:28:45 +2026-09-14 17:04:09 | INFO | train rank=0 | step=3600/50000 progress=7.2% loss=0.134745 lr=9.986773e-05 lr_backbone=9.986773e-05 elapsed=03:17:42 avg_step=3.30s recent_step=3.32s eta=42:28:15 +2026-09-14 17:04:44 | INFO | train rank=0 | step=3610/50000 progress=7.2% loss=0.123393 lr=9.986532e-05 lr_backbone=9.986532e-05 elapsed=03:18:17 avg_step=3.30s recent_step=3.49s eta=42:28:06 +2026-09-14 17:05:35 | INFO | train rank=0 | step=3620/50000 progress=7.2% loss=0.136033 lr=9.986288e-05 lr_backbone=9.986288e-05 elapsed=03:19:08 avg_step=3.30s recent_step=5.07s eta=42:31:20 +2026-09-14 17:06:10 | INFO | train rank=0 | step=3630/50000 progress=7.3% loss=0.125557 lr=9.986043e-05 lr_backbone=9.986043e-05 elapsed=03:19:43 avg_step=3.30s recent_step=3.53s eta=42:31:16 +2026-09-14 17:06:48 | INFO | train rank=0 | step=3640/50000 progress=7.3% loss=0.120454 lr=9.985795e-05 lr_backbone=9.985795e-05 elapsed=03:20:21 avg_step=3.30s recent_step=3.79s eta=42:31:46 +2026-09-14 17:07:14 | INFO | train rank=0 | step=3650/50000 progress=7.3% loss=0.134933 lr=9.985544e-05 lr_backbone=9.985544e-05 elapsed=03:20:47 avg_step=3.30s recent_step=2.62s eta=42:29:47 +2026-09-14 17:08:01 | INFO | train rank=0 | step=3660/50000 progress=7.3% loss=0.128138 lr=9.985292e-05 lr_backbone=9.985292e-05 elapsed=03:21:33 avg_step=3.30s recent_step=4.64s eta=42:32:03 +2026-09-14 17:08:31 | INFO | train rank=0 | step=3670/50000 progress=7.3% loss=0.145790 lr=9.985037e-05 lr_backbone=9.985037e-05 elapsed=03:22:04 avg_step=3.30s recent_step=3.05s eta=42:30:57 +2026-09-14 17:08:55 | INFO | train rank=0 | step=3680/50000 progress=7.4% loss=0.166308 lr=9.984781e-05 lr_backbone=9.984781e-05 elapsed=03:22:28 avg_step=3.30s recent_step=2.37s eta=42:28:27 +2026-09-14 17:09:31 | INFO | train rank=0 | step=3690/50000 progress=7.4% loss=0.130251 lr=9.984522e-05 lr_backbone=9.984522e-05 elapsed=03:23:04 avg_step=3.30s recent_step=3.64s eta=42:28:36 +2026-09-14 17:10:27 | INFO | train rank=0 | step=3700/50000 progress=7.4% loss=0.151399 lr=9.984261e-05 lr_backbone=9.984261e-05 elapsed=03:24:00 avg_step=3.31s recent_step=5.59s eta=42:32:49 +2026-09-14 17:11:03 | INFO | train rank=0 | step=3710/50000 progress=7.4% loss=0.162911 lr=9.983997e-05 lr_backbone=9.983997e-05 elapsed=03:24:36 avg_step=3.31s recent_step=3.58s eta=42:32:50 +2026-09-14 17:11:45 | INFO | train rank=0 | step=3720/50000 progress=7.4% loss=0.168574 lr=9.983732e-05 lr_backbone=9.983732e-05 elapsed=03:25:17 avg_step=3.31s recent_step=4.17s eta=42:34:04 +2026-09-14 17:12:14 | INFO | train rank=0 | step=3730/50000 progress=7.5% loss=0.140902 lr=9.983464e-05 lr_backbone=9.983464e-05 elapsed=03:25:47 avg_step=3.31s recent_step=2.96s eta=42:32:48 +2026-09-14 17:13:08 | INFO | train rank=0 | step=3740/50000 progress=7.5% loss=0.144755 lr=9.983194e-05 lr_backbone=9.983194e-05 elapsed=03:26:40 avg_step=3.32s recent_step=5.34s eta=42:36:25 +2026-09-14 17:13:33 | INFO | train rank=0 | step=3750/50000 progress=7.5% loss=0.120432 lr=9.982922e-05 lr_backbone=9.982922e-05 elapsed=03:27:06 avg_step=3.31s recent_step=2.59s eta=42:34:22 +2026-09-14 17:14:06 | INFO | train rank=0 | step=3760/50000 progress=7.5% loss=0.176803 lr=9.982648e-05 lr_backbone=9.982648e-05 elapsed=03:27:39 avg_step=3.31s recent_step=3.24s eta=42:33:40 +2026-09-14 17:14:34 | INFO | train rank=0 | step=3770/50000 progress=7.5% loss=0.115264 lr=9.982372e-05 lr_backbone=9.982372e-05 elapsed=03:28:07 avg_step=3.31s recent_step=2.80s eta=42:32:04 +2026-09-14 17:15:17 | INFO | train rank=0 | step=3780/50000 progress=7.6% loss=0.138184 lr=9.982093e-05 lr_backbone=9.982093e-05 elapsed=03:28:50 avg_step=3.31s recent_step=4.35s eta=42:33:38 +2026-09-14 17:15:50 | INFO | train rank=0 | step=3790/50000 progress=7.6% loss=0.166199 lr=9.981813e-05 lr_backbone=9.981813e-05 elapsed=03:29:22 avg_step=3.31s recent_step=3.22s eta=42:32:53 +2026-09-14 17:16:20 | INFO | train rank=0 | step=3800/50000 progress=7.6% loss=0.176653 lr=9.981530e-05 lr_backbone=9.981530e-05 elapsed=03:29:52 avg_step=3.31s recent_step=3.00s eta=42:31:42 +2026-09-14 17:16:59 | INFO | train rank=0 | step=3810/50000 progress=7.6% loss=0.123373 lr=9.981245e-05 lr_backbone=9.981245e-05 elapsed=03:30:32 avg_step=3.32s recent_step=3.96s eta=42:32:27 +2026-09-14 17:17:36 | INFO | train rank=0 | step=3820/50000 progress=7.6% loss=0.164213 lr=9.980957e-05 lr_backbone=9.980957e-05 elapsed=03:31:08 avg_step=3.32s recent_step=3.65s eta=42:32:34 +2026-09-14 17:18:06 | INFO | train rank=0 | step=3830/50000 progress=7.7% loss=0.154014 lr=9.980668e-05 lr_backbone=9.980668e-05 elapsed=03:31:39 avg_step=3.32s recent_step=3.01s eta=42:31:24 +2026-09-14 17:18:42 | INFO | train rank=0 | step=3840/50000 progress=7.7% loss=0.154499 lr=9.980376e-05 lr_backbone=9.980376e-05 elapsed=03:32:15 avg_step=3.32s recent_step=3.66s eta=42:31:32 +2026-09-14 17:19:26 | INFO | train rank=0 | step=3850/50000 progress=7.7% loss=0.121730 lr=9.980083e-05 lr_backbone=9.980083e-05 elapsed=03:32:58 avg_step=3.32s recent_step=4.33s eta=42:33:00 +2026-09-14 17:20:01 | INFO | train rank=0 | step=3860/50000 progress=7.7% loss=0.132926 lr=9.979787e-05 lr_backbone=9.979787e-05 elapsed=03:33:33 avg_step=3.32s recent_step=3.51s eta=42:32:50 +2026-09-14 17:20:32 | INFO | train rank=0 | step=3870/50000 progress=7.7% loss=0.137585 lr=9.979489e-05 lr_backbone=9.979489e-05 elapsed=03:34:04 avg_step=3.32s recent_step=3.09s eta=42:31:49 +2026-09-14 17:21:07 | INFO | train rank=0 | step=3880/50000 progress=7.8% loss=0.142797 lr=9.979188e-05 lr_backbone=9.979188e-05 elapsed=03:34:40 avg_step=3.32s recent_step=3.56s eta=42:31:44 +2026-09-14 17:22:00 | INFO | train rank=0 | step=3890/50000 progress=7.8% loss=0.119681 lr=9.978886e-05 lr_backbone=9.978886e-05 elapsed=03:35:33 avg_step=3.32s recent_step=5.27s eta=42:35:02 +2026-09-14 17:22:34 | INFO | train rank=0 | step=3900/50000 progress=7.8% loss=0.126535 lr=9.978581e-05 lr_backbone=9.978581e-05 elapsed=03:36:06 avg_step=3.32s recent_step=3.38s eta=42:34:35 +2026-09-14 17:23:13 | INFO | train rank=0 | step=3910/50000 progress=7.8% loss=0.111169 lr=9.978274e-05 lr_backbone=9.978274e-05 elapsed=03:36:45 avg_step=3.33s recent_step=3.90s eta=42:35:09 +2026-09-14 17:23:43 | INFO | train rank=0 | step=3920/50000 progress=7.8% loss=0.130016 lr=9.977965e-05 lr_backbone=9.977965e-05 elapsed=03:37:15 avg_step=3.33s recent_step=2.99s eta=42:33:57 +2026-09-14 17:24:24 | INFO | train rank=0 | step=3930/50000 progress=7.9% loss=0.151015 lr=9.977654e-05 lr_backbone=9.977654e-05 elapsed=03:37:57 avg_step=3.33s recent_step=4.14s eta=42:34:59 +2026-09-14 17:25:01 | INFO | train rank=0 | step=3940/50000 progress=7.9% loss=0.167249 lr=9.977341e-05 lr_backbone=9.977341e-05 elapsed=03:38:34 avg_step=3.33s recent_step=3.70s eta=42:35:09 +2026-09-14 17:25:33 | INFO | train rank=0 | step=3950/50000 progress=7.9% loss=0.127559 lr=9.977025e-05 lr_backbone=9.977025e-05 elapsed=03:39:06 avg_step=3.33s recent_step=3.23s eta=42:34:24 +2026-09-14 17:26:03 | INFO | train rank=0 | step=3960/50000 progress=7.9% loss=0.140614 lr=9.976707e-05 lr_backbone=9.976707e-05 elapsed=03:39:36 avg_step=3.33s recent_step=2.95s eta=42:33:07 +2026-09-14 17:26:33 | INFO | train rank=0 | step=3970/50000 progress=7.9% loss=0.113656 lr=9.976387e-05 lr_backbone=9.976387e-05 elapsed=03:40:06 avg_step=3.33s recent_step=3.03s eta=42:31:59 +2026-09-14 17:27:33 | INFO | train rank=0 | step=3980/50000 progress=8.0% loss=0.153500 lr=9.976065e-05 lr_backbone=9.976065e-05 elapsed=03:41:06 avg_step=3.33s recent_step=5.99s eta=42:36:33 +2026-09-14 17:28:04 | INFO | train rank=0 | step=3990/50000 progress=8.0% loss=0.135409 lr=9.975741e-05 lr_backbone=9.975741e-05 elapsed=03:41:37 avg_step=3.33s recent_step=3.10s eta=42:35:33 +2026-09-14 17:28:38 | INFO | train rank=0 | step=4000/50000 progress=8.0% loss=0.156765 lr=9.975415e-05 lr_backbone=9.975415e-05 elapsed=03:42:10 avg_step=3.33s recent_step=3.38s eta=42:35:04 +2026-09-14 17:29:14 | INFO | train rank=0 | step=4010/50000 progress=8.0% loss=0.138041 lr=9.975086e-05 lr_backbone=9.975086e-05 elapsed=03:42:47 avg_step=3.33s recent_step=3.65s eta=42:35:07 +2026-09-14 17:30:07 | INFO | train rank=0 | step=4020/50000 progress=8.0% loss=0.113748 lr=9.974755e-05 lr_backbone=9.974755e-05 elapsed=03:43:40 avg_step=3.34s recent_step=5.34s eta=42:38:23 +2026-09-14 17:30:40 | INFO | train rank=0 | step=4030/50000 progress=8.1% loss=0.102938 lr=9.974422e-05 lr_backbone=9.974422e-05 elapsed=03:44:12 avg_step=3.34s recent_step=3.22s eta=42:37:36 +2026-09-14 17:31:13 | INFO | train rank=0 | step=4040/50000 progress=8.1% loss=0.131042 lr=9.974087e-05 lr_backbone=9.974087e-05 elapsed=03:44:46 avg_step=3.34s recent_step=3.35s eta=42:37:04 +2026-09-14 17:31:46 | INFO | train rank=0 | step=4050/50000 progress=8.1% loss=0.160846 lr=9.973750e-05 lr_backbone=9.973750e-05 elapsed=03:45:19 avg_step=3.34s recent_step=3.31s eta=42:36:28 +2026-09-14 17:32:33 | INFO | train rank=0 | step=4060/50000 progress=8.1% loss=0.153407 lr=9.973410e-05 lr_backbone=9.973410e-05 elapsed=03:46:06 avg_step=3.34s recent_step=4.68s eta=42:38:27 +2026-09-14 17:33:08 | INFO | train rank=0 | step=4070/50000 progress=8.1% loss=0.175761 lr=9.973068e-05 lr_backbone=9.973068e-05 elapsed=03:46:41 avg_step=3.34s recent_step=3.49s eta=42:38:10 +2026-09-14 17:33:39 | INFO | train rank=0 | step=4080/50000 progress=8.2% loss=0.106390 lr=9.972725e-05 lr_backbone=9.972725e-05 elapsed=03:47:12 avg_step=3.34s recent_step=3.12s eta=42:37:12 +2026-09-14 17:34:05 | INFO | train rank=0 | step=4090/50000 progress=8.2% loss=0.142522 lr=9.972379e-05 lr_backbone=9.972379e-05 elapsed=03:47:37 avg_step=3.34s recent_step=2.53s eta=42:35:08 +2026-09-14 17:34:50 | INFO | train rank=0 | step=4100/50000 progress=8.2% loss=0.130650 lr=9.972030e-05 lr_backbone=9.972030e-05 elapsed=03:48:22 avg_step=3.34s recent_step=4.51s eta=42:36:45 +2026-09-14 17:35:24 | INFO | train rank=0 | step=4110/50000 progress=8.2% loss=0.143421 lr=9.971680e-05 lr_backbone=9.971680e-05 elapsed=03:48:56 avg_step=3.34s recent_step=3.40s eta=42:36:18 +2026-09-14 17:35:56 | INFO | train rank=0 | step=4120/50000 progress=8.2% loss=0.141197 lr=9.971327e-05 lr_backbone=9.971327e-05 elapsed=03:49:28 avg_step=3.34s recent_step=3.21s eta=42:35:29 +2026-09-14 17:36:33 | INFO | train rank=0 | step=4130/50000 progress=8.3% loss=0.130302 lr=9.970973e-05 lr_backbone=9.970973e-05 elapsed=03:50:06 avg_step=3.34s recent_step=3.71s eta=42:35:37 +2026-09-14 17:37:19 | INFO | train rank=0 | step=4140/50000 progress=8.3% loss=0.155271 lr=9.970616e-05 lr_backbone=9.970616e-05 elapsed=03:50:52 avg_step=3.35s recent_step=4.66s eta=42:37:29 +2026-09-14 17:37:50 | INFO | train rank=0 | step=4150/50000 progress=8.3% loss=0.117313 lr=9.970257e-05 lr_backbone=9.970257e-05 elapsed=03:51:23 avg_step=3.35s recent_step=3.09s eta=42:36:27 +2026-09-14 17:38:20 | INFO | train rank=0 | step=4160/50000 progress=8.3% loss=0.139379 lr=9.969895e-05 lr_backbone=9.969895e-05 elapsed=03:51:53 avg_step=3.34s recent_step=2.96s eta=42:35:11 +2026-09-14 17:38:54 | INFO | train rank=0 | step=4170/50000 progress=8.3% loss=0.151863 lr=9.969532e-05 lr_backbone=9.969532e-05 elapsed=03:52:27 avg_step=3.34s recent_step=3.46s eta=42:34:50 +2026-09-14 17:39:29 | INFO | train rank=0 | step=4180/50000 progress=8.4% loss=0.149042 lr=9.969166e-05 lr_backbone=9.969166e-05 elapsed=03:53:02 avg_step=3.34s recent_step=3.44s eta=42:34:27 +2026-09-14 17:40:00 | INFO | train rank=0 | step=4190/50000 progress=8.4% loss=0.134961 lr=9.968799e-05 lr_backbone=9.968799e-05 elapsed=03:53:33 avg_step=3.34s recent_step=3.11s eta=42:33:27 +2026-09-14 17:40:36 | INFO | train rank=0 | step=4200/50000 progress=8.4% loss=0.140332 lr=9.968429e-05 lr_backbone=9.968429e-05 elapsed=03:54:09 avg_step=3.35s recent_step=3.60s eta=42:33:21 +2026-09-14 17:41:20 | INFO | train rank=0 | step=4210/50000 progress=8.4% loss=0.128261 lr=9.968057e-05 lr_backbone=9.968057e-05 elapsed=03:54:53 avg_step=3.35s recent_step=4.42s eta=42:34:45 +2026-09-14 17:41:55 | INFO | train rank=0 | step=4220/50000 progress=8.4% loss=0.152501 lr=9.967682e-05 lr_backbone=9.967682e-05 elapsed=03:55:27 avg_step=3.35s recent_step=3.47s eta=42:34:24 +2026-09-14 17:42:31 | INFO | train rank=0 | step=4230/50000 progress=8.5% loss=0.168643 lr=9.967306e-05 lr_backbone=9.967306e-05 elapsed=03:56:04 avg_step=3.35s recent_step=3.67s eta=42:34:26 +2026-09-14 17:43:11 | INFO | train rank=0 | step=4240/50000 progress=8.5% loss=0.133706 lr=9.966927e-05 lr_backbone=9.966927e-05 elapsed=03:56:44 avg_step=3.35s recent_step=3.96s eta=42:34:59 +2026-09-14 17:43:56 | INFO | train rank=0 | step=4250/50000 progress=8.5% loss=0.230895 lr=9.966546e-05 lr_backbone=9.966546e-05 elapsed=03:57:29 avg_step=3.35s recent_step=4.52s eta=42:36:32 +2026-09-14 17:44:33 | INFO | train rank=0 | step=4260/50000 progress=8.5% loss=0.134604 lr=9.966163e-05 lr_backbone=9.966163e-05 elapsed=03:58:06 avg_step=3.35s recent_step=3.65s eta=42:36:30 +2026-09-14 17:45:00 | INFO | train rank=0 | step=4270/50000 progress=8.5% loss=0.136876 lr=9.965778e-05 lr_backbone=9.965778e-05 elapsed=03:58:33 avg_step=3.35s recent_step=2.75s eta=42:34:52 +2026-09-14 17:45:39 | INFO | train rank=0 | step=4280/50000 progress=8.6% loss=0.142109 lr=9.965391e-05 lr_backbone=9.965391e-05 elapsed=03:59:12 avg_step=3.35s recent_step=3.85s eta=42:35:13 +2026-09-14 17:46:19 | INFO | train rank=0 | step=4290/50000 progress=8.6% loss=0.156105 lr=9.965001e-05 lr_backbone=9.965001e-05 elapsed=03:59:52 avg_step=3.35s recent_step=3.99s eta=42:35:47 +2026-09-14 17:46:55 | INFO | train rank=0 | step=4300/50000 progress=8.6% loss=0.184306 lr=9.964610e-05 lr_backbone=9.964610e-05 elapsed=04:00:28 avg_step=3.36s recent_step=3.67s eta=42:35:46 +2026-09-14 17:47:28 | INFO | train rank=0 | step=4310/50000 progress=8.6% loss=0.123412 lr=9.964216e-05 lr_backbone=9.964216e-05 elapsed=04:01:00 avg_step=3.36s recent_step=3.23s eta=42:34:59 +2026-09-14 17:48:02 | INFO | train rank=0 | step=4320/50000 progress=8.6% loss=0.155353 lr=9.963820e-05 lr_backbone=9.963820e-05 elapsed=04:01:35 avg_step=3.36s recent_step=3.47s eta=42:34:38 +2026-09-14 17:48:58 | INFO | train rank=0 | step=4330/50000 progress=8.7% loss=0.100425 lr=9.963422e-05 lr_backbone=9.963422e-05 elapsed=04:02:31 avg_step=3.36s recent_step=5.59s eta=42:38:01 +2026-09-14 17:49:29 | INFO | train rank=0 | step=4340/50000 progress=8.7% loss=0.139296 lr=9.963021e-05 lr_backbone=9.963021e-05 elapsed=04:03:02 avg_step=3.36s recent_step=3.04s eta=42:36:54 +2026-09-14 17:50:06 | INFO | train rank=0 | step=4350/50000 progress=8.7% loss=0.146398 lr=9.962619e-05 lr_backbone=9.962619e-05 elapsed=04:03:39 avg_step=3.36s recent_step=3.72s eta=42:36:58 +2026-09-14 17:50:37 | INFO | train rank=0 | step=4360/50000 progress=8.7% loss=0.139718 lr=9.962214e-05 lr_backbone=9.962214e-05 elapsed=04:04:10 avg_step=3.36s recent_step=3.09s eta=42:35:55 +2026-09-14 17:51:28 | INFO | train rank=0 | step=4370/50000 progress=8.7% loss=0.127858 lr=9.961807e-05 lr_backbone=9.961807e-05 elapsed=04:05:00 avg_step=3.36s recent_step=5.08s eta=42:38:21 +2026-09-14 17:51:55 | INFO | train rank=0 | step=4380/50000 progress=8.8% loss=0.132977 lr=9.961398e-05 lr_backbone=9.961398e-05 elapsed=04:05:28 avg_step=3.36s recent_step=2.77s eta=42:36:46 +2026-09-14 17:52:26 | INFO | train rank=0 | step=4390/50000 progress=8.8% loss=0.121042 lr=9.960987e-05 lr_backbone=9.960987e-05 elapsed=04:05:59 avg_step=3.36s recent_step=3.07s eta=42:35:42 +2026-09-14 17:52:55 | INFO | train rank=0 | step=4400/50000 progress=8.8% loss=0.158385 lr=9.960574e-05 lr_backbone=9.960574e-05 elapsed=04:06:28 avg_step=3.36s recent_step=2.92s eta=42:34:22 +2026-09-14 17:53:40 | INFO | train rank=0 | step=4410/50000 progress=8.8% loss=0.135858 lr=9.960158e-05 lr_backbone=9.960158e-05 elapsed=04:07:13 avg_step=3.36s recent_step=4.47s eta=42:35:44 +2026-09-14 17:54:15 | INFO | train rank=0 | step=4420/50000 progress=8.8% loss=0.139943 lr=9.959740e-05 lr_backbone=9.959740e-05 elapsed=04:07:48 avg_step=3.36s recent_step=3.48s eta=42:35:22 +2026-09-14 17:54:45 | INFO | train rank=0 | step=4430/50000 progress=8.9% loss=0.138874 lr=9.959320e-05 lr_backbone=9.959320e-05 elapsed=04:08:18 avg_step=3.36s recent_step=3.01s eta=42:34:12 +2026-09-14 17:55:18 | INFO | train rank=0 | step=4440/50000 progress=8.9% loss=0.148455 lr=9.958898e-05 lr_backbone=9.958898e-05 elapsed=04:08:51 avg_step=3.36s recent_step=3.32s eta=42:33:34 +2026-09-14 17:55:55 | INFO | train rank=0 | step=4450/50000 progress=8.9% loss=0.150913 lr=9.958474e-05 lr_backbone=9.958474e-05 elapsed=04:09:27 avg_step=3.36s recent_step=3.65s eta=42:33:30 +2026-09-14 17:56:26 | INFO | train rank=0 | step=4460/50000 progress=8.9% loss=0.149868 lr=9.958048e-05 lr_backbone=9.958048e-05 elapsed=04:09:59 avg_step=3.36s recent_step=3.14s eta=42:32:34 +2026-09-14 17:57:00 | INFO | train rank=0 | step=4470/50000 progress=8.9% loss=0.152792 lr=9.957619e-05 lr_backbone=9.957619e-05 elapsed=04:10:33 avg_step=3.36s recent_step=3.40s eta=42:32:04 +2026-09-14 17:57:32 | INFO | train rank=0 | step=4480/50000 progress=9.0% loss=0.168451 lr=9.957188e-05 lr_backbone=9.957188e-05 elapsed=04:11:05 avg_step=3.36s recent_step=3.20s eta=42:31:13 +2026-09-14 17:58:18 | INFO | train rank=0 | step=4490/50000 progress=9.0% loss=0.123455 lr=9.956755e-05 lr_backbone=9.956755e-05 elapsed=04:11:51 avg_step=3.37s recent_step=4.63s eta=42:32:48 +2026-09-14 17:58:52 | INFO | train rank=0 | step=4500/50000 progress=9.0% loss=0.123225 lr=9.956320e-05 lr_backbone=9.956320e-05 elapsed=04:12:25 avg_step=3.37s recent_step=3.38s eta=42:32:16 +2026-09-14 17:59:31 | INFO | train rank=0 | step=4510/50000 progress=9.0% loss=0.148442 lr=9.955883e-05 lr_backbone=9.955883e-05 elapsed=04:13:04 avg_step=3.37s recent_step=3.89s eta=42:32:36 +2026-09-14 18:00:02 | INFO | train rank=0 | step=4520/50000 progress=9.0% loss=0.102462 lr=9.955444e-05 lr_backbone=9.955444e-05 elapsed=04:13:35 avg_step=3.37s recent_step=3.14s eta=42:31:39 +2026-09-14 18:00:49 | INFO | train rank=0 | step=4530/50000 progress=9.1% loss=0.129572 lr=9.955002e-05 lr_backbone=9.955002e-05 elapsed=04:14:22 avg_step=3.37s recent_step=4.69s eta=42:33:18 +2026-09-14 18:01:21 | INFO | train rank=0 | step=4540/50000 progress=9.1% loss=0.118392 lr=9.954558e-05 lr_backbone=9.954558e-05 elapsed=04:14:54 avg_step=3.37s recent_step=3.17s eta=42:32:24 +2026-09-14 18:01:54 | INFO | train rank=0 | step=4550/50000 progress=9.1% loss=0.162617 lr=9.954112e-05 lr_backbone=9.954112e-05 elapsed=04:15:26 avg_step=3.37s recent_step=3.26s eta=42:31:39 +2026-09-14 18:02:30 | INFO | train rank=0 | step=4560/50000 progress=9.1% loss=0.173704 lr=9.953664e-05 lr_backbone=9.953664e-05 elapsed=04:16:03 avg_step=3.37s recent_step=3.67s eta=42:31:35 +2026-09-14 18:03:10 | INFO | train rank=0 | step=4570/50000 progress=9.1% loss=0.201192 lr=9.953214e-05 lr_backbone=9.953214e-05 elapsed=04:16:42 avg_step=3.37s recent_step=3.94s eta=42:31:58 +2026-09-14 18:03:45 | INFO | train rank=0 | step=4580/50000 progress=9.2% loss=0.132897 lr=9.952762e-05 lr_backbone=9.952762e-05 elapsed=04:17:18 avg_step=3.37s recent_step=3.52s eta=42:31:40 +2026-09-14 18:04:16 | INFO | train rank=0 | step=4590/50000 progress=9.2% loss=0.115069 lr=9.952307e-05 lr_backbone=9.952307e-05 elapsed=04:17:49 avg_step=3.37s recent_step=3.17s eta=42:30:46 +2026-09-14 18:04:51 | INFO | train rank=0 | step=4600/50000 progress=9.2% loss=0.133444 lr=9.951850e-05 lr_backbone=9.951850e-05 elapsed=04:18:23 avg_step=3.37s recent_step=3.41s eta=42:30:16 +2026-09-14 18:05:37 | INFO | train rank=0 | step=4610/50000 progress=9.2% loss=0.125441 lr=9.951391e-05 lr_backbone=9.951391e-05 elapsed=04:19:10 avg_step=3.37s recent_step=4.65s eta=42:31:48 +2026-09-14 18:06:06 | INFO | train rank=0 | step=4620/50000 progress=9.2% loss=0.107253 lr=9.950930e-05 lr_backbone=9.950930e-05 elapsed=04:19:38 avg_step=3.37s recent_step=2.86s eta=42:30:24 +2026-09-14 18:06:34 | INFO | train rank=0 | step=4630/50000 progress=9.3% loss=0.200940 lr=9.950467e-05 lr_backbone=9.950467e-05 elapsed=04:20:07 avg_step=3.37s recent_step=2.85s eta=42:28:59 +2026-09-14 18:07:25 | INFO | train rank=0 | step=4640/50000 progress=9.3% loss=0.124016 lr=9.950002e-05 lr_backbone=9.950002e-05 elapsed=04:20:58 avg_step=3.37s recent_step=5.08s eta=42:31:12 +2026-09-14 18:08:00 | INFO | train rank=0 | step=4650/50000 progress=9.3% loss=0.130971 lr=9.949534e-05 lr_backbone=9.949534e-05 elapsed=04:21:33 avg_step=3.37s recent_step=3.54s eta=42:30:55 +2026-09-14 18:08:30 | INFO | train rank=0 | step=4660/50000 progress=9.3% loss=0.121919 lr=9.949064e-05 lr_backbone=9.949064e-05 elapsed=04:22:02 avg_step=3.37s recent_step=2.93s eta=42:29:37 +2026-09-14 18:09:02 | INFO | train rank=0 | step=4670/50000 progress=9.3% loss=0.102581 lr=9.948593e-05 lr_backbone=9.948593e-05 elapsed=04:22:35 avg_step=3.37s recent_step=3.28s eta=42:28:55 +2026-09-14 18:09:58 | INFO | train rank=0 | step=4680/50000 progress=9.4% loss=0.153390 lr=9.948118e-05 lr_backbone=9.948118e-05 elapsed=04:23:31 avg_step=3.38s recent_step=5.57s eta=42:31:53 +2026-09-14 18:10:34 | INFO | train rank=0 | step=4690/50000 progress=9.4% loss=0.133858 lr=9.947642e-05 lr_backbone=9.947642e-05 elapsed=04:24:07 avg_step=3.38s recent_step=3.58s eta=42:31:39 +2026-09-14 18:11:00 | INFO | train rank=0 | step=4700/50000 progress=9.4% loss=0.131406 lr=9.947164e-05 lr_backbone=9.947164e-05 elapsed=04:24:33 avg_step=3.38s recent_step=2.63s eta=42:29:53 +2026-09-14 18:11:37 | INFO | train rank=0 | step=4710/50000 progress=9.4% loss=0.148749 lr=9.946683e-05 lr_backbone=9.946683e-05 elapsed=04:25:10 avg_step=3.38s recent_step=3.71s eta=42:29:51 +2026-09-14 18:12:18 | INFO | train rank=0 | step=4720/50000 progress=9.4% loss=0.125198 lr=9.946201e-05 lr_backbone=9.946201e-05 elapsed=04:25:51 avg_step=3.38s recent_step=4.09s eta=42:30:26 +2026-09-14 18:12:53 | INFO | train rank=0 | step=4730/50000 progress=9.5% loss=0.139744 lr=9.945716e-05 lr_backbone=9.945716e-05 elapsed=04:26:26 avg_step=3.38s recent_step=3.45s eta=42:29:59 +2026-09-14 18:13:25 | INFO | train rank=0 | step=4740/50000 progress=9.5% loss=0.145712 lr=9.945229e-05 lr_backbone=9.945229e-05 elapsed=04:26:58 avg_step=3.38s recent_step=3.24s eta=42:29:11 +2026-09-14 18:14:01 | INFO | train rank=0 | step=4750/50000 progress=9.5% loss=0.141110 lr=9.944739e-05 lr_backbone=9.944739e-05 elapsed=04:27:33 avg_step=3.38s recent_step=3.55s eta=42:28:54 +2026-09-14 18:14:39 | INFO | train rank=0 | step=4760/50000 progress=9.5% loss=0.151206 lr=9.944248e-05 lr_backbone=9.944248e-05 elapsed=04:28:11 avg_step=3.38s recent_step=3.81s eta=42:29:01 +2026-09-14 18:15:19 | INFO | train rank=0 | step=4770/50000 progress=9.5% loss=0.135613 lr=9.943754e-05 lr_backbone=9.943754e-05 elapsed=04:28:52 avg_step=3.38s recent_step=4.02s eta=42:29:27 +2026-09-14 18:15:50 | INFO | train rank=0 | step=4780/50000 progress=9.6% loss=0.128772 lr=9.943259e-05 lr_backbone=9.943259e-05 elapsed=04:29:22 avg_step=3.38s recent_step=3.07s eta=42:28:24 +2026-09-14 18:16:27 | INFO | train rank=0 | step=4790/50000 progress=9.6% loss=0.174558 lr=9.942761e-05 lr_backbone=9.942761e-05 elapsed=04:30:00 avg_step=3.38s recent_step=3.76s eta=42:28:25 +2026-09-14 18:17:12 | INFO | train rank=0 | step=4800/50000 progress=9.6% loss=0.124087 lr=9.942261e-05 lr_backbone=9.942261e-05 elapsed=04:30:44 avg_step=3.38s recent_step=4.45s eta=42:29:32 +2026-09-14 18:17:48 | INFO | train rank=0 | step=4810/50000 progress=9.6% loss=0.128842 lr=9.941759e-05 lr_backbone=9.941759e-05 elapsed=04:31:21 avg_step=3.38s recent_step=3.66s eta=42:29:24 +2026-09-14 18:18:22 | INFO | train rank=0 | step=4820/50000 progress=9.6% loss=0.143948 lr=9.941254e-05 lr_backbone=9.941254e-05 elapsed=04:31:55 avg_step=3.39s recent_step=3.43s eta=42:28:54 +2026-09-14 18:18:48 | INFO | train rank=0 | step=4830/50000 progress=9.7% loss=0.121069 lr=9.940748e-05 lr_backbone=9.940748e-05 elapsed=04:32:21 avg_step=3.38s recent_step=2.57s eta=42:27:04 +2026-09-14 18:19:48 | INFO | train rank=0 | step=4840/50000 progress=9.7% loss=0.120929 lr=9.940239e-05 lr_backbone=9.940239e-05 elapsed=04:33:20 avg_step=3.39s recent_step=5.95s eta=42:30:30 +2026-09-14 18:20:12 | INFO | train rank=0 | step=4850/50000 progress=9.7% loss=0.108768 lr=9.939728e-05 lr_backbone=9.939728e-05 elapsed=04:33:45 avg_step=3.39s recent_step=2.44s eta=42:28:27 +2026-09-14 18:20:45 | INFO | train rank=0 | step=4860/50000 progress=9.7% loss=0.112348 lr=9.939215e-05 lr_backbone=9.939215e-05 elapsed=04:34:18 avg_step=3.39s recent_step=3.32s eta=42:27:47 +2026-09-14 18:21:22 | INFO | train rank=0 | step=4870/50000 progress=9.7% loss=0.164466 lr=9.938700e-05 lr_backbone=9.938700e-05 elapsed=04:34:55 avg_step=3.39s recent_step=3.71s eta=42:27:44 +2026-09-14 18:22:12 | INFO | train rank=0 | step=4880/50000 progress=9.8% loss=0.172308 lr=9.938183e-05 lr_backbone=9.938183e-05 elapsed=04:35:45 avg_step=3.39s recent_step=4.99s eta=42:29:38 +2026-09-14 18:22:49 | INFO | train rank=0 | step=4890/50000 progress=9.8% loss=0.158004 lr=9.937663e-05 lr_backbone=9.937663e-05 elapsed=04:36:22 avg_step=3.39s recent_step=3.66s eta=42:29:28 +2026-09-14 18:23:18 | INFO | train rank=0 | step=4900/50000 progress=9.8% loss=0.202151 lr=9.937142e-05 lr_backbone=9.937142e-05 elapsed=04:36:51 avg_step=3.39s recent_step=2.97s eta=42:28:15 +2026-09-14 18:23:54 | INFO | train rank=0 | step=4910/50000 progress=9.8% loss=0.129008 lr=9.936618e-05 lr_backbone=9.936618e-05 elapsed=04:37:27 avg_step=3.39s recent_step=3.55s eta=42:27:56 +2026-09-14 18:24:33 | INFO | train rank=0 | step=4920/50000 progress=9.8% loss=0.148058 lr=9.936092e-05 lr_backbone=9.936092e-05 elapsed=04:38:06 avg_step=3.39s recent_step=3.94s eta=42:28:12 +2026-09-14 18:25:03 | INFO | train rank=0 | step=4930/50000 progress=9.9% loss=0.099218 lr=9.935564e-05 lr_backbone=9.935564e-05 elapsed=04:38:36 avg_step=3.39s recent_step=2.97s eta=42:27:00 +2026-09-14 18:25:40 | INFO | train rank=0 | step=4940/50000 progress=9.9% loss=0.112648 lr=9.935034e-05 lr_backbone=9.935034e-05 elapsed=04:39:13 avg_step=3.39s recent_step=3.73s eta=42:26:57 +2026-09-14 18:26:17 | INFO | train rank=0 | step=4950/50000 progress=9.9% loss=0.126947 lr=9.934501e-05 lr_backbone=9.934501e-05 elapsed=04:39:49 avg_step=3.39s recent_step=3.63s eta=42:26:45 +2026-09-14 18:27:14 | INFO | train rank=0 | step=4960/50000 progress=9.9% loss=0.140525 lr=9.933966e-05 lr_backbone=9.933966e-05 elapsed=04:40:47 avg_step=3.40s recent_step=5.72s eta=42:29:42 +2026-09-14 18:27:49 | INFO | train rank=0 | step=4970/50000 progress=9.9% loss=0.137761 lr=9.933430e-05 lr_backbone=9.933430e-05 elapsed=04:41:22 avg_step=3.40s recent_step=3.51s eta=42:29:18 +2026-09-14 18:28:25 | INFO | train rank=0 | step=4980/50000 progress=10.0% loss=0.096086 lr=9.932891e-05 lr_backbone=9.932891e-05 elapsed=04:41:58 avg_step=3.40s recent_step=3.60s eta=42:29:03 +2026-09-14 18:28:51 | INFO | train rank=0 | step=4990/50000 progress=10.0% loss=0.114752 lr=9.932350e-05 lr_backbone=9.932350e-05 elapsed=04:42:24 avg_step=3.40s recent_step=2.66s eta=42:27:22 +2026-09-14 18:29:37 | INFO | train rank=0 | step=5000/50000 progress=10.0% loss=0.128582 lr=9.931807e-05 lr_backbone=9.931807e-05 elapsed=04:43:10 avg_step=3.40s recent_step=4.59s eta=42:28:36 +2026-09-14 18:30:14 | INFO | train rank=0 | step=5010/50000 progress=10.0% loss=0.121906 lr=9.931261e-05 lr_backbone=9.931261e-05 elapsed=04:43:47 avg_step=3.40s recent_step=3.70s eta=42:28:29 +2026-09-14 18:30:43 | INFO | train rank=0 | step=5020/50000 progress=10.0% loss=0.108067 lr=9.930714e-05 lr_backbone=9.930714e-05 elapsed=04:44:16 avg_step=3.40s recent_step=2.87s eta=42:27:08 +2026-09-14 18:31:18 | INFO | train rank=0 | step=5030/50000 progress=10.1% loss=0.182623 lr=9.930164e-05 lr_backbone=9.930164e-05 elapsed=04:44:51 avg_step=3.40s recent_step=3.53s eta=42:26:46 +2026-09-14 18:32:06 | INFO | train rank=0 | step=5040/50000 progress=10.1% loss=0.144554 lr=9.929612e-05 lr_backbone=9.929612e-05 elapsed=04:45:39 avg_step=3.40s recent_step=4.79s eta=42:28:16 +2026-09-14 18:32:39 | INFO | train rank=0 | step=5050/50000 progress=10.1% loss=0.140626 lr=9.929058e-05 lr_backbone=9.929058e-05 elapsed=04:46:12 avg_step=3.40s recent_step=3.25s eta=42:27:29 +2026-09-14 18:33:13 | INFO | train rank=0 | step=5060/50000 progress=10.1% loss=0.124357 lr=9.928502e-05 lr_backbone=9.928502e-05 elapsed=04:46:46 avg_step=3.40s recent_step=3.44s eta=42:26:58 +2026-09-14 18:33:49 | INFO | train rank=0 | step=5070/50000 progress=10.1% loss=0.115570 lr=9.927944e-05 lr_backbone=9.927944e-05 elapsed=04:47:22 avg_step=3.40s recent_step=3.60s eta=42:26:42 +2026-09-14 18:34:30 | INFO | train rank=0 | step=5080/50000 progress=10.2% loss=0.131652 lr=9.927383e-05 lr_backbone=9.927383e-05 elapsed=04:48:03 avg_step=3.40s recent_step=4.12s eta=42:27:11 +2026-09-14 18:35:00 | INFO | train rank=0 | step=5090/50000 progress=10.2% loss=0.179402 lr=9.926820e-05 lr_backbone=9.926820e-05 elapsed=04:48:33 avg_step=3.40s recent_step=2.99s eta=42:26:01 +2026-09-14 18:35:37 | INFO | train rank=0 | step=5100/50000 progress=10.2% loss=0.120471 lr=9.926256e-05 lr_backbone=9.926256e-05 elapsed=04:49:10 avg_step=3.40s recent_step=3.68s eta=42:25:52 +2026-09-14 18:36:09 | INFO | train rank=0 | step=5110/50000 progress=10.2% loss=0.158145 lr=9.925689e-05 lr_backbone=9.925689e-05 elapsed=04:49:42 avg_step=3.40s recent_step=3.17s eta=42:24:57 +2026-09-14 18:36:58 | INFO | train rank=0 | step=5120/50000 progress=10.2% loss=0.130235 lr=9.925120e-05 lr_backbone=9.925120e-05 elapsed=04:50:30 avg_step=3.40s recent_step=4.87s eta=42:26:31 +2026-09-14 18:37:29 | INFO | train rank=0 | step=5130/50000 progress=10.3% loss=0.093378 lr=9.924548e-05 lr_backbone=9.924548e-05 elapsed=04:51:01 avg_step=3.40s recent_step=3.10s eta=42:25:31 +2026-09-14 18:38:08 | INFO | train rank=0 | step=5140/50000 progress=10.3% loss=0.133122 lr=9.923975e-05 lr_backbone=9.923975e-05 elapsed=04:51:41 avg_step=3.40s recent_step=3.93s eta=42:25:43 +2026-09-14 18:38:36 | INFO | train rank=0 | step=5150/50000 progress=10.3% loss=0.131242 lr=9.923399e-05 lr_backbone=9.923399e-05 elapsed=04:52:08 avg_step=3.40s recent_step=2.79s eta=42:24:15 +2026-09-14 18:39:22 | INFO | train rank=0 | step=5160/50000 progress=10.3% loss=0.115464 lr=9.922822e-05 lr_backbone=9.922822e-05 elapsed=04:52:54 avg_step=3.41s recent_step=4.59s eta=42:25:24 +2026-09-14 18:40:01 | INFO | train rank=0 | step=5170/50000 progress=10.3% loss=0.156526 lr=9.922242e-05 lr_backbone=9.922242e-05 elapsed=04:53:34 avg_step=3.41s recent_step=3.97s eta=42:25:39 +2026-09-14 18:40:35 | INFO | train rank=0 | step=5180/50000 progress=10.4% loss=0.103646 lr=9.921660e-05 lr_backbone=9.921660e-05 elapsed=04:54:07 avg_step=3.41s recent_step=3.33s eta=42:24:58 +2026-09-14 18:41:04 | INFO | train rank=0 | step=5190/50000 progress=10.4% loss=0.149178 lr=9.921076e-05 lr_backbone=9.921076e-05 elapsed=04:54:36 avg_step=3.41s recent_step=2.90s eta=42:23:40 +2026-09-14 18:41:50 | INFO | train rank=0 | step=5200/50000 progress=10.4% loss=0.135319 lr=9.920489e-05 lr_backbone=9.920489e-05 elapsed=04:55:22 avg_step=3.41s recent_step=4.60s eta=42:24:49 +2026-09-14 18:42:30 | INFO | train rank=0 | step=5210/50000 progress=10.4% loss=0.104770 lr=9.919901e-05 lr_backbone=9.919901e-05 elapsed=04:56:03 avg_step=3.41s recent_step=4.08s eta=42:25:13 +2026-09-14 18:42:59 | INFO | train rank=0 | step=5220/50000 progress=10.4% loss=0.133164 lr=9.919310e-05 lr_backbone=9.919310e-05 elapsed=04:56:32 avg_step=3.41s recent_step=2.86s eta=42:23:51 +2026-09-14 18:43:33 | INFO | train rank=0 | step=5230/50000 progress=10.5% loss=0.104377 lr=9.918717e-05 lr_backbone=9.918717e-05 elapsed=04:57:05 avg_step=3.41s recent_step=3.36s eta=42:23:13 +2026-09-14 18:44:26 | INFO | train rank=0 | step=5240/50000 progress=10.5% loss=0.187903 lr=9.918122e-05 lr_backbone=9.918122e-05 elapsed=04:57:58 avg_step=3.41s recent_step=5.29s eta=42:25:20 +2026-09-14 18:45:02 | INFO | train rank=0 | step=5250/50000 progress=10.5% loss=0.157430 lr=9.917525e-05 lr_backbone=9.917525e-05 elapsed=04:58:35 avg_step=3.41s recent_step=3.68s eta=42:25:09 +2026-09-14 18:45:34 | INFO | train rank=0 | step=5260/50000 progress=10.5% loss=0.171058 lr=9.916926e-05 lr_backbone=9.916926e-05 elapsed=04:59:07 avg_step=3.41s recent_step=3.16s eta=42:24:13 +2026-09-14 18:46:12 | INFO | train rank=0 | step=5270/50000 progress=10.5% loss=0.126266 lr=9.916325e-05 lr_backbone=9.916325e-05 elapsed=04:59:45 avg_step=3.41s recent_step=3.78s eta=42:24:10 +2026-09-14 18:46:58 | INFO | train rank=0 | step=5280/50000 progress=10.6% loss=0.134015 lr=9.915721e-05 lr_backbone=9.915721e-05 elapsed=05:00:30 avg_step=3.41s recent_step=4.58s eta=42:25:15 +2026-09-14 18:47:28 | INFO | train rank=0 | step=5290/50000 progress=10.6% loss=0.152472 lr=9.915116e-05 lr_backbone=9.915116e-05 elapsed=05:01:01 avg_step=3.41s recent_step=3.03s eta=42:24:08 +2026-09-14 18:47:58 | INFO | train rank=0 | step=5300/50000 progress=10.6% loss=0.131732 lr=9.914508e-05 lr_backbone=9.914508e-05 elapsed=05:01:30 avg_step=3.41s recent_step=2.99s eta=42:22:58 +2026-09-14 18:48:35 | INFO | train rank=0 | step=5310/50000 progress=10.6% loss=0.156865 lr=9.913898e-05 lr_backbone=9.913898e-05 elapsed=05:02:08 avg_step=3.41s recent_step=3.76s eta=42:22:53 +2026-09-14 18:49:18 | INFO | train rank=0 | step=5320/50000 progress=10.6% loss=0.123894 lr=9.913286e-05 lr_backbone=9.913286e-05 elapsed=05:02:51 avg_step=3.42s recent_step=4.31s eta=42:23:34 +2026-09-14 18:49:52 | INFO | train rank=0 | step=5330/50000 progress=10.7% loss=0.173579 lr=9.912671e-05 lr_backbone=9.912671e-05 elapsed=05:03:25 avg_step=3.42s recent_step=3.37s eta=42:22:56 +2026-09-14 18:50:29 | INFO | train rank=0 | step=5340/50000 progress=10.7% loss=0.115980 lr=9.912055e-05 lr_backbone=9.912055e-05 elapsed=05:04:02 avg_step=3.42s recent_step=3.68s eta=42:22:44 +2026-09-14 18:51:02 | INFO | train rank=0 | step=5350/50000 progress=10.7% loss=0.130454 lr=9.911436e-05 lr_backbone=9.911436e-05 elapsed=05:04:35 avg_step=3.42s recent_step=3.33s eta=42:22:03 +2026-09-14 18:51:47 | INFO | train rank=0 | step=5360/50000 progress=10.7% loss=0.129647 lr=9.910816e-05 lr_backbone=9.910816e-05 elapsed=05:05:20 avg_step=3.42s recent_step=4.49s eta=42:22:58 +2026-09-14 18:52:18 | INFO | train rank=0 | step=5370/50000 progress=10.7% loss=0.125204 lr=9.910193e-05 lr_backbone=9.910193e-05 elapsed=05:05:51 avg_step=3.42s recent_step=3.11s eta=42:21:58 +2026-09-14 18:52:46 | INFO | train rank=0 | step=5380/50000 progress=10.8% loss=0.150802 lr=9.909568e-05 lr_backbone=9.909568e-05 elapsed=05:06:19 avg_step=3.42s recent_step=2.77s eta=42:20:31 +2026-09-14 18:53:18 | INFO | train rank=0 | step=5390/50000 progress=10.8% loss=0.104871 lr=9.908940e-05 lr_backbone=9.908940e-05 elapsed=05:06:50 avg_step=3.42s recent_step=3.17s eta=42:19:36 +2026-09-14 18:54:06 | INFO | train rank=0 | step=5400/50000 progress=10.8% loss=0.143564 lr=9.908311e-05 lr_backbone=9.908311e-05 elapsed=05:07:39 avg_step=3.42s recent_step=4.89s eta=42:21:04 +2026-09-14 18:54:45 | INFO | train rank=0 | step=5410/50000 progress=10.8% loss=0.109002 lr=9.907680e-05 lr_backbone=9.907680e-05 elapsed=05:08:18 avg_step=3.42s recent_step=3.84s eta=42:21:04 +2026-09-14 18:55:25 | INFO | train rank=0 | step=5420/50000 progress=10.8% loss=0.092563 lr=9.907046e-05 lr_backbone=9.907046e-05 elapsed=05:08:58 avg_step=3.42s recent_step=4.01s eta=42:21:19 +2026-09-14 18:55:51 | INFO | train rank=0 | step=5430/50000 progress=10.9% loss=0.119864 lr=9.906410e-05 lr_backbone=9.906410e-05 elapsed=05:09:24 avg_step=3.42s recent_step=2.59s eta=42:19:36 +2026-09-14 18:56:24 | INFO | train rank=0 | step=5440/50000 progress=10.9% loss=0.114971 lr=9.905772e-05 lr_backbone=9.905772e-05 elapsed=05:09:56 avg_step=3.42s recent_step=3.27s eta=42:18:50 +2026-09-14 18:56:56 | INFO | train rank=0 | step=5450/50000 progress=10.9% loss=0.130183 lr=9.905132e-05 lr_backbone=9.905132e-05 elapsed=05:10:28 avg_step=3.42s recent_step=3.21s eta=42:17:58 +2026-09-14 18:57:27 | INFO | train rank=0 | step=5460/50000 progress=10.9% loss=0.111438 lr=9.904490e-05 lr_backbone=9.904490e-05 elapsed=05:10:59 avg_step=3.42s recent_step=3.10s eta=42:16:58 +2026-09-14 18:58:03 | INFO | train rank=0 | step=5470/50000 progress=10.9% loss=0.131670 lr=9.903846e-05 lr_backbone=9.903846e-05 elapsed=05:11:36 avg_step=3.42s recent_step=3.62s eta=42:16:40 +2026-09-14 18:59:03 | INFO | train rank=0 | step=5480/50000 progress=11.0% loss=0.128646 lr=9.903199e-05 lr_backbone=9.903199e-05 elapsed=05:12:35 avg_step=3.42s recent_step=5.97s eta=42:19:33 +2026-09-14 18:59:35 | INFO | train rank=0 | step=5490/50000 progress=11.0% loss=0.130174 lr=9.902551e-05 lr_backbone=9.902551e-05 elapsed=05:13:08 avg_step=3.42s recent_step=3.27s eta=42:18:47 +2026-09-14 19:00:09 | INFO | train rank=0 | step=5500/50000 progress=11.0% loss=0.142711 lr=9.901900e-05 lr_backbone=9.901900e-05 elapsed=05:13:42 avg_step=3.42s recent_step=3.38s eta=42:18:09 +2026-09-14 19:00:47 | INFO | train rank=0 | step=5510/50000 progress=11.0% loss=0.140367 lr=9.901247e-05 lr_backbone=9.901247e-05 elapsed=05:14:19 avg_step=3.42s recent_step=3.76s eta=42:18:02 +2026-09-14 19:01:40 | INFO | train rank=0 | step=5520/50000 progress=11.0% loss=0.128845 lr=9.900592e-05 lr_backbone=9.900592e-05 elapsed=05:15:13 avg_step=3.43s recent_step=5.35s eta=42:20:03 +2026-09-14 19:02:12 | INFO | train rank=0 | step=5530/50000 progress=11.1% loss=0.162509 lr=9.899935e-05 lr_backbone=9.899935e-05 elapsed=05:15:45 avg_step=3.43s recent_step=3.24s eta=42:19:14 +2026-09-14 19:02:46 | INFO | train rank=0 | step=5540/50000 progress=11.1% loss=0.128552 lr=9.899275e-05 lr_backbone=9.899275e-05 elapsed=05:16:19 avg_step=3.43s recent_step=3.35s eta=42:18:33 +2026-09-14 19:03:25 | INFO | train rank=0 | step=5550/50000 progress=11.1% loss=0.112428 lr=9.898614e-05 lr_backbone=9.898614e-05 elapsed=05:16:58 avg_step=3.43s recent_step=3.88s eta=42:18:35 +2026-09-14 19:04:07 | INFO | train rank=0 | step=5560/50000 progress=11.1% loss=0.131487 lr=9.897950e-05 lr_backbone=9.897950e-05 elapsed=05:17:40 avg_step=3.43s recent_step=4.21s eta=42:19:03 +2026-09-14 19:04:38 | INFO | train rank=0 | step=5570/50000 progress=11.1% loss=0.115742 lr=9.897284e-05 lr_backbone=9.897284e-05 elapsed=05:18:11 avg_step=3.43s recent_step=3.15s eta=42:18:07 +2026-09-14 19:05:09 | INFO | train rank=0 | step=5580/50000 progress=11.2% loss=0.157841 lr=9.896616e-05 lr_backbone=9.896616e-05 elapsed=05:18:42 avg_step=3.43s recent_step=3.08s eta=42:17:05 +2026-09-14 19:05:46 | INFO | train rank=0 | step=5590/50000 progress=11.2% loss=0.093585 lr=9.895946e-05 lr_backbone=9.895946e-05 elapsed=05:19:19 avg_step=3.43s recent_step=3.72s eta=42:16:53 +2026-09-14 19:06:31 | INFO | train rank=0 | step=5600/50000 progress=11.2% loss=0.131286 lr=9.895274e-05 lr_backbone=9.895274e-05 elapsed=05:20:04 avg_step=3.43s recent_step=4.47s eta=42:17:42 +2026-09-14 19:07:01 | INFO | train rank=0 | step=5610/50000 progress=11.2% loss=0.129999 lr=9.894600e-05 lr_backbone=9.894600e-05 elapsed=05:20:34 avg_step=3.43s recent_step=3.01s eta=42:16:34 +2026-09-14 19:07:30 | INFO | train rank=0 | step=5620/50000 progress=11.2% loss=0.114357 lr=9.893923e-05 lr_backbone=9.893923e-05 elapsed=05:21:02 avg_step=3.43s recent_step=2.85s eta=42:15:14 +2026-09-14 19:08:00 | INFO | train rank=0 | step=5630/50000 progress=11.3% loss=0.108781 lr=9.893245e-05 lr_backbone=9.893245e-05 elapsed=05:21:33 avg_step=3.43s recent_step=3.07s eta=42:14:12 +2026-09-14 19:08:47 | INFO | train rank=0 | step=5640/50000 progress=11.3% loss=0.109213 lr=9.892564e-05 lr_backbone=9.892564e-05 elapsed=05:22:20 avg_step=3.43s recent_step=4.69s eta=42:15:17 +2026-09-14 19:09:25 | INFO | train rank=0 | step=5650/50000 progress=11.3% loss=0.091412 lr=9.891881e-05 lr_backbone=9.891881e-05 elapsed=05:22:58 avg_step=3.43s recent_step=3.77s eta=42:15:10 +2026-09-14 19:09:54 | INFO | train rank=0 | step=5660/50000 progress=11.3% loss=0.121746 lr=9.891196e-05 lr_backbone=9.891196e-05 elapsed=05:23:27 avg_step=3.43s recent_step=2.90s eta=42:13:54 +2026-09-14 19:10:21 | INFO | train rank=0 | step=5670/50000 progress=11.3% loss=0.133140 lr=9.890509e-05 lr_backbone=9.890509e-05 elapsed=05:23:54 avg_step=3.43s recent_step=2.72s eta=42:12:24 +2026-09-14 19:11:07 | INFO | train rank=0 | step=5680/50000 progress=11.4% loss=0.115447 lr=9.889819e-05 lr_backbone=9.889819e-05 elapsed=05:24:39 avg_step=3.43s recent_step=4.55s eta=42:13:17 +2026-09-14 19:11:43 | INFO | train rank=0 | step=5690/50000 progress=11.4% loss=0.116874 lr=9.889128e-05 lr_backbone=9.889128e-05 elapsed=05:25:16 avg_step=3.43s recent_step=3.69s eta=42:13:03 +2026-09-14 19:12:23 | INFO | train rank=0 | step=5700/50000 progress=11.4% loss=0.139176 lr=9.888434e-05 lr_backbone=9.888434e-05 elapsed=05:25:56 avg_step=3.43s recent_step=3.94s eta=42:13:08 +2026-09-14 19:12:54 | INFO | train rank=0 | step=5710/50000 progress=11.4% loss=0.135372 lr=9.887739e-05 lr_backbone=9.887739e-05 elapsed=05:26:27 avg_step=3.43s recent_step=3.15s eta=42:12:12 +2026-09-14 19:13:41 | INFO | train rank=0 | step=5720/50000 progress=11.4% loss=0.115874 lr=9.887041e-05 lr_backbone=9.887041e-05 elapsed=05:27:14 avg_step=3.43s recent_step=4.64s eta=42:13:11 +2026-09-14 19:14:06 | INFO | train rank=0 | step=5730/50000 progress=11.5% loss=0.159203 lr=9.886341e-05 lr_backbone=9.886341e-05 elapsed=05:27:39 avg_step=3.43s recent_step=2.57s eta=42:11:30 +2026-09-14 19:14:31 | INFO | train rank=0 | step=5740/50000 progress=11.5% loss=0.145730 lr=9.885638e-05 lr_backbone=9.885638e-05 elapsed=05:28:03 avg_step=3.43s recent_step=2.43s eta=42:09:39 +2026-09-14 19:15:08 | INFO | train rank=0 | step=5750/50000 progress=11.5% loss=0.123435 lr=9.884934e-05 lr_backbone=9.884934e-05 elapsed=05:28:41 avg_step=3.43s recent_step=3.73s eta=42:09:28 +2026-09-14 19:15:50 | INFO | train rank=0 | step=5760/50000 progress=11.5% loss=0.151262 lr=9.884228e-05 lr_backbone=9.884228e-05 elapsed=05:29:23 avg_step=3.43s recent_step=4.19s eta=42:09:52 +2026-09-14 19:16:26 | INFO | train rank=0 | step=5770/50000 progress=11.5% loss=0.077086 lr=9.883519e-05 lr_backbone=9.883519e-05 elapsed=05:29:59 avg_step=3.43s recent_step=3.62s eta=42:09:32 +2026-09-14 19:16:57 | INFO | train rank=0 | step=5780/50000 progress=11.6% loss=0.123432 lr=9.882808e-05 lr_backbone=9.882808e-05 elapsed=05:30:29 avg_step=3.43s recent_step=3.05s eta=42:08:28 +2026-09-14 19:17:27 | INFO | train rank=0 | step=5790/50000 progress=11.6% loss=0.111846 lr=9.882096e-05 lr_backbone=9.882096e-05 elapsed=05:31:00 avg_step=3.43s recent_step=3.06s eta=42:07:26 +2026-09-14 19:18:10 | INFO | train rank=0 | step=5800/50000 progress=11.6% loss=0.129366 lr=9.881381e-05 lr_backbone=9.881381e-05 elapsed=05:31:43 avg_step=3.43s recent_step=4.32s eta=42:07:59 +2026-09-14 19:18:49 | INFO | train rank=0 | step=5810/50000 progress=11.6% loss=0.142901 lr=9.880663e-05 lr_backbone=9.880663e-05 elapsed=05:32:22 avg_step=3.43s recent_step=3.86s eta=42:07:57 +2026-09-14 19:19:22 | INFO | train rank=0 | step=5820/50000 progress=11.6% loss=0.153285 lr=9.879944e-05 lr_backbone=9.879944e-05 elapsed=05:32:54 avg_step=3.43s recent_step=3.27s eta=42:07:10 +2026-09-14 19:19:56 | INFO | train rank=0 | step=5830/50000 progress=11.7% loss=0.096761 lr=9.879223e-05 lr_backbone=9.879223e-05 elapsed=05:33:28 avg_step=3.43s recent_step=3.39s eta=42:06:33 +2026-09-14 19:20:44 | INFO | train rank=0 | step=5840/50000 progress=11.7% loss=0.094168 lr=9.878499e-05 lr_backbone=9.878499e-05 elapsed=05:34:17 avg_step=3.43s recent_step=4.83s eta=42:07:44 +2026-09-14 19:21:11 | INFO | train rank=0 | step=5850/50000 progress=11.7% loss=0.123176 lr=9.877774e-05 lr_backbone=9.877774e-05 elapsed=05:34:43 avg_step=3.43s recent_step=2.69s eta=42:06:13 +2026-09-14 19:21:45 | INFO | train rank=0 | step=5860/50000 progress=11.7% loss=0.119589 lr=9.877046e-05 lr_backbone=9.877046e-05 elapsed=05:35:18 avg_step=3.43s recent_step=3.42s eta=42:05:38 +2026-09-14 19:22:07 | INFO | train rank=0 | step=5870/50000 progress=11.7% loss=0.128627 lr=9.876316e-05 lr_backbone=9.876316e-05 elapsed=05:35:40 avg_step=3.43s recent_step=2.23s eta=42:03:33 +2026-09-14 19:22:54 | INFO | train rank=0 | step=5880/50000 progress=11.8% loss=0.110454 lr=9.875584e-05 lr_backbone=9.875584e-05 elapsed=05:36:27 avg_step=3.43s recent_step=4.66s eta=42:04:31 +2026-09-14 19:23:28 | INFO | train rank=0 | step=5890/50000 progress=11.8% loss=0.136618 lr=9.874850e-05 lr_backbone=9.874850e-05 elapsed=05:37:01 avg_step=3.43s recent_step=3.41s eta=42:03:55 +2026-09-14 19:24:10 | INFO | train rank=0 | step=5900/50000 progress=11.8% loss=0.077991 lr=9.874113e-05 lr_backbone=9.874113e-05 elapsed=05:37:42 avg_step=3.43s recent_step=4.17s eta=42:04:17 +2026-09-14 19:24:47 | INFO | train rank=0 | step=5910/50000 progress=11.8% loss=0.172271 lr=9.873375e-05 lr_backbone=9.873375e-05 elapsed=05:38:20 avg_step=3.43s recent_step=3.78s eta=42:04:08 +2026-09-14 19:25:17 | INFO | train rank=0 | step=5920/50000 progress=11.8% loss=0.093119 lr=9.872634e-05 lr_backbone=9.872634e-05 elapsed=05:38:49 avg_step=3.43s recent_step=2.91s eta=42:02:54 +2026-09-14 19:25:53 | INFO | train rank=0 | step=5930/50000 progress=11.9% loss=0.127362 lr=9.871892e-05 lr_backbone=9.871892e-05 elapsed=05:39:26 avg_step=3.43s recent_step=3.65s eta=42:02:36 +2026-09-14 19:26:30 | INFO | train rank=0 | step=5940/50000 progress=11.9% loss=0.160597 lr=9.871147e-05 lr_backbone=9.871147e-05 elapsed=05:40:03 avg_step=3.43s recent_step=3.72s eta=42:02:23 +2026-09-14 19:27:08 | INFO | train rank=0 | step=5950/50000 progress=11.9% loss=0.142682 lr=9.870400e-05 lr_backbone=9.870400e-05 elapsed=05:40:41 avg_step=3.44s recent_step=3.81s eta=42:02:17 +2026-09-14 19:27:46 | INFO | train rank=0 | step=5960/50000 progress=11.9% loss=0.127230 lr=9.869651e-05 lr_backbone=9.869651e-05 elapsed=05:41:19 avg_step=3.44s recent_step=3.80s eta=42:02:09 +2026-09-14 19:28:20 | INFO | train rank=0 | step=5970/50000 progress=11.9% loss=0.170612 lr=9.868899e-05 lr_backbone=9.868899e-05 elapsed=05:41:52 avg_step=3.44s recent_step=3.32s eta=42:01:26 +2026-09-14 19:29:06 | INFO | train rank=0 | step=5980/50000 progress=12.0% loss=0.096079 lr=9.868146e-05 lr_backbone=9.868146e-05 elapsed=05:42:39 avg_step=3.44s recent_step=4.64s eta=42:02:20 +2026-09-14 19:29:40 | INFO | train rank=0 | step=5990/50000 progress=12.0% loss=0.113928 lr=9.867391e-05 lr_backbone=9.867391e-05 elapsed=05:43:13 avg_step=3.44s recent_step=3.45s eta=42:01:47 +2026-09-14 19:30:24 | INFO | train rank=0 | step=6000/50000 progress=12.0% loss=0.140292 lr=9.866633e-05 lr_backbone=9.866633e-05 elapsed=05:43:57 avg_step=3.44s recent_step=4.33s eta=42:02:18 +2026-09-14 19:31:01 | INFO | train rank=0 | step=6010/50000 progress=12.0% loss=0.202893 lr=9.865873e-05 lr_backbone=9.865873e-05 elapsed=05:44:34 avg_step=3.44s recent_step=3.70s eta=42:02:03 +2026-09-14 19:31:35 | INFO | train rank=0 | step=6020/50000 progress=12.0% loss=0.131844 lr=9.865111e-05 lr_backbone=9.865111e-05 elapsed=05:45:08 avg_step=3.44s recent_step=3.44s eta=42:01:28 +2026-09-14 19:32:09 | INFO | train rank=0 | step=6030/50000 progress=12.1% loss=0.159206 lr=9.864347e-05 lr_backbone=9.864347e-05 elapsed=05:45:42 avg_step=3.44s recent_step=3.42s eta=42:00:52 +2026-09-14 19:32:53 | INFO | train rank=0 | step=6040/50000 progress=12.1% loss=0.134648 lr=9.863581e-05 lr_backbone=9.863581e-05 elapsed=05:46:26 avg_step=3.44s recent_step=4.39s eta=42:01:27 +2026-09-14 19:33:30 | INFO | train rank=0 | step=6050/50000 progress=12.1% loss=0.144877 lr=9.862813e-05 lr_backbone=9.862813e-05 elapsed=05:47:03 avg_step=3.44s recent_step=3.72s eta=42:01:13 +2026-09-14 19:34:06 | INFO | train rank=0 | step=6060/50000 progress=12.1% loss=0.153513 lr=9.862043e-05 lr_backbone=9.862043e-05 elapsed=05:47:39 avg_step=3.44s recent_step=3.55s eta=42:00:46 +2026-09-14 19:34:42 | INFO | train rank=0 | step=6070/50000 progress=12.1% loss=0.100965 lr=9.861270e-05 lr_backbone=9.861270e-05 elapsed=05:48:15 avg_step=3.44s recent_step=3.62s eta=42:00:25 +2026-09-14 19:35:15 | INFO | train rank=0 | step=6080/50000 progress=12.2% loss=0.127921 lr=9.860495e-05 lr_backbone=9.860495e-05 elapsed=05:48:47 avg_step=3.44s recent_step=3.24s eta=41:59:36 +2026-09-14 19:35:48 | INFO | train rank=0 | step=6090/50000 progress=12.2% loss=0.140859 lr=9.859719e-05 lr_backbone=9.859719e-05 elapsed=05:49:21 avg_step=3.44s recent_step=3.36s eta=41:58:56 +2026-09-14 19:36:24 | INFO | train rank=0 | step=6100/50000 progress=12.2% loss=0.107316 lr=9.858940e-05 lr_backbone=9.858940e-05 elapsed=05:49:57 avg_step=3.44s recent_step=3.62s eta=41:58:34 +2026-09-14 19:37:09 | INFO | train rank=0 | step=6110/50000 progress=12.2% loss=0.090544 lr=9.858159e-05 lr_backbone=9.858159e-05 elapsed=05:50:41 avg_step=3.44s recent_step=4.43s eta=41:59:10 +2026-09-14 19:37:40 | INFO | train rank=0 | step=6120/50000 progress=12.2% loss=0.123092 lr=9.857376e-05 lr_backbone=9.857376e-05 elapsed=05:51:13 avg_step=3.44s recent_step=3.14s eta=41:58:14 +2026-09-14 19:38:13 | INFO | train rank=0 | step=6130/50000 progress=12.3% loss=0.121180 lr=9.856590e-05 lr_backbone=9.856590e-05 elapsed=05:51:45 avg_step=3.44s recent_step=3.26s eta=41:57:26 +2026-09-14 19:38:56 | INFO | train rank=0 | step=6140/50000 progress=12.3% loss=0.121822 lr=9.855803e-05 lr_backbone=9.855803e-05 elapsed=05:52:29 avg_step=3.44s recent_step=4.35s eta=41:57:57 +2026-09-14 19:39:29 | INFO | train rank=0 | step=6150/50000 progress=12.3% loss=0.133765 lr=9.855013e-05 lr_backbone=9.855013e-05 elapsed=05:53:02 avg_step=3.44s recent_step=3.31s eta=41:57:13 +2026-09-14 19:40:03 | INFO | train rank=0 | step=6160/50000 progress=12.3% loss=0.087149 lr=9.854222e-05 lr_backbone=9.854222e-05 elapsed=05:53:35 avg_step=3.44s recent_step=3.34s eta=41:56:31 +2026-09-14 19:40:35 | INFO | train rank=0 | step=6170/50000 progress=12.3% loss=0.133064 lr=9.853428e-05 lr_backbone=9.853428e-05 elapsed=05:54:08 avg_step=3.44s recent_step=3.22s eta=41:55:40 +2026-09-14 19:41:11 | INFO | train rank=0 | step=6180/50000 progress=12.4% loss=0.118487 lr=9.852632e-05 lr_backbone=9.852632e-05 elapsed=05:54:44 avg_step=3.44s recent_step=3.63s eta=41:55:19 +2026-09-14 19:42:02 | INFO | train rank=0 | step=6190/50000 progress=12.4% loss=0.138490 lr=9.851834e-05 lr_backbone=9.851834e-05 elapsed=05:55:35 avg_step=3.45s recent_step=5.11s eta=41:56:43 +2026-09-14 19:42:34 | INFO | train rank=0 | step=6200/50000 progress=12.4% loss=0.120459 lr=9.851034e-05 lr_backbone=9.851034e-05 elapsed=05:56:07 avg_step=3.45s recent_step=3.21s eta=41:55:51 +2026-09-14 19:43:14 | INFO | train rank=0 | step=6210/50000 progress=12.4% loss=0.126584 lr=9.850232e-05 lr_backbone=9.850232e-05 elapsed=05:56:46 avg_step=3.45s recent_step=3.92s eta=41:55:50 +2026-09-14 19:43:56 | INFO | train rank=0 | step=6220/50000 progress=12.4% loss=0.110355 lr=9.849427e-05 lr_backbone=9.849427e-05 elapsed=05:57:29 avg_step=3.45s recent_step=4.24s eta=41:56:12 +2026-09-14 19:44:33 | INFO | train rank=0 | step=6230/50000 progress=12.5% loss=0.197728 lr=9.848621e-05 lr_backbone=9.848621e-05 elapsed=05:58:05 avg_step=3.45s recent_step=3.67s eta=41:55:53 +2026-09-14 19:45:06 | INFO | train rank=0 | step=6240/50000 progress=12.5% loss=0.139740 lr=9.847812e-05 lr_backbone=9.847812e-05 elapsed=05:58:38 avg_step=3.45s recent_step=3.29s eta=41:55:07 +2026-09-14 19:45:44 | INFO | train rank=0 | step=6250/50000 progress=12.5% loss=0.119267 lr=9.847001e-05 lr_backbone=9.847001e-05 elapsed=05:59:17 avg_step=3.45s recent_step=3.84s eta=41:55:00 +2026-09-14 19:46:19 | INFO | train rank=0 | step=6260/50000 progress=12.5% loss=0.123980 lr=9.846188e-05 lr_backbone=9.846188e-05 elapsed=05:59:52 avg_step=3.45s recent_step=3.51s eta=41:54:30 +2026-09-14 19:47:15 | INFO | train rank=0 | step=6270/50000 progress=12.5% loss=0.126771 lr=9.845373e-05 lr_backbone=9.845373e-05 elapsed=06:00:48 avg_step=3.45s recent_step=5.56s eta=41:56:23 +2026-09-14 19:47:51 | INFO | train rank=0 | step=6280/50000 progress=12.6% loss=0.103099 lr=9.844556e-05 lr_backbone=9.844556e-05 elapsed=06:01:24 avg_step=3.45s recent_step=3.65s eta=41:56:02 +2026-09-14 19:48:26 | INFO | train rank=0 | step=6290/50000 progress=12.6% loss=0.134098 lr=9.843737e-05 lr_backbone=9.843737e-05 elapsed=06:01:59 avg_step=3.45s recent_step=3.49s eta=41:55:31 +2026-09-14 19:49:03 | INFO | train rank=0 | step=6300/50000 progress=12.6% loss=0.085081 lr=9.842916e-05 lr_backbone=9.842916e-05 elapsed=06:02:36 avg_step=3.45s recent_step=3.67s eta=41:55:11 +2026-09-14 19:49:51 | INFO | train rank=0 | step=6310/50000 progress=12.6% loss=0.109210 lr=9.842092e-05 lr_backbone=9.842092e-05 elapsed=06:03:24 avg_step=3.46s recent_step=4.81s eta=41:56:10 +2026-09-14 19:50:26 | INFO | train rank=0 | step=6320/50000 progress=12.6% loss=0.164917 lr=9.841267e-05 lr_backbone=9.841267e-05 elapsed=06:03:59 avg_step=3.46s recent_step=3.50s eta=41:55:39 +2026-09-14 19:50:59 | INFO | train rank=0 | step=6330/50000 progress=12.7% loss=0.133108 lr=9.840439e-05 lr_backbone=9.840439e-05 elapsed=06:04:32 avg_step=3.46s recent_step=3.29s eta=41:54:53 +2026-09-14 19:51:28 | INFO | train rank=0 | step=6340/50000 progress=12.7% loss=0.162650 lr=9.839609e-05 lr_backbone=9.839609e-05 elapsed=06:05:01 avg_step=3.45s recent_step=2.96s eta=41:53:44 +2026-09-14 19:52:17 | INFO | train rank=0 | step=6350/50000 progress=12.7% loss=0.149679 lr=9.838777e-05 lr_backbone=9.838777e-05 elapsed=06:05:50 avg_step=3.46s recent_step=4.85s eta=41:54:45 +2026-09-14 19:52:52 | INFO | train rank=0 | step=6360/50000 progress=12.7% loss=0.104792 lr=9.837943e-05 lr_backbone=9.837943e-05 elapsed=06:06:25 avg_step=3.46s recent_step=3.49s eta=41:54:13 +2026-09-14 19:53:26 | INFO | train rank=0 | step=6370/50000 progress=12.7% loss=0.127964 lr=9.837107e-05 lr_backbone=9.837107e-05 elapsed=06:06:59 avg_step=3.46s recent_step=3.46s eta=41:53:38 +2026-09-14 19:54:02 | INFO | train rank=0 | step=6380/50000 progress=12.8% loss=0.129990 lr=9.836269e-05 lr_backbone=9.836269e-05 elapsed=06:07:35 avg_step=3.46s recent_step=3.55s eta=41:53:10 +2026-09-14 19:54:51 | INFO | train rank=0 | step=6390/50000 progress=12.8% loss=0.119241 lr=9.835428e-05 lr_backbone=9.835428e-05 elapsed=06:08:24 avg_step=3.46s recent_step=4.95s eta=41:54:18 +2026-09-14 19:55:27 | INFO | train rank=0 | step=6400/50000 progress=12.8% loss=0.126697 lr=9.834586e-05 lr_backbone=9.834586e-05 elapsed=06:08:59 avg_step=3.46s recent_step=3.52s eta=41:53:47 +2026-09-14 19:55:58 | INFO | train rank=0 | step=6410/50000 progress=12.8% loss=0.094944 lr=9.833741e-05 lr_backbone=9.833741e-05 elapsed=06:09:31 avg_step=3.46s recent_step=3.13s eta=41:52:51 +2026-09-14 19:56:23 | INFO | train rank=0 | step=6420/50000 progress=12.8% loss=0.156019 lr=9.832895e-05 lr_backbone=9.832895e-05 elapsed=06:09:56 avg_step=3.46s recent_step=2.50s eta=41:51:11 +2026-09-14 19:57:17 | INFO | train rank=0 | step=6430/50000 progress=12.9% loss=0.120983 lr=9.832046e-05 lr_backbone=9.832046e-05 elapsed=06:10:49 avg_step=3.46s recent_step=5.36s eta=41:52:46 +2026-09-14 19:57:51 | INFO | train rank=0 | step=6440/50000 progress=12.9% loss=0.111780 lr=9.831195e-05 lr_backbone=9.831195e-05 elapsed=06:11:24 avg_step=3.46s recent_step=3.43s eta=41:52:09 +2026-09-14 19:58:17 | INFO | train rank=0 | step=6450/50000 progress=12.9% loss=0.142764 lr=9.830342e-05 lr_backbone=9.830342e-05 elapsed=06:11:50 avg_step=3.46s recent_step=2.61s eta=41:50:37 +2026-09-14 19:58:49 | INFO | train rank=0 | step=6460/50000 progress=12.9% loss=0.126592 lr=9.829486e-05 lr_backbone=9.829486e-05 elapsed=06:12:22 avg_step=3.46s recent_step=3.24s eta=41:49:48 +2026-09-14 19:59:33 | INFO | train rank=0 | step=6470/50000 progress=12.9% loss=0.144441 lr=9.828629e-05 lr_backbone=9.828629e-05 elapsed=06:13:06 avg_step=3.46s recent_step=4.33s eta=41:50:12 +2026-09-14 20:00:07 | INFO | train rank=0 | step=6480/50000 progress=13.0% loss=0.096958 lr=9.827770e-05 lr_backbone=9.827770e-05 elapsed=06:13:40 avg_step=3.46s recent_step=3.44s eta=41:49:37 +2026-09-14 20:00:42 | INFO | train rank=0 | step=6490/50000 progress=13.0% loss=0.106233 lr=9.826908e-05 lr_backbone=9.826908e-05 elapsed=06:14:15 avg_step=3.46s recent_step=3.53s eta=41:49:07 +2026-09-14 20:01:14 | INFO | train rank=0 | step=6500/50000 progress=13.0% loss=0.114684 lr=9.826045e-05 lr_backbone=9.826045e-05 elapsed=06:14:47 avg_step=3.46s recent_step=3.14s eta=41:48:11 +2026-09-14 20:01:51 | INFO | train rank=0 | step=6510/50000 progress=13.0% loss=0.136052 lr=9.825179e-05 lr_backbone=9.825179e-05 elapsed=06:15:23 avg_step=3.46s recent_step=3.67s eta=41:47:50 +2026-09-14 20:02:28 | INFO | train rank=0 | step=6520/50000 progress=13.0% loss=0.190782 lr=9.824311e-05 lr_backbone=9.824311e-05 elapsed=06:16:01 avg_step=3.46s recent_step=3.72s eta=41:47:33 +2026-09-14 20:03:04 | INFO | train rank=0 | step=6530/50000 progress=13.1% loss=0.124224 lr=9.823441e-05 lr_backbone=9.823441e-05 elapsed=06:16:37 avg_step=3.46s recent_step=3.60s eta=41:47:08 +2026-09-14 20:03:36 | INFO | train rank=0 | step=6540/50000 progress=13.1% loss=0.111803 lr=9.822569e-05 lr_backbone=9.822569e-05 elapsed=06:17:09 avg_step=3.46s recent_step=3.19s eta=41:46:15 +2026-09-14 20:04:13 | INFO | train rank=0 | step=6550/50000 progress=13.1% loss=0.095776 lr=9.821695e-05 lr_backbone=9.821695e-05 elapsed=06:17:45 avg_step=3.46s recent_step=3.69s eta=41:45:56 +2026-09-14 20:04:45 | INFO | train rank=0 | step=6560/50000 progress=13.1% loss=0.086688 lr=9.820818e-05 lr_backbone=9.820818e-05 elapsed=06:18:18 avg_step=3.46s recent_step=3.21s eta=41:45:05 +2026-09-14 20:05:18 | INFO | train rank=0 | step=6570/50000 progress=13.1% loss=0.148439 lr=9.819940e-05 lr_backbone=9.819940e-05 elapsed=06:18:51 avg_step=3.46s recent_step=3.32s eta=41:44:21 +2026-09-14 20:05:49 | INFO | train rank=0 | step=6580/50000 progress=13.2% loss=0.101996 lr=9.819059e-05 lr_backbone=9.819059e-05 elapsed=06:19:22 avg_step=3.46s recent_step=3.14s eta=41:43:25 +2026-09-14 20:06:35 | INFO | train rank=0 | step=6590/50000 progress=13.2% loss=0.141108 lr=9.818177e-05 lr_backbone=9.818177e-05 elapsed=06:20:08 avg_step=3.46s recent_step=4.58s eta=41:44:05 +2026-09-14 20:07:02 | INFO | train rank=0 | step=6600/50000 progress=13.2% loss=0.152195 lr=9.817292e-05 lr_backbone=9.817292e-05 elapsed=06:20:35 avg_step=3.46s recent_step=2.72s eta=41:42:41 +2026-09-14 20:07:33 | INFO | train rank=0 | step=6610/50000 progress=13.2% loss=0.138360 lr=9.816405e-05 lr_backbone=9.816405e-05 elapsed=06:21:05 avg_step=3.46s recent_step=3.02s eta=41:41:38 +2026-09-14 20:08:11 | INFO | train rank=0 | step=6620/50000 progress=13.2% loss=0.111111 lr=9.815516e-05 lr_backbone=9.815516e-05 elapsed=06:21:43 avg_step=3.46s recent_step=3.80s eta=41:41:26 +2026-09-14 20:08:55 | INFO | train rank=0 | step=6630/50000 progress=13.3% loss=0.092182 lr=9.814625e-05 lr_backbone=9.814625e-05 elapsed=06:22:28 avg_step=3.46s recent_step=4.42s eta=41:41:54 +2026-09-14 20:09:33 | INFO | train rank=0 | step=6640/50000 progress=13.3% loss=0.152516 lr=9.813732e-05 lr_backbone=9.813732e-05 elapsed=06:23:06 avg_step=3.46s recent_step=3.86s eta=41:41:46 +2026-09-14 20:10:04 | INFO | train rank=0 | step=6650/50000 progress=13.3% loss=0.090000 lr=9.812837e-05 lr_backbone=9.812837e-05 elapsed=06:23:37 avg_step=3.46s recent_step=3.06s eta=41:40:45 +2026-09-14 20:10:28 | INFO | train rank=0 | step=6660/50000 progress=13.3% loss=0.120839 lr=9.811939e-05 lr_backbone=9.811939e-05 elapsed=06:24:01 avg_step=3.46s recent_step=2.40s eta=41:39:01 +2026-09-14 20:11:05 | INFO | train rank=0 | step=6670/50000 progress=13.3% loss=0.126796 lr=9.811040e-05 lr_backbone=9.811040e-05 elapsed=06:24:38 avg_step=3.46s recent_step=3.68s eta=41:38:40 +2026-09-14 20:11:56 | INFO | train rank=0 | step=6680/50000 progress=13.4% loss=0.119814 lr=9.810138e-05 lr_backbone=9.810138e-05 elapsed=06:25:29 avg_step=3.46s recent_step=5.17s eta=41:39:57 +2026-09-14 20:12:29 | INFO | train rank=0 | step=6690/50000 progress=13.4% loss=0.103179 lr=9.809235e-05 lr_backbone=9.809235e-05 elapsed=06:26:01 avg_step=3.46s recent_step=3.22s eta=41:39:06 +2026-09-14 20:12:59 | INFO | train rank=0 | step=6700/50000 progress=13.4% loss=0.126730 lr=9.808329e-05 lr_backbone=9.808329e-05 elapsed=06:26:32 avg_step=3.46s recent_step=3.08s eta=41:38:07 +2026-09-14 20:13:42 | INFO | train rank=0 | step=6710/50000 progress=13.4% loss=0.121113 lr=9.807421e-05 lr_backbone=9.807421e-05 elapsed=06:27:15 avg_step=3.46s recent_step=4.30s eta=41:38:27 +2026-09-14 20:14:15 | INFO | train rank=0 | step=6720/50000 progress=13.4% loss=0.112845 lr=9.806511e-05 lr_backbone=9.806511e-05 elapsed=06:27:48 avg_step=3.46s recent_step=3.24s eta=41:37:38 +2026-09-14 20:14:53 | INFO | train rank=0 | step=6730/50000 progress=13.5% loss=0.115043 lr=9.805599e-05 lr_backbone=9.805599e-05 elapsed=06:28:25 avg_step=3.46s recent_step=3.78s eta=41:37:24 +2026-09-14 20:15:26 | INFO | train rank=0 | step=6740/50000 progress=13.5% loss=0.134901 lr=9.804685e-05 lr_backbone=9.804685e-05 elapsed=06:28:58 avg_step=3.46s recent_step=3.28s eta=41:36:37 +2026-09-14 20:16:11 | INFO | train rank=0 | step=6750/50000 progress=13.5% loss=0.098375 lr=9.803768e-05 lr_backbone=9.803768e-05 elapsed=06:29:43 avg_step=3.46s recent_step=4.50s eta=41:37:09 +2026-09-14 20:16:48 | INFO | train rank=0 | step=6760/50000 progress=13.5% loss=0.121179 lr=9.802850e-05 lr_backbone=9.802850e-05 elapsed=06:30:21 avg_step=3.46s recent_step=3.79s eta=41:36:55 +2026-09-14 20:17:23 | INFO | train rank=0 | step=6770/50000 progress=13.5% loss=0.114653 lr=9.801929e-05 lr_backbone=9.801929e-05 elapsed=06:30:56 avg_step=3.46s recent_step=3.50s eta=41:36:23 +2026-09-14 20:18:03 | INFO | train rank=0 | step=6780/50000 progress=13.6% loss=0.104765 lr=9.801007e-05 lr_backbone=9.801007e-05 elapsed=06:31:36 avg_step=3.47s recent_step=4.00s eta=41:36:22 +2026-09-14 20:18:42 | INFO | train rank=0 | step=6790/50000 progress=13.6% loss=0.090398 lr=9.800082e-05 lr_backbone=9.800082e-05 elapsed=06:32:15 avg_step=3.47s recent_step=3.91s eta=41:36:16 +2026-09-14 20:19:21 | INFO | train rank=0 | step=6800/50000 progress=13.6% loss=0.115757 lr=9.799155e-05 lr_backbone=9.799155e-05 elapsed=06:32:54 avg_step=3.47s recent_step=3.89s eta=41:36:08 +2026-09-14 20:19:57 | INFO | train rank=0 | step=6810/50000 progress=13.6% loss=0.107459 lr=9.798226e-05 lr_backbone=9.798226e-05 elapsed=06:33:30 avg_step=3.47s recent_step=3.55s eta=41:35:38 +2026-09-14 20:20:28 | INFO | train rank=0 | step=6820/50000 progress=13.6% loss=0.111214 lr=9.797295e-05 lr_backbone=9.797295e-05 elapsed=06:34:01 avg_step=3.47s recent_step=3.11s eta=41:34:41 +2026-09-14 20:21:12 | INFO | train rank=0 | step=6830/50000 progress=13.7% loss=0.091059 lr=9.796362e-05 lr_backbone=9.796362e-05 elapsed=06:34:45 avg_step=3.47s recent_step=4.39s eta=41:35:05 +2026-09-14 20:21:49 | INFO | train rank=0 | step=6840/50000 progress=13.7% loss=0.120872 lr=9.795427e-05 lr_backbone=9.795427e-05 elapsed=06:35:22 avg_step=3.47s recent_step=3.73s eta=41:34:47 +2026-09-14 20:22:21 | INFO | train rank=0 | step=6850/50000 progress=13.7% loss=0.132110 lr=9.794490e-05 lr_backbone=9.794490e-05 elapsed=06:35:54 avg_step=3.47s recent_step=3.16s eta=41:33:53 +2026-09-14 20:22:55 | INFO | train rank=0 | step=6860/50000 progress=13.7% loss=0.151908 lr=9.793550e-05 lr_backbone=9.793550e-05 elapsed=06:36:28 avg_step=3.47s recent_step=3.46s eta=41:33:18 +2026-09-14 20:23:44 | INFO | train rank=0 | step=6870/50000 progress=13.7% loss=0.123946 lr=9.792609e-05 lr_backbone=9.792609e-05 elapsed=06:37:17 avg_step=3.47s recent_step=4.87s eta=41:34:11 +2026-09-14 20:24:15 | INFO | train rank=0 | step=6880/50000 progress=13.8% loss=0.113096 lr=9.791665e-05 lr_backbone=9.791665e-05 elapsed=06:37:48 avg_step=3.47s recent_step=3.12s eta=41:33:15 +2026-09-14 20:24:42 | INFO | train rank=0 | step=6890/50000 progress=13.8% loss=0.137079 lr=9.790720e-05 lr_backbone=9.790720e-05 elapsed=06:38:14 avg_step=3.47s recent_step=2.61s eta=41:31:47 +2026-09-14 20:25:32 | INFO | train rank=0 | step=6900/50000 progress=13.8% loss=0.095881 lr=9.789772e-05 lr_backbone=9.789772e-05 elapsed=06:39:04 avg_step=3.47s recent_step=5.01s eta=41:32:49 +2026-09-14 20:26:10 | INFO | train rank=0 | step=6910/50000 progress=13.8% loss=0.116547 lr=9.788822e-05 lr_backbone=9.788822e-05 elapsed=06:39:43 avg_step=3.47s recent_step=3.83s eta=41:32:36 +2026-09-14 20:26:41 | INFO | train rank=0 | step=6920/50000 progress=13.8% loss=0.161458 lr=9.787870e-05 lr_backbone=9.787870e-05 elapsed=06:40:13 avg_step=3.47s recent_step=3.07s eta=41:31:37 +2026-09-14 20:27:12 | INFO | train rank=0 | step=6930/50000 progress=13.9% loss=0.174193 lr=9.786916e-05 lr_backbone=9.786916e-05 elapsed=06:40:45 avg_step=3.47s recent_step=3.14s eta=41:30:41 +2026-09-14 20:28:15 | INFO | train rank=0 | step=6940/50000 progress=13.9% loss=0.108037 lr=9.785960e-05 lr_backbone=9.785960e-05 elapsed=06:41:48 avg_step=3.47s recent_step=6.30s eta=41:33:02 +2026-09-14 20:28:45 | INFO | train rank=0 | step=6950/50000 progress=13.9% loss=0.112713 lr=9.785002e-05 lr_backbone=9.785002e-05 elapsed=06:42:18 avg_step=3.47s recent_step=2.98s eta=41:31:57 +2026-09-14 20:29:17 | INFO | train rank=0 | step=6960/50000 progress=13.9% loss=0.108975 lr=9.784041e-05 lr_backbone=9.784041e-05 elapsed=06:42:50 avg_step=3.47s recent_step=3.20s eta=41:31:05 +2026-09-14 20:29:54 | INFO | train rank=0 | step=6970/50000 progress=13.9% loss=0.124566 lr=9.783079e-05 lr_backbone=9.783079e-05 elapsed=06:43:26 avg_step=3.47s recent_step=3.68s eta=41:30:43 +2026-09-14 20:30:39 | INFO | train rank=0 | step=6980/50000 progress=14.0% loss=0.135543 lr=9.782114e-05 lr_backbone=9.782114e-05 elapsed=06:44:12 avg_step=3.47s recent_step=4.55s eta=41:31:15 +2026-09-14 20:31:16 | INFO | train rank=0 | step=6990/50000 progress=14.0% loss=0.115021 lr=9.781148e-05 lr_backbone=9.781148e-05 elapsed=06:44:49 avg_step=3.47s recent_step=3.70s eta=41:30:54 +2026-09-14 20:31:47 | INFO | train rank=0 | step=7000/50000 progress=14.0% loss=0.137735 lr=9.780179e-05 lr_backbone=9.780179e-05 elapsed=06:45:20 avg_step=3.47s recent_step=3.10s eta=41:29:56 +2026-09-14 20:32:21 | INFO | train rank=0 | step=7010/50000 progress=14.0% loss=0.108268 lr=9.779208e-05 lr_backbone=9.779208e-05 elapsed=06:45:53 avg_step=3.47s recent_step=3.36s eta=41:29:14 +2026-09-14 20:33:04 | INFO | train rank=0 | step=7020/50000 progress=14.0% loss=0.109232 lr=9.778235e-05 lr_backbone=9.778235e-05 elapsed=06:46:37 avg_step=3.48s recent_step=4.31s eta=41:29:31 +2026-09-14 20:33:35 | INFO | train rank=0 | step=7030/50000 progress=14.1% loss=0.153982 lr=9.777260e-05 lr_backbone=9.777260e-05 elapsed=06:47:08 avg_step=3.47s recent_step=3.11s eta=41:28:34 +2026-09-14 20:34:11 | INFO | train rank=0 | step=7040/50000 progress=14.1% loss=0.148574 lr=9.776283e-05 lr_backbone=9.776283e-05 elapsed=06:47:44 avg_step=3.48s recent_step=3.62s eta=41:28:08 +2026-09-14 20:34:40 | INFO | train rank=0 | step=7050/50000 progress=14.1% loss=0.124494 lr=9.775304e-05 lr_backbone=9.775304e-05 elapsed=06:48:13 avg_step=3.47s recent_step=2.88s eta=41:26:57 +2026-09-14 20:35:24 | INFO | train rank=0 | step=7060/50000 progress=14.1% loss=0.132302 lr=9.774323e-05 lr_backbone=9.774323e-05 elapsed=06:48:57 avg_step=3.48s recent_step=4.42s eta=41:27:19 +2026-09-14 20:35:54 | INFO | train rank=0 | step=7070/50000 progress=14.1% loss=0.116718 lr=9.773339e-05 lr_backbone=9.773339e-05 elapsed=06:49:26 avg_step=3.47s recent_step=2.95s eta=41:26:13 +2026-09-14 20:36:28 | INFO | train rank=0 | step=7080/50000 progress=14.2% loss=0.126417 lr=9.772354e-05 lr_backbone=9.772354e-05 elapsed=06:50:01 avg_step=3.47s recent_step=3.45s eta=41:25:37 +2026-09-14 20:37:10 | INFO | train rank=0 | step=7090/50000 progress=14.2% loss=0.124258 lr=9.771366e-05 lr_backbone=9.771366e-05 elapsed=06:50:43 avg_step=3.48s recent_step=4.21s eta=41:25:46 +2026-09-14 20:37:48 | INFO | train rank=0 | step=7100/50000 progress=14.2% loss=0.147610 lr=9.770377e-05 lr_backbone=9.770377e-05 elapsed=06:51:20 avg_step=3.48s recent_step=3.74s eta=41:25:27 +2026-09-14 20:38:31 | INFO | train rank=0 | step=7110/50000 progress=14.2% loss=0.069762 lr=9.769385e-05 lr_backbone=9.769385e-05 elapsed=06:52:04 avg_step=3.48s recent_step=4.39s eta=41:25:48 +2026-09-14 20:39:00 | INFO | train rank=0 | step=7120/50000 progress=14.2% loss=0.113932 lr=9.768391e-05 lr_backbone=9.768391e-05 elapsed=06:52:33 avg_step=3.48s recent_step=2.84s eta=41:24:34 +2026-09-14 20:39:31 | INFO | train rank=0 | step=7130/50000 progress=14.3% loss=0.122691 lr=9.767395e-05 lr_backbone=9.767395e-05 elapsed=06:53:04 avg_step=3.48s recent_step=3.10s eta=41:23:37 +2026-09-14 20:40:09 | INFO | train rank=0 | step=7140/50000 progress=14.3% loss=0.142374 lr=9.766398e-05 lr_backbone=9.766398e-05 elapsed=06:53:42 avg_step=3.48s recent_step=3.82s eta=41:23:23 +2026-09-14 20:40:52 | INFO | train rank=0 | step=7150/50000 progress=14.3% loss=0.107340 lr=9.765398e-05 lr_backbone=9.765398e-05 elapsed=06:54:25 avg_step=3.48s recent_step=4.32s eta=41:23:39 +2026-09-14 20:41:30 | INFO | train rank=0 | step=7160/50000 progress=14.3% loss=0.108672 lr=9.764395e-05 lr_backbone=9.764395e-05 elapsed=06:55:03 avg_step=3.48s recent_step=3.80s eta=41:23:23 +2026-09-14 20:42:10 | INFO | train rank=0 | step=7170/50000 progress=14.3% loss=0.116365 lr=9.763391e-05 lr_backbone=9.763391e-05 elapsed=06:55:43 avg_step=3.48s recent_step=3.96s eta=41:23:17 +2026-09-14 20:42:47 | INFO | train rank=0 | step=7180/50000 progress=14.4% loss=0.130496 lr=9.762385e-05 lr_backbone=9.762385e-05 elapsed=06:56:20 avg_step=3.48s recent_step=3.69s eta=41:22:55 +2026-09-14 20:43:39 | INFO | train rank=0 | step=7190/50000 progress=14.4% loss=0.122419 lr=9.761377e-05 lr_backbone=9.761377e-05 elapsed=06:57:11 avg_step=3.48s recent_step=5.19s eta=41:24:02 +2026-09-14 20:44:09 | INFO | train rank=0 | step=7200/50000 progress=14.4% loss=0.124893 lr=9.760366e-05 lr_backbone=9.760366e-05 elapsed=06:57:42 avg_step=3.48s recent_step=3.01s eta=41:22:59 +2026-09-14 20:44:41 | INFO | train rank=0 | step=7210/50000 progress=14.4% loss=0.141238 lr=9.759354e-05 lr_backbone=9.759354e-05 elapsed=06:58:14 avg_step=3.48s recent_step=3.27s eta=41:22:12 +2026-09-14 20:45:14 | INFO | train rank=0 | step=7220/50000 progress=14.4% loss=0.116696 lr=9.758339e-05 lr_backbone=9.758339e-05 elapsed=06:58:47 avg_step=3.48s recent_step=3.29s eta=41:21:26 +2026-09-14 20:45:55 | INFO | train rank=0 | step=7230/50000 progress=14.5% loss=0.154721 lr=9.757322e-05 lr_backbone=9.757322e-05 elapsed=06:59:28 avg_step=3.48s recent_step=4.07s eta=41:21:26 +2026-09-14 20:46:22 | INFO | train rank=0 | step=7240/50000 progress=14.5% loss=0.134994 lr=9.756303e-05 lr_backbone=9.756303e-05 elapsed=06:59:55 avg_step=3.48s recent_step=2.72s eta=41:20:06 +2026-09-14 20:46:55 | INFO | train rank=0 | step=7250/50000 progress=14.5% loss=0.115730 lr=9.755283e-05 lr_backbone=9.755283e-05 elapsed=07:00:28 avg_step=3.48s recent_step=3.28s eta=41:19:19 +2026-09-14 20:47:36 | INFO | train rank=0 | step=7260/50000 progress=14.5% loss=0.118864 lr=9.754260e-05 lr_backbone=9.754260e-05 elapsed=07:01:08 avg_step=3.48s recent_step=4.07s eta=41:19:19 +2026-09-14 20:48:31 | INFO | train rank=0 | step=7270/50000 progress=14.5% loss=0.119419 lr=9.753235e-05 lr_backbone=9.753235e-05 elapsed=07:02:03 avg_step=3.48s recent_step=5.50s eta=41:20:43 +2026-09-14 20:48:59 | INFO | train rank=0 | step=7280/50000 progress=14.6% loss=0.167732 lr=9.752208e-05 lr_backbone=9.752208e-05 elapsed=07:02:32 avg_step=3.48s recent_step=2.81s eta=41:19:29 +2026-09-14 20:49:34 | INFO | train rank=0 | step=7290/50000 progress=14.6% loss=0.133932 lr=9.751178e-05 lr_backbone=9.751178e-05 elapsed=07:03:07 avg_step=3.48s recent_step=3.54s eta=41:18:57 +2026-09-14 20:50:04 | INFO | train rank=0 | step=7300/50000 progress=14.6% loss=0.104590 lr=9.750147e-05 lr_backbone=9.750147e-05 elapsed=07:03:37 avg_step=3.48s recent_step=2.98s eta=41:17:53 +2026-09-14 20:51:00 | INFO | train rank=0 | step=7310/50000 progress=14.6% loss=0.133223 lr=9.749114e-05 lr_backbone=9.749114e-05 elapsed=07:04:33 avg_step=3.48s recent_step=5.61s eta=41:19:23 +2026-09-14 20:51:25 | INFO | train rank=0 | step=7320/50000 progress=14.6% loss=0.081002 lr=9.748078e-05 lr_backbone=9.748078e-05 elapsed=07:04:58 avg_step=3.48s recent_step=2.47s eta=41:17:49 +2026-09-14 20:51:59 | INFO | train rank=0 | step=7330/50000 progress=14.7% loss=0.114891 lr=9.747041e-05 lr_backbone=9.747041e-05 elapsed=07:05:32 avg_step=3.48s recent_step=3.45s eta=41:17:12 +2026-09-14 20:52:36 | INFO | train rank=0 | step=7340/50000 progress=14.7% loss=0.105607 lr=9.746001e-05 lr_backbone=9.746001e-05 elapsed=07:06:08 avg_step=3.48s recent_step=3.62s eta=41:16:45 +2026-09-14 20:53:16 | INFO | train rank=0 | step=7350/50000 progress=14.7% loss=0.111186 lr=9.744960e-05 lr_backbone=9.744960e-05 elapsed=07:06:49 avg_step=3.48s recent_step=4.04s eta=41:16:42 +2026-09-14 20:53:49 | INFO | train rank=0 | step=7360/50000 progress=14.7% loss=0.109435 lr=9.743916e-05 lr_backbone=9.743916e-05 elapsed=07:07:22 avg_step=3.48s recent_step=3.29s eta=41:15:56 +2026-09-14 20:54:24 | INFO | train rank=0 | step=7370/50000 progress=14.7% loss=0.163938 lr=9.742870e-05 lr_backbone=9.742870e-05 elapsed=07:07:57 avg_step=3.48s recent_step=3.52s eta=41:15:23 +2026-09-14 20:54:59 | INFO | train rank=0 | step=7380/50000 progress=14.8% loss=0.131717 lr=9.741822e-05 lr_backbone=9.741822e-05 elapsed=07:08:32 avg_step=3.48s recent_step=3.52s eta=41:14:50 +2026-09-14 20:55:49 | INFO | train rank=0 | step=7390/50000 progress=14.8% loss=0.094599 lr=9.740772e-05 lr_backbone=9.740772e-05 elapsed=07:09:22 avg_step=3.49s recent_step=5.01s eta=41:15:43 +2026-09-14 20:56:25 | INFO | train rank=0 | step=7400/50000 progress=14.8% loss=0.139836 lr=9.739720e-05 lr_backbone=9.739720e-05 elapsed=07:09:58 avg_step=3.49s recent_step=3.58s eta=41:15:13 +2026-09-14 20:56:55 | INFO | train rank=0 | step=7410/50000 progress=14.8% loss=0.144922 lr=9.738666e-05 lr_backbone=9.738666e-05 elapsed=07:10:27 avg_step=3.49s recent_step=2.97s eta=41:14:09 +2026-09-14 20:57:30 | INFO | train rank=0 | step=7420/50000 progress=14.8% loss=0.129372 lr=9.737610e-05 lr_backbone=9.737610e-05 elapsed=07:11:03 avg_step=3.49s recent_step=3.51s eta=41:13:36 +2026-09-14 20:58:15 | INFO | train rank=0 | step=7430/50000 progress=14.9% loss=0.144833 lr=9.736552e-05 lr_backbone=9.736552e-05 elapsed=07:11:48 avg_step=3.49s recent_step=4.56s eta=41:14:03 +2026-09-14 20:58:45 | INFO | train rank=0 | step=7440/50000 progress=14.9% loss=0.094949 lr=9.735492e-05 lr_backbone=9.735492e-05 elapsed=07:12:18 avg_step=3.49s recent_step=2.97s eta=41:12:58 +2026-09-14 20:59:15 | INFO | train rank=0 | step=7450/50000 progress=14.9% loss=0.102595 lr=9.734429e-05 lr_backbone=9.734429e-05 elapsed=07:12:48 avg_step=3.49s recent_step=3.01s eta=41:11:56 +2026-09-14 20:59:43 | INFO | train rank=0 | step=7460/50000 progress=14.9% loss=0.132987 lr=9.733365e-05 lr_backbone=9.733365e-05 elapsed=07:13:16 avg_step=3.48s recent_step=2.78s eta=41:10:41 +2026-09-14 21:00:34 | INFO | train rank=0 | step=7470/50000 progress=14.9% loss=0.132405 lr=9.732298e-05 lr_backbone=9.732298e-05 elapsed=07:14:07 avg_step=3.49s recent_step=5.14s eta=41:11:40 +2026-09-14 21:01:06 | INFO | train rank=0 | step=7480/50000 progress=15.0% loss=0.092639 lr=9.731230e-05 lr_backbone=9.731230e-05 elapsed=07:14:39 avg_step=3.49s recent_step=3.14s eta=41:10:46 +2026-09-14 21:01:40 | INFO | train rank=0 | step=7490/50000 progress=15.0% loss=0.126202 lr=9.730159e-05 lr_backbone=9.730159e-05 elapsed=07:15:12 avg_step=3.49s recent_step=3.38s eta=41:10:05 +2026-09-14 21:02:16 | INFO | train rank=0 | step=7500/50000 progress=15.0% loss=0.119845 lr=9.729086e-05 lr_backbone=9.729086e-05 elapsed=07:15:49 avg_step=3.49s recent_step=3.65s eta=41:09:39 +2026-09-14 21:02:55 | INFO | train rank=0 | step=7510/50000 progress=15.0% loss=0.147294 lr=9.728011e-05 lr_backbone=9.728011e-05 elapsed=07:16:28 avg_step=3.49s recent_step=3.91s eta=41:09:28 +2026-09-14 21:03:29 | INFO | train rank=0 | step=7520/50000 progress=15.0% loss=0.124213 lr=9.726935e-05 lr_backbone=9.726935e-05 elapsed=07:17:02 avg_step=3.49s recent_step=3.42s eta=41:08:49 +2026-09-14 21:04:05 | INFO | train rank=0 | step=7530/50000 progress=15.1% loss=0.127025 lr=9.725856e-05 lr_backbone=9.725856e-05 elapsed=07:17:38 avg_step=3.49s recent_step=3.56s eta=41:08:19 +2026-09-14 21:04:47 | INFO | train rank=0 | step=7540/50000 progress=15.1% loss=0.139840 lr=9.724775e-05 lr_backbone=9.724775e-05 elapsed=07:18:20 avg_step=3.49s recent_step=4.19s eta=41:08:24 +2026-09-14 21:05:21 | INFO | train rank=0 | step=7550/50000 progress=15.1% loss=0.123409 lr=9.723692e-05 lr_backbone=9.723692e-05 elapsed=07:18:54 avg_step=3.49s recent_step=3.44s eta=41:07:46 +2026-09-14 21:05:58 | INFO | train rank=0 | step=7560/50000 progress=15.1% loss=0.128834 lr=9.722606e-05 lr_backbone=9.722606e-05 elapsed=07:19:31 avg_step=3.49s recent_step=3.71s eta=41:07:24 +2026-09-14 21:06:33 | INFO | train rank=0 | step=7570/50000 progress=15.1% loss=0.095876 lr=9.721519e-05 lr_backbone=9.721519e-05 elapsed=07:20:06 avg_step=3.49s recent_step=3.45s eta=41:06:47 +2026-09-14 21:07:23 | INFO | train rank=0 | step=7580/50000 progress=15.2% loss=0.113317 lr=9.720430e-05 lr_backbone=9.720430e-05 elapsed=07:20:56 avg_step=3.49s recent_step=5.04s eta=41:07:39 +2026-09-14 21:07:58 | INFO | train rank=0 | step=7590/50000 progress=15.2% loss=0.108046 lr=9.719339e-05 lr_backbone=9.719339e-05 elapsed=07:21:31 avg_step=3.49s recent_step=3.46s eta=41:07:02 +2026-09-14 21:08:39 | INFO | train rank=0 | step=7600/50000 progress=15.2% loss=0.114456 lr=9.718245e-05 lr_backbone=9.718245e-05 elapsed=07:22:12 avg_step=3.49s recent_step=4.12s eta=41:07:03 +2026-09-14 21:09:13 | INFO | train rank=0 | step=7610/50000 progress=15.2% loss=0.102676 lr=9.717150e-05 lr_backbone=9.717150e-05 elapsed=07:22:46 avg_step=3.49s recent_step=3.35s eta=41:06:20 +2026-09-14 21:09:59 | INFO | train rank=0 | step=7620/50000 progress=15.2% loss=0.155358 lr=9.716052e-05 lr_backbone=9.716052e-05 elapsed=07:23:32 avg_step=3.49s recent_step=4.63s eta=41:06:48 +2026-09-14 21:10:27 | INFO | train rank=0 | step=7630/50000 progress=15.3% loss=0.143957 lr=9.714953e-05 lr_backbone=9.714953e-05 elapsed=07:24:00 avg_step=3.49s recent_step=2.84s eta=41:05:37 +2026-09-14 21:10:57 | INFO | train rank=0 | step=7640/50000 progress=15.3% loss=0.147726 lr=9.713851e-05 lr_backbone=9.713851e-05 elapsed=07:24:30 avg_step=3.49s recent_step=3.01s eta=41:04:35 +2026-09-14 21:11:31 | INFO | train rank=0 | step=7650/50000 progress=15.3% loss=0.108573 lr=9.712747e-05 lr_backbone=9.712747e-05 elapsed=07:25:04 avg_step=3.49s recent_step=3.33s eta=41:03:52 +2026-09-14 21:12:23 | INFO | train rank=0 | step=7660/50000 progress=15.3% loss=0.124284 lr=9.711642e-05 lr_backbone=9.711642e-05 elapsed=07:25:56 avg_step=3.49s recent_step=5.26s eta=41:04:54 +2026-09-14 21:12:52 | INFO | train rank=0 | step=7670/50000 progress=15.3% loss=0.098163 lr=9.710534e-05 lr_backbone=9.710534e-05 elapsed=07:26:25 avg_step=3.49s recent_step=2.87s eta=41:03:45 +2026-09-14 21:13:23 | INFO | train rank=0 | step=7680/50000 progress=15.4% loss=0.114098 lr=9.709424e-05 lr_backbone=9.709424e-05 elapsed=07:26:56 avg_step=3.49s recent_step=3.11s eta=41:02:49 +2026-09-14 21:13:56 | INFO | train rank=0 | step=7690/50000 progress=15.4% loss=0.119210 lr=9.708312e-05 lr_backbone=9.708312e-05 elapsed=07:27:29 avg_step=3.49s recent_step=3.31s eta=41:02:04 +2026-09-14 21:14:49 | INFO | train rank=0 | step=7700/50000 progress=15.4% loss=0.114167 lr=9.707198e-05 lr_backbone=9.707198e-05 elapsed=07:28:22 avg_step=3.49s recent_step=5.33s eta=41:03:10 +2026-09-14 21:15:21 | INFO | train rank=0 | step=7710/50000 progress=15.4% loss=0.101690 lr=9.706082e-05 lr_backbone=9.706082e-05 elapsed=07:28:53 avg_step=3.49s recent_step=3.11s eta=41:02:14 +2026-09-14 21:15:55 | INFO | train rank=0 | step=7720/50000 progress=15.4% loss=0.126067 lr=9.704964e-05 lr_backbone=9.704964e-05 elapsed=07:29:28 avg_step=3.49s recent_step=3.47s eta=41:01:37 +2026-09-14 21:16:22 | INFO | train rank=0 | step=7730/50000 progress=15.5% loss=0.137403 lr=9.703843e-05 lr_backbone=9.703843e-05 elapsed=07:29:55 avg_step=3.49s recent_step=2.68s eta=41:00:18 +2026-09-14 21:16:55 | INFO | train rank=0 | step=7740/50000 progress=15.5% loss=0.096307 lr=9.702721e-05 lr_backbone=9.702721e-05 elapsed=07:30:28 avg_step=3.49s recent_step=3.30s eta=40:59:32 +2026-09-14 21:17:36 | INFO | train rank=0 | step=7750/50000 progress=15.5% loss=0.103077 lr=9.701597e-05 lr_backbone=9.701597e-05 elapsed=07:31:09 avg_step=3.49s recent_step=4.15s eta=40:59:33 +2026-09-14 21:18:12 | INFO | train rank=0 | step=7760/50000 progress=15.5% loss=0.120790 lr=9.700471e-05 lr_backbone=9.700471e-05 elapsed=07:31:44 avg_step=3.49s recent_step=3.51s eta=40:58:59 +2026-09-14 21:18:47 | INFO | train rank=0 | step=7770/50000 progress=15.5% loss=0.102280 lr=9.699342e-05 lr_backbone=9.699342e-05 elapsed=07:32:20 avg_step=3.49s recent_step=3.54s eta=40:58:27 +2026-09-14 21:19:47 | INFO | train rank=0 | step=7780/50000 progress=15.6% loss=0.129810 lr=9.698212e-05 lr_backbone=9.698212e-05 elapsed=07:33:19 avg_step=3.50s recent_step=5.98s eta=41:00:07 +2026-09-14 21:20:14 | INFO | train rank=0 | step=7790/50000 progress=15.6% loss=0.118529 lr=9.697079e-05 lr_backbone=9.697079e-05 elapsed=07:33:47 avg_step=3.50s recent_step=2.77s eta=40:58:53 +2026-09-14 21:20:51 | INFO | train rank=0 | step=7800/50000 progress=15.6% loss=0.133433 lr=9.695945e-05 lr_backbone=9.695945e-05 elapsed=07:34:23 avg_step=3.50s recent_step=3.62s eta=40:58:24 +2026-09-14 21:21:25 | INFO | train rank=0 | step=7810/50000 progress=15.6% loss=0.127254 lr=9.694808e-05 lr_backbone=9.694808e-05 elapsed=07:34:58 avg_step=3.50s recent_step=3.42s eta=40:57:45 +2026-09-14 21:22:19 | INFO | train rank=0 | step=7820/50000 progress=15.6% loss=0.127331 lr=9.693669e-05 lr_backbone=9.693669e-05 elapsed=07:35:52 avg_step=3.50s recent_step=5.41s eta=40:58:53 +2026-09-14 21:22:53 | INFO | train rank=0 | step=7830/50000 progress=15.7% loss=0.127777 lr=9.692529e-05 lr_backbone=9.692529e-05 elapsed=07:36:26 avg_step=3.50s recent_step=3.41s eta=40:58:14 +2026-09-14 21:23:23 | INFO | train rank=0 | step=7840/50000 progress=15.7% loss=0.105209 lr=9.691386e-05 lr_backbone=9.691386e-05 elapsed=07:36:56 avg_step=3.50s recent_step=3.01s eta=40:57:13 +2026-09-14 21:23:53 | INFO | train rank=0 | step=7850/50000 progress=15.7% loss=0.154375 lr=9.690241e-05 lr_backbone=9.690241e-05 elapsed=07:37:26 avg_step=3.50s recent_step=3.03s eta=40:56:13 +2026-09-14 21:24:29 | INFO | train rank=0 | step=7860/50000 progress=15.7% loss=0.120717 lr=9.689094e-05 lr_backbone=9.689094e-05 elapsed=07:38:02 avg_step=3.50s recent_step=3.58s eta=40:55:42 +2026-09-14 21:25:02 | INFO | train rank=0 | step=7870/50000 progress=15.7% loss=0.111957 lr=9.687945e-05 lr_backbone=9.687945e-05 elapsed=07:38:34 avg_step=3.50s recent_step=3.25s eta=40:54:54 +2026-09-14 21:25:40 | INFO | train rank=0 | step=7880/50000 progress=15.8% loss=0.137028 lr=9.686794e-05 lr_backbone=9.686794e-05 elapsed=07:39:13 avg_step=3.50s recent_step=3.81s eta=40:54:36 +2026-09-14 21:26:11 | INFO | train rank=0 | step=7890/50000 progress=15.8% loss=0.149728 lr=9.685641e-05 lr_backbone=9.685641e-05 elapsed=07:39:44 avg_step=3.50s recent_step=3.09s eta=40:53:39 +2026-09-14 21:26:58 | INFO | train rank=0 | step=7900/50000 progress=15.8% loss=0.136604 lr=9.684486e-05 lr_backbone=9.684486e-05 elapsed=07:40:31 avg_step=3.50s recent_step=4.72s eta=40:54:10 +2026-09-14 21:27:30 | INFO | train rank=0 | step=7910/50000 progress=15.8% loss=0.169646 lr=9.683329e-05 lr_backbone=9.683329e-05 elapsed=07:41:02 avg_step=3.50s recent_step=3.15s eta=40:53:16 +2026-09-14 21:28:06 | INFO | train rank=0 | step=7920/50000 progress=15.8% loss=0.117781 lr=9.682170e-05 lr_backbone=9.682170e-05 elapsed=07:41:39 avg_step=3.50s recent_step=3.64s eta=40:52:49 +2026-09-14 21:28:43 | INFO | train rank=0 | step=7930/50000 progress=15.9% loss=0.091725 lr=9.681008e-05 lr_backbone=9.681008e-05 elapsed=07:42:16 avg_step=3.50s recent_step=3.69s eta=40:52:24 +2026-09-14 21:29:26 | INFO | train rank=0 | step=7940/50000 progress=15.9% loss=0.136936 lr=9.679845e-05 lr_backbone=9.679845e-05 elapsed=07:42:58 avg_step=3.50s recent_step=4.28s eta=40:52:31 +2026-09-14 21:30:02 | INFO | train rank=0 | step=7950/50000 progress=15.9% loss=0.104055 lr=9.678680e-05 lr_backbone=9.678680e-05 elapsed=07:43:35 avg_step=3.50s recent_step=3.64s eta=40:52:03 +2026-09-14 21:30:31 | INFO | train rank=0 | step=7960/50000 progress=15.9% loss=0.111599 lr=9.677512e-05 lr_backbone=9.677512e-05 elapsed=07:44:04 avg_step=3.50s recent_step=2.88s eta=40:50:55 +2026-09-14 21:31:04 | INFO | train rank=0 | step=7970/50000 progress=15.9% loss=0.123556 lr=9.676343e-05 lr_backbone=9.676343e-05 elapsed=07:44:37 avg_step=3.50s recent_step=3.37s eta=40:50:13 +2026-09-14 21:32:05 | INFO | train rank=0 | step=7980/50000 progress=16.0% loss=0.122593 lr=9.675171e-05 lr_backbone=9.675171e-05 elapsed=07:45:38 avg_step=3.50s recent_step=6.08s eta=40:51:54 +2026-09-14 21:32:38 | INFO | train rank=0 | step=7990/50000 progress=16.0% loss=0.133946 lr=9.673998e-05 lr_backbone=9.673998e-05 elapsed=07:46:11 avg_step=3.50s recent_step=3.29s eta=40:51:08 +2026-09-14 21:33:06 | INFO | train rank=0 | step=8000/50000 progress=16.0% loss=0.139312 lr=9.672822e-05 lr_backbone=9.672822e-05 elapsed=07:46:39 avg_step=3.50s recent_step=2.78s eta=40:49:55 +2026-09-14 21:33:41 | INFO | train rank=0 | step=8010/50000 progress=16.0% loss=0.108319 lr=9.671645e-05 lr_backbone=9.671645e-05 elapsed=07:47:14 avg_step=3.50s recent_step=3.54s eta=40:49:22 +2026-09-14 21:34:32 | INFO | train rank=0 | step=8020/50000 progress=16.0% loss=0.095545 lr=9.670465e-05 lr_backbone=9.670465e-05 elapsed=07:48:05 avg_step=3.50s recent_step=5.06s eta=40:50:09 +2026-09-14 21:35:04 | INFO | train rank=0 | step=8030/50000 progress=16.1% loss=0.084081 lr=9.669283e-05 lr_backbone=9.669283e-05 elapsed=07:48:37 avg_step=3.50s recent_step=3.21s eta=40:49:19 +2026-09-14 21:35:46 | INFO | train rank=0 | step=8040/50000 progress=16.1% loss=0.103231 lr=9.668100e-05 lr_backbone=9.668100e-05 elapsed=07:49:19 avg_step=3.50s recent_step=4.17s eta=40:49:19 +2026-09-14 21:36:12 | INFO | train rank=0 | step=8050/50000 progress=16.1% loss=0.122829 lr=9.666914e-05 lr_backbone=9.666914e-05 elapsed=07:49:45 avg_step=3.50s recent_step=2.65s eta=40:47:59 +2026-09-14 21:36:58 | INFO | train rank=0 | step=8060/50000 progress=16.1% loss=0.118801 lr=9.665726e-05 lr_backbone=9.665726e-05 elapsed=07:50:30 avg_step=3.50s recent_step=4.54s eta=40:48:18 +2026-09-14 21:37:32 | INFO | train rank=0 | step=8070/50000 progress=16.1% loss=0.105078 lr=9.664536e-05 lr_backbone=9.664536e-05 elapsed=07:51:05 avg_step=3.50s recent_step=3.48s eta=40:47:42 +2026-09-14 21:38:09 | INFO | train rank=0 | step=8080/50000 progress=16.2% loss=0.139408 lr=9.663344e-05 lr_backbone=9.663344e-05 elapsed=07:51:42 avg_step=3.50s recent_step=3.65s eta=40:47:15 +2026-09-14 21:38:47 | INFO | train rank=0 | step=8090/50000 progress=16.2% loss=0.129411 lr=9.662150e-05 lr_backbone=9.662150e-05 elapsed=07:52:20 avg_step=3.50s recent_step=3.80s eta=40:46:55 +2026-09-14 21:39:41 | INFO | train rank=0 | step=8100/50000 progress=16.2% loss=0.117748 lr=9.660954e-05 lr_backbone=9.660954e-05 elapsed=07:53:14 avg_step=3.51s recent_step=5.40s eta=40:47:58 +2026-09-14 21:40:12 | INFO | train rank=0 | step=8110/50000 progress=16.2% loss=0.125572 lr=9.659756e-05 lr_backbone=9.659756e-05 elapsed=07:53:45 avg_step=3.51s recent_step=3.14s eta=40:47:04 +2026-09-14 21:40:45 | INFO | train rank=0 | step=8120/50000 progress=16.2% loss=0.124092 lr=9.658556e-05 lr_backbone=9.658556e-05 elapsed=07:54:18 avg_step=3.50s recent_step=3.31s eta=40:46:19 +2026-09-14 21:41:16 | INFO | train rank=0 | step=8130/50000 progress=16.3% loss=0.164445 lr=9.657354e-05 lr_backbone=9.657354e-05 elapsed=07:54:49 avg_step=3.50s recent_step=3.09s eta=40:45:23 +2026-09-14 21:41:58 | INFO | train rank=0 | step=8140/50000 progress=16.3% loss=0.150992 lr=9.656150e-05 lr_backbone=9.656150e-05 elapsed=07:55:31 avg_step=3.51s recent_step=4.19s eta=40:45:23 +2026-09-14 21:42:32 | INFO | train rank=0 | step=8150/50000 progress=16.3% loss=0.105831 lr=9.654944e-05 lr_backbone=9.654944e-05 elapsed=07:56:04 avg_step=3.50s recent_step=3.34s eta=40:44:39 +2026-09-14 21:43:03 | INFO | train rank=0 | step=8160/50000 progress=16.3% loss=0.138748 lr=9.653736e-05 lr_backbone=9.653736e-05 elapsed=07:56:36 avg_step=3.50s recent_step=3.12s eta=40:43:45 +2026-09-14 21:43:34 | INFO | train rank=0 | step=8170/50000 progress=16.3% loss=0.092104 lr=9.652525e-05 lr_backbone=9.652525e-05 elapsed=07:57:06 avg_step=3.50s recent_step=3.08s eta=40:42:48 +2026-09-14 21:44:24 | INFO | train rank=0 | step=8180/50000 progress=16.4% loss=0.154272 lr=9.651313e-05 lr_backbone=9.651313e-05 elapsed=07:57:57 avg_step=3.51s recent_step=5.03s eta=40:43:31 +2026-09-14 21:45:01 | INFO | train rank=0 | step=8190/50000 progress=16.4% loss=0.133685 lr=9.650099e-05 lr_backbone=9.650099e-05 elapsed=07:58:34 avg_step=3.51s recent_step=3.73s eta=40:43:07 +2026-09-14 21:45:34 | INFO | train rank=0 | step=8200/50000 progress=16.4% loss=0.114941 lr=9.648882e-05 lr_backbone=9.648882e-05 elapsed=07:59:07 avg_step=3.51s recent_step=3.28s eta=40:42:21 +2026-09-14 21:46:04 | INFO | train rank=0 | step=8210/50000 progress=16.4% loss=0.178891 lr=9.647664e-05 lr_backbone=9.647664e-05 elapsed=07:59:37 avg_step=3.51s recent_step=3.00s eta=40:41:20 +2026-09-14 21:46:51 | INFO | train rank=0 | step=8220/50000 progress=16.4% loss=0.116288 lr=9.646444e-05 lr_backbone=9.646444e-05 elapsed=08:00:24 avg_step=3.51s recent_step=4.73s eta=40:41:47 +2026-09-14 21:47:28 | INFO | train rank=0 | step=8230/50000 progress=16.5% loss=0.100816 lr=9.645221e-05 lr_backbone=9.645221e-05 elapsed=08:01:01 avg_step=3.51s recent_step=3.70s eta=40:41:22 +2026-09-14 21:47:54 | INFO | train rank=0 | step=8240/50000 progress=16.5% loss=0.118248 lr=9.643997e-05 lr_backbone=9.643997e-05 elapsed=08:01:27 avg_step=3.51s recent_step=2.59s eta=40:40:01 +2026-09-14 21:48:30 | INFO | train rank=0 | step=8250/50000 progress=16.5% loss=0.121850 lr=9.642770e-05 lr_backbone=9.642770e-05 elapsed=08:02:03 avg_step=3.51s recent_step=3.56s eta=40:39:28 +2026-09-14 21:49:23 | INFO | train rank=0 | step=8260/50000 progress=16.5% loss=0.180911 lr=9.641542e-05 lr_backbone=9.641542e-05 elapsed=08:02:55 avg_step=3.51s recent_step=5.27s eta=40:40:23 +2026-09-14 21:49:45 | INFO | train rank=0 | step=8270/50000 progress=16.5% loss=0.120352 lr=9.640311e-05 lr_backbone=9.640311e-05 elapsed=08:03:18 avg_step=3.51s recent_step=2.28s eta=40:38:45 +2026-09-14 21:50:14 | INFO | train rank=0 | step=8280/50000 progress=16.6% loss=0.130279 lr=9.639078e-05 lr_backbone=9.639078e-05 elapsed=08:03:47 avg_step=3.51s recent_step=2.88s eta=40:37:39 +2026-09-14 21:50:46 | INFO | train rank=0 | step=8290/50000 progress=16.6% loss=0.111753 lr=9.637844e-05 lr_backbone=9.637844e-05 elapsed=08:04:19 avg_step=3.51s recent_step=3.18s eta=40:36:48 +2026-09-14 21:51:39 | INFO | train rank=0 | step=8300/50000 progress=16.6% loss=0.148383 lr=9.636607e-05 lr_backbone=9.636607e-05 elapsed=08:05:12 avg_step=3.51s recent_step=5.29s eta=40:37:42 +2026-09-14 21:52:09 | INFO | train rank=0 | step=8310/50000 progress=16.6% loss=0.132087 lr=9.635368e-05 lr_backbone=9.635368e-05 elapsed=08:05:42 avg_step=3.51s recent_step=3.02s eta=40:36:43 +2026-09-14 21:52:34 | INFO | train rank=0 | step=8320/50000 progress=16.6% loss=0.097291 lr=9.634128e-05 lr_backbone=9.634128e-05 elapsed=08:06:07 avg_step=3.51s recent_step=2.45s eta=40:35:15 +2026-09-14 21:53:11 | INFO | train rank=0 | step=8330/50000 progress=16.7% loss=0.146458 lr=9.632885e-05 lr_backbone=9.632885e-05 elapsed=08:06:44 avg_step=3.51s recent_step=3.74s eta=40:34:52 +2026-09-14 21:54:08 | INFO | train rank=0 | step=8340/50000 progress=16.7% loss=0.113044 lr=9.631640e-05 lr_backbone=9.631640e-05 elapsed=08:07:41 avg_step=3.51s recent_step=5.70s eta=40:36:07 +2026-09-14 21:54:29 | INFO | train rank=0 | step=8350/50000 progress=16.7% loss=0.104906 lr=9.630393e-05 lr_backbone=9.630393e-05 elapsed=08:08:02 avg_step=3.51s recent_step=2.08s eta=40:34:20 +2026-09-14 21:55:01 | INFO | train rank=0 | step=8360/50000 progress=16.7% loss=0.107153 lr=9.629145e-05 lr_backbone=9.629145e-05 elapsed=08:08:34 avg_step=3.51s recent_step=3.19s eta=40:33:30 +2026-09-14 21:55:33 | INFO | train rank=0 | step=8370/50000 progress=16.7% loss=0.128737 lr=9.627894e-05 lr_backbone=9.627894e-05 elapsed=08:09:06 avg_step=3.51s recent_step=3.21s eta=40:32:40 +2026-09-14 21:56:20 | INFO | train rank=0 | step=8380/50000 progress=16.8% loss=0.130740 lr=9.626641e-05 lr_backbone=9.626641e-05 elapsed=08:09:53 avg_step=3.51s recent_step=4.74s eta=40:33:06 +2026-09-14 21:56:54 | INFO | train rank=0 | step=8390/50000 progress=16.8% loss=0.113216 lr=9.625386e-05 lr_backbone=9.625386e-05 elapsed=08:10:27 avg_step=3.51s recent_step=3.33s eta=40:32:22 +2026-09-14 21:57:30 | INFO | train rank=0 | step=8400/50000 progress=16.8% loss=0.082745 lr=9.624129e-05 lr_backbone=9.624129e-05 elapsed=08:11:03 avg_step=3.51s recent_step=3.62s eta=40:31:53 +2026-09-14 21:58:06 | INFO | train rank=0 | step=8410/50000 progress=16.8% loss=0.162188 lr=9.622870e-05 lr_backbone=9.622870e-05 elapsed=08:11:39 avg_step=3.51s recent_step=3.59s eta=40:31:22 +2026-09-14 21:58:55 | INFO | train rank=0 | step=8420/50000 progress=16.8% loss=0.094708 lr=9.621609e-05 lr_backbone=9.621609e-05 elapsed=08:12:27 avg_step=3.51s recent_step=4.87s eta=40:31:54 +2026-09-14 21:59:26 | INFO | train rank=0 | step=8430/50000 progress=16.9% loss=0.112363 lr=9.620346e-05 lr_backbone=9.620346e-05 elapsed=08:12:59 avg_step=3.51s recent_step=3.19s eta=40:31:03 +2026-09-14 22:00:04 | INFO | train rank=0 | step=8440/50000 progress=16.9% loss=0.124035 lr=9.619081e-05 lr_backbone=9.619081e-05 elapsed=08:13:37 avg_step=3.51s recent_step=3.74s eta=40:30:40 +2026-09-14 22:00:32 | INFO | train rank=0 | step=8450/50000 progress=16.9% loss=0.135897 lr=9.617814e-05 lr_backbone=9.617814e-05 elapsed=08:14:05 avg_step=3.51s recent_step=2.82s eta=40:29:30 +2026-09-14 22:01:16 | INFO | train rank=0 | step=8460/50000 progress=16.9% loss=0.116018 lr=9.616545e-05 lr_backbone=9.616545e-05 elapsed=08:14:49 avg_step=3.51s recent_step=4.38s eta=40:29:38 +2026-09-14 22:01:51 | INFO | train rank=0 | step=8470/50000 progress=16.9% loss=0.117923 lr=9.615274e-05 lr_backbone=9.615274e-05 elapsed=08:15:24 avg_step=3.51s recent_step=3.49s eta=40:29:02 +2026-09-14 22:02:29 | INFO | train rank=0 | step=8480/50000 progress=17.0% loss=0.111893 lr=9.614001e-05 lr_backbone=9.614001e-05 elapsed=08:16:02 avg_step=3.51s recent_step=3.82s eta=40:28:42 +2026-09-14 22:03:02 | INFO | train rank=0 | step=8490/50000 progress=17.0% loss=0.147640 lr=9.612726e-05 lr_backbone=9.612726e-05 elapsed=08:16:35 avg_step=3.51s recent_step=3.34s eta=40:27:59 +2026-09-14 22:03:49 | INFO | train rank=0 | step=8500/50000 progress=17.0% loss=0.128232 lr=9.611449e-05 lr_backbone=9.611449e-05 elapsed=08:17:22 avg_step=3.51s recent_step=4.67s eta=40:28:20 +2026-09-14 22:04:26 | INFO | train rank=0 | step=8510/50000 progress=17.0% loss=0.131796 lr=9.610170e-05 lr_backbone=9.610170e-05 elapsed=08:17:59 avg_step=3.51s recent_step=3.73s eta=40:27:56 +2026-09-14 22:04:56 | INFO | train rank=0 | step=8520/50000 progress=17.0% loss=0.093201 lr=9.608888e-05 lr_backbone=9.608888e-05 elapsed=08:18:29 avg_step=3.51s recent_step=3.00s eta=40:26:56 +2026-09-14 22:05:28 | INFO | train rank=0 | step=8530/50000 progress=17.1% loss=0.122736 lr=9.607605e-05 lr_backbone=9.607605e-05 elapsed=08:19:01 avg_step=3.51s recent_step=3.18s eta=40:26:04 +2026-09-14 22:06:20 | INFO | train rank=0 | step=8540/50000 progress=17.1% loss=0.140505 lr=9.606320e-05 lr_backbone=9.606320e-05 elapsed=08:19:53 avg_step=3.51s recent_step=5.17s eta=40:26:50 +2026-09-14 22:06:55 | INFO | train rank=0 | step=8550/50000 progress=17.1% loss=0.132698 lr=9.605033e-05 lr_backbone=9.605033e-05 elapsed=08:20:27 avg_step=3.51s recent_step=3.48s eta=40:26:13 +2026-09-14 22:07:26 | INFO | train rank=0 | step=8560/50000 progress=17.1% loss=0.086617 lr=9.603744e-05 lr_backbone=9.603744e-05 elapsed=08:20:59 avg_step=3.51s recent_step=3.13s eta=40:25:20 +2026-09-14 22:08:02 | INFO | train rank=0 | step=8570/50000 progress=17.1% loss=0.120264 lr=9.602452e-05 lr_backbone=9.602452e-05 elapsed=08:21:34 avg_step=3.51s recent_step=3.57s eta=40:24:48 +2026-09-14 22:08:51 | INFO | train rank=0 | step=8580/50000 progress=17.2% loss=0.129443 lr=9.601159e-05 lr_backbone=9.601159e-05 elapsed=08:22:24 avg_step=3.51s recent_step=4.95s eta=40:25:22 +2026-09-14 22:09:25 | INFO | train rank=0 | step=8590/50000 progress=17.2% loss=0.089035 lr=9.599864e-05 lr_backbone=9.599864e-05 elapsed=08:22:58 avg_step=3.51s recent_step=3.38s eta=40:24:40 +2026-09-14 22:09:56 | INFO | train rank=0 | step=8600/50000 progress=17.2% loss=0.124410 lr=9.598567e-05 lr_backbone=9.598567e-05 elapsed=08:23:29 avg_step=3.51s recent_step=3.15s eta=40:23:47 +2026-09-14 22:10:23 | INFO | train rank=0 | step=8610/50000 progress=17.2% loss=0.114068 lr=9.597267e-05 lr_backbone=9.597267e-05 elapsed=08:23:56 avg_step=3.51s recent_step=2.64s eta=40:22:31 +2026-09-14 22:11:16 | INFO | train rank=0 | step=8620/50000 progress=17.2% loss=0.131326 lr=9.595966e-05 lr_backbone=9.595966e-05 elapsed=08:24:49 avg_step=3.51s recent_step=5.29s eta=40:23:21 +2026-09-14 22:11:48 | INFO | train rank=0 | step=8630/50000 progress=17.3% loss=0.134239 lr=9.594663e-05 lr_backbone=9.594663e-05 elapsed=08:25:21 avg_step=3.51s recent_step=3.24s eta=40:22:33 +2026-09-14 22:12:21 | INFO | train rank=0 | step=8640/50000 progress=17.3% loss=0.141952 lr=9.593358e-05 lr_backbone=9.593358e-05 elapsed=08:25:53 avg_step=3.51s recent_step=3.24s eta=40:21:45 +2026-09-14 22:13:02 | INFO | train rank=0 | step=8650/50000 progress=17.3% loss=0.093993 lr=9.592050e-05 lr_backbone=9.592050e-05 elapsed=08:26:34 avg_step=3.51s recent_step=4.10s eta=40:21:38 +2026-09-14 22:13:46 | INFO | train rank=0 | step=8660/50000 progress=17.3% loss=0.115797 lr=9.590741e-05 lr_backbone=9.590741e-05 elapsed=08:27:18 avg_step=3.51s recent_step=4.40s eta=40:21:45 +2026-09-14 22:14:13 | INFO | train rank=0 | step=8670/50000 progress=17.3% loss=0.144517 lr=9.589430e-05 lr_backbone=9.589430e-05 elapsed=08:27:45 avg_step=3.51s recent_step=2.69s eta=40:20:31 +2026-09-14 22:14:41 | INFO | train rank=0 | step=8680/50000 progress=17.4% loss=0.167701 lr=9.588116e-05 lr_backbone=9.588116e-05 elapsed=08:28:14 avg_step=3.51s recent_step=2.83s eta=40:19:23 +2026-09-14 22:15:17 | INFO | train rank=0 | step=8690/50000 progress=17.4% loss=0.111048 lr=9.586801e-05 lr_backbone=9.586801e-05 elapsed=08:28:50 avg_step=3.51s recent_step=3.64s eta=40:18:54 +2026-09-14 22:16:09 | INFO | train rank=0 | step=8700/50000 progress=17.4% loss=0.101304 lr=9.585483e-05 lr_backbone=9.585483e-05 elapsed=08:29:42 avg_step=3.52s recent_step=5.18s eta=40:19:38 +2026-09-14 22:16:46 | INFO | train rank=0 | step=8710/50000 progress=17.4% loss=0.105895 lr=9.584164e-05 lr_backbone=9.584164e-05 elapsed=08:30:18 avg_step=3.52s recent_step=3.65s eta=40:19:09 +2026-09-14 22:17:19 | INFO | train rank=0 | step=8720/50000 progress=17.4% loss=0.128625 lr=9.582843e-05 lr_backbone=9.582843e-05 elapsed=08:30:52 avg_step=3.52s recent_step=3.35s eta=40:18:26 +2026-09-14 22:17:47 | INFO | train rank=0 | step=8730/50000 progress=17.5% loss=0.127457 lr=9.581519e-05 lr_backbone=9.581519e-05 elapsed=08:31:19 avg_step=3.51s recent_step=2.75s eta=40:17:15 +2026-09-14 22:18:42 | INFO | train rank=0 | step=8740/50000 progress=17.5% loss=0.107133 lr=9.580194e-05 lr_backbone=9.580194e-05 elapsed=08:32:14 avg_step=3.52s recent_step=5.50s eta=40:18:14 +2026-09-14 22:19:17 | INFO | train rank=0 | step=8750/50000 progress=17.5% loss=0.129728 lr=9.578867e-05 lr_backbone=9.578867e-05 elapsed=08:32:50 avg_step=3.52s recent_step=3.53s eta=40:17:39 +2026-09-14 22:19:52 | INFO | train rank=0 | step=8760/50000 progress=17.5% loss=0.157098 lr=9.577537e-05 lr_backbone=9.577537e-05 elapsed=08:33:25 avg_step=3.52s recent_step=3.55s eta=40:17:05 +2026-09-14 22:20:22 | INFO | train rank=0 | step=8770/50000 progress=17.5% loss=0.100302 lr=9.576206e-05 lr_backbone=9.576206e-05 elapsed=08:33:55 avg_step=3.52s recent_step=2.97s eta=40:16:04 +2026-09-14 22:21:16 | INFO | train rank=0 | step=8780/50000 progress=17.6% loss=0.120732 lr=9.574872e-05 lr_backbone=9.574872e-05 elapsed=08:34:49 avg_step=3.52s recent_step=5.44s eta=40:17:00 +2026-09-14 22:21:47 | INFO | train rank=0 | step=8790/50000 progress=17.6% loss=0.101683 lr=9.573537e-05 lr_backbone=9.573537e-05 elapsed=08:35:19 avg_step=3.52s recent_step=3.02s eta=40:16:01 +2026-09-14 22:22:18 | INFO | train rank=0 | step=8800/50000 progress=17.6% loss=0.100464 lr=9.572200e-05 lr_backbone=9.572200e-05 elapsed=08:35:51 avg_step=3.52s recent_step=3.15s eta=40:15:09 +2026-09-14 22:22:52 | INFO | train rank=0 | step=8810/50000 progress=17.6% loss=0.144323 lr=9.570860e-05 lr_backbone=9.570860e-05 elapsed=08:36:25 avg_step=3.52s recent_step=3.37s eta=40:14:27 +2026-09-14 22:23:49 | INFO | train rank=0 | step=8820/50000 progress=17.6% loss=0.097900 lr=9.569519e-05 lr_backbone=9.569519e-05 elapsed=08:37:22 avg_step=3.52s recent_step=5.72s eta=40:15:35 +2026-09-14 22:24:26 | INFO | train rank=0 | step=8830/50000 progress=17.7% loss=0.132397 lr=9.568176e-05 lr_backbone=9.568176e-05 elapsed=08:37:59 avg_step=3.52s recent_step=3.71s eta=40:15:08 +2026-09-14 22:24:59 | INFO | train rank=0 | step=8840/50000 progress=17.7% loss=0.091131 lr=9.566830e-05 lr_backbone=9.566830e-05 elapsed=08:38:32 avg_step=3.52s recent_step=3.31s eta=40:14:23 +2026-09-14 22:25:25 | INFO | train rank=0 | step=8850/50000 progress=17.7% loss=0.088390 lr=9.565483e-05 lr_backbone=9.565483e-05 elapsed=08:38:58 avg_step=3.52s recent_step=2.56s eta=40:13:04 +2026-09-14 22:26:12 | INFO | train rank=0 | step=8860/50000 progress=17.7% loss=0.116624 lr=9.564133e-05 lr_backbone=9.564133e-05 elapsed=08:39:45 avg_step=3.52s recent_step=4.72s eta=40:13:24 +2026-09-14 22:26:52 | INFO | train rank=0 | step=8870/50000 progress=17.7% loss=0.160886 lr=9.562782e-05 lr_backbone=9.562782e-05 elapsed=08:40:25 avg_step=3.52s recent_step=4.03s eta=40:13:12 +2026-09-14 22:27:28 | INFO | train rank=0 | step=8880/50000 progress=17.8% loss=0.100909 lr=9.561429e-05 lr_backbone=9.561429e-05 elapsed=08:41:00 avg_step=3.52s recent_step=3.52s eta=40:12:37 +2026-09-14 22:28:01 | INFO | train rank=0 | step=8890/50000 progress=17.8% loss=0.111852 lr=9.560073e-05 lr_backbone=9.560073e-05 elapsed=08:41:34 avg_step=3.52s recent_step=3.36s eta=40:11:55 +2026-09-14 22:28:56 | INFO | train rank=0 | step=8900/50000 progress=17.8% loss=0.094658 lr=9.558716e-05 lr_backbone=9.558716e-05 elapsed=08:42:29 avg_step=3.52s recent_step=5.51s eta=40:12:51 +2026-09-14 22:29:34 | INFO | train rank=0 | step=8910/50000 progress=17.8% loss=0.092435 lr=9.557357e-05 lr_backbone=9.557357e-05 elapsed=08:43:07 avg_step=3.52s recent_step=3.79s eta=40:12:29 +2026-09-14 22:29:59 | INFO | train rank=0 | step=8920/50000 progress=17.8% loss=0.146070 lr=9.555995e-05 lr_backbone=9.555995e-05 elapsed=08:43:32 avg_step=3.52s recent_step=2.47s eta=40:11:05 +2026-09-14 22:30:18 | INFO | train rank=0 | step=8930/50000 progress=17.9% loss=0.094813 lr=9.554632e-05 lr_backbone=9.554632e-05 elapsed=08:43:50 avg_step=3.52s recent_step=1.86s eta=40:09:14 +2026-09-14 22:31:32 | INFO | train rank=0 | step=8940/50000 progress=17.9% loss=0.100732 lr=9.553267e-05 lr_backbone=9.553267e-05 elapsed=08:45:05 avg_step=3.52s recent_step=7.45s eta=40:11:39 +2026-09-14 22:32:06 | INFO | train rank=0 | step=8950/50000 progress=17.9% loss=0.131684 lr=9.551899e-05 lr_backbone=9.551899e-05 elapsed=08:45:39 avg_step=3.52s recent_step=3.43s eta=40:10:59 +2026-09-14 22:32:41 | INFO | train rank=0 | step=8960/50000 progress=17.9% loss=0.118251 lr=9.550530e-05 lr_backbone=9.550530e-05 elapsed=08:46:14 avg_step=3.52s recent_step=3.50s eta=40:10:23 +2026-09-14 22:33:11 | INFO | train rank=0 | step=8970/50000 progress=17.9% loss=0.134499 lr=9.549159e-05 lr_backbone=9.549159e-05 elapsed=08:46:44 avg_step=3.52s recent_step=3.00s eta=40:09:24 +2026-09-14 22:34:08 | INFO | train rank=0 | step=8980/50000 progress=18.0% loss=0.135563 lr=9.547785e-05 lr_backbone=9.547785e-05 elapsed=08:47:41 avg_step=3.53s recent_step=5.63s eta=40:10:25 +2026-09-14 22:34:39 | INFO | train rank=0 | step=8990/50000 progress=18.0% loss=0.129154 lr=9.546410e-05 lr_backbone=9.546410e-05 elapsed=08:48:12 avg_step=3.53s recent_step=3.12s eta=40:09:31 +2026-09-14 22:35:18 | INFO | train rank=0 | step=9000/50000 progress=18.0% loss=0.129448 lr=9.545033e-05 lr_backbone=9.545033e-05 elapsed=08:48:51 avg_step=3.53s recent_step=3.94s eta=40:09:14 +2026-09-14 22:35:49 | INFO | train rank=0 | step=9010/50000 progress=18.0% loss=0.133663 lr=9.543653e-05 lr_backbone=9.543653e-05 elapsed=08:49:22 avg_step=3.53s recent_step=3.05s eta=40:08:18 +2026-09-14 22:36:29 | INFO | train rank=0 | step=9020/50000 progress=18.0% loss=0.108974 lr=9.542272e-05 lr_backbone=9.542272e-05 elapsed=08:50:02 avg_step=3.53s recent_step=4.01s eta=40:08:04 +2026-09-14 22:37:07 | INFO | train rank=0 | step=9030/50000 progress=18.1% loss=0.103141 lr=9.540889e-05 lr_backbone=9.540889e-05 elapsed=08:50:40 avg_step=3.53s recent_step=3.85s eta=40:07:44 +2026-09-14 22:37:43 | INFO | train rank=0 | step=9040/50000 progress=18.1% loss=0.103507 lr=9.539504e-05 lr_backbone=9.539504e-05 elapsed=08:51:16 avg_step=3.53s recent_step=3.58s eta=40:07:11 +2026-09-14 22:38:16 | INFO | train rank=0 | step=9050/50000 progress=18.1% loss=0.165578 lr=9.538116e-05 lr_backbone=9.538116e-05 elapsed=08:51:49 avg_step=3.53s recent_step=3.30s eta=40:06:26 +2026-09-14 22:38:58 | INFO | train rank=0 | step=9060/50000 progress=18.1% loss=0.095245 lr=9.536727e-05 lr_backbone=9.536727e-05 elapsed=08:52:31 avg_step=3.53s recent_step=4.16s eta=40:06:19 +2026-09-14 22:39:33 | INFO | train rank=0 | step=9070/50000 progress=18.1% loss=0.155391 lr=9.535336e-05 lr_backbone=9.535336e-05 elapsed=08:53:06 avg_step=3.53s recent_step=3.55s eta=40:05:45 +2026-09-14 22:40:11 | INFO | train rank=0 | step=9080/50000 progress=18.2% loss=0.116356 lr=9.533943e-05 lr_backbone=9.533943e-05 elapsed=08:53:44 avg_step=3.53s recent_step=3.77s eta=40:05:20 +2026-09-14 22:40:40 | INFO | train rank=0 | step=9090/50000 progress=18.2% loss=0.140788 lr=9.532548e-05 lr_backbone=9.532548e-05 elapsed=08:54:13 avg_step=3.53s recent_step=2.91s eta=40:04:18 +2026-09-14 22:41:31 | INFO | train rank=0 | step=9100/50000 progress=18.2% loss=0.109597 lr=9.531151e-05 lr_backbone=9.531151e-05 elapsed=08:55:04 avg_step=3.53s recent_step=5.10s eta=40:04:53 +2026-09-14 22:42:14 | INFO | train rank=0 | step=9110/50000 progress=18.2% loss=0.112619 lr=9.529752e-05 lr_backbone=9.529752e-05 elapsed=08:55:47 avg_step=3.53s recent_step=4.27s eta=40:04:51 +2026-09-14 22:42:41 | INFO | train rank=0 | step=9120/50000 progress=18.2% loss=0.130960 lr=9.528350e-05 lr_backbone=9.528350e-05 elapsed=08:56:14 avg_step=3.53s recent_step=2.70s eta=40:03:38 +2026-09-14 22:43:08 | INFO | train rank=0 | step=9130/50000 progress=18.3% loss=0.113462 lr=9.526947e-05 lr_backbone=9.526947e-05 elapsed=08:56:41 avg_step=3.53s recent_step=2.71s eta=40:02:26 +2026-09-14 22:43:58 | INFO | train rank=0 | step=9140/50000 progress=18.3% loss=0.106204 lr=9.525542e-05 lr_backbone=9.525542e-05 elapsed=08:57:31 avg_step=3.53s recent_step=5.05s eta=40:02:59 +2026-09-14 22:44:33 | INFO | train rank=0 | step=9150/50000 progress=18.3% loss=0.115167 lr=9.524135e-05 lr_backbone=9.524135e-05 elapsed=08:58:06 avg_step=3.53s recent_step=3.46s eta=40:02:21 +2026-09-14 22:45:03 | INFO | train rank=0 | step=9160/50000 progress=18.3% loss=0.142299 lr=9.522726e-05 lr_backbone=9.522726e-05 elapsed=08:58:36 avg_step=3.53s recent_step=3.04s eta=40:01:24 +2026-09-14 22:45:31 | INFO | train rank=0 | step=9170/50000 progress=18.3% loss=0.105388 lr=9.521315e-05 lr_backbone=9.521315e-05 elapsed=08:59:04 avg_step=3.53s recent_step=2.74s eta=40:00:13 +2026-09-14 22:46:27 | INFO | train rank=0 | step=9180/50000 progress=18.4% loss=0.130502 lr=9.519902e-05 lr_backbone=9.519902e-05 elapsed=09:00:00 avg_step=3.53s recent_step=5.62s eta=40:01:11 +2026-09-14 22:46:53 | INFO | train rank=0 | step=9190/50000 progress=18.4% loss=0.115522 lr=9.518487e-05 lr_backbone=9.518487e-05 elapsed=09:00:26 avg_step=3.53s recent_step=2.63s eta=39:59:56 +2026-09-14 22:47:23 | INFO | train rank=0 | step=9200/50000 progress=18.4% loss=0.110644 lr=9.517070e-05 lr_backbone=9.517070e-05 elapsed=09:00:56 avg_step=3.53s recent_step=2.94s eta=39:58:55 +2026-09-14 22:48:00 | INFO | train rank=0 | step=9210/50000 progress=18.4% loss=0.101613 lr=9.515652e-05 lr_backbone=9.515652e-05 elapsed=09:01:32 avg_step=3.53s recent_step=3.69s eta=39:58:27 +2026-09-14 22:48:46 | INFO | train rank=0 | step=9220/50000 progress=18.4% loss=0.100148 lr=9.514231e-05 lr_backbone=9.514231e-05 elapsed=09:02:18 avg_step=3.53s recent_step=4.58s eta=39:58:38 +2026-09-14 22:49:19 | INFO | train rank=0 | step=9230/50000 progress=18.5% loss=0.162133 lr=9.512808e-05 lr_backbone=9.512808e-05 elapsed=09:02:52 avg_step=3.53s recent_step=3.36s eta=39:57:56 +2026-09-14 22:49:46 | INFO | train rank=0 | step=9240/50000 progress=18.5% loss=0.111238 lr=9.511383e-05 lr_backbone=9.511383e-05 elapsed=09:03:19 avg_step=3.53s recent_step=2.72s eta=39:56:45 +2026-09-14 22:50:17 | INFO | train rank=0 | step=9250/50000 progress=18.5% loss=0.097486 lr=9.509956e-05 lr_backbone=9.509956e-05 elapsed=09:03:50 avg_step=3.53s recent_step=3.06s eta=39:55:49 +2026-09-14 22:51:05 | INFO | train rank=0 | step=9260/50000 progress=18.5% loss=0.119634 lr=9.508527e-05 lr_backbone=9.508527e-05 elapsed=09:04:37 avg_step=3.53s recent_step=4.77s eta=39:56:08 +2026-09-14 22:51:44 | INFO | train rank=0 | step=9270/50000 progress=18.5% loss=0.091876 lr=9.507097e-05 lr_backbone=9.507097e-05 elapsed=09:05:16 avg_step=3.53s recent_step=3.89s eta=39:55:49 +2026-09-14 22:52:14 | INFO | train rank=0 | step=9280/50000 progress=18.6% loss=0.115898 lr=9.505664e-05 lr_backbone=9.505664e-05 elapsed=09:05:47 avg_step=3.53s recent_step=3.06s eta=39:54:53 +2026-09-14 22:52:46 | INFO | train rank=0 | step=9290/50000 progress=18.6% loss=0.135955 lr=9.504229e-05 lr_backbone=9.504229e-05 elapsed=09:06:19 avg_step=3.53s recent_step=3.18s eta=39:54:02 +2026-09-14 22:53:39 | INFO | train rank=0 | step=9300/50000 progress=18.6% loss=0.114101 lr=9.502793e-05 lr_backbone=9.502793e-05 elapsed=09:07:12 avg_step=3.53s recent_step=5.31s eta=39:54:45 +2026-09-14 22:54:04 | INFO | train rank=0 | step=9310/50000 progress=18.6% loss=0.094675 lr=9.501354e-05 lr_backbone=9.501354e-05 elapsed=09:07:37 avg_step=3.53s recent_step=2.50s eta=39:53:24 +2026-09-14 22:54:39 | INFO | train rank=0 | step=9320/50000 progress=18.6% loss=0.129368 lr=9.499913e-05 lr_backbone=9.499913e-05 elapsed=09:08:12 avg_step=3.53s recent_step=3.48s eta=39:52:47 +2026-09-14 22:55:09 | INFO | train rank=0 | step=9330/50000 progress=18.7% loss=0.119279 lr=9.498471e-05 lr_backbone=9.498471e-05 elapsed=09:08:42 avg_step=3.53s recent_step=3.04s eta=39:51:51 +2026-09-14 22:55:58 | INFO | train rank=0 | step=9340/50000 progress=18.7% loss=0.103093 lr=9.497026e-05 lr_backbone=9.497026e-05 elapsed=09:09:30 avg_step=3.53s recent_step=4.84s eta=39:52:12 +2026-09-14 22:56:33 | INFO | train rank=0 | step=9350/50000 progress=18.7% loss=0.125294 lr=9.495580e-05 lr_backbone=9.495580e-05 elapsed=09:10:06 avg_step=3.53s recent_step=3.57s eta=39:51:39 +2026-09-14 22:57:03 | INFO | train rank=0 | step=9360/50000 progress=18.7% loss=0.123187 lr=9.494131e-05 lr_backbone=9.494131e-05 elapsed=09:10:36 avg_step=3.53s recent_step=3.00s eta=39:50:40 +2026-09-14 22:57:29 | INFO | train rank=0 | step=9370/50000 progress=18.7% loss=0.163986 lr=9.492681e-05 lr_backbone=9.492681e-05 elapsed=09:11:02 avg_step=3.53s recent_step=2.62s eta=39:49:26 +2026-09-14 22:58:12 | INFO | train rank=0 | step=9380/50000 progress=18.8% loss=0.114812 lr=9.491228e-05 lr_backbone=9.491228e-05 elapsed=09:11:45 avg_step=3.53s recent_step=4.25s eta=39:49:22 +2026-09-14 22:58:49 | INFO | train rank=0 | step=9390/50000 progress=18.8% loss=0.104196 lr=9.489774e-05 lr_backbone=9.489774e-05 elapsed=09:12:21 avg_step=3.53s recent_step=3.66s eta=39:48:52 +2026-09-14 22:59:22 | INFO | train rank=0 | step=9400/50000 progress=18.8% loss=0.154984 lr=9.488318e-05 lr_backbone=9.488318e-05 elapsed=09:12:55 avg_step=3.53s recent_step=3.38s eta=39:48:10 +2026-09-14 22:59:51 | INFO | train rank=0 | step=9410/50000 progress=18.8% loss=0.130586 lr=9.486859e-05 lr_backbone=9.486859e-05 elapsed=09:13:23 avg_step=3.53s recent_step=2.82s eta=39:47:04 +2026-09-14 23:00:31 | INFO | train rank=0 | step=9420/50000 progress=18.8% loss=0.116815 lr=9.485399e-05 lr_backbone=9.485399e-05 elapsed=09:14:04 avg_step=3.53s recent_step=4.04s eta=39:46:51 +2026-09-14 23:01:06 | INFO | train rank=0 | step=9430/50000 progress=18.9% loss=0.116859 lr=9.483937e-05 lr_backbone=9.483937e-05 elapsed=09:14:39 avg_step=3.53s recent_step=3.54s eta=39:46:16 +2026-09-14 23:01:30 | INFO | train rank=0 | step=9440/50000 progress=18.9% loss=0.114976 lr=9.482473e-05 lr_backbone=9.482473e-05 elapsed=09:15:03 avg_step=3.53s recent_step=2.38s eta=39:44:52 +2026-09-14 23:02:06 | INFO | train rank=0 | step=9450/50000 progress=18.9% loss=0.107822 lr=9.481007e-05 lr_backbone=9.481007e-05 elapsed=09:15:39 avg_step=3.53s recent_step=3.57s eta=39:44:18 +2026-09-14 23:02:49 | INFO | train rank=0 | step=9460/50000 progress=18.9% loss=0.130698 lr=9.479539e-05 lr_backbone=9.479539e-05 elapsed=09:16:22 avg_step=3.53s recent_step=4.34s eta=39:44:18 +2026-09-14 23:03:19 | INFO | train rank=0 | step=9470/50000 progress=18.9% loss=0.134011 lr=9.478069e-05 lr_backbone=9.478069e-05 elapsed=09:16:52 avg_step=3.53s recent_step=2.97s eta=39:43:19 +2026-09-14 23:04:01 | INFO | train rank=0 | step=9480/50000 progress=19.0% loss=0.116575 lr=9.476597e-05 lr_backbone=9.476597e-05 elapsed=09:17:34 avg_step=3.53s recent_step=4.18s eta=39:43:11 +2026-09-14 23:04:30 | INFO | train rank=0 | step=9490/50000 progress=19.0% loss=0.135228 lr=9.475123e-05 lr_backbone=9.475123e-05 elapsed=09:18:03 avg_step=3.53s recent_step=2.93s eta=39:42:10 +2026-09-14 23:05:12 | INFO | train rank=0 | step=9500/50000 progress=19.0% loss=0.172449 lr=9.473647e-05 lr_backbone=9.473647e-05 elapsed=09:18:45 avg_step=3.53s recent_step=4.18s eta=39:42:03 +2026-09-14 23:05:48 | INFO | train rank=0 | step=9510/50000 progress=19.0% loss=0.146305 lr=9.472169e-05 lr_backbone=9.472169e-05 elapsed=09:19:21 avg_step=3.53s recent_step=3.64s eta=39:41:32 +2026-09-14 23:06:29 | INFO | train rank=0 | step=9520/50000 progress=19.0% loss=0.100650 lr=9.470689e-05 lr_backbone=9.470689e-05 elapsed=09:20:02 avg_step=3.53s recent_step=4.08s eta=39:41:21 +2026-09-14 23:06:57 | INFO | train rank=0 | step=9530/50000 progress=19.1% loss=0.121897 lr=9.469207e-05 lr_backbone=9.469207e-05 elapsed=09:20:30 avg_step=3.53s recent_step=2.78s eta=39:40:14 +2026-09-14 23:07:46 | INFO | train rank=0 | step=9540/50000 progress=19.1% loss=0.130004 lr=9.467723e-05 lr_backbone=9.467723e-05 elapsed=09:21:19 avg_step=3.53s recent_step=4.89s eta=39:40:36 +2026-09-14 23:08:20 | INFO | train rank=0 | step=9550/50000 progress=19.1% loss=0.103375 lr=9.466238e-05 lr_backbone=9.466238e-05 elapsed=09:21:53 avg_step=3.53s recent_step=3.43s eta=39:39:56 +2026-09-14 23:08:51 | INFO | train rank=0 | step=9560/50000 progress=19.1% loss=0.128201 lr=9.464750e-05 lr_backbone=9.464750e-05 elapsed=09:22:24 avg_step=3.53s recent_step=3.13s eta=39:39:04 +2026-09-14 23:09:30 | INFO | train rank=0 | step=9570/50000 progress=19.1% loss=0.098656 lr=9.463260e-05 lr_backbone=9.463260e-05 elapsed=09:23:03 avg_step=3.53s recent_step=3.84s eta=39:38:41 +2026-09-14 23:10:08 | INFO | train rank=0 | step=9580/50000 progress=19.2% loss=0.125650 lr=9.461769e-05 lr_backbone=9.461769e-05 elapsed=09:23:41 avg_step=3.53s recent_step=3.82s eta=39:38:18 +2026-09-14 23:10:39 | INFO | train rank=0 | step=9590/50000 progress=19.2% loss=0.093632 lr=9.460275e-05 lr_backbone=9.460275e-05 elapsed=09:24:12 avg_step=3.53s recent_step=3.14s eta=39:37:27 +2026-09-14 23:11:19 | INFO | train rank=0 | step=9600/50000 progress=19.2% loss=0.134458 lr=9.458780e-05 lr_backbone=9.458780e-05 elapsed=09:24:51 avg_step=3.53s recent_step=3.92s eta=39:37:08 +2026-09-14 23:11:57 | INFO | train rank=0 | step=9610/50000 progress=19.2% loss=0.111359 lr=9.457282e-05 lr_backbone=9.457282e-05 elapsed=09:25:30 avg_step=3.53s recent_step=3.87s eta=39:36:46 +2026-09-14 23:12:49 | INFO | train rank=0 | step=9620/50000 progress=19.2% loss=0.146907 lr=9.455783e-05 lr_backbone=9.455783e-05 elapsed=09:26:21 avg_step=3.53s recent_step=5.13s eta=39:37:18 +2026-09-14 23:13:22 | INFO | train rank=0 | step=9630/50000 progress=19.3% loss=0.093918 lr=9.454282e-05 lr_backbone=9.454282e-05 elapsed=09:26:55 avg_step=3.53s recent_step=3.38s eta=39:36:37 +2026-09-14 23:13:55 | INFO | train rank=0 | step=9640/50000 progress=19.3% loss=0.117121 lr=9.452778e-05 lr_backbone=9.452778e-05 elapsed=09:27:27 avg_step=3.53s recent_step=3.23s eta=39:35:48 +2026-09-14 23:14:31 | INFO | train rank=0 | step=9650/50000 progress=19.3% loss=0.104349 lr=9.451273e-05 lr_backbone=9.451273e-05 elapsed=09:28:03 avg_step=3.53s recent_step=3.60s eta=39:35:16 +2026-09-14 23:15:12 | INFO | train rank=0 | step=9660/50000 progress=19.3% loss=0.083623 lr=9.449766e-05 lr_backbone=9.449766e-05 elapsed=09:28:45 avg_step=3.53s recent_step=4.19s eta=39:35:08 +2026-09-14 23:15:46 | INFO | train rank=0 | step=9670/50000 progress=19.3% loss=0.127932 lr=9.448257e-05 lr_backbone=9.448257e-05 elapsed=09:29:18 avg_step=3.53s recent_step=3.33s eta=39:34:24 +2026-09-14 23:16:28 | INFO | train rank=0 | step=9680/50000 progress=19.4% loss=0.127662 lr=9.446746e-05 lr_backbone=9.446746e-05 elapsed=09:30:01 avg_step=3.53s recent_step=4.21s eta=39:34:17 +2026-09-14 23:17:01 | INFO | train rank=0 | step=9690/50000 progress=19.4% loss=0.113976 lr=9.445233e-05 lr_backbone=9.445233e-05 elapsed=09:30:33 avg_step=3.53s recent_step=3.28s eta=39:33:31 +2026-09-14 23:17:41 | INFO | train rank=0 | step=9700/50000 progress=19.4% loss=0.109841 lr=9.443718e-05 lr_backbone=9.443718e-05 elapsed=09:31:14 avg_step=3.53s recent_step=4.05s eta=39:33:17 +2026-09-14 23:18:10 | INFO | train rank=0 | step=9710/50000 progress=19.4% loss=0.134655 lr=9.442201e-05 lr_backbone=9.442201e-05 elapsed=09:31:43 avg_step=3.53s recent_step=2.92s eta=39:32:17 +2026-09-14 23:18:57 | INFO | train rank=0 | step=9720/50000 progress=19.4% loss=0.115524 lr=9.440682e-05 lr_backbone=9.440682e-05 elapsed=09:32:30 avg_step=3.53s recent_step=4.70s eta=39:32:30 +2026-09-14 23:19:22 | INFO | train rank=0 | step=9730/50000 progress=19.5% loss=0.159619 lr=9.439161e-05 lr_backbone=9.439161e-05 elapsed=09:32:54 avg_step=3.53s recent_step=2.42s eta=39:31:08 +2026-09-14 23:19:56 | INFO | train rank=0 | step=9740/50000 progress=19.5% loss=0.097587 lr=9.437639e-05 lr_backbone=9.437639e-05 elapsed=09:33:29 avg_step=3.53s recent_step=3.48s eta=39:30:31 +2026-09-14 23:20:40 | INFO | train rank=0 | step=9750/50000 progress=19.5% loss=0.129019 lr=9.436114e-05 lr_backbone=9.436114e-05 elapsed=09:34:12 avg_step=3.53s recent_step=4.32s eta=39:30:28 +2026-09-14 23:21:18 | INFO | train rank=0 | step=9760/50000 progress=19.5% loss=0.108514 lr=9.434587e-05 lr_backbone=9.434587e-05 elapsed=09:34:51 avg_step=3.53s recent_step=3.82s eta=39:30:04 +2026-09-14 23:21:54 | INFO | train rank=0 | step=9770/50000 progress=19.5% loss=0.110704 lr=9.433059e-05 lr_backbone=9.433059e-05 elapsed=09:35:27 avg_step=3.53s recent_step=3.67s eta=39:29:35 +2026-09-14 23:22:26 | INFO | train rank=0 | step=9780/50000 progress=19.6% loss=0.081836 lr=9.431528e-05 lr_backbone=9.431528e-05 elapsed=09:35:59 avg_step=3.53s recent_step=3.16s eta=39:28:44 +2026-09-14 23:23:04 | INFO | train rank=0 | step=9790/50000 progress=19.6% loss=0.092336 lr=9.429996e-05 lr_backbone=9.429996e-05 elapsed=09:36:37 avg_step=3.53s recent_step=3.79s eta=39:28:19 +2026-09-14 23:23:49 | INFO | train rank=0 | step=9800/50000 progress=19.6% loss=0.158786 lr=9.428462e-05 lr_backbone=9.428462e-05 elapsed=09:37:22 avg_step=3.53s recent_step=4.51s eta=39:28:24 +2026-09-14 23:24:16 | INFO | train rank=0 | step=9810/50000 progress=19.6% loss=0.121133 lr=9.426925e-05 lr_backbone=9.426925e-05 elapsed=09:37:49 avg_step=3.53s recent_step=2.68s eta=39:27:14 +2026-09-14 23:24:50 | INFO | train rank=0 | step=9820/50000 progress=19.6% loss=0.138981 lr=9.425387e-05 lr_backbone=9.425387e-05 elapsed=09:38:23 avg_step=3.53s recent_step=3.46s eta=39:26:35 +2026-09-14 23:25:42 | INFO | train rank=0 | step=9830/50000 progress=19.7% loss=0.124171 lr=9.423847e-05 lr_backbone=9.423847e-05 elapsed=09:39:15 avg_step=3.54s recent_step=5.16s eta=39:27:06 +2026-09-14 23:26:09 | INFO | train rank=0 | step=9840/50000 progress=19.7% loss=0.155061 lr=9.422305e-05 lr_backbone=9.422305e-05 elapsed=09:39:42 avg_step=3.53s recent_step=2.71s eta=39:25:57 +2026-09-14 23:26:47 | INFO | train rank=0 | step=9850/50000 progress=19.7% loss=0.136910 lr=9.420761e-05 lr_backbone=9.420761e-05 elapsed=09:40:20 avg_step=3.54s recent_step=3.79s eta=39:25:32 +2026-09-14 23:27:24 | INFO | train rank=0 | step=9860/50000 progress=19.7% loss=0.129344 lr=9.419215e-05 lr_backbone=9.419215e-05 elapsed=09:40:57 avg_step=3.54s recent_step=3.71s eta=39:25:04 +2026-09-14 23:28:14 | INFO | train rank=0 | step=9870/50000 progress=19.7% loss=0.123781 lr=9.417667e-05 lr_backbone=9.417667e-05 elapsed=09:41:47 avg_step=3.54s recent_step=5.03s eta=39:25:29 +2026-09-14 23:28:51 | INFO | train rank=0 | step=9880/50000 progress=19.8% loss=0.090509 lr=9.416117e-05 lr_backbone=9.416117e-05 elapsed=09:42:24 avg_step=3.54s recent_step=3.67s eta=39:24:59 +2026-09-14 23:29:25 | INFO | train rank=0 | step=9890/50000 progress=19.8% loss=0.098697 lr=9.414565e-05 lr_backbone=9.414565e-05 elapsed=09:42:57 avg_step=3.54s recent_step=3.35s eta=39:24:16 +2026-09-14 23:29:54 | INFO | train rank=0 | step=9900/50000 progress=19.8% loss=0.097775 lr=9.413012e-05 lr_backbone=9.413012e-05 elapsed=09:43:27 avg_step=3.54s recent_step=2.96s eta=39:23:18 +2026-09-14 23:30:42 | INFO | train rank=0 | step=9910/50000 progress=19.8% loss=0.086540 lr=9.411456e-05 lr_backbone=9.411456e-05 elapsed=09:44:14 avg_step=3.54s recent_step=4.72s eta=39:23:30 +2026-09-14 23:31:16 | INFO | train rank=0 | step=9920/50000 progress=19.8% loss=0.102500 lr=9.409899e-05 lr_backbone=9.409899e-05 elapsed=09:44:49 avg_step=3.54s recent_step=3.44s eta=39:22:51 +2026-09-14 23:31:56 | INFO | train rank=0 | step=9930/50000 progress=19.9% loss=0.120434 lr=9.408339e-05 lr_backbone=9.408339e-05 elapsed=09:45:29 avg_step=3.54s recent_step=4.01s eta=39:22:35 +2026-09-14 23:32:23 | INFO | train rank=0 | step=9940/50000 progress=19.9% loss=0.117814 lr=9.406778e-05 lr_backbone=9.406778e-05 elapsed=09:45:55 avg_step=3.54s recent_step=2.66s eta=39:21:24 +2026-09-14 23:33:02 | INFO | train rank=0 | step=9950/50000 progress=19.9% loss=0.079217 lr=9.405214e-05 lr_backbone=9.405214e-05 elapsed=09:46:34 avg_step=3.54s recent_step=3.89s eta=39:21:03 +2026-09-14 23:33:38 | INFO | train rank=0 | step=9960/50000 progress=19.9% loss=0.137843 lr=9.403649e-05 lr_backbone=9.403649e-05 elapsed=09:47:11 avg_step=3.54s recent_step=3.65s eta=39:20:32 +2026-09-14 23:34:08 | INFO | train rank=0 | step=9970/50000 progress=19.9% loss=0.087535 lr=9.402082e-05 lr_backbone=9.402082e-05 elapsed=09:47:41 avg_step=3.54s recent_step=2.99s eta=39:19:35 +2026-09-14 23:34:46 | INFO | train rank=0 | step=9980/50000 progress=20.0% loss=0.183501 lr=9.400513e-05 lr_backbone=9.400513e-05 elapsed=09:48:19 avg_step=3.54s recent_step=3.83s eta=39:19:11 +2026-09-14 23:35:20 | INFO | train rank=0 | step=9990/50000 progress=20.0% loss=0.105732 lr=9.398942e-05 lr_backbone=9.398942e-05 elapsed=09:48:52 avg_step=3.54s recent_step=3.33s eta=39:18:28 +2026-09-14 23:35:58 | INFO | train rank=0 | step=10000/50000 progress=20.0% loss=0.149167 lr=9.397369e-05 lr_backbone=9.397369e-05 elapsed=09:49:30 avg_step=3.54s recent_step=3.80s eta=39:18:03 +2026-09-14 23:36:29 | INFO | train rank=0 | step=10010/50000 progress=20.0% loss=0.152337 lr=9.395794e-05 lr_backbone=9.395794e-05 elapsed=09:50:02 avg_step=3.54s recent_step=3.16s eta=39:17:12 +2026-09-14 23:36:56 | INFO | train rank=0 | step=10020/50000 progress=20.0% loss=0.143458 lr=9.394217e-05 lr_backbone=9.394217e-05 elapsed=09:50:28 avg_step=3.54s recent_step=2.66s eta=39:16:02 +2026-09-14 23:37:37 | INFO | train rank=0 | step=10030/50000 progress=20.1% loss=0.077435 lr=9.392638e-05 lr_backbone=9.392638e-05 elapsed=09:51:09 avg_step=3.54s recent_step=4.09s eta=39:15:49 +2026-09-14 23:38:20 | INFO | train rank=0 | step=10040/50000 progress=20.1% loss=0.104373 lr=9.391058e-05 lr_backbone=9.391058e-05 elapsed=09:51:53 avg_step=3.54s recent_step=4.34s eta=39:15:45 +2026-09-14 23:38:57 | INFO | train rank=0 | step=10050/50000 progress=20.1% loss=0.134616 lr=9.389475e-05 lr_backbone=9.389475e-05 elapsed=09:52:30 avg_step=3.54s recent_step=3.69s eta=39:15:16 +2026-09-14 23:39:25 | INFO | train rank=0 | step=10060/50000 progress=20.1% loss=0.118639 lr=9.387891e-05 lr_backbone=9.387891e-05 elapsed=09:52:58 avg_step=3.54s recent_step=2.83s eta=39:14:13 +2026-09-14 23:40:06 | INFO | train rank=0 | step=10070/50000 progress=20.1% loss=0.113979 lr=9.386304e-05 lr_backbone=9.386304e-05 elapsed=09:53:39 avg_step=3.54s recent_step=4.08s eta=39:13:59 +2026-09-14 23:40:54 | INFO | train rank=0 | step=10080/50000 progress=20.2% loss=0.132897 lr=9.384716e-05 lr_backbone=9.384716e-05 elapsed=09:54:27 avg_step=3.54s recent_step=4.83s eta=39:14:15 +2026-09-14 23:41:27 | INFO | train rank=0 | step=10090/50000 progress=20.2% loss=0.152222 lr=9.383126e-05 lr_backbone=9.383126e-05 elapsed=09:55:00 avg_step=3.54s recent_step=3.27s eta=39:13:29 +2026-09-14 23:41:56 | INFO | train rank=0 | step=10100/50000 progress=20.2% loss=0.185888 lr=9.381533e-05 lr_backbone=9.381533e-05 elapsed=09:55:29 avg_step=3.54s recent_step=2.91s eta=39:12:28 +2026-09-14 23:42:31 | INFO | train rank=0 | step=10110/50000 progress=20.2% loss=0.138061 lr=9.379939e-05 lr_backbone=9.379939e-05 elapsed=09:56:04 avg_step=3.54s recent_step=3.50s eta=39:11:52 +2026-09-14 23:43:13 | INFO | train rank=0 | step=10120/50000 progress=20.2% loss=0.118199 lr=9.378343e-05 lr_backbone=9.378343e-05 elapsed=09:56:46 avg_step=3.54s recent_step=4.19s eta=39:11:42 +2026-09-14 23:43:44 | INFO | train rank=0 | step=10130/50000 progress=20.3% loss=0.158576 lr=9.376745e-05 lr_backbone=9.376745e-05 elapsed=09:57:17 avg_step=3.54s recent_step=3.08s eta=39:10:48 +2026-09-14 23:44:14 | INFO | train rank=0 | step=10140/50000 progress=20.3% loss=0.137872 lr=9.375146e-05 lr_backbone=9.375146e-05 elapsed=09:57:47 avg_step=3.54s recent_step=3.05s eta=39:09:54 +2026-09-14 23:45:04 | INFO | train rank=0 | step=10150/50000 progress=20.3% loss=0.096712 lr=9.373544e-05 lr_backbone=9.373544e-05 elapsed=09:58:37 avg_step=3.54s recent_step=4.95s eta=39:10:14 +2026-09-14 23:45:35 | INFO | train rank=0 | step=10160/50000 progress=20.3% loss=0.120004 lr=9.371940e-05 lr_backbone=9.371940e-05 elapsed=09:59:08 avg_step=3.54s recent_step=3.08s eta=39:09:21 +2026-09-14 23:46:12 | INFO | train rank=0 | step=10170/50000 progress=20.3% loss=0.099547 lr=9.370335e-05 lr_backbone=9.370335e-05 elapsed=09:59:44 avg_step=3.54s recent_step=3.68s eta=39:08:51 +2026-09-14 23:46:48 | INFO | train rank=0 | step=10180/50000 progress=20.4% loss=0.110942 lr=9.368727e-05 lr_backbone=9.368727e-05 elapsed=10:00:21 avg_step=3.54s recent_step=3.69s eta=39:08:22 +2026-09-14 23:47:23 | INFO | train rank=0 | step=10190/50000 progress=20.4% loss=0.106736 lr=9.367118e-05 lr_backbone=9.367118e-05 elapsed=10:00:56 avg_step=3.54s recent_step=3.50s eta=39:07:45 +2026-09-14 23:48:11 | INFO | train rank=0 | step=10200/50000 progress=20.4% loss=0.098772 lr=9.365506e-05 lr_backbone=9.365506e-05 elapsed=10:01:44 avg_step=3.54s recent_step=4.77s eta=39:07:57 +2026-09-14 23:48:46 | INFO | train rank=0 | step=10210/50000 progress=20.4% loss=0.118426 lr=9.363893e-05 lr_backbone=9.363893e-05 elapsed=10:02:19 avg_step=3.54s recent_step=3.53s eta=39:07:21 +2026-09-14 23:49:11 | INFO | train rank=0 | step=10220/50000 progress=20.4% loss=0.081323 lr=9.362278e-05 lr_backbone=9.362278e-05 elapsed=10:02:44 avg_step=3.54s recent_step=2.45s eta=39:06:04 +2026-09-14 23:50:02 | INFO | train rank=0 | step=10230/50000 progress=20.5% loss=0.107097 lr=9.360661e-05 lr_backbone=9.360661e-05 elapsed=10:03:35 avg_step=3.54s recent_step=5.14s eta=39:06:31 +2026-09-14 23:50:31 | INFO | train rank=0 | step=10240/50000 progress=20.5% loss=0.122924 lr=9.359042e-05 lr_backbone=9.359042e-05 elapsed=10:04:04 avg_step=3.54s recent_step=2.87s eta=39:05:29 +2026-09-14 23:51:13 | INFO | train rank=0 | step=10250/50000 progress=20.5% loss=0.134239 lr=9.357421e-05 lr_backbone=9.357421e-05 elapsed=10:04:45 avg_step=3.54s recent_step=4.17s eta=39:05:18 +2026-09-14 23:51:45 | INFO | train rank=0 | step=10260/50000 progress=20.5% loss=0.082064 lr=9.355798e-05 lr_backbone=9.355798e-05 elapsed=10:05:18 avg_step=3.54s recent_step=3.28s eta=39:04:33 +2026-09-14 23:52:45 | INFO | train rank=0 | step=10270/50000 progress=20.5% loss=0.106304 lr=9.354174e-05 lr_backbone=9.354174e-05 elapsed=10:06:18 avg_step=3.54s recent_step=6.00s eta=39:05:32 +2026-09-14 23:53:17 | INFO | train rank=0 | step=10280/50000 progress=20.6% loss=0.114409 lr=9.352547e-05 lr_backbone=9.352547e-05 elapsed=10:06:50 avg_step=3.54s recent_step=3.19s eta=39:04:43 +2026-09-14 23:53:47 | INFO | train rank=0 | step=10290/50000 progress=20.6% loss=0.094238 lr=9.350919e-05 lr_backbone=9.350919e-05 elapsed=10:07:19 avg_step=3.54s recent_step=2.92s eta=39:03:44 +2026-09-14 23:54:21 | INFO | train rank=0 | step=10300/50000 progress=20.6% loss=0.089653 lr=9.349288e-05 lr_backbone=9.349288e-05 elapsed=10:07:54 avg_step=3.54s recent_step=3.47s eta=39:03:06 +2026-09-14 23:55:04 | INFO | train rank=0 | step=10310/50000 progress=20.6% loss=0.147457 lr=9.347656e-05 lr_backbone=9.347656e-05 elapsed=10:08:37 avg_step=3.54s recent_step=4.29s eta=39:02:59 +2026-09-14 23:55:41 | INFO | train rank=0 | step=10320/50000 progress=20.6% loss=0.134323 lr=9.346022e-05 lr_backbone=9.346022e-05 elapsed=10:09:13 avg_step=3.54s recent_step=3.64s eta=39:02:27 +2026-09-14 23:56:06 | INFO | train rank=0 | step=10330/50000 progress=20.7% loss=0.110788 lr=9.344386e-05 lr_backbone=9.344386e-05 elapsed=10:09:39 avg_step=3.54s recent_step=2.58s eta=39:01:15 +2026-09-14 23:56:39 | INFO | train rank=0 | step=10340/50000 progress=20.7% loss=0.129127 lr=9.342748e-05 lr_backbone=9.342748e-05 elapsed=10:10:12 avg_step=3.54s recent_step=3.28s eta=39:00:30 +2026-09-14 23:57:19 | INFO | train rank=0 | step=10350/50000 progress=20.7% loss=0.116813 lr=9.341108e-05 lr_backbone=9.341108e-05 elapsed=10:10:52 avg_step=3.54s recent_step=4.03s eta=39:00:13 +2026-09-14 23:57:58 | INFO | train rank=0 | step=10360/50000 progress=20.7% loss=0.119053 lr=9.339466e-05 lr_backbone=9.339466e-05 elapsed=10:11:31 avg_step=3.54s recent_step=3.84s eta=38:59:49 +2026-09-14 23:58:34 | INFO | train rank=0 | step=10370/50000 progress=20.7% loss=0.108711 lr=9.337823e-05 lr_backbone=9.337823e-05 elapsed=10:12:07 avg_step=3.54s recent_step=3.58s eta=38:59:16 +2026-09-14 23:59:10 | INFO | train rank=0 | step=10380/50000 progress=20.8% loss=0.113446 lr=9.336177e-05 lr_backbone=9.336177e-05 elapsed=10:12:43 avg_step=3.54s recent_step=3.63s eta=38:58:44 +2026-09-15 00:00:00 | INFO | train rank=0 | step=10390/50000 progress=20.8% loss=0.090443 lr=9.334530e-05 lr_backbone=9.334530e-05 elapsed=10:13:33 avg_step=3.54s recent_step=5.03s eta=38:59:05 +2026-09-15 00:00:35 | INFO | train rank=0 | step=10400/50000 progress=20.8% loss=0.108737 lr=9.332880e-05 lr_backbone=9.332880e-05 elapsed=10:14:08 avg_step=3.54s recent_step=3.50s eta=38:58:28 +2026-09-15 00:01:09 | INFO | train rank=0 | step=10410/50000 progress=20.8% loss=0.126026 lr=9.331229e-05 lr_backbone=9.331229e-05 elapsed=10:14:42 avg_step=3.54s recent_step=3.36s eta=38:57:45 +2026-09-15 00:01:45 | INFO | train rank=0 | step=10420/50000 progress=20.8% loss=0.104952 lr=9.329576e-05 lr_backbone=9.329576e-05 elapsed=10:15:18 avg_step=3.54s recent_step=3.64s eta=38:57:14 +2026-09-15 00:02:39 | INFO | train rank=0 | step=10430/50000 progress=20.9% loss=0.111214 lr=9.327921e-05 lr_backbone=9.327921e-05 elapsed=10:16:12 avg_step=3.54s recent_step=5.40s eta=38:57:49 +2026-09-15 00:03:04 | INFO | train rank=0 | step=10440/50000 progress=20.9% loss=0.115171 lr=9.326264e-05 lr_backbone=9.326264e-05 elapsed=10:16:37 avg_step=3.54s recent_step=2.51s eta=38:56:34 +2026-09-15 00:03:39 | INFO | train rank=0 | step=10450/50000 progress=20.9% loss=0.085484 lr=9.324605e-05 lr_backbone=9.324605e-05 elapsed=10:17:12 avg_step=3.54s recent_step=3.48s eta=38:55:56 +2026-09-15 00:04:20 | INFO | train rank=0 | step=10460/50000 progress=20.9% loss=0.084857 lr=9.322944e-05 lr_backbone=9.322944e-05 elapsed=10:17:53 avg_step=3.54s recent_step=4.12s eta=38:55:42 +2026-09-15 00:04:54 | INFO | train rank=0 | step=10470/50000 progress=20.9% loss=0.130935 lr=9.321282e-05 lr_backbone=9.321282e-05 elapsed=10:18:27 avg_step=3.54s recent_step=3.36s eta=38:55:00 +2026-09-15 00:05:39 | INFO | train rank=0 | step=10480/50000 progress=21.0% loss=0.120739 lr=9.319617e-05 lr_backbone=9.319617e-05 elapsed=10:19:12 avg_step=3.55s recent_step=4.53s eta=38:55:02 +2026-09-15 00:06:06 | INFO | train rank=0 | step=10490/50000 progress=21.0% loss=0.130361 lr=9.317951e-05 lr_backbone=9.317951e-05 elapsed=10:19:39 avg_step=3.54s recent_step=2.69s eta=38:53:54 +2026-09-15 00:06:43 | INFO | train rank=0 | step=10500/50000 progress=21.0% loss=0.105240 lr=9.316282e-05 lr_backbone=9.316282e-05 elapsed=10:20:16 avg_step=3.54s recent_step=3.68s eta=38:53:23 +2026-09-15 00:07:24 | INFO | train rank=0 | step=10510/50000 progress=21.0% loss=0.117624 lr=9.314612e-05 lr_backbone=9.314612e-05 elapsed=10:20:56 avg_step=3.54s recent_step=4.07s eta=38:53:08 +2026-09-15 00:08:10 | INFO | train rank=0 | step=10520/50000 progress=21.0% loss=0.112112 lr=9.312940e-05 lr_backbone=9.312940e-05 elapsed=10:21:43 avg_step=3.55s recent_step=4.64s eta=38:53:14 +2026-09-15 00:08:39 | INFO | train rank=0 | step=10530/50000 progress=21.1% loss=0.092039 lr=9.311266e-05 lr_backbone=9.311266e-05 elapsed=10:22:12 avg_step=3.55s recent_step=2.89s eta=38:52:14 +2026-09-15 00:09:13 | INFO | train rank=0 | step=10540/50000 progress=21.1% loss=0.137471 lr=9.309590e-05 lr_backbone=9.309590e-05 elapsed=10:22:46 avg_step=3.55s recent_step=3.40s eta=38:51:33 +2026-09-15 00:09:38 | INFO | train rank=0 | step=10550/50000 progress=21.1% loss=0.122781 lr=9.307913e-05 lr_backbone=9.307913e-05 elapsed=10:23:11 avg_step=3.54s recent_step=2.50s eta=38:50:18 +2026-09-15 00:10:22 | INFO | train rank=0 | step=10560/50000 progress=21.1% loss=0.121846 lr=9.306233e-05 lr_backbone=9.306233e-05 elapsed=10:23:55 avg_step=3.55s recent_step=4.42s eta=38:50:15 +2026-09-15 00:10:51 | INFO | train rank=0 | step=10570/50000 progress=21.1% loss=0.104046 lr=9.304552e-05 lr_backbone=9.304552e-05 elapsed=10:24:24 avg_step=3.54s recent_step=2.86s eta=38:49:15 +2026-09-15 00:11:23 | INFO | train rank=0 | step=10580/50000 progress=21.2% loss=0.103066 lr=9.302868e-05 lr_backbone=9.302868e-05 elapsed=10:24:56 avg_step=3.54s recent_step=3.22s eta=38:48:27 +2026-09-15 00:12:07 | INFO | train rank=0 | step=10590/50000 progress=21.2% loss=0.158694 lr=9.301183e-05 lr_backbone=9.301183e-05 elapsed=10:25:40 avg_step=3.54s recent_step=4.40s eta=38:48:23 +2026-09-15 00:12:52 | INFO | train rank=0 | step=10600/50000 progress=21.2% loss=0.102132 lr=9.299496e-05 lr_backbone=9.299496e-05 elapsed=10:26:25 avg_step=3.55s recent_step=4.48s eta=38:48:23 +2026-09-15 00:13:19 | INFO | train rank=0 | step=10610/50000 progress=21.2% loss=0.131358 lr=9.297807e-05 lr_backbone=9.297807e-05 elapsed=10:26:52 avg_step=3.54s recent_step=2.72s eta=38:47:17 +2026-09-15 00:13:57 | INFO | train rank=0 | step=10620/50000 progress=21.2% loss=0.100015 lr=9.296116e-05 lr_backbone=9.296116e-05 elapsed=10:27:30 avg_step=3.55s recent_step=3.83s eta=38:46:52 +2026-09-15 00:14:52 | INFO | train rank=0 | step=10630/50000 progress=21.3% loss=0.114544 lr=9.294423e-05 lr_backbone=9.294423e-05 elapsed=10:28:25 avg_step=3.55s recent_step=5.49s eta=38:47:28 +2026-09-15 00:15:29 | INFO | train rank=0 | step=10640/50000 progress=21.3% loss=0.094676 lr=9.292728e-05 lr_backbone=9.292728e-05 elapsed=10:29:02 avg_step=3.55s recent_step=3.71s eta=38:46:59 +2026-09-15 00:16:04 | INFO | train rank=0 | step=10650/50000 progress=21.3% loss=0.128222 lr=9.291032e-05 lr_backbone=9.291032e-05 elapsed=10:29:37 avg_step=3.55s recent_step=3.49s eta=38:46:21 +2026-09-15 00:16:42 | INFO | train rank=0 | step=10660/50000 progress=21.3% loss=0.093177 lr=9.289334e-05 lr_backbone=9.289334e-05 elapsed=10:30:15 avg_step=3.55s recent_step=3.77s eta=38:45:54 +2026-09-15 00:17:27 | INFO | train rank=0 | step=10670/50000 progress=21.3% loss=0.123275 lr=9.287633e-05 lr_backbone=9.287633e-05 elapsed=10:31:00 avg_step=3.55s recent_step=4.55s eta=38:45:55 +2026-09-15 00:18:05 | INFO | train rank=0 | step=10680/50000 progress=21.4% loss=0.115263 lr=9.285931e-05 lr_backbone=9.285931e-05 elapsed=10:31:38 avg_step=3.55s recent_step=3.78s eta=38:45:28 +2026-09-15 00:18:36 | INFO | train rank=0 | step=10690/50000 progress=21.4% loss=0.111951 lr=9.284227e-05 lr_backbone=9.284227e-05 elapsed=10:32:09 avg_step=3.55s recent_step=3.12s eta=38:44:37 +2026-09-15 00:19:17 | INFO | train rank=0 | step=10700/50000 progress=21.4% loss=0.141908 lr=9.282521e-05 lr_backbone=9.282521e-05 elapsed=10:32:50 avg_step=3.55s recent_step=4.08s eta=38:44:21 +2026-09-15 00:20:02 | INFO | train rank=0 | step=10710/50000 progress=21.4% loss=0.145312 lr=9.280813e-05 lr_backbone=9.280813e-05 elapsed=10:33:35 avg_step=3.55s recent_step=4.48s eta=38:44:20 +2026-09-15 00:20:28 | INFO | train rank=0 | step=10720/50000 progress=21.4% loss=0.110066 lr=9.279104e-05 lr_backbone=9.279104e-05 elapsed=10:34:01 avg_step=3.55s recent_step=2.59s eta=38:43:09 +2026-09-15 00:21:05 | INFO | train rank=0 | step=10730/50000 progress=21.5% loss=0.091858 lr=9.277392e-05 lr_backbone=9.277392e-05 elapsed=10:34:38 avg_step=3.55s recent_step=3.75s eta=38:42:41 +2026-09-15 00:21:35 | INFO | train rank=0 | step=10740/50000 progress=21.5% loss=0.103660 lr=9.275679e-05 lr_backbone=9.275679e-05 elapsed=10:35:08 avg_step=3.55s recent_step=2.95s eta=38:41:44 +2026-09-15 00:22:23 | INFO | train rank=0 | step=10750/50000 progress=21.5% loss=0.113415 lr=9.273964e-05 lr_backbone=9.273964e-05 elapsed=10:35:56 avg_step=3.55s recent_step=4.81s eta=38:41:54 +2026-09-15 00:22:54 | INFO | train rank=0 | step=10760/50000 progress=21.5% loss=0.120379 lr=9.272246e-05 lr_backbone=9.272246e-05 elapsed=10:36:27 avg_step=3.55s recent_step=3.13s eta=38:41:04 +2026-09-15 00:23:30 | INFO | train rank=0 | step=10770/50000 progress=21.5% loss=0.093149 lr=9.270527e-05 lr_backbone=9.270527e-05 elapsed=10:37:02 avg_step=3.55s recent_step=3.53s eta=38:40:28 +2026-09-15 00:24:02 | INFO | train rank=0 | step=10780/50000 progress=21.6% loss=0.108833 lr=9.268807e-05 lr_backbone=9.268807e-05 elapsed=10:37:35 avg_step=3.55s recent_step=3.28s eta=38:39:42 +2026-09-15 00:24:48 | INFO | train rank=0 | step=10790/50000 progress=21.6% loss=0.139641 lr=9.267084e-05 lr_backbone=9.267084e-05 elapsed=10:38:21 avg_step=3.55s recent_step=4.57s eta=38:39:44 +2026-09-15 00:25:28 | INFO | train rank=0 | step=10800/50000 progress=21.6% loss=0.087807 lr=9.265359e-05 lr_backbone=9.265359e-05 elapsed=10:39:00 avg_step=3.55s recent_step=3.94s eta=38:39:23 +2026-09-15 00:26:06 | INFO | train rank=0 | step=10810/50000 progress=21.6% loss=0.106967 lr=9.263633e-05 lr_backbone=9.263633e-05 elapsed=10:39:39 avg_step=3.55s recent_step=3.87s eta=38:38:59 +2026-09-15 00:26:46 | INFO | train rank=0 | step=10820/50000 progress=21.6% loss=0.076298 lr=9.261904e-05 lr_backbone=9.261904e-05 elapsed=10:40:19 avg_step=3.55s recent_step=3.99s eta=38:38:39 +2026-09-15 00:27:45 | INFO | train rank=0 | step=10830/50000 progress=21.7% loss=0.154219 lr=9.260174e-05 lr_backbone=9.260174e-05 elapsed=10:41:17 avg_step=3.55s recent_step=5.85s eta=38:39:27 +2026-09-15 00:28:12 | INFO | train rank=0 | step=10840/50000 progress=21.7% loss=0.118443 lr=9.258442e-05 lr_backbone=9.258442e-05 elapsed=10:41:45 avg_step=3.55s recent_step=2.76s eta=38:38:23 +2026-09-15 00:28:49 | INFO | train rank=0 | step=10850/50000 progress=21.7% loss=0.106988 lr=9.256708e-05 lr_backbone=9.256708e-05 elapsed=10:42:22 avg_step=3.55s recent_step=3.72s eta=38:37:53 +2026-09-15 00:29:24 | INFO | train rank=0 | step=10860/50000 progress=21.7% loss=0.113953 lr=9.254972e-05 lr_backbone=9.254972e-05 elapsed=10:42:57 avg_step=3.55s recent_step=3.43s eta=38:37:13 +2026-09-15 00:29:59 | INFO | train rank=0 | step=10870/50000 progress=21.7% loss=0.129860 lr=9.253235e-05 lr_backbone=9.253235e-05 elapsed=10:43:31 avg_step=3.55s recent_step=3.47s eta=38:36:35 +2026-09-15 00:30:35 | INFO | train rank=0 | step=10880/50000 progress=21.8% loss=0.109170 lr=9.251495e-05 lr_backbone=9.251495e-05 elapsed=10:44:08 avg_step=3.55s recent_step=3.69s eta=38:36:04 +2026-09-15 00:31:06 | INFO | train rank=0 | step=10890/50000 progress=21.8% loss=0.119838 lr=9.249754e-05 lr_backbone=9.249754e-05 elapsed=10:44:38 avg_step=3.55s recent_step=3.01s eta=38:35:09 +2026-09-15 00:31:42 | INFO | train rank=0 | step=10900/50000 progress=21.8% loss=0.126689 lr=9.248011e-05 lr_backbone=9.248011e-05 elapsed=10:45:14 avg_step=3.55s recent_step=3.61s eta=38:34:36 +2026-09-15 00:32:28 | INFO | train rank=0 | step=10910/50000 progress=21.8% loss=0.123390 lr=9.246266e-05 lr_backbone=9.246266e-05 elapsed=10:46:00 avg_step=3.55s recent_step=4.60s eta=38:34:38 +2026-09-15 00:33:06 | INFO | train rank=0 | step=10920/50000 progress=21.8% loss=0.109667 lr=9.244519e-05 lr_backbone=9.244519e-05 elapsed=10:46:38 avg_step=3.55s recent_step=3.79s eta=38:34:11 +2026-09-15 00:33:33 | INFO | train rank=0 | step=10930/50000 progress=21.9% loss=0.102625 lr=9.242770e-05 lr_backbone=9.242770e-05 elapsed=10:47:06 avg_step=3.55s recent_step=2.72s eta=38:33:06 +2026-09-15 00:34:05 | INFO | train rank=0 | step=10940/50000 progress=21.9% loss=0.076303 lr=9.241019e-05 lr_backbone=9.241019e-05 elapsed=10:47:38 avg_step=3.55s recent_step=3.27s eta=38:32:20 +2026-09-15 00:35:01 | INFO | train rank=0 | step=10950/50000 progress=21.9% loss=0.111564 lr=9.239267e-05 lr_backbone=9.239267e-05 elapsed=10:48:33 avg_step=3.55s recent_step=5.51s eta=38:32:54 +2026-09-15 00:35:29 | INFO | train rank=0 | step=10960/50000 progress=21.9% loss=0.126274 lr=9.237512e-05 lr_backbone=9.237512e-05 elapsed=10:49:02 avg_step=3.55s recent_step=2.85s eta=38:31:54 +2026-09-15 00:36:02 | INFO | train rank=0 | step=10970/50000 progress=21.9% loss=0.113925 lr=9.235756e-05 lr_backbone=9.235756e-05 elapsed=10:49:35 avg_step=3.55s recent_step=3.34s eta=38:31:11 +2026-09-15 00:36:38 | INFO | train rank=0 | step=10980/50000 progress=22.0% loss=0.089837 lr=9.233998e-05 lr_backbone=9.233998e-05 elapsed=10:50:11 avg_step=3.55s recent_step=3.57s eta=38:30:36 +2026-09-15 00:37:25 | INFO | train rank=0 | step=10990/50000 progress=22.0% loss=0.096922 lr=9.232238e-05 lr_backbone=9.232238e-05 elapsed=10:50:58 avg_step=3.55s recent_step=4.72s eta=38:30:42 +2026-09-15 00:37:59 | INFO | train rank=0 | step=11000/50000 progress=22.0% loss=0.094351 lr=9.230476e-05 lr_backbone=9.230476e-05 elapsed=10:51:32 avg_step=3.55s recent_step=3.41s eta=38:30:01 +2026-09-15 00:38:27 | INFO | train rank=0 | step=11010/50000 progress=22.0% loss=0.118072 lr=9.228713e-05 lr_backbone=9.228713e-05 elapsed=10:52:00 avg_step=3.55s recent_step=2.73s eta=38:28:56 +2026-09-15 00:38:55 | INFO | train rank=0 | step=11020/50000 progress=22.0% loss=0.128310 lr=9.226947e-05 lr_backbone=9.226947e-05 elapsed=10:52:28 avg_step=3.55s recent_step=2.80s eta=38:27:54 +2026-09-15 00:39:53 | INFO | train rank=0 | step=11030/50000 progress=22.1% loss=0.087407 lr=9.225180e-05 lr_backbone=9.225180e-05 elapsed=10:53:26 avg_step=3.55s recent_step=5.84s eta=38:28:40 +2026-09-15 00:40:30 | INFO | train rank=0 | step=11040/50000 progress=22.1% loss=0.119908 lr=9.223411e-05 lr_backbone=9.223411e-05 elapsed=10:54:03 avg_step=3.55s recent_step=3.67s eta=38:28:08 +2026-09-15 00:41:00 | INFO | train rank=0 | step=11050/50000 progress=22.1% loss=0.088756 lr=9.221640e-05 lr_backbone=9.221640e-05 elapsed=10:54:33 avg_step=3.55s recent_step=3.06s eta=38:27:15 +2026-09-15 00:41:35 | INFO | train rank=0 | step=11060/50000 progress=22.1% loss=0.109208 lr=9.219867e-05 lr_backbone=9.219867e-05 elapsed=10:55:08 avg_step=3.55s recent_step=3.46s eta=38:26:36 +2026-09-15 00:42:19 | INFO | train rank=0 | step=11070/50000 progress=22.1% loss=0.108306 lr=9.218092e-05 lr_backbone=9.218092e-05 elapsed=10:55:52 avg_step=3.55s recent_step=4.37s eta=38:26:29 +2026-09-15 00:42:55 | INFO | train rank=0 | step=11080/50000 progress=22.2% loss=0.133725 lr=9.216315e-05 lr_backbone=9.216315e-05 elapsed=10:56:28 avg_step=3.55s recent_step=3.67s eta=38:25:58 +2026-09-15 00:43:22 | INFO | train rank=0 | step=11090/50000 progress=22.2% loss=0.113965 lr=9.214537e-05 lr_backbone=9.214537e-05 elapsed=10:56:55 avg_step=3.55s recent_step=2.70s eta=38:24:52 +2026-09-15 00:43:56 | INFO | train rank=0 | step=11100/50000 progress=22.2% loss=0.127615 lr=9.212757e-05 lr_backbone=9.212757e-05 elapsed=10:57:28 avg_step=3.55s recent_step=3.33s eta=38:24:09 +2026-09-15 00:44:53 | INFO | train rank=0 | step=11110/50000 progress=22.2% loss=0.115877 lr=9.210975e-05 lr_backbone=9.210975e-05 elapsed=10:58:26 avg_step=3.56s recent_step=5.77s eta=38:24:51 +2026-09-15 00:45:28 | INFO | train rank=0 | step=11120/50000 progress=22.2% loss=0.148324 lr=9.209191e-05 lr_backbone=9.209191e-05 elapsed=10:59:01 avg_step=3.56s recent_step=3.44s eta=38:24:11 +2026-09-15 00:46:03 | INFO | train rank=0 | step=11130/50000 progress=22.3% loss=0.116473 lr=9.207405e-05 lr_backbone=9.207405e-05 elapsed=10:59:36 avg_step=3.56s recent_step=3.53s eta=38:23:34 +2026-09-15 00:46:39 | INFO | train rank=0 | step=11140/50000 progress=22.3% loss=0.102018 lr=9.205617e-05 lr_backbone=9.205617e-05 elapsed=11:00:12 avg_step=3.56s recent_step=3.64s eta=38:23:02 +2026-09-15 00:47:26 | INFO | train rank=0 | step=11150/50000 progress=22.3% loss=0.100395 lr=9.203828e-05 lr_backbone=9.203828e-05 elapsed=11:00:59 avg_step=3.56s recent_step=4.65s eta=38:23:04 +2026-09-15 00:47:46 | INFO | train rank=0 | step=11160/50000 progress=22.3% loss=0.104281 lr=9.202037e-05 lr_backbone=9.202037e-05 elapsed=11:01:19 avg_step=3.56s recent_step=2.04s eta=38:21:36 +2026-09-15 00:48:28 | INFO | train rank=0 | step=11170/50000 progress=22.3% loss=0.096461 lr=9.200244e-05 lr_backbone=9.200244e-05 elapsed=11:02:00 avg_step=3.56s recent_step=4.13s eta=38:21:20 +2026-09-15 00:49:01 | INFO | train rank=0 | step=11180/50000 progress=22.4% loss=0.116870 lr=9.198449e-05 lr_backbone=9.198449e-05 elapsed=11:02:33 avg_step=3.56s recent_step=3.30s eta=38:20:36 +2026-09-15 00:50:04 | INFO | train rank=0 | step=11190/50000 progress=22.4% loss=0.096517 lr=9.196652e-05 lr_backbone=9.196652e-05 elapsed=11:03:36 avg_step=3.56s recent_step=6.31s eta=38:21:36 +2026-09-15 00:50:32 | INFO | train rank=0 | step=11200/50000 progress=22.4% loss=0.130438 lr=9.194853e-05 lr_backbone=9.194853e-05 elapsed=11:04:05 avg_step=3.56s recent_step=2.82s eta=38:20:35 +2026-09-15 00:51:08 | INFO | train rank=0 | step=11210/50000 progress=22.4% loss=0.105250 lr=9.193053e-05 lr_backbone=9.193053e-05 elapsed=11:04:41 avg_step=3.56s recent_step=3.58s eta=38:20:00 +2026-09-15 00:51:40 | INFO | train rank=0 | step=11220/50000 progress=22.4% loss=0.125796 lr=9.191250e-05 lr_backbone=9.191250e-05 elapsed=11:05:12 avg_step=3.56s recent_step=3.18s eta=38:19:12 +2026-09-15 00:52:27 | INFO | train rank=0 | step=11230/50000 progress=22.5% loss=0.124501 lr=9.189446e-05 lr_backbone=9.189446e-05 elapsed=11:05:59 avg_step=3.56s recent_step=4.70s eta=38:19:15 +2026-09-15 00:52:55 | INFO | train rank=0 | step=11240/50000 progress=22.5% loss=0.145673 lr=9.187640e-05 lr_backbone=9.187640e-05 elapsed=11:06:27 avg_step=3.56s recent_step=2.80s eta=38:18:14 +2026-09-15 00:53:27 | INFO | train rank=0 | step=11250/50000 progress=22.5% loss=0.095991 lr=9.185832e-05 lr_backbone=9.185832e-05 elapsed=11:06:59 avg_step=3.56s recent_step=3.19s eta=38:17:26 +2026-09-15 00:54:05 | INFO | train rank=0 | step=11260/50000 progress=22.5% loss=0.135022 lr=9.184023e-05 lr_backbone=9.184023e-05 elapsed=11:07:37 avg_step=3.56s recent_step=3.81s eta=38:16:59 +2026-09-15 00:54:53 | INFO | train rank=0 | step=11270/50000 progress=22.5% loss=0.109683 lr=9.182211e-05 lr_backbone=9.182211e-05 elapsed=11:08:26 avg_step=3.56s recent_step=4.86s eta=38:17:08 +2026-09-15 00:55:24 | INFO | train rank=0 | step=11280/50000 progress=22.6% loss=0.085488 lr=9.180398e-05 lr_backbone=9.180398e-05 elapsed=11:08:57 avg_step=3.56s recent_step=3.06s eta=38:16:15 +2026-09-15 00:56:05 | INFO | train rank=0 | step=11290/50000 progress=22.6% loss=0.095322 lr=9.178583e-05 lr_backbone=9.178583e-05 elapsed=11:09:38 avg_step=3.56s recent_step=4.12s eta=38:15:59 +2026-09-15 00:56:34 | INFO | train rank=0 | step=11300/50000 progress=22.6% loss=0.094279 lr=9.176766e-05 lr_backbone=9.176766e-05 elapsed=11:10:07 avg_step=3.56s recent_step=2.92s eta=38:15:02 +2026-09-15 00:57:20 | INFO | train rank=0 | step=11310/50000 progress=22.6% loss=0.123092 lr=9.174947e-05 lr_backbone=9.174947e-05 elapsed=11:10:53 avg_step=3.56s recent_step=4.62s eta=38:15:02 +2026-09-15 00:57:54 | INFO | train rank=0 | step=11320/50000 progress=22.6% loss=0.126802 lr=9.173127e-05 lr_backbone=9.173127e-05 elapsed=11:11:27 avg_step=3.56s recent_step=3.38s eta=38:14:20 +2026-09-15 00:58:29 | INFO | train rank=0 | step=11330/50000 progress=22.7% loss=0.106017 lr=9.171304e-05 lr_backbone=9.171304e-05 elapsed=11:12:02 avg_step=3.56s recent_step=3.45s eta=38:13:41 +2026-09-15 00:59:01 | INFO | train rank=0 | step=11340/50000 progress=22.7% loss=0.136625 lr=9.169480e-05 lr_backbone=9.169480e-05 elapsed=11:12:33 avg_step=3.56s recent_step=3.18s eta=38:12:53 +2026-09-15 00:59:47 | INFO | train rank=0 | step=11350/50000 progress=22.7% loss=0.145494 lr=9.167654e-05 lr_backbone=9.167654e-05 elapsed=11:13:20 avg_step=3.56s recent_step=4.62s eta=38:12:53 +2026-09-15 01:00:20 | INFO | train rank=0 | step=11360/50000 progress=22.7% loss=0.164649 lr=9.165826e-05 lr_backbone=9.165826e-05 elapsed=11:13:52 avg_step=3.56s recent_step=3.29s eta=38:12:09 +2026-09-15 01:00:54 | INFO | train rank=0 | step=11370/50000 progress=22.7% loss=0.083397 lr=9.163996e-05 lr_backbone=9.163996e-05 elapsed=11:14:26 avg_step=3.56s recent_step=3.38s eta=38:11:27 +2026-09-15 01:01:29 | INFO | train rank=0 | step=11380/50000 progress=22.8% loss=0.129511 lr=9.162165e-05 lr_backbone=9.162165e-05 elapsed=11:15:02 avg_step=3.56s recent_step=3.56s eta=38:10:51 +2026-09-15 01:02:18 | INFO | train rank=0 | step=11390/50000 progress=22.8% loss=0.134641 lr=9.160331e-05 lr_backbone=9.160331e-05 elapsed=11:15:51 avg_step=3.56s recent_step=4.87s eta=38:11:00 +2026-09-15 01:02:49 | INFO | train rank=0 | step=11400/50000 progress=22.8% loss=0.114926 lr=9.158496e-05 lr_backbone=9.158496e-05 elapsed=11:16:22 avg_step=3.56s recent_step=3.17s eta=38:10:11 +2026-09-15 01:03:17 | INFO | train rank=0 | step=11410/50000 progress=22.8% loss=0.182391 lr=9.156659e-05 lr_backbone=9.156659e-05 elapsed=11:16:50 avg_step=3.56s recent_step=2.77s eta=38:09:09 +2026-09-15 01:03:50 | INFO | train rank=0 | step=11420/50000 progress=22.8% loss=0.124611 lr=9.154820e-05 lr_backbone=9.154820e-05 elapsed=11:17:23 avg_step=3.56s recent_step=3.34s eta=38:08:26 +2026-09-15 01:04:31 | INFO | train rank=0 | step=11430/50000 progress=22.9% loss=0.125450 lr=9.152979e-05 lr_backbone=9.152979e-05 elapsed=11:18:04 avg_step=3.56s recent_step=4.09s eta=38:08:08 +2026-09-15 01:04:58 | INFO | train rank=0 | step=11440/50000 progress=22.9% loss=0.103828 lr=9.151137e-05 lr_backbone=9.151137e-05 elapsed=11:18:31 avg_step=3.56s recent_step=2.63s eta=38:07:01 +2026-09-15 01:05:36 | INFO | train rank=0 | step=11450/50000 progress=22.9% loss=0.101611 lr=9.149293e-05 lr_backbone=9.149293e-05 elapsed=11:19:09 avg_step=3.56s recent_step=3.82s eta=38:06:34 +2026-09-15 01:06:17 | INFO | train rank=0 | step=11460/50000 progress=22.9% loss=0.166904 lr=9.147447e-05 lr_backbone=9.147447e-05 elapsed=11:19:50 avg_step=3.56s recent_step=4.08s eta=38:06:16 +2026-09-15 01:07:08 | INFO | train rank=0 | step=11470/50000 progress=22.9% loss=0.098599 lr=9.145599e-05 lr_backbone=9.145599e-05 elapsed=11:20:41 avg_step=3.56s recent_step=5.16s eta=38:06:35 +2026-09-15 01:07:40 | INFO | train rank=0 | step=11480/50000 progress=23.0% loss=0.112359 lr=9.143749e-05 lr_backbone=9.143749e-05 elapsed=11:21:13 avg_step=3.56s recent_step=3.14s eta=38:05:45 +2026-09-15 01:08:12 | INFO | train rank=0 | step=11490/50000 progress=23.0% loss=0.131786 lr=9.141898e-05 lr_backbone=9.141898e-05 elapsed=11:21:44 avg_step=3.56s recent_step=3.18s eta=38:04:56 +2026-09-15 01:08:45 | INFO | train rank=0 | step=11500/50000 progress=23.0% loss=0.120421 lr=9.140044e-05 lr_backbone=9.140044e-05 elapsed=11:22:17 avg_step=3.56s recent_step=3.31s eta=38:04:13 +2026-09-15 01:09:32 | INFO | train rank=0 | step=11510/50000 progress=23.0% loss=0.115843 lr=9.138189e-05 lr_backbone=9.138189e-05 elapsed=11:23:05 avg_step=3.56s recent_step=4.71s eta=38:04:16 +2026-09-15 01:09:57 | INFO | train rank=0 | step=11520/50000 progress=23.0% loss=0.106692 lr=9.136332e-05 lr_backbone=9.136332e-05 elapsed=11:23:29 avg_step=3.56s recent_step=2.48s eta=38:03:04 +2026-09-15 01:10:30 | INFO | train rank=0 | step=11530/50000 progress=23.1% loss=0.127771 lr=9.134473e-05 lr_backbone=9.134473e-05 elapsed=11:24:03 avg_step=3.56s recent_step=3.32s eta=38:02:20 +2026-09-15 01:11:02 | INFO | train rank=0 | step=11540/50000 progress=23.1% loss=0.120037 lr=9.132613e-05 lr_backbone=9.132613e-05 elapsed=11:24:35 avg_step=3.56s recent_step=3.25s eta=38:01:34 +2026-09-15 01:11:46 | INFO | train rank=0 | step=11550/50000 progress=23.1% loss=0.096664 lr=9.130750e-05 lr_backbone=9.130750e-05 elapsed=11:25:19 avg_step=3.56s recent_step=4.34s eta=38:01:25 +2026-09-15 01:12:18 | INFO | train rank=0 | step=11560/50000 progress=23.1% loss=0.091469 lr=9.128886e-05 lr_backbone=9.128886e-05 elapsed=11:25:51 avg_step=3.56s recent_step=3.24s eta=38:00:39 +2026-09-15 01:12:58 | INFO | train rank=0 | step=11570/50000 progress=23.1% loss=0.102981 lr=9.127020e-05 lr_backbone=9.127020e-05 elapsed=11:26:30 avg_step=3.56s recent_step=3.95s eta=38:00:16 +2026-09-15 01:13:39 | INFO | train rank=0 | step=11580/50000 progress=23.2% loss=0.131186 lr=9.125152e-05 lr_backbone=9.125152e-05 elapsed=11:27:12 avg_step=3.56s recent_step=4.19s eta=38:00:01 +2026-09-15 01:14:24 | INFO | train rank=0 | step=11590/50000 progress=23.2% loss=0.099205 lr=9.123283e-05 lr_backbone=9.123283e-05 elapsed=11:27:57 avg_step=3.56s recent_step=4.48s eta=37:59:56 +2026-09-15 01:14:59 | INFO | train rank=0 | step=11600/50000 progress=23.2% loss=0.083363 lr=9.121411e-05 lr_backbone=9.121411e-05 elapsed=11:28:32 avg_step=3.56s recent_step=3.47s eta=37:59:17 +2026-09-15 01:15:31 | INFO | train rank=0 | step=11610/50000 progress=23.2% loss=0.112798 lr=9.119538e-05 lr_backbone=9.119538e-05 elapsed=11:29:04 avg_step=3.56s recent_step=3.21s eta=37:58:30 +2026-09-15 01:16:02 | INFO | train rank=0 | step=11620/50000 progress=23.2% loss=0.096214 lr=9.117663e-05 lr_backbone=9.117663e-05 elapsed=11:29:35 avg_step=3.56s recent_step=3.14s eta=37:57:40 +2026-09-15 01:16:43 | INFO | train rank=0 | step=11630/50000 progress=23.3% loss=0.116937 lr=9.115786e-05 lr_backbone=9.115786e-05 elapsed=11:30:16 avg_step=3.56s recent_step=4.07s eta=37:57:22 +2026-09-15 01:17:11 | INFO | train rank=0 | step=11640/50000 progress=23.3% loss=0.093730 lr=9.113908e-05 lr_backbone=9.113908e-05 elapsed=11:30:44 avg_step=3.56s recent_step=2.77s eta=37:56:20 +2026-09-15 01:18:02 | INFO | train rank=0 | step=11650/50000 progress=23.3% loss=0.095033 lr=9.112027e-05 lr_backbone=9.112027e-05 elapsed=11:31:35 avg_step=3.56s recent_step=5.11s eta=37:56:35 +2026-09-15 01:18:40 | INFO | train rank=0 | step=11660/50000 progress=23.3% loss=0.093400 lr=9.110145e-05 lr_backbone=9.110145e-05 elapsed=11:32:13 avg_step=3.56s recent_step=3.82s eta=37:56:08 +2026-09-15 01:19:21 | INFO | train rank=0 | step=11670/50000 progress=23.3% loss=0.130548 lr=9.108261e-05 lr_backbone=9.108261e-05 elapsed=11:32:54 avg_step=3.56s recent_step=4.07s eta=37:55:49 +2026-09-15 01:19:53 | INFO | train rank=0 | step=11680/50000 progress=23.4% loss=0.095335 lr=9.106375e-05 lr_backbone=9.106375e-05 elapsed=11:33:26 avg_step=3.56s recent_step=3.26s eta=37:55:04 +2026-09-15 01:20:34 | INFO | train rank=0 | step=11690/50000 progress=23.4% loss=0.137474 lr=9.104487e-05 lr_backbone=9.104487e-05 elapsed=11:34:07 avg_step=3.56s recent_step=4.09s eta=37:54:45 +2026-09-15 01:21:18 | INFO | train rank=0 | step=11700/50000 progress=23.4% loss=0.110377 lr=9.102598e-05 lr_backbone=9.102598e-05 elapsed=11:34:51 avg_step=3.56s recent_step=4.38s eta=37:54:36 +2026-09-15 01:21:55 | INFO | train rank=0 | step=11710/50000 progress=23.4% loss=0.103336 lr=9.100707e-05 lr_backbone=9.100707e-05 elapsed=11:35:28 avg_step=3.56s recent_step=3.69s eta=37:54:05 +2026-09-15 01:22:23 | INFO | train rank=0 | step=11720/50000 progress=23.4% loss=0.127138 lr=9.098814e-05 lr_backbone=9.098814e-05 elapsed=11:35:56 avg_step=3.56s recent_step=2.79s eta=37:53:04 +2026-09-15 01:22:59 | INFO | train rank=0 | step=11730/50000 progress=23.5% loss=0.119105 lr=9.096919e-05 lr_backbone=9.096919e-05 elapsed=11:36:32 avg_step=3.56s recent_step=3.63s eta=37:52:30 +2026-09-15 01:23:37 | INFO | train rank=0 | step=11740/50000 progress=23.5% loss=0.134265 lr=9.095022e-05 lr_backbone=9.095022e-05 elapsed=11:37:09 avg_step=3.56s recent_step=3.75s eta=37:52:01 +2026-09-15 01:24:20 | INFO | train rank=0 | step=11750/50000 progress=23.5% loss=0.109153 lr=9.093124e-05 lr_backbone=9.093124e-05 elapsed=11:37:52 avg_step=3.56s recent_step=4.31s eta=37:51:49 +2026-09-15 01:24:53 | INFO | train rank=0 | step=11760/50000 progress=23.5% loss=0.125800 lr=9.091224e-05 lr_backbone=9.091224e-05 elapsed=11:38:25 avg_step=3.56s recent_step=3.30s eta=37:51:05 +2026-09-15 01:25:31 | INFO | train rank=0 | step=11770/50000 progress=23.5% loss=0.102329 lr=9.089322e-05 lr_backbone=9.089322e-05 elapsed=11:39:04 avg_step=3.56s recent_step=3.84s eta=37:50:39 +2026-09-15 01:26:08 | INFO | train rank=0 | step=11780/50000 progress=23.6% loss=0.117612 lr=9.087418e-05 lr_backbone=9.087418e-05 elapsed=11:39:41 avg_step=3.56s recent_step=3.69s eta=37:50:07 +2026-09-15 01:26:56 | INFO | train rank=0 | step=11790/50000 progress=23.6% loss=0.099156 lr=9.085513e-05 lr_backbone=9.085513e-05 elapsed=11:40:29 avg_step=3.56s recent_step=4.81s eta=37:50:12 +2026-09-15 01:27:22 | INFO | train rank=0 | step=11800/50000 progress=23.6% loss=0.114470 lr=9.083605e-05 lr_backbone=9.083605e-05 elapsed=11:40:55 avg_step=3.56s recent_step=2.56s eta=37:49:04 +2026-09-15 01:27:49 | INFO | train rank=0 | step=11810/50000 progress=23.6% loss=0.124225 lr=9.081696e-05 lr_backbone=9.081696e-05 elapsed=11:41:22 avg_step=3.56s recent_step=2.74s eta=37:48:02 +2026-09-15 01:28:23 | INFO | train rank=0 | step=11820/50000 progress=23.6% loss=0.101724 lr=9.079785e-05 lr_backbone=9.079785e-05 elapsed=11:41:56 avg_step=3.56s recent_step=3.41s eta=37:47:21 +2026-09-15 01:29:11 | INFO | train rank=0 | step=11830/50000 progress=23.7% loss=0.121837 lr=9.077873e-05 lr_backbone=9.077873e-05 elapsed=11:42:44 avg_step=3.56s recent_step=4.78s eta=37:47:25 +2026-09-15 01:29:47 | INFO | train rank=0 | step=11840/50000 progress=23.7% loss=0.146553 lr=9.075958e-05 lr_backbone=9.075958e-05 elapsed=11:43:20 avg_step=3.56s recent_step=3.56s eta=37:46:49 +2026-09-15 01:30:30 | INFO | train rank=0 | step=11850/50000 progress=23.7% loss=0.115904 lr=9.074042e-05 lr_backbone=9.074042e-05 elapsed=11:44:02 avg_step=3.56s recent_step=4.28s eta=37:46:36 +2026-09-15 01:31:07 | INFO | train rank=0 | step=11860/50000 progress=23.7% loss=0.094698 lr=9.072124e-05 lr_backbone=9.072124e-05 elapsed=11:44:40 avg_step=3.56s recent_step=3.72s eta=37:46:06 +2026-09-15 01:32:00 | INFO | train rank=0 | step=11870/50000 progress=23.7% loss=0.102084 lr=9.070204e-05 lr_backbone=9.070204e-05 elapsed=11:45:33 avg_step=3.57s recent_step=5.35s eta=37:46:27 +2026-09-15 01:32:28 | INFO | train rank=0 | step=11880/50000 progress=23.8% loss=0.130500 lr=9.068283e-05 lr_backbone=9.068283e-05 elapsed=11:46:01 avg_step=3.57s recent_step=2.78s eta=37:45:27 +2026-09-15 01:32:59 | INFO | train rank=0 | step=11890/50000 progress=23.8% loss=0.122719 lr=9.066359e-05 lr_backbone=9.066359e-05 elapsed=11:46:31 avg_step=3.57s recent_step=3.05s eta=37:44:34 +2026-09-15 01:33:34 | INFO | train rank=0 | step=11900/50000 progress=23.8% loss=0.147902 lr=9.064434e-05 lr_backbone=9.064434e-05 elapsed=11:47:06 avg_step=3.57s recent_step=3.50s eta=37:43:57 +2026-09-15 01:34:28 | INFO | train rank=0 | step=11910/50000 progress=23.8% loss=0.087157 lr=9.062507e-05 lr_backbone=9.062507e-05 elapsed=11:48:01 avg_step=3.57s recent_step=5.46s eta=37:44:22 +2026-09-15 01:34:59 | INFO | train rank=0 | step=11920/50000 progress=23.8% loss=0.160811 lr=9.060578e-05 lr_backbone=9.060578e-05 elapsed=11:48:32 avg_step=3.57s recent_step=3.08s eta=37:43:30 +2026-09-15 01:35:27 | INFO | train rank=0 | step=11930/50000 progress=23.9% loss=0.120109 lr=9.058648e-05 lr_backbone=9.058648e-05 elapsed=11:49:00 avg_step=3.57s recent_step=2.82s eta=37:42:31 +2026-09-15 01:35:53 | INFO | train rank=0 | step=11940/50000 progress=23.9% loss=0.100648 lr=9.056716e-05 lr_backbone=9.056716e-05 elapsed=11:49:26 avg_step=3.57s recent_step=2.57s eta=37:41:24 +2026-09-15 01:36:46 | INFO | train rank=0 | step=11950/50000 progress=23.9% loss=0.083647 lr=9.054782e-05 lr_backbone=9.054782e-05 elapsed=11:50:18 avg_step=3.57s recent_step=5.28s eta=37:41:42 +2026-09-15 01:37:24 | INFO | train rank=0 | step=11960/50000 progress=23.9% loss=0.133697 lr=9.052846e-05 lr_backbone=9.052846e-05 elapsed=11:50:57 avg_step=3.57s recent_step=3.84s eta=37:41:15 +2026-09-15 01:37:53 | INFO | train rank=0 | step=11970/50000 progress=23.9% loss=0.099514 lr=9.050908e-05 lr_backbone=9.050908e-05 elapsed=11:51:26 avg_step=3.57s recent_step=2.92s eta=37:40:19 +2026-09-15 01:38:35 | INFO | train rank=0 | step=11980/50000 progress=24.0% loss=0.126984 lr=9.048969e-05 lr_backbone=9.048969e-05 elapsed=11:52:07 avg_step=3.57s recent_step=4.14s eta=37:40:02 +2026-09-15 01:39:26 | INFO | train rank=0 | step=11990/50000 progress=24.0% loss=0.114704 lr=9.047028e-05 lr_backbone=9.047028e-05 elapsed=11:52:58 avg_step=3.57s recent_step=5.09s eta=37:40:14 +2026-09-15 01:39:59 | INFO | train rank=0 | step=12000/50000 progress=24.0% loss=0.095091 lr=9.045085e-05 lr_backbone=9.045085e-05 elapsed=11:53:32 avg_step=3.57s recent_step=3.37s eta=37:39:33 +2026-09-15 01:40:42 | INFO | train rank=0 | step=12010/50000 progress=24.0% loss=0.126551 lr=9.043140e-05 lr_backbone=9.043140e-05 elapsed=11:54:15 avg_step=3.57s recent_step=4.25s eta=37:39:18 +2026-09-15 01:41:07 | INFO | train rank=0 | step=12020/50000 progress=24.0% loss=0.096860 lr=9.041194e-05 lr_backbone=9.041194e-05 elapsed=11:54:39 avg_step=3.57s recent_step=2.49s eta=37:38:09 +2026-09-15 01:41:56 | INFO | train rank=0 | step=12030/50000 progress=24.1% loss=0.156182 lr=9.039246e-05 lr_backbone=9.039246e-05 elapsed=11:55:29 avg_step=3.57s recent_step=4.91s eta=37:38:15 +2026-09-15 01:42:28 | INFO | train rank=0 | step=12040/50000 progress=24.1% loss=0.167740 lr=9.037296e-05 lr_backbone=9.037296e-05 elapsed=11:56:01 avg_step=3.57s recent_step=3.26s eta=37:37:30 +2026-09-15 01:42:57 | INFO | train rank=0 | step=12050/50000 progress=24.1% loss=0.151101 lr=9.035344e-05 lr_backbone=9.035344e-05 elapsed=11:56:30 avg_step=3.57s recent_step=2.86s eta=37:36:32 +2026-09-15 01:43:28 | INFO | train rank=0 | step=12060/50000 progress=24.1% loss=0.084093 lr=9.033391e-05 lr_backbone=9.033391e-05 elapsed=11:57:01 avg_step=3.57s recent_step=3.13s eta=37:35:43 +2026-09-15 01:44:07 | INFO | train rank=0 | step=12070/50000 progress=24.1% loss=0.096645 lr=9.031435e-05 lr_backbone=9.031435e-05 elapsed=11:57:40 avg_step=3.57s recent_step=3.86s eta=37:35:16 +2026-09-15 01:44:38 | INFO | train rank=0 | step=12080/50000 progress=24.2% loss=0.116749 lr=9.029478e-05 lr_backbone=9.029478e-05 elapsed=11:58:11 avg_step=3.57s recent_step=3.09s eta=37:34:25 +2026-09-15 01:45:17 | INFO | train rank=0 | step=12090/50000 progress=24.2% loss=0.101716 lr=9.027519e-05 lr_backbone=9.027519e-05 elapsed=11:58:50 avg_step=3.57s recent_step=3.96s eta=37:34:02 +2026-09-15 01:45:53 | INFO | train rank=0 | step=12100/50000 progress=24.2% loss=0.109170 lr=9.025559e-05 lr_backbone=9.025559e-05 elapsed=11:59:25 avg_step=3.57s recent_step=3.51s eta=37:33:25 +2026-09-15 01:46:36 | INFO | train rank=0 | step=12110/50000 progress=24.2% loss=0.112145 lr=9.023597e-05 lr_backbone=9.023597e-05 elapsed=12:00:09 avg_step=3.57s recent_step=4.34s eta=37:33:14 +2026-09-15 01:47:12 | INFO | train rank=0 | step=12120/50000 progress=24.2% loss=0.118521 lr=9.021633e-05 lr_backbone=9.021633e-05 elapsed=12:00:44 avg_step=3.57s recent_step=3.56s eta=37:32:38 +2026-09-15 01:47:40 | INFO | train rank=0 | step=12130/50000 progress=24.3% loss=0.075168 lr=9.019667e-05 lr_backbone=9.019667e-05 elapsed=12:01:13 avg_step=3.57s recent_step=2.88s eta=37:31:40 +2026-09-15 01:48:17 | INFO | train rank=0 | step=12140/50000 progress=24.3% loss=0.126409 lr=9.017699e-05 lr_backbone=9.017699e-05 elapsed=12:01:50 avg_step=3.57s recent_step=3.65s eta=37:31:07 +2026-09-15 01:49:12 | INFO | train rank=0 | step=12150/50000 progress=24.3% loss=0.093387 lr=9.015730e-05 lr_backbone=9.015730e-05 elapsed=12:02:45 avg_step=3.57s recent_step=5.52s eta=37:31:32 +2026-09-15 01:49:46 | INFO | train rank=0 | step=12160/50000 progress=24.3% loss=0.105941 lr=9.013759e-05 lr_backbone=9.013759e-05 elapsed=12:03:19 avg_step=3.57s recent_step=3.40s eta=37:30:51 +2026-09-15 01:50:17 | INFO | train rank=0 | step=12170/50000 progress=24.3% loss=0.143674 lr=9.011786e-05 lr_backbone=9.011786e-05 elapsed=12:03:50 avg_step=3.57s recent_step=3.07s eta=37:30:00 +2026-09-15 01:50:54 | INFO | train rank=0 | step=12180/50000 progress=24.4% loss=0.096774 lr=9.009811e-05 lr_backbone=9.009811e-05 elapsed=12:04:27 avg_step=3.57s recent_step=3.72s eta=37:29:29 +2026-09-15 01:51:36 | INFO | train rank=0 | step=12190/50000 progress=24.4% loss=0.103544 lr=9.007835e-05 lr_backbone=9.007835e-05 elapsed=12:05:09 avg_step=3.57s recent_step=4.21s eta=37:29:13 +2026-09-15 01:52:01 | INFO | train rank=0 | step=12200/50000 progress=24.4% loss=0.124716 lr=9.005857e-05 lr_backbone=9.005857e-05 elapsed=12:05:34 avg_step=3.57s recent_step=2.53s eta=37:28:05 +2026-09-15 01:52:37 | INFO | train rank=0 | step=12210/50000 progress=24.4% loss=0.096921 lr=9.003877e-05 lr_backbone=9.003877e-05 elapsed=12:06:10 avg_step=3.57s recent_step=3.61s eta=37:27:31 +2026-09-15 01:53:15 | INFO | train rank=0 | step=12220/50000 progress=24.4% loss=0.102096 lr=9.001895e-05 lr_backbone=9.001895e-05 elapsed=12:06:48 avg_step=3.57s recent_step=3.81s eta=37:27:03 +2026-09-15 01:54:00 | INFO | train rank=0 | step=12230/50000 progress=24.5% loss=0.090370 lr=8.999912e-05 lr_backbone=8.999912e-05 elapsed=12:07:33 avg_step=3.57s recent_step=4.49s eta=37:26:55 +2026-09-15 01:54:34 | INFO | train rank=0 | step=12240/50000 progress=24.5% loss=0.118373 lr=8.997927e-05 lr_backbone=8.997927e-05 elapsed=12:08:07 avg_step=3.57s recent_step=3.41s eta=37:26:15 +2026-09-15 01:55:06 | INFO | train rank=0 | step=12250/50000 progress=24.5% loss=0.151463 lr=8.995940e-05 lr_backbone=8.995940e-05 elapsed=12:08:39 avg_step=3.57s recent_step=3.19s eta=37:25:27 +2026-09-15 01:55:33 | INFO | train rank=0 | step=12260/50000 progress=24.5% loss=0.095148 lr=8.993951e-05 lr_backbone=8.993951e-05 elapsed=12:09:06 avg_step=3.57s recent_step=2.70s eta=37:24:25 +2026-09-15 01:56:25 | INFO | train rank=0 | step=12270/50000 progress=24.5% loss=0.123361 lr=8.991961e-05 lr_backbone=8.991961e-05 elapsed=12:09:57 avg_step=3.57s recent_step=5.13s eta=37:24:37 +2026-09-15 01:57:02 | INFO | train rank=0 | step=12280/50000 progress=24.6% loss=0.088985 lr=8.989969e-05 lr_backbone=8.989969e-05 elapsed=12:10:34 avg_step=3.57s recent_step=3.69s eta=37:24:05 +2026-09-15 01:57:31 | INFO | train rank=0 | step=12290/50000 progress=24.6% loss=0.118422 lr=8.987975e-05 lr_backbone=8.987975e-05 elapsed=12:11:04 avg_step=3.57s recent_step=2.93s eta=37:23:10 +2026-09-15 01:58:07 | INFO | train rank=0 | step=12300/50000 progress=24.6% loss=0.113333 lr=8.985979e-05 lr_backbone=8.985979e-05 elapsed=12:11:40 avg_step=3.57s recent_step=3.64s eta=37:22:36 +2026-09-15 01:59:05 | INFO | train rank=0 | step=12310/50000 progress=24.6% loss=0.081257 lr=8.983982e-05 lr_backbone=8.983982e-05 elapsed=12:12:38 avg_step=3.57s recent_step=5.78s eta=37:23:08 +2026-09-15 01:59:40 | INFO | train rank=0 | step=12320/50000 progress=24.6% loss=0.112105 lr=8.981983e-05 lr_backbone=8.981983e-05 elapsed=12:13:13 avg_step=3.57s recent_step=3.50s eta=37:22:31 +2026-09-15 02:00:06 | INFO | train rank=0 | step=12330/50000 progress=24.7% loss=0.102379 lr=8.979982e-05 lr_backbone=8.979982e-05 elapsed=12:13:39 avg_step=3.57s recent_step=2.58s eta=37:21:25 +2026-09-15 02:00:43 | INFO | train rank=0 | step=12340/50000 progress=24.7% loss=0.132199 lr=8.977980e-05 lr_backbone=8.977980e-05 elapsed=12:14:16 avg_step=3.57s recent_step=3.68s eta=37:20:52 +2026-09-15 02:01:33 | INFO | train rank=0 | step=12350/50000 progress=24.7% loss=0.105347 lr=8.975975e-05 lr_backbone=8.975975e-05 elapsed=12:15:06 avg_step=3.57s recent_step=5.06s eta=37:21:02 +2026-09-15 02:02:05 | INFO | train rank=0 | step=12360/50000 progress=24.7% loss=0.118866 lr=8.973969e-05 lr_backbone=8.973969e-05 elapsed=12:15:38 avg_step=3.57s recent_step=3.15s eta=37:20:14 +2026-09-15 02:02:34 | INFO | train rank=0 | step=12370/50000 progress=24.7% loss=0.136031 lr=8.971962e-05 lr_backbone=8.971962e-05 elapsed=12:16:07 avg_step=3.57s recent_step=2.91s eta=37:19:18 +2026-09-15 02:03:01 | INFO | train rank=0 | step=12380/50000 progress=24.8% loss=0.128180 lr=8.969952e-05 lr_backbone=8.969952e-05 elapsed=12:16:34 avg_step=3.57s recent_step=2.69s eta=37:18:15 +2026-09-15 02:03:46 | INFO | train rank=0 | step=12390/50000 progress=24.8% loss=0.110618 lr=8.967941e-05 lr_backbone=8.967941e-05 elapsed=12:17:19 avg_step=3.57s recent_step=4.52s eta=37:18:08 +2026-09-15 02:04:24 | INFO | train rank=0 | step=12400/50000 progress=24.8% loss=0.118682 lr=8.965928e-05 lr_backbone=8.965928e-05 elapsed=12:17:57 avg_step=3.57s recent_step=3.77s eta=37:17:39 +2026-09-15 02:04:51 | INFO | train rank=0 | step=12410/50000 progress=24.8% loss=0.111162 lr=8.963913e-05 lr_backbone=8.963913e-05 elapsed=12:18:24 avg_step=3.57s recent_step=2.71s eta=37:16:37 +2026-09-15 02:05:21 | INFO | train rank=0 | step=12420/50000 progress=24.8% loss=0.133366 lr=8.961897e-05 lr_backbone=8.961897e-05 elapsed=12:18:53 avg_step=3.57s recent_step=2.99s eta=37:15:44 +2026-09-15 02:06:07 | INFO | train rank=0 | step=12430/50000 progress=24.9% loss=0.095707 lr=8.959878e-05 lr_backbone=8.959878e-05 elapsed=12:19:40 avg_step=3.57s recent_step=4.64s eta=37:15:40 +2026-09-15 02:06:39 | INFO | train rank=0 | step=12440/50000 progress=24.9% loss=0.096611 lr=8.957858e-05 lr_backbone=8.957858e-05 elapsed=12:20:11 avg_step=3.57s recent_step=3.16s eta=37:14:52 +2026-09-15 02:07:12 | INFO | train rank=0 | step=12450/50000 progress=24.9% loss=0.113132 lr=8.955837e-05 lr_backbone=8.955837e-05 elapsed=12:20:45 avg_step=3.57s recent_step=3.31s eta=37:14:09 +2026-09-15 02:07:48 | INFO | train rank=0 | step=12460/50000 progress=24.9% loss=0.091856 lr=8.953813e-05 lr_backbone=8.953813e-05 elapsed=12:21:20 avg_step=3.57s recent_step=3.59s eta=37:13:34 +2026-09-15 02:08:39 | INFO | train rank=0 | step=12470/50000 progress=24.9% loss=0.078538 lr=8.951788e-05 lr_backbone=8.951788e-05 elapsed=12:22:12 avg_step=3.57s recent_step=5.11s eta=37:13:44 +2026-09-15 02:09:12 | INFO | train rank=0 | step=12480/50000 progress=25.0% loss=0.110564 lr=8.949761e-05 lr_backbone=8.949761e-05 elapsed=12:22:44 avg_step=3.57s recent_step=3.29s eta=37:13:00 +2026-09-15 02:09:50 | INFO | train rank=0 | step=12490/50000 progress=25.0% loss=0.107129 lr=8.947733e-05 lr_backbone=8.947733e-05 elapsed=12:23:23 avg_step=3.57s recent_step=3.86s eta=37:12:33 +2026-09-15 02:10:24 | INFO | train rank=0 | step=12500/50000 progress=25.0% loss=0.135922 lr=8.945703e-05 lr_backbone=8.945703e-05 elapsed=12:23:56 avg_step=3.57s recent_step=3.34s eta=37:11:50 +2026-09-15 02:11:16 | INFO | train rank=0 | step=12510/50000 progress=25.0% loss=0.107701 lr=8.943671e-05 lr_backbone=8.943671e-05 elapsed=12:24:48 avg_step=3.57s recent_step=5.20s eta=37:12:03 +2026-09-15 02:11:48 | INFO | train rank=0 | step=12520/50000 progress=25.0% loss=0.080804 lr=8.941637e-05 lr_backbone=8.941637e-05 elapsed=12:25:20 avg_step=3.57s recent_step=3.19s eta=37:11:16 +2026-09-15 02:12:31 | INFO | train rank=0 | step=12530/50000 progress=25.1% loss=0.094948 lr=8.939601e-05 lr_backbone=8.939601e-05 elapsed=12:26:03 avg_step=3.57s recent_step=4.30s eta=37:11:02 +2026-09-15 02:13:04 | INFO | train rank=0 | step=12540/50000 progress=25.1% loss=0.094437 lr=8.937564e-05 lr_backbone=8.937564e-05 elapsed=12:26:37 avg_step=3.57s recent_step=3.34s eta=37:10:19 +2026-09-15 02:14:00 | INFO | train rank=0 | step=12550/50000 progress=25.1% loss=0.112815 lr=8.935525e-05 lr_backbone=8.935525e-05 elapsed=12:27:33 avg_step=3.57s recent_step=5.60s eta=37:10:44 +2026-09-15 02:14:37 | INFO | train rank=0 | step=12560/50000 progress=25.1% loss=0.166404 lr=8.933485e-05 lr_backbone=8.933485e-05 elapsed=12:28:09 avg_step=3.57s recent_step=3.66s eta=37:10:11 +2026-09-15 02:15:03 | INFO | train rank=0 | step=12570/50000 progress=25.1% loss=0.108824 lr=8.931442e-05 lr_backbone=8.931442e-05 elapsed=12:28:35 avg_step=3.57s recent_step=2.62s eta=37:09:07 +2026-09-15 02:15:35 | INFO | train rank=0 | step=12580/50000 progress=25.2% loss=0.117277 lr=8.929398e-05 lr_backbone=8.929398e-05 elapsed=12:29:08 avg_step=3.57s recent_step=3.25s eta=37:08:22 +2026-09-15 02:16:22 | INFO | train rank=0 | step=12590/50000 progress=25.2% loss=0.100081 lr=8.927352e-05 lr_backbone=8.927352e-05 elapsed=12:29:54 avg_step=3.57s recent_step=4.65s eta=37:08:18 +2026-09-15 02:16:55 | INFO | train rank=0 | step=12600/50000 progress=25.2% loss=0.138411 lr=8.925305e-05 lr_backbone=8.925305e-05 elapsed=12:30:27 avg_step=3.57s recent_step=3.29s eta=37:07:34 +2026-09-15 02:17:33 | INFO | train rank=0 | step=12610/50000 progress=25.2% loss=0.157423 lr=8.923256e-05 lr_backbone=8.923256e-05 elapsed=12:31:06 avg_step=3.57s recent_step=3.86s eta=37:07:06 +2026-09-15 02:18:07 | INFO | train rank=0 | step=12620/50000 progress=25.2% loss=0.103881 lr=8.921205e-05 lr_backbone=8.921205e-05 elapsed=12:31:39 avg_step=3.57s recent_step=3.35s eta=37:06:24 +2026-09-15 02:18:52 | INFO | train rank=0 | step=12630/50000 progress=25.3% loss=0.115003 lr=8.919152e-05 lr_backbone=8.919152e-05 elapsed=12:32:25 avg_step=3.57s recent_step=4.54s eta=37:06:16 +2026-09-15 02:19:20 | INFO | train rank=0 | step=12640/50000 progress=25.3% loss=0.152663 lr=8.917098e-05 lr_backbone=8.917098e-05 elapsed=12:32:53 avg_step=3.57s recent_step=2.79s eta=37:05:18 +2026-09-15 02:19:51 | INFO | train rank=0 | step=12650/50000 progress=25.3% loss=0.099590 lr=8.915041e-05 lr_backbone=8.915041e-05 elapsed=12:33:24 avg_step=3.57s recent_step=3.12s eta=37:04:28 +2026-09-15 02:20:32 | INFO | train rank=0 | step=12660/50000 progress=25.3% loss=0.107617 lr=8.912984e-05 lr_backbone=8.912984e-05 elapsed=12:34:04 avg_step=3.57s recent_step=4.06s eta=37:04:07 +2026-09-15 02:21:21 | INFO | train rank=0 | step=12670/50000 progress=25.3% loss=0.080758 lr=8.910924e-05 lr_backbone=8.910924e-05 elapsed=12:34:54 avg_step=3.57s recent_step=4.93s eta=37:04:11 +2026-09-15 02:21:54 | INFO | train rank=0 | step=12680/50000 progress=25.4% loss=0.103390 lr=8.908863e-05 lr_backbone=8.908863e-05 elapsed=12:35:27 avg_step=3.57s recent_step=3.31s eta=37:03:28 +2026-09-15 02:22:27 | INFO | train rank=0 | step=12690/50000 progress=25.4% loss=0.146465 lr=8.906800e-05 lr_backbone=8.906800e-05 elapsed=12:36:00 avg_step=3.57s recent_step=3.27s eta=37:02:43 +2026-09-15 02:23:01 | INFO | train rank=0 | step=12700/50000 progress=25.4% loss=0.116390 lr=8.904735e-05 lr_backbone=8.904735e-05 elapsed=12:36:34 avg_step=3.57s recent_step=3.45s eta=37:02:04 +2026-09-15 02:23:47 | INFO | train rank=0 | step=12710/50000 progress=25.4% loss=0.111141 lr=8.902669e-05 lr_backbone=8.902669e-05 elapsed=12:37:20 avg_step=3.58s recent_step=4.61s eta=37:01:59 +2026-09-15 02:24:21 | INFO | train rank=0 | step=12720/50000 progress=25.4% loss=0.130085 lr=8.900601e-05 lr_backbone=8.900601e-05 elapsed=12:37:54 avg_step=3.58s recent_step=3.39s eta=37:01:18 +2026-09-15 02:24:58 | INFO | train rank=0 | step=12730/50000 progress=25.5% loss=0.125289 lr=8.898531e-05 lr_backbone=8.898531e-05 elapsed=12:38:31 avg_step=3.58s recent_step=3.65s eta=37:00:44 +2026-09-15 02:25:34 | INFO | train rank=0 | step=12740/50000 progress=25.5% loss=0.103801 lr=8.896460e-05 lr_backbone=8.896460e-05 elapsed=12:39:07 avg_step=3.58s recent_step=3.59s eta=37:00:09 +2026-09-15 02:26:21 | INFO | train rank=0 | step=12750/50000 progress=25.5% loss=0.098476 lr=8.894386e-05 lr_backbone=8.894386e-05 elapsed=12:39:54 avg_step=3.58s recent_step=4.70s eta=37:00:06 +2026-09-15 02:26:53 | INFO | train rank=0 | step=12760/50000 progress=25.5% loss=0.137062 lr=8.892312e-05 lr_backbone=8.892312e-05 elapsed=12:40:25 avg_step=3.58s recent_step=3.19s eta=36:59:19 +2026-09-15 02:27:28 | INFO | train rank=0 | step=12770/50000 progress=25.5% loss=0.092585 lr=8.890235e-05 lr_backbone=8.890235e-05 elapsed=12:41:00 avg_step=3.58s recent_step=3.49s eta=36:58:40 +2026-09-15 02:28:04 | INFO | train rank=0 | step=12780/50000 progress=25.6% loss=0.117594 lr=8.888157e-05 lr_backbone=8.888157e-05 elapsed=12:41:37 avg_step=3.58s recent_step=3.64s eta=36:58:06 +2026-09-15 02:28:53 | INFO | train rank=0 | step=12790/50000 progress=25.6% loss=0.111980 lr=8.886077e-05 lr_backbone=8.886077e-05 elapsed=12:42:25 avg_step=3.58s recent_step=4.87s eta=36:58:08 +2026-09-15 02:29:14 | INFO | train rank=0 | step=12800/50000 progress=25.6% loss=0.109071 lr=8.883995e-05 lr_backbone=8.883995e-05 elapsed=12:42:47 avg_step=3.58s recent_step=2.18s eta=36:56:52 +2026-09-15 02:29:48 | INFO | train rank=0 | step=12810/50000 progress=25.6% loss=0.118969 lr=8.881912e-05 lr_backbone=8.881912e-05 elapsed=12:43:21 avg_step=3.58s recent_step=3.38s eta=36:56:11 +2026-09-15 02:30:16 | INFO | train rank=0 | step=12820/50000 progress=25.6% loss=0.112187 lr=8.879826e-05 lr_backbone=8.879826e-05 elapsed=12:43:49 avg_step=3.57s recent_step=2.75s eta=36:55:11 +2026-09-15 02:31:09 | INFO | train rank=0 | step=12830/50000 progress=25.7% loss=0.109632 lr=8.877740e-05 lr_backbone=8.877740e-05 elapsed=12:44:42 avg_step=3.58s recent_step=5.33s eta=36:55:26 +2026-09-15 02:31:35 | INFO | train rank=0 | step=12840/50000 progress=25.7% loss=0.123909 lr=8.875651e-05 lr_backbone=8.875651e-05 elapsed=12:45:08 avg_step=3.58s recent_step=2.65s eta=36:54:23 +2026-09-15 02:31:59 | INFO | train rank=0 | step=12850/50000 progress=25.7% loss=0.122848 lr=8.873561e-05 lr_backbone=8.873561e-05 elapsed=12:45:32 avg_step=3.57s recent_step=2.33s eta=36:53:11 +2026-09-15 02:32:30 | INFO | train rank=0 | step=12860/50000 progress=25.7% loss=0.119252 lr=8.871469e-05 lr_backbone=8.871469e-05 elapsed=12:46:03 avg_step=3.57s recent_step=3.13s eta=36:52:23 +2026-09-15 02:33:18 | INFO | train rank=0 | step=12870/50000 progress=25.7% loss=0.126770 lr=8.869376e-05 lr_backbone=8.869376e-05 elapsed=12:46:51 avg_step=3.58s recent_step=4.77s eta=36:52:21 +2026-09-15 02:33:50 | INFO | train rank=0 | step=12880/50000 progress=25.8% loss=0.115845 lr=8.867280e-05 lr_backbone=8.867280e-05 elapsed=12:47:23 avg_step=3.57s recent_step=3.25s eta=36:51:36 +2026-09-15 02:34:28 | INFO | train rank=0 | step=12890/50000 progress=25.8% loss=0.115299 lr=8.865183e-05 lr_backbone=8.865183e-05 elapsed=12:48:01 avg_step=3.57s recent_step=3.81s eta=36:51:07 +2026-09-15 02:34:56 | INFO | train rank=0 | step=12900/50000 progress=25.8% loss=0.104626 lr=8.863085e-05 lr_backbone=8.863085e-05 elapsed=12:48:29 avg_step=3.57s recent_step=2.74s eta=36:50:08 +2026-09-15 02:35:41 | INFO | train rank=0 | step=12910/50000 progress=25.8% loss=0.078808 lr=8.860984e-05 lr_backbone=8.860984e-05 elapsed=12:49:14 avg_step=3.58s recent_step=4.50s eta=36:49:59 +2026-09-15 02:36:04 | INFO | train rank=0 | step=12920/50000 progress=25.8% loss=0.158620 lr=8.858882e-05 lr_backbone=8.858882e-05 elapsed=12:49:37 avg_step=3.57s recent_step=2.37s eta=36:48:48 +2026-09-15 02:36:41 | INFO | train rank=0 | step=12930/50000 progress=25.9% loss=0.096497 lr=8.856779e-05 lr_backbone=8.856779e-05 elapsed=12:50:14 avg_step=3.57s recent_step=3.68s eta=36:48:15 +2026-09-15 02:37:14 | INFO | train rank=0 | step=12940/50000 progress=25.9% loss=0.172275 lr=8.854673e-05 lr_backbone=8.854673e-05 elapsed=12:50:47 avg_step=3.57s recent_step=3.33s eta=36:47:33 +2026-09-15 02:38:09 | INFO | train rank=0 | step=12950/50000 progress=25.9% loss=0.097497 lr=8.852566e-05 lr_backbone=8.852566e-05 elapsed=12:51:42 avg_step=3.58s recent_step=5.45s eta=36:47:51 +2026-09-15 02:38:39 | INFO | train rank=0 | step=12960/50000 progress=25.9% loss=0.149886 lr=8.850457e-05 lr_backbone=8.850457e-05 elapsed=12:52:12 avg_step=3.58s recent_step=2.99s eta=36:46:58 +2026-09-15 02:39:14 | INFO | train rank=0 | step=12970/50000 progress=25.9% loss=0.092361 lr=8.848347e-05 lr_backbone=8.848347e-05 elapsed=12:52:47 avg_step=3.57s recent_step=3.54s eta=36:46:21 +2026-09-15 02:39:49 | INFO | train rank=0 | step=12980/50000 progress=26.0% loss=0.117346 lr=8.846235e-05 lr_backbone=8.846235e-05 elapsed=12:53:22 avg_step=3.57s recent_step=3.46s eta=36:45:42 +2026-09-15 02:40:23 | INFO | train rank=0 | step=12990/50000 progress=26.0% loss=0.116240 lr=8.844121e-05 lr_backbone=8.844121e-05 elapsed=12:53:56 avg_step=3.57s recent_step=3.40s eta=36:45:01 +2026-09-15 02:40:58 | INFO | train rank=0 | step=13000/50000 progress=26.0% loss=0.098570 lr=8.842006e-05 lr_backbone=8.842006e-05 elapsed=12:54:30 avg_step=3.57s recent_step=3.47s eta=36:44:23 +2026-09-15 02:41:31 | INFO | train rank=0 | step=13010/50000 progress=26.0% loss=0.106834 lr=8.839888e-05 lr_backbone=8.839888e-05 elapsed=12:55:04 avg_step=3.57s recent_step=3.32s eta=36:43:40 +2026-09-15 02:42:03 | INFO | train rank=0 | step=13020/50000 progress=26.0% loss=0.084635 lr=8.837770e-05 lr_backbone=8.837770e-05 elapsed=12:55:36 avg_step=3.57s recent_step=3.24s eta=36:42:54 +2026-09-15 02:42:37 | INFO | train rank=0 | step=13030/50000 progress=26.1% loss=0.113822 lr=8.835649e-05 lr_backbone=8.835649e-05 elapsed=12:56:10 avg_step=3.57s recent_step=3.43s eta=36:42:15 +2026-09-15 02:43:12 | INFO | train rank=0 | step=13040/50000 progress=26.1% loss=0.143241 lr=8.833527e-05 lr_backbone=8.833527e-05 elapsed=12:56:45 avg_step=3.57s recent_step=3.49s eta=36:41:37 +2026-09-15 02:43:56 | INFO | train rank=0 | step=13050/50000 progress=26.1% loss=0.131747 lr=8.831403e-05 lr_backbone=8.831403e-05 elapsed=12:57:29 avg_step=3.57s recent_step=4.37s eta=36:41:23 +2026-09-15 02:44:23 | INFO | train rank=0 | step=13060/50000 progress=26.1% loss=0.122704 lr=8.829277e-05 lr_backbone=8.829277e-05 elapsed=12:57:56 avg_step=3.57s recent_step=2.73s eta=36:40:24 +2026-09-15 02:45:00 | INFO | train rank=0 | step=13070/50000 progress=26.1% loss=0.099205 lr=8.827150e-05 lr_backbone=8.827150e-05 elapsed=12:58:33 avg_step=3.57s recent_step=3.69s eta=36:39:51 +2026-09-15 02:45:23 | INFO | train rank=0 | step=13080/50000 progress=26.2% loss=0.089269 lr=8.825021e-05 lr_backbone=8.825021e-05 elapsed=12:58:56 avg_step=3.57s recent_step=2.29s eta=36:38:39 +2026-09-15 02:46:03 | INFO | train rank=0 | step=13090/50000 progress=26.2% loss=0.140013 lr=8.822891e-05 lr_backbone=8.822891e-05 elapsed=12:59:36 avg_step=3.57s recent_step=3.96s eta=36:38:14 +2026-09-15 02:46:39 | INFO | train rank=0 | step=13100/50000 progress=26.2% loss=0.105869 lr=8.820758e-05 lr_backbone=8.820758e-05 elapsed=13:00:12 avg_step=3.57s recent_step=3.64s eta=36:37:41 +2026-09-15 02:47:12 | INFO | train rank=0 | step=13110/50000 progress=26.2% loss=0.116451 lr=8.818624e-05 lr_backbone=8.818624e-05 elapsed=13:00:44 avg_step=3.57s recent_step=3.25s eta=36:36:56 +2026-09-15 02:47:51 | INFO | train rank=0 | step=13120/50000 progress=26.2% loss=0.112288 lr=8.816489e-05 lr_backbone=8.816489e-05 elapsed=13:01:24 avg_step=3.57s recent_step=3.94s eta=36:36:30 +2026-09-15 02:48:39 | INFO | train rank=0 | step=13130/50000 progress=26.3% loss=0.118928 lr=8.814351e-05 lr_backbone=8.814351e-05 elapsed=13:02:12 avg_step=3.57s recent_step=4.79s eta=36:36:29 +2026-09-15 02:49:20 | INFO | train rank=0 | step=13140/50000 progress=26.3% loss=0.109397 lr=8.812213e-05 lr_backbone=8.812213e-05 elapsed=13:02:53 avg_step=3.57s recent_step=4.11s eta=36:36:08 +2026-09-15 02:50:10 | INFO | train rank=0 | step=13150/50000 progress=26.3% loss=0.090632 lr=8.810072e-05 lr_backbone=8.810072e-05 elapsed=13:03:42 avg_step=3.58s recent_step=4.95s eta=36:36:11 +2026-09-15 02:51:14 | INFO | train rank=0 | step=13160/50000 progress=26.3% loss=0.094085 lr=8.807930e-05 lr_backbone=8.807930e-05 elapsed=13:04:47 avg_step=3.58s recent_step=6.42s eta=36:36:55 +2026-09-15 02:52:19 | INFO | train rank=0 | step=13170/50000 progress=26.3% loss=0.103862 lr=8.805786e-05 lr_backbone=8.805786e-05 elapsed=13:05:51 avg_step=3.58s recent_step=6.49s eta=36:37:40 +2026-09-15 02:53:24 | INFO | train rank=0 | step=13180/50000 progress=26.4% loss=0.103064 lr=8.803640e-05 lr_backbone=8.803640e-05 elapsed=13:06:57 avg_step=3.58s recent_step=6.58s eta=36:38:28 +2026-09-15 02:54:35 | INFO | train rank=0 | step=13190/50000 progress=26.4% loss=0.114525 lr=8.801493e-05 lr_backbone=8.801493e-05 elapsed=13:08:08 avg_step=3.59s recent_step=7.03s eta=36:39:29 +2026-09-15 02:55:45 | INFO | train rank=0 | step=13200/50000 progress=26.4% loss=0.091639 lr=8.799344e-05 lr_backbone=8.799344e-05 elapsed=13:09:18 avg_step=3.59s recent_step=7.03s eta=36:40:29 +2026-09-15 02:56:49 | INFO | train rank=0 | step=13210/50000 progress=26.4% loss=0.119366 lr=8.797193e-05 lr_backbone=8.797193e-05 elapsed=13:10:22 avg_step=3.59s recent_step=6.37s eta=36:41:11 +2026-09-15 02:57:54 | INFO | train rank=0 | step=13220/50000 progress=26.4% loss=0.118105 lr=8.795041e-05 lr_backbone=8.795041e-05 elapsed=13:11:27 avg_step=3.59s recent_step=6.52s eta=36:41:56 +2026-09-15 02:58:50 | INFO | train rank=0 | step=13230/50000 progress=26.5% loss=0.087124 lr=8.792887e-05 lr_backbone=8.792887e-05 elapsed=13:12:22 avg_step=3.59s recent_step=5.56s eta=36:42:15 +2026-09-15 02:59:57 | INFO | train rank=0 | step=13240/50000 progress=26.5% loss=0.114392 lr=8.790732e-05 lr_backbone=8.790732e-05 elapsed=13:13:30 avg_step=3.60s recent_step=6.79s eta=36:43:08 +2026-09-15 03:01:00 | INFO | train rank=0 | step=13250/50000 progress=26.5% loss=0.110624 lr=8.788574e-05 lr_backbone=8.788574e-05 elapsed=13:14:32 avg_step=3.60s recent_step=6.21s eta=36:43:44 +2026-09-15 03:02:16 | INFO | train rank=0 | step=13260/50000 progress=26.5% loss=0.089079 lr=8.786415e-05 lr_backbone=8.786415e-05 elapsed=13:15:49 avg_step=3.60s recent_step=7.68s eta=36:45:01 +2026-09-15 03:03:28 | INFO | train rank=0 | step=13270/50000 progress=26.5% loss=0.116880 lr=8.784255e-05 lr_backbone=8.784255e-05 elapsed=13:17:01 avg_step=3.60s recent_step=7.20s eta=36:46:05 +2026-09-15 03:04:30 | INFO | train rank=0 | step=13280/50000 progress=26.6% loss=0.086164 lr=8.782093e-05 lr_backbone=8.782093e-05 elapsed=13:18:03 avg_step=3.61s recent_step=6.14s eta=36:46:39 +2026-09-15 03:05:34 | INFO | train rank=0 | step=13290/50000 progress=26.6% loss=0.143691 lr=8.779929e-05 lr_backbone=8.779929e-05 elapsed=13:19:07 avg_step=3.61s recent_step=6.41s eta=36:47:21 +2026-09-15 03:06:53 | INFO | train rank=0 | step=13300/50000 progress=26.6% loss=0.103210 lr=8.777763e-05 lr_backbone=8.777763e-05 elapsed=13:20:26 avg_step=3.61s recent_step=7.93s eta=36:48:44 +2026-09-15 03:08:01 | INFO | train rank=0 | step=13310/50000 progress=26.6% loss=0.103568 lr=8.775596e-05 lr_backbone=8.775596e-05 elapsed=13:21:34 avg_step=3.61s recent_step=6.78s eta=36:49:35 +2026-09-15 03:09:11 | INFO | train rank=0 | step=13320/50000 progress=26.6% loss=0.099281 lr=8.773427e-05 lr_backbone=8.773427e-05 elapsed=13:22:44 avg_step=3.62s recent_step=7.00s eta=36:50:32 +2026-09-15 03:10:16 | INFO | train rank=0 | step=13330/50000 progress=26.7% loss=0.097939 lr=8.771257e-05 lr_backbone=8.771257e-05 elapsed=13:23:49 avg_step=3.62s recent_step=6.49s eta=36:51:15 +2026-09-15 03:11:20 | INFO | train rank=0 | step=13340/50000 progress=26.7% loss=0.106018 lr=8.769085e-05 lr_backbone=8.769085e-05 elapsed=13:24:53 avg_step=3.62s recent_step=6.42s eta=36:51:56 +2026-09-15 03:12:32 | INFO | train rank=0 | step=13350/50000 progress=26.7% loss=0.133977 lr=8.766911e-05 lr_backbone=8.766911e-05 elapsed=13:26:05 avg_step=3.62s recent_step=7.20s eta=36:52:58 +2026-09-15 03:13:37 | INFO | train rank=0 | step=13360/50000 progress=26.7% loss=0.123506 lr=8.764736e-05 lr_backbone=8.764736e-05 elapsed=13:27:09 avg_step=3.62s recent_step=6.45s eta=36:53:39 +2026-09-15 03:14:34 | INFO | train rank=0 | step=13370/50000 progress=26.7% loss=0.102271 lr=8.762559e-05 lr_backbone=8.762559e-05 elapsed=13:28:07 avg_step=3.63s recent_step=5.79s eta=36:54:02 +2026-09-15 03:15:38 | INFO | train rank=0 | step=13380/50000 progress=26.8% loss=0.119130 lr=8.760380e-05 lr_backbone=8.760380e-05 elapsed=13:29:11 avg_step=3.63s recent_step=6.34s eta=36:54:40 +2026-09-15 03:16:46 | INFO | train rank=0 | step=13390/50000 progress=26.8% loss=0.100930 lr=8.758200e-05 lr_backbone=8.758200e-05 elapsed=13:30:19 avg_step=3.63s recent_step=6.84s eta=36:55:32 +2026-09-15 03:17:44 | INFO | train rank=0 | step=13400/50000 progress=26.8% loss=0.104040 lr=8.756018e-05 lr_backbone=8.756018e-05 elapsed=13:31:16 avg_step=3.63s recent_step=5.72s eta=36:55:53 +2026-09-15 03:18:52 | INFO | train rank=0 | step=13410/50000 progress=26.8% loss=0.097107 lr=8.753834e-05 lr_backbone=8.753834e-05 elapsed=13:32:25 avg_step=3.63s recent_step=6.83s eta=36:56:44 +2026-09-15 03:19:51 | INFO | train rank=0 | step=13420/50000 progress=26.8% loss=0.142773 lr=8.751649e-05 lr_backbone=8.751649e-05 elapsed=13:33:24 avg_step=3.64s recent_step=5.90s eta=36:57:09 +2026-09-15 03:21:02 | INFO | train rank=0 | step=13430/50000 progress=26.9% loss=0.096426 lr=8.749462e-05 lr_backbone=8.749462e-05 elapsed=13:34:34 avg_step=3.64s recent_step=7.08s eta=36:58:06 +2026-09-15 03:22:09 | INFO | train rank=0 | step=13440/50000 progress=26.9% loss=0.082784 lr=8.747273e-05 lr_backbone=8.747273e-05 elapsed=13:35:41 avg_step=3.64s recent_step=6.70s eta=36:58:53 +2026-09-15 03:23:15 | INFO | train rank=0 | step=13450/50000 progress=26.9% loss=0.133846 lr=8.745083e-05 lr_backbone=8.745083e-05 elapsed=13:36:47 avg_step=3.64s recent_step=6.59s eta=36:59:37 +2026-09-15 03:24:24 | INFO | train rank=0 | step=13460/50000 progress=26.9% loss=0.112222 lr=8.742891e-05 lr_backbone=8.742891e-05 elapsed=13:37:57 avg_step=3.65s recent_step=6.98s eta=37:00:31 +2026-09-15 03:25:36 | INFO | train rank=0 | step=13470/50000 progress=26.9% loss=0.145435 lr=8.740698e-05 lr_backbone=8.740698e-05 elapsed=13:39:09 avg_step=3.65s recent_step=7.15s eta=37:01:29 +2026-09-15 03:26:41 | INFO | train rank=0 | step=13480/50000 progress=27.0% loss=0.105578 lr=8.738503e-05 lr_backbone=8.738503e-05 elapsed=13:40:14 avg_step=3.65s recent_step=6.57s eta=37:02:12 +2026-09-15 03:27:53 | INFO | train rank=0 | step=13490/50000 progress=27.0% loss=0.106029 lr=8.736306e-05 lr_backbone=8.736306e-05 elapsed=13:41:25 avg_step=3.65s recent_step=7.11s eta=37:03:09 +2026-09-15 03:29:02 | INFO | train rank=0 | step=13500/50000 progress=27.0% loss=0.139061 lr=8.734107e-05 lr_backbone=8.734107e-05 elapsed=13:42:35 avg_step=3.66s recent_step=6.91s eta=37:04:01 +2026-09-15 03:30:01 | INFO | train rank=0 | step=13510/50000 progress=27.0% loss=0.133708 lr=8.731907e-05 lr_backbone=8.731907e-05 elapsed=13:43:34 avg_step=3.66s recent_step=5.90s eta=37:04:25 +2026-09-15 03:31:02 | INFO | train rank=0 | step=13520/50000 progress=27.0% loss=0.089338 lr=8.729706e-05 lr_backbone=8.729706e-05 elapsed=13:44:35 avg_step=3.66s recent_step=6.16s eta=37:04:56 +2026-09-15 03:32:45 | INFO | train rank=0 | step=13530/50000 progress=27.1% loss=0.135291 lr=8.727502e-05 lr_backbone=8.727502e-05 elapsed=13:46:18 avg_step=3.66s recent_step=10.25s eta=37:07:17 +2026-09-15 03:34:35 | INFO | train rank=0 | step=13540/50000 progress=27.1% loss=0.138033 lr=8.725298e-05 lr_backbone=8.725298e-05 elapsed=13:48:08 avg_step=3.67s recent_step=11.03s eta=37:09:59 +2026-09-15 03:36:37 | INFO | train rank=0 | step=13550/50000 progress=27.1% loss=0.112817 lr=8.723091e-05 lr_backbone=8.723091e-05 elapsed=13:50:09 avg_step=3.68s recent_step=12.14s eta=37:13:10 +2026-09-15 03:38:22 | INFO | train rank=0 | step=13560/50000 progress=27.1% loss=0.137172 lr=8.720883e-05 lr_backbone=8.720883e-05 elapsed=13:51:55 avg_step=3.68s recent_step=10.57s eta=37:15:38 +2026-09-15 03:39:58 | INFO | train rank=0 | step=13570/50000 progress=27.1% loss=0.104407 lr=8.718673e-05 lr_backbone=8.718673e-05 elapsed=13:53:31 avg_step=3.69s recent_step=9.58s eta=37:17:40 +2026-09-15 03:41:35 | INFO | train rank=0 | step=13580/50000 progress=27.2% loss=0.077267 lr=8.716462e-05 lr_backbone=8.716462e-05 elapsed=13:55:08 avg_step=3.69s recent_step=9.69s eta=37:19:44 +2026-09-15 03:43:03 | INFO | train rank=0 | step=13590/50000 progress=27.2% loss=0.108070 lr=8.714249e-05 lr_backbone=8.714249e-05 elapsed=13:56:36 avg_step=3.69s recent_step=8.82s eta=37:21:25 +2026-09-15 03:44:30 | INFO | train rank=0 | step=13600/50000 progress=27.2% loss=0.108481 lr=8.712034e-05 lr_backbone=8.712034e-05 elapsed=13:58:03 avg_step=3.70s recent_step=8.68s eta=37:23:01 +2026-09-15 03:47:01 | INFO | train rank=0 | step=13610/50000 progress=27.2% loss=0.127851 lr=8.709818e-05 lr_backbone=8.709818e-05 elapsed=14:00:34 avg_step=3.71s recent_step=15.13s eta=37:27:30 +2026-09-15 03:47:30 | INFO | train rank=0 | step=13620/50000 progress=27.2% loss=0.081011 lr=8.707600e-05 lr_backbone=8.707600e-05 elapsed=14:01:03 avg_step=3.71s recent_step=2.86s eta=37:26:30 +2026-09-15 03:47:59 | INFO | train rank=0 | step=13630/50000 progress=27.3% loss=0.096253 lr=8.705380e-05 lr_backbone=8.705380e-05 elapsed=14:01:32 avg_step=3.70s recent_step=2.92s eta=37:25:32 +2026-09-15 03:48:36 | INFO | train rank=0 | step=13640/50000 progress=27.3% loss=0.091675 lr=8.703159e-05 lr_backbone=8.703159e-05 elapsed=14:02:09 avg_step=3.70s recent_step=3.74s eta=37:24:56 +2026-09-15 03:49:03 | INFO | train rank=0 | step=13650/50000 progress=27.3% loss=0.104218 lr=8.700936e-05 lr_backbone=8.700936e-05 elapsed=14:02:36 avg_step=3.70s recent_step=2.66s eta=37:23:51 +2026-09-15 03:49:30 | INFO | train rank=0 | step=13660/50000 progress=27.3% loss=0.112035 lr=8.698712e-05 lr_backbone=8.698712e-05 elapsed=14:03:03 avg_step=3.70s recent_step=2.75s eta=37:22:49 +2026-09-15 03:49:58 | INFO | train rank=0 | step=13670/50000 progress=27.3% loss=0.122154 lr=8.696486e-05 lr_backbone=8.696486e-05 elapsed=14:03:31 avg_step=3.70s recent_step=2.80s eta=37:21:48 +2026-09-15 03:50:36 | INFO | train rank=0 | step=13680/50000 progress=27.4% loss=0.105601 lr=8.694258e-05 lr_backbone=8.694258e-05 elapsed=14:04:09 avg_step=3.70s recent_step=3.79s eta=37:21:13 +2026-09-15 03:51:01 | INFO | train rank=0 | step=13690/50000 progress=27.4% loss=0.111657 lr=8.692029e-05 lr_backbone=8.692029e-05 elapsed=14:04:34 avg_step=3.70s recent_step=2.51s eta=37:20:04 +2026-09-15 03:51:30 | INFO | train rank=0 | step=13700/50000 progress=27.4% loss=0.108215 lr=8.689798e-05 lr_backbone=8.689798e-05 elapsed=14:05:03 avg_step=3.70s recent_step=2.89s eta=37:19:06 +2026-09-15 03:51:57 | INFO | train rank=0 | step=13710/50000 progress=27.4% loss=0.108154 lr=8.687566e-05 lr_backbone=8.687566e-05 elapsed=14:05:30 avg_step=3.70s recent_step=2.71s eta=37:18:02 +2026-09-15 03:52:29 | INFO | train rank=0 | step=13720/50000 progress=27.4% loss=0.094486 lr=8.685331e-05 lr_backbone=8.685331e-05 elapsed=14:06:02 avg_step=3.70s recent_step=3.14s eta=37:17:11 +2026-09-15 03:52:55 | INFO | train rank=0 | step=13730/50000 progress=27.5% loss=0.095051 lr=8.683096e-05 lr_backbone=8.683096e-05 elapsed=14:06:28 avg_step=3.70s recent_step=2.63s eta=37:16:05 +2026-09-15 03:53:22 | INFO | train rank=0 | step=13740/50000 progress=27.5% loss=0.123425 lr=8.680858e-05 lr_backbone=8.680858e-05 elapsed=14:06:55 avg_step=3.70s recent_step=2.72s eta=37:15:03 +2026-09-15 03:53:48 | INFO | train rank=0 | step=13750/50000 progress=27.5% loss=0.106237 lr=8.678620e-05 lr_backbone=8.678620e-05 elapsed=14:07:21 avg_step=3.70s recent_step=2.55s eta=37:13:55 +2026-09-15 03:54:25 | INFO | train rank=0 | step=13760/50000 progress=27.5% loss=0.100947 lr=8.676379e-05 lr_backbone=8.676379e-05 elapsed=14:07:58 avg_step=3.70s recent_step=3.74s eta=37:13:19 +2026-09-15 03:54:53 | INFO | train rank=0 | step=13770/50000 progress=27.5% loss=0.090775 lr=8.674137e-05 lr_backbone=8.674137e-05 elapsed=14:08:26 avg_step=3.70s recent_step=2.76s eta=37:12:18 +2026-09-15 03:55:21 | INFO | train rank=0 | step=13780/50000 progress=27.6% loss=0.101718 lr=8.671893e-05 lr_backbone=8.671893e-05 elapsed=14:08:54 avg_step=3.70s recent_step=2.83s eta=37:11:18 +2026-09-15 03:55:48 | INFO | train rank=0 | step=13790/50000 progress=27.6% loss=0.106754 lr=8.669648e-05 lr_backbone=8.669648e-05 elapsed=14:09:21 avg_step=3.70s recent_step=2.70s eta=37:10:15 +2026-09-15 03:56:16 | INFO | train rank=0 | step=13800/50000 progress=27.6% loss=0.069108 lr=8.667401e-05 lr_backbone=8.667401e-05 elapsed=14:09:49 avg_step=3.69s recent_step=2.82s eta=37:09:15 +2026-09-15 03:56:42 | INFO | train rank=0 | step=13810/50000 progress=27.6% loss=0.088001 lr=8.665152e-05 lr_backbone=8.665152e-05 elapsed=14:10:15 avg_step=3.69s recent_step=2.56s eta=37:08:08 +2026-09-15 03:57:11 | INFO | train rank=0 | step=13820/50000 progress=27.6% loss=0.121755 lr=8.662902e-05 lr_backbone=8.662902e-05 elapsed=14:10:44 avg_step=3.69s recent_step=2.88s eta=37:07:10 +2026-09-15 03:57:37 | INFO | train rank=0 | step=13830/50000 progress=27.7% loss=0.113212 lr=8.660650e-05 lr_backbone=8.660650e-05 elapsed=14:11:09 avg_step=3.69s recent_step=2.58s eta=37:06:04 +2026-09-15 03:58:03 | INFO | train rank=0 | step=13840/50000 progress=27.7% loss=0.125594 lr=8.658397e-05 lr_backbone=8.658397e-05 elapsed=14:11:36 avg_step=3.69s recent_step=2.69s eta=37:05:01 +2026-09-15 03:58:30 | INFO | train rank=0 | step=13850/50000 progress=27.7% loss=0.122587 lr=8.656142e-05 lr_backbone=8.656142e-05 elapsed=14:12:03 avg_step=3.69s recent_step=2.64s eta=37:03:57 +2026-09-15 03:59:01 | INFO | train rank=0 | step=13860/50000 progress=27.7% loss=0.127946 lr=8.653885e-05 lr_backbone=8.653885e-05 elapsed=14:12:34 avg_step=3.69s recent_step=3.13s eta=37:03:05 +2026-09-15 03:59:27 | INFO | train rank=0 | step=13870/50000 progress=27.7% loss=0.145554 lr=8.651627e-05 lr_backbone=8.651627e-05 elapsed=14:13:00 avg_step=3.69s recent_step=2.59s eta=37:01:59 +2026-09-15 04:00:02 | INFO | train rank=0 | step=13880/50000 progress=27.8% loss=0.109204 lr=8.649367e-05 lr_backbone=8.649367e-05 elapsed=14:13:34 avg_step=3.69s recent_step=3.45s eta=37:01:16 +2026-09-15 04:00:22 | INFO | train rank=0 | step=13890/50000 progress=27.8% loss=0.117143 lr=8.647106e-05 lr_backbone=8.647106e-05 elapsed=14:13:55 avg_step=3.69s recent_step=2.03s eta=36:59:56 +2026-09-15 04:00:59 | INFO | train rank=0 | step=13900/50000 progress=27.8% loss=0.084570 lr=8.644843e-05 lr_backbone=8.644843e-05 elapsed=14:14:32 avg_step=3.69s recent_step=3.71s eta=36:59:20 +2026-09-15 04:01:31 | INFO | train rank=0 | step=13910/50000 progress=27.8% loss=0.128776 lr=8.642579e-05 lr_backbone=8.642579e-05 elapsed=14:15:04 avg_step=3.69s recent_step=3.19s eta=36:58:30 +2026-09-15 04:01:57 | INFO | train rank=0 | step=13920/50000 progress=27.8% loss=0.142876 lr=8.640312e-05 lr_backbone=8.640312e-05 elapsed=14:15:30 avg_step=3.69s recent_step=2.66s eta=36:57:26 +2026-09-15 04:02:31 | INFO | train rank=0 | step=13930/50000 progress=27.9% loss=0.129561 lr=8.638045e-05 lr_backbone=8.638045e-05 elapsed=14:16:03 avg_step=3.69s recent_step=3.31s eta=36:56:40 +2026-09-15 04:03:01 | INFO | train rank=0 | step=13940/50000 progress=27.9% loss=0.161480 lr=8.635775e-05 lr_backbone=8.635775e-05 elapsed=14:16:34 avg_step=3.69s recent_step=3.05s eta=36:55:46 +2026-09-15 04:03:25 | INFO | train rank=0 | step=13950/50000 progress=27.9% loss=0.128849 lr=8.633504e-05 lr_backbone=8.633504e-05 elapsed=14:16:57 avg_step=3.69s recent_step=2.35s eta=36:54:35 +2026-09-15 04:03:49 | INFO | train rank=0 | step=13960/50000 progress=27.9% loss=0.120859 lr=8.631232e-05 lr_backbone=8.631232e-05 elapsed=14:17:22 avg_step=3.68s recent_step=2.45s eta=36:53:26 +2026-09-15 04:04:16 | INFO | train rank=0 | step=13970/50000 progress=27.9% loss=0.101633 lr=8.628958e-05 lr_backbone=8.628958e-05 elapsed=14:17:49 avg_step=3.68s recent_step=2.72s eta=36:52:24 +2026-09-15 04:04:50 | INFO | train rank=0 | step=13980/50000 progress=28.0% loss=0.105639 lr=8.626682e-05 lr_backbone=8.626682e-05 elapsed=14:18:23 avg_step=3.68s recent_step=3.37s eta=36:51:39 +2026-09-15 04:05:17 | INFO | train rank=0 | step=13990/50000 progress=28.0% loss=0.115751 lr=8.624405e-05 lr_backbone=8.624405e-05 elapsed=14:18:49 avg_step=3.68s recent_step=2.66s eta=36:50:36 +2026-09-15 04:05:43 | INFO | train rank=0 | step=14000/50000 progress=28.0% loss=0.090338 lr=8.622126e-05 lr_backbone=8.622126e-05 elapsed=14:19:16 avg_step=3.68s recent_step=2.67s eta=36:49:33 +2026-09-15 04:06:10 | INFO | train rank=0 | step=14010/50000 progress=28.0% loss=0.148538 lr=8.619846e-05 lr_backbone=8.619846e-05 elapsed=14:19:42 avg_step=3.68s recent_step=2.63s eta=36:48:30 +2026-09-15 04:06:37 | INFO | train rank=0 | step=14020/50000 progress=28.0% loss=0.124127 lr=8.617564e-05 lr_backbone=8.617564e-05 elapsed=14:20:10 avg_step=3.68s recent_step=2.73s eta=36:47:28 +2026-09-15 04:07:06 | INFO | train rank=0 | step=14030/50000 progress=28.1% loss=0.085107 lr=8.615280e-05 lr_backbone=8.615280e-05 elapsed=14:20:39 avg_step=3.68s recent_step=2.90s eta=36:46:32 +2026-09-15 04:07:45 | INFO | train rank=0 | step=14040/50000 progress=28.1% loss=0.097497 lr=8.612995e-05 lr_backbone=8.612995e-05 elapsed=14:21:18 avg_step=3.68s recent_step=3.90s eta=36:46:01 +2026-09-15 04:08:16 | INFO | train rank=0 | step=14050/50000 progress=28.1% loss=0.125429 lr=8.610708e-05 lr_backbone=8.610708e-05 elapsed=14:21:49 avg_step=3.68s recent_step=3.10s eta=36:45:09 +2026-09-15 04:08:41 | INFO | train rank=0 | step=14060/50000 progress=28.1% loss=0.096160 lr=8.608420e-05 lr_backbone=8.608420e-05 elapsed=14:22:13 avg_step=3.68s recent_step=2.47s eta=36:44:01 +2026-09-15 04:09:02 | INFO | train rank=0 | step=14070/50000 progress=28.1% loss=0.147417 lr=8.606130e-05 lr_backbone=8.606130e-05 elapsed=14:22:35 avg_step=3.68s recent_step=2.19s eta=36:42:46 +2026-09-15 04:09:48 | INFO | train rank=0 | step=14080/50000 progress=28.2% loss=0.100590 lr=8.603838e-05 lr_backbone=8.603838e-05 elapsed=14:23:21 avg_step=3.68s recent_step=4.53s eta=36:42:31 +2026-09-15 04:10:09 | INFO | train rank=0 | step=14090/50000 progress=28.2% loss=0.096258 lr=8.601545e-05 lr_backbone=8.601545e-05 elapsed=14:23:42 avg_step=3.68s recent_step=2.12s eta=36:41:14 +2026-09-15 04:10:37 | INFO | train rank=0 | step=14100/50000 progress=28.2% loss=0.086188 lr=8.599250e-05 lr_backbone=8.599250e-05 elapsed=14:24:10 avg_step=3.68s recent_step=2.80s eta=36:40:15 +2026-09-15 04:11:02 | INFO | train rank=0 | step=14110/50000 progress=28.2% loss=0.096170 lr=8.596954e-05 lr_backbone=8.596954e-05 elapsed=14:24:34 avg_step=3.68s recent_step=2.47s eta=36:39:08 +2026-09-15 04:11:33 | INFO | train rank=0 | step=14120/50000 progress=28.2% loss=0.125909 lr=8.594656e-05 lr_backbone=8.594656e-05 elapsed=14:25:06 avg_step=3.68s recent_step=3.11s eta=36:38:17 +2026-09-15 04:12:01 | INFO | train rank=0 | step=14130/50000 progress=28.3% loss=0.100067 lr=8.592357e-05 lr_backbone=8.592357e-05 elapsed=14:25:34 avg_step=3.68s recent_step=2.84s eta=36:37:19 +2026-09-15 04:12:28 | INFO | train rank=0 | step=14140/50000 progress=28.3% loss=0.107068 lr=8.590056e-05 lr_backbone=8.590056e-05 elapsed=14:26:01 avg_step=3.67s recent_step=2.72s eta=36:36:18 +2026-09-15 04:13:00 | INFO | train rank=0 | step=14150/50000 progress=28.3% loss=0.115205 lr=8.587754e-05 lr_backbone=8.587754e-05 elapsed=14:26:32 avg_step=3.67s recent_step=3.12s eta=36:35:27 +2026-09-15 04:13:27 | INFO | train rank=0 | step=14160/50000 progress=28.3% loss=0.075388 lr=8.585449e-05 lr_backbone=8.585449e-05 elapsed=14:27:00 avg_step=3.67s recent_step=2.78s eta=36:34:28 +2026-09-15 04:13:53 | INFO | train rank=0 | step=14170/50000 progress=28.3% loss=0.085136 lr=8.583144e-05 lr_backbone=8.583144e-05 elapsed=14:27:26 avg_step=3.67s recent_step=2.58s eta=36:33:23 +2026-09-15 04:14:20 | INFO | train rank=0 | step=14180/50000 progress=28.4% loss=0.128548 lr=8.580837e-05 lr_backbone=8.580837e-05 elapsed=14:27:52 avg_step=3.67s recent_step=2.63s eta=36:32:20 +2026-09-15 04:14:43 | INFO | train rank=0 | step=14190/50000 progress=28.4% loss=0.084394 lr=8.578528e-05 lr_backbone=8.578528e-05 elapsed=14:28:16 avg_step=3.67s recent_step=2.39s eta=36:31:11 +2026-09-15 04:15:16 | INFO | train rank=0 | step=14200/50000 progress=28.4% loss=0.146908 lr=8.576217e-05 lr_backbone=8.576217e-05 elapsed=14:28:49 avg_step=3.67s recent_step=3.28s eta=36:30:25 +2026-09-15 04:15:46 | INFO | train rank=0 | step=14210/50000 progress=28.4% loss=0.115005 lr=8.573906e-05 lr_backbone=8.573906e-05 elapsed=14:29:18 avg_step=3.67s recent_step=2.94s eta=36:29:30 +2026-09-15 04:16:18 | INFO | train rank=0 | step=14220/50000 progress=28.4% loss=0.094132 lr=8.571592e-05 lr_backbone=8.571592e-05 elapsed=14:29:50 avg_step=3.67s recent_step=3.19s eta=36:28:41 +2026-09-15 04:16:39 | INFO | train rank=0 | step=14230/50000 progress=28.5% loss=0.096830 lr=8.569277e-05 lr_backbone=8.569277e-05 elapsed=14:30:12 avg_step=3.67s recent_step=2.14s eta=36:27:26 +2026-09-15 04:17:10 | INFO | train rank=0 | step=14240/50000 progress=28.5% loss=0.104729 lr=8.566960e-05 lr_backbone=8.566960e-05 elapsed=14:30:43 avg_step=3.67s recent_step=3.14s eta=36:26:36 +2026-09-15 04:17:30 | INFO | train rank=0 | step=14250/50000 progress=28.5% loss=0.103575 lr=8.564642e-05 lr_backbone=8.564642e-05 elapsed=14:31:03 avg_step=3.67s recent_step=2.01s eta=36:25:17 +2026-09-15 04:17:58 | INFO | train rank=0 | step=14260/50000 progress=28.5% loss=0.089318 lr=8.562323e-05 lr_backbone=8.562323e-05 elapsed=14:31:31 avg_step=3.67s recent_step=2.73s eta=36:24:17 +2026-09-15 04:18:25 | INFO | train rank=0 | step=14270/50000 progress=28.5% loss=0.104910 lr=8.560001e-05 lr_backbone=8.560001e-05 elapsed=14:31:58 avg_step=3.67s recent_step=2.69s eta=36:23:16 +2026-09-15 04:19:03 | INFO | train rank=0 | step=14280/50000 progress=28.6% loss=0.135653 lr=8.557678e-05 lr_backbone=8.557678e-05 elapsed=14:32:36 avg_step=3.67s recent_step=3.87s eta=36:22:45 +2026-09-15 04:19:29 | INFO | train rank=0 | step=14290/50000 progress=28.6% loss=0.142601 lr=8.555354e-05 lr_backbone=8.555354e-05 elapsed=14:33:02 avg_step=3.67s recent_step=2.61s eta=36:21:41 +2026-09-15 04:20:00 | INFO | train rank=0 | step=14300/50000 progress=28.6% loss=0.105484 lr=8.553028e-05 lr_backbone=8.553028e-05 elapsed=14:33:33 avg_step=3.67s recent_step=3.04s eta=36:20:49 +2026-09-15 04:20:26 | INFO | train rank=0 | step=14310/50000 progress=28.6% loss=0.095753 lr=8.550701e-05 lr_backbone=8.550701e-05 elapsed=14:33:59 avg_step=3.66s recent_step=2.65s eta=36:19:47 +2026-09-15 04:20:55 | INFO | train rank=0 | step=14320/50000 progress=28.6% loss=0.119433 lr=8.548371e-05 lr_backbone=8.548371e-05 elapsed=14:34:28 avg_step=3.66s recent_step=2.87s eta=36:18:51 +2026-09-15 04:21:20 | INFO | train rank=0 | step=14330/50000 progress=28.7% loss=0.107790 lr=8.546041e-05 lr_backbone=8.546041e-05 elapsed=14:34:53 avg_step=3.66s recent_step=2.53s eta=36:17:46 +2026-09-15 04:21:46 | INFO | train rank=0 | step=14340/50000 progress=28.7% loss=0.115935 lr=8.543709e-05 lr_backbone=8.543709e-05 elapsed=14:35:18 avg_step=3.66s recent_step=2.53s eta=36:16:41 +2026-09-15 04:22:18 | INFO | train rank=0 | step=14350/50000 progress=28.7% loss=0.108670 lr=8.541375e-05 lr_backbone=8.541375e-05 elapsed=14:35:50 avg_step=3.66s recent_step=3.21s eta=36:15:53 +2026-09-15 04:22:44 | INFO | train rank=0 | step=14360/50000 progress=28.7% loss=0.100706 lr=8.539040e-05 lr_backbone=8.539040e-05 elapsed=14:36:17 avg_step=3.66s recent_step=2.65s eta=36:14:51 +2026-09-15 04:23:13 | INFO | train rank=0 | step=14370/50000 progress=28.7% loss=0.127960 lr=8.536703e-05 lr_backbone=8.536703e-05 elapsed=14:36:46 avg_step=3.66s recent_step=2.89s eta=36:13:56 +2026-09-15 04:23:38 | INFO | train rank=0 | step=14380/50000 progress=28.8% loss=0.118164 lr=8.534365e-05 lr_backbone=8.534365e-05 elapsed=14:37:11 avg_step=3.66s recent_step=2.48s eta=36:12:50 +2026-09-15 04:24:05 | INFO | train rank=0 | step=14390/50000 progress=28.8% loss=0.132872 lr=8.532025e-05 lr_backbone=8.532025e-05 elapsed=14:37:38 avg_step=3.66s recent_step=2.76s eta=36:11:51 +2026-09-15 04:24:26 | INFO | train rank=0 | step=14400/50000 progress=28.8% loss=0.124942 lr=8.529683e-05 lr_backbone=8.529683e-05 elapsed=14:37:59 avg_step=3.66s recent_step=2.04s eta=36:10:34 +2026-09-15 04:24:46 | INFO | train rank=0 | step=14410/50000 progress=28.8% loss=0.137258 lr=8.527340e-05 lr_backbone=8.527340e-05 elapsed=14:38:19 avg_step=3.66s recent_step=1.99s eta=36:09:16 +2026-09-15 04:25:16 | INFO | train rank=0 | step=14420/50000 progress=28.8% loss=0.097288 lr=8.524996e-05 lr_backbone=8.524996e-05 elapsed=14:38:49 avg_step=3.66s recent_step=3.05s eta=36:08:25 +2026-09-15 04:25:38 | INFO | train rank=0 | step=14430/50000 progress=28.9% loss=0.127866 lr=8.522650e-05 lr_backbone=8.522650e-05 elapsed=14:39:11 avg_step=3.66s recent_step=2.21s eta=36:07:12 +2026-09-15 04:26:04 | INFO | train rank=0 | step=14440/50000 progress=28.9% loss=0.114882 lr=8.520302e-05 lr_backbone=8.520302e-05 elapsed=14:39:37 avg_step=3.65s recent_step=2.55s eta=36:06:09 +2026-09-15 04:26:27 | INFO | train rank=0 | step=14450/50000 progress=28.9% loss=0.110249 lr=8.517953e-05 lr_backbone=8.517953e-05 elapsed=14:40:00 avg_step=3.65s recent_step=2.35s eta=36:05:00 +2026-09-15 04:27:02 | INFO | train rank=0 | step=14460/50000 progress=28.9% loss=0.143673 lr=8.515602e-05 lr_backbone=8.515602e-05 elapsed=14:40:34 avg_step=3.65s recent_step=3.44s eta=36:04:18 +2026-09-15 04:27:26 | INFO | train rank=0 | step=14470/50000 progress=28.9% loss=0.077231 lr=8.513250e-05 lr_backbone=8.513250e-05 elapsed=14:40:59 avg_step=3.65s recent_step=2.46s eta=36:03:12 +2026-09-15 04:27:53 | INFO | train rank=0 | step=14480/50000 progress=29.0% loss=0.089619 lr=8.510896e-05 lr_backbone=8.510896e-05 elapsed=14:41:26 avg_step=3.65s recent_step=2.69s eta=36:02:12 +2026-09-15 04:28:20 | INFO | train rank=0 | step=14490/50000 progress=29.0% loss=0.136429 lr=8.508541e-05 lr_backbone=8.508541e-05 elapsed=14:41:53 avg_step=3.65s recent_step=2.69s eta=36:01:12 +2026-09-15 04:28:46 | INFO | train rank=0 | step=14500/50000 progress=29.0% loss=0.100691 lr=8.506184e-05 lr_backbone=8.506184e-05 elapsed=14:42:19 avg_step=3.65s recent_step=2.60s eta=36:00:10 +2026-09-15 04:29:29 | INFO | train rank=0 | step=14510/50000 progress=29.0% loss=0.153623 lr=8.503826e-05 lr_backbone=8.503826e-05 elapsed=14:43:01 avg_step=3.65s recent_step=4.24s eta=35:59:48 +2026-09-15 04:29:52 | INFO | train rank=0 | step=14520/50000 progress=29.0% loss=0.094148 lr=8.501466e-05 lr_backbone=8.501466e-05 elapsed=14:43:25 avg_step=3.65s recent_step=2.35s eta=35:58:40 +2026-09-15 04:30:22 | INFO | train rank=0 | step=14530/50000 progress=29.1% loss=0.077104 lr=8.499104e-05 lr_backbone=8.499104e-05 elapsed=14:43:55 avg_step=3.65s recent_step=2.99s eta=35:57:47 +2026-09-15 04:30:51 | INFO | train rank=0 | step=14540/50000 progress=29.1% loss=0.146733 lr=8.496741e-05 lr_backbone=8.496741e-05 elapsed=14:44:23 avg_step=3.65s recent_step=2.86s eta=35:56:51 +2026-09-15 04:31:10 | INFO | train rank=0 | step=14550/50000 progress=29.1% loss=0.098002 lr=8.494377e-05 lr_backbone=8.494377e-05 elapsed=14:44:43 avg_step=3.65s recent_step=1.94s eta=35:55:33 +2026-09-15 04:31:36 | INFO | train rank=0 | step=14560/50000 progress=29.1% loss=0.110141 lr=8.492011e-05 lr_backbone=8.492011e-05 elapsed=14:45:09 avg_step=3.65s recent_step=2.64s eta=35:54:32 +2026-09-15 04:32:04 | INFO | train rank=0 | step=14570/50000 progress=29.1% loss=0.074796 lr=8.489643e-05 lr_backbone=8.489643e-05 elapsed=14:45:37 avg_step=3.65s recent_step=2.80s eta=35:53:35 +2026-09-15 04:32:24 | INFO | train rank=0 | step=14580/50000 progress=29.2% loss=0.115320 lr=8.487274e-05 lr_backbone=8.487274e-05 elapsed=14:45:57 avg_step=3.65s recent_step=1.96s eta=35:52:17 +2026-09-15 04:32:56 | INFO | train rank=0 | step=14590/50000 progress=29.2% loss=0.103432 lr=8.484903e-05 lr_backbone=8.484903e-05 elapsed=14:46:29 avg_step=3.65s recent_step=3.24s eta=35:51:31 +2026-09-15 04:33:24 | INFO | train rank=0 | step=14600/50000 progress=29.2% loss=0.108563 lr=8.482531e-05 lr_backbone=8.482531e-05 elapsed=14:46:57 avg_step=3.65s recent_step=2.79s eta=35:50:34 +2026-09-15 04:33:51 | INFO | train rank=0 | step=14610/50000 progress=29.2% loss=0.117530 lr=8.480158e-05 lr_backbone=8.480158e-05 elapsed=14:47:24 avg_step=3.64s recent_step=2.70s eta=35:49:34 +2026-09-15 04:34:14 | INFO | train rank=0 | step=14620/50000 progress=29.2% loss=0.097156 lr=8.477782e-05 lr_backbone=8.477782e-05 elapsed=14:47:47 avg_step=3.64s recent_step=2.28s eta=35:48:25 +2026-09-15 04:34:48 | INFO | train rank=0 | step=14630/50000 progress=29.3% loss=0.102163 lr=8.475406e-05 lr_backbone=8.475406e-05 elapsed=14:48:21 avg_step=3.64s recent_step=3.44s eta=35:47:44 +2026-09-15 04:35:12 | INFO | train rank=0 | step=14640/50000 progress=29.3% loss=0.108761 lr=8.473028e-05 lr_backbone=8.473028e-05 elapsed=14:48:45 avg_step=3.64s recent_step=2.35s eta=35:46:36 +2026-09-15 04:35:34 | INFO | train rank=0 | step=14650/50000 progress=29.3% loss=0.093556 lr=8.470648e-05 lr_backbone=8.470648e-05 elapsed=14:49:06 avg_step=3.64s recent_step=2.18s eta=35:45:24 +2026-09-15 04:36:00 | INFO | train rank=0 | step=14660/50000 progress=29.3% loss=0.104182 lr=8.468267e-05 lr_backbone=8.468267e-05 elapsed=14:49:33 avg_step=3.64s recent_step=2.67s eta=35:44:24 +2026-09-15 04:36:28 | INFO | train rank=0 | step=14670/50000 progress=29.3% loss=0.112798 lr=8.465884e-05 lr_backbone=8.465884e-05 elapsed=14:50:01 avg_step=3.64s recent_step=2.75s eta=35:43:26 +2026-09-15 04:36:57 | INFO | train rank=0 | step=14680/50000 progress=29.4% loss=0.090773 lr=8.463499e-05 lr_backbone=8.463499e-05 elapsed=14:50:30 avg_step=3.64s recent_step=2.95s eta=35:42:33 +2026-09-15 04:37:19 | INFO | train rank=0 | step=14690/50000 progress=29.4% loss=0.142505 lr=8.461114e-05 lr_backbone=8.461114e-05 elapsed=14:50:52 avg_step=3.64s recent_step=2.15s eta=35:41:21 +2026-09-15 04:37:55 | INFO | train rank=0 | step=14700/50000 progress=29.4% loss=0.119266 lr=8.458726e-05 lr_backbone=8.458726e-05 elapsed=14:51:28 avg_step=3.64s recent_step=3.67s eta=35:40:45 +2026-09-15 04:38:26 | INFO | train rank=0 | step=14710/50000 progress=29.4% loss=0.099160 lr=8.456337e-05 lr_backbone=8.456337e-05 elapsed=14:51:59 avg_step=3.64s recent_step=3.04s eta=35:39:55 +2026-09-15 04:39:09 | INFO | train rank=0 | step=14720/50000 progress=29.4% loss=0.093315 lr=8.453947e-05 lr_backbone=8.453947e-05 elapsed=14:52:42 avg_step=3.64s recent_step=4.33s eta=35:39:35 +2026-09-15 04:39:40 | INFO | train rank=0 | step=14730/50000 progress=29.5% loss=0.098078 lr=8.451555e-05 lr_backbone=8.451555e-05 elapsed=14:53:13 avg_step=3.64s recent_step=3.06s eta=35:38:45 +2026-09-15 04:40:13 | INFO | train rank=0 | step=14740/50000 progress=29.5% loss=0.104293 lr=8.449162e-05 lr_backbone=8.449162e-05 elapsed=14:53:46 avg_step=3.64s recent_step=3.33s eta=35:38:01 +2026-09-15 04:41:00 | INFO | train rank=0 | step=14750/50000 progress=29.5% loss=0.098552 lr=8.446767e-05 lr_backbone=8.446767e-05 elapsed=14:54:33 avg_step=3.64s recent_step=4.71s eta=35:37:50 +2026-09-15 04:41:34 | INFO | train rank=0 | step=14760/50000 progress=29.5% loss=0.143008 lr=8.444371e-05 lr_backbone=8.444371e-05 elapsed=14:55:06 avg_step=3.64s recent_step=3.34s eta=35:37:07 +2026-09-15 04:42:12 | INFO | train rank=0 | step=14770/50000 progress=29.5% loss=0.137875 lr=8.441973e-05 lr_backbone=8.441973e-05 elapsed=14:55:45 avg_step=3.64s recent_step=3.85s eta=35:36:35 +2026-09-15 04:42:46 | INFO | train rank=0 | step=14780/50000 progress=29.6% loss=0.114111 lr=8.439573e-05 lr_backbone=8.439573e-05 elapsed=14:56:19 avg_step=3.64s recent_step=3.43s eta=35:35:54 +2026-09-15 04:43:40 | INFO | train rank=0 | step=14790/50000 progress=29.6% loss=0.123766 lr=8.437172e-05 lr_backbone=8.437172e-05 elapsed=14:57:13 avg_step=3.64s recent_step=5.36s eta=35:35:59 +2026-09-15 04:44:15 | INFO | train rank=0 | step=14800/50000 progress=29.6% loss=0.109323 lr=8.434770e-05 lr_backbone=8.434770e-05 elapsed=14:57:48 avg_step=3.64s recent_step=3.48s eta=35:35:18 +2026-09-15 04:44:49 | INFO | train rank=0 | step=14810/50000 progress=29.6% loss=0.118217 lr=8.432366e-05 lr_backbone=8.432366e-05 elapsed=14:58:21 avg_step=3.64s recent_step=3.38s eta=35:34:36 +2026-09-15 04:45:25 | INFO | train rank=0 | step=14820/50000 progress=29.6% loss=0.079266 lr=8.429961e-05 lr_backbone=8.429961e-05 elapsed=14:58:58 avg_step=3.64s recent_step=3.65s eta=35:34:00 +2026-09-15 04:46:16 | INFO | train rank=0 | step=14830/50000 progress=29.7% loss=0.098621 lr=8.427554e-05 lr_backbone=8.427554e-05 elapsed=14:59:48 avg_step=3.64s recent_step=5.03s eta=35:33:56 +2026-09-15 04:46:52 | INFO | train rank=0 | step=14840/50000 progress=29.7% loss=0.161168 lr=8.425145e-05 lr_backbone=8.425145e-05 elapsed=15:00:25 avg_step=3.64s recent_step=3.66s eta=35:33:20 +2026-09-15 04:47:30 | INFO | train rank=0 | step=14850/50000 progress=29.7% loss=0.093896 lr=8.422736e-05 lr_backbone=8.422736e-05 elapsed=15:01:03 avg_step=3.64s recent_step=3.77s eta=35:32:47 +2026-09-15 04:48:01 | INFO | train rank=0 | step=14860/50000 progress=29.7% loss=0.095386 lr=8.420324e-05 lr_backbone=8.420324e-05 elapsed=15:01:34 avg_step=3.64s recent_step=3.14s eta=35:31:59 +2026-09-15 04:48:49 | INFO | train rank=0 | step=14870/50000 progress=29.7% loss=0.076670 lr=8.417911e-05 lr_backbone=8.417911e-05 elapsed=15:02:22 avg_step=3.64s recent_step=4.80s eta=35:31:50 +2026-09-15 04:49:13 | INFO | train rank=0 | step=14880/50000 progress=29.8% loss=0.109012 lr=8.415497e-05 lr_backbone=8.415497e-05 elapsed=15:02:45 avg_step=3.64s recent_step=2.34s eta=35:30:43 +2026-09-15 04:49:39 | INFO | train rank=0 | step=14890/50000 progress=29.8% loss=0.152862 lr=8.413081e-05 lr_backbone=8.413081e-05 elapsed=15:03:12 avg_step=3.64s recent_step=2.61s eta=35:29:42 +2026-09-15 04:50:17 | INFO | train rank=0 | step=14900/50000 progress=29.8% loss=0.101707 lr=8.410664e-05 lr_backbone=8.410664e-05 elapsed=15:03:50 avg_step=3.64s recent_step=3.84s eta=35:29:11 +2026-09-15 04:50:58 | INFO | train rank=0 | step=14910/50000 progress=29.8% loss=0.096022 lr=8.408245e-05 lr_backbone=8.408245e-05 elapsed=15:04:31 avg_step=3.64s recent_step=4.09s eta=35:28:45 +2026-09-15 04:51:35 | INFO | train rank=0 | step=14920/50000 progress=29.8% loss=0.097312 lr=8.405824e-05 lr_backbone=8.405824e-05 elapsed=15:05:08 avg_step=3.64s recent_step=3.68s eta=35:28:09 +2026-09-15 04:52:08 | INFO | train rank=0 | step=14930/50000 progress=29.9% loss=0.121340 lr=8.403402e-05 lr_backbone=8.403402e-05 elapsed=15:05:41 avg_step=3.64s recent_step=3.30s eta=35:27:25 +2026-09-15 04:52:39 | INFO | train rank=0 | step=14940/50000 progress=29.9% loss=0.106452 lr=8.400979e-05 lr_backbone=8.400979e-05 elapsed=15:06:12 avg_step=3.64s recent_step=3.14s eta=35:26:37 +2026-09-15 04:53:27 | INFO | train rank=0 | step=14950/50000 progress=29.9% loss=0.092404 lr=8.398554e-05 lr_backbone=8.398554e-05 elapsed=15:07:00 avg_step=3.64s recent_step=4.78s eta=35:26:27 +2026-09-15 04:54:06 | INFO | train rank=0 | step=14960/50000 progress=29.9% loss=0.120377 lr=8.396128e-05 lr_backbone=8.396128e-05 elapsed=15:07:39 avg_step=3.64s recent_step=3.87s eta=35:25:56 +2026-09-15 04:54:47 | INFO | train rank=0 | step=14970/50000 progress=29.9% loss=0.118859 lr=8.393700e-05 lr_backbone=8.393700e-05 elapsed=15:08:20 avg_step=3.64s recent_step=4.09s eta=35:25:30 +2026-09-15 04:55:24 | INFO | train rank=0 | step=14980/50000 progress=30.0% loss=0.092049 lr=8.391271e-05 lr_backbone=8.391271e-05 elapsed=15:08:57 avg_step=3.64s recent_step=3.76s eta=35:24:57 +2026-09-15 04:56:08 | INFO | train rank=0 | step=14990/50000 progress=30.0% loss=0.083321 lr=8.388840e-05 lr_backbone=8.388840e-05 elapsed=15:09:41 avg_step=3.64s recent_step=4.42s eta=35:24:38 +2026-09-15 04:56:42 | INFO | train rank=0 | step=15000/50000 progress=30.0% loss=0.108856 lr=8.386408e-05 lr_backbone=8.386408e-05 elapsed=15:10:15 avg_step=3.64s recent_step=3.39s eta=35:23:56 +2026-09-15 04:57:17 | INFO | train rank=0 | step=15010/50000 progress=30.0% loss=0.116950 lr=8.383974e-05 lr_backbone=8.383974e-05 elapsed=15:10:50 avg_step=3.64s recent_step=3.50s eta=35:23:16 +2026-09-15 04:57:44 | INFO | train rank=0 | step=15020/50000 progress=30.0% loss=0.109600 lr=8.381539e-05 lr_backbone=8.381539e-05 elapsed=15:11:17 avg_step=3.64s recent_step=2.63s eta=35:22:17 +2026-09-15 04:58:33 | INFO | train rank=0 | step=15030/50000 progress=30.1% loss=0.079887 lr=8.379102e-05 lr_backbone=8.379102e-05 elapsed=15:12:06 avg_step=3.64s recent_step=4.93s eta=35:22:10 +2026-09-15 04:59:02 | INFO | train rank=0 | step=15040/50000 progress=30.1% loss=0.138600 lr=8.376664e-05 lr_backbone=8.376664e-05 elapsed=15:12:35 avg_step=3.64s recent_step=2.91s eta=35:21:17 +2026-09-15 04:59:40 | INFO | train rank=0 | step=15050/50000 progress=30.1% loss=0.098911 lr=8.374224e-05 lr_backbone=8.374224e-05 elapsed=15:13:12 avg_step=3.64s recent_step=3.75s eta=35:20:43 +2026-09-15 05:00:08 | INFO | train rank=0 | step=15060/50000 progress=30.1% loss=0.075941 lr=8.371783e-05 lr_backbone=8.371783e-05 elapsed=15:13:41 avg_step=3.64s recent_step=2.86s eta=35:19:48 +2026-09-15 05:01:05 | INFO | train rank=0 | step=15070/50000 progress=30.1% loss=0.098289 lr=8.369341e-05 lr_backbone=8.369341e-05 elapsed=15:14:37 avg_step=3.64s recent_step=5.64s eta=35:19:58 +2026-09-15 05:01:42 | INFO | train rank=0 | step=15080/50000 progress=30.2% loss=0.166250 lr=8.366897e-05 lr_backbone=8.366897e-05 elapsed=15:15:14 avg_step=3.64s recent_step=3.71s eta=35:19:23 +2026-09-15 05:02:13 | INFO | train rank=0 | step=15090/50000 progress=30.2% loss=0.116867 lr=8.364451e-05 lr_backbone=8.364451e-05 elapsed=15:15:46 avg_step=3.64s recent_step=3.14s eta=35:18:35 +2026-09-15 05:02:50 | INFO | train rank=0 | step=15100/50000 progress=30.2% loss=0.090919 lr=8.362004e-05 lr_backbone=8.362004e-05 elapsed=15:16:22 avg_step=3.64s recent_step=3.65s eta=35:17:59 +2026-09-15 05:03:47 | INFO | train rank=0 | step=15110/50000 progress=30.2% loss=0.095158 lr=8.359556e-05 lr_backbone=8.359556e-05 elapsed=15:17:20 avg_step=3.64s recent_step=5.74s eta=35:18:11 +2026-09-15 05:04:17 | INFO | train rank=0 | step=15120/50000 progress=30.2% loss=0.072357 lr=8.357106e-05 lr_backbone=8.357106e-05 elapsed=15:17:50 avg_step=3.64s recent_step=3.01s eta=35:17:20 +2026-09-15 05:04:44 | INFO | train rank=0 | step=15130/50000 progress=30.3% loss=0.109364 lr=8.354654e-05 lr_backbone=8.354654e-05 elapsed=15:18:16 avg_step=3.64s recent_step=2.66s eta=35:16:21 +2026-09-15 05:05:22 | INFO | train rank=0 | step=15140/50000 progress=30.3% loss=0.130632 lr=8.352201e-05 lr_backbone=8.352201e-05 elapsed=15:18:55 avg_step=3.64s recent_step=3.83s eta=35:15:49 +2026-09-15 05:06:03 | INFO | train rank=0 | step=15150/50000 progress=30.3% loss=0.098257 lr=8.349747e-05 lr_backbone=8.349747e-05 elapsed=15:19:36 avg_step=3.64s recent_step=4.13s eta=35:15:23 +2026-09-15 05:06:36 | INFO | train rank=0 | step=15160/50000 progress=30.3% loss=0.124481 lr=8.347291e-05 lr_backbone=8.347291e-05 elapsed=15:20:09 avg_step=3.64s recent_step=3.30s eta=35:14:39 +2026-09-15 05:07:05 | INFO | train rank=0 | step=15170/50000 progress=30.3% loss=0.129116 lr=8.344834e-05 lr_backbone=8.344834e-05 elapsed=15:20:38 avg_step=3.64s recent_step=2.92s eta=35:13:46 +2026-09-15 05:07:39 | INFO | train rank=0 | step=15180/50000 progress=30.4% loss=0.118460 lr=8.342375e-05 lr_backbone=8.342375e-05 elapsed=15:21:12 avg_step=3.64s recent_step=3.41s eta=35:13:04 +2026-09-15 05:08:30 | INFO | train rank=0 | step=15190/50000 progress=30.4% loss=0.137124 lr=8.339915e-05 lr_backbone=8.339915e-05 elapsed=15:22:03 avg_step=3.64s recent_step=5.11s eta=35:13:02 +2026-09-15 05:09:00 | INFO | train rank=0 | step=15200/50000 progress=30.4% loss=0.083829 lr=8.337453e-05 lr_backbone=8.337453e-05 elapsed=15:22:33 avg_step=3.64s recent_step=3.00s eta=35:12:10 +2026-09-15 05:09:32 | INFO | train rank=0 | step=15210/50000 progress=30.4% loss=0.098715 lr=8.334990e-05 lr_backbone=8.334990e-05 elapsed=15:23:04 avg_step=3.64s recent_step=3.11s eta=35:11:22 +2026-09-15 05:10:02 | INFO | train rank=0 | step=15220/50000 progress=30.4% loss=0.109751 lr=8.332525e-05 lr_backbone=8.332525e-05 elapsed=15:23:35 avg_step=3.64s recent_step=3.09s eta=35:10:33 +2026-09-15 05:10:57 | INFO | train rank=0 | step=15230/50000 progress=30.5% loss=0.096015 lr=8.330059e-05 lr_backbone=8.330059e-05 elapsed=15:24:30 avg_step=3.64s recent_step=5.43s eta=35:10:37 +2026-09-15 05:11:34 | INFO | train rank=0 | step=15240/50000 progress=30.5% loss=0.120312 lr=8.327592e-05 lr_backbone=8.327592e-05 elapsed=15:25:07 avg_step=3.64s recent_step=3.72s eta=35:10:03 +2026-09-15 05:12:03 | INFO | train rank=0 | step=15250/50000 progress=30.5% loss=0.096284 lr=8.325123e-05 lr_backbone=8.325123e-05 elapsed=15:25:36 avg_step=3.64s recent_step=2.90s eta=35:09:09 +2026-09-15 05:12:38 | INFO | train rank=0 | step=15260/50000 progress=30.5% loss=0.140476 lr=8.322652e-05 lr_backbone=8.322652e-05 elapsed=15:26:11 avg_step=3.64s recent_step=3.55s eta=35:08:31 +2026-09-15 05:13:18 | INFO | train rank=0 | step=15270/50000 progress=30.5% loss=0.109620 lr=8.320181e-05 lr_backbone=8.320181e-05 elapsed=15:26:50 avg_step=3.64s recent_step=3.93s eta=35:08:01 +2026-09-15 05:13:57 | INFO | train rank=0 | step=15280/50000 progress=30.6% loss=0.103296 lr=8.317707e-05 lr_backbone=8.317707e-05 elapsed=15:27:30 avg_step=3.64s recent_step=3.94s eta=35:07:31 +2026-09-15 05:14:33 | INFO | train rank=0 | step=15290/50000 progress=30.6% loss=0.092134 lr=8.315232e-05 lr_backbone=8.315232e-05 elapsed=15:28:06 avg_step=3.64s recent_step=3.64s eta=35:06:54 +2026-09-15 05:15:05 | INFO | train rank=0 | step=15300/50000 progress=30.6% loss=0.109053 lr=8.312756e-05 lr_backbone=8.312756e-05 elapsed=15:28:38 avg_step=3.64s recent_step=3.14s eta=35:06:07 +2026-09-15 05:15:51 | INFO | train rank=0 | step=15310/50000 progress=30.6% loss=0.114901 lr=8.310279e-05 lr_backbone=8.310279e-05 elapsed=15:29:24 avg_step=3.64s recent_step=4.64s eta=35:05:53 +2026-09-15 05:16:27 | INFO | train rank=0 | step=15320/50000 progress=30.6% loss=0.095401 lr=8.307799e-05 lr_backbone=8.307799e-05 elapsed=15:30:00 avg_step=3.64s recent_step=3.58s eta=35:05:15 +2026-09-15 05:16:59 | INFO | train rank=0 | step=15330/50000 progress=30.7% loss=0.094473 lr=8.305319e-05 lr_backbone=8.305319e-05 elapsed=15:30:32 avg_step=3.64s recent_step=3.24s eta=35:04:30 +2026-09-15 05:17:32 | INFO | train rank=0 | step=15340/50000 progress=30.7% loss=0.119732 lr=8.302837e-05 lr_backbone=8.302837e-05 elapsed=15:31:05 avg_step=3.64s recent_step=3.28s eta=35:03:45 +2026-09-15 05:18:17 | INFO | train rank=0 | step=15350/50000 progress=30.7% loss=0.086928 lr=8.300353e-05 lr_backbone=8.300353e-05 elapsed=15:31:50 avg_step=3.64s recent_step=4.51s eta=35:03:28 +2026-09-15 05:18:54 | INFO | train rank=0 | step=15360/50000 progress=30.7% loss=0.101431 lr=8.297868e-05 lr_backbone=8.297868e-05 elapsed=15:32:27 avg_step=3.64s recent_step=3.65s eta=35:02:52 +2026-09-15 05:19:29 | INFO | train rank=0 | step=15370/50000 progress=30.7% loss=0.095948 lr=8.295382e-05 lr_backbone=8.295382e-05 elapsed=15:33:02 avg_step=3.64s recent_step=3.49s eta=35:02:12 +2026-09-15 05:20:06 | INFO | train rank=0 | step=15380/50000 progress=30.8% loss=0.144902 lr=8.292894e-05 lr_backbone=8.292894e-05 elapsed=15:33:39 avg_step=3.64s recent_step=3.73s eta=35:01:38 +2026-09-15 05:20:41 | INFO | train rank=0 | step=15390/50000 progress=30.8% loss=0.104751 lr=8.290405e-05 lr_backbone=8.290405e-05 elapsed=15:34:14 avg_step=3.64s recent_step=3.50s eta=35:00:58 +2026-09-15 05:21:26 | INFO | train rank=0 | step=15400/50000 progress=30.8% loss=0.121389 lr=8.287914e-05 lr_backbone=8.287914e-05 elapsed=15:34:59 avg_step=3.64s recent_step=4.48s eta=35:00:40 +2026-09-15 05:22:03 | INFO | train rank=0 | step=15410/50000 progress=30.8% loss=0.116074 lr=8.285422e-05 lr_backbone=8.285422e-05 elapsed=15:35:36 avg_step=3.64s recent_step=3.77s eta=35:00:07 +2026-09-15 05:22:32 | INFO | train rank=0 | step=15420/50000 progress=30.8% loss=0.119387 lr=8.282929e-05 lr_backbone=8.282929e-05 elapsed=15:36:05 avg_step=3.64s recent_step=2.85s eta=34:59:12 +2026-09-15 05:23:25 | INFO | train rank=0 | step=15430/50000 progress=30.9% loss=0.098672 lr=8.280434e-05 lr_backbone=8.280434e-05 elapsed=15:36:58 avg_step=3.64s recent_step=5.29s eta=34:59:13 +2026-09-15 05:24:00 | INFO | train rank=0 | step=15440/50000 progress=30.9% loss=0.097650 lr=8.277937e-05 lr_backbone=8.277937e-05 elapsed=15:37:33 avg_step=3.64s recent_step=3.51s eta=34:58:34 +2026-09-15 05:24:36 | INFO | train rank=0 | step=15450/50000 progress=30.9% loss=0.131905 lr=8.275440e-05 lr_backbone=8.275440e-05 elapsed=15:38:09 avg_step=3.64s recent_step=3.60s eta=34:57:56 +2026-09-15 05:25:07 | INFO | train rank=0 | step=15460/50000 progress=30.9% loss=0.109733 lr=8.272940e-05 lr_backbone=8.272940e-05 elapsed=15:38:39 avg_step=3.64s recent_step=3.05s eta=34:57:07 +2026-09-15 05:25:51 | INFO | train rank=0 | step=15470/50000 progress=30.9% loss=0.117129 lr=8.270440e-05 lr_backbone=8.270440e-05 elapsed=15:39:24 avg_step=3.64s recent_step=4.45s eta=34:56:48 +2026-09-15 05:26:30 | INFO | train rank=0 | step=15480/50000 progress=31.0% loss=0.110523 lr=8.267937e-05 lr_backbone=8.267937e-05 elapsed=15:40:02 avg_step=3.64s recent_step=3.87s eta=34:56:17 +2026-09-15 05:27:06 | INFO | train rank=0 | step=15490/50000 progress=31.0% loss=0.111321 lr=8.265434e-05 lr_backbone=8.265434e-05 elapsed=15:40:38 avg_step=3.64s recent_step=3.59s eta=34:55:39 +2026-09-15 05:27:35 | INFO | train rank=0 | step=15500/50000 progress=31.0% loss=0.094982 lr=8.262929e-05 lr_backbone=8.262929e-05 elapsed=15:41:08 avg_step=3.64s recent_step=2.97s eta=34:54:48 +2026-09-15 05:28:15 | INFO | train rank=0 | step=15510/50000 progress=31.0% loss=0.076492 lr=8.260422e-05 lr_backbone=8.260422e-05 elapsed=15:41:48 avg_step=3.64s recent_step=4.00s eta=34:54:19 +2026-09-15 05:28:51 | INFO | train rank=0 | step=15520/50000 progress=31.0% loss=0.110031 lr=8.257915e-05 lr_backbone=8.257915e-05 elapsed=15:42:24 avg_step=3.64s recent_step=3.60s eta=34:53:42 +2026-09-15 05:29:31 | INFO | train rank=0 | step=15530/50000 progress=31.1% loss=0.099868 lr=8.255405e-05 lr_backbone=8.255405e-05 elapsed=15:43:04 avg_step=3.64s recent_step=3.97s eta=34:53:13 +2026-09-15 05:30:02 | INFO | train rank=0 | step=15540/50000 progress=31.1% loss=0.109279 lr=8.252895e-05 lr_backbone=8.252895e-05 elapsed=15:43:34 avg_step=3.64s recent_step=3.06s eta=34:52:23 +2026-09-15 05:30:48 | INFO | train rank=0 | step=15550/50000 progress=31.1% loss=0.105180 lr=8.250382e-05 lr_backbone=8.250382e-05 elapsed=15:44:21 avg_step=3.64s recent_step=4.62s eta=34:52:09 +2026-09-15 05:31:19 | INFO | train rank=0 | step=15560/50000 progress=31.1% loss=0.104694 lr=8.247869e-05 lr_backbone=8.247869e-05 elapsed=15:44:52 avg_step=3.64s recent_step=3.12s eta=34:51:21 +2026-09-15 05:31:52 | INFO | train rank=0 | step=15570/50000 progress=31.1% loss=0.117410 lr=8.245354e-05 lr_backbone=8.245354e-05 elapsed=15:45:24 avg_step=3.64s recent_step=3.26s eta=34:50:36 +2026-09-15 05:32:32 | INFO | train rank=0 | step=15580/50000 progress=31.2% loss=0.093406 lr=8.242837e-05 lr_backbone=8.242837e-05 elapsed=15:46:05 avg_step=3.64s recent_step=4.07s eta=34:50:09 +2026-09-15 05:33:17 | INFO | train rank=0 | step=15590/50000 progress=31.2% loss=0.103907 lr=8.240320e-05 lr_backbone=8.240320e-05 elapsed=15:46:50 avg_step=3.64s recent_step=4.50s eta=34:49:51 +2026-09-15 05:33:56 | INFO | train rank=0 | step=15600/50000 progress=31.2% loss=0.104743 lr=8.237800e-05 lr_backbone=8.237800e-05 elapsed=15:47:29 avg_step=3.64s recent_step=3.84s eta=34:49:19 +2026-09-15 05:34:34 | INFO | train rank=0 | step=15610/50000 progress=31.2% loss=0.081583 lr=8.235280e-05 lr_backbone=8.235280e-05 elapsed=15:48:07 avg_step=3.64s recent_step=3.84s eta=34:48:47 +2026-09-15 05:34:57 | INFO | train rank=0 | step=15620/50000 progress=31.2% loss=0.104381 lr=8.232758e-05 lr_backbone=8.232758e-05 elapsed=15:48:29 avg_step=3.64s recent_step=2.24s eta=34:47:40 +2026-09-15 05:35:44 | INFO | train rank=0 | step=15630/50000 progress=31.3% loss=0.078998 lr=8.230234e-05 lr_backbone=8.230234e-05 elapsed=15:49:17 avg_step=3.64s recent_step=4.73s eta=34:47:27 +2026-09-15 05:36:24 | INFO | train rank=0 | step=15640/50000 progress=31.3% loss=0.096372 lr=8.227709e-05 lr_backbone=8.227709e-05 elapsed=15:49:57 avg_step=3.64s recent_step=4.05s eta=34:46:59 +2026-09-15 05:37:04 | INFO | train rank=0 | step=15650/50000 progress=31.3% loss=0.131475 lr=8.225183e-05 lr_backbone=8.225183e-05 elapsed=15:50:37 avg_step=3.64s recent_step=4.01s eta=34:46:31 +2026-09-15 05:37:39 | INFO | train rank=0 | step=15660/50000 progress=31.3% loss=0.102513 lr=8.222655e-05 lr_backbone=8.222655e-05 elapsed=15:51:12 avg_step=3.64s recent_step=3.46s eta=34:45:50 +2026-09-15 05:38:28 | INFO | train rank=0 | step=15670/50000 progress=31.3% loss=0.098192 lr=8.220126e-05 lr_backbone=8.220126e-05 elapsed=15:52:01 avg_step=3.65s recent_step=4.91s eta=34:45:42 +2026-09-15 05:38:58 | INFO | train rank=0 | step=15680/50000 progress=31.4% loss=0.114777 lr=8.217596e-05 lr_backbone=8.217596e-05 elapsed=15:52:31 avg_step=3.64s recent_step=2.98s eta=34:44:51 +2026-09-15 05:39:19 | INFO | train rank=0 | step=15690/50000 progress=31.4% loss=0.115537 lr=8.215064e-05 lr_backbone=8.215064e-05 elapsed=15:52:52 avg_step=3.64s recent_step=2.15s eta=34:43:42 +2026-09-15 05:39:49 | INFO | train rank=0 | step=15700/50000 progress=31.4% loss=0.099342 lr=8.212530e-05 lr_backbone=8.212530e-05 elapsed=15:53:22 avg_step=3.64s recent_step=2.93s eta=34:42:50 +2026-09-15 05:40:38 | INFO | train rank=0 | step=15710/50000 progress=31.4% loss=0.112180 lr=8.209996e-05 lr_backbone=8.209996e-05 elapsed=15:54:11 avg_step=3.64s recent_step=4.97s eta=34:42:42 +2026-09-15 05:41:18 | INFO | train rank=0 | step=15720/50000 progress=31.4% loss=0.118886 lr=8.207460e-05 lr_backbone=8.207460e-05 elapsed=15:54:50 avg_step=3.64s recent_step=3.92s eta=34:42:12 +2026-09-15 05:41:52 | INFO | train rank=0 | step=15730/50000 progress=31.5% loss=0.071048 lr=8.204922e-05 lr_backbone=8.204922e-05 elapsed=15:55:24 avg_step=3.64s recent_step=3.40s eta=34:41:30 +2026-09-15 05:42:27 | INFO | train rank=0 | step=15740/50000 progress=31.5% loss=0.104301 lr=8.202383e-05 lr_backbone=8.202383e-05 elapsed=15:56:00 avg_step=3.64s recent_step=3.54s eta=34:40:51 +2026-09-15 05:43:09 | INFO | train rank=0 | step=15750/50000 progress=31.5% loss=0.148753 lr=8.199843e-05 lr_backbone=8.199843e-05 elapsed=15:56:42 avg_step=3.64s recent_step=4.23s eta=34:40:28 +2026-09-15 05:43:34 | INFO | train rank=0 | step=15760/50000 progress=31.5% loss=0.121771 lr=8.197301e-05 lr_backbone=8.197301e-05 elapsed=15:57:07 avg_step=3.64s recent_step=2.47s eta=34:39:26 +2026-09-15 05:44:08 | INFO | train rank=0 | step=15770/50000 progress=31.5% loss=0.123708 lr=8.194758e-05 lr_backbone=8.194758e-05 elapsed=15:57:41 avg_step=3.64s recent_step=3.42s eta=34:38:44 +2026-09-15 05:44:37 | INFO | train rank=0 | step=15780/50000 progress=31.6% loss=0.110237 lr=8.192213e-05 lr_backbone=8.192213e-05 elapsed=15:58:09 avg_step=3.64s recent_step=2.85s eta=34:37:50 +2026-09-15 05:45:20 | INFO | train rank=0 | step=15790/50000 progress=31.6% loss=0.101088 lr=8.189667e-05 lr_backbone=8.189667e-05 elapsed=15:58:52 avg_step=3.64s recent_step=4.29s eta=34:37:28 +2026-09-15 05:45:57 | INFO | train rank=0 | step=15800/50000 progress=31.6% loss=0.114770 lr=8.187120e-05 lr_backbone=8.187120e-05 elapsed=15:59:30 avg_step=3.64s recent_step=3.73s eta=34:36:53 +2026-09-15 05:46:28 | INFO | train rank=0 | step=15810/50000 progress=31.6% loss=0.120109 lr=8.184571e-05 lr_backbone=8.184571e-05 elapsed=16:00:01 avg_step=3.64s recent_step=3.13s eta=34:36:06 +2026-09-15 05:47:05 | INFO | train rank=0 | step=15820/50000 progress=31.6% loss=0.080242 lr=8.182021e-05 lr_backbone=8.182021e-05 elapsed=16:00:38 avg_step=3.64s recent_step=3.69s eta=34:35:31 +2026-09-15 05:47:56 | INFO | train rank=0 | step=15830/50000 progress=31.7% loss=0.111584 lr=8.179470e-05 lr_backbone=8.179470e-05 elapsed=16:01:29 avg_step=3.64s recent_step=5.06s eta=34:35:25 +2026-09-15 05:48:28 | INFO | train rank=0 | step=15840/50000 progress=31.7% loss=0.100932 lr=8.176917e-05 lr_backbone=8.176917e-05 elapsed=16:02:01 avg_step=3.64s recent_step=3.26s eta=34:34:40 +2026-09-15 05:48:56 | INFO | train rank=0 | step=15850/50000 progress=31.7% loss=0.098815 lr=8.174362e-05 lr_backbone=8.174362e-05 elapsed=16:02:29 avg_step=3.64s recent_step=2.78s eta=34:33:45 +2026-09-15 05:49:29 | INFO | train rank=0 | step=15860/50000 progress=31.7% loss=0.110197 lr=8.171807e-05 lr_backbone=8.171807e-05 elapsed=16:03:02 avg_step=3.64s recent_step=3.27s eta=34:33:00 +2026-09-15 05:50:17 | INFO | train rank=0 | step=15870/50000 progress=31.7% loss=0.098644 lr=8.169250e-05 lr_backbone=8.169250e-05 elapsed=16:03:50 avg_step=3.64s recent_step=4.81s eta=34:32:49 +2026-09-15 05:50:49 | INFO | train rank=0 | step=15880/50000 progress=31.8% loss=0.110073 lr=8.166691e-05 lr_backbone=8.166691e-05 elapsed=16:04:22 avg_step=3.64s recent_step=3.25s eta=34:32:04 +2026-09-15 05:51:23 | INFO | train rank=0 | step=15890/50000 progress=31.8% loss=0.117551 lr=8.164131e-05 lr_backbone=8.164131e-05 elapsed=16:04:56 avg_step=3.64s recent_step=3.35s eta=34:31:21 +2026-09-15 05:52:01 | INFO | train rank=0 | step=15900/50000 progress=31.8% loss=0.116527 lr=8.161570e-05 lr_backbone=8.161570e-05 elapsed=16:05:34 avg_step=3.64s recent_step=3.81s eta=34:30:49 +2026-09-15 05:52:47 | INFO | train rank=0 | step=15910/50000 progress=31.8% loss=0.111421 lr=8.159007e-05 lr_backbone=8.159007e-05 elapsed=16:06:20 avg_step=3.64s recent_step=4.59s eta=34:30:32 +2026-09-15 05:53:20 | INFO | train rank=0 | step=15920/50000 progress=31.8% loss=0.137021 lr=8.156443e-05 lr_backbone=8.156443e-05 elapsed=16:06:52 avg_step=3.64s recent_step=3.28s eta=34:29:48 +2026-09-15 05:53:59 | INFO | train rank=0 | step=15930/50000 progress=31.9% loss=0.093447 lr=8.153878e-05 lr_backbone=8.153878e-05 elapsed=16:07:31 avg_step=3.64s recent_step=3.89s eta=34:29:17 +2026-09-15 05:54:29 | INFO | train rank=0 | step=15940/50000 progress=31.9% loss=0.114341 lr=8.151311e-05 lr_backbone=8.151311e-05 elapsed=16:08:02 avg_step=3.64s recent_step=3.03s eta=34:28:27 +2026-09-15 05:55:15 | INFO | train rank=0 | step=15950/50000 progress=31.9% loss=0.078759 lr=8.148743e-05 lr_backbone=8.148743e-05 elapsed=16:08:48 avg_step=3.64s recent_step=4.65s eta=34:28:12 +2026-09-15 05:55:57 | INFO | train rank=0 | step=15960/50000 progress=31.9% loss=0.112049 lr=8.146174e-05 lr_backbone=8.146174e-05 elapsed=16:09:30 avg_step=3.64s recent_step=4.18s eta=34:27:47 +2026-09-15 05:56:28 | INFO | train rank=0 | step=15970/50000 progress=31.9% loss=0.095808 lr=8.143603e-05 lr_backbone=8.143603e-05 elapsed=16:10:01 avg_step=3.64s recent_step=3.11s eta=34:27:00 +2026-09-15 05:56:58 | INFO | train rank=0 | step=15980/50000 progress=32.0% loss=0.109188 lr=8.141030e-05 lr_backbone=8.141030e-05 elapsed=16:10:31 avg_step=3.64s recent_step=2.97s eta=34:26:09 +2026-09-15 05:57:46 | INFO | train rank=0 | step=15990/50000 progress=32.0% loss=0.068403 lr=8.138457e-05 lr_backbone=8.138457e-05 elapsed=16:11:19 avg_step=3.64s recent_step=4.78s eta=34:25:56 +2026-09-15 05:58:18 | INFO | train rank=0 | step=16000/50000 progress=32.0% loss=0.103202 lr=8.135882e-05 lr_backbone=8.135882e-05 elapsed=16:11:51 avg_step=3.64s recent_step=3.27s eta=34:25:12 +2026-09-15 05:58:49 | INFO | train rank=0 | step=16010/50000 progress=32.0% loss=0.086491 lr=8.133305e-05 lr_backbone=8.133305e-05 elapsed=16:12:22 avg_step=3.64s recent_step=3.06s eta=34:24:23 +2026-09-15 05:59:19 | INFO | train rank=0 | step=16020/50000 progress=32.0% loss=0.116558 lr=8.130728e-05 lr_backbone=8.130728e-05 elapsed=16:12:52 avg_step=3.64s recent_step=2.97s eta=34:23:32 +2026-09-15 06:00:12 | INFO | train rank=0 | step=16030/50000 progress=32.1% loss=0.086217 lr=8.128149e-05 lr_backbone=8.128149e-05 elapsed=16:13:44 avg_step=3.64s recent_step=5.28s eta=34:23:31 +2026-09-15 06:00:43 | INFO | train rank=0 | step=16040/50000 progress=32.1% loss=0.146504 lr=8.125568e-05 lr_backbone=8.125568e-05 elapsed=16:14:16 avg_step=3.64s recent_step=3.16s eta=34:22:44 +2026-09-15 06:01:16 | INFO | train rank=0 | step=16050/50000 progress=32.1% loss=0.091026 lr=8.122986e-05 lr_backbone=8.122986e-05 elapsed=16:14:48 avg_step=3.64s recent_step=3.25s eta=34:21:59 +2026-09-15 06:01:49 | INFO | train rank=0 | step=16060/50000 progress=32.1% loss=0.104531 lr=8.120403e-05 lr_backbone=8.120403e-05 elapsed=16:15:22 avg_step=3.64s recent_step=3.38s eta=34:21:17 +2026-09-15 06:02:40 | INFO | train rank=0 | step=16070/50000 progress=32.1% loss=0.122804 lr=8.117818e-05 lr_backbone=8.117818e-05 elapsed=16:16:13 avg_step=3.64s recent_step=5.04s eta=34:21:10 +2026-09-15 06:03:15 | INFO | train rank=0 | step=16080/50000 progress=32.2% loss=0.064690 lr=8.115232e-05 lr_backbone=8.115232e-05 elapsed=16:16:48 avg_step=3.64s recent_step=3.52s eta=34:20:31 +2026-09-15 06:03:51 | INFO | train rank=0 | step=16090/50000 progress=32.2% loss=0.113306 lr=8.112645e-05 lr_backbone=8.112645e-05 elapsed=16:17:24 avg_step=3.64s recent_step=3.63s eta=34:19:54 +2026-09-15 06:04:23 | INFO | train rank=0 | step=16100/50000 progress=32.2% loss=0.116788 lr=8.110056e-05 lr_backbone=8.110056e-05 elapsed=16:17:55 avg_step=3.64s recent_step=3.13s eta=34:19:07 +2026-09-15 06:05:10 | INFO | train rank=0 | step=16110/50000 progress=32.2% loss=0.113024 lr=8.107466e-05 lr_backbone=8.107466e-05 elapsed=16:18:43 avg_step=3.65s recent_step=4.76s eta=34:18:54 +2026-09-15 06:05:42 | INFO | train rank=0 | step=16120/50000 progress=32.2% loss=0.095234 lr=8.104875e-05 lr_backbone=8.104875e-05 elapsed=16:19:14 avg_step=3.64s recent_step=3.14s eta=34:18:07 +2026-09-15 06:06:14 | INFO | train rank=0 | step=16130/50000 progress=32.3% loss=0.099928 lr=8.102282e-05 lr_backbone=8.102282e-05 elapsed=16:19:47 avg_step=3.64s recent_step=3.26s eta=34:17:22 +2026-09-15 06:06:40 | INFO | train rank=0 | step=16140/50000 progress=32.3% loss=0.130520 lr=8.099688e-05 lr_backbone=8.099688e-05 elapsed=16:20:13 avg_step=3.64s recent_step=2.57s eta=34:16:23 +2026-09-15 06:07:28 | INFO | train rank=0 | step=16150/50000 progress=32.3% loss=0.135004 lr=8.097093e-05 lr_backbone=8.097093e-05 elapsed=16:21:00 avg_step=3.64s recent_step=4.77s eta=34:16:11 +2026-09-15 06:08:17 | INFO | train rank=0 | step=16160/50000 progress=32.3% loss=0.128753 lr=8.094496e-05 lr_backbone=8.094496e-05 elapsed=16:21:50 avg_step=3.65s recent_step=4.96s eta=34:16:02 +2026-09-15 06:08:49 | INFO | train rank=0 | step=16170/50000 progress=32.3% loss=0.118202 lr=8.091898e-05 lr_backbone=8.091898e-05 elapsed=16:22:22 avg_step=3.65s recent_step=3.18s eta=34:15:16 +2026-09-15 06:09:27 | INFO | train rank=0 | step=16180/50000 progress=32.4% loss=0.106846 lr=8.089298e-05 lr_backbone=8.089298e-05 elapsed=16:23:00 avg_step=3.65s recent_step=3.78s eta=34:14:42 +2026-09-15 06:10:02 | INFO | train rank=0 | step=16190/50000 progress=32.4% loss=0.160812 lr=8.086697e-05 lr_backbone=8.086697e-05 elapsed=16:23:35 avg_step=3.65s recent_step=3.50s eta=34:14:03 +2026-09-15 06:10:50 | INFO | train rank=0 | step=16200/50000 progress=32.4% loss=0.119008 lr=8.084095e-05 lr_backbone=8.084095e-05 elapsed=16:24:23 avg_step=3.65s recent_step=4.78s eta=34:13:50 +2026-09-15 06:11:19 | INFO | train rank=0 | step=16210/50000 progress=32.4% loss=0.127663 lr=8.081491e-05 lr_backbone=8.081491e-05 elapsed=16:24:52 avg_step=3.65s recent_step=2.96s eta=34:12:59 +2026-09-15 06:11:55 | INFO | train rank=0 | step=16220/50000 progress=32.4% loss=0.115933 lr=8.078886e-05 lr_backbone=8.078886e-05 elapsed=16:25:28 avg_step=3.65s recent_step=3.55s eta=34:12:21 +2026-09-15 06:12:27 | INFO | train rank=0 | step=16230/50000 progress=32.5% loss=0.122136 lr=8.076280e-05 lr_backbone=8.076280e-05 elapsed=16:26:00 avg_step=3.65s recent_step=3.26s eta=34:11:36 +2026-09-15 06:13:10 | INFO | train rank=0 | step=16240/50000 progress=32.5% loss=0.105434 lr=8.073673e-05 lr_backbone=8.073673e-05 elapsed=16:26:43 avg_step=3.65s recent_step=4.26s eta=34:11:13 +2026-09-15 06:13:53 | INFO | train rank=0 | step=16250/50000 progress=32.5% loss=0.113611 lr=8.071064e-05 lr_backbone=8.071064e-05 elapsed=16:27:26 avg_step=3.65s recent_step=4.26s eta=34:10:49 +2026-09-15 06:14:28 | INFO | train rank=0 | step=16260/50000 progress=32.5% loss=0.105421 lr=8.068453e-05 lr_backbone=8.068453e-05 elapsed=16:28:01 avg_step=3.65s recent_step=3.52s eta=34:10:10 +2026-09-15 06:15:08 | INFO | train rank=0 | step=16270/50000 progress=32.5% loss=0.130514 lr=8.065842e-05 lr_backbone=8.065842e-05 elapsed=16:28:40 avg_step=3.65s recent_step=3.96s eta=34:09:40 +2026-09-15 06:16:03 | INFO | train rank=0 | step=16280/50000 progress=32.6% loss=0.109596 lr=8.063229e-05 lr_backbone=8.063229e-05 elapsed=16:29:36 avg_step=3.65s recent_step=5.54s eta=34:09:43 +2026-09-15 06:16:26 | INFO | train rank=0 | step=16290/50000 progress=32.6% loss=0.089763 lr=8.060614e-05 lr_backbone=8.060614e-05 elapsed=16:29:59 avg_step=3.65s recent_step=2.36s eta=34:08:39 +2026-09-15 06:16:57 | INFO | train rank=0 | step=16300/50000 progress=32.6% loss=0.068810 lr=8.057999e-05 lr_backbone=8.057999e-05 elapsed=16:30:30 avg_step=3.65s recent_step=3.08s eta=34:07:51 +2026-09-15 06:17:33 | INFO | train rank=0 | step=16310/50000 progress=32.6% loss=0.126803 lr=8.055382e-05 lr_backbone=8.055382e-05 elapsed=16:31:06 avg_step=3.65s recent_step=3.57s eta=34:07:13 +2026-09-15 06:18:21 | INFO | train rank=0 | step=16320/50000 progress=32.6% loss=0.082291 lr=8.052763e-05 lr_backbone=8.052763e-05 elapsed=16:31:53 avg_step=3.65s recent_step=4.75s eta=34:07:00 +2026-09-15 06:18:55 | INFO | train rank=0 | step=16330/50000 progress=32.7% loss=0.098472 lr=8.050144e-05 lr_backbone=8.050144e-05 elapsed=16:32:28 avg_step=3.65s recent_step=3.44s eta=34:06:19 +2026-09-15 06:19:24 | INFO | train rank=0 | step=16340/50000 progress=32.7% loss=0.094489 lr=8.047523e-05 lr_backbone=8.047523e-05 elapsed=16:32:56 avg_step=3.65s recent_step=2.87s eta=34:05:27 +2026-09-15 06:19:59 | INFO | train rank=0 | step=16350/50000 progress=32.7% loss=0.116621 lr=8.044900e-05 lr_backbone=8.044900e-05 elapsed=16:33:32 avg_step=3.65s recent_step=3.55s eta=34:04:48 +2026-09-15 06:20:50 | INFO | train rank=0 | step=16360/50000 progress=32.7% loss=0.101114 lr=8.042277e-05 lr_backbone=8.042277e-05 elapsed=16:34:23 avg_step=3.65s recent_step=5.10s eta=34:04:41 +2026-09-15 06:21:19 | INFO | train rank=0 | step=16370/50000 progress=32.7% loss=0.107385 lr=8.039651e-05 lr_backbone=8.039651e-05 elapsed=16:34:51 avg_step=3.65s recent_step=2.85s eta=34:03:49 +2026-09-15 06:21:56 | INFO | train rank=0 | step=16380/50000 progress=32.8% loss=0.107302 lr=8.037025e-05 lr_backbone=8.037025e-05 elapsed=16:35:29 avg_step=3.65s recent_step=3.71s eta=34:03:14 +2026-09-15 06:22:31 | INFO | train rank=0 | step=16390/50000 progress=32.8% loss=0.115413 lr=8.034397e-05 lr_backbone=8.034397e-05 elapsed=16:36:03 avg_step=3.65s recent_step=3.48s eta=34:02:34 +2026-09-15 06:23:21 | INFO | train rank=0 | step=16400/50000 progress=32.8% loss=0.112749 lr=8.031768e-05 lr_backbone=8.031768e-05 elapsed=16:36:54 avg_step=3.65s recent_step=5.05s eta=34:02:26 +2026-09-15 06:23:59 | INFO | train rank=0 | step=16410/50000 progress=32.8% loss=0.182075 lr=8.029138e-05 lr_backbone=8.029138e-05 elapsed=16:37:32 avg_step=3.65s recent_step=3.79s eta=34:01:52 +2026-09-15 06:24:34 | INFO | train rank=0 | step=16420/50000 progress=32.8% loss=0.125927 lr=8.026507e-05 lr_backbone=8.026507e-05 elapsed=16:38:07 avg_step=3.65s recent_step=3.56s eta=34:01:14 +2026-09-15 06:25:02 | INFO | train rank=0 | step=16430/50000 progress=32.9% loss=0.075131 lr=8.023874e-05 lr_backbone=8.023874e-05 elapsed=16:38:35 avg_step=3.65s recent_step=2.78s eta=34:00:20 +2026-09-15 06:26:01 | INFO | train rank=0 | step=16440/50000 progress=32.9% loss=0.103619 lr=8.021239e-05 lr_backbone=8.021239e-05 elapsed=16:39:34 avg_step=3.65s recent_step=5.86s eta=34:00:28 +2026-09-15 06:26:31 | INFO | train rank=0 | step=16450/50000 progress=32.9% loss=0.114677 lr=8.018604e-05 lr_backbone=8.018604e-05 elapsed=16:40:04 avg_step=3.65s recent_step=3.04s eta=33:59:40 +2026-09-15 06:27:03 | INFO | train rank=0 | step=16460/50000 progress=32.9% loss=0.078950 lr=8.015967e-05 lr_backbone=8.015967e-05 elapsed=16:40:35 avg_step=3.65s recent_step=3.14s eta=33:58:53 +2026-09-15 06:27:31 | INFO | train rank=0 | step=16470/50000 progress=32.9% loss=0.090221 lr=8.013328e-05 lr_backbone=8.013328e-05 elapsed=16:41:03 avg_step=3.65s recent_step=2.80s eta=33:57:59 +2026-09-15 06:28:17 | INFO | train rank=0 | step=16480/50000 progress=33.0% loss=0.118109 lr=8.010689e-05 lr_backbone=8.010689e-05 elapsed=16:41:50 avg_step=3.65s recent_step=4.68s eta=33:57:44 +2026-09-15 06:28:51 | INFO | train rank=0 | step=16490/50000 progress=33.0% loss=0.091156 lr=8.008048e-05 lr_backbone=8.008048e-05 elapsed=16:42:24 avg_step=3.65s recent_step=3.36s eta=33:57:01 +2026-09-15 06:29:24 | INFO | train rank=0 | step=16500/50000 progress=33.0% loss=0.094751 lr=8.005406e-05 lr_backbone=8.005406e-05 elapsed=16:42:57 avg_step=3.65s recent_step=3.28s eta=33:56:17 +2026-09-15 06:29:59 | INFO | train rank=0 | step=16510/50000 progress=33.0% loss=0.118239 lr=8.002762e-05 lr_backbone=8.002762e-05 elapsed=16:43:32 avg_step=3.65s recent_step=3.56s eta=33:55:39 +2026-09-15 06:30:48 | INFO | train rank=0 | step=16520/50000 progress=33.0% loss=0.104944 lr=8.000117e-05 lr_backbone=8.000117e-05 elapsed=16:44:20 avg_step=3.65s recent_step=4.81s eta=33:55:26 +2026-09-15 06:31:19 | INFO | train rank=0 | step=16530/50000 progress=33.1% loss=0.112695 lr=7.997471e-05 lr_backbone=7.997471e-05 elapsed=16:44:52 avg_step=3.65s recent_step=3.17s eta=33:54:40 +2026-09-15 06:31:53 | INFO | train rank=0 | step=16540/50000 progress=33.1% loss=0.095964 lr=7.994824e-05 lr_backbone=7.994824e-05 elapsed=16:45:26 avg_step=3.65s recent_step=3.37s eta=33:53:58 +2026-09-15 06:32:14 | INFO | train rank=0 | step=16550/50000 progress=33.1% loss=0.118777 lr=7.992175e-05 lr_backbone=7.992175e-05 elapsed=16:45:47 avg_step=3.65s recent_step=2.12s eta=33:52:51 +2026-09-15 06:33:04 | INFO | train rank=0 | step=16560/50000 progress=33.1% loss=0.097810 lr=7.989525e-05 lr_backbone=7.989525e-05 elapsed=16:46:36 avg_step=3.65s recent_step=4.94s eta=33:52:41 +2026-09-15 06:33:40 | INFO | train rank=0 | step=16570/50000 progress=33.1% loss=0.110764 lr=7.986874e-05 lr_backbone=7.986874e-05 elapsed=16:47:13 avg_step=3.65s recent_step=3.62s eta=33:52:03 +2026-09-15 06:34:22 | INFO | train rank=0 | step=16580/50000 progress=33.2% loss=0.096174 lr=7.984221e-05 lr_backbone=7.984221e-05 elapsed=16:47:55 avg_step=3.65s recent_step=4.20s eta=33:51:38 +2026-09-15 06:34:53 | INFO | train rank=0 | step=16590/50000 progress=33.2% loss=0.129032 lr=7.981567e-05 lr_backbone=7.981567e-05 elapsed=16:48:26 avg_step=3.65s recent_step=3.11s eta=33:50:51 +2026-09-15 06:35:49 | INFO | train rank=0 | step=16600/50000 progress=33.2% loss=0.134968 lr=7.978912e-05 lr_backbone=7.978912e-05 elapsed=16:49:22 avg_step=3.65s recent_step=5.58s eta=33:50:53 +2026-09-15 06:36:10 | INFO | train rank=0 | step=16610/50000 progress=33.2% loss=0.095686 lr=7.976255e-05 lr_backbone=7.976255e-05 elapsed=16:49:42 avg_step=3.65s recent_step=2.08s eta=33:49:45 +2026-09-15 06:36:40 | INFO | train rank=0 | step=16620/50000 progress=33.2% loss=0.090925 lr=7.973597e-05 lr_backbone=7.973597e-05 elapsed=16:50:13 avg_step=3.65s recent_step=3.04s eta=33:48:57 +2026-09-15 06:37:14 | INFO | train rank=0 | step=16630/50000 progress=33.3% loss=0.124561 lr=7.970938e-05 lr_backbone=7.970938e-05 elapsed=16:50:47 avg_step=3.65s recent_step=3.38s eta=33:48:15 +2026-09-15 06:38:02 | INFO | train rank=0 | step=16640/50000 progress=33.3% loss=0.109778 lr=7.968277e-05 lr_backbone=7.968277e-05 elapsed=16:51:35 avg_step=3.65s recent_step=4.82s eta=33:48:02 +2026-09-15 06:38:32 | INFO | train rank=0 | step=16650/50000 progress=33.3% loss=0.139651 lr=7.965615e-05 lr_backbone=7.965615e-05 elapsed=16:52:05 avg_step=3.65s recent_step=3.00s eta=33:47:12 +2026-09-15 06:39:11 | INFO | train rank=0 | step=16660/50000 progress=33.3% loss=0.104572 lr=7.962952e-05 lr_backbone=7.962952e-05 elapsed=16:52:44 avg_step=3.65s recent_step=3.89s eta=33:46:41 +2026-09-15 06:39:40 | INFO | train rank=0 | step=16670/50000 progress=33.3% loss=0.111408 lr=7.960288e-05 lr_backbone=7.960288e-05 elapsed=16:53:13 avg_step=3.65s recent_step=2.96s eta=33:45:50 +2026-09-15 06:40:36 | INFO | train rank=0 | step=16680/50000 progress=33.4% loss=0.086635 lr=7.957622e-05 lr_backbone=7.957622e-05 elapsed=16:54:09 avg_step=3.65s recent_step=5.60s eta=33:45:53 +2026-09-15 06:41:05 | INFO | train rank=0 | step=16690/50000 progress=33.4% loss=0.148924 lr=7.954955e-05 lr_backbone=7.954955e-05 elapsed=16:54:38 avg_step=3.65s recent_step=2.89s eta=33:45:01 +2026-09-15 06:41:36 | INFO | train rank=0 | step=16700/50000 progress=33.4% loss=0.092081 lr=7.952287e-05 lr_backbone=7.952287e-05 elapsed=16:55:08 avg_step=3.65s recent_step=3.03s eta=33:44:13 +2026-09-15 06:42:08 | INFO | train rank=0 | step=16710/50000 progress=33.4% loss=0.124074 lr=7.949617e-05 lr_backbone=7.949617e-05 elapsed=16:55:41 avg_step=3.65s recent_step=3.26s eta=33:43:29 +2026-09-15 06:42:52 | INFO | train rank=0 | step=16720/50000 progress=33.4% loss=0.117290 lr=7.946947e-05 lr_backbone=7.946947e-05 elapsed=16:56:24 avg_step=3.65s recent_step=4.32s eta=33:43:06 +2026-09-15 06:43:21 | INFO | train rank=0 | step=16730/50000 progress=33.5% loss=0.125532 lr=7.944274e-05 lr_backbone=7.944274e-05 elapsed=16:56:54 avg_step=3.65s recent_step=2.93s eta=33:42:15 +2026-09-15 06:43:52 | INFO | train rank=0 | step=16740/50000 progress=33.5% loss=0.099468 lr=7.941601e-05 lr_backbone=7.941601e-05 elapsed=16:57:25 avg_step=3.65s recent_step=3.09s eta=33:41:27 +2026-09-15 06:44:24 | INFO | train rank=0 | step=16750/50000 progress=33.5% loss=0.120974 lr=7.938926e-05 lr_backbone=7.938926e-05 elapsed=16:57:57 avg_step=3.65s recent_step=3.23s eta=33:40:43 +2026-09-15 06:45:06 | INFO | train rank=0 | step=16760/50000 progress=33.5% loss=0.091708 lr=7.936250e-05 lr_backbone=7.936250e-05 elapsed=16:58:39 avg_step=3.65s recent_step=4.23s eta=33:40:18 +2026-09-15 06:45:40 | INFO | train rank=0 | step=16770/50000 progress=33.5% loss=0.102767 lr=7.933573e-05 lr_backbone=7.933573e-05 elapsed=16:59:13 avg_step=3.65s recent_step=3.38s eta=33:39:36 +2026-09-15 06:46:15 | INFO | train rank=0 | step=16780/50000 progress=33.6% loss=0.105410 lr=7.930894e-05 lr_backbone=7.930894e-05 elapsed=16:59:48 avg_step=3.65s recent_step=3.52s eta=33:38:57 +2026-09-15 06:46:50 | INFO | train rank=0 | step=16790/50000 progress=33.6% loss=0.111960 lr=7.928215e-05 lr_backbone=7.928215e-05 elapsed=17:00:23 avg_step=3.65s recent_step=3.45s eta=33:38:16 +2026-09-15 06:47:46 | INFO | train rank=0 | step=16800/50000 progress=33.6% loss=0.108253 lr=7.925533e-05 lr_backbone=7.925533e-05 elapsed=17:01:19 avg_step=3.65s recent_step=5.66s eta=33:38:20 +2026-09-15 06:48:12 | INFO | train rank=0 | step=16810/50000 progress=33.6% loss=0.103743 lr=7.922851e-05 lr_backbone=7.922851e-05 elapsed=17:01:45 avg_step=3.65s recent_step=2.54s eta=33:37:21 +2026-09-15 06:48:48 | INFO | train rank=0 | step=16820/50000 progress=33.6% loss=0.101008 lr=7.920167e-05 lr_backbone=7.920167e-05 elapsed=17:02:21 avg_step=3.65s recent_step=3.62s eta=33:36:44 +2026-09-15 06:49:17 | INFO | train rank=0 | step=16830/50000 progress=33.7% loss=0.109665 lr=7.917482e-05 lr_backbone=7.917482e-05 elapsed=17:02:49 avg_step=3.65s recent_step=2.87s eta=33:35:53 +2026-09-15 06:50:01 | INFO | train rank=0 | step=16840/50000 progress=33.7% loss=0.128103 lr=7.914796e-05 lr_backbone=7.914796e-05 elapsed=17:03:33 avg_step=3.65s recent_step=4.41s eta=33:35:31 +2026-09-15 06:50:37 | INFO | train rank=0 | step=16850/50000 progress=33.7% loss=0.115889 lr=7.912108e-05 lr_backbone=7.912108e-05 elapsed=17:04:10 avg_step=3.65s recent_step=3.61s eta=33:34:54 +2026-09-15 06:51:13 | INFO | train rank=0 | step=16860/50000 progress=33.7% loss=0.090699 lr=7.909420e-05 lr_backbone=7.909420e-05 elapsed=17:04:46 avg_step=3.65s recent_step=3.63s eta=33:34:17 +2026-09-15 06:51:48 | INFO | train rank=0 | step=16870/50000 progress=33.7% loss=0.082267 lr=7.906730e-05 lr_backbone=7.906730e-05 elapsed=17:05:21 avg_step=3.65s recent_step=3.48s eta=33:33:37 +2026-09-15 06:52:29 | INFO | train rank=0 | step=16880/50000 progress=33.8% loss=0.101691 lr=7.904038e-05 lr_backbone=7.904038e-05 elapsed=17:06:02 avg_step=3.65s recent_step=4.10s eta=33:33:10 +2026-09-15 06:53:05 | INFO | train rank=0 | step=16890/50000 progress=33.8% loss=0.116339 lr=7.901346e-05 lr_backbone=7.901346e-05 elapsed=17:06:38 avg_step=3.65s recent_step=3.62s eta=33:32:33 +2026-09-15 06:53:44 | INFO | train rank=0 | step=16900/50000 progress=33.8% loss=0.090586 lr=7.898652e-05 lr_backbone=7.898652e-05 elapsed=17:07:16 avg_step=3.65s recent_step=3.86s eta=33:32:00 +2026-09-15 06:54:19 | INFO | train rank=0 | step=16910/50000 progress=33.8% loss=0.098853 lr=7.895957e-05 lr_backbone=7.895957e-05 elapsed=17:07:52 avg_step=3.65s recent_step=3.52s eta=33:31:21 +2026-09-15 06:54:59 | INFO | train rank=0 | step=16920/50000 progress=33.8% loss=0.099911 lr=7.893260e-05 lr_backbone=7.893260e-05 elapsed=17:08:32 avg_step=3.65s recent_step=4.03s eta=33:30:53 +2026-09-15 06:55:31 | INFO | train rank=0 | step=16930/50000 progress=33.9% loss=0.100183 lr=7.890562e-05 lr_backbone=7.890562e-05 elapsed=17:09:03 avg_step=3.65s recent_step=3.13s eta=33:30:06 +2026-09-15 06:56:07 | INFO | train rank=0 | step=16940/50000 progress=33.9% loss=0.082622 lr=7.887864e-05 lr_backbone=7.887864e-05 elapsed=17:09:40 avg_step=3.65s recent_step=3.64s eta=33:29:29 +2026-09-15 06:56:48 | INFO | train rank=0 | step=16950/50000 progress=33.9% loss=0.120929 lr=7.885163e-05 lr_backbone=7.885163e-05 elapsed=17:10:21 avg_step=3.65s recent_step=4.08s eta=33:29:01 +2026-09-15 06:57:39 | INFO | train rank=0 | step=16960/50000 progress=33.9% loss=0.111324 lr=7.882462e-05 lr_backbone=7.882462e-05 elapsed=17:11:12 avg_step=3.65s recent_step=5.14s eta=33:28:54 +2026-09-15 06:58:13 | INFO | train rank=0 | step=16970/50000 progress=33.9% loss=0.115419 lr=7.879759e-05 lr_backbone=7.879759e-05 elapsed=17:11:45 avg_step=3.65s recent_step=3.34s eta=33:28:12 +2026-09-15 06:58:46 | INFO | train rank=0 | step=16980/50000 progress=34.0% loss=0.067048 lr=7.877055e-05 lr_backbone=7.877055e-05 elapsed=17:12:19 avg_step=3.65s recent_step=3.31s eta=33:27:29 +2026-09-15 06:59:17 | INFO | train rank=0 | step=16990/50000 progress=34.0% loss=0.130930 lr=7.874350e-05 lr_backbone=7.874350e-05 elapsed=17:12:50 avg_step=3.65s recent_step=3.14s eta=33:26:42 +2026-09-15 07:00:04 | INFO | train rank=0 | step=17000/50000 progress=34.0% loss=0.094995 lr=7.871643e-05 lr_backbone=7.871643e-05 elapsed=17:13:37 avg_step=3.65s recent_step=4.69s eta=33:26:26 +2026-09-15 07:00:46 | INFO | train rank=0 | step=17010/50000 progress=34.0% loss=0.087054 lr=7.868936e-05 lr_backbone=7.868936e-05 elapsed=17:14:19 avg_step=3.65s recent_step=4.17s eta=33:26:00 +2026-09-15 07:01:15 | INFO | train rank=0 | step=17020/50000 progress=34.0% loss=0.100686 lr=7.866227e-05 lr_backbone=7.866227e-05 elapsed=17:14:47 avg_step=3.65s recent_step=2.87s eta=33:25:08 +2026-09-15 07:01:56 | INFO | train rank=0 | step=17030/50000 progress=34.1% loss=0.070875 lr=7.863516e-05 lr_backbone=7.863516e-05 elapsed=17:15:29 avg_step=3.65s recent_step=4.16s eta=33:24:42 +2026-09-15 07:02:39 | INFO | train rank=0 | step=17040/50000 progress=34.1% loss=0.119195 lr=7.860805e-05 lr_backbone=7.860805e-05 elapsed=17:16:12 avg_step=3.65s recent_step=4.28s eta=33:24:17 +2026-09-15 07:03:06 | INFO | train rank=0 | step=17050/50000 progress=34.1% loss=0.061623 lr=7.858092e-05 lr_backbone=7.858092e-05 elapsed=17:16:38 avg_step=3.65s recent_step=2.66s eta=33:23:22 +2026-09-15 07:03:49 | INFO | train rank=0 | step=17060/50000 progress=34.1% loss=0.115515 lr=7.855378e-05 lr_backbone=7.855378e-05 elapsed=17:17:22 avg_step=3.65s recent_step=4.35s eta=33:22:59 +2026-09-15 07:04:17 | INFO | train rank=0 | step=17070/50000 progress=34.1% loss=0.099781 lr=7.852663e-05 lr_backbone=7.852663e-05 elapsed=17:17:50 avg_step=3.65s recent_step=2.84s eta=33:22:07 +2026-09-15 07:05:03 | INFO | train rank=0 | step=17080/50000 progress=34.2% loss=0.110560 lr=7.849946e-05 lr_backbone=7.849946e-05 elapsed=17:18:36 avg_step=3.65s recent_step=4.60s eta=33:21:49 +2026-09-15 07:05:37 | INFO | train rank=0 | step=17090/50000 progress=34.2% loss=0.075788 lr=7.847228e-05 lr_backbone=7.847228e-05 elapsed=17:19:09 avg_step=3.65s recent_step=3.31s eta=33:21:06 +2026-09-15 07:06:13 | INFO | train rank=0 | step=17100/50000 progress=34.2% loss=0.118993 lr=7.844509e-05 lr_backbone=7.844509e-05 elapsed=17:19:46 avg_step=3.65s recent_step=3.68s eta=33:20:30 +2026-09-15 07:06:38 | INFO | train rank=0 | step=17110/50000 progress=34.2% loss=0.115434 lr=7.841789e-05 lr_backbone=7.841789e-05 elapsed=17:20:11 avg_step=3.65s recent_step=2.49s eta=33:19:31 +2026-09-15 07:07:26 | INFO | train rank=0 | step=17120/50000 progress=34.2% loss=0.114596 lr=7.839067e-05 lr_backbone=7.839067e-05 elapsed=17:20:59 avg_step=3.65s recent_step=4.76s eta=33:19:16 +2026-09-15 07:08:02 | INFO | train rank=0 | step=17130/50000 progress=34.3% loss=0.080673 lr=7.836345e-05 lr_backbone=7.836345e-05 elapsed=17:21:35 avg_step=3.65s recent_step=3.65s eta=33:18:39 +2026-09-15 07:08:35 | INFO | train rank=0 | step=17140/50000 progress=34.3% loss=0.118562 lr=7.833621e-05 lr_backbone=7.833621e-05 elapsed=17:22:07 avg_step=3.65s recent_step=3.24s eta=33:17:55 +2026-09-15 07:09:09 | INFO | train rank=0 | step=17150/50000 progress=34.3% loss=0.131704 lr=7.830896e-05 lr_backbone=7.830896e-05 elapsed=17:22:42 avg_step=3.65s recent_step=3.44s eta=33:17:15 +2026-09-15 07:09:55 | INFO | train rank=0 | step=17160/50000 progress=34.3% loss=0.118196 lr=7.828169e-05 lr_backbone=7.828169e-05 elapsed=17:23:27 avg_step=3.65s recent_step=4.56s eta=33:16:56 +2026-09-15 07:10:26 | INFO | train rank=0 | step=17170/50000 progress=34.3% loss=0.087048 lr=7.825441e-05 lr_backbone=7.825441e-05 elapsed=17:23:59 avg_step=3.65s recent_step=3.17s eta=33:16:10 +2026-09-15 07:10:54 | INFO | train rank=0 | step=17180/50000 progress=34.4% loss=0.095701 lr=7.822712e-05 lr_backbone=7.822712e-05 elapsed=17:24:27 avg_step=3.65s recent_step=2.79s eta=33:15:17 +2026-09-15 07:11:34 | INFO | train rank=0 | step=17190/50000 progress=34.4% loss=0.132578 lr=7.819982e-05 lr_backbone=7.819982e-05 elapsed=17:25:07 avg_step=3.65s recent_step=4.00s eta=33:14:47 +2026-09-15 07:12:17 | INFO | train rank=0 | step=17200/50000 progress=34.4% loss=0.107438 lr=7.817251e-05 lr_backbone=7.817251e-05 elapsed=17:25:50 avg_step=3.65s recent_step=4.25s eta=33:14:23 +2026-09-15 07:13:01 | INFO | train rank=0 | step=17210/50000 progress=34.4% loss=0.122079 lr=7.814518e-05 lr_backbone=7.814518e-05 elapsed=17:26:34 avg_step=3.65s recent_step=4.40s eta=33:14:00 +2026-09-15 07:13:31 | INFO | train rank=0 | step=17220/50000 progress=34.4% loss=0.104175 lr=7.811784e-05 lr_backbone=7.811784e-05 elapsed=17:27:04 avg_step=3.65s recent_step=3.03s eta=33:13:12 +2026-09-15 07:14:03 | INFO | train rank=0 | step=17230/50000 progress=34.5% loss=0.104924 lr=7.809049e-05 lr_backbone=7.809049e-05 elapsed=17:27:36 avg_step=3.65s recent_step=3.22s eta=33:12:27 +2026-09-15 07:14:49 | INFO | train rank=0 | step=17240/50000 progress=34.5% loss=0.114833 lr=7.806313e-05 lr_backbone=7.806313e-05 elapsed=17:28:21 avg_step=3.65s recent_step=4.53s eta=33:12:08 +2026-09-15 07:15:30 | INFO | train rank=0 | step=17250/50000 progress=34.5% loss=0.130146 lr=7.803575e-05 lr_backbone=7.803575e-05 elapsed=17:29:02 avg_step=3.65s recent_step=4.10s eta=33:11:40 +2026-09-15 07:16:05 | INFO | train rank=0 | step=17260/50000 progress=34.5% loss=0.104753 lr=7.800836e-05 lr_backbone=7.800836e-05 elapsed=17:29:38 avg_step=3.65s recent_step=3.53s eta=33:11:01 +2026-09-15 07:16:38 | INFO | train rank=0 | step=17270/50000 progress=34.5% loss=0.117110 lr=7.798096e-05 lr_backbone=7.798096e-05 elapsed=17:30:11 avg_step=3.65s recent_step=3.31s eta=33:10:18 +2026-09-15 07:17:24 | INFO | train rank=0 | step=17280/50000 progress=34.6% loss=0.136448 lr=7.795355e-05 lr_backbone=7.795355e-05 elapsed=17:30:57 avg_step=3.65s recent_step=4.57s eta=33:09:59 +2026-09-15 07:17:56 | INFO | train rank=0 | step=17290/50000 progress=34.6% loss=0.094212 lr=7.792613e-05 lr_backbone=7.792613e-05 elapsed=17:31:29 avg_step=3.65s recent_step=3.26s eta=33:09:15 +2026-09-15 07:18:33 | INFO | train rank=0 | step=17300/50000 progress=34.6% loss=0.123427 lr=7.789869e-05 lr_backbone=7.789869e-05 elapsed=17:32:06 avg_step=3.65s recent_step=3.69s eta=33:08:40 +2026-09-15 07:19:02 | INFO | train rank=0 | step=17310/50000 progress=34.6% loss=0.086718 lr=7.787124e-05 lr_backbone=7.787124e-05 elapsed=17:32:35 avg_step=3.65s recent_step=2.87s eta=33:07:48 +2026-09-15 07:19:51 | INFO | train rank=0 | step=17320/50000 progress=34.6% loss=0.106253 lr=7.784378e-05 lr_backbone=7.784378e-05 elapsed=17:33:23 avg_step=3.65s recent_step=4.87s eta=33:07:35 +2026-09-15 07:20:17 | INFO | train rank=0 | step=17330/50000 progress=34.7% loss=0.098155 lr=7.781631e-05 lr_backbone=7.781631e-05 elapsed=17:33:49 avg_step=3.65s recent_step=2.62s eta=33:06:39 +2026-09-15 07:20:53 | INFO | train rank=0 | step=17340/50000 progress=34.7% loss=0.100425 lr=7.778882e-05 lr_backbone=7.778882e-05 elapsed=17:34:26 avg_step=3.65s recent_step=3.64s eta=33:06:02 +2026-09-15 07:21:34 | INFO | train rank=0 | step=17350/50000 progress=34.7% loss=0.166458 lr=7.776132e-05 lr_backbone=7.776132e-05 elapsed=17:35:07 avg_step=3.65s recent_step=4.12s eta=33:05:35 +2026-09-15 07:22:28 | INFO | train rank=0 | step=17360/50000 progress=34.7% loss=0.107698 lr=7.773381e-05 lr_backbone=7.773381e-05 elapsed=17:36:01 avg_step=3.65s recent_step=5.38s eta=33:05:31 +2026-09-15 07:23:02 | INFO | train rank=0 | step=17370/50000 progress=34.7% loss=0.120531 lr=7.770629e-05 lr_backbone=7.770629e-05 elapsed=17:36:34 avg_step=3.65s recent_step=3.34s eta=33:04:48 +2026-09-15 07:23:35 | INFO | train rank=0 | step=17380/50000 progress=34.8% loss=0.101538 lr=7.767876e-05 lr_backbone=7.767876e-05 elapsed=17:37:08 avg_step=3.65s recent_step=3.39s eta=33:04:07 +2026-09-15 07:24:06 | INFO | train rank=0 | step=17390/50000 progress=34.8% loss=0.124987 lr=7.765121e-05 lr_backbone=7.765121e-05 elapsed=17:37:38 avg_step=3.65s recent_step=3.01s eta=33:03:18 +2026-09-15 07:24:55 | INFO | train rank=0 | step=17400/50000 progress=34.8% loss=0.083651 lr=7.762365e-05 lr_backbone=7.762365e-05 elapsed=17:38:27 avg_step=3.65s recent_step=4.90s eta=33:03:05 +2026-09-15 07:25:26 | INFO | train rank=0 | step=17410/50000 progress=34.8% loss=0.112005 lr=7.759608e-05 lr_backbone=7.759608e-05 elapsed=17:38:59 avg_step=3.65s recent_step=3.17s eta=33:02:20 +2026-09-15 07:26:03 | INFO | train rank=0 | step=17420/50000 progress=34.8% loss=0.122846 lr=7.756850e-05 lr_backbone=7.756850e-05 elapsed=17:39:36 avg_step=3.65s recent_step=3.67s eta=33:01:44 +2026-09-15 07:26:31 | INFO | train rank=0 | step=17430/50000 progress=34.9% loss=0.075057 lr=7.754091e-05 lr_backbone=7.754091e-05 elapsed=17:40:04 avg_step=3.65s recent_step=2.84s eta=33:00:52 +2026-09-15 07:27:29 | INFO | train rank=0 | step=17440/50000 progress=34.9% loss=0.097720 lr=7.751330e-05 lr_backbone=7.751330e-05 elapsed=17:41:02 avg_step=3.65s recent_step=5.77s eta=33:00:55 +2026-09-15 07:27:59 | INFO | train rank=0 | step=17450/50000 progress=34.9% loss=0.111791 lr=7.748568e-05 lr_backbone=7.748568e-05 elapsed=17:41:32 avg_step=3.65s recent_step=3.03s eta=33:00:07 +2026-09-15 07:28:31 | INFO | train rank=0 | step=17460/50000 progress=34.9% loss=0.084955 lr=7.745805e-05 lr_backbone=7.745805e-05 elapsed=17:42:04 avg_step=3.65s recent_step=3.15s eta=32:59:21 +2026-09-15 07:29:04 | INFO | train rank=0 | step=17470/50000 progress=34.9% loss=0.119209 lr=7.743041e-05 lr_backbone=7.743041e-05 elapsed=17:42:36 avg_step=3.65s recent_step=3.28s eta=32:58:38 +2026-09-15 07:30:01 | INFO | train rank=0 | step=17480/50000 progress=35.0% loss=0.107852 lr=7.740275e-05 lr_backbone=7.740275e-05 elapsed=17:43:33 avg_step=3.65s recent_step=5.71s eta=32:58:40 +2026-09-15 07:30:29 | INFO | train rank=0 | step=17490/50000 progress=35.0% loss=0.110346 lr=7.737509e-05 lr_backbone=7.737509e-05 elapsed=17:44:01 avg_step=3.65s recent_step=2.80s eta=32:57:47 +2026-09-15 07:31:07 | INFO | train rank=0 | step=17500/50000 progress=35.0% loss=0.101910 lr=7.734741e-05 lr_backbone=7.734741e-05 elapsed=17:44:40 avg_step=3.65s recent_step=3.81s eta=32:57:14 +2026-09-15 07:31:41 | INFO | train rank=0 | step=17510/50000 progress=35.0% loss=0.112235 lr=7.731972e-05 lr_backbone=7.731972e-05 elapsed=17:45:14 avg_step=3.65s recent_step=3.42s eta=32:56:33 +2026-09-15 07:32:30 | INFO | train rank=0 | step=17520/50000 progress=35.0% loss=0.097147 lr=7.729201e-05 lr_backbone=7.729201e-05 elapsed=17:46:03 avg_step=3.65s recent_step=4.94s eta=32:56:20 +2026-09-15 07:33:01 | INFO | train rank=0 | step=17530/50000 progress=35.1% loss=0.087395 lr=7.726430e-05 lr_backbone=7.726430e-05 elapsed=17:46:34 avg_step=3.65s recent_step=3.07s eta=32:55:33 +2026-09-15 07:33:35 | INFO | train rank=0 | step=17540/50000 progress=35.1% loss=0.133854 lr=7.723657e-05 lr_backbone=7.723657e-05 elapsed=17:47:07 avg_step=3.65s recent_step=3.36s eta=32:54:51 +2026-09-15 07:34:16 | INFO | train rank=0 | step=17550/50000 progress=35.1% loss=0.081668 lr=7.720884e-05 lr_backbone=7.720884e-05 elapsed=17:47:49 avg_step=3.65s recent_step=4.11s eta=32:54:23 +2026-09-15 07:35:00 | INFO | train rank=0 | step=17560/50000 progress=35.1% loss=0.141201 lr=7.718109e-05 lr_backbone=7.718109e-05 elapsed=17:48:33 avg_step=3.65s recent_step=4.40s eta=32:54:01 +2026-09-15 07:35:28 | INFO | train rank=0 | step=17570/50000 progress=35.1% loss=0.078859 lr=7.715332e-05 lr_backbone=7.715332e-05 elapsed=17:49:01 avg_step=3.65s recent_step=2.82s eta=32:53:09 +2026-09-15 07:36:01 | INFO | train rank=0 | step=17580/50000 progress=35.2% loss=0.105385 lr=7.712555e-05 lr_backbone=7.712555e-05 elapsed=17:49:34 avg_step=3.65s recent_step=3.29s eta=32:52:26 +2026-09-15 07:36:29 | INFO | train rank=0 | step=17590/50000 progress=35.2% loss=0.091444 lr=7.709776e-05 lr_backbone=7.709776e-05 elapsed=17:50:02 avg_step=3.65s recent_step=2.82s eta=32:51:34 +2026-09-15 07:37:11 | INFO | train rank=0 | step=17600/50000 progress=35.2% loss=0.100274 lr=7.706997e-05 lr_backbone=7.706997e-05 elapsed=17:50:43 avg_step=3.65s recent_step=4.16s eta=32:51:07 +2026-09-15 07:37:41 | INFO | train rank=0 | step=17610/50000 progress=35.2% loss=0.115264 lr=7.704216e-05 lr_backbone=7.704216e-05 elapsed=17:51:14 avg_step=3.65s recent_step=3.02s eta=32:50:19 +2026-09-15 07:38:12 | INFO | train rank=0 | step=17620/50000 progress=35.2% loss=0.097805 lr=7.701434e-05 lr_backbone=7.701434e-05 elapsed=17:51:44 avg_step=3.65s recent_step=3.07s eta=32:49:31 +2026-09-15 07:38:47 | INFO | train rank=0 | step=17630/50000 progress=35.3% loss=0.146243 lr=7.698650e-05 lr_backbone=7.698650e-05 elapsed=17:52:19 avg_step=3.65s recent_step=3.51s eta=32:48:52 +2026-09-15 07:39:33 | INFO | train rank=0 | step=17640/50000 progress=35.3% loss=0.140062 lr=7.695866e-05 lr_backbone=7.695866e-05 elapsed=17:53:06 avg_step=3.65s recent_step=4.64s eta=32:48:34 +2026-09-15 07:40:11 | INFO | train rank=0 | step=17650/50000 progress=35.3% loss=0.108351 lr=7.693080e-05 lr_backbone=7.693080e-05 elapsed=17:53:43 avg_step=3.65s recent_step=3.75s eta=32:48:00 +2026-09-15 07:40:52 | INFO | train rank=0 | step=17660/50000 progress=35.3% loss=0.089426 lr=7.690293e-05 lr_backbone=7.690293e-05 elapsed=17:54:25 avg_step=3.65s recent_step=4.14s eta=32:47:32 +2026-09-15 07:41:23 | INFO | train rank=0 | step=17670/50000 progress=35.3% loss=0.100194 lr=7.687505e-05 lr_backbone=7.687505e-05 elapsed=17:54:55 avg_step=3.65s recent_step=3.07s eta=32:46:45 +2026-09-15 07:42:17 | INFO | train rank=0 | step=17680/50000 progress=35.4% loss=0.086882 lr=7.684716e-05 lr_backbone=7.684716e-05 elapsed=17:55:50 avg_step=3.65s recent_step=5.41s eta=32:46:41 +2026-09-15 07:42:51 | INFO | train rank=0 | step=17690/50000 progress=35.4% loss=0.093423 lr=7.681926e-05 lr_backbone=7.681926e-05 elapsed=17:56:24 avg_step=3.65s recent_step=3.41s eta=32:46:00 +2026-09-15 07:43:27 | INFO | train rank=0 | step=17700/50000 progress=35.4% loss=0.102259 lr=7.679134e-05 lr_backbone=7.679134e-05 elapsed=17:57:00 avg_step=3.65s recent_step=3.62s eta=32:45:23 +2026-09-15 07:44:05 | INFO | train rank=0 | step=17710/50000 progress=35.4% loss=0.135430 lr=7.676341e-05 lr_backbone=7.676341e-05 elapsed=17:57:38 avg_step=3.65s recent_step=3.84s eta=32:44:49 +2026-09-15 07:44:48 | INFO | train rank=0 | step=17720/50000 progress=35.4% loss=0.110893 lr=7.673547e-05 lr_backbone=7.673547e-05 elapsed=17:58:21 avg_step=3.65s recent_step=4.28s eta=32:44:24 +2026-09-15 07:45:24 | INFO | train rank=0 | step=17730/50000 progress=35.5% loss=0.120633 lr=7.670752e-05 lr_backbone=7.670752e-05 elapsed=17:58:57 avg_step=3.65s recent_step=3.59s eta=32:43:47 +2026-09-15 07:45:55 | INFO | train rank=0 | step=17740/50000 progress=35.5% loss=0.091405 lr=7.667956e-05 lr_backbone=7.667956e-05 elapsed=17:59:27 avg_step=3.65s recent_step=3.05s eta=32:42:59 +2026-09-15 07:46:31 | INFO | train rank=0 | step=17750/50000 progress=35.5% loss=0.102673 lr=7.665159e-05 lr_backbone=7.665159e-05 elapsed=18:00:04 avg_step=3.65s recent_step=3.68s eta=32:42:23 +2026-09-15 07:47:20 | INFO | train rank=0 | step=17760/50000 progress=35.5% loss=0.159878 lr=7.662360e-05 lr_backbone=7.662360e-05 elapsed=18:00:52 avg_step=3.65s recent_step=4.83s eta=32:42:08 +2026-09-15 07:47:45 | INFO | train rank=0 | step=17770/50000 progress=35.5% loss=0.090655 lr=7.659560e-05 lr_backbone=7.659560e-05 elapsed=18:01:18 avg_step=3.65s recent_step=2.55s eta=32:41:12 +2026-09-15 07:48:21 | INFO | train rank=0 | step=17780/50000 progress=35.6% loss=0.122744 lr=7.656759e-05 lr_backbone=7.656759e-05 elapsed=18:01:54 avg_step=3.65s recent_step=3.60s eta=32:40:34 +2026-09-15 07:48:57 | INFO | train rank=0 | step=17790/50000 progress=35.6% loss=0.077042 lr=7.653957e-05 lr_backbone=7.653957e-05 elapsed=18:02:30 avg_step=3.65s recent_step=3.58s eta=32:39:56 +2026-09-15 07:49:36 | INFO | train rank=0 | step=17800/50000 progress=35.6% loss=0.137178 lr=7.651154e-05 lr_backbone=7.651154e-05 elapsed=18:03:09 avg_step=3.65s recent_step=3.88s eta=32:39:24 +2026-09-15 07:50:14 | INFO | train rank=0 | step=17810/50000 progress=35.6% loss=0.104944 lr=7.648350e-05 lr_backbone=7.648350e-05 elapsed=18:03:47 avg_step=3.65s recent_step=3.84s eta=32:38:51 +2026-09-15 07:50:53 | INFO | train rank=0 | step=17820/50000 progress=35.6% loss=0.113133 lr=7.645544e-05 lr_backbone=7.645544e-05 elapsed=18:04:26 avg_step=3.65s recent_step=3.91s eta=32:38:19 +2026-09-15 07:51:28 | INFO | train rank=0 | step=17830/50000 progress=35.7% loss=0.117089 lr=7.642738e-05 lr_backbone=7.642738e-05 elapsed=18:05:01 avg_step=3.65s recent_step=3.47s eta=32:37:39 +2026-09-15 07:52:14 | INFO | train rank=0 | step=17840/50000 progress=35.7% loss=0.115991 lr=7.639930e-05 lr_backbone=7.639930e-05 elapsed=18:05:47 avg_step=3.65s recent_step=4.64s eta=32:37:20 +2026-09-15 07:52:46 | INFO | train rank=0 | step=17850/50000 progress=35.7% loss=0.098113 lr=7.637121e-05 lr_backbone=7.637121e-05 elapsed=18:06:18 avg_step=3.65s recent_step=3.14s eta=32:36:35 +2026-09-15 07:53:25 | INFO | train rank=0 | step=17860/50000 progress=35.7% loss=0.094390 lr=7.634311e-05 lr_backbone=7.634311e-05 elapsed=18:06:58 avg_step=3.65s recent_step=3.93s eta=32:36:03 +2026-09-15 07:54:01 | INFO | train rank=0 | step=17870/50000 progress=35.7% loss=0.117738 lr=7.631499e-05 lr_backbone=7.631499e-05 elapsed=18:07:34 avg_step=3.65s recent_step=3.62s eta=32:35:26 +2026-09-15 07:54:46 | INFO | train rank=0 | step=17880/50000 progress=35.8% loss=0.076827 lr=7.628687e-05 lr_backbone=7.628687e-05 elapsed=18:08:19 avg_step=3.65s recent_step=4.46s eta=32:35:04 +2026-09-15 07:55:12 | INFO | train rank=0 | step=17890/50000 progress=35.8% loss=0.104324 lr=7.625873e-05 lr_backbone=7.625873e-05 elapsed=18:08:45 avg_step=3.65s recent_step=2.64s eta=32:34:09 +2026-09-15 07:55:53 | INFO | train rank=0 | step=17900/50000 progress=35.8% loss=0.101942 lr=7.623058e-05 lr_backbone=7.623058e-05 elapsed=18:09:25 avg_step=3.65s recent_step=4.04s eta=32:33:40 +2026-09-15 07:56:26 | INFO | train rank=0 | step=17910/50000 progress=35.8% loss=0.088522 lr=7.620242e-05 lr_backbone=7.620242e-05 elapsed=18:09:59 avg_step=3.65s recent_step=3.35s eta=32:32:58 +2026-09-15 07:57:13 | INFO | train rank=0 | step=17920/50000 progress=35.8% loss=0.104292 lr=7.617425e-05 lr_backbone=7.617425e-05 elapsed=18:10:45 avg_step=3.65s recent_step=4.64s eta=32:32:39 +2026-09-15 07:57:48 | INFO | train rank=0 | step=17930/50000 progress=35.9% loss=0.113131 lr=7.614607e-05 lr_backbone=7.614607e-05 elapsed=18:11:21 avg_step=3.65s recent_step=3.54s eta=32:32:01 +2026-09-15 07:58:25 | INFO | train rank=0 | step=17940/50000 progress=35.9% loss=0.089794 lr=7.611788e-05 lr_backbone=7.611788e-05 elapsed=18:11:57 avg_step=3.65s recent_step=3.66s eta=32:31:24 +2026-09-15 07:59:04 | INFO | train rank=0 | step=17950/50000 progress=35.9% loss=0.119331 lr=7.608967e-05 lr_backbone=7.608967e-05 elapsed=18:12:37 avg_step=3.65s recent_step=3.98s eta=32:30:54 +2026-09-15 07:59:45 | INFO | train rank=0 | step=17960/50000 progress=35.9% loss=0.136668 lr=7.606146e-05 lr_backbone=7.606146e-05 elapsed=18:13:18 avg_step=3.65s recent_step=4.07s eta=32:30:25 +2026-09-15 08:00:21 | INFO | train rank=0 | step=17970/50000 progress=35.9% loss=0.103100 lr=7.603323e-05 lr_backbone=7.603323e-05 elapsed=18:13:54 avg_step=3.65s recent_step=3.61s eta=32:29:47 +2026-09-15 08:00:47 | INFO | train rank=0 | step=17980/50000 progress=36.0% loss=0.090295 lr=7.600499e-05 lr_backbone=7.600499e-05 elapsed=18:14:20 avg_step=3.65s recent_step=2.58s eta=32:28:52 +2026-09-15 08:01:25 | INFO | train rank=0 | step=17990/50000 progress=36.0% loss=0.081531 lr=7.597674e-05 lr_backbone=7.597674e-05 elapsed=18:14:58 avg_step=3.65s recent_step=3.84s eta=32:28:18 +2026-09-15 08:02:11 | INFO | train rank=0 | step=18000/50000 progress=36.0% loss=0.082060 lr=7.594848e-05 lr_backbone=7.594848e-05 elapsed=18:15:44 avg_step=3.65s recent_step=4.58s eta=32:27:58 +2026-09-15 08:02:46 | INFO | train rank=0 | step=18010/50000 progress=36.0% loss=0.121825 lr=7.592021e-05 lr_backbone=7.592021e-05 elapsed=18:16:19 avg_step=3.65s recent_step=3.46s eta=32:27:19 +2026-09-15 08:03:19 | INFO | train rank=0 | step=18020/50000 progress=36.0% loss=0.151489 lr=7.589192e-05 lr_backbone=7.589192e-05 elapsed=18:16:52 avg_step=3.65s recent_step=3.34s eta=32:26:36 +2026-09-15 08:03:51 | INFO | train rank=0 | step=18030/50000 progress=36.1% loss=0.106474 lr=7.586363e-05 lr_backbone=7.586363e-05 elapsed=18:17:24 avg_step=3.65s recent_step=3.18s eta=32:25:52 +2026-09-15 08:04:33 | INFO | train rank=0 | step=18040/50000 progress=36.1% loss=0.101998 lr=7.583532e-05 lr_backbone=7.583532e-05 elapsed=18:18:06 avg_step=3.65s recent_step=4.26s eta=32:25:26 +2026-09-15 08:05:07 | INFO | train rank=0 | step=18050/50000 progress=36.1% loss=0.114970 lr=7.580700e-05 lr_backbone=7.580700e-05 elapsed=18:18:40 avg_step=3.65s recent_step=3.33s eta=32:24:43 +2026-09-15 08:05:37 | INFO | train rank=0 | step=18060/50000 progress=36.1% loss=0.083372 lr=7.577867e-05 lr_backbone=7.577867e-05 elapsed=18:19:10 avg_step=3.65s recent_step=3.04s eta=32:23:56 +2026-09-15 08:06:14 | INFO | train rank=0 | step=18070/50000 progress=36.1% loss=0.080969 lr=7.575033e-05 lr_backbone=7.575033e-05 elapsed=18:19:47 avg_step=3.65s recent_step=3.66s eta=32:23:20 +2026-09-15 08:07:03 | INFO | train rank=0 | step=18080/50000 progress=36.2% loss=0.098393 lr=7.572198e-05 lr_backbone=7.572198e-05 elapsed=18:20:36 avg_step=3.65s recent_step=4.95s eta=32:23:06 +2026-09-15 08:07:37 | INFO | train rank=0 | step=18090/50000 progress=36.2% loss=0.129288 lr=7.569361e-05 lr_backbone=7.569361e-05 elapsed=18:21:09 avg_step=3.65s recent_step=3.33s eta=32:22:24 +2026-09-15 08:08:11 | INFO | train rank=0 | step=18100/50000 progress=36.2% loss=0.106981 lr=7.566524e-05 lr_backbone=7.566524e-05 elapsed=18:21:44 avg_step=3.65s recent_step=3.41s eta=32:21:43 +2026-09-15 08:08:43 | INFO | train rank=0 | step=18110/50000 progress=36.2% loss=0.090237 lr=7.563685e-05 lr_backbone=7.563685e-05 elapsed=18:22:16 avg_step=3.65s recent_step=3.26s eta=32:21:00 +2026-09-15 08:09:17 | INFO | train rank=0 | step=18120/50000 progress=36.2% loss=0.096133 lr=7.560846e-05 lr_backbone=7.560846e-05 elapsed=18:22:50 avg_step=3.65s recent_step=3.39s eta=32:20:19 +2026-09-15 08:09:52 | INFO | train rank=0 | step=18130/50000 progress=36.3% loss=0.097669 lr=7.558005e-05 lr_backbone=7.558005e-05 elapsed=18:23:25 avg_step=3.65s recent_step=3.52s eta=32:19:40 +2026-09-15 08:10:23 | INFO | train rank=0 | step=18140/50000 progress=36.3% loss=0.091829 lr=7.555163e-05 lr_backbone=7.555163e-05 elapsed=18:23:56 avg_step=3.65s recent_step=3.03s eta=32:18:53 +2026-09-15 08:10:57 | INFO | train rank=0 | step=18150/50000 progress=36.3% loss=0.141876 lr=7.552320e-05 lr_backbone=7.552320e-05 elapsed=18:24:30 avg_step=3.65s recent_step=3.42s eta=32:18:12 +2026-09-15 08:11:50 | INFO | train rank=0 | step=18160/50000 progress=36.3% loss=0.123575 lr=7.549475e-05 lr_backbone=7.549475e-05 elapsed=18:25:23 avg_step=3.65s recent_step=5.33s eta=32:18:05 +2026-09-15 08:12:22 | INFO | train rank=0 | step=18170/50000 progress=36.3% loss=0.101177 lr=7.546630e-05 lr_backbone=7.546630e-05 elapsed=18:25:55 avg_step=3.65s recent_step=3.21s eta=32:17:21 +2026-09-15 08:12:55 | INFO | train rank=0 | step=18180/50000 progress=36.4% loss=0.083256 lr=7.543784e-05 lr_backbone=7.543784e-05 elapsed=18:26:28 avg_step=3.65s recent_step=3.30s eta=32:16:38 +2026-09-15 08:13:31 | INFO | train rank=0 | step=18190/50000 progress=36.4% loss=0.105569 lr=7.540936e-05 lr_backbone=7.540936e-05 elapsed=18:27:04 avg_step=3.65s recent_step=3.56s eta=32:16:00 +2026-09-15 08:14:14 | INFO | train rank=0 | step=18200/50000 progress=36.4% loss=0.139713 lr=7.538088e-05 lr_backbone=7.538088e-05 elapsed=18:27:47 avg_step=3.65s recent_step=4.30s eta=32:15:35 +2026-09-15 08:14:48 | INFO | train rank=0 | step=18210/50000 progress=36.4% loss=0.114877 lr=7.535238e-05 lr_backbone=7.535238e-05 elapsed=18:28:21 avg_step=3.65s recent_step=3.44s eta=32:14:54 +2026-09-15 08:15:21 | INFO | train rank=0 | step=18220/50000 progress=36.4% loss=0.108038 lr=7.532387e-05 lr_backbone=7.532387e-05 elapsed=18:28:53 avg_step=3.65s recent_step=3.21s eta=32:14:10 +2026-09-15 08:15:56 | INFO | train rank=0 | step=18230/50000 progress=36.5% loss=0.117134 lr=7.529535e-05 lr_backbone=7.529535e-05 elapsed=18:29:29 avg_step=3.65s recent_step=3.55s eta=32:13:32 +2026-09-15 08:16:41 | INFO | train rank=0 | step=18240/50000 progress=36.5% loss=0.144568 lr=7.526682e-05 lr_backbone=7.526682e-05 elapsed=18:30:14 avg_step=3.65s recent_step=4.52s eta=32:13:11 +2026-09-15 08:17:23 | INFO | train rank=0 | step=18250/50000 progress=36.5% loss=0.114290 lr=7.523828e-05 lr_backbone=7.523828e-05 elapsed=18:30:56 avg_step=3.65s recent_step=4.17s eta=32:12:43 +2026-09-15 08:17:51 | INFO | train rank=0 | step=18260/50000 progress=36.5% loss=0.098056 lr=7.520972e-05 lr_backbone=7.520972e-05 elapsed=18:31:24 avg_step=3.65s recent_step=2.78s eta=32:11:51 +2026-09-15 08:18:23 | INFO | train rank=0 | step=18270/50000 progress=36.5% loss=0.114813 lr=7.518116e-05 lr_backbone=7.518116e-05 elapsed=18:31:56 avg_step=3.65s recent_step=3.22s eta=32:11:07 +2026-09-15 08:19:25 | INFO | train rank=0 | step=18280/50000 progress=36.6% loss=0.113484 lr=7.515259e-05 lr_backbone=7.515259e-05 elapsed=18:32:58 avg_step=3.65s recent_step=6.19s eta=32:11:15 +2026-09-15 08:19:53 | INFO | train rank=0 | step=18290/50000 progress=36.6% loss=0.093381 lr=7.512400e-05 lr_backbone=7.512400e-05 elapsed=18:33:26 avg_step=3.65s recent_step=2.80s eta=32:10:24 +2026-09-15 08:20:32 | INFO | train rank=0 | step=18300/50000 progress=36.6% loss=0.109705 lr=7.509540e-05 lr_backbone=7.509540e-05 elapsed=18:34:05 avg_step=3.65s recent_step=3.95s eta=32:09:52 +2026-09-15 08:21:04 | INFO | train rank=0 | step=18310/50000 progress=36.6% loss=0.102198 lr=7.506679e-05 lr_backbone=7.506679e-05 elapsed=18:34:36 avg_step=3.65s recent_step=3.13s eta=32:09:07 +2026-09-15 08:21:41 | INFO | train rank=0 | step=18320/50000 progress=36.6% loss=0.122434 lr=7.503818e-05 lr_backbone=7.503818e-05 elapsed=18:35:14 avg_step=3.65s recent_step=3.78s eta=32:08:32 +2026-09-15 08:22:13 | INFO | train rank=0 | step=18330/50000 progress=36.7% loss=0.105692 lr=7.500955e-05 lr_backbone=7.500955e-05 elapsed=18:35:46 avg_step=3.65s recent_step=3.19s eta=32:07:48 +2026-09-15 08:22:56 | INFO | train rank=0 | step=18340/50000 progress=36.7% loss=0.099788 lr=7.498090e-05 lr_backbone=7.498090e-05 elapsed=18:36:29 avg_step=3.65s recent_step=4.27s eta=32:07:22 +2026-09-15 08:23:35 | INFO | train rank=0 | step=18350/50000 progress=36.7% loss=0.110601 lr=7.495225e-05 lr_backbone=7.495225e-05 elapsed=18:37:08 avg_step=3.65s recent_step=3.90s eta=32:06:50 +2026-09-15 08:24:23 | INFO | train rank=0 | step=18360/50000 progress=36.7% loss=0.124343 lr=7.492359e-05 lr_backbone=7.492359e-05 elapsed=18:37:56 avg_step=3.65s recent_step=4.83s eta=32:06:33 +2026-09-15 08:24:51 | INFO | train rank=0 | step=18370/50000 progress=36.7% loss=0.090146 lr=7.489492e-05 lr_backbone=7.489492e-05 elapsed=18:38:23 avg_step=3.65s recent_step=2.73s eta=32:05:41 +2026-09-15 08:25:30 | INFO | train rank=0 | step=18380/50000 progress=36.8% loss=0.145941 lr=7.486623e-05 lr_backbone=7.486623e-05 elapsed=18:39:03 avg_step=3.65s recent_step=3.94s eta=32:05:09 +2026-09-15 08:26:06 | INFO | train rank=0 | step=18390/50000 progress=36.8% loss=0.123110 lr=7.483754e-05 lr_backbone=7.483754e-05 elapsed=18:39:39 avg_step=3.65s recent_step=3.60s eta=32:04:32 +2026-09-15 08:26:54 | INFO | train rank=0 | step=18400/50000 progress=36.8% loss=0.108298 lr=7.480883e-05 lr_backbone=7.480883e-05 elapsed=18:40:27 avg_step=3.65s recent_step=4.82s eta=32:04:15 +2026-09-15 08:27:27 | INFO | train rank=0 | step=18410/50000 progress=36.8% loss=0.111574 lr=7.478011e-05 lr_backbone=7.478011e-05 elapsed=18:40:59 avg_step=3.65s recent_step=3.24s eta=32:03:32 +2026-09-15 08:27:58 | INFO | train rank=0 | step=18420/50000 progress=36.8% loss=0.102296 lr=7.475139e-05 lr_backbone=7.475139e-05 elapsed=18:41:31 avg_step=3.65s recent_step=3.14s eta=32:02:46 +2026-09-15 08:28:26 | INFO | train rank=0 | step=18430/50000 progress=36.9% loss=0.088184 lr=7.472265e-05 lr_backbone=7.472265e-05 elapsed=18:41:59 avg_step=3.65s recent_step=2.81s eta=32:01:55 +2026-09-15 08:29:16 | INFO | train rank=0 | step=18440/50000 progress=36.9% loss=0.113132 lr=7.469390e-05 lr_backbone=7.469390e-05 elapsed=18:42:49 avg_step=3.65s recent_step=4.98s eta=32:01:42 +2026-09-15 08:29:53 | INFO | train rank=0 | step=18450/50000 progress=36.9% loss=0.116120 lr=7.466514e-05 lr_backbone=7.466514e-05 elapsed=18:43:26 avg_step=3.65s recent_step=3.70s eta=32:01:06 +2026-09-15 08:30:29 | INFO | train rank=0 | step=18460/50000 progress=36.9% loss=0.092932 lr=7.463637e-05 lr_backbone=7.463637e-05 elapsed=18:44:02 avg_step=3.65s recent_step=3.62s eta=32:00:29 +2026-09-15 08:31:08 | INFO | train rank=0 | step=18470/50000 progress=36.9% loss=0.150531 lr=7.460759e-05 lr_backbone=7.460759e-05 elapsed=18:44:41 avg_step=3.65s recent_step=3.89s eta=31:59:56 +2026-09-15 08:31:42 | INFO | train rank=0 | step=18480/50000 progress=37.0% loss=0.097628 lr=7.457879e-05 lr_backbone=7.457879e-05 elapsed=18:45:15 avg_step=3.65s recent_step=3.46s eta=31:59:16 +2026-09-15 08:32:23 | INFO | train rank=0 | step=18490/50000 progress=37.0% loss=0.104314 lr=7.454999e-05 lr_backbone=7.454999e-05 elapsed=18:45:56 avg_step=3.65s recent_step=4.09s eta=31:58:47 +2026-09-15 08:32:55 | INFO | train rank=0 | step=18500/50000 progress=37.0% loss=0.110826 lr=7.452118e-05 lr_backbone=7.452118e-05 elapsed=18:46:28 avg_step=3.65s recent_step=3.19s eta=31:58:03 +2026-09-15 08:33:30 | INFO | train rank=0 | step=18510/50000 progress=37.0% loss=0.101018 lr=7.449235e-05 lr_backbone=7.449235e-05 elapsed=18:47:03 avg_step=3.65s recent_step=3.49s eta=31:57:23 +2026-09-15 08:34:17 | INFO | train rank=0 | step=18520/50000 progress=37.0% loss=0.114421 lr=7.446351e-05 lr_backbone=7.446351e-05 elapsed=18:47:49 avg_step=3.65s recent_step=4.63s eta=31:57:04 +2026-09-15 08:34:54 | INFO | train rank=0 | step=18530/50000 progress=37.1% loss=0.110076 lr=7.443467e-05 lr_backbone=7.443467e-05 elapsed=18:48:27 avg_step=3.65s recent_step=3.79s eta=31:56:29 +2026-09-15 08:35:28 | INFO | train rank=0 | step=18540/50000 progress=37.1% loss=0.101633 lr=7.440581e-05 lr_backbone=7.440581e-05 elapsed=18:49:01 avg_step=3.65s recent_step=3.39s eta=31:55:48 +2026-09-15 08:36:03 | INFO | train rank=0 | step=18550/50000 progress=37.1% loss=0.100821 lr=7.437694e-05 lr_backbone=7.437694e-05 elapsed=18:49:36 avg_step=3.65s recent_step=3.52s eta=31:55:09 +2026-09-15 08:36:46 | INFO | train rank=0 | step=18560/50000 progress=37.1% loss=0.116118 lr=7.434807e-05 lr_backbone=7.434807e-05 elapsed=18:50:19 avg_step=3.65s recent_step=4.28s eta=31:54:43 +2026-09-15 08:37:18 | INFO | train rank=0 | step=18570/50000 progress=37.1% loss=0.084523 lr=7.431918e-05 lr_backbone=7.431918e-05 elapsed=18:50:51 avg_step=3.65s recent_step=3.15s eta=31:53:58 +2026-09-15 08:37:51 | INFO | train rank=0 | step=18580/50000 progress=37.2% loss=0.104318 lr=7.429028e-05 lr_backbone=7.429028e-05 elapsed=18:51:24 avg_step=3.65s recent_step=3.31s eta=31:53:16 +2026-09-15 08:38:23 | INFO | train rank=0 | step=18590/50000 progress=37.2% loss=0.133655 lr=7.426137e-05 lr_backbone=7.426137e-05 elapsed=18:51:55 avg_step=3.65s recent_step=3.18s eta=31:52:31 +2026-09-15 08:39:19 | INFO | train rank=0 | step=18600/50000 progress=37.2% loss=0.093340 lr=7.423245e-05 lr_backbone=7.423245e-05 elapsed=18:52:52 avg_step=3.65s recent_step=5.63s eta=31:52:28 +2026-09-15 08:39:45 | INFO | train rank=0 | step=18610/50000 progress=37.2% loss=0.091768 lr=7.420352e-05 lr_backbone=7.420352e-05 elapsed=18:53:18 avg_step=3.65s recent_step=2.66s eta=31:51:35 +2026-09-15 08:40:21 | INFO | train rank=0 | step=18620/50000 progress=37.2% loss=0.097625 lr=7.417457e-05 lr_backbone=7.417457e-05 elapsed=18:53:54 avg_step=3.65s recent_step=3.55s eta=31:50:57 +2026-09-15 08:40:55 | INFO | train rank=0 | step=18630/50000 progress=37.3% loss=0.104223 lr=7.414562e-05 lr_backbone=7.414562e-05 elapsed=18:54:28 avg_step=3.65s recent_step=3.42s eta=31:50:16 +2026-09-15 08:41:43 | INFO | train rank=0 | step=18640/50000 progress=37.3% loss=0.115662 lr=7.411666e-05 lr_backbone=7.411666e-05 elapsed=18:55:16 avg_step=3.65s recent_step=4.80s eta=31:49:59 +2026-09-15 08:42:09 | INFO | train rank=0 | step=18650/50000 progress=37.3% loss=0.089950 lr=7.408768e-05 lr_backbone=7.408768e-05 elapsed=18:55:42 avg_step=3.65s recent_step=2.55s eta=31:49:04 +2026-09-15 08:42:42 | INFO | train rank=0 | step=18660/50000 progress=37.3% loss=0.090132 lr=7.405870e-05 lr_backbone=7.405870e-05 elapsed=18:56:14 avg_step=3.65s recent_step=3.29s eta=31:48:21 +2026-09-15 08:43:08 | INFO | train rank=0 | step=18670/50000 progress=37.3% loss=0.108577 lr=7.402970e-05 lr_backbone=7.402970e-05 elapsed=18:56:41 avg_step=3.65s recent_step=2.65s eta=31:47:28 +2026-09-15 08:44:05 | INFO | train rank=0 | step=18680/50000 progress=37.4% loss=0.098462 lr=7.400070e-05 lr_backbone=7.400070e-05 elapsed=18:57:38 avg_step=3.65s recent_step=5.66s eta=31:47:25 +2026-09-15 08:44:36 | INFO | train rank=0 | step=18690/50000 progress=37.4% loss=0.138728 lr=7.397168e-05 lr_backbone=7.397168e-05 elapsed=18:58:09 avg_step=3.65s recent_step=3.17s eta=31:46:40 +2026-09-15 08:45:12 | INFO | train rank=0 | step=18700/50000 progress=37.4% loss=0.133246 lr=7.394266e-05 lr_backbone=7.394266e-05 elapsed=18:58:45 avg_step=3.65s recent_step=3.59s eta=31:46:03 +2026-09-15 08:45:44 | INFO | train rank=0 | step=18710/50000 progress=37.4% loss=0.109435 lr=7.391362e-05 lr_backbone=7.391362e-05 elapsed=18:59:17 avg_step=3.65s recent_step=3.16s eta=31:45:18 +2026-09-15 08:46:28 | INFO | train rank=0 | step=18720/50000 progress=37.4% loss=0.113764 lr=7.388457e-05 lr_backbone=7.388457e-05 elapsed=19:00:01 avg_step=3.65s recent_step=4.41s eta=31:44:54 +2026-09-15 08:47:07 | INFO | train rank=0 | step=18730/50000 progress=37.5% loss=0.105908 lr=7.385552e-05 lr_backbone=7.385552e-05 elapsed=19:00:40 avg_step=3.65s recent_step=3.90s eta=31:44:22 +2026-09-15 08:47:46 | INFO | train rank=0 | step=18740/50000 progress=37.5% loss=0.092734 lr=7.382645e-05 lr_backbone=7.382645e-05 elapsed=19:01:19 avg_step=3.65s recent_step=3.89s eta=31:43:49 +2026-09-15 08:48:23 | INFO | train rank=0 | step=18750/50000 progress=37.5% loss=0.101674 lr=7.379737e-05 lr_backbone=7.379737e-05 elapsed=19:01:56 avg_step=3.65s recent_step=3.71s eta=31:43:13 +2026-09-15 08:49:11 | INFO | train rank=0 | step=18760/50000 progress=37.5% loss=0.105984 lr=7.376828e-05 lr_backbone=7.376828e-05 elapsed=19:02:44 avg_step=3.65s recent_step=4.78s eta=31:42:56 +2026-09-15 08:49:41 | INFO | train rank=0 | step=18770/50000 progress=37.5% loss=0.078088 lr=7.373918e-05 lr_backbone=7.373918e-05 elapsed=19:03:14 avg_step=3.65s recent_step=3.00s eta=31:42:08 +2026-09-15 08:50:11 | INFO | train rank=0 | step=18780/50000 progress=37.6% loss=0.102736 lr=7.371007e-05 lr_backbone=7.371007e-05 elapsed=19:03:44 avg_step=3.65s recent_step=3.01s eta=31:41:21 +2026-09-15 08:50:36 | INFO | train rank=0 | step=18790/50000 progress=37.6% loss=0.112082 lr=7.368095e-05 lr_backbone=7.368095e-05 elapsed=19:04:09 avg_step=3.65s recent_step=2.48s eta=31:40:25 +2026-09-15 08:51:28 | INFO | train rank=0 | step=18800/50000 progress=37.6% loss=0.107693 lr=7.365182e-05 lr_backbone=7.365182e-05 elapsed=19:05:01 avg_step=3.65s recent_step=5.23s eta=31:40:15 +2026-09-15 08:52:04 | INFO | train rank=0 | step=18810/50000 progress=37.6% loss=0.090911 lr=7.362268e-05 lr_backbone=7.362268e-05 elapsed=19:05:36 avg_step=3.65s recent_step=3.53s eta=31:39:36 +2026-09-15 08:52:39 | INFO | train rank=0 | step=18820/50000 progress=37.6% loss=0.109771 lr=7.359353e-05 lr_backbone=7.359353e-05 elapsed=19:06:12 avg_step=3.65s recent_step=3.58s eta=31:38:58 +2026-09-15 08:53:12 | INFO | train rank=0 | step=18830/50000 progress=37.7% loss=0.113503 lr=7.356437e-05 lr_backbone=7.356437e-05 elapsed=19:06:45 avg_step=3.65s recent_step=3.28s eta=31:38:16 +2026-09-15 08:54:03 | INFO | train rank=0 | step=18840/50000 progress=37.7% loss=0.120761 lr=7.353520e-05 lr_backbone=7.353520e-05 elapsed=19:07:36 avg_step=3.65s recent_step=5.08s eta=31:38:03 +2026-09-15 08:54:41 | INFO | train rank=0 | step=18850/50000 progress=37.7% loss=0.134334 lr=7.350601e-05 lr_backbone=7.350601e-05 elapsed=19:08:14 avg_step=3.65s recent_step=3.78s eta=31:37:28 +2026-09-15 08:55:13 | INFO | train rank=0 | step=18860/50000 progress=37.7% loss=0.116884 lr=7.347682e-05 lr_backbone=7.347682e-05 elapsed=19:08:46 avg_step=3.65s recent_step=3.20s eta=31:36:44 +2026-09-15 08:55:53 | INFO | train rank=0 | step=18870/50000 progress=37.7% loss=0.133323 lr=7.344762e-05 lr_backbone=7.344762e-05 elapsed=19:09:26 avg_step=3.65s recent_step=4.02s eta=31:36:14 +2026-09-15 08:56:36 | INFO | train rank=0 | step=18880/50000 progress=37.8% loss=0.109603 lr=7.341841e-05 lr_backbone=7.341841e-05 elapsed=19:10:08 avg_step=3.66s recent_step=4.27s eta=31:35:47 +2026-09-15 08:57:01 | INFO | train rank=0 | step=18890/50000 progress=37.8% loss=0.094830 lr=7.338918e-05 lr_backbone=7.338918e-05 elapsed=19:10:34 avg_step=3.65s recent_step=2.58s eta=31:34:53 +2026-09-15 08:57:30 | INFO | train rank=0 | step=18900/50000 progress=37.8% loss=0.106657 lr=7.335995e-05 lr_backbone=7.335995e-05 elapsed=19:11:03 avg_step=3.65s recent_step=2.87s eta=31:34:03 +2026-09-15 08:58:11 | INFO | train rank=0 | step=18910/50000 progress=37.8% loss=0.094750 lr=7.333071e-05 lr_backbone=7.333071e-05 elapsed=19:11:44 avg_step=3.65s recent_step=4.12s eta=31:33:35 +2026-09-15 08:59:07 | INFO | train rank=0 | step=18920/50000 progress=37.8% loss=0.111535 lr=7.330145e-05 lr_backbone=7.330145e-05 elapsed=19:12:40 avg_step=3.66s recent_step=5.57s eta=31:33:29 +2026-09-15 08:59:38 | INFO | train rank=0 | step=18930/50000 progress=37.9% loss=0.106308 lr=7.327219e-05 lr_backbone=7.327219e-05 elapsed=19:13:11 avg_step=3.66s recent_step=3.09s eta=31:32:43 +2026-09-15 09:00:09 | INFO | train rank=0 | step=18940/50000 progress=37.9% loss=0.091563 lr=7.324292e-05 lr_backbone=7.324292e-05 elapsed=19:13:41 avg_step=3.65s recent_step=3.08s eta=31:31:58 +2026-09-15 09:00:41 | INFO | train rank=0 | step=18950/50000 progress=37.9% loss=0.127947 lr=7.321363e-05 lr_backbone=7.321363e-05 elapsed=19:14:14 avg_step=3.65s recent_step=3.27s eta=31:31:15 +2026-09-15 09:01:35 | INFO | train rank=0 | step=18960/50000 progress=37.9% loss=0.099410 lr=7.318434e-05 lr_backbone=7.318434e-05 elapsed=19:15:08 avg_step=3.66s recent_step=5.38s eta=31:31:06 +2026-09-15 09:02:14 | INFO | train rank=0 | step=18970/50000 progress=37.9% loss=0.082775 lr=7.315503e-05 lr_backbone=7.315503e-05 elapsed=19:15:46 avg_step=3.66s recent_step=3.85s eta=31:30:33 +2026-09-15 09:02:46 | INFO | train rank=0 | step=18980/50000 progress=38.0% loss=0.109070 lr=7.312572e-05 lr_backbone=7.312572e-05 elapsed=19:16:19 avg_step=3.66s recent_step=3.26s eta=31:29:50 +2026-09-15 09:03:23 | INFO | train rank=0 | step=18990/50000 progress=38.0% loss=0.110695 lr=7.309639e-05 lr_backbone=7.309639e-05 elapsed=19:16:56 avg_step=3.66s recent_step=3.65s eta=31:29:13 +2026-09-15 09:04:16 | INFO | train rank=0 | step=19000/50000 progress=38.0% loss=0.106341 lr=7.306706e-05 lr_backbone=7.306706e-05 elapsed=19:17:49 avg_step=3.66s recent_step=5.30s eta=31:29:04 +2026-09-15 09:04:45 | INFO | train rank=0 | step=19010/50000 progress=38.0% loss=0.077943 lr=7.303771e-05 lr_backbone=7.303771e-05 elapsed=19:18:18 avg_step=3.66s recent_step=2.91s eta=31:28:15 +2026-09-15 09:05:28 | INFO | train rank=0 | step=19020/50000 progress=38.0% loss=0.083477 lr=7.300836e-05 lr_backbone=7.300836e-05 elapsed=19:19:01 avg_step=3.66s recent_step=4.33s eta=31:27:50 +2026-09-15 09:06:02 | INFO | train rank=0 | step=19030/50000 progress=38.1% loss=0.101783 lr=7.297899e-05 lr_backbone=7.297899e-05 elapsed=19:19:35 avg_step=3.66s recent_step=3.36s eta=31:27:08 +2026-09-15 09:06:55 | INFO | train rank=0 | step=19040/50000 progress=38.1% loss=0.137304 lr=7.294962e-05 lr_backbone=7.294962e-05 elapsed=19:20:27 avg_step=3.66s recent_step=5.29s eta=31:26:58 +2026-09-15 09:07:23 | INFO | train rank=0 | step=19050/50000 progress=38.1% loss=0.105124 lr=7.292023e-05 lr_backbone=7.292023e-05 elapsed=19:20:56 avg_step=3.66s recent_step=2.87s eta=31:26:09 +2026-09-15 09:07:56 | INFO | train rank=0 | step=19060/50000 progress=38.1% loss=0.096090 lr=7.289084e-05 lr_backbone=7.289084e-05 elapsed=19:21:29 avg_step=3.66s recent_step=3.26s eta=31:25:26 +2026-09-15 09:08:30 | INFO | train rank=0 | step=19070/50000 progress=38.1% loss=0.110735 lr=7.286143e-05 lr_backbone=7.286143e-05 elapsed=19:22:03 avg_step=3.66s recent_step=3.43s eta=31:24:45 +2026-09-15 09:09:07 | INFO | train rank=0 | step=19080/50000 progress=38.2% loss=0.099201 lr=7.283202e-05 lr_backbone=7.283202e-05 elapsed=19:22:40 avg_step=3.66s recent_step=3.72s eta=31:24:10 +2026-09-15 09:09:35 | INFO | train rank=0 | step=19090/50000 progress=38.2% loss=0.085175 lr=7.280259e-05 lr_backbone=7.280259e-05 elapsed=19:23:08 avg_step=3.66s recent_step=2.77s eta=31:23:19 +2026-09-15 09:10:09 | INFO | train rank=0 | step=19100/50000 progress=38.2% loss=0.094098 lr=7.277316e-05 lr_backbone=7.277316e-05 elapsed=19:23:42 avg_step=3.66s recent_step=3.38s eta=31:22:38 +2026-09-15 09:10:47 | INFO | train rank=0 | step=19110/50000 progress=38.2% loss=0.097819 lr=7.274371e-05 lr_backbone=7.274371e-05 elapsed=19:24:20 avg_step=3.66s recent_step=3.81s eta=31:22:04 +2026-09-15 09:11:40 | INFO | train rank=0 | step=19120/50000 progress=38.2% loss=0.122405 lr=7.271426e-05 lr_backbone=7.271426e-05 elapsed=19:25:13 avg_step=3.66s recent_step=5.28s eta=31:21:54 +2026-09-15 09:12:07 | INFO | train rank=0 | step=19130/50000 progress=38.3% loss=0.100017 lr=7.268479e-05 lr_backbone=7.268479e-05 elapsed=19:25:40 avg_step=3.66s recent_step=2.72s eta=31:21:02 +2026-09-15 09:12:48 | INFO | train rank=0 | step=19140/50000 progress=38.3% loss=0.085971 lr=7.265532e-05 lr_backbone=7.265532e-05 elapsed=19:26:21 avg_step=3.66s recent_step=4.08s eta=31:20:32 +2026-09-15 09:13:15 | INFO | train rank=0 | step=19150/50000 progress=38.3% loss=0.105618 lr=7.262583e-05 lr_backbone=7.262583e-05 elapsed=19:26:47 avg_step=3.66s recent_step=2.67s eta=31:19:40 +2026-09-15 09:13:57 | INFO | train rank=0 | step=19160/50000 progress=38.3% loss=0.097204 lr=7.259634e-05 lr_backbone=7.259634e-05 elapsed=19:27:30 avg_step=3.66s recent_step=4.29s eta=31:19:13 +2026-09-15 09:14:21 | INFO | train rank=0 | step=19170/50000 progress=38.3% loss=0.109766 lr=7.256683e-05 lr_backbone=7.256683e-05 elapsed=19:27:54 avg_step=3.66s recent_step=2.34s eta=31:18:16 +2026-09-15 09:15:02 | INFO | train rank=0 | step=19180/50000 progress=38.4% loss=0.110473 lr=7.253732e-05 lr_backbone=7.253732e-05 elapsed=19:28:35 avg_step=3.66s recent_step=4.10s eta=31:17:46 +2026-09-15 09:15:39 | INFO | train rank=0 | step=19190/50000 progress=38.4% loss=0.109619 lr=7.250779e-05 lr_backbone=7.250779e-05 elapsed=19:29:12 avg_step=3.66s recent_step=3.75s eta=31:17:11 +2026-09-15 09:16:31 | INFO | train rank=0 | step=19200/50000 progress=38.4% loss=0.162965 lr=7.247826e-05 lr_backbone=7.247826e-05 elapsed=19:30:04 avg_step=3.66s recent_step=5.20s eta=31:17:00 +2026-09-15 09:17:06 | INFO | train rank=0 | step=19210/50000 progress=38.4% loss=0.089585 lr=7.244871e-05 lr_backbone=7.244871e-05 elapsed=19:30:39 avg_step=3.66s recent_step=3.46s eta=31:16:20 +2026-09-15 09:17:39 | INFO | train rank=0 | step=19220/50000 progress=38.4% loss=0.101698 lr=7.241916e-05 lr_backbone=7.241916e-05 elapsed=19:31:12 avg_step=3.66s recent_step=3.27s eta=31:15:37 +2026-09-15 09:18:08 | INFO | train rank=0 | step=19230/50000 progress=38.5% loss=0.093404 lr=7.238960e-05 lr_backbone=7.238960e-05 elapsed=19:31:41 avg_step=3.66s recent_step=2.89s eta=31:14:48 +2026-09-15 09:18:59 | INFO | train rank=0 | step=19240/50000 progress=38.5% loss=0.114963 lr=7.236002e-05 lr_backbone=7.236002e-05 elapsed=19:32:32 avg_step=3.66s recent_step=5.10s eta=31:14:35 +2026-09-15 09:19:37 | INFO | train rank=0 | step=19250/50000 progress=38.5% loss=0.113642 lr=7.233044e-05 lr_backbone=7.233044e-05 elapsed=19:33:10 avg_step=3.66s recent_step=3.87s eta=31:14:02 +2026-09-15 09:20:13 | INFO | train rank=0 | step=19260/50000 progress=38.5% loss=0.093833 lr=7.230085e-05 lr_backbone=7.230085e-05 elapsed=19:33:46 avg_step=3.66s recent_step=3.53s eta=31:13:23 +2026-09-15 09:20:40 | INFO | train rank=0 | step=19270/50000 progress=38.5% loss=0.109429 lr=7.227124e-05 lr_backbone=7.227124e-05 elapsed=19:34:13 avg_step=3.66s recent_step=2.75s eta=31:12:32 +2026-09-15 09:21:27 | INFO | train rank=0 | step=19280/50000 progress=38.6% loss=0.086947 lr=7.224163e-05 lr_backbone=7.224163e-05 elapsed=19:35:00 avg_step=3.66s recent_step=4.72s eta=31:12:13 +2026-09-15 09:22:01 | INFO | train rank=0 | step=19290/50000 progress=38.6% loss=0.093404 lr=7.221201e-05 lr_backbone=7.221201e-05 elapsed=19:35:34 avg_step=3.66s recent_step=3.40s eta=31:11:32 +2026-09-15 09:22:33 | INFO | train rank=0 | step=19300/50000 progress=38.6% loss=0.103236 lr=7.218238e-05 lr_backbone=7.218238e-05 elapsed=19:36:06 avg_step=3.66s recent_step=3.20s eta=31:10:48 +2026-09-15 09:23:10 | INFO | train rank=0 | step=19310/50000 progress=38.6% loss=0.091756 lr=7.215274e-05 lr_backbone=7.215274e-05 elapsed=19:36:42 avg_step=3.66s recent_step=3.61s eta=31:10:11 +2026-09-15 09:23:56 | INFO | train rank=0 | step=19320/50000 progress=38.6% loss=0.134644 lr=7.212308e-05 lr_backbone=7.212308e-05 elapsed=19:37:29 avg_step=3.66s recent_step=4.63s eta=31:09:50 +2026-09-15 09:24:21 | INFO | train rank=0 | step=19330/50000 progress=38.7% loss=0.081750 lr=7.209342e-05 lr_backbone=7.209342e-05 elapsed=19:37:54 avg_step=3.66s recent_step=2.51s eta=31:08:55 +2026-09-15 09:24:50 | INFO | train rank=0 | step=19340/50000 progress=38.7% loss=0.101173 lr=7.206375e-05 lr_backbone=7.206375e-05 elapsed=19:38:23 avg_step=3.66s recent_step=2.87s eta=31:08:06 +2026-09-15 09:25:29 | INFO | train rank=0 | step=19350/50000 progress=38.7% loss=0.096783 lr=7.203407e-05 lr_backbone=7.203407e-05 elapsed=19:39:02 avg_step=3.66s recent_step=3.91s eta=31:07:34 +2026-09-15 09:26:17 | INFO | train rank=0 | step=19360/50000 progress=38.7% loss=0.098477 lr=7.200438e-05 lr_backbone=7.200438e-05 elapsed=19:39:50 avg_step=3.66s recent_step=4.80s eta=31:07:15 +2026-09-15 09:26:58 | INFO | train rank=0 | step=19370/50000 progress=38.7% loss=0.103840 lr=7.197468e-05 lr_backbone=7.197468e-05 elapsed=19:40:31 avg_step=3.66s recent_step=4.13s eta=31:06:46 +2026-09-15 09:27:26 | INFO | train rank=0 | step=19380/50000 progress=38.8% loss=0.113680 lr=7.194497e-05 lr_backbone=7.194497e-05 elapsed=19:40:59 avg_step=3.66s recent_step=2.80s eta=31:05:56 +2026-09-15 09:27:56 | INFO | train rank=0 | step=19390/50000 progress=38.8% loss=0.099491 lr=7.191525e-05 lr_backbone=7.191525e-05 elapsed=19:41:29 avg_step=3.66s recent_step=3.00s eta=31:05:09 +2026-09-15 09:28:53 | INFO | train rank=0 | step=19400/50000 progress=38.8% loss=0.086794 lr=7.188553e-05 lr_backbone=7.188553e-05 elapsed=19:42:26 avg_step=3.66s recent_step=5.71s eta=31:05:05 +2026-09-15 09:29:22 | INFO | train rank=0 | step=19410/50000 progress=38.8% loss=0.126706 lr=7.185579e-05 lr_backbone=7.185579e-05 elapsed=19:42:55 avg_step=3.66s recent_step=2.89s eta=31:04:16 +2026-09-15 09:29:53 | INFO | train rank=0 | step=19420/50000 progress=38.8% loss=0.104018 lr=7.182604e-05 lr_backbone=7.182604e-05 elapsed=19:43:25 avg_step=3.66s recent_step=3.04s eta=31:03:30 +2026-09-15 09:30:28 | INFO | train rank=0 | step=19430/50000 progress=38.9% loss=0.116974 lr=7.179628e-05 lr_backbone=7.179628e-05 elapsed=19:44:01 avg_step=3.66s recent_step=3.56s eta=31:02:52 +2026-09-15 09:31:22 | INFO | train rank=0 | step=19440/50000 progress=38.9% loss=0.116849 lr=7.176652e-05 lr_backbone=7.176652e-05 elapsed=19:44:54 avg_step=3.66s recent_step=5.35s eta=31:02:42 +2026-09-15 09:31:54 | INFO | train rank=0 | step=19450/50000 progress=38.9% loss=0.096536 lr=7.173674e-05 lr_backbone=7.173674e-05 elapsed=19:45:27 avg_step=3.66s recent_step=3.27s eta=31:01:59 +2026-09-15 09:32:28 | INFO | train rank=0 | step=19460/50000 progress=38.9% loss=0.083011 lr=7.170696e-05 lr_backbone=7.170696e-05 elapsed=19:46:01 avg_step=3.66s recent_step=3.39s eta=31:01:19 +2026-09-15 09:33:05 | INFO | train rank=0 | step=19470/50000 progress=38.9% loss=0.135879 lr=7.167716e-05 lr_backbone=7.167716e-05 elapsed=19:46:38 avg_step=3.66s recent_step=3.65s eta=31:00:42 +2026-09-15 09:34:01 | INFO | train rank=0 | step=19480/50000 progress=39.0% loss=0.101375 lr=7.164736e-05 lr_backbone=7.164736e-05 elapsed=19:47:33 avg_step=3.66s recent_step=5.58s eta=31:00:35 +2026-09-15 09:34:35 | INFO | train rank=0 | step=19490/50000 progress=39.0% loss=0.075289 lr=7.161754e-05 lr_backbone=7.161754e-05 elapsed=19:48:08 avg_step=3.66s recent_step=3.46s eta=30:59:56 +2026-09-15 09:34:59 | INFO | train rank=0 | step=19500/50000 progress=39.0% loss=0.096535 lr=7.158772e-05 lr_backbone=7.158772e-05 elapsed=19:48:32 avg_step=3.66s recent_step=2.39s eta=30:58:59 +2026-09-15 09:35:36 | INFO | train rank=0 | step=19510/50000 progress=39.0% loss=0.077849 lr=7.155788e-05 lr_backbone=7.155788e-05 elapsed=19:49:09 avg_step=3.66s recent_step=3.73s eta=30:58:24 +2026-09-15 09:36:32 | INFO | train rank=0 | step=19520/50000 progress=39.0% loss=0.110841 lr=7.152804e-05 lr_backbone=7.152804e-05 elapsed=19:50:05 avg_step=3.66s recent_step=5.53s eta=30:58:17 +2026-09-15 09:37:04 | INFO | train rank=0 | step=19530/50000 progress=39.1% loss=0.089508 lr=7.149819e-05 lr_backbone=7.149819e-05 elapsed=19:50:37 avg_step=3.66s recent_step=3.24s eta=30:57:34 +2026-09-15 09:37:38 | INFO | train rank=0 | step=19540/50000 progress=39.1% loss=0.107929 lr=7.146833e-05 lr_backbone=7.146833e-05 elapsed=19:51:11 avg_step=3.66s recent_step=3.42s eta=30:56:53 +2026-09-15 09:38:13 | INFO | train rank=0 | step=19550/50000 progress=39.1% loss=0.120621 lr=7.143846e-05 lr_backbone=7.143846e-05 elapsed=19:51:45 avg_step=3.66s recent_step=3.43s eta=30:56:13 +2026-09-15 09:38:50 | INFO | train rank=0 | step=19560/50000 progress=39.1% loss=0.093075 lr=7.140858e-05 lr_backbone=7.140858e-05 elapsed=19:52:22 avg_step=3.66s recent_step=3.71s eta=30:55:37 +2026-09-15 09:39:16 | INFO | train rank=0 | step=19570/50000 progress=39.1% loss=0.101213 lr=7.137869e-05 lr_backbone=7.137869e-05 elapsed=19:52:49 avg_step=3.66s recent_step=2.64s eta=30:54:45 +2026-09-15 09:39:43 | INFO | train rank=0 | step=19580/50000 progress=39.2% loss=0.110179 lr=7.134879e-05 lr_backbone=7.134879e-05 elapsed=19:53:16 avg_step=3.66s recent_step=2.70s eta=30:53:54 +2026-09-15 09:40:13 | INFO | train rank=0 | step=19590/50000 progress=39.2% loss=0.113476 lr=7.131888e-05 lr_backbone=7.131888e-05 elapsed=19:53:46 avg_step=3.66s recent_step=3.01s eta=30:53:07 +2026-09-15 09:40:52 | INFO | train rank=0 | step=19600/50000 progress=39.2% loss=0.071357 lr=7.128896e-05 lr_backbone=7.128896e-05 elapsed=19:54:25 avg_step=3.66s recent_step=3.91s eta=30:52:34 +2026-09-15 09:41:24 | INFO | train rank=0 | step=19610/50000 progress=39.2% loss=0.094533 lr=7.125904e-05 lr_backbone=7.125904e-05 elapsed=19:54:57 avg_step=3.66s recent_step=3.16s eta=30:51:50 +2026-09-15 09:41:49 | INFO | train rank=0 | step=19620/50000 progress=39.2% loss=0.093924 lr=7.122910e-05 lr_backbone=7.122910e-05 elapsed=19:55:22 avg_step=3.66s recent_step=2.51s eta=30:50:56 +2026-09-15 09:42:26 | INFO | train rank=0 | step=19630/50000 progress=39.3% loss=0.068467 lr=7.119916e-05 lr_backbone=7.119916e-05 elapsed=19:55:59 avg_step=3.66s recent_step=3.71s eta=30:50:20 +2026-09-15 09:43:08 | INFO | train rank=0 | step=19640/50000 progress=39.3% loss=0.150724 lr=7.116920e-05 lr_backbone=7.116920e-05 elapsed=19:56:41 avg_step=3.66s recent_step=4.18s eta=30:49:52 +2026-09-15 09:43:56 | INFO | train rank=0 | step=19650/50000 progress=39.3% loss=0.095745 lr=7.113924e-05 lr_backbone=7.113924e-05 elapsed=19:57:29 avg_step=3.66s recent_step=4.78s eta=30:49:32 +2026-09-15 09:44:27 | INFO | train rank=0 | step=19660/50000 progress=39.3% loss=0.097780 lr=7.110926e-05 lr_backbone=7.110926e-05 elapsed=19:58:00 avg_step=3.66s recent_step=3.15s eta=30:48:48 +2026-09-15 09:45:03 | INFO | train rank=0 | step=19670/50000 progress=39.3% loss=0.132151 lr=7.107928e-05 lr_backbone=7.107928e-05 elapsed=19:58:36 avg_step=3.66s recent_step=3.57s eta=30:48:10 +2026-09-15 09:45:34 | INFO | train rank=0 | step=19680/50000 progress=39.4% loss=0.109667 lr=7.104929e-05 lr_backbone=7.104929e-05 elapsed=19:59:07 avg_step=3.66s recent_step=3.08s eta=30:47:25 +2026-09-15 09:46:16 | INFO | train rank=0 | step=19690/50000 progress=39.4% loss=0.119052 lr=7.101929e-05 lr_backbone=7.101929e-05 elapsed=19:59:49 avg_step=3.66s recent_step=4.21s eta=30:46:57 +2026-09-15 09:46:51 | INFO | train rank=0 | step=19700/50000 progress=39.4% loss=0.093409 lr=7.098928e-05 lr_backbone=7.098928e-05 elapsed=20:00:24 avg_step=3.66s recent_step=3.51s eta=30:46:18 +2026-09-15 09:47:29 | INFO | train rank=0 | step=19710/50000 progress=39.4% loss=0.116668 lr=7.095926e-05 lr_backbone=7.095926e-05 elapsed=20:01:02 avg_step=3.66s recent_step=3.82s eta=30:45:44 +2026-09-15 09:48:02 | INFO | train rank=0 | step=19720/50000 progress=39.4% loss=0.090295 lr=7.092923e-05 lr_backbone=7.092923e-05 elapsed=20:01:35 avg_step=3.66s recent_step=3.29s eta=30:45:02 +2026-09-15 09:48:45 | INFO | train rank=0 | step=19730/50000 progress=39.5% loss=0.094210 lr=7.089920e-05 lr_backbone=7.089920e-05 elapsed=20:02:18 avg_step=3.66s recent_step=4.31s eta=30:44:35 +2026-09-15 09:49:17 | INFO | train rank=0 | step=19740/50000 progress=39.5% loss=0.090006 lr=7.086915e-05 lr_backbone=7.086915e-05 elapsed=20:02:50 avg_step=3.66s recent_step=3.20s eta=30:43:52 +2026-09-15 09:49:55 | INFO | train rank=0 | step=19750/50000 progress=39.5% loss=0.123897 lr=7.083909e-05 lr_backbone=7.083909e-05 elapsed=20:03:27 avg_step=3.66s recent_step=3.76s eta=30:43:17 +2026-09-15 09:50:29 | INFO | train rank=0 | step=19760/50000 progress=39.5% loss=0.084909 lr=7.080903e-05 lr_backbone=7.080903e-05 elapsed=20:04:02 avg_step=3.66s recent_step=3.47s eta=30:42:37 +2026-09-15 09:51:14 | INFO | train rank=0 | step=19770/50000 progress=39.5% loss=0.100211 lr=7.077895e-05 lr_backbone=7.077895e-05 elapsed=20:04:46 avg_step=3.66s recent_step=4.42s eta=30:42:12 +2026-09-15 09:51:44 | INFO | train rank=0 | step=19780/50000 progress=39.6% loss=0.100601 lr=7.074887e-05 lr_backbone=7.074887e-05 elapsed=20:05:16 avg_step=3.66s recent_step=3.01s eta=30:41:26 +2026-09-15 09:52:26 | INFO | train rank=0 | step=19790/50000 progress=39.6% loss=0.106179 lr=7.071878e-05 lr_backbone=7.071878e-05 elapsed=20:05:59 avg_step=3.66s recent_step=4.28s eta=30:40:59 +2026-09-15 09:53:05 | INFO | train rank=0 | step=19800/50000 progress=39.6% loss=0.095950 lr=7.068868e-05 lr_backbone=7.068868e-05 elapsed=20:06:38 avg_step=3.66s recent_step=3.89s eta=30:40:26 +2026-09-15 09:53:42 | INFO | train rank=0 | step=19810/50000 progress=39.6% loss=0.100275 lr=7.065857e-05 lr_backbone=7.065857e-05 elapsed=20:07:14 avg_step=3.66s recent_step=3.61s eta=30:39:49 +2026-09-15 09:54:17 | INFO | train rank=0 | step=19820/50000 progress=39.6% loss=0.077629 lr=7.062845e-05 lr_backbone=7.062845e-05 elapsed=20:07:50 avg_step=3.66s recent_step=3.55s eta=30:39:10 +2026-09-15 09:55:01 | INFO | train rank=0 | step=19830/50000 progress=39.7% loss=0.099224 lr=7.059832e-05 lr_backbone=7.059832e-05 elapsed=20:08:34 avg_step=3.66s recent_step=4.38s eta=30:38:45 +2026-09-15 09:55:44 | INFO | train rank=0 | step=19840/50000 progress=39.7% loss=0.088429 lr=7.056818e-05 lr_backbone=7.056818e-05 elapsed=20:09:17 avg_step=3.66s recent_step=4.33s eta=30:38:18 +2026-09-15 09:56:14 | INFO | train rank=0 | step=19850/50000 progress=39.7% loss=0.129506 lr=7.053804e-05 lr_backbone=7.053804e-05 elapsed=20:09:47 avg_step=3.66s recent_step=3.01s eta=30:37:32 +2026-09-15 09:56:43 | INFO | train rank=0 | step=19860/50000 progress=39.7% loss=0.082205 lr=7.050788e-05 lr_backbone=7.050788e-05 elapsed=20:10:16 avg_step=3.66s recent_step=2.90s eta=30:36:44 +2026-09-15 09:57:28 | INFO | train rank=0 | step=19870/50000 progress=39.7% loss=0.089438 lr=7.047772e-05 lr_backbone=7.047772e-05 elapsed=20:11:01 avg_step=3.66s recent_step=4.45s eta=30:36:20 +2026-09-15 09:58:07 | INFO | train rank=0 | step=19880/50000 progress=39.8% loss=0.131241 lr=7.044754e-05 lr_backbone=7.044754e-05 elapsed=20:11:40 avg_step=3.66s recent_step=3.93s eta=30:35:47 +2026-09-15 09:58:40 | INFO | train rank=0 | step=19890/50000 progress=39.8% loss=0.130714 lr=7.041736e-05 lr_backbone=7.041736e-05 elapsed=20:12:12 avg_step=3.66s recent_step=3.25s eta=30:35:04 +2026-09-15 09:59:13 | INFO | train rank=0 | step=19900/50000 progress=39.8% loss=0.099394 lr=7.038717e-05 lr_backbone=7.038717e-05 elapsed=20:12:46 avg_step=3.66s recent_step=3.33s eta=30:34:23 +2026-09-15 09:59:58 | INFO | train rank=0 | step=19910/50000 progress=39.8% loss=0.094329 lr=7.035697e-05 lr_backbone=7.035697e-05 elapsed=20:13:31 avg_step=3.66s recent_step=4.50s eta=30:33:59 +2026-09-15 10:00:36 | INFO | train rank=0 | step=19920/50000 progress=39.8% loss=0.132918 lr=7.032676e-05 lr_backbone=7.032676e-05 elapsed=20:14:09 avg_step=3.66s recent_step=3.81s eta=30:33:25 +2026-09-15 10:01:15 | INFO | train rank=0 | step=19930/50000 progress=39.9% loss=0.099835 lr=7.029654e-05 lr_backbone=7.029654e-05 elapsed=20:14:48 avg_step=3.66s recent_step=3.89s eta=30:32:52 +2026-09-15 10:01:56 | INFO | train rank=0 | step=19940/50000 progress=39.9% loss=0.102879 lr=7.026632e-05 lr_backbone=7.026632e-05 elapsed=20:15:29 avg_step=3.66s recent_step=4.13s eta=30:32:22 +2026-09-15 10:02:31 | INFO | train rank=0 | step=19950/50000 progress=39.9% loss=0.109834 lr=7.023608e-05 lr_backbone=7.023608e-05 elapsed=20:16:04 avg_step=3.66s recent_step=3.48s eta=30:31:43 +2026-09-15 10:03:28 | INFO | train rank=0 | step=19960/50000 progress=39.9% loss=0.093428 lr=7.020584e-05 lr_backbone=7.020584e-05 elapsed=20:17:01 avg_step=3.66s recent_step=5.71s eta=30:31:38 +2026-09-15 10:04:06 | INFO | train rank=0 | step=19970/50000 progress=39.9% loss=0.111665 lr=7.017558e-05 lr_backbone=7.017558e-05 elapsed=20:17:39 avg_step=3.66s recent_step=3.79s eta=30:31:03 +2026-09-15 10:04:39 | INFO | train rank=0 | step=19980/50000 progress=40.0% loss=0.111801 lr=7.014532e-05 lr_backbone=7.014532e-05 elapsed=20:18:12 avg_step=3.66s recent_step=3.32s eta=30:30:21 +2026-09-15 10:05:13 | INFO | train rank=0 | step=19990/50000 progress=40.0% loss=0.102781 lr=7.011505e-05 lr_backbone=7.011505e-05 elapsed=20:18:46 avg_step=3.66s recent_step=3.39s eta=30:29:41 +2026-09-15 10:05:45 | INFO | train rank=0 | step=20000/50000 progress=40.0% loss=0.091743 lr=7.008477e-05 lr_backbone=7.008477e-05 elapsed=20:19:18 avg_step=3.66s recent_step=3.17s eta=30:28:57 +2026-09-15 10:06:20 | INFO | train rank=0 | step=20010/50000 progress=40.0% loss=0.139109 lr=7.005448e-05 lr_backbone=7.005448e-05 elapsed=20:19:52 avg_step=3.66s recent_step=3.48s eta=30:28:17 +2026-09-15 10:06:57 | INFO | train rank=0 | step=20020/50000 progress=40.0% loss=0.120097 lr=7.002419e-05 lr_backbone=7.002419e-05 elapsed=20:20:30 avg_step=3.66s recent_step=3.77s eta=30:27:43 +2026-09-15 10:07:54 | INFO | train rank=0 | step=20030/50000 progress=40.1% loss=0.111496 lr=6.999388e-05 lr_backbone=6.999388e-05 elapsed=20:21:27 avg_step=3.66s recent_step=5.64s eta=30:27:36 +2026-09-15 10:08:23 | INFO | train rank=0 | step=20040/50000 progress=40.1% loss=0.102437 lr=6.996356e-05 lr_backbone=6.996356e-05 elapsed=20:21:56 avg_step=3.66s recent_step=2.90s eta=30:26:48 +2026-09-15 10:08:56 | INFO | train rank=0 | step=20050/50000 progress=40.1% loss=0.102138 lr=6.993324e-05 lr_backbone=6.993324e-05 elapsed=20:22:29 avg_step=3.66s recent_step=3.30s eta=30:26:06 +2026-09-15 10:09:37 | INFO | train rank=0 | step=20060/50000 progress=40.1% loss=0.106866 lr=6.990291e-05 lr_backbone=6.990291e-05 elapsed=20:23:09 avg_step=3.66s recent_step=4.09s eta=30:25:36 +2026-09-15 10:10:28 | INFO | train rank=0 | step=20070/50000 progress=40.1% loss=0.093270 lr=6.987257e-05 lr_backbone=6.987257e-05 elapsed=20:24:01 avg_step=3.66s recent_step=5.14s eta=30:25:21 +2026-09-15 10:11:07 | INFO | train rank=0 | step=20080/50000 progress=40.2% loss=0.074152 lr=6.984222e-05 lr_backbone=6.984222e-05 elapsed=20:24:39 avg_step=3.66s recent_step=3.86s eta=30:24:48 +2026-09-15 10:11:44 | INFO | train rank=0 | step=20090/50000 progress=40.2% loss=0.111856 lr=6.981186e-05 lr_backbone=6.981186e-05 elapsed=20:25:16 avg_step=3.66s recent_step=3.69s eta=30:24:11 +2026-09-15 10:12:06 | INFO | train rank=0 | step=20100/50000 progress=40.2% loss=0.136455 lr=6.978149e-05 lr_backbone=6.978149e-05 elapsed=20:25:39 avg_step=3.66s recent_step=2.25s eta=30:23:14 +2026-09-15 10:12:49 | INFO | train rank=0 | step=20110/50000 progress=40.2% loss=0.094635 lr=6.975112e-05 lr_backbone=6.975112e-05 elapsed=20:26:22 avg_step=3.66s recent_step=4.29s eta=30:22:47 +2026-09-15 10:13:26 | INFO | train rank=0 | step=20120/50000 progress=40.2% loss=0.108652 lr=6.972073e-05 lr_backbone=6.972073e-05 elapsed=20:26:59 avg_step=3.66s recent_step=3.74s eta=30:22:11 +2026-09-15 10:14:00 | INFO | train rank=0 | step=20130/50000 progress=40.3% loss=0.138119 lr=6.969034e-05 lr_backbone=6.969034e-05 elapsed=20:27:33 avg_step=3.66s recent_step=3.41s eta=30:21:31 +2026-09-15 10:14:33 | INFO | train rank=0 | step=20140/50000 progress=40.3% loss=0.084320 lr=6.965994e-05 lr_backbone=6.965994e-05 elapsed=20:28:06 avg_step=3.66s recent_step=3.28s eta=30:20:49 +2026-09-15 10:15:25 | INFO | train rank=0 | step=20150/50000 progress=40.3% loss=0.094955 lr=6.962953e-05 lr_backbone=6.962953e-05 elapsed=20:28:58 avg_step=3.66s recent_step=5.19s eta=30:20:35 +2026-09-15 10:16:01 | INFO | train rank=0 | step=20160/50000 progress=40.3% loss=0.081693 lr=6.959911e-05 lr_backbone=6.959911e-05 elapsed=20:29:34 avg_step=3.66s recent_step=3.59s eta=30:19:57 +2026-09-15 10:16:30 | INFO | train rank=0 | step=20170/50000 progress=40.3% loss=0.110687 lr=6.956868e-05 lr_backbone=6.956868e-05 elapsed=20:30:02 avg_step=3.66s recent_step=2.86s eta=30:19:09 +2026-09-15 10:17:07 | INFO | train rank=0 | step=20180/50000 progress=40.4% loss=0.116987 lr=6.953825e-05 lr_backbone=6.953825e-05 elapsed=20:30:40 avg_step=3.66s recent_step=3.74s eta=30:18:33 +2026-09-15 10:17:56 | INFO | train rank=0 | step=20190/50000 progress=40.4% loss=0.086450 lr=6.950780e-05 lr_backbone=6.950780e-05 elapsed=20:31:29 avg_step=3.66s recent_step=4.94s eta=30:18:16 +2026-09-15 10:18:33 | INFO | train rank=0 | step=20200/50000 progress=40.4% loss=0.128816 lr=6.947735e-05 lr_backbone=6.947735e-05 elapsed=20:32:06 avg_step=3.66s recent_step=3.70s eta=30:17:40 +2026-09-15 10:19:08 | INFO | train rank=0 | step=20210/50000 progress=40.4% loss=0.087031 lr=6.944689e-05 lr_backbone=6.944689e-05 elapsed=20:32:41 avg_step=3.66s recent_step=3.49s eta=30:17:01 +2026-09-15 10:19:37 | INFO | train rank=0 | step=20220/50000 progress=40.4% loss=0.093275 lr=6.941642e-05 lr_backbone=6.941642e-05 elapsed=20:33:10 avg_step=3.66s recent_step=2.87s eta=30:16:12 +2026-09-15 10:20:25 | INFO | train rank=0 | step=20230/50000 progress=40.5% loss=0.100242 lr=6.938594e-05 lr_backbone=6.938594e-05 elapsed=20:33:57 avg_step=3.66s recent_step=4.75s eta=30:15:52 +2026-09-15 10:20:57 | INFO | train rank=0 | step=20240/50000 progress=40.5% loss=0.116664 lr=6.935545e-05 lr_backbone=6.935545e-05 elapsed=20:34:29 avg_step=3.66s recent_step=3.21s eta=30:15:09 +2026-09-15 10:21:31 | INFO | train rank=0 | step=20250/50000 progress=40.5% loss=0.096749 lr=6.932496e-05 lr_backbone=6.932496e-05 elapsed=20:35:04 avg_step=3.66s recent_step=3.42s eta=30:14:28 +2026-09-15 10:22:07 | INFO | train rank=0 | step=20260/50000 progress=40.5% loss=0.115846 lr=6.929445e-05 lr_backbone=6.929445e-05 elapsed=20:35:40 avg_step=3.66s recent_step=3.66s eta=30:13:52 +2026-09-15 10:23:05 | INFO | train rank=0 | step=20270/50000 progress=40.5% loss=0.082729 lr=6.926394e-05 lr_backbone=6.926394e-05 elapsed=20:36:38 avg_step=3.66s recent_step=5.80s eta=30:13:47 +2026-09-15 10:23:34 | INFO | train rank=0 | step=20280/50000 progress=40.6% loss=0.143812 lr=6.923342e-05 lr_backbone=6.923342e-05 elapsed=20:37:07 avg_step=3.66s recent_step=2.89s eta=30:12:59 +2026-09-15 10:24:11 | INFO | train rank=0 | step=20290/50000 progress=40.6% loss=0.087040 lr=6.920289e-05 lr_backbone=6.920289e-05 elapsed=20:37:44 avg_step=3.66s recent_step=3.65s eta=30:12:22 +2026-09-15 10:24:37 | INFO | train rank=0 | step=20300/50000 progress=40.6% loss=0.109794 lr=6.917236e-05 lr_backbone=6.917236e-05 elapsed=20:38:10 avg_step=3.66s recent_step=2.61s eta=30:11:30 +2026-09-15 10:25:35 | INFO | train rank=0 | step=20310/50000 progress=40.6% loss=0.111289 lr=6.914181e-05 lr_backbone=6.914181e-05 elapsed=20:39:08 avg_step=3.66s recent_step=5.81s eta=30:11:25 +2026-09-15 10:26:06 | INFO | train rank=0 | step=20320/50000 progress=40.6% loss=0.105521 lr=6.911126e-05 lr_backbone=6.911126e-05 elapsed=20:39:39 avg_step=3.66s recent_step=3.08s eta=30:10:40 +2026-09-15 10:26:40 | INFO | train rank=0 | step=20330/50000 progress=40.7% loss=0.123022 lr=6.908069e-05 lr_backbone=6.908069e-05 elapsed=20:40:13 avg_step=3.66s recent_step=3.42s eta=30:10:00 +2026-09-15 10:27:13 | INFO | train rank=0 | step=20340/50000 progress=40.7% loss=0.115540 lr=6.905012e-05 lr_backbone=6.905012e-05 elapsed=20:40:46 avg_step=3.66s recent_step=3.32s eta=30:09:18 +2026-09-15 10:28:07 | INFO | train rank=0 | step=20350/50000 progress=40.7% loss=0.123524 lr=6.901954e-05 lr_backbone=6.901954e-05 elapsed=20:41:39 avg_step=3.66s recent_step=5.36s eta=30:09:06 +2026-09-15 10:28:37 | INFO | train rank=0 | step=20360/50000 progress=40.7% loss=0.087665 lr=6.898895e-05 lr_backbone=6.898895e-05 elapsed=20:42:09 avg_step=3.66s recent_step=2.98s eta=30:08:20 +2026-09-15 10:29:11 | INFO | train rank=0 | step=20370/50000 progress=40.7% loss=0.072740 lr=6.895836e-05 lr_backbone=6.895836e-05 elapsed=20:42:44 avg_step=3.66s recent_step=3.44s eta=30:07:40 +2026-09-15 10:29:42 | INFO | train rank=0 | step=20380/50000 progress=40.8% loss=0.094963 lr=6.892775e-05 lr_backbone=6.892775e-05 elapsed=20:43:15 avg_step=3.66s recent_step=3.16s eta=30:06:56 +2026-09-15 10:30:28 | INFO | train rank=0 | step=20390/50000 progress=40.8% loss=0.087971 lr=6.889714e-05 lr_backbone=6.889714e-05 elapsed=20:44:01 avg_step=3.66s recent_step=4.57s eta=30:06:33 +2026-09-15 10:31:07 | INFO | train rank=0 | step=20400/50000 progress=40.8% loss=0.148410 lr=6.886652e-05 lr_backbone=6.886652e-05 elapsed=20:44:40 avg_step=3.66s recent_step=3.91s eta=30:06:00 +2026-09-15 10:31:42 | INFO | train rank=0 | step=20410/50000 progress=40.8% loss=0.128015 lr=6.883589e-05 lr_backbone=6.883589e-05 elapsed=20:45:15 avg_step=3.66s recent_step=3.47s eta=30:05:20 +2026-09-15 10:32:15 | INFO | train rank=0 | step=20420/50000 progress=40.8% loss=0.060172 lr=6.880526e-05 lr_backbone=6.880526e-05 elapsed=20:45:47 avg_step=3.66s recent_step=3.25s eta=30:04:38 +2026-09-15 10:33:00 | INFO | train rank=0 | step=20430/50000 progress=40.9% loss=0.074481 lr=6.877461e-05 lr_backbone=6.877461e-05 elapsed=20:46:32 avg_step=3.66s recent_step=4.50s eta=30:04:13 +2026-09-15 10:33:27 | INFO | train rank=0 | step=20440/50000 progress=40.9% loss=0.096287 lr=6.874396e-05 lr_backbone=6.874396e-05 elapsed=20:47:00 avg_step=3.66s recent_step=2.79s eta=30:03:24 +2026-09-15 10:34:01 | INFO | train rank=0 | step=20450/50000 progress=40.9% loss=0.139038 lr=6.871329e-05 lr_backbone=6.871329e-05 elapsed=20:47:34 avg_step=3.66s recent_step=3.34s eta=30:02:43 +2026-09-15 10:34:33 | INFO | train rank=0 | step=20460/50000 progress=40.9% loss=0.118052 lr=6.868262e-05 lr_backbone=6.868262e-05 elapsed=20:48:06 avg_step=3.66s recent_step=3.19s eta=30:01:59 +2026-09-15 10:35:24 | INFO | train rank=0 | step=20470/50000 progress=40.9% loss=0.079477 lr=6.865195e-05 lr_backbone=6.865195e-05 elapsed=20:48:56 avg_step=3.66s recent_step=5.09s eta=30:01:43 +2026-09-15 10:35:59 | INFO | train rank=0 | step=20480/50000 progress=41.0% loss=0.129125 lr=6.862126e-05 lr_backbone=6.862126e-05 elapsed=20:49:32 avg_step=3.66s recent_step=3.56s eta=30:01:05 +2026-09-15 10:36:35 | INFO | train rank=0 | step=20490/50000 progress=41.0% loss=0.118624 lr=6.859057e-05 lr_backbone=6.859057e-05 elapsed=20:50:08 avg_step=3.66s recent_step=3.62s eta=30:00:28 +2026-09-15 10:37:03 | INFO | train rank=0 | step=20500/50000 progress=41.0% loss=0.088403 lr=6.855986e-05 lr_backbone=6.855986e-05 elapsed=20:50:36 avg_step=3.66s recent_step=2.80s eta=29:59:39 +2026-09-15 10:37:49 | INFO | train rank=0 | step=20510/50000 progress=41.0% loss=0.120623 lr=6.852915e-05 lr_backbone=6.852915e-05 elapsed=20:51:22 avg_step=3.66s recent_step=4.57s eta=29:59:16 +2026-09-15 10:38:27 | INFO | train rank=0 | step=20520/50000 progress=41.0% loss=0.108101 lr=6.849843e-05 lr_backbone=6.849843e-05 elapsed=20:52:00 avg_step=3.66s recent_step=3.82s eta=29:58:41 +2026-09-15 10:39:00 | INFO | train rank=0 | step=20530/50000 progress=41.1% loss=0.090489 lr=6.846771e-05 lr_backbone=6.846771e-05 elapsed=20:52:33 avg_step=3.66s recent_step=3.27s eta=29:57:59 +2026-09-15 10:39:29 | INFO | train rank=0 | step=20540/50000 progress=41.1% loss=0.144595 lr=6.843697e-05 lr_backbone=6.843697e-05 elapsed=20:53:01 avg_step=3.66s recent_step=2.86s eta=29:57:11 +2026-09-15 10:40:19 | INFO | train rank=0 | step=20550/50000 progress=41.1% loss=0.091021 lr=6.840623e-05 lr_backbone=6.840623e-05 elapsed=20:53:52 avg_step=3.66s recent_step=5.05s eta=29:56:54 +2026-09-15 10:40:57 | INFO | train rank=0 | step=20560/50000 progress=41.1% loss=0.112559 lr=6.837548e-05 lr_backbone=6.837548e-05 elapsed=20:54:30 avg_step=3.66s recent_step=3.77s eta=29:56:19 +2026-09-15 10:41:30 | INFO | train rank=0 | step=20570/50000 progress=41.1% loss=0.120454 lr=6.834472e-05 lr_backbone=6.834472e-05 elapsed=20:55:03 avg_step=3.66s recent_step=3.36s eta=29:55:38 +2026-09-15 10:42:03 | INFO | train rank=0 | step=20580/50000 progress=41.2% loss=0.110410 lr=6.831395e-05 lr_backbone=6.831395e-05 elapsed=20:55:36 avg_step=3.66s recent_step=3.26s eta=29:54:56 +2026-09-15 10:42:42 | INFO | train rank=0 | step=20590/50000 progress=41.2% loss=0.074169 lr=6.828317e-05 lr_backbone=6.828317e-05 elapsed=20:56:15 avg_step=3.66s recent_step=3.95s eta=29:54:23 +2026-09-15 10:43:20 | INFO | train rank=0 | step=20600/50000 progress=41.2% loss=0.087606 lr=6.825239e-05 lr_backbone=6.825239e-05 elapsed=20:56:52 avg_step=3.66s recent_step=3.71s eta=29:53:48 +2026-09-15 10:44:00 | INFO | train rank=0 | step=20610/50000 progress=41.2% loss=0.122927 lr=6.822160e-05 lr_backbone=6.822160e-05 elapsed=20:57:33 avg_step=3.66s recent_step=4.06s eta=29:53:17 +2026-09-15 10:44:31 | INFO | train rank=0 | step=20620/50000 progress=41.2% loss=0.096688 lr=6.819080e-05 lr_backbone=6.819080e-05 elapsed=20:58:04 avg_step=3.66s recent_step=3.12s eta=29:52:32 +2026-09-15 10:45:07 | INFO | train rank=0 | step=20630/50000 progress=41.3% loss=0.079165 lr=6.815999e-05 lr_backbone=6.815999e-05 elapsed=20:58:40 avg_step=3.66s recent_step=3.58s eta=29:51:54 +2026-09-15 10:45:44 | INFO | train rank=0 | step=20640/50000 progress=41.3% loss=0.083978 lr=6.812918e-05 lr_backbone=6.812918e-05 elapsed=20:59:17 avg_step=3.66s recent_step=3.73s eta=29:51:19 +2026-09-15 10:46:25 | INFO | train rank=0 | step=20650/50000 progress=41.3% loss=0.123104 lr=6.809836e-05 lr_backbone=6.809836e-05 elapsed=20:59:58 avg_step=3.66s recent_step=4.06s eta=29:50:48 +2026-09-15 10:46:57 | INFO | train rank=0 | step=20660/50000 progress=41.3% loss=0.092133 lr=6.806752e-05 lr_backbone=6.806752e-05 elapsed=21:00:29 avg_step=3.66s recent_step=3.16s eta=29:50:04 +2026-09-15 10:47:46 | INFO | train rank=0 | step=20670/50000 progress=41.3% loss=0.119285 lr=6.803669e-05 lr_backbone=6.803669e-05 elapsed=21:01:19 avg_step=3.66s recent_step=4.98s eta=29:49:46 +2026-09-15 10:48:17 | INFO | train rank=0 | step=20680/50000 progress=41.4% loss=0.094719 lr=6.800584e-05 lr_backbone=6.800584e-05 elapsed=21:01:50 avg_step=3.66s recent_step=3.09s eta=29:49:02 +2026-09-15 10:48:51 | INFO | train rank=0 | step=20690/50000 progress=41.4% loss=0.111489 lr=6.797498e-05 lr_backbone=6.797498e-05 elapsed=21:02:24 avg_step=3.66s recent_step=3.39s eta=29:48:21 +2026-09-15 10:49:26 | INFO | train rank=0 | step=20700/50000 progress=41.4% loss=0.088983 lr=6.794412e-05 lr_backbone=6.794412e-05 elapsed=21:02:59 avg_step=3.66s recent_step=3.52s eta=29:47:43 +2026-09-15 10:50:02 | INFO | train rank=0 | step=20710/50000 progress=41.4% loss=0.142463 lr=6.791325e-05 lr_backbone=6.791325e-05 elapsed=21:03:35 avg_step=3.66s recent_step=3.55s eta=29:47:04 +2026-09-15 10:50:39 | INFO | train rank=0 | step=20720/50000 progress=41.4% loss=0.079058 lr=6.788237e-05 lr_backbone=6.788237e-05 elapsed=21:04:11 avg_step=3.66s recent_step=3.66s eta=29:46:28 +2026-09-15 10:51:09 | INFO | train rank=0 | step=20730/50000 progress=41.5% loss=0.139858 lr=6.785149e-05 lr_backbone=6.785149e-05 elapsed=21:04:42 avg_step=3.66s recent_step=3.03s eta=29:45:42 +2026-09-15 10:51:47 | INFO | train rank=0 | step=20740/50000 progress=41.5% loss=0.142755 lr=6.782059e-05 lr_backbone=6.782059e-05 elapsed=21:05:20 avg_step=3.66s recent_step=3.80s eta=29:45:08 +2026-09-15 10:52:39 | INFO | train rank=0 | step=20750/50000 progress=41.5% loss=0.104059 lr=6.778969e-05 lr_backbone=6.778969e-05 elapsed=21:06:12 avg_step=3.66s recent_step=5.25s eta=29:44:54 +2026-09-15 10:53:06 | INFO | train rank=0 | step=20760/50000 progress=41.5% loss=0.116024 lr=6.775878e-05 lr_backbone=6.775878e-05 elapsed=21:06:39 avg_step=3.66s recent_step=2.69s eta=29:44:03 +2026-09-15 10:53:42 | INFO | train rank=0 | step=20770/50000 progress=41.5% loss=0.102413 lr=6.772787e-05 lr_backbone=6.772787e-05 elapsed=21:07:15 avg_step=3.66s recent_step=3.58s eta=29:43:26 +2026-09-15 10:54:18 | INFO | train rank=0 | step=20780/50000 progress=41.6% loss=0.086988 lr=6.769694e-05 lr_backbone=6.769694e-05 elapsed=21:07:51 avg_step=3.66s recent_step=3.60s eta=29:42:48 +2026-09-15 10:54:58 | INFO | train rank=0 | step=20790/50000 progress=41.6% loss=0.093636 lr=6.766601e-05 lr_backbone=6.766601e-05 elapsed=21:08:31 avg_step=3.66s recent_step=4.00s eta=29:42:16 +2026-09-15 10:55:29 | INFO | train rank=0 | step=20800/50000 progress=41.6% loss=0.076095 lr=6.763507e-05 lr_backbone=6.763507e-05 elapsed=21:09:01 avg_step=3.66s recent_step=3.04s eta=29:41:31 +2026-09-15 10:55:59 | INFO | train rank=0 | step=20810/50000 progress=41.6% loss=0.115096 lr=6.760412e-05 lr_backbone=6.760412e-05 elapsed=21:09:32 avg_step=3.66s recent_step=3.08s eta=29:40:46 +2026-09-15 10:56:44 | INFO | train rank=0 | step=20820/50000 progress=41.6% loss=0.082854 lr=6.757316e-05 lr_backbone=6.757316e-05 elapsed=21:10:16 avg_step=3.66s recent_step=4.44s eta=29:40:20 +2026-09-15 10:57:21 | INFO | train rank=0 | step=20830/50000 progress=41.7% loss=0.124703 lr=6.754220e-05 lr_backbone=6.754220e-05 elapsed=21:10:54 avg_step=3.66s recent_step=3.71s eta=29:39:45 +2026-09-15 10:57:52 | INFO | train rank=0 | step=20840/50000 progress=41.7% loss=0.080077 lr=6.751123e-05 lr_backbone=6.751123e-05 elapsed=21:11:25 avg_step=3.66s recent_step=3.13s eta=29:39:01 +2026-09-15 10:58:30 | INFO | train rank=0 | step=20850/50000 progress=41.7% loss=0.093336 lr=6.748025e-05 lr_backbone=6.748025e-05 elapsed=21:12:02 avg_step=3.66s recent_step=3.74s eta=29:38:25 +2026-09-15 10:59:18 | INFO | train rank=0 | step=20860/50000 progress=41.7% loss=0.087451 lr=6.744926e-05 lr_backbone=6.744926e-05 elapsed=21:12:51 avg_step=3.66s recent_step=4.84s eta=29:38:05 +2026-09-15 10:59:54 | INFO | train rank=0 | step=20870/50000 progress=41.7% loss=0.093583 lr=6.741827e-05 lr_backbone=6.741827e-05 elapsed=21:13:27 avg_step=3.66s recent_step=3.60s eta=29:37:27 +2026-09-15 11:00:31 | INFO | train rank=0 | step=20880/50000 progress=41.8% loss=0.109822 lr=6.738727e-05 lr_backbone=6.738727e-05 elapsed=21:14:04 avg_step=3.66s recent_step=3.74s eta=29:36:52 +2026-09-15 11:01:05 | INFO | train rank=0 | step=20890/50000 progress=41.8% loss=0.100355 lr=6.735626e-05 lr_backbone=6.735626e-05 elapsed=21:14:38 avg_step=3.66s recent_step=3.41s eta=29:36:12 +2026-09-15 11:01:53 | INFO | train rank=0 | step=20900/50000 progress=41.8% loss=0.095016 lr=6.732524e-05 lr_backbone=6.732524e-05 elapsed=21:15:26 avg_step=3.66s recent_step=4.77s eta=29:35:51 +2026-09-15 11:02:32 | INFO | train rank=0 | step=20910/50000 progress=41.8% loss=0.139372 lr=6.729422e-05 lr_backbone=6.729422e-05 elapsed=21:16:05 avg_step=3.66s recent_step=3.87s eta=29:35:17 +2026-09-15 11:02:59 | INFO | train rank=0 | step=20920/50000 progress=41.8% loss=0.115717 lr=6.726319e-05 lr_backbone=6.726319e-05 elapsed=21:16:32 avg_step=3.66s recent_step=2.75s eta=29:34:28 +2026-09-15 11:03:33 | INFO | train rank=0 | step=20930/50000 progress=41.9% loss=0.085657 lr=6.723215e-05 lr_backbone=6.723215e-05 elapsed=21:17:05 avg_step=3.66s recent_step=3.33s eta=29:33:46 +2026-09-15 11:04:24 | INFO | train rank=0 | step=20940/50000 progress=41.9% loss=0.132066 lr=6.720110e-05 lr_backbone=6.720110e-05 elapsed=21:17:57 avg_step=3.66s recent_step=5.14s eta=29:33:30 +2026-09-15 11:04:52 | INFO | train rank=0 | step=20950/50000 progress=41.9% loss=0.099026 lr=6.717004e-05 lr_backbone=6.717004e-05 elapsed=21:18:25 avg_step=3.66s recent_step=2.84s eta=29:32:42 +2026-09-15 11:05:28 | INFO | train rank=0 | step=20960/50000 progress=41.9% loss=0.146653 lr=6.713898e-05 lr_backbone=6.713898e-05 elapsed=21:19:00 avg_step=3.66s recent_step=3.52s eta=29:32:04 +2026-09-15 11:05:58 | INFO | train rank=0 | step=20970/50000 progress=41.9% loss=0.120860 lr=6.710791e-05 lr_backbone=6.710791e-05 elapsed=21:19:31 avg_step=3.66s recent_step=3.04s eta=29:31:18 +2026-09-15 11:06:51 | INFO | train rank=0 | step=20980/50000 progress=42.0% loss=0.074024 lr=6.707684e-05 lr_backbone=6.707684e-05 elapsed=21:20:24 avg_step=3.66s recent_step=5.28s eta=29:31:04 +2026-09-15 11:07:21 | INFO | train rank=0 | step=20990/50000 progress=42.0% loss=0.103540 lr=6.704575e-05 lr_backbone=6.704575e-05 elapsed=21:20:54 avg_step=3.66s recent_step=3.03s eta=29:30:19 +2026-09-15 11:07:55 | INFO | train rank=0 | step=21000/50000 progress=42.0% loss=0.117775 lr=6.701466e-05 lr_backbone=6.701466e-05 elapsed=21:21:28 avg_step=3.66s recent_step=3.41s eta=29:29:39 +2026-09-15 11:08:32 | INFO | train rank=0 | step=21010/50000 progress=42.0% loss=0.097308 lr=6.698356e-05 lr_backbone=6.698356e-05 elapsed=21:22:05 avg_step=3.66s recent_step=3.68s eta=29:29:03 +2026-09-15 11:09:17 | INFO | train rank=0 | step=21020/50000 progress=42.0% loss=0.132117 lr=6.695245e-05 lr_backbone=6.695245e-05 elapsed=21:22:50 avg_step=3.66s recent_step=4.55s eta=29:28:38 +2026-09-15 11:09:57 | INFO | train rank=0 | step=21030/50000 progress=42.1% loss=0.099254 lr=6.692134e-05 lr_backbone=6.692134e-05 elapsed=21:23:30 avg_step=3.66s recent_step=4.00s eta=29:28:06 +2026-09-15 11:10:33 | INFO | train rank=0 | step=21040/50000 progress=42.1% loss=0.076467 lr=6.689022e-05 lr_backbone=6.689022e-05 elapsed=21:24:05 avg_step=3.66s recent_step=3.50s eta=29:27:27 +2026-09-15 11:11:08 | INFO | train rank=0 | step=21050/50000 progress=42.1% loss=0.101815 lr=6.685909e-05 lr_backbone=6.685909e-05 elapsed=21:24:41 avg_step=3.66s recent_step=3.56s eta=29:26:49 +2026-09-15 11:11:58 | INFO | train rank=0 | step=21060/50000 progress=42.1% loss=0.105577 lr=6.682795e-05 lr_backbone=6.682795e-05 elapsed=21:25:30 avg_step=3.66s recent_step=4.95s eta=29:26:30 +2026-09-15 11:12:34 | INFO | train rank=0 | step=21070/50000 progress=42.1% loss=0.088698 lr=6.679681e-05 lr_backbone=6.679681e-05 elapsed=21:26:07 avg_step=3.66s recent_step=3.62s eta=29:25:53 +2026-09-15 11:13:09 | INFO | train rank=0 | step=21080/50000 progress=42.2% loss=0.098467 lr=6.676566e-05 lr_backbone=6.676566e-05 elapsed=21:26:42 avg_step=3.66s recent_step=3.50s eta=29:25:14 +2026-09-15 11:13:49 | INFO | train rank=0 | step=21090/50000 progress=42.2% loss=0.118005 lr=6.673450e-05 lr_backbone=6.673450e-05 elapsed=21:27:22 avg_step=3.66s recent_step=4.03s eta=29:24:43 +2026-09-15 11:14:38 | INFO | train rank=0 | step=21100/50000 progress=42.2% loss=0.108362 lr=6.670333e-05 lr_backbone=6.670333e-05 elapsed=21:28:11 avg_step=3.66s recent_step=4.91s eta=29:24:23 +2026-09-15 11:15:05 | INFO | train rank=0 | step=21110/50000 progress=42.2% loss=0.123668 lr=6.667216e-05 lr_backbone=6.667216e-05 elapsed=21:28:37 avg_step=3.66s recent_step=2.64s eta=29:23:33 +2026-09-15 11:15:35 | INFO | train rank=0 | step=21120/50000 progress=42.2% loss=0.081045 lr=6.664098e-05 lr_backbone=6.664098e-05 elapsed=21:29:07 avg_step=3.66s recent_step=2.99s eta=29:22:47 +2026-09-15 11:16:11 | INFO | train rank=0 | step=21130/50000 progress=42.3% loss=0.136112 lr=6.660979e-05 lr_backbone=6.660979e-05 elapsed=21:29:43 avg_step=3.66s recent_step=3.61s eta=29:22:10 +2026-09-15 11:16:52 | INFO | train rank=0 | step=21140/50000 progress=42.3% loss=0.092106 lr=6.657859e-05 lr_backbone=6.657859e-05 elapsed=21:30:25 avg_step=3.66s recent_step=4.11s eta=29:21:39 +2026-09-15 11:17:21 | INFO | train rank=0 | step=21150/50000 progress=42.3% loss=0.099575 lr=6.654739e-05 lr_backbone=6.654739e-05 elapsed=21:30:54 avg_step=3.66s recent_step=2.91s eta=29:20:52 +2026-09-15 11:18:01 | INFO | train rank=0 | step=21160/50000 progress=42.3% loss=0.132187 lr=6.651618e-05 lr_backbone=6.651618e-05 elapsed=21:31:34 avg_step=3.66s recent_step=4.00s eta=29:20:20 +2026-09-15 11:18:36 | INFO | train rank=0 | step=21170/50000 progress=42.3% loss=0.090017 lr=6.648497e-05 lr_backbone=6.648497e-05 elapsed=21:32:09 avg_step=3.66s recent_step=3.56s eta=29:19:42 +2026-09-15 11:19:27 | INFO | train rank=0 | step=21180/50000 progress=42.4% loss=0.091312 lr=6.645374e-05 lr_backbone=6.645374e-05 elapsed=21:33:00 avg_step=3.66s recent_step=5.04s eta=29:19:24 +2026-09-15 11:20:00 | INFO | train rank=0 | step=21190/50000 progress=42.4% loss=0.086709 lr=6.642251e-05 lr_backbone=6.642251e-05 elapsed=21:33:33 avg_step=3.66s recent_step=3.32s eta=29:18:43 +2026-09-15 11:20:37 | INFO | train rank=0 | step=21200/50000 progress=42.4% loss=0.086923 lr=6.639127e-05 lr_backbone=6.639127e-05 elapsed=21:34:10 avg_step=3.66s recent_step=3.72s eta=29:18:07 +2026-09-15 11:21:14 | INFO | train rank=0 | step=21210/50000 progress=42.4% loss=0.125675 lr=6.636003e-05 lr_backbone=6.636003e-05 elapsed=21:34:47 avg_step=3.66s recent_step=3.69s eta=29:17:31 +2026-09-15 11:22:00 | INFO | train rank=0 | step=21220/50000 progress=42.4% loss=0.069448 lr=6.632877e-05 lr_backbone=6.632877e-05 elapsed=21:35:33 avg_step=3.66s recent_step=4.60s eta=29:17:07 +2026-09-15 11:22:33 | INFO | train rank=0 | step=21230/50000 progress=42.5% loss=0.095999 lr=6.629751e-05 lr_backbone=6.629751e-05 elapsed=21:36:05 avg_step=3.66s recent_step=3.23s eta=29:16:24 +2026-09-15 11:23:05 | INFO | train rank=0 | step=21240/50000 progress=42.5% loss=0.131998 lr=6.626625e-05 lr_backbone=6.626625e-05 elapsed=21:36:38 avg_step=3.66s recent_step=3.24s eta=29:15:42 +2026-09-15 11:23:36 | INFO | train rank=0 | step=21250/50000 progress=42.5% loss=0.097916 lr=6.623497e-05 lr_backbone=6.623497e-05 elapsed=21:37:08 avg_step=3.66s recent_step=3.06s eta=29:14:57 +2026-09-15 11:24:15 | INFO | train rank=0 | step=21260/50000 progress=42.5% loss=0.132156 lr=6.620369e-05 lr_backbone=6.620369e-05 elapsed=21:37:48 avg_step=3.66s recent_step=3.97s eta=29:14:25 +2026-09-15 11:24:42 | INFO | train rank=0 | step=21270/50000 progress=42.5% loss=0.112099 lr=6.617240e-05 lr_backbone=6.617240e-05 elapsed=21:38:15 avg_step=3.66s recent_step=2.68s eta=29:13:35 +2026-09-15 11:25:18 | INFO | train rank=0 | step=21280/50000 progress=42.6% loss=0.113242 lr=6.614111e-05 lr_backbone=6.614111e-05 elapsed=21:38:51 avg_step=3.66s recent_step=3.63s eta=29:12:58 +2026-09-15 11:25:49 | INFO | train rank=0 | step=21290/50000 progress=42.6% loss=0.108459 lr=6.610981e-05 lr_backbone=6.610981e-05 elapsed=21:39:21 avg_step=3.66s recent_step=3.02s eta=29:12:13 +2026-09-15 11:26:41 | INFO | train rank=0 | step=21300/50000 progress=42.6% loss=0.094528 lr=6.607850e-05 lr_backbone=6.607850e-05 elapsed=21:40:14 avg_step=3.66s recent_step=5.23s eta=29:11:57 +2026-09-15 11:27:21 | INFO | train rank=0 | step=21310/50000 progress=42.6% loss=0.109027 lr=6.604718e-05 lr_backbone=6.604718e-05 elapsed=21:40:54 avg_step=3.66s recent_step=4.01s eta=29:11:25 +2026-09-15 11:27:55 | INFO | train rank=0 | step=21320/50000 progress=42.6% loss=0.090103 lr=6.601586e-05 lr_backbone=6.601586e-05 elapsed=21:41:28 avg_step=3.66s recent_step=3.38s eta=29:10:45 +2026-09-15 11:28:31 | INFO | train rank=0 | step=21330/50000 progress=42.7% loss=0.115816 lr=6.598453e-05 lr_backbone=6.598453e-05 elapsed=21:42:03 avg_step=3.66s recent_step=3.58s eta=29:10:07 +2026-09-15 11:29:23 | INFO | train rank=0 | step=21340/50000 progress=42.7% loss=0.098410 lr=6.595319e-05 lr_backbone=6.595319e-05 elapsed=21:42:56 avg_step=3.66s recent_step=5.22s eta=29:09:51 +2026-09-15 11:30:02 | INFO | train rank=0 | step=21350/50000 progress=42.7% loss=0.128294 lr=6.592184e-05 lr_backbone=6.592184e-05 elapsed=21:43:34 avg_step=3.66s recent_step=3.88s eta=29:09:18 +2026-09-15 11:30:28 | INFO | train rank=0 | step=21360/50000 progress=42.7% loss=0.102414 lr=6.589049e-05 lr_backbone=6.589049e-05 elapsed=21:44:01 avg_step=3.66s recent_step=2.64s eta=29:08:27 +2026-09-15 11:31:01 | INFO | train rank=0 | step=21370/50000 progress=42.7% loss=0.114185 lr=6.585913e-05 lr_backbone=6.585913e-05 elapsed=21:44:34 avg_step=3.66s recent_step=3.33s eta=29:07:46 +2026-09-15 11:31:36 | INFO | train rank=0 | step=21380/50000 progress=42.8% loss=0.111942 lr=6.582777e-05 lr_backbone=6.582777e-05 elapsed=21:45:09 avg_step=3.66s recent_step=3.50s eta=29:07:07 +2026-09-15 11:32:01 | INFO | train rank=0 | step=21390/50000 progress=42.8% loss=0.107227 lr=6.579640e-05 lr_backbone=6.579640e-05 elapsed=21:45:34 avg_step=3.66s recent_step=2.45s eta=29:06:15 +2026-09-15 11:32:38 | INFO | train rank=0 | step=21400/50000 progress=42.8% loss=0.098522 lr=6.576502e-05 lr_backbone=6.576502e-05 elapsed=21:46:10 avg_step=3.66s recent_step=3.68s eta=29:05:38 +2026-09-15 11:33:06 | INFO | train rank=0 | step=21410/50000 progress=42.8% loss=0.067042 lr=6.573363e-05 lr_backbone=6.573363e-05 elapsed=21:46:39 avg_step=3.66s recent_step=2.86s eta=29:04:51 +2026-09-15 11:33:54 | INFO | train rank=0 | step=21420/50000 progress=42.8% loss=0.124098 lr=6.570224e-05 lr_backbone=6.570224e-05 elapsed=21:47:27 avg_step=3.66s recent_step=4.76s eta=29:04:29 +2026-09-15 11:34:22 | INFO | train rank=0 | step=21430/50000 progress=42.9% loss=0.120280 lr=6.567084e-05 lr_backbone=6.567084e-05 elapsed=21:47:55 avg_step=3.66s recent_step=2.84s eta=29:03:41 +2026-09-15 11:34:54 | INFO | train rank=0 | step=21440/50000 progress=42.9% loss=0.104389 lr=6.563943e-05 lr_backbone=6.563943e-05 elapsed=21:48:27 avg_step=3.66s recent_step=3.17s eta=29:02:58 +2026-09-15 11:35:24 | INFO | train rank=0 | step=21450/50000 progress=42.9% loss=0.111608 lr=6.560802e-05 lr_backbone=6.560802e-05 elapsed=21:48:57 avg_step=3.66s recent_step=3.01s eta=29:02:13 +2026-09-15 11:35:59 | INFO | train rank=0 | step=21460/50000 progress=42.9% loss=0.098262 lr=6.557660e-05 lr_backbone=6.557660e-05 elapsed=21:49:32 avg_step=3.66s recent_step=3.54s eta=29:01:34 +2026-09-15 11:36:53 | INFO | train rank=0 | step=21470/50000 progress=42.9% loss=0.082381 lr=6.554517e-05 lr_backbone=6.554517e-05 elapsed=21:50:26 avg_step=3.66s recent_step=5.40s eta=29:01:21 +2026-09-15 11:37:22 | INFO | train rank=0 | step=21480/50000 progress=43.0% loss=0.105761 lr=6.551374e-05 lr_backbone=6.551374e-05 elapsed=21:50:54 avg_step=3.66s recent_step=2.82s eta=29:00:33 +2026-09-15 11:37:52 | INFO | train rank=0 | step=21490/50000 progress=43.0% loss=0.116125 lr=6.548230e-05 lr_backbone=6.548230e-05 elapsed=21:51:25 avg_step=3.66s recent_step=3.09s eta=28:59:49 +2026-09-15 11:38:35 | INFO | train rank=0 | step=21500/50000 progress=43.0% loss=0.092399 lr=6.545085e-05 lr_backbone=6.545085e-05 elapsed=21:52:07 avg_step=3.66s recent_step=4.21s eta=28:59:20 +2026-09-15 11:39:11 | INFO | train rank=0 | step=21510/50000 progress=43.0% loss=0.120821 lr=6.541940e-05 lr_backbone=6.541940e-05 elapsed=21:52:44 avg_step=3.66s recent_step=3.67s eta=28:58:43 +2026-09-15 11:39:46 | INFO | train rank=0 | step=21520/50000 progress=43.0% loss=0.097136 lr=6.538793e-05 lr_backbone=6.538793e-05 elapsed=21:53:19 avg_step=3.66s recent_step=3.49s eta=28:58:04 +2026-09-15 11:40:20 | INFO | train rank=0 | step=21530/50000 progress=43.1% loss=0.088155 lr=6.535647e-05 lr_backbone=6.535647e-05 elapsed=21:53:53 avg_step=3.66s recent_step=3.39s eta=28:57:24 +2026-09-15 11:41:02 | INFO | train rank=0 | step=21540/50000 progress=43.1% loss=0.105083 lr=6.532499e-05 lr_backbone=6.532499e-05 elapsed=21:54:35 avg_step=3.66s recent_step=4.23s eta=28:56:55 +2026-09-15 11:41:52 | INFO | train rank=0 | step=21550/50000 progress=43.1% loss=0.074180 lr=6.529351e-05 lr_backbone=6.529351e-05 elapsed=21:55:25 avg_step=3.66s recent_step=5.01s eta=28:56:36 +2026-09-15 11:42:15 | INFO | train rank=0 | step=21560/50000 progress=43.1% loss=0.073960 lr=6.526202e-05 lr_backbone=6.526202e-05 elapsed=21:55:48 avg_step=3.66s recent_step=2.28s eta=28:55:41 +2026-09-15 11:42:44 | INFO | train rank=0 | step=21570/50000 progress=43.1% loss=0.111391 lr=6.523053e-05 lr_backbone=6.523053e-05 elapsed=21:56:17 avg_step=3.66s recent_step=2.92s eta=28:54:55 +2026-09-15 11:43:27 | INFO | train rank=0 | step=21580/50000 progress=43.2% loss=0.090028 lr=6.519903e-05 lr_backbone=6.519903e-05 elapsed=21:57:00 avg_step=3.66s recent_step=4.26s eta=28:54:26 +2026-09-15 11:44:10 | INFO | train rank=0 | step=21590/50000 progress=43.2% loss=0.088242 lr=6.516752e-05 lr_backbone=6.516752e-05 elapsed=21:57:42 avg_step=3.66s recent_step=4.26s eta=28:53:57 +2026-09-15 11:44:43 | INFO | train rank=0 | step=21600/50000 progress=43.2% loss=0.075308 lr=6.513601e-05 lr_backbone=6.513601e-05 elapsed=21:58:16 avg_step=3.66s recent_step=3.30s eta=28:53:16 +2026-09-15 11:45:23 | INFO | train rank=0 | step=21610/50000 progress=43.2% loss=0.088334 lr=6.510448e-05 lr_backbone=6.510448e-05 elapsed=21:58:56 avg_step=3.66s recent_step=4.02s eta=28:52:44 +2026-09-15 11:45:54 | INFO | train rank=0 | step=21620/50000 progress=43.2% loss=0.097087 lr=6.507296e-05 lr_backbone=6.507296e-05 elapsed=21:59:27 avg_step=3.66s recent_step=3.08s eta=28:52:00 +2026-09-15 11:46:32 | INFO | train rank=0 | step=21630/50000 progress=43.3% loss=0.094317 lr=6.504142e-05 lr_backbone=6.504142e-05 elapsed=22:00:05 avg_step=3.66s recent_step=3.82s eta=28:51:25 +2026-09-15 11:47:08 | INFO | train rank=0 | step=21640/50000 progress=43.3% loss=0.113056 lr=6.500988e-05 lr_backbone=6.500988e-05 elapsed=22:00:40 avg_step=3.66s recent_step=3.56s eta=28:50:47 +2026-09-15 11:47:44 | INFO | train rank=0 | step=21650/50000 progress=43.3% loss=0.099180 lr=6.497833e-05 lr_backbone=6.497833e-05 elapsed=22:01:17 avg_step=3.66s recent_step=3.67s eta=28:50:11 +2026-09-15 11:48:26 | INFO | train rank=0 | step=21660/50000 progress=43.3% loss=0.093250 lr=6.494678e-05 lr_backbone=6.494678e-05 elapsed=22:01:58 avg_step=3.66s recent_step=4.13s eta=28:49:41 +2026-09-15 11:49:07 | INFO | train rank=0 | step=21670/50000 progress=43.3% loss=0.101060 lr=6.491522e-05 lr_backbone=6.491522e-05 elapsed=22:02:40 avg_step=3.66s recent_step=4.12s eta=28:49:10 +2026-09-15 11:49:39 | INFO | train rank=0 | step=21680/50000 progress=43.4% loss=0.115814 lr=6.488365e-05 lr_backbone=6.488365e-05 elapsed=22:03:12 avg_step=3.66s recent_step=3.21s eta=28:48:27 +2026-09-15 11:50:17 | INFO | train rank=0 | step=21690/50000 progress=43.4% loss=0.106347 lr=6.485208e-05 lr_backbone=6.485208e-05 elapsed=22:03:50 avg_step=3.66s recent_step=3.85s eta=28:47:53 +2026-09-15 11:51:00 | INFO | train rank=0 | step=21700/50000 progress=43.4% loss=0.088415 lr=6.482050e-05 lr_backbone=6.482050e-05 elapsed=22:04:33 avg_step=3.66s recent_step=4.31s eta=28:47:25 +2026-09-15 11:51:46 | INFO | train rank=0 | step=21710/50000 progress=43.4% loss=0.165083 lr=6.478891e-05 lr_backbone=6.478891e-05 elapsed=22:05:18 avg_step=3.66s recent_step=4.52s eta=28:47:00 +2026-09-15 11:52:27 | INFO | train rank=0 | step=21720/50000 progress=43.4% loss=0.085463 lr=6.475732e-05 lr_backbone=6.475732e-05 elapsed=22:05:59 avg_step=3.66s recent_step=4.09s eta=28:46:29 +2026-09-15 11:52:58 | INFO | train rank=0 | step=21730/50000 progress=43.5% loss=0.077552 lr=6.472572e-05 lr_backbone=6.472572e-05 elapsed=22:06:31 avg_step=3.66s recent_step=3.13s eta=28:45:45 +2026-09-15 11:53:39 | INFO | train rank=0 | step=21740/50000 progress=43.5% loss=0.126152 lr=6.469411e-05 lr_backbone=6.469411e-05 elapsed=22:07:11 avg_step=3.66s recent_step=4.07s eta=28:45:14 +2026-09-15 11:54:24 | INFO | train rank=0 | step=21750/50000 progress=43.5% loss=0.125828 lr=6.466250e-05 lr_backbone=6.466250e-05 elapsed=22:07:56 avg_step=3.66s recent_step=4.50s eta=28:44:48 +2026-09-15 11:54:55 | INFO | train rank=0 | step=21760/50000 progress=43.5% loss=0.098872 lr=6.463088e-05 lr_backbone=6.463088e-05 elapsed=22:08:28 avg_step=3.66s recent_step=3.15s eta=28:44:05 +2026-09-15 11:55:31 | INFO | train rank=0 | step=21770/50000 progress=43.5% loss=0.102513 lr=6.459926e-05 lr_backbone=6.459926e-05 elapsed=22:09:03 avg_step=3.66s recent_step=3.54s eta=28:43:26 +2026-09-15 11:56:09 | INFO | train rank=0 | step=21780/50000 progress=43.6% loss=0.134475 lr=6.456763e-05 lr_backbone=6.456763e-05 elapsed=22:09:41 avg_step=3.66s recent_step=3.80s eta=28:42:52 +2026-09-15 11:56:52 | INFO | train rank=0 | step=21790/50000 progress=43.6% loss=0.090750 lr=6.453599e-05 lr_backbone=6.453599e-05 elapsed=22:10:25 avg_step=3.66s recent_step=4.34s eta=28:42:24 +2026-09-15 11:57:29 | INFO | train rank=0 | step=21800/50000 progress=43.6% loss=0.107195 lr=6.450434e-05 lr_backbone=6.450434e-05 elapsed=22:11:02 avg_step=3.66s recent_step=3.72s eta=28:41:48 +2026-09-15 11:57:58 | INFO | train rank=0 | step=21810/50000 progress=43.6% loss=0.140866 lr=6.447269e-05 lr_backbone=6.447269e-05 elapsed=22:11:30 avg_step=3.66s recent_step=2.84s eta=28:41:01 +2026-09-15 11:58:36 | INFO | train rank=0 | step=21820/50000 progress=43.6% loss=0.109739 lr=6.444104e-05 lr_backbone=6.444104e-05 elapsed=22:12:09 avg_step=3.66s recent_step=3.81s eta=28:40:26 +2026-09-15 11:59:24 | INFO | train rank=0 | step=21830/50000 progress=43.7% loss=0.079170 lr=6.440937e-05 lr_backbone=6.440937e-05 elapsed=22:12:57 avg_step=3.66s recent_step=4.85s eta=28:40:05 +2026-09-15 12:00:01 | INFO | train rank=0 | step=21840/50000 progress=43.7% loss=0.109294 lr=6.437770e-05 lr_backbone=6.437770e-05 elapsed=22:13:34 avg_step=3.66s recent_step=3.69s eta=28:39:28 +2026-09-15 12:00:39 | INFO | train rank=0 | step=21850/50000 progress=43.7% loss=0.086177 lr=6.434603e-05 lr_backbone=6.434603e-05 elapsed=22:14:12 avg_step=3.66s recent_step=3.82s eta=28:38:54 +2026-09-15 12:01:14 | INFO | train rank=0 | step=21860/50000 progress=43.7% loss=0.079075 lr=6.431435e-05 lr_backbone=6.431435e-05 elapsed=22:14:47 avg_step=3.66s recent_step=3.45s eta=28:38:14 +2026-09-15 12:02:02 | INFO | train rank=0 | step=21870/50000 progress=43.7% loss=0.080789 lr=6.428266e-05 lr_backbone=6.428266e-05 elapsed=22:15:34 avg_step=3.66s recent_step=4.77s eta=28:37:52 +2026-09-15 12:02:27 | INFO | train rank=0 | step=21880/50000 progress=43.8% loss=0.096614 lr=6.425096e-05 lr_backbone=6.425096e-05 elapsed=22:15:59 avg_step=3.66s recent_step=2.51s eta=28:37:00 +2026-09-15 12:02:54 | INFO | train rank=0 | step=21890/50000 progress=43.8% loss=0.092426 lr=6.421926e-05 lr_backbone=6.421926e-05 elapsed=22:16:27 avg_step=3.66s recent_step=2.72s eta=28:36:12 +2026-09-15 12:03:29 | INFO | train rank=0 | step=21900/50000 progress=43.8% loss=0.091512 lr=6.418756e-05 lr_backbone=6.418756e-05 elapsed=22:17:02 avg_step=3.66s recent_step=3.51s eta=28:35:33 +2026-09-15 12:04:22 | INFO | train rank=0 | step=21910/50000 progress=43.8% loss=0.127145 lr=6.415584e-05 lr_backbone=6.415584e-05 elapsed=22:17:55 avg_step=3.66s recent_step=5.30s eta=28:35:17 +2026-09-15 12:04:56 | INFO | train rank=0 | step=21920/50000 progress=43.8% loss=0.092815 lr=6.412412e-05 lr_backbone=6.412412e-05 elapsed=22:18:28 avg_step=3.66s recent_step=3.36s eta=28:34:37 +2026-09-15 12:05:35 | INFO | train rank=0 | step=21930/50000 progress=43.9% loss=0.092211 lr=6.409240e-05 lr_backbone=6.409240e-05 elapsed=22:19:07 avg_step=3.66s recent_step=3.91s eta=28:34:03 +2026-09-15 12:06:05 | INFO | train rank=0 | step=21940/50000 progress=43.9% loss=0.139146 lr=6.406066e-05 lr_backbone=6.406066e-05 elapsed=22:19:38 avg_step=3.66s recent_step=3.08s eta=28:33:19 +2026-09-15 12:06:43 | INFO | train rank=0 | step=21950/50000 progress=43.9% loss=0.075001 lr=6.402893e-05 lr_backbone=6.402893e-05 elapsed=22:20:16 avg_step=3.66s recent_step=3.78s eta=28:32:44 +2026-09-15 12:07:24 | INFO | train rank=0 | step=21960/50000 progress=43.9% loss=0.079325 lr=6.399718e-05 lr_backbone=6.399718e-05 elapsed=22:20:56 avg_step=3.66s recent_step=4.02s eta=28:32:12 +2026-09-15 12:07:54 | INFO | train rank=0 | step=21970/50000 progress=43.9% loss=0.101617 lr=6.396543e-05 lr_backbone=6.396543e-05 elapsed=22:21:27 avg_step=3.66s recent_step=3.07s eta=28:31:28 +2026-09-15 12:08:27 | INFO | train rank=0 | step=21980/50000 progress=44.0% loss=0.074086 lr=6.393368e-05 lr_backbone=6.393368e-05 elapsed=22:22:00 avg_step=3.66s recent_step=3.31s eta=28:30:47 +2026-09-15 12:09:04 | INFO | train rank=0 | step=21990/50000 progress=44.0% loss=0.084524 lr=6.390191e-05 lr_backbone=6.390191e-05 elapsed=22:22:37 avg_step=3.66s recent_step=3.65s eta=28:30:10 +2026-09-15 12:09:46 | INFO | train rank=0 | step=22000/50000 progress=44.0% loss=0.105852 lr=6.387015e-05 lr_backbone=6.387015e-05 elapsed=22:23:18 avg_step=3.66s recent_step=4.18s eta=28:29:40 +2026-09-15 12:10:18 | INFO | train rank=0 | step=22010/50000 progress=44.0% loss=0.079929 lr=6.383837e-05 lr_backbone=6.383837e-05 elapsed=22:23:51 avg_step=3.66s recent_step=3.28s eta=28:28:58 +2026-09-15 12:10:58 | INFO | train rank=0 | step=22020/50000 progress=44.0% loss=0.088342 lr=6.380659e-05 lr_backbone=6.380659e-05 elapsed=22:24:31 avg_step=3.66s recent_step=3.97s eta=28:28:26 +2026-09-15 12:11:49 | INFO | train rank=0 | step=22030/50000 progress=44.1% loss=0.105449 lr=6.377480e-05 lr_backbone=6.377480e-05 elapsed=22:25:22 avg_step=3.66s recent_step=5.13s eta=28:28:07 +2026-09-15 12:12:25 | INFO | train rank=0 | step=22040/50000 progress=44.1% loss=0.101430 lr=6.374301e-05 lr_backbone=6.374301e-05 elapsed=22:25:57 avg_step=3.66s recent_step=3.53s eta=28:27:29 +2026-09-15 12:13:02 | INFO | train rank=0 | step=22050/50000 progress=44.1% loss=0.152420 lr=6.371121e-05 lr_backbone=6.371121e-05 elapsed=22:26:35 avg_step=3.66s recent_step=3.78s eta=28:26:54 +2026-09-15 12:13:37 | INFO | train rank=0 | step=22060/50000 progress=44.1% loss=0.084520 lr=6.367941e-05 lr_backbone=6.367941e-05 elapsed=22:27:10 avg_step=3.66s recent_step=3.44s eta=28:26:15 +2026-09-15 12:14:30 | INFO | train rank=0 | step=22070/50000 progress=44.1% loss=0.078885 lr=6.364760e-05 lr_backbone=6.364760e-05 elapsed=22:28:02 avg_step=3.66s recent_step=5.28s eta=28:25:58 +2026-09-15 12:15:05 | INFO | train rank=0 | step=22080/50000 progress=44.2% loss=0.106947 lr=6.361578e-05 lr_backbone=6.361578e-05 elapsed=22:28:38 avg_step=3.66s recent_step=3.53s eta=28:25:20 +2026-09-15 12:15:39 | INFO | train rank=0 | step=22090/50000 progress=44.2% loss=0.120781 lr=6.358396e-05 lr_backbone=6.358396e-05 elapsed=22:29:12 avg_step=3.66s recent_step=3.38s eta=28:24:40 +2026-09-15 12:16:18 | INFO | train rank=0 | step=22100/50000 progress=44.2% loss=0.097733 lr=6.355213e-05 lr_backbone=6.355213e-05 elapsed=22:29:51 avg_step=3.66s recent_step=3.93s eta=28:24:06 +2026-09-15 12:17:09 | INFO | train rank=0 | step=22110/50000 progress=44.2% loss=0.088649 lr=6.352029e-05 lr_backbone=6.352029e-05 elapsed=22:30:42 avg_step=3.67s recent_step=5.12s eta=28:23:48 +2026-09-15 12:17:36 | INFO | train rank=0 | step=22120/50000 progress=44.2% loss=0.098537 lr=6.348845e-05 lr_backbone=6.348845e-05 elapsed=22:31:09 avg_step=3.66s recent_step=2.70s eta=28:22:59 +2026-09-15 12:18:09 | INFO | train rank=0 | step=22130/50000 progress=44.3% loss=0.085211 lr=6.345661e-05 lr_backbone=6.345661e-05 elapsed=22:31:42 avg_step=3.66s recent_step=3.25s eta=28:22:18 +2026-09-15 12:18:43 | INFO | train rank=0 | step=22140/50000 progress=44.3% loss=0.123633 lr=6.342476e-05 lr_backbone=6.342476e-05 elapsed=22:32:16 avg_step=3.66s recent_step=3.47s eta=28:21:38 +2026-09-15 12:19:30 | INFO | train rank=0 | step=22150/50000 progress=44.3% loss=0.109322 lr=6.339290e-05 lr_backbone=6.339290e-05 elapsed=22:33:02 avg_step=3.67s recent_step=4.61s eta=28:21:14 +2026-09-15 12:20:12 | INFO | train rank=0 | step=22160/50000 progress=44.3% loss=0.121351 lr=6.336103e-05 lr_backbone=6.336103e-05 elapsed=22:33:45 avg_step=3.67s recent_step=4.23s eta=28:20:44 +2026-09-15 12:20:49 | INFO | train rank=0 | step=22170/50000 progress=44.3% loss=0.078826 lr=6.332916e-05 lr_backbone=6.332916e-05 elapsed=22:34:21 avg_step=3.67s recent_step=3.68s eta=28:20:08 +2026-09-15 12:21:14 | INFO | train rank=0 | step=22180/50000 progress=44.4% loss=0.098768 lr=6.329729e-05 lr_backbone=6.329729e-05 elapsed=22:34:47 avg_step=3.66s recent_step=2.57s eta=28:19:17 +2026-09-15 12:22:03 | INFO | train rank=0 | step=22190/50000 progress=44.4% loss=0.113091 lr=6.326541e-05 lr_backbone=6.326541e-05 elapsed=22:35:35 avg_step=3.67s recent_step=4.82s eta=28:18:55 +2026-09-15 12:22:36 | INFO | train rank=0 | step=22200/50000 progress=44.4% loss=0.101298 lr=6.323352e-05 lr_backbone=6.323352e-05 elapsed=22:36:09 avg_step=3.67s recent_step=3.38s eta=28:18:15 +2026-09-15 12:23:05 | INFO | train rank=0 | step=22210/50000 progress=44.4% loss=0.094683 lr=6.320163e-05 lr_backbone=6.320163e-05 elapsed=22:36:38 avg_step=3.66s recent_step=2.86s eta=28:17:28 +2026-09-15 12:23:34 | INFO | train rank=0 | step=22220/50000 progress=44.4% loss=0.085914 lr=6.316973e-05 lr_backbone=6.316973e-05 elapsed=22:37:07 avg_step=3.66s recent_step=2.94s eta=28:16:42 +2026-09-15 12:24:13 | INFO | train rank=0 | step=22230/50000 progress=44.5% loss=0.123793 lr=6.313782e-05 lr_backbone=6.313782e-05 elapsed=22:37:46 avg_step=3.66s recent_step=3.86s eta=28:16:08 +2026-09-15 12:24:44 | INFO | train rank=0 | step=22240/50000 progress=44.5% loss=0.098340 lr=6.310591e-05 lr_backbone=6.310591e-05 elapsed=22:38:17 avg_step=3.66s recent_step=3.12s eta=28:15:25 +2026-09-15 12:25:18 | INFO | train rank=0 | step=22250/50000 progress=44.5% loss=0.095661 lr=6.307400e-05 lr_backbone=6.307400e-05 elapsed=22:38:51 avg_step=3.66s recent_step=3.37s eta=28:14:44 +2026-09-15 12:25:52 | INFO | train rank=0 | step=22260/50000 progress=44.5% loss=0.120187 lr=6.304208e-05 lr_backbone=6.304208e-05 elapsed=22:39:25 avg_step=3.66s recent_step=3.45s eta=28:14:05 +2026-09-15 12:26:44 | INFO | train rank=0 | step=22270/50000 progress=44.5% loss=0.127683 lr=6.301015e-05 lr_backbone=6.301015e-05 elapsed=22:40:17 avg_step=3.66s recent_step=5.21s eta=28:13:48 +2026-09-15 12:27:17 | INFO | train rank=0 | step=22280/50000 progress=44.6% loss=0.089375 lr=6.297821e-05 lr_backbone=6.297821e-05 elapsed=22:40:50 avg_step=3.66s recent_step=3.26s eta=28:13:06 +2026-09-15 12:27:49 | INFO | train rank=0 | step=22290/50000 progress=44.6% loss=0.119649 lr=6.294628e-05 lr_backbone=6.294628e-05 elapsed=22:41:22 avg_step=3.66s recent_step=3.23s eta=28:12:24 +2026-09-15 12:28:28 | INFO | train rank=0 | step=22300/50000 progress=44.6% loss=0.112186 lr=6.291433e-05 lr_backbone=6.291433e-05 elapsed=22:42:00 avg_step=3.66s recent_step=3.82s eta=28:11:49 +2026-09-15 12:29:23 | INFO | train rank=0 | step=22310/50000 progress=44.6% loss=0.074448 lr=6.288238e-05 lr_backbone=6.288238e-05 elapsed=22:42:55 avg_step=3.67s recent_step=5.50s eta=28:11:35 +2026-09-15 12:29:49 | INFO | train rank=0 | step=22320/50000 progress=44.6% loss=0.091138 lr=6.285043e-05 lr_backbone=6.285043e-05 elapsed=22:43:22 avg_step=3.66s recent_step=2.65s eta=28:10:46 +2026-09-15 12:30:20 | INFO | train rank=0 | step=22330/50000 progress=44.7% loss=0.135406 lr=6.281846e-05 lr_backbone=6.281846e-05 elapsed=22:43:53 avg_step=3.66s recent_step=3.12s eta=28:10:03 +2026-09-15 12:31:00 | INFO | train rank=0 | step=22340/50000 progress=44.7% loss=0.103492 lr=6.278650e-05 lr_backbone=6.278650e-05 elapsed=22:44:33 avg_step=3.66s recent_step=4.00s eta=28:09:30 +2026-09-15 12:31:53 | INFO | train rank=0 | step=22350/50000 progress=44.7% loss=0.125251 lr=6.275452e-05 lr_backbone=6.275452e-05 elapsed=22:45:26 avg_step=3.67s recent_step=5.27s eta=28:09:13 +2026-09-15 12:32:30 | INFO | train rank=0 | step=22360/50000 progress=44.7% loss=0.111080 lr=6.272255e-05 lr_backbone=6.272255e-05 elapsed=22:46:03 avg_step=3.67s recent_step=3.71s eta=28:08:37 +2026-09-15 12:32:57 | INFO | train rank=0 | step=22370/50000 progress=44.7% loss=0.113379 lr=6.269056e-05 lr_backbone=6.269056e-05 elapsed=22:46:30 avg_step=3.67s recent_step=2.71s eta=28:07:49 +2026-09-15 12:33:25 | INFO | train rank=0 | step=22380/50000 progress=44.8% loss=0.117562 lr=6.265857e-05 lr_backbone=6.265857e-05 elapsed=22:46:58 avg_step=3.66s recent_step=2.83s eta=28:07:02 +2026-09-15 12:34:10 | INFO | train rank=0 | step=22390/50000 progress=44.8% loss=0.125305 lr=6.262658e-05 lr_backbone=6.262658e-05 elapsed=22:47:43 avg_step=3.67s recent_step=4.51s eta=28:06:36 +2026-09-15 12:34:50 | INFO | train rank=0 | step=22400/50000 progress=44.8% loss=0.101921 lr=6.259458e-05 lr_backbone=6.259458e-05 elapsed=22:48:23 avg_step=3.67s recent_step=4.00s eta=28:06:03 +2026-09-15 12:35:21 | INFO | train rank=0 | step=22410/50000 progress=44.8% loss=0.099080 lr=6.256257e-05 lr_backbone=6.256257e-05 elapsed=22:48:53 avg_step=3.67s recent_step=3.02s eta=28:05:18 +2026-09-15 12:35:50 | INFO | train rank=0 | step=22420/50000 progress=44.8% loss=0.122090 lr=6.253056e-05 lr_backbone=6.253056e-05 elapsed=22:49:23 avg_step=3.66s recent_step=2.94s eta=28:04:33 +2026-09-15 12:36:35 | INFO | train rank=0 | step=22430/50000 progress=44.9% loss=0.104264 lr=6.249854e-05 lr_backbone=6.249854e-05 elapsed=22:50:07 avg_step=3.67s recent_step=4.45s eta=28:04:06 +2026-09-15 12:37:11 | INFO | train rank=0 | step=22440/50000 progress=44.9% loss=0.114782 lr=6.246652e-05 lr_backbone=6.246652e-05 elapsed=22:50:44 avg_step=3.67s recent_step=3.64s eta=28:03:29 +2026-09-15 12:37:45 | INFO | train rank=0 | step=22450/50000 progress=44.9% loss=0.106009 lr=6.243449e-05 lr_backbone=6.243449e-05 elapsed=22:51:18 avg_step=3.66s recent_step=3.39s eta=28:02:49 +2026-09-15 12:38:19 | INFO | train rank=0 | step=22460/50000 progress=44.9% loss=0.099271 lr=6.240246e-05 lr_backbone=6.240246e-05 elapsed=22:51:52 avg_step=3.66s recent_step=3.41s eta=28:02:09 +2026-09-15 12:39:15 | INFO | train rank=0 | step=22470/50000 progress=44.9% loss=0.120482 lr=6.237042e-05 lr_backbone=6.237042e-05 elapsed=22:52:48 avg_step=3.67s recent_step=5.58s eta=28:01:56 +2026-09-15 12:39:44 | INFO | train rank=0 | step=22480/50000 progress=45.0% loss=0.112241 lr=6.233838e-05 lr_backbone=6.233838e-05 elapsed=22:53:17 avg_step=3.67s recent_step=2.97s eta=28:01:11 +2026-09-15 12:40:24 | INFO | train rank=0 | step=22490/50000 progress=45.0% loss=0.117584 lr=6.230633e-05 lr_backbone=6.230633e-05 elapsed=22:53:57 avg_step=3.67s recent_step=4.01s eta=28:00:38 +2026-09-15 12:40:54 | INFO | train rank=0 | step=22500/50000 progress=45.0% loss=0.092581 lr=6.227427e-05 lr_backbone=6.227427e-05 elapsed=22:54:27 avg_step=3.67s recent_step=2.97s eta=27:59:53 +2026-09-15 12:41:47 | INFO | train rank=0 | step=22510/50000 progress=45.0% loss=0.106007 lr=6.224221e-05 lr_backbone=6.224221e-05 elapsed=22:55:20 avg_step=3.67s recent_step=5.27s eta=27:59:36 +2026-09-15 12:42:19 | INFO | train rank=0 | step=22520/50000 progress=45.0% loss=0.095431 lr=6.221015e-05 lr_backbone=6.221015e-05 elapsed=22:55:52 avg_step=3.67s recent_step=3.21s eta=27:58:54 +2026-09-15 12:42:57 | INFO | train rank=0 | step=22530/50000 progress=45.1% loss=0.092396 lr=6.217808e-05 lr_backbone=6.217808e-05 elapsed=22:56:30 avg_step=3.67s recent_step=3.79s eta=27:58:19 +2026-09-15 12:43:29 | INFO | train rank=0 | step=22540/50000 progress=45.1% loss=0.107721 lr=6.214600e-05 lr_backbone=6.214600e-05 elapsed=22:57:02 avg_step=3.67s recent_step=3.21s eta=27:57:36 +2026-09-15 12:44:22 | INFO | train rank=0 | step=22550/50000 progress=45.1% loss=0.124411 lr=6.211392e-05 lr_backbone=6.211392e-05 elapsed=22:57:55 avg_step=3.67s recent_step=5.27s eta=27:57:19 +2026-09-15 12:44:53 | INFO | train rank=0 | step=22560/50000 progress=45.1% loss=0.081886 lr=6.208183e-05 lr_backbone=6.208183e-05 elapsed=22:58:26 avg_step=3.67s recent_step=3.17s eta=27:56:37 +2026-09-15 12:45:27 | INFO | train rank=0 | step=22570/50000 progress=45.1% loss=0.148576 lr=6.204974e-05 lr_backbone=6.204974e-05 elapsed=22:59:00 avg_step=3.67s recent_step=3.37s eta=27:55:56 +2026-09-15 12:46:02 | INFO | train rank=0 | step=22580/50000 progress=45.2% loss=0.097680 lr=6.201764e-05 lr_backbone=6.201764e-05 elapsed=22:59:34 avg_step=3.67s recent_step=3.46s eta=27:55:17 +2026-09-15 12:46:52 | INFO | train rank=0 | step=22590/50000 progress=45.2% loss=0.097682 lr=6.198554e-05 lr_backbone=6.198554e-05 elapsed=23:00:24 avg_step=3.67s recent_step=4.99s eta=27:54:57 +2026-09-15 12:47:27 | INFO | train rank=0 | step=22600/50000 progress=45.2% loss=0.114883 lr=6.195343e-05 lr_backbone=6.195343e-05 elapsed=23:01:00 avg_step=3.67s recent_step=3.58s eta=27:54:19 +2026-09-15 12:48:00 | INFO | train rank=0 | step=22610/50000 progress=45.2% loss=0.077789 lr=6.192132e-05 lr_backbone=6.192132e-05 elapsed=23:01:32 avg_step=3.67s recent_step=3.24s eta=27:53:37 +2026-09-15 12:48:36 | INFO | train rank=0 | step=22620/50000 progress=45.2% loss=0.100698 lr=6.188920e-05 lr_backbone=6.188920e-05 elapsed=23:02:09 avg_step=3.67s recent_step=3.65s eta=27:53:00 +2026-09-15 12:49:27 | INFO | train rank=0 | step=22630/50000 progress=45.3% loss=0.086034 lr=6.185708e-05 lr_backbone=6.185708e-05 elapsed=23:03:00 avg_step=3.67s recent_step=5.13s eta=27:52:41 +2026-09-15 12:50:04 | INFO | train rank=0 | step=22640/50000 progress=45.3% loss=0.137180 lr=6.182495e-05 lr_backbone=6.182495e-05 elapsed=23:03:37 avg_step=3.67s recent_step=3.65s eta=27:52:04 +2026-09-15 12:50:37 | INFO | train rank=0 | step=22650/50000 progress=45.3% loss=0.092853 lr=6.179282e-05 lr_backbone=6.179282e-05 elapsed=23:04:09 avg_step=3.67s recent_step=3.26s eta=27:51:23 +2026-09-15 12:51:07 | INFO | train rank=0 | step=22660/50000 progress=45.3% loss=0.095546 lr=6.176068e-05 lr_backbone=6.176068e-05 elapsed=23:04:40 avg_step=3.67s recent_step=3.07s eta=27:50:39 +2026-09-15 12:51:51 | INFO | train rank=0 | step=22670/50000 progress=45.3% loss=0.077597 lr=6.172853e-05 lr_backbone=6.172853e-05 elapsed=23:05:23 avg_step=3.67s recent_step=4.34s eta=27:50:10 +2026-09-15 12:52:23 | INFO | train rank=0 | step=22680/50000 progress=45.4% loss=0.115748 lr=6.169638e-05 lr_backbone=6.169638e-05 elapsed=23:05:56 avg_step=3.67s recent_step=3.25s eta=27:49:29 +2026-09-15 12:52:49 | INFO | train rank=0 | step=22690/50000 progress=45.4% loss=0.127403 lr=6.166423e-05 lr_backbone=6.166423e-05 elapsed=23:06:22 avg_step=3.67s recent_step=2.60s eta=27:48:39 +2026-09-15 12:53:23 | INFO | train rank=0 | step=22700/50000 progress=45.4% loss=0.115985 lr=6.163207e-05 lr_backbone=6.163207e-05 elapsed=23:06:56 avg_step=3.67s recent_step=3.39s eta=27:47:59 +2026-09-15 12:54:04 | INFO | train rank=0 | step=22710/50000 progress=45.4% loss=0.103795 lr=6.159991e-05 lr_backbone=6.159991e-05 elapsed=23:07:37 avg_step=3.67s recent_step=4.09s eta=27:47:28 +2026-09-15 12:54:41 | INFO | train rank=0 | step=22720/50000 progress=45.4% loss=0.117283 lr=6.156774e-05 lr_backbone=6.156774e-05 elapsed=23:08:14 avg_step=3.67s recent_step=3.74s eta=27:46:52 +2026-09-15 12:55:13 | INFO | train rank=0 | step=22730/50000 progress=45.5% loss=0.115082 lr=6.153556e-05 lr_backbone=6.153556e-05 elapsed=23:08:46 avg_step=3.67s recent_step=3.20s eta=27:46:10 +2026-09-15 12:55:50 | INFO | train rank=0 | step=22740/50000 progress=45.5% loss=0.089317 lr=6.150338e-05 lr_backbone=6.150338e-05 elapsed=23:09:23 avg_step=3.67s recent_step=3.64s eta=27:45:33 +2026-09-15 12:56:32 | INFO | train rank=0 | step=22750/50000 progress=45.5% loss=0.070538 lr=6.147120e-05 lr_backbone=6.147120e-05 elapsed=23:10:05 avg_step=3.67s recent_step=4.25s eta=27:45:03 +2026-09-15 12:57:21 | INFO | train rank=0 | step=22760/50000 progress=45.5% loss=0.087292 lr=6.143901e-05 lr_backbone=6.143901e-05 elapsed=23:10:54 avg_step=3.67s recent_step=4.90s eta=27:44:41 +2026-09-15 12:57:49 | INFO | train rank=0 | step=22770/50000 progress=45.5% loss=0.113080 lr=6.140681e-05 lr_backbone=6.140681e-05 elapsed=23:11:21 avg_step=3.67s recent_step=2.74s eta=27:43:53 +2026-09-15 12:58:15 | INFO | train rank=0 | step=22780/50000 progress=45.6% loss=0.107683 lr=6.137461e-05 lr_backbone=6.137461e-05 elapsed=23:11:48 avg_step=3.67s recent_step=2.65s eta=27:43:04 +2026-09-15 12:58:56 | INFO | train rank=0 | step=22790/50000 progress=45.6% loss=0.090984 lr=6.134241e-05 lr_backbone=6.134241e-05 elapsed=23:12:28 avg_step=3.67s recent_step=4.03s eta=27:42:32 +2026-09-15 12:59:47 | INFO | train rank=0 | step=22800/50000 progress=45.6% loss=0.130751 lr=6.131020e-05 lr_backbone=6.131020e-05 elapsed=23:13:20 avg_step=3.67s recent_step=5.15s eta=27:42:13 +2026-09-15 13:00:19 | INFO | train rank=0 | step=22810/50000 progress=45.6% loss=0.081018 lr=6.127798e-05 lr_backbone=6.127798e-05 elapsed=23:13:52 avg_step=3.67s recent_step=3.23s eta=27:41:31 +2026-09-15 13:00:57 | INFO | train rank=0 | step=22820/50000 progress=45.6% loss=0.117690 lr=6.124576e-05 lr_backbone=6.124576e-05 elapsed=23:14:30 avg_step=3.67s recent_step=3.80s eta=27:40:56 +2026-09-15 13:01:30 | INFO | train rank=0 | step=22830/50000 progress=45.7% loss=0.074088 lr=6.121354e-05 lr_backbone=6.121354e-05 elapsed=23:15:03 avg_step=3.67s recent_step=3.31s eta=27:40:15 +2026-09-15 13:02:21 | INFO | train rank=0 | step=22840/50000 progress=45.7% loss=0.109754 lr=6.118131e-05 lr_backbone=6.118131e-05 elapsed=23:15:54 avg_step=3.67s recent_step=5.07s eta=27:39:55 +2026-09-15 13:02:52 | INFO | train rank=0 | step=22850/50000 progress=45.7% loss=0.098274 lr=6.114907e-05 lr_backbone=6.114907e-05 elapsed=23:16:25 avg_step=3.67s recent_step=3.10s eta=27:39:12 +2026-09-15 13:03:16 | INFO | train rank=0 | step=22860/50000 progress=45.7% loss=0.116746 lr=6.111683e-05 lr_backbone=6.111683e-05 elapsed=23:16:49 avg_step=3.67s recent_step=2.40s eta=27:38:20 +2026-09-15 13:03:56 | INFO | train rank=0 | step=22870/50000 progress=45.7% loss=0.132254 lr=6.108459e-05 lr_backbone=6.108459e-05 elapsed=23:17:28 avg_step=3.67s recent_step=3.94s eta=27:37:47 +2026-09-15 13:04:46 | INFO | train rank=0 | step=22880/50000 progress=45.8% loss=0.099240 lr=6.105234e-05 lr_backbone=6.105234e-05 elapsed=23:18:18 avg_step=3.67s recent_step=5.02s eta=27:37:26 +2026-09-15 13:05:22 | INFO | train rank=0 | step=22890/50000 progress=45.8% loss=0.082279 lr=6.102009e-05 lr_backbone=6.102009e-05 elapsed=23:18:54 avg_step=3.67s recent_step=3.60s eta=27:36:49 +2026-09-15 13:05:57 | INFO | train rank=0 | step=22900/50000 progress=45.8% loss=0.082140 lr=6.098783e-05 lr_backbone=6.098783e-05 elapsed=23:19:30 avg_step=3.67s recent_step=3.58s eta=27:36:11 +2026-09-15 13:06:36 | INFO | train rank=0 | step=22910/50000 progress=45.8% loss=0.109850 lr=6.095557e-05 lr_backbone=6.095557e-05 elapsed=23:20:08 avg_step=3.67s recent_step=3.82s eta=27:35:36 +2026-09-15 13:07:26 | INFO | train rank=0 | step=22920/50000 progress=45.8% loss=0.077406 lr=6.092330e-05 lr_backbone=6.092330e-05 elapsed=23:20:59 avg_step=3.67s recent_step=5.06s eta=27:35:16 +2026-09-15 13:08:01 | INFO | train rank=0 | step=22930/50000 progress=45.9% loss=0.099441 lr=6.089102e-05 lr_backbone=6.089102e-05 elapsed=23:21:34 avg_step=3.67s recent_step=3.50s eta=27:34:37 +2026-09-15 13:08:30 | INFO | train rank=0 | step=22940/50000 progress=45.9% loss=0.110575 lr=6.085875e-05 lr_backbone=6.085875e-05 elapsed=23:22:03 avg_step=3.67s recent_step=2.88s eta=27:33:51 +2026-09-15 13:08:59 | INFO | train rank=0 | step=22950/50000 progress=45.9% loss=0.110038 lr=6.082646e-05 lr_backbone=6.082646e-05 elapsed=23:22:32 avg_step=3.67s recent_step=2.91s eta=27:33:06 +2026-09-15 13:09:49 | INFO | train rank=0 | step=22960/50000 progress=45.9% loss=0.089697 lr=6.079418e-05 lr_backbone=6.079418e-05 elapsed=23:23:21 avg_step=3.67s recent_step=4.94s eta=27:32:44 +2026-09-15 13:10:26 | INFO | train rank=0 | step=22970/50000 progress=45.9% loss=0.116391 lr=6.076189e-05 lr_backbone=6.076189e-05 elapsed=23:23:59 avg_step=3.67s recent_step=3.76s eta=27:32:09 +2026-09-15 13:11:01 | INFO | train rank=0 | step=22980/50000 progress=46.0% loss=0.110388 lr=6.072959e-05 lr_backbone=6.072959e-05 elapsed=23:24:34 avg_step=3.67s recent_step=3.48s eta=27:31:30 +2026-09-15 13:11:32 | INFO | train rank=0 | step=22990/50000 progress=46.0% loss=0.094051 lr=6.069729e-05 lr_backbone=6.069729e-05 elapsed=23:25:05 avg_step=3.67s recent_step=3.08s eta=27:30:46 +2026-09-15 13:12:11 | INFO | train rank=0 | step=23000/50000 progress=46.0% loss=0.083376 lr=6.066498e-05 lr_backbone=6.066498e-05 elapsed=23:25:44 avg_step=3.67s recent_step=3.90s eta=27:30:12 +2026-09-15 13:12:44 | INFO | train rank=0 | step=23010/50000 progress=46.0% loss=0.083803 lr=6.063267e-05 lr_backbone=6.063267e-05 elapsed=23:26:17 avg_step=3.67s recent_step=3.31s eta=27:29:31 +2026-09-15 13:13:16 | INFO | train rank=0 | step=23020/50000 progress=46.0% loss=0.096253 lr=6.060036e-05 lr_backbone=6.060036e-05 elapsed=23:26:49 avg_step=3.67s recent_step=3.19s eta=27:28:49 +2026-09-15 13:13:41 | INFO | train rank=0 | step=23030/50000 progress=46.1% loss=0.102671 lr=6.056804e-05 lr_backbone=6.056804e-05 elapsed=23:27:14 avg_step=3.67s recent_step=2.55s eta=27:27:59 +2026-09-15 13:14:30 | INFO | train rank=0 | step=23040/50000 progress=46.1% loss=0.102049 lr=6.053571e-05 lr_backbone=6.053571e-05 elapsed=23:28:02 avg_step=3.67s recent_step=4.83s eta=27:27:36 +2026-09-15 13:15:02 | INFO | train rank=0 | step=23050/50000 progress=46.1% loss=0.118172 lr=6.050338e-05 lr_backbone=6.050338e-05 elapsed=23:28:35 avg_step=3.67s recent_step=3.23s eta=27:26:54 +2026-09-15 13:15:36 | INFO | train rank=0 | step=23060/50000 progress=46.1% loss=0.105275 lr=6.047105e-05 lr_backbone=6.047105e-05 elapsed=23:29:08 avg_step=3.67s recent_step=3.38s eta=27:26:14 +2026-09-15 13:16:04 | INFO | train rank=0 | step=23070/50000 progress=46.1% loss=0.111103 lr=6.043871e-05 lr_backbone=6.043871e-05 elapsed=23:29:36 avg_step=3.67s recent_step=2.81s eta=27:25:28 +2026-09-15 13:16:48 | INFO | train rank=0 | step=23080/50000 progress=46.2% loss=0.122162 lr=6.040637e-05 lr_backbone=6.040637e-05 elapsed=23:30:21 avg_step=3.67s recent_step=4.42s eta=27:25:00 +2026-09-15 13:17:24 | INFO | train rank=0 | step=23090/50000 progress=46.2% loss=0.100729 lr=6.037402e-05 lr_backbone=6.037402e-05 elapsed=23:30:57 avg_step=3.67s recent_step=3.63s eta=27:24:23 +2026-09-15 13:17:59 | INFO | train rank=0 | step=23100/50000 progress=46.2% loss=0.086888 lr=6.034167e-05 lr_backbone=6.034167e-05 elapsed=23:31:32 avg_step=3.67s recent_step=3.46s eta=27:23:44 +2026-09-15 13:18:34 | INFO | train rank=0 | step=23110/50000 progress=46.2% loss=0.094766 lr=6.030931e-05 lr_backbone=6.030931e-05 elapsed=23:32:07 avg_step=3.67s recent_step=3.54s eta=27:23:06 +2026-09-15 13:19:22 | INFO | train rank=0 | step=23120/50000 progress=46.2% loss=0.116034 lr=6.027695e-05 lr_backbone=6.027695e-05 elapsed=23:32:54 avg_step=3.67s recent_step=4.73s eta=27:22:41 +2026-09-15 13:19:48 | INFO | train rank=0 | step=23130/50000 progress=46.3% loss=0.130312 lr=6.024458e-05 lr_backbone=6.024458e-05 elapsed=23:33:21 avg_step=3.67s recent_step=2.64s eta=27:21:53 +2026-09-15 13:20:24 | INFO | train rank=0 | step=23140/50000 progress=46.3% loss=0.102179 lr=6.021221e-05 lr_backbone=6.021221e-05 elapsed=23:33:57 avg_step=3.67s recent_step=3.64s eta=27:21:16 +2026-09-15 13:21:07 | INFO | train rank=0 | step=23150/50000 progress=46.3% loss=0.093328 lr=6.017984e-05 lr_backbone=6.017984e-05 elapsed=23:34:40 avg_step=3.67s recent_step=4.31s eta=27:20:46 +2026-09-15 13:22:00 | INFO | train rank=0 | step=23160/50000 progress=46.3% loss=0.099184 lr=6.014746e-05 lr_backbone=6.014746e-05 elapsed=23:35:33 avg_step=3.67s recent_step=5.27s eta=27:20:28 +2026-09-15 13:22:28 | INFO | train rank=0 | step=23170/50000 progress=46.3% loss=0.099136 lr=6.011508e-05 lr_backbone=6.011508e-05 elapsed=23:36:01 avg_step=3.67s recent_step=2.83s eta=27:19:42 +2026-09-15 13:23:05 | INFO | train rank=0 | step=23180/50000 progress=46.4% loss=0.094088 lr=6.008269e-05 lr_backbone=6.008269e-05 elapsed=23:36:37 avg_step=3.67s recent_step=3.63s eta=27:19:05 +2026-09-15 13:23:44 | INFO | train rank=0 | step=23190/50000 progress=46.4% loss=0.105667 lr=6.005030e-05 lr_backbone=6.005030e-05 elapsed=23:37:17 avg_step=3.67s recent_step=3.91s eta=27:18:31 +2026-09-15 13:24:38 | INFO | train rank=0 | step=23200/50000 progress=46.4% loss=0.090090 lr=6.001790e-05 lr_backbone=6.001790e-05 elapsed=23:38:11 avg_step=3.67s recent_step=5.47s eta=27:18:15 +2026-09-15 13:25:19 | INFO | train rank=0 | step=23210/50000 progress=46.4% loss=0.133045 lr=5.998550e-05 lr_backbone=5.998550e-05 elapsed=23:38:51 avg_step=3.67s recent_step=4.02s eta=27:17:43 +2026-09-15 13:25:53 | INFO | train rank=0 | step=23220/50000 progress=46.4% loss=0.105027 lr=5.995309e-05 lr_backbone=5.995309e-05 elapsed=23:39:26 avg_step=3.67s recent_step=3.41s eta=27:17:03 +2026-09-15 13:26:27 | INFO | train rank=0 | step=23230/50000 progress=46.5% loss=0.100300 lr=5.992068e-05 lr_backbone=5.992068e-05 elapsed=23:40:00 avg_step=3.67s recent_step=3.47s eta=27:16:24 +2026-09-15 13:27:09 | INFO | train rank=0 | step=23240/50000 progress=46.5% loss=0.115855 lr=5.988827e-05 lr_backbone=5.988827e-05 elapsed=23:40:41 avg_step=3.67s recent_step=4.12s eta=27:15:52 +2026-09-15 13:27:43 | INFO | train rank=0 | step=23250/50000 progress=46.5% loss=0.109291 lr=5.985585e-05 lr_backbone=5.985585e-05 elapsed=23:41:16 avg_step=3.67s recent_step=3.47s eta=27:15:13 +2026-09-15 13:28:21 | INFO | train rank=0 | step=23260/50000 progress=46.5% loss=0.078068 lr=5.982343e-05 lr_backbone=5.982343e-05 elapsed=23:41:54 avg_step=3.67s recent_step=3.75s eta=27:14:38 +2026-09-15 13:28:51 | INFO | train rank=0 | step=23270/50000 progress=46.5% loss=0.100878 lr=5.979100e-05 lr_backbone=5.979100e-05 elapsed=23:42:24 avg_step=3.67s recent_step=3.01s eta=27:13:54 +2026-09-15 13:29:31 | INFO | train rank=0 | step=23280/50000 progress=46.6% loss=0.092720 lr=5.975857e-05 lr_backbone=5.975857e-05 elapsed=23:43:04 avg_step=3.67s recent_step=4.05s eta=27:13:21 +2026-09-15 13:30:05 | INFO | train rank=0 | step=23290/50000 progress=46.6% loss=0.094629 lr=5.972613e-05 lr_backbone=5.972613e-05 elapsed=23:43:38 avg_step=3.67s recent_step=3.34s eta=27:12:41 +2026-09-15 13:30:42 | INFO | train rank=0 | step=23300/50000 progress=46.6% loss=0.120949 lr=5.969369e-05 lr_backbone=5.969369e-05 elapsed=23:44:14 avg_step=3.67s recent_step=3.68s eta=27:12:04 +2026-09-15 13:31:16 | INFO | train rank=0 | step=23310/50000 progress=46.6% loss=0.112356 lr=5.966125e-05 lr_backbone=5.966125e-05 elapsed=23:44:49 avg_step=3.67s recent_step=3.46s eta=27:11:25 +2026-09-15 13:32:00 | INFO | train rank=0 | step=23320/50000 progress=46.6% loss=0.122711 lr=5.962880e-05 lr_backbone=5.962880e-05 elapsed=23:45:33 avg_step=3.67s recent_step=4.35s eta=27:10:56 +2026-09-15 13:32:36 | INFO | train rank=0 | step=23330/50000 progress=46.7% loss=0.097813 lr=5.959635e-05 lr_backbone=5.959635e-05 elapsed=23:46:09 avg_step=3.67s recent_step=3.64s eta=27:10:19 +2026-09-15 13:33:08 | INFO | train rank=0 | step=23340/50000 progress=46.7% loss=0.104754 lr=5.956389e-05 lr_backbone=5.956389e-05 elapsed=23:46:41 avg_step=3.67s recent_step=3.21s eta=27:09:38 +2026-09-15 13:33:45 | INFO | train rank=0 | step=23350/50000 progress=46.7% loss=0.077900 lr=5.953143e-05 lr_backbone=5.953143e-05 elapsed=23:47:18 avg_step=3.67s recent_step=3.64s eta=27:09:01 +2026-09-15 13:34:37 | INFO | train rank=0 | step=23360/50000 progress=46.7% loss=0.100936 lr=5.949897e-05 lr_backbone=5.949897e-05 elapsed=23:48:10 avg_step=3.67s recent_step=5.21s eta=27:08:42 +2026-09-15 13:35:08 | INFO | train rank=0 | step=23370/50000 progress=46.7% loss=0.096860 lr=5.946650e-05 lr_backbone=5.946650e-05 elapsed=23:48:41 avg_step=3.67s recent_step=3.10s eta=27:07:58 +2026-09-15 13:35:46 | INFO | train rank=0 | step=23380/50000 progress=46.8% loss=0.085262 lr=5.943402e-05 lr_backbone=5.943402e-05 elapsed=23:49:18 avg_step=3.67s recent_step=3.77s eta=27:07:23 +2026-09-15 13:36:21 | INFO | train rank=0 | step=23390/50000 progress=46.8% loss=0.148328 lr=5.940155e-05 lr_backbone=5.940155e-05 elapsed=23:49:54 avg_step=3.67s recent_step=3.51s eta=27:06:44 +2026-09-15 13:37:11 | INFO | train rank=0 | step=23400/50000 progress=46.8% loss=0.075909 lr=5.936907e-05 lr_backbone=5.936907e-05 elapsed=23:50:44 avg_step=3.67s recent_step=5.01s eta=27:06:23 +2026-09-15 13:37:46 | INFO | train rank=0 | step=23410/50000 progress=46.8% loss=0.083845 lr=5.933658e-05 lr_backbone=5.933658e-05 elapsed=23:51:19 avg_step=3.67s recent_step=3.52s eta=27:05:45 +2026-09-15 13:38:20 | INFO | train rank=0 | step=23420/50000 progress=46.8% loss=0.105921 lr=5.930409e-05 lr_backbone=5.930409e-05 elapsed=23:51:53 avg_step=3.67s recent_step=3.40s eta=27:05:05 +2026-09-15 13:39:02 | INFO | train rank=0 | step=23430/50000 progress=46.9% loss=0.110507 lr=5.927160e-05 lr_backbone=5.927160e-05 elapsed=23:52:35 avg_step=3.67s recent_step=4.21s eta=27:04:34 +2026-09-15 13:39:54 | INFO | train rank=0 | step=23440/50000 progress=46.9% loss=0.096917 lr=5.923910e-05 lr_backbone=5.923910e-05 elapsed=23:53:27 avg_step=3.67s recent_step=5.19s eta=27:04:15 +2026-09-15 13:40:23 | INFO | train rank=0 | step=23450/50000 progress=46.9% loss=0.111423 lr=5.920660e-05 lr_backbone=5.920660e-05 elapsed=23:53:56 avg_step=3.67s recent_step=2.94s eta=27:03:30 +2026-09-15 13:40:51 | INFO | train rank=0 | step=23460/50000 progress=46.9% loss=0.116305 lr=5.917409e-05 lr_backbone=5.917409e-05 elapsed=23:54:24 avg_step=3.67s recent_step=2.80s eta=27:02:43 +2026-09-15 13:41:23 | INFO | train rank=0 | step=23470/50000 progress=46.9% loss=0.090231 lr=5.914158e-05 lr_backbone=5.914158e-05 elapsed=23:54:56 avg_step=3.67s recent_step=3.17s eta=27:02:01 +2026-09-15 13:42:09 | INFO | train rank=0 | step=23480/50000 progress=47.0% loss=0.106022 lr=5.910907e-05 lr_backbone=5.910907e-05 elapsed=23:55:41 avg_step=3.67s recent_step=4.56s eta=27:01:34 +2026-09-15 13:42:48 | INFO | train rank=0 | step=23490/50000 progress=47.0% loss=0.079601 lr=5.907655e-05 lr_backbone=5.907655e-05 elapsed=23:56:21 avg_step=3.67s recent_step=3.97s eta=27:01:01 +2026-09-15 13:43:21 | INFO | train rank=0 | step=23500/50000 progress=47.0% loss=0.104579 lr=5.904403e-05 lr_backbone=5.904403e-05 elapsed=23:56:53 avg_step=3.67s recent_step=3.22s eta=27:00:19 +2026-09-15 13:43:54 | INFO | train rank=0 | step=23510/50000 progress=47.0% loss=0.107100 lr=5.901150e-05 lr_backbone=5.901150e-05 elapsed=23:57:26 avg_step=3.67s recent_step=3.30s eta=26:59:39 +2026-09-15 13:44:45 | INFO | train rank=0 | step=23520/50000 progress=47.0% loss=0.105522 lr=5.897897e-05 lr_backbone=5.897897e-05 elapsed=23:58:18 avg_step=3.67s recent_step=5.13s eta=26:59:18 +2026-09-15 13:45:09 | INFO | train rank=0 | step=23530/50000 progress=47.1% loss=0.108649 lr=5.894644e-05 lr_backbone=5.894644e-05 elapsed=23:58:42 avg_step=3.67s recent_step=2.43s eta=26:58:28 +2026-09-15 13:45:49 | INFO | train rank=0 | step=23540/50000 progress=47.1% loss=0.095019 lr=5.891390e-05 lr_backbone=5.891390e-05 elapsed=23:59:21 avg_step=3.67s recent_step=3.94s eta=26:57:54 +2026-09-15 13:46:19 | INFO | train rank=0 | step=23550/50000 progress=47.1% loss=0.098603 lr=5.888136e-05 lr_backbone=5.888136e-05 elapsed=23:59:52 avg_step=3.67s recent_step=3.06s eta=26:57:11 +2026-09-15 13:47:02 | INFO | train rank=0 | step=23560/50000 progress=47.1% loss=0.105258 lr=5.884881e-05 lr_backbone=5.884881e-05 elapsed=24:00:35 avg_step=3.67s recent_step=4.31s eta=26:56:41 +2026-09-15 13:47:40 | INFO | train rank=0 | step=23570/50000 progress=47.1% loss=0.104627 lr=5.881626e-05 lr_backbone=5.881626e-05 elapsed=24:01:13 avg_step=3.67s recent_step=3.79s eta=26:56:06 +2026-09-15 13:48:11 | INFO | train rank=0 | step=23580/50000 progress=47.2% loss=0.110510 lr=5.878371e-05 lr_backbone=5.878371e-05 elapsed=24:01:44 avg_step=3.67s recent_step=3.13s eta=26:55:23 +2026-09-15 13:48:43 | INFO | train rank=0 | step=23590/50000 progress=47.2% loss=0.112850 lr=5.875115e-05 lr_backbone=5.875115e-05 elapsed=24:02:16 avg_step=3.67s recent_step=3.17s eta=26:54:41 +2026-09-15 13:49:42 | INFO | train rank=0 | step=23600/50000 progress=47.2% loss=0.113608 lr=5.871859e-05 lr_backbone=5.871859e-05 elapsed=24:03:15 avg_step=3.67s recent_step=5.87s eta=26:54:29 +2026-09-15 13:50:17 | INFO | train rank=0 | step=23610/50000 progress=47.2% loss=0.120218 lr=5.868603e-05 lr_backbone=5.868603e-05 elapsed=24:03:50 avg_step=3.67s recent_step=3.52s eta=26:53:50 +2026-09-15 13:50:48 | INFO | train rank=0 | step=23620/50000 progress=47.2% loss=0.096619 lr=5.865346e-05 lr_backbone=5.865346e-05 elapsed=24:04:21 avg_step=3.67s recent_step=3.11s eta=26:53:07 +2026-09-15 13:51:21 | INFO | train rank=0 | step=23630/50000 progress=47.3% loss=0.134277 lr=5.862089e-05 lr_backbone=5.862089e-05 elapsed=24:04:54 avg_step=3.67s recent_step=3.30s eta=26:52:26 +2026-09-15 13:52:10 | INFO | train rank=0 | step=23640/50000 progress=47.3% loss=0.091886 lr=5.858831e-05 lr_backbone=5.858831e-05 elapsed=24:05:43 avg_step=3.67s recent_step=4.91s eta=26:52:03 +2026-09-15 13:52:39 | INFO | train rank=0 | step=23650/50000 progress=47.3% loss=0.085284 lr=5.855573e-05 lr_backbone=5.855573e-05 elapsed=24:06:11 avg_step=3.67s recent_step=2.86s eta=26:51:18 +2026-09-15 13:53:14 | INFO | train rank=0 | step=23660/50000 progress=47.3% loss=0.110677 lr=5.852315e-05 lr_backbone=5.852315e-05 elapsed=24:06:47 avg_step=3.67s recent_step=3.53s eta=26:50:40 +2026-09-15 13:53:46 | INFO | train rank=0 | step=23670/50000 progress=47.3% loss=0.079508 lr=5.849056e-05 lr_backbone=5.849056e-05 elapsed=24:07:19 avg_step=3.67s recent_step=3.24s eta=26:49:58 +2026-09-15 13:54:32 | INFO | train rank=0 | step=23680/50000 progress=47.4% loss=0.112670 lr=5.845797e-05 lr_backbone=5.845797e-05 elapsed=24:08:05 avg_step=3.67s recent_step=4.53s eta=26:49:31 +2026-09-15 13:55:07 | INFO | train rank=0 | step=23690/50000 progress=47.4% loss=0.094700 lr=5.842537e-05 lr_backbone=5.842537e-05 elapsed=24:08:40 avg_step=3.67s recent_step=3.57s eta=26:48:53 +2026-09-15 13:55:41 | INFO | train rank=0 | step=23700/50000 progress=47.4% loss=0.094114 lr=5.839278e-05 lr_backbone=5.839278e-05 elapsed=24:09:13 avg_step=3.67s recent_step=3.32s eta=26:48:13 +2026-09-15 13:56:11 | INFO | train rank=0 | step=23710/50000 progress=47.4% loss=0.102670 lr=5.836017e-05 lr_backbone=5.836017e-05 elapsed=24:09:44 avg_step=3.67s recent_step=3.03s eta=26:47:29 +2026-09-15 13:57:04 | INFO | train rank=0 | step=23720/50000 progress=47.4% loss=0.089063 lr=5.832757e-05 lr_backbone=5.832757e-05 elapsed=24:10:37 avg_step=3.67s recent_step=5.34s eta=26:47:11 +2026-09-15 13:57:32 | INFO | train rank=0 | step=23730/50000 progress=47.5% loss=0.109195 lr=5.829496e-05 lr_backbone=5.829496e-05 elapsed=24:11:05 avg_step=3.67s recent_step=2.80s eta=26:46:24 +2026-09-15 13:58:01 | INFO | train rank=0 | step=23740/50000 progress=47.5% loss=0.086068 lr=5.826235e-05 lr_backbone=5.826235e-05 elapsed=24:11:34 avg_step=3.67s recent_step=2.85s eta=26:45:39 +2026-09-15 13:58:32 | INFO | train rank=0 | step=23750/50000 progress=47.5% loss=0.091245 lr=5.822973e-05 lr_backbone=5.822973e-05 elapsed=24:12:05 avg_step=3.67s recent_step=3.14s eta=26:44:56 +2026-09-15 13:59:22 | INFO | train rank=0 | step=23760/50000 progress=47.5% loss=0.086497 lr=5.819711e-05 lr_backbone=5.819711e-05 elapsed=24:12:55 avg_step=3.67s recent_step=4.99s eta=26:44:34 +2026-09-15 14:00:01 | INFO | train rank=0 | step=23770/50000 progress=47.5% loss=0.075166 lr=5.816449e-05 lr_backbone=5.816449e-05 elapsed=24:13:34 avg_step=3.67s recent_step=3.86s eta=26:44:00 +2026-09-15 14:00:41 | INFO | train rank=0 | step=23780/50000 progress=47.6% loss=0.113610 lr=5.813186e-05 lr_backbone=5.813186e-05 elapsed=24:14:13 avg_step=3.67s recent_step=3.98s eta=26:43:26 +2026-09-15 14:01:12 | INFO | train rank=0 | step=23790/50000 progress=47.6% loss=0.083964 lr=5.809923e-05 lr_backbone=5.809923e-05 elapsed=24:14:44 avg_step=3.67s recent_step=3.09s eta=26:42:43 +2026-09-15 14:01:59 | INFO | train rank=0 | step=23800/50000 progress=47.6% loss=0.098606 lr=5.806659e-05 lr_backbone=5.806659e-05 elapsed=24:15:31 avg_step=3.67s recent_step=4.70s eta=26:42:18 +2026-09-15 14:02:35 | INFO | train rank=0 | step=23810/50000 progress=47.6% loss=0.087063 lr=5.803395e-05 lr_backbone=5.803395e-05 elapsed=24:16:07 avg_step=3.67s recent_step=3.60s eta=26:41:40 +2026-09-15 14:03:11 | INFO | train rank=0 | step=23820/50000 progress=47.6% loss=0.073859 lr=5.800131e-05 lr_backbone=5.800131e-05 elapsed=24:16:43 avg_step=3.67s recent_step=3.61s eta=26:41:03 +2026-09-15 14:03:49 | INFO | train rank=0 | step=23830/50000 progress=47.7% loss=0.119747 lr=5.796867e-05 lr_backbone=5.796867e-05 elapsed=24:17:22 avg_step=3.67s recent_step=3.86s eta=26:40:29 +2026-09-15 14:04:45 | INFO | train rank=0 | step=23840/50000 progress=47.7% loss=0.128796 lr=5.793602e-05 lr_backbone=5.793602e-05 elapsed=24:18:18 avg_step=3.67s recent_step=5.55s eta=26:40:12 +2026-09-15 14:05:23 | INFO | train rank=0 | step=23850/50000 progress=47.7% loss=0.107979 lr=5.790337e-05 lr_backbone=5.790337e-05 elapsed=24:18:56 avg_step=3.67s recent_step=3.85s eta=26:39:38 +2026-09-15 14:05:54 | INFO | train rank=0 | step=23860/50000 progress=47.7% loss=0.120601 lr=5.787071e-05 lr_backbone=5.787071e-05 elapsed=24:19:27 avg_step=3.67s recent_step=3.05s eta=26:38:54 +2026-09-15 14:06:30 | INFO | train rank=0 | step=23870/50000 progress=47.7% loss=0.095167 lr=5.783805e-05 lr_backbone=5.783805e-05 elapsed=24:20:03 avg_step=3.67s recent_step=3.62s eta=26:38:17 +2026-09-15 14:07:08 | INFO | train rank=0 | step=23880/50000 progress=47.8% loss=0.101017 lr=5.780539e-05 lr_backbone=5.780539e-05 elapsed=24:20:41 avg_step=3.67s recent_step=3.80s eta=26:37:42 +2026-09-15 14:07:45 | INFO | train rank=0 | step=23890/50000 progress=47.8% loss=0.116641 lr=5.777273e-05 lr_backbone=5.777273e-05 elapsed=24:21:17 avg_step=3.67s recent_step=3.67s eta=26:37:05 +2026-09-15 14:08:16 | INFO | train rank=0 | step=23900/50000 progress=47.8% loss=0.103108 lr=5.774006e-05 lr_backbone=5.774006e-05 elapsed=24:21:49 avg_step=3.67s recent_step=3.12s eta=26:36:22 +2026-09-15 14:08:52 | INFO | train rank=0 | step=23910/50000 progress=47.8% loss=0.082491 lr=5.770738e-05 lr_backbone=5.770738e-05 elapsed=24:22:24 avg_step=3.67s recent_step=3.59s eta=26:35:45 +2026-09-15 14:09:29 | INFO | train rank=0 | step=23920/50000 progress=47.8% loss=0.107086 lr=5.767471e-05 lr_backbone=5.767471e-05 elapsed=24:23:02 avg_step=3.67s recent_step=3.73s eta=26:35:09 +2026-09-15 14:09:58 | INFO | train rank=0 | step=23930/50000 progress=47.9% loss=0.078958 lr=5.764203e-05 lr_backbone=5.764203e-05 elapsed=24:23:31 avg_step=3.67s recent_step=2.95s eta=26:34:24 +2026-09-15 14:10:40 | INFO | train rank=0 | step=23940/50000 progress=47.9% loss=0.156088 lr=5.760935e-05 lr_backbone=5.760935e-05 elapsed=24:24:13 avg_step=3.67s recent_step=4.18s eta=26:33:53 +2026-09-15 14:11:12 | INFO | train rank=0 | step=23950/50000 progress=47.9% loss=0.085830 lr=5.757666e-05 lr_backbone=5.757666e-05 elapsed=24:24:44 avg_step=3.67s recent_step=3.14s eta=26:33:10 +2026-09-15 14:12:05 | INFO | train rank=0 | step=23960/50000 progress=47.9% loss=0.096136 lr=5.754397e-05 lr_backbone=5.754397e-05 elapsed=24:25:38 avg_step=3.67s recent_step=5.31s eta=26:32:52 +2026-09-15 14:12:40 | INFO | train rank=0 | step=23970/50000 progress=47.9% loss=0.098585 lr=5.751128e-05 lr_backbone=5.751128e-05 elapsed=24:26:13 avg_step=3.67s recent_step=3.56s eta=26:32:14 +2026-09-15 14:13:13 | INFO | train rank=0 | step=23980/50000 progress=48.0% loss=0.124243 lr=5.747858e-05 lr_backbone=5.747858e-05 elapsed=24:26:45 avg_step=3.67s recent_step=3.22s eta=26:31:32 +2026-09-15 14:13:48 | INFO | train rank=0 | step=23990/50000 progress=48.0% loss=0.092945 lr=5.744588e-05 lr_backbone=5.744588e-05 elapsed=24:27:21 avg_step=3.67s recent_step=3.52s eta=26:30:54 +2026-09-15 14:14:37 | INFO | train rank=0 | step=24000/50000 progress=48.0% loss=0.058132 lr=5.741318e-05 lr_backbone=5.741318e-05 elapsed=24:28:09 avg_step=3.67s recent_step=4.88s eta=26:30:30 +2026-09-15 14:15:10 | INFO | train rank=0 | step=24010/50000 progress=48.0% loss=0.098347 lr=5.738048e-05 lr_backbone=5.738048e-05 elapsed=24:28:43 avg_step=3.67s recent_step=3.34s eta=26:29:50 +2026-09-15 14:15:41 | INFO | train rank=0 | step=24020/50000 progress=48.0% loss=0.109313 lr=5.734777e-05 lr_backbone=5.734777e-05 elapsed=24:29:14 avg_step=3.67s recent_step=3.13s eta=26:29:07 +2026-09-15 14:16:19 | INFO | train rank=0 | step=24030/50000 progress=48.1% loss=0.094563 lr=5.731506e-05 lr_backbone=5.731506e-05 elapsed=24:29:52 avg_step=3.67s recent_step=3.79s eta=26:28:32 +2026-09-15 14:17:10 | INFO | train rank=0 | step=24040/50000 progress=48.1% loss=0.089429 lr=5.728234e-05 lr_backbone=5.728234e-05 elapsed=24:30:43 avg_step=3.67s recent_step=5.07s eta=26:28:10 +2026-09-15 14:17:50 | INFO | train rank=0 | step=24050/50000 progress=48.1% loss=0.096218 lr=5.724962e-05 lr_backbone=5.724962e-05 elapsed=24:31:23 avg_step=3.67s recent_step=4.05s eta=26:27:38 +2026-09-15 14:18:20 | INFO | train rank=0 | step=24060/50000 progress=48.1% loss=0.115121 lr=5.721690e-05 lr_backbone=5.721690e-05 elapsed=24:31:53 avg_step=3.67s recent_step=2.95s eta=26:26:53 +2026-09-15 14:18:58 | INFO | train rank=0 | step=24070/50000 progress=48.1% loss=0.101445 lr=5.718418e-05 lr_backbone=5.718418e-05 elapsed=24:32:30 avg_step=3.67s recent_step=3.76s eta=26:26:18 +2026-09-15 14:19:47 | INFO | train rank=0 | step=24080/50000 progress=48.2% loss=0.088357 lr=5.715145e-05 lr_backbone=5.715145e-05 elapsed=24:33:19 avg_step=3.67s recent_step=4.91s eta=26:25:54 +2026-09-15 14:20:23 | INFO | train rank=0 | step=24090/50000 progress=48.2% loss=0.094212 lr=5.711872e-05 lr_backbone=5.711872e-05 elapsed=24:33:56 avg_step=3.67s recent_step=3.61s eta=26:25:17 +2026-09-15 14:20:56 | INFO | train rank=0 | step=24100/50000 progress=48.2% loss=0.100274 lr=5.708598e-05 lr_backbone=5.708598e-05 elapsed=24:34:28 avg_step=3.67s recent_step=3.29s eta=26:24:36 +2026-09-15 14:21:29 | INFO | train rank=0 | step=24110/50000 progress=48.2% loss=0.113854 lr=5.705325e-05 lr_backbone=5.705325e-05 elapsed=24:35:02 avg_step=3.67s recent_step=3.37s eta=26:23:56 +2026-09-15 14:22:21 | INFO | train rank=0 | step=24120/50000 progress=48.2% loss=0.095222 lr=5.702051e-05 lr_backbone=5.702051e-05 elapsed=24:35:54 avg_step=3.67s recent_step=5.15s eta=26:23:35 +2026-09-15 14:22:53 | INFO | train rank=0 | step=24130/50000 progress=48.3% loss=0.118287 lr=5.698776e-05 lr_backbone=5.698776e-05 elapsed=24:36:26 avg_step=3.67s recent_step=3.24s eta=26:22:54 +2026-09-15 14:23:34 | INFO | train rank=0 | step=24140/50000 progress=48.3% loss=0.097884 lr=5.695502e-05 lr_backbone=5.695502e-05 elapsed=24:37:07 avg_step=3.67s recent_step=4.05s eta=26:22:21 +2026-09-15 14:24:04 | INFO | train rank=0 | step=24150/50000 progress=48.3% loss=0.101489 lr=5.692227e-05 lr_backbone=5.692227e-05 elapsed=24:37:37 avg_step=3.67s recent_step=3.03s eta=26:21:38 +2026-09-15 14:24:52 | INFO | train rank=0 | step=24160/50000 progress=48.3% loss=0.066746 lr=5.688951e-05 lr_backbone=5.688951e-05 elapsed=24:38:25 avg_step=3.67s recent_step=4.76s eta=26:21:13 +2026-09-15 14:25:20 | INFO | train rank=0 | step=24170/50000 progress=48.3% loss=0.108176 lr=5.685676e-05 lr_backbone=5.685676e-05 elapsed=24:38:53 avg_step=3.67s recent_step=2.88s eta=26:20:28 +2026-09-15 14:25:57 | INFO | train rank=0 | step=24180/50000 progress=48.4% loss=0.117030 lr=5.682400e-05 lr_backbone=5.682400e-05 elapsed=24:39:30 avg_step=3.67s recent_step=3.68s eta=26:19:51 +2026-09-15 14:26:31 | INFO | train rank=0 | step=24190/50000 progress=48.4% loss=0.115827 lr=5.679124e-05 lr_backbone=5.679124e-05 elapsed=24:40:03 avg_step=3.67s recent_step=3.32s eta=26:19:11 +2026-09-15 14:27:08 | INFO | train rank=0 | step=24200/50000 progress=48.4% loss=0.089818 lr=5.675847e-05 lr_backbone=5.675847e-05 elapsed=24:40:41 avg_step=3.67s recent_step=3.73s eta=26:18:34 +2026-09-15 14:27:37 | INFO | train rank=0 | step=24210/50000 progress=48.4% loss=0.113702 lr=5.672571e-05 lr_backbone=5.672571e-05 elapsed=24:41:10 avg_step=3.67s recent_step=2.95s eta=26:17:50 +2026-09-15 14:28:00 | INFO | train rank=0 | step=24220/50000 progress=48.4% loss=0.099914 lr=5.669294e-05 lr_backbone=5.669294e-05 elapsed=24:41:33 avg_step=3.67s recent_step=2.32s eta=26:16:59 +2026-09-15 14:28:27 | INFO | train rank=0 | step=24230/50000 progress=48.5% loss=0.106437 lr=5.666016e-05 lr_backbone=5.666016e-05 elapsed=24:42:00 avg_step=3.67s recent_step=2.66s eta=26:16:11 +2026-09-15 14:29:21 | INFO | train rank=0 | step=24240/50000 progress=48.5% loss=0.092483 lr=5.662739e-05 lr_backbone=5.662739e-05 elapsed=24:42:54 avg_step=3.67s recent_step=5.43s eta=26:15:53 +2026-09-15 14:29:57 | INFO | train rank=0 | step=24250/50000 progress=48.5% loss=0.120382 lr=5.659461e-05 lr_backbone=5.659461e-05 elapsed=24:43:29 avg_step=3.67s recent_step=3.53s eta=26:15:15 +2026-09-15 14:30:27 | INFO | train rank=0 | step=24260/50000 progress=48.5% loss=0.125452 lr=5.656183e-05 lr_backbone=5.656183e-05 elapsed=24:44:00 avg_step=3.67s recent_step=3.06s eta=26:14:32 +2026-09-15 14:30:59 | INFO | train rank=0 | step=24270/50000 progress=48.5% loss=0.079555 lr=5.652904e-05 lr_backbone=5.652904e-05 elapsed=24:44:32 avg_step=3.67s recent_step=3.15s eta=26:13:50 +2026-09-15 14:31:50 | INFO | train rank=0 | step=24280/50000 progress=48.6% loss=0.102608 lr=5.649625e-05 lr_backbone=5.649625e-05 elapsed=24:45:22 avg_step=3.67s recent_step=5.07s eta=26:13:28 +2026-09-15 14:32:18 | INFO | train rank=0 | step=24290/50000 progress=48.6% loss=0.086803 lr=5.646346e-05 lr_backbone=5.646346e-05 elapsed=24:45:51 avg_step=3.67s recent_step=2.88s eta=26:12:43 +2026-09-15 14:32:44 | INFO | train rank=0 | step=24300/50000 progress=48.6% loss=0.121828 lr=5.643067e-05 lr_backbone=5.643067e-05 elapsed=24:46:17 avg_step=3.67s recent_step=2.58s eta=26:11:55 +2026-09-15 14:33:18 | INFO | train rank=0 | step=24310/50000 progress=48.6% loss=0.107762 lr=5.639787e-05 lr_backbone=5.639787e-05 elapsed=24:46:51 avg_step=3.67s recent_step=3.39s eta=26:11:15 +2026-09-15 14:34:14 | INFO | train rank=0 | step=24320/50000 progress=48.6% loss=0.085851 lr=5.636508e-05 lr_backbone=5.636508e-05 elapsed=24:47:47 avg_step=3.67s recent_step=5.59s eta=26:10:59 +2026-09-15 14:34:46 | INFO | train rank=0 | step=24330/50000 progress=48.7% loss=0.112767 lr=5.633227e-05 lr_backbone=5.633227e-05 elapsed=24:48:19 avg_step=3.67s recent_step=3.20s eta=26:10:17 +2026-09-15 14:35:20 | INFO | train rank=0 | step=24340/50000 progress=48.7% loss=0.134322 lr=5.629947e-05 lr_backbone=5.629947e-05 elapsed=24:48:53 avg_step=3.67s recent_step=3.40s eta=26:09:37 +2026-09-15 14:35:49 | INFO | train rank=0 | step=24350/50000 progress=48.7% loss=0.107971 lr=5.626666e-05 lr_backbone=5.626666e-05 elapsed=24:49:22 avg_step=3.67s recent_step=2.95s eta=26:08:53 +2026-09-15 14:36:36 | INFO | train rank=0 | step=24360/50000 progress=48.7% loss=0.085225 lr=5.623385e-05 lr_backbone=5.623385e-05 elapsed=24:50:09 avg_step=3.67s recent_step=4.70s eta=26:08:27 +2026-09-15 14:37:11 | INFO | train rank=0 | step=24370/50000 progress=48.7% loss=0.116435 lr=5.620104e-05 lr_backbone=5.620104e-05 elapsed=24:50:43 avg_step=3.67s recent_step=3.41s eta=26:07:48 +2026-09-15 14:37:50 | INFO | train rank=0 | step=24380/50000 progress=48.8% loss=0.101186 lr=5.616822e-05 lr_backbone=5.616822e-05 elapsed=24:51:22 avg_step=3.67s recent_step=3.89s eta=26:07:14 +2026-09-15 14:38:20 | INFO | train rank=0 | step=24390/50000 progress=48.8% loss=0.079710 lr=5.613541e-05 lr_backbone=5.613541e-05 elapsed=24:51:53 avg_step=3.67s recent_step=3.04s eta=26:06:30 +2026-09-15 14:39:08 | INFO | train rank=0 | step=24400/50000 progress=48.8% loss=0.110523 lr=5.610258e-05 lr_backbone=5.610258e-05 elapsed=24:52:41 avg_step=3.67s recent_step=4.82s eta=26:06:06 +2026-09-15 14:39:44 | INFO | train rank=0 | step=24410/50000 progress=48.8% loss=0.100892 lr=5.606976e-05 lr_backbone=5.606976e-05 elapsed=24:53:17 avg_step=3.67s recent_step=3.56s eta=26:05:28 +2026-09-15 14:40:17 | INFO | train rank=0 | step=24420/50000 progress=48.8% loss=0.083888 lr=5.603693e-05 lr_backbone=5.603693e-05 elapsed=24:53:50 avg_step=3.67s recent_step=3.30s eta=26:04:47 +2026-09-15 14:40:49 | INFO | train rank=0 | step=24430/50000 progress=48.9% loss=0.088184 lr=5.600411e-05 lr_backbone=5.600411e-05 elapsed=24:54:21 avg_step=3.67s recent_step=3.19s eta=26:04:05 +2026-09-15 14:41:31 | INFO | train rank=0 | step=24440/50000 progress=48.9% loss=0.114104 lr=5.597127e-05 lr_backbone=5.597127e-05 elapsed=24:55:04 avg_step=3.67s recent_step=4.23s eta=26:03:35 +2026-09-15 14:42:03 | INFO | train rank=0 | step=24450/50000 progress=48.9% loss=0.122747 lr=5.593844e-05 lr_backbone=5.593844e-05 elapsed=24:55:36 avg_step=3.67s recent_step=3.24s eta=26:02:53 +2026-09-15 14:42:49 | INFO | train rank=0 | step=24460/50000 progress=48.9% loss=0.127189 lr=5.590560e-05 lr_backbone=5.590560e-05 elapsed=24:56:22 avg_step=3.67s recent_step=4.62s eta=26:02:27 +2026-09-15 14:43:23 | INFO | train rank=0 | step=24470/50000 progress=48.9% loss=0.111696 lr=5.587276e-05 lr_backbone=5.587276e-05 elapsed=24:56:55 avg_step=3.67s recent_step=3.31s eta=26:01:46 +2026-09-15 14:44:05 | INFO | train rank=0 | step=24480/50000 progress=49.0% loss=0.095279 lr=5.583992e-05 lr_backbone=5.583992e-05 elapsed=24:57:38 avg_step=3.67s recent_step=4.29s eta=26:01:16 +2026-09-15 14:44:41 | INFO | train rank=0 | step=24490/50000 progress=49.0% loss=0.088814 lr=5.580708e-05 lr_backbone=5.580708e-05 elapsed=24:58:14 avg_step=3.67s recent_step=3.56s eta=26:00:38 +2026-09-15 14:45:26 | INFO | train rank=0 | step=24500/50000 progress=49.0% loss=0.111886 lr=5.577423e-05 lr_backbone=5.577423e-05 elapsed=24:58:59 avg_step=3.67s recent_step=4.49s eta=26:00:10 +2026-09-15 14:45:57 | INFO | train rank=0 | step=24510/50000 progress=49.0% loss=0.097656 lr=5.574138e-05 lr_backbone=5.574138e-05 elapsed=24:59:30 avg_step=3.67s recent_step=3.15s eta=25:59:28 +2026-09-15 14:46:42 | INFO | train rank=0 | step=24520/50000 progress=49.0% loss=0.112707 lr=5.570853e-05 lr_backbone=5.570853e-05 elapsed=25:00:14 avg_step=3.67s recent_step=4.43s eta=25:58:59 +2026-09-15 14:47:10 | INFO | train rank=0 | step=24530/50000 progress=49.1% loss=0.096778 lr=5.567568e-05 lr_backbone=5.567568e-05 elapsed=25:00:43 avg_step=3.67s recent_step=2.88s eta=25:58:14 +2026-09-15 14:48:02 | INFO | train rank=0 | step=24540/50000 progress=49.1% loss=0.114334 lr=5.564282e-05 lr_backbone=5.564282e-05 elapsed=25:01:35 avg_step=3.67s recent_step=5.18s eta=25:57:53 +2026-09-15 14:48:38 | INFO | train rank=0 | step=24550/50000 progress=49.1% loss=0.112107 lr=5.560996e-05 lr_backbone=5.560996e-05 elapsed=25:02:11 avg_step=3.67s recent_step=3.57s eta=25:57:15 +2026-09-15 14:49:16 | INFO | train rank=0 | step=24560/50000 progress=49.1% loss=0.097273 lr=5.557710e-05 lr_backbone=5.557710e-05 elapsed=25:02:49 avg_step=3.67s recent_step=3.78s eta=25:56:39 +2026-09-15 14:49:46 | INFO | train rank=0 | step=24570/50000 progress=49.1% loss=0.084012 lr=5.554423e-05 lr_backbone=5.554423e-05 elapsed=25:03:19 avg_step=3.67s recent_step=3.03s eta=25:55:56 +2026-09-15 14:50:37 | INFO | train rank=0 | step=24580/50000 progress=49.2% loss=0.082590 lr=5.551137e-05 lr_backbone=5.551137e-05 elapsed=25:04:10 avg_step=3.67s recent_step=5.14s eta=25:55:35 +2026-09-15 14:51:08 | INFO | train rank=0 | step=24590/50000 progress=49.2% loss=0.090443 lr=5.547850e-05 lr_backbone=5.547850e-05 elapsed=25:04:41 avg_step=3.67s recent_step=3.04s eta=25:54:51 +2026-09-15 14:51:43 | INFO | train rank=0 | step=24600/50000 progress=49.2% loss=0.096391 lr=5.544563e-05 lr_backbone=5.544563e-05 elapsed=25:05:16 avg_step=3.67s recent_step=3.55s eta=25:54:13 +2026-09-15 14:52:22 | INFO | train rank=0 | step=24610/50000 progress=49.2% loss=0.103591 lr=5.541275e-05 lr_backbone=5.541275e-05 elapsed=25:05:55 avg_step=3.67s recent_step=3.84s eta=25:53:38 +2026-09-15 14:53:17 | INFO | train rank=0 | step=24620/50000 progress=49.2% loss=0.088582 lr=5.537988e-05 lr_backbone=5.537988e-05 elapsed=25:06:49 avg_step=3.67s recent_step=5.48s eta=25:53:20 +2026-09-15 14:53:45 | INFO | train rank=0 | step=24630/50000 progress=49.3% loss=0.127257 lr=5.534700e-05 lr_backbone=5.534700e-05 elapsed=25:07:18 avg_step=3.67s recent_step=2.85s eta=25:52:35 +2026-09-15 14:54:20 | INFO | train rank=0 | step=24640/50000 progress=49.3% loss=0.093186 lr=5.531412e-05 lr_backbone=5.531412e-05 elapsed=25:07:53 avg_step=3.67s recent_step=3.49s eta=25:51:56 +2026-09-15 14:54:52 | INFO | train rank=0 | step=24650/50000 progress=49.3% loss=0.085446 lr=5.528123e-05 lr_backbone=5.528123e-05 elapsed=25:08:25 avg_step=3.67s recent_step=3.19s eta=25:51:15 +2026-09-15 14:55:45 | INFO | train rank=0 | step=24660/50000 progress=49.3% loss=0.109794 lr=5.524835e-05 lr_backbone=5.524835e-05 elapsed=25:09:18 avg_step=3.67s recent_step=5.34s eta=25:50:55 +2026-09-15 14:56:22 | INFO | train rank=0 | step=24670/50000 progress=49.3% loss=0.116191 lr=5.521546e-05 lr_backbone=5.521546e-05 elapsed=25:09:54 avg_step=3.67s recent_step=3.63s eta=25:50:18 +2026-09-15 14:56:56 | INFO | train rank=0 | step=24680/50000 progress=49.4% loss=0.089473 lr=5.518257e-05 lr_backbone=5.518257e-05 elapsed=25:10:29 avg_step=3.67s recent_step=3.49s eta=25:49:39 +2026-09-15 14:57:31 | INFO | train rank=0 | step=24690/50000 progress=49.4% loss=0.127863 lr=5.514968e-05 lr_backbone=5.514968e-05 elapsed=25:11:04 avg_step=3.67s recent_step=3.43s eta=25:49:00 +2026-09-15 14:58:23 | INFO | train rank=0 | step=24700/50000 progress=49.4% loss=0.110430 lr=5.511678e-05 lr_backbone=5.511678e-05 elapsed=25:11:56 avg_step=3.67s recent_step=5.26s eta=25:48:40 +2026-09-15 14:58:54 | INFO | train rank=0 | step=24710/50000 progress=49.4% loss=0.081624 lr=5.508389e-05 lr_backbone=5.508389e-05 elapsed=25:12:27 avg_step=3.67s recent_step=3.05s eta=25:47:57 +2026-09-15 14:59:27 | INFO | train rank=0 | step=24720/50000 progress=49.4% loss=0.112494 lr=5.505099e-05 lr_backbone=5.505099e-05 elapsed=25:12:59 avg_step=3.67s recent_step=3.28s eta=25:47:16 +2026-09-15 14:59:54 | INFO | train rank=0 | step=24730/50000 progress=49.5% loss=0.075706 lr=5.501809e-05 lr_backbone=5.501809e-05 elapsed=25:13:26 avg_step=3.67s recent_step=2.70s eta=25:46:29 +2026-09-15 15:00:41 | INFO | train rank=0 | step=24740/50000 progress=49.5% loss=0.095684 lr=5.498518e-05 lr_backbone=5.498518e-05 elapsed=25:14:14 avg_step=3.67s recent_step=4.72s eta=25:46:03 +2026-09-15 15:01:15 | INFO | train rank=0 | step=24750/50000 progress=49.5% loss=0.107456 lr=5.495228e-05 lr_backbone=5.495228e-05 elapsed=25:14:48 avg_step=3.67s recent_step=3.42s eta=25:45:24 +2026-09-15 15:01:51 | INFO | train rank=0 | step=24760/50000 progress=49.5% loss=0.100465 lr=5.491937e-05 lr_backbone=5.491937e-05 elapsed=25:15:24 avg_step=3.67s recent_step=3.61s eta=25:44:47 +2026-09-15 15:02:23 | INFO | train rank=0 | step=24770/50000 progress=49.5% loss=0.097238 lr=5.488646e-05 lr_backbone=5.488646e-05 elapsed=25:15:56 avg_step=3.67s recent_step=3.20s eta=25:44:05 +2026-09-15 15:03:11 | INFO | train rank=0 | step=24780/50000 progress=49.6% loss=0.084386 lr=5.485355e-05 lr_backbone=5.485355e-05 elapsed=25:16:44 avg_step=3.67s recent_step=4.79s eta=25:43:40 +2026-09-15 15:03:46 | INFO | train rank=0 | step=24790/50000 progress=49.6% loss=0.095210 lr=5.482063e-05 lr_backbone=5.482063e-05 elapsed=25:17:18 avg_step=3.67s recent_step=3.44s eta=25:43:01 +2026-09-15 15:04:17 | INFO | train rank=0 | step=24800/50000 progress=49.6% loss=0.092540 lr=5.478772e-05 lr_backbone=5.478772e-05 elapsed=25:17:50 avg_step=3.67s recent_step=3.17s eta=25:42:19 +2026-09-15 15:04:57 | INFO | train rank=0 | step=24810/50000 progress=49.6% loss=0.091491 lr=5.475480e-05 lr_backbone=5.475480e-05 elapsed=25:18:29 avg_step=3.67s recent_step=3.94s eta=25:41:45 +2026-09-15 15:05:49 | INFO | train rank=0 | step=24820/50000 progress=49.6% loss=0.113493 lr=5.472188e-05 lr_backbone=5.472188e-05 elapsed=25:19:22 avg_step=3.67s recent_step=5.24s eta=25:41:24 +2026-09-15 15:06:23 | INFO | train rank=0 | step=24830/50000 progress=49.7% loss=0.118979 lr=5.468895e-05 lr_backbone=5.468895e-05 elapsed=25:19:56 avg_step=3.67s recent_step=3.43s eta=25:40:45 +2026-09-15 15:06:54 | INFO | train rank=0 | step=24840/50000 progress=49.7% loss=0.075435 lr=5.465603e-05 lr_backbone=5.465603e-05 elapsed=25:20:27 avg_step=3.67s recent_step=3.11s eta=25:40:02 +2026-09-15 15:07:28 | INFO | train rank=0 | step=24850/50000 progress=49.7% loss=0.088644 lr=5.462310e-05 lr_backbone=5.462310e-05 elapsed=25:21:01 avg_step=3.67s recent_step=3.39s eta=25:39:23 +2026-09-15 15:08:06 | INFO | train rank=0 | step=24860/50000 progress=49.7% loss=0.089452 lr=5.459017e-05 lr_backbone=5.459017e-05 elapsed=25:21:39 avg_step=3.67s recent_step=3.79s eta=25:38:47 +2026-09-15 15:08:44 | INFO | train rank=0 | step=24870/50000 progress=49.7% loss=0.068282 lr=5.455724e-05 lr_backbone=5.455724e-05 elapsed=25:22:17 avg_step=3.67s recent_step=3.79s eta=25:38:12 +2026-09-15 15:09:16 | INFO | train rank=0 | step=24880/50000 progress=49.8% loss=0.077049 lr=5.452431e-05 lr_backbone=5.452431e-05 elapsed=25:22:49 avg_step=3.67s recent_step=3.21s eta=25:37:30 +2026-09-15 15:09:47 | INFO | train rank=0 | step=24890/50000 progress=49.8% loss=0.089236 lr=5.449138e-05 lr_backbone=5.449138e-05 elapsed=25:23:20 avg_step=3.67s recent_step=3.10s eta=25:36:48 +2026-09-15 15:10:45 | INFO | train rank=0 | step=24900/50000 progress=49.8% loss=0.094535 lr=5.445844e-05 lr_backbone=5.445844e-05 elapsed=25:24:18 avg_step=3.67s recent_step=5.77s eta=25:36:32 +2026-09-15 15:11:19 | INFO | train rank=0 | step=24910/50000 progress=49.8% loss=0.078620 lr=5.442550e-05 lr_backbone=5.442550e-05 elapsed=25:24:51 avg_step=3.67s recent_step=3.36s eta=25:35:53 +2026-09-15 15:11:52 | INFO | train rank=0 | step=24920/50000 progress=49.8% loss=0.103129 lr=5.439256e-05 lr_backbone=5.439256e-05 elapsed=25:25:25 avg_step=3.67s recent_step=3.34s eta=25:35:12 +2026-09-15 15:12:27 | INFO | train rank=0 | step=24930/50000 progress=49.9% loss=0.100526 lr=5.435962e-05 lr_backbone=5.435962e-05 elapsed=25:26:00 avg_step=3.67s recent_step=3.47s eta=25:34:34 +2026-09-15 15:13:17 | INFO | train rank=0 | step=24940/50000 progress=49.9% loss=0.147683 lr=5.432667e-05 lr_backbone=5.432667e-05 elapsed=25:26:50 avg_step=3.67s recent_step=5.03s eta=25:34:11 +2026-09-15 15:13:52 | INFO | train rank=0 | step=24950/50000 progress=49.9% loss=0.095713 lr=5.429373e-05 lr_backbone=5.429373e-05 elapsed=25:27:25 avg_step=3.67s recent_step=3.50s eta=25:33:32 +2026-09-15 15:14:23 | INFO | train rank=0 | step=24960/50000 progress=49.9% loss=0.100451 lr=5.426078e-05 lr_backbone=5.426078e-05 elapsed=25:27:56 avg_step=3.67s recent_step=3.13s eta=25:32:50 +2026-09-15 15:14:58 | INFO | train rank=0 | step=24970/50000 progress=49.9% loss=0.104771 lr=5.422783e-05 lr_backbone=5.422783e-05 elapsed=25:28:31 avg_step=3.67s recent_step=3.46s eta=25:32:11 +2026-09-15 15:15:49 | INFO | train rank=0 | step=24980/50000 progress=50.0% loss=0.100611 lr=5.419488e-05 lr_backbone=5.419488e-05 elapsed=25:29:22 avg_step=3.67s recent_step=5.14s eta=25:31:49 +2026-09-15 15:16:20 | INFO | train rank=0 | step=24990/50000 progress=50.0% loss=0.090728 lr=5.416192e-05 lr_backbone=5.416192e-05 elapsed=25:29:52 avg_step=3.67s recent_step=3.02s eta=25:31:06 +2026-09-15 15:16:52 | INFO | train rank=0 | step=25000/50000 progress=50.0% loss=0.105296 lr=5.412897e-05 lr_backbone=5.412897e-05 elapsed=25:30:25 avg_step=3.67s recent_step=3.29s eta=25:30:25 +2026-09-15 15:17:27 | INFO | train rank=0 | step=25010/50000 progress=50.0% loss=0.129893 lr=5.409601e-05 lr_backbone=5.409601e-05 elapsed=25:31:00 avg_step=3.67s recent_step=3.48s eta=25:29:46 +2026-09-15 15:18:12 | INFO | train rank=0 | step=25020/50000 progress=50.0% loss=0.104603 lr=5.406305e-05 lr_backbone=5.406305e-05 elapsed=25:31:45 avg_step=3.67s recent_step=4.51s eta=25:29:18 +2026-09-15 15:18:44 | INFO | train rank=0 | step=25030/50000 progress=50.1% loss=0.095839 lr=5.403009e-05 lr_backbone=5.403009e-05 elapsed=25:32:17 avg_step=3.67s recent_step=3.18s eta=25:28:36 +2026-09-15 15:19:21 | INFO | train rank=0 | step=25040/50000 progress=50.1% loss=0.104829 lr=5.399713e-05 lr_backbone=5.399713e-05 elapsed=25:32:54 avg_step=3.67s recent_step=3.73s eta=25:28:00 +2026-09-15 15:19:50 | INFO | train rank=0 | step=25050/50000 progress=50.1% loss=0.170563 lr=5.396416e-05 lr_backbone=5.396416e-05 elapsed=25:33:23 avg_step=3.67s recent_step=2.84s eta=25:27:15 +2026-09-15 15:20:44 | INFO | train rank=0 | step=25060/50000 progress=50.1% loss=0.092960 lr=5.393120e-05 lr_backbone=5.393120e-05 elapsed=25:34:17 avg_step=3.67s recent_step=5.46s eta=25:26:56 +2026-09-15 15:21:17 | INFO | train rank=0 | step=25070/50000 progress=50.1% loss=0.092512 lr=5.389823e-05 lr_backbone=5.389823e-05 elapsed=25:34:50 avg_step=3.67s recent_step=3.24s eta=25:26:15 +2026-09-15 15:21:51 | INFO | train rank=0 | step=25080/50000 progress=50.2% loss=0.110048 lr=5.386526e-05 lr_backbone=5.386526e-05 elapsed=25:35:23 avg_step=3.67s recent_step=3.39s eta=25:25:36 +2026-09-15 15:22:29 | INFO | train rank=0 | step=25090/50000 progress=50.2% loss=0.103902 lr=5.383229e-05 lr_backbone=5.383229e-05 elapsed=25:36:02 avg_step=3.67s recent_step=3.87s eta=25:25:01 +2026-09-15 15:23:20 | INFO | train rank=0 | step=25100/50000 progress=50.2% loss=0.079617 lr=5.379931e-05 lr_backbone=5.379931e-05 elapsed=25:36:53 avg_step=3.67s recent_step=5.06s eta=25:24:38 +2026-09-15 15:23:57 | INFO | train rank=0 | step=25110/50000 progress=50.2% loss=0.090226 lr=5.376634e-05 lr_backbone=5.376634e-05 elapsed=25:37:30 avg_step=3.67s recent_step=3.70s eta=25:24:01 +2026-09-15 15:24:21 | INFO | train rank=0 | step=25120/50000 progress=50.2% loss=0.101025 lr=5.373336e-05 lr_backbone=5.373336e-05 elapsed=25:37:54 avg_step=3.67s recent_step=2.40s eta=25:23:12 +2026-09-15 15:24:50 | INFO | train rank=0 | step=25130/50000 progress=50.3% loss=0.117898 lr=5.370039e-05 lr_backbone=5.370039e-05 elapsed=25:38:22 avg_step=3.67s recent_step=2.88s eta=25:22:27 +2026-09-15 15:25:34 | INFO | train rank=0 | step=25140/50000 progress=50.3% loss=0.124252 lr=5.366741e-05 lr_backbone=5.366741e-05 elapsed=25:39:07 avg_step=3.67s recent_step=4.42s eta=25:21:58 +2026-09-15 15:26:14 | INFO | train rank=0 | step=25150/50000 progress=50.3% loss=0.076398 lr=5.363443e-05 lr_backbone=5.363443e-05 elapsed=25:39:46 avg_step=3.67s recent_step=3.97s eta=25:21:24 +2026-09-15 15:26:39 | INFO | train rank=0 | step=25160/50000 progress=50.3% loss=0.075415 lr=5.360144e-05 lr_backbone=5.360144e-05 elapsed=25:40:11 avg_step=3.67s recent_step=2.51s eta=25:20:36 +2026-09-15 15:27:09 | INFO | train rank=0 | step=25170/50000 progress=50.3% loss=0.093668 lr=5.356846e-05 lr_backbone=5.356846e-05 elapsed=25:40:41 avg_step=3.67s recent_step=2.99s eta=25:19:53 +2026-09-15 15:27:58 | INFO | train rank=0 | step=25180/50000 progress=50.4% loss=0.089310 lr=5.353547e-05 lr_backbone=5.353547e-05 elapsed=25:41:31 avg_step=3.67s recent_step=4.95s eta=25:19:29 +2026-09-15 15:28:36 | INFO | train rank=0 | step=25190/50000 progress=50.4% loss=0.125167 lr=5.350249e-05 lr_backbone=5.350249e-05 elapsed=25:42:09 avg_step=3.67s recent_step=3.79s eta=25:18:53 +2026-09-15 15:29:06 | INFO | train rank=0 | step=25200/50000 progress=50.4% loss=0.090564 lr=5.346950e-05 lr_backbone=5.346950e-05 elapsed=25:42:39 avg_step=3.67s recent_step=3.02s eta=25:18:10 +2026-09-15 15:29:48 | INFO | train rank=0 | step=25210/50000 progress=50.4% loss=0.087735 lr=5.343651e-05 lr_backbone=5.343651e-05 elapsed=25:43:21 avg_step=3.67s recent_step=4.16s eta=25:17:38 +2026-09-15 15:30:35 | INFO | train rank=0 | step=25220/50000 progress=50.4% loss=0.108546 lr=5.340351e-05 lr_backbone=5.340351e-05 elapsed=25:44:08 avg_step=3.67s recent_step=4.75s eta=25:17:12 +2026-09-15 15:31:16 | INFO | train rank=0 | step=25230/50000 progress=50.5% loss=0.094179 lr=5.337052e-05 lr_backbone=5.337052e-05 elapsed=25:44:49 avg_step=3.67s recent_step=4.06s eta=25:16:39 +2026-09-15 15:31:38 | INFO | train rank=0 | step=25240/50000 progress=50.5% loss=0.109713 lr=5.333753e-05 lr_backbone=5.333753e-05 elapsed=25:45:11 avg_step=3.67s recent_step=2.25s eta=25:15:48 +2026-09-15 15:32:15 | INFO | train rank=0 | step=25250/50000 progress=50.5% loss=0.084117 lr=5.330453e-05 lr_backbone=5.330453e-05 elapsed=25:45:48 avg_step=3.67s recent_step=3.67s eta=25:15:11 +2026-09-15 15:33:04 | INFO | train rank=0 | step=25260/50000 progress=50.5% loss=0.087662 lr=5.327153e-05 lr_backbone=5.327153e-05 elapsed=25:46:37 avg_step=3.67s recent_step=4.90s eta=25:14:47 +2026-09-15 15:33:33 | INFO | train rank=0 | step=25270/50000 progress=50.5% loss=0.113678 lr=5.323853e-05 lr_backbone=5.323853e-05 elapsed=25:47:06 avg_step=3.67s recent_step=2.87s eta=25:14:02 +2026-09-15 15:34:07 | INFO | train rank=0 | step=25280/50000 progress=50.6% loss=0.121665 lr=5.320553e-05 lr_backbone=5.320553e-05 elapsed=25:47:40 avg_step=3.67s recent_step=3.42s eta=25:13:23 +2026-09-15 15:34:46 | INFO | train rank=0 | step=25290/50000 progress=50.6% loss=0.088876 lr=5.317253e-05 lr_backbone=5.317253e-05 elapsed=25:48:19 avg_step=3.67s recent_step=3.92s eta=25:12:48 +2026-09-15 15:35:29 | INFO | train rank=0 | step=25300/50000 progress=50.6% loss=0.086407 lr=5.313953e-05 lr_backbone=5.313953e-05 elapsed=25:49:02 avg_step=3.67s recent_step=4.33s eta=25:12:18 +2026-09-15 15:35:57 | INFO | train rank=0 | step=25310/50000 progress=50.6% loss=0.103666 lr=5.310652e-05 lr_backbone=5.310652e-05 elapsed=25:49:30 avg_step=3.67s recent_step=2.79s eta=25:11:33 +2026-09-15 15:36:29 | INFO | train rank=0 | step=25320/50000 progress=50.6% loss=0.091794 lr=5.307351e-05 lr_backbone=5.307351e-05 elapsed=25:50:02 avg_step=3.67s recent_step=3.18s eta=25:10:51 +2026-09-15 15:36:59 | INFO | train rank=0 | step=25330/50000 progress=50.7% loss=0.117273 lr=5.304051e-05 lr_backbone=5.304051e-05 elapsed=25:50:32 avg_step=3.67s recent_step=2.98s eta=25:10:08 +2026-09-15 15:37:35 | INFO | train rank=0 | step=25340/50000 progress=50.7% loss=0.098279 lr=5.300750e-05 lr_backbone=5.300750e-05 elapsed=25:51:08 avg_step=3.67s recent_step=3.60s eta=25:09:30 +2026-09-15 15:38:12 | INFO | train rank=0 | step=25350/50000 progress=50.7% loss=0.103103 lr=5.297449e-05 lr_backbone=5.297449e-05 elapsed=25:51:45 avg_step=3.67s recent_step=3.72s eta=25:08:54 +2026-09-15 15:38:43 | INFO | train rank=0 | step=25360/50000 progress=50.7% loss=0.087096 lr=5.294148e-05 lr_backbone=5.294148e-05 elapsed=25:52:16 avg_step=3.67s recent_step=3.07s eta=25:08:11 +2026-09-15 15:39:14 | INFO | train rank=0 | step=25370/50000 progress=50.7% loss=0.109713 lr=5.290846e-05 lr_backbone=5.290846e-05 elapsed=25:52:47 avg_step=3.67s recent_step=3.10s eta=25:07:29 +2026-09-15 15:40:15 | INFO | train rank=0 | step=25380/50000 progress=50.8% loss=0.110140 lr=5.287545e-05 lr_backbone=5.287545e-05 elapsed=25:53:47 avg_step=3.67s recent_step=6.08s eta=25:07:16 +2026-09-15 15:40:44 | INFO | train rank=0 | step=25390/50000 progress=50.8% loss=0.098218 lr=5.284243e-05 lr_backbone=5.284243e-05 elapsed=25:54:17 avg_step=3.67s recent_step=2.91s eta=25:06:32 +2026-09-15 15:41:07 | INFO | train rank=0 | step=25400/50000 progress=50.8% loss=0.101833 lr=5.280942e-05 lr_backbone=5.280942e-05 elapsed=25:54:40 avg_step=3.67s recent_step=2.30s eta=25:05:42 +2026-09-15 15:41:48 | INFO | train rank=0 | step=25410/50000 progress=50.8% loss=0.079742 lr=5.277640e-05 lr_backbone=5.277640e-05 elapsed=25:55:20 avg_step=3.67s recent_step=4.08s eta=25:05:09 +2026-09-15 15:42:39 | INFO | train rank=0 | step=25420/50000 progress=50.8% loss=0.094002 lr=5.274338e-05 lr_backbone=5.274338e-05 elapsed=25:56:12 avg_step=3.67s recent_step=5.12s eta=25:04:46 +2026-09-15 15:43:18 | INFO | train rank=0 | step=25430/50000 progress=50.9% loss=0.117273 lr=5.271036e-05 lr_backbone=5.271036e-05 elapsed=25:56:51 avg_step=3.67s recent_step=3.89s eta=25:04:12 +2026-09-15 15:43:52 | INFO | train rank=0 | step=25440/50000 progress=50.9% loss=0.116515 lr=5.267734e-05 lr_backbone=5.267734e-05 elapsed=25:57:25 avg_step=3.67s recent_step=3.43s eta=25:03:32 +2026-09-15 15:44:28 | INFO | train rank=0 | step=25450/50000 progress=50.9% loss=0.079556 lr=5.264432e-05 lr_backbone=5.264432e-05 elapsed=25:58:01 avg_step=3.67s recent_step=3.58s eta=25:02:55 +2026-09-15 15:45:08 | INFO | train rank=0 | step=25460/50000 progress=50.9% loss=0.095364 lr=5.261129e-05 lr_backbone=5.261129e-05 elapsed=25:58:41 avg_step=3.67s recent_step=4.04s eta=25:02:22 +2026-09-15 15:45:46 | INFO | train rank=0 | step=25470/50000 progress=50.9% loss=0.092055 lr=5.257827e-05 lr_backbone=5.257827e-05 elapsed=25:59:19 avg_step=3.67s recent_step=3.83s eta=25:01:46 +2026-09-15 15:46:21 | INFO | train rank=0 | step=25480/50000 progress=51.0% loss=0.108794 lr=5.254524e-05 lr_backbone=5.254524e-05 elapsed=25:59:54 avg_step=3.67s recent_step=3.43s eta=25:01:07 +2026-09-15 15:46:54 | INFO | train rank=0 | step=25490/50000 progress=51.0% loss=0.075150 lr=5.251222e-05 lr_backbone=5.251222e-05 elapsed=26:00:26 avg_step=3.67s recent_step=3.28s eta=25:00:27 +2026-09-15 15:47:48 | INFO | train rank=0 | step=25500/50000 progress=51.0% loss=0.097687 lr=5.247919e-05 lr_backbone=5.247919e-05 elapsed=26:01:21 avg_step=3.67s recent_step=5.49s eta=25:00:07 +2026-09-15 15:48:20 | INFO | train rank=0 | step=25510/50000 progress=51.0% loss=0.101445 lr=5.244616e-05 lr_backbone=5.244616e-05 elapsed=26:01:52 avg_step=3.67s recent_step=3.11s eta=24:59:25 +2026-09-15 15:48:51 | INFO | train rank=0 | step=25520/50000 progress=51.0% loss=0.078696 lr=5.241313e-05 lr_backbone=5.241313e-05 elapsed=26:02:24 avg_step=3.67s recent_step=3.13s eta=24:58:43 +2026-09-15 15:49:30 | INFO | train rank=0 | step=25530/50000 progress=51.1% loss=0.107979 lr=5.238010e-05 lr_backbone=5.238010e-05 elapsed=26:03:03 avg_step=3.67s recent_step=3.90s eta=24:58:09 +2026-09-15 15:50:30 | INFO | train rank=0 | step=25540/50000 progress=51.1% loss=0.076613 lr=5.234706e-05 lr_backbone=5.234706e-05 elapsed=26:04:03 avg_step=3.67s recent_step=6.00s eta=24:57:54 +2026-09-15 15:51:03 | INFO | train rank=0 | step=25550/50000 progress=51.1% loss=0.072195 lr=5.231403e-05 lr_backbone=5.231403e-05 elapsed=26:04:35 avg_step=3.67s recent_step=3.27s eta=24:57:14 +2026-09-15 15:51:33 | INFO | train rank=0 | step=25560/50000 progress=51.1% loss=0.104995 lr=5.228100e-05 lr_backbone=5.228100e-05 elapsed=26:05:06 avg_step=3.67s recent_step=3.01s eta=24:56:31 +2026-09-15 15:52:12 | INFO | train rank=0 | step=25570/50000 progress=51.1% loss=0.137745 lr=5.224796e-05 lr_backbone=5.224796e-05 elapsed=26:05:45 avg_step=3.67s recent_step=3.91s eta=24:55:56 +2026-09-15 15:53:03 | INFO | train rank=0 | step=25580/50000 progress=51.2% loss=0.089250 lr=5.221492e-05 lr_backbone=5.221492e-05 elapsed=26:06:35 avg_step=3.67s recent_step=5.07s eta=24:55:33 +2026-09-15 15:53:37 | INFO | train rank=0 | step=25590/50000 progress=51.2% loss=0.132725 lr=5.218189e-05 lr_backbone=5.218189e-05 elapsed=26:07:10 avg_step=3.67s recent_step=3.42s eta=24:54:54 +2026-09-15 15:54:07 | INFO | train rank=0 | step=25600/50000 progress=51.2% loss=0.080564 lr=5.214885e-05 lr_backbone=5.214885e-05 elapsed=26:07:40 avg_step=3.67s recent_step=3.02s eta=24:54:11 +2026-09-15 15:54:38 | INFO | train rank=0 | step=25610/50000 progress=51.2% loss=0.091156 lr=5.211581e-05 lr_backbone=5.211581e-05 elapsed=26:08:11 avg_step=3.67s recent_step=3.09s eta=24:53:28 +2026-09-15 15:55:20 | INFO | train rank=0 | step=25620/50000 progress=51.2% loss=0.087852 lr=5.208277e-05 lr_backbone=5.208277e-05 elapsed=26:08:53 avg_step=3.67s recent_step=4.21s eta=24:52:57 +2026-09-15 15:55:54 | INFO | train rank=0 | step=25630/50000 progress=51.3% loss=0.109395 lr=5.204973e-05 lr_backbone=5.204973e-05 elapsed=26:09:27 avg_step=3.67s recent_step=3.43s eta=24:52:18 +2026-09-15 15:56:26 | INFO | train rank=0 | step=25640/50000 progress=51.3% loss=0.106374 lr=5.201669e-05 lr_backbone=5.201669e-05 elapsed=26:09:59 avg_step=3.67s recent_step=3.16s eta=24:51:36 +2026-09-15 15:57:15 | INFO | train rank=0 | step=25650/50000 progress=51.3% loss=0.065065 lr=5.198364e-05 lr_backbone=5.198364e-05 elapsed=26:10:48 avg_step=3.67s recent_step=4.95s eta=24:51:11 +2026-09-15 15:57:50 | INFO | train rank=0 | step=25660/50000 progress=51.3% loss=0.075875 lr=5.195060e-05 lr_backbone=5.195060e-05 elapsed=26:11:22 avg_step=3.67s recent_step=3.42s eta=24:50:32 +2026-09-15 15:58:22 | INFO | train rank=0 | step=25670/50000 progress=51.3% loss=0.094409 lr=5.191755e-05 lr_backbone=5.191755e-05 elapsed=26:11:55 avg_step=3.67s recent_step=3.26s eta=24:49:52 +2026-09-15 15:58:58 | INFO | train rank=0 | step=25680/50000 progress=51.4% loss=0.105930 lr=5.188451e-05 lr_backbone=5.188451e-05 elapsed=26:12:31 avg_step=3.67s recent_step=3.63s eta=24:49:14 +2026-09-15 15:59:33 | INFO | train rank=0 | step=25690/50000 progress=51.4% loss=0.125702 lr=5.185146e-05 lr_backbone=5.185146e-05 elapsed=26:13:06 avg_step=3.67s recent_step=3.49s eta=24:48:36 +2026-09-15 16:00:24 | INFO | train rank=0 | step=25700/50000 progress=51.4% loss=0.102527 lr=5.181842e-05 lr_backbone=5.181842e-05 elapsed=26:13:57 avg_step=3.67s recent_step=5.06s eta=24:48:12 +2026-09-15 16:00:44 | INFO | train rank=0 | step=25710/50000 progress=51.4% loss=0.123598 lr=5.178537e-05 lr_backbone=5.178537e-05 elapsed=26:14:17 avg_step=3.67s recent_step=2.00s eta=24:47:20 +2026-09-15 16:01:17 | INFO | train rank=0 | step=25720/50000 progress=51.4% loss=0.089279 lr=5.175232e-05 lr_backbone=5.175232e-05 elapsed=26:14:50 avg_step=3.67s recent_step=3.29s eta=24:46:39 +2026-09-15 16:02:07 | INFO | train rank=0 | step=25730/50000 progress=51.5% loss=0.102243 lr=5.171927e-05 lr_backbone=5.171927e-05 elapsed=26:15:39 avg_step=3.67s recent_step=4.96s eta=24:46:15 +2026-09-15 16:02:44 | INFO | train rank=0 | step=25740/50000 progress=51.5% loss=0.107729 lr=5.168622e-05 lr_backbone=5.168622e-05 elapsed=26:16:16 avg_step=3.67s recent_step=3.71s eta=24:45:38 +2026-09-15 16:03:16 | INFO | train rank=0 | step=25750/50000 progress=51.5% loss=0.094826 lr=5.165317e-05 lr_backbone=5.165317e-05 elapsed=26:16:49 avg_step=3.67s recent_step=3.21s eta=24:44:57 +2026-09-15 16:03:54 | INFO | train rank=0 | step=25760/50000 progress=51.5% loss=0.098416 lr=5.162012e-05 lr_backbone=5.162012e-05 elapsed=26:17:27 avg_step=3.67s recent_step=3.80s eta=24:44:22 +2026-09-15 16:04:45 | INFO | train rank=0 | step=25770/50000 progress=51.5% loss=0.087127 lr=5.158706e-05 lr_backbone=5.158706e-05 elapsed=26:18:18 avg_step=3.67s recent_step=5.17s eta=24:43:59 +2026-09-15 16:05:20 | INFO | train rank=0 | step=25780/50000 progress=51.6% loss=0.140251 lr=5.155401e-05 lr_backbone=5.155401e-05 elapsed=26:18:53 avg_step=3.67s recent_step=3.44s eta=24:43:20 +2026-09-15 16:05:53 | INFO | train rank=0 | step=25790/50000 progress=51.6% loss=0.107481 lr=5.152096e-05 lr_backbone=5.152096e-05 elapsed=26:19:26 avg_step=3.67s recent_step=3.31s eta=24:42:40 +2026-09-15 16:06:25 | INFO | train rank=0 | step=25800/50000 progress=51.6% loss=0.101577 lr=5.148790e-05 lr_backbone=5.148790e-05 elapsed=26:19:58 avg_step=3.67s recent_step=3.24s eta=24:41:59 +2026-09-15 16:07:02 | INFO | train rank=0 | step=25810/50000 progress=51.6% loss=0.114390 lr=5.145485e-05 lr_backbone=5.145485e-05 elapsed=26:20:34 avg_step=3.67s recent_step=3.63s eta=24:41:22 +2026-09-15 16:07:42 | INFO | train rank=0 | step=25820/50000 progress=51.6% loss=0.089259 lr=5.142179e-05 lr_backbone=5.142179e-05 elapsed=26:21:15 avg_step=3.67s recent_step=4.04s eta=24:40:49 +2026-09-15 16:08:17 | INFO | train rank=0 | step=25830/50000 progress=51.7% loss=0.108981 lr=5.138874e-05 lr_backbone=5.138874e-05 elapsed=26:21:50 avg_step=3.67s recent_step=3.52s eta=24:40:10 +2026-09-15 16:08:59 | INFO | train rank=0 | step=25840/50000 progress=51.7% loss=0.110297 lr=5.135568e-05 lr_backbone=5.135568e-05 elapsed=26:22:32 avg_step=3.67s recent_step=4.16s eta=24:39:38 +2026-09-15 16:09:46 | INFO | train rank=0 | step=25850/50000 progress=51.7% loss=0.094732 lr=5.132262e-05 lr_backbone=5.132262e-05 elapsed=26:23:18 avg_step=3.68s recent_step=4.69s eta=24:39:11 +2026-09-15 16:10:21 | INFO | train rank=0 | step=25860/50000 progress=51.7% loss=0.062761 lr=5.128956e-05 lr_backbone=5.128956e-05 elapsed=26:23:53 avg_step=3.67s recent_step=3.50s eta=24:38:33 +2026-09-15 16:10:51 | INFO | train rank=0 | step=25870/50000 progress=51.7% loss=0.122869 lr=5.125650e-05 lr_backbone=5.125650e-05 elapsed=26:24:24 avg_step=3.67s recent_step=3.03s eta=24:37:50 +2026-09-15 16:11:27 | INFO | train rank=0 | step=25880/50000 progress=51.8% loss=0.110088 lr=5.122345e-05 lr_backbone=5.122345e-05 elapsed=26:25:00 avg_step=3.67s recent_step=3.66s eta=24:37:13 +2026-09-15 16:12:21 | INFO | train rank=0 | step=25890/50000 progress=51.8% loss=0.106721 lr=5.119039e-05 lr_backbone=5.119039e-05 elapsed=26:25:54 avg_step=3.68s recent_step=5.40s eta=24:36:52 +2026-09-15 16:12:48 | INFO | train rank=0 | step=25900/50000 progress=51.8% loss=0.094108 lr=5.115733e-05 lr_backbone=5.115733e-05 elapsed=26:26:21 avg_step=3.67s recent_step=2.64s eta=24:36:06 +2026-09-15 16:13:16 | INFO | train rank=0 | step=25910/50000 progress=51.8% loss=0.097273 lr=5.112426e-05 lr_backbone=5.112426e-05 elapsed=26:26:48 avg_step=3.67s recent_step=2.78s eta=24:35:21 +2026-09-15 16:13:45 | INFO | train rank=0 | step=25920/50000 progress=51.8% loss=0.125674 lr=5.109120e-05 lr_backbone=5.109120e-05 elapsed=26:27:18 avg_step=3.67s recent_step=2.98s eta=24:34:38 +2026-09-15 16:14:25 | INFO | train rank=0 | step=25930/50000 progress=51.9% loss=0.110940 lr=5.105814e-05 lr_backbone=5.105814e-05 elapsed=26:27:58 avg_step=3.67s recent_step=3.97s eta=24:34:03 +2026-09-15 16:15:12 | INFO | train rank=0 | step=25940/50000 progress=51.9% loss=0.076653 lr=5.102508e-05 lr_backbone=5.102508e-05 elapsed=26:28:45 avg_step=3.67s recent_step=4.72s eta=24:33:36 +2026-09-15 16:15:48 | INFO | train rank=0 | step=25950/50000 progress=51.9% loss=0.093101 lr=5.099202e-05 lr_backbone=5.099202e-05 elapsed=26:29:21 avg_step=3.67s recent_step=3.61s eta=24:32:59 +2026-09-15 16:16:24 | INFO | train rank=0 | step=25960/50000 progress=51.9% loss=0.086537 lr=5.095895e-05 lr_backbone=5.095895e-05 elapsed=26:29:57 avg_step=3.67s recent_step=3.54s eta=24:32:21 +2026-09-15 16:16:53 | INFO | train rank=0 | step=25970/50000 progress=51.9% loss=0.088348 lr=5.092589e-05 lr_backbone=5.092589e-05 elapsed=26:30:25 avg_step=3.67s recent_step=2.88s eta=24:31:37 +2026-09-15 16:17:36 | INFO | train rank=0 | step=25980/50000 progress=52.0% loss=0.068436 lr=5.089283e-05 lr_backbone=5.089283e-05 elapsed=26:31:09 avg_step=3.67s recent_step=4.35s eta=24:31:07 +2026-09-15 16:18:05 | INFO | train rank=0 | step=25990/50000 progress=52.0% loss=0.097210 lr=5.085976e-05 lr_backbone=5.085976e-05 elapsed=26:31:38 avg_step=3.67s recent_step=2.90s eta=24:30:23 +2026-09-15 16:18:38 | INFO | train rank=0 | step=26000/50000 progress=52.0% loss=0.111717 lr=5.082670e-05 lr_backbone=5.082670e-05 elapsed=26:32:11 avg_step=3.67s recent_step=3.30s eta=24:29:42 +2026-09-15 16:19:32 | INFO | train rank=0 | step=26010/50000 progress=52.0% loss=0.115528 lr=5.079363e-05 lr_backbone=5.079363e-05 elapsed=26:33:05 avg_step=3.67s recent_step=5.39s eta=24:29:22 +2026-09-15 16:20:01 | INFO | train rank=0 | step=26020/50000 progress=52.0% loss=0.090162 lr=5.076057e-05 lr_backbone=5.076057e-05 elapsed=26:33:34 avg_step=3.67s recent_step=2.90s eta=24:28:38 +2026-09-15 16:20:31 | INFO | train rank=0 | step=26030/50000 progress=52.1% loss=0.115580 lr=5.072750e-05 lr_backbone=5.072750e-05 elapsed=26:34:04 avg_step=3.67s recent_step=2.99s eta=24:27:55 +2026-09-15 16:21:04 | INFO | train rank=0 | step=26040/50000 progress=52.1% loss=0.099851 lr=5.069443e-05 lr_backbone=5.069443e-05 elapsed=26:34:37 avg_step=3.67s recent_step=3.34s eta=24:27:15 +2026-09-15 16:21:44 | INFO | train rank=0 | step=26050/50000 progress=52.1% loss=0.099293 lr=5.066137e-05 lr_backbone=5.066137e-05 elapsed=26:35:16 avg_step=3.67s recent_step=3.91s eta=24:26:40 +2026-09-15 16:22:22 | INFO | train rank=0 | step=26060/50000 progress=52.1% loss=0.103660 lr=5.062830e-05 lr_backbone=5.062830e-05 elapsed=26:35:55 avg_step=3.67s recent_step=3.82s eta=24:26:05 +2026-09-15 16:23:04 | INFO | train rank=0 | step=26070/50000 progress=52.1% loss=0.090848 lr=5.059524e-05 lr_backbone=5.059524e-05 elapsed=26:36:36 avg_step=3.67s recent_step=4.18s eta=24:25:33 +2026-09-15 16:23:47 | INFO | train rank=0 | step=26080/50000 progress=52.2% loss=0.081274 lr=5.056217e-05 lr_backbone=5.056217e-05 elapsed=26:37:19 avg_step=3.67s recent_step=4.30s eta=24:25:02 +2026-09-15 16:24:22 | INFO | train rank=0 | step=26090/50000 progress=52.2% loss=0.093957 lr=5.052910e-05 lr_backbone=5.052910e-05 elapsed=26:37:54 avg_step=3.67s recent_step=3.50s eta=24:24:23 +2026-09-15 16:25:04 | INFO | train rank=0 | step=26100/50000 progress=52.2% loss=0.100272 lr=5.049603e-05 lr_backbone=5.049603e-05 elapsed=26:38:36 avg_step=3.67s recent_step=4.20s eta=24:23:51 +2026-09-15 16:25:39 | INFO | train rank=0 | step=26110/50000 progress=52.2% loss=0.152593 lr=5.046296e-05 lr_backbone=5.046296e-05 elapsed=26:39:11 avg_step=3.67s recent_step=3.51s eta=24:23:13 +2026-09-15 16:26:15 | INFO | train rank=0 | step=26120/50000 progress=52.2% loss=0.092698 lr=5.042990e-05 lr_backbone=5.042990e-05 elapsed=26:39:48 avg_step=3.67s recent_step=3.61s eta=24:22:36 +2026-09-15 16:26:52 | INFO | train rank=0 | step=26130/50000 progress=52.3% loss=0.113978 lr=5.039683e-05 lr_backbone=5.039683e-05 elapsed=26:40:25 avg_step=3.67s recent_step=3.71s eta=24:21:59 +2026-09-15 16:27:33 | INFO | train rank=0 | step=26140/50000 progress=52.3% loss=0.109052 lr=5.036376e-05 lr_backbone=5.036376e-05 elapsed=26:41:06 avg_step=3.68s recent_step=4.16s eta=24:21:27 +2026-09-15 16:28:07 | INFO | train rank=0 | step=26150/50000 progress=52.3% loss=0.099281 lr=5.033069e-05 lr_backbone=5.033069e-05 elapsed=26:41:40 avg_step=3.67s recent_step=3.34s eta=24:20:47 +2026-09-15 16:28:58 | INFO | train rank=0 | step=26160/50000 progress=52.3% loss=0.103178 lr=5.029762e-05 lr_backbone=5.029762e-05 elapsed=26:42:31 avg_step=3.68s recent_step=5.14s eta=24:20:24 +2026-09-15 16:29:28 | INFO | train rank=0 | step=26170/50000 progress=52.3% loss=0.086017 lr=5.026455e-05 lr_backbone=5.026455e-05 elapsed=26:43:00 avg_step=3.68s recent_step=2.94s eta=24:19:40 +2026-09-15 16:30:04 | INFO | train rank=0 | step=26180/50000 progress=52.4% loss=0.106483 lr=5.023148e-05 lr_backbone=5.023148e-05 elapsed=26:43:36 avg_step=3.68s recent_step=3.60s eta=24:19:03 +2026-09-15 16:30:36 | INFO | train rank=0 | step=26190/50000 progress=52.4% loss=0.100778 lr=5.019842e-05 lr_backbone=5.019842e-05 elapsed=26:44:09 avg_step=3.68s recent_step=3.22s eta=24:18:22 +2026-09-15 16:31:12 | INFO | train rank=0 | step=26200/50000 progress=52.4% loss=0.131438 lr=5.016535e-05 lr_backbone=5.016535e-05 elapsed=26:44:44 avg_step=3.67s recent_step=3.56s eta=24:17:44 +2026-09-15 16:31:50 | INFO | train rank=0 | step=26210/50000 progress=52.4% loss=0.076372 lr=5.013228e-05 lr_backbone=5.013228e-05 elapsed=26:45:23 avg_step=3.68s recent_step=3.83s eta=24:17:09 +2026-09-15 16:32:22 | INFO | train rank=0 | step=26220/50000 progress=52.4% loss=0.118267 lr=5.009921e-05 lr_backbone=5.009921e-05 elapsed=26:45:55 avg_step=3.67s recent_step=3.24s eta=24:16:28 +2026-09-15 16:32:59 | INFO | train rank=0 | step=26230/50000 progress=52.5% loss=0.085935 lr=5.006614e-05 lr_backbone=5.006614e-05 elapsed=26:46:32 avg_step=3.67s recent_step=3.70s eta=24:15:52 +2026-09-15 16:33:50 | INFO | train rank=0 | step=26240/50000 progress=52.5% loss=0.104149 lr=5.003307e-05 lr_backbone=5.003307e-05 elapsed=26:47:23 avg_step=3.68s recent_step=5.06s eta=24:15:28 +2026-09-15 16:34:33 | INFO | train rank=0 | step=26250/50000 progress=52.5% loss=0.082050 lr=5.000000e-05 lr_backbone=5.000000e-05 elapsed=26:48:06 avg_step=3.68s recent_step=4.30s eta=24:14:56 +2026-09-15 16:35:05 | INFO | train rank=0 | step=26260/50000 progress=52.5% loss=0.113279 lr=4.996693e-05 lr_backbone=4.996693e-05 elapsed=26:48:38 avg_step=3.68s recent_step=3.23s eta=24:14:16 +2026-09-15 16:35:37 | INFO | train rank=0 | step=26270/50000 progress=52.5% loss=0.107541 lr=4.993386e-05 lr_backbone=4.993386e-05 elapsed=26:49:10 avg_step=3.68s recent_step=3.18s eta=24:13:34 +2026-09-15 16:36:20 | INFO | train rank=0 | step=26280/50000 progress=52.6% loss=0.103258 lr=4.990079e-05 lr_backbone=4.990079e-05 elapsed=26:49:53 avg_step=3.68s recent_step=4.31s eta=24:13:03 +2026-09-15 16:36:57 | INFO | train rank=0 | step=26290/50000 progress=52.6% loss=0.086291 lr=4.986772e-05 lr_backbone=4.986772e-05 elapsed=26:50:30 avg_step=3.68s recent_step=3.71s eta=24:12:27 +2026-09-15 16:37:33 | INFO | train rank=0 | step=26300/50000 progress=52.6% loss=0.102024 lr=4.983465e-05 lr_backbone=4.983465e-05 elapsed=26:51:06 avg_step=3.68s recent_step=3.59s eta=24:11:49 +2026-09-15 16:38:10 | INFO | train rank=0 | step=26310/50000 progress=52.6% loss=0.099066 lr=4.980158e-05 lr_backbone=4.980158e-05 elapsed=26:51:43 avg_step=3.68s recent_step=3.74s eta=24:11:13 +2026-09-15 16:38:42 | INFO | train rank=0 | step=26320/50000 progress=52.6% loss=0.083184 lr=4.976852e-05 lr_backbone=4.976852e-05 elapsed=26:52:15 avg_step=3.68s recent_step=3.14s eta=24:10:32 +2026-09-15 16:39:24 | INFO | train rank=0 | step=26330/50000 progress=52.7% loss=0.117959 lr=4.973545e-05 lr_backbone=4.973545e-05 elapsed=26:52:57 avg_step=3.68s recent_step=4.22s eta=24:10:00 +2026-09-15 16:39:57 | INFO | train rank=0 | step=26340/50000 progress=52.7% loss=0.123065 lr=4.970238e-05 lr_backbone=4.970238e-05 elapsed=26:53:30 avg_step=3.68s recent_step=3.34s eta=24:09:20 +2026-09-15 16:40:34 | INFO | train rank=0 | step=26350/50000 progress=52.7% loss=0.077650 lr=4.966931e-05 lr_backbone=4.966931e-05 elapsed=26:54:07 avg_step=3.68s recent_step=3.63s eta=24:08:43 +2026-09-15 16:41:14 | INFO | train rank=0 | step=26360/50000 progress=52.7% loss=0.087724 lr=4.963624e-05 lr_backbone=4.963624e-05 elapsed=26:54:47 avg_step=3.68s recent_step=4.06s eta=24:08:10 +2026-09-15 16:41:54 | INFO | train rank=0 | step=26370/50000 progress=52.7% loss=0.098089 lr=4.960317e-05 lr_backbone=4.960317e-05 elapsed=26:55:27 avg_step=3.68s recent_step=3.96s eta=24:07:36 +2026-09-15 16:42:25 | INFO | train rank=0 | step=26380/50000 progress=52.8% loss=0.093036 lr=4.957010e-05 lr_backbone=4.957010e-05 elapsed=26:55:58 avg_step=3.68s recent_step=3.14s eta=24:06:54 +2026-09-15 16:43:08 | INFO | train rank=0 | step=26390/50000 progress=52.8% loss=0.091228 lr=4.953704e-05 lr_backbone=4.953704e-05 elapsed=26:56:41 avg_step=3.68s recent_step=4.23s eta=24:06:22 +2026-09-15 16:43:39 | INFO | train rank=0 | step=26400/50000 progress=52.8% loss=0.118739 lr=4.950397e-05 lr_backbone=4.950397e-05 elapsed=26:57:11 avg_step=3.68s recent_step=3.08s eta=24:05:40 +2026-09-15 16:44:24 | INFO | train rank=0 | step=26410/50000 progress=52.8% loss=0.113286 lr=4.947090e-05 lr_backbone=4.947090e-05 elapsed=26:57:57 avg_step=3.68s recent_step=4.58s eta=24:05:11 +2026-09-15 16:44:53 | INFO | train rank=0 | step=26420/50000 progress=52.8% loss=0.092066 lr=4.943783e-05 lr_backbone=4.943783e-05 elapsed=26:58:26 avg_step=3.68s recent_step=2.87s eta=24:04:27 +2026-09-15 16:45:38 | INFO | train rank=0 | step=26430/50000 progress=52.9% loss=0.091338 lr=4.940476e-05 lr_backbone=4.940476e-05 elapsed=26:59:11 avg_step=3.68s recent_step=4.50s eta=24:03:58 +2026-09-15 16:46:08 | INFO | train rank=0 | step=26440/50000 progress=52.9% loss=0.110854 lr=4.937170e-05 lr_backbone=4.937170e-05 elapsed=26:59:41 avg_step=3.68s recent_step=3.03s eta=24:03:16 +2026-09-15 16:46:51 | INFO | train rank=0 | step=26450/50000 progress=52.9% loss=0.098352 lr=4.933863e-05 lr_backbone=4.933863e-05 elapsed=27:00:23 avg_step=3.68s recent_step=4.23s eta=24:02:44 +2026-09-15 16:47:24 | INFO | train rank=0 | step=26460/50000 progress=52.9% loss=0.093250 lr=4.930557e-05 lr_backbone=4.930557e-05 elapsed=27:00:57 avg_step=3.68s recent_step=3.31s eta=24:02:04 +2026-09-15 16:48:08 | INFO | train rank=0 | step=26470/50000 progress=52.9% loss=0.113941 lr=4.927250e-05 lr_backbone=4.927250e-05 elapsed=27:01:41 avg_step=3.68s recent_step=4.46s eta=24:01:34 +2026-09-15 16:48:47 | INFO | train rank=0 | step=26480/50000 progress=53.0% loss=0.086940 lr=4.923943e-05 lr_backbone=4.923943e-05 elapsed=27:02:20 avg_step=3.68s recent_step=3.84s eta=24:00:59 +2026-09-15 16:49:18 | INFO | train rank=0 | step=26490/50000 progress=53.0% loss=0.109710 lr=4.920637e-05 lr_backbone=4.920637e-05 elapsed=27:02:51 avg_step=3.68s recent_step=3.12s eta=24:00:17 +2026-09-15 16:49:52 | INFO | train rank=0 | step=26500/50000 progress=53.0% loss=0.103528 lr=4.917330e-05 lr_backbone=4.917330e-05 elapsed=27:03:25 avg_step=3.68s recent_step=3.39s eta=23:59:38 +2026-09-15 16:50:30 | INFO | train rank=0 | step=26510/50000 progress=53.0% loss=0.114830 lr=4.914024e-05 lr_backbone=4.914024e-05 elapsed=27:04:03 avg_step=3.68s recent_step=3.81s eta=23:59:02 +2026-09-15 16:51:05 | INFO | train rank=0 | step=26520/50000 progress=53.0% loss=0.109052 lr=4.910717e-05 lr_backbone=4.910717e-05 elapsed=27:04:38 avg_step=3.68s recent_step=3.53s eta=23:58:24 +2026-09-15 16:51:51 | INFO | train rank=0 | step=26530/50000 progress=53.1% loss=0.105070 lr=4.907411e-05 lr_backbone=4.907411e-05 elapsed=27:05:24 avg_step=3.68s recent_step=4.61s eta=23:57:55 +2026-09-15 16:52:27 | INFO | train rank=0 | step=26540/50000 progress=53.1% loss=0.113968 lr=4.904105e-05 lr_backbone=4.904105e-05 elapsed=27:05:59 avg_step=3.68s recent_step=3.54s eta=23:57:17 +2026-09-15 16:53:09 | INFO | train rank=0 | step=26550/50000 progress=53.1% loss=0.099631 lr=4.900798e-05 lr_backbone=4.900798e-05 elapsed=27:06:42 avg_step=3.68s recent_step=4.26s eta=23:56:46 +2026-09-15 16:53:40 | INFO | train rank=0 | step=26560/50000 progress=53.1% loss=0.094163 lr=4.897492e-05 lr_backbone=4.897492e-05 elapsed=27:07:13 avg_step=3.68s recent_step=3.11s eta=23:56:04 +2026-09-15 16:54:27 | INFO | train rank=0 | step=26570/50000 progress=53.1% loss=0.077669 lr=4.894186e-05 lr_backbone=4.894186e-05 elapsed=27:07:59 avg_step=3.68s recent_step=4.65s eta=23:55:36 +2026-09-15 16:55:02 | INFO | train rank=0 | step=26580/50000 progress=53.2% loss=0.107434 lr=4.890880e-05 lr_backbone=4.890880e-05 elapsed=27:08:34 avg_step=3.68s recent_step=3.49s eta=23:54:57 +2026-09-15 16:55:38 | INFO | train rank=0 | step=26590/50000 progress=53.2% loss=0.098095 lr=4.887574e-05 lr_backbone=4.887574e-05 elapsed=27:09:11 avg_step=3.68s recent_step=3.68s eta=23:54:21 +2026-09-15 16:56:11 | INFO | train rank=0 | step=26600/50000 progress=53.2% loss=0.114235 lr=4.884267e-05 lr_backbone=4.884267e-05 elapsed=27:09:44 avg_step=3.68s recent_step=3.31s eta=23:53:41 +2026-09-15 16:56:56 | INFO | train rank=0 | step=26610/50000 progress=53.2% loss=0.120340 lr=4.880961e-05 lr_backbone=4.880961e-05 elapsed=27:10:28 avg_step=3.68s recent_step=4.42s eta=23:53:10 +2026-09-15 16:57:29 | INFO | train rank=0 | step=26620/50000 progress=53.2% loss=0.095351 lr=4.877655e-05 lr_backbone=4.877655e-05 elapsed=27:11:02 avg_step=3.68s recent_step=3.34s eta=23:52:31 +2026-09-15 16:58:03 | INFO | train rank=0 | step=26630/50000 progress=53.3% loss=0.093062 lr=4.874350e-05 lr_backbone=4.874350e-05 elapsed=27:11:36 avg_step=3.68s recent_step=3.44s eta=23:51:52 +2026-09-15 16:58:32 | INFO | train rank=0 | step=26640/50000 progress=53.3% loss=0.116859 lr=4.871044e-05 lr_backbone=4.871044e-05 elapsed=27:12:05 avg_step=3.68s recent_step=2.83s eta=23:51:08 +2026-09-15 16:59:03 | INFO | train rank=0 | step=26650/50000 progress=53.3% loss=0.116465 lr=4.867738e-05 lr_backbone=4.867738e-05 elapsed=27:12:36 avg_step=3.68s recent_step=3.17s eta=23:50:27 +2026-09-15 16:59:39 | INFO | train rank=0 | step=26660/50000 progress=53.3% loss=0.104196 lr=4.864432e-05 lr_backbone=4.864432e-05 elapsed=27:13:12 avg_step=3.68s recent_step=3.60s eta=23:49:49 +2026-09-15 17:00:06 | INFO | train rank=0 | step=26670/50000 progress=53.3% loss=0.088025 lr=4.861126e-05 lr_backbone=4.861126e-05 elapsed=27:13:39 avg_step=3.68s recent_step=2.64s eta=23:49:03 +2026-09-15 17:00:38 | INFO | train rank=0 | step=26680/50000 progress=53.4% loss=0.093419 lr=4.857821e-05 lr_backbone=4.857821e-05 elapsed=27:14:11 avg_step=3.68s recent_step=3.21s eta=23:48:22 +2026-09-15 17:01:10 | INFO | train rank=0 | step=26690/50000 progress=53.4% loss=0.103407 lr=4.854515e-05 lr_backbone=4.854515e-05 elapsed=27:14:43 avg_step=3.67s recent_step=3.22s eta=23:47:42 +2026-09-15 17:01:51 | INFO | train rank=0 | step=26700/50000 progress=53.4% loss=0.076028 lr=4.851210e-05 lr_backbone=4.851210e-05 elapsed=27:15:24 avg_step=3.68s recent_step=4.07s eta=23:47:08 +2026-09-15 17:02:25 | INFO | train rank=0 | step=26710/50000 progress=53.4% loss=0.087181 lr=4.847904e-05 lr_backbone=4.847904e-05 elapsed=27:15:58 avg_step=3.67s recent_step=3.46s eta=23:46:30 +2026-09-15 17:02:50 | INFO | train rank=0 | step=26720/50000 progress=53.4% loss=0.115009 lr=4.844599e-05 lr_backbone=4.844599e-05 elapsed=27:16:23 avg_step=3.67s recent_step=2.49s eta=23:45:43 +2026-09-15 17:03:28 | INFO | train rank=0 | step=26730/50000 progress=53.5% loss=0.078898 lr=4.841294e-05 lr_backbone=4.841294e-05 elapsed=27:17:01 avg_step=3.67s recent_step=3.81s eta=23:45:07 +2026-09-15 17:04:24 | INFO | train rank=0 | step=26740/50000 progress=53.5% loss=0.131519 lr=4.837988e-05 lr_backbone=4.837988e-05 elapsed=27:17:57 avg_step=3.68s recent_step=5.58s eta=23:44:47 +2026-09-15 17:05:17 | INFO | train rank=0 | step=26750/50000 progress=53.5% loss=0.128876 lr=4.834683e-05 lr_backbone=4.834683e-05 elapsed=27:18:50 avg_step=3.68s recent_step=5.31s eta=23:44:25 +2026-09-15 17:06:12 | INFO | train rank=0 | step=26760/50000 progress=53.5% loss=0.097911 lr=4.831378e-05 lr_backbone=4.831378e-05 elapsed=27:19:45 avg_step=3.68s recent_step=5.50s eta=23:44:04 +2026-09-15 17:07:09 | INFO | train rank=0 | step=26770/50000 progress=53.5% loss=0.075348 lr=4.828073e-05 lr_backbone=4.828073e-05 elapsed=27:20:42 avg_step=3.68s recent_step=5.65s eta=23:43:44 +2026-09-15 17:08:16 | INFO | train rank=0 | step=26780/50000 progress=53.6% loss=0.102902 lr=4.824768e-05 lr_backbone=4.824768e-05 elapsed=27:21:49 avg_step=3.68s recent_step=6.69s eta=23:43:33 +2026-09-15 17:09:23 | INFO | train rank=0 | step=26790/50000 progress=53.6% loss=0.124125 lr=4.821463e-05 lr_backbone=4.821463e-05 elapsed=27:22:56 avg_step=3.68s recent_step=6.73s eta=23:43:23 +2026-09-15 17:10:32 | INFO | train rank=0 | step=26800/50000 progress=53.6% loss=0.141975 lr=4.818158e-05 lr_backbone=4.818158e-05 elapsed=27:24:05 avg_step=3.68s recent_step=6.92s eta=23:43:14 +2026-09-15 17:11:36 | INFO | train rank=0 | step=26810/50000 progress=53.6% loss=0.112667 lr=4.814854e-05 lr_backbone=4.814854e-05 elapsed=27:25:09 avg_step=3.68s recent_step=6.37s eta=23:43:01 +2026-09-15 17:12:46 | INFO | train rank=0 | step=26820/50000 progress=53.6% loss=0.096222 lr=4.811549e-05 lr_backbone=4.811549e-05 elapsed=27:26:19 avg_step=3.68s recent_step=6.97s eta=23:42:52 +2026-09-15 17:13:58 | INFO | train rank=0 | step=26830/50000 progress=53.7% loss=0.086036 lr=4.808245e-05 lr_backbone=4.808245e-05 elapsed=27:27:31 avg_step=3.68s recent_step=7.21s eta=23:42:46 +2026-09-15 17:15:01 | INFO | train rank=0 | step=26840/50000 progress=53.7% loss=0.071309 lr=4.804940e-05 lr_backbone=4.804940e-05 elapsed=27:28:34 avg_step=3.69s recent_step=6.31s eta=23:42:32 +2026-09-15 17:16:05 | INFO | train rank=0 | step=26850/50000 progress=53.7% loss=0.094747 lr=4.801636e-05 lr_backbone=4.801636e-05 elapsed=27:29:38 avg_step=3.69s recent_step=6.44s eta=23:42:19 +2026-09-15 17:17:13 | INFO | train rank=0 | step=26860/50000 progress=53.7% loss=0.098018 lr=4.798331e-05 lr_backbone=4.798331e-05 elapsed=27:30:46 avg_step=3.69s recent_step=6.75s eta=23:42:08 +2026-09-15 17:18:28 | INFO | train rank=0 | step=26870/50000 progress=53.7% loss=0.099871 lr=4.795027e-05 lr_backbone=4.795027e-05 elapsed=27:32:01 avg_step=3.69s recent_step=7.55s eta=23:42:05 +2026-09-15 17:19:36 | INFO | train rank=0 | step=26880/50000 progress=53.8% loss=0.088534 lr=4.791723e-05 lr_backbone=4.791723e-05 elapsed=27:33:09 avg_step=3.69s recent_step=6.78s eta=23:41:54 +2026-09-15 17:20:42 | INFO | train rank=0 | step=26890/50000 progress=53.8% loss=0.082648 lr=4.788419e-05 lr_backbone=4.788419e-05 elapsed=27:34:15 avg_step=3.69s recent_step=6.61s eta=23:41:42 +2026-09-15 17:21:52 | INFO | train rank=0 | step=26900/50000 progress=53.8% loss=0.069753 lr=4.785115e-05 lr_backbone=4.785115e-05 elapsed=27:35:25 avg_step=3.69s recent_step=6.97s eta=23:41:34 +2026-09-15 17:22:55 | INFO | train rank=0 | step=26910/50000 progress=53.8% loss=0.080689 lr=4.781811e-05 lr_backbone=4.781811e-05 elapsed=27:36:27 avg_step=3.69s recent_step=6.25s eta=23:41:19 +2026-09-15 17:24:01 | INFO | train rank=0 | step=26920/50000 progress=53.8% loss=0.083783 lr=4.778508e-05 lr_backbone=4.778508e-05 elapsed=27:37:34 avg_step=3.69s recent_step=6.62s eta=23:41:07 +2026-09-15 17:25:06 | INFO | train rank=0 | step=26930/50000 progress=53.9% loss=0.108047 lr=4.775204e-05 lr_backbone=4.775204e-05 elapsed=27:38:38 avg_step=3.70s recent_step=6.48s eta=23:40:54 +2026-09-15 17:26:11 | INFO | train rank=0 | step=26940/50000 progress=53.9% loss=0.099146 lr=4.771900e-05 lr_backbone=4.771900e-05 elapsed=27:39:43 avg_step=3.70s recent_step=6.51s eta=23:40:41 +2026-09-15 17:27:05 | INFO | train rank=0 | step=26950/50000 progress=53.9% loss=0.099781 lr=4.768597e-05 lr_backbone=4.768597e-05 elapsed=27:40:38 avg_step=3.70s recent_step=5.42s eta=23:40:19 +2026-09-15 17:28:14 | INFO | train rank=0 | step=26960/50000 progress=53.9% loss=0.095759 lr=4.765294e-05 lr_backbone=4.765294e-05 elapsed=27:41:46 avg_step=3.70s recent_step=6.88s eta=23:40:09 +2026-09-15 17:29:18 | INFO | train rank=0 | step=26970/50000 progress=53.9% loss=0.084563 lr=4.761990e-05 lr_backbone=4.761990e-05 elapsed=27:42:50 avg_step=3.70s recent_step=6.41s eta=23:39:55 +2026-09-15 17:30:25 | INFO | train rank=0 | step=26980/50000 progress=54.0% loss=0.108527 lr=4.758687e-05 lr_backbone=4.758687e-05 elapsed=27:43:58 avg_step=3.70s recent_step=6.71s eta=23:39:44 +2026-09-15 17:31:31 | INFO | train rank=0 | step=26990/50000 progress=54.0% loss=0.085733 lr=4.755384e-05 lr_backbone=4.755384e-05 elapsed=27:45:04 avg_step=3.70s recent_step=6.63s eta=23:39:32 +2026-09-15 17:32:34 | INFO | train rank=0 | step=27000/50000 progress=54.0% loss=0.082067 lr=4.752081e-05 lr_backbone=4.752081e-05 elapsed=27:46:06 avg_step=3.70s recent_step=6.25s eta=23:39:16 +2026-09-15 17:33:38 | INFO | train rank=0 | step=27010/50000 progress=54.0% loss=0.121137 lr=4.748778e-05 lr_backbone=4.748778e-05 elapsed=27:47:11 avg_step=3.70s recent_step=6.47s eta=23:39:03 +2026-09-15 17:34:55 | INFO | train rank=0 | step=27020/50000 progress=54.0% loss=0.085870 lr=4.745476e-05 lr_backbone=4.745476e-05 elapsed=27:48:27 avg_step=3.70s recent_step=7.62s eta=23:38:59 +2026-09-15 17:36:10 | INFO | train rank=0 | step=27030/50000 progress=54.1% loss=0.110755 lr=4.742173e-05 lr_backbone=4.742173e-05 elapsed=27:49:43 avg_step=3.71s recent_step=7.54s eta=23:38:55 +2026-09-15 17:37:15 | INFO | train rank=0 | step=27040/50000 progress=54.1% loss=0.083795 lr=4.738871e-05 lr_backbone=4.738871e-05 elapsed=27:50:47 avg_step=3.71s recent_step=6.46s eta=23:38:41 +2026-09-15 17:38:17 | INFO | train rank=0 | step=27050/50000 progress=54.1% loss=0.115234 lr=4.735568e-05 lr_backbone=4.735568e-05 elapsed=27:51:49 avg_step=3.71s recent_step=6.20s eta=23:38:25 +2026-09-15 17:39:25 | INFO | train rank=0 | step=27060/50000 progress=54.1% loss=0.076757 lr=4.732266e-05 lr_backbone=4.732266e-05 elapsed=27:52:58 avg_step=3.71s recent_step=6.85s eta=23:38:15 +2026-09-15 17:40:26 | INFO | train rank=0 | step=27070/50000 progress=54.1% loss=0.105090 lr=4.728964e-05 lr_backbone=4.728964e-05 elapsed=27:53:59 avg_step=3.71s recent_step=6.09s eta=23:37:58 +2026-09-15 17:41:32 | INFO | train rank=0 | step=27080/50000 progress=54.2% loss=0.128347 lr=4.725662e-05 lr_backbone=4.725662e-05 elapsed=27:55:05 avg_step=3.71s recent_step=6.60s eta=23:37:45 +2026-09-15 17:42:40 | INFO | train rank=0 | step=27090/50000 progress=54.2% loss=0.094881 lr=4.722360e-05 lr_backbone=4.722360e-05 elapsed=27:56:12 avg_step=3.71s recent_step=6.76s eta=23:37:34 +2026-09-15 17:43:44 | INFO | train rank=0 | step=27100/50000 progress=54.2% loss=0.104327 lr=4.719058e-05 lr_backbone=4.719058e-05 elapsed=27:57:17 avg_step=3.71s recent_step=6.47s eta=23:37:20 +2026-09-15 17:44:46 | INFO | train rank=0 | step=27110/50000 progress=54.2% loss=0.099096 lr=4.715757e-05 lr_backbone=4.715757e-05 elapsed=27:58:19 avg_step=3.71s recent_step=6.20s eta=23:37:04 +2026-09-15 17:45:46 | INFO | train rank=0 | step=27120/50000 progress=54.2% loss=0.142449 lr=4.712455e-05 lr_backbone=4.712455e-05 elapsed=27:59:19 avg_step=3.72s recent_step=5.97s eta=23:36:46 +2026-09-15 17:46:55 | INFO | train rank=0 | step=27130/50000 progress=54.3% loss=0.072124 lr=4.709154e-05 lr_backbone=4.709154e-05 elapsed=28:00:27 avg_step=3.72s recent_step=6.87s eta=23:36:35 +2026-09-15 17:48:49 | INFO | train rank=0 | step=27140/50000 progress=54.3% loss=0.096474 lr=4.705852e-05 lr_backbone=4.705852e-05 elapsed=28:02:22 avg_step=3.72s recent_step=11.47s eta=23:37:03 +2026-09-15 17:50:48 | INFO | train rank=0 | step=27150/50000 progress=54.3% loss=0.077440 lr=4.702551e-05 lr_backbone=4.702551e-05 elapsed=28:04:20 avg_step=3.72s recent_step=11.83s eta=23:37:35 +2026-09-15 17:52:53 | INFO | train rank=0 | step=27160/50000 progress=54.3% loss=0.108988 lr=4.699250e-05 lr_backbone=4.699250e-05 elapsed=28:06:26 avg_step=3.73s recent_step=12.55s eta=23:38:11 +2026-09-15 17:54:37 | INFO | train rank=0 | step=27170/50000 progress=54.3% loss=0.102509 lr=4.695949e-05 lr_backbone=4.695949e-05 elapsed=28:08:10 avg_step=3.73s recent_step=10.37s eta=23:38:30 +2026-09-15 17:56:31 | INFO | train rank=0 | step=27180/50000 progress=54.4% loss=0.082267 lr=4.692649e-05 lr_backbone=4.692649e-05 elapsed=28:10:04 avg_step=3.73s recent_step=11.42s eta=23:38:57 +2026-09-15 17:58:05 | INFO | train rank=0 | step=27190/50000 progress=54.4% loss=0.102119 lr=4.689348e-05 lr_backbone=4.689348e-05 elapsed=28:11:38 avg_step=3.73s recent_step=9.43s eta=23:39:08 +2026-09-15 17:59:37 | INFO | train rank=0 | step=27200/50000 progress=54.4% loss=0.087674 lr=4.686047e-05 lr_backbone=4.686047e-05 elapsed=28:13:10 avg_step=3.73s recent_step=9.21s eta=23:39:16 +2026-09-15 18:01:10 | INFO | train rank=0 | step=27210/50000 progress=54.4% loss=0.102406 lr=4.682747e-05 lr_backbone=4.682747e-05 elapsed=28:14:42 avg_step=3.74s recent_step=9.22s eta=23:39:25 +2026-09-15 18:03:21 | INFO | train rank=0 | step=27220/50000 progress=54.4% loss=0.119153 lr=4.679447e-05 lr_backbone=4.679447e-05 elapsed=28:16:54 avg_step=3.74s recent_step=13.15s eta=23:40:06 +2026-09-15 18:03:55 | INFO | train rank=0 | step=27230/50000 progress=54.5% loss=0.096373 lr=4.676147e-05 lr_backbone=4.676147e-05 elapsed=28:17:28 avg_step=3.74s recent_step=3.38s eta=23:39:26 +2026-09-15 18:04:25 | INFO | train rank=0 | step=27240/50000 progress=54.5% loss=0.112469 lr=4.672847e-05 lr_backbone=4.672847e-05 elapsed=28:17:58 avg_step=3.74s recent_step=3.06s eta=23:38:43 +2026-09-15 18:04:54 | INFO | train rank=0 | step=27250/50000 progress=54.5% loss=0.103479 lr=4.669547e-05 lr_backbone=4.669547e-05 elapsed=28:18:27 avg_step=3.74s recent_step=2.90s eta=23:37:58 +2026-09-15 18:05:20 | INFO | train rank=0 | step=27260/50000 progress=54.5% loss=0.114253 lr=4.666247e-05 lr_backbone=4.666247e-05 elapsed=28:18:53 avg_step=3.74s recent_step=2.57s eta=23:37:11 +2026-09-15 18:05:56 | INFO | train rank=0 | step=27270/50000 progress=54.5% loss=0.086627 lr=4.662948e-05 lr_backbone=4.662948e-05 elapsed=28:19:29 avg_step=3.74s recent_step=3.58s eta=23:36:33 +2026-09-15 18:06:20 | INFO | train rank=0 | step=27280/50000 progress=54.6% loss=0.096813 lr=4.659649e-05 lr_backbone=4.659649e-05 elapsed=28:19:52 avg_step=3.74s recent_step=2.37s eta=23:35:44 +2026-09-15 18:06:54 | INFO | train rank=0 | step=27290/50000 progress=54.6% loss=0.094665 lr=4.656349e-05 lr_backbone=4.656349e-05 elapsed=28:20:27 avg_step=3.74s recent_step=3.44s eta=23:35:04 +2026-09-15 18:07:21 | INFO | train rank=0 | step=27300/50000 progress=54.6% loss=0.102553 lr=4.653050e-05 lr_backbone=4.653050e-05 elapsed=28:20:54 avg_step=3.74s recent_step=2.73s eta=23:34:18 +2026-09-15 18:07:48 | INFO | train rank=0 | step=27310/50000 progress=54.6% loss=0.098011 lr=4.649751e-05 lr_backbone=4.649751e-05 elapsed=28:21:21 avg_step=3.74s recent_step=2.71s eta=23:33:32 +2026-09-15 18:08:16 | INFO | train rank=0 | step=27320/50000 progress=54.6% loss=0.084461 lr=4.646453e-05 lr_backbone=4.646453e-05 elapsed=28:21:49 avg_step=3.74s recent_step=2.81s eta=23:32:47 +2026-09-15 18:08:43 | INFO | train rank=0 | step=27330/50000 progress=54.7% loss=0.118628 lr=4.643154e-05 lr_backbone=4.643154e-05 elapsed=28:22:16 avg_step=3.74s recent_step=2.67s eta=23:32:01 +2026-09-15 18:09:14 | INFO | train rank=0 | step=27340/50000 progress=54.7% loss=0.095598 lr=4.639856e-05 lr_backbone=4.639856e-05 elapsed=28:22:46 avg_step=3.74s recent_step=3.04s eta=23:31:18 +2026-09-15 18:09:52 | INFO | train rank=0 | step=27350/50000 progress=54.7% loss=0.121047 lr=4.636557e-05 lr_backbone=4.636557e-05 elapsed=28:23:25 avg_step=3.74s recent_step=3.85s eta=23:30:41 +2026-09-15 18:10:18 | INFO | train rank=0 | step=27360/50000 progress=54.7% loss=0.103397 lr=4.633259e-05 lr_backbone=4.633259e-05 elapsed=28:23:51 avg_step=3.74s recent_step=2.62s eta=23:29:55 +2026-09-15 18:10:40 | INFO | train rank=0 | step=27370/50000 progress=54.7% loss=0.094028 lr=4.629961e-05 lr_backbone=4.629961e-05 elapsed=28:24:13 avg_step=3.74s recent_step=2.23s eta=23:29:05 +2026-09-15 18:11:05 | INFO | train rank=0 | step=27380/50000 progress=54.8% loss=0.124881 lr=4.626664e-05 lr_backbone=4.626664e-05 elapsed=28:24:38 avg_step=3.74s recent_step=2.48s eta=23:28:17 +2026-09-15 18:11:42 | INFO | train rank=0 | step=27390/50000 progress=54.8% loss=0.090631 lr=4.623366e-05 lr_backbone=4.623366e-05 elapsed=28:25:15 avg_step=3.74s recent_step=3.66s eta=23:27:39 +2026-09-15 18:12:09 | INFO | train rank=0 | step=27400/50000 progress=54.8% loss=0.104187 lr=4.620069e-05 lr_backbone=4.620069e-05 elapsed=28:25:41 avg_step=3.74s recent_step=2.67s eta=23:26:53 +2026-09-15 18:12:33 | INFO | train rank=0 | step=27410/50000 progress=54.8% loss=0.088768 lr=4.616771e-05 lr_backbone=4.616771e-05 elapsed=28:26:06 avg_step=3.73s recent_step=2.47s eta=23:26:05 +2026-09-15 18:13:01 | INFO | train rank=0 | step=27420/50000 progress=54.8% loss=0.088292 lr=4.613474e-05 lr_backbone=4.613474e-05 elapsed=28:26:34 avg_step=3.73s recent_step=2.77s eta=23:25:20 +2026-09-15 18:13:34 | INFO | train rank=0 | step=27430/50000 progress=54.9% loss=0.114012 lr=4.610177e-05 lr_backbone=4.610177e-05 elapsed=28:27:06 avg_step=3.73s recent_step=3.26s eta=23:24:39 +2026-09-15 18:13:59 | INFO | train rank=0 | step=27440/50000 progress=54.9% loss=0.108394 lr=4.606880e-05 lr_backbone=4.606880e-05 elapsed=28:27:32 avg_step=3.73s recent_step=2.56s eta=23:23:52 +2026-09-15 18:14:22 | INFO | train rank=0 | step=27450/50000 progress=54.9% loss=0.101368 lr=4.603584e-05 lr_backbone=4.603584e-05 elapsed=28:27:55 avg_step=3.73s recent_step=2.27s eta=23:23:02 +2026-09-15 18:14:54 | INFO | train rank=0 | step=27460/50000 progress=54.9% loss=0.085072 lr=4.600287e-05 lr_backbone=4.600287e-05 elapsed=28:28:26 avg_step=3.73s recent_step=3.18s eta=23:22:20 +2026-09-15 18:15:32 | INFO | train rank=0 | step=27470/50000 progress=54.9% loss=0.091472 lr=4.596991e-05 lr_backbone=4.596991e-05 elapsed=28:29:04 avg_step=3.73s recent_step=3.80s eta=23:21:44 +2026-09-15 18:16:00 | INFO | train rank=0 | step=27480/50000 progress=55.0% loss=0.073150 lr=4.593695e-05 lr_backbone=4.593695e-05 elapsed=28:29:33 avg_step=3.73s recent_step=2.81s eta=23:20:59 +2026-09-15 18:16:24 | INFO | train rank=0 | step=27490/50000 progress=55.0% loss=0.103997 lr=4.590399e-05 lr_backbone=4.590399e-05 elapsed=28:29:57 avg_step=3.73s recent_step=2.41s eta=23:20:10 +2026-09-15 18:16:50 | INFO | train rank=0 | step=27500/50000 progress=55.0% loss=0.088597 lr=4.587103e-05 lr_backbone=4.587103e-05 elapsed=28:30:23 avg_step=3.73s recent_step=2.61s eta=23:19:24 +2026-09-15 18:17:24 | INFO | train rank=0 | step=27510/50000 progress=55.0% loss=0.094606 lr=4.583808e-05 lr_backbone=4.583808e-05 elapsed=28:30:57 avg_step=3.73s recent_step=3.41s eta=23:18:44 +2026-09-15 18:17:49 | INFO | train rank=0 | step=27520/50000 progress=55.0% loss=0.076497 lr=4.580512e-05 lr_backbone=4.580512e-05 elapsed=28:31:21 avg_step=3.73s recent_step=2.46s eta=23:17:56 +2026-09-15 18:18:17 | INFO | train rank=0 | step=27530/50000 progress=55.1% loss=0.101014 lr=4.577217e-05 lr_backbone=4.577217e-05 elapsed=28:31:50 avg_step=3.73s recent_step=2.86s eta=23:17:12 +2026-09-15 18:18:43 | INFO | train rank=0 | step=27540/50000 progress=55.1% loss=0.118393 lr=4.573922e-05 lr_backbone=4.573922e-05 elapsed=28:32:16 avg_step=3.73s recent_step=2.60s eta=23:16:25 +2026-09-15 18:19:18 | INFO | train rank=0 | step=27550/50000 progress=55.1% loss=0.098453 lr=4.570627e-05 lr_backbone=4.570627e-05 elapsed=28:32:51 avg_step=3.73s recent_step=3.48s eta=23:15:46 +2026-09-15 18:19:46 | INFO | train rank=0 | step=27560/50000 progress=55.1% loss=0.097154 lr=4.567333e-05 lr_backbone=4.567333e-05 elapsed=28:33:19 avg_step=3.73s recent_step=2.83s eta=23:15:01 +2026-09-15 18:20:11 | INFO | train rank=0 | step=27570/50000 progress=55.1% loss=0.083647 lr=4.564038e-05 lr_backbone=4.564038e-05 elapsed=28:33:43 avg_step=3.73s recent_step=2.42s eta=23:14:13 +2026-09-15 18:20:37 | INFO | train rank=0 | step=27580/50000 progress=55.2% loss=0.099561 lr=4.560744e-05 lr_backbone=4.560744e-05 elapsed=28:34:10 avg_step=3.73s recent_step=2.65s eta=23:13:27 +2026-09-15 18:21:13 | INFO | train rank=0 | step=27590/50000 progress=55.2% loss=0.091963 lr=4.557450e-05 lr_backbone=4.557450e-05 elapsed=28:34:46 avg_step=3.73s recent_step=3.64s eta=23:12:49 +2026-09-15 18:21:42 | INFO | train rank=0 | step=27600/50000 progress=55.2% loss=0.080701 lr=4.554156e-05 lr_backbone=4.554156e-05 elapsed=28:35:15 avg_step=3.73s recent_step=2.87s eta=23:12:05 +2026-09-15 18:22:17 | INFO | train rank=0 | step=27610/50000 progress=55.2% loss=0.088587 lr=4.550862e-05 lr_backbone=4.550862e-05 elapsed=28:35:50 avg_step=3.73s recent_step=3.51s eta=23:11:26 +2026-09-15 18:22:42 | INFO | train rank=0 | step=27620/50000 progress=55.2% loss=0.093735 lr=4.547569e-05 lr_backbone=4.547569e-05 elapsed=28:36:15 avg_step=3.73s recent_step=2.51s eta=23:10:39 +2026-09-15 18:23:16 | INFO | train rank=0 | step=27630/50000 progress=55.3% loss=0.097983 lr=4.544276e-05 lr_backbone=4.544276e-05 elapsed=28:36:49 avg_step=3.73s recent_step=3.42s eta=23:09:59 +2026-09-15 18:23:43 | INFO | train rank=0 | step=27640/50000 progress=55.3% loss=0.114802 lr=4.540983e-05 lr_backbone=4.540983e-05 elapsed=28:37:16 avg_step=3.73s recent_step=2.67s eta=23:09:13 +2026-09-15 18:24:08 | INFO | train rank=0 | step=27650/50000 progress=55.3% loss=0.094223 lr=4.537690e-05 lr_backbone=4.537690e-05 elapsed=28:37:41 avg_step=3.73s recent_step=2.53s eta=23:08:26 +2026-09-15 18:24:37 | INFO | train rank=0 | step=27660/50000 progress=55.3% loss=0.121575 lr=4.534397e-05 lr_backbone=4.534397e-05 elapsed=28:38:10 avg_step=3.73s recent_step=2.89s eta=23:07:42 +2026-09-15 18:25:09 | INFO | train rank=0 | step=27670/50000 progress=55.3% loss=0.110071 lr=4.531105e-05 lr_backbone=4.531105e-05 elapsed=28:38:41 avg_step=3.73s recent_step=3.13s eta=23:07:00 +2026-09-15 18:25:39 | INFO | train rank=0 | step=27680/50000 progress=55.4% loss=0.111817 lr=4.527812e-05 lr_backbone=4.527812e-05 elapsed=28:39:12 avg_step=3.73s recent_step=3.02s eta=23:06:17 +2026-09-15 18:26:06 | INFO | train rank=0 | step=27690/50000 progress=55.4% loss=0.075997 lr=4.524520e-05 lr_backbone=4.524520e-05 elapsed=28:39:39 avg_step=3.73s recent_step=2.70s eta=23:05:32 +2026-09-15 18:26:27 | INFO | train rank=0 | step=27700/50000 progress=55.4% loss=0.093375 lr=4.521228e-05 lr_backbone=4.521228e-05 elapsed=28:40:00 avg_step=3.73s recent_step=2.09s eta=23:04:41 +2026-09-15 18:27:06 | INFO | train rank=0 | step=27710/50000 progress=55.4% loss=0.093759 lr=4.517937e-05 lr_backbone=4.517937e-05 elapsed=28:40:38 avg_step=3.73s recent_step=3.89s eta=23:04:05 +2026-09-15 18:27:31 | INFO | train rank=0 | step=27720/50000 progress=55.4% loss=0.107555 lr=4.514645e-05 lr_backbone=4.514645e-05 elapsed=28:41:04 avg_step=3.73s recent_step=2.58s eta=23:03:19 +2026-09-15 18:27:57 | INFO | train rank=0 | step=27730/50000 progress=55.5% loss=0.106223 lr=4.511354e-05 lr_backbone=4.511354e-05 elapsed=28:41:30 avg_step=3.72s recent_step=2.58s eta=23:02:32 +2026-09-15 18:28:23 | INFO | train rank=0 | step=27740/50000 progress=55.5% loss=0.122467 lr=4.508063e-05 lr_backbone=4.508063e-05 elapsed=28:41:56 avg_step=3.72s recent_step=2.57s eta=23:01:46 +2026-09-15 18:29:00 | INFO | train rank=0 | step=27750/50000 progress=55.5% loss=0.100037 lr=4.504772e-05 lr_backbone=4.504772e-05 elapsed=28:42:33 avg_step=3.72s recent_step=3.70s eta=23:01:08 +2026-09-15 18:29:28 | INFO | train rank=0 | step=27760/50000 progress=55.5% loss=0.074280 lr=4.501482e-05 lr_backbone=4.501482e-05 elapsed=28:43:01 avg_step=3.72s recent_step=2.80s eta=23:00:24 +2026-09-15 18:30:00 | INFO | train rank=0 | step=27770/50000 progress=55.5% loss=0.106378 lr=4.498191e-05 lr_backbone=4.498191e-05 elapsed=28:43:33 avg_step=3.72s recent_step=3.20s eta=22:59:42 +2026-09-15 18:30:28 | INFO | train rank=0 | step=27780/50000 progress=55.6% loss=0.075261 lr=4.494901e-05 lr_backbone=4.494901e-05 elapsed=28:44:00 avg_step=3.72s recent_step=2.75s eta=22:58:57 +2026-09-15 18:31:06 | INFO | train rank=0 | step=27790/50000 progress=55.6% loss=0.137659 lr=4.491611e-05 lr_backbone=4.491611e-05 elapsed=28:44:39 avg_step=3.72s recent_step=3.85s eta=22:58:21 +2026-09-15 18:31:33 | INFO | train rank=0 | step=27800/50000 progress=55.6% loss=0.106144 lr=4.488322e-05 lr_backbone=4.488322e-05 elapsed=28:45:06 avg_step=3.72s recent_step=2.67s eta=22:57:35 +2026-09-15 18:31:54 | INFO | train rank=0 | step=27810/50000 progress=55.6% loss=0.070410 lr=4.485032e-05 lr_backbone=4.485032e-05 elapsed=28:45:27 avg_step=3.72s recent_step=2.10s eta=22:56:45 +2026-09-15 18:32:19 | INFO | train rank=0 | step=27820/50000 progress=55.6% loss=0.080757 lr=4.481743e-05 lr_backbone=4.481743e-05 elapsed=28:45:52 avg_step=3.72s recent_step=2.52s eta=22:55:58 +2026-09-15 18:32:53 | INFO | train rank=0 | step=27830/50000 progress=55.7% loss=0.078303 lr=4.478454e-05 lr_backbone=4.478454e-05 elapsed=28:46:25 avg_step=3.72s recent_step=3.37s eta=22:55:18 +2026-09-15 18:33:20 | INFO | train rank=0 | step=27840/50000 progress=55.7% loss=0.090971 lr=4.475165e-05 lr_backbone=4.475165e-05 elapsed=28:46:53 avg_step=3.72s recent_step=2.72s eta=22:54:33 +2026-09-15 18:33:39 | INFO | train rank=0 | step=27850/50000 progress=55.7% loss=0.084532 lr=4.471877e-05 lr_backbone=4.471877e-05 elapsed=28:47:12 avg_step=3.72s recent_step=1.88s eta=22:53:41 +2026-09-15 18:34:05 | INFO | train rank=0 | step=27860/50000 progress=55.7% loss=0.106402 lr=4.468588e-05 lr_backbone=4.468588e-05 elapsed=28:47:37 avg_step=3.72s recent_step=2.59s eta=22:52:55 +2026-09-15 18:34:38 | INFO | train rank=0 | step=27870/50000 progress=55.7% loss=0.140795 lr=4.465300e-05 lr_backbone=4.465300e-05 elapsed=28:48:11 avg_step=3.72s recent_step=3.36s eta=22:52:15 +2026-09-15 18:34:57 | INFO | train rank=0 | step=27880/50000 progress=55.8% loss=0.115321 lr=4.462012e-05 lr_backbone=4.462012e-05 elapsed=28:48:30 avg_step=3.72s recent_step=1.90s eta=22:51:23 +2026-09-15 18:35:31 | INFO | train rank=0 | step=27890/50000 progress=55.8% loss=0.083761 lr=4.458725e-05 lr_backbone=4.458725e-05 elapsed=28:49:04 avg_step=3.72s recent_step=3.42s eta=22:50:44 +2026-09-15 18:35:59 | INFO | train rank=0 | step=27900/50000 progress=55.8% loss=0.100778 lr=4.455437e-05 lr_backbone=4.455437e-05 elapsed=28:49:32 avg_step=3.72s recent_step=2.74s eta=22:49:59 +2026-09-15 18:36:29 | INFO | train rank=0 | step=27910/50000 progress=55.8% loss=0.073884 lr=4.452150e-05 lr_backbone=4.452150e-05 elapsed=28:50:02 avg_step=3.72s recent_step=3.01s eta=22:49:16 +2026-09-15 18:36:52 | INFO | train rank=0 | step=27920/50000 progress=55.8% loss=0.091001 lr=4.448863e-05 lr_backbone=4.448863e-05 elapsed=28:50:25 avg_step=3.72s recent_step=2.30s eta=22:48:28 +2026-09-15 18:37:21 | INFO | train rank=0 | step=27930/50000 progress=55.9% loss=0.101323 lr=4.445577e-05 lr_backbone=4.445577e-05 elapsed=28:50:54 avg_step=3.72s recent_step=2.97s eta=22:47:44 +2026-09-15 18:37:54 | INFO | train rank=0 | step=27940/50000 progress=55.9% loss=0.089669 lr=4.442290e-05 lr_backbone=4.442290e-05 elapsed=28:51:26 avg_step=3.72s recent_step=3.21s eta=22:47:03 +2026-09-15 18:38:20 | INFO | train rank=0 | step=27950/50000 progress=55.9% loss=0.121751 lr=4.439004e-05 lr_backbone=4.439004e-05 elapsed=28:51:52 avg_step=3.72s recent_step=2.60s eta=22:46:17 +2026-09-15 18:38:52 | INFO | train rank=0 | step=27960/50000 progress=55.9% loss=0.099633 lr=4.435718e-05 lr_backbone=4.435718e-05 elapsed=28:52:25 avg_step=3.72s recent_step=3.22s eta=22:45:36 +2026-09-15 18:39:16 | INFO | train rank=0 | step=27970/50000 progress=55.9% loss=0.102423 lr=4.432432e-05 lr_backbone=4.432432e-05 elapsed=28:52:49 avg_step=3.72s recent_step=2.44s eta=22:44:49 +2026-09-15 18:39:42 | INFO | train rank=0 | step=27980/50000 progress=56.0% loss=0.116167 lr=4.429147e-05 lr_backbone=4.429147e-05 elapsed=28:53:14 avg_step=3.72s recent_step=2.55s eta=22:44:03 +2026-09-15 18:40:02 | INFO | train rank=0 | step=27990/50000 progress=56.0% loss=0.142741 lr=4.425862e-05 lr_backbone=4.425862e-05 elapsed=28:53:35 avg_step=3.72s recent_step=2.06s eta=22:43:12 +2026-09-15 18:40:30 | INFO | train rank=0 | step=28000/50000 progress=56.0% loss=0.094667 lr=4.422577e-05 lr_backbone=4.422577e-05 elapsed=28:54:03 avg_step=3.72s recent_step=2.76s eta=22:42:28 +2026-09-15 18:41:03 | INFO | train rank=0 | step=28010/50000 progress=56.0% loss=0.075205 lr=4.419292e-05 lr_backbone=4.419292e-05 elapsed=28:54:36 avg_step=3.72s recent_step=3.34s eta=22:41:48 +2026-09-15 18:41:26 | INFO | train rank=0 | step=28020/50000 progress=56.0% loss=0.118490 lr=4.416008e-05 lr_backbone=4.416008e-05 elapsed=28:54:59 avg_step=3.72s recent_step=2.26s eta=22:40:59 +2026-09-15 18:42:01 | INFO | train rank=0 | step=28030/50000 progress=56.1% loss=0.086338 lr=4.412724e-05 lr_backbone=4.412724e-05 elapsed=28:55:34 avg_step=3.72s recent_step=3.50s eta=22:40:20 +2026-09-15 18:42:22 | INFO | train rank=0 | step=28040/50000 progress=56.1% loss=0.097427 lr=4.409440e-05 lr_backbone=4.409440e-05 elapsed=28:55:54 avg_step=3.71s recent_step=2.09s eta=22:39:30 +2026-09-15 18:42:49 | INFO | train rank=0 | step=28050/50000 progress=56.1% loss=0.077823 lr=4.406156e-05 lr_backbone=4.406156e-05 elapsed=28:56:22 avg_step=3.71s recent_step=2.74s eta=22:38:45 +2026-09-15 18:43:17 | INFO | train rank=0 | step=28060/50000 progress=56.1% loss=0.109425 lr=4.402873e-05 lr_backbone=4.402873e-05 elapsed=28:56:50 avg_step=3.71s recent_step=2.84s eta=22:38:01 +2026-09-15 18:43:48 | INFO | train rank=0 | step=28070/50000 progress=56.1% loss=0.091014 lr=4.399589e-05 lr_backbone=4.399589e-05 elapsed=28:57:20 avg_step=3.71s recent_step=3.02s eta=22:37:19 +2026-09-15 18:44:12 | INFO | train rank=0 | step=28080/50000 progress=56.2% loss=0.120955 lr=4.396307e-05 lr_backbone=4.396307e-05 elapsed=28:57:45 avg_step=3.71s recent_step=2.44s eta=22:36:32 +2026-09-15 18:44:41 | INFO | train rank=0 | step=28090/50000 progress=56.2% loss=0.100804 lr=4.393024e-05 lr_backbone=4.393024e-05 elapsed=28:58:13 avg_step=3.71s recent_step=2.86s eta=22:35:48 +2026-09-15 18:45:04 | INFO | train rank=0 | step=28100/50000 progress=56.2% loss=0.100385 lr=4.389742e-05 lr_backbone=4.389742e-05 elapsed=28:58:37 avg_step=3.71s recent_step=2.33s eta=22:35:00 +2026-09-15 18:45:33 | INFO | train rank=0 | step=28110/50000 progress=56.2% loss=0.120772 lr=4.386459e-05 lr_backbone=4.386459e-05 elapsed=28:59:06 avg_step=3.71s recent_step=2.92s eta=22:34:17 +2026-09-15 18:45:59 | INFO | train rank=0 | step=28120/50000 progress=56.2% loss=0.088067 lr=4.383178e-05 lr_backbone=4.383178e-05 elapsed=28:59:32 avg_step=3.71s recent_step=2.57s eta=22:33:31 +2026-09-15 18:46:29 | INFO | train rank=0 | step=28130/50000 progress=56.3% loss=0.087077 lr=4.379896e-05 lr_backbone=4.379896e-05 elapsed=29:00:02 avg_step=3.71s recent_step=3.06s eta=22:32:49 +2026-09-15 18:47:00 | INFO | train rank=0 | step=28140/50000 progress=56.3% loss=0.101921 lr=4.376615e-05 lr_backbone=4.376615e-05 elapsed=29:00:33 avg_step=3.71s recent_step=3.04s eta=22:32:06 +2026-09-15 18:47:26 | INFO | train rank=0 | step=28150/50000 progress=56.3% loss=0.117758 lr=4.373334e-05 lr_backbone=4.373334e-05 elapsed=29:00:59 avg_step=3.71s recent_step=2.60s eta=22:31:21 +2026-09-15 18:47:51 | INFO | train rank=0 | step=28160/50000 progress=56.3% loss=0.094037 lr=4.370053e-05 lr_backbone=4.370053e-05 elapsed=29:01:24 avg_step=3.71s recent_step=2.50s eta=22:30:34 +2026-09-15 18:48:18 | INFO | train rank=0 | step=28170/50000 progress=56.3% loss=0.101622 lr=4.366773e-05 lr_backbone=4.366773e-05 elapsed=29:01:51 avg_step=3.71s recent_step=2.73s eta=22:29:49 +2026-09-15 18:48:43 | INFO | train rank=0 | step=28180/50000 progress=56.4% loss=0.087721 lr=4.363492e-05 lr_backbone=4.363492e-05 elapsed=29:02:16 avg_step=3.71s recent_step=2.52s eta=22:29:03 +2026-09-15 18:49:17 | INFO | train rank=0 | step=28190/50000 progress=56.4% loss=0.076469 lr=4.360213e-05 lr_backbone=4.360213e-05 elapsed=29:02:50 avg_step=3.71s recent_step=3.37s eta=22:28:23 +2026-09-15 18:49:45 | INFO | train rank=0 | step=28200/50000 progress=56.4% loss=0.101100 lr=4.356933e-05 lr_backbone=4.356933e-05 elapsed=29:03:18 avg_step=3.71s recent_step=2.85s eta=22:27:40 +2026-09-15 18:50:13 | INFO | train rank=0 | step=28210/50000 progress=56.4% loss=0.106257 lr=4.353654e-05 lr_backbone=4.353654e-05 elapsed=29:03:46 avg_step=3.71s recent_step=2.80s eta=22:26:55 +2026-09-15 18:50:33 | INFO | train rank=0 | step=28220/50000 progress=56.4% loss=0.083662 lr=4.350375e-05 lr_backbone=4.350375e-05 elapsed=29:04:06 avg_step=3.71s recent_step=1.95s eta=22:26:05 +2026-09-15 18:51:09 | INFO | train rank=0 | step=28230/50000 progress=56.5% loss=0.106362 lr=4.347096e-05 lr_backbone=4.347096e-05 elapsed=29:04:41 avg_step=3.71s recent_step=3.58s eta=22:25:27 +2026-09-15 18:51:32 | INFO | train rank=0 | step=28240/50000 progress=56.5% loss=0.150017 lr=4.343817e-05 lr_backbone=4.343817e-05 elapsed=29:05:05 avg_step=3.71s recent_step=2.35s eta=22:24:39 +2026-09-15 18:52:00 | INFO | train rank=0 | step=28250/50000 progress=56.5% loss=0.069892 lr=4.340539e-05 lr_backbone=4.340539e-05 elapsed=29:05:33 avg_step=3.71s recent_step=2.82s eta=22:23:55 +2026-09-15 18:52:26 | INFO | train rank=0 | step=28260/50000 progress=56.5% loss=0.085657 lr=4.337261e-05 lr_backbone=4.337261e-05 elapsed=29:05:59 avg_step=3.71s recent_step=2.58s eta=22:23:09 +2026-09-15 18:52:50 | INFO | train rank=0 | step=28270/50000 progress=56.5% loss=0.097175 lr=4.333984e-05 lr_backbone=4.333984e-05 elapsed=29:06:23 avg_step=3.71s recent_step=2.41s eta=22:22:22 +2026-09-15 18:53:16 | INFO | train rank=0 | step=28280/50000 progress=56.6% loss=0.096308 lr=4.330706e-05 lr_backbone=4.330706e-05 elapsed=29:06:49 avg_step=3.71s recent_step=2.57s eta=22:21:37 +2026-09-15 18:53:47 | INFO | train rank=0 | step=28290/50000 progress=56.6% loss=0.098393 lr=4.327429e-05 lr_backbone=4.327429e-05 elapsed=29:07:20 avg_step=3.71s recent_step=3.11s eta=22:20:55 +2026-09-15 18:54:14 | INFO | train rank=0 | step=28300/50000 progress=56.6% loss=0.091420 lr=4.324153e-05 lr_backbone=4.324153e-05 elapsed=29:07:47 avg_step=3.71s recent_step=2.70s eta=22:20:10 +2026-09-15 18:54:36 | INFO | train rank=0 | step=28310/50000 progress=56.6% loss=0.068812 lr=4.320876e-05 lr_backbone=4.320876e-05 elapsed=29:08:09 avg_step=3.71s recent_step=2.22s eta=22:19:22 +2026-09-15 18:55:01 | INFO | train rank=0 | step=28320/50000 progress=56.6% loss=0.088779 lr=4.317600e-05 lr_backbone=4.317600e-05 elapsed=29:08:34 avg_step=3.70s recent_step=2.47s eta=22:18:35 +2026-09-15 18:55:29 | INFO | train rank=0 | step=28330/50000 progress=56.7% loss=0.083446 lr=4.314324e-05 lr_backbone=4.314324e-05 elapsed=29:09:02 avg_step=3.70s recent_step=2.77s eta=22:17:51 +2026-09-15 18:55:56 | INFO | train rank=0 | step=28340/50000 progress=56.7% loss=0.084861 lr=4.311049e-05 lr_backbone=4.311049e-05 elapsed=29:09:29 avg_step=3.70s recent_step=2.73s eta=22:17:07 +2026-09-15 18:56:23 | INFO | train rank=0 | step=28350/50000 progress=56.7% loss=0.089679 lr=4.307773e-05 lr_backbone=4.307773e-05 elapsed=29:09:55 avg_step=3.70s recent_step=2.66s eta=22:16:22 +2026-09-15 18:56:50 | INFO | train rank=0 | step=28360/50000 progress=56.7% loss=0.094451 lr=4.304498e-05 lr_backbone=4.304498e-05 elapsed=29:10:23 avg_step=3.70s recent_step=2.77s eta=22:15:37 +2026-09-15 18:57:47 | INFO | train rank=0 | step=28370/50000 progress=56.7% loss=0.091399 lr=4.301224e-05 lr_backbone=4.301224e-05 elapsed=29:11:20 avg_step=3.70s recent_step=5.71s eta=22:15:16 +2026-09-15 18:58:15 | INFO | train rank=0 | step=28380/50000 progress=56.8% loss=0.103406 lr=4.297949e-05 lr_backbone=4.297949e-05 elapsed=29:11:47 avg_step=3.70s recent_step=2.72s eta=22:14:31 +2026-09-15 18:58:44 | INFO | train rank=0 | step=28390/50000 progress=56.8% loss=0.143409 lr=4.294675e-05 lr_backbone=4.294675e-05 elapsed=29:12:16 avg_step=3.70s recent_step=2.89s eta=22:13:48 +2026-09-15 18:59:14 | INFO | train rank=0 | step=28400/50000 progress=56.8% loss=0.106856 lr=4.291402e-05 lr_backbone=4.291402e-05 elapsed=29:12:47 avg_step=3.70s recent_step=3.02s eta=22:13:06 +2026-09-15 18:59:56 | INFO | train rank=0 | step=28410/50000 progress=56.8% loss=0.105557 lr=4.288128e-05 lr_backbone=4.288128e-05 elapsed=29:13:29 avg_step=3.70s recent_step=4.20s eta=22:12:32 +2026-09-15 19:00:27 | INFO | train rank=0 | step=28420/50000 progress=56.8% loss=0.105698 lr=4.284855e-05 lr_backbone=4.284855e-05 elapsed=29:14:00 avg_step=3.70s recent_step=3.14s eta=22:11:51 +2026-09-15 19:01:21 | INFO | train rank=0 | step=28430/50000 progress=56.9% loss=0.088329 lr=4.281582e-05 lr_backbone=4.281582e-05 elapsed=29:14:54 avg_step=3.70s recent_step=5.42s eta=22:11:27 +2026-09-15 19:01:53 | INFO | train rank=0 | step=28440/50000 progress=56.9% loss=0.102507 lr=4.278310e-05 lr_backbone=4.278310e-05 elapsed=29:15:26 avg_step=3.70s recent_step=3.14s eta=22:10:46 +2026-09-15 19:02:34 | INFO | train rank=0 | step=28450/50000 progress=56.9% loss=0.077295 lr=4.275038e-05 lr_backbone=4.275038e-05 elapsed=29:16:06 avg_step=3.70s recent_step=4.08s eta=22:10:12 +2026-09-15 19:03:10 | INFO | train rank=0 | step=28460/50000 progress=56.9% loss=0.100549 lr=4.271766e-05 lr_backbone=4.271766e-05 elapsed=29:16:43 avg_step=3.70s recent_step=3.69s eta=22:09:34 +2026-09-15 19:03:59 | INFO | train rank=0 | step=28470/50000 progress=56.9% loss=0.111892 lr=4.268494e-05 lr_backbone=4.268494e-05 elapsed=29:17:32 avg_step=3.70s recent_step=4.87s eta=22:09:06 +2026-09-15 19:04:37 | INFO | train rank=0 | step=28480/50000 progress=57.0% loss=0.096235 lr=4.265223e-05 lr_backbone=4.265223e-05 elapsed=29:18:10 avg_step=3.70s recent_step=3.80s eta=22:08:30 +2026-09-15 19:05:06 | INFO | train rank=0 | step=28490/50000 progress=57.0% loss=0.120008 lr=4.261952e-05 lr_backbone=4.261952e-05 elapsed=29:18:39 avg_step=3.70s recent_step=2.87s eta=22:07:47 +2026-09-15 19:05:38 | INFO | train rank=0 | step=28500/50000 progress=57.0% loss=0.084518 lr=4.258682e-05 lr_backbone=4.258682e-05 elapsed=29:19:11 avg_step=3.70s recent_step=3.26s eta=22:07:06 +2026-09-15 19:06:26 | INFO | train rank=0 | step=28510/50000 progress=57.0% loss=0.078087 lr=4.255412e-05 lr_backbone=4.255412e-05 elapsed=29:19:59 avg_step=3.70s recent_step=4.73s eta=22:06:37 +2026-09-15 19:07:01 | INFO | train rank=0 | step=28520/50000 progress=57.0% loss=0.097390 lr=4.252142e-05 lr_backbone=4.252142e-05 elapsed=29:20:34 avg_step=3.70s recent_step=3.51s eta=22:05:58 +2026-09-15 19:07:25 | INFO | train rank=0 | step=28530/50000 progress=57.1% loss=0.087986 lr=4.248872e-05 lr_backbone=4.248872e-05 elapsed=29:20:57 avg_step=3.70s recent_step=2.38s eta=22:05:11 +2026-09-15 19:08:11 | INFO | train rank=0 | step=28540/50000 progress=57.1% loss=0.068516 lr=4.245603e-05 lr_backbone=4.245603e-05 elapsed=29:21:44 avg_step=3.70s recent_step=4.64s eta=22:04:42 +2026-09-15 19:08:55 | INFO | train rank=0 | step=28550/50000 progress=57.1% loss=0.083877 lr=4.242334e-05 lr_backbone=4.242334e-05 elapsed=29:22:27 avg_step=3.70s recent_step=4.36s eta=22:04:09 +2026-09-15 19:09:33 | INFO | train rank=0 | step=28560/50000 progress=57.1% loss=0.095662 lr=4.239065e-05 lr_backbone=4.239065e-05 elapsed=29:23:05 avg_step=3.70s recent_step=3.80s eta=22:03:33 +2026-09-15 19:10:12 | INFO | train rank=0 | step=28570/50000 progress=57.1% loss=0.102881 lr=4.235797e-05 lr_backbone=4.235797e-05 elapsed=29:23:45 avg_step=3.70s recent_step=3.95s eta=22:02:58 +2026-09-15 19:10:42 | INFO | train rank=0 | step=28580/50000 progress=57.2% loss=0.122256 lr=4.232529e-05 lr_backbone=4.232529e-05 elapsed=29:24:15 avg_step=3.70s recent_step=2.97s eta=22:02:15 +2026-09-15 19:11:41 | INFO | train rank=0 | step=28590/50000 progress=57.2% loss=0.097812 lr=4.229262e-05 lr_backbone=4.229262e-05 elapsed=29:25:14 avg_step=3.70s recent_step=5.94s eta=22:01:55 +2026-09-15 19:12:12 | INFO | train rank=0 | step=28600/50000 progress=57.2% loss=0.079883 lr=4.225994e-05 lr_backbone=4.225994e-05 elapsed=29:25:44 avg_step=3.70s recent_step=3.04s eta=22:01:13 +2026-09-15 19:12:52 | INFO | train rank=0 | step=28610/50000 progress=57.2% loss=0.110300 lr=4.222727e-05 lr_backbone=4.222727e-05 elapsed=29:26:25 avg_step=3.70s recent_step=4.09s eta=22:00:39 +2026-09-15 19:13:20 | INFO | train rank=0 | step=28620/50000 progress=57.2% loss=0.080304 lr=4.219461e-05 lr_backbone=4.219461e-05 elapsed=29:26:52 avg_step=3.70s recent_step=2.71s eta=21:59:54 +2026-09-15 19:14:17 | INFO | train rank=0 | step=28630/50000 progress=57.3% loss=0.097389 lr=4.216195e-05 lr_backbone=4.216195e-05 elapsed=29:27:50 avg_step=3.70s recent_step=5.72s eta=21:59:32 +2026-09-15 19:14:46 | INFO | train rank=0 | step=28640/50000 progress=57.3% loss=0.091277 lr=4.212929e-05 lr_backbone=4.212929e-05 elapsed=29:28:19 avg_step=3.70s recent_step=2.93s eta=21:58:49 +2026-09-15 19:15:22 | INFO | train rank=0 | step=28650/50000 progress=57.3% loss=0.102816 lr=4.209663e-05 lr_backbone=4.209663e-05 elapsed=29:28:55 avg_step=3.70s recent_step=3.62s eta=21:58:12 +2026-09-15 19:15:59 | INFO | train rank=0 | step=28660/50000 progress=57.3% loss=0.106537 lr=4.206398e-05 lr_backbone=4.206398e-05 elapsed=29:29:32 avg_step=3.70s recent_step=3.65s eta=21:57:34 +2026-09-15 19:16:51 | INFO | train rank=0 | step=28670/50000 progress=57.3% loss=0.132564 lr=4.203133e-05 lr_backbone=4.203133e-05 elapsed=29:30:24 avg_step=3.71s recent_step=5.22s eta=21:57:09 +2026-09-15 19:17:19 | INFO | train rank=0 | step=28680/50000 progress=57.4% loss=0.096023 lr=4.199869e-05 lr_backbone=4.199869e-05 elapsed=29:30:52 avg_step=3.70s recent_step=2.85s eta=21:56:25 +2026-09-15 19:17:55 | INFO | train rank=0 | step=28690/50000 progress=57.4% loss=0.086668 lr=4.196605e-05 lr_backbone=4.196605e-05 elapsed=29:31:27 avg_step=3.70s recent_step=3.52s eta=21:55:47 +2026-09-15 19:18:29 | INFO | train rank=0 | step=28700/50000 progress=57.4% loss=0.058238 lr=4.193341e-05 lr_backbone=4.193341e-05 elapsed=29:32:01 avg_step=3.70s recent_step=3.39s eta=21:55:07 +2026-09-15 19:19:17 | INFO | train rank=0 | step=28710/50000 progress=57.4% loss=0.097745 lr=4.190077e-05 lr_backbone=4.190077e-05 elapsed=29:32:50 avg_step=3.70s recent_step=4.85s eta=21:54:39 +2026-09-15 19:19:44 | INFO | train rank=0 | step=28720/50000 progress=57.4% loss=0.102944 lr=4.186814e-05 lr_backbone=4.186814e-05 elapsed=29:33:17 avg_step=3.70s recent_step=2.71s eta=21:53:54 +2026-09-15 19:20:22 | INFO | train rank=0 | step=28730/50000 progress=57.5% loss=0.100115 lr=4.183551e-05 lr_backbone=4.183551e-05 elapsed=29:33:55 avg_step=3.70s recent_step=3.76s eta=21:53:18 +2026-09-15 19:20:56 | INFO | train rank=0 | step=28740/50000 progress=57.5% loss=0.086173 lr=4.180289e-05 lr_backbone=4.180289e-05 elapsed=29:34:29 avg_step=3.70s recent_step=3.47s eta=21:52:39 +2026-09-15 19:21:36 | INFO | train rank=0 | step=28750/50000 progress=57.5% loss=0.104747 lr=4.177027e-05 lr_backbone=4.177027e-05 elapsed=29:35:09 avg_step=3.70s recent_step=3.99s eta=21:52:04 +2026-09-15 19:22:08 | INFO | train rank=0 | step=28760/50000 progress=57.5% loss=0.125950 lr=4.173765e-05 lr_backbone=4.173765e-05 elapsed=29:35:41 avg_step=3.70s recent_step=3.20s eta=21:51:23 +2026-09-15 19:22:41 | INFO | train rank=0 | step=28770/50000 progress=57.5% loss=0.083889 lr=4.170504e-05 lr_backbone=4.170504e-05 elapsed=29:36:13 avg_step=3.70s recent_step=3.23s eta=21:50:43 +2026-09-15 19:23:20 | INFO | train rank=0 | step=28780/50000 progress=57.6% loss=0.122898 lr=4.167243e-05 lr_backbone=4.167243e-05 elapsed=29:36:53 avg_step=3.70s recent_step=3.92s eta=21:50:07 +2026-09-15 19:24:07 | INFO | train rank=0 | step=28790/50000 progress=57.6% loss=0.103646 lr=4.163983e-05 lr_backbone=4.163983e-05 elapsed=29:37:39 avg_step=3.70s recent_step=4.67s eta=21:49:37 +2026-09-15 19:24:39 | INFO | train rank=0 | step=28800/50000 progress=57.6% loss=0.078122 lr=4.160722e-05 lr_backbone=4.160722e-05 elapsed=29:38:12 avg_step=3.70s recent_step=3.24s eta=21:48:57 +2026-09-15 19:25:16 | INFO | train rank=0 | step=28810/50000 progress=57.6% loss=0.094244 lr=4.157463e-05 lr_backbone=4.157463e-05 elapsed=29:38:49 avg_step=3.70s recent_step=3.67s eta=21:48:20 +2026-09-15 19:25:49 | INFO | train rank=0 | step=28820/50000 progress=57.6% loss=0.092737 lr=4.154203e-05 lr_backbone=4.154203e-05 elapsed=29:39:22 avg_step=3.70s recent_step=3.31s eta=21:47:40 +2026-09-15 19:26:35 | INFO | train rank=0 | step=28830/50000 progress=57.7% loss=0.105162 lr=4.150944e-05 lr_backbone=4.150944e-05 elapsed=29:40:08 avg_step=3.70s recent_step=4.59s eta=21:47:09 +2026-09-15 19:27:12 | INFO | train rank=0 | step=28840/50000 progress=57.7% loss=0.100808 lr=4.147685e-05 lr_backbone=4.147685e-05 elapsed=29:40:45 avg_step=3.70s recent_step=3.78s eta=21:46:33 +2026-09-15 19:27:46 | INFO | train rank=0 | step=28850/50000 progress=57.7% loss=0.087972 lr=4.144427e-05 lr_backbone=4.144427e-05 elapsed=29:41:19 avg_step=3.70s recent_step=3.34s eta=21:45:53 +2026-09-15 19:28:23 | INFO | train rank=0 | step=28860/50000 progress=57.7% loss=0.103393 lr=4.141169e-05 lr_backbone=4.141169e-05 elapsed=29:41:56 avg_step=3.70s recent_step=3.74s eta=21:45:16 +2026-09-15 19:29:07 | INFO | train rank=0 | step=28870/50000 progress=57.7% loss=0.099740 lr=4.137911e-05 lr_backbone=4.137911e-05 elapsed=29:42:40 avg_step=3.70s recent_step=4.36s eta=21:44:44 +2026-09-15 19:29:38 | INFO | train rank=0 | step=28880/50000 progress=57.8% loss=0.091602 lr=4.134654e-05 lr_backbone=4.134654e-05 elapsed=29:43:11 avg_step=3.70s recent_step=3.10s eta=21:44:02 +2026-09-15 19:30:11 | INFO | train rank=0 | step=28890/50000 progress=57.8% loss=0.087464 lr=4.131397e-05 lr_backbone=4.131397e-05 elapsed=29:43:43 avg_step=3.70s recent_step=3.28s eta=21:43:22 +2026-09-15 19:30:51 | INFO | train rank=0 | step=28900/50000 progress=57.8% loss=0.087271 lr=4.128141e-05 lr_backbone=4.128141e-05 elapsed=29:44:24 avg_step=3.70s recent_step=4.06s eta=21:42:48 +2026-09-15 19:31:43 | INFO | train rank=0 | step=28910/50000 progress=57.8% loss=0.104811 lr=4.124885e-05 lr_backbone=4.124885e-05 elapsed=29:45:16 avg_step=3.71s recent_step=5.19s eta=21:42:22 +2026-09-15 19:32:16 | INFO | train rank=0 | step=28920/50000 progress=57.8% loss=0.104964 lr=4.121629e-05 lr_backbone=4.121629e-05 elapsed=29:45:48 avg_step=3.71s recent_step=3.24s eta=21:41:41 +2026-09-15 19:32:47 | INFO | train rank=0 | step=28930/50000 progress=57.9% loss=0.112719 lr=4.118374e-05 lr_backbone=4.118374e-05 elapsed=29:46:20 avg_step=3.70s recent_step=3.18s eta=21:41:00 +2026-09-15 19:33:22 | INFO | train rank=0 | step=28940/50000 progress=57.9% loss=0.110253 lr=4.115119e-05 lr_backbone=4.115119e-05 elapsed=29:46:55 avg_step=3.70s recent_step=3.44s eta=21:40:21 +2026-09-15 19:34:18 | INFO | train rank=0 | step=28950/50000 progress=57.9% loss=0.062627 lr=4.111864e-05 lr_backbone=4.111864e-05 elapsed=29:47:51 avg_step=3.71s recent_step=5.60s eta=21:39:58 +2026-09-15 19:34:48 | INFO | train rank=0 | step=28960/50000 progress=57.9% loss=0.092631 lr=4.108610e-05 lr_backbone=4.108610e-05 elapsed=29:48:21 avg_step=3.71s recent_step=3.04s eta=21:39:16 +2026-09-15 19:35:20 | INFO | train rank=0 | step=28970/50000 progress=57.9% loss=0.103305 lr=4.105356e-05 lr_backbone=4.105356e-05 elapsed=29:48:53 avg_step=3.70s recent_step=3.20s eta=21:38:35 +2026-09-15 19:35:56 | INFO | train rank=0 | step=28980/50000 progress=58.0% loss=0.123076 lr=4.102103e-05 lr_backbone=4.102103e-05 elapsed=29:49:29 avg_step=3.70s recent_step=3.56s eta=21:37:57 +2026-09-15 19:36:46 | INFO | train rank=0 | step=28990/50000 progress=58.0% loss=0.087175 lr=4.098850e-05 lr_backbone=4.098850e-05 elapsed=29:50:19 avg_step=3.71s recent_step=4.99s eta=21:37:30 +2026-09-15 19:37:09 | INFO | train rank=0 | step=29000/50000 progress=58.0% loss=0.096924 lr=4.095597e-05 lr_backbone=4.095597e-05 elapsed=29:50:42 avg_step=3.70s recent_step=2.36s eta=21:36:43 +2026-09-15 19:37:40 | INFO | train rank=0 | step=29010/50000 progress=58.0% loss=0.085357 lr=4.092345e-05 lr_backbone=4.092345e-05 elapsed=29:51:13 avg_step=3.70s recent_step=3.09s eta=21:36:01 +2026-09-15 19:38:03 | INFO | train rank=0 | step=29020/50000 progress=58.0% loss=0.063519 lr=4.089093e-05 lr_backbone=4.089093e-05 elapsed=29:51:36 avg_step=3.70s recent_step=2.28s eta=21:35:14 +2026-09-15 19:39:00 | INFO | train rank=0 | step=29030/50000 progress=58.1% loss=0.097081 lr=4.085842e-05 lr_backbone=4.085842e-05 elapsed=29:52:33 avg_step=3.70s recent_step=5.68s eta=21:34:51 +2026-09-15 19:39:26 | INFO | train rank=0 | step=29040/50000 progress=58.1% loss=0.149541 lr=4.082591e-05 lr_backbone=4.082591e-05 elapsed=29:52:59 avg_step=3.70s recent_step=2.65s eta=21:34:07 +2026-09-15 19:40:05 | INFO | train rank=0 | step=29050/50000 progress=58.1% loss=0.099162 lr=4.079340e-05 lr_backbone=4.079340e-05 elapsed=29:53:38 avg_step=3.70s recent_step=3.90s eta=21:33:31 +2026-09-15 19:40:37 | INFO | train rank=0 | step=29060/50000 progress=58.1% loss=0.096367 lr=4.076090e-05 lr_backbone=4.076090e-05 elapsed=29:54:10 avg_step=3.70s recent_step=3.16s eta=21:32:50 +2026-09-15 19:41:25 | INFO | train rank=0 | step=29070/50000 progress=58.1% loss=0.075844 lr=4.072840e-05 lr_backbone=4.072840e-05 elapsed=29:54:58 avg_step=3.70s recent_step=4.85s eta=21:32:21 +2026-09-15 19:42:03 | INFO | train rank=0 | step=29080/50000 progress=58.2% loss=0.116015 lr=4.069591e-05 lr_backbone=4.069591e-05 elapsed=29:55:36 avg_step=3.70s recent_step=3.77s eta=21:31:44 +2026-09-15 19:42:39 | INFO | train rank=0 | step=29090/50000 progress=58.2% loss=0.124055 lr=4.066342e-05 lr_backbone=4.066342e-05 elapsed=29:56:12 avg_step=3.70s recent_step=3.57s eta=21:31:06 +2026-09-15 19:43:10 | INFO | train rank=0 | step=29100/50000 progress=58.2% loss=0.096737 lr=4.063093e-05 lr_backbone=4.063093e-05 elapsed=29:56:43 avg_step=3.70s recent_step=3.15s eta=21:30:25 +2026-09-15 19:44:02 | INFO | train rank=0 | step=29110/50000 progress=58.2% loss=0.093763 lr=4.059845e-05 lr_backbone=4.059845e-05 elapsed=29:57:35 avg_step=3.71s recent_step=5.16s eta=21:29:59 +2026-09-15 19:44:38 | INFO | train rank=0 | step=29120/50000 progress=58.2% loss=0.105786 lr=4.056598e-05 lr_backbone=4.056598e-05 elapsed=29:58:11 avg_step=3.71s recent_step=3.64s eta=21:29:21 +2026-09-15 19:45:10 | INFO | train rank=0 | step=29130/50000 progress=58.3% loss=0.134426 lr=4.053350e-05 lr_backbone=4.053350e-05 elapsed=29:58:43 avg_step=3.70s recent_step=3.21s eta=21:28:41 +2026-09-15 19:45:42 | INFO | train rank=0 | step=29140/50000 progress=58.3% loss=0.121868 lr=4.050103e-05 lr_backbone=4.050103e-05 elapsed=29:59:15 avg_step=3.70s recent_step=3.18s eta=21:28:00 +2026-09-15 19:46:36 | INFO | train rank=0 | step=29150/50000 progress=58.3% loss=0.088710 lr=4.046857e-05 lr_backbone=4.046857e-05 elapsed=30:00:09 avg_step=3.71s recent_step=5.39s eta=21:27:35 +2026-09-15 19:47:15 | INFO | train rank=0 | step=29160/50000 progress=58.3% loss=0.123022 lr=4.043611e-05 lr_backbone=4.043611e-05 elapsed=30:00:48 avg_step=3.71s recent_step=3.89s eta=21:26:59 +2026-09-15 19:47:43 | INFO | train rank=0 | step=29170/50000 progress=58.3% loss=0.095984 lr=4.040365e-05 lr_backbone=4.040365e-05 elapsed=30:01:16 avg_step=3.71s recent_step=2.84s eta=21:26:16 +2026-09-15 19:48:25 | INFO | train rank=0 | step=29180/50000 progress=58.4% loss=0.098843 lr=4.037120e-05 lr_backbone=4.037120e-05 elapsed=30:01:57 avg_step=3.71s recent_step=4.13s eta=21:25:42 +2026-09-15 19:49:11 | INFO | train rank=0 | step=29190/50000 progress=58.4% loss=0.104597 lr=4.033875e-05 lr_backbone=4.033875e-05 elapsed=30:02:44 avg_step=3.71s recent_step=4.65s eta=21:25:12 +2026-09-15 19:49:44 | INFO | train rank=0 | step=29200/50000 progress=58.4% loss=0.079792 lr=4.030631e-05 lr_backbone=4.030631e-05 elapsed=30:03:17 avg_step=3.71s recent_step=3.32s eta=21:24:32 +2026-09-15 19:50:11 | INFO | train rank=0 | step=29210/50000 progress=58.4% loss=0.102326 lr=4.027387e-05 lr_backbone=4.027387e-05 elapsed=30:03:43 avg_step=3.71s recent_step=2.62s eta=21:23:47 +2026-09-15 19:50:39 | INFO | train rank=0 | step=29220/50000 progress=58.4% loss=0.111043 lr=4.024143e-05 lr_backbone=4.024143e-05 elapsed=30:04:12 avg_step=3.70s recent_step=2.85s eta=21:23:04 +2026-09-15 19:51:18 | INFO | train rank=0 | step=29230/50000 progress=58.5% loss=0.089656 lr=4.020900e-05 lr_backbone=4.020900e-05 elapsed=30:04:51 avg_step=3.70s recent_step=3.94s eta=21:22:28 +2026-09-15 19:51:48 | INFO | train rank=0 | step=29240/50000 progress=58.5% loss=0.100542 lr=4.017657e-05 lr_backbone=4.017657e-05 elapsed=30:05:20 avg_step=3.70s recent_step=2.93s eta=21:21:46 +2026-09-15 19:52:26 | INFO | train rank=0 | step=29250/50000 progress=58.5% loss=0.129955 lr=4.014415e-05 lr_backbone=4.014415e-05 elapsed=30:05:59 avg_step=3.70s recent_step=3.82s eta=21:21:10 +2026-09-15 19:52:56 | INFO | train rank=0 | step=29260/50000 progress=58.5% loss=0.133085 lr=4.011173e-05 lr_backbone=4.011173e-05 elapsed=30:06:28 avg_step=3.70s recent_step=2.97s eta=21:20:27 +2026-09-15 19:53:38 | INFO | train rank=0 | step=29270/50000 progress=58.5% loss=0.109335 lr=4.007932e-05 lr_backbone=4.007932e-05 elapsed=30:07:11 avg_step=3.70s recent_step=4.24s eta=21:19:54 +2026-09-15 19:54:09 | INFO | train rank=0 | step=29280/50000 progress=58.6% loss=0.087167 lr=4.004691e-05 lr_backbone=4.004691e-05 elapsed=30:07:42 avg_step=3.70s recent_step=3.12s eta=21:19:13 +2026-09-15 19:54:45 | INFO | train rank=0 | step=29290/50000 progress=58.6% loss=0.076258 lr=4.001450e-05 lr_backbone=4.001450e-05 elapsed=30:08:18 avg_step=3.70s recent_step=3.62s eta=21:18:35 +2026-09-15 19:55:21 | INFO | train rank=0 | step=29300/50000 progress=58.6% loss=0.099160 lr=3.998210e-05 lr_backbone=3.998210e-05 elapsed=30:08:54 avg_step=3.70s recent_step=3.57s eta=21:17:57 +2026-09-15 19:56:09 | INFO | train rank=0 | step=29310/50000 progress=58.6% loss=0.094106 lr=3.994970e-05 lr_backbone=3.994970e-05 elapsed=30:09:41 avg_step=3.70s recent_step=4.75s eta=21:17:28 +2026-09-15 19:56:43 | INFO | train rank=0 | step=29320/50000 progress=58.6% loss=0.128869 lr=3.991731e-05 lr_backbone=3.991731e-05 elapsed=30:10:16 avg_step=3.70s recent_step=3.44s eta=21:16:49 +2026-09-15 19:57:21 | INFO | train rank=0 | step=29330/50000 progress=58.7% loss=0.105760 lr=3.988492e-05 lr_backbone=3.988492e-05 elapsed=30:10:54 avg_step=3.70s recent_step=3.82s eta=21:16:12 +2026-09-15 19:57:58 | INFO | train rank=0 | step=29340/50000 progress=58.7% loss=0.096284 lr=3.985254e-05 lr_backbone=3.985254e-05 elapsed=30:11:31 avg_step=3.70s recent_step=3.68s eta=21:15:35 +2026-09-15 19:58:44 | INFO | train rank=0 | step=29350/50000 progress=58.7% loss=0.083601 lr=3.982016e-05 lr_backbone=3.982016e-05 elapsed=30:12:17 avg_step=3.70s recent_step=4.65s eta=21:15:05 +2026-09-15 19:59:17 | INFO | train rank=0 | step=29360/50000 progress=58.7% loss=0.070440 lr=3.978779e-05 lr_backbone=3.978779e-05 elapsed=30:12:49 avg_step=3.70s recent_step=3.23s eta=21:14:24 +2026-09-15 19:59:58 | INFO | train rank=0 | step=29370/50000 progress=58.7% loss=0.112765 lr=3.975542e-05 lr_backbone=3.975542e-05 elapsed=30:13:30 avg_step=3.70s recent_step=4.10s eta=21:13:50 +2026-09-15 20:00:33 | INFO | train rank=0 | step=29380/50000 progress=58.8% loss=0.118973 lr=3.972305e-05 lr_backbone=3.972305e-05 elapsed=30:14:06 avg_step=3.70s recent_step=3.52s eta=21:13:12 +2026-09-15 20:01:21 | INFO | train rank=0 | step=29390/50000 progress=58.8% loss=0.153375 lr=3.969069e-05 lr_backbone=3.969069e-05 elapsed=30:14:54 avg_step=3.71s recent_step=4.86s eta=21:12:43 +2026-09-15 20:02:00 | INFO | train rank=0 | step=29400/50000 progress=58.8% loss=0.080719 lr=3.965833e-05 lr_backbone=3.965833e-05 elapsed=30:15:32 avg_step=3.71s recent_step=3.81s eta=21:12:07 +2026-09-15 20:02:25 | INFO | train rank=0 | step=29410/50000 progress=58.8% loss=0.109907 lr=3.962598e-05 lr_backbone=3.962598e-05 elapsed=30:15:58 avg_step=3.70s recent_step=2.54s eta=21:11:21 +2026-09-15 20:03:04 | INFO | train rank=0 | step=29420/50000 progress=58.8% loss=0.138638 lr=3.959363e-05 lr_backbone=3.959363e-05 elapsed=30:16:37 avg_step=3.70s recent_step=3.90s eta=21:10:46 +2026-09-15 20:03:53 | INFO | train rank=0 | step=29430/50000 progress=58.9% loss=0.067735 lr=3.956129e-05 lr_backbone=3.956129e-05 elapsed=30:17:26 avg_step=3.71s recent_step=4.92s eta=21:10:17 +2026-09-15 20:04:20 | INFO | train rank=0 | step=29440/50000 progress=58.9% loss=0.118307 lr=3.952895e-05 lr_backbone=3.952895e-05 elapsed=30:17:53 avg_step=3.70s recent_step=2.72s eta=21:09:33 +2026-09-15 20:04:54 | INFO | train rank=0 | step=29450/50000 progress=58.9% loss=0.100881 lr=3.949662e-05 lr_backbone=3.949662e-05 elapsed=30:18:27 avg_step=3.70s recent_step=3.36s eta=21:08:54 +2026-09-15 20:05:16 | INFO | train rank=0 | step=29460/50000 progress=58.9% loss=0.096688 lr=3.946429e-05 lr_backbone=3.946429e-05 elapsed=30:18:49 avg_step=3.70s recent_step=2.22s eta=21:08:06 +2026-09-15 20:06:03 | INFO | train rank=0 | step=29470/50000 progress=58.9% loss=0.083115 lr=3.943196e-05 lr_backbone=3.943196e-05 elapsed=30:19:36 avg_step=3.70s recent_step=4.67s eta=21:07:36 +2026-09-15 20:06:35 | INFO | train rank=0 | step=29480/50000 progress=59.0% loss=0.126115 lr=3.939964e-05 lr_backbone=3.939964e-05 elapsed=30:20:07 avg_step=3.70s recent_step=3.17s eta=21:06:55 +2026-09-15 20:07:15 | INFO | train rank=0 | step=29490/50000 progress=59.0% loss=0.098412 lr=3.936733e-05 lr_backbone=3.936733e-05 elapsed=30:20:48 avg_step=3.70s recent_step=4.07s eta=21:06:21 +2026-09-15 20:07:48 | INFO | train rank=0 | step=29500/50000 progress=59.0% loss=0.095017 lr=3.933502e-05 lr_backbone=3.933502e-05 elapsed=30:21:21 avg_step=3.70s recent_step=3.24s eta=21:05:41 +2026-09-15 20:08:39 | INFO | train rank=0 | step=29510/50000 progress=59.0% loss=0.091020 lr=3.930271e-05 lr_backbone=3.930271e-05 elapsed=30:22:11 avg_step=3.70s recent_step=5.10s eta=21:05:13 +2026-09-15 20:09:17 | INFO | train rank=0 | step=29520/50000 progress=59.0% loss=0.068395 lr=3.927041e-05 lr_backbone=3.927041e-05 elapsed=30:22:50 avg_step=3.70s recent_step=3.81s eta=21:04:37 +2026-09-15 20:09:57 | INFO | train rank=0 | step=29530/50000 progress=59.1% loss=0.103731 lr=3.923811e-05 lr_backbone=3.923811e-05 elapsed=30:23:29 avg_step=3.71s recent_step=3.98s eta=21:04:02 +2026-09-15 20:10:25 | INFO | train rank=0 | step=29540/50000 progress=59.1% loss=0.087630 lr=3.920582e-05 lr_backbone=3.920582e-05 elapsed=30:23:58 avg_step=3.70s recent_step=2.88s eta=21:03:19 +2026-09-15 20:11:21 | INFO | train rank=0 | step=29550/50000 progress=59.1% loss=0.131900 lr=3.917354e-05 lr_backbone=3.917354e-05 elapsed=30:24:54 avg_step=3.71s recent_step=5.54s eta=21:02:55 +2026-09-15 20:11:54 | INFO | train rank=0 | step=29560/50000 progress=59.1% loss=0.089792 lr=3.914125e-05 lr_backbone=3.914125e-05 elapsed=30:25:27 avg_step=3.71s recent_step=3.31s eta=21:02:15 +2026-09-15 20:12:27 | INFO | train rank=0 | step=29570/50000 progress=59.1% loss=0.099035 lr=3.910898e-05 lr_backbone=3.910898e-05 elapsed=30:26:00 avg_step=3.71s recent_step=3.27s eta=21:01:35 +2026-09-15 20:13:02 | INFO | train rank=0 | step=29580/50000 progress=59.2% loss=0.128628 lr=3.907670e-05 lr_backbone=3.907670e-05 elapsed=30:26:35 avg_step=3.71s recent_step=3.51s eta=21:00:56 +2026-09-15 20:13:45 | INFO | train rank=0 | step=29590/50000 progress=59.2% loss=0.104309 lr=3.904443e-05 lr_backbone=3.904443e-05 elapsed=30:27:17 avg_step=3.71s recent_step=4.27s eta=21:00:23 +2026-09-15 20:14:13 | INFO | train rank=0 | step=29600/50000 progress=59.2% loss=0.085564 lr=3.901217e-05 lr_backbone=3.901217e-05 elapsed=30:27:45 avg_step=3.70s recent_step=2.80s eta=20:59:40 +2026-09-15 20:14:41 | INFO | train rank=0 | step=29610/50000 progress=59.2% loss=0.112953 lr=3.897991e-05 lr_backbone=3.897991e-05 elapsed=30:28:13 avg_step=3.70s recent_step=2.81s eta=20:58:57 +2026-09-15 20:15:14 | INFO | train rank=0 | step=29620/50000 progress=59.2% loss=0.078148 lr=3.894766e-05 lr_backbone=3.894766e-05 elapsed=30:28:47 avg_step=3.70s recent_step=3.37s eta=20:58:17 +2026-09-15 20:15:59 | INFO | train rank=0 | step=29630/50000 progress=59.3% loss=0.088451 lr=3.891541e-05 lr_backbone=3.891541e-05 elapsed=30:29:32 avg_step=3.70s recent_step=4.44s eta=20:57:45 +2026-09-15 20:16:25 | INFO | train rank=0 | step=29640/50000 progress=59.3% loss=0.104521 lr=3.888317e-05 lr_backbone=3.888317e-05 elapsed=30:29:58 avg_step=3.70s recent_step=2.63s eta=20:57:01 +2026-09-15 20:17:24 | INFO | train rank=0 | step=29650/50000 progress=59.3% loss=0.092353 lr=3.885093e-05 lr_backbone=3.885093e-05 elapsed=30:30:57 avg_step=3.71s recent_step=5.93s eta=20:56:39 +2026-09-15 20:18:00 | INFO | train rank=0 | step=29660/50000 progress=59.3% loss=0.085490 lr=3.881869e-05 lr_backbone=3.881869e-05 elapsed=30:31:32 avg_step=3.71s recent_step=3.54s eta=20:56:01 +2026-09-15 20:18:34 | INFO | train rank=0 | step=29670/50000 progress=59.3% loss=0.107639 lr=3.878646e-05 lr_backbone=3.878646e-05 elapsed=30:32:07 avg_step=3.70s recent_step=3.42s eta=20:55:22 +2026-09-15 20:19:13 | INFO | train rank=0 | step=29680/50000 progress=59.4% loss=0.105251 lr=3.875424e-05 lr_backbone=3.875424e-05 elapsed=30:32:46 avg_step=3.71s recent_step=3.94s eta=20:54:47 +2026-09-15 20:19:51 | INFO | train rank=0 | step=29690/50000 progress=59.4% loss=0.095416 lr=3.872202e-05 lr_backbone=3.872202e-05 elapsed=30:33:24 avg_step=3.71s recent_step=3.75s eta=20:54:10 +2026-09-15 20:20:22 | INFO | train rank=0 | step=29700/50000 progress=59.4% loss=0.092618 lr=3.868980e-05 lr_backbone=3.868980e-05 elapsed=30:33:55 avg_step=3.70s recent_step=3.12s eta=20:53:29 +2026-09-15 20:21:09 | INFO | train rank=0 | step=29710/50000 progress=59.4% loss=0.084893 lr=3.865759e-05 lr_backbone=3.865759e-05 elapsed=30:34:42 avg_step=3.71s recent_step=4.70s eta=20:52:59 +2026-09-15 20:21:45 | INFO | train rank=0 | step=29720/50000 progress=59.4% loss=0.102577 lr=3.862539e-05 lr_backbone=3.862539e-05 elapsed=30:35:18 avg_step=3.71s recent_step=3.64s eta=20:52:21 +2026-09-15 20:22:25 | INFO | train rank=0 | step=29730/50000 progress=59.5% loss=0.098916 lr=3.859319e-05 lr_backbone=3.859319e-05 elapsed=30:35:58 avg_step=3.71s recent_step=4.00s eta=20:51:46 +2026-09-15 20:23:03 | INFO | train rank=0 | step=29740/50000 progress=59.5% loss=0.095462 lr=3.856099e-05 lr_backbone=3.856099e-05 elapsed=30:36:36 avg_step=3.71s recent_step=3.74s eta=20:51:09 +2026-09-15 20:23:43 | INFO | train rank=0 | step=29750/50000 progress=59.5% loss=0.111894 lr=3.852880e-05 lr_backbone=3.852880e-05 elapsed=30:37:15 avg_step=3.71s recent_step=3.97s eta=20:50:34 +2026-09-15 20:24:17 | INFO | train rank=0 | step=29760/50000 progress=59.5% loss=0.116033 lr=3.849662e-05 lr_backbone=3.849662e-05 elapsed=30:37:50 avg_step=3.71s recent_step=3.47s eta=20:49:55 +2026-09-15 20:24:53 | INFO | train rank=0 | step=29770/50000 progress=59.5% loss=0.105306 lr=3.846444e-05 lr_backbone=3.846444e-05 elapsed=30:38:26 avg_step=3.71s recent_step=3.56s eta=20:49:17 +2026-09-15 20:25:28 | INFO | train rank=0 | step=29780/50000 progress=59.6% loss=0.103049 lr=3.843226e-05 lr_backbone=3.843226e-05 elapsed=30:39:00 avg_step=3.71s recent_step=3.45s eta=20:48:39 +2026-09-15 20:26:13 | INFO | train rank=0 | step=29790/50000 progress=59.6% loss=0.095316 lr=3.840009e-05 lr_backbone=3.840009e-05 elapsed=30:39:46 avg_step=3.71s recent_step=4.56s eta=20:48:07 +2026-09-15 20:26:51 | INFO | train rank=0 | step=29800/50000 progress=59.6% loss=0.101363 lr=3.836793e-05 lr_backbone=3.836793e-05 elapsed=30:40:24 avg_step=3.71s recent_step=3.83s eta=20:47:31 +2026-09-15 20:27:20 | INFO | train rank=0 | step=29810/50000 progress=59.6% loss=0.100776 lr=3.833577e-05 lr_backbone=3.833577e-05 elapsed=30:40:53 avg_step=3.71s recent_step=2.84s eta=20:46:48 +2026-09-15 20:27:59 | INFO | train rank=0 | step=29820/50000 progress=59.6% loss=0.109161 lr=3.830362e-05 lr_backbone=3.830362e-05 elapsed=30:41:32 avg_step=3.71s recent_step=3.94s eta=20:46:13 +2026-09-15 20:28:52 | INFO | train rank=0 | step=29830/50000 progress=59.7% loss=0.074869 lr=3.827147e-05 lr_backbone=3.827147e-05 elapsed=30:42:25 avg_step=3.71s recent_step=5.26s eta=20:45:46 +2026-09-15 20:29:30 | INFO | train rank=0 | step=29840/50000 progress=59.7% loss=0.071462 lr=3.823932e-05 lr_backbone=3.823932e-05 elapsed=30:43:03 avg_step=3.71s recent_step=3.83s eta=20:45:10 +2026-09-15 20:30:04 | INFO | train rank=0 | step=29850/50000 progress=59.7% loss=0.120351 lr=3.820718e-05 lr_backbone=3.820718e-05 elapsed=30:43:37 avg_step=3.71s recent_step=3.42s eta=20:44:31 +2026-09-15 20:30:43 | INFO | train rank=0 | step=29860/50000 progress=59.7% loss=0.105604 lr=3.817505e-05 lr_backbone=3.817505e-05 elapsed=30:44:16 avg_step=3.71s recent_step=3.86s eta=20:43:55 +2026-09-15 20:31:27 | INFO | train rank=0 | step=29870/50000 progress=59.7% loss=0.093612 lr=3.814292e-05 lr_backbone=3.814292e-05 elapsed=30:45:00 avg_step=3.71s recent_step=4.43s eta=20:43:23 +2026-09-15 20:31:58 | INFO | train rank=0 | step=29880/50000 progress=59.8% loss=0.102195 lr=3.811080e-05 lr_backbone=3.811080e-05 elapsed=30:45:31 avg_step=3.71s recent_step=3.11s eta=20:42:42 +2026-09-15 20:32:35 | INFO | train rank=0 | step=29890/50000 progress=59.8% loss=0.099938 lr=3.807868e-05 lr_backbone=3.807868e-05 elapsed=30:46:07 avg_step=3.71s recent_step=3.63s eta=20:42:04 +2026-09-15 20:33:06 | INFO | train rank=0 | step=29900/50000 progress=59.8% loss=0.096800 lr=3.804657e-05 lr_backbone=3.804657e-05 elapsed=30:46:39 avg_step=3.71s recent_step=3.11s eta=20:41:23 +2026-09-15 20:33:52 | INFO | train rank=0 | step=29910/50000 progress=59.8% loss=0.085806 lr=3.801446e-05 lr_backbone=3.801446e-05 elapsed=30:47:25 avg_step=3.71s recent_step=4.63s eta=20:40:52 +2026-09-15 20:34:27 | INFO | train rank=0 | step=29920/50000 progress=59.8% loss=0.080809 lr=3.798236e-05 lr_backbone=3.798236e-05 elapsed=30:48:00 avg_step=3.71s recent_step=3.54s eta=20:40:14 +2026-09-15 20:34:59 | INFO | train rank=0 | step=29930/50000 progress=59.9% loss=0.066744 lr=3.795026e-05 lr_backbone=3.795026e-05 elapsed=30:48:32 avg_step=3.71s recent_step=3.16s eta=20:39:33 +2026-09-15 20:35:32 | INFO | train rank=0 | step=29940/50000 progress=59.9% loss=0.088304 lr=3.791817e-05 lr_backbone=3.791817e-05 elapsed=30:49:04 avg_step=3.71s recent_step=3.26s eta=20:38:53 +2026-09-15 20:36:28 | INFO | train rank=0 | step=29950/50000 progress=59.9% loss=0.097467 lr=3.788608e-05 lr_backbone=3.788608e-05 elapsed=30:50:01 avg_step=3.71s recent_step=5.66s eta=20:38:29 +2026-09-15 20:37:00 | INFO | train rank=0 | step=29960/50000 progress=59.9% loss=0.073452 lr=3.785400e-05 lr_backbone=3.785400e-05 elapsed=30:50:33 avg_step=3.71s recent_step=3.21s eta=20:37:49 +2026-09-15 20:37:38 | INFO | train rank=0 | step=29970/50000 progress=59.9% loss=0.083428 lr=3.782192e-05 lr_backbone=3.782192e-05 elapsed=30:51:11 avg_step=3.71s recent_step=3.79s eta=20:37:12 +2026-09-15 20:38:13 | INFO | train rank=0 | step=29980/50000 progress=60.0% loss=0.090250 lr=3.778985e-05 lr_backbone=3.778985e-05 elapsed=30:51:46 avg_step=3.71s recent_step=3.45s eta=20:36:34 +2026-09-15 20:38:57 | INFO | train rank=0 | step=29990/50000 progress=60.0% loss=0.103325 lr=3.775779e-05 lr_backbone=3.775779e-05 elapsed=30:52:30 avg_step=3.71s recent_step=4.45s eta=20:36:02 +2026-09-15 20:39:33 | INFO | train rank=0 | step=30000/50000 progress=60.0% loss=0.116475 lr=3.772573e-05 lr_backbone=3.772573e-05 elapsed=30:53:05 avg_step=3.71s recent_step=3.54s eta=20:35:23 +2026-09-15 20:40:07 | INFO | train rank=0 | step=30010/50000 progress=60.0% loss=0.128982 lr=3.769367e-05 lr_backbone=3.769367e-05 elapsed=30:53:40 avg_step=3.71s recent_step=3.43s eta=20:34:45 +2026-09-15 20:40:40 | INFO | train rank=0 | step=30020/50000 progress=60.0% loss=0.068416 lr=3.766162e-05 lr_backbone=3.766162e-05 elapsed=30:54:13 avg_step=3.71s recent_step=3.28s eta=20:34:05 +2026-09-15 20:41:41 | INFO | train rank=0 | step=30030/50000 progress=60.1% loss=0.102008 lr=3.762958e-05 lr_backbone=3.762958e-05 elapsed=30:55:13 avg_step=3.71s recent_step=6.09s eta=20:33:43 +2026-09-15 20:42:07 | INFO | train rank=0 | step=30040/50000 progress=60.1% loss=0.080841 lr=3.759754e-05 lr_backbone=3.759754e-05 elapsed=30:55:40 avg_step=3.71s recent_step=2.67s eta=20:33:00 +2026-09-15 20:42:53 | INFO | train rank=0 | step=30050/50000 progress=60.1% loss=0.103665 lr=3.756551e-05 lr_backbone=3.756551e-05 elapsed=30:56:25 avg_step=3.71s recent_step=4.53s eta=20:32:28 +2026-09-15 20:43:16 | INFO | train rank=0 | step=30060/50000 progress=60.1% loss=0.098628 lr=3.753348e-05 lr_backbone=3.753348e-05 elapsed=30:56:49 avg_step=3.71s recent_step=2.36s eta=20:31:42 +2026-09-15 20:44:03 | INFO | train rank=0 | step=30070/50000 progress=60.1% loss=0.098547 lr=3.750146e-05 lr_backbone=3.750146e-05 elapsed=30:57:35 avg_step=3.71s recent_step=4.63s eta=20:31:11 +2026-09-15 20:44:27 | INFO | train rank=0 | step=30080/50000 progress=60.2% loss=0.113512 lr=3.746944e-05 lr_backbone=3.746944e-05 elapsed=30:58:00 avg_step=3.71s recent_step=2.49s eta=20:30:26 +2026-09-15 20:44:57 | INFO | train rank=0 | step=30090/50000 progress=60.2% loss=0.088235 lr=3.743743e-05 lr_backbone=3.743743e-05 elapsed=30:58:29 avg_step=3.71s recent_step=2.91s eta=20:29:44 +2026-09-15 20:45:30 | INFO | train rank=0 | step=30100/50000 progress=60.2% loss=0.074873 lr=3.740542e-05 lr_backbone=3.740542e-05 elapsed=30:59:03 avg_step=3.71s recent_step=3.39s eta=20:29:04 +2026-09-15 20:46:34 | INFO | train rank=0 | step=30110/50000 progress=60.2% loss=0.083647 lr=3.737342e-05 lr_backbone=3.737342e-05 elapsed=31:00:07 avg_step=3.71s recent_step=6.33s eta=20:28:45 +2026-09-15 20:47:00 | INFO | train rank=0 | step=30120/50000 progress=60.2% loss=0.115999 lr=3.734143e-05 lr_backbone=3.734143e-05 elapsed=31:00:33 avg_step=3.71s recent_step=2.64s eta=20:28:01 +2026-09-15 20:47:34 | INFO | train rank=0 | step=30130/50000 progress=60.3% loss=0.064797 lr=3.730944e-05 lr_backbone=3.730944e-05 elapsed=31:01:06 avg_step=3.71s recent_step=3.34s eta=20:27:21 +2026-09-15 20:48:02 | INFO | train rank=0 | step=30140/50000 progress=60.3% loss=0.099938 lr=3.727745e-05 lr_backbone=3.727745e-05 elapsed=31:01:35 avg_step=3.71s recent_step=2.84s eta=20:26:38 +2026-09-15 20:48:40 | INFO | train rank=0 | step=30150/50000 progress=60.3% loss=0.088536 lr=3.724548e-05 lr_backbone=3.724548e-05 elapsed=31:02:13 avg_step=3.71s recent_step=3.80s eta=20:26:02 +2026-09-15 20:49:12 | INFO | train rank=0 | step=30160/50000 progress=60.3% loss=0.124415 lr=3.721350e-05 lr_backbone=3.721350e-05 elapsed=31:02:45 avg_step=3.71s recent_step=3.19s eta=20:25:21 +2026-09-15 20:49:52 | INFO | train rank=0 | step=30170/50000 progress=60.3% loss=0.090140 lr=3.718154e-05 lr_backbone=3.718154e-05 elapsed=31:03:25 avg_step=3.71s recent_step=4.02s eta=20:24:46 +2026-09-15 20:50:18 | INFO | train rank=0 | step=30180/50000 progress=60.4% loss=0.169070 lr=3.714957e-05 lr_backbone=3.714957e-05 elapsed=31:03:51 avg_step=3.71s recent_step=2.64s eta=20:24:02 +2026-09-15 20:51:13 | INFO | train rank=0 | step=30190/50000 progress=60.4% loss=0.057827 lr=3.711762e-05 lr_backbone=3.711762e-05 elapsed=31:04:46 avg_step=3.71s recent_step=5.43s eta=20:23:37 +2026-09-15 20:51:54 | INFO | train rank=0 | step=30200/50000 progress=60.4% loss=0.098877 lr=3.708567e-05 lr_backbone=3.708567e-05 elapsed=31:05:26 avg_step=3.71s recent_step=4.07s eta=20:23:02 +2026-09-15 20:52:32 | INFO | train rank=0 | step=30210/50000 progress=60.4% loss=0.091828 lr=3.705372e-05 lr_backbone=3.705372e-05 elapsed=31:06:05 avg_step=3.71s recent_step=3.82s eta=20:22:26 +2026-09-15 20:53:06 | INFO | train rank=0 | step=30220/50000 progress=60.4% loss=0.101707 lr=3.702179e-05 lr_backbone=3.702179e-05 elapsed=31:06:38 avg_step=3.71s recent_step=3.39s eta=20:21:47 +2026-09-15 20:53:54 | INFO | train rank=0 | step=30230/50000 progress=60.5% loss=0.075687 lr=3.698985e-05 lr_backbone=3.698985e-05 elapsed=31:07:27 avg_step=3.71s recent_step=4.87s eta=20:21:17 +2026-09-15 20:54:30 | INFO | train rank=0 | step=30240/50000 progress=60.5% loss=0.085000 lr=3.695792e-05 lr_backbone=3.695792e-05 elapsed=31:08:02 avg_step=3.71s recent_step=3.52s eta=20:20:39 +2026-09-15 20:55:06 | INFO | train rank=0 | step=30250/50000 progress=60.5% loss=0.074333 lr=3.692600e-05 lr_backbone=3.692600e-05 elapsed=31:08:39 avg_step=3.71s recent_step=3.68s eta=20:20:02 +2026-09-15 20:55:42 | INFO | train rank=0 | step=30260/50000 progress=60.5% loss=0.101214 lr=3.689409e-05 lr_backbone=3.689409e-05 elapsed=31:09:15 avg_step=3.71s recent_step=3.60s eta=20:19:24 +2026-09-15 20:56:26 | INFO | train rank=0 | step=30270/50000 progress=60.5% loss=0.097451 lr=3.686218e-05 lr_backbone=3.686218e-05 elapsed=31:09:58 avg_step=3.71s recent_step=4.33s eta=20:18:51 +2026-09-15 20:56:57 | INFO | train rank=0 | step=30280/50000 progress=60.6% loss=0.098431 lr=3.683027e-05 lr_backbone=3.683027e-05 elapsed=31:10:30 avg_step=3.71s recent_step=3.12s eta=20:18:10 +2026-09-15 20:57:42 | INFO | train rank=0 | step=30290/50000 progress=60.6% loss=0.085963 lr=3.679837e-05 lr_backbone=3.679837e-05 elapsed=31:11:14 avg_step=3.71s recent_step=4.48s eta=20:17:38 +2026-09-15 20:58:09 | INFO | train rank=0 | step=30300/50000 progress=60.6% loss=0.102524 lr=3.676648e-05 lr_backbone=3.676648e-05 elapsed=31:11:42 avg_step=3.71s recent_step=2.77s eta=20:16:55 +2026-09-15 20:59:00 | INFO | train rank=0 | step=30310/50000 progress=60.6% loss=0.100803 lr=3.673459e-05 lr_backbone=3.673459e-05 elapsed=31:12:33 avg_step=3.71s recent_step=5.06s eta=20:16:26 +2026-09-15 20:59:33 | INFO | train rank=0 | step=30320/50000 progress=60.6% loss=0.077233 lr=3.670271e-05 lr_backbone=3.670271e-05 elapsed=31:13:05 avg_step=3.71s recent_step=3.26s eta=20:15:47 +2026-09-15 21:00:03 | INFO | train rank=0 | step=30330/50000 progress=60.7% loss=0.097629 lr=3.667084e-05 lr_backbone=3.667084e-05 elapsed=31:13:36 avg_step=3.71s recent_step=3.06s eta=20:15:05 +2026-09-15 21:00:34 | INFO | train rank=0 | step=30340/50000 progress=60.7% loss=0.080296 lr=3.663897e-05 lr_backbone=3.663897e-05 elapsed=31:14:07 avg_step=3.71s recent_step=3.11s eta=20:14:24 +2026-09-15 21:01:25 | INFO | train rank=0 | step=30350/50000 progress=60.7% loss=0.094577 lr=3.660710e-05 lr_backbone=3.660710e-05 elapsed=31:14:58 avg_step=3.71s recent_step=5.09s eta=20:13:56 +2026-09-15 21:01:57 | INFO | train rank=0 | step=30360/50000 progress=60.7% loss=0.108441 lr=3.657524e-05 lr_backbone=3.657524e-05 elapsed=31:15:30 avg_step=3.71s recent_step=3.20s eta=20:13:16 +2026-09-15 21:02:25 | INFO | train rank=0 | step=30370/50000 progress=60.7% loss=0.117583 lr=3.654339e-05 lr_backbone=3.654339e-05 elapsed=31:15:57 avg_step=3.71s recent_step=2.76s eta=20:12:33 +2026-09-15 21:02:55 | INFO | train rank=0 | step=30380/50000 progress=60.8% loss=0.081109 lr=3.651155e-05 lr_backbone=3.651155e-05 elapsed=31:16:28 avg_step=3.71s recent_step=3.00s eta=20:11:51 +2026-09-15 21:03:45 | INFO | train rank=0 | step=30390/50000 progress=60.8% loss=0.114779 lr=3.647971e-05 lr_backbone=3.647971e-05 elapsed=31:17:18 avg_step=3.71s recent_step=5.08s eta=20:11:23 +2026-09-15 21:04:15 | INFO | train rank=0 | step=30400/50000 progress=60.8% loss=0.096351 lr=3.644787e-05 lr_backbone=3.644787e-05 elapsed=31:17:48 avg_step=3.71s recent_step=3.00s eta=20:10:41 +2026-09-15 21:04:38 | INFO | train rank=0 | step=30410/50000 progress=60.8% loss=0.089856 lr=3.641604e-05 lr_backbone=3.641604e-05 elapsed=31:18:11 avg_step=3.71s recent_step=2.25s eta=20:09:55 +2026-09-15 21:05:11 | INFO | train rank=0 | step=30420/50000 progress=60.8% loss=0.100570 lr=3.638422e-05 lr_backbone=3.638422e-05 elapsed=31:18:43 avg_step=3.71s recent_step=3.27s eta=20:09:15 +2026-09-15 21:06:11 | INFO | train rank=0 | step=30430/50000 progress=60.9% loss=0.081028 lr=3.635240e-05 lr_backbone=3.635240e-05 elapsed=31:19:43 avg_step=3.71s recent_step=6.00s eta=20:08:53 +2026-09-15 21:06:46 | INFO | train rank=0 | step=30440/50000 progress=60.9% loss=0.103398 lr=3.632059e-05 lr_backbone=3.632059e-05 elapsed=31:20:19 avg_step=3.71s recent_step=3.56s eta=20:08:15 +2026-09-15 21:07:19 | INFO | train rank=0 | step=30450/50000 progress=60.9% loss=0.081686 lr=3.628879e-05 lr_backbone=3.628879e-05 elapsed=31:20:52 avg_step=3.71s recent_step=3.25s eta=20:07:35 +2026-09-15 21:07:54 | INFO | train rank=0 | step=30460/50000 progress=60.9% loss=0.097877 lr=3.625699e-05 lr_backbone=3.625699e-05 elapsed=31:21:26 avg_step=3.71s recent_step=3.49s eta=20:06:56 +2026-09-15 21:08:46 | INFO | train rank=0 | step=30470/50000 progress=60.9% loss=0.089349 lr=3.622520e-05 lr_backbone=3.622520e-05 elapsed=31:22:19 avg_step=3.71s recent_step=5.28s eta=20:06:29 +2026-09-15 21:09:19 | INFO | train rank=0 | step=30480/50000 progress=61.0% loss=0.082465 lr=3.619341e-05 lr_backbone=3.619341e-05 elapsed=31:22:52 avg_step=3.71s recent_step=3.27s eta=20:05:49 +2026-09-15 21:09:50 | INFO | train rank=0 | step=30490/50000 progress=61.0% loss=0.095718 lr=3.616163e-05 lr_backbone=3.616163e-05 elapsed=31:23:23 avg_step=3.71s recent_step=3.06s eta=20:05:08 +2026-09-15 21:10:18 | INFO | train rank=0 | step=30500/50000 progress=61.0% loss=0.096348 lr=3.612985e-05 lr_backbone=3.612985e-05 elapsed=31:23:51 avg_step=3.71s recent_step=2.81s eta=20:04:25 +2026-09-15 21:11:23 | INFO | train rank=0 | step=30510/50000 progress=61.0% loss=0.076840 lr=3.609809e-05 lr_backbone=3.609809e-05 elapsed=31:24:56 avg_step=3.71s recent_step=6.50s eta=20:04:06 +2026-09-15 21:11:50 | INFO | train rank=0 | step=30520/50000 progress=61.0% loss=0.094493 lr=3.606632e-05 lr_backbone=3.606632e-05 elapsed=31:25:23 avg_step=3.71s recent_step=2.71s eta=20:03:23 +2026-09-15 21:12:17 | INFO | train rank=0 | step=30530/50000 progress=61.1% loss=0.106901 lr=3.603457e-05 lr_backbone=3.603457e-05 elapsed=31:25:50 avg_step=3.71s recent_step=2.68s eta=20:02:39 +2026-09-15 21:12:47 | INFO | train rank=0 | step=30540/50000 progress=61.1% loss=0.109141 lr=3.600282e-05 lr_backbone=3.600282e-05 elapsed=31:26:20 avg_step=3.71s recent_step=3.07s eta=20:01:58 +2026-09-15 21:13:34 | INFO | train rank=0 | step=30550/50000 progress=61.1% loss=0.076950 lr=3.597107e-05 lr_backbone=3.597107e-05 elapsed=31:27:07 avg_step=3.71s recent_step=4.65s eta=20:01:27 +2026-09-15 21:14:07 | INFO | train rank=0 | step=30560/50000 progress=61.1% loss=0.091771 lr=3.593934e-05 lr_backbone=3.593934e-05 elapsed=31:27:40 avg_step=3.71s recent_step=3.28s eta=20:00:47 +2026-09-15 21:14:40 | INFO | train rank=0 | step=30570/50000 progress=61.1% loss=0.082224 lr=3.590760e-05 lr_backbone=3.590760e-05 elapsed=31:28:12 avg_step=3.71s recent_step=3.30s eta=20:00:07 +2026-09-15 21:15:08 | INFO | train rank=0 | step=30580/50000 progress=61.2% loss=0.103392 lr=3.587588e-05 lr_backbone=3.587588e-05 elapsed=31:28:41 avg_step=3.71s recent_step=2.82s eta=19:59:25 +2026-09-15 21:16:02 | INFO | train rank=0 | step=30590/50000 progress=61.2% loss=0.105521 lr=3.584416e-05 lr_backbone=3.584416e-05 elapsed=31:29:35 avg_step=3.71s recent_step=5.43s eta=19:58:59 +2026-09-15 21:16:36 | INFO | train rank=0 | step=30600/50000 progress=61.2% loss=0.074779 lr=3.581244e-05 lr_backbone=3.581244e-05 elapsed=31:30:09 avg_step=3.71s recent_step=3.42s eta=19:58:20 +2026-09-15 21:17:10 | INFO | train rank=0 | step=30610/50000 progress=61.2% loss=0.066822 lr=3.578074e-05 lr_backbone=3.578074e-05 elapsed=31:30:43 avg_step=3.71s recent_step=3.36s eta=19:57:40 +2026-09-15 21:17:49 | INFO | train rank=0 | step=30620/50000 progress=61.2% loss=0.100600 lr=3.574904e-05 lr_backbone=3.574904e-05 elapsed=31:31:22 avg_step=3.71s recent_step=3.95s eta=19:57:05 +2026-09-15 21:18:47 | INFO | train rank=0 | step=30630/50000 progress=61.3% loss=0.075317 lr=3.571734e-05 lr_backbone=3.571734e-05 elapsed=31:32:19 avg_step=3.71s recent_step=5.71s eta=19:56:41 +2026-09-15 21:19:18 | INFO | train rank=0 | step=30640/50000 progress=61.3% loss=0.108889 lr=3.568565e-05 lr_backbone=3.568565e-05 elapsed=31:32:51 avg_step=3.71s recent_step=3.17s eta=19:56:00 +2026-09-15 21:19:49 | INFO | train rank=0 | step=30650/50000 progress=61.3% loss=0.121156 lr=3.565397e-05 lr_backbone=3.565397e-05 elapsed=31:33:22 avg_step=3.71s recent_step=3.05s eta=19:55:19 +2026-09-15 21:20:27 | INFO | train rank=0 | step=30660/50000 progress=61.3% loss=0.082376 lr=3.562230e-05 lr_backbone=3.562230e-05 elapsed=31:33:59 avg_step=3.71s recent_step=3.78s eta=19:54:42 +2026-09-15 21:21:03 | INFO | train rank=0 | step=30670/50000 progress=61.3% loss=0.106602 lr=3.559063e-05 lr_backbone=3.559063e-05 elapsed=31:34:36 avg_step=3.71s recent_step=3.66s eta=19:54:05 +2026-09-15 21:21:45 | INFO | train rank=0 | step=30680/50000 progress=61.4% loss=0.075374 lr=3.555896e-05 lr_backbone=3.555896e-05 elapsed=31:35:18 avg_step=3.71s recent_step=4.17s eta=19:53:31 +2026-09-15 21:22:14 | INFO | train rank=0 | step=30690/50000 progress=61.4% loss=0.118004 lr=3.552731e-05 lr_backbone=3.552731e-05 elapsed=31:35:46 avg_step=3.71s recent_step=2.89s eta=19:52:49 +2026-09-15 21:22:49 | INFO | train rank=0 | step=30700/50000 progress=61.4% loss=0.109401 lr=3.549566e-05 lr_backbone=3.549566e-05 elapsed=31:36:22 avg_step=3.71s recent_step=3.53s eta=19:52:10 +2026-09-15 21:23:27 | INFO | train rank=0 | step=30710/50000 progress=61.4% loss=0.104952 lr=3.546401e-05 lr_backbone=3.546401e-05 elapsed=31:36:59 avg_step=3.71s recent_step=3.76s eta=19:51:34 +2026-09-15 21:24:02 | INFO | train rank=0 | step=30720/50000 progress=61.4% loss=0.100661 lr=3.543237e-05 lr_backbone=3.543237e-05 elapsed=31:37:35 avg_step=3.71s recent_step=3.53s eta=19:50:55 +2026-09-15 21:24:36 | INFO | train rank=0 | step=30730/50000 progress=61.5% loss=0.096227 lr=3.540074e-05 lr_backbone=3.540074e-05 elapsed=31:38:08 avg_step=3.71s recent_step=3.38s eta=19:50:16 +2026-09-15 21:25:12 | INFO | train rank=0 | step=30740/50000 progress=61.5% loss=0.085696 lr=3.536912e-05 lr_backbone=3.536912e-05 elapsed=31:38:45 avg_step=3.71s recent_step=3.63s eta=19:49:39 +2026-09-15 21:25:59 | INFO | train rank=0 | step=30750/50000 progress=61.5% loss=0.107484 lr=3.533750e-05 lr_backbone=3.533750e-05 elapsed=31:39:32 avg_step=3.71s recent_step=4.73s eta=19:49:08 +2026-09-15 21:26:38 | INFO | train rank=0 | step=30760/50000 progress=61.5% loss=0.084805 lr=3.530589e-05 lr_backbone=3.530589e-05 elapsed=31:40:11 avg_step=3.71s recent_step=3.86s eta=19:48:32 +2026-09-15 21:27:06 | INFO | train rank=0 | step=30770/50000 progress=61.5% loss=0.126917 lr=3.527428e-05 lr_backbone=3.527428e-05 elapsed=31:40:39 avg_step=3.71s recent_step=2.83s eta=19:47:50 +2026-09-15 21:27:32 | INFO | train rank=0 | step=30780/50000 progress=61.6% loss=0.102144 lr=3.524268e-05 lr_backbone=3.524268e-05 elapsed=31:41:05 avg_step=3.71s recent_step=2.61s eta=19:47:06 +2026-09-15 21:28:27 | INFO | train rank=0 | step=30790/50000 progress=61.6% loss=0.062158 lr=3.521109e-05 lr_backbone=3.521109e-05 elapsed=31:42:00 avg_step=3.71s recent_step=5.52s eta=19:46:40 +2026-09-15 21:29:00 | INFO | train rank=0 | step=30800/50000 progress=61.6% loss=0.117654 lr=3.517950e-05 lr_backbone=3.517950e-05 elapsed=31:42:32 avg_step=3.71s recent_step=3.21s eta=19:46:00 +2026-09-15 21:29:35 | INFO | train rank=0 | step=30810/50000 progress=61.6% loss=0.132180 lr=3.514792e-05 lr_backbone=3.514792e-05 elapsed=31:43:08 avg_step=3.71s recent_step=3.56s eta=19:45:22 +2026-09-15 21:30:11 | INFO | train rank=0 | step=30820/50000 progress=61.6% loss=0.117656 lr=3.511635e-05 lr_backbone=3.511635e-05 elapsed=31:43:44 avg_step=3.71s recent_step=3.56s eta=19:44:44 +2026-09-15 21:30:53 | INFO | train rank=0 | step=30830/50000 progress=61.7% loss=0.115031 lr=3.508478e-05 lr_backbone=3.508478e-05 elapsed=31:44:26 avg_step=3.71s recent_step=4.20s eta=19:44:10 +2026-09-15 21:31:32 | INFO | train rank=0 | step=30840/50000 progress=61.7% loss=0.075667 lr=3.505322e-05 lr_backbone=3.505322e-05 elapsed=31:45:04 avg_step=3.71s recent_step=3.87s eta=19:43:34 +2026-09-15 21:32:01 | INFO | train rank=0 | step=30850/50000 progress=61.7% loss=0.102114 lr=3.502167e-05 lr_backbone=3.502167e-05 elapsed=31:45:34 avg_step=3.71s recent_step=2.95s eta=19:42:52 +2026-09-15 21:32:44 | INFO | train rank=0 | step=30860/50000 progress=61.7% loss=0.096783 lr=3.499012e-05 lr_backbone=3.499012e-05 elapsed=31:46:16 avg_step=3.71s recent_step=4.25s eta=19:42:18 +2026-09-15 21:33:37 | INFO | train rank=0 | step=30870/50000 progress=61.7% loss=0.090039 lr=3.495858e-05 lr_backbone=3.495858e-05 elapsed=31:47:10 avg_step=3.71s recent_step=5.34s eta=19:41:51 +2026-09-15 21:34:09 | INFO | train rank=0 | step=30880/50000 progress=61.8% loss=0.099905 lr=3.492704e-05 lr_backbone=3.492704e-05 elapsed=31:47:42 avg_step=3.71s recent_step=3.22s eta=19:41:11 +2026-09-15 21:34:41 | INFO | train rank=0 | step=30890/50000 progress=61.8% loss=0.107300 lr=3.489552e-05 lr_backbone=3.489552e-05 elapsed=31:48:13 avg_step=3.71s recent_step=3.14s eta=19:40:31 +2026-09-15 21:35:12 | INFO | train rank=0 | step=30900/50000 progress=61.8% loss=0.077995 lr=3.486399e-05 lr_backbone=3.486399e-05 elapsed=31:48:44 avg_step=3.71s recent_step=3.10s eta=19:39:50 +2026-09-15 21:36:03 | INFO | train rank=0 | step=30910/50000 progress=61.8% loss=0.080531 lr=3.483248e-05 lr_backbone=3.483248e-05 elapsed=31:49:36 avg_step=3.71s recent_step=5.13s eta=19:39:22 +2026-09-15 21:36:41 | INFO | train rank=0 | step=30920/50000 progress=61.8% loss=0.095642 lr=3.480097e-05 lr_backbone=3.480097e-05 elapsed=31:50:14 avg_step=3.71s recent_step=3.82s eta=19:38:45 +2026-09-15 21:37:16 | INFO | train rank=0 | step=30930/50000 progress=61.9% loss=0.088564 lr=3.476947e-05 lr_backbone=3.476947e-05 elapsed=31:50:49 avg_step=3.71s recent_step=3.51s eta=19:38:07 +2026-09-15 21:37:47 | INFO | train rank=0 | step=30940/50000 progress=61.9% loss=0.092437 lr=3.473798e-05 lr_backbone=3.473798e-05 elapsed=31:51:20 avg_step=3.71s recent_step=3.13s eta=19:37:26 +2026-09-15 21:38:38 | INFO | train rank=0 | step=30950/50000 progress=61.9% loss=0.068092 lr=3.470649e-05 lr_backbone=3.470649e-05 elapsed=31:52:11 avg_step=3.71s recent_step=5.03s eta=19:36:58 +2026-09-15 21:39:09 | INFO | train rank=0 | step=30960/50000 progress=61.9% loss=0.094855 lr=3.467501e-05 lr_backbone=3.467501e-05 elapsed=31:52:42 avg_step=3.71s recent_step=3.13s eta=19:36:17 +2026-09-15 21:39:41 | INFO | train rank=0 | step=30970/50000 progress=61.9% loss=0.120018 lr=3.464353e-05 lr_backbone=3.464353e-05 elapsed=31:53:14 avg_step=3.71s recent_step=3.24s eta=19:35:37 +2026-09-15 21:40:12 | INFO | train rank=0 | step=30980/50000 progress=62.0% loss=0.111405 lr=3.461207e-05 lr_backbone=3.461207e-05 elapsed=31:53:45 avg_step=3.71s recent_step=3.02s eta=19:34:56 +2026-09-15 21:40:49 | INFO | train rank=0 | step=30990/50000 progress=62.0% loss=0.098928 lr=3.458060e-05 lr_backbone=3.458060e-05 elapsed=31:54:21 avg_step=3.71s recent_step=3.68s eta=19:34:18 +2026-09-15 21:41:22 | INFO | train rank=0 | step=31000/50000 progress=62.0% loss=0.093856 lr=3.454915e-05 lr_backbone=3.454915e-05 elapsed=31:54:55 avg_step=3.71s recent_step=3.39s eta=19:33:39 +2026-09-15 21:41:57 | INFO | train rank=0 | step=31010/50000 progress=62.0% loss=0.085690 lr=3.451770e-05 lr_backbone=3.451770e-05 elapsed=31:55:29 avg_step=3.71s recent_step=3.42s eta=19:33:01 +2026-09-15 21:42:33 | INFO | train rank=0 | step=31020/50000 progress=62.0% loss=0.098276 lr=3.448626e-05 lr_backbone=3.448626e-05 elapsed=31:56:05 avg_step=3.71s recent_step=3.59s eta=19:32:23 +2026-09-15 21:43:25 | INFO | train rank=0 | step=31030/50000 progress=62.1% loss=0.098466 lr=3.445483e-05 lr_backbone=3.445483e-05 elapsed=31:56:58 avg_step=3.71s recent_step=5.24s eta=19:31:55 +2026-09-15 21:44:00 | INFO | train rank=0 | step=31040/50000 progress=62.1% loss=0.067613 lr=3.442340e-05 lr_backbone=3.442340e-05 elapsed=31:57:33 avg_step=3.71s recent_step=3.48s eta=19:31:17 +2026-09-15 21:44:35 | INFO | train rank=0 | step=31050/50000 progress=62.1% loss=0.102878 lr=3.439198e-05 lr_backbone=3.439198e-05 elapsed=31:58:08 avg_step=3.71s recent_step=3.51s eta=19:30:38 +2026-09-15 21:45:09 | INFO | train rank=0 | step=31060/50000 progress=62.1% loss=0.070103 lr=3.436057e-05 lr_backbone=3.436057e-05 elapsed=31:58:42 avg_step=3.71s recent_step=3.41s eta=19:30:00 +2026-09-15 21:45:52 | INFO | train rank=0 | step=31070/50000 progress=62.1% loss=0.091626 lr=3.432916e-05 lr_backbone=3.432916e-05 elapsed=31:59:25 avg_step=3.71s recent_step=4.29s eta=19:29:26 +2026-09-15 21:46:29 | INFO | train rank=0 | step=31080/50000 progress=62.2% loss=0.093825 lr=3.429776e-05 lr_backbone=3.429776e-05 elapsed=32:00:02 avg_step=3.71s recent_step=3.73s eta=19:28:49 +2026-09-15 21:47:03 | INFO | train rank=0 | step=31090/50000 progress=62.2% loss=0.096953 lr=3.426637e-05 lr_backbone=3.426637e-05 elapsed=32:00:36 avg_step=3.71s recent_step=3.37s eta=19:28:10 +2026-09-15 21:47:40 | INFO | train rank=0 | step=31100/50000 progress=62.2% loss=0.122099 lr=3.423498e-05 lr_backbone=3.423498e-05 elapsed=32:01:13 avg_step=3.71s recent_step=3.75s eta=19:27:33 +2026-09-15 21:48:09 | INFO | train rank=0 | step=31110/50000 progress=62.2% loss=0.092936 lr=3.420360e-05 lr_backbone=3.420360e-05 elapsed=32:01:42 avg_step=3.71s recent_step=2.88s eta=19:26:51 +2026-09-15 21:48:53 | INFO | train rank=0 | step=31120/50000 progress=62.2% loss=0.070839 lr=3.417223e-05 lr_backbone=3.417223e-05 elapsed=32:02:26 avg_step=3.71s recent_step=4.40s eta=19:26:18 +2026-09-15 21:49:20 | INFO | train rank=0 | step=31130/50000 progress=62.3% loss=0.092242 lr=3.414087e-05 lr_backbone=3.414087e-05 elapsed=32:02:52 avg_step=3.71s recent_step=2.66s eta=19:25:35 +2026-09-15 21:49:55 | INFO | train rank=0 | step=31140/50000 progress=62.3% loss=0.073193 lr=3.410951e-05 lr_backbone=3.410951e-05 elapsed=32:03:28 avg_step=3.71s recent_step=3.55s eta=19:24:57 +2026-09-15 21:50:39 | INFO | train rank=0 | step=31150/50000 progress=62.3% loss=0.100222 lr=3.407816e-05 lr_backbone=3.407816e-05 elapsed=32:04:11 avg_step=3.71s recent_step=4.35s eta=19:24:24 +2026-09-15 21:51:12 | INFO | train rank=0 | step=31160/50000 progress=62.3% loss=0.084687 lr=3.404681e-05 lr_backbone=3.404681e-05 elapsed=32:04:44 avg_step=3.71s recent_step=3.30s eta=19:23:44 +2026-09-15 21:51:51 | INFO | train rank=0 | step=31170/50000 progress=62.3% loss=0.115748 lr=3.401547e-05 lr_backbone=3.401547e-05 elapsed=32:05:24 avg_step=3.71s recent_step=3.98s eta=19:23:09 +2026-09-15 21:52:26 | INFO | train rank=0 | step=31180/50000 progress=62.4% loss=0.090417 lr=3.398414e-05 lr_backbone=3.398414e-05 elapsed=32:05:59 avg_step=3.71s recent_step=3.48s eta=19:22:30 +2026-09-15 21:53:18 | INFO | train rank=0 | step=31190/50000 progress=62.4% loss=0.065151 lr=3.395282e-05 lr_backbone=3.395282e-05 elapsed=32:06:51 avg_step=3.71s recent_step=5.19s eta=19:22:02 +2026-09-15 21:53:46 | INFO | train rank=0 | step=31200/50000 progress=62.4% loss=0.097481 lr=3.392150e-05 lr_backbone=3.392150e-05 elapsed=32:07:19 avg_step=3.71s recent_step=2.81s eta=19:21:20 +2026-09-15 21:54:22 | INFO | train rank=0 | step=31210/50000 progress=62.4% loss=0.075951 lr=3.389019e-05 lr_backbone=3.389019e-05 elapsed=32:07:55 avg_step=3.71s recent_step=3.57s eta=19:20:42 +2026-09-15 21:54:49 | INFO | train rank=0 | step=31220/50000 progress=62.4% loss=0.076071 lr=3.385889e-05 lr_backbone=3.385889e-05 elapsed=32:08:22 avg_step=3.71s recent_step=2.68s eta=19:19:59 +2026-09-15 21:55:33 | INFO | train rank=0 | step=31230/50000 progress=62.5% loss=0.072749 lr=3.382760e-05 lr_backbone=3.382760e-05 elapsed=32:09:06 avg_step=3.71s recent_step=4.42s eta=19:19:26 +2026-09-15 21:56:18 | INFO | train rank=0 | step=31240/50000 progress=62.5% loss=0.088290 lr=3.379631e-05 lr_backbone=3.379631e-05 elapsed=32:09:51 avg_step=3.71s recent_step=4.54s eta=19:18:54 +2026-09-15 21:56:52 | INFO | train rank=0 | step=31250/50000 progress=62.5% loss=0.068718 lr=3.376503e-05 lr_backbone=3.376503e-05 elapsed=32:10:25 avg_step=3.71s recent_step=3.35s eta=19:18:15 +2026-09-15 21:57:29 | INFO | train rank=0 | step=31260/50000 progress=62.5% loss=0.073594 lr=3.373375e-05 lr_backbone=3.373375e-05 elapsed=32:11:02 avg_step=3.71s recent_step=3.70s eta=19:17:37 +2026-09-15 21:58:09 | INFO | train rank=0 | step=31270/50000 progress=62.5% loss=0.100391 lr=3.370249e-05 lr_backbone=3.370249e-05 elapsed=32:11:41 avg_step=3.71s recent_step=3.98s eta=19:17:02 +2026-09-15 21:58:56 | INFO | train rank=0 | step=31280/50000 progress=62.6% loss=0.089232 lr=3.367123e-05 lr_backbone=3.367123e-05 elapsed=32:12:28 avg_step=3.71s recent_step=4.69s eta=19:16:31 +2026-09-15 21:59:29 | INFO | train rank=0 | step=31290/50000 progress=62.6% loss=0.102113 lr=3.363997e-05 lr_backbone=3.363997e-05 elapsed=32:13:02 avg_step=3.71s recent_step=3.36s eta=19:15:52 +2026-09-15 22:00:00 | INFO | train rank=0 | step=31300/50000 progress=62.6% loss=0.116511 lr=3.360873e-05 lr_backbone=3.360873e-05 elapsed=32:13:33 avg_step=3.71s recent_step=3.07s eta=19:15:11 +2026-09-15 22:00:41 | INFO | train rank=0 | step=31310/50000 progress=62.6% loss=0.127438 lr=3.357749e-05 lr_backbone=3.357749e-05 elapsed=32:14:14 avg_step=3.71s recent_step=4.12s eta=19:14:36 +2026-09-15 22:01:34 | INFO | train rank=0 | step=31320/50000 progress=62.6% loss=0.076856 lr=3.354626e-05 lr_backbone=3.354626e-05 elapsed=32:15:07 avg_step=3.71s recent_step=5.26s eta=19:14:09 +2026-09-15 22:02:09 | INFO | train rank=0 | step=31330/50000 progress=62.7% loss=0.089643 lr=3.351503e-05 lr_backbone=3.351503e-05 elapsed=32:15:42 avg_step=3.71s recent_step=3.52s eta=19:13:30 +2026-09-15 22:02:42 | INFO | train rank=0 | step=31340/50000 progress=62.7% loss=0.110681 lr=3.348382e-05 lr_backbone=3.348382e-05 elapsed=32:16:15 avg_step=3.71s recent_step=3.32s eta=19:12:51 +2026-09-15 22:03:12 | INFO | train rank=0 | step=31350/50000 progress=62.7% loss=0.097161 lr=3.345261e-05 lr_backbone=3.345261e-05 elapsed=32:16:45 avg_step=3.71s recent_step=3.02s eta=19:12:10 +2026-09-15 22:04:01 | INFO | train rank=0 | step=31360/50000 progress=62.7% loss=0.102549 lr=3.342141e-05 lr_backbone=3.342141e-05 elapsed=32:17:33 avg_step=3.71s recent_step=4.82s eta=19:11:39 +2026-09-15 22:04:33 | INFO | train rank=0 | step=31370/50000 progress=62.7% loss=0.090346 lr=3.339021e-05 lr_backbone=3.339021e-05 elapsed=32:18:06 avg_step=3.71s recent_step=3.25s eta=19:11:00 +2026-09-15 22:05:10 | INFO | train rank=0 | step=31380/50000 progress=62.8% loss=0.091811 lr=3.335902e-05 lr_backbone=3.335902e-05 elapsed=32:18:43 avg_step=3.71s recent_step=3.69s eta=19:10:22 +2026-09-15 22:05:47 | INFO | train rank=0 | step=31390/50000 progress=62.8% loss=0.074217 lr=3.332784e-05 lr_backbone=3.332784e-05 elapsed=32:19:20 avg_step=3.71s recent_step=3.71s eta=19:09:45 +2026-09-15 22:06:43 | INFO | train rank=0 | step=31400/50000 progress=62.8% loss=0.108340 lr=3.329667e-05 lr_backbone=3.329667e-05 elapsed=32:20:16 avg_step=3.71s recent_step=5.58s eta=19:09:19 +2026-09-15 22:07:10 | INFO | train rank=0 | step=31410/50000 progress=62.8% loss=0.076462 lr=3.326550e-05 lr_backbone=3.326550e-05 elapsed=32:20:43 avg_step=3.71s recent_step=2.73s eta=19:08:37 +2026-09-15 22:07:45 | INFO | train rank=0 | step=31420/50000 progress=62.8% loss=0.108556 lr=3.323434e-05 lr_backbone=3.323434e-05 elapsed=32:21:18 avg_step=3.71s recent_step=3.45s eta=19:07:58 +2026-09-15 22:08:24 | INFO | train rank=0 | step=31430/50000 progress=62.9% loss=0.085313 lr=3.320319e-05 lr_backbone=3.320319e-05 elapsed=32:21:57 avg_step=3.71s recent_step=3.92s eta=19:07:22 +2026-09-15 22:09:10 | INFO | train rank=0 | step=31440/50000 progress=62.9% loss=0.079754 lr=3.317205e-05 lr_backbone=3.317205e-05 elapsed=32:22:43 avg_step=3.71s recent_step=4.65s eta=19:06:51 +2026-09-15 22:09:42 | INFO | train rank=0 | step=31450/50000 progress=62.9% loss=0.095223 lr=3.314091e-05 lr_backbone=3.314091e-05 elapsed=32:23:15 avg_step=3.71s recent_step=3.18s eta=19:06:10 +2026-09-15 22:10:14 | INFO | train rank=0 | step=31460/50000 progress=62.9% loss=0.087984 lr=3.310978e-05 lr_backbone=3.310978e-05 elapsed=32:23:47 avg_step=3.71s recent_step=3.20s eta=19:05:30 +2026-09-15 22:10:47 | INFO | train rank=0 | step=31470/50000 progress=62.9% loss=0.089191 lr=3.307866e-05 lr_backbone=3.307866e-05 elapsed=32:24:19 avg_step=3.71s recent_step=3.23s eta=19:04:51 +2026-09-15 22:11:39 | INFO | train rank=0 | step=31480/50000 progress=63.0% loss=0.090580 lr=3.304755e-05 lr_backbone=3.304755e-05 elapsed=32:25:11 avg_step=3.71s recent_step=5.20s eta=19:04:22 +2026-09-15 22:12:14 | INFO | train rank=0 | step=31490/50000 progress=63.0% loss=0.110104 lr=3.301644e-05 lr_backbone=3.301644e-05 elapsed=32:25:47 avg_step=3.71s recent_step=3.58s eta=19:03:44 +2026-09-15 22:12:43 | INFO | train rank=0 | step=31500/50000 progress=63.0% loss=0.092556 lr=3.298534e-05 lr_backbone=3.298534e-05 elapsed=32:26:16 avg_step=3.71s recent_step=2.86s eta=19:03:02 +2026-09-15 22:13:16 | INFO | train rank=0 | step=31510/50000 progress=63.0% loss=0.100057 lr=3.295425e-05 lr_backbone=3.295425e-05 elapsed=32:26:49 avg_step=3.71s recent_step=3.31s eta=19:02:23 +2026-09-15 22:14:07 | INFO | train rank=0 | step=31520/50000 progress=63.0% loss=0.072558 lr=3.292316e-05 lr_backbone=3.292316e-05 elapsed=32:27:40 avg_step=3.71s recent_step=5.06s eta=19:01:54 +2026-09-15 22:14:39 | INFO | train rank=0 | step=31530/50000 progress=63.1% loss=0.128894 lr=3.289209e-05 lr_backbone=3.289209e-05 elapsed=32:28:11 avg_step=3.71s recent_step=3.18s eta=19:01:14 +2026-09-15 22:15:17 | INFO | train rank=0 | step=31540/50000 progress=63.1% loss=0.143824 lr=3.286102e-05 lr_backbone=3.286102e-05 elapsed=32:28:50 avg_step=3.71s recent_step=3.89s eta=19:00:38 +2026-09-15 22:15:52 | INFO | train rank=0 | step=31550/50000 progress=63.1% loss=0.085087 lr=3.282996e-05 lr_backbone=3.282996e-05 elapsed=32:29:25 avg_step=3.71s recent_step=3.46s eta=18:59:59 +2026-09-15 22:16:44 | INFO | train rank=0 | step=31560/50000 progress=63.1% loss=0.077595 lr=3.279890e-05 lr_backbone=3.279890e-05 elapsed=32:30:17 avg_step=3.71s recent_step=5.19s eta=18:59:31 +2026-09-15 22:17:22 | INFO | train rank=0 | step=31570/50000 progress=63.1% loss=0.067545 lr=3.276785e-05 lr_backbone=3.276785e-05 elapsed=32:30:55 avg_step=3.71s recent_step=3.82s eta=18:58:54 +2026-09-15 22:17:43 | INFO | train rank=0 | step=31580/50000 progress=63.2% loss=0.102391 lr=3.273681e-05 lr_backbone=3.273681e-05 elapsed=32:31:16 avg_step=3.71s recent_step=2.11s eta=18:58:08 +2026-09-15 22:18:23 | INFO | train rank=0 | step=31590/50000 progress=63.2% loss=0.112188 lr=3.270578e-05 lr_backbone=3.270578e-05 elapsed=32:31:56 avg_step=3.71s recent_step=3.99s eta=18:57:33 +2026-09-15 22:19:16 | INFO | train rank=0 | step=31600/50000 progress=63.2% loss=0.079834 lr=3.267476e-05 lr_backbone=3.267476e-05 elapsed=32:32:48 avg_step=3.71s recent_step=5.25s eta=18:57:04 +2026-09-15 22:19:47 | INFO | train rank=0 | step=31610/50000 progress=63.2% loss=0.118388 lr=3.264374e-05 lr_backbone=3.264374e-05 elapsed=32:33:20 avg_step=3.71s recent_step=3.17s eta=18:56:24 +2026-09-15 22:20:23 | INFO | train rank=0 | step=31620/50000 progress=63.2% loss=0.093784 lr=3.261273e-05 lr_backbone=3.261273e-05 elapsed=32:33:56 avg_step=3.71s recent_step=3.58s eta=18:55:46 +2026-09-15 22:20:55 | INFO | train rank=0 | step=31630/50000 progress=63.3% loss=0.085932 lr=3.258173e-05 lr_backbone=3.258173e-05 elapsed=32:34:28 avg_step=3.71s recent_step=3.23s eta=18:55:07 +2026-09-15 22:21:48 | INFO | train rank=0 | step=31640/50000 progress=63.3% loss=0.157611 lr=3.255074e-05 lr_backbone=3.255074e-05 elapsed=32:35:21 avg_step=3.71s recent_step=5.26s eta=18:54:39 +2026-09-15 22:22:22 | INFO | train rank=0 | step=31650/50000 progress=63.3% loss=0.090697 lr=3.251975e-05 lr_backbone=3.251975e-05 elapsed=32:35:54 avg_step=3.71s recent_step=3.36s eta=18:53:59 +2026-09-15 22:22:58 | INFO | train rank=0 | step=31660/50000 progress=63.3% loss=0.112803 lr=3.248877e-05 lr_backbone=3.248877e-05 elapsed=32:36:31 avg_step=3.71s recent_step=3.61s eta=18:53:22 +2026-09-15 22:23:26 | INFO | train rank=0 | step=31670/50000 progress=63.3% loss=0.096354 lr=3.245780e-05 lr_backbone=3.245780e-05 elapsed=32:36:58 avg_step=3.71s recent_step=2.78s eta=18:52:39 +2026-09-15 22:23:56 | INFO | train rank=0 | step=31680/50000 progress=63.4% loss=0.105970 lr=3.242684e-05 lr_backbone=3.242684e-05 elapsed=32:37:29 avg_step=3.71s recent_step=3.07s eta=18:51:59 +2026-09-15 22:24:28 | INFO | train rank=0 | step=31690/50000 progress=63.4% loss=0.098304 lr=3.239588e-05 lr_backbone=3.239588e-05 elapsed=32:38:01 avg_step=3.71s recent_step=3.16s eta=18:51:18 +2026-09-15 22:25:00 | INFO | train rank=0 | step=31700/50000 progress=63.4% loss=0.074824 lr=3.236493e-05 lr_backbone=3.236493e-05 elapsed=32:38:33 avg_step=3.71s recent_step=3.24s eta=18:50:39 +2026-09-15 22:25:31 | INFO | train rank=0 | step=31710/50000 progress=63.4% loss=0.120188 lr=3.233399e-05 lr_backbone=3.233399e-05 elapsed=32:39:04 avg_step=3.71s recent_step=3.07s eta=18:49:58 +2026-09-15 22:26:15 | INFO | train rank=0 | step=31720/50000 progress=63.4% loss=0.126348 lr=3.230306e-05 lr_backbone=3.230306e-05 elapsed=32:39:48 avg_step=3.71s recent_step=4.43s eta=18:49:25 +2026-09-15 22:26:53 | INFO | train rank=0 | step=31730/50000 progress=63.5% loss=0.100767 lr=3.227213e-05 lr_backbone=3.227213e-05 elapsed=32:40:26 avg_step=3.71s recent_step=3.81s eta=18:48:48 +2026-09-15 22:27:23 | INFO | train rank=0 | step=31740/50000 progress=63.5% loss=0.085392 lr=3.224122e-05 lr_backbone=3.224122e-05 elapsed=32:40:56 avg_step=3.71s recent_step=2.99s eta=18:48:07 +2026-09-15 22:27:52 | INFO | train rank=0 | step=31750/50000 progress=63.5% loss=0.080735 lr=3.221031e-05 lr_backbone=3.221031e-05 elapsed=32:41:25 avg_step=3.71s recent_step=2.90s eta=18:47:25 +2026-09-15 22:28:30 | INFO | train rank=0 | step=31760/50000 progress=63.5% loss=0.094490 lr=3.217941e-05 lr_backbone=3.217941e-05 elapsed=32:42:03 avg_step=3.71s recent_step=3.77s eta=18:46:49 +2026-09-15 22:29:04 | INFO | train rank=0 | step=31770/50000 progress=63.5% loss=0.110666 lr=3.214851e-05 lr_backbone=3.214851e-05 elapsed=32:42:37 avg_step=3.71s recent_step=3.42s eta=18:46:10 +2026-09-15 22:29:31 | INFO | train rank=0 | step=31780/50000 progress=63.6% loss=0.087097 lr=3.211763e-05 lr_backbone=3.211763e-05 elapsed=32:43:04 avg_step=3.71s recent_step=2.69s eta=18:45:27 +2026-09-15 22:30:03 | INFO | train rank=0 | step=31790/50000 progress=63.6% loss=0.112165 lr=3.208675e-05 lr_backbone=3.208675e-05 elapsed=32:43:36 avg_step=3.71s recent_step=3.25s eta=18:44:47 +2026-09-15 22:30:55 | INFO | train rank=0 | step=31800/50000 progress=63.6% loss=0.107610 lr=3.205588e-05 lr_backbone=3.205588e-05 elapsed=32:44:27 avg_step=3.71s recent_step=5.12s eta=18:44:19 +2026-09-15 22:31:27 | INFO | train rank=0 | step=31810/50000 progress=63.6% loss=0.092933 lr=3.202502e-05 lr_backbone=3.202502e-05 elapsed=32:44:59 avg_step=3.71s recent_step=3.20s eta=18:43:39 +2026-09-15 22:32:09 | INFO | train rank=0 | step=31820/50000 progress=63.6% loss=0.083729 lr=3.199416e-05 lr_backbone=3.199416e-05 elapsed=32:45:42 avg_step=3.71s recent_step=4.23s eta=18:43:04 +2026-09-15 22:32:46 | INFO | train rank=0 | step=31830/50000 progress=63.7% loss=0.076656 lr=3.196331e-05 lr_backbone=3.196331e-05 elapsed=32:46:19 avg_step=3.71s recent_step=3.68s eta=18:42:27 +2026-09-15 22:33:41 | INFO | train rank=0 | step=31840/50000 progress=63.7% loss=0.083185 lr=3.193248e-05 lr_backbone=3.193248e-05 elapsed=32:47:14 avg_step=3.71s recent_step=5.58s eta=18:42:01 +2026-09-15 22:34:07 | INFO | train rank=0 | step=31850/50000 progress=63.7% loss=0.077538 lr=3.190164e-05 lr_backbone=3.190164e-05 elapsed=32:47:40 avg_step=3.71s recent_step=2.58s eta=18:41:17 +2026-09-15 22:34:45 | INFO | train rank=0 | step=31860/50000 progress=63.7% loss=0.121600 lr=3.187082e-05 lr_backbone=3.187082e-05 elapsed=32:48:18 avg_step=3.71s recent_step=3.75s eta=18:40:41 +2026-09-15 22:35:21 | INFO | train rank=0 | step=31870/50000 progress=63.7% loss=0.114161 lr=3.184001e-05 lr_backbone=3.184001e-05 elapsed=32:48:54 avg_step=3.71s recent_step=3.66s eta=18:40:03 +2026-09-15 22:36:10 | INFO | train rank=0 | step=31880/50000 progress=63.8% loss=0.089858 lr=3.180920e-05 lr_backbone=3.180920e-05 elapsed=32:49:42 avg_step=3.71s recent_step=4.82s eta=18:39:32 +2026-09-15 22:36:45 | INFO | train rank=0 | step=31890/50000 progress=63.8% loss=0.113691 lr=3.177840e-05 lr_backbone=3.177840e-05 elapsed=32:50:18 avg_step=3.71s recent_step=3.55s eta=18:38:54 +2026-09-15 22:37:16 | INFO | train rank=0 | step=31900/50000 progress=63.8% loss=0.112841 lr=3.174761e-05 lr_backbone=3.174761e-05 elapsed=32:50:49 avg_step=3.71s recent_step=3.09s eta=18:38:14 +2026-09-15 22:37:47 | INFO | train rank=0 | step=31910/50000 progress=63.8% loss=0.093678 lr=3.171683e-05 lr_backbone=3.171683e-05 elapsed=32:51:20 avg_step=3.71s recent_step=3.13s eta=18:37:34 +2026-09-15 22:38:34 | INFO | train rank=0 | step=31920/50000 progress=63.8% loss=0.098001 lr=3.168605e-05 lr_backbone=3.168605e-05 elapsed=32:52:07 avg_step=3.71s recent_step=4.69s eta=18:37:02 +2026-09-15 22:39:10 | INFO | train rank=0 | step=31930/50000 progress=63.9% loss=0.094926 lr=3.165528e-05 lr_backbone=3.165528e-05 elapsed=32:52:43 avg_step=3.71s recent_step=3.58s eta=18:36:24 +2026-09-15 22:39:43 | INFO | train rank=0 | step=31940/50000 progress=63.9% loss=0.088744 lr=3.162452e-05 lr_backbone=3.162452e-05 elapsed=32:53:16 avg_step=3.71s recent_step=3.30s eta=18:35:45 +2026-09-15 22:40:11 | INFO | train rank=0 | step=31950/50000 progress=63.9% loss=0.088774 lr=3.159377e-05 lr_backbone=3.159377e-05 elapsed=32:53:43 avg_step=3.71s recent_step=2.76s eta=18:35:02 +2026-09-15 22:40:53 | INFO | train rank=0 | step=31960/50000 progress=63.9% loss=0.120855 lr=3.156303e-05 lr_backbone=3.156303e-05 elapsed=32:54:25 avg_step=3.71s recent_step=4.21s eta=18:34:28 +2026-09-15 22:41:26 | INFO | train rank=0 | step=31970/50000 progress=63.9% loss=0.109056 lr=3.153229e-05 lr_backbone=3.153229e-05 elapsed=32:54:59 avg_step=3.71s recent_step=3.35s eta=18:33:49 +2026-09-15 22:41:58 | INFO | train rank=0 | step=31980/50000 progress=64.0% loss=0.132298 lr=3.150157e-05 lr_backbone=3.150157e-05 elapsed=32:55:31 avg_step=3.71s recent_step=3.17s eta=18:33:09 +2026-09-15 22:42:28 | INFO | train rank=0 | step=31990/50000 progress=64.0% loss=0.122486 lr=3.147085e-05 lr_backbone=3.147085e-05 elapsed=32:56:01 avg_step=3.71s recent_step=3.00s eta=18:32:28 +2026-09-15 22:43:09 | INFO | train rank=0 | step=32000/50000 progress=64.0% loss=0.080699 lr=3.144014e-05 lr_backbone=3.144014e-05 elapsed=32:56:42 avg_step=3.71s recent_step=4.16s eta=18:31:54 +2026-09-15 22:43:46 | INFO | train rank=0 | step=32010/50000 progress=64.0% loss=0.091956 lr=3.140943e-05 lr_backbone=3.140943e-05 elapsed=32:57:19 avg_step=3.71s recent_step=3.64s eta=18:31:16 +2026-09-15 22:44:18 | INFO | train rank=0 | step=32020/50000 progress=64.0% loss=0.093638 lr=3.137874e-05 lr_backbone=3.137874e-05 elapsed=32:57:51 avg_step=3.71s recent_step=3.26s eta=18:30:37 +2026-09-15 22:44:53 | INFO | train rank=0 | step=32030/50000 progress=64.1% loss=0.102161 lr=3.134805e-05 lr_backbone=3.134805e-05 elapsed=32:58:26 avg_step=3.71s recent_step=3.45s eta=18:29:58 +2026-09-15 22:45:38 | INFO | train rank=0 | step=32040/50000 progress=64.1% loss=0.078987 lr=3.131738e-05 lr_backbone=3.131738e-05 elapsed=32:59:11 avg_step=3.71s recent_step=4.50s eta=18:29:25 +2026-09-15 22:46:22 | INFO | train rank=0 | step=32050/50000 progress=64.1% loss=0.088330 lr=3.128671e-05 lr_backbone=3.128671e-05 elapsed=32:59:55 avg_step=3.71s recent_step=4.38s eta=18:28:52 +2026-09-15 22:46:56 | INFO | train rank=0 | step=32060/50000 progress=64.1% loss=0.087311 lr=3.125604e-05 lr_backbone=3.125604e-05 elapsed=33:00:29 avg_step=3.71s recent_step=3.41s eta=18:28:13 +2026-09-15 22:47:31 | INFO | train rank=0 | step=32070/50000 progress=64.1% loss=0.095216 lr=3.122539e-05 lr_backbone=3.122539e-05 elapsed=33:01:03 avg_step=3.71s recent_step=3.47s eta=18:27:35 +2026-09-15 22:48:08 | INFO | train rank=0 | step=32080/50000 progress=64.2% loss=0.127824 lr=3.119474e-05 lr_backbone=3.119474e-05 elapsed=33:01:40 avg_step=3.71s recent_step=3.69s eta=18:26:58 +2026-09-15 22:48:59 | INFO | train rank=0 | step=32090/50000 progress=64.2% loss=0.086768 lr=3.116411e-05 lr_backbone=3.116411e-05 elapsed=33:02:32 avg_step=3.71s recent_step=5.16s eta=18:26:29 +2026-09-15 22:49:34 | INFO | train rank=0 | step=32100/50000 progress=64.2% loss=0.101884 lr=3.113348e-05 lr_backbone=3.113348e-05 elapsed=33:03:07 avg_step=3.71s recent_step=3.53s eta=18:25:51 +2026-09-15 22:50:06 | INFO | train rank=0 | step=32110/50000 progress=64.2% loss=0.085795 lr=3.110286e-05 lr_backbone=3.110286e-05 elapsed=33:03:39 avg_step=3.71s recent_step=3.20s eta=18:25:11 +2026-09-15 22:50:38 | INFO | train rank=0 | step=32120/50000 progress=64.2% loss=0.113567 lr=3.107225e-05 lr_backbone=3.107225e-05 elapsed=33:04:11 avg_step=3.71s recent_step=3.13s eta=18:24:31 +2026-09-15 22:51:14 | INFO | train rank=0 | step=32130/50000 progress=64.3% loss=0.102486 lr=3.104164e-05 lr_backbone=3.104164e-05 elapsed=33:04:47 avg_step=3.71s recent_step=3.66s eta=18:23:53 +2026-09-15 22:52:00 | INFO | train rank=0 | step=32140/50000 progress=64.3% loss=0.112979 lr=3.101105e-05 lr_backbone=3.101105e-05 elapsed=33:05:33 avg_step=3.71s recent_step=4.62s eta=18:23:21 +2026-09-15 22:52:35 | INFO | train rank=0 | step=32150/50000 progress=64.3% loss=0.069682 lr=3.098046e-05 lr_backbone=3.098046e-05 elapsed=33:06:08 avg_step=3.71s recent_step=3.43s eta=18:22:43 +2026-09-15 22:53:15 | INFO | train rank=0 | step=32160/50000 progress=64.3% loss=0.092821 lr=3.094988e-05 lr_backbone=3.094988e-05 elapsed=33:06:48 avg_step=3.71s recent_step=4.06s eta=18:22:08 +2026-09-15 22:53:56 | INFO | train rank=0 | step=32170/50000 progress=64.3% loss=0.114071 lr=3.091931e-05 lr_backbone=3.091931e-05 elapsed=33:07:28 avg_step=3.71s recent_step=4.02s eta=18:21:32 +2026-09-15 22:54:32 | INFO | train rank=0 | step=32180/50000 progress=64.4% loss=0.098472 lr=3.088874e-05 lr_backbone=3.088874e-05 elapsed=33:08:05 avg_step=3.71s recent_step=3.61s eta=18:20:55 +2026-09-15 22:55:13 | INFO | train rank=0 | step=32190/50000 progress=64.4% loss=0.085967 lr=3.085819e-05 lr_backbone=3.085819e-05 elapsed=33:08:46 avg_step=3.71s recent_step=4.14s eta=18:20:20 +2026-09-15 22:55:43 | INFO | train rank=0 | step=32200/50000 progress=64.4% loss=0.094928 lr=3.082764e-05 lr_backbone=3.082764e-05 elapsed=33:09:16 avg_step=3.71s recent_step=3.03s eta=18:19:39 +2026-09-15 22:56:18 | INFO | train rank=0 | step=32210/50000 progress=64.4% loss=0.115599 lr=3.079711e-05 lr_backbone=3.079711e-05 elapsed=33:09:50 avg_step=3.71s recent_step=3.42s eta=18:19:01 +2026-09-15 22:57:10 | INFO | train rank=0 | step=32220/50000 progress=64.4% loss=0.081518 lr=3.076658e-05 lr_backbone=3.076658e-05 elapsed=33:10:42 avg_step=3.71s recent_step=5.20s eta=18:18:32 +2026-09-15 22:57:48 | INFO | train rank=0 | step=32230/50000 progress=64.5% loss=0.073886 lr=3.073606e-05 lr_backbone=3.073606e-05 elapsed=33:11:21 avg_step=3.71s recent_step=3.87s eta=18:17:56 +2026-09-15 22:58:16 | INFO | train rank=0 | step=32240/50000 progress=64.5% loss=0.087753 lr=3.070555e-05 lr_backbone=3.070555e-05 elapsed=33:11:49 avg_step=3.71s recent_step=2.78s eta=18:17:14 +2026-09-15 22:58:53 | INFO | train rank=0 | step=32250/50000 progress=64.5% loss=0.088984 lr=3.067504e-05 lr_backbone=3.067504e-05 elapsed=33:12:26 avg_step=3.71s recent_step=3.67s eta=18:16:36 +2026-09-15 22:59:47 | INFO | train rank=0 | step=32260/50000 progress=64.5% loss=0.074672 lr=3.064455e-05 lr_backbone=3.064455e-05 elapsed=33:13:20 avg_step=3.71s recent_step=5.41s eta=18:16:09 +2026-09-15 23:00:18 | INFO | train rank=0 | step=32270/50000 progress=64.5% loss=0.095855 lr=3.061406e-05 lr_backbone=3.061406e-05 elapsed=33:13:51 avg_step=3.71s recent_step=3.12s eta=18:15:28 +2026-09-15 23:01:02 | INFO | train rank=0 | step=32280/50000 progress=64.6% loss=0.076397 lr=3.058358e-05 lr_backbone=3.058358e-05 elapsed=33:14:35 avg_step=3.71s recent_step=4.43s eta=18:14:55 +2026-09-15 23:01:33 | INFO | train rank=0 | step=32290/50000 progress=64.6% loss=0.096653 lr=3.055311e-05 lr_backbone=3.055311e-05 elapsed=33:15:06 avg_step=3.71s recent_step=3.07s eta=18:14:15 +2026-09-15 23:02:21 | INFO | train rank=0 | step=32300/50000 progress=64.6% loss=0.070193 lr=3.052265e-05 lr_backbone=3.052265e-05 elapsed=33:15:54 avg_step=3.71s recent_step=4.79s eta=18:13:43 +2026-09-15 23:02:55 | INFO | train rank=0 | step=32310/50000 progress=64.6% loss=0.068114 lr=3.049220e-05 lr_backbone=3.049220e-05 elapsed=33:16:28 avg_step=3.71s recent_step=3.41s eta=18:13:05 +2026-09-15 23:03:21 | INFO | train rank=0 | step=32320/50000 progress=64.6% loss=0.097216 lr=3.046175e-05 lr_backbone=3.046175e-05 elapsed=33:16:54 avg_step=3.71s recent_step=2.62s eta=18:12:22 +2026-09-15 23:03:50 | INFO | train rank=0 | step=32330/50000 progress=64.7% loss=0.083362 lr=3.043132e-05 lr_backbone=3.043132e-05 elapsed=33:17:23 avg_step=3.71s recent_step=2.91s eta=18:11:40 +2026-09-15 23:04:49 | INFO | train rank=0 | step=32340/50000 progress=64.7% loss=0.088275 lr=3.040089e-05 lr_backbone=3.040089e-05 elapsed=33:18:21 avg_step=3.71s recent_step=5.83s eta=18:11:15 +2026-09-15 23:05:21 | INFO | train rank=0 | step=32350/50000 progress=64.7% loss=0.105570 lr=3.037047e-05 lr_backbone=3.037047e-05 elapsed=33:18:54 avg_step=3.71s recent_step=3.24s eta=18:10:35 +2026-09-15 23:06:00 | INFO | train rank=0 | step=32360/50000 progress=64.7% loss=0.130423 lr=3.034006e-05 lr_backbone=3.034006e-05 elapsed=33:19:32 avg_step=3.71s recent_step=3.86s eta=18:09:59 +2026-09-15 23:06:31 | INFO | train rank=0 | step=32370/50000 progress=64.7% loss=0.114161 lr=3.030966e-05 lr_backbone=3.030966e-05 elapsed=33:20:04 avg_step=3.71s recent_step=3.18s eta=18:09:19 +2026-09-15 23:07:30 | INFO | train rank=0 | step=32380/50000 progress=64.8% loss=0.121917 lr=3.027927e-05 lr_backbone=3.027927e-05 elapsed=33:21:03 avg_step=3.71s recent_step=5.86s eta=18:08:54 +2026-09-15 23:08:01 | INFO | train rank=0 | step=32390/50000 progress=64.8% loss=0.088454 lr=3.024888e-05 lr_backbone=3.024888e-05 elapsed=33:21:34 avg_step=3.71s recent_step=3.07s eta=18:08:13 +2026-09-15 23:08:37 | INFO | train rank=0 | step=32400/50000 progress=64.8% loss=0.098837 lr=3.021851e-05 lr_backbone=3.021851e-05 elapsed=33:22:10 avg_step=3.71s recent_step=3.63s eta=18:07:36 +2026-09-15 23:09:11 | INFO | train rank=0 | step=32410/50000 progress=64.8% loss=0.119942 lr=3.018814e-05 lr_backbone=3.018814e-05 elapsed=33:22:43 avg_step=3.71s recent_step=3.36s eta=18:06:57 +2026-09-15 23:09:53 | INFO | train rank=0 | step=32420/50000 progress=64.8% loss=0.098032 lr=3.015778e-05 lr_backbone=3.015778e-05 elapsed=33:23:26 avg_step=3.71s recent_step=4.22s eta=18:06:22 +2026-09-15 23:10:25 | INFO | train rank=0 | step=32430/50000 progress=64.9% loss=0.086108 lr=3.012743e-05 lr_backbone=3.012743e-05 elapsed=33:23:57 avg_step=3.71s recent_step=3.17s eta=18:05:42 +2026-09-15 23:10:53 | INFO | train rank=0 | step=32440/50000 progress=64.9% loss=0.082153 lr=3.009709e-05 lr_backbone=3.009709e-05 elapsed=33:24:26 avg_step=3.71s recent_step=2.83s eta=18:05:00 +2026-09-15 23:11:27 | INFO | train rank=0 | step=32450/50000 progress=64.9% loss=0.102571 lr=3.006676e-05 lr_backbone=3.006676e-05 elapsed=33:25:00 avg_step=3.71s recent_step=3.41s eta=18:04:22 +2026-09-15 23:12:13 | INFO | train rank=0 | step=32460/50000 progress=64.9% loss=0.079781 lr=3.003644e-05 lr_backbone=3.003644e-05 elapsed=33:25:45 avg_step=3.71s recent_step=4.56s eta=18:03:49 +2026-09-15 23:12:45 | INFO | train rank=0 | step=32470/50000 progress=64.9% loss=0.127504 lr=3.000612e-05 lr_backbone=3.000612e-05 elapsed=33:26:18 avg_step=3.71s recent_step=3.27s eta=18:03:10 +2026-09-15 23:13:18 | INFO | train rank=0 | step=32480/50000 progress=65.0% loss=0.107521 lr=2.997581e-05 lr_backbone=2.997581e-05 elapsed=33:26:50 avg_step=3.71s recent_step=3.24s eta=18:02:30 +2026-09-15 23:13:47 | INFO | train rank=0 | step=32490/50000 progress=65.0% loss=0.088810 lr=2.994552e-05 lr_backbone=2.994552e-05 elapsed=33:27:20 avg_step=3.71s recent_step=2.95s eta=18:01:49 +2026-09-15 23:14:28 | INFO | train rank=0 | step=32500/50000 progress=65.0% loss=0.073859 lr=2.991523e-05 lr_backbone=2.991523e-05 elapsed=33:28:01 avg_step=3.71s recent_step=4.06s eta=18:01:14 +2026-09-15 23:15:02 | INFO | train rank=0 | step=32510/50000 progress=65.0% loss=0.116689 lr=2.988495e-05 lr_backbone=2.988495e-05 elapsed=33:28:35 avg_step=3.71s recent_step=3.45s eta=18:00:35 +2026-09-15 23:15:38 | INFO | train rank=0 | step=32520/50000 progress=65.0% loss=0.105492 lr=2.985468e-05 lr_backbone=2.985468e-05 elapsed=33:29:11 avg_step=3.71s recent_step=3.60s eta=17:59:58 +2026-09-15 23:16:09 | INFO | train rank=0 | step=32530/50000 progress=65.1% loss=0.121729 lr=2.982442e-05 lr_backbone=2.982442e-05 elapsed=33:29:41 avg_step=3.71s recent_step=3.02s eta=17:59:17 +2026-09-15 23:16:57 | INFO | train rank=0 | step=32540/50000 progress=65.1% loss=0.123306 lr=2.979416e-05 lr_backbone=2.979416e-05 elapsed=33:30:30 avg_step=3.71s recent_step=4.83s eta=17:58:46 +2026-09-15 23:17:37 | INFO | train rank=0 | step=32550/50000 progress=65.1% loss=0.102856 lr=2.976392e-05 lr_backbone=2.976392e-05 elapsed=33:31:10 avg_step=3.71s recent_step=4.04s eta=17:58:11 +2026-09-15 23:18:13 | INFO | train rank=0 | step=32560/50000 progress=65.1% loss=0.090865 lr=2.973368e-05 lr_backbone=2.973368e-05 elapsed=33:31:45 avg_step=3.71s recent_step=3.54s eta=17:57:33 +2026-09-15 23:18:47 | INFO | train rank=0 | step=32570/50000 progress=65.1% loss=0.101391 lr=2.970346e-05 lr_backbone=2.970346e-05 elapsed=33:32:20 avg_step=3.71s recent_step=3.45s eta=17:56:54 +2026-09-15 23:19:38 | INFO | train rank=0 | step=32580/50000 progress=65.2% loss=0.063140 lr=2.967324e-05 lr_backbone=2.967324e-05 elapsed=33:33:10 avg_step=3.71s recent_step=5.05s eta=17:56:24 +2026-09-15 23:20:13 | INFO | train rank=0 | step=32590/50000 progress=65.2% loss=0.099563 lr=2.964303e-05 lr_backbone=2.964303e-05 elapsed=33:33:45 avg_step=3.71s recent_step=3.49s eta=17:55:46 +2026-09-15 23:20:53 | INFO | train rank=0 | step=32600/50000 progress=65.2% loss=0.089391 lr=2.961283e-05 lr_backbone=2.961283e-05 elapsed=33:34:25 avg_step=3.71s recent_step=4.01s eta=17:55:11 +2026-09-15 23:21:24 | INFO | train rank=0 | step=32610/50000 progress=65.2% loss=0.076719 lr=2.958264e-05 lr_backbone=2.958264e-05 elapsed=33:34:56 avg_step=3.71s recent_step=3.10s eta=17:54:30 +2026-09-15 23:22:07 | INFO | train rank=0 | step=32620/50000 progress=65.2% loss=0.124846 lr=2.955246e-05 lr_backbone=2.955246e-05 elapsed=33:35:40 avg_step=3.71s recent_step=4.33s eta=17:53:57 +2026-09-15 23:22:37 | INFO | train rank=0 | step=32630/50000 progress=65.3% loss=0.082942 lr=2.952228e-05 lr_backbone=2.952228e-05 elapsed=33:36:10 avg_step=3.71s recent_step=3.05s eta=17:53:16 +2026-09-15 23:23:15 | INFO | train rank=0 | step=32640/50000 progress=65.3% loss=0.107913 lr=2.949212e-05 lr_backbone=2.949212e-05 elapsed=33:36:48 avg_step=3.71s recent_step=3.75s eta=17:52:39 +2026-09-15 23:23:51 | INFO | train rank=0 | step=32650/50000 progress=65.3% loss=0.118544 lr=2.946196e-05 lr_backbone=2.946196e-05 elapsed=33:37:24 avg_step=3.71s recent_step=3.59s eta=17:52:02 +2026-09-15 23:24:36 | INFO | train rank=0 | step=32660/50000 progress=65.3% loss=0.124679 lr=2.943182e-05 lr_backbone=2.943182e-05 elapsed=33:38:09 avg_step=3.71s recent_step=4.53s eta=17:51:29 +2026-09-15 23:25:16 | INFO | train rank=0 | step=32670/50000 progress=65.3% loss=0.103743 lr=2.940168e-05 lr_backbone=2.940168e-05 elapsed=33:38:49 avg_step=3.71s recent_step=4.00s eta=17:50:53 +2026-09-15 23:25:53 | INFO | train rank=0 | step=32680/50000 progress=65.4% loss=0.094790 lr=2.937155e-05 lr_backbone=2.937155e-05 elapsed=33:39:26 avg_step=3.71s recent_step=3.68s eta=17:50:16 +2026-09-15 23:26:25 | INFO | train rank=0 | step=32690/50000 progress=65.4% loss=0.112326 lr=2.934143e-05 lr_backbone=2.934143e-05 elapsed=33:39:57 avg_step=3.71s recent_step=3.17s eta=17:49:36 +2026-09-15 23:27:08 | INFO | train rank=0 | step=32700/50000 progress=65.4% loss=0.106469 lr=2.931132e-05 lr_backbone=2.931132e-05 elapsed=33:40:41 avg_step=3.71s recent_step=4.38s eta=17:49:03 +2026-09-15 23:27:48 | INFO | train rank=0 | step=32710/50000 progress=65.4% loss=0.083427 lr=2.928122e-05 lr_backbone=2.928122e-05 elapsed=33:41:20 avg_step=3.71s recent_step=3.93s eta=17:48:27 +2026-09-15 23:28:23 | INFO | train rank=0 | step=32720/50000 progress=65.4% loss=0.094558 lr=2.925113e-05 lr_backbone=2.925113e-05 elapsed=33:41:56 avg_step=3.71s recent_step=3.51s eta=17:47:49 +2026-09-15 23:28:55 | INFO | train rank=0 | step=32730/50000 progress=65.5% loss=0.106133 lr=2.922105e-05 lr_backbone=2.922105e-05 elapsed=33:42:28 avg_step=3.71s recent_step=3.23s eta=17:47:09 +2026-09-15 23:29:42 | INFO | train rank=0 | step=32740/50000 progress=65.5% loss=0.088931 lr=2.919097e-05 lr_backbone=2.919097e-05 elapsed=33:43:15 avg_step=3.71s recent_step=4.73s eta=17:46:37 +2026-09-15 23:30:21 | INFO | train rank=0 | step=32750/50000 progress=65.5% loss=0.089564 lr=2.916091e-05 lr_backbone=2.916091e-05 elapsed=33:43:53 avg_step=3.71s recent_step=3.81s eta=17:46:01 +2026-09-15 23:30:51 | INFO | train rank=0 | step=32760/50000 progress=65.5% loss=0.132054 lr=2.913085e-05 lr_backbone=2.913085e-05 elapsed=33:44:24 avg_step=3.71s recent_step=3.02s eta=17:45:20 +2026-09-15 23:31:22 | INFO | train rank=0 | step=32770/50000 progress=65.5% loss=0.101853 lr=2.910080e-05 lr_backbone=2.910080e-05 elapsed=33:44:55 avg_step=3.71s recent_step=3.16s eta=17:44:40 +2026-09-15 23:32:17 | INFO | train rank=0 | step=32780/50000 progress=65.6% loss=0.095575 lr=2.907077e-05 lr_backbone=2.907077e-05 elapsed=33:45:50 avg_step=3.71s recent_step=5.43s eta=17:44:12 +2026-09-15 23:32:46 | INFO | train rank=0 | step=32790/50000 progress=65.6% loss=0.088580 lr=2.904074e-05 lr_backbone=2.904074e-05 elapsed=33:46:18 avg_step=3.71s recent_step=2.88s eta=17:43:31 +2026-09-15 23:33:24 | INFO | train rank=0 | step=32800/50000 progress=65.6% loss=0.116355 lr=2.901072e-05 lr_backbone=2.901072e-05 elapsed=33:46:56 avg_step=3.71s recent_step=3.79s eta=17:42:54 +2026-09-15 23:33:55 | INFO | train rank=0 | step=32810/50000 progress=65.6% loss=0.100816 lr=2.898071e-05 lr_backbone=2.898071e-05 elapsed=33:47:28 avg_step=3.71s recent_step=3.20s eta=17:42:14 +2026-09-15 23:34:51 | INFO | train rank=0 | step=32820/50000 progress=65.6% loss=0.071954 lr=2.895071e-05 lr_backbone=2.895071e-05 elapsed=33:48:24 avg_step=3.71s recent_step=5.58s eta=17:41:47 +2026-09-15 23:35:20 | INFO | train rank=0 | step=32830/50000 progress=65.7% loss=0.071329 lr=2.892072e-05 lr_backbone=2.892072e-05 elapsed=33:48:53 avg_step=3.71s recent_step=2.91s eta=17:41:06 +2026-09-15 23:35:55 | INFO | train rank=0 | step=32840/50000 progress=65.7% loss=0.110327 lr=2.889074e-05 lr_backbone=2.889074e-05 elapsed=33:49:28 avg_step=3.71s recent_step=3.47s eta=17:40:28 +2026-09-15 23:36:31 | INFO | train rank=0 | step=32850/50000 progress=65.7% loss=0.077746 lr=2.886076e-05 lr_backbone=2.886076e-05 elapsed=33:50:04 avg_step=3.71s recent_step=3.57s eta=17:39:50 +2026-09-15 23:37:11 | INFO | train rank=0 | step=32860/50000 progress=65.7% loss=0.096497 lr=2.883080e-05 lr_backbone=2.883080e-05 elapsed=33:50:44 avg_step=3.71s recent_step=4.00s eta=17:39:14 +2026-09-15 23:37:40 | INFO | train rank=0 | step=32870/50000 progress=65.7% loss=0.088002 lr=2.880084e-05 lr_backbone=2.880084e-05 elapsed=33:51:13 avg_step=3.71s recent_step=2.91s eta=17:38:33 +2026-09-15 23:38:10 | INFO | train rank=0 | step=32880/50000 progress=65.8% loss=0.109439 lr=2.877090e-05 lr_backbone=2.877090e-05 elapsed=33:51:43 avg_step=3.71s recent_step=3.01s eta=17:37:52 +2026-09-15 23:38:44 | INFO | train rank=0 | step=32890/50000 progress=65.8% loss=0.060551 lr=2.874096e-05 lr_backbone=2.874096e-05 elapsed=33:52:17 avg_step=3.71s recent_step=3.39s eta=17:37:13 +2026-09-15 23:39:28 | INFO | train rank=0 | step=32900/50000 progress=65.8% loss=0.096023 lr=2.871104e-05 lr_backbone=2.871104e-05 elapsed=33:53:01 avg_step=3.71s recent_step=4.40s eta=17:36:40 +2026-09-15 23:40:02 | INFO | train rank=0 | step=32910/50000 progress=65.8% loss=0.117930 lr=2.868112e-05 lr_backbone=2.868112e-05 elapsed=33:53:34 avg_step=3.71s recent_step=3.39s eta=17:36:01 +2026-09-15 23:40:43 | INFO | train rank=0 | step=32920/50000 progress=65.8% loss=0.096274 lr=2.865121e-05 lr_backbone=2.865121e-05 elapsed=33:54:16 avg_step=3.71s recent_step=4.13s eta=17:35:26 +2026-09-15 23:41:16 | INFO | train rank=0 | step=32930/50000 progress=65.9% loss=0.066779 lr=2.862131e-05 lr_backbone=2.862131e-05 elapsed=33:54:48 avg_step=3.71s recent_step=3.25s eta=17:34:47 +2026-09-15 23:42:05 | INFO | train rank=0 | step=32940/50000 progress=65.9% loss=0.094643 lr=2.859142e-05 lr_backbone=2.859142e-05 elapsed=33:55:38 avg_step=3.71s recent_step=4.92s eta=17:34:16 +2026-09-15 23:42:40 | INFO | train rank=0 | step=32950/50000 progress=65.9% loss=0.129149 lr=2.856154e-05 lr_backbone=2.856154e-05 elapsed=33:56:12 avg_step=3.71s recent_step=3.47s eta=17:33:38 +2026-09-15 23:43:11 | INFO | train rank=0 | step=32960/50000 progress=65.9% loss=0.095138 lr=2.853167e-05 lr_backbone=2.853167e-05 elapsed=33:56:43 avg_step=3.71s recent_step=3.10s eta=17:32:58 +2026-09-15 23:43:47 | INFO | train rank=0 | step=32970/50000 progress=65.9% loss=0.079950 lr=2.850181e-05 lr_backbone=2.850181e-05 elapsed=33:57:20 avg_step=3.71s recent_step=3.69s eta=17:32:21 +2026-09-15 23:44:30 | INFO | train rank=0 | step=32980/50000 progress=66.0% loss=0.089965 lr=2.847196e-05 lr_backbone=2.847196e-05 elapsed=33:58:03 avg_step=3.71s recent_step=4.27s eta=17:31:46 +2026-09-15 23:45:09 | INFO | train rank=0 | step=32990/50000 progress=66.0% loss=0.107512 lr=2.844212e-05 lr_backbone=2.844212e-05 elapsed=33:58:42 avg_step=3.71s recent_step=3.88s eta=17:31:10 +2026-09-15 23:45:41 | INFO | train rank=0 | step=33000/50000 progress=66.0% loss=0.079858 lr=2.841228e-05 lr_backbone=2.841228e-05 elapsed=33:59:14 avg_step=3.71s recent_step=3.19s eta=17:30:30 +2026-09-15 23:46:15 | INFO | train rank=0 | step=33010/50000 progress=66.0% loss=0.076062 lr=2.838246e-05 lr_backbone=2.838246e-05 elapsed=33:59:48 avg_step=3.71s recent_step=3.40s eta=17:29:52 +2026-09-15 23:47:03 | INFO | train rank=0 | step=33020/50000 progress=66.0% loss=0.089042 lr=2.835264e-05 lr_backbone=2.835264e-05 elapsed=34:00:36 avg_step=3.71s recent_step=4.82s eta=17:29:20 +2026-09-15 23:47:36 | INFO | train rank=0 | step=33030/50000 progress=66.1% loss=0.072497 lr=2.832284e-05 lr_backbone=2.832284e-05 elapsed=34:01:09 avg_step=3.71s recent_step=3.33s eta=17:28:41 +2026-09-15 23:48:09 | INFO | train rank=0 | step=33040/50000 progress=66.1% loss=0.085529 lr=2.829304e-05 lr_backbone=2.829304e-05 elapsed=34:01:42 avg_step=3.71s recent_step=3.25s eta=17:28:02 +2026-09-15 23:48:44 | INFO | train rank=0 | step=33050/50000 progress=66.1% loss=0.096474 lr=2.826326e-05 lr_backbone=2.826326e-05 elapsed=34:02:17 avg_step=3.71s recent_step=3.50s eta=17:27:24 +2026-09-15 23:49:25 | INFO | train rank=0 | step=33060/50000 progress=66.1% loss=0.105726 lr=2.823348e-05 lr_backbone=2.823348e-05 elapsed=34:02:58 avg_step=3.71s recent_step=4.13s eta=17:26:49 +2026-09-15 23:50:01 | INFO | train rank=0 | step=33070/50000 progress=66.1% loss=0.102200 lr=2.820372e-05 lr_backbone=2.820372e-05 elapsed=34:03:33 avg_step=3.71s recent_step=3.54s eta=17:26:11 +2026-09-15 23:50:35 | INFO | train rank=0 | step=33080/50000 progress=66.2% loss=0.078434 lr=2.817396e-05 lr_backbone=2.817396e-05 elapsed=34:04:08 avg_step=3.71s recent_step=3.46s eta=17:25:33 +2026-09-15 23:51:04 | INFO | train rank=0 | step=33090/50000 progress=66.2% loss=0.095570 lr=2.814421e-05 lr_backbone=2.814421e-05 elapsed=34:04:37 avg_step=3.71s recent_step=2.88s eta=17:24:51 +2026-09-15 23:51:55 | INFO | train rank=0 | step=33100/50000 progress=66.2% loss=0.089044 lr=2.811447e-05 lr_backbone=2.811447e-05 elapsed=34:05:27 avg_step=3.71s recent_step=5.07s eta=17:24:21 +2026-09-15 23:52:26 | INFO | train rank=0 | step=33110/50000 progress=66.2% loss=0.075738 lr=2.808475e-05 lr_backbone=2.808475e-05 elapsed=34:05:58 avg_step=3.71s recent_step=3.11s eta=17:23:41 +2026-09-15 23:52:50 | INFO | train rank=0 | step=33120/50000 progress=66.2% loss=0.082931 lr=2.805503e-05 lr_backbone=2.805503e-05 elapsed=34:06:22 avg_step=3.71s recent_step=2.39s eta=17:22:57 +2026-09-15 23:53:25 | INFO | train rank=0 | step=33130/50000 progress=66.3% loss=0.100568 lr=2.802532e-05 lr_backbone=2.802532e-05 elapsed=34:06:58 avg_step=3.71s recent_step=3.52s eta=17:22:19 +2026-09-15 23:54:14 | INFO | train rank=0 | step=33140/50000 progress=66.3% loss=0.091761 lr=2.799562e-05 lr_backbone=2.799562e-05 elapsed=34:07:47 avg_step=3.71s recent_step=4.93s eta=17:21:48 +2026-09-15 23:54:48 | INFO | train rank=0 | step=33150/50000 progress=66.3% loss=0.082935 lr=2.796593e-05 lr_backbone=2.796593e-05 elapsed=34:08:21 avg_step=3.71s recent_step=3.39s eta=17:21:10 +2026-09-15 23:55:25 | INFO | train rank=0 | step=33160/50000 progress=66.3% loss=0.099706 lr=2.793625e-05 lr_backbone=2.793625e-05 elapsed=34:08:57 avg_step=3.71s recent_step=3.66s eta=17:20:32 +2026-09-15 23:55:59 | INFO | train rank=0 | step=33170/50000 progress=66.3% loss=0.056615 lr=2.790658e-05 lr_backbone=2.790658e-05 elapsed=34:09:32 avg_step=3.71s recent_step=3.50s eta=17:19:54 +2026-09-15 23:56:59 | INFO | train rank=0 | step=33180/50000 progress=66.4% loss=0.102228 lr=2.787692e-05 lr_backbone=2.787692e-05 elapsed=34:10:32 avg_step=3.71s recent_step=5.96s eta=17:19:29 +2026-09-15 23:57:38 | INFO | train rank=0 | step=33190/50000 progress=66.4% loss=0.131492 lr=2.784726e-05 lr_backbone=2.784726e-05 elapsed=34:11:11 avg_step=3.71s recent_step=3.92s eta=17:18:53 +2026-09-15 23:58:12 | INFO | train rank=0 | step=33200/50000 progress=66.4% loss=0.090722 lr=2.781762e-05 lr_backbone=2.781762e-05 elapsed=34:11:45 avg_step=3.71s recent_step=3.38s eta=17:18:14 +2026-09-15 23:58:43 | INFO | train rank=0 | step=33210/50000 progress=66.4% loss=0.103205 lr=2.778799e-05 lr_backbone=2.778799e-05 elapsed=34:12:16 avg_step=3.71s recent_step=3.13s eta=17:17:34 +2026-09-15 23:59:33 | INFO | train rank=0 | step=33220/50000 progress=66.4% loss=0.091330 lr=2.775837e-05 lr_backbone=2.775837e-05 elapsed=34:13:06 avg_step=3.71s recent_step=4.94s eta=17:17:03 +2026-09-16 00:00:03 | INFO | train rank=0 | step=33230/50000 progress=66.5% loss=0.120835 lr=2.772876e-05 lr_backbone=2.772876e-05 elapsed=34:13:35 avg_step=3.71s recent_step=2.98s eta=17:16:22 +2026-09-16 00:00:37 | INFO | train rank=0 | step=33240/50000 progress=66.5% loss=0.085551 lr=2.769915e-05 lr_backbone=2.769915e-05 elapsed=34:14:10 avg_step=3.71s recent_step=3.49s eta=17:15:44 +2026-09-16 00:01:09 | INFO | train rank=0 | step=33250/50000 progress=66.5% loss=0.098563 lr=2.766956e-05 lr_backbone=2.766956e-05 elapsed=34:14:42 avg_step=3.71s recent_step=3.16s eta=17:15:04 +2026-09-16 00:02:01 | INFO | train rank=0 | step=33260/50000 progress=66.5% loss=0.125502 lr=2.763998e-05 lr_backbone=2.763998e-05 elapsed=34:15:34 avg_step=3.71s recent_step=5.19s eta=17:14:35 +2026-09-16 00:02:29 | INFO | train rank=0 | step=33270/50000 progress=66.5% loss=0.102414 lr=2.761040e-05 lr_backbone=2.761040e-05 elapsed=34:16:02 avg_step=3.71s recent_step=2.83s eta=17:13:53 +2026-09-16 00:03:00 | INFO | train rank=0 | step=33280/50000 progress=66.6% loss=0.086060 lr=2.758084e-05 lr_backbone=2.758084e-05 elapsed=34:16:33 avg_step=3.71s recent_step=3.09s eta=17:13:13 +2026-09-16 00:03:30 | INFO | train rank=0 | step=33290/50000 progress=66.6% loss=0.087398 lr=2.755129e-05 lr_backbone=2.755129e-05 elapsed=34:17:03 avg_step=3.71s recent_step=2.95s eta=17:12:32 +2026-09-16 00:04:15 | INFO | train rank=0 | step=33300/50000 progress=66.6% loss=0.108501 lr=2.752174e-05 lr_backbone=2.752174e-05 elapsed=34:17:48 avg_step=3.71s recent_step=4.54s eta=17:11:59 +2026-09-16 00:04:42 | INFO | train rank=0 | step=33310/50000 progress=66.6% loss=0.102193 lr=2.749221e-05 lr_backbone=2.749221e-05 elapsed=34:18:15 avg_step=3.71s recent_step=2.68s eta=17:11:17 +2026-09-16 00:05:17 | INFO | train rank=0 | step=33320/50000 progress=66.6% loss=0.070672 lr=2.746268e-05 lr_backbone=2.746268e-05 elapsed=34:18:50 avg_step=3.71s recent_step=3.51s eta=17:10:39 +2026-09-16 00:05:51 | INFO | train rank=0 | step=33330/50000 progress=66.7% loss=0.110428 lr=2.743317e-05 lr_backbone=2.743317e-05 elapsed=34:19:24 avg_step=3.71s recent_step=3.38s eta=17:10:00 +2026-09-16 00:06:40 | INFO | train rank=0 | step=33340/50000 progress=66.7% loss=0.099688 lr=2.740366e-05 lr_backbone=2.740366e-05 elapsed=34:20:12 avg_step=3.71s recent_step=4.88s eta=17:09:29 +2026-09-16 00:07:13 | INFO | train rank=0 | step=33350/50000 progress=66.7% loss=0.128148 lr=2.737417e-05 lr_backbone=2.737417e-05 elapsed=34:20:46 avg_step=3.71s recent_step=3.35s eta=17:08:50 +2026-09-16 00:07:48 | INFO | train rank=0 | step=33360/50000 progress=66.7% loss=0.083991 lr=2.734468e-05 lr_backbone=2.734468e-05 elapsed=34:21:21 avg_step=3.71s recent_step=3.53s eta=17:08:12 +2026-09-16 00:08:23 | INFO | train rank=0 | step=33370/50000 progress=66.7% loss=0.114947 lr=2.731521e-05 lr_backbone=2.731521e-05 elapsed=34:21:56 avg_step=3.71s recent_step=3.45s eta=17:07:34 +2026-09-16 00:09:17 | INFO | train rank=0 | step=33380/50000 progress=66.8% loss=0.066649 lr=2.728574e-05 lr_backbone=2.728574e-05 elapsed=34:22:50 avg_step=3.71s recent_step=5.41s eta=17:07:05 +2026-09-16 00:09:48 | INFO | train rank=0 | step=33390/50000 progress=66.8% loss=0.097448 lr=2.725629e-05 lr_backbone=2.725629e-05 elapsed=34:23:20 avg_step=3.71s recent_step=3.06s eta=17:06:25 +2026-09-16 00:10:20 | INFO | train rank=0 | step=33400/50000 progress=66.8% loss=0.121725 lr=2.722684e-05 lr_backbone=2.722684e-05 elapsed=34:23:53 avg_step=3.71s recent_step=3.22s eta=17:05:45 +2026-09-16 00:10:56 | INFO | train rank=0 | step=33410/50000 progress=66.8% loss=0.094333 lr=2.719741e-05 lr_backbone=2.719741e-05 elapsed=34:24:29 avg_step=3.71s recent_step=3.65s eta=17:05:08 +2026-09-16 00:11:44 | INFO | train rank=0 | step=33420/50000 progress=66.8% loss=0.069524 lr=2.716798e-05 lr_backbone=2.716798e-05 elapsed=34:25:17 avg_step=3.71s recent_step=4.74s eta=17:04:36 +2026-09-16 00:12:15 | INFO | train rank=0 | step=33430/50000 progress=66.9% loss=0.136757 lr=2.713857e-05 lr_backbone=2.713857e-05 elapsed=34:25:47 avg_step=3.71s recent_step=3.08s eta=17:03:56 +2026-09-16 00:12:45 | INFO | train rank=0 | step=33440/50000 progress=66.9% loss=0.068228 lr=2.710916e-05 lr_backbone=2.710916e-05 elapsed=34:26:17 avg_step=3.71s recent_step=3.00s eta=17:03:15 +2026-09-16 00:13:37 | INFO | train rank=0 | step=33450/50000 progress=66.9% loss=0.122359 lr=2.707977e-05 lr_backbone=2.707977e-05 elapsed=34:27:10 avg_step=3.71s recent_step=5.27s eta=17:02:46 +2026-09-16 00:14:20 | INFO | train rank=0 | step=33460/50000 progress=66.9% loss=0.094361 lr=2.705038e-05 lr_backbone=2.705038e-05 elapsed=34:27:53 avg_step=3.71s recent_step=4.29s eta=17:02:12 +2026-09-16 00:14:50 | INFO | train rank=0 | step=33470/50000 progress=66.9% loss=0.070372 lr=2.702101e-05 lr_backbone=2.702101e-05 elapsed=34:28:23 avg_step=3.71s recent_step=3.00s eta=17:01:31 +2026-09-16 00:15:22 | INFO | train rank=0 | step=33480/50000 progress=67.0% loss=0.071307 lr=2.699164e-05 lr_backbone=2.699164e-05 elapsed=34:28:55 avg_step=3.71s recent_step=3.16s eta=17:00:51 +2026-09-16 00:16:16 | INFO | train rank=0 | step=33490/50000 progress=67.0% loss=0.072661 lr=2.696229e-05 lr_backbone=2.696229e-05 elapsed=34:29:49 avg_step=3.71s recent_step=5.43s eta=17:00:23 +2026-09-16 00:16:47 | INFO | train rank=0 | step=33500/50000 progress=67.0% loss=0.123709 lr=2.693294e-05 lr_backbone=2.693294e-05 elapsed=34:30:20 avg_step=3.71s recent_step=3.11s eta=16:59:43 +2026-09-16 00:17:16 | INFO | train rank=0 | step=33510/50000 progress=67.0% loss=0.089681 lr=2.690361e-05 lr_backbone=2.690361e-05 elapsed=34:30:48 avg_step=3.71s recent_step=2.85s eta=16:59:01 +2026-09-16 00:17:54 | INFO | train rank=0 | step=33520/50000 progress=67.0% loss=0.078815 lr=2.687428e-05 lr_backbone=2.687428e-05 elapsed=34:31:27 avg_step=3.71s recent_step=3.83s eta=16:58:25 +2026-09-16 00:18:50 | INFO | train rank=0 | step=33530/50000 progress=67.1% loss=0.096417 lr=2.684497e-05 lr_backbone=2.684497e-05 elapsed=34:32:23 avg_step=3.71s recent_step=5.60s eta=16:57:57 +2026-09-16 00:19:19 | INFO | train rank=0 | step=33540/50000 progress=67.1% loss=0.095350 lr=2.681566e-05 lr_backbone=2.681566e-05 elapsed=34:32:52 avg_step=3.71s recent_step=2.95s eta=16:57:16 +2026-09-16 00:19:45 | INFO | train rank=0 | step=33550/50000 progress=67.1% loss=0.115149 lr=2.678637e-05 lr_backbone=2.678637e-05 elapsed=34:33:18 avg_step=3.71s recent_step=2.53s eta=16:56:34 +2026-09-16 00:20:19 | INFO | train rank=0 | step=33560/50000 progress=67.1% loss=0.081898 lr=2.675708e-05 lr_backbone=2.675708e-05 elapsed=34:33:52 avg_step=3.71s recent_step=3.43s eta=16:55:55 +2026-09-16 00:21:23 | INFO | train rank=0 | step=33570/50000 progress=67.1% loss=0.096301 lr=2.672781e-05 lr_backbone=2.672781e-05 elapsed=34:34:56 avg_step=3.71s recent_step=6.41s eta=16:55:31 +2026-09-16 00:21:58 | INFO | train rank=0 | step=33580/50000 progress=67.2% loss=0.088673 lr=2.669855e-05 lr_backbone=2.669855e-05 elapsed=34:35:31 avg_step=3.71s recent_step=3.51s eta=16:54:53 +2026-09-16 00:22:36 | INFO | train rank=0 | step=33590/50000 progress=67.2% loss=0.118282 lr=2.666929e-05 lr_backbone=2.666929e-05 elapsed=34:36:09 avg_step=3.71s recent_step=3.76s eta=16:54:16 +2026-09-16 00:23:07 | INFO | train rank=0 | step=33600/50000 progress=67.2% loss=0.110903 lr=2.664005e-05 lr_backbone=2.664005e-05 elapsed=34:36:40 avg_step=3.71s recent_step=3.15s eta=16:53:36 +2026-09-16 00:23:57 | INFO | train rank=0 | step=33610/50000 progress=67.2% loss=0.097433 lr=2.661082e-05 lr_backbone=2.661082e-05 elapsed=34:37:30 avg_step=3.71s recent_step=4.99s eta=16:53:06 +2026-09-16 00:24:25 | INFO | train rank=0 | step=33620/50000 progress=67.2% loss=0.100799 lr=2.658159e-05 lr_backbone=2.658159e-05 elapsed=34:37:58 avg_step=3.71s recent_step=2.77s eta=16:52:24 +2026-09-16 00:24:57 | INFO | train rank=0 | step=33630/50000 progress=67.3% loss=0.118006 lr=2.655238e-05 lr_backbone=2.655238e-05 elapsed=34:38:29 avg_step=3.71s recent_step=3.15s eta=16:51:44 +2026-09-16 00:25:34 | INFO | train rank=0 | step=33640/50000 progress=67.3% loss=0.078650 lr=2.652318e-05 lr_backbone=2.652318e-05 elapsed=34:39:07 avg_step=3.71s recent_step=3.74s eta=16:51:07 +2026-09-16 00:26:34 | INFO | train rank=0 | step=33650/50000 progress=67.3% loss=0.073944 lr=2.649399e-05 lr_backbone=2.649399e-05 elapsed=34:40:07 avg_step=3.71s recent_step=6.04s eta=16:50:42 +2026-09-16 00:27:08 | INFO | train rank=0 | step=33660/50000 progress=67.3% loss=0.107753 lr=2.646480e-05 lr_backbone=2.646480e-05 elapsed=34:40:41 avg_step=3.71s recent_step=3.39s eta=16:50:03 +2026-09-16 00:27:47 | INFO | train rank=0 | step=33670/50000 progress=67.3% loss=0.075228 lr=2.643563e-05 lr_backbone=2.643563e-05 elapsed=34:41:19 avg_step=3.71s recent_step=3.84s eta=16:49:26 +2026-09-16 00:28:16 | INFO | train rank=0 | step=33680/50000 progress=67.4% loss=0.053316 lr=2.640647e-05 lr_backbone=2.640647e-05 elapsed=34:41:48 avg_step=3.71s recent_step=2.90s eta=16:48:45 +2026-09-16 00:29:06 | INFO | train rank=0 | step=33690/50000 progress=67.4% loss=0.099249 lr=2.637732e-05 lr_backbone=2.637732e-05 elapsed=34:42:39 avg_step=3.71s recent_step=5.09s eta=16:48:15 +2026-09-16 00:29:45 | INFO | train rank=0 | step=33700/50000 progress=67.4% loss=0.102067 lr=2.634818e-05 lr_backbone=2.634818e-05 elapsed=34:43:18 avg_step=3.71s recent_step=3.89s eta=16:47:39 +2026-09-16 00:30:13 | INFO | train rank=0 | step=33710/50000 progress=67.4% loss=0.126299 lr=2.631905e-05 lr_backbone=2.631905e-05 elapsed=34:43:46 avg_step=3.71s recent_step=2.81s eta=16:46:57 +2026-09-16 00:30:48 | INFO | train rank=0 | step=33720/50000 progress=67.4% loss=0.111213 lr=2.628993e-05 lr_backbone=2.628993e-05 elapsed=34:44:20 avg_step=3.71s recent_step=3.41s eta=16:46:19 +2026-09-16 00:31:27 | INFO | train rank=0 | step=33730/50000 progress=67.5% loss=0.115492 lr=2.626082e-05 lr_backbone=2.626082e-05 elapsed=34:45:00 avg_step=3.71s recent_step=3.94s eta=16:45:43 +2026-09-16 00:31:58 | INFO | train rank=0 | step=33740/50000 progress=67.5% loss=0.102568 lr=2.623172e-05 lr_backbone=2.623172e-05 elapsed=34:45:31 avg_step=3.71s recent_step=3.12s eta=16:45:03 +2026-09-16 00:32:41 | INFO | train rank=0 | step=33750/50000 progress=67.5% loss=0.101756 lr=2.620263e-05 lr_backbone=2.620263e-05 elapsed=34:46:13 avg_step=3.71s recent_step=4.25s eta=16:44:28 +2026-09-16 00:33:15 | INFO | train rank=0 | step=33760/50000 progress=67.5% loss=0.078835 lr=2.617355e-05 lr_backbone=2.617355e-05 elapsed=34:46:48 avg_step=3.71s recent_step=3.46s eta=16:43:50 +2026-09-16 00:34:09 | INFO | train rank=0 | step=33770/50000 progress=67.5% loss=0.122042 lr=2.614448e-05 lr_backbone=2.614448e-05 elapsed=34:47:41 avg_step=3.71s recent_step=5.34s eta=16:43:21 +2026-09-16 00:34:36 | INFO | train rank=0 | step=33780/50000 progress=67.6% loss=0.085581 lr=2.611543e-05 lr_backbone=2.611543e-05 elapsed=34:48:09 avg_step=3.71s recent_step=2.76s eta=16:42:39 +2026-09-16 00:35:14 | INFO | train rank=0 | step=33790/50000 progress=67.6% loss=0.113622 lr=2.608638e-05 lr_backbone=2.608638e-05 elapsed=34:48:47 avg_step=3.71s recent_step=3.78s eta=16:42:02 +2026-09-16 00:35:45 | INFO | train rank=0 | step=33800/50000 progress=67.6% loss=0.133658 lr=2.605734e-05 lr_backbone=2.605734e-05 elapsed=34:49:18 avg_step=3.71s recent_step=3.14s eta=16:41:23 +2026-09-16 00:36:34 | INFO | train rank=0 | step=33810/50000 progress=67.6% loss=0.074535 lr=2.602832e-05 lr_backbone=2.602832e-05 elapsed=34:50:07 avg_step=3.71s recent_step=4.89s eta=16:40:51 +2026-09-16 00:37:07 | INFO | train rank=0 | step=33820/50000 progress=67.6% loss=0.105766 lr=2.599930e-05 lr_backbone=2.599930e-05 elapsed=34:50:40 avg_step=3.71s recent_step=3.26s eta=16:40:12 +2026-09-16 00:37:44 | INFO | train rank=0 | step=33830/50000 progress=67.7% loss=0.129450 lr=2.597030e-05 lr_backbone=2.597030e-05 elapsed=34:51:16 avg_step=3.71s recent_step=3.69s eta=16:39:35 +2026-09-16 00:38:19 | INFO | train rank=0 | step=33840/50000 progress=67.7% loss=0.117085 lr=2.594130e-05 lr_backbone=2.594130e-05 elapsed=34:51:52 avg_step=3.71s recent_step=3.53s eta=16:38:57 +2026-09-16 00:39:03 | INFO | train rank=0 | step=33850/50000 progress=67.7% loss=0.107030 lr=2.591232e-05 lr_backbone=2.591232e-05 elapsed=34:52:36 avg_step=3.71s recent_step=4.42s eta=16:38:23 +2026-09-16 00:39:35 | INFO | train rank=0 | step=33860/50000 progress=67.7% loss=0.100823 lr=2.588334e-05 lr_backbone=2.588334e-05 elapsed=34:53:07 avg_step=3.71s recent_step=3.14s eta=16:37:43 +2026-09-16 00:40:04 | INFO | train rank=0 | step=33870/50000 progress=67.7% loss=0.079544 lr=2.585438e-05 lr_backbone=2.585438e-05 elapsed=34:53:36 avg_step=3.71s recent_step=2.90s eta=16:37:02 +2026-09-16 00:40:35 | INFO | train rank=0 | step=33880/50000 progress=67.8% loss=0.129094 lr=2.582543e-05 lr_backbone=2.582543e-05 elapsed=34:54:08 avg_step=3.71s recent_step=3.13s eta=16:36:23 +2026-09-16 00:41:31 | INFO | train rank=0 | step=33890/50000 progress=67.8% loss=0.083063 lr=2.579648e-05 lr_backbone=2.579648e-05 elapsed=34:55:03 avg_step=3.71s recent_step=5.58s eta=16:35:54 +2026-09-16 00:42:04 | INFO | train rank=0 | step=33900/50000 progress=67.8% loss=0.083283 lr=2.576755e-05 lr_backbone=2.576755e-05 elapsed=34:55:37 avg_step=3.71s recent_step=3.35s eta=16:35:16 +2026-09-16 00:42:39 | INFO | train rank=0 | step=33910/50000 progress=67.8% loss=0.088329 lr=2.573863e-05 lr_backbone=2.573863e-05 elapsed=34:56:12 avg_step=3.71s recent_step=3.51s eta=16:34:38 +2026-09-16 00:43:09 | INFO | train rank=0 | step=33920/50000 progress=67.8% loss=0.114754 lr=2.570972e-05 lr_backbone=2.570972e-05 elapsed=34:56:42 avg_step=3.71s recent_step=2.97s eta=16:33:57 +2026-09-16 00:43:53 | INFO | train rank=0 | step=33930/50000 progress=67.9% loss=0.082369 lr=2.568082e-05 lr_backbone=2.568082e-05 elapsed=34:57:26 avg_step=3.71s recent_step=4.42s eta=16:33:23 +2026-09-16 00:44:22 | INFO | train rank=0 | step=33940/50000 progress=67.9% loss=0.069724 lr=2.565193e-05 lr_backbone=2.565193e-05 elapsed=34:57:55 avg_step=3.71s recent_step=2.89s eta=16:32:42 +2026-09-16 00:44:57 | INFO | train rank=0 | step=33950/50000 progress=67.9% loss=0.086810 lr=2.562306e-05 lr_backbone=2.562306e-05 elapsed=34:58:30 avg_step=3.71s recent_step=3.51s eta=16:32:04 +2026-09-16 00:45:35 | INFO | train rank=0 | step=33960/50000 progress=67.9% loss=0.061089 lr=2.559419e-05 lr_backbone=2.559419e-05 elapsed=34:59:07 avg_step=3.71s recent_step=3.73s eta=16:31:27 +2026-09-16 00:46:20 | INFO | train rank=0 | step=33970/50000 progress=67.9% loss=0.105416 lr=2.556533e-05 lr_backbone=2.556533e-05 elapsed=34:59:53 avg_step=3.71s recent_step=4.56s eta=16:30:54 +2026-09-16 00:46:57 | INFO | train rank=0 | step=33980/50000 progress=68.0% loss=0.086833 lr=2.553649e-05 lr_backbone=2.553649e-05 elapsed=35:00:29 avg_step=3.71s recent_step=3.64s eta=16:30:17 +2026-09-16 00:47:30 | INFO | train rank=0 | step=33990/50000 progress=68.0% loss=0.109870 lr=2.550765e-05 lr_backbone=2.550765e-05 elapsed=35:01:03 avg_step=3.71s recent_step=3.33s eta=16:29:38 +2026-09-16 00:47:57 | INFO | train rank=0 | step=34000/50000 progress=68.0% loss=0.062447 lr=2.547882e-05 lr_backbone=2.547882e-05 elapsed=35:01:29 avg_step=3.71s recent_step=2.68s eta=16:28:56 +2026-09-16 00:48:52 | INFO | train rank=0 | step=34010/50000 progress=68.0% loss=0.083983 lr=2.545001e-05 lr_backbone=2.545001e-05 elapsed=35:02:24 avg_step=3.71s recent_step=5.51s eta=16:28:27 +2026-09-16 00:49:25 | INFO | train rank=0 | step=34020/50000 progress=68.0% loss=0.090760 lr=2.542121e-05 lr_backbone=2.542121e-05 elapsed=35:02:58 avg_step=3.71s recent_step=3.36s eta=16:27:49 +2026-09-16 00:50:00 | INFO | train rank=0 | step=34030/50000 progress=68.1% loss=0.098866 lr=2.539241e-05 lr_backbone=2.539241e-05 elapsed=35:03:33 avg_step=3.71s recent_step=3.48s eta=16:27:10 +2026-09-16 00:50:35 | INFO | train rank=0 | step=34040/50000 progress=68.1% loss=0.099203 lr=2.536363e-05 lr_backbone=2.536363e-05 elapsed=35:04:07 avg_step=3.71s recent_step=3.44s eta=16:26:32 +2026-09-16 00:51:22 | INFO | train rank=0 | step=34050/50000 progress=68.1% loss=0.094718 lr=2.533486e-05 lr_backbone=2.533486e-05 elapsed=35:04:55 avg_step=3.71s recent_step=4.75s eta=16:26:00 +2026-09-16 00:51:58 | INFO | train rank=0 | step=34060/50000 progress=68.1% loss=0.080954 lr=2.530610e-05 lr_backbone=2.530610e-05 elapsed=35:05:31 avg_step=3.71s recent_step=3.58s eta=16:25:22 +2026-09-16 00:52:27 | INFO | train rank=0 | step=34070/50000 progress=68.1% loss=0.118808 lr=2.527735e-05 lr_backbone=2.527735e-05 elapsed=35:05:59 avg_step=3.71s recent_step=2.87s eta=16:24:41 +2026-09-16 00:53:09 | INFO | train rank=0 | step=34080/50000 progress=68.2% loss=0.094089 lr=2.524861e-05 lr_backbone=2.524861e-05 elapsed=35:06:41 avg_step=3.71s recent_step=4.19s eta=16:24:06 +2026-09-16 00:53:52 | INFO | train rank=0 | step=34090/50000 progress=68.2% loss=0.116328 lr=2.521989e-05 lr_backbone=2.521989e-05 elapsed=35:07:25 avg_step=3.71s recent_step=4.33s eta=16:23:32 +2026-09-16 00:54:36 | INFO | train rank=0 | step=34100/50000 progress=68.2% loss=0.072590 lr=2.519117e-05 lr_backbone=2.519117e-05 elapsed=35:08:09 avg_step=3.71s recent_step=4.38s eta=16:22:58 +2026-09-16 00:55:10 | INFO | train rank=0 | step=34110/50000 progress=68.2% loss=0.083325 lr=2.516246e-05 lr_backbone=2.516246e-05 elapsed=35:08:43 avg_step=3.71s recent_step=3.45s eta=16:22:20 +2026-09-16 00:55:48 | INFO | train rank=0 | step=34120/50000 progress=68.2% loss=0.086467 lr=2.513377e-05 lr_backbone=2.513377e-05 elapsed=35:09:21 avg_step=3.71s recent_step=3.79s eta=16:21:43 +2026-09-16 00:56:30 | INFO | train rank=0 | step=34130/50000 progress=68.3% loss=0.076982 lr=2.510508e-05 lr_backbone=2.510508e-05 elapsed=35:10:03 avg_step=3.71s recent_step=4.20s eta=16:21:08 +2026-09-16 00:57:10 | INFO | train rank=0 | step=34140/50000 progress=68.3% loss=0.080227 lr=2.507641e-05 lr_backbone=2.507641e-05 elapsed=35:10:43 avg_step=3.71s recent_step=4.03s eta=16:20:33 +2026-09-16 00:57:40 | INFO | train rank=0 | step=34150/50000 progress=68.3% loss=0.084335 lr=2.504775e-05 lr_backbone=2.504775e-05 elapsed=35:11:13 avg_step=3.71s recent_step=2.99s eta=16:19:52 +2026-09-16 00:58:16 | INFO | train rank=0 | step=34160/50000 progress=68.3% loss=0.079385 lr=2.501910e-05 lr_backbone=2.501910e-05 elapsed=35:11:49 avg_step=3.71s recent_step=3.58s eta=16:19:15 +2026-09-16 00:59:07 | INFO | train rank=0 | step=34170/50000 progress=68.3% loss=0.105112 lr=2.499045e-05 lr_backbone=2.499045e-05 elapsed=35:12:39 avg_step=3.71s recent_step=5.05s eta=16:18:44 +2026-09-16 00:59:43 | INFO | train rank=0 | step=34180/50000 progress=68.4% loss=0.070002 lr=2.496182e-05 lr_backbone=2.496182e-05 elapsed=35:13:16 avg_step=3.71s recent_step=3.62s eta=16:18:06 +2026-09-16 01:00:12 | INFO | train rank=0 | step=34190/50000 progress=68.4% loss=0.108261 lr=2.493321e-05 lr_backbone=2.493321e-05 elapsed=35:13:45 avg_step=3.71s recent_step=2.94s eta=16:17:26 +2026-09-16 01:00:44 | INFO | train rank=0 | step=34200/50000 progress=68.4% loss=0.094476 lr=2.490460e-05 lr_backbone=2.490460e-05 elapsed=35:14:17 avg_step=3.71s recent_step=3.14s eta=16:16:46 +2026-09-16 01:01:30 | INFO | train rank=0 | step=34210/50000 progress=68.4% loss=0.103003 lr=2.487600e-05 lr_backbone=2.487600e-05 elapsed=35:15:03 avg_step=3.71s recent_step=4.62s eta=16:16:13 +2026-09-16 01:02:06 | INFO | train rank=0 | step=34220/50000 progress=68.4% loss=0.101689 lr=2.484741e-05 lr_backbone=2.484741e-05 elapsed=35:15:38 avg_step=3.71s recent_step=3.55s eta=16:15:35 +2026-09-16 01:02:44 | INFO | train rank=0 | step=34230/50000 progress=68.5% loss=0.063178 lr=2.481884e-05 lr_backbone=2.481884e-05 elapsed=35:16:17 avg_step=3.71s recent_step=3.82s eta=16:14:59 +2026-09-16 01:03:08 | INFO | train rank=0 | step=34240/50000 progress=68.5% loss=0.109540 lr=2.479028e-05 lr_backbone=2.479028e-05 elapsed=35:16:41 avg_step=3.71s recent_step=2.41s eta=16:14:16 +2026-09-16 01:04:04 | INFO | train rank=0 | step=34250/50000 progress=68.5% loss=0.096744 lr=2.476172e-05 lr_backbone=2.476172e-05 elapsed=35:17:37 avg_step=3.71s recent_step=5.65s eta=16:13:47 +2026-09-16 01:04:41 | INFO | train rank=0 | step=34260/50000 progress=68.5% loss=0.083644 lr=2.473318e-05 lr_backbone=2.473318e-05 elapsed=35:18:14 avg_step=3.71s recent_step=3.66s eta=16:13:10 +2026-09-16 01:05:14 | INFO | train rank=0 | step=34270/50000 progress=68.5% loss=0.089070 lr=2.470465e-05 lr_backbone=2.470465e-05 elapsed=35:18:47 avg_step=3.71s recent_step=3.30s eta=16:12:31 +2026-09-16 01:05:48 | INFO | train rank=0 | step=34280/50000 progress=68.6% loss=0.093492 lr=2.467613e-05 lr_backbone=2.467613e-05 elapsed=35:19:20 avg_step=3.71s recent_step=3.37s eta=16:11:52 +2026-09-16 01:06:39 | INFO | train rank=0 | step=34290/50000 progress=68.6% loss=0.089716 lr=2.464762e-05 lr_backbone=2.464762e-05 elapsed=35:20:12 avg_step=3.71s recent_step=5.19s eta=16:11:22 +2026-09-16 01:07:06 | INFO | train rank=0 | step=34300/50000 progress=68.6% loss=0.103922 lr=2.461912e-05 lr_backbone=2.461912e-05 elapsed=35:20:39 avg_step=3.71s recent_step=2.64s eta=16:10:40 +2026-09-16 01:07:35 | INFO | train rank=0 | step=34310/50000 progress=68.6% loss=0.143534 lr=2.459064e-05 lr_backbone=2.459064e-05 elapsed=35:21:08 avg_step=3.71s recent_step=2.92s eta=16:09:59 +2026-09-16 01:08:08 | INFO | train rank=0 | step=34320/50000 progress=68.6% loss=0.109231 lr=2.456216e-05 lr_backbone=2.456216e-05 elapsed=35:21:41 avg_step=3.71s recent_step=3.30s eta=16:09:20 +2026-09-16 01:09:10 | INFO | train rank=0 | step=34330/50000 progress=68.7% loss=0.078825 lr=2.453370e-05 lr_backbone=2.453370e-05 elapsed=35:22:43 avg_step=3.71s recent_step=6.19s eta=16:08:55 +2026-09-16 01:09:45 | INFO | train rank=0 | step=34340/50000 progress=68.7% loss=0.071334 lr=2.450525e-05 lr_backbone=2.450525e-05 elapsed=35:23:18 avg_step=3.71s recent_step=3.50s eta=16:08:17 +2026-09-16 01:10:21 | INFO | train rank=0 | step=34350/50000 progress=68.7% loss=0.074448 lr=2.447680e-05 lr_backbone=2.447680e-05 elapsed=35:23:54 avg_step=3.71s recent_step=3.63s eta=16:07:39 +2026-09-16 01:10:56 | INFO | train rank=0 | step=34360/50000 progress=68.7% loss=0.111062 lr=2.444837e-05 lr_backbone=2.444837e-05 elapsed=35:24:29 avg_step=3.71s recent_step=3.54s eta=16:07:01 +2026-09-16 01:11:50 | INFO | train rank=0 | step=34370/50000 progress=68.7% loss=0.093675 lr=2.441995e-05 lr_backbone=2.441995e-05 elapsed=35:25:22 avg_step=3.71s recent_step=5.32s eta=16:06:31 +2026-09-16 01:12:18 | INFO | train rank=0 | step=34380/50000 progress=68.8% loss=0.087787 lr=2.439154e-05 lr_backbone=2.439154e-05 elapsed=35:25:50 avg_step=3.71s recent_step=2.79s eta=16:05:50 +2026-09-16 01:12:53 | INFO | train rank=0 | step=34390/50000 progress=68.8% loss=0.101345 lr=2.436315e-05 lr_backbone=2.436315e-05 elapsed=35:26:25 avg_step=3.71s recent_step=3.52s eta=16:05:12 +2026-09-16 01:13:20 | INFO | train rank=0 | step=34400/50000 progress=68.8% loss=0.108739 lr=2.433476e-05 lr_backbone=2.433476e-05 elapsed=35:26:53 avg_step=3.71s recent_step=2.76s eta=16:04:31 +2026-09-16 01:14:14 | INFO | train rank=0 | step=34410/50000 progress=68.8% loss=0.100384 lr=2.430639e-05 lr_backbone=2.430639e-05 elapsed=35:27:47 avg_step=3.71s recent_step=5.40s eta=16:04:01 +2026-09-16 01:14:47 | INFO | train rank=0 | step=34420/50000 progress=68.8% loss=0.079661 lr=2.427802e-05 lr_backbone=2.427802e-05 elapsed=35:28:20 avg_step=3.71s recent_step=3.32s eta=16:03:22 +2026-09-16 01:15:23 | INFO | train rank=0 | step=34430/50000 progress=68.9% loss=0.077262 lr=2.424967e-05 lr_backbone=2.424967e-05 elapsed=35:28:56 avg_step=3.71s recent_step=3.54s eta=16:02:45 +2026-09-16 01:15:55 | INFO | train rank=0 | step=34440/50000 progress=68.9% loss=0.133634 lr=2.422133e-05 lr_backbone=2.422133e-05 elapsed=35:29:28 avg_step=3.71s recent_step=3.22s eta=16:02:05 +2026-09-16 01:16:44 | INFO | train rank=0 | step=34450/50000 progress=68.9% loss=0.133977 lr=2.419300e-05 lr_backbone=2.419300e-05 elapsed=35:30:17 avg_step=3.71s recent_step=4.92s eta=16:01:34 +2026-09-16 01:17:18 | INFO | train rank=0 | step=34460/50000 progress=68.9% loss=0.110543 lr=2.416468e-05 lr_backbone=2.416468e-05 elapsed=35:30:51 avg_step=3.71s recent_step=3.38s eta=16:00:55 +2026-09-16 01:17:52 | INFO | train rank=0 | step=34470/50000 progress=68.9% loss=0.081611 lr=2.413637e-05 lr_backbone=2.413637e-05 elapsed=35:31:25 avg_step=3.71s recent_step=3.41s eta=16:00:17 +2026-09-16 01:18:28 | INFO | train rank=0 | step=34480/50000 progress=69.0% loss=0.106844 lr=2.410808e-05 lr_backbone=2.410808e-05 elapsed=35:32:00 avg_step=3.71s recent_step=3.55s eta=15:59:39 +2026-09-16 01:19:21 | INFO | train rank=0 | step=34490/50000 progress=69.0% loss=0.087501 lr=2.407979e-05 lr_backbone=2.407979e-05 elapsed=35:32:54 avg_step=3.71s recent_step=5.31s eta=15:59:09 +2026-09-16 01:19:54 | INFO | train rank=0 | step=34500/50000 progress=69.0% loss=0.113387 lr=2.405152e-05 lr_backbone=2.405152e-05 elapsed=35:33:27 avg_step=3.71s recent_step=3.37s eta=15:58:30 +2026-09-16 01:20:25 | INFO | train rank=0 | step=34510/50000 progress=69.0% loss=0.074292 lr=2.402326e-05 lr_backbone=2.402326e-05 elapsed=35:33:58 avg_step=3.71s recent_step=3.09s eta=15:57:50 +2026-09-16 01:20:58 | INFO | train rank=0 | step=34520/50000 progress=69.0% loss=0.122109 lr=2.399501e-05 lr_backbone=2.399501e-05 elapsed=35:34:31 avg_step=3.71s recent_step=3.31s eta=15:57:11 +2026-09-16 01:21:45 | INFO | train rank=0 | step=34530/50000 progress=69.1% loss=0.073527 lr=2.396677e-05 lr_backbone=2.396677e-05 elapsed=35:35:18 avg_step=3.71s recent_step=4.67s eta=15:56:39 +2026-09-16 01:22:17 | INFO | train rank=0 | step=34540/50000 progress=69.1% loss=0.117711 lr=2.393854e-05 lr_backbone=2.393854e-05 elapsed=35:35:50 avg_step=3.71s recent_step=3.23s eta=15:55:59 +2026-09-16 01:22:58 | INFO | train rank=0 | step=34550/50000 progress=69.1% loss=0.103351 lr=2.391033e-05 lr_backbone=2.391033e-05 elapsed=35:36:31 avg_step=3.71s recent_step=4.06s eta=15:55:24 +2026-09-16 01:23:35 | INFO | train rank=0 | step=34560/50000 progress=69.1% loss=0.131922 lr=2.388212e-05 lr_backbone=2.388212e-05 elapsed=35:37:08 avg_step=3.71s recent_step=3.72s eta=15:54:47 +2026-09-16 01:24:30 | INFO | train rank=0 | step=34570/50000 progress=69.1% loss=0.098369 lr=2.385393e-05 lr_backbone=2.385393e-05 elapsed=35:38:03 avg_step=3.71s recent_step=5.48s eta=15:54:18 +2026-09-16 01:24:58 | INFO | train rank=0 | step=34580/50000 progress=69.2% loss=0.103834 lr=2.382575e-05 lr_backbone=2.382575e-05 elapsed=35:38:31 avg_step=3.71s recent_step=2.81s eta=15:53:36 +2026-09-16 01:25:32 | INFO | train rank=0 | step=34590/50000 progress=69.2% loss=0.091818 lr=2.379758e-05 lr_backbone=2.379758e-05 elapsed=35:39:05 avg_step=3.71s recent_step=3.37s eta=15:52:58 +2026-09-16 01:26:06 | INFO | train rank=0 | step=34600/50000 progress=69.2% loss=0.084007 lr=2.376942e-05 lr_backbone=2.376942e-05 elapsed=35:39:39 avg_step=3.71s recent_step=3.46s eta=15:52:20 +2026-09-16 01:26:49 | INFO | train rank=0 | step=34610/50000 progress=69.2% loss=0.092644 lr=2.374127e-05 lr_backbone=2.374127e-05 elapsed=35:40:22 avg_step=3.71s recent_step=4.30s eta=15:51:45 +2026-09-16 01:27:25 | INFO | train rank=0 | step=34620/50000 progress=69.2% loss=0.098401 lr=2.371313e-05 lr_backbone=2.371313e-05 elapsed=35:40:58 avg_step=3.71s recent_step=3.60s eta=15:51:08 +2026-09-16 01:27:53 | INFO | train rank=0 | step=34630/50000 progress=69.3% loss=0.073096 lr=2.368501e-05 lr_backbone=2.368501e-05 elapsed=35:41:25 avg_step=3.71s recent_step=2.72s eta=15:50:26 +2026-09-16 01:28:29 | INFO | train rank=0 | step=34640/50000 progress=69.3% loss=0.121574 lr=2.365689e-05 lr_backbone=2.365689e-05 elapsed=35:42:02 avg_step=3.71s recent_step=3.62s eta=15:49:49 +2026-09-16 01:29:15 | INFO | train rank=0 | step=34650/50000 progress=69.3% loss=0.087399 lr=2.362879e-05 lr_backbone=2.362879e-05 elapsed=35:42:48 avg_step=3.71s recent_step=4.66s eta=15:49:16 +2026-09-16 01:29:52 | INFO | train rank=0 | step=34660/50000 progress=69.3% loss=0.077435 lr=2.360070e-05 lr_backbone=2.360070e-05 elapsed=35:43:25 avg_step=3.71s recent_step=3.66s eta=15:48:38 +2026-09-16 01:30:25 | INFO | train rank=0 | step=34670/50000 progress=69.3% loss=0.106930 lr=2.357262e-05 lr_backbone=2.357262e-05 elapsed=35:43:58 avg_step=3.71s recent_step=3.33s eta=15:48:00 +2026-09-16 01:30:58 | INFO | train rank=0 | step=34680/50000 progress=69.4% loss=0.125027 lr=2.354456e-05 lr_backbone=2.354456e-05 elapsed=35:44:31 avg_step=3.71s recent_step=3.28s eta=15:47:21 +2026-09-16 01:31:43 | INFO | train rank=0 | step=34690/50000 progress=69.4% loss=0.102928 lr=2.351650e-05 lr_backbone=2.351650e-05 elapsed=35:45:16 avg_step=3.71s recent_step=4.51s eta=15:46:47 +2026-09-16 01:32:19 | INFO | train rank=0 | step=34700/50000 progress=69.4% loss=0.085928 lr=2.348846e-05 lr_backbone=2.348846e-05 elapsed=35:45:52 avg_step=3.71s recent_step=3.56s eta=15:46:09 +2026-09-16 01:32:55 | INFO | train rank=0 | step=34710/50000 progress=69.4% loss=0.103049 lr=2.346043e-05 lr_backbone=2.346043e-05 elapsed=35:46:28 avg_step=3.71s recent_step=3.63s eta=15:45:32 +2026-09-16 01:33:27 | INFO | train rank=0 | step=34720/50000 progress=69.4% loss=0.097875 lr=2.343241e-05 lr_backbone=2.343241e-05 elapsed=35:47:00 avg_step=3.71s recent_step=3.22s eta=15:44:52 +2026-09-16 01:34:15 | INFO | train rank=0 | step=34730/50000 progress=69.5% loss=0.097400 lr=2.340440e-05 lr_backbone=2.340440e-05 elapsed=35:47:48 avg_step=3.71s recent_step=4.81s eta=15:44:20 +2026-09-16 01:34:52 | INFO | train rank=0 | step=34740/50000 progress=69.5% loss=0.073736 lr=2.337640e-05 lr_backbone=2.337640e-05 elapsed=35:48:25 avg_step=3.71s recent_step=3.69s eta=15:43:43 +2026-09-16 01:35:30 | INFO | train rank=0 | step=34750/50000 progress=69.5% loss=0.108894 lr=2.334841e-05 lr_backbone=2.334841e-05 elapsed=35:49:03 avg_step=3.71s recent_step=3.76s eta=15:43:06 +2026-09-16 01:36:05 | INFO | train rank=0 | step=34760/50000 progress=69.5% loss=0.103434 lr=2.332044e-05 lr_backbone=2.332044e-05 elapsed=35:49:38 avg_step=3.71s recent_step=3.51s eta=15:42:28 +2026-09-16 01:36:56 | INFO | train rank=0 | step=34770/50000 progress=69.5% loss=0.079757 lr=2.329248e-05 lr_backbone=2.329248e-05 elapsed=35:50:29 avg_step=3.71s recent_step=5.07s eta=15:41:57 +2026-09-16 01:37:33 | INFO | train rank=0 | step=34780/50000 progress=69.6% loss=0.122417 lr=2.326453e-05 lr_backbone=2.326453e-05 elapsed=35:51:05 avg_step=3.71s recent_step=3.69s eta=15:41:20 +2026-09-16 01:38:04 | INFO | train rank=0 | step=34790/50000 progress=69.6% loss=0.075834 lr=2.323659e-05 lr_backbone=2.323659e-05 elapsed=35:51:37 avg_step=3.71s recent_step=3.16s eta=15:40:40 +2026-09-16 01:38:42 | INFO | train rank=0 | step=34800/50000 progress=69.6% loss=0.096959 lr=2.320866e-05 lr_backbone=2.320866e-05 elapsed=35:52:14 avg_step=3.71s recent_step=3.74s eta=15:40:03 +2026-09-16 01:39:24 | INFO | train rank=0 | step=34810/50000 progress=69.6% loss=0.086970 lr=2.318074e-05 lr_backbone=2.318074e-05 elapsed=35:52:56 avg_step=3.71s recent_step=4.19s eta=15:39:28 +2026-09-16 01:39:59 | INFO | train rank=0 | step=34820/50000 progress=69.6% loss=0.092427 lr=2.315284e-05 lr_backbone=2.315284e-05 elapsed=35:53:31 avg_step=3.71s recent_step=3.50s eta=15:38:50 +2026-09-16 01:40:26 | INFO | train rank=0 | step=34830/50000 progress=69.7% loss=0.084749 lr=2.312495e-05 lr_backbone=2.312495e-05 elapsed=35:53:58 avg_step=3.71s recent_step=2.70s eta=15:38:09 +2026-09-16 01:41:00 | INFO | train rank=0 | step=34840/50000 progress=69.7% loss=0.110507 lr=2.309707e-05 lr_backbone=2.309707e-05 elapsed=35:54:33 avg_step=3.71s recent_step=3.48s eta=15:37:31 +2026-09-16 01:41:44 | INFO | train rank=0 | step=34850/50000 progress=69.7% loss=0.106672 lr=2.306920e-05 lr_backbone=2.306920e-05 elapsed=35:55:17 avg_step=3.71s recent_step=4.39s eta=15:36:56 +2026-09-16 01:42:16 | INFO | train rank=0 | step=34860/50000 progress=69.7% loss=0.092941 lr=2.304134e-05 lr_backbone=2.304134e-05 elapsed=35:55:49 avg_step=3.71s recent_step=3.22s eta=15:36:17 +2026-09-16 01:42:47 | INFO | train rank=0 | step=34870/50000 progress=69.7% loss=0.089327 lr=2.301350e-05 lr_backbone=2.301350e-05 elapsed=35:56:20 avg_step=3.71s recent_step=3.05s eta=15:35:37 +2026-09-16 01:43:20 | INFO | train rank=0 | step=34880/50000 progress=69.8% loss=0.097478 lr=2.298566e-05 lr_backbone=2.298566e-05 elapsed=35:56:53 avg_step=3.71s recent_step=3.35s eta=15:34:59 +2026-09-16 01:44:05 | INFO | train rank=0 | step=34890/50000 progress=69.8% loss=0.117278 lr=2.295784e-05 lr_backbone=2.295784e-05 elapsed=35:57:38 avg_step=3.71s recent_step=4.50s eta=15:34:25 +2026-09-16 01:44:41 | INFO | train rank=0 | step=34900/50000 progress=69.8% loss=0.097730 lr=2.293003e-05 lr_backbone=2.293003e-05 elapsed=35:58:14 avg_step=3.71s recent_step=3.57s eta=15:33:47 +2026-09-16 01:45:14 | INFO | train rank=0 | step=34910/50000 progress=69.8% loss=0.110193 lr=2.290224e-05 lr_backbone=2.290224e-05 elapsed=35:58:47 avg_step=3.71s recent_step=3.27s eta=15:33:08 +2026-09-16 01:45:52 | INFO | train rank=0 | step=34920/50000 progress=69.8% loss=0.099050 lr=2.287445e-05 lr_backbone=2.287445e-05 elapsed=35:59:25 avg_step=3.71s recent_step=3.86s eta=15:32:32 +2026-09-16 01:46:48 | INFO | train rank=0 | step=34930/50000 progress=69.9% loss=0.101256 lr=2.284668e-05 lr_backbone=2.284668e-05 elapsed=36:00:21 avg_step=3.71s recent_step=5.59s eta=15:32:03 +2026-09-16 01:47:17 | INFO | train rank=0 | step=34940/50000 progress=69.9% loss=0.121075 lr=2.281891e-05 lr_backbone=2.281891e-05 elapsed=36:00:50 avg_step=3.71s recent_step=2.86s eta=15:31:22 +2026-09-16 01:47:53 | INFO | train rank=0 | step=34950/50000 progress=69.9% loss=0.070603 lr=2.279116e-05 lr_backbone=2.279116e-05 elapsed=36:01:25 avg_step=3.71s recent_step=3.56s eta=15:30:44 +2026-09-16 01:48:13 | INFO | train rank=0 | step=34960/50000 progress=69.9% loss=0.121486 lr=2.276343e-05 lr_backbone=2.276343e-05 elapsed=36:01:46 avg_step=3.71s recent_step=2.08s eta=15:30:00 +2026-09-16 01:48:52 | INFO | train rank=0 | step=34970/50000 progress=69.9% loss=0.106509 lr=2.273570e-05 lr_backbone=2.273570e-05 elapsed=36:02:25 avg_step=3.71s recent_step=3.90s eta=15:29:24 +2026-09-16 01:49:28 | INFO | train rank=0 | step=34980/50000 progress=70.0% loss=0.077656 lr=2.270799e-05 lr_backbone=2.270799e-05 elapsed=36:03:01 avg_step=3.71s recent_step=3.61s eta=15:28:46 +2026-09-16 01:49:59 | INFO | train rank=0 | step=34990/50000 progress=70.0% loss=0.101757 lr=2.268028e-05 lr_backbone=2.268028e-05 elapsed=36:03:32 avg_step=3.71s recent_step=3.05s eta=15:28:06 +2026-09-16 01:50:32 | INFO | train rank=0 | step=35000/50000 progress=70.0% loss=0.099761 lr=2.265259e-05 lr_backbone=2.265259e-05 elapsed=36:04:04 avg_step=3.71s recent_step=3.26s eta=15:27:27 +2026-09-16 01:51:29 | INFO | train rank=0 | step=35010/50000 progress=70.0% loss=0.083437 lr=2.262491e-05 lr_backbone=2.262491e-05 elapsed=36:05:02 avg_step=3.71s recent_step=5.72s eta=15:26:59 +2026-09-16 01:51:59 | INFO | train rank=0 | step=35020/50000 progress=70.0% loss=0.080412 lr=2.259725e-05 lr_backbone=2.259725e-05 elapsed=36:05:32 avg_step=3.71s recent_step=3.05s eta=15:26:19 +2026-09-16 01:52:35 | INFO | train rank=0 | step=35030/50000 progress=70.1% loss=0.104970 lr=2.256959e-05 lr_backbone=2.256959e-05 elapsed=36:06:07 avg_step=3.71s recent_step=3.54s eta=15:25:41 +2026-09-16 01:53:11 | INFO | train rank=0 | step=35040/50000 progress=70.1% loss=0.074893 lr=2.254195e-05 lr_backbone=2.254195e-05 elapsed=36:06:44 avg_step=3.71s recent_step=3.63s eta=15:25:04 +2026-09-16 01:53:59 | INFO | train rank=0 | step=35050/50000 progress=70.1% loss=0.083908 lr=2.251432e-05 lr_backbone=2.251432e-05 elapsed=36:07:32 avg_step=3.71s recent_step=4.85s eta=15:24:31 +2026-09-16 01:54:32 | INFO | train rank=0 | step=35060/50000 progress=70.1% loss=0.121706 lr=2.248670e-05 lr_backbone=2.248670e-05 elapsed=36:08:04 avg_step=3.71s recent_step=3.23s eta=15:23:52 +2026-09-16 01:55:08 | INFO | train rank=0 | step=35070/50000 progress=70.1% loss=0.146035 lr=2.245909e-05 lr_backbone=2.245909e-05 elapsed=36:08:41 avg_step=3.71s recent_step=3.67s eta=15:23:15 +2026-09-16 01:55:42 | INFO | train rank=0 | step=35080/50000 progress=70.2% loss=0.109315 lr=2.243150e-05 lr_backbone=2.243150e-05 elapsed=36:09:15 avg_step=3.71s recent_step=3.40s eta=15:22:36 +2026-09-16 01:56:31 | INFO | train rank=0 | step=35090/50000 progress=70.2% loss=0.107595 lr=2.240392e-05 lr_backbone=2.240392e-05 elapsed=36:10:04 avg_step=3.71s recent_step=4.92s eta=15:22:04 +2026-09-16 01:57:09 | INFO | train rank=0 | step=35100/50000 progress=70.2% loss=0.082752 lr=2.237635e-05 lr_backbone=2.237635e-05 elapsed=36:10:41 avg_step=3.71s recent_step=3.72s eta=15:21:27 +2026-09-16 01:57:30 | INFO | train rank=0 | step=35110/50000 progress=70.2% loss=0.068020 lr=2.234879e-05 lr_backbone=2.234879e-05 elapsed=36:11:03 avg_step=3.71s recent_step=2.13s eta=15:20:44 +2026-09-16 01:58:13 | INFO | train rank=0 | step=35120/50000 progress=70.2% loss=0.067034 lr=2.232124e-05 lr_backbone=2.232124e-05 elapsed=36:11:46 avg_step=3.71s recent_step=4.31s eta=15:20:09 +2026-09-16 01:59:00 | INFO | train rank=0 | step=35130/50000 progress=70.3% loss=0.078970 lr=2.229371e-05 lr_backbone=2.229371e-05 elapsed=36:12:33 avg_step=3.71s recent_step=4.69s eta=15:19:36 +2026-09-16 01:59:31 | INFO | train rank=0 | step=35140/50000 progress=70.3% loss=0.082151 lr=2.226619e-05 lr_backbone=2.226619e-05 elapsed=36:13:04 avg_step=3.71s recent_step=3.14s eta=15:18:57 +2026-09-16 02:00:11 | INFO | train rank=0 | step=35150/50000 progress=70.3% loss=0.104759 lr=2.223868e-05 lr_backbone=2.223868e-05 elapsed=36:13:44 avg_step=3.71s recent_step=3.96s eta=15:18:20 +2026-09-16 02:00:50 | INFO | train rank=0 | step=35160/50000 progress=70.3% loss=0.105832 lr=2.221118e-05 lr_backbone=2.221118e-05 elapsed=36:14:23 avg_step=3.71s recent_step=3.88s eta=15:17:44 +2026-09-16 02:01:38 | INFO | train rank=0 | step=35170/50000 progress=70.3% loss=0.112934 lr=2.218369e-05 lr_backbone=2.218369e-05 elapsed=36:15:11 avg_step=3.71s recent_step=4.81s eta=15:17:12 +2026-09-16 02:02:25 | INFO | train rank=0 | step=35180/50000 progress=70.4% loss=0.096070 lr=2.215622e-05 lr_backbone=2.215622e-05 elapsed=36:15:58 avg_step=3.71s recent_step=4.69s eta=15:16:39 +2026-09-16 02:02:58 | INFO | train rank=0 | step=35190/50000 progress=70.4% loss=0.120373 lr=2.212876e-05 lr_backbone=2.212876e-05 elapsed=36:16:30 avg_step=3.71s recent_step=3.28s eta=15:16:00 +2026-09-16 02:03:30 | INFO | train rank=0 | step=35200/50000 progress=70.4% loss=0.079724 lr=2.210131e-05 lr_backbone=2.210131e-05 elapsed=36:17:03 avg_step=3.71s recent_step=3.24s eta=15:15:21 +2026-09-16 02:04:13 | INFO | train rank=0 | step=35210/50000 progress=70.4% loss=0.069934 lr=2.207387e-05 lr_backbone=2.207387e-05 elapsed=36:17:46 avg_step=3.71s recent_step=4.34s eta=15:14:46 +2026-09-16 02:04:45 | INFO | train rank=0 | step=35220/50000 progress=70.4% loss=0.089010 lr=2.204645e-05 lr_backbone=2.204645e-05 elapsed=36:18:18 avg_step=3.71s recent_step=3.15s eta=15:14:07 +2026-09-16 02:05:18 | INFO | train rank=0 | step=35230/50000 progress=70.5% loss=0.097069 lr=2.201904e-05 lr_backbone=2.201904e-05 elapsed=36:18:51 avg_step=3.71s recent_step=3.31s eta=15:13:28 +2026-09-16 02:05:55 | INFO | train rank=0 | step=35240/50000 progress=70.5% loss=0.104944 lr=2.199164e-05 lr_backbone=2.199164e-05 elapsed=36:19:28 avg_step=3.71s recent_step=3.70s eta=15:12:51 +2026-09-16 02:06:43 | INFO | train rank=0 | step=35250/50000 progress=70.5% loss=0.087684 lr=2.196425e-05 lr_backbone=2.196425e-05 elapsed=36:20:16 avg_step=3.71s recent_step=4.81s eta=15:12:18 +2026-09-16 02:07:13 | INFO | train rank=0 | step=35260/50000 progress=70.5% loss=0.101550 lr=2.193687e-05 lr_backbone=2.193687e-05 elapsed=36:20:46 avg_step=3.71s recent_step=2.99s eta=15:11:38 +2026-09-16 02:07:47 | INFO | train rank=0 | step=35270/50000 progress=70.5% loss=0.097353 lr=2.190951e-05 lr_backbone=2.190951e-05 elapsed=36:21:20 avg_step=3.71s recent_step=3.40s eta=15:11:00 +2026-09-16 02:08:18 | INFO | train rank=0 | step=35280/50000 progress=70.6% loss=0.099019 lr=2.188216e-05 lr_backbone=2.188216e-05 elapsed=36:21:50 avg_step=3.71s recent_step=3.06s eta=15:10:20 +2026-09-16 02:09:10 | INFO | train rank=0 | step=35290/50000 progress=70.6% loss=0.070582 lr=2.185482e-05 lr_backbone=2.185482e-05 elapsed=36:22:43 avg_step=3.71s recent_step=5.23s eta=15:09:49 +2026-09-16 02:09:45 | INFO | train rank=0 | step=35300/50000 progress=70.6% loss=0.088671 lr=2.182749e-05 lr_backbone=2.182749e-05 elapsed=36:23:18 avg_step=3.71s recent_step=3.50s eta=15:09:11 +2026-09-16 02:10:24 | INFO | train rank=0 | step=35310/50000 progress=70.6% loss=0.115370 lr=2.180018e-05 lr_backbone=2.180018e-05 elapsed=36:23:57 avg_step=3.71s recent_step=3.90s eta=15:08:35 +2026-09-16 02:10:59 | INFO | train rank=0 | step=35320/50000 progress=70.6% loss=0.099066 lr=2.177288e-05 lr_backbone=2.177288e-05 elapsed=36:24:31 avg_step=3.71s recent_step=3.48s eta=15:07:57 +2026-09-16 02:11:48 | INFO | train rank=0 | step=35330/50000 progress=70.7% loss=0.105495 lr=2.174559e-05 lr_backbone=2.174559e-05 elapsed=36:25:21 avg_step=3.71s recent_step=4.95s eta=15:07:25 +2026-09-16 02:12:23 | INFO | train rank=0 | step=35340/50000 progress=70.7% loss=0.055762 lr=2.171831e-05 lr_backbone=2.171831e-05 elapsed=36:25:56 avg_step=3.71s recent_step=3.45s eta=15:06:47 +2026-09-16 02:12:53 | INFO | train rank=0 | step=35350/50000 progress=70.7% loss=0.111405 lr=2.169104e-05 lr_backbone=2.169104e-05 elapsed=36:26:25 avg_step=3.71s recent_step=2.98s eta=15:06:06 +2026-09-16 02:13:24 | INFO | train rank=0 | step=35360/50000 progress=70.7% loss=0.135894 lr=2.166379e-05 lr_backbone=2.166379e-05 elapsed=36:26:57 avg_step=3.71s recent_step=3.13s eta=15:05:27 +2026-09-16 02:14:11 | INFO | train rank=0 | step=35370/50000 progress=70.7% loss=0.141794 lr=2.163655e-05 lr_backbone=2.163655e-05 elapsed=36:27:44 avg_step=3.71s recent_step=4.75s eta=15:04:54 +2026-09-16 02:14:44 | INFO | train rank=0 | step=35380/50000 progress=70.8% loss=0.092963 lr=2.160933e-05 lr_backbone=2.160933e-05 elapsed=36:28:17 avg_step=3.71s recent_step=3.28s eta=15:04:15 +2026-09-16 02:15:19 | INFO | train rank=0 | step=35390/50000 progress=70.8% loss=0.083333 lr=2.158211e-05 lr_backbone=2.158211e-05 elapsed=36:28:52 avg_step=3.71s recent_step=3.46s eta=15:03:37 +2026-09-16 02:15:51 | INFO | train rank=0 | step=35400/50000 progress=70.8% loss=0.108925 lr=2.155491e-05 lr_backbone=2.155491e-05 elapsed=36:29:24 avg_step=3.71s recent_step=3.23s eta=15:02:58 +2026-09-16 02:16:37 | INFO | train rank=0 | step=35410/50000 progress=70.8% loss=0.076426 lr=2.152772e-05 lr_backbone=2.152772e-05 elapsed=36:30:10 avg_step=3.71s recent_step=4.59s eta=15:02:25 +2026-09-16 02:17:17 | INFO | train rank=0 | step=35420/50000 progress=70.8% loss=0.093657 lr=2.150054e-05 lr_backbone=2.150054e-05 elapsed=36:30:50 avg_step=3.71s recent_step=3.99s eta=15:01:49 +2026-09-16 02:17:51 | INFO | train rank=0 | step=35430/50000 progress=70.9% loss=0.118031 lr=2.147337e-05 lr_backbone=2.147337e-05 elapsed=36:31:24 avg_step=3.71s recent_step=3.40s eta=15:01:10 +2026-09-16 02:18:26 | INFO | train rank=0 | step=35440/50000 progress=70.9% loss=0.080043 lr=2.144622e-05 lr_backbone=2.144622e-05 elapsed=36:31:59 avg_step=3.71s recent_step=3.50s eta=15:00:32 +2026-09-16 02:19:10 | INFO | train rank=0 | step=35450/50000 progress=70.9% loss=0.079852 lr=2.141908e-05 lr_backbone=2.141908e-05 elapsed=36:32:43 avg_step=3.71s recent_step=4.40s eta=14:59:58 +2026-09-16 02:19:34 | INFO | train rank=0 | step=35460/50000 progress=70.9% loss=0.079636 lr=2.139195e-05 lr_backbone=2.139195e-05 elapsed=36:33:07 avg_step=3.71s recent_step=2.41s eta=14:59:15 +2026-09-16 02:20:09 | INFO | train rank=0 | step=35470/50000 progress=70.9% loss=0.086695 lr=2.136484e-05 lr_backbone=2.136484e-05 elapsed=36:33:42 avg_step=3.71s recent_step=3.50s eta=14:58:37 +2026-09-16 02:20:42 | INFO | train rank=0 | step=35480/50000 progress=71.0% loss=0.078822 lr=2.133773e-05 lr_backbone=2.133773e-05 elapsed=36:34:15 avg_step=3.71s recent_step=3.34s eta=14:57:59 +2026-09-16 02:21:39 | INFO | train rank=0 | step=35490/50000 progress=71.0% loss=0.106157 lr=2.131064e-05 lr_backbone=2.131064e-05 elapsed=36:35:12 avg_step=3.71s recent_step=5.65s eta=14:57:30 +2026-09-16 02:22:10 | INFO | train rank=0 | step=35500/50000 progress=71.0% loss=0.090130 lr=2.128357e-05 lr_backbone=2.128357e-05 elapsed=36:35:43 avg_step=3.71s recent_step=3.13s eta=14:56:50 +2026-09-16 02:22:37 | INFO | train rank=0 | step=35510/50000 progress=71.0% loss=0.084131 lr=2.125650e-05 lr_backbone=2.125650e-05 elapsed=36:36:09 avg_step=3.71s recent_step=2.63s eta=14:56:09 +2026-09-16 02:23:09 | INFO | train rank=0 | step=35520/50000 progress=71.0% loss=0.075949 lr=2.122945e-05 lr_backbone=2.122945e-05 elapsed=36:36:42 avg_step=3.71s recent_step=3.22s eta=14:55:30 +2026-09-16 02:23:58 | INFO | train rank=0 | step=35530/50000 progress=71.1% loss=0.091115 lr=2.120241e-05 lr_backbone=2.120241e-05 elapsed=36:37:30 avg_step=3.71s recent_step=4.87s eta=14:54:57 +2026-09-16 02:24:30 | INFO | train rank=0 | step=35540/50000 progress=71.1% loss=0.074616 lr=2.117538e-05 lr_backbone=2.117538e-05 elapsed=36:38:03 avg_step=3.71s recent_step=3.27s eta=14:54:18 +2026-09-16 02:25:05 | INFO | train rank=0 | step=35550/50000 progress=71.1% loss=0.107327 lr=2.114837e-05 lr_backbone=2.114837e-05 elapsed=36:38:38 avg_step=3.71s recent_step=3.53s eta=14:53:40 +2026-09-16 02:25:42 | INFO | train rank=0 | step=35560/50000 progress=71.1% loss=0.104040 lr=2.112136e-05 lr_backbone=2.112136e-05 elapsed=36:39:15 avg_step=3.71s recent_step=3.64s eta=14:53:03 +2026-09-16 02:26:32 | INFO | train rank=0 | step=35570/50000 progress=71.1% loss=0.074803 lr=2.109438e-05 lr_backbone=2.109438e-05 elapsed=36:40:05 avg_step=3.71s recent_step=4.98s eta=14:52:31 +2026-09-16 02:26:59 | INFO | train rank=0 | step=35580/50000 progress=71.2% loss=0.112351 lr=2.106740e-05 lr_backbone=2.106740e-05 elapsed=36:40:31 avg_step=3.71s recent_step=2.69s eta=14:51:50 +2026-09-16 02:27:24 | INFO | train rank=0 | step=35590/50000 progress=71.2% loss=0.125392 lr=2.104043e-05 lr_backbone=2.104043e-05 elapsed=36:40:57 avg_step=3.71s recent_step=2.52s eta=14:51:08 +2026-09-16 02:27:56 | INFO | train rank=0 | step=35600/50000 progress=71.2% loss=0.072044 lr=2.101348e-05 lr_backbone=2.101348e-05 elapsed=36:41:29 avg_step=3.71s recent_step=3.21s eta=14:50:29 +2026-09-16 02:28:54 | INFO | train rank=0 | step=35610/50000 progress=71.2% loss=0.087561 lr=2.098654e-05 lr_backbone=2.098654e-05 elapsed=36:42:27 avg_step=3.71s recent_step=5.79s eta=14:50:00 +2026-09-16 02:29:21 | INFO | train rank=0 | step=35620/50000 progress=71.2% loss=0.104183 lr=2.095962e-05 lr_backbone=2.095962e-05 elapsed=36:42:54 avg_step=3.71s recent_step=2.70s eta=14:49:19 +2026-09-16 02:29:46 | INFO | train rank=0 | step=35630/50000 progress=71.3% loss=0.066032 lr=2.093270e-05 lr_backbone=2.093270e-05 elapsed=36:43:18 avg_step=3.71s recent_step=2.49s eta=14:48:37 +2026-09-16 02:30:13 | INFO | train rank=0 | step=35640/50000 progress=71.3% loss=0.070977 lr=2.090580e-05 lr_backbone=2.090580e-05 elapsed=36:43:46 avg_step=3.71s recent_step=2.75s eta=14:47:56 +2026-09-16 02:31:05 | INFO | train rank=0 | step=35650/50000 progress=71.3% loss=0.098480 lr=2.087892e-05 lr_backbone=2.087892e-05 elapsed=36:44:38 avg_step=3.71s recent_step=5.16s eta=14:47:25 +2026-09-16 02:31:40 | INFO | train rank=0 | step=35660/50000 progress=71.3% loss=0.103105 lr=2.085204e-05 lr_backbone=2.085204e-05 elapsed=36:45:12 avg_step=3.71s recent_step=3.47s eta=14:46:47 +2026-09-16 02:32:12 | INFO | train rank=0 | step=35670/50000 progress=71.3% loss=0.099793 lr=2.082518e-05 lr_backbone=2.082518e-05 elapsed=36:45:44 avg_step=3.71s recent_step=3.21s eta=14:46:08 +2026-09-16 02:32:53 | INFO | train rank=0 | step=35680/50000 progress=71.4% loss=0.063881 lr=2.079833e-05 lr_backbone=2.079833e-05 elapsed=36:46:26 avg_step=3.71s recent_step=4.17s eta=14:45:32 +2026-09-16 02:33:35 | INFO | train rank=0 | step=35690/50000 progress=71.4% loss=0.083294 lr=2.077149e-05 lr_backbone=2.077149e-05 elapsed=36:47:08 avg_step=3.71s recent_step=4.19s eta=14:44:57 +2026-09-16 02:34:09 | INFO | train rank=0 | step=35700/50000 progress=71.4% loss=0.085007 lr=2.074467e-05 lr_backbone=2.074467e-05 elapsed=36:47:42 avg_step=3.71s recent_step=3.36s eta=14:44:19 +2026-09-16 02:34:40 | INFO | train rank=0 | step=35710/50000 progress=71.4% loss=0.104867 lr=2.071785e-05 lr_backbone=2.071785e-05 elapsed=36:48:13 avg_step=3.71s recent_step=3.11s eta=14:43:39 +2026-09-16 02:35:21 | INFO | train rank=0 | step=35720/50000 progress=71.4% loss=0.106225 lr=2.069106e-05 lr_backbone=2.069106e-05 elapsed=36:48:54 avg_step=3.71s recent_step=4.15s eta=14:43:04 +2026-09-16 02:36:00 | INFO | train rank=0 | step=35730/50000 progress=71.5% loss=0.082797 lr=2.066427e-05 lr_backbone=2.066427e-05 elapsed=36:49:33 avg_step=3.71s recent_step=3.91s eta=14:42:27 +2026-09-16 02:36:33 | INFO | train rank=0 | step=35740/50000 progress=71.5% loss=0.097593 lr=2.063750e-05 lr_backbone=2.063750e-05 elapsed=36:50:06 avg_step=3.71s recent_step=3.30s eta=14:41:49 +2026-09-16 02:37:08 | INFO | train rank=0 | step=35750/50000 progress=71.5% loss=0.072407 lr=2.061074e-05 lr_backbone=2.061074e-05 elapsed=36:50:41 avg_step=3.71s recent_step=3.46s eta=14:41:11 +2026-09-16 02:37:39 | INFO | train rank=0 | step=35760/50000 progress=71.5% loss=0.098424 lr=2.058399e-05 lr_backbone=2.058399e-05 elapsed=36:51:12 avg_step=3.71s recent_step=3.08s eta=14:40:31 +2026-09-16 02:38:20 | INFO | train rank=0 | step=35770/50000 progress=71.5% loss=0.082270 lr=2.055726e-05 lr_backbone=2.055726e-05 elapsed=36:51:52 avg_step=3.71s recent_step=4.09s eta=14:39:55 +2026-09-16 02:39:05 | INFO | train rank=0 | step=35780/50000 progress=71.6% loss=0.072628 lr=2.053053e-05 lr_backbone=2.053053e-05 elapsed=36:52:38 avg_step=3.71s recent_step=4.52s eta=14:39:21 +2026-09-16 02:39:41 | INFO | train rank=0 | step=35790/50000 progress=71.6% loss=0.080629 lr=2.050383e-05 lr_backbone=2.050383e-05 elapsed=36:53:14 avg_step=3.71s recent_step=3.59s eta=14:38:44 +2026-09-16 02:40:21 | INFO | train rank=0 | step=35800/50000 progress=71.6% loss=0.094727 lr=2.047713e-05 lr_backbone=2.047713e-05 elapsed=36:53:53 avg_step=3.71s recent_step=3.98s eta=14:38:08 +2026-09-16 02:41:16 | INFO | train rank=0 | step=35810/50000 progress=71.6% loss=0.077506 lr=2.045045e-05 lr_backbone=2.045045e-05 elapsed=36:54:49 avg_step=3.71s recent_step=5.52s eta=14:37:38 +2026-09-16 02:41:43 | INFO | train rank=0 | step=35820/50000 progress=71.6% loss=0.089748 lr=2.042378e-05 lr_backbone=2.042378e-05 elapsed=36:55:15 avg_step=3.71s recent_step=2.69s eta=14:36:57 +2026-09-16 02:42:19 | INFO | train rank=0 | step=35830/50000 progress=71.7% loss=0.063308 lr=2.039712e-05 lr_backbone=2.039712e-05 elapsed=36:55:52 avg_step=3.71s recent_step=3.66s eta=14:36:19 +2026-09-16 02:43:00 | INFO | train rank=0 | step=35840/50000 progress=71.7% loss=0.139613 lr=2.037048e-05 lr_backbone=2.037048e-05 elapsed=36:56:33 avg_step=3.71s recent_step=4.10s eta=14:35:44 +2026-09-16 02:43:47 | INFO | train rank=0 | step=35850/50000 progress=71.7% loss=0.099731 lr=2.034385e-05 lr_backbone=2.034385e-05 elapsed=36:57:20 avg_step=3.71s recent_step=4.66s eta=14:35:10 +2026-09-16 02:44:19 | INFO | train rank=0 | step=35860/50000 progress=71.7% loss=0.118720 lr=2.031723e-05 lr_backbone=2.031723e-05 elapsed=36:57:52 avg_step=3.71s recent_step=3.24s eta=14:34:32 +2026-09-16 02:44:53 | INFO | train rank=0 | step=35870/50000 progress=71.7% loss=0.085130 lr=2.029062e-05 lr_backbone=2.029062e-05 elapsed=36:58:26 avg_step=3.71s recent_step=3.36s eta=14:33:53 +2026-09-16 02:45:29 | INFO | train rank=0 | step=35880/50000 progress=71.8% loss=0.103497 lr=2.026403e-05 lr_backbone=2.026403e-05 elapsed=36:59:02 avg_step=3.71s recent_step=3.58s eta=14:33:15 +2026-09-16 02:46:19 | INFO | train rank=0 | step=35890/50000 progress=71.8% loss=0.111482 lr=2.023745e-05 lr_backbone=2.023745e-05 elapsed=36:59:52 avg_step=3.71s recent_step=5.02s eta=14:32:43 +2026-09-16 02:47:03 | INFO | train rank=0 | step=35900/50000 progress=71.8% loss=0.076222 lr=2.021088e-05 lr_backbone=2.021088e-05 elapsed=37:00:36 avg_step=3.71s recent_step=4.38s eta=14:32:09 +2026-09-16 02:47:39 | INFO | train rank=0 | step=35910/50000 progress=71.8% loss=0.129108 lr=2.018433e-05 lr_backbone=2.018433e-05 elapsed=37:01:12 avg_step=3.71s recent_step=3.63s eta=14:31:32 +2026-09-16 02:48:04 | INFO | train rank=0 | step=35920/50000 progress=71.8% loss=0.118007 lr=2.015779e-05 lr_backbone=2.015779e-05 elapsed=37:01:37 avg_step=3.71s recent_step=2.51s eta=14:30:50 +2026-09-16 02:48:54 | INFO | train rank=0 | step=35930/50000 progress=71.9% loss=0.107014 lr=2.013126e-05 lr_backbone=2.013126e-05 elapsed=37:02:27 avg_step=3.71s recent_step=4.97s eta=14:30:18 +2026-09-16 02:49:23 | INFO | train rank=0 | step=35940/50000 progress=71.9% loss=0.092754 lr=2.010475e-05 lr_backbone=2.010475e-05 elapsed=37:02:55 avg_step=3.71s recent_step=2.87s eta=14:29:37 +2026-09-16 02:49:51 | INFO | train rank=0 | step=35950/50000 progress=71.9% loss=0.092771 lr=2.007825e-05 lr_backbone=2.007825e-05 elapsed=37:03:23 avg_step=3.71s recent_step=2.81s eta=14:28:57 +2026-09-16 02:50:27 | INFO | train rank=0 | step=35960/50000 progress=71.9% loss=0.116267 lr=2.005176e-05 lr_backbone=2.005176e-05 elapsed=37:03:59 avg_step=3.71s recent_step=3.60s eta=14:28:19 +2026-09-16 02:51:16 | INFO | train rank=0 | step=35970/50000 progress=71.9% loss=0.117348 lr=2.002529e-05 lr_backbone=2.002529e-05 elapsed=37:04:48 avg_step=3.71s recent_step=4.90s eta=14:27:47 +2026-09-16 02:51:55 | INFO | train rank=0 | step=35980/50000 progress=72.0% loss=0.088482 lr=1.999883e-05 lr_backbone=1.999883e-05 elapsed=37:05:28 avg_step=3.71s recent_step=3.93s eta=14:27:10 +2026-09-16 02:52:29 | INFO | train rank=0 | step=35990/50000 progress=72.0% loss=0.099656 lr=1.997238e-05 lr_backbone=1.997238e-05 elapsed=37:06:02 avg_step=3.71s recent_step=3.43s eta=14:26:32 +2026-09-16 02:53:04 | INFO | train rank=0 | step=36000/50000 progress=72.0% loss=0.085157 lr=1.994594e-05 lr_backbone=1.994594e-05 elapsed=37:06:37 avg_step=3.71s recent_step=3.46s eta=14:25:54 +2026-09-16 02:53:46 | INFO | train rank=0 | step=36010/50000 progress=72.0% loss=0.090336 lr=1.991952e-05 lr_backbone=1.991952e-05 elapsed=37:07:19 avg_step=3.71s recent_step=4.19s eta=14:25:19 +2026-09-16 02:54:24 | INFO | train rank=0 | step=36020/50000 progress=72.0% loss=0.129904 lr=1.989311e-05 lr_backbone=1.989311e-05 elapsed=37:07:57 avg_step=3.71s recent_step=3.83s eta=14:24:42 +2026-09-16 02:54:59 | INFO | train rank=0 | step=36030/50000 progress=72.1% loss=0.126833 lr=1.986672e-05 lr_backbone=1.986672e-05 elapsed=37:08:31 avg_step=3.71s recent_step=3.45s eta=14:24:04 +2026-09-16 02:55:36 | INFO | train rank=0 | step=36040/50000 progress=72.1% loss=0.085888 lr=1.984033e-05 lr_backbone=1.984033e-05 elapsed=37:09:09 avg_step=3.71s recent_step=3.71s eta=14:23:27 +2026-09-16 02:56:25 | INFO | train rank=0 | step=36050/50000 progress=72.1% loss=0.100155 lr=1.981396e-05 lr_backbone=1.981396e-05 elapsed=37:09:58 avg_step=3.71s recent_step=4.91s eta=14:22:54 +2026-09-16 02:56:53 | INFO | train rank=0 | step=36060/50000 progress=72.1% loss=0.084175 lr=1.978761e-05 lr_backbone=1.978761e-05 elapsed=37:10:26 avg_step=3.71s recent_step=2.82s eta=14:22:14 +2026-09-16 02:57:30 | INFO | train rank=0 | step=36070/50000 progress=72.1% loss=0.113817 lr=1.976126e-05 lr_backbone=1.976126e-05 elapsed=37:11:02 avg_step=3.71s recent_step=3.64s eta=14:21:36 +2026-09-16 02:57:57 | INFO | train rank=0 | step=36080/50000 progress=72.2% loss=0.110523 lr=1.973493e-05 lr_backbone=1.973493e-05 elapsed=37:11:30 avg_step=3.71s recent_step=2.78s eta=14:20:56 +2026-09-16 02:59:01 | INFO | train rank=0 | step=36090/50000 progress=72.2% loss=0.090435 lr=1.970862e-05 lr_backbone=1.970862e-05 elapsed=37:12:34 avg_step=3.71s recent_step=6.38s eta=14:20:29 +2026-09-16 02:59:30 | INFO | train rank=0 | step=36100/50000 progress=72.2% loss=0.101696 lr=1.968232e-05 lr_backbone=1.968232e-05 elapsed=37:13:02 avg_step=3.71s recent_step=2.85s eta=14:19:48 +2026-09-16 03:00:01 | INFO | train rank=0 | step=36110/50000 progress=72.2% loss=0.080162 lr=1.965603e-05 lr_backbone=1.965603e-05 elapsed=37:13:34 avg_step=3.71s recent_step=3.18s eta=14:19:09 +2026-09-16 03:00:39 | INFO | train rank=0 | step=36120/50000 progress=72.2% loss=0.073306 lr=1.962975e-05 lr_backbone=1.962975e-05 elapsed=37:14:12 avg_step=3.71s recent_step=3.76s eta=14:18:32 +2026-09-16 03:01:17 | INFO | train rank=0 | step=36130/50000 progress=72.3% loss=0.108234 lr=1.960349e-05 lr_backbone=1.960349e-05 elapsed=37:14:50 avg_step=3.71s recent_step=3.82s eta=14:17:56 +2026-09-16 03:01:56 | INFO | train rank=0 | step=36140/50000 progress=72.3% loss=0.080226 lr=1.957723e-05 lr_backbone=1.957723e-05 elapsed=37:15:29 avg_step=3.71s recent_step=3.91s eta=14:17:19 +2026-09-16 03:02:35 | INFO | train rank=0 | step=36150/50000 progress=72.3% loss=0.081672 lr=1.955100e-05 lr_backbone=1.955100e-05 elapsed=37:16:08 avg_step=3.71s recent_step=3.92s eta=14:16:43 +2026-09-16 03:03:13 | INFO | train rank=0 | step=36160/50000 progress=72.3% loss=0.107896 lr=1.952477e-05 lr_backbone=1.952477e-05 elapsed=37:16:46 avg_step=3.71s recent_step=3.75s eta=14:16:06 +2026-09-16 03:04:01 | INFO | train rank=0 | step=36170/50000 progress=72.3% loss=0.093212 lr=1.949856e-05 lr_backbone=1.949856e-05 elapsed=37:17:33 avg_step=3.71s recent_step=4.76s eta=14:15:33 +2026-09-16 03:04:39 | INFO | train rank=0 | step=36180/50000 progress=72.4% loss=0.092651 lr=1.947237e-05 lr_backbone=1.947237e-05 elapsed=37:18:11 avg_step=3.71s recent_step=3.80s eta=14:14:56 +2026-09-16 03:05:17 | INFO | train rank=0 | step=36190/50000 progress=72.4% loss=0.076704 lr=1.944618e-05 lr_backbone=1.944618e-05 elapsed=37:18:50 avg_step=3.71s recent_step=3.87s eta=14:14:20 +2026-09-16 03:05:46 | INFO | train rank=0 | step=36200/50000 progress=72.4% loss=0.100669 lr=1.942001e-05 lr_backbone=1.942001e-05 elapsed=37:19:19 avg_step=3.71s recent_step=2.85s eta=14:13:39 +2026-09-16 03:06:31 | INFO | train rank=0 | step=36210/50000 progress=72.4% loss=0.103269 lr=1.939386e-05 lr_backbone=1.939386e-05 elapsed=37:20:04 avg_step=3.71s recent_step=4.56s eta=14:13:05 +2026-09-16 03:07:02 | INFO | train rank=0 | step=36220/50000 progress=72.4% loss=0.074289 lr=1.936771e-05 lr_backbone=1.936771e-05 elapsed=37:20:35 avg_step=3.71s recent_step=3.08s eta=14:12:26 +2026-09-16 03:07:36 | INFO | train rank=0 | step=36230/50000 progress=72.5% loss=0.078861 lr=1.934158e-05 lr_backbone=1.934158e-05 elapsed=37:21:09 avg_step=3.71s recent_step=3.42s eta=14:11:48 +2026-09-16 03:08:12 | INFO | train rank=0 | step=36240/50000 progress=72.5% loss=0.069314 lr=1.931547e-05 lr_backbone=1.931547e-05 elapsed=37:21:45 avg_step=3.71s recent_step=3.55s eta=14:11:10 +2026-09-16 03:09:01 | INFO | train rank=0 | step=36250/50000 progress=72.5% loss=0.086826 lr=1.928936e-05 lr_backbone=1.928936e-05 elapsed=37:22:34 avg_step=3.71s recent_step=4.91s eta=14:10:37 +2026-09-16 03:09:37 | INFO | train rank=0 | step=36260/50000 progress=72.5% loss=0.093460 lr=1.926327e-05 lr_backbone=1.926327e-05 elapsed=37:23:10 avg_step=3.71s recent_step=3.65s eta=14:10:00 +2026-09-16 03:10:07 | INFO | train rank=0 | step=36270/50000 progress=72.5% loss=0.112251 lr=1.923720e-05 lr_backbone=1.923720e-05 elapsed=37:23:40 avg_step=3.71s recent_step=2.99s eta=14:09:20 +2026-09-16 03:10:38 | INFO | train rank=0 | step=36280/50000 progress=72.6% loss=0.072599 lr=1.921114e-05 lr_backbone=1.921114e-05 elapsed=37:24:10 avg_step=3.71s recent_step=3.03s eta=14:08:40 +2026-09-16 03:11:29 | INFO | train rank=0 | step=36290/50000 progress=72.6% loss=0.066478 lr=1.918509e-05 lr_backbone=1.918509e-05 elapsed=37:25:02 avg_step=3.71s recent_step=5.15s eta=14:08:09 +2026-09-16 03:11:59 | INFO | train rank=0 | step=36300/50000 progress=72.6% loss=0.088647 lr=1.915905e-05 lr_backbone=1.915905e-05 elapsed=37:25:32 avg_step=3.71s recent_step=2.97s eta=14:07:29 +2026-09-16 03:12:33 | INFO | train rank=0 | step=36310/50000 progress=72.6% loss=0.114042 lr=1.913303e-05 lr_backbone=1.913303e-05 elapsed=37:26:06 avg_step=3.71s recent_step=3.44s eta=14:06:51 +2026-09-16 03:13:09 | INFO | train rank=0 | step=36320/50000 progress=72.6% loss=0.081077 lr=1.910702e-05 lr_backbone=1.910702e-05 elapsed=37:26:42 avg_step=3.71s recent_step=3.55s eta=14:06:13 +2026-09-16 03:13:45 | INFO | train rank=0 | step=36330/50000 progress=72.7% loss=0.099017 lr=1.908102e-05 lr_backbone=1.908102e-05 elapsed=37:27:18 avg_step=3.71s recent_step=3.62s eta=14:05:35 +2026-09-16 03:14:19 | INFO | train rank=0 | step=36340/50000 progress=72.7% loss=0.093347 lr=1.905504e-05 lr_backbone=1.905504e-05 elapsed=37:27:52 avg_step=3.71s recent_step=3.44s eta=14:04:57 +2026-09-16 03:14:46 | INFO | train rank=0 | step=36350/50000 progress=72.7% loss=0.076018 lr=1.902907e-05 lr_backbone=1.902907e-05 elapsed=37:28:19 avg_step=3.71s recent_step=2.69s eta=14:04:16 +2026-09-16 03:15:16 | INFO | train rank=0 | step=36360/50000 progress=72.7% loss=0.099841 lr=1.900312e-05 lr_backbone=1.900312e-05 elapsed=37:28:48 avg_step=3.71s recent_step=2.93s eta=14:03:36 +2026-09-16 03:16:10 | INFO | train rank=0 | step=36370/50000 progress=72.7% loss=0.092222 lr=1.897718e-05 lr_backbone=1.897718e-05 elapsed=37:29:42 avg_step=3.71s recent_step=5.40s eta=14:03:06 +2026-09-16 03:16:45 | INFO | train rank=0 | step=36380/50000 progress=72.8% loss=0.109102 lr=1.895125e-05 lr_backbone=1.895125e-05 elapsed=37:30:18 avg_step=3.71s recent_step=3.55s eta=14:02:28 +2026-09-16 03:17:18 | INFO | train rank=0 | step=36390/50000 progress=72.8% loss=0.128744 lr=1.892534e-05 lr_backbone=1.892534e-05 elapsed=37:30:51 avg_step=3.71s recent_step=3.34s eta=14:01:49 +2026-09-16 03:17:49 | INFO | train rank=0 | step=36400/50000 progress=72.8% loss=0.109005 lr=1.889944e-05 lr_backbone=1.889944e-05 elapsed=37:31:22 avg_step=3.71s recent_step=3.07s eta=14:01:10 +2026-09-16 03:18:45 | INFO | train rank=0 | step=36410/50000 progress=72.8% loss=0.094606 lr=1.887355e-05 lr_backbone=1.887355e-05 elapsed=37:32:18 avg_step=3.71s recent_step=5.60s eta=14:00:40 +2026-09-16 03:19:15 | INFO | train rank=0 | step=36420/50000 progress=72.8% loss=0.103877 lr=1.884768e-05 lr_backbone=1.884768e-05 elapsed=37:32:47 avg_step=3.71s recent_step=2.95s eta=14:00:00 +2026-09-16 03:19:53 | INFO | train rank=0 | step=36430/50000 progress=72.9% loss=0.085990 lr=1.882182e-05 lr_backbone=1.882182e-05 elapsed=37:33:26 avg_step=3.71s recent_step=3.83s eta=13:59:23 +2026-09-16 03:20:24 | INFO | train rank=0 | step=36440/50000 progress=72.9% loss=0.098993 lr=1.879597e-05 lr_backbone=1.879597e-05 elapsed=37:33:56 avg_step=3.71s recent_step=3.07s eta=13:58:44 +2026-09-16 03:21:21 | INFO | train rank=0 | step=36450/50000 progress=72.9% loss=0.067974 lr=1.877014e-05 lr_backbone=1.877014e-05 elapsed=37:34:54 avg_step=3.71s recent_step=5.79s eta=13:58:14 +2026-09-16 03:21:57 | INFO | train rank=0 | step=36460/50000 progress=72.9% loss=0.093566 lr=1.874432e-05 lr_backbone=1.874432e-05 elapsed=37:35:29 avg_step=3.71s recent_step=3.51s eta=13:57:36 +2026-09-16 03:22:34 | INFO | train rank=0 | step=36470/50000 progress=72.9% loss=0.077181 lr=1.871851e-05 lr_backbone=1.871851e-05 elapsed=37:36:07 avg_step=3.71s recent_step=3.73s eta=13:56:59 +2026-09-16 03:23:12 | INFO | train rank=0 | step=36480/50000 progress=73.0% loss=0.115112 lr=1.869272e-05 lr_backbone=1.869272e-05 elapsed=37:36:45 avg_step=3.71s recent_step=3.84s eta=13:56:23 +2026-09-16 03:23:57 | INFO | train rank=0 | step=36490/50000 progress=73.0% loss=0.095653 lr=1.866695e-05 lr_backbone=1.866695e-05 elapsed=37:37:30 avg_step=3.71s recent_step=4.49s eta=13:55:48 +2026-09-16 03:24:33 | INFO | train rank=0 | step=36500/50000 progress=73.0% loss=0.090701 lr=1.864118e-05 lr_backbone=1.864118e-05 elapsed=37:38:06 avg_step=3.71s recent_step=3.62s eta=13:55:11 +2026-09-16 03:25:06 | INFO | train rank=0 | step=36510/50000 progress=73.0% loss=0.128323 lr=1.861543e-05 lr_backbone=1.861543e-05 elapsed=37:38:39 avg_step=3.71s recent_step=3.27s eta=13:54:32 +2026-09-16 03:25:33 | INFO | train rank=0 | step=36520/50000 progress=73.0% loss=0.073517 lr=1.858970e-05 lr_backbone=1.858970e-05 elapsed=37:39:06 avg_step=3.71s recent_step=2.72s eta=13:53:51 +2026-09-16 03:26:19 | INFO | train rank=0 | step=36530/50000 progress=73.1% loss=0.104240 lr=1.856397e-05 lr_backbone=1.856397e-05 elapsed=37:39:52 avg_step=3.71s recent_step=4.56s eta=13:53:17 +2026-09-16 03:26:54 | INFO | train rank=0 | step=36540/50000 progress=73.1% loss=0.087771 lr=1.853826e-05 lr_backbone=1.853826e-05 elapsed=37:40:27 avg_step=3.71s recent_step=3.52s eta=13:52:40 +2026-09-16 03:27:21 | INFO | train rank=0 | step=36550/50000 progress=73.1% loss=0.092128 lr=1.851257e-05 lr_backbone=1.851257e-05 elapsed=37:40:54 avg_step=3.71s recent_step=2.71s eta=13:51:59 +2026-09-16 03:28:00 | INFO | train rank=0 | step=36560/50000 progress=73.1% loss=0.078404 lr=1.848689e-05 lr_backbone=1.848689e-05 elapsed=37:41:32 avg_step=3.71s recent_step=3.86s eta=13:51:22 +2026-09-16 03:29:01 | INFO | train rank=0 | step=36570/50000 progress=73.1% loss=0.069803 lr=1.846122e-05 lr_backbone=1.846122e-05 elapsed=37:42:34 avg_step=3.71s recent_step=6.16s eta=13:50:54 +2026-09-16 03:29:41 | INFO | train rank=0 | step=36580/50000 progress=73.2% loss=0.097086 lr=1.843557e-05 lr_backbone=1.843557e-05 elapsed=37:43:14 avg_step=3.71s recent_step=3.99s eta=13:50:18 +2026-09-16 03:30:09 | INFO | train rank=0 | step=36590/50000 progress=73.2% loss=0.112546 lr=1.840993e-05 lr_backbone=1.840993e-05 elapsed=37:43:41 avg_step=3.71s recent_step=2.73s eta=13:49:37 +2026-09-16 03:30:47 | INFO | train rank=0 | step=36600/50000 progress=73.2% loss=0.102140 lr=1.838430e-05 lr_backbone=1.838430e-05 elapsed=37:44:20 avg_step=3.71s recent_step=3.82s eta=13:49:01 +2026-09-16 03:31:27 | INFO | train rank=0 | step=36610/50000 progress=73.2% loss=0.082864 lr=1.835869e-05 lr_backbone=1.835869e-05 elapsed=37:45:00 avg_step=3.71s recent_step=4.06s eta=13:48:25 +2026-09-16 03:32:07 | INFO | train rank=0 | step=36620/50000 progress=73.2% loss=0.109616 lr=1.833309e-05 lr_backbone=1.833309e-05 elapsed=37:45:40 avg_step=3.71s recent_step=3.98s eta=13:47:49 +2026-09-16 03:32:31 | INFO | train rank=0 | step=36630/50000 progress=73.3% loss=0.097794 lr=1.830750e-05 lr_backbone=1.830750e-05 elapsed=37:46:04 avg_step=3.71s recent_step=2.37s eta=13:47:07 +2026-09-16 03:33:05 | INFO | train rank=0 | step=36640/50000 progress=73.3% loss=0.112974 lr=1.828193e-05 lr_backbone=1.828193e-05 elapsed=37:46:38 avg_step=3.71s recent_step=3.43s eta=13:46:28 +2026-09-16 03:33:59 | INFO | train rank=0 | step=36650/50000 progress=73.3% loss=0.074800 lr=1.825638e-05 lr_backbone=1.825638e-05 elapsed=37:47:31 avg_step=3.71s recent_step=5.34s eta=13:45:57 +2026-09-16 03:34:35 | INFO | train rank=0 | step=36660/50000 progress=73.3% loss=0.107621 lr=1.823083e-05 lr_backbone=1.823083e-05 elapsed=37:48:08 avg_step=3.71s recent_step=3.66s eta=13:45:20 +2026-09-16 03:35:12 | INFO | train rank=0 | step=36670/50000 progress=73.3% loss=0.076544 lr=1.820530e-05 lr_backbone=1.820530e-05 elapsed=37:48:45 avg_step=3.71s recent_step=3.68s eta=13:44:43 +2026-09-16 03:35:47 | INFO | train rank=0 | step=36680/50000 progress=73.4% loss=0.087377 lr=1.817979e-05 lr_backbone=1.817979e-05 elapsed=37:49:20 avg_step=3.71s recent_step=3.53s eta=13:44:05 +2026-09-16 03:36:34 | INFO | train rank=0 | step=36690/50000 progress=73.4% loss=0.097762 lr=1.815429e-05 lr_backbone=1.815429e-05 elapsed=37:50:07 avg_step=3.71s recent_step=4.65s eta=13:43:31 +2026-09-16 03:37:10 | INFO | train rank=0 | step=36700/50000 progress=73.4% loss=0.085876 lr=1.812880e-05 lr_backbone=1.812880e-05 elapsed=37:50:43 avg_step=3.71s recent_step=3.64s eta=13:42:54 +2026-09-16 03:37:36 | INFO | train rank=0 | step=36710/50000 progress=73.4% loss=0.098013 lr=1.810333e-05 lr_backbone=1.810333e-05 elapsed=37:51:09 avg_step=3.71s recent_step=2.58s eta=13:42:13 +2026-09-16 03:38:07 | INFO | train rank=0 | step=36720/50000 progress=73.4% loss=0.111198 lr=1.807787e-05 lr_backbone=1.807787e-05 elapsed=37:51:40 avg_step=3.71s recent_step=3.11s eta=13:41:33 +2026-09-16 03:38:51 | INFO | train rank=0 | step=36730/50000 progress=73.5% loss=0.083929 lr=1.805242e-05 lr_backbone=1.805242e-05 elapsed=37:52:24 avg_step=3.71s recent_step=4.44s eta=13:40:59 +2026-09-16 03:39:29 | INFO | train rank=0 | step=36740/50000 progress=73.5% loss=0.104967 lr=1.802699e-05 lr_backbone=1.802699e-05 elapsed=37:53:02 avg_step=3.71s recent_step=3.77s eta=13:40:22 +2026-09-16 03:40:08 | INFO | train rank=0 | step=36750/50000 progress=73.5% loss=0.076815 lr=1.800157e-05 lr_backbone=1.800157e-05 elapsed=37:53:41 avg_step=3.71s recent_step=3.87s eta=13:39:45 +2026-09-16 03:40:51 | INFO | train rank=0 | step=36760/50000 progress=73.5% loss=0.077898 lr=1.797617e-05 lr_backbone=1.797617e-05 elapsed=37:54:24 avg_step=3.71s recent_step=4.31s eta=13:39:10 +2026-09-16 03:41:35 | INFO | train rank=0 | step=36770/50000 progress=73.5% loss=0.091974 lr=1.795078e-05 lr_backbone=1.795078e-05 elapsed=37:55:07 avg_step=3.71s recent_step=4.35s eta=13:38:36 +2026-09-16 03:42:06 | INFO | train rank=0 | step=36780/50000 progress=73.6% loss=0.093405 lr=1.792540e-05 lr_backbone=1.792540e-05 elapsed=37:55:39 avg_step=3.71s recent_step=3.15s eta=13:37:56 +2026-09-16 03:42:46 | INFO | train rank=0 | step=36790/50000 progress=73.6% loss=0.118348 lr=1.790004e-05 lr_backbone=1.790004e-05 elapsed=37:56:19 avg_step=3.71s recent_step=4.02s eta=13:37:20 +2026-09-16 03:43:22 | INFO | train rank=0 | step=36800/50000 progress=73.6% loss=0.099894 lr=1.787470e-05 lr_backbone=1.787470e-05 elapsed=37:56:54 avg_step=3.71s recent_step=3.53s eta=13:36:43 +2026-09-16 03:44:13 | INFO | train rank=0 | step=36810/50000 progress=73.6% loss=0.103563 lr=1.784936e-05 lr_backbone=1.784936e-05 elapsed=37:57:46 avg_step=3.71s recent_step=5.15s eta=13:36:11 +2026-09-16 03:44:39 | INFO | train rank=0 | step=36820/50000 progress=73.6% loss=0.119660 lr=1.782404e-05 lr_backbone=1.782404e-05 elapsed=37:58:12 avg_step=3.71s recent_step=2.62s eta=13:35:30 +2026-09-16 03:45:07 | INFO | train rank=0 | step=36830/50000 progress=73.7% loss=0.083924 lr=1.779874e-05 lr_backbone=1.779874e-05 elapsed=37:58:40 avg_step=3.71s recent_step=2.76s eta=13:34:49 +2026-09-16 03:45:39 | INFO | train rank=0 | step=36840/50000 progress=73.7% loss=0.098216 lr=1.777345e-05 lr_backbone=1.777345e-05 elapsed=37:59:12 avg_step=3.71s recent_step=3.22s eta=13:34:10 +2026-09-16 03:46:29 | INFO | train rank=0 | step=36850/50000 progress=73.7% loss=0.158984 lr=1.774817e-05 lr_backbone=1.774817e-05 elapsed=38:00:02 avg_step=3.71s recent_step=4.96s eta=13:33:38 +2026-09-16 03:47:05 | INFO | train rank=0 | step=36860/50000 progress=73.7% loss=0.085678 lr=1.772291e-05 lr_backbone=1.772291e-05 elapsed=38:00:38 avg_step=3.71s recent_step=3.64s eta=13:33:00 +2026-09-16 03:47:33 | INFO | train rank=0 | step=36870/50000 progress=73.7% loss=0.088518 lr=1.769766e-05 lr_backbone=1.769766e-05 elapsed=38:01:06 avg_step=3.71s recent_step=2.78s eta=13:32:20 +2026-09-16 03:48:09 | INFO | train rank=0 | step=36880/50000 progress=73.8% loss=0.085338 lr=1.767242e-05 lr_backbone=1.767242e-05 elapsed=38:01:42 avg_step=3.71s recent_step=3.59s eta=13:31:42 +2026-09-16 03:49:01 | INFO | train rank=0 | step=36890/50000 progress=73.8% loss=0.122819 lr=1.764720e-05 lr_backbone=1.764720e-05 elapsed=38:02:33 avg_step=3.71s recent_step=5.16s eta=13:31:10 +2026-09-16 03:49:28 | INFO | train rank=0 | step=36900/50000 progress=73.8% loss=0.086822 lr=1.762200e-05 lr_backbone=1.762200e-05 elapsed=38:03:01 avg_step=3.71s recent_step=2.71s eta=13:30:30 +2026-09-16 03:50:02 | INFO | train rank=0 | step=36910/50000 progress=73.8% loss=0.082741 lr=1.759680e-05 lr_backbone=1.759680e-05 elapsed=38:03:34 avg_step=3.71s recent_step=3.38s eta=13:29:51 +2026-09-16 03:50:30 | INFO | train rank=0 | step=36920/50000 progress=73.8% loss=0.106462 lr=1.757163e-05 lr_backbone=1.757163e-05 elapsed=38:04:03 avg_step=3.71s recent_step=2.89s eta=13:29:11 +2026-09-16 03:51:29 | INFO | train rank=0 | step=36930/50000 progress=73.9% loss=0.079852 lr=1.754646e-05 lr_backbone=1.754646e-05 elapsed=38:05:02 avg_step=3.71s recent_step=5.85s eta=13:28:42 +2026-09-16 03:52:04 | INFO | train rank=0 | step=36940/50000 progress=73.9% loss=0.122191 lr=1.752131e-05 lr_backbone=1.752131e-05 elapsed=38:05:37 avg_step=3.71s recent_step=3.49s eta=13:28:04 +2026-09-16 03:52:36 | INFO | train rank=0 | step=36950/50000 progress=73.9% loss=0.086303 lr=1.749618e-05 lr_backbone=1.749618e-05 elapsed=38:06:09 avg_step=3.71s recent_step=3.26s eta=13:27:25 +2026-09-16 03:53:07 | INFO | train rank=0 | step=36960/50000 progress=73.9% loss=0.100347 lr=1.747105e-05 lr_backbone=1.747105e-05 elapsed=38:06:40 avg_step=3.71s recent_step=3.08s eta=13:26:46 +2026-09-16 03:53:57 | INFO | train rank=0 | step=36970/50000 progress=73.9% loss=0.069003 lr=1.744595e-05 lr_backbone=1.744595e-05 elapsed=38:07:30 avg_step=3.71s recent_step=4.99s eta=13:26:13 +2026-09-16 03:54:36 | INFO | train rank=0 | step=36980/50000 progress=74.0% loss=0.101908 lr=1.742085e-05 lr_backbone=1.742085e-05 elapsed=38:08:09 avg_step=3.71s recent_step=3.92s eta=13:25:37 +2026-09-16 03:55:05 | INFO | train rank=0 | step=36990/50000 progress=74.0% loss=0.120087 lr=1.739578e-05 lr_backbone=1.739578e-05 elapsed=38:08:38 avg_step=3.71s recent_step=2.92s eta=13:24:57 +2026-09-16 03:55:39 | INFO | train rank=0 | step=37000/50000 progress=74.0% loss=0.068936 lr=1.737071e-05 lr_backbone=1.737071e-05 elapsed=38:09:12 avg_step=3.71s recent_step=3.40s eta=13:24:19 +2026-09-16 03:56:33 | INFO | train rank=0 | step=37010/50000 progress=74.0% loss=0.077568 lr=1.734566e-05 lr_backbone=1.734566e-05 elapsed=38:10:05 avg_step=3.71s recent_step=5.32s eta=13:23:47 +2026-09-16 03:57:02 | INFO | train rank=0 | step=37020/50000 progress=74.0% loss=0.075910 lr=1.732063e-05 lr_backbone=1.732063e-05 elapsed=38:10:35 avg_step=3.71s recent_step=2.90s eta=13:23:07 +2026-09-16 03:57:46 | INFO | train rank=0 | step=37030/50000 progress=74.1% loss=0.085571 lr=1.729560e-05 lr_backbone=1.729560e-05 elapsed=38:11:18 avg_step=3.71s recent_step=4.38s eta=13:22:32 +2026-09-16 03:58:14 | INFO | train rank=0 | step=37040/50000 progress=74.1% loss=0.079859 lr=1.727060e-05 lr_backbone=1.727060e-05 elapsed=38:11:47 avg_step=3.71s recent_step=2.82s eta=13:21:52 +2026-09-16 03:59:02 | INFO | train rank=0 | step=37050/50000 progress=74.1% loss=0.094513 lr=1.724560e-05 lr_backbone=1.724560e-05 elapsed=38:12:34 avg_step=3.71s recent_step=4.79s eta=13:21:19 +2026-09-16 03:59:31 | INFO | train rank=0 | step=37060/50000 progress=74.1% loss=0.091846 lr=1.722063e-05 lr_backbone=1.722063e-05 elapsed=38:13:03 avg_step=3.71s recent_step=2.91s eta=13:20:39 +2026-09-16 04:00:00 | INFO | train rank=0 | step=37070/50000 progress=74.1% loss=0.097802 lr=1.719566e-05 lr_backbone=1.719566e-05 elapsed=38:13:33 avg_step=3.71s recent_step=2.95s eta=13:19:59 +2026-09-16 04:00:32 | INFO | train rank=0 | step=37080/50000 progress=74.2% loss=0.112293 lr=1.717071e-05 lr_backbone=1.717071e-05 elapsed=38:14:05 avg_step=3.71s recent_step=3.22s eta=13:19:20 +2026-09-16 04:01:13 | INFO | train rank=0 | step=37090/50000 progress=74.2% loss=0.073072 lr=1.714578e-05 lr_backbone=1.714578e-05 elapsed=38:14:46 avg_step=3.71s recent_step=4.05s eta=13:18:44 +2026-09-16 04:01:52 | INFO | train rank=0 | step=37100/50000 progress=74.2% loss=0.086104 lr=1.712086e-05 lr_backbone=1.712086e-05 elapsed=38:15:25 avg_step=3.71s recent_step=3.94s eta=13:18:08 +2026-09-16 04:02:30 | INFO | train rank=0 | step=37110/50000 progress=74.2% loss=0.099538 lr=1.709595e-05 lr_backbone=1.709595e-05 elapsed=38:16:03 avg_step=3.71s recent_step=3.80s eta=13:17:31 +2026-09-16 04:03:04 | INFO | train rank=0 | step=37120/50000 progress=74.2% loss=0.113842 lr=1.707106e-05 lr_backbone=1.707106e-05 elapsed=38:16:37 avg_step=3.71s recent_step=3.39s eta=13:16:53 +2026-09-16 04:03:59 | INFO | train rank=0 | step=37130/50000 progress=74.3% loss=0.084553 lr=1.704618e-05 lr_backbone=1.704618e-05 elapsed=38:17:32 avg_step=3.71s recent_step=5.48s eta=13:16:22 +2026-09-16 04:04:33 | INFO | train rank=0 | step=37140/50000 progress=74.3% loss=0.071026 lr=1.702132e-05 lr_backbone=1.702132e-05 elapsed=38:18:06 avg_step=3.71s recent_step=3.45s eta=13:15:44 +2026-09-16 04:05:09 | INFO | train rank=0 | step=37150/50000 progress=74.3% loss=0.080255 lr=1.699647e-05 lr_backbone=1.699647e-05 elapsed=38:18:41 avg_step=3.71s recent_step=3.54s eta=13:15:06 +2026-09-16 04:05:45 | INFO | train rank=0 | step=37160/50000 progress=74.3% loss=0.086281 lr=1.697163e-05 lr_backbone=1.697163e-05 elapsed=38:19:17 avg_step=3.71s recent_step=3.58s eta=13:14:28 +2026-09-16 04:06:16 | INFO | train rank=0 | step=37170/50000 progress=74.3% loss=0.114367 lr=1.694681e-05 lr_backbone=1.694681e-05 elapsed=38:19:49 avg_step=3.71s recent_step=3.17s eta=13:13:49 +2026-09-16 04:06:55 | INFO | train rank=0 | step=37180/50000 progress=74.4% loss=0.103566 lr=1.692201e-05 lr_backbone=1.692201e-05 elapsed=38:20:27 avg_step=3.71s recent_step=3.85s eta=13:13:13 +2026-09-16 04:07:25 | INFO | train rank=0 | step=37190/50000 progress=74.4% loss=0.082475 lr=1.689721e-05 lr_backbone=1.689721e-05 elapsed=38:20:58 avg_step=3.71s recent_step=3.02s eta=13:12:33 +2026-09-16 04:07:54 | INFO | train rank=0 | step=37200/50000 progress=74.4% loss=0.080701 lr=1.687244e-05 lr_backbone=1.687244e-05 elapsed=38:21:26 avg_step=3.71s recent_step=2.88s eta=13:11:53 +2026-09-16 04:08:33 | INFO | train rank=0 | step=37210/50000 progress=74.4% loss=0.087315 lr=1.684768e-05 lr_backbone=1.684768e-05 elapsed=38:22:06 avg_step=3.71s recent_step=3.95s eta=13:11:17 +2026-09-16 04:09:07 | INFO | train rank=0 | step=37220/50000 progress=74.4% loss=0.078049 lr=1.682293e-05 lr_backbone=1.682293e-05 elapsed=38:22:40 avg_step=3.71s recent_step=3.36s eta=13:10:39 +2026-09-16 04:09:40 | INFO | train rank=0 | step=37230/50000 progress=74.5% loss=0.112162 lr=1.679819e-05 lr_backbone=1.679819e-05 elapsed=38:23:13 avg_step=3.71s recent_step=3.36s eta=13:10:00 +2026-09-16 04:10:14 | INFO | train rank=0 | step=37240/50000 progress=74.5% loss=0.107189 lr=1.677348e-05 lr_backbone=1.677348e-05 elapsed=38:23:47 avg_step=3.71s recent_step=3.36s eta=13:09:22 +2026-09-16 04:11:07 | INFO | train rank=0 | step=37250/50000 progress=74.5% loss=0.091626 lr=1.674877e-05 lr_backbone=1.674877e-05 elapsed=38:24:40 avg_step=3.71s recent_step=5.32s eta=13:08:50 +2026-09-16 04:11:43 | INFO | train rank=0 | step=37260/50000 progress=74.5% loss=0.128676 lr=1.672408e-05 lr_backbone=1.672408e-05 elapsed=38:25:15 avg_step=3.71s recent_step=3.55s eta=13:08:13 +2026-09-16 04:12:16 | INFO | train rank=0 | step=37270/50000 progress=74.5% loss=0.060160 lr=1.669941e-05 lr_backbone=1.669941e-05 elapsed=38:25:48 avg_step=3.71s recent_step=3.29s eta=13:07:34 +2026-09-16 04:12:54 | INFO | train rank=0 | step=37280/50000 progress=74.6% loss=0.083554 lr=1.667475e-05 lr_backbone=1.667475e-05 elapsed=38:26:27 avg_step=3.71s recent_step=3.86s eta=13:06:58 +2026-09-16 04:13:49 | INFO | train rank=0 | step=37290/50000 progress=74.6% loss=0.077281 lr=1.665010e-05 lr_backbone=1.665010e-05 elapsed=38:27:21 avg_step=3.71s recent_step=5.45s eta=13:06:26 +2026-09-16 04:14:19 | INFO | train rank=0 | step=37300/50000 progress=74.6% loss=0.082965 lr=1.662547e-05 lr_backbone=1.662547e-05 elapsed=38:27:52 avg_step=3.71s recent_step=3.06s eta=13:05:47 +2026-09-16 04:14:57 | INFO | train rank=0 | step=37310/50000 progress=74.6% loss=0.102419 lr=1.660085e-05 lr_backbone=1.660085e-05 elapsed=38:28:30 avg_step=3.71s recent_step=3.83s eta=13:05:10 +2026-09-16 04:15:31 | INFO | train rank=0 | step=37320/50000 progress=74.6% loss=0.087384 lr=1.657625e-05 lr_backbone=1.657625e-05 elapsed=38:29:04 avg_step=3.71s recent_step=3.39s eta=13:04:32 +2026-09-16 04:16:21 | INFO | train rank=0 | step=37330/50000 progress=74.7% loss=0.086434 lr=1.655166e-05 lr_backbone=1.655166e-05 elapsed=38:29:54 avg_step=3.71s recent_step=4.93s eta=13:03:59 +2026-09-16 04:17:00 | INFO | train rank=0 | step=37340/50000 progress=74.7% loss=0.056452 lr=1.652709e-05 lr_backbone=1.652709e-05 elapsed=38:30:33 avg_step=3.71s recent_step=3.97s eta=13:03:23 +2026-09-16 04:17:31 | INFO | train rank=0 | step=37350/50000 progress=74.7% loss=0.089953 lr=1.650253e-05 lr_backbone=1.650253e-05 elapsed=38:31:04 avg_step=3.71s recent_step=3.06s eta=13:02:43 +2026-09-16 04:18:02 | INFO | train rank=0 | step=37360/50000 progress=74.7% loss=0.070470 lr=1.647799e-05 lr_backbone=1.647799e-05 elapsed=38:31:35 avg_step=3.71s recent_step=3.12s eta=13:02:04 +2026-09-16 04:18:43 | INFO | train rank=0 | step=37370/50000 progress=74.7% loss=0.086510 lr=1.645346e-05 lr_backbone=1.645346e-05 elapsed=38:32:15 avg_step=3.71s recent_step=4.04s eta=13:01:28 +2026-09-16 04:19:17 | INFO | train rank=0 | step=37380/50000 progress=74.8% loss=0.088437 lr=1.642894e-05 lr_backbone=1.642894e-05 elapsed=38:32:49 avg_step=3.71s recent_step=3.39s eta=13:00:50 +2026-09-16 04:19:49 | INFO | train rank=0 | step=37390/50000 progress=74.8% loss=0.122378 lr=1.640444e-05 lr_backbone=1.640444e-05 elapsed=38:33:22 avg_step=3.71s recent_step=3.25s eta=13:00:11 +2026-09-16 04:20:16 | INFO | train rank=0 | step=37400/50000 progress=74.8% loss=0.082372 lr=1.637996e-05 lr_backbone=1.637996e-05 elapsed=38:33:49 avg_step=3.71s recent_step=2.72s eta=12:59:31 +2026-09-16 04:21:05 | INFO | train rank=0 | step=37410/50000 progress=74.8% loss=0.108091 lr=1.635549e-05 lr_backbone=1.635549e-05 elapsed=38:34:38 avg_step=3.71s recent_step=4.92s eta=12:58:58 +2026-09-16 04:21:38 | INFO | train rank=0 | step=37420/50000 progress=74.8% loss=0.100685 lr=1.633103e-05 lr_backbone=1.633103e-05 elapsed=38:35:10 avg_step=3.71s recent_step=3.22s eta=12:58:19 +2026-09-16 04:22:08 | INFO | train rank=0 | step=37430/50000 progress=74.9% loss=0.112192 lr=1.630659e-05 lr_backbone=1.630659e-05 elapsed=38:35:40 avg_step=3.71s recent_step=2.99s eta=12:57:40 +2026-09-16 04:22:41 | INFO | train rank=0 | step=37440/50000 progress=74.9% loss=0.080287 lr=1.628217e-05 lr_backbone=1.628217e-05 elapsed=38:36:14 avg_step=3.71s recent_step=3.33s eta=12:57:01 +2026-09-16 04:23:34 | INFO | train rank=0 | step=37450/50000 progress=74.9% loss=0.117447 lr=1.625776e-05 lr_backbone=1.625776e-05 elapsed=38:37:07 avg_step=3.71s recent_step=5.35s eta=12:56:30 +2026-09-16 04:24:12 | INFO | train rank=0 | step=37460/50000 progress=74.9% loss=0.071772 lr=1.623336e-05 lr_backbone=1.623336e-05 elapsed=38:37:45 avg_step=3.71s recent_step=3.78s eta=12:55:53 +2026-09-16 04:24:44 | INFO | train rank=0 | step=37470/50000 progress=74.9% loss=0.085577 lr=1.620898e-05 lr_backbone=1.620898e-05 elapsed=38:38:17 avg_step=3.71s recent_step=3.21s eta=12:55:14 +2026-09-16 04:25:21 | INFO | train rank=0 | step=37480/50000 progress=75.0% loss=0.088292 lr=1.618461e-05 lr_backbone=1.618461e-05 elapsed=38:38:54 avg_step=3.71s recent_step=3.71s eta=12:54:37 +2026-09-16 04:26:06 | INFO | train rank=0 | step=37490/50000 progress=75.0% loss=0.064360 lr=1.616026e-05 lr_backbone=1.616026e-05 elapsed=38:39:39 avg_step=3.71s recent_step=4.48s eta=12:54:02 +2026-09-16 04:26:34 | INFO | train rank=0 | step=37500/50000 progress=75.0% loss=0.095873 lr=1.613592e-05 lr_backbone=1.613592e-05 elapsed=38:40:06 avg_step=3.71s recent_step=2.72s eta=12:53:22 +2026-09-16 04:27:03 | INFO | train rank=0 | step=37510/50000 progress=75.0% loss=0.086077 lr=1.611160e-05 lr_backbone=1.611160e-05 elapsed=38:40:36 avg_step=3.71s recent_step=2.94s eta=12:52:42 +2026-09-16 04:27:35 | INFO | train rank=0 | step=37520/50000 progress=75.0% loss=0.085186 lr=1.608729e-05 lr_backbone=1.608729e-05 elapsed=38:41:08 avg_step=3.71s recent_step=3.23s eta=12:52:03 +2026-09-16 04:28:19 | INFO | train rank=0 | step=37530/50000 progress=75.1% loss=0.079171 lr=1.606300e-05 lr_backbone=1.606300e-05 elapsed=38:41:51 avg_step=3.71s recent_step=4.34s eta=12:51:28 +2026-09-16 04:28:51 | INFO | train rank=0 | step=37540/50000 progress=75.1% loss=0.092719 lr=1.603872e-05 lr_backbone=1.603872e-05 elapsed=38:42:24 avg_step=3.71s recent_step=3.25s eta=12:50:50 +2026-09-16 04:29:21 | INFO | train rank=0 | step=37550/50000 progress=75.1% loss=0.093934 lr=1.601446e-05 lr_backbone=1.601446e-05 elapsed=38:42:54 avg_step=3.71s recent_step=2.98s eta=12:50:10 +2026-09-16 04:30:03 | INFO | train rank=0 | step=37560/50000 progress=75.1% loss=0.068760 lr=1.599021e-05 lr_backbone=1.599021e-05 elapsed=38:43:36 avg_step=3.71s recent_step=4.22s eta=12:49:35 +2026-09-16 04:31:01 | INFO | train rank=0 | step=37570/50000 progress=75.1% loss=0.073050 lr=1.596598e-05 lr_backbone=1.596598e-05 elapsed=38:44:34 avg_step=3.71s recent_step=5.77s eta=12:49:04 +2026-09-16 04:31:39 | INFO | train rank=0 | step=37580/50000 progress=75.2% loss=0.095782 lr=1.594176e-05 lr_backbone=1.594176e-05 elapsed=38:45:12 avg_step=3.71s recent_step=3.86s eta=12:48:28 +2026-09-16 04:32:19 | INFO | train rank=0 | step=37590/50000 progress=75.2% loss=0.105084 lr=1.591755e-05 lr_backbone=1.591755e-05 elapsed=38:45:52 avg_step=3.71s recent_step=3.94s eta=12:47:51 +2026-09-16 04:32:47 | INFO | train rank=0 | step=37600/50000 progress=75.2% loss=0.102916 lr=1.589336e-05 lr_backbone=1.589336e-05 elapsed=38:46:20 avg_step=3.71s recent_step=2.81s eta=12:47:11 +2026-09-16 04:33:47 | INFO | train rank=0 | step=37610/50000 progress=75.2% loss=0.076873 lr=1.586919e-05 lr_backbone=1.586919e-05 elapsed=38:47:20 avg_step=3.71s recent_step=6.05s eta=12:46:42 +2026-09-16 04:34:22 | INFO | train rank=0 | step=37620/50000 progress=75.2% loss=0.100735 lr=1.584503e-05 lr_backbone=1.584503e-05 elapsed=38:47:54 avg_step=3.71s recent_step=3.43s eta=12:46:04 +2026-09-16 04:34:54 | INFO | train rank=0 | step=37630/50000 progress=75.3% loss=0.130850 lr=1.582089e-05 lr_backbone=1.582089e-05 elapsed=38:48:26 avg_step=3.71s recent_step=3.21s eta=12:45:25 +2026-09-16 04:35:28 | INFO | train rank=0 | step=37640/50000 progress=75.3% loss=0.078201 lr=1.579676e-05 lr_backbone=1.579676e-05 elapsed=38:49:00 avg_step=3.71s recent_step=3.39s eta=12:44:47 +2026-09-16 04:36:16 | INFO | train rank=0 | step=37650/50000 progress=75.3% loss=0.083314 lr=1.577264e-05 lr_backbone=1.577264e-05 elapsed=38:49:48 avg_step=3.71s recent_step=4.80s eta=12:44:13 +2026-09-16 04:36:52 | INFO | train rank=0 | step=37660/50000 progress=75.3% loss=0.095308 lr=1.574855e-05 lr_backbone=1.574855e-05 elapsed=38:50:24 avg_step=3.71s recent_step=3.61s eta=12:43:36 +2026-09-16 04:37:18 | INFO | train rank=0 | step=37670/50000 progress=75.3% loss=0.067388 lr=1.572446e-05 lr_backbone=1.572446e-05 elapsed=38:50:50 avg_step=3.71s recent_step=2.59s eta=12:42:55 +2026-09-16 04:37:52 | INFO | train rank=0 | step=37680/50000 progress=75.4% loss=0.065027 lr=1.570039e-05 lr_backbone=1.570039e-05 elapsed=38:51:25 avg_step=3.71s recent_step=3.46s eta=12:42:17 +2026-09-16 04:38:35 | INFO | train rank=0 | step=37690/50000 progress=75.4% loss=0.082828 lr=1.567634e-05 lr_backbone=1.567634e-05 elapsed=38:52:08 avg_step=3.71s recent_step=4.33s eta=12:41:42 +2026-09-16 04:39:19 | INFO | train rank=0 | step=37700/50000 progress=75.4% loss=0.073952 lr=1.565230e-05 lr_backbone=1.565230e-05 elapsed=38:52:52 avg_step=3.71s recent_step=4.37s eta=12:41:07 +2026-09-16 04:39:45 | INFO | train rank=0 | step=37710/50000 progress=75.4% loss=0.086501 lr=1.562828e-05 lr_backbone=1.562828e-05 elapsed=38:53:18 avg_step=3.71s recent_step=2.57s eta=12:40:26 +2026-09-16 04:40:19 | INFO | train rank=0 | step=37720/50000 progress=75.4% loss=0.082423 lr=1.560427e-05 lr_backbone=1.560427e-05 elapsed=38:53:52 avg_step=3.71s recent_step=3.43s eta=12:39:48 +2026-09-16 04:40:55 | INFO | train rank=0 | step=37730/50000 progress=75.5% loss=0.087398 lr=1.558027e-05 lr_backbone=1.558027e-05 elapsed=38:54:28 avg_step=3.71s recent_step=3.64s eta=12:39:11 +2026-09-16 04:41:39 | INFO | train rank=0 | step=37740/50000 progress=75.5% loss=0.073853 lr=1.555629e-05 lr_backbone=1.555629e-05 elapsed=38:55:12 avg_step=3.71s recent_step=4.36s eta=12:38:36 +2026-09-16 04:42:18 | INFO | train rank=0 | step=37750/50000 progress=75.5% loss=0.119300 lr=1.553233e-05 lr_backbone=1.553233e-05 elapsed=38:55:51 avg_step=3.71s recent_step=3.93s eta=12:37:59 +2026-09-16 04:42:48 | INFO | train rank=0 | step=37760/50000 progress=75.5% loss=0.069536 lr=1.550838e-05 lr_backbone=1.550838e-05 elapsed=38:56:21 avg_step=3.71s recent_step=3.00s eta=12:37:20 +2026-09-16 04:43:20 | INFO | train rank=0 | step=37770/50000 progress=75.5% loss=0.089553 lr=1.548445e-05 lr_backbone=1.548445e-05 elapsed=38:56:53 avg_step=3.71s recent_step=3.18s eta=12:36:41 +2026-09-16 04:44:14 | INFO | train rank=0 | step=37780/50000 progress=75.6% loss=0.103035 lr=1.546053e-05 lr_backbone=1.546053e-05 elapsed=38:57:47 avg_step=3.71s recent_step=5.37s eta=12:36:09 +2026-09-16 04:44:51 | INFO | train rank=0 | step=37790/50000 progress=75.6% loss=0.095689 lr=1.543663e-05 lr_backbone=1.543663e-05 elapsed=38:58:24 avg_step=3.71s recent_step=3.71s eta=12:35:32 +2026-09-16 04:45:19 | INFO | train rank=0 | step=37800/50000 progress=75.6% loss=0.055425 lr=1.541274e-05 lr_backbone=1.541274e-05 elapsed=38:58:52 avg_step=3.71s recent_step=2.82s eta=12:34:52 +2026-09-16 04:45:57 | INFO | train rank=0 | step=37810/50000 progress=75.6% loss=0.095074 lr=1.538886e-05 lr_backbone=1.538886e-05 elapsed=38:59:30 avg_step=3.71s recent_step=3.80s eta=12:34:15 +2026-09-16 04:46:38 | INFO | train rank=0 | step=37820/50000 progress=75.6% loss=0.082754 lr=1.536501e-05 lr_backbone=1.536501e-05 elapsed=39:00:11 avg_step=3.71s recent_step=4.08s eta=12:33:39 +2026-09-16 04:47:13 | INFO | train rank=0 | step=37830/50000 progress=75.7% loss=0.095682 lr=1.534116e-05 lr_backbone=1.534116e-05 elapsed=39:00:46 avg_step=3.71s recent_step=3.53s eta=12:33:02 +2026-09-16 04:47:45 | INFO | train rank=0 | step=37840/50000 progress=75.7% loss=0.100908 lr=1.531733e-05 lr_backbone=1.531733e-05 elapsed=39:01:18 avg_step=3.71s recent_step=3.21s eta=12:32:23 +2026-09-16 04:48:33 | INFO | train rank=0 | step=37850/50000 progress=75.7% loss=0.104604 lr=1.529352e-05 lr_backbone=1.529352e-05 elapsed=39:02:06 avg_step=3.71s recent_step=4.79s eta=12:31:49 +2026-09-16 04:49:09 | INFO | train rank=0 | step=37860/50000 progress=75.7% loss=0.119683 lr=1.526972e-05 lr_backbone=1.526972e-05 elapsed=39:02:42 avg_step=3.71s recent_step=3.59s eta=12:31:12 +2026-09-16 04:49:44 | INFO | train rank=0 | step=37870/50000 progress=75.7% loss=0.107472 lr=1.524594e-05 lr_backbone=1.524594e-05 elapsed=39:03:17 avg_step=3.71s recent_step=3.47s eta=12:30:34 +2026-09-16 04:50:24 | INFO | train rank=0 | step=37880/50000 progress=75.8% loss=0.103898 lr=1.522218e-05 lr_backbone=1.522218e-05 elapsed=39:03:56 avg_step=3.71s recent_step=3.96s eta=12:29:57 +2026-09-16 04:51:14 | INFO | train rank=0 | step=37890/50000 progress=75.8% loss=0.075607 lr=1.519842e-05 lr_backbone=1.519842e-05 elapsed=39:04:47 avg_step=3.71s recent_step=5.04s eta=12:29:24 +2026-09-16 04:51:48 | INFO | train rank=0 | step=37900/50000 progress=75.8% loss=0.086969 lr=1.517469e-05 lr_backbone=1.517469e-05 elapsed=39:05:21 avg_step=3.71s recent_step=3.39s eta=12:28:46 +2026-09-16 04:52:19 | INFO | train rank=0 | step=37910/50000 progress=75.8% loss=0.081515 lr=1.515097e-05 lr_backbone=1.515097e-05 elapsed=39:05:51 avg_step=3.71s recent_step=3.06s eta=12:28:07 +2026-09-16 04:52:45 | INFO | train rank=0 | step=37920/50000 progress=75.8% loss=0.124128 lr=1.512726e-05 lr_backbone=1.512726e-05 elapsed=39:06:17 avg_step=3.71s recent_step=2.61s eta=12:27:26 +2026-09-16 04:53:27 | INFO | train rank=0 | step=37930/50000 progress=75.9% loss=0.087046 lr=1.510357e-05 lr_backbone=1.510357e-05 elapsed=39:06:59 avg_step=3.71s recent_step=4.19s eta=12:26:51 +2026-09-16 04:54:06 | INFO | train rank=0 | step=37940/50000 progress=75.9% loss=0.069086 lr=1.507989e-05 lr_backbone=1.507989e-05 elapsed=39:07:39 avg_step=3.71s recent_step=3.95s eta=12:26:15 +2026-09-16 04:54:43 | INFO | train rank=0 | step=37950/50000 progress=75.9% loss=0.087104 lr=1.505623e-05 lr_backbone=1.505623e-05 elapsed=39:08:15 avg_step=3.71s recent_step=3.65s eta=12:25:37 +2026-09-16 04:55:20 | INFO | train rank=0 | step=37960/50000 progress=75.9% loss=0.082595 lr=1.503259e-05 lr_backbone=1.503259e-05 elapsed=39:08:53 avg_step=3.71s recent_step=3.78s eta=12:25:00 +2026-09-16 04:56:13 | INFO | train rank=0 | step=37970/50000 progress=75.9% loss=0.124491 lr=1.500896e-05 lr_backbone=1.500896e-05 elapsed=39:09:46 avg_step=3.71s recent_step=5.26s eta=12:24:28 +2026-09-16 04:56:49 | INFO | train rank=0 | step=37980/50000 progress=76.0% loss=0.090834 lr=1.498534e-05 lr_backbone=1.498534e-05 elapsed=39:10:22 avg_step=3.71s recent_step=3.58s eta=12:23:51 +2026-09-16 04:57:20 | INFO | train rank=0 | step=37990/50000 progress=76.0% loss=0.089533 lr=1.496174e-05 lr_backbone=1.496174e-05 elapsed=39:10:53 avg_step=3.71s recent_step=3.12s eta=12:23:12 +2026-09-16 04:57:59 | INFO | train rank=0 | step=38000/50000 progress=76.0% loss=0.086134 lr=1.493816e-05 lr_backbone=1.493816e-05 elapsed=39:11:31 avg_step=3.71s recent_step=3.85s eta=12:22:35 +2026-09-16 04:58:53 | INFO | train rank=0 | step=38010/50000 progress=76.0% loss=0.096603 lr=1.491459e-05 lr_backbone=1.491459e-05 elapsed=39:12:25 avg_step=3.71s recent_step=5.40s eta=12:22:03 +2026-09-16 04:59:26 | INFO | train rank=0 | step=38020/50000 progress=76.0% loss=0.087373 lr=1.489104e-05 lr_backbone=1.489104e-05 elapsed=39:12:59 avg_step=3.71s recent_step=3.39s eta=12:21:25 +2026-09-16 05:00:06 | INFO | train rank=0 | step=38030/50000 progress=76.1% loss=0.078593 lr=1.486750e-05 lr_backbone=1.486750e-05 elapsed=39:13:38 avg_step=3.71s recent_step=3.91s eta=12:20:48 +2026-09-16 05:00:28 | INFO | train rank=0 | step=38040/50000 progress=76.1% loss=0.115241 lr=1.484398e-05 lr_backbone=1.484398e-05 elapsed=39:14:00 avg_step=3.71s recent_step=2.20s eta=12:20:06 +2026-09-16 05:01:14 | INFO | train rank=0 | step=38050/50000 progress=76.1% loss=0.114347 lr=1.482047e-05 lr_backbone=1.482047e-05 elapsed=39:14:47 avg_step=3.71s recent_step=4.67s eta=12:19:32 +2026-09-16 05:01:46 | INFO | train rank=0 | step=38060/50000 progress=76.1% loss=0.104681 lr=1.479698e-05 lr_backbone=1.479698e-05 elapsed=39:15:19 avg_step=3.71s recent_step=3.16s eta=12:18:53 +2026-09-16 05:02:12 | INFO | train rank=0 | step=38070/50000 progress=76.1% loss=0.066062 lr=1.477350e-05 lr_backbone=1.477350e-05 elapsed=39:15:45 avg_step=3.71s recent_step=2.66s eta=12:18:13 +2026-09-16 05:02:51 | INFO | train rank=0 | step=38080/50000 progress=76.2% loss=0.111581 lr=1.475004e-05 lr_backbone=1.475004e-05 elapsed=39:16:24 avg_step=3.71s recent_step=3.88s eta=12:17:36 +2026-09-16 05:03:44 | INFO | train rank=0 | step=38090/50000 progress=76.2% loss=0.120046 lr=1.472660e-05 lr_backbone=1.472660e-05 elapsed=39:17:17 avg_step=3.71s recent_step=5.28s eta=12:17:04 +2026-09-16 05:04:23 | INFO | train rank=0 | step=38100/50000 progress=76.2% loss=0.071799 lr=1.470317e-05 lr_backbone=1.470317e-05 elapsed=39:17:55 avg_step=3.71s recent_step=3.85s eta=12:16:27 +2026-09-16 05:04:57 | INFO | train rank=0 | step=38110/50000 progress=76.2% loss=0.077331 lr=1.467975e-05 lr_backbone=1.467975e-05 elapsed=39:18:30 avg_step=3.71s recent_step=3.48s eta=12:15:50 +2026-09-16 05:05:33 | INFO | train rank=0 | step=38120/50000 progress=76.2% loss=0.118807 lr=1.465635e-05 lr_backbone=1.465635e-05 elapsed=39:19:06 avg_step=3.71s recent_step=3.57s eta=12:15:12 +2026-09-16 05:06:23 | INFO | train rank=0 | step=38130/50000 progress=76.3% loss=0.073908 lr=1.463297e-05 lr_backbone=1.463297e-05 elapsed=39:19:56 avg_step=3.71s recent_step=5.04s eta=12:14:39 +2026-09-16 05:06:59 | INFO | train rank=0 | step=38140/50000 progress=76.3% loss=0.081774 lr=1.460960e-05 lr_backbone=1.460960e-05 elapsed=39:20:32 avg_step=3.71s recent_step=3.54s eta=12:14:01 +2026-09-16 05:07:34 | INFO | train rank=0 | step=38150/50000 progress=76.3% loss=0.090281 lr=1.458625e-05 lr_backbone=1.458625e-05 elapsed=39:21:06 avg_step=3.71s recent_step=3.48s eta=12:13:24 +2026-09-16 05:08:09 | INFO | train rank=0 | step=38160/50000 progress=76.3% loss=0.073481 lr=1.456291e-05 lr_backbone=1.456291e-05 elapsed=39:21:42 avg_step=3.71s recent_step=3.57s eta=12:12:46 +2026-09-16 05:08:53 | INFO | train rank=0 | step=38170/50000 progress=76.3% loss=0.083000 lr=1.453959e-05 lr_backbone=1.453959e-05 elapsed=39:22:26 avg_step=3.71s recent_step=4.38s eta=12:12:11 +2026-09-16 05:09:23 | INFO | train rank=0 | step=38180/50000 progress=76.4% loss=0.095144 lr=1.451629e-05 lr_backbone=1.451629e-05 elapsed=39:22:55 avg_step=3.71s recent_step=2.93s eta=12:11:31 +2026-09-16 05:09:51 | INFO | train rank=0 | step=38190/50000 progress=76.4% loss=0.092163 lr=1.449299e-05 lr_backbone=1.449299e-05 elapsed=39:23:24 avg_step=3.71s recent_step=2.85s eta=12:10:52 +2026-09-16 05:10:29 | INFO | train rank=0 | step=38200/50000 progress=76.4% loss=0.104847 lr=1.446972e-05 lr_backbone=1.446972e-05 elapsed=39:24:02 avg_step=3.71s recent_step=3.79s eta=12:10:15 +2026-09-16 05:11:21 | INFO | train rank=0 | step=38210/50000 progress=76.4% loss=0.085101 lr=1.444646e-05 lr_backbone=1.444646e-05 elapsed=39:24:53 avg_step=3.71s recent_step=5.16s eta=12:09:42 +2026-09-16 05:12:07 | INFO | train rank=0 | step=38220/50000 progress=76.4% loss=0.084214 lr=1.442322e-05 lr_backbone=1.442322e-05 elapsed=39:25:40 avg_step=3.71s recent_step=4.62s eta=12:09:08 +2026-09-16 05:12:35 | INFO | train rank=0 | step=38230/50000 progress=76.5% loss=0.076625 lr=1.439999e-05 lr_backbone=1.439999e-05 elapsed=39:26:08 avg_step=3.71s recent_step=2.84s eta=12:08:28 +2026-09-16 05:13:15 | INFO | train rank=0 | step=38240/50000 progress=76.5% loss=0.073504 lr=1.437677e-05 lr_backbone=1.437677e-05 elapsed=39:26:47 avg_step=3.71s recent_step=3.94s eta=12:07:51 +2026-09-16 05:14:02 | INFO | train rank=0 | step=38250/50000 progress=76.5% loss=0.093769 lr=1.435358e-05 lr_backbone=1.435358e-05 elapsed=39:27:35 avg_step=3.71s recent_step=4.76s eta=12:07:17 +2026-09-16 05:14:38 | INFO | train rank=0 | step=38260/50000 progress=76.5% loss=0.078100 lr=1.433040e-05 lr_backbone=1.433040e-05 elapsed=39:28:11 avg_step=3.71s recent_step=3.56s eta=12:06:40 +2026-09-16 05:15:07 | INFO | train rank=0 | step=38270/50000 progress=76.5% loss=0.087970 lr=1.430723e-05 lr_backbone=1.430723e-05 elapsed=39:28:39 avg_step=3.71s recent_step=2.87s eta=12:06:00 +2026-09-16 05:15:48 | INFO | train rank=0 | step=38280/50000 progress=76.6% loss=0.084321 lr=1.428408e-05 lr_backbone=1.428408e-05 elapsed=39:29:21 avg_step=3.71s recent_step=4.13s eta=12:05:24 +2026-09-16 05:16:33 | INFO | train rank=0 | step=38290/50000 progress=76.6% loss=0.064385 lr=1.426094e-05 lr_backbone=1.426094e-05 elapsed=39:30:06 avg_step=3.71s recent_step=4.54s eta=12:04:50 +2026-09-16 05:17:07 | INFO | train rank=0 | step=38300/50000 progress=76.6% loss=0.079607 lr=1.423783e-05 lr_backbone=1.423783e-05 elapsed=39:30:40 avg_step=3.71s recent_step=3.41s eta=12:04:12 +2026-09-16 05:17:31 | INFO | train rank=0 | step=38310/50000 progress=76.6% loss=0.089543 lr=1.421472e-05 lr_backbone=1.421472e-05 elapsed=39:31:04 avg_step=3.71s recent_step=2.39s eta=12:03:30 +2026-09-16 05:18:04 | INFO | train rank=0 | step=38320/50000 progress=76.6% loss=0.076337 lr=1.419163e-05 lr_backbone=1.419163e-05 elapsed=39:31:37 avg_step=3.71s recent_step=3.26s eta=12:02:52 +2026-09-16 05:18:59 | INFO | train rank=0 | step=38330/50000 progress=76.7% loss=0.097862 lr=1.416856e-05 lr_backbone=1.416856e-05 elapsed=39:32:32 avg_step=3.71s recent_step=5.54s eta=12:02:20 +2026-09-16 05:19:34 | INFO | train rank=0 | step=38340/50000 progress=76.7% loss=0.128499 lr=1.414551e-05 lr_backbone=1.414551e-05 elapsed=39:33:07 avg_step=3.71s recent_step=3.47s eta=12:01:42 +2026-09-16 05:20:07 | INFO | train rank=0 | step=38350/50000 progress=76.7% loss=0.072807 lr=1.412246e-05 lr_backbone=1.412246e-05 elapsed=39:33:40 avg_step=3.71s recent_step=3.31s eta=12:01:04 +2026-09-16 05:20:36 | INFO | train rank=0 | step=38360/50000 progress=76.7% loss=0.116883 lr=1.409944e-05 lr_backbone=1.409944e-05 elapsed=39:34:09 avg_step=3.71s recent_step=2.93s eta=12:00:25 +2026-09-16 05:21:27 | INFO | train rank=0 | step=38370/50000 progress=76.7% loss=0.088451 lr=1.407643e-05 lr_backbone=1.407643e-05 elapsed=39:35:00 avg_step=3.71s recent_step=5.07s eta=11:59:52 +2026-09-16 05:21:56 | INFO | train rank=0 | step=38380/50000 progress=76.8% loss=0.089634 lr=1.405344e-05 lr_backbone=1.405344e-05 elapsed=39:35:29 avg_step=3.71s recent_step=2.92s eta=11:59:12 +2026-09-16 05:22:29 | INFO | train rank=0 | step=38390/50000 progress=76.8% loss=0.080994 lr=1.403046e-05 lr_backbone=1.403046e-05 elapsed=39:36:02 avg_step=3.71s recent_step=3.26s eta=11:58:34 +2026-09-16 05:23:08 | INFO | train rank=0 | step=38400/50000 progress=76.8% loss=0.084321 lr=1.400750e-05 lr_backbone=1.400750e-05 elapsed=39:36:41 avg_step=3.71s recent_step=3.91s eta=11:57:57 +2026-09-16 05:23:59 | INFO | train rank=0 | step=38410/50000 progress=76.8% loss=0.102566 lr=1.398455e-05 lr_backbone=1.398455e-05 elapsed=39:37:32 avg_step=3.71s recent_step=5.11s eta=11:57:24 +2026-09-16 05:24:28 | INFO | train rank=0 | step=38420/50000 progress=76.8% loss=0.125994 lr=1.396162e-05 lr_backbone=1.396162e-05 elapsed=39:38:01 avg_step=3.71s recent_step=2.93s eta=11:56:45 +2026-09-16 05:25:05 | INFO | train rank=0 | step=38430/50000 progress=76.9% loss=0.090472 lr=1.393870e-05 lr_backbone=1.393870e-05 elapsed=39:38:38 avg_step=3.71s recent_step=3.70s eta=11:56:07 +2026-09-16 05:25:35 | INFO | train rank=0 | step=38440/50000 progress=76.9% loss=0.081320 lr=1.391580e-05 lr_backbone=1.391580e-05 elapsed=39:39:08 avg_step=3.71s recent_step=2.96s eta=11:55:28 +2026-09-16 05:26:29 | INFO | train rank=0 | step=38450/50000 progress=76.9% loss=0.077367 lr=1.389292e-05 lr_backbone=1.389292e-05 elapsed=39:40:02 avg_step=3.71s recent_step=5.39s eta=11:54:56 +2026-09-16 05:27:01 | INFO | train rank=0 | step=38460/50000 progress=76.9% loss=0.073729 lr=1.387005e-05 lr_backbone=1.387005e-05 elapsed=39:40:34 avg_step=3.71s recent_step=3.24s eta=11:54:17 +2026-09-16 05:27:38 | INFO | train rank=0 | step=38470/50000 progress=76.9% loss=0.095227 lr=1.384720e-05 lr_backbone=1.384720e-05 elapsed=39:41:11 avg_step=3.71s recent_step=3.71s eta=11:53:40 +2026-09-16 05:28:11 | INFO | train rank=0 | step=38480/50000 progress=77.0% loss=0.111902 lr=1.382436e-05 lr_backbone=1.382436e-05 elapsed=39:41:44 avg_step=3.71s recent_step=3.26s eta=11:53:02 +2026-09-16 05:29:00 | INFO | train rank=0 | step=38490/50000 progress=77.0% loss=0.104951 lr=1.380154e-05 lr_backbone=1.380154e-05 elapsed=39:42:33 avg_step=3.71s recent_step=4.94s eta=11:52:28 +2026-09-16 05:29:33 | INFO | train rank=0 | step=38500/50000 progress=77.0% loss=0.079732 lr=1.377874e-05 lr_backbone=1.377874e-05 elapsed=39:43:06 avg_step=3.71s recent_step=3.28s eta=11:51:50 +2026-09-16 05:30:02 | INFO | train rank=0 | step=38510/50000 progress=77.0% loss=0.073757 lr=1.375595e-05 lr_backbone=1.375595e-05 elapsed=39:43:35 avg_step=3.71s recent_step=2.86s eta=11:51:10 +2026-09-16 05:30:41 | INFO | train rank=0 | step=38520/50000 progress=77.0% loss=0.086391 lr=1.373318e-05 lr_backbone=1.373318e-05 elapsed=39:44:14 avg_step=3.71s recent_step=3.89s eta=11:50:33 +2026-09-16 05:31:27 | INFO | train rank=0 | step=38530/50000 progress=77.1% loss=0.091346 lr=1.371042e-05 lr_backbone=1.371042e-05 elapsed=39:45:00 avg_step=3.71s recent_step=4.62s eta=11:49:59 +2026-09-16 05:31:56 | INFO | train rank=0 | step=38540/50000 progress=77.1% loss=0.102727 lr=1.368768e-05 lr_backbone=1.368768e-05 elapsed=39:45:28 avg_step=3.71s recent_step=2.85s eta=11:49:19 +2026-09-16 05:32:28 | INFO | train rank=0 | step=38550/50000 progress=77.1% loss=0.067538 lr=1.366496e-05 lr_backbone=1.366496e-05 elapsed=39:46:00 avg_step=3.71s recent_step=3.20s eta=11:48:41 +2026-09-16 05:33:03 | INFO | train rank=0 | step=38560/50000 progress=77.1% loss=0.074473 lr=1.364225e-05 lr_backbone=1.364225e-05 elapsed=39:46:36 avg_step=3.71s recent_step=3.54s eta=11:48:03 +2026-09-16 05:34:00 | INFO | train rank=0 | step=38570/50000 progress=77.1% loss=0.070304 lr=1.361955e-05 lr_backbone=1.361955e-05 elapsed=39:47:33 avg_step=3.71s recent_step=5.73s eta=11:47:32 +2026-09-16 05:34:44 | INFO | train rank=0 | step=38580/50000 progress=77.2% loss=0.095569 lr=1.359688e-05 lr_backbone=1.359688e-05 elapsed=39:48:17 avg_step=3.71s recent_step=4.39s eta=11:46:57 +2026-09-16 05:35:09 | INFO | train rank=0 | step=38590/50000 progress=77.2% loss=0.120254 lr=1.357421e-05 lr_backbone=1.357421e-05 elapsed=39:48:41 avg_step=3.71s recent_step=2.45s eta=11:46:16 +2026-09-16 05:35:44 | INFO | train rank=0 | step=38600/50000 progress=77.2% loss=0.101289 lr=1.355157e-05 lr_backbone=1.355157e-05 elapsed=39:49:17 avg_step=3.71s recent_step=3.56s eta=11:45:38 +2026-09-16 05:36:38 | INFO | train rank=0 | step=38610/50000 progress=77.2% loss=0.085934 lr=1.352894e-05 lr_backbone=1.352894e-05 elapsed=39:50:11 avg_step=3.71s recent_step=5.36s eta=11:45:06 +2026-09-16 05:37:10 | INFO | train rank=0 | step=38620/50000 progress=77.2% loss=0.091532 lr=1.350633e-05 lr_backbone=1.350633e-05 elapsed=39:50:43 avg_step=3.71s recent_step=3.21s eta=11:44:27 +2026-09-16 05:37:39 | INFO | train rank=0 | step=38630/50000 progress=77.3% loss=0.081735 lr=1.348373e-05 lr_backbone=1.348373e-05 elapsed=39:51:12 avg_step=3.71s recent_step=2.91s eta=11:43:48 +2026-09-16 05:38:14 | INFO | train rank=0 | step=38640/50000 progress=77.3% loss=0.117666 lr=1.346115e-05 lr_backbone=1.346115e-05 elapsed=39:51:46 avg_step=3.71s recent_step=3.47s eta=11:43:10 +2026-09-16 05:39:03 | INFO | train rank=0 | step=38650/50000 progress=77.3% loss=0.087666 lr=1.343858e-05 lr_backbone=1.343858e-05 elapsed=39:52:36 avg_step=3.71s recent_step=4.95s eta=11:42:36 +2026-09-16 05:39:37 | INFO | train rank=0 | step=38660/50000 progress=77.3% loss=0.091231 lr=1.341603e-05 lr_backbone=1.341603e-05 elapsed=39:53:10 avg_step=3.71s recent_step=3.40s eta=11:41:58 +2026-09-16 05:40:11 | INFO | train rank=0 | step=38670/50000 progress=77.3% loss=0.083130 lr=1.339350e-05 lr_backbone=1.339350e-05 elapsed=39:53:43 avg_step=3.71s recent_step=3.34s eta=11:41:20 +2026-09-16 05:40:46 | INFO | train rank=0 | step=38680/50000 progress=77.4% loss=0.101282 lr=1.337098e-05 lr_backbone=1.337098e-05 elapsed=39:54:19 avg_step=3.71s recent_step=3.52s eta=11:40:42 +2026-09-16 05:41:40 | INFO | train rank=0 | step=38690/50000 progress=77.4% loss=0.089401 lr=1.334848e-05 lr_backbone=1.334848e-05 elapsed=39:55:13 avg_step=3.71s recent_step=5.42s eta=11:40:10 +2026-09-16 05:42:13 | INFO | train rank=0 | step=38700/50000 progress=77.4% loss=0.111164 lr=1.332599e-05 lr_backbone=1.332599e-05 elapsed=39:55:46 avg_step=3.71s recent_step=3.34s eta=11:39:32 +2026-09-16 05:42:46 | INFO | train rank=0 | step=38710/50000 progress=77.4% loss=0.090744 lr=1.330352e-05 lr_backbone=1.330352e-05 elapsed=39:56:19 avg_step=3.71s recent_step=3.29s eta=11:38:54 +2026-09-16 05:43:19 | INFO | train rank=0 | step=38720/50000 progress=77.4% loss=0.105217 lr=1.328107e-05 lr_backbone=1.328107e-05 elapsed=39:56:51 avg_step=3.71s recent_step=3.24s eta=11:38:15 +2026-09-16 05:44:13 | INFO | train rank=0 | step=38730/50000 progress=77.5% loss=0.077979 lr=1.325863e-05 lr_backbone=1.325863e-05 elapsed=39:57:46 avg_step=3.71s recent_step=5.47s eta=11:37:43 +2026-09-16 05:44:43 | INFO | train rank=0 | step=38740/50000 progress=77.5% loss=0.082598 lr=1.323621e-05 lr_backbone=1.323621e-05 elapsed=39:58:16 avg_step=3.71s recent_step=3.00s eta=11:37:04 +2026-09-16 05:45:13 | INFO | train rank=0 | step=38750/50000 progress=77.5% loss=0.088274 lr=1.321380e-05 lr_backbone=1.321380e-05 elapsed=39:58:46 avg_step=3.71s recent_step=2.97s eta=11:36:25 +2026-09-16 05:45:43 | INFO | train rank=0 | step=38760/50000 progress=77.5% loss=0.082849 lr=1.319142e-05 lr_backbone=1.319142e-05 elapsed=39:59:15 avg_step=3.71s recent_step=2.96s eta=11:35:45 +2026-09-16 05:46:19 | INFO | train rank=0 | step=38770/50000 progress=77.5% loss=0.088764 lr=1.316904e-05 lr_backbone=1.316904e-05 elapsed=39:59:52 avg_step=3.71s recent_step=3.62s eta=11:35:08 +2026-09-16 05:47:00 | INFO | train rank=0 | step=38780/50000 progress=77.6% loss=0.127353 lr=1.314669e-05 lr_backbone=1.314669e-05 elapsed=40:00:33 avg_step=3.71s recent_step=4.16s eta=11:34:32 +2026-09-16 05:47:34 | INFO | train rank=0 | step=38790/50000 progress=77.6% loss=0.095171 lr=1.312434e-05 lr_backbone=1.312434e-05 elapsed=40:01:07 avg_step=3.71s recent_step=3.36s eta=11:33:54 +2026-09-16 05:48:05 | INFO | train rank=0 | step=38800/50000 progress=77.6% loss=0.124748 lr=1.310202e-05 lr_backbone=1.310202e-05 elapsed=40:01:38 avg_step=3.71s recent_step=3.07s eta=11:33:15 +2026-09-16 05:48:49 | INFO | train rank=0 | step=38810/50000 progress=77.6% loss=0.070953 lr=1.307971e-05 lr_backbone=1.307971e-05 elapsed=40:02:22 avg_step=3.71s recent_step=4.45s eta=11:32:40 +2026-09-16 05:49:20 | INFO | train rank=0 | step=38820/50000 progress=77.6% loss=0.094068 lr=1.305742e-05 lr_backbone=1.305742e-05 elapsed=40:02:52 avg_step=3.71s recent_step=3.03s eta=11:32:01 +2026-09-16 05:49:52 | INFO | train rank=0 | step=38830/50000 progress=77.7% loss=0.085374 lr=1.303514e-05 lr_backbone=1.303514e-05 elapsed=40:03:25 avg_step=3.71s recent_step=3.29s eta=11:31:22 +2026-09-16 05:50:29 | INFO | train rank=0 | step=38840/50000 progress=77.7% loss=0.120763 lr=1.301288e-05 lr_backbone=1.301288e-05 elapsed=40:04:02 avg_step=3.71s recent_step=3.71s eta=11:30:45 +2026-09-16 05:51:24 | INFO | train rank=0 | step=38850/50000 progress=77.7% loss=0.102951 lr=1.299064e-05 lr_backbone=1.299064e-05 elapsed=40:04:57 avg_step=3.71s recent_step=5.47s eta=11:30:13 +2026-09-16 05:51:50 | INFO | train rank=0 | step=38860/50000 progress=77.7% loss=0.136889 lr=1.296841e-05 lr_backbone=1.296841e-05 elapsed=40:05:23 avg_step=3.71s recent_step=2.63s eta=11:29:33 +2026-09-16 05:52:23 | INFO | train rank=0 | step=38870/50000 progress=77.7% loss=0.062491 lr=1.294620e-05 lr_backbone=1.294620e-05 elapsed=40:05:56 avg_step=3.71s recent_step=3.30s eta=11:28:54 +2026-09-16 05:52:58 | INFO | train rank=0 | step=38880/50000 progress=77.8% loss=0.091575 lr=1.292400e-05 lr_backbone=1.292400e-05 elapsed=40:06:31 avg_step=3.71s recent_step=3.45s eta=11:28:17 +2026-09-16 05:53:43 | INFO | train rank=0 | step=38890/50000 progress=77.8% loss=0.082520 lr=1.290182e-05 lr_backbone=1.290182e-05 elapsed=40:07:16 avg_step=3.71s recent_step=4.54s eta=11:27:42 +2026-09-16 05:54:21 | INFO | train rank=0 | step=38900/50000 progress=77.8% loss=0.084930 lr=1.287966e-05 lr_backbone=1.287966e-05 elapsed=40:07:54 avg_step=3.71s recent_step=3.80s eta=11:27:05 +2026-09-16 05:54:54 | INFO | train rank=0 | step=38910/50000 progress=77.8% loss=0.107309 lr=1.285751e-05 lr_backbone=1.285751e-05 elapsed=40:08:27 avg_step=3.71s recent_step=3.24s eta=11:26:26 +2026-09-16 05:55:23 | INFO | train rank=0 | step=38920/50000 progress=77.8% loss=0.116590 lr=1.283538e-05 lr_backbone=1.283538e-05 elapsed=40:08:56 avg_step=3.71s recent_step=2.95s eta=11:25:47 +2026-09-16 05:56:07 | INFO | train rank=0 | step=38930/50000 progress=77.9% loss=0.083799 lr=1.281327e-05 lr_backbone=1.281327e-05 elapsed=40:09:40 avg_step=3.71s recent_step=4.39s eta=11:25:12 +2026-09-16 05:56:33 | INFO | train rank=0 | step=38940/50000 progress=77.9% loss=0.075436 lr=1.279117e-05 lr_backbone=1.279117e-05 elapsed=40:10:06 avg_step=3.71s recent_step=2.58s eta=11:24:32 +2026-09-16 05:57:09 | INFO | train rank=0 | step=38950/50000 progress=77.9% loss=0.091122 lr=1.276909e-05 lr_backbone=1.276909e-05 elapsed=40:10:42 avg_step=3.71s recent_step=3.58s eta=11:23:54 +2026-09-16 05:57:37 | INFO | train rank=0 | step=38960/50000 progress=77.9% loss=0.087364 lr=1.274702e-05 lr_backbone=1.274702e-05 elapsed=40:11:10 avg_step=3.71s recent_step=2.84s eta=11:23:14 +2026-09-16 05:58:17 | INFO | train rank=0 | step=38970/50000 progress=77.9% loss=0.080664 lr=1.272498e-05 lr_backbone=1.272498e-05 elapsed=40:11:50 avg_step=3.71s recent_step=3.99s eta=11:22:38 +2026-09-16 05:58:46 | INFO | train rank=0 | step=38980/50000 progress=78.0% loss=0.100775 lr=1.270294e-05 lr_backbone=1.270294e-05 elapsed=40:12:19 avg_step=3.71s recent_step=2.92s eta=11:21:59 +2026-09-16 05:59:32 | INFO | train rank=0 | step=38990/50000 progress=78.0% loss=0.090050 lr=1.268093e-05 lr_backbone=1.268093e-05 elapsed=40:13:05 avg_step=3.71s recent_step=4.55s eta=11:21:24 +2026-09-16 06:00:04 | INFO | train rank=0 | step=39000/50000 progress=78.0% loss=0.111566 lr=1.265893e-05 lr_backbone=1.265893e-05 elapsed=40:13:37 avg_step=3.71s recent_step=3.22s eta=11:20:45 +2026-09-16 06:00:41 | INFO | train rank=0 | step=39010/50000 progress=78.0% loss=0.127024 lr=1.263694e-05 lr_backbone=1.263694e-05 elapsed=40:14:14 avg_step=3.71s recent_step=3.68s eta=11:20:08 +2026-09-16 06:01:19 | INFO | train rank=0 | step=39020/50000 progress=78.0% loss=0.073251 lr=1.261497e-05 lr_backbone=1.261497e-05 elapsed=40:14:51 avg_step=3.71s recent_step=3.77s eta=11:19:31 +2026-09-16 06:01:55 | INFO | train rank=0 | step=39030/50000 progress=78.1% loss=0.105490 lr=1.259302e-05 lr_backbone=1.259302e-05 elapsed=40:15:28 avg_step=3.71s recent_step=3.67s eta=11:18:54 +2026-09-16 06:02:34 | INFO | train rank=0 | step=39040/50000 progress=78.1% loss=0.074800 lr=1.257109e-05 lr_backbone=1.257109e-05 elapsed=40:16:07 avg_step=3.71s recent_step=3.86s eta=11:18:17 +2026-09-16 06:03:20 | INFO | train rank=0 | step=39050/50000 progress=78.1% loss=0.106594 lr=1.254917e-05 lr_backbone=1.254917e-05 elapsed=40:16:53 avg_step=3.71s recent_step=4.59s eta=11:17:43 +2026-09-16 06:03:59 | INFO | train rank=0 | step=39060/50000 progress=78.1% loss=0.093522 lr=1.252727e-05 lr_backbone=1.252727e-05 elapsed=40:17:32 avg_step=3.71s recent_step=3.96s eta=11:17:06 +2026-09-16 06:04:35 | INFO | train rank=0 | step=39070/50000 progress=78.1% loss=0.098571 lr=1.250538e-05 lr_backbone=1.250538e-05 elapsed=40:18:08 avg_step=3.71s recent_step=3.53s eta=11:16:28 +2026-09-16 06:05:09 | INFO | train rank=0 | step=39080/50000 progress=78.2% loss=0.110080 lr=1.248351e-05 lr_backbone=1.248351e-05 elapsed=40:18:41 avg_step=3.71s recent_step=3.40s eta=11:15:50 +2026-09-16 06:05:51 | INFO | train rank=0 | step=39090/50000 progress=78.2% loss=0.099139 lr=1.246166e-05 lr_backbone=1.246166e-05 elapsed=40:19:24 avg_step=3.71s recent_step=4.23s eta=11:15:15 +2026-09-16 06:06:32 | INFO | train rank=0 | step=39100/50000 progress=78.2% loss=0.075027 lr=1.243982e-05 lr_backbone=1.243982e-05 elapsed=40:20:05 avg_step=3.71s recent_step=4.14s eta=11:14:39 +2026-09-16 06:07:11 | INFO | train rank=0 | step=39110/50000 progress=78.2% loss=0.129780 lr=1.241800e-05 lr_backbone=1.241800e-05 elapsed=40:20:44 avg_step=3.71s recent_step=3.90s eta=11:14:02 +2026-09-16 06:07:46 | INFO | train rank=0 | step=39120/50000 progress=78.2% loss=0.118982 lr=1.239620e-05 lr_backbone=1.239620e-05 elapsed=40:21:19 avg_step=3.71s recent_step=3.47s eta=11:13:24 +2026-09-16 06:08:20 | INFO | train rank=0 | step=39130/50000 progress=78.3% loss=0.099550 lr=1.237441e-05 lr_backbone=1.237441e-05 elapsed=40:21:53 avg_step=3.71s recent_step=3.42s eta=11:12:46 +2026-09-16 06:09:04 | INFO | train rank=0 | step=39140/50000 progress=78.3% loss=0.098403 lr=1.235264e-05 lr_backbone=1.235264e-05 elapsed=40:22:37 avg_step=3.71s recent_step=4.41s eta=11:12:11 +2026-09-16 06:09:32 | INFO | train rank=0 | step=39150/50000 progress=78.3% loss=0.088230 lr=1.233089e-05 lr_backbone=1.233089e-05 elapsed=40:23:05 avg_step=3.71s recent_step=2.76s eta=11:11:31 +2026-09-16 06:10:10 | INFO | train rank=0 | step=39160/50000 progress=78.3% loss=0.114176 lr=1.230915e-05 lr_backbone=1.230915e-05 elapsed=40:23:43 avg_step=3.71s recent_step=3.82s eta=11:10:55 +2026-09-16 06:10:46 | INFO | train rank=0 | step=39170/50000 progress=78.3% loss=0.084442 lr=1.228743e-05 lr_backbone=1.228743e-05 elapsed=40:24:19 avg_step=3.71s recent_step=3.62s eta=11:10:17 +2026-09-16 06:11:22 | INFO | train rank=0 | step=39180/50000 progress=78.4% loss=0.077167 lr=1.226573e-05 lr_backbone=1.226573e-05 elapsed=40:24:55 avg_step=3.71s recent_step=3.58s eta=11:09:40 +2026-09-16 06:12:03 | INFO | train rank=0 | step=39190/50000 progress=78.4% loss=0.083120 lr=1.224404e-05 lr_backbone=1.224404e-05 elapsed=40:25:36 avg_step=3.71s recent_step=4.10s eta=11:09:04 +2026-09-16 06:12:41 | INFO | train rank=0 | step=39200/50000 progress=78.4% loss=0.117282 lr=1.222237e-05 lr_backbone=1.222237e-05 elapsed=40:26:14 avg_step=3.71s recent_step=3.79s eta=11:08:27 +2026-09-16 06:13:21 | INFO | train rank=0 | step=39210/50000 progress=78.4% loss=0.075561 lr=1.220071e-05 lr_backbone=1.220071e-05 elapsed=40:26:54 avg_step=3.71s recent_step=4.02s eta=11:07:50 +2026-09-16 06:13:55 | INFO | train rank=0 | step=39220/50000 progress=78.4% loss=0.091344 lr=1.217907e-05 lr_backbone=1.217907e-05 elapsed=40:27:28 avg_step=3.71s recent_step=3.36s eta=11:07:12 +2026-09-16 06:14:38 | INFO | train rank=0 | step=39230/50000 progress=78.5% loss=0.079493 lr=1.215745e-05 lr_backbone=1.215745e-05 elapsed=40:28:11 avg_step=3.71s recent_step=4.31s eta=11:06:37 +2026-09-16 06:15:16 | INFO | train rank=0 | step=39240/50000 progress=78.5% loss=0.090626 lr=1.213585e-05 lr_backbone=1.213585e-05 elapsed=40:28:49 avg_step=3.71s recent_step=3.78s eta=11:06:00 +2026-09-16 06:15:46 | INFO | train rank=0 | step=39250/50000 progress=78.5% loss=0.072466 lr=1.211426e-05 lr_backbone=1.211426e-05 elapsed=40:29:18 avg_step=3.71s recent_step=2.96s eta=11:05:21 +2026-09-16 06:16:26 | INFO | train rank=0 | step=39260/50000 progress=78.5% loss=0.089888 lr=1.209268e-05 lr_backbone=1.209268e-05 elapsed=40:29:59 avg_step=3.71s recent_step=4.09s eta=11:04:45 +2026-09-16 06:17:12 | INFO | train rank=0 | step=39270/50000 progress=78.5% loss=0.113720 lr=1.207113e-05 lr_backbone=1.207113e-05 elapsed=40:30:44 avg_step=3.71s recent_step=4.53s eta=11:04:10 +2026-09-16 06:17:51 | INFO | train rank=0 | step=39280/50000 progress=78.6% loss=0.089408 lr=1.204959e-05 lr_backbone=1.204959e-05 elapsed=40:31:23 avg_step=3.71s recent_step=3.90s eta=11:03:33 +2026-09-16 06:18:22 | INFO | train rank=0 | step=39290/50000 progress=78.6% loss=0.064192 lr=1.202807e-05 lr_backbone=1.202807e-05 elapsed=40:31:55 avg_step=3.71s recent_step=3.14s eta=11:02:54 +2026-09-16 06:18:56 | INFO | train rank=0 | step=39300/50000 progress=78.6% loss=0.109120 lr=1.200656e-05 lr_backbone=1.200656e-05 elapsed=40:32:29 avg_step=3.71s recent_step=3.39s eta=11:02:16 +2026-09-16 06:19:40 | INFO | train rank=0 | step=39310/50000 progress=78.6% loss=0.089448 lr=1.198507e-05 lr_backbone=1.198507e-05 elapsed=40:33:13 avg_step=3.71s recent_step=4.42s eta=11:01:41 +2026-09-16 06:20:22 | INFO | train rank=0 | step=39320/50000 progress=78.6% loss=0.109286 lr=1.196360e-05 lr_backbone=1.196360e-05 elapsed=40:33:55 avg_step=3.71s recent_step=4.16s eta=11:01:05 +2026-09-16 06:20:54 | INFO | train rank=0 | step=39330/50000 progress=78.7% loss=0.069020 lr=1.194214e-05 lr_backbone=1.194214e-05 elapsed=40:34:26 avg_step=3.71s recent_step=3.19s eta=11:00:27 +2026-09-16 06:21:22 | INFO | train rank=0 | step=39340/50000 progress=78.7% loss=0.100117 lr=1.192070e-05 lr_backbone=1.192070e-05 elapsed=40:34:55 avg_step=3.71s recent_step=2.81s eta=10:59:47 +2026-09-16 06:22:12 | INFO | train rank=0 | step=39350/50000 progress=78.7% loss=0.089455 lr=1.189928e-05 lr_backbone=1.189928e-05 elapsed=40:35:45 avg_step=3.71s recent_step=5.02s eta=10:59:13 +2026-09-16 06:22:41 | INFO | train rank=0 | step=39360/50000 progress=78.7% loss=0.104960 lr=1.187787e-05 lr_backbone=1.187787e-05 elapsed=40:36:14 avg_step=3.71s recent_step=2.93s eta=10:58:34 +2026-09-16 06:23:12 | INFO | train rank=0 | step=39370/50000 progress=78.7% loss=0.087726 lr=1.185649e-05 lr_backbone=1.185649e-05 elapsed=40:36:45 avg_step=3.71s recent_step=3.11s eta=10:57:55 +2026-09-16 06:23:50 | INFO | train rank=0 | step=39380/50000 progress=78.8% loss=0.076547 lr=1.183511e-05 lr_backbone=1.183511e-05 elapsed=40:37:23 avg_step=3.71s recent_step=3.78s eta=10:57:18 +2026-09-16 06:24:43 | INFO | train rank=0 | step=39390/50000 progress=78.8% loss=0.088503 lr=1.181376e-05 lr_backbone=1.181376e-05 elapsed=40:38:16 avg_step=3.71s recent_step=5.29s eta=10:56:46 +2026-09-16 06:25:13 | INFO | train rank=0 | step=39400/50000 progress=78.8% loss=0.095584 lr=1.179242e-05 lr_backbone=1.179242e-05 elapsed=40:38:46 avg_step=3.71s recent_step=3.03s eta=10:56:07 +2026-09-16 06:25:53 | INFO | train rank=0 | step=39410/50000 progress=78.8% loss=0.085136 lr=1.177109e-05 lr_backbone=1.177109e-05 elapsed=40:39:26 avg_step=3.71s recent_step=4.01s eta=10:55:30 +2026-09-16 06:26:28 | INFO | train rank=0 | step=39420/50000 progress=78.8% loss=0.076753 lr=1.174979e-05 lr_backbone=1.174979e-05 elapsed=40:40:01 avg_step=3.71s recent_step=3.51s eta=10:54:53 +2026-09-16 06:27:15 | INFO | train rank=0 | step=39430/50000 progress=78.9% loss=0.103184 lr=1.172850e-05 lr_backbone=1.172850e-05 elapsed=40:40:48 avg_step=3.71s recent_step=4.64s eta=10:54:18 +2026-09-16 06:27:59 | INFO | train rank=0 | step=39440/50000 progress=78.9% loss=0.121785 lr=1.170723e-05 lr_backbone=1.170723e-05 elapsed=40:41:32 avg_step=3.71s recent_step=4.45s eta=10:53:43 +2026-09-16 06:28:38 | INFO | train rank=0 | step=39450/50000 progress=78.9% loss=0.103652 lr=1.168597e-05 lr_backbone=1.168597e-05 elapsed=40:42:10 avg_step=3.71s recent_step=3.81s eta=10:53:06 +2026-09-16 06:29:13 | INFO | train rank=0 | step=39460/50000 progress=78.9% loss=0.141869 lr=1.166473e-05 lr_backbone=1.166473e-05 elapsed=40:42:46 avg_step=3.71s recent_step=3.54s eta=10:52:28 +2026-09-16 06:29:44 | INFO | train rank=0 | step=39470/50000 progress=78.9% loss=0.097045 lr=1.164351e-05 lr_backbone=1.164351e-05 elapsed=40:43:16 avg_step=3.71s recent_step=3.06s eta=10:51:49 +2026-09-16 06:30:26 | INFO | train rank=0 | step=39480/50000 progress=79.0% loss=0.105067 lr=1.162230e-05 lr_backbone=1.162230e-05 elapsed=40:43:58 avg_step=3.71s recent_step=4.20s eta=10:51:13 +2026-09-16 06:30:55 | INFO | train rank=0 | step=39490/50000 progress=79.0% loss=0.077172 lr=1.160112e-05 lr_backbone=1.160112e-05 elapsed=40:44:28 avg_step=3.71s recent_step=2.94s eta=10:50:34 +2026-09-16 06:31:22 | INFO | train rank=0 | step=39500/50000 progress=79.0% loss=0.063030 lr=1.157994e-05 lr_backbone=1.157994e-05 elapsed=40:44:55 avg_step=3.71s recent_step=2.72s eta=10:49:54 +2026-09-16 06:31:59 | INFO | train rank=0 | step=39510/50000 progress=79.0% loss=0.090503 lr=1.155879e-05 lr_backbone=1.155879e-05 elapsed=40:45:32 avg_step=3.71s recent_step=3.71s eta=10:49:17 +2026-09-16 06:32:58 | INFO | train rank=0 | step=39520/50000 progress=79.0% loss=0.087674 lr=1.153765e-05 lr_backbone=1.153765e-05 elapsed=40:46:31 avg_step=3.71s recent_step=5.90s eta=10:48:46 +2026-09-16 06:33:37 | INFO | train rank=0 | step=39530/50000 progress=79.1% loss=0.117645 lr=1.151653e-05 lr_backbone=1.151653e-05 elapsed=40:47:09 avg_step=3.71s recent_step=3.83s eta=10:48:09 +2026-09-16 06:34:10 | INFO | train rank=0 | step=39540/50000 progress=79.1% loss=0.088920 lr=1.149543e-05 lr_backbone=1.149543e-05 elapsed=40:47:42 avg_step=3.71s recent_step=3.31s eta=10:47:31 +2026-09-16 06:34:46 | INFO | train rank=0 | step=39550/50000 progress=79.1% loss=0.135210 lr=1.147434e-05 lr_backbone=1.147434e-05 elapsed=40:48:19 avg_step=3.71s recent_step=3.63s eta=10:46:54 +2026-09-16 06:35:35 | INFO | train rank=0 | step=39560/50000 progress=79.1% loss=0.093680 lr=1.145327e-05 lr_backbone=1.145327e-05 elapsed=40:49:08 avg_step=3.71s recent_step=4.90s eta=10:46:20 +2026-09-16 06:36:10 | INFO | train rank=0 | step=39570/50000 progress=79.1% loss=0.086563 lr=1.143221e-05 lr_backbone=1.143221e-05 elapsed=40:49:43 avg_step=3.71s recent_step=3.55s eta=10:45:42 +2026-09-16 06:36:39 | INFO | train rank=0 | step=39580/50000 progress=79.2% loss=0.084203 lr=1.141118e-05 lr_backbone=1.141118e-05 elapsed=40:50:12 avg_step=3.71s recent_step=2.85s eta=10:45:03 +2026-09-16 06:37:12 | INFO | train rank=0 | step=39590/50000 progress=79.2% loss=0.065648 lr=1.139016e-05 lr_backbone=1.139016e-05 elapsed=40:50:45 avg_step=3.71s recent_step=3.31s eta=10:44:24 +2026-09-16 06:37:58 | INFO | train rank=0 | step=39600/50000 progress=79.2% loss=0.097861 lr=1.136915e-05 lr_backbone=1.136915e-05 elapsed=40:51:31 avg_step=3.71s recent_step=4.62s eta=10:43:50 +2026-09-16 06:38:28 | INFO | train rank=0 | step=39610/50000 progress=79.2% loss=0.070932 lr=1.134817e-05 lr_backbone=1.134817e-05 elapsed=40:52:01 avg_step=3.71s recent_step=2.97s eta=10:43:11 +2026-09-16 06:39:03 | INFO | train rank=0 | step=39620/50000 progress=79.2% loss=0.093784 lr=1.132720e-05 lr_backbone=1.132720e-05 elapsed=40:52:36 avg_step=3.71s recent_step=3.51s eta=10:42:33 +2026-09-16 06:39:40 | INFO | train rank=0 | step=39630/50000 progress=79.3% loss=0.089578 lr=1.130624e-05 lr_backbone=1.130624e-05 elapsed=40:53:12 avg_step=3.71s recent_step=3.65s eta=10:41:56 +2026-09-16 06:40:34 | INFO | train rank=0 | step=39640/50000 progress=79.3% loss=0.067330 lr=1.128531e-05 lr_backbone=1.128531e-05 elapsed=40:54:07 avg_step=3.71s recent_step=5.43s eta=10:41:23 +2026-09-16 06:41:05 | INFO | train rank=0 | step=39650/50000 progress=79.3% loss=0.075357 lr=1.126439e-05 lr_backbone=1.126439e-05 elapsed=40:54:38 avg_step=3.71s recent_step=3.11s eta=10:40:44 +2026-09-16 06:41:39 | INFO | train rank=0 | step=39660/50000 progress=79.3% loss=0.081013 lr=1.124349e-05 lr_backbone=1.124349e-05 elapsed=40:55:11 avg_step=3.71s recent_step=3.37s eta=10:40:06 +2026-09-16 06:42:16 | INFO | train rank=0 | step=39670/50000 progress=79.3% loss=0.075574 lr=1.122260e-05 lr_backbone=1.122260e-05 elapsed=40:55:48 avg_step=3.71s recent_step=3.69s eta=10:39:29 +2026-09-16 06:43:12 | INFO | train rank=0 | step=39680/50000 progress=79.4% loss=0.084216 lr=1.120174e-05 lr_backbone=1.120174e-05 elapsed=40:56:45 avg_step=3.71s recent_step=5.66s eta=10:38:57 +2026-09-16 06:43:44 | INFO | train rank=0 | step=39690/50000 progress=79.4% loss=0.087654 lr=1.118088e-05 lr_backbone=1.118088e-05 elapsed=40:57:16 avg_step=3.71s recent_step=3.14s eta=10:38:18 +2026-09-16 06:44:24 | INFO | train rank=0 | step=39700/50000 progress=79.4% loss=0.089138 lr=1.116005e-05 lr_backbone=1.116005e-05 elapsed=40:57:56 avg_step=3.71s recent_step=4.01s eta=10:37:42 +2026-09-16 06:44:56 | INFO | train rank=0 | step=39710/50000 progress=79.4% loss=0.090369 lr=1.113923e-05 lr_backbone=1.113923e-05 elapsed=40:58:29 avg_step=3.71s recent_step=3.26s eta=10:37:03 +2026-09-16 06:45:45 | INFO | train rank=0 | step=39720/50000 progress=79.4% loss=0.081300 lr=1.111843e-05 lr_backbone=1.111843e-05 elapsed=40:59:18 avg_step=3.71s recent_step=4.88s eta=10:36:29 +2026-09-16 06:46:17 | INFO | train rank=0 | step=39730/50000 progress=79.5% loss=0.064263 lr=1.109765e-05 lr_backbone=1.109765e-05 elapsed=40:59:50 avg_step=3.71s recent_step=3.23s eta=10:35:51 +2026-09-16 06:46:54 | INFO | train rank=0 | step=39740/50000 progress=79.5% loss=0.111754 lr=1.107688e-05 lr_backbone=1.107688e-05 elapsed=41:00:26 avg_step=3.71s recent_step=3.65s eta=10:35:14 +2026-09-16 06:47:34 | INFO | train rank=0 | step=39750/50000 progress=79.5% loss=0.098262 lr=1.105614e-05 lr_backbone=1.105614e-05 elapsed=41:01:07 avg_step=3.71s recent_step=4.07s eta=10:34:37 +2026-09-16 06:48:21 | INFO | train rank=0 | step=39760/50000 progress=79.5% loss=0.118410 lr=1.103540e-05 lr_backbone=1.103540e-05 elapsed=41:01:54 avg_step=3.72s recent_step=4.70s eta=10:34:03 +2026-09-16 06:48:55 | INFO | train rank=0 | step=39770/50000 progress=79.5% loss=0.063421 lr=1.101469e-05 lr_backbone=1.101469e-05 elapsed=41:02:28 avg_step=3.72s recent_step=3.37s eta=10:33:25 +2026-09-16 06:49:32 | INFO | train rank=0 | step=39780/50000 progress=79.6% loss=0.090818 lr=1.099399e-05 lr_backbone=1.099399e-05 elapsed=41:03:05 avg_step=3.72s recent_step=3.72s eta=10:32:48 +2026-09-16 06:49:55 | INFO | train rank=0 | step=39790/50000 progress=79.6% loss=0.082453 lr=1.097331e-05 lr_backbone=1.097331e-05 elapsed=41:03:28 avg_step=3.71s recent_step=2.27s eta=10:32:07 +2026-09-16 06:50:46 | INFO | train rank=0 | step=39800/50000 progress=79.6% loss=0.092798 lr=1.095265e-05 lr_backbone=1.095265e-05 elapsed=41:04:19 avg_step=3.72s recent_step=5.08s eta=10:31:33 +2026-09-16 06:51:20 | INFO | train rank=0 | step=39810/50000 progress=79.6% loss=0.120776 lr=1.093200e-05 lr_backbone=1.093200e-05 elapsed=41:04:53 avg_step=3.71s recent_step=3.42s eta=10:30:55 +2026-09-16 06:51:53 | INFO | train rank=0 | step=39820/50000 progress=79.6% loss=0.084290 lr=1.091137e-05 lr_backbone=1.091137e-05 elapsed=41:05:26 avg_step=3.71s recent_step=3.31s eta=10:30:17 +2026-09-16 06:52:28 | INFO | train rank=0 | step=39830/50000 progress=79.7% loss=0.085830 lr=1.089076e-05 lr_backbone=1.089076e-05 elapsed=41:06:01 avg_step=3.71s recent_step=3.47s eta=10:29:39 +2026-09-16 06:53:16 | INFO | train rank=0 | step=39840/50000 progress=79.7% loss=0.088892 lr=1.087016e-05 lr_backbone=1.087016e-05 elapsed=41:06:49 avg_step=3.72s recent_step=4.84s eta=10:29:05 +2026-09-16 06:53:52 | INFO | train rank=0 | step=39850/50000 progress=79.7% loss=0.075868 lr=1.084959e-05 lr_backbone=1.084959e-05 elapsed=41:07:25 avg_step=3.72s recent_step=3.57s eta=10:28:27 +2026-09-16 06:54:20 | INFO | train rank=0 | step=39860/50000 progress=79.7% loss=0.087137 lr=1.082902e-05 lr_backbone=1.082902e-05 elapsed=41:07:52 avg_step=3.71s recent_step=2.77s eta=10:27:48 +2026-09-16 06:54:51 | INFO | train rank=0 | step=39870/50000 progress=79.7% loss=0.100966 lr=1.080848e-05 lr_backbone=1.080848e-05 elapsed=41:08:24 avg_step=3.71s recent_step=3.16s eta=10:27:09 +2026-09-16 06:55:40 | INFO | train rank=0 | step=39880/50000 progress=79.8% loss=0.086161 lr=1.078795e-05 lr_backbone=1.078795e-05 elapsed=41:09:13 avg_step=3.71s recent_step=4.94s eta=10:26:35 +2026-09-16 06:56:16 | INFO | train rank=0 | step=39890/50000 progress=79.8% loss=0.074688 lr=1.076744e-05 lr_backbone=1.076744e-05 elapsed=41:09:49 avg_step=3.71s recent_step=3.60s eta=10:25:58 +2026-09-16 06:56:40 | INFO | train rank=0 | step=39900/50000 progress=79.8% loss=0.110784 lr=1.074695e-05 lr_backbone=1.074695e-05 elapsed=41:10:13 avg_step=3.71s recent_step=2.37s eta=10:25:17 +2026-09-16 06:57:15 | INFO | train rank=0 | step=39910/50000 progress=79.8% loss=0.079553 lr=1.072648e-05 lr_backbone=1.072648e-05 elapsed=41:10:48 avg_step=3.71s recent_step=3.50s eta=10:24:39 +2026-09-16 06:58:03 | INFO | train rank=0 | step=39920/50000 progress=79.8% loss=0.104718 lr=1.070602e-05 lr_backbone=1.070602e-05 elapsed=41:11:36 avg_step=3.71s recent_step=4.83s eta=10:24:05 +2026-09-16 06:58:35 | INFO | train rank=0 | step=39930/50000 progress=79.9% loss=0.081631 lr=1.068558e-05 lr_backbone=1.068558e-05 elapsed=41:12:08 avg_step=3.71s recent_step=3.14s eta=10:23:27 +2026-09-16 06:59:10 | INFO | train rank=0 | step=39940/50000 progress=79.9% loss=0.086758 lr=1.066515e-05 lr_backbone=1.066515e-05 elapsed=41:12:43 avg_step=3.71s recent_step=3.52s eta=10:22:49 +2026-09-16 06:59:40 | INFO | train rank=0 | step=39950/50000 progress=79.9% loss=0.120378 lr=1.064475e-05 lr_backbone=1.064475e-05 elapsed=41:13:12 avg_step=3.71s recent_step=2.96s eta=10:22:10 +2026-09-16 07:00:30 | INFO | train rank=0 | step=39960/50000 progress=79.9% loss=0.083516 lr=1.062436e-05 lr_backbone=1.062436e-05 elapsed=41:14:02 avg_step=3.71s recent_step=5.00s eta=10:21:36 +2026-09-16 07:01:04 | INFO | train rank=0 | step=39970/50000 progress=79.9% loss=0.095338 lr=1.060399e-05 lr_backbone=1.060399e-05 elapsed=41:14:36 avg_step=3.71s recent_step=3.39s eta=10:20:58 +2026-09-16 07:01:43 | INFO | train rank=0 | step=39980/50000 progress=80.0% loss=0.074405 lr=1.058363e-05 lr_backbone=1.058363e-05 elapsed=41:15:16 avg_step=3.71s recent_step=3.93s eta=10:20:21 +2026-09-16 07:02:16 | INFO | train rank=0 | step=39990/50000 progress=80.0% loss=0.085966 lr=1.056329e-05 lr_backbone=1.056329e-05 elapsed=41:15:49 avg_step=3.71s recent_step=3.29s eta=10:19:43 +2026-09-16 07:02:55 | INFO | train rank=0 | step=40000/50000 progress=80.0% loss=0.095449 lr=1.054297e-05 lr_backbone=1.054297e-05 elapsed=41:16:28 avg_step=3.71s recent_step=3.92s eta=10:19:07 +2026-09-16 07:03:23 | INFO | train rank=0 | step=40010/50000 progress=80.0% loss=0.073250 lr=1.052267e-05 lr_backbone=1.052267e-05 elapsed=41:16:56 avg_step=3.71s recent_step=2.81s eta=10:18:27 +2026-09-16 07:03:56 | INFO | train rank=0 | step=40020/50000 progress=80.0% loss=0.092150 lr=1.050239e-05 lr_backbone=1.050239e-05 elapsed=41:17:28 avg_step=3.71s recent_step=3.25s eta=10:17:49 +2026-09-16 07:04:27 | INFO | train rank=0 | step=40030/50000 progress=80.1% loss=0.125587 lr=1.048212e-05 lr_backbone=1.048212e-05 elapsed=41:17:59 avg_step=3.71s recent_step=3.10s eta=10:17:10 +2026-09-16 07:05:05 | INFO | train rank=0 | step=40040/50000 progress=80.1% loss=0.087272 lr=1.046187e-05 lr_backbone=1.046187e-05 elapsed=41:18:38 avg_step=3.71s recent_step=3.87s eta=10:16:33 +2026-09-16 07:05:39 | INFO | train rank=0 | step=40050/50000 progress=80.1% loss=0.079285 lr=1.044163e-05 lr_backbone=1.044163e-05 elapsed=41:19:11 avg_step=3.71s recent_step=3.33s eta=10:15:55 +2026-09-16 07:06:12 | INFO | train rank=0 | step=40060/50000 progress=80.1% loss=0.117911 lr=1.042142e-05 lr_backbone=1.042142e-05 elapsed=41:19:45 avg_step=3.71s recent_step=3.37s eta=10:15:17 +2026-09-16 07:06:44 | INFO | train rank=0 | step=40070/50000 progress=80.1% loss=0.081079 lr=1.040122e-05 lr_backbone=1.040122e-05 elapsed=41:20:17 avg_step=3.71s recent_step=3.21s eta=10:14:39 +2026-09-16 07:07:35 | INFO | train rank=0 | step=40080/50000 progress=80.2% loss=0.079216 lr=1.038103e-05 lr_backbone=1.038103e-05 elapsed=41:21:08 avg_step=3.71s recent_step=5.09s eta=10:14:05 +2026-09-16 07:08:06 | INFO | train rank=0 | step=40090/50000 progress=80.2% loss=0.087341 lr=1.036087e-05 lr_backbone=1.036087e-05 elapsed=41:21:39 avg_step=3.71s recent_step=3.08s eta=10:13:26 +2026-09-16 07:08:40 | INFO | train rank=0 | step=40100/50000 progress=80.2% loss=0.098726 lr=1.034072e-05 lr_backbone=1.034072e-05 elapsed=41:22:13 avg_step=3.71s recent_step=3.39s eta=10:12:49 +2026-09-16 07:09:28 | INFO | train rank=0 | step=40110/50000 progress=80.2% loss=0.073238 lr=1.032059e-05 lr_backbone=1.032059e-05 elapsed=41:23:01 avg_step=3.71s recent_step=4.86s eta=10:12:14 +2026-09-16 07:10:09 | INFO | train rank=0 | step=40120/50000 progress=80.2% loss=0.073757 lr=1.030048e-05 lr_backbone=1.030048e-05 elapsed=41:23:42 avg_step=3.71s recent_step=4.09s eta=10:11:38 +2026-09-16 07:10:41 | INFO | train rank=0 | step=40130/50000 progress=80.3% loss=0.087703 lr=1.028038e-05 lr_backbone=1.028038e-05 elapsed=41:24:14 avg_step=3.71s recent_step=3.19s eta=10:11:00 +2026-09-16 07:11:17 | INFO | train rank=0 | step=40140/50000 progress=80.3% loss=0.099551 lr=1.026031e-05 lr_backbone=1.026031e-05 elapsed=41:24:50 avg_step=3.71s recent_step=3.58s eta=10:10:22 +2026-09-16 07:11:53 | INFO | train rank=0 | step=40150/50000 progress=80.3% loss=0.103190 lr=1.024025e-05 lr_backbone=1.024025e-05 elapsed=41:25:26 avg_step=3.71s recent_step=3.59s eta=10:09:45 +2026-09-16 07:12:39 | INFO | train rank=0 | step=40160/50000 progress=80.3% loss=0.076063 lr=1.022020e-05 lr_backbone=1.022020e-05 elapsed=41:26:12 avg_step=3.71s recent_step=4.60s eta=10:09:10 +2026-09-16 07:13:17 | INFO | train rank=0 | step=40170/50000 progress=80.3% loss=0.100739 lr=1.020018e-05 lr_backbone=1.020018e-05 elapsed=41:26:50 avg_step=3.71s recent_step=3.83s eta=10:08:33 +2026-09-16 07:13:53 | INFO | train rank=0 | step=40180/50000 progress=80.4% loss=0.114781 lr=1.018017e-05 lr_backbone=1.018017e-05 elapsed=41:27:26 avg_step=3.71s recent_step=3.61s eta=10:07:55 +2026-09-16 07:14:28 | INFO | train rank=0 | step=40190/50000 progress=80.4% loss=0.111800 lr=1.016018e-05 lr_backbone=1.016018e-05 elapsed=41:28:00 avg_step=3.71s recent_step=3.41s eta=10:07:18 +2026-09-16 07:15:21 | INFO | train rank=0 | step=40200/50000 progress=80.4% loss=0.107748 lr=1.014021e-05 lr_backbone=1.014021e-05 elapsed=41:28:54 avg_step=3.71s recent_step=5.32s eta=10:06:44 +2026-09-16 07:15:50 | INFO | train rank=0 | step=40210/50000 progress=80.4% loss=0.088507 lr=1.012025e-05 lr_backbone=1.012025e-05 elapsed=41:29:23 avg_step=3.71s recent_step=2.95s eta=10:06:05 +2026-09-16 07:16:22 | INFO | train rank=0 | step=40220/50000 progress=80.4% loss=0.104562 lr=1.010031e-05 lr_backbone=1.010031e-05 elapsed=41:29:55 avg_step=3.71s recent_step=3.17s eta=10:05:27 +2026-09-16 07:16:55 | INFO | train rank=0 | step=40230/50000 progress=80.5% loss=0.114458 lr=1.008039e-05 lr_backbone=1.008039e-05 elapsed=41:30:27 avg_step=3.71s recent_step=3.26s eta=10:04:49 +2026-09-16 07:17:30 | INFO | train rank=0 | step=40240/50000 progress=80.5% loss=0.104226 lr=1.006049e-05 lr_backbone=1.006049e-05 elapsed=41:31:03 avg_step=3.71s recent_step=3.51s eta=10:04:11 +2026-09-16 07:17:57 | INFO | train rank=0 | step=40250/50000 progress=80.5% loss=0.064491 lr=1.004060e-05 lr_backbone=1.004060e-05 elapsed=41:31:30 avg_step=3.71s recent_step=2.72s eta=10:03:31 +2026-09-16 07:18:30 | INFO | train rank=0 | step=40260/50000 progress=80.5% loss=0.125699 lr=1.002073e-05 lr_backbone=1.002073e-05 elapsed=41:32:03 avg_step=3.71s recent_step=3.33s eta=10:02:53 +2026-09-16 07:18:59 | INFO | train rank=0 | step=40270/50000 progress=80.5% loss=0.095073 lr=1.000088e-05 lr_backbone=1.000088e-05 elapsed=41:32:32 avg_step=3.71s recent_step=2.92s eta=10:02:14 +2026-09-16 07:19:30 | INFO | train rank=0 | step=40280/50000 progress=80.6% loss=0.091958 lr=9.981047e-06 lr_backbone=9.981047e-06 elapsed=41:33:03 avg_step=3.71s recent_step=3.05s eta=10:01:36 +2026-09-16 07:20:06 | INFO | train rank=0 | step=40290/50000 progress=80.6% loss=0.107548 lr=9.961231e-06 lr_backbone=9.961231e-06 elapsed=41:33:38 avg_step=3.71s recent_step=3.57s eta=10:00:58 +2026-09-16 07:20:44 | INFO | train rank=0 | step=40300/50000 progress=80.6% loss=0.098339 lr=9.941432e-06 lr_backbone=9.941432e-06 elapsed=41:34:16 avg_step=3.71s recent_step=3.80s eta=10:00:21 +2026-09-16 07:21:12 | INFO | train rank=0 | step=40310/50000 progress=80.6% loss=0.083846 lr=9.921651e-06 lr_backbone=9.921651e-06 elapsed=41:34:45 avg_step=3.71s recent_step=2.83s eta=09:59:42 +2026-09-16 07:21:47 | INFO | train rank=0 | step=40320/50000 progress=80.6% loss=0.082229 lr=9.901887e-06 lr_backbone=9.901887e-06 elapsed=41:35:20 avg_step=3.71s recent_step=3.52s eta=09:59:04 +2026-09-16 07:22:24 | INFO | train rank=0 | step=40330/50000 progress=80.7% loss=0.092997 lr=9.882141e-06 lr_backbone=9.882141e-06 elapsed=41:35:57 avg_step=3.71s recent_step=3.67s eta=09:58:27 +2026-09-16 07:23:10 | INFO | train rank=0 | step=40340/50000 progress=80.7% loss=0.136039 lr=9.862413e-06 lr_backbone=9.862413e-06 elapsed=41:36:43 avg_step=3.71s recent_step=4.67s eta=09:57:52 +2026-09-16 07:24:04 | INFO | train rank=0 | step=40350/50000 progress=80.7% loss=0.062557 lr=9.842702e-06 lr_backbone=9.842702e-06 elapsed=41:37:36 avg_step=3.71s recent_step=5.32s eta=09:57:19 +2026-09-16 07:24:51 | INFO | train rank=0 | step=40360/50000 progress=80.7% loss=0.108061 lr=9.823008e-06 lr_backbone=9.823008e-06 elapsed=41:38:24 avg_step=3.71s recent_step=4.72s eta=09:56:44 +2026-09-16 07:25:52 | INFO | train rank=0 | step=40370/50000 progress=80.7% loss=0.083331 lr=9.803332e-06 lr_backbone=9.803332e-06 elapsed=41:39:25 avg_step=3.71s recent_step=6.11s eta=09:56:13 +2026-09-16 07:26:59 | INFO | train rank=0 | step=40380/50000 progress=80.8% loss=0.066700 lr=9.783674e-06 lr_backbone=9.783674e-06 elapsed=41:40:32 avg_step=3.72s recent_step=6.69s eta=09:55:43 +2026-09-16 07:28:12 | INFO | train rank=0 | step=40390/50000 progress=80.8% loss=0.089013 lr=9.764034e-06 lr_backbone=9.764034e-06 elapsed=41:41:45 avg_step=3.72s recent_step=7.33s eta=09:55:14 +2026-09-16 07:29:24 | INFO | train rank=0 | step=40400/50000 progress=80.8% loss=0.077161 lr=9.744411e-06 lr_backbone=9.744411e-06 elapsed=41:42:57 avg_step=3.72s recent_step=7.18s eta=09:54:45 +2026-09-16 07:30:31 | INFO | train rank=0 | step=40410/50000 progress=80.8% loss=0.089108 lr=9.724805e-06 lr_backbone=9.724805e-06 elapsed=41:44:04 avg_step=3.72s recent_step=6.73s eta=09:54:15 +2026-09-16 07:31:34 | INFO | train rank=0 | step=40420/50000 progress=80.8% loss=0.083601 lr=9.705217e-06 lr_backbone=9.705217e-06 elapsed=41:45:06 avg_step=3.72s recent_step=6.23s eta=09:53:44 +2026-09-16 07:32:41 | INFO | train rank=0 | step=40430/50000 progress=80.9% loss=0.080984 lr=9.685647e-06 lr_backbone=9.685647e-06 elapsed=41:46:13 avg_step=3.72s recent_step=6.70s eta=09:53:14 +2026-09-16 07:33:54 | INFO | train rank=0 | step=40440/50000 progress=80.9% loss=0.079644 lr=9.666095e-06 lr_backbone=9.666095e-06 elapsed=41:47:27 avg_step=3.72s recent_step=7.38s eta=09:52:45 +2026-09-16 07:34:57 | INFO | train rank=0 | step=40450/50000 progress=80.9% loss=0.071174 lr=9.646560e-06 lr_backbone=9.646560e-06 elapsed=41:48:29 avg_step=3.72s recent_step=6.23s eta=09:52:14 +2026-09-16 07:36:06 | INFO | train rank=0 | step=40460/50000 progress=80.9% loss=0.072308 lr=9.627043e-06 lr_backbone=9.627043e-06 elapsed=41:49:39 avg_step=3.72s recent_step=6.90s eta=09:51:44 +2026-09-16 07:37:07 | INFO | train rank=0 | step=40470/50000 progress=80.9% loss=0.086718 lr=9.607543e-06 lr_backbone=9.607543e-06 elapsed=41:50:40 avg_step=3.72s recent_step=6.12s eta=09:51:13 +2026-09-16 07:38:17 | INFO | train rank=0 | step=40480/50000 progress=81.0% loss=0.134888 lr=9.588061e-06 lr_backbone=9.588061e-06 elapsed=41:51:49 avg_step=3.72s recent_step=6.96s eta=09:50:43 +2026-09-16 07:39:24 | INFO | train rank=0 | step=40490/50000 progress=81.0% loss=0.112793 lr=9.568597e-06 lr_backbone=9.568597e-06 elapsed=41:52:56 avg_step=3.72s recent_step=6.70s eta=09:50:13 +2026-09-16 07:40:41 | INFO | train rank=0 | step=40500/50000 progress=81.0% loss=0.063799 lr=9.549150e-06 lr_backbone=9.549150e-06 elapsed=41:54:14 avg_step=3.72s recent_step=7.77s eta=09:49:45 +2026-09-16 07:41:53 | INFO | train rank=0 | step=40510/50000 progress=81.0% loss=0.079807 lr=9.529721e-06 lr_backbone=9.529721e-06 elapsed=41:55:26 avg_step=3.73s recent_step=7.20s eta=09:49:16 +2026-09-16 07:42:52 | INFO | train rank=0 | step=40520/50000 progress=81.0% loss=0.061738 lr=9.510310e-06 lr_backbone=9.510310e-06 elapsed=41:56:25 avg_step=3.73s recent_step=5.86s eta=09:48:44 +2026-09-16 07:44:01 | INFO | train rank=0 | step=40530/50000 progress=81.1% loss=0.094980 lr=9.490917e-06 lr_backbone=9.490917e-06 elapsed=41:57:34 avg_step=3.73s recent_step=6.96s eta=09:48:14 +2026-09-16 07:45:12 | INFO | train rank=0 | step=40540/50000 progress=81.1% loss=0.066390 lr=9.471541e-06 lr_backbone=9.471541e-06 elapsed=41:58:45 avg_step=3.73s recent_step=7.03s eta=09:47:44 +2026-09-16 07:46:09 | INFO | train rank=0 | step=40550/50000 progress=81.1% loss=0.091498 lr=9.452183e-06 lr_backbone=9.452183e-06 elapsed=41:59:42 avg_step=3.73s recent_step=5.76s eta=09:47:12 +2026-09-16 07:47:20 | INFO | train rank=0 | step=40560/50000 progress=81.1% loss=0.103735 lr=9.432843e-06 lr_backbone=9.432843e-06 elapsed=42:00:53 avg_step=3.73s recent_step=7.05s eta=09:46:42 +2026-09-16 07:48:32 | INFO | train rank=0 | step=40570/50000 progress=81.1% loss=0.086889 lr=9.413520e-06 lr_backbone=9.413520e-06 elapsed=42:02:04 avg_step=3.73s recent_step=7.18s eta=09:46:13 +2026-09-16 07:49:32 | INFO | train rank=0 | step=40580/50000 progress=81.2% loss=0.108440 lr=9.394216e-06 lr_backbone=9.394216e-06 elapsed=42:03:05 avg_step=3.73s recent_step=6.04s eta=09:45:41 +2026-09-16 07:50:35 | INFO | train rank=0 | step=40590/50000 progress=81.2% loss=0.087967 lr=9.374929e-06 lr_backbone=9.374929e-06 elapsed=42:04:08 avg_step=3.73s recent_step=6.26s eta=09:45:10 +2026-09-16 07:51:44 | INFO | train rank=0 | step=40600/50000 progress=81.2% loss=0.119029 lr=9.355659e-06 lr_backbone=9.355659e-06 elapsed=42:05:17 avg_step=3.73s recent_step=6.94s eta=09:44:40 +2026-09-16 07:52:52 | INFO | train rank=0 | step=40610/50000 progress=81.2% loss=0.114920 lr=9.336408e-06 lr_backbone=9.336408e-06 elapsed=42:06:25 avg_step=3.73s recent_step=6.79s eta=09:44:10 +2026-09-16 07:54:12 | INFO | train rank=0 | step=40620/50000 progress=81.2% loss=0.062488 lr=9.317174e-06 lr_backbone=9.317174e-06 elapsed=42:07:45 avg_step=3.73s recent_step=7.97s eta=09:43:42 +2026-09-16 07:55:23 | INFO | train rank=0 | step=40630/50000 progress=81.3% loss=0.088876 lr=9.297958e-06 lr_backbone=9.297958e-06 elapsed=42:08:56 avg_step=3.73s recent_step=7.13s eta=09:43:13 +2026-09-16 07:56:36 | INFO | train rank=0 | step=40640/50000 progress=81.3% loss=0.105489 lr=9.278760e-06 lr_backbone=9.278760e-06 elapsed=42:10:09 avg_step=3.74s recent_step=7.34s eta=09:42:44 +2026-09-16 07:57:40 | INFO | train rank=0 | step=40650/50000 progress=81.3% loss=0.083666 lr=9.259580e-06 lr_backbone=9.259580e-06 elapsed=42:11:12 avg_step=3.74s recent_step=6.31s eta=09:42:12 +2026-09-16 07:58:41 | INFO | train rank=0 | step=40660/50000 progress=81.3% loss=0.095321 lr=9.240418e-06 lr_backbone=9.240418e-06 elapsed=42:12:13 avg_step=3.74s recent_step=6.11s eta=09:41:40 +2026-09-16 07:59:48 | INFO | train rank=0 | step=40670/50000 progress=81.3% loss=0.091367 lr=9.221273e-06 lr_backbone=9.221273e-06 elapsed=42:13:21 avg_step=3.74s recent_step=6.78s eta=09:41:10 +2026-09-16 08:00:52 | INFO | train rank=0 | step=40680/50000 progress=81.4% loss=0.115027 lr=9.202146e-06 lr_backbone=9.202146e-06 elapsed=42:14:24 avg_step=3.74s recent_step=6.31s eta=09:40:38 +2026-09-16 08:02:07 | INFO | train rank=0 | step=40690/50000 progress=81.4% loss=0.082921 lr=9.183037e-06 lr_backbone=9.183037e-06 elapsed=42:15:40 avg_step=3.74s recent_step=7.55s eta=09:40:10 +2026-09-16 08:03:20 | INFO | train rank=0 | step=40700/50000 progress=81.4% loss=0.067965 lr=9.163946e-06 lr_backbone=9.163946e-06 elapsed=42:16:53 avg_step=3.74s recent_step=7.29s eta=09:39:40 +2026-09-16 08:04:18 | INFO | train rank=0 | step=40710/50000 progress=81.4% loss=0.104910 lr=9.144873e-06 lr_backbone=9.144873e-06 elapsed=42:17:50 avg_step=3.74s recent_step=5.77s eta=09:39:08 +2026-09-16 08:05:21 | INFO | train rank=0 | step=40720/50000 progress=81.4% loss=0.102934 lr=9.125818e-06 lr_backbone=9.125818e-06 elapsed=42:18:53 avg_step=3.74s recent_step=6.30s eta=09:38:36 +2026-09-16 08:06:22 | INFO | train rank=0 | step=40730/50000 progress=81.5% loss=0.118109 lr=9.106780e-06 lr_backbone=9.106780e-06 elapsed=42:19:55 avg_step=3.74s recent_step=6.16s eta=09:38:04 +2026-09-16 08:08:01 | INFO | train rank=0 | step=40740/50000 progress=81.5% loss=0.071189 lr=9.087761e-06 lr_backbone=9.087761e-06 elapsed=42:21:34 avg_step=3.74s recent_step=9.86s eta=09:37:41 +2026-09-16 08:09:59 | INFO | train rank=0 | step=40750/50000 progress=81.5% loss=0.093898 lr=9.068759e-06 lr_backbone=9.068759e-06 elapsed=42:23:32 avg_step=3.75s recent_step=11.79s eta=09:37:21 +2026-09-16 08:12:06 | INFO | train rank=0 | step=40760/50000 progress=81.5% loss=0.097127 lr=9.049776e-06 lr_backbone=9.049776e-06 elapsed=42:25:39 avg_step=3.75s recent_step=12.72s eta=09:37:04 +2026-09-16 08:14:05 | INFO | train rank=0 | step=40770/50000 progress=81.5% loss=0.104992 lr=9.030810e-06 lr_backbone=9.030810e-06 elapsed=42:27:38 avg_step=3.75s recent_step=11.94s eta=09:36:45 +2026-09-16 08:15:54 | INFO | train rank=0 | step=40780/50000 progress=81.6% loss=0.089921 lr=9.011862e-06 lr_backbone=9.011862e-06 elapsed=42:29:27 avg_step=3.75s recent_step=10.89s eta=09:36:24 +2026-09-16 08:17:48 | INFO | train rank=0 | step=40790/50000 progress=81.6% loss=0.089230 lr=8.992932e-06 lr_backbone=8.992932e-06 elapsed=42:31:21 avg_step=3.75s recent_step=11.38s eta=09:36:04 +2026-09-16 08:19:23 | INFO | train rank=0 | step=40800/50000 progress=81.6% loss=0.119472 lr=8.974020e-06 lr_backbone=8.974020e-06 elapsed=42:32:56 avg_step=3.75s recent_step=9.54s eta=09:35:39 +2026-09-16 08:20:52 | INFO | train rank=0 | step=40810/50000 progress=81.6% loss=0.073908 lr=8.955126e-06 lr_backbone=8.955126e-06 elapsed=42:34:25 avg_step=3.76s recent_step=8.89s eta=09:35:13 +2026-09-16 08:23:48 | INFO | train rank=0 | step=40820/50000 progress=81.6% loss=0.096311 lr=8.936249e-06 lr_backbone=8.936249e-06 elapsed=42:37:21 avg_step=3.76s recent_step=17.59s eta=09:35:07 +2026-09-16 08:24:16 | INFO | train rank=0 | step=40830/50000 progress=81.7% loss=0.089861 lr=8.917391e-06 lr_backbone=8.917391e-06 elapsed=42:37:49 avg_step=3.76s recent_step=2.83s eta=09:34:27 +2026-09-16 08:24:46 | INFO | train rank=0 | step=40840/50000 progress=81.7% loss=0.097777 lr=8.898551e-06 lr_backbone=8.898551e-06 elapsed=42:38:18 avg_step=3.76s recent_step=2.92s eta=09:33:48 +2026-09-16 08:25:12 | INFO | train rank=0 | step=40850/50000 progress=81.7% loss=0.084653 lr=8.879729e-06 lr_backbone=8.879729e-06 elapsed=42:38:45 avg_step=3.76s recent_step=2.64s eta=09:33:08 +2026-09-16 08:25:53 | INFO | train rank=0 | step=40860/50000 progress=81.7% loss=0.082495 lr=8.860925e-06 lr_backbone=8.860925e-06 elapsed=42:39:25 avg_step=3.76s recent_step=4.06s eta=09:32:31 +2026-09-16 08:26:23 | INFO | train rank=0 | step=40870/50000 progress=81.7% loss=0.097896 lr=8.842138e-06 lr_backbone=8.842138e-06 elapsed=42:39:56 avg_step=3.76s recent_step=3.06s eta=09:31:52 +2026-09-16 08:26:45 | INFO | train rank=0 | step=40880/50000 progress=81.8% loss=0.084936 lr=8.823370e-06 lr_backbone=8.823370e-06 elapsed=42:40:18 avg_step=3.76s recent_step=2.22s eta=09:31:11 +2026-09-16 08:27:11 | INFO | train rank=0 | step=40890/50000 progress=81.8% loss=0.077682 lr=8.804620e-06 lr_backbone=8.804620e-06 elapsed=42:40:44 avg_step=3.76s recent_step=2.56s eta=09:30:30 +2026-09-16 08:27:46 | INFO | train rank=0 | step=40900/50000 progress=81.8% loss=0.106097 lr=8.785888e-06 lr_backbone=8.785888e-06 elapsed=42:41:19 avg_step=3.76s recent_step=3.47s eta=09:29:52 +2026-09-16 08:28:14 | INFO | train rank=0 | step=40910/50000 progress=81.8% loss=0.086117 lr=8.767173e-06 lr_backbone=8.767173e-06 elapsed=42:41:47 avg_step=3.76s recent_step=2.83s eta=09:29:13 +2026-09-16 08:28:41 | INFO | train rank=0 | step=40920/50000 progress=81.8% loss=0.074544 lr=8.748477e-06 lr_backbone=8.748477e-06 elapsed=42:42:14 avg_step=3.76s recent_step=2.67s eta=09:28:33 +2026-09-16 08:29:08 | INFO | train rank=0 | step=40930/50000 progress=81.9% loss=0.097569 lr=8.729799e-06 lr_backbone=8.729799e-06 elapsed=42:42:41 avg_step=3.76s recent_step=2.74s eta=09:27:53 +2026-09-16 08:29:59 | INFO | train rank=0 | step=40940/50000 progress=81.9% loss=0.086739 lr=8.711139e-06 lr_backbone=8.711139e-06 elapsed=42:43:31 avg_step=3.76s recent_step=5.03s eta=09:27:18 +2026-09-16 08:30:27 | INFO | train rank=0 | step=40950/50000 progress=81.9% loss=0.062781 lr=8.692497e-06 lr_backbone=8.692497e-06 elapsed=42:44:00 avg_step=3.76s recent_step=2.83s eta=09:26:38 +2026-09-16 08:30:50 | INFO | train rank=0 | step=40960/50000 progress=81.9% loss=0.080081 lr=8.673873e-06 lr_backbone=8.673873e-06 elapsed=42:44:23 avg_step=3.76s recent_step=2.35s eta=09:25:58 +2026-09-16 08:31:17 | INFO | train rank=0 | step=40970/50000 progress=81.9% loss=0.086691 lr=8.655267e-06 lr_backbone=8.655267e-06 elapsed=42:44:50 avg_step=3.76s recent_step=2.68s eta=09:25:18 +2026-09-16 08:31:50 | INFO | train rank=0 | step=40980/50000 progress=82.0% loss=0.104808 lr=8.636680e-06 lr_backbone=8.636680e-06 elapsed=42:45:23 avg_step=3.76s recent_step=3.26s eta=09:24:39 +2026-09-16 08:32:15 | INFO | train rank=0 | step=40990/50000 progress=82.0% loss=0.078605 lr=8.618110e-06 lr_backbone=8.618110e-06 elapsed=42:45:48 avg_step=3.76s recent_step=2.52s eta=09:23:59 +2026-09-16 08:32:38 | INFO | train rank=0 | step=41000/50000 progress=82.0% loss=0.063811 lr=8.599558e-06 lr_backbone=8.599558e-06 elapsed=42:46:11 avg_step=3.76s recent_step=2.35s eta=09:23:18 +2026-09-16 08:33:03 | INFO | train rank=0 | step=41010/50000 progress=82.0% loss=0.057462 lr=8.581025e-06 lr_backbone=8.581025e-06 elapsed=42:46:36 avg_step=3.76s recent_step=2.50s eta=09:22:38 +2026-09-16 08:33:47 | INFO | train rank=0 | step=41020/50000 progress=82.0% loss=0.073087 lr=8.562510e-06 lr_backbone=8.562510e-06 elapsed=42:47:20 avg_step=3.76s recent_step=4.36s eta=09:22:02 +2026-09-16 08:34:10 | INFO | train rank=0 | step=41030/50000 progress=82.1% loss=0.087969 lr=8.544012e-06 lr_backbone=8.544012e-06 elapsed=42:47:43 avg_step=3.75s recent_step=2.29s eta=09:21:21 +2026-09-16 08:34:34 | INFO | train rank=0 | step=41040/50000 progress=82.1% loss=0.104023 lr=8.525533e-06 lr_backbone=8.525533e-06 elapsed=42:48:07 avg_step=3.75s recent_step=2.39s eta=09:20:40 +2026-09-16 08:35:02 | INFO | train rank=0 | step=41050/50000 progress=82.1% loss=0.101849 lr=8.507072e-06 lr_backbone=8.507072e-06 elapsed=42:48:34 avg_step=3.75s recent_step=2.79s eta=09:20:01 +2026-09-16 08:35:41 | INFO | train rank=0 | step=41060/50000 progress=82.1% loss=0.060668 lr=8.488630e-06 lr_backbone=8.488630e-06 elapsed=42:49:14 avg_step=3.75s recent_step=3.96s eta=09:19:24 +2026-09-16 08:36:07 | INFO | train rank=0 | step=41070/50000 progress=82.1% loss=0.063202 lr=8.470205e-06 lr_backbone=8.470205e-06 elapsed=42:49:40 avg_step=3.75s recent_step=2.61s eta=09:18:44 +2026-09-16 08:36:31 | INFO | train rank=0 | step=41080/50000 progress=82.2% loss=0.116111 lr=8.451799e-06 lr_backbone=8.451799e-06 elapsed=42:50:04 avg_step=3.75s recent_step=2.34s eta=09:18:03 +2026-09-16 08:37:03 | INFO | train rank=0 | step=41090/50000 progress=82.2% loss=0.074152 lr=8.433410e-06 lr_backbone=8.433410e-06 elapsed=42:50:36 avg_step=3.75s recent_step=3.20s eta=09:17:24 +2026-09-16 08:37:36 | INFO | train rank=0 | step=41100/50000 progress=82.2% loss=0.079351 lr=8.415040e-06 lr_backbone=8.415040e-06 elapsed=42:51:09 avg_step=3.75s recent_step=3.35s eta=09:16:46 +2026-09-16 08:38:02 | INFO | train rank=0 | step=41110/50000 progress=82.2% loss=0.122202 lr=8.396688e-06 lr_backbone=8.396688e-06 elapsed=42:51:35 avg_step=3.75s recent_step=2.55s eta=09:16:06 +2026-09-16 08:38:31 | INFO | train rank=0 | step=41120/50000 progress=82.2% loss=0.082401 lr=8.378355e-06 lr_backbone=8.378355e-06 elapsed=42:52:04 avg_step=3.75s recent_step=2.91s eta=09:15:26 +2026-09-16 08:38:55 | INFO | train rank=0 | step=41130/50000 progress=82.3% loss=0.097232 lr=8.360039e-06 lr_backbone=8.360039e-06 elapsed=42:52:27 avg_step=3.75s recent_step=2.38s eta=09:14:46 +2026-09-16 08:39:34 | INFO | train rank=0 | step=41140/50000 progress=82.3% loss=0.110344 lr=8.341742e-06 lr_backbone=8.341742e-06 elapsed=42:53:06 avg_step=3.75s recent_step=3.90s eta=09:14:09 +2026-09-16 08:39:56 | INFO | train rank=0 | step=41150/50000 progress=82.3% loss=0.084556 lr=8.323463e-06 lr_backbone=8.323463e-06 elapsed=42:53:29 avg_step=3.75s recent_step=2.23s eta=09:13:28 +2026-09-16 08:40:24 | INFO | train rank=0 | step=41160/50000 progress=82.3% loss=0.104596 lr=8.305202e-06 lr_backbone=8.305202e-06 elapsed=42:53:57 avg_step=3.75s recent_step=2.84s eta=09:12:48 +2026-09-16 08:40:50 | INFO | train rank=0 | step=41170/50000 progress=82.3% loss=0.103304 lr=8.286959e-06 lr_backbone=8.286959e-06 elapsed=42:54:23 avg_step=3.75s recent_step=2.54s eta=09:12:08 +2026-09-16 08:41:24 | INFO | train rank=0 | step=41180/50000 progress=82.4% loss=0.068541 lr=8.268735e-06 lr_backbone=8.268735e-06 elapsed=42:54:57 avg_step=3.75s recent_step=3.43s eta=09:11:30 +2026-09-16 08:41:54 | INFO | train rank=0 | step=41190/50000 progress=82.4% loss=0.161686 lr=8.250529e-06 lr_backbone=8.250529e-06 elapsed=42:55:26 avg_step=3.75s recent_step=2.97s eta=09:10:51 +2026-09-16 08:42:14 | INFO | train rank=0 | step=41200/50000 progress=82.4% loss=0.095732 lr=8.232341e-06 lr_backbone=8.232341e-06 elapsed=42:55:47 avg_step=3.75s recent_step=2.02s eta=09:10:10 +2026-09-16 08:42:44 | INFO | train rank=0 | step=41210/50000 progress=82.4% loss=0.097190 lr=8.214171e-06 lr_backbone=8.214171e-06 elapsed=42:56:16 avg_step=3.75s recent_step=2.97s eta=09:09:30 +2026-09-16 08:43:14 | INFO | train rank=0 | step=41220/50000 progress=82.4% loss=0.115216 lr=8.196020e-06 lr_backbone=8.196020e-06 elapsed=42:56:47 avg_step=3.75s recent_step=3.03s eta=09:08:51 +2026-09-16 08:43:35 | INFO | train rank=0 | step=41230/50000 progress=82.5% loss=0.079500 lr=8.177887e-06 lr_backbone=8.177887e-06 elapsed=42:57:08 avg_step=3.75s recent_step=2.11s eta=09:08:10 +2026-09-16 08:44:00 | INFO | train rank=0 | step=41240/50000 progress=82.5% loss=0.138480 lr=8.159772e-06 lr_backbone=8.159772e-06 elapsed=42:57:33 avg_step=3.75s recent_step=2.54s eta=09:07:30 +2026-09-16 08:44:27 | INFO | train rank=0 | step=41250/50000 progress=82.5% loss=0.081967 lr=8.141676e-06 lr_backbone=8.141676e-06 elapsed=42:58:00 avg_step=3.75s recent_step=2.67s eta=09:06:50 +2026-09-16 08:45:05 | INFO | train rank=0 | step=41260/50000 progress=82.5% loss=0.096753 lr=8.123598e-06 lr_backbone=8.123598e-06 elapsed=42:58:38 avg_step=3.75s recent_step=3.80s eta=09:06:13 +2026-09-16 08:45:27 | INFO | train rank=0 | step=41270/50000 progress=82.5% loss=0.111314 lr=8.105538e-06 lr_backbone=8.105538e-06 elapsed=42:59:00 avg_step=3.75s recent_step=2.21s eta=09:05:32 +2026-09-16 08:45:54 | INFO | train rank=0 | step=41280/50000 progress=82.6% loss=0.085843 lr=8.087497e-06 lr_backbone=8.087497e-06 elapsed=42:59:27 avg_step=3.75s recent_step=2.70s eta=09:04:53 +2026-09-16 08:46:23 | INFO | train rank=0 | step=41290/50000 progress=82.6% loss=0.059190 lr=8.069474e-06 lr_backbone=8.069474e-06 elapsed=42:59:56 avg_step=3.75s recent_step=2.92s eta=09:04:13 +2026-09-16 08:46:56 | INFO | train rank=0 | step=41300/50000 progress=82.6% loss=0.069145 lr=8.051469e-06 lr_backbone=8.051469e-06 elapsed=43:00:29 avg_step=3.75s recent_step=3.28s eta=09:03:35 +2026-09-16 08:47:20 | INFO | train rank=0 | step=41310/50000 progress=82.6% loss=0.097827 lr=8.033482e-06 lr_backbone=8.033482e-06 elapsed=43:00:53 avg_step=3.75s recent_step=2.44s eta=09:02:55 +2026-09-16 08:47:49 | INFO | train rank=0 | step=41320/50000 progress=82.6% loss=0.098082 lr=8.015514e-06 lr_backbone=8.015514e-06 elapsed=43:01:21 avg_step=3.75s recent_step=2.82s eta=09:02:15 +2026-09-16 08:48:11 | INFO | train rank=0 | step=41330/50000 progress=82.7% loss=0.086867 lr=7.997565e-06 lr_backbone=7.997565e-06 elapsed=43:01:44 avg_step=3.75s recent_step=2.27s eta=09:01:35 +2026-09-16 08:48:42 | INFO | train rank=0 | step=41340/50000 progress=82.7% loss=0.094810 lr=7.979633e-06 lr_backbone=7.979633e-06 elapsed=43:02:15 avg_step=3.75s recent_step=3.11s eta=09:00:56 +2026-09-16 08:49:10 | INFO | train rank=0 | step=41350/50000 progress=82.7% loss=0.117380 lr=7.961720e-06 lr_backbone=7.961720e-06 elapsed=43:02:42 avg_step=3.75s recent_step=2.73s eta=09:00:16 +2026-09-16 08:49:34 | INFO | train rank=0 | step=41360/50000 progress=82.7% loss=0.115617 lr=7.943826e-06 lr_backbone=7.943826e-06 elapsed=43:03:06 avg_step=3.75s recent_step=2.40s eta=08:59:36 +2026-09-16 08:50:08 | INFO | train rank=0 | step=41370/50000 progress=82.7% loss=0.112441 lr=7.925950e-06 lr_backbone=7.925950e-06 elapsed=43:03:40 avg_step=3.75s recent_step=3.40s eta=08:58:58 +2026-09-16 08:50:45 | INFO | train rank=0 | step=41380/50000 progress=82.8% loss=0.098424 lr=7.908092e-06 lr_backbone=7.908092e-06 elapsed=43:04:18 avg_step=3.75s recent_step=3.71s eta=08:58:20 +2026-09-16 08:51:08 | INFO | train rank=0 | step=41390/50000 progress=82.8% loss=0.090991 lr=7.890253e-06 lr_backbone=7.890253e-06 elapsed=43:04:41 avg_step=3.75s recent_step=2.32s eta=08:57:40 +2026-09-16 08:51:35 | INFO | train rank=0 | step=41400/50000 progress=82.8% loss=0.101693 lr=7.872432e-06 lr_backbone=7.872432e-06 elapsed=43:05:08 avg_step=3.75s recent_step=2.75s eta=08:57:00 +2026-09-16 08:52:08 | INFO | train rank=0 | step=41410/50000 progress=82.8% loss=0.103636 lr=7.854629e-06 lr_backbone=7.854629e-06 elapsed=43:05:41 avg_step=3.75s recent_step=3.23s eta=08:56:22 +2026-09-16 08:52:37 | INFO | train rank=0 | step=41420/50000 progress=82.8% loss=0.099782 lr=7.836845e-06 lr_backbone=7.836845e-06 elapsed=43:06:10 avg_step=3.75s recent_step=2.95s eta=08:55:42 +2026-09-16 08:53:06 | INFO | train rank=0 | step=41430/50000 progress=82.9% loss=0.089899 lr=7.819080e-06 lr_backbone=7.819080e-06 elapsed=43:06:39 avg_step=3.75s recent_step=2.86s eta=08:55:03 +2026-09-16 08:53:26 | INFO | train rank=0 | step=41440/50000 progress=82.9% loss=0.085761 lr=7.801332e-06 lr_backbone=7.801332e-06 elapsed=43:06:59 avg_step=3.75s recent_step=2.01s eta=08:54:22 +2026-09-16 08:54:01 | INFO | train rank=0 | step=41450/50000 progress=82.9% loss=0.087256 lr=7.783604e-06 lr_backbone=7.783604e-06 elapsed=43:07:34 avg_step=3.75s recent_step=3.50s eta=08:53:44 +2026-09-16 08:54:31 | INFO | train rank=0 | step=41460/50000 progress=82.9% loss=0.115723 lr=7.765893e-06 lr_backbone=7.765893e-06 elapsed=43:08:03 avg_step=3.75s recent_step=2.97s eta=08:53:05 +2026-09-16 08:54:54 | INFO | train rank=0 | step=41470/50000 progress=82.9% loss=0.089792 lr=7.748202e-06 lr_backbone=7.748202e-06 elapsed=43:08:27 avg_step=3.75s recent_step=2.32s eta=08:52:25 +2026-09-16 08:55:23 | INFO | train rank=0 | step=41480/50000 progress=83.0% loss=0.086304 lr=7.730528e-06 lr_backbone=7.730528e-06 elapsed=43:08:56 avg_step=3.74s recent_step=2.93s eta=08:51:46 +2026-09-16 08:55:56 | INFO | train rank=0 | step=41490/50000 progress=83.0% loss=0.088339 lr=7.712874e-06 lr_backbone=7.712874e-06 elapsed=43:09:29 avg_step=3.74s recent_step=3.28s eta=08:51:07 +2026-09-16 08:56:17 | INFO | train rank=0 | step=41500/50000 progress=83.0% loss=0.079289 lr=7.695237e-06 lr_backbone=7.695237e-06 elapsed=43:09:50 avg_step=3.74s recent_step=2.10s eta=08:50:26 +2026-09-16 08:56:46 | INFO | train rank=0 | step=41510/50000 progress=83.0% loss=0.097911 lr=7.677620e-06 lr_backbone=7.677620e-06 elapsed=43:10:19 avg_step=3.74s recent_step=2.94s eta=08:49:47 +2026-09-16 08:57:13 | INFO | train rank=0 | step=41520/50000 progress=83.0% loss=0.070717 lr=7.660020e-06 lr_backbone=7.660020e-06 elapsed=43:10:46 avg_step=3.74s recent_step=2.69s eta=08:49:08 +2026-09-16 08:57:46 | INFO | train rank=0 | step=41530/50000 progress=83.1% loss=0.089793 lr=7.642440e-06 lr_backbone=7.642440e-06 elapsed=43:11:19 avg_step=3.74s recent_step=3.29s eta=08:48:29 +2026-09-16 08:58:14 | INFO | train rank=0 | step=41540/50000 progress=83.1% loss=0.123441 lr=7.624877e-06 lr_backbone=7.624877e-06 elapsed=43:11:46 avg_step=3.74s recent_step=2.74s eta=08:47:50 +2026-09-16 08:58:42 | INFO | train rank=0 | step=41550/50000 progress=83.1% loss=0.097724 lr=7.607334e-06 lr_backbone=7.607334e-06 elapsed=43:12:14 avg_step=3.74s recent_step=2.80s eta=08:47:11 +2026-09-16 08:59:05 | INFO | train rank=0 | step=41560/50000 progress=83.1% loss=0.063185 lr=7.589809e-06 lr_backbone=7.589809e-06 elapsed=43:12:37 avg_step=3.74s recent_step=2.31s eta=08:46:30 +2026-09-16 08:59:32 | INFO | train rank=0 | step=41570/50000 progress=83.1% loss=0.105937 lr=7.572302e-06 lr_backbone=7.572302e-06 elapsed=43:13:04 avg_step=3.74s recent_step=2.69s eta=08:45:51 +2026-09-16 09:00:01 | INFO | train rank=0 | step=41580/50000 progress=83.2% loss=0.091006 lr=7.554814e-06 lr_backbone=7.554814e-06 elapsed=43:13:34 avg_step=3.74s recent_step=2.99s eta=08:45:12 +2026-09-16 09:00:27 | INFO | train rank=0 | step=41590/50000 progress=83.2% loss=0.101048 lr=7.537344e-06 lr_backbone=7.537344e-06 elapsed=43:14:00 avg_step=3.74s recent_step=2.55s eta=08:44:32 +2026-09-16 09:00:50 | INFO | train rank=0 | step=41600/50000 progress=83.2% loss=0.078386 lr=7.519894e-06 lr_backbone=7.519894e-06 elapsed=43:14:23 avg_step=3.74s recent_step=2.30s eta=08:43:52 +2026-09-16 09:01:24 | INFO | train rank=0 | step=41610/50000 progress=83.2% loss=0.072436 lr=7.502461e-06 lr_backbone=7.502461e-06 elapsed=43:14:57 avg_step=3.74s recent_step=3.45s eta=08:43:13 +2026-09-16 09:01:48 | INFO | train rank=0 | step=41620/50000 progress=83.2% loss=0.093118 lr=7.485048e-06 lr_backbone=7.485048e-06 elapsed=43:15:20 avg_step=3.74s recent_step=2.31s eta=08:42:33 +2026-09-16 09:02:15 | INFO | train rank=0 | step=41630/50000 progress=83.3% loss=0.075522 lr=7.467652e-06 lr_backbone=7.467652e-06 elapsed=43:15:48 avg_step=3.74s recent_step=2.75s eta=08:41:54 +2026-09-16 09:02:45 | INFO | train rank=0 | step=41640/50000 progress=83.3% loss=0.078893 lr=7.450276e-06 lr_backbone=7.450276e-06 elapsed=43:16:18 avg_step=3.74s recent_step=3.03s eta=08:41:15 +2026-09-16 09:03:09 | INFO | train rank=0 | step=41650/50000 progress=83.3% loss=0.090909 lr=7.432918e-06 lr_backbone=7.432918e-06 elapsed=43:16:42 avg_step=3.74s recent_step=2.41s eta=08:40:35 +2026-09-16 09:03:37 | INFO | train rank=0 | step=41660/50000 progress=83.3% loss=0.070811 lr=7.415579e-06 lr_backbone=7.415579e-06 elapsed=43:17:09 avg_step=3.74s recent_step=2.73s eta=08:39:55 +2026-09-16 09:04:04 | INFO | train rank=0 | step=41670/50000 progress=83.3% loss=0.126898 lr=7.398258e-06 lr_backbone=7.398258e-06 elapsed=43:17:37 avg_step=3.74s recent_step=2.72s eta=08:39:16 +2026-09-16 09:04:31 | INFO | train rank=0 | step=41680/50000 progress=83.4% loss=0.080037 lr=7.380956e-06 lr_backbone=7.380956e-06 elapsed=43:18:03 avg_step=3.74s recent_step=2.67s eta=08:38:36 +2026-09-16 09:05:00 | INFO | train rank=0 | step=41690/50000 progress=83.4% loss=0.097610 lr=7.363673e-06 lr_backbone=7.363673e-06 elapsed=43:18:33 avg_step=3.74s recent_step=2.95s eta=08:37:57 +2026-09-16 09:05:25 | INFO | train rank=0 | step=41700/50000 progress=83.4% loss=0.094585 lr=7.346408e-06 lr_backbone=7.346408e-06 elapsed=43:18:58 avg_step=3.74s recent_step=2.54s eta=08:37:18 +2026-09-16 09:05:51 | INFO | train rank=0 | step=41710/50000 progress=83.4% loss=0.085830 lr=7.329162e-06 lr_backbone=7.329162e-06 elapsed=43:19:23 avg_step=3.74s recent_step=2.51s eta=08:36:38 +2026-09-16 09:06:17 | INFO | train rank=0 | step=41720/50000 progress=83.4% loss=0.067783 lr=7.311934e-06 lr_backbone=7.311934e-06 elapsed=43:19:49 avg_step=3.74s recent_step=2.61s eta=08:35:58 +2026-09-16 09:06:49 | INFO | train rank=0 | step=41730/50000 progress=83.5% loss=0.187192 lr=7.294726e-06 lr_backbone=7.294726e-06 elapsed=43:20:22 avg_step=3.74s recent_step=3.22s eta=08:35:20 +2026-09-16 09:07:16 | INFO | train rank=0 | step=41740/50000 progress=83.5% loss=0.089417 lr=7.277536e-06 lr_backbone=7.277536e-06 elapsed=43:20:49 avg_step=3.74s recent_step=2.73s eta=08:34:40 +2026-09-16 09:07:39 | INFO | train rank=0 | step=41750/50000 progress=83.5% loss=0.114099 lr=7.260364e-06 lr_backbone=7.260364e-06 elapsed=43:21:12 avg_step=3.74s recent_step=2.30s eta=08:34:00 +2026-09-16 09:08:05 | INFO | train rank=0 | step=41760/50000 progress=83.5% loss=0.078607 lr=7.243212e-06 lr_backbone=7.243212e-06 elapsed=43:21:37 avg_step=3.74s recent_step=2.54s eta=08:33:20 +2026-09-16 09:08:41 | INFO | train rank=0 | step=41770/50000 progress=83.5% loss=0.118862 lr=7.226078e-06 lr_backbone=7.226078e-06 elapsed=43:22:14 avg_step=3.74s recent_step=3.61s eta=08:32:43 +2026-09-16 09:09:02 | INFO | train rank=0 | step=41780/50000 progress=83.6% loss=0.093944 lr=7.208963e-06 lr_backbone=7.208963e-06 elapsed=43:22:35 avg_step=3.74s recent_step=2.15s eta=08:32:02 +2026-09-16 09:09:30 | INFO | train rank=0 | step=41790/50000 progress=83.6% loss=0.079434 lr=7.191866e-06 lr_backbone=7.191866e-06 elapsed=43:23:02 avg_step=3.74s recent_step=2.74s eta=08:31:23 +2026-09-16 09:09:56 | INFO | train rank=0 | step=41800/50000 progress=83.6% loss=0.093263 lr=7.174788e-06 lr_backbone=7.174788e-06 elapsed=43:23:28 avg_step=3.74s recent_step=2.60s eta=08:30:43 +2026-09-16 09:10:22 | INFO | train rank=0 | step=41810/50000 progress=83.6% loss=0.111625 lr=7.157729e-06 lr_backbone=7.157729e-06 elapsed=43:23:55 avg_step=3.74s recent_step=2.65s eta=08:30:04 +2026-09-16 09:10:47 | INFO | train rank=0 | step=41820/50000 progress=83.6% loss=0.111278 lr=7.140689e-06 lr_backbone=7.140689e-06 elapsed=43:24:19 avg_step=3.74s recent_step=2.45s eta=08:29:24 +2026-09-16 09:11:16 | INFO | train rank=0 | step=41830/50000 progress=83.7% loss=0.078956 lr=7.123667e-06 lr_backbone=7.123667e-06 elapsed=43:24:49 avg_step=3.74s recent_step=2.97s eta=08:28:45 +2026-09-16 09:11:43 | INFO | train rank=0 | step=41840/50000 progress=83.7% loss=0.085182 lr=7.106664e-06 lr_backbone=7.106664e-06 elapsed=43:25:16 avg_step=3.74s recent_step=2.68s eta=08:28:06 +2026-09-16 09:12:11 | INFO | train rank=0 | step=41850/50000 progress=83.7% loss=0.079194 lr=7.089680e-06 lr_backbone=7.089680e-06 elapsed=43:25:44 avg_step=3.74s recent_step=2.84s eta=08:27:27 +2026-09-16 09:12:36 | INFO | train rank=0 | step=41860/50000 progress=83.7% loss=0.080356 lr=7.072715e-06 lr_backbone=7.072715e-06 elapsed=43:26:08 avg_step=3.74s recent_step=2.43s eta=08:26:47 +2026-09-16 09:13:03 | INFO | train rank=0 | step=41870/50000 progress=83.7% loss=0.072403 lr=7.055768e-06 lr_backbone=7.055768e-06 elapsed=43:26:36 avg_step=3.74s recent_step=2.75s eta=08:26:07 +2026-09-16 09:13:29 | INFO | train rank=0 | step=41880/50000 progress=83.8% loss=0.116313 lr=7.038841e-06 lr_backbone=7.038841e-06 elapsed=43:27:02 avg_step=3.74s recent_step=2.60s eta=08:25:28 +2026-09-16 09:13:57 | INFO | train rank=0 | step=41890/50000 progress=83.8% loss=0.099053 lr=7.021932e-06 lr_backbone=7.021932e-06 elapsed=43:27:30 avg_step=3.73s recent_step=2.76s eta=08:24:49 +2026-09-16 09:14:24 | INFO | train rank=0 | step=41900/50000 progress=83.8% loss=0.114276 lr=7.005042e-06 lr_backbone=7.005042e-06 elapsed=43:27:57 avg_step=3.73s recent_step=2.71s eta=08:24:09 +2026-09-16 09:14:53 | INFO | train rank=0 | step=41910/50000 progress=83.8% loss=0.072916 lr=6.988170e-06 lr_backbone=6.988170e-06 elapsed=43:28:26 avg_step=3.73s recent_step=2.88s eta=08:23:30 +2026-09-16 09:15:17 | INFO | train rank=0 | step=41920/50000 progress=83.8% loss=0.074861 lr=6.971318e-06 lr_backbone=6.971318e-06 elapsed=43:28:49 avg_step=3.73s recent_step=2.40s eta=08:22:50 +2026-09-16 09:15:43 | INFO | train rank=0 | step=41930/50000 progress=83.9% loss=0.096470 lr=6.954484e-06 lr_backbone=6.954484e-06 elapsed=43:29:15 avg_step=3.73s recent_step=2.59s eta=08:22:11 +2026-09-16 09:16:10 | INFO | train rank=0 | step=41940/50000 progress=83.9% loss=0.070762 lr=6.937669e-06 lr_backbone=6.937669e-06 elapsed=43:29:42 avg_step=3.73s recent_step=2.69s eta=08:21:31 +2026-09-16 09:16:31 | INFO | train rank=0 | step=41950/50000 progress=83.9% loss=0.078117 lr=6.920873e-06 lr_backbone=6.920873e-06 elapsed=43:30:04 avg_step=3.73s recent_step=2.18s eta=08:20:51 +2026-09-16 09:16:55 | INFO | train rank=0 | step=41960/50000 progress=83.9% loss=0.105593 lr=6.904096e-06 lr_backbone=6.904096e-06 elapsed=43:30:28 avg_step=3.73s recent_step=2.42s eta=08:20:11 +2026-09-16 09:17:25 | INFO | train rank=0 | step=41970/50000 progress=83.9% loss=0.089149 lr=6.887338e-06 lr_backbone=6.887338e-06 elapsed=43:30:58 avg_step=3.73s recent_step=2.94s eta=08:19:32 +2026-09-16 09:17:50 | INFO | train rank=0 | step=41980/50000 progress=84.0% loss=0.087734 lr=6.870598e-06 lr_backbone=6.870598e-06 elapsed=43:31:23 avg_step=3.73s recent_step=2.54s eta=08:18:53 +2026-09-16 09:18:15 | INFO | train rank=0 | step=41990/50000 progress=84.0% loss=0.053332 lr=6.853878e-06 lr_backbone=6.853878e-06 elapsed=43:31:47 avg_step=3.73s recent_step=2.44s eta=08:18:13 +2026-09-16 09:18:47 | INFO | train rank=0 | step=42000/50000 progress=84.0% loss=0.096702 lr=6.837176e-06 lr_backbone=6.837176e-06 elapsed=43:32:19 avg_step=3.73s recent_step=3.18s eta=08:17:35 +2026-09-16 09:19:35 | INFO | train rank=0 | step=42010/50000 progress=84.0% loss=0.089176 lr=6.820493e-06 lr_backbone=6.820493e-06 elapsed=43:33:07 avg_step=3.73s recent_step=4.82s eta=08:16:59 +2026-09-16 09:20:08 | INFO | train rank=0 | step=42020/50000 progress=84.0% loss=0.082199 lr=6.803829e-06 lr_backbone=6.803829e-06 elapsed=43:33:41 avg_step=3.73s recent_step=3.34s eta=08:16:21 +2026-09-16 09:20:44 | INFO | train rank=0 | step=42030/50000 progress=84.1% loss=0.094678 lr=6.787184e-06 lr_backbone=6.787184e-06 elapsed=43:34:16 avg_step=3.73s recent_step=3.56s eta=08:15:44 +2026-09-16 09:21:17 | INFO | train rank=0 | step=42040/50000 progress=84.1% loss=0.089990 lr=6.770558e-06 lr_backbone=6.770558e-06 elapsed=43:34:50 avg_step=3.73s recent_step=3.31s eta=08:15:06 +2026-09-16 09:22:14 | INFO | train rank=0 | step=42050/50000 progress=84.1% loss=0.068393 lr=6.753951e-06 lr_backbone=6.753951e-06 elapsed=43:35:47 avg_step=3.73s recent_step=5.69s eta=08:14:32 +2026-09-16 09:22:44 | INFO | train rank=0 | step=42060/50000 progress=84.1% loss=0.097017 lr=6.737362e-06 lr_backbone=6.737362e-06 elapsed=43:36:16 avg_step=3.73s recent_step=2.99s eta=08:13:53 +2026-09-16 09:23:14 | INFO | train rank=0 | step=42070/50000 progress=84.1% loss=0.086897 lr=6.720793e-06 lr_backbone=6.720793e-06 elapsed=43:36:47 avg_step=3.73s recent_step=3.03s eta=08:13:15 +2026-09-16 09:23:58 | INFO | train rank=0 | step=42080/50000 progress=84.2% loss=0.092079 lr=6.704242e-06 lr_backbone=6.704242e-06 elapsed=43:37:31 avg_step=3.73s recent_step=4.41s eta=08:12:39 +2026-09-16 09:24:30 | INFO | train rank=0 | step=42090/50000 progress=84.2% loss=0.073277 lr=6.687711e-06 lr_backbone=6.687711e-06 elapsed=43:38:03 avg_step=3.73s recent_step=3.17s eta=08:12:00 +2026-09-16 09:25:20 | INFO | train rank=0 | step=42100/50000 progress=84.2% loss=0.117715 lr=6.671198e-06 lr_backbone=6.671198e-06 elapsed=43:38:53 avg_step=3.73s recent_step=5.04s eta=08:11:25 +2026-09-16 09:25:56 | INFO | train rank=0 | step=42110/50000 progress=84.2% loss=0.083919 lr=6.654705e-06 lr_backbone=6.654705e-06 elapsed=43:39:29 avg_step=3.73s recent_step=3.56s eta=08:10:48 +2026-09-16 09:26:25 | INFO | train rank=0 | step=42120/50000 progress=84.2% loss=0.080958 lr=6.638230e-06 lr_backbone=6.638230e-06 elapsed=43:39:57 avg_step=3.73s recent_step=2.87s eta=08:10:09 +2026-09-16 09:26:57 | INFO | train rank=0 | step=42130/50000 progress=84.3% loss=0.068390 lr=6.621774e-06 lr_backbone=6.621774e-06 elapsed=43:40:30 avg_step=3.73s recent_step=3.25s eta=08:09:31 +2026-09-16 09:27:55 | INFO | train rank=0 | step=42140/50000 progress=84.3% loss=0.076997 lr=6.605338e-06 lr_backbone=6.605338e-06 elapsed=43:41:28 avg_step=3.73s recent_step=5.77s eta=08:08:57 +2026-09-16 09:28:24 | INFO | train rank=0 | step=42150/50000 progress=84.3% loss=0.093641 lr=6.588920e-06 lr_backbone=6.588920e-06 elapsed=43:41:57 avg_step=3.73s recent_step=2.94s eta=08:08:18 +2026-09-16 09:28:55 | INFO | train rank=0 | step=42160/50000 progress=84.3% loss=0.080894 lr=6.572521e-06 lr_backbone=6.572521e-06 elapsed=43:42:28 avg_step=3.73s recent_step=3.11s eta=08:07:40 +2026-09-16 09:29:25 | INFO | train rank=0 | step=42170/50000 progress=84.3% loss=0.095112 lr=6.556141e-06 lr_backbone=6.556141e-06 elapsed=43:42:57 avg_step=3.73s recent_step=2.93s eta=08:07:01 +2026-09-16 09:30:12 | INFO | train rank=0 | step=42180/50000 progress=84.4% loss=0.073116 lr=6.539780e-06 lr_backbone=6.539780e-06 elapsed=43:43:45 avg_step=3.73s recent_step=4.77s eta=08:06:26 +2026-09-16 09:30:49 | INFO | train rank=0 | step=42190/50000 progress=84.4% loss=0.079933 lr=6.523439e-06 lr_backbone=6.523439e-06 elapsed=43:44:21 avg_step=3.73s recent_step=3.65s eta=08:05:48 +2026-09-16 09:31:34 | INFO | train rank=0 | step=42200/50000 progress=84.4% loss=0.064602 lr=6.507116e-06 lr_backbone=6.507116e-06 elapsed=43:45:06 avg_step=3.73s recent_step=4.49s eta=08:05:12 +2026-09-16 09:32:07 | INFO | train rank=0 | step=42210/50000 progress=84.4% loss=0.078307 lr=6.490812e-06 lr_backbone=6.490812e-06 elapsed=43:45:39 avg_step=3.73s recent_step=3.31s eta=08:04:34 +2026-09-16 09:32:59 | INFO | train rank=0 | step=42220/50000 progress=84.4% loss=0.097379 lr=6.474528e-06 lr_backbone=6.474528e-06 elapsed=43:46:32 avg_step=3.73s recent_step=5.28s eta=08:04:00 +2026-09-16 09:33:32 | INFO | train rank=0 | step=42230/50000 progress=84.5% loss=0.079155 lr=6.458262e-06 lr_backbone=6.458262e-06 elapsed=43:47:04 avg_step=3.73s recent_step=3.21s eta=08:03:21 +2026-09-16 09:34:10 | INFO | train rank=0 | step=42240/50000 progress=84.5% loss=0.083259 lr=6.442015e-06 lr_backbone=6.442015e-06 elapsed=43:47:43 avg_step=3.73s recent_step=3.88s eta=08:02:44 +2026-09-16 09:34:42 | INFO | train rank=0 | step=42250/50000 progress=84.5% loss=0.108116 lr=6.425788e-06 lr_backbone=6.425788e-06 elapsed=43:48:14 avg_step=3.73s recent_step=3.13s eta=08:02:06 +2026-09-16 09:35:33 | INFO | train rank=0 | step=42260/50000 progress=84.5% loss=0.083849 lr=6.409579e-06 lr_backbone=6.409579e-06 elapsed=43:49:06 avg_step=3.73s recent_step=5.16s eta=08:01:31 +2026-09-16 09:36:07 | INFO | train rank=0 | step=42270/50000 progress=84.5% loss=0.087067 lr=6.393390e-06 lr_backbone=6.393390e-06 elapsed=43:49:40 avg_step=3.73s recent_step=3.40s eta=08:00:53 +2026-09-16 09:36:34 | INFO | train rank=0 | step=42280/50000 progress=84.6% loss=0.066247 lr=6.377220e-06 lr_backbone=6.377220e-06 elapsed=43:50:07 avg_step=3.73s recent_step=2.68s eta=08:00:14 +2026-09-16 09:37:14 | INFO | train rank=0 | step=42290/50000 progress=84.6% loss=0.103608 lr=6.361068e-06 lr_backbone=6.361068e-06 elapsed=43:50:47 avg_step=3.73s recent_step=4.00s eta=07:59:37 +2026-09-16 09:38:11 | INFO | train rank=0 | step=42300/50000 progress=84.6% loss=0.081164 lr=6.344936e-06 lr_backbone=6.344936e-06 elapsed=43:51:44 avg_step=3.73s recent_step=5.71s eta=07:59:03 +2026-09-16 09:38:39 | INFO | train rank=0 | step=42310/50000 progress=84.6% loss=0.093400 lr=6.328823e-06 lr_backbone=6.328823e-06 elapsed=43:52:12 avg_step=3.73s recent_step=2.82s eta=07:58:24 +2026-09-16 09:39:11 | INFO | train rank=0 | step=42320/50000 progress=84.6% loss=0.090208 lr=6.312729e-06 lr_backbone=6.312729e-06 elapsed=43:52:43 avg_step=3.73s recent_step=3.13s eta=07:57:46 +2026-09-16 09:39:46 | INFO | train rank=0 | step=42330/50000 progress=84.7% loss=0.088602 lr=6.296654e-06 lr_backbone=6.296654e-06 elapsed=43:53:19 avg_step=3.73s recent_step=3.57s eta=07:57:08 +2026-09-16 09:40:32 | INFO | train rank=0 | step=42340/50000 progress=84.7% loss=0.084911 lr=6.280599e-06 lr_backbone=6.280599e-06 elapsed=43:54:04 avg_step=3.73s recent_step=4.53s eta=07:56:32 +2026-09-16 09:41:05 | INFO | train rank=0 | step=42350/50000 progress=84.7% loss=0.062536 lr=6.264562e-06 lr_backbone=6.264562e-06 elapsed=43:54:38 avg_step=3.73s recent_step=3.37s eta=07:55:54 +2026-09-16 09:41:41 | INFO | train rank=0 | step=42360/50000 progress=84.7% loss=0.110097 lr=6.248544e-06 lr_backbone=6.248544e-06 elapsed=43:55:14 avg_step=3.73s recent_step=3.56s eta=07:55:17 +2026-09-16 09:42:20 | INFO | train rank=0 | step=42370/50000 progress=84.7% loss=0.092628 lr=6.232546e-06 lr_backbone=6.232546e-06 elapsed=43:55:53 avg_step=3.73s recent_step=3.92s eta=07:54:40 +2026-09-16 09:43:09 | INFO | train rank=0 | step=42380/50000 progress=84.8% loss=0.088006 lr=6.216567e-06 lr_backbone=6.216567e-06 elapsed=43:56:42 avg_step=3.73s recent_step=4.86s eta=07:54:05 +2026-09-16 09:43:40 | INFO | train rank=0 | step=42390/50000 progress=84.8% loss=0.116683 lr=6.200607e-06 lr_backbone=6.200607e-06 elapsed=43:57:13 avg_step=3.73s recent_step=3.13s eta=07:53:26 +2026-09-16 09:44:07 | INFO | train rank=0 | step=42400/50000 progress=84.8% loss=0.082687 lr=6.184666e-06 lr_backbone=6.184666e-06 elapsed=43:57:40 avg_step=3.73s recent_step=2.73s eta=07:52:47 +2026-09-16 09:44:37 | INFO | train rank=0 | step=42410/50000 progress=84.8% loss=0.118391 lr=6.168744e-06 lr_backbone=6.168744e-06 elapsed=43:58:10 avg_step=3.73s recent_step=3.01s eta=07:52:08 +2026-09-16 09:45:33 | INFO | train rank=0 | step=42420/50000 progress=84.8% loss=0.087501 lr=6.152842e-06 lr_backbone=6.152842e-06 elapsed=43:59:06 avg_step=3.73s recent_step=5.57s eta=07:51:34 +2026-09-16 09:46:05 | INFO | train rank=0 | step=42430/50000 progress=84.9% loss=0.099405 lr=6.136958e-06 lr_backbone=6.136958e-06 elapsed=43:59:38 avg_step=3.73s recent_step=3.17s eta=07:50:56 +2026-09-16 09:46:42 | INFO | train rank=0 | step=42440/50000 progress=84.9% loss=0.091516 lr=6.121094e-06 lr_backbone=6.121094e-06 elapsed=44:00:14 avg_step=3.73s recent_step=3.67s eta=07:50:19 +2026-09-16 09:47:17 | INFO | train rank=0 | step=42450/50000 progress=84.9% loss=0.089756 lr=6.105249e-06 lr_backbone=6.105249e-06 elapsed=44:00:50 avg_step=3.73s recent_step=3.57s eta=07:49:41 +2026-09-16 09:48:05 | INFO | train rank=0 | step=42460/50000 progress=84.9% loss=0.099899 lr=6.089423e-06 lr_backbone=6.089423e-06 elapsed=44:01:38 avg_step=3.73s recent_step=4.74s eta=07:49:05 +2026-09-16 09:48:44 | INFO | train rank=0 | step=42470/50000 progress=84.9% loss=0.086607 lr=6.073617e-06 lr_backbone=6.073617e-06 elapsed=44:02:17 avg_step=3.73s recent_step=3.96s eta=07:48:28 +2026-09-16 09:49:19 | INFO | train rank=0 | step=42480/50000 progress=85.0% loss=0.088026 lr=6.057829e-06 lr_backbone=6.057829e-06 elapsed=44:02:52 avg_step=3.73s recent_step=3.50s eta=07:47:51 +2026-09-16 09:49:49 | INFO | train rank=0 | step=42490/50000 progress=85.0% loss=0.115766 lr=6.042061e-06 lr_backbone=6.042061e-06 elapsed=44:03:22 avg_step=3.73s recent_step=2.94s eta=07:47:12 +2026-09-16 09:50:24 | INFO | train rank=0 | step=42500/50000 progress=85.0% loss=0.070029 lr=6.026312e-06 lr_backbone=6.026312e-06 elapsed=44:03:57 avg_step=3.73s recent_step=3.57s eta=07:46:34 +2026-09-16 09:51:05 | INFO | train rank=0 | step=42510/50000 progress=85.0% loss=0.086707 lr=6.010583e-06 lr_backbone=6.010583e-06 elapsed=44:04:38 avg_step=3.73s recent_step=4.03s eta=07:45:58 +2026-09-16 09:51:41 | INFO | train rank=0 | step=42520/50000 progress=85.0% loss=0.099198 lr=5.994872e-06 lr_backbone=5.994872e-06 elapsed=44:05:14 avg_step=3.73s recent_step=3.62s eta=07:45:20 +2026-09-16 09:52:07 | INFO | train rank=0 | step=42530/50000 progress=85.1% loss=0.131320 lr=5.979181e-06 lr_backbone=5.979181e-06 elapsed=44:05:40 avg_step=3.73s recent_step=2.64s eta=07:44:41 +2026-09-16 09:53:00 | INFO | train rank=0 | step=42540/50000 progress=85.1% loss=0.111592 lr=5.963509e-06 lr_backbone=5.963509e-06 elapsed=44:06:33 avg_step=3.73s recent_step=5.25s eta=07:44:06 +2026-09-16 09:53:32 | INFO | train rank=0 | step=42550/50000 progress=85.1% loss=0.083551 lr=5.947857e-06 lr_backbone=5.947857e-06 elapsed=44:07:05 avg_step=3.73s recent_step=3.19s eta=07:43:28 +2026-09-16 09:54:01 | INFO | train rank=0 | step=42560/50000 progress=85.1% loss=0.100877 lr=5.932223e-06 lr_backbone=5.932223e-06 elapsed=44:07:34 avg_step=3.73s recent_step=2.95s eta=07:42:49 +2026-09-16 09:54:32 | INFO | train rank=0 | step=42570/50000 progress=85.1% loss=0.094448 lr=5.916609e-06 lr_backbone=5.916609e-06 elapsed=44:08:04 avg_step=3.73s recent_step=3.04s eta=07:42:11 +2026-09-16 09:55:28 | INFO | train rank=0 | step=42580/50000 progress=85.2% loss=0.067129 lr=5.901014e-06 lr_backbone=5.901014e-06 elapsed=44:09:00 avg_step=3.73s recent_step=5.60s eta=07:41:37 +2026-09-16 09:55:59 | INFO | train rank=0 | step=42590/50000 progress=85.2% loss=0.077412 lr=5.885439e-06 lr_backbone=5.885439e-06 elapsed=44:09:32 avg_step=3.73s recent_step=3.16s eta=07:40:58 +2026-09-16 09:56:38 | INFO | train rank=0 | step=42600/50000 progress=85.2% loss=0.089348 lr=5.869882e-06 lr_backbone=5.869882e-06 elapsed=44:10:11 avg_step=3.73s recent_step=3.88s eta=07:40:21 +2026-09-16 09:57:11 | INFO | train rank=0 | step=42610/50000 progress=85.2% loss=0.110293 lr=5.854345e-06 lr_backbone=5.854345e-06 elapsed=44:10:44 avg_step=3.73s recent_step=3.32s eta=07:39:43 +2026-09-16 09:57:50 | INFO | train rank=0 | step=42620/50000 progress=85.2% loss=0.088363 lr=5.838828e-06 lr_backbone=5.838828e-06 elapsed=44:11:23 avg_step=3.73s recent_step=3.88s eta=07:39:06 +2026-09-16 09:58:20 | INFO | train rank=0 | step=42630/50000 progress=85.3% loss=0.071117 lr=5.823330e-06 lr_backbone=5.823330e-06 elapsed=44:11:53 avg_step=3.73s recent_step=3.03s eta=07:38:28 +2026-09-16 09:58:56 | INFO | train rank=0 | step=42640/50000 progress=85.3% loss=0.109814 lr=5.807851e-06 lr_backbone=5.807851e-06 elapsed=44:12:29 avg_step=3.73s recent_step=3.60s eta=07:37:50 +2026-09-16 09:59:31 | INFO | train rank=0 | step=42650/50000 progress=85.3% loss=0.100996 lr=5.792391e-06 lr_backbone=5.792391e-06 elapsed=44:13:04 avg_step=3.73s recent_step=3.45s eta=07:37:12 +2026-09-16 10:00:20 | INFO | train rank=0 | step=42660/50000 progress=85.3% loss=0.081400 lr=5.776951e-06 lr_backbone=5.776951e-06 elapsed=44:13:52 avg_step=3.73s recent_step=4.87s eta=07:36:37 +2026-09-16 10:00:54 | INFO | train rank=0 | step=42670/50000 progress=85.3% loss=0.071908 lr=5.761530e-06 lr_backbone=5.761530e-06 elapsed=44:14:27 avg_step=3.73s recent_step=3.45s eta=07:35:59 +2026-09-16 10:01:29 | INFO | train rank=0 | step=42680/50000 progress=85.4% loss=0.103805 lr=5.746128e-06 lr_backbone=5.746128e-06 elapsed=44:15:02 avg_step=3.73s recent_step=3.54s eta=07:35:21 +2026-09-16 10:02:04 | INFO | train rank=0 | step=42690/50000 progress=85.4% loss=0.101647 lr=5.730746e-06 lr_backbone=5.730746e-06 elapsed=44:15:36 avg_step=3.73s recent_step=3.42s eta=07:34:43 +2026-09-16 10:02:50 | INFO | train rank=0 | step=42700/50000 progress=85.4% loss=0.118018 lr=5.715383e-06 lr_backbone=5.715383e-06 elapsed=44:16:23 avg_step=3.73s recent_step=4.65s eta=07:34:08 +2026-09-16 10:03:26 | INFO | train rank=0 | step=42710/50000 progress=85.4% loss=0.125370 lr=5.700039e-06 lr_backbone=5.700039e-06 elapsed=44:16:59 avg_step=3.73s recent_step=3.57s eta=07:33:30 +2026-09-16 10:03:58 | INFO | train rank=0 | step=42720/50000 progress=85.4% loss=0.154096 lr=5.684715e-06 lr_backbone=5.684715e-06 elapsed=44:17:31 avg_step=3.73s recent_step=3.25s eta=07:32:52 +2026-09-16 10:04:29 | INFO | train rank=0 | step=42730/50000 progress=85.5% loss=0.103710 lr=5.669410e-06 lr_backbone=5.669410e-06 elapsed=44:18:02 avg_step=3.73s recent_step=3.04s eta=07:32:13 +2026-09-16 10:05:13 | INFO | train rank=0 | step=42740/50000 progress=85.5% loss=0.072715 lr=5.654125e-06 lr_backbone=5.654125e-06 elapsed=44:18:45 avg_step=3.73s recent_step=4.39s eta=07:31:37 +2026-09-16 10:05:43 | INFO | train rank=0 | step=42750/50000 progress=85.5% loss=0.074497 lr=5.638859e-06 lr_backbone=5.638859e-06 elapsed=44:19:16 avg_step=3.73s recent_step=3.06s eta=07:30:59 +2026-09-16 10:06:11 | INFO | train rank=0 | step=42760/50000 progress=85.5% loss=0.096066 lr=5.623612e-06 lr_backbone=5.623612e-06 elapsed=44:19:43 avg_step=3.73s recent_step=2.75s eta=07:30:20 +2026-09-16 10:06:42 | INFO | train rank=0 | step=42770/50000 progress=85.5% loss=0.056227 lr=5.608385e-06 lr_backbone=5.608385e-06 elapsed=44:20:15 avg_step=3.73s recent_step=3.14s eta=07:29:41 +2026-09-16 10:07:36 | INFO | train rank=0 | step=42780/50000 progress=85.6% loss=0.098292 lr=5.593178e-06 lr_backbone=5.593178e-06 elapsed=44:21:09 avg_step=3.73s recent_step=5.42s eta=07:29:07 +2026-09-16 10:08:05 | INFO | train rank=0 | step=42790/50000 progress=85.6% loss=0.084524 lr=5.577989e-06 lr_backbone=5.577989e-06 elapsed=44:21:37 avg_step=3.73s recent_step=2.82s eta=07:28:28 +2026-09-16 10:08:38 | INFO | train rank=0 | step=42800/50000 progress=85.6% loss=0.080505 lr=5.562820e-06 lr_backbone=5.562820e-06 elapsed=44:22:11 avg_step=3.73s recent_step=3.32s eta=07:27:50 +2026-09-16 10:09:12 | INFO | train rank=0 | step=42810/50000 progress=85.6% loss=0.061803 lr=5.547671e-06 lr_backbone=5.547671e-06 elapsed=44:22:45 avg_step=3.73s recent_step=3.41s eta=07:27:12 +2026-09-16 10:10:08 | INFO | train rank=0 | step=42820/50000 progress=85.6% loss=0.097876 lr=5.532541e-06 lr_backbone=5.532541e-06 elapsed=44:23:41 avg_step=3.73s recent_step=5.61s eta=07:26:38 +2026-09-16 10:10:40 | INFO | train rank=0 | step=42830/50000 progress=85.7% loss=0.092383 lr=5.517430e-06 lr_backbone=5.517430e-06 elapsed=44:24:13 avg_step=3.73s recent_step=3.23s eta=07:26:00 +2026-09-16 10:11:17 | INFO | train rank=0 | step=42840/50000 progress=85.7% loss=0.074240 lr=5.502339e-06 lr_backbone=5.502339e-06 elapsed=44:24:49 avg_step=3.73s recent_step=3.64s eta=07:25:22 +2026-09-16 10:12:03 | INFO | train rank=0 | step=42850/50000 progress=85.7% loss=0.081417 lr=5.487268e-06 lr_backbone=5.487268e-06 elapsed=44:25:35 avg_step=3.73s recent_step=4.60s eta=07:24:47 +2026-09-16 10:12:38 | INFO | train rank=0 | step=42860/50000 progress=85.7% loss=0.078350 lr=5.472215e-06 lr_backbone=5.472215e-06 elapsed=44:26:10 avg_step=3.73s recent_step=3.50s eta=07:24:09 +2026-09-16 10:13:12 | INFO | train rank=0 | step=42870/50000 progress=85.7% loss=0.065734 lr=5.457183e-06 lr_backbone=5.457183e-06 elapsed=44:26:45 avg_step=3.73s recent_step=3.42s eta=07:23:31 +2026-09-16 10:14:00 | INFO | train rank=0 | step=42880/50000 progress=85.8% loss=0.091861 lr=5.442170e-06 lr_backbone=5.442170e-06 elapsed=44:27:33 avg_step=3.73s recent_step=4.80s eta=07:22:55 +2026-09-16 10:14:25 | INFO | train rank=0 | step=42890/50000 progress=85.8% loss=0.082374 lr=5.427176e-06 lr_backbone=5.427176e-06 elapsed=44:27:58 avg_step=3.73s recent_step=2.50s eta=07:22:16 +2026-09-16 10:14:58 | INFO | train rank=0 | step=42900/50000 progress=85.8% loss=0.065887 lr=5.412202e-06 lr_backbone=5.412202e-06 elapsed=44:28:31 avg_step=3.73s recent_step=3.35s eta=07:21:38 +2026-09-16 10:15:30 | INFO | train rank=0 | step=42910/50000 progress=85.8% loss=0.074801 lr=5.397247e-06 lr_backbone=5.397247e-06 elapsed=44:29:03 avg_step=3.73s recent_step=3.16s eta=07:21:00 +2026-09-16 10:16:18 | INFO | train rank=0 | step=42920/50000 progress=85.8% loss=0.109646 lr=5.382312e-06 lr_backbone=5.382312e-06 elapsed=44:29:51 avg_step=3.73s recent_step=4.80s eta=07:20:24 +2026-09-16 10:16:55 | INFO | train rank=0 | step=42930/50000 progress=85.9% loss=0.108123 lr=5.367396e-06 lr_backbone=5.367396e-06 elapsed=44:30:27 avg_step=3.73s recent_step=3.66s eta=07:19:47 +2026-09-16 10:17:25 | INFO | train rank=0 | step=42940/50000 progress=85.9% loss=0.093896 lr=5.352500e-06 lr_backbone=5.352500e-06 elapsed=44:30:58 avg_step=3.73s recent_step=3.06s eta=07:19:08 +2026-09-16 10:18:01 | INFO | train rank=0 | step=42950/50000 progress=85.9% loss=0.125304 lr=5.337623e-06 lr_backbone=5.337623e-06 elapsed=44:31:34 avg_step=3.73s recent_step=3.61s eta=07:18:31 +2026-09-16 10:18:54 | INFO | train rank=0 | step=42960/50000 progress=85.9% loss=0.093826 lr=5.322766e-06 lr_backbone=5.322766e-06 elapsed=44:32:27 avg_step=3.73s recent_step=5.28s eta=07:17:56 +2026-09-16 10:19:23 | INFO | train rank=0 | step=42970/50000 progress=85.9% loss=0.105274 lr=5.307929e-06 lr_backbone=5.307929e-06 elapsed=44:32:56 avg_step=3.73s recent_step=2.93s eta=07:17:18 +2026-09-16 10:19:55 | INFO | train rank=0 | step=42980/50000 progress=86.0% loss=0.088823 lr=5.293111e-06 lr_backbone=5.293111e-06 elapsed=44:33:28 avg_step=3.73s recent_step=3.17s eta=07:16:39 +2026-09-16 10:20:29 | INFO | train rank=0 | step=42990/50000 progress=86.0% loss=0.082653 lr=5.278312e-06 lr_backbone=5.278312e-06 elapsed=44:34:02 avg_step=3.73s recent_step=3.43s eta=07:16:01 +2026-09-16 10:21:14 | INFO | train rank=0 | step=43000/50000 progress=86.0% loss=0.099797 lr=5.263534e-06 lr_backbone=5.263534e-06 elapsed=44:34:47 avg_step=3.73s recent_step=4.46s eta=07:15:25 +2026-09-16 10:21:53 | INFO | train rank=0 | step=43010/50000 progress=86.0% loss=0.060012 lr=5.248774e-06 lr_backbone=5.248774e-06 elapsed=44:35:26 avg_step=3.73s recent_step=3.89s eta=07:14:48 +2026-09-16 10:22:29 | INFO | train rank=0 | step=43020/50000 progress=86.0% loss=0.121384 lr=5.234034e-06 lr_backbone=5.234034e-06 elapsed=44:36:02 avg_step=3.73s recent_step=3.61s eta=07:14:11 +2026-09-16 10:22:57 | INFO | train rank=0 | step=43030/50000 progress=86.1% loss=0.075290 lr=5.219314e-06 lr_backbone=5.219314e-06 elapsed=44:36:30 avg_step=3.73s recent_step=2.81s eta=07:13:32 +2026-09-16 10:23:48 | INFO | train rank=0 | step=43040/50000 progress=86.1% loss=0.106176 lr=5.204614e-06 lr_backbone=5.204614e-06 elapsed=44:37:20 avg_step=3.73s recent_step=5.07s eta=07:12:57 +2026-09-16 10:24:22 | INFO | train rank=0 | step=43050/50000 progress=86.1% loss=0.076995 lr=5.189933e-06 lr_backbone=5.189933e-06 elapsed=44:37:54 avg_step=3.73s recent_step=3.39s eta=07:12:19 +2026-09-16 10:24:56 | INFO | train rank=0 | step=43060/50000 progress=86.1% loss=0.076692 lr=5.175271e-06 lr_backbone=5.175271e-06 elapsed=44:38:29 avg_step=3.73s recent_step=3.43s eta=07:11:41 +2026-09-16 10:25:32 | INFO | train rank=0 | step=43070/50000 progress=86.1% loss=0.092444 lr=5.160630e-06 lr_backbone=5.160630e-06 elapsed=44:39:05 avg_step=3.73s recent_step=3.59s eta=07:11:04 +2026-09-16 10:26:28 | INFO | train rank=0 | step=43080/50000 progress=86.2% loss=0.106816 lr=5.146008e-06 lr_backbone=5.146008e-06 elapsed=44:40:01 avg_step=3.73s recent_step=5.59s eta=07:10:29 +2026-09-16 10:27:02 | INFO | train rank=0 | step=43090/50000 progress=86.2% loss=0.091986 lr=5.131405e-06 lr_backbone=5.131405e-06 elapsed=44:40:35 avg_step=3.73s recent_step=3.42s eta=07:09:51 +2026-09-16 10:27:35 | INFO | train rank=0 | step=43100/50000 progress=86.2% loss=0.113256 lr=5.116822e-06 lr_backbone=5.116822e-06 elapsed=44:41:07 avg_step=3.73s recent_step=3.27s eta=07:09:13 +2026-09-16 10:28:05 | INFO | train rank=0 | step=43110/50000 progress=86.2% loss=0.086627 lr=5.102259e-06 lr_backbone=5.102259e-06 elapsed=44:41:38 avg_step=3.73s recent_step=3.03s eta=07:08:35 +2026-09-16 10:28:40 | INFO | train rank=0 | step=43120/50000 progress=86.2% loss=0.073754 lr=5.087715e-06 lr_backbone=5.087715e-06 elapsed=44:42:13 avg_step=3.73s recent_step=3.54s eta=07:07:57 +2026-09-16 10:29:20 | INFO | train rank=0 | step=43130/50000 progress=86.3% loss=0.060849 lr=5.073191e-06 lr_backbone=5.073191e-06 elapsed=44:42:53 avg_step=3.73s recent_step=3.99s eta=07:07:20 +2026-09-16 10:30:15 | INFO | train rank=0 | step=43140/50000 progress=86.3% loss=0.114861 lr=5.058687e-06 lr_backbone=5.058687e-06 elapsed=44:43:48 avg_step=3.73s recent_step=5.47s eta=07:06:46 +2026-09-16 10:30:53 | INFO | train rank=0 | step=43150/50000 progress=86.3% loss=0.103787 lr=5.044202e-06 lr_backbone=5.044202e-06 elapsed=44:44:26 avg_step=3.73s recent_step=3.80s eta=07:06:09 +2026-09-16 10:31:25 | INFO | train rank=0 | step=43160/50000 progress=86.3% loss=0.068735 lr=5.029737e-06 lr_backbone=5.029737e-06 elapsed=44:44:58 avg_step=3.73s recent_step=3.25s eta=07:05:30 +2026-09-16 10:32:05 | INFO | train rank=0 | step=43170/50000 progress=86.3% loss=0.084170 lr=5.015292e-06 lr_backbone=5.015292e-06 elapsed=44:45:38 avg_step=3.73s recent_step=3.96s eta=07:04:53 +2026-09-16 10:32:37 | INFO | train rank=0 | step=43180/50000 progress=86.4% loss=0.104183 lr=5.000866e-06 lr_backbone=5.000866e-06 elapsed=44:46:10 avg_step=3.73s recent_step=3.21s eta=07:04:15 +2026-09-16 10:33:08 | INFO | train rank=0 | step=43190/50000 progress=86.4% loss=0.081225 lr=4.986461e-06 lr_backbone=4.986461e-06 elapsed=44:46:41 avg_step=3.73s recent_step=3.07s eta=07:03:37 +2026-09-16 10:33:57 | INFO | train rank=0 | step=43200/50000 progress=86.4% loss=0.134832 lr=4.972074e-06 lr_backbone=4.972074e-06 elapsed=44:47:30 avg_step=3.73s recent_step=4.93s eta=07:03:02 +2026-09-16 10:34:25 | INFO | train rank=0 | step=43210/50000 progress=86.4% loss=0.069908 lr=4.957708e-06 lr_backbone=4.957708e-06 elapsed=44:47:58 avg_step=3.73s recent_step=2.78s eta=07:02:23 +2026-09-16 10:35:01 | INFO | train rank=0 | step=43220/50000 progress=86.4% loss=0.098235 lr=4.943361e-06 lr_backbone=4.943361e-06 elapsed=44:48:34 avg_step=3.73s recent_step=3.58s eta=07:01:45 +2026-09-16 10:35:32 | INFO | train rank=0 | step=43230/50000 progress=86.5% loss=0.102416 lr=4.929034e-06 lr_backbone=4.929034e-06 elapsed=44:49:04 avg_step=3.73s recent_step=3.07s eta=07:01:07 +2026-09-16 10:36:21 | INFO | train rank=0 | step=43240/50000 progress=86.5% loss=0.122233 lr=4.914726e-06 lr_backbone=4.914726e-06 elapsed=44:49:54 avg_step=3.73s recent_step=4.95s eta=07:00:31 +2026-09-16 10:37:02 | INFO | train rank=0 | step=43250/50000 progress=86.5% loss=0.047255 lr=4.900438e-06 lr_backbone=4.900438e-06 elapsed=44:50:35 avg_step=3.73s recent_step=4.09s eta=06:59:55 +2026-09-16 10:37:33 | INFO | train rank=0 | step=43260/50000 progress=86.5% loss=0.089128 lr=4.886171e-06 lr_backbone=4.886171e-06 elapsed=44:51:06 avg_step=3.73s recent_step=3.14s eta=06:59:16 +2026-09-16 10:38:14 | INFO | train rank=0 | step=43270/50000 progress=86.5% loss=0.080285 lr=4.871922e-06 lr_backbone=4.871922e-06 elapsed=44:51:47 avg_step=3.73s recent_step=4.11s eta=06:58:40 +2026-09-16 10:38:49 | INFO | train rank=0 | step=43280/50000 progress=86.6% loss=0.088950 lr=4.857694e-06 lr_backbone=4.857694e-06 elapsed=44:52:21 avg_step=3.73s recent_step=3.42s eta=06:58:02 +2026-09-16 10:39:23 | INFO | train rank=0 | step=43290/50000 progress=86.6% loss=0.080736 lr=4.843485e-06 lr_backbone=4.843485e-06 elapsed=44:52:56 avg_step=3.73s recent_step=3.42s eta=06:57:24 +2026-09-16 10:40:03 | INFO | train rank=0 | step=43300/50000 progress=86.6% loss=0.071128 lr=4.829296e-06 lr_backbone=4.829296e-06 elapsed=44:53:36 avg_step=3.73s recent_step=4.05s eta=06:56:47 +2026-09-16 10:40:40 | INFO | train rank=0 | step=43310/50000 progress=86.6% loss=0.074450 lr=4.815127e-06 lr_backbone=4.815127e-06 elapsed=44:54:13 avg_step=3.73s recent_step=3.67s eta=06:56:10 +2026-09-16 10:41:14 | INFO | train rank=0 | step=43320/50000 progress=86.6% loss=0.081116 lr=4.800977e-06 lr_backbone=4.800977e-06 elapsed=44:54:47 avg_step=3.73s recent_step=3.45s eta=06:55:32 +2026-09-16 10:41:46 | INFO | train rank=0 | step=43330/50000 progress=86.7% loss=0.111322 lr=4.786847e-06 lr_backbone=4.786847e-06 elapsed=44:55:19 avg_step=3.73s recent_step=3.19s eta=06:54:54 +2026-09-16 10:42:38 | INFO | train rank=0 | step=43340/50000 progress=86.7% loss=0.074315 lr=4.772738e-06 lr_backbone=4.772738e-06 elapsed=44:56:10 avg_step=3.73s recent_step=5.11s eta=06:54:19 +2026-09-16 10:43:08 | INFO | train rank=0 | step=43350/50000 progress=86.7% loss=0.078143 lr=4.758647e-06 lr_backbone=4.758647e-06 elapsed=44:56:41 avg_step=3.73s recent_step=3.09s eta=06:53:40 +2026-09-16 10:43:40 | INFO | train rank=0 | step=43360/50000 progress=86.7% loss=0.112789 lr=4.744577e-06 lr_backbone=4.744577e-06 elapsed=44:57:13 avg_step=3.73s recent_step=3.15s eta=06:53:02 +2026-09-16 10:44:18 | INFO | train rank=0 | step=43370/50000 progress=86.7% loss=0.125463 lr=4.730526e-06 lr_backbone=4.730526e-06 elapsed=44:57:51 avg_step=3.73s recent_step=3.85s eta=06:52:25 +2026-09-16 10:45:07 | INFO | train rank=0 | step=43380/50000 progress=86.8% loss=0.072202 lr=4.716496e-06 lr_backbone=4.716496e-06 elapsed=44:58:39 avg_step=3.73s recent_step=4.82s eta=06:51:49 +2026-09-16 10:45:39 | INFO | train rank=0 | step=43390/50000 progress=86.8% loss=0.105687 lr=4.702485e-06 lr_backbone=4.702485e-06 elapsed=44:59:11 avg_step=3.73s recent_step=3.19s eta=06:51:11 +2026-09-16 10:46:20 | INFO | train rank=0 | step=43400/50000 progress=86.8% loss=0.087507 lr=4.688494e-06 lr_backbone=4.688494e-06 elapsed=44:59:52 avg_step=3.73s recent_step=4.12s eta=06:50:34 +2026-09-16 10:46:51 | INFO | train rank=0 | step=43410/50000 progress=86.8% loss=0.093605 lr=4.674522e-06 lr_backbone=4.674522e-06 elapsed=45:00:23 avg_step=3.73s recent_step=3.09s eta=06:49:56 +2026-09-16 10:47:32 | INFO | train rank=0 | step=43420/50000 progress=86.8% loss=0.116971 lr=4.660571e-06 lr_backbone=4.660571e-06 elapsed=45:01:05 avg_step=3.73s recent_step=4.15s eta=06:49:19 +2026-09-16 10:48:02 | INFO | train rank=0 | step=43430/50000 progress=86.9% loss=0.094176 lr=4.646639e-06 lr_backbone=4.646639e-06 elapsed=45:01:34 avg_step=3.73s recent_step=2.95s eta=06:48:41 +2026-09-16 10:48:35 | INFO | train rank=0 | step=43440/50000 progress=86.9% loss=0.082701 lr=4.632727e-06 lr_backbone=4.632727e-06 elapsed=45:02:08 avg_step=3.73s recent_step=3.32s eta=06:48:03 +2026-09-16 10:49:12 | INFO | train rank=0 | step=43450/50000 progress=86.9% loss=0.098332 lr=4.618835e-06 lr_backbone=4.618835e-06 elapsed=45:02:45 avg_step=3.73s recent_step=3.71s eta=06:47:26 +2026-09-16 10:49:52 | INFO | train rank=0 | step=43460/50000 progress=86.9% loss=0.101724 lr=4.604963e-06 lr_backbone=4.604963e-06 elapsed=45:03:25 avg_step=3.73s recent_step=4.03s eta=06:46:49 +2026-09-16 10:50:26 | INFO | train rank=0 | step=43470/50000 progress=86.9% loss=0.084777 lr=4.591111e-06 lr_backbone=4.591111e-06 elapsed=45:03:59 avg_step=3.73s recent_step=3.40s eta=06:46:11 +2026-09-16 10:51:00 | INFO | train rank=0 | step=43480/50000 progress=87.0% loss=0.078680 lr=4.577278e-06 lr_backbone=4.577278e-06 elapsed=45:04:33 avg_step=3.73s recent_step=3.37s eta=06:45:33 +2026-09-16 10:51:42 | INFO | train rank=0 | step=43490/50000 progress=87.0% loss=0.134954 lr=4.563466e-06 lr_backbone=4.563466e-06 elapsed=45:05:14 avg_step=3.73s recent_step=4.19s eta=06:44:56 +2026-09-16 10:52:14 | INFO | train rank=0 | step=43500/50000 progress=87.0% loss=0.101042 lr=4.549673e-06 lr_backbone=4.549673e-06 elapsed=45:05:47 avg_step=3.73s recent_step=3.25s eta=06:44:18 +2026-09-16 10:53:04 | INFO | train rank=0 | step=43510/50000 progress=87.0% loss=0.090018 lr=4.535900e-06 lr_backbone=4.535900e-06 elapsed=45:06:36 avg_step=3.73s recent_step=4.95s eta=06:43:43 +2026-09-16 10:53:46 | INFO | train rank=0 | step=43520/50000 progress=87.0% loss=0.100220 lr=4.522148e-06 lr_backbone=4.522148e-06 elapsed=45:07:18 avg_step=3.73s recent_step=4.19s eta=06:43:06 +2026-09-16 10:54:18 | INFO | train rank=0 | step=43530/50000 progress=87.1% loss=0.108610 lr=4.508415e-06 lr_backbone=4.508415e-06 elapsed=45:07:51 avg_step=3.73s recent_step=3.22s eta=06:42:28 +2026-09-16 10:55:01 | INFO | train rank=0 | step=43540/50000 progress=87.1% loss=0.116871 lr=4.494701e-06 lr_backbone=4.494701e-06 elapsed=45:08:33 avg_step=3.73s recent_step=4.29s eta=06:41:52 +2026-09-16 10:55:44 | INFO | train rank=0 | step=43550/50000 progress=87.1% loss=0.078424 lr=4.481008e-06 lr_backbone=4.481008e-06 elapsed=45:09:17 avg_step=3.73s recent_step=4.34s eta=06:41:15 +2026-09-16 10:56:18 | INFO | train rank=0 | step=43560/50000 progress=87.1% loss=0.110811 lr=4.467335e-06 lr_backbone=4.467335e-06 elapsed=45:09:51 avg_step=3.73s recent_step=3.39s eta=06:40:37 +2026-09-16 10:56:55 | INFO | train rank=0 | step=43570/50000 progress=87.1% loss=0.081608 lr=4.453682e-06 lr_backbone=4.453682e-06 elapsed=45:10:28 avg_step=3.73s recent_step=3.70s eta=06:40:00 +2026-09-16 10:57:23 | INFO | train rank=0 | step=43580/50000 progress=87.2% loss=0.082591 lr=4.440048e-06 lr_backbone=4.440048e-06 elapsed=45:10:56 avg_step=3.73s recent_step=2.78s eta=06:39:21 +2026-09-16 10:58:02 | INFO | train rank=0 | step=43590/50000 progress=87.2% loss=0.086335 lr=4.426435e-06 lr_backbone=4.426435e-06 elapsed=45:11:34 avg_step=3.73s recent_step=3.87s eta=06:38:44 +2026-09-16 10:58:39 | INFO | train rank=0 | step=43600/50000 progress=87.2% loss=0.073210 lr=4.412841e-06 lr_backbone=4.412841e-06 elapsed=45:12:12 avg_step=3.73s recent_step=3.76s eta=06:38:07 +2026-09-16 10:59:16 | INFO | train rank=0 | step=43610/50000 progress=87.2% loss=0.083797 lr=4.399267e-06 lr_backbone=4.399267e-06 elapsed=45:12:49 avg_step=3.73s recent_step=3.73s eta=06:37:30 +2026-09-16 10:59:51 | INFO | train rank=0 | step=43620/50000 progress=87.2% loss=0.110494 lr=4.385714e-06 lr_backbone=4.385714e-06 elapsed=45:13:24 avg_step=3.73s recent_step=3.47s eta=06:36:52 +2026-09-16 11:00:28 | INFO | train rank=0 | step=43630/50000 progress=87.3% loss=0.120133 lr=4.372180e-06 lr_backbone=4.372180e-06 elapsed=45:14:01 avg_step=3.73s recent_step=3.67s eta=06:36:14 +2026-09-16 11:01:12 | INFO | train rank=0 | step=43640/50000 progress=87.3% loss=0.077677 lr=4.358666e-06 lr_backbone=4.358666e-06 elapsed=45:14:45 avg_step=3.73s recent_step=4.44s eta=06:35:38 +2026-09-16 11:01:45 | INFO | train rank=0 | step=43650/50000 progress=87.3% loss=0.082312 lr=4.345172e-06 lr_backbone=4.345172e-06 elapsed=45:15:17 avg_step=3.73s recent_step=3.23s eta=06:35:00 +2026-09-16 11:02:36 | INFO | train rank=0 | step=43660/50000 progress=87.3% loss=0.110865 lr=4.331699e-06 lr_backbone=4.331699e-06 elapsed=45:16:09 avg_step=3.73s recent_step=5.18s eta=06:34:25 +2026-09-16 11:03:16 | INFO | train rank=0 | step=43670/50000 progress=87.3% loss=0.087685 lr=4.318245e-06 lr_backbone=4.318245e-06 elapsed=45:16:49 avg_step=3.73s recent_step=3.96s eta=06:33:48 +2026-09-16 11:03:51 | INFO | train rank=0 | step=43680/50000 progress=87.4% loss=0.083014 lr=4.304811e-06 lr_backbone=4.304811e-06 elapsed=45:17:24 avg_step=3.73s recent_step=3.51s eta=06:33:10 +2026-09-16 11:04:21 | INFO | train rank=0 | step=43690/50000 progress=87.4% loss=0.100417 lr=4.291397e-06 lr_backbone=4.291397e-06 elapsed=45:17:54 avg_step=3.73s recent_step=3.03s eta=06:32:32 +2026-09-16 11:05:12 | INFO | train rank=0 | step=43700/50000 progress=87.4% loss=0.097485 lr=4.278003e-06 lr_backbone=4.278003e-06 elapsed=45:18:45 avg_step=3.73s recent_step=5.03s eta=06:31:56 +2026-09-16 11:05:48 | INFO | train rank=0 | step=43710/50000 progress=87.4% loss=0.120625 lr=4.264629e-06 lr_backbone=4.264629e-06 elapsed=45:19:21 avg_step=3.73s recent_step=3.67s eta=06:31:19 +2026-09-16 11:06:16 | INFO | train rank=0 | step=43720/50000 progress=87.4% loss=0.073755 lr=4.251275e-06 lr_backbone=4.251275e-06 elapsed=45:19:49 avg_step=3.73s recent_step=2.79s eta=06:30:40 +2026-09-16 11:06:53 | INFO | train rank=0 | step=43730/50000 progress=87.5% loss=0.095603 lr=4.237941e-06 lr_backbone=4.237941e-06 elapsed=45:20:26 avg_step=3.73s recent_step=3.64s eta=06:30:03 +2026-09-16 11:07:38 | INFO | train rank=0 | step=43740/50000 progress=87.5% loss=0.094279 lr=4.224628e-06 lr_backbone=4.224628e-06 elapsed=45:21:11 avg_step=3.73s recent_step=4.50s eta=06:29:27 +2026-09-16 11:08:23 | INFO | train rank=0 | step=43750/50000 progress=87.5% loss=0.063982 lr=4.211334e-06 lr_backbone=4.211334e-06 elapsed=45:21:56 avg_step=3.73s recent_step=4.54s eta=06:28:50 +2026-09-16 11:08:56 | INFO | train rank=0 | step=43760/50000 progress=87.5% loss=0.072808 lr=4.198060e-06 lr_backbone=4.198060e-06 elapsed=45:22:29 avg_step=3.73s recent_step=3.26s eta=06:28:12 +2026-09-16 11:09:31 | INFO | train rank=0 | step=43770/50000 progress=87.5% loss=0.091484 lr=4.184806e-06 lr_backbone=4.184806e-06 elapsed=45:23:04 avg_step=3.73s recent_step=3.50s eta=06:27:35 +2026-09-16 11:10:03 | INFO | train rank=0 | step=43780/50000 progress=87.6% loss=0.113343 lr=4.171572e-06 lr_backbone=4.171572e-06 elapsed=45:23:35 avg_step=3.73s recent_step=3.18s eta=06:26:57 +2026-09-16 11:10:56 | INFO | train rank=0 | step=43790/50000 progress=87.6% loss=0.097288 lr=4.158359e-06 lr_backbone=4.158359e-06 elapsed=45:24:29 avg_step=3.73s recent_step=5.38s eta=06:26:22 +2026-09-16 11:11:36 | INFO | train rank=0 | step=43800/50000 progress=87.6% loss=0.110644 lr=4.145165e-06 lr_backbone=4.145165e-06 elapsed=45:25:09 avg_step=3.73s recent_step=3.97s eta=06:25:45 +2026-09-16 11:12:08 | INFO | train rank=0 | step=43810/50000 progress=87.6% loss=0.124499 lr=4.131991e-06 lr_backbone=4.131991e-06 elapsed=45:25:41 avg_step=3.73s recent_step=3.20s eta=06:25:07 +2026-09-16 11:12:38 | INFO | train rank=0 | step=43820/50000 progress=87.6% loss=0.097702 lr=4.118838e-06 lr_backbone=4.118838e-06 elapsed=45:26:11 avg_step=3.73s recent_step=3.02s eta=06:24:28 +2026-09-16 11:13:24 | INFO | train rank=0 | step=43830/50000 progress=87.7% loss=0.083059 lr=4.105705e-06 lr_backbone=4.105705e-06 elapsed=45:26:56 avg_step=3.73s recent_step=4.53s eta=06:23:52 +2026-09-16 11:14:00 | INFO | train rank=0 | step=43840/50000 progress=87.7% loss=0.075810 lr=4.092591e-06 lr_backbone=4.092591e-06 elapsed=45:27:33 avg_step=3.73s recent_step=3.67s eta=06:23:15 +2026-09-16 11:14:36 | INFO | train rank=0 | step=43850/50000 progress=87.7% loss=0.097050 lr=4.079498e-06 lr_backbone=4.079498e-06 elapsed=45:28:09 avg_step=3.73s recent_step=3.58s eta=06:22:37 +2026-09-16 11:15:11 | INFO | train rank=0 | step=43860/50000 progress=87.7% loss=0.079915 lr=4.066425e-06 lr_backbone=4.066425e-06 elapsed=45:28:44 avg_step=3.73s recent_step=3.50s eta=06:21:59 +2026-09-16 11:15:53 | INFO | train rank=0 | step=43870/50000 progress=87.7% loss=0.100829 lr=4.053372e-06 lr_backbone=4.053372e-06 elapsed=45:29:26 avg_step=3.73s recent_step=4.23s eta=06:21:23 +2026-09-16 11:16:33 | INFO | train rank=0 | step=43880/50000 progress=87.8% loss=0.072190 lr=4.040339e-06 lr_backbone=4.040339e-06 elapsed=45:30:06 avg_step=3.73s recent_step=3.97s eta=06:20:46 +2026-09-16 11:17:08 | INFO | train rank=0 | step=43890/50000 progress=87.8% loss=0.074934 lr=4.027326e-06 lr_backbone=4.027326e-06 elapsed=45:30:41 avg_step=3.73s recent_step=3.49s eta=06:20:08 +2026-09-16 11:17:46 | INFO | train rank=0 | step=43900/50000 progress=87.8% loss=0.116461 lr=4.014333e-06 lr_backbone=4.014333e-06 elapsed=45:31:18 avg_step=3.73s recent_step=3.76s eta=06:19:31 +2026-09-16 11:18:39 | INFO | train rank=0 | step=43910/50000 progress=87.8% loss=0.092778 lr=4.001360e-06 lr_backbone=4.001360e-06 elapsed=45:32:11 avg_step=3.73s recent_step=5.31s eta=06:18:56 +2026-09-16 11:19:12 | INFO | train rank=0 | step=43920/50000 progress=87.8% loss=0.089531 lr=3.988408e-06 lr_backbone=3.988408e-06 elapsed=45:32:44 avg_step=3.73s recent_step=3.29s eta=06:18:18 +2026-09-16 11:19:50 | INFO | train rank=0 | step=43930/50000 progress=87.9% loss=0.103629 lr=3.975475e-06 lr_backbone=3.975475e-06 elapsed=45:33:22 avg_step=3.73s recent_step=3.80s eta=06:17:40 +2026-09-16 11:20:29 | INFO | train rank=0 | step=43940/50000 progress=87.9% loss=0.067781 lr=3.962563e-06 lr_backbone=3.962563e-06 elapsed=45:34:02 avg_step=3.73s recent_step=3.96s eta=06:17:03 +2026-09-16 11:21:11 | INFO | train rank=0 | step=43950/50000 progress=87.9% loss=0.069739 lr=3.949671e-06 lr_backbone=3.949671e-06 elapsed=45:34:44 avg_step=3.73s recent_step=4.20s eta=06:16:27 +2026-09-16 11:21:46 | INFO | train rank=0 | step=43960/50000 progress=87.9% loss=0.066367 lr=3.936799e-06 lr_backbone=3.936799e-06 elapsed=45:35:19 avg_step=3.73s recent_step=3.52s eta=06:15:49 +2026-09-16 11:22:24 | INFO | train rank=0 | step=43970/50000 progress=87.9% loss=0.101311 lr=3.923947e-06 lr_backbone=3.923947e-06 elapsed=45:35:57 avg_step=3.73s recent_step=3.78s eta=06:15:12 +2026-09-16 11:22:58 | INFO | train rank=0 | step=43980/50000 progress=88.0% loss=0.084236 lr=3.911115e-06 lr_backbone=3.911115e-06 elapsed=45:36:31 avg_step=3.73s recent_step=3.38s eta=06:14:34 +2026-09-16 11:23:42 | INFO | train rank=0 | step=43990/50000 progress=88.0% loss=0.082497 lr=3.898304e-06 lr_backbone=3.898304e-06 elapsed=45:37:15 avg_step=3.73s recent_step=4.47s eta=06:13:58 +2026-09-16 11:24:16 | INFO | train rank=0 | step=44000/50000 progress=88.0% loss=0.098072 lr=3.885512e-06 lr_backbone=3.885512e-06 elapsed=45:37:49 avg_step=3.73s recent_step=3.37s eta=06:13:20 +2026-09-16 11:24:52 | INFO | train rank=0 | step=44010/50000 progress=88.0% loss=0.089261 lr=3.872741e-06 lr_backbone=3.872741e-06 elapsed=45:38:25 avg_step=3.73s recent_step=3.59s eta=06:12:42 +2026-09-16 11:25:28 | INFO | train rank=0 | step=44020/50000 progress=88.0% loss=0.074970 lr=3.859990e-06 lr_backbone=3.859990e-06 elapsed=45:39:01 avg_step=3.73s recent_step=3.63s eta=06:12:05 +2026-09-16 11:26:14 | INFO | train rank=0 | step=44030/50000 progress=88.1% loss=0.087454 lr=3.847259e-06 lr_backbone=3.847259e-06 elapsed=45:39:46 avg_step=3.73s recent_step=4.53s eta=06:11:29 +2026-09-16 11:26:38 | INFO | train rank=0 | step=44040/50000 progress=88.1% loss=0.099045 lr=3.834549e-06 lr_backbone=3.834549e-06 elapsed=45:40:11 avg_step=3.73s recent_step=2.41s eta=06:10:49 +2026-09-16 11:27:07 | INFO | train rank=0 | step=44050/50000 progress=88.1% loss=0.090444 lr=3.821858e-06 lr_backbone=3.821858e-06 elapsed=45:40:40 avg_step=3.73s recent_step=2.94s eta=06:10:11 +2026-09-16 11:27:44 | INFO | train rank=0 | step=44060/50000 progress=88.1% loss=0.124814 lr=3.809188e-06 lr_backbone=3.809188e-06 elapsed=45:41:16 avg_step=3.73s recent_step=3.64s eta=06:09:34 +2026-09-16 11:28:25 | INFO | train rank=0 | step=44070/50000 progress=88.1% loss=0.104219 lr=3.796538e-06 lr_backbone=3.796538e-06 elapsed=45:41:58 avg_step=3.73s recent_step=4.15s eta=06:08:57 +2026-09-16 11:28:54 | INFO | train rank=0 | step=44080/50000 progress=88.2% loss=0.085210 lr=3.783908e-06 lr_backbone=3.783908e-06 elapsed=45:42:27 avg_step=3.73s recent_step=2.87s eta=06:08:18 +2026-09-16 11:29:21 | INFO | train rank=0 | step=44090/50000 progress=88.2% loss=0.076414 lr=3.771298e-06 lr_backbone=3.771298e-06 elapsed=45:42:53 avg_step=3.73s recent_step=2.69s eta=06:07:40 +2026-09-16 11:29:52 | INFO | train rank=0 | step=44100/50000 progress=88.2% loss=0.087150 lr=3.758709e-06 lr_backbone=3.758709e-06 elapsed=45:43:24 avg_step=3.73s recent_step=3.10s eta=06:07:01 +2026-09-16 11:30:41 | INFO | train rank=0 | step=44110/50000 progress=88.2% loss=0.098841 lr=3.746140e-06 lr_backbone=3.746140e-06 elapsed=45:44:14 avg_step=3.73s recent_step=4.96s eta=06:06:26 +2026-09-16 11:31:06 | INFO | train rank=0 | step=44120/50000 progress=88.2% loss=0.075237 lr=3.733591e-06 lr_backbone=3.733591e-06 elapsed=45:44:39 avg_step=3.73s recent_step=2.51s eta=06:05:47 +2026-09-16 11:31:47 | INFO | train rank=0 | step=44130/50000 progress=88.3% loss=0.102890 lr=3.721062e-06 lr_backbone=3.721062e-06 elapsed=45:45:20 avg_step=3.73s recent_step=4.05s eta=06:05:10 +2026-09-16 11:32:26 | INFO | train rank=0 | step=44140/50000 progress=88.3% loss=0.096451 lr=3.708554e-06 lr_backbone=3.708554e-06 elapsed=45:45:59 avg_step=3.73s recent_step=3.89s eta=06:04:33 +2026-09-16 11:33:19 | INFO | train rank=0 | step=44150/50000 progress=88.3% loss=0.100249 lr=3.696065e-06 lr_backbone=3.696065e-06 elapsed=45:46:52 avg_step=3.73s recent_step=5.33s eta=06:03:58 +2026-09-16 11:33:55 | INFO | train rank=0 | step=44160/50000 progress=88.3% loss=0.069766 lr=3.683597e-06 lr_backbone=3.683597e-06 elapsed=45:47:28 avg_step=3.73s recent_step=3.61s eta=06:03:20 +2026-09-16 11:34:29 | INFO | train rank=0 | step=44170/50000 progress=88.3% loss=0.101155 lr=3.671150e-06 lr_backbone=3.671150e-06 elapsed=45:48:02 avg_step=3.73s recent_step=3.40s eta=06:02:42 +2026-09-16 11:35:00 | INFO | train rank=0 | step=44180/50000 progress=88.4% loss=0.098092 lr=3.658722e-06 lr_backbone=3.658722e-06 elapsed=45:48:33 avg_step=3.73s recent_step=3.06s eta=06:02:04 +2026-09-16 11:35:39 | INFO | train rank=0 | step=44190/50000 progress=88.4% loss=0.091176 lr=3.646315e-06 lr_backbone=3.646315e-06 elapsed=45:49:12 avg_step=3.73s recent_step=3.90s eta=06:01:27 +2026-09-16 11:36:10 | INFO | train rank=0 | step=44200/50000 progress=88.4% loss=0.077987 lr=3.633928e-06 lr_backbone=3.633928e-06 elapsed=45:49:43 avg_step=3.73s recent_step=3.11s eta=06:00:49 +2026-09-16 11:36:43 | INFO | train rank=0 | step=44210/50000 progress=88.4% loss=0.080994 lr=3.621562e-06 lr_backbone=3.621562e-06 elapsed=45:50:16 avg_step=3.73s recent_step=3.32s eta=06:00:11 +2026-09-16 11:37:12 | INFO | train rank=0 | step=44220/50000 progress=88.4% loss=0.091222 lr=3.609215e-06 lr_backbone=3.609215e-06 elapsed=45:50:45 avg_step=3.73s recent_step=2.89s eta=05:59:33 +2026-09-16 11:37:59 | INFO | train rank=0 | step=44230/50000 progress=88.5% loss=0.100023 lr=3.596889e-06 lr_backbone=3.596889e-06 elapsed=45:51:32 avg_step=3.73s recent_step=4.68s eta=05:58:56 +2026-09-16 11:38:38 | INFO | train rank=0 | step=44240/50000 progress=88.5% loss=0.076877 lr=3.584583e-06 lr_backbone=3.584583e-06 elapsed=45:52:11 avg_step=3.73s recent_step=3.97s eta=05:58:19 +2026-09-16 11:39:13 | INFO | train rank=0 | step=44250/50000 progress=88.5% loss=0.130760 lr=3.572298e-06 lr_backbone=3.572298e-06 elapsed=45:52:46 avg_step=3.73s recent_step=3.45s eta=05:57:42 +2026-09-16 11:39:43 | INFO | train rank=0 | step=44260/50000 progress=88.5% loss=0.077757 lr=3.560033e-06 lr_backbone=3.560033e-06 elapsed=45:53:15 avg_step=3.73s recent_step=2.97s eta=05:57:03 +2026-09-16 11:40:32 | INFO | train rank=0 | step=44270/50000 progress=88.5% loss=0.122891 lr=3.547788e-06 lr_backbone=3.547788e-06 elapsed=45:54:05 avg_step=3.73s recent_step=4.96s eta=05:56:28 +2026-09-16 11:41:11 | INFO | train rank=0 | step=44280/50000 progress=88.6% loss=0.102173 lr=3.535564e-06 lr_backbone=3.535564e-06 elapsed=45:54:44 avg_step=3.73s recent_step=3.87s eta=05:55:51 +2026-09-16 11:41:47 | INFO | train rank=0 | step=44290/50000 progress=88.6% loss=0.085431 lr=3.523360e-06 lr_backbone=3.523360e-06 elapsed=45:55:20 avg_step=3.73s recent_step=3.59s eta=05:55:13 +2026-09-16 11:42:26 | INFO | train rank=0 | step=44300/50000 progress=88.6% loss=0.068626 lr=3.511176e-06 lr_backbone=3.511176e-06 elapsed=45:55:59 avg_step=3.73s recent_step=3.95s eta=05:54:36 +2026-09-16 11:43:21 | INFO | train rank=0 | step=44310/50000 progress=88.6% loss=0.088005 lr=3.499012e-06 lr_backbone=3.499012e-06 elapsed=45:56:54 avg_step=3.73s recent_step=5.46s eta=05:54:01 +2026-09-16 11:43:54 | INFO | train rank=0 | step=44320/50000 progress=88.6% loss=0.067374 lr=3.486869e-06 lr_backbone=3.486869e-06 elapsed=45:57:26 avg_step=3.73s recent_step=3.27s eta=05:53:23 +2026-09-16 11:44:25 | INFO | train rank=0 | step=44330/50000 progress=88.7% loss=0.087619 lr=3.474746e-06 lr_backbone=3.474746e-06 elapsed=45:57:57 avg_step=3.73s recent_step=3.09s eta=05:52:45 +2026-09-16 11:44:59 | INFO | train rank=0 | step=44340/50000 progress=88.7% loss=0.081198 lr=3.462644e-06 lr_backbone=3.462644e-06 elapsed=45:58:32 avg_step=3.73s recent_step=3.46s eta=05:52:07 +2026-09-16 11:45:48 | INFO | train rank=0 | step=44350/50000 progress=88.7% loss=0.061200 lr=3.450562e-06 lr_backbone=3.450562e-06 elapsed=45:59:20 avg_step=3.73s recent_step=4.85s eta=05:51:31 +2026-09-16 11:46:22 | INFO | train rank=0 | step=44360/50000 progress=88.7% loss=0.104640 lr=3.438500e-06 lr_backbone=3.438500e-06 elapsed=45:59:55 avg_step=3.73s recent_step=3.45s eta=05:50:54 +2026-09-16 11:46:52 | INFO | train rank=0 | step=44370/50000 progress=88.7% loss=0.086626 lr=3.426459e-06 lr_backbone=3.426459e-06 elapsed=46:00:25 avg_step=3.73s recent_step=3.03s eta=05:50:15 +2026-09-16 11:47:17 | INFO | train rank=0 | step=44380/50000 progress=88.8% loss=0.076237 lr=3.414438e-06 lr_backbone=3.414438e-06 elapsed=46:00:50 avg_step=3.73s recent_step=2.45s eta=05:49:36 +2026-09-16 11:48:17 | INFO | train rank=0 | step=44390/50000 progress=88.8% loss=0.089019 lr=3.402437e-06 lr_backbone=3.402437e-06 elapsed=46:01:50 avg_step=3.73s recent_step=6.00s eta=05:49:02 +2026-09-16 11:48:49 | INFO | train rank=0 | step=44400/50000 progress=88.8% loss=0.127722 lr=3.390457e-06 lr_backbone=3.390457e-06 elapsed=46:02:21 avg_step=3.73s recent_step=3.17s eta=05:48:24 +2026-09-16 11:49:20 | INFO | train rank=0 | step=44410/50000 progress=88.8% loss=0.083729 lr=3.378497e-06 lr_backbone=3.378497e-06 elapsed=46:02:53 avg_step=3.73s recent_step=3.17s eta=05:47:46 +2026-09-16 11:49:52 | INFO | train rank=0 | step=44420/50000 progress=88.8% loss=0.085687 lr=3.366558e-06 lr_backbone=3.366558e-06 elapsed=46:03:25 avg_step=3.73s recent_step=3.15s eta=05:47:08 +2026-09-16 11:50:41 | INFO | train rank=0 | step=44430/50000 progress=88.9% loss=0.095150 lr=3.354638e-06 lr_backbone=3.354638e-06 elapsed=46:04:13 avg_step=3.73s recent_step=4.88s eta=05:46:32 +2026-09-16 11:51:15 | INFO | train rank=0 | step=44440/50000 progress=88.9% loss=0.092421 lr=3.342740e-06 lr_backbone=3.342740e-06 elapsed=46:04:48 avg_step=3.73s recent_step=3.44s eta=05:45:54 +2026-09-16 11:51:47 | INFO | train rank=0 | step=44450/50000 progress=88.9% loss=0.099085 lr=3.330862e-06 lr_backbone=3.330862e-06 elapsed=46:05:20 avg_step=3.73s recent_step=3.24s eta=05:45:16 +2026-09-16 11:52:22 | INFO | train rank=0 | step=44460/50000 progress=88.9% loss=0.100291 lr=3.319004e-06 lr_backbone=3.319004e-06 elapsed=46:05:55 avg_step=3.73s recent_step=3.46s eta=05:44:39 +2026-09-16 11:53:03 | INFO | train rank=0 | step=44470/50000 progress=88.9% loss=0.081559 lr=3.307166e-06 lr_backbone=3.307166e-06 elapsed=46:06:36 avg_step=3.73s recent_step=4.12s eta=05:44:02 +2026-09-16 11:53:38 | INFO | train rank=0 | step=44480/50000 progress=89.0% loss=0.102020 lr=3.295349e-06 lr_backbone=3.295349e-06 elapsed=46:07:11 avg_step=3.73s recent_step=3.53s eta=05:43:24 +2026-09-16 11:54:16 | INFO | train rank=0 | step=44490/50000 progress=89.0% loss=0.159901 lr=3.283553e-06 lr_backbone=3.283553e-06 elapsed=46:07:49 avg_step=3.73s recent_step=3.74s eta=05:42:47 +2026-09-16 11:54:53 | INFO | train rank=0 | step=44500/50000 progress=89.0% loss=0.096004 lr=3.271777e-06 lr_backbone=3.271777e-06 elapsed=46:08:26 avg_step=3.73s recent_step=3.73s eta=05:42:10 +2026-09-16 11:55:41 | INFO | train rank=0 | step=44510/50000 progress=89.0% loss=0.084956 lr=3.260021e-06 lr_backbone=3.260021e-06 elapsed=46:09:14 avg_step=3.73s recent_step=4.77s eta=05:41:33 +2026-09-16 11:56:08 | INFO | train rank=0 | step=44520/50000 progress=89.0% loss=0.101831 lr=3.248286e-06 lr_backbone=3.248286e-06 elapsed=46:09:41 avg_step=3.73s recent_step=2.74s eta=05:40:55 +2026-09-16 11:56:51 | INFO | train rank=0 | step=44530/50000 progress=89.1% loss=0.086850 lr=3.236571e-06 lr_backbone=3.236571e-06 elapsed=46:10:24 avg_step=3.73s recent_step=4.32s eta=05:40:18 +2026-09-16 11:57:25 | INFO | train rank=0 | step=44540/50000 progress=89.1% loss=0.086235 lr=3.224877e-06 lr_backbone=3.224877e-06 elapsed=46:10:58 avg_step=3.73s recent_step=3.38s eta=05:39:41 +2026-09-16 11:58:20 | INFO | train rank=0 | step=44550/50000 progress=89.1% loss=0.078078 lr=3.213203e-06 lr_backbone=3.213203e-06 elapsed=46:11:53 avg_step=3.73s recent_step=5.48s eta=05:39:05 +2026-09-16 11:58:47 | INFO | train rank=0 | step=44560/50000 progress=89.1% loss=0.134480 lr=3.201550e-06 lr_backbone=3.201550e-06 elapsed=46:12:20 avg_step=3.73s recent_step=2.69s eta=05:38:27 +2026-09-16 11:59:23 | INFO | train rank=0 | step=44570/50000 progress=89.1% loss=0.096958 lr=3.189917e-06 lr_backbone=3.189917e-06 elapsed=46:12:55 avg_step=3.73s recent_step=3.55s eta=05:37:49 +2026-09-16 12:00:06 | INFO | train rank=0 | step=44580/50000 progress=89.2% loss=0.089742 lr=3.178304e-06 lr_backbone=3.178304e-06 elapsed=46:13:39 avg_step=3.73s recent_step=4.34s eta=05:37:13 +2026-09-16 12:00:54 | INFO | train rank=0 | step=44590/50000 progress=89.2% loss=0.069096 lr=3.166712e-06 lr_backbone=3.166712e-06 elapsed=46:14:26 avg_step=3.73s recent_step=4.75s eta=05:36:37 +2026-09-16 12:01:28 | INFO | train rank=0 | step=44600/50000 progress=89.2% loss=0.093627 lr=3.155141e-06 lr_backbone=3.155141e-06 elapsed=46:15:01 avg_step=3.73s recent_step=3.45s eta=05:35:59 +2026-09-16 12:01:59 | INFO | train rank=0 | step=44610/50000 progress=89.2% loss=0.082877 lr=3.143590e-06 lr_backbone=3.143590e-06 elapsed=46:15:32 avg_step=3.73s recent_step=3.11s eta=05:35:21 +2026-09-16 12:02:40 | INFO | train rank=0 | step=44620/50000 progress=89.2% loss=0.089274 lr=3.132059e-06 lr_backbone=3.132059e-06 elapsed=46:16:13 avg_step=3.73s recent_step=4.08s eta=05:34:44 +2026-09-16 12:03:21 | INFO | train rank=0 | step=44630/50000 progress=89.3% loss=0.091679 lr=3.120549e-06 lr_backbone=3.120549e-06 elapsed=46:16:53 avg_step=3.73s recent_step=4.06s eta=05:34:07 +2026-09-16 12:03:55 | INFO | train rank=0 | step=44640/50000 progress=89.3% loss=0.099658 lr=3.109060e-06 lr_backbone=3.109060e-06 elapsed=46:17:28 avg_step=3.73s recent_step=3.49s eta=05:33:29 +2026-09-16 12:04:28 | INFO | train rank=0 | step=44650/50000 progress=89.3% loss=0.095099 lr=3.097591e-06 lr_backbone=3.097591e-06 elapsed=46:18:01 avg_step=3.73s recent_step=3.27s eta=05:32:51 +2026-09-16 12:05:15 | INFO | train rank=0 | step=44660/50000 progress=89.3% loss=0.092900 lr=3.086142e-06 lr_backbone=3.086142e-06 elapsed=46:18:48 avg_step=3.73s recent_step=4.69s eta=05:32:15 +2026-09-16 12:05:57 | INFO | train rank=0 | step=44670/50000 progress=89.3% loss=0.085506 lr=3.074715e-06 lr_backbone=3.074715e-06 elapsed=46:19:29 avg_step=3.73s recent_step=4.17s eta=05:31:38 +2026-09-16 12:06:29 | INFO | train rank=0 | step=44680/50000 progress=89.4% loss=0.085529 lr=3.063307e-06 lr_backbone=3.063307e-06 elapsed=46:20:02 avg_step=3.73s recent_step=3.22s eta=05:31:00 +2026-09-16 12:06:59 | INFO | train rank=0 | step=44690/50000 progress=89.4% loss=0.079956 lr=3.051920e-06 lr_backbone=3.051920e-06 elapsed=46:20:32 avg_step=3.73s recent_step=3.01s eta=05:30:22 +2026-09-16 12:07:44 | INFO | train rank=0 | step=44700/50000 progress=89.4% loss=0.108388 lr=3.040554e-06 lr_backbone=3.040554e-06 elapsed=46:21:17 avg_step=3.73s recent_step=4.54s eta=05:29:46 +2026-09-16 12:08:36 | INFO | train rank=0 | step=44710/50000 progress=89.4% loss=0.088081 lr=3.029208e-06 lr_backbone=3.029208e-06 elapsed=46:22:08 avg_step=3.73s recent_step=5.13s eta=05:29:10 +2026-09-16 12:09:17 | INFO | train rank=0 | step=44720/50000 progress=89.4% loss=0.089942 lr=3.017883e-06 lr_backbone=3.017883e-06 elapsed=46:22:49 avg_step=3.73s recent_step=4.09s eta=05:28:33 +2026-09-16 12:09:47 | INFO | train rank=0 | step=44730/50000 progress=89.5% loss=0.082772 lr=3.006578e-06 lr_backbone=3.006578e-06 elapsed=46:23:20 avg_step=3.73s recent_step=3.09s eta=05:27:55 +2026-09-16 12:10:19 | INFO | train rank=0 | step=44740/50000 progress=89.5% loss=0.108746 lr=2.995294e-06 lr_backbone=2.995294e-06 elapsed=46:23:52 avg_step=3.73s recent_step=3.16s eta=05:27:17 +2026-09-16 12:11:06 | INFO | train rank=0 | step=44750/50000 progress=89.5% loss=0.098715 lr=2.984030e-06 lr_backbone=2.984030e-06 elapsed=46:24:39 avg_step=3.73s recent_step=4.67s eta=05:26:41 +2026-09-16 12:11:39 | INFO | train rank=0 | step=44760/50000 progress=89.5% loss=0.057456 lr=2.972787e-06 lr_backbone=2.972787e-06 elapsed=46:25:12 avg_step=3.73s recent_step=3.33s eta=05:26:03 +2026-09-16 12:12:15 | INFO | train rank=0 | step=44770/50000 progress=89.5% loss=0.094900 lr=2.961565e-06 lr_backbone=2.961565e-06 elapsed=46:25:48 avg_step=3.73s recent_step=3.63s eta=05:25:26 +2026-09-16 12:12:49 | INFO | train rank=0 | step=44780/50000 progress=89.6% loss=0.125989 lr=2.950363e-06 lr_backbone=2.950363e-06 elapsed=46:26:22 avg_step=3.73s recent_step=3.38s eta=05:24:48 +2026-09-16 12:13:40 | INFO | train rank=0 | step=44790/50000 progress=89.6% loss=0.058630 lr=2.939182e-06 lr_backbone=2.939182e-06 elapsed=46:27:13 avg_step=3.73s recent_step=5.09s eta=05:24:12 +2026-09-16 12:14:14 | INFO | train rank=0 | step=44800/50000 progress=89.6% loss=0.107144 lr=2.928021e-06 lr_backbone=2.928021e-06 elapsed=46:27:47 avg_step=3.73s recent_step=3.41s eta=05:23:34 +2026-09-16 12:14:44 | INFO | train rank=0 | step=44810/50000 progress=89.6% loss=0.080389 lr=2.916881e-06 lr_backbone=2.916881e-06 elapsed=46:28:17 avg_step=3.73s recent_step=2.98s eta=05:22:56 +2026-09-16 12:15:16 | INFO | train rank=0 | step=44820/50000 progress=89.6% loss=0.103119 lr=2.905762e-06 lr_backbone=2.905762e-06 elapsed=46:28:49 avg_step=3.73s recent_step=3.17s eta=05:22:18 +2026-09-16 12:16:06 | INFO | train rank=0 | step=44830/50000 progress=89.7% loss=0.094248 lr=2.894663e-06 lr_backbone=2.894663e-06 elapsed=46:29:39 avg_step=3.73s recent_step=5.04s eta=05:21:42 +2026-09-16 12:16:41 | INFO | train rank=0 | step=44840/50000 progress=89.7% loss=0.082284 lr=2.883584e-06 lr_backbone=2.883584e-06 elapsed=46:30:14 avg_step=3.73s recent_step=3.50s eta=05:21:05 +2026-09-16 12:17:11 | INFO | train rank=0 | step=44850/50000 progress=89.7% loss=0.093501 lr=2.872527e-06 lr_backbone=2.872527e-06 elapsed=46:30:44 avg_step=3.73s recent_step=2.98s eta=05:20:27 +2026-09-16 12:17:46 | INFO | train rank=0 | step=44860/50000 progress=89.7% loss=0.061162 lr=2.861490e-06 lr_backbone=2.861490e-06 elapsed=46:31:19 avg_step=3.73s recent_step=3.54s eta=05:19:49 +2026-09-16 12:18:35 | INFO | train rank=0 | step=44870/50000 progress=89.7% loss=0.092984 lr=2.850473e-06 lr_backbone=2.850473e-06 elapsed=46:32:08 avg_step=3.73s recent_step=4.88s eta=05:19:13 +2026-09-16 12:19:08 | INFO | train rank=0 | step=44880/50000 progress=89.8% loss=0.097799 lr=2.839477e-06 lr_backbone=2.839477e-06 elapsed=46:32:40 avg_step=3.73s recent_step=3.25s eta=05:18:35 +2026-09-16 12:19:43 | INFO | train rank=0 | step=44890/50000 progress=89.8% loss=0.084463 lr=2.828502e-06 lr_backbone=2.828502e-06 elapsed=46:33:16 avg_step=3.73s recent_step=3.58s eta=05:17:58 +2026-09-16 12:20:11 | INFO | train rank=0 | step=44900/50000 progress=89.8% loss=0.070342 lr=2.817548e-06 lr_backbone=2.817548e-06 elapsed=46:33:44 avg_step=3.73s recent_step=2.74s eta=05:17:19 +2026-09-16 12:20:48 | INFO | train rank=0 | step=44910/50000 progress=89.8% loss=0.114649 lr=2.806614e-06 lr_backbone=2.806614e-06 elapsed=46:34:21 avg_step=3.73s recent_step=3.75s eta=05:16:42 +2026-09-16 12:21:19 | INFO | train rank=0 | step=44920/50000 progress=89.8% loss=0.053046 lr=2.795700e-06 lr_backbone=2.795700e-06 elapsed=46:34:52 avg_step=3.73s recent_step=3.04s eta=05:16:04 +2026-09-16 12:21:58 | INFO | train rank=0 | step=44930/50000 progress=89.9% loss=0.084051 lr=2.784808e-06 lr_backbone=2.784808e-06 elapsed=46:35:31 avg_step=3.73s recent_step=3.96s eta=05:15:27 +2026-09-16 12:22:31 | INFO | train rank=0 | step=44940/50000 progress=89.9% loss=0.085183 lr=2.773936e-06 lr_backbone=2.773936e-06 elapsed=46:36:04 avg_step=3.73s recent_step=3.25s eta=05:14:49 +2026-09-16 12:23:12 | INFO | train rank=0 | step=44950/50000 progress=89.9% loss=0.075680 lr=2.763084e-06 lr_backbone=2.763084e-06 elapsed=46:36:45 avg_step=3.73s recent_step=4.14s eta=05:14:12 +2026-09-16 12:23:39 | INFO | train rank=0 | step=44960/50000 progress=89.9% loss=0.104814 lr=2.752254e-06 lr_backbone=2.752254e-06 elapsed=46:37:12 avg_step=3.73s recent_step=2.67s eta=05:13:33 +2026-09-16 12:24:11 | INFO | train rank=0 | step=44970/50000 progress=89.9% loss=0.104107 lr=2.741444e-06 lr_backbone=2.741444e-06 elapsed=46:37:44 avg_step=3.73s recent_step=3.21s eta=05:12:56 +2026-09-16 12:24:52 | INFO | train rank=0 | step=44980/50000 progress=90.0% loss=0.082487 lr=2.730655e-06 lr_backbone=2.730655e-06 elapsed=46:38:25 avg_step=3.73s recent_step=4.11s eta=05:12:19 +2026-09-16 12:25:39 | INFO | train rank=0 | step=44990/50000 progress=90.0% loss=0.112053 lr=2.719886e-06 lr_backbone=2.719886e-06 elapsed=46:39:12 avg_step=3.73s recent_step=4.67s eta=05:11:42 +2026-09-16 12:26:16 | INFO | train rank=0 | step=45000/50000 progress=90.0% loss=0.092164 lr=2.709138e-06 lr_backbone=2.709138e-06 elapsed=46:39:48 avg_step=3.73s recent_step=3.68s eta=05:11:05 +2026-09-16 12:26:57 | INFO | train rank=0 | step=45010/50000 progress=90.0% loss=0.071190 lr=2.698411e-06 lr_backbone=2.698411e-06 elapsed=46:40:30 avg_step=3.73s recent_step=4.15s eta=05:10:28 +2026-09-16 12:27:29 | INFO | train rank=0 | step=45020/50000 progress=90.0% loss=0.077731 lr=2.687704e-06 lr_backbone=2.687704e-06 elapsed=46:41:02 avg_step=3.73s recent_step=3.24s eta=05:09:50 +2026-09-16 12:28:03 | INFO | train rank=0 | step=45030/50000 progress=90.1% loss=0.090350 lr=2.677018e-06 lr_backbone=2.677018e-06 elapsed=46:41:36 avg_step=3.73s recent_step=3.36s eta=05:09:12 +2026-09-16 12:28:35 | INFO | train rank=0 | step=45040/50000 progress=90.1% loss=0.089191 lr=2.666353e-06 lr_backbone=2.666353e-06 elapsed=46:42:08 avg_step=3.73s recent_step=3.21s eta=05:08:35 +2026-09-16 12:29:10 | INFO | train rank=0 | step=45050/50000 progress=90.1% loss=0.081954 lr=2.655709e-06 lr_backbone=2.655709e-06 elapsed=46:42:43 avg_step=3.73s recent_step=3.47s eta=05:07:57 +2026-09-16 12:29:37 | INFO | train rank=0 | step=45060/50000 progress=90.1% loss=0.100513 lr=2.645085e-06 lr_backbone=2.645085e-06 elapsed=46:43:10 avg_step=3.73s recent_step=2.76s eta=05:07:19 +2026-09-16 12:30:37 | INFO | train rank=0 | step=45070/50000 progress=90.1% loss=0.102250 lr=2.634482e-06 lr_backbone=2.634482e-06 elapsed=46:44:10 avg_step=3.73s recent_step=5.94s eta=05:06:44 +2026-09-16 12:31:17 | INFO | train rank=0 | step=45080/50000 progress=90.2% loss=0.075749 lr=2.623899e-06 lr_backbone=2.623899e-06 elapsed=46:44:50 avg_step=3.73s recent_step=4.00s eta=05:06:07 +2026-09-16 12:31:46 | INFO | train rank=0 | step=45090/50000 progress=90.2% loss=0.087889 lr=2.613338e-06 lr_backbone=2.613338e-06 elapsed=46:45:19 avg_step=3.73s recent_step=2.95s eta=05:05:28 +2026-09-16 12:32:20 | INFO | train rank=0 | step=45100/50000 progress=90.2% loss=0.097722 lr=2.602797e-06 lr_backbone=2.602797e-06 elapsed=46:45:53 avg_step=3.73s recent_step=3.39s eta=05:04:51 +2026-09-16 12:33:11 | INFO | train rank=0 | step=45110/50000 progress=90.2% loss=0.093792 lr=2.592277e-06 lr_backbone=2.592277e-06 elapsed=46:46:44 avg_step=3.73s recent_step=5.07s eta=05:04:15 +2026-09-16 12:33:47 | INFO | train rank=0 | step=45120/50000 progress=90.2% loss=0.109587 lr=2.581777e-06 lr_backbone=2.581777e-06 elapsed=46:47:20 avg_step=3.73s recent_step=3.62s eta=05:03:37 +2026-09-16 12:34:19 | INFO | train rank=0 | step=45130/50000 progress=90.3% loss=0.090604 lr=2.571299e-06 lr_backbone=2.571299e-06 elapsed=46:47:52 avg_step=3.73s recent_step=3.16s eta=05:02:59 +2026-09-16 12:34:53 | INFO | train rank=0 | step=45140/50000 progress=90.3% loss=0.087674 lr=2.560841e-06 lr_backbone=2.560841e-06 elapsed=46:48:26 avg_step=3.73s recent_step=3.45s eta=05:02:22 +2026-09-16 12:35:39 | INFO | train rank=0 | step=45150/50000 progress=90.3% loss=0.080775 lr=2.550404e-06 lr_backbone=2.550404e-06 elapsed=46:49:11 avg_step=3.73s recent_step=4.51s eta=05:01:45 +2026-09-16 12:36:11 | INFO | train rank=0 | step=45160/50000 progress=90.3% loss=0.067613 lr=2.539987e-06 lr_backbone=2.539987e-06 elapsed=46:49:43 avg_step=3.73s recent_step=3.20s eta=05:01:07 +2026-09-16 12:36:45 | INFO | train rank=0 | step=45170/50000 progress=90.3% loss=0.109886 lr=2.529591e-06 lr_backbone=2.529591e-06 elapsed=46:50:18 avg_step=3.73s recent_step=3.45s eta=05:00:30 +2026-09-16 12:37:18 | INFO | train rank=0 | step=45180/50000 progress=90.4% loss=0.094863 lr=2.519217e-06 lr_backbone=2.519217e-06 elapsed=46:50:51 avg_step=3.73s recent_step=3.30s eta=04:59:52 +2026-09-16 12:38:23 | INFO | train rank=0 | step=45190/50000 progress=90.4% loss=0.078222 lr=2.508862e-06 lr_backbone=2.508862e-06 elapsed=46:51:56 avg_step=3.73s recent_step=6.54s eta=04:59:18 +2026-09-16 12:38:55 | INFO | train rank=0 | step=45200/50000 progress=90.4% loss=0.087605 lr=2.498529e-06 lr_backbone=2.498529e-06 elapsed=46:52:28 avg_step=3.73s recent_step=3.14s eta=04:58:40 +2026-09-16 12:39:27 | INFO | train rank=0 | step=45210/50000 progress=90.4% loss=0.092298 lr=2.488217e-06 lr_backbone=2.488217e-06 elapsed=46:53:00 avg_step=3.73s recent_step=3.24s eta=04:58:02 +2026-09-16 12:40:04 | INFO | train rank=0 | step=45220/50000 progress=90.4% loss=0.086005 lr=2.477925e-06 lr_backbone=2.477925e-06 elapsed=46:53:36 avg_step=3.73s recent_step=3.63s eta=04:57:24 +2026-09-16 12:40:56 | INFO | train rank=0 | step=45230/50000 progress=90.5% loss=0.072372 lr=2.467654e-06 lr_backbone=2.467654e-06 elapsed=46:54:29 avg_step=3.73s recent_step=5.25s eta=04:56:49 +2026-09-16 12:41:27 | INFO | train rank=0 | step=45240/50000 progress=90.5% loss=0.104164 lr=2.457404e-06 lr_backbone=2.457404e-06 elapsed=46:55:00 avg_step=3.73s recent_step=3.15s eta=04:56:11 +2026-09-16 12:42:01 | INFO | train rank=0 | step=45250/50000 progress=90.5% loss=0.113781 lr=2.447174e-06 lr_backbone=2.447174e-06 elapsed=46:55:33 avg_step=3.73s recent_step=3.31s eta=04:55:33 +2026-09-16 12:42:40 | INFO | train rank=0 | step=45260/50000 progress=90.5% loss=0.089049 lr=2.436966e-06 lr_backbone=2.436966e-06 elapsed=46:56:13 avg_step=3.73s recent_step=3.95s eta=04:54:56 +2026-09-16 12:43:23 | INFO | train rank=0 | step=45270/50000 progress=90.5% loss=0.081807 lr=2.426778e-06 lr_backbone=2.426778e-06 elapsed=46:56:56 avg_step=3.73s recent_step=4.29s eta=04:54:19 +2026-09-16 12:44:00 | INFO | train rank=0 | step=45280/50000 progress=90.6% loss=0.099119 lr=2.416611e-06 lr_backbone=2.416611e-06 elapsed=46:57:33 avg_step=3.73s recent_step=3.70s eta=04:53:42 +2026-09-16 12:44:33 | INFO | train rank=0 | step=45290/50000 progress=90.6% loss=0.086270 lr=2.406465e-06 lr_backbone=2.406465e-06 elapsed=46:58:06 avg_step=3.73s recent_step=3.33s eta=04:53:04 +2026-09-16 12:45:02 | INFO | train rank=0 | step=45300/50000 progress=90.6% loss=0.085907 lr=2.396339e-06 lr_backbone=2.396339e-06 elapsed=46:58:35 avg_step=3.73s recent_step=2.85s eta=04:52:26 +2026-09-16 12:45:55 | INFO | train rank=0 | step=45310/50000 progress=90.6% loss=0.071234 lr=2.386235e-06 lr_backbone=2.386235e-06 elapsed=46:59:28 avg_step=3.73s recent_step=5.32s eta=04:51:50 +2026-09-16 12:46:22 | INFO | train rank=0 | step=45320/50000 progress=90.6% loss=0.114417 lr=2.376151e-06 lr_backbone=2.376151e-06 elapsed=46:59:55 avg_step=3.73s recent_step=2.72s eta=04:51:12 +2026-09-16 12:47:05 | INFO | train rank=0 | step=45330/50000 progress=90.7% loss=0.113973 lr=2.366088e-06 lr_backbone=2.366088e-06 elapsed=47:00:37 avg_step=3.73s recent_step=4.24s eta=04:50:35 +2026-09-16 12:47:36 | INFO | train rank=0 | step=45340/50000 progress=90.7% loss=0.068931 lr=2.356046e-06 lr_backbone=2.356046e-06 elapsed=47:01:09 avg_step=3.73s recent_step=3.18s eta=04:49:57 +2026-09-16 12:48:15 | INFO | train rank=0 | step=45350/50000 progress=90.7% loss=0.081323 lr=2.346025e-06 lr_backbone=2.346025e-06 elapsed=47:01:47 avg_step=3.73s recent_step=3.82s eta=04:49:20 +2026-09-16 12:48:51 | INFO | train rank=0 | step=45360/50000 progress=90.7% loss=0.101045 lr=2.336025e-06 lr_backbone=2.336025e-06 elapsed=47:02:24 avg_step=3.73s recent_step=3.62s eta=04:48:42 +2026-09-16 12:49:30 | INFO | train rank=0 | step=45370/50000 progress=90.7% loss=0.083529 lr=2.326045e-06 lr_backbone=2.326045e-06 elapsed=47:03:03 avg_step=3.73s recent_step=3.90s eta=04:48:05 +2026-09-16 12:50:04 | INFO | train rank=0 | step=45380/50000 progress=90.8% loss=0.090422 lr=2.316086e-06 lr_backbone=2.316086e-06 elapsed=47:03:37 avg_step=3.73s recent_step=3.41s eta=04:47:27 +2026-09-16 12:50:51 | INFO | train rank=0 | step=45390/50000 progress=90.8% loss=0.070976 lr=2.306149e-06 lr_backbone=2.306149e-06 elapsed=47:04:24 avg_step=3.73s recent_step=4.76s eta=04:46:51 +2026-09-16 12:51:18 | INFO | train rank=0 | step=45400/50000 progress=90.8% loss=0.075879 lr=2.296232e-06 lr_backbone=2.296232e-06 elapsed=47:04:51 avg_step=3.73s recent_step=2.66s eta=04:46:13 +2026-09-16 12:51:59 | INFO | train rank=0 | step=45410/50000 progress=90.8% loss=0.081157 lr=2.286336e-06 lr_backbone=2.286336e-06 elapsed=47:05:31 avg_step=3.73s recent_step=4.05s eta=04:45:36 +2026-09-16 12:52:25 | INFO | train rank=0 | step=45420/50000 progress=90.8% loss=0.095424 lr=2.276460e-06 lr_backbone=2.276460e-06 elapsed=47:05:58 avg_step=3.73s recent_step=2.66s eta=04:44:57 +2026-09-16 12:53:10 | INFO | train rank=0 | step=45430/50000 progress=90.9% loss=0.099539 lr=2.266606e-06 lr_backbone=2.266606e-06 elapsed=47:06:43 avg_step=3.73s recent_step=4.50s eta=04:44:21 +2026-09-16 12:53:47 | INFO | train rank=0 | step=45440/50000 progress=90.9% loss=0.076660 lr=2.256773e-06 lr_backbone=2.256773e-06 elapsed=47:07:20 avg_step=3.73s recent_step=3.67s eta=04:43:43 +2026-09-16 12:54:33 | INFO | train rank=0 | step=45450/50000 progress=90.9% loss=0.088770 lr=2.246960e-06 lr_backbone=2.246960e-06 elapsed=47:08:06 avg_step=3.73s recent_step=4.58s eta=04:43:07 +2026-09-16 12:55:12 | INFO | train rank=0 | step=45460/50000 progress=90.9% loss=0.079194 lr=2.237169e-06 lr_backbone=2.237169e-06 elapsed=47:08:45 avg_step=3.73s recent_step=3.96s eta=04:42:30 +2026-09-16 12:55:41 | INFO | train rank=0 | step=45470/50000 progress=90.9% loss=0.086073 lr=2.227398e-06 lr_backbone=2.227398e-06 elapsed=47:09:13 avg_step=3.73s recent_step=2.82s eta=04:41:51 +2026-09-16 12:56:09 | INFO | train rank=0 | step=45480/50000 progress=91.0% loss=0.109285 lr=2.217648e-06 lr_backbone=2.217648e-06 elapsed=47:09:41 avg_step=3.73s recent_step=2.82s eta=04:41:13 +2026-09-16 12:57:07 | INFO | train rank=0 | step=45490/50000 progress=91.0% loss=0.074552 lr=2.207919e-06 lr_backbone=2.207919e-06 elapsed=47:10:40 avg_step=3.73s recent_step=5.83s eta=04:40:38 +2026-09-16 12:57:40 | INFO | train rank=0 | step=45500/50000 progress=91.0% loss=0.082022 lr=2.198211e-06 lr_backbone=2.198211e-06 elapsed=47:11:13 avg_step=3.73s recent_step=3.30s eta=04:40:00 +2026-09-16 12:58:17 | INFO | train rank=0 | step=45510/50000 progress=91.0% loss=0.101590 lr=2.188524e-06 lr_backbone=2.188524e-06 elapsed=47:11:49 avg_step=3.73s recent_step=3.66s eta=04:39:23 +2026-09-16 12:58:52 | INFO | train rank=0 | step=45520/50000 progress=91.0% loss=0.091370 lr=2.178858e-06 lr_backbone=2.178858e-06 elapsed=47:12:25 avg_step=3.73s recent_step=3.58s eta=04:38:45 +2026-09-16 12:59:46 | INFO | train rank=0 | step=45530/50000 progress=91.1% loss=0.065414 lr=2.169212e-06 lr_backbone=2.169212e-06 elapsed=47:13:19 avg_step=3.73s recent_step=5.34s eta=04:38:10 +2026-09-16 13:00:26 | INFO | train rank=0 | step=45540/50000 progress=91.1% loss=0.110198 lr=2.159588e-06 lr_backbone=2.159588e-06 elapsed=47:13:59 avg_step=3.73s recent_step=4.00s eta=04:37:32 +2026-09-16 13:01:07 | INFO | train rank=0 | step=45550/50000 progress=91.1% loss=0.075895 lr=2.149984e-06 lr_backbone=2.149984e-06 elapsed=47:14:40 avg_step=3.73s recent_step=4.14s eta=04:36:55 +2026-09-16 13:01:33 | INFO | train rank=0 | step=45560/50000 progress=91.1% loss=0.102118 lr=2.140402e-06 lr_backbone=2.140402e-06 elapsed=47:15:06 avg_step=3.73s recent_step=2.62s eta=04:36:17 +2026-09-16 13:02:15 | INFO | train rank=0 | step=45570/50000 progress=91.1% loss=0.087491 lr=2.130840e-06 lr_backbone=2.130840e-06 elapsed=47:15:48 avg_step=3.73s recent_step=4.19s eta=04:35:40 +2026-09-16 13:02:47 | INFO | train rank=0 | step=45580/50000 progress=91.2% loss=0.082203 lr=2.121300e-06 lr_backbone=2.121300e-06 elapsed=47:16:20 avg_step=3.73s recent_step=3.14s eta=04:35:02 +2026-09-16 13:03:18 | INFO | train rank=0 | step=45590/50000 progress=91.2% loss=0.063279 lr=2.111780e-06 lr_backbone=2.111780e-06 elapsed=47:16:50 avg_step=3.73s recent_step=3.09s eta=04:34:24 +2026-09-16 13:03:53 | INFO | train rank=0 | step=45600/50000 progress=91.2% loss=0.089979 lr=2.102281e-06 lr_backbone=2.102281e-06 elapsed=47:17:26 avg_step=3.73s recent_step=3.53s eta=04:33:47 +2026-09-16 13:04:42 | INFO | train rank=0 | step=45610/50000 progress=91.2% loss=0.101291 lr=2.092803e-06 lr_backbone=2.092803e-06 elapsed=47:18:15 avg_step=3.73s recent_step=4.94s eta=04:33:11 +2026-09-16 13:05:06 | INFO | train rank=0 | step=45620/50000 progress=91.2% loss=0.091042 lr=2.083346e-06 lr_backbone=2.083346e-06 elapsed=47:18:39 avg_step=3.73s recent_step=2.38s eta=04:32:32 +2026-09-16 13:05:42 | INFO | train rank=0 | step=45630/50000 progress=91.3% loss=0.098912 lr=2.073911e-06 lr_backbone=2.073911e-06 elapsed=47:19:15 avg_step=3.73s recent_step=3.59s eta=04:31:54 +2026-09-16 13:06:20 | INFO | train rank=0 | step=45640/50000 progress=91.3% loss=0.078398 lr=2.064496e-06 lr_backbone=2.064496e-06 elapsed=47:19:53 avg_step=3.73s recent_step=3.82s eta=04:31:17 +2026-09-16 13:07:00 | INFO | train rank=0 | step=45650/50000 progress=91.3% loss=0.094185 lr=2.055102e-06 lr_backbone=2.055102e-06 elapsed=47:20:33 avg_step=3.73s recent_step=4.03s eta=04:30:40 +2026-09-16 13:07:31 | INFO | train rank=0 | step=45660/50000 progress=91.3% loss=0.128570 lr=2.045729e-06 lr_backbone=2.045729e-06 elapsed=47:21:04 avg_step=3.73s recent_step=3.08s eta=04:30:02 +2026-09-16 13:08:08 | INFO | train rank=0 | step=45670/50000 progress=91.3% loss=0.074834 lr=2.036377e-06 lr_backbone=2.036377e-06 elapsed=47:21:40 avg_step=3.73s recent_step=3.63s eta=04:29:25 +2026-09-16 13:08:45 | INFO | train rank=0 | step=45680/50000 progress=91.4% loss=0.098545 lr=2.027046e-06 lr_backbone=2.027046e-06 elapsed=47:22:18 avg_step=3.73s recent_step=3.74s eta=04:28:47 +2026-09-16 13:09:46 | INFO | train rank=0 | step=45690/50000 progress=91.4% loss=0.081907 lr=2.017736e-06 lr_backbone=2.017736e-06 elapsed=47:23:19 avg_step=3.73s recent_step=6.10s eta=04:28:12 +2026-09-16 13:10:12 | INFO | train rank=0 | step=45700/50000 progress=91.4% loss=0.121315 lr=2.008447e-06 lr_backbone=2.008447e-06 elapsed=47:23:45 avg_step=3.73s recent_step=2.65s eta=04:27:34 +2026-09-16 13:10:50 | INFO | train rank=0 | step=45710/50000 progress=91.4% loss=0.098631 lr=1.999178e-06 lr_backbone=1.999178e-06 elapsed=47:24:23 avg_step=3.73s recent_step=3.76s eta=04:26:57 +2026-09-16 13:11:22 | INFO | train rank=0 | step=45720/50000 progress=91.4% loss=0.110502 lr=1.989931e-06 lr_backbone=1.989931e-06 elapsed=47:24:55 avg_step=3.73s recent_step=3.20s eta=04:26:19 +2026-09-16 13:12:05 | INFO | train rank=0 | step=45730/50000 progress=91.5% loss=0.065232 lr=1.980705e-06 lr_backbone=1.980705e-06 elapsed=47:25:38 avg_step=3.73s recent_step=4.29s eta=04:25:42 +2026-09-16 13:12:39 | INFO | train rank=0 | step=45740/50000 progress=91.5% loss=0.096525 lr=1.971500e-06 lr_backbone=1.971500e-06 elapsed=47:26:12 avg_step=3.73s recent_step=3.43s eta=04:25:04 +2026-09-16 13:13:09 | INFO | train rank=0 | step=45750/50000 progress=91.5% loss=0.073221 lr=1.962316e-06 lr_backbone=1.962316e-06 elapsed=47:26:42 avg_step=3.73s recent_step=2.96s eta=04:24:26 +2026-09-16 13:13:40 | INFO | train rank=0 | step=45760/50000 progress=91.5% loss=0.081405 lr=1.953153e-06 lr_backbone=1.953153e-06 elapsed=47:27:12 avg_step=3.73s recent_step=3.07s eta=04:23:48 +2026-09-16 13:14:35 | INFO | train rank=0 | step=45770/50000 progress=91.5% loss=0.107529 lr=1.944011e-06 lr_backbone=1.944011e-06 elapsed=47:28:08 avg_step=3.73s recent_step=5.56s eta=04:23:13 +2026-09-16 13:15:02 | INFO | train rank=0 | step=45780/50000 progress=91.6% loss=0.096525 lr=1.934890e-06 lr_backbone=1.934890e-06 elapsed=47:28:35 avg_step=3.73s recent_step=2.72s eta=04:22:34 +2026-09-16 13:15:36 | INFO | train rank=0 | step=45790/50000 progress=91.6% loss=0.070974 lr=1.925790e-06 lr_backbone=1.925790e-06 elapsed=47:29:09 avg_step=3.73s recent_step=3.37s eta=04:21:57 +2026-09-16 13:16:10 | INFO | train rank=0 | step=45800/50000 progress=91.6% loss=0.109493 lr=1.916711e-06 lr_backbone=1.916711e-06 elapsed=47:29:42 avg_step=3.73s recent_step=3.37s eta=04:21:19 +2026-09-16 13:17:07 | INFO | train rank=0 | step=45810/50000 progress=91.6% loss=0.106101 lr=1.907653e-06 lr_backbone=1.907653e-06 elapsed=47:30:39 avg_step=3.73s recent_step=5.71s eta=04:20:44 +2026-09-16 13:17:43 | INFO | train rank=0 | step=45820/50000 progress=91.6% loss=0.105476 lr=1.898616e-06 lr_backbone=1.898616e-06 elapsed=47:31:16 avg_step=3.73s recent_step=3.60s eta=04:20:06 +2026-09-16 13:18:18 | INFO | train rank=0 | step=45830/50000 progress=91.7% loss=0.076310 lr=1.889601e-06 lr_backbone=1.889601e-06 elapsed=47:31:51 avg_step=3.73s recent_step=3.52s eta=04:19:29 +2026-09-16 13:18:49 | INFO | train rank=0 | step=45840/50000 progress=91.7% loss=0.103248 lr=1.880606e-06 lr_backbone=1.880606e-06 elapsed=47:32:22 avg_step=3.73s recent_step=3.14s eta=04:18:51 +2026-09-16 13:19:34 | INFO | train rank=0 | step=45850/50000 progress=91.7% loss=0.091656 lr=1.871632e-06 lr_backbone=1.871632e-06 elapsed=47:33:07 avg_step=3.73s recent_step=4.49s eta=04:18:14 +2026-09-16 13:20:06 | INFO | train rank=0 | step=45860/50000 progress=91.7% loss=0.066634 lr=1.862679e-06 lr_backbone=1.862679e-06 elapsed=47:33:39 avg_step=3.73s recent_step=3.17s eta=04:17:36 +2026-09-16 13:20:43 | INFO | train rank=0 | step=45870/50000 progress=91.7% loss=0.089650 lr=1.853748e-06 lr_backbone=1.853748e-06 elapsed=47:34:16 avg_step=3.73s recent_step=3.71s eta=04:16:59 +2026-09-16 13:21:18 | INFO | train rank=0 | step=45880/50000 progress=91.8% loss=0.087534 lr=1.844837e-06 lr_backbone=1.844837e-06 elapsed=47:34:50 avg_step=3.73s recent_step=3.46s eta=04:16:21 +2026-09-16 13:22:14 | INFO | train rank=0 | step=45890/50000 progress=91.8% loss=0.086059 lr=1.835948e-06 lr_backbone=1.835948e-06 elapsed=47:35:46 avg_step=3.73s recent_step=5.60s eta=04:15:46 +2026-09-16 13:22:53 | INFO | train rank=0 | step=45900/50000 progress=91.8% loss=0.093996 lr=1.827079e-06 lr_backbone=1.827079e-06 elapsed=47:36:26 avg_step=3.73s recent_step=3.97s eta=04:15:09 +2026-09-16 13:23:23 | INFO | train rank=0 | step=45910/50000 progress=91.8% loss=0.064108 lr=1.818232e-06 lr_backbone=1.818232e-06 elapsed=47:36:55 avg_step=3.73s recent_step=2.93s eta=04:14:30 +2026-09-16 13:23:58 | INFO | train rank=0 | step=45920/50000 progress=91.8% loss=0.097538 lr=1.809406e-06 lr_backbone=1.809406e-06 elapsed=47:37:31 avg_step=3.73s recent_step=3.55s eta=04:13:53 +2026-09-16 13:24:47 | INFO | train rank=0 | step=45930/50000 progress=91.9% loss=0.087003 lr=1.800600e-06 lr_backbone=1.800600e-06 elapsed=47:38:20 avg_step=3.73s recent_step=4.93s eta=04:13:17 +2026-09-16 13:25:19 | INFO | train rank=0 | step=45940/50000 progress=91.9% loss=0.126046 lr=1.791816e-06 lr_backbone=1.791816e-06 elapsed=47:38:51 avg_step=3.73s recent_step=3.13s eta=04:12:39 +2026-09-16 13:25:47 | INFO | train rank=0 | step=45950/50000 progress=91.9% loss=0.082694 lr=1.783053e-06 lr_backbone=1.783053e-06 elapsed=47:39:20 avg_step=3.73s recent_step=2.83s eta=04:12:01 +2026-09-16 13:26:13 | INFO | train rank=0 | step=45960/50000 progress=91.9% loss=0.086839 lr=1.774311e-06 lr_backbone=1.774311e-06 elapsed=47:39:46 avg_step=3.73s recent_step=2.61s eta=04:11:22 +2026-09-16 13:26:53 | INFO | train rank=0 | step=45970/50000 progress=91.9% loss=0.097269 lr=1.765590e-06 lr_backbone=1.765590e-06 elapsed=47:40:26 avg_step=3.73s recent_step=4.04s eta=04:10:45 +2026-09-16 13:27:32 | INFO | train rank=0 | step=45980/50000 progress=92.0% loss=0.100885 lr=1.756891e-06 lr_backbone=1.756891e-06 elapsed=47:41:05 avg_step=3.73s recent_step=3.82s eta=04:10:08 +2026-09-16 13:28:04 | INFO | train rank=0 | step=45990/50000 progress=92.0% loss=0.068787 lr=1.748212e-06 lr_backbone=1.748212e-06 elapsed=47:41:37 avg_step=3.73s recent_step=3.25s eta=04:09:30 +2026-09-16 13:28:43 | INFO | train rank=0 | step=46000/50000 progress=92.0% loss=0.051186 lr=1.739554e-06 lr_backbone=1.739554e-06 elapsed=47:42:16 avg_step=3.73s recent_step=3.90s eta=04:08:53 +2026-09-16 13:29:27 | INFO | train rank=0 | step=46010/50000 progress=92.0% loss=0.104376 lr=1.730918e-06 lr_backbone=1.730918e-06 elapsed=47:42:59 avg_step=3.73s recent_step=4.35s eta=04:08:16 +2026-09-16 13:29:58 | INFO | train rank=0 | step=46020/50000 progress=92.0% loss=0.088443 lr=1.722303e-06 lr_backbone=1.722303e-06 elapsed=47:43:31 avg_step=3.73s recent_step=3.17s eta=04:07:38 +2026-09-16 13:30:27 | INFO | train rank=0 | step=46030/50000 progress=92.1% loss=0.072483 lr=1.713709e-06 lr_backbone=1.713709e-06 elapsed=47:44:00 avg_step=3.73s recent_step=2.87s eta=04:07:00 +2026-09-16 13:31:03 | INFO | train rank=0 | step=46040/50000 progress=92.1% loss=0.109376 lr=1.705135e-06 lr_backbone=1.705135e-06 elapsed=47:44:35 avg_step=3.73s recent_step=3.55s eta=04:06:23 +2026-09-16 13:31:57 | INFO | train rank=0 | step=46050/50000 progress=92.1% loss=0.088807 lr=1.696584e-06 lr_backbone=1.696584e-06 elapsed=47:45:29 avg_step=3.73s recent_step=5.40s eta=04:05:47 +2026-09-16 13:32:30 | INFO | train rank=0 | step=46060/50000 progress=92.1% loss=0.085552 lr=1.688053e-06 lr_backbone=1.688053e-06 elapsed=47:46:03 avg_step=3.73s recent_step=3.38s eta=04:05:09 +2026-09-16 13:33:05 | INFO | train rank=0 | step=46070/50000 progress=92.1% loss=0.109390 lr=1.679543e-06 lr_backbone=1.679543e-06 elapsed=47:46:38 avg_step=3.73s recent_step=3.48s eta=04:04:32 +2026-09-16 13:33:33 | INFO | train rank=0 | step=46080/50000 progress=92.2% loss=0.089949 lr=1.671054e-06 lr_backbone=1.671054e-06 elapsed=47:47:06 avg_step=3.73s recent_step=2.75s eta=04:03:54 +2026-09-16 13:34:23 | INFO | train rank=0 | step=46090/50000 progress=92.2% loss=0.084113 lr=1.662587e-06 lr_backbone=1.662587e-06 elapsed=47:47:56 avg_step=3.73s recent_step=5.07s eta=04:03:17 +2026-09-16 13:34:53 | INFO | train rank=0 | step=46100/50000 progress=92.2% loss=0.075349 lr=1.654141e-06 lr_backbone=1.654141e-06 elapsed=47:48:26 avg_step=3.73s recent_step=2.93s eta=04:02:39 +2026-09-16 13:35:19 | INFO | train rank=0 | step=46110/50000 progress=92.2% loss=0.066256 lr=1.645716e-06 lr_backbone=1.645716e-06 elapsed=47:48:51 avg_step=3.73s recent_step=2.58s eta=04:02:01 +2026-09-16 13:35:56 | INFO | train rank=0 | step=46120/50000 progress=92.2% loss=0.066365 lr=1.637312e-06 lr_backbone=1.637312e-06 elapsed=47:49:29 avg_step=3.73s recent_step=3.72s eta=04:01:24 +2026-09-16 13:36:49 | INFO | train rank=0 | step=46130/50000 progress=92.3% loss=0.101526 lr=1.628929e-06 lr_backbone=1.628929e-06 elapsed=47:50:22 avg_step=3.73s recent_step=5.30s eta=04:00:48 +2026-09-16 13:37:22 | INFO | train rank=0 | step=46140/50000 progress=92.3% loss=0.091232 lr=1.620567e-06 lr_backbone=1.620567e-06 elapsed=47:50:55 avg_step=3.73s recent_step=3.30s eta=04:00:10 +2026-09-16 13:37:56 | INFO | train rank=0 | step=46150/50000 progress=92.3% loss=0.083600 lr=1.612227e-06 lr_backbone=1.612227e-06 elapsed=47:51:29 avg_step=3.73s recent_step=3.39s eta=03:59:32 +2026-09-16 13:38:31 | INFO | train rank=0 | step=46160/50000 progress=92.3% loss=0.078382 lr=1.603907e-06 lr_backbone=1.603907e-06 elapsed=47:52:04 avg_step=3.73s recent_step=3.53s eta=03:58:55 +2026-09-16 13:39:25 | INFO | train rank=0 | step=46170/50000 progress=92.3% loss=0.087663 lr=1.595609e-06 lr_backbone=1.595609e-06 elapsed=47:52:58 avg_step=3.73s recent_step=5.37s eta=03:58:19 +2026-09-16 13:40:00 | INFO | train rank=0 | step=46180/50000 progress=92.4% loss=0.091524 lr=1.587332e-06 lr_backbone=1.587332e-06 elapsed=47:53:33 avg_step=3.73s recent_step=3.50s eta=03:57:41 +2026-09-16 13:40:25 | INFO | train rank=0 | step=46190/50000 progress=92.4% loss=0.079551 lr=1.579077e-06 lr_backbone=1.579077e-06 elapsed=47:53:57 avg_step=3.73s recent_step=2.46s eta=03:57:03 +2026-09-16 13:41:04 | INFO | train rank=0 | step=46200/50000 progress=92.4% loss=0.077398 lr=1.570842e-06 lr_backbone=1.570842e-06 elapsed=47:54:36 avg_step=3.73s recent_step=3.91s eta=03:56:26 +2026-09-16 13:41:50 | INFO | train rank=0 | step=46210/50000 progress=92.4% loss=0.102148 lr=1.562629e-06 lr_backbone=1.562629e-06 elapsed=47:55:23 avg_step=3.73s recent_step=4.65s eta=03:55:49 +2026-09-16 13:42:20 | INFO | train rank=0 | step=46220/50000 progress=92.4% loss=0.096005 lr=1.554436e-06 lr_backbone=1.554436e-06 elapsed=47:55:53 avg_step=3.73s recent_step=3.00s eta=03:55:11 +2026-09-16 13:42:52 | INFO | train rank=0 | step=46230/50000 progress=92.5% loss=0.084962 lr=1.546265e-06 lr_backbone=1.546265e-06 elapsed=47:56:25 avg_step=3.73s recent_step=3.22s eta=03:54:34 +2026-09-16 13:43:25 | INFO | train rank=0 | step=46240/50000 progress=92.5% loss=0.088009 lr=1.538115e-06 lr_backbone=1.538115e-06 elapsed=47:56:58 avg_step=3.73s recent_step=3.30s eta=03:53:56 +2026-09-16 13:44:12 | INFO | train rank=0 | step=46250/50000 progress=92.5% loss=0.067423 lr=1.529987e-06 lr_backbone=1.529987e-06 elapsed=47:57:45 avg_step=3.73s recent_step=4.71s eta=03:53:19 +2026-09-16 13:44:46 | INFO | train rank=0 | step=46260/50000 progress=92.5% loss=0.097445 lr=1.521879e-06 lr_backbone=1.521879e-06 elapsed=47:58:18 avg_step=3.73s recent_step=3.32s eta=03:52:42 +2026-09-16 13:45:21 | INFO | train rank=0 | step=46270/50000 progress=92.5% loss=0.088855 lr=1.513793e-06 lr_backbone=1.513793e-06 elapsed=47:58:54 avg_step=3.73s recent_step=3.55s eta=03:52:04 +2026-09-16 13:45:57 | INFO | train rank=0 | step=46280/50000 progress=92.6% loss=0.101801 lr=1.505728e-06 lr_backbone=1.505728e-06 elapsed=47:59:30 avg_step=3.73s recent_step=3.62s eta=03:51:27 +2026-09-16 13:46:56 | INFO | train rank=0 | step=46290/50000 progress=92.6% loss=0.116376 lr=1.497684e-06 lr_backbone=1.497684e-06 elapsed=48:00:28 avg_step=3.73s recent_step=5.81s eta=03:50:51 +2026-09-16 13:47:29 | INFO | train rank=0 | step=46300/50000 progress=92.6% loss=0.105725 lr=1.489662e-06 lr_backbone=1.489662e-06 elapsed=48:01:01 avg_step=3.73s recent_step=3.31s eta=03:50:14 +2026-09-16 13:47:58 | INFO | train rank=0 | step=46310/50000 progress=92.6% loss=0.088577 lr=1.481660e-06 lr_backbone=1.481660e-06 elapsed=48:01:31 avg_step=3.73s recent_step=2.94s eta=03:49:36 +2026-09-16 13:48:34 | INFO | train rank=0 | step=46320/50000 progress=92.6% loss=0.121421 lr=1.473680e-06 lr_backbone=1.473680e-06 elapsed=48:02:07 avg_step=3.73s recent_step=3.62s eta=03:48:58 +2026-09-16 13:49:18 | INFO | train rank=0 | step=46330/50000 progress=92.7% loss=0.075663 lr=1.465721e-06 lr_backbone=1.465721e-06 elapsed=48:02:51 avg_step=3.73s recent_step=4.40s eta=03:48:21 +2026-09-16 13:49:58 | INFO | train rank=0 | step=46340/50000 progress=92.7% loss=0.082923 lr=1.457783e-06 lr_backbone=1.457783e-06 elapsed=48:03:31 avg_step=3.73s recent_step=3.97s eta=03:47:44 +2026-09-16 13:50:38 | INFO | train rank=0 | step=46350/50000 progress=92.7% loss=0.072840 lr=1.449867e-06 lr_backbone=1.449867e-06 elapsed=48:04:10 avg_step=3.73s recent_step=3.96s eta=03:47:07 +2026-09-16 13:51:08 | INFO | train rank=0 | step=46360/50000 progress=92.7% loss=0.130011 lr=1.441972e-06 lr_backbone=1.441972e-06 elapsed=48:04:41 avg_step=3.73s recent_step=3.09s eta=03:46:29 +2026-09-16 13:51:57 | INFO | train rank=0 | step=46370/50000 progress=92.7% loss=0.069684 lr=1.434098e-06 lr_backbone=1.434098e-06 elapsed=48:05:30 avg_step=3.73s recent_step=4.87s eta=03:45:53 +2026-09-16 13:52:28 | INFO | train rank=0 | step=46380/50000 progress=92.8% loss=0.099823 lr=1.426245e-06 lr_backbone=1.426245e-06 elapsed=48:06:00 avg_step=3.73s recent_step=3.05s eta=03:45:15 +2026-09-16 13:53:05 | INFO | train rank=0 | step=46390/50000 progress=92.8% loss=0.116233 lr=1.418413e-06 lr_backbone=1.418413e-06 elapsed=48:06:38 avg_step=3.73s recent_step=3.71s eta=03:44:38 +2026-09-16 13:53:41 | INFO | train rank=0 | step=46400/50000 progress=92.8% loss=0.098868 lr=1.410603e-06 lr_backbone=1.410603e-06 elapsed=48:07:14 avg_step=3.73s recent_step=3.67s eta=03:44:00 +2026-09-16 13:54:27 | INFO | train rank=0 | step=46410/50000 progress=92.8% loss=0.089920 lr=1.402814e-06 lr_backbone=1.402814e-06 elapsed=48:08:00 avg_step=3.73s recent_step=4.58s eta=03:43:23 +2026-09-16 13:55:04 | INFO | train rank=0 | step=46420/50000 progress=92.8% loss=0.081267 lr=1.395046e-06 lr_backbone=1.395046e-06 elapsed=48:08:37 avg_step=3.73s recent_step=3.68s eta=03:42:46 +2026-09-16 13:55:34 | INFO | train rank=0 | step=46430/50000 progress=92.9% loss=0.101649 lr=1.387300e-06 lr_backbone=1.387300e-06 elapsed=48:09:07 avg_step=3.73s recent_step=3.03s eta=03:42:08 +2026-09-16 13:56:10 | INFO | train rank=0 | step=46440/50000 progress=92.9% loss=0.100833 lr=1.379575e-06 lr_backbone=1.379575e-06 elapsed=48:09:42 avg_step=3.73s recent_step=3.54s eta=03:41:31 +2026-09-16 13:56:59 | INFO | train rank=0 | step=46450/50000 progress=92.9% loss=0.073953 lr=1.371871e-06 lr_backbone=1.371871e-06 elapsed=48:10:32 avg_step=3.73s recent_step=4.93s eta=03:40:54 +2026-09-16 13:57:31 | INFO | train rank=0 | step=46460/50000 progress=92.9% loss=0.097795 lr=1.364188e-06 lr_backbone=1.364188e-06 elapsed=48:11:03 avg_step=3.73s recent_step=3.16s eta=03:40:17 +2026-09-16 13:58:09 | INFO | train rank=0 | step=46470/50000 progress=92.9% loss=0.098818 lr=1.356527e-06 lr_backbone=1.356527e-06 elapsed=48:11:42 avg_step=3.73s recent_step=3.81s eta=03:39:39 +2026-09-16 13:58:41 | INFO | train rank=0 | step=46480/50000 progress=93.0% loss=0.082395 lr=1.348887e-06 lr_backbone=1.348887e-06 elapsed=48:12:14 avg_step=3.73s recent_step=3.25s eta=03:39:02 +2026-09-16 13:59:19 | INFO | train rank=0 | step=46490/50000 progress=93.0% loss=0.097102 lr=1.341268e-06 lr_backbone=1.341268e-06 elapsed=48:12:51 avg_step=3.73s recent_step=3.75s eta=03:38:24 +2026-09-16 13:59:57 | INFO | train rank=0 | step=46500/50000 progress=93.0% loss=0.100669 lr=1.333670e-06 lr_backbone=1.333670e-06 elapsed=48:13:30 avg_step=3.73s recent_step=3.81s eta=03:37:47 +2026-09-16 14:00:29 | INFO | train rank=0 | step=46510/50000 progress=93.0% loss=0.076859 lr=1.326094e-06 lr_backbone=1.326094e-06 elapsed=48:14:02 avg_step=3.73s recent_step=3.23s eta=03:37:09 +2026-09-16 14:01:06 | INFO | train rank=0 | step=46520/50000 progress=93.0% loss=0.105408 lr=1.318539e-06 lr_backbone=1.318539e-06 elapsed=48:14:38 avg_step=3.73s recent_step=3.67s eta=03:36:32 +2026-09-16 14:02:02 | INFO | train rank=0 | step=46530/50000 progress=93.1% loss=0.119474 lr=1.311005e-06 lr_backbone=1.311005e-06 elapsed=48:15:34 avg_step=3.73s recent_step=5.58s eta=03:35:56 +2026-09-16 14:02:36 | INFO | train rank=0 | step=46540/50000 progress=93.1% loss=0.080000 lr=1.303493e-06 lr_backbone=1.303493e-06 elapsed=48:16:09 avg_step=3.73s recent_step=3.43s eta=03:35:18 +2026-09-16 14:03:02 | INFO | train rank=0 | step=46550/50000 progress=93.1% loss=0.105972 lr=1.296002e-06 lr_backbone=1.296002e-06 elapsed=48:16:34 avg_step=3.73s recent_step=2.59s eta=03:34:40 +2026-09-16 14:03:36 | INFO | train rank=0 | step=46560/50000 progress=93.1% loss=0.079225 lr=1.288532e-06 lr_backbone=1.288532e-06 elapsed=48:17:08 avg_step=3.73s recent_step=3.39s eta=03:34:03 +2026-09-16 14:04:19 | INFO | train rank=0 | step=46570/50000 progress=93.1% loss=0.071866 lr=1.281084e-06 lr_backbone=1.281084e-06 elapsed=48:17:52 avg_step=3.73s recent_step=4.33s eta=03:33:26 +2026-09-16 14:04:53 | INFO | train rank=0 | step=46580/50000 progress=93.2% loss=0.075441 lr=1.273656e-06 lr_backbone=1.273656e-06 elapsed=48:18:26 avg_step=3.73s recent_step=3.44s eta=03:32:48 +2026-09-16 14:05:44 | INFO | train rank=0 | step=46590/50000 progress=93.2% loss=0.087237 lr=1.266251e-06 lr_backbone=1.266251e-06 elapsed=48:19:17 avg_step=3.73s recent_step=5.09s eta=03:32:12 +2026-09-16 14:06:10 | INFO | train rank=0 | step=46600/50000 progress=93.2% loss=0.065650 lr=1.258866e-06 lr_backbone=1.258866e-06 elapsed=48:19:42 avg_step=3.73s recent_step=2.55s eta=03:31:34 +2026-09-16 14:06:43 | INFO | train rank=0 | step=46610/50000 progress=93.2% loss=0.080278 lr=1.251503e-06 lr_backbone=1.251503e-06 elapsed=48:20:16 avg_step=3.73s recent_step=3.34s eta=03:30:56 +2026-09-16 14:07:18 | INFO | train rank=0 | step=46620/50000 progress=93.2% loss=0.082867 lr=1.244161e-06 lr_backbone=1.244161e-06 elapsed=48:20:51 avg_step=3.73s recent_step=3.49s eta=03:30:18 +2026-09-16 14:07:57 | INFO | train rank=0 | step=46630/50000 progress=93.3% loss=0.069453 lr=1.236840e-06 lr_backbone=1.236840e-06 elapsed=48:21:30 avg_step=3.73s recent_step=3.95s eta=03:29:41 +2026-09-16 14:08:29 | INFO | train rank=0 | step=46640/50000 progress=93.3% loss=0.069750 lr=1.229541e-06 lr_backbone=1.229541e-06 elapsed=48:22:02 avg_step=3.73s recent_step=3.20s eta=03:29:04 +2026-09-16 14:09:16 | INFO | train rank=0 | step=46650/50000 progress=93.3% loss=0.107400 lr=1.222263e-06 lr_backbone=1.222263e-06 elapsed=48:22:49 avg_step=3.73s recent_step=4.69s eta=03:28:27 +2026-09-16 14:09:52 | INFO | train rank=0 | step=46660/50000 progress=93.3% loss=0.084144 lr=1.215007e-06 lr_backbone=1.215007e-06 elapsed=48:23:25 avg_step=3.73s recent_step=3.58s eta=03:27:49 +2026-09-16 14:10:42 | INFO | train rank=0 | step=46670/50000 progress=93.3% loss=0.098866 lr=1.207772e-06 lr_backbone=1.207772e-06 elapsed=48:24:15 avg_step=3.73s recent_step=5.02s eta=03:27:13 +2026-09-16 14:11:16 | INFO | train rank=0 | step=46680/50000 progress=93.4% loss=0.088208 lr=1.200558e-06 lr_backbone=1.200558e-06 elapsed=48:24:49 avg_step=3.73s recent_step=3.40s eta=03:26:35 +2026-09-16 14:11:54 | INFO | train rank=0 | step=46690/50000 progress=93.4% loss=0.084084 lr=1.193365e-06 lr_backbone=1.193365e-06 elapsed=48:25:27 avg_step=3.73s recent_step=3.78s eta=03:25:58 +2026-09-16 14:12:30 | INFO | train rank=0 | step=46700/50000 progress=93.4% loss=0.083073 lr=1.186194e-06 lr_backbone=1.186194e-06 elapsed=48:26:03 avg_step=3.73s recent_step=3.62s eta=03:25:21 +2026-09-16 14:13:17 | INFO | train rank=0 | step=46710/50000 progress=93.4% loss=0.070514 lr=1.179044e-06 lr_backbone=1.179044e-06 elapsed=48:26:50 avg_step=3.73s recent_step=4.64s eta=03:24:44 +2026-09-16 14:13:52 | INFO | train rank=0 | step=46720/50000 progress=93.4% loss=0.075468 lr=1.171916e-06 lr_backbone=1.171916e-06 elapsed=48:27:24 avg_step=3.73s recent_step=3.47s eta=03:24:06 +2026-09-16 14:14:24 | INFO | train rank=0 | step=46730/50000 progress=93.5% loss=0.082825 lr=1.164809e-06 lr_backbone=1.164809e-06 elapsed=48:27:57 avg_step=3.73s recent_step=3.23s eta=03:23:29 +2026-09-16 14:14:57 | INFO | train rank=0 | step=46740/50000 progress=93.5% loss=0.108881 lr=1.157723e-06 lr_backbone=1.157723e-06 elapsed=48:28:29 avg_step=3.73s recent_step=3.29s eta=03:22:51 +2026-09-16 14:15:45 | INFO | train rank=0 | step=46750/50000 progress=93.5% loss=0.094287 lr=1.150658e-06 lr_backbone=1.150658e-06 elapsed=48:29:18 avg_step=3.73s recent_step=4.85s eta=03:22:15 +2026-09-16 14:16:24 | INFO | train rank=0 | step=46760/50000 progress=93.5% loss=0.068689 lr=1.143615e-06 lr_backbone=1.143615e-06 elapsed=48:29:57 avg_step=3.73s recent_step=3.91s eta=03:21:37 +2026-09-16 14:16:57 | INFO | train rank=0 | step=46770/50000 progress=93.5% loss=0.085647 lr=1.136594e-06 lr_backbone=1.136594e-06 elapsed=48:30:30 avg_step=3.73s recent_step=3.27s eta=03:21:00 +2026-09-16 14:17:37 | INFO | train rank=0 | step=46780/50000 progress=93.6% loss=0.104146 lr=1.129594e-06 lr_backbone=1.129594e-06 elapsed=48:31:10 avg_step=3.73s recent_step=4.01s eta=03:20:23 +2026-09-16 14:18:26 | INFO | train rank=0 | step=46790/50000 progress=93.6% loss=0.119369 lr=1.122615e-06 lr_backbone=1.122615e-06 elapsed=48:31:58 avg_step=3.73s recent_step=4.85s eta=03:19:46 +2026-09-16 14:18:55 | INFO | train rank=0 | step=46800/50000 progress=93.6% loss=0.089075 lr=1.115657e-06 lr_backbone=1.115657e-06 elapsed=48:32:28 avg_step=3.73s recent_step=2.91s eta=03:19:08 +2026-09-16 14:19:34 | INFO | train rank=0 | step=46810/50000 progress=93.6% loss=0.093109 lr=1.108721e-06 lr_backbone=1.108721e-06 elapsed=48:33:07 avg_step=3.73s recent_step=3.89s eta=03:18:31 +2026-09-16 14:20:09 | INFO | train rank=0 | step=46820/50000 progress=93.6% loss=0.095983 lr=1.101806e-06 lr_backbone=1.101806e-06 elapsed=48:33:42 avg_step=3.73s recent_step=3.57s eta=03:17:53 +2026-09-16 14:20:58 | INFO | train rank=0 | step=46830/50000 progress=93.7% loss=0.077624 lr=1.094913e-06 lr_backbone=1.094913e-06 elapsed=48:34:31 avg_step=3.73s recent_step=4.85s eta=03:17:17 +2026-09-16 14:21:32 | INFO | train rank=0 | step=46840/50000 progress=93.7% loss=0.093295 lr=1.088041e-06 lr_backbone=1.088041e-06 elapsed=48:35:05 avg_step=3.73s recent_step=3.38s eta=03:16:39 +2026-09-16 14:22:05 | INFO | train rank=0 | step=46850/50000 progress=93.7% loss=0.093241 lr=1.081190e-06 lr_backbone=1.081190e-06 elapsed=48:35:38 avg_step=3.73s recent_step=3.32s eta=03:16:02 +2026-09-16 14:22:40 | INFO | train rank=0 | step=46860/50000 progress=93.7% loss=0.105790 lr=1.074361e-06 lr_backbone=1.074361e-06 elapsed=48:36:13 avg_step=3.73s recent_step=3.50s eta=03:15:24 +2026-09-16 14:23:32 | INFO | train rank=0 | step=46870/50000 progress=93.7% loss=0.101262 lr=1.067554e-06 lr_backbone=1.067554e-06 elapsed=48:37:05 avg_step=3.73s recent_step=5.24s eta=03:14:48 +2026-09-16 14:23:56 | INFO | train rank=0 | step=46880/50000 progress=93.8% loss=0.064311 lr=1.060767e-06 lr_backbone=1.060767e-06 elapsed=48:37:29 avg_step=3.73s recent_step=2.41s eta=03:14:10 +2026-09-16 14:24:37 | INFO | train rank=0 | step=46890/50000 progress=93.8% loss=0.103679 lr=1.054002e-06 lr_backbone=1.054002e-06 elapsed=48:38:09 avg_step=3.73s recent_step=4.02s eta=03:13:32 +2026-09-16 14:25:06 | INFO | train rank=0 | step=46900/50000 progress=93.8% loss=0.096656 lr=1.047259e-06 lr_backbone=1.047259e-06 elapsed=48:38:38 avg_step=3.73s recent_step=2.91s eta=03:12:55 +2026-09-16 14:25:52 | INFO | train rank=0 | step=46910/50000 progress=93.8% loss=0.077192 lr=1.040537e-06 lr_backbone=1.040537e-06 elapsed=48:39:25 avg_step=3.73s recent_step=4.62s eta=03:12:18 +2026-09-16 14:26:18 | INFO | train rank=0 | step=46920/50000 progress=93.8% loss=0.095989 lr=1.033836e-06 lr_backbone=1.033836e-06 elapsed=48:39:51 avg_step=3.73s recent_step=2.64s eta=03:11:40 +2026-09-16 14:26:53 | INFO | train rank=0 | step=46930/50000 progress=93.9% loss=0.072397 lr=1.027157e-06 lr_backbone=1.027157e-06 elapsed=48:40:25 avg_step=3.73s recent_step=3.44s eta=03:11:02 +2026-09-16 14:27:32 | INFO | train rank=0 | step=46940/50000 progress=93.9% loss=0.098740 lr=1.020499e-06 lr_backbone=1.020499e-06 elapsed=48:41:05 avg_step=3.73s recent_step=3.92s eta=03:10:25 +2026-09-16 14:28:24 | INFO | train rank=0 | step=46950/50000 progress=93.9% loss=0.074636 lr=1.013862e-06 lr_backbone=1.013862e-06 elapsed=48:41:57 avg_step=3.73s recent_step=5.19s eta=03:09:49 +2026-09-16 14:28:54 | INFO | train rank=0 | step=46960/50000 progress=93.9% loss=0.082809 lr=1.007247e-06 lr_backbone=1.007247e-06 elapsed=48:42:27 avg_step=3.73s recent_step=3.06s eta=03:09:11 +2026-09-16 14:29:33 | INFO | train rank=0 | step=46970/50000 progress=93.9% loss=0.085460 lr=1.000654e-06 lr_backbone=1.000654e-06 elapsed=48:43:06 avg_step=3.73s recent_step=3.86s eta=03:08:34 +2026-09-16 14:30:08 | INFO | train rank=0 | step=46980/50000 progress=94.0% loss=0.093652 lr=9.940817e-07 lr_backbone=9.940817e-07 elapsed=48:43:41 avg_step=3.73s recent_step=3.52s eta=03:07:56 +2026-09-16 14:30:58 | INFO | train rank=0 | step=46990/50000 progress=94.0% loss=0.084017 lr=9.875310e-07 lr_backbone=9.875310e-07 elapsed=48:44:31 avg_step=3.73s recent_step=5.00s eta=03:07:20 +2026-09-16 14:31:27 | INFO | train rank=0 | step=47000/50000 progress=94.0% loss=0.071367 lr=9.810017e-07 lr_backbone=9.810017e-07 elapsed=48:45:00 avg_step=3.73s recent_step=2.92s eta=03:06:42 +2026-09-16 14:31:54 | INFO | train rank=0 | step=47010/50000 progress=94.0% loss=0.073676 lr=9.744939e-07 lr_backbone=9.744939e-07 elapsed=48:45:27 avg_step=3.73s recent_step=2.68s eta=03:06:04 +2026-09-16 14:32:28 | INFO | train rank=0 | step=47020/50000 progress=94.0% loss=0.106890 lr=9.680075e-07 lr_backbone=9.680075e-07 elapsed=48:46:01 avg_step=3.73s recent_step=3.43s eta=03:05:26 +2026-09-16 14:33:14 | INFO | train rank=0 | step=47030/50000 progress=94.1% loss=0.095779 lr=9.615426e-07 lr_backbone=9.615426e-07 elapsed=48:46:46 avg_step=3.73s recent_step=4.51s eta=03:04:49 +2026-09-16 14:33:45 | INFO | train rank=0 | step=47040/50000 progress=94.1% loss=0.108370 lr=9.550991e-07 lr_backbone=9.550991e-07 elapsed=48:47:17 avg_step=3.73s recent_step=3.09s eta=03:04:12 +2026-09-16 14:34:18 | INFO | train rank=0 | step=47050/50000 progress=94.1% loss=0.080518 lr=9.486771e-07 lr_backbone=9.486771e-07 elapsed=48:47:51 avg_step=3.73s recent_step=3.33s eta=03:03:34 +2026-09-16 14:34:56 | INFO | train rank=0 | step=47060/50000 progress=94.1% loss=0.108045 lr=9.422765e-07 lr_backbone=9.422765e-07 elapsed=48:48:29 avg_step=3.73s recent_step=3.81s eta=03:02:57 +2026-09-16 14:35:31 | INFO | train rank=0 | step=47070/50000 progress=94.1% loss=0.081246 lr=9.358974e-07 lr_backbone=9.358974e-07 elapsed=48:49:03 avg_step=3.73s recent_step=3.46s eta=03:02:19 +2026-09-16 14:36:05 | INFO | train rank=0 | step=47080/50000 progress=94.2% loss=0.126831 lr=9.295398e-07 lr_backbone=9.295398e-07 elapsed=48:49:37 avg_step=3.73s recent_step=3.40s eta=03:01:42 +2026-09-16 14:36:50 | INFO | train rank=0 | step=47090/50000 progress=94.2% loss=0.100577 lr=9.232036e-07 lr_backbone=9.232036e-07 elapsed=48:50:23 avg_step=3.73s recent_step=4.54s eta=03:01:05 +2026-09-16 14:37:20 | INFO | train rank=0 | step=47100/50000 progress=94.2% loss=0.091596 lr=9.168889e-07 lr_backbone=9.168889e-07 elapsed=48:50:53 avg_step=3.73s recent_step=3.00s eta=03:00:27 +2026-09-16 14:38:01 | INFO | train rank=0 | step=47110/50000 progress=94.2% loss=0.088831 lr=9.105957e-07 lr_backbone=9.105957e-07 elapsed=48:51:33 avg_step=3.73s recent_step=4.08s eta=02:59:50 +2026-09-16 14:38:27 | INFO | train rank=0 | step=47120/50000 progress=94.2% loss=0.090661 lr=9.043239e-07 lr_backbone=9.043239e-07 elapsed=48:52:00 avg_step=3.73s recent_step=2.63s eta=02:59:12 +2026-09-16 14:39:11 | INFO | train rank=0 | step=47130/50000 progress=94.3% loss=0.133664 lr=8.980736e-07 lr_backbone=8.980736e-07 elapsed=48:52:44 avg_step=3.73s recent_step=4.45s eta=02:58:35 +2026-09-16 14:39:43 | INFO | train rank=0 | step=47140/50000 progress=94.3% loss=0.082339 lr=8.918448e-07 lr_backbone=8.918448e-07 elapsed=48:53:16 avg_step=3.73s recent_step=3.16s eta=02:57:57 +2026-09-16 14:40:38 | INFO | train rank=0 | step=47150/50000 progress=94.3% loss=0.080180 lr=8.856375e-07 lr_backbone=8.856375e-07 elapsed=48:54:11 avg_step=3.73s recent_step=5.49s eta=02:57:21 +2026-09-16 14:41:10 | INFO | train rank=0 | step=47160/50000 progress=94.3% loss=0.086907 lr=8.794516e-07 lr_backbone=8.794516e-07 elapsed=48:54:42 avg_step=3.73s recent_step=3.16s eta=02:56:43 +2026-09-16 14:41:54 | INFO | train rank=0 | step=47170/50000 progress=94.3% loss=0.066839 lr=8.732873e-07 lr_backbone=8.732873e-07 elapsed=48:55:26 avg_step=3.73s recent_step=4.41s eta=02:56:06 +2026-09-16 14:42:24 | INFO | train rank=0 | step=47180/50000 progress=94.4% loss=0.113502 lr=8.671444e-07 lr_backbone=8.671444e-07 elapsed=48:55:57 avg_step=3.73s recent_step=3.05s eta=02:55:29 +2026-09-16 14:43:05 | INFO | train rank=0 | step=47190/50000 progress=94.4% loss=0.086457 lr=8.610230e-07 lr_backbone=8.610230e-07 elapsed=48:56:38 avg_step=3.73s recent_step=4.06s eta=02:54:51 +2026-09-16 14:43:42 | INFO | train rank=0 | step=47200/50000 progress=94.4% loss=0.104092 lr=8.549231e-07 lr_backbone=8.549231e-07 elapsed=48:57:15 avg_step=3.73s recent_step=3.77s eta=02:54:14 +2026-09-16 14:44:19 | INFO | train rank=0 | step=47210/50000 progress=94.4% loss=0.092051 lr=8.488448e-07 lr_backbone=8.488448e-07 elapsed=48:57:52 avg_step=3.73s recent_step=3.66s eta=02:53:37 +2026-09-16 14:44:59 | INFO | train rank=0 | step=47220/50000 progress=94.4% loss=0.100634 lr=8.427879e-07 lr_backbone=8.427879e-07 elapsed=48:58:32 avg_step=3.73s recent_step=4.00s eta=02:53:00 +2026-09-16 14:45:41 | INFO | train rank=0 | step=47230/50000 progress=94.5% loss=0.098756 lr=8.367525e-07 lr_backbone=8.367525e-07 elapsed=48:59:14 avg_step=3.73s recent_step=4.18s eta=02:52:23 +2026-09-16 14:46:18 | INFO | train rank=0 | step=47240/50000 progress=94.5% loss=0.100738 lr=8.307386e-07 lr_backbone=8.307386e-07 elapsed=48:59:51 avg_step=3.73s recent_step=3.76s eta=02:51:45 +2026-09-16 14:46:57 | INFO | train rank=0 | step=47250/50000 progress=94.5% loss=0.091824 lr=8.247463e-07 lr_backbone=8.247463e-07 elapsed=49:00:30 avg_step=3.73s recent_step=3.91s eta=02:51:08 +2026-09-16 14:47:27 | INFO | train rank=0 | step=47260/50000 progress=94.5% loss=0.076488 lr=8.187754e-07 lr_backbone=8.187754e-07 elapsed=49:01:00 avg_step=3.73s recent_step=2.98s eta=02:50:30 +2026-09-16 14:48:22 | INFO | train rank=0 | step=47270/50000 progress=94.5% loss=0.090224 lr=8.128261e-07 lr_backbone=8.128261e-07 elapsed=49:01:55 avg_step=3.73s recent_step=5.48s eta=02:49:54 +2026-09-16 14:48:52 | INFO | train rank=0 | step=47280/50000 progress=94.6% loss=0.118998 lr=8.068982e-07 lr_backbone=8.068982e-07 elapsed=49:02:25 avg_step=3.73s recent_step=3.05s eta=02:49:16 +2026-09-16 14:49:27 | INFO | train rank=0 | step=47290/50000 progress=94.6% loss=0.111823 lr=8.009919e-07 lr_backbone=8.009919e-07 elapsed=49:03:00 avg_step=3.73s recent_step=3.51s eta=02:48:39 +2026-09-16 14:50:02 | INFO | train rank=0 | step=47300/50000 progress=94.6% loss=0.083842 lr=7.951071e-07 lr_backbone=7.951071e-07 elapsed=49:03:35 avg_step=3.73s recent_step=3.44s eta=02:48:01 +2026-09-16 14:50:49 | INFO | train rank=0 | step=47310/50000 progress=94.6% loss=0.083503 lr=7.892439e-07 lr_backbone=7.892439e-07 elapsed=49:04:21 avg_step=3.73s recent_step=4.68s eta=02:47:24 +2026-09-16 14:51:30 | INFO | train rank=0 | step=47320/50000 progress=94.6% loss=0.108044 lr=7.834022e-07 lr_backbone=7.834022e-07 elapsed=49:05:02 avg_step=3.73s recent_step=4.10s eta=02:46:47 +2026-09-16 14:51:54 | INFO | train rank=0 | step=47330/50000 progress=94.7% loss=0.066577 lr=7.775819e-07 lr_backbone=7.775819e-07 elapsed=49:05:26 avg_step=3.73s recent_step=2.38s eta=02:46:09 +2026-09-16 14:52:28 | INFO | train rank=0 | step=47340/50000 progress=94.7% loss=0.081140 lr=7.717833e-07 lr_backbone=7.717833e-07 elapsed=49:06:01 avg_step=3.73s recent_step=3.49s eta=02:45:32 +2026-09-16 14:53:17 | INFO | train rank=0 | step=47350/50000 progress=94.7% loss=0.094356 lr=7.660061e-07 lr_backbone=7.660061e-07 elapsed=49:06:50 avg_step=3.73s recent_step=4.84s eta=02:44:55 +2026-09-16 14:53:48 | INFO | train rank=0 | step=47360/50000 progress=94.7% loss=0.108639 lr=7.602505e-07 lr_backbone=7.602505e-07 elapsed=49:07:20 avg_step=3.73s recent_step=3.09s eta=02:44:17 +2026-09-16 14:54:29 | INFO | train rank=0 | step=47370/50000 progress=94.7% loss=0.092668 lr=7.545165e-07 lr_backbone=7.545165e-07 elapsed=49:08:01 avg_step=3.73s recent_step=4.09s eta=02:43:40 +2026-09-16 14:55:04 | INFO | train rank=0 | step=47380/50000 progress=94.8% loss=0.112889 lr=7.488039e-07 lr_backbone=7.488039e-07 elapsed=49:08:37 avg_step=3.73s recent_step=3.58s eta=02:43:03 +2026-09-16 14:55:48 | INFO | train rank=0 | step=47390/50000 progress=94.8% loss=0.063026 lr=7.431130e-07 lr_backbone=7.431130e-07 elapsed=49:09:20 avg_step=3.73s recent_step=4.32s eta=02:42:26 +2026-09-16 14:56:29 | INFO | train rank=0 | step=47400/50000 progress=94.8% loss=0.102661 lr=7.374435e-07 lr_backbone=7.374435e-07 elapsed=49:10:02 avg_step=3.73s recent_step=4.11s eta=02:41:48 +2026-09-16 14:57:01 | INFO | train rank=0 | step=47410/50000 progress=94.8% loss=0.092923 lr=7.317957e-07 lr_backbone=7.317957e-07 elapsed=49:10:34 avg_step=3.73s recent_step=3.22s eta=02:41:11 +2026-09-16 14:57:34 | INFO | train rank=0 | step=47420/50000 progress=94.8% loss=0.079728 lr=7.261693e-07 lr_backbone=7.261693e-07 elapsed=49:11:07 avg_step=3.73s recent_step=3.31s eta=02:40:33 +2026-09-16 14:58:29 | INFO | train rank=0 | step=47430/50000 progress=94.9% loss=0.102164 lr=7.205645e-07 lr_backbone=7.205645e-07 elapsed=49:12:02 avg_step=3.73s recent_step=5.50s eta=02:39:57 +2026-09-16 14:59:06 | INFO | train rank=0 | step=47440/50000 progress=94.9% loss=0.080662 lr=7.149813e-07 lr_backbone=7.149813e-07 elapsed=49:12:39 avg_step=3.73s recent_step=3.68s eta=02:39:20 +2026-09-16 14:59:40 | INFO | train rank=0 | step=47450/50000 progress=94.9% loss=0.112217 lr=7.094197e-07 lr_backbone=7.094197e-07 elapsed=49:13:12 avg_step=3.73s recent_step=3.39s eta=02:38:42 +2026-09-16 15:00:06 | INFO | train rank=0 | step=47460/50000 progress=94.9% loss=0.088785 lr=7.038796e-07 lr_backbone=7.038796e-07 elapsed=49:13:38 avg_step=3.73s recent_step=2.58s eta=02:38:04 +2026-09-16 15:00:58 | INFO | train rank=0 | step=47470/50000 progress=94.9% loss=0.085342 lr=6.983610e-07 lr_backbone=6.983610e-07 elapsed=49:14:31 avg_step=3.73s recent_step=5.22s eta=02:37:27 +2026-09-16 15:01:30 | INFO | train rank=0 | step=47480/50000 progress=95.0% loss=0.089690 lr=6.928641e-07 lr_backbone=6.928641e-07 elapsed=49:15:02 avg_step=3.73s recent_step=3.19s eta=02:36:50 +2026-09-16 15:02:02 | INFO | train rank=0 | step=47490/50000 progress=95.0% loss=0.082851 lr=6.873887e-07 lr_backbone=6.873887e-07 elapsed=49:15:35 avg_step=3.73s recent_step=3.25s eta=02:36:12 +2026-09-16 15:02:36 | INFO | train rank=0 | step=47500/50000 progress=95.0% loss=0.080840 lr=6.819348e-07 lr_backbone=6.819348e-07 elapsed=49:16:09 avg_step=3.73s recent_step=3.35s eta=02:35:35 +2026-09-16 15:03:21 | INFO | train rank=0 | step=47510/50000 progress=95.0% loss=0.083397 lr=6.765026e-07 lr_backbone=6.765026e-07 elapsed=49:16:54 avg_step=3.73s recent_step=4.52s eta=02:34:58 +2026-09-16 15:03:50 | INFO | train rank=0 | step=47520/50000 progress=95.0% loss=0.082885 lr=6.710919e-07 lr_backbone=6.710919e-07 elapsed=49:17:23 avg_step=3.73s recent_step=2.91s eta=02:34:20 +2026-09-16 15:04:32 | INFO | train rank=0 | step=47530/50000 progress=95.1% loss=0.067649 lr=6.657028e-07 lr_backbone=6.657028e-07 elapsed=49:18:05 avg_step=3.73s recent_step=4.21s eta=02:33:43 +2026-09-16 15:05:08 | INFO | train rank=0 | step=47540/50000 progress=95.1% loss=0.088736 lr=6.603353e-07 lr_backbone=6.603353e-07 elapsed=49:18:41 avg_step=3.73s recent_step=3.59s eta=02:33:05 +2026-09-16 15:05:49 | INFO | train rank=0 | step=47550/50000 progress=95.1% loss=0.056395 lr=6.549893e-07 lr_backbone=6.549893e-07 elapsed=49:19:22 avg_step=3.73s recent_step=4.13s eta=02:32:28 +2026-09-16 15:06:23 | INFO | train rank=0 | step=47560/50000 progress=95.1% loss=0.084948 lr=6.496650e-07 lr_backbone=6.496650e-07 elapsed=49:19:56 avg_step=3.73s recent_step=3.35s eta=02:31:51 +2026-09-16 15:07:06 | INFO | train rank=0 | step=47570/50000 progress=95.1% loss=0.095324 lr=6.443622e-07 lr_backbone=6.443622e-07 elapsed=49:20:39 avg_step=3.73s recent_step=4.31s eta=02:31:14 +2026-09-16 15:07:34 | INFO | train rank=0 | step=47580/50000 progress=95.2% loss=0.081503 lr=6.390810e-07 lr_backbone=6.390810e-07 elapsed=49:21:07 avg_step=3.73s recent_step=2.85s eta=02:30:36 +2026-09-16 15:08:08 | INFO | train rank=0 | step=47590/50000 progress=95.2% loss=0.088361 lr=6.338214e-07 lr_backbone=6.338214e-07 elapsed=49:21:41 avg_step=3.73s recent_step=3.34s eta=02:29:58 +2026-09-16 15:08:37 | INFO | train rank=0 | step=47600/50000 progress=95.2% loss=0.095681 lr=6.285835e-07 lr_backbone=6.285835e-07 elapsed=49:22:10 avg_step=3.73s recent_step=2.96s eta=02:29:21 +2026-09-16 15:09:30 | INFO | train rank=0 | step=47610/50000 progress=95.2% loss=0.093670 lr=6.233671e-07 lr_backbone=6.233671e-07 elapsed=49:23:03 avg_step=3.73s recent_step=5.25s eta=02:28:44 +2026-09-16 15:10:04 | INFO | train rank=0 | step=47620/50000 progress=95.2% loss=0.099170 lr=6.181723e-07 lr_backbone=6.181723e-07 elapsed=49:23:37 avg_step=3.73s recent_step=3.40s eta=02:28:07 +2026-09-16 15:10:47 | INFO | train rank=0 | step=47630/50000 progress=95.3% loss=0.061532 lr=6.129991e-07 lr_backbone=6.129991e-07 elapsed=49:24:20 avg_step=3.73s recent_step=4.34s eta=02:27:30 +2026-09-16 15:11:15 | INFO | train rank=0 | step=47640/50000 progress=95.3% loss=0.086178 lr=6.078475e-07 lr_backbone=6.078475e-07 elapsed=49:24:48 avg_step=3.73s recent_step=2.81s eta=02:26:52 +2026-09-16 15:11:57 | INFO | train rank=0 | step=47650/50000 progress=95.3% loss=0.088918 lr=6.027175e-07 lr_backbone=6.027175e-07 elapsed=49:25:30 avg_step=3.73s recent_step=4.14s eta=02:26:15 +2026-09-16 15:12:28 | INFO | train rank=0 | step=47660/50000 progress=95.3% loss=0.072163 lr=5.976091e-07 lr_backbone=5.976091e-07 elapsed=49:26:00 avg_step=3.73s recent_step=3.09s eta=02:25:37 +2026-09-16 15:12:58 | INFO | train rank=0 | step=47670/50000 progress=95.3% loss=0.078731 lr=5.925224e-07 lr_backbone=5.925224e-07 elapsed=49:26:31 avg_step=3.73s recent_step=3.03s eta=02:24:59 +2026-09-16 15:13:46 | INFO | train rank=0 | step=47680/50000 progress=95.4% loss=0.074937 lr=5.874572e-07 lr_backbone=5.874572e-07 elapsed=49:27:19 avg_step=3.73s recent_step=4.78s eta=02:24:22 +2026-09-16 15:14:23 | INFO | train rank=0 | step=47690/50000 progress=95.4% loss=0.073019 lr=5.824137e-07 lr_backbone=5.824137e-07 elapsed=49:27:56 avg_step=3.73s recent_step=3.72s eta=02:23:45 +2026-09-16 15:14:52 | INFO | train rank=0 | step=47700/50000 progress=95.4% loss=0.072683 lr=5.773917e-07 lr_backbone=5.773917e-07 elapsed=49:28:24 avg_step=3.73s recent_step=2.86s eta=02:23:07 +2026-09-16 15:15:37 | INFO | train rank=0 | step=47710/50000 progress=95.4% loss=0.078789 lr=5.723914e-07 lr_backbone=5.723914e-07 elapsed=49:29:10 avg_step=3.73s recent_step=4.54s eta=02:22:30 +2026-09-16 15:16:29 | INFO | train rank=0 | step=47720/50000 progress=95.4% loss=0.079170 lr=5.674128e-07 lr_backbone=5.674128e-07 elapsed=49:30:02 avg_step=3.73s recent_step=5.18s eta=02:21:54 +2026-09-16 15:17:03 | INFO | train rank=0 | step=47730/50000 progress=95.5% loss=0.077824 lr=5.624557e-07 lr_backbone=5.624557e-07 elapsed=49:30:35 avg_step=3.73s recent_step=3.38s eta=02:21:16 +2026-09-16 15:17:36 | INFO | train rank=0 | step=47740/50000 progress=95.5% loss=0.106339 lr=5.575203e-07 lr_backbone=5.575203e-07 elapsed=49:31:09 avg_step=3.73s recent_step=3.36s eta=02:20:39 +2026-09-16 15:18:14 | INFO | train rank=0 | step=47750/50000 progress=95.5% loss=0.083757 lr=5.526065e-07 lr_backbone=5.526065e-07 elapsed=49:31:47 avg_step=3.73s recent_step=3.79s eta=02:20:01 +2026-09-16 15:18:56 | INFO | train rank=0 | step=47760/50000 progress=95.5% loss=0.082479 lr=5.477143e-07 lr_backbone=5.477143e-07 elapsed=49:32:29 avg_step=3.73s recent_step=4.19s eta=02:19:24 +2026-09-16 15:19:27 | INFO | train rank=0 | step=47770/50000 progress=95.5% loss=0.110576 lr=5.428438e-07 lr_backbone=5.428438e-07 elapsed=49:33:00 avg_step=3.73s recent_step=3.11s eta=02:18:47 +2026-09-16 15:20:03 | INFO | train rank=0 | step=47780/50000 progress=95.6% loss=0.079294 lr=5.379948e-07 lr_backbone=5.379948e-07 elapsed=49:33:36 avg_step=3.73s recent_step=3.61s eta=02:18:09 +2026-09-16 15:20:53 | INFO | train rank=0 | step=47790/50000 progress=95.6% loss=0.100023 lr=5.331676e-07 lr_backbone=5.331676e-07 elapsed=49:34:26 avg_step=3.73s recent_step=4.99s eta=02:17:32 +2026-09-16 15:21:27 | INFO | train rank=0 | step=47800/50000 progress=95.6% loss=0.088362 lr=5.283619e-07 lr_backbone=5.283619e-07 elapsed=49:35:00 avg_step=3.73s recent_step=3.42s eta=02:16:55 +2026-09-16 15:22:09 | INFO | train rank=0 | step=47810/50000 progress=95.6% loss=0.096056 lr=5.235780e-07 lr_backbone=5.235780e-07 elapsed=49:35:41 avg_step=3.73s recent_step=4.12s eta=02:16:18 +2026-09-16 15:22:40 | INFO | train rank=0 | step=47820/50000 progress=95.6% loss=0.076569 lr=5.188156e-07 lr_backbone=5.188156e-07 elapsed=49:36:13 avg_step=3.73s recent_step=3.12s eta=02:15:40 +2026-09-16 15:23:33 | INFO | train rank=0 | step=47830/50000 progress=95.7% loss=0.097327 lr=5.140749e-07 lr_backbone=5.140749e-07 elapsed=49:37:06 avg_step=3.73s recent_step=5.32s eta=02:15:04 +2026-09-16 15:24:04 | INFO | train rank=0 | step=47840/50000 progress=95.7% loss=0.059131 lr=5.093558e-07 lr_backbone=5.093558e-07 elapsed=49:37:37 avg_step=3.73s recent_step=3.08s eta=02:14:26 +2026-09-16 15:24:36 | INFO | train rank=0 | step=47850/50000 progress=95.7% loss=0.084272 lr=5.046584e-07 lr_backbone=5.046584e-07 elapsed=49:38:09 avg_step=3.73s recent_step=3.21s eta=02:13:48 +2026-09-16 15:25:13 | INFO | train rank=0 | step=47860/50000 progress=95.7% loss=0.105120 lr=4.999827e-07 lr_backbone=4.999827e-07 elapsed=49:38:46 avg_step=3.73s recent_step=3.71s eta=02:13:11 +2026-09-16 15:26:01 | INFO | train rank=0 | step=47870/50000 progress=95.7% loss=0.093085 lr=4.953286e-07 lr_backbone=4.953286e-07 elapsed=49:39:34 avg_step=3.73s recent_step=4.80s eta=02:12:34 +2026-09-16 15:26:42 | INFO | train rank=0 | step=47880/50000 progress=95.8% loss=0.138945 lr=4.906961e-07 lr_backbone=4.906961e-07 elapsed=49:40:15 avg_step=3.73s recent_step=4.10s eta=02:11:57 +2026-09-16 15:27:20 | INFO | train rank=0 | step=47890/50000 progress=95.8% loss=0.098183 lr=4.860853e-07 lr_backbone=4.860853e-07 elapsed=49:40:52 avg_step=3.73s recent_step=3.76s eta=02:11:20 +2026-09-16 15:27:52 | INFO | train rank=0 | step=47900/50000 progress=95.8% loss=0.076171 lr=4.814962e-07 lr_backbone=4.814962e-07 elapsed=49:41:25 avg_step=3.73s recent_step=3.26s eta=02:10:42 +2026-09-16 15:28:36 | INFO | train rank=0 | step=47910/50000 progress=95.8% loss=0.095928 lr=4.769287e-07 lr_backbone=4.769287e-07 elapsed=49:42:09 avg_step=3.73s recent_step=4.38s eta=02:10:05 +2026-09-16 15:29:16 | INFO | train rank=0 | step=47920/50000 progress=95.8% loss=0.072844 lr=4.723829e-07 lr_backbone=4.723829e-07 elapsed=49:42:48 avg_step=3.73s recent_step=3.95s eta=02:09:28 +2026-09-16 15:29:45 | INFO | train rank=0 | step=47930/50000 progress=95.9% loss=0.088230 lr=4.678588e-07 lr_backbone=4.678588e-07 elapsed=49:43:18 avg_step=3.73s recent_step=2.93s eta=02:08:50 +2026-09-16 15:30:22 | INFO | train rank=0 | step=47940/50000 progress=95.9% loss=0.115306 lr=4.633563e-07 lr_backbone=4.633563e-07 elapsed=49:43:55 avg_step=3.73s recent_step=3.69s eta=02:08:13 +2026-09-16 15:30:58 | INFO | train rank=0 | step=47950/50000 progress=95.9% loss=0.094072 lr=4.588755e-07 lr_backbone=4.588755e-07 elapsed=49:44:31 avg_step=3.73s recent_step=3.62s eta=02:07:35 +2026-09-16 15:31:41 | INFO | train rank=0 | step=47960/50000 progress=95.9% loss=0.063822 lr=4.544163e-07 lr_backbone=4.544163e-07 elapsed=49:45:14 avg_step=3.73s recent_step=4.32s eta=02:06:58 +2026-09-16 15:32:13 | INFO | train rank=0 | step=47970/50000 progress=95.9% loss=0.090403 lr=4.499789e-07 lr_backbone=4.499789e-07 elapsed=49:45:46 avg_step=3.73s recent_step=3.21s eta=02:06:21 +2026-09-16 15:32:45 | INFO | train rank=0 | step=47980/50000 progress=96.0% loss=0.098776 lr=4.455631e-07 lr_backbone=4.455631e-07 elapsed=49:46:18 avg_step=3.73s recent_step=3.21s eta=02:05:43 +2026-09-16 15:33:21 | INFO | train rank=0 | step=47990/50000 progress=96.0% loss=0.087485 lr=4.411690e-07 lr_backbone=4.411690e-07 elapsed=49:46:53 avg_step=3.73s recent_step=3.54s eta=02:05:06 +2026-09-16 15:34:19 | INFO | train rank=0 | step=48000/50000 progress=96.0% loss=0.062650 lr=4.367965e-07 lr_backbone=4.367965e-07 elapsed=49:47:52 avg_step=3.73s recent_step=5.86s eta=02:04:29 +2026-09-16 15:34:49 | INFO | train rank=0 | step=48010/50000 progress=96.0% loss=0.075907 lr=4.324458e-07 lr_backbone=4.324458e-07 elapsed=49:48:22 avg_step=3.73s recent_step=2.98s eta=02:03:52 +2026-09-16 15:35:15 | INFO | train rank=0 | step=48020/50000 progress=96.0% loss=0.060770 lr=4.281167e-07 lr_backbone=4.281167e-07 elapsed=49:48:47 avg_step=3.73s recent_step=2.56s eta=02:03:14 +2026-09-16 15:35:57 | INFO | train rank=0 | step=48030/50000 progress=96.1% loss=0.109677 lr=4.238093e-07 lr_backbone=4.238093e-07 elapsed=49:49:29 avg_step=3.73s recent_step=4.18s eta=02:02:37 +2026-09-16 15:36:44 | INFO | train rank=0 | step=48040/50000 progress=96.1% loss=0.095099 lr=4.195236e-07 lr_backbone=4.195236e-07 elapsed=49:50:17 avg_step=3.73s recent_step=4.72s eta=02:02:00 +2026-09-16 15:37:15 | INFO | train rank=0 | step=48050/50000 progress=96.1% loss=0.060493 lr=4.152596e-07 lr_backbone=4.152596e-07 elapsed=49:50:48 avg_step=3.73s recent_step=3.17s eta=02:01:22 +2026-09-16 15:37:38 | INFO | train rank=0 | step=48060/50000 progress=96.1% loss=0.075494 lr=4.110173e-07 lr_backbone=4.110173e-07 elapsed=49:51:11 avg_step=3.73s recent_step=2.27s eta=02:00:44 +2026-09-16 15:38:09 | INFO | train rank=0 | step=48070/50000 progress=96.1% loss=0.068745 lr=4.067967e-07 lr_backbone=4.067967e-07 elapsed=49:51:42 avg_step=3.73s recent_step=3.11s eta=02:00:06 +2026-09-16 15:39:06 | INFO | train rank=0 | step=48080/50000 progress=96.2% loss=0.116395 lr=4.025977e-07 lr_backbone=4.025977e-07 elapsed=49:52:38 avg_step=3.73s recent_step=5.65s eta=01:59:30 +2026-09-16 15:39:43 | INFO | train rank=0 | step=48090/50000 progress=96.2% loss=0.086171 lr=3.984205e-07 lr_backbone=3.984205e-07 elapsed=49:53:16 avg_step=3.73s recent_step=3.75s eta=01:58:53 +2026-09-16 15:40:13 | INFO | train rank=0 | step=48100/50000 progress=96.2% loss=0.091188 lr=3.942649e-07 lr_backbone=3.942649e-07 elapsed=49:53:46 avg_step=3.73s recent_step=3.02s eta=01:58:15 +2026-09-16 15:40:44 | INFO | train rank=0 | step=48110/50000 progress=96.2% loss=0.088493 lr=3.901311e-07 lr_backbone=3.901311e-07 elapsed=49:54:17 avg_step=3.73s recent_step=3.07s eta=01:57:37 +2026-09-16 15:41:36 | INFO | train rank=0 | step=48120/50000 progress=96.2% loss=0.107973 lr=3.860189e-07 lr_backbone=3.860189e-07 elapsed=49:55:09 avg_step=3.73s recent_step=5.17s eta=01:57:01 +2026-09-16 15:42:05 | INFO | train rank=0 | step=48130/50000 progress=96.3% loss=0.078996 lr=3.819285e-07 lr_backbone=3.819285e-07 elapsed=49:55:38 avg_step=3.73s recent_step=2.94s eta=01:56:23 +2026-09-16 15:42:40 | INFO | train rank=0 | step=48140/50000 progress=96.3% loss=0.056971 lr=3.778598e-07 lr_backbone=3.778598e-07 elapsed=49:56:13 avg_step=3.73s recent_step=3.46s eta=01:55:45 +2026-09-16 15:43:10 | INFO | train rank=0 | step=48150/50000 progress=96.3% loss=0.090072 lr=3.738127e-07 lr_backbone=3.738127e-07 elapsed=49:56:43 avg_step=3.73s recent_step=3.00s eta=01:55:08 +2026-09-16 15:43:59 | INFO | train rank=0 | step=48160/50000 progress=96.3% loss=0.133883 lr=3.697874e-07 lr_backbone=3.697874e-07 elapsed=49:57:32 avg_step=3.73s recent_step=4.93s eta=01:54:31 +2026-09-16 15:44:31 | INFO | train rank=0 | step=48170/50000 progress=96.3% loss=0.077191 lr=3.657838e-07 lr_backbone=3.657838e-07 elapsed=49:58:04 avg_step=3.73s recent_step=3.23s eta=01:53:53 +2026-09-16 15:44:56 | INFO | train rank=0 | step=48180/50000 progress=96.4% loss=0.085580 lr=3.618019e-07 lr_backbone=3.618019e-07 elapsed=49:58:29 avg_step=3.73s recent_step=2.50s eta=01:53:16 +2026-09-16 15:45:29 | INFO | train rank=0 | step=48190/50000 progress=96.4% loss=0.069540 lr=3.578417e-07 lr_backbone=3.578417e-07 elapsed=49:59:02 avg_step=3.73s recent_step=3.25s eta=01:52:38 +2026-09-16 15:46:09 | INFO | train rank=0 | step=48200/50000 progress=96.4% loss=0.078465 lr=3.539033e-07 lr_backbone=3.539033e-07 elapsed=49:59:42 avg_step=3.73s recent_step=4.03s eta=01:52:01 +2026-09-16 15:46:43 | INFO | train rank=0 | step=48210/50000 progress=96.4% loss=0.091799 lr=3.499865e-07 lr_backbone=3.499865e-07 elapsed=50:00:16 avg_step=3.73s recent_step=3.39s eta=01:51:23 +2026-09-16 15:47:18 | INFO | train rank=0 | step=48220/50000 progress=96.4% loss=0.109552 lr=3.460915e-07 lr_backbone=3.460915e-07 elapsed=50:00:51 avg_step=3.73s recent_step=3.47s eta=01:50:46 +2026-09-16 15:47:55 | INFO | train rank=0 | step=48230/50000 progress=96.5% loss=0.073677 lr=3.422181e-07 lr_backbone=3.422181e-07 elapsed=50:01:28 avg_step=3.73s recent_step=3.72s eta=01:50:09 +2026-09-16 15:48:28 | INFO | train rank=0 | step=48240/50000 progress=96.5% loss=0.132869 lr=3.383666e-07 lr_backbone=3.383666e-07 elapsed=50:02:01 avg_step=3.73s recent_step=3.29s eta=01:49:31 +2026-09-16 15:49:04 | INFO | train rank=0 | step=48250/50000 progress=96.5% loss=0.120057 lr=3.345367e-07 lr_backbone=3.345367e-07 elapsed=50:02:37 avg_step=3.73s recent_step=3.62s eta=01:48:54 +2026-09-16 15:49:41 | INFO | train rank=0 | step=48260/50000 progress=96.5% loss=0.109149 lr=3.307285e-07 lr_backbone=3.307285e-07 elapsed=50:03:14 avg_step=3.73s recent_step=3.74s eta=01:48:16 +2026-09-16 15:50:17 | INFO | train rank=0 | step=48270/50000 progress=96.5% loss=0.074424 lr=3.269421e-07 lr_backbone=3.269421e-07 elapsed=50:03:50 avg_step=3.73s recent_step=3.55s eta=01:47:39 +2026-09-16 15:50:57 | INFO | train rank=0 | step=48280/50000 progress=96.6% loss=0.101267 lr=3.231774e-07 lr_backbone=3.231774e-07 elapsed=50:04:30 avg_step=3.73s recent_step=3.99s eta=01:47:02 +2026-09-16 15:52:04 | INFO | train rank=0 | step=48290/50000 progress=96.6% loss=0.091274 lr=3.194345e-07 lr_backbone=3.194345e-07 elapsed=50:05:37 avg_step=3.73s recent_step=6.72s eta=01:46:25 +2026-09-16 15:52:33 | INFO | train rank=0 | step=48300/50000 progress=96.6% loss=0.122670 lr=3.157132e-07 lr_backbone=3.157132e-07 elapsed=50:06:05 avg_step=3.73s recent_step=2.85s eta=01:45:48 +2026-09-16 15:53:00 | INFO | train rank=0 | step=48310/50000 progress=96.6% loss=0.083545 lr=3.120138e-07 lr_backbone=3.120138e-07 elapsed=50:06:33 avg_step=3.73s recent_step=2.75s eta=01:45:10 +2026-09-16 15:53:28 | INFO | train rank=0 | step=48320/50000 progress=96.6% loss=0.082612 lr=3.083360e-07 lr_backbone=3.083360e-07 elapsed=50:07:00 avg_step=3.73s recent_step=2.76s eta=01:44:32 +2026-09-16 15:54:15 | INFO | train rank=0 | step=48330/50000 progress=96.7% loss=0.066447 lr=3.046800e-07 lr_backbone=3.046800e-07 elapsed=50:07:48 avg_step=3.73s recent_step=4.71s eta=01:43:55 +2026-09-16 15:54:52 | INFO | train rank=0 | step=48340/50000 progress=96.7% loss=0.067306 lr=3.010457e-07 lr_backbone=3.010457e-07 elapsed=50:08:25 avg_step=3.73s recent_step=3.72s eta=01:43:18 +2026-09-16 15:55:30 | INFO | train rank=0 | step=48350/50000 progress=96.7% loss=0.091813 lr=2.974332e-07 lr_backbone=2.974332e-07 elapsed=50:09:03 avg_step=3.73s recent_step=3.84s eta=01:42:41 +2026-09-16 15:56:04 | INFO | train rank=0 | step=48360/50000 progress=96.7% loss=0.084628 lr=2.938424e-07 lr_backbone=2.938424e-07 elapsed=50:09:37 avg_step=3.73s recent_step=3.36s eta=01:42:03 +2026-09-16 15:56:42 | INFO | train rank=0 | step=48370/50000 progress=96.7% loss=0.091891 lr=2.902733e-07 lr_backbone=2.902733e-07 elapsed=50:10:15 avg_step=3.73s recent_step=3.81s eta=01:41:26 +2026-09-16 15:57:15 | INFO | train rank=0 | step=48380/50000 progress=96.8% loss=0.083564 lr=2.867260e-07 lr_backbone=2.867260e-07 elapsed=50:10:48 avg_step=3.73s recent_step=3.30s eta=01:40:48 +2026-09-16 15:57:59 | INFO | train rank=0 | step=48390/50000 progress=96.8% loss=0.070951 lr=2.832004e-07 lr_backbone=2.832004e-07 elapsed=50:11:31 avg_step=3.73s recent_step=4.35s eta=01:40:11 +2026-09-16 15:58:27 | INFO | train rank=0 | step=48400/50000 progress=96.8% loss=0.074006 lr=2.796966e-07 lr_backbone=2.796966e-07 elapsed=50:12:00 avg_step=3.73s recent_step=2.89s eta=01:39:34 +2026-09-16 15:59:17 | INFO | train rank=0 | step=48410/50000 progress=96.8% loss=0.095837 lr=2.762145e-07 lr_backbone=2.762145e-07 elapsed=50:12:50 avg_step=3.73s recent_step=4.98s eta=01:38:57 +2026-09-16 15:59:51 | INFO | train rank=0 | step=48420/50000 progress=96.8% loss=0.100601 lr=2.727542e-07 lr_backbone=2.727542e-07 elapsed=50:13:24 avg_step=3.73s recent_step=3.35s eta=01:38:19 +2026-09-16 16:00:25 | INFO | train rank=0 | step=48430/50000 progress=96.9% loss=0.072338 lr=2.693156e-07 lr_backbone=2.693156e-07 elapsed=50:13:58 avg_step=3.73s recent_step=3.43s eta=01:37:42 +2026-09-16 16:01:06 | INFO | train rank=0 | step=48440/50000 progress=96.9% loss=0.091160 lr=2.658988e-07 lr_backbone=2.658988e-07 elapsed=50:14:39 avg_step=3.73s recent_step=4.12s eta=01:37:05 +2026-09-16 16:01:47 | INFO | train rank=0 | step=48450/50000 progress=96.9% loss=0.070872 lr=2.625038e-07 lr_backbone=2.625038e-07 elapsed=50:15:20 avg_step=3.73s recent_step=4.09s eta=01:36:27 +2026-09-16 16:02:25 | INFO | train rank=0 | step=48460/50000 progress=96.9% loss=0.078213 lr=2.591305e-07 lr_backbone=2.591305e-07 elapsed=50:15:57 avg_step=3.73s recent_step=3.74s eta=01:35:50 +2026-09-16 16:03:05 | INFO | train rank=0 | step=48470/50000 progress=96.9% loss=0.082318 lr=2.557789e-07 lr_backbone=2.557789e-07 elapsed=50:16:37 avg_step=3.73s recent_step=4.00s eta=01:35:13 +2026-09-16 16:03:46 | INFO | train rank=0 | step=48480/50000 progress=97.0% loss=0.065912 lr=2.524492e-07 lr_backbone=2.524492e-07 elapsed=50:17:19 avg_step=3.73s recent_step=4.14s eta=01:34:36 +2026-09-16 16:04:21 | INFO | train rank=0 | step=48490/50000 progress=97.0% loss=0.116159 lr=2.491411e-07 lr_backbone=2.491411e-07 elapsed=50:17:54 avg_step=3.73s recent_step=3.49s eta=01:33:58 +2026-09-16 16:04:50 | INFO | train rank=0 | step=48500/50000 progress=97.0% loss=0.098200 lr=2.458549e-07 lr_backbone=2.458549e-07 elapsed=50:18:23 avg_step=3.73s recent_step=2.89s eta=01:33:21 +2026-09-16 16:05:20 | INFO | train rank=0 | step=48510/50000 progress=97.0% loss=0.107446 lr=2.425904e-07 lr_backbone=2.425904e-07 elapsed=50:18:53 avg_step=3.73s recent_step=3.02s eta=01:32:43 +2026-09-16 16:06:08 | INFO | train rank=0 | step=48520/50000 progress=97.0% loss=0.120545 lr=2.393477e-07 lr_backbone=2.393477e-07 elapsed=50:19:41 avg_step=3.73s recent_step=4.80s eta=01:32:06 +2026-09-16 16:06:44 | INFO | train rank=0 | step=48530/50000 progress=97.1% loss=0.070030 lr=2.361267e-07 lr_backbone=2.361267e-07 elapsed=50:20:17 avg_step=3.73s recent_step=3.60s eta=01:31:29 +2026-09-16 16:07:20 | INFO | train rank=0 | step=48540/50000 progress=97.1% loss=0.093744 lr=2.329275e-07 lr_backbone=2.329275e-07 elapsed=50:20:52 avg_step=3.73s recent_step=3.57s eta=01:30:51 +2026-09-16 16:07:56 | INFO | train rank=0 | step=48550/50000 progress=97.1% loss=0.068090 lr=2.297501e-07 lr_backbone=2.297501e-07 elapsed=50:21:29 avg_step=3.73s recent_step=3.66s eta=01:30:14 +2026-09-16 16:08:33 | INFO | train rank=0 | step=48560/50000 progress=97.1% loss=0.088914 lr=2.265944e-07 lr_backbone=2.265944e-07 elapsed=50:22:05 avg_step=3.73s recent_step=3.64s eta=01:29:37 +2026-09-16 16:09:09 | INFO | train rank=0 | step=48570/50000 progress=97.1% loss=0.097336 lr=2.234605e-07 lr_backbone=2.234605e-07 elapsed=50:22:42 avg_step=3.73s recent_step=3.66s eta=01:28:59 +2026-09-16 16:09:45 | INFO | train rank=0 | step=48580/50000 progress=97.2% loss=0.098503 lr=2.203484e-07 lr_backbone=2.203484e-07 elapsed=50:23:18 avg_step=3.73s recent_step=3.55s eta=01:28:22 +2026-09-16 16:10:22 | INFO | train rank=0 | step=48590/50000 progress=97.2% loss=0.109680 lr=2.172581e-07 lr_backbone=2.172581e-07 elapsed=50:23:55 avg_step=3.73s recent_step=3.72s eta=01:27:44 +2026-09-16 16:11:21 | INFO | train rank=0 | step=48600/50000 progress=97.2% loss=0.124482 lr=2.141896e-07 lr_backbone=2.141896e-07 elapsed=50:24:54 avg_step=3.73s recent_step=5.94s eta=01:27:08 +2026-09-16 16:11:54 | INFO | train rank=0 | step=48610/50000 progress=97.2% loss=0.074930 lr=2.111428e-07 lr_backbone=2.111428e-07 elapsed=50:25:27 avg_step=3.73s recent_step=3.30s eta=01:26:30 +2026-09-16 16:12:28 | INFO | train rank=0 | step=48620/50000 progress=97.2% loss=0.059399 lr=2.081178e-07 lr_backbone=2.081178e-07 elapsed=50:26:01 avg_step=3.73s recent_step=3.41s eta=01:25:53 +2026-09-16 16:12:59 | INFO | train rank=0 | step=48630/50000 progress=97.3% loss=0.084302 lr=2.051146e-07 lr_backbone=2.051146e-07 elapsed=50:26:32 avg_step=3.73s recent_step=3.04s eta=01:25:15 +2026-09-16 16:13:41 | INFO | train rank=0 | step=48640/50000 progress=97.3% loss=0.104277 lr=2.021331e-07 lr_backbone=2.021331e-07 elapsed=50:27:14 avg_step=3.73s recent_step=4.19s eta=01:24:38 +2026-09-16 16:14:27 | INFO | train rank=0 | step=48650/50000 progress=97.3% loss=0.080804 lr=1.991735e-07 lr_backbone=1.991735e-07 elapsed=50:27:59 avg_step=3.73s recent_step=4.58s eta=01:24:01 +2026-09-16 16:14:56 | INFO | train rank=0 | step=48660/50000 progress=97.3% loss=0.077105 lr=1.962356e-07 lr_backbone=1.962356e-07 elapsed=50:28:29 avg_step=3.73s recent_step=2.93s eta=01:23:23 +2026-09-16 16:15:39 | INFO | train rank=0 | step=48670/50000 progress=97.3% loss=0.072394 lr=1.933195e-07 lr_backbone=1.933195e-07 elapsed=50:29:12 avg_step=3.73s recent_step=4.35s eta=01:22:46 +2026-09-16 16:16:13 | INFO | train rank=0 | step=48680/50000 progress=97.4% loss=0.109412 lr=1.904253e-07 lr_backbone=1.904253e-07 elapsed=50:29:46 avg_step=3.73s recent_step=3.41s eta=01:22:09 +2026-09-16 16:17:06 | INFO | train rank=0 | step=48690/50000 progress=97.4% loss=0.105150 lr=1.875527e-07 lr_backbone=1.875527e-07 elapsed=50:30:39 avg_step=3.73s recent_step=5.25s eta=01:21:32 +2026-09-16 16:17:36 | INFO | train rank=0 | step=48700/50000 progress=97.4% loss=0.073325 lr=1.847020e-07 lr_backbone=1.847020e-07 elapsed=50:31:09 avg_step=3.73s recent_step=3.05s eta=01:20:54 +2026-09-16 16:18:07 | INFO | train rank=0 | step=48710/50000 progress=97.4% loss=0.076962 lr=1.818731e-07 lr_backbone=1.818731e-07 elapsed=50:31:40 avg_step=3.73s recent_step=3.06s eta=01:20:17 +2026-09-16 16:18:33 | INFO | train rank=0 | step=48720/50000 progress=97.4% loss=0.081073 lr=1.790660e-07 lr_backbone=1.790660e-07 elapsed=50:32:06 avg_step=3.73s recent_step=2.62s eta=01:19:39 +2026-09-16 16:19:15 | INFO | train rank=0 | step=48730/50000 progress=97.5% loss=0.167992 lr=1.762806e-07 lr_backbone=1.762806e-07 elapsed=50:32:47 avg_step=3.73s recent_step=4.14s eta=01:19:02 +2026-09-16 16:19:51 | INFO | train rank=0 | step=48740/50000 progress=97.5% loss=0.097297 lr=1.735171e-07 lr_backbone=1.735171e-07 elapsed=50:33:24 avg_step=3.73s recent_step=3.63s eta=01:18:25 +2026-09-16 16:20:30 | INFO | train rank=0 | step=48750/50000 progress=97.5% loss=0.103742 lr=1.707753e-07 lr_backbone=1.707753e-07 elapsed=50:34:03 avg_step=3.73s recent_step=3.92s eta=01:17:47 +2026-09-16 16:21:03 | INFO | train rank=0 | step=48760/50000 progress=97.5% loss=0.079180 lr=1.680554e-07 lr_backbone=1.680554e-07 elapsed=50:34:36 avg_step=3.73s recent_step=3.27s eta=01:17:10 +2026-09-16 16:21:49 | INFO | train rank=0 | step=48770/50000 progress=97.5% loss=0.082406 lr=1.653572e-07 lr_backbone=1.653572e-07 elapsed=50:35:22 avg_step=3.73s recent_step=4.61s eta=01:16:33 +2026-09-16 16:22:25 | INFO | train rank=0 | step=48780/50000 progress=97.6% loss=0.091981 lr=1.626809e-07 lr_backbone=1.626809e-07 elapsed=50:35:58 avg_step=3.73s recent_step=3.61s eta=01:15:55 +2026-09-16 16:22:51 | INFO | train rank=0 | step=48790/50000 progress=97.6% loss=0.154797 lr=1.600263e-07 lr_backbone=1.600263e-07 elapsed=50:36:24 avg_step=3.73s recent_step=2.60s eta=01:15:18 +2026-09-16 16:23:21 | INFO | train rank=0 | step=48800/50000 progress=97.6% loss=0.101241 lr=1.573936e-07 lr_backbone=1.573936e-07 elapsed=50:36:53 avg_step=3.73s recent_step=2.96s eta=01:14:40 +2026-09-16 16:24:16 | INFO | train rank=0 | step=48810/50000 progress=97.6% loss=0.071625 lr=1.547826e-07 lr_backbone=1.547826e-07 elapsed=50:37:49 avg_step=3.73s recent_step=5.53s eta=01:14:03 +2026-09-16 16:24:50 | INFO | train rank=0 | step=48820/50000 progress=97.6% loss=0.121460 lr=1.521935e-07 lr_backbone=1.521935e-07 elapsed=50:38:23 avg_step=3.73s recent_step=3.43s eta=01:13:26 +2026-09-16 16:25:24 | INFO | train rank=0 | step=48830/50000 progress=97.7% loss=0.093358 lr=1.496262e-07 lr_backbone=1.496262e-07 elapsed=50:38:57 avg_step=3.73s recent_step=3.40s eta=01:12:48 +2026-09-16 16:26:01 | INFO | train rank=0 | step=48840/50000 progress=97.7% loss=0.083037 lr=1.470806e-07 lr_backbone=1.470806e-07 elapsed=50:39:33 avg_step=3.73s recent_step=3.65s eta=01:12:11 +2026-09-16 16:26:38 | INFO | train rank=0 | step=48850/50000 progress=97.7% loss=0.079000 lr=1.445569e-07 lr_backbone=1.445569e-07 elapsed=50:40:10 avg_step=3.73s recent_step=3.70s eta=01:11:34 +2026-09-16 16:27:15 | INFO | train rank=0 | step=48860/50000 progress=97.7% loss=0.095248 lr=1.420550e-07 lr_backbone=1.420550e-07 elapsed=50:40:48 avg_step=3.73s recent_step=3.75s eta=01:10:56 +2026-09-16 16:27:56 | INFO | train rank=0 | step=48870/50000 progress=97.7% loss=0.079330 lr=1.395749e-07 lr_backbone=1.395749e-07 elapsed=50:41:28 avg_step=3.73s recent_step=4.03s eta=01:10:19 +2026-09-16 16:28:31 | INFO | train rank=0 | step=48880/50000 progress=97.8% loss=0.102243 lr=1.371166e-07 lr_backbone=1.371166e-07 elapsed=50:42:04 avg_step=3.73s recent_step=3.54s eta=01:09:42 +2026-09-16 16:29:22 | INFO | train rank=0 | step=48890/50000 progress=97.8% loss=0.104551 lr=1.346801e-07 lr_backbone=1.346801e-07 elapsed=50:42:55 avg_step=3.73s recent_step=5.09s eta=01:09:05 +2026-09-16 16:29:50 | INFO | train rank=0 | step=48900/50000 progress=97.8% loss=0.096049 lr=1.322654e-07 lr_backbone=1.322654e-07 elapsed=50:43:23 avg_step=3.73s recent_step=2.86s eta=01:08:27 +2026-09-16 16:30:28 | INFO | train rank=0 | step=48910/50000 progress=97.8% loss=0.093856 lr=1.298726e-07 lr_backbone=1.298726e-07 elapsed=50:44:01 avg_step=3.73s recent_step=3.73s eta=01:07:50 +2026-09-16 16:31:02 | INFO | train rank=0 | step=48920/50000 progress=97.8% loss=0.097378 lr=1.275015e-07 lr_backbone=1.275015e-07 elapsed=50:44:34 avg_step=3.73s recent_step=3.39s eta=01:07:12 +2026-09-16 16:31:42 | INFO | train rank=0 | step=48930/50000 progress=97.9% loss=0.096010 lr=1.251523e-07 lr_backbone=1.251523e-07 elapsed=50:45:15 avg_step=3.73s recent_step=4.05s eta=01:06:35 +2026-09-16 16:32:11 | INFO | train rank=0 | step=48940/50000 progress=97.9% loss=0.095177 lr=1.228249e-07 lr_backbone=1.228249e-07 elapsed=50:45:43 avg_step=3.73s recent_step=2.83s eta=01:05:58 +2026-09-16 16:32:59 | INFO | train rank=0 | step=48950/50000 progress=97.9% loss=0.098451 lr=1.205193e-07 lr_backbone=1.205193e-07 elapsed=50:46:32 avg_step=3.73s recent_step=4.85s eta=01:05:20 +2026-09-16 16:33:37 | INFO | train rank=0 | step=48960/50000 progress=97.9% loss=0.084899 lr=1.182355e-07 lr_backbone=1.182355e-07 elapsed=50:47:09 avg_step=3.73s recent_step=3.75s eta=01:04:43 +2026-09-16 16:34:11 | INFO | train rank=0 | step=48970/50000 progress=97.9% loss=0.106996 lr=1.159736e-07 lr_backbone=1.159736e-07 elapsed=50:47:44 avg_step=3.73s recent_step=3.46s eta=01:04:06 +2026-09-16 16:34:51 | INFO | train rank=0 | step=48980/50000 progress=98.0% loss=0.072911 lr=1.137334e-07 lr_backbone=1.137334e-07 elapsed=50:48:24 avg_step=3.73s recent_step=3.97s eta=01:03:28 +2026-09-16 16:35:27 | INFO | train rank=0 | step=48990/50000 progress=98.0% loss=0.115809 lr=1.115151e-07 lr_backbone=1.115151e-07 elapsed=50:48:59 avg_step=3.73s recent_step=3.58s eta=01:02:51 +2026-09-16 16:36:03 | INFO | train rank=0 | step=49000/50000 progress=98.0% loss=0.105316 lr=1.093186e-07 lr_backbone=1.093186e-07 elapsed=50:49:35 avg_step=3.73s recent_step=3.60s eta=01:02:14 +2026-09-16 16:36:44 | INFO | train rank=0 | step=49010/50000 progress=98.0% loss=0.094980 lr=1.071440e-07 lr_backbone=1.071440e-07 elapsed=50:50:17 avg_step=3.73s recent_step=4.18s eta=01:01:36 +2026-09-16 16:37:20 | INFO | train rank=0 | step=49020/50000 progress=98.0% loss=0.097239 lr=1.049911e-07 lr_backbone=1.049911e-07 elapsed=50:50:53 avg_step=3.73s recent_step=3.59s eta=01:00:59 +2026-09-16 16:37:56 | INFO | train rank=0 | step=49030/50000 progress=98.1% loss=0.098948 lr=1.028601e-07 lr_backbone=1.028601e-07 elapsed=50:51:29 avg_step=3.73s recent_step=3.57s eta=01:00:22 +2026-09-16 16:38:29 | INFO | train rank=0 | step=49040/50000 progress=98.1% loss=0.141006 lr=1.007509e-07 lr_backbone=1.007509e-07 elapsed=50:52:02 avg_step=3.73s recent_step=3.29s eta=00:59:44 +2026-09-16 16:39:17 | INFO | train rank=0 | step=49050/50000 progress=98.1% loss=0.082103 lr=9.866358e-08 lr_backbone=9.866358e-08 elapsed=50:52:50 avg_step=3.73s recent_step=4.80s eta=00:59:07 +2026-09-16 16:39:52 | INFO | train rank=0 | step=49060/50000 progress=98.1% loss=0.097850 lr=9.659805e-08 lr_backbone=9.659805e-08 elapsed=50:53:25 avg_step=3.73s recent_step=3.49s eta=00:58:30 +2026-09-16 16:40:23 | INFO | train rank=0 | step=49070/50000 progress=98.1% loss=0.117116 lr=9.455435e-08 lr_backbone=9.455435e-08 elapsed=50:53:56 avg_step=3.73s recent_step=3.12s eta=00:57:52 +2026-09-16 16:41:00 | INFO | train rank=0 | step=49080/50000 progress=98.2% loss=0.071360 lr=9.253248e-08 lr_backbone=9.253248e-08 elapsed=50:54:32 avg_step=3.73s recent_step=3.67s eta=00:57:15 +2026-09-16 16:41:48 | INFO | train rank=0 | step=49090/50000 progress=98.2% loss=0.105810 lr=9.053244e-08 lr_backbone=9.053244e-08 elapsed=50:55:21 avg_step=3.73s recent_step=4.87s eta=00:56:38 +2026-09-16 16:42:23 | INFO | train rank=0 | step=49100/50000 progress=98.2% loss=0.104774 lr=8.855423e-08 lr_backbone=8.855423e-08 elapsed=50:55:55 avg_step=3.73s recent_step=3.43s eta=00:56:00 +2026-09-16 16:42:58 | INFO | train rank=0 | step=49110/50000 progress=98.2% loss=0.062625 lr=8.659786e-08 lr_backbone=8.659786e-08 elapsed=50:56:31 avg_step=3.73s recent_step=3.51s eta=00:55:23 +2026-09-16 16:43:27 | INFO | train rank=0 | step=49120/50000 progress=98.2% loss=0.125742 lr=8.466332e-08 lr_backbone=8.466332e-08 elapsed=50:57:00 avg_step=3.73s recent_step=2.97s eta=00:54:46 +2026-09-16 16:44:21 | INFO | train rank=0 | step=49130/50000 progress=98.3% loss=0.125616 lr=8.275061e-08 lr_backbone=8.275061e-08 elapsed=50:57:53 avg_step=3.73s recent_step=5.31s eta=00:54:08 +2026-09-16 16:45:00 | INFO | train rank=0 | step=49140/50000 progress=98.3% loss=0.106042 lr=8.085974e-08 lr_backbone=8.085974e-08 elapsed=50:58:33 avg_step=3.73s recent_step=3.93s eta=00:53:31 +2026-09-16 16:45:34 | INFO | train rank=0 | step=49150/50000 progress=98.3% loss=0.089655 lr=7.899071e-08 lr_backbone=7.899071e-08 elapsed=50:59:07 avg_step=3.73s recent_step=3.43s eta=00:52:54 +2026-09-16 16:46:11 | INFO | train rank=0 | step=49160/50000 progress=98.3% loss=0.087845 lr=7.714351e-08 lr_backbone=7.714351e-08 elapsed=50:59:44 avg_step=3.73s recent_step=3.71s eta=00:52:16 +2026-09-16 16:46:57 | INFO | train rank=0 | step=49170/50000 progress=98.3% loss=0.077364 lr=7.531815e-08 lr_backbone=7.531815e-08 elapsed=51:00:29 avg_step=3.73s recent_step=4.55s eta=00:51:39 +2026-09-16 16:47:29 | INFO | train rank=0 | step=49180/50000 progress=98.4% loss=0.096863 lr=7.351463e-08 lr_backbone=7.351463e-08 elapsed=51:01:02 avg_step=3.73s recent_step=3.22s eta=00:51:02 +2026-09-16 16:48:07 | INFO | train rank=0 | step=49190/50000 progress=98.4% loss=0.074096 lr=7.173295e-08 lr_backbone=7.173295e-08 elapsed=51:01:39 avg_step=3.73s recent_step=3.77s eta=00:50:24 +2026-09-16 16:48:42 | INFO | train rank=0 | step=49200/50000 progress=98.4% loss=0.099524 lr=6.997311e-08 lr_backbone=6.997311e-08 elapsed=51:02:15 avg_step=3.73s recent_step=3.57s eta=00:49:47 +2026-09-16 16:49:31 | INFO | train rank=0 | step=49210/50000 progress=98.4% loss=0.102101 lr=6.823511e-08 lr_backbone=6.823511e-08 elapsed=51:03:04 avg_step=3.73s recent_step=4.89s eta=00:49:10 +2026-09-16 16:50:08 | INFO | train rank=0 | step=49220/50000 progress=98.4% loss=0.097844 lr=6.651896e-08 lr_backbone=6.651896e-08 elapsed=51:03:40 avg_step=3.73s recent_step=3.64s eta=00:48:33 +2026-09-16 16:50:40 | INFO | train rank=0 | step=49230/50000 progress=98.5% loss=0.106214 lr=6.482464e-08 lr_backbone=6.482464e-08 elapsed=51:04:13 avg_step=3.73s recent_step=3.27s eta=00:47:55 +2026-09-16 16:51:19 | INFO | train rank=0 | step=49240/50000 progress=98.5% loss=0.095490 lr=6.315217e-08 lr_backbone=6.315217e-08 elapsed=51:04:51 avg_step=3.73s recent_step=3.83s eta=00:47:18 +2026-09-16 16:52:03 | INFO | train rank=0 | step=49250/50000 progress=98.5% loss=0.095656 lr=6.150154e-08 lr_backbone=6.150154e-08 elapsed=51:05:36 avg_step=3.73s recent_step=4.48s eta=00:46:41 +2026-09-16 16:52:39 | INFO | train rank=0 | step=49260/50000 progress=98.5% loss=0.084879 lr=5.987276e-08 lr_backbone=5.987276e-08 elapsed=51:06:12 avg_step=3.73s recent_step=3.58s eta=00:46:03 +2026-09-16 16:53:18 | INFO | train rank=0 | step=49270/50000 progress=98.5% loss=0.116038 lr=5.826582e-08 lr_backbone=5.826582e-08 elapsed=51:06:51 avg_step=3.73s recent_step=3.87s eta=00:45:26 +2026-09-16 16:53:49 | INFO | train rank=0 | step=49280/50000 progress=98.6% loss=0.099093 lr=5.668073e-08 lr_backbone=5.668073e-08 elapsed=51:07:22 avg_step=3.73s recent_step=3.09s eta=00:44:48 +2026-09-16 16:54:30 | INFO | train rank=0 | step=49290/50000 progress=98.6% loss=0.108638 lr=5.511749e-08 lr_backbone=5.511749e-08 elapsed=51:08:03 avg_step=3.73s recent_step=4.14s eta=00:44:11 +2026-09-16 16:54:58 | INFO | train rank=0 | step=49300/50000 progress=98.6% loss=0.075664 lr=5.357609e-08 lr_backbone=5.357609e-08 elapsed=51:08:31 avg_step=3.73s recent_step=2.78s eta=00:43:34 +2026-09-16 16:55:36 | INFO | train rank=0 | step=49310/50000 progress=98.6% loss=0.081271 lr=5.205655e-08 lr_backbone=5.205655e-08 elapsed=51:09:08 avg_step=3.73s recent_step=3.76s eta=00:42:56 +2026-09-16 16:56:03 | INFO | train rank=0 | step=49320/50000 progress=98.6% loss=0.086798 lr=5.055885e-08 lr_backbone=5.055885e-08 elapsed=51:09:35 avg_step=3.73s recent_step=2.70s eta=00:42:19 +2026-09-16 16:56:50 | INFO | train rank=0 | step=49330/50000 progress=98.7% loss=0.107206 lr=4.908300e-08 lr_backbone=4.908300e-08 elapsed=51:10:23 avg_step=3.73s recent_step=4.72s eta=00:41:42 +2026-09-16 16:57:24 | INFO | train rank=0 | step=49340/50000 progress=98.7% loss=0.112335 lr=4.762900e-08 lr_backbone=4.762900e-08 elapsed=51:10:57 avg_step=3.73s recent_step=3.47s eta=00:41:04 +2026-09-16 16:57:58 | INFO | train rank=0 | step=49350/50000 progress=98.7% loss=0.073248 lr=4.619685e-08 lr_backbone=4.619685e-08 elapsed=51:11:31 avg_step=3.73s recent_step=3.39s eta=00:40:27 +2026-09-16 16:58:37 | INFO | train rank=0 | step=49360/50000 progress=98.7% loss=0.096500 lr=4.478655e-08 lr_backbone=4.478655e-08 elapsed=51:12:10 avg_step=3.73s recent_step=3.85s eta=00:39:50 +2026-09-16 16:59:21 | INFO | train rank=0 | step=49370/50000 progress=98.7% loss=0.058256 lr=4.339811e-08 lr_backbone=4.339811e-08 elapsed=51:12:54 avg_step=3.73s recent_step=4.39s eta=00:39:12 +2026-09-16 16:59:51 | INFO | train rank=0 | step=49380/50000 progress=98.8% loss=0.087521 lr=4.203152e-08 lr_backbone=4.203152e-08 elapsed=51:13:23 avg_step=3.73s recent_step=2.98s eta=00:38:35 +2026-09-16 17:00:36 | INFO | train rank=0 | step=49390/50000 progress=98.8% loss=0.079819 lr=4.068678e-08 lr_backbone=4.068678e-08 elapsed=51:14:09 avg_step=3.73s recent_step=4.54s eta=00:37:58 +2026-09-16 17:01:12 | INFO | train rank=0 | step=49400/50000 progress=98.8% loss=0.094611 lr=3.936389e-08 lr_backbone=3.936389e-08 elapsed=51:14:45 avg_step=3.73s recent_step=3.66s eta=00:37:20 +2026-09-16 17:02:00 | INFO | train rank=0 | step=49410/50000 progress=98.8% loss=0.071564 lr=3.806286e-08 lr_backbone=3.806286e-08 elapsed=51:15:33 avg_step=3.73s recent_step=4.73s eta=00:36:43 +2026-09-16 17:02:36 | INFO | train rank=0 | step=49420/50000 progress=98.8% loss=0.119028 lr=3.678369e-08 lr_backbone=3.678369e-08 elapsed=51:16:09 avg_step=3.73s recent_step=3.61s eta=00:36:06 +2026-09-16 17:03:08 | INFO | train rank=0 | step=49430/50000 progress=98.9% loss=0.068559 lr=3.552637e-08 lr_backbone=3.552637e-08 elapsed=51:16:41 avg_step=3.73s recent_step=3.22s eta=00:35:28 +2026-09-16 17:03:43 | INFO | train rank=0 | step=49440/50000 progress=98.9% loss=0.075465 lr=3.429090e-08 lr_backbone=3.429090e-08 elapsed=51:17:16 avg_step=3.73s recent_step=3.50s eta=00:34:51 +2026-09-16 17:04:19 | INFO | train rank=0 | step=49450/50000 progress=98.9% loss=0.090533 lr=3.307730e-08 lr_backbone=3.307730e-08 elapsed=51:17:52 avg_step=3.73s recent_step=3.65s eta=00:34:13 +2026-09-16 17:04:57 | INFO | train rank=0 | step=49460/50000 progress=98.9% loss=0.102575 lr=3.188555e-08 lr_backbone=3.188555e-08 elapsed=51:18:29 avg_step=3.73s recent_step=3.71s eta=00:33:36 +2026-09-16 17:05:46 | INFO | train rank=0 | step=49470/50000 progress=98.9% loss=0.111038 lr=3.071566e-08 lr_backbone=3.071566e-08 elapsed=51:19:19 avg_step=3.73s recent_step=4.99s eta=00:32:59 +2026-09-16 17:06:22 | INFO | train rank=0 | step=49480/50000 progress=99.0% loss=0.119911 lr=2.956762e-08 lr_backbone=2.956762e-08 elapsed=51:19:55 avg_step=3.73s recent_step=3.57s eta=00:32:22 +2026-09-16 17:06:51 | INFO | train rank=0 | step=49490/50000 progress=99.0% loss=0.081065 lr=2.844145e-08 lr_backbone=2.844145e-08 elapsed=51:20:24 avg_step=3.73s recent_step=2.89s eta=00:31:44 +2026-09-16 17:07:26 | INFO | train rank=0 | step=49500/50000 progress=99.0% loss=0.089167 lr=2.733713e-08 lr_backbone=2.733713e-08 elapsed=51:20:59 avg_step=3.73s recent_step=3.47s eta=00:31:07 +2026-09-16 17:08:18 | INFO | train rank=0 | step=49510/50000 progress=99.0% loss=0.091395 lr=2.625468e-08 lr_backbone=2.625468e-08 elapsed=51:21:51 avg_step=3.73s recent_step=5.26s eta=00:30:30 +2026-09-16 17:08:55 | INFO | train rank=0 | step=49520/50000 progress=99.0% loss=0.060692 lr=2.519408e-08 lr_backbone=2.519408e-08 elapsed=51:22:28 avg_step=3.73s recent_step=3.66s eta=00:29:52 +2026-09-16 17:09:34 | INFO | train rank=0 | step=49530/50000 progress=99.1% loss=0.074559 lr=2.415535e-08 lr_backbone=2.415535e-08 elapsed=51:23:06 avg_step=3.73s recent_step=3.85s eta=00:29:15 +2026-09-16 17:10:09 | INFO | train rank=0 | step=49540/50000 progress=99.1% loss=0.085781 lr=2.313847e-08 lr_backbone=2.313847e-08 elapsed=51:23:42 avg_step=3.73s recent_step=3.54s eta=00:28:38 +2026-09-16 17:10:55 | INFO | train rank=0 | step=49550/50000 progress=99.1% loss=0.108062 lr=2.214346e-08 lr_backbone=2.214346e-08 elapsed=51:24:27 avg_step=3.73s recent_step=4.56s eta=00:28:00 +2026-09-16 17:11:25 | INFO | train rank=0 | step=49560/50000 progress=99.1% loss=0.068302 lr=2.117031e-08 lr_backbone=2.117031e-08 elapsed=51:24:58 avg_step=3.73s recent_step=3.04s eta=00:27:23 +2026-09-16 17:11:59 | INFO | train rank=0 | step=49570/50000 progress=99.1% loss=0.084934 lr=2.021902e-08 lr_backbone=2.021902e-08 elapsed=51:25:32 avg_step=3.73s recent_step=3.39s eta=00:26:45 +2026-09-16 17:12:28 | INFO | train rank=0 | step=49580/50000 progress=99.2% loss=0.073336 lr=1.928960e-08 lr_backbone=1.928960e-08 elapsed=51:26:01 avg_step=3.73s recent_step=2.92s eta=00:26:08 +2026-09-16 17:13:18 | INFO | train rank=0 | step=49590/50000 progress=99.2% loss=0.103442 lr=1.838204e-08 lr_backbone=1.838204e-08 elapsed=51:26:51 avg_step=3.73s recent_step=5.03s eta=00:25:31 +2026-09-16 17:13:52 | INFO | train rank=0 | step=49600/50000 progress=99.2% loss=0.089320 lr=1.749634e-08 lr_backbone=1.749634e-08 elapsed=51:27:25 avg_step=3.73s recent_step=3.41s eta=00:24:53 +2026-09-16 17:14:27 | INFO | train rank=0 | step=49610/50000 progress=99.2% loss=0.092615 lr=1.663251e-08 lr_backbone=1.663251e-08 elapsed=51:27:59 avg_step=3.73s recent_step=3.43s eta=00:24:16 +2026-09-16 17:15:04 | INFO | train rank=0 | step=49620/50000 progress=99.2% loss=0.100285 lr=1.579054e-08 lr_backbone=1.579054e-08 elapsed=51:28:36 avg_step=3.73s recent_step=3.70s eta=00:23:39 +2026-09-16 17:15:47 | INFO | train rank=0 | step=49630/50000 progress=99.3% loss=0.073339 lr=1.497043e-08 lr_backbone=1.497043e-08 elapsed=51:29:20 avg_step=3.73s recent_step=4.35s eta=00:23:01 +2026-09-16 17:16:17 | INFO | train rank=0 | step=49640/50000 progress=99.3% loss=0.076686 lr=1.417219e-08 lr_backbone=1.417219e-08 elapsed=51:29:50 avg_step=3.73s recent_step=3.04s eta=00:22:24 +2026-09-16 17:16:57 | INFO | train rank=0 | step=49650/50000 progress=99.3% loss=0.088007 lr=1.339582e-08 lr_backbone=1.339582e-08 elapsed=51:30:29 avg_step=3.73s recent_step=3.92s eta=00:21:47 +2026-09-16 17:17:29 | INFO | train rank=0 | step=49660/50000 progress=99.3% loss=0.083589 lr=1.264131e-08 lr_backbone=1.264131e-08 elapsed=51:31:02 avg_step=3.73s recent_step=3.25s eta=00:21:09 +2026-09-16 17:18:17 | INFO | train rank=0 | step=49670/50000 progress=99.3% loss=0.129467 lr=1.190867e-08 lr_backbone=1.190867e-08 elapsed=51:31:50 avg_step=3.73s recent_step=4.78s eta=00:20:32 +2026-09-16 17:18:51 | INFO | train rank=0 | step=49680/50000 progress=99.4% loss=0.099500 lr=1.119789e-08 lr_backbone=1.119789e-08 elapsed=51:32:24 avg_step=3.73s recent_step=3.45s eta=00:19:55 +2026-09-16 17:19:32 | INFO | train rank=0 | step=49690/50000 progress=99.4% loss=0.087240 lr=1.050898e-08 lr_backbone=1.050898e-08 elapsed=51:33:05 avg_step=3.73s recent_step=4.10s eta=00:19:17 +2026-09-16 17:19:55 | INFO | train rank=0 | step=49700/50000 progress=99.4% loss=0.088360 lr=9.841942e-09 lr_backbone=9.841942e-09 elapsed=51:33:28 avg_step=3.73s recent_step=2.26s eta=00:18:40 +2026-09-16 17:20:50 | INFO | train rank=0 | step=49710/50000 progress=99.4% loss=0.083536 lr=9.196768e-09 lr_backbone=9.196768e-09 elapsed=51:34:23 avg_step=3.73s recent_step=5.47s eta=00:18:03 +2026-09-16 17:21:21 | INFO | train rank=0 | step=49720/50000 progress=99.4% loss=0.061944 lr=8.573461e-09 lr_backbone=8.573461e-09 elapsed=51:34:54 avg_step=3.73s recent_step=3.17s eta=00:17:25 +2026-09-16 17:21:58 | INFO | train rank=0 | step=49730/50000 progress=99.5% loss=0.077684 lr=7.972023e-09 lr_backbone=7.972023e-09 elapsed=51:35:31 avg_step=3.73s recent_step=3.66s eta=00:16:48 +2026-09-16 17:22:35 | INFO | train rank=0 | step=49740/50000 progress=99.5% loss=0.102743 lr=7.392452e-09 lr_backbone=7.392452e-09 elapsed=51:36:08 avg_step=3.73s recent_step=3.67s eta=00:16:11 +2026-09-16 17:23:30 | INFO | train rank=0 | step=49750/50000 progress=99.5% loss=0.105622 lr=6.834750e-09 lr_backbone=6.834750e-09 elapsed=51:37:02 avg_step=3.74s recent_step=5.49s eta=00:15:33 +2026-09-16 17:24:02 | INFO | train rank=0 | step=49760/50000 progress=99.5% loss=0.106474 lr=6.298917e-09 lr_backbone=6.298917e-09 elapsed=51:37:34 avg_step=3.74s recent_step=3.20s eta=00:14:56 +2026-09-16 17:24:37 | INFO | train rank=0 | step=49770/50000 progress=99.5% loss=0.081416 lr=5.784953e-09 lr_backbone=5.784953e-09 elapsed=51:38:09 avg_step=3.73s recent_step=3.50s eta=00:14:19 +2026-09-16 17:25:09 | INFO | train rank=0 | step=49780/50000 progress=99.6% loss=0.084830 lr=5.292858e-09 lr_backbone=5.292858e-09 elapsed=51:38:42 avg_step=3.73s recent_step=3.22s eta=00:13:41 +2026-09-16 17:26:05 | INFO | train rank=0 | step=49790/50000 progress=99.6% loss=0.094256 lr=4.822632e-09 lr_backbone=4.822632e-09 elapsed=51:39:37 avg_step=3.74s recent_step=5.56s eta=00:13:04 +2026-09-16 17:26:40 | INFO | train rank=0 | step=49800/50000 progress=99.6% loss=0.082912 lr=4.374276e-09 lr_backbone=4.374276e-09 elapsed=51:40:13 avg_step=3.74s recent_step=3.54s eta=00:12:27 +2026-09-16 17:27:16 | INFO | train rank=0 | step=49810/50000 progress=99.6% loss=0.088138 lr=3.947790e-09 lr_backbone=3.947790e-09 elapsed=51:40:49 avg_step=3.74s recent_step=3.58s eta=00:11:49 +2026-09-16 17:27:49 | INFO | train rank=0 | step=49820/50000 progress=99.6% loss=0.117827 lr=3.543173e-09 lr_backbone=3.543173e-09 elapsed=51:41:22 avg_step=3.74s recent_step=3.32s eta=00:11:12 +2026-09-16 17:28:32 | INFO | train rank=0 | step=49830/50000 progress=99.7% loss=0.088866 lr=3.160427e-09 lr_backbone=3.160427e-09 elapsed=51:42:05 avg_step=3.74s recent_step=4.30s eta=00:10:34 +2026-09-16 17:29:13 | INFO | train rank=0 | step=49840/50000 progress=99.7% loss=0.078022 lr=2.799551e-09 lr_backbone=2.799551e-09 elapsed=51:42:45 avg_step=3.74s recent_step=4.07s eta=00:09:57 +2026-09-16 17:29:41 | INFO | train rank=0 | step=49850/50000 progress=99.7% loss=0.076673 lr=2.460546e-09 lr_backbone=2.460546e-09 elapsed=51:43:14 avg_step=3.74s recent_step=2.83s eta=00:09:20 +2026-09-16 17:30:11 | INFO | train rank=0 | step=49860/50000 progress=99.7% loss=0.093227 lr=2.143411e-09 lr_backbone=2.143411e-09 elapsed=51:43:44 avg_step=3.73s recent_step=3.06s eta=00:08:42 +2026-09-16 17:31:06 | INFO | train rank=0 | step=49870/50000 progress=99.7% loss=0.087461 lr=1.848147e-09 lr_backbone=1.848147e-09 elapsed=51:44:39 avg_step=3.74s recent_step=5.46s eta=00:08:05 +2026-09-16 17:31:44 | INFO | train rank=0 | step=49880/50000 progress=99.8% loss=0.072218 lr=1.574754e-09 lr_backbone=1.574754e-09 elapsed=51:45:17 avg_step=3.74s recent_step=3.77s eta=00:07:28 +2026-09-16 17:32:17 | INFO | train rank=0 | step=49890/50000 progress=99.8% loss=0.090955 lr=1.323232e-09 lr_backbone=1.323232e-09 elapsed=51:45:50 avg_step=3.74s recent_step=3.31s eta=00:06:50 +2026-09-16 17:32:54 | INFO | train rank=0 | step=49900/50000 progress=99.8% loss=0.132379 lr=1.093581e-09 lr_backbone=1.093581e-09 elapsed=51:46:27 avg_step=3.74s recent_step=3.69s eta=00:06:13 +2026-09-16 17:33:43 | INFO | train rank=0 | step=49910/50000 progress=99.8% loss=0.066054 lr=8.858012e-10 lr_backbone=8.858012e-10 elapsed=51:47:16 avg_step=3.74s recent_step=4.89s eta=00:05:36 +2026-09-16 17:34:14 | INFO | train rank=0 | step=49920/50000 progress=99.8% loss=0.108818 lr=6.998928e-10 lr_backbone=6.998928e-10 elapsed=51:47:47 avg_step=3.74s recent_step=3.14s eta=00:04:58 +2026-09-16 17:34:49 | INFO | train rank=0 | step=49930/50000 progress=99.9% loss=0.077332 lr=5.358557e-10 lr_backbone=5.358557e-10 elapsed=51:48:22 avg_step=3.74s recent_step=3.53s eta=00:04:21 +2026-09-16 17:35:20 | INFO | train rank=0 | step=49940/50000 progress=99.9% loss=0.086497 lr=3.936901e-10 lr_backbone=3.936901e-10 elapsed=51:48:53 avg_step=3.74s recent_step=3.05s eta=00:03:44 +2026-09-16 17:36:03 | INFO | train rank=0 | step=49950/50000 progress=99.9% loss=0.108402 lr=2.733960e-10 lr_backbone=2.733960e-10 elapsed=51:49:36 avg_step=3.74s recent_step=4.33s eta=00:03:06 +2026-09-16 17:36:36 | INFO | train rank=0 | step=49960/50000 progress=99.9% loss=0.106112 lr=1.749735e-10 lr_backbone=1.749735e-10 elapsed=51:50:09 avg_step=3.74s recent_step=3.25s eta=00:02:29 +2026-09-16 17:37:10 | INFO | train rank=0 | step=49970/50000 progress=99.9% loss=0.082950 lr=9.842262e-11 lr_backbone=9.842262e-11 elapsed=51:50:42 avg_step=3.74s recent_step=3.39s eta=00:01:52 +2026-09-16 17:37:39 | INFO | train rank=0 | step=49980/50000 progress=100.0% loss=0.118901 lr=4.374339e-11 lr_backbone=4.374339e-11 elapsed=51:51:12 avg_step=3.73s recent_step=2.97s eta=00:01:14 +2026-09-16 17:38:17 | INFO | train rank=0 | step=49990/50000 progress=100.0% loss=0.075177 lr=1.093585e-11 lr_backbone=1.093585e-11 elapsed=51:51:50 avg_step=3.73s recent_step=3.76s eta=00:00:37 +2026-09-16 17:38:49 | INFO | train rank=0 | step=50000/50000 progress=100.0% loss=0.105713 lr=0.000000e+00 lr_backbone=0.000000e+00 elapsed=51:52:22 avg_step=3.73s recent_step=3.25s eta=00:00:00 +2026-09-16 17:39:33 | INFO | train rank=0 | Saved DeepSpeed checkpoint to /data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_50000 +2026-09-16 17:39:33 | INFO | train rank=0 | Finished training final_step=50000/50000 elapsed=51:53:06 avg_step=3.74s +2026-09-16 17:39:33 | INFO | train rank=0 | Training finished +2026-09-16 17:40:09 | INFO | train rank=0 | Saved final DeepSpeed checkpoint to /data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask/ds_checkpoint_final diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..76c72c09fdddc84465be9688098b634952617ba6 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3b2c5c69dce98a2cf8335980444c2e348684320fd7a08cb7d2ba33081a6629eb +size 8066685992 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..6d077851fbe2c85d4d3e0b29401de84e7efc8c57 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:995a00042643cb0d40d29c38865a508786566428a556377230f7e5e3610ff24c +size 8066667688 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..060936b569900f2496b305c3f444d24733bf9534 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b94be9c4a3abf6fe10e95d05fff3bb235f87b5a312b1aafead647a779a4073d2 +size 8066672104 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..0ae304e44c231be1c33dde5827fb7680ac7973ca --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c1e5e3f1a08be7d7f38723be85783df028bf8f74192559eb3b29570df5bc62f +size 8066671208 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_50000/global_step50000/mp_rank_00_model_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_50000/global_step50000/mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..faf2b439930397d80568ea86ad6f089d8ac1c174 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_50000/global_step50000/mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:db17858086cb75834dae5bc85e09a99012b0dbcbead2318014177a6c2f2f9a08 +size 5378167768 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_50000/latest b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_50000/latest new file mode 100644 index 0000000000000000000000000000000000000000..59a558c0ee32a055bd8fc16bf0d99a5950c59ac1 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_50000/latest @@ -0,0 +1 @@ +global_step50000 \ No newline at end of file diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_50000/zero_to_fp32.py b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_50000/zero_to_fp32.py new file mode 100644 index 0000000000000000000000000000000000000000..f40f1f0acbbff30e8e8d29906b56aae4e8b8a138 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_50000/zero_to_fp32.py @@ -0,0 +1,790 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: +# python zero_to_fp32.py . output_dir/ +# or +# python zero_to_fp32.py . output_dir/ --safe_serialization + +import argparse +import torch +import glob +import math +import os +import re +import gc +import json +import numpy as np +from tqdm import tqdm +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS, AUTOEP_LAYERS_KEY, + AUTOEP_LAYERS_KEY_LEGACY, AUTOEP_ZERO3_EXPERT_STATE_FORMAT_KEY, + AUTOEP_ZERO3_PARTITIONED_EXPERT_STATE_FORMAT) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def _has_autoep_zero3_partitioned_metadata(state_dict): + autoep_layers = state_dict.get(AUTOEP_LAYERS_KEY) + if autoep_layers is None: + autoep_layers = state_dict.get(AUTOEP_LAYERS_KEY_LEGACY) + if not isinstance(autoep_layers, list): + return False + return any( + isinstance(entry, dict) + and entry.get(AUTOEP_ZERO3_EXPERT_STATE_FORMAT_KEY) == AUTOEP_ZERO3_PARTITIONED_EXPERT_STATE_FORMAT + for entry in autoep_layers) + + +def _raise_if_autoep_zero3_partitioned_state(state_dict): + if _has_autoep_zero3_partitioned_metadata(state_dict): + raise NotImplementedError("zero_to_fp32 does not support AutoEP ZeRO-3 partition-native checkpoints. " + "AutoEP expert parameters are partitioned over expert replica groups, so " + "global data-parallel consolidation would produce incomplete expert tensors. " + "Use ds_to_universal.py for expert-aware conversion.") + + +def _raise_if_autoep_zero3_partitioned_checkpoint(model_files): + for file in model_files: + state_dict = torch.load(file, map_location=device, weights_only=False) + _raise_if_autoep_zero3_partitioned_state(state_dict) + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device, weights_only=False) + _raise_if_autoep_zero3_partitioned_state(state_dict) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + total_files = len(files) + state_dicts = [] + for f in tqdm(files, desc='Loading checkpoint shards'): + state_dict = torch.load(f, map_location=device, mmap=True, weights_only=False) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if ZERO_STAGE not in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + # parse_model_states rejects AutoEP ZeRO-3 partition-native checkpoints + # before the expensive optimizer-shard load below. + model_files = get_model_state_files(ds_checkpoint_dir) + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +class GatheredTensor: + """ + A pseudo tensor that collects partitioned weights. + It is more memory efficient when there are multiple groups. + """ + + def __init__(self, flat_groups, flat_groups_offset, offset, partitioned_numel, shape): + self.flat_groups = flat_groups + self.flat_groups_offset = flat_groups_offset + self.offset = offset + self.partitioned_numel = partitioned_numel + self.shape = shape + self.dtype = self.flat_groups[0][0].dtype + + def contiguous(self): + """ + Merge partitioned weights from flat_groups into a single tensor. + """ + end_idx = self.offset + self.partitioned_numel + world_size = len(self.flat_groups) + pad_flat_param_chunks = [] + + for rank_i in range(world_size): + # for each rank, we need to collect weights from related group/groups + flat_groups_at_rank_i = self.flat_groups[rank_i] + start_group_id = None + end_group_id = None + for group_id in range(len(self.flat_groups_offset)): + if self.flat_groups_offset[group_id] <= self.offset < self.flat_groups_offset[group_id + 1]: + start_group_id = group_id + if self.flat_groups_offset[group_id] < end_idx <= self.flat_groups_offset[group_id + 1]: + end_group_id = group_id + break + # collect weights from related group/groups + for group_id in range(start_group_id, end_group_id + 1): + flat_tensor = flat_groups_at_rank_i[group_id] + start_offset = self.offset - self.flat_groups_offset[group_id] + end_offset = min(end_idx, self.flat_groups_offset[group_id + 1]) - self.flat_groups_offset[group_id] + pad_flat_param_chunks.append(flat_tensor[start_offset:end_offset]) + + # collect weights from all ranks + pad_flat_param = torch.cat(pad_flat_param_chunks, dim=0) + param = pad_flat_param[:self.shape.numel()].view(self.shape).contiguous() + return param + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = sum([flat_group.numel() for flat_group in fp32_flat_groups[0]]) * world_size + + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + flat_groups_offset = [0] + list(np.cumsum([flat_tensor.numel() for flat_tensor in fp32_flat_groups[0]])) + for name, shape in tqdm(param_shapes.items(), desc='Gathering sharded weights'): + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # memory efficient tensor + tensor = GatheredTensor(fp32_flat_groups, flat_groups_offset, offset, partitioned_numel, shape) + state_dict[name] = tensor + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def to_torch_tensor(state_dict, return_empty_tensor=False): + """ + Convert state_dict of GatheredTensor to torch tensor + """ + torch_state_dict = {} + converted_tensors = {} + for name, tensor in state_dict.items(): + tensor_id = id(tensor) + if tensor_id in converted_tensors: # shared tensors + shared_tensor = torch_state_dict[converted_tensors[tensor_id]] + torch_state_dict[name] = shared_tensor + else: + converted_tensors[tensor_id] = name + if return_empty_tensor: + torch_state_dict[name] = torch.empty(tensor.shape, dtype=tensor.dtype) + else: + torch_state_dict[name] = tensor.contiguous() + return torch_state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, + tag=None, + exclude_frozen_parameters=False, + lazy_mode=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + - ``lazy_mode``: get state_dict in lazy mode. It returns a dict of pesduo tensor instead of torch tensor, which is more memory efficient. + Convert the pesduo tensor to torch tensor by ``.contiguous()`` + + Returns: + - pytorch ``state_dict`` + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + Note: the above usage may not work if your application doesn't have sufficient free CPU memory. + You may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. Or you can load state_dict in lazy mode :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, lazy_mode=True) # not on cpu + for name, lazy_tensor in state_dict.item(): + tensor = lazy_tensor.contiguous() # to cpu + print(name, tensor) + # del tensor to release memory if it no longer in use + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + state_dict = _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + if lazy_mode: + return state_dict + else: + return to_torch_tensor(state_dict) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, + output_dir, + max_shard_size="5GB", + safe_serialization=False, + tag=None, + exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_dir``: directory to the pytorch fp32 state_dict output files + - ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB + - ``safe_serialization``: whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`). + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + # Dependency pre-check + if safe_serialization: + try: + from safetensors.torch import save_file + except ImportError: + print('If you want to use `safe_serialization`, please `pip install safetensors`') + raise + if max_shard_size is not None: + try: + from huggingface_hub import split_torch_state_dict_into_shards + except ImportError: + print('If you want to use `max_shard_size`, please `pip install huggingface_hub`') + raise + + # Convert zero checkpoint to state_dict + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, + tag, + exclude_frozen_parameters, + lazy_mode=True) + + # Shard the model if it is too big. + weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin" + if max_shard_size is not None: + filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors") + # an memory-efficient approach for sharding + empty_state_dict = to_torch_tensor(state_dict, return_empty_tensor=True) + state_dict_split = split_torch_state_dict_into_shards(empty_state_dict, + filename_pattern=filename_pattern, + max_shard_size=max_shard_size) + else: + from collections import namedtuple + StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"]) + state_dict_split = StateDictSplit(is_sharded=False, + filename_to_tensors={weights_name: list(state_dict.keys())}) + + # Save the model by shard + os.makedirs(output_dir, exist_ok=True) + filename_to_tensors = state_dict_split.filename_to_tensors.items() + for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"): + shard_state_dict = {tensor_name: state_dict[tensor_name] for tensor_name in tensors} + shard_state_dict = to_torch_tensor(shard_state_dict) + output_path = os.path.join(output_dir, shard_file) + if safe_serialization: + save_file(shard_state_dict, output_path, metadata={"format": "pt"}) + else: + torch.save(shard_state_dict, output_path) + # release the memory of current shard + for tensor_name in list(shard_state_dict.keys()): + del state_dict[tensor_name] + del shard_state_dict[tensor_name] + del shard_state_dict + gc.collect() + + # Save index if sharded + if state_dict_split.is_sharded: + index = { + "metadata": state_dict_split.metadata, + "weight_map": state_dict_split.tensor_to_filename, + } + save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json" + save_index_file = os.path.join(output_dir, save_index_file) + with open(save_index_file, "w", encoding="utf-8") as f: + content = json.dumps(index, indent=2, sort_keys=True) + "\n" + f.write(content) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info("Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info("Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument("output_dir", + type=str, + help="directory to the pytorch fp32 state_dict output files" + "(e.g. path/checkpoint-12-output/)") + parser.add_argument( + "--max_shard_size", + type=str, + default="5GB", + help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size" + "lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`" + "We default it to 5GB in order for models to be able to run easily on free-tier google colab instances" + "without CPU OOM issues.") + parser.add_argument( + "--safe_serialization", + default=False, + action='store_true', + help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_dir, + max_shard_size=args.max_shard_size, + safe_serialization=args.safe_serialization, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..76c72c09fdddc84465be9688098b634952617ba6 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3b2c5c69dce98a2cf8335980444c2e348684320fd7a08cb7d2ba33081a6629eb +size 8066685992 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..6d077851fbe2c85d4d3e0b29401de84e7efc8c57 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:995a00042643cb0d40d29c38865a508786566428a556377230f7e5e3610ff24c +size 8066667688 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..060936b569900f2496b305c3f444d24733bf9534 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b94be9c4a3abf6fe10e95d05fff3bb235f87b5a312b1aafead647a779a4073d2 +size 8066672104 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..0ae304e44c231be1c33dde5827fb7680ac7973ca --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c1e5e3f1a08be7d7f38723be85783df028bf8f74192559eb3b29570df5bc62f +size 8066671208 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_final/global_step50000/mp_rank_00_model_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_final/global_step50000/mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..faf2b439930397d80568ea86ad6f089d8ac1c174 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_final/global_step50000/mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:db17858086cb75834dae5bc85e09a99012b0dbcbead2318014177a6c2f2f9a08 +size 5378167768 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_final/latest b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_final/latest new file mode 100644 index 0000000000000000000000000000000000000000..59a558c0ee32a055bd8fc16bf0d99a5950c59ac1 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_final/latest @@ -0,0 +1 @@ +global_step50000 \ No newline at end of file diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_final/zero_to_fp32.py b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_final/zero_to_fp32.py new file mode 100644 index 0000000000000000000000000000000000000000..f40f1f0acbbff30e8e8d29906b56aae4e8b8a138 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_final/zero_to_fp32.py @@ -0,0 +1,790 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: +# python zero_to_fp32.py . output_dir/ +# or +# python zero_to_fp32.py . output_dir/ --safe_serialization + +import argparse +import torch +import glob +import math +import os +import re +import gc +import json +import numpy as np +from tqdm import tqdm +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS, AUTOEP_LAYERS_KEY, + AUTOEP_LAYERS_KEY_LEGACY, AUTOEP_ZERO3_EXPERT_STATE_FORMAT_KEY, + AUTOEP_ZERO3_PARTITIONED_EXPERT_STATE_FORMAT) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def _has_autoep_zero3_partitioned_metadata(state_dict): + autoep_layers = state_dict.get(AUTOEP_LAYERS_KEY) + if autoep_layers is None: + autoep_layers = state_dict.get(AUTOEP_LAYERS_KEY_LEGACY) + if not isinstance(autoep_layers, list): + return False + return any( + isinstance(entry, dict) + and entry.get(AUTOEP_ZERO3_EXPERT_STATE_FORMAT_KEY) == AUTOEP_ZERO3_PARTITIONED_EXPERT_STATE_FORMAT + for entry in autoep_layers) + + +def _raise_if_autoep_zero3_partitioned_state(state_dict): + if _has_autoep_zero3_partitioned_metadata(state_dict): + raise NotImplementedError("zero_to_fp32 does not support AutoEP ZeRO-3 partition-native checkpoints. " + "AutoEP expert parameters are partitioned over expert replica groups, so " + "global data-parallel consolidation would produce incomplete expert tensors. " + "Use ds_to_universal.py for expert-aware conversion.") + + +def _raise_if_autoep_zero3_partitioned_checkpoint(model_files): + for file in model_files: + state_dict = torch.load(file, map_location=device, weights_only=False) + _raise_if_autoep_zero3_partitioned_state(state_dict) + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device, weights_only=False) + _raise_if_autoep_zero3_partitioned_state(state_dict) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + total_files = len(files) + state_dicts = [] + for f in tqdm(files, desc='Loading checkpoint shards'): + state_dict = torch.load(f, map_location=device, mmap=True, weights_only=False) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if ZERO_STAGE not in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + # parse_model_states rejects AutoEP ZeRO-3 partition-native checkpoints + # before the expensive optimizer-shard load below. + model_files = get_model_state_files(ds_checkpoint_dir) + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +class GatheredTensor: + """ + A pseudo tensor that collects partitioned weights. + It is more memory efficient when there are multiple groups. + """ + + def __init__(self, flat_groups, flat_groups_offset, offset, partitioned_numel, shape): + self.flat_groups = flat_groups + self.flat_groups_offset = flat_groups_offset + self.offset = offset + self.partitioned_numel = partitioned_numel + self.shape = shape + self.dtype = self.flat_groups[0][0].dtype + + def contiguous(self): + """ + Merge partitioned weights from flat_groups into a single tensor. + """ + end_idx = self.offset + self.partitioned_numel + world_size = len(self.flat_groups) + pad_flat_param_chunks = [] + + for rank_i in range(world_size): + # for each rank, we need to collect weights from related group/groups + flat_groups_at_rank_i = self.flat_groups[rank_i] + start_group_id = None + end_group_id = None + for group_id in range(len(self.flat_groups_offset)): + if self.flat_groups_offset[group_id] <= self.offset < self.flat_groups_offset[group_id + 1]: + start_group_id = group_id + if self.flat_groups_offset[group_id] < end_idx <= self.flat_groups_offset[group_id + 1]: + end_group_id = group_id + break + # collect weights from related group/groups + for group_id in range(start_group_id, end_group_id + 1): + flat_tensor = flat_groups_at_rank_i[group_id] + start_offset = self.offset - self.flat_groups_offset[group_id] + end_offset = min(end_idx, self.flat_groups_offset[group_id + 1]) - self.flat_groups_offset[group_id] + pad_flat_param_chunks.append(flat_tensor[start_offset:end_offset]) + + # collect weights from all ranks + pad_flat_param = torch.cat(pad_flat_param_chunks, dim=0) + param = pad_flat_param[:self.shape.numel()].view(self.shape).contiguous() + return param + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = sum([flat_group.numel() for flat_group in fp32_flat_groups[0]]) * world_size + + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + flat_groups_offset = [0] + list(np.cumsum([flat_tensor.numel() for flat_tensor in fp32_flat_groups[0]])) + for name, shape in tqdm(param_shapes.items(), desc='Gathering sharded weights'): + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # memory efficient tensor + tensor = GatheredTensor(fp32_flat_groups, flat_groups_offset, offset, partitioned_numel, shape) + state_dict[name] = tensor + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def to_torch_tensor(state_dict, return_empty_tensor=False): + """ + Convert state_dict of GatheredTensor to torch tensor + """ + torch_state_dict = {} + converted_tensors = {} + for name, tensor in state_dict.items(): + tensor_id = id(tensor) + if tensor_id in converted_tensors: # shared tensors + shared_tensor = torch_state_dict[converted_tensors[tensor_id]] + torch_state_dict[name] = shared_tensor + else: + converted_tensors[tensor_id] = name + if return_empty_tensor: + torch_state_dict[name] = torch.empty(tensor.shape, dtype=tensor.dtype) + else: + torch_state_dict[name] = tensor.contiguous() + return torch_state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, + tag=None, + exclude_frozen_parameters=False, + lazy_mode=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + - ``lazy_mode``: get state_dict in lazy mode. It returns a dict of pesduo tensor instead of torch tensor, which is more memory efficient. + Convert the pesduo tensor to torch tensor by ``.contiguous()`` + + Returns: + - pytorch ``state_dict`` + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + Note: the above usage may not work if your application doesn't have sufficient free CPU memory. + You may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. Or you can load state_dict in lazy mode :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, lazy_mode=True) # not on cpu + for name, lazy_tensor in state_dict.item(): + tensor = lazy_tensor.contiguous() # to cpu + print(name, tensor) + # del tensor to release memory if it no longer in use + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + state_dict = _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + if lazy_mode: + return state_dict + else: + return to_torch_tensor(state_dict) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, + output_dir, + max_shard_size="5GB", + safe_serialization=False, + tag=None, + exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_dir``: directory to the pytorch fp32 state_dict output files + - ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB + - ``safe_serialization``: whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`). + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + # Dependency pre-check + if safe_serialization: + try: + from safetensors.torch import save_file + except ImportError: + print('If you want to use `safe_serialization`, please `pip install safetensors`') + raise + if max_shard_size is not None: + try: + from huggingface_hub import split_torch_state_dict_into_shards + except ImportError: + print('If you want to use `max_shard_size`, please `pip install huggingface_hub`') + raise + + # Convert zero checkpoint to state_dict + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, + tag, + exclude_frozen_parameters, + lazy_mode=True) + + # Shard the model if it is too big. + weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin" + if max_shard_size is not None: + filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors") + # an memory-efficient approach for sharding + empty_state_dict = to_torch_tensor(state_dict, return_empty_tensor=True) + state_dict_split = split_torch_state_dict_into_shards(empty_state_dict, + filename_pattern=filename_pattern, + max_shard_size=max_shard_size) + else: + from collections import namedtuple + StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"]) + state_dict_split = StateDictSplit(is_sharded=False, + filename_to_tensors={weights_name: list(state_dict.keys())}) + + # Save the model by shard + os.makedirs(output_dir, exist_ok=True) + filename_to_tensors = state_dict_split.filename_to_tensors.items() + for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"): + shard_state_dict = {tensor_name: state_dict[tensor_name] for tensor_name in tensors} + shard_state_dict = to_torch_tensor(shard_state_dict) + output_path = os.path.join(output_dir, shard_file) + if safe_serialization: + save_file(shard_state_dict, output_path, metadata={"format": "pt"}) + else: + torch.save(shard_state_dict, output_path) + # release the memory of current shard + for tensor_name in list(shard_state_dict.keys()): + del state_dict[tensor_name] + del shard_state_dict[tensor_name] + del shard_state_dict + gc.collect() + + # Save index if sharded + if state_dict_split.is_sharded: + index = { + "metadata": state_dict_split.metadata, + "weight_map": state_dict_split.tensor_to_filename, + } + save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json" + save_index_file = os.path.join(output_dir, save_index_file) + with open(save_index_file, "w", encoding="utf-8") as f: + content = json.dumps(index, indent=2, sort_keys=True) + "\n" + f.write(content) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info("Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info("Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument("output_dir", + type=str, + help="directory to the pytorch fp32 state_dict output files" + "(e.g. path/checkpoint-12-output/)") + parser.add_argument( + "--max_shard_size", + type=str, + default="5GB", + help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size" + "lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`" + "We default it to 5GB in order for models to be able to run easily on free-tier google colab instances" + "without CPU OOM issues.") + parser.add_argument( + "--safe_serialization", + default=False, + action='store_true', + help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_dir, + max_shard_size=args.max_shard_size, + safe_serialization=args.safe_serialization, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/train.log b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/train.log new file mode 100644 index 0000000000000000000000000000000000000000..fe9ac21db3b48f90d52526fb50eb00e9c9b9b783 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/train.log @@ -0,0 +1,5020 @@ +2026-09-14 13:54:59 | INFO | train rank=0 | Starting training project=codebase_v8_droid_joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16 save_dir=/data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16 distributed=True deepspeed=True world_size=4 device=cuda:0 +2026-09-14 13:55:21 | INFO | train rank=0 | No pretrained checkpoint provided; training from scratch +2026-09-14 13:55:21 | INFO | train rank=0 | Initializing Droid LeRobot Dataset: path=/data/NTU_slab/draven/data/DROID fps=15.00 use_relabelled_instruction=False +2026-09-14 13:55:21 | INFO | train rank=0 | Batch config: total_batch_size=128 world_size=4 grad_acc_steps=1 per_device_batch_size=32 +2026-09-14 13:55:21 | INFO | train rank=0 | Dataset image_resize_size=320 +2026-09-14 13:55:21 | INFO | train rank=0 | Droid uses parquet-shard streaming with a per-rank shuffle buffer: batch_size=32 buffer_size=1024 num_workers=8 drop_last=True +2026-09-14 14:25:53 | INFO | train rank=0 | Starting training project=codebase_v8_droid_joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16 save_dir=/data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16 distributed=True deepspeed=True world_size=4 device=cuda:0 +2026-09-14 14:26:15 | INFO | train rank=0 | No pretrained checkpoint provided; training from scratch +2026-09-14 14:26:15 | INFO | train rank=0 | Initializing Droid LeRobot Dataset: path=/data/NTU_slab/draven/data/DROID fps=15.00 use_relabelled_instruction=False +2026-09-14 14:26:15 | INFO | train rank=0 | Batch config: total_batch_size=128 world_size=4 grad_acc_steps=1 per_device_batch_size=32 +2026-09-14 14:26:15 | INFO | train rank=0 | Dataset image_resize_size=320 +2026-09-14 14:26:15 | INFO | train rank=0 | Droid uses parquet-shard streaming with a per-rank shuffle buffer: batch_size=32 buffer_size=1024 num_workers=8 drop_last=True +2026-09-14 14:26:23 | INFO | train rank=0 | Droid streaming layout: parquet_shards=50 active_workers=8 worker_buffer_min=128 worker_buffer_max=128 max_active_shards=2 +2026-09-14 14:26:23 | INFO | train rank=0 | Resolved Droid joint action stats from /data/NTU_slab/draven/data/DROID/meta/joint_delta_stats_h16.json: q01=[-0.3050941824913025, -0.47471314668655396, -0.3020089268684387, -0.493502140045166, -0.4844675362110138, -0.4544465243816376, -0.550822913646698] q99=[0.30148017406463623, 0.5795503258705139, 0.3001617193222046, 0.5117637515068054, 0.4750978946685791, 0.5400477647781372, 0.5597710609436035] +2026-09-14 14:26:23 | INFO | train rank=0 | Trainable params: 2,688,879,624 / 2,688,879,624 +2026-09-14 14:30:21 | INFO | train rank=0 | Starting training start_step=0 max_steps=50000 remaining_steps=50000 log_interval=10 save_interval=50000 +2026-09-14 14:33:20 | INFO | train rank=0 | step=10/50000 progress=0.0% loss=1.202225 lr=3.600000e-07 lr_backbone=3.600000e-07 elapsed=00:02:58 avg_step=17.90s recent_step=17.90s eta=248:32:41 +2026-09-14 14:33:50 | INFO | train rank=0 | step=20/50000 progress=0.0% loss=1.201931 lr=7.600000e-07 lr_backbone=7.600000e-07 elapsed=00:03:28 avg_step=10.45s recent_step=2.99s eta=145:00:48 +2026-09-14 14:34:28 | INFO | train rank=0 | step=30/50000 progress=0.1% loss=1.214134 lr=1.160000e-06 lr_backbone=1.160000e-06 elapsed=00:04:06 avg_step=8.21s recent_step=3.73s eta=113:54:52 +2026-09-14 14:34:59 | INFO | train rank=0 | step=40/50000 progress=0.1% loss=1.232838 lr=1.560000e-06 lr_backbone=1.560000e-06 elapsed=00:04:37 avg_step=6.95s recent_step=3.17s eta=96:24:14 +2026-09-14 14:35:45 | INFO | train rank=0 | step=50/50000 progress=0.1% loss=1.199577 lr=1.960000e-06 lr_backbone=1.960000e-06 elapsed=00:05:23 avg_step=6.48s recent_step=4.60s eta=89:51:57 +2026-09-14 14:36:15 | INFO | train rank=0 | step=60/50000 progress=0.1% loss=1.156020 lr=2.360000e-06 lr_backbone=2.360000e-06 elapsed=00:05:53 avg_step=5.89s recent_step=2.93s eta=81:39:30 +2026-09-14 14:36:41 | INFO | train rank=0 | step=70/50000 progress=0.1% loss=1.121820 lr=2.760000e-06 lr_backbone=2.760000e-06 elapsed=00:06:19 avg_step=5.42s recent_step=2.60s eta=75:07:38 +2026-09-14 14:37:09 | INFO | train rank=0 | step=80/50000 progress=0.2% loss=1.167464 lr=3.160000e-06 lr_backbone=3.160000e-06 elapsed=00:06:47 avg_step=5.10s recent_step=2.84s eta=70:39:07 +2026-09-14 14:37:55 | INFO | train rank=0 | step=90/50000 progress=0.2% loss=1.170982 lr=3.560000e-06 lr_backbone=3.560000e-06 elapsed=00:07:33 avg_step=5.04s recent_step=4.59s eta=69:51:51 +2026-09-14 14:38:30 | INFO | train rank=0 | step=100/50000 progress=0.2% loss=1.129605 lr=3.960000e-06 lr_backbone=3.960000e-06 elapsed=00:08:08 avg_step=4.88s recent_step=3.46s eta=67:39:51 +2026-09-14 14:39:08 | INFO | train rank=0 | step=110/50000 progress=0.2% loss=1.124686 lr=4.360000e-06 lr_backbone=4.360000e-06 elapsed=00:08:46 avg_step=4.79s recent_step=3.87s eta=66:22:49 +2026-09-14 14:39:41 | INFO | train rank=0 | step=120/50000 progress=0.2% loss=1.121937 lr=4.760000e-06 lr_backbone=4.760000e-06 elapsed=00:09:19 avg_step=4.66s recent_step=3.28s eta=64:37:09 +2026-09-14 14:40:28 | INFO | train rank=0 | step=130/50000 progress=0.3% loss=1.040055 lr=5.160000e-06 lr_backbone=5.160000e-06 elapsed=00:10:06 avg_step=4.66s recent_step=4.68s eta=64:37:14 +2026-09-14 14:40:59 | INFO | train rank=0 | step=140/50000 progress=0.3% loss=1.043003 lr=5.560000e-06 lr_backbone=5.560000e-06 elapsed=00:10:37 avg_step=4.55s recent_step=3.10s eta=63:03:38 +2026-09-14 14:41:26 | INFO | train rank=0 | step=150/50000 progress=0.3% loss=1.014271 lr=5.960000e-06 lr_backbone=5.960000e-06 elapsed=00:11:04 avg_step=4.43s recent_step=2.71s eta=61:20:44 +2026-09-14 14:42:00 | INFO | train rank=0 | step=160/50000 progress=0.3% loss=0.974995 lr=6.360000e-06 lr_backbone=6.360000e-06 elapsed=00:11:38 avg_step=4.37s recent_step=3.39s eta=60:26:01 +2026-09-14 14:42:44 | INFO | train rank=0 | step=170/50000 progress=0.3% loss=1.000194 lr=6.760000e-06 lr_backbone=6.760000e-06 elapsed=00:12:22 avg_step=4.37s recent_step=4.38s eta=60:26:15 +2026-09-14 14:43:18 | INFO | train rank=0 | step=180/50000 progress=0.4% loss=0.882384 lr=7.160000e-06 lr_backbone=7.160000e-06 elapsed=00:12:56 avg_step=4.31s recent_step=3.40s eta=59:40:53 +2026-09-14 14:44:02 | INFO | train rank=0 | step=190/50000 progress=0.4% loss=0.870770 lr=7.560000e-06 lr_backbone=7.560000e-06 elapsed=00:13:40 avg_step=4.32s recent_step=4.37s eta=59:42:53 +2026-09-14 14:44:27 | INFO | train rank=0 | step=200/50000 progress=0.4% loss=0.874388 lr=7.960000e-06 lr_backbone=7.960000e-06 elapsed=00:14:05 avg_step=4.23s recent_step=2.56s eta=58:29:07 +2026-09-14 14:45:09 | INFO | train rank=0 | step=210/50000 progress=0.4% loss=0.830836 lr=8.360000e-06 lr_backbone=8.360000e-06 elapsed=00:14:47 avg_step=4.23s recent_step=4.24s eta=58:29:00 +2026-09-14 14:45:42 | INFO | train rank=0 | step=220/50000 progress=0.4% loss=0.754512 lr=8.760000e-06 lr_backbone=8.760000e-06 elapsed=00:15:20 avg_step=4.18s recent_step=3.27s eta=57:51:59 +2026-09-14 14:46:12 | INFO | train rank=0 | step=230/50000 progress=0.5% loss=0.677468 lr=9.160000e-06 lr_backbone=9.160000e-06 elapsed=00:15:50 avg_step=4.13s recent_step=3.01s eta=57:08:48 +2026-09-14 14:46:39 | INFO | train rank=0 | step=240/50000 progress=0.5% loss=0.641256 lr=9.560000e-06 lr_backbone=9.560000e-06 elapsed=00:16:17 avg_step=4.07s recent_step=2.71s eta=56:18:53 +2026-09-14 14:47:22 | INFO | train rank=0 | step=250/50000 progress=0.5% loss=0.514854 lr=9.960000e-06 lr_backbone=9.960000e-06 elapsed=00:17:00 avg_step=4.08s recent_step=4.25s eta=56:23:58 +2026-09-14 14:47:56 | INFO | train rank=0 | step=260/50000 progress=0.5% loss=0.470147 lr=1.036000e-05 lr_backbone=1.036000e-05 elapsed=00:17:34 avg_step=4.05s recent_step=3.39s eta=56:01:06 +2026-09-14 14:48:31 | INFO | train rank=0 | step=270/50000 progress=0.5% loss=0.467129 lr=1.076000e-05 lr_backbone=1.076000e-05 elapsed=00:18:09 avg_step=4.03s recent_step=3.50s eta=55:43:20 +2026-09-14 14:48:58 | INFO | train rank=0 | step=280/50000 progress=0.6% loss=0.369339 lr=1.116000e-05 lr_backbone=1.116000e-05 elapsed=00:18:36 avg_step=3.99s recent_step=2.76s eta=55:05:02 +2026-09-14 14:49:45 | INFO | train rank=0 | step=290/50000 progress=0.6% loss=0.346153 lr=1.156000e-05 lr_backbone=1.156000e-05 elapsed=00:19:23 avg_step=4.01s recent_step=4.66s eta=55:23:29 +2026-09-14 14:50:12 | INFO | train rank=0 | step=300/50000 progress=0.6% loss=0.337980 lr=1.196000e-05 lr_backbone=1.196000e-05 elapsed=00:19:50 avg_step=3.97s recent_step=2.70s eta=54:46:40 +2026-09-14 14:50:47 | INFO | train rank=0 | step=310/50000 progress=0.6% loss=0.315017 lr=1.236000e-05 lr_backbone=1.236000e-05 elapsed=00:20:25 avg_step=3.95s recent_step=3.47s eta=54:32:45 +2026-09-14 14:51:20 | INFO | train rank=0 | step=320/50000 progress=0.6% loss=0.351939 lr=1.276000e-05 lr_backbone=1.276000e-05 elapsed=00:20:58 avg_step=3.93s recent_step=3.34s eta=54:16:18 +2026-09-14 14:52:07 | INFO | train rank=0 | step=330/50000 progress=0.7% loss=0.283984 lr=1.316000e-05 lr_backbone=1.316000e-05 elapsed=00:21:45 avg_step=3.95s recent_step=4.66s eta=54:33:57 +2026-09-14 14:52:39 | INFO | train rank=0 | step=340/50000 progress=0.7% loss=0.285372 lr=1.356000e-05 lr_backbone=1.356000e-05 elapsed=00:22:17 avg_step=3.93s recent_step=3.28s eta=54:16:48 +2026-09-14 14:53:11 | INFO | train rank=0 | step=350/50000 progress=0.7% loss=0.245635 lr=1.396000e-05 lr_backbone=1.396000e-05 elapsed=00:22:49 avg_step=3.91s recent_step=3.19s eta=53:58:34 +2026-09-14 14:53:49 | INFO | train rank=0 | step=360/50000 progress=0.7% loss=0.251484 lr=1.436000e-05 lr_backbone=1.436000e-05 elapsed=00:23:27 avg_step=3.91s recent_step=3.74s eta=53:53:57 +2026-09-14 14:54:26 | INFO | train rank=0 | step=370/50000 progress=0.7% loss=0.288055 lr=1.476000e-05 lr_backbone=1.476000e-05 elapsed=00:24:04 avg_step=3.90s recent_step=3.69s eta=53:48:24 +2026-09-14 14:54:58 | INFO | train rank=0 | step=380/50000 progress=0.8% loss=0.239937 lr=1.516000e-05 lr_backbone=1.516000e-05 elapsed=00:24:36 avg_step=3.89s recent_step=3.24s eta=53:33:19 +2026-09-14 14:55:42 | INFO | train rank=0 | step=390/50000 progress=0.8% loss=0.273216 lr=1.556000e-05 lr_backbone=1.556000e-05 elapsed=00:25:20 avg_step=3.90s recent_step=4.36s eta=53:42:41 +2026-09-14 14:56:13 | INFO | train rank=0 | step=400/50000 progress=0.8% loss=0.241954 lr=1.596000e-05 lr_backbone=1.596000e-05 elapsed=00:25:51 avg_step=3.88s recent_step=3.14s eta=53:26:27 +2026-09-14 14:56:52 | INFO | train rank=0 | step=410/50000 progress=0.8% loss=0.261590 lr=1.636000e-05 lr_backbone=1.636000e-05 elapsed=00:26:30 avg_step=3.88s recent_step=3.94s eta=53:27:05 +2026-09-14 14:57:27 | INFO | train rank=0 | step=420/50000 progress=0.8% loss=0.258727 lr=1.676000e-05 lr_backbone=1.676000e-05 elapsed=00:27:05 avg_step=3.87s recent_step=3.46s eta=53:18:12 +2026-09-14 14:58:06 | INFO | train rank=0 | step=430/50000 progress=0.9% loss=0.283424 lr=1.716000e-05 lr_backbone=1.716000e-05 elapsed=00:27:44 avg_step=3.87s recent_step=3.94s eta=53:18:49 +2026-09-14 14:58:36 | INFO | train rank=0 | step=440/50000 progress=0.9% loss=0.215101 lr=1.756000e-05 lr_backbone=1.756000e-05 elapsed=00:28:14 avg_step=3.85s recent_step=2.96s eta=53:00:59 +2026-09-14 14:59:12 | INFO | train rank=0 | step=450/50000 progress=0.9% loss=0.204605 lr=1.796000e-05 lr_backbone=1.796000e-05 elapsed=00:28:50 avg_step=3.85s recent_step=3.63s eta=52:56:12 +2026-09-14 14:59:47 | INFO | train rank=0 | step=460/50000 progress=0.9% loss=0.198649 lr=1.836000e-05 lr_backbone=1.836000e-05 elapsed=00:29:25 avg_step=3.84s recent_step=3.43s eta=52:48:08 +2026-09-14 15:00:38 | INFO | train rank=0 | step=470/50000 progress=0.9% loss=0.191035 lr=1.876000e-05 lr_backbone=1.876000e-05 elapsed=00:30:16 avg_step=3.87s recent_step=5.16s eta=53:10:48 +2026-09-14 15:01:12 | INFO | train rank=0 | step=480/50000 progress=1.0% loss=0.183043 lr=1.916000e-05 lr_backbone=1.916000e-05 elapsed=00:30:50 avg_step=3.85s recent_step=3.34s eta=53:01:12 +2026-09-14 15:01:47 | INFO | train rank=0 | step=490/50000 progress=1.0% loss=0.218179 lr=1.956000e-05 lr_backbone=1.956000e-05 elapsed=00:31:25 avg_step=3.85s recent_step=3.55s eta=52:55:29 +2026-09-14 15:02:18 | INFO | train rank=0 | step=500/50000 progress=1.0% loss=0.169423 lr=1.996000e-05 lr_backbone=1.996000e-05 elapsed=00:31:56 avg_step=3.83s recent_step=3.11s eta=52:42:43 +2026-09-14 15:02:56 | INFO | train rank=0 | step=510/50000 progress=1.0% loss=0.176128 lr=2.036000e-05 lr_backbone=2.036000e-05 elapsed=00:32:34 avg_step=3.83s recent_step=3.77s eta=52:41:03 +2026-09-14 15:03:32 | INFO | train rank=0 | step=520/50000 progress=1.0% loss=0.222637 lr=2.076000e-05 lr_backbone=2.076000e-05 elapsed=00:33:10 avg_step=3.83s recent_step=3.64s eta=52:37:23 +2026-09-14 15:04:10 | INFO | train rank=0 | step=530/50000 progress=1.1% loss=0.184744 lr=2.116000e-05 lr_backbone=2.116000e-05 elapsed=00:33:48 avg_step=3.83s recent_step=3.77s eta=52:35:48 +2026-09-14 15:04:42 | INFO | train rank=0 | step=540/50000 progress=1.1% loss=0.203500 lr=2.156000e-05 lr_backbone=2.156000e-05 elapsed=00:34:20 avg_step=3.82s recent_step=3.18s eta=52:25:16 +2026-09-14 15:05:27 | INFO | train rank=0 | step=550/50000 progress=1.1% loss=0.204685 lr=2.196000e-05 lr_backbone=2.196000e-05 elapsed=00:35:05 avg_step=3.83s recent_step=4.50s eta=52:34:51 +2026-09-14 15:05:58 | INFO | train rank=0 | step=560/50000 progress=1.1% loss=0.238735 lr=2.236000e-05 lr_backbone=2.236000e-05 elapsed=00:35:36 avg_step=3.81s recent_step=3.08s eta=52:23:16 +2026-09-14 15:06:29 | INFO | train rank=0 | step=570/50000 progress=1.1% loss=0.181208 lr=2.276000e-05 lr_backbone=2.276000e-05 elapsed=00:36:07 avg_step=3.80s recent_step=3.16s eta=52:13:13 +2026-09-14 15:07:04 | INFO | train rank=0 | step=580/50000 progress=1.2% loss=0.206925 lr=2.316000e-05 lr_backbone=2.316000e-05 elapsed=00:36:42 avg_step=3.80s recent_step=3.49s eta=52:08:07 +2026-09-14 15:07:39 | INFO | train rank=0 | step=590/50000 progress=1.2% loss=0.215958 lr=2.356000e-05 lr_backbone=2.356000e-05 elapsed=00:37:17 avg_step=3.79s recent_step=3.47s eta=52:02:52 +2026-09-14 15:08:07 | INFO | train rank=0 | step=600/50000 progress=1.2% loss=0.203502 lr=2.396000e-05 lr_backbone=2.396000e-05 elapsed=00:37:45 avg_step=3.78s recent_step=2.79s eta=51:48:25 +2026-09-14 15:09:03 | INFO | train rank=0 | step=610/50000 progress=1.2% loss=0.161059 lr=2.436000e-05 lr_backbone=2.436000e-05 elapsed=00:38:41 avg_step=3.80s recent_step=5.58s eta=52:12:08 +2026-09-14 15:09:32 | INFO | train rank=0 | step=620/50000 progress=1.2% loss=0.217555 lr=2.476000e-05 lr_backbone=2.476000e-05 elapsed=00:39:10 avg_step=3.79s recent_step=2.96s eta=52:00:14 +2026-09-14 15:10:02 | INFO | train rank=0 | step=630/50000 progress=1.3% loss=0.188285 lr=2.516000e-05 lr_backbone=2.516000e-05 elapsed=00:39:40 avg_step=3.78s recent_step=2.94s eta=51:48:32 +2026-09-14 15:10:35 | INFO | train rank=0 | step=640/50000 progress=1.3% loss=0.215863 lr=2.556000e-05 lr_backbone=2.556000e-05 elapsed=00:40:13 avg_step=3.77s recent_step=3.32s eta=51:42:00 +2026-09-14 15:11:20 | INFO | train rank=0 | step=650/50000 progress=1.3% loss=0.161677 lr=2.596000e-05 lr_backbone=2.596000e-05 elapsed=00:40:58 avg_step=3.78s recent_step=4.51s eta=51:50:45 +2026-09-14 15:11:47 | INFO | train rank=0 | step=660/50000 progress=1.3% loss=0.214136 lr=2.636000e-05 lr_backbone=2.636000e-05 elapsed=00:41:25 avg_step=3.77s recent_step=2.70s eta=51:36:36 +2026-09-14 15:12:18 | INFO | train rank=0 | step=670/50000 progress=1.3% loss=0.179368 lr=2.676000e-05 lr_backbone=2.676000e-05 elapsed=00:41:56 avg_step=3.76s recent_step=3.07s eta=51:27:27 +2026-09-14 15:12:53 | INFO | train rank=0 | step=680/50000 progress=1.4% loss=0.263313 lr=2.716000e-05 lr_backbone=2.716000e-05 elapsed=00:42:31 avg_step=3.75s recent_step=3.53s eta=51:24:05 +2026-09-14 15:13:42 | INFO | train rank=0 | step=690/50000 progress=1.4% loss=0.167403 lr=2.756000e-05 lr_backbone=2.756000e-05 elapsed=00:43:20 avg_step=3.77s recent_step=4.88s eta=51:36:54 +2026-09-14 15:14:12 | INFO | train rank=0 | step=700/50000 progress=1.4% loss=0.190120 lr=2.796000e-05 lr_backbone=2.796000e-05 elapsed=00:43:50 avg_step=3.76s recent_step=3.05s eta=51:27:49 +2026-09-14 15:14:38 | INFO | train rank=0 | step=710/50000 progress=1.4% loss=0.156023 lr=2.836000e-05 lr_backbone=2.836000e-05 elapsed=00:44:16 avg_step=3.74s recent_step=2.61s eta=51:13:53 +2026-09-14 15:15:16 | INFO | train rank=0 | step=720/50000 progress=1.4% loss=0.203428 lr=2.876000e-05 lr_backbone=2.876000e-05 elapsed=00:44:54 avg_step=3.74s recent_step=3.74s eta=51:13:14 +2026-09-14 15:15:51 | INFO | train rank=0 | step=730/50000 progress=1.5% loss=0.191914 lr=2.916000e-05 lr_backbone=2.916000e-05 elapsed=00:45:29 avg_step=3.74s recent_step=3.52s eta=51:10:06 +2026-09-14 15:16:23 | INFO | train rank=0 | step=740/50000 progress=1.5% loss=0.182338 lr=2.956000e-05 lr_backbone=2.956000e-05 elapsed=00:46:01 avg_step=3.73s recent_step=3.20s eta=51:03:30 +2026-09-14 15:16:53 | INFO | train rank=0 | step=750/50000 progress=1.5% loss=0.215083 lr=2.996000e-05 lr_backbone=2.996000e-05 elapsed=00:46:31 avg_step=3.72s recent_step=3.05s eta=50:55:27 +2026-09-14 15:17:28 | INFO | train rank=0 | step=760/50000 progress=1.5% loss=0.204341 lr=3.036000e-05 lr_backbone=3.036000e-05 elapsed=00:47:06 avg_step=3.72s recent_step=3.48s eta=50:52:10 +2026-09-14 15:18:09 | INFO | train rank=0 | step=770/50000 progress=1.5% loss=0.191354 lr=3.076000e-05 lr_backbone=3.076000e-05 elapsed=00:47:47 avg_step=3.72s recent_step=4.11s eta=50:55:44 +2026-09-14 15:18:45 | INFO | train rank=0 | step=780/50000 progress=1.6% loss=0.187846 lr=3.116000e-05 lr_backbone=3.116000e-05 elapsed=00:48:23 avg_step=3.72s recent_step=3.54s eta=50:53:08 +2026-09-14 15:19:17 | INFO | train rank=0 | step=790/50000 progress=1.6% loss=0.197273 lr=3.156000e-05 lr_backbone=3.156000e-05 elapsed=00:48:55 avg_step=3.72s recent_step=3.24s eta=50:47:31 +2026-09-14 15:19:45 | INFO | train rank=0 | step=800/50000 progress=1.6% loss=0.167343 lr=3.196000e-05 lr_backbone=3.196000e-05 elapsed=00:49:23 avg_step=3.70s recent_step=2.79s eta=50:37:25 +2026-09-14 15:20:27 | INFO | train rank=0 | step=810/50000 progress=1.6% loss=0.196600 lr=3.236000e-05 lr_backbone=3.236000e-05 elapsed=00:50:05 avg_step=3.71s recent_step=4.27s eta=50:42:29 +2026-09-14 15:20:59 | INFO | train rank=0 | step=820/50000 progress=1.6% loss=0.215608 lr=3.276000e-05 lr_backbone=3.276000e-05 elapsed=00:50:37 avg_step=3.70s recent_step=3.16s eta=50:36:22 +2026-09-14 15:21:32 | INFO | train rank=0 | step=830/50000 progress=1.7% loss=0.176806 lr=3.316000e-05 lr_backbone=3.316000e-05 elapsed=00:51:10 avg_step=3.70s recent_step=3.27s eta=50:31:30 +2026-09-14 15:22:13 | INFO | train rank=0 | step=840/50000 progress=1.7% loss=0.168129 lr=3.356000e-05 lr_backbone=3.356000e-05 elapsed=00:51:51 avg_step=3.70s recent_step=4.07s eta=50:34:29 +2026-09-14 15:22:59 | INFO | train rank=0 | step=850/50000 progress=1.7% loss=0.183252 lr=3.396000e-05 lr_backbone=3.396000e-05 elapsed=00:52:37 avg_step=3.71s recent_step=4.62s eta=50:42:41 +2026-09-14 15:23:27 | INFO | train rank=0 | step=860/50000 progress=1.7% loss=0.166464 lr=3.436000e-05 lr_backbone=3.436000e-05 elapsed=00:53:05 avg_step=3.70s recent_step=2.84s eta=50:33:42 +2026-09-14 15:24:04 | INFO | train rank=0 | step=870/50000 progress=1.7% loss=0.158284 lr=3.476000e-05 lr_backbone=3.476000e-05 elapsed=00:53:42 avg_step=3.70s recent_step=3.72s eta=50:33:12 +2026-09-14 15:24:41 | INFO | train rank=0 | step=880/50000 progress=1.8% loss=0.156265 lr=3.516000e-05 lr_backbone=3.516000e-05 elapsed=00:54:19 avg_step=3.70s recent_step=3.64s eta=50:31:57 +2026-09-14 15:25:31 | INFO | train rank=0 | step=890/50000 progress=1.8% loss=0.179028 lr=3.556000e-05 lr_backbone=3.556000e-05 elapsed=00:55:09 avg_step=3.72s recent_step=5.04s eta=50:43:35 +2026-09-14 15:26:04 | INFO | train rank=0 | step=900/50000 progress=1.8% loss=0.180321 lr=3.596000e-05 lr_backbone=3.596000e-05 elapsed=00:55:42 avg_step=3.71s recent_step=3.32s eta=50:39:23 +2026-09-14 15:26:36 | INFO | train rank=0 | step=910/50000 progress=1.8% loss=0.185187 lr=3.636000e-05 lr_backbone=3.636000e-05 elapsed=00:56:14 avg_step=3.71s recent_step=3.21s eta=50:34:15 +2026-09-14 15:27:17 | INFO | train rank=0 | step=920/50000 progress=1.8% loss=0.154584 lr=3.676000e-05 lr_backbone=3.676000e-05 elapsed=00:56:55 avg_step=3.71s recent_step=4.04s eta=50:36:32 +2026-09-14 15:28:03 | INFO | train rank=0 | step=930/50000 progress=1.9% loss=0.232857 lr=3.716000e-05 lr_backbone=3.716000e-05 elapsed=00:57:41 avg_step=3.72s recent_step=4.66s eta=50:44:17 +2026-09-14 15:28:34 | INFO | train rank=0 | step=940/50000 progress=1.9% loss=0.157625 lr=3.756000e-05 lr_backbone=3.756000e-05 elapsed=00:58:12 avg_step=3.72s recent_step=3.10s eta=50:38:14 +2026-09-14 15:29:11 | INFO | train rank=0 | step=950/50000 progress=1.9% loss=0.179614 lr=3.796000e-05 lr_backbone=3.796000e-05 elapsed=00:58:49 avg_step=3.72s recent_step=3.70s eta=50:37:29 +2026-09-14 15:29:41 | INFO | train rank=0 | step=960/50000 progress=1.9% loss=0.196458 lr=3.836000e-05 lr_backbone=3.836000e-05 elapsed=00:59:19 avg_step=3.71s recent_step=3.01s eta=50:30:53 +2026-09-14 15:30:32 | INFO | train rank=0 | step=970/50000 progress=1.9% loss=0.178953 lr=3.876000e-05 lr_backbone=3.876000e-05 elapsed=01:00:10 avg_step=3.72s recent_step=5.02s eta=50:41:17 +2026-09-14 15:31:04 | INFO | train rank=0 | step=980/50000 progress=2.0% loss=0.179071 lr=3.916000e-05 lr_backbone=3.916000e-05 elapsed=01:00:42 avg_step=3.72s recent_step=3.23s eta=50:36:36 +2026-09-14 15:31:42 | INFO | train rank=0 | step=990/50000 progress=2.0% loss=0.147620 lr=3.956000e-05 lr_backbone=3.956000e-05 elapsed=01:01:20 avg_step=3.72s recent_step=3.83s eta=50:36:56 +2026-09-14 15:32:08 | INFO | train rank=0 | step=1000/50000 progress=2.0% loss=0.188798 lr=3.996000e-05 lr_backbone=3.996000e-05 elapsed=01:01:46 avg_step=3.71s recent_step=2.53s eta=50:26:36 +2026-09-14 15:32:55 | INFO | train rank=0 | step=1010/50000 progress=2.0% loss=0.196840 lr=4.036000e-05 lr_backbone=4.036000e-05 elapsed=01:02:33 avg_step=3.72s recent_step=4.72s eta=50:34:11 +2026-09-14 15:33:19 | INFO | train rank=0 | step=1020/50000 progress=2.0% loss=0.162017 lr=4.076000e-05 lr_backbone=4.076000e-05 elapsed=01:02:57 avg_step=3.70s recent_step=2.43s eta=50:23:16 +2026-09-14 15:33:56 | INFO | train rank=0 | step=1030/50000 progress=2.1% loss=0.212881 lr=4.116000e-05 lr_backbone=4.116000e-05 elapsed=01:03:34 avg_step=3.70s recent_step=3.68s eta=50:22:26 +2026-09-14 15:34:29 | INFO | train rank=0 | step=1040/50000 progress=2.1% loss=0.204140 lr=4.156000e-05 lr_backbone=4.156000e-05 elapsed=01:04:07 avg_step=3.70s recent_step=3.30s eta=50:18:42 +2026-09-14 15:35:16 | INFO | train rank=0 | step=1050/50000 progress=2.1% loss=0.202473 lr=4.196000e-05 lr_backbone=4.196000e-05 elapsed=01:04:55 avg_step=3.71s recent_step=4.76s eta=50:26:21 +2026-09-14 15:35:48 | INFO | train rank=0 | step=1060/50000 progress=2.1% loss=0.176627 lr=4.236000e-05 lr_backbone=4.236000e-05 elapsed=01:05:26 avg_step=3.70s recent_step=3.17s eta=50:21:34 +2026-09-14 15:36:18 | INFO | train rank=0 | step=1070/50000 progress=2.1% loss=0.176796 lr=4.276000e-05 lr_backbone=4.276000e-05 elapsed=01:05:56 avg_step=3.70s recent_step=3.00s eta=50:15:36 +2026-09-14 15:36:53 | INFO | train rank=0 | step=1080/50000 progress=2.2% loss=0.172949 lr=4.316000e-05 lr_backbone=4.316000e-05 elapsed=01:06:31 avg_step=3.70s recent_step=3.46s eta=50:13:11 +2026-09-14 15:37:31 | INFO | train rank=0 | step=1090/50000 progress=2.2% loss=0.203999 lr=4.356000e-05 lr_backbone=4.356000e-05 elapsed=01:07:09 avg_step=3.70s recent_step=3.80s eta=50:13:20 +2026-09-14 15:38:02 | INFO | train rank=0 | step=1100/50000 progress=2.2% loss=0.181866 lr=4.396000e-05 lr_backbone=4.396000e-05 elapsed=01:07:40 avg_step=3.69s recent_step=3.16s eta=50:08:46 +2026-09-14 15:38:33 | INFO | train rank=0 | step=1110/50000 progress=2.2% loss=0.164285 lr=4.436000e-05 lr_backbone=4.436000e-05 elapsed=01:08:11 avg_step=3.69s recent_step=3.06s eta=50:03:32 +2026-09-14 15:39:07 | INFO | train rank=0 | step=1120/50000 progress=2.2% loss=0.159918 lr=4.476000e-05 lr_backbone=4.476000e-05 elapsed=01:08:45 avg_step=3.68s recent_step=3.35s eta=50:00:29 +2026-09-14 15:39:43 | INFO | train rank=0 | step=1130/50000 progress=2.3% loss=0.156350 lr=4.516000e-05 lr_backbone=4.516000e-05 elapsed=01:09:21 avg_step=3.68s recent_step=3.65s eta=49:59:36 +2026-09-14 15:40:18 | INFO | train rank=0 | step=1140/50000 progress=2.3% loss=0.181939 lr=4.556000e-05 lr_backbone=4.556000e-05 elapsed=01:09:56 avg_step=3.68s recent_step=3.54s eta=49:57:59 +2026-09-14 15:40:56 | INFO | train rank=0 | step=1150/50000 progress=2.3% loss=0.192403 lr=4.596000e-05 lr_backbone=4.596000e-05 elapsed=01:10:34 avg_step=3.68s recent_step=3.77s eta=49:57:58 +2026-09-14 15:41:24 | INFO | train rank=0 | step=1160/50000 progress=2.3% loss=0.179524 lr=4.636000e-05 lr_backbone=4.636000e-05 elapsed=01:11:02 avg_step=3.67s recent_step=2.79s eta=49:51:04 +2026-09-14 15:41:54 | INFO | train rank=0 | step=1170/50000 progress=2.3% loss=0.200119 lr=4.676000e-05 lr_backbone=4.676000e-05 elapsed=01:11:32 avg_step=3.67s recent_step=3.04s eta=49:46:01 +2026-09-14 15:42:28 | INFO | train rank=0 | step=1180/50000 progress=2.4% loss=0.173303 lr=4.716000e-05 lr_backbone=4.716000e-05 elapsed=01:12:06 avg_step=3.67s recent_step=3.41s eta=49:43:39 +2026-09-14 15:43:03 | INFO | train rank=0 | step=1190/50000 progress=2.4% loss=0.176426 lr=4.756000e-05 lr_backbone=4.756000e-05 elapsed=01:12:41 avg_step=3.66s recent_step=3.44s eta=49:41:28 +2026-09-14 15:43:46 | INFO | train rank=0 | step=1200/50000 progress=2.4% loss=0.186763 lr=4.796000e-05 lr_backbone=4.796000e-05 elapsed=01:13:24 avg_step=3.67s recent_step=4.29s eta=49:45:07 +2026-09-14 15:44:30 | INFO | train rank=0 | step=1210/50000 progress=2.4% loss=0.162088 lr=4.836000e-05 lr_backbone=4.836000e-05 elapsed=01:14:08 avg_step=3.68s recent_step=4.37s eta=49:49:14 +2026-09-14 15:45:06 | INFO | train rank=0 | step=1220/50000 progress=2.4% loss=0.160914 lr=4.876000e-05 lr_backbone=4.876000e-05 elapsed=01:14:44 avg_step=3.68s recent_step=3.66s eta=49:48:32 +2026-09-14 15:45:37 | INFO | train rank=0 | step=1230/50000 progress=2.5% loss=0.174100 lr=4.916000e-05 lr_backbone=4.916000e-05 elapsed=01:15:15 avg_step=3.67s recent_step=3.06s eta=49:43:51 +2026-09-14 15:46:24 | INFO | train rank=0 | step=1240/50000 progress=2.5% loss=0.146434 lr=4.956000e-05 lr_backbone=4.956000e-05 elapsed=01:16:02 avg_step=3.68s recent_step=4.70s eta=49:50:01 +2026-09-14 15:46:55 | INFO | train rank=0 | step=1250/50000 progress=2.5% loss=0.163214 lr=4.996000e-05 lr_backbone=4.996000e-05 elapsed=01:16:33 avg_step=3.67s recent_step=3.13s eta=49:45:48 +2026-09-14 15:47:29 | INFO | train rank=0 | step=1260/50000 progress=2.5% loss=0.172414 lr=5.036000e-05 lr_backbone=5.036000e-05 elapsed=01:17:07 avg_step=3.67s recent_step=3.36s eta=49:43:10 +2026-09-14 15:48:03 | INFO | train rank=0 | step=1270/50000 progress=2.5% loss=0.158865 lr=5.076000e-05 lr_backbone=5.076000e-05 elapsed=01:17:41 avg_step=3.67s recent_step=3.43s eta=49:41:02 +2026-09-14 15:48:44 | INFO | train rank=0 | step=1280/50000 progress=2.6% loss=0.180789 lr=5.116000e-05 lr_backbone=5.116000e-05 elapsed=01:18:22 avg_step=3.67s recent_step=4.10s eta=49:43:09 +2026-09-14 15:49:25 | INFO | train rank=0 | step=1290/50000 progress=2.6% loss=0.173888 lr=5.156000e-05 lr_backbone=5.156000e-05 elapsed=01:19:03 avg_step=3.68s recent_step=4.14s eta=49:45:29 +2026-09-14 15:50:02 | INFO | train rank=0 | step=1300/50000 progress=2.6% loss=0.202639 lr=5.196000e-05 lr_backbone=5.196000e-05 elapsed=01:19:40 avg_step=3.68s recent_step=3.68s eta=49:44:53 +2026-09-14 15:50:36 | INFO | train rank=0 | step=1310/50000 progress=2.6% loss=0.162162 lr=5.236000e-05 lr_backbone=5.236000e-05 elapsed=01:20:14 avg_step=3.68s recent_step=3.40s eta=49:42:33 +2026-09-14 15:51:09 | INFO | train rank=0 | step=1320/50000 progress=2.6% loss=0.205439 lr=5.276000e-05 lr_backbone=5.276000e-05 elapsed=01:20:47 avg_step=3.67s recent_step=3.26s eta=49:39:22 +2026-09-14 15:52:02 | INFO | train rank=0 | step=1330/50000 progress=2.7% loss=0.171545 lr=5.316000e-05 lr_backbone=5.316000e-05 elapsed=01:21:40 avg_step=3.68s recent_step=5.34s eta=49:48:55 +2026-09-14 15:52:36 | INFO | train rank=0 | step=1340/50000 progress=2.7% loss=0.156701 lr=5.356000e-05 lr_backbone=5.356000e-05 elapsed=01:22:14 avg_step=3.68s recent_step=3.41s eta=49:46:38 +2026-09-14 15:53:13 | INFO | train rank=0 | step=1350/50000 progress=2.7% loss=0.172750 lr=5.396000e-05 lr_backbone=5.396000e-05 elapsed=01:22:51 avg_step=3.68s recent_step=3.63s eta=49:45:41 +2026-09-14 15:53:41 | INFO | train rank=0 | step=1360/50000 progress=2.7% loss=0.157070 lr=5.436000e-05 lr_backbone=5.436000e-05 elapsed=01:23:19 avg_step=3.68s recent_step=2.83s eta=49:40:00 +2026-09-14 15:54:27 | INFO | train rank=0 | step=1370/50000 progress=2.7% loss=0.220158 lr=5.476000e-05 lr_backbone=5.476000e-05 elapsed=01:24:05 avg_step=3.68s recent_step=4.57s eta=49:44:42 +2026-09-14 15:55:07 | INFO | train rank=0 | step=1380/50000 progress=2.8% loss=0.154585 lr=5.516000e-05 lr_backbone=5.516000e-05 elapsed=01:24:45 avg_step=3.69s recent_step=4.09s eta=49:46:28 +2026-09-14 15:55:38 | INFO | train rank=0 | step=1390/50000 progress=2.8% loss=0.187382 lr=5.556000e-05 lr_backbone=5.556000e-05 elapsed=01:25:16 avg_step=3.68s recent_step=3.03s eta=49:42:01 +2026-09-14 15:56:09 | INFO | train rank=0 | step=1400/50000 progress=2.8% loss=0.206354 lr=5.596000e-05 lr_backbone=5.596000e-05 elapsed=01:25:47 avg_step=3.68s recent_step=3.12s eta=49:38:08 +2026-09-14 15:56:41 | INFO | train rank=0 | step=1410/50000 progress=2.8% loss=0.155464 lr=5.636000e-05 lr_backbone=5.636000e-05 elapsed=01:26:19 avg_step=3.67s recent_step=3.21s eta=49:34:49 +2026-09-14 15:57:32 | INFO | train rank=0 | step=1420/50000 progress=2.8% loss=0.138839 lr=5.676000e-05 lr_backbone=5.676000e-05 elapsed=01:27:10 avg_step=3.68s recent_step=5.11s eta=49:42:25 +2026-09-14 15:58:06 | INFO | train rank=0 | step=1430/50000 progress=2.9% loss=0.142837 lr=5.716000e-05 lr_backbone=5.716000e-05 elapsed=01:27:44 avg_step=3.68s recent_step=3.39s eta=49:40:08 +2026-09-14 15:58:35 | INFO | train rank=0 | step=1440/50000 progress=2.9% loss=0.148837 lr=5.756000e-05 lr_backbone=5.756000e-05 elapsed=01:28:13 avg_step=3.68s recent_step=2.93s eta=49:35:17 +2026-09-14 15:59:13 | INFO | train rank=0 | step=1450/50000 progress=2.9% loss=0.138192 lr=5.796000e-05 lr_backbone=5.796000e-05 elapsed=01:28:51 avg_step=3.68s recent_step=3.74s eta=49:35:01 +2026-09-14 15:59:57 | INFO | train rank=0 | step=1460/50000 progress=2.9% loss=0.192068 lr=5.836000e-05 lr_backbone=5.836000e-05 elapsed=01:29:35 avg_step=3.68s recent_step=4.46s eta=49:38:45 +2026-09-14 16:00:30 | INFO | train rank=0 | step=1470/50000 progress=2.9% loss=0.185218 lr=5.876000e-05 lr_backbone=5.876000e-05 elapsed=01:30:08 avg_step=3.68s recent_step=3.26s eta=49:35:47 +2026-09-14 16:00:58 | INFO | train rank=0 | step=1480/50000 progress=3.0% loss=0.173707 lr=5.916000e-05 lr_backbone=5.916000e-05 elapsed=01:30:36 avg_step=3.67s recent_step=2.80s eta=49:30:21 +2026-09-14 16:01:38 | INFO | train rank=0 | step=1490/50000 progress=3.0% loss=0.219002 lr=5.956000e-05 lr_backbone=5.956000e-05 elapsed=01:31:16 avg_step=3.68s recent_step=4.02s eta=49:31:36 +2026-09-14 16:02:29 | INFO | train rank=0 | step=1500/50000 progress=3.0% loss=0.157522 lr=5.996000e-05 lr_backbone=5.996000e-05 elapsed=01:32:07 avg_step=3.68s recent_step=5.07s eta=49:38:31 +2026-09-14 16:02:59 | INFO | train rank=0 | step=1510/50000 progress=3.0% loss=0.142781 lr=6.036000e-05 lr_backbone=6.036000e-05 elapsed=01:32:37 avg_step=3.68s recent_step=3.03s eta=49:34:23 +2026-09-14 16:03:24 | INFO | train rank=0 | step=1520/50000 progress=3.0% loss=0.188818 lr=6.076000e-05 lr_backbone=6.076000e-05 elapsed=01:33:02 avg_step=3.67s recent_step=2.52s eta=49:27:35 +2026-09-14 16:03:59 | INFO | train rank=0 | step=1530/50000 progress=3.1% loss=0.175991 lr=6.116000e-05 lr_backbone=6.116000e-05 elapsed=01:33:37 avg_step=3.67s recent_step=3.48s eta=49:25:58 +2026-09-14 16:04:51 | INFO | train rank=0 | step=1540/50000 progress=3.1% loss=0.156560 lr=6.156000e-05 lr_backbone=6.156000e-05 elapsed=01:34:29 avg_step=3.68s recent_step=5.20s eta=49:33:23 +2026-09-14 16:05:24 | INFO | train rank=0 | step=1550/50000 progress=3.1% loss=0.163326 lr=6.196000e-05 lr_backbone=6.196000e-05 elapsed=01:35:02 avg_step=3.68s recent_step=3.26s eta=49:30:36 +2026-09-14 16:05:55 | INFO | train rank=0 | step=1560/50000 progress=3.1% loss=0.150236 lr=6.236000e-05 lr_backbone=6.236000e-05 elapsed=01:35:33 avg_step=3.68s recent_step=3.18s eta=49:27:25 +2026-09-14 16:06:22 | INFO | train rank=0 | step=1570/50000 progress=3.1% loss=0.157616 lr=6.276000e-05 lr_backbone=6.276000e-05 elapsed=01:36:00 avg_step=3.67s recent_step=2.70s eta=49:21:46 +2026-09-14 16:07:16 | INFO | train rank=0 | step=1580/50000 progress=3.2% loss=0.185888 lr=6.316000e-05 lr_backbone=6.316000e-05 elapsed=01:36:54 avg_step=3.68s recent_step=5.39s eta=49:29:57 +2026-09-14 16:07:41 | INFO | train rank=0 | step=1590/50000 progress=3.2% loss=0.168087 lr=6.356000e-05 lr_backbone=6.356000e-05 elapsed=01:37:19 avg_step=3.67s recent_step=2.44s eta=49:23:01 +2026-09-14 16:08:20 | INFO | train rank=0 | step=1600/50000 progress=3.2% loss=0.171326 lr=6.396000e-05 lr_backbone=6.396000e-05 elapsed=01:37:58 avg_step=3.67s recent_step=3.97s eta=49:23:53 +2026-09-14 16:08:52 | INFO | train rank=0 | step=1610/50000 progress=3.2% loss=0.143701 lr=6.436000e-05 lr_backbone=6.436000e-05 elapsed=01:38:30 avg_step=3.67s recent_step=3.18s eta=49:20:48 +2026-09-14 16:09:30 | INFO | train rank=0 | step=1620/50000 progress=3.2% loss=0.168299 lr=6.476000e-05 lr_backbone=6.476000e-05 elapsed=01:39:08 avg_step=3.67s recent_step=3.84s eta=49:21:01 +2026-09-14 16:10:01 | INFO | train rank=0 | step=1630/50000 progress=3.3% loss=0.142287 lr=6.516000e-05 lr_backbone=6.516000e-05 elapsed=01:39:39 avg_step=3.67s recent_step=3.01s eta=49:17:08 +2026-09-14 16:10:33 | INFO | train rank=0 | step=1640/50000 progress=3.3% loss=0.178480 lr=6.556000e-05 lr_backbone=6.556000e-05 elapsed=01:40:11 avg_step=3.67s recent_step=3.29s eta=49:14:39 +2026-09-14 16:11:06 | INFO | train rank=0 | step=1650/50000 progress=3.3% loss=0.177915 lr=6.596000e-05 lr_backbone=6.596000e-05 elapsed=01:40:44 avg_step=3.66s recent_step=3.27s eta=49:12:07 +2026-09-14 16:11:55 | INFO | train rank=0 | step=1660/50000 progress=3.3% loss=0.181842 lr=6.636000e-05 lr_backbone=6.636000e-05 elapsed=01:41:33 avg_step=3.67s recent_step=4.86s eta=49:17:20 +2026-09-14 16:12:23 | INFO | train rank=0 | step=1670/50000 progress=3.3% loss=0.171957 lr=6.676000e-05 lr_backbone=6.676000e-05 elapsed=01:42:01 avg_step=3.67s recent_step=2.80s eta=49:12:30 +2026-09-14 16:13:01 | INFO | train rank=0 | step=1680/50000 progress=3.4% loss=0.160079 lr=6.716000e-05 lr_backbone=6.716000e-05 elapsed=01:42:39 avg_step=3.67s recent_step=3.80s eta=49:12:31 +2026-09-14 16:13:32 | INFO | train rank=0 | step=1690/50000 progress=3.4% loss=0.177359 lr=6.756000e-05 lr_backbone=6.756000e-05 elapsed=01:43:10 avg_step=3.66s recent_step=3.15s eta=49:09:27 +2026-09-14 16:14:22 | INFO | train rank=0 | step=1700/50000 progress=3.4% loss=0.157348 lr=6.796000e-05 lr_backbone=6.796000e-05 elapsed=01:44:00 avg_step=3.67s recent_step=4.97s eta=49:15:00 +2026-09-14 16:14:54 | INFO | train rank=0 | step=1710/50000 progress=3.4% loss=0.188555 lr=6.836000e-05 lr_backbone=6.836000e-05 elapsed=01:44:32 avg_step=3.67s recent_step=3.16s eta=49:12:00 +2026-09-14 16:15:30 | INFO | train rank=0 | step=1720/50000 progress=3.4% loss=0.171070 lr=6.876000e-05 lr_backbone=6.876000e-05 elapsed=01:45:08 avg_step=3.67s recent_step=3.60s eta=49:11:05 +2026-09-14 16:16:06 | INFO | train rank=0 | step=1730/50000 progress=3.5% loss=0.149275 lr=6.916000e-05 lr_backbone=6.916000e-05 elapsed=01:45:44 avg_step=3.67s recent_step=3.68s eta=49:10:31 +2026-09-14 16:16:53 | INFO | train rank=0 | step=1740/50000 progress=3.5% loss=0.171981 lr=6.956000e-05 lr_backbone=6.956000e-05 elapsed=01:46:31 avg_step=3.67s recent_step=4.66s eta=49:14:30 +2026-09-14 16:17:31 | INFO | train rank=0 | step=1750/50000 progress=3.5% loss=0.151986 lr=6.996000e-05 lr_backbone=6.996000e-05 elapsed=01:47:09 avg_step=3.67s recent_step=3.77s eta=49:14:21 +2026-09-14 16:18:03 | INFO | train rank=0 | step=1760/50000 progress=3.5% loss=0.180906 lr=7.036000e-05 lr_backbone=7.036000e-05 elapsed=01:47:41 avg_step=3.67s recent_step=3.28s eta=49:11:57 +2026-09-14 16:18:29 | INFO | train rank=0 | step=1770/50000 progress=3.5% loss=0.158940 lr=7.076000e-05 lr_backbone=7.076000e-05 elapsed=01:48:07 avg_step=3.67s recent_step=2.60s eta=49:06:27 +2026-09-14 16:19:14 | INFO | train rank=0 | step=1780/50000 progress=3.6% loss=0.175237 lr=7.116000e-05 lr_backbone=7.116000e-05 elapsed=01:48:52 avg_step=3.67s recent_step=4.50s eta=49:09:38 +2026-09-14 16:19:40 | INFO | train rank=0 | step=1790/50000 progress=3.6% loss=0.167444 lr=7.156000e-05 lr_backbone=7.156000e-05 elapsed=01:49:18 avg_step=3.66s recent_step=2.55s eta=49:03:58 +2026-09-14 16:20:20 | INFO | train rank=0 | step=1800/50000 progress=3.6% loss=0.155058 lr=7.196000e-05 lr_backbone=7.196000e-05 elapsed=01:49:58 avg_step=3.67s recent_step=3.98s eta=49:04:47 +2026-09-14 16:20:54 | INFO | train rank=0 | step=1810/50000 progress=3.6% loss=0.155002 lr=7.236000e-05 lr_backbone=7.236000e-05 elapsed=01:50:32 avg_step=3.66s recent_step=3.46s eta=49:03:16 +2026-09-14 16:21:40 | INFO | train rank=0 | step=1820/50000 progress=3.6% loss=0.138843 lr=7.276000e-05 lr_backbone=7.276000e-05 elapsed=01:51:18 avg_step=3.67s recent_step=4.52s eta=49:06:26 +2026-09-14 16:22:10 | INFO | train rank=0 | step=1830/50000 progress=3.7% loss=0.172610 lr=7.316000e-05 lr_backbone=7.316000e-05 elapsed=01:51:48 avg_step=3.67s recent_step=2.99s eta=49:02:51 +2026-09-14 16:22:53 | INFO | train rank=0 | step=1840/50000 progress=3.7% loss=0.159027 lr=7.356000e-05 lr_backbone=7.356000e-05 elapsed=01:52:31 avg_step=3.67s recent_step=4.30s eta=49:05:00 +2026-09-14 16:23:34 | INFO | train rank=0 | step=1850/50000 progress=3.7% loss=0.178782 lr=7.396000e-05 lr_backbone=7.396000e-05 elapsed=01:53:12 avg_step=3.67s recent_step=4.16s eta=49:06:30 +2026-09-14 16:24:05 | INFO | train rank=0 | step=1860/50000 progress=3.7% loss=0.165892 lr=7.436000e-05 lr_backbone=7.436000e-05 elapsed=01:53:43 avg_step=3.67s recent_step=3.13s eta=49:03:33 +2026-09-14 16:24:37 | INFO | train rank=0 | step=1870/50000 progress=3.7% loss=0.149154 lr=7.476000e-05 lr_backbone=7.476000e-05 elapsed=01:54:15 avg_step=3.67s recent_step=3.21s eta=49:00:58 +2026-09-14 16:25:23 | INFO | train rank=0 | step=1880/50000 progress=3.8% loss=0.153459 lr=7.516000e-05 lr_backbone=7.516000e-05 elapsed=01:55:01 avg_step=3.67s recent_step=4.58s eta=49:04:16 +2026-09-14 16:25:52 | INFO | train rank=0 | step=1890/50000 progress=3.8% loss=0.158634 lr=7.556000e-05 lr_backbone=7.556000e-05 elapsed=01:55:30 avg_step=3.67s recent_step=2.91s eta=49:00:27 +2026-09-14 16:26:24 | INFO | train rank=0 | step=1900/50000 progress=3.8% loss=0.178513 lr=7.596000e-05 lr_backbone=7.596000e-05 elapsed=01:56:02 avg_step=3.66s recent_step=3.18s eta=48:57:47 +2026-09-14 16:26:57 | INFO | train rank=0 | step=1910/50000 progress=3.8% loss=0.203602 lr=7.636000e-05 lr_backbone=7.636000e-05 elapsed=01:56:35 avg_step=3.66s recent_step=3.25s eta=48:55:26 +2026-09-14 16:27:41 | INFO | train rank=0 | step=1920/50000 progress=3.8% loss=0.162281 lr=7.676000e-05 lr_backbone=7.676000e-05 elapsed=01:57:19 avg_step=3.67s recent_step=4.42s eta=48:58:00 +2026-09-14 16:28:11 | INFO | train rank=0 | step=1930/50000 progress=3.9% loss=0.159196 lr=7.716000e-05 lr_backbone=7.716000e-05 elapsed=01:57:49 avg_step=3.66s recent_step=3.05s eta=48:54:50 +2026-09-14 16:28:47 | INFO | train rank=0 | step=1940/50000 progress=3.9% loss=0.172746 lr=7.756000e-05 lr_backbone=7.756000e-05 elapsed=01:58:25 avg_step=3.66s recent_step=3.57s eta=48:53:49 +2026-09-14 16:29:14 | INFO | train rank=0 | step=1950/50000 progress=3.9% loss=0.168617 lr=7.796000e-05 lr_backbone=7.796000e-05 elapsed=01:58:52 avg_step=3.66s recent_step=2.67s eta=48:49:06 +2026-09-14 16:30:03 | INFO | train rank=0 | step=1960/50000 progress=3.9% loss=0.149344 lr=7.836000e-05 lr_backbone=7.836000e-05 elapsed=01:59:41 avg_step=3.66s recent_step=4.91s eta=48:53:37 +2026-09-14 16:30:36 | INFO | train rank=0 | step=1970/50000 progress=3.9% loss=0.136905 lr=7.876000e-05 lr_backbone=7.876000e-05 elapsed=02:00:14 avg_step=3.66s recent_step=3.28s eta=48:51:28 +2026-09-14 16:31:15 | INFO | train rank=0 | step=1980/50000 progress=4.0% loss=0.164196 lr=7.916000e-05 lr_backbone=7.916000e-05 elapsed=02:00:53 avg_step=3.66s recent_step=3.97s eta=48:52:05 +2026-09-14 16:31:52 | INFO | train rank=0 | step=1990/50000 progress=4.0% loss=0.209080 lr=7.956000e-05 lr_backbone=7.956000e-05 elapsed=02:01:30 avg_step=3.66s recent_step=3.66s eta=48:51:27 +2026-09-14 16:32:40 | INFO | train rank=0 | step=2000/50000 progress=4.0% loss=0.177012 lr=7.996000e-05 lr_backbone=7.996000e-05 elapsed=02:02:18 avg_step=3.67s recent_step=4.76s eta=48:55:14 +2026-09-14 16:33:14 | INFO | train rank=0 | step=2010/50000 progress=4.0% loss=0.146109 lr=8.036000e-05 lr_backbone=8.036000e-05 elapsed=02:02:52 avg_step=3.67s recent_step=3.49s eta=48:53:54 +2026-09-14 16:33:38 | INFO | train rank=0 | step=2020/50000 progress=4.0% loss=0.211169 lr=8.076000e-05 lr_backbone=8.076000e-05 elapsed=02:03:16 avg_step=3.66s recent_step=2.36s eta=48:48:06 +2026-09-14 16:34:19 | INFO | train rank=0 | step=2030/50000 progress=4.1% loss=0.185117 lr=8.116000e-05 lr_backbone=8.116000e-05 elapsed=02:03:57 avg_step=3.66s recent_step=4.05s eta=48:49:02 +2026-09-14 16:35:13 | INFO | train rank=0 | step=2040/50000 progress=4.1% loss=0.159783 lr=8.156000e-05 lr_backbone=8.156000e-05 elapsed=02:04:51 avg_step=3.67s recent_step=5.44s eta=48:55:22 +2026-09-14 16:35:48 | INFO | train rank=0 | step=2050/50000 progress=4.1% loss=0.154371 lr=8.196000e-05 lr_backbone=8.196000e-05 elapsed=02:05:26 avg_step=3.67s recent_step=3.55s eta=48:54:18 +2026-09-14 16:36:18 | INFO | train rank=0 | step=2060/50000 progress=4.1% loss=0.166174 lr=8.236000e-05 lr_backbone=8.236000e-05 elapsed=02:05:56 avg_step=3.67s recent_step=2.98s eta=48:51:00 +2026-09-14 16:36:50 | INFO | train rank=0 | step=2070/50000 progress=4.1% loss=0.153190 lr=8.276000e-05 lr_backbone=8.276000e-05 elapsed=02:06:28 avg_step=3.67s recent_step=3.19s eta=48:48:32 +2026-09-14 16:37:36 | INFO | train rank=0 | step=2080/50000 progress=4.2% loss=0.169419 lr=8.316000e-05 lr_backbone=8.316000e-05 elapsed=02:07:14 avg_step=3.67s recent_step=4.57s eta=48:51:24 +2026-09-14 16:37:58 | INFO | train rank=0 | step=2090/50000 progress=4.2% loss=0.148718 lr=8.356000e-05 lr_backbone=8.356000e-05 elapsed=02:07:36 avg_step=3.66s recent_step=2.21s eta=48:45:12 +2026-09-14 16:38:27 | INFO | train rank=0 | step=2100/50000 progress=4.2% loss=0.158239 lr=8.396000e-05 lr_backbone=8.396000e-05 elapsed=02:08:05 avg_step=3.66s recent_step=2.93s eta=48:41:49 +2026-09-14 16:38:58 | INFO | train rank=0 | step=2110/50000 progress=4.2% loss=0.189954 lr=8.436000e-05 lr_backbone=8.436000e-05 elapsed=02:08:36 avg_step=3.66s recent_step=3.06s eta=48:38:56 +2026-09-14 16:39:50 | INFO | train rank=0 | step=2120/50000 progress=4.2% loss=0.143484 lr=8.476000e-05 lr_backbone=8.476000e-05 elapsed=02:09:28 avg_step=3.66s recent_step=5.18s eta=48:44:03 +2026-09-14 16:40:20 | INFO | train rank=0 | step=2130/50000 progress=4.3% loss=0.176652 lr=8.516000e-05 lr_backbone=8.516000e-05 elapsed=02:09:58 avg_step=3.66s recent_step=3.05s eta=48:41:08 +2026-09-14 16:41:00 | INFO | train rank=0 | step=2140/50000 progress=4.3% loss=0.154858 lr=8.556000e-05 lr_backbone=8.556000e-05 elapsed=02:10:39 avg_step=3.66s recent_step=4.04s eta=48:41:55 +2026-09-14 16:41:33 | INFO | train rank=0 | step=2150/50000 progress=4.3% loss=0.169065 lr=8.596000e-05 lr_backbone=8.596000e-05 elapsed=02:11:11 avg_step=3.66s recent_step=3.28s eta=48:39:53 +2026-09-14 16:42:06 | INFO | train rank=0 | step=2160/50000 progress=4.3% loss=0.207290 lr=8.636000e-05 lr_backbone=8.636000e-05 elapsed=02:11:44 avg_step=3.66s recent_step=3.24s eta=48:37:43 +2026-09-14 16:42:38 | INFO | train rank=0 | step=2170/50000 progress=4.3% loss=0.160640 lr=8.676000e-05 lr_backbone=8.676000e-05 elapsed=02:12:16 avg_step=3.66s recent_step=3.26s eta=48:35:39 +2026-09-14 16:43:11 | INFO | train rank=0 | step=2180/50000 progress=4.4% loss=0.160348 lr=8.716000e-05 lr_backbone=8.716000e-05 elapsed=02:12:49 avg_step=3.66s recent_step=3.23s eta=48:33:29 +2026-09-14 16:43:48 | INFO | train rank=0 | step=2190/50000 progress=4.4% loss=0.159066 lr=8.756000e-05 lr_backbone=8.756000e-05 elapsed=02:13:26 avg_step=3.66s recent_step=3.69s eta=48:33:00 +2026-09-14 16:44:30 | INFO | train rank=0 | step=2200/50000 progress=4.4% loss=0.153691 lr=8.796000e-05 lr_backbone=8.796000e-05 elapsed=02:14:08 avg_step=3.66s recent_step=4.26s eta=48:34:35 +2026-09-14 16:45:04 | INFO | train rank=0 | step=2210/50000 progress=4.4% loss=0.141104 lr=8.836000e-05 lr_backbone=8.836000e-05 elapsed=02:14:42 avg_step=3.66s recent_step=3.34s eta=48:32:50 +2026-09-14 16:45:40 | INFO | train rank=0 | step=2220/50000 progress=4.4% loss=0.148912 lr=8.876000e-05 lr_backbone=8.876000e-05 elapsed=02:15:18 avg_step=3.66s recent_step=3.62s eta=48:32:06 +2026-09-14 16:46:27 | INFO | train rank=0 | step=2230/50000 progress=4.5% loss=0.164915 lr=8.916000e-05 lr_backbone=8.916000e-05 elapsed=02:16:05 avg_step=3.66s recent_step=4.69s eta=48:35:10 +2026-09-14 16:47:11 | INFO | train rank=0 | step=2240/50000 progress=4.5% loss=0.181479 lr=8.956000e-05 lr_backbone=8.956000e-05 elapsed=02:16:50 avg_step=3.67s recent_step=4.48s eta=48:37:29 +2026-09-14 16:47:40 | INFO | train rank=0 | step=2250/50000 progress=4.5% loss=0.173935 lr=8.996000e-05 lr_backbone=8.996000e-05 elapsed=02:17:18 avg_step=3.66s recent_step=2.90s eta=48:34:08 +2026-09-14 16:48:09 | INFO | train rank=0 | step=2260/50000 progress=4.5% loss=0.188523 lr=9.036000e-05 lr_backbone=9.036000e-05 elapsed=02:17:47 avg_step=3.66s recent_step=2.81s eta=48:30:33 +2026-09-14 16:48:45 | INFO | train rank=0 | step=2270/50000 progress=4.5% loss=0.167035 lr=9.076000e-05 lr_backbone=9.076000e-05 elapsed=02:18:24 avg_step=3.66s recent_step=3.69s eta=48:30:03 +2026-09-14 16:49:36 | INFO | train rank=0 | step=2280/50000 progress=4.6% loss=0.171725 lr=9.116000e-05 lr_backbone=9.116000e-05 elapsed=02:19:14 avg_step=3.66s recent_step=5.05s eta=48:34:19 +2026-09-14 16:50:08 | INFO | train rank=0 | step=2290/50000 progress=4.6% loss=0.157613 lr=9.156000e-05 lr_backbone=9.156000e-05 elapsed=02:19:46 avg_step=3.66s recent_step=3.23s eta=48:32:11 +2026-09-14 16:50:42 | INFO | train rank=0 | step=2300/50000 progress=4.6% loss=0.193127 lr=9.196000e-05 lr_backbone=9.196000e-05 elapsed=02:20:20 avg_step=3.66s recent_step=3.40s eta=48:30:41 +2026-09-14 16:51:18 | INFO | train rank=0 | step=2310/50000 progress=4.6% loss=0.203043 lr=9.236000e-05 lr_backbone=9.236000e-05 elapsed=02:20:56 avg_step=3.66s recent_step=3.61s eta=48:29:54 +2026-09-14 16:51:57 | INFO | train rank=0 | step=2320/50000 progress=4.6% loss=0.166052 lr=9.276000e-05 lr_backbone=9.276000e-05 elapsed=02:21:35 avg_step=3.66s recent_step=3.87s eta=48:30:00 +2026-09-14 16:52:33 | INFO | train rank=0 | step=2330/50000 progress=4.7% loss=0.175234 lr=9.316000e-05 lr_backbone=9.316000e-05 elapsed=02:22:11 avg_step=3.66s recent_step=3.62s eta=48:29:16 +2026-09-14 16:53:12 | INFO | train rank=0 | step=2340/50000 progress=4.7% loss=0.169714 lr=9.356000e-05 lr_backbone=9.356000e-05 elapsed=02:22:50 avg_step=3.66s recent_step=3.81s eta=48:29:11 +2026-09-14 16:53:51 | INFO | train rank=0 | step=2350/50000 progress=4.7% loss=0.198438 lr=9.396000e-05 lr_backbone=9.396000e-05 elapsed=02:23:29 avg_step=3.66s recent_step=3.98s eta=48:29:39 +2026-09-14 16:54:25 | INFO | train rank=0 | step=2360/50000 progress=4.7% loss=0.143198 lr=9.436000e-05 lr_backbone=9.436000e-05 elapsed=02:24:03 avg_step=3.66s recent_step=3.32s eta=48:27:54 +2026-09-14 16:55:14 | INFO | train rank=0 | step=2370/50000 progress=4.7% loss=0.183860 lr=9.476000e-05 lr_backbone=9.476000e-05 elapsed=02:24:52 avg_step=3.67s recent_step=4.95s eta=48:31:35 +2026-09-14 16:55:41 | INFO | train rank=0 | step=2380/50000 progress=4.8% loss=0.159244 lr=9.516000e-05 lr_backbone=9.516000e-05 elapsed=02:25:19 avg_step=3.66s recent_step=2.65s eta=48:27:34 +2026-09-14 16:56:22 | INFO | train rank=0 | step=2390/50000 progress=4.8% loss=0.162534 lr=9.556000e-05 lr_backbone=9.556000e-05 elapsed=02:26:00 avg_step=3.67s recent_step=4.19s eta=48:28:42 +2026-09-14 16:56:57 | INFO | train rank=0 | step=2400/50000 progress=4.8% loss=0.168788 lr=9.596000e-05 lr_backbone=9.596000e-05 elapsed=02:26:35 avg_step=3.66s recent_step=3.49s eta=48:27:31 +2026-09-14 16:57:30 | INFO | train rank=0 | step=2410/50000 progress=4.8% loss=0.171913 lr=9.636000e-05 lr_backbone=9.636000e-05 elapsed=02:27:08 avg_step=3.66s recent_step=3.26s eta=48:25:34 +2026-09-14 16:58:01 | INFO | train rank=0 | step=2420/50000 progress=4.8% loss=0.170796 lr=9.676000e-05 lr_backbone=9.676000e-05 elapsed=02:27:39 avg_step=3.66s recent_step=3.15s eta=48:23:16 +2026-09-14 16:58:39 | INFO | train rank=0 | step=2430/50000 progress=4.9% loss=0.152053 lr=9.716000e-05 lr_backbone=9.716000e-05 elapsed=02:28:17 avg_step=3.66s recent_step=3.77s eta=48:23:01 +2026-09-14 16:59:19 | INFO | train rank=0 | step=2440/50000 progress=4.9% loss=0.131939 lr=9.756000e-05 lr_backbone=9.756000e-05 elapsed=02:28:57 avg_step=3.66s recent_step=3.97s eta=48:23:24 +2026-09-14 17:00:08 | INFO | train rank=0 | step=2450/50000 progress=4.9% loss=0.150363 lr=9.796000e-05 lr_backbone=9.796000e-05 elapsed=02:29:46 avg_step=3.67s recent_step=4.89s eta=48:26:46 +2026-09-14 17:00:38 | INFO | train rank=0 | step=2460/50000 progress=4.9% loss=0.165130 lr=9.836000e-05 lr_backbone=9.836000e-05 elapsed=02:30:16 avg_step=3.67s recent_step=3.04s eta=48:24:08 +2026-09-14 17:01:19 | INFO | train rank=0 | step=2470/50000 progress=4.9% loss=0.177373 lr=9.876000e-05 lr_backbone=9.876000e-05 elapsed=02:30:57 avg_step=3.67s recent_step=4.10s eta=48:24:55 +2026-09-14 17:01:55 | INFO | train rank=0 | step=2480/50000 progress=5.0% loss=0.162242 lr=9.916000e-05 lr_backbone=9.916000e-05 elapsed=02:31:33 avg_step=3.67s recent_step=3.55s eta=48:23:55 +2026-09-14 17:02:31 | INFO | train rank=0 | step=2490/50000 progress=5.0% loss=0.142040 lr=9.956000e-05 lr_backbone=9.956000e-05 elapsed=02:32:09 avg_step=3.67s recent_step=3.67s eta=48:23:19 +2026-09-14 17:03:07 | INFO | train rank=0 | step=2500/50000 progress=5.0% loss=0.158152 lr=9.996000e-05 lr_backbone=9.996000e-05 elapsed=02:32:45 avg_step=3.67s recent_step=3.54s eta=48:22:18 +2026-09-14 17:03:31 | INFO | train rank=0 | step=2510/50000 progress=5.0% loss=0.181933 lr=9.999999e-05 lr_backbone=9.999999e-05 elapsed=02:33:09 avg_step=3.66s recent_step=2.40s eta=48:17:43 +2026-09-14 17:04:06 | INFO | train rank=0 | step=2520/50000 progress=5.0% loss=0.136752 lr=9.999996e-05 lr_backbone=9.999996e-05 elapsed=02:33:44 avg_step=3.66s recent_step=3.49s eta=48:16:35 +2026-09-14 17:04:50 | INFO | train rank=0 | step=2530/50000 progress=5.1% loss=0.201310 lr=9.999990e-05 lr_backbone=9.999990e-05 elapsed=02:34:28 avg_step=3.66s recent_step=4.39s eta=48:18:16 +2026-09-14 17:05:25 | INFO | train rank=0 | step=2540/50000 progress=5.1% loss=0.166289 lr=9.999983e-05 lr_backbone=9.999983e-05 elapsed=02:35:03 avg_step=3.66s recent_step=3.50s eta=48:17:08 +2026-09-14 17:05:55 | INFO | train rank=0 | step=2550/50000 progress=5.1% loss=0.152031 lr=9.999973e-05 lr_backbone=9.999973e-05 elapsed=02:35:33 avg_step=3.66s recent_step=3.00s eta=48:14:28 +2026-09-14 17:06:40 | INFO | train rank=0 | step=2560/50000 progress=5.1% loss=0.155723 lr=9.999961e-05 lr_backbone=9.999961e-05 elapsed=02:36:18 avg_step=3.66s recent_step=4.58s eta=48:16:42 +2026-09-14 17:07:20 | INFO | train rank=0 | step=2570/50000 progress=5.1% loss=0.160116 lr=9.999946e-05 lr_backbone=9.999946e-05 elapsed=02:36:58 avg_step=3.66s recent_step=3.99s eta=48:17:06 +2026-09-14 17:07:51 | INFO | train rank=0 | step=2580/50000 progress=5.2% loss=0.167382 lr=9.999930e-05 lr_backbone=9.999930e-05 elapsed=02:37:29 avg_step=3.66s recent_step=3.09s eta=48:14:42 +2026-09-14 17:08:20 | INFO | train rank=0 | step=2590/50000 progress=5.2% loss=0.176995 lr=9.999911e-05 lr_backbone=9.999911e-05 elapsed=02:37:58 avg_step=3.66s recent_step=2.93s eta=48:11:52 +2026-09-14 17:08:59 | INFO | train rank=0 | step=2600/50000 progress=5.2% loss=0.162869 lr=9.999891e-05 lr_backbone=9.999891e-05 elapsed=02:38:37 avg_step=3.66s recent_step=3.88s eta=48:11:56 +2026-09-14 17:09:40 | INFO | train rank=0 | step=2610/50000 progress=5.2% loss=0.141837 lr=9.999868e-05 lr_backbone=9.999868e-05 elapsed=02:39:18 avg_step=3.66s recent_step=4.08s eta=48:12:37 +2026-09-14 17:10:08 | INFO | train rank=0 | step=2620/50000 progress=5.2% loss=0.145123 lr=9.999843e-05 lr_backbone=9.999843e-05 elapsed=02:39:46 avg_step=3.66s recent_step=2.83s eta=48:09:29 +2026-09-14 17:10:42 | INFO | train rank=0 | step=2630/50000 progress=5.3% loss=0.162799 lr=9.999815e-05 lr_backbone=9.999815e-05 elapsed=02:40:20 avg_step=3.66s recent_step=3.35s eta=48:07:58 +2026-09-14 17:11:24 | INFO | train rank=0 | step=2640/50000 progress=5.3% loss=0.153864 lr=9.999786e-05 lr_backbone=9.999786e-05 elapsed=02:41:02 avg_step=3.66s recent_step=4.17s eta=48:08:52 +2026-09-14 17:12:00 | INFO | train rank=0 | step=2650/50000 progress=5.3% loss=0.147769 lr=9.999754e-05 lr_backbone=9.999754e-05 elapsed=02:41:38 avg_step=3.66s recent_step=3.65s eta=48:08:14 +2026-09-14 17:12:33 | INFO | train rank=0 | step=2660/50000 progress=5.3% loss=0.142824 lr=9.999720e-05 lr_backbone=9.999720e-05 elapsed=02:42:11 avg_step=3.66s recent_step=3.33s eta=48:06:39 +2026-09-14 17:13:13 | INFO | train rank=0 | step=2670/50000 progress=5.3% loss=0.133482 lr=9.999684e-05 lr_backbone=9.999684e-05 elapsed=02:42:51 avg_step=3.66s recent_step=3.96s eta=48:06:56 +2026-09-14 17:13:54 | INFO | train rank=0 | step=2680/50000 progress=5.4% loss=0.164602 lr=9.999646e-05 lr_backbone=9.999646e-05 elapsed=02:43:32 avg_step=3.66s recent_step=4.05s eta=48:07:29 +2026-09-14 17:14:22 | INFO | train rank=0 | step=2690/50000 progress=5.4% loss=0.166211 lr=9.999605e-05 lr_backbone=9.999605e-05 elapsed=02:44:00 avg_step=3.66s recent_step=2.81s eta=48:04:22 +2026-09-14 17:14:56 | INFO | train rank=0 | step=2700/50000 progress=5.4% loss=0.133816 lr=9.999563e-05 lr_backbone=9.999563e-05 elapsed=02:44:34 avg_step=3.66s recent_step=3.45s eta=48:03:09 +2026-09-14 17:15:40 | INFO | train rank=0 | step=2710/50000 progress=5.4% loss=0.149728 lr=9.999518e-05 lr_backbone=9.999518e-05 elapsed=02:45:18 avg_step=3.66s recent_step=4.39s eta=48:04:41 +2026-09-14 17:16:18 | INFO | train rank=0 | step=2720/50000 progress=5.4% loss=0.139175 lr=9.999471e-05 lr_backbone=9.999471e-05 elapsed=02:45:56 avg_step=3.66s recent_step=3.78s eta=48:04:26 +2026-09-14 17:16:54 | INFO | train rank=0 | step=2730/50000 progress=5.5% loss=0.137127 lr=9.999422e-05 lr_backbone=9.999422e-05 elapsed=02:46:32 avg_step=3.66s recent_step=3.62s eta=48:03:42 +2026-09-14 17:17:27 | INFO | train rank=0 | step=2740/50000 progress=5.5% loss=0.162997 lr=9.999370e-05 lr_backbone=9.999370e-05 elapsed=02:47:05 avg_step=3.66s recent_step=3.30s eta=48:02:05 +2026-09-14 17:18:16 | INFO | train rank=0 | step=2750/50000 progress=5.5% loss=0.166192 lr=9.999317e-05 lr_backbone=9.999317e-05 elapsed=02:47:54 avg_step=3.66s recent_step=4.90s eta=48:05:01 +2026-09-14 17:18:47 | INFO | train rank=0 | step=2760/50000 progress=5.5% loss=0.125477 lr=9.999261e-05 lr_backbone=9.999261e-05 elapsed=02:48:25 avg_step=3.66s recent_step=3.05s eta=48:02:39 +2026-09-14 17:19:24 | INFO | train rank=0 | step=2770/50000 progress=5.5% loss=0.179544 lr=9.999203e-05 lr_backbone=9.999203e-05 elapsed=02:49:02 avg_step=3.66s recent_step=3.71s eta=48:02:10 +2026-09-14 17:20:01 | INFO | train rank=0 | step=2780/50000 progress=5.6% loss=0.155384 lr=9.999143e-05 lr_backbone=9.999143e-05 elapsed=02:49:39 avg_step=3.66s recent_step=3.76s eta=48:01:50 +2026-09-14 17:20:52 | INFO | train rank=0 | step=2790/50000 progress=5.6% loss=0.129787 lr=9.999080e-05 lr_backbone=9.999080e-05 elapsed=02:50:30 avg_step=3.67s recent_step=5.07s eta=48:05:12 +2026-09-14 17:21:23 | INFO | train rank=0 | step=2800/50000 progress=5.6% loss=0.121887 lr=9.999016e-05 lr_backbone=9.999016e-05 elapsed=02:51:01 avg_step=3.66s recent_step=3.13s eta=48:03:05 +2026-09-14 17:21:52 | INFO | train rank=0 | step=2810/50000 progress=5.6% loss=0.166771 lr=9.998949e-05 lr_backbone=9.998949e-05 elapsed=02:51:30 avg_step=3.66s recent_step=2.87s eta=48:00:15 +2026-09-14 17:22:29 | INFO | train rank=0 | step=2820/50000 progress=5.6% loss=0.125396 lr=9.998880e-05 lr_backbone=9.998880e-05 elapsed=02:52:07 avg_step=3.66s recent_step=3.71s eta=47:59:46 +2026-09-14 17:23:17 | INFO | train rank=0 | step=2830/50000 progress=5.7% loss=0.157924 lr=9.998809e-05 lr_backbone=9.998809e-05 elapsed=02:52:55 avg_step=3.67s recent_step=4.76s eta=48:02:13 +2026-09-14 17:23:49 | INFO | train rank=0 | step=2840/50000 progress=5.7% loss=0.150975 lr=9.998736e-05 lr_backbone=9.998736e-05 elapsed=02:53:27 avg_step=3.66s recent_step=3.24s eta=48:00:25 +2026-09-14 17:24:24 | INFO | train rank=0 | step=2850/50000 progress=5.7% loss=0.204926 lr=9.998660e-05 lr_backbone=9.998660e-05 elapsed=02:54:02 avg_step=3.66s recent_step=3.51s eta=47:59:24 +2026-09-14 17:24:57 | INFO | train rank=0 | step=2860/50000 progress=5.7% loss=0.145323 lr=9.998583e-05 lr_backbone=9.998583e-05 elapsed=02:54:35 avg_step=3.66s recent_step=3.27s eta=47:57:42 +2026-09-14 17:25:47 | INFO | train rank=0 | step=2870/50000 progress=5.7% loss=0.165659 lr=9.998503e-05 lr_backbone=9.998503e-05 elapsed=02:55:25 avg_step=3.67s recent_step=4.98s eta=48:00:42 +2026-09-14 17:26:22 | INFO | train rank=0 | step=2880/50000 progress=5.8% loss=0.159364 lr=9.998421e-05 lr_backbone=9.998421e-05 elapsed=02:56:00 avg_step=3.67s recent_step=3.49s eta=47:59:36 +2026-09-14 17:26:50 | INFO | train rank=0 | step=2890/50000 progress=5.8% loss=0.143652 lr=9.998337e-05 lr_backbone=9.998337e-05 elapsed=02:56:28 avg_step=3.66s recent_step=2.81s eta=47:56:39 +2026-09-14 17:27:19 | INFO | train rank=0 | step=2900/50000 progress=5.8% loss=0.161389 lr=9.998250e-05 lr_backbone=9.998250e-05 elapsed=02:56:57 avg_step=3.66s recent_step=2.88s eta=47:53:55 +2026-09-14 17:28:10 | INFO | train rank=0 | step=2910/50000 progress=5.8% loss=0.141821 lr=9.998162e-05 lr_backbone=9.998162e-05 elapsed=02:57:48 avg_step=3.67s recent_step=5.15s eta=47:57:20 +2026-09-14 17:28:39 | INFO | train rank=0 | step=2920/50000 progress=5.8% loss=0.158849 lr=9.998071e-05 lr_backbone=9.998071e-05 elapsed=02:58:17 avg_step=3.66s recent_step=2.87s eta=47:54:35 +2026-09-14 17:29:13 | INFO | train rank=0 | step=2930/50000 progress=5.9% loss=0.116333 lr=9.997978e-05 lr_backbone=9.997978e-05 elapsed=02:58:51 avg_step=3.66s recent_step=3.40s eta=47:53:16 +2026-09-14 17:29:43 | INFO | train rank=0 | step=2940/50000 progress=5.9% loss=0.159840 lr=9.997883e-05 lr_backbone=9.997883e-05 elapsed=02:59:21 avg_step=3.66s recent_step=3.01s eta=47:50:56 +2026-09-14 17:30:34 | INFO | train rank=0 | step=2950/50000 progress=5.9% loss=0.147406 lr=9.997786e-05 lr_backbone=9.997786e-05 elapsed=03:00:12 avg_step=3.67s recent_step=5.06s eta=47:54:03 +2026-09-14 17:31:00 | INFO | train rank=0 | step=2960/50000 progress=5.9% loss=0.144964 lr=9.997686e-05 lr_backbone=9.997686e-05 elapsed=03:00:38 avg_step=3.66s recent_step=2.64s eta=47:50:43 +2026-09-14 17:31:29 | INFO | train rank=0 | step=2970/50000 progress=5.9% loss=0.159267 lr=9.997584e-05 lr_backbone=9.997584e-05 elapsed=03:01:07 avg_step=3.66s recent_step=2.95s eta=47:48:14 +2026-09-14 17:32:05 | INFO | train rank=0 | step=2980/50000 progress=6.0% loss=0.165141 lr=9.997481e-05 lr_backbone=9.997481e-05 elapsed=03:01:44 avg_step=3.66s recent_step=3.60s eta=47:47:29 +2026-09-14 17:33:03 | INFO | train rank=0 | step=2990/50000 progress=6.0% loss=0.167668 lr=9.997375e-05 lr_backbone=9.997375e-05 elapsed=03:02:41 avg_step=3.67s recent_step=5.77s eta=47:52:23 +2026-09-14 17:33:36 | INFO | train rank=0 | step=3000/50000 progress=6.0% loss=0.173619 lr=9.997266e-05 lr_backbone=9.997266e-05 elapsed=03:03:14 avg_step=3.66s recent_step=3.30s eta=47:50:50 +2026-09-14 17:34:07 | INFO | train rank=0 | step=3010/50000 progress=6.0% loss=0.148448 lr=9.997156e-05 lr_backbone=9.997156e-05 elapsed=03:03:45 avg_step=3.66s recent_step=3.12s eta=47:48:48 +2026-09-14 17:34:41 | INFO | train rank=0 | step=3020/50000 progress=6.0% loss=0.178203 lr=9.997043e-05 lr_backbone=9.997043e-05 elapsed=03:04:19 avg_step=3.66s recent_step=3.40s eta=47:47:31 +2026-09-14 17:35:33 | INFO | train rank=0 | step=3030/50000 progress=6.1% loss=0.113289 lr=9.996928e-05 lr_backbone=9.996928e-05 elapsed=03:05:11 avg_step=3.67s recent_step=5.14s eta=47:50:43 +2026-09-14 17:35:59 | INFO | train rank=0 | step=3040/50000 progress=6.1% loss=0.172410 lr=9.996811e-05 lr_backbone=9.996811e-05 elapsed=03:05:37 avg_step=3.66s recent_step=2.66s eta=47:47:31 +2026-09-14 17:36:38 | INFO | train rank=0 | step=3050/50000 progress=6.1% loss=0.137931 lr=9.996692e-05 lr_backbone=9.996692e-05 elapsed=03:06:16 avg_step=3.66s recent_step=3.84s eta=47:47:21 +2026-09-14 17:37:14 | INFO | train rank=0 | step=3060/50000 progress=6.1% loss=0.172474 lr=9.996571e-05 lr_backbone=9.996571e-05 elapsed=03:06:52 avg_step=3.66s recent_step=3.59s eta=47:46:33 +2026-09-14 17:38:04 | INFO | train rank=0 | step=3070/50000 progress=6.1% loss=0.160466 lr=9.996447e-05 lr_backbone=9.996447e-05 elapsed=03:07:42 avg_step=3.67s recent_step=4.99s eta=47:49:18 +2026-09-14 17:38:41 | INFO | train rank=0 | step=3080/50000 progress=6.2% loss=0.143878 lr=9.996322e-05 lr_backbone=9.996322e-05 elapsed=03:08:19 avg_step=3.67s recent_step=3.76s eta=47:48:55 +2026-09-14 17:39:12 | INFO | train rank=0 | step=3090/50000 progress=6.2% loss=0.173343 lr=9.996194e-05 lr_backbone=9.996194e-05 elapsed=03:08:50 avg_step=3.67s recent_step=3.11s eta=47:46:53 +2026-09-14 17:39:47 | INFO | train rank=0 | step=3100/50000 progress=6.2% loss=0.141916 lr=9.996064e-05 lr_backbone=9.996064e-05 elapsed=03:09:25 avg_step=3.67s recent_step=3.51s eta=47:45:53 +2026-09-14 17:40:25 | INFO | train rank=0 | step=3110/50000 progress=6.2% loss=0.140081 lr=9.995931e-05 lr_backbone=9.995931e-05 elapsed=03:10:03 avg_step=3.67s recent_step=3.82s eta=47:45:39 +2026-09-14 17:40:58 | INFO | train rank=0 | step=3120/50000 progress=6.2% loss=0.151890 lr=9.995797e-05 lr_backbone=9.995797e-05 elapsed=03:10:36 avg_step=3.67s recent_step=3.26s eta=47:44:01 +2026-09-14 17:41:32 | INFO | train rank=0 | step=3130/50000 progress=6.3% loss=0.162619 lr=9.995660e-05 lr_backbone=9.995660e-05 elapsed=03:11:10 avg_step=3.66s recent_step=3.37s eta=47:42:40 +2026-09-14 17:42:02 | INFO | train rank=0 | step=3140/50000 progress=6.3% loss=0.116416 lr=9.995521e-05 lr_backbone=9.995521e-05 elapsed=03:11:40 avg_step=3.66s recent_step=3.02s eta=47:40:27 +2026-09-14 17:42:48 | INFO | train rank=0 | step=3150/50000 progress=6.3% loss=0.138294 lr=9.995380e-05 lr_backbone=9.995380e-05 elapsed=03:12:26 avg_step=3.67s recent_step=4.65s eta=47:42:18 +2026-09-14 17:43:26 | INFO | train rank=0 | step=3160/50000 progress=6.3% loss=0.185505 lr=9.995237e-05 lr_backbone=9.995237e-05 elapsed=03:13:04 avg_step=3.67s recent_step=3.74s eta=47:41:52 +2026-09-14 17:43:50 | INFO | train rank=0 | step=3170/50000 progress=6.3% loss=0.177768 lr=9.995092e-05 lr_backbone=9.995092e-05 elapsed=03:13:28 avg_step=3.66s recent_step=2.39s eta=47:38:07 +2026-09-14 17:44:23 | INFO | train rank=0 | step=3180/50000 progress=6.4% loss=0.136257 lr=9.994944e-05 lr_backbone=9.994944e-05 elapsed=03:14:01 avg_step=3.66s recent_step=3.35s eta=47:36:45 +2026-09-14 17:45:12 | INFO | train rank=0 | step=3190/50000 progress=6.4% loss=0.154288 lr=9.994794e-05 lr_backbone=9.994794e-05 elapsed=03:14:50 avg_step=3.66s recent_step=4.85s eta=47:39:03 +2026-09-14 17:45:41 | INFO | train rank=0 | step=3200/50000 progress=6.4% loss=0.157211 lr=9.994642e-05 lr_backbone=9.994642e-05 elapsed=03:15:19 avg_step=3.66s recent_step=2.97s eta=47:36:44 +2026-09-14 17:46:19 | INFO | train rank=0 | step=3210/50000 progress=6.4% loss=0.137847 lr=9.994488e-05 lr_backbone=9.994488e-05 elapsed=03:15:57 avg_step=3.66s recent_step=3.75s eta=47:36:20 +2026-09-14 17:46:53 | INFO | train rank=0 | step=3220/50000 progress=6.4% loss=0.119442 lr=9.994332e-05 lr_backbone=9.994332e-05 elapsed=03:16:31 avg_step=3.66s recent_step=3.38s eta=47:35:02 +2026-09-14 17:47:40 | INFO | train rank=0 | step=3230/50000 progress=6.5% loss=0.166622 lr=9.994173e-05 lr_backbone=9.994173e-05 elapsed=03:17:18 avg_step=3.67s recent_step=4.74s eta=47:37:02 +2026-09-14 17:48:06 | INFO | train rank=0 | step=3240/50000 progress=6.5% loss=0.138192 lr=9.994013e-05 lr_backbone=9.994013e-05 elapsed=03:17:44 avg_step=3.66s recent_step=2.58s eta=47:33:49 +2026-09-14 17:48:34 | INFO | train rank=0 | step=3250/50000 progress=6.5% loss=0.147963 lr=9.993850e-05 lr_backbone=9.993850e-05 elapsed=03:18:12 avg_step=3.66s recent_step=2.85s eta=47:31:15 +2026-09-14 17:49:20 | INFO | train rank=0 | step=3260/50000 progress=6.5% loss=0.156572 lr=9.993685e-05 lr_backbone=9.993685e-05 elapsed=03:18:58 avg_step=3.66s recent_step=4.60s eta=47:32:53 +2026-09-14 17:49:58 | INFO | train rank=0 | step=3270/50000 progress=6.5% loss=0.112946 lr=9.993518e-05 lr_backbone=9.993518e-05 elapsed=03:19:36 avg_step=3.66s recent_step=3.73s eta=47:32:26 +2026-09-14 17:50:34 | INFO | train rank=0 | step=3280/50000 progress=6.6% loss=0.131776 lr=9.993348e-05 lr_backbone=9.993348e-05 elapsed=03:20:12 avg_step=3.66s recent_step=3.63s eta=47:31:45 +2026-09-14 17:51:12 | INFO | train rank=0 | step=3290/50000 progress=6.6% loss=0.128936 lr=9.993176e-05 lr_backbone=9.993176e-05 elapsed=03:20:50 avg_step=3.66s recent_step=3.84s eta=47:31:33 +2026-09-14 17:51:47 | INFO | train rank=0 | step=3300/50000 progress=6.6% loss=0.150917 lr=9.993003e-05 lr_backbone=9.993003e-05 elapsed=03:21:25 avg_step=3.66s recent_step=3.44s eta=47:30:25 +2026-09-14 17:52:45 | INFO | train rank=0 | step=3310/50000 progress=6.6% loss=0.134676 lr=9.992827e-05 lr_backbone=9.992827e-05 elapsed=03:22:23 avg_step=3.67s recent_step=5.79s eta=47:34:49 +2026-09-14 17:53:16 | INFO | train rank=0 | step=3320/50000 progress=6.6% loss=0.172490 lr=9.992649e-05 lr_backbone=9.992649e-05 elapsed=03:22:54 avg_step=3.67s recent_step=3.10s eta=47:32:52 +2026-09-14 17:53:51 | INFO | train rank=0 | step=3330/50000 progress=6.7% loss=0.147270 lr=9.992468e-05 lr_backbone=9.992468e-05 elapsed=03:23:29 avg_step=3.67s recent_step=3.54s eta=47:31:57 +2026-09-14 17:54:30 | INFO | train rank=0 | step=3340/50000 progress=6.7% loss=0.150025 lr=9.992286e-05 lr_backbone=9.992286e-05 elapsed=03:24:08 avg_step=3.67s recent_step=3.87s eta=47:31:50 +2026-09-14 17:55:15 | INFO | train rank=0 | step=3350/50000 progress=6.7% loss=0.130402 lr=9.992101e-05 lr_backbone=9.992101e-05 elapsed=03:24:53 avg_step=3.67s recent_step=4.48s eta=47:33:06 +2026-09-14 17:55:47 | INFO | train rank=0 | step=3360/50000 progress=6.7% loss=0.161426 lr=9.991914e-05 lr_backbone=9.991914e-05 elapsed=03:25:25 avg_step=3.67s recent_step=3.19s eta=47:31:24 +2026-09-14 17:56:17 | INFO | train rank=0 | step=3370/50000 progress=6.7% loss=0.140466 lr=9.991725e-05 lr_backbone=9.991725e-05 elapsed=03:25:55 avg_step=3.67s recent_step=3.00s eta=47:29:15 +2026-09-14 17:56:46 | INFO | train rank=0 | step=3380/50000 progress=6.8% loss=0.144585 lr=9.991534e-05 lr_backbone=9.991534e-05 elapsed=03:26:24 avg_step=3.66s recent_step=2.93s eta=47:26:58 +2026-09-14 17:57:33 | INFO | train rank=0 | step=3390/50000 progress=6.8% loss=0.136314 lr=9.991340e-05 lr_backbone=9.991340e-05 elapsed=03:27:11 avg_step=3.67s recent_step=4.71s eta=47:28:45 +2026-09-14 17:58:04 | INFO | train rank=0 | step=3400/50000 progress=6.8% loss=0.133742 lr=9.991145e-05 lr_backbone=9.991145e-05 elapsed=03:27:42 avg_step=3.67s recent_step=3.09s eta=47:26:50 +2026-09-14 17:58:35 | INFO | train rank=0 | step=3410/50000 progress=6.8% loss=0.128921 lr=9.990947e-05 lr_backbone=9.990947e-05 elapsed=03:28:13 avg_step=3.66s recent_step=3.12s eta=47:24:59 +2026-09-14 17:59:07 | INFO | train rank=0 | step=3420/50000 progress=6.8% loss=0.121125 lr=9.990747e-05 lr_backbone=9.990747e-05 elapsed=03:28:45 avg_step=3.66s recent_step=3.13s eta=47:23:09 +2026-09-14 17:59:43 | INFO | train rank=0 | step=3430/50000 progress=6.9% loss=0.142113 lr=9.990545e-05 lr_backbone=9.990545e-05 elapsed=03:29:21 avg_step=3.66s recent_step=3.68s eta=47:22:35 +2026-09-14 18:00:20 | INFO | train rank=0 | step=3440/50000 progress=6.9% loss=0.181489 lr=9.990340e-05 lr_backbone=9.990340e-05 elapsed=03:29:58 avg_step=3.66s recent_step=3.63s eta=47:21:54 +2026-09-14 18:00:52 | INFO | train rank=0 | step=3450/50000 progress=6.9% loss=0.160236 lr=9.990134e-05 lr_backbone=9.990134e-05 elapsed=03:30:30 avg_step=3.66s recent_step=3.27s eta=47:20:25 +2026-09-14 18:01:29 | INFO | train rank=0 | step=3460/50000 progress=6.9% loss=0.147674 lr=9.989925e-05 lr_backbone=9.989925e-05 elapsed=03:31:07 avg_step=3.66s recent_step=3.69s eta=47:19:52 +2026-09-14 18:02:12 | INFO | train rank=0 | step=3470/50000 progress=6.9% loss=0.138916 lr=9.989714e-05 lr_backbone=9.989714e-05 elapsed=03:31:50 avg_step=3.66s recent_step=4.26s eta=47:20:36 +2026-09-14 18:02:46 | INFO | train rank=0 | step=3480/50000 progress=7.0% loss=0.170483 lr=9.989501e-05 lr_backbone=9.989501e-05 elapsed=03:32:24 avg_step=3.66s recent_step=3.41s eta=47:19:26 +2026-09-14 18:03:18 | INFO | train rank=0 | step=3490/50000 progress=7.0% loss=0.128379 lr=9.989286e-05 lr_backbone=9.989286e-05 elapsed=03:32:56 avg_step=3.66s recent_step=3.20s eta=47:17:48 +2026-09-14 18:03:57 | INFO | train rank=0 | step=3500/50000 progress=7.0% loss=0.136525 lr=9.989068e-05 lr_backbone=9.989068e-05 elapsed=03:33:35 avg_step=3.66s recent_step=3.94s eta=47:17:48 +2026-09-14 18:04:28 | INFO | train rank=0 | step=3510/50000 progress=7.0% loss=0.154394 lr=9.988848e-05 lr_backbone=9.988848e-05 elapsed=03:34:06 avg_step=3.66s recent_step=3.06s eta=47:15:51 +2026-09-14 18:05:03 | INFO | train rank=0 | step=3520/50000 progress=7.0% loss=0.138014 lr=9.988627e-05 lr_backbone=9.988627e-05 elapsed=03:34:41 avg_step=3.66s recent_step=3.54s eta=47:14:59 +2026-09-14 18:05:35 | INFO | train rank=0 | step=3530/50000 progress=7.1% loss=0.155338 lr=9.988403e-05 lr_backbone=9.988403e-05 elapsed=03:35:13 avg_step=3.66s recent_step=3.16s eta=47:13:17 +2026-09-14 18:06:24 | INFO | train rank=0 | step=3540/50000 progress=7.1% loss=0.123411 lr=9.988176e-05 lr_backbone=9.988176e-05 elapsed=03:36:02 avg_step=3.66s recent_step=4.88s eta=47:15:21 +2026-09-14 18:06:56 | INFO | train rank=0 | step=3550/50000 progress=7.1% loss=0.141076 lr=9.987948e-05 lr_backbone=9.987948e-05 elapsed=03:36:34 avg_step=3.66s recent_step=3.22s eta=47:13:46 +2026-09-14 18:07:33 | INFO | train rank=0 | step=3560/50000 progress=7.1% loss=0.137276 lr=9.987718e-05 lr_backbone=9.987718e-05 elapsed=03:37:11 avg_step=3.66s recent_step=3.74s eta=47:13:20 +2026-09-14 18:08:12 | INFO | train rank=0 | step=3570/50000 progress=7.1% loss=0.154147 lr=9.987485e-05 lr_backbone=9.987485e-05 elapsed=03:37:50 avg_step=3.66s recent_step=3.87s eta=47:13:10 +2026-09-14 18:08:40 | INFO | train rank=0 | step=3580/50000 progress=7.2% loss=0.143328 lr=9.987250e-05 lr_backbone=9.987250e-05 elapsed=03:38:18 avg_step=3.66s recent_step=2.77s eta=47:10:37 +2026-09-14 18:09:31 | INFO | train rank=0 | step=3590/50000 progress=7.2% loss=0.141763 lr=9.987013e-05 lr_backbone=9.987013e-05 elapsed=03:39:09 avg_step=3.66s recent_step=5.12s eta=47:13:10 +2026-09-14 18:10:09 | INFO | train rank=0 | step=3600/50000 progress=7.2% loss=0.129908 lr=9.986773e-05 lr_backbone=9.986773e-05 elapsed=03:39:47 avg_step=3.66s recent_step=3.79s eta=47:12:50 +2026-09-14 18:10:40 | INFO | train rank=0 | step=3610/50000 progress=7.2% loss=0.130822 lr=9.986532e-05 lr_backbone=9.986532e-05 elapsed=03:40:18 avg_step=3.66s recent_step=3.16s eta=47:11:09 +2026-09-14 18:11:19 | INFO | train rank=0 | step=3620/50000 progress=7.2% loss=0.126388 lr=9.986288e-05 lr_backbone=9.986288e-05 elapsed=03:40:57 avg_step=3.66s recent_step=3.84s eta=47:10:56 +2026-09-14 18:12:14 | INFO | train rank=0 | step=3630/50000 progress=7.3% loss=0.117819 lr=9.986043e-05 lr_backbone=9.986043e-05 elapsed=03:41:52 avg_step=3.67s recent_step=5.49s eta=47:14:12 +2026-09-14 18:12:51 | INFO | train rank=0 | step=3640/50000 progress=7.3% loss=0.128881 lr=9.985795e-05 lr_backbone=9.985795e-05 elapsed=03:42:29 avg_step=3.67s recent_step=3.72s eta=47:13:43 +2026-09-14 18:13:16 | INFO | train rank=0 | step=3650/50000 progress=7.3% loss=0.134927 lr=9.985544e-05 lr_backbone=9.985544e-05 elapsed=03:42:54 avg_step=3.66s recent_step=2.46s eta=47:10:32 +2026-09-14 18:13:49 | INFO | train rank=0 | step=3660/50000 progress=7.3% loss=0.140093 lr=9.985292e-05 lr_backbone=9.985292e-05 elapsed=03:43:27 avg_step=3.66s recent_step=3.37s eta=47:09:18 +2026-09-14 18:14:32 | INFO | train rank=0 | step=3670/50000 progress=7.3% loss=0.138577 lr=9.985037e-05 lr_backbone=9.985037e-05 elapsed=03:44:10 avg_step=3.66s recent_step=4.24s eta=47:09:54 +2026-09-14 18:14:59 | INFO | train rank=0 | step=3680/50000 progress=7.4% loss=0.150947 lr=9.984781e-05 lr_backbone=9.984781e-05 elapsed=03:44:37 avg_step=3.66s recent_step=2.73s eta=47:07:21 +2026-09-14 18:15:28 | INFO | train rank=0 | step=3690/50000 progress=7.4% loss=0.129691 lr=9.984522e-05 lr_backbone=9.984522e-05 elapsed=03:45:06 avg_step=3.66s recent_step=2.86s eta=47:05:03 +2026-09-14 18:16:06 | INFO | train rank=0 | step=3700/50000 progress=7.4% loss=0.149573 lr=9.984261e-05 lr_backbone=9.984261e-05 elapsed=03:45:44 avg_step=3.66s recent_step=3.86s eta=47:04:51 +2026-09-14 18:16:52 | INFO | train rank=0 | step=3710/50000 progress=7.4% loss=0.146880 lr=9.983997e-05 lr_backbone=9.983997e-05 elapsed=03:46:30 avg_step=3.66s recent_step=4.60s eta=47:06:12 +2026-09-14 18:17:24 | INFO | train rank=0 | step=3720/50000 progress=7.4% loss=0.143059 lr=9.983732e-05 lr_backbone=9.983732e-05 elapsed=03:47:02 avg_step=3.66s recent_step=3.22s eta=47:04:41 +2026-09-14 18:18:01 | INFO | train rank=0 | step=3730/50000 progress=7.5% loss=0.133615 lr=9.983464e-05 lr_backbone=9.983464e-05 elapsed=03:47:39 avg_step=3.66s recent_step=3.68s eta=47:04:06 +2026-09-14 18:18:47 | INFO | train rank=0 | step=3740/50000 progress=7.5% loss=0.129462 lr=9.983194e-05 lr_backbone=9.983194e-05 elapsed=03:48:25 avg_step=3.66s recent_step=4.56s eta=47:05:20 +2026-09-14 18:19:15 | INFO | train rank=0 | step=3750/50000 progress=7.5% loss=0.136251 lr=9.982922e-05 lr_backbone=9.982922e-05 elapsed=03:48:54 avg_step=3.66s recent_step=2.87s eta=47:03:06 +2026-09-14 18:19:48 | INFO | train rank=0 | step=3760/50000 progress=7.5% loss=0.159186 lr=9.982648e-05 lr_backbone=9.982648e-05 elapsed=03:49:26 avg_step=3.66s recent_step=3.29s eta=47:01:43 +2026-09-14 18:20:44 | INFO | train rank=0 | step=3770/50000 progress=7.5% loss=0.129788 lr=9.982372e-05 lr_backbone=9.982372e-05 elapsed=03:50:22 avg_step=3.67s recent_step=5.54s eta=47:04:57 +2026-09-14 18:21:16 | INFO | train rank=0 | step=3780/50000 progress=7.6% loss=0.120845 lr=9.982093e-05 lr_backbone=9.982093e-05 elapsed=03:50:54 avg_step=3.67s recent_step=3.25s eta=47:03:29 +2026-09-14 18:21:51 | INFO | train rank=0 | step=3790/50000 progress=7.6% loss=0.161176 lr=9.981813e-05 lr_backbone=9.981813e-05 elapsed=03:51:29 avg_step=3.66s recent_step=3.45s eta=47:02:27 +2026-09-14 18:22:20 | INFO | train rank=0 | step=3800/50000 progress=7.6% loss=0.172310 lr=9.981530e-05 lr_backbone=9.981530e-05 elapsed=03:51:58 avg_step=3.66s recent_step=2.92s eta=47:00:20 +2026-09-14 18:23:04 | INFO | train rank=0 | step=3810/50000 progress=7.6% loss=0.104557 lr=9.981245e-05 lr_backbone=9.981245e-05 elapsed=03:52:42 avg_step=3.66s recent_step=4.42s eta=47:01:15 +2026-09-14 18:23:41 | INFO | train rank=0 | step=3820/50000 progress=7.6% loss=0.148717 lr=9.980957e-05 lr_backbone=9.980957e-05 elapsed=03:53:19 avg_step=3.66s recent_step=3.64s eta=47:00:35 +2026-09-14 18:24:11 | INFO | train rank=0 | step=3830/50000 progress=7.7% loss=0.160715 lr=9.980668e-05 lr_backbone=9.980668e-05 elapsed=03:53:49 avg_step=3.66s recent_step=3.01s eta=46:58:39 +2026-09-14 18:24:40 | INFO | train rank=0 | step=3840/50000 progress=7.7% loss=0.140192 lr=9.980376e-05 lr_backbone=9.980376e-05 elapsed=03:54:18 avg_step=3.66s recent_step=2.93s eta=46:56:35 +2026-09-14 18:25:34 | INFO | train rank=0 | step=3850/50000 progress=7.7% loss=0.112142 lr=9.980083e-05 lr_backbone=9.980083e-05 elapsed=03:55:12 avg_step=3.67s recent_step=5.39s eta=46:59:25 +2026-09-14 18:26:02 | INFO | train rank=0 | step=3860/50000 progress=7.7% loss=0.136294 lr=9.979787e-05 lr_backbone=9.979787e-05 elapsed=03:55:40 avg_step=3.66s recent_step=2.76s eta=46:57:00 +2026-09-14 18:26:37 | INFO | train rank=0 | step=3870/50000 progress=7.7% loss=0.125178 lr=9.979489e-05 lr_backbone=9.979489e-05 elapsed=03:56:15 avg_step=3.66s recent_step=3.56s eta=46:56:11 +2026-09-14 18:27:13 | INFO | train rank=0 | step=3880/50000 progress=7.8% loss=0.141874 lr=9.979188e-05 lr_backbone=9.979188e-05 elapsed=03:56:51 avg_step=3.66s recent_step=3.56s eta=46:55:22 +2026-09-14 18:28:05 | INFO | train rank=0 | step=3890/50000 progress=7.8% loss=0.121993 lr=9.978886e-05 lr_backbone=9.978886e-05 elapsed=03:57:43 avg_step=3.67s recent_step=5.21s eta=46:57:49 +2026-09-14 18:28:39 | INFO | train rank=0 | step=3900/50000 progress=7.8% loss=0.141541 lr=9.978581e-05 lr_backbone=9.978581e-05 elapsed=03:58:17 avg_step=3.67s recent_step=3.44s eta=46:56:46 +2026-09-14 18:29:17 | INFO | train rank=0 | step=3910/50000 progress=7.8% loss=0.133274 lr=9.978274e-05 lr_backbone=9.978274e-05 elapsed=03:58:55 avg_step=3.67s recent_step=3.79s eta=46:56:24 +2026-09-14 18:29:46 | INFO | train rank=0 | step=3920/50000 progress=7.8% loss=0.119017 lr=9.977965e-05 lr_backbone=9.977965e-05 elapsed=03:59:24 avg_step=3.66s recent_step=2.90s eta=46:54:17 +2026-09-14 18:30:28 | INFO | train rank=0 | step=3930/50000 progress=7.9% loss=0.146772 lr=9.977654e-05 lr_backbone=9.977654e-05 elapsed=04:00:06 avg_step=3.67s recent_step=4.18s eta=46:54:41 +2026-09-14 18:30:57 | INFO | train rank=0 | step=3940/50000 progress=7.9% loss=0.153977 lr=9.977341e-05 lr_backbone=9.977341e-05 elapsed=04:00:35 avg_step=3.66s recent_step=2.92s eta=46:52:37 +2026-09-14 18:31:26 | INFO | train rank=0 | step=3950/50000 progress=7.9% loss=0.125685 lr=9.977025e-05 lr_backbone=9.977025e-05 elapsed=04:01:04 avg_step=3.66s recent_step=2.86s eta=46:50:27 +2026-09-14 18:31:55 | INFO | train rank=0 | step=3960/50000 progress=7.9% loss=0.129883 lr=9.976707e-05 lr_backbone=9.976707e-05 elapsed=04:01:33 avg_step=3.66s recent_step=2.96s eta=46:48:28 +2026-09-14 18:32:35 | INFO | train rank=0 | step=3970/50000 progress=7.9% loss=0.122214 lr=9.976387e-05 lr_backbone=9.976387e-05 elapsed=04:02:13 avg_step=3.66s recent_step=4.01s eta=46:48:33 +2026-09-14 18:33:18 | INFO | train rank=0 | step=3980/50000 progress=8.0% loss=0.135321 lr=9.976065e-05 lr_backbone=9.976065e-05 elapsed=04:02:56 avg_step=3.66s recent_step=4.24s eta=46:49:03 +2026-09-14 18:34:03 | INFO | train rank=0 | step=3990/50000 progress=8.0% loss=0.142366 lr=9.975741e-05 lr_backbone=9.975741e-05 elapsed=04:03:41 avg_step=3.66s recent_step=4.52s eta=46:50:04 +2026-09-14 18:34:36 | INFO | train rank=0 | step=4000/50000 progress=8.0% loss=0.139377 lr=9.975415e-05 lr_backbone=9.975415e-05 elapsed=04:04:14 avg_step=3.66s recent_step=3.32s eta=46:48:48 +2026-09-14 18:35:04 | INFO | train rank=0 | step=4010/50000 progress=8.0% loss=0.122544 lr=9.975086e-05 lr_backbone=9.975086e-05 elapsed=04:04:42 avg_step=3.66s recent_step=2.83s eta=46:46:36 +2026-09-14 18:35:56 | INFO | train rank=0 | step=4020/50000 progress=8.0% loss=0.111665 lr=9.974755e-05 lr_backbone=9.974755e-05 elapsed=04:05:34 avg_step=3.67s recent_step=5.10s eta=46:48:44 +2026-09-14 18:36:30 | INFO | train rank=0 | step=4030/50000 progress=8.1% loss=0.104391 lr=9.974422e-05 lr_backbone=9.974422e-05 elapsed=04:06:08 avg_step=3.66s recent_step=3.43s eta=46:47:41 +2026-09-14 18:37:04 | INFO | train rank=0 | step=4040/50000 progress=8.1% loss=0.122427 lr=9.974087e-05 lr_backbone=9.974087e-05 elapsed=04:06:42 avg_step=3.66s recent_step=3.40s eta=46:46:34 +2026-09-14 18:37:35 | INFO | train rank=0 | step=4050/50000 progress=8.1% loss=0.137520 lr=9.973750e-05 lr_backbone=9.973750e-05 elapsed=04:07:13 avg_step=3.66s recent_step=3.17s eta=46:45:01 +2026-09-14 18:38:21 | INFO | train rank=0 | step=4060/50000 progress=8.1% loss=0.143363 lr=9.973410e-05 lr_backbone=9.973410e-05 elapsed=04:07:59 avg_step=3.66s recent_step=4.54s eta=46:46:03 +2026-09-14 18:38:55 | INFO | train rank=0 | step=4070/50000 progress=8.1% loss=0.132776 lr=9.973068e-05 lr_backbone=9.973068e-05 elapsed=04:08:33 avg_step=3.66s recent_step=3.42s eta=46:44:59 +2026-09-14 18:39:26 | INFO | train rank=0 | step=4080/50000 progress=8.2% loss=0.105266 lr=9.972725e-05 lr_backbone=9.972725e-05 elapsed=04:09:04 avg_step=3.66s recent_step=3.12s eta=46:43:21 +2026-09-14 18:40:02 | INFO | train rank=0 | step=4090/50000 progress=8.2% loss=0.130162 lr=9.972379e-05 lr_backbone=9.972379e-05 elapsed=04:09:40 avg_step=3.66s recent_step=3.57s eta=46:42:34 +2026-09-14 18:40:38 | INFO | train rank=0 | step=4100/50000 progress=8.2% loss=0.118848 lr=9.972030e-05 lr_backbone=9.972030e-05 elapsed=04:10:16 avg_step=3.66s recent_step=3.62s eta=46:41:53 +2026-09-14 18:41:12 | INFO | train rank=0 | step=4110/50000 progress=8.2% loss=0.126899 lr=9.971680e-05 lr_backbone=9.971680e-05 elapsed=04:10:50 avg_step=3.66s recent_step=3.41s eta=46:40:48 +2026-09-14 18:41:45 | INFO | train rank=0 | step=4120/50000 progress=8.2% loss=0.144549 lr=9.971327e-05 lr_backbone=9.971327e-05 elapsed=04:11:23 avg_step=3.66s recent_step=3.31s eta=46:39:32 +2026-09-14 18:42:39 | INFO | train rank=0 | step=4130/50000 progress=8.3% loss=0.116934 lr=9.970973e-05 lr_backbone=9.970973e-05 elapsed=04:12:17 avg_step=3.67s recent_step=5.32s eta=46:42:00 +2026-09-14 18:43:10 | INFO | train rank=0 | step=4140/50000 progress=8.3% loss=0.142790 lr=9.970616e-05 lr_backbone=9.970616e-05 elapsed=04:12:48 avg_step=3.66s recent_step=3.11s eta=46:40:22 +2026-09-14 18:43:40 | INFO | train rank=0 | step=4150/50000 progress=8.3% loss=0.119852 lr=9.970257e-05 lr_backbone=9.970257e-05 elapsed=04:13:18 avg_step=3.66s recent_step=2.99s eta=46:38:31 +2026-09-14 18:44:14 | INFO | train rank=0 | step=4160/50000 progress=8.3% loss=0.126983 lr=9.969895e-05 lr_backbone=9.969895e-05 elapsed=04:13:52 avg_step=3.66s recent_step=3.42s eta=46:37:27 +2026-09-14 18:45:05 | INFO | train rank=0 | step=4170/50000 progress=8.3% loss=0.143649 lr=9.969532e-05 lr_backbone=9.969532e-05 elapsed=04:14:43 avg_step=3.67s recent_step=5.16s eta=46:39:36 +2026-09-14 18:45:38 | INFO | train rank=0 | step=4180/50000 progress=8.4% loss=0.149500 lr=9.969166e-05 lr_backbone=9.969166e-05 elapsed=04:15:16 avg_step=3.66s recent_step=3.23s eta=46:38:11 +2026-09-14 18:46:10 | INFO | train rank=0 | step=4190/50000 progress=8.4% loss=0.132524 lr=9.968799e-05 lr_backbone=9.968799e-05 elapsed=04:15:48 avg_step=3.66s recent_step=3.24s eta=46:36:48 +2026-09-14 18:46:45 | INFO | train rank=0 | step=4200/50000 progress=8.4% loss=0.140588 lr=9.968429e-05 lr_backbone=9.968429e-05 elapsed=04:16:23 avg_step=3.66s recent_step=3.45s eta=46:35:48 +2026-09-14 18:47:33 | INFO | train rank=0 | step=4210/50000 progress=8.4% loss=0.125517 lr=9.968057e-05 lr_backbone=9.968057e-05 elapsed=04:17:11 avg_step=3.67s recent_step=4.80s eta=46:37:15 +2026-09-14 18:48:04 | INFO | train rank=0 | step=4220/50000 progress=8.4% loss=0.127165 lr=9.967682e-05 lr_backbone=9.967682e-05 elapsed=04:17:42 avg_step=3.66s recent_step=3.16s eta=46:35:44 +2026-09-14 18:48:45 | INFO | train rank=0 | step=4230/50000 progress=8.5% loss=0.145264 lr=9.967306e-05 lr_backbone=9.967306e-05 elapsed=04:18:23 avg_step=3.67s recent_step=4.05s eta=46:35:49 +2026-09-14 18:49:22 | INFO | train rank=0 | step=4240/50000 progress=8.5% loss=0.127893 lr=9.966927e-05 lr_backbone=9.966927e-05 elapsed=04:19:00 avg_step=3.67s recent_step=3.75s eta=46:35:22 +2026-09-14 18:50:04 | INFO | train rank=0 | step=4250/50000 progress=8.5% loss=0.171955 lr=9.966546e-05 lr_backbone=9.966546e-05 elapsed=04:19:42 avg_step=3.67s recent_step=4.23s eta=46:35:46 +2026-09-14 18:50:42 | INFO | train rank=0 | step=4260/50000 progress=8.5% loss=0.115808 lr=9.966163e-05 lr_backbone=9.966163e-05 elapsed=04:20:20 avg_step=3.67s recent_step=3.77s eta=46:35:20 +2026-09-14 18:51:10 | INFO | train rank=0 | step=4270/50000 progress=8.5% loss=0.124290 lr=9.965778e-05 lr_backbone=9.965778e-05 elapsed=04:20:48 avg_step=3.66s recent_step=2.78s eta=46:33:08 +2026-09-14 18:51:47 | INFO | train rank=0 | step=4280/50000 progress=8.6% loss=0.127776 lr=9.965391e-05 lr_backbone=9.965391e-05 elapsed=04:21:25 avg_step=3.66s recent_step=3.66s eta=46:32:31 +2026-09-14 18:52:26 | INFO | train rank=0 | step=4290/50000 progress=8.6% loss=0.157254 lr=9.965001e-05 lr_backbone=9.965001e-05 elapsed=04:22:04 avg_step=3.67s recent_step=3.96s eta=46:32:25 +2026-09-14 18:53:05 | INFO | train rank=0 | step=4300/50000 progress=8.6% loss=0.196056 lr=9.964610e-05 lr_backbone=9.964610e-05 elapsed=04:22:43 avg_step=3.67s recent_step=3.89s eta=46:32:12 +2026-09-14 18:53:37 | INFO | train rank=0 | step=4310/50000 progress=8.6% loss=0.130929 lr=9.964216e-05 lr_backbone=9.964216e-05 elapsed=04:23:15 avg_step=3.66s recent_step=3.20s eta=46:30:47 +2026-09-14 18:54:11 | INFO | train rank=0 | step=4320/50000 progress=8.6% loss=0.143629 lr=9.963820e-05 lr_backbone=9.963820e-05 elapsed=04:23:49 avg_step=3.66s recent_step=3.45s eta=46:29:47 +2026-09-14 18:55:06 | INFO | train rank=0 | step=4330/50000 progress=8.7% loss=0.101545 lr=9.963422e-05 lr_backbone=9.963422e-05 elapsed=04:24:44 avg_step=3.67s recent_step=5.45s eta=46:32:19 +2026-09-14 18:55:36 | INFO | train rank=0 | step=4340/50000 progress=8.7% loss=0.127305 lr=9.963021e-05 lr_backbone=9.963021e-05 elapsed=04:25:14 avg_step=3.67s recent_step=3.00s eta=46:30:32 +2026-09-14 18:56:12 | INFO | train rank=0 | step=4350/50000 progress=8.7% loss=0.141648 lr=9.962619e-05 lr_backbone=9.962619e-05 elapsed=04:25:50 avg_step=3.67s recent_step=3.61s eta=46:29:49 +2026-09-14 18:56:44 | INFO | train rank=0 | step=4360/50000 progress=8.7% loss=0.125707 lr=9.962214e-05 lr_backbone=9.962214e-05 elapsed=04:26:22 avg_step=3.67s recent_step=3.14s eta=46:28:18 +2026-09-14 18:57:36 | INFO | train rank=0 | step=4370/50000 progress=8.7% loss=0.149386 lr=9.961807e-05 lr_backbone=9.961807e-05 elapsed=04:27:14 avg_step=3.67s recent_step=5.22s eta=46:30:24 +2026-09-14 18:58:05 | INFO | train rank=0 | step=4380/50000 progress=8.8% loss=0.127239 lr=9.961398e-05 lr_backbone=9.961398e-05 elapsed=04:27:43 avg_step=3.67s recent_step=2.97s eta=46:28:34 +2026-09-14 18:58:37 | INFO | train rank=0 | step=4390/50000 progress=8.8% loss=0.107158 lr=9.960987e-05 lr_backbone=9.960987e-05 elapsed=04:28:15 avg_step=3.67s recent_step=3.10s eta=46:26:59 +2026-09-14 18:59:10 | INFO | train rank=0 | step=4400/50000 progress=8.8% loss=0.128269 lr=9.960574e-05 lr_backbone=9.960574e-05 elapsed=04:28:48 avg_step=3.67s recent_step=3.32s eta=46:25:47 +2026-09-14 18:59:55 | INFO | train rank=0 | step=4410/50000 progress=8.8% loss=0.127148 lr=9.960158e-05 lr_backbone=9.960158e-05 elapsed=04:29:33 avg_step=3.67s recent_step=4.55s eta=46:26:42 +2026-09-14 19:00:32 | INFO | train rank=0 | step=4420/50000 progress=8.8% loss=0.141296 lr=9.959740e-05 lr_backbone=9.959740e-05 elapsed=04:30:10 avg_step=3.67s recent_step=3.63s eta=46:26:02 +2026-09-14 19:01:04 | INFO | train rank=0 | step=4430/50000 progress=8.9% loss=0.109859 lr=9.959320e-05 lr_backbone=9.959320e-05 elapsed=04:30:42 avg_step=3.67s recent_step=3.21s eta=46:24:38 +2026-09-14 19:01:36 | INFO | train rank=0 | step=4440/50000 progress=8.9% loss=0.131039 lr=9.958898e-05 lr_backbone=9.958898e-05 elapsed=04:31:14 avg_step=3.67s recent_step=3.22s eta=46:23:16 +2026-09-14 19:02:13 | INFO | train rank=0 | step=4450/50000 progress=8.9% loss=0.167111 lr=9.958474e-05 lr_backbone=9.958474e-05 elapsed=04:31:51 avg_step=3.67s recent_step=3.70s eta=46:22:43 +2026-09-14 19:02:45 | INFO | train rank=0 | step=4460/50000 progress=8.9% loss=0.133536 lr=9.958048e-05 lr_backbone=9.958048e-05 elapsed=04:32:23 avg_step=3.66s recent_step=3.16s eta=46:21:15 +2026-09-14 19:03:12 | INFO | train rank=0 | step=4470/50000 progress=8.9% loss=0.142030 lr=9.957619e-05 lr_backbone=9.957619e-05 elapsed=04:32:50 avg_step=3.66s recent_step=2.71s eta=46:19:01 +2026-09-14 19:03:52 | INFO | train rank=0 | step=4480/50000 progress=9.0% loss=0.163940 lr=9.957188e-05 lr_backbone=9.957188e-05 elapsed=04:33:30 avg_step=3.66s recent_step=4.06s eta=46:19:05 +2026-09-14 19:04:41 | INFO | train rank=0 | step=4490/50000 progress=9.0% loss=0.117314 lr=9.956755e-05 lr_backbone=9.956755e-05 elapsed=04:34:19 avg_step=3.67s recent_step=4.84s eta=46:20:28 +2026-09-14 19:05:14 | INFO | train rank=0 | step=4500/50000 progress=9.0% loss=0.121833 lr=9.956320e-05 lr_backbone=9.956320e-05 elapsed=04:34:52 avg_step=3.66s recent_step=3.31s eta=46:19:16 +2026-09-14 19:05:51 | INFO | train rank=0 | step=4510/50000 progress=9.0% loss=0.148514 lr=9.955883e-05 lr_backbone=9.955883e-05 elapsed=04:35:29 avg_step=3.67s recent_step=3.73s eta=46:18:45 +2026-09-14 19:06:23 | INFO | train rank=0 | step=4520/50000 progress=9.0% loss=0.112303 lr=9.955444e-05 lr_backbone=9.955444e-05 elapsed=04:36:01 avg_step=3.66s recent_step=3.17s eta=46:17:18 +2026-09-14 19:07:11 | INFO | train rank=0 | step=4530/50000 progress=9.1% loss=0.125196 lr=9.955002e-05 lr_backbone=9.955002e-05 elapsed=04:36:49 avg_step=3.67s recent_step=4.85s eta=46:18:41 +2026-09-14 19:07:45 | INFO | train rank=0 | step=4540/50000 progress=9.1% loss=0.120941 lr=9.954558e-05 lr_backbone=9.954558e-05 elapsed=04:37:23 avg_step=3.67s recent_step=3.37s eta=46:17:35 +2026-09-14 19:08:16 | INFO | train rank=0 | step=4550/50000 progress=9.1% loss=0.146020 lr=9.954112e-05 lr_backbone=9.954112e-05 elapsed=04:37:54 avg_step=3.66s recent_step=3.14s eta=46:16:06 +2026-09-14 19:08:52 | INFO | train rank=0 | step=4560/50000 progress=9.1% loss=0.178557 lr=9.953664e-05 lr_backbone=9.953664e-05 elapsed=04:38:30 avg_step=3.66s recent_step=3.53s eta=46:15:15 +2026-09-14 19:09:27 | INFO | train rank=0 | step=4570/50000 progress=9.1% loss=0.174490 lr=9.953214e-05 lr_backbone=9.953214e-05 elapsed=04:39:05 avg_step=3.66s recent_step=3.49s eta=46:14:21 +2026-09-14 19:10:07 | INFO | train rank=0 | step=4580/50000 progress=9.2% loss=0.147521 lr=9.952762e-05 lr_backbone=9.952762e-05 elapsed=04:39:45 avg_step=3.67s recent_step=4.08s eta=46:14:26 +2026-09-14 19:10:43 | INFO | train rank=0 | step=4590/50000 progress=9.2% loss=0.105569 lr=9.952307e-05 lr_backbone=9.952307e-05 elapsed=04:40:21 avg_step=3.66s recent_step=3.55s eta=46:13:38 +2026-09-14 19:11:18 | INFO | train rank=0 | step=4600/50000 progress=9.2% loss=0.148438 lr=9.951850e-05 lr_backbone=9.951850e-05 elapsed=04:40:56 avg_step=3.66s recent_step=3.54s eta=46:12:49 +2026-09-14 19:11:47 | INFO | train rank=0 | step=4610/50000 progress=9.2% loss=0.112463 lr=9.951391e-05 lr_backbone=9.951391e-05 elapsed=04:41:25 avg_step=3.66s recent_step=2.85s eta=46:10:53 +2026-09-14 19:12:33 | INFO | train rank=0 | step=4620/50000 progress=9.2% loss=0.119184 lr=9.950930e-05 lr_backbone=9.950930e-05 elapsed=04:42:11 avg_step=3.66s recent_step=4.62s eta=46:11:51 +2026-09-14 19:12:58 | INFO | train rank=0 | step=4630/50000 progress=9.3% loss=0.172065 lr=9.950467e-05 lr_backbone=9.950467e-05 elapsed=04:42:36 avg_step=3.66s recent_step=2.48s eta=46:09:18 +2026-09-14 19:13:42 | INFO | train rank=0 | step=4640/50000 progress=9.3% loss=0.122385 lr=9.950002e-05 lr_backbone=9.950002e-05 elapsed=04:43:20 avg_step=3.66s recent_step=4.40s eta=46:09:54 +2026-09-14 19:14:19 | INFO | train rank=0 | step=4650/50000 progress=9.3% loss=0.128103 lr=9.949534e-05 lr_backbone=9.949534e-05 elapsed=04:43:57 avg_step=3.66s recent_step=3.69s eta=46:09:19 +2026-09-14 19:14:51 | INFO | train rank=0 | step=4660/50000 progress=9.3% loss=0.117631 lr=9.949064e-05 lr_backbone=9.949064e-05 elapsed=04:44:29 avg_step=3.66s recent_step=3.27s eta=46:08:04 +2026-09-14 19:15:30 | INFO | train rank=0 | step=4670/50000 progress=9.3% loss=0.091507 lr=9.948593e-05 lr_backbone=9.948593e-05 elapsed=04:45:08 avg_step=3.66s recent_step=3.86s eta=46:07:47 +2026-09-14 19:16:15 | INFO | train rank=0 | step=4680/50000 progress=9.4% loss=0.142823 lr=9.948118e-05 lr_backbone=9.948118e-05 elapsed=04:45:53 avg_step=3.67s recent_step=4.51s eta=46:08:32 +2026-09-14 19:16:51 | INFO | train rank=0 | step=4690/50000 progress=9.4% loss=0.121753 lr=9.947642e-05 lr_backbone=9.947642e-05 elapsed=04:46:29 avg_step=3.67s recent_step=3.54s eta=46:07:43 +2026-09-14 19:17:20 | INFO | train rank=0 | step=4700/50000 progress=9.4% loss=0.122790 lr=9.947164e-05 lr_backbone=9.947164e-05 elapsed=04:46:58 avg_step=3.66s recent_step=2.92s eta=46:05:55 +2026-09-14 19:17:53 | INFO | train rank=0 | step=4710/50000 progress=9.4% loss=0.121767 lr=9.946683e-05 lr_backbone=9.946683e-05 elapsed=04:47:31 avg_step=3.66s recent_step=3.35s eta=46:04:48 +2026-09-14 19:18:33 | INFO | train rank=0 | step=4720/50000 progress=9.4% loss=0.120689 lr=9.946201e-05 lr_backbone=9.946201e-05 elapsed=04:48:11 avg_step=3.66s recent_step=4.00s eta=46:04:44 +2026-09-14 19:19:23 | INFO | train rank=0 | step=4730/50000 progress=9.5% loss=0.127353 lr=9.945716e-05 lr_backbone=9.945716e-05 elapsed=04:49:01 avg_step=3.67s recent_step=5.01s eta=46:06:16 +2026-09-14 19:19:58 | INFO | train rank=0 | step=4740/50000 progress=9.5% loss=0.127659 lr=9.945229e-05 lr_backbone=9.945229e-05 elapsed=04:49:36 avg_step=3.67s recent_step=3.45s eta=46:05:19 +2026-09-14 19:20:28 | INFO | train rank=0 | step=4750/50000 progress=9.5% loss=0.124027 lr=9.944739e-05 lr_backbone=9.944739e-05 elapsed=04:50:06 avg_step=3.66s recent_step=3.02s eta=46:03:40 +2026-09-14 19:21:01 | INFO | train rank=0 | step=4760/50000 progress=9.5% loss=0.127619 lr=9.944248e-05 lr_backbone=9.944248e-05 elapsed=04:50:39 avg_step=3.66s recent_step=3.25s eta=46:02:24 +2026-09-14 19:21:49 | INFO | train rank=0 | step=4770/50000 progress=9.5% loss=0.135473 lr=9.943754e-05 lr_backbone=9.943754e-05 elapsed=04:51:27 avg_step=3.67s recent_step=4.85s eta=46:03:40 +2026-09-14 19:22:21 | INFO | train rank=0 | step=4780/50000 progress=9.6% loss=0.126009 lr=9.943259e-05 lr_backbone=9.943259e-05 elapsed=04:51:59 avg_step=3.67s recent_step=3.17s eta=46:02:17 +2026-09-14 19:22:59 | INFO | train rank=0 | step=4790/50000 progress=9.6% loss=0.164495 lr=9.942761e-05 lr_backbone=9.942761e-05 elapsed=04:52:37 avg_step=3.67s recent_step=3.78s eta=46:01:51 +2026-09-14 19:23:37 | INFO | train rank=0 | step=4800/50000 progress=9.6% loss=0.116383 lr=9.942261e-05 lr_backbone=9.942261e-05 elapsed=04:53:15 avg_step=3.67s recent_step=3.88s eta=46:01:34 +2026-09-14 19:24:08 | INFO | train rank=0 | step=4810/50000 progress=9.6% loss=0.132062 lr=9.941759e-05 lr_backbone=9.941759e-05 elapsed=04:53:46 avg_step=3.66s recent_step=3.08s eta=46:00:03 +2026-09-14 19:24:42 | INFO | train rank=0 | step=4820/50000 progress=9.6% loss=0.121262 lr=9.941254e-05 lr_backbone=9.941254e-05 elapsed=04:54:20 avg_step=3.66s recent_step=3.40s eta=45:59:01 +2026-09-14 19:25:25 | INFO | train rank=0 | step=4830/50000 progress=9.7% loss=0.124585 lr=9.940748e-05 lr_backbone=9.940748e-05 elapsed=04:55:03 avg_step=3.67s recent_step=4.29s eta=45:59:23 +2026-09-14 19:26:11 | INFO | train rank=0 | step=4840/50000 progress=9.7% loss=0.103926 lr=9.940239e-05 lr_backbone=9.940239e-05 elapsed=04:55:49 avg_step=3.67s recent_step=4.62s eta=46:00:16 +2026-09-14 19:26:37 | INFO | train rank=0 | step=4850/50000 progress=9.7% loss=0.113059 lr=9.939728e-05 lr_backbone=9.939728e-05 elapsed=04:56:15 avg_step=3.67s recent_step=2.57s eta=45:57:58 +2026-09-14 19:27:12 | INFO | train rank=0 | step=4860/50000 progress=9.7% loss=0.120368 lr=9.939215e-05 lr_backbone=9.939215e-05 elapsed=04:56:50 avg_step=3.66s recent_step=3.48s eta=45:57:04 +2026-09-14 19:27:54 | INFO | train rank=0 | step=4870/50000 progress=9.7% loss=0.146492 lr=9.938700e-05 lr_backbone=9.938700e-05 elapsed=04:57:32 avg_step=3.67s recent_step=4.20s eta=45:57:17 +2026-09-14 19:28:39 | INFO | train rank=0 | step=4880/50000 progress=9.8% loss=0.167145 lr=9.938183e-05 lr_backbone=9.938183e-05 elapsed=04:58:17 avg_step=3.67s recent_step=4.51s eta=45:57:58 +2026-09-14 19:29:16 | INFO | train rank=0 | step=4890/50000 progress=9.8% loss=0.149584 lr=9.937663e-05 lr_backbone=9.937663e-05 elapsed=04:58:54 avg_step=3.67s recent_step=3.69s eta=45:57:23 +2026-09-14 19:29:47 | INFO | train rank=0 | step=4900/50000 progress=9.8% loss=0.186356 lr=9.937142e-05 lr_backbone=9.937142e-05 elapsed=04:59:25 avg_step=3.67s recent_step=3.10s eta=45:55:55 +2026-09-14 19:30:34 | INFO | train rank=0 | step=4910/50000 progress=9.8% loss=0.118365 lr=9.936618e-05 lr_backbone=9.936618e-05 elapsed=05:00:12 avg_step=3.67s recent_step=4.70s eta=45:56:53 +2026-09-14 19:31:02 | INFO | train rank=0 | step=4920/50000 progress=9.8% loss=0.124510 lr=9.936092e-05 lr_backbone=9.936092e-05 elapsed=05:00:40 avg_step=3.67s recent_step=2.78s eta=45:54:55 +2026-09-14 19:31:39 | INFO | train rank=0 | step=4930/50000 progress=9.9% loss=0.107079 lr=9.935564e-05 lr_backbone=9.935564e-05 elapsed=05:01:17 avg_step=3.67s recent_step=3.74s eta=45:54:26 +2026-09-14 19:32:08 | INFO | train rank=0 | step=4940/50000 progress=9.9% loss=0.101619 lr=9.935034e-05 lr_backbone=9.935034e-05 elapsed=05:01:46 avg_step=3.67s recent_step=2.91s eta=45:52:40 +2026-09-14 19:32:50 | INFO | train rank=0 | step=4950/50000 progress=9.9% loss=0.107982 lr=9.934501e-05 lr_backbone=9.934501e-05 elapsed=05:02:28 avg_step=3.67s recent_step=4.13s eta=45:52:46 +2026-09-14 19:33:39 | INFO | train rank=0 | step=4960/50000 progress=9.9% loss=0.131058 lr=9.933966e-05 lr_backbone=9.933966e-05 elapsed=05:03:17 avg_step=3.67s recent_step=4.95s eta=45:54:05 +2026-09-14 19:34:12 | INFO | train rank=0 | step=4970/50000 progress=9.9% loss=0.104649 lr=9.933430e-05 lr_backbone=9.933430e-05 elapsed=05:03:50 avg_step=3.67s recent_step=3.34s eta=45:52:59 +2026-09-14 19:34:50 | INFO | train rank=0 | step=4980/50000 progress=10.0% loss=0.101773 lr=9.932891e-05 lr_backbone=9.932891e-05 elapsed=05:04:28 avg_step=3.67s recent_step=3.77s eta=45:52:32 +2026-09-14 19:35:15 | INFO | train rank=0 | step=4990/50000 progress=10.0% loss=0.117405 lr=9.932350e-05 lr_backbone=9.932350e-05 elapsed=05:04:53 avg_step=3.67s recent_step=2.52s eta=45:50:11 +2026-09-14 19:36:00 | INFO | train rank=0 | step=5000/50000 progress=10.0% loss=0.140061 lr=9.931807e-05 lr_backbone=9.931807e-05 elapsed=05:05:38 avg_step=3.67s recent_step=4.42s eta=45:50:43 +2026-09-14 19:36:39 | INFO | train rank=0 | step=5010/50000 progress=10.0% loss=0.109432 lr=9.931261e-05 lr_backbone=9.931261e-05 elapsed=05:06:17 avg_step=3.67s recent_step=3.93s eta=45:50:30 +2026-09-14 19:37:07 | INFO | train rank=0 | step=5020/50000 progress=10.0% loss=0.108725 lr=9.930714e-05 lr_backbone=9.930714e-05 elapsed=05:06:45 avg_step=3.67s recent_step=2.76s eta=45:48:31 +2026-09-14 19:37:44 | INFO | train rank=0 | step=5030/50000 progress=10.1% loss=0.130307 lr=9.930164e-05 lr_backbone=9.930164e-05 elapsed=05:07:22 avg_step=3.67s recent_step=3.77s eta=45:48:04 +2026-09-14 19:38:31 | INFO | train rank=0 | step=5040/50000 progress=10.1% loss=0.130698 lr=9.929612e-05 lr_backbone=9.929612e-05 elapsed=05:08:09 avg_step=3.67s recent_step=4.67s eta=45:48:57 +2026-09-14 19:39:02 | INFO | train rank=0 | step=5050/50000 progress=10.1% loss=0.124180 lr=9.929058e-05 lr_backbone=9.929058e-05 elapsed=05:08:40 avg_step=3.67s recent_step=3.11s eta=45:47:31 +2026-09-14 19:39:36 | INFO | train rank=0 | step=5060/50000 progress=10.1% loss=0.129495 lr=9.928502e-05 lr_backbone=9.928502e-05 elapsed=05:09:14 avg_step=3.67s recent_step=3.38s eta=45:46:29 +2026-09-14 19:40:12 | INFO | train rank=0 | step=5070/50000 progress=10.1% loss=0.097368 lr=9.927944e-05 lr_backbone=9.927944e-05 elapsed=05:09:50 avg_step=3.67s recent_step=3.66s eta=45:45:51 +2026-09-14 19:40:52 | INFO | train rank=0 | step=5080/50000 progress=10.2% loss=0.127074 lr=9.927383e-05 lr_backbone=9.927383e-05 elapsed=05:10:30 avg_step=3.67s recent_step=3.96s eta=45:45:40 +2026-09-14 19:41:21 | INFO | train rank=0 | step=5090/50000 progress=10.2% loss=0.140259 lr=9.926820e-05 lr_backbone=9.926820e-05 elapsed=05:10:59 avg_step=3.67s recent_step=2.94s eta=45:44:00 +2026-09-14 19:41:56 | INFO | train rank=0 | step=5100/50000 progress=10.2% loss=0.125188 lr=9.926256e-05 lr_backbone=9.926256e-05 elapsed=05:11:34 avg_step=3.67s recent_step=3.50s eta=45:43:09 +2026-09-14 19:42:43 | INFO | train rank=0 | step=5110/50000 progress=10.2% loss=0.128255 lr=9.925689e-05 lr_backbone=9.925689e-05 elapsed=05:12:21 avg_step=3.67s recent_step=4.64s eta=45:43:57 +2026-09-14 19:43:18 | INFO | train rank=0 | step=5120/50000 progress=10.2% loss=0.136851 lr=9.925120e-05 lr_backbone=9.925120e-05 elapsed=05:12:56 avg_step=3.67s recent_step=3.51s eta=45:43:06 +2026-09-14 19:43:48 | INFO | train rank=0 | step=5130/50000 progress=10.3% loss=0.091846 lr=9.924548e-05 lr_backbone=9.924548e-05 elapsed=05:13:26 avg_step=3.67s recent_step=3.04s eta=45:41:35 +2026-09-14 19:44:27 | INFO | train rank=0 | step=5140/50000 progress=10.3% loss=0.118405 lr=9.923975e-05 lr_backbone=9.923975e-05 elapsed=05:14:05 avg_step=3.67s recent_step=3.89s eta=45:41:17 +2026-09-14 19:44:55 | INFO | train rank=0 | step=5150/50000 progress=10.3% loss=0.121878 lr=9.923399e-05 lr_backbone=9.923399e-05 elapsed=05:14:33 avg_step=3.66s recent_step=2.75s eta=45:39:21 +2026-09-14 19:45:37 | INFO | train rank=0 | step=5160/50000 progress=10.3% loss=0.109802 lr=9.922822e-05 lr_backbone=9.922822e-05 elapsed=05:15:15 avg_step=3.67s recent_step=4.28s eta=45:39:38 +2026-09-14 19:46:17 | INFO | train rank=0 | step=5170/50000 progress=10.3% loss=0.129397 lr=9.922242e-05 lr_backbone=9.922242e-05 elapsed=05:15:55 avg_step=3.67s recent_step=3.93s eta=45:39:24 +2026-09-14 19:46:49 | INFO | train rank=0 | step=5180/50000 progress=10.4% loss=0.125307 lr=9.921660e-05 lr_backbone=9.921660e-05 elapsed=05:16:27 avg_step=3.67s recent_step=3.23s eta=45:38:10 +2026-09-14 19:47:34 | INFO | train rank=0 | step=5190/50000 progress=10.4% loss=0.132522 lr=9.921076e-05 lr_backbone=9.921076e-05 elapsed=05:17:12 avg_step=3.67s recent_step=4.46s eta=45:38:42 +2026-09-14 19:48:07 | INFO | train rank=0 | step=5200/50000 progress=10.4% loss=0.120739 lr=9.920489e-05 lr_backbone=9.920489e-05 elapsed=05:17:45 avg_step=3.67s recent_step=3.35s eta=45:37:37 +2026-09-14 19:48:44 | INFO | train rank=0 | step=5210/50000 progress=10.4% loss=0.101240 lr=9.919901e-05 lr_backbone=9.919901e-05 elapsed=05:18:22 avg_step=3.67s recent_step=3.68s eta=45:37:02 +2026-09-14 19:49:15 | INFO | train rank=0 | step=5220/50000 progress=10.4% loss=0.128992 lr=9.919310e-05 lr_backbone=9.919310e-05 elapsed=05:18:53 avg_step=3.67s recent_step=3.12s eta=45:35:38 +2026-09-14 19:50:04 | INFO | train rank=0 | step=5230/50000 progress=10.5% loss=0.101543 lr=9.918717e-05 lr_backbone=9.918717e-05 elapsed=05:19:42 avg_step=3.67s recent_step=4.88s eta=45:36:45 +2026-09-14 19:50:43 | INFO | train rank=0 | step=5240/50000 progress=10.5% loss=0.153605 lr=9.918122e-05 lr_backbone=9.918122e-05 elapsed=05:20:21 avg_step=3.67s recent_step=3.86s eta=45:36:26 +2026-09-14 19:51:19 | INFO | train rank=0 | step=5250/50000 progress=10.5% loss=0.129422 lr=9.917525e-05 lr_backbone=9.917525e-05 elapsed=05:20:58 avg_step=3.67s recent_step=3.69s eta=45:35:51 +2026-09-14 19:51:51 | INFO | train rank=0 | step=5260/50000 progress=10.5% loss=0.160522 lr=9.916926e-05 lr_backbone=9.916926e-05 elapsed=05:21:29 avg_step=3.67s recent_step=3.10s eta=45:34:26 +2026-09-14 19:52:29 | INFO | train rank=0 | step=5270/50000 progress=10.5% loss=0.124935 lr=9.916325e-05 lr_backbone=9.916325e-05 elapsed=05:22:07 avg_step=3.67s recent_step=3.80s eta=45:34:01 +2026-09-14 19:53:15 | INFO | train rank=0 | step=5280/50000 progress=10.6% loss=0.127607 lr=9.915721e-05 lr_backbone=9.915721e-05 elapsed=05:22:53 avg_step=3.67s recent_step=4.60s eta=45:34:44 +2026-09-14 19:53:45 | INFO | train rank=0 | step=5290/50000 progress=10.6% loss=0.133074 lr=9.915116e-05 lr_backbone=9.915116e-05 elapsed=05:23:23 avg_step=3.67s recent_step=3.03s eta=45:33:13 +2026-09-14 19:54:14 | INFO | train rank=0 | step=5300/50000 progress=10.6% loss=0.126745 lr=9.914508e-05 lr_backbone=9.914508e-05 elapsed=05:23:52 avg_step=3.67s recent_step=2.90s eta=45:31:31 +2026-09-14 19:54:52 | INFO | train rank=0 | step=5310/50000 progress=10.6% loss=0.131511 lr=9.913898e-05 lr_backbone=9.913898e-05 elapsed=05:24:30 avg_step=3.67s recent_step=3.81s eta=45:31:07 +2026-09-14 19:55:31 | INFO | train rank=0 | step=5320/50000 progress=10.6% loss=0.125919 lr=9.913286e-05 lr_backbone=9.913286e-05 elapsed=05:25:09 avg_step=3.67s recent_step=3.95s eta=45:30:54 +2026-09-14 19:56:07 | INFO | train rank=0 | step=5330/50000 progress=10.7% loss=0.140816 lr=9.912671e-05 lr_backbone=9.912671e-05 elapsed=05:25:45 avg_step=3.67s recent_step=3.54s eta=45:30:07 +2026-09-14 19:56:47 | INFO | train rank=0 | step=5340/50000 progress=10.7% loss=0.116357 lr=9.912055e-05 lr_backbone=9.912055e-05 elapsed=05:26:25 avg_step=3.67s recent_step=3.99s eta=45:29:57 +2026-09-14 19:57:18 | INFO | train rank=0 | step=5350/50000 progress=10.7% loss=0.119287 lr=9.911436e-05 lr_backbone=9.911436e-05 elapsed=05:26:56 avg_step=3.67s recent_step=3.12s eta=45:28:35 +2026-09-14 19:57:59 | INFO | train rank=0 | step=5360/50000 progress=10.7% loss=0.116035 lr=9.910816e-05 lr_backbone=9.910816e-05 elapsed=05:27:37 avg_step=3.67s recent_step=4.08s eta=45:28:33 +2026-09-14 19:58:30 | INFO | train rank=0 | step=5370/50000 progress=10.7% loss=0.119810 lr=9.910193e-05 lr_backbone=9.910193e-05 elapsed=05:28:08 avg_step=3.67s recent_step=3.13s eta=45:27:12 +2026-09-14 19:58:58 | INFO | train rank=0 | step=5380/50000 progress=10.8% loss=0.135992 lr=9.909568e-05 lr_backbone=9.909568e-05 elapsed=05:28:36 avg_step=3.66s recent_step=2.83s eta=45:25:25 +2026-09-14 19:59:30 | INFO | train rank=0 | step=5390/50000 progress=10.8% loss=0.103570 lr=9.908940e-05 lr_backbone=9.908940e-05 elapsed=05:29:08 avg_step=3.66s recent_step=3.15s eta=45:24:06 +2026-09-14 20:00:20 | INFO | train rank=0 | step=5400/50000 progress=10.8% loss=0.112905 lr=9.908311e-05 lr_backbone=9.908311e-05 elapsed=05:29:58 avg_step=3.67s recent_step=4.98s eta=45:25:18 +2026-09-14 20:00:59 | INFO | train rank=0 | step=5410/50000 progress=10.8% loss=0.112273 lr=9.907680e-05 lr_backbone=9.907680e-05 elapsed=05:30:37 avg_step=3.67s recent_step=3.96s eta=45:25:05 +2026-09-14 20:01:40 | INFO | train rank=0 | step=5420/50000 progress=10.8% loss=0.088336 lr=9.907046e-05 lr_backbone=9.907046e-05 elapsed=05:31:18 avg_step=3.67s recent_step=4.03s eta=45:24:58 +2026-09-14 20:02:05 | INFO | train rank=0 | step=5430/50000 progress=10.9% loss=0.119921 lr=9.906410e-05 lr_backbone=9.906410e-05 elapsed=05:31:43 avg_step=3.67s recent_step=2.53s eta=45:22:48 +2026-09-14 20:02:39 | INFO | train rank=0 | step=5440/50000 progress=10.9% loss=0.125837 lr=9.905772e-05 lr_backbone=9.905772e-05 elapsed=05:32:17 avg_step=3.66s recent_step=3.42s eta=45:21:51 +2026-09-14 20:03:12 | INFO | train rank=0 | step=5450/50000 progress=10.9% loss=0.113903 lr=9.905132e-05 lr_backbone=9.905132e-05 elapsed=05:32:50 avg_step=3.66s recent_step=3.32s eta=45:20:47 +2026-09-14 20:03:43 | INFO | train rank=0 | step=5460/50000 progress=10.9% loss=0.108014 lr=9.904490e-05 lr_backbone=9.904490e-05 elapsed=05:33:21 avg_step=3.66s recent_step=3.10s eta=45:19:24 +2026-09-14 20:04:19 | INFO | train rank=0 | step=5470/50000 progress=10.9% loss=0.117847 lr=9.903846e-05 lr_backbone=9.903846e-05 elapsed=05:33:57 avg_step=3.66s recent_step=3.58s eta=45:18:40 +2026-09-14 20:05:20 | INFO | train rank=0 | step=5480/50000 progress=11.0% loss=0.128803 lr=9.903199e-05 lr_backbone=9.903199e-05 elapsed=05:34:58 avg_step=3.67s recent_step=6.09s eta=45:21:21 +2026-09-14 20:05:55 | INFO | train rank=0 | step=5490/50000 progress=11.0% loss=0.116607 lr=9.902551e-05 lr_backbone=9.902551e-05 elapsed=05:35:33 avg_step=3.67s recent_step=3.52s eta=45:20:32 +2026-09-14 20:06:27 | INFO | train rank=0 | step=5500/50000 progress=11.0% loss=0.128629 lr=9.901900e-05 lr_backbone=9.901900e-05 elapsed=05:36:05 avg_step=3.67s recent_step=3.22s eta=45:19:19 +2026-09-14 20:07:07 | INFO | train rank=0 | step=5510/50000 progress=11.0% loss=0.119595 lr=9.901247e-05 lr_backbone=9.901247e-05 elapsed=05:36:45 avg_step=3.67s recent_step=4.00s eta=45:19:09 +2026-09-14 20:07:58 | INFO | train rank=0 | step=5520/50000 progress=11.0% loss=0.105931 lr=9.900592e-05 lr_backbone=9.900592e-05 elapsed=05:37:36 avg_step=3.67s recent_step=5.07s eta=45:20:25 +2026-09-14 20:08:29 | INFO | train rank=0 | step=5530/50000 progress=11.1% loss=0.126778 lr=9.899935e-05 lr_backbone=9.899935e-05 elapsed=05:38:07 avg_step=3.67s recent_step=3.10s eta=45:19:03 +2026-09-14 20:09:00 | INFO | train rank=0 | step=5540/50000 progress=11.1% loss=0.131285 lr=9.899275e-05 lr_backbone=9.899275e-05 elapsed=05:38:38 avg_step=3.67s recent_step=3.10s eta=45:17:41 +2026-09-14 20:09:41 | INFO | train rank=0 | step=5550/50000 progress=11.1% loss=0.108536 lr=9.898614e-05 lr_backbone=9.898614e-05 elapsed=05:39:19 avg_step=3.67s recent_step=4.12s eta=45:17:41 +2026-09-14 20:10:22 | INFO | train rank=0 | step=5560/50000 progress=11.1% loss=0.117365 lr=9.897950e-05 lr_backbone=9.897950e-05 elapsed=05:40:00 avg_step=3.67s recent_step=4.13s eta=45:17:40 +2026-09-14 20:10:54 | INFO | train rank=0 | step=5570/50000 progress=11.1% loss=0.106971 lr=9.897284e-05 lr_backbone=9.897284e-05 elapsed=05:40:32 avg_step=3.67s recent_step=3.15s eta=45:16:22 +2026-09-14 20:11:27 | INFO | train rank=0 | step=5580/50000 progress=11.2% loss=0.140317 lr=9.896616e-05 lr_backbone=9.896616e-05 elapsed=05:41:05 avg_step=3.67s recent_step=3.26s eta=45:15:13 +2026-09-14 20:12:05 | INFO | train rank=0 | step=5590/50000 progress=11.2% loss=0.103529 lr=9.895946e-05 lr_backbone=9.895946e-05 elapsed=05:41:43 avg_step=3.67s recent_step=3.80s eta=45:14:47 +2026-09-14 20:12:49 | INFO | train rank=0 | step=5600/50000 progress=11.2% loss=0.115883 lr=9.895274e-05 lr_backbone=9.895274e-05 elapsed=05:42:27 avg_step=3.67s recent_step=4.42s eta=45:15:10 +2026-09-14 20:13:21 | INFO | train rank=0 | step=5610/50000 progress=11.2% loss=0.119286 lr=9.894600e-05 lr_backbone=9.894600e-05 elapsed=05:42:59 avg_step=3.67s recent_step=3.26s eta=45:14:00 +2026-09-14 20:13:50 | INFO | train rank=0 | step=5620/50000 progress=11.2% loss=0.117321 lr=9.893923e-05 lr_backbone=9.893923e-05 elapsed=05:43:28 avg_step=3.67s recent_step=2.84s eta=45:12:18 +2026-09-14 20:14:21 | INFO | train rank=0 | step=5630/50000 progress=11.3% loss=0.101377 lr=9.893245e-05 lr_backbone=9.893245e-05 elapsed=05:43:59 avg_step=3.67s recent_step=3.11s eta=45:10:57 +2026-09-14 20:15:09 | INFO | train rank=0 | step=5640/50000 progress=11.3% loss=0.103779 lr=9.892564e-05 lr_backbone=9.892564e-05 elapsed=05:44:47 avg_step=3.67s recent_step=4.81s eta=45:11:51 +2026-09-14 20:15:47 | INFO | train rank=0 | step=5650/50000 progress=11.3% loss=0.097384 lr=9.891881e-05 lr_backbone=9.891881e-05 elapsed=05:45:25 avg_step=3.67s recent_step=3.85s eta=45:11:28 +2026-09-14 20:16:15 | INFO | train rank=0 | step=5660/50000 progress=11.3% loss=0.127539 lr=9.891196e-05 lr_backbone=9.891196e-05 elapsed=05:45:53 avg_step=3.67s recent_step=2.77s eta=45:09:41 +2026-09-14 20:16:46 | INFO | train rank=0 | step=5670/50000 progress=11.3% loss=0.127767 lr=9.890509e-05 lr_backbone=9.890509e-05 elapsed=05:46:24 avg_step=3.67s recent_step=3.12s eta=45:08:21 +2026-09-14 20:17:33 | INFO | train rank=0 | step=5680/50000 progress=11.4% loss=0.101987 lr=9.889819e-05 lr_backbone=9.889819e-05 elapsed=05:47:11 avg_step=3.67s recent_step=4.63s eta=45:09:00 +2026-09-14 20:18:07 | INFO | train rank=0 | step=5690/50000 progress=11.4% loss=0.112669 lr=9.889128e-05 lr_backbone=9.889128e-05 elapsed=05:47:45 avg_step=3.67s recent_step=3.45s eta=45:08:07 +2026-09-14 20:18:46 | INFO | train rank=0 | step=5700/50000 progress=11.4% loss=0.131003 lr=9.888434e-05 lr_backbone=9.888434e-05 elapsed=05:48:24 avg_step=3.67s recent_step=3.86s eta=45:07:45 +2026-09-14 20:19:16 | INFO | train rank=0 | step=5710/50000 progress=11.4% loss=0.136690 lr=9.887739e-05 lr_backbone=9.887739e-05 elapsed=05:48:54 avg_step=3.67s recent_step=3.07s eta=45:06:22 +2026-09-14 20:20:01 | INFO | train rank=0 | step=5720/50000 progress=11.4% loss=0.118459 lr=9.887041e-05 lr_backbone=9.887041e-05 elapsed=05:49:39 avg_step=3.67s recent_step=4.45s eta=45:06:46 +2026-09-14 20:20:28 | INFO | train rank=0 | step=5730/50000 progress=11.5% loss=0.143412 lr=9.886341e-05 lr_backbone=9.886341e-05 elapsed=05:50:06 avg_step=3.67s recent_step=2.69s eta=45:04:54 +2026-09-14 20:20:52 | INFO | train rank=0 | step=5740/50000 progress=11.5% loss=0.154965 lr=9.885638e-05 lr_backbone=9.885638e-05 elapsed=05:50:30 avg_step=3.66s recent_step=2.43s eta=45:02:42 +2026-09-14 20:21:29 | INFO | train rank=0 | step=5750/50000 progress=11.5% loss=0.118635 lr=9.884934e-05 lr_backbone=9.884934e-05 elapsed=05:51:07 avg_step=3.66s recent_step=3.65s eta=45:02:04 +2026-09-14 20:22:13 | INFO | train rank=0 | step=5760/50000 progress=11.5% loss=0.131064 lr=9.884228e-05 lr_backbone=9.884228e-05 elapsed=05:51:51 avg_step=3.67s recent_step=4.45s eta=45:02:28 +2026-09-14 20:22:50 | INFO | train rank=0 | step=5770/50000 progress=11.5% loss=0.085858 lr=9.883519e-05 lr_backbone=9.883519e-05 elapsed=05:52:28 avg_step=3.67s recent_step=3.68s eta=45:01:52 +2026-09-14 20:23:20 | INFO | train rank=0 | step=5780/50000 progress=11.6% loss=0.114947 lr=9.882808e-05 lr_backbone=9.882808e-05 elapsed=05:52:58 avg_step=3.66s recent_step=3.03s eta=45:00:27 +2026-09-14 20:23:49 | INFO | train rank=0 | step=5790/50000 progress=11.6% loss=0.106721 lr=9.882096e-05 lr_backbone=9.882096e-05 elapsed=05:53:27 avg_step=3.66s recent_step=2.90s eta=44:58:52 +2026-09-14 20:24:32 | INFO | train rank=0 | step=5800/50000 progress=11.6% loss=0.127532 lr=9.881381e-05 lr_backbone=9.881381e-05 elapsed=05:54:10 avg_step=3.66s recent_step=4.29s eta=44:59:03 +2026-09-14 20:25:11 | INFO | train rank=0 | step=5810/50000 progress=11.6% loss=0.110594 lr=9.880663e-05 lr_backbone=9.880663e-05 elapsed=05:54:49 avg_step=3.66s recent_step=3.85s eta=44:58:41 +2026-09-14 20:25:44 | INFO | train rank=0 | step=5820/50000 progress=11.6% loss=0.143548 lr=9.879944e-05 lr_backbone=9.879944e-05 elapsed=05:55:22 avg_step=3.66s recent_step=3.35s eta=44:57:40 +2026-09-14 20:26:17 | INFO | train rank=0 | step=5830/50000 progress=11.7% loss=0.099664 lr=9.879223e-05 lr_backbone=9.879223e-05 elapsed=05:55:55 avg_step=3.66s recent_step=3.31s eta=44:56:37 +2026-09-14 20:27:02 | INFO | train rank=0 | step=5840/50000 progress=11.7% loss=0.096538 lr=9.878499e-05 lr_backbone=9.878499e-05 elapsed=05:56:40 avg_step=3.66s recent_step=4.48s eta=44:57:03 +2026-09-14 20:27:35 | INFO | train rank=0 | step=5850/50000 progress=11.7% loss=0.121344 lr=9.877774e-05 lr_backbone=9.877774e-05 elapsed=05:57:13 avg_step=3.66s recent_step=3.27s eta=44:55:56 +2026-09-14 20:28:04 | INFO | train rank=0 | step=5860/50000 progress=11.7% loss=0.093015 lr=9.877046e-05 lr_backbone=9.877046e-05 elapsed=05:57:42 avg_step=3.66s recent_step=2.97s eta=44:54:27 +2026-09-14 20:28:35 | INFO | train rank=0 | step=5870/50000 progress=11.7% loss=0.116429 lr=9.876316e-05 lr_backbone=9.876316e-05 elapsed=05:58:13 avg_step=3.66s recent_step=3.04s eta=44:53:04 +2026-09-14 20:29:13 | INFO | train rank=0 | step=5880/50000 progress=11.8% loss=0.113464 lr=9.875584e-05 lr_backbone=9.875584e-05 elapsed=05:58:51 avg_step=3.66s recent_step=3.78s eta=44:52:36 +2026-09-14 20:29:46 | INFO | train rank=0 | step=5890/50000 progress=11.8% loss=0.117922 lr=9.874850e-05 lr_backbone=9.874850e-05 elapsed=05:59:24 avg_step=3.66s recent_step=3.32s eta=44:51:33 +2026-09-14 20:30:25 | INFO | train rank=0 | step=5900/50000 progress=11.8% loss=0.080875 lr=9.874113e-05 lr_backbone=9.874113e-05 elapsed=06:00:03 avg_step=3.66s recent_step=3.95s eta=44:51:18 +2026-09-14 20:31:03 | INFO | train rank=0 | step=5910/50000 progress=11.8% loss=0.146070 lr=9.873375e-05 lr_backbone=9.873375e-05 elapsed=06:00:41 avg_step=3.66s recent_step=3.75s eta=44:50:48 +2026-09-14 20:31:48 | INFO | train rank=0 | step=5920/50000 progress=11.8% loss=0.093046 lr=9.872634e-05 lr_backbone=9.872634e-05 elapsed=06:01:26 avg_step=3.66s recent_step=4.57s eta=44:51:19 +2026-09-14 20:32:19 | INFO | train rank=0 | step=5930/50000 progress=11.9% loss=0.127010 lr=9.871892e-05 lr_backbone=9.871892e-05 elapsed=06:01:57 avg_step=3.66s recent_step=3.10s eta=44:50:00 +2026-09-14 20:33:05 | INFO | train rank=0 | step=5940/50000 progress=11.9% loss=0.139755 lr=9.871147e-05 lr_backbone=9.871147e-05 elapsed=06:02:43 avg_step=3.66s recent_step=4.55s eta=44:50:30 +2026-09-14 20:33:42 | INFO | train rank=0 | step=5950/50000 progress=11.9% loss=0.128340 lr=9.870400e-05 lr_backbone=9.870400e-05 elapsed=06:03:20 avg_step=3.66s recent_step=3.67s eta=44:49:54 +2026-09-14 20:34:24 | INFO | train rank=0 | step=5960/50000 progress=11.9% loss=0.138401 lr=9.869651e-05 lr_backbone=9.869651e-05 elapsed=06:04:02 avg_step=3.66s recent_step=4.24s eta=44:50:00 +2026-09-14 20:34:52 | INFO | train rank=0 | step=5970/50000 progress=11.9% loss=0.143490 lr=9.868899e-05 lr_backbone=9.868899e-05 elapsed=06:04:30 avg_step=3.66s recent_step=2.77s eta=44:48:17 +2026-09-14 20:35:42 | INFO | train rank=0 | step=5980/50000 progress=12.0% loss=0.087950 lr=9.868146e-05 lr_backbone=9.868146e-05 elapsed=06:05:20 avg_step=3.67s recent_step=5.05s eta=44:49:23 +2026-09-14 20:36:12 | INFO | train rank=0 | step=5990/50000 progress=12.0% loss=0.108453 lr=9.867391e-05 lr_backbone=9.867391e-05 elapsed=06:05:50 avg_step=3.66s recent_step=3.00s eta=44:47:57 +2026-09-14 20:36:51 | INFO | train rank=0 | step=6000/50000 progress=12.0% loss=0.125561 lr=9.866633e-05 lr_backbone=9.866633e-05 elapsed=06:06:29 avg_step=3.66s recent_step=3.89s eta=44:47:37 +2026-09-14 20:37:21 | INFO | train rank=0 | step=6010/50000 progress=12.0% loss=0.182914 lr=9.865873e-05 lr_backbone=9.865873e-05 elapsed=06:06:59 avg_step=3.66s recent_step=2.98s eta=44:46:10 +2026-09-14 20:38:12 | INFO | train rank=0 | step=6020/50000 progress=12.0% loss=0.118791 lr=9.865111e-05 lr_backbone=9.865111e-05 elapsed=06:07:50 avg_step=3.67s recent_step=5.08s eta=44:47:17 +2026-09-14 20:38:43 | INFO | train rank=0 | step=6030/50000 progress=12.1% loss=0.144826 lr=9.864347e-05 lr_backbone=9.864347e-05 elapsed=06:08:21 avg_step=3.67s recent_step=3.12s eta=44:46:00 +2026-09-14 20:39:22 | INFO | train rank=0 | step=6040/50000 progress=12.1% loss=0.127394 lr=9.863581e-05 lr_backbone=9.863581e-05 elapsed=06:09:00 avg_step=3.67s recent_step=3.90s eta=44:45:40 +2026-09-14 20:39:54 | INFO | train rank=0 | step=6050/50000 progress=12.1% loss=0.125461 lr=9.862813e-05 lr_backbone=9.862813e-05 elapsed=06:09:32 avg_step=3.66s recent_step=3.22s eta=44:44:32 +2026-09-14 20:40:45 | INFO | train rank=0 | step=6060/50000 progress=12.1% loss=0.142824 lr=9.862043e-05 lr_backbone=9.862043e-05 elapsed=06:10:23 avg_step=3.67s recent_step=5.13s eta=44:45:41 +2026-09-14 20:41:13 | INFO | train rank=0 | step=6070/50000 progress=12.1% loss=0.102924 lr=9.861270e-05 lr_backbone=9.861270e-05 elapsed=06:10:51 avg_step=3.67s recent_step=2.71s eta=44:43:55 +2026-09-14 20:41:48 | INFO | train rank=0 | step=6080/50000 progress=12.2% loss=0.111577 lr=9.860495e-05 lr_backbone=9.860495e-05 elapsed=06:11:26 avg_step=3.67s recent_step=3.51s eta=44:43:08 +2026-09-14 20:42:21 | INFO | train rank=0 | step=6090/50000 progress=12.2% loss=0.142038 lr=9.859719e-05 lr_backbone=9.859719e-05 elapsed=06:11:59 avg_step=3.67s recent_step=3.38s eta=44:42:10 +2026-09-14 20:43:10 | INFO | train rank=0 | step=6100/50000 progress=12.2% loss=0.118646 lr=9.858940e-05 lr_backbone=9.858940e-05 elapsed=06:12:48 avg_step=3.67s recent_step=4.86s eta=44:43:00 +2026-09-14 20:43:44 | INFO | train rank=0 | step=6110/50000 progress=12.2% loss=0.093742 lr=9.858159e-05 lr_backbone=9.858159e-05 elapsed=06:13:22 avg_step=3.67s recent_step=3.42s eta=44:42:05 +2026-09-14 20:44:25 | INFO | train rank=0 | step=6120/50000 progress=12.2% loss=0.127672 lr=9.857376e-05 lr_backbone=9.857376e-05 elapsed=06:14:03 avg_step=3.67s recent_step=4.10s eta=44:41:59 +2026-09-14 20:45:00 | INFO | train rank=0 | step=6130/50000 progress=12.3% loss=0.107866 lr=9.856590e-05 lr_backbone=9.856590e-05 elapsed=06:14:38 avg_step=3.67s recent_step=3.50s eta=44:41:10 +2026-09-14 20:45:37 | INFO | train rank=0 | step=6140/50000 progress=12.3% loss=0.124612 lr=9.855803e-05 lr_backbone=9.855803e-05 elapsed=06:15:15 avg_step=3.67s recent_step=3.71s eta=44:40:37 +2026-09-14 20:46:09 | INFO | train rank=0 | step=6150/50000 progress=12.3% loss=0.137926 lr=9.855013e-05 lr_backbone=9.855013e-05 elapsed=06:15:47 avg_step=3.67s recent_step=3.18s eta=44:39:25 +2026-09-14 20:46:49 | INFO | train rank=0 | step=6160/50000 progress=12.3% loss=0.095418 lr=9.854222e-05 lr_backbone=9.854222e-05 elapsed=06:16:27 avg_step=3.67s recent_step=4.03s eta=44:39:14 +2026-09-14 20:47:24 | INFO | train rank=0 | step=6170/50000 progress=12.3% loss=0.111263 lr=9.853428e-05 lr_backbone=9.853428e-05 elapsed=06:17:02 avg_step=3.67s recent_step=3.47s eta=44:38:24 +2026-09-14 20:48:02 | INFO | train rank=0 | step=6180/50000 progress=12.4% loss=0.092888 lr=9.852632e-05 lr_backbone=9.852632e-05 elapsed=06:17:40 avg_step=3.67s recent_step=3.76s eta=44:37:54 +2026-09-14 20:48:43 | INFO | train rank=0 | step=6190/50000 progress=12.4% loss=0.140539 lr=9.851834e-05 lr_backbone=9.851834e-05 elapsed=06:18:21 avg_step=3.67s recent_step=4.15s eta=44:37:52 +2026-09-14 20:49:23 | INFO | train rank=0 | step=6200/50000 progress=12.4% loss=0.121290 lr=9.851034e-05 lr_backbone=9.851034e-05 elapsed=06:19:01 avg_step=3.67s recent_step=4.01s eta=44:37:39 +2026-09-14 20:50:02 | INFO | train rank=0 | step=6210/50000 progress=12.4% loss=0.121059 lr=9.850232e-05 lr_backbone=9.850232e-05 elapsed=06:19:40 avg_step=3.67s recent_step=3.90s eta=44:37:18 +2026-09-14 20:50:45 | INFO | train rank=0 | step=6220/50000 progress=12.4% loss=0.097636 lr=9.849427e-05 lr_backbone=9.849427e-05 elapsed=06:20:23 avg_step=3.67s recent_step=4.30s eta=44:37:26 +2026-09-14 20:51:18 | INFO | train rank=0 | step=6230/50000 progress=12.5% loss=0.168930 lr=9.848621e-05 lr_backbone=9.848621e-05 elapsed=06:20:56 avg_step=3.67s recent_step=3.26s eta=44:36:21 +2026-09-14 20:51:46 | INFO | train rank=0 | step=6240/50000 progress=12.5% loss=0.112280 lr=9.847812e-05 lr_backbone=9.847812e-05 elapsed=06:21:24 avg_step=3.67s recent_step=2.82s eta=44:34:45 +2026-09-14 20:52:18 | INFO | train rank=0 | step=6250/50000 progress=12.5% loss=0.112775 lr=9.847001e-05 lr_backbone=9.847001e-05 elapsed=06:21:56 avg_step=3.67s recent_step=3.20s eta=44:33:35 +2026-09-14 20:53:10 | INFO | train rank=0 | step=6260/50000 progress=12.5% loss=0.117020 lr=9.846188e-05 lr_backbone=9.846188e-05 elapsed=06:22:48 avg_step=3.67s recent_step=5.19s eta=44:34:45 +2026-09-14 20:53:39 | INFO | train rank=0 | step=6270/50000 progress=12.5% loss=0.128195 lr=9.845373e-05 lr_backbone=9.845373e-05 elapsed=06:23:17 avg_step=3.67s recent_step=2.92s eta=44:33:16 +2026-09-14 20:54:12 | INFO | train rank=0 | step=6280/50000 progress=12.6% loss=0.105729 lr=9.844556e-05 lr_backbone=9.844556e-05 elapsed=06:23:50 avg_step=3.67s recent_step=3.26s eta=44:32:11 +2026-09-14 20:54:49 | INFO | train rank=0 | step=6290/50000 progress=12.6% loss=0.118961 lr=9.843737e-05 lr_backbone=9.843737e-05 elapsed=06:24:27 avg_step=3.67s recent_step=3.68s eta=44:31:35 +2026-09-14 20:55:34 | INFO | train rank=0 | step=6300/50000 progress=12.6% loss=0.093373 lr=9.842916e-05 lr_backbone=9.842916e-05 elapsed=06:25:12 avg_step=3.67s recent_step=4.51s eta=44:31:57 +2026-09-14 20:56:02 | INFO | train rank=0 | step=6310/50000 progress=12.6% loss=0.111318 lr=9.842092e-05 lr_backbone=9.842092e-05 elapsed=06:25:40 avg_step=3.67s recent_step=2.82s eta=44:30:21 +2026-09-14 20:56:46 | INFO | train rank=0 | step=6320/50000 progress=12.6% loss=0.158254 lr=9.841267e-05 lr_backbone=9.841267e-05 elapsed=06:26:24 avg_step=3.67s recent_step=4.44s eta=44:30:38 +2026-09-14 20:57:15 | INFO | train rank=0 | step=6330/50000 progress=12.7% loss=0.117103 lr=9.840439e-05 lr_backbone=9.840439e-05 elapsed=06:26:53 avg_step=3.67s recent_step=2.89s eta=44:29:08 +2026-09-14 20:58:00 | INFO | train rank=0 | step=6340/50000 progress=12.7% loss=0.144824 lr=9.839609e-05 lr_backbone=9.839609e-05 elapsed=06:27:38 avg_step=3.67s recent_step=4.49s eta=44:29:28 +2026-09-14 20:58:39 | INFO | train rank=0 | step=6350/50000 progress=12.7% loss=0.120721 lr=9.838777e-05 lr_backbone=9.838777e-05 elapsed=06:28:18 avg_step=3.67s recent_step=3.95s eta=44:29:10 +2026-09-14 20:59:15 | INFO | train rank=0 | step=6360/50000 progress=12.7% loss=0.098424 lr=9.837943e-05 lr_backbone=9.837943e-05 elapsed=06:28:53 avg_step=3.67s recent_step=3.52s eta=44:28:24 +2026-09-14 20:59:45 | INFO | train rank=0 | step=6370/50000 progress=12.7% loss=0.138324 lr=9.837107e-05 lr_backbone=9.837107e-05 elapsed=06:29:23 avg_step=3.67s recent_step=3.07s eta=44:27:06 +2026-09-14 21:00:33 | INFO | train rank=0 | step=6380/50000 progress=12.8% loss=0.111487 lr=9.836269e-05 lr_backbone=9.836269e-05 elapsed=06:30:11 avg_step=3.67s recent_step=4.76s eta=44:27:44 +2026-09-14 21:01:03 | INFO | train rank=0 | step=6390/50000 progress=12.8% loss=0.104907 lr=9.835428e-05 lr_backbone=9.835428e-05 elapsed=06:30:41 avg_step=3.67s recent_step=3.00s eta=44:26:22 +2026-09-14 21:01:42 | INFO | train rank=0 | step=6400/50000 progress=12.8% loss=0.110486 lr=9.834586e-05 lr_backbone=9.834586e-05 elapsed=06:31:20 avg_step=3.67s recent_step=3.87s eta=44:25:59 +2026-09-14 21:02:17 | INFO | train rank=0 | step=6410/50000 progress=12.8% loss=0.107372 lr=9.833741e-05 lr_backbone=9.833741e-05 elapsed=06:31:55 avg_step=3.67s recent_step=3.48s eta=44:25:09 +2026-09-14 21:03:00 | INFO | train rank=0 | step=6420/50000 progress=12.8% loss=0.133402 lr=9.832895e-05 lr_backbone=9.832895e-05 elapsed=06:32:38 avg_step=3.67s recent_step=4.39s eta=44:25:22 +2026-09-14 21:03:34 | INFO | train rank=0 | step=6430/50000 progress=12.9% loss=0.112437 lr=9.832046e-05 lr_backbone=9.832046e-05 elapsed=06:33:12 avg_step=3.67s recent_step=3.34s eta=44:24:23 +2026-09-14 21:04:13 | INFO | train rank=0 | step=6440/50000 progress=12.9% loss=0.112805 lr=9.831195e-05 lr_backbone=9.831195e-05 elapsed=06:33:51 avg_step=3.67s recent_step=3.89s eta=44:24:01 +2026-09-14 21:04:40 | INFO | train rank=0 | step=6450/50000 progress=12.9% loss=0.121144 lr=9.830342e-05 lr_backbone=9.830342e-05 elapsed=06:34:18 avg_step=3.67s recent_step=2.69s eta=44:22:18 +2026-09-14 21:05:22 | INFO | train rank=0 | step=6460/50000 progress=12.9% loss=0.118938 lr=9.829486e-05 lr_backbone=9.829486e-05 elapsed=06:35:00 avg_step=3.67s recent_step=4.22s eta=44:22:19 +2026-09-14 21:05:48 | INFO | train rank=0 | step=6470/50000 progress=12.9% loss=0.140137 lr=9.828629e-05 lr_backbone=9.828629e-05 elapsed=06:35:26 avg_step=3.67s recent_step=2.58s eta=44:20:28 +2026-09-14 21:06:40 | INFO | train rank=0 | step=6480/50000 progress=13.0% loss=0.104725 lr=9.827770e-05 lr_backbone=9.827770e-05 elapsed=06:36:18 avg_step=3.67s recent_step=5.24s eta=44:21:37 +2026-09-14 21:07:16 | INFO | train rank=0 | step=6490/50000 progress=13.0% loss=0.101534 lr=9.826908e-05 lr_backbone=9.826908e-05 elapsed=06:36:54 avg_step=3.67s recent_step=3.59s eta=44:20:55 +2026-09-14 21:07:53 | INFO | train rank=0 | step=6500/50000 progress=13.0% loss=0.120316 lr=9.826045e-05 lr_backbone=9.826045e-05 elapsed=06:37:31 avg_step=3.67s recent_step=3.74s eta=44:20:23 +2026-09-14 21:08:29 | INFO | train rank=0 | step=6510/50000 progress=13.0% loss=0.128055 lr=9.825179e-05 lr_backbone=9.825179e-05 elapsed=06:38:07 avg_step=3.67s recent_step=3.53s eta=44:19:37 +2026-09-14 21:09:13 | INFO | train rank=0 | step=6520/50000 progress=13.0% loss=0.157181 lr=9.824311e-05 lr_backbone=9.824311e-05 elapsed=06:38:51 avg_step=3.67s recent_step=4.48s eta=44:19:54 +2026-09-14 21:09:47 | INFO | train rank=0 | step=6530/50000 progress=13.1% loss=0.114444 lr=9.823441e-05 lr_backbone=9.823441e-05 elapsed=06:39:25 avg_step=3.67s recent_step=3.40s eta=44:19:00 +2026-09-14 21:10:20 | INFO | train rank=0 | step=6540/50000 progress=13.1% loss=0.098464 lr=9.822569e-05 lr_backbone=9.822569e-05 elapsed=06:39:58 avg_step=3.67s recent_step=3.29s eta=44:17:57 +2026-09-14 21:10:54 | INFO | train rank=0 | step=6550/50000 progress=13.1% loss=0.105089 lr=9.821695e-05 lr_backbone=9.821695e-05 elapsed=06:40:32 avg_step=3.67s recent_step=3.34s eta=44:16:59 +2026-09-14 21:11:40 | INFO | train rank=0 | step=6560/50000 progress=13.1% loss=0.087390 lr=9.820818e-05 lr_backbone=9.820818e-05 elapsed=06:41:18 avg_step=3.67s recent_step=4.61s eta=44:17:25 +2026-09-14 21:12:17 | INFO | train rank=0 | step=6570/50000 progress=13.1% loss=0.126728 lr=9.819940e-05 lr_backbone=9.819940e-05 elapsed=06:41:55 avg_step=3.67s recent_step=3.75s eta=44:16:53 +2026-09-14 21:12:49 | INFO | train rank=0 | step=6580/50000 progress=13.2% loss=0.094640 lr=9.819059e-05 lr_backbone=9.819059e-05 elapsed=06:42:27 avg_step=3.67s recent_step=3.18s eta=44:15:44 +2026-09-14 21:13:32 | INFO | train rank=0 | step=6590/50000 progress=13.2% loss=0.128734 lr=9.818177e-05 lr_backbone=9.818177e-05 elapsed=06:43:10 avg_step=3.67s recent_step=4.28s eta=44:15:48 +2026-09-14 21:14:14 | INFO | train rank=0 | step=6600/50000 progress=13.2% loss=0.138108 lr=9.817292e-05 lr_backbone=9.817292e-05 elapsed=06:43:52 avg_step=3.67s recent_step=4.18s eta=44:15:44 +2026-09-14 21:14:51 | INFO | train rank=0 | step=6610/50000 progress=13.2% loss=0.123970 lr=9.816405e-05 lr_backbone=9.816405e-05 elapsed=06:44:29 avg_step=3.67s recent_step=3.78s eta=44:15:15 +2026-09-14 21:15:25 | INFO | train rank=0 | step=6620/50000 progress=13.2% loss=0.104148 lr=9.815516e-05 lr_backbone=9.815516e-05 elapsed=06:45:03 avg_step=3.67s recent_step=3.37s eta=44:14:19 +2026-09-14 21:16:03 | INFO | train rank=0 | step=6630/50000 progress=13.3% loss=0.112674 lr=9.814625e-05 lr_backbone=9.814625e-05 elapsed=06:45:41 avg_step=3.67s recent_step=3.81s eta=44:13:51 +2026-09-14 21:16:51 | INFO | train rank=0 | step=6640/50000 progress=13.3% loss=0.139350 lr=9.813732e-05 lr_backbone=9.813732e-05 elapsed=06:46:29 avg_step=3.67s recent_step=4.78s eta=44:14:27 +2026-09-14 21:17:23 | INFO | train rank=0 | step=6650/50000 progress=13.3% loss=0.089559 lr=9.812837e-05 lr_backbone=9.812837e-05 elapsed=06:47:01 avg_step=3.67s recent_step=3.21s eta=44:13:20 +2026-09-14 21:17:49 | INFO | train rank=0 | step=6660/50000 progress=13.3% loss=0.110346 lr=9.811939e-05 lr_backbone=9.811939e-05 elapsed=06:47:27 avg_step=3.67s recent_step=2.53s eta=44:11:29 +2026-09-14 21:18:25 | INFO | train rank=0 | step=6670/50000 progress=13.3% loss=0.107094 lr=9.811040e-05 lr_backbone=9.811040e-05 elapsed=06:48:03 avg_step=3.67s recent_step=3.64s eta=44:10:51 +2026-09-14 21:19:19 | INFO | train rank=0 | step=6680/50000 progress=13.4% loss=0.116832 lr=9.810138e-05 lr_backbone=9.810138e-05 elapsed=06:48:57 avg_step=3.67s recent_step=5.40s eta=44:12:06 +2026-09-14 21:19:49 | INFO | train rank=0 | step=6690/50000 progress=13.4% loss=0.115194 lr=9.809235e-05 lr_backbone=9.809235e-05 elapsed=06:49:27 avg_step=3.67s recent_step=3.03s eta=44:10:48 +2026-09-14 21:20:21 | INFO | train rank=0 | step=6700/50000 progress=13.4% loss=0.114740 lr=9.808329e-05 lr_backbone=9.808329e-05 elapsed=06:49:59 avg_step=3.67s recent_step=3.13s eta=44:09:36 +2026-09-14 21:20:50 | INFO | train rank=0 | step=6710/50000 progress=13.4% loss=0.106258 lr=9.807421e-05 lr_backbone=9.807421e-05 elapsed=06:50:28 avg_step=3.67s recent_step=2.95s eta=44:08:13 +2026-09-14 21:21:26 | INFO | train rank=0 | step=6720/50000 progress=13.4% loss=0.112445 lr=9.806511e-05 lr_backbone=9.806511e-05 elapsed=06:51:04 avg_step=3.67s recent_step=3.60s eta=44:07:32 +2026-09-14 21:21:56 | INFO | train rank=0 | step=6730/50000 progress=13.5% loss=0.107138 lr=9.805599e-05 lr_backbone=9.805599e-05 elapsed=06:51:34 avg_step=3.67s recent_step=3.01s eta=44:06:13 +2026-09-14 21:22:39 | INFO | train rank=0 | step=6740/50000 progress=13.5% loss=0.127404 lr=9.804685e-05 lr_backbone=9.804685e-05 elapsed=06:52:17 avg_step=3.67s recent_step=4.28s eta=44:06:15 +2026-09-14 21:23:17 | INFO | train rank=0 | step=6750/50000 progress=13.5% loss=0.091305 lr=9.803768e-05 lr_backbone=9.803768e-05 elapsed=06:52:55 avg_step=3.67s recent_step=3.76s eta=44:05:44 +2026-09-14 21:24:00 | INFO | train rank=0 | step=6760/50000 progress=13.5% loss=0.132963 lr=9.802850e-05 lr_backbone=9.802850e-05 elapsed=06:53:38 avg_step=3.67s recent_step=4.37s eta=44:05:53 +2026-09-14 21:24:34 | INFO | train rank=0 | step=6770/50000 progress=13.5% loss=0.104797 lr=9.801929e-05 lr_backbone=9.801929e-05 elapsed=06:54:12 avg_step=3.67s recent_step=3.35s eta=44:04:55 +2026-09-14 21:25:10 | INFO | train rank=0 | step=6780/50000 progress=13.6% loss=0.115482 lr=9.801007e-05 lr_backbone=9.801007e-05 elapsed=06:54:48 avg_step=3.67s recent_step=3.61s eta=44:04:15 +2026-09-14 21:25:43 | INFO | train rank=0 | step=6790/50000 progress=13.6% loss=0.094389 lr=9.800082e-05 lr_backbone=9.800082e-05 elapsed=06:55:21 avg_step=3.67s recent_step=3.30s eta=44:03:14 +2026-09-14 21:26:27 | INFO | train rank=0 | step=6800/50000 progress=13.6% loss=0.106501 lr=9.799155e-05 lr_backbone=9.799155e-05 elapsed=06:56:05 avg_step=3.67s recent_step=4.43s eta=44:03:26 +2026-09-14 21:27:02 | INFO | train rank=0 | step=6810/50000 progress=13.6% loss=0.102188 lr=9.798226e-05 lr_backbone=9.798226e-05 elapsed=06:56:40 avg_step=3.67s recent_step=3.48s eta=44:02:37 +2026-09-14 21:27:34 | INFO | train rank=0 | step=6820/50000 progress=13.6% loss=0.106887 lr=9.797295e-05 lr_backbone=9.797295e-05 elapsed=06:57:12 avg_step=3.67s recent_step=3.20s eta=44:01:30 +2026-09-14 21:28:05 | INFO | train rank=0 | step=6830/50000 progress=13.7% loss=0.101469 lr=9.796362e-05 lr_backbone=9.796362e-05 elapsed=06:57:43 avg_step=3.67s recent_step=3.10s eta=44:00:17 +2026-09-14 21:28:56 | INFO | train rank=0 | step=6840/50000 progress=13.7% loss=0.113729 lr=9.795427e-05 lr_backbone=9.795427e-05 elapsed=06:58:34 avg_step=3.67s recent_step=5.05s eta=44:01:08 +2026-09-14 21:29:28 | INFO | train rank=0 | step=6850/50000 progress=13.7% loss=0.120797 lr=9.794490e-05 lr_backbone=9.794490e-05 elapsed=06:59:06 avg_step=3.67s recent_step=3.25s eta=44:00:04 +2026-09-14 21:30:02 | INFO | train rank=0 | step=6860/50000 progress=13.7% loss=0.136496 lr=9.793550e-05 lr_backbone=9.793550e-05 elapsed=06:59:40 avg_step=3.67s recent_step=3.42s eta=43:59:12 +2026-09-14 21:30:32 | INFO | train rank=0 | step=6870/50000 progress=13.7% loss=0.134854 lr=9.792609e-05 lr_backbone=9.792609e-05 elapsed=07:00:10 avg_step=3.67s recent_step=3.01s eta=43:57:54 +2026-09-14 21:31:08 | INFO | train rank=0 | step=6880/50000 progress=13.8% loss=0.098129 lr=9.791665e-05 lr_backbone=9.791665e-05 elapsed=07:00:46 avg_step=3.67s recent_step=3.55s eta=43:57:10 +2026-09-14 21:31:39 | INFO | train rank=0 | step=6890/50000 progress=13.8% loss=0.114299 lr=9.790720e-05 lr_backbone=9.790720e-05 elapsed=07:01:17 avg_step=3.67s recent_step=3.07s eta=43:55:56 +2026-09-14 21:32:13 | INFO | train rank=0 | step=6900/50000 progress=13.8% loss=0.092008 lr=9.789772e-05 lr_backbone=9.789772e-05 elapsed=07:01:51 avg_step=3.67s recent_step=3.42s eta=43:55:04 +2026-09-14 21:32:38 | INFO | train rank=0 | step=6910/50000 progress=13.8% loss=0.102390 lr=9.788822e-05 lr_backbone=9.788822e-05 elapsed=07:02:16 avg_step=3.67s recent_step=2.50s eta=43:53:14 +2026-09-14 21:33:21 | INFO | train rank=0 | step=6920/50000 progress=13.8% loss=0.116469 lr=9.787870e-05 lr_backbone=9.787870e-05 elapsed=07:02:59 avg_step=3.67s recent_step=4.28s eta=43:53:16 +2026-09-14 21:33:40 | INFO | train rank=0 | step=6930/50000 progress=13.9% loss=0.153192 lr=9.786916e-05 lr_backbone=9.786916e-05 elapsed=07:03:18 avg_step=3.67s recent_step=1.98s eta=43:50:54 +2026-09-14 21:34:46 | INFO | train rank=0 | step=6940/50000 progress=13.9% loss=0.095132 lr=9.785960e-05 lr_backbone=9.785960e-05 elapsed=07:04:24 avg_step=3.67s recent_step=6.52s eta=43:53:15 +2026-09-14 21:35:15 | INFO | train rank=0 | step=6950/50000 progress=13.9% loss=0.113364 lr=9.785002e-05 lr_backbone=9.785002e-05 elapsed=07:04:53 avg_step=3.67s recent_step=2.96s eta=43:51:54 +2026-09-14 21:35:47 | INFO | train rank=0 | step=6960/50000 progress=13.9% loss=0.102205 lr=9.784041e-05 lr_backbone=9.784041e-05 elapsed=07:05:25 avg_step=3.67s recent_step=3.19s eta=43:50:47 +2026-09-14 21:36:24 | INFO | train rank=0 | step=6970/50000 progress=13.9% loss=0.103022 lr=9.783079e-05 lr_backbone=9.783079e-05 elapsed=07:06:02 avg_step=3.67s recent_step=3.70s eta=43:50:12 +2026-09-14 21:37:11 | INFO | train rank=0 | step=6980/50000 progress=14.0% loss=0.137687 lr=9.782114e-05 lr_backbone=9.782114e-05 elapsed=07:06:49 avg_step=3.67s recent_step=4.73s eta=43:50:41 +2026-09-14 21:37:49 | INFO | train rank=0 | step=6990/50000 progress=14.0% loss=0.124021 lr=9.781148e-05 lr_backbone=9.781148e-05 elapsed=07:07:27 avg_step=3.67s recent_step=3.75s eta=43:50:09 +2026-09-14 21:38:21 | INFO | train rank=0 | step=7000/50000 progress=14.0% loss=0.118148 lr=9.780179e-05 lr_backbone=9.780179e-05 elapsed=07:07:59 avg_step=3.67s recent_step=3.20s eta=43:49:04 +2026-09-14 21:38:53 | INFO | train rank=0 | step=7010/50000 progress=14.0% loss=0.112886 lr=9.779208e-05 lr_backbone=9.779208e-05 elapsed=07:08:31 avg_step=3.67s recent_step=3.22s eta=43:48:00 +2026-09-14 21:39:35 | INFO | train rank=0 | step=7020/50000 progress=14.0% loss=0.114862 lr=9.778235e-05 lr_backbone=9.778235e-05 elapsed=07:09:13 avg_step=3.67s recent_step=4.21s eta=43:47:56 +2026-09-14 21:40:07 | INFO | train rank=0 | step=7030/50000 progress=14.1% loss=0.125975 lr=9.777260e-05 lr_backbone=9.777260e-05 elapsed=07:09:45 avg_step=3.67s recent_step=3.17s eta=43:46:49 +2026-09-14 21:40:50 | INFO | train rank=0 | step=7040/50000 progress=14.1% loss=0.144890 lr=9.776283e-05 lr_backbone=9.776283e-05 elapsed=07:10:28 avg_step=3.67s recent_step=4.33s eta=43:46:53 +2026-09-14 21:41:22 | INFO | train rank=0 | step=7050/50000 progress=14.1% loss=0.116994 lr=9.775304e-05 lr_backbone=9.775304e-05 elapsed=07:11:00 avg_step=3.67s recent_step=3.18s eta=43:45:46 +2026-09-14 21:41:56 | INFO | train rank=0 | step=7060/50000 progress=14.1% loss=0.118522 lr=9.774323e-05 lr_backbone=9.774323e-05 elapsed=07:11:34 avg_step=3.67s recent_step=3.42s eta=43:44:54 +2026-09-14 21:42:26 | INFO | train rank=0 | step=7070/50000 progress=14.1% loss=0.098578 lr=9.773339e-05 lr_backbone=9.773339e-05 elapsed=07:12:04 avg_step=3.67s recent_step=3.02s eta=43:43:38 +2026-09-14 21:43:06 | INFO | train rank=0 | step=7080/50000 progress=14.2% loss=0.127621 lr=9.772354e-05 lr_backbone=9.772354e-05 elapsed=07:12:44 avg_step=3.67s recent_step=3.92s eta=43:43:17 +2026-09-14 21:43:40 | INFO | train rank=0 | step=7090/50000 progress=14.2% loss=0.110258 lr=9.771366e-05 lr_backbone=9.771366e-05 elapsed=07:13:18 avg_step=3.67s recent_step=3.43s eta=43:42:26 +2026-09-14 21:44:11 | INFO | train rank=0 | step=7100/50000 progress=14.2% loss=0.132634 lr=9.770377e-05 lr_backbone=9.770377e-05 elapsed=07:13:49 avg_step=3.67s recent_step=3.14s eta=43:41:18 +2026-09-14 21:44:55 | INFO | train rank=0 | step=7110/50000 progress=14.2% loss=0.087843 lr=9.769385e-05 lr_backbone=9.769385e-05 elapsed=07:14:33 avg_step=3.67s recent_step=4.38s eta=43:41:25 +2026-09-14 21:45:23 | INFO | train rank=0 | step=7120/50000 progress=14.2% loss=0.104339 lr=9.768391e-05 lr_backbone=9.768391e-05 elapsed=07:15:01 avg_step=3.67s recent_step=2.79s eta=43:39:55 +2026-09-14 21:45:57 | INFO | train rank=0 | step=7130/50000 progress=14.3% loss=0.104427 lr=9.767395e-05 lr_backbone=9.767395e-05 elapsed=07:15:35 avg_step=3.67s recent_step=3.42s eta=43:39:04 +2026-09-14 21:46:37 | INFO | train rank=0 | step=7140/50000 progress=14.3% loss=0.140014 lr=9.766398e-05 lr_backbone=9.766398e-05 elapsed=07:16:15 avg_step=3.67s recent_step=3.98s eta=43:38:46 +2026-09-14 21:47:19 | INFO | train rank=0 | step=7150/50000 progress=14.3% loss=0.108835 lr=9.765398e-05 lr_backbone=9.765398e-05 elapsed=07:16:57 avg_step=3.67s recent_step=4.17s eta=43:38:40 +2026-09-14 21:47:57 | INFO | train rank=0 | step=7160/50000 progress=14.3% loss=0.106717 lr=9.764395e-05 lr_backbone=9.764395e-05 elapsed=07:17:35 avg_step=3.67s recent_step=3.82s eta=43:38:12 +2026-09-14 21:48:37 | INFO | train rank=0 | step=7170/50000 progress=14.3% loss=0.114117 lr=9.763391e-05 lr_backbone=9.763391e-05 elapsed=07:18:15 avg_step=3.67s recent_step=4.03s eta=43:37:57 +2026-09-14 21:49:20 | INFO | train rank=0 | step=7180/50000 progress=14.4% loss=0.131450 lr=9.762385e-05 lr_backbone=9.762385e-05 elapsed=07:18:58 avg_step=3.67s recent_step=4.32s eta=43:37:59 +2026-09-14 21:50:09 | INFO | train rank=0 | step=7190/50000 progress=14.4% loss=0.113192 lr=9.761377e-05 lr_backbone=9.761377e-05 elapsed=07:19:47 avg_step=3.67s recent_step=4.84s eta=43:38:32 +2026-09-14 21:50:39 | INFO | train rank=0 | step=7200/50000 progress=14.4% loss=0.112679 lr=9.760366e-05 lr_backbone=9.760366e-05 elapsed=07:20:17 avg_step=3.67s recent_step=3.06s eta=43:37:20 +2026-09-14 21:51:14 | INFO | train rank=0 | step=7210/50000 progress=14.4% loss=0.117404 lr=9.759354e-05 lr_backbone=9.759354e-05 elapsed=07:20:52 avg_step=3.67s recent_step=3.47s eta=43:36:31 +2026-09-14 21:51:47 | INFO | train rank=0 | step=7220/50000 progress=14.4% loss=0.113961 lr=9.758339e-05 lr_backbone=9.758339e-05 elapsed=07:21:25 avg_step=3.67s recent_step=3.30s eta=43:35:32 +2026-09-14 21:52:27 | INFO | train rank=0 | step=7230/50000 progress=14.5% loss=0.124158 lr=9.757322e-05 lr_backbone=9.757322e-05 elapsed=07:22:05 avg_step=3.67s recent_step=4.00s eta=43:35:16 +2026-09-14 21:52:56 | INFO | train rank=0 | step=7240/50000 progress=14.5% loss=0.138909 lr=9.756303e-05 lr_backbone=9.756303e-05 elapsed=07:22:34 avg_step=3.67s recent_step=2.86s eta=43:33:51 +2026-09-14 21:53:32 | INFO | train rank=0 | step=7250/50000 progress=14.5% loss=0.109378 lr=9.755283e-05 lr_backbone=9.755283e-05 elapsed=07:23:10 avg_step=3.67s recent_step=3.61s eta=43:33:11 +2026-09-14 21:54:13 | INFO | train rank=0 | step=7260/50000 progress=14.5% loss=0.112963 lr=9.754260e-05 lr_backbone=9.754260e-05 elapsed=07:23:51 avg_step=3.67s recent_step=4.11s eta=43:33:00 +2026-09-14 21:55:07 | INFO | train rank=0 | step=7270/50000 progress=14.5% loss=0.119974 lr=9.753235e-05 lr_backbone=9.753235e-05 elapsed=07:24:45 avg_step=3.67s recent_step=5.45s eta=43:34:08 +2026-09-14 21:55:36 | INFO | train rank=0 | step=7280/50000 progress=14.6% loss=0.154767 lr=9.752208e-05 lr_backbone=9.752208e-05 elapsed=07:25:14 avg_step=3.67s recent_step=2.83s eta=43:32:42 +2026-09-14 21:56:12 | INFO | train rank=0 | step=7290/50000 progress=14.6% loss=0.126197 lr=9.751178e-05 lr_backbone=9.751178e-05 elapsed=07:25:50 avg_step=3.67s recent_step=3.61s eta=43:32:02 +2026-09-14 21:56:42 | INFO | train rank=0 | step=7300/50000 progress=14.6% loss=0.098215 lr=9.750147e-05 lr_backbone=9.750147e-05 elapsed=07:26:20 avg_step=3.67s recent_step=3.06s eta=43:30:49 +2026-09-14 21:57:36 | INFO | train rank=0 | step=7310/50000 progress=14.6% loss=0.126412 lr=9.749114e-05 lr_backbone=9.749114e-05 elapsed=07:27:14 avg_step=3.67s recent_step=5.39s eta=43:31:53 +2026-09-14 21:58:02 | INFO | train rank=0 | step=7320/50000 progress=14.6% loss=0.093398 lr=9.748078e-05 lr_backbone=9.748078e-05 elapsed=07:27:40 avg_step=3.67s recent_step=2.61s eta=43:30:15 +2026-09-14 21:58:39 | INFO | train rank=0 | step=7330/50000 progress=14.7% loss=0.106950 lr=9.747041e-05 lr_backbone=9.747041e-05 elapsed=07:28:17 avg_step=3.67s recent_step=3.64s eta=43:29:37 +2026-09-14 21:59:11 | INFO | train rank=0 | step=7340/50000 progress=14.7% loss=0.108349 lr=9.746001e-05 lr_backbone=9.746001e-05 elapsed=07:28:49 avg_step=3.67s recent_step=3.25s eta=43:28:35 +2026-09-14 21:59:59 | INFO | train rank=0 | step=7350/50000 progress=14.7% loss=0.114911 lr=9.744960e-05 lr_backbone=9.744960e-05 elapsed=07:29:37 avg_step=3.67s recent_step=4.78s eta=43:29:03 +2026-09-14 22:00:34 | INFO | train rank=0 | step=7360/50000 progress=14.7% loss=0.102994 lr=9.743916e-05 lr_backbone=9.743916e-05 elapsed=07:30:12 avg_step=3.67s recent_step=3.53s eta=43:28:18 +2026-09-14 22:01:07 | INFO | train rank=0 | step=7370/50000 progress=14.7% loss=0.149240 lr=9.742870e-05 lr_backbone=9.742870e-05 elapsed=07:30:45 avg_step=3.67s recent_step=3.21s eta=43:27:15 +2026-09-14 22:01:42 | INFO | train rank=0 | step=7380/50000 progress=14.8% loss=0.125325 lr=9.741822e-05 lr_backbone=9.741822e-05 elapsed=07:31:20 avg_step=3.67s recent_step=3.55s eta=43:26:31 +2026-09-14 22:02:29 | INFO | train rank=0 | step=7390/50000 progress=14.8% loss=0.111012 lr=9.740772e-05 lr_backbone=9.740772e-05 elapsed=07:32:07 avg_step=3.67s recent_step=4.72s eta=43:26:55 +2026-09-14 22:03:05 | INFO | train rank=0 | step=7400/50000 progress=14.8% loss=0.125346 lr=9.739720e-05 lr_backbone=9.739720e-05 elapsed=07:32:43 avg_step=3.67s recent_step=3.58s eta=43:26:13 +2026-09-14 22:03:35 | INFO | train rank=0 | step=7410/50000 progress=14.8% loss=0.125473 lr=9.738666e-05 lr_backbone=9.738666e-05 elapsed=07:33:13 avg_step=3.67s recent_step=2.95s eta=43:24:55 +2026-09-14 22:04:08 | INFO | train rank=0 | step=7420/50000 progress=14.8% loss=0.122869 lr=9.737610e-05 lr_backbone=9.737610e-05 elapsed=07:33:46 avg_step=3.67s recent_step=3.34s eta=43:24:00 +2026-09-14 22:04:55 | INFO | train rank=0 | step=7430/50000 progress=14.9% loss=0.136780 lr=9.736552e-05 lr_backbone=9.736552e-05 elapsed=07:34:33 avg_step=3.67s recent_step=4.70s eta=43:24:22 +2026-09-14 22:05:24 | INFO | train rank=0 | step=7440/50000 progress=14.9% loss=0.093995 lr=9.735492e-05 lr_backbone=9.735492e-05 elapsed=07:35:02 avg_step=3.67s recent_step=2.88s eta=43:23:00 +2026-09-14 22:05:54 | INFO | train rank=0 | step=7450/50000 progress=14.9% loss=0.103499 lr=9.734429e-05 lr_backbone=9.734429e-05 elapsed=07:35:32 avg_step=3.67s recent_step=2.98s eta=43:21:44 +2026-09-14 22:06:31 | INFO | train rank=0 | step=7460/50000 progress=14.9% loss=0.137591 lr=9.733365e-05 lr_backbone=9.733365e-05 elapsed=07:36:09 avg_step=3.67s recent_step=3.74s eta=43:21:12 +2026-09-14 22:07:13 | INFO | train rank=0 | step=7470/50000 progress=14.9% loss=0.123644 lr=9.732298e-05 lr_backbone=9.732298e-05 elapsed=07:36:51 avg_step=3.67s recent_step=4.16s eta=43:21:03 +2026-09-14 22:07:54 | INFO | train rank=0 | step=7480/50000 progress=15.0% loss=0.102674 lr=9.731230e-05 lr_backbone=9.731230e-05 elapsed=07:37:32 avg_step=3.67s recent_step=4.17s eta=43:20:54 +2026-09-14 22:08:27 | INFO | train rank=0 | step=7490/50000 progress=15.0% loss=0.107433 lr=9.730159e-05 lr_backbone=9.730159e-05 elapsed=07:38:05 avg_step=3.67s recent_step=3.26s eta=43:19:54 +2026-09-14 22:09:13 | INFO | train rank=0 | step=7500/50000 progress=15.0% loss=0.091765 lr=9.729086e-05 lr_backbone=9.729086e-05 elapsed=07:38:51 avg_step=3.67s recent_step=4.58s eta=43:20:09 +2026-09-14 22:09:43 | INFO | train rank=0 | step=7510/50000 progress=15.0% loss=0.137176 lr=9.728011e-05 lr_backbone=9.728011e-05 elapsed=07:39:21 avg_step=3.67s recent_step=3.02s eta=43:18:56 +2026-09-14 22:10:21 | INFO | train rank=0 | step=7520/50000 progress=15.0% loss=0.115249 lr=9.726935e-05 lr_backbone=9.726935e-05 elapsed=07:39:59 avg_step=3.67s recent_step=3.86s eta=43:18:30 +2026-09-14 22:10:54 | INFO | train rank=0 | step=7530/50000 progress=15.1% loss=0.125202 lr=9.725856e-05 lr_backbone=9.725856e-05 elapsed=07:40:32 avg_step=3.67s recent_step=3.24s eta=43:17:29 +2026-09-14 22:11:43 | INFO | train rank=0 | step=7540/50000 progress=15.1% loss=0.130516 lr=9.724775e-05 lr_backbone=9.724775e-05 elapsed=07:41:21 avg_step=3.67s recent_step=4.95s eta=43:18:04 +2026-09-14 22:12:17 | INFO | train rank=0 | step=7550/50000 progress=15.1% loss=0.111236 lr=9.723692e-05 lr_backbone=9.723692e-05 elapsed=07:41:55 avg_step=3.67s recent_step=3.34s eta=43:17:09 +2026-09-14 22:12:55 | INFO | train rank=0 | step=7560/50000 progress=15.1% loss=0.136244 lr=9.722606e-05 lr_backbone=9.722606e-05 elapsed=07:42:33 avg_step=3.67s recent_step=3.82s eta=43:16:40 +2026-09-14 22:13:27 | INFO | train rank=0 | step=7570/50000 progress=15.1% loss=0.096161 lr=9.721519e-05 lr_backbone=9.721519e-05 elapsed=07:43:05 avg_step=3.67s recent_step=3.27s eta=43:15:41 +2026-09-14 22:14:15 | INFO | train rank=0 | step=7580/50000 progress=15.2% loss=0.114059 lr=9.720430e-05 lr_backbone=9.720430e-05 elapsed=07:43:53 avg_step=3.67s recent_step=4.79s eta=43:16:06 +2026-09-14 22:14:50 | INFO | train rank=0 | step=7590/50000 progress=15.2% loss=0.092487 lr=9.719339e-05 lr_backbone=9.719339e-05 elapsed=07:44:28 avg_step=3.67s recent_step=3.49s eta=43:15:19 +2026-09-14 22:15:30 | INFO | train rank=0 | step=7600/50000 progress=15.2% loss=0.102669 lr=9.718245e-05 lr_backbone=9.718245e-05 elapsed=07:45:08 avg_step=3.67s recent_step=4.02s eta=43:15:02 +2026-09-14 22:16:04 | INFO | train rank=0 | step=7610/50000 progress=15.2% loss=0.107493 lr=9.717150e-05 lr_backbone=9.717150e-05 elapsed=07:45:42 avg_step=3.67s recent_step=3.32s eta=43:14:05 +2026-09-14 22:16:49 | INFO | train rank=0 | step=7620/50000 progress=15.2% loss=0.136022 lr=9.716052e-05 lr_backbone=9.716052e-05 elapsed=07:46:27 avg_step=3.67s recent_step=4.55s eta=43:14:18 +2026-09-14 22:17:18 | INFO | train rank=0 | step=7630/50000 progress=15.3% loss=0.149858 lr=9.714953e-05 lr_backbone=9.714953e-05 elapsed=07:46:56 avg_step=3.67s recent_step=2.93s eta=43:13:00 +2026-09-14 22:17:49 | INFO | train rank=0 | step=7640/50000 progress=15.3% loss=0.147348 lr=9.713851e-05 lr_backbone=9.713851e-05 elapsed=07:47:27 avg_step=3.67s recent_step=3.07s eta=43:11:50 +2026-09-14 22:18:22 | INFO | train rank=0 | step=7650/50000 progress=15.3% loss=0.106244 lr=9.712747e-05 lr_backbone=9.712747e-05 elapsed=07:48:00 avg_step=3.67s recent_step=3.30s eta=43:10:53 +2026-09-14 22:19:14 | INFO | train rank=0 | step=7660/50000 progress=15.3% loss=0.110062 lr=9.711642e-05 lr_backbone=9.711642e-05 elapsed=07:48:52 avg_step=3.67s recent_step=5.17s eta=43:11:39 +2026-09-14 22:19:43 | INFO | train rank=0 | step=7670/50000 progress=15.3% loss=0.091971 lr=9.710534e-05 lr_backbone=9.710534e-05 elapsed=07:49:21 avg_step=3.67s recent_step=2.88s eta=43:10:18 +2026-09-14 22:20:14 | INFO | train rank=0 | step=7680/50000 progress=15.4% loss=0.106068 lr=9.709424e-05 lr_backbone=9.709424e-05 elapsed=07:49:52 avg_step=3.67s recent_step=3.15s eta=43:09:13 +2026-09-14 22:20:46 | INFO | train rank=0 | step=7690/50000 progress=15.4% loss=0.133562 lr=9.708312e-05 lr_backbone=9.708312e-05 elapsed=07:50:24 avg_step=3.67s recent_step=3.19s eta=43:08:10 +2026-09-14 22:21:36 | INFO | train rank=0 | step=7700/50000 progress=15.4% loss=0.124687 lr=9.707198e-05 lr_backbone=9.707198e-05 elapsed=07:51:14 avg_step=3.67s recent_step=4.99s eta=43:08:45 +2026-09-14 22:22:07 | INFO | train rank=0 | step=7710/50000 progress=15.4% loss=0.108940 lr=9.706082e-05 lr_backbone=9.706082e-05 elapsed=07:51:45 avg_step=3.67s recent_step=3.08s eta=43:07:36 +2026-09-14 22:22:41 | INFO | train rank=0 | step=7720/50000 progress=15.4% loss=0.128579 lr=9.704964e-05 lr_backbone=9.704964e-05 elapsed=07:52:19 avg_step=3.67s recent_step=3.40s eta=43:06:45 +2026-09-14 22:23:10 | INFO | train rank=0 | step=7730/50000 progress=15.5% loss=0.130473 lr=9.703843e-05 lr_backbone=9.703843e-05 elapsed=07:52:48 avg_step=3.67s recent_step=2.87s eta=43:05:25 +2026-09-14 22:23:48 | INFO | train rank=0 | step=7740/50000 progress=15.5% loss=0.091458 lr=9.702721e-05 lr_backbone=9.702721e-05 elapsed=07:53:26 avg_step=3.67s recent_step=3.88s eta=43:05:00 +2026-09-14 22:24:26 | INFO | train rank=0 | step=7750/50000 progress=15.5% loss=0.102286 lr=9.701597e-05 lr_backbone=9.701597e-05 elapsed=07:54:04 avg_step=3.67s recent_step=3.80s eta=43:04:30 +2026-09-14 22:24:56 | INFO | train rank=0 | step=7760/50000 progress=15.5% loss=0.112480 lr=9.700471e-05 lr_backbone=9.700471e-05 elapsed=07:54:34 avg_step=3.67s recent_step=2.98s eta=43:03:16 +2026-09-14 22:25:34 | INFO | train rank=0 | step=7770/50000 progress=15.5% loss=0.117652 lr=9.699342e-05 lr_backbone=9.699342e-05 elapsed=07:55:12 avg_step=3.67s recent_step=3.79s eta=43:02:46 +2026-09-14 22:26:41 | INFO | train rank=0 | step=7780/50000 progress=15.6% loss=0.129953 lr=9.698212e-05 lr_backbone=9.698212e-05 elapsed=07:56:19 avg_step=3.67s recent_step=6.71s eta=43:04:54 +2026-09-14 22:27:11 | INFO | train rank=0 | step=7790/50000 progress=15.6% loss=0.121289 lr=9.697079e-05 lr_backbone=9.697079e-05 elapsed=07:56:49 avg_step=3.67s recent_step=2.95s eta=43:03:38 +2026-09-14 22:27:47 | INFO | train rank=0 | step=7800/50000 progress=15.6% loss=0.116056 lr=9.695945e-05 lr_backbone=9.695945e-05 elapsed=07:57:25 avg_step=3.67s recent_step=3.68s eta=43:03:01 +2026-09-14 22:28:22 | INFO | train rank=0 | step=7810/50000 progress=15.6% loss=0.110963 lr=9.694808e-05 lr_backbone=9.694808e-05 elapsed=07:58:00 avg_step=3.67s recent_step=3.50s eta=43:02:15 +2026-09-14 22:29:15 | INFO | train rank=0 | step=7820/50000 progress=15.6% loss=0.112765 lr=9.693669e-05 lr_backbone=9.693669e-05 elapsed=07:58:53 avg_step=3.67s recent_step=5.27s eta=43:03:04 +2026-09-14 22:29:48 | INFO | train rank=0 | step=7830/50000 progress=15.7% loss=0.102620 lr=9.692529e-05 lr_backbone=9.692529e-05 elapsed=07:59:26 avg_step=3.67s recent_step=3.25s eta=43:02:05 +2026-09-14 22:30:21 | INFO | train rank=0 | step=7840/50000 progress=15.7% loss=0.088199 lr=9.691386e-05 lr_backbone=9.691386e-05 elapsed=07:59:59 avg_step=3.67s recent_step=3.32s eta=43:01:09 +2026-09-14 22:31:11 | INFO | train rank=0 | step=7850/50000 progress=15.7% loss=0.137558 lr=9.690241e-05 lr_backbone=9.690241e-05 elapsed=08:00:49 avg_step=3.68s recent_step=5.05s eta=43:01:46 +2026-09-14 22:31:45 | INFO | train rank=0 | step=7860/50000 progress=15.7% loss=0.095943 lr=9.689094e-05 lr_backbone=9.689094e-05 elapsed=08:01:23 avg_step=3.67s recent_step=3.35s eta=43:00:52 +2026-09-14 22:32:23 | INFO | train rank=0 | step=7870/50000 progress=15.7% loss=0.112390 lr=9.687945e-05 lr_backbone=9.687945e-05 elapsed=08:02:01 avg_step=3.67s recent_step=3.80s eta=43:00:22 +2026-09-14 22:32:54 | INFO | train rank=0 | step=7880/50000 progress=15.8% loss=0.119823 lr=9.686794e-05 lr_backbone=9.686794e-05 elapsed=08:02:32 avg_step=3.67s recent_step=3.11s eta=42:59:15 +2026-09-14 22:33:30 | INFO | train rank=0 | step=7890/50000 progress=15.8% loss=0.142705 lr=9.685641e-05 lr_backbone=9.685641e-05 elapsed=08:03:08 avg_step=3.67s recent_step=3.58s eta=42:58:33 +2026-09-14 22:34:11 | INFO | train rank=0 | step=7900/50000 progress=15.8% loss=0.125839 lr=9.684486e-05 lr_backbone=9.684486e-05 elapsed=08:03:49 avg_step=3.67s recent_step=4.13s eta=42:58:20 +2026-09-14 22:34:43 | INFO | train rank=0 | step=7910/50000 progress=15.8% loss=0.150543 lr=9.683329e-05 lr_backbone=9.683329e-05 elapsed=08:04:21 avg_step=3.67s recent_step=3.17s eta=42:57:17 +2026-09-14 22:35:24 | INFO | train rank=0 | step=7920/50000 progress=15.8% loss=0.104492 lr=9.682170e-05 lr_backbone=9.682170e-05 elapsed=08:05:02 avg_step=3.67s recent_step=4.15s eta=42:57:06 +2026-09-14 22:36:03 | INFO | train rank=0 | step=7930/50000 progress=15.9% loss=0.095395 lr=9.681008e-05 lr_backbone=9.681008e-05 elapsed=08:05:41 avg_step=3.67s recent_step=3.86s eta=42:56:38 +2026-09-14 22:36:54 | INFO | train rank=0 | step=7940/50000 progress=15.9% loss=0.124923 lr=9.679845e-05 lr_backbone=9.679845e-05 elapsed=08:06:32 avg_step=3.68s recent_step=5.18s eta=42:57:21 +2026-09-14 22:37:35 | INFO | train rank=0 | step=7950/50000 progress=15.9% loss=0.099545 lr=9.678680e-05 lr_backbone=9.678680e-05 elapsed=08:07:13 avg_step=3.68s recent_step=4.09s eta=42:57:06 +2026-09-14 22:38:02 | INFO | train rank=0 | step=7960/50000 progress=15.9% loss=0.103815 lr=9.677512e-05 lr_backbone=9.677512e-05 elapsed=08:07:40 avg_step=3.68s recent_step=2.70s eta=42:55:38 +2026-09-14 22:38:35 | INFO | train rank=0 | step=7970/50000 progress=15.9% loss=0.118671 lr=9.676343e-05 lr_backbone=9.676343e-05 elapsed=08:08:13 avg_step=3.68s recent_step=3.25s eta=42:54:38 +2026-09-14 22:39:26 | INFO | train rank=0 | step=7980/50000 progress=16.0% loss=0.123688 lr=9.675171e-05 lr_backbone=9.675171e-05 elapsed=08:09:04 avg_step=3.68s recent_step=5.16s eta=42:55:20 +2026-09-14 22:40:00 | INFO | train rank=0 | step=7990/50000 progress=16.0% loss=0.130774 lr=9.673998e-05 lr_backbone=9.673998e-05 elapsed=08:09:38 avg_step=3.68s recent_step=3.40s eta=42:54:29 +2026-09-14 22:40:31 | INFO | train rank=0 | step=8000/50000 progress=16.0% loss=0.131709 lr=9.672822e-05 lr_backbone=9.672822e-05 elapsed=08:10:09 avg_step=3.68s recent_step=3.06s eta=42:53:19 +2026-09-14 22:41:03 | INFO | train rank=0 | step=8010/50000 progress=16.0% loss=0.116401 lr=9.671645e-05 lr_backbone=9.671645e-05 elapsed=08:10:41 avg_step=3.68s recent_step=3.18s eta=42:52:17 +2026-09-14 22:41:41 | INFO | train rank=0 | step=8020/50000 progress=16.0% loss=0.102807 lr=9.670465e-05 lr_backbone=9.670465e-05 elapsed=08:11:19 avg_step=3.68s recent_step=3.84s eta=42:51:49 +2026-09-14 22:42:24 | INFO | train rank=0 | step=8030/50000 progress=16.1% loss=0.085835 lr=9.669283e-05 lr_backbone=9.669283e-05 elapsed=08:12:02 avg_step=3.68s recent_step=4.30s eta=42:51:45 +2026-09-14 22:42:53 | INFO | train rank=0 | step=8040/50000 progress=16.1% loss=0.098170 lr=9.668100e-05 lr_backbone=9.668100e-05 elapsed=08:12:31 avg_step=3.68s recent_step=2.91s eta=42:50:28 +2026-09-14 22:43:33 | INFO | train rank=0 | step=8050/50000 progress=16.1% loss=0.110688 lr=9.666914e-05 lr_backbone=9.666914e-05 elapsed=08:13:11 avg_step=3.68s recent_step=3.94s eta=42:50:05 +2026-09-14 22:44:15 | INFO | train rank=0 | step=8060/50000 progress=16.1% loss=0.113941 lr=9.665726e-05 lr_backbone=9.665726e-05 elapsed=08:13:53 avg_step=3.68s recent_step=4.22s eta=42:49:56 +2026-09-14 22:44:55 | INFO | train rank=0 | step=8070/50000 progress=16.1% loss=0.094169 lr=9.664536e-05 lr_backbone=9.664536e-05 elapsed=08:14:33 avg_step=3.68s recent_step=4.05s eta=42:49:39 +2026-09-14 22:45:23 | INFO | train rank=0 | step=8080/50000 progress=16.2% loss=0.127960 lr=9.663344e-05 lr_backbone=9.663344e-05 elapsed=08:15:01 avg_step=3.68s recent_step=2.73s eta=42:48:13 +2026-09-14 22:46:00 | INFO | train rank=0 | step=8090/50000 progress=16.2% loss=0.122616 lr=9.662150e-05 lr_backbone=9.662150e-05 elapsed=08:15:38 avg_step=3.68s recent_step=3.71s eta=42:47:38 +2026-09-14 22:46:47 | INFO | train rank=0 | step=8100/50000 progress=16.2% loss=0.106267 lr=9.660954e-05 lr_backbone=9.660954e-05 elapsed=08:16:25 avg_step=3.68s recent_step=4.71s eta=42:47:55 +2026-09-14 22:47:21 | INFO | train rank=0 | step=8110/50000 progress=16.2% loss=0.101063 lr=9.659756e-05 lr_backbone=9.659756e-05 elapsed=08:16:59 avg_step=3.68s recent_step=3.36s eta=42:47:02 +2026-09-14 22:47:54 | INFO | train rank=0 | step=8120/50000 progress=16.2% loss=0.102909 lr=9.658556e-05 lr_backbone=9.658556e-05 elapsed=08:17:32 avg_step=3.68s recent_step=3.34s eta=42:46:08 +2026-09-14 22:48:30 | INFO | train rank=0 | step=8130/50000 progress=16.3% loss=0.134955 lr=9.657354e-05 lr_backbone=9.657354e-05 elapsed=08:18:08 avg_step=3.68s recent_step=3.57s eta=42:45:26 +2026-09-14 22:49:17 | INFO | train rank=0 | step=8140/50000 progress=16.3% loss=0.135033 lr=9.656150e-05 lr_backbone=9.656150e-05 elapsed=08:18:55 avg_step=3.68s recent_step=4.74s eta=42:45:43 +2026-09-14 22:49:48 | INFO | train rank=0 | step=8150/50000 progress=16.3% loss=0.112345 lr=9.654944e-05 lr_backbone=9.654944e-05 elapsed=08:19:26 avg_step=3.68s recent_step=3.08s eta=42:44:36 +2026-09-14 22:50:20 | INFO | train rank=0 | step=8160/50000 progress=16.3% loss=0.121432 lr=9.653736e-05 lr_backbone=9.653736e-05 elapsed=08:19:58 avg_step=3.68s recent_step=3.23s eta=42:43:36 +2026-09-14 22:50:55 | INFO | train rank=0 | step=8170/50000 progress=16.3% loss=0.090082 lr=9.652525e-05 lr_backbone=9.652525e-05 elapsed=08:20:33 avg_step=3.68s recent_step=3.51s eta=42:42:51 +2026-09-14 22:51:48 | INFO | train rank=0 | step=8180/50000 progress=16.4% loss=0.127675 lr=9.651313e-05 lr_backbone=9.651313e-05 elapsed=08:21:26 avg_step=3.68s recent_step=5.30s eta=42:43:37 +2026-09-14 22:52:17 | INFO | train rank=0 | step=8190/50000 progress=16.4% loss=0.127379 lr=9.650099e-05 lr_backbone=9.650099e-05 elapsed=08:21:55 avg_step=3.68s recent_step=2.84s eta=42:42:18 +2026-09-14 22:52:53 | INFO | train rank=0 | step=8200/50000 progress=16.4% loss=0.106969 lr=9.648882e-05 lr_backbone=9.648882e-05 elapsed=08:22:31 avg_step=3.68s recent_step=3.62s eta=42:41:38 +2026-09-14 22:53:21 | INFO | train rank=0 | step=8210/50000 progress=16.4% loss=0.137957 lr=9.647664e-05 lr_backbone=9.647664e-05 elapsed=08:22:59 avg_step=3.68s recent_step=2.82s eta=42:40:17 +2026-09-14 22:54:11 | INFO | train rank=0 | step=8220/50000 progress=16.4% loss=0.098496 lr=9.646444e-05 lr_backbone=9.646444e-05 elapsed=08:23:49 avg_step=3.68s recent_step=4.95s eta=42:40:45 +2026-09-14 22:54:48 | INFO | train rank=0 | step=8230/50000 progress=16.5% loss=0.096954 lr=9.645221e-05 lr_backbone=9.645221e-05 elapsed=08:24:26 avg_step=3.68s recent_step=3.77s eta=42:40:13 +2026-09-14 22:55:26 | INFO | train rank=0 | step=8240/50000 progress=16.5% loss=0.121289 lr=9.643997e-05 lr_backbone=9.643997e-05 elapsed=08:25:04 avg_step=3.68s recent_step=3.75s eta=42:39:40 +2026-09-14 22:55:57 | INFO | train rank=0 | step=8250/50000 progress=16.5% loss=0.110534 lr=9.642770e-05 lr_backbone=9.642770e-05 elapsed=08:25:35 avg_step=3.68s recent_step=3.10s eta=42:38:34 +2026-09-14 22:56:52 | INFO | train rank=0 | step=8260/50000 progress=16.5% loss=0.170394 lr=9.641542e-05 lr_backbone=9.641542e-05 elapsed=08:26:30 avg_step=3.68s recent_step=5.49s eta=42:39:29 +2026-09-14 22:57:18 | INFO | train rank=0 | step=8270/50000 progress=16.5% loss=0.114909 lr=9.640311e-05 lr_backbone=9.640311e-05 elapsed=08:26:56 avg_step=3.68s recent_step=2.60s eta=42:37:58 +2026-09-14 22:57:43 | INFO | train rank=0 | step=8280/50000 progress=16.6% loss=0.110587 lr=9.639078e-05 lr_backbone=9.639078e-05 elapsed=08:27:21 avg_step=3.68s recent_step=2.58s eta=42:36:26 +2026-09-14 22:58:19 | INFO | train rank=0 | step=8290/50000 progress=16.6% loss=0.099922 lr=9.637844e-05 lr_backbone=9.637844e-05 elapsed=08:27:57 avg_step=3.68s recent_step=3.50s eta=42:35:40 +2026-09-14 22:59:08 | INFO | train rank=0 | step=8300/50000 progress=16.6% loss=0.115445 lr=9.636607e-05 lr_backbone=9.636607e-05 elapsed=08:28:46 avg_step=3.68s recent_step=4.91s eta=42:36:06 +2026-09-14 22:59:34 | INFO | train rank=0 | step=8310/50000 progress=16.6% loss=0.121576 lr=9.635368e-05 lr_backbone=9.635368e-05 elapsed=08:29:12 avg_step=3.68s recent_step=2.65s eta=42:34:38 +2026-09-14 23:00:12 | INFO | train rank=0 | step=8320/50000 progress=16.6% loss=0.097222 lr=9.634128e-05 lr_backbone=9.634128e-05 elapsed=08:29:50 avg_step=3.68s recent_step=3.78s eta=42:34:06 +2026-09-14 23:00:52 | INFO | train rank=0 | step=8330/50000 progress=16.7% loss=0.130979 lr=9.632885e-05 lr_backbone=9.632885e-05 elapsed=08:30:30 avg_step=3.68s recent_step=3.98s eta=42:33:44 +2026-09-14 23:01:46 | INFO | train rank=0 | step=8340/50000 progress=16.7% loss=0.113210 lr=9.631640e-05 lr_backbone=9.631640e-05 elapsed=08:31:24 avg_step=3.68s recent_step=5.46s eta=42:34:36 +2026-09-14 23:02:07 | INFO | train rank=0 | step=8350/50000 progress=16.7% loss=0.091494 lr=9.630393e-05 lr_backbone=9.630393e-05 elapsed=08:31:45 avg_step=3.68s recent_step=2.02s eta=42:32:37 +2026-09-14 23:02:39 | INFO | train rank=0 | step=8360/50000 progress=16.7% loss=0.105098 lr=9.629145e-05 lr_backbone=9.629145e-05 elapsed=08:32:17 avg_step=3.68s recent_step=3.28s eta=42:31:40 +2026-09-14 23:03:09 | INFO | train rank=0 | step=8370/50000 progress=16.7% loss=0.106784 lr=9.627894e-05 lr_backbone=9.627894e-05 elapsed=08:32:47 avg_step=3.68s recent_step=2.99s eta=42:30:29 +2026-09-14 23:03:54 | INFO | train rank=0 | step=8380/50000 progress=16.8% loss=0.115492 lr=9.626641e-05 lr_backbone=9.626641e-05 elapsed=08:33:32 avg_step=3.68s recent_step=4.50s eta=42:30:33 +2026-09-14 23:04:33 | INFO | train rank=0 | step=8390/50000 progress=16.8% loss=0.102572 lr=9.625386e-05 lr_backbone=9.625386e-05 elapsed=08:34:11 avg_step=3.68s recent_step=3.86s eta=42:30:05 +2026-09-14 23:05:10 | INFO | train rank=0 | step=8400/50000 progress=16.8% loss=0.088713 lr=9.624129e-05 lr_backbone=9.624129e-05 elapsed=08:34:48 avg_step=3.68s recent_step=3.72s eta=42:29:31 +2026-09-14 23:05:36 | INFO | train rank=0 | step=8410/50000 progress=16.8% loss=0.151637 lr=9.622870e-05 lr_backbone=9.622870e-05 elapsed=08:35:14 avg_step=3.68s recent_step=2.62s eta=42:28:02 +2026-09-14 23:06:27 | INFO | train rank=0 | step=8420/50000 progress=16.8% loss=0.091669 lr=9.621609e-05 lr_backbone=9.621609e-05 elapsed=08:36:05 avg_step=3.68s recent_step=5.12s eta=42:28:36 +2026-09-14 23:07:03 | INFO | train rank=0 | step=8430/50000 progress=16.9% loss=0.107734 lr=9.620346e-05 lr_backbone=9.620346e-05 elapsed=08:36:41 avg_step=3.68s recent_step=3.52s eta=42:27:52 +2026-09-14 23:07:33 | INFO | train rank=0 | step=8440/50000 progress=16.9% loss=0.112791 lr=9.619081e-05 lr_backbone=9.619081e-05 elapsed=08:37:11 avg_step=3.68s recent_step=3.06s eta=42:26:45 +2026-09-14 23:08:03 | INFO | train rank=0 | step=8450/50000 progress=16.9% loss=0.109730 lr=9.617814e-05 lr_backbone=9.617814e-05 elapsed=08:37:41 avg_step=3.68s recent_step=2.95s eta=42:25:32 +2026-09-14 23:08:57 | INFO | train rank=0 | step=8460/50000 progress=16.9% loss=0.107868 lr=9.616545e-05 lr_backbone=9.616545e-05 elapsed=08:38:35 avg_step=3.68s recent_step=5.41s eta=42:26:20 +2026-09-14 23:09:33 | INFO | train rank=0 | step=8470/50000 progress=16.9% loss=0.106977 lr=9.615274e-05 lr_backbone=9.615274e-05 elapsed=08:39:11 avg_step=3.68s recent_step=3.61s eta=42:25:41 +2026-09-14 23:10:13 | INFO | train rank=0 | step=8480/50000 progress=17.0% loss=0.118141 lr=9.614001e-05 lr_backbone=9.614001e-05 elapsed=08:39:51 avg_step=3.68s recent_step=4.06s eta=42:25:23 +2026-09-14 23:10:48 | INFO | train rank=0 | step=8490/50000 progress=17.0% loss=0.137694 lr=9.612726e-05 lr_backbone=9.612726e-05 elapsed=08:40:26 avg_step=3.68s recent_step=3.47s eta=42:24:35 +2026-09-14 23:11:36 | INFO | train rank=0 | step=8500/50000 progress=17.0% loss=0.117252 lr=9.611449e-05 lr_backbone=9.611449e-05 elapsed=08:41:14 avg_step=3.68s recent_step=4.83s eta=42:24:55 +2026-09-14 23:12:14 | INFO | train rank=0 | step=8510/50000 progress=17.0% loss=0.108790 lr=9.610170e-05 lr_backbone=9.610170e-05 elapsed=08:41:52 avg_step=3.68s recent_step=3.74s eta=42:24:21 +2026-09-14 23:12:43 | INFO | train rank=0 | step=8520/50000 progress=17.0% loss=0.113127 lr=9.608888e-05 lr_backbone=9.608888e-05 elapsed=08:42:21 avg_step=3.68s recent_step=2.88s eta=42:23:05 +2026-09-14 23:13:14 | INFO | train rank=0 | step=8530/50000 progress=17.1% loss=0.114092 lr=9.607605e-05 lr_backbone=9.607605e-05 elapsed=08:42:53 avg_step=3.68s recent_step=3.19s eta=42:22:05 +2026-09-14 23:13:55 | INFO | train rank=0 | step=8540/50000 progress=17.1% loss=0.141009 lr=9.606320e-05 lr_backbone=9.606320e-05 elapsed=08:43:33 avg_step=3.68s recent_step=4.08s eta=42:21:47 +2026-09-14 23:14:26 | INFO | train rank=0 | step=8550/50000 progress=17.1% loss=0.119963 lr=9.605033e-05 lr_backbone=9.605033e-05 elapsed=08:44:04 avg_step=3.68s recent_step=3.03s eta=42:20:39 +2026-09-14 23:15:00 | INFO | train rank=0 | step=8560/50000 progress=17.1% loss=0.094831 lr=9.603744e-05 lr_backbone=9.603744e-05 elapsed=08:44:38 avg_step=3.68s recent_step=3.44s eta=42:19:50 +2026-09-14 23:15:31 | INFO | train rank=0 | step=8570/50000 progress=17.1% loss=0.120648 lr=9.602452e-05 lr_backbone=9.602452e-05 elapsed=08:45:09 avg_step=3.68s recent_step=3.06s eta=42:18:44 +2026-09-14 23:16:19 | INFO | train rank=0 | step=8580/50000 progress=17.2% loss=0.127084 lr=9.601159e-05 lr_backbone=9.601159e-05 elapsed=08:45:57 avg_step=3.68s recent_step=4.82s eta=42:19:02 +2026-09-14 23:16:52 | INFO | train rank=0 | step=8590/50000 progress=17.2% loss=0.098901 lr=9.599864e-05 lr_backbone=9.599864e-05 elapsed=08:46:31 avg_step=3.68s recent_step=3.38s eta=42:18:11 +2026-09-14 23:17:26 | INFO | train rank=0 | step=8600/50000 progress=17.2% loss=0.107897 lr=9.598567e-05 lr_backbone=9.598567e-05 elapsed=08:47:04 avg_step=3.68s recent_step=3.32s eta=42:17:17 +2026-09-14 23:17:54 | INFO | train rank=0 | step=8610/50000 progress=17.2% loss=0.105250 lr=9.597267e-05 lr_backbone=9.597267e-05 elapsed=08:47:32 avg_step=3.68s recent_step=2.83s eta=42:16:00 +2026-09-14 23:18:44 | INFO | train rank=0 | step=8620/50000 progress=17.2% loss=0.115811 lr=9.595966e-05 lr_backbone=9.595966e-05 elapsed=08:48:22 avg_step=3.68s recent_step=4.99s eta=42:16:26 +2026-09-14 23:19:15 | INFO | train rank=0 | step=8630/50000 progress=17.3% loss=0.142060 lr=9.594663e-05 lr_backbone=9.594663e-05 elapsed=08:48:53 avg_step=3.68s recent_step=3.09s eta=42:15:21 +2026-09-14 23:19:49 | INFO | train rank=0 | step=8640/50000 progress=17.3% loss=0.110520 lr=9.593358e-05 lr_backbone=9.593358e-05 elapsed=08:49:27 avg_step=3.68s recent_step=3.36s eta=42:14:29 +2026-09-14 23:20:29 | INFO | train rank=0 | step=8650/50000 progress=17.3% loss=0.077809 lr=9.592050e-05 lr_backbone=9.592050e-05 elapsed=08:50:07 avg_step=3.68s recent_step=4.08s eta=42:14:12 +2026-09-14 23:21:15 | INFO | train rank=0 | step=8660/50000 progress=17.3% loss=0.129279 lr=9.590741e-05 lr_backbone=9.590741e-05 elapsed=08:50:53 avg_step=3.68s recent_step=4.55s eta=42:14:17 +2026-09-14 23:21:43 | INFO | train rank=0 | step=8670/50000 progress=17.3% loss=0.120508 lr=9.589430e-05 lr_backbone=9.589430e-05 elapsed=08:51:21 avg_step=3.68s recent_step=2.81s eta=42:12:58 +2026-09-14 23:22:23 | INFO | train rank=0 | step=8680/50000 progress=17.4% loss=0.151977 lr=9.588116e-05 lr_backbone=9.588116e-05 elapsed=08:52:01 avg_step=3.68s recent_step=4.05s eta=42:12:39 +2026-09-14 23:22:51 | INFO | train rank=0 | step=8690/50000 progress=17.4% loss=0.092885 lr=9.586801e-05 lr_backbone=9.586801e-05 elapsed=08:52:29 avg_step=3.68s recent_step=2.79s eta=42:11:20 +2026-09-14 23:23:49 | INFO | train rank=0 | step=8700/50000 progress=17.4% loss=0.107617 lr=9.585483e-05 lr_backbone=9.585483e-05 elapsed=08:53:27 avg_step=3.68s recent_step=5.77s eta=42:12:23 +2026-09-14 23:24:32 | INFO | train rank=0 | step=8710/50000 progress=17.4% loss=0.113255 lr=9.584164e-05 lr_backbone=9.584164e-05 elapsed=08:54:10 avg_step=3.68s recent_step=4.28s eta=42:12:15 +2026-09-14 23:24:54 | INFO | train rank=0 | step=8720/50000 progress=17.4% loss=0.110048 lr=9.582843e-05 lr_backbone=9.582843e-05 elapsed=08:54:32 avg_step=3.68s recent_step=2.22s eta=42:10:29 +2026-09-14 23:25:34 | INFO | train rank=0 | step=8730/50000 progress=17.5% loss=0.114389 lr=9.581519e-05 lr_backbone=9.581519e-05 elapsed=08:55:12 avg_step=3.68s recent_step=3.96s eta=42:10:05 +2026-09-14 23:26:19 | INFO | train rank=0 | step=8740/50000 progress=17.5% loss=0.107532 lr=9.580194e-05 lr_backbone=9.580194e-05 elapsed=08:55:57 avg_step=3.68s recent_step=4.57s eta=42:10:11 +2026-09-14 23:27:00 | INFO | train rank=0 | step=8750/50000 progress=17.5% loss=0.111830 lr=9.578867e-05 lr_backbone=9.578867e-05 elapsed=08:56:38 avg_step=3.68s recent_step=4.03s eta=42:09:51 +2026-09-14 23:27:29 | INFO | train rank=0 | step=8760/50000 progress=17.5% loss=0.146760 lr=9.577537e-05 lr_backbone=9.577537e-05 elapsed=08:57:07 avg_step=3.68s recent_step=2.90s eta=42:08:37 +2026-09-14 23:28:00 | INFO | train rank=0 | step=8770/50000 progress=17.5% loss=0.093239 lr=9.576206e-05 lr_backbone=9.576206e-05 elapsed=08:57:38 avg_step=3.68s recent_step=3.10s eta=42:07:33 +2026-09-14 23:28:56 | INFO | train rank=0 | step=8780/50000 progress=17.6% loss=0.109966 lr=9.574872e-05 lr_backbone=9.574872e-05 elapsed=08:58:34 avg_step=3.68s recent_step=5.62s eta=42:08:27 +2026-09-14 23:29:38 | INFO | train rank=0 | step=8790/50000 progress=17.6% loss=0.107862 lr=9.573537e-05 lr_backbone=9.573537e-05 elapsed=08:59:16 avg_step=3.68s recent_step=4.26s eta=42:08:17 +2026-09-14 23:30:05 | INFO | train rank=0 | step=8800/50000 progress=17.6% loss=0.103712 lr=9.572200e-05 lr_backbone=9.572200e-05 elapsed=08:59:43 avg_step=3.68s recent_step=2.69s eta=42:06:54 +2026-09-14 23:30:41 | INFO | train rank=0 | step=8810/50000 progress=17.6% loss=0.119973 lr=9.570860e-05 lr_backbone=9.570860e-05 elapsed=09:00:19 avg_step=3.68s recent_step=3.59s eta=42:06:13 +2026-09-14 23:31:22 | INFO | train rank=0 | step=8820/50000 progress=17.6% loss=0.092178 lr=9.569519e-05 lr_backbone=9.569519e-05 elapsed=09:01:00 avg_step=3.68s recent_step=4.07s eta=42:05:54 +2026-09-14 23:31:57 | INFO | train rank=0 | step=8830/50000 progress=17.7% loss=0.111214 lr=9.568176e-05 lr_backbone=9.568176e-05 elapsed=09:01:35 avg_step=3.68s recent_step=3.57s eta=42:05:12 +2026-09-14 23:32:34 | INFO | train rank=0 | step=8840/50000 progress=17.7% loss=0.089634 lr=9.566830e-05 lr_backbone=9.566830e-05 elapsed=09:02:12 avg_step=3.68s recent_step=3.71s eta=42:04:37 +2026-09-14 23:33:03 | INFO | train rank=0 | step=8850/50000 progress=17.7% loss=0.081973 lr=9.565483e-05 lr_backbone=9.565483e-05 elapsed=09:02:41 avg_step=3.68s recent_step=2.81s eta=42:03:19 +2026-09-14 23:33:56 | INFO | train rank=0 | step=8860/50000 progress=17.7% loss=0.112999 lr=9.564133e-05 lr_backbone=9.564133e-05 elapsed=09:03:34 avg_step=3.68s recent_step=5.32s eta=42:03:59 +2026-09-14 23:34:25 | INFO | train rank=0 | step=8870/50000 progress=17.7% loss=0.144918 lr=9.562782e-05 lr_backbone=9.562782e-05 elapsed=09:04:03 avg_step=3.68s recent_step=2.97s eta=42:02:49 +2026-09-14 23:35:01 | INFO | train rank=0 | step=8880/50000 progress=17.8% loss=0.100168 lr=9.561429e-05 lr_backbone=9.561429e-05 elapsed=09:04:39 avg_step=3.68s recent_step=3.55s eta=42:02:06 +2026-09-14 23:35:34 | INFO | train rank=0 | step=8890/50000 progress=17.8% loss=0.109201 lr=9.560073e-05 lr_backbone=9.560073e-05 elapsed=09:05:12 avg_step=3.68s recent_step=3.30s eta=42:01:11 +2026-09-14 23:36:30 | INFO | train rank=0 | step=8900/50000 progress=17.8% loss=0.092880 lr=9.558716e-05 lr_backbone=9.558716e-05 elapsed=09:06:08 avg_step=3.68s recent_step=5.65s eta=42:02:05 +2026-09-14 23:37:09 | INFO | train rank=0 | step=8910/50000 progress=17.8% loss=0.089896 lr=9.557357e-05 lr_backbone=9.557357e-05 elapsed=09:06:47 avg_step=3.68s recent_step=3.88s eta=42:01:38 +2026-09-14 23:37:35 | INFO | train rank=0 | step=8920/50000 progress=17.8% loss=0.106636 lr=9.555995e-05 lr_backbone=9.555995e-05 elapsed=09:07:13 avg_step=3.68s recent_step=2.54s eta=42:00:08 +2026-09-14 23:37:53 | INFO | train rank=0 | step=8930/50000 progress=17.9% loss=0.104850 lr=9.554632e-05 lr_backbone=9.554632e-05 elapsed=09:07:31 avg_step=3.68s recent_step=1.85s eta=41:58:08 +2026-09-14 23:38:45 | INFO | train rank=0 | step=8940/50000 progress=17.9% loss=0.097487 lr=9.553267e-05 lr_backbone=9.553267e-05 elapsed=09:08:23 avg_step=3.68s recent_step=5.21s eta=41:58:41 +2026-09-14 23:39:21 | INFO | train rank=0 | step=8950/50000 progress=17.9% loss=0.125935 lr=9.551899e-05 lr_backbone=9.551899e-05 elapsed=09:08:59 avg_step=3.68s recent_step=3.58s eta=41:58:00 +2026-09-14 23:39:56 | INFO | train rank=0 | step=8960/50000 progress=17.9% loss=0.111972 lr=9.550530e-05 lr_backbone=9.550530e-05 elapsed=09:09:34 avg_step=3.68s recent_step=3.46s eta=41:57:13 +2026-09-14 23:40:26 | INFO | train rank=0 | step=8970/50000 progress=17.9% loss=0.139577 lr=9.549159e-05 lr_backbone=9.549159e-05 elapsed=09:10:04 avg_step=3.68s recent_step=3.03s eta=41:56:06 +2026-09-14 23:41:26 | INFO | train rank=0 | step=8980/50000 progress=18.0% loss=0.117384 lr=9.547785e-05 lr_backbone=9.547785e-05 elapsed=09:11:04 avg_step=3.68s recent_step=6.05s eta=41:57:18 +2026-09-14 23:41:56 | INFO | train rank=0 | step=8990/50000 progress=18.0% loss=0.101980 lr=9.546410e-05 lr_backbone=9.546410e-05 elapsed=09:11:34 avg_step=3.68s recent_step=2.96s eta=41:56:08 +2026-09-14 23:42:35 | INFO | train rank=0 | step=9000/50000 progress=18.0% loss=0.122415 lr=9.545033e-05 lr_backbone=9.545033e-05 elapsed=09:12:13 avg_step=3.68s recent_step=3.87s eta=41:55:40 +2026-09-14 23:43:03 | INFO | train rank=0 | step=9010/50000 progress=18.0% loss=0.118000 lr=9.543653e-05 lr_backbone=9.543653e-05 elapsed=09:12:41 avg_step=3.68s recent_step=2.86s eta=41:54:26 +2026-09-14 23:43:43 | INFO | train rank=0 | step=9020/50000 progress=18.0% loss=0.104108 lr=9.542272e-05 lr_backbone=9.542272e-05 elapsed=09:13:21 avg_step=3.68s recent_step=3.98s eta=41:54:03 +2026-09-14 23:44:22 | INFO | train rank=0 | step=9030/50000 progress=18.1% loss=0.102898 lr=9.540889e-05 lr_backbone=9.540889e-05 elapsed=09:14:00 avg_step=3.68s recent_step=3.84s eta=41:53:33 +2026-09-14 23:45:00 | INFO | train rank=0 | step=9040/50000 progress=18.1% loss=0.119142 lr=9.539504e-05 lr_backbone=9.539504e-05 elapsed=09:14:38 avg_step=3.68s recent_step=3.82s eta=41:53:03 +2026-09-14 23:45:33 | INFO | train rank=0 | step=9050/50000 progress=18.1% loss=0.129410 lr=9.538116e-05 lr_backbone=9.538116e-05 elapsed=09:15:11 avg_step=3.68s recent_step=3.28s eta=41:52:08 +2026-09-14 23:46:18 | INFO | train rank=0 | step=9060/50000 progress=18.1% loss=0.091158 lr=9.536727e-05 lr_backbone=9.536727e-05 elapsed=09:15:56 avg_step=3.68s recent_step=4.50s eta=41:52:08 +2026-09-14 23:46:55 | INFO | train rank=0 | step=9070/50000 progress=18.1% loss=0.139599 lr=9.535336e-05 lr_backbone=9.535336e-05 elapsed=09:16:33 avg_step=3.68s recent_step=3.71s eta=41:51:32 +2026-09-14 23:47:32 | INFO | train rank=0 | step=9080/50000 progress=18.2% loss=0.107568 lr=9.533943e-05 lr_backbone=9.533943e-05 elapsed=09:17:10 avg_step=3.68s recent_step=3.75s eta=41:50:59 +2026-09-14 23:48:01 | INFO | train rank=0 | step=9090/50000 progress=18.2% loss=0.124664 lr=9.532548e-05 lr_backbone=9.532548e-05 elapsed=09:17:39 avg_step=3.68s recent_step=2.84s eta=41:49:44 +2026-09-14 23:48:50 | INFO | train rank=0 | step=9100/50000 progress=18.2% loss=0.104018 lr=9.531151e-05 lr_backbone=9.531151e-05 elapsed=09:18:28 avg_step=3.68s recent_step=4.97s eta=41:50:05 +2026-09-14 23:49:33 | INFO | train rank=0 | step=9110/50000 progress=18.2% loss=0.115990 lr=9.529752e-05 lr_backbone=9.529752e-05 elapsed=09:19:11 avg_step=3.68s recent_step=4.27s eta=41:49:55 +2026-09-14 23:50:02 | INFO | train rank=0 | step=9120/50000 progress=18.2% loss=0.126831 lr=9.528350e-05 lr_backbone=9.528350e-05 elapsed=09:19:40 avg_step=3.68s recent_step=2.90s eta=41:48:43 +2026-09-14 23:50:28 | INFO | train rank=0 | step=9130/50000 progress=18.3% loss=0.105571 lr=9.526947e-05 lr_backbone=9.526947e-05 elapsed=09:20:06 avg_step=3.68s recent_step=2.62s eta=41:47:18 +2026-09-14 23:51:19 | INFO | train rank=0 | step=9140/50000 progress=18.3% loss=0.103925 lr=9.525542e-05 lr_backbone=9.525542e-05 elapsed=09:20:57 avg_step=3.68s recent_step=5.04s eta=41:47:42 +2026-09-14 23:51:52 | INFO | train rank=0 | step=9150/50000 progress=18.3% loss=0.106007 lr=9.524135e-05 lr_backbone=9.524135e-05 elapsed=09:21:30 avg_step=3.68s recent_step=3.32s eta=41:46:50 +2026-09-14 23:52:23 | INFO | train rank=0 | step=9160/50000 progress=18.3% loss=0.129580 lr=9.522726e-05 lr_backbone=9.522726e-05 elapsed=09:22:01 avg_step=3.68s recent_step=3.13s eta=41:45:48 +2026-09-14 23:52:52 | INFO | train rank=0 | step=9170/50000 progress=18.3% loss=0.118704 lr=9.521315e-05 lr_backbone=9.521315e-05 elapsed=09:22:30 avg_step=3.68s recent_step=2.84s eta=41:44:34 +2026-09-14 23:53:47 | INFO | train rank=0 | step=9180/50000 progress=18.4% loss=0.119132 lr=9.519902e-05 lr_backbone=9.519902e-05 elapsed=09:23:25 avg_step=3.68s recent_step=5.52s eta=41:45:19 +2026-09-14 23:54:16 | INFO | train rank=0 | step=9190/50000 progress=18.4% loss=0.104329 lr=9.518487e-05 lr_backbone=9.518487e-05 elapsed=09:23:54 avg_step=3.68s recent_step=2.96s eta=41:44:10 +2026-09-14 23:54:45 | INFO | train rank=0 | step=9200/50000 progress=18.4% loss=0.108461 lr=9.517070e-05 lr_backbone=9.517070e-05 elapsed=09:24:23 avg_step=3.68s recent_step=2.85s eta=41:42:56 +2026-09-14 23:55:21 | INFO | train rank=0 | step=9210/50000 progress=18.4% loss=0.088374 lr=9.515652e-05 lr_backbone=9.515652e-05 elapsed=09:24:59 avg_step=3.68s recent_step=3.60s eta=41:42:16 +2026-09-14 23:56:06 | INFO | train rank=0 | step=9220/50000 progress=18.4% loss=0.092188 lr=9.514231e-05 lr_backbone=9.514231e-05 elapsed=09:25:44 avg_step=3.68s recent_step=4.55s eta=41:42:17 +2026-09-14 23:56:40 | INFO | train rank=0 | step=9230/50000 progress=18.5% loss=0.144577 lr=9.512808e-05 lr_backbone=9.512808e-05 elapsed=09:26:18 avg_step=3.68s recent_step=3.32s eta=41:41:25 +2026-09-14 23:57:07 | INFO | train rank=0 | step=9240/50000 progress=18.5% loss=0.112063 lr=9.511383e-05 lr_backbone=9.511383e-05 elapsed=09:26:45 avg_step=3.68s recent_step=2.77s eta=41:40:08 +2026-09-14 23:57:40 | INFO | train rank=0 | step=9250/50000 progress=18.5% loss=0.094774 lr=9.509956e-05 lr_backbone=9.509956e-05 elapsed=09:27:18 avg_step=3.68s recent_step=3.24s eta=41:39:11 +2026-09-14 23:58:27 | INFO | train rank=0 | step=9260/50000 progress=18.5% loss=0.105209 lr=9.508527e-05 lr_backbone=9.508527e-05 elapsed=09:28:05 avg_step=3.68s recent_step=4.70s eta=41:39:20 +2026-09-14 23:59:08 | INFO | train rank=0 | step=9270/50000 progress=18.5% loss=0.084937 lr=9.507097e-05 lr_backbone=9.507097e-05 elapsed=09:28:46 avg_step=3.68s recent_step=4.15s eta=41:39:03 +2026-09-14 23:59:40 | INFO | train rank=0 | step=9280/50000 progress=18.6% loss=0.098085 lr=9.505664e-05 lr_backbone=9.505664e-05 elapsed=09:29:18 avg_step=3.68s recent_step=3.17s eta=41:38:04 +2026-09-15 00:00:13 | INFO | train rank=0 | step=9290/50000 progress=18.6% loss=0.111673 lr=9.504229e-05 lr_backbone=9.504229e-05 elapsed=09:29:51 avg_step=3.68s recent_step=3.31s eta=41:37:11 +2026-09-15 00:01:05 | INFO | train rank=0 | step=9300/50000 progress=18.6% loss=0.110421 lr=9.502793e-05 lr_backbone=9.502793e-05 elapsed=09:30:43 avg_step=3.68s recent_step=5.21s eta=41:37:41 +2026-09-15 00:01:30 | INFO | train rank=0 | step=9310/50000 progress=18.6% loss=0.087950 lr=9.501354e-05 lr_backbone=9.501354e-05 elapsed=09:31:08 avg_step=3.68s recent_step=2.54s eta=41:36:14 +2026-09-15 00:02:05 | INFO | train rank=0 | step=9320/50000 progress=18.6% loss=0.105496 lr=9.499913e-05 lr_backbone=9.499913e-05 elapsed=09:31:43 avg_step=3.68s recent_step=3.42s eta=41:35:26 +2026-09-15 00:02:38 | INFO | train rank=0 | step=9330/50000 progress=18.7% loss=0.113957 lr=9.498471e-05 lr_backbone=9.498471e-05 elapsed=09:32:16 avg_step=3.68s recent_step=3.30s eta=41:34:32 +2026-09-15 00:03:25 | INFO | train rank=0 | step=9340/50000 progress=18.7% loss=0.116943 lr=9.497026e-05 lr_backbone=9.497026e-05 elapsed=09:33:03 avg_step=3.68s recent_step=4.77s eta=41:34:43 +2026-09-15 00:04:03 | INFO | train rank=0 | step=9350/50000 progress=18.7% loss=0.121180 lr=9.495580e-05 lr_backbone=9.495580e-05 elapsed=09:33:41 avg_step=3.68s recent_step=3.74s eta=41:34:09 +2026-09-15 00:04:33 | INFO | train rank=0 | step=9360/50000 progress=18.7% loss=0.120235 lr=9.494131e-05 lr_backbone=9.494131e-05 elapsed=09:34:11 avg_step=3.68s recent_step=3.02s eta=41:33:03 +2026-09-15 00:05:00 | INFO | train rank=0 | step=9370/50000 progress=18.7% loss=0.132671 lr=9.492681e-05 lr_backbone=9.492681e-05 elapsed=09:34:38 avg_step=3.68s recent_step=2.67s eta=41:31:43 +2026-09-15 00:05:41 | INFO | train rank=0 | step=9380/50000 progress=18.8% loss=0.100806 lr=9.491228e-05 lr_backbone=9.491228e-05 elapsed=09:35:19 avg_step=3.68s recent_step=4.14s eta=41:31:26 +2026-09-15 00:06:15 | INFO | train rank=0 | step=9390/50000 progress=18.8% loss=0.117008 lr=9.489774e-05 lr_backbone=9.489774e-05 elapsed=09:35:53 avg_step=3.68s recent_step=3.44s eta=41:30:39 +2026-09-15 00:06:49 | INFO | train rank=0 | step=9400/50000 progress=18.8% loss=0.144911 lr=9.488318e-05 lr_backbone=9.488318e-05 elapsed=09:36:27 avg_step=3.68s recent_step=3.32s eta=41:29:46 +2026-09-15 00:07:16 | INFO | train rank=0 | step=9410/50000 progress=18.8% loss=0.123952 lr=9.486859e-05 lr_backbone=9.486859e-05 elapsed=09:36:54 avg_step=3.68s recent_step=2.74s eta=41:28:29 +2026-09-15 00:07:56 | INFO | train rank=0 | step=9420/50000 progress=18.8% loss=0.115195 lr=9.485399e-05 lr_backbone=9.485399e-05 elapsed=09:37:34 avg_step=3.68s recent_step=3.98s eta=41:28:05 +2026-09-15 00:08:31 | INFO | train rank=0 | step=9430/50000 progress=18.9% loss=0.104318 lr=9.483937e-05 lr_backbone=9.483937e-05 elapsed=09:38:09 avg_step=3.68s recent_step=3.55s eta=41:27:23 +2026-09-15 00:08:58 | INFO | train rank=0 | step=9440/50000 progress=18.9% loss=0.109772 lr=9.482473e-05 lr_backbone=9.482473e-05 elapsed=09:38:36 avg_step=3.68s recent_step=2.67s eta=41:26:03 +2026-09-15 00:09:31 | INFO | train rank=0 | step=9450/50000 progress=18.9% loss=0.091881 lr=9.481007e-05 lr_backbone=9.481007e-05 elapsed=09:39:09 avg_step=3.68s recent_step=3.30s eta=41:25:10 +2026-09-15 00:10:13 | INFO | train rank=0 | step=9460/50000 progress=18.9% loss=0.107079 lr=9.479539e-05 lr_backbone=9.479539e-05 elapsed=09:39:51 avg_step=3.68s recent_step=4.20s eta=41:24:55 +2026-09-15 00:10:48 | INFO | train rank=0 | step=9470/50000 progress=18.9% loss=0.125042 lr=9.478069e-05 lr_backbone=9.478069e-05 elapsed=09:40:26 avg_step=3.68s recent_step=3.48s eta=41:24:10 +2026-09-15 00:11:26 | INFO | train rank=0 | step=9480/50000 progress=19.0% loss=0.104263 lr=9.476597e-05 lr_backbone=9.476597e-05 elapsed=09:41:04 avg_step=3.68s recent_step=3.81s eta=41:23:39 +2026-09-15 00:12:00 | INFO | train rank=0 | step=9490/50000 progress=19.0% loss=0.119045 lr=9.475123e-05 lr_backbone=9.475123e-05 elapsed=09:41:38 avg_step=3.68s recent_step=3.44s eta=41:22:52 +2026-09-15 00:12:41 | INFO | train rank=0 | step=9500/50000 progress=19.0% loss=0.155484 lr=9.473647e-05 lr_backbone=9.473647e-05 elapsed=09:42:19 avg_step=3.68s recent_step=4.05s eta=41:22:31 +2026-09-15 00:13:12 | INFO | train rank=0 | step=9510/50000 progress=19.0% loss=0.129190 lr=9.472169e-05 lr_backbone=9.472169e-05 elapsed=09:42:50 avg_step=3.68s recent_step=3.08s eta=41:21:29 +2026-09-15 00:13:45 | INFO | train rank=0 | step=9520/50000 progress=19.0% loss=0.089433 lr=9.470689e-05 lr_backbone=9.470689e-05 elapsed=09:43:23 avg_step=3.68s recent_step=3.31s eta=41:20:37 +2026-09-15 00:14:25 | INFO | train rank=0 | step=9530/50000 progress=19.1% loss=0.119025 lr=9.469207e-05 lr_backbone=9.469207e-05 elapsed=09:44:03 avg_step=3.68s recent_step=4.07s eta=41:20:16 +2026-09-15 00:15:12 | INFO | train rank=0 | step=9540/50000 progress=19.1% loss=0.112128 lr=9.467723e-05 lr_backbone=9.467723e-05 elapsed=09:44:50 avg_step=3.68s recent_step=4.62s eta=41:20:20 +2026-09-15 00:15:46 | INFO | train rank=0 | step=9550/50000 progress=19.1% loss=0.104427 lr=9.466238e-05 lr_backbone=9.466238e-05 elapsed=09:45:24 avg_step=3.68s recent_step=3.39s eta=41:19:31 +2026-09-15 00:16:23 | INFO | train rank=0 | step=9560/50000 progress=19.1% loss=0.144188 lr=9.464750e-05 lr_backbone=9.464750e-05 elapsed=09:46:01 avg_step=3.68s recent_step=3.71s eta=41:18:55 +2026-09-15 00:17:03 | INFO | train rank=0 | step=9570/50000 progress=19.1% loss=0.103339 lr=9.463260e-05 lr_backbone=9.463260e-05 elapsed=09:46:41 avg_step=3.68s recent_step=4.02s eta=41:18:33 +2026-09-15 00:17:41 | INFO | train rank=0 | step=9580/50000 progress=19.2% loss=0.122859 lr=9.461769e-05 lr_backbone=9.461769e-05 elapsed=09:47:19 avg_step=3.68s recent_step=3.78s eta=41:18:00 +2026-09-15 00:18:12 | INFO | train rank=0 | step=9590/50000 progress=19.2% loss=0.101037 lr=9.460275e-05 lr_backbone=9.460275e-05 elapsed=09:47:50 avg_step=3.68s recent_step=3.12s eta=41:17:00 +2026-09-15 00:18:53 | INFO | train rank=0 | step=9600/50000 progress=19.2% loss=0.120293 lr=9.458780e-05 lr_backbone=9.458780e-05 elapsed=09:48:31 avg_step=3.68s recent_step=4.09s eta=41:16:41 +2026-09-15 00:19:32 | INFO | train rank=0 | step=9610/50000 progress=19.2% loss=0.100019 lr=9.457282e-05 lr_backbone=9.457282e-05 elapsed=09:49:10 avg_step=3.68s recent_step=3.91s eta=41:16:14 +2026-09-15 00:20:24 | INFO | train rank=0 | step=9620/50000 progress=19.2% loss=0.134716 lr=9.455783e-05 lr_backbone=9.455783e-05 elapsed=09:50:02 avg_step=3.68s recent_step=5.25s eta=41:16:43 +2026-09-15 00:20:58 | INFO | train rank=0 | step=9630/50000 progress=19.3% loss=0.084468 lr=9.454282e-05 lr_backbone=9.454282e-05 elapsed=09:50:36 avg_step=3.68s recent_step=3.34s eta=41:15:52 +2026-09-15 00:21:30 | INFO | train rank=0 | step=9640/50000 progress=19.3% loss=0.110408 lr=9.452778e-05 lr_backbone=9.452778e-05 elapsed=09:51:08 avg_step=3.68s recent_step=3.19s eta=41:14:55 +2026-09-15 00:22:01 | INFO | train rank=0 | step=9650/50000 progress=19.3% loss=0.103213 lr=9.451273e-05 lr_backbone=9.451273e-05 elapsed=09:51:39 avg_step=3.68s recent_step=3.11s eta=41:13:54 +2026-09-15 00:22:47 | INFO | train rank=0 | step=9660/50000 progress=19.3% loss=0.092190 lr=9.449766e-05 lr_backbone=9.449766e-05 elapsed=09:52:25 avg_step=3.68s recent_step=4.64s eta=41:13:58 +2026-09-15 00:23:22 | INFO | train rank=0 | step=9670/50000 progress=19.3% loss=0.118261 lr=9.448257e-05 lr_backbone=9.448257e-05 elapsed=09:53:00 avg_step=3.68s recent_step=3.48s eta=41:13:13 +2026-09-15 00:23:50 | INFO | train rank=0 | step=9680/50000 progress=19.4% loss=0.130334 lr=9.446746e-05 lr_backbone=9.446746e-05 elapsed=09:53:28 avg_step=3.68s recent_step=2.80s eta=41:11:59 +2026-09-15 00:24:26 | INFO | train rank=0 | step=9690/50000 progress=19.4% loss=0.119913 lr=9.445233e-05 lr_backbone=9.445233e-05 elapsed=09:54:04 avg_step=3.68s recent_step=3.61s eta=41:11:20 +2026-09-15 00:25:24 | INFO | train rank=0 | step=9700/50000 progress=19.4% loss=0.093435 lr=9.443718e-05 lr_backbone=9.443718e-05 elapsed=09:55:02 avg_step=3.68s recent_step=5.77s eta=41:12:10 +2026-09-15 00:25:55 | INFO | train rank=0 | step=9710/50000 progress=19.4% loss=0.123899 lr=9.442201e-05 lr_backbone=9.442201e-05 elapsed=09:55:33 avg_step=3.68s recent_step=3.16s eta=41:11:11 +2026-09-15 00:26:21 | INFO | train rank=0 | step=9720/50000 progress=19.4% loss=0.100623 lr=9.440682e-05 lr_backbone=9.440682e-05 elapsed=09:55:59 avg_step=3.68s recent_step=2.58s eta=41:09:49 +2026-09-15 00:26:54 | INFO | train rank=0 | step=9730/50000 progress=19.5% loss=0.139635 lr=9.439161e-05 lr_backbone=9.439161e-05 elapsed=09:56:32 avg_step=3.68s recent_step=3.26s eta=41:08:55 +2026-09-15 00:27:42 | INFO | train rank=0 | step=9740/50000 progress=19.5% loss=0.094139 lr=9.437639e-05 lr_backbone=9.437639e-05 elapsed=09:57:20 avg_step=3.68s recent_step=4.77s eta=41:09:03 +2026-09-15 00:28:24 | INFO | train rank=0 | step=9750/50000 progress=19.5% loss=0.138038 lr=9.436114e-05 lr_backbone=9.436114e-05 elapsed=09:58:02 avg_step=3.68s recent_step=4.20s eta=41:08:48 +2026-09-15 00:28:56 | INFO | train rank=0 | step=9760/50000 progress=19.5% loss=0.110177 lr=9.434587e-05 lr_backbone=9.434587e-05 elapsed=09:58:34 avg_step=3.68s recent_step=3.29s eta=41:07:55 +2026-09-15 00:29:34 | INFO | train rank=0 | step=9770/50000 progress=19.5% loss=0.121828 lr=9.433059e-05 lr_backbone=9.433059e-05 elapsed=09:59:12 avg_step=3.68s recent_step=3.71s eta=41:07:20 +2026-09-15 00:30:11 | INFO | train rank=0 | step=9780/50000 progress=19.6% loss=0.091404 lr=9.431528e-05 lr_backbone=9.431528e-05 elapsed=09:59:49 avg_step=3.68s recent_step=3.74s eta=41:06:45 +2026-09-15 00:30:43 | INFO | train rank=0 | step=9790/50000 progress=19.6% loss=0.093354 lr=9.429996e-05 lr_backbone=9.429996e-05 elapsed=10:00:21 avg_step=3.68s recent_step=3.20s eta=41:05:49 +2026-09-15 00:31:19 | INFO | train rank=0 | step=9800/50000 progress=19.6% loss=0.127642 lr=9.428462e-05 lr_backbone=9.428462e-05 elapsed=10:00:57 avg_step=3.68s recent_step=3.59s eta=41:05:08 +2026-09-15 00:31:53 | INFO | train rank=0 | step=9810/50000 progress=19.6% loss=0.116599 lr=9.426925e-05 lr_backbone=9.426925e-05 elapsed=10:01:31 avg_step=3.68s recent_step=3.41s eta=41:04:20 +2026-09-15 00:32:43 | INFO | train rank=0 | step=9820/50000 progress=19.6% loss=0.132418 lr=9.425387e-05 lr_backbone=9.425387e-05 elapsed=10:02:21 avg_step=3.68s recent_step=5.02s eta=41:04:38 +2026-09-15 00:33:18 | INFO | train rank=0 | step=9830/50000 progress=19.7% loss=0.108175 lr=9.423847e-05 lr_backbone=9.423847e-05 elapsed=10:02:56 avg_step=3.68s recent_step=3.47s eta=41:03:53 +2026-09-15 00:33:50 | INFO | train rank=0 | step=9840/50000 progress=19.7% loss=0.125842 lr=9.422305e-05 lr_backbone=9.422305e-05 elapsed=10:03:28 avg_step=3.68s recent_step=3.18s eta=41:02:56 +2026-09-15 00:34:22 | INFO | train rank=0 | step=9850/50000 progress=19.7% loss=0.117286 lr=9.420761e-05 lr_backbone=9.420761e-05 elapsed=10:04:00 avg_step=3.68s recent_step=3.20s eta=41:01:59 +2026-09-15 00:35:02 | INFO | train rank=0 | step=9860/50000 progress=19.7% loss=0.118088 lr=9.419215e-05 lr_backbone=9.419215e-05 elapsed=10:04:40 avg_step=3.68s recent_step=4.02s eta=41:01:37 +2026-09-15 00:35:37 | INFO | train rank=0 | step=9870/50000 progress=19.7% loss=0.104339 lr=9.417667e-05 lr_backbone=9.417667e-05 elapsed=10:05:15 avg_step=3.68s recent_step=3.49s eta=41:00:52 +2026-09-15 00:36:25 | INFO | train rank=0 | step=9880/50000 progress=19.8% loss=0.098861 lr=9.416117e-05 lr_backbone=9.416117e-05 elapsed=10:06:03 avg_step=3.68s recent_step=4.87s eta=41:01:04 +2026-09-15 00:36:56 | INFO | train rank=0 | step=9890/50000 progress=19.8% loss=0.100059 lr=9.414565e-05 lr_backbone=9.414565e-05 elapsed=10:06:34 avg_step=3.68s recent_step=3.02s eta=41:00:00 +2026-09-15 00:37:34 | INFO | train rank=0 | step=9900/50000 progress=19.8% loss=0.100582 lr=9.413012e-05 lr_backbone=9.413012e-05 elapsed=10:07:12 avg_step=3.68s recent_step=3.83s eta=40:59:29 +2026-09-15 00:38:06 | INFO | train rank=0 | step=9910/50000 progress=19.8% loss=0.093737 lr=9.411456e-05 lr_backbone=9.411456e-05 elapsed=10:07:44 avg_step=3.68s recent_step=3.19s eta=40:58:32 +2026-09-15 00:38:51 | INFO | train rank=0 | step=9920/50000 progress=19.8% loss=0.104477 lr=9.409899e-05 lr_backbone=9.409899e-05 elapsed=10:08:29 avg_step=3.68s recent_step=4.56s eta=40:58:31 +2026-09-15 00:39:28 | INFO | train rank=0 | step=9930/50000 progress=19.9% loss=0.111546 lr=9.408339e-05 lr_backbone=9.408339e-05 elapsed=10:09:06 avg_step=3.68s recent_step=3.67s eta=40:57:54 +2026-09-15 00:40:07 | INFO | train rank=0 | step=9940/50000 progress=19.9% loss=0.105836 lr=9.406778e-05 lr_backbone=9.406778e-05 elapsed=10:09:45 avg_step=3.68s recent_step=3.93s eta=40:57:28 +2026-09-15 00:40:44 | INFO | train rank=0 | step=9950/50000 progress=19.9% loss=0.092090 lr=9.405214e-05 lr_backbone=9.405214e-05 elapsed=10:10:22 avg_step=3.68s recent_step=3.61s eta=40:56:48 +2026-09-15 00:41:22 | INFO | train rank=0 | step=9960/50000 progress=19.9% loss=0.133768 lr=9.403649e-05 lr_backbone=9.403649e-05 elapsed=10:11:00 avg_step=3.68s recent_step=3.87s eta=40:56:19 +2026-09-15 00:41:50 | INFO | train rank=0 | step=9970/50000 progress=19.9% loss=0.086342 lr=9.402082e-05 lr_backbone=9.402082e-05 elapsed=10:11:28 avg_step=3.68s recent_step=2.76s eta=40:55:05 +2026-09-15 00:42:34 | INFO | train rank=0 | step=9980/50000 progress=20.0% loss=0.155516 lr=9.400513e-05 lr_backbone=9.400513e-05 elapsed=10:12:12 avg_step=3.68s recent_step=4.41s eta=40:54:58 +2026-09-15 00:43:05 | INFO | train rank=0 | step=9990/50000 progress=20.0% loss=0.117285 lr=9.398942e-05 lr_backbone=9.398942e-05 elapsed=10:12:43 avg_step=3.68s recent_step=3.10s eta=40:53:58 +2026-09-15 00:43:39 | INFO | train rank=0 | step=10000/50000 progress=20.0% loss=0.119463 lr=9.397369e-05 lr_backbone=9.397369e-05 elapsed=10:13:17 avg_step=3.68s recent_step=3.40s eta=40:53:10 +2026-09-15 00:44:15 | INFO | train rank=0 | step=10010/50000 progress=20.0% loss=0.132188 lr=9.395794e-05 lr_backbone=9.395794e-05 elapsed=10:13:53 avg_step=3.68s recent_step=3.57s eta=40:52:29 +2026-09-15 00:45:07 | INFO | train rank=0 | step=10020/50000 progress=20.0% loss=0.123579 lr=9.394217e-05 lr_backbone=9.394217e-05 elapsed=10:14:45 avg_step=3.68s recent_step=5.23s eta=40:52:54 +2026-09-15 00:45:46 | INFO | train rank=0 | step=10030/50000 progress=20.1% loss=0.090516 lr=9.392638e-05 lr_backbone=9.392638e-05 elapsed=10:15:24 avg_step=3.68s recent_step=3.90s eta=40:52:26 +2026-09-15 00:46:22 | INFO | train rank=0 | step=10040/50000 progress=20.1% loss=0.110676 lr=9.391058e-05 lr_backbone=9.391058e-05 elapsed=10:16:00 avg_step=3.68s recent_step=3.63s eta=40:51:47 +2026-09-15 00:47:00 | INFO | train rank=0 | step=10050/50000 progress=20.1% loss=0.113571 lr=9.389475e-05 lr_backbone=9.389475e-05 elapsed=10:16:38 avg_step=3.68s recent_step=3.74s eta=40:51:12 +2026-09-15 00:47:55 | INFO | train rank=0 | step=10060/50000 progress=20.1% loss=0.131129 lr=9.387891e-05 lr_backbone=9.387891e-05 elapsed=10:17:33 avg_step=3.68s recent_step=5.48s eta=40:51:47 +2026-09-15 00:48:14 | INFO | train rank=0 | step=10070/50000 progress=20.1% loss=0.099824 lr=9.386304e-05 lr_backbone=9.386304e-05 elapsed=10:17:52 avg_step=3.68s recent_step=1.90s eta=40:49:59 +2026-09-15 00:48:45 | INFO | train rank=0 | step=10080/50000 progress=20.2% loss=0.126462 lr=9.384716e-05 lr_backbone=9.384716e-05 elapsed=10:18:23 avg_step=3.68s recent_step=3.09s eta=40:48:59 +2026-09-15 00:49:16 | INFO | train rank=0 | step=10090/50000 progress=20.2% loss=0.120730 lr=9.383126e-05 lr_backbone=9.383126e-05 elapsed=10:18:54 avg_step=3.68s recent_step=3.15s eta=40:48:02 +2026-09-15 00:50:12 | INFO | train rank=0 | step=10100/50000 progress=20.2% loss=0.151001 lr=9.381533e-05 lr_backbone=9.381533e-05 elapsed=10:19:50 avg_step=3.68s recent_step=5.62s eta=40:48:42 +2026-09-15 00:50:49 | INFO | train rank=0 | step=10110/50000 progress=20.2% loss=0.129897 lr=9.379939e-05 lr_backbone=9.379939e-05 elapsed=10:20:27 avg_step=3.68s recent_step=3.69s eta=40:48:05 +2026-09-15 00:51:18 | INFO | train rank=0 | step=10120/50000 progress=20.2% loss=0.101845 lr=9.378343e-05 lr_backbone=9.378343e-05 elapsed=10:20:56 avg_step=3.68s recent_step=2.88s eta=40:46:57 +2026-09-15 00:51:52 | INFO | train rank=0 | step=10130/50000 progress=20.3% loss=0.143809 lr=9.376745e-05 lr_backbone=9.376745e-05 elapsed=10:21:30 avg_step=3.68s recent_step=3.39s eta=40:46:08 +2026-09-15 00:52:40 | INFO | train rank=0 | step=10140/50000 progress=20.3% loss=0.135733 lr=9.375146e-05 lr_backbone=9.375146e-05 elapsed=10:22:18 avg_step=3.68s recent_step=4.78s eta=40:46:15 +2026-09-15 00:53:11 | INFO | train rank=0 | step=10150/50000 progress=20.3% loss=0.087523 lr=9.373544e-05 lr_backbone=9.373544e-05 elapsed=10:22:49 avg_step=3.68s recent_step=3.10s eta=40:45:15 +2026-09-15 00:53:48 | INFO | train rank=0 | step=10160/50000 progress=20.3% loss=0.108952 lr=9.371940e-05 lr_backbone=9.371940e-05 elapsed=10:23:26 avg_step=3.68s recent_step=3.68s eta=40:44:38 +2026-09-15 00:54:22 | INFO | train rank=0 | step=10170/50000 progress=20.3% loss=0.095077 lr=9.370335e-05 lr_backbone=9.370335e-05 elapsed=10:24:00 avg_step=3.68s recent_step=3.43s eta=40:43:52 +2026-09-15 00:55:04 | INFO | train rank=0 | step=10180/50000 progress=20.4% loss=0.115071 lr=9.368727e-05 lr_backbone=9.368727e-05 elapsed=10:24:42 avg_step=3.68s recent_step=4.24s eta=40:43:37 +2026-09-15 00:55:45 | INFO | train rank=0 | step=10190/50000 progress=20.4% loss=0.100740 lr=9.367118e-05 lr_backbone=9.367118e-05 elapsed=10:25:23 avg_step=3.68s recent_step=4.03s eta=40:43:14 +2026-09-15 00:56:20 | INFO | train rank=0 | step=10200/50000 progress=20.4% loss=0.099180 lr=9.365506e-05 lr_backbone=9.365506e-05 elapsed=10:25:58 avg_step=3.68s recent_step=3.56s eta=40:42:32 +2026-09-15 00:56:53 | INFO | train rank=0 | step=10210/50000 progress=20.4% loss=0.118345 lr=9.363893e-05 lr_backbone=9.363893e-05 elapsed=10:26:31 avg_step=3.68s recent_step=3.25s eta=40:41:38 +2026-09-15 00:57:40 | INFO | train rank=0 | step=10220/50000 progress=20.4% loss=0.096167 lr=9.362278e-05 lr_backbone=9.362278e-05 elapsed=10:27:18 avg_step=3.68s recent_step=4.75s eta=40:41:43 +2026-09-15 00:58:15 | INFO | train rank=0 | step=10230/50000 progress=20.5% loss=0.101312 lr=9.360661e-05 lr_backbone=9.360661e-05 elapsed=10:27:53 avg_step=3.68s recent_step=3.49s eta=40:40:59 +2026-09-15 00:58:52 | INFO | train rank=0 | step=10240/50000 progress=20.5% loss=0.113024 lr=9.359042e-05 lr_backbone=9.359042e-05 elapsed=10:28:30 avg_step=3.68s recent_step=3.72s eta=40:40:24 +2026-09-15 00:59:28 | INFO | train rank=0 | step=10250/50000 progress=20.5% loss=0.110381 lr=9.357421e-05 lr_backbone=9.357421e-05 elapsed=10:29:06 avg_step=3.68s recent_step=3.54s eta=40:39:41 +2026-09-15 01:00:05 | INFO | train rank=0 | step=10260/50000 progress=20.5% loss=0.074530 lr=9.355798e-05 lr_backbone=9.355798e-05 elapsed=10:29:43 avg_step=3.68s recent_step=3.69s eta=40:39:05 +2026-09-15 01:00:44 | INFO | train rank=0 | step=10270/50000 progress=20.5% loss=0.103227 lr=9.354174e-05 lr_backbone=9.354174e-05 elapsed=10:30:22 avg_step=3.68s recent_step=3.91s eta=40:38:37 +2026-09-15 01:01:15 | INFO | train rank=0 | step=10280/50000 progress=20.6% loss=0.120613 lr=9.352547e-05 lr_backbone=9.352547e-05 elapsed=10:30:53 avg_step=3.68s recent_step=3.10s eta=40:37:37 +2026-09-15 01:01:54 | INFO | train rank=0 | step=10290/50000 progress=20.6% loss=0.084619 lr=9.350919e-05 lr_backbone=9.350919e-05 elapsed=10:31:32 avg_step=3.68s recent_step=3.97s eta=40:37:12 +2026-09-15 01:02:51 | INFO | train rank=0 | step=10300/50000 progress=20.6% loss=0.082969 lr=9.349288e-05 lr_backbone=9.349288e-05 elapsed=10:32:29 avg_step=3.68s recent_step=5.66s eta=40:37:51 +2026-09-15 01:03:25 | INFO | train rank=0 | step=10310/50000 progress=20.6% loss=0.140441 lr=9.347656e-05 lr_backbone=9.347656e-05 elapsed=10:33:03 avg_step=3.68s recent_step=3.42s eta=40:37:04 +2026-09-15 01:03:56 | INFO | train rank=0 | step=10320/50000 progress=20.6% loss=0.119590 lr=9.346022e-05 lr_backbone=9.346022e-05 elapsed=10:33:34 avg_step=3.68s recent_step=3.11s eta=40:36:05 +2026-09-15 01:04:32 | INFO | train rank=0 | step=10330/50000 progress=20.7% loss=0.101456 lr=9.344386e-05 lr_backbone=9.344386e-05 elapsed=10:34:10 avg_step=3.68s recent_step=3.54s eta=40:35:23 +2026-09-15 01:05:27 | INFO | train rank=0 | step=10340/50000 progress=20.7% loss=0.118202 lr=9.342748e-05 lr_backbone=9.342748e-05 elapsed=10:35:05 avg_step=3.69s recent_step=5.54s eta=40:35:57 +2026-09-15 01:05:59 | INFO | train rank=0 | step=10350/50000 progress=20.7% loss=0.111076 lr=9.341108e-05 lr_backbone=9.341108e-05 elapsed=10:35:37 avg_step=3.68s recent_step=3.17s eta=40:35:00 +2026-09-15 01:06:35 | INFO | train rank=0 | step=10360/50000 progress=20.7% loss=0.116155 lr=9.339466e-05 lr_backbone=9.339466e-05 elapsed=10:36:13 avg_step=3.68s recent_step=3.65s eta=40:34:22 +2026-09-15 01:07:08 | INFO | train rank=0 | step=10370/50000 progress=20.7% loss=0.106431 lr=9.337823e-05 lr_backbone=9.337823e-05 elapsed=10:36:46 avg_step=3.68s recent_step=3.22s eta=40:33:28 +2026-09-15 01:07:49 | INFO | train rank=0 | step=10380/50000 progress=20.8% loss=0.104673 lr=9.336177e-05 lr_backbone=9.336177e-05 elapsed=10:37:27 avg_step=3.68s recent_step=4.16s eta=40:33:09 +2026-09-15 01:08:15 | INFO | train rank=0 | step=10390/50000 progress=20.8% loss=0.093975 lr=9.334530e-05 lr_backbone=9.334530e-05 elapsed=10:37:53 avg_step=3.68s recent_step=2.61s eta=40:31:51 +2026-09-15 01:08:54 | INFO | train rank=0 | step=10400/50000 progress=20.8% loss=0.108064 lr=9.332880e-05 lr_backbone=9.332880e-05 elapsed=10:38:32 avg_step=3.68s recent_step=3.88s eta=40:31:22 +2026-09-15 01:09:25 | INFO | train rank=0 | step=10410/50000 progress=20.8% loss=0.101032 lr=9.331229e-05 lr_backbone=9.331229e-05 elapsed=10:39:03 avg_step=3.68s recent_step=3.08s eta=40:30:22 +2026-09-15 01:09:58 | INFO | train rank=0 | step=10420/50000 progress=20.8% loss=0.100498 lr=9.329576e-05 lr_backbone=9.329576e-05 elapsed=10:39:36 avg_step=3.68s recent_step=3.34s eta=40:29:32 +2026-09-15 01:10:30 | INFO | train rank=0 | step=10430/50000 progress=20.9% loss=0.106255 lr=9.327921e-05 lr_backbone=9.327921e-05 elapsed=10:40:08 avg_step=3.68s recent_step=3.23s eta=40:28:38 +2026-09-15 01:11:00 | INFO | train rank=0 | step=10440/50000 progress=20.9% loss=0.110235 lr=9.326264e-05 lr_backbone=9.326264e-05 elapsed=10:40:38 avg_step=3.68s recent_step=2.95s eta=40:27:33 +2026-09-15 01:11:35 | INFO | train rank=0 | step=10450/50000 progress=20.9% loss=0.085708 lr=9.324605e-05 lr_backbone=9.324605e-05 elapsed=10:41:13 avg_step=3.68s recent_step=3.53s eta=40:26:50 +2026-09-15 01:12:43 | INFO | train rank=0 | step=10460/50000 progress=20.9% loss=0.089675 lr=9.322944e-05 lr_backbone=9.322944e-05 elapsed=10:42:21 avg_step=3.68s recent_step=6.75s eta=40:28:10 +2026-09-15 01:13:17 | INFO | train rank=0 | step=10470/50000 progress=20.9% loss=0.125090 lr=9.321282e-05 lr_backbone=9.321282e-05 elapsed=10:42:55 avg_step=3.68s recent_step=3.41s eta=40:27:22 +2026-09-15 01:13:46 | INFO | train rank=0 | step=10480/50000 progress=21.0% loss=0.111255 lr=9.319617e-05 lr_backbone=9.319617e-05 elapsed=10:43:24 avg_step=3.68s recent_step=2.90s eta=40:26:16 +2026-09-15 01:14:19 | INFO | train rank=0 | step=10490/50000 progress=21.0% loss=0.119234 lr=9.317951e-05 lr_backbone=9.317951e-05 elapsed=10:43:57 avg_step=3.68s recent_step=3.29s eta=40:25:24 +2026-09-15 01:14:59 | INFO | train rank=0 | step=10500/50000 progress=21.0% loss=0.093926 lr=9.316282e-05 lr_backbone=9.316282e-05 elapsed=10:44:37 avg_step=3.68s recent_step=4.03s eta=40:25:00 +2026-09-15 01:15:33 | INFO | train rank=0 | step=10510/50000 progress=21.0% loss=0.095298 lr=9.314612e-05 lr_backbone=9.314612e-05 elapsed=10:45:11 avg_step=3.68s recent_step=3.39s eta=40:24:12 +2026-09-15 01:16:05 | INFO | train rank=0 | step=10520/50000 progress=21.0% loss=0.110317 lr=9.312940e-05 lr_backbone=9.312940e-05 elapsed=10:45:43 avg_step=3.68s recent_step=3.18s eta=40:23:17 +2026-09-15 01:16:31 | INFO | train rank=0 | step=10530/50000 progress=21.1% loss=0.085062 lr=9.311266e-05 lr_backbone=9.311266e-05 elapsed=10:46:09 avg_step=3.68s recent_step=2.65s eta=40:22:01 +2026-09-15 01:17:09 | INFO | train rank=0 | step=10540/50000 progress=21.1% loss=0.112922 lr=9.309590e-05 lr_backbone=9.309590e-05 elapsed=10:46:47 avg_step=3.68s recent_step=3.81s eta=40:21:29 +2026-09-15 01:17:43 | INFO | train rank=0 | step=10550/50000 progress=21.1% loss=0.111866 lr=9.307913e-05 lr_backbone=9.307913e-05 elapsed=10:47:21 avg_step=3.68s recent_step=3.34s eta=40:20:40 +2026-09-15 01:18:12 | INFO | train rank=0 | step=10560/50000 progress=21.1% loss=0.115587 lr=9.306233e-05 lr_backbone=9.306233e-05 elapsed=10:47:50 avg_step=3.68s recent_step=2.90s eta=40:19:33 +2026-09-15 01:18:48 | INFO | train rank=0 | step=10570/50000 progress=21.1% loss=0.101979 lr=9.304552e-05 lr_backbone=9.304552e-05 elapsed=10:48:26 avg_step=3.68s recent_step=3.64s eta=40:18:55 +2026-09-15 01:19:34 | INFO | train rank=0 | step=10580/50000 progress=21.2% loss=0.098308 lr=9.302868e-05 lr_backbone=9.302868e-05 elapsed=10:49:12 avg_step=3.68s recent_step=4.58s eta=40:18:52 +2026-09-15 01:20:03 | INFO | train rank=0 | step=10590/50000 progress=21.2% loss=0.144211 lr=9.301183e-05 lr_backbone=9.301183e-05 elapsed=10:49:41 avg_step=3.68s recent_step=2.97s eta=40:17:48 +2026-09-15 01:20:39 | INFO | train rank=0 | step=10600/50000 progress=21.2% loss=0.095744 lr=9.299496e-05 lr_backbone=9.299496e-05 elapsed=10:50:17 avg_step=3.68s recent_step=3.58s eta=40:17:08 +2026-09-15 01:21:15 | INFO | train rank=0 | step=10610/50000 progress=21.2% loss=0.107163 lr=9.297807e-05 lr_backbone=9.297807e-05 elapsed=10:50:53 avg_step=3.68s recent_step=3.58s eta=40:16:27 +2026-09-15 01:21:54 | INFO | train rank=0 | step=10620/50000 progress=21.2% loss=0.098782 lr=9.296116e-05 lr_backbone=9.296116e-05 elapsed=10:51:32 avg_step=3.68s recent_step=3.91s eta=40:15:59 +2026-09-15 01:22:40 | INFO | train rank=0 | step=10630/50000 progress=21.3% loss=0.102905 lr=9.294423e-05 lr_backbone=9.294423e-05 elapsed=10:52:18 avg_step=3.68s recent_step=4.64s eta=40:15:57 +2026-09-15 01:23:18 | INFO | train rank=0 | step=10640/50000 progress=21.3% loss=0.104183 lr=9.292728e-05 lr_backbone=9.292728e-05 elapsed=10:52:56 avg_step=3.68s recent_step=3.71s eta=40:15:22 +2026-09-15 01:23:53 | INFO | train rank=0 | step=10650/50000 progress=21.3% loss=0.139692 lr=9.291032e-05 lr_backbone=9.291032e-05 elapsed=10:53:31 avg_step=3.68s recent_step=3.54s eta=40:14:39 +2026-09-15 01:24:31 | INFO | train rank=0 | step=10660/50000 progress=21.3% loss=0.093017 lr=9.289334e-05 lr_backbone=9.289334e-05 elapsed=10:54:09 avg_step=3.68s recent_step=3.79s eta=40:14:07 +2026-09-15 01:25:18 | INFO | train rank=0 | step=10670/50000 progress=21.3% loss=0.113691 lr=9.287633e-05 lr_backbone=9.287633e-05 elapsed=10:54:56 avg_step=3.68s recent_step=4.72s eta=40:14:08 +2026-09-15 01:25:58 | INFO | train rank=0 | step=10680/50000 progress=21.4% loss=0.110800 lr=9.285931e-05 lr_backbone=9.285931e-05 elapsed=10:55:36 avg_step=3.68s recent_step=3.96s eta=40:13:41 +2026-09-15 01:26:30 | INFO | train rank=0 | step=10690/50000 progress=21.4% loss=0.093301 lr=9.284227e-05 lr_backbone=9.284227e-05 elapsed=10:56:08 avg_step=3.68s recent_step=3.25s eta=40:12:49 +2026-09-15 01:27:09 | INFO | train rank=0 | step=10700/50000 progress=21.4% loss=0.128282 lr=9.282521e-05 lr_backbone=9.282521e-05 elapsed=10:56:47 avg_step=3.68s recent_step=3.93s eta=40:12:21 +2026-09-15 01:27:56 | INFO | train rank=0 | step=10710/50000 progress=21.4% loss=0.133150 lr=9.280813e-05 lr_backbone=9.280813e-05 elapsed=10:57:34 avg_step=3.68s recent_step=4.65s eta=40:12:19 +2026-09-15 01:28:24 | INFO | train rank=0 | step=10720/50000 progress=21.4% loss=0.088902 lr=9.279104e-05 lr_backbone=9.279104e-05 elapsed=10:58:02 avg_step=3.68s recent_step=2.84s eta=40:11:12 +2026-09-15 01:28:56 | INFO | train rank=0 | step=10730/50000 progress=21.5% loss=0.098573 lr=9.277392e-05 lr_backbone=9.277392e-05 elapsed=10:58:34 avg_step=3.68s recent_step=3.20s eta=40:10:17 +2026-09-15 01:29:50 | INFO | train rank=0 | step=10740/50000 progress=21.5% loss=0.091434 lr=9.275679e-05 lr_backbone=9.275679e-05 elapsed=10:59:28 avg_step=3.68s recent_step=5.41s eta=40:10:44 +2026-09-15 01:30:31 | INFO | train rank=0 | step=10750/50000 progress=21.5% loss=0.093934 lr=9.273964e-05 lr_backbone=9.273964e-05 elapsed=11:00:09 avg_step=3.68s recent_step=4.06s eta=40:10:20 +2026-09-15 01:30:58 | INFO | train rank=0 | step=10760/50000 progress=21.5% loss=0.125090 lr=9.272246e-05 lr_backbone=9.272246e-05 elapsed=11:00:36 avg_step=3.68s recent_step=2.69s eta=40:09:07 +2026-09-15 01:31:28 | INFO | train rank=0 | step=10770/50000 progress=21.5% loss=0.092776 lr=9.270527e-05 lr_backbone=9.270527e-05 elapsed=11:01:06 avg_step=3.68s recent_step=3.01s eta=40:08:06 +2026-09-15 01:32:20 | INFO | train rank=0 | step=10780/50000 progress=21.6% loss=0.104792 lr=9.268807e-05 lr_backbone=9.268807e-05 elapsed=11:01:58 avg_step=3.68s recent_step=5.19s eta=40:08:24 +2026-09-15 01:32:56 | INFO | train rank=0 | step=10790/50000 progress=21.6% loss=0.125756 lr=9.267084e-05 lr_backbone=9.267084e-05 elapsed=11:02:34 avg_step=3.68s recent_step=3.58s eta=40:07:43 +2026-09-15 01:33:33 | INFO | train rank=0 | step=10800/50000 progress=21.6% loss=0.096142 lr=9.265359e-05 lr_backbone=9.265359e-05 elapsed=11:03:11 avg_step=3.68s recent_step=3.75s eta=40:07:09 +2026-09-15 01:34:08 | INFO | train rank=0 | step=10810/50000 progress=21.6% loss=0.107342 lr=9.263633e-05 lr_backbone=9.263633e-05 elapsed=11:03:46 avg_step=3.68s recent_step=3.50s eta=40:06:25 +2026-09-15 01:34:54 | INFO | train rank=0 | step=10820/50000 progress=21.6% loss=0.082686 lr=9.261904e-05 lr_backbone=9.261904e-05 elapsed=11:04:32 avg_step=3.69s recent_step=4.59s eta=40:06:21 +2026-09-15 01:35:34 | INFO | train rank=0 | step=10830/50000 progress=21.7% loss=0.124121 lr=9.260174e-05 lr_backbone=9.260174e-05 elapsed=11:05:12 avg_step=3.69s recent_step=4.03s eta=40:05:57 +2026-09-15 01:36:06 | INFO | train rank=0 | step=10840/50000 progress=21.7% loss=0.104453 lr=9.258442e-05 lr_backbone=9.258442e-05 elapsed=11:05:44 avg_step=3.68s recent_step=3.12s eta=40:05:00 +2026-09-15 01:36:41 | INFO | train rank=0 | step=10850/50000 progress=21.7% loss=0.100128 lr=9.256708e-05 lr_backbone=9.256708e-05 elapsed=11:06:19 avg_step=3.68s recent_step=3.49s eta=40:04:16 +2026-09-15 01:37:24 | INFO | train rank=0 | step=10860/50000 progress=21.7% loss=0.103776 lr=9.254972e-05 lr_backbone=9.254972e-05 elapsed=11:07:02 avg_step=3.69s recent_step=4.34s eta=40:04:02 +2026-09-15 01:38:00 | INFO | train rank=0 | step=10870/50000 progress=21.7% loss=0.124449 lr=9.253235e-05 lr_backbone=9.253235e-05 elapsed=11:07:38 avg_step=3.69s recent_step=3.60s eta=40:03:23 +2026-09-15 01:38:36 | INFO | train rank=0 | step=10880/50000 progress=21.8% loss=0.102479 lr=9.251495e-05 lr_backbone=9.251495e-05 elapsed=11:08:14 avg_step=3.69s recent_step=3.56s eta=40:02:41 +2026-09-15 01:39:01 | INFO | train rank=0 | step=10890/50000 progress=21.8% loss=0.119216 lr=9.249754e-05 lr_backbone=9.249754e-05 elapsed=11:08:39 avg_step=3.68s recent_step=2.52s eta=40:01:22 +2026-09-15 01:39:38 | INFO | train rank=0 | step=10900/50000 progress=21.8% loss=0.112711 lr=9.248011e-05 lr_backbone=9.248011e-05 elapsed=11:09:16 avg_step=3.68s recent_step=3.76s eta=40:00:48 +2026-09-15 01:40:22 | INFO | train rank=0 | step=10910/50000 progress=21.8% loss=0.104929 lr=9.246266e-05 lr_backbone=9.246266e-05 elapsed=11:10:00 avg_step=3.68s recent_step=4.33s eta=40:00:35 +2026-09-15 01:40:54 | INFO | train rank=0 | step=10920/50000 progress=21.8% loss=0.097759 lr=9.244519e-05 lr_backbone=9.244519e-05 elapsed=11:10:32 avg_step=3.68s recent_step=3.21s eta=39:59:41 +2026-09-15 01:41:25 | INFO | train rank=0 | step=10930/50000 progress=21.9% loss=0.085735 lr=9.242770e-05 lr_backbone=9.242770e-05 elapsed=11:11:03 avg_step=3.68s recent_step=3.16s eta=39:58:46 +2026-09-15 01:42:22 | INFO | train rank=0 | step=10940/50000 progress=21.9% loss=0.074789 lr=9.241019e-05 lr_backbone=9.241019e-05 elapsed=11:12:00 avg_step=3.69s recent_step=5.69s eta=39:59:21 +2026-09-15 01:42:54 | INFO | train rank=0 | step=10950/50000 progress=21.9% loss=0.089281 lr=9.239267e-05 lr_backbone=9.239267e-05 elapsed=11:12:32 avg_step=3.69s recent_step=3.18s eta=39:58:26 +2026-09-15 01:43:36 | INFO | train rank=0 | step=10960/50000 progress=21.9% loss=0.116790 lr=9.237512e-05 lr_backbone=9.237512e-05 elapsed=11:13:14 avg_step=3.69s recent_step=4.19s eta=39:58:07 +2026-09-15 01:44:04 | INFO | train rank=0 | step=10970/50000 progress=21.9% loss=0.104512 lr=9.235756e-05 lr_backbone=9.235756e-05 elapsed=11:13:42 avg_step=3.68s recent_step=2.78s eta=39:56:58 +2026-09-15 01:44:54 | INFO | train rank=0 | step=10980/50000 progress=22.0% loss=0.097059 lr=9.233998e-05 lr_backbone=9.233998e-05 elapsed=11:14:32 avg_step=3.69s recent_step=4.97s eta=39:57:06 +2026-09-15 01:45:20 | INFO | train rank=0 | step=10990/50000 progress=22.0% loss=0.096366 lr=9.232238e-05 lr_backbone=9.232238e-05 elapsed=11:14:58 avg_step=3.69s recent_step=2.65s eta=39:55:53 +2026-09-15 01:45:42 | INFO | train rank=0 | step=11000/50000 progress=22.0% loss=0.096316 lr=9.230476e-05 lr_backbone=9.230476e-05 elapsed=11:15:20 avg_step=3.68s recent_step=2.21s eta=39:54:24 +2026-09-15 01:46:16 | INFO | train rank=0 | step=11010/50000 progress=22.0% loss=0.109448 lr=9.228713e-05 lr_backbone=9.228713e-05 elapsed=11:15:54 avg_step=3.68s recent_step=3.39s eta=39:53:37 +2026-09-15 01:47:11 | INFO | train rank=0 | step=11020/50000 progress=22.0% loss=0.114377 lr=9.226947e-05 lr_backbone=9.226947e-05 elapsed=11:16:49 avg_step=3.69s recent_step=5.53s eta=39:54:05 +2026-09-15 01:47:49 | INFO | train rank=0 | step=11030/50000 progress=22.1% loss=0.095606 lr=9.225180e-05 lr_backbone=9.225180e-05 elapsed=11:17:27 avg_step=3.69s recent_step=3.77s eta=39:53:31 +2026-09-15 01:48:23 | INFO | train rank=0 | step=11040/50000 progress=22.1% loss=0.094350 lr=9.223411e-05 lr_backbone=9.223411e-05 elapsed=11:18:01 avg_step=3.68s recent_step=3.36s eta=39:52:43 +2026-09-15 01:48:54 | INFO | train rank=0 | step=11050/50000 progress=22.1% loss=0.090944 lr=9.221640e-05 lr_backbone=9.221640e-05 elapsed=11:18:32 avg_step=3.68s recent_step=3.09s eta=39:51:45 +2026-09-15 01:49:45 | INFO | train rank=0 | step=11060/50000 progress=22.1% loss=0.107829 lr=9.219867e-05 lr_backbone=9.219867e-05 elapsed=11:19:23 avg_step=3.69s recent_step=5.17s eta=39:52:00 +2026-09-15 01:50:23 | INFO | train rank=0 | step=11070/50000 progress=22.1% loss=0.100533 lr=9.218092e-05 lr_backbone=9.218092e-05 elapsed=11:20:01 avg_step=3.69s recent_step=3.78s eta=39:51:27 +2026-09-15 01:50:52 | INFO | train rank=0 | step=11080/50000 progress=22.2% loss=0.115641 lr=9.216315e-05 lr_backbone=9.216315e-05 elapsed=11:20:30 avg_step=3.69s recent_step=2.94s eta=39:50:24 +2026-09-15 01:51:23 | INFO | train rank=0 | step=11090/50000 progress=22.2% loss=0.105775 lr=9.214537e-05 lr_backbone=9.214537e-05 elapsed=11:21:01 avg_step=3.68s recent_step=3.09s eta=39:49:26 +2026-09-15 01:52:20 | INFO | train rank=0 | step=11100/50000 progress=22.2% loss=0.128549 lr=9.212757e-05 lr_backbone=9.212757e-05 elapsed=11:21:58 avg_step=3.69s recent_step=5.62s eta=39:49:57 +2026-09-15 01:52:50 | INFO | train rank=0 | step=11110/50000 progress=22.2% loss=0.104511 lr=9.210975e-05 lr_backbone=9.210975e-05 elapsed=11:22:28 avg_step=3.69s recent_step=3.07s eta=39:48:59 +2026-09-15 01:53:22 | INFO | train rank=0 | step=11120/50000 progress=22.2% loss=0.138303 lr=9.209191e-05 lr_backbone=9.209191e-05 elapsed=11:23:00 avg_step=3.69s recent_step=3.14s eta=39:48:03 +2026-09-15 01:53:49 | INFO | train rank=0 | step=11130/50000 progress=22.3% loss=0.106957 lr=9.207405e-05 lr_backbone=9.207405e-05 elapsed=11:23:27 avg_step=3.68s recent_step=2.76s eta=39:46:54 +2026-09-15 01:54:36 | INFO | train rank=0 | step=11140/50000 progress=22.3% loss=0.092557 lr=9.205617e-05 lr_backbone=9.205617e-05 elapsed=11:24:14 avg_step=3.69s recent_step=4.67s eta=39:46:51 +2026-09-15 01:55:12 | INFO | train rank=0 | step=11150/50000 progress=22.3% loss=0.098265 lr=9.203828e-05 lr_backbone=9.203828e-05 elapsed=11:24:50 avg_step=3.69s recent_step=3.59s eta=39:46:11 +2026-09-15 01:55:38 | INFO | train rank=0 | step=11160/50000 progress=22.3% loss=0.102757 lr=9.202037e-05 lr_backbone=9.202037e-05 elapsed=11:25:16 avg_step=3.68s recent_step=2.59s eta=39:44:56 +2026-09-15 01:56:08 | INFO | train rank=0 | step=11170/50000 progress=22.3% loss=0.097441 lr=9.200244e-05 lr_backbone=9.200244e-05 elapsed=11:25:46 avg_step=3.68s recent_step=3.00s eta=39:43:55 +2026-09-15 01:57:06 | INFO | train rank=0 | step=11180/50000 progress=22.4% loss=0.094143 lr=9.198449e-05 lr_backbone=9.198449e-05 elapsed=11:26:44 avg_step=3.69s recent_step=5.83s eta=39:44:33 +2026-09-15 01:57:48 | INFO | train rank=0 | step=11190/50000 progress=22.4% loss=0.103488 lr=9.196652e-05 lr_backbone=9.196652e-05 elapsed=11:27:26 avg_step=3.69s recent_step=4.15s eta=39:44:12 +2026-09-15 01:58:19 | INFO | train rank=0 | step=11200/50000 progress=22.4% loss=0.109377 lr=9.194853e-05 lr_backbone=9.194853e-05 elapsed=11:27:57 avg_step=3.69s recent_step=3.12s eta=39:43:16 +2026-09-15 01:58:53 | INFO | train rank=0 | step=11210/50000 progress=22.4% loss=0.105764 lr=9.193053e-05 lr_backbone=9.193053e-05 elapsed=11:28:31 avg_step=3.69s recent_step=3.44s eta=39:42:31 +2026-09-15 01:59:28 | INFO | train rank=0 | step=11220/50000 progress=22.4% loss=0.108869 lr=9.191250e-05 lr_backbone=9.191250e-05 elapsed=11:29:06 avg_step=3.69s recent_step=3.51s eta=39:41:48 +2026-09-15 02:00:15 | INFO | train rank=0 | step=11230/50000 progress=22.5% loss=0.109593 lr=9.189446e-05 lr_backbone=9.189446e-05 elapsed=11:29:53 avg_step=3.69s recent_step=4.71s eta=39:41:46 +2026-09-15 02:00:45 | INFO | train rank=0 | step=11240/50000 progress=22.5% loss=0.117511 lr=9.187640e-05 lr_backbone=9.187640e-05 elapsed=11:30:23 avg_step=3.69s recent_step=3.01s eta=39:40:46 +2026-09-15 02:01:19 | INFO | train rank=0 | step=11250/50000 progress=22.5% loss=0.100694 lr=9.185832e-05 lr_backbone=9.185832e-05 elapsed=11:30:57 avg_step=3.69s recent_step=3.33s eta=39:39:57 +2026-09-15 02:01:56 | INFO | train rank=0 | step=11260/50000 progress=22.5% loss=0.125504 lr=9.184023e-05 lr_backbone=9.184023e-05 elapsed=11:31:34 avg_step=3.69s recent_step=3.69s eta=39:39:20 +2026-09-15 02:02:46 | INFO | train rank=0 | step=11270/50000 progress=22.5% loss=0.105708 lr=9.182211e-05 lr_backbone=9.182211e-05 elapsed=11:32:24 avg_step=3.69s recent_step=4.98s eta=39:39:28 +2026-09-15 02:03:19 | INFO | train rank=0 | step=11280/50000 progress=22.6% loss=0.085333 lr=9.180398e-05 lr_backbone=9.180398e-05 elapsed=11:32:57 avg_step=3.69s recent_step=3.38s eta=39:38:41 +2026-09-15 02:04:00 | INFO | train rank=0 | step=11290/50000 progress=22.6% loss=0.078743 lr=9.178583e-05 lr_backbone=9.178583e-05 elapsed=11:33:38 avg_step=3.69s recent_step=4.11s eta=39:38:18 +2026-09-15 02:04:31 | INFO | train rank=0 | step=11300/50000 progress=22.6% loss=0.098607 lr=9.176766e-05 lr_backbone=9.176766e-05 elapsed=11:34:09 avg_step=3.69s recent_step=3.01s eta=39:37:18 +2026-09-15 02:05:17 | INFO | train rank=0 | step=11310/50000 progress=22.6% loss=0.124598 lr=9.174947e-05 lr_backbone=9.174947e-05 elapsed=11:34:55 avg_step=3.69s recent_step=4.63s eta=39:37:14 +2026-09-15 02:05:52 | INFO | train rank=0 | step=11320/50000 progress=22.6% loss=0.097958 lr=9.173127e-05 lr_backbone=9.173127e-05 elapsed=11:35:30 avg_step=3.69s recent_step=3.53s eta=39:36:32 +2026-09-15 02:06:24 | INFO | train rank=0 | step=11330/50000 progress=22.7% loss=0.095081 lr=9.171304e-05 lr_backbone=9.171304e-05 elapsed=11:36:02 avg_step=3.69s recent_step=3.18s eta=39:35:38 +2026-09-15 02:06:56 | INFO | train rank=0 | step=11340/50000 progress=22.7% loss=0.121564 lr=9.169480e-05 lr_backbone=9.169480e-05 elapsed=11:36:34 avg_step=3.69s recent_step=3.22s eta=39:34:45 +2026-09-15 02:07:40 | INFO | train rank=0 | step=11350/50000 progress=22.7% loss=0.152959 lr=9.167654e-05 lr_backbone=9.167654e-05 elapsed=11:37:18 avg_step=3.69s recent_step=4.42s eta=39:34:33 +2026-09-15 02:08:12 | INFO | train rank=0 | step=11360/50000 progress=22.7% loss=0.139491 lr=9.165826e-05 lr_backbone=9.165826e-05 elapsed=11:37:50 avg_step=3.69s recent_step=3.15s eta=39:33:38 +2026-09-15 02:08:47 | INFO | train rank=0 | step=11370/50000 progress=22.7% loss=0.081714 lr=9.163996e-05 lr_backbone=9.163996e-05 elapsed=11:38:25 avg_step=3.69s recent_step=3.47s eta=39:32:54 +2026-09-15 02:09:22 | INFO | train rank=0 | step=11380/50000 progress=22.8% loss=0.106202 lr=9.162165e-05 lr_backbone=9.162165e-05 elapsed=11:39:00 avg_step=3.69s recent_step=3.55s eta=39:32:12 +2026-09-15 02:10:09 | INFO | train rank=0 | step=11390/50000 progress=22.8% loss=0.128606 lr=9.160331e-05 lr_backbone=9.160331e-05 elapsed=11:39:47 avg_step=3.69s recent_step=4.72s eta=39:32:10 +2026-09-15 02:10:41 | INFO | train rank=0 | step=11400/50000 progress=22.8% loss=0.106496 lr=9.158496e-05 lr_backbone=9.158496e-05 elapsed=11:40:19 avg_step=3.69s recent_step=3.22s eta=39:31:18 +2026-09-15 02:11:12 | INFO | train rank=0 | step=11410/50000 progress=22.8% loss=0.156887 lr=9.156659e-05 lr_backbone=9.156659e-05 elapsed=11:40:50 avg_step=3.69s recent_step=3.06s eta=39:30:20 +2026-09-15 02:11:45 | INFO | train rank=0 | step=11420/50000 progress=22.8% loss=0.106198 lr=9.154820e-05 lr_backbone=9.154820e-05 elapsed=11:41:23 avg_step=3.69s recent_step=3.31s eta=39:29:30 +2026-09-15 02:12:26 | INFO | train rank=0 | step=11430/50000 progress=22.9% loss=0.109217 lr=9.152979e-05 lr_backbone=9.152979e-05 elapsed=11:42:04 avg_step=3.69s recent_step=4.10s eta=39:29:07 +2026-09-15 02:12:59 | INFO | train rank=0 | step=11440/50000 progress=22.9% loss=0.118321 lr=9.151137e-05 lr_backbone=9.151137e-05 elapsed=11:42:37 avg_step=3.69s recent_step=3.31s eta=39:28:18 +2026-09-15 02:13:34 | INFO | train rank=0 | step=11450/50000 progress=22.9% loss=0.090817 lr=9.149293e-05 lr_backbone=9.149293e-05 elapsed=11:43:12 avg_step=3.68s recent_step=3.49s eta=39:27:34 +2026-09-15 02:14:14 | INFO | train rank=0 | step=11460/50000 progress=22.9% loss=0.151118 lr=9.147447e-05 lr_backbone=9.147447e-05 elapsed=11:43:52 avg_step=3.69s recent_step=4.01s eta=39:27:08 +2026-09-15 02:15:09 | INFO | train rank=0 | step=11470/50000 progress=22.9% loss=0.101490 lr=9.145599e-05 lr_backbone=9.145599e-05 elapsed=11:44:47 avg_step=3.69s recent_step=5.45s eta=39:27:31 +2026-09-15 02:15:45 | INFO | train rank=0 | step=11480/50000 progress=23.0% loss=0.103726 lr=9.143749e-05 lr_backbone=9.143749e-05 elapsed=11:45:23 avg_step=3.69s recent_step=3.58s eta=39:26:50 +2026-09-15 02:16:21 | INFO | train rank=0 | step=11490/50000 progress=23.0% loss=0.124994 lr=9.141898e-05 lr_backbone=9.141898e-05 elapsed=11:45:59 avg_step=3.69s recent_step=3.60s eta=39:26:11 +2026-09-15 02:16:52 | INFO | train rank=0 | step=11500/50000 progress=23.0% loss=0.115895 lr=9.140044e-05 lr_backbone=9.140044e-05 elapsed=11:46:30 avg_step=3.69s recent_step=3.17s eta=39:25:17 +2026-09-15 02:17:41 | INFO | train rank=0 | step=11510/50000 progress=23.0% loss=0.115053 lr=9.138189e-05 lr_backbone=9.138189e-05 elapsed=11:47:19 avg_step=3.69s recent_step=4.90s eta=39:25:20 +2026-09-15 02:18:12 | INFO | train rank=0 | step=11520/50000 progress=23.0% loss=0.101003 lr=9.136332e-05 lr_backbone=9.136332e-05 elapsed=11:47:50 avg_step=3.69s recent_step=3.02s eta=39:24:21 +2026-09-15 02:18:48 | INFO | train rank=0 | step=11530/50000 progress=23.1% loss=0.124645 lr=9.134473e-05 lr_backbone=9.134473e-05 elapsed=11:48:27 avg_step=3.69s recent_step=3.69s eta=39:23:45 +2026-09-15 02:19:23 | INFO | train rank=0 | step=11540/50000 progress=23.1% loss=0.110495 lr=9.132613e-05 lr_backbone=9.132613e-05 elapsed=11:49:01 avg_step=3.69s recent_step=3.45s eta=39:23:00 +2026-09-15 02:20:04 | INFO | train rank=0 | step=11550/50000 progress=23.1% loss=0.099197 lr=9.130750e-05 lr_backbone=9.130750e-05 elapsed=11:49:42 avg_step=3.69s recent_step=4.08s eta=39:22:36 +2026-09-15 02:20:41 | INFO | train rank=0 | step=11560/50000 progress=23.1% loss=0.099599 lr=9.128886e-05 lr_backbone=9.128886e-05 elapsed=11:50:19 avg_step=3.69s recent_step=3.68s eta=39:21:59 +2026-09-15 02:21:22 | INFO | train rank=0 | step=11570/50000 progress=23.1% loss=0.105086 lr=9.127020e-05 lr_backbone=9.127020e-05 elapsed=11:51:00 avg_step=3.69s recent_step=4.18s eta=39:21:39 +2026-09-15 02:22:01 | INFO | train rank=0 | step=11580/50000 progress=23.2% loss=0.117478 lr=9.125152e-05 lr_backbone=9.125152e-05 elapsed=11:51:39 avg_step=3.69s recent_step=3.86s eta=39:21:07 +2026-09-15 02:22:43 | INFO | train rank=0 | step=11590/50000 progress=23.2% loss=0.101444 lr=9.123283e-05 lr_backbone=9.123283e-05 elapsed=11:52:21 avg_step=3.69s recent_step=4.16s eta=39:20:46 +2026-09-15 02:23:16 | INFO | train rank=0 | step=11600/50000 progress=23.2% loss=0.085265 lr=9.121411e-05 lr_backbone=9.121411e-05 elapsed=11:52:54 avg_step=3.69s recent_step=3.30s eta=39:19:57 +2026-09-15 02:23:49 | INFO | train rank=0 | step=11610/50000 progress=23.2% loss=0.105359 lr=9.119538e-05 lr_backbone=9.119538e-05 elapsed=11:53:27 avg_step=3.69s recent_step=3.31s eta=39:19:07 +2026-09-15 02:24:21 | INFO | train rank=0 | step=11620/50000 progress=23.2% loss=0.093787 lr=9.117663e-05 lr_backbone=9.117663e-05 elapsed=11:53:59 avg_step=3.69s recent_step=3.26s eta=39:18:16 +2026-09-15 02:25:00 | INFO | train rank=0 | step=11630/50000 progress=23.3% loss=0.109435 lr=9.115786e-05 lr_backbone=9.115786e-05 elapsed=11:54:38 avg_step=3.69s recent_step=3.82s eta=39:17:44 +2026-09-15 02:25:30 | INFO | train rank=0 | step=11640/50000 progress=23.3% loss=0.089754 lr=9.113908e-05 lr_backbone=9.113908e-05 elapsed=11:55:08 avg_step=3.69s recent_step=3.00s eta=39:16:44 +2026-09-15 02:26:08 | INFO | train rank=0 | step=11650/50000 progress=23.3% loss=0.102524 lr=9.112027e-05 lr_backbone=9.112027e-05 elapsed=11:55:46 avg_step=3.69s recent_step=3.84s eta=39:16:13 +2026-09-15 02:26:43 | INFO | train rank=0 | step=11660/50000 progress=23.3% loss=0.086860 lr=9.110145e-05 lr_backbone=9.110145e-05 elapsed=11:56:21 avg_step=3.69s recent_step=3.55s eta=39:15:31 +2026-09-15 02:27:37 | INFO | train rank=0 | step=11670/50000 progress=23.3% loss=0.142301 lr=9.108261e-05 lr_backbone=9.108261e-05 elapsed=11:57:15 avg_step=3.69s recent_step=5.37s eta=39:15:49 +2026-09-15 02:28:10 | INFO | train rank=0 | step=11680/50000 progress=23.4% loss=0.102328 lr=9.106375e-05 lr_backbone=9.106375e-05 elapsed=11:57:48 avg_step=3.69s recent_step=3.27s eta=39:14:59 +2026-09-15 02:28:42 | INFO | train rank=0 | step=11690/50000 progress=23.4% loss=0.121700 lr=9.104487e-05 lr_backbone=9.104487e-05 elapsed=11:58:20 avg_step=3.69s recent_step=3.24s eta=39:14:07 +2026-09-15 02:29:18 | INFO | train rank=0 | step=11700/50000 progress=23.4% loss=0.098814 lr=9.102598e-05 lr_backbone=9.102598e-05 elapsed=11:58:56 avg_step=3.69s recent_step=3.55s eta=39:13:26 +2026-09-15 02:30:09 | INFO | train rank=0 | step=11710/50000 progress=23.4% loss=0.100326 lr=9.100707e-05 lr_backbone=9.100707e-05 elapsed=11:59:47 avg_step=3.69s recent_step=5.11s eta=39:13:36 +2026-09-15 02:30:34 | INFO | train rank=0 | step=11720/50000 progress=23.4% loss=0.117013 lr=9.098814e-05 lr_backbone=9.098814e-05 elapsed=12:00:12 avg_step=3.69s recent_step=2.48s eta=39:12:19 +2026-09-15 02:31:12 | INFO | train rank=0 | step=11730/50000 progress=23.5% loss=0.097833 lr=9.096919e-05 lr_backbone=9.096919e-05 elapsed=12:00:50 avg_step=3.69s recent_step=3.80s eta=39:11:46 +2026-09-15 02:31:49 | INFO | train rank=0 | step=11740/50000 progress=23.5% loss=0.126752 lr=9.095022e-05 lr_backbone=9.095022e-05 elapsed=12:01:27 avg_step=3.69s recent_step=3.69s eta=39:11:10 +2026-09-15 02:32:33 | INFO | train rank=0 | step=11750/50000 progress=23.5% loss=0.112309 lr=9.093124e-05 lr_backbone=9.093124e-05 elapsed=12:02:11 avg_step=3.69s recent_step=4.44s eta=39:10:57 +2026-09-15 02:33:06 | INFO | train rank=0 | step=11760/50000 progress=23.5% loss=0.102144 lr=9.091224e-05 lr_backbone=9.091224e-05 elapsed=12:02:44 avg_step=3.69s recent_step=3.34s eta=39:10:09 +2026-09-15 02:33:42 | INFO | train rank=0 | step=11770/50000 progress=23.5% loss=0.108562 lr=9.089322e-05 lr_backbone=9.089322e-05 elapsed=12:03:20 avg_step=3.69s recent_step=3.56s eta=39:09:28 +2026-09-15 02:34:18 | INFO | train rank=0 | step=11780/50000 progress=23.6% loss=0.105233 lr=9.087418e-05 lr_backbone=9.087418e-05 elapsed=12:03:56 avg_step=3.69s recent_step=3.60s eta=39:08:49 +2026-09-15 02:35:08 | INFO | train rank=0 | step=11790/50000 progress=23.6% loss=0.095033 lr=9.085513e-05 lr_backbone=9.085513e-05 elapsed=12:04:47 avg_step=3.69s recent_step=5.05s eta=39:08:56 +2026-09-15 02:35:36 | INFO | train rank=0 | step=11800/50000 progress=23.6% loss=0.114871 lr=9.083605e-05 lr_backbone=9.083605e-05 elapsed=12:05:14 avg_step=3.69s recent_step=2.72s eta=39:07:48 +2026-09-15 02:36:02 | INFO | train rank=0 | step=11810/50000 progress=23.6% loss=0.100237 lr=9.081696e-05 lr_backbone=9.081696e-05 elapsed=12:05:40 avg_step=3.69s recent_step=2.66s eta=39:06:37 +2026-09-15 02:36:38 | INFO | train rank=0 | step=11820/50000 progress=23.6% loss=0.107834 lr=9.079785e-05 lr_backbone=9.079785e-05 elapsed=12:06:16 avg_step=3.69s recent_step=3.53s eta=39:05:56 +2026-09-15 02:37:22 | INFO | train rank=0 | step=11830/50000 progress=23.7% loss=0.115143 lr=9.077873e-05 lr_backbone=9.077873e-05 elapsed=12:07:00 avg_step=3.69s recent_step=4.46s eta=39:05:44 +2026-09-15 02:38:00 | INFO | train rank=0 | step=11840/50000 progress=23.7% loss=0.128311 lr=9.075958e-05 lr_backbone=9.075958e-05 elapsed=12:07:38 avg_step=3.69s recent_step=3.74s eta=39:05:08 +2026-09-15 02:38:41 | INFO | train rank=0 | step=11850/50000 progress=23.7% loss=0.112869 lr=9.074042e-05 lr_backbone=9.074042e-05 elapsed=12:08:19 avg_step=3.69s recent_step=4.17s eta=39:04:47 +2026-09-15 02:39:19 | INFO | train rank=0 | step=11860/50000 progress=23.7% loss=0.089745 lr=9.072124e-05 lr_backbone=9.072124e-05 elapsed=12:08:57 avg_step=3.69s recent_step=3.76s eta=39:04:13 +2026-09-15 02:40:14 | INFO | train rank=0 | step=11870/50000 progress=23.7% loss=0.099726 lr=9.070204e-05 lr_backbone=9.070204e-05 elapsed=12:09:52 avg_step=3.69s recent_step=5.52s eta=39:04:35 +2026-09-15 02:40:42 | INFO | train rank=0 | step=11880/50000 progress=23.8% loss=0.117063 lr=9.068283e-05 lr_backbone=9.068283e-05 elapsed=12:10:20 avg_step=3.69s recent_step=2.78s eta=39:03:29 +2026-09-15 02:41:11 | INFO | train rank=0 | step=11890/50000 progress=23.8% loss=0.105999 lr=9.066359e-05 lr_backbone=9.066359e-05 elapsed=12:10:49 avg_step=3.69s recent_step=2.92s eta=39:02:27 +2026-09-15 02:41:45 | INFO | train rank=0 | step=11900/50000 progress=23.8% loss=0.123660 lr=9.064434e-05 lr_backbone=9.064434e-05 elapsed=12:11:23 avg_step=3.69s recent_step=3.35s eta=39:01:39 +2026-09-15 02:42:36 | INFO | train rank=0 | step=11910/50000 progress=23.8% loss=0.081650 lr=9.062507e-05 lr_backbone=9.062507e-05 elapsed=12:12:14 avg_step=3.69s recent_step=5.12s eta=39:01:49 +2026-09-15 02:43:07 | INFO | train rank=0 | step=11920/50000 progress=23.8% loss=0.131989 lr=9.060578e-05 lr_backbone=9.060578e-05 elapsed=12:12:45 avg_step=3.69s recent_step=3.15s eta=39:00:54 +2026-09-15 02:43:35 | INFO | train rank=0 | step=11930/50000 progress=23.9% loss=0.108106 lr=9.058648e-05 lr_backbone=9.058648e-05 elapsed=12:13:13 avg_step=3.69s recent_step=2.78s eta=38:59:49 +2026-09-15 02:44:00 | INFO | train rank=0 | step=11940/50000 progress=23.9% loss=0.096831 lr=9.056716e-05 lr_backbone=9.056716e-05 elapsed=12:13:38 avg_step=3.69s recent_step=2.45s eta=38:58:32 +2026-09-15 02:44:45 | INFO | train rank=0 | step=11950/50000 progress=23.9% loss=0.082898 lr=9.054782e-05 lr_backbone=9.054782e-05 elapsed=12:14:23 avg_step=3.69s recent_step=4.58s eta=38:58:24 +2026-09-15 02:45:19 | INFO | train rank=0 | step=11960/50000 progress=23.9% loss=0.108514 lr=9.052846e-05 lr_backbone=9.052846e-05 elapsed=12:14:57 avg_step=3.69s recent_step=3.37s eta=38:57:37 +2026-09-15 02:45:45 | INFO | train rank=0 | step=11970/50000 progress=23.9% loss=0.111671 lr=9.050908e-05 lr_backbone=9.050908e-05 elapsed=12:15:23 avg_step=3.69s recent_step=2.56s eta=38:56:24 +2026-09-15 02:46:21 | INFO | train rank=0 | step=11980/50000 progress=24.0% loss=0.120820 lr=9.048969e-05 lr_backbone=9.048969e-05 elapsed=12:15:59 avg_step=3.69s recent_step=3.62s eta=38:55:45 +2026-09-15 02:47:01 | INFO | train rank=0 | step=11990/50000 progress=24.0% loss=0.090594 lr=9.047028e-05 lr_backbone=9.047028e-05 elapsed=12:16:39 avg_step=3.69s recent_step=3.96s eta=38:55:17 +2026-09-15 02:47:29 | INFO | train rank=0 | step=12000/50000 progress=24.0% loss=0.106372 lr=9.045085e-05 lr_backbone=9.045085e-05 elapsed=12:17:07 avg_step=3.69s recent_step=2.86s eta=38:54:14 +2026-09-15 02:48:04 | INFO | train rank=0 | step=12010/50000 progress=24.0% loss=0.122741 lr=9.043140e-05 lr_backbone=9.043140e-05 elapsed=12:17:43 avg_step=3.69s recent_step=3.54s eta=38:53:32 +2026-09-15 02:48:26 | INFO | train rank=0 | step=12020/50000 progress=24.0% loss=0.094540 lr=9.041194e-05 lr_backbone=9.041194e-05 elapsed=12:18:04 avg_step=3.68s recent_step=2.11s eta=38:52:06 +2026-09-15 02:49:05 | INFO | train rank=0 | step=12030/50000 progress=24.1% loss=0.133750 lr=9.039246e-05 lr_backbone=9.039246e-05 elapsed=12:18:43 avg_step=3.68s recent_step=3.91s eta=38:51:36 +2026-09-15 02:49:31 | INFO | train rank=0 | step=12040/50000 progress=24.1% loss=0.148098 lr=9.037296e-05 lr_backbone=9.037296e-05 elapsed=12:19:09 avg_step=3.68s recent_step=2.66s eta=38:50:27 +2026-09-15 02:49:56 | INFO | train rank=0 | step=12050/50000 progress=24.1% loss=0.121899 lr=9.035344e-05 lr_backbone=9.035344e-05 elapsed=12:19:34 avg_step=3.68s recent_step=2.51s eta=38:49:13 +2026-09-15 02:50:23 | INFO | train rank=0 | step=12060/50000 progress=24.1% loss=0.095812 lr=9.033391e-05 lr_backbone=9.033391e-05 elapsed=12:20:01 avg_step=3.68s recent_step=2.69s eta=38:48:05 +2026-09-15 02:50:51 | INFO | train rank=0 | step=12070/50000 progress=24.1% loss=0.090293 lr=9.031435e-05 lr_backbone=9.031435e-05 elapsed=12:20:29 avg_step=3.68s recent_step=2.73s eta=38:46:58 +2026-09-15 02:51:14 | INFO | train rank=0 | step=12080/50000 progress=24.2% loss=0.104102 lr=9.029478e-05 lr_backbone=9.029478e-05 elapsed=12:20:52 avg_step=3.68s recent_step=2.38s eta=38:45:40 +2026-09-15 02:51:46 | INFO | train rank=0 | step=12090/50000 progress=24.2% loss=0.100451 lr=9.027519e-05 lr_backbone=9.027519e-05 elapsed=12:21:24 avg_step=3.68s recent_step=3.12s eta=38:44:46 +2026-09-15 02:52:15 | INFO | train rank=0 | step=12100/50000 progress=24.2% loss=0.112609 lr=9.025559e-05 lr_backbone=9.025559e-05 elapsed=12:21:53 avg_step=3.68s recent_step=2.91s eta=38:43:45 +2026-09-15 02:52:43 | INFO | train rank=0 | step=12110/50000 progress=24.2% loss=0.112377 lr=9.023597e-05 lr_backbone=9.023597e-05 elapsed=12:22:21 avg_step=3.68s recent_step=2.85s eta=38:42:42 +2026-09-15 02:53:11 | INFO | train rank=0 | step=12120/50000 progress=24.2% loss=0.098755 lr=9.021633e-05 lr_backbone=9.021633e-05 elapsed=12:22:49 avg_step=3.68s recent_step=2.83s eta=38:41:39 +2026-09-15 02:53:36 | INFO | train rank=0 | step=12130/50000 progress=24.3% loss=0.085051 lr=9.019667e-05 lr_backbone=9.019667e-05 elapsed=12:23:14 avg_step=3.68s recent_step=2.44s eta=38:40:24 +2026-09-15 02:54:06 | INFO | train rank=0 | step=12140/50000 progress=24.3% loss=0.102925 lr=9.017699e-05 lr_backbone=9.017699e-05 elapsed=12:23:44 avg_step=3.68s recent_step=3.02s eta=38:39:26 +2026-09-15 02:54:44 | INFO | train rank=0 | step=12150/50000 progress=24.3% loss=0.097154 lr=9.015730e-05 lr_backbone=9.015730e-05 elapsed=12:24:22 avg_step=3.68s recent_step=3.77s eta=38:38:52 +2026-09-15 02:55:11 | INFO | train rank=0 | step=12160/50000 progress=24.3% loss=0.109835 lr=9.013759e-05 lr_backbone=9.013759e-05 elapsed=12:24:49 avg_step=3.68s recent_step=2.74s eta=38:37:47 +2026-09-15 02:55:36 | INFO | train rank=0 | step=12170/50000 progress=24.3% loss=0.120723 lr=9.011786e-05 lr_backbone=9.011786e-05 elapsed=12:25:14 avg_step=3.67s recent_step=2.44s eta=38:36:31 +2026-09-15 02:56:05 | INFO | train rank=0 | step=12180/50000 progress=24.4% loss=0.095521 lr=9.009811e-05 lr_backbone=9.009811e-05 elapsed=12:25:43 avg_step=3.67s recent_step=2.93s eta=38:35:32 +2026-09-15 02:56:38 | INFO | train rank=0 | step=12190/50000 progress=24.4% loss=0.097316 lr=9.007835e-05 lr_backbone=9.007835e-05 elapsed=12:26:16 avg_step=3.67s recent_step=3.32s eta=38:34:44 +2026-09-15 02:56:58 | INFO | train rank=0 | step=12200/50000 progress=24.4% loss=0.106261 lr=9.005857e-05 lr_backbone=9.005857e-05 elapsed=12:26:36 avg_step=3.67s recent_step=2.01s eta=38:33:16 +2026-09-15 02:57:28 | INFO | train rank=0 | step=12210/50000 progress=24.4% loss=0.090624 lr=9.003877e-05 lr_backbone=9.003877e-05 elapsed=12:27:06 avg_step=3.67s recent_step=3.01s eta=38:32:19 +2026-09-15 02:57:56 | INFO | train rank=0 | step=12220/50000 progress=24.4% loss=0.106468 lr=9.001895e-05 lr_backbone=9.001895e-05 elapsed=12:27:34 avg_step=3.67s recent_step=2.74s eta=38:31:13 +2026-09-15 02:58:28 | INFO | train rank=0 | step=12230/50000 progress=24.5% loss=0.091751 lr=8.999912e-05 lr_backbone=8.999912e-05 elapsed=12:28:06 avg_step=3.67s recent_step=3.26s eta=38:30:24 +2026-09-15 02:58:56 | INFO | train rank=0 | step=12240/50000 progress=24.5% loss=0.118163 lr=8.997927e-05 lr_backbone=8.997927e-05 elapsed=12:28:34 avg_step=3.67s recent_step=2.75s eta=38:29:19 +2026-09-15 02:59:21 | INFO | train rank=0 | step=12250/50000 progress=24.5% loss=0.136529 lr=8.995940e-05 lr_backbone=8.995940e-05 elapsed=12:28:59 avg_step=3.67s recent_step=2.57s eta=38:28:08 +2026-09-15 02:59:43 | INFO | train rank=0 | step=12260/50000 progress=24.5% loss=0.093869 lr=8.993951e-05 lr_backbone=8.993951e-05 elapsed=12:29:21 avg_step=3.67s recent_step=2.20s eta=38:26:46 +2026-09-15 03:00:20 | INFO | train rank=0 | step=12270/50000 progress=24.5% loss=0.115862 lr=8.991961e-05 lr_backbone=8.991961e-05 elapsed=12:29:58 avg_step=3.67s recent_step=3.64s eta=38:26:09 +2026-09-15 03:00:48 | INFO | train rank=0 | step=12280/50000 progress=24.6% loss=0.099251 lr=8.989969e-05 lr_backbone=8.989969e-05 elapsed=12:30:26 avg_step=3.67s recent_step=2.82s eta=38:25:06 +2026-09-15 03:01:12 | INFO | train rank=0 | step=12290/50000 progress=24.6% loss=0.096989 lr=8.987975e-05 lr_backbone=8.987975e-05 elapsed=12:30:50 avg_step=3.67s recent_step=2.40s eta=38:23:50 +2026-09-15 03:01:42 | INFO | train rank=0 | step=12300/50000 progress=24.6% loss=0.096716 lr=8.985979e-05 lr_backbone=8.985979e-05 elapsed=12:31:20 avg_step=3.67s recent_step=3.01s eta=38:22:53 +2026-09-15 03:02:21 | INFO | train rank=0 | step=12310/50000 progress=24.6% loss=0.081551 lr=8.983982e-05 lr_backbone=8.983982e-05 elapsed=12:31:59 avg_step=3.67s recent_step=3.89s eta=38:22:24 +2026-09-15 03:02:49 | INFO | train rank=0 | step=12320/50000 progress=24.6% loss=0.109278 lr=8.981983e-05 lr_backbone=8.981983e-05 elapsed=12:32:27 avg_step=3.66s recent_step=2.82s eta=38:21:21 +2026-09-15 03:03:10 | INFO | train rank=0 | step=12330/50000 progress=24.7% loss=0.105198 lr=8.979982e-05 lr_backbone=8.979982e-05 elapsed=12:32:48 avg_step=3.66s recent_step=2.11s eta=38:19:57 +2026-09-15 03:03:37 | INFO | train rank=0 | step=12340/50000 progress=24.7% loss=0.117549 lr=8.977980e-05 lr_backbone=8.977980e-05 elapsed=12:33:15 avg_step=3.66s recent_step=2.72s eta=38:18:51 +2026-09-15 03:04:12 | INFO | train rank=0 | step=12350/50000 progress=24.7% loss=0.102396 lr=8.975975e-05 lr_backbone=8.975975e-05 elapsed=12:33:50 avg_step=3.66s recent_step=3.41s eta=38:18:07 +2026-09-15 03:04:38 | INFO | train rank=0 | step=12360/50000 progress=24.7% loss=0.110364 lr=8.973969e-05 lr_backbone=8.973969e-05 elapsed=12:34:16 avg_step=3.66s recent_step=2.65s eta=38:17:00 +2026-09-15 03:05:01 | INFO | train rank=0 | step=12370/50000 progress=24.7% loss=0.135486 lr=8.971962e-05 lr_backbone=8.971962e-05 elapsed=12:34:39 avg_step=3.66s recent_step=2.29s eta=38:15:41 +2026-09-15 03:05:23 | INFO | train rank=0 | step=12380/50000 progress=24.8% loss=0.121300 lr=8.969952e-05 lr_backbone=8.969952e-05 elapsed=12:35:01 avg_step=3.66s recent_step=2.24s eta=38:14:22 +2026-09-15 03:05:57 | INFO | train rank=0 | step=12390/50000 progress=24.8% loss=0.097698 lr=8.967941e-05 lr_backbone=8.967941e-05 elapsed=12:35:35 avg_step=3.66s recent_step=3.39s eta=38:13:37 +2026-09-15 03:06:29 | INFO | train rank=0 | step=12400/50000 progress=24.8% loss=0.117871 lr=8.965928e-05 lr_backbone=8.965928e-05 elapsed=12:36:07 avg_step=3.66s recent_step=3.13s eta=38:12:44 +2026-09-15 03:06:52 | INFO | train rank=0 | step=12410/50000 progress=24.8% loss=0.117081 lr=8.963913e-05 lr_backbone=8.963913e-05 elapsed=12:36:30 avg_step=3.66s recent_step=2.32s eta=38:11:27 +2026-09-15 03:07:18 | INFO | train rank=0 | step=12420/50000 progress=24.8% loss=0.126595 lr=8.961897e-05 lr_backbone=8.961897e-05 elapsed=12:36:56 avg_step=3.66s recent_step=2.64s eta=38:10:20 +2026-09-15 03:07:50 | INFO | train rank=0 | step=12430/50000 progress=24.9% loss=0.091825 lr=8.959878e-05 lr_backbone=8.959878e-05 elapsed=12:37:28 avg_step=3.66s recent_step=3.18s eta=38:09:29 +2026-09-15 03:08:13 | INFO | train rank=0 | step=12440/50000 progress=24.9% loss=0.099397 lr=8.957858e-05 lr_backbone=8.957858e-05 elapsed=12:37:51 avg_step=3.66s recent_step=2.34s eta=38:08:13 +2026-09-15 03:08:40 | INFO | train rank=0 | step=12450/50000 progress=24.9% loss=0.096992 lr=8.955837e-05 lr_backbone=8.955837e-05 elapsed=12:38:18 avg_step=3.65s recent_step=2.69s eta=38:07:07 +2026-09-15 03:09:09 | INFO | train rank=0 | step=12460/50000 progress=24.9% loss=0.086026 lr=8.953813e-05 lr_backbone=8.953813e-05 elapsed=12:38:47 avg_step=3.65s recent_step=2.86s eta=38:06:07 +2026-09-15 03:09:45 | INFO | train rank=0 | step=12470/50000 progress=24.9% loss=0.085515 lr=8.951788e-05 lr_backbone=8.951788e-05 elapsed=12:39:23 avg_step=3.65s recent_step=3.62s eta=38:05:29 +2026-09-15 03:10:14 | INFO | train rank=0 | step=12480/50000 progress=25.0% loss=0.106969 lr=8.949761e-05 lr_backbone=8.949761e-05 elapsed=12:39:52 avg_step=3.65s recent_step=2.85s eta=38:04:28 +2026-09-15 03:10:42 | INFO | train rank=0 | step=12490/50000 progress=25.0% loss=0.104426 lr=8.947733e-05 lr_backbone=8.947733e-05 elapsed=12:40:20 avg_step=3.65s recent_step=2.84s eta=38:03:27 +2026-09-15 03:11:10 | INFO | train rank=0 | step=12500/50000 progress=25.0% loss=0.114412 lr=8.945703e-05 lr_backbone=8.945703e-05 elapsed=12:40:48 avg_step=3.65s recent_step=2.75s eta=38:02:24 +2026-09-15 03:11:49 | INFO | train rank=0 | step=12510/50000 progress=25.0% loss=0.094901 lr=8.943671e-05 lr_backbone=8.943671e-05 elapsed=12:41:27 avg_step=3.65s recent_step=3.95s eta=38:01:56 +2026-09-15 03:12:17 | INFO | train rank=0 | step=12520/50000 progress=25.0% loss=0.088884 lr=8.941637e-05 lr_backbone=8.941637e-05 elapsed=12:41:55 avg_step=3.65s recent_step=2.75s eta=38:00:53 +2026-09-15 03:12:45 | INFO | train rank=0 | step=12530/50000 progress=25.1% loss=0.088143 lr=8.939601e-05 lr_backbone=8.939601e-05 elapsed=12:42:23 avg_step=3.65s recent_step=2.81s eta=37:59:51 +2026-09-15 03:13:15 | INFO | train rank=0 | step=12540/50000 progress=25.1% loss=0.093101 lr=8.937564e-05 lr_backbone=8.937564e-05 elapsed=12:42:53 avg_step=3.65s recent_step=3.02s eta=37:58:56 +2026-09-15 03:13:58 | INFO | train rank=0 | step=12550/50000 progress=25.1% loss=0.120568 lr=8.935525e-05 lr_backbone=8.935525e-05 elapsed=12:43:36 avg_step=3.65s recent_step=4.33s eta=37:58:40 +2026-09-15 03:14:24 | INFO | train rank=0 | step=12560/50000 progress=25.1% loss=0.153515 lr=8.933485e-05 lr_backbone=8.933485e-05 elapsed=12:44:02 avg_step=3.65s recent_step=2.56s eta=37:57:31 +2026-09-15 03:14:47 | INFO | train rank=0 | step=12570/50000 progress=25.1% loss=0.092079 lr=8.931442e-05 lr_backbone=8.931442e-05 elapsed=12:44:25 avg_step=3.65s recent_step=2.31s eta=37:56:14 +2026-09-15 03:15:12 | INFO | train rank=0 | step=12580/50000 progress=25.2% loss=0.108221 lr=8.929398e-05 lr_backbone=8.929398e-05 elapsed=12:44:50 avg_step=3.65s recent_step=2.52s eta=37:55:04 +2026-09-15 03:15:44 | INFO | train rank=0 | step=12590/50000 progress=25.2% loss=0.096065 lr=8.927352e-05 lr_backbone=8.927352e-05 elapsed=12:45:22 avg_step=3.65s recent_step=3.20s eta=37:54:14 +2026-09-15 03:16:09 | INFO | train rank=0 | step=12600/50000 progress=25.2% loss=0.121332 lr=8.925305e-05 lr_backbone=8.925305e-05 elapsed=12:45:47 avg_step=3.65s recent_step=2.48s eta=37:53:03 +2026-09-15 03:16:37 | INFO | train rank=0 | step=12610/50000 progress=25.2% loss=0.122372 lr=8.923256e-05 lr_backbone=8.923256e-05 elapsed=12:46:15 avg_step=3.65s recent_step=2.79s eta=37:52:01 +2026-09-15 03:17:04 | INFO | train rank=0 | step=12620/50000 progress=25.2% loss=0.093401 lr=8.921205e-05 lr_backbone=8.921205e-05 elapsed=12:46:42 avg_step=3.65s recent_step=2.72s eta=37:50:57 +2026-09-15 03:17:38 | INFO | train rank=0 | step=12630/50000 progress=25.3% loss=0.110475 lr=8.919152e-05 lr_backbone=8.919152e-05 elapsed=12:47:16 avg_step=3.65s recent_step=3.42s eta=37:50:14 +2026-09-15 03:18:03 | INFO | train rank=0 | step=12640/50000 progress=25.3% loss=0.118831 lr=8.917098e-05 lr_backbone=8.917098e-05 elapsed=12:47:41 avg_step=3.64s recent_step=2.51s eta=37:49:04 +2026-09-15 03:18:26 | INFO | train rank=0 | step=12650/50000 progress=25.3% loss=0.084242 lr=8.915041e-05 lr_backbone=8.915041e-05 elapsed=12:48:04 avg_step=3.64s recent_step=2.29s eta=37:47:48 +2026-09-15 03:18:58 | INFO | train rank=0 | step=12660/50000 progress=25.3% loss=0.103357 lr=8.912984e-05 lr_backbone=8.912984e-05 elapsed=12:48:36 avg_step=3.64s recent_step=3.20s eta=37:46:58 +2026-09-15 03:19:33 | INFO | train rank=0 | step=12670/50000 progress=25.3% loss=0.091254 lr=8.910924e-05 lr_backbone=8.910924e-05 elapsed=12:49:11 avg_step=3.64s recent_step=3.44s eta=37:46:16 +2026-09-15 03:19:58 | INFO | train rank=0 | step=12680/50000 progress=25.4% loss=0.094506 lr=8.908863e-05 lr_backbone=8.908863e-05 elapsed=12:49:36 avg_step=3.64s recent_step=2.52s eta=37:45:06 +2026-09-15 03:20:25 | INFO | train rank=0 | step=12690/50000 progress=25.4% loss=0.132454 lr=8.906800e-05 lr_backbone=8.906800e-05 elapsed=12:50:03 avg_step=3.64s recent_step=2.68s eta=37:44:02 +2026-09-15 03:20:52 | INFO | train rank=0 | step=12700/50000 progress=25.4% loss=0.105780 lr=8.904735e-05 lr_backbone=8.904735e-05 elapsed=12:50:30 avg_step=3.64s recent_step=2.79s eta=37:43:00 +2026-09-15 03:21:26 | INFO | train rank=0 | step=12710/50000 progress=25.4% loss=0.104233 lr=8.902669e-05 lr_backbone=8.902669e-05 elapsed=12:51:04 avg_step=3.64s recent_step=3.34s eta=37:42:15 +2026-09-15 03:21:54 | INFO | train rank=0 | step=12720/50000 progress=25.4% loss=0.106500 lr=8.900601e-05 lr_backbone=8.900601e-05 elapsed=12:51:32 avg_step=3.64s recent_step=2.82s eta=37:41:14 +2026-09-15 03:22:23 | INFO | train rank=0 | step=12730/50000 progress=25.5% loss=0.109846 lr=8.898531e-05 lr_backbone=8.898531e-05 elapsed=12:52:01 avg_step=3.64s recent_step=2.87s eta=37:40:15 +2026-09-15 03:22:51 | INFO | train rank=0 | step=12740/50000 progress=25.5% loss=0.092779 lr=8.896460e-05 lr_backbone=8.896460e-05 elapsed=12:52:29 avg_step=3.64s recent_step=2.86s eta=37:39:16 +2026-09-15 03:23:26 | INFO | train rank=0 | step=12750/50000 progress=25.5% loss=0.089869 lr=8.894386e-05 lr_backbone=8.894386e-05 elapsed=12:53:04 avg_step=3.64s recent_step=3.44s eta=37:38:34 +2026-09-15 03:23:50 | INFO | train rank=0 | step=12760/50000 progress=25.5% loss=0.138151 lr=8.892312e-05 lr_backbone=8.892312e-05 elapsed=12:53:28 avg_step=3.64s recent_step=2.45s eta=37:37:23 +2026-09-15 03:24:20 | INFO | train rank=0 | step=12770/50000 progress=25.5% loss=0.101718 lr=8.890235e-05 lr_backbone=8.890235e-05 elapsed=12:53:58 avg_step=3.64s recent_step=2.95s eta=37:36:26 +2026-09-15 03:24:48 | INFO | train rank=0 | step=12780/50000 progress=25.6% loss=0.121038 lr=8.888157e-05 lr_backbone=8.888157e-05 elapsed=12:54:26 avg_step=3.64s recent_step=2.83s eta=37:35:26 +2026-09-15 03:25:24 | INFO | train rank=0 | step=12790/50000 progress=25.6% loss=0.092288 lr=8.886077e-05 lr_backbone=8.886077e-05 elapsed=12:55:02 avg_step=3.64s recent_step=3.59s eta=37:34:49 +2026-09-15 03:25:42 | INFO | train rank=0 | step=12800/50000 progress=25.6% loss=0.109533 lr=8.883995e-05 lr_backbone=8.883995e-05 elapsed=12:55:20 avg_step=3.63s recent_step=1.79s eta=37:33:19 +2026-09-15 03:26:08 | INFO | train rank=0 | step=12810/50000 progress=25.6% loss=0.114149 lr=8.881912e-05 lr_backbone=8.881912e-05 elapsed=12:55:46 avg_step=3.63s recent_step=2.65s eta=37:32:14 +2026-09-15 03:26:33 | INFO | train rank=0 | step=12820/50000 progress=25.6% loss=0.112724 lr=8.879826e-05 lr_backbone=8.879826e-05 elapsed=12:56:11 avg_step=3.63s recent_step=2.52s eta=37:31:05 +2026-09-15 03:27:11 | INFO | train rank=0 | step=12830/50000 progress=25.7% loss=0.116088 lr=8.877740e-05 lr_backbone=8.877740e-05 elapsed=12:56:49 avg_step=3.63s recent_step=3.80s eta=37:30:34 +2026-09-15 03:27:33 | INFO | train rank=0 | step=12840/50000 progress=25.7% loss=0.121607 lr=8.875651e-05 lr_backbone=8.875651e-05 elapsed=12:57:11 avg_step=3.63s recent_step=2.21s eta=37:29:16 +2026-09-15 03:27:55 | INFO | train rank=0 | step=12850/50000 progress=25.7% loss=0.117349 lr=8.873561e-05 lr_backbone=8.873561e-05 elapsed=12:57:33 avg_step=3.63s recent_step=2.14s eta=37:27:57 +2026-09-15 03:28:18 | INFO | train rank=0 | step=12860/50000 progress=25.7% loss=0.106991 lr=8.871469e-05 lr_backbone=8.871469e-05 elapsed=12:57:56 avg_step=3.63s recent_step=2.28s eta=37:26:41 +2026-09-15 03:28:55 | INFO | train rank=0 | step=12870/50000 progress=25.7% loss=0.121406 lr=8.869376e-05 lr_backbone=8.869376e-05 elapsed=12:58:33 avg_step=3.63s recent_step=3.70s eta=37:26:07 +2026-09-15 03:29:22 | INFO | train rank=0 | step=12880/50000 progress=25.8% loss=0.120151 lr=8.867280e-05 lr_backbone=8.867280e-05 elapsed=12:59:00 avg_step=3.63s recent_step=2.74s eta=37:25:05 +2026-09-15 03:29:51 | INFO | train rank=0 | step=12890/50000 progress=25.8% loss=0.100376 lr=8.865183e-05 lr_backbone=8.865183e-05 elapsed=12:59:29 avg_step=3.63s recent_step=2.89s eta=37:24:08 +2026-09-15 03:30:14 | INFO | train rank=0 | step=12900/50000 progress=25.8% loss=0.104585 lr=8.863085e-05 lr_backbone=8.863085e-05 elapsed=12:59:52 avg_step=3.63s recent_step=2.33s eta=37:22:54 +2026-09-15 03:30:45 | INFO | train rank=0 | step=12910/50000 progress=25.8% loss=0.085472 lr=8.860984e-05 lr_backbone=8.860984e-05 elapsed=13:00:23 avg_step=3.63s recent_step=3.03s eta=37:22:01 +2026-09-15 03:31:03 | INFO | train rank=0 | step=12920/50000 progress=25.8% loss=0.147312 lr=8.858882e-05 lr_backbone=8.858882e-05 elapsed=13:00:41 avg_step=3.63s recent_step=1.83s eta=37:20:33 +2026-09-15 03:31:30 | INFO | train rank=0 | step=12930/50000 progress=25.9% loss=0.096972 lr=8.856779e-05 lr_backbone=8.856779e-05 elapsed=13:01:08 avg_step=3.62s recent_step=2.71s eta=37:19:30 +2026-09-15 03:31:56 | INFO | train rank=0 | step=12940/50000 progress=25.9% loss=0.151107 lr=8.854673e-05 lr_backbone=8.854673e-05 elapsed=13:01:34 avg_step=3.62s recent_step=2.62s eta=37:18:25 +2026-09-15 03:32:34 | INFO | train rank=0 | step=12950/50000 progress=25.9% loss=0.087982 lr=8.852566e-05 lr_backbone=8.852566e-05 elapsed=13:02:12 avg_step=3.62s recent_step=3.77s eta=37:17:53 +2026-09-15 03:32:58 | INFO | train rank=0 | step=12960/50000 progress=25.9% loss=0.130096 lr=8.850457e-05 lr_backbone=8.850457e-05 elapsed=13:02:36 avg_step=3.62s recent_step=2.41s eta=37:16:42 +2026-09-15 03:33:23 | INFO | train rank=0 | step=12970/50000 progress=25.9% loss=0.089955 lr=8.848347e-05 lr_backbone=8.848347e-05 elapsed=13:03:01 avg_step=3.62s recent_step=2.52s eta=37:15:35 +2026-09-15 03:33:50 | INFO | train rank=0 | step=12980/50000 progress=26.0% loss=0.111447 lr=8.846235e-05 lr_backbone=8.846235e-05 elapsed=13:03:28 avg_step=3.62s recent_step=2.68s eta=37:14:32 +2026-09-15 03:34:14 | INFO | train rank=0 | step=12990/50000 progress=26.0% loss=0.101675 lr=8.844121e-05 lr_backbone=8.844121e-05 elapsed=13:03:52 avg_step=3.62s recent_step=2.44s eta=37:13:22 +2026-09-15 03:34:41 | INFO | train rank=0 | step=13000/50000 progress=26.0% loss=0.093678 lr=8.842006e-05 lr_backbone=8.842006e-05 elapsed=13:04:19 avg_step=3.62s recent_step=2.62s eta=37:12:17 +2026-09-15 03:35:01 | INFO | train rank=0 | step=13010/50000 progress=26.0% loss=0.109358 lr=8.839888e-05 lr_backbone=8.839888e-05 elapsed=13:04:39 avg_step=3.62s recent_step=2.00s eta=37:10:55 +2026-09-15 03:35:19 | INFO | train rank=0 | step=13020/50000 progress=26.0% loss=0.087018 lr=8.837770e-05 lr_backbone=8.837770e-05 elapsed=13:04:57 avg_step=3.62s recent_step=1.82s eta=37:09:27 +2026-09-15 03:35:46 | INFO | train rank=0 | step=13030/50000 progress=26.1% loss=0.109700 lr=8.835649e-05 lr_backbone=8.835649e-05 elapsed=13:05:24 avg_step=3.62s recent_step=2.71s eta=37:08:26 +2026-09-15 03:36:10 | INFO | train rank=0 | step=13040/50000 progress=26.1% loss=0.127024 lr=8.833527e-05 lr_backbone=8.833527e-05 elapsed=13:05:48 avg_step=3.62s recent_step=2.41s eta=37:07:15 +2026-09-15 03:36:37 | INFO | train rank=0 | step=13050/50000 progress=26.1% loss=0.122679 lr=8.831403e-05 lr_backbone=8.831403e-05 elapsed=13:06:15 avg_step=3.62s recent_step=2.75s eta=37:06:14 +2026-09-15 03:36:59 | INFO | train rank=0 | step=13060/50000 progress=26.1% loss=0.119427 lr=8.829277e-05 lr_backbone=8.829277e-05 elapsed=13:06:37 avg_step=3.61s recent_step=2.15s eta=37:04:57 +2026-09-15 03:37:22 | INFO | train rank=0 | step=13070/50000 progress=26.1% loss=0.102509 lr=8.827150e-05 lr_backbone=8.827150e-05 elapsed=13:07:00 avg_step=3.61s recent_step=2.34s eta=37:03:45 +2026-09-15 03:37:40 | INFO | train rank=0 | step=13080/50000 progress=26.2% loss=0.086180 lr=8.825021e-05 lr_backbone=8.825021e-05 elapsed=13:07:18 avg_step=3.61s recent_step=1.78s eta=37:02:17 +2026-09-15 03:38:06 | INFO | train rank=0 | step=13090/50000 progress=26.2% loss=0.120652 lr=8.822891e-05 lr_backbone=8.822891e-05 elapsed=13:07:44 avg_step=3.61s recent_step=2.59s eta=37:01:12 +2026-09-15 03:38:32 | INFO | train rank=0 | step=13100/50000 progress=26.2% loss=0.098483 lr=8.820758e-05 lr_backbone=8.820758e-05 elapsed=13:08:10 avg_step=3.61s recent_step=2.62s eta=37:00:08 +2026-09-15 03:38:56 | INFO | train rank=0 | step=13110/50000 progress=26.2% loss=0.126441 lr=8.818624e-05 lr_backbone=8.818624e-05 elapsed=13:08:34 avg_step=3.61s recent_step=2.36s eta=36:58:56 +2026-09-15 03:39:24 | INFO | train rank=0 | step=13120/50000 progress=26.2% loss=0.108606 lr=8.816489e-05 lr_backbone=8.816489e-05 elapsed=13:09:02 avg_step=3.61s recent_step=2.84s eta=36:57:59 +2026-09-15 03:40:00 | INFO | train rank=0 | step=13130/50000 progress=26.3% loss=0.117614 lr=8.814351e-05 lr_backbone=8.814351e-05 elapsed=13:09:38 avg_step=3.61s recent_step=3.56s eta=36:57:21 +2026-09-15 03:40:31 | INFO | train rank=0 | step=13140/50000 progress=26.3% loss=0.096638 lr=8.812213e-05 lr_backbone=8.812213e-05 elapsed=13:10:09 avg_step=3.61s recent_step=3.07s eta=36:56:30 +2026-09-15 03:41:07 | INFO | train rank=0 | step=13150/50000 progress=26.3% loss=0.100592 lr=8.810072e-05 lr_backbone=8.810072e-05 elapsed=13:10:45 avg_step=3.61s recent_step=3.61s eta=36:55:54 +2026-09-15 03:41:53 | INFO | train rank=0 | step=13160/50000 progress=26.3% loss=0.099884 lr=8.807930e-05 lr_backbone=8.807930e-05 elapsed=13:11:31 avg_step=3.61s recent_step=4.67s eta=36:55:48 +2026-09-15 03:42:39 | INFO | train rank=0 | step=13170/50000 progress=26.3% loss=0.090966 lr=8.805786e-05 lr_backbone=8.805786e-05 elapsed=13:12:17 avg_step=3.61s recent_step=4.57s eta=36:55:39 +2026-09-15 03:43:26 | INFO | train rank=0 | step=13180/50000 progress=26.4% loss=0.100729 lr=8.803640e-05 lr_backbone=8.803640e-05 elapsed=13:13:04 avg_step=3.61s recent_step=4.67s eta=36:55:32 +2026-09-15 03:44:15 | INFO | train rank=0 | step=13190/50000 progress=26.4% loss=0.099115 lr=8.801493e-05 lr_backbone=8.801493e-05 elapsed=13:13:53 avg_step=3.61s recent_step=4.88s eta=36:55:32 +2026-09-15 03:45:05 | INFO | train rank=0 | step=13200/50000 progress=26.4% loss=0.093536 lr=8.799344e-05 lr_backbone=8.799344e-05 elapsed=13:14:43 avg_step=3.61s recent_step=5.00s eta=36:55:34 +2026-09-15 03:46:09 | INFO | train rank=0 | step=13210/50000 progress=26.4% loss=0.119305 lr=8.797193e-05 lr_backbone=8.797193e-05 elapsed=13:15:47 avg_step=3.61s recent_step=6.40s eta=36:56:16 +2026-09-15 03:47:15 | INFO | train rank=0 | step=13220/50000 progress=26.4% loss=0.114782 lr=8.795041e-05 lr_backbone=8.795041e-05 elapsed=13:16:53 avg_step=3.62s recent_step=6.60s eta=36:57:03 +2026-09-15 03:48:11 | INFO | train rank=0 | step=13230/50000 progress=26.5% loss=0.088253 lr=8.792887e-05 lr_backbone=8.792887e-05 elapsed=13:17:49 avg_step=3.62s recent_step=5.61s eta=36:57:22 +2026-09-15 03:49:20 | INFO | train rank=0 | step=13240/50000 progress=26.5% loss=0.095913 lr=8.790732e-05 lr_backbone=8.790732e-05 elapsed=13:18:58 avg_step=3.62s recent_step=6.95s eta=36:58:18 +2026-09-15 03:50:22 | INFO | train rank=0 | step=13250/50000 progress=26.5% loss=0.108743 lr=8.788574e-05 lr_backbone=8.788574e-05 elapsed=13:20:00 avg_step=3.62s recent_step=6.22s eta=36:58:54 +2026-09-15 03:51:39 | INFO | train rank=0 | step=13260/50000 progress=26.5% loss=0.097249 lr=8.786415e-05 lr_backbone=8.786415e-05 elapsed=13:21:18 avg_step=3.63s recent_step=7.70s eta=37:00:11 +2026-09-15 03:52:52 | INFO | train rank=0 | step=13270/50000 progress=26.5% loss=0.107663 lr=8.784255e-05 lr_backbone=8.784255e-05 elapsed=13:22:30 avg_step=3.63s recent_step=7.26s eta=37:01:15 +2026-09-15 03:53:53 | INFO | train rank=0 | step=13280/50000 progress=26.6% loss=0.091767 lr=8.782093e-05 lr_backbone=8.782093e-05 elapsed=13:23:31 avg_step=3.63s recent_step=6.06s eta=37:01:46 +2026-09-15 03:54:58 | INFO | train rank=0 | step=13290/50000 progress=26.6% loss=0.124721 lr=8.779929e-05 lr_backbone=8.779929e-05 elapsed=13:24:36 avg_step=3.63s recent_step=6.56s eta=37:02:31 +2026-09-15 03:56:18 | INFO | train rank=0 | step=13300/50000 progress=26.6% loss=0.095861 lr=8.777763e-05 lr_backbone=8.777763e-05 elapsed=13:25:56 avg_step=3.64s recent_step=7.98s eta=37:03:55 +2026-09-15 03:57:26 | INFO | train rank=0 | step=13310/50000 progress=26.6% loss=0.098936 lr=8.775596e-05 lr_backbone=8.775596e-05 elapsed=13:27:04 avg_step=3.64s recent_step=6.75s eta=37:04:44 +2026-09-15 03:58:38 | INFO | train rank=0 | step=13320/50000 progress=26.6% loss=0.084643 lr=8.773427e-05 lr_backbone=8.773427e-05 elapsed=13:28:16 avg_step=3.64s recent_step=7.23s eta=37:05:47 +2026-09-15 03:59:44 | INFO | train rank=0 | step=13330/50000 progress=26.7% loss=0.092182 lr=8.771257e-05 lr_backbone=8.771257e-05 elapsed=13:29:22 avg_step=3.64s recent_step=6.63s eta=37:06:33 +2026-09-15 04:00:48 | INFO | train rank=0 | step=13340/50000 progress=26.7% loss=0.104977 lr=8.769085e-05 lr_backbone=8.769085e-05 elapsed=13:30:26 avg_step=3.65s recent_step=6.42s eta=37:07:13 +2026-09-15 04:02:01 | INFO | train rank=0 | step=13350/50000 progress=26.7% loss=0.120788 lr=8.766911e-05 lr_backbone=8.766911e-05 elapsed=13:31:39 avg_step=3.65s recent_step=7.24s eta=37:08:15 +2026-09-15 04:03:07 | INFO | train rank=0 | step=13360/50000 progress=26.7% loss=0.106120 lr=8.764736e-05 lr_backbone=8.764736e-05 elapsed=13:32:45 avg_step=3.65s recent_step=6.59s eta=37:08:59 +2026-09-15 04:04:04 | INFO | train rank=0 | step=13370/50000 progress=26.7% loss=0.110841 lr=8.762559e-05 lr_backbone=8.762559e-05 elapsed=13:33:42 avg_step=3.65s recent_step=5.75s eta=37:09:20 +2026-09-15 04:05:09 | INFO | train rank=0 | step=13380/50000 progress=26.8% loss=0.104239 lr=8.760380e-05 lr_backbone=8.760380e-05 elapsed=13:34:47 avg_step=3.65s recent_step=6.51s eta=37:10:02 +2026-09-15 04:06:20 | INFO | train rank=0 | step=13390/50000 progress=26.8% loss=0.092343 lr=8.758200e-05 lr_backbone=8.758200e-05 elapsed=13:35:58 avg_step=3.66s recent_step=7.02s eta=37:10:57 +2026-09-15 04:07:24 | INFO | train rank=0 | step=13400/50000 progress=26.8% loss=0.099803 lr=8.756018e-05 lr_backbone=8.756018e-05 elapsed=13:37:02 avg_step=3.66s recent_step=6.46s eta=37:11:37 +2026-09-15 04:08:36 | INFO | train rank=0 | step=13410/50000 progress=26.8% loss=0.095091 lr=8.753834e-05 lr_backbone=8.753834e-05 elapsed=13:38:14 avg_step=3.66s recent_step=7.20s eta=37:12:37 +2026-09-15 04:09:40 | INFO | train rank=0 | step=13420/50000 progress=26.8% loss=0.128407 lr=8.751649e-05 lr_backbone=8.751649e-05 elapsed=13:39:18 avg_step=3.66s recent_step=6.39s eta=37:13:15 +2026-09-15 04:10:53 | INFO | train rank=0 | step=13430/50000 progress=26.9% loss=0.091362 lr=8.749462e-05 lr_backbone=8.749462e-05 elapsed=13:40:31 avg_step=3.67s recent_step=7.28s eta=37:14:17 +2026-09-15 04:11:59 | INFO | train rank=0 | step=13440/50000 progress=26.9% loss=0.079334 lr=8.747273e-05 lr_backbone=8.747273e-05 elapsed=13:41:37 avg_step=3.67s recent_step=6.65s eta=37:15:01 +2026-09-15 04:13:08 | INFO | train rank=0 | step=13450/50000 progress=26.9% loss=0.111416 lr=8.745083e-05 lr_backbone=8.745083e-05 elapsed=13:42:46 avg_step=3.67s recent_step=6.82s eta=37:15:50 +2026-09-15 04:14:17 | INFO | train rank=0 | step=13460/50000 progress=26.9% loss=0.109416 lr=8.742891e-05 lr_backbone=8.742891e-05 elapsed=13:43:55 avg_step=3.67s recent_step=6.98s eta=37:16:44 +2026-09-15 04:15:29 | INFO | train rank=0 | step=13470/50000 progress=26.9% loss=0.124633 lr=8.740698e-05 lr_backbone=8.740698e-05 elapsed=13:45:07 avg_step=3.68s recent_step=7.14s eta=37:17:41 +2026-09-15 04:16:36 | INFO | train rank=0 | step=13480/50000 progress=27.0% loss=0.108555 lr=8.738503e-05 lr_backbone=8.738503e-05 elapsed=13:46:14 avg_step=3.68s recent_step=6.71s eta=37:18:26 +2026-09-15 04:17:51 | INFO | train rank=0 | step=13490/50000 progress=27.0% loss=0.097971 lr=8.736306e-05 lr_backbone=8.736306e-05 elapsed=13:47:29 avg_step=3.68s recent_step=7.54s eta=37:19:34 +2026-09-15 04:19:03 | INFO | train rank=0 | step=13500/50000 progress=27.0% loss=0.127145 lr=8.734107e-05 lr_backbone=8.734107e-05 elapsed=13:48:41 avg_step=3.68s recent_step=7.20s eta=37:20:32 +2026-09-15 04:20:07 | INFO | train rank=0 | step=13510/50000 progress=27.0% loss=0.119146 lr=8.731907e-05 lr_backbone=8.731907e-05 elapsed=13:49:45 avg_step=3.69s recent_step=6.35s eta=37:21:08 +2026-09-15 04:21:10 | INFO | train rank=0 | step=13520/50000 progress=27.0% loss=0.091304 lr=8.729706e-05 lr_backbone=8.729706e-05 elapsed=13:50:48 avg_step=3.69s recent_step=6.32s eta=37:21:42 +2026-09-15 04:22:56 | INFO | train rank=0 | step=13530/50000 progress=27.1% loss=0.115738 lr=8.727502e-05 lr_backbone=8.727502e-05 elapsed=13:52:34 avg_step=3.69s recent_step=10.63s eta=37:24:12 +2026-09-15 04:24:46 | INFO | train rank=0 | step=13540/50000 progress=27.1% loss=0.112398 lr=8.725298e-05 lr_backbone=8.725298e-05 elapsed=13:54:24 avg_step=3.70s recent_step=10.95s eta=37:26:51 +2026-09-15 04:26:50 | INFO | train rank=0 | step=13550/50000 progress=27.1% loss=0.093332 lr=8.723091e-05 lr_backbone=8.723091e-05 elapsed=13:56:28 avg_step=3.70s recent_step=12.43s eta=37:30:09 +2026-09-15 04:28:50 | INFO | train rank=0 | step=13560/50000 progress=27.1% loss=0.109650 lr=8.720883e-05 lr_backbone=8.720883e-05 elapsed=13:58:28 avg_step=3.71s recent_step=12.01s eta=37:33:15 +2026-09-15 04:30:49 | INFO | train rank=0 | step=13570/50000 progress=27.1% loss=0.103050 lr=8.718673e-05 lr_backbone=8.718673e-05 elapsed=14:00:27 avg_step=3.72s recent_step=11.83s eta=37:36:16 +2026-09-15 04:32:50 | INFO | train rank=0 | step=13580/50000 progress=27.2% loss=0.084294 lr=8.716462e-05 lr_backbone=8.716462e-05 elapsed=14:02:28 avg_step=3.72s recent_step=12.17s eta=37:39:25 +2026-09-15 04:34:43 | INFO | train rank=0 | step=13590/50000 progress=27.2% loss=0.094862 lr=8.714249e-05 lr_backbone=8.714249e-05 elapsed=14:04:21 avg_step=3.73s recent_step=11.24s eta=37:42:10 +2026-09-15 04:36:38 | INFO | train rank=0 | step=13600/50000 progress=27.2% loss=0.103246 lr=8.712034e-05 lr_backbone=8.712034e-05 elapsed=14:06:16 avg_step=3.73s recent_step=11.52s eta=37:45:01 +2026-09-15 04:40:01 | INFO | train rank=0 | step=13610/50000 progress=27.2% loss=0.117139 lr=8.709818e-05 lr_backbone=8.709818e-05 elapsed=14:09:39 avg_step=3.75s recent_step=20.29s eta=37:51:46 +2026-09-15 04:40:38 | INFO | train rank=0 | step=13620/50000 progress=27.2% loss=0.096042 lr=8.707600e-05 lr_backbone=8.707600e-05 elapsed=14:10:16 avg_step=3.75s recent_step=3.76s eta=37:51:09 +2026-09-15 04:41:15 | INFO | train rank=0 | step=13630/50000 progress=27.3% loss=0.098071 lr=8.705380e-05 lr_backbone=8.705380e-05 elapsed=14:10:53 avg_step=3.75s recent_step=3.70s eta=37:50:30 +2026-09-15 04:42:06 | INFO | train rank=0 | step=13640/50000 progress=27.3% loss=0.089030 lr=8.703159e-05 lr_backbone=8.703159e-05 elapsed=14:11:44 avg_step=3.75s recent_step=5.08s eta=37:50:28 +2026-09-15 04:42:41 | INFO | train rank=0 | step=13650/50000 progress=27.3% loss=0.100365 lr=8.700936e-05 lr_backbone=8.700936e-05 elapsed=14:12:19 avg_step=3.75s recent_step=3.53s eta=37:49:45 +2026-09-15 04:43:17 | INFO | train rank=0 | step=13660/50000 progress=27.3% loss=0.104289 lr=8.698712e-05 lr_backbone=8.698712e-05 elapsed=14:12:55 avg_step=3.75s recent_step=3.56s eta=37:49:03 +2026-09-15 04:43:52 | INFO | train rank=0 | step=13670/50000 progress=27.3% loss=0.129710 lr=8.696486e-05 lr_backbone=8.696486e-05 elapsed=14:13:30 avg_step=3.75s recent_step=3.49s eta=37:48:19 +2026-09-15 04:44:47 | INFO | train rank=0 | step=13680/50000 progress=27.4% loss=0.098847 lr=8.694258e-05 lr_backbone=8.694258e-05 elapsed=14:14:25 avg_step=3.75s recent_step=5.51s eta=37:48:28 +2026-09-15 04:45:18 | INFO | train rank=0 | step=13690/50000 progress=27.4% loss=0.103038 lr=8.692029e-05 lr_backbone=8.692029e-05 elapsed=14:14:56 avg_step=3.75s recent_step=3.12s eta=37:47:34 +2026-09-15 04:45:57 | INFO | train rank=0 | step=13700/50000 progress=27.4% loss=0.109267 lr=8.689798e-05 lr_backbone=8.689798e-05 elapsed=14:15:35 avg_step=3.75s recent_step=3.83s eta=37:46:59 +2026-09-15 04:46:31 | INFO | train rank=0 | step=13710/50000 progress=27.4% loss=0.092358 lr=8.687566e-05 lr_backbone=8.687566e-05 elapsed=14:16:09 avg_step=3.75s recent_step=3.47s eta=37:46:14 +2026-09-15 04:47:18 | INFO | train rank=0 | step=13720/50000 progress=27.4% loss=0.100299 lr=8.685331e-05 lr_backbone=8.685331e-05 elapsed=14:16:56 avg_step=3.75s recent_step=4.68s eta=37:46:01 +2026-09-15 04:47:48 | INFO | train rank=0 | step=13730/50000 progress=27.5% loss=0.097313 lr=8.683096e-05 lr_backbone=8.683096e-05 elapsed=14:17:26 avg_step=3.75s recent_step=3.02s eta=37:45:04 +2026-09-15 04:48:22 | INFO | train rank=0 | step=13740/50000 progress=27.5% loss=0.105313 lr=8.680858e-05 lr_backbone=8.680858e-05 elapsed=14:18:00 avg_step=3.75s recent_step=3.40s eta=37:44:18 +2026-09-15 04:48:54 | INFO | train rank=0 | step=13750/50000 progress=27.5% loss=0.113673 lr=8.678620e-05 lr_backbone=8.678620e-05 elapsed=14:18:32 avg_step=3.75s recent_step=3.21s eta=37:43:26 +2026-09-15 04:49:51 | INFO | train rank=0 | step=13760/50000 progress=27.5% loss=0.099389 lr=8.676379e-05 lr_backbone=8.676379e-05 elapsed=14:19:29 avg_step=3.75s recent_step=5.64s eta=37:43:39 +2026-09-15 04:50:26 | INFO | train rank=0 | step=13770/50000 progress=27.5% loss=0.092623 lr=8.674137e-05 lr_backbone=8.674137e-05 elapsed=14:20:04 avg_step=3.75s recent_step=3.52s eta=37:42:55 +2026-09-15 04:50:59 | INFO | train rank=0 | step=13780/50000 progress=27.6% loss=0.106346 lr=8.671893e-05 lr_backbone=8.671893e-05 elapsed=14:20:37 avg_step=3.75s recent_step=3.34s eta=37:42:07 +2026-09-15 04:51:31 | INFO | train rank=0 | step=13790/50000 progress=27.6% loss=0.098071 lr=8.669648e-05 lr_backbone=8.669648e-05 elapsed=14:21:09 avg_step=3.75s recent_step=3.15s eta=37:41:14 +2026-09-15 04:52:13 | INFO | train rank=0 | step=13800/50000 progress=27.6% loss=0.089570 lr=8.667401e-05 lr_backbone=8.667401e-05 elapsed=14:21:51 avg_step=3.75s recent_step=4.23s eta=37:40:49 +2026-09-15 04:52:46 | INFO | train rank=0 | step=13810/50000 progress=27.6% loss=0.092564 lr=8.665152e-05 lr_backbone=8.665152e-05 elapsed=14:22:24 avg_step=3.75s recent_step=3.27s eta=37:39:59 +2026-09-15 04:53:22 | INFO | train rank=0 | step=13820/50000 progress=27.6% loss=0.107505 lr=8.662902e-05 lr_backbone=8.662902e-05 elapsed=14:23:00 avg_step=3.75s recent_step=3.62s eta=37:39:18 +2026-09-15 04:53:57 | INFO | train rank=0 | step=13830/50000 progress=27.7% loss=0.110706 lr=8.660650e-05 lr_backbone=8.660650e-05 elapsed=14:23:35 avg_step=3.75s recent_step=3.45s eta=37:38:33 +2026-09-15 04:54:33 | INFO | train rank=0 | step=13840/50000 progress=27.7% loss=0.125416 lr=8.658397e-05 lr_backbone=8.658397e-05 elapsed=14:24:11 avg_step=3.75s recent_step=3.68s eta=37:37:54 +2026-09-15 04:55:04 | INFO | train rank=0 | step=13850/50000 progress=27.7% loss=0.116480 lr=8.656142e-05 lr_backbone=8.656142e-05 elapsed=14:24:42 avg_step=3.75s recent_step=3.03s eta=37:36:58 +2026-09-15 04:55:35 | INFO | train rank=0 | step=13860/50000 progress=27.7% loss=0.134711 lr=8.653885e-05 lr_backbone=8.653885e-05 elapsed=14:25:13 avg_step=3.75s recent_step=3.13s eta=37:36:04 +2026-09-15 04:56:09 | INFO | train rank=0 | step=13870/50000 progress=27.7% loss=0.128936 lr=8.651627e-05 lr_backbone=8.651627e-05 elapsed=14:25:47 avg_step=3.75s recent_step=3.38s eta=37:35:17 +2026-09-15 04:56:57 | INFO | train rank=0 | step=13880/50000 progress=27.8% loss=0.107028 lr=8.649367e-05 lr_backbone=8.649367e-05 elapsed=14:26:35 avg_step=3.75s recent_step=4.85s eta=37:35:08 +2026-09-15 04:57:33 | INFO | train rank=0 | step=13890/50000 progress=27.8% loss=0.098933 lr=8.647106e-05 lr_backbone=8.647106e-05 elapsed=14:27:11 avg_step=3.75s recent_step=3.59s eta=37:34:27 +2026-09-15 04:58:03 | INFO | train rank=0 | step=13900/50000 progress=27.8% loss=0.081321 lr=8.644843e-05 lr_backbone=8.644843e-05 elapsed=14:27:41 avg_step=3.75s recent_step=3.00s eta=37:33:30 +2026-09-15 04:58:40 | INFO | train rank=0 | step=13910/50000 progress=27.8% loss=0.107638 lr=8.642579e-05 lr_backbone=8.642579e-05 elapsed=14:28:18 avg_step=3.75s recent_step=3.66s eta=37:32:50 +2026-09-15 04:59:28 | INFO | train rank=0 | step=13920/50000 progress=27.8% loss=0.123539 lr=8.640312e-05 lr_backbone=8.640312e-05 elapsed=14:29:06 avg_step=3.75s recent_step=4.86s eta=37:32:42 +2026-09-15 05:00:02 | INFO | train rank=0 | step=13930/50000 progress=27.9% loss=0.136290 lr=8.638045e-05 lr_backbone=8.638045e-05 elapsed=14:29:40 avg_step=3.75s recent_step=3.36s eta=37:31:54 +2026-09-15 05:00:46 | INFO | train rank=0 | step=13940/50000 progress=27.9% loss=0.140274 lr=8.635775e-05 lr_backbone=8.635775e-05 elapsed=14:30:25 avg_step=3.75s recent_step=4.46s eta=37:31:35 +2026-09-15 05:01:14 | INFO | train rank=0 | step=13950/50000 progress=27.9% loss=0.116440 lr=8.633504e-05 lr_backbone=8.633504e-05 elapsed=14:30:52 avg_step=3.75s recent_step=2.79s eta=37:30:33 +2026-09-15 05:01:57 | INFO | train rank=0 | step=13960/50000 progress=27.9% loss=0.121160 lr=8.631232e-05 lr_backbone=8.631232e-05 elapsed=14:31:35 avg_step=3.75s recent_step=4.24s eta=37:30:08 +2026-09-15 05:02:36 | INFO | train rank=0 | step=13970/50000 progress=27.9% loss=0.098840 lr=8.628958e-05 lr_backbone=8.628958e-05 elapsed=14:32:14 avg_step=3.75s recent_step=3.89s eta=37:29:35 +2026-09-15 05:03:18 | INFO | train rank=0 | step=13980/50000 progress=28.0% loss=0.096261 lr=8.626682e-05 lr_backbone=8.626682e-05 elapsed=14:32:56 avg_step=3.75s recent_step=4.24s eta=37:29:10 +2026-09-15 05:03:57 | INFO | train rank=0 | step=13990/50000 progress=28.0% loss=0.102326 lr=8.624405e-05 lr_backbone=8.624405e-05 elapsed=14:33:35 avg_step=3.75s recent_step=3.86s eta=37:28:35 +2026-09-15 05:04:35 | INFO | train rank=0 | step=14000/50000 progress=28.0% loss=0.093801 lr=8.622126e-05 lr_backbone=8.622126e-05 elapsed=14:34:13 avg_step=3.75s recent_step=3.85s eta=37:28:01 +2026-09-15 05:05:19 | INFO | train rank=0 | step=14010/50000 progress=28.0% loss=0.130720 lr=8.619846e-05 lr_backbone=8.619846e-05 elapsed=14:34:57 avg_step=3.75s recent_step=4.41s eta=37:27:40 +2026-09-15 05:05:50 | INFO | train rank=0 | step=14020/50000 progress=28.0% loss=0.111672 lr=8.617564e-05 lr_backbone=8.617564e-05 elapsed=14:35:28 avg_step=3.75s recent_step=3.08s eta=37:26:45 +2026-09-15 05:06:27 | INFO | train rank=0 | step=14030/50000 progress=28.1% loss=0.094192 lr=8.615280e-05 lr_backbone=8.615280e-05 elapsed=14:36:05 avg_step=3.75s recent_step=3.71s eta=37:26:07 +2026-09-15 05:07:17 | INFO | train rank=0 | step=14040/50000 progress=28.1% loss=0.094887 lr=8.612995e-05 lr_backbone=8.612995e-05 elapsed=14:36:55 avg_step=3.75s recent_step=5.00s eta=37:26:02 +2026-09-15 05:07:45 | INFO | train rank=0 | step=14050/50000 progress=28.1% loss=0.116067 lr=8.610708e-05 lr_backbone=8.610708e-05 elapsed=14:37:23 avg_step=3.75s recent_step=2.82s eta=37:25:01 +2026-09-15 05:08:19 | INFO | train rank=0 | step=14060/50000 progress=28.1% loss=0.093126 lr=8.608420e-05 lr_backbone=8.608420e-05 elapsed=14:37:57 avg_step=3.75s recent_step=3.41s eta=37:24:14 +2026-09-15 05:09:02 | INFO | train rank=0 | step=14070/50000 progress=28.1% loss=0.135403 lr=8.606130e-05 lr_backbone=8.606130e-05 elapsed=14:38:40 avg_step=3.75s recent_step=4.25s eta=37:23:50 +2026-09-15 05:09:44 | INFO | train rank=0 | step=14080/50000 progress=28.2% loss=0.099580 lr=8.603838e-05 lr_backbone=8.603838e-05 elapsed=14:39:22 avg_step=3.75s recent_step=4.24s eta=37:23:25 +2026-09-15 05:10:20 | INFO | train rank=0 | step=14090/50000 progress=28.2% loss=0.098588 lr=8.601545e-05 lr_backbone=8.601545e-05 elapsed=14:39:58 avg_step=3.75s recent_step=3.54s eta=37:22:42 +2026-09-15 05:10:58 | INFO | train rank=0 | step=14100/50000 progress=28.2% loss=0.086498 lr=8.599250e-05 lr_backbone=8.599250e-05 elapsed=14:40:36 avg_step=3.75s recent_step=3.83s eta=37:22:07 +2026-09-15 05:11:29 | INFO | train rank=0 | step=14110/50000 progress=28.2% loss=0.103049 lr=8.596954e-05 lr_backbone=8.596954e-05 elapsed=14:41:07 avg_step=3.75s recent_step=3.11s eta=37:21:13 +2026-09-15 05:12:22 | INFO | train rank=0 | step=14120/50000 progress=28.2% loss=0.114521 lr=8.594656e-05 lr_backbone=8.594656e-05 elapsed=14:42:00 avg_step=3.75s recent_step=5.24s eta=37:21:13 +2026-09-15 05:12:47 | INFO | train rank=0 | step=14130/50000 progress=28.3% loss=0.110348 lr=8.592357e-05 lr_backbone=8.592357e-05 elapsed=14:42:25 avg_step=3.75s recent_step=2.52s eta=37:20:05 +2026-09-15 05:13:16 | INFO | train rank=0 | step=14140/50000 progress=28.3% loss=0.102622 lr=8.590056e-05 lr_backbone=8.590056e-05 elapsed=14:42:54 avg_step=3.75s recent_step=2.89s eta=37:19:06 +2026-09-15 05:13:56 | INFO | train rank=0 | step=14150/50000 progress=28.3% loss=0.115946 lr=8.587754e-05 lr_backbone=8.587754e-05 elapsed=14:43:35 avg_step=3.75s recent_step=4.08s eta=37:18:37 +2026-09-15 05:14:52 | INFO | train rank=0 | step=14160/50000 progress=28.3% loss=0.087464 lr=8.585449e-05 lr_backbone=8.585449e-05 elapsed=14:44:30 avg_step=3.75s recent_step=5.58s eta=37:18:46 +2026-09-15 05:15:30 | INFO | train rank=0 | step=14170/50000 progress=28.3% loss=0.092449 lr=8.583144e-05 lr_backbone=8.583144e-05 elapsed=14:45:08 avg_step=3.75s recent_step=3.76s eta=37:18:08 +2026-09-15 05:15:49 | INFO | train rank=0 | step=14180/50000 progress=28.4% loss=0.120577 lr=8.580837e-05 lr_backbone=8.580837e-05 elapsed=14:45:27 avg_step=3.75s recent_step=1.96s eta=37:16:46 +2026-09-15 05:16:20 | INFO | train rank=0 | step=14190/50000 progress=28.4% loss=0.088863 lr=8.578528e-05 lr_backbone=8.578528e-05 elapsed=14:45:58 avg_step=3.75s recent_step=3.08s eta=37:15:51 +2026-09-15 05:17:11 | INFO | train rank=0 | step=14200/50000 progress=28.4% loss=0.111121 lr=8.576217e-05 lr_backbone=8.576217e-05 elapsed=14:46:49 avg_step=3.75s recent_step=5.11s eta=37:15:48 +2026-09-15 05:17:39 | INFO | train rank=0 | step=14210/50000 progress=28.4% loss=0.101556 lr=8.573906e-05 lr_backbone=8.573906e-05 elapsed=14:47:17 avg_step=3.75s recent_step=2.80s eta=37:14:47 +2026-09-15 05:18:09 | INFO | train rank=0 | step=14220/50000 progress=28.4% loss=0.092348 lr=8.571592e-05 lr_backbone=8.571592e-05 elapsed=14:47:47 avg_step=3.75s recent_step=2.98s eta=37:13:50 +2026-09-15 05:18:35 | INFO | train rank=0 | step=14230/50000 progress=28.5% loss=0.113731 lr=8.569277e-05 lr_backbone=8.569277e-05 elapsed=14:48:13 avg_step=3.75s recent_step=2.54s eta=37:12:42 +2026-09-15 05:19:21 | INFO | train rank=0 | step=14240/50000 progress=28.5% loss=0.104320 lr=8.566960e-05 lr_backbone=8.566960e-05 elapsed=14:48:59 avg_step=3.75s recent_step=4.63s eta=37:12:27 +2026-09-15 05:19:55 | INFO | train rank=0 | step=14250/50000 progress=28.5% loss=0.091833 lr=8.564642e-05 lr_backbone=8.564642e-05 elapsed=14:49:33 avg_step=3.75s recent_step=3.43s eta=37:11:42 +2026-09-15 05:20:34 | INFO | train rank=0 | step=14260/50000 progress=28.5% loss=0.083506 lr=8.562323e-05 lr_backbone=8.562323e-05 elapsed=14:50:12 avg_step=3.75s recent_step=3.91s eta=37:11:09 +2026-09-15 05:21:03 | INFO | train rank=0 | step=14270/50000 progress=28.5% loss=0.093673 lr=8.560001e-05 lr_backbone=8.560001e-05 elapsed=14:50:41 avg_step=3.75s recent_step=2.84s eta=37:10:08 +2026-09-15 05:21:47 | INFO | train rank=0 | step=14280/50000 progress=28.6% loss=0.110584 lr=8.557678e-05 lr_backbone=8.557678e-05 elapsed=14:51:25 avg_step=3.75s recent_step=4.46s eta=37:09:49 +2026-09-15 05:22:17 | INFO | train rank=0 | step=14290/50000 progress=28.6% loss=0.114728 lr=8.555354e-05 lr_backbone=8.555354e-05 elapsed=14:51:55 avg_step=3.74s recent_step=2.95s eta=37:08:52 +2026-09-15 05:22:45 | INFO | train rank=0 | step=14300/50000 progress=28.6% loss=0.101095 lr=8.553028e-05 lr_backbone=8.553028e-05 elapsed=14:52:23 avg_step=3.74s recent_step=2.81s eta=37:07:51 +2026-09-15 05:23:27 | INFO | train rank=0 | step=14310/50000 progress=28.6% loss=0.091921 lr=8.550701e-05 lr_backbone=8.550701e-05 elapsed=14:53:05 avg_step=3.74s recent_step=4.21s eta=37:07:25 +2026-09-15 05:24:09 | INFO | train rank=0 | step=14320/50000 progress=28.6% loss=0.107631 lr=8.548371e-05 lr_backbone=8.548371e-05 elapsed=14:53:47 avg_step=3.74s recent_step=4.22s eta=37:07:00 +2026-09-15 05:24:43 | INFO | train rank=0 | step=14330/50000 progress=28.7% loss=0.107376 lr=8.546041e-05 lr_backbone=8.546041e-05 elapsed=14:54:21 avg_step=3.74s recent_step=3.36s eta=37:06:13 +2026-09-15 05:25:20 | INFO | train rank=0 | step=14340/50000 progress=28.7% loss=0.105495 lr=8.543709e-05 lr_backbone=8.543709e-05 elapsed=14:54:58 avg_step=3.74s recent_step=3.72s eta=37:05:34 +2026-09-15 05:26:06 | INFO | train rank=0 | step=14350/50000 progress=28.7% loss=0.112094 lr=8.541375e-05 lr_backbone=8.541375e-05 elapsed=14:55:44 avg_step=3.75s recent_step=4.58s eta=37:05:18 +2026-09-15 05:27:00 | INFO | train rank=0 | step=14360/50000 progress=28.7% loss=0.092385 lr=8.539040e-05 lr_backbone=8.539040e-05 elapsed=14:56:38 avg_step=3.75s recent_step=5.44s eta=37:05:22 +2026-09-15 05:27:37 | INFO | train rank=0 | step=14370/50000 progress=28.7% loss=0.125356 lr=8.536703e-05 lr_backbone=8.536703e-05 elapsed=14:57:15 avg_step=3.75s recent_step=3.64s eta=37:04:42 +2026-09-15 05:28:13 | INFO | train rank=0 | step=14380/50000 progress=28.8% loss=0.120251 lr=8.534365e-05 lr_backbone=8.534365e-05 elapsed=14:57:51 avg_step=3.75s recent_step=3.67s eta=37:04:03 +2026-09-15 05:28:47 | INFO | train rank=0 | step=14390/50000 progress=28.8% loss=0.137407 lr=8.532025e-05 lr_backbone=8.532025e-05 elapsed=14:58:25 avg_step=3.75s recent_step=3.36s eta=37:03:16 +2026-09-15 05:29:36 | INFO | train rank=0 | step=14400/50000 progress=28.8% loss=0.113709 lr=8.529683e-05 lr_backbone=8.529683e-05 elapsed=14:59:14 avg_step=3.75s recent_step=4.89s eta=37:03:07 +2026-09-15 05:30:07 | INFO | train rank=0 | step=14410/50000 progress=28.8% loss=0.131584 lr=8.527340e-05 lr_backbone=8.527340e-05 elapsed=14:59:45 avg_step=3.75s recent_step=3.15s eta=37:02:15 +2026-09-15 05:30:38 | INFO | train rank=0 | step=14420/50000 progress=28.8% loss=0.108319 lr=8.524996e-05 lr_backbone=8.524996e-05 elapsed=15:00:16 avg_step=3.75s recent_step=3.09s eta=37:01:21 +2026-09-15 05:31:10 | INFO | train rank=0 | step=14430/50000 progress=28.9% loss=0.111627 lr=8.522650e-05 lr_backbone=8.522650e-05 elapsed=15:00:48 avg_step=3.75s recent_step=3.13s eta=37:00:28 +2026-09-15 05:31:58 | INFO | train rank=0 | step=14440/50000 progress=28.9% loss=0.094946 lr=8.520302e-05 lr_backbone=8.520302e-05 elapsed=15:01:36 avg_step=3.75s recent_step=4.79s eta=37:00:17 +2026-09-15 05:32:32 | INFO | train rank=0 | step=14450/50000 progress=28.9% loss=0.101253 lr=8.517953e-05 lr_backbone=8.517953e-05 elapsed=15:02:10 avg_step=3.75s recent_step=3.41s eta=36:59:31 +2026-09-15 05:33:01 | INFO | train rank=0 | step=14460/50000 progress=28.9% loss=0.110623 lr=8.515602e-05 lr_backbone=8.515602e-05 elapsed=15:02:39 avg_step=3.75s recent_step=2.97s eta=36:58:35 +2026-09-15 05:33:34 | INFO | train rank=0 | step=14470/50000 progress=28.9% loss=0.084964 lr=8.513250e-05 lr_backbone=8.513250e-05 elapsed=15:03:12 avg_step=3.75s recent_step=3.30s eta=36:57:46 +2026-09-15 05:34:15 | INFO | train rank=0 | step=14480/50000 progress=29.0% loss=0.082307 lr=8.510896e-05 lr_backbone=8.510896e-05 elapsed=15:03:53 avg_step=3.75s recent_step=4.06s eta=36:57:17 +2026-09-15 05:34:50 | INFO | train rank=0 | step=14490/50000 progress=29.0% loss=0.119223 lr=8.508541e-05 lr_backbone=8.508541e-05 elapsed=15:04:28 avg_step=3.75s recent_step=3.45s eta=36:56:32 +2026-09-15 05:35:37 | INFO | train rank=0 | step=14500/50000 progress=29.0% loss=0.109723 lr=8.506184e-05 lr_backbone=8.506184e-05 elapsed=15:05:15 avg_step=3.75s recent_step=4.75s eta=36:56:19 +2026-09-15 05:36:25 | INFO | train rank=0 | step=14510/50000 progress=29.0% loss=0.131589 lr=8.503826e-05 lr_backbone=8.503826e-05 elapsed=15:06:03 avg_step=3.75s recent_step=4.80s eta=36:56:07 +2026-09-15 05:36:54 | INFO | train rank=0 | step=14520/50000 progress=29.0% loss=0.096830 lr=8.501466e-05 lr_backbone=8.501466e-05 elapsed=15:06:32 avg_step=3.75s recent_step=2.92s eta=36:55:10 +2026-09-15 05:37:33 | INFO | train rank=0 | step=14530/50000 progress=29.1% loss=0.078848 lr=8.499104e-05 lr_backbone=8.499104e-05 elapsed=15:07:11 avg_step=3.75s recent_step=3.90s eta=36:54:36 +2026-09-15 05:38:19 | INFO | train rank=0 | step=14540/50000 progress=29.1% loss=0.113126 lr=8.496741e-05 lr_backbone=8.496741e-05 elapsed=15:07:57 avg_step=3.75s recent_step=4.57s eta=36:54:19 +2026-09-15 05:38:56 | INFO | train rank=0 | step=14550/50000 progress=29.1% loss=0.093958 lr=8.494377e-05 lr_backbone=8.494377e-05 elapsed=15:08:34 avg_step=3.75s recent_step=3.66s eta=36:53:39 +2026-09-15 05:39:33 | INFO | train rank=0 | step=14560/50000 progress=29.1% loss=0.104916 lr=8.492011e-05 lr_backbone=8.492011e-05 elapsed=15:09:11 avg_step=3.75s recent_step=3.75s eta=36:53:02 +2026-09-15 05:40:11 | INFO | train rank=0 | step=14570/50000 progress=29.1% loss=0.072015 lr=8.489643e-05 lr_backbone=8.489643e-05 elapsed=15:09:49 avg_step=3.75s recent_step=3.77s eta=36:52:25 +2026-09-15 05:40:45 | INFO | train rank=0 | step=14580/50000 progress=29.2% loss=0.104648 lr=8.487274e-05 lr_backbone=8.487274e-05 elapsed=15:10:23 avg_step=3.75s recent_step=3.38s eta=36:51:38 +2026-09-15 05:41:26 | INFO | train rank=0 | step=14590/50000 progress=29.2% loss=0.105262 lr=8.484903e-05 lr_backbone=8.484903e-05 elapsed=15:11:04 avg_step=3.75s recent_step=4.10s eta=36:51:09 +2026-09-15 05:42:03 | INFO | train rank=0 | step=14600/50000 progress=29.2% loss=0.110858 lr=8.482531e-05 lr_backbone=8.482531e-05 elapsed=15:11:41 avg_step=3.75s recent_step=3.72s eta=36:50:31 +2026-09-15 05:42:41 | INFO | train rank=0 | step=14610/50000 progress=29.2% loss=0.107887 lr=8.480158e-05 lr_backbone=8.480158e-05 elapsed=15:12:19 avg_step=3.75s recent_step=3.79s eta=36:49:55 +2026-09-15 05:43:33 | INFO | train rank=0 | step=14620/50000 progress=29.2% loss=0.102762 lr=8.477782e-05 lr_backbone=8.477782e-05 elapsed=15:13:11 avg_step=3.75s recent_step=5.26s eta=36:49:54 +2026-09-15 05:44:10 | INFO | train rank=0 | step=14630/50000 progress=29.3% loss=0.092975 lr=8.475406e-05 lr_backbone=8.475406e-05 elapsed=15:13:48 avg_step=3.75s recent_step=3.64s eta=36:49:14 +2026-09-15 05:44:42 | INFO | train rank=0 | step=14640/50000 progress=29.3% loss=0.100517 lr=8.473028e-05 lr_backbone=8.473028e-05 elapsed=15:14:20 avg_step=3.75s recent_step=3.27s eta=36:48:25 +2026-09-15 05:45:16 | INFO | train rank=0 | step=14650/50000 progress=29.3% loss=0.100462 lr=8.470648e-05 lr_backbone=8.470648e-05 elapsed=15:14:54 avg_step=3.75s recent_step=3.35s eta=36:47:38 +2026-09-15 05:46:07 | INFO | train rank=0 | step=14660/50000 progress=29.3% loss=0.108719 lr=8.468267e-05 lr_backbone=8.468267e-05 elapsed=15:15:45 avg_step=3.75s recent_step=5.08s eta=36:47:33 +2026-09-15 05:46:42 | INFO | train rank=0 | step=14670/50000 progress=29.3% loss=0.105987 lr=8.465884e-05 lr_backbone=8.465884e-05 elapsed=15:16:20 avg_step=3.75s recent_step=3.56s eta=36:46:50 +2026-09-15 05:47:12 | INFO | train rank=0 | step=14680/50000 progress=29.4% loss=0.089202 lr=8.463499e-05 lr_backbone=8.463499e-05 elapsed=15:16:50 avg_step=3.75s recent_step=3.02s eta=36:45:55 +2026-09-15 05:47:47 | INFO | train rank=0 | step=14690/50000 progress=29.4% loss=0.098897 lr=8.461114e-05 lr_backbone=8.461114e-05 elapsed=15:17:25 avg_step=3.75s recent_step=3.47s eta=36:45:11 +2026-09-15 05:48:36 | INFO | train rank=0 | step=14700/50000 progress=29.4% loss=0.111439 lr=8.458726e-05 lr_backbone=8.458726e-05 elapsed=15:18:14 avg_step=3.75s recent_step=4.85s eta=36:45:00 +2026-09-15 05:48:57 | INFO | train rank=0 | step=14710/50000 progress=29.4% loss=0.086551 lr=8.456337e-05 lr_backbone=8.456337e-05 elapsed=15:18:35 avg_step=3.75s recent_step=2.13s eta=36:43:44 +2026-09-15 05:49:31 | INFO | train rank=0 | step=14720/50000 progress=29.4% loss=0.097418 lr=8.453947e-05 lr_backbone=8.453947e-05 elapsed=15:19:09 avg_step=3.75s recent_step=3.45s eta=36:42:59 +2026-09-15 05:50:09 | INFO | train rank=0 | step=14730/50000 progress=29.5% loss=0.090717 lr=8.451555e-05 lr_backbone=8.451555e-05 elapsed=15:19:47 avg_step=3.75s recent_step=3.81s eta=36:42:23 +2026-09-15 05:50:56 | INFO | train rank=0 | step=14740/50000 progress=29.5% loss=0.105113 lr=8.449162e-05 lr_backbone=8.449162e-05 elapsed=15:20:34 avg_step=3.75s recent_step=4.62s eta=36:42:07 +2026-09-15 05:51:24 | INFO | train rank=0 | step=14750/50000 progress=29.5% loss=0.100414 lr=8.446767e-05 lr_backbone=8.446767e-05 elapsed=15:21:02 avg_step=3.75s recent_step=2.82s eta=36:41:07 +2026-09-15 05:51:57 | INFO | train rank=0 | step=14760/50000 progress=29.5% loss=0.136592 lr=8.444371e-05 lr_backbone=8.444371e-05 elapsed=15:21:35 avg_step=3.75s recent_step=3.27s eta=36:40:18 +2026-09-15 05:52:30 | INFO | train rank=0 | step=14770/50000 progress=29.5% loss=0.127242 lr=8.441973e-05 lr_backbone=8.441973e-05 elapsed=15:22:08 avg_step=3.75s recent_step=3.32s eta=36:39:31 +2026-09-15 05:53:24 | INFO | train rank=0 | step=14780/50000 progress=29.6% loss=0.117817 lr=8.439573e-05 lr_backbone=8.439573e-05 elapsed=15:23:02 avg_step=3.75s recent_step=5.41s eta=36:39:33 +2026-09-15 05:53:47 | INFO | train rank=0 | step=14790/50000 progress=29.6% loss=0.123193 lr=8.437172e-05 lr_backbone=8.437172e-05 elapsed=15:23:25 avg_step=3.75s recent_step=2.32s eta=36:38:21 +2026-09-15 05:54:20 | INFO | train rank=0 | step=14800/50000 progress=29.6% loss=0.101209 lr=8.434770e-05 lr_backbone=8.434770e-05 elapsed=15:23:58 avg_step=3.75s recent_step=3.25s eta=36:37:32 +2026-09-15 05:54:57 | INFO | train rank=0 | step=14810/50000 progress=29.6% loss=0.100954 lr=8.432366e-05 lr_backbone=8.432366e-05 elapsed=15:24:35 avg_step=3.75s recent_step=3.70s eta=36:36:54 +2026-09-15 05:55:42 | INFO | train rank=0 | step=14820/50000 progress=29.6% loss=0.095182 lr=8.429961e-05 lr_backbone=8.429961e-05 elapsed=15:25:20 avg_step=3.75s recent_step=4.55s eta=36:36:35 +2026-09-15 05:56:10 | INFO | train rank=0 | step=14830/50000 progress=29.7% loss=0.085853 lr=8.427554e-05 lr_backbone=8.427554e-05 elapsed=15:25:48 avg_step=3.75s recent_step=2.74s eta=36:35:34 +2026-09-15 05:56:41 | INFO | train rank=0 | step=14840/50000 progress=29.7% loss=0.142889 lr=8.425145e-05 lr_backbone=8.425145e-05 elapsed=15:26:19 avg_step=3.75s recent_step=3.11s eta=36:34:42 +2026-09-15 05:57:16 | INFO | train rank=0 | step=14850/50000 progress=29.7% loss=0.100442 lr=8.422736e-05 lr_backbone=8.422736e-05 elapsed=15:26:54 avg_step=3.75s recent_step=3.50s eta=36:33:58 +2026-09-15 05:58:04 | INFO | train rank=0 | step=14860/50000 progress=29.7% loss=0.085722 lr=8.420324e-05 lr_backbone=8.420324e-05 elapsed=15:27:42 avg_step=3.75s recent_step=4.84s eta=36:33:47 +2026-09-15 05:58:35 | INFO | train rank=0 | step=14870/50000 progress=29.7% loss=0.079757 lr=8.417911e-05 lr_backbone=8.417911e-05 elapsed=15:28:13 avg_step=3.75s recent_step=3.09s eta=36:32:54 +2026-09-15 05:59:16 | INFO | train rank=0 | step=14880/50000 progress=29.8% loss=0.102976 lr=8.415497e-05 lr_backbone=8.415497e-05 elapsed=15:28:54 avg_step=3.75s recent_step=4.09s eta=36:32:25 +2026-09-15 05:59:50 | INFO | train rank=0 | step=14890/50000 progress=29.8% loss=0.130766 lr=8.413081e-05 lr_backbone=8.413081e-05 elapsed=15:29:28 avg_step=3.75s recent_step=3.46s eta=36:31:40 +2026-09-15 06:00:37 | INFO | train rank=0 | step=14900/50000 progress=29.8% loss=0.087701 lr=8.410664e-05 lr_backbone=8.410664e-05 elapsed=15:30:15 avg_step=3.75s recent_step=4.69s eta=36:31:25 +2026-09-15 06:01:06 | INFO | train rank=0 | step=14910/50000 progress=29.8% loss=0.093331 lr=8.408245e-05 lr_backbone=8.408245e-05 elapsed=15:30:44 avg_step=3.75s recent_step=2.81s eta=36:30:26 +2026-09-15 06:01:49 | INFO | train rank=0 | step=14920/50000 progress=29.8% loss=0.110263 lr=8.405824e-05 lr_backbone=8.405824e-05 elapsed=15:31:27 avg_step=3.75s recent_step=4.37s eta=36:30:03 +2026-09-15 06:02:13 | INFO | train rank=0 | step=14930/50000 progress=29.9% loss=0.094950 lr=8.403402e-05 lr_backbone=8.403402e-05 elapsed=15:31:51 avg_step=3.74s recent_step=2.41s eta=36:28:54 +2026-09-15 06:03:01 | INFO | train rank=0 | step=14940/50000 progress=29.9% loss=0.098302 lr=8.400979e-05 lr_backbone=8.400979e-05 elapsed=15:32:39 avg_step=3.75s recent_step=4.82s eta=36:28:42 +2026-09-15 06:03:30 | INFO | train rank=0 | step=14950/50000 progress=29.9% loss=0.092438 lr=8.398554e-05 lr_backbone=8.398554e-05 elapsed=15:33:08 avg_step=3.75s recent_step=2.86s eta=36:27:44 +2026-09-15 06:04:03 | INFO | train rank=0 | step=14960/50000 progress=29.9% loss=0.095478 lr=8.396128e-05 lr_backbone=8.396128e-05 elapsed=15:33:41 avg_step=3.74s recent_step=3.28s eta=36:26:55 +2026-09-15 06:04:33 | INFO | train rank=0 | step=14970/50000 progress=29.9% loss=0.108629 lr=8.393700e-05 lr_backbone=8.393700e-05 elapsed=15:34:11 avg_step=3.74s recent_step=3.00s eta=36:26:01 +2026-09-15 06:05:17 | INFO | train rank=0 | step=14980/50000 progress=30.0% loss=0.094919 lr=8.391271e-05 lr_backbone=8.391271e-05 elapsed=15:34:55 avg_step=3.74s recent_step=4.45s eta=36:25:40 +2026-09-15 06:05:52 | INFO | train rank=0 | step=14990/50000 progress=30.0% loss=0.085684 lr=8.388840e-05 lr_backbone=8.388840e-05 elapsed=15:35:30 avg_step=3.74s recent_step=3.46s eta=36:24:56 +2026-09-15 06:06:36 | INFO | train rank=0 | step=15000/50000 progress=30.0% loss=0.100123 lr=8.386408e-05 lr_backbone=8.386408e-05 elapsed=15:36:14 avg_step=3.74s recent_step=4.40s eta=36:24:34 +2026-09-15 06:07:14 | INFO | train rank=0 | step=15010/50000 progress=30.0% loss=0.100569 lr=8.383974e-05 lr_backbone=8.383974e-05 elapsed=15:36:52 avg_step=3.74s recent_step=3.78s eta=36:23:57 +2026-09-15 06:07:58 | INFO | train rank=0 | step=15020/50000 progress=30.0% loss=0.094022 lr=8.381539e-05 lr_backbone=8.381539e-05 elapsed=15:37:36 avg_step=3.75s recent_step=4.44s eta=36:23:36 +2026-09-15 06:08:30 | INFO | train rank=0 | step=15030/50000 progress=30.1% loss=0.091788 lr=8.379102e-05 lr_backbone=8.379102e-05 elapsed=15:38:08 avg_step=3.75s recent_step=3.19s eta=36:22:45 +2026-09-15 06:09:12 | INFO | train rank=0 | step=15040/50000 progress=30.1% loss=0.118079 lr=8.376664e-05 lr_backbone=8.376664e-05 elapsed=15:38:50 avg_step=3.75s recent_step=4.22s eta=36:22:19 +2026-09-15 06:09:46 | INFO | train rank=0 | step=15050/50000 progress=30.1% loss=0.108995 lr=8.374224e-05 lr_backbone=8.374224e-05 elapsed=15:39:24 avg_step=3.75s recent_step=3.36s eta=36:21:33 +2026-09-15 06:10:23 | INFO | train rank=0 | step=15060/50000 progress=30.1% loss=0.081167 lr=8.371783e-05 lr_backbone=8.371783e-05 elapsed=15:40:01 avg_step=3.75s recent_step=3.68s eta=36:20:53 +2026-09-15 06:10:59 | INFO | train rank=0 | step=15070/50000 progress=30.1% loss=0.094357 lr=8.369341e-05 lr_backbone=8.369341e-05 elapsed=15:40:37 avg_step=3.75s recent_step=3.62s eta=36:20:13 +2026-09-15 06:11:36 | INFO | train rank=0 | step=15080/50000 progress=30.2% loss=0.129708 lr=8.366897e-05 lr_backbone=8.366897e-05 elapsed=15:41:14 avg_step=3.74s recent_step=3.70s eta=36:19:35 +2026-09-15 06:12:08 | INFO | train rank=0 | step=15090/50000 progress=30.2% loss=0.118047 lr=8.364451e-05 lr_backbone=8.364451e-05 elapsed=15:41:46 avg_step=3.74s recent_step=3.23s eta=36:18:45 +2026-09-15 06:12:47 | INFO | train rank=0 | step=15100/50000 progress=30.2% loss=0.102010 lr=8.362004e-05 lr_backbone=8.362004e-05 elapsed=15:42:25 avg_step=3.74s recent_step=3.92s eta=36:18:12 +2026-09-15 06:13:08 | INFO | train rank=0 | step=15110/50000 progress=30.2% loss=0.094938 lr=8.359556e-05 lr_backbone=8.359556e-05 elapsed=15:42:46 avg_step=3.74s recent_step=2.05s eta=36:16:55 +2026-09-15 06:14:03 | INFO | train rank=0 | step=15120/50000 progress=30.2% loss=0.075979 lr=8.357106e-05 lr_backbone=8.357106e-05 elapsed=15:43:41 avg_step=3.74s recent_step=5.46s eta=36:16:57 +2026-09-15 06:14:38 | INFO | train rank=0 | step=15130/50000 progress=30.3% loss=0.112211 lr=8.354654e-05 lr_backbone=8.354654e-05 elapsed=15:44:16 avg_step=3.74s recent_step=3.49s eta=36:16:14 +2026-09-15 06:15:10 | INFO | train rank=0 | step=15140/50000 progress=30.3% loss=0.113233 lr=8.352201e-05 lr_backbone=8.352201e-05 elapsed=15:44:49 avg_step=3.74s recent_step=3.30s eta=36:15:27 +2026-09-15 06:15:51 | INFO | train rank=0 | step=15150/50000 progress=30.3% loss=0.108222 lr=8.349747e-05 lr_backbone=8.349747e-05 elapsed=15:45:29 avg_step=3.74s recent_step=4.01s eta=36:14:55 +2026-09-15 06:16:39 | INFO | train rank=0 | step=15160/50000 progress=30.3% loss=0.107823 lr=8.347291e-05 lr_backbone=8.347291e-05 elapsed=15:46:17 avg_step=3.75s recent_step=4.82s eta=36:14:43 +2026-09-15 06:17:12 | INFO | train rank=0 | step=15170/50000 progress=30.3% loss=0.104727 lr=8.344834e-05 lr_backbone=8.344834e-05 elapsed=15:46:50 avg_step=3.74s recent_step=3.35s eta=36:13:56 +2026-09-15 06:17:49 | INFO | train rank=0 | step=15180/50000 progress=30.4% loss=0.113173 lr=8.342375e-05 lr_backbone=8.342375e-05 elapsed=15:47:27 avg_step=3.74s recent_step=3.64s eta=36:13:16 +2026-09-15 06:18:27 | INFO | train rank=0 | step=15190/50000 progress=30.4% loss=0.120103 lr=8.339915e-05 lr_backbone=8.339915e-05 elapsed=15:48:05 avg_step=3.74s recent_step=3.86s eta=36:12:42 +2026-09-15 06:19:02 | INFO | train rank=0 | step=15200/50000 progress=30.4% loss=0.075047 lr=8.337453e-05 lr_backbone=8.337453e-05 elapsed=15:48:40 avg_step=3.74s recent_step=3.51s eta=36:11:59 +2026-09-15 06:19:53 | INFO | train rank=0 | step=15210/50000 progress=30.4% loss=0.092554 lr=8.334990e-05 lr_backbone=8.334990e-05 elapsed=15:49:31 avg_step=3.75s recent_step=5.09s eta=36:11:52 +2026-09-15 06:20:26 | INFO | train rank=0 | step=15220/50000 progress=30.4% loss=0.104288 lr=8.332525e-05 lr_backbone=8.332525e-05 elapsed=15:50:04 avg_step=3.75s recent_step=3.24s eta=36:11:03 +2026-09-15 06:20:54 | INFO | train rank=0 | step=15230/50000 progress=30.5% loss=0.105196 lr=8.330059e-05 lr_backbone=8.330059e-05 elapsed=15:50:32 avg_step=3.74s recent_step=2.84s eta=36:10:05 +2026-09-15 06:21:33 | INFO | train rank=0 | step=15240/50000 progress=30.5% loss=0.105194 lr=8.327592e-05 lr_backbone=8.327592e-05 elapsed=15:51:11 avg_step=3.74s recent_step=3.88s eta=36:09:30 +2026-09-15 06:22:20 | INFO | train rank=0 | step=15250/50000 progress=30.5% loss=0.096717 lr=8.325123e-05 lr_backbone=8.325123e-05 elapsed=15:51:58 avg_step=3.75s recent_step=4.66s eta=36:09:14 +2026-09-15 06:22:58 | INFO | train rank=0 | step=15260/50000 progress=30.5% loss=0.117938 lr=8.322652e-05 lr_backbone=8.322652e-05 elapsed=15:52:36 avg_step=3.75s recent_step=3.85s eta=36:08:39 +2026-09-15 06:23:33 | INFO | train rank=0 | step=15270/50000 progress=30.5% loss=0.109248 lr=8.320181e-05 lr_backbone=8.320181e-05 elapsed=15:53:11 avg_step=3.75s recent_step=3.49s eta=36:07:55 +2026-09-15 06:24:11 | INFO | train rank=0 | step=15280/50000 progress=30.6% loss=0.102851 lr=8.317707e-05 lr_backbone=8.317707e-05 elapsed=15:53:49 avg_step=3.75s recent_step=3.80s eta=36:07:19 +2026-09-15 06:24:48 | INFO | train rank=0 | step=15290/50000 progress=30.6% loss=0.096561 lr=8.315232e-05 lr_backbone=8.315232e-05 elapsed=15:54:26 avg_step=3.75s recent_step=3.74s eta=36:06:41 +2026-09-15 06:25:21 | INFO | train rank=0 | step=15300/50000 progress=30.6% loss=0.097229 lr=8.312756e-05 lr_backbone=8.312756e-05 elapsed=15:54:59 avg_step=3.75s recent_step=3.25s eta=36:05:53 +2026-09-15 06:25:59 | INFO | train rank=0 | step=15310/50000 progress=30.6% loss=0.116884 lr=8.310279e-05 lr_backbone=8.310279e-05 elapsed=15:55:37 avg_step=3.75s recent_step=3.81s eta=36:05:17 +2026-09-15 06:26:31 | INFO | train rank=0 | step=15320/50000 progress=30.6% loss=0.082170 lr=8.307799e-05 lr_backbone=8.307799e-05 elapsed=15:56:09 avg_step=3.74s recent_step=3.19s eta=36:04:27 +2026-09-15 06:27:17 | INFO | train rank=0 | step=15330/50000 progress=30.7% loss=0.096412 lr=8.305319e-05 lr_backbone=8.305319e-05 elapsed=15:56:55 avg_step=3.75s recent_step=4.60s eta=36:04:09 +2026-09-15 06:27:57 | INFO | train rank=0 | step=15340/50000 progress=30.7% loss=0.101078 lr=8.302837e-05 lr_backbone=8.302837e-05 elapsed=15:57:35 avg_step=3.75s recent_step=4.00s eta=36:03:37 +2026-09-15 06:28:27 | INFO | train rank=0 | step=15350/50000 progress=30.7% loss=0.086611 lr=8.300353e-05 lr_backbone=8.300353e-05 elapsed=15:58:05 avg_step=3.75s recent_step=3.05s eta=36:02:44 +2026-09-15 06:29:04 | INFO | train rank=0 | step=15360/50000 progress=30.7% loss=0.093003 lr=8.297868e-05 lr_backbone=8.297868e-05 elapsed=15:58:42 avg_step=3.74s recent_step=3.67s eta=36:02:05 +2026-09-15 06:29:48 | INFO | train rank=0 | step=15370/50000 progress=30.7% loss=0.083299 lr=8.295382e-05 lr_backbone=8.295382e-05 elapsed=15:59:26 avg_step=3.75s recent_step=4.36s eta=36:01:41 +2026-09-15 06:30:25 | INFO | train rank=0 | step=15380/50000 progress=30.8% loss=0.121704 lr=8.292894e-05 lr_backbone=8.292894e-05 elapsed=16:00:03 avg_step=3.75s recent_step=3.70s eta=36:01:03 +2026-09-15 06:30:51 | INFO | train rank=0 | step=15390/50000 progress=30.8% loss=0.098716 lr=8.290405e-05 lr_backbone=8.290405e-05 elapsed=16:00:29 avg_step=3.74s recent_step=2.59s eta=35:59:59 +2026-09-15 06:31:25 | INFO | train rank=0 | step=15400/50000 progress=30.8% loss=0.097160 lr=8.287914e-05 lr_backbone=8.287914e-05 elapsed=16:01:03 avg_step=3.74s recent_step=3.40s eta=35:59:14 +2026-09-15 06:32:22 | INFO | train rank=0 | step=15410/50000 progress=30.8% loss=0.115037 lr=8.285422e-05 lr_backbone=8.285422e-05 elapsed=16:02:00 avg_step=3.75s recent_step=5.73s eta=35:59:21 +2026-09-15 06:32:49 | INFO | train rank=0 | step=15420/50000 progress=30.8% loss=0.106003 lr=8.282929e-05 lr_backbone=8.282929e-05 elapsed=16:02:27 avg_step=3.74s recent_step=2.73s eta=35:58:21 +2026-09-15 06:33:22 | INFO | train rank=0 | step=15430/50000 progress=30.9% loss=0.111465 lr=8.280434e-05 lr_backbone=8.280434e-05 elapsed=16:03:00 avg_step=3.74s recent_step=3.25s eta=35:57:33 +2026-09-15 06:33:55 | INFO | train rank=0 | step=15440/50000 progress=30.9% loss=0.075707 lr=8.277937e-05 lr_backbone=8.277937e-05 elapsed=16:03:33 avg_step=3.74s recent_step=3.35s eta=35:56:46 +2026-09-15 06:34:39 | INFO | train rank=0 | step=15450/50000 progress=30.9% loss=0.122515 lr=8.275440e-05 lr_backbone=8.275440e-05 elapsed=16:04:17 avg_step=3.74s recent_step=4.41s eta=35:56:24 +2026-09-15 06:35:16 | INFO | train rank=0 | step=15460/50000 progress=30.9% loss=0.100512 lr=8.272940e-05 lr_backbone=8.272940e-05 elapsed=16:04:54 avg_step=3.74s recent_step=3.66s eta=35:55:44 +2026-09-15 06:35:49 | INFO | train rank=0 | step=15470/50000 progress=30.9% loss=0.103015 lr=8.270440e-05 lr_backbone=8.270440e-05 elapsed=16:05:27 avg_step=3.74s recent_step=3.32s eta=35:54:57 +2026-09-15 06:36:24 | INFO | train rank=0 | step=15480/50000 progress=31.0% loss=0.102441 lr=8.267937e-05 lr_backbone=8.267937e-05 elapsed=16:06:02 avg_step=3.74s recent_step=3.53s eta=35:54:15 +2026-09-15 06:37:20 | INFO | train rank=0 | step=15490/50000 progress=31.0% loss=0.096738 lr=8.265434e-05 lr_backbone=8.265434e-05 elapsed=16:06:58 avg_step=3.75s recent_step=5.52s eta=35:54:17 +2026-09-15 06:37:50 | INFO | train rank=0 | step=15500/50000 progress=31.0% loss=0.086432 lr=8.262929e-05 lr_backbone=8.262929e-05 elapsed=16:07:28 avg_step=3.75s recent_step=3.00s eta=35:53:23 +2026-09-15 06:38:19 | INFO | train rank=0 | step=15510/50000 progress=31.0% loss=0.084036 lr=8.260422e-05 lr_backbone=8.260422e-05 elapsed=16:07:57 avg_step=3.74s recent_step=2.93s eta=35:52:28 +2026-09-15 06:39:00 | INFO | train rank=0 | step=15520/50000 progress=31.0% loss=0.104752 lr=8.257915e-05 lr_backbone=8.257915e-05 elapsed=16:08:38 avg_step=3.74s recent_step=4.15s eta=35:51:59 +2026-09-15 06:39:36 | INFO | train rank=0 | step=15530/50000 progress=31.1% loss=0.089191 lr=8.255405e-05 lr_backbone=8.255405e-05 elapsed=16:09:14 avg_step=3.74s recent_step=3.52s eta=35:51:17 +2026-09-15 06:40:29 | INFO | train rank=0 | step=15540/50000 progress=31.1% loss=0.099724 lr=8.252895e-05 lr_backbone=8.252895e-05 elapsed=16:10:07 avg_step=3.75s recent_step=5.36s eta=35:51:15 +2026-09-15 06:40:59 | INFO | train rank=0 | step=15550/50000 progress=31.1% loss=0.100566 lr=8.250382e-05 lr_backbone=8.250382e-05 elapsed=16:10:37 avg_step=3.75s recent_step=2.98s eta=35:50:21 +2026-09-15 06:41:34 | INFO | train rank=0 | step=15560/50000 progress=31.1% loss=0.101305 lr=8.247869e-05 lr_backbone=8.247869e-05 elapsed=16:11:12 avg_step=3.75s recent_step=3.51s eta=35:49:38 +2026-09-15 06:42:10 | INFO | train rank=0 | step=15570/50000 progress=31.1% loss=0.138969 lr=8.245354e-05 lr_backbone=8.245354e-05 elapsed=16:11:48 avg_step=3.74s recent_step=3.58s eta=35:48:57 +2026-09-15 06:43:03 | INFO | train rank=0 | step=15580/50000 progress=31.2% loss=0.109573 lr=8.242837e-05 lr_backbone=8.242837e-05 elapsed=16:12:41 avg_step=3.75s recent_step=5.27s eta=35:48:53 +2026-09-15 06:43:37 | INFO | train rank=0 | step=15590/50000 progress=31.2% loss=0.095493 lr=8.240320e-05 lr_backbone=8.240320e-05 elapsed=16:13:15 avg_step=3.75s recent_step=3.44s eta=35:48:09 +2026-09-15 06:44:14 | INFO | train rank=0 | step=15600/50000 progress=31.2% loss=0.107335 lr=8.237800e-05 lr_backbone=8.237800e-05 elapsed=16:13:52 avg_step=3.75s recent_step=3.75s eta=35:47:32 +2026-09-15 06:44:47 | INFO | train rank=0 | step=15610/50000 progress=31.2% loss=0.081931 lr=8.235280e-05 lr_backbone=8.235280e-05 elapsed=16:14:25 avg_step=3.75s recent_step=3.28s eta=35:46:44 +2026-09-15 06:45:39 | INFO | train rank=0 | step=15620/50000 progress=31.2% loss=0.103663 lr=8.232758e-05 lr_backbone=8.232758e-05 elapsed=16:15:17 avg_step=3.75s recent_step=5.15s eta=35:46:38 +2026-09-15 06:46:19 | INFO | train rank=0 | step=15630/50000 progress=31.3% loss=0.078715 lr=8.230234e-05 lr_backbone=8.230234e-05 elapsed=16:15:57 avg_step=3.75s recent_step=4.01s eta=35:46:06 +2026-09-15 06:46:38 | INFO | train rank=0 | step=15640/50000 progress=31.3% loss=0.093402 lr=8.227709e-05 lr_backbone=8.227709e-05 elapsed=16:16:16 avg_step=3.75s recent_step=1.92s eta=35:44:48 +2026-09-15 06:47:11 | INFO | train rank=0 | step=15650/50000 progress=31.3% loss=0.094827 lr=8.225183e-05 lr_backbone=8.225183e-05 elapsed=16:16:49 avg_step=3.75s recent_step=3.32s eta=35:44:02 +2026-09-15 06:48:00 | INFO | train rank=0 | step=15660/50000 progress=31.3% loss=0.103364 lr=8.222655e-05 lr_backbone=8.222655e-05 elapsed=16:17:38 avg_step=3.75s recent_step=4.87s eta=35:43:49 +2026-09-15 06:48:30 | INFO | train rank=0 | step=15670/50000 progress=31.3% loss=0.100115 lr=8.220126e-05 lr_backbone=8.220126e-05 elapsed=16:18:08 avg_step=3.75s recent_step=3.03s eta=35:42:56 +2026-09-15 06:49:05 | INFO | train rank=0 | step=15680/50000 progress=31.4% loss=0.104305 lr=8.217596e-05 lr_backbone=8.217596e-05 elapsed=16:18:43 avg_step=3.75s recent_step=3.44s eta=35:42:11 +2026-09-15 06:49:40 | INFO | train rank=0 | step=15690/50000 progress=31.4% loss=0.110498 lr=8.215064e-05 lr_backbone=8.215064e-05 elapsed=16:19:18 avg_step=3.74s recent_step=3.50s eta=35:41:29 +2026-09-15 06:50:31 | INFO | train rank=0 | step=15700/50000 progress=31.4% loss=0.096510 lr=8.212530e-05 lr_backbone=8.212530e-05 elapsed=16:20:09 avg_step=3.75s recent_step=5.12s eta=35:41:21 +2026-09-15 06:51:10 | INFO | train rank=0 | step=15710/50000 progress=31.4% loss=0.091810 lr=8.209996e-05 lr_backbone=8.209996e-05 elapsed=16:20:48 avg_step=3.75s recent_step=3.92s eta=35:40:48 +2026-09-15 06:51:39 | INFO | train rank=0 | step=15720/50000 progress=31.4% loss=0.116214 lr=8.207460e-05 lr_backbone=8.207460e-05 elapsed=16:21:17 avg_step=3.75s recent_step=2.90s eta=35:39:52 +2026-09-15 06:52:14 | INFO | train rank=0 | step=15730/50000 progress=31.5% loss=0.085592 lr=8.204922e-05 lr_backbone=8.204922e-05 elapsed=16:21:52 avg_step=3.75s recent_step=3.49s eta=35:39:09 +2026-09-15 06:52:59 | INFO | train rank=0 | step=15740/50000 progress=31.5% loss=0.093055 lr=8.202383e-05 lr_backbone=8.202383e-05 elapsed=16:22:37 avg_step=3.75s recent_step=4.49s eta=35:38:47 +2026-09-15 06:53:37 | INFO | train rank=0 | step=15750/50000 progress=31.5% loss=0.142682 lr=8.199843e-05 lr_backbone=8.199843e-05 elapsed=16:23:15 avg_step=3.75s recent_step=3.78s eta=35:38:11 +2026-09-15 06:54:12 | INFO | train rank=0 | step=15760/50000 progress=31.5% loss=0.102099 lr=8.197301e-05 lr_backbone=8.197301e-05 elapsed=16:23:50 avg_step=3.75s recent_step=3.52s eta=35:37:28 +2026-09-15 06:54:52 | INFO | train rank=0 | step=15770/50000 progress=31.5% loss=0.105486 lr=8.194758e-05 lr_backbone=8.194758e-05 elapsed=16:24:31 avg_step=3.75s recent_step=4.06s eta=35:36:58 +2026-09-15 06:55:43 | INFO | train rank=0 | step=15780/50000 progress=31.6% loss=0.098207 lr=8.192213e-05 lr_backbone=8.192213e-05 elapsed=16:25:21 avg_step=3.75s recent_step=5.06s eta=35:36:49 +2026-09-15 06:56:17 | INFO | train rank=0 | step=15790/50000 progress=31.6% loss=0.099067 lr=8.189667e-05 lr_backbone=8.189667e-05 elapsed=16:25:55 avg_step=3.75s recent_step=3.38s eta=35:36:03 +2026-09-15 06:56:48 | INFO | train rank=0 | step=15800/50000 progress=31.6% loss=0.098993 lr=8.187120e-05 lr_backbone=8.187120e-05 elapsed=16:26:26 avg_step=3.75s recent_step=3.11s eta=35:35:12 +2026-09-15 06:57:14 | INFO | train rank=0 | step=15810/50000 progress=31.6% loss=0.105168 lr=8.184571e-05 lr_backbone=8.184571e-05 elapsed=16:26:52 avg_step=3.75s recent_step=2.64s eta=35:34:10 +2026-09-15 06:57:58 | INFO | train rank=0 | step=15820/50000 progress=31.6% loss=0.089822 lr=8.182021e-05 lr_backbone=8.182021e-05 elapsed=16:27:36 avg_step=3.75s recent_step=4.33s eta=35:33:46 +2026-09-15 06:58:38 | INFO | train rank=0 | step=15830/50000 progress=31.7% loss=0.107265 lr=8.179470e-05 lr_backbone=8.179470e-05 elapsed=16:28:16 avg_step=3.75s recent_step=4.05s eta=35:33:15 +2026-09-15 06:59:12 | INFO | train rank=0 | step=15840/50000 progress=31.7% loss=0.094660 lr=8.176917e-05 lr_backbone=8.176917e-05 elapsed=16:28:50 avg_step=3.75s recent_step=3.38s eta=35:32:29 +2026-09-15 06:59:50 | INFO | train rank=0 | step=15850/50000 progress=31.7% loss=0.094764 lr=8.174362e-05 lr_backbone=8.174362e-05 elapsed=16:29:28 avg_step=3.75s recent_step=3.80s eta=35:31:53 +2026-09-15 07:00:38 | INFO | train rank=0 | step=15860/50000 progress=31.7% loss=0.092639 lr=8.171807e-05 lr_backbone=8.171807e-05 elapsed=16:30:16 avg_step=3.75s recent_step=4.85s eta=35:31:39 +2026-09-15 07:01:11 | INFO | train rank=0 | step=15870/50000 progress=31.7% loss=0.096910 lr=8.169250e-05 lr_backbone=8.169250e-05 elapsed=16:30:49 avg_step=3.75s recent_step=3.29s eta=35:30:52 +2026-09-15 07:01:42 | INFO | train rank=0 | step=15880/50000 progress=31.8% loss=0.096448 lr=8.166691e-05 lr_backbone=8.166691e-05 elapsed=16:31:20 avg_step=3.75s recent_step=3.05s eta=35:30:00 +2026-09-15 07:02:22 | INFO | train rank=0 | step=15890/50000 progress=31.8% loss=0.113344 lr=8.164131e-05 lr_backbone=8.164131e-05 elapsed=16:32:00 avg_step=3.75s recent_step=4.01s eta=35:29:28 +2026-09-15 07:03:11 | INFO | train rank=0 | step=15900/50000 progress=31.8% loss=0.111082 lr=8.161570e-05 lr_backbone=8.161570e-05 elapsed=16:32:49 avg_step=3.75s recent_step=4.92s eta=35:29:15 +2026-09-15 07:03:47 | INFO | train rank=0 | step=15910/50000 progress=31.8% loss=0.105857 lr=8.159007e-05 lr_backbone=8.159007e-05 elapsed=16:33:25 avg_step=3.75s recent_step=3.59s eta=35:28:35 +2026-09-15 07:04:18 | INFO | train rank=0 | step=15920/50000 progress=31.8% loss=0.127671 lr=8.156443e-05 lr_backbone=8.156443e-05 elapsed=16:33:56 avg_step=3.75s recent_step=3.06s eta=35:27:43 +2026-09-15 07:04:46 | INFO | train rank=0 | step=15930/50000 progress=31.9% loss=0.089073 lr=8.153878e-05 lr_backbone=8.153878e-05 elapsed=16:34:24 avg_step=3.75s recent_step=2.85s eta=35:26:46 +2026-09-15 07:05:34 | INFO | train rank=0 | step=15940/50000 progress=31.9% loss=0.102557 lr=8.151311e-05 lr_backbone=8.151311e-05 elapsed=16:35:12 avg_step=3.75s recent_step=4.80s eta=35:26:31 +2026-09-15 07:06:02 | INFO | train rank=0 | step=15950/50000 progress=31.9% loss=0.090345 lr=8.148743e-05 lr_backbone=8.148743e-05 elapsed=16:35:40 avg_step=3.75s recent_step=2.81s eta=35:25:33 +2026-09-15 07:06:41 | INFO | train rank=0 | step=15960/50000 progress=31.9% loss=0.099747 lr=8.146174e-05 lr_backbone=8.146174e-05 elapsed=16:36:19 avg_step=3.75s recent_step=3.91s eta=35:25:00 +2026-09-15 07:07:22 | INFO | train rank=0 | step=15970/50000 progress=31.9% loss=0.098772 lr=8.143603e-05 lr_backbone=8.143603e-05 elapsed=16:37:00 avg_step=3.75s recent_step=4.04s eta=35:24:28 +2026-09-15 07:08:02 | INFO | train rank=0 | step=15980/50000 progress=32.0% loss=0.096893 lr=8.141030e-05 lr_backbone=8.141030e-05 elapsed=16:37:40 avg_step=3.75s recent_step=4.00s eta=35:23:56 +2026-09-15 07:08:43 | INFO | train rank=0 | step=15990/50000 progress=32.0% loss=0.076164 lr=8.138457e-05 lr_backbone=8.138457e-05 elapsed=16:38:21 avg_step=3.75s recent_step=4.14s eta=35:23:27 +2026-09-15 07:09:10 | INFO | train rank=0 | step=16000/50000 progress=32.0% loss=0.091209 lr=8.135882e-05 lr_backbone=8.135882e-05 elapsed=16:38:48 avg_step=3.75s recent_step=2.72s eta=35:22:28 +2026-09-15 07:09:50 | INFO | train rank=0 | step=16010/50000 progress=32.0% loss=0.085836 lr=8.133305e-05 lr_backbone=8.133305e-05 elapsed=16:39:28 avg_step=3.75s recent_step=4.02s eta=35:21:56 +2026-09-15 07:10:36 | INFO | train rank=0 | step=16020/50000 progress=32.0% loss=0.107279 lr=8.130728e-05 lr_backbone=8.130728e-05 elapsed=16:40:14 avg_step=3.75s recent_step=4.59s eta=35:21:37 +2026-09-15 07:11:05 | INFO | train rank=0 | step=16030/50000 progress=32.1% loss=0.091058 lr=8.128149e-05 lr_backbone=8.128149e-05 elapsed=16:40:43 avg_step=3.75s recent_step=2.83s eta=35:20:40 +2026-09-15 07:11:42 | INFO | train rank=0 | step=16040/50000 progress=32.1% loss=0.120247 lr=8.125568e-05 lr_backbone=8.125568e-05 elapsed=16:41:20 avg_step=3.75s recent_step=3.72s eta=35:20:02 +2026-09-15 07:12:11 | INFO | train rank=0 | step=16050/50000 progress=32.1% loss=0.093021 lr=8.122986e-05 lr_backbone=8.122986e-05 elapsed=16:41:49 avg_step=3.75s recent_step=2.94s eta=35:19:08 +2026-09-15 07:13:06 | INFO | train rank=0 | step=16060/50000 progress=32.1% loss=0.096024 lr=8.120403e-05 lr_backbone=8.120403e-05 elapsed=16:42:44 avg_step=3.75s recent_step=5.45s eta=35:19:06 +2026-09-15 07:13:42 | INFO | train rank=0 | step=16070/50000 progress=32.1% loss=0.128652 lr=8.117818e-05 lr_backbone=8.117818e-05 elapsed=16:43:20 avg_step=3.75s recent_step=3.66s eta=35:18:27 +2026-09-15 07:14:19 | INFO | train rank=0 | step=16080/50000 progress=32.2% loss=0.075532 lr=8.115232e-05 lr_backbone=8.115232e-05 elapsed=16:43:57 avg_step=3.75s recent_step=3.69s eta=35:17:48 +2026-09-15 07:14:54 | INFO | train rank=0 | step=16090/50000 progress=32.2% loss=0.101852 lr=8.112645e-05 lr_backbone=8.112645e-05 elapsed=16:44:32 avg_step=3.75s recent_step=3.46s eta=35:17:05 +2026-09-15 07:15:46 | INFO | train rank=0 | step=16100/50000 progress=32.2% loss=0.110847 lr=8.110056e-05 lr_backbone=8.110056e-05 elapsed=16:45:24 avg_step=3.75s recent_step=5.19s eta=35:16:58 +2026-09-15 07:16:15 | INFO | train rank=0 | step=16110/50000 progress=32.2% loss=0.105490 lr=8.107466e-05 lr_backbone=8.107466e-05 elapsed=16:45:53 avg_step=3.75s recent_step=2.91s eta=35:16:03 +2026-09-15 07:16:47 | INFO | train rank=0 | step=16120/50000 progress=32.2% loss=0.090967 lr=8.104875e-05 lr_backbone=8.104875e-05 elapsed=16:46:25 avg_step=3.75s recent_step=3.17s eta=35:15:13 +2026-09-15 07:17:09 | INFO | train rank=0 | step=16130/50000 progress=32.3% loss=0.082549 lr=8.102282e-05 lr_backbone=8.102282e-05 elapsed=16:46:47 avg_step=3.75s recent_step=2.28s eta=35:14:05 +2026-09-15 07:17:59 | INFO | train rank=0 | step=16140/50000 progress=32.3% loss=0.106142 lr=8.099688e-05 lr_backbone=8.099688e-05 elapsed=16:47:37 avg_step=3.75s recent_step=4.98s eta=35:13:53 +2026-09-15 07:18:30 | INFO | train rank=0 | step=16150/50000 progress=32.3% loss=0.125486 lr=8.097093e-05 lr_backbone=8.097093e-05 elapsed=16:48:08 avg_step=3.75s recent_step=3.04s eta=35:13:01 +2026-09-15 07:18:58 | INFO | train rank=0 | step=16160/50000 progress=32.3% loss=0.099441 lr=8.094496e-05 lr_backbone=8.094496e-05 elapsed=16:48:36 avg_step=3.74s recent_step=2.80s eta=35:12:04 +2026-09-15 07:19:39 | INFO | train rank=0 | step=16170/50000 progress=32.3% loss=0.111549 lr=8.091898e-05 lr_backbone=8.091898e-05 elapsed=16:49:17 avg_step=3.75s recent_step=4.08s eta=35:11:34 +2026-09-15 07:20:30 | INFO | train rank=0 | step=16180/50000 progress=32.4% loss=0.101767 lr=8.089298e-05 lr_backbone=8.089298e-05 elapsed=16:50:08 avg_step=3.75s recent_step=5.13s eta=35:11:25 +2026-09-15 07:21:02 | INFO | train rank=0 | step=16190/50000 progress=32.4% loss=0.124334 lr=8.086697e-05 lr_backbone=8.086697e-05 elapsed=16:50:40 avg_step=3.75s recent_step=3.23s eta=35:10:37 +2026-09-15 07:21:33 | INFO | train rank=0 | step=16200/50000 progress=32.4% loss=0.111520 lr=8.084095e-05 lr_backbone=8.084095e-05 elapsed=16:51:11 avg_step=3.75s recent_step=3.06s eta=35:09:45 +2026-09-15 07:22:02 | INFO | train rank=0 | step=16210/50000 progress=32.4% loss=0.119837 lr=8.081491e-05 lr_backbone=8.081491e-05 elapsed=16:51:40 avg_step=3.74s recent_step=2.88s eta=35:08:50 +2026-09-15 07:22:56 | INFO | train rank=0 | step=16220/50000 progress=32.4% loss=0.097147 lr=8.078886e-05 lr_backbone=8.078886e-05 elapsed=16:52:34 avg_step=3.75s recent_step=5.42s eta=35:08:47 +2026-09-15 07:23:31 | INFO | train rank=0 | step=16230/50000 progress=32.5% loss=0.109052 lr=8.076280e-05 lr_backbone=8.076280e-05 elapsed=16:53:09 avg_step=3.75s recent_step=3.57s eta=35:08:06 +2026-09-15 07:24:06 | INFO | train rank=0 | step=16240/50000 progress=32.5% loss=0.097034 lr=8.073673e-05 lr_backbone=8.073673e-05 elapsed=16:53:44 avg_step=3.75s recent_step=3.40s eta=35:07:22 +2026-09-15 07:24:37 | INFO | train rank=0 | step=16250/50000 progress=32.5% loss=0.116419 lr=8.071064e-05 lr_backbone=8.071064e-05 elapsed=16:54:15 avg_step=3.74s recent_step=3.17s eta=35:06:32 +2026-09-15 07:25:28 | INFO | train rank=0 | step=16260/50000 progress=32.5% loss=0.094280 lr=8.068453e-05 lr_backbone=8.068453e-05 elapsed=16:55:06 avg_step=3.75s recent_step=5.08s eta=35:06:23 +2026-09-15 07:26:04 | INFO | train rank=0 | step=16270/50000 progress=32.5% loss=0.119398 lr=8.065842e-05 lr_backbone=8.065842e-05 elapsed=16:55:42 avg_step=3.75s recent_step=3.64s eta=35:05:43 +2026-09-15 07:26:30 | INFO | train rank=0 | step=16280/50000 progress=32.6% loss=0.104312 lr=8.063229e-05 lr_backbone=8.063229e-05 elapsed=16:56:08 avg_step=3.74s recent_step=2.56s eta=35:04:41 +2026-09-15 07:27:05 | INFO | train rank=0 | step=16290/50000 progress=32.6% loss=0.100307 lr=8.060614e-05 lr_backbone=8.060614e-05 elapsed=16:56:43 avg_step=3.74s recent_step=3.47s eta=35:03:58 +2026-09-15 07:27:52 | INFO | train rank=0 | step=16300/50000 progress=32.6% loss=0.066521 lr=8.057999e-05 lr_backbone=8.057999e-05 elapsed=16:57:30 avg_step=3.75s recent_step=4.72s eta=35:03:40 +2026-09-15 07:28:22 | INFO | train rank=0 | step=16310/50000 progress=32.6% loss=0.116274 lr=8.055382e-05 lr_backbone=8.055382e-05 elapsed=16:58:00 avg_step=3.74s recent_step=2.96s eta=35:02:47 +2026-09-15 07:29:00 | INFO | train rank=0 | step=16320/50000 progress=32.6% loss=0.085403 lr=8.052763e-05 lr_backbone=8.052763e-05 elapsed=16:58:38 avg_step=3.74s recent_step=3.80s eta=35:02:10 +2026-09-15 07:29:30 | INFO | train rank=0 | step=16330/50000 progress=32.7% loss=0.101630 lr=8.050144e-05 lr_backbone=8.050144e-05 elapsed=16:59:08 avg_step=3.74s recent_step=3.02s eta=35:01:18 +2026-09-15 07:30:08 | INFO | train rank=0 | step=16340/50000 progress=32.7% loss=0.093753 lr=8.047523e-05 lr_backbone=8.047523e-05 elapsed=16:59:46 avg_step=3.74s recent_step=3.81s eta=35:00:42 +2026-09-15 07:30:46 | INFO | train rank=0 | step=16350/50000 progress=32.7% loss=0.101163 lr=8.044900e-05 lr_backbone=8.044900e-05 elapsed=17:00:24 avg_step=3.74s recent_step=3.77s eta=35:00:05 +2026-09-15 07:31:16 | INFO | train rank=0 | step=16360/50000 progress=32.7% loss=0.084933 lr=8.042277e-05 lr_backbone=8.042277e-05 elapsed=17:00:54 avg_step=3.74s recent_step=3.04s eta=34:59:13 +2026-09-15 07:31:45 | INFO | train rank=0 | step=16370/50000 progress=32.7% loss=0.097274 lr=8.039651e-05 lr_backbone=8.039651e-05 elapsed=17:01:23 avg_step=3.74s recent_step=2.87s eta=34:58:18 +2026-09-15 07:32:34 | INFO | train rank=0 | step=16380/50000 progress=32.8% loss=0.104582 lr=8.037025e-05 lr_backbone=8.037025e-05 elapsed=17:02:12 avg_step=3.74s recent_step=4.95s eta=34:58:05 +2026-09-15 07:33:16 | INFO | train rank=0 | step=16390/50000 progress=32.8% loss=0.090629 lr=8.034397e-05 lr_backbone=8.034397e-05 elapsed=17:02:54 avg_step=3.74s recent_step=4.19s eta=34:57:37 +2026-09-15 07:33:49 | INFO | train rank=0 | step=16400/50000 progress=32.8% loss=0.098598 lr=8.031768e-05 lr_backbone=8.031768e-05 elapsed=17:03:27 avg_step=3.74s recent_step=3.29s eta=34:56:50 +2026-09-15 07:34:24 | INFO | train rank=0 | step=16410/50000 progress=32.8% loss=0.163198 lr=8.029138e-05 lr_backbone=8.029138e-05 elapsed=17:04:02 avg_step=3.74s recent_step=3.54s eta=34:56:09 +2026-09-15 07:35:14 | INFO | train rank=0 | step=16420/50000 progress=32.8% loss=0.128276 lr=8.026507e-05 lr_backbone=8.026507e-05 elapsed=17:04:52 avg_step=3.74s recent_step=4.95s eta=34:55:56 +2026-09-15 07:35:44 | INFO | train rank=0 | step=16430/50000 progress=32.9% loss=0.086112 lr=8.023874e-05 lr_backbone=8.023874e-05 elapsed=17:05:22 avg_step=3.74s recent_step=3.03s eta=34:55:04 +2026-09-15 07:36:15 | INFO | train rank=0 | step=16440/50000 progress=32.9% loss=0.092051 lr=8.021239e-05 lr_backbone=8.021239e-05 elapsed=17:05:53 avg_step=3.74s recent_step=3.02s eta=34:54:12 +2026-09-15 07:36:46 | INFO | train rank=0 | step=16450/50000 progress=32.9% loss=0.106902 lr=8.018604e-05 lr_backbone=8.018604e-05 elapsed=17:06:24 avg_step=3.74s recent_step=3.10s eta=34:53:21 +2026-09-15 07:37:38 | INFO | train rank=0 | step=16460/50000 progress=32.9% loss=0.084318 lr=8.015967e-05 lr_backbone=8.015967e-05 elapsed=17:07:16 avg_step=3.74s recent_step=5.25s eta=34:53:14 +2026-09-15 07:38:05 | INFO | train rank=0 | step=16470/50000 progress=32.9% loss=0.096218 lr=8.013328e-05 lr_backbone=8.013328e-05 elapsed=17:07:43 avg_step=3.74s recent_step=2.70s eta=34:52:16 +2026-09-15 07:38:41 | INFO | train rank=0 | step=16480/50000 progress=33.0% loss=0.118414 lr=8.010689e-05 lr_backbone=8.010689e-05 elapsed=17:08:19 avg_step=3.74s recent_step=3.57s eta=34:51:35 +2026-09-15 07:39:19 | INFO | train rank=0 | step=16490/50000 progress=33.0% loss=0.085550 lr=8.008048e-05 lr_backbone=8.008048e-05 elapsed=17:08:57 avg_step=3.74s recent_step=3.86s eta=34:50:59 +2026-09-15 07:40:00 | INFO | train rank=0 | step=16500/50000 progress=33.0% loss=0.093915 lr=8.005406e-05 lr_backbone=8.005406e-05 elapsed=17:09:38 avg_step=3.74s recent_step=4.07s eta=34:50:29 +2026-09-15 07:40:27 | INFO | train rank=0 | step=16510/50000 progress=33.0% loss=0.103946 lr=8.002762e-05 lr_backbone=8.002762e-05 elapsed=17:10:05 avg_step=3.74s recent_step=2.70s eta=34:49:30 +2026-09-15 07:40:59 | INFO | train rank=0 | step=16520/50000 progress=33.0% loss=0.091391 lr=8.000117e-05 lr_backbone=8.000117e-05 elapsed=17:10:37 avg_step=3.74s recent_step=3.20s eta=34:48:42 +2026-09-15 07:41:32 | INFO | train rank=0 | step=16530/50000 progress=33.1% loss=0.094930 lr=7.997471e-05 lr_backbone=7.997471e-05 elapsed=17:11:10 avg_step=3.74s recent_step=3.33s eta=34:47:56 +2026-09-15 07:42:11 | INFO | train rank=0 | step=16540/50000 progress=33.1% loss=0.096389 lr=7.994824e-05 lr_backbone=7.994824e-05 elapsed=17:11:49 avg_step=3.74s recent_step=3.89s eta=34:47:21 +2026-09-15 07:42:45 | INFO | train rank=0 | step=16550/50000 progress=33.1% loss=0.111929 lr=7.992175e-05 lr_backbone=7.992175e-05 elapsed=17:12:23 avg_step=3.74s recent_step=3.38s eta=34:46:37 +2026-09-15 07:43:18 | INFO | train rank=0 | step=16560/50000 progress=33.1% loss=0.099772 lr=7.989525e-05 lr_backbone=7.989525e-05 elapsed=17:12:56 avg_step=3.74s recent_step=3.28s eta=34:45:50 +2026-09-15 07:43:52 | INFO | train rank=0 | step=16570/50000 progress=33.1% loss=0.102309 lr=7.986874e-05 lr_backbone=7.986874e-05 elapsed=17:13:30 avg_step=3.74s recent_step=3.45s eta=34:45:07 +2026-09-15 07:44:33 | INFO | train rank=0 | step=16580/50000 progress=33.2% loss=0.102634 lr=7.984221e-05 lr_backbone=7.984221e-05 elapsed=17:14:11 avg_step=3.74s recent_step=4.04s eta=34:44:35 +2026-09-15 07:45:05 | INFO | train rank=0 | step=16590/50000 progress=33.2% loss=0.124765 lr=7.981567e-05 lr_backbone=7.981567e-05 elapsed=17:14:43 avg_step=3.74s recent_step=3.19s eta=34:43:47 +2026-09-15 07:46:00 | INFO | train rank=0 | step=16600/50000 progress=33.2% loss=0.096762 lr=7.978912e-05 lr_backbone=7.978912e-05 elapsed=17:15:38 avg_step=3.74s recent_step=5.58s eta=34:43:46 +2026-09-15 07:46:40 | INFO | train rank=0 | step=16610/50000 progress=33.2% loss=0.103497 lr=7.976255e-05 lr_backbone=7.976255e-05 elapsed=17:16:18 avg_step=3.74s recent_step=3.97s eta=34:43:13 +2026-09-15 07:47:14 | INFO | train rank=0 | step=16620/50000 progress=33.2% loss=0.088266 lr=7.973597e-05 lr_backbone=7.973597e-05 elapsed=17:16:52 avg_step=3.74s recent_step=3.40s eta=34:42:29 +2026-09-15 07:47:55 | INFO | train rank=0 | step=16630/50000 progress=33.3% loss=0.112429 lr=7.970938e-05 lr_backbone=7.970938e-05 elapsed=17:17:33 avg_step=3.74s recent_step=4.12s eta=34:41:59 +2026-09-15 07:48:27 | INFO | train rank=0 | step=16640/50000 progress=33.3% loss=0.103299 lr=7.968277e-05 lr_backbone=7.968277e-05 elapsed=17:18:05 avg_step=3.74s recent_step=3.12s eta=34:41:09 +2026-09-15 07:49:19 | INFO | train rank=0 | step=16650/50000 progress=33.3% loss=0.141183 lr=7.965615e-05 lr_backbone=7.965615e-05 elapsed=17:18:57 avg_step=3.74s recent_step=5.24s eta=34:41:02 +2026-09-15 07:49:53 | INFO | train rank=0 | step=16660/50000 progress=33.3% loss=0.093781 lr=7.962952e-05 lr_backbone=7.962952e-05 elapsed=17:19:31 avg_step=3.74s recent_step=3.41s eta=34:40:17 +2026-09-15 07:50:33 | INFO | train rank=0 | step=16670/50000 progress=33.3% loss=0.099410 lr=7.960288e-05 lr_backbone=7.960288e-05 elapsed=17:20:11 avg_step=3.74s recent_step=4.00s eta=34:39:45 +2026-09-15 07:51:08 | INFO | train rank=0 | step=16680/50000 progress=33.4% loss=0.089548 lr=7.957622e-05 lr_backbone=7.957622e-05 elapsed=17:20:46 avg_step=3.74s recent_step=3.51s eta=34:39:03 +2026-09-15 07:51:50 | INFO | train rank=0 | step=16690/50000 progress=33.4% loss=0.124225 lr=7.954955e-05 lr_backbone=7.954955e-05 elapsed=17:21:28 avg_step=3.74s recent_step=4.18s eta=34:38:34 +2026-09-15 07:52:24 | INFO | train rank=0 | step=16700/50000 progress=33.4% loss=0.090086 lr=7.952287e-05 lr_backbone=7.952287e-05 elapsed=17:22:02 avg_step=3.74s recent_step=3.40s eta=34:37:50 +2026-09-15 07:52:54 | INFO | train rank=0 | step=16710/50000 progress=33.4% loss=0.131854 lr=7.949617e-05 lr_backbone=7.949617e-05 elapsed=17:22:32 avg_step=3.74s recent_step=3.02s eta=34:36:58 +2026-09-15 07:53:31 | INFO | train rank=0 | step=16720/50000 progress=33.4% loss=0.105587 lr=7.946947e-05 lr_backbone=7.946947e-05 elapsed=17:23:09 avg_step=3.74s recent_step=3.73s eta=34:36:20 +2026-09-15 07:54:26 | INFO | train rank=0 | step=16730/50000 progress=33.5% loss=0.104419 lr=7.944274e-05 lr_backbone=7.944274e-05 elapsed=17:24:04 avg_step=3.74s recent_step=5.44s eta=34:36:17 +2026-09-15 07:55:06 | INFO | train rank=0 | step=16740/50000 progress=33.5% loss=0.100659 lr=7.941601e-05 lr_backbone=7.941601e-05 elapsed=17:24:44 avg_step=3.74s recent_step=4.05s eta=34:35:45 +2026-09-15 07:55:33 | INFO | train rank=0 | step=16750/50000 progress=33.5% loss=0.106133 lr=7.938926e-05 lr_backbone=7.938926e-05 elapsed=17:25:11 avg_step=3.74s recent_step=2.72s eta=34:34:47 +2026-09-15 07:56:09 | INFO | train rank=0 | step=16760/50000 progress=33.5% loss=0.092295 lr=7.936250e-05 lr_backbone=7.936250e-05 elapsed=17:25:47 avg_step=3.74s recent_step=3.53s eta=34:34:06 +2026-09-15 07:56:54 | INFO | train rank=0 | step=16770/50000 progress=33.5% loss=0.095167 lr=7.933573e-05 lr_backbone=7.933573e-05 elapsed=17:26:32 avg_step=3.74s recent_step=4.51s eta=34:33:43 +2026-09-15 07:57:20 | INFO | train rank=0 | step=16780/50000 progress=33.6% loss=0.086749 lr=7.930894e-05 lr_backbone=7.930894e-05 elapsed=17:26:58 avg_step=3.74s recent_step=2.64s eta=34:32:44 +2026-09-15 07:57:55 | INFO | train rank=0 | step=16790/50000 progress=33.6% loss=0.099736 lr=7.928215e-05 lr_backbone=7.928215e-05 elapsed=17:27:33 avg_step=3.74s recent_step=3.48s eta=34:32:01 +2026-09-15 07:58:30 | INFO | train rank=0 | step=16800/50000 progress=33.6% loss=0.093999 lr=7.925533e-05 lr_backbone=7.925533e-05 elapsed=17:28:08 avg_step=3.74s recent_step=3.47s eta=34:31:18 +2026-09-15 07:59:20 | INFO | train rank=0 | step=16810/50000 progress=33.6% loss=0.103498 lr=7.922851e-05 lr_backbone=7.922851e-05 elapsed=17:28:59 avg_step=3.74s recent_step=5.09s eta=34:31:08 +2026-09-15 07:59:58 | INFO | train rank=0 | step=16820/50000 progress=33.6% loss=0.107535 lr=7.920167e-05 lr_backbone=7.920167e-05 elapsed=17:29:36 avg_step=3.74s recent_step=3.70s eta=34:30:29 +2026-09-15 08:00:33 | INFO | train rank=0 | step=16830/50000 progress=33.7% loss=0.099807 lr=7.917482e-05 lr_backbone=7.917482e-05 elapsed=17:30:11 avg_step=3.74s recent_step=3.50s eta=34:29:47 +2026-09-15 08:01:13 | INFO | train rank=0 | step=16840/50000 progress=33.7% loss=0.126492 lr=7.914796e-05 lr_backbone=7.914796e-05 elapsed=17:30:51 avg_step=3.74s recent_step=4.00s eta=34:29:15 +2026-09-15 08:01:56 | INFO | train rank=0 | step=16850/50000 progress=33.7% loss=0.110476 lr=7.912108e-05 lr_backbone=7.912108e-05 elapsed=17:31:34 avg_step=3.74s recent_step=4.38s eta=34:28:50 +2026-09-15 08:02:30 | INFO | train rank=0 | step=16860/50000 progress=33.7% loss=0.087425 lr=7.909420e-05 lr_backbone=7.909420e-05 elapsed=17:32:08 avg_step=3.74s recent_step=3.35s eta=34:28:05 +2026-09-15 08:03:03 | INFO | train rank=0 | step=16870/50000 progress=33.7% loss=0.084337 lr=7.906730e-05 lr_backbone=7.906730e-05 elapsed=17:32:41 avg_step=3.74s recent_step=3.35s eta=34:27:19 +2026-09-15 08:03:35 | INFO | train rank=0 | step=16880/50000 progress=33.8% loss=0.099878 lr=7.904038e-05 lr_backbone=7.904038e-05 elapsed=17:33:13 avg_step=3.74s recent_step=3.19s eta=34:26:31 +2026-09-15 08:04:31 | INFO | train rank=0 | step=16890/50000 progress=33.8% loss=0.116778 lr=7.901346e-05 lr_backbone=7.901346e-05 elapsed=17:34:09 avg_step=3.74s recent_step=5.58s eta=34:26:29 +2026-09-15 08:05:00 | INFO | train rank=0 | step=16900/50000 progress=33.8% loss=0.090511 lr=7.898652e-05 lr_backbone=7.898652e-05 elapsed=17:34:38 avg_step=3.74s recent_step=2.92s eta=34:25:36 +2026-09-15 08:05:32 | INFO | train rank=0 | step=16910/50000 progress=33.8% loss=0.102959 lr=7.895957e-05 lr_backbone=7.895957e-05 elapsed=17:35:10 avg_step=3.74s recent_step=3.18s eta=34:24:47 +2026-09-15 08:06:09 | INFO | train rank=0 | step=16920/50000 progress=33.8% loss=0.095200 lr=7.893260e-05 lr_backbone=7.893260e-05 elapsed=17:35:47 avg_step=3.74s recent_step=3.75s eta=34:24:10 +2026-09-15 08:06:57 | INFO | train rank=0 | step=16930/50000 progress=33.9% loss=0.095554 lr=7.890562e-05 lr_backbone=7.890562e-05 elapsed=17:36:35 avg_step=3.74s recent_step=4.80s eta=34:23:53 +2026-09-15 08:07:28 | INFO | train rank=0 | step=16940/50000 progress=33.9% loss=0.084297 lr=7.887864e-05 lr_backbone=7.887864e-05 elapsed=17:37:06 avg_step=3.74s recent_step=3.05s eta=34:23:02 +2026-09-15 08:08:01 | INFO | train rank=0 | step=16950/50000 progress=33.9% loss=0.117705 lr=7.885163e-05 lr_backbone=7.885163e-05 elapsed=17:37:39 avg_step=3.74s recent_step=3.34s eta=34:22:17 +2026-09-15 08:08:34 | INFO | train rank=0 | step=16960/50000 progress=33.9% loss=0.112241 lr=7.882462e-05 lr_backbone=7.882462e-05 elapsed=17:38:12 avg_step=3.74s recent_step=3.30s eta=34:21:31 +2026-09-15 08:09:23 | INFO | train rank=0 | step=16970/50000 progress=33.9% loss=0.117110 lr=7.879759e-05 lr_backbone=7.879759e-05 elapsed=17:39:01 avg_step=3.74s recent_step=4.85s eta=34:21:15 +2026-09-15 08:09:49 | INFO | train rank=0 | step=16980/50000 progress=34.0% loss=0.074022 lr=7.877055e-05 lr_backbone=7.877055e-05 elapsed=17:39:27 avg_step=3.74s recent_step=2.64s eta=34:20:16 +2026-09-15 08:10:30 | INFO | train rank=0 | step=16990/50000 progress=34.0% loss=0.095407 lr=7.874350e-05 lr_backbone=7.874350e-05 elapsed=17:40:08 avg_step=3.74s recent_step=4.06s eta=34:19:45 +2026-09-15 08:10:54 | INFO | train rank=0 | step=17000/50000 progress=34.0% loss=0.080304 lr=7.871643e-05 lr_backbone=7.871643e-05 elapsed=17:40:32 avg_step=3.74s recent_step=2.40s eta=34:18:41 +2026-09-15 08:11:41 | INFO | train rank=0 | step=17010/50000 progress=34.0% loss=0.080437 lr=7.868936e-05 lr_backbone=7.868936e-05 elapsed=17:41:19 avg_step=3.74s recent_step=4.72s eta=34:18:23 +2026-09-15 08:12:13 | INFO | train rank=0 | step=17020/50000 progress=34.0% loss=0.106503 lr=7.866227e-05 lr_backbone=7.866227e-05 elapsed=17:41:51 avg_step=3.74s recent_step=3.18s eta=34:17:35 +2026-09-15 08:12:44 | INFO | train rank=0 | step=17030/50000 progress=34.1% loss=0.067074 lr=7.863516e-05 lr_backbone=7.863516e-05 elapsed=17:42:22 avg_step=3.74s recent_step=3.14s eta=34:16:45 +2026-09-15 08:13:25 | INFO | train rank=0 | step=17040/50000 progress=34.1% loss=0.111798 lr=7.860805e-05 lr_backbone=7.860805e-05 elapsed=17:43:03 avg_step=3.74s recent_step=4.03s eta=34:16:14 +2026-09-15 08:14:12 | INFO | train rank=0 | step=17050/50000 progress=34.1% loss=0.087333 lr=7.858092e-05 lr_backbone=7.858092e-05 elapsed=17:43:50 avg_step=3.74s recent_step=4.69s eta=34:15:54 +2026-09-15 08:14:40 | INFO | train rank=0 | step=17060/50000 progress=34.1% loss=0.106255 lr=7.855378e-05 lr_backbone=7.855378e-05 elapsed=17:44:18 avg_step=3.74s recent_step=2.86s eta=34:15:00 +2026-09-15 08:15:31 | INFO | train rank=0 | step=17070/50000 progress=34.1% loss=0.098151 lr=7.852663e-05 lr_backbone=7.852663e-05 elapsed=17:45:09 avg_step=3.74s recent_step=5.08s eta=34:14:48 +2026-09-15 08:16:03 | INFO | train rank=0 | step=17080/50000 progress=34.2% loss=0.105986 lr=7.849946e-05 lr_backbone=7.849946e-05 elapsed=17:45:41 avg_step=3.74s recent_step=3.16s eta=34:13:59 +2026-09-15 08:16:41 | INFO | train rank=0 | step=17090/50000 progress=34.2% loss=0.081432 lr=7.847228e-05 lr_backbone=7.847228e-05 elapsed=17:46:19 avg_step=3.74s recent_step=3.82s eta=34:13:24 +2026-09-15 08:17:17 | INFO | train rank=0 | step=17100/50000 progress=34.2% loss=0.107703 lr=7.844509e-05 lr_backbone=7.844509e-05 elapsed=17:46:55 avg_step=3.74s recent_step=3.61s eta=34:12:43 +2026-09-15 08:17:50 | INFO | train rank=0 | step=17110/50000 progress=34.2% loss=0.111151 lr=7.841789e-05 lr_backbone=7.841789e-05 elapsed=17:47:28 avg_step=3.74s recent_step=3.35s eta=34:11:59 +2026-09-15 08:18:30 | INFO | train rank=0 | step=17120/50000 progress=34.2% loss=0.103825 lr=7.839067e-05 lr_backbone=7.839067e-05 elapsed=17:48:08 avg_step=3.74s recent_step=3.95s eta=34:11:25 +2026-09-15 08:19:25 | INFO | train rank=0 | step=17130/50000 progress=34.3% loss=0.091640 lr=7.836345e-05 lr_backbone=7.836345e-05 elapsed=17:49:03 avg_step=3.74s recent_step=5.52s eta=34:11:22 +2026-09-15 08:19:59 | INFO | train rank=0 | step=17140/50000 progress=34.3% loss=0.100670 lr=7.833621e-05 lr_backbone=7.833621e-05 elapsed=17:49:37 avg_step=3.74s recent_step=3.36s eta=34:10:37 +2026-09-15 08:20:28 | INFO | train rank=0 | step=17150/50000 progress=34.3% loss=0.101557 lr=7.830896e-05 lr_backbone=7.830896e-05 elapsed=17:50:06 avg_step=3.74s recent_step=2.94s eta=34:09:44 +2026-09-15 08:21:00 | INFO | train rank=0 | step=17160/50000 progress=34.3% loss=0.107356 lr=7.828169e-05 lr_backbone=7.828169e-05 elapsed=17:50:38 avg_step=3.74s recent_step=3.16s eta=34:08:55 +2026-09-15 08:21:49 | INFO | train rank=0 | step=17170/50000 progress=34.3% loss=0.092554 lr=7.825441e-05 lr_backbone=7.825441e-05 elapsed=17:51:27 avg_step=3.74s recent_step=4.98s eta=34:08:42 +2026-09-15 08:22:23 | INFO | train rank=0 | step=17180/50000 progress=34.4% loss=0.104626 lr=7.822712e-05 lr_backbone=7.822712e-05 elapsed=17:52:01 avg_step=3.74s recent_step=3.32s eta=34:07:56 +2026-09-15 08:23:01 | INFO | train rank=0 | step=17190/50000 progress=34.4% loss=0.112343 lr=7.819982e-05 lr_backbone=7.819982e-05 elapsed=17:52:39 avg_step=3.74s recent_step=3.86s eta=34:07:21 +2026-09-15 08:23:37 | INFO | train rank=0 | step=17200/50000 progress=34.4% loss=0.093271 lr=7.817251e-05 lr_backbone=7.817251e-05 elapsed=17:53:15 avg_step=3.74s recent_step=3.53s eta=34:06:39 +2026-09-15 08:24:26 | INFO | train rank=0 | step=17210/50000 progress=34.4% loss=0.119694 lr=7.814518e-05 lr_backbone=7.814518e-05 elapsed=17:54:04 avg_step=3.74s recent_step=4.93s eta=34:06:25 +2026-09-15 08:25:04 | INFO | train rank=0 | step=17220/50000 progress=34.4% loss=0.105578 lr=7.811784e-05 lr_backbone=7.811784e-05 elapsed=17:54:42 avg_step=3.74s recent_step=3.86s eta=34:05:49 +2026-09-15 08:25:33 | INFO | train rank=0 | step=17230/50000 progress=34.5% loss=0.108426 lr=7.809049e-05 lr_backbone=7.809049e-05 elapsed=17:55:11 avg_step=3.74s recent_step=2.88s eta=34:04:55 +2026-09-15 08:26:01 | INFO | train rank=0 | step=17240/50000 progress=34.5% loss=0.108846 lr=7.806313e-05 lr_backbone=7.806313e-05 elapsed=17:55:39 avg_step=3.74s recent_step=2.81s eta=34:04:00 +2026-09-15 08:26:56 | INFO | train rank=0 | step=17250/50000 progress=34.5% loss=0.111121 lr=7.803575e-05 lr_backbone=7.803575e-05 elapsed=17:56:34 avg_step=3.74s recent_step=5.48s eta=34:03:56 +2026-09-15 08:27:36 | INFO | train rank=0 | step=17260/50000 progress=34.5% loss=0.093385 lr=7.800836e-05 lr_backbone=7.800836e-05 elapsed=17:57:14 avg_step=3.74s recent_step=3.96s eta=34:03:22 +2026-09-15 08:28:06 | INFO | train rank=0 | step=17270/50000 progress=34.5% loss=0.103991 lr=7.798096e-05 lr_backbone=7.798096e-05 elapsed=17:57:44 avg_step=3.74s recent_step=3.01s eta=34:02:31 +2026-09-15 08:28:34 | INFO | train rank=0 | step=17280/50000 progress=34.6% loss=0.118649 lr=7.795355e-05 lr_backbone=7.795355e-05 elapsed=17:58:12 avg_step=3.74s recent_step=2.82s eta=34:01:36 +2026-09-15 08:29:21 | INFO | train rank=0 | step=17290/50000 progress=34.6% loss=0.104054 lr=7.792613e-05 lr_backbone=7.792613e-05 elapsed=17:58:59 avg_step=3.74s recent_step=4.67s eta=34:01:16 +2026-09-15 08:29:49 | INFO | train rank=0 | step=17300/50000 progress=34.6% loss=0.112354 lr=7.789869e-05 lr_backbone=7.789869e-05 elapsed=17:59:27 avg_step=3.74s recent_step=2.84s eta=34:00:22 +2026-09-15 08:30:18 | INFO | train rank=0 | step=17310/50000 progress=34.6% loss=0.089143 lr=7.787124e-05 lr_backbone=7.787124e-05 elapsed=17:59:56 avg_step=3.74s recent_step=2.89s eta=33:59:28 +2026-09-15 08:30:51 | INFO | train rank=0 | step=17320/50000 progress=34.6% loss=0.111117 lr=7.784378e-05 lr_backbone=7.784378e-05 elapsed=18:00:29 avg_step=3.74s recent_step=3.31s eta=33:58:42 +2026-09-15 08:31:40 | INFO | train rank=0 | step=17330/50000 progress=34.7% loss=0.095860 lr=7.781631e-05 lr_backbone=7.781631e-05 elapsed=18:01:18 avg_step=3.74s recent_step=4.87s eta=33:58:26 +2026-09-15 08:32:20 | INFO | train rank=0 | step=17340/50000 progress=34.7% loss=0.091828 lr=7.778882e-05 lr_backbone=7.778882e-05 elapsed=18:01:58 avg_step=3.74s recent_step=4.02s eta=33:57:54 +2026-09-15 08:32:53 | INFO | train rank=0 | step=17350/50000 progress=34.7% loss=0.153536 lr=7.776132e-05 lr_backbone=7.776132e-05 elapsed=18:02:31 avg_step=3.74s recent_step=3.25s eta=33:57:07 +2026-09-15 08:33:24 | INFO | train rank=0 | step=17360/50000 progress=34.7% loss=0.120158 lr=7.773381e-05 lr_backbone=7.773381e-05 elapsed=18:03:02 avg_step=3.74s recent_step=3.19s eta=33:56:19 +2026-09-15 08:34:06 | INFO | train rank=0 | step=17370/50000 progress=34.7% loss=0.096164 lr=7.770629e-05 lr_backbone=7.770629e-05 elapsed=18:03:44 avg_step=3.74s recent_step=4.20s eta=33:55:51 +2026-09-15 08:34:36 | INFO | train rank=0 | step=17380/50000 progress=34.8% loss=0.089181 lr=7.767876e-05 lr_backbone=7.767876e-05 elapsed=18:04:14 avg_step=3.74s recent_step=2.94s eta=33:54:58 +2026-09-15 08:35:11 | INFO | train rank=0 | step=17390/50000 progress=34.8% loss=0.118122 lr=7.765121e-05 lr_backbone=7.765121e-05 elapsed=18:04:49 avg_step=3.74s recent_step=3.48s eta=33:54:16 +2026-09-15 08:35:45 | INFO | train rank=0 | step=17400/50000 progress=34.8% loss=0.081276 lr=7.762365e-05 lr_backbone=7.762365e-05 elapsed=18:05:23 avg_step=3.74s recent_step=3.43s eta=33:53:32 +2026-09-15 08:36:35 | INFO | train rank=0 | step=17410/50000 progress=34.8% loss=0.101282 lr=7.759608e-05 lr_backbone=7.759608e-05 elapsed=18:06:13 avg_step=3.74s recent_step=4.99s eta=33:53:18 +2026-09-15 08:37:14 | INFO | train rank=0 | step=17420/50000 progress=34.8% loss=0.103264 lr=7.756850e-05 lr_backbone=7.756850e-05 elapsed=18:06:52 avg_step=3.74s recent_step=3.94s eta=33:52:45 +2026-09-15 08:37:51 | INFO | train rank=0 | step=17430/50000 progress=34.9% loss=0.082551 lr=7.754091e-05 lr_backbone=7.754091e-05 elapsed=18:07:29 avg_step=3.74s recent_step=3.69s eta=33:52:06 +2026-09-15 08:38:28 | INFO | train rank=0 | step=17440/50000 progress=34.9% loss=0.110982 lr=7.751330e-05 lr_backbone=7.751330e-05 elapsed=18:08:06 avg_step=3.74s recent_step=3.73s eta=33:51:28 +2026-09-15 08:39:16 | INFO | train rank=0 | step=17450/50000 progress=34.9% loss=0.100928 lr=7.748568e-05 lr_backbone=7.748568e-05 elapsed=18:08:54 avg_step=3.74s recent_step=4.76s eta=33:51:10 +2026-09-15 08:39:51 | INFO | train rank=0 | step=17460/50000 progress=34.9% loss=0.099639 lr=7.745805e-05 lr_backbone=7.745805e-05 elapsed=18:09:29 avg_step=3.74s recent_step=3.48s eta=33:50:27 +2026-09-15 08:40:24 | INFO | train rank=0 | step=17470/50000 progress=34.9% loss=0.105823 lr=7.743041e-05 lr_backbone=7.743041e-05 elapsed=18:10:02 avg_step=3.74s recent_step=3.36s eta=33:49:43 +2026-09-15 08:41:00 | INFO | train rank=0 | step=17480/50000 progress=35.0% loss=0.101162 lr=7.740275e-05 lr_backbone=7.740275e-05 elapsed=18:10:38 avg_step=3.74s recent_step=3.53s eta=33:49:01 +2026-09-15 08:41:51 | INFO | train rank=0 | step=17490/50000 progress=35.0% loss=0.102460 lr=7.737509e-05 lr_backbone=7.737509e-05 elapsed=18:11:29 avg_step=3.74s recent_step=5.18s eta=33:48:51 +2026-09-15 08:42:28 | INFO | train rank=0 | step=17500/50000 progress=35.0% loss=0.094360 lr=7.734741e-05 lr_backbone=7.734741e-05 elapsed=18:12:06 avg_step=3.74s recent_step=3.68s eta=33:48:12 +2026-09-15 08:42:55 | INFO | train rank=0 | step=17510/50000 progress=35.0% loss=0.092752 lr=7.731972e-05 lr_backbone=7.731972e-05 elapsed=18:12:33 avg_step=3.74s recent_step=2.71s eta=33:47:15 +2026-09-15 08:43:29 | INFO | train rank=0 | step=17520/50000 progress=35.0% loss=0.101191 lr=7.729201e-05 lr_backbone=7.729201e-05 elapsed=18:13:07 avg_step=3.74s recent_step=3.34s eta=33:46:30 +2026-09-15 08:44:19 | INFO | train rank=0 | step=17530/50000 progress=35.1% loss=0.087932 lr=7.726430e-05 lr_backbone=7.726430e-05 elapsed=18:13:57 avg_step=3.74s recent_step=5.00s eta=33:46:16 +2026-09-15 08:44:55 | INFO | train rank=0 | step=17540/50000 progress=35.1% loss=0.128620 lr=7.723657e-05 lr_backbone=7.723657e-05 elapsed=18:14:33 avg_step=3.74s recent_step=3.63s eta=33:45:37 +2026-09-15 08:45:22 | INFO | train rank=0 | step=17550/50000 progress=35.1% loss=0.094212 lr=7.720884e-05 lr_backbone=7.720884e-05 elapsed=18:15:00 avg_step=3.74s recent_step=2.74s eta=33:44:41 +2026-09-15 08:46:01 | INFO | train rank=0 | step=17560/50000 progress=35.1% loss=0.111702 lr=7.718109e-05 lr_backbone=7.718109e-05 elapsed=18:15:39 avg_step=3.74s recent_step=3.83s eta=33:44:05 +2026-09-15 08:46:38 | INFO | train rank=0 | step=17570/50000 progress=35.1% loss=0.083885 lr=7.715332e-05 lr_backbone=7.715332e-05 elapsed=18:16:16 avg_step=3.74s recent_step=3.70s eta=33:43:27 +2026-09-15 08:47:15 | INFO | train rank=0 | step=17580/50000 progress=35.2% loss=0.113292 lr=7.712555e-05 lr_backbone=7.712555e-05 elapsed=18:16:53 avg_step=3.74s recent_step=3.69s eta=33:42:48 +2026-09-15 08:47:48 | INFO | train rank=0 | step=17590/50000 progress=35.2% loss=0.093262 lr=7.709776e-05 lr_backbone=7.709776e-05 elapsed=18:17:26 avg_step=3.74s recent_step=3.37s eta=33:42:04 +2026-09-15 08:48:22 | INFO | train rank=0 | step=17600/50000 progress=35.2% loss=0.107402 lr=7.706997e-05 lr_backbone=7.706997e-05 elapsed=18:18:00 avg_step=3.74s recent_step=3.34s eta=33:41:19 +2026-09-15 08:49:10 | INFO | train rank=0 | step=17610/50000 progress=35.2% loss=0.099132 lr=7.704216e-05 lr_backbone=7.704216e-05 elapsed=18:18:48 avg_step=3.74s recent_step=4.85s eta=33:41:02 +2026-09-15 08:49:50 | INFO | train rank=0 | step=17620/50000 progress=35.2% loss=0.088104 lr=7.701434e-05 lr_backbone=7.701434e-05 elapsed=18:19:28 avg_step=3.74s recent_step=3.95s eta=33:40:28 +2026-09-15 08:50:17 | INFO | train rank=0 | step=17630/50000 progress=35.3% loss=0.124313 lr=7.698650e-05 lr_backbone=7.698650e-05 elapsed=18:19:56 avg_step=3.74s recent_step=2.79s eta=33:39:33 +2026-09-15 08:50:54 | INFO | train rank=0 | step=17640/50000 progress=35.3% loss=0.128233 lr=7.695866e-05 lr_backbone=7.695866e-05 elapsed=18:20:32 avg_step=3.74s recent_step=3.66s eta=33:38:54 +2026-09-15 08:51:36 | INFO | train rank=0 | step=17650/50000 progress=35.3% loss=0.094477 lr=7.693080e-05 lr_backbone=7.693080e-05 elapsed=18:21:14 avg_step=3.74s recent_step=4.19s eta=33:38:25 +2026-09-15 08:52:04 | INFO | train rank=0 | step=17660/50000 progress=35.3% loss=0.094392 lr=7.690293e-05 lr_backbone=7.690293e-05 elapsed=18:21:42 avg_step=3.74s recent_step=2.84s eta=33:37:31 +2026-09-15 08:52:36 | INFO | train rank=0 | step=17670/50000 progress=35.3% loss=0.090676 lr=7.687505e-05 lr_backbone=7.687505e-05 elapsed=18:22:14 avg_step=3.74s recent_step=3.14s eta=33:36:43 +2026-09-15 08:53:09 | INFO | train rank=0 | step=17680/50000 progress=35.4% loss=0.088736 lr=7.684716e-05 lr_backbone=7.684716e-05 elapsed=18:22:47 avg_step=3.74s recent_step=3.28s eta=33:35:57 +2026-09-15 08:54:03 | INFO | train rank=0 | step=17690/50000 progress=35.4% loss=0.101429 lr=7.681926e-05 lr_backbone=7.681926e-05 elapsed=18:23:41 avg_step=3.74s recent_step=5.47s eta=33:35:51 +2026-09-15 08:54:39 | INFO | train rank=0 | step=17700/50000 progress=35.4% loss=0.100995 lr=7.679134e-05 lr_backbone=7.679134e-05 elapsed=18:24:17 avg_step=3.74s recent_step=3.56s eta=33:35:10 +2026-09-15 08:55:05 | INFO | train rank=0 | step=17710/50000 progress=35.4% loss=0.119571 lr=7.676341e-05 lr_backbone=7.676341e-05 elapsed=18:24:43 avg_step=3.74s recent_step=2.64s eta=33:34:12 +2026-09-15 08:55:41 | INFO | train rank=0 | step=17720/50000 progress=35.4% loss=0.094800 lr=7.673547e-05 lr_backbone=7.673547e-05 elapsed=18:25:19 avg_step=3.74s recent_step=3.58s eta=33:33:32 +2026-09-15 08:56:28 | INFO | train rank=0 | step=17730/50000 progress=35.5% loss=0.115144 lr=7.670752e-05 lr_backbone=7.670752e-05 elapsed=18:26:06 avg_step=3.74s recent_step=4.70s eta=33:33:12 +2026-09-15 08:56:59 | INFO | train rank=0 | step=17740/50000 progress=35.5% loss=0.103350 lr=7.667956e-05 lr_backbone=7.667956e-05 elapsed=18:26:37 avg_step=3.74s recent_step=3.07s eta=33:32:22 +2026-09-15 08:57:25 | INFO | train rank=0 | step=17750/50000 progress=35.5% loss=0.097954 lr=7.665159e-05 lr_backbone=7.665159e-05 elapsed=18:27:03 avg_step=3.74s recent_step=2.58s eta=33:31:24 +2026-09-15 08:58:01 | INFO | train rank=0 | step=17760/50000 progress=35.5% loss=0.119717 lr=7.662360e-05 lr_backbone=7.662360e-05 elapsed=18:27:39 avg_step=3.74s recent_step=3.61s eta=33:30:44 +2026-09-15 08:58:57 | INFO | train rank=0 | step=17770/50000 progress=35.5% loss=0.084724 lr=7.659560e-05 lr_backbone=7.659560e-05 elapsed=18:28:35 avg_step=3.74s recent_step=5.58s eta=33:30:40 +2026-09-15 08:59:33 | INFO | train rank=0 | step=17780/50000 progress=35.6% loss=0.128566 lr=7.656759e-05 lr_backbone=7.656759e-05 elapsed=18:29:11 avg_step=3.74s recent_step=3.61s eta=33:30:00 +2026-09-15 09:00:10 | INFO | train rank=0 | step=17790/50000 progress=35.6% loss=0.089266 lr=7.653957e-05 lr_backbone=7.653957e-05 elapsed=18:29:48 avg_step=3.74s recent_step=3.71s eta=33:29:22 +2026-09-15 09:00:34 | INFO | train rank=0 | step=17800/50000 progress=35.6% loss=0.115276 lr=7.651154e-05 lr_backbone=7.651154e-05 elapsed=18:30:12 avg_step=3.74s recent_step=2.43s eta=33:28:21 +2026-09-15 09:01:25 | INFO | train rank=0 | step=17810/50000 progress=35.6% loss=0.092972 lr=7.648350e-05 lr_backbone=7.648350e-05 elapsed=18:31:03 avg_step=3.74s recent_step=5.06s eta=33:28:07 +2026-09-15 09:01:53 | INFO | train rank=0 | step=17820/50000 progress=35.6% loss=0.111728 lr=7.645544e-05 lr_backbone=7.645544e-05 elapsed=18:31:31 avg_step=3.74s recent_step=2.85s eta=33:27:14 +2026-09-15 09:02:28 | INFO | train rank=0 | step=17830/50000 progress=35.7% loss=0.108458 lr=7.642738e-05 lr_backbone=7.642738e-05 elapsed=18:32:06 avg_step=3.74s recent_step=3.51s eta=33:26:32 +2026-09-15 09:03:05 | INFO | train rank=0 | step=17840/50000 progress=35.7% loss=0.108730 lr=7.639930e-05 lr_backbone=7.639930e-05 elapsed=18:32:43 avg_step=3.74s recent_step=3.61s eta=33:25:53 +2026-09-15 09:03:50 | INFO | train rank=0 | step=17850/50000 progress=35.7% loss=0.086222 lr=7.637121e-05 lr_backbone=7.637121e-05 elapsed=18:33:28 avg_step=3.74s recent_step=4.52s eta=33:25:29 +2026-09-15 09:04:20 | INFO | train rank=0 | step=17860/50000 progress=35.7% loss=0.087791 lr=7.634311e-05 lr_backbone=7.634311e-05 elapsed=18:33:58 avg_step=3.74s recent_step=2.99s eta=33:24:38 +2026-09-15 09:04:51 | INFO | train rank=0 | step=17870/50000 progress=35.7% loss=0.103265 lr=7.631499e-05 lr_backbone=7.631499e-05 elapsed=18:34:29 avg_step=3.74s recent_step=3.17s eta=33:23:50 +2026-09-15 09:05:34 | INFO | train rank=0 | step=17880/50000 progress=35.8% loss=0.086454 lr=7.628687e-05 lr_backbone=7.628687e-05 elapsed=18:35:12 avg_step=3.74s recent_step=4.27s eta=33:23:22 +2026-09-15 09:06:14 | INFO | train rank=0 | step=17890/50000 progress=35.8% loss=0.094109 lr=7.625873e-05 lr_backbone=7.625873e-05 elapsed=18:35:52 avg_step=3.74s recent_step=3.98s eta=33:22:49 +2026-09-15 09:06:48 | INFO | train rank=0 | step=17900/50000 progress=35.8% loss=0.097534 lr=7.623058e-05 lr_backbone=7.623058e-05 elapsed=18:36:26 avg_step=3.74s recent_step=3.39s eta=33:22:05 +2026-09-15 09:07:20 | INFO | train rank=0 | step=17910/50000 progress=35.8% loss=0.093399 lr=7.620242e-05 lr_backbone=7.620242e-05 elapsed=18:36:58 avg_step=3.74s recent_step=3.29s eta=33:21:20 +2026-09-15 09:07:59 | INFO | train rank=0 | step=17920/50000 progress=35.8% loss=0.099511 lr=7.617425e-05 lr_backbone=7.617425e-05 elapsed=18:37:37 avg_step=3.74s recent_step=3.86s eta=33:20:45 +2026-09-15 09:08:46 | INFO | train rank=0 | step=17930/50000 progress=35.9% loss=0.104340 lr=7.614607e-05 lr_backbone=7.614607e-05 elapsed=18:38:24 avg_step=3.74s recent_step=4.69s eta=33:20:24 +2026-09-15 09:09:15 | INFO | train rank=0 | step=17940/50000 progress=35.9% loss=0.093516 lr=7.611788e-05 lr_backbone=7.611788e-05 elapsed=18:38:53 avg_step=3.74s recent_step=2.86s eta=33:19:31 +2026-09-15 09:09:52 | INFO | train rank=0 | step=17950/50000 progress=35.9% loss=0.110832 lr=7.608967e-05 lr_backbone=7.608967e-05 elapsed=18:39:30 avg_step=3.74s recent_step=3.74s eta=33:18:53 +2026-09-15 09:10:29 | INFO | train rank=0 | step=17960/50000 progress=35.9% loss=0.117745 lr=7.606146e-05 lr_backbone=7.606146e-05 elapsed=18:40:07 avg_step=3.74s recent_step=3.68s eta=33:18:15 +2026-09-15 09:11:24 | INFO | train rank=0 | step=17970/50000 progress=35.9% loss=0.090551 lr=7.603323e-05 lr_backbone=7.603323e-05 elapsed=18:41:02 avg_step=3.74s recent_step=5.50s eta=33:18:09 +2026-09-15 09:12:00 | INFO | train rank=0 | step=17980/50000 progress=36.0% loss=0.085529 lr=7.600499e-05 lr_backbone=7.600499e-05 elapsed=18:41:38 avg_step=3.74s recent_step=3.61s eta=33:17:29 +2026-09-15 09:12:25 | INFO | train rank=0 | step=17990/50000 progress=36.0% loss=0.079703 lr=7.597674e-05 lr_backbone=7.597674e-05 elapsed=18:42:03 avg_step=3.74s recent_step=2.53s eta=33:16:30 +2026-09-15 09:13:09 | INFO | train rank=0 | step=18000/50000 progress=36.0% loss=0.078031 lr=7.594848e-05 lr_backbone=7.594848e-05 elapsed=18:42:47 avg_step=3.74s recent_step=4.41s eta=33:16:04 +2026-09-15 09:13:52 | INFO | train rank=0 | step=18010/50000 progress=36.0% loss=0.112497 lr=7.592021e-05 lr_backbone=7.592021e-05 elapsed=18:43:30 avg_step=3.74s recent_step=4.28s eta=33:15:36 +2026-09-15 09:14:20 | INFO | train rank=0 | step=18020/50000 progress=36.0% loss=0.121681 lr=7.589192e-05 lr_backbone=7.589192e-05 elapsed=18:43:58 avg_step=3.74s recent_step=2.85s eta=33:14:43 +2026-09-15 09:14:57 | INFO | train rank=0 | step=18030/50000 progress=36.1% loss=0.101515 lr=7.586363e-05 lr_backbone=7.586363e-05 elapsed=18:44:35 avg_step=3.74s recent_step=3.66s eta=33:14:04 +2026-09-15 09:15:56 | INFO | train rank=0 | step=18040/50000 progress=36.1% loss=0.113766 lr=7.583532e-05 lr_backbone=7.583532e-05 elapsed=18:45:34 avg_step=3.74s recent_step=5.89s eta=33:14:05 +2026-09-15 09:16:25 | INFO | train rank=0 | step=18050/50000 progress=36.1% loss=0.107525 lr=7.580700e-05 lr_backbone=7.580700e-05 elapsed=18:46:03 avg_step=3.74s recent_step=2.86s eta=33:13:12 +2026-09-15 09:17:03 | INFO | train rank=0 | step=18060/50000 progress=36.1% loss=0.087837 lr=7.577867e-05 lr_backbone=7.577867e-05 elapsed=18:46:41 avg_step=3.74s recent_step=3.84s eta=33:12:36 +2026-09-15 09:17:35 | INFO | train rank=0 | step=18070/50000 progress=36.1% loss=0.081641 lr=7.575033e-05 lr_backbone=7.575033e-05 elapsed=18:47:13 avg_step=3.74s recent_step=3.20s eta=33:11:49 +2026-09-15 09:18:12 | INFO | train rank=0 | step=18080/50000 progress=36.2% loss=0.094886 lr=7.572198e-05 lr_backbone=7.572198e-05 elapsed=18:47:50 avg_step=3.74s recent_step=3.69s eta=33:11:11 +2026-09-15 09:18:42 | INFO | train rank=0 | step=18090/50000 progress=36.2% loss=0.130216 lr=7.569361e-05 lr_backbone=7.569361e-05 elapsed=18:48:20 avg_step=3.74s recent_step=3.05s eta=33:10:21 +2026-09-15 09:19:20 | INFO | train rank=0 | step=18100/50000 progress=36.2% loss=0.097825 lr=7.566524e-05 lr_backbone=7.566524e-05 elapsed=18:48:58 avg_step=3.74s recent_step=3.72s eta=33:09:43 +2026-09-15 09:19:56 | INFO | train rank=0 | step=18110/50000 progress=36.2% loss=0.092295 lr=7.563685e-05 lr_backbone=7.563685e-05 elapsed=18:49:34 avg_step=3.74s recent_step=3.67s eta=33:09:04 +2026-09-15 09:20:46 | INFO | train rank=0 | step=18120/50000 progress=36.2% loss=0.100908 lr=7.560846e-05 lr_backbone=7.560846e-05 elapsed=18:50:24 avg_step=3.74s recent_step=5.02s eta=33:08:50 +2026-09-15 09:21:24 | INFO | train rank=0 | step=18130/50000 progress=36.3% loss=0.091964 lr=7.558005e-05 lr_backbone=7.558005e-05 elapsed=18:51:02 avg_step=3.74s recent_step=3.72s eta=33:08:12 +2026-09-15 09:21:58 | INFO | train rank=0 | step=18140/50000 progress=36.3% loss=0.085156 lr=7.555163e-05 lr_backbone=7.555163e-05 elapsed=18:51:36 avg_step=3.74s recent_step=3.42s eta=33:07:29 +2026-09-15 09:22:26 | INFO | train rank=0 | step=18150/50000 progress=36.3% loss=0.125207 lr=7.552320e-05 lr_backbone=7.552320e-05 elapsed=18:52:04 avg_step=3.74s recent_step=2.80s eta=33:06:35 +2026-09-15 09:23:06 | INFO | train rank=0 | step=18160/50000 progress=36.3% loss=0.108982 lr=7.549475e-05 lr_backbone=7.549475e-05 elapsed=18:52:44 avg_step=3.74s recent_step=3.99s eta=33:06:02 +2026-09-15 09:23:38 | INFO | train rank=0 | step=18170/50000 progress=36.3% loss=0.100793 lr=7.546630e-05 lr_backbone=7.546630e-05 elapsed=18:53:16 avg_step=3.74s recent_step=3.18s eta=33:05:14 +2026-09-15 09:24:13 | INFO | train rank=0 | step=18180/50000 progress=36.4% loss=0.077487 lr=7.543784e-05 lr_backbone=7.543784e-05 elapsed=18:53:51 avg_step=3.74s recent_step=3.53s eta=33:04:33 +2026-09-15 09:24:54 | INFO | train rank=0 | step=18190/50000 progress=36.4% loss=0.108380 lr=7.540936e-05 lr_backbone=7.540936e-05 elapsed=18:54:32 avg_step=3.74s recent_step=4.09s eta=33:04:02 +2026-09-15 09:25:30 | INFO | train rank=0 | step=18200/50000 progress=36.4% loss=0.118785 lr=7.538088e-05 lr_backbone=7.538088e-05 elapsed=18:55:08 avg_step=3.74s recent_step=3.60s eta=33:03:22 +2026-09-15 09:26:07 | INFO | train rank=0 | step=18210/50000 progress=36.4% loss=0.114635 lr=7.535238e-05 lr_backbone=7.535238e-05 elapsed=18:55:45 avg_step=3.74s recent_step=3.72s eta=33:02:44 +2026-09-15 09:26:35 | INFO | train rank=0 | step=18220/50000 progress=36.4% loss=0.101711 lr=7.532387e-05 lr_backbone=7.532387e-05 elapsed=18:56:13 avg_step=3.74s recent_step=2.77s eta=33:01:50 +2026-09-15 09:27:07 | INFO | train rank=0 | step=18230/50000 progress=36.5% loss=0.118114 lr=7.529535e-05 lr_backbone=7.529535e-05 elapsed=18:56:45 avg_step=3.74s recent_step=3.29s eta=33:01:04 +2026-09-15 09:27:53 | INFO | train rank=0 | step=18240/50000 progress=36.5% loss=0.120881 lr=7.526682e-05 lr_backbone=7.526682e-05 elapsed=18:57:31 avg_step=3.74s recent_step=4.58s eta=33:00:41 +2026-09-15 09:28:34 | INFO | train rank=0 | step=18250/50000 progress=36.5% loss=0.120281 lr=7.523828e-05 lr_backbone=7.523828e-05 elapsed=18:58:12 avg_step=3.74s recent_step=4.06s eta=33:00:09 +2026-09-15 09:29:02 | INFO | train rank=0 | step=18260/50000 progress=36.5% loss=0.093252 lr=7.520972e-05 lr_backbone=7.520972e-05 elapsed=18:58:40 avg_step=3.74s recent_step=2.78s eta=32:59:15 +2026-09-15 09:29:50 | INFO | train rank=0 | step=18270/50000 progress=36.5% loss=0.107322 lr=7.518116e-05 lr_backbone=7.518116e-05 elapsed=18:59:28 avg_step=3.74s recent_step=4.86s eta=32:58:57 +2026-09-15 09:30:27 | INFO | train rank=0 | step=18280/50000 progress=36.6% loss=0.106605 lr=7.515259e-05 lr_backbone=7.515259e-05 elapsed=19:00:05 avg_step=3.74s recent_step=3.63s eta=32:58:18 +2026-09-15 09:31:09 | INFO | train rank=0 | step=18290/50000 progress=36.6% loss=0.095919 lr=7.512400e-05 lr_backbone=7.512400e-05 elapsed=19:00:47 avg_step=3.74s recent_step=4.28s eta=32:57:50 +2026-09-15 09:31:41 | INFO | train rank=0 | step=18300/50000 progress=36.6% loss=0.110202 lr=7.509540e-05 lr_backbone=7.509540e-05 elapsed=19:01:19 avg_step=3.74s recent_step=3.20s eta=32:57:03 +2026-09-15 09:32:11 | INFO | train rank=0 | step=18310/50000 progress=36.6% loss=0.110642 lr=7.506679e-05 lr_backbone=7.506679e-05 elapsed=19:01:49 avg_step=3.74s recent_step=3.00s eta=32:56:13 +2026-09-15 09:32:48 | INFO | train rank=0 | step=18320/50000 progress=36.6% loss=0.100846 lr=7.503818e-05 lr_backbone=7.503818e-05 elapsed=19:02:26 avg_step=3.74s recent_step=3.64s eta=32:55:34 +2026-09-15 09:33:28 | INFO | train rank=0 | step=18330/50000 progress=36.7% loss=0.101094 lr=7.500955e-05 lr_backbone=7.500955e-05 elapsed=19:03:06 avg_step=3.74s recent_step=4.05s eta=32:55:02 +2026-09-15 09:34:04 | INFO | train rank=0 | step=18340/50000 progress=36.7% loss=0.092529 lr=7.498090e-05 lr_backbone=7.498090e-05 elapsed=19:03:42 avg_step=3.74s recent_step=3.56s eta=32:54:21 +2026-09-15 09:34:41 | INFO | train rank=0 | step=18350/50000 progress=36.7% loss=0.091165 lr=7.495225e-05 lr_backbone=7.495225e-05 elapsed=19:04:19 avg_step=3.74s recent_step=3.75s eta=32:53:44 +2026-09-15 09:35:25 | INFO | train rank=0 | step=18360/50000 progress=36.7% loss=0.130548 lr=7.492359e-05 lr_backbone=7.492359e-05 elapsed=19:05:03 avg_step=3.74s recent_step=4.39s eta=32:53:18 +2026-09-15 09:36:07 | INFO | train rank=0 | step=18370/50000 progress=36.7% loss=0.089731 lr=7.489492e-05 lr_backbone=7.489492e-05 elapsed=19:05:45 avg_step=3.74s recent_step=4.22s eta=32:52:48 +2026-09-15 09:36:40 | INFO | train rank=0 | step=18380/50000 progress=36.8% loss=0.126640 lr=7.486623e-05 lr_backbone=7.486623e-05 elapsed=19:06:18 avg_step=3.74s recent_step=3.26s eta=32:52:03 +2026-09-15 09:37:16 | INFO | train rank=0 | step=18390/50000 progress=36.8% loss=0.117999 lr=7.483754e-05 lr_backbone=7.483754e-05 elapsed=19:06:54 avg_step=3.74s recent_step=3.60s eta=32:51:23 +2026-09-15 09:37:57 | INFO | train rank=0 | step=18400/50000 progress=36.8% loss=0.103486 lr=7.480883e-05 lr_backbone=7.480883e-05 elapsed=19:07:35 avg_step=3.74s recent_step=4.07s eta=32:50:51 +2026-09-15 09:38:34 | INFO | train rank=0 | step=18410/50000 progress=36.8% loss=0.102465 lr=7.478011e-05 lr_backbone=7.478011e-05 elapsed=19:08:12 avg_step=3.74s recent_step=3.68s eta=32:50:12 +2026-09-15 09:39:06 | INFO | train rank=0 | step=18420/50000 progress=36.8% loss=0.099340 lr=7.475139e-05 lr_backbone=7.475139e-05 elapsed=19:08:44 avg_step=3.74s recent_step=3.29s eta=32:49:27 +2026-09-15 09:39:40 | INFO | train rank=0 | step=18430/50000 progress=36.9% loss=0.090164 lr=7.472265e-05 lr_backbone=7.472265e-05 elapsed=19:09:18 avg_step=3.74s recent_step=3.40s eta=32:48:44 +2026-09-15 09:40:22 | INFO | train rank=0 | step=18440/50000 progress=36.9% loss=0.107836 lr=7.469390e-05 lr_backbone=7.469390e-05 elapsed=19:10:00 avg_step=3.74s recent_step=4.13s eta=32:48:13 +2026-09-15 09:41:10 | INFO | train rank=0 | step=18450/50000 progress=36.9% loss=0.098636 lr=7.466514e-05 lr_backbone=7.466514e-05 elapsed=19:10:48 avg_step=3.74s recent_step=4.79s eta=32:47:54 +2026-09-15 09:41:48 | INFO | train rank=0 | step=18460/50000 progress=36.9% loss=0.081168 lr=7.463637e-05 lr_backbone=7.463637e-05 elapsed=19:11:26 avg_step=3.74s recent_step=3.84s eta=32:47:18 +2026-09-15 09:42:20 | INFO | train rank=0 | step=18470/50000 progress=36.9% loss=0.122853 lr=7.460759e-05 lr_backbone=7.460759e-05 elapsed=19:11:58 avg_step=3.74s recent_step=3.23s eta=32:46:32 +2026-09-15 09:42:50 | INFO | train rank=0 | step=18480/50000 progress=37.0% loss=0.092574 lr=7.457879e-05 lr_backbone=7.457879e-05 elapsed=19:12:28 avg_step=3.74s recent_step=2.94s eta=32:45:41 +2026-09-15 09:43:40 | INFO | train rank=0 | step=18490/50000 progress=37.0% loss=0.086099 lr=7.454999e-05 lr_backbone=7.454999e-05 elapsed=19:13:18 avg_step=3.74s recent_step=5.04s eta=32:45:25 +2026-09-15 09:44:13 | INFO | train rank=0 | step=18500/50000 progress=37.0% loss=0.093121 lr=7.452118e-05 lr_backbone=7.452118e-05 elapsed=19:13:51 avg_step=3.74s recent_step=3.25s eta=32:44:39 +2026-09-15 09:44:47 | INFO | train rank=0 | step=18510/50000 progress=37.0% loss=0.093524 lr=7.449235e-05 lr_backbone=7.449235e-05 elapsed=19:14:25 avg_step=3.74s recent_step=3.47s eta=32:43:57 +2026-09-15 09:45:20 | INFO | train rank=0 | step=18520/50000 progress=37.0% loss=0.106227 lr=7.446351e-05 lr_backbone=7.446351e-05 elapsed=19:14:58 avg_step=3.74s recent_step=3.27s eta=32:43:12 +2026-09-15 09:46:10 | INFO | train rank=0 | step=18530/50000 progress=37.1% loss=0.105288 lr=7.443467e-05 lr_backbone=7.443467e-05 elapsed=19:15:49 avg_step=3.74s recent_step=5.05s eta=32:42:57 +2026-09-15 09:46:44 | INFO | train rank=0 | step=18540/50000 progress=37.1% loss=0.097673 lr=7.440581e-05 lr_backbone=7.440581e-05 elapsed=19:16:22 avg_step=3.74s recent_step=3.40s eta=32:42:14 +2026-09-15 09:47:21 | INFO | train rank=0 | step=18550/50000 progress=37.1% loss=0.090383 lr=7.437694e-05 lr_backbone=7.437694e-05 elapsed=19:16:59 avg_step=3.74s recent_step=3.61s eta=32:41:34 +2026-09-15 09:47:56 | INFO | train rank=0 | step=18560/50000 progress=37.1% loss=0.104116 lr=7.434807e-05 lr_backbone=7.434807e-05 elapsed=19:17:34 avg_step=3.74s recent_step=3.52s eta=32:40:53 +2026-09-15 09:48:40 | INFO | train rank=0 | step=18570/50000 progress=37.1% loss=0.097532 lr=7.431918e-05 lr_backbone=7.431918e-05 elapsed=19:18:18 avg_step=3.74s recent_step=4.44s eta=32:40:27 +2026-09-15 09:49:19 | INFO | train rank=0 | step=18580/50000 progress=37.2% loss=0.093636 lr=7.429028e-05 lr_backbone=7.429028e-05 elapsed=19:18:57 avg_step=3.74s recent_step=3.89s eta=32:39:52 +2026-09-15 09:49:53 | INFO | train rank=0 | step=18590/50000 progress=37.2% loss=0.136818 lr=7.426137e-05 lr_backbone=7.426137e-05 elapsed=19:19:31 avg_step=3.74s recent_step=3.35s eta=32:39:08 +2026-09-15 09:50:26 | INFO | train rank=0 | step=18600/50000 progress=37.2% loss=0.087130 lr=7.423245e-05 lr_backbone=7.423245e-05 elapsed=19:20:04 avg_step=3.74s recent_step=3.31s eta=32:38:23 +2026-09-15 09:51:05 | INFO | train rank=0 | step=18610/50000 progress=37.2% loss=0.091186 lr=7.420352e-05 lr_backbone=7.420352e-05 elapsed=19:20:43 avg_step=3.74s recent_step=3.89s eta=32:37:48 +2026-09-15 09:51:38 | INFO | train rank=0 | step=18620/50000 progress=37.2% loss=0.087357 lr=7.417457e-05 lr_backbone=7.417457e-05 elapsed=19:21:16 avg_step=3.74s recent_step=3.35s eta=32:37:04 +2026-09-15 09:52:13 | INFO | train rank=0 | step=18630/50000 progress=37.3% loss=0.093587 lr=7.414562e-05 lr_backbone=7.414562e-05 elapsed=19:21:51 avg_step=3.74s recent_step=3.46s eta=32:36:22 +2026-09-15 09:52:48 | INFO | train rank=0 | step=18640/50000 progress=37.3% loss=0.111661 lr=7.411666e-05 lr_backbone=7.411666e-05 elapsed=19:22:26 avg_step=3.74s recent_step=3.55s eta=32:35:42 +2026-09-15 09:53:32 | INFO | train rank=0 | step=18650/50000 progress=37.3% loss=0.078633 lr=7.408768e-05 lr_backbone=7.408768e-05 elapsed=19:23:10 avg_step=3.74s recent_step=4.34s eta=32:35:14 +2026-09-15 09:54:08 | INFO | train rank=0 | step=18660/50000 progress=37.3% loss=0.096185 lr=7.405870e-05 lr_backbone=7.405870e-05 elapsed=19:23:46 avg_step=3.74s recent_step=3.59s eta=32:34:34 +2026-09-15 09:54:51 | INFO | train rank=0 | step=18670/50000 progress=37.3% loss=0.117232 lr=7.402970e-05 lr_backbone=7.402970e-05 elapsed=19:24:29 avg_step=3.74s recent_step=4.36s eta=32:34:07 +2026-09-15 09:55:31 | INFO | train rank=0 | step=18680/50000 progress=37.4% loss=0.094444 lr=7.400070e-05 lr_backbone=7.400070e-05 elapsed=19:25:09 avg_step=3.74s recent_step=4.02s eta=32:33:35 +2026-09-15 09:56:02 | INFO | train rank=0 | step=18690/50000 progress=37.4% loss=0.113288 lr=7.397168e-05 lr_backbone=7.397168e-05 elapsed=19:25:40 avg_step=3.74s recent_step=3.10s eta=32:32:46 +2026-09-15 09:56:36 | INFO | train rank=0 | step=18700/50000 progress=37.4% loss=0.135435 lr=7.394266e-05 lr_backbone=7.394266e-05 elapsed=19:26:14 avg_step=3.74s recent_step=3.40s eta=32:32:03 +2026-09-15 09:57:24 | INFO | train rank=0 | step=18710/50000 progress=37.4% loss=0.109427 lr=7.391362e-05 lr_backbone=7.391362e-05 elapsed=19:27:02 avg_step=3.74s recent_step=4.77s eta=32:31:43 +2026-09-15 09:57:55 | INFO | train rank=0 | step=18720/50000 progress=37.4% loss=0.104149 lr=7.388457e-05 lr_backbone=7.388457e-05 elapsed=19:27:33 avg_step=3.74s recent_step=3.13s eta=32:30:55 +2026-09-15 09:58:33 | INFO | train rank=0 | step=18730/50000 progress=37.5% loss=0.089743 lr=7.385552e-05 lr_backbone=7.385552e-05 elapsed=19:28:11 avg_step=3.74s recent_step=3.80s eta=32:30:19 +2026-09-15 09:59:05 | INFO | train rank=0 | step=18740/50000 progress=37.5% loss=0.091341 lr=7.382645e-05 lr_backbone=7.382645e-05 elapsed=19:28:43 avg_step=3.74s recent_step=3.14s eta=32:29:32 +2026-09-15 09:59:38 | INFO | train rank=0 | step=18750/50000 progress=37.5% loss=0.104493 lr=7.379737e-05 lr_backbone=7.379737e-05 elapsed=19:29:16 avg_step=3.74s recent_step=3.35s eta=32:28:48 +2026-09-15 10:00:23 | INFO | train rank=0 | step=18760/50000 progress=37.5% loss=0.095625 lr=7.376828e-05 lr_backbone=7.376828e-05 elapsed=19:30:01 avg_step=3.74s recent_step=4.50s eta=32:28:23 +2026-09-15 10:01:05 | INFO | train rank=0 | step=18770/50000 progress=37.5% loss=0.078098 lr=7.373918e-05 lr_backbone=7.373918e-05 elapsed=19:30:43 avg_step=3.74s recent_step=4.16s eta=32:27:52 +2026-09-15 10:01:40 | INFO | train rank=0 | step=18780/50000 progress=37.6% loss=0.098945 lr=7.371007e-05 lr_backbone=7.371007e-05 elapsed=19:31:18 avg_step=3.74s recent_step=3.50s eta=32:27:11 +2026-09-15 10:02:18 | INFO | train rank=0 | step=18790/50000 progress=37.6% loss=0.123486 lr=7.368095e-05 lr_backbone=7.368095e-05 elapsed=19:31:56 avg_step=3.74s recent_step=3.76s eta=32:26:34 +2026-09-15 10:02:51 | INFO | train rank=0 | step=18800/50000 progress=37.6% loss=0.101243 lr=7.365182e-05 lr_backbone=7.365182e-05 elapsed=19:32:29 avg_step=3.74s recent_step=3.34s eta=32:25:50 +2026-09-15 10:03:20 | INFO | train rank=0 | step=18810/50000 progress=37.6% loss=0.095283 lr=7.362268e-05 lr_backbone=7.362268e-05 elapsed=19:32:58 avg_step=3.74s recent_step=2.92s eta=32:24:59 +2026-09-15 10:04:13 | INFO | train rank=0 | step=18820/50000 progress=37.6% loss=0.106648 lr=7.359353e-05 lr_backbone=7.359353e-05 elapsed=19:33:51 avg_step=3.74s recent_step=5.26s eta=32:24:46 +2026-09-15 10:04:37 | INFO | train rank=0 | step=18830/50000 progress=37.7% loss=0.109522 lr=7.356437e-05 lr_backbone=7.356437e-05 elapsed=19:34:15 avg_step=3.74s recent_step=2.39s eta=32:23:46 +2026-09-15 10:05:08 | INFO | train rank=0 | step=18840/50000 progress=37.7% loss=0.107773 lr=7.353520e-05 lr_backbone=7.353520e-05 elapsed=19:34:46 avg_step=3.74s recent_step=3.11s eta=32:22:59 +2026-09-15 10:05:44 | INFO | train rank=0 | step=18850/50000 progress=37.7% loss=0.146711 lr=7.350601e-05 lr_backbone=7.350601e-05 elapsed=19:35:22 avg_step=3.74s recent_step=3.62s eta=32:22:19 +2026-09-15 10:06:38 | INFO | train rank=0 | step=18860/50000 progress=37.7% loss=0.107264 lr=7.347682e-05 lr_backbone=7.347682e-05 elapsed=19:36:16 avg_step=3.74s recent_step=5.44s eta=32:22:10 +2026-09-15 10:07:15 | INFO | train rank=0 | step=18870/50000 progress=37.7% loss=0.119581 lr=7.344762e-05 lr_backbone=7.344762e-05 elapsed=19:36:53 avg_step=3.74s recent_step=3.69s eta=32:21:31 +2026-09-15 10:07:52 | INFO | train rank=0 | step=18880/50000 progress=37.8% loss=0.107679 lr=7.341841e-05 lr_backbone=7.341841e-05 elapsed=19:37:30 avg_step=3.74s recent_step=3.69s eta=32:20:53 +2026-09-15 10:08:27 | INFO | train rank=0 | step=18890/50000 progress=37.8% loss=0.087405 lr=7.338918e-05 lr_backbone=7.338918e-05 elapsed=19:38:05 avg_step=3.74s recent_step=3.45s eta=32:20:11 +2026-09-15 10:09:10 | INFO | train rank=0 | step=18900/50000 progress=37.8% loss=0.096888 lr=7.335995e-05 lr_backbone=7.335995e-05 elapsed=19:38:48 avg_step=3.74s recent_step=4.35s eta=32:19:44 +2026-09-15 10:09:51 | INFO | train rank=0 | step=18910/50000 progress=37.8% loss=0.089965 lr=7.333071e-05 lr_backbone=7.333071e-05 elapsed=19:39:29 avg_step=3.74s recent_step=4.12s eta=32:19:12 +2026-09-15 10:10:16 | INFO | train rank=0 | step=18920/50000 progress=37.8% loss=0.097252 lr=7.330145e-05 lr_backbone=7.330145e-05 elapsed=19:39:54 avg_step=3.74s recent_step=2.51s eta=32:18:15 +2026-09-15 10:10:54 | INFO | train rank=0 | step=18930/50000 progress=37.9% loss=0.093610 lr=7.327219e-05 lr_backbone=7.327219e-05 elapsed=19:40:32 avg_step=3.74s recent_step=3.72s eta=32:17:37 +2026-09-15 10:11:31 | INFO | train rank=0 | step=18940/50000 progress=37.9% loss=0.094064 lr=7.324292e-05 lr_backbone=7.324292e-05 elapsed=19:41:09 avg_step=3.74s recent_step=3.78s eta=32:17:00 +2026-09-15 10:12:05 | INFO | train rank=0 | step=18950/50000 progress=37.9% loss=0.136894 lr=7.321363e-05 lr_backbone=7.321363e-05 elapsed=19:41:43 avg_step=3.74s recent_step=3.38s eta=32:16:17 +2026-09-15 10:12:35 | INFO | train rank=0 | step=18960/50000 progress=37.9% loss=0.091521 lr=7.318434e-05 lr_backbone=7.318434e-05 elapsed=19:42:13 avg_step=3.74s recent_step=2.96s eta=32:15:27 +2026-09-15 10:13:10 | INFO | train rank=0 | step=18970/50000 progress=37.9% loss=0.094642 lr=7.315503e-05 lr_backbone=7.315503e-05 elapsed=19:42:48 avg_step=3.74s recent_step=3.46s eta=32:14:45 +2026-09-15 10:13:55 | INFO | train rank=0 | step=18980/50000 progress=38.0% loss=0.085456 lr=7.312572e-05 lr_backbone=7.312572e-05 elapsed=19:43:33 avg_step=3.74s recent_step=4.54s eta=32:14:20 +2026-09-15 10:14:27 | INFO | train rank=0 | step=18990/50000 progress=38.0% loss=0.103927 lr=7.309639e-05 lr_backbone=7.309639e-05 elapsed=19:44:05 avg_step=3.74s recent_step=3.23s eta=32:13:35 +2026-09-15 10:15:04 | INFO | train rank=0 | step=19000/50000 progress=38.0% loss=0.103457 lr=7.306706e-05 lr_backbone=7.306706e-05 elapsed=19:44:42 avg_step=3.74s recent_step=3.70s eta=32:12:57 +2026-09-15 10:15:37 | INFO | train rank=0 | step=19010/50000 progress=38.0% loss=0.079910 lr=7.303771e-05 lr_backbone=7.303771e-05 elapsed=19:45:15 avg_step=3.74s recent_step=3.23s eta=32:12:11 +2026-09-15 10:16:21 | INFO | train rank=0 | step=19020/50000 progress=38.0% loss=0.077039 lr=7.300836e-05 lr_backbone=7.300836e-05 elapsed=19:45:59 avg_step=3.74s recent_step=4.47s eta=32:11:45 +2026-09-15 10:17:00 | INFO | train rank=0 | step=19030/50000 progress=38.1% loss=0.094647 lr=7.297899e-05 lr_backbone=7.297899e-05 elapsed=19:46:38 avg_step=3.74s recent_step=3.89s eta=32:11:10 +2026-09-15 10:17:42 | INFO | train rank=0 | step=19040/50000 progress=38.1% loss=0.119153 lr=7.294962e-05 lr_backbone=7.294962e-05 elapsed=19:47:21 avg_step=3.74s recent_step=4.23s eta=32:10:41 +2026-09-15 10:18:11 | INFO | train rank=0 | step=19050/50000 progress=38.1% loss=0.093635 lr=7.292023e-05 lr_backbone=7.292023e-05 elapsed=19:47:49 avg_step=3.74s recent_step=2.85s eta=32:09:49 +2026-09-15 10:18:51 | INFO | train rank=0 | step=19060/50000 progress=38.1% loss=0.085901 lr=7.289084e-05 lr_backbone=7.289084e-05 elapsed=19:48:29 avg_step=3.74s recent_step=3.96s eta=32:09:15 +2026-09-15 10:19:23 | INFO | train rank=0 | step=19070/50000 progress=38.1% loss=0.103817 lr=7.286143e-05 lr_backbone=7.286143e-05 elapsed=19:49:01 avg_step=3.74s recent_step=3.22s eta=32:08:29 +2026-09-15 10:20:09 | INFO | train rank=0 | step=19080/50000 progress=38.2% loss=0.087509 lr=7.283202e-05 lr_backbone=7.283202e-05 elapsed=19:49:47 avg_step=3.74s recent_step=4.62s eta=32:08:06 +2026-09-15 10:20:48 | INFO | train rank=0 | step=19090/50000 progress=38.2% loss=0.100423 lr=7.280259e-05 lr_backbone=7.280259e-05 elapsed=19:50:26 avg_step=3.74s recent_step=3.91s eta=32:07:31 +2026-09-15 10:21:32 | INFO | train rank=0 | step=19100/50000 progress=38.2% loss=0.099657 lr=7.277316e-05 lr_backbone=7.277316e-05 elapsed=19:51:10 avg_step=3.74s recent_step=4.35s eta=32:07:04 +2026-09-15 10:22:06 | INFO | train rank=0 | step=19110/50000 progress=38.2% loss=0.102577 lr=7.274371e-05 lr_backbone=7.274371e-05 elapsed=19:51:44 avg_step=3.74s recent_step=3.42s eta=32:06:21 +2026-09-15 10:22:55 | INFO | train rank=0 | step=19120/50000 progress=38.2% loss=0.107766 lr=7.271426e-05 lr_backbone=7.271426e-05 elapsed=19:52:33 avg_step=3.74s recent_step=4.90s eta=32:06:03 +2026-09-15 10:23:24 | INFO | train rank=0 | step=19130/50000 progress=38.3% loss=0.100600 lr=7.268479e-05 lr_backbone=7.268479e-05 elapsed=19:53:02 avg_step=3.74s recent_step=2.94s eta=32:05:12 +2026-09-15 10:24:00 | INFO | train rank=0 | step=19140/50000 progress=38.3% loss=0.086343 lr=7.265532e-05 lr_backbone=7.265532e-05 elapsed=19:53:38 avg_step=3.74s recent_step=3.56s eta=32:04:32 +2026-09-15 10:24:35 | INFO | train rank=0 | step=19150/50000 progress=38.3% loss=0.094788 lr=7.262583e-05 lr_backbone=7.262583e-05 elapsed=19:54:13 avg_step=3.74s recent_step=3.56s eta=32:03:52 +2026-09-15 10:25:21 | INFO | train rank=0 | step=19160/50000 progress=38.3% loss=0.095018 lr=7.259634e-05 lr_backbone=7.259634e-05 elapsed=19:54:59 avg_step=3.74s recent_step=4.61s eta=32:03:28 +2026-09-15 10:25:54 | INFO | train rank=0 | step=19170/50000 progress=38.3% loss=0.096500 lr=7.256683e-05 lr_backbone=7.256683e-05 elapsed=19:55:32 avg_step=3.74s recent_step=3.30s eta=32:02:43 +2026-09-15 10:26:16 | INFO | train rank=0 | step=19180/50000 progress=38.4% loss=0.102620 lr=7.253732e-05 lr_backbone=7.253732e-05 elapsed=19:55:54 avg_step=3.74s recent_step=2.11s eta=32:01:40 +2026-09-15 10:27:10 | INFO | train rank=0 | step=19190/50000 progress=38.4% loss=0.099688 lr=7.250779e-05 lr_backbone=7.250779e-05 elapsed=19:56:48 avg_step=3.74s recent_step=5.41s eta=32:01:29 +2026-09-15 10:27:45 | INFO | train rank=0 | step=19200/50000 progress=38.4% loss=0.152857 lr=7.247826e-05 lr_backbone=7.247826e-05 elapsed=19:57:23 avg_step=3.74s recent_step=3.53s eta=32:00:49 +2026-09-15 10:28:10 | INFO | train rank=0 | step=19210/50000 progress=38.4% loss=0.094975 lr=7.244871e-05 lr_backbone=7.244871e-05 elapsed=19:57:48 avg_step=3.74s recent_step=2.53s eta=31:59:52 +2026-09-15 10:28:50 | INFO | train rank=0 | step=19220/50000 progress=38.4% loss=0.089469 lr=7.241916e-05 lr_backbone=7.241916e-05 elapsed=19:58:28 avg_step=3.74s recent_step=3.92s eta=31:59:17 +2026-09-15 10:29:25 | INFO | train rank=0 | step=19230/50000 progress=38.5% loss=0.088405 lr=7.238960e-05 lr_backbone=7.238960e-05 elapsed=19:59:03 avg_step=3.74s recent_step=3.56s eta=31:58:37 +2026-09-15 10:30:05 | INFO | train rank=0 | step=19240/50000 progress=38.5% loss=0.109328 lr=7.236002e-05 lr_backbone=7.236002e-05 elapsed=19:59:43 avg_step=3.74s recent_step=4.01s eta=31:58:04 +2026-09-15 10:30:45 | INFO | train rank=0 | step=19250/50000 progress=38.5% loss=0.103394 lr=7.233044e-05 lr_backbone=7.233044e-05 elapsed=20:00:23 avg_step=3.74s recent_step=4.01s eta=31:57:31 +2026-09-15 10:31:33 | INFO | train rank=0 | step=19260/50000 progress=38.5% loss=0.085377 lr=7.230085e-05 lr_backbone=7.230085e-05 elapsed=20:01:11 avg_step=3.74s recent_step=4.78s eta=31:57:10 +2026-09-15 10:32:07 | INFO | train rank=0 | step=19270/50000 progress=38.5% loss=0.099134 lr=7.227124e-05 lr_backbone=7.227124e-05 elapsed=20:01:45 avg_step=3.74s recent_step=3.38s eta=31:56:27 +2026-09-15 10:32:39 | INFO | train rank=0 | step=19280/50000 progress=38.6% loss=0.083281 lr=7.224163e-05 lr_backbone=7.224163e-05 elapsed=20:02:17 avg_step=3.74s recent_step=3.22s eta=31:55:41 +2026-09-15 10:33:15 | INFO | train rank=0 | step=19290/50000 progress=38.6% loss=0.082206 lr=7.221201e-05 lr_backbone=7.221201e-05 elapsed=20:02:53 avg_step=3.74s recent_step=3.57s eta=31:55:01 +2026-09-15 10:34:08 | INFO | train rank=0 | step=19300/50000 progress=38.6% loss=0.094410 lr=7.218238e-05 lr_backbone=7.218238e-05 elapsed=20:03:46 avg_step=3.74s recent_step=5.35s eta=31:54:49 +2026-09-15 10:34:39 | INFO | train rank=0 | step=19310/50000 progress=38.6% loss=0.087274 lr=7.215274e-05 lr_backbone=7.215274e-05 elapsed=20:04:17 avg_step=3.74s recent_step=3.02s eta=31:54:00 +2026-09-15 10:35:11 | INFO | train rank=0 | step=19320/50000 progress=38.6% loss=0.125038 lr=7.212308e-05 lr_backbone=7.212308e-05 elapsed=20:04:49 avg_step=3.74s recent_step=3.23s eta=31:53:15 +2026-09-15 10:35:42 | INFO | train rank=0 | step=19330/50000 progress=38.7% loss=0.089663 lr=7.209342e-05 lr_backbone=7.209342e-05 elapsed=20:05:20 avg_step=3.74s recent_step=3.08s eta=31:52:27 +2026-09-15 10:36:33 | INFO | train rank=0 | step=19340/50000 progress=38.7% loss=0.111835 lr=7.206375e-05 lr_backbone=7.206375e-05 elapsed=20:06:11 avg_step=3.74s recent_step=5.17s eta=31:52:12 +2026-09-15 10:37:06 | INFO | train rank=0 | step=19350/50000 progress=38.7% loss=0.087636 lr=7.203407e-05 lr_backbone=7.203407e-05 elapsed=20:06:44 avg_step=3.74s recent_step=3.26s eta=31:51:27 +2026-09-15 10:37:39 | INFO | train rank=0 | step=19360/50000 progress=38.7% loss=0.095008 lr=7.200438e-05 lr_backbone=7.200438e-05 elapsed=20:07:17 avg_step=3.74s recent_step=3.29s eta=31:50:42 +2026-09-15 10:38:15 | INFO | train rank=0 | step=19370/50000 progress=38.7% loss=0.101233 lr=7.197468e-05 lr_backbone=7.197468e-05 elapsed=20:07:53 avg_step=3.74s recent_step=3.59s eta=31:50:02 +2026-09-15 10:39:09 | INFO | train rank=0 | step=19380/50000 progress=38.8% loss=0.102822 lr=7.194497e-05 lr_backbone=7.194497e-05 elapsed=20:08:47 avg_step=3.74s recent_step=5.45s eta=31:49:52 +2026-09-15 10:39:34 | INFO | train rank=0 | step=19390/50000 progress=38.8% loss=0.098537 lr=7.191525e-05 lr_backbone=7.191525e-05 elapsed=20:09:12 avg_step=3.74s recent_step=2.51s eta=31:48:55 +2026-09-15 10:40:05 | INFO | train rank=0 | step=19400/50000 progress=38.8% loss=0.091893 lr=7.188553e-05 lr_backbone=7.188553e-05 elapsed=20:09:43 avg_step=3.74s recent_step=3.09s eta=31:48:07 +2026-09-15 10:40:41 | INFO | train rank=0 | step=19410/50000 progress=38.8% loss=0.126945 lr=7.185579e-05 lr_backbone=7.185579e-05 elapsed=20:10:19 avg_step=3.74s recent_step=3.58s eta=31:47:27 +2026-09-15 10:41:31 | INFO | train rank=0 | step=19420/50000 progress=38.8% loss=0.099398 lr=7.182604e-05 lr_backbone=7.182604e-05 elapsed=20:11:09 avg_step=3.74s recent_step=4.99s eta=31:47:10 +2026-09-15 10:42:00 | INFO | train rank=0 | step=19430/50000 progress=38.9% loss=0.110006 lr=7.179628e-05 lr_backbone=7.179628e-05 elapsed=20:11:38 avg_step=3.74s recent_step=2.85s eta=31:46:18 +2026-09-15 10:42:35 | INFO | train rank=0 | step=19440/50000 progress=38.9% loss=0.110820 lr=7.176652e-05 lr_backbone=7.176652e-05 elapsed=20:12:14 avg_step=3.74s recent_step=3.60s eta=31:45:39 +2026-09-15 10:43:12 | INFO | train rank=0 | step=19450/50000 progress=38.9% loss=0.080872 lr=7.173674e-05 lr_backbone=7.173674e-05 elapsed=20:12:50 avg_step=3.74s recent_step=3.69s eta=31:45:00 +2026-09-15 10:44:06 | INFO | train rank=0 | step=19460/50000 progress=38.9% loss=0.089731 lr=7.170696e-05 lr_backbone=7.170696e-05 elapsed=20:13:44 avg_step=3.74s recent_step=5.34s eta=31:44:48 +2026-09-15 10:44:41 | INFO | train rank=0 | step=19470/50000 progress=38.9% loss=0.114849 lr=7.167716e-05 lr_backbone=7.167716e-05 elapsed=20:14:19 avg_step=3.74s recent_step=3.50s eta=31:44:07 +2026-09-15 10:45:16 | INFO | train rank=0 | step=19480/50000 progress=39.0% loss=0.096636 lr=7.164736e-05 lr_backbone=7.164736e-05 elapsed=20:14:54 avg_step=3.74s recent_step=3.47s eta=31:43:25 +2026-09-15 10:45:52 | INFO | train rank=0 | step=19490/50000 progress=39.0% loss=0.078779 lr=7.161754e-05 lr_backbone=7.161754e-05 elapsed=20:15:30 avg_step=3.74s recent_step=3.61s eta=31:42:46 +2026-09-15 10:46:35 | INFO | train rank=0 | step=19500/50000 progress=39.0% loss=0.092447 lr=7.158772e-05 lr_backbone=7.158772e-05 elapsed=20:16:13 avg_step=3.74s recent_step=4.35s eta=31:42:18 +2026-09-15 10:47:01 | INFO | train rank=0 | step=19510/50000 progress=39.0% loss=0.089990 lr=7.155788e-05 lr_backbone=7.155788e-05 elapsed=20:16:39 avg_step=3.74s recent_step=2.55s eta=31:41:22 +2026-09-15 10:47:32 | INFO | train rank=0 | step=19520/50000 progress=39.0% loss=0.093200 lr=7.152804e-05 lr_backbone=7.152804e-05 elapsed=20:17:10 avg_step=3.74s recent_step=3.14s eta=31:40:35 +2026-09-15 10:48:09 | INFO | train rank=0 | step=19530/50000 progress=39.1% loss=0.075657 lr=7.149819e-05 lr_backbone=7.149819e-05 elapsed=20:17:47 avg_step=3.74s recent_step=3.67s eta=31:39:56 +2026-09-15 10:48:58 | INFO | train rank=0 | step=19540/50000 progress=39.1% loss=0.092547 lr=7.146833e-05 lr_backbone=7.146833e-05 elapsed=20:18:36 avg_step=3.74s recent_step=4.96s eta=31:39:38 +2026-09-15 10:49:28 | INFO | train rank=0 | step=19550/50000 progress=39.1% loss=0.118630 lr=7.143846e-05 lr_backbone=7.143846e-05 elapsed=20:19:06 avg_step=3.74s recent_step=3.02s eta=31:38:49 +2026-09-15 10:50:09 | INFO | train rank=0 | step=19560/50000 progress=39.1% loss=0.090987 lr=7.140858e-05 lr_backbone=7.140858e-05 elapsed=20:19:47 avg_step=3.74s recent_step=4.06s eta=31:38:17 +2026-09-15 10:50:47 | INFO | train rank=0 | step=19570/50000 progress=39.1% loss=0.091969 lr=7.137869e-05 lr_backbone=7.137869e-05 elapsed=20:20:25 avg_step=3.74s recent_step=3.81s eta=31:37:40 +2026-09-15 10:51:42 | INFO | train rank=0 | step=19580/50000 progress=39.2% loss=0.094649 lr=7.134879e-05 lr_backbone=7.134879e-05 elapsed=20:21:20 avg_step=3.74s recent_step=5.48s eta=31:37:30 +2026-09-15 10:52:15 | INFO | train rank=0 | step=19590/50000 progress=39.2% loss=0.087659 lr=7.131888e-05 lr_backbone=7.131888e-05 elapsed=20:21:53 avg_step=3.74s recent_step=3.31s eta=31:36:46 +2026-09-15 10:52:50 | INFO | train rank=0 | step=19600/50000 progress=39.2% loss=0.075258 lr=7.128896e-05 lr_backbone=7.128896e-05 elapsed=20:22:28 avg_step=3.74s recent_step=3.50s eta=31:36:04 +2026-09-15 10:53:22 | INFO | train rank=0 | step=19610/50000 progress=39.2% loss=0.078109 lr=7.125904e-05 lr_backbone=7.125904e-05 elapsed=20:23:00 avg_step=3.74s recent_step=3.22s eta=31:35:19 +2026-09-15 10:54:15 | INFO | train rank=0 | step=19620/50000 progress=39.2% loss=0.089933 lr=7.122910e-05 lr_backbone=7.122910e-05 elapsed=20:23:53 avg_step=3.74s recent_step=5.32s eta=31:35:06 +2026-09-15 10:54:52 | INFO | train rank=0 | step=19630/50000 progress=39.3% loss=0.088217 lr=7.119916e-05 lr_backbone=7.119916e-05 elapsed=20:24:30 avg_step=3.74s recent_step=3.67s eta=31:34:27 +2026-09-15 10:55:29 | INFO | train rank=0 | step=19640/50000 progress=39.3% loss=0.146846 lr=7.116920e-05 lr_backbone=7.116920e-05 elapsed=20:25:07 avg_step=3.74s recent_step=3.75s eta=31:33:50 +2026-09-15 10:55:57 | INFO | train rank=0 | step=19650/50000 progress=39.3% loss=0.107031 lr=7.113924e-05 lr_backbone=7.113924e-05 elapsed=20:25:35 avg_step=3.74s recent_step=2.72s eta=31:32:57 +2026-09-15 10:56:50 | INFO | train rank=0 | step=19660/50000 progress=39.3% loss=0.099939 lr=7.110926e-05 lr_backbone=7.110926e-05 elapsed=20:26:28 avg_step=3.74s recent_step=5.38s eta=31:32:45 +2026-09-15 10:57:20 | INFO | train rank=0 | step=19670/50000 progress=39.3% loss=0.107275 lr=7.107928e-05 lr_backbone=7.107928e-05 elapsed=20:26:58 avg_step=3.74s recent_step=2.94s eta=31:31:55 +2026-09-15 10:57:53 | INFO | train rank=0 | step=19680/50000 progress=39.4% loss=0.091350 lr=7.104929e-05 lr_backbone=7.104929e-05 elapsed=20:27:31 avg_step=3.74s recent_step=3.29s eta=31:31:10 +2026-09-15 10:58:30 | INFO | train rank=0 | step=19690/50000 progress=39.4% loss=0.103949 lr=7.101929e-05 lr_backbone=7.101929e-05 elapsed=20:28:08 avg_step=3.74s recent_step=3.69s eta=31:30:32 +2026-09-15 10:59:14 | INFO | train rank=0 | step=19700/50000 progress=39.4% loss=0.097018 lr=7.098928e-05 lr_backbone=7.098928e-05 elapsed=20:28:52 avg_step=3.74s recent_step=4.40s eta=31:30:05 +2026-09-15 10:59:45 | INFO | train rank=0 | step=19710/50000 progress=39.4% loss=0.092610 lr=7.095926e-05 lr_backbone=7.095926e-05 elapsed=20:29:23 avg_step=3.74s recent_step=3.12s eta=31:29:18 +2026-09-15 11:00:15 | INFO | train rank=0 | step=19720/50000 progress=39.4% loss=0.087322 lr=7.092923e-05 lr_backbone=7.092923e-05 elapsed=20:29:53 avg_step=3.74s recent_step=3.05s eta=31:28:30 +2026-09-15 11:00:42 | INFO | train rank=0 | step=19730/50000 progress=39.5% loss=0.102649 lr=7.089920e-05 lr_backbone=7.089920e-05 elapsed=20:30:20 avg_step=3.74s recent_step=2.68s eta=31:27:36 +2026-09-15 11:01:29 | INFO | train rank=0 | step=19740/50000 progress=39.5% loss=0.099131 lr=7.086915e-05 lr_backbone=7.086915e-05 elapsed=20:31:07 avg_step=3.74s recent_step=4.75s eta=31:27:14 +2026-09-15 11:02:10 | INFO | train rank=0 | step=19750/50000 progress=39.5% loss=0.116199 lr=7.083909e-05 lr_backbone=7.083909e-05 elapsed=20:31:48 avg_step=3.74s recent_step=4.08s eta=31:26:42 +2026-09-15 11:02:48 | INFO | train rank=0 | step=19760/50000 progress=39.5% loss=0.091078 lr=7.080903e-05 lr_backbone=7.080903e-05 elapsed=20:32:26 avg_step=3.74s recent_step=3.82s eta=31:26:05 +2026-09-15 11:03:22 | INFO | train rank=0 | step=19770/50000 progress=39.5% loss=0.098611 lr=7.077895e-05 lr_backbone=7.077895e-05 elapsed=20:33:00 avg_step=3.74s recent_step=3.31s eta=31:25:21 +2026-09-15 11:04:12 | INFO | train rank=0 | step=19780/50000 progress=39.6% loss=0.097918 lr=7.074887e-05 lr_backbone=7.074887e-05 elapsed=20:33:50 avg_step=3.74s recent_step=5.01s eta=31:25:03 +2026-09-15 11:04:44 | INFO | train rank=0 | step=19790/50000 progress=39.6% loss=0.098118 lr=7.071878e-05 lr_backbone=7.071878e-05 elapsed=20:34:22 avg_step=3.74s recent_step=3.22s eta=31:24:18 +2026-09-15 11:05:15 | INFO | train rank=0 | step=19800/50000 progress=39.6% loss=0.098013 lr=7.068868e-05 lr_backbone=7.068868e-05 elapsed=20:34:53 avg_step=3.74s recent_step=3.10s eta=31:23:31 +2026-09-15 11:05:49 | INFO | train rank=0 | step=19810/50000 progress=39.6% loss=0.098876 lr=7.065857e-05 lr_backbone=7.065857e-05 elapsed=20:35:27 avg_step=3.74s recent_step=3.38s eta=31:22:48 +2026-09-15 11:06:44 | INFO | train rank=0 | step=19820/50000 progress=39.6% loss=0.088573 lr=7.062845e-05 lr_backbone=7.062845e-05 elapsed=20:36:22 avg_step=3.74s recent_step=5.52s eta=31:22:37 +2026-09-15 11:07:19 | INFO | train rank=0 | step=19830/50000 progress=39.7% loss=0.088207 lr=7.059832e-05 lr_backbone=7.059832e-05 elapsed=20:36:57 avg_step=3.74s recent_step=3.50s eta=31:21:56 +2026-09-15 11:07:45 | INFO | train rank=0 | step=19840/50000 progress=39.7% loss=0.081078 lr=7.056818e-05 lr_backbone=7.056818e-05 elapsed=20:37:23 avg_step=3.74s recent_step=2.57s eta=31:21:01 +2026-09-15 11:08:16 | INFO | train rank=0 | step=19850/50000 progress=39.7% loss=0.108252 lr=7.053804e-05 lr_backbone=7.053804e-05 elapsed=20:37:54 avg_step=3.74s recent_step=3.11s eta=31:20:14 +2026-09-15 11:09:11 | INFO | train rank=0 | step=19860/50000 progress=39.7% loss=0.083355 lr=7.050788e-05 lr_backbone=7.050788e-05 elapsed=20:38:49 avg_step=3.74s recent_step=5.53s eta=31:20:04 +2026-09-15 11:09:47 | INFO | train rank=0 | step=19870/50000 progress=39.7% loss=0.090631 lr=7.047772e-05 lr_backbone=7.047772e-05 elapsed=20:39:25 avg_step=3.74s recent_step=3.57s eta=31:19:24 +2026-09-15 11:10:17 | INFO | train rank=0 | step=19880/50000 progress=39.8% loss=0.112399 lr=7.044754e-05 lr_backbone=7.044754e-05 elapsed=20:39:56 avg_step=3.74s recent_step=3.08s eta=31:18:36 +2026-09-15 11:10:54 | INFO | train rank=0 | step=19890/50000 progress=39.8% loss=0.141281 lr=7.041736e-05 lr_backbone=7.041736e-05 elapsed=20:40:32 avg_step=3.74s recent_step=3.63s eta=31:17:57 +2026-09-15 11:11:35 | INFO | train rank=0 | step=19900/50000 progress=39.8% loss=0.105150 lr=7.038717e-05 lr_backbone=7.038717e-05 elapsed=20:41:13 avg_step=3.74s recent_step=4.08s eta=31:17:25 +2026-09-15 11:12:05 | INFO | train rank=0 | step=19910/50000 progress=39.8% loss=0.097889 lr=7.035697e-05 lr_backbone=7.035697e-05 elapsed=20:41:43 avg_step=3.74s recent_step=3.08s eta=31:16:37 +2026-09-15 11:12:41 | INFO | train rank=0 | step=19920/50000 progress=39.8% loss=0.103798 lr=7.032676e-05 lr_backbone=7.032676e-05 elapsed=20:42:19 avg_step=3.74s recent_step=3.59s eta=31:15:58 +2026-09-15 11:13:13 | INFO | train rank=0 | step=19930/50000 progress=39.9% loss=0.090260 lr=7.029654e-05 lr_backbone=7.029654e-05 elapsed=20:42:51 avg_step=3.74s recent_step=3.17s eta=31:15:12 +2026-09-15 11:14:04 | INFO | train rank=0 | step=19940/50000 progress=39.9% loss=0.100915 lr=7.026632e-05 lr_backbone=7.026632e-05 elapsed=20:43:42 avg_step=3.74s recent_step=5.13s eta=31:14:55 +2026-09-15 11:14:44 | INFO | train rank=0 | step=19950/50000 progress=39.9% loss=0.116861 lr=7.023608e-05 lr_backbone=7.023608e-05 elapsed=20:44:22 avg_step=3.74s recent_step=4.01s eta=31:14:22 +2026-09-15 11:15:18 | INFO | train rank=0 | step=19960/50000 progress=39.9% loss=0.095902 lr=7.020584e-05 lr_backbone=7.020584e-05 elapsed=20:44:56 avg_step=3.74s recent_step=3.36s eta=31:13:39 +2026-09-15 11:15:52 | INFO | train rank=0 | step=19970/50000 progress=39.9% loss=0.105825 lr=7.017558e-05 lr_backbone=7.017558e-05 elapsed=20:45:30 avg_step=3.74s recent_step=3.37s eta=31:12:56 +2026-09-15 11:16:45 | INFO | train rank=0 | step=19980/50000 progress=40.0% loss=0.097931 lr=7.014532e-05 lr_backbone=7.014532e-05 elapsed=20:46:23 avg_step=3.74s recent_step=5.31s eta=31:12:42 +2026-09-15 11:17:19 | INFO | train rank=0 | step=19990/50000 progress=40.0% loss=0.079606 lr=7.011505e-05 lr_backbone=7.011505e-05 elapsed=20:46:57 avg_step=3.74s recent_step=3.45s eta=31:12:00 +2026-09-15 11:17:47 | INFO | train rank=0 | step=20000/50000 progress=40.0% loss=0.103231 lr=7.008477e-05 lr_backbone=7.008477e-05 elapsed=20:47:25 avg_step=3.74s recent_step=2.79s eta=31:11:08 +2026-09-15 11:18:26 | INFO | train rank=0 | step=20010/50000 progress=40.0% loss=0.110432 lr=7.005448e-05 lr_backbone=7.005448e-05 elapsed=20:48:04 avg_step=3.74s recent_step=3.82s eta=31:10:32 +2026-09-15 11:19:18 | INFO | train rank=0 | step=20020/50000 progress=40.0% loss=0.104573 lr=7.002419e-05 lr_backbone=7.002419e-05 elapsed=20:48:56 avg_step=3.74s recent_step=5.24s eta=31:10:17 +2026-09-15 11:19:47 | INFO | train rank=0 | step=20030/50000 progress=40.1% loss=0.110305 lr=6.999388e-05 lr_backbone=6.999388e-05 elapsed=20:49:25 avg_step=3.74s recent_step=2.93s eta=31:09:28 +2026-09-15 11:20:19 | INFO | train rank=0 | step=20040/50000 progress=40.1% loss=0.097651 lr=6.996356e-05 lr_backbone=6.996356e-05 elapsed=20:49:57 avg_step=3.74s recent_step=3.14s eta=31:08:41 +2026-09-15 11:20:52 | INFO | train rank=0 | step=20050/50000 progress=40.1% loss=0.092250 lr=6.993324e-05 lr_backbone=6.993324e-05 elapsed=20:50:30 avg_step=3.74s recent_step=3.29s eta=31:07:57 +2026-09-15 11:21:46 | INFO | train rank=0 | step=20060/50000 progress=40.1% loss=0.119915 lr=6.990291e-05 lr_backbone=6.990291e-05 elapsed=20:51:24 avg_step=3.74s recent_step=5.47s eta=31:07:45 +2026-09-15 11:22:11 | INFO | train rank=0 | step=20070/50000 progress=40.1% loss=0.091521 lr=6.987257e-05 lr_backbone=6.987257e-05 elapsed=20:51:49 avg_step=3.74s recent_step=2.48s eta=31:06:49 +2026-09-15 11:22:47 | INFO | train rank=0 | step=20080/50000 progress=40.2% loss=0.081924 lr=6.984222e-05 lr_backbone=6.984222e-05 elapsed=20:52:25 avg_step=3.74s recent_step=3.60s eta=31:06:09 +2026-09-15 11:23:26 | INFO | train rank=0 | step=20090/50000 progress=40.2% loss=0.109987 lr=6.981186e-05 lr_backbone=6.981186e-05 elapsed=20:53:04 avg_step=3.74s recent_step=3.93s eta=31:05:35 +2026-09-15 11:24:13 | INFO | train rank=0 | step=20100/50000 progress=40.2% loss=0.110074 lr=6.978149e-05 lr_backbone=6.978149e-05 elapsed=20:53:51 avg_step=3.74s recent_step=4.66s eta=31:05:11 +2026-09-15 11:24:44 | INFO | train rank=0 | step=20110/50000 progress=40.2% loss=0.081277 lr=6.975112e-05 lr_backbone=6.975112e-05 elapsed=20:54:22 avg_step=3.74s recent_step=3.08s eta=31:04:24 +2026-09-15 11:25:21 | INFO | train rank=0 | step=20120/50000 progress=40.2% loss=0.107203 lr=6.972073e-05 lr_backbone=6.972073e-05 elapsed=20:54:59 avg_step=3.74s recent_step=3.68s eta=31:03:45 +2026-09-15 11:25:46 | INFO | train rank=0 | step=20130/50000 progress=40.3% loss=0.123439 lr=6.969034e-05 lr_backbone=6.969034e-05 elapsed=20:55:24 avg_step=3.74s recent_step=2.53s eta=31:02:50 +2026-09-15 11:26:31 | INFO | train rank=0 | step=20140/50000 progress=40.3% loss=0.080651 lr=6.965994e-05 lr_backbone=6.965994e-05 elapsed=20:56:09 avg_step=3.74s recent_step=4.46s eta=31:02:23 +2026-09-15 11:27:05 | INFO | train rank=0 | step=20150/50000 progress=40.3% loss=0.086141 lr=6.962953e-05 lr_backbone=6.962953e-05 elapsed=20:56:43 avg_step=3.74s recent_step=3.41s eta=31:01:41 +2026-09-15 11:27:38 | INFO | train rank=0 | step=20160/50000 progress=40.3% loss=0.085274 lr=6.959911e-05 lr_backbone=6.959911e-05 elapsed=20:57:16 avg_step=3.74s recent_step=3.32s eta=31:00:57 +2026-09-15 11:28:12 | INFO | train rank=0 | step=20170/50000 progress=40.3% loss=0.105782 lr=6.956868e-05 lr_backbone=6.956868e-05 elapsed=20:57:50 avg_step=3.74s recent_step=3.42s eta=31:00:15 +2026-09-15 11:28:57 | INFO | train rank=0 | step=20180/50000 progress=40.4% loss=0.109141 lr=6.953825e-05 lr_backbone=6.953825e-05 elapsed=20:58:35 avg_step=3.74s recent_step=4.48s eta=30:59:49 +2026-09-15 11:29:33 | INFO | train rank=0 | step=20190/50000 progress=40.4% loss=0.086527 lr=6.950780e-05 lr_backbone=6.950780e-05 elapsed=20:59:11 avg_step=3.74s recent_step=3.60s eta=30:59:09 +2026-09-15 11:30:08 | INFO | train rank=0 | step=20200/50000 progress=40.4% loss=0.109566 lr=6.947735e-05 lr_backbone=6.947735e-05 elapsed=20:59:46 avg_step=3.74s recent_step=3.55s eta=30:58:29 +2026-09-15 11:30:43 | INFO | train rank=0 | step=20210/50000 progress=40.4% loss=0.085604 lr=6.944689e-05 lr_backbone=6.944689e-05 elapsed=21:00:21 avg_step=3.74s recent_step=3.45s eta=30:57:47 +2026-09-15 11:31:30 | INFO | train rank=0 | step=20220/50000 progress=40.4% loss=0.084885 lr=6.941642e-05 lr_backbone=6.941642e-05 elapsed=21:01:08 avg_step=3.74s recent_step=4.68s eta=30:57:23 +2026-09-15 11:31:58 | INFO | train rank=0 | step=20230/50000 progress=40.5% loss=0.099240 lr=6.938594e-05 lr_backbone=6.938594e-05 elapsed=21:01:36 avg_step=3.74s recent_step=2.87s eta=30:56:33 +2026-09-15 11:32:34 | INFO | train rank=0 | step=20240/50000 progress=40.5% loss=0.112999 lr=6.935545e-05 lr_backbone=6.935545e-05 elapsed=21:02:12 avg_step=3.74s recent_step=3.59s eta=30:55:54 +2026-09-15 11:33:09 | INFO | train rank=0 | step=20250/50000 progress=40.5% loss=0.085913 lr=6.932496e-05 lr_backbone=6.932496e-05 elapsed=21:02:47 avg_step=3.74s recent_step=3.45s eta=30:55:12 +2026-09-15 11:34:06 | INFO | train rank=0 | step=20260/50000 progress=40.5% loss=0.106092 lr=6.929445e-05 lr_backbone=6.929445e-05 elapsed=21:03:44 avg_step=3.74s recent_step=5.70s eta=30:55:03 +2026-09-15 11:34:36 | INFO | train rank=0 | step=20270/50000 progress=40.5% loss=0.077996 lr=6.926394e-05 lr_backbone=6.926394e-05 elapsed=21:04:14 avg_step=3.74s recent_step=3.04s eta=30:54:16 +2026-09-15 11:35:06 | INFO | train rank=0 | step=20280/50000 progress=40.6% loss=0.116019 lr=6.923342e-05 lr_backbone=6.923342e-05 elapsed=21:04:44 avg_step=3.74s recent_step=2.98s eta=30:53:27 +2026-09-15 11:35:31 | INFO | train rank=0 | step=20290/50000 progress=40.6% loss=0.099901 lr=6.920289e-05 lr_backbone=6.920289e-05 elapsed=21:05:09 avg_step=3.74s recent_step=2.54s eta=30:52:32 +2026-09-15 11:36:14 | INFO | train rank=0 | step=20300/50000 progress=40.6% loss=0.106404 lr=6.917236e-05 lr_backbone=6.917236e-05 elapsed=21:05:52 avg_step=3.74s recent_step=4.29s eta=30:52:03 +2026-09-15 11:36:50 | INFO | train rank=0 | step=20310/50000 progress=40.6% loss=0.106913 lr=6.914181e-05 lr_backbone=6.914181e-05 elapsed=21:06:28 avg_step=3.74s recent_step=3.56s eta=30:51:23 +2026-09-15 11:37:25 | INFO | train rank=0 | step=20320/50000 progress=40.6% loss=0.115381 lr=6.911126e-05 lr_backbone=6.911126e-05 elapsed=21:07:03 avg_step=3.74s recent_step=3.49s eta=30:50:41 +2026-09-15 11:37:57 | INFO | train rank=0 | step=20330/50000 progress=40.7% loss=0.108136 lr=6.908069e-05 lr_backbone=6.908069e-05 elapsed=21:07:35 avg_step=3.74s recent_step=3.18s eta=30:49:56 +2026-09-15 11:38:41 | INFO | train rank=0 | step=20340/50000 progress=40.7% loss=0.092197 lr=6.905012e-05 lr_backbone=6.905012e-05 elapsed=21:08:19 avg_step=3.74s recent_step=4.42s eta=30:49:28 +2026-09-15 11:39:20 | INFO | train rank=0 | step=20350/50000 progress=40.7% loss=0.103159 lr=6.901954e-05 lr_backbone=6.901954e-05 elapsed=21:08:58 avg_step=3.74s recent_step=3.92s eta=30:48:53 +2026-09-15 11:39:54 | INFO | train rank=0 | step=20360/50000 progress=40.7% loss=0.084045 lr=6.898895e-05 lr_backbone=6.898895e-05 elapsed=21:09:32 avg_step=3.74s recent_step=3.42s eta=30:48:11 +2026-09-15 11:40:29 | INFO | train rank=0 | step=20370/50000 progress=40.7% loss=0.078467 lr=6.895836e-05 lr_backbone=6.895836e-05 elapsed=21:10:07 avg_step=3.74s recent_step=3.49s eta=30:47:30 +2026-09-15 11:41:16 | INFO | train rank=0 | step=20380/50000 progress=40.8% loss=0.089875 lr=6.892775e-05 lr_backbone=6.892775e-05 elapsed=21:10:54 avg_step=3.74s recent_step=4.65s eta=30:47:06 +2026-09-15 11:41:49 | INFO | train rank=0 | step=20390/50000 progress=40.8% loss=0.105299 lr=6.889714e-05 lr_backbone=6.889714e-05 elapsed=21:11:27 avg_step=3.74s recent_step=3.32s eta=30:46:23 +2026-09-15 11:42:24 | INFO | train rank=0 | step=20400/50000 progress=40.8% loss=0.132397 lr=6.886652e-05 lr_backbone=6.886652e-05 elapsed=21:12:02 avg_step=3.74s recent_step=3.55s eta=30:45:43 +2026-09-15 11:42:56 | INFO | train rank=0 | step=20410/50000 progress=40.8% loss=0.118287 lr=6.883589e-05 lr_backbone=6.883589e-05 elapsed=21:12:34 avg_step=3.74s recent_step=3.15s eta=30:44:57 +2026-09-15 11:43:48 | INFO | train rank=0 | step=20420/50000 progress=40.8% loss=0.075097 lr=6.880526e-05 lr_backbone=6.880526e-05 elapsed=21:13:26 avg_step=3.74s recent_step=5.23s eta=30:44:41 +2026-09-15 11:44:27 | INFO | train rank=0 | step=20430/50000 progress=40.9% loss=0.077990 lr=6.877461e-05 lr_backbone=6.877461e-05 elapsed=21:14:05 avg_step=3.74s recent_step=3.86s eta=30:44:05 +2026-09-15 11:45:03 | INFO | train rank=0 | step=20440/50000 progress=40.9% loss=0.106603 lr=6.874396e-05 lr_backbone=6.874396e-05 elapsed=21:14:41 avg_step=3.74s recent_step=3.63s eta=30:43:26 +2026-09-15 11:45:42 | INFO | train rank=0 | step=20450/50000 progress=40.9% loss=0.120589 lr=6.871329e-05 lr_backbone=6.871329e-05 elapsed=21:15:20 avg_step=3.74s recent_step=3.93s eta=30:42:51 +2026-09-15 11:46:35 | INFO | train rank=0 | step=20460/50000 progress=40.9% loss=0.118837 lr=6.868262e-05 lr_backbone=6.868262e-05 elapsed=21:16:13 avg_step=3.74s recent_step=5.31s eta=30:42:36 +2026-09-15 11:47:02 | INFO | train rank=0 | step=20470/50000 progress=40.9% loss=0.093820 lr=6.865195e-05 lr_backbone=6.865195e-05 elapsed=21:16:40 avg_step=3.74s recent_step=2.64s eta=30:41:43 +2026-09-15 11:47:37 | INFO | train rank=0 | step=20480/50000 progress=41.0% loss=0.136042 lr=6.862126e-05 lr_backbone=6.862126e-05 elapsed=21:17:16 avg_step=3.74s recent_step=3.57s eta=30:41:03 +2026-09-15 11:48:09 | INFO | train rank=0 | step=20490/50000 progress=41.0% loss=0.117045 lr=6.859057e-05 lr_backbone=6.859057e-05 elapsed=21:17:47 avg_step=3.74s recent_step=3.19s eta=30:40:18 +2026-09-15 11:49:04 | INFO | train rank=0 | step=20500/50000 progress=41.0% loss=0.085952 lr=6.855986e-05 lr_backbone=6.855986e-05 elapsed=21:18:42 avg_step=3.74s recent_step=5.42s eta=30:40:04 +2026-09-15 11:49:36 | INFO | train rank=0 | step=20510/50000 progress=41.0% loss=0.106963 lr=6.852915e-05 lr_backbone=6.852915e-05 elapsed=21:19:14 avg_step=3.74s recent_step=3.23s eta=30:39:20 +2026-09-15 11:50:15 | INFO | train rank=0 | step=20520/50000 progress=41.0% loss=0.097389 lr=6.849843e-05 lr_backbone=6.849843e-05 elapsed=21:19:53 avg_step=3.74s recent_step=3.89s eta=30:38:44 +2026-09-15 11:50:43 | INFO | train rank=0 | step=20530/50000 progress=41.1% loss=0.087876 lr=6.846771e-05 lr_backbone=6.846771e-05 elapsed=21:20:21 avg_step=3.74s recent_step=2.77s eta=30:37:53 +2026-09-15 11:51:21 | INFO | train rank=0 | step=20540/50000 progress=41.1% loss=0.134267 lr=6.843697e-05 lr_backbone=6.843697e-05 elapsed=21:20:59 avg_step=3.74s recent_step=3.84s eta=30:37:17 +2026-09-15 11:51:57 | INFO | train rank=0 | step=20550/50000 progress=41.1% loss=0.088993 lr=6.840623e-05 lr_backbone=6.840623e-05 elapsed=21:21:35 avg_step=3.74s recent_step=3.59s eta=30:36:37 +2026-09-15 11:52:32 | INFO | train rank=0 | step=20560/50000 progress=41.1% loss=0.094868 lr=6.837548e-05 lr_backbone=6.837548e-05 elapsed=21:22:10 avg_step=3.74s recent_step=3.54s eta=30:35:57 +2026-09-15 11:53:12 | INFO | train rank=0 | step=20570/50000 progress=41.1% loss=0.103836 lr=6.834472e-05 lr_backbone=6.834472e-05 elapsed=21:22:50 avg_step=3.74s recent_step=3.94s eta=30:35:23 +2026-09-15 11:53:59 | INFO | train rank=0 | step=20580/50000 progress=41.2% loss=0.100559 lr=6.831395e-05 lr_backbone=6.831395e-05 elapsed=21:23:37 avg_step=3.74s recent_step=4.69s eta=30:34:59 +2026-09-15 11:54:32 | INFO | train rank=0 | step=20590/50000 progress=41.2% loss=0.081858 lr=6.828317e-05 lr_backbone=6.828317e-05 elapsed=21:24:10 avg_step=3.74s recent_step=3.32s eta=30:34:15 +2026-09-15 11:55:10 | INFO | train rank=0 | step=20600/50000 progress=41.2% loss=0.092994 lr=6.825239e-05 lr_backbone=6.825239e-05 elapsed=21:24:48 avg_step=3.74s recent_step=3.82s eta=30:33:39 +2026-09-15 11:55:48 | INFO | train rank=0 | step=20610/50000 progress=41.2% loss=0.112298 lr=6.822160e-05 lr_backbone=6.822160e-05 elapsed=21:25:26 avg_step=3.74s recent_step=3.86s eta=30:33:03 +2026-09-15 11:56:35 | INFO | train rank=0 | step=20620/50000 progress=41.2% loss=0.099498 lr=6.819080e-05 lr_backbone=6.819080e-05 elapsed=21:26:13 avg_step=3.74s recent_step=4.61s eta=30:32:38 +2026-09-15 11:57:00 | INFO | train rank=0 | step=20630/50000 progress=41.3% loss=0.089872 lr=6.815999e-05 lr_backbone=6.815999e-05 elapsed=21:26:38 avg_step=3.74s recent_step=2.56s eta=30:31:44 +2026-09-15 11:57:31 | INFO | train rank=0 | step=20640/50000 progress=41.3% loss=0.092831 lr=6.812918e-05 lr_backbone=6.812918e-05 elapsed=21:27:09 avg_step=3.74s recent_step=3.10s eta=30:30:57 +2026-09-15 11:58:06 | INFO | train rank=0 | step=20650/50000 progress=41.3% loss=0.123981 lr=6.809836e-05 lr_backbone=6.809836e-05 elapsed=21:27:44 avg_step=3.74s recent_step=3.50s eta=30:30:16 +2026-09-15 11:58:47 | INFO | train rank=0 | step=20660/50000 progress=41.3% loss=0.084651 lr=6.806752e-05 lr_backbone=6.806752e-05 elapsed=21:28:25 avg_step=3.74s recent_step=4.09s eta=30:29:44 +2026-09-15 11:59:23 | INFO | train rank=0 | step=20670/50000 progress=41.3% loss=0.114175 lr=6.803669e-05 lr_backbone=6.803669e-05 elapsed=21:29:01 avg_step=3.74s recent_step=3.58s eta=30:29:04 +2026-09-15 12:00:08 | INFO | train rank=0 | step=20680/50000 progress=41.4% loss=0.088214 lr=6.800584e-05 lr_backbone=6.800584e-05 elapsed=21:29:46 avg_step=3.74s recent_step=4.52s eta=30:28:38 +2026-09-15 12:00:49 | INFO | train rank=0 | step=20690/50000 progress=41.4% loss=0.105927 lr=6.797498e-05 lr_backbone=6.797498e-05 elapsed=21:30:27 avg_step=3.74s recent_step=4.09s eta=30:28:05 +2026-09-15 12:01:41 | INFO | train rank=0 | step=20700/50000 progress=41.4% loss=0.096672 lr=6.794412e-05 lr_backbone=6.794412e-05 elapsed=21:31:19 avg_step=3.74s recent_step=5.24s eta=30:27:49 +2026-09-15 12:02:20 | INFO | train rank=0 | step=20710/50000 progress=41.4% loss=0.118268 lr=6.791325e-05 lr_backbone=6.791325e-05 elapsed=21:31:58 avg_step=3.74s recent_step=3.83s eta=30:27:13 +2026-09-15 12:02:53 | INFO | train rank=0 | step=20720/50000 progress=41.4% loss=0.079767 lr=6.788237e-05 lr_backbone=6.788237e-05 elapsed=21:32:32 avg_step=3.74s recent_step=3.38s eta=30:26:30 +2026-09-15 12:03:28 | INFO | train rank=0 | step=20730/50000 progress=41.5% loss=0.117824 lr=6.785149e-05 lr_backbone=6.785149e-05 elapsed=21:33:06 avg_step=3.74s recent_step=3.41s eta=30:25:48 +2026-09-15 12:04:11 | INFO | train rank=0 | step=20740/50000 progress=41.5% loss=0.116545 lr=6.782059e-05 lr_backbone=6.782059e-05 elapsed=21:33:49 avg_step=3.74s recent_step=4.31s eta=30:25:19 +2026-09-15 12:04:42 | INFO | train rank=0 | step=20750/50000 progress=41.5% loss=0.096258 lr=6.778969e-05 lr_backbone=6.778969e-05 elapsed=21:34:20 avg_step=3.74s recent_step=3.15s eta=30:24:33 +2026-09-15 12:05:17 | INFO | train rank=0 | step=20760/50000 progress=41.5% loss=0.107103 lr=6.775878e-05 lr_backbone=6.775878e-05 elapsed=21:34:55 avg_step=3.74s recent_step=3.46s eta=30:23:52 +2026-09-15 12:05:49 | INFO | train rank=0 | step=20770/50000 progress=41.5% loss=0.086495 lr=6.772787e-05 lr_backbone=6.772787e-05 elapsed=21:35:27 avg_step=3.74s recent_step=3.26s eta=30:23:07 +2026-09-15 12:06:46 | INFO | train rank=0 | step=20780/50000 progress=41.6% loss=0.085446 lr=6.769694e-05 lr_backbone=6.769694e-05 elapsed=21:36:24 avg_step=3.74s recent_step=5.64s eta=30:22:57 +2026-09-15 12:07:24 | INFO | train rank=0 | step=20790/50000 progress=41.6% loss=0.091018 lr=6.766601e-05 lr_backbone=6.766601e-05 elapsed=21:37:02 avg_step=3.74s recent_step=3.85s eta=30:22:21 +2026-09-15 12:07:58 | INFO | train rank=0 | step=20800/50000 progress=41.6% loss=0.074480 lr=6.763507e-05 lr_backbone=6.763507e-05 elapsed=21:37:36 avg_step=3.74s recent_step=3.33s eta=30:21:37 +2026-09-15 12:08:29 | INFO | train rank=0 | step=20810/50000 progress=41.6% loss=0.103263 lr=6.760412e-05 lr_backbone=6.760412e-05 elapsed=21:38:07 avg_step=3.74s recent_step=3.17s eta=30:20:52 +2026-09-15 12:09:15 | INFO | train rank=0 | step=20820/50000 progress=41.6% loss=0.087890 lr=6.757316e-05 lr_backbone=6.757316e-05 elapsed=21:38:53 avg_step=3.74s recent_step=4.56s eta=30:20:26 +2026-09-15 12:09:48 | INFO | train rank=0 | step=20830/50000 progress=41.7% loss=0.114478 lr=6.754220e-05 lr_backbone=6.754220e-05 elapsed=21:39:26 avg_step=3.74s recent_step=3.35s eta=30:19:43 +2026-09-15 12:10:26 | INFO | train rank=0 | step=20840/50000 progress=41.7% loss=0.079137 lr=6.751123e-05 lr_backbone=6.751123e-05 elapsed=21:40:04 avg_step=3.74s recent_step=3.74s eta=30:19:06 +2026-09-15 12:10:58 | INFO | train rank=0 | step=20850/50000 progress=41.7% loss=0.090021 lr=6.748025e-05 lr_backbone=6.748025e-05 elapsed=21:40:36 avg_step=3.74s recent_step=3.18s eta=30:18:20 +2026-09-15 12:11:48 | INFO | train rank=0 | step=20860/50000 progress=41.7% loss=0.077404 lr=6.744926e-05 lr_backbone=6.744926e-05 elapsed=21:41:26 avg_step=3.74s recent_step=5.07s eta=30:18:02 +2026-09-15 12:12:26 | INFO | train rank=0 | step=20870/50000 progress=41.7% loss=0.106326 lr=6.741827e-05 lr_backbone=6.741827e-05 elapsed=21:42:04 avg_step=3.74s recent_step=3.72s eta=30:17:24 +2026-09-15 12:13:04 | INFO | train rank=0 | step=20880/50000 progress=41.8% loss=0.102478 lr=6.738727e-05 lr_backbone=6.738727e-05 elapsed=21:42:42 avg_step=3.74s recent_step=3.82s eta=30:16:47 +2026-09-15 12:13:33 | INFO | train rank=0 | step=20890/50000 progress=41.8% loss=0.086540 lr=6.735626e-05 lr_backbone=6.735626e-05 elapsed=21:43:11 avg_step=3.74s recent_step=2.87s eta=30:15:58 +2026-09-15 12:14:26 | INFO | train rank=0 | step=20900/50000 progress=41.8% loss=0.104055 lr=6.732524e-05 lr_backbone=6.732524e-05 elapsed=21:44:04 avg_step=3.74s recent_step=5.37s eta=30:15:43 +2026-09-15 12:14:51 | INFO | train rank=0 | step=20910/50000 progress=41.8% loss=0.114673 lr=6.729422e-05 lr_backbone=6.729422e-05 elapsed=21:44:29 avg_step=3.74s recent_step=2.49s eta=30:14:48 +2026-09-15 12:15:21 | INFO | train rank=0 | step=20920/50000 progress=41.8% loss=0.105631 lr=6.726319e-05 lr_backbone=6.726319e-05 elapsed=21:44:59 avg_step=3.74s recent_step=2.95s eta=30:14:00 +2026-09-15 12:15:54 | INFO | train rank=0 | step=20930/50000 progress=41.9% loss=0.089530 lr=6.723215e-05 lr_backbone=6.723215e-05 elapsed=21:45:32 avg_step=3.74s recent_step=3.38s eta=30:13:17 +2026-09-15 12:16:46 | INFO | train rank=0 | step=20940/50000 progress=41.9% loss=0.114127 lr=6.720110e-05 lr_backbone=6.720110e-05 elapsed=21:46:24 avg_step=3.74s recent_step=5.20s eta=30:13:00 +2026-09-15 12:17:09 | INFO | train rank=0 | step=20950/50000 progress=41.9% loss=0.090998 lr=6.717004e-05 lr_backbone=6.717004e-05 elapsed=21:46:47 avg_step=3.74s recent_step=2.22s eta=30:12:02 +2026-09-15 12:17:32 | INFO | train rank=0 | step=20960/50000 progress=41.9% loss=0.118676 lr=6.713898e-05 lr_backbone=6.713898e-05 elapsed=21:47:10 avg_step=3.74s recent_step=2.35s eta=30:11:05 +2026-09-15 12:18:09 | INFO | train rank=0 | step=20970/50000 progress=41.9% loss=0.112949 lr=6.710791e-05 lr_backbone=6.710791e-05 elapsed=21:47:47 avg_step=3.74s recent_step=3.70s eta=30:10:27 +2026-09-15 12:19:04 | INFO | train rank=0 | step=20980/50000 progress=42.0% loss=0.085525 lr=6.707684e-05 lr_backbone=6.707684e-05 elapsed=21:48:42 avg_step=3.74s recent_step=5.47s eta=30:10:13 +2026-09-15 12:19:34 | INFO | train rank=0 | step=20990/50000 progress=42.0% loss=0.099214 lr=6.704575e-05 lr_backbone=6.704575e-05 elapsed=21:49:12 avg_step=3.74s recent_step=3.05s eta=30:09:26 +2026-09-15 12:20:12 | INFO | train rank=0 | step=21000/50000 progress=42.0% loss=0.106332 lr=6.701466e-05 lr_backbone=6.701466e-05 elapsed=21:49:50 avg_step=3.74s recent_step=3.75s eta=30:08:49 +2026-09-15 12:20:42 | INFO | train rank=0 | step=21010/50000 progress=42.0% loss=0.102734 lr=6.698356e-05 lr_backbone=6.698356e-05 elapsed=21:50:20 avg_step=3.74s recent_step=3.06s eta=30:08:02 +2026-09-15 12:21:35 | INFO | train rank=0 | step=21020/50000 progress=42.0% loss=0.120011 lr=6.695245e-05 lr_backbone=6.695245e-05 elapsed=21:51:13 avg_step=3.74s recent_step=5.28s eta=30:07:46 +2026-09-15 12:22:06 | INFO | train rank=0 | step=21030/50000 progress=42.1% loss=0.097392 lr=6.692134e-05 lr_backbone=6.692134e-05 elapsed=21:51:44 avg_step=3.74s recent_step=3.10s eta=30:07:00 +2026-09-15 12:22:37 | INFO | train rank=0 | step=21040/50000 progress=42.1% loss=0.077148 lr=6.689022e-05 lr_backbone=6.689022e-05 elapsed=21:52:15 avg_step=3.74s recent_step=3.11s eta=30:06:13 +2026-09-15 12:23:13 | INFO | train rank=0 | step=21050/50000 progress=42.1% loss=0.092691 lr=6.685909e-05 lr_backbone=6.685909e-05 elapsed=21:52:51 avg_step=3.74s recent_step=3.54s eta=30:05:33 +2026-09-15 12:23:57 | INFO | train rank=0 | step=21060/50000 progress=42.1% loss=0.106763 lr=6.682795e-05 lr_backbone=6.682795e-05 elapsed=21:53:35 avg_step=3.74s recent_step=4.48s eta=30:05:06 +2026-09-15 12:24:27 | INFO | train rank=0 | step=21070/50000 progress=42.1% loss=0.101753 lr=6.679681e-05 lr_backbone=6.679681e-05 elapsed=21:54:05 avg_step=3.74s recent_step=2.93s eta=30:04:17 +2026-09-15 12:25:02 | INFO | train rank=0 | step=21080/50000 progress=42.2% loss=0.092693 lr=6.676566e-05 lr_backbone=6.676566e-05 elapsed=21:54:40 avg_step=3.74s recent_step=3.50s eta=30:03:37 +2026-09-15 12:25:30 | INFO | train rank=0 | step=21090/50000 progress=42.2% loss=0.094957 lr=6.673450e-05 lr_backbone=6.673450e-05 elapsed=21:55:08 avg_step=3.74s recent_step=2.85s eta=30:02:47 +2026-09-15 12:26:07 | INFO | train rank=0 | step=21100/50000 progress=42.2% loss=0.102731 lr=6.670333e-05 lr_backbone=6.670333e-05 elapsed=21:55:45 avg_step=3.74s recent_step=3.66s eta=30:02:08 +2026-09-15 12:26:35 | INFO | train rank=0 | step=21110/50000 progress=42.2% loss=0.101562 lr=6.667216e-05 lr_backbone=6.667216e-05 elapsed=21:56:13 avg_step=3.74s recent_step=2.79s eta=30:01:18 +2026-09-15 12:27:10 | INFO | train rank=0 | step=21120/50000 progress=42.2% loss=0.085065 lr=6.664098e-05 lr_backbone=6.664098e-05 elapsed=21:56:48 avg_step=3.74s recent_step=3.50s eta=30:00:37 +2026-09-15 12:27:42 | INFO | train rank=0 | step=21130/50000 progress=42.3% loss=0.125434 lr=6.660979e-05 lr_backbone=6.660979e-05 elapsed=21:57:20 avg_step=3.74s recent_step=3.26s eta=29:59:53 +2026-09-15 12:28:31 | INFO | train rank=0 | step=21140/50000 progress=42.3% loss=0.087418 lr=6.657859e-05 lr_backbone=6.657859e-05 elapsed=21:58:09 avg_step=3.74s recent_step=4.89s eta=29:59:31 +2026-09-15 12:29:01 | INFO | train rank=0 | step=21150/50000 progress=42.3% loss=0.090720 lr=6.654739e-05 lr_backbone=6.654739e-05 elapsed=21:58:39 avg_step=3.74s recent_step=2.97s eta=29:58:44 +2026-09-15 12:29:33 | INFO | train rank=0 | step=21160/50000 progress=42.3% loss=0.113901 lr=6.651618e-05 lr_backbone=6.651618e-05 elapsed=21:59:11 avg_step=3.74s recent_step=3.18s eta=29:57:59 +2026-09-15 12:30:20 | INFO | train rank=0 | step=21170/50000 progress=42.3% loss=0.088581 lr=6.648497e-05 lr_backbone=6.648497e-05 elapsed=21:59:58 avg_step=3.74s recent_step=4.75s eta=29:57:35 +2026-09-15 12:31:04 | INFO | train rank=0 | step=21180/50000 progress=42.4% loss=0.092079 lr=6.645374e-05 lr_backbone=6.645374e-05 elapsed=22:00:42 avg_step=3.74s recent_step=4.40s eta=29:57:06 +2026-09-15 12:31:41 | INFO | train rank=0 | step=21190/50000 progress=42.4% loss=0.090761 lr=6.642251e-05 lr_backbone=6.642251e-05 elapsed=22:01:19 avg_step=3.74s recent_step=3.67s eta=29:56:28 +2026-09-15 12:32:18 | INFO | train rank=0 | step=21200/50000 progress=42.4% loss=0.093316 lr=6.639127e-05 lr_backbone=6.639127e-05 elapsed=22:01:56 avg_step=3.74s recent_step=3.75s eta=29:55:51 +2026-09-15 12:32:50 | INFO | train rank=0 | step=21210/50000 progress=42.4% loss=0.109701 lr=6.636003e-05 lr_backbone=6.636003e-05 elapsed=22:02:28 avg_step=3.74s recent_step=3.11s eta=29:55:05 +2026-09-15 12:33:44 | INFO | train rank=0 | step=21220/50000 progress=42.4% loss=0.076792 lr=6.632877e-05 lr_backbone=6.632877e-05 elapsed=22:03:22 avg_step=3.74s recent_step=5.41s eta=29:54:50 +2026-09-15 12:34:19 | INFO | train rank=0 | step=21230/50000 progress=42.5% loss=0.084824 lr=6.629751e-05 lr_backbone=6.629751e-05 elapsed=22:03:57 avg_step=3.74s recent_step=3.57s eta=29:54:10 +2026-09-15 12:34:53 | INFO | train rank=0 | step=21240/50000 progress=42.5% loss=0.123027 lr=6.626625e-05 lr_backbone=6.626625e-05 elapsed=22:04:31 avg_step=3.74s recent_step=3.36s eta=29:53:28 +2026-09-15 12:35:32 | INFO | train rank=0 | step=21250/50000 progress=42.5% loss=0.096357 lr=6.623497e-05 lr_backbone=6.623497e-05 elapsed=22:05:10 avg_step=3.74s recent_step=3.92s eta=29:52:53 +2026-09-15 12:36:22 | INFO | train rank=0 | step=21260/50000 progress=42.5% loss=0.125035 lr=6.620369e-05 lr_backbone=6.620369e-05 elapsed=22:06:00 avg_step=3.74s recent_step=4.98s eta=29:52:32 +2026-09-15 12:36:51 | INFO | train rank=0 | step=21270/50000 progress=42.5% loss=0.088392 lr=6.617240e-05 lr_backbone=6.617240e-05 elapsed=22:06:29 avg_step=3.74s recent_step=2.89s eta=29:51:43 +2026-09-15 12:37:22 | INFO | train rank=0 | step=21280/50000 progress=42.6% loss=0.099195 lr=6.614111e-05 lr_backbone=6.614111e-05 elapsed=22:07:00 avg_step=3.74s recent_step=3.11s eta=29:50:57 +2026-09-15 12:38:00 | INFO | train rank=0 | step=21290/50000 progress=42.6% loss=0.103866 lr=6.610981e-05 lr_backbone=6.610981e-05 elapsed=22:07:38 avg_step=3.74s recent_step=3.78s eta=29:50:20 +2026-09-15 12:38:44 | INFO | train rank=0 | step=21300/50000 progress=42.6% loss=0.091073 lr=6.607850e-05 lr_backbone=6.607850e-05 elapsed=22:08:22 avg_step=3.74s recent_step=4.45s eta=29:49:52 +2026-09-15 12:39:18 | INFO | train rank=0 | step=21310/50000 progress=42.6% loss=0.116753 lr=6.604718e-05 lr_backbone=6.604718e-05 elapsed=22:08:56 avg_step=3.74s recent_step=3.38s eta=29:49:10 +2026-09-15 12:39:51 | INFO | train rank=0 | step=21320/50000 progress=42.6% loss=0.094886 lr=6.601586e-05 lr_backbone=6.601586e-05 elapsed=22:09:29 avg_step=3.74s recent_step=3.29s eta=29:48:27 +2026-09-15 12:40:20 | INFO | train rank=0 | step=21330/50000 progress=42.7% loss=0.094945 lr=6.598453e-05 lr_backbone=6.598453e-05 elapsed=22:09:58 avg_step=3.74s recent_step=2.87s eta=29:47:37 +2026-09-15 12:41:14 | INFO | train rank=0 | step=21340/50000 progress=42.7% loss=0.088967 lr=6.595319e-05 lr_backbone=6.595319e-05 elapsed=22:10:52 avg_step=3.74s recent_step=5.47s eta=29:47:23 +2026-09-15 12:41:45 | INFO | train rank=0 | step=21350/50000 progress=42.7% loss=0.124746 lr=6.592184e-05 lr_backbone=6.592184e-05 elapsed=22:11:23 avg_step=3.74s recent_step=3.05s eta=29:46:36 +2026-09-15 12:42:26 | INFO | train rank=0 | step=21360/50000 progress=42.7% loss=0.093925 lr=6.589049e-05 lr_backbone=6.589049e-05 elapsed=22:12:04 avg_step=3.74s recent_step=4.15s eta=29:46:04 +2026-09-15 12:43:01 | INFO | train rank=0 | step=21370/50000 progress=42.7% loss=0.101697 lr=6.585913e-05 lr_backbone=6.585913e-05 elapsed=22:12:39 avg_step=3.74s recent_step=3.50s eta=29:45:24 +2026-09-15 12:43:48 | INFO | train rank=0 | step=21380/50000 progress=42.8% loss=0.101872 lr=6.582777e-05 lr_backbone=6.582777e-05 elapsed=22:13:26 avg_step=3.74s recent_step=4.69s eta=29:44:59 +2026-09-15 12:44:21 | INFO | train rank=0 | step=21390/50000 progress=42.8% loss=0.105037 lr=6.579640e-05 lr_backbone=6.579640e-05 elapsed=22:13:59 avg_step=3.74s recent_step=3.27s eta=29:44:15 +2026-09-15 12:44:54 | INFO | train rank=0 | step=21400/50000 progress=42.8% loss=0.090854 lr=6.576502e-05 lr_backbone=6.576502e-05 elapsed=22:14:32 avg_step=3.74s recent_step=3.35s eta=29:43:33 +2026-09-15 12:45:30 | INFO | train rank=0 | step=21410/50000 progress=42.8% loss=0.082965 lr=6.573363e-05 lr_backbone=6.573363e-05 elapsed=22:15:08 avg_step=3.74s recent_step=3.54s eta=29:42:53 +2026-09-15 12:46:24 | INFO | train rank=0 | step=21420/50000 progress=42.8% loss=0.114019 lr=6.570224e-05 lr_backbone=6.570224e-05 elapsed=22:16:02 avg_step=3.74s recent_step=5.43s eta=29:42:38 +2026-09-15 12:46:55 | INFO | train rank=0 | step=21430/50000 progress=42.9% loss=0.108850 lr=6.567084e-05 lr_backbone=6.567084e-05 elapsed=22:16:33 avg_step=3.74s recent_step=3.11s eta=29:41:52 +2026-09-15 12:47:23 | INFO | train rank=0 | step=21440/50000 progress=42.9% loss=0.096308 lr=6.563943e-05 lr_backbone=6.563943e-05 elapsed=22:17:01 avg_step=3.74s recent_step=2.75s eta=29:41:01 +2026-09-15 12:47:58 | INFO | train rank=0 | step=21450/50000 progress=42.9% loss=0.086959 lr=6.560802e-05 lr_backbone=6.560802e-05 elapsed=22:17:36 avg_step=3.74s recent_step=3.52s eta=29:40:21 +2026-09-15 12:48:52 | INFO | train rank=0 | step=21460/50000 progress=42.9% loss=0.085433 lr=6.557660e-05 lr_backbone=6.557660e-05 elapsed=22:18:30 avg_step=3.74s recent_step=5.42s eta=29:40:06 +2026-09-15 12:49:22 | INFO | train rank=0 | step=21470/50000 progress=42.9% loss=0.084843 lr=6.554517e-05 lr_backbone=6.554517e-05 elapsed=22:19:00 avg_step=3.74s recent_step=3.01s eta=29:39:19 +2026-09-15 12:50:03 | INFO | train rank=0 | step=21480/50000 progress=43.0% loss=0.108094 lr=6.551374e-05 lr_backbone=6.551374e-05 elapsed=22:19:41 avg_step=3.74s recent_step=4.10s eta=29:38:46 +2026-09-15 12:50:33 | INFO | train rank=0 | step=21490/50000 progress=43.0% loss=0.108428 lr=6.548230e-05 lr_backbone=6.548230e-05 elapsed=22:20:11 avg_step=3.74s recent_step=3.00s eta=29:37:59 +2026-09-15 12:51:16 | INFO | train rank=0 | step=21500/50000 progress=43.0% loss=0.084865 lr=6.545085e-05 lr_backbone=6.545085e-05 elapsed=22:20:54 avg_step=3.74s recent_step=4.33s eta=29:37:29 +2026-09-15 12:51:49 | INFO | train rank=0 | step=21510/50000 progress=43.0% loss=0.101529 lr=6.541940e-05 lr_backbone=6.541940e-05 elapsed=22:21:28 avg_step=3.74s recent_step=3.31s eta=29:36:46 +2026-09-15 12:52:26 | INFO | train rank=0 | step=21520/50000 progress=43.0% loss=0.104464 lr=6.538793e-05 lr_backbone=6.538793e-05 elapsed=22:22:04 avg_step=3.74s recent_step=3.69s eta=29:36:08 +2026-09-15 12:53:01 | INFO | train rank=0 | step=21530/50000 progress=43.1% loss=0.086200 lr=6.535647e-05 lr_backbone=6.535647e-05 elapsed=22:22:39 avg_step=3.74s recent_step=3.46s eta=29:35:27 +2026-09-15 12:54:01 | INFO | train rank=0 | step=21540/50000 progress=43.1% loss=0.098861 lr=6.532499e-05 lr_backbone=6.532499e-05 elapsed=22:23:39 avg_step=3.74s recent_step=5.98s eta=29:35:19 +2026-09-15 12:54:21 | INFO | train rank=0 | step=21550/50000 progress=43.1% loss=0.076538 lr=6.529351e-05 lr_backbone=6.529351e-05 elapsed=22:23:59 avg_step=3.74s recent_step=2.04s eta=29:34:19 +2026-09-15 12:54:55 | INFO | train rank=0 | step=21560/50000 progress=43.1% loss=0.077472 lr=6.526202e-05 lr_backbone=6.526202e-05 elapsed=22:24:33 avg_step=3.74s recent_step=3.38s eta=29:33:37 +2026-09-15 12:55:35 | INFO | train rank=0 | step=21570/50000 progress=43.1% loss=0.106687 lr=6.523053e-05 lr_backbone=6.523053e-05 elapsed=22:25:13 avg_step=3.74s recent_step=3.96s eta=29:33:02 +2026-09-15 12:56:19 | INFO | train rank=0 | step=21580/50000 progress=43.2% loss=0.097914 lr=6.519903e-05 lr_backbone=6.519903e-05 elapsed=22:25:57 avg_step=3.74s recent_step=4.48s eta=29:32:34 +2026-09-15 12:56:56 | INFO | train rank=0 | step=21590/50000 progress=43.2% loss=0.098212 lr=6.516752e-05 lr_backbone=6.516752e-05 elapsed=22:26:34 avg_step=3.74s recent_step=3.63s eta=29:31:55 +2026-09-15 12:57:31 | INFO | train rank=0 | step=21600/50000 progress=43.2% loss=0.085023 lr=6.513601e-05 lr_backbone=6.513601e-05 elapsed=22:27:09 avg_step=3.74s recent_step=3.51s eta=29:31:15 +2026-09-15 12:58:21 | INFO | train rank=0 | step=21610/50000 progress=43.2% loss=0.094751 lr=6.510448e-05 lr_backbone=6.510448e-05 elapsed=22:27:59 avg_step=3.74s recent_step=5.07s eta=29:30:55 +2026-09-15 12:58:53 | INFO | train rank=0 | step=21620/50000 progress=43.2% loss=0.093259 lr=6.507296e-05 lr_backbone=6.507296e-05 elapsed=22:28:31 avg_step=3.74s recent_step=3.20s eta=29:30:11 +2026-09-15 12:59:19 | INFO | train rank=0 | step=21630/50000 progress=43.3% loss=0.099433 lr=6.504142e-05 lr_backbone=6.504142e-05 elapsed=22:28:57 avg_step=3.74s recent_step=2.54s eta=29:29:17 +2026-09-15 12:59:54 | INFO | train rank=0 | step=21640/50000 progress=43.3% loss=0.098181 lr=6.500988e-05 lr_backbone=6.500988e-05 elapsed=22:29:32 avg_step=3.74s recent_step=3.51s eta=29:28:37 +2026-09-15 13:00:41 | INFO | train rank=0 | step=21650/50000 progress=43.3% loss=0.091155 lr=6.497833e-05 lr_backbone=6.497833e-05 elapsed=22:30:19 avg_step=3.74s recent_step=4.69s eta=29:28:12 +2026-09-15 13:01:22 | INFO | train rank=0 | step=21660/50000 progress=43.3% loss=0.085687 lr=6.494678e-05 lr_backbone=6.494678e-05 elapsed=22:31:00 avg_step=3.74s recent_step=4.10s eta=29:27:39 +2026-09-15 13:02:01 | INFO | train rank=0 | step=21670/50000 progress=43.3% loss=0.118064 lr=6.491522e-05 lr_backbone=6.491522e-05 elapsed=22:31:39 avg_step=3.74s recent_step=3.87s eta=29:27:03 +2026-09-15 13:02:38 | INFO | train rank=0 | step=21680/50000 progress=43.4% loss=0.113102 lr=6.488365e-05 lr_backbone=6.488365e-05 elapsed=22:32:16 avg_step=3.74s recent_step=3.79s eta=29:26:27 +2026-09-15 13:03:24 | INFO | train rank=0 | step=21690/50000 progress=43.4% loss=0.106731 lr=6.485208e-05 lr_backbone=6.485208e-05 elapsed=22:33:02 avg_step=3.74s recent_step=4.56s eta=29:26:00 +2026-09-15 13:03:51 | INFO | train rank=0 | step=21700/50000 progress=43.4% loss=0.102283 lr=6.482050e-05 lr_backbone=6.482050e-05 elapsed=22:33:29 avg_step=3.74s recent_step=2.68s eta=29:25:09 +2026-09-15 13:04:22 | INFO | train rank=0 | step=21710/50000 progress=43.4% loss=0.148115 lr=6.478891e-05 lr_backbone=6.478891e-05 elapsed=22:34:00 avg_step=3.74s recent_step=3.13s eta=29:24:23 +2026-09-15 13:04:52 | INFO | train rank=0 | step=21720/50000 progress=43.4% loss=0.092941 lr=6.475732e-05 lr_backbone=6.475732e-05 elapsed=22:34:30 avg_step=3.74s recent_step=2.94s eta=29:23:35 +2026-09-15 13:05:44 | INFO | train rank=0 | step=21730/50000 progress=43.5% loss=0.069242 lr=6.472572e-05 lr_backbone=6.472572e-05 elapsed=22:35:22 avg_step=3.74s recent_step=5.21s eta=29:23:17 +2026-09-15 13:06:22 | INFO | train rank=0 | step=21740/50000 progress=43.5% loss=0.117553 lr=6.469411e-05 lr_backbone=6.469411e-05 elapsed=22:36:00 avg_step=3.74s recent_step=3.78s eta=29:22:40 +2026-09-15 13:06:59 | INFO | train rank=0 | step=21750/50000 progress=43.5% loss=0.128441 lr=6.466250e-05 lr_backbone=6.466250e-05 elapsed=22:36:37 avg_step=3.74s recent_step=3.72s eta=29:22:02 +2026-09-15 13:07:33 | INFO | train rank=0 | step=21760/50000 progress=43.5% loss=0.116077 lr=6.463088e-05 lr_backbone=6.463088e-05 elapsed=22:37:11 avg_step=3.74s recent_step=3.45s eta=29:21:21 +2026-09-15 13:08:31 | INFO | train rank=0 | step=21770/50000 progress=43.5% loss=0.089258 lr=6.459926e-05 lr_backbone=6.459926e-05 elapsed=22:38:10 avg_step=3.74s recent_step=5.82s eta=29:21:11 +2026-09-15 13:09:01 | INFO | train rank=0 | step=21780/50000 progress=43.6% loss=0.107485 lr=6.456763e-05 lr_backbone=6.456763e-05 elapsed=22:38:39 avg_step=3.74s recent_step=2.99s eta=29:20:24 +2026-09-15 13:09:32 | INFO | train rank=0 | step=21790/50000 progress=43.6% loss=0.107120 lr=6.453599e-05 lr_backbone=6.453599e-05 elapsed=22:39:10 avg_step=3.74s recent_step=3.02s eta=29:19:37 +2026-09-15 13:10:13 | INFO | train rank=0 | step=21800/50000 progress=43.6% loss=0.098311 lr=6.450434e-05 lr_backbone=6.450434e-05 elapsed=22:39:51 avg_step=3.74s recent_step=4.19s eta=29:19:05 +2026-09-15 13:10:58 | INFO | train rank=0 | step=21810/50000 progress=43.6% loss=0.135553 lr=6.447269e-05 lr_backbone=6.447269e-05 elapsed=22:40:37 avg_step=3.74s recent_step=4.51s eta=29:18:37 +2026-09-15 13:11:22 | INFO | train rank=0 | step=21820/50000 progress=43.6% loss=0.104294 lr=6.444104e-05 lr_backbone=6.444104e-05 elapsed=22:41:00 avg_step=3.74s recent_step=2.39s eta=29:17:43 +2026-09-15 13:11:52 | INFO | train rank=0 | step=21830/50000 progress=43.7% loss=0.079312 lr=6.440937e-05 lr_backbone=6.440937e-05 elapsed=22:41:30 avg_step=3.74s recent_step=2.95s eta=29:16:55 +2026-09-15 13:12:22 | INFO | train rank=0 | step=21840/50000 progress=43.7% loss=0.109250 lr=6.437770e-05 lr_backbone=6.437770e-05 elapsed=22:42:00 avg_step=3.74s recent_step=3.04s eta=29:16:08 +2026-09-15 13:13:20 | INFO | train rank=0 | step=21850/50000 progress=43.7% loss=0.084227 lr=6.434603e-05 lr_backbone=6.434603e-05 elapsed=22:42:58 avg_step=3.74s recent_step=5.79s eta=29:15:57 +2026-09-15 13:13:56 | INFO | train rank=0 | step=21860/50000 progress=43.7% loss=0.085725 lr=6.431435e-05 lr_backbone=6.431435e-05 elapsed=22:43:34 avg_step=3.74s recent_step=3.56s eta=29:15:18 +2026-09-15 13:14:27 | INFO | train rank=0 | step=21870/50000 progress=43.7% loss=0.078504 lr=6.428266e-05 lr_backbone=6.428266e-05 elapsed=22:44:05 avg_step=3.74s recent_step=3.08s eta=29:14:32 +2026-09-15 13:15:01 | INFO | train rank=0 | step=21880/50000 progress=43.8% loss=0.111465 lr=6.425096e-05 lr_backbone=6.425096e-05 elapsed=22:44:39 avg_step=3.74s recent_step=3.38s eta=29:13:50 +2026-09-15 13:15:49 | INFO | train rank=0 | step=21890/50000 progress=43.8% loss=0.091170 lr=6.421926e-05 lr_backbone=6.421926e-05 elapsed=22:45:27 avg_step=3.74s recent_step=4.89s eta=29:13:27 +2026-09-15 13:16:21 | INFO | train rank=0 | step=21900/50000 progress=43.8% loss=0.078462 lr=6.418756e-05 lr_backbone=6.418756e-05 elapsed=22:45:59 avg_step=3.74s recent_step=3.14s eta=29:12:42 +2026-09-15 13:17:00 | INFO | train rank=0 | step=21910/50000 progress=43.8% loss=0.114338 lr=6.415584e-05 lr_backbone=6.415584e-05 elapsed=22:46:38 avg_step=3.74s recent_step=3.88s eta=29:12:06 +2026-09-15 13:17:34 | INFO | train rank=0 | step=21920/50000 progress=43.8% loss=0.092971 lr=6.412412e-05 lr_backbone=6.412412e-05 elapsed=22:47:12 avg_step=3.74s recent_step=3.45s eta=29:11:25 +2026-09-15 13:18:23 | INFO | train rank=0 | step=21930/50000 progress=43.9% loss=0.085271 lr=6.409240e-05 lr_backbone=6.409240e-05 elapsed=22:48:01 avg_step=3.74s recent_step=4.88s eta=29:11:02 +2026-09-15 13:19:02 | INFO | train rank=0 | step=21940/50000 progress=43.9% loss=0.116607 lr=6.406066e-05 lr_backbone=6.406066e-05 elapsed=22:48:40 avg_step=3.74s recent_step=3.87s eta=29:10:26 +2026-09-15 13:19:34 | INFO | train rank=0 | step=21950/50000 progress=43.9% loss=0.080712 lr=6.402893e-05 lr_backbone=6.402893e-05 elapsed=22:49:12 avg_step=3.74s recent_step=3.28s eta=29:09:43 +2026-09-15 13:20:12 | INFO | train rank=0 | step=21960/50000 progress=43.9% loss=0.084877 lr=6.399718e-05 lr_backbone=6.399718e-05 elapsed=22:49:50 avg_step=3.74s recent_step=3.79s eta=29:09:06 +2026-09-15 13:21:04 | INFO | train rank=0 | step=21970/50000 progress=43.9% loss=0.080750 lr=6.396543e-05 lr_backbone=6.396543e-05 elapsed=22:50:42 avg_step=3.74s recent_step=5.19s eta=29:08:47 +2026-09-15 13:21:37 | INFO | train rank=0 | step=21980/50000 progress=44.0% loss=0.083718 lr=6.393368e-05 lr_backbone=6.393368e-05 elapsed=22:51:15 avg_step=3.74s recent_step=3.26s eta=29:08:04 +2026-09-15 13:22:13 | INFO | train rank=0 | step=21990/50000 progress=44.0% loss=0.075213 lr=6.390191e-05 lr_backbone=6.390191e-05 elapsed=22:51:51 avg_step=3.74s recent_step=3.64s eta=29:07:25 +2026-09-15 13:22:51 | INFO | train rank=0 | step=22000/50000 progress=44.0% loss=0.087460 lr=6.387015e-05 lr_backbone=6.387015e-05 elapsed=22:52:29 avg_step=3.74s recent_step=3.75s eta=29:06:48 +2026-09-15 13:23:17 | INFO | train rank=0 | step=22010/50000 progress=44.0% loss=0.082906 lr=6.383837e-05 lr_backbone=6.383837e-05 elapsed=22:52:55 avg_step=3.74s recent_step=2.65s eta=29:05:56 +2026-09-15 13:23:50 | INFO | train rank=0 | step=22020/50000 progress=44.0% loss=0.081414 lr=6.380659e-05 lr_backbone=6.380659e-05 elapsed=22:53:28 avg_step=3.74s recent_step=3.25s eta=29:05:13 +2026-09-15 13:24:19 | INFO | train rank=0 | step=22030/50000 progress=44.1% loss=0.113051 lr=6.377480e-05 lr_backbone=6.377480e-05 elapsed=22:53:57 avg_step=3.74s recent_step=2.94s eta=29:04:25 +2026-09-15 13:24:51 | INFO | train rank=0 | step=22040/50000 progress=44.1% loss=0.097901 lr=6.374301e-05 lr_backbone=6.374301e-05 elapsed=22:54:29 avg_step=3.74s recent_step=3.14s eta=29:03:40 +2026-09-15 13:25:42 | INFO | train rank=0 | step=22050/50000 progress=44.1% loss=0.120778 lr=6.371121e-05 lr_backbone=6.371121e-05 elapsed=22:55:20 avg_step=3.74s recent_step=5.16s eta=29:03:20 +2026-09-15 13:26:13 | INFO | train rank=0 | step=22060/50000 progress=44.1% loss=0.080107 lr=6.367941e-05 lr_backbone=6.367941e-05 elapsed=22:55:51 avg_step=3.74s recent_step=3.05s eta=29:02:34 +2026-09-15 13:26:42 | INFO | train rank=0 | step=22070/50000 progress=44.1% loss=0.085964 lr=6.364760e-05 lr_backbone=6.364760e-05 elapsed=22:56:20 avg_step=3.74s recent_step=2.97s eta=29:01:47 +2026-09-15 13:27:16 | INFO | train rank=0 | step=22080/50000 progress=44.2% loss=0.100200 lr=6.361578e-05 lr_backbone=6.361578e-05 elapsed=22:56:54 avg_step=3.74s recent_step=3.41s eta=29:01:06 +2026-09-15 13:28:03 | INFO | train rank=0 | step=22090/50000 progress=44.2% loss=0.110969 lr=6.358396e-05 lr_backbone=6.358396e-05 elapsed=22:57:41 avg_step=3.74s recent_step=4.66s eta=29:00:40 +2026-09-15 13:28:39 | INFO | train rank=0 | step=22100/50000 progress=44.2% loss=0.086258 lr=6.355213e-05 lr_backbone=6.355213e-05 elapsed=22:58:17 avg_step=3.74s recent_step=3.59s eta=29:00:00 +2026-09-15 13:29:08 | INFO | train rank=0 | step=22110/50000 progress=44.2% loss=0.087375 lr=6.352029e-05 lr_backbone=6.352029e-05 elapsed=22:58:46 avg_step=3.74s recent_step=2.87s eta=28:59:12 +2026-09-15 13:29:38 | INFO | train rank=0 | step=22120/50000 progress=44.2% loss=0.104679 lr=6.348845e-05 lr_backbone=6.348845e-05 elapsed=22:59:16 avg_step=3.74s recent_step=3.06s eta=28:58:26 +2026-09-15 13:30:30 | INFO | train rank=0 | step=22130/50000 progress=44.3% loss=0.088382 lr=6.345661e-05 lr_backbone=6.345661e-05 elapsed=23:00:08 avg_step=3.74s recent_step=5.20s eta=28:58:07 +2026-09-15 13:31:05 | INFO | train rank=0 | step=22140/50000 progress=44.3% loss=0.110209 lr=6.342476e-05 lr_backbone=6.342476e-05 elapsed=23:00:43 avg_step=3.74s recent_step=3.49s eta=28:57:26 +2026-09-15 13:31:38 | INFO | train rank=0 | step=22150/50000 progress=44.3% loss=0.105929 lr=6.339290e-05 lr_backbone=6.339290e-05 elapsed=23:01:16 avg_step=3.74s recent_step=3.23s eta=28:56:43 +2026-09-15 13:32:22 | INFO | train rank=0 | step=22160/50000 progress=44.3% loss=0.118464 lr=6.336103e-05 lr_backbone=6.336103e-05 elapsed=23:02:00 avg_step=3.74s recent_step=4.47s eta=28:56:14 +2026-09-15 13:33:02 | INFO | train rank=0 | step=22170/50000 progress=44.3% loss=0.080283 lr=6.332916e-05 lr_backbone=6.332916e-05 elapsed=23:02:40 avg_step=3.74s recent_step=4.01s eta=28:55:40 +2026-09-15 13:33:36 | INFO | train rank=0 | step=22180/50000 progress=44.4% loss=0.083666 lr=6.329729e-05 lr_backbone=6.329729e-05 elapsed=23:03:14 avg_step=3.74s recent_step=3.41s eta=28:54:59 +2026-09-15 13:34:18 | INFO | train rank=0 | step=22190/50000 progress=44.4% loss=0.107960 lr=6.326541e-05 lr_backbone=6.326541e-05 elapsed=23:03:56 avg_step=3.74s recent_step=4.12s eta=28:54:26 +2026-09-15 13:35:02 | INFO | train rank=0 | step=22200/50000 progress=44.4% loss=0.100404 lr=6.323352e-05 lr_backbone=6.323352e-05 elapsed=23:04:40 avg_step=3.74s recent_step=4.47s eta=28:53:58 +2026-09-15 13:35:39 | INFO | train rank=0 | step=22210/50000 progress=44.4% loss=0.094153 lr=6.320163e-05 lr_backbone=6.320163e-05 elapsed=23:05:17 avg_step=3.74s recent_step=3.61s eta=28:53:19 +2026-09-15 13:36:14 | INFO | train rank=0 | step=22220/50000 progress=44.4% loss=0.092145 lr=6.316973e-05 lr_backbone=6.316973e-05 elapsed=23:05:52 avg_step=3.74s recent_step=3.59s eta=28:52:39 +2026-09-15 13:36:43 | INFO | train rank=0 | step=22230/50000 progress=44.5% loss=0.098263 lr=6.313782e-05 lr_backbone=6.313782e-05 elapsed=23:06:21 avg_step=3.74s recent_step=2.87s eta=28:51:51 +2026-09-15 13:37:28 | INFO | train rank=0 | step=22240/50000 progress=44.5% loss=0.104013 lr=6.310591e-05 lr_backbone=6.310591e-05 elapsed=23:07:06 avg_step=3.74s recent_step=4.49s eta=28:51:23 +2026-09-15 13:38:03 | INFO | train rank=0 | step=22250/50000 progress=44.5% loss=0.085279 lr=6.307400e-05 lr_backbone=6.307400e-05 elapsed=23:07:41 avg_step=3.74s recent_step=3.50s eta=28:50:43 +2026-09-15 13:38:40 | INFO | train rank=0 | step=22260/50000 progress=44.5% loss=0.103618 lr=6.304208e-05 lr_backbone=6.304208e-05 elapsed=23:08:18 avg_step=3.74s recent_step=3.71s eta=28:50:05 +2026-09-15 13:39:08 | INFO | train rank=0 | step=22270/50000 progress=44.5% loss=0.100828 lr=6.301015e-05 lr_backbone=6.301015e-05 elapsed=23:08:46 avg_step=3.74s recent_step=2.79s eta=28:49:15 +2026-09-15 13:39:52 | INFO | train rank=0 | step=22280/50000 progress=44.6% loss=0.111057 lr=6.297821e-05 lr_backbone=6.297821e-05 elapsed=23:09:30 avg_step=3.74s recent_step=4.42s eta=28:48:47 +2026-09-15 13:40:27 | INFO | train rank=0 | step=22290/50000 progress=44.6% loss=0.115786 lr=6.294628e-05 lr_backbone=6.294628e-05 elapsed=23:10:05 avg_step=3.74s recent_step=3.52s eta=28:48:06 +2026-09-15 13:41:12 | INFO | train rank=0 | step=22300/50000 progress=44.6% loss=0.109511 lr=6.291433e-05 lr_backbone=6.291433e-05 elapsed=23:10:50 avg_step=3.74s recent_step=4.41s eta=28:47:37 +2026-09-15 13:41:49 | INFO | train rank=0 | step=22310/50000 progress=44.6% loss=0.079040 lr=6.288238e-05 lr_backbone=6.288238e-05 elapsed=23:11:27 avg_step=3.74s recent_step=3.73s eta=28:47:00 +2026-09-15 13:42:28 | INFO | train rank=0 | step=22320/50000 progress=44.6% loss=0.093250 lr=6.285043e-05 lr_backbone=6.285043e-05 elapsed=23:12:06 avg_step=3.74s recent_step=3.88s eta=28:46:24 +2026-09-15 13:42:57 | INFO | train rank=0 | step=22330/50000 progress=44.7% loss=0.116142 lr=6.281846e-05 lr_backbone=6.281846e-05 elapsed=23:12:35 avg_step=3.74s recent_step=2.90s eta=28:45:36 +2026-09-15 13:43:41 | INFO | train rank=0 | step=22340/50000 progress=44.7% loss=0.106148 lr=6.278650e-05 lr_backbone=6.278650e-05 elapsed=23:13:19 avg_step=3.74s recent_step=4.42s eta=28:45:07 +2026-09-15 13:44:23 | INFO | train rank=0 | step=22350/50000 progress=44.7% loss=0.116327 lr=6.275452e-05 lr_backbone=6.275452e-05 elapsed=23:14:01 avg_step=3.74s recent_step=4.25s eta=28:44:36 +2026-09-15 13:45:01 | INFO | train rank=0 | step=22360/50000 progress=44.7% loss=0.111421 lr=6.272255e-05 lr_backbone=6.272255e-05 elapsed=23:14:39 avg_step=3.74s recent_step=3.75s eta=28:43:59 +2026-09-15 13:45:29 | INFO | train rank=0 | step=22370/50000 progress=44.7% loss=0.104056 lr=6.269056e-05 lr_backbone=6.269056e-05 elapsed=23:15:07 avg_step=3.74s recent_step=2.80s eta=28:43:10 +2026-09-15 13:46:11 | INFO | train rank=0 | step=22380/50000 progress=44.8% loss=0.120097 lr=6.265857e-05 lr_backbone=6.265857e-05 elapsed=23:15:49 avg_step=3.74s recent_step=4.16s eta=28:42:37 +2026-09-15 13:46:46 | INFO | train rank=0 | step=22390/50000 progress=44.8% loss=0.111911 lr=6.262658e-05 lr_backbone=6.262658e-05 elapsed=23:16:25 avg_step=3.74s recent_step=3.60s eta=28:41:58 +2026-09-15 13:47:40 | INFO | train rank=0 | step=22400/50000 progress=44.8% loss=0.092321 lr=6.259458e-05 lr_backbone=6.259458e-05 elapsed=23:17:18 avg_step=3.74s recent_step=5.37s eta=28:41:41 +2026-09-15 13:48:11 | INFO | train rank=0 | step=22410/50000 progress=44.8% loss=0.093930 lr=6.256257e-05 lr_backbone=6.256257e-05 elapsed=23:17:49 avg_step=3.74s recent_step=3.04s eta=28:40:55 +2026-09-15 13:48:41 | INFO | train rank=0 | step=22420/50000 progress=44.8% loss=0.119383 lr=6.253056e-05 lr_backbone=6.253056e-05 elapsed=23:18:19 avg_step=3.74s recent_step=3.06s eta=28:40:09 +2026-09-15 13:49:17 | INFO | train rank=0 | step=22430/50000 progress=44.9% loss=0.097628 lr=6.249854e-05 lr_backbone=6.249854e-05 elapsed=23:18:55 avg_step=3.74s recent_step=3.53s eta=28:39:29 +2026-09-15 13:50:05 | INFO | train rank=0 | step=22440/50000 progress=44.9% loss=0.110442 lr=6.246652e-05 lr_backbone=6.246652e-05 elapsed=23:19:43 avg_step=3.74s recent_step=4.81s eta=28:39:05 +2026-09-15 13:50:28 | INFO | train rank=0 | step=22450/50000 progress=44.9% loss=0.100811 lr=6.243449e-05 lr_backbone=6.243449e-05 elapsed=23:20:06 avg_step=3.74s recent_step=2.29s eta=28:38:09 +2026-09-15 13:51:17 | INFO | train rank=0 | step=22460/50000 progress=44.9% loss=0.089367 lr=6.240246e-05 lr_backbone=6.240246e-05 elapsed=23:20:55 avg_step=3.74s recent_step=4.95s eta=28:37:47 +2026-09-15 13:51:45 | INFO | train rank=0 | step=22470/50000 progress=44.9% loss=0.117690 lr=6.237042e-05 lr_backbone=6.237042e-05 elapsed=23:21:23 avg_step=3.74s recent_step=2.75s eta=28:36:57 +2026-09-15 13:52:23 | INFO | train rank=0 | step=22480/50000 progress=45.0% loss=0.093274 lr=6.233838e-05 lr_backbone=6.233838e-05 elapsed=23:22:01 avg_step=3.74s recent_step=3.87s eta=28:36:21 +2026-09-15 13:52:58 | INFO | train rank=0 | step=22490/50000 progress=45.0% loss=0.093792 lr=6.230633e-05 lr_backbone=6.230633e-05 elapsed=23:22:36 avg_step=3.74s recent_step=3.44s eta=28:35:40 +2026-09-15 13:53:44 | INFO | train rank=0 | step=22500/50000 progress=45.0% loss=0.089850 lr=6.227427e-05 lr_backbone=6.227427e-05 elapsed=23:23:22 avg_step=3.74s recent_step=4.59s eta=28:35:13 +2026-09-15 13:54:19 | INFO | train rank=0 | step=22510/50000 progress=45.0% loss=0.106345 lr=6.224221e-05 lr_backbone=6.224221e-05 elapsed=23:23:57 avg_step=3.74s recent_step=3.56s eta=28:34:33 +2026-09-15 13:54:53 | INFO | train rank=0 | step=22520/50000 progress=45.0% loss=0.096852 lr=6.221015e-05 lr_backbone=6.221015e-05 elapsed=23:24:31 avg_step=3.74s recent_step=3.37s eta=28:33:52 +2026-09-15 13:55:30 | INFO | train rank=0 | step=22530/50000 progress=45.1% loss=0.079330 lr=6.217808e-05 lr_backbone=6.217808e-05 elapsed=23:25:08 avg_step=3.74s recent_step=3.73s eta=28:33:14 +2026-09-15 13:56:06 | INFO | train rank=0 | step=22540/50000 progress=45.1% loss=0.110940 lr=6.214600e-05 lr_backbone=6.214600e-05 elapsed=23:25:44 avg_step=3.74s recent_step=3.59s eta=28:32:35 +2026-09-15 13:57:01 | INFO | train rank=0 | step=22550/50000 progress=45.1% loss=0.109741 lr=6.211392e-05 lr_backbone=6.211392e-05 elapsed=23:26:39 avg_step=3.74s recent_step=5.51s eta=28:32:19 +2026-09-15 13:57:32 | INFO | train rank=0 | step=22560/50000 progress=45.1% loss=0.082436 lr=6.208183e-05 lr_backbone=6.208183e-05 elapsed=23:27:10 avg_step=3.74s recent_step=3.09s eta=28:31:34 +2026-09-15 13:58:06 | INFO | train rank=0 | step=22570/50000 progress=45.1% loss=0.133051 lr=6.204974e-05 lr_backbone=6.204974e-05 elapsed=23:27:44 avg_step=3.74s recent_step=3.39s eta=28:30:52 +2026-09-15 13:58:40 | INFO | train rank=0 | step=22580/50000 progress=45.2% loss=0.092520 lr=6.201764e-05 lr_backbone=6.201764e-05 elapsed=23:28:18 avg_step=3.74s recent_step=3.40s eta=28:30:10 +2026-09-15 13:59:31 | INFO | train rank=0 | step=22590/50000 progress=45.2% loss=0.088952 lr=6.198554e-05 lr_backbone=6.198554e-05 elapsed=23:29:09 avg_step=3.74s recent_step=5.10s eta=28:29:49 +2026-09-15 14:00:10 | INFO | train rank=0 | step=22600/50000 progress=45.2% loss=0.096423 lr=6.195343e-05 lr_backbone=6.195343e-05 elapsed=23:29:48 avg_step=3.74s recent_step=3.86s eta=28:29:13 +2026-09-15 14:00:43 | INFO | train rank=0 | step=22610/50000 progress=45.2% loss=0.086818 lr=6.192132e-05 lr_backbone=6.192132e-05 elapsed=23:30:21 avg_step=3.74s recent_step=3.32s eta=28:28:31 +2026-09-15 14:01:17 | INFO | train rank=0 | step=22620/50000 progress=45.2% loss=0.098417 lr=6.188920e-05 lr_backbone=6.188920e-05 elapsed=23:30:55 avg_step=3.74s recent_step=3.39s eta=28:27:49 +2026-09-15 14:02:12 | INFO | train rank=0 | step=22630/50000 progress=45.3% loss=0.105127 lr=6.185708e-05 lr_backbone=6.185708e-05 elapsed=23:31:50 avg_step=3.74s recent_step=5.49s eta=28:27:33 +2026-09-15 14:02:48 | INFO | train rank=0 | step=22640/50000 progress=45.3% loss=0.112064 lr=6.182495e-05 lr_backbone=6.182495e-05 elapsed=23:32:26 avg_step=3.74s recent_step=3.68s eta=28:26:54 +2026-09-15 14:03:20 | INFO | train rank=0 | step=22650/50000 progress=45.3% loss=0.099078 lr=6.179282e-05 lr_backbone=6.179282e-05 elapsed=23:32:58 avg_step=3.74s recent_step=3.17s eta=28:26:10 +2026-09-15 14:03:52 | INFO | train rank=0 | step=22660/50000 progress=45.3% loss=0.091570 lr=6.176068e-05 lr_backbone=6.176068e-05 elapsed=23:33:30 avg_step=3.74s recent_step=3.19s eta=28:25:26 +2026-09-15 14:04:36 | INFO | train rank=0 | step=22670/50000 progress=45.3% loss=0.079269 lr=6.172853e-05 lr_backbone=6.172853e-05 elapsed=23:34:14 avg_step=3.74s recent_step=4.44s eta=28:24:57 +2026-09-15 14:05:09 | INFO | train rank=0 | step=22680/50000 progress=45.4% loss=0.109963 lr=6.169638e-05 lr_backbone=6.169638e-05 elapsed=23:34:47 avg_step=3.74s recent_step=3.26s eta=28:24:14 +2026-09-15 14:05:33 | INFO | train rank=0 | step=22690/50000 progress=45.4% loss=0.112718 lr=6.166423e-05 lr_backbone=6.166423e-05 elapsed=23:35:11 avg_step=3.74s recent_step=2.38s eta=28:23:20 +2026-09-15 14:06:07 | INFO | train rank=0 | step=22700/50000 progress=45.4% loss=0.131788 lr=6.163207e-05 lr_backbone=6.163207e-05 elapsed=23:35:45 avg_step=3.74s recent_step=3.36s eta=28:22:38 +2026-09-15 14:06:50 | INFO | train rank=0 | step=22710/50000 progress=45.4% loss=0.096323 lr=6.159991e-05 lr_backbone=6.159991e-05 elapsed=23:36:28 avg_step=3.74s recent_step=4.39s eta=28:22:08 +2026-09-15 14:07:21 | INFO | train rank=0 | step=22720/50000 progress=45.4% loss=0.107251 lr=6.156774e-05 lr_backbone=6.156774e-05 elapsed=23:36:59 avg_step=3.74s recent_step=3.10s eta=28:21:23 +2026-09-15 14:07:57 | INFO | train rank=0 | step=22730/50000 progress=45.5% loss=0.085469 lr=6.153556e-05 lr_backbone=6.153556e-05 elapsed=23:37:35 avg_step=3.74s recent_step=3.61s eta=28:20:44 +2026-09-15 14:08:35 | INFO | train rank=0 | step=22740/50000 progress=45.5% loss=0.087546 lr=6.150338e-05 lr_backbone=6.150338e-05 elapsed=23:38:13 avg_step=3.74s recent_step=3.74s eta=28:20:07 +2026-09-15 14:09:21 | INFO | train rank=0 | step=22750/50000 progress=45.5% loss=0.084633 lr=6.147120e-05 lr_backbone=6.147120e-05 elapsed=23:38:59 avg_step=3.74s recent_step=4.63s eta=28:19:40 +2026-09-15 14:09:49 | INFO | train rank=0 | step=22760/50000 progress=45.5% loss=0.082448 lr=6.143901e-05 lr_backbone=6.143901e-05 elapsed=23:39:27 avg_step=3.74s recent_step=2.82s eta=28:18:52 +2026-09-15 14:10:21 | INFO | train rank=0 | step=22770/50000 progress=45.5% loss=0.096420 lr=6.140681e-05 lr_backbone=6.140681e-05 elapsed=23:39:59 avg_step=3.74s recent_step=3.16s eta=28:18:07 +2026-09-15 14:10:53 | INFO | train rank=0 | step=22780/50000 progress=45.6% loss=0.108883 lr=6.137461e-05 lr_backbone=6.137461e-05 elapsed=23:40:31 avg_step=3.74s recent_step=3.25s eta=28:17:24 +2026-09-15 14:11:42 | INFO | train rank=0 | step=22790/50000 progress=45.6% loss=0.090758 lr=6.134241e-05 lr_backbone=6.134241e-05 elapsed=23:41:20 avg_step=3.74s recent_step=4.89s eta=28:17:00 +2026-09-15 14:12:22 | INFO | train rank=0 | step=22800/50000 progress=45.6% loss=0.137202 lr=6.131020e-05 lr_backbone=6.131020e-05 elapsed=23:42:00 avg_step=3.74s recent_step=4.01s eta=28:16:26 +2026-09-15 14:12:59 | INFO | train rank=0 | step=22810/50000 progress=45.6% loss=0.086648 lr=6.127798e-05 lr_backbone=6.127798e-05 elapsed=23:42:37 avg_step=3.74s recent_step=3.63s eta=28:15:47 +2026-09-15 14:13:35 | INFO | train rank=0 | step=22820/50000 progress=45.6% loss=0.109572 lr=6.124576e-05 lr_backbone=6.124576e-05 elapsed=23:43:13 avg_step=3.74s recent_step=3.61s eta=28:15:08 +2026-09-15 14:14:27 | INFO | train rank=0 | step=22830/50000 progress=45.7% loss=0.075487 lr=6.121354e-05 lr_backbone=6.121354e-05 elapsed=23:44:05 avg_step=3.74s recent_step=5.25s eta=28:14:49 +2026-09-15 14:15:00 | INFO | train rank=0 | step=22840/50000 progress=45.7% loss=0.094808 lr=6.118131e-05 lr_backbone=6.118131e-05 elapsed=23:44:38 avg_step=3.74s recent_step=3.23s eta=28:14:05 +2026-09-15 14:15:36 | INFO | train rank=0 | step=22850/50000 progress=45.7% loss=0.087638 lr=6.114907e-05 lr_backbone=6.114907e-05 elapsed=23:45:14 avg_step=3.74s recent_step=3.65s eta=28:13:27 +2026-09-15 14:16:07 | INFO | train rank=0 | step=22860/50000 progress=45.7% loss=0.124714 lr=6.111683e-05 lr_backbone=6.111683e-05 elapsed=23:45:45 avg_step=3.74s recent_step=3.11s eta=28:12:42 +2026-09-15 14:16:58 | INFO | train rank=0 | step=22870/50000 progress=45.7% loss=0.131405 lr=6.108459e-05 lr_backbone=6.108459e-05 elapsed=23:46:36 avg_step=3.74s recent_step=5.03s eta=28:12:20 +2026-09-15 14:17:28 | INFO | train rank=0 | step=22880/50000 progress=45.8% loss=0.092162 lr=6.105234e-05 lr_backbone=6.105234e-05 elapsed=23:47:06 avg_step=3.74s recent_step=3.01s eta=28:11:33 +2026-09-15 14:17:56 | INFO | train rank=0 | step=22890/50000 progress=45.8% loss=0.072295 lr=6.102009e-05 lr_backbone=6.102009e-05 elapsed=23:47:34 avg_step=3.74s recent_step=2.81s eta=28:10:45 +2026-09-15 14:18:33 | INFO | train rank=0 | step=22900/50000 progress=45.8% loss=0.094860 lr=6.098783e-05 lr_backbone=6.098783e-05 elapsed=23:48:11 avg_step=3.74s recent_step=3.71s eta=28:10:07 +2026-09-15 14:19:12 | INFO | train rank=0 | step=22910/50000 progress=45.8% loss=0.101192 lr=6.095557e-05 lr_backbone=6.095557e-05 elapsed=23:48:50 avg_step=3.74s recent_step=3.95s eta=28:09:32 +2026-09-15 14:19:47 | INFO | train rank=0 | step=22920/50000 progress=45.8% loss=0.075664 lr=6.092330e-05 lr_backbone=6.092330e-05 elapsed=23:49:25 avg_step=3.74s recent_step=3.46s eta=28:08:51 +2026-09-15 14:20:13 | INFO | train rank=0 | step=22930/50000 progress=45.9% loss=0.092388 lr=6.089102e-05 lr_backbone=6.089102e-05 elapsed=23:49:51 avg_step=3.74s recent_step=2.57s eta=28:08:00 +2026-09-15 14:20:41 | INFO | train rank=0 | step=22940/50000 progress=45.9% loss=0.100064 lr=6.085875e-05 lr_backbone=6.085875e-05 elapsed=23:50:19 avg_step=3.74s recent_step=2.85s eta=28:07:12 +2026-09-15 14:21:36 | INFO | train rank=0 | step=22950/50000 progress=45.9% loss=0.099961 lr=6.082646e-05 lr_backbone=6.082646e-05 elapsed=23:51:14 avg_step=3.74s recent_step=5.52s eta=28:06:56 +2026-09-15 14:22:14 | INFO | train rank=0 | step=22960/50000 progress=45.9% loss=0.091092 lr=6.079418e-05 lr_backbone=6.079418e-05 elapsed=23:51:52 avg_step=3.74s recent_step=3.80s eta=28:06:19 +2026-09-15 14:22:49 | INFO | train rank=0 | step=22970/50000 progress=45.9% loss=0.107248 lr=6.076189e-05 lr_backbone=6.076189e-05 elapsed=23:52:27 avg_step=3.74s recent_step=3.47s eta=28:05:38 +2026-09-15 14:23:26 | INFO | train rank=0 | step=22980/50000 progress=46.0% loss=0.094194 lr=6.072959e-05 lr_backbone=6.072959e-05 elapsed=23:53:04 avg_step=3.74s recent_step=3.72s eta=28:05:01 +2026-09-15 14:24:27 | INFO | train rank=0 | step=22990/50000 progress=46.0% loss=0.094747 lr=6.069729e-05 lr_backbone=6.069729e-05 elapsed=23:54:05 avg_step=3.74s recent_step=6.04s eta=28:04:50 +2026-09-15 14:25:00 | INFO | train rank=0 | step=23000/50000 progress=46.0% loss=0.090343 lr=6.066498e-05 lr_backbone=6.066498e-05 elapsed=23:54:38 avg_step=3.74s recent_step=3.34s eta=28:04:08 +2026-09-15 14:25:31 | INFO | train rank=0 | step=23010/50000 progress=46.0% loss=0.094829 lr=6.063267e-05 lr_backbone=6.063267e-05 elapsed=23:55:09 avg_step=3.74s recent_step=3.10s eta=28:03:23 +2026-09-15 14:26:03 | INFO | train rank=0 | step=23020/50000 progress=46.0% loss=0.079444 lr=6.060036e-05 lr_backbone=6.060036e-05 elapsed=23:55:41 avg_step=3.74s recent_step=3.19s eta=28:02:39 +2026-09-15 14:26:50 | INFO | train rank=0 | step=23030/50000 progress=46.1% loss=0.091450 lr=6.056804e-05 lr_backbone=6.056804e-05 elapsed=23:56:28 avg_step=3.74s recent_step=4.70s eta=28:02:13 +2026-09-15 14:27:23 | INFO | train rank=0 | step=23040/50000 progress=46.1% loss=0.094648 lr=6.053571e-05 lr_backbone=6.053571e-05 elapsed=23:57:01 avg_step=3.74s recent_step=3.26s eta=28:01:30 +2026-09-15 14:27:57 | INFO | train rank=0 | step=23050/50000 progress=46.1% loss=0.111954 lr=6.050338e-05 lr_backbone=6.050338e-05 elapsed=23:57:35 avg_step=3.74s recent_step=3.49s eta=28:00:50 +2026-09-15 14:28:26 | INFO | train rank=0 | step=23060/50000 progress=46.1% loss=0.096946 lr=6.047105e-05 lr_backbone=6.047105e-05 elapsed=23:58:04 avg_step=3.74s recent_step=2.85s eta=28:00:02 +2026-09-15 14:29:18 | INFO | train rank=0 | step=23070/50000 progress=46.1% loss=0.095255 lr=6.043871e-05 lr_backbone=6.043871e-05 elapsed=23:58:56 avg_step=3.74s recent_step=5.22s eta=27:59:42 +2026-09-15 14:29:56 | INFO | train rank=0 | step=23080/50000 progress=46.2% loss=0.136550 lr=6.040637e-05 lr_backbone=6.040637e-05 elapsed=23:59:34 avg_step=3.74s recent_step=3.80s eta=27:59:05 +2026-09-15 14:30:23 | INFO | train rank=0 | step=23090/50000 progress=46.2% loss=0.106422 lr=6.037402e-05 lr_backbone=6.037402e-05 elapsed=24:00:01 avg_step=3.74s recent_step=2.70s eta=27:58:15 +2026-09-15 14:31:00 | INFO | train rank=0 | step=23100/50000 progress=46.2% loss=0.095030 lr=6.034167e-05 lr_backbone=6.034167e-05 elapsed=24:00:38 avg_step=3.74s recent_step=3.66s eta=27:57:37 +2026-09-15 14:31:54 | INFO | train rank=0 | step=23110/50000 progress=46.2% loss=0.083168 lr=6.030931e-05 lr_backbone=6.030931e-05 elapsed=24:01:32 avg_step=3.74s recent_step=5.43s eta=27:57:19 +2026-09-15 14:32:27 | INFO | train rank=0 | step=23120/50000 progress=46.2% loss=0.116214 lr=6.027695e-05 lr_backbone=6.027695e-05 elapsed=24:02:05 avg_step=3.74s recent_step=3.29s eta=27:56:36 +2026-09-15 14:33:05 | INFO | train rank=0 | step=23130/50000 progress=46.3% loss=0.111843 lr=6.024458e-05 lr_backbone=6.024458e-05 elapsed=24:02:43 avg_step=3.74s recent_step=3.81s eta=27:56:00 +2026-09-15 14:33:31 | INFO | train rank=0 | step=23140/50000 progress=46.3% loss=0.091828 lr=6.021221e-05 lr_backbone=6.021221e-05 elapsed=24:03:09 avg_step=3.74s recent_step=2.56s eta=27:55:09 +2026-09-15 14:34:17 | INFO | train rank=0 | step=23150/50000 progress=46.3% loss=0.093883 lr=6.017984e-05 lr_backbone=6.017984e-05 elapsed=24:03:55 avg_step=3.74s recent_step=4.69s eta=27:54:42 +2026-09-15 14:34:46 | INFO | train rank=0 | step=23160/50000 progress=46.3% loss=0.103738 lr=6.014746e-05 lr_backbone=6.014746e-05 elapsed=24:04:24 avg_step=3.74s recent_step=2.85s eta=27:53:54 +2026-09-15 14:35:21 | INFO | train rank=0 | step=23170/50000 progress=46.3% loss=0.094966 lr=6.011508e-05 lr_backbone=6.011508e-05 elapsed=24:04:59 avg_step=3.74s recent_step=3.52s eta=27:53:14 +2026-09-15 14:36:00 | INFO | train rank=0 | step=23180/50000 progress=46.4% loss=0.084177 lr=6.008269e-05 lr_backbone=6.008269e-05 elapsed=24:05:38 avg_step=3.74s recent_step=3.87s eta=27:52:39 +2026-09-15 14:36:47 | INFO | train rank=0 | step=23190/50000 progress=46.4% loss=0.091137 lr=6.005030e-05 lr_backbone=6.005030e-05 elapsed=24:06:25 avg_step=3.74s recent_step=4.73s eta=27:52:13 +2026-09-15 14:37:21 | INFO | train rank=0 | step=23200/50000 progress=46.4% loss=0.096473 lr=6.001790e-05 lr_backbone=6.001790e-05 elapsed=24:06:59 avg_step=3.74s recent_step=3.40s eta=27:51:31 +2026-09-15 14:37:53 | INFO | train rank=0 | step=23210/50000 progress=46.4% loss=0.107163 lr=5.998550e-05 lr_backbone=5.998550e-05 elapsed=24:07:31 avg_step=3.74s recent_step=3.14s eta=27:50:47 +2026-09-15 14:38:23 | INFO | train rank=0 | step=23220/50000 progress=46.4% loss=0.086831 lr=5.995309e-05 lr_backbone=5.995309e-05 elapsed=24:08:01 avg_step=3.74s recent_step=3.04s eta=27:50:01 +2026-09-15 14:39:20 | INFO | train rank=0 | step=23230/50000 progress=46.5% loss=0.102867 lr=5.992068e-05 lr_backbone=5.992068e-05 elapsed=24:08:58 avg_step=3.74s recent_step=5.68s eta=27:49:46 +2026-09-15 14:39:51 | INFO | train rank=0 | step=23240/50000 progress=46.5% loss=0.096416 lr=5.988827e-05 lr_backbone=5.988827e-05 elapsed=24:09:29 avg_step=3.74s recent_step=3.13s eta=27:49:02 +2026-09-15 14:40:20 | INFO | train rank=0 | step=23250/50000 progress=46.5% loss=0.111120 lr=5.985585e-05 lr_backbone=5.985585e-05 elapsed=24:09:58 avg_step=3.74s recent_step=2.89s eta=27:48:14 +2026-09-15 14:40:51 | INFO | train rank=0 | step=23260/50000 progress=46.5% loss=0.088545 lr=5.982343e-05 lr_backbone=5.982343e-05 elapsed=24:10:29 avg_step=3.74s recent_step=3.10s eta=27:47:30 +2026-09-15 14:41:50 | INFO | train rank=0 | step=23270/50000 progress=46.5% loss=0.109623 lr=5.979100e-05 lr_backbone=5.979100e-05 elapsed=24:11:28 avg_step=3.74s recent_step=5.93s eta=27:47:17 +2026-09-15 14:42:23 | INFO | train rank=0 | step=23280/50000 progress=46.6% loss=0.091528 lr=5.975857e-05 lr_backbone=5.975857e-05 elapsed=24:12:01 avg_step=3.74s recent_step=3.29s eta=27:46:35 +2026-09-15 14:43:02 | INFO | train rank=0 | step=23290/50000 progress=46.6% loss=0.092019 lr=5.972613e-05 lr_backbone=5.972613e-05 elapsed=24:12:40 avg_step=3.74s recent_step=3.89s eta=27:45:59 +2026-09-15 14:43:37 | INFO | train rank=0 | step=23300/50000 progress=46.6% loss=0.095362 lr=5.969369e-05 lr_backbone=5.969369e-05 elapsed=24:13:15 avg_step=3.74s recent_step=3.48s eta=27:45:19 +2026-09-15 14:44:12 | INFO | train rank=0 | step=23310/50000 progress=46.6% loss=0.103179 lr=5.966125e-05 lr_backbone=5.966125e-05 elapsed=24:13:50 avg_step=3.74s recent_step=3.51s eta=27:44:38 +2026-09-15 14:44:48 | INFO | train rank=0 | step=23320/50000 progress=46.6% loss=0.105953 lr=5.962880e-05 lr_backbone=5.962880e-05 elapsed=24:14:26 avg_step=3.74s recent_step=3.65s eta=27:44:00 +2026-09-15 14:45:17 | INFO | train rank=0 | step=23330/50000 progress=46.7% loss=0.093692 lr=5.959635e-05 lr_backbone=5.959635e-05 elapsed=24:14:55 avg_step=3.74s recent_step=2.88s eta=27:43:13 +2026-09-15 14:45:50 | INFO | train rank=0 | step=23340/50000 progress=46.7% loss=0.091788 lr=5.956389e-05 lr_backbone=5.956389e-05 elapsed=24:15:28 avg_step=3.74s recent_step=3.25s eta=27:42:30 +2026-09-15 14:46:25 | INFO | train rank=0 | step=23350/50000 progress=46.7% loss=0.073909 lr=5.953143e-05 lr_backbone=5.953143e-05 elapsed=24:16:03 avg_step=3.74s recent_step=3.53s eta=27:41:50 +2026-09-15 14:46:53 | INFO | train rank=0 | step=23360/50000 progress=46.7% loss=0.099624 lr=5.949897e-05 lr_backbone=5.949897e-05 elapsed=24:16:31 avg_step=3.74s recent_step=2.77s eta=27:41:01 +2026-09-15 14:47:29 | INFO | train rank=0 | step=23370/50000 progress=46.7% loss=0.092009 lr=5.946650e-05 lr_backbone=5.946650e-05 elapsed=24:17:07 avg_step=3.74s recent_step=3.62s eta=27:40:23 +2026-09-15 14:48:07 | INFO | train rank=0 | step=23380/50000 progress=46.8% loss=0.088517 lr=5.943402e-05 lr_backbone=5.943402e-05 elapsed=24:17:45 avg_step=3.74s recent_step=3.79s eta=27:39:46 +2026-09-15 14:48:55 | INFO | train rank=0 | step=23390/50000 progress=46.8% loss=0.126448 lr=5.940155e-05 lr_backbone=5.940155e-05 elapsed=24:18:33 avg_step=3.74s recent_step=4.86s eta=27:39:21 +2026-09-15 14:49:28 | INFO | train rank=0 | step=23400/50000 progress=46.8% loss=0.078326 lr=5.936907e-05 lr_backbone=5.936907e-05 elapsed=24:19:06 avg_step=3.74s recent_step=3.24s eta=27:38:38 +2026-09-15 14:50:08 | INFO | train rank=0 | step=23410/50000 progress=46.8% loss=0.085904 lr=5.933658e-05 lr_backbone=5.933658e-05 elapsed=24:19:46 avg_step=3.74s recent_step=4.05s eta=27:38:04 +2026-09-15 14:50:46 | INFO | train rank=0 | step=23420/50000 progress=46.8% loss=0.101486 lr=5.930409e-05 lr_backbone=5.930409e-05 elapsed=24:20:24 avg_step=3.74s recent_step=3.76s eta=27:37:27 +2026-09-15 14:51:38 | INFO | train rank=0 | step=23430/50000 progress=46.9% loss=0.101535 lr=5.927160e-05 lr_backbone=5.927160e-05 elapsed=24:21:16 avg_step=3.74s recent_step=5.25s eta=27:37:06 +2026-09-15 14:52:04 | INFO | train rank=0 | step=23440/50000 progress=46.9% loss=0.094154 lr=5.923910e-05 lr_backbone=5.923910e-05 elapsed=24:21:42 avg_step=3.74s recent_step=2.56s eta=27:36:16 +2026-09-15 14:52:41 | INFO | train rank=0 | step=23450/50000 progress=46.9% loss=0.107321 lr=5.920660e-05 lr_backbone=5.920660e-05 elapsed=24:22:19 avg_step=3.74s recent_step=3.72s eta=27:35:38 +2026-09-15 14:53:18 | INFO | train rank=0 | step=23460/50000 progress=46.9% loss=0.103330 lr=5.917409e-05 lr_backbone=5.917409e-05 elapsed=24:22:56 avg_step=3.74s recent_step=3.68s eta=27:35:00 +2026-09-15 14:54:13 | INFO | train rank=0 | step=23470/50000 progress=46.9% loss=0.079984 lr=5.914158e-05 lr_backbone=5.914158e-05 elapsed=24:23:51 avg_step=3.74s recent_step=5.49s eta=27:34:42 +2026-09-15 14:54:40 | INFO | train rank=0 | step=23480/50000 progress=47.0% loss=0.093819 lr=5.910907e-05 lr_backbone=5.910907e-05 elapsed=24:24:18 avg_step=3.74s recent_step=2.70s eta=27:33:53 +2026-09-15 14:55:18 | INFO | train rank=0 | step=23490/50000 progress=47.0% loss=0.077008 lr=5.907655e-05 lr_backbone=5.907655e-05 elapsed=24:24:56 avg_step=3.74s recent_step=3.79s eta=27:33:16 +2026-09-15 14:55:49 | INFO | train rank=0 | step=23500/50000 progress=47.0% loss=0.091567 lr=5.904403e-05 lr_backbone=5.904403e-05 elapsed=24:25:27 avg_step=3.74s recent_step=3.16s eta=27:32:32 +2026-09-15 14:56:37 | INFO | train rank=0 | step=23510/50000 progress=47.0% loss=0.096617 lr=5.901150e-05 lr_backbone=5.901150e-05 elapsed=24:26:15 avg_step=3.74s recent_step=4.73s eta=27:32:06 +2026-09-15 14:57:11 | INFO | train rank=0 | step=23520/50000 progress=47.0% loss=0.092488 lr=5.897897e-05 lr_backbone=5.897897e-05 elapsed=24:26:49 avg_step=3.74s recent_step=3.38s eta=27:31:25 +2026-09-15 14:57:45 | INFO | train rank=0 | step=23530/50000 progress=47.1% loss=0.113362 lr=5.894644e-05 lr_backbone=5.894644e-05 elapsed=24:27:23 avg_step=3.74s recent_step=3.43s eta=27:30:44 +2026-09-15 14:58:23 | INFO | train rank=0 | step=23540/50000 progress=47.1% loss=0.083673 lr=5.891390e-05 lr_backbone=5.891390e-05 elapsed=24:28:01 avg_step=3.74s recent_step=3.81s eta=27:30:07 +2026-09-15 14:59:01 | INFO | train rank=0 | step=23550/50000 progress=47.1% loss=0.097366 lr=5.888136e-05 lr_backbone=5.888136e-05 elapsed=24:28:39 avg_step=3.74s recent_step=3.76s eta=27:29:30 +2026-09-15 14:59:37 | INFO | train rank=0 | step=23560/50000 progress=47.1% loss=0.096594 lr=5.884881e-05 lr_backbone=5.884881e-05 elapsed=24:29:15 avg_step=3.74s recent_step=3.62s eta=27:28:51 +2026-09-15 15:00:35 | INFO | train rank=0 | step=23570/50000 progress=47.1% loss=0.094654 lr=5.881626e-05 lr_backbone=5.881626e-05 elapsed=24:30:13 avg_step=3.74s recent_step=5.85s eta=27:28:37 +2026-09-15 15:01:07 | INFO | train rank=0 | step=23580/50000 progress=47.2% loss=0.098880 lr=5.878371e-05 lr_backbone=5.878371e-05 elapsed=24:30:45 avg_step=3.74s recent_step=3.15s eta=27:27:53 +2026-09-15 15:01:42 | INFO | train rank=0 | step=23590/50000 progress=47.2% loss=0.113384 lr=5.875115e-05 lr_backbone=5.875115e-05 elapsed=24:31:20 avg_step=3.74s recent_step=3.51s eta=27:27:13 +2026-09-15 15:02:14 | INFO | train rank=0 | step=23600/50000 progress=47.2% loss=0.093320 lr=5.871859e-05 lr_backbone=5.871859e-05 elapsed=24:31:52 avg_step=3.74s recent_step=3.18s eta=27:26:29 +2026-09-15 15:03:04 | INFO | train rank=0 | step=23610/50000 progress=47.2% loss=0.106379 lr=5.868603e-05 lr_backbone=5.868603e-05 elapsed=24:32:42 avg_step=3.74s recent_step=5.01s eta=27:26:06 +2026-09-15 15:03:31 | INFO | train rank=0 | step=23620/50000 progress=47.2% loss=0.083486 lr=5.865346e-05 lr_backbone=5.865346e-05 elapsed=24:33:09 avg_step=3.74s recent_step=2.67s eta=27:25:17 +2026-09-15 15:04:03 | INFO | train rank=0 | step=23630/50000 progress=47.3% loss=0.109719 lr=5.862089e-05 lr_backbone=5.862089e-05 elapsed=24:33:41 avg_step=3.74s recent_step=3.20s eta=27:24:33 +2026-09-15 15:04:37 | INFO | train rank=0 | step=23640/50000 progress=47.3% loss=0.105046 lr=5.858831e-05 lr_backbone=5.858831e-05 elapsed=24:34:15 avg_step=3.74s recent_step=3.40s eta=27:23:52 +2026-09-15 15:05:23 | INFO | train rank=0 | step=23650/50000 progress=47.3% loss=0.085376 lr=5.855573e-05 lr_backbone=5.855573e-05 elapsed=24:35:01 avg_step=3.74s recent_step=4.66s eta=27:23:25 +2026-09-15 15:05:51 | INFO | train rank=0 | step=23660/50000 progress=47.3% loss=0.081450 lr=5.852315e-05 lr_backbone=5.852315e-05 elapsed=24:35:29 avg_step=3.74s recent_step=2.76s eta=27:22:37 +2026-09-15 15:06:18 | INFO | train rank=0 | step=23670/50000 progress=47.3% loss=0.078993 lr=5.849056e-05 lr_backbone=5.849056e-05 elapsed=24:35:56 avg_step=3.74s recent_step=2.72s eta=27:21:48 +2026-09-15 15:06:52 | INFO | train rank=0 | step=23680/50000 progress=47.4% loss=0.103383 lr=5.845797e-05 lr_backbone=5.845797e-05 elapsed=24:36:30 avg_step=3.74s recent_step=3.40s eta=27:21:07 +2026-09-15 15:07:49 | INFO | train rank=0 | step=23690/50000 progress=47.4% loss=0.087570 lr=5.842537e-05 lr_backbone=5.842537e-05 elapsed=24:37:27 avg_step=3.74s recent_step=5.73s eta=27:20:51 +2026-09-15 15:08:25 | INFO | train rank=0 | step=23700/50000 progress=47.4% loss=0.094324 lr=5.839278e-05 lr_backbone=5.839278e-05 elapsed=24:38:03 avg_step=3.74s recent_step=3.59s eta=27:20:12 +2026-09-15 15:08:59 | INFO | train rank=0 | step=23710/50000 progress=47.4% loss=0.095791 lr=5.836017e-05 lr_backbone=5.836017e-05 elapsed=24:38:37 avg_step=3.74s recent_step=3.35s eta=27:19:30 +2026-09-15 15:09:35 | INFO | train rank=0 | step=23720/50000 progress=47.4% loss=0.087873 lr=5.832757e-05 lr_backbone=5.832757e-05 elapsed=24:39:13 avg_step=3.74s recent_step=3.68s eta=27:18:52 +2026-09-15 15:10:25 | INFO | train rank=0 | step=23730/50000 progress=47.5% loss=0.098952 lr=5.829496e-05 lr_backbone=5.829496e-05 elapsed=24:40:03 avg_step=3.74s recent_step=4.94s eta=27:18:28 +2026-09-15 15:10:57 | INFO | train rank=0 | step=23740/50000 progress=47.5% loss=0.079260 lr=5.826235e-05 lr_backbone=5.826235e-05 elapsed=24:40:35 avg_step=3.74s recent_step=3.20s eta=27:17:45 +2026-09-15 15:11:26 | INFO | train rank=0 | step=23750/50000 progress=47.5% loss=0.076018 lr=5.822973e-05 lr_backbone=5.822973e-05 elapsed=24:41:04 avg_step=3.74s recent_step=2.95s eta=27:16:59 +2026-09-15 15:11:54 | INFO | train rank=0 | step=23760/50000 progress=47.5% loss=0.089426 lr=5.819711e-05 lr_backbone=5.819711e-05 elapsed=24:41:32 avg_step=3.74s recent_step=2.76s eta=27:16:10 +2026-09-15 15:12:51 | INFO | train rank=0 | step=23770/50000 progress=47.5% loss=0.081984 lr=5.816449e-05 lr_backbone=5.816449e-05 elapsed=24:42:29 avg_step=3.74s recent_step=5.71s eta=27:15:55 +2026-09-15 15:13:21 | INFO | train rank=0 | step=23780/50000 progress=47.6% loss=0.105180 lr=5.813186e-05 lr_backbone=5.813186e-05 elapsed=24:42:59 avg_step=3.74s recent_step=2.98s eta=27:15:09 +2026-09-15 15:13:58 | INFO | train rank=0 | step=23790/50000 progress=47.6% loss=0.088598 lr=5.809923e-05 lr_backbone=5.809923e-05 elapsed=24:43:36 avg_step=3.74s recent_step=3.72s eta=27:14:31 +2026-09-15 15:14:27 | INFO | train rank=0 | step=23800/50000 progress=47.6% loss=0.095676 lr=5.806659e-05 lr_backbone=5.806659e-05 elapsed=24:44:05 avg_step=3.74s recent_step=2.86s eta=27:13:44 +2026-09-15 15:15:21 | INFO | train rank=0 | step=23810/50000 progress=47.6% loss=0.089311 lr=5.803395e-05 lr_backbone=5.803395e-05 elapsed=24:44:59 avg_step=3.74s recent_step=5.46s eta=27:13:25 +2026-09-15 15:15:44 | INFO | train rank=0 | step=23820/50000 progress=47.6% loss=0.080958 lr=5.800131e-05 lr_backbone=5.800131e-05 elapsed=24:45:22 avg_step=3.74s recent_step=2.26s eta=27:12:32 +2026-09-15 15:16:15 | INFO | train rank=0 | step=23830/50000 progress=47.7% loss=0.100859 lr=5.796867e-05 lr_backbone=5.796867e-05 elapsed=24:45:53 avg_step=3.74s recent_step=3.10s eta=27:11:47 +2026-09-15 15:16:50 | INFO | train rank=0 | step=23840/50000 progress=47.7% loss=0.102071 lr=5.793602e-05 lr_backbone=5.793602e-05 elapsed=24:46:28 avg_step=3.74s recent_step=3.53s eta=27:11:07 +2026-09-15 15:17:35 | INFO | train rank=0 | step=23850/50000 progress=47.7% loss=0.104742 lr=5.790337e-05 lr_backbone=5.790337e-05 elapsed=24:47:13 avg_step=3.74s recent_step=4.52s eta=27:10:39 +2026-09-15 15:18:08 | INFO | train rank=0 | step=23860/50000 progress=47.7% loss=0.112678 lr=5.787071e-05 lr_backbone=5.787071e-05 elapsed=24:47:46 avg_step=3.74s recent_step=3.29s eta=27:09:56 +2026-09-15 15:18:37 | INFO | train rank=0 | step=23870/50000 progress=47.7% loss=0.089114 lr=5.783805e-05 lr_backbone=5.783805e-05 elapsed=24:48:15 avg_step=3.74s recent_step=2.91s eta=27:09:10 +2026-09-15 15:19:15 | INFO | train rank=0 | step=23880/50000 progress=47.8% loss=0.093627 lr=5.780539e-05 lr_backbone=5.780539e-05 elapsed=24:48:53 avg_step=3.74s recent_step=3.79s eta=27:08:33 +2026-09-15 15:20:03 | INFO | train rank=0 | step=23890/50000 progress=47.8% loss=0.106387 lr=5.777273e-05 lr_backbone=5.777273e-05 elapsed=24:49:41 avg_step=3.74s recent_step=4.82s eta=27:08:07 +2026-09-15 15:20:36 | INFO | train rank=0 | step=23900/50000 progress=47.8% loss=0.101810 lr=5.774006e-05 lr_backbone=5.774006e-05 elapsed=24:50:14 avg_step=3.74s recent_step=3.23s eta=27:07:24 +2026-09-15 15:21:10 | INFO | train rank=0 | step=23910/50000 progress=47.8% loss=0.076890 lr=5.770738e-05 lr_backbone=5.770738e-05 elapsed=24:50:48 avg_step=3.74s recent_step=3.42s eta=27:06:43 +2026-09-15 15:21:40 | INFO | train rank=0 | step=23920/50000 progress=47.8% loss=0.094570 lr=5.767471e-05 lr_backbone=5.767471e-05 elapsed=24:51:18 avg_step=3.74s recent_step=3.00s eta=27:05:58 +2026-09-15 15:22:25 | INFO | train rank=0 | step=23930/50000 progress=47.9% loss=0.078057 lr=5.764203e-05 lr_backbone=5.764203e-05 elapsed=24:52:03 avg_step=3.74s recent_step=4.51s eta=27:05:29 +2026-09-15 15:23:02 | INFO | train rank=0 | step=23940/50000 progress=47.9% loss=0.135915 lr=5.760935e-05 lr_backbone=5.760935e-05 elapsed=24:52:40 avg_step=3.74s recent_step=3.69s eta=27:04:51 +2026-09-15 15:23:39 | INFO | train rank=0 | step=23950/50000 progress=47.9% loss=0.076009 lr=5.757666e-05 lr_backbone=5.757666e-05 elapsed=24:53:17 avg_step=3.74s recent_step=3.70s eta=27:04:13 +2026-09-15 15:24:11 | INFO | train rank=0 | step=23960/50000 progress=47.9% loss=0.091661 lr=5.754397e-05 lr_backbone=5.754397e-05 elapsed=24:53:49 avg_step=3.74s recent_step=3.21s eta=27:03:30 +2026-09-15 15:25:01 | INFO | train rank=0 | step=23970/50000 progress=47.9% loss=0.088739 lr=5.751128e-05 lr_backbone=5.751128e-05 elapsed=24:54:39 avg_step=3.74s recent_step=4.98s eta=27:03:06 +2026-09-15 15:25:32 | INFO | train rank=0 | step=23980/50000 progress=48.0% loss=0.100479 lr=5.747858e-05 lr_backbone=5.747858e-05 elapsed=24:55:10 avg_step=3.74s recent_step=3.10s eta=27:02:22 +2026-09-15 15:26:09 | INFO | train rank=0 | step=23990/50000 progress=48.0% loss=0.088534 lr=5.744588e-05 lr_backbone=5.744588e-05 elapsed=24:55:47 avg_step=3.74s recent_step=3.73s eta=27:01:44 +2026-09-15 15:26:48 | INFO | train rank=0 | step=24000/50000 progress=48.0% loss=0.072049 lr=5.741318e-05 lr_backbone=5.741318e-05 elapsed=24:56:26 avg_step=3.74s recent_step=3.85s eta=27:01:08 +2026-09-15 15:27:27 | INFO | train rank=0 | step=24010/50000 progress=48.0% loss=0.091176 lr=5.738048e-05 lr_backbone=5.738048e-05 elapsed=24:57:05 avg_step=3.74s recent_step=3.94s eta=27:00:33 +2026-09-15 15:28:02 | INFO | train rank=0 | step=24020/50000 progress=48.0% loss=0.113812 lr=5.734777e-05 lr_backbone=5.734777e-05 elapsed=24:57:40 avg_step=3.74s recent_step=3.54s eta=26:59:53 +2026-09-15 15:28:35 | INFO | train rank=0 | step=24030/50000 progress=48.1% loss=0.086602 lr=5.731506e-05 lr_backbone=5.731506e-05 elapsed=24:58:13 avg_step=3.74s recent_step=3.21s eta=26:59:10 +2026-09-15 15:29:07 | INFO | train rank=0 | step=24040/50000 progress=48.1% loss=0.097038 lr=5.728234e-05 lr_backbone=5.728234e-05 elapsed=24:58:45 avg_step=3.74s recent_step=3.20s eta=26:58:27 +2026-09-15 15:30:06 | INFO | train rank=0 | step=24050/50000 progress=48.1% loss=0.087514 lr=5.724962e-05 lr_backbone=5.724962e-05 elapsed=24:59:44 avg_step=3.74s recent_step=5.92s eta=26:58:13 +2026-09-15 15:30:43 | INFO | train rank=0 | step=24060/50000 progress=48.1% loss=0.117258 lr=5.721690e-05 lr_backbone=5.721690e-05 elapsed=25:00:21 avg_step=3.74s recent_step=3.73s eta=26:57:35 +2026-09-15 15:31:16 | INFO | train rank=0 | step=24070/50000 progress=48.1% loss=0.095120 lr=5.718418e-05 lr_backbone=5.718418e-05 elapsed=25:00:54 avg_step=3.74s recent_step=3.31s eta=26:56:53 +2026-09-15 15:31:46 | INFO | train rank=0 | step=24080/50000 progress=48.2% loss=0.089793 lr=5.715145e-05 lr_backbone=5.715145e-05 elapsed=25:01:24 avg_step=3.74s recent_step=2.98s eta=26:56:08 +2026-09-15 15:32:34 | INFO | train rank=0 | step=24090/50000 progress=48.2% loss=0.091798 lr=5.711872e-05 lr_backbone=5.711872e-05 elapsed=25:02:12 avg_step=3.74s recent_step=4.82s eta=26:55:42 +2026-09-15 15:33:05 | INFO | train rank=0 | step=24100/50000 progress=48.2% loss=0.095768 lr=5.708598e-05 lr_backbone=5.708598e-05 elapsed=25:02:43 avg_step=3.74s recent_step=3.06s eta=26:54:57 +2026-09-15 15:33:47 | INFO | train rank=0 | step=24110/50000 progress=48.2% loss=0.111593 lr=5.705325e-05 lr_backbone=5.705325e-05 elapsed=25:03:25 avg_step=3.74s recent_step=4.27s eta=26:54:25 +2026-09-15 15:34:11 | INFO | train rank=0 | step=24120/50000 progress=48.2% loss=0.092389 lr=5.702051e-05 lr_backbone=5.702051e-05 elapsed=25:03:49 avg_step=3.74s recent_step=2.35s eta=26:53:33 +2026-09-15 15:35:04 | INFO | train rank=0 | step=24130/50000 progress=48.3% loss=0.108307 lr=5.698776e-05 lr_backbone=5.698776e-05 elapsed=25:04:42 avg_step=3.74s recent_step=5.30s eta=26:53:12 +2026-09-15 15:35:32 | INFO | train rank=0 | step=24140/50000 progress=48.3% loss=0.104626 lr=5.695502e-05 lr_backbone=5.695502e-05 elapsed=25:05:10 avg_step=3.74s recent_step=2.85s eta=26:52:25 +2026-09-15 15:36:01 | INFO | train rank=0 | step=24150/50000 progress=48.3% loss=0.100437 lr=5.692227e-05 lr_backbone=5.692227e-05 elapsed=25:05:39 avg_step=3.74s recent_step=2.88s eta=26:51:39 +2026-09-15 15:36:36 | INFO | train rank=0 | step=24160/50000 progress=48.3% loss=0.079275 lr=5.688951e-05 lr_backbone=5.688951e-05 elapsed=25:06:14 avg_step=3.74s recent_step=3.47s eta=26:50:58 +2026-09-15 15:37:17 | INFO | train rank=0 | step=24170/50000 progress=48.3% loss=0.107469 lr=5.685676e-05 lr_backbone=5.685676e-05 elapsed=25:06:55 avg_step=3.74s recent_step=4.12s eta=26:50:25 +2026-09-15 15:37:45 | INFO | train rank=0 | step=24180/50000 progress=48.4% loss=0.096756 lr=5.682400e-05 lr_backbone=5.682400e-05 elapsed=25:07:23 avg_step=3.74s recent_step=2.76s eta=26:49:37 +2026-09-15 15:38:26 | INFO | train rank=0 | step=24190/50000 progress=48.4% loss=0.100403 lr=5.679124e-05 lr_backbone=5.679124e-05 elapsed=25:08:04 avg_step=3.74s recent_step=4.14s eta=26:49:04 +2026-09-15 15:39:00 | INFO | train rank=0 | step=24200/50000 progress=48.4% loss=0.086079 lr=5.675847e-05 lr_backbone=5.675847e-05 elapsed=25:08:38 avg_step=3.74s recent_step=3.39s eta=26:48:23 +2026-09-15 15:39:45 | INFO | train rank=0 | step=24210/50000 progress=48.4% loss=0.102452 lr=5.672571e-05 lr_backbone=5.672571e-05 elapsed=25:09:23 avg_step=3.74s recent_step=4.52s eta=26:47:54 +2026-09-15 15:40:11 | INFO | train rank=0 | step=24220/50000 progress=48.4% loss=0.100769 lr=5.669294e-05 lr_backbone=5.669294e-05 elapsed=25:09:49 avg_step=3.74s recent_step=2.54s eta=26:47:03 +2026-09-15 15:40:59 | INFO | train rank=0 | step=24230/50000 progress=48.5% loss=0.095313 lr=5.666016e-05 lr_backbone=5.666016e-05 elapsed=25:10:37 avg_step=3.74s recent_step=4.88s eta=26:46:38 +2026-09-15 15:41:28 | INFO | train rank=0 | step=24240/50000 progress=48.5% loss=0.082565 lr=5.662739e-05 lr_backbone=5.662739e-05 elapsed=25:11:06 avg_step=3.74s recent_step=2.90s eta=26:45:52 +2026-09-15 15:42:10 | INFO | train rank=0 | step=24250/50000 progress=48.5% loss=0.104130 lr=5.659461e-05 lr_backbone=5.659461e-05 elapsed=25:11:48 avg_step=3.74s recent_step=4.14s eta=26:45:19 +2026-09-15 15:42:48 | INFO | train rank=0 | step=24260/50000 progress=48.5% loss=0.123856 lr=5.656183e-05 lr_backbone=5.656183e-05 elapsed=25:12:26 avg_step=3.74s recent_step=3.82s eta=26:44:42 +2026-09-15 15:43:41 | INFO | train rank=0 | step=24270/50000 progress=48.5% loss=0.077169 lr=5.652904e-05 lr_backbone=5.652904e-05 elapsed=25:13:19 avg_step=3.74s recent_step=5.30s eta=26:44:21 +2026-09-15 15:44:13 | INFO | train rank=0 | step=24280/50000 progress=48.6% loss=0.089491 lr=5.649625e-05 lr_backbone=5.649625e-05 elapsed=25:13:51 avg_step=3.74s recent_step=3.23s eta=26:43:38 +2026-09-15 15:44:43 | INFO | train rank=0 | step=24290/50000 progress=48.6% loss=0.083788 lr=5.646346e-05 lr_backbone=5.646346e-05 elapsed=25:14:21 avg_step=3.74s recent_step=2.94s eta=26:42:52 +2026-09-15 15:45:11 | INFO | train rank=0 | step=24300/50000 progress=48.6% loss=0.096141 lr=5.643067e-05 lr_backbone=5.643067e-05 elapsed=25:14:49 avg_step=3.74s recent_step=2.83s eta=26:42:05 +2026-09-15 15:46:02 | INFO | train rank=0 | step=24310/50000 progress=48.6% loss=0.089199 lr=5.639787e-05 lr_backbone=5.639787e-05 elapsed=25:15:40 avg_step=3.74s recent_step=5.12s eta=26:41:43 +2026-09-15 15:46:41 | INFO | train rank=0 | step=24320/50000 progress=48.6% loss=0.103313 lr=5.636508e-05 lr_backbone=5.636508e-05 elapsed=25:16:19 avg_step=3.74s recent_step=3.93s eta=26:41:07 +2026-09-15 15:47:08 | INFO | train rank=0 | step=24330/50000 progress=48.7% loss=0.102096 lr=5.633227e-05 lr_backbone=5.633227e-05 elapsed=25:16:46 avg_step=3.74s recent_step=2.65s eta=26:40:18 +2026-09-15 15:47:42 | INFO | train rank=0 | step=24340/50000 progress=48.7% loss=0.119143 lr=5.629947e-05 lr_backbone=5.629947e-05 elapsed=25:17:20 avg_step=3.74s recent_step=3.39s eta=26:39:37 +2026-09-15 15:48:28 | INFO | train rank=0 | step=24350/50000 progress=48.7% loss=0.111792 lr=5.626666e-05 lr_backbone=5.626666e-05 elapsed=25:18:06 avg_step=3.74s recent_step=4.59s eta=26:39:09 +2026-09-15 15:48:59 | INFO | train rank=0 | step=24360/50000 progress=48.7% loss=0.079646 lr=5.623385e-05 lr_backbone=5.623385e-05 elapsed=25:18:37 avg_step=3.74s recent_step=3.10s eta=26:38:25 +2026-09-15 15:49:31 | INFO | train rank=0 | step=24370/50000 progress=48.7% loss=0.109886 lr=5.620104e-05 lr_backbone=5.620104e-05 elapsed=25:19:09 avg_step=3.74s recent_step=3.19s eta=26:37:42 +2026-09-15 15:50:10 | INFO | train rank=0 | step=24380/50000 progress=48.8% loss=0.091166 lr=5.616822e-05 lr_backbone=5.616822e-05 elapsed=25:19:48 avg_step=3.74s recent_step=3.91s eta=26:37:06 +2026-09-15 15:50:50 | INFO | train rank=0 | step=24390/50000 progress=48.8% loss=0.082705 lr=5.613541e-05 lr_backbone=5.613541e-05 elapsed=25:20:28 avg_step=3.74s recent_step=3.99s eta=26:36:31 +2026-09-15 15:51:24 | INFO | train rank=0 | step=24400/50000 progress=48.8% loss=0.096626 lr=5.610258e-05 lr_backbone=5.610258e-05 elapsed=25:21:02 avg_step=3.74s recent_step=3.40s eta=26:35:50 +2026-09-15 15:52:00 | INFO | train rank=0 | step=24410/50000 progress=48.8% loss=0.103061 lr=5.606976e-05 lr_backbone=5.606976e-05 elapsed=25:21:38 avg_step=3.74s recent_step=3.66s eta=26:35:12 +2026-09-15 15:52:52 | INFO | train rank=0 | step=24420/50000 progress=48.8% loss=0.098592 lr=5.603693e-05 lr_backbone=5.603693e-05 elapsed=25:22:30 avg_step=3.74s recent_step=5.16s eta=26:34:49 +2026-09-15 15:53:23 | INFO | train rank=0 | step=24430/50000 progress=48.9% loss=0.089647 lr=5.600411e-05 lr_backbone=5.600411e-05 elapsed=25:23:01 avg_step=3.74s recent_step=3.08s eta=26:34:05 +2026-09-15 15:53:53 | INFO | train rank=0 | step=24440/50000 progress=48.9% loss=0.097877 lr=5.597127e-05 lr_backbone=5.597127e-05 elapsed=25:23:31 avg_step=3.74s recent_step=3.06s eta=26:33:20 +2026-09-15 15:54:29 | INFO | train rank=0 | step=24450/50000 progress=48.9% loss=0.107890 lr=5.593844e-05 lr_backbone=5.593844e-05 elapsed=25:24:07 avg_step=3.74s recent_step=3.54s eta=26:32:41 +2026-09-15 15:55:12 | INFO | train rank=0 | step=24460/50000 progress=48.9% loss=0.116636 lr=5.590560e-05 lr_backbone=5.590560e-05 elapsed=25:24:50 avg_step=3.74s recent_step=4.29s eta=26:32:09 +2026-09-15 15:55:48 | INFO | train rank=0 | step=24470/50000 progress=48.9% loss=0.114229 lr=5.587276e-05 lr_backbone=5.587276e-05 elapsed=25:25:26 avg_step=3.74s recent_step=3.64s eta=26:31:31 +2026-09-15 15:56:22 | INFO | train rank=0 | step=24480/50000 progress=49.0% loss=0.096740 lr=5.583992e-05 lr_backbone=5.583992e-05 elapsed=25:26:00 avg_step=3.74s recent_step=3.40s eta=26:30:50 +2026-09-15 15:56:53 | INFO | train rank=0 | step=24490/50000 progress=49.0% loss=0.091231 lr=5.580708e-05 lr_backbone=5.580708e-05 elapsed=25:26:31 avg_step=3.74s recent_step=3.14s eta=26:30:06 +2026-09-15 15:57:51 | INFO | train rank=0 | step=24500/50000 progress=49.0% loss=0.098016 lr=5.577423e-05 lr_backbone=5.577423e-05 elapsed=25:27:29 avg_step=3.74s recent_step=5.71s eta=26:29:49 +2026-09-15 15:58:25 | INFO | train rank=0 | step=24510/50000 progress=49.0% loss=0.102744 lr=5.574138e-05 lr_backbone=5.574138e-05 elapsed=25:28:03 avg_step=3.74s recent_step=3.45s eta=26:29:09 +2026-09-15 15:59:01 | INFO | train rank=0 | step=24520/50000 progress=49.0% loss=0.090715 lr=5.570853e-05 lr_backbone=5.570853e-05 elapsed=25:28:39 avg_step=3.74s recent_step=3.65s eta=26:28:30 +2026-09-15 15:59:35 | INFO | train rank=0 | step=24530/50000 progress=49.1% loss=0.085473 lr=5.567568e-05 lr_backbone=5.567568e-05 elapsed=25:29:13 avg_step=3.74s recent_step=3.31s eta=26:27:49 +2026-09-15 16:00:19 | INFO | train rank=0 | step=24540/50000 progress=49.1% loss=0.105310 lr=5.564282e-05 lr_backbone=5.564282e-05 elapsed=25:29:57 avg_step=3.74s recent_step=4.47s eta=26:27:19 +2026-09-15 16:00:48 | INFO | train rank=0 | step=24550/50000 progress=49.1% loss=0.086926 lr=5.560996e-05 lr_backbone=5.560996e-05 elapsed=25:30:27 avg_step=3.74s recent_step=2.92s eta=26:26:33 +2026-09-15 16:01:25 | INFO | train rank=0 | step=24560/50000 progress=49.1% loss=0.102752 lr=5.557710e-05 lr_backbone=5.557710e-05 elapsed=25:31:03 avg_step=3.74s recent_step=3.69s eta=26:25:55 +2026-09-15 16:02:03 | INFO | train rank=0 | step=24570/50000 progress=49.1% loss=0.095510 lr=5.554423e-05 lr_backbone=5.554423e-05 elapsed=25:31:41 avg_step=3.74s recent_step=3.73s eta=26:25:17 +2026-09-15 16:02:51 | INFO | train rank=0 | step=24580/50000 progress=49.2% loss=0.079054 lr=5.551137e-05 lr_backbone=5.551137e-05 elapsed=25:32:29 avg_step=3.74s recent_step=4.88s eta=26:24:52 +2026-09-15 16:03:25 | INFO | train rank=0 | step=24590/50000 progress=49.2% loss=0.085420 lr=5.547850e-05 lr_backbone=5.547850e-05 elapsed=25:33:03 avg_step=3.74s recent_step=3.39s eta=26:24:11 +2026-09-15 16:03:56 | INFO | train rank=0 | step=24600/50000 progress=49.2% loss=0.080955 lr=5.544563e-05 lr_backbone=5.544563e-05 elapsed=25:33:34 avg_step=3.74s recent_step=3.08s eta=26:23:27 +2026-09-15 16:04:41 | INFO | train rank=0 | step=24610/50000 progress=49.2% loss=0.103209 lr=5.541275e-05 lr_backbone=5.541275e-05 elapsed=25:34:19 avg_step=3.74s recent_step=4.46s eta=26:22:57 +2026-09-15 16:05:17 | INFO | train rank=0 | step=24620/50000 progress=49.2% loss=0.087449 lr=5.537988e-05 lr_backbone=5.537988e-05 elapsed=25:34:55 avg_step=3.74s recent_step=3.63s eta=26:22:18 +2026-09-15 16:05:48 | INFO | train rank=0 | step=24630/50000 progress=49.3% loss=0.116151 lr=5.534700e-05 lr_backbone=5.534700e-05 elapsed=25:35:26 avg_step=3.74s recent_step=3.14s eta=26:21:34 +2026-09-15 16:06:21 | INFO | train rank=0 | step=24640/50000 progress=49.3% loss=0.087740 lr=5.531412e-05 lr_backbone=5.531412e-05 elapsed=25:35:59 avg_step=3.74s recent_step=3.21s eta=26:20:52 +2026-09-15 16:07:14 | INFO | train rank=0 | step=24650/50000 progress=49.3% loss=0.076062 lr=5.528123e-05 lr_backbone=5.528123e-05 elapsed=25:36:52 avg_step=3.74s recent_step=5.37s eta=26:20:31 +2026-09-15 16:07:50 | INFO | train rank=0 | step=24660/50000 progress=49.3% loss=0.099114 lr=5.524835e-05 lr_backbone=5.524835e-05 elapsed=25:37:28 avg_step=3.74s recent_step=3.55s eta=26:19:52 +2026-09-15 16:08:24 | INFO | train rank=0 | step=24670/50000 progress=49.3% loss=0.098299 lr=5.521546e-05 lr_backbone=5.521546e-05 elapsed=25:38:02 avg_step=3.74s recent_step=3.44s eta=26:19:11 +2026-09-15 16:09:02 | INFO | train rank=0 | step=24680/50000 progress=49.4% loss=0.086371 lr=5.518257e-05 lr_backbone=5.518257e-05 elapsed=25:38:40 avg_step=3.74s recent_step=3.77s eta=26:18:34 +2026-09-15 16:09:49 | INFO | train rank=0 | step=24690/50000 progress=49.4% loss=0.110810 lr=5.514968e-05 lr_backbone=5.514968e-05 elapsed=25:39:27 avg_step=3.74s recent_step=4.67s eta=26:18:06 +2026-09-15 16:10:27 | INFO | train rank=0 | step=24700/50000 progress=49.4% loss=0.097437 lr=5.511678e-05 lr_backbone=5.511678e-05 elapsed=25:40:05 avg_step=3.74s recent_step=3.82s eta=26:17:29 +2026-09-15 16:10:59 | INFO | train rank=0 | step=24710/50000 progress=49.4% loss=0.071616 lr=5.508389e-05 lr_backbone=5.508389e-05 elapsed=25:40:37 avg_step=3.74s recent_step=3.18s eta=26:16:46 +2026-09-15 16:11:31 | INFO | train rank=0 | step=24720/50000 progress=49.4% loss=0.111548 lr=5.505099e-05 lr_backbone=5.505099e-05 elapsed=25:41:09 avg_step=3.74s recent_step=3.23s eta=26:16:04 +2026-09-15 16:12:14 | INFO | train rank=0 | step=24730/50000 progress=49.5% loss=0.090250 lr=5.501809e-05 lr_backbone=5.501809e-05 elapsed=25:41:52 avg_step=3.74s recent_step=4.31s eta=26:15:32 +2026-09-15 16:12:46 | INFO | train rank=0 | step=24740/50000 progress=49.5% loss=0.090429 lr=5.498518e-05 lr_backbone=5.498518e-05 elapsed=25:42:24 avg_step=3.74s recent_step=3.17s eta=26:14:49 +2026-09-15 16:13:21 | INFO | train rank=0 | step=24750/50000 progress=49.5% loss=0.103743 lr=5.495228e-05 lr_backbone=5.495228e-05 elapsed=25:42:59 avg_step=3.74s recent_step=3.51s eta=26:14:09 +2026-09-15 16:13:59 | INFO | train rank=0 | step=24760/50000 progress=49.5% loss=0.106782 lr=5.491937e-05 lr_backbone=5.491937e-05 elapsed=25:43:37 avg_step=3.74s recent_step=3.78s eta=26:13:32 +2026-09-15 16:14:33 | INFO | train rank=0 | step=24770/50000 progress=49.5% loss=0.101352 lr=5.488646e-05 lr_backbone=5.488646e-05 elapsed=25:44:11 avg_step=3.74s recent_step=3.41s eta=26:12:51 +2026-09-15 16:15:32 | INFO | train rank=0 | step=24780/50000 progress=49.6% loss=0.089416 lr=5.485355e-05 lr_backbone=5.485355e-05 elapsed=25:45:10 avg_step=3.74s recent_step=5.91s eta=26:12:36 +2026-09-15 16:15:55 | INFO | train rank=0 | step=24790/50000 progress=49.6% loss=0.081305 lr=5.482063e-05 lr_backbone=5.482063e-05 elapsed=25:45:33 avg_step=3.74s recent_step=2.37s eta=26:11:44 +2026-09-15 16:16:21 | INFO | train rank=0 | step=24800/50000 progress=49.6% loss=0.100123 lr=5.478772e-05 lr_backbone=5.478772e-05 elapsed=25:45:59 avg_step=3.74s recent_step=2.61s eta=26:10:56 +2026-09-15 16:16:57 | INFO | train rank=0 | step=24810/50000 progress=49.6% loss=0.098179 lr=5.475480e-05 lr_backbone=5.475480e-05 elapsed=25:46:35 avg_step=3.74s recent_step=3.56s eta=26:10:16 +2026-09-15 16:17:45 | INFO | train rank=0 | step=24820/50000 progress=49.6% loss=0.099954 lr=5.472188e-05 lr_backbone=5.472188e-05 elapsed=25:47:23 avg_step=3.74s recent_step=4.74s eta=26:09:49 +2026-09-15 16:18:20 | INFO | train rank=0 | step=24830/50000 progress=49.7% loss=0.096236 lr=5.468895e-05 lr_backbone=5.468895e-05 elapsed=25:47:58 avg_step=3.74s recent_step=3.59s eta=26:09:10 +2026-09-15 16:18:49 | INFO | train rank=0 | step=24840/50000 progress=49.7% loss=0.073518 lr=5.465603e-05 lr_backbone=5.465603e-05 elapsed=25:48:27 avg_step=3.74s recent_step=2.87s eta=26:08:24 +2026-09-15 16:19:32 | INFO | train rank=0 | step=24850/50000 progress=49.7% loss=0.102327 lr=5.462310e-05 lr_backbone=5.462310e-05 elapsed=25:49:10 avg_step=3.74s recent_step=4.32s eta=26:07:53 +2026-09-15 16:20:12 | INFO | train rank=0 | step=24860/50000 progress=49.7% loss=0.093301 lr=5.459017e-05 lr_backbone=5.459017e-05 elapsed=25:49:50 avg_step=3.74s recent_step=3.97s eta=26:07:17 +2026-09-15 16:20:45 | INFO | train rank=0 | step=24870/50000 progress=49.7% loss=0.072716 lr=5.455724e-05 lr_backbone=5.455724e-05 elapsed=25:50:23 avg_step=3.74s recent_step=3.27s eta=26:06:35 +2026-09-15 16:21:22 | INFO | train rank=0 | step=24880/50000 progress=49.8% loss=0.087504 lr=5.452431e-05 lr_backbone=5.452431e-05 elapsed=25:51:00 avg_step=3.74s recent_step=3.72s eta=26:05:58 +2026-09-15 16:22:01 | INFO | train rank=0 | step=24890/50000 progress=49.8% loss=0.081396 lr=5.449138e-05 lr_backbone=5.449138e-05 elapsed=25:51:39 avg_step=3.74s recent_step=3.89s eta=26:05:22 +2026-09-15 16:22:45 | INFO | train rank=0 | step=24900/50000 progress=49.8% loss=0.089986 lr=5.445844e-05 lr_backbone=5.445844e-05 elapsed=25:52:23 avg_step=3.74s recent_step=4.45s eta=26:04:51 +2026-09-15 16:23:22 | INFO | train rank=0 | step=24910/50000 progress=49.8% loss=0.085421 lr=5.442550e-05 lr_backbone=5.442550e-05 elapsed=25:53:00 avg_step=3.74s recent_step=3.70s eta=26:04:14 +2026-09-15 16:23:54 | INFO | train rank=0 | step=24920/50000 progress=49.8% loss=0.090679 lr=5.439256e-05 lr_backbone=5.439256e-05 elapsed=25:53:32 avg_step=3.74s recent_step=3.18s eta=26:03:30 +2026-09-15 16:24:32 | INFO | train rank=0 | step=24930/50000 progress=49.9% loss=0.094005 lr=5.435962e-05 lr_backbone=5.435962e-05 elapsed=25:54:10 avg_step=3.74s recent_step=3.76s eta=26:02:53 +2026-09-15 16:25:16 | INFO | train rank=0 | step=24940/50000 progress=49.9% loss=0.119133 lr=5.432667e-05 lr_backbone=5.432667e-05 elapsed=25:54:54 avg_step=3.74s recent_step=4.47s eta=26:02:23 +2026-09-15 16:25:49 | INFO | train rank=0 | step=24950/50000 progress=49.9% loss=0.082941 lr=5.429373e-05 lr_backbone=5.429373e-05 elapsed=25:55:27 avg_step=3.74s recent_step=3.24s eta=26:01:41 +2026-09-15 16:26:19 | INFO | train rank=0 | step=24960/50000 progress=49.9% loss=0.099024 lr=5.426078e-05 lr_backbone=5.426078e-05 elapsed=25:55:57 avg_step=3.74s recent_step=3.05s eta=26:00:56 +2026-09-15 16:27:02 | INFO | train rank=0 | step=24970/50000 progress=49.9% loss=0.106284 lr=5.422783e-05 lr_backbone=5.422783e-05 elapsed=25:56:40 avg_step=3.74s recent_step=4.25s eta=26:00:24 +2026-09-15 16:27:47 | INFO | train rank=0 | step=24980/50000 progress=50.0% loss=0.098156 lr=5.419488e-05 lr_backbone=5.419488e-05 elapsed=25:57:25 avg_step=3.74s recent_step=4.49s eta=25:59:54 +2026-09-15 16:28:27 | INFO | train rank=0 | step=24990/50000 progress=50.0% loss=0.088184 lr=5.416192e-05 lr_backbone=5.416192e-05 elapsed=25:58:05 avg_step=3.74s recent_step=4.08s eta=25:59:20 +2026-09-15 16:28:55 | INFO | train rank=0 | step=25000/50000 progress=50.0% loss=0.100787 lr=5.412897e-05 lr_backbone=5.412897e-05 elapsed=25:58:33 avg_step=3.74s recent_step=2.74s eta=25:58:33 +2026-09-15 16:29:28 | INFO | train rank=0 | step=25010/50000 progress=50.0% loss=0.127674 lr=5.409601e-05 lr_backbone=5.409601e-05 elapsed=25:59:06 avg_step=3.74s recent_step=3.37s eta=25:57:52 +2026-09-15 16:30:09 | INFO | train rank=0 | step=25020/50000 progress=50.0% loss=0.118264 lr=5.406305e-05 lr_backbone=5.406305e-05 elapsed=25:59:47 avg_step=3.74s recent_step=4.06s eta=25:57:17 +2026-09-15 16:30:41 | INFO | train rank=0 | step=25030/50000 progress=50.1% loss=0.088415 lr=5.403009e-05 lr_backbone=5.403009e-05 elapsed=26:00:19 avg_step=3.74s recent_step=3.17s eta=25:56:34 +2026-09-15 16:31:17 | INFO | train rank=0 | step=25040/50000 progress=50.1% loss=0.092655 lr=5.399713e-05 lr_backbone=5.399713e-05 elapsed=26:00:55 avg_step=3.74s recent_step=3.62s eta=25:55:56 +2026-09-15 16:31:55 | INFO | train rank=0 | step=25050/50000 progress=50.1% loss=0.145130 lr=5.396416e-05 lr_backbone=5.396416e-05 elapsed=26:01:33 avg_step=3.74s recent_step=3.85s eta=25:55:19 +2026-09-15 16:32:50 | INFO | train rank=0 | step=25060/50000 progress=50.1% loss=0.084429 lr=5.393120e-05 lr_backbone=5.393120e-05 elapsed=26:02:28 avg_step=3.74s recent_step=5.47s eta=25:54:59 +2026-09-15 16:33:26 | INFO | train rank=0 | step=25070/50000 progress=50.1% loss=0.097893 lr=5.389823e-05 lr_backbone=5.389823e-05 elapsed=26:03:04 avg_step=3.74s recent_step=3.61s eta=25:54:21 +2026-09-15 16:33:54 | INFO | train rank=0 | step=25080/50000 progress=50.2% loss=0.108525 lr=5.386526e-05 lr_backbone=5.386526e-05 elapsed=26:03:32 avg_step=3.74s recent_step=2.73s eta=25:53:33 +2026-09-15 16:34:27 | INFO | train rank=0 | step=25090/50000 progress=50.2% loss=0.082777 lr=5.383229e-05 lr_backbone=5.383229e-05 elapsed=26:04:05 avg_step=3.74s recent_step=3.36s eta=25:52:52 +2026-09-15 16:35:22 | INFO | train rank=0 | step=25100/50000 progress=50.2% loss=0.083332 lr=5.379931e-05 lr_backbone=5.379931e-05 elapsed=26:05:00 avg_step=3.74s recent_step=5.46s eta=25:52:32 +2026-09-15 16:36:00 | INFO | train rank=0 | step=25110/50000 progress=50.2% loss=0.096710 lr=5.376634e-05 lr_backbone=5.376634e-05 elapsed=26:05:38 avg_step=3.74s recent_step=3.82s eta=25:51:55 +2026-09-15 16:36:28 | INFO | train rank=0 | step=25120/50000 progress=50.2% loss=0.080533 lr=5.373336e-05 lr_backbone=5.373336e-05 elapsed=26:06:06 avg_step=3.74s recent_step=2.79s eta=25:51:08 +2026-09-15 16:36:56 | INFO | train rank=0 | step=25130/50000 progress=50.3% loss=0.109620 lr=5.370039e-05 lr_backbone=5.370039e-05 elapsed=26:06:34 avg_step=3.74s recent_step=2.85s eta=25:50:22 +2026-09-15 16:37:39 | INFO | train rank=0 | step=25140/50000 progress=50.3% loss=0.105155 lr=5.366741e-05 lr_backbone=5.366741e-05 elapsed=26:07:17 avg_step=3.74s recent_step=4.28s eta=25:49:50 +2026-09-15 16:38:18 | INFO | train rank=0 | step=25150/50000 progress=50.3% loss=0.077556 lr=5.363443e-05 lr_backbone=5.363443e-05 elapsed=26:07:56 avg_step=3.74s recent_step=3.92s eta=25:49:14 +2026-09-15 16:38:46 | INFO | train rank=0 | step=25160/50000 progress=50.3% loss=0.072298 lr=5.360144e-05 lr_backbone=5.360144e-05 elapsed=26:08:24 avg_step=3.74s recent_step=2.78s eta=25:48:27 +2026-09-15 16:39:22 | INFO | train rank=0 | step=25170/50000 progress=50.3% loss=0.087261 lr=5.356846e-05 lr_backbone=5.356846e-05 elapsed=26:09:00 avg_step=3.74s recent_step=3.63s eta=25:47:49 +2026-09-15 16:40:01 | INFO | train rank=0 | step=25180/50000 progress=50.4% loss=0.087355 lr=5.353547e-05 lr_backbone=5.353547e-05 elapsed=26:09:39 avg_step=3.74s recent_step=3.81s eta=25:47:12 +2026-09-15 16:40:37 | INFO | train rank=0 | step=25190/50000 progress=50.4% loss=0.112234 lr=5.350249e-05 lr_backbone=5.350249e-05 elapsed=26:10:15 avg_step=3.74s recent_step=3.67s eta=25:46:34 +2026-09-15 16:41:22 | INFO | train rank=0 | step=25200/50000 progress=50.4% loss=0.080480 lr=5.346950e-05 lr_backbone=5.346950e-05 elapsed=26:11:00 avg_step=3.74s recent_step=4.44s eta=25:46:03 +2026-09-15 16:42:01 | INFO | train rank=0 | step=25210/50000 progress=50.4% loss=0.090715 lr=5.343651e-05 lr_backbone=5.343651e-05 elapsed=26:11:39 avg_step=3.74s recent_step=3.90s eta=25:45:28 +2026-09-15 16:42:33 | INFO | train rank=0 | step=25220/50000 progress=50.4% loss=0.091178 lr=5.340351e-05 lr_backbone=5.340351e-05 elapsed=26:12:11 avg_step=3.74s recent_step=3.28s eta=25:44:46 +2026-09-15 16:43:15 | INFO | train rank=0 | step=25230/50000 progress=50.5% loss=0.085207 lr=5.337052e-05 lr_backbone=5.337052e-05 elapsed=26:12:53 avg_step=3.74s recent_step=4.12s eta=25:44:12 +2026-09-15 16:44:04 | INFO | train rank=0 | step=25240/50000 progress=50.5% loss=0.098551 lr=5.333753e-05 lr_backbone=5.333753e-05 elapsed=26:13:42 avg_step=3.74s recent_step=4.97s eta=25:43:47 +2026-09-15 16:44:39 | INFO | train rank=0 | step=25250/50000 progress=50.5% loss=0.080213 lr=5.330453e-05 lr_backbone=5.330453e-05 elapsed=26:14:18 avg_step=3.74s recent_step=3.52s eta=25:43:07 +2026-09-15 16:45:08 | INFO | train rank=0 | step=25260/50000 progress=50.5% loss=0.085687 lr=5.327153e-05 lr_backbone=5.327153e-05 elapsed=26:14:46 avg_step=3.74s recent_step=2.89s eta=25:42:21 +2026-09-15 16:45:37 | INFO | train rank=0 | step=25270/50000 progress=50.5% loss=0.101569 lr=5.323853e-05 lr_backbone=5.323853e-05 elapsed=26:15:15 avg_step=3.74s recent_step=2.88s eta=25:41:35 +2026-09-15 16:46:23 | INFO | train rank=0 | step=25280/50000 progress=50.6% loss=0.101728 lr=5.320553e-05 lr_backbone=5.320553e-05 elapsed=26:16:01 avg_step=3.74s recent_step=4.54s eta=25:41:06 +2026-09-15 16:46:56 | INFO | train rank=0 | step=25290/50000 progress=50.6% loss=0.090361 lr=5.317253e-05 lr_backbone=5.317253e-05 elapsed=26:16:34 avg_step=3.74s recent_step=3.29s eta=25:40:24 +2026-09-15 16:47:42 | INFO | train rank=0 | step=25300/50000 progress=50.6% loss=0.088947 lr=5.313953e-05 lr_backbone=5.313953e-05 elapsed=26:17:20 avg_step=3.74s recent_step=4.67s eta=25:39:56 +2026-09-15 16:48:11 | INFO | train rank=0 | step=25310/50000 progress=50.6% loss=0.089502 lr=5.310652e-05 lr_backbone=5.310652e-05 elapsed=26:17:49 avg_step=3.74s recent_step=2.87s eta=25:39:10 +2026-09-15 16:48:52 | INFO | train rank=0 | step=25320/50000 progress=50.6% loss=0.092710 lr=5.307351e-05 lr_backbone=5.307351e-05 elapsed=26:18:30 avg_step=3.74s recent_step=4.08s eta=25:38:36 +2026-09-15 16:49:25 | INFO | train rank=0 | step=25330/50000 progress=50.7% loss=0.099501 lr=5.304051e-05 lr_backbone=5.304051e-05 elapsed=26:19:03 avg_step=3.74s recent_step=3.34s eta=25:37:55 +2026-09-15 16:50:07 | INFO | train rank=0 | step=25340/50000 progress=50.7% loss=0.099493 lr=5.300750e-05 lr_backbone=5.300750e-05 elapsed=26:19:45 avg_step=3.74s recent_step=4.20s eta=25:37:22 +2026-09-15 16:50:44 | INFO | train rank=0 | step=25350/50000 progress=50.7% loss=0.094123 lr=5.297449e-05 lr_backbone=5.297449e-05 elapsed=26:20:22 avg_step=3.74s recent_step=3.73s eta=25:36:44 +2026-09-15 16:51:22 | INFO | train rank=0 | step=25360/50000 progress=50.7% loss=0.086183 lr=5.294148e-05 lr_backbone=5.294148e-05 elapsed=26:21:00 avg_step=3.74s recent_step=3.72s eta=25:36:07 +2026-09-15 16:52:01 | INFO | train rank=0 | step=25370/50000 progress=50.7% loss=0.098327 lr=5.290846e-05 lr_backbone=5.290846e-05 elapsed=26:21:39 avg_step=3.74s recent_step=3.93s eta=25:35:31 +2026-09-15 16:52:44 | INFO | train rank=0 | step=25380/50000 progress=50.8% loss=0.117665 lr=5.287545e-05 lr_backbone=5.287545e-05 elapsed=26:22:22 avg_step=3.74s recent_step=4.34s eta=25:34:59 +2026-09-15 16:53:18 | INFO | train rank=0 | step=25390/50000 progress=50.8% loss=0.099498 lr=5.284243e-05 lr_backbone=5.284243e-05 elapsed=26:22:56 avg_step=3.74s recent_step=3.39s eta=25:34:19 +2026-09-15 16:53:47 | INFO | train rank=0 | step=25400/50000 progress=50.8% loss=0.102526 lr=5.280942e-05 lr_backbone=5.280942e-05 elapsed=26:23:25 avg_step=3.74s recent_step=2.92s eta=25:33:33 +2026-09-15 16:54:22 | INFO | train rank=0 | step=25410/50000 progress=50.8% loss=0.084373 lr=5.277640e-05 lr_backbone=5.277640e-05 elapsed=26:24:00 avg_step=3.74s recent_step=3.41s eta=25:32:53 +2026-09-15 16:55:14 | INFO | train rank=0 | step=25420/50000 progress=50.8% loss=0.088708 lr=5.274338e-05 lr_backbone=5.274338e-05 elapsed=26:24:52 avg_step=3.74s recent_step=5.25s eta=25:32:30 +2026-09-15 16:55:49 | INFO | train rank=0 | step=25430/50000 progress=50.9% loss=0.099704 lr=5.271036e-05 lr_backbone=5.271036e-05 elapsed=26:25:27 avg_step=3.74s recent_step=3.54s eta=25:31:50 +2026-09-15 16:56:23 | INFO | train rank=0 | step=25440/50000 progress=50.9% loss=0.096360 lr=5.267734e-05 lr_backbone=5.267734e-05 elapsed=26:26:01 avg_step=3.74s recent_step=3.30s eta=25:31:09 +2026-09-15 16:56:54 | INFO | train rank=0 | step=25450/50000 progress=50.9% loss=0.080248 lr=5.264432e-05 lr_backbone=5.264432e-05 elapsed=26:26:32 avg_step=3.74s recent_step=3.15s eta=25:30:26 +2026-09-15 16:57:48 | INFO | train rank=0 | step=25460/50000 progress=50.9% loss=0.094057 lr=5.261129e-05 lr_backbone=5.261129e-05 elapsed=26:27:26 avg_step=3.74s recent_step=5.38s eta=25:30:04 +2026-09-15 16:58:23 | INFO | train rank=0 | step=25470/50000 progress=50.9% loss=0.093807 lr=5.257827e-05 lr_backbone=5.257827e-05 elapsed=26:28:01 avg_step=3.74s recent_step=3.51s eta=25:29:24 +2026-09-15 16:58:51 | INFO | train rank=0 | step=25480/50000 progress=51.0% loss=0.096308 lr=5.254524e-05 lr_backbone=5.254524e-05 elapsed=26:28:29 avg_step=3.74s recent_step=2.82s eta=25:28:38 +2026-09-15 16:59:28 | INFO | train rank=0 | step=25490/50000 progress=51.0% loss=0.072493 lr=5.251222e-05 lr_backbone=5.251222e-05 elapsed=26:29:06 avg_step=3.74s recent_step=3.67s eta=25:28:00 +2026-09-15 17:00:10 | INFO | train rank=0 | step=25500/50000 progress=51.0% loss=0.098081 lr=5.247919e-05 lr_backbone=5.247919e-05 elapsed=26:29:48 avg_step=3.74s recent_step=4.19s eta=25:27:27 +2026-09-15 17:00:43 | INFO | train rank=0 | step=25510/50000 progress=51.0% loss=0.104084 lr=5.244616e-05 lr_backbone=5.244616e-05 elapsed=26:30:21 avg_step=3.74s recent_step=3.37s eta=25:26:46 +2026-09-15 17:01:15 | INFO | train rank=0 | step=25520/50000 progress=51.0% loss=0.075974 lr=5.241313e-05 lr_backbone=5.241313e-05 elapsed=26:30:53 avg_step=3.74s recent_step=3.13s eta=25:26:03 +2026-09-15 17:01:49 | INFO | train rank=0 | step=25530/50000 progress=51.1% loss=0.082595 lr=5.238010e-05 lr_backbone=5.238010e-05 elapsed=26:31:27 avg_step=3.74s recent_step=3.43s eta=25:25:22 +2026-09-15 17:02:23 | INFO | train rank=0 | step=25540/50000 progress=51.1% loss=0.075582 lr=5.234706e-05 lr_backbone=5.234706e-05 elapsed=26:32:01 avg_step=3.74s recent_step=3.41s eta=25:24:42 +2026-09-15 17:02:50 | INFO | train rank=0 | step=25550/50000 progress=51.1% loss=0.076982 lr=5.231403e-05 lr_backbone=5.231403e-05 elapsed=26:32:28 avg_step=3.74s recent_step=2.71s eta=25:23:54 +2026-09-15 17:03:14 | INFO | train rank=0 | step=25560/50000 progress=51.1% loss=0.109832 lr=5.228100e-05 lr_backbone=5.228100e-05 elapsed=26:32:52 avg_step=3.74s recent_step=2.37s eta=25:23:04 +2026-09-15 17:03:40 | INFO | train rank=0 | step=25570/50000 progress=51.1% loss=0.103945 lr=5.224796e-05 lr_backbone=5.224796e-05 elapsed=26:33:18 avg_step=3.74s recent_step=2.62s eta=25:22:16 +2026-09-15 17:04:28 | INFO | train rank=0 | step=25580/50000 progress=51.2% loss=0.083516 lr=5.221492e-05 lr_backbone=5.221492e-05 elapsed=26:34:06 avg_step=3.74s recent_step=4.76s eta=25:21:48 +2026-09-15 17:04:49 | INFO | train rank=0 | step=25590/50000 progress=51.2% loss=0.111685 lr=5.218189e-05 lr_backbone=5.218189e-05 elapsed=26:34:27 avg_step=3.74s recent_step=2.16s eta=25:20:56 +2026-09-15 17:05:15 | INFO | train rank=0 | step=25600/50000 progress=51.2% loss=0.084187 lr=5.214885e-05 lr_backbone=5.214885e-05 elapsed=26:34:53 avg_step=3.74s recent_step=2.57s eta=25:20:07 +2026-09-15 17:05:41 | INFO | train rank=0 | step=25610/50000 progress=51.2% loss=0.089951 lr=5.211581e-05 lr_backbone=5.211581e-05 elapsed=26:35:19 avg_step=3.74s recent_step=2.58s eta=25:19:19 +2026-09-15 17:06:14 | INFO | train rank=0 | step=25620/50000 progress=51.2% loss=0.091211 lr=5.208277e-05 lr_backbone=5.208277e-05 elapsed=26:35:52 avg_step=3.74s recent_step=3.35s eta=25:18:38 +2026-09-15 17:06:39 | INFO | train rank=0 | step=25630/50000 progress=51.3% loss=0.106295 lr=5.204973e-05 lr_backbone=5.204973e-05 elapsed=26:36:17 avg_step=3.74s recent_step=2.45s eta=25:17:48 +2026-09-15 17:07:06 | INFO | train rank=0 | step=25640/50000 progress=51.3% loss=0.087764 lr=5.201669e-05 lr_backbone=5.201669e-05 elapsed=26:36:45 avg_step=3.74s recent_step=2.79s eta=25:17:02 +2026-09-15 17:07:31 | INFO | train rank=0 | step=25650/50000 progress=51.3% loss=0.087690 lr=5.198364e-05 lr_backbone=5.198364e-05 elapsed=26:37:09 avg_step=3.74s recent_step=2.42s eta=25:16:12 +2026-09-15 17:08:09 | INFO | train rank=0 | step=25660/50000 progress=51.3% loss=0.082623 lr=5.195060e-05 lr_backbone=5.195060e-05 elapsed=26:37:47 avg_step=3.74s recent_step=3.88s eta=25:15:36 +2026-09-15 17:08:32 | INFO | train rank=0 | step=25670/50000 progress=51.3% loss=0.097125 lr=5.191755e-05 lr_backbone=5.191755e-05 elapsed=26:38:10 avg_step=3.74s recent_step=2.27s eta=25:14:45 +2026-09-15 17:08:59 | INFO | train rank=0 | step=25680/50000 progress=51.4% loss=0.097120 lr=5.188451e-05 lr_backbone=5.188451e-05 elapsed=26:38:37 avg_step=3.74s recent_step=2.64s eta=25:13:57 +2026-09-15 17:09:17 | INFO | train rank=0 | step=25690/50000 progress=51.4% loss=0.121116 lr=5.185146e-05 lr_backbone=5.185146e-05 elapsed=26:38:55 avg_step=3.73s recent_step=1.84s eta=25:13:02 +2026-09-15 17:09:56 | INFO | train rank=0 | step=25700/50000 progress=51.4% loss=0.114501 lr=5.181842e-05 lr_backbone=5.181842e-05 elapsed=26:39:35 avg_step=3.73s recent_step=3.95s eta=25:12:26 +2026-09-15 17:10:25 | INFO | train rank=0 | step=25710/50000 progress=51.4% loss=0.112247 lr=5.178537e-05 lr_backbone=5.178537e-05 elapsed=26:40:03 avg_step=3.73s recent_step=2.86s eta=25:11:41 +2026-09-15 17:10:54 | INFO | train rank=0 | step=25720/50000 progress=51.4% loss=0.078069 lr=5.175232e-05 lr_backbone=5.175232e-05 elapsed=26:40:32 avg_step=3.73s recent_step=2.92s eta=25:10:56 +2026-09-15 17:11:23 | INFO | train rank=0 | step=25730/50000 progress=51.5% loss=0.099896 lr=5.171927e-05 lr_backbone=5.171927e-05 elapsed=26:41:01 avg_step=3.73s recent_step=2.90s eta=25:10:10 +2026-09-15 17:11:59 | INFO | train rank=0 | step=25740/50000 progress=51.5% loss=0.098607 lr=5.168622e-05 lr_backbone=5.168622e-05 elapsed=26:41:37 avg_step=3.73s recent_step=3.53s eta=25:09:31 +2026-09-15 17:12:25 | INFO | train rank=0 | step=25750/50000 progress=51.5% loss=0.091003 lr=5.165317e-05 lr_backbone=5.165317e-05 elapsed=26:42:03 avg_step=3.73s recent_step=2.68s eta=25:08:44 +2026-09-15 17:12:52 | INFO | train rank=0 | step=25760/50000 progress=51.5% loss=0.102768 lr=5.162012e-05 lr_backbone=5.162012e-05 elapsed=26:42:30 avg_step=3.73s recent_step=2.61s eta=25:07:56 +2026-09-15 17:13:22 | INFO | train rank=0 | step=25770/50000 progress=51.5% loss=0.087808 lr=5.158706e-05 lr_backbone=5.158706e-05 elapsed=26:43:00 avg_step=3.73s recent_step=3.03s eta=25:07:12 +2026-09-15 17:14:01 | INFO | train rank=0 | step=25780/50000 progress=51.6% loss=0.129653 lr=5.155401e-05 lr_backbone=5.155401e-05 elapsed=26:43:39 avg_step=3.73s recent_step=3.93s eta=25:06:37 +2026-09-15 17:14:22 | INFO | train rank=0 | step=25790/50000 progress=51.6% loss=0.087749 lr=5.152096e-05 lr_backbone=5.152096e-05 elapsed=26:44:00 avg_step=3.73s recent_step=2.11s eta=25:05:44 +2026-09-15 17:14:50 | INFO | train rank=0 | step=25800/50000 progress=51.6% loss=0.093730 lr=5.148790e-05 lr_backbone=5.148790e-05 elapsed=26:44:28 avg_step=3.73s recent_step=2.79s eta=25:04:58 +2026-09-15 17:15:15 | INFO | train rank=0 | step=25810/50000 progress=51.6% loss=0.099929 lr=5.145485e-05 lr_backbone=5.145485e-05 elapsed=26:44:53 avg_step=3.73s recent_step=2.47s eta=25:04:09 +2026-09-15 17:15:54 | INFO | train rank=0 | step=25820/50000 progress=51.6% loss=0.092731 lr=5.142179e-05 lr_backbone=5.142179e-05 elapsed=26:45:32 avg_step=3.73s recent_step=3.92s eta=25:03:33 +2026-09-15 17:16:20 | INFO | train rank=0 | step=25830/50000 progress=51.7% loss=0.092012 lr=5.138874e-05 lr_backbone=5.138874e-05 elapsed=26:45:58 avg_step=3.73s recent_step=2.60s eta=25:02:45 +2026-09-15 17:16:46 | INFO | train rank=0 | step=25840/50000 progress=51.7% loss=0.101507 lr=5.135568e-05 lr_backbone=5.135568e-05 elapsed=26:46:24 avg_step=3.73s recent_step=2.59s eta=25:01:57 +2026-09-15 17:17:12 | INFO | train rank=0 | step=25850/50000 progress=51.7% loss=0.080200 lr=5.132262e-05 lr_backbone=5.132262e-05 elapsed=26:46:50 avg_step=3.73s recent_step=2.62s eta=25:01:10 +2026-09-15 17:17:48 | INFO | train rank=0 | step=25860/50000 progress=51.7% loss=0.076929 lr=5.128956e-05 lr_backbone=5.128956e-05 elapsed=26:47:26 avg_step=3.73s recent_step=3.55s eta=25:00:31 +2026-09-15 17:18:14 | INFO | train rank=0 | step=25870/50000 progress=51.7% loss=0.109621 lr=5.125650e-05 lr_backbone=5.125650e-05 elapsed=26:47:52 avg_step=3.73s recent_step=2.67s eta=24:59:44 +2026-09-15 17:18:39 | INFO | train rank=0 | step=25880/50000 progress=51.8% loss=0.091668 lr=5.122345e-05 lr_backbone=5.122345e-05 elapsed=26:48:17 avg_step=3.73s recent_step=2.52s eta=24:58:55 +2026-09-15 17:19:00 | INFO | train rank=0 | step=25890/50000 progress=51.8% loss=0.081218 lr=5.119039e-05 lr_backbone=5.119039e-05 elapsed=26:48:38 avg_step=3.73s recent_step=2.02s eta=24:58:02 +2026-09-15 17:19:39 | INFO | train rank=0 | step=25900/50000 progress=51.8% loss=0.095141 lr=5.115733e-05 lr_backbone=5.115733e-05 elapsed=26:49:17 avg_step=3.73s recent_step=3.97s eta=24:57:27 +2026-09-15 17:20:11 | INFO | train rank=0 | step=25910/50000 progress=51.8% loss=0.096246 lr=5.112426e-05 lr_backbone=5.112426e-05 elapsed=26:49:49 avg_step=3.73s recent_step=3.19s eta=24:56:44 +2026-09-15 17:20:40 | INFO | train rank=0 | step=25920/50000 progress=51.8% loss=0.106355 lr=5.109120e-05 lr_backbone=5.109120e-05 elapsed=26:50:18 avg_step=3.73s recent_step=2.88s eta=24:55:59 +2026-09-15 17:21:05 | INFO | train rank=0 | step=25930/50000 progress=51.9% loss=0.111711 lr=5.105814e-05 lr_backbone=5.105814e-05 elapsed=26:50:43 avg_step=3.73s recent_step=2.55s eta=24:55:11 +2026-09-15 17:21:36 | INFO | train rank=0 | step=25940/50000 progress=51.9% loss=0.085690 lr=5.102508e-05 lr_backbone=5.102508e-05 elapsed=26:51:14 avg_step=3.73s recent_step=3.09s eta=24:54:28 +2026-09-15 17:21:59 | INFO | train rank=0 | step=25950/50000 progress=51.9% loss=0.092733 lr=5.099202e-05 lr_backbone=5.099202e-05 elapsed=26:51:37 avg_step=3.73s recent_step=2.24s eta=24:53:37 +2026-09-15 17:22:26 | INFO | train rank=0 | step=25960/50000 progress=51.9% loss=0.088102 lr=5.095895e-05 lr_backbone=5.095895e-05 elapsed=26:52:04 avg_step=3.73s recent_step=2.76s eta=24:52:51 +2026-09-15 17:22:57 | INFO | train rank=0 | step=25970/50000 progress=51.9% loss=0.093846 lr=5.092589e-05 lr_backbone=5.092589e-05 elapsed=26:52:35 avg_step=3.73s recent_step=3.04s eta=24:52:07 +2026-09-15 17:23:29 | INFO | train rank=0 | step=25980/50000 progress=52.0% loss=0.070239 lr=5.089283e-05 lr_backbone=5.089283e-05 elapsed=26:53:07 avg_step=3.73s recent_step=3.25s eta=24:51:25 +2026-09-15 17:23:49 | INFO | train rank=0 | step=25990/50000 progress=52.0% loss=0.086101 lr=5.085976e-05 lr_backbone=5.085976e-05 elapsed=26:53:27 avg_step=3.72s recent_step=1.96s eta=24:50:32 +2026-09-15 17:24:12 | INFO | train rank=0 | step=26000/50000 progress=52.0% loss=0.101927 lr=5.082670e-05 lr_backbone=5.082670e-05 elapsed=26:53:50 avg_step=3.72s recent_step=2.34s eta=24:49:42 +2026-09-15 17:24:43 | INFO | train rank=0 | step=26010/50000 progress=52.0% loss=0.110315 lr=5.079363e-05 lr_backbone=5.079363e-05 elapsed=26:54:21 avg_step=3.72s recent_step=3.04s eta=24:48:58 +2026-09-15 17:25:17 | INFO | train rank=0 | step=26020/50000 progress=52.0% loss=0.090594 lr=5.076057e-05 lr_backbone=5.076057e-05 elapsed=26:54:55 avg_step=3.72s recent_step=3.40s eta=24:48:18 +2026-09-15 17:25:45 | INFO | train rank=0 | step=26030/50000 progress=52.1% loss=0.104674 lr=5.072750e-05 lr_backbone=5.072750e-05 elapsed=26:55:23 avg_step=3.72s recent_step=2.83s eta=24:47:33 +2026-09-15 17:26:12 | INFO | train rank=0 | step=26040/50000 progress=52.1% loss=0.105096 lr=5.069443e-05 lr_backbone=5.069443e-05 elapsed=26:55:50 avg_step=3.72s recent_step=2.71s eta=24:46:46 +2026-09-15 17:26:42 | INFO | train rank=0 | step=26050/50000 progress=52.1% loss=0.093324 lr=5.066137e-05 lr_backbone=5.066137e-05 elapsed=26:56:20 avg_step=3.72s recent_step=2.94s eta=24:46:02 +2026-09-15 17:27:12 | INFO | train rank=0 | step=26060/50000 progress=52.1% loss=0.092955 lr=5.062830e-05 lr_backbone=5.062830e-05 elapsed=26:56:51 avg_step=3.72s recent_step=3.10s eta=24:45:19 +2026-09-15 17:27:41 | INFO | train rank=0 | step=26070/50000 progress=52.1% loss=0.084064 lr=5.059524e-05 lr_backbone=5.059524e-05 elapsed=26:57:19 avg_step=3.72s recent_step=2.83s eta=24:44:33 +2026-09-15 17:28:04 | INFO | train rank=0 | step=26080/50000 progress=52.2% loss=0.092607 lr=5.056217e-05 lr_backbone=5.056217e-05 elapsed=26:57:42 avg_step=3.72s recent_step=2.32s eta=24:43:43 +2026-09-15 17:28:36 | INFO | train rank=0 | step=26090/50000 progress=52.2% loss=0.087281 lr=5.052910e-05 lr_backbone=5.052910e-05 elapsed=26:58:14 avg_step=3.72s recent_step=3.16s eta=24:43:01 +2026-09-15 17:29:13 | INFO | train rank=0 | step=26100/50000 progress=52.2% loss=0.107452 lr=5.049603e-05 lr_backbone=5.049603e-05 elapsed=26:58:51 avg_step=3.72s recent_step=3.75s eta=24:42:24 +2026-09-15 17:29:41 | INFO | train rank=0 | step=26110/50000 progress=52.2% loss=0.130733 lr=5.046296e-05 lr_backbone=5.046296e-05 elapsed=26:59:19 avg_step=3.72s recent_step=2.76s eta=24:41:38 +2026-09-15 17:30:10 | INFO | train rank=0 | step=26120/50000 progress=52.2% loss=0.099050 lr=5.042990e-05 lr_backbone=5.042990e-05 elapsed=26:59:48 avg_step=3.72s recent_step=2.93s eta=24:40:53 +2026-09-15 17:30:36 | INFO | train rank=0 | step=26130/50000 progress=52.3% loss=0.103221 lr=5.039683e-05 lr_backbone=5.039683e-05 elapsed=27:00:14 avg_step=3.72s recent_step=2.62s eta=24:40:06 +2026-09-15 17:31:10 | INFO | train rank=0 | step=26140/50000 progress=52.3% loss=0.106799 lr=5.036376e-05 lr_backbone=5.036376e-05 elapsed=27:00:48 avg_step=3.72s recent_step=3.43s eta=24:39:26 +2026-09-15 17:31:45 | INFO | train rank=0 | step=26150/50000 progress=52.3% loss=0.098349 lr=5.033069e-05 lr_backbone=5.033069e-05 elapsed=27:01:23 avg_step=3.72s recent_step=3.46s eta=24:38:47 +2026-09-15 17:32:14 | INFO | train rank=0 | step=26160/50000 progress=52.3% loss=0.090140 lr=5.029762e-05 lr_backbone=5.029762e-05 elapsed=27:01:52 avg_step=3.72s recent_step=2.90s eta=24:38:02 +2026-09-15 17:32:42 | INFO | train rank=0 | step=26170/50000 progress=52.3% loss=0.083130 lr=5.026455e-05 lr_backbone=5.026455e-05 elapsed=27:02:20 avg_step=3.72s recent_step=2.75s eta=24:37:16 +2026-09-15 17:33:11 | INFO | train rank=0 | step=26180/50000 progress=52.4% loss=0.100882 lr=5.023148e-05 lr_backbone=5.023148e-05 elapsed=27:02:49 avg_step=3.72s recent_step=2.91s eta=24:36:31 +2026-09-15 17:33:34 | INFO | train rank=0 | step=26190/50000 progress=52.4% loss=0.097859 lr=5.019842e-05 lr_backbone=5.019842e-05 elapsed=27:03:12 avg_step=3.72s recent_step=2.31s eta=24:35:41 +2026-09-15 17:34:00 | INFO | train rank=0 | step=26200/50000 progress=52.4% loss=0.110411 lr=5.016535e-05 lr_backbone=5.016535e-05 elapsed=27:03:38 avg_step=3.72s recent_step=2.66s eta=24:34:55 +2026-09-15 17:34:30 | INFO | train rank=0 | step=26210/50000 progress=52.4% loss=0.085634 lr=5.013228e-05 lr_backbone=5.013228e-05 elapsed=27:04:08 avg_step=3.72s recent_step=2.94s eta=24:34:10 +2026-09-15 17:35:09 | INFO | train rank=0 | step=26220/50000 progress=52.4% loss=0.104485 lr=5.009921e-05 lr_backbone=5.009921e-05 elapsed=27:04:47 avg_step=3.72s recent_step=3.91s eta=24:33:35 +2026-09-15 17:35:27 | INFO | train rank=0 | step=26230/50000 progress=52.5% loss=0.090223 lr=5.006614e-05 lr_backbone=5.006614e-05 elapsed=27:05:05 avg_step=3.72s recent_step=1.82s eta=24:32:40 +2026-09-15 17:35:52 | INFO | train rank=0 | step=26240/50000 progress=52.5% loss=0.093859 lr=5.003307e-05 lr_backbone=5.003307e-05 elapsed=27:05:30 avg_step=3.72s recent_step=2.52s eta=24:31:52 +2026-09-15 17:36:15 | INFO | train rank=0 | step=26250/50000 progress=52.5% loss=0.085649 lr=5.000000e-05 lr_backbone=5.000000e-05 elapsed=27:05:53 avg_step=3.72s recent_step=2.30s eta=24:31:02 +2026-09-15 17:36:52 | INFO | train rank=0 | step=26260/50000 progress=52.5% loss=0.106795 lr=4.996693e-05 lr_backbone=4.996693e-05 elapsed=27:06:30 avg_step=3.72s recent_step=3.64s eta=24:30:25 +2026-09-15 17:37:19 | INFO | train rank=0 | step=26270/50000 progress=52.5% loss=0.080680 lr=4.993386e-05 lr_backbone=4.993386e-05 elapsed=27:06:57 avg_step=3.72s recent_step=2.77s eta=24:29:39 +2026-09-15 17:37:45 | INFO | train rank=0 | step=26280/50000 progress=52.6% loss=0.117228 lr=4.990079e-05 lr_backbone=4.990079e-05 elapsed=27:07:24 avg_step=3.72s recent_step=2.62s eta=24:28:52 +2026-09-15 17:38:16 | INFO | train rank=0 | step=26290/50000 progress=52.6% loss=0.084664 lr=4.986772e-05 lr_backbone=4.986772e-05 elapsed=27:07:54 avg_step=3.72s recent_step=3.09s eta=24:28:09 +2026-09-15 17:38:37 | INFO | train rank=0 | step=26300/50000 progress=52.6% loss=0.089113 lr=4.983465e-05 lr_backbone=4.983465e-05 elapsed=27:08:15 avg_step=3.71s recent_step=2.09s eta=24:27:17 +2026-09-15 17:39:07 | INFO | train rank=0 | step=26310/50000 progress=52.6% loss=0.095459 lr=4.980158e-05 lr_backbone=4.980158e-05 elapsed=27:08:45 avg_step=3.71s recent_step=2.95s eta=24:26:33 +2026-09-15 17:39:35 | INFO | train rank=0 | step=26320/50000 progress=52.6% loss=0.084200 lr=4.976852e-05 lr_backbone=4.976852e-05 elapsed=27:09:13 avg_step=3.71s recent_step=2.81s eta=24:25:48 +2026-09-15 17:40:03 | INFO | train rank=0 | step=26330/50000 progress=52.7% loss=0.118555 lr=4.973545e-05 lr_backbone=4.973545e-05 elapsed=27:09:41 avg_step=3.71s recent_step=2.79s eta=24:25:02 +2026-09-15 17:40:28 | INFO | train rank=0 | step=26340/50000 progress=52.7% loss=0.125922 lr=4.970238e-05 lr_backbone=4.970238e-05 elapsed=27:10:06 avg_step=3.71s recent_step=2.53s eta=24:24:15 +2026-09-15 17:40:55 | INFO | train rank=0 | step=26350/50000 progress=52.7% loss=0.095439 lr=4.966931e-05 lr_backbone=4.966931e-05 elapsed=27:10:33 avg_step=3.71s recent_step=2.67s eta=24:23:28 +2026-09-15 17:41:18 | INFO | train rank=0 | step=26360/50000 progress=52.7% loss=0.080438 lr=4.963624e-05 lr_backbone=4.963624e-05 elapsed=27:10:56 avg_step=3.71s recent_step=2.37s eta=24:22:39 +2026-09-15 17:41:48 | INFO | train rank=0 | step=26370/50000 progress=52.7% loss=0.099130 lr=4.960317e-05 lr_backbone=4.960317e-05 elapsed=27:11:26 avg_step=3.71s recent_step=2.98s eta=24:21:55 +2026-09-15 17:42:22 | INFO | train rank=0 | step=26380/50000 progress=52.8% loss=0.097465 lr=4.957010e-05 lr_backbone=4.957010e-05 elapsed=27:12:00 avg_step=3.71s recent_step=3.36s eta=24:21:15 +2026-09-15 17:42:51 | INFO | train rank=0 | step=26390/50000 progress=52.8% loss=0.087096 lr=4.953704e-05 lr_backbone=4.953704e-05 elapsed=27:12:29 avg_step=3.71s recent_step=2.89s eta=24:20:30 +2026-09-15 17:43:17 | INFO | train rank=0 | step=26400/50000 progress=52.8% loss=0.117436 lr=4.950397e-05 lr_backbone=4.950397e-05 elapsed=27:12:55 avg_step=3.71s recent_step=2.59s eta=24:19:43 +2026-09-15 17:43:57 | INFO | train rank=0 | step=26410/50000 progress=52.8% loss=0.109340 lr=4.947090e-05 lr_backbone=4.947090e-05 elapsed=27:13:35 avg_step=3.71s recent_step=4.01s eta=24:19:09 +2026-09-15 17:44:25 | INFO | train rank=0 | step=26420/50000 progress=52.8% loss=0.090009 lr=4.943783e-05 lr_backbone=4.943783e-05 elapsed=27:14:03 avg_step=3.71s recent_step=2.83s eta=24:18:24 +2026-09-15 17:44:51 | INFO | train rank=0 | step=26430/50000 progress=52.9% loss=0.088557 lr=4.940476e-05 lr_backbone=4.940476e-05 elapsed=27:14:29 avg_step=3.71s recent_step=2.59s eta=24:17:37 +2026-09-15 17:45:15 | INFO | train rank=0 | step=26440/50000 progress=52.9% loss=0.103371 lr=4.937170e-05 lr_backbone=4.937170e-05 elapsed=27:14:53 avg_step=3.71s recent_step=2.37s eta=24:16:48 +2026-09-15 17:45:48 | INFO | train rank=0 | step=26450/50000 progress=52.9% loss=0.089045 lr=4.933863e-05 lr_backbone=4.933863e-05 elapsed=27:15:26 avg_step=3.71s recent_step=3.38s eta=24:16:08 +2026-09-15 17:46:14 | INFO | train rank=0 | step=26460/50000 progress=52.9% loss=0.084940 lr=4.930557e-05 lr_backbone=4.930557e-05 elapsed=27:15:52 avg_step=3.71s recent_step=2.59s eta=24:15:21 +2026-09-15 17:46:42 | INFO | train rank=0 | step=26470/50000 progress=52.9% loss=0.115324 lr=4.927250e-05 lr_backbone=4.927250e-05 elapsed=27:16:20 avg_step=3.71s recent_step=2.74s eta=24:14:35 +2026-09-15 17:47:09 | INFO | train rank=0 | step=26480/50000 progress=53.0% loss=0.082978 lr=4.923943e-05 lr_backbone=4.923943e-05 elapsed=27:16:47 avg_step=3.71s recent_step=2.78s eta=24:13:50 +2026-09-15 17:47:42 | INFO | train rank=0 | step=26490/50000 progress=53.0% loss=0.105993 lr=4.920637e-05 lr_backbone=4.920637e-05 elapsed=27:17:20 avg_step=3.71s recent_step=3.24s eta=24:13:08 +2026-09-15 17:48:10 | INFO | train rank=0 | step=26500/50000 progress=53.0% loss=0.088121 lr=4.917330e-05 lr_backbone=4.917330e-05 elapsed=27:17:48 avg_step=3.71s recent_step=2.77s eta=24:12:23 +2026-09-15 17:48:32 | INFO | train rank=0 | step=26510/50000 progress=53.0% loss=0.102586 lr=4.914024e-05 lr_backbone=4.914024e-05 elapsed=27:18:11 avg_step=3.71s recent_step=2.28s eta=24:11:33 +2026-09-15 17:48:57 | INFO | train rank=0 | step=26520/50000 progress=53.0% loss=0.114830 lr=4.910717e-05 lr_backbone=4.910717e-05 elapsed=27:18:35 avg_step=3.71s recent_step=2.47s eta=24:10:45 +2026-09-15 17:49:34 | INFO | train rank=0 | step=26530/50000 progress=53.1% loss=0.096356 lr=4.907411e-05 lr_backbone=4.907411e-05 elapsed=27:19:12 avg_step=3.71s recent_step=3.65s eta=24:10:08 +2026-09-15 17:50:00 | INFO | train rank=0 | step=26540/50000 progress=53.1% loss=0.105600 lr=4.904105e-05 lr_backbone=4.904105e-05 elapsed=27:19:38 avg_step=3.71s recent_step=2.61s eta=24:09:21 +2026-09-15 17:50:26 | INFO | train rank=0 | step=26550/50000 progress=53.1% loss=0.089486 lr=4.900798e-05 lr_backbone=4.900798e-05 elapsed=27:20:04 avg_step=3.71s recent_step=2.61s eta=24:08:34 +2026-09-15 17:50:53 | INFO | train rank=0 | step=26560/50000 progress=53.1% loss=0.096396 lr=4.897492e-05 lr_backbone=4.897492e-05 elapsed=27:20:31 avg_step=3.71s recent_step=2.71s eta=24:07:48 +2026-09-15 17:51:26 | INFO | train rank=0 | step=26570/50000 progress=53.1% loss=0.080494 lr=4.894186e-05 lr_backbone=4.894186e-05 elapsed=27:21:04 avg_step=3.71s recent_step=3.27s eta=24:07:07 +2026-09-15 17:51:50 | INFO | train rank=0 | step=26580/50000 progress=53.2% loss=0.113631 lr=4.890880e-05 lr_backbone=4.890880e-05 elapsed=27:21:28 avg_step=3.71s recent_step=2.45s eta=24:06:19 +2026-09-15 17:52:18 | INFO | train rank=0 | step=26590/50000 progress=53.2% loss=0.094196 lr=4.887574e-05 lr_backbone=4.887574e-05 elapsed=27:21:56 avg_step=3.71s recent_step=2.76s eta=24:05:34 +2026-09-15 17:52:41 | INFO | train rank=0 | step=26600/50000 progress=53.2% loss=0.106027 lr=4.884267e-05 lr_backbone=4.884267e-05 elapsed=27:22:19 avg_step=3.70s recent_step=2.35s eta=24:04:45 +2026-09-15 17:53:08 | INFO | train rank=0 | step=26610/50000 progress=53.2% loss=0.094109 lr=4.880961e-05 lr_backbone=4.880961e-05 elapsed=27:22:46 avg_step=3.70s recent_step=2.71s eta=24:03:59 +2026-09-15 17:53:33 | INFO | train rank=0 | step=26620/50000 progress=53.2% loss=0.096003 lr=4.877655e-05 lr_backbone=4.877655e-05 elapsed=27:23:11 avg_step=3.70s recent_step=2.50s eta=24:03:12 +2026-09-15 17:53:55 | INFO | train rank=0 | step=26630/50000 progress=53.3% loss=0.091237 lr=4.874350e-05 lr_backbone=4.874350e-05 elapsed=27:23:33 avg_step=3.70s recent_step=2.17s eta=24:02:21 +2026-09-15 17:54:13 | INFO | train rank=0 | step=26640/50000 progress=53.3% loss=0.117912 lr=4.871044e-05 lr_backbone=4.871044e-05 elapsed=27:23:51 avg_step=3.70s recent_step=1.82s eta=24:01:27 +2026-09-15 17:54:38 | INFO | train rank=0 | step=26650/50000 progress=53.3% loss=0.104653 lr=4.867738e-05 lr_backbone=4.867738e-05 elapsed=27:24:16 avg_step=3.70s recent_step=2.45s eta=24:00:39 +2026-09-15 17:55:01 | INFO | train rank=0 | step=26660/50000 progress=53.3% loss=0.103236 lr=4.864432e-05 lr_backbone=4.864432e-05 elapsed=27:24:39 avg_step=3.70s recent_step=2.31s eta=23:59:50 +2026-09-15 17:55:21 | INFO | train rank=0 | step=26670/50000 progress=53.3% loss=0.088770 lr=4.861126e-05 lr_backbone=4.861126e-05 elapsed=27:24:59 avg_step=3.70s recent_step=2.00s eta=23:58:58 +2026-09-15 17:55:38 | INFO | train rank=0 | step=26680/50000 progress=53.4% loss=0.091394 lr=4.857821e-05 lr_backbone=4.857821e-05 elapsed=27:25:16 avg_step=3.70s recent_step=1.75s eta=23:58:04 +2026-09-15 17:55:59 | INFO | train rank=0 | step=26690/50000 progress=53.4% loss=0.096873 lr=4.854515e-05 lr_backbone=4.854515e-05 elapsed=27:25:37 avg_step=3.70s recent_step=2.09s eta=23:57:13 +2026-09-15 17:56:20 | INFO | train rank=0 | step=26700/50000 progress=53.4% loss=0.076781 lr=4.851210e-05 lr_backbone=4.851210e-05 elapsed=27:25:58 avg_step=3.70s recent_step=2.06s eta=23:56:22 +2026-09-15 17:56:39 | INFO | train rank=0 | step=26710/50000 progress=53.4% loss=0.078387 lr=4.847904e-05 lr_backbone=4.847904e-05 elapsed=27:26:17 avg_step=3.70s recent_step=1.92s eta=23:55:29 +2026-09-15 17:56:57 | INFO | train rank=0 | step=26720/50000 progress=53.4% loss=0.094221 lr=4.844599e-05 lr_backbone=4.844599e-05 elapsed=27:26:35 avg_step=3.70s recent_step=1.79s eta=23:54:36 +2026-09-15 17:57:20 | INFO | train rank=0 | step=26730/50000 progress=53.5% loss=0.085116 lr=4.841294e-05 lr_backbone=4.841294e-05 elapsed=27:26:58 avg_step=3.70s recent_step=2.28s eta=23:53:47 +2026-09-15 17:57:58 | INFO | train rank=0 | step=26740/50000 progress=53.5% loss=0.116486 lr=4.837988e-05 lr_backbone=4.837988e-05 elapsed=27:27:36 avg_step=3.70s recent_step=3.80s eta=23:53:10 +2026-09-15 17:58:33 | INFO | train rank=0 | step=26750/50000 progress=53.5% loss=0.107963 lr=4.834683e-05 lr_backbone=4.834683e-05 elapsed=27:28:11 avg_step=3.70s recent_step=3.52s eta=23:52:32 +2026-09-15 17:59:11 | INFO | train rank=0 | step=26760/50000 progress=53.5% loss=0.089362 lr=4.831378e-05 lr_backbone=4.831378e-05 elapsed=27:28:49 avg_step=3.70s recent_step=3.79s eta=23:51:56 +2026-09-15 17:59:58 | INFO | train rank=0 | step=26770/50000 progress=53.5% loss=0.075157 lr=4.828073e-05 lr_backbone=4.828073e-05 elapsed=27:29:36 avg_step=3.70s recent_step=4.74s eta=23:51:28 +2026-09-15 18:00:49 | INFO | train rank=0 | step=26780/50000 progress=53.6% loss=0.089219 lr=4.824768e-05 lr_backbone=4.824768e-05 elapsed=27:30:27 avg_step=3.70s recent_step=5.09s eta=23:51:03 +2026-09-15 18:01:40 | INFO | train rank=0 | step=26790/50000 progress=53.6% loss=0.116023 lr=4.821463e-05 lr_backbone=4.821463e-05 elapsed=27:31:18 avg_step=3.70s recent_step=5.07s eta=23:50:38 +2026-09-15 18:02:46 | INFO | train rank=0 | step=26800/50000 progress=53.6% loss=0.117997 lr=4.818158e-05 lr_backbone=4.818158e-05 elapsed=27:32:24 avg_step=3.70s recent_step=6.62s eta=23:50:26 +2026-09-15 18:04:00 | INFO | train rank=0 | step=26810/50000 progress=53.6% loss=0.099501 lr=4.814854e-05 lr_backbone=4.814854e-05 elapsed=27:33:38 avg_step=3.70s recent_step=7.35s eta=23:50:21 +2026-09-15 18:05:07 | INFO | train rank=0 | step=26820/50000 progress=53.6% loss=0.085145 lr=4.811549e-05 lr_backbone=4.811549e-05 elapsed=27:34:45 avg_step=3.70s recent_step=6.68s eta=23:50:10 +2026-09-15 18:06:10 | INFO | train rank=0 | step=26830/50000 progress=53.7% loss=0.079418 lr=4.808245e-05 lr_backbone=4.808245e-05 elapsed=27:35:48 avg_step=3.70s recent_step=6.35s eta=23:49:55 +2026-09-15 18:07:18 | INFO | train rank=0 | step=26840/50000 progress=53.7% loss=0.072636 lr=4.804940e-05 lr_backbone=4.804940e-05 elapsed=27:36:56 avg_step=3.70s recent_step=6.76s eta=23:49:45 +2026-09-15 18:08:24 | INFO | train rank=0 | step=26850/50000 progress=53.7% loss=0.083720 lr=4.801636e-05 lr_backbone=4.801636e-05 elapsed=27:38:02 avg_step=3.71s recent_step=6.67s eta=23:49:33 +2026-09-15 18:09:26 | INFO | train rank=0 | step=26860/50000 progress=53.7% loss=0.088330 lr=4.798331e-05 lr_backbone=4.798331e-05 elapsed=27:39:04 avg_step=3.71s recent_step=6.18s eta=23:49:18 +2026-09-15 18:10:37 | INFO | train rank=0 | step=26870/50000 progress=53.7% loss=0.092050 lr=4.795027e-05 lr_backbone=4.795027e-05 elapsed=27:40:15 avg_step=3.71s recent_step=7.06s eta=23:49:09 +2026-09-15 18:11:43 | INFO | train rank=0 | step=26880/50000 progress=53.8% loss=0.089697 lr=4.791723e-05 lr_backbone=4.791723e-05 elapsed=27:41:21 avg_step=3.71s recent_step=6.63s eta=23:48:57 +2026-09-15 18:12:58 | INFO | train rank=0 | step=26890/50000 progress=53.8% loss=0.076306 lr=4.788419e-05 lr_backbone=4.788419e-05 elapsed=27:42:36 avg_step=3.71s recent_step=7.48s eta=23:48:53 +2026-09-15 18:14:09 | INFO | train rank=0 | step=26900/50000 progress=53.8% loss=0.067748 lr=4.785115e-05 lr_backbone=4.785115e-05 elapsed=27:43:47 avg_step=3.71s recent_step=7.11s eta=23:48:45 +2026-09-15 18:15:22 | INFO | train rank=0 | step=26910/50000 progress=53.8% loss=0.083547 lr=4.781811e-05 lr_backbone=4.781811e-05 elapsed=27:45:00 avg_step=3.71s recent_step=7.27s eta=23:48:38 +2026-09-15 18:16:31 | INFO | train rank=0 | step=26920/50000 progress=53.8% loss=0.090273 lr=4.778508e-05 lr_backbone=4.778508e-05 elapsed=27:46:09 avg_step=3.71s recent_step=6.90s eta=23:48:29 +2026-09-15 18:17:46 | INFO | train rank=0 | step=26930/50000 progress=53.9% loss=0.083155 lr=4.775204e-05 lr_backbone=4.775204e-05 elapsed=27:47:24 avg_step=3.71s recent_step=7.54s eta=23:48:24 +2026-09-15 18:18:40 | INFO | train rank=0 | step=26940/50000 progress=53.9% loss=0.093284 lr=4.771900e-05 lr_backbone=4.771900e-05 elapsed=27:48:18 avg_step=3.72s recent_step=5.39s eta=23:48:01 +2026-09-15 18:19:45 | INFO | train rank=0 | step=26950/50000 progress=53.9% loss=0.093128 lr=4.768597e-05 lr_backbone=4.768597e-05 elapsed=27:49:23 avg_step=3.72s recent_step=6.55s eta=23:47:48 +2026-09-15 18:20:52 | INFO | train rank=0 | step=26960/50000 progress=53.9% loss=0.092040 lr=4.765294e-05 lr_backbone=4.765294e-05 elapsed=27:50:30 avg_step=3.72s recent_step=6.67s eta=23:47:37 +2026-09-15 18:21:59 | INFO | train rank=0 | step=26970/50000 progress=53.9% loss=0.082947 lr=4.761990e-05 lr_backbone=4.761990e-05 elapsed=27:51:37 avg_step=3.72s recent_step=6.67s eta=23:47:25 +2026-09-15 18:23:05 | INFO | train rank=0 | step=26980/50000 progress=54.0% loss=0.085729 lr=4.758687e-05 lr_backbone=4.758687e-05 elapsed=27:52:43 avg_step=3.72s recent_step=6.65s eta=23:47:12 +2026-09-15 18:24:16 | INFO | train rank=0 | step=26990/50000 progress=54.0% loss=0.085129 lr=4.755384e-05 lr_backbone=4.755384e-05 elapsed=27:53:54 avg_step=3.72s recent_step=7.04s eta=23:47:03 +2026-09-15 18:25:29 | INFO | train rank=0 | step=27000/50000 progress=54.0% loss=0.076947 lr=4.752081e-05 lr_backbone=4.752081e-05 elapsed=27:55:07 avg_step=3.72s recent_step=7.31s eta=23:46:57 +2026-09-15 18:26:38 | INFO | train rank=0 | step=27010/50000 progress=54.0% loss=0.113080 lr=4.748778e-05 lr_backbone=4.748778e-05 elapsed=27:56:16 avg_step=3.72s recent_step=6.89s eta=23:46:47 +2026-09-15 18:27:54 | INFO | train rank=0 | step=27020/50000 progress=54.0% loss=0.091408 lr=4.745476e-05 lr_backbone=4.745476e-05 elapsed=27:57:32 avg_step=3.73s recent_step=7.66s eta=23:46:43 +2026-09-15 18:29:08 | INFO | train rank=0 | step=27030/50000 progress=54.1% loss=0.088378 lr=4.742173e-05 lr_backbone=4.742173e-05 elapsed=27:58:46 avg_step=3.73s recent_step=7.38s eta=23:46:37 +2026-09-15 18:30:13 | INFO | train rank=0 | step=27040/50000 progress=54.1% loss=0.089683 lr=4.738871e-05 lr_backbone=4.738871e-05 elapsed=27:59:51 avg_step=3.73s recent_step=6.50s eta=23:46:23 +2026-09-15 18:31:15 | INFO | train rank=0 | step=27050/50000 progress=54.1% loss=0.093482 lr=4.735568e-05 lr_backbone=4.735568e-05 elapsed=28:00:53 avg_step=3.73s recent_step=6.15s eta=23:46:06 +2026-09-15 18:32:28 | INFO | train rank=0 | step=27060/50000 progress=54.1% loss=0.083644 lr=4.732266e-05 lr_backbone=4.732266e-05 elapsed=28:02:06 avg_step=3.73s recent_step=7.32s eta=23:45:59 +2026-09-15 18:33:32 | INFO | train rank=0 | step=27070/50000 progress=54.1% loss=0.109290 lr=4.728964e-05 lr_backbone=4.728964e-05 elapsed=28:03:10 avg_step=3.73s recent_step=6.46s eta=23:45:45 +2026-09-15 18:34:38 | INFO | train rank=0 | step=27080/50000 progress=54.2% loss=0.120812 lr=4.725662e-05 lr_backbone=4.725662e-05 elapsed=28:04:16 avg_step=3.73s recent_step=6.53s eta=23:45:32 +2026-09-15 18:35:49 | INFO | train rank=0 | step=27090/50000 progress=54.2% loss=0.090665 lr=4.722360e-05 lr_backbone=4.722360e-05 elapsed=28:05:27 avg_step=3.73s recent_step=7.15s eta=23:45:23 +2026-09-15 18:36:59 | INFO | train rank=0 | step=27100/50000 progress=54.2% loss=0.106551 lr=4.719058e-05 lr_backbone=4.719058e-05 elapsed=28:06:37 avg_step=3.73s recent_step=6.98s eta=23:45:13 +2026-09-15 18:37:59 | INFO | train rank=0 | step=27110/50000 progress=54.2% loss=0.093915 lr=4.715757e-05 lr_backbone=4.715757e-05 elapsed=28:07:37 avg_step=3.74s recent_step=5.98s eta=23:44:55 +2026-09-15 18:38:55 | INFO | train rank=0 | step=27120/50000 progress=54.2% loss=0.109243 lr=4.712455e-05 lr_backbone=4.712455e-05 elapsed=28:08:33 avg_step=3.74s recent_step=5.65s eta=23:44:34 +2026-09-15 18:40:16 | INFO | train rank=0 | step=27130/50000 progress=54.3% loss=0.077633 lr=4.709154e-05 lr_backbone=4.709154e-05 elapsed=28:09:54 avg_step=3.74s recent_step=8.11s eta=23:44:33 +2026-09-15 18:42:10 | INFO | train rank=0 | step=27140/50000 progress=54.3% loss=0.098792 lr=4.705852e-05 lr_backbone=4.705852e-05 elapsed=28:11:48 avg_step=3.74s recent_step=11.34s eta=23:45:00 +2026-09-15 18:44:14 | INFO | train rank=0 | step=27150/50000 progress=54.3% loss=0.069889 lr=4.702551e-05 lr_backbone=4.702551e-05 elapsed=28:13:52 avg_step=3.74s recent_step=12.41s eta=23:45:35 +2026-09-15 18:46:24 | INFO | train rank=0 | step=27160/50000 progress=54.3% loss=0.101381 lr=4.699250e-05 lr_backbone=4.699250e-05 elapsed=28:16:02 avg_step=3.75s recent_step=13.01s eta=23:46:16 +2026-09-15 18:48:14 | INFO | train rank=0 | step=27170/50000 progress=54.3% loss=0.101815 lr=4.695949e-05 lr_backbone=4.695949e-05 elapsed=28:17:52 avg_step=3.75s recent_step=11.01s eta=23:46:39 +2026-09-15 18:50:20 | INFO | train rank=0 | step=27180/50000 progress=54.4% loss=0.082907 lr=4.692649e-05 lr_backbone=4.692649e-05 elapsed=28:19:58 avg_step=3.75s recent_step=12.54s eta=23:47:16 +2026-09-15 18:52:17 | INFO | train rank=0 | step=27190/50000 progress=54.4% loss=0.095566 lr=4.689348e-05 lr_backbone=4.689348e-05 elapsed=28:21:55 avg_step=3.76s recent_step=11.72s eta=23:47:45 +2026-09-15 18:54:20 | INFO | train rank=0 | step=27200/50000 progress=54.4% loss=0.079005 lr=4.686047e-05 lr_backbone=4.686047e-05 elapsed=28:23:58 avg_step=3.76s recent_step=12.30s eta=23:48:19 +2026-09-15 18:56:18 | INFO | train rank=0 | step=27210/50000 progress=54.4% loss=0.091517 lr=4.682747e-05 lr_backbone=4.682747e-05 elapsed=28:25:56 avg_step=3.76s recent_step=11.82s eta=23:48:49 +2026-09-15 18:59:06 | INFO | train rank=0 | step=27220/50000 progress=54.4% loss=0.107875 lr=4.679447e-05 lr_backbone=4.679447e-05 elapsed=28:28:44 avg_step=3.77s recent_step=16.85s eta=23:50:01 +2026-09-15 18:59:53 | INFO | train rank=0 | step=27230/50000 progress=54.5% loss=0.097722 lr=4.676147e-05 lr_backbone=4.676147e-05 elapsed=28:29:31 avg_step=3.77s recent_step=4.70s eta=23:49:31 +2026-09-15 19:00:36 | INFO | train rank=0 | step=27240/50000 progress=54.5% loss=0.109837 lr=4.672847e-05 lr_backbone=4.672847e-05 elapsed=28:30:14 avg_step=3.77s recent_step=4.27s eta=23:48:58 +2026-09-15 19:01:13 | INFO | train rank=0 | step=27250/50000 progress=54.5% loss=0.097221 lr=4.669547e-05 lr_backbone=4.669547e-05 elapsed=28:30:51 avg_step=3.77s recent_step=3.71s eta=23:48:20 +2026-09-15 19:01:41 | INFO | train rank=0 | step=27260/50000 progress=54.5% loss=0.098641 lr=4.666247e-05 lr_backbone=4.666247e-05 elapsed=28:31:19 avg_step=3.77s recent_step=2.75s eta=23:47:33 +2026-09-15 19:02:39 | INFO | train rank=0 | step=27270/50000 progress=54.5% loss=0.082941 lr=4.662948e-05 lr_backbone=4.662948e-05 elapsed=28:32:17 avg_step=3.77s recent_step=5.87s eta=23:47:13 +2026-09-15 19:03:08 | INFO | train rank=0 | step=27280/50000 progress=54.6% loss=0.098247 lr=4.659649e-05 lr_backbone=4.659649e-05 elapsed=28:32:46 avg_step=3.77s recent_step=2.83s eta=23:46:28 +2026-09-15 19:03:50 | INFO | train rank=0 | step=27290/50000 progress=54.6% loss=0.092199 lr=4.656349e-05 lr_backbone=4.656349e-05 elapsed=28:33:28 avg_step=3.77s recent_step=4.29s eta=23:45:54 +2026-09-15 19:04:22 | INFO | train rank=0 | step=27300/50000 progress=54.6% loss=0.098151 lr=4.653050e-05 lr_backbone=4.653050e-05 elapsed=28:34:00 avg_step=3.77s recent_step=3.16s eta=23:45:12 +2026-09-15 19:04:58 | INFO | train rank=0 | step=27310/50000 progress=54.6% loss=0.111608 lr=4.649751e-05 lr_backbone=4.649751e-05 elapsed=28:34:36 avg_step=3.77s recent_step=3.58s eta=23:44:32 +2026-09-15 19:05:32 | INFO | train rank=0 | step=27320/50000 progress=54.6% loss=0.090782 lr=4.646453e-05 lr_backbone=4.646453e-05 elapsed=28:35:10 avg_step=3.77s recent_step=3.37s eta=23:43:51 +2026-09-15 19:06:07 | INFO | train rank=0 | step=27330/50000 progress=54.7% loss=0.109177 lr=4.643154e-05 lr_backbone=4.643154e-05 elapsed=28:35:45 avg_step=3.77s recent_step=3.55s eta=23:43:12 +2026-09-15 19:06:39 | INFO | train rank=0 | step=27340/50000 progress=54.7% loss=0.084434 lr=4.639856e-05 lr_backbone=4.639856e-05 elapsed=28:36:17 avg_step=3.77s recent_step=3.21s eta=23:42:30 +2026-09-15 19:07:40 | INFO | train rank=0 | step=27350/50000 progress=54.7% loss=0.100937 lr=4.636557e-05 lr_backbone=4.636557e-05 elapsed=28:37:18 avg_step=3.77s recent_step=6.10s eta=23:42:11 +2026-09-15 19:08:13 | INFO | train rank=0 | step=27360/50000 progress=54.7% loss=0.086380 lr=4.633259e-05 lr_backbone=4.633259e-05 elapsed=28:37:51 avg_step=3.77s recent_step=3.33s eta=23:41:30 +2026-09-15 19:08:43 | INFO | train rank=0 | step=27370/50000 progress=54.7% loss=0.085463 lr=4.629961e-05 lr_backbone=4.629961e-05 elapsed=28:38:21 avg_step=3.77s recent_step=2.93s eta=23:40:45 +2026-09-15 19:09:18 | INFO | train rank=0 | step=27380/50000 progress=54.8% loss=0.112523 lr=4.626664e-05 lr_backbone=4.626664e-05 elapsed=28:38:56 avg_step=3.77s recent_step=3.52s eta=23:40:06 +2026-09-15 19:10:16 | INFO | train rank=0 | step=27390/50000 progress=54.8% loss=0.089624 lr=4.623366e-05 lr_backbone=4.623366e-05 elapsed=28:39:54 avg_step=3.77s recent_step=5.84s eta=23:39:45 +2026-09-15 19:10:50 | INFO | train rank=0 | step=27400/50000 progress=54.8% loss=0.092148 lr=4.620069e-05 lr_backbone=4.620069e-05 elapsed=28:40:28 avg_step=3.77s recent_step=3.32s eta=23:39:04 +2026-09-15 19:11:25 | INFO | train rank=0 | step=27410/50000 progress=54.8% loss=0.080488 lr=4.616771e-05 lr_backbone=4.616771e-05 elapsed=28:41:03 avg_step=3.77s recent_step=3.50s eta=23:38:24 +2026-09-15 19:11:57 | INFO | train rank=0 | step=27420/50000 progress=54.8% loss=0.102566 lr=4.613474e-05 lr_backbone=4.613474e-05 elapsed=28:41:35 avg_step=3.77s recent_step=3.26s eta=23:37:42 +2026-09-15 19:12:37 | INFO | train rank=0 | step=27430/50000 progress=54.9% loss=0.093695 lr=4.610177e-05 lr_backbone=4.610177e-05 elapsed=28:42:15 avg_step=3.77s recent_step=3.99s eta=23:37:06 +2026-09-15 19:13:08 | INFO | train rank=0 | step=27440/50000 progress=54.9% loss=0.096447 lr=4.606880e-05 lr_backbone=4.606880e-05 elapsed=28:42:46 avg_step=3.77s recent_step=3.11s eta=23:36:23 +2026-09-15 19:13:36 | INFO | train rank=0 | step=27450/50000 progress=54.9% loss=0.110615 lr=4.603584e-05 lr_backbone=4.603584e-05 elapsed=28:43:14 avg_step=3.77s recent_step=2.77s eta=23:35:37 +2026-09-15 19:14:17 | INFO | train rank=0 | step=27460/50000 progress=54.9% loss=0.085523 lr=4.600287e-05 lr_backbone=4.600287e-05 elapsed=28:43:55 avg_step=3.77s recent_step=4.14s eta=23:35:03 +2026-09-15 19:15:10 | INFO | train rank=0 | step=27470/50000 progress=54.9% loss=0.083368 lr=4.596991e-05 lr_backbone=4.596991e-05 elapsed=28:44:48 avg_step=3.77s recent_step=5.28s eta=23:34:37 +2026-09-15 19:15:43 | INFO | train rank=0 | step=27480/50000 progress=55.0% loss=0.075796 lr=4.593695e-05 lr_backbone=4.593695e-05 elapsed=28:45:21 avg_step=3.77s recent_step=3.34s eta=23:33:56 +2026-09-15 19:16:20 | INFO | train rank=0 | step=27490/50000 progress=55.0% loss=0.086835 lr=4.590399e-05 lr_backbone=4.590399e-05 elapsed=28:45:58 avg_step=3.77s recent_step=3.64s eta=23:33:18 +2026-09-15 19:16:51 | INFO | train rank=0 | step=27500/50000 progress=55.0% loss=0.084384 lr=4.587103e-05 lr_backbone=4.587103e-05 elapsed=28:46:29 avg_step=3.77s recent_step=3.15s eta=23:32:35 +2026-09-15 19:17:38 | INFO | train rank=0 | step=27510/50000 progress=55.0% loss=0.091469 lr=4.583808e-05 lr_backbone=4.583808e-05 elapsed=28:47:16 avg_step=3.77s recent_step=4.70s eta=23:32:05 +2026-09-15 19:18:10 | INFO | train rank=0 | step=27520/50000 progress=55.0% loss=0.078690 lr=4.580512e-05 lr_backbone=4.580512e-05 elapsed=28:47:48 avg_step=3.77s recent_step=3.20s eta=23:31:23 +2026-09-15 19:18:38 | INFO | train rank=0 | step=27530/50000 progress=55.1% loss=0.083895 lr=4.577217e-05 lr_backbone=4.577217e-05 elapsed=28:48:16 avg_step=3.77s recent_step=2.77s eta=23:30:37 +2026-09-15 19:19:11 | INFO | train rank=0 | step=27540/50000 progress=55.1% loss=0.103541 lr=4.573922e-05 lr_backbone=4.573922e-05 elapsed=28:48:49 avg_step=3.77s recent_step=3.31s eta=23:29:55 +2026-09-15 19:20:02 | INFO | train rank=0 | step=27550/50000 progress=55.1% loss=0.088829 lr=4.570627e-05 lr_backbone=4.570627e-05 elapsed=28:49:40 avg_step=3.77s recent_step=5.09s eta=23:29:28 +2026-09-15 19:20:41 | INFO | train rank=0 | step=27560/50000 progress=55.1% loss=0.090300 lr=4.567333e-05 lr_backbone=4.567333e-05 elapsed=28:50:19 avg_step=3.77s recent_step=3.88s eta=23:28:52 +2026-09-15 19:21:13 | INFO | train rank=0 | step=27570/50000 progress=55.1% loss=0.078477 lr=4.564038e-05 lr_backbone=4.564038e-05 elapsed=28:50:51 avg_step=3.77s recent_step=3.24s eta=23:28:10 +2026-09-15 19:21:44 | INFO | train rank=0 | step=27580/50000 progress=55.2% loss=0.086047 lr=4.560744e-05 lr_backbone=4.560744e-05 elapsed=28:51:22 avg_step=3.77s recent_step=3.10s eta=23:27:27 +2026-09-15 19:22:29 | INFO | train rank=0 | step=27590/50000 progress=55.2% loss=0.088432 lr=4.557450e-05 lr_backbone=4.557450e-05 elapsed=28:52:07 avg_step=3.77s recent_step=4.51s eta=23:26:55 +2026-09-15 19:23:05 | INFO | train rank=0 | step=27600/50000 progress=55.2% loss=0.089106 lr=4.554156e-05 lr_backbone=4.554156e-05 elapsed=28:52:43 avg_step=3.77s recent_step=3.60s eta=23:26:16 +2026-09-15 19:23:38 | INFO | train rank=0 | step=27610/50000 progress=55.2% loss=0.093848 lr=4.550862e-05 lr_backbone=4.550862e-05 elapsed=28:53:16 avg_step=3.77s recent_step=3.22s eta=23:25:34 +2026-09-15 19:24:09 | INFO | train rank=0 | step=27620/50000 progress=55.2% loss=0.101565 lr=4.547569e-05 lr_backbone=4.547569e-05 elapsed=28:53:47 avg_step=3.77s recent_step=3.15s eta=23:24:51 +2026-09-15 19:24:58 | INFO | train rank=0 | step=27630/50000 progress=55.3% loss=0.097173 lr=4.544276e-05 lr_backbone=4.544276e-05 elapsed=28:54:36 avg_step=3.77s recent_step=4.94s eta=23:24:23 +2026-09-15 19:25:34 | INFO | train rank=0 | step=27640/50000 progress=55.3% loss=0.102018 lr=4.540983e-05 lr_backbone=4.540983e-05 elapsed=28:55:12 avg_step=3.77s recent_step=3.51s eta=23:23:43 +2026-09-15 19:26:04 | INFO | train rank=0 | step=27650/50000 progress=55.3% loss=0.092072 lr=4.537690e-05 lr_backbone=4.537690e-05 elapsed=28:55:42 avg_step=3.77s recent_step=3.04s eta=23:23:00 +2026-09-15 19:26:40 | INFO | train rank=0 | step=27660/50000 progress=55.3% loss=0.106620 lr=4.534397e-05 lr_backbone=4.534397e-05 elapsed=28:56:18 avg_step=3.77s recent_step=3.59s eta=23:22:21 +2026-09-15 19:27:31 | INFO | train rank=0 | step=27670/50000 progress=55.3% loss=0.088822 lr=4.531105e-05 lr_backbone=4.531105e-05 elapsed=28:57:09 avg_step=3.77s recent_step=5.09s eta=23:21:54 +2026-09-15 19:28:12 | INFO | train rank=0 | step=27680/50000 progress=55.4% loss=0.100356 lr=4.527812e-05 lr_backbone=4.527812e-05 elapsed=28:57:50 avg_step=3.77s recent_step=4.14s eta=23:21:19 +2026-09-15 19:28:50 | INFO | train rank=0 | step=27690/50000 progress=55.4% loss=0.073197 lr=4.524520e-05 lr_backbone=4.524520e-05 elapsed=28:58:28 avg_step=3.77s recent_step=3.77s eta=23:20:41 +2026-09-15 19:29:25 | INFO | train rank=0 | step=27700/50000 progress=55.4% loss=0.085836 lr=4.521228e-05 lr_backbone=4.521228e-05 elapsed=28:59:03 avg_step=3.77s recent_step=3.51s eta=23:20:02 +2026-09-15 19:30:06 | INFO | train rank=0 | step=27710/50000 progress=55.4% loss=0.085665 lr=4.517937e-05 lr_backbone=4.517937e-05 elapsed=28:59:44 avg_step=3.77s recent_step=4.10s eta=23:19:27 +2026-09-15 19:30:40 | INFO | train rank=0 | step=27720/50000 progress=55.4% loss=0.091478 lr=4.514645e-05 lr_backbone=4.514645e-05 elapsed=29:00:18 avg_step=3.77s recent_step=3.40s eta=23:18:46 +2026-09-15 19:31:15 | INFO | train rank=0 | step=27730/50000 progress=55.5% loss=0.097821 lr=4.511354e-05 lr_backbone=4.511354e-05 elapsed=29:00:53 avg_step=3.77s recent_step=3.55s eta=23:18:07 +2026-09-15 19:31:58 | INFO | train rank=0 | step=27740/50000 progress=55.5% loss=0.115079 lr=4.508063e-05 lr_backbone=4.508063e-05 elapsed=29:01:36 avg_step=3.77s recent_step=4.25s eta=23:17:33 +2026-09-15 19:32:51 | INFO | train rank=0 | step=27750/50000 progress=55.5% loss=0.100434 lr=4.504772e-05 lr_backbone=4.504772e-05 elapsed=29:02:29 avg_step=3.77s recent_step=5.29s eta=23:17:07 +2026-09-15 19:33:26 | INFO | train rank=0 | step=27760/50000 progress=55.5% loss=0.075655 lr=4.501482e-05 lr_backbone=4.501482e-05 elapsed=29:03:04 avg_step=3.77s recent_step=3.47s eta=23:16:27 +2026-09-15 19:33:58 | INFO | train rank=0 | step=27770/50000 progress=55.5% loss=0.111807 lr=4.498191e-05 lr_backbone=4.498191e-05 elapsed=29:03:36 avg_step=3.77s recent_step=3.26s eta=23:15:46 +2026-09-15 19:34:24 | INFO | train rank=0 | step=27780/50000 progress=55.6% loss=0.089154 lr=4.494901e-05 lr_backbone=4.494901e-05 elapsed=29:04:02 avg_step=3.77s recent_step=2.61s eta=23:14:59 +2026-09-15 19:35:23 | INFO | train rank=0 | step=27790/50000 progress=55.6% loss=0.127337 lr=4.491611e-05 lr_backbone=4.491611e-05 elapsed=29:05:01 avg_step=3.77s recent_step=5.92s eta=23:14:38 +2026-09-15 19:35:56 | INFO | train rank=0 | step=27800/50000 progress=55.6% loss=0.114160 lr=4.488322e-05 lr_backbone=4.488322e-05 elapsed=29:05:34 avg_step=3.77s recent_step=3.25s eta=23:13:56 +2026-09-15 19:36:25 | INFO | train rank=0 | step=27810/50000 progress=55.6% loss=0.086019 lr=4.485032e-05 lr_backbone=4.485032e-05 elapsed=29:06:03 avg_step=3.77s recent_step=2.91s eta=23:13:12 +2026-09-15 19:37:05 | INFO | train rank=0 | step=27820/50000 progress=55.6% loss=0.082575 lr=4.481743e-05 lr_backbone=4.481743e-05 elapsed=29:06:43 avg_step=3.77s recent_step=3.98s eta=23:12:36 +2026-09-15 19:37:51 | INFO | train rank=0 | step=27830/50000 progress=55.7% loss=0.075692 lr=4.478454e-05 lr_backbone=4.478454e-05 elapsed=29:07:29 avg_step=3.77s recent_step=4.60s eta=23:12:05 +2026-09-15 19:38:23 | INFO | train rank=0 | step=27840/50000 progress=55.7% loss=0.093589 lr=4.475165e-05 lr_backbone=4.475165e-05 elapsed=29:08:01 avg_step=3.77s recent_step=3.18s eta=23:11:22 +2026-09-15 19:38:49 | INFO | train rank=0 | step=27850/50000 progress=55.7% loss=0.077869 lr=4.471877e-05 lr_backbone=4.471877e-05 elapsed=29:08:27 avg_step=3.77s recent_step=2.63s eta=23:10:36 +2026-09-15 19:39:26 | INFO | train rank=0 | step=27860/50000 progress=55.7% loss=0.090815 lr=4.468588e-05 lr_backbone=4.468588e-05 elapsed=29:09:04 avg_step=3.77s recent_step=3.67s eta=23:09:57 +2026-09-15 19:40:20 | INFO | train rank=0 | step=27870/50000 progress=55.7% loss=0.111185 lr=4.465300e-05 lr_backbone=4.465300e-05 elapsed=29:09:58 avg_step=3.77s recent_step=5.44s eta=23:09:33 +2026-09-15 19:40:51 | INFO | train rank=0 | step=27880/50000 progress=55.8% loss=0.102324 lr=4.462012e-05 lr_backbone=4.462012e-05 elapsed=29:10:29 avg_step=3.77s recent_step=3.12s eta=23:08:50 +2026-09-15 19:41:33 | INFO | train rank=0 | step=27890/50000 progress=55.8% loss=0.098209 lr=4.458725e-05 lr_backbone=4.458725e-05 elapsed=29:11:11 avg_step=3.77s recent_step=4.22s eta=23:08:16 +2026-09-15 19:42:14 | INFO | train rank=0 | step=27900/50000 progress=55.8% loss=0.111843 lr=4.455437e-05 lr_backbone=4.455437e-05 elapsed=29:11:52 avg_step=3.77s recent_step=4.04s eta=23:07:40 +2026-09-15 19:42:55 | INFO | train rank=0 | step=27910/50000 progress=55.8% loss=0.085220 lr=4.452150e-05 lr_backbone=4.452150e-05 elapsed=29:12:33 avg_step=3.77s recent_step=4.12s eta=23:07:06 +2026-09-15 19:43:25 | INFO | train rank=0 | step=27920/50000 progress=55.8% loss=0.092055 lr=4.448863e-05 lr_backbone=4.448863e-05 elapsed=29:13:03 avg_step=3.77s recent_step=3.05s eta=23:06:22 +2026-09-15 19:43:58 | INFO | train rank=0 | step=27930/50000 progress=55.9% loss=0.098251 lr=4.445577e-05 lr_backbone=4.445577e-05 elapsed=29:13:36 avg_step=3.77s recent_step=3.27s eta=23:05:41 +2026-09-15 19:44:29 | INFO | train rank=0 | step=27940/50000 progress=55.9% loss=0.094877 lr=4.442290e-05 lr_backbone=4.442290e-05 elapsed=29:14:07 avg_step=3.77s recent_step=3.11s eta=23:04:58 +2026-09-15 19:45:16 | INFO | train rank=0 | step=27950/50000 progress=55.9% loss=0.109636 lr=4.439004e-05 lr_backbone=4.439004e-05 elapsed=29:14:54 avg_step=3.77s recent_step=4.66s eta=23:04:27 +2026-09-15 19:45:44 | INFO | train rank=0 | step=27960/50000 progress=55.9% loss=0.088860 lr=4.435718e-05 lr_backbone=4.435718e-05 elapsed=29:15:22 avg_step=3.77s recent_step=2.79s eta=23:03:42 +2026-09-15 19:46:14 | INFO | train rank=0 | step=27970/50000 progress=55.9% loss=0.101048 lr=4.432432e-05 lr_backbone=4.432432e-05 elapsed=29:15:52 avg_step=3.77s recent_step=3.04s eta=23:02:58 +2026-09-15 19:46:46 | INFO | train rank=0 | step=27980/50000 progress=56.0% loss=0.085540 lr=4.429147e-05 lr_backbone=4.429147e-05 elapsed=29:16:24 avg_step=3.77s recent_step=3.22s eta=23:02:16 +2026-09-15 19:47:46 | INFO | train rank=0 | step=27990/50000 progress=56.0% loss=0.129892 lr=4.425862e-05 lr_backbone=4.425862e-05 elapsed=29:17:24 avg_step=3.77s recent_step=5.98s eta=23:01:56 +2026-09-15 19:48:15 | INFO | train rank=0 | step=28000/50000 progress=56.0% loss=0.095233 lr=4.422577e-05 lr_backbone=4.422577e-05 elapsed=29:17:53 avg_step=3.77s recent_step=2.86s eta=23:01:11 +2026-09-15 19:48:48 | INFO | train rank=0 | step=28010/50000 progress=56.0% loss=0.086328 lr=4.419292e-05 lr_backbone=4.419292e-05 elapsed=29:18:26 avg_step=3.77s recent_step=3.32s eta=23:00:30 +2026-09-15 19:49:15 | INFO | train rank=0 | step=28020/50000 progress=56.0% loss=0.108014 lr=4.416008e-05 lr_backbone=4.416008e-05 elapsed=29:18:53 avg_step=3.77s recent_step=2.69s eta=22:59:44 +2026-09-15 19:50:10 | INFO | train rank=0 | step=28030/50000 progress=56.1% loss=0.101805 lr=4.412724e-05 lr_backbone=4.412724e-05 elapsed=29:19:48 avg_step=3.77s recent_step=5.50s eta=22:59:20 +2026-09-15 19:50:45 | INFO | train rank=0 | step=28040/50000 progress=56.1% loss=0.093684 lr=4.409440e-05 lr_backbone=4.409440e-05 elapsed=29:20:23 avg_step=3.77s recent_step=3.52s eta=22:58:40 +2026-09-15 19:51:22 | INFO | train rank=0 | step=28050/50000 progress=56.1% loss=0.075513 lr=4.406156e-05 lr_backbone=4.406156e-05 elapsed=29:21:00 avg_step=3.77s recent_step=3.74s eta=22:58:03 +2026-09-15 19:51:48 | INFO | train rank=0 | step=28060/50000 progress=56.1% loss=0.100393 lr=4.402873e-05 lr_backbone=4.402873e-05 elapsed=29:21:26 avg_step=3.77s recent_step=2.55s eta=22:57:15 +2026-09-15 19:52:45 | INFO | train rank=0 | step=28070/50000 progress=56.1% loss=0.092926 lr=4.399589e-05 lr_backbone=4.399589e-05 elapsed=29:22:23 avg_step=3.77s recent_step=5.72s eta=22:56:53 +2026-09-15 19:53:17 | INFO | train rank=0 | step=28080/50000 progress=56.2% loss=0.107488 lr=4.396307e-05 lr_backbone=4.396307e-05 elapsed=29:22:55 avg_step=3.77s recent_step=3.16s eta=22:56:10 +2026-09-15 19:53:52 | INFO | train rank=0 | step=28090/50000 progress=56.2% loss=0.095025 lr=4.393024e-05 lr_backbone=4.393024e-05 elapsed=29:23:30 avg_step=3.77s recent_step=3.50s eta=22:55:31 +2026-09-15 19:54:23 | INFO | train rank=0 | step=28100/50000 progress=56.2% loss=0.088348 lr=4.389742e-05 lr_backbone=4.389742e-05 elapsed=29:24:01 avg_step=3.77s recent_step=3.13s eta=22:54:48 +2026-09-15 19:55:16 | INFO | train rank=0 | step=28110/50000 progress=56.2% loss=0.106072 lr=4.386459e-05 lr_backbone=4.386459e-05 elapsed=29:24:54 avg_step=3.77s recent_step=5.31s eta=22:54:22 +2026-09-15 19:55:52 | INFO | train rank=0 | step=28120/50000 progress=56.2% loss=0.084467 lr=4.383178e-05 lr_backbone=4.383178e-05 elapsed=29:25:30 avg_step=3.77s recent_step=3.55s eta=22:53:43 +2026-09-15 19:56:25 | INFO | train rank=0 | step=28130/50000 progress=56.3% loss=0.091266 lr=4.379896e-05 lr_backbone=4.379896e-05 elapsed=29:26:03 avg_step=3.77s recent_step=3.30s eta=22:53:02 +2026-09-15 19:56:58 | INFO | train rank=0 | step=28140/50000 progress=56.3% loss=0.091981 lr=4.376615e-05 lr_backbone=4.376615e-05 elapsed=29:26:36 avg_step=3.77s recent_step=3.33s eta=22:52:21 +2026-09-15 19:57:43 | INFO | train rank=0 | step=28150/50000 progress=56.3% loss=0.108479 lr=4.373334e-05 lr_backbone=4.373334e-05 elapsed=29:27:21 avg_step=3.77s recent_step=4.55s eta=22:51:49 +2026-09-15 19:58:15 | INFO | train rank=0 | step=28160/50000 progress=56.3% loss=0.096332 lr=4.370053e-05 lr_backbone=4.370053e-05 elapsed=29:27:53 avg_step=3.77s recent_step=3.13s eta=22:51:06 +2026-09-15 19:58:46 | INFO | train rank=0 | step=28170/50000 progress=56.3% loss=0.096829 lr=4.366773e-05 lr_backbone=4.366773e-05 elapsed=29:28:24 avg_step=3.77s recent_step=3.18s eta=22:50:24 +2026-09-15 19:59:19 | INFO | train rank=0 | step=28180/50000 progress=56.4% loss=0.085717 lr=4.363492e-05 lr_backbone=4.363492e-05 elapsed=29:28:57 avg_step=3.77s recent_step=3.25s eta=22:49:43 +2026-09-15 20:00:01 | INFO | train rank=0 | step=28190/50000 progress=56.4% loss=0.083026 lr=4.360213e-05 lr_backbone=4.360213e-05 elapsed=29:29:39 avg_step=3.77s recent_step=4.21s eta=22:49:08 +2026-09-15 20:00:34 | INFO | train rank=0 | step=28200/50000 progress=56.4% loss=0.100390 lr=4.356933e-05 lr_backbone=4.356933e-05 elapsed=29:30:12 avg_step=3.77s recent_step=3.31s eta=22:48:27 +2026-09-15 20:01:11 | INFO | train rank=0 | step=28210/50000 progress=56.4% loss=0.103064 lr=4.353654e-05 lr_backbone=4.353654e-05 elapsed=29:30:49 avg_step=3.77s recent_step=3.67s eta=22:47:49 +2026-09-15 20:01:49 | INFO | train rank=0 | step=28220/50000 progress=56.4% loss=0.071215 lr=4.350375e-05 lr_backbone=4.350375e-05 elapsed=29:31:27 avg_step=3.77s recent_step=3.82s eta=22:47:12 +2026-09-15 20:02:31 | INFO | train rank=0 | step=28230/50000 progress=56.5% loss=0.093920 lr=4.347096e-05 lr_backbone=4.347096e-05 elapsed=29:32:09 avg_step=3.77s recent_step=4.15s eta=22:46:37 +2026-09-15 20:03:05 | INFO | train rank=0 | step=28240/50000 progress=56.5% loss=0.128187 lr=4.343817e-05 lr_backbone=4.343817e-05 elapsed=29:32:44 avg_step=3.77s recent_step=3.49s eta=22:45:57 +2026-09-15 20:03:34 | INFO | train rank=0 | step=28250/50000 progress=56.5% loss=0.080586 lr=4.340539e-05 lr_backbone=4.340539e-05 elapsed=29:33:12 avg_step=3.77s recent_step=2.86s eta=22:45:12 +2026-09-15 20:04:09 | INFO | train rank=0 | step=28260/50000 progress=56.5% loss=0.080422 lr=4.337261e-05 lr_backbone=4.337261e-05 elapsed=29:33:47 avg_step=3.77s recent_step=3.50s eta=22:44:33 +2026-09-15 20:05:02 | INFO | train rank=0 | step=28270/50000 progress=56.5% loss=0.089605 lr=4.333984e-05 lr_backbone=4.333984e-05 elapsed=29:34:40 avg_step=3.77s recent_step=5.26s eta=22:44:06 +2026-09-15 20:05:35 | INFO | train rank=0 | step=28280/50000 progress=56.6% loss=0.083825 lr=4.330706e-05 lr_backbone=4.330706e-05 elapsed=29:35:13 avg_step=3.77s recent_step=3.30s eta=22:43:25 +2026-09-15 20:06:10 | INFO | train rank=0 | step=28290/50000 progress=56.6% loss=0.082463 lr=4.327429e-05 lr_backbone=4.327429e-05 elapsed=29:35:48 avg_step=3.77s recent_step=3.50s eta=22:42:45 +2026-09-15 20:06:42 | INFO | train rank=0 | step=28300/50000 progress=56.6% loss=0.089265 lr=4.324153e-05 lr_backbone=4.324153e-05 elapsed=29:36:20 avg_step=3.77s recent_step=3.20s eta=22:42:03 +2026-09-15 20:07:39 | INFO | train rank=0 | step=28310/50000 progress=56.6% loss=0.080045 lr=4.320876e-05 lr_backbone=4.320876e-05 elapsed=29:37:17 avg_step=3.77s recent_step=5.77s eta=22:41:41 +2026-09-15 20:08:13 | INFO | train rank=0 | step=28320/50000 progress=56.6% loss=0.090198 lr=4.317600e-05 lr_backbone=4.317600e-05 elapsed=29:37:51 avg_step=3.77s recent_step=3.34s eta=22:41:00 +2026-09-15 20:08:50 | INFO | train rank=0 | step=28330/50000 progress=56.7% loss=0.093792 lr=4.314324e-05 lr_backbone=4.314324e-05 elapsed=29:38:28 avg_step=3.77s recent_step=3.77s eta=22:40:23 +2026-09-15 20:09:20 | INFO | train rank=0 | step=28340/50000 progress=56.7% loss=0.087224 lr=4.311049e-05 lr_backbone=4.311049e-05 elapsed=29:38:58 avg_step=3.77s recent_step=2.97s eta=22:39:39 +2026-09-15 20:10:03 | INFO | train rank=0 | step=28350/50000 progress=56.7% loss=0.083550 lr=4.307773e-05 lr_backbone=4.307773e-05 elapsed=29:39:41 avg_step=3.77s recent_step=4.28s eta=22:39:05 +2026-09-15 20:10:26 | INFO | train rank=0 | step=28360/50000 progress=56.7% loss=0.079796 lr=4.304498e-05 lr_backbone=4.304498e-05 elapsed=29:40:04 avg_step=3.77s recent_step=2.30s eta=22:38:16 +2026-09-15 20:10:58 | INFO | train rank=0 | step=28370/50000 progress=56.7% loss=0.094203 lr=4.301224e-05 lr_backbone=4.301224e-05 elapsed=29:40:36 avg_step=3.77s recent_step=3.20s eta=22:37:34 +2026-09-15 20:11:32 | INFO | train rank=0 | step=28380/50000 progress=56.8% loss=0.099181 lr=4.297949e-05 lr_backbone=4.297949e-05 elapsed=29:41:10 avg_step=3.77s recent_step=3.43s eta=22:36:54 +2026-09-15 20:12:18 | INFO | train rank=0 | step=28390/50000 progress=56.8% loss=0.117136 lr=4.294675e-05 lr_backbone=4.294675e-05 elapsed=29:41:57 avg_step=3.77s recent_step=4.63s eta=22:36:23 +2026-09-15 20:12:50 | INFO | train rank=0 | step=28400/50000 progress=56.8% loss=0.094830 lr=4.291402e-05 lr_backbone=4.291402e-05 elapsed=29:42:28 avg_step=3.77s recent_step=3.16s eta=22:35:41 +2026-09-15 20:13:23 | INFO | train rank=0 | step=28410/50000 progress=56.8% loss=0.102288 lr=4.288128e-05 lr_backbone=4.288128e-05 elapsed=29:43:01 avg_step=3.77s recent_step=3.27s eta=22:34:59 +2026-09-15 20:14:01 | INFO | train rank=0 | step=28420/50000 progress=56.8% loss=0.085905 lr=4.284855e-05 lr_backbone=4.284855e-05 elapsed=29:43:39 avg_step=3.77s recent_step=3.78s eta=22:34:22 +2026-09-15 20:14:41 | INFO | train rank=0 | step=28430/50000 progress=56.9% loss=0.093951 lr=4.281582e-05 lr_backbone=4.281582e-05 elapsed=29:44:19 avg_step=3.77s recent_step=4.02s eta=22:33:46 +2026-09-15 20:15:24 | INFO | train rank=0 | step=28440/50000 progress=56.9% loss=0.102268 lr=4.278310e-05 lr_backbone=4.278310e-05 elapsed=29:45:02 avg_step=3.77s recent_step=4.33s eta=22:33:13 +2026-09-15 20:15:51 | INFO | train rank=0 | step=28450/50000 progress=56.9% loss=0.084673 lr=4.275038e-05 lr_backbone=4.275038e-05 elapsed=29:45:29 avg_step=3.77s recent_step=2.65s eta=22:32:26 +2026-09-15 20:16:28 | INFO | train rank=0 | step=28460/50000 progress=56.9% loss=0.095316 lr=4.271766e-05 lr_backbone=4.271766e-05 elapsed=29:46:06 avg_step=3.77s recent_step=3.70s eta=22:31:48 +2026-09-15 20:17:03 | INFO | train rank=0 | step=28470/50000 progress=56.9% loss=0.100853 lr=4.268494e-05 lr_backbone=4.268494e-05 elapsed=29:46:41 avg_step=3.77s recent_step=3.57s eta=22:31:09 +2026-09-15 20:17:48 | INFO | train rank=0 | step=28480/50000 progress=57.0% loss=0.098677 lr=4.265223e-05 lr_backbone=4.265223e-05 elapsed=29:47:26 avg_step=3.77s recent_step=4.50s eta=22:30:37 +2026-09-15 20:18:21 | INFO | train rank=0 | step=28490/50000 progress=57.0% loss=0.095593 lr=4.261952e-05 lr_backbone=4.261952e-05 elapsed=29:47:59 avg_step=3.77s recent_step=3.31s eta=22:29:56 +2026-09-15 20:18:51 | INFO | train rank=0 | step=28500/50000 progress=57.0% loss=0.087883 lr=4.258682e-05 lr_backbone=4.258682e-05 elapsed=29:48:29 avg_step=3.77s recent_step=2.94s eta=22:29:12 +2026-09-15 20:19:34 | INFO | train rank=0 | step=28510/50000 progress=57.0% loss=0.087524 lr=4.255412e-05 lr_backbone=4.255412e-05 elapsed=29:49:12 avg_step=3.77s recent_step=4.28s eta=22:28:38 +2026-09-15 20:20:13 | INFO | train rank=0 | step=28520/50000 progress=57.0% loss=0.095262 lr=4.252142e-05 lr_backbone=4.252142e-05 elapsed=29:49:51 avg_step=3.77s recent_step=3.98s eta=22:28:02 +2026-09-15 20:20:45 | INFO | train rank=0 | step=28530/50000 progress=57.1% loss=0.084761 lr=4.248872e-05 lr_backbone=4.248872e-05 elapsed=29:50:23 avg_step=3.77s recent_step=3.20s eta=22:27:20 +2026-09-15 20:21:22 | INFO | train rank=0 | step=28540/50000 progress=57.1% loss=0.073697 lr=4.245603e-05 lr_backbone=4.245603e-05 elapsed=29:51:00 avg_step=3.77s recent_step=3.69s eta=22:26:42 +2026-09-15 20:22:09 | INFO | train rank=0 | step=28550/50000 progress=57.1% loss=0.083524 lr=4.242334e-05 lr_backbone=4.242334e-05 elapsed=29:51:47 avg_step=3.77s recent_step=4.68s eta=22:26:11 +2026-09-15 20:22:56 | INFO | train rank=0 | step=28560/50000 progress=57.1% loss=0.103898 lr=4.239065e-05 lr_backbone=4.239065e-05 elapsed=29:52:34 avg_step=3.77s recent_step=4.68s eta=22:25:41 +2026-09-15 20:23:28 | INFO | train rank=0 | step=28570/50000 progress=57.1% loss=0.102470 lr=4.235797e-05 lr_backbone=4.235797e-05 elapsed=29:53:06 avg_step=3.77s recent_step=3.18s eta=22:24:58 +2026-09-15 20:24:03 | INFO | train rank=0 | step=28580/50000 progress=57.2% loss=0.112063 lr=4.232529e-05 lr_backbone=4.232529e-05 elapsed=29:53:41 avg_step=3.77s recent_step=3.59s eta=22:24:19 +2026-09-15 20:24:43 | INFO | train rank=0 | step=28590/50000 progress=57.2% loss=0.079978 lr=4.229262e-05 lr_backbone=4.229262e-05 elapsed=29:54:21 avg_step=3.77s recent_step=3.99s eta=22:23:43 +2026-09-15 20:25:16 | INFO | train rank=0 | step=28600/50000 progress=57.2% loss=0.086552 lr=4.225994e-05 lr_backbone=4.225994e-05 elapsed=29:54:54 avg_step=3.77s recent_step=3.27s eta=22:23:02 +2026-09-15 20:25:52 | INFO | train rank=0 | step=28610/50000 progress=57.2% loss=0.098841 lr=4.222727e-05 lr_backbone=4.222727e-05 elapsed=29:55:30 avg_step=3.77s recent_step=3.64s eta=22:22:24 +2026-09-15 20:26:24 | INFO | train rank=0 | step=28620/50000 progress=57.2% loss=0.086131 lr=4.219461e-05 lr_backbone=4.219461e-05 elapsed=29:56:02 avg_step=3.77s recent_step=3.15s eta=22:21:41 +2026-09-15 20:27:15 | INFO | train rank=0 | step=28630/50000 progress=57.3% loss=0.089059 lr=4.216195e-05 lr_backbone=4.216195e-05 elapsed=29:56:53 avg_step=3.77s recent_step=5.07s eta=22:21:13 +2026-09-15 20:27:55 | INFO | train rank=0 | step=28640/50000 progress=57.3% loss=0.078688 lr=4.212929e-05 lr_backbone=4.212929e-05 elapsed=29:57:33 avg_step=3.77s recent_step=4.01s eta=22:20:38 +2026-09-15 20:28:31 | INFO | train rank=0 | step=28650/50000 progress=57.3% loss=0.083953 lr=4.209663e-05 lr_backbone=4.209663e-05 elapsed=29:58:09 avg_step=3.77s recent_step=3.64s eta=22:19:59 +2026-09-15 20:29:04 | INFO | train rank=0 | step=28660/50000 progress=57.3% loss=0.099912 lr=4.206398e-05 lr_backbone=4.206398e-05 elapsed=29:58:42 avg_step=3.77s recent_step=3.25s eta=22:19:17 +2026-09-15 20:29:51 | INFO | train rank=0 | step=28670/50000 progress=57.3% loss=0.119645 lr=4.203133e-05 lr_backbone=4.203133e-05 elapsed=29:59:29 avg_step=3.77s recent_step=4.77s eta=22:18:47 +2026-09-15 20:30:28 | INFO | train rank=0 | step=28680/50000 progress=57.4% loss=0.091906 lr=4.199869e-05 lr_backbone=4.199869e-05 elapsed=30:00:06 avg_step=3.77s recent_step=3.68s eta=22:18:09 +2026-09-15 20:31:03 | INFO | train rank=0 | step=28690/50000 progress=57.4% loss=0.091441 lr=4.196605e-05 lr_backbone=4.196605e-05 elapsed=30:00:41 avg_step=3.77s recent_step=3.45s eta=22:17:29 +2026-09-15 20:31:30 | INFO | train rank=0 | step=28700/50000 progress=57.4% loss=0.066715 lr=4.193341e-05 lr_backbone=4.193341e-05 elapsed=30:01:08 avg_step=3.77s recent_step=2.69s eta=22:16:43 +2026-09-15 20:32:21 | INFO | train rank=0 | step=28710/50000 progress=57.4% loss=0.089329 lr=4.190077e-05 lr_backbone=4.190077e-05 elapsed=30:01:59 avg_step=3.77s recent_step=5.14s eta=22:16:16 +2026-09-15 20:32:49 | INFO | train rank=0 | step=28720/50000 progress=57.4% loss=0.094344 lr=4.186814e-05 lr_backbone=4.186814e-05 elapsed=30:02:27 avg_step=3.77s recent_step=2.82s eta=22:15:31 +2026-09-15 20:33:29 | INFO | train rank=0 | step=28730/50000 progress=57.5% loss=0.096806 lr=4.183551e-05 lr_backbone=4.183551e-05 elapsed=30:03:07 avg_step=3.77s recent_step=3.94s eta=22:14:55 +2026-09-15 20:34:01 | INFO | train rank=0 | step=28740/50000 progress=57.5% loss=0.082494 lr=4.180289e-05 lr_backbone=4.180289e-05 elapsed=30:03:39 avg_step=3.77s recent_step=3.29s eta=22:14:14 +2026-09-15 20:34:44 | INFO | train rank=0 | step=28750/50000 progress=57.5% loss=0.106918 lr=4.177027e-05 lr_backbone=4.177027e-05 elapsed=30:04:22 avg_step=3.77s recent_step=4.30s eta=22:13:40 +2026-09-15 20:35:17 | INFO | train rank=0 | step=28760/50000 progress=57.5% loss=0.101463 lr=4.173765e-05 lr_backbone=4.173765e-05 elapsed=30:04:55 avg_step=3.77s recent_step=3.30s eta=22:12:59 +2026-09-15 20:35:52 | INFO | train rank=0 | step=28770/50000 progress=57.5% loss=0.082475 lr=4.170504e-05 lr_backbone=4.170504e-05 elapsed=30:05:30 avg_step=3.77s recent_step=3.42s eta=22:12:19 +2026-09-15 20:36:26 | INFO | train rank=0 | step=28780/50000 progress=57.6% loss=0.108008 lr=4.167243e-05 lr_backbone=4.167243e-05 elapsed=30:06:04 avg_step=3.77s recent_step=3.40s eta=22:11:38 +2026-09-15 20:37:12 | INFO | train rank=0 | step=28790/50000 progress=57.6% loss=0.108455 lr=4.163983e-05 lr_backbone=4.163983e-05 elapsed=30:06:50 avg_step=3.77s recent_step=4.61s eta=22:11:07 +2026-09-15 20:37:47 | INFO | train rank=0 | step=28800/50000 progress=57.6% loss=0.083470 lr=4.160722e-05 lr_backbone=4.160722e-05 elapsed=30:07:25 avg_step=3.77s recent_step=3.55s eta=22:10:28 +2026-09-15 20:38:22 | INFO | train rank=0 | step=28810/50000 progress=57.6% loss=0.080057 lr=4.157463e-05 lr_backbone=4.157463e-05 elapsed=30:08:00 avg_step=3.77s recent_step=3.42s eta=22:09:48 +2026-09-15 20:38:59 | INFO | train rank=0 | step=28820/50000 progress=57.6% loss=0.082751 lr=4.154203e-05 lr_backbone=4.154203e-05 elapsed=30:08:37 avg_step=3.77s recent_step=3.74s eta=22:09:10 +2026-09-15 20:39:36 | INFO | train rank=0 | step=28830/50000 progress=57.7% loss=0.112949 lr=4.150944e-05 lr_backbone=4.150944e-05 elapsed=30:09:14 avg_step=3.77s recent_step=3.74s eta=22:08:32 +2026-09-15 20:40:28 | INFO | train rank=0 | step=28840/50000 progress=57.7% loss=0.096055 lr=4.147685e-05 lr_backbone=4.147685e-05 elapsed=30:10:06 avg_step=3.77s recent_step=5.22s eta=22:08:05 +2026-09-15 20:41:05 | INFO | train rank=0 | step=28850/50000 progress=57.7% loss=0.083763 lr=4.144427e-05 lr_backbone=4.144427e-05 elapsed=30:10:43 avg_step=3.77s recent_step=3.69s eta=22:07:27 +2026-09-15 20:41:39 | INFO | train rank=0 | step=28860/50000 progress=57.7% loss=0.105881 lr=4.141169e-05 lr_backbone=4.141169e-05 elapsed=30:11:17 avg_step=3.77s recent_step=3.32s eta=22:06:46 +2026-09-15 20:42:10 | INFO | train rank=0 | step=28870/50000 progress=57.7% loss=0.095854 lr=4.137911e-05 lr_backbone=4.137911e-05 elapsed=30:11:48 avg_step=3.77s recent_step=3.11s eta=22:06:03 +2026-09-15 20:42:52 | INFO | train rank=0 | step=28880/50000 progress=57.8% loss=0.088538 lr=4.134654e-05 lr_backbone=4.134654e-05 elapsed=30:12:30 avg_step=3.77s recent_step=4.22s eta=22:05:29 +2026-09-15 20:43:36 | INFO | train rank=0 | step=28890/50000 progress=57.8% loss=0.089476 lr=4.131397e-05 lr_backbone=4.131397e-05 elapsed=30:13:14 avg_step=3.77s recent_step=4.45s eta=22:04:56 +2026-09-15 20:44:06 | INFO | train rank=0 | step=28900/50000 progress=57.8% loss=0.087339 lr=4.128141e-05 lr_backbone=4.128141e-05 elapsed=30:13:44 avg_step=3.77s recent_step=2.94s eta=22:04:13 +2026-09-15 20:44:45 | INFO | train rank=0 | step=28910/50000 progress=57.8% loss=0.093632 lr=4.124885e-05 lr_backbone=4.124885e-05 elapsed=30:14:23 avg_step=3.77s recent_step=3.92s eta=22:03:36 +2026-09-15 20:45:34 | INFO | train rank=0 | step=28920/50000 progress=57.8% loss=0.104341 lr=4.121629e-05 lr_backbone=4.121629e-05 elapsed=30:15:12 avg_step=3.77s recent_step=4.89s eta=22:03:07 +2026-09-15 20:46:14 | INFO | train rank=0 | step=28930/50000 progress=57.9% loss=0.105653 lr=4.118374e-05 lr_backbone=4.118374e-05 elapsed=30:15:52 avg_step=3.77s recent_step=3.99s eta=22:02:30 +2026-09-15 20:46:49 | INFO | train rank=0 | step=28940/50000 progress=57.9% loss=0.104399 lr=4.115119e-05 lr_backbone=4.115119e-05 elapsed=30:16:27 avg_step=3.77s recent_step=3.50s eta=22:01:51 +2026-09-15 20:47:24 | INFO | train rank=0 | step=28950/50000 progress=57.9% loss=0.070436 lr=4.111864e-05 lr_backbone=4.111864e-05 elapsed=30:17:02 avg_step=3.77s recent_step=3.56s eta=22:01:12 +2026-09-15 20:48:12 | INFO | train rank=0 | step=28960/50000 progress=57.9% loss=0.086385 lr=4.108610e-05 lr_backbone=4.108610e-05 elapsed=30:17:50 avg_step=3.77s recent_step=4.80s eta=22:00:42 +2026-09-15 20:48:49 | INFO | train rank=0 | step=28970/50000 progress=57.9% loss=0.105824 lr=4.105356e-05 lr_backbone=4.105356e-05 elapsed=30:18:27 avg_step=3.77s recent_step=3.63s eta=22:00:03 +2026-09-15 20:49:15 | INFO | train rank=0 | step=28980/50000 progress=58.0% loss=0.101453 lr=4.102103e-05 lr_backbone=4.102103e-05 elapsed=30:18:53 avg_step=3.77s recent_step=2.68s eta=21:59:17 +2026-09-15 20:49:48 | INFO | train rank=0 | step=28990/50000 progress=58.0% loss=0.086697 lr=4.098850e-05 lr_backbone=4.098850e-05 elapsed=30:19:26 avg_step=3.77s recent_step=3.28s eta=21:58:36 +2026-09-15 20:50:34 | INFO | train rank=0 | step=29000/50000 progress=58.0% loss=0.088148 lr=4.095597e-05 lr_backbone=4.095597e-05 elapsed=30:20:12 avg_step=3.77s recent_step=4.53s eta=21:58:04 +2026-09-15 20:51:12 | INFO | train rank=0 | step=29010/50000 progress=58.0% loss=0.090599 lr=4.092345e-05 lr_backbone=4.092345e-05 elapsed=30:20:50 avg_step=3.77s recent_step=3.89s eta=21:57:27 +2026-09-15 20:51:49 | INFO | train rank=0 | step=29020/50000 progress=58.0% loss=0.076124 lr=4.089093e-05 lr_backbone=4.089093e-05 elapsed=30:21:27 avg_step=3.77s recent_step=3.66s eta=21:56:49 +2026-09-15 20:52:25 | INFO | train rank=0 | step=29030/50000 progress=58.1% loss=0.088967 lr=4.085842e-05 lr_backbone=4.085842e-05 elapsed=30:22:03 avg_step=3.77s recent_step=3.58s eta=21:56:10 +2026-09-15 20:53:09 | INFO | train rank=0 | step=29040/50000 progress=58.1% loss=0.125564 lr=4.082591e-05 lr_backbone=4.082591e-05 elapsed=30:22:47 avg_step=3.77s recent_step=4.40s eta=21:55:37 +2026-09-15 20:53:41 | INFO | train rank=0 | step=29050/50000 progress=58.1% loss=0.092376 lr=4.079340e-05 lr_backbone=4.079340e-05 elapsed=30:23:19 avg_step=3.77s recent_step=3.17s eta=21:54:55 +2026-09-15 20:54:03 | INFO | train rank=0 | step=29060/50000 progress=58.1% loss=0.092507 lr=4.076090e-05 lr_backbone=4.076090e-05 elapsed=30:23:41 avg_step=3.77s recent_step=2.27s eta=21:54:07 +2026-09-15 20:54:41 | INFO | train rank=0 | step=29070/50000 progress=58.1% loss=0.077273 lr=4.072840e-05 lr_backbone=4.072840e-05 elapsed=30:24:19 avg_step=3.77s recent_step=3.74s eta=21:53:29 +2026-09-15 20:55:30 | INFO | train rank=0 | step=29080/50000 progress=58.2% loss=0.109217 lr=4.069591e-05 lr_backbone=4.069591e-05 elapsed=30:25:08 avg_step=3.77s recent_step=4.96s eta=21:53:00 +2026-09-15 20:55:55 | INFO | train rank=0 | step=29090/50000 progress=58.2% loss=0.122953 lr=4.066342e-05 lr_backbone=4.066342e-05 elapsed=30:25:33 avg_step=3.77s recent_step=2.42s eta=21:52:12 +2026-09-15 20:56:27 | INFO | train rank=0 | step=29100/50000 progress=58.2% loss=0.088414 lr=4.063093e-05 lr_backbone=4.063093e-05 elapsed=30:26:05 avg_step=3.77s recent_step=3.26s eta=21:51:31 +2026-09-15 20:56:59 | INFO | train rank=0 | step=29110/50000 progress=58.2% loss=0.090470 lr=4.059845e-05 lr_backbone=4.059845e-05 elapsed=30:26:37 avg_step=3.76s recent_step=3.23s eta=21:50:49 +2026-09-15 20:57:53 | INFO | train rank=0 | step=29120/50000 progress=58.2% loss=0.089926 lr=4.056598e-05 lr_backbone=4.056598e-05 elapsed=30:27:31 avg_step=3.77s recent_step=5.39s eta=21:50:23 +2026-09-15 20:58:21 | INFO | train rank=0 | step=29130/50000 progress=58.3% loss=0.110769 lr=4.053350e-05 lr_backbone=4.053350e-05 elapsed=30:27:59 avg_step=3.77s recent_step=2.73s eta=21:49:38 +2026-09-15 20:58:51 | INFO | train rank=0 | step=29140/50000 progress=58.3% loss=0.112499 lr=4.050103e-05 lr_backbone=4.050103e-05 elapsed=30:28:29 avg_step=3.76s recent_step=3.07s eta=21:48:56 +2026-09-15 20:59:25 | INFO | train rank=0 | step=29150/50000 progress=58.3% loss=0.094279 lr=4.046857e-05 lr_backbone=4.046857e-05 elapsed=30:29:03 avg_step=3.76s recent_step=3.37s eta=21:48:15 +2026-09-15 21:00:23 | INFO | train rank=0 | step=29160/50000 progress=58.3% loss=0.124530 lr=4.043611e-05 lr_backbone=4.043611e-05 elapsed=30:30:01 avg_step=3.77s recent_step=5.78s eta=21:47:52 +2026-09-15 21:00:57 | INFO | train rank=0 | step=29170/50000 progress=58.3% loss=0.097617 lr=4.040365e-05 lr_backbone=4.040365e-05 elapsed=30:30:35 avg_step=3.77s recent_step=3.37s eta=21:47:12 +2026-09-15 21:01:31 | INFO | train rank=0 | step=29180/50000 progress=58.4% loss=0.089302 lr=4.037120e-05 lr_backbone=4.037120e-05 elapsed=30:31:09 avg_step=3.77s recent_step=3.45s eta=21:46:32 +2026-09-15 21:02:00 | INFO | train rank=0 | step=29190/50000 progress=58.4% loss=0.094308 lr=4.033875e-05 lr_backbone=4.033875e-05 elapsed=30:31:39 avg_step=3.76s recent_step=2.95s eta=21:45:48 +2026-09-15 21:02:54 | INFO | train rank=0 | step=29200/50000 progress=58.4% loss=0.077788 lr=4.030631e-05 lr_backbone=4.030631e-05 elapsed=30:32:32 avg_step=3.77s recent_step=5.37s eta=21:45:22 +2026-09-15 21:03:27 | INFO | train rank=0 | step=29210/50000 progress=58.4% loss=0.090019 lr=4.027387e-05 lr_backbone=4.027387e-05 elapsed=30:33:05 avg_step=3.77s recent_step=3.28s eta=21:44:41 +2026-09-15 21:04:01 | INFO | train rank=0 | step=29220/50000 progress=58.4% loss=0.103148 lr=4.024143e-05 lr_backbone=4.024143e-05 elapsed=30:33:39 avg_step=3.77s recent_step=3.39s eta=21:44:01 +2026-09-15 21:04:45 | INFO | train rank=0 | step=29230/50000 progress=58.5% loss=0.086537 lr=4.020900e-05 lr_backbone=4.020900e-05 elapsed=30:34:23 avg_step=3.77s recent_step=4.45s eta=21:43:28 +2026-09-15 21:05:31 | INFO | train rank=0 | step=29240/50000 progress=58.5% loss=0.103631 lr=4.017657e-05 lr_backbone=4.017657e-05 elapsed=30:35:10 avg_step=3.77s recent_step=4.61s eta=21:42:56 +2026-09-15 21:05:59 | INFO | train rank=0 | step=29250/50000 progress=58.5% loss=0.115279 lr=4.014415e-05 lr_backbone=4.014415e-05 elapsed=30:35:37 avg_step=3.77s recent_step=2.79s eta=21:42:12 +2026-09-15 21:06:33 | INFO | train rank=0 | step=29260/50000 progress=58.5% loss=0.113635 lr=4.011173e-05 lr_backbone=4.011173e-05 elapsed=30:36:11 avg_step=3.77s recent_step=3.33s eta=21:41:31 +2026-09-15 21:07:12 | INFO | train rank=0 | step=29270/50000 progress=58.5% loss=0.100853 lr=4.007932e-05 lr_backbone=4.007932e-05 elapsed=30:36:50 avg_step=3.77s recent_step=3.88s eta=21:40:54 +2026-09-15 21:08:01 | INFO | train rank=0 | step=29280/50000 progress=58.6% loss=0.081149 lr=4.004691e-05 lr_backbone=4.004691e-05 elapsed=30:37:39 avg_step=3.77s recent_step=4.99s eta=21:40:25 +2026-09-15 21:08:37 | INFO | train rank=0 | step=29290/50000 progress=58.6% loss=0.070326 lr=4.001450e-05 lr_backbone=4.001450e-05 elapsed=30:38:15 avg_step=3.77s recent_step=3.56s eta=21:39:46 +2026-09-15 21:09:12 | INFO | train rank=0 | step=29300/50000 progress=58.6% loss=0.090888 lr=3.998210e-05 lr_backbone=3.998210e-05 elapsed=30:38:50 avg_step=3.77s recent_step=3.50s eta=21:39:06 +2026-09-15 21:09:51 | INFO | train rank=0 | step=29310/50000 progress=58.6% loss=0.106318 lr=3.994970e-05 lr_backbone=3.994970e-05 elapsed=30:39:29 avg_step=3.77s recent_step=3.87s eta=21:38:30 +2026-09-15 21:10:39 | INFO | train rank=0 | step=29320/50000 progress=58.6% loss=0.108511 lr=3.991731e-05 lr_backbone=3.991731e-05 elapsed=30:40:17 avg_step=3.77s recent_step=4.79s eta=21:37:59 +2026-09-15 21:11:09 | INFO | train rank=0 | step=29330/50000 progress=58.7% loss=0.099737 lr=3.988492e-05 lr_backbone=3.988492e-05 elapsed=30:40:47 avg_step=3.77s recent_step=3.01s eta=21:37:16 +2026-09-15 21:11:40 | INFO | train rank=0 | step=29340/50000 progress=58.7% loss=0.099888 lr=3.985254e-05 lr_backbone=3.985254e-05 elapsed=30:41:18 avg_step=3.77s recent_step=3.07s eta=21:36:34 +2026-09-15 21:12:16 | INFO | train rank=0 | step=29350/50000 progress=58.7% loss=0.088514 lr=3.982016e-05 lr_backbone=3.982016e-05 elapsed=30:41:54 avg_step=3.77s recent_step=3.60s eta=21:35:55 +2026-09-15 21:13:04 | INFO | train rank=0 | step=29360/50000 progress=58.7% loss=0.068663 lr=3.978779e-05 lr_backbone=3.978779e-05 elapsed=30:42:42 avg_step=3.77s recent_step=4.81s eta=21:35:24 +2026-09-15 21:13:37 | INFO | train rank=0 | step=29370/50000 progress=58.7% loss=0.101969 lr=3.975542e-05 lr_backbone=3.975542e-05 elapsed=30:43:15 avg_step=3.77s recent_step=3.31s eta=21:34:44 +2026-09-15 21:14:13 | INFO | train rank=0 | step=29380/50000 progress=58.8% loss=0.108615 lr=3.972305e-05 lr_backbone=3.972305e-05 elapsed=30:43:51 avg_step=3.77s recent_step=3.62s eta=21:34:05 +2026-09-15 21:14:47 | INFO | train rank=0 | step=29390/50000 progress=58.8% loss=0.134975 lr=3.969069e-05 lr_backbone=3.969069e-05 elapsed=30:44:25 avg_step=3.77s recent_step=3.39s eta=21:33:25 +2026-09-15 21:15:34 | INFO | train rank=0 | step=29400/50000 progress=58.8% loss=0.087022 lr=3.965833e-05 lr_backbone=3.965833e-05 elapsed=30:45:12 avg_step=3.77s recent_step=4.67s eta=21:32:53 +2026-09-15 21:15:58 | INFO | train rank=0 | step=29410/50000 progress=58.8% loss=0.105329 lr=3.962598e-05 lr_backbone=3.962598e-05 elapsed=30:45:36 avg_step=3.77s recent_step=2.46s eta=21:32:07 +2026-09-15 21:16:36 | INFO | train rank=0 | step=29420/50000 progress=58.8% loss=0.125546 lr=3.959363e-05 lr_backbone=3.959363e-05 elapsed=30:46:14 avg_step=3.77s recent_step=3.75s eta=21:31:29 +2026-09-15 21:17:10 | INFO | train rank=0 | step=29430/50000 progress=58.9% loss=0.077356 lr=3.956129e-05 lr_backbone=3.956129e-05 elapsed=30:46:48 avg_step=3.77s recent_step=3.45s eta=21:30:49 +2026-09-15 21:18:04 | INFO | train rank=0 | step=29440/50000 progress=58.9% loss=0.125251 lr=3.952895e-05 lr_backbone=3.952895e-05 elapsed=30:47:42 avg_step=3.77s recent_step=5.35s eta=21:30:22 +2026-09-15 21:18:35 | INFO | train rank=0 | step=29450/50000 progress=58.9% loss=0.101413 lr=3.949662e-05 lr_backbone=3.949662e-05 elapsed=30:48:13 avg_step=3.77s recent_step=3.13s eta=21:29:40 +2026-09-15 21:19:01 | INFO | train rank=0 | step=29460/50000 progress=58.9% loss=0.082165 lr=3.946429e-05 lr_backbone=3.946429e-05 elapsed=30:48:39 avg_step=3.77s recent_step=2.55s eta=21:28:54 +2026-09-15 21:19:31 | INFO | train rank=0 | step=29470/50000 progress=58.9% loss=0.084803 lr=3.943196e-05 lr_backbone=3.943196e-05 elapsed=30:49:09 avg_step=3.76s recent_step=3.00s eta=21:28:11 +2026-09-15 21:20:25 | INFO | train rank=0 | step=29480/50000 progress=59.0% loss=0.105153 lr=3.939964e-05 lr_backbone=3.939964e-05 elapsed=30:50:03 avg_step=3.77s recent_step=5.37s eta=21:27:45 +2026-09-15 21:20:55 | INFO | train rank=0 | step=29490/50000 progress=59.0% loss=0.092975 lr=3.936733e-05 lr_backbone=3.936733e-05 elapsed=30:50:33 avg_step=3.77s recent_step=3.09s eta=21:27:02 +2026-09-15 21:21:32 | INFO | train rank=0 | step=29500/50000 progress=59.0% loss=0.081869 lr=3.933502e-05 lr_backbone=3.933502e-05 elapsed=30:51:10 avg_step=3.77s recent_step=3.66s eta=21:26:24 +2026-09-15 21:22:07 | INFO | train rank=0 | step=29510/50000 progress=59.0% loss=0.093922 lr=3.930271e-05 lr_backbone=3.930271e-05 elapsed=30:51:45 avg_step=3.77s recent_step=3.51s eta=21:25:45 +2026-09-15 21:23:04 | INFO | train rank=0 | step=29520/50000 progress=59.0% loss=0.073605 lr=3.927041e-05 lr_backbone=3.927041e-05 elapsed=30:52:42 avg_step=3.77s recent_step=5.68s eta=21:25:20 +2026-09-15 21:23:33 | INFO | train rank=0 | step=29530/50000 progress=59.1% loss=0.109221 lr=3.923811e-05 lr_backbone=3.923811e-05 elapsed=30:53:11 avg_step=3.77s recent_step=2.86s eta=21:24:36 +2026-09-15 21:23:58 | INFO | train rank=0 | step=29540/50000 progress=59.1% loss=0.083478 lr=3.920582e-05 lr_backbone=3.920582e-05 elapsed=30:53:36 avg_step=3.76s recent_step=2.59s eta=21:23:51 +2026-09-15 21:24:29 | INFO | train rank=0 | step=29550/50000 progress=59.1% loss=0.137564 lr=3.917354e-05 lr_backbone=3.917354e-05 elapsed=30:54:07 avg_step=3.76s recent_step=3.01s eta=21:23:08 +2026-09-15 21:25:14 | INFO | train rank=0 | step=29560/50000 progress=59.1% loss=0.091246 lr=3.914125e-05 lr_backbone=3.914125e-05 elapsed=30:54:52 avg_step=3.76s recent_step=4.50s eta=21:22:35 +2026-09-15 21:25:48 | INFO | train rank=0 | step=29570/50000 progress=59.1% loss=0.088449 lr=3.910898e-05 lr_backbone=3.910898e-05 elapsed=30:55:26 avg_step=3.76s recent_step=3.42s eta=21:21:55 +2026-09-15 21:26:19 | INFO | train rank=0 | step=29580/50000 progress=59.2% loss=0.118892 lr=3.907670e-05 lr_backbone=3.907670e-05 elapsed=30:55:57 avg_step=3.76s recent_step=3.09s eta=21:21:13 +2026-09-15 21:27:00 | INFO | train rank=0 | step=29590/50000 progress=59.2% loss=0.093902 lr=3.904443e-05 lr_backbone=3.904443e-05 elapsed=30:56:38 avg_step=3.76s recent_step=4.14s eta=21:20:38 +2026-09-15 21:27:49 | INFO | train rank=0 | step=29600/50000 progress=59.2% loss=0.091936 lr=3.901217e-05 lr_backbone=3.901217e-05 elapsed=30:57:27 avg_step=3.77s recent_step=4.89s eta=21:20:08 +2026-09-15 21:28:17 | INFO | train rank=0 | step=29610/50000 progress=59.2% loss=0.110289 lr=3.897991e-05 lr_backbone=3.897991e-05 elapsed=30:57:55 avg_step=3.76s recent_step=2.83s eta=21:19:24 +2026-09-15 21:28:52 | INFO | train rank=0 | step=29620/50000 progress=59.2% loss=0.074865 lr=3.894766e-05 lr_backbone=3.894766e-05 elapsed=30:58:30 avg_step=3.76s recent_step=3.51s eta=21:18:44 +2026-09-15 21:29:32 | INFO | train rank=0 | step=29630/50000 progress=59.3% loss=0.102831 lr=3.891541e-05 lr_backbone=3.891541e-05 elapsed=30:59:10 avg_step=3.76s recent_step=3.99s eta=21:18:08 +2026-09-15 21:30:22 | INFO | train rank=0 | step=29640/50000 progress=59.3% loss=0.083718 lr=3.888317e-05 lr_backbone=3.888317e-05 elapsed=31:00:00 avg_step=3.77s recent_step=4.95s eta=21:17:39 +2026-09-15 21:31:01 | INFO | train rank=0 | step=29650/50000 progress=59.3% loss=0.101576 lr=3.885093e-05 lr_backbone=3.885093e-05 elapsed=31:00:39 avg_step=3.77s recent_step=3.92s eta=21:17:02 +2026-09-15 21:31:37 | INFO | train rank=0 | step=29660/50000 progress=59.3% loss=0.081220 lr=3.881869e-05 lr_backbone=3.881869e-05 elapsed=31:01:15 avg_step=3.77s recent_step=3.61s eta=21:16:23 +2026-09-15 21:32:12 | INFO | train rank=0 | step=29670/50000 progress=59.3% loss=0.093260 lr=3.878646e-05 lr_backbone=3.878646e-05 elapsed=31:01:50 avg_step=3.77s recent_step=3.48s eta=21:15:44 +2026-09-15 21:32:52 | INFO | train rank=0 | step=29680/50000 progress=59.4% loss=0.092874 lr=3.875424e-05 lr_backbone=3.875424e-05 elapsed=31:02:30 avg_step=3.77s recent_step=4.04s eta=21:15:08 +2026-09-15 21:33:31 | INFO | train rank=0 | step=29690/50000 progress=59.4% loss=0.093336 lr=3.872202e-05 lr_backbone=3.872202e-05 elapsed=31:03:09 avg_step=3.77s recent_step=3.87s eta=21:14:31 +2026-09-15 21:34:02 | INFO | train rank=0 | step=29700/50000 progress=59.4% loss=0.086240 lr=3.868980e-05 lr_backbone=3.868980e-05 elapsed=31:03:40 avg_step=3.77s recent_step=3.15s eta=21:13:49 +2026-09-15 21:34:48 | INFO | train rank=0 | step=29710/50000 progress=59.4% loss=0.075166 lr=3.865759e-05 lr_backbone=3.865759e-05 elapsed=31:04:26 avg_step=3.77s recent_step=4.62s eta=21:13:17 +2026-09-15 21:35:30 | INFO | train rank=0 | step=29720/50000 progress=59.4% loss=0.089361 lr=3.862539e-05 lr_backbone=3.862539e-05 elapsed=31:05:08 avg_step=3.77s recent_step=4.12s eta=21:12:42 +2026-09-15 21:36:01 | INFO | train rank=0 | step=29730/50000 progress=59.5% loss=0.095536 lr=3.859319e-05 lr_backbone=3.859319e-05 elapsed=31:05:39 avg_step=3.77s recent_step=3.10s eta=21:12:00 +2026-09-15 21:36:38 | INFO | train rank=0 | step=29740/50000 progress=59.5% loss=0.085833 lr=3.856099e-05 lr_backbone=3.856099e-05 elapsed=31:06:16 avg_step=3.77s recent_step=3.72s eta=21:11:22 +2026-09-15 21:37:20 | INFO | train rank=0 | step=29750/50000 progress=59.5% loss=0.104717 lr=3.852880e-05 lr_backbone=3.852880e-05 elapsed=31:06:58 avg_step=3.77s recent_step=4.24s eta=21:10:48 +2026-09-15 21:37:49 | INFO | train rank=0 | step=29760/50000 progress=59.5% loss=0.094979 lr=3.849662e-05 lr_backbone=3.849662e-05 elapsed=31:07:27 avg_step=3.77s recent_step=2.92s eta=21:10:04 +2026-09-15 21:38:30 | INFO | train rank=0 | step=29770/50000 progress=59.5% loss=0.096603 lr=3.846444e-05 lr_backbone=3.846444e-05 elapsed=31:08:08 avg_step=3.77s recent_step=4.06s eta=21:09:28 +2026-09-15 21:39:03 | INFO | train rank=0 | step=29780/50000 progress=59.6% loss=0.112794 lr=3.843226e-05 lr_backbone=3.843226e-05 elapsed=31:08:41 avg_step=3.77s recent_step=3.33s eta=21:08:48 +2026-09-15 21:39:50 | INFO | train rank=0 | step=29790/50000 progress=59.6% loss=0.083815 lr=3.840009e-05 lr_backbone=3.840009e-05 elapsed=31:09:28 avg_step=3.77s recent_step=4.70s eta=21:08:17 +2026-09-15 21:40:26 | INFO | train rank=0 | step=29800/50000 progress=59.6% loss=0.093556 lr=3.836793e-05 lr_backbone=3.836793e-05 elapsed=31:10:04 avg_step=3.77s recent_step=3.53s eta=21:07:37 +2026-09-15 21:41:02 | INFO | train rank=0 | step=29810/50000 progress=59.6% loss=0.083390 lr=3.833577e-05 lr_backbone=3.833577e-05 elapsed=31:10:40 avg_step=3.77s recent_step=3.61s eta=21:06:59 +2026-09-15 21:41:34 | INFO | train rank=0 | step=29820/50000 progress=59.6% loss=0.109631 lr=3.830362e-05 lr_backbone=3.830362e-05 elapsed=31:11:12 avg_step=3.77s recent_step=3.25s eta=21:06:17 +2026-09-15 21:42:31 | INFO | train rank=0 | step=29830/50000 progress=59.7% loss=0.081623 lr=3.827147e-05 lr_backbone=3.827147e-05 elapsed=31:12:09 avg_step=3.77s recent_step=5.66s eta=21:05:53 +2026-09-15 21:43:02 | INFO | train rank=0 | step=29840/50000 progress=59.7% loss=0.094496 lr=3.823932e-05 lr_backbone=3.823932e-05 elapsed=31:12:40 avg_step=3.77s recent_step=3.17s eta=21:05:11 +2026-09-15 21:43:39 | INFO | train rank=0 | step=29850/50000 progress=59.7% loss=0.112374 lr=3.820718e-05 lr_backbone=3.820718e-05 elapsed=31:13:17 avg_step=3.77s recent_step=3.64s eta=21:04:32 +2026-09-15 21:44:12 | INFO | train rank=0 | step=29860/50000 progress=59.7% loss=0.099805 lr=3.817505e-05 lr_backbone=3.817505e-05 elapsed=31:13:50 avg_step=3.77s recent_step=3.35s eta=21:03:52 +2026-09-15 21:45:07 | INFO | train rank=0 | step=29870/50000 progress=59.7% loss=0.081038 lr=3.814292e-05 lr_backbone=3.814292e-05 elapsed=31:14:45 avg_step=3.77s recent_step=5.47s eta=21:03:26 +2026-09-15 21:45:33 | INFO | train rank=0 | step=29880/50000 progress=59.8% loss=0.094125 lr=3.811080e-05 lr_backbone=3.811080e-05 elapsed=31:15:11 avg_step=3.77s recent_step=2.59s eta=21:02:40 +2026-09-15 21:46:03 | INFO | train rank=0 | step=29890/50000 progress=59.8% loss=0.095700 lr=3.807868e-05 lr_backbone=3.807868e-05 elapsed=31:15:41 avg_step=3.77s recent_step=3.00s eta=21:01:58 +2026-09-15 21:46:33 | INFO | train rank=0 | step=29900/50000 progress=59.8% loss=0.084878 lr=3.804657e-05 lr_backbone=3.804657e-05 elapsed=31:16:11 avg_step=3.76s recent_step=3.04s eta=21:01:15 +2026-09-15 21:47:16 | INFO | train rank=0 | step=29910/50000 progress=59.8% loss=0.076181 lr=3.801446e-05 lr_backbone=3.801446e-05 elapsed=31:16:54 avg_step=3.77s recent_step=4.21s eta=21:00:40 +2026-09-15 21:47:54 | INFO | train rank=0 | step=29920/50000 progress=59.8% loss=0.096180 lr=3.798236e-05 lr_backbone=3.798236e-05 elapsed=31:17:32 avg_step=3.77s recent_step=3.86s eta=21:00:03 +2026-09-15 21:48:27 | INFO | train rank=0 | step=29930/50000 progress=59.9% loss=0.065181 lr=3.795026e-05 lr_backbone=3.795026e-05 elapsed=31:18:05 avg_step=3.76s recent_step=3.30s eta=20:59:23 +2026-09-15 21:49:02 | INFO | train rank=0 | step=29940/50000 progress=59.9% loss=0.081838 lr=3.791817e-05 lr_backbone=3.791817e-05 elapsed=31:18:40 avg_step=3.76s recent_step=3.48s eta=20:58:43 +2026-09-15 21:49:52 | INFO | train rank=0 | step=29950/50000 progress=59.9% loss=0.093329 lr=3.788608e-05 lr_backbone=3.788608e-05 elapsed=31:19:30 avg_step=3.77s recent_step=5.01s eta=20:58:14 +2026-09-15 21:50:26 | INFO | train rank=0 | step=29960/50000 progress=59.9% loss=0.076350 lr=3.785400e-05 lr_backbone=3.785400e-05 elapsed=31:20:04 avg_step=3.77s recent_step=3.39s eta=20:57:34 +2026-09-15 21:51:02 | INFO | train rank=0 | step=29970/50000 progress=59.9% loss=0.082368 lr=3.782192e-05 lr_backbone=3.782192e-05 elapsed=31:20:40 avg_step=3.77s recent_step=3.59s eta=20:56:55 +2026-09-15 21:51:32 | INFO | train rank=0 | step=29980/50000 progress=60.0% loss=0.080084 lr=3.778985e-05 lr_backbone=3.778985e-05 elapsed=31:21:10 avg_step=3.76s recent_step=2.98s eta=20:56:12 +2026-09-15 21:52:21 | INFO | train rank=0 | step=29990/50000 progress=60.0% loss=0.093198 lr=3.775779e-05 lr_backbone=3.775779e-05 elapsed=31:21:59 avg_step=3.77s recent_step=4.97s eta=20:55:42 +2026-09-15 21:52:56 | INFO | train rank=0 | step=30000/50000 progress=60.0% loss=0.100375 lr=3.772573e-05 lr_backbone=3.772573e-05 elapsed=31:22:34 avg_step=3.77s recent_step=3.48s eta=20:55:03 +2026-09-15 21:53:26 | INFO | train rank=0 | step=30010/50000 progress=60.0% loss=0.128908 lr=3.769367e-05 lr_backbone=3.769367e-05 elapsed=31:23:04 avg_step=3.76s recent_step=2.98s eta=20:54:20 +2026-09-15 21:54:09 | INFO | train rank=0 | step=30020/50000 progress=60.0% loss=0.074623 lr=3.766162e-05 lr_backbone=3.766162e-05 elapsed=31:23:47 avg_step=3.77s recent_step=4.25s eta=20:53:45 +2026-09-15 21:54:50 | INFO | train rank=0 | step=30030/50000 progress=60.1% loss=0.090464 lr=3.762958e-05 lr_backbone=3.762958e-05 elapsed=31:24:28 avg_step=3.77s recent_step=4.11s eta=20:53:10 +2026-09-15 21:55:24 | INFO | train rank=0 | step=30040/50000 progress=60.1% loss=0.071287 lr=3.759754e-05 lr_backbone=3.759754e-05 elapsed=31:25:02 avg_step=3.77s recent_step=3.48s eta=20:52:30 +2026-09-15 21:56:02 | INFO | train rank=0 | step=30050/50000 progress=60.1% loss=0.102255 lr=3.756551e-05 lr_backbone=3.756551e-05 elapsed=31:25:40 avg_step=3.77s recent_step=3.72s eta=20:51:53 +2026-09-15 21:56:36 | INFO | train rank=0 | step=30060/50000 progress=60.1% loss=0.090924 lr=3.753348e-05 lr_backbone=3.753348e-05 elapsed=31:26:14 avg_step=3.76s recent_step=3.41s eta=20:51:13 +2026-09-15 21:57:25 | INFO | train rank=0 | step=30070/50000 progress=60.1% loss=0.107692 lr=3.750146e-05 lr_backbone=3.750146e-05 elapsed=31:27:03 avg_step=3.77s recent_step=4.94s eta=20:50:43 +2026-09-15 21:58:00 | INFO | train rank=0 | step=30080/50000 progress=60.2% loss=0.103727 lr=3.746944e-05 lr_backbone=3.746944e-05 elapsed=31:27:38 avg_step=3.77s recent_step=3.44s eta=20:50:03 +2026-09-15 21:58:44 | INFO | train rank=0 | step=30090/50000 progress=60.2% loss=0.079185 lr=3.743743e-05 lr_backbone=3.743743e-05 elapsed=31:28:22 avg_step=3.77s recent_step=4.40s eta=20:49:29 +2026-09-15 21:59:17 | INFO | train rank=0 | step=30100/50000 progress=60.2% loss=0.082445 lr=3.740542e-05 lr_backbone=3.740542e-05 elapsed=31:28:55 avg_step=3.77s recent_step=3.32s eta=20:48:49 +2026-09-15 21:59:50 | INFO | train rank=0 | step=30110/50000 progress=60.2% loss=0.090833 lr=3.737342e-05 lr_backbone=3.737342e-05 elapsed=31:29:28 avg_step=3.77s recent_step=3.31s eta=20:48:08 +2026-09-15 22:00:40 | INFO | train rank=0 | step=30120/50000 progress=60.2% loss=0.094862 lr=3.734143e-05 lr_backbone=3.734143e-05 elapsed=31:30:18 avg_step=3.77s recent_step=5.04s eta=20:47:39 +2026-09-15 22:01:08 | INFO | train rank=0 | step=30130/50000 progress=60.3% loss=0.069694 lr=3.730944e-05 lr_backbone=3.730944e-05 elapsed=31:30:46 avg_step=3.77s recent_step=2.77s eta=20:46:55 +2026-09-15 22:01:47 | INFO | train rank=0 | step=30140/50000 progress=60.3% loss=0.090067 lr=3.727745e-05 lr_backbone=3.727745e-05 elapsed=31:31:25 avg_step=3.77s recent_step=3.93s eta=20:46:18 +2026-09-15 22:02:32 | INFO | train rank=0 | step=30150/50000 progress=60.3% loss=0.082244 lr=3.724548e-05 lr_backbone=3.724548e-05 elapsed=31:32:10 avg_step=3.77s recent_step=4.45s eta=20:45:45 +2026-09-15 22:03:23 | INFO | train rank=0 | step=30160/50000 progress=60.3% loss=0.096280 lr=3.721350e-05 lr_backbone=3.721350e-05 elapsed=31:33:01 avg_step=3.77s recent_step=5.16s eta=20:45:16 +2026-09-15 22:03:56 | INFO | train rank=0 | step=30170/50000 progress=60.3% loss=0.089427 lr=3.718154e-05 lr_backbone=3.718154e-05 elapsed=31:33:34 avg_step=3.77s recent_step=3.29s eta=20:44:36 +2026-09-15 22:04:28 | INFO | train rank=0 | step=30180/50000 progress=60.4% loss=0.135562 lr=3.714957e-05 lr_backbone=3.714957e-05 elapsed=31:34:06 avg_step=3.77s recent_step=3.15s eta=20:43:54 +2026-09-15 22:05:01 | INFO | train rank=0 | step=30190/50000 progress=60.4% loss=0.072627 lr=3.711762e-05 lr_backbone=3.711762e-05 elapsed=31:34:39 avg_step=3.77s recent_step=3.30s eta=20:43:13 +2026-09-15 22:05:56 | INFO | train rank=0 | step=30200/50000 progress=60.4% loss=0.098728 lr=3.708567e-05 lr_backbone=3.708567e-05 elapsed=31:35:34 avg_step=3.77s recent_step=5.49s eta=20:42:47 +2026-09-15 22:06:26 | INFO | train rank=0 | step=30210/50000 progress=60.4% loss=0.084205 lr=3.705372e-05 lr_backbone=3.705372e-05 elapsed=31:36:04 avg_step=3.77s recent_step=3.00s eta=20:42:04 +2026-09-15 22:07:05 | INFO | train rank=0 | step=30220/50000 progress=60.4% loss=0.099299 lr=3.702179e-05 lr_backbone=3.702179e-05 elapsed=31:36:43 avg_step=3.77s recent_step=3.93s eta=20:41:28 +2026-09-15 22:07:42 | INFO | train rank=0 | step=30230/50000 progress=60.5% loss=0.085053 lr=3.698985e-05 lr_backbone=3.698985e-05 elapsed=31:37:20 avg_step=3.77s recent_step=3.70s eta=20:40:50 +2026-09-15 22:08:19 | INFO | train rank=0 | step=30240/50000 progress=60.5% loss=0.091723 lr=3.695792e-05 lr_backbone=3.695792e-05 elapsed=31:37:57 avg_step=3.77s recent_step=3.70s eta=20:40:12 +2026-09-15 22:08:56 | INFO | train rank=0 | step=30250/50000 progress=60.5% loss=0.065235 lr=3.692600e-05 lr_backbone=3.692600e-05 elapsed=31:38:34 avg_step=3.77s recent_step=3.69s eta=20:39:33 +2026-09-15 22:09:31 | INFO | train rank=0 | step=30260/50000 progress=60.5% loss=0.087903 lr=3.689409e-05 lr_backbone=3.689409e-05 elapsed=31:39:09 avg_step=3.77s recent_step=3.49s eta=20:38:54 +2026-09-15 22:10:02 | INFO | train rank=0 | step=30270/50000 progress=60.5% loss=0.085402 lr=3.686218e-05 lr_backbone=3.686218e-05 elapsed=31:39:40 avg_step=3.77s recent_step=3.16s eta=20:38:12 +2026-09-15 22:10:45 | INFO | train rank=0 | step=30280/50000 progress=60.6% loss=0.088854 lr=3.683027e-05 lr_backbone=3.683027e-05 elapsed=31:40:23 avg_step=3.77s recent_step=4.23s eta=20:37:38 +2026-09-15 22:11:18 | INFO | train rank=0 | step=30290/50000 progress=60.6% loss=0.088598 lr=3.679837e-05 lr_backbone=3.679837e-05 elapsed=31:40:56 avg_step=3.77s recent_step=3.33s eta=20:36:57 +2026-09-15 22:11:56 | INFO | train rank=0 | step=30300/50000 progress=60.6% loss=0.085285 lr=3.676648e-05 lr_backbone=3.676648e-05 elapsed=31:41:34 avg_step=3.77s recent_step=3.76s eta=20:36:20 +2026-09-15 22:12:25 | INFO | train rank=0 | step=30310/50000 progress=60.6% loss=0.087648 lr=3.673459e-05 lr_backbone=3.673459e-05 elapsed=31:42:03 avg_step=3.77s recent_step=2.98s eta=20:35:37 +2026-09-15 22:13:18 | INFO | train rank=0 | step=30320/50000 progress=60.6% loss=0.078508 lr=3.670271e-05 lr_backbone=3.670271e-05 elapsed=31:42:56 avg_step=3.77s recent_step=5.29s eta=20:35:09 +2026-09-15 22:13:55 | INFO | train rank=0 | step=30330/50000 progress=60.7% loss=0.100667 lr=3.667084e-05 lr_backbone=3.667084e-05 elapsed=31:43:33 avg_step=3.77s recent_step=3.63s eta=20:34:30 +2026-09-15 22:14:35 | INFO | train rank=0 | step=30340/50000 progress=60.7% loss=0.076010 lr=3.663897e-05 lr_backbone=3.663897e-05 elapsed=31:44:13 avg_step=3.77s recent_step=4.00s eta=20:33:54 +2026-09-15 22:15:07 | INFO | train rank=0 | step=30350/50000 progress=60.7% loss=0.090746 lr=3.660710e-05 lr_backbone=3.660710e-05 elapsed=31:44:45 avg_step=3.77s recent_step=3.24s eta=20:33:13 +2026-09-15 22:15:57 | INFO | train rank=0 | step=30360/50000 progress=60.7% loss=0.108543 lr=3.657524e-05 lr_backbone=3.657524e-05 elapsed=31:45:35 avg_step=3.77s recent_step=4.96s eta=20:32:43 +2026-09-15 22:16:27 | INFO | train rank=0 | step=30370/50000 progress=60.7% loss=0.098674 lr=3.654339e-05 lr_backbone=3.654339e-05 elapsed=31:46:05 avg_step=3.77s recent_step=3.07s eta=20:32:01 +2026-09-15 22:17:05 | INFO | train rank=0 | step=30380/50000 progress=60.8% loss=0.082897 lr=3.651155e-05 lr_backbone=3.651155e-05 elapsed=31:46:43 avg_step=3.77s recent_step=3.75s eta=20:31:23 +2026-09-15 22:17:37 | INFO | train rank=0 | step=30390/50000 progress=60.8% loss=0.102338 lr=3.647971e-05 lr_backbone=3.647971e-05 elapsed=31:47:15 avg_step=3.77s recent_step=3.22s eta=20:30:42 +2026-09-15 22:18:28 | INFO | train rank=0 | step=30400/50000 progress=60.8% loss=0.089311 lr=3.644787e-05 lr_backbone=3.644787e-05 elapsed=31:48:06 avg_step=3.77s recent_step=5.11s eta=20:30:13 +2026-09-15 22:18:57 | INFO | train rank=0 | step=30410/50000 progress=60.8% loss=0.083255 lr=3.641604e-05 lr_backbone=3.641604e-05 elapsed=31:48:35 avg_step=3.77s recent_step=2.90s eta=20:29:30 +2026-09-15 22:19:23 | INFO | train rank=0 | step=30420/50000 progress=60.8% loss=0.091961 lr=3.638422e-05 lr_backbone=3.638422e-05 elapsed=31:49:01 avg_step=3.77s recent_step=2.55s eta=20:28:44 +2026-09-15 22:19:54 | INFO | train rank=0 | step=30430/50000 progress=60.9% loss=0.081423 lr=3.635240e-05 lr_backbone=3.635240e-05 elapsed=31:49:32 avg_step=3.77s recent_step=3.18s eta=20:28:03 +2026-09-15 22:20:38 | INFO | train rank=0 | step=30440/50000 progress=60.9% loss=0.088853 lr=3.632059e-05 lr_backbone=3.632059e-05 elapsed=31:50:16 avg_step=3.77s recent_step=4.38s eta=20:27:29 +2026-09-15 22:21:19 | INFO | train rank=0 | step=30450/50000 progress=60.9% loss=0.084943 lr=3.628879e-05 lr_backbone=3.628879e-05 elapsed=31:50:57 avg_step=3.77s recent_step=4.08s eta=20:26:54 +2026-09-15 22:21:51 | INFO | train rank=0 | step=30460/50000 progress=60.9% loss=0.095199 lr=3.625699e-05 lr_backbone=3.625699e-05 elapsed=31:51:29 avg_step=3.77s recent_step=3.21s eta=20:26:12 +2026-09-15 22:22:27 | INFO | train rank=0 | step=30470/50000 progress=60.9% loss=0.087894 lr=3.622520e-05 lr_backbone=3.622520e-05 elapsed=31:52:05 avg_step=3.77s recent_step=3.64s eta=20:25:34 +2026-09-15 22:23:25 | INFO | train rank=0 | step=30480/50000 progress=61.0% loss=0.089696 lr=3.619341e-05 lr_backbone=3.619341e-05 elapsed=31:53:03 avg_step=3.77s recent_step=5.73s eta=20:25:09 +2026-09-15 22:23:56 | INFO | train rank=0 | step=30490/50000 progress=61.0% loss=0.085703 lr=3.616163e-05 lr_backbone=3.616163e-05 elapsed=31:53:34 avg_step=3.77s recent_step=3.15s eta=20:24:27 +2026-09-15 22:24:29 | INFO | train rank=0 | step=30500/50000 progress=61.0% loss=0.088656 lr=3.612985e-05 lr_backbone=3.612985e-05 elapsed=31:54:07 avg_step=3.77s recent_step=3.25s eta=20:23:46 +2026-09-15 22:25:00 | INFO | train rank=0 | step=30510/50000 progress=61.0% loss=0.067268 lr=3.609809e-05 lr_backbone=3.609809e-05 elapsed=31:54:38 avg_step=3.77s recent_step=3.16s eta=20:23:05 +2026-09-15 22:25:44 | INFO | train rank=0 | step=30520/50000 progress=61.0% loss=0.095239 lr=3.606632e-05 lr_backbone=3.606632e-05 elapsed=31:55:22 avg_step=3.77s recent_step=4.33s eta=20:22:31 +2026-09-15 22:26:15 | INFO | train rank=0 | step=30530/50000 progress=61.1% loss=0.107977 lr=3.603457e-05 lr_backbone=3.603457e-05 elapsed=31:55:53 avg_step=3.77s recent_step=3.13s eta=20:21:49 +2026-09-15 22:26:47 | INFO | train rank=0 | step=30540/50000 progress=61.1% loss=0.090448 lr=3.600282e-05 lr_backbone=3.600282e-05 elapsed=31:56:25 avg_step=3.77s recent_step=3.19s eta=20:21:08 +2026-09-15 22:27:19 | INFO | train rank=0 | step=30550/50000 progress=61.1% loss=0.069959 lr=3.597107e-05 lr_backbone=3.597107e-05 elapsed=31:56:57 avg_step=3.76s recent_step=3.22s eta=20:20:27 +2026-09-15 22:28:04 | INFO | train rank=0 | step=30560/50000 progress=61.1% loss=0.081294 lr=3.593934e-05 lr_backbone=3.593934e-05 elapsed=31:57:42 avg_step=3.77s recent_step=4.51s eta=20:19:54 +2026-09-15 22:28:35 | INFO | train rank=0 | step=30570/50000 progress=61.1% loss=0.076683 lr=3.590760e-05 lr_backbone=3.590760e-05 elapsed=31:58:13 avg_step=3.76s recent_step=3.13s eta=20:19:12 +2026-09-15 22:29:13 | INFO | train rank=0 | step=30580/50000 progress=61.2% loss=0.103779 lr=3.587588e-05 lr_backbone=3.587588e-05 elapsed=31:58:51 avg_step=3.76s recent_step=3.78s eta=20:18:34 +2026-09-15 22:29:40 | INFO | train rank=0 | step=30590/50000 progress=61.2% loss=0.090126 lr=3.584416e-05 lr_backbone=3.584416e-05 elapsed=31:59:18 avg_step=3.76s recent_step=2.73s eta=20:17:50 +2026-09-15 22:30:26 | INFO | train rank=0 | step=30600/50000 progress=61.2% loss=0.082473 lr=3.581244e-05 lr_backbone=3.581244e-05 elapsed=32:00:04 avg_step=3.76s recent_step=4.60s eta=20:17:18 +2026-09-15 22:30:58 | INFO | train rank=0 | step=30610/50000 progress=61.2% loss=0.071540 lr=3.578074e-05 lr_backbone=3.578074e-05 elapsed=32:00:36 avg_step=3.76s recent_step=3.17s eta=20:16:36 +2026-09-15 22:31:38 | INFO | train rank=0 | step=30620/50000 progress=61.2% loss=0.099564 lr=3.574904e-05 lr_backbone=3.574904e-05 elapsed=32:01:16 avg_step=3.76s recent_step=4.01s eta=20:16:00 +2026-09-15 22:32:12 | INFO | train rank=0 | step=30630/50000 progress=61.3% loss=0.086992 lr=3.571734e-05 lr_backbone=3.571734e-05 elapsed=32:01:50 avg_step=3.76s recent_step=3.39s eta=20:15:20 +2026-09-15 22:32:55 | INFO | train rank=0 | step=30640/50000 progress=61.3% loss=0.100341 lr=3.568565e-05 lr_backbone=3.568565e-05 elapsed=32:02:33 avg_step=3.76s recent_step=4.31s eta=20:14:46 +2026-09-15 22:33:25 | INFO | train rank=0 | step=30650/50000 progress=61.3% loss=0.095614 lr=3.565397e-05 lr_backbone=3.565397e-05 elapsed=32:03:03 avg_step=3.76s recent_step=3.03s eta=20:14:04 +2026-09-15 22:33:58 | INFO | train rank=0 | step=30660/50000 progress=61.3% loss=0.087330 lr=3.562230e-05 lr_backbone=3.562230e-05 elapsed=32:03:36 avg_step=3.76s recent_step=3.27s eta=20:13:23 +2026-09-15 22:34:38 | INFO | train rank=0 | step=30670/50000 progress=61.3% loss=0.097235 lr=3.559063e-05 lr_backbone=3.559063e-05 elapsed=32:04:16 avg_step=3.76s recent_step=4.03s eta=20:12:47 +2026-09-15 22:35:36 | INFO | train rank=0 | step=30680/50000 progress=61.4% loss=0.078887 lr=3.555896e-05 lr_backbone=3.555896e-05 elapsed=32:05:14 avg_step=3.77s recent_step=5.74s eta=20:12:22 +2026-09-15 22:35:59 | INFO | train rank=0 | step=30690/50000 progress=61.4% loss=0.105213 lr=3.552731e-05 lr_backbone=3.552731e-05 elapsed=32:05:37 avg_step=3.76s recent_step=2.34s eta=20:11:35 +2026-09-15 22:36:39 | INFO | train rank=0 | step=30700/50000 progress=61.4% loss=0.093873 lr=3.549566e-05 lr_backbone=3.549566e-05 elapsed=32:06:17 avg_step=3.76s recent_step=3.99s eta=20:10:59 +2026-09-15 22:37:14 | INFO | train rank=0 | step=30710/50000 progress=61.4% loss=0.094944 lr=3.546401e-05 lr_backbone=3.546401e-05 elapsed=32:06:52 avg_step=3.76s recent_step=3.48s eta=20:10:20 +2026-09-15 22:38:12 | INFO | train rank=0 | step=30720/50000 progress=61.4% loss=0.085501 lr=3.543237e-05 lr_backbone=3.543237e-05 elapsed=32:07:50 avg_step=3.77s recent_step=5.79s eta=20:09:55 +2026-09-15 22:38:47 | INFO | train rank=0 | step=30730/50000 progress=61.5% loss=0.092853 lr=3.540074e-05 lr_backbone=3.540074e-05 elapsed=32:08:25 avg_step=3.77s recent_step=3.48s eta=20:09:15 +2026-09-15 22:39:21 | INFO | train rank=0 | step=30740/50000 progress=61.5% loss=0.089087 lr=3.536912e-05 lr_backbone=3.536912e-05 elapsed=32:08:59 avg_step=3.77s recent_step=3.40s eta=20:08:35 +2026-09-15 22:39:52 | INFO | train rank=0 | step=30750/50000 progress=61.5% loss=0.104685 lr=3.533750e-05 lr_backbone=3.533750e-05 elapsed=32:09:30 avg_step=3.76s recent_step=3.15s eta=20:07:54 +2026-09-15 22:40:43 | INFO | train rank=0 | step=30760/50000 progress=61.5% loss=0.078303 lr=3.530589e-05 lr_backbone=3.530589e-05 elapsed=32:10:21 avg_step=3.77s recent_step=5.08s eta=20:07:24 +2026-09-15 22:41:15 | INFO | train rank=0 | step=30770/50000 progress=61.5% loss=0.112510 lr=3.527428e-05 lr_backbone=3.527428e-05 elapsed=32:10:53 avg_step=3.77s recent_step=3.21s eta=20:06:43 +2026-09-15 22:41:51 | INFO | train rank=0 | step=30780/50000 progress=61.6% loss=0.109120 lr=3.524268e-05 lr_backbone=3.524268e-05 elapsed=32:11:29 avg_step=3.77s recent_step=3.61s eta=20:06:05 +2026-09-15 22:42:30 | INFO | train rank=0 | step=30790/50000 progress=61.6% loss=0.071883 lr=3.521109e-05 lr_backbone=3.521109e-05 elapsed=32:12:08 avg_step=3.77s recent_step=3.87s eta=20:05:28 +2026-09-15 22:43:16 | INFO | train rank=0 | step=30800/50000 progress=61.6% loss=0.118463 lr=3.517950e-05 lr_backbone=3.517950e-05 elapsed=32:12:54 avg_step=3.77s recent_step=4.58s eta=20:04:55 +2026-09-15 22:43:54 | INFO | train rank=0 | step=30810/50000 progress=61.6% loss=0.109390 lr=3.514792e-05 lr_backbone=3.514792e-05 elapsed=32:13:32 avg_step=3.77s recent_step=3.86s eta=20:04:18 +2026-09-15 22:44:27 | INFO | train rank=0 | step=30820/50000 progress=61.6% loss=0.117591 lr=3.511635e-05 lr_backbone=3.511635e-05 elapsed=32:14:05 avg_step=3.77s recent_step=3.26s eta=20:03:37 +2026-09-15 22:45:05 | INFO | train rank=0 | step=30830/50000 progress=61.7% loss=0.091614 lr=3.508478e-05 lr_backbone=3.508478e-05 elapsed=32:14:43 avg_step=3.77s recent_step=3.77s eta=20:03:00 +2026-09-15 22:45:51 | INFO | train rank=0 | step=30840/50000 progress=61.7% loss=0.083747 lr=3.505322e-05 lr_backbone=3.505322e-05 elapsed=32:15:29 avg_step=3.77s recent_step=4.62s eta=20:02:27 +2026-09-15 22:46:22 | INFO | train rank=0 | step=30850/50000 progress=61.7% loss=0.087882 lr=3.502167e-05 lr_backbone=3.502167e-05 elapsed=32:16:00 avg_step=3.77s recent_step=3.17s eta=20:01:46 +2026-09-15 22:46:54 | INFO | train rank=0 | step=30860/50000 progress=61.7% loss=0.092049 lr=3.499012e-05 lr_backbone=3.499012e-05 elapsed=32:16:32 avg_step=3.77s recent_step=3.16s eta=20:01:05 +2026-09-15 22:47:34 | INFO | train rank=0 | step=30870/50000 progress=61.7% loss=0.091605 lr=3.495858e-05 lr_backbone=3.495858e-05 elapsed=32:17:12 avg_step=3.77s recent_step=4.01s eta=20:00:28 +2026-09-15 22:48:28 | INFO | train rank=0 | step=30880/50000 progress=61.8% loss=0.085226 lr=3.492704e-05 lr_backbone=3.492704e-05 elapsed=32:18:06 avg_step=3.77s recent_step=5.35s eta=20:00:01 +2026-09-15 22:49:07 | INFO | train rank=0 | step=30890/50000 progress=61.8% loss=0.103357 lr=3.489552e-05 lr_backbone=3.489552e-05 elapsed=32:18:45 avg_step=3.77s recent_step=3.90s eta=19:59:24 +2026-09-15 22:49:31 | INFO | train rank=0 | step=30900/50000 progress=61.8% loss=0.084069 lr=3.486399e-05 lr_backbone=3.486399e-05 elapsed=32:19:09 avg_step=3.77s recent_step=2.44s eta=19:58:38 +2026-09-15 22:50:10 | INFO | train rank=0 | step=30910/50000 progress=61.8% loss=0.086676 lr=3.483248e-05 lr_backbone=3.483248e-05 elapsed=32:19:48 avg_step=3.77s recent_step=3.85s eta=19:58:01 +2026-09-15 22:51:01 | INFO | train rank=0 | step=30920/50000 progress=61.8% loss=0.080588 lr=3.480097e-05 lr_backbone=3.480097e-05 elapsed=32:20:39 avg_step=3.77s recent_step=5.16s eta=19:57:32 +2026-09-15 22:51:29 | INFO | train rank=0 | step=30930/50000 progress=61.9% loss=0.090166 lr=3.476947e-05 lr_backbone=3.476947e-05 elapsed=32:21:07 avg_step=3.77s recent_step=2.83s eta=19:56:48 +2026-09-15 22:51:58 | INFO | train rank=0 | step=30940/50000 progress=61.9% loss=0.077414 lr=3.473798e-05 lr_backbone=3.473798e-05 elapsed=32:21:36 avg_step=3.77s recent_step=2.85s eta=19:56:05 +2026-09-15 22:52:31 | INFO | train rank=0 | step=30950/50000 progress=61.9% loss=0.073872 lr=3.470649e-05 lr_backbone=3.470649e-05 elapsed=32:22:09 avg_step=3.77s recent_step=3.28s eta=19:55:24 +2026-09-15 22:53:25 | INFO | train rank=0 | step=30960/50000 progress=61.9% loss=0.089670 lr=3.467501e-05 lr_backbone=3.467501e-05 elapsed=32:23:03 avg_step=3.77s recent_step=5.41s eta=19:54:57 +2026-09-15 22:54:01 | INFO | train rank=0 | step=30970/50000 progress=61.9% loss=0.102593 lr=3.464353e-05 lr_backbone=3.464353e-05 elapsed=32:23:39 avg_step=3.77s recent_step=3.66s eta=19:54:18 +2026-09-15 22:54:28 | INFO | train rank=0 | step=30980/50000 progress=62.0% loss=0.121365 lr=3.461207e-05 lr_backbone=3.461207e-05 elapsed=32:24:06 avg_step=3.77s recent_step=2.63s eta=19:53:34 +2026-09-15 22:54:59 | INFO | train rank=0 | step=30990/50000 progress=62.0% loss=0.088349 lr=3.458060e-05 lr_backbone=3.458060e-05 elapsed=32:24:37 avg_step=3.76s recent_step=3.10s eta=19:52:52 +2026-09-15 22:55:36 | INFO | train rank=0 | step=31000/50000 progress=62.0% loss=0.095248 lr=3.454915e-05 lr_backbone=3.454915e-05 elapsed=32:25:14 avg_step=3.76s recent_step=3.76s eta=19:52:14 +2026-09-15 22:56:09 | INFO | train rank=0 | step=31010/50000 progress=62.0% loss=0.087520 lr=3.451770e-05 lr_backbone=3.451770e-05 elapsed=32:25:47 avg_step=3.76s recent_step=3.23s eta=19:51:34 +2026-09-15 22:56:43 | INFO | train rank=0 | step=31020/50000 progress=62.0% loss=0.092137 lr=3.448626e-05 lr_backbone=3.448626e-05 elapsed=32:26:21 avg_step=3.76s recent_step=3.43s eta=19:50:54 +2026-09-15 22:57:18 | INFO | train rank=0 | step=31030/50000 progress=62.1% loss=0.088189 lr=3.445483e-05 lr_backbone=3.445483e-05 elapsed=32:26:56 avg_step=3.76s recent_step=3.47s eta=19:50:14 +2026-09-15 22:58:14 | INFO | train rank=0 | step=31040/50000 progress=62.1% loss=0.090166 lr=3.442340e-05 lr_backbone=3.442340e-05 elapsed=32:27:52 avg_step=3.77s recent_step=5.64s eta=19:49:48 +2026-09-15 22:58:45 | INFO | train rank=0 | step=31050/50000 progress=62.1% loss=0.106050 lr=3.439198e-05 lr_backbone=3.439198e-05 elapsed=32:28:23 avg_step=3.76s recent_step=3.05s eta=19:49:06 +2026-09-15 22:59:21 | INFO | train rank=0 | step=31060/50000 progress=62.1% loss=0.074435 lr=3.436057e-05 lr_backbone=3.436057e-05 elapsed=32:28:59 avg_step=3.76s recent_step=3.61s eta=19:48:28 +2026-09-15 22:59:55 | INFO | train rank=0 | step=31070/50000 progress=62.1% loss=0.082102 lr=3.432916e-05 lr_backbone=3.432916e-05 elapsed=32:29:33 avg_step=3.76s recent_step=3.39s eta=19:47:48 +2026-09-15 23:00:53 | INFO | train rank=0 | step=31080/50000 progress=62.2% loss=0.092296 lr=3.429776e-05 lr_backbone=3.429776e-05 elapsed=32:30:31 avg_step=3.77s recent_step=5.88s eta=19:47:23 +2026-09-15 23:01:25 | INFO | train rank=0 | step=31090/50000 progress=62.2% loss=0.089243 lr=3.426637e-05 lr_backbone=3.426637e-05 elapsed=32:31:03 avg_step=3.77s recent_step=3.19s eta=19:46:42 +2026-09-15 23:02:04 | INFO | train rank=0 | step=31100/50000 progress=62.2% loss=0.104143 lr=3.423498e-05 lr_backbone=3.423498e-05 elapsed=32:31:42 avg_step=3.77s recent_step=3.87s eta=19:46:05 +2026-09-15 23:02:29 | INFO | train rank=0 | step=31110/50000 progress=62.2% loss=0.083224 lr=3.420360e-05 lr_backbone=3.420360e-05 elapsed=32:32:07 avg_step=3.76s recent_step=2.47s eta=19:45:19 +2026-09-15 23:03:17 | INFO | train rank=0 | step=31120/50000 progress=62.2% loss=0.085042 lr=3.417223e-05 lr_backbone=3.417223e-05 elapsed=32:32:55 avg_step=3.77s recent_step=4.81s eta=19:44:48 +2026-09-15 23:03:44 | INFO | train rank=0 | step=31130/50000 progress=62.3% loss=0.090207 lr=3.414087e-05 lr_backbone=3.414087e-05 elapsed=32:33:22 avg_step=3.76s recent_step=2.72s eta=19:44:04 +2026-09-15 23:04:20 | INFO | train rank=0 | step=31140/50000 progress=62.3% loss=0.071522 lr=3.410951e-05 lr_backbone=3.410951e-05 elapsed=32:33:58 avg_step=3.76s recent_step=3.62s eta=19:43:25 +2026-09-15 23:05:00 | INFO | train rank=0 | step=31150/50000 progress=62.3% loss=0.093518 lr=3.407816e-05 lr_backbone=3.407816e-05 elapsed=32:34:38 avg_step=3.76s recent_step=3.96s eta=19:42:49 +2026-09-15 23:05:41 | INFO | train rank=0 | step=31160/50000 progress=62.3% loss=0.088413 lr=3.404681e-05 lr_backbone=3.404681e-05 elapsed=32:35:19 avg_step=3.77s recent_step=4.08s eta=19:42:13 +2026-09-15 23:06:06 | INFO | train rank=0 | step=31170/50000 progress=62.3% loss=0.100167 lr=3.401547e-05 lr_backbone=3.401547e-05 elapsed=32:35:44 avg_step=3.76s recent_step=2.56s eta=19:41:28 +2026-09-15 23:06:37 | INFO | train rank=0 | step=31180/50000 progress=62.4% loss=0.086756 lr=3.398414e-05 lr_backbone=3.398414e-05 elapsed=32:36:15 avg_step=3.76s recent_step=3.14s eta=19:40:47 +2026-09-15 23:06:59 | INFO | train rank=0 | step=31190/50000 progress=62.4% loss=0.071780 lr=3.395282e-05 lr_backbone=3.395282e-05 elapsed=32:36:37 avg_step=3.76s recent_step=2.15s eta=19:39:59 +2026-09-15 23:07:58 | INFO | train rank=0 | step=31200/50000 progress=62.4% loss=0.098013 lr=3.392150e-05 lr_backbone=3.392150e-05 elapsed=32:37:36 avg_step=3.76s recent_step=5.86s eta=19:39:34 +2026-09-15 23:08:34 | INFO | train rank=0 | step=31210/50000 progress=62.4% loss=0.082194 lr=3.389019e-05 lr_backbone=3.389019e-05 elapsed=32:38:12 avg_step=3.76s recent_step=3.66s eta=19:38:56 +2026-09-15 23:09:13 | INFO | train rank=0 | step=31220/50000 progress=62.4% loss=0.078005 lr=3.385889e-05 lr_backbone=3.385889e-05 elapsed=32:38:51 avg_step=3.76s recent_step=3.86s eta=19:38:19 +2026-09-15 23:09:46 | INFO | train rank=0 | step=31230/50000 progress=62.5% loss=0.080901 lr=3.382760e-05 lr_backbone=3.382760e-05 elapsed=32:39:24 avg_step=3.76s recent_step=3.34s eta=19:37:39 +2026-09-15 23:10:51 | INFO | train rank=0 | step=31240/50000 progress=62.5% loss=0.090489 lr=3.379631e-05 lr_backbone=3.379631e-05 elapsed=32:40:29 avg_step=3.77s recent_step=6.46s eta=19:37:17 +2026-09-15 23:11:18 | INFO | train rank=0 | step=31250/50000 progress=62.5% loss=0.077983 lr=3.376503e-05 lr_backbone=3.376503e-05 elapsed=32:40:56 avg_step=3.76s recent_step=2.69s eta=19:36:33 +2026-09-15 23:11:53 | INFO | train rank=0 | step=31260/50000 progress=62.5% loss=0.077898 lr=3.373375e-05 lr_backbone=3.373375e-05 elapsed=32:41:31 avg_step=3.76s recent_step=3.55s eta=19:35:54 +2026-09-15 23:12:34 | INFO | train rank=0 | step=31270/50000 progress=62.5% loss=0.078873 lr=3.370249e-05 lr_backbone=3.370249e-05 elapsed=32:42:12 avg_step=3.77s recent_step=4.06s eta=19:35:18 +2026-09-15 23:13:18 | INFO | train rank=0 | step=31280/50000 progress=62.6% loss=0.093048 lr=3.367123e-05 lr_backbone=3.367123e-05 elapsed=32:42:56 avg_step=3.77s recent_step=4.45s eta=19:34:45 +2026-09-15 23:13:49 | INFO | train rank=0 | step=31290/50000 progress=62.6% loss=0.105461 lr=3.363997e-05 lr_backbone=3.363997e-05 elapsed=32:43:27 avg_step=3.77s recent_step=3.07s eta=19:34:03 +2026-09-15 23:14:21 | INFO | train rank=0 | step=31300/50000 progress=62.6% loss=0.101826 lr=3.360873e-05 lr_backbone=3.360873e-05 elapsed=32:43:59 avg_step=3.76s recent_step=3.21s eta=19:33:22 +2026-09-15 23:15:06 | INFO | train rank=0 | step=31310/50000 progress=62.6% loss=0.109258 lr=3.357749e-05 lr_backbone=3.357749e-05 elapsed=32:44:44 avg_step=3.77s recent_step=4.53s eta=19:32:49 +2026-09-15 23:15:52 | INFO | train rank=0 | step=31320/50000 progress=62.6% loss=0.075815 lr=3.354626e-05 lr_backbone=3.354626e-05 elapsed=32:45:30 avg_step=3.77s recent_step=4.59s eta=19:32:16 +2026-09-15 23:16:23 | INFO | train rank=0 | step=31330/50000 progress=62.7% loss=0.089726 lr=3.351503e-05 lr_backbone=3.351503e-05 elapsed=32:46:01 avg_step=3.77s recent_step=3.05s eta=19:31:34 +2026-09-15 23:16:59 | INFO | train rank=0 | step=31340/50000 progress=62.7% loss=0.099374 lr=3.348382e-05 lr_backbone=3.348382e-05 elapsed=32:46:37 avg_step=3.77s recent_step=3.65s eta=19:30:56 +2026-09-15 23:17:31 | INFO | train rank=0 | step=31350/50000 progress=62.7% loss=0.085440 lr=3.345261e-05 lr_backbone=3.345261e-05 elapsed=32:47:09 avg_step=3.76s recent_step=3.17s eta=19:30:15 +2026-09-15 23:18:09 | INFO | train rank=0 | step=31360/50000 progress=62.7% loss=0.091928 lr=3.342141e-05 lr_backbone=3.342141e-05 elapsed=32:47:47 avg_step=3.76s recent_step=3.83s eta=19:29:38 +2026-09-15 23:18:43 | INFO | train rank=0 | step=31370/50000 progress=62.7% loss=0.076732 lr=3.339021e-05 lr_backbone=3.339021e-05 elapsed=32:48:21 avg_step=3.76s recent_step=3.42s eta=19:28:58 +2026-09-15 23:19:13 | INFO | train rank=0 | step=31380/50000 progress=62.8% loss=0.094348 lr=3.335902e-05 lr_backbone=3.335902e-05 elapsed=32:48:51 avg_step=3.76s recent_step=2.98s eta=19:28:16 +2026-09-15 23:19:50 | INFO | train rank=0 | step=31390/50000 progress=62.8% loss=0.067670 lr=3.332784e-05 lr_backbone=3.332784e-05 elapsed=32:49:28 avg_step=3.76s recent_step=3.65s eta=19:27:37 +2026-09-15 23:20:39 | INFO | train rank=0 | step=31400/50000 progress=62.8% loss=0.098216 lr=3.329667e-05 lr_backbone=3.329667e-05 elapsed=32:50:17 avg_step=3.76s recent_step=4.90s eta=19:27:06 +2026-09-15 23:21:12 | INFO | train rank=0 | step=31410/50000 progress=62.8% loss=0.073277 lr=3.326550e-05 lr_backbone=3.326550e-05 elapsed=32:50:50 avg_step=3.76s recent_step=3.34s eta=19:26:26 +2026-09-15 23:21:47 | INFO | train rank=0 | step=31420/50000 progress=62.8% loss=0.108448 lr=3.323434e-05 lr_backbone=3.323434e-05 elapsed=32:51:25 avg_step=3.76s recent_step=3.46s eta=19:25:47 +2026-09-15 23:22:25 | INFO | train rank=0 | step=31430/50000 progress=62.9% loss=0.093273 lr=3.320319e-05 lr_backbone=3.320319e-05 elapsed=32:52:03 avg_step=3.76s recent_step=3.78s eta=19:25:09 +2026-09-15 23:23:14 | INFO | train rank=0 | step=31440/50000 progress=62.9% loss=0.079057 lr=3.317205e-05 lr_backbone=3.317205e-05 elapsed=32:52:52 avg_step=3.77s recent_step=4.99s eta=19:24:39 +2026-09-15 23:23:48 | INFO | train rank=0 | step=31450/50000 progress=62.9% loss=0.097505 lr=3.314091e-05 lr_backbone=3.314091e-05 elapsed=32:53:26 avg_step=3.76s recent_step=3.38s eta=19:23:59 +2026-09-15 23:24:21 | INFO | train rank=0 | step=31460/50000 progress=62.9% loss=0.081264 lr=3.310978e-05 lr_backbone=3.310978e-05 elapsed=32:53:59 avg_step=3.76s recent_step=3.27s eta=19:23:18 +2026-09-15 23:24:56 | INFO | train rank=0 | step=31470/50000 progress=62.9% loss=0.102982 lr=3.307866e-05 lr_backbone=3.307866e-05 elapsed=32:54:34 avg_step=3.76s recent_step=3.48s eta=19:22:39 +2026-09-15 23:25:46 | INFO | train rank=0 | step=31480/50000 progress=63.0% loss=0.091665 lr=3.304755e-05 lr_backbone=3.304755e-05 elapsed=32:55:24 avg_step=3.77s recent_step=5.00s eta=19:22:09 +2026-09-15 23:26:22 | INFO | train rank=0 | step=31490/50000 progress=63.0% loss=0.106517 lr=3.301644e-05 lr_backbone=3.301644e-05 elapsed=32:56:00 avg_step=3.77s recent_step=3.57s eta=19:21:30 +2026-09-15 23:26:53 | INFO | train rank=0 | step=31500/50000 progress=63.0% loss=0.079402 lr=3.298534e-05 lr_backbone=3.298534e-05 elapsed=32:56:31 avg_step=3.76s recent_step=3.18s eta=19:20:49 +2026-09-15 23:27:31 | INFO | train rank=0 | step=31510/50000 progress=63.0% loss=0.079039 lr=3.295425e-05 lr_backbone=3.295425e-05 elapsed=32:57:09 avg_step=3.76s recent_step=3.80s eta=19:20:11 +2026-09-15 23:28:19 | INFO | train rank=0 | step=31520/50000 progress=63.0% loss=0.077215 lr=3.292316e-05 lr_backbone=3.292316e-05 elapsed=32:57:57 avg_step=3.77s recent_step=4.76s eta=19:19:39 +2026-09-15 23:28:50 | INFO | train rank=0 | step=31530/50000 progress=63.1% loss=0.117814 lr=3.289209e-05 lr_backbone=3.289209e-05 elapsed=32:58:28 avg_step=3.76s recent_step=3.08s eta=19:18:58 +2026-09-15 23:29:27 | INFO | train rank=0 | step=31540/50000 progress=63.1% loss=0.143989 lr=3.286102e-05 lr_backbone=3.286102e-05 elapsed=32:59:05 avg_step=3.76s recent_step=3.70s eta=19:18:20 +2026-09-15 23:29:53 | INFO | train rank=0 | step=31550/50000 progress=63.1% loss=0.085814 lr=3.282996e-05 lr_backbone=3.282996e-05 elapsed=32:59:31 avg_step=3.76s recent_step=2.58s eta=19:17:35 +2026-09-15 23:30:33 | INFO | train rank=0 | step=31560/50000 progress=63.1% loss=0.077835 lr=3.279890e-05 lr_backbone=3.279890e-05 elapsed=33:00:11 avg_step=3.76s recent_step=4.04s eta=19:16:59 +2026-09-15 23:31:13 | INFO | train rank=0 | step=31570/50000 progress=63.1% loss=0.067999 lr=3.276785e-05 lr_backbone=3.276785e-05 elapsed=33:00:51 avg_step=3.76s recent_step=4.04s eta=19:16:23 +2026-09-15 23:31:44 | INFO | train rank=0 | step=31580/50000 progress=63.2% loss=0.100236 lr=3.273681e-05 lr_backbone=3.273681e-05 elapsed=33:01:22 avg_step=3.76s recent_step=3.09s eta=19:15:41 +2026-09-15 23:32:18 | INFO | train rank=0 | step=31590/50000 progress=63.2% loss=0.108172 lr=3.270578e-05 lr_backbone=3.270578e-05 elapsed=33:01:56 avg_step=3.76s recent_step=3.42s eta=19:15:02 +2026-09-15 23:33:13 | INFO | train rank=0 | step=31600/50000 progress=63.2% loss=0.069331 lr=3.267476e-05 lr_backbone=3.267476e-05 elapsed=33:02:51 avg_step=3.76s recent_step=5.47s eta=19:14:34 +2026-09-15 23:33:45 | INFO | train rank=0 | step=31610/50000 progress=63.2% loss=0.100161 lr=3.264374e-05 lr_backbone=3.264374e-05 elapsed=33:03:23 avg_step=3.76s recent_step=3.23s eta=19:13:53 +2026-09-15 23:34:22 | INFO | train rank=0 | step=31620/50000 progress=63.2% loss=0.103880 lr=3.261273e-05 lr_backbone=3.261273e-05 elapsed=33:04:00 avg_step=3.76s recent_step=3.65s eta=19:13:15 +2026-09-15 23:34:58 | INFO | train rank=0 | step=31630/50000 progress=63.3% loss=0.088071 lr=3.258173e-05 lr_backbone=3.258173e-05 elapsed=33:04:36 avg_step=3.76s recent_step=3.56s eta=19:12:36 +2026-09-15 23:35:48 | INFO | train rank=0 | step=31640/50000 progress=63.3% loss=0.140188 lr=3.255074e-05 lr_backbone=3.255074e-05 elapsed=33:05:26 avg_step=3.77s recent_step=5.01s eta=19:12:06 +2026-09-15 23:36:22 | INFO | train rank=0 | step=31650/50000 progress=63.3% loss=0.088556 lr=3.251975e-05 lr_backbone=3.251975e-05 elapsed=33:06:00 avg_step=3.76s recent_step=3.44s eta=19:11:26 +2026-09-15 23:36:58 | INFO | train rank=0 | step=31660/50000 progress=63.3% loss=0.103552 lr=3.248877e-05 lr_backbone=3.248877e-05 elapsed=33:06:36 avg_step=3.76s recent_step=3.55s eta=19:10:47 +2026-09-15 23:37:25 | INFO | train rank=0 | step=31670/50000 progress=63.3% loss=0.103284 lr=3.245780e-05 lr_backbone=3.245780e-05 elapsed=33:07:03 avg_step=3.76s recent_step=2.72s eta=19:10:04 +2026-09-15 23:38:02 | INFO | train rank=0 | step=31680/50000 progress=63.4% loss=0.092604 lr=3.242684e-05 lr_backbone=3.242684e-05 elapsed=33:07:40 avg_step=3.76s recent_step=3.74s eta=19:09:26 +2026-09-15 23:38:36 | INFO | train rank=0 | step=31690/50000 progress=63.4% loss=0.093741 lr=3.239588e-05 lr_backbone=3.239588e-05 elapsed=33:08:14 avg_step=3.76s recent_step=3.34s eta=19:08:46 +2026-09-15 23:39:14 | INFO | train rank=0 | step=31700/50000 progress=63.4% loss=0.075600 lr=3.236493e-05 lr_backbone=3.236493e-05 elapsed=33:08:52 avg_step=3.76s recent_step=3.83s eta=19:08:09 +2026-09-15 23:39:50 | INFO | train rank=0 | step=31710/50000 progress=63.4% loss=0.114208 lr=3.233399e-05 lr_backbone=3.233399e-05 elapsed=33:09:28 avg_step=3.76s recent_step=3.60s eta=19:07:30 +2026-09-15 23:40:37 | INFO | train rank=0 | step=31720/50000 progress=63.4% loss=0.116346 lr=3.230306e-05 lr_backbone=3.230306e-05 elapsed=33:10:15 avg_step=3.76s recent_step=4.73s eta=19:06:58 +2026-09-15 23:41:06 | INFO | train rank=0 | step=31730/50000 progress=63.5% loss=0.088272 lr=3.227213e-05 lr_backbone=3.227213e-05 elapsed=33:10:44 avg_step=3.76s recent_step=2.87s eta=19:06:15 +2026-09-15 23:41:37 | INFO | train rank=0 | step=31740/50000 progress=63.5% loss=0.079196 lr=3.224122e-05 lr_backbone=3.224122e-05 elapsed=33:11:15 avg_step=3.76s recent_step=3.12s eta=19:05:34 +2026-09-15 23:42:13 | INFO | train rank=0 | step=31750/50000 progress=63.5% loss=0.084583 lr=3.221031e-05 lr_backbone=3.221031e-05 elapsed=33:11:51 avg_step=3.76s recent_step=3.64s eta=19:04:55 +2026-09-15 23:43:12 | INFO | train rank=0 | step=31760/50000 progress=63.5% loss=0.077065 lr=3.217941e-05 lr_backbone=3.217941e-05 elapsed=33:12:50 avg_step=3.76s recent_step=5.88s eta=19:04:30 +2026-09-15 23:43:48 | INFO | train rank=0 | step=31770/50000 progress=63.5% loss=0.102838 lr=3.214851e-05 lr_backbone=3.214851e-05 elapsed=33:13:26 avg_step=3.76s recent_step=3.59s eta=19:03:51 +2026-09-15 23:44:26 | INFO | train rank=0 | step=31780/50000 progress=63.6% loss=0.083714 lr=3.211763e-05 lr_backbone=3.211763e-05 elapsed=33:14:04 avg_step=3.76s recent_step=3.81s eta=19:03:14 +2026-09-15 23:44:54 | INFO | train rank=0 | step=31790/50000 progress=63.6% loss=0.091762 lr=3.208675e-05 lr_backbone=3.208675e-05 elapsed=33:14:32 avg_step=3.76s recent_step=2.76s eta=19:02:30 +2026-09-15 23:45:40 | INFO | train rank=0 | step=31800/50000 progress=63.6% loss=0.090499 lr=3.205588e-05 lr_backbone=3.205588e-05 elapsed=33:15:18 avg_step=3.76s recent_step=4.62s eta=19:01:58 +2026-09-15 23:46:17 | INFO | train rank=0 | step=31810/50000 progress=63.6% loss=0.098595 lr=3.202502e-05 lr_backbone=3.202502e-05 elapsed=33:15:55 avg_step=3.76s recent_step=3.70s eta=19:01:20 +2026-09-15 23:46:51 | INFO | train rank=0 | step=31820/50000 progress=63.6% loss=0.086689 lr=3.199416e-05 lr_backbone=3.199416e-05 elapsed=33:16:29 avg_step=3.76s recent_step=3.39s eta=19:00:40 +2026-09-15 23:47:19 | INFO | train rank=0 | step=31830/50000 progress=63.7% loss=0.072124 lr=3.196331e-05 lr_backbone=3.196331e-05 elapsed=33:16:57 avg_step=3.76s recent_step=2.79s eta=18:59:57 +2026-09-15 23:48:11 | INFO | train rank=0 | step=31840/50000 progress=63.7% loss=0.091767 lr=3.193248e-05 lr_backbone=3.193248e-05 elapsed=33:17:49 avg_step=3.76s recent_step=5.26s eta=18:59:27 +2026-09-15 23:48:43 | INFO | train rank=0 | step=31850/50000 progress=63.7% loss=0.074146 lr=3.190164e-05 lr_backbone=3.190164e-05 elapsed=33:18:21 avg_step=3.76s recent_step=3.18s eta=18:58:47 +2026-09-15 23:49:19 | INFO | train rank=0 | step=31860/50000 progress=63.7% loss=0.105544 lr=3.187082e-05 lr_backbone=3.187082e-05 elapsed=33:18:57 avg_step=3.76s recent_step=3.56s eta=18:58:08 +2026-09-15 23:49:45 | INFO | train rank=0 | step=31870/50000 progress=63.7% loss=0.112806 lr=3.184001e-05 lr_backbone=3.184001e-05 elapsed=33:19:23 avg_step=3.76s recent_step=2.63s eta=18:57:24 +2026-09-15 23:50:35 | INFO | train rank=0 | step=31880/50000 progress=63.8% loss=0.092880 lr=3.180920e-05 lr_backbone=3.180920e-05 elapsed=33:20:13 avg_step=3.76s recent_step=4.97s eta=18:56:53 +2026-09-15 23:51:12 | INFO | train rank=0 | step=31890/50000 progress=63.8% loss=0.103174 lr=3.177840e-05 lr_backbone=3.177840e-05 elapsed=33:20:50 avg_step=3.76s recent_step=3.75s eta=18:56:15 +2026-09-15 23:51:48 | INFO | train rank=0 | step=31900/50000 progress=63.8% loss=0.104197 lr=3.174761e-05 lr_backbone=3.174761e-05 elapsed=33:21:26 avg_step=3.76s recent_step=3.55s eta=18:55:36 +2026-09-15 23:52:20 | INFO | train rank=0 | step=31910/50000 progress=63.8% loss=0.089562 lr=3.171683e-05 lr_backbone=3.171683e-05 elapsed=33:21:58 avg_step=3.76s recent_step=3.20s eta=18:54:55 +2026-09-15 23:53:11 | INFO | train rank=0 | step=31920/50000 progress=63.8% loss=0.089479 lr=3.168605e-05 lr_backbone=3.168605e-05 elapsed=33:22:49 avg_step=3.76s recent_step=5.11s eta=18:54:25 +2026-09-15 23:53:40 | INFO | train rank=0 | step=31930/50000 progress=63.9% loss=0.092575 lr=3.165528e-05 lr_backbone=3.165528e-05 elapsed=33:23:18 avg_step=3.76s recent_step=2.94s eta=18:53:43 +2026-09-15 23:54:17 | INFO | train rank=0 | step=31940/50000 progress=63.9% loss=0.088502 lr=3.162452e-05 lr_backbone=3.162452e-05 elapsed=33:23:55 avg_step=3.76s recent_step=3.69s eta=18:53:05 +2026-09-15 23:54:50 | INFO | train rank=0 | step=31950/50000 progress=63.9% loss=0.106794 lr=3.159377e-05 lr_backbone=3.159377e-05 elapsed=33:24:28 avg_step=3.76s recent_step=3.25s eta=18:52:24 +2026-09-15 23:55:50 | INFO | train rank=0 | step=31960/50000 progress=63.9% loss=0.106761 lr=3.156303e-05 lr_backbone=3.156303e-05 elapsed=33:25:28 avg_step=3.76s recent_step=5.99s eta=18:51:59 +2026-09-15 23:56:20 | INFO | train rank=0 | step=31970/50000 progress=63.9% loss=0.100051 lr=3.153229e-05 lr_backbone=3.153229e-05 elapsed=33:25:58 avg_step=3.76s recent_step=3.08s eta=18:51:18 +2026-09-15 23:56:55 | INFO | train rank=0 | step=31980/50000 progress=64.0% loss=0.121484 lr=3.150157e-05 lr_backbone=3.150157e-05 elapsed=33:26:33 avg_step=3.76s recent_step=3.47s eta=18:50:39 +2026-09-15 23:57:30 | INFO | train rank=0 | step=31990/50000 progress=64.0% loss=0.108081 lr=3.147085e-05 lr_backbone=3.147085e-05 elapsed=33:27:08 avg_step=3.76s recent_step=3.48s eta=18:49:59 +2026-09-15 23:58:12 | INFO | train rank=0 | step=32000/50000 progress=64.0% loss=0.097470 lr=3.144014e-05 lr_backbone=3.144014e-05 elapsed=33:27:50 avg_step=3.76s recent_step=4.22s eta=18:49:24 +2026-09-15 23:58:52 | INFO | train rank=0 | step=32010/50000 progress=64.0% loss=0.094241 lr=3.140943e-05 lr_backbone=3.140943e-05 elapsed=33:28:30 avg_step=3.76s recent_step=3.97s eta=18:48:48 +2026-09-15 23:59:29 | INFO | train rank=0 | step=32020/50000 progress=64.0% loss=0.087058 lr=3.137874e-05 lr_backbone=3.137874e-05 elapsed=33:29:07 avg_step=3.76s recent_step=3.75s eta=18:48:10 +2026-09-15 23:59:58 | INFO | train rank=0 | step=32030/50000 progress=64.1% loss=0.085380 lr=3.134805e-05 lr_backbone=3.134805e-05 elapsed=33:29:36 avg_step=3.76s recent_step=2.90s eta=18:47:28 +2026-09-16 00:00:38 | INFO | train rank=0 | step=32040/50000 progress=64.1% loss=0.069927 lr=3.131738e-05 lr_backbone=3.131738e-05 elapsed=33:30:16 avg_step=3.76s recent_step=3.92s eta=18:46:51 +2026-09-16 00:01:11 | INFO | train rank=0 | step=32050/50000 progress=64.1% loss=0.074965 lr=3.128671e-05 lr_backbone=3.128671e-05 elapsed=33:30:49 avg_step=3.76s recent_step=3.29s eta=18:46:10 +2026-09-16 00:01:51 | INFO | train rank=0 | step=32060/50000 progress=64.1% loss=0.084258 lr=3.125604e-05 lr_backbone=3.125604e-05 elapsed=33:31:29 avg_step=3.76s recent_step=4.01s eta=18:45:34 +2026-09-16 00:02:28 | INFO | train rank=0 | step=32070/50000 progress=64.1% loss=0.094661 lr=3.122539e-05 lr_backbone=3.122539e-05 elapsed=33:32:06 avg_step=3.76s recent_step=3.76s eta=18:44:57 +2026-09-16 00:03:16 | INFO | train rank=0 | step=32080/50000 progress=64.2% loss=0.121353 lr=3.119474e-05 lr_backbone=3.119474e-05 elapsed=33:32:54 avg_step=3.76s recent_step=4.78s eta=18:44:25 +2026-09-16 00:03:46 | INFO | train rank=0 | step=32090/50000 progress=64.2% loss=0.101440 lr=3.116411e-05 lr_backbone=3.116411e-05 elapsed=33:33:24 avg_step=3.76s recent_step=3.00s eta=18:43:43 +2026-09-16 00:04:24 | INFO | train rank=0 | step=32100/50000 progress=64.2% loss=0.100888 lr=3.113348e-05 lr_backbone=3.113348e-05 elapsed=33:34:02 avg_step=3.76s recent_step=3.78s eta=18:43:05 +2026-09-16 00:04:55 | INFO | train rank=0 | step=32110/50000 progress=64.2% loss=0.084676 lr=3.110286e-05 lr_backbone=3.110286e-05 elapsed=33:34:33 avg_step=3.76s recent_step=3.16s eta=18:42:24 +2026-09-16 00:05:49 | INFO | train rank=0 | step=32120/50000 progress=64.2% loss=0.102712 lr=3.107225e-05 lr_backbone=3.107225e-05 elapsed=33:35:27 avg_step=3.76s recent_step=5.35s eta=18:41:55 +2026-09-16 00:06:21 | INFO | train rank=0 | step=32130/50000 progress=64.3% loss=0.098933 lr=3.104164e-05 lr_backbone=3.104164e-05 elapsed=33:35:59 avg_step=3.76s recent_step=3.20s eta=18:41:14 +2026-09-16 00:06:54 | INFO | train rank=0 | step=32140/50000 progress=64.3% loss=0.100454 lr=3.101105e-05 lr_backbone=3.101105e-05 elapsed=33:36:32 avg_step=3.76s recent_step=3.33s eta=18:40:34 +2026-09-16 00:07:31 | INFO | train rank=0 | step=32150/50000 progress=64.3% loss=0.080284 lr=3.098046e-05 lr_backbone=3.098046e-05 elapsed=33:37:09 avg_step=3.76s recent_step=3.68s eta=18:39:56 +2026-09-16 00:08:13 | INFO | train rank=0 | step=32160/50000 progress=64.3% loss=0.088358 lr=3.094988e-05 lr_backbone=3.094988e-05 elapsed=33:37:51 avg_step=3.76s recent_step=4.19s eta=18:39:21 +2026-09-16 00:08:50 | INFO | train rank=0 | step=32170/50000 progress=64.3% loss=0.093829 lr=3.091931e-05 lr_backbone=3.091931e-05 elapsed=33:38:28 avg_step=3.76s recent_step=3.67s eta=18:38:43 +2026-09-16 00:09:21 | INFO | train rank=0 | step=32180/50000 progress=64.4% loss=0.095072 lr=3.088874e-05 lr_backbone=3.088874e-05 elapsed=33:38:59 avg_step=3.76s recent_step=3.14s eta=18:38:02 +2026-09-16 00:09:55 | INFO | train rank=0 | step=32190/50000 progress=64.4% loss=0.083172 lr=3.085819e-05 lr_backbone=3.085819e-05 elapsed=33:39:33 avg_step=3.76s recent_step=3.36s eta=18:37:22 +2026-09-16 00:10:48 | INFO | train rank=0 | step=32200/50000 progress=64.4% loss=0.108529 lr=3.082764e-05 lr_backbone=3.082764e-05 elapsed=33:40:26 avg_step=3.76s recent_step=5.30s eta=18:36:53 +2026-09-16 00:11:23 | INFO | train rank=0 | step=32210/50000 progress=64.4% loss=0.116868 lr=3.079711e-05 lr_backbone=3.079711e-05 elapsed=33:41:01 avg_step=3.76s recent_step=3.50s eta=18:36:14 +2026-09-16 00:11:52 | INFO | train rank=0 | step=32220/50000 progress=64.4% loss=0.079060 lr=3.076658e-05 lr_backbone=3.076658e-05 elapsed=33:41:30 avg_step=3.76s recent_step=2.97s eta=18:35:32 +2026-09-16 00:12:27 | INFO | train rank=0 | step=32230/50000 progress=64.5% loss=0.077840 lr=3.073606e-05 lr_backbone=3.073606e-05 elapsed=33:42:05 avg_step=3.76s recent_step=3.48s eta=18:34:52 +2026-09-16 00:13:05 | INFO | train rank=0 | step=32240/50000 progress=64.5% loss=0.093251 lr=3.070555e-05 lr_backbone=3.070555e-05 elapsed=33:42:43 avg_step=3.76s recent_step=3.82s eta=18:34:15 +2026-09-16 00:13:40 | INFO | train rank=0 | step=32250/50000 progress=64.5% loss=0.083677 lr=3.067504e-05 lr_backbone=3.067504e-05 elapsed=33:43:18 avg_step=3.76s recent_step=3.47s eta=18:33:36 +2026-09-16 00:14:29 | INFO | train rank=0 | step=32260/50000 progress=64.5% loss=0.068516 lr=3.064455e-05 lr_backbone=3.064455e-05 elapsed=33:44:07 avg_step=3.76s recent_step=4.86s eta=18:33:04 +2026-09-16 00:15:02 | INFO | train rank=0 | step=32270/50000 progress=64.5% loss=0.088356 lr=3.061406e-05 lr_backbone=3.061406e-05 elapsed=33:44:40 avg_step=3.76s recent_step=3.37s eta=18:32:24 +2026-09-16 00:15:49 | INFO | train rank=0 | step=32280/50000 progress=64.6% loss=0.075609 lr=3.058358e-05 lr_backbone=3.058358e-05 elapsed=33:45:27 avg_step=3.76s recent_step=4.65s eta=18:31:52 +2026-09-16 00:16:20 | INFO | train rank=0 | step=32290/50000 progress=64.6% loss=0.095827 lr=3.055311e-05 lr_backbone=3.055311e-05 elapsed=33:45:58 avg_step=3.76s recent_step=3.11s eta=18:31:10 +2026-09-16 00:16:57 | INFO | train rank=0 | step=32300/50000 progress=64.6% loss=0.078645 lr=3.052265e-05 lr_backbone=3.052265e-05 elapsed=33:46:35 avg_step=3.76s recent_step=3.66s eta=18:30:32 +2026-09-16 00:17:29 | INFO | train rank=0 | step=32310/50000 progress=64.6% loss=0.078281 lr=3.049220e-05 lr_backbone=3.049220e-05 elapsed=33:47:07 avg_step=3.76s recent_step=3.21s eta=18:29:51 +2026-09-16 00:18:26 | INFO | train rank=0 | step=32320/50000 progress=64.6% loss=0.096055 lr=3.046175e-05 lr_backbone=3.046175e-05 elapsed=33:48:04 avg_step=3.77s recent_step=5.77s eta=18:29:25 +2026-09-16 00:18:58 | INFO | train rank=0 | step=32330/50000 progress=64.7% loss=0.089374 lr=3.043132e-05 lr_backbone=3.043132e-05 elapsed=33:48:36 avg_step=3.76s recent_step=3.14s eta=18:28:44 +2026-09-16 00:19:30 | INFO | train rank=0 | step=32340/50000 progress=64.7% loss=0.083975 lr=3.040089e-05 lr_backbone=3.040089e-05 elapsed=33:49:08 avg_step=3.76s recent_step=3.20s eta=18:28:03 +2026-09-16 00:19:55 | INFO | train rank=0 | step=32350/50000 progress=64.7% loss=0.092282 lr=3.037047e-05 lr_backbone=3.037047e-05 elapsed=33:49:33 avg_step=3.76s recent_step=2.52s eta=18:27:18 +2026-09-16 00:20:50 | INFO | train rank=0 | step=32360/50000 progress=64.7% loss=0.135863 lr=3.034006e-05 lr_backbone=3.034006e-05 elapsed=33:50:28 avg_step=3.76s recent_step=5.48s eta=18:26:50 +2026-09-16 00:21:32 | INFO | train rank=0 | step=32370/50000 progress=64.7% loss=0.109040 lr=3.030966e-05 lr_backbone=3.030966e-05 elapsed=33:51:10 avg_step=3.76s recent_step=4.20s eta=18:26:15 +2026-09-16 00:21:55 | INFO | train rank=0 | step=32380/50000 progress=64.8% loss=0.103565 lr=3.027927e-05 lr_backbone=3.027927e-05 elapsed=33:51:33 avg_step=3.76s recent_step=2.30s eta=18:25:29 +2026-09-16 00:22:33 | INFO | train rank=0 | step=32390/50000 progress=64.8% loss=0.078827 lr=3.024888e-05 lr_backbone=3.024888e-05 elapsed=33:52:11 avg_step=3.76s recent_step=3.77s eta=18:24:52 +2026-09-16 00:23:11 | INFO | train rank=0 | step=32400/50000 progress=64.8% loss=0.078784 lr=3.021851e-05 lr_backbone=3.021851e-05 elapsed=33:52:49 avg_step=3.76s recent_step=3.90s eta=18:24:15 +2026-09-16 00:23:44 | INFO | train rank=0 | step=32410/50000 progress=64.8% loss=0.099079 lr=3.018814e-05 lr_backbone=3.018814e-05 elapsed=33:53:22 avg_step=3.76s recent_step=3.23s eta=18:23:34 +2026-09-16 00:24:21 | INFO | train rank=0 | step=32420/50000 progress=64.8% loss=0.097768 lr=3.015778e-05 lr_backbone=3.015778e-05 elapsed=33:53:59 avg_step=3.76s recent_step=3.77s eta=18:22:57 +2026-09-16 00:25:02 | INFO | train rank=0 | step=32430/50000 progress=64.9% loss=0.076091 lr=3.012743e-05 lr_backbone=3.012743e-05 elapsed=33:54:40 avg_step=3.76s recent_step=4.09s eta=18:22:21 +2026-09-16 00:25:43 | INFO | train rank=0 | step=32440/50000 progress=64.9% loss=0.088052 lr=3.009709e-05 lr_backbone=3.009709e-05 elapsed=33:55:21 avg_step=3.76s recent_step=4.06s eta=18:21:45 +2026-09-16 00:26:13 | INFO | train rank=0 | step=32450/50000 progress=64.9% loss=0.096492 lr=3.006676e-05 lr_backbone=3.006676e-05 elapsed=33:55:51 avg_step=3.76s recent_step=2.97s eta=18:21:03 +2026-09-16 00:26:56 | INFO | train rank=0 | step=32460/50000 progress=64.9% loss=0.077145 lr=3.003644e-05 lr_backbone=3.003644e-05 elapsed=33:56:34 avg_step=3.76s recent_step=4.35s eta=18:20:28 +2026-09-16 00:27:29 | INFO | train rank=0 | step=32470/50000 progress=64.9% loss=0.120332 lr=3.000612e-05 lr_backbone=3.000612e-05 elapsed=33:57:07 avg_step=3.76s recent_step=3.26s eta=18:19:48 +2026-09-16 00:28:06 | INFO | train rank=0 | step=32480/50000 progress=65.0% loss=0.085712 lr=2.997581e-05 lr_backbone=2.997581e-05 elapsed=33:57:44 avg_step=3.76s recent_step=3.68s eta=18:19:10 +2026-09-16 00:28:42 | INFO | train rank=0 | step=32490/50000 progress=65.0% loss=0.080910 lr=2.994552e-05 lr_backbone=2.994552e-05 elapsed=33:58:20 avg_step=3.76s recent_step=3.67s eta=18:18:32 +2026-09-16 00:29:17 | INFO | train rank=0 | step=32500/50000 progress=65.0% loss=0.082932 lr=2.991523e-05 lr_backbone=2.991523e-05 elapsed=33:58:55 avg_step=3.76s recent_step=3.51s eta=18:17:53 +2026-09-16 00:29:52 | INFO | train rank=0 | step=32510/50000 progress=65.0% loss=0.118973 lr=2.988495e-05 lr_backbone=2.988495e-05 elapsed=33:59:30 avg_step=3.76s recent_step=3.43s eta=18:17:13 +2026-09-16 00:30:29 | INFO | train rank=0 | step=32520/50000 progress=65.0% loss=0.107844 lr=2.985468e-05 lr_backbone=2.985468e-05 elapsed=34:00:07 avg_step=3.76s recent_step=3.77s eta=18:16:36 +2026-09-16 00:31:07 | INFO | train rank=0 | step=32530/50000 progress=65.1% loss=0.115379 lr=2.982442e-05 lr_backbone=2.982442e-05 elapsed=34:00:45 avg_step=3.76s recent_step=3.80s eta=18:15:58 +2026-09-16 00:31:59 | INFO | train rank=0 | step=32540/50000 progress=65.1% loss=0.094800 lr=2.979416e-05 lr_backbone=2.979416e-05 elapsed=34:01:37 avg_step=3.76s recent_step=5.16s eta=18:15:28 +2026-09-16 00:32:38 | INFO | train rank=0 | step=32550/50000 progress=65.1% loss=0.087815 lr=2.976392e-05 lr_backbone=2.976392e-05 elapsed=34:02:16 avg_step=3.76s recent_step=3.89s eta=18:14:51 +2026-09-16 00:33:13 | INFO | train rank=0 | step=32560/50000 progress=65.1% loss=0.094223 lr=2.973368e-05 lr_backbone=2.973368e-05 elapsed=34:02:51 avg_step=3.76s recent_step=3.47s eta=18:14:12 +2026-09-16 00:33:48 | INFO | train rank=0 | step=32570/50000 progress=65.1% loss=0.087708 lr=2.970346e-05 lr_backbone=2.970346e-05 elapsed=34:03:26 avg_step=3.76s recent_step=3.53s eta=18:13:33 +2026-09-16 00:34:29 | INFO | train rank=0 | step=32580/50000 progress=65.2% loss=0.074963 lr=2.967324e-05 lr_backbone=2.967324e-05 elapsed=34:04:07 avg_step=3.76s recent_step=4.12s eta=18:12:57 +2026-09-16 00:35:09 | INFO | train rank=0 | step=32590/50000 progress=65.2% loss=0.099356 lr=2.964303e-05 lr_backbone=2.964303e-05 elapsed=34:04:47 avg_step=3.76s recent_step=3.97s eta=18:12:21 +2026-09-16 00:35:39 | INFO | train rank=0 | step=32600/50000 progress=65.2% loss=0.093680 lr=2.961283e-05 lr_backbone=2.961283e-05 elapsed=34:05:17 avg_step=3.76s recent_step=3.04s eta=18:11:39 +2026-09-16 00:36:03 | INFO | train rank=0 | step=32610/50000 progress=65.2% loss=0.077268 lr=2.958264e-05 lr_backbone=2.958264e-05 elapsed=34:05:41 avg_step=3.76s recent_step=2.41s eta=18:10:54 +2026-09-16 00:37:00 | INFO | train rank=0 | step=32620/50000 progress=65.2% loss=0.115016 lr=2.955246e-05 lr_backbone=2.955246e-05 elapsed=34:06:38 avg_step=3.76s recent_step=5.66s eta=18:10:27 +2026-09-16 00:37:44 | INFO | train rank=0 | step=32630/50000 progress=65.3% loss=0.077526 lr=2.952228e-05 lr_backbone=2.952228e-05 elapsed=34:07:22 avg_step=3.76s recent_step=4.42s eta=18:09:53 +2026-09-16 00:38:22 | INFO | train rank=0 | step=32640/50000 progress=65.3% loss=0.082625 lr=2.949212e-05 lr_backbone=2.949212e-05 elapsed=34:08:00 avg_step=3.76s recent_step=3.80s eta=18:09:15 +2026-09-16 00:38:51 | INFO | train rank=0 | step=32650/50000 progress=65.3% loss=0.110329 lr=2.946196e-05 lr_backbone=2.946196e-05 elapsed=34:08:29 avg_step=3.76s recent_step=2.92s eta=18:08:33 +2026-09-16 00:39:40 | INFO | train rank=0 | step=32660/50000 progress=65.3% loss=0.102685 lr=2.943182e-05 lr_backbone=2.943182e-05 elapsed=34:09:18 avg_step=3.76s recent_step=4.84s eta=18:08:01 +2026-09-16 00:40:15 | INFO | train rank=0 | step=32670/50000 progress=65.3% loss=0.091808 lr=2.940168e-05 lr_backbone=2.940168e-05 elapsed=34:09:53 avg_step=3.76s recent_step=3.53s eta=18:07:22 +2026-09-16 00:40:59 | INFO | train rank=0 | step=32680/50000 progress=65.4% loss=0.096070 lr=2.937155e-05 lr_backbone=2.937155e-05 elapsed=34:10:37 avg_step=3.76s recent_step=4.39s eta=18:06:48 +2026-09-16 00:41:35 | INFO | train rank=0 | step=32690/50000 progress=65.4% loss=0.103990 lr=2.934143e-05 lr_backbone=2.934143e-05 elapsed=34:11:13 avg_step=3.76s recent_step=3.58s eta=18:06:09 +2026-09-16 00:42:18 | INFO | train rank=0 | step=32700/50000 progress=65.4% loss=0.100871 lr=2.931132e-05 lr_backbone=2.931132e-05 elapsed=34:11:56 avg_step=3.77s recent_step=4.37s eta=18:05:35 +2026-09-16 00:42:58 | INFO | train rank=0 | step=32710/50000 progress=65.4% loss=0.081141 lr=2.928122e-05 lr_backbone=2.928122e-05 elapsed=34:12:36 avg_step=3.77s recent_step=3.91s eta=18:04:58 +2026-09-16 00:43:33 | INFO | train rank=0 | step=32720/50000 progress=65.4% loss=0.087706 lr=2.925113e-05 lr_backbone=2.925113e-05 elapsed=34:13:11 avg_step=3.77s recent_step=3.58s eta=18:04:19 +2026-09-16 00:44:06 | INFO | train rank=0 | step=32730/50000 progress=65.5% loss=0.098500 lr=2.922105e-05 lr_backbone=2.922105e-05 elapsed=34:13:44 avg_step=3.76s recent_step=3.26s eta=18:03:39 +2026-09-16 00:44:52 | INFO | train rank=0 | step=32740/50000 progress=65.5% loss=0.078671 lr=2.919097e-05 lr_backbone=2.919097e-05 elapsed=34:14:30 avg_step=3.77s recent_step=4.60s eta=18:03:06 +2026-09-16 00:45:33 | INFO | train rank=0 | step=32750/50000 progress=65.5% loss=0.087173 lr=2.916091e-05 lr_backbone=2.916091e-05 elapsed=34:15:11 avg_step=3.77s recent_step=4.11s eta=18:02:30 +2026-09-16 00:46:03 | INFO | train rank=0 | step=32760/50000 progress=65.5% loss=0.109998 lr=2.913085e-05 lr_backbone=2.913085e-05 elapsed=34:15:41 avg_step=3.76s recent_step=2.96s eta=18:01:48 +2026-09-16 00:46:32 | INFO | train rank=0 | step=32770/50000 progress=65.5% loss=0.102093 lr=2.910080e-05 lr_backbone=2.910080e-05 elapsed=34:16:10 avg_step=3.76s recent_step=2.98s eta=18:01:06 +2026-09-16 00:47:27 | INFO | train rank=0 | step=32780/50000 progress=65.6% loss=0.087143 lr=2.907077e-05 lr_backbone=2.907077e-05 elapsed=34:17:05 avg_step=3.77s recent_step=5.46s eta=18:00:37 +2026-09-16 00:47:59 | INFO | train rank=0 | step=32790/50000 progress=65.6% loss=0.084081 lr=2.904074e-05 lr_backbone=2.904074e-05 elapsed=34:17:37 avg_step=3.77s recent_step=3.21s eta=17:59:57 +2026-09-16 00:48:35 | INFO | train rank=0 | step=32800/50000 progress=65.6% loss=0.116971 lr=2.901072e-05 lr_backbone=2.901072e-05 elapsed=34:18:13 avg_step=3.77s recent_step=3.61s eta=17:59:18 +2026-09-16 00:49:12 | INFO | train rank=0 | step=32810/50000 progress=65.6% loss=0.097538 lr=2.898071e-05 lr_backbone=2.898071e-05 elapsed=34:18:50 avg_step=3.77s recent_step=3.72s eta=17:58:41 +2026-09-16 00:50:01 | INFO | train rank=0 | step=32820/50000 progress=65.6% loss=0.072691 lr=2.895071e-05 lr_backbone=2.895071e-05 elapsed=34:19:39 avg_step=3.77s recent_step=4.82s eta=17:58:08 +2026-09-16 00:50:37 | INFO | train rank=0 | step=32830/50000 progress=65.7% loss=0.073985 lr=2.892072e-05 lr_backbone=2.892072e-05 elapsed=34:20:15 avg_step=3.77s recent_step=3.63s eta=17:57:30 +2026-09-16 00:51:14 | INFO | train rank=0 | step=32840/50000 progress=65.7% loss=0.106296 lr=2.889074e-05 lr_backbone=2.889074e-05 elapsed=34:20:52 avg_step=3.77s recent_step=3.73s eta=17:56:52 +2026-09-16 00:51:50 | INFO | train rank=0 | step=32850/50000 progress=65.7% loss=0.077937 lr=2.886076e-05 lr_backbone=2.886076e-05 elapsed=34:21:28 avg_step=3.77s recent_step=3.61s eta=17:56:14 +2026-09-16 00:52:30 | INFO | train rank=0 | step=32860/50000 progress=65.7% loss=0.088674 lr=2.883080e-05 lr_backbone=2.883080e-05 elapsed=34:22:08 avg_step=3.77s recent_step=3.96s eta=17:55:37 +2026-09-16 00:52:58 | INFO | train rank=0 | step=32870/50000 progress=65.7% loss=0.080226 lr=2.880084e-05 lr_backbone=2.880084e-05 elapsed=34:22:36 avg_step=3.77s recent_step=2.83s eta=17:54:55 +2026-09-16 00:53:30 | INFO | train rank=0 | step=32880/50000 progress=65.8% loss=0.094368 lr=2.877090e-05 lr_backbone=2.877090e-05 elapsed=34:23:08 avg_step=3.76s recent_step=3.13s eta=17:54:14 +2026-09-16 00:54:04 | INFO | train rank=0 | step=32890/50000 progress=65.8% loss=0.067867 lr=2.874096e-05 lr_backbone=2.874096e-05 elapsed=34:23:42 avg_step=3.76s recent_step=3.48s eta=17:53:35 +2026-09-16 00:54:48 | INFO | train rank=0 | step=32900/50000 progress=65.8% loss=0.090897 lr=2.871104e-05 lr_backbone=2.871104e-05 elapsed=34:24:26 avg_step=3.76s recent_step=4.32s eta=17:53:00 +2026-09-16 00:55:22 | INFO | train rank=0 | step=32910/50000 progress=65.8% loss=0.113336 lr=2.868112e-05 lr_backbone=2.868112e-05 elapsed=34:25:00 avg_step=3.76s recent_step=3.39s eta=17:52:20 +2026-09-16 00:56:02 | INFO | train rank=0 | step=32920/50000 progress=65.8% loss=0.082313 lr=2.865121e-05 lr_backbone=2.865121e-05 elapsed=34:25:40 avg_step=3.76s recent_step=4.08s eta=17:51:44 +2026-09-16 00:56:34 | INFO | train rank=0 | step=32930/50000 progress=65.9% loss=0.070720 lr=2.862131e-05 lr_backbone=2.862131e-05 elapsed=34:26:12 avg_step=3.76s recent_step=3.16s eta=17:51:03 +2026-09-16 00:57:16 | INFO | train rank=0 | step=32940/50000 progress=65.9% loss=0.089595 lr=2.859142e-05 lr_backbone=2.859142e-05 elapsed=34:26:54 avg_step=3.76s recent_step=4.20s eta=17:50:28 +2026-09-16 00:57:44 | INFO | train rank=0 | step=32950/50000 progress=65.9% loss=0.114996 lr=2.856154e-05 lr_backbone=2.856154e-05 elapsed=34:27:22 avg_step=3.76s recent_step=2.76s eta=17:49:45 +2026-09-16 00:58:21 | INFO | train rank=0 | step=32960/50000 progress=65.9% loss=0.093622 lr=2.853167e-05 lr_backbone=2.853167e-05 elapsed=34:27:59 avg_step=3.76s recent_step=3.74s eta=17:49:07 +2026-09-16 00:58:44 | INFO | train rank=0 | step=32970/50000 progress=65.9% loss=0.088288 lr=2.850181e-05 lr_backbone=2.850181e-05 elapsed=34:28:22 avg_step=3.76s recent_step=2.33s eta=17:48:22 +2026-09-16 00:59:33 | INFO | train rank=0 | step=32980/50000 progress=66.0% loss=0.082517 lr=2.847196e-05 lr_backbone=2.847196e-05 elapsed=34:29:11 avg_step=3.76s recent_step=4.83s eta=17:47:50 +2026-09-16 01:00:08 | INFO | train rank=0 | step=32990/50000 progress=66.0% loss=0.108397 lr=2.844212e-05 lr_backbone=2.844212e-05 elapsed=34:29:46 avg_step=3.76s recent_step=3.50s eta=17:47:11 +2026-09-16 01:00:41 | INFO | train rank=0 | step=33000/50000 progress=66.0% loss=0.088611 lr=2.841228e-05 lr_backbone=2.841228e-05 elapsed=34:30:19 avg_step=3.76s recent_step=3.37s eta=17:46:32 +2026-09-16 01:01:25 | INFO | train rank=0 | step=33010/50000 progress=66.0% loss=0.071451 lr=2.838246e-05 lr_backbone=2.838246e-05 elapsed=34:31:04 avg_step=3.76s recent_step=4.41s eta=17:45:57 +2026-09-16 01:02:05 | INFO | train rank=0 | step=33020/50000 progress=66.0% loss=0.087865 lr=2.835264e-05 lr_backbone=2.835264e-05 elapsed=34:31:43 avg_step=3.76s recent_step=3.92s eta=17:45:20 +2026-09-16 01:02:43 | INFO | train rank=0 | step=33030/50000 progress=66.1% loss=0.072520 lr=2.832284e-05 lr_backbone=2.832284e-05 elapsed=34:32:21 avg_step=3.76s recent_step=3.87s eta=17:44:43 +2026-09-16 01:03:23 | INFO | train rank=0 | step=33040/50000 progress=66.1% loss=0.080448 lr=2.829304e-05 lr_backbone=2.829304e-05 elapsed=34:33:01 avg_step=3.76s recent_step=3.97s eta=17:44:07 +2026-09-16 01:03:51 | INFO | train rank=0 | step=33050/50000 progress=66.1% loss=0.092121 lr=2.826326e-05 lr_backbone=2.826326e-05 elapsed=34:33:29 avg_step=3.76s recent_step=2.76s eta=17:43:24 +2026-09-16 01:04:33 | INFO | train rank=0 | step=33060/50000 progress=66.1% loss=0.092695 lr=2.823348e-05 lr_backbone=2.823348e-05 elapsed=34:34:11 avg_step=3.76s recent_step=4.22s eta=17:42:49 +2026-09-16 01:05:07 | INFO | train rank=0 | step=33070/50000 progress=66.1% loss=0.097101 lr=2.820372e-05 lr_backbone=2.820372e-05 elapsed=34:34:45 avg_step=3.76s recent_step=3.40s eta=17:42:09 +2026-09-16 01:05:47 | INFO | train rank=0 | step=33080/50000 progress=66.2% loss=0.073985 lr=2.817396e-05 lr_backbone=2.817396e-05 elapsed=34:35:25 avg_step=3.76s recent_step=4.01s eta=17:41:33 +2026-09-16 01:06:17 | INFO | train rank=0 | step=33090/50000 progress=66.2% loss=0.095952 lr=2.814421e-05 lr_backbone=2.814421e-05 elapsed=34:35:55 avg_step=3.76s recent_step=2.95s eta=17:40:51 +2026-09-16 01:07:08 | INFO | train rank=0 | step=33100/50000 progress=66.2% loss=0.080091 lr=2.811447e-05 lr_backbone=2.811447e-05 elapsed=34:36:46 avg_step=3.76s recent_step=5.12s eta=17:40:20 +2026-09-16 01:07:42 | INFO | train rank=0 | step=33110/50000 progress=66.2% loss=0.073021 lr=2.808475e-05 lr_backbone=2.808475e-05 elapsed=34:37:20 avg_step=3.76s recent_step=3.45s eta=17:39:41 +2026-09-16 01:08:03 | INFO | train rank=0 | step=33120/50000 progress=66.2% loss=0.086437 lr=2.805503e-05 lr_backbone=2.805503e-05 elapsed=34:37:41 avg_step=3.76s recent_step=2.09s eta=17:38:55 +2026-09-16 01:08:38 | INFO | train rank=0 | step=33130/50000 progress=66.3% loss=0.107092 lr=2.802532e-05 lr_backbone=2.802532e-05 elapsed=34:38:16 avg_step=3.76s recent_step=3.49s eta=17:38:16 +2026-09-16 01:09:30 | INFO | train rank=0 | step=33140/50000 progress=66.3% loss=0.096723 lr=2.799562e-05 lr_backbone=2.799562e-05 elapsed=34:39:08 avg_step=3.76s recent_step=5.19s eta=17:37:45 +2026-09-16 01:10:01 | INFO | train rank=0 | step=33150/50000 progress=66.3% loss=0.084348 lr=2.796593e-05 lr_backbone=2.796593e-05 elapsed=34:39:39 avg_step=3.76s recent_step=3.07s eta=17:37:04 +2026-09-16 01:10:29 | INFO | train rank=0 | step=33160/50000 progress=66.3% loss=0.099915 lr=2.793625e-05 lr_backbone=2.793625e-05 elapsed=34:40:07 avg_step=3.76s recent_step=2.80s eta=17:36:22 +2026-09-16 01:10:57 | INFO | train rank=0 | step=33170/50000 progress=66.3% loss=0.066980 lr=2.790658e-05 lr_backbone=2.790658e-05 elapsed=34:40:35 avg_step=3.76s recent_step=2.78s eta=17:35:39 +2026-09-16 01:11:58 | INFO | train rank=0 | step=33180/50000 progress=66.4% loss=0.106258 lr=2.787692e-05 lr_backbone=2.787692e-05 elapsed=34:41:36 avg_step=3.76s recent_step=6.12s eta=17:35:13 +2026-09-16 01:12:33 | INFO | train rank=0 | step=33190/50000 progress=66.4% loss=0.112091 lr=2.784726e-05 lr_backbone=2.784726e-05 elapsed=34:42:11 avg_step=3.76s recent_step=3.55s eta=17:34:35 +2026-09-16 01:13:00 | INFO | train rank=0 | step=33200/50000 progress=66.4% loss=0.084921 lr=2.781762e-05 lr_backbone=2.781762e-05 elapsed=34:42:38 avg_step=3.76s recent_step=2.63s eta=17:33:51 +2026-09-16 01:13:37 | INFO | train rank=0 | step=33210/50000 progress=66.4% loss=0.098890 lr=2.778799e-05 lr_backbone=2.778799e-05 elapsed=34:43:15 avg_step=3.76s recent_step=3.74s eta=17:33:14 +2026-09-16 01:14:26 | INFO | train rank=0 | step=33220/50000 progress=66.4% loss=0.092761 lr=2.775837e-05 lr_backbone=2.775837e-05 elapsed=34:44:04 avg_step=3.76s recent_step=4.91s eta=17:32:42 +2026-09-16 01:15:01 | INFO | train rank=0 | step=33230/50000 progress=66.5% loss=0.116381 lr=2.772876e-05 lr_backbone=2.772876e-05 elapsed=34:44:39 avg_step=3.76s recent_step=3.51s eta=17:32:03 +2026-09-16 01:15:30 | INFO | train rank=0 | step=33240/50000 progress=66.5% loss=0.079759 lr=2.769915e-05 lr_backbone=2.769915e-05 elapsed=34:45:08 avg_step=3.76s recent_step=2.93s eta=17:31:21 +2026-09-16 01:16:02 | INFO | train rank=0 | step=33250/50000 progress=66.5% loss=0.090065 lr=2.766956e-05 lr_backbone=2.766956e-05 elapsed=34:45:40 avg_step=3.76s recent_step=3.17s eta=17:30:40 +2026-09-16 01:17:04 | INFO | train rank=0 | step=33260/50000 progress=66.5% loss=0.108441 lr=2.763998e-05 lr_backbone=2.763998e-05 elapsed=34:46:42 avg_step=3.76s recent_step=6.18s eta=17:30:15 +2026-09-16 01:17:27 | INFO | train rank=0 | step=33270/50000 progress=66.5% loss=0.091091 lr=2.761040e-05 lr_backbone=2.761040e-05 elapsed=34:47:05 avg_step=3.76s recent_step=2.27s eta=17:29:30 +2026-09-16 01:18:05 | INFO | train rank=0 | step=33280/50000 progress=66.6% loss=0.080933 lr=2.758084e-05 lr_backbone=2.758084e-05 elapsed=34:47:43 avg_step=3.76s recent_step=3.83s eta=17:28:52 +2026-09-16 01:18:37 | INFO | train rank=0 | step=33290/50000 progress=66.6% loss=0.083855 lr=2.755129e-05 lr_backbone=2.755129e-05 elapsed=34:48:15 avg_step=3.76s recent_step=3.17s eta=17:28:12 +2026-09-16 01:19:23 | INFO | train rank=0 | step=33300/50000 progress=66.6% loss=0.094947 lr=2.752174e-05 lr_backbone=2.752174e-05 elapsed=34:49:01 avg_step=3.76s recent_step=4.67s eta=17:27:39 +2026-09-16 01:19:56 | INFO | train rank=0 | step=33310/50000 progress=66.6% loss=0.083116 lr=2.749221e-05 lr_backbone=2.749221e-05 elapsed=34:49:34 avg_step=3.76s recent_step=3.22s eta=17:26:58 +2026-09-16 01:20:35 | INFO | train rank=0 | step=33320/50000 progress=66.6% loss=0.072264 lr=2.746268e-05 lr_backbone=2.746268e-05 elapsed=34:50:13 avg_step=3.76s recent_step=3.96s eta=17:26:22 +2026-09-16 01:21:09 | INFO | train rank=0 | step=33330/50000 progress=66.7% loss=0.091556 lr=2.743317e-05 lr_backbone=2.743317e-05 elapsed=34:50:47 avg_step=3.76s recent_step=3.35s eta=17:25:42 +2026-09-16 01:22:03 | INFO | train rank=0 | step=33340/50000 progress=66.7% loss=0.082236 lr=2.740366e-05 lr_backbone=2.740366e-05 elapsed=34:51:41 avg_step=3.76s recent_step=5.45s eta=17:25:13 +2026-09-16 01:22:42 | INFO | train rank=0 | step=33350/50000 progress=66.7% loss=0.118836 lr=2.737417e-05 lr_backbone=2.737417e-05 elapsed=34:52:20 avg_step=3.76s recent_step=3.93s eta=17:24:36 +2026-09-16 01:23:11 | INFO | train rank=0 | step=33360/50000 progress=66.7% loss=0.073484 lr=2.734468e-05 lr_backbone=2.734468e-05 elapsed=34:52:49 avg_step=3.76s recent_step=2.85s eta=17:23:54 +2026-09-16 01:23:41 | INFO | train rank=0 | step=33370/50000 progress=66.7% loss=0.108818 lr=2.731521e-05 lr_backbone=2.731521e-05 elapsed=34:53:19 avg_step=3.76s recent_step=3.02s eta=17:23:12 +2026-09-16 01:24:44 | INFO | train rank=0 | step=33380/50000 progress=66.8% loss=0.082647 lr=2.728574e-05 lr_backbone=2.728574e-05 elapsed=34:54:22 avg_step=3.76s recent_step=6.25s eta=17:22:47 +2026-09-16 01:25:14 | INFO | train rank=0 | step=33390/50000 progress=66.8% loss=0.086897 lr=2.725629e-05 lr_backbone=2.725629e-05 elapsed=34:54:52 avg_step=3.76s recent_step=3.05s eta=17:22:06 +2026-09-16 01:25:38 | INFO | train rank=0 | step=33400/50000 progress=66.8% loss=0.117471 lr=2.722684e-05 lr_backbone=2.722684e-05 elapsed=34:55:16 avg_step=3.76s recent_step=2.41s eta=17:21:21 +2026-09-16 01:26:10 | INFO | train rank=0 | step=33410/50000 progress=66.8% loss=0.085413 lr=2.719741e-05 lr_backbone=2.719741e-05 elapsed=34:55:48 avg_step=3.76s recent_step=3.13s eta=17:20:41 +2026-09-16 01:26:59 | INFO | train rank=0 | step=33420/50000 progress=66.8% loss=0.078308 lr=2.716798e-05 lr_backbone=2.716798e-05 elapsed=34:56:37 avg_step=3.76s recent_step=4.94s eta=17:20:09 +2026-09-16 01:27:30 | INFO | train rank=0 | step=33430/50000 progress=66.9% loss=0.107135 lr=2.713857e-05 lr_backbone=2.713857e-05 elapsed=34:57:08 avg_step=3.76s recent_step=3.09s eta=17:19:28 +2026-09-16 01:28:03 | INFO | train rank=0 | step=33440/50000 progress=66.9% loss=0.079763 lr=2.710916e-05 lr_backbone=2.710916e-05 elapsed=34:57:41 avg_step=3.76s recent_step=3.31s eta=17:18:48 +2026-09-16 01:28:41 | INFO | train rank=0 | step=33450/50000 progress=66.9% loss=0.100780 lr=2.707977e-05 lr_backbone=2.707977e-05 elapsed=34:58:19 avg_step=3.76s recent_step=3.80s eta=17:18:11 +2026-09-16 01:29:43 | INFO | train rank=0 | step=33460/50000 progress=66.9% loss=0.084299 lr=2.705038e-05 lr_backbone=2.705038e-05 elapsed=34:59:21 avg_step=3.76s recent_step=6.16s eta=17:17:45 +2026-09-16 01:30:24 | INFO | train rank=0 | step=33470/50000 progress=66.9% loss=0.071275 lr=2.702101e-05 lr_backbone=2.702101e-05 elapsed=35:00:02 avg_step=3.76s recent_step=4.14s eta=17:17:09 +2026-09-16 01:30:56 | INFO | train rank=0 | step=33480/50000 progress=67.0% loss=0.079822 lr=2.699164e-05 lr_backbone=2.699164e-05 elapsed=35:00:34 avg_step=3.76s recent_step=3.20s eta=17:16:29 +2026-09-16 01:31:19 | INFO | train rank=0 | step=33490/50000 progress=67.0% loss=0.076858 lr=2.696229e-05 lr_backbone=2.696229e-05 elapsed=35:00:57 avg_step=3.76s recent_step=2.33s eta=17:15:44 +2026-09-16 01:32:10 | INFO | train rank=0 | step=33500/50000 progress=67.0% loss=0.109063 lr=2.693294e-05 lr_backbone=2.693294e-05 elapsed=35:01:48 avg_step=3.76s recent_step=5.06s eta=17:15:13 +2026-09-16 01:32:44 | INFO | train rank=0 | step=33510/50000 progress=67.0% loss=0.092544 lr=2.690361e-05 lr_backbone=2.690361e-05 elapsed=35:02:22 avg_step=3.76s recent_step=3.36s eta=17:14:33 +2026-09-16 01:33:22 | INFO | train rank=0 | step=33520/50000 progress=67.0% loss=0.077598 lr=2.687428e-05 lr_backbone=2.687428e-05 elapsed=35:03:00 avg_step=3.76s recent_step=3.83s eta=17:13:56 +2026-09-16 01:33:51 | INFO | train rank=0 | step=33530/50000 progress=67.1% loss=0.075095 lr=2.684497e-05 lr_backbone=2.684497e-05 elapsed=35:03:29 avg_step=3.76s recent_step=2.88s eta=17:13:14 +2026-09-16 01:34:46 | INFO | train rank=0 | step=33540/50000 progress=67.1% loss=0.098826 lr=2.681566e-05 lr_backbone=2.681566e-05 elapsed=35:04:24 avg_step=3.76s recent_step=5.49s eta=17:12:45 +2026-09-16 01:35:20 | INFO | train rank=0 | step=33550/50000 progress=67.1% loss=0.109912 lr=2.678637e-05 lr_backbone=2.678637e-05 elapsed=35:04:58 avg_step=3.76s recent_step=3.44s eta=17:12:05 +2026-09-16 01:35:55 | INFO | train rank=0 | step=33560/50000 progress=67.1% loss=0.079412 lr=2.675708e-05 lr_backbone=2.675708e-05 elapsed=35:05:33 avg_step=3.76s recent_step=3.49s eta=17:11:26 +2026-09-16 01:36:16 | INFO | train rank=0 | step=33570/50000 progress=67.1% loss=0.098373 lr=2.672781e-05 lr_backbone=2.672781e-05 elapsed=35:05:54 avg_step=3.76s recent_step=2.12s eta=17:10:41 +2026-09-16 01:37:10 | INFO | train rank=0 | step=33580/50000 progress=67.2% loss=0.080843 lr=2.669855e-05 lr_backbone=2.669855e-05 elapsed=35:06:48 avg_step=3.76s recent_step=5.43s eta=17:10:11 +2026-09-16 01:37:46 | INFO | train rank=0 | step=33590/50000 progress=67.2% loss=0.092993 lr=2.666929e-05 lr_backbone=2.666929e-05 elapsed=35:07:24 avg_step=3.76s recent_step=3.52s eta=17:09:32 +2026-09-16 01:38:16 | INFO | train rank=0 | step=33600/50000 progress=67.2% loss=0.113332 lr=2.664005e-05 lr_backbone=2.664005e-05 elapsed=35:07:54 avg_step=3.76s recent_step=3.07s eta=17:08:51 +2026-09-16 01:38:49 | INFO | train rank=0 | step=33610/50000 progress=67.2% loss=0.090982 lr=2.661082e-05 lr_backbone=2.661082e-05 elapsed=35:08:27 avg_step=3.76s recent_step=3.32s eta=17:08:11 +2026-09-16 01:39:24 | INFO | train rank=0 | step=33620/50000 progress=67.2% loss=0.085394 lr=2.658159e-05 lr_backbone=2.658159e-05 elapsed=35:09:02 avg_step=3.76s recent_step=3.42s eta=17:07:32 +2026-09-16 01:40:16 | INFO | train rank=0 | step=33630/50000 progress=67.3% loss=0.098675 lr=2.655238e-05 lr_backbone=2.655238e-05 elapsed=35:09:54 avg_step=3.76s recent_step=5.20s eta=17:07:01 +2026-09-16 01:40:41 | INFO | train rank=0 | step=33640/50000 progress=67.3% loss=0.075747 lr=2.652318e-05 lr_backbone=2.652318e-05 elapsed=35:10:19 avg_step=3.76s recent_step=2.58s eta=17:06:18 +2026-09-16 01:41:35 | INFO | train rank=0 | step=33650/50000 progress=67.3% loss=0.083599 lr=2.649399e-05 lr_backbone=2.649399e-05 elapsed=35:11:13 avg_step=3.76s recent_step=5.37s eta=17:05:48 +2026-09-16 01:42:09 | INFO | train rank=0 | step=33660/50000 progress=67.3% loss=0.091560 lr=2.646480e-05 lr_backbone=2.646480e-05 elapsed=35:11:47 avg_step=3.76s recent_step=3.38s eta=17:05:09 +2026-09-16 01:42:47 | INFO | train rank=0 | step=33670/50000 progress=67.3% loss=0.074132 lr=2.643563e-05 lr_backbone=2.643563e-05 elapsed=35:12:25 avg_step=3.76s recent_step=3.85s eta=17:04:31 +2026-09-16 01:43:31 | INFO | train rank=0 | step=33680/50000 progress=67.4% loss=0.066638 lr=2.640647e-05 lr_backbone=2.640647e-05 elapsed=35:13:09 avg_step=3.76s recent_step=4.38s eta=17:03:57 +2026-09-16 01:44:19 | INFO | train rank=0 | step=33690/50000 progress=67.4% loss=0.097560 lr=2.637732e-05 lr_backbone=2.637732e-05 elapsed=35:13:58 avg_step=3.76s recent_step=4.83s eta=17:03:24 +2026-09-16 01:44:53 | INFO | train rank=0 | step=33700/50000 progress=67.4% loss=0.093986 lr=2.634818e-05 lr_backbone=2.634818e-05 elapsed=35:14:31 avg_step=3.76s recent_step=3.30s eta=17:02:44 +2026-09-16 01:45:22 | INFO | train rank=0 | step=33710/50000 progress=67.4% loss=0.095312 lr=2.631905e-05 lr_backbone=2.631905e-05 elapsed=35:15:00 avg_step=3.76s recent_step=2.99s eta=17:02:03 +2026-09-16 01:45:53 | INFO | train rank=0 | step=33720/50000 progress=67.4% loss=0.094629 lr=2.628993e-05 lr_backbone=2.628993e-05 elapsed=35:15:31 avg_step=3.76s recent_step=3.10s eta=17:01:22 +2026-09-16 01:46:43 | INFO | train rank=0 | step=33730/50000 progress=67.5% loss=0.105020 lr=2.626082e-05 lr_backbone=2.626082e-05 elapsed=35:16:21 avg_step=3.76s recent_step=4.93s eta=17:00:50 +2026-09-16 01:47:22 | INFO | train rank=0 | step=33740/50000 progress=67.5% loss=0.092573 lr=2.623172e-05 lr_backbone=2.623172e-05 elapsed=35:17:00 avg_step=3.76s recent_step=3.98s eta=17:00:14 +2026-09-16 01:47:50 | INFO | train rank=0 | step=33750/50000 progress=67.5% loss=0.086232 lr=2.620263e-05 lr_backbone=2.620263e-05 elapsed=35:17:28 avg_step=3.76s recent_step=2.79s eta=16:59:31 +2026-09-16 01:48:32 | INFO | train rank=0 | step=33760/50000 progress=67.5% loss=0.082313 lr=2.617355e-05 lr_backbone=2.617355e-05 elapsed=35:18:10 avg_step=3.76s recent_step=4.13s eta=16:58:55 +2026-09-16 01:49:36 | INFO | train rank=0 | step=33770/50000 progress=67.5% loss=0.111616 lr=2.614448e-05 lr_backbone=2.614448e-05 elapsed=35:19:14 avg_step=3.77s recent_step=6.40s eta=16:58:30 +2026-09-16 01:50:05 | INFO | train rank=0 | step=33780/50000 progress=67.6% loss=0.090971 lr=2.611543e-05 lr_backbone=2.611543e-05 elapsed=35:19:43 avg_step=3.77s recent_step=2.98s eta=16:57:49 +2026-09-16 01:50:43 | INFO | train rank=0 | step=33790/50000 progress=67.6% loss=0.099358 lr=2.608638e-05 lr_backbone=2.608638e-05 elapsed=35:20:21 avg_step=3.77s recent_step=3.72s eta=16:57:11 +2026-09-16 01:51:13 | INFO | train rank=0 | step=33800/50000 progress=67.6% loss=0.112453 lr=2.605734e-05 lr_backbone=2.605734e-05 elapsed=35:20:51 avg_step=3.76s recent_step=3.01s eta=16:56:30 +2026-09-16 01:52:03 | INFO | train rank=0 | step=33810/50000 progress=67.6% loss=0.083988 lr=2.602832e-05 lr_backbone=2.602832e-05 elapsed=35:21:41 avg_step=3.77s recent_step=5.01s eta=16:55:58 +2026-09-16 01:52:37 | INFO | train rank=0 | step=33820/50000 progress=67.6% loss=0.083644 lr=2.599930e-05 lr_backbone=2.599930e-05 elapsed=35:22:15 avg_step=3.77s recent_step=3.46s eta=16:55:19 +2026-09-16 01:53:15 | INFO | train rank=0 | step=33830/50000 progress=67.7% loss=0.126693 lr=2.597030e-05 lr_backbone=2.597030e-05 elapsed=35:22:53 avg_step=3.77s recent_step=3.73s eta=16:54:41 +2026-09-16 01:53:46 | INFO | train rank=0 | step=33840/50000 progress=67.7% loss=0.124472 lr=2.594130e-05 lr_backbone=2.594130e-05 elapsed=35:23:24 avg_step=3.76s recent_step=3.15s eta=16:54:01 +2026-09-16 01:54:38 | INFO | train rank=0 | step=33850/50000 progress=67.7% loss=0.086679 lr=2.591232e-05 lr_backbone=2.591232e-05 elapsed=35:24:16 avg_step=3.77s recent_step=5.21s eta=16:53:30 +2026-09-16 01:55:13 | INFO | train rank=0 | step=33860/50000 progress=67.7% loss=0.097759 lr=2.588334e-05 lr_backbone=2.588334e-05 elapsed=35:24:51 avg_step=3.77s recent_step=3.43s eta=16:52:51 +2026-09-16 01:55:50 | INFO | train rank=0 | step=33870/50000 progress=67.7% loss=0.075394 lr=2.585438e-05 lr_backbone=2.585438e-05 elapsed=35:25:28 avg_step=3.77s recent_step=3.72s eta=16:52:13 +2026-09-16 01:56:20 | INFO | train rank=0 | step=33880/50000 progress=67.8% loss=0.106530 lr=2.582543e-05 lr_backbone=2.582543e-05 elapsed=35:25:58 avg_step=3.77s recent_step=3.03s eta=16:51:32 +2026-09-16 01:57:14 | INFO | train rank=0 | step=33890/50000 progress=67.8% loss=0.075360 lr=2.579648e-05 lr_backbone=2.579648e-05 elapsed=35:26:52 avg_step=3.77s recent_step=5.42s eta=16:51:02 +2026-09-16 01:57:43 | INFO | train rank=0 | step=33900/50000 progress=67.8% loss=0.078738 lr=2.576755e-05 lr_backbone=2.576755e-05 elapsed=35:27:21 avg_step=3.77s recent_step=2.82s eta=16:50:20 +2026-09-16 01:58:21 | INFO | train rank=0 | step=33910/50000 progress=67.8% loss=0.088594 lr=2.573863e-05 lr_backbone=2.573863e-05 elapsed=35:27:59 avg_step=3.77s recent_step=3.85s eta=16:49:42 +2026-09-16 01:58:44 | INFO | train rank=0 | step=33920/50000 progress=67.8% loss=0.098626 lr=2.570972e-05 lr_backbone=2.570972e-05 elapsed=35:28:22 avg_step=3.76s recent_step=2.26s eta=16:48:58 +2026-09-16 01:59:31 | INFO | train rank=0 | step=33930/50000 progress=67.9% loss=0.084847 lr=2.568082e-05 lr_backbone=2.568082e-05 elapsed=35:29:09 avg_step=3.77s recent_step=4.74s eta=16:48:25 +2026-09-16 02:00:06 | INFO | train rank=0 | step=33940/50000 progress=67.9% loss=0.080125 lr=2.565193e-05 lr_backbone=2.565193e-05 elapsed=35:29:44 avg_step=3.77s recent_step=3.46s eta=16:47:45 +2026-09-16 02:00:44 | INFO | train rank=0 | step=33950/50000 progress=67.9% loss=0.076194 lr=2.562306e-05 lr_backbone=2.562306e-05 elapsed=35:30:22 avg_step=3.77s recent_step=3.80s eta=16:47:08 +2026-09-16 02:01:17 | INFO | train rank=0 | step=33960/50000 progress=67.9% loss=0.071450 lr=2.559419e-05 lr_backbone=2.559419e-05 elapsed=35:30:55 avg_step=3.76s recent_step=3.34s eta=16:46:28 +2026-09-16 02:02:04 | INFO | train rank=0 | step=33970/50000 progress=67.9% loss=0.088480 lr=2.556533e-05 lr_backbone=2.556533e-05 elapsed=35:31:42 avg_step=3.77s recent_step=4.73s eta=16:45:55 +2026-09-16 02:02:36 | INFO | train rank=0 | step=33980/50000 progress=68.0% loss=0.079040 lr=2.553649e-05 lr_backbone=2.553649e-05 elapsed=35:32:14 avg_step=3.76s recent_step=3.15s eta=16:45:15 +2026-09-16 02:03:06 | INFO | train rank=0 | step=33990/50000 progress=68.0% loss=0.106351 lr=2.550765e-05 lr_backbone=2.550765e-05 elapsed=35:32:44 avg_step=3.76s recent_step=3.05s eta=16:44:34 +2026-09-16 02:03:43 | INFO | train rank=0 | step=34000/50000 progress=68.0% loss=0.068901 lr=2.547882e-05 lr_backbone=2.547882e-05 elapsed=35:33:21 avg_step=3.76s recent_step=3.65s eta=16:43:56 +2026-09-16 02:04:35 | INFO | train rank=0 | step=34010/50000 progress=68.0% loss=0.084868 lr=2.545001e-05 lr_backbone=2.545001e-05 elapsed=35:34:13 avg_step=3.77s recent_step=5.19s eta=16:43:25 +2026-09-16 02:05:19 | INFO | train rank=0 | step=34020/50000 progress=68.0% loss=0.089447 lr=2.542121e-05 lr_backbone=2.542121e-05 elapsed=35:34:57 avg_step=3.77s recent_step=4.37s eta=16:42:50 +2026-09-16 02:05:43 | INFO | train rank=0 | step=34030/50000 progress=68.1% loss=0.086201 lr=2.539241e-05 lr_backbone=2.539241e-05 elapsed=35:35:21 avg_step=3.76s recent_step=2.49s eta=16:42:06 +2026-09-16 02:06:16 | INFO | train rank=0 | step=34040/50000 progress=68.1% loss=0.107861 lr=2.536363e-05 lr_backbone=2.536363e-05 elapsed=35:35:54 avg_step=3.76s recent_step=3.25s eta=16:41:26 +2026-09-16 02:07:03 | INFO | train rank=0 | step=34050/50000 progress=68.1% loss=0.092129 lr=2.533486e-05 lr_backbone=2.533486e-05 elapsed=35:36:41 avg_step=3.77s recent_step=4.69s eta=16:40:53 +2026-09-16 02:07:39 | INFO | train rank=0 | step=34060/50000 progress=68.1% loss=0.082237 lr=2.530610e-05 lr_backbone=2.530610e-05 elapsed=35:37:17 avg_step=3.77s recent_step=3.63s eta=16:40:14 +2026-09-16 02:08:20 | INFO | train rank=0 | step=34070/50000 progress=68.1% loss=0.111838 lr=2.527735e-05 lr_backbone=2.527735e-05 elapsed=35:37:58 avg_step=3.77s recent_step=4.12s eta=16:39:38 +2026-09-16 02:08:50 | INFO | train rank=0 | step=34080/50000 progress=68.2% loss=0.082510 lr=2.524861e-05 lr_backbone=2.524861e-05 elapsed=35:38:28 avg_step=3.76s recent_step=2.95s eta=16:38:57 +2026-09-16 02:09:38 | INFO | train rank=0 | step=34090/50000 progress=68.2% loss=0.105862 lr=2.521989e-05 lr_backbone=2.521989e-05 elapsed=35:39:16 avg_step=3.77s recent_step=4.83s eta=16:38:24 +2026-09-16 02:10:18 | INFO | train rank=0 | step=34100/50000 progress=68.2% loss=0.077636 lr=2.519117e-05 lr_backbone=2.519117e-05 elapsed=35:39:56 avg_step=3.77s recent_step=3.96s eta=16:37:48 +2026-09-16 02:10:42 | INFO | train rank=0 | step=34110/50000 progress=68.2% loss=0.075151 lr=2.516246e-05 lr_backbone=2.516246e-05 elapsed=35:40:20 avg_step=3.76s recent_step=2.45s eta=16:37:04 +2026-09-16 02:11:21 | INFO | train rank=0 | step=34120/50000 progress=68.2% loss=0.089319 lr=2.513377e-05 lr_backbone=2.513377e-05 elapsed=35:40:59 avg_step=3.76s recent_step=3.82s eta=16:36:26 +2026-09-16 02:12:05 | INFO | train rank=0 | step=34130/50000 progress=68.3% loss=0.083604 lr=2.510508e-05 lr_backbone=2.510508e-05 elapsed=35:41:43 avg_step=3.77s recent_step=4.40s eta=16:35:52 +2026-09-16 02:12:38 | INFO | train rank=0 | step=34140/50000 progress=68.3% loss=0.083537 lr=2.507641e-05 lr_backbone=2.507641e-05 elapsed=35:42:16 avg_step=3.76s recent_step=3.29s eta=16:35:12 +2026-09-16 02:13:09 | INFO | train rank=0 | step=34150/50000 progress=68.3% loss=0.081531 lr=2.504775e-05 lr_backbone=2.504775e-05 elapsed=35:42:47 avg_step=3.76s recent_step=3.13s eta=16:34:31 +2026-09-16 02:13:43 | INFO | train rank=0 | step=34160/50000 progress=68.3% loss=0.073155 lr=2.501910e-05 lr_backbone=2.501910e-05 elapsed=35:43:21 avg_step=3.76s recent_step=3.39s eta=16:33:52 +2026-09-16 02:14:35 | INFO | train rank=0 | step=34170/50000 progress=68.3% loss=0.106170 lr=2.499045e-05 lr_backbone=2.499045e-05 elapsed=35:44:13 avg_step=3.77s recent_step=5.28s eta=16:33:21 +2026-09-16 02:15:11 | INFO | train rank=0 | step=34180/50000 progress=68.4% loss=0.068968 lr=2.496182e-05 lr_backbone=2.496182e-05 elapsed=35:44:49 avg_step=3.77s recent_step=3.57s eta=16:32:43 +2026-09-16 02:15:47 | INFO | train rank=0 | step=34190/50000 progress=68.4% loss=0.112647 lr=2.493321e-05 lr_backbone=2.493321e-05 elapsed=35:45:25 avg_step=3.76s recent_step=3.54s eta=16:32:04 +2026-09-16 02:16:16 | INFO | train rank=0 | step=34200/50000 progress=68.4% loss=0.083477 lr=2.490460e-05 lr_backbone=2.490460e-05 elapsed=35:45:54 avg_step=3.76s recent_step=2.90s eta=16:31:22 +2026-09-16 02:17:08 | INFO | train rank=0 | step=34210/50000 progress=68.4% loss=0.096977 lr=2.487600e-05 lr_backbone=2.487600e-05 elapsed=35:46:46 avg_step=3.77s recent_step=5.24s eta=16:30:52 +2026-09-16 02:17:41 | INFO | train rank=0 | step=34220/50000 progress=68.4% loss=0.081075 lr=2.484741e-05 lr_backbone=2.484741e-05 elapsed=35:47:19 avg_step=3.77s recent_step=3.35s eta=16:30:12 +2026-09-16 02:18:14 | INFO | train rank=0 | step=34230/50000 progress=68.5% loss=0.068418 lr=2.481884e-05 lr_backbone=2.481884e-05 elapsed=35:47:52 avg_step=3.76s recent_step=3.23s eta=16:29:32 +2026-09-16 02:18:55 | INFO | train rank=0 | step=34240/50000 progress=68.5% loss=0.092954 lr=2.479028e-05 lr_backbone=2.479028e-05 elapsed=35:48:33 avg_step=3.77s recent_step=4.15s eta=16:28:56 +2026-09-16 02:19:33 | INFO | train rank=0 | step=34250/50000 progress=68.5% loss=0.086437 lr=2.476172e-05 lr_backbone=2.476172e-05 elapsed=35:49:11 avg_step=3.76s recent_step=3.74s eta=16:28:18 +2026-09-16 02:20:09 | INFO | train rank=0 | step=34260/50000 progress=68.5% loss=0.079886 lr=2.473318e-05 lr_backbone=2.473318e-05 elapsed=35:49:47 avg_step=3.76s recent_step=3.66s eta=16:27:40 +2026-09-16 02:20:42 | INFO | train rank=0 | step=34270/50000 progress=68.5% loss=0.086579 lr=2.470465e-05 lr_backbone=2.470465e-05 elapsed=35:50:20 avg_step=3.76s recent_step=3.32s eta=16:27:00 +2026-09-16 02:21:17 | INFO | train rank=0 | step=34280/50000 progress=68.6% loss=0.101426 lr=2.467613e-05 lr_backbone=2.467613e-05 elapsed=35:50:55 avg_step=3.76s recent_step=3.45s eta=16:26:21 +2026-09-16 02:22:13 | INFO | train rank=0 | step=34290/50000 progress=68.6% loss=0.088813 lr=2.464762e-05 lr_backbone=2.464762e-05 elapsed=35:51:51 avg_step=3.77s recent_step=5.63s eta=16:25:52 +2026-09-16 02:22:52 | INFO | train rank=0 | step=34300/50000 progress=68.6% loss=0.099850 lr=2.461912e-05 lr_backbone=2.461912e-05 elapsed=35:52:30 avg_step=3.77s recent_step=3.83s eta=16:25:15 +2026-09-16 02:23:24 | INFO | train rank=0 | step=34310/50000 progress=68.6% loss=0.128695 lr=2.459064e-05 lr_backbone=2.459064e-05 elapsed=35:53:02 avg_step=3.77s recent_step=3.21s eta=16:24:35 +2026-09-16 02:23:56 | INFO | train rank=0 | step=34320/50000 progress=68.6% loss=0.088560 lr=2.456216e-05 lr_backbone=2.456216e-05 elapsed=35:53:34 avg_step=3.76s recent_step=3.23s eta=16:23:55 +2026-09-16 02:24:41 | INFO | train rank=0 | step=34330/50000 progress=68.7% loss=0.078406 lr=2.453370e-05 lr_backbone=2.453370e-05 elapsed=35:54:19 avg_step=3.77s recent_step=4.47s eta=16:23:20 +2026-09-16 02:25:11 | INFO | train rank=0 | step=34340/50000 progress=68.7% loss=0.073514 lr=2.450525e-05 lr_backbone=2.450525e-05 elapsed=35:54:49 avg_step=3.76s recent_step=3.05s eta=16:22:39 +2026-09-16 02:25:46 | INFO | train rank=0 | step=34350/50000 progress=68.7% loss=0.078442 lr=2.447680e-05 lr_backbone=2.447680e-05 elapsed=35:55:24 avg_step=3.76s recent_step=3.49s eta=16:22:00 +2026-09-16 02:26:18 | INFO | train rank=0 | step=34360/50000 progress=68.7% loss=0.107115 lr=2.444837e-05 lr_backbone=2.444837e-05 elapsed=35:55:56 avg_step=3.76s recent_step=3.15s eta=16:21:20 +2026-09-16 02:27:05 | INFO | train rank=0 | step=34370/50000 progress=68.7% loss=0.097093 lr=2.441995e-05 lr_backbone=2.441995e-05 elapsed=35:56:43 avg_step=3.77s recent_step=4.71s eta=16:20:47 +2026-09-16 02:27:33 | INFO | train rank=0 | step=34380/50000 progress=68.8% loss=0.092290 lr=2.439154e-05 lr_backbone=2.439154e-05 elapsed=35:57:11 avg_step=3.76s recent_step=2.79s eta=16:20:04 +2026-09-16 02:28:08 | INFO | train rank=0 | step=34390/50000 progress=68.8% loss=0.088427 lr=2.436315e-05 lr_backbone=2.436315e-05 elapsed=35:57:46 avg_step=3.76s recent_step=3.57s eta=16:19:26 +2026-09-16 02:28:41 | INFO | train rank=0 | step=34400/50000 progress=68.8% loss=0.103409 lr=2.433476e-05 lr_backbone=2.433476e-05 elapsed=35:58:19 avg_step=3.76s recent_step=3.29s eta=16:18:46 +2026-09-16 02:29:35 | INFO | train rank=0 | step=34410/50000 progress=68.8% loss=0.091643 lr=2.430639e-05 lr_backbone=2.430639e-05 elapsed=35:59:13 avg_step=3.76s recent_step=5.36s eta=16:18:16 +2026-09-16 02:30:13 | INFO | train rank=0 | step=34420/50000 progress=68.8% loss=0.073201 lr=2.427802e-05 lr_backbone=2.427802e-05 elapsed=35:59:51 avg_step=3.77s recent_step=3.80s eta=16:17:38 +2026-09-16 02:30:50 | INFO | train rank=0 | step=34430/50000 progress=68.9% loss=0.082113 lr=2.424967e-05 lr_backbone=2.424967e-05 elapsed=36:00:28 avg_step=3.76s recent_step=3.74s eta=16:17:00 +2026-09-16 02:31:29 | INFO | train rank=0 | step=34440/50000 progress=68.9% loss=0.133803 lr=2.422133e-05 lr_backbone=2.422133e-05 elapsed=36:01:07 avg_step=3.77s recent_step=3.87s eta=16:16:23 +2026-09-16 02:32:05 | INFO | train rank=0 | step=34450/50000 progress=68.9% loss=0.110931 lr=2.419300e-05 lr_backbone=2.419300e-05 elapsed=36:01:43 avg_step=3.76s recent_step=3.57s eta=16:15:45 +2026-09-16 02:32:38 | INFO | train rank=0 | step=34460/50000 progress=68.9% loss=0.090785 lr=2.416468e-05 lr_backbone=2.416468e-05 elapsed=36:02:16 avg_step=3.76s recent_step=3.29s eta=16:15:05 +2026-09-16 02:33:09 | INFO | train rank=0 | step=34470/50000 progress=68.9% loss=0.080799 lr=2.413637e-05 lr_backbone=2.413637e-05 elapsed=36:02:47 avg_step=3.76s recent_step=3.12s eta=16:14:24 +2026-09-16 02:33:44 | INFO | train rank=0 | step=34480/50000 progress=69.0% loss=0.099922 lr=2.410808e-05 lr_backbone=2.410808e-05 elapsed=36:03:22 avg_step=3.76s recent_step=3.56s eta=16:13:46 +2026-09-16 02:34:43 | INFO | train rank=0 | step=34490/50000 progress=69.0% loss=0.080504 lr=2.407979e-05 lr_backbone=2.407979e-05 elapsed=36:04:21 avg_step=3.77s recent_step=5.89s eta=16:13:18 +2026-09-16 02:35:15 | INFO | train rank=0 | step=34500/50000 progress=69.0% loss=0.089056 lr=2.405152e-05 lr_backbone=2.405152e-05 elapsed=36:04:53 avg_step=3.77s recent_step=3.14s eta=16:12:37 +2026-09-16 02:35:47 | INFO | train rank=0 | step=34510/50000 progress=69.0% loss=0.078177 lr=2.402326e-05 lr_backbone=2.402326e-05 elapsed=36:05:25 avg_step=3.76s recent_step=3.22s eta=16:11:57 +2026-09-16 02:36:18 | INFO | train rank=0 | step=34520/50000 progress=69.0% loss=0.111143 lr=2.399501e-05 lr_backbone=2.399501e-05 elapsed=36:05:56 avg_step=3.76s recent_step=3.14s eta=16:11:17 +2026-09-16 02:37:12 | INFO | train rank=0 | step=34530/50000 progress=69.1% loss=0.084187 lr=2.396677e-05 lr_backbone=2.396677e-05 elapsed=36:06:50 avg_step=3.77s recent_step=5.34s eta=16:10:46 +2026-09-16 02:37:40 | INFO | train rank=0 | step=34540/50000 progress=69.1% loss=0.107707 lr=2.393854e-05 lr_backbone=2.393854e-05 elapsed=36:07:18 avg_step=3.76s recent_step=2.80s eta=16:10:04 +2026-09-16 02:38:15 | INFO | train rank=0 | step=34550/50000 progress=69.1% loss=0.091768 lr=2.391033e-05 lr_backbone=2.391033e-05 elapsed=36:07:53 avg_step=3.76s recent_step=3.55s eta=16:09:26 +2026-09-16 02:38:47 | INFO | train rank=0 | step=34560/50000 progress=69.1% loss=0.123824 lr=2.388212e-05 lr_backbone=2.388212e-05 elapsed=36:08:25 avg_step=3.76s recent_step=3.15s eta=16:08:45 +2026-09-16 02:39:40 | INFO | train rank=0 | step=34570/50000 progress=69.1% loss=0.090625 lr=2.385393e-05 lr_backbone=2.385393e-05 elapsed=36:09:18 avg_step=3.77s recent_step=5.35s eta=16:08:15 +2026-09-16 02:40:09 | INFO | train rank=0 | step=34580/50000 progress=69.2% loss=0.098273 lr=2.382575e-05 lr_backbone=2.382575e-05 elapsed=36:09:47 avg_step=3.76s recent_step=2.84s eta=16:07:33 +2026-09-16 02:40:31 | INFO | train rank=0 | step=34590/50000 progress=69.2% loss=0.086811 lr=2.379758e-05 lr_backbone=2.379758e-05 elapsed=36:10:09 avg_step=3.76s recent_step=2.28s eta=16:06:49 +2026-09-16 02:41:07 | INFO | train rank=0 | step=34600/50000 progress=69.2% loss=0.096459 lr=2.376942e-05 lr_backbone=2.376942e-05 elapsed=36:10:45 avg_step=3.76s recent_step=3.61s eta=16:06:10 +2026-09-16 02:41:55 | INFO | train rank=0 | step=34610/50000 progress=69.2% loss=0.092228 lr=2.374127e-05 lr_backbone=2.374127e-05 elapsed=36:11:33 avg_step=3.76s recent_step=4.75s eta=16:05:37 +2026-09-16 02:42:32 | INFO | train rank=0 | step=34620/50000 progress=69.2% loss=0.096629 lr=2.371313e-05 lr_backbone=2.371313e-05 elapsed=36:12:10 avg_step=3.76s recent_step=3.66s eta=16:04:59 +2026-09-16 02:43:07 | INFO | train rank=0 | step=34630/50000 progress=69.3% loss=0.082710 lr=2.368501e-05 lr_backbone=2.368501e-05 elapsed=36:12:45 avg_step=3.76s recent_step=3.52s eta=16:04:20 +2026-09-16 02:43:41 | INFO | train rank=0 | step=34640/50000 progress=69.3% loss=0.101855 lr=2.365689e-05 lr_backbone=2.365689e-05 elapsed=36:13:19 avg_step=3.76s recent_step=3.39s eta=16:03:41 +2026-09-16 02:44:26 | INFO | train rank=0 | step=34650/50000 progress=69.3% loss=0.074535 lr=2.362879e-05 lr_backbone=2.362879e-05 elapsed=36:14:04 avg_step=3.76s recent_step=4.54s eta=16:03:07 +2026-09-16 02:44:59 | INFO | train rank=0 | step=34660/50000 progress=69.3% loss=0.082849 lr=2.360070e-05 lr_backbone=2.360070e-05 elapsed=36:14:37 avg_step=3.76s recent_step=3.24s eta=16:02:27 +2026-09-16 02:45:36 | INFO | train rank=0 | step=34670/50000 progress=69.3% loss=0.090334 lr=2.357262e-05 lr_backbone=2.357262e-05 elapsed=36:15:14 avg_step=3.76s recent_step=3.70s eta=16:01:49 +2026-09-16 02:46:12 | INFO | train rank=0 | step=34680/50000 progress=69.4% loss=0.111457 lr=2.354456e-05 lr_backbone=2.354456e-05 elapsed=36:15:50 avg_step=3.76s recent_step=3.68s eta=16:01:11 +2026-09-16 02:46:53 | INFO | train rank=0 | step=34690/50000 progress=69.4% loss=0.085427 lr=2.351650e-05 lr_backbone=2.351650e-05 elapsed=36:16:31 avg_step=3.76s recent_step=4.11s eta=16:00:35 +2026-09-16 02:47:33 | INFO | train rank=0 | step=34700/50000 progress=69.4% loss=0.088583 lr=2.348846e-05 lr_backbone=2.348846e-05 elapsed=36:17:11 avg_step=3.76s recent_step=3.92s eta=15:59:58 +2026-09-16 02:48:01 | INFO | train rank=0 | step=34710/50000 progress=69.4% loss=0.088096 lr=2.346043e-05 lr_backbone=2.346043e-05 elapsed=36:17:39 avg_step=3.76s recent_step=2.87s eta=15:59:16 +2026-09-16 02:48:35 | INFO | train rank=0 | step=34720/50000 progress=69.4% loss=0.096558 lr=2.343241e-05 lr_backbone=2.343241e-05 elapsed=36:18:13 avg_step=3.76s recent_step=3.40s eta=15:58:37 +2026-09-16 02:49:28 | INFO | train rank=0 | step=34730/50000 progress=69.5% loss=0.092409 lr=2.340440e-05 lr_backbone=2.340440e-05 elapsed=36:19:06 avg_step=3.76s recent_step=5.22s eta=15:58:06 +2026-09-16 02:50:10 | INFO | train rank=0 | step=34740/50000 progress=69.5% loss=0.074694 lr=2.337640e-05 lr_backbone=2.337640e-05 elapsed=36:19:48 avg_step=3.76s recent_step=4.25s eta=15:57:30 +2026-09-16 02:50:38 | INFO | train rank=0 | step=34750/50000 progress=69.5% loss=0.083568 lr=2.334841e-05 lr_backbone=2.334841e-05 elapsed=36:20:16 avg_step=3.76s recent_step=2.80s eta=15:56:48 +2026-09-16 02:51:10 | INFO | train rank=0 | step=34760/50000 progress=69.5% loss=0.095418 lr=2.332044e-05 lr_backbone=2.332044e-05 elapsed=36:20:48 avg_step=3.76s recent_step=3.22s eta=15:56:08 +2026-09-16 02:52:03 | INFO | train rank=0 | step=34770/50000 progress=69.5% loss=0.079511 lr=2.329248e-05 lr_backbone=2.329248e-05 elapsed=36:21:41 avg_step=3.76s recent_step=5.29s eta=15:55:37 +2026-09-16 02:52:37 | INFO | train rank=0 | step=34780/50000 progress=69.6% loss=0.102843 lr=2.326453e-05 lr_backbone=2.326453e-05 elapsed=36:22:15 avg_step=3.76s recent_step=3.34s eta=15:54:58 +2026-09-16 02:53:13 | INFO | train rank=0 | step=34790/50000 progress=69.6% loss=0.091838 lr=2.323659e-05 lr_backbone=2.323659e-05 elapsed=36:22:51 avg_step=3.76s recent_step=3.67s eta=15:54:20 +2026-09-16 02:53:40 | INFO | train rank=0 | step=34800/50000 progress=69.6% loss=0.081612 lr=2.320866e-05 lr_backbone=2.320866e-05 elapsed=36:23:18 avg_step=3.76s recent_step=2.65s eta=15:53:37 +2026-09-16 02:54:26 | INFO | train rank=0 | step=34810/50000 progress=69.6% loss=0.098303 lr=2.318074e-05 lr_backbone=2.318074e-05 elapsed=36:24:04 avg_step=3.76s recent_step=4.67s eta=15:53:03 +2026-09-16 02:55:01 | INFO | train rank=0 | step=34820/50000 progress=69.6% loss=0.084601 lr=2.315284e-05 lr_backbone=2.315284e-05 elapsed=36:24:39 avg_step=3.76s recent_step=3.49s eta=15:52:25 +2026-09-16 02:55:38 | INFO | train rank=0 | step=34830/50000 progress=69.7% loss=0.088047 lr=2.312495e-05 lr_backbone=2.312495e-05 elapsed=36:25:16 avg_step=3.76s recent_step=3.69s eta=15:51:47 +2026-09-16 02:56:11 | INFO | train rank=0 | step=34840/50000 progress=69.7% loss=0.095614 lr=2.309707e-05 lr_backbone=2.309707e-05 elapsed=36:25:49 avg_step=3.76s recent_step=3.29s eta=15:51:07 +2026-09-16 02:56:56 | INFO | train rank=0 | step=34850/50000 progress=69.7% loss=0.090823 lr=2.306920e-05 lr_backbone=2.306920e-05 elapsed=36:26:34 avg_step=3.76s recent_step=4.46s eta=15:50:32 +2026-09-16 02:57:28 | INFO | train rank=0 | step=34860/50000 progress=69.7% loss=0.099408 lr=2.304134e-05 lr_backbone=2.304134e-05 elapsed=36:27:06 avg_step=3.76s recent_step=3.20s eta=15:49:52 +2026-09-16 02:58:00 | INFO | train rank=0 | step=34870/50000 progress=69.7% loss=0.096378 lr=2.301350e-05 lr_backbone=2.301350e-05 elapsed=36:27:38 avg_step=3.76s recent_step=3.23s eta=15:49:12 +2026-09-16 02:58:33 | INFO | train rank=0 | step=34880/50000 progress=69.8% loss=0.086437 lr=2.298566e-05 lr_backbone=2.298566e-05 elapsed=36:28:11 avg_step=3.76s recent_step=3.26s eta=15:48:32 +2026-09-16 02:59:14 | INFO | train rank=0 | step=34890/50000 progress=69.8% loss=0.105052 lr=2.295784e-05 lr_backbone=2.295784e-05 elapsed=36:28:52 avg_step=3.76s recent_step=4.14s eta=15:47:56 +2026-09-16 02:59:46 | INFO | train rank=0 | step=34900/50000 progress=69.8% loss=0.085402 lr=2.293003e-05 lr_backbone=2.293003e-05 elapsed=36:29:24 avg_step=3.76s recent_step=3.17s eta=15:47:16 +2026-09-16 03:00:21 | INFO | train rank=0 | step=34910/50000 progress=69.8% loss=0.112450 lr=2.290224e-05 lr_backbone=2.290224e-05 elapsed=36:29:59 avg_step=3.76s recent_step=3.53s eta=15:46:38 +2026-09-16 03:00:57 | INFO | train rank=0 | step=34920/50000 progress=69.8% loss=0.093535 lr=2.287445e-05 lr_backbone=2.287445e-05 elapsed=36:30:35 avg_step=3.76s recent_step=3.64s eta=15:45:59 +2026-09-16 03:01:58 | INFO | train rank=0 | step=34930/50000 progress=69.9% loss=0.101530 lr=2.284668e-05 lr_backbone=2.284668e-05 elapsed=36:31:36 avg_step=3.76s recent_step=6.01s eta=15:45:31 +2026-09-16 03:02:38 | INFO | train rank=0 | step=34940/50000 progress=69.9% loss=0.109083 lr=2.281891e-05 lr_backbone=2.281891e-05 elapsed=36:32:16 avg_step=3.76s recent_step=4.01s eta=15:44:55 +2026-09-16 03:03:05 | INFO | train rank=0 | step=34950/50000 progress=69.9% loss=0.068514 lr=2.279116e-05 lr_backbone=2.279116e-05 elapsed=36:32:43 avg_step=3.76s recent_step=2.74s eta=15:44:13 +2026-09-16 03:03:37 | INFO | train rank=0 | step=34960/50000 progress=69.9% loss=0.100603 lr=2.276343e-05 lr_backbone=2.276343e-05 elapsed=36:33:15 avg_step=3.76s recent_step=3.15s eta=15:43:32 +2026-09-16 03:04:29 | INFO | train rank=0 | step=34970/50000 progress=69.9% loss=0.091713 lr=2.273570e-05 lr_backbone=2.273570e-05 elapsed=36:34:07 avg_step=3.76s recent_step=5.23s eta=15:43:01 +2026-09-16 03:05:03 | INFO | train rank=0 | step=34980/50000 progress=70.0% loss=0.080532 lr=2.270799e-05 lr_backbone=2.270799e-05 elapsed=36:34:41 avg_step=3.76s recent_step=3.38s eta=15:42:22 +2026-09-16 03:05:36 | INFO | train rank=0 | step=34990/50000 progress=70.0% loss=0.085737 lr=2.268028e-05 lr_backbone=2.268028e-05 elapsed=36:35:14 avg_step=3.76s recent_step=3.35s eta=15:41:42 +2026-09-16 03:06:06 | INFO | train rank=0 | step=35000/50000 progress=70.0% loss=0.091693 lr=2.265259e-05 lr_backbone=2.265259e-05 elapsed=36:35:44 avg_step=3.76s recent_step=2.95s eta=15:41:01 +2026-09-16 03:07:01 | INFO | train rank=0 | step=35010/50000 progress=70.0% loss=0.088764 lr=2.262491e-05 lr_backbone=2.262491e-05 elapsed=36:36:39 avg_step=3.76s recent_step=5.54s eta=15:40:31 +2026-09-16 03:07:39 | INFO | train rank=0 | step=35020/50000 progress=70.0% loss=0.083920 lr=2.259725e-05 lr_backbone=2.259725e-05 elapsed=36:37:17 avg_step=3.76s recent_step=3.76s eta=15:39:54 +2026-09-16 03:08:14 | INFO | train rank=0 | step=35030/50000 progress=70.1% loss=0.091986 lr=2.256959e-05 lr_backbone=2.256959e-05 elapsed=36:37:52 avg_step=3.76s recent_step=3.56s eta=15:39:15 +2026-09-16 03:08:46 | INFO | train rank=0 | step=35040/50000 progress=70.1% loss=0.075210 lr=2.254195e-05 lr_backbone=2.254195e-05 elapsed=36:38:24 avg_step=3.76s recent_step=3.19s eta=15:38:35 +2026-09-16 03:09:33 | INFO | train rank=0 | step=35050/50000 progress=70.1% loss=0.083952 lr=2.251432e-05 lr_backbone=2.251432e-05 elapsed=36:39:11 avg_step=3.76s recent_step=4.64s eta=15:38:01 +2026-09-16 03:10:05 | INFO | train rank=0 | step=35060/50000 progress=70.1% loss=0.109180 lr=2.248670e-05 lr_backbone=2.248670e-05 elapsed=36:39:43 avg_step=3.76s recent_step=3.20s eta=15:37:21 +2026-09-16 03:10:41 | INFO | train rank=0 | step=35070/50000 progress=70.1% loss=0.139643 lr=2.245909e-05 lr_backbone=2.245909e-05 elapsed=36:40:19 avg_step=3.76s recent_step=3.66s eta=15:36:43 +2026-09-16 03:11:04 | INFO | train rank=0 | step=35080/50000 progress=70.2% loss=0.092797 lr=2.243150e-05 lr_backbone=2.243150e-05 elapsed=36:40:42 avg_step=3.76s recent_step=2.25s eta=15:35:59 +2026-09-16 03:11:45 | INFO | train rank=0 | step=35090/50000 progress=70.2% loss=0.092970 lr=2.240392e-05 lr_backbone=2.240392e-05 elapsed=36:41:23 avg_step=3.76s recent_step=4.12s eta=15:35:23 +2026-09-16 03:12:18 | INFO | train rank=0 | step=35100/50000 progress=70.2% loss=0.094283 lr=2.237635e-05 lr_backbone=2.237635e-05 elapsed=36:41:56 avg_step=3.76s recent_step=3.27s eta=15:34:43 +2026-09-16 03:12:57 | INFO | train rank=0 | step=35110/50000 progress=70.2% loss=0.070642 lr=2.234879e-05 lr_backbone=2.234879e-05 elapsed=36:42:35 avg_step=3.76s recent_step=3.91s eta=15:34:06 +2026-09-16 03:13:29 | INFO | train rank=0 | step=35120/50000 progress=70.2% loss=0.081141 lr=2.232124e-05 lr_backbone=2.232124e-05 elapsed=36:43:07 avg_step=3.76s recent_step=3.22s eta=15:33:26 +2026-09-16 03:14:27 | INFO | train rank=0 | step=35130/50000 progress=70.3% loss=0.083204 lr=2.229371e-05 lr_backbone=2.229371e-05 elapsed=36:44:05 avg_step=3.76s recent_step=5.84s eta=15:32:57 +2026-09-16 03:15:00 | INFO | train rank=0 | step=35140/50000 progress=70.3% loss=0.078153 lr=2.226619e-05 lr_backbone=2.226619e-05 elapsed=36:44:38 avg_step=3.76s recent_step=3.31s eta=15:32:18 +2026-09-16 03:15:29 | INFO | train rank=0 | step=35150/50000 progress=70.3% loss=0.093974 lr=2.223868e-05 lr_backbone=2.223868e-05 elapsed=36:45:07 avg_step=3.76s recent_step=2.89s eta=15:31:36 +2026-09-16 03:15:57 | INFO | train rank=0 | step=35160/50000 progress=70.3% loss=0.097082 lr=2.221118e-05 lr_backbone=2.221118e-05 elapsed=36:45:35 avg_step=3.76s recent_step=2.81s eta=15:30:55 +2026-09-16 03:16:41 | INFO | train rank=0 | step=35170/50000 progress=70.3% loss=0.108045 lr=2.218369e-05 lr_backbone=2.218369e-05 elapsed=36:46:19 avg_step=3.76s recent_step=4.35s eta=15:30:19 +2026-09-16 03:17:05 | INFO | train rank=0 | step=35180/50000 progress=70.4% loss=0.094235 lr=2.215622e-05 lr_backbone=2.215622e-05 elapsed=36:46:43 avg_step=3.76s recent_step=2.45s eta=15:29:36 +2026-09-16 03:17:39 | INFO | train rank=0 | step=35190/50000 progress=70.4% loss=0.107790 lr=2.212876e-05 lr_backbone=2.212876e-05 elapsed=36:47:17 avg_step=3.76s recent_step=3.34s eta=15:28:57 +2026-09-16 03:18:28 | INFO | train rank=0 | step=35200/50000 progress=70.4% loss=0.083760 lr=2.210131e-05 lr_backbone=2.210131e-05 elapsed=36:48:06 avg_step=3.76s recent_step=4.94s eta=15:28:24 +2026-09-16 03:19:05 | INFO | train rank=0 | step=35210/50000 progress=70.4% loss=0.078392 lr=2.207387e-05 lr_backbone=2.207387e-05 elapsed=36:48:43 avg_step=3.76s recent_step=3.67s eta=15:27:46 +2026-09-16 03:19:44 | INFO | train rank=0 | step=35220/50000 progress=70.4% loss=0.092658 lr=2.204645e-05 lr_backbone=2.204645e-05 elapsed=36:49:22 avg_step=3.76s recent_step=3.87s eta=15:27:09 +2026-09-16 03:20:15 | INFO | train rank=0 | step=35230/50000 progress=70.5% loss=0.092030 lr=2.201904e-05 lr_backbone=2.201904e-05 elapsed=36:49:53 avg_step=3.76s recent_step=3.10s eta=15:26:29 +2026-09-16 03:21:15 | INFO | train rank=0 | step=35240/50000 progress=70.5% loss=0.107235 lr=2.199164e-05 lr_backbone=2.199164e-05 elapsed=36:50:53 avg_step=3.76s recent_step=6.01s eta=15:26:00 +2026-09-16 03:21:54 | INFO | train rank=0 | step=35250/50000 progress=70.5% loss=0.099825 lr=2.196425e-05 lr_backbone=2.196425e-05 elapsed=36:51:32 avg_step=3.76s recent_step=3.90s eta=15:25:23 +2026-09-16 03:22:24 | INFO | train rank=0 | step=35260/50000 progress=70.5% loss=0.083392 lr=2.193687e-05 lr_backbone=2.193687e-05 elapsed=36:52:02 avg_step=3.76s recent_step=3.00s eta=15:24:42 +2026-09-16 03:22:55 | INFO | train rank=0 | step=35270/50000 progress=70.5% loss=0.092806 lr=2.190951e-05 lr_backbone=2.190951e-05 elapsed=36:52:33 avg_step=3.76s recent_step=3.15s eta=15:24:02 +2026-09-16 03:23:39 | INFO | train rank=0 | step=35280/50000 progress=70.6% loss=0.085404 lr=2.188216e-05 lr_backbone=2.188216e-05 elapsed=36:53:17 avg_step=3.76s recent_step=4.34s eta=15:23:27 +2026-09-16 03:24:17 | INFO | train rank=0 | step=35290/50000 progress=70.6% loss=0.081168 lr=2.185482e-05 lr_backbone=2.185482e-05 elapsed=36:53:55 avg_step=3.76s recent_step=3.87s eta=15:22:50 +2026-09-16 03:24:43 | INFO | train rank=0 | step=35300/50000 progress=70.6% loss=0.084860 lr=2.182749e-05 lr_backbone=2.182749e-05 elapsed=36:54:21 avg_step=3.76s recent_step=2.62s eta=15:22:07 +2026-09-16 03:25:18 | INFO | train rank=0 | step=35310/50000 progress=70.6% loss=0.113281 lr=2.180018e-05 lr_backbone=2.180018e-05 elapsed=36:54:56 avg_step=3.76s recent_step=3.45s eta=15:21:28 +2026-09-16 03:26:01 | INFO | train rank=0 | step=35320/50000 progress=70.6% loss=0.100679 lr=2.177288e-05 lr_backbone=2.177288e-05 elapsed=36:55:39 avg_step=3.76s recent_step=4.35s eta=15:20:53 +2026-09-16 03:26:40 | INFO | train rank=0 | step=35330/50000 progress=70.7% loss=0.091869 lr=2.174559e-05 lr_backbone=2.174559e-05 elapsed=36:56:18 avg_step=3.76s recent_step=3.89s eta=15:20:16 +2026-09-16 03:27:13 | INFO | train rank=0 | step=35340/50000 progress=70.7% loss=0.070292 lr=2.171831e-05 lr_backbone=2.171831e-05 elapsed=36:56:51 avg_step=3.76s recent_step=3.27s eta=15:19:36 +2026-09-16 03:27:51 | INFO | train rank=0 | step=35350/50000 progress=70.7% loss=0.100061 lr=2.169104e-05 lr_backbone=2.169104e-05 elapsed=36:57:29 avg_step=3.76s recent_step=3.77s eta=15:18:59 +2026-09-16 03:28:34 | INFO | train rank=0 | step=35360/50000 progress=70.7% loss=0.135111 lr=2.166379e-05 lr_backbone=2.166379e-05 elapsed=36:58:12 avg_step=3.76s recent_step=4.33s eta=15:18:23 +2026-09-16 03:29:13 | INFO | train rank=0 | step=35370/50000 progress=70.7% loss=0.124015 lr=2.163655e-05 lr_backbone=2.163655e-05 elapsed=36:58:51 avg_step=3.76s recent_step=3.94s eta=15:17:47 +2026-09-16 03:29:45 | INFO | train rank=0 | step=35380/50000 progress=70.8% loss=0.089502 lr=2.160933e-05 lr_backbone=2.160933e-05 elapsed=36:59:23 avg_step=3.76s recent_step=3.10s eta=15:17:06 +2026-09-16 03:30:15 | INFO | train rank=0 | step=35390/50000 progress=70.8% loss=0.078234 lr=2.158211e-05 lr_backbone=2.158211e-05 elapsed=36:59:53 avg_step=3.76s recent_step=3.02s eta=15:16:25 +2026-09-16 03:31:10 | INFO | train rank=0 | step=35400/50000 progress=70.8% loss=0.095504 lr=2.155491e-05 lr_backbone=2.155491e-05 elapsed=37:00:48 avg_step=3.76s recent_step=5.58s eta=15:15:55 +2026-09-16 03:31:48 | INFO | train rank=0 | step=35410/50000 progress=70.8% loss=0.088801 lr=2.152772e-05 lr_backbone=2.152772e-05 elapsed=37:01:26 avg_step=3.76s recent_step=3.75s eta=15:15:18 +2026-09-16 03:32:25 | INFO | train rank=0 | step=35420/50000 progress=70.8% loss=0.085677 lr=2.150054e-05 lr_backbone=2.150054e-05 elapsed=37:02:03 avg_step=3.76s recent_step=3.66s eta=15:14:40 +2026-09-16 03:32:51 | INFO | train rank=0 | step=35430/50000 progress=70.9% loss=0.107369 lr=2.147337e-05 lr_backbone=2.147337e-05 elapsed=37:02:29 avg_step=3.76s recent_step=2.64s eta=15:13:57 +2026-09-16 03:33:46 | INFO | train rank=0 | step=35440/50000 progress=70.9% loss=0.084325 lr=2.144622e-05 lr_backbone=2.144622e-05 elapsed=37:03:24 avg_step=3.76s recent_step=5.54s eta=15:13:27 +2026-09-16 03:34:24 | INFO | train rank=0 | step=35450/50000 progress=70.9% loss=0.082713 lr=2.141908e-05 lr_backbone=2.141908e-05 elapsed=37:04:02 avg_step=3.76s recent_step=3.77s eta=15:12:49 +2026-09-16 03:34:49 | INFO | train rank=0 | step=35460/50000 progress=70.9% loss=0.085396 lr=2.139195e-05 lr_backbone=2.139195e-05 elapsed=37:04:27 avg_step=3.76s recent_step=2.52s eta=15:12:07 +2026-09-16 03:35:24 | INFO | train rank=0 | step=35470/50000 progress=70.9% loss=0.081218 lr=2.136484e-05 lr_backbone=2.136484e-05 elapsed=37:05:02 avg_step=3.76s recent_step=3.47s eta=15:11:28 +2026-09-16 03:36:17 | INFO | train rank=0 | step=35480/50000 progress=71.0% loss=0.092783 lr=2.133773e-05 lr_backbone=2.133773e-05 elapsed=37:05:55 avg_step=3.76s recent_step=5.31s eta=15:10:56 +2026-09-16 03:36:55 | INFO | train rank=0 | step=35490/50000 progress=71.0% loss=0.099564 lr=2.131064e-05 lr_backbone=2.131064e-05 elapsed=37:06:33 avg_step=3.76s recent_step=3.78s eta=15:10:19 +2026-09-16 03:37:25 | INFO | train rank=0 | step=35500/50000 progress=71.0% loss=0.081062 lr=2.128357e-05 lr_backbone=2.128357e-05 elapsed=37:07:03 avg_step=3.76s recent_step=3.06s eta=15:09:38 +2026-09-16 03:37:56 | INFO | train rank=0 | step=35510/50000 progress=71.0% loss=0.081939 lr=2.125650e-05 lr_backbone=2.125650e-05 elapsed=37:07:34 avg_step=3.76s recent_step=3.07s eta=15:08:58 +2026-09-16 03:38:36 | INFO | train rank=0 | step=35520/50000 progress=71.0% loss=0.068577 lr=2.122945e-05 lr_backbone=2.122945e-05 elapsed=37:08:14 avg_step=3.76s recent_step=3.99s eta=15:08:21 +2026-09-16 03:39:12 | INFO | train rank=0 | step=35530/50000 progress=71.1% loss=0.099838 lr=2.120241e-05 lr_backbone=2.120241e-05 elapsed=37:08:50 avg_step=3.76s recent_step=3.60s eta=15:07:43 +2026-09-16 03:39:49 | INFO | train rank=0 | step=35540/50000 progress=71.1% loss=0.085124 lr=2.117538e-05 lr_backbone=2.117538e-05 elapsed=37:09:27 avg_step=3.76s recent_step=3.68s eta=15:07:05 +2026-09-16 03:40:24 | INFO | train rank=0 | step=35550/50000 progress=71.1% loss=0.110427 lr=2.114837e-05 lr_backbone=2.114837e-05 elapsed=37:10:02 avg_step=3.76s recent_step=3.48s eta=15:06:26 +2026-09-16 03:41:07 | INFO | train rank=0 | step=35560/50000 progress=71.1% loss=0.108467 lr=2.112136e-05 lr_backbone=2.112136e-05 elapsed=37:10:45 avg_step=3.76s recent_step=4.28s eta=15:05:51 +2026-09-16 03:41:44 | INFO | train rank=0 | step=35570/50000 progress=71.1% loss=0.089705 lr=2.109438e-05 lr_backbone=2.109438e-05 elapsed=37:11:22 avg_step=3.76s recent_step=3.70s eta=15:05:13 +2026-09-16 03:42:19 | INFO | train rank=0 | step=35580/50000 progress=71.2% loss=0.087329 lr=2.106740e-05 lr_backbone=2.106740e-05 elapsed=37:11:57 avg_step=3.76s recent_step=3.58s eta=15:04:34 +2026-09-16 03:42:49 | INFO | train rank=0 | step=35590/50000 progress=71.2% loss=0.113224 lr=2.104043e-05 lr_backbone=2.104043e-05 elapsed=37:12:27 avg_step=3.76s recent_step=2.94s eta=15:03:53 +2026-09-16 03:43:46 | INFO | train rank=0 | step=35600/50000 progress=71.2% loss=0.081398 lr=2.101348e-05 lr_backbone=2.101348e-05 elapsed=37:13:24 avg_step=3.76s recent_step=5.76s eta=15:03:24 +2026-09-16 03:44:12 | INFO | train rank=0 | step=35610/50000 progress=71.2% loss=0.081942 lr=2.098654e-05 lr_backbone=2.098654e-05 elapsed=37:13:50 avg_step=3.76s recent_step=2.57s eta=15:02:41 +2026-09-16 03:44:50 | INFO | train rank=0 | step=35620/50000 progress=71.2% loss=0.108098 lr=2.095962e-05 lr_backbone=2.095962e-05 elapsed=37:14:28 avg_step=3.76s recent_step=3.80s eta=15:02:04 +2026-09-16 03:45:24 | INFO | train rank=0 | step=35630/50000 progress=71.3% loss=0.073283 lr=2.093270e-05 lr_backbone=2.093270e-05 elapsed=37:15:02 avg_step=3.76s recent_step=3.36s eta=15:01:25 +2026-09-16 03:46:11 | INFO | train rank=0 | step=35640/50000 progress=71.3% loss=0.070027 lr=2.090580e-05 lr_backbone=2.090580e-05 elapsed=37:15:49 avg_step=3.76s recent_step=4.73s eta=15:00:51 +2026-09-16 03:46:49 | INFO | train rank=0 | step=35650/50000 progress=71.3% loss=0.087088 lr=2.087892e-05 lr_backbone=2.087892e-05 elapsed=37:16:27 avg_step=3.76s recent_step=3.80s eta=15:00:13 +2026-09-16 03:47:22 | INFO | train rank=0 | step=35660/50000 progress=71.3% loss=0.091555 lr=2.085204e-05 lr_backbone=2.085204e-05 elapsed=37:17:00 avg_step=3.76s recent_step=3.34s eta=14:59:34 +2026-09-16 03:48:01 | INFO | train rank=0 | step=35670/50000 progress=71.3% loss=0.114083 lr=2.082518e-05 lr_backbone=2.082518e-05 elapsed=37:17:39 avg_step=3.76s recent_step=3.83s eta=14:58:57 +2026-09-16 03:48:57 | INFO | train rank=0 | step=35680/50000 progress=71.4% loss=0.075157 lr=2.079833e-05 lr_backbone=2.079833e-05 elapsed=37:18:35 avg_step=3.76s recent_step=5.60s eta=14:58:26 +2026-09-16 03:49:36 | INFO | train rank=0 | step=35690/50000 progress=71.4% loss=0.083586 lr=2.077149e-05 lr_backbone=2.077149e-05 elapsed=37:19:14 avg_step=3.76s recent_step=3.96s eta=14:57:49 +2026-09-16 03:50:07 | INFO | train rank=0 | step=35700/50000 progress=71.4% loss=0.085383 lr=2.074467e-05 lr_backbone=2.074467e-05 elapsed=37:19:45 avg_step=3.76s recent_step=3.04s eta=14:57:09 +2026-09-16 03:50:38 | INFO | train rank=0 | step=35710/50000 progress=71.4% loss=0.086501 lr=2.071785e-05 lr_backbone=2.071785e-05 elapsed=37:20:16 avg_step=3.76s recent_step=3.16s eta=14:56:29 +2026-09-16 03:51:27 | INFO | train rank=0 | step=35720/50000 progress=71.4% loss=0.119252 lr=2.069106e-05 lr_backbone=2.069106e-05 elapsed=37:21:05 avg_step=3.76s recent_step=4.89s eta=14:55:56 +2026-09-16 03:51:58 | INFO | train rank=0 | step=35730/50000 progress=71.5% loss=0.080196 lr=2.066427e-05 lr_backbone=2.066427e-05 elapsed=37:21:36 avg_step=3.76s recent_step=3.09s eta=14:55:15 +2026-09-16 03:52:30 | INFO | train rank=0 | step=35740/50000 progress=71.5% loss=0.077658 lr=2.063750e-05 lr_backbone=2.063750e-05 elapsed=37:22:08 avg_step=3.76s recent_step=3.20s eta=14:54:35 +2026-09-16 03:53:05 | INFO | train rank=0 | step=35750/50000 progress=71.5% loss=0.067428 lr=2.061074e-05 lr_backbone=2.061074e-05 elapsed=37:22:43 avg_step=3.76s recent_step=3.46s eta=14:53:57 +2026-09-16 03:53:44 | INFO | train rank=0 | step=35760/50000 progress=71.5% loss=0.091328 lr=2.058399e-05 lr_backbone=2.058399e-05 elapsed=37:23:22 avg_step=3.76s recent_step=3.94s eta=14:53:20 +2026-09-16 03:54:14 | INFO | train rank=0 | step=35770/50000 progress=71.5% loss=0.076603 lr=2.055726e-05 lr_backbone=2.055726e-05 elapsed=37:23:52 avg_step=3.76s recent_step=2.96s eta=14:52:39 +2026-09-16 03:54:50 | INFO | train rank=0 | step=35780/50000 progress=71.6% loss=0.075186 lr=2.053053e-05 lr_backbone=2.053053e-05 elapsed=37:24:28 avg_step=3.76s recent_step=3.61s eta=14:52:01 +2026-09-16 03:55:18 | INFO | train rank=0 | step=35790/50000 progress=71.6% loss=0.084917 lr=2.050383e-05 lr_backbone=2.050383e-05 elapsed=37:24:56 avg_step=3.76s recent_step=2.78s eta=14:51:19 +2026-09-16 03:56:02 | INFO | train rank=0 | step=35800/50000 progress=71.6% loss=0.086405 lr=2.047713e-05 lr_backbone=2.047713e-05 elapsed=37:25:40 avg_step=3.76s recent_step=4.46s eta=14:50:44 +2026-09-16 03:56:37 | INFO | train rank=0 | step=35810/50000 progress=71.6% loss=0.077612 lr=2.045045e-05 lr_backbone=2.045045e-05 elapsed=37:26:15 avg_step=3.76s recent_step=3.43s eta=14:50:05 +2026-09-16 03:57:01 | INFO | train rank=0 | step=35820/50000 progress=71.6% loss=0.109403 lr=2.042378e-05 lr_backbone=2.042378e-05 elapsed=37:26:39 avg_step=3.76s recent_step=2.44s eta=14:49:22 +2026-09-16 03:57:33 | INFO | train rank=0 | step=35830/50000 progress=71.7% loss=0.074033 lr=2.039712e-05 lr_backbone=2.039712e-05 elapsed=37:27:11 avg_step=3.76s recent_step=3.16s eta=14:48:42 +2026-09-16 03:58:13 | INFO | train rank=0 | step=35840/50000 progress=71.7% loss=0.116085 lr=2.037048e-05 lr_backbone=2.037048e-05 elapsed=37:27:51 avg_step=3.76s recent_step=4.01s eta=14:48:06 +2026-09-16 03:58:41 | INFO | train rank=0 | step=35850/50000 progress=71.7% loss=0.080320 lr=2.034385e-05 lr_backbone=2.034385e-05 elapsed=37:28:19 avg_step=3.76s recent_step=2.86s eta=14:47:24 +2026-09-16 03:59:17 | INFO | train rank=0 | step=35860/50000 progress=71.7% loss=0.122078 lr=2.031723e-05 lr_backbone=2.031723e-05 elapsed=37:28:55 avg_step=3.76s recent_step=3.58s eta=14:46:46 +2026-09-16 03:59:56 | INFO | train rank=0 | step=35870/50000 progress=71.7% loss=0.088895 lr=2.029062e-05 lr_backbone=2.029062e-05 elapsed=37:29:34 avg_step=3.76s recent_step=3.90s eta=14:46:09 +2026-09-16 04:00:59 | INFO | train rank=0 | step=35880/50000 progress=71.8% loss=0.094344 lr=2.026403e-05 lr_backbone=2.026403e-05 elapsed=37:30:37 avg_step=3.76s recent_step=6.31s eta=14:45:41 +2026-09-16 04:01:31 | INFO | train rank=0 | step=35890/50000 progress=71.8% loss=0.099748 lr=2.023745e-05 lr_backbone=2.023745e-05 elapsed=37:31:09 avg_step=3.76s recent_step=3.15s eta=14:45:01 +2026-09-16 04:02:01 | INFO | train rank=0 | step=35900/50000 progress=71.8% loss=0.079748 lr=2.021088e-05 lr_backbone=2.021088e-05 elapsed=37:31:39 avg_step=3.76s recent_step=3.06s eta=14:44:21 +2026-09-16 04:02:34 | INFO | train rank=0 | step=35910/50000 progress=71.8% loss=0.107657 lr=2.018433e-05 lr_backbone=2.018433e-05 elapsed=37:32:12 avg_step=3.76s recent_step=3.26s eta=14:43:41 +2026-09-16 04:03:27 | INFO | train rank=0 | step=35920/50000 progress=71.8% loss=0.087375 lr=2.015779e-05 lr_backbone=2.015779e-05 elapsed=37:33:05 avg_step=3.76s recent_step=5.34s eta=14:43:10 +2026-09-16 04:04:01 | INFO | train rank=0 | step=35930/50000 progress=71.9% loss=0.107348 lr=2.013126e-05 lr_backbone=2.013126e-05 elapsed=37:33:39 avg_step=3.76s recent_step=3.37s eta=14:42:31 +2026-09-16 04:04:38 | INFO | train rank=0 | step=35940/50000 progress=71.9% loss=0.090186 lr=2.010475e-05 lr_backbone=2.010475e-05 elapsed=37:34:16 avg_step=3.76s recent_step=3.66s eta=14:41:53 +2026-09-16 04:05:12 | INFO | train rank=0 | step=35950/50000 progress=71.9% loss=0.086660 lr=2.007825e-05 lr_backbone=2.007825e-05 elapsed=37:34:50 avg_step=3.76s recent_step=3.43s eta=14:41:14 +2026-09-16 04:06:08 | INFO | train rank=0 | step=35960/50000 progress=71.9% loss=0.100406 lr=2.005176e-05 lr_backbone=2.005176e-05 elapsed=37:35:46 avg_step=3.76s recent_step=5.60s eta=14:40:43 +2026-09-16 04:06:42 | INFO | train rank=0 | step=35970/50000 progress=71.9% loss=0.114396 lr=2.002529e-05 lr_backbone=2.002529e-05 elapsed=37:36:20 avg_step=3.76s recent_step=3.47s eta=14:40:04 +2026-09-16 04:07:12 | INFO | train rank=0 | step=35980/50000 progress=72.0% loss=0.084667 lr=1.999883e-05 lr_backbone=1.999883e-05 elapsed=37:36:50 avg_step=3.76s recent_step=2.96s eta=14:39:24 +2026-09-16 04:07:49 | INFO | train rank=0 | step=35990/50000 progress=72.0% loss=0.096083 lr=1.997238e-05 lr_backbone=1.997238e-05 elapsed=37:37:27 avg_step=3.76s recent_step=3.68s eta=14:38:46 +2026-09-16 04:08:44 | INFO | train rank=0 | step=36000/50000 progress=72.0% loss=0.091566 lr=1.994594e-05 lr_backbone=1.994594e-05 elapsed=37:38:22 avg_step=3.76s recent_step=5.51s eta=14:38:15 +2026-09-16 04:09:26 | INFO | train rank=0 | step=36010/50000 progress=72.0% loss=0.087014 lr=1.991952e-05 lr_backbone=1.991952e-05 elapsed=37:39:04 avg_step=3.76s recent_step=4.16s eta=14:37:39 +2026-09-16 04:10:02 | INFO | train rank=0 | step=36020/50000 progress=72.0% loss=0.106221 lr=1.989311e-05 lr_backbone=1.989311e-05 elapsed=37:39:41 avg_step=3.76s recent_step=3.70s eta=14:37:01 +2026-09-16 04:10:35 | INFO | train rank=0 | step=36030/50000 progress=72.1% loss=0.102626 lr=1.986672e-05 lr_backbone=1.986672e-05 elapsed=37:40:13 avg_step=3.76s recent_step=3.23s eta=14:36:21 +2026-09-16 04:11:32 | INFO | train rank=0 | step=36040/50000 progress=72.1% loss=0.074465 lr=1.984033e-05 lr_backbone=1.984033e-05 elapsed=37:41:10 avg_step=3.76s recent_step=5.72s eta=14:35:51 +2026-09-16 04:12:08 | INFO | train rank=0 | step=36050/50000 progress=72.1% loss=0.107473 lr=1.981396e-05 lr_backbone=1.981396e-05 elapsed=37:41:47 avg_step=3.76s recent_step=3.65s eta=14:35:13 +2026-09-16 04:12:41 | INFO | train rank=0 | step=36060/50000 progress=72.1% loss=0.077426 lr=1.978761e-05 lr_backbone=1.978761e-05 elapsed=37:42:19 avg_step=3.76s recent_step=3.29s eta=14:34:34 +2026-09-16 04:13:12 | INFO | train rank=0 | step=36070/50000 progress=72.1% loss=0.103585 lr=1.976126e-05 lr_backbone=1.976126e-05 elapsed=37:42:50 avg_step=3.76s recent_step=3.07s eta=14:33:53 +2026-09-16 04:13:53 | INFO | train rank=0 | step=36080/50000 progress=72.2% loss=0.100946 lr=1.973493e-05 lr_backbone=1.973493e-05 elapsed=37:43:31 avg_step=3.76s recent_step=4.07s eta=14:33:17 +2026-09-16 04:14:32 | INFO | train rank=0 | step=36090/50000 progress=72.2% loss=0.081379 lr=1.970862e-05 lr_backbone=1.970862e-05 elapsed=37:44:10 avg_step=3.76s recent_step=3.88s eta=14:32:40 +2026-09-16 04:15:03 | INFO | train rank=0 | step=36100/50000 progress=72.2% loss=0.112698 lr=1.968232e-05 lr_backbone=1.968232e-05 elapsed=37:44:41 avg_step=3.76s recent_step=3.10s eta=14:31:59 +2026-09-16 04:15:32 | INFO | train rank=0 | step=36110/50000 progress=72.2% loss=0.078154 lr=1.965603e-05 lr_backbone=1.965603e-05 elapsed=37:45:10 avg_step=3.76s recent_step=2.97s eta=14:31:19 +2026-09-16 04:16:20 | INFO | train rank=0 | step=36120/50000 progress=72.2% loss=0.079872 lr=1.962975e-05 lr_backbone=1.962975e-05 elapsed=37:45:58 avg_step=3.76s recent_step=4.74s eta=14:30:45 +2026-09-16 04:16:57 | INFO | train rank=0 | step=36130/50000 progress=72.3% loss=0.089822 lr=1.960349e-05 lr_backbone=1.960349e-05 elapsed=37:46:35 avg_step=3.76s recent_step=3.71s eta=14:30:07 +2026-09-16 04:17:31 | INFO | train rank=0 | step=36140/50000 progress=72.3% loss=0.083087 lr=1.957723e-05 lr_backbone=1.957723e-05 elapsed=37:47:09 avg_step=3.76s recent_step=3.48s eta=14:29:28 +2026-09-16 04:18:11 | INFO | train rank=0 | step=36150/50000 progress=72.3% loss=0.080865 lr=1.955100e-05 lr_backbone=1.955100e-05 elapsed=37:47:49 avg_step=3.76s recent_step=4.00s eta=14:28:51 +2026-09-16 04:19:00 | INFO | train rank=0 | step=36160/50000 progress=72.3% loss=0.095389 lr=1.952477e-05 lr_backbone=1.952477e-05 elapsed=37:48:38 avg_step=3.76s recent_step=4.80s eta=14:28:18 +2026-09-16 04:19:37 | INFO | train rank=0 | step=36170/50000 progress=72.3% loss=0.081732 lr=1.949856e-05 lr_backbone=1.949856e-05 elapsed=37:49:15 avg_step=3.76s recent_step=3.70s eta=14:27:40 +2026-09-16 04:20:07 | INFO | train rank=0 | step=36180/50000 progress=72.4% loss=0.083386 lr=1.947237e-05 lr_backbone=1.947237e-05 elapsed=37:49:45 avg_step=3.76s recent_step=3.08s eta=14:27:00 +2026-09-16 04:20:41 | INFO | train rank=0 | step=36190/50000 progress=72.4% loss=0.068903 lr=1.944618e-05 lr_backbone=1.944618e-05 elapsed=37:50:19 avg_step=3.76s recent_step=3.32s eta=14:26:20 +2026-09-16 04:21:19 | INFO | train rank=0 | step=36200/50000 progress=72.4% loss=0.099277 lr=1.942001e-05 lr_backbone=1.942001e-05 elapsed=37:50:57 avg_step=3.76s recent_step=3.80s eta=14:25:43 +2026-09-16 04:21:50 | INFO | train rank=0 | step=36210/50000 progress=72.4% loss=0.092646 lr=1.939386e-05 lr_backbone=1.939386e-05 elapsed=37:51:28 avg_step=3.76s recent_step=3.16s eta=14:25:03 +2026-09-16 04:22:25 | INFO | train rank=0 | step=36220/50000 progress=72.4% loss=0.076630 lr=1.936771e-05 lr_backbone=1.936771e-05 elapsed=37:52:03 avg_step=3.76s recent_step=3.45s eta=14:24:24 +2026-09-16 04:23:04 | INFO | train rank=0 | step=36230/50000 progress=72.5% loss=0.072354 lr=1.934158e-05 lr_backbone=1.934158e-05 elapsed=37:52:42 avg_step=3.76s recent_step=3.96s eta=14:23:47 +2026-09-16 04:23:44 | INFO | train rank=0 | step=36240/50000 progress=72.5% loss=0.074672 lr=1.931547e-05 lr_backbone=1.931547e-05 elapsed=37:53:22 avg_step=3.76s recent_step=4.00s eta=14:23:10 +2026-09-16 04:24:14 | INFO | train rank=0 | step=36250/50000 progress=72.5% loss=0.093653 lr=1.928936e-05 lr_backbone=1.928936e-05 elapsed=37:53:52 avg_step=3.76s recent_step=3.02s eta=14:22:30 +2026-09-16 04:24:50 | INFO | train rank=0 | step=36260/50000 progress=72.5% loss=0.084167 lr=1.926327e-05 lr_backbone=1.926327e-05 elapsed=37:54:28 avg_step=3.76s recent_step=3.53s eta=14:21:51 +2026-09-16 04:25:40 | INFO | train rank=0 | step=36270/50000 progress=72.5% loss=0.114432 lr=1.923720e-05 lr_backbone=1.923720e-05 elapsed=37:55:18 avg_step=3.76s recent_step=4.99s eta=14:21:18 +2026-09-16 04:26:14 | INFO | train rank=0 | step=36280/50000 progress=72.6% loss=0.064616 lr=1.921114e-05 lr_backbone=1.921114e-05 elapsed=37:55:52 avg_step=3.76s recent_step=3.44s eta=14:20:40 +2026-09-16 04:26:46 | INFO | train rank=0 | step=36290/50000 progress=72.6% loss=0.072965 lr=1.918509e-05 lr_backbone=1.918509e-05 elapsed=37:56:24 avg_step=3.76s recent_step=3.22s eta=14:20:00 +2026-09-16 04:27:19 | INFO | train rank=0 | step=36300/50000 progress=72.6% loss=0.082559 lr=1.915905e-05 lr_backbone=1.915905e-05 elapsed=37:56:57 avg_step=3.76s recent_step=3.27s eta=14:19:20 +2026-09-16 04:27:56 | INFO | train rank=0 | step=36310/50000 progress=72.6% loss=0.104082 lr=1.913303e-05 lr_backbone=1.913303e-05 elapsed=37:57:34 avg_step=3.76s recent_step=3.70s eta=14:18:43 +2026-09-16 04:28:47 | INFO | train rank=0 | step=36320/50000 progress=72.6% loss=0.085017 lr=1.910702e-05 lr_backbone=1.910702e-05 elapsed=37:58:25 avg_step=3.76s recent_step=5.12s eta=14:18:10 +2026-09-16 04:29:17 | INFO | train rank=0 | step=36330/50000 progress=72.7% loss=0.090637 lr=1.908102e-05 lr_backbone=1.908102e-05 elapsed=37:58:55 avg_step=3.76s recent_step=2.99s eta=14:17:29 +2026-09-16 04:29:54 | INFO | train rank=0 | step=36340/50000 progress=72.7% loss=0.094779 lr=1.905504e-05 lr_backbone=1.905504e-05 elapsed=37:59:32 avg_step=3.76s recent_step=3.66s eta=14:16:51 +2026-09-16 04:30:26 | INFO | train rank=0 | step=36350/50000 progress=72.7% loss=0.075786 lr=1.902907e-05 lr_backbone=1.902907e-05 elapsed=38:00:04 avg_step=3.76s recent_step=3.22s eta=14:16:12 +2026-09-16 04:31:05 | INFO | train rank=0 | step=36360/50000 progress=72.7% loss=0.097046 lr=1.900312e-05 lr_backbone=1.900312e-05 elapsed=38:00:43 avg_step=3.76s recent_step=3.88s eta=14:15:35 +2026-09-16 04:31:40 | INFO | train rank=0 | step=36370/50000 progress=72.7% loss=0.094442 lr=1.897718e-05 lr_backbone=1.897718e-05 elapsed=38:01:18 avg_step=3.76s recent_step=3.50s eta=14:14:56 +2026-09-16 04:32:15 | INFO | train rank=0 | step=36380/50000 progress=72.8% loss=0.090136 lr=1.895125e-05 lr_backbone=1.895125e-05 elapsed=38:01:53 avg_step=3.76s recent_step=3.47s eta=14:14:17 +2026-09-16 04:33:02 | INFO | train rank=0 | step=36390/50000 progress=72.8% loss=0.109207 lr=1.892534e-05 lr_backbone=1.892534e-05 elapsed=38:02:40 avg_step=3.76s recent_step=4.75s eta=14:13:43 +2026-09-16 04:33:46 | INFO | train rank=0 | step=36400/50000 progress=72.8% loss=0.093281 lr=1.889944e-05 lr_backbone=1.889944e-05 elapsed=38:03:24 avg_step=3.76s recent_step=4.41s eta=14:13:08 +2026-09-16 04:34:28 | INFO | train rank=0 | step=36410/50000 progress=72.8% loss=0.091595 lr=1.887355e-05 lr_backbone=1.887355e-05 elapsed=38:04:06 avg_step=3.76s recent_step=4.22s eta=14:12:32 +2026-09-16 04:34:55 | INFO | train rank=0 | step=36420/50000 progress=72.8% loss=0.106044 lr=1.884768e-05 lr_backbone=1.884768e-05 elapsed=38:04:33 avg_step=3.76s recent_step=2.65s eta=14:11:50 +2026-09-16 04:35:22 | INFO | train rank=0 | step=36430/50000 progress=72.9% loss=0.072710 lr=1.882182e-05 lr_backbone=1.882182e-05 elapsed=38:05:00 avg_step=3.76s recent_step=2.70s eta=14:11:09 +2026-09-16 04:36:07 | INFO | train rank=0 | step=36440/50000 progress=72.9% loss=0.083009 lr=1.879597e-05 lr_backbone=1.879597e-05 elapsed=38:05:45 avg_step=3.76s recent_step=4.51s eta=14:10:34 +2026-09-16 04:36:43 | INFO | train rank=0 | step=36450/50000 progress=72.9% loss=0.074088 lr=1.877014e-05 lr_backbone=1.877014e-05 elapsed=38:06:21 avg_step=3.76s recent_step=3.59s eta=14:09:56 +2026-09-16 04:37:22 | INFO | train rank=0 | step=36460/50000 progress=72.9% loss=0.086508 lr=1.874432e-05 lr_backbone=1.874432e-05 elapsed=38:07:00 avg_step=3.76s recent_step=3.95s eta=14:09:19 +2026-09-16 04:38:18 | INFO | train rank=0 | step=36470/50000 progress=72.9% loss=0.075192 lr=1.871851e-05 lr_backbone=1.871851e-05 elapsed=38:07:56 avg_step=3.76s recent_step=5.61s eta=14:08:48 +2026-09-16 04:38:55 | INFO | train rank=0 | step=36480/50000 progress=73.0% loss=0.111930 lr=1.869272e-05 lr_backbone=1.869272e-05 elapsed=38:08:33 avg_step=3.76s recent_step=3.67s eta=14:08:10 +2026-09-16 04:39:26 | INFO | train rank=0 | step=36490/50000 progress=73.0% loss=0.098347 lr=1.866695e-05 lr_backbone=1.866695e-05 elapsed=38:09:04 avg_step=3.76s recent_step=3.06s eta=14:07:30 +2026-09-16 04:40:04 | INFO | train rank=0 | step=36500/50000 progress=73.0% loss=0.085476 lr=1.864118e-05 lr_backbone=1.864118e-05 elapsed=38:09:42 avg_step=3.76s recent_step=3.81s eta=14:06:52 +2026-09-16 04:40:34 | INFO | train rank=0 | step=36510/50000 progress=73.0% loss=0.112847 lr=1.861543e-05 lr_backbone=1.861543e-05 elapsed=38:10:12 avg_step=3.76s recent_step=3.04s eta=14:06:12 +2026-09-16 04:41:32 | INFO | train rank=0 | step=36520/50000 progress=73.0% loss=0.071251 lr=1.858970e-05 lr_backbone=1.858970e-05 elapsed=38:11:10 avg_step=3.76s recent_step=5.80s eta=14:05:42 +2026-09-16 04:42:08 | INFO | train rank=0 | step=36530/50000 progress=73.1% loss=0.103537 lr=1.856397e-05 lr_backbone=1.856397e-05 elapsed=38:11:46 avg_step=3.76s recent_step=3.56s eta=14:05:03 +2026-09-16 04:42:39 | INFO | train rank=0 | step=36540/50000 progress=73.1% loss=0.090060 lr=1.853826e-05 lr_backbone=1.853826e-05 elapsed=38:12:17 avg_step=3.76s recent_step=3.11s eta=14:04:23 +2026-09-16 04:43:09 | INFO | train rank=0 | step=36550/50000 progress=73.1% loss=0.086140 lr=1.851257e-05 lr_backbone=1.851257e-05 elapsed=38:12:47 avg_step=3.76s recent_step=2.97s eta=14:03:43 +2026-09-16 04:44:02 | INFO | train rank=0 | step=36560/50000 progress=73.1% loss=0.083289 lr=1.848689e-05 lr_backbone=1.848689e-05 elapsed=38:13:40 avg_step=3.76s recent_step=5.38s eta=14:03:11 +2026-09-16 04:44:34 | INFO | train rank=0 | step=36570/50000 progress=73.1% loss=0.075217 lr=1.846122e-05 lr_backbone=1.846122e-05 elapsed=38:14:12 avg_step=3.76s recent_step=3.16s eta=14:02:31 +2026-09-16 04:45:00 | INFO | train rank=0 | step=36580/50000 progress=73.2% loss=0.095393 lr=1.843557e-05 lr_backbone=1.843557e-05 elapsed=38:14:38 avg_step=3.76s recent_step=2.64s eta=14:01:49 +2026-09-16 04:45:34 | INFO | train rank=0 | step=36590/50000 progress=73.2% loss=0.100724 lr=1.840993e-05 lr_backbone=1.840993e-05 elapsed=38:15:12 avg_step=3.76s recent_step=3.37s eta=14:01:10 +2026-09-16 04:46:26 | INFO | train rank=0 | step=36600/50000 progress=73.2% loss=0.092033 lr=1.838430e-05 lr_backbone=1.838430e-05 elapsed=38:16:04 avg_step=3.76s recent_step=5.20s eta=14:00:38 +2026-09-16 04:47:04 | INFO | train rank=0 | step=36610/50000 progress=73.2% loss=0.080792 lr=1.835869e-05 lr_backbone=1.835869e-05 elapsed=38:16:42 avg_step=3.76s recent_step=3.77s eta=14:00:00 +2026-09-16 04:47:31 | INFO | train rank=0 | step=36620/50000 progress=73.2% loss=0.111006 lr=1.833309e-05 lr_backbone=1.833309e-05 elapsed=38:17:09 avg_step=3.76s recent_step=2.75s eta=13:59:19 +2026-09-16 04:48:03 | INFO | train rank=0 | step=36630/50000 progress=73.3% loss=0.095718 lr=1.830750e-05 lr_backbone=1.830750e-05 elapsed=38:17:41 avg_step=3.76s recent_step=3.14s eta=13:58:39 +2026-09-16 04:48:45 | INFO | train rank=0 | step=36640/50000 progress=73.3% loss=0.094541 lr=1.828193e-05 lr_backbone=1.828193e-05 elapsed=38:18:23 avg_step=3.76s recent_step=4.22s eta=13:58:03 +2026-09-16 04:49:21 | INFO | train rank=0 | step=36650/50000 progress=73.3% loss=0.075201 lr=1.825638e-05 lr_backbone=1.825638e-05 elapsed=38:18:59 avg_step=3.76s recent_step=3.60s eta=13:57:25 +2026-09-16 04:50:00 | INFO | train rank=0 | step=36660/50000 progress=73.3% loss=0.100463 lr=1.823083e-05 lr_backbone=1.823083e-05 elapsed=38:19:38 avg_step=3.76s recent_step=3.92s eta=13:56:48 +2026-09-16 04:50:42 | INFO | train rank=0 | step=36670/50000 progress=73.3% loss=0.078352 lr=1.820530e-05 lr_backbone=1.820530e-05 elapsed=38:20:20 avg_step=3.76s recent_step=4.17s eta=13:56:12 +2026-09-16 04:51:26 | INFO | train rank=0 | step=36680/50000 progress=73.4% loss=0.081263 lr=1.817979e-05 lr_backbone=1.817979e-05 elapsed=38:21:04 avg_step=3.76s recent_step=4.45s eta=13:55:36 +2026-09-16 04:52:00 | INFO | train rank=0 | step=36690/50000 progress=73.4% loss=0.093547 lr=1.815429e-05 lr_backbone=1.815429e-05 elapsed=38:21:38 avg_step=3.76s recent_step=3.41s eta=13:54:57 +2026-09-16 04:52:36 | INFO | train rank=0 | step=36700/50000 progress=73.4% loss=0.086233 lr=1.812880e-05 lr_backbone=1.812880e-05 elapsed=38:22:14 avg_step=3.76s recent_step=3.52s eta=13:54:19 +2026-09-16 04:53:05 | INFO | train rank=0 | step=36710/50000 progress=73.4% loss=0.088716 lr=1.810333e-05 lr_backbone=1.810333e-05 elapsed=38:22:43 avg_step=3.76s recent_step=2.93s eta=13:53:38 +2026-09-16 04:53:55 | INFO | train rank=0 | step=36720/50000 progress=73.4% loss=0.107487 lr=1.807787e-05 lr_backbone=1.807787e-05 elapsed=38:23:33 avg_step=3.76s recent_step=4.99s eta=13:53:05 +2026-09-16 04:54:36 | INFO | train rank=0 | step=36730/50000 progress=73.5% loss=0.082076 lr=1.805242e-05 lr_backbone=1.805242e-05 elapsed=38:24:14 avg_step=3.76s recent_step=4.16s eta=13:52:29 +2026-09-16 04:55:09 | INFO | train rank=0 | step=36740/50000 progress=73.5% loss=0.096420 lr=1.802699e-05 lr_backbone=1.802699e-05 elapsed=38:24:47 avg_step=3.76s recent_step=3.22s eta=13:51:49 +2026-09-16 04:55:43 | INFO | train rank=0 | step=36750/50000 progress=73.5% loss=0.077539 lr=1.800157e-05 lr_backbone=1.800157e-05 elapsed=38:25:21 avg_step=3.76s recent_step=3.47s eta=13:51:11 +2026-09-16 04:56:28 | INFO | train rank=0 | step=36760/50000 progress=73.5% loss=0.086049 lr=1.797617e-05 lr_backbone=1.797617e-05 elapsed=38:26:06 avg_step=3.76s recent_step=4.46s eta=13:50:35 +2026-09-16 04:56:56 | INFO | train rank=0 | step=36770/50000 progress=73.5% loss=0.072375 lr=1.795078e-05 lr_backbone=1.795078e-05 elapsed=38:26:34 avg_step=3.76s recent_step=2.79s eta=13:49:54 +2026-09-16 04:57:35 | INFO | train rank=0 | step=36780/50000 progress=73.6% loss=0.085010 lr=1.792540e-05 lr_backbone=1.792540e-05 elapsed=38:27:13 avg_step=3.76s recent_step=3.96s eta=13:49:17 +2026-09-16 04:58:02 | INFO | train rank=0 | step=36790/50000 progress=73.6% loss=0.101095 lr=1.790004e-05 lr_backbone=1.790004e-05 elapsed=38:27:40 avg_step=3.76s recent_step=2.67s eta=13:48:36 +2026-09-16 04:58:52 | INFO | train rank=0 | step=36800/50000 progress=73.6% loss=0.095307 lr=1.787470e-05 lr_backbone=1.787470e-05 elapsed=38:28:30 avg_step=3.76s recent_step=4.95s eta=13:48:03 +2026-09-16 04:59:23 | INFO | train rank=0 | step=36810/50000 progress=73.6% loss=0.099832 lr=1.784936e-05 lr_backbone=1.784936e-05 elapsed=38:29:01 avg_step=3.76s recent_step=3.16s eta=13:47:23 +2026-09-16 04:59:56 | INFO | train rank=0 | step=36820/50000 progress=73.6% loss=0.109164 lr=1.782404e-05 lr_backbone=1.782404e-05 elapsed=38:29:34 avg_step=3.76s recent_step=3.23s eta=13:46:43 +2026-09-16 05:00:29 | INFO | train rank=0 | step=36830/50000 progress=73.7% loss=0.080697 lr=1.779874e-05 lr_backbone=1.779874e-05 elapsed=38:30:07 avg_step=3.76s recent_step=3.33s eta=13:46:04 +2026-09-16 05:01:21 | INFO | train rank=0 | step=36840/50000 progress=73.7% loss=0.084217 lr=1.777345e-05 lr_backbone=1.777345e-05 elapsed=38:30:59 avg_step=3.76s recent_step=5.24s eta=13:45:32 +2026-09-16 05:01:57 | INFO | train rank=0 | step=36850/50000 progress=73.7% loss=0.124076 lr=1.774817e-05 lr_backbone=1.774817e-05 elapsed=38:31:35 avg_step=3.76s recent_step=3.59s eta=13:44:53 +2026-09-16 05:02:31 | INFO | train rank=0 | step=36860/50000 progress=73.7% loss=0.090136 lr=1.772291e-05 lr_backbone=1.772291e-05 elapsed=38:32:09 avg_step=3.76s recent_step=3.35s eta=13:44:14 +2026-09-16 05:03:01 | INFO | train rank=0 | step=36870/50000 progress=73.7% loss=0.096861 lr=1.769766e-05 lr_backbone=1.769766e-05 elapsed=38:32:39 avg_step=3.76s recent_step=3.09s eta=13:43:34 +2026-09-16 05:03:42 | INFO | train rank=0 | step=36880/50000 progress=73.8% loss=0.083843 lr=1.767242e-05 lr_backbone=1.767242e-05 elapsed=38:33:20 avg_step=3.76s recent_step=4.08s eta=13:42:58 +2026-09-16 05:04:10 | INFO | train rank=0 | step=36890/50000 progress=73.8% loss=0.105245 lr=1.764720e-05 lr_backbone=1.764720e-05 elapsed=38:33:48 avg_step=3.76s recent_step=2.76s eta=13:42:16 +2026-09-16 05:04:37 | INFO | train rank=0 | step=36900/50000 progress=73.8% loss=0.094433 lr=1.762200e-05 lr_backbone=1.762200e-05 elapsed=38:34:15 avg_step=3.76s recent_step=2.70s eta=13:41:35 +2026-09-16 05:05:17 | INFO | train rank=0 | step=36910/50000 progress=73.8% loss=0.081609 lr=1.759680e-05 lr_backbone=1.759680e-05 elapsed=38:34:55 avg_step=3.76s recent_step=4.05s eta=13:40:58 +2026-09-16 05:05:58 | INFO | train rank=0 | step=36920/50000 progress=73.8% loss=0.086587 lr=1.757163e-05 lr_backbone=1.757163e-05 elapsed=38:35:36 avg_step=3.76s recent_step=4.04s eta=13:40:22 +2026-09-16 05:06:35 | INFO | train rank=0 | step=36930/50000 progress=73.9% loss=0.088523 lr=1.754646e-05 lr_backbone=1.754646e-05 elapsed=38:36:13 avg_step=3.76s recent_step=3.71s eta=13:39:44 +2026-09-16 05:07:11 | INFO | train rank=0 | step=36940/50000 progress=73.9% loss=0.122826 lr=1.752131e-05 lr_backbone=1.752131e-05 elapsed=38:36:49 avg_step=3.76s recent_step=3.58s eta=13:39:06 +2026-09-16 05:07:46 | INFO | train rank=0 | step=36950/50000 progress=73.9% loss=0.090757 lr=1.749618e-05 lr_backbone=1.749618e-05 elapsed=38:37:24 avg_step=3.76s recent_step=3.50s eta=13:38:27 +2026-09-16 05:08:29 | INFO | train rank=0 | step=36960/50000 progress=73.9% loss=0.091843 lr=1.747105e-05 lr_backbone=1.747105e-05 elapsed=38:38:07 avg_step=3.76s recent_step=4.32s eta=13:37:51 +2026-09-16 05:09:01 | INFO | train rank=0 | step=36970/50000 progress=73.9% loss=0.075880 lr=1.744595e-05 lr_backbone=1.744595e-05 elapsed=38:38:39 avg_step=3.76s recent_step=3.24s eta=13:37:12 +2026-09-16 05:09:37 | INFO | train rank=0 | step=36980/50000 progress=74.0% loss=0.086612 lr=1.742085e-05 lr_backbone=1.742085e-05 elapsed=38:39:15 avg_step=3.76s recent_step=3.56s eta=13:36:34 +2026-09-16 05:10:07 | INFO | train rank=0 | step=36990/50000 progress=74.0% loss=0.107534 lr=1.739578e-05 lr_backbone=1.739578e-05 elapsed=38:39:45 avg_step=3.76s recent_step=3.00s eta=13:35:53 +2026-09-16 05:11:01 | INFO | train rank=0 | step=37000/50000 progress=74.0% loss=0.081166 lr=1.737071e-05 lr_backbone=1.737071e-05 elapsed=38:40:39 avg_step=3.76s recent_step=5.37s eta=13:35:21 +2026-09-16 05:11:33 | INFO | train rank=0 | step=37010/50000 progress=74.0% loss=0.082484 lr=1.734566e-05 lr_backbone=1.734566e-05 elapsed=38:41:11 avg_step=3.76s recent_step=3.22s eta=13:34:42 +2026-09-16 05:12:11 | INFO | train rank=0 | step=37020/50000 progress=74.0% loss=0.072621 lr=1.732063e-05 lr_backbone=1.732063e-05 elapsed=38:41:49 avg_step=3.76s recent_step=3.85s eta=13:34:04 +2026-09-16 05:12:45 | INFO | train rank=0 | step=37030/50000 progress=74.1% loss=0.079633 lr=1.729560e-05 lr_backbone=1.729560e-05 elapsed=38:42:23 avg_step=3.76s recent_step=3.35s eta=13:33:25 +2026-09-16 05:13:41 | INFO | train rank=0 | step=37040/50000 progress=74.1% loss=0.074700 lr=1.727060e-05 lr_backbone=1.727060e-05 elapsed=38:43:19 avg_step=3.76s recent_step=5.63s eta=13:32:54 +2026-09-16 05:14:14 | INFO | train rank=0 | step=37050/50000 progress=74.1% loss=0.087112 lr=1.724560e-05 lr_backbone=1.724560e-05 elapsed=38:43:52 avg_step=3.76s recent_step=3.26s eta=13:32:15 +2026-09-16 05:14:47 | INFO | train rank=0 | step=37060/50000 progress=74.1% loss=0.081456 lr=1.722063e-05 lr_backbone=1.722063e-05 elapsed=38:44:25 avg_step=3.76s recent_step=3.29s eta=13:31:36 +2026-09-16 05:15:25 | INFO | train rank=0 | step=37070/50000 progress=74.1% loss=0.102335 lr=1.719566e-05 lr_backbone=1.719566e-05 elapsed=38:45:03 avg_step=3.76s recent_step=3.88s eta=13:30:58 +2026-09-16 05:16:16 | INFO | train rank=0 | step=37080/50000 progress=74.2% loss=0.110407 lr=1.717071e-05 lr_backbone=1.717071e-05 elapsed=38:45:54 avg_step=3.76s recent_step=5.07s eta=13:30:25 +2026-09-16 05:16:46 | INFO | train rank=0 | step=37090/50000 progress=74.2% loss=0.080646 lr=1.714578e-05 lr_backbone=1.714578e-05 elapsed=38:46:24 avg_step=3.76s recent_step=3.01s eta=13:29:45 +2026-09-16 05:17:17 | INFO | train rank=0 | step=37100/50000 progress=74.2% loss=0.091490 lr=1.712086e-05 lr_backbone=1.712086e-05 elapsed=38:46:55 avg_step=3.76s recent_step=3.09s eta=13:29:05 +2026-09-16 05:17:52 | INFO | train rank=0 | step=37110/50000 progress=74.2% loss=0.099505 lr=1.709595e-05 lr_backbone=1.709595e-05 elapsed=38:47:30 avg_step=3.76s recent_step=3.51s eta=13:28:27 +2026-09-16 05:18:40 | INFO | train rank=0 | step=37120/50000 progress=74.2% loss=0.104171 lr=1.707106e-05 lr_backbone=1.707106e-05 elapsed=38:48:18 avg_step=3.76s recent_step=4.74s eta=13:27:52 +2026-09-16 05:19:15 | INFO | train rank=0 | step=37130/50000 progress=74.3% loss=0.080776 lr=1.704618e-05 lr_backbone=1.704618e-05 elapsed=38:48:53 avg_step=3.76s recent_step=3.52s eta=13:27:14 +2026-09-16 05:19:48 | INFO | train rank=0 | step=37140/50000 progress=74.3% loss=0.074265 lr=1.702132e-05 lr_backbone=1.702132e-05 elapsed=38:49:26 avg_step=3.76s recent_step=3.31s eta=13:26:35 +2026-09-16 05:20:20 | INFO | train rank=0 | step=37150/50000 progress=74.3% loss=0.084703 lr=1.699647e-05 lr_backbone=1.699647e-05 elapsed=38:49:58 avg_step=3.76s recent_step=3.22s eta=13:25:55 +2026-09-16 05:21:00 | INFO | train rank=0 | step=37160/50000 progress=74.3% loss=0.099036 lr=1.697163e-05 lr_backbone=1.697163e-05 elapsed=38:50:38 avg_step=3.76s recent_step=3.97s eta=13:25:18 +2026-09-16 05:21:30 | INFO | train rank=0 | step=37170/50000 progress=74.3% loss=0.101324 lr=1.694681e-05 lr_backbone=1.694681e-05 elapsed=38:51:08 avg_step=3.76s recent_step=3.04s eta=13:24:38 +2026-09-16 05:22:09 | INFO | train rank=0 | step=37180/50000 progress=74.4% loss=0.088471 lr=1.692201e-05 lr_backbone=1.692201e-05 elapsed=38:51:47 avg_step=3.76s recent_step=3.91s eta=13:24:01 +2026-09-16 05:22:48 | INFO | train rank=0 | step=37190/50000 progress=74.4% loss=0.094671 lr=1.689721e-05 lr_backbone=1.689721e-05 elapsed=38:52:26 avg_step=3.76s recent_step=3.84s eta=13:23:24 +2026-09-16 05:23:23 | INFO | train rank=0 | step=37200/50000 progress=74.4% loss=0.081184 lr=1.687244e-05 lr_backbone=1.687244e-05 elapsed=38:53:01 avg_step=3.76s recent_step=3.48s eta=13:22:45 +2026-09-16 05:23:55 | INFO | train rank=0 | step=37210/50000 progress=74.4% loss=0.083674 lr=1.684768e-05 lr_backbone=1.684768e-05 elapsed=38:53:33 avg_step=3.76s recent_step=3.22s eta=13:22:06 +2026-09-16 05:24:32 | INFO | train rank=0 | step=37220/50000 progress=74.4% loss=0.087144 lr=1.682293e-05 lr_backbone=1.682293e-05 elapsed=38:54:10 avg_step=3.76s recent_step=3.75s eta=13:21:28 +2026-09-16 05:25:14 | INFO | train rank=0 | step=37230/50000 progress=74.5% loss=0.100989 lr=1.679819e-05 lr_backbone=1.679819e-05 elapsed=38:54:52 avg_step=3.76s recent_step=4.21s eta=13:20:52 +2026-09-16 05:25:41 | INFO | train rank=0 | step=37240/50000 progress=74.5% loss=0.089328 lr=1.677348e-05 lr_backbone=1.677348e-05 elapsed=38:55:19 avg_step=3.76s recent_step=2.66s eta=13:20:10 +2026-09-16 05:26:15 | INFO | train rank=0 | step=37250/50000 progress=74.5% loss=0.090082 lr=1.674877e-05 lr_backbone=1.674877e-05 elapsed=38:55:53 avg_step=3.76s recent_step=3.36s eta=13:19:31 +2026-09-16 05:26:48 | INFO | train rank=0 | step=37260/50000 progress=74.5% loss=0.128305 lr=1.672408e-05 lr_backbone=1.672408e-05 elapsed=38:56:26 avg_step=3.76s recent_step=3.32s eta=13:18:52 +2026-09-16 05:27:22 | INFO | train rank=0 | step=37270/50000 progress=74.5% loss=0.069473 lr=1.669941e-05 lr_backbone=1.669941e-05 elapsed=38:57:00 avg_step=3.76s recent_step=3.47s eta=13:18:14 +2026-09-16 05:28:00 | INFO | train rank=0 | step=37280/50000 progress=74.6% loss=0.098121 lr=1.667475e-05 lr_backbone=1.667475e-05 elapsed=38:57:38 avg_step=3.76s recent_step=3.80s eta=13:17:36 +2026-09-16 05:28:59 | INFO | train rank=0 | step=37290/50000 progress=74.6% loss=0.070773 lr=1.665010e-05 lr_backbone=1.665010e-05 elapsed=38:58:37 avg_step=3.76s recent_step=5.83s eta=13:17:06 +2026-09-16 05:29:35 | INFO | train rank=0 | step=37300/50000 progress=74.6% loss=0.075782 lr=1.662547e-05 lr_backbone=1.662547e-05 elapsed=38:59:13 avg_step=3.76s recent_step=3.63s eta=13:16:27 +2026-09-16 05:30:10 | INFO | train rank=0 | step=37310/50000 progress=74.6% loss=0.094329 lr=1.660085e-05 lr_backbone=1.660085e-05 elapsed=38:59:48 avg_step=3.76s recent_step=3.46s eta=13:15:49 +2026-09-16 05:30:39 | INFO | train rank=0 | step=37320/50000 progress=74.6% loss=0.096648 lr=1.657625e-05 lr_backbone=1.657625e-05 elapsed=39:00:17 avg_step=3.76s recent_step=2.90s eta=13:15:08 +2026-09-16 05:31:25 | INFO | train rank=0 | step=37330/50000 progress=74.7% loss=0.075061 lr=1.655166e-05 lr_backbone=1.655166e-05 elapsed=39:01:03 avg_step=3.76s recent_step=4.59s eta=13:14:33 +2026-09-16 05:31:54 | INFO | train rank=0 | step=37340/50000 progress=74.7% loss=0.063266 lr=1.652709e-05 lr_backbone=1.652709e-05 elapsed=39:01:32 avg_step=3.76s recent_step=2.95s eta=13:13:53 +2026-09-16 05:32:33 | INFO | train rank=0 | step=37350/50000 progress=74.7% loss=0.088000 lr=1.650253e-05 lr_backbone=1.650253e-05 elapsed=39:02:11 avg_step=3.76s recent_step=3.90s eta=13:13:16 +2026-09-16 05:33:06 | INFO | train rank=0 | step=37360/50000 progress=74.7% loss=0.075083 lr=1.647799e-05 lr_backbone=1.647799e-05 elapsed=39:02:44 avg_step=3.76s recent_step=3.27s eta=13:12:37 +2026-09-16 05:33:54 | INFO | train rank=0 | step=37370/50000 progress=74.7% loss=0.086858 lr=1.645346e-05 lr_backbone=1.645346e-05 elapsed=39:03:32 avg_step=3.76s recent_step=4.80s eta=13:12:02 +2026-09-16 05:34:25 | INFO | train rank=0 | step=37380/50000 progress=74.8% loss=0.079466 lr=1.642894e-05 lr_backbone=1.642894e-05 elapsed=39:04:03 avg_step=3.76s recent_step=3.10s eta=13:11:23 +2026-09-16 05:34:56 | INFO | train rank=0 | step=37390/50000 progress=74.8% loss=0.106490 lr=1.640444e-05 lr_backbone=1.640444e-05 elapsed=39:04:34 avg_step=3.76s recent_step=3.12s eta=13:10:43 +2026-09-16 05:35:24 | INFO | train rank=0 | step=37400/50000 progress=74.8% loss=0.073404 lr=1.637996e-05 lr_backbone=1.637996e-05 elapsed=39:05:02 avg_step=3.76s recent_step=2.77s eta=13:10:02 +2026-09-16 05:36:06 | INFO | train rank=0 | step=37410/50000 progress=74.8% loss=0.100520 lr=1.635549e-05 lr_backbone=1.635549e-05 elapsed=39:05:44 avg_step=3.76s recent_step=4.23s eta=13:09:26 +2026-09-16 05:36:45 | INFO | train rank=0 | step=37420/50000 progress=74.8% loss=0.088247 lr=1.633103e-05 lr_backbone=1.633103e-05 elapsed=39:06:23 avg_step=3.76s recent_step=3.94s eta=13:08:49 +2026-09-16 05:37:17 | INFO | train rank=0 | step=37430/50000 progress=74.9% loss=0.102545 lr=1.630659e-05 lr_backbone=1.630659e-05 elapsed=39:06:55 avg_step=3.76s recent_step=3.14s eta=13:08:09 +2026-09-16 05:37:45 | INFO | train rank=0 | step=37440/50000 progress=74.9% loss=0.081794 lr=1.628217e-05 lr_backbone=1.628217e-05 elapsed=39:07:23 avg_step=3.76s recent_step=2.85s eta=13:07:28 +2026-09-16 05:38:33 | INFO | train rank=0 | step=37450/50000 progress=74.9% loss=0.117356 lr=1.625776e-05 lr_backbone=1.625776e-05 elapsed=39:08:11 avg_step=3.76s recent_step=4.78s eta=13:06:54 +2026-09-16 05:39:05 | INFO | train rank=0 | step=37460/50000 progress=74.9% loss=0.081796 lr=1.623336e-05 lr_backbone=1.623336e-05 elapsed=39:08:43 avg_step=3.76s recent_step=3.17s eta=13:06:15 +2026-09-16 05:39:35 | INFO | train rank=0 | step=37470/50000 progress=74.9% loss=0.081431 lr=1.620898e-05 lr_backbone=1.620898e-05 elapsed=39:09:13 avg_step=3.76s recent_step=3.03s eta=13:05:35 +2026-09-16 05:40:14 | INFO | train rank=0 | step=37480/50000 progress=75.0% loss=0.079068 lr=1.618461e-05 lr_backbone=1.618461e-05 elapsed=39:09:52 avg_step=3.76s recent_step=3.84s eta=13:04:57 +2026-09-16 05:41:05 | INFO | train rank=0 | step=37490/50000 progress=75.0% loss=0.070478 lr=1.616026e-05 lr_backbone=1.616026e-05 elapsed=39:10:43 avg_step=3.76s recent_step=5.18s eta=13:04:24 +2026-09-16 05:41:38 | INFO | train rank=0 | step=37500/50000 progress=75.0% loss=0.080522 lr=1.613592e-05 lr_backbone=1.613592e-05 elapsed=39:11:16 avg_step=3.76s recent_step=3.22s eta=13:03:45 +2026-09-16 05:42:11 | INFO | train rank=0 | step=37510/50000 progress=75.0% loss=0.094766 lr=1.611160e-05 lr_backbone=1.611160e-05 elapsed=39:11:49 avg_step=3.76s recent_step=3.34s eta=13:03:06 +2026-09-16 05:42:43 | INFO | train rank=0 | step=37520/50000 progress=75.0% loss=0.078061 lr=1.608729e-05 lr_backbone=1.608729e-05 elapsed=39:12:21 avg_step=3.76s recent_step=3.18s eta=13:02:26 +2026-09-16 05:43:40 | INFO | train rank=0 | step=37530/50000 progress=75.1% loss=0.088232 lr=1.606300e-05 lr_backbone=1.606300e-05 elapsed=39:13:19 avg_step=3.76s recent_step=5.77s eta=13:01:55 +2026-09-16 05:44:21 | INFO | train rank=0 | step=37540/50000 progress=75.1% loss=0.084276 lr=1.603872e-05 lr_backbone=1.603872e-05 elapsed=39:13:59 avg_step=3.76s recent_step=4.05s eta=13:01:19 +2026-09-16 05:44:54 | INFO | train rank=0 | step=37550/50000 progress=75.1% loss=0.096737 lr=1.601446e-05 lr_backbone=1.601446e-05 elapsed=39:14:32 avg_step=3.76s recent_step=3.34s eta=13:00:40 +2026-09-16 05:45:29 | INFO | train rank=0 | step=37560/50000 progress=75.1% loss=0.070528 lr=1.599021e-05 lr_backbone=1.599021e-05 elapsed=39:15:07 avg_step=3.76s recent_step=3.48s eta=13:00:01 +2026-09-16 05:46:17 | INFO | train rank=0 | step=37570/50000 progress=75.1% loss=0.061889 lr=1.596598e-05 lr_backbone=1.596598e-05 elapsed=39:15:55 avg_step=3.76s recent_step=4.77s eta=12:59:27 +2026-09-16 05:46:48 | INFO | train rank=0 | step=37580/50000 progress=75.2% loss=0.089277 lr=1.594176e-05 lr_backbone=1.594176e-05 elapsed=39:16:26 avg_step=3.76s recent_step=3.07s eta=12:58:47 +2026-09-16 05:47:24 | INFO | train rank=0 | step=37590/50000 progress=75.2% loss=0.089219 lr=1.591755e-05 lr_backbone=1.591755e-05 elapsed=39:17:02 avg_step=3.76s recent_step=3.61s eta=12:58:09 +2026-09-16 05:48:14 | INFO | train rank=0 | step=37600/50000 progress=75.2% loss=0.079753 lr=1.589336e-05 lr_backbone=1.589336e-05 elapsed=39:17:52 avg_step=3.76s recent_step=5.02s eta=12:57:35 +2026-09-16 05:48:45 | INFO | train rank=0 | step=37610/50000 progress=75.2% loss=0.083161 lr=1.586919e-05 lr_backbone=1.586919e-05 elapsed=39:18:23 avg_step=3.76s recent_step=3.15s eta=12:56:56 +2026-09-16 05:49:22 | INFO | train rank=0 | step=37620/50000 progress=75.2% loss=0.093657 lr=1.584503e-05 lr_backbone=1.584503e-05 elapsed=39:19:00 avg_step=3.76s recent_step=3.71s eta=12:56:18 +2026-09-16 05:49:54 | INFO | train rank=0 | step=37630/50000 progress=75.3% loss=0.098601 lr=1.582089e-05 lr_backbone=1.582089e-05 elapsed=39:19:32 avg_step=3.76s recent_step=3.17s eta=12:55:38 +2026-09-16 05:50:49 | INFO | train rank=0 | step=37640/50000 progress=75.3% loss=0.090196 lr=1.579676e-05 lr_backbone=1.579676e-05 elapsed=39:20:27 avg_step=3.76s recent_step=5.47s eta=12:55:06 +2026-09-16 05:51:22 | INFO | train rank=0 | step=37650/50000 progress=75.3% loss=0.083459 lr=1.577264e-05 lr_backbone=1.577264e-05 elapsed=39:21:00 avg_step=3.76s recent_step=3.37s eta=12:54:27 +2026-09-16 05:51:52 | INFO | train rank=0 | step=37660/50000 progress=75.3% loss=0.082424 lr=1.574855e-05 lr_backbone=1.574855e-05 elapsed=39:21:30 avg_step=3.76s recent_step=2.99s eta=12:53:47 +2026-09-16 05:52:32 | INFO | train rank=0 | step=37670/50000 progress=75.3% loss=0.074148 lr=1.572446e-05 lr_backbone=1.572446e-05 elapsed=39:22:10 avg_step=3.76s recent_step=3.91s eta=12:53:10 +2026-09-16 05:53:06 | INFO | train rank=0 | step=37680/50000 progress=75.4% loss=0.081228 lr=1.570039e-05 lr_backbone=1.570039e-05 elapsed=39:22:44 avg_step=3.76s recent_step=3.42s eta=12:52:31 +2026-09-16 05:53:56 | INFO | train rank=0 | step=37690/50000 progress=75.4% loss=0.085772 lr=1.567634e-05 lr_backbone=1.567634e-05 elapsed=39:23:34 avg_step=3.76s recent_step=4.99s eta=12:51:58 +2026-09-16 05:54:33 | INFO | train rank=0 | step=37700/50000 progress=75.4% loss=0.076246 lr=1.565230e-05 lr_backbone=1.565230e-05 elapsed=39:24:11 avg_step=3.76s recent_step=3.71s eta=12:51:20 +2026-09-16 05:55:07 | INFO | train rank=0 | step=37710/50000 progress=75.4% loss=0.083328 lr=1.562828e-05 lr_backbone=1.562828e-05 elapsed=39:24:45 avg_step=3.76s recent_step=3.46s eta=12:50:41 +2026-09-16 05:55:51 | INFO | train rank=0 | step=37720/50000 progress=75.4% loss=0.083005 lr=1.560427e-05 lr_backbone=1.560427e-05 elapsed=39:25:29 avg_step=3.76s recent_step=4.33s eta=12:50:05 +2026-09-16 05:56:31 | INFO | train rank=0 | step=37730/50000 progress=75.5% loss=0.099428 lr=1.558027e-05 lr_backbone=1.558027e-05 elapsed=39:26:09 avg_step=3.76s recent_step=4.01s eta=12:49:29 +2026-09-16 05:56:54 | INFO | train rank=0 | step=37740/50000 progress=75.5% loss=0.068905 lr=1.555629e-05 lr_backbone=1.555629e-05 elapsed=39:26:32 avg_step=3.76s recent_step=2.35s eta=12:48:46 +2026-09-16 05:57:22 | INFO | train rank=0 | step=37750/50000 progress=75.5% loss=0.107752 lr=1.553233e-05 lr_backbone=1.553233e-05 elapsed=39:27:00 avg_step=3.76s recent_step=2.80s eta=12:48:06 +2026-09-16 05:58:13 | INFO | train rank=0 | step=37760/50000 progress=75.5% loss=0.077926 lr=1.550838e-05 lr_backbone=1.550838e-05 elapsed=39:27:51 avg_step=3.76s recent_step=5.12s eta=12:47:32 +2026-09-16 05:58:53 | INFO | train rank=0 | step=37770/50000 progress=75.5% loss=0.094408 lr=1.548445e-05 lr_backbone=1.548445e-05 elapsed=39:28:31 avg_step=3.76s recent_step=3.94s eta=12:46:55 +2026-09-16 05:59:30 | INFO | train rank=0 | step=37780/50000 progress=75.6% loss=0.089252 lr=1.546053e-05 lr_backbone=1.546053e-05 elapsed=39:29:08 avg_step=3.76s recent_step=3.69s eta=12:46:18 +2026-09-16 05:59:59 | INFO | train rank=0 | step=37790/50000 progress=75.6% loss=0.088195 lr=1.543663e-05 lr_backbone=1.543663e-05 elapsed=39:29:37 avg_step=3.76s recent_step=2.94s eta=12:45:37 +2026-09-16 06:00:40 | INFO | train rank=0 | step=37800/50000 progress=75.6% loss=0.063694 lr=1.541274e-05 lr_backbone=1.541274e-05 elapsed=39:30:18 avg_step=3.76s recent_step=4.08s eta=12:45:01 +2026-09-16 06:01:16 | INFO | train rank=0 | step=37810/50000 progress=75.6% loss=0.094240 lr=1.538886e-05 lr_backbone=1.538886e-05 elapsed=39:30:54 avg_step=3.76s recent_step=3.61s eta=12:44:23 +2026-09-16 06:01:47 | INFO | train rank=0 | step=37820/50000 progress=75.6% loss=0.072667 lr=1.536501e-05 lr_backbone=1.536501e-05 elapsed=39:31:25 avg_step=3.76s recent_step=3.09s eta=12:43:43 +2026-09-16 06:02:21 | INFO | train rank=0 | step=37830/50000 progress=75.7% loss=0.089034 lr=1.534116e-05 lr_backbone=1.534116e-05 elapsed=39:31:59 avg_step=3.76s recent_step=3.43s eta=12:43:04 +2026-09-16 06:03:16 | INFO | train rank=0 | step=37840/50000 progress=75.7% loss=0.102927 lr=1.531733e-05 lr_backbone=1.531733e-05 elapsed=39:32:54 avg_step=3.76s recent_step=5.44s eta=12:42:32 +2026-09-16 06:04:02 | INFO | train rank=0 | step=37850/50000 progress=75.7% loss=0.093398 lr=1.529352e-05 lr_backbone=1.529352e-05 elapsed=39:33:40 avg_step=3.76s recent_step=4.60s eta=12:41:57 +2026-09-16 06:04:34 | INFO | train rank=0 | step=37860/50000 progress=75.7% loss=0.121609 lr=1.526972e-05 lr_backbone=1.526972e-05 elapsed=39:34:12 avg_step=3.76s recent_step=3.20s eta=12:41:17 +2026-09-16 06:05:04 | INFO | train rank=0 | step=37870/50000 progress=75.7% loss=0.091875 lr=1.524594e-05 lr_backbone=1.524594e-05 elapsed=39:34:42 avg_step=3.76s recent_step=3.02s eta=12:40:37 +2026-09-16 06:05:41 | INFO | train rank=0 | step=37880/50000 progress=75.8% loss=0.095999 lr=1.522218e-05 lr_backbone=1.522218e-05 elapsed=39:35:19 avg_step=3.76s recent_step=3.72s eta=12:40:00 +2026-09-16 06:06:35 | INFO | train rank=0 | step=37890/50000 progress=75.8% loss=0.071450 lr=1.519842e-05 lr_backbone=1.519842e-05 elapsed=39:36:13 avg_step=3.76s recent_step=5.37s eta=12:39:27 +2026-09-16 06:07:02 | INFO | train rank=0 | step=37900/50000 progress=75.8% loss=0.082302 lr=1.517469e-05 lr_backbone=1.517469e-05 elapsed=39:36:40 avg_step=3.76s recent_step=2.76s eta=12:38:46 +2026-09-16 06:07:48 | INFO | train rank=0 | step=37910/50000 progress=75.8% loss=0.080414 lr=1.515097e-05 lr_backbone=1.515097e-05 elapsed=39:37:26 avg_step=3.76s recent_step=4.62s eta=12:38:12 +2026-09-16 06:08:19 | INFO | train rank=0 | step=37920/50000 progress=75.8% loss=0.113232 lr=1.512726e-05 lr_backbone=1.512726e-05 elapsed=39:37:57 avg_step=3.76s recent_step=3.06s eta=12:37:32 +2026-09-16 06:09:10 | INFO | train rank=0 | step=37930/50000 progress=75.9% loss=0.082411 lr=1.510357e-05 lr_backbone=1.510357e-05 elapsed=39:38:48 avg_step=3.76s recent_step=5.07s eta=12:36:58 +2026-09-16 06:09:45 | INFO | train rank=0 | step=37940/50000 progress=75.9% loss=0.074181 lr=1.507989e-05 lr_backbone=1.507989e-05 elapsed=39:39:23 avg_step=3.76s recent_step=3.48s eta=12:36:20 +2026-09-16 06:10:24 | INFO | train rank=0 | step=37950/50000 progress=75.9% loss=0.094730 lr=1.505623e-05 lr_backbone=1.505623e-05 elapsed=39:40:02 avg_step=3.76s recent_step=3.90s eta=12:35:42 +2026-09-16 06:10:55 | INFO | train rank=0 | step=37960/50000 progress=75.9% loss=0.074643 lr=1.503259e-05 lr_backbone=1.503259e-05 elapsed=39:40:33 avg_step=3.76s recent_step=3.14s eta=12:35:03 +2026-09-16 06:11:43 | INFO | train rank=0 | step=37970/50000 progress=75.9% loss=0.109477 lr=1.500896e-05 lr_backbone=1.500896e-05 elapsed=39:41:21 avg_step=3.76s recent_step=4.79s eta=12:34:28 +2026-09-16 06:12:13 | INFO | train rank=0 | step=37980/50000 progress=76.0% loss=0.093363 lr=1.498534e-05 lr_backbone=1.498534e-05 elapsed=39:41:51 avg_step=3.76s recent_step=2.98s eta=12:33:48 +2026-09-16 06:12:47 | INFO | train rank=0 | step=37990/50000 progress=76.0% loss=0.099997 lr=1.496174e-05 lr_backbone=1.496174e-05 elapsed=39:42:25 avg_step=3.76s recent_step=3.43s eta=12:33:10 +2026-09-16 06:13:17 | INFO | train rank=0 | step=38000/50000 progress=76.0% loss=0.080313 lr=1.493816e-05 lr_backbone=1.493816e-05 elapsed=39:42:55 avg_step=3.76s recent_step=2.98s eta=12:32:30 +2026-09-16 06:14:02 | INFO | train rank=0 | step=38010/50000 progress=76.0% loss=0.096253 lr=1.491459e-05 lr_backbone=1.491459e-05 elapsed=39:43:40 avg_step=3.76s recent_step=4.48s eta=12:31:54 +2026-09-16 06:14:33 | INFO | train rank=0 | step=38020/50000 progress=76.0% loss=0.080315 lr=1.489104e-05 lr_backbone=1.489104e-05 elapsed=39:44:11 avg_step=3.76s recent_step=3.17s eta=12:31:15 +2026-09-16 06:15:04 | INFO | train rank=0 | step=38030/50000 progress=76.1% loss=0.084881 lr=1.486750e-05 lr_backbone=1.486750e-05 elapsed=39:44:42 avg_step=3.76s recent_step=3.08s eta=12:30:35 +2026-09-16 06:15:38 | INFO | train rank=0 | step=38040/50000 progress=76.1% loss=0.099373 lr=1.484398e-05 lr_backbone=1.484398e-05 elapsed=39:45:16 avg_step=3.76s recent_step=3.35s eta=12:29:56 +2026-09-16 06:16:25 | INFO | train rank=0 | step=38050/50000 progress=76.1% loss=0.114792 lr=1.482047e-05 lr_backbone=1.482047e-05 elapsed=39:46:03 avg_step=3.76s recent_step=4.76s eta=12:29:22 +2026-09-16 06:17:04 | INFO | train rank=0 | step=38060/50000 progress=76.1% loss=0.087933 lr=1.479698e-05 lr_backbone=1.479698e-05 elapsed=39:46:42 avg_step=3.76s recent_step=3.84s eta=12:28:44 +2026-09-16 06:17:31 | INFO | train rank=0 | step=38070/50000 progress=76.1% loss=0.075121 lr=1.477350e-05 lr_backbone=1.477350e-05 elapsed=39:47:09 avg_step=3.76s recent_step=2.77s eta=12:28:03 +2026-09-16 06:18:04 | INFO | train rank=0 | step=38080/50000 progress=76.2% loss=0.108993 lr=1.475004e-05 lr_backbone=1.475004e-05 elapsed=39:47:42 avg_step=3.76s recent_step=3.24s eta=12:27:24 +2026-09-16 06:18:52 | INFO | train rank=0 | step=38090/50000 progress=76.2% loss=0.096700 lr=1.472660e-05 lr_backbone=1.472660e-05 elapsed=39:48:30 avg_step=3.76s recent_step=4.86s eta=12:26:50 +2026-09-16 06:19:20 | INFO | train rank=0 | step=38100/50000 progress=76.2% loss=0.084071 lr=1.470317e-05 lr_backbone=1.470317e-05 elapsed=39:48:58 avg_step=3.76s recent_step=2.78s eta=12:26:09 +2026-09-16 06:19:51 | INFO | train rank=0 | step=38110/50000 progress=76.2% loss=0.091880 lr=1.467975e-05 lr_backbone=1.467975e-05 elapsed=39:49:29 avg_step=3.76s recent_step=3.04s eta=12:25:29 +2026-09-16 06:20:47 | INFO | train rank=0 | step=38120/50000 progress=76.2% loss=0.113471 lr=1.465635e-05 lr_backbone=1.465635e-05 elapsed=39:50:25 avg_step=3.76s recent_step=5.62s eta=12:24:58 +2026-09-16 06:21:15 | INFO | train rank=0 | step=38130/50000 progress=76.3% loss=0.079532 lr=1.463297e-05 lr_backbone=1.463297e-05 elapsed=39:50:53 avg_step=3.76s recent_step=2.86s eta=12:24:17 +2026-09-16 06:21:55 | INFO | train rank=0 | step=38140/50000 progress=76.3% loss=0.078344 lr=1.460960e-05 lr_backbone=1.460960e-05 elapsed=39:51:33 avg_step=3.76s recent_step=3.94s eta=12:23:40 +2026-09-16 06:22:20 | INFO | train rank=0 | step=38150/50000 progress=76.3% loss=0.086173 lr=1.458625e-05 lr_backbone=1.458625e-05 elapsed=39:51:58 avg_step=3.76s recent_step=2.56s eta=12:22:59 +2026-09-16 06:23:21 | INFO | train rank=0 | step=38160/50000 progress=76.3% loss=0.074326 lr=1.456291e-05 lr_backbone=1.456291e-05 elapsed=39:52:59 avg_step=3.76s recent_step=6.06s eta=12:22:28 +2026-09-16 06:23:51 | INFO | train rank=0 | step=38170/50000 progress=76.3% loss=0.085516 lr=1.453959e-05 lr_backbone=1.453959e-05 elapsed=39:53:29 avg_step=3.76s recent_step=3.00s eta=12:21:48 +2026-09-16 06:24:30 | INFO | train rank=0 | step=38180/50000 progress=76.4% loss=0.100049 lr=1.451629e-05 lr_backbone=1.451629e-05 elapsed=39:54:08 avg_step=3.76s recent_step=3.92s eta=12:21:11 +2026-09-16 06:25:02 | INFO | train rank=0 | step=38190/50000 progress=76.4% loss=0.077689 lr=1.449299e-05 lr_backbone=1.449299e-05 elapsed=39:54:40 avg_step=3.76s recent_step=3.15s eta=12:20:32 +2026-09-16 06:25:59 | INFO | train rank=0 | step=38200/50000 progress=76.4% loss=0.103771 lr=1.446972e-05 lr_backbone=1.446972e-05 elapsed=39:55:37 avg_step=3.76s recent_step=5.77s eta=12:20:00 +2026-09-16 06:26:30 | INFO | train rank=0 | step=38210/50000 progress=76.4% loss=0.083030 lr=1.444646e-05 lr_backbone=1.444646e-05 elapsed=39:56:08 avg_step=3.76s recent_step=3.08s eta=12:19:20 +2026-09-16 06:27:05 | INFO | train rank=0 | step=38220/50000 progress=76.4% loss=0.090035 lr=1.442322e-05 lr_backbone=1.442322e-05 elapsed=39:56:43 avg_step=3.76s recent_step=3.46s eta=12:18:42 +2026-09-16 06:27:37 | INFO | train rank=0 | step=38230/50000 progress=76.5% loss=0.069945 lr=1.439999e-05 lr_backbone=1.439999e-05 elapsed=39:57:15 avg_step=3.76s recent_step=3.24s eta=12:18:03 +2026-09-16 06:28:13 | INFO | train rank=0 | step=38240/50000 progress=76.5% loss=0.087570 lr=1.437677e-05 lr_backbone=1.437677e-05 elapsed=39:57:51 avg_step=3.76s recent_step=3.61s eta=12:17:25 +2026-09-16 06:28:46 | INFO | train rank=0 | step=38250/50000 progress=76.5% loss=0.094703 lr=1.435358e-05 lr_backbone=1.435358e-05 elapsed=39:58:24 avg_step=3.76s recent_step=3.31s eta=12:16:46 +2026-09-16 06:29:17 | INFO | train rank=0 | step=38260/50000 progress=76.5% loss=0.075806 lr=1.433040e-05 lr_backbone=1.433040e-05 elapsed=39:58:55 avg_step=3.76s recent_step=3.09s eta=12:16:06 +2026-09-16 06:29:52 | INFO | train rank=0 | step=38270/50000 progress=76.5% loss=0.083926 lr=1.430723e-05 lr_backbone=1.430723e-05 elapsed=39:59:30 avg_step=3.76s recent_step=3.49s eta=12:15:27 +2026-09-16 06:30:44 | INFO | train rank=0 | step=38280/50000 progress=76.6% loss=0.092016 lr=1.428408e-05 lr_backbone=1.428408e-05 elapsed=40:00:22 avg_step=3.76s recent_step=5.20s eta=12:14:54 +2026-09-16 06:31:21 | INFO | train rank=0 | step=38290/50000 progress=76.6% loss=0.075652 lr=1.426094e-05 lr_backbone=1.426094e-05 elapsed=40:00:59 avg_step=3.76s recent_step=3.65s eta=12:14:16 +2026-09-16 06:31:58 | INFO | train rank=0 | step=38300/50000 progress=76.6% loss=0.074471 lr=1.423783e-05 lr_backbone=1.423783e-05 elapsed=40:01:36 avg_step=3.76s recent_step=3.70s eta=12:13:38 +2026-09-16 06:32:30 | INFO | train rank=0 | step=38310/50000 progress=76.6% loss=0.081396 lr=1.421472e-05 lr_backbone=1.421472e-05 elapsed=40:02:08 avg_step=3.76s recent_step=3.26s eta=12:12:59 +2026-09-16 06:33:16 | INFO | train rank=0 | step=38320/50000 progress=76.6% loss=0.090270 lr=1.419163e-05 lr_backbone=1.419163e-05 elapsed=40:02:54 avg_step=3.76s recent_step=4.62s eta=12:12:24 +2026-09-16 06:33:47 | INFO | train rank=0 | step=38330/50000 progress=76.7% loss=0.090863 lr=1.416856e-05 lr_backbone=1.416856e-05 elapsed=40:03:25 avg_step=3.76s recent_step=3.03s eta=12:11:44 +2026-09-16 06:34:17 | INFO | train rank=0 | step=38340/50000 progress=76.7% loss=0.112478 lr=1.414551e-05 lr_backbone=1.414551e-05 elapsed=40:03:55 avg_step=3.76s recent_step=2.98s eta=12:11:04 +2026-09-16 06:34:52 | INFO | train rank=0 | step=38350/50000 progress=76.7% loss=0.074880 lr=1.412246e-05 lr_backbone=1.412246e-05 elapsed=40:04:30 avg_step=3.76s recent_step=3.59s eta=12:10:26 +2026-09-16 06:35:30 | INFO | train rank=0 | step=38360/50000 progress=76.7% loss=0.097055 lr=1.409944e-05 lr_backbone=1.409944e-05 elapsed=40:05:08 avg_step=3.76s recent_step=3.78s eta=12:09:49 +2026-09-16 06:36:17 | INFO | train rank=0 | step=38370/50000 progress=76.7% loss=0.093899 lr=1.407643e-05 lr_backbone=1.407643e-05 elapsed=40:05:55 avg_step=3.76s recent_step=4.73s eta=12:09:14 +2026-09-16 06:36:45 | INFO | train rank=0 | step=38380/50000 progress=76.8% loss=0.078174 lr=1.405344e-05 lr_backbone=1.405344e-05 elapsed=40:06:23 avg_step=3.76s recent_step=2.76s eta=12:08:33 +2026-09-16 06:37:20 | INFO | train rank=0 | step=38390/50000 progress=76.8% loss=0.080434 lr=1.403046e-05 lr_backbone=1.403046e-05 elapsed=40:06:58 avg_step=3.76s recent_step=3.45s eta=12:07:55 +2026-09-16 06:37:57 | INFO | train rank=0 | step=38400/50000 progress=76.8% loss=0.080325 lr=1.400750e-05 lr_backbone=1.400750e-05 elapsed=40:07:35 avg_step=3.76s recent_step=3.73s eta=12:07:17 +2026-09-16 06:38:50 | INFO | train rank=0 | step=38410/50000 progress=76.8% loss=0.094546 lr=1.398455e-05 lr_backbone=1.398455e-05 elapsed=40:08:28 avg_step=3.76s recent_step=5.32s eta=12:06:44 +2026-09-16 06:39:19 | INFO | train rank=0 | step=38420/50000 progress=76.8% loss=0.096069 lr=1.396162e-05 lr_backbone=1.396162e-05 elapsed=40:08:57 avg_step=3.76s recent_step=2.92s eta=12:06:04 +2026-09-16 06:39:57 | INFO | train rank=0 | step=38430/50000 progress=76.9% loss=0.094306 lr=1.393870e-05 lr_backbone=1.393870e-05 elapsed=40:09:35 avg_step=3.76s recent_step=3.76s eta=12:05:26 +2026-09-16 06:40:26 | INFO | train rank=0 | step=38440/50000 progress=76.9% loss=0.073193 lr=1.391580e-05 lr_backbone=1.391580e-05 elapsed=40:10:04 avg_step=3.76s recent_step=2.96s eta=12:04:46 +2026-09-16 06:41:24 | INFO | train rank=0 | step=38450/50000 progress=76.9% loss=0.073280 lr=1.389292e-05 lr_backbone=1.389292e-05 elapsed=40:11:02 avg_step=3.76s recent_step=5.77s eta=12:04:15 +2026-09-16 06:41:55 | INFO | train rank=0 | step=38460/50000 progress=76.9% loss=0.079169 lr=1.387005e-05 lr_backbone=1.387005e-05 elapsed=40:11:33 avg_step=3.76s recent_step=3.14s eta=12:03:35 +2026-09-16 06:42:33 | INFO | train rank=0 | step=38470/50000 progress=76.9% loss=0.094975 lr=1.384720e-05 lr_backbone=1.384720e-05 elapsed=40:12:11 avg_step=3.76s recent_step=3.78s eta=12:02:58 +2026-09-16 06:43:09 | INFO | train rank=0 | step=38480/50000 progress=77.0% loss=0.103140 lr=1.382436e-05 lr_backbone=1.382436e-05 elapsed=40:12:47 avg_step=3.76s recent_step=3.52s eta=12:02:19 +2026-09-16 06:43:55 | INFO | train rank=0 | step=38490/50000 progress=77.0% loss=0.098370 lr=1.380154e-05 lr_backbone=1.380154e-05 elapsed=40:13:33 avg_step=3.76s recent_step=4.62s eta=12:01:44 +2026-09-16 06:44:27 | INFO | train rank=0 | step=38500/50000 progress=77.0% loss=0.081285 lr=1.377874e-05 lr_backbone=1.377874e-05 elapsed=40:14:05 avg_step=3.76s recent_step=3.25s eta=12:01:05 +2026-09-16 06:44:57 | INFO | train rank=0 | step=38510/50000 progress=77.0% loss=0.069546 lr=1.375595e-05 lr_backbone=1.375595e-05 elapsed=40:14:35 avg_step=3.76s recent_step=2.97s eta=12:00:25 +2026-09-16 06:45:35 | INFO | train rank=0 | step=38520/50000 progress=77.0% loss=0.086807 lr=1.373318e-05 lr_backbone=1.373318e-05 elapsed=40:15:13 avg_step=3.76s recent_step=3.84s eta=11:59:48 +2026-09-16 06:46:26 | INFO | train rank=0 | step=38530/50000 progress=77.1% loss=0.078277 lr=1.371042e-05 lr_backbone=1.371042e-05 elapsed=40:16:04 avg_step=3.76s recent_step=5.06s eta=11:59:14 +2026-09-16 06:46:54 | INFO | train rank=0 | step=38540/50000 progress=77.1% loss=0.092309 lr=1.368768e-05 lr_backbone=1.368768e-05 elapsed=40:16:32 avg_step=3.76s recent_step=2.80s eta=11:58:34 +2026-09-16 06:47:22 | INFO | train rank=0 | step=38550/50000 progress=77.1% loss=0.080002 lr=1.366496e-05 lr_backbone=1.366496e-05 elapsed=40:17:00 avg_step=3.76s recent_step=2.85s eta=11:57:53 +2026-09-16 06:47:57 | INFO | train rank=0 | step=38560/50000 progress=77.1% loss=0.078761 lr=1.364225e-05 lr_backbone=1.364225e-05 elapsed=40:17:35 avg_step=3.76s recent_step=3.47s eta=11:57:15 +2026-09-16 06:48:53 | INFO | train rank=0 | step=38570/50000 progress=77.1% loss=0.080838 lr=1.361955e-05 lr_backbone=1.361955e-05 elapsed=40:18:31 avg_step=3.76s recent_step=5.58s eta=11:56:42 +2026-09-16 06:49:36 | INFO | train rank=0 | step=38580/50000 progress=77.2% loss=0.093080 lr=1.359688e-05 lr_backbone=1.359688e-05 elapsed=40:19:14 avg_step=3.76s recent_step=4.30s eta=11:56:06 +2026-09-16 06:50:03 | INFO | train rank=0 | step=38590/50000 progress=77.2% loss=0.105647 lr=1.357421e-05 lr_backbone=1.357421e-05 elapsed=40:19:41 avg_step=3.76s recent_step=2.73s eta=11:55:26 +2026-09-16 06:50:39 | INFO | train rank=0 | step=38600/50000 progress=77.2% loss=0.103030 lr=1.355157e-05 lr_backbone=1.355157e-05 elapsed=40:20:17 avg_step=3.76s recent_step=3.59s eta=11:54:48 +2026-09-16 06:51:30 | INFO | train rank=0 | step=38610/50000 progress=77.2% loss=0.071923 lr=1.352894e-05 lr_backbone=1.352894e-05 elapsed=40:21:08 avg_step=3.76s recent_step=5.07s eta=11:54:14 +2026-09-16 06:52:02 | INFO | train rank=0 | step=38620/50000 progress=77.2% loss=0.098977 lr=1.350633e-05 lr_backbone=1.350633e-05 elapsed=40:21:40 avg_step=3.76s recent_step=3.21s eta=11:53:35 +2026-09-16 06:52:29 | INFO | train rank=0 | step=38630/50000 progress=77.3% loss=0.084137 lr=1.348373e-05 lr_backbone=1.348373e-05 elapsed=40:22:07 avg_step=3.76s recent_step=2.72s eta=11:52:54 +2026-09-16 06:53:02 | INFO | train rank=0 | step=38640/50000 progress=77.3% loss=0.083920 lr=1.346115e-05 lr_backbone=1.346115e-05 elapsed=40:22:40 avg_step=3.76s recent_step=3.29s eta=11:52:15 +2026-09-16 06:53:50 | INFO | train rank=0 | step=38650/50000 progress=77.3% loss=0.097172 lr=1.343858e-05 lr_backbone=1.343858e-05 elapsed=40:23:28 avg_step=3.76s recent_step=4.84s eta=11:51:40 +2026-09-16 06:54:25 | INFO | train rank=0 | step=38660/50000 progress=77.3% loss=0.077629 lr=1.341603e-05 lr_backbone=1.341603e-05 elapsed=40:24:03 avg_step=3.76s recent_step=3.47s eta=11:51:02 +2026-09-16 06:54:59 | INFO | train rank=0 | step=38670/50000 progress=77.3% loss=0.094761 lr=1.339350e-05 lr_backbone=1.339350e-05 elapsed=40:24:37 avg_step=3.76s recent_step=3.43s eta=11:50:23 +2026-09-16 06:55:35 | INFO | train rank=0 | step=38680/50000 progress=77.4% loss=0.086563 lr=1.337098e-05 lr_backbone=1.337098e-05 elapsed=40:25:13 avg_step=3.76s recent_step=3.53s eta=11:49:45 +2026-09-16 06:56:33 | INFO | train rank=0 | step=38690/50000 progress=77.4% loss=0.084725 lr=1.334848e-05 lr_backbone=1.334848e-05 elapsed=40:26:11 avg_step=3.76s recent_step=5.84s eta=11:49:14 +2026-09-16 06:57:05 | INFO | train rank=0 | step=38700/50000 progress=77.4% loss=0.096448 lr=1.332599e-05 lr_backbone=1.332599e-05 elapsed=40:26:43 avg_step=3.76s recent_step=3.22s eta=11:48:34 +2026-09-16 06:57:40 | INFO | train rank=0 | step=38710/50000 progress=77.4% loss=0.081189 lr=1.330352e-05 lr_backbone=1.330352e-05 elapsed=40:27:18 avg_step=3.76s recent_step=3.51s eta=11:47:56 +2026-09-16 06:58:10 | INFO | train rank=0 | step=38720/50000 progress=77.4% loss=0.101698 lr=1.328107e-05 lr_backbone=1.328107e-05 elapsed=40:27:48 avg_step=3.76s recent_step=2.99s eta=11:47:16 +2026-09-16 06:59:05 | INFO | train rank=0 | step=38730/50000 progress=77.5% loss=0.087241 lr=1.325863e-05 lr_backbone=1.325863e-05 elapsed=40:28:43 avg_step=3.76s recent_step=5.46s eta=11:46:43 +2026-09-16 06:59:37 | INFO | train rank=0 | step=38740/50000 progress=77.5% loss=0.093422 lr=1.323621e-05 lr_backbone=1.323621e-05 elapsed=40:29:15 avg_step=3.76s recent_step=3.18s eta=11:46:04 +2026-09-16 07:00:06 | INFO | train rank=0 | step=38750/50000 progress=77.5% loss=0.075303 lr=1.321380e-05 lr_backbone=1.321380e-05 elapsed=40:29:44 avg_step=3.76s recent_step=2.94s eta=11:45:24 +2026-09-16 07:00:36 | INFO | train rank=0 | step=38760/50000 progress=77.5% loss=0.086713 lr=1.319142e-05 lr_backbone=1.319142e-05 elapsed=40:30:14 avg_step=3.76s recent_step=3.00s eta=11:44:44 +2026-09-16 07:01:08 | INFO | train rank=0 | step=38770/50000 progress=77.5% loss=0.089973 lr=1.316904e-05 lr_backbone=1.316904e-05 elapsed=40:30:46 avg_step=3.76s recent_step=3.16s eta=11:44:05 +2026-09-16 07:01:48 | INFO | train rank=0 | step=38780/50000 progress=77.6% loss=0.092045 lr=1.314669e-05 lr_backbone=1.314669e-05 elapsed=40:31:26 avg_step=3.76s recent_step=4.08s eta=11:43:28 +2026-09-16 07:02:24 | INFO | train rank=0 | step=38790/50000 progress=77.6% loss=0.077175 lr=1.312434e-05 lr_backbone=1.312434e-05 elapsed=40:32:02 avg_step=3.76s recent_step=3.58s eta=11:42:50 +2026-09-16 07:02:55 | INFO | train rank=0 | step=38800/50000 progress=77.6% loss=0.112577 lr=1.310202e-05 lr_backbone=1.310202e-05 elapsed=40:32:33 avg_step=3.76s recent_step=3.11s eta=11:42:11 +2026-09-16 07:03:42 | INFO | train rank=0 | step=38810/50000 progress=77.6% loss=0.075852 lr=1.307971e-05 lr_backbone=1.307971e-05 elapsed=40:33:20 avg_step=3.76s recent_step=4.62s eta=11:41:35 +2026-09-16 07:04:11 | INFO | train rank=0 | step=38820/50000 progress=77.6% loss=0.103309 lr=1.305742e-05 lr_backbone=1.305742e-05 elapsed=40:33:49 avg_step=3.76s recent_step=2.99s eta=11:40:56 +2026-09-16 07:04:43 | INFO | train rank=0 | step=38830/50000 progress=77.7% loss=0.080829 lr=1.303514e-05 lr_backbone=1.303514e-05 elapsed=40:34:21 avg_step=3.76s recent_step=3.14s eta=11:40:16 +2026-09-16 07:05:26 | INFO | train rank=0 | step=38840/50000 progress=77.7% loss=0.117681 lr=1.301288e-05 lr_backbone=1.301288e-05 elapsed=40:35:04 avg_step=3.76s recent_step=4.32s eta=11:39:40 +2026-09-16 07:06:11 | INFO | train rank=0 | step=38850/50000 progress=77.7% loss=0.088594 lr=1.299064e-05 lr_backbone=1.299064e-05 elapsed=40:35:49 avg_step=3.76s recent_step=4.47s eta=11:39:05 +2026-09-16 07:06:41 | INFO | train rank=0 | step=38860/50000 progress=77.7% loss=0.127140 lr=1.296841e-05 lr_backbone=1.296841e-05 elapsed=40:36:19 avg_step=3.76s recent_step=3.00s eta=11:38:25 +2026-09-16 07:07:20 | INFO | train rank=0 | step=38870/50000 progress=77.7% loss=0.058739 lr=1.294620e-05 lr_backbone=1.294620e-05 elapsed=40:36:58 avg_step=3.76s recent_step=3.90s eta=11:37:47 +2026-09-16 07:07:54 | INFO | train rank=0 | step=38880/50000 progress=77.8% loss=0.097911 lr=1.292400e-05 lr_backbone=1.292400e-05 elapsed=40:37:32 avg_step=3.76s recent_step=3.41s eta=11:37:09 +2026-09-16 07:08:32 | INFO | train rank=0 | step=38890/50000 progress=77.8% loss=0.085682 lr=1.290182e-05 lr_backbone=1.290182e-05 elapsed=40:38:10 avg_step=3.76s recent_step=3.83s eta=11:36:31 +2026-09-16 07:09:07 | INFO | train rank=0 | step=38900/50000 progress=77.8% loss=0.082477 lr=1.287966e-05 lr_backbone=1.287966e-05 elapsed=40:38:45 avg_step=3.76s recent_step=3.48s eta=11:35:53 +2026-09-16 07:09:50 | INFO | train rank=0 | step=38910/50000 progress=77.8% loss=0.113870 lr=1.285751e-05 lr_backbone=1.285751e-05 elapsed=40:39:28 avg_step=3.76s recent_step=4.35s eta=11:35:17 +2026-09-16 07:10:31 | INFO | train rank=0 | step=38920/50000 progress=77.8% loss=0.098962 lr=1.283538e-05 lr_backbone=1.283538e-05 elapsed=40:40:09 avg_step=3.76s recent_step=4.06s eta=11:34:40 +2026-09-16 07:11:07 | INFO | train rank=0 | step=38930/50000 progress=77.9% loss=0.081230 lr=1.281327e-05 lr_backbone=1.281327e-05 elapsed=40:40:45 avg_step=3.76s recent_step=3.58s eta=11:34:02 +2026-09-16 07:11:40 | INFO | train rank=0 | step=38940/50000 progress=77.9% loss=0.079855 lr=1.279117e-05 lr_backbone=1.279117e-05 elapsed=40:41:18 avg_step=3.76s recent_step=3.31s eta=11:33:23 +2026-09-16 07:12:26 | INFO | train rank=0 | step=38950/50000 progress=77.9% loss=0.076851 lr=1.276909e-05 lr_backbone=1.276909e-05 elapsed=40:42:04 avg_step=3.76s recent_step=4.60s eta=11:32:48 +2026-09-16 07:13:05 | INFO | train rank=0 | step=38960/50000 progress=77.9% loss=0.083247 lr=1.274702e-05 lr_backbone=1.274702e-05 elapsed=40:42:43 avg_step=3.76s recent_step=3.88s eta=11:32:11 +2026-09-16 07:13:42 | INFO | train rank=0 | step=38970/50000 progress=77.9% loss=0.081952 lr=1.272498e-05 lr_backbone=1.272498e-05 elapsed=40:43:20 avg_step=3.76s recent_step=3.74s eta=11:31:33 +2026-09-16 07:14:18 | INFO | train rank=0 | step=38980/50000 progress=78.0% loss=0.084048 lr=1.270294e-05 lr_backbone=1.270294e-05 elapsed=40:43:56 avg_step=3.76s recent_step=3.55s eta=11:30:55 +2026-09-16 07:15:06 | INFO | train rank=0 | step=38990/50000 progress=78.0% loss=0.087626 lr=1.268093e-05 lr_backbone=1.268093e-05 elapsed=40:44:44 avg_step=3.76s recent_step=4.87s eta=11:30:20 +2026-09-16 07:15:40 | INFO | train rank=0 | step=39000/50000 progress=78.0% loss=0.106062 lr=1.265893e-05 lr_backbone=1.265893e-05 elapsed=40:45:18 avg_step=3.76s recent_step=3.32s eta=11:29:42 +2026-09-16 07:16:19 | INFO | train rank=0 | step=39010/50000 progress=78.0% loss=0.113992 lr=1.263694e-05 lr_backbone=1.263694e-05 elapsed=40:45:57 avg_step=3.76s recent_step=3.90s eta=11:29:04 +2026-09-16 07:16:49 | INFO | train rank=0 | step=39020/50000 progress=78.0% loss=0.079159 lr=1.261497e-05 lr_backbone=1.261497e-05 elapsed=40:46:27 avg_step=3.76s recent_step=3.09s eta=11:28:25 +2026-09-16 07:17:27 | INFO | train rank=0 | step=39030/50000 progress=78.1% loss=0.106369 lr=1.259302e-05 lr_backbone=1.259302e-05 elapsed=40:47:05 avg_step=3.76s recent_step=3.73s eta=11:27:47 +2026-09-16 07:17:56 | INFO | train rank=0 | step=39040/50000 progress=78.1% loss=0.075444 lr=1.257109e-05 lr_backbone=1.257109e-05 elapsed=40:47:34 avg_step=3.76s recent_step=2.94s eta=11:27:07 +2026-09-16 07:18:30 | INFO | train rank=0 | step=39050/50000 progress=78.1% loss=0.087006 lr=1.254917e-05 lr_backbone=1.254917e-05 elapsed=40:48:08 avg_step=3.76s recent_step=3.39s eta=11:26:28 +2026-09-16 07:18:59 | INFO | train rank=0 | step=39060/50000 progress=78.1% loss=0.078388 lr=1.252727e-05 lr_backbone=1.252727e-05 elapsed=40:48:37 avg_step=3.76s recent_step=2.87s eta=11:25:48 +2026-09-16 07:19:51 | INFO | train rank=0 | step=39070/50000 progress=78.1% loss=0.090516 lr=1.250538e-05 lr_backbone=1.250538e-05 elapsed=40:49:29 avg_step=3.76s recent_step=5.26s eta=11:25:15 +2026-09-16 07:20:17 | INFO | train rank=0 | step=39080/50000 progress=78.2% loss=0.092177 lr=1.248351e-05 lr_backbone=1.248351e-05 elapsed=40:49:55 avg_step=3.76s recent_step=2.62s eta=11:24:34 +2026-09-16 07:20:52 | INFO | train rank=0 | step=39090/50000 progress=78.2% loss=0.096325 lr=1.246166e-05 lr_backbone=1.246166e-05 elapsed=40:50:30 avg_step=3.76s recent_step=3.43s eta=11:23:56 +2026-09-16 07:21:18 | INFO | train rank=0 | step=39100/50000 progress=78.2% loss=0.075905 lr=1.243982e-05 lr_backbone=1.243982e-05 elapsed=40:50:56 avg_step=3.76s recent_step=2.66s eta=11:23:15 +2026-09-16 07:21:44 | INFO | train rank=0 | step=39110/50000 progress=78.2% loss=0.108603 lr=1.241800e-05 lr_backbone=1.241800e-05 elapsed=40:51:22 avg_step=3.76s recent_step=2.56s eta=11:22:34 +2026-09-16 07:22:09 | INFO | train rank=0 | step=39120/50000 progress=78.2% loss=0.097611 lr=1.239620e-05 lr_backbone=1.239620e-05 elapsed=40:51:47 avg_step=3.76s recent_step=2.54s eta=11:21:53 +2026-09-16 07:22:49 | INFO | train rank=0 | step=39130/50000 progress=78.3% loss=0.095769 lr=1.237441e-05 lr_backbone=1.237441e-05 elapsed=40:52:27 avg_step=3.76s recent_step=3.94s eta=11:21:16 +2026-09-16 07:23:18 | INFO | train rank=0 | step=39140/50000 progress=78.3% loss=0.095385 lr=1.235264e-05 lr_backbone=1.235264e-05 elapsed=40:52:56 avg_step=3.76s recent_step=2.94s eta=11:20:36 +2026-09-16 07:23:47 | INFO | train rank=0 | step=39150/50000 progress=78.3% loss=0.089208 lr=1.233089e-05 lr_backbone=1.233089e-05 elapsed=40:53:25 avg_step=3.76s recent_step=2.92s eta=11:19:56 +2026-09-16 07:24:12 | INFO | train rank=0 | step=39160/50000 progress=78.3% loss=0.116665 lr=1.230915e-05 lr_backbone=1.230915e-05 elapsed=40:53:50 avg_step=3.76s recent_step=2.50s eta=11:19:15 +2026-09-16 07:24:42 | INFO | train rank=0 | step=39170/50000 progress=78.3% loss=0.084249 lr=1.228743e-05 lr_backbone=1.228743e-05 elapsed=40:54:20 avg_step=3.76s recent_step=2.96s eta=11:18:35 +2026-09-16 07:25:06 | INFO | train rank=0 | step=39180/50000 progress=78.4% loss=0.079872 lr=1.226573e-05 lr_backbone=1.226573e-05 elapsed=40:54:44 avg_step=3.76s recent_step=2.40s eta=11:17:54 +2026-09-16 07:25:38 | INFO | train rank=0 | step=39190/50000 progress=78.4% loss=0.087373 lr=1.224404e-05 lr_backbone=1.224404e-05 elapsed=40:55:16 avg_step=3.76s recent_step=3.15s eta=11:17:15 +2026-09-16 07:26:11 | INFO | train rank=0 | step=39200/50000 progress=78.4% loss=0.102267 lr=1.222237e-05 lr_backbone=1.222237e-05 elapsed=40:55:49 avg_step=3.76s recent_step=3.37s eta=11:16:36 +2026-09-16 07:26:42 | INFO | train rank=0 | step=39210/50000 progress=78.4% loss=0.079720 lr=1.220071e-05 lr_backbone=1.220071e-05 elapsed=40:56:20 avg_step=3.76s recent_step=3.05s eta=11:15:56 +2026-09-16 07:27:08 | INFO | train rank=0 | step=39220/50000 progress=78.4% loss=0.088491 lr=1.217907e-05 lr_backbone=1.217907e-05 elapsed=40:56:46 avg_step=3.76s recent_step=2.64s eta=11:15:16 +2026-09-16 07:27:31 | INFO | train rank=0 | step=39230/50000 progress=78.5% loss=0.084123 lr=1.215745e-05 lr_backbone=1.215745e-05 elapsed=40:57:09 avg_step=3.76s recent_step=2.33s eta=11:14:34 +2026-09-16 07:28:09 | INFO | train rank=0 | step=39240/50000 progress=78.5% loss=0.089067 lr=1.213585e-05 lr_backbone=1.213585e-05 elapsed=40:57:47 avg_step=3.76s recent_step=3.76s eta=11:13:57 +2026-09-16 07:28:41 | INFO | train rank=0 | step=39250/50000 progress=78.5% loss=0.071531 lr=1.211426e-05 lr_backbone=1.211426e-05 elapsed=40:58:19 avg_step=3.76s recent_step=3.22s eta=11:13:18 +2026-09-16 07:29:12 | INFO | train rank=0 | step=39260/50000 progress=78.5% loss=0.089927 lr=1.209268e-05 lr_backbone=1.209268e-05 elapsed=40:58:50 avg_step=3.76s recent_step=3.04s eta=11:12:38 +2026-09-16 07:29:38 | INFO | train rank=0 | step=39270/50000 progress=78.5% loss=0.098755 lr=1.207113e-05 lr_backbone=1.207113e-05 elapsed=40:59:16 avg_step=3.76s recent_step=2.61s eta=11:11:57 +2026-09-16 07:30:05 | INFO | train rank=0 | step=39280/50000 progress=78.6% loss=0.079376 lr=1.204959e-05 lr_backbone=1.204959e-05 elapsed=40:59:43 avg_step=3.76s recent_step=2.77s eta=11:11:17 +2026-09-16 07:30:35 | INFO | train rank=0 | step=39290/50000 progress=78.6% loss=0.068503 lr=1.202807e-05 lr_backbone=1.202807e-05 elapsed=41:00:13 avg_step=3.76s recent_step=2.97s eta=11:10:37 +2026-09-16 07:30:59 | INFO | train rank=0 | step=39300/50000 progress=78.6% loss=0.089756 lr=1.200656e-05 lr_backbone=1.200656e-05 elapsed=41:00:37 avg_step=3.76s recent_step=2.43s eta=11:09:56 +2026-09-16 07:31:27 | INFO | train rank=0 | step=39310/50000 progress=78.6% loss=0.094219 lr=1.198507e-05 lr_backbone=1.198507e-05 elapsed=41:01:05 avg_step=3.76s recent_step=2.78s eta=11:09:16 +2026-09-16 07:32:03 | INFO | train rank=0 | step=39320/50000 progress=78.6% loss=0.097240 lr=1.196360e-05 lr_backbone=1.196360e-05 elapsed=41:01:41 avg_step=3.76s recent_step=3.59s eta=11:08:38 +2026-09-16 07:32:30 | INFO | train rank=0 | step=39330/50000 progress=78.7% loss=0.073705 lr=1.194214e-05 lr_backbone=1.194214e-05 elapsed=41:02:08 avg_step=3.76s recent_step=2.66s eta=11:07:57 +2026-09-16 07:32:55 | INFO | train rank=0 | step=39340/50000 progress=78.7% loss=0.093587 lr=1.192070e-05 lr_backbone=1.192070e-05 elapsed=41:02:33 avg_step=3.76s recent_step=2.58s eta=11:07:17 +2026-09-16 07:33:19 | INFO | train rank=0 | step=39350/50000 progress=78.7% loss=0.082189 lr=1.189928e-05 lr_backbone=1.189928e-05 elapsed=41:02:57 avg_step=3.76s recent_step=2.33s eta=11:06:35 +2026-09-16 07:33:48 | INFO | train rank=0 | step=39360/50000 progress=78.7% loss=0.105955 lr=1.187787e-05 lr_backbone=1.187787e-05 elapsed=41:03:26 avg_step=3.76s recent_step=2.96s eta=11:05:55 +2026-09-16 07:34:22 | INFO | train rank=0 | step=39370/50000 progress=78.7% loss=0.078934 lr=1.185649e-05 lr_backbone=1.185649e-05 elapsed=41:04:00 avg_step=3.76s recent_step=3.31s eta=11:05:17 +2026-09-16 07:34:54 | INFO | train rank=0 | step=39380/50000 progress=78.8% loss=0.087478 lr=1.183511e-05 lr_backbone=1.183511e-05 elapsed=41:04:32 avg_step=3.76s recent_step=3.26s eta=11:04:38 +2026-09-16 07:35:24 | INFO | train rank=0 | step=39390/50000 progress=78.8% loss=0.093004 lr=1.181376e-05 lr_backbone=1.181376e-05 elapsed=41:05:02 avg_step=3.75s recent_step=2.96s eta=11:03:58 +2026-09-16 07:35:50 | INFO | train rank=0 | step=39400/50000 progress=78.8% loss=0.096322 lr=1.179242e-05 lr_backbone=1.179242e-05 elapsed=41:05:28 avg_step=3.75s recent_step=2.64s eta=11:03:18 +2026-09-16 07:36:22 | INFO | train rank=0 | step=39410/50000 progress=78.8% loss=0.092428 lr=1.177109e-05 lr_backbone=1.177109e-05 elapsed=41:06:00 avg_step=3.75s recent_step=3.21s eta=11:02:39 +2026-09-16 07:36:52 | INFO | train rank=0 | step=39420/50000 progress=78.8% loss=0.083275 lr=1.174979e-05 lr_backbone=1.174979e-05 elapsed=41:06:30 avg_step=3.75s recent_step=2.96s eta=11:01:59 +2026-09-16 07:37:21 | INFO | train rank=0 | step=39430/50000 progress=78.9% loss=0.089890 lr=1.172850e-05 lr_backbone=1.172850e-05 elapsed=41:06:59 avg_step=3.75s recent_step=2.96s eta=11:01:19 +2026-09-16 07:37:52 | INFO | train rank=0 | step=39440/50000 progress=78.9% loss=0.098687 lr=1.170723e-05 lr_backbone=1.170723e-05 elapsed=41:07:30 avg_step=3.75s recent_step=3.03s eta=11:00:40 +2026-09-16 07:38:22 | INFO | train rank=0 | step=39450/50000 progress=78.9% loss=0.095113 lr=1.168597e-05 lr_backbone=1.168597e-05 elapsed=41:08:00 avg_step=3.75s recent_step=3.03s eta=11:00:00 +2026-09-16 07:38:50 | INFO | train rank=0 | step=39460/50000 progress=78.9% loss=0.119669 lr=1.166473e-05 lr_backbone=1.166473e-05 elapsed=41:08:28 avg_step=3.75s recent_step=2.76s eta=10:59:20 +2026-09-16 07:39:22 | INFO | train rank=0 | step=39470/50000 progress=78.9% loss=0.087345 lr=1.164351e-05 lr_backbone=1.164351e-05 elapsed=41:09:00 avg_step=3.75s recent_step=3.26s eta=10:58:41 +2026-09-16 07:39:50 | INFO | train rank=0 | step=39480/50000 progress=79.0% loss=0.100363 lr=1.162230e-05 lr_backbone=1.162230e-05 elapsed=41:09:28 avg_step=3.75s recent_step=2.72s eta=10:58:01 +2026-09-16 07:40:20 | INFO | train rank=0 | step=39490/50000 progress=79.0% loss=0.093731 lr=1.160112e-05 lr_backbone=1.160112e-05 elapsed=41:09:58 avg_step=3.75s recent_step=3.06s eta=10:57:22 +2026-09-16 07:40:47 | INFO | train rank=0 | step=39500/50000 progress=79.0% loss=0.068637 lr=1.157994e-05 lr_backbone=1.157994e-05 elapsed=41:10:25 avg_step=3.75s recent_step=2.65s eta=10:56:41 +2026-09-16 07:41:17 | INFO | train rank=0 | step=39510/50000 progress=79.0% loss=0.084955 lr=1.155879e-05 lr_backbone=1.155879e-05 elapsed=41:10:55 avg_step=3.75s recent_step=3.02s eta=10:56:02 +2026-09-16 07:41:41 | INFO | train rank=0 | step=39520/50000 progress=79.0% loss=0.085695 lr=1.153765e-05 lr_backbone=1.153765e-05 elapsed=41:11:19 avg_step=3.75s recent_step=2.41s eta=10:55:21 +2026-09-16 07:42:18 | INFO | train rank=0 | step=39530/50000 progress=79.1% loss=0.094964 lr=1.151653e-05 lr_backbone=1.151653e-05 elapsed=41:11:56 avg_step=3.75s recent_step=3.75s eta=10:54:43 +2026-09-16 07:42:51 | INFO | train rank=0 | step=39540/50000 progress=79.1% loss=0.075947 lr=1.149543e-05 lr_backbone=1.149543e-05 elapsed=41:12:29 avg_step=3.75s recent_step=3.23s eta=10:54:04 +2026-09-16 07:43:11 | INFO | train rank=0 | step=39550/50000 progress=79.1% loss=0.107921 lr=1.147434e-05 lr_backbone=1.147434e-05 elapsed=41:12:49 avg_step=3.75s recent_step=2.01s eta=10:53:22 +2026-09-16 07:43:38 | INFO | train rank=0 | step=39560/50000 progress=79.1% loss=0.086381 lr=1.145327e-05 lr_backbone=1.145327e-05 elapsed=41:13:16 avg_step=3.75s recent_step=2.68s eta=10:52:42 +2026-09-16 07:44:15 | INFO | train rank=0 | step=39570/50000 progress=79.1% loss=0.076995 lr=1.143221e-05 lr_backbone=1.143221e-05 elapsed=41:13:53 avg_step=3.75s recent_step=3.69s eta=10:52:04 +2026-09-16 07:44:40 | INFO | train rank=0 | step=39580/50000 progress=79.2% loss=0.092984 lr=1.141118e-05 lr_backbone=1.141118e-05 elapsed=41:14:18 avg_step=3.75s recent_step=2.56s eta=10:51:23 +2026-09-16 07:44:59 | INFO | train rank=0 | step=39590/50000 progress=79.2% loss=0.069956 lr=1.139016e-05 lr_backbone=1.139016e-05 elapsed=41:14:37 avg_step=3.75s recent_step=1.85s eta=10:50:41 +2026-09-16 07:45:18 | INFO | train rank=0 | step=39600/50000 progress=79.2% loss=0.107512 lr=1.136915e-05 lr_backbone=1.136915e-05 elapsed=41:14:56 avg_step=3.75s recent_step=1.97s eta=10:49:59 +2026-09-16 07:45:58 | INFO | train rank=0 | step=39610/50000 progress=79.2% loss=0.066913 lr=1.134817e-05 lr_backbone=1.134817e-05 elapsed=41:15:36 avg_step=3.75s recent_step=3.93s eta=10:49:22 +2026-09-16 07:46:23 | INFO | train rank=0 | step=39620/50000 progress=79.2% loss=0.093632 lr=1.132720e-05 lr_backbone=1.132720e-05 elapsed=41:16:01 avg_step=3.75s recent_step=2.57s eta=10:48:41 +2026-09-16 07:46:53 | INFO | train rank=0 | step=39630/50000 progress=79.3% loss=0.081478 lr=1.130624e-05 lr_backbone=1.130624e-05 elapsed=41:16:31 avg_step=3.75s recent_step=2.91s eta=10:48:01 +2026-09-16 07:47:18 | INFO | train rank=0 | step=39640/50000 progress=79.3% loss=0.061642 lr=1.128531e-05 lr_backbone=1.128531e-05 elapsed=41:16:57 avg_step=3.75s recent_step=2.60s eta=10:47:21 +2026-09-16 07:47:56 | INFO | train rank=0 | step=39650/50000 progress=79.3% loss=0.086427 lr=1.126439e-05 lr_backbone=1.126439e-05 elapsed=41:17:34 avg_step=3.75s recent_step=3.79s eta=10:46:44 +2026-09-16 07:48:15 | INFO | train rank=0 | step=39660/50000 progress=79.3% loss=0.082896 lr=1.124349e-05 lr_backbone=1.124349e-05 elapsed=41:17:53 avg_step=3.75s recent_step=1.85s eta=10:46:01 +2026-09-16 07:48:42 | INFO | train rank=0 | step=39670/50000 progress=79.3% loss=0.070202 lr=1.122260e-05 lr_backbone=1.122260e-05 elapsed=41:18:20 avg_step=3.75s recent_step=2.72s eta=10:45:21 +2026-09-16 07:49:22 | INFO | train rank=0 | step=39680/50000 progress=79.4% loss=0.066510 lr=1.120174e-05 lr_backbone=1.120174e-05 elapsed=41:19:00 avg_step=3.75s recent_step=3.96s eta=10:44:44 +2026-09-16 07:49:49 | INFO | train rank=0 | step=39690/50000 progress=79.4% loss=0.075575 lr=1.118088e-05 lr_backbone=1.118088e-05 elapsed=41:19:27 avg_step=3.75s recent_step=2.73s eta=10:44:04 +2026-09-16 07:50:23 | INFO | train rank=0 | step=39700/50000 progress=79.4% loss=0.085790 lr=1.116005e-05 lr_backbone=1.116005e-05 elapsed=41:20:01 avg_step=3.75s recent_step=3.39s eta=10:43:25 +2026-09-16 07:50:51 | INFO | train rank=0 | step=39710/50000 progress=79.4% loss=0.086127 lr=1.113923e-05 lr_backbone=1.113923e-05 elapsed=41:20:29 avg_step=3.75s recent_step=2.77s eta=10:42:45 +2026-09-16 07:51:25 | INFO | train rank=0 | step=39720/50000 progress=79.4% loss=0.079845 lr=1.111843e-05 lr_backbone=1.111843e-05 elapsed=41:21:03 avg_step=3.75s recent_step=3.39s eta=10:42:07 +2026-09-16 07:51:52 | INFO | train rank=0 | step=39730/50000 progress=79.5% loss=0.073696 lr=1.109765e-05 lr_backbone=1.109765e-05 elapsed=41:21:30 avg_step=3.75s recent_step=2.70s eta=10:41:27 +2026-09-16 07:52:19 | INFO | train rank=0 | step=39740/50000 progress=79.5% loss=0.094452 lr=1.107688e-05 lr_backbone=1.107688e-05 elapsed=41:21:57 avg_step=3.75s recent_step=2.73s eta=10:40:47 +2026-09-16 07:52:52 | INFO | train rank=0 | step=39750/50000 progress=79.5% loss=0.097153 lr=1.105614e-05 lr_backbone=1.105614e-05 elapsed=41:22:30 avg_step=3.75s recent_step=3.35s eta=10:40:08 +2026-09-16 07:53:28 | INFO | train rank=0 | step=39760/50000 progress=79.5% loss=0.118924 lr=1.103540e-05 lr_backbone=1.103540e-05 elapsed=41:23:06 avg_step=3.75s recent_step=3.58s eta=10:39:30 +2026-09-16 07:53:57 | INFO | train rank=0 | step=39770/50000 progress=79.5% loss=0.068122 lr=1.101469e-05 lr_backbone=1.101469e-05 elapsed=41:23:35 avg_step=3.75s recent_step=2.85s eta=10:38:51 +2026-09-16 07:54:25 | INFO | train rank=0 | step=39780/50000 progress=79.6% loss=0.080028 lr=1.099399e-05 lr_backbone=1.099399e-05 elapsed=41:24:03 avg_step=3.75s recent_step=2.86s eta=10:38:11 +2026-09-16 07:54:45 | INFO | train rank=0 | step=39790/50000 progress=79.6% loss=0.082341 lr=1.097331e-05 lr_backbone=1.097331e-05 elapsed=41:24:23 avg_step=3.75s recent_step=1.93s eta=10:37:29 +2026-09-16 07:55:20 | INFO | train rank=0 | step=39800/50000 progress=79.6% loss=0.083094 lr=1.095265e-05 lr_backbone=1.095265e-05 elapsed=41:24:58 avg_step=3.75s recent_step=3.55s eta=10:36:51 +2026-09-16 07:55:46 | INFO | train rank=0 | step=39810/50000 progress=79.6% loss=0.116021 lr=1.093200e-05 lr_backbone=1.093200e-05 elapsed=41:25:24 avg_step=3.75s recent_step=2.61s eta=10:36:10 +2026-09-16 07:56:13 | INFO | train rank=0 | step=39820/50000 progress=79.6% loss=0.078240 lr=1.091137e-05 lr_backbone=1.091137e-05 elapsed=41:25:51 avg_step=3.75s recent_step=2.70s eta=10:35:30 +2026-09-16 07:56:42 | INFO | train rank=0 | step=39830/50000 progress=79.7% loss=0.084921 lr=1.089076e-05 lr_backbone=1.089076e-05 elapsed=41:26:20 avg_step=3.75s recent_step=2.92s eta=10:34:51 +2026-09-16 07:57:15 | INFO | train rank=0 | step=39840/50000 progress=79.7% loss=0.078517 lr=1.087016e-05 lr_backbone=1.087016e-05 elapsed=41:26:53 avg_step=3.75s recent_step=3.28s eta=10:34:12 +2026-09-16 07:57:43 | INFO | train rank=0 | step=39850/50000 progress=79.7% loss=0.081534 lr=1.084959e-05 lr_backbone=1.084959e-05 elapsed=41:27:21 avg_step=3.75s recent_step=2.79s eta=10:33:32 +2026-09-16 07:58:05 | INFO | train rank=0 | step=39860/50000 progress=79.7% loss=0.082384 lr=1.082902e-05 lr_backbone=1.082902e-05 elapsed=41:27:43 avg_step=3.74s recent_step=2.23s eta=10:32:51 +2026-09-16 07:58:32 | INFO | train rank=0 | step=39870/50000 progress=79.7% loss=0.104104 lr=1.080848e-05 lr_backbone=1.080848e-05 elapsed=41:28:10 avg_step=3.74s recent_step=2.69s eta=10:32:11 +2026-09-16 07:59:06 | INFO | train rank=0 | step=39880/50000 progress=79.8% loss=0.080612 lr=1.078795e-05 lr_backbone=1.078795e-05 elapsed=41:28:44 avg_step=3.74s recent_step=3.38s eta=10:31:32 +2026-09-16 07:59:35 | INFO | train rank=0 | step=39890/50000 progress=79.8% loss=0.072953 lr=1.076744e-05 lr_backbone=1.076744e-05 elapsed=41:29:13 avg_step=3.74s recent_step=2.85s eta=10:30:53 +2026-09-16 07:59:54 | INFO | train rank=0 | step=39900/50000 progress=79.8% loss=0.105824 lr=1.074695e-05 lr_backbone=1.074695e-05 elapsed=41:29:32 avg_step=3.74s recent_step=1.98s eta=10:30:11 +2026-09-16 08:00:22 | INFO | train rank=0 | step=39910/50000 progress=79.8% loss=0.079138 lr=1.072648e-05 lr_backbone=1.072648e-05 elapsed=41:30:00 avg_step=3.74s recent_step=2.78s eta=10:29:31 +2026-09-16 08:00:56 | INFO | train rank=0 | step=39920/50000 progress=79.8% loss=0.096189 lr=1.070602e-05 lr_backbone=1.070602e-05 elapsed=41:30:34 avg_step=3.74s recent_step=3.39s eta=10:28:52 +2026-09-16 08:01:23 | INFO | train rank=0 | step=39930/50000 progress=79.9% loss=0.082301 lr=1.068558e-05 lr_backbone=1.068558e-05 elapsed=41:31:01 avg_step=3.74s recent_step=2.65s eta=10:28:12 +2026-09-16 08:01:51 | INFO | train rank=0 | step=39940/50000 progress=79.9% loss=0.097806 lr=1.066515e-05 lr_backbone=1.066515e-05 elapsed=41:31:29 avg_step=3.74s recent_step=2.88s eta=10:27:33 +2026-09-16 08:02:17 | INFO | train rank=0 | step=39950/50000 progress=79.9% loss=0.098451 lr=1.064475e-05 lr_backbone=1.064475e-05 elapsed=41:31:55 avg_step=3.74s recent_step=2.55s eta=10:26:52 +2026-09-16 08:02:53 | INFO | train rank=0 | step=39960/50000 progress=79.9% loss=0.081268 lr=1.062436e-05 lr_backbone=1.062436e-05 elapsed=41:32:31 avg_step=3.74s recent_step=3.58s eta=10:26:14 +2026-09-16 08:03:19 | INFO | train rank=0 | step=39970/50000 progress=79.9% loss=0.092061 lr=1.060399e-05 lr_backbone=1.060399e-05 elapsed=41:32:57 avg_step=3.74s recent_step=2.63s eta=10:25:34 +2026-09-16 08:03:50 | INFO | train rank=0 | step=39980/50000 progress=80.0% loss=0.077713 lr=1.058363e-05 lr_backbone=1.058363e-05 elapsed=41:33:28 avg_step=3.74s recent_step=3.06s eta=10:24:55 +2026-09-16 08:04:16 | INFO | train rank=0 | step=39990/50000 progress=80.0% loss=0.097826 lr=1.056329e-05 lr_backbone=1.056329e-05 elapsed=41:33:54 avg_step=3.74s recent_step=2.62s eta=10:24:15 +2026-09-16 08:04:43 | INFO | train rank=0 | step=40000/50000 progress=80.0% loss=0.091597 lr=1.054297e-05 lr_backbone=1.054297e-05 elapsed=41:34:21 avg_step=3.74s recent_step=2.68s eta=10:23:35 +2026-09-16 08:05:03 | INFO | train rank=0 | step=40010/50000 progress=80.0% loss=0.070996 lr=1.052267e-05 lr_backbone=1.052267e-05 elapsed=41:34:41 avg_step=3.74s recent_step=2.07s eta=10:22:53 +2026-09-16 08:05:28 | INFO | train rank=0 | step=40020/50000 progress=80.0% loss=0.088043 lr=1.050239e-05 lr_backbone=1.050239e-05 elapsed=41:35:06 avg_step=3.74s recent_step=2.51s eta=10:22:13 +2026-09-16 08:05:52 | INFO | train rank=0 | step=40030/50000 progress=80.1% loss=0.106988 lr=1.048212e-05 lr_backbone=1.048212e-05 elapsed=41:35:30 avg_step=3.74s recent_step=2.37s eta=10:21:32 +2026-09-16 08:06:17 | INFO | train rank=0 | step=40040/50000 progress=80.1% loss=0.080102 lr=1.046187e-05 lr_backbone=1.046187e-05 elapsed=41:35:55 avg_step=3.74s recent_step=2.44s eta=10:20:51 +2026-09-16 08:06:41 | INFO | train rank=0 | step=40050/50000 progress=80.1% loss=0.092486 lr=1.044163e-05 lr_backbone=1.044163e-05 elapsed=41:36:19 avg_step=3.74s recent_step=2.49s eta=10:20:11 +2026-09-16 08:07:10 | INFO | train rank=0 | step=40060/50000 progress=80.1% loss=0.104880 lr=1.042142e-05 lr_backbone=1.042142e-05 elapsed=41:36:48 avg_step=3.74s recent_step=2.83s eta=10:19:31 +2026-09-16 08:07:34 | INFO | train rank=0 | step=40070/50000 progress=80.1% loss=0.083947 lr=1.040122e-05 lr_backbone=1.040122e-05 elapsed=41:37:12 avg_step=3.74s recent_step=2.41s eta=10:18:50 +2026-09-16 08:08:07 | INFO | train rank=0 | step=40080/50000 progress=80.2% loss=0.086744 lr=1.038103e-05 lr_backbone=1.038103e-05 elapsed=41:37:45 avg_step=3.74s recent_step=3.32s eta=10:18:12 +2026-09-16 08:08:32 | INFO | train rank=0 | step=40090/50000 progress=80.2% loss=0.082888 lr=1.036087e-05 lr_backbone=1.036087e-05 elapsed=41:38:10 avg_step=3.74s recent_step=2.51s eta=10:17:32 +2026-09-16 08:09:00 | INFO | train rank=0 | step=40100/50000 progress=80.2% loss=0.099730 lr=1.034072e-05 lr_backbone=1.034072e-05 elapsed=41:38:38 avg_step=3.74s recent_step=2.82s eta=10:16:52 +2026-09-16 08:09:26 | INFO | train rank=0 | step=40110/50000 progress=80.2% loss=0.077124 lr=1.032059e-05 lr_backbone=1.032059e-05 elapsed=41:39:05 avg_step=3.74s recent_step=2.60s eta=10:16:12 +2026-09-16 08:09:59 | INFO | train rank=0 | step=40120/50000 progress=80.2% loss=0.071293 lr=1.030048e-05 lr_backbone=1.030048e-05 elapsed=41:39:37 avg_step=3.74s recent_step=3.24s eta=10:15:33 +2026-09-16 08:10:23 | INFO | train rank=0 | step=40130/50000 progress=80.3% loss=0.094734 lr=1.028038e-05 lr_backbone=1.028038e-05 elapsed=41:40:01 avg_step=3.74s recent_step=2.36s eta=10:14:52 +2026-09-16 08:10:49 | INFO | train rank=0 | step=40140/50000 progress=80.3% loss=0.096537 lr=1.026031e-05 lr_backbone=1.026031e-05 elapsed=41:40:27 avg_step=3.74s recent_step=2.68s eta=10:14:12 +2026-09-16 08:11:16 | INFO | train rank=0 | step=40150/50000 progress=80.3% loss=0.093391 lr=1.024025e-05 lr_backbone=1.024025e-05 elapsed=41:40:54 avg_step=3.74s recent_step=2.68s eta=10:13:32 +2026-09-16 08:11:50 | INFO | train rank=0 | step=40160/50000 progress=80.3% loss=0.077326 lr=1.022020e-05 lr_backbone=1.022020e-05 elapsed=41:41:28 avg_step=3.74s recent_step=3.40s eta=10:12:54 +2026-09-16 08:12:16 | INFO | train rank=0 | step=40170/50000 progress=80.3% loss=0.085911 lr=1.020018e-05 lr_backbone=1.020018e-05 elapsed=41:41:54 avg_step=3.74s recent_step=2.61s eta=10:12:14 +2026-09-16 08:12:44 | INFO | train rank=0 | step=40180/50000 progress=80.4% loss=0.115372 lr=1.018017e-05 lr_backbone=1.018017e-05 elapsed=41:42:22 avg_step=3.74s recent_step=2.75s eta=10:11:34 +2026-09-16 08:13:10 | INFO | train rank=0 | step=40190/50000 progress=80.4% loss=0.101501 lr=1.016018e-05 lr_backbone=1.016018e-05 elapsed=41:42:48 avg_step=3.74s recent_step=2.65s eta=10:10:54 +2026-09-16 08:13:47 | INFO | train rank=0 | step=40200/50000 progress=80.4% loss=0.101308 lr=1.014021e-05 lr_backbone=1.014021e-05 elapsed=41:43:25 avg_step=3.74s recent_step=3.69s eta=10:10:17 +2026-09-16 08:14:09 | INFO | train rank=0 | step=40210/50000 progress=80.4% loss=0.081712 lr=1.012025e-05 lr_backbone=1.012025e-05 elapsed=41:43:47 avg_step=3.74s recent_step=2.16s eta=10:09:36 +2026-09-16 08:14:31 | INFO | train rank=0 | step=40220/50000 progress=80.4% loss=0.093360 lr=1.010031e-05 lr_backbone=1.010031e-05 elapsed=41:44:09 avg_step=3.74s recent_step=2.19s eta=10:08:54 +2026-09-16 08:14:54 | INFO | train rank=0 | step=40230/50000 progress=80.5% loss=0.100704 lr=1.008039e-05 lr_backbone=1.008039e-05 elapsed=41:44:32 avg_step=3.74s recent_step=2.33s eta=10:08:14 +2026-09-16 08:15:15 | INFO | train rank=0 | step=40240/50000 progress=80.5% loss=0.100016 lr=1.006049e-05 lr_backbone=1.006049e-05 elapsed=41:44:53 avg_step=3.73s recent_step=2.06s eta=10:07:32 +2026-09-16 08:15:33 | INFO | train rank=0 | step=40250/50000 progress=80.5% loss=0.062713 lr=1.004060e-05 lr_backbone=1.004060e-05 elapsed=41:45:11 avg_step=3.73s recent_step=1.88s eta=10:06:50 +2026-09-16 08:15:55 | INFO | train rank=0 | step=40260/50000 progress=80.5% loss=0.106861 lr=1.002073e-05 lr_backbone=1.002073e-05 elapsed=41:45:33 avg_step=3.73s recent_step=2.11s eta=10:06:09 +2026-09-16 08:16:16 | INFO | train rank=0 | step=40270/50000 progress=80.5% loss=0.091634 lr=1.000088e-05 lr_backbone=1.000088e-05 elapsed=41:45:54 avg_step=3.73s recent_step=2.15s eta=10:05:28 +2026-09-16 08:16:36 | INFO | train rank=0 | step=40280/50000 progress=80.6% loss=0.090012 lr=9.981047e-06 lr_backbone=9.981047e-06 elapsed=41:46:14 avg_step=3.73s recent_step=2.03s eta=10:04:47 +2026-09-16 08:17:01 | INFO | train rank=0 | step=40290/50000 progress=80.6% loss=0.110671 lr=9.961231e-06 lr_backbone=9.961231e-06 elapsed=41:46:39 avg_step=3.73s recent_step=2.45s eta=10:04:06 +2026-09-16 08:17:26 | INFO | train rank=0 | step=40300/50000 progress=80.6% loss=0.094839 lr=9.941432e-06 lr_backbone=9.941432e-06 elapsed=41:47:04 avg_step=3.73s recent_step=2.56s eta=10:03:26 +2026-09-16 08:17:47 | INFO | train rank=0 | step=40310/50000 progress=80.6% loss=0.079903 lr=9.921651e-06 lr_backbone=9.921651e-06 elapsed=41:47:25 avg_step=3.73s recent_step=2.03s eta=10:02:45 +2026-09-16 08:18:10 | INFO | train rank=0 | step=40320/50000 progress=80.6% loss=0.083153 lr=9.901887e-06 lr_backbone=9.901887e-06 elapsed=41:47:48 avg_step=3.73s recent_step=2.35s eta=10:02:04 +2026-09-16 08:18:37 | INFO | train rank=0 | step=40330/50000 progress=80.7% loss=0.075212 lr=9.882141e-06 lr_backbone=9.882141e-06 elapsed=41:48:15 avg_step=3.73s recent_step=2.72s eta=10:01:24 +2026-09-16 08:19:11 | INFO | train rank=0 | step=40340/50000 progress=80.7% loss=0.134422 lr=9.862413e-06 lr_backbone=9.862413e-06 elapsed=41:48:49 avg_step=3.73s recent_step=3.33s eta=10:00:46 +2026-09-16 08:19:45 | INFO | train rank=0 | step=40350/50000 progress=80.7% loss=0.076957 lr=9.842702e-06 lr_backbone=9.842702e-06 elapsed=41:49:23 avg_step=3.73s recent_step=3.45s eta=10:00:08 +2026-09-16 08:20:19 | INFO | train rank=0 | step=40360/50000 progress=80.7% loss=0.098519 lr=9.823008e-06 lr_backbone=9.823008e-06 elapsed=41:49:57 avg_step=3.73s recent_step=3.42s eta=09:59:30 +2026-09-16 08:21:02 | INFO | train rank=0 | step=40370/50000 progress=80.7% loss=0.078436 lr=9.803332e-06 lr_backbone=9.803332e-06 elapsed=41:50:40 avg_step=3.73s recent_step=4.23s eta=09:58:54 +2026-09-16 08:21:47 | INFO | train rank=0 | step=40380/50000 progress=80.8% loss=0.070785 lr=9.783674e-06 lr_backbone=9.783674e-06 elapsed=41:51:25 avg_step=3.73s recent_step=4.51s eta=09:58:18 +2026-09-16 08:22:58 | INFO | train rank=0 | step=40390/50000 progress=80.8% loss=0.097268 lr=9.764034e-06 lr_backbone=9.764034e-06 elapsed=41:52:36 avg_step=3.73s recent_step=7.16s eta=09:57:49 +2026-09-16 08:24:10 | INFO | train rank=0 | step=40400/50000 progress=80.8% loss=0.083222 lr=9.744411e-06 lr_backbone=9.744411e-06 elapsed=41:53:48 avg_step=3.73s recent_step=7.22s eta=09:57:20 +2026-09-16 08:25:16 | INFO | train rank=0 | step=40410/50000 progress=80.8% loss=0.082156 lr=9.724805e-06 lr_backbone=9.724805e-06 elapsed=41:54:54 avg_step=3.73s recent_step=6.59s eta=09:56:49 +2026-09-16 08:26:19 | INFO | train rank=0 | step=40420/50000 progress=80.8% loss=0.080851 lr=9.705217e-06 lr_backbone=9.705217e-06 elapsed=41:55:57 avg_step=3.73s recent_step=6.26s eta=09:56:18 +2026-09-16 08:27:25 | INFO | train rank=0 | step=40430/50000 progress=80.9% loss=0.077806 lr=9.685647e-06 lr_backbone=9.685647e-06 elapsed=41:57:03 avg_step=3.74s recent_step=6.59s eta=09:55:48 +2026-09-16 08:28:39 | INFO | train rank=0 | step=40440/50000 progress=80.9% loss=0.083035 lr=9.666095e-06 lr_backbone=9.666095e-06 elapsed=41:58:17 avg_step=3.74s recent_step=7.46s eta=09:55:19 +2026-09-16 08:29:42 | INFO | train rank=0 | step=40450/50000 progress=80.9% loss=0.078147 lr=9.646560e-06 lr_backbone=9.646560e-06 elapsed=41:59:20 avg_step=3.74s recent_step=6.25s eta=09:54:48 +2026-09-16 08:30:51 | INFO | train rank=0 | step=40460/50000 progress=80.9% loss=0.076824 lr=9.627043e-06 lr_backbone=9.627043e-06 elapsed=42:00:29 avg_step=3.74s recent_step=6.95s eta=09:54:18 +2026-09-16 08:31:54 | INFO | train rank=0 | step=40470/50000 progress=80.9% loss=0.083752 lr=9.607543e-06 lr_backbone=9.607543e-06 elapsed=42:01:32 avg_step=3.74s recent_step=6.27s eta=09:53:46 +2026-09-16 08:33:10 | INFO | train rank=0 | step=40480/50000 progress=81.0% loss=0.121280 lr=9.588061e-06 lr_backbone=9.588061e-06 elapsed=42:02:48 avg_step=3.74s recent_step=7.59s eta=09:53:18 +2026-09-16 08:34:16 | INFO | train rank=0 | step=40490/50000 progress=81.0% loss=0.095604 lr=9.568597e-06 lr_backbone=9.568597e-06 elapsed=42:03:54 avg_step=3.74s recent_step=6.57s eta=09:52:47 +2026-09-16 08:35:32 | INFO | train rank=0 | step=40500/50000 progress=81.0% loss=0.071914 lr=9.549150e-06 lr_backbone=9.549150e-06 elapsed=42:05:10 avg_step=3.74s recent_step=7.64s eta=09:52:19 +2026-09-16 08:36:45 | INFO | train rank=0 | step=40510/50000 progress=81.0% loss=0.080506 lr=9.529721e-06 lr_backbone=9.529721e-06 elapsed=42:06:23 avg_step=3.74s recent_step=7.31s eta=09:51:50 +2026-09-16 08:37:46 | INFO | train rank=0 | step=40520/50000 progress=81.0% loss=0.061154 lr=9.510310e-06 lr_backbone=9.510310e-06 elapsed=42:07:24 avg_step=3.74s recent_step=6.13s eta=09:51:18 +2026-09-16 08:38:55 | INFO | train rank=0 | step=40530/50000 progress=81.1% loss=0.104610 lr=9.490917e-06 lr_backbone=9.490917e-06 elapsed=42:08:33 avg_step=3.74s recent_step=6.91s eta=09:50:48 +2026-09-16 08:40:05 | INFO | train rank=0 | step=40540/50000 progress=81.1% loss=0.068654 lr=9.471541e-06 lr_backbone=9.471541e-06 elapsed=42:09:43 avg_step=3.74s recent_step=6.93s eta=09:50:18 +2026-09-16 08:41:01 | INFO | train rank=0 | step=40550/50000 progress=81.1% loss=0.082882 lr=9.452183e-06 lr_backbone=9.452183e-06 elapsed=42:10:39 avg_step=3.74s recent_step=5.61s eta=09:49:45 +2026-09-16 08:42:06 | INFO | train rank=0 | step=40560/50000 progress=81.1% loss=0.103496 lr=9.432843e-06 lr_backbone=9.432843e-06 elapsed=42:11:44 avg_step=3.75s recent_step=6.53s eta=09:49:14 +2026-09-16 08:43:15 | INFO | train rank=0 | step=40570/50000 progress=81.1% loss=0.082556 lr=9.413520e-06 lr_backbone=9.413520e-06 elapsed=42:12:53 avg_step=3.75s recent_step=6.93s eta=09:48:44 +2026-09-16 08:44:16 | INFO | train rank=0 | step=40580/50000 progress=81.2% loss=0.095251 lr=9.394216e-06 lr_backbone=9.394216e-06 elapsed=42:13:54 avg_step=3.75s recent_step=6.05s eta=09:48:12 +2026-09-16 08:45:23 | INFO | train rank=0 | step=40590/50000 progress=81.2% loss=0.085610 lr=9.374929e-06 lr_backbone=9.374929e-06 elapsed=42:15:01 avg_step=3.75s recent_step=6.75s eta=09:47:41 +2026-09-16 08:46:37 | INFO | train rank=0 | step=40600/50000 progress=81.2% loss=0.099263 lr=9.355659e-06 lr_backbone=9.355659e-06 elapsed=42:16:15 avg_step=3.75s recent_step=7.31s eta=09:47:12 +2026-09-16 08:47:46 | INFO | train rank=0 | step=40610/50000 progress=81.2% loss=0.123749 lr=9.336408e-06 lr_backbone=9.336408e-06 elapsed=42:17:24 avg_step=3.75s recent_step=6.92s eta=09:46:42 +2026-09-16 08:49:05 | INFO | train rank=0 | step=40620/50000 progress=81.2% loss=0.066287 lr=9.317174e-06 lr_backbone=9.317174e-06 elapsed=42:18:43 avg_step=3.75s recent_step=7.90s eta=09:46:14 +2026-09-16 08:50:14 | INFO | train rank=0 | step=40630/50000 progress=81.3% loss=0.092868 lr=9.297958e-06 lr_backbone=9.297958e-06 elapsed=42:19:52 avg_step=3.75s recent_step=6.97s eta=09:45:44 +2026-09-16 08:51:35 | INFO | train rank=0 | step=40640/50000 progress=81.3% loss=0.096314 lr=9.278760e-06 lr_backbone=9.278760e-06 elapsed=42:21:13 avg_step=3.75s recent_step=8.01s eta=09:45:16 +2026-09-16 08:52:40 | INFO | train rank=0 | step=40650/50000 progress=81.3% loss=0.073989 lr=9.259580e-06 lr_backbone=9.259580e-06 elapsed=42:22:18 avg_step=3.75s recent_step=6.55s eta=09:44:45 +2026-09-16 08:53:41 | INFO | train rank=0 | step=40660/50000 progress=81.3% loss=0.096584 lr=9.240418e-06 lr_backbone=9.240418e-06 elapsed=42:23:19 avg_step=3.75s recent_step=6.12s eta=09:44:13 +2026-09-16 08:54:49 | INFO | train rank=0 | step=40670/50000 progress=81.3% loss=0.078140 lr=9.221273e-06 lr_backbone=9.221273e-06 elapsed=42:24:27 avg_step=3.75s recent_step=6.80s eta=09:43:43 +2026-09-16 08:55:49 | INFO | train rank=0 | step=40680/50000 progress=81.4% loss=0.114205 lr=9.202146e-06 lr_backbone=9.202146e-06 elapsed=42:25:27 avg_step=3.75s recent_step=6.01s eta=09:43:10 +2026-09-16 08:57:03 | INFO | train rank=0 | step=40690/50000 progress=81.4% loss=0.080080 lr=9.183037e-06 lr_backbone=9.183037e-06 elapsed=42:26:41 avg_step=3.76s recent_step=7.33s eta=09:42:41 +2026-09-16 08:58:12 | INFO | train rank=0 | step=40700/50000 progress=81.4% loss=0.070826 lr=9.163946e-06 lr_backbone=9.163946e-06 elapsed=42:27:50 avg_step=3.76s recent_step=6.95s eta=09:42:11 +2026-09-16 08:59:13 | INFO | train rank=0 | step=40710/50000 progress=81.4% loss=0.107212 lr=9.144873e-06 lr_backbone=9.144873e-06 elapsed=42:28:51 avg_step=3.76s recent_step=6.09s eta=09:41:38 +2026-09-16 09:00:17 | INFO | train rank=0 | step=40720/50000 progress=81.4% loss=0.110573 lr=9.125818e-06 lr_backbone=9.125818e-06 elapsed=42:29:55 avg_step=3.76s recent_step=6.38s eta=09:41:07 +2026-09-16 09:01:19 | INFO | train rank=0 | step=40730/50000 progress=81.5% loss=0.109691 lr=9.106780e-06 lr_backbone=9.106780e-06 elapsed=42:30:57 avg_step=3.76s recent_step=6.26s eta=09:40:35 +2026-09-16 09:03:00 | INFO | train rank=0 | step=40740/50000 progress=81.5% loss=0.072730 lr=9.087761e-06 lr_backbone=9.087761e-06 elapsed=42:32:38 avg_step=3.76s recent_step=10.03s eta=09:40:12 +2026-09-16 09:04:53 | INFO | train rank=0 | step=40750/50000 progress=81.5% loss=0.087037 lr=9.068759e-06 lr_backbone=9.068759e-06 elapsed=42:34:31 avg_step=3.76s recent_step=11.36s eta=09:39:51 +2026-09-16 09:06:57 | INFO | train rank=0 | step=40760/50000 progress=81.5% loss=0.090645 lr=9.049776e-06 lr_backbone=9.049776e-06 elapsed=42:36:35 avg_step=3.76s recent_step=12.37s eta=09:39:33 +2026-09-16 09:08:59 | INFO | train rank=0 | step=40770/50000 progress=81.5% loss=0.100698 lr=9.030810e-06 lr_backbone=9.030810e-06 elapsed=42:38:37 avg_step=3.77s recent_step=12.17s eta=09:39:15 +2026-09-16 09:10:47 | INFO | train rank=0 | step=40780/50000 progress=81.6% loss=0.079677 lr=9.011862e-06 lr_backbone=9.011862e-06 elapsed=42:40:25 avg_step=3.77s recent_step=10.80s eta=09:38:53 +2026-09-16 09:12:54 | INFO | train rank=0 | step=40790/50000 progress=81.6% loss=0.078285 lr=8.992932e-06 lr_backbone=8.992932e-06 elapsed=42:42:32 avg_step=3.77s recent_step=12.73s eta=09:38:35 +2026-09-16 09:14:53 | INFO | train rank=0 | step=40800/50000 progress=81.6% loss=0.110544 lr=8.974020e-06 lr_backbone=8.974020e-06 elapsed=42:44:31 avg_step=3.77s recent_step=11.90s eta=09:38:16 +2026-09-16 09:16:51 | INFO | train rank=0 | step=40810/50000 progress=81.6% loss=0.077922 lr=8.955126e-06 lr_backbone=8.955126e-06 elapsed=42:46:29 avg_step=3.77s recent_step=11.84s eta=09:37:56 +2026-09-16 09:20:43 | INFO | train rank=0 | step=40820/50000 progress=81.6% loss=0.085643 lr=8.936249e-06 lr_backbone=8.936249e-06 elapsed=42:50:21 avg_step=3.78s recent_step=23.11s eta=09:38:02 +2026-09-16 09:21:17 | INFO | train rank=0 | step=40830/50000 progress=81.7% loss=0.091129 lr=8.917391e-06 lr_backbone=8.917391e-06 elapsed=42:50:55 avg_step=3.78s recent_step=3.43s eta=09:37:24 +2026-09-16 09:21:54 | INFO | train rank=0 | step=40840/50000 progress=81.7% loss=0.087808 lr=8.898551e-06 lr_backbone=8.898551e-06 elapsed=42:51:32 avg_step=3.78s recent_step=3.72s eta=09:36:46 +2026-09-16 09:22:28 | INFO | train rank=0 | step=40850/50000 progress=81.7% loss=0.092541 lr=8.879729e-06 lr_backbone=8.879729e-06 elapsed=42:52:06 avg_step=3.78s recent_step=3.37s eta=09:36:07 +2026-09-16 09:23:24 | INFO | train rank=0 | step=40860/50000 progress=81.7% loss=0.088932 lr=8.860925e-06 lr_backbone=8.860925e-06 elapsed=42:53:02 avg_step=3.78s recent_step=5.66s eta=09:35:33 +2026-09-16 09:24:04 | INFO | train rank=0 | step=40870/50000 progress=81.7% loss=0.085787 lr=8.842138e-06 lr_backbone=8.842138e-06 elapsed=42:53:42 avg_step=3.78s recent_step=3.99s eta=09:34:56 +2026-09-16 09:24:32 | INFO | train rank=0 | step=40880/50000 progress=81.8% loss=0.082960 lr=8.823370e-06 lr_backbone=8.823370e-06 elapsed=42:54:10 avg_step=3.78s recent_step=2.80s eta=09:34:16 +2026-09-16 09:25:06 | INFO | train rank=0 | step=40890/50000 progress=81.8% loss=0.086976 lr=8.804620e-06 lr_backbone=8.804620e-06 elapsed=42:54:44 avg_step=3.78s recent_step=3.34s eta=09:33:37 +2026-09-16 09:25:48 | INFO | train rank=0 | step=40900/50000 progress=81.8% loss=0.093740 lr=8.785888e-06 lr_backbone=8.785888e-06 elapsed=42:55:26 avg_step=3.78s recent_step=4.27s eta=09:33:01 +2026-09-16 09:26:25 | INFO | train rank=0 | step=40910/50000 progress=81.8% loss=0.086616 lr=8.767173e-06 lr_backbone=8.767173e-06 elapsed=42:56:03 avg_step=3.78s recent_step=3.67s eta=09:32:23 +2026-09-16 09:26:58 | INFO | train rank=0 | step=40920/50000 progress=81.8% loss=0.082423 lr=8.748477e-06 lr_backbone=8.748477e-06 elapsed=42:56:36 avg_step=3.78s recent_step=3.28s eta=09:31:44 +2026-09-16 09:27:30 | INFO | train rank=0 | step=40930/50000 progress=81.9% loss=0.097212 lr=8.729799e-06 lr_backbone=8.729799e-06 elapsed=42:57:08 avg_step=3.78s recent_step=3.19s eta=09:31:05 +2026-09-16 09:28:37 | INFO | train rank=0 | step=40940/50000 progress=81.9% loss=0.083737 lr=8.711139e-06 lr_backbone=8.711139e-06 elapsed=42:58:15 avg_step=3.78s recent_step=6.78s eta=09:30:34 +2026-09-16 09:29:12 | INFO | train rank=0 | step=40950/50000 progress=81.9% loss=0.069211 lr=8.692497e-06 lr_backbone=8.692497e-06 elapsed=42:58:50 avg_step=3.78s recent_step=3.42s eta=09:29:55 +2026-09-16 09:29:42 | INFO | train rank=0 | step=40960/50000 progress=81.9% loss=0.090800 lr=8.673873e-06 lr_backbone=8.673873e-06 elapsed=42:59:20 avg_step=3.78s recent_step=3.09s eta=09:29:16 +2026-09-16 09:30:17 | INFO | train rank=0 | step=40970/50000 progress=81.9% loss=0.084711 lr=8.655267e-06 lr_backbone=8.655267e-06 elapsed=42:59:55 avg_step=3.78s recent_step=3.43s eta=09:28:37 +2026-09-16 09:31:05 | INFO | train rank=0 | step=40980/50000 progress=82.0% loss=0.099452 lr=8.636680e-06 lr_backbone=8.636680e-06 elapsed=43:00:43 avg_step=3.78s recent_step=4.79s eta=09:28:02 +2026-09-16 09:31:44 | INFO | train rank=0 | step=40990/50000 progress=82.0% loss=0.066817 lr=8.618110e-06 lr_backbone=8.618110e-06 elapsed=43:01:22 avg_step=3.78s recent_step=3.90s eta=09:27:24 +2026-09-16 09:32:17 | INFO | train rank=0 | step=41000/50000 progress=82.0% loss=0.066248 lr=8.599558e-06 lr_backbone=8.599558e-06 elapsed=43:01:55 avg_step=3.78s recent_step=3.38s eta=09:26:45 +2026-09-16 09:32:51 | INFO | train rank=0 | step=41010/50000 progress=82.0% loss=0.061897 lr=8.581025e-06 lr_backbone=8.581025e-06 elapsed=43:02:29 avg_step=3.78s recent_step=3.37s eta=09:26:07 +2026-09-16 09:33:53 | INFO | train rank=0 | step=41020/50000 progress=82.0% loss=0.082728 lr=8.562510e-06 lr_backbone=8.562510e-06 elapsed=43:03:31 avg_step=3.78s recent_step=6.17s eta=09:25:34 +2026-09-16 09:34:22 | INFO | train rank=0 | step=41030/50000 progress=82.1% loss=0.099103 lr=8.544012e-06 lr_backbone=8.544012e-06 elapsed=43:04:00 avg_step=3.78s recent_step=2.89s eta=09:24:54 +2026-09-16 09:34:51 | INFO | train rank=0 | step=41040/50000 progress=82.1% loss=0.096698 lr=8.525533e-06 lr_backbone=8.525533e-06 elapsed=43:04:29 avg_step=3.78s recent_step=2.92s eta=09:24:15 +2026-09-16 09:35:26 | INFO | train rank=0 | step=41050/50000 progress=82.1% loss=0.093848 lr=8.507072e-06 lr_backbone=8.507072e-06 elapsed=43:05:05 avg_step=3.78s recent_step=3.56s eta=09:23:37 +2026-09-16 09:36:20 | INFO | train rank=0 | step=41060/50000 progress=82.1% loss=0.063963 lr=8.488630e-06 lr_backbone=8.488630e-06 elapsed=43:05:58 avg_step=3.78s recent_step=5.32s eta=09:23:02 +2026-09-16 09:36:48 | INFO | train rank=0 | step=41070/50000 progress=82.1% loss=0.075518 lr=8.470205e-06 lr_backbone=8.470205e-06 elapsed=43:06:26 avg_step=3.78s recent_step=2.85s eta=09:22:22 +2026-09-16 09:37:29 | INFO | train rank=0 | step=41080/50000 progress=82.2% loss=0.127798 lr=8.451799e-06 lr_backbone=8.451799e-06 elapsed=43:07:07 avg_step=3.78s recent_step=4.06s eta=09:21:45 +2026-09-16 09:38:09 | INFO | train rank=0 | step=41090/50000 progress=82.2% loss=0.075606 lr=8.433410e-06 lr_backbone=8.433410e-06 elapsed=43:07:47 avg_step=3.78s recent_step=4.04s eta=09:21:08 +2026-09-16 09:38:57 | INFO | train rank=0 | step=41100/50000 progress=82.2% loss=0.085608 lr=8.415040e-06 lr_backbone=8.415040e-06 elapsed=43:08:35 avg_step=3.78s recent_step=4.76s eta=09:20:32 +2026-09-16 09:39:29 | INFO | train rank=0 | step=41110/50000 progress=82.2% loss=0.093736 lr=8.396688e-06 lr_backbone=8.396688e-06 elapsed=43:09:07 avg_step=3.78s recent_step=3.20s eta=09:19:53 +2026-09-16 09:40:06 | INFO | train rank=0 | step=41120/50000 progress=82.2% loss=0.072009 lr=8.378355e-06 lr_backbone=8.378355e-06 elapsed=43:09:44 avg_step=3.78s recent_step=3.71s eta=09:19:15 +2026-09-16 09:40:35 | INFO | train rank=0 | step=41130/50000 progress=82.3% loss=0.078671 lr=8.360039e-06 lr_backbone=8.360039e-06 elapsed=43:10:13 avg_step=3.78s recent_step=2.96s eta=09:18:36 +2026-09-16 09:41:21 | INFO | train rank=0 | step=41140/50000 progress=82.3% loss=0.098114 lr=8.341742e-06 lr_backbone=8.341742e-06 elapsed=43:10:59 avg_step=3.78s recent_step=4.59s eta=09:18:00 +2026-09-16 09:41:51 | INFO | train rank=0 | step=41150/50000 progress=82.3% loss=0.072247 lr=8.323463e-06 lr_backbone=8.323463e-06 elapsed=43:11:29 avg_step=3.78s recent_step=2.93s eta=09:17:20 +2026-09-16 09:42:30 | INFO | train rank=0 | step=41160/50000 progress=82.3% loss=0.089486 lr=8.305202e-06 lr_backbone=8.305202e-06 elapsed=43:12:08 avg_step=3.78s recent_step=3.90s eta=09:16:43 +2026-09-16 09:43:07 | INFO | train rank=0 | step=41170/50000 progress=82.3% loss=0.089291 lr=8.286959e-06 lr_backbone=8.286959e-06 elapsed=43:12:45 avg_step=3.78s recent_step=3.73s eta=09:16:05 +2026-09-16 09:43:49 | INFO | train rank=0 | step=41180/50000 progress=82.4% loss=0.073164 lr=8.268735e-06 lr_backbone=8.268735e-06 elapsed=43:13:27 avg_step=3.78s recent_step=4.25s eta=09:15:28 +2026-09-16 09:44:21 | INFO | train rank=0 | step=41190/50000 progress=82.4% loss=0.128682 lr=8.250529e-06 lr_backbone=8.250529e-06 elapsed=43:13:59 avg_step=3.78s recent_step=3.14s eta=09:14:49 +2026-09-16 09:44:53 | INFO | train rank=0 | step=41200/50000 progress=82.4% loss=0.085805 lr=8.232341e-06 lr_backbone=8.232341e-06 elapsed=43:14:31 avg_step=3.78s recent_step=3.18s eta=09:14:10 +2026-09-16 09:45:30 | INFO | train rank=0 | step=41210/50000 progress=82.4% loss=0.081822 lr=8.214171e-06 lr_backbone=8.214171e-06 elapsed=43:15:08 avg_step=3.78s recent_step=3.70s eta=09:13:32 +2026-09-16 09:46:18 | INFO | train rank=0 | step=41220/50000 progress=82.4% loss=0.103662 lr=8.196020e-06 lr_backbone=8.196020e-06 elapsed=43:15:56 avg_step=3.78s recent_step=4.85s eta=09:12:56 +2026-09-16 09:46:51 | INFO | train rank=0 | step=41230/50000 progress=82.5% loss=0.077089 lr=8.177887e-06 lr_backbone=8.177887e-06 elapsed=43:16:29 avg_step=3.78s recent_step=3.25s eta=09:12:17 +2026-09-16 09:47:27 | INFO | train rank=0 | step=41240/50000 progress=82.5% loss=0.130335 lr=8.159772e-06 lr_backbone=8.159772e-06 elapsed=43:17:05 avg_step=3.78s recent_step=3.58s eta=09:11:39 +2026-09-16 09:47:55 | INFO | train rank=0 | step=41250/50000 progress=82.5% loss=0.087516 lr=8.141676e-06 lr_backbone=8.141676e-06 elapsed=43:17:33 avg_step=3.78s recent_step=2.80s eta=09:10:59 +2026-09-16 09:48:45 | INFO | train rank=0 | step=41260/50000 progress=82.5% loss=0.092720 lr=8.123598e-06 lr_backbone=8.123598e-06 elapsed=43:18:23 avg_step=3.78s recent_step=5.03s eta=09:10:24 +2026-09-16 09:49:16 | INFO | train rank=0 | step=41270/50000 progress=82.5% loss=0.094862 lr=8.105538e-06 lr_backbone=8.105538e-06 elapsed=43:18:54 avg_step=3.78s recent_step=3.14s eta=09:09:45 +2026-09-16 09:49:47 | INFO | train rank=0 | step=41280/50000 progress=82.6% loss=0.080754 lr=8.087497e-06 lr_backbone=8.087497e-06 elapsed=43:19:25 avg_step=3.78s recent_step=3.07s eta=09:09:06 +2026-09-16 09:50:28 | INFO | train rank=0 | step=41290/50000 progress=82.6% loss=0.064018 lr=8.069474e-06 lr_backbone=8.069474e-06 elapsed=43:20:06 avg_step=3.78s recent_step=4.11s eta=09:08:29 +2026-09-16 09:51:27 | INFO | train rank=0 | step=41300/50000 progress=82.6% loss=0.084018 lr=8.051469e-06 lr_backbone=8.051469e-06 elapsed=43:21:05 avg_step=3.78s recent_step=5.88s eta=09:07:55 +2026-09-16 09:51:57 | INFO | train rank=0 | step=41310/50000 progress=82.6% loss=0.082642 lr=8.033482e-06 lr_backbone=8.033482e-06 elapsed=43:21:35 avg_step=3.78s recent_step=3.03s eta=09:07:16 +2026-09-16 09:52:32 | INFO | train rank=0 | step=41320/50000 progress=82.6% loss=0.097923 lr=8.015514e-06 lr_backbone=8.015514e-06 elapsed=43:22:10 avg_step=3.78s recent_step=3.48s eta=09:06:37 +2026-09-16 09:53:00 | INFO | train rank=0 | step=41330/50000 progress=82.7% loss=0.086704 lr=7.997565e-06 lr_backbone=7.997565e-06 elapsed=43:22:38 avg_step=3.78s recent_step=2.79s eta=09:05:58 +2026-09-16 09:53:47 | INFO | train rank=0 | step=41340/50000 progress=82.7% loss=0.087344 lr=7.979633e-06 lr_backbone=7.979633e-06 elapsed=43:23:25 avg_step=3.78s recent_step=4.70s eta=09:05:22 +2026-09-16 09:54:23 | INFO | train rank=0 | step=41350/50000 progress=82.7% loss=0.106759 lr=7.961720e-06 lr_backbone=7.961720e-06 elapsed=43:24:01 avg_step=3.78s recent_step=3.58s eta=09:04:44 +2026-09-16 09:54:53 | INFO | train rank=0 | step=41360/50000 progress=82.7% loss=0.097157 lr=7.943826e-06 lr_backbone=7.943826e-06 elapsed=43:24:31 avg_step=3.78s recent_step=3.02s eta=09:04:04 +2026-09-16 09:55:33 | INFO | train rank=0 | step=41370/50000 progress=82.7% loss=0.096675 lr=7.925950e-06 lr_backbone=7.925950e-06 elapsed=43:25:11 avg_step=3.78s recent_step=4.03s eta=09:03:27 +2026-09-16 09:56:30 | INFO | train rank=0 | step=41380/50000 progress=82.8% loss=0.088057 lr=7.908092e-06 lr_backbone=7.908092e-06 elapsed=43:26:08 avg_step=3.78s recent_step=5.71s eta=09:02:53 +2026-09-16 09:56:58 | INFO | train rank=0 | step=41390/50000 progress=82.8% loss=0.094121 lr=7.890253e-06 lr_backbone=7.890253e-06 elapsed=43:26:36 avg_step=3.78s recent_step=2.74s eta=09:02:13 +2026-09-16 09:57:32 | INFO | train rank=0 | step=41400/50000 progress=82.8% loss=0.097974 lr=7.872432e-06 lr_backbone=7.872432e-06 elapsed=43:27:10 avg_step=3.78s recent_step=3.46s eta=09:01:35 +2026-09-16 09:58:13 | INFO | train rank=0 | step=41410/50000 progress=82.8% loss=0.099186 lr=7.854629e-06 lr_backbone=7.854629e-06 elapsed=43:27:51 avg_step=3.78s recent_step=4.10s eta=09:00:58 +2026-09-16 09:58:54 | INFO | train rank=0 | step=41420/50000 progress=82.8% loss=0.084010 lr=7.836845e-06 lr_backbone=7.836845e-06 elapsed=43:28:32 avg_step=3.78s recent_step=4.11s eta=09:00:21 +2026-09-16 09:59:31 | INFO | train rank=0 | step=41430/50000 progress=82.9% loss=0.085784 lr=7.819080e-06 lr_backbone=7.819080e-06 elapsed=43:29:09 avg_step=3.78s recent_step=3.65s eta=08:59:43 +2026-09-16 09:59:58 | INFO | train rank=0 | step=41440/50000 progress=82.9% loss=0.081282 lr=7.801332e-06 lr_backbone=7.801332e-06 elapsed=43:29:36 avg_step=3.78s recent_step=2.67s eta=08:59:02 +2026-09-16 10:00:32 | INFO | train rank=0 | step=41450/50000 progress=82.9% loss=0.069315 lr=7.783604e-06 lr_backbone=7.783604e-06 elapsed=43:30:10 avg_step=3.78s recent_step=3.46s eta=08:58:24 +2026-09-16 10:01:20 | INFO | train rank=0 | step=41460/50000 progress=82.9% loss=0.092998 lr=7.765893e-06 lr_backbone=7.765893e-06 elapsed=43:30:58 avg_step=3.78s recent_step=4.75s eta=08:57:48 +2026-09-16 10:01:52 | INFO | train rank=0 | step=41470/50000 progress=82.9% loss=0.085642 lr=7.748202e-06 lr_backbone=7.748202e-06 elapsed=43:31:30 avg_step=3.78s recent_step=3.27s eta=08:57:09 +2026-09-16 10:02:33 | INFO | train rank=0 | step=41480/50000 progress=83.0% loss=0.083884 lr=7.730528e-06 lr_backbone=7.730528e-06 elapsed=43:32:11 avg_step=3.78s recent_step=4.07s eta=08:56:32 +2026-09-16 10:03:17 | INFO | train rank=0 | step=41490/50000 progress=83.0% loss=0.093383 lr=7.712874e-06 lr_backbone=7.712874e-06 elapsed=43:32:55 avg_step=3.78s recent_step=4.39s eta=08:55:56 +2026-09-16 10:03:47 | INFO | train rank=0 | step=41500/50000 progress=83.0% loss=0.076653 lr=7.695237e-06 lr_backbone=7.695237e-06 elapsed=43:33:25 avg_step=3.78s recent_step=3.01s eta=08:55:16 +2026-09-16 10:04:26 | INFO | train rank=0 | step=41510/50000 progress=83.0% loss=0.081752 lr=7.677620e-06 lr_backbone=7.677620e-06 elapsed=43:34:04 avg_step=3.78s recent_step=3.93s eta=08:54:39 +2026-09-16 10:05:02 | INFO | train rank=0 | step=41520/50000 progress=83.0% loss=0.078497 lr=7.660020e-06 lr_backbone=7.660020e-06 elapsed=43:34:40 avg_step=3.78s recent_step=3.62s eta=08:54:01 +2026-09-16 10:05:34 | INFO | train rank=0 | step=41530/50000 progress=83.1% loss=0.086334 lr=7.642440e-06 lr_backbone=7.642440e-06 elapsed=43:35:12 avg_step=3.78s recent_step=3.14s eta=08:53:22 +2026-09-16 10:06:22 | INFO | train rank=0 | step=41540/50000 progress=83.1% loss=0.094059 lr=7.624877e-06 lr_backbone=7.624877e-06 elapsed=43:36:00 avg_step=3.78s recent_step=4.86s eta=08:52:46 +2026-09-16 10:06:47 | INFO | train rank=0 | step=41550/50000 progress=83.1% loss=0.107203 lr=7.607334e-06 lr_backbone=7.607334e-06 elapsed=43:36:25 avg_step=3.78s recent_step=2.42s eta=08:52:05 +2026-09-16 10:07:24 | INFO | train rank=0 | step=41560/50000 progress=83.1% loss=0.067494 lr=7.589809e-06 lr_backbone=7.589809e-06 elapsed=43:37:02 avg_step=3.78s recent_step=3.74s eta=08:51:28 +2026-09-16 10:08:00 | INFO | train rank=0 | step=41570/50000 progress=83.1% loss=0.103077 lr=7.572302e-06 lr_backbone=7.572302e-06 elapsed=43:37:38 avg_step=3.78s recent_step=3.58s eta=08:50:49 +2026-09-16 10:08:51 | INFO | train rank=0 | step=41580/50000 progress=83.2% loss=0.084267 lr=7.554814e-06 lr_backbone=7.554814e-06 elapsed=43:38:29 avg_step=3.78s recent_step=5.10s eta=08:50:14 +2026-09-16 10:09:25 | INFO | train rank=0 | step=41590/50000 progress=83.2% loss=0.094598 lr=7.537344e-06 lr_backbone=7.537344e-06 elapsed=43:39:03 avg_step=3.78s recent_step=3.43s eta=08:49:36 +2026-09-16 10:09:56 | INFO | train rank=0 | step=41600/50000 progress=83.2% loss=0.076139 lr=7.519894e-06 lr_backbone=7.519894e-06 elapsed=43:39:34 avg_step=3.78s recent_step=3.03s eta=08:48:57 +2026-09-16 10:10:36 | INFO | train rank=0 | step=41610/50000 progress=83.2% loss=0.076442 lr=7.502461e-06 lr_backbone=7.502461e-06 elapsed=43:40:14 avg_step=3.78s recent_step=4.05s eta=08:48:19 +2026-09-16 10:11:12 | INFO | train rank=0 | step=41620/50000 progress=83.2% loss=0.083786 lr=7.485048e-06 lr_backbone=7.485048e-06 elapsed=43:40:50 avg_step=3.78s recent_step=3.64s eta=08:47:41 +2026-09-16 10:11:49 | INFO | train rank=0 | step=41630/50000 progress=83.3% loss=0.079439 lr=7.467652e-06 lr_backbone=7.467652e-06 elapsed=43:41:27 avg_step=3.78s recent_step=3.66s eta=08:47:03 +2026-09-16 10:12:20 | INFO | train rank=0 | step=41640/50000 progress=83.3% loss=0.080096 lr=7.450276e-06 lr_backbone=7.450276e-06 elapsed=43:41:58 avg_step=3.78s recent_step=3.10s eta=08:46:24 +2026-09-16 10:13:04 | INFO | train rank=0 | step=41650/50000 progress=83.3% loss=0.090527 lr=7.432918e-06 lr_backbone=7.432918e-06 elapsed=43:42:42 avg_step=3.78s recent_step=4.38s eta=08:45:48 +2026-09-16 10:13:45 | INFO | train rank=0 | step=41660/50000 progress=83.3% loss=0.077695 lr=7.415579e-06 lr_backbone=7.415579e-06 elapsed=43:43:23 avg_step=3.78s recent_step=4.14s eta=08:45:10 +2026-09-16 10:14:19 | INFO | train rank=0 | step=41670/50000 progress=83.3% loss=0.107606 lr=7.398258e-06 lr_backbone=7.398258e-06 elapsed=43:43:57 avg_step=3.78s recent_step=3.34s eta=08:44:32 +2026-09-16 10:14:54 | INFO | train rank=0 | step=41680/50000 progress=83.4% loss=0.083550 lr=7.380956e-06 lr_backbone=7.380956e-06 elapsed=43:44:32 avg_step=3.78s recent_step=3.58s eta=08:43:54 +2026-09-16 10:15:40 | INFO | train rank=0 | step=41690/50000 progress=83.4% loss=0.098072 lr=7.363673e-06 lr_backbone=7.363673e-06 elapsed=43:45:18 avg_step=3.78s recent_step=4.53s eta=08:43:17 +2026-09-16 10:16:10 | INFO | train rank=0 | step=41700/50000 progress=83.4% loss=0.091800 lr=7.346408e-06 lr_backbone=7.346408e-06 elapsed=43:45:48 avg_step=3.78s recent_step=3.03s eta=08:42:38 +2026-09-16 10:16:45 | INFO | train rank=0 | step=41710/50000 progress=83.4% loss=0.081706 lr=7.329162e-06 lr_backbone=7.329162e-06 elapsed=43:46:23 avg_step=3.78s recent_step=3.45s eta=08:42:00 +2026-09-16 10:17:22 | INFO | train rank=0 | step=41720/50000 progress=83.4% loss=0.067865 lr=7.311934e-06 lr_backbone=7.311934e-06 elapsed=43:47:00 avg_step=3.78s recent_step=3.74s eta=08:41:22 +2026-09-16 10:18:10 | INFO | train rank=0 | step=41730/50000 progress=83.5% loss=0.160891 lr=7.294726e-06 lr_backbone=7.294726e-06 elapsed=43:47:48 avg_step=3.78s recent_step=4.79s eta=08:40:46 +2026-09-16 10:18:47 | INFO | train rank=0 | step=41740/50000 progress=83.5% loss=0.093540 lr=7.277536e-06 lr_backbone=7.277536e-06 elapsed=43:48:25 avg_step=3.78s recent_step=3.71s eta=08:40:08 +2026-09-16 10:19:22 | INFO | train rank=0 | step=41750/50000 progress=83.5% loss=0.121064 lr=7.260364e-06 lr_backbone=7.260364e-06 elapsed=43:49:00 avg_step=3.78s recent_step=3.52s eta=08:39:30 +2026-09-16 10:19:53 | INFO | train rank=0 | step=41760/50000 progress=83.5% loss=0.072861 lr=7.243212e-06 lr_backbone=7.243212e-06 elapsed=43:49:32 avg_step=3.78s recent_step=3.14s eta=08:38:51 +2026-09-16 10:20:29 | INFO | train rank=0 | step=41770/50000 progress=83.5% loss=0.108321 lr=7.226078e-06 lr_backbone=7.226078e-06 elapsed=43:50:07 avg_step=3.78s recent_step=3.59s eta=08:38:13 +2026-09-16 10:21:30 | INFO | train rank=0 | step=41780/50000 progress=83.6% loss=0.083049 lr=7.208963e-06 lr_backbone=7.208963e-06 elapsed=43:51:08 avg_step=3.78s recent_step=6.10s eta=08:37:39 +2026-09-16 10:21:57 | INFO | train rank=0 | step=41790/50000 progress=83.6% loss=0.079015 lr=7.191866e-06 lr_backbone=7.191866e-06 elapsed=43:51:35 avg_step=3.78s recent_step=2.64s eta=08:36:59 +2026-09-16 10:22:34 | INFO | train rank=0 | step=41800/50000 progress=83.6% loss=0.099433 lr=7.174788e-06 lr_backbone=7.174788e-06 elapsed=43:52:12 avg_step=3.78s recent_step=3.72s eta=08:36:21 +2026-09-16 10:23:07 | INFO | train rank=0 | step=41810/50000 progress=83.6% loss=0.103120 lr=7.157729e-06 lr_backbone=7.157729e-06 elapsed=43:52:45 avg_step=3.78s recent_step=3.32s eta=08:35:43 +2026-09-16 10:23:55 | INFO | train rank=0 | step=41820/50000 progress=83.6% loss=0.105273 lr=7.140689e-06 lr_backbone=7.140689e-06 elapsed=43:53:33 avg_step=3.78s recent_step=4.77s eta=08:35:07 +2026-09-16 10:24:35 | INFO | train rank=0 | step=41830/50000 progress=83.7% loss=0.083315 lr=7.123667e-06 lr_backbone=7.123667e-06 elapsed=43:54:13 avg_step=3.78s recent_step=4.04s eta=08:34:30 +2026-09-16 10:25:10 | INFO | train rank=0 | step=41840/50000 progress=83.7% loss=0.073813 lr=7.106664e-06 lr_backbone=7.106664e-06 elapsed=43:54:48 avg_step=3.78s recent_step=3.45s eta=08:33:51 +2026-09-16 10:25:44 | INFO | train rank=0 | step=41850/50000 progress=83.7% loss=0.091409 lr=7.089680e-06 lr_backbone=7.089680e-06 elapsed=43:55:22 avg_step=3.78s recent_step=3.46s eta=08:33:13 +2026-09-16 10:26:37 | INFO | train rank=0 | step=41860/50000 progress=83.7% loss=0.070504 lr=7.072715e-06 lr_backbone=7.072715e-06 elapsed=43:56:15 avg_step=3.78s recent_step=5.22s eta=08:32:38 +2026-09-16 10:27:11 | INFO | train rank=0 | step=41870/50000 progress=83.7% loss=0.076675 lr=7.055768e-06 lr_backbone=7.055768e-06 elapsed=43:56:49 avg_step=3.78s recent_step=3.46s eta=08:31:59 +2026-09-16 10:27:45 | INFO | train rank=0 | step=41880/50000 progress=83.8% loss=0.103434 lr=7.038841e-06 lr_backbone=7.038841e-06 elapsed=43:57:23 avg_step=3.78s recent_step=3.37s eta=08:31:21 +2026-09-16 10:28:23 | INFO | train rank=0 | step=41890/50000 progress=83.8% loss=0.090524 lr=7.021932e-06 lr_backbone=7.021932e-06 elapsed=43:58:01 avg_step=3.78s recent_step=3.79s eta=08:30:43 +2026-09-16 10:29:05 | INFO | train rank=0 | step=41900/50000 progress=83.8% loss=0.113290 lr=7.005042e-06 lr_backbone=7.005042e-06 elapsed=43:58:43 avg_step=3.78s recent_step=4.24s eta=08:30:06 +2026-09-16 10:29:45 | INFO | train rank=0 | step=41910/50000 progress=83.8% loss=0.071708 lr=6.988170e-06 lr_backbone=6.988170e-06 elapsed=43:59:23 avg_step=3.78s recent_step=3.96s eta=08:29:29 +2026-09-16 10:30:18 | INFO | train rank=0 | step=41920/50000 progress=83.8% loss=0.076391 lr=6.971318e-06 lr_backbone=6.971318e-06 elapsed=43:59:56 avg_step=3.78s recent_step=3.30s eta=08:28:50 +2026-09-16 10:30:52 | INFO | train rank=0 | step=41930/50000 progress=83.9% loss=0.088873 lr=6.954484e-06 lr_backbone=6.954484e-06 elapsed=44:00:30 avg_step=3.78s recent_step=3.45s eta=08:28:12 +2026-09-16 10:31:35 | INFO | train rank=0 | step=41940/50000 progress=83.9% loss=0.072543 lr=6.937669e-06 lr_backbone=6.937669e-06 elapsed=44:01:13 avg_step=3.78s recent_step=4.26s eta=08:27:35 +2026-09-16 10:32:05 | INFO | train rank=0 | step=41950/50000 progress=83.9% loss=0.086737 lr=6.920873e-06 lr_backbone=6.920873e-06 elapsed=44:01:43 avg_step=3.78s recent_step=3.01s eta=08:26:56 +2026-09-16 10:32:35 | INFO | train rank=0 | step=41960/50000 progress=83.9% loss=0.100955 lr=6.904096e-06 lr_backbone=6.904096e-06 elapsed=44:02:13 avg_step=3.78s recent_step=3.02s eta=08:26:16 +2026-09-16 10:33:07 | INFO | train rank=0 | step=41970/50000 progress=83.9% loss=0.095119 lr=6.887338e-06 lr_backbone=6.887338e-06 elapsed=44:02:45 avg_step=3.78s recent_step=3.24s eta=08:25:37 +2026-09-16 10:33:51 | INFO | train rank=0 | step=41980/50000 progress=84.0% loss=0.078648 lr=6.870598e-06 lr_backbone=6.870598e-06 elapsed=44:03:29 avg_step=3.78s recent_step=4.37s eta=08:25:01 +2026-09-16 10:34:28 | INFO | train rank=0 | step=41990/50000 progress=84.0% loss=0.068006 lr=6.853878e-06 lr_backbone=6.853878e-06 elapsed=44:04:06 avg_step=3.78s recent_step=3.65s eta=08:24:23 +2026-09-16 10:35:01 | INFO | train rank=0 | step=42000/50000 progress=84.0% loss=0.094829 lr=6.837176e-06 lr_backbone=6.837176e-06 elapsed=44:04:39 avg_step=3.78s recent_step=3.34s eta=08:23:44 +2026-09-16 10:35:34 | INFO | train rank=0 | step=42010/50000 progress=84.0% loss=0.081231 lr=6.820493e-06 lr_backbone=6.820493e-06 elapsed=44:05:12 avg_step=3.78s recent_step=3.32s eta=08:23:06 +2026-09-16 10:36:19 | INFO | train rank=0 | step=42020/50000 progress=84.0% loss=0.087987 lr=6.803829e-06 lr_backbone=6.803829e-06 elapsed=44:05:57 avg_step=3.78s recent_step=4.44s eta=08:22:29 +2026-09-16 10:36:53 | INFO | train rank=0 | step=42030/50000 progress=84.1% loss=0.081592 lr=6.787184e-06 lr_backbone=6.787184e-06 elapsed=44:06:31 avg_step=3.78s recent_step=3.49s eta=08:21:51 +2026-09-16 10:37:34 | INFO | train rank=0 | step=42040/50000 progress=84.1% loss=0.100742 lr=6.770558e-06 lr_backbone=6.770558e-06 elapsed=44:07:12 avg_step=3.78s recent_step=4.06s eta=08:21:13 +2026-09-16 10:38:11 | INFO | train rank=0 | step=42050/50000 progress=84.1% loss=0.063049 lr=6.753951e-06 lr_backbone=6.753951e-06 elapsed=44:07:49 avg_step=3.78s recent_step=3.71s eta=08:20:35 +2026-09-16 10:39:06 | INFO | train rank=0 | step=42060/50000 progress=84.1% loss=0.092050 lr=6.737362e-06 lr_backbone=6.737362e-06 elapsed=44:08:44 avg_step=3.78s recent_step=5.46s eta=08:20:01 +2026-09-16 10:39:38 | INFO | train rank=0 | step=42070/50000 progress=84.1% loss=0.073995 lr=6.720793e-06 lr_backbone=6.720793e-06 elapsed=44:09:16 avg_step=3.78s recent_step=3.18s eta=08:19:22 +2026-09-16 10:40:24 | INFO | train rank=0 | step=42080/50000 progress=84.2% loss=0.084900 lr=6.704242e-06 lr_backbone=6.704242e-06 elapsed=44:10:02 avg_step=3.78s recent_step=4.64s eta=08:18:46 +2026-09-16 10:40:53 | INFO | train rank=0 | step=42090/50000 progress=84.2% loss=0.073415 lr=6.687711e-06 lr_backbone=6.687711e-06 elapsed=44:10:31 avg_step=3.78s recent_step=2.93s eta=08:18:06 +2026-09-16 10:41:47 | INFO | train rank=0 | step=42100/50000 progress=84.2% loss=0.100707 lr=6.671198e-06 lr_backbone=6.671198e-06 elapsed=44:11:25 avg_step=3.78s recent_step=5.41s eta=08:17:32 +2026-09-16 10:42:19 | INFO | train rank=0 | step=42110/50000 progress=84.2% loss=0.080005 lr=6.654705e-06 lr_backbone=6.654705e-06 elapsed=44:11:57 avg_step=3.78s recent_step=3.11s eta=08:16:53 +2026-09-16 10:42:47 | INFO | train rank=0 | step=42120/50000 progress=84.2% loss=0.080422 lr=6.638230e-06 lr_backbone=6.638230e-06 elapsed=44:12:25 avg_step=3.78s recent_step=2.88s eta=08:16:13 +2026-09-16 10:43:22 | INFO | train rank=0 | step=42130/50000 progress=84.3% loss=0.080302 lr=6.621774e-06 lr_backbone=6.621774e-06 elapsed=44:13:00 avg_step=3.78s recent_step=3.42s eta=08:15:35 +2026-09-16 10:44:16 | INFO | train rank=0 | step=42140/50000 progress=84.3% loss=0.083545 lr=6.605338e-06 lr_backbone=6.605338e-06 elapsed=44:13:54 avg_step=3.78s recent_step=5.43s eta=08:15:00 +2026-09-16 10:44:44 | INFO | train rank=0 | step=42150/50000 progress=84.3% loss=0.069589 lr=6.588920e-06 lr_backbone=6.588920e-06 elapsed=44:14:22 avg_step=3.78s recent_step=2.86s eta=08:14:21 +2026-09-16 10:45:12 | INFO | train rank=0 | step=42160/50000 progress=84.3% loss=0.088530 lr=6.572521e-06 lr_backbone=6.572521e-06 elapsed=44:14:50 avg_step=3.78s recent_step=2.79s eta=08:13:41 +2026-09-16 10:45:39 | INFO | train rank=0 | step=42170/50000 progress=84.3% loss=0.088276 lr=6.556141e-06 lr_backbone=6.556141e-06 elapsed=44:15:17 avg_step=3.78s recent_step=2.67s eta=08:13:01 +2026-09-16 10:46:30 | INFO | train rank=0 | step=42180/50000 progress=84.4% loss=0.068185 lr=6.539780e-06 lr_backbone=6.539780e-06 elapsed=44:16:08 avg_step=3.78s recent_step=5.12s eta=08:12:26 +2026-09-16 10:47:10 | INFO | train rank=0 | step=42190/50000 progress=84.4% loss=0.086186 lr=6.523439e-06 lr_backbone=6.523439e-06 elapsed=44:16:48 avg_step=3.78s recent_step=3.93s eta=08:11:48 +2026-09-16 10:47:43 | INFO | train rank=0 | step=42200/50000 progress=84.4% loss=0.070848 lr=6.507116e-06 lr_backbone=6.507116e-06 elapsed=44:17:21 avg_step=3.78s recent_step=3.30s eta=08:11:10 +2026-09-16 10:48:20 | INFO | train rank=0 | step=42210/50000 progress=84.4% loss=0.079555 lr=6.490812e-06 lr_backbone=6.490812e-06 elapsed=44:17:58 avg_step=3.78s recent_step=3.78s eta=08:10:32 +2026-09-16 10:49:13 | INFO | train rank=0 | step=42220/50000 progress=84.4% loss=0.100420 lr=6.474528e-06 lr_backbone=6.474528e-06 elapsed=44:18:52 avg_step=3.78s recent_step=5.32s eta=08:09:57 +2026-09-16 10:49:46 | INFO | train rank=0 | step=42230/50000 progress=84.5% loss=0.075521 lr=6.458262e-06 lr_backbone=6.458262e-06 elapsed=44:19:24 avg_step=3.78s recent_step=3.24s eta=08:09:18 +2026-09-16 10:50:24 | INFO | train rank=0 | step=42240/50000 progress=84.5% loss=0.080681 lr=6.442015e-06 lr_backbone=6.442015e-06 elapsed=44:20:02 avg_step=3.78s recent_step=3.85s eta=08:08:40 +2026-09-16 10:50:55 | INFO | train rank=0 | step=42250/50000 progress=84.5% loss=0.103316 lr=6.425788e-06 lr_backbone=6.425788e-06 elapsed=44:20:33 avg_step=3.78s recent_step=3.03s eta=08:08:01 +2026-09-16 10:51:45 | INFO | train rank=0 | step=42260/50000 progress=84.5% loss=0.088783 lr=6.409579e-06 lr_backbone=6.409579e-06 elapsed=44:21:23 avg_step=3.78s recent_step=5.01s eta=08:07:26 +2026-09-16 10:52:19 | INFO | train rank=0 | step=42270/50000 progress=84.5% loss=0.076334 lr=6.393390e-06 lr_backbone=6.393390e-06 elapsed=44:21:57 avg_step=3.78s recent_step=3.44s eta=08:06:47 +2026-09-16 10:52:46 | INFO | train rank=0 | step=42280/50000 progress=84.6% loss=0.071624 lr=6.377220e-06 lr_backbone=6.377220e-06 elapsed=44:22:24 avg_step=3.78s recent_step=2.72s eta=08:06:08 +2026-09-16 10:53:24 | INFO | train rank=0 | step=42290/50000 progress=84.6% loss=0.096102 lr=6.361068e-06 lr_backbone=6.361068e-06 elapsed=44:23:03 avg_step=3.78s recent_step=3.83s eta=08:05:30 +2026-09-16 10:54:21 | INFO | train rank=0 | step=42300/50000 progress=84.6% loss=0.095963 lr=6.344936e-06 lr_backbone=6.344936e-06 elapsed=44:23:59 avg_step=3.78s recent_step=5.65s eta=08:04:56 +2026-09-16 10:54:51 | INFO | train rank=0 | step=42310/50000 progress=84.6% loss=0.092994 lr=6.328823e-06 lr_backbone=6.328823e-06 elapsed=44:24:29 avg_step=3.78s recent_step=2.98s eta=08:04:16 +2026-09-16 10:55:22 | INFO | train rank=0 | step=42320/50000 progress=84.6% loss=0.082748 lr=6.312729e-06 lr_backbone=6.312729e-06 elapsed=44:25:00 avg_step=3.78s recent_step=3.17s eta=08:03:37 +2026-09-16 10:55:58 | INFO | train rank=0 | step=42330/50000 progress=84.7% loss=0.089186 lr=6.296654e-06 lr_backbone=6.296654e-06 elapsed=44:25:36 avg_step=3.78s recent_step=3.53s eta=08:02:59 +2026-09-16 10:56:45 | INFO | train rank=0 | step=42340/50000 progress=84.7% loss=0.079335 lr=6.280599e-06 lr_backbone=6.280599e-06 elapsed=44:26:23 avg_step=3.78s recent_step=4.69s eta=08:02:23 +2026-09-16 10:57:18 | INFO | train rank=0 | step=42350/50000 progress=84.7% loss=0.068425 lr=6.264562e-06 lr_backbone=6.264562e-06 elapsed=44:26:56 avg_step=3.78s recent_step=3.30s eta=08:01:44 +2026-09-16 10:57:51 | INFO | train rank=0 | step=42360/50000 progress=84.7% loss=0.083831 lr=6.248544e-06 lr_backbone=6.248544e-06 elapsed=44:27:29 avg_step=3.78s recent_step=3.36s eta=08:01:06 +2026-09-16 10:58:32 | INFO | train rank=0 | step=42370/50000 progress=84.7% loss=0.077864 lr=6.232546e-06 lr_backbone=6.232546e-06 elapsed=44:28:10 avg_step=3.78s recent_step=4.08s eta=08:00:29 +2026-09-16 10:59:22 | INFO | train rank=0 | step=42380/50000 progress=84.8% loss=0.081994 lr=6.216567e-06 lr_backbone=6.216567e-06 elapsed=44:29:00 avg_step=3.78s recent_step=4.97s eta=07:59:53 +2026-09-16 10:59:53 | INFO | train rank=0 | step=42390/50000 progress=84.8% loss=0.089106 lr=6.200607e-06 lr_backbone=6.200607e-06 elapsed=44:29:31 avg_step=3.78s recent_step=3.12s eta=07:59:14 +2026-09-16 11:00:20 | INFO | train rank=0 | step=42400/50000 progress=84.8% loss=0.091598 lr=6.184666e-06 lr_backbone=6.184666e-06 elapsed=44:29:58 avg_step=3.78s recent_step=2.73s eta=07:58:34 +2026-09-16 11:00:51 | INFO | train rank=0 | step=42410/50000 progress=84.8% loss=0.091744 lr=6.168744e-06 lr_backbone=6.168744e-06 elapsed=44:30:29 avg_step=3.78s recent_step=3.11s eta=07:57:55 +2026-09-16 11:01:48 | INFO | train rank=0 | step=42420/50000 progress=84.8% loss=0.090064 lr=6.152842e-06 lr_backbone=6.152842e-06 elapsed=44:31:26 avg_step=3.78s recent_step=5.68s eta=07:57:21 +2026-09-16 11:02:20 | INFO | train rank=0 | step=42430/50000 progress=84.9% loss=0.094500 lr=6.136958e-06 lr_backbone=6.136958e-06 elapsed=44:31:59 avg_step=3.78s recent_step=3.24s eta=07:56:42 +2026-09-16 11:02:54 | INFO | train rank=0 | step=42440/50000 progress=84.9% loss=0.084493 lr=6.121094e-06 lr_backbone=6.121094e-06 elapsed=44:32:32 avg_step=3.78s recent_step=3.34s eta=07:56:04 +2026-09-16 11:03:29 | INFO | train rank=0 | step=42450/50000 progress=84.9% loss=0.080608 lr=6.105249e-06 lr_backbone=6.105249e-06 elapsed=44:33:07 avg_step=3.78s recent_step=3.55s eta=07:55:25 +2026-09-16 11:04:15 | INFO | train rank=0 | step=42460/50000 progress=84.9% loss=0.090115 lr=6.089423e-06 lr_backbone=6.089423e-06 elapsed=44:33:53 avg_step=3.78s recent_step=4.53s eta=07:54:49 +2026-09-16 11:04:54 | INFO | train rank=0 | step=42470/50000 progress=84.9% loss=0.084660 lr=6.073617e-06 lr_backbone=6.073617e-06 elapsed=44:34:32 avg_step=3.78s recent_step=3.96s eta=07:54:12 +2026-09-16 11:05:30 | INFO | train rank=0 | step=42480/50000 progress=85.0% loss=0.100961 lr=6.057829e-06 lr_backbone=6.057829e-06 elapsed=44:35:08 avg_step=3.78s recent_step=3.55s eta=07:53:33 +2026-09-16 11:05:59 | INFO | train rank=0 | step=42490/50000 progress=85.0% loss=0.099580 lr=6.042061e-06 lr_backbone=6.042061e-06 elapsed=44:35:37 avg_step=3.78s recent_step=2.92s eta=07:52:54 +2026-09-16 11:06:35 | INFO | train rank=0 | step=42500/50000 progress=85.0% loss=0.075425 lr=6.026312e-06 lr_backbone=6.026312e-06 elapsed=44:36:13 avg_step=3.78s recent_step=3.61s eta=07:52:16 +2026-09-16 11:07:15 | INFO | train rank=0 | step=42510/50000 progress=85.0% loss=0.092500 lr=6.010583e-06 lr_backbone=6.010583e-06 elapsed=44:36:53 avg_step=3.78s recent_step=4.02s eta=07:51:39 +2026-09-16 11:07:52 | INFO | train rank=0 | step=42520/50000 progress=85.0% loss=0.096334 lr=5.994872e-06 lr_backbone=5.994872e-06 elapsed=44:37:30 avg_step=3.78s recent_step=3.70s eta=07:51:01 +2026-09-16 11:08:16 | INFO | train rank=0 | step=42530/50000 progress=85.1% loss=0.106834 lr=5.979181e-06 lr_backbone=5.979181e-06 elapsed=44:37:54 avg_step=3.78s recent_step=2.39s eta=07:50:21 +2026-09-16 11:09:13 | INFO | train rank=0 | step=42540/50000 progress=85.1% loss=0.098787 lr=5.963509e-06 lr_backbone=5.963509e-06 elapsed=44:38:51 avg_step=3.78s recent_step=5.66s eta=07:49:46 +2026-09-16 11:09:44 | INFO | train rank=0 | step=42550/50000 progress=85.1% loss=0.081091 lr=5.947857e-06 lr_backbone=5.947857e-06 elapsed=44:39:23 avg_step=3.78s recent_step=3.17s eta=07:49:07 +2026-09-16 11:10:15 | INFO | train rank=0 | step=42560/50000 progress=85.1% loss=0.100821 lr=5.932223e-06 lr_backbone=5.932223e-06 elapsed=44:39:53 avg_step=3.78s recent_step=3.08s eta=07:48:28 +2026-09-16 11:10:47 | INFO | train rank=0 | step=42570/50000 progress=85.1% loss=0.080922 lr=5.916609e-06 lr_backbone=5.916609e-06 elapsed=44:40:25 avg_step=3.78s recent_step=3.19s eta=07:47:49 +2026-09-16 11:11:44 | INFO | train rank=0 | step=42580/50000 progress=85.2% loss=0.068805 lr=5.901014e-06 lr_backbone=5.901014e-06 elapsed=44:41:22 avg_step=3.78s recent_step=5.67s eta=07:47:15 +2026-09-16 11:12:16 | INFO | train rank=0 | step=42590/50000 progress=85.2% loss=0.071887 lr=5.885439e-06 lr_backbone=5.885439e-06 elapsed=44:41:54 avg_step=3.78s recent_step=3.19s eta=07:46:36 +2026-09-16 11:12:55 | INFO | train rank=0 | step=42600/50000 progress=85.2% loss=0.089255 lr=5.869882e-06 lr_backbone=5.869882e-06 elapsed=44:42:33 avg_step=3.78s recent_step=3.93s eta=07:45:59 +2026-09-16 11:13:26 | INFO | train rank=0 | step=42610/50000 progress=85.2% loss=0.096672 lr=5.854345e-06 lr_backbone=5.854345e-06 elapsed=44:43:04 avg_step=3.78s recent_step=3.11s eta=07:45:20 +2026-09-16 11:14:06 | INFO | train rank=0 | step=42620/50000 progress=85.2% loss=0.084351 lr=5.838828e-06 lr_backbone=5.838828e-06 elapsed=44:43:44 avg_step=3.78s recent_step=4.01s eta=07:44:42 +2026-09-16 11:14:39 | INFO | train rank=0 | step=42630/50000 progress=85.3% loss=0.069353 lr=5.823330e-06 lr_backbone=5.823330e-06 elapsed=44:44:17 avg_step=3.78s recent_step=3.25s eta=07:44:04 +2026-09-16 11:15:14 | INFO | train rank=0 | step=42640/50000 progress=85.3% loss=0.095804 lr=5.807851e-06 lr_backbone=5.807851e-06 elapsed=44:44:52 avg_step=3.78s recent_step=3.56s eta=07:43:25 +2026-09-16 11:15:50 | INFO | train rank=0 | step=42650/50000 progress=85.3% loss=0.096815 lr=5.792391e-06 lr_backbone=5.792391e-06 elapsed=44:45:28 avg_step=3.78s recent_step=3.59s eta=07:42:47 +2026-09-16 11:16:39 | INFO | train rank=0 | step=42660/50000 progress=85.3% loss=0.070538 lr=5.776951e-06 lr_backbone=5.776951e-06 elapsed=44:46:17 avg_step=3.78s recent_step=4.88s eta=07:42:11 +2026-09-16 11:17:11 | INFO | train rank=0 | step=42670/50000 progress=85.3% loss=0.073333 lr=5.761530e-06 lr_backbone=5.761530e-06 elapsed=44:46:49 avg_step=3.78s recent_step=3.17s eta=07:41:33 +2026-09-16 11:17:48 | INFO | train rank=0 | step=42680/50000 progress=85.4% loss=0.084448 lr=5.746128e-06 lr_backbone=5.746128e-06 elapsed=44:47:26 avg_step=3.78s recent_step=3.68s eta=07:40:55 +2026-09-16 11:18:21 | INFO | train rank=0 | step=42690/50000 progress=85.4% loss=0.099811 lr=5.730746e-06 lr_backbone=5.730746e-06 elapsed=44:47:59 avg_step=3.78s recent_step=3.28s eta=07:40:16 +2026-09-16 11:19:09 | INFO | train rank=0 | step=42700/50000 progress=85.4% loss=0.094426 lr=5.715383e-06 lr_backbone=5.715383e-06 elapsed=44:48:47 avg_step=3.78s recent_step=4.81s eta=07:39:40 +2026-09-16 11:19:44 | INFO | train rank=0 | step=42710/50000 progress=85.4% loss=0.113540 lr=5.700039e-06 lr_backbone=5.700039e-06 elapsed=44:49:22 avg_step=3.78s recent_step=3.57s eta=07:39:02 +2026-09-16 11:20:18 | INFO | train rank=0 | step=42720/50000 progress=85.4% loss=0.131200 lr=5.684715e-06 lr_backbone=5.684715e-06 elapsed=44:49:56 avg_step=3.78s recent_step=3.40s eta=07:38:23 +2026-09-16 11:20:47 | INFO | train rank=0 | step=42730/50000 progress=85.5% loss=0.086137 lr=5.669410e-06 lr_backbone=5.669410e-06 elapsed=44:50:25 avg_step=3.78s recent_step=2.90s eta=07:37:44 +2026-09-16 11:21:31 | INFO | train rank=0 | step=42740/50000 progress=85.5% loss=0.069791 lr=5.654125e-06 lr_backbone=5.654125e-06 elapsed=44:51:09 avg_step=3.78s recent_step=4.37s eta=07:37:07 +2026-09-16 11:22:03 | INFO | train rank=0 | step=42750/50000 progress=85.5% loss=0.077128 lr=5.638859e-06 lr_backbone=5.638859e-06 elapsed=44:51:41 avg_step=3.78s recent_step=3.16s eta=07:36:29 +2026-09-16 11:22:30 | INFO | train rank=0 | step=42760/50000 progress=85.5% loss=0.092103 lr=5.623612e-06 lr_backbone=5.623612e-06 elapsed=44:52:08 avg_step=3.78s recent_step=2.75s eta=07:35:49 +2026-09-16 11:23:01 | INFO | train rank=0 | step=42770/50000 progress=85.5% loss=0.067197 lr=5.608385e-06 lr_backbone=5.608385e-06 elapsed=44:52:39 avg_step=3.78s recent_step=3.13s eta=07:35:10 +2026-09-16 11:23:54 | INFO | train rank=0 | step=42780/50000 progress=85.6% loss=0.094860 lr=5.593178e-06 lr_backbone=5.593178e-06 elapsed=44:53:32 avg_step=3.78s recent_step=5.29s eta=07:34:35 +2026-09-16 11:24:23 | INFO | train rank=0 | step=42790/50000 progress=85.6% loss=0.074693 lr=5.577989e-06 lr_backbone=5.577989e-06 elapsed=44:54:01 avg_step=3.78s recent_step=2.81s eta=07:33:56 +2026-09-16 11:24:56 | INFO | train rank=0 | step=42800/50000 progress=85.6% loss=0.082447 lr=5.562820e-06 lr_backbone=5.562820e-06 elapsed=44:54:34 avg_step=3.78s recent_step=3.34s eta=07:33:17 +2026-09-16 11:25:31 | INFO | train rank=0 | step=42810/50000 progress=85.6% loss=0.076550 lr=5.547671e-06 lr_backbone=5.547671e-06 elapsed=44:55:09 avg_step=3.78s recent_step=3.49s eta=07:32:39 +2026-09-16 11:26:26 | INFO | train rank=0 | step=42820/50000 progress=85.6% loss=0.082282 lr=5.532541e-06 lr_backbone=5.532541e-06 elapsed=44:56:04 avg_step=3.78s recent_step=5.49s eta=07:32:04 +2026-09-16 11:26:58 | INFO | train rank=0 | step=42830/50000 progress=85.7% loss=0.090838 lr=5.517430e-06 lr_backbone=5.517430e-06 elapsed=44:56:36 avg_step=3.78s recent_step=3.23s eta=07:31:25 +2026-09-16 11:27:31 | INFO | train rank=0 | step=42840/50000 progress=85.7% loss=0.081673 lr=5.502339e-06 lr_backbone=5.502339e-06 elapsed=44:57:09 avg_step=3.78s recent_step=3.29s eta=07:30:47 +2026-09-16 11:28:04 | INFO | train rank=0 | step=42850/50000 progress=85.7% loss=0.082911 lr=5.487268e-06 lr_backbone=5.487268e-06 elapsed=44:57:42 avg_step=3.78s recent_step=3.30s eta=07:30:08 +2026-09-16 11:28:57 | INFO | train rank=0 | step=42860/50000 progress=85.7% loss=0.085314 lr=5.472215e-06 lr_backbone=5.472215e-06 elapsed=44:58:35 avg_step=3.78s recent_step=5.27s eta=07:29:33 +2026-09-16 11:29:31 | INFO | train rank=0 | step=42870/50000 progress=85.7% loss=0.058877 lr=5.457183e-06 lr_backbone=5.457183e-06 elapsed=44:59:09 avg_step=3.78s recent_step=3.46s eta=07:28:54 +2026-09-16 11:30:02 | INFO | train rank=0 | step=42880/50000 progress=85.8% loss=0.091512 lr=5.442170e-06 lr_backbone=5.442170e-06 elapsed=44:59:40 avg_step=3.78s recent_step=3.11s eta=07:28:16 +2026-09-16 11:30:31 | INFO | train rank=0 | step=42890/50000 progress=85.8% loss=0.074414 lr=5.427176e-06 lr_backbone=5.427176e-06 elapsed=45:00:09 avg_step=3.78s recent_step=2.90s eta=07:27:36 +2026-09-16 11:31:18 | INFO | train rank=0 | step=42900/50000 progress=85.8% loss=0.062165 lr=5.412202e-06 lr_backbone=5.412202e-06 elapsed=45:00:56 avg_step=3.78s recent_step=4.66s eta=07:27:00 +2026-09-16 11:31:53 | INFO | train rank=0 | step=42910/50000 progress=85.8% loss=0.071866 lr=5.397247e-06 lr_backbone=5.397247e-06 elapsed=45:01:31 avg_step=3.78s recent_step=3.55s eta=07:26:22 +2026-09-16 11:32:30 | INFO | train rank=0 | step=42920/50000 progress=85.8% loss=0.095421 lr=5.382312e-06 lr_backbone=5.382312e-06 elapsed=45:02:08 avg_step=3.78s recent_step=3.71s eta=07:25:44 +2026-09-16 11:33:11 | INFO | train rank=0 | step=42930/50000 progress=85.9% loss=0.092442 lr=5.367396e-06 lr_backbone=5.367396e-06 elapsed=45:02:49 avg_step=3.78s recent_step=4.04s eta=07:25:07 +2026-09-16 11:34:00 | INFO | train rank=0 | step=42940/50000 progress=85.9% loss=0.090868 lr=5.352500e-06 lr_backbone=5.352500e-06 elapsed=45:03:38 avg_step=3.78s recent_step=4.97s eta=07:24:31 +2026-09-16 11:34:39 | INFO | train rank=0 | step=42950/50000 progress=85.9% loss=0.109682 lr=5.337623e-06 lr_backbone=5.337623e-06 elapsed=45:04:17 avg_step=3.78s recent_step=3.88s eta=07:23:53 +2026-09-16 11:35:10 | INFO | train rank=0 | step=42960/50000 progress=85.9% loss=0.085087 lr=5.322766e-06 lr_backbone=5.322766e-06 elapsed=45:04:48 avg_step=3.78s recent_step=3.09s eta=07:23:14 +2026-09-16 11:35:45 | INFO | train rank=0 | step=42970/50000 progress=85.9% loss=0.093766 lr=5.307929e-06 lr_backbone=5.307929e-06 elapsed=45:05:23 avg_step=3.78s recent_step=3.49s eta=07:22:36 +2026-09-16 11:36:23 | INFO | train rank=0 | step=42980/50000 progress=86.0% loss=0.088714 lr=5.293111e-06 lr_backbone=5.293111e-06 elapsed=45:06:01 avg_step=3.78s recent_step=3.84s eta=07:21:58 +2026-09-16 11:36:50 | INFO | train rank=0 | step=42990/50000 progress=86.0% loss=0.087907 lr=5.278312e-06 lr_backbone=5.278312e-06 elapsed=45:06:28 avg_step=3.78s recent_step=2.69s eta=07:21:19 +2026-09-16 11:37:26 | INFO | train rank=0 | step=43000/50000 progress=86.0% loss=0.130944 lr=5.263534e-06 lr_backbone=5.263534e-06 elapsed=45:07:04 avg_step=3.78s recent_step=3.55s eta=07:20:41 +2026-09-16 11:38:03 | INFO | train rank=0 | step=43010/50000 progress=86.0% loss=0.077306 lr=5.248774e-06 lr_backbone=5.248774e-06 elapsed=45:07:41 avg_step=3.78s recent_step=3.68s eta=07:20:03 +2026-09-16 11:38:48 | INFO | train rank=0 | step=43020/50000 progress=86.0% loss=0.119546 lr=5.234034e-06 lr_backbone=5.234034e-06 elapsed=45:08:26 avg_step=3.78s recent_step=4.57s eta=07:19:26 +2026-09-16 11:39:23 | INFO | train rank=0 | step=43030/50000 progress=86.1% loss=0.077341 lr=5.219314e-06 lr_backbone=5.219314e-06 elapsed=45:09:01 avg_step=3.78s recent_step=3.46s eta=07:18:48 +2026-09-16 11:40:00 | INFO | train rank=0 | step=43040/50000 progress=86.1% loss=0.098266 lr=5.204614e-06 lr_backbone=5.204614e-06 elapsed=45:09:38 avg_step=3.78s recent_step=3.67s eta=07:18:10 +2026-09-16 11:40:36 | INFO | train rank=0 | step=43050/50000 progress=86.1% loss=0.080409 lr=5.189933e-06 lr_backbone=5.189933e-06 elapsed=45:10:14 avg_step=3.78s recent_step=3.63s eta=07:17:32 +2026-09-16 11:41:28 | INFO | train rank=0 | step=43060/50000 progress=86.1% loss=0.082745 lr=5.175271e-06 lr_backbone=5.175271e-06 elapsed=45:11:06 avg_step=3.78s recent_step=5.20s eta=07:16:57 +2026-09-16 11:42:04 | INFO | train rank=0 | step=43070/50000 progress=86.1% loss=0.104085 lr=5.160630e-06 lr_backbone=5.160630e-06 elapsed=45:11:42 avg_step=3.78s recent_step=3.61s eta=07:16:18 +2026-09-16 11:42:41 | INFO | train rank=0 | step=43080/50000 progress=86.2% loss=0.110530 lr=5.146008e-06 lr_backbone=5.146008e-06 elapsed=45:12:19 avg_step=3.78s recent_step=3.72s eta=07:15:41 +2026-09-16 11:43:14 | INFO | train rank=0 | step=43090/50000 progress=86.2% loss=0.084014 lr=5.131405e-06 lr_backbone=5.131405e-06 elapsed=45:12:52 avg_step=3.78s recent_step=3.30s eta=07:15:02 +2026-09-16 11:44:08 | INFO | train rank=0 | step=43100/50000 progress=86.2% loss=0.093836 lr=5.116822e-06 lr_backbone=5.116822e-06 elapsed=45:13:47 avg_step=3.78s recent_step=5.41s eta=07:14:27 +2026-09-16 11:44:46 | INFO | train rank=0 | step=43110/50000 progress=86.2% loss=0.090356 lr=5.102259e-06 lr_backbone=5.102259e-06 elapsed=45:14:24 avg_step=3.78s recent_step=3.75s eta=07:13:49 +2026-09-16 11:45:19 | INFO | train rank=0 | step=43120/50000 progress=86.2% loss=0.066817 lr=5.087715e-06 lr_backbone=5.087715e-06 elapsed=45:14:57 avg_step=3.78s recent_step=3.35s eta=07:13:11 +2026-09-16 11:45:55 | INFO | train rank=0 | step=43130/50000 progress=86.3% loss=0.064447 lr=5.073191e-06 lr_backbone=5.073191e-06 elapsed=45:15:33 avg_step=3.78s recent_step=3.56s eta=07:12:33 +2026-09-16 11:46:56 | INFO | train rank=0 | step=43140/50000 progress=86.3% loss=0.099195 lr=5.058687e-06 lr_backbone=5.058687e-06 elapsed=45:16:34 avg_step=3.78s recent_step=6.13s eta=07:11:58 +2026-09-16 11:47:32 | INFO | train rank=0 | step=43150/50000 progress=86.3% loss=0.089128 lr=5.044202e-06 lr_backbone=5.044202e-06 elapsed=45:17:10 avg_step=3.78s recent_step=3.54s eta=07:11:20 +2026-09-16 11:48:01 | INFO | train rank=0 | step=43160/50000 progress=86.3% loss=0.071020 lr=5.029737e-06 lr_backbone=5.029737e-06 elapsed=45:17:39 avg_step=3.78s recent_step=2.91s eta=07:10:41 +2026-09-16 11:48:39 | INFO | train rank=0 | step=43170/50000 progress=86.3% loss=0.095673 lr=5.015292e-06 lr_backbone=5.015292e-06 elapsed=45:18:17 avg_step=3.78s recent_step=3.84s eta=07:10:03 +2026-09-16 11:49:11 | INFO | train rank=0 | step=43180/50000 progress=86.4% loss=0.099851 lr=5.000866e-06 lr_backbone=5.000866e-06 elapsed=45:18:49 avg_step=3.78s recent_step=3.15s eta=07:09:25 +2026-09-16 11:49:37 | INFO | train rank=0 | step=43190/50000 progress=86.4% loss=0.085779 lr=4.986461e-06 lr_backbone=4.986461e-06 elapsed=45:19:15 avg_step=3.78s recent_step=2.64s eta=07:08:45 +2026-09-16 11:50:14 | INFO | train rank=0 | step=43200/50000 progress=86.4% loss=0.116973 lr=4.972074e-06 lr_backbone=4.972074e-06 elapsed=45:19:52 avg_step=3.78s recent_step=3.65s eta=07:08:07 +2026-09-16 11:50:58 | INFO | train rank=0 | step=43210/50000 progress=86.4% loss=0.077124 lr=4.957708e-06 lr_backbone=4.957708e-06 elapsed=45:20:36 avg_step=3.78s recent_step=4.48s eta=07:07:30 +2026-09-16 11:51:38 | INFO | train rank=0 | step=43220/50000 progress=86.4% loss=0.111542 lr=4.943361e-06 lr_backbone=4.943361e-06 elapsed=45:21:16 avg_step=3.78s recent_step=3.96s eta=07:06:53 +2026-09-16 11:52:07 | INFO | train rank=0 | step=43230/50000 progress=86.5% loss=0.087808 lr=4.929034e-06 lr_backbone=4.929034e-06 elapsed=45:21:45 avg_step=3.78s recent_step=2.94s eta=07:06:14 +2026-09-16 11:52:41 | INFO | train rank=0 | step=43240/50000 progress=86.5% loss=0.096874 lr=4.914726e-06 lr_backbone=4.914726e-06 elapsed=45:22:19 avg_step=3.78s recent_step=3.40s eta=07:05:36 +2026-09-16 11:53:41 | INFO | train rank=0 | step=43250/50000 progress=86.5% loss=0.060454 lr=4.900438e-06 lr_backbone=4.900438e-06 elapsed=45:23:19 avg_step=3.78s recent_step=5.91s eta=07:05:01 +2026-09-16 11:54:13 | INFO | train rank=0 | step=43260/50000 progress=86.5% loss=0.088290 lr=4.886171e-06 lr_backbone=4.886171e-06 elapsed=45:23:51 avg_step=3.78s recent_step=3.21s eta=07:04:22 +2026-09-16 11:54:54 | INFO | train rank=0 | step=43270/50000 progress=86.5% loss=0.081149 lr=4.871922e-06 lr_backbone=4.871922e-06 elapsed=45:24:32 avg_step=3.78s recent_step=4.15s eta=07:03:45 +2026-09-16 11:55:28 | INFO | train rank=0 | step=43280/50000 progress=86.6% loss=0.094016 lr=4.857694e-06 lr_backbone=4.857694e-06 elapsed=45:25:06 avg_step=3.78s recent_step=3.34s eta=07:03:07 +2026-09-16 11:56:01 | INFO | train rank=0 | step=43290/50000 progress=86.6% loss=0.078636 lr=4.843485e-06 lr_backbone=4.843485e-06 elapsed=45:25:39 avg_step=3.78s recent_step=3.33s eta=07:02:28 +2026-09-16 11:56:46 | INFO | train rank=0 | step=43300/50000 progress=86.6% loss=0.075594 lr=4.829296e-06 lr_backbone=4.829296e-06 elapsed=45:26:24 avg_step=3.78s recent_step=4.50s eta=07:01:52 +2026-09-16 11:57:22 | INFO | train rank=0 | step=43310/50000 progress=86.6% loss=0.082844 lr=4.815127e-06 lr_backbone=4.815127e-06 elapsed=45:27:00 avg_step=3.78s recent_step=3.65s eta=07:01:14 +2026-09-16 11:57:57 | INFO | train rank=0 | step=43320/50000 progress=86.6% loss=0.075831 lr=4.800977e-06 lr_backbone=4.800977e-06 elapsed=45:27:35 avg_step=3.78s recent_step=3.50s eta=07:00:35 +2026-09-16 11:58:28 | INFO | train rank=0 | step=43330/50000 progress=86.7% loss=0.086097 lr=4.786847e-06 lr_backbone=4.786847e-06 elapsed=45:28:06 avg_step=3.78s recent_step=3.01s eta=06:59:56 +2026-09-16 11:59:21 | INFO | train rank=0 | step=43340/50000 progress=86.7% loss=0.073025 lr=4.772738e-06 lr_backbone=4.772738e-06 elapsed=45:28:59 avg_step=3.78s recent_step=5.36s eta=06:59:21 +2026-09-16 11:59:51 | INFO | train rank=0 | step=43350/50000 progress=86.7% loss=0.090907 lr=4.758647e-06 lr_backbone=4.758647e-06 elapsed=45:29:29 avg_step=3.78s recent_step=2.94s eta=06:58:42 +2026-09-16 12:00:24 | INFO | train rank=0 | step=43360/50000 progress=86.7% loss=0.090732 lr=4.744577e-06 lr_backbone=4.744577e-06 elapsed=45:30:02 avg_step=3.78s recent_step=3.31s eta=06:58:04 +2026-09-16 12:01:03 | INFO | train rank=0 | step=43370/50000 progress=86.7% loss=0.127298 lr=4.730526e-06 lr_backbone=4.730526e-06 elapsed=45:30:41 avg_step=3.78s recent_step=3.91s eta=06:57:26 +2026-09-16 12:01:54 | INFO | train rank=0 | step=43380/50000 progress=86.8% loss=0.077755 lr=4.716496e-06 lr_backbone=4.716496e-06 elapsed=45:31:32 avg_step=3.78s recent_step=5.09s eta=06:56:50 +2026-09-16 12:02:26 | INFO | train rank=0 | step=43390/50000 progress=86.8% loss=0.100469 lr=4.702485e-06 lr_backbone=4.702485e-06 elapsed=45:32:04 avg_step=3.78s recent_step=3.20s eta=06:56:12 +2026-09-16 12:03:06 | INFO | train rank=0 | step=43400/50000 progress=86.8% loss=0.079141 lr=4.688494e-06 lr_backbone=4.688494e-06 elapsed=45:32:44 avg_step=3.78s recent_step=4.03s eta=06:55:34 +2026-09-16 12:03:36 | INFO | train rank=0 | step=43410/50000 progress=86.8% loss=0.086473 lr=4.674522e-06 lr_backbone=4.674522e-06 elapsed=45:33:14 avg_step=3.78s recent_step=2.98s eta=06:54:55 +2026-09-16 12:04:15 | INFO | train rank=0 | step=43420/50000 progress=86.8% loss=0.110659 lr=4.660571e-06 lr_backbone=4.660571e-06 elapsed=45:33:53 avg_step=3.78s recent_step=3.96s eta=06:54:18 +2026-09-16 12:04:45 | INFO | train rank=0 | step=43430/50000 progress=86.9% loss=0.085329 lr=4.646639e-06 lr_backbone=4.646639e-06 elapsed=45:34:23 avg_step=3.78s recent_step=2.97s eta=06:53:39 +2026-09-16 12:05:14 | INFO | train rank=0 | step=43440/50000 progress=86.9% loss=0.074923 lr=4.632727e-06 lr_backbone=4.632727e-06 elapsed=45:34:52 avg_step=3.78s recent_step=2.90s eta=06:53:00 +2026-09-16 12:05:53 | INFO | train rank=0 | step=43450/50000 progress=86.9% loss=0.089953 lr=4.618835e-06 lr_backbone=4.618835e-06 elapsed=45:35:31 avg_step=3.78s recent_step=3.95s eta=06:52:22 +2026-09-16 12:06:40 | INFO | train rank=0 | step=43460/50000 progress=86.9% loss=0.085735 lr=4.604963e-06 lr_backbone=4.604963e-06 elapsed=45:36:18 avg_step=3.78s recent_step=4.68s eta=06:51:46 +2026-09-16 12:07:09 | INFO | train rank=0 | step=43470/50000 progress=86.9% loss=0.085940 lr=4.591111e-06 lr_backbone=4.591111e-06 elapsed=45:36:47 avg_step=3.78s recent_step=2.90s eta=06:51:07 +2026-09-16 12:07:35 | INFO | train rank=0 | step=43480/50000 progress=87.0% loss=0.070705 lr=4.577278e-06 lr_backbone=4.577278e-06 elapsed=45:37:13 avg_step=3.78s recent_step=2.58s eta=06:50:27 +2026-09-16 12:08:28 | INFO | train rank=0 | step=43490/50000 progress=87.0% loss=0.112515 lr=4.563466e-06 lr_backbone=4.563466e-06 elapsed=45:38:06 avg_step=3.78s recent_step=5.30s eta=06:49:51 +2026-09-16 12:08:59 | INFO | train rank=0 | step=43500/50000 progress=87.0% loss=0.103397 lr=4.549673e-06 lr_backbone=4.549673e-06 elapsed=45:38:37 avg_step=3.78s recent_step=3.05s eta=06:49:13 +2026-09-16 12:09:36 | INFO | train rank=0 | step=43510/50000 progress=87.0% loss=0.094975 lr=4.535900e-06 lr_backbone=4.535900e-06 elapsed=45:39:14 avg_step=3.78s recent_step=3.74s eta=06:48:35 +2026-09-16 12:10:01 | INFO | train rank=0 | step=43520/50000 progress=87.0% loss=0.076808 lr=4.522148e-06 lr_backbone=4.522148e-06 elapsed=45:39:39 avg_step=3.78s recent_step=2.47s eta=06:47:55 +2026-09-16 12:10:42 | INFO | train rank=0 | step=43530/50000 progress=87.1% loss=0.094675 lr=4.508415e-06 lr_backbone=4.508415e-06 elapsed=45:40:20 avg_step=3.78s recent_step=4.13s eta=06:47:18 +2026-09-16 12:11:22 | INFO | train rank=0 | step=43540/50000 progress=87.1% loss=0.116416 lr=4.494701e-06 lr_backbone=4.494701e-06 elapsed=45:41:00 avg_step=3.78s recent_step=4.01s eta=06:46:40 +2026-09-16 12:11:57 | INFO | train rank=0 | step=43550/50000 progress=87.1% loss=0.080156 lr=4.481008e-06 lr_backbone=4.481008e-06 elapsed=45:41:35 avg_step=3.78s recent_step=3.53s eta=06:46:02 +2026-09-16 12:12:30 | INFO | train rank=0 | step=43560/50000 progress=87.1% loss=0.101079 lr=4.467335e-06 lr_backbone=4.467335e-06 elapsed=45:42:08 avg_step=3.78s recent_step=3.22s eta=06:45:24 +2026-09-16 12:13:27 | INFO | train rank=0 | step=43570/50000 progress=87.1% loss=0.081113 lr=4.453682e-06 lr_backbone=4.453682e-06 elapsed=45:43:05 avg_step=3.78s recent_step=5.71s eta=06:44:49 +2026-09-16 12:13:56 | INFO | train rank=0 | step=43580/50000 progress=87.2% loss=0.085728 lr=4.440048e-06 lr_backbone=4.440048e-06 elapsed=45:43:34 avg_step=3.78s recent_step=2.94s eta=06:44:10 +2026-09-16 12:14:28 | INFO | train rank=0 | step=43590/50000 progress=87.2% loss=0.083590 lr=4.426435e-06 lr_backbone=4.426435e-06 elapsed=45:44:06 avg_step=3.78s recent_step=3.15s eta=06:43:31 +2026-09-16 12:15:03 | INFO | train rank=0 | step=43600/50000 progress=87.2% loss=0.091192 lr=4.412841e-06 lr_backbone=4.412841e-06 elapsed=45:44:41 avg_step=3.78s recent_step=3.51s eta=06:42:53 +2026-09-16 12:16:05 | INFO | train rank=0 | step=43610/50000 progress=87.2% loss=0.078321 lr=4.399267e-06 lr_backbone=4.399267e-06 elapsed=45:45:43 avg_step=3.78s recent_step=6.26s eta=06:42:19 +2026-09-16 12:16:34 | INFO | train rank=0 | step=43620/50000 progress=87.2% loss=0.089141 lr=4.385714e-06 lr_backbone=4.385714e-06 elapsed=45:46:12 avg_step=3.78s recent_step=2.90s eta=06:41:40 +2026-09-16 12:17:09 | INFO | train rank=0 | step=43630/50000 progress=87.3% loss=0.120957 lr=4.372180e-06 lr_backbone=4.372180e-06 elapsed=45:46:47 avg_step=3.78s recent_step=3.45s eta=06:41:01 +2026-09-16 12:17:35 | INFO | train rank=0 | step=43640/50000 progress=87.3% loss=0.071403 lr=4.358666e-06 lr_backbone=4.358666e-06 elapsed=45:47:13 avg_step=3.78s recent_step=2.66s eta=06:40:22 +2026-09-16 12:18:25 | INFO | train rank=0 | step=43650/50000 progress=87.3% loss=0.080365 lr=4.345172e-06 lr_backbone=4.345172e-06 elapsed=45:48:03 avg_step=3.78s recent_step=5.00s eta=06:39:46 +2026-09-16 12:19:02 | INFO | train rank=0 | step=43660/50000 progress=87.3% loss=0.110120 lr=4.331699e-06 lr_backbone=4.331699e-06 elapsed=45:48:40 avg_step=3.78s recent_step=3.68s eta=06:39:08 +2026-09-16 12:19:39 | INFO | train rank=0 | step=43670/50000 progress=87.3% loss=0.081013 lr=4.318245e-06 lr_backbone=4.318245e-06 elapsed=45:49:17 avg_step=3.78s recent_step=3.64s eta=06:38:30 +2026-09-16 12:20:16 | INFO | train rank=0 | step=43680/50000 progress=87.4% loss=0.085395 lr=4.304811e-06 lr_backbone=4.304811e-06 elapsed=45:49:54 avg_step=3.78s recent_step=3.70s eta=06:37:52 +2026-09-16 12:21:05 | INFO | train rank=0 | step=43690/50000 progress=87.4% loss=0.093024 lr=4.291397e-06 lr_backbone=4.291397e-06 elapsed=45:50:43 avg_step=3.78s recent_step=4.93s eta=06:37:16 +2026-09-16 12:21:35 | INFO | train rank=0 | step=43700/50000 progress=87.4% loss=0.085985 lr=4.278003e-06 lr_backbone=4.278003e-06 elapsed=45:51:13 avg_step=3.78s recent_step=3.00s eta=06:36:37 +2026-09-16 12:22:12 | INFO | train rank=0 | step=43710/50000 progress=87.4% loss=0.103792 lr=4.264629e-06 lr_backbone=4.264629e-06 elapsed=45:51:50 avg_step=3.78s recent_step=3.72s eta=06:35:59 +2026-09-16 12:22:39 | INFO | train rank=0 | step=43720/50000 progress=87.4% loss=0.073936 lr=4.251275e-06 lr_backbone=4.251275e-06 elapsed=45:52:17 avg_step=3.78s recent_step=2.64s eta=06:35:20 +2026-09-16 12:23:32 | INFO | train rank=0 | step=43730/50000 progress=87.5% loss=0.093411 lr=4.237941e-06 lr_backbone=4.237941e-06 elapsed=45:53:10 avg_step=3.78s recent_step=5.38s eta=06:34:45 +2026-09-16 12:24:04 | INFO | train rank=0 | step=43740/50000 progress=87.5% loss=0.097786 lr=4.224628e-06 lr_backbone=4.224628e-06 elapsed=45:53:42 avg_step=3.78s recent_step=3.15s eta=06:34:06 +2026-09-16 12:24:45 | INFO | train rank=0 | step=43750/50000 progress=87.5% loss=0.067473 lr=4.211334e-06 lr_backbone=4.211334e-06 elapsed=45:54:23 avg_step=3.78s recent_step=4.08s eta=06:33:29 +2026-09-16 12:25:18 | INFO | train rank=0 | step=43760/50000 progress=87.5% loss=0.079111 lr=4.198060e-06 lr_backbone=4.198060e-06 elapsed=45:54:56 avg_step=3.78s recent_step=3.32s eta=06:32:50 +2026-09-16 12:26:02 | INFO | train rank=0 | step=43770/50000 progress=87.5% loss=0.095378 lr=4.184806e-06 lr_backbone=4.184806e-06 elapsed=45:55:40 avg_step=3.78s recent_step=4.42s eta=06:32:13 +2026-09-16 12:26:36 | INFO | train rank=0 | step=43780/50000 progress=87.6% loss=0.097254 lr=4.171572e-06 lr_backbone=4.171572e-06 elapsed=45:56:14 avg_step=3.78s recent_step=3.36s eta=06:31:35 +2026-09-16 12:27:04 | INFO | train rank=0 | step=43790/50000 progress=87.6% loss=0.086197 lr=4.158359e-06 lr_backbone=4.158359e-06 elapsed=45:56:42 avg_step=3.78s recent_step=2.81s eta=06:30:56 +2026-09-16 12:27:43 | INFO | train rank=0 | step=43800/50000 progress=87.6% loss=0.100554 lr=4.145165e-06 lr_backbone=4.145165e-06 elapsed=45:57:21 avg_step=3.78s recent_step=3.97s eta=06:30:18 +2026-09-16 12:28:28 | INFO | train rank=0 | step=43810/50000 progress=87.6% loss=0.099361 lr=4.131991e-06 lr_backbone=4.131991e-06 elapsed=45:58:06 avg_step=3.78s recent_step=4.41s eta=06:29:41 +2026-09-16 12:29:08 | INFO | train rank=0 | step=43820/50000 progress=87.6% loss=0.088992 lr=4.118838e-06 lr_backbone=4.118838e-06 elapsed=45:58:46 avg_step=3.78s recent_step=4.02s eta=06:29:04 +2026-09-16 12:29:38 | INFO | train rank=0 | step=43830/50000 progress=87.7% loss=0.077975 lr=4.105705e-06 lr_backbone=4.105705e-06 elapsed=45:59:16 avg_step=3.78s recent_step=2.99s eta=06:28:25 +2026-09-16 12:30:22 | INFO | train rank=0 | step=43840/50000 progress=87.7% loss=0.076984 lr=4.092591e-06 lr_backbone=4.092591e-06 elapsed=46:00:00 avg_step=3.78s recent_step=4.43s eta=06:27:48 +2026-09-16 12:31:06 | INFO | train rank=0 | step=43850/50000 progress=87.7% loss=0.089580 lr=4.079498e-06 lr_backbone=4.079498e-06 elapsed=46:00:44 avg_step=3.78s recent_step=4.40s eta=06:27:11 +2026-09-16 12:31:40 | INFO | train rank=0 | step=43860/50000 progress=87.7% loss=0.073737 lr=4.066425e-06 lr_backbone=4.066425e-06 elapsed=46:01:18 avg_step=3.78s recent_step=3.41s eta=06:26:33 +2026-09-16 12:32:12 | INFO | train rank=0 | step=43870/50000 progress=87.7% loss=0.091243 lr=4.053372e-06 lr_backbone=4.053372e-06 elapsed=46:01:50 avg_step=3.78s recent_step=3.20s eta=06:25:54 +2026-09-16 12:32:55 | INFO | train rank=0 | step=43880/50000 progress=87.8% loss=0.057378 lr=4.040339e-06 lr_backbone=4.040339e-06 elapsed=46:02:33 avg_step=3.78s recent_step=4.28s eta=06:25:17 +2026-09-16 12:33:35 | INFO | train rank=0 | step=43890/50000 progress=87.8% loss=0.079449 lr=4.027326e-06 lr_backbone=4.027326e-06 elapsed=46:03:13 avg_step=3.78s recent_step=3.98s eta=06:24:40 +2026-09-16 12:34:07 | INFO | train rank=0 | step=43900/50000 progress=87.8% loss=0.094241 lr=4.014333e-06 lr_backbone=4.014333e-06 elapsed=46:03:45 avg_step=3.78s recent_step=3.19s eta=06:24:01 +2026-09-16 12:34:37 | INFO | train rank=0 | step=43910/50000 progress=87.8% loss=0.090062 lr=4.001360e-06 lr_backbone=4.001360e-06 elapsed=46:04:15 avg_step=3.78s recent_step=3.07s eta=06:23:23 +2026-09-16 12:35:11 | INFO | train rank=0 | step=43920/50000 progress=87.8% loss=0.081137 lr=3.988408e-06 lr_backbone=3.988408e-06 elapsed=46:04:49 avg_step=3.78s recent_step=3.35s eta=06:22:44 +2026-09-16 12:36:04 | INFO | train rank=0 | step=43930/50000 progress=87.9% loss=0.095051 lr=3.975475e-06 lr_backbone=3.975475e-06 elapsed=46:05:42 avg_step=3.78s recent_step=5.36s eta=06:22:09 +2026-09-16 12:36:41 | INFO | train rank=0 | step=43940/50000 progress=87.9% loss=0.074029 lr=3.962563e-06 lr_backbone=3.962563e-06 elapsed=46:06:19 avg_step=3.78s recent_step=3.67s eta=06:21:31 +2026-09-16 12:37:11 | INFO | train rank=0 | step=43950/50000 progress=87.9% loss=0.072452 lr=3.949671e-06 lr_backbone=3.949671e-06 elapsed=46:06:49 avg_step=3.78s recent_step=2.97s eta=06:20:52 +2026-09-16 12:37:44 | INFO | train rank=0 | step=43960/50000 progress=87.9% loss=0.067196 lr=3.936799e-06 lr_backbone=3.936799e-06 elapsed=46:07:22 avg_step=3.78s recent_step=3.30s eta=06:20:13 +2026-09-16 12:38:42 | INFO | train rank=0 | step=43970/50000 progress=87.9% loss=0.115465 lr=3.923947e-06 lr_backbone=3.923947e-06 elapsed=46:08:20 avg_step=3.78s recent_step=5.78s eta=06:19:38 +2026-09-16 12:39:04 | INFO | train rank=0 | step=43980/50000 progress=88.0% loss=0.093933 lr=3.911115e-06 lr_backbone=3.911115e-06 elapsed=46:08:42 avg_step=3.78s recent_step=2.18s eta=06:18:58 +2026-09-16 12:39:34 | INFO | train rank=0 | step=43990/50000 progress=88.0% loss=0.090149 lr=3.898304e-06 lr_backbone=3.898304e-06 elapsed=46:09:12 avg_step=3.78s recent_step=3.10s eta=06:18:20 +2026-09-16 12:40:11 | INFO | train rank=0 | step=44000/50000 progress=88.0% loss=0.095544 lr=3.885512e-06 lr_backbone=3.885512e-06 elapsed=46:09:49 avg_step=3.78s recent_step=3.61s eta=06:17:42 +2026-09-16 12:41:05 | INFO | train rank=0 | step=44010/50000 progress=88.0% loss=0.075432 lr=3.872741e-06 lr_backbone=3.872741e-06 elapsed=46:10:43 avg_step=3.78s recent_step=5.41s eta=06:17:06 +2026-09-16 12:41:40 | INFO | train rank=0 | step=44020/50000 progress=88.0% loss=0.072469 lr=3.859990e-06 lr_backbone=3.859990e-06 elapsed=46:11:18 avg_step=3.78s recent_step=3.58s eta=06:16:28 +2026-09-16 12:42:18 | INFO | train rank=0 | step=44030/50000 progress=88.1% loss=0.099679 lr=3.847259e-06 lr_backbone=3.847259e-06 elapsed=46:11:56 avg_step=3.78s recent_step=3.76s eta=06:15:50 +2026-09-16 12:42:52 | INFO | train rank=0 | step=44040/50000 progress=88.1% loss=0.082677 lr=3.834549e-06 lr_backbone=3.834549e-06 elapsed=46:12:30 avg_step=3.78s recent_step=3.41s eta=06:15:12 +2026-09-16 12:43:41 | INFO | train rank=0 | step=44050/50000 progress=88.1% loss=0.081335 lr=3.821858e-06 lr_backbone=3.821858e-06 elapsed=46:13:19 avg_step=3.78s recent_step=4.88s eta=06:14:36 +2026-09-16 12:44:11 | INFO | train rank=0 | step=44060/50000 progress=88.1% loss=0.116505 lr=3.809188e-06 lr_backbone=3.809188e-06 elapsed=46:13:49 avg_step=3.78s recent_step=3.05s eta=06:13:57 +2026-09-16 12:44:45 | INFO | train rank=0 | step=44070/50000 progress=88.1% loss=0.097657 lr=3.796538e-06 lr_backbone=3.796538e-06 elapsed=46:14:23 avg_step=3.78s recent_step=3.39s eta=06:13:19 +2026-09-16 12:45:16 | INFO | train rank=0 | step=44080/50000 progress=88.2% loss=0.086456 lr=3.783908e-06 lr_backbone=3.783908e-06 elapsed=46:14:54 avg_step=3.78s recent_step=3.04s eta=06:12:40 +2026-09-16 12:46:07 | INFO | train rank=0 | step=44090/50000 progress=88.2% loss=0.075910 lr=3.771298e-06 lr_backbone=3.771298e-06 elapsed=46:15:45 avg_step=3.78s recent_step=5.13s eta=06:12:04 +2026-09-16 12:46:41 | INFO | train rank=0 | step=44100/50000 progress=88.2% loss=0.080744 lr=3.758709e-06 lr_backbone=3.758709e-06 elapsed=46:16:19 avg_step=3.78s recent_step=3.42s eta=06:11:26 +2026-09-16 12:47:12 | INFO | train rank=0 | step=44110/50000 progress=88.2% loss=0.112680 lr=3.746140e-06 lr_backbone=3.746140e-06 elapsed=46:16:50 avg_step=3.78s recent_step=3.08s eta=06:10:47 +2026-09-16 12:47:49 | INFO | train rank=0 | step=44120/50000 progress=88.2% loss=0.078737 lr=3.733591e-06 lr_backbone=3.733591e-06 elapsed=46:17:27 avg_step=3.78s recent_step=3.72s eta=06:10:09 +2026-09-16 12:48:38 | INFO | train rank=0 | step=44130/50000 progress=88.3% loss=0.101832 lr=3.721062e-06 lr_backbone=3.721062e-06 elapsed=46:18:16 avg_step=3.78s recent_step=4.87s eta=06:09:33 +2026-09-16 12:49:10 | INFO | train rank=0 | step=44140/50000 progress=88.3% loss=0.096668 lr=3.708554e-06 lr_backbone=3.708554e-06 elapsed=46:18:48 avg_step=3.78s recent_step=3.25s eta=06:08:54 +2026-09-16 12:49:39 | INFO | train rank=0 | step=44150/50000 progress=88.3% loss=0.078316 lr=3.696065e-06 lr_backbone=3.696065e-06 elapsed=46:19:17 avg_step=3.78s recent_step=2.88s eta=06:08:15 +2026-09-16 12:50:15 | INFO | train rank=0 | step=44160/50000 progress=88.3% loss=0.071424 lr=3.683597e-06 lr_backbone=3.683597e-06 elapsed=46:19:53 avg_step=3.78s recent_step=3.54s eta=06:07:37 +2026-09-16 12:51:03 | INFO | train rank=0 | step=44170/50000 progress=88.3% loss=0.092099 lr=3.671150e-06 lr_backbone=3.671150e-06 elapsed=46:20:41 avg_step=3.78s recent_step=4.80s eta=06:07:01 +2026-09-16 12:51:36 | INFO | train rank=0 | step=44180/50000 progress=88.4% loss=0.100341 lr=3.658722e-06 lr_backbone=3.658722e-06 elapsed=46:21:14 avg_step=3.78s recent_step=3.32s eta=06:06:22 +2026-09-16 12:52:07 | INFO | train rank=0 | step=44190/50000 progress=88.4% loss=0.079122 lr=3.646315e-06 lr_backbone=3.646315e-06 elapsed=46:21:45 avg_step=3.78s recent_step=3.14s eta=06:05:44 +2026-09-16 12:52:44 | INFO | train rank=0 | step=44200/50000 progress=88.4% loss=0.083210 lr=3.633928e-06 lr_backbone=3.633928e-06 elapsed=46:22:22 avg_step=3.78s recent_step=3.63s eta=06:05:06 +2026-09-16 12:53:36 | INFO | train rank=0 | step=44210/50000 progress=88.4% loss=0.085126 lr=3.621562e-06 lr_backbone=3.621562e-06 elapsed=46:23:14 avg_step=3.78s recent_step=5.20s eta=06:04:30 +2026-09-16 12:54:10 | INFO | train rank=0 | step=44220/50000 progress=88.4% loss=0.091880 lr=3.609215e-06 lr_backbone=3.609215e-06 elapsed=46:23:48 avg_step=3.78s recent_step=3.47s eta=06:03:52 +2026-09-16 12:54:47 | INFO | train rank=0 | step=44230/50000 progress=88.5% loss=0.105727 lr=3.596889e-06 lr_backbone=3.596889e-06 elapsed=46:24:25 avg_step=3.78s recent_step=3.68s eta=06:03:14 +2026-09-16 12:55:15 | INFO | train rank=0 | step=44240/50000 progress=88.5% loss=0.073864 lr=3.584583e-06 lr_backbone=3.584583e-06 elapsed=46:24:53 avg_step=3.78s recent_step=2.82s eta=06:02:35 +2026-09-16 12:56:01 | INFO | train rank=0 | step=44250/50000 progress=88.5% loss=0.106253 lr=3.572298e-06 lr_backbone=3.572298e-06 elapsed=46:25:39 avg_step=3.78s recent_step=4.54s eta=06:01:58 +2026-09-16 12:56:40 | INFO | train rank=0 | step=44260/50000 progress=88.5% loss=0.083114 lr=3.560033e-06 lr_backbone=3.560033e-06 elapsed=46:26:18 avg_step=3.78s recent_step=3.93s eta=06:01:21 +2026-09-16 12:57:14 | INFO | train rank=0 | step=44270/50000 progress=88.5% loss=0.098690 lr=3.547788e-06 lr_backbone=3.547788e-06 elapsed=46:26:52 avg_step=3.78s recent_step=3.37s eta=06:00:42 +2026-09-16 12:57:56 | INFO | train rank=0 | step=44280/50000 progress=88.6% loss=0.091958 lr=3.535564e-06 lr_backbone=3.535564e-06 elapsed=46:27:34 avg_step=3.78s recent_step=4.24s eta=06:00:05 +2026-09-16 12:58:52 | INFO | train rank=0 | step=44290/50000 progress=88.6% loss=0.092368 lr=3.523360e-06 lr_backbone=3.523360e-06 elapsed=46:28:30 avg_step=3.78s recent_step=5.58s eta=05:59:30 +2026-09-16 12:59:16 | INFO | train rank=0 | step=44300/50000 progress=88.6% loss=0.075436 lr=3.511176e-06 lr_backbone=3.511176e-06 elapsed=46:28:54 avg_step=3.78s recent_step=2.40s eta=05:58:50 +2026-09-16 12:59:42 | INFO | train rank=0 | step=44310/50000 progress=88.6% loss=0.089119 lr=3.499012e-06 lr_backbone=3.499012e-06 elapsed=46:29:20 avg_step=3.78s recent_step=2.65s eta=05:58:11 +2026-09-16 13:00:11 | INFO | train rank=0 | step=44320/50000 progress=88.6% loss=0.080321 lr=3.486869e-06 lr_backbone=3.486869e-06 elapsed=46:29:49 avg_step=3.78s recent_step=2.90s eta=05:57:32 +2026-09-16 13:00:52 | INFO | train rank=0 | step=44330/50000 progress=88.7% loss=0.075983 lr=3.474746e-06 lr_backbone=3.474746e-06 elapsed=46:30:30 avg_step=3.78s recent_step=4.11s eta=05:56:55 +2026-09-16 13:01:14 | INFO | train rank=0 | step=44340/50000 progress=88.7% loss=0.074156 lr=3.462644e-06 lr_backbone=3.462644e-06 elapsed=46:30:52 avg_step=3.78s recent_step=2.16s eta=05:56:15 +2026-09-16 13:01:43 | INFO | train rank=0 | step=44350/50000 progress=88.7% loss=0.062118 lr=3.450562e-06 lr_backbone=3.450562e-06 elapsed=46:31:21 avg_step=3.78s recent_step=2.87s eta=05:55:36 +2026-09-16 13:02:29 | INFO | train rank=0 | step=44360/50000 progress=88.7% loss=0.115097 lr=3.438500e-06 lr_backbone=3.438500e-06 elapsed=46:32:07 avg_step=3.78s recent_step=4.63s eta=05:54:59 +2026-09-16 13:03:12 | INFO | train rank=0 | step=44370/50000 progress=88.7% loss=0.083726 lr=3.426459e-06 lr_backbone=3.426459e-06 elapsed=46:32:50 avg_step=3.78s recent_step=4.26s eta=05:54:22 +2026-09-16 13:03:43 | INFO | train rank=0 | step=44380/50000 progress=88.8% loss=0.067292 lr=3.414438e-06 lr_backbone=3.414438e-06 elapsed=46:33:21 avg_step=3.78s recent_step=3.12s eta=05:53:43 +2026-09-16 13:04:18 | INFO | train rank=0 | step=44390/50000 progress=88.8% loss=0.088797 lr=3.402437e-06 lr_backbone=3.402437e-06 elapsed=46:33:56 avg_step=3.78s recent_step=3.55s eta=05:53:05 +2026-09-16 13:04:58 | INFO | train rank=0 | step=44400/50000 progress=88.8% loss=0.116381 lr=3.390457e-06 lr_backbone=3.390457e-06 elapsed=46:34:36 avg_step=3.78s recent_step=3.97s eta=05:52:28 +2026-09-16 13:05:33 | INFO | train rank=0 | step=44410/50000 progress=88.8% loss=0.082412 lr=3.378497e-06 lr_backbone=3.378497e-06 elapsed=46:35:11 avg_step=3.78s recent_step=3.54s eta=05:51:50 +2026-09-16 13:06:11 | INFO | train rank=0 | step=44420/50000 progress=88.8% loss=0.079839 lr=3.366558e-06 lr_backbone=3.366558e-06 elapsed=46:35:49 avg_step=3.78s recent_step=3.80s eta=05:51:12 +2026-09-16 13:06:49 | INFO | train rank=0 | step=44430/50000 progress=88.9% loss=0.084471 lr=3.354638e-06 lr_backbone=3.354638e-06 elapsed=46:36:27 avg_step=3.78s recent_step=3.78s eta=05:50:34 +2026-09-16 13:07:32 | INFO | train rank=0 | step=44440/50000 progress=88.9% loss=0.085511 lr=3.342740e-06 lr_backbone=3.342740e-06 elapsed=46:37:10 avg_step=3.78s recent_step=4.28s eta=05:49:57 +2026-09-16 13:08:21 | INFO | train rank=0 | step=44450/50000 progress=88.9% loss=0.102140 lr=3.330862e-06 lr_backbone=3.330862e-06 elapsed=46:37:59 avg_step=3.78s recent_step=4.94s eta=05:49:21 +2026-09-16 13:08:53 | INFO | train rank=0 | step=44460/50000 progress=88.9% loss=0.084661 lr=3.319004e-06 lr_backbone=3.319004e-06 elapsed=46:38:31 avg_step=3.78s recent_step=3.12s eta=05:48:42 +2026-09-16 13:09:20 | INFO | train rank=0 | step=44470/50000 progress=88.9% loss=0.081063 lr=3.307166e-06 lr_backbone=3.307166e-06 elapsed=46:38:58 avg_step=3.78s recent_step=2.74s eta=05:48:03 +2026-09-16 13:10:06 | INFO | train rank=0 | step=44480/50000 progress=89.0% loss=0.083437 lr=3.295349e-06 lr_backbone=3.295349e-06 elapsed=46:39:44 avg_step=3.78s recent_step=4.62s eta=05:47:27 +2026-09-16 13:10:45 | INFO | train rank=0 | step=44490/50000 progress=89.0% loss=0.130572 lr=3.283553e-06 lr_backbone=3.283553e-06 elapsed=46:40:23 avg_step=3.78s recent_step=3.86s eta=05:46:49 +2026-09-16 13:11:21 | INFO | train rank=0 | step=44500/50000 progress=89.0% loss=0.103881 lr=3.271777e-06 lr_backbone=3.271777e-06 elapsed=46:40:59 avg_step=3.78s recent_step=3.63s eta=05:46:11 +2026-09-16 13:11:54 | INFO | train rank=0 | step=44510/50000 progress=89.0% loss=0.083687 lr=3.260021e-06 lr_backbone=3.260021e-06 elapsed=46:41:32 avg_step=3.78s recent_step=3.27s eta=05:45:33 +2026-09-16 13:12:35 | INFO | train rank=0 | step=44520/50000 progress=89.0% loss=0.091866 lr=3.248286e-06 lr_backbone=3.248286e-06 elapsed=46:42:13 avg_step=3.78s recent_step=4.11s eta=05:44:55 +2026-09-16 13:13:15 | INFO | train rank=0 | step=44530/50000 progress=89.1% loss=0.078283 lr=3.236571e-06 lr_backbone=3.236571e-06 elapsed=46:42:53 avg_step=3.78s recent_step=4.03s eta=05:44:18 +2026-09-16 13:14:02 | INFO | train rank=0 | step=44540/50000 progress=89.1% loss=0.082719 lr=3.224877e-06 lr_backbone=3.224877e-06 elapsed=46:43:40 avg_step=3.78s recent_step=4.63s eta=05:43:41 +2026-09-16 13:14:33 | INFO | train rank=0 | step=44550/50000 progress=89.1% loss=0.074742 lr=3.213203e-06 lr_backbone=3.213203e-06 elapsed=46:44:11 avg_step=3.78s recent_step=3.12s eta=05:43:02 +2026-09-16 13:15:19 | INFO | train rank=0 | step=44560/50000 progress=89.1% loss=0.115369 lr=3.201550e-06 lr_backbone=3.201550e-06 elapsed=46:44:57 avg_step=3.78s recent_step=4.64s eta=05:42:26 +2026-09-16 13:15:54 | INFO | train rank=0 | step=44570/50000 progress=89.1% loss=0.095374 lr=3.189917e-06 lr_backbone=3.189917e-06 elapsed=46:45:33 avg_step=3.78s recent_step=3.53s eta=05:41:48 +2026-09-16 13:16:29 | INFO | train rank=0 | step=44580/50000 progress=89.2% loss=0.091820 lr=3.178304e-06 lr_backbone=3.178304e-06 elapsed=46:46:07 avg_step=3.78s recent_step=3.48s eta=05:41:10 +2026-09-16 13:17:02 | INFO | train rank=0 | step=44590/50000 progress=89.2% loss=0.073921 lr=3.166712e-06 lr_backbone=3.166712e-06 elapsed=46:46:40 avg_step=3.78s recent_step=3.32s eta=05:40:31 +2026-09-16 13:17:50 | INFO | train rank=0 | step=44600/50000 progress=89.2% loss=0.086284 lr=3.155141e-06 lr_backbone=3.155141e-06 elapsed=46:47:28 avg_step=3.78s recent_step=4.77s eta=05:39:55 +2026-09-16 13:18:22 | INFO | train rank=0 | step=44610/50000 progress=89.2% loss=0.083692 lr=3.143590e-06 lr_backbone=3.143590e-06 elapsed=46:48:00 avg_step=3.78s recent_step=3.18s eta=05:39:16 +2026-09-16 13:18:56 | INFO | train rank=0 | step=44620/50000 progress=89.2% loss=0.085934 lr=3.132059e-06 lr_backbone=3.132059e-06 elapsed=46:48:34 avg_step=3.78s recent_step=3.39s eta=05:38:38 +2026-09-16 13:19:31 | INFO | train rank=0 | step=44630/50000 progress=89.3% loss=0.098223 lr=3.120549e-06 lr_backbone=3.120549e-06 elapsed=46:49:09 avg_step=3.78s recent_step=3.54s eta=05:38:00 +2026-09-16 13:20:20 | INFO | train rank=0 | step=44640/50000 progress=89.3% loss=0.095452 lr=3.109060e-06 lr_backbone=3.109060e-06 elapsed=46:49:58 avg_step=3.78s recent_step=4.88s eta=05:37:23 +2026-09-16 13:20:55 | INFO | train rank=0 | step=44650/50000 progress=89.3% loss=0.096383 lr=3.097591e-06 lr_backbone=3.097591e-06 elapsed=46:50:33 avg_step=3.78s recent_step=3.49s eta=05:36:45 +2026-09-16 13:21:36 | INFO | train rank=0 | step=44660/50000 progress=89.3% loss=0.094672 lr=3.086142e-06 lr_backbone=3.086142e-06 elapsed=46:51:14 avg_step=3.78s recent_step=4.11s eta=05:36:08 +2026-09-16 13:22:08 | INFO | train rank=0 | step=44670/50000 progress=89.3% loss=0.079477 lr=3.074715e-06 lr_backbone=3.074715e-06 elapsed=46:51:46 avg_step=3.78s recent_step=3.22s eta=05:35:29 +2026-09-16 13:22:40 | INFO | train rank=0 | step=44680/50000 progress=89.4% loss=0.086796 lr=3.063307e-06 lr_backbone=3.063307e-06 elapsed=46:52:18 avg_step=3.78s recent_step=3.15s eta=05:34:51 +2026-09-16 13:23:15 | INFO | train rank=0 | step=44690/50000 progress=89.4% loss=0.077781 lr=3.051920e-06 lr_backbone=3.051920e-06 elapsed=46:52:53 avg_step=3.78s recent_step=3.54s eta=05:34:13 +2026-09-16 13:24:05 | INFO | train rank=0 | step=44700/50000 progress=89.4% loss=0.105014 lr=3.040554e-06 lr_backbone=3.040554e-06 elapsed=46:53:43 avg_step=3.78s recent_step=4.95s eta=05:33:37 +2026-09-16 13:24:45 | INFO | train rank=0 | step=44710/50000 progress=89.4% loss=0.085748 lr=3.029208e-06 lr_backbone=3.029208e-06 elapsed=46:54:23 avg_step=3.78s recent_step=4.05s eta=05:32:59 +2026-09-16 13:25:19 | INFO | train rank=0 | step=44720/50000 progress=89.4% loss=0.079248 lr=3.017883e-06 lr_backbone=3.017883e-06 elapsed=46:54:57 avg_step=3.78s recent_step=3.36s eta=05:32:21 +2026-09-16 13:25:53 | INFO | train rank=0 | step=44730/50000 progress=89.5% loss=0.091875 lr=3.006578e-06 lr_backbone=3.006578e-06 elapsed=46:55:31 avg_step=3.78s recent_step=3.42s eta=05:31:43 +2026-09-16 13:26:40 | INFO | train rank=0 | step=44740/50000 progress=89.5% loss=0.104058 lr=2.995294e-06 lr_backbone=2.995294e-06 elapsed=46:56:18 avg_step=3.78s recent_step=4.68s eta=05:31:06 +2026-09-16 13:27:11 | INFO | train rank=0 | step=44750/50000 progress=89.5% loss=0.095062 lr=2.984030e-06 lr_backbone=2.984030e-06 elapsed=46:56:49 avg_step=3.78s recent_step=3.14s eta=05:30:27 +2026-09-16 13:27:47 | INFO | train rank=0 | step=44760/50000 progress=89.5% loss=0.066180 lr=2.972787e-06 lr_backbone=2.972787e-06 elapsed=46:57:26 avg_step=3.78s recent_step=3.63s eta=05:29:50 +2026-09-16 13:28:22 | INFO | train rank=0 | step=44770/50000 progress=89.5% loss=0.106459 lr=2.961565e-06 lr_backbone=2.961565e-06 elapsed=46:58:00 avg_step=3.78s recent_step=3.45s eta=05:29:11 +2026-09-16 13:29:03 | INFO | train rank=0 | step=44780/50000 progress=89.6% loss=0.098159 lr=2.950363e-06 lr_backbone=2.950363e-06 elapsed=46:58:41 avg_step=3.78s recent_step=4.15s eta=05:28:34 +2026-09-16 13:29:39 | INFO | train rank=0 | step=44790/50000 progress=89.6% loss=0.076393 lr=2.939182e-06 lr_backbone=2.939182e-06 elapsed=46:59:17 avg_step=3.78s recent_step=3.59s eta=05:27:56 +2026-09-16 13:30:17 | INFO | train rank=0 | step=44800/50000 progress=89.6% loss=0.107684 lr=2.928021e-06 lr_backbone=2.928021e-06 elapsed=46:59:55 avg_step=3.78s recent_step=3.77s eta=05:27:18 +2026-09-16 13:30:46 | INFO | train rank=0 | step=44810/50000 progress=89.6% loss=0.077887 lr=2.916881e-06 lr_backbone=2.916881e-06 elapsed=47:00:24 avg_step=3.78s recent_step=2.93s eta=05:26:40 +2026-09-16 13:31:19 | INFO | train rank=0 | step=44820/50000 progress=89.6% loss=0.088244 lr=2.905762e-06 lr_backbone=2.905762e-06 elapsed=47:00:57 avg_step=3.78s recent_step=3.25s eta=05:26:01 +2026-09-16 13:32:08 | INFO | train rank=0 | step=44830/50000 progress=89.7% loss=0.104869 lr=2.894663e-06 lr_backbone=2.894663e-06 elapsed=47:01:46 avg_step=3.78s recent_step=4.96s eta=05:25:25 +2026-09-16 13:32:46 | INFO | train rank=0 | step=44840/50000 progress=89.7% loss=0.069277 lr=2.883584e-06 lr_backbone=2.883584e-06 elapsed=47:02:24 avg_step=3.78s recent_step=3.79s eta=05:24:47 +2026-09-16 13:33:27 | INFO | train rank=0 | step=44850/50000 progress=89.7% loss=0.083317 lr=2.872527e-06 lr_backbone=2.872527e-06 elapsed=47:03:05 avg_step=3.78s recent_step=4.07s eta=05:24:10 +2026-09-16 13:33:58 | INFO | train rank=0 | step=44860/50000 progress=89.7% loss=0.069170 lr=2.861490e-06 lr_backbone=2.861490e-06 elapsed=47:03:36 avg_step=3.78s recent_step=3.09s eta=05:23:31 +2026-09-16 13:34:42 | INFO | train rank=0 | step=44870/50000 progress=89.7% loss=0.087071 lr=2.850473e-06 lr_backbone=2.850473e-06 elapsed=47:04:20 avg_step=3.78s recent_step=4.37s eta=05:22:54 +2026-09-16 13:35:19 | INFO | train rank=0 | step=44880/50000 progress=89.8% loss=0.081660 lr=2.839477e-06 lr_backbone=2.839477e-06 elapsed=47:04:57 avg_step=3.78s recent_step=3.70s eta=05:22:16 +2026-09-16 13:35:57 | INFO | train rank=0 | step=44890/50000 progress=89.8% loss=0.098097 lr=2.828502e-06 lr_backbone=2.828502e-06 elapsed=47:05:35 avg_step=3.78s recent_step=3.85s eta=05:21:38 +2026-09-16 13:36:32 | INFO | train rank=0 | step=44900/50000 progress=89.8% loss=0.072877 lr=2.817548e-06 lr_backbone=2.817548e-06 elapsed=47:06:10 avg_step=3.78s recent_step=3.45s eta=05:21:00 +2026-09-16 13:37:26 | INFO | train rank=0 | step=44910/50000 progress=89.8% loss=0.112606 lr=2.806614e-06 lr_backbone=2.806614e-06 elapsed=47:07:04 avg_step=3.78s recent_step=5.38s eta=05:20:24 +2026-09-16 13:37:58 | INFO | train rank=0 | step=44920/50000 progress=89.8% loss=0.062957 lr=2.795700e-06 lr_backbone=2.795700e-06 elapsed=47:07:36 avg_step=3.78s recent_step=3.20s eta=05:19:46 +2026-09-16 13:38:26 | INFO | train rank=0 | step=44930/50000 progress=89.9% loss=0.088304 lr=2.784808e-06 lr_backbone=2.784808e-06 elapsed=47:08:04 avg_step=3.78s recent_step=2.83s eta=05:19:07 +2026-09-16 13:38:59 | INFO | train rank=0 | step=44940/50000 progress=89.9% loss=0.078398 lr=2.773936e-06 lr_backbone=2.773936e-06 elapsed=47:08:37 avg_step=3.78s recent_step=3.29s eta=05:18:29 +2026-09-16 13:39:46 | INFO | train rank=0 | step=44950/50000 progress=89.9% loss=0.074581 lr=2.763084e-06 lr_backbone=2.763084e-06 elapsed=47:09:24 avg_step=3.78s recent_step=4.67s eta=05:17:52 +2026-09-16 13:40:16 | INFO | train rank=0 | step=44960/50000 progress=89.9% loss=0.093424 lr=2.752254e-06 lr_backbone=2.752254e-06 elapsed=47:09:54 avg_step=3.78s recent_step=3.03s eta=05:17:13 +2026-09-16 13:41:00 | INFO | train rank=0 | step=44970/50000 progress=89.9% loss=0.084478 lr=2.741444e-06 lr_backbone=2.741444e-06 elapsed=47:10:38 avg_step=3.78s recent_step=4.40s eta=05:16:36 +2026-09-16 13:41:31 | INFO | train rank=0 | step=44980/50000 progress=90.0% loss=0.077124 lr=2.730655e-06 lr_backbone=2.730655e-06 elapsed=47:11:09 avg_step=3.78s recent_step=3.09s eta=05:15:58 +2026-09-16 13:42:10 | INFO | train rank=0 | step=44990/50000 progress=90.0% loss=0.115193 lr=2.719886e-06 lr_backbone=2.719886e-06 elapsed=47:11:48 avg_step=3.78s recent_step=3.87s eta=05:15:20 +2026-09-16 13:42:53 | INFO | train rank=0 | step=45000/50000 progress=90.0% loss=0.085835 lr=2.709138e-06 lr_backbone=2.709138e-06 elapsed=47:12:31 avg_step=3.78s recent_step=4.34s eta=05:14:43 +2026-09-16 13:43:42 | INFO | train rank=0 | step=45010/50000 progress=90.0% loss=0.076924 lr=2.698411e-06 lr_backbone=2.698411e-06 elapsed=47:13:21 avg_step=3.78s recent_step=4.95s eta=05:14:07 +2026-09-16 13:44:13 | INFO | train rank=0 | step=45020/50000 progress=90.0% loss=0.072607 lr=2.687704e-06 lr_backbone=2.687704e-06 elapsed=47:13:51 avg_step=3.78s recent_step=3.09s eta=05:13:28 +2026-09-16 13:44:46 | INFO | train rank=0 | step=45030/50000 progress=90.1% loss=0.092910 lr=2.677018e-06 lr_backbone=2.677018e-06 elapsed=47:14:24 avg_step=3.78s recent_step=3.23s eta=05:12:50 +2026-09-16 13:45:20 | INFO | train rank=0 | step=45040/50000 progress=90.1% loss=0.079284 lr=2.666353e-06 lr_backbone=2.666353e-06 elapsed=47:14:58 avg_step=3.78s recent_step=3.39s eta=05:12:11 +2026-09-16 13:45:59 | INFO | train rank=0 | step=45050/50000 progress=90.1% loss=0.076949 lr=2.655709e-06 lr_backbone=2.655709e-06 elapsed=47:15:37 avg_step=3.78s recent_step=3.91s eta=05:11:34 +2026-09-16 13:46:27 | INFO | train rank=0 | step=45060/50000 progress=90.1% loss=0.090580 lr=2.645085e-06 lr_backbone=2.645085e-06 elapsed=47:16:05 avg_step=3.78s recent_step=2.84s eta=05:10:55 +2026-09-16 13:47:01 | INFO | train rank=0 | step=45070/50000 progress=90.1% loss=0.103168 lr=2.634482e-06 lr_backbone=2.634482e-06 elapsed=47:16:39 avg_step=3.78s recent_step=3.38s eta=05:10:17 +2026-09-16 13:47:32 | INFO | train rank=0 | step=45080/50000 progress=90.2% loss=0.072497 lr=2.623899e-06 lr_backbone=2.623899e-06 elapsed=47:17:10 avg_step=3.78s recent_step=3.11s eta=05:09:38 +2026-09-16 13:48:08 | INFO | train rank=0 | step=45090/50000 progress=90.2% loss=0.099409 lr=2.613338e-06 lr_backbone=2.613338e-06 elapsed=47:17:46 avg_step=3.78s recent_step=3.62s eta=05:09:00 +2026-09-16 13:48:47 | INFO | train rank=0 | step=45100/50000 progress=90.2% loss=0.089402 lr=2.602797e-06 lr_backbone=2.602797e-06 elapsed=47:18:25 avg_step=3.78s recent_step=3.88s eta=05:08:23 +2026-09-16 13:49:48 | INFO | train rank=0 | step=45110/50000 progress=90.2% loss=0.079148 lr=2.592277e-06 lr_backbone=2.592277e-06 elapsed=47:19:26 avg_step=3.78s recent_step=6.10s eta=05:07:48 +2026-09-16 13:50:20 | INFO | train rank=0 | step=45120/50000 progress=90.2% loss=0.106841 lr=2.581777e-06 lr_backbone=2.581777e-06 elapsed=47:19:58 avg_step=3.78s recent_step=3.17s eta=05:07:09 +2026-09-16 13:50:51 | INFO | train rank=0 | step=45130/50000 progress=90.3% loss=0.089341 lr=2.571299e-06 lr_backbone=2.571299e-06 elapsed=47:20:29 avg_step=3.78s recent_step=3.09s eta=05:06:31 +2026-09-16 13:51:28 | INFO | train rank=0 | step=45140/50000 progress=90.3% loss=0.098005 lr=2.560841e-06 lr_backbone=2.560841e-06 elapsed=47:21:06 avg_step=3.78s recent_step=3.73s eta=05:05:53 +2026-09-16 13:52:20 | INFO | train rank=0 | step=45150/50000 progress=90.3% loss=0.071379 lr=2.550404e-06 lr_backbone=2.550404e-06 elapsed=47:21:58 avg_step=3.78s recent_step=5.16s eta=05:05:16 +2026-09-16 13:52:58 | INFO | train rank=0 | step=45160/50000 progress=90.3% loss=0.076163 lr=2.539987e-06 lr_backbone=2.539987e-06 elapsed=47:22:36 avg_step=3.78s recent_step=3.89s eta=05:04:39 +2026-09-16 13:53:31 | INFO | train rank=0 | step=45170/50000 progress=90.3% loss=0.104415 lr=2.529591e-06 lr_backbone=2.529591e-06 elapsed=47:23:09 avg_step=3.78s recent_step=3.22s eta=05:04:00 +2026-09-16 13:53:58 | INFO | train rank=0 | step=45180/50000 progress=90.4% loss=0.100735 lr=2.519217e-06 lr_backbone=2.519217e-06 elapsed=47:23:36 avg_step=3.78s recent_step=2.73s eta=05:03:22 +2026-09-16 13:54:45 | INFO | train rank=0 | step=45190/50000 progress=90.4% loss=0.068072 lr=2.508862e-06 lr_backbone=2.508862e-06 elapsed=47:24:23 avg_step=3.78s recent_step=4.74s eta=05:02:45 +2026-09-16 13:55:14 | INFO | train rank=0 | step=45200/50000 progress=90.4% loss=0.093370 lr=2.498529e-06 lr_backbone=2.498529e-06 elapsed=47:24:52 avg_step=3.78s recent_step=2.90s eta=05:02:06 +2026-09-16 13:55:49 | INFO | train rank=0 | step=45210/50000 progress=90.4% loss=0.090719 lr=2.488217e-06 lr_backbone=2.488217e-06 elapsed=47:25:27 avg_step=3.78s recent_step=3.46s eta=05:01:28 +2026-09-16 13:56:20 | INFO | train rank=0 | step=45220/50000 progress=90.4% loss=0.074013 lr=2.477925e-06 lr_backbone=2.477925e-06 elapsed=47:25:58 avg_step=3.78s recent_step=3.10s eta=05:00:50 +2026-09-16 13:57:22 | INFO | train rank=0 | step=45230/50000 progress=90.5% loss=0.081611 lr=2.467654e-06 lr_backbone=2.467654e-06 elapsed=47:27:00 avg_step=3.78s recent_step=6.22s eta=05:00:14 +2026-09-16 13:57:53 | INFO | train rank=0 | step=45240/50000 progress=90.5% loss=0.085415 lr=2.457404e-06 lr_backbone=2.457404e-06 elapsed=47:27:31 avg_step=3.78s recent_step=3.09s eta=04:59:36 +2026-09-16 13:58:29 | INFO | train rank=0 | step=45250/50000 progress=90.5% loss=0.100118 lr=2.447174e-06 lr_backbone=2.447174e-06 elapsed=47:28:07 avg_step=3.78s recent_step=3.60s eta=04:58:58 +2026-09-16 13:59:08 | INFO | train rank=0 | step=45260/50000 progress=90.5% loss=0.072269 lr=2.436966e-06 lr_backbone=2.436966e-06 elapsed=47:28:46 avg_step=3.78s recent_step=3.89s eta=04:58:20 +2026-09-16 13:59:59 | INFO | train rank=0 | step=45270/50000 progress=90.5% loss=0.091040 lr=2.426778e-06 lr_backbone=2.426778e-06 elapsed=47:29:37 avg_step=3.78s recent_step=5.12s eta=04:57:44 +2026-09-16 14:00:37 | INFO | train rank=0 | step=45280/50000 progress=90.6% loss=0.095716 lr=2.416611e-06 lr_backbone=2.416611e-06 elapsed=47:30:15 avg_step=3.78s recent_step=3.74s eta=04:57:06 +2026-09-16 14:01:05 | INFO | train rank=0 | step=45290/50000 progress=90.6% loss=0.090938 lr=2.406465e-06 lr_backbone=2.406465e-06 elapsed=47:30:43 avg_step=3.78s recent_step=2.88s eta=04:56:27 +2026-09-16 14:01:42 | INFO | train rank=0 | step=45300/50000 progress=90.6% loss=0.088705 lr=2.396339e-06 lr_backbone=2.396339e-06 elapsed=47:31:20 avg_step=3.78s recent_step=3.61s eta=04:55:50 +2026-09-16 14:02:30 | INFO | train rank=0 | step=45310/50000 progress=90.6% loss=0.077006 lr=2.386235e-06 lr_backbone=2.386235e-06 elapsed=47:32:08 avg_step=3.78s recent_step=4.88s eta=04:55:13 +2026-09-16 14:03:08 | INFO | train rank=0 | step=45320/50000 progress=90.6% loss=0.091475 lr=2.376151e-06 lr_backbone=2.376151e-06 elapsed=47:32:46 avg_step=3.78s recent_step=3.74s eta=04:54:35 +2026-09-16 14:03:42 | INFO | train rank=0 | step=45330/50000 progress=90.7% loss=0.107988 lr=2.366088e-06 lr_backbone=2.366088e-06 elapsed=47:33:20 avg_step=3.78s recent_step=3.40s eta=04:53:57 +2026-09-16 14:04:20 | INFO | train rank=0 | step=45340/50000 progress=90.7% loss=0.074799 lr=2.356046e-06 lr_backbone=2.356046e-06 elapsed=47:33:58 avg_step=3.78s recent_step=3.78s eta=04:53:19 +2026-09-16 14:05:03 | INFO | train rank=0 | step=45350/50000 progress=90.7% loss=0.072160 lr=2.346025e-06 lr_backbone=2.346025e-06 elapsed=47:34:41 avg_step=3.78s recent_step=4.34s eta=04:52:42 +2026-09-16 14:05:34 | INFO | train rank=0 | step=45360/50000 progress=90.7% loss=0.098847 lr=2.336025e-06 lr_backbone=2.336025e-06 elapsed=47:35:12 avg_step=3.78s recent_step=3.15s eta=04:52:04 +2026-09-16 14:06:08 | INFO | train rank=0 | step=45370/50000 progress=90.7% loss=0.085576 lr=2.326045e-06 lr_backbone=2.326045e-06 elapsed=47:35:46 avg_step=3.78s recent_step=3.37s eta=04:51:25 +2026-09-16 14:06:39 | INFO | train rank=0 | step=45380/50000 progress=90.8% loss=0.091597 lr=2.316086e-06 lr_backbone=2.316086e-06 elapsed=47:36:17 avg_step=3.78s recent_step=3.10s eta=04:50:47 +2026-09-16 14:07:34 | INFO | train rank=0 | step=45390/50000 progress=90.8% loss=0.073231 lr=2.306149e-06 lr_backbone=2.306149e-06 elapsed=47:37:12 avg_step=3.78s recent_step=5.44s eta=04:50:11 +2026-09-16 14:08:06 | INFO | train rank=0 | step=45400/50000 progress=90.8% loss=0.075286 lr=2.296232e-06 lr_backbone=2.296232e-06 elapsed=47:37:44 avg_step=3.78s recent_step=3.22s eta=04:49:33 +2026-09-16 14:08:36 | INFO | train rank=0 | step=45410/50000 progress=90.8% loss=0.074750 lr=2.286336e-06 lr_backbone=2.286336e-06 elapsed=47:38:14 avg_step=3.78s recent_step=2.99s eta=04:48:54 +2026-09-16 14:09:15 | INFO | train rank=0 | step=45420/50000 progress=90.8% loss=0.093327 lr=2.276460e-06 lr_backbone=2.276460e-06 elapsed=47:38:53 avg_step=3.78s recent_step=3.88s eta=04:48:16 +2026-09-16 14:09:58 | INFO | train rank=0 | step=45430/50000 progress=90.9% loss=0.109604 lr=2.266606e-06 lr_backbone=2.266606e-06 elapsed=47:39:36 avg_step=3.78s recent_step=4.38s eta=04:47:39 +2026-09-16 14:10:28 | INFO | train rank=0 | step=45440/50000 progress=90.9% loss=0.065771 lr=2.256773e-06 lr_backbone=2.256773e-06 elapsed=47:40:06 avg_step=3.78s recent_step=3.00s eta=04:47:01 +2026-09-16 14:11:06 | INFO | train rank=0 | step=45450/50000 progress=90.9% loss=0.092645 lr=2.246960e-06 lr_backbone=2.246960e-06 elapsed=47:40:44 avg_step=3.78s recent_step=3.78s eta=04:46:23 +2026-09-16 14:11:33 | INFO | train rank=0 | step=45460/50000 progress=90.9% loss=0.086782 lr=2.237169e-06 lr_backbone=2.237169e-06 elapsed=47:41:11 avg_step=3.78s recent_step=2.67s eta=04:45:44 +2026-09-16 14:12:25 | INFO | train rank=0 | step=45470/50000 progress=90.9% loss=0.088281 lr=2.227398e-06 lr_backbone=2.227398e-06 elapsed=47:42:03 avg_step=3.78s recent_step=5.25s eta=04:45:08 +2026-09-16 14:13:05 | INFO | train rank=0 | step=45480/50000 progress=91.0% loss=0.091851 lr=2.217648e-06 lr_backbone=2.217648e-06 elapsed=47:42:43 avg_step=3.78s recent_step=3.94s eta=04:44:30 +2026-09-16 14:13:43 | INFO | train rank=0 | step=45490/50000 progress=91.0% loss=0.076192 lr=2.207919e-06 lr_backbone=2.207919e-06 elapsed=47:43:21 avg_step=3.78s recent_step=3.81s eta=04:43:52 +2026-09-16 14:14:07 | INFO | train rank=0 | step=45500/50000 progress=91.0% loss=0.088021 lr=2.198211e-06 lr_backbone=2.198211e-06 elapsed=47:43:45 avg_step=3.78s recent_step=2.40s eta=04:43:13 +2026-09-16 14:15:00 | INFO | train rank=0 | step=45510/50000 progress=91.0% loss=0.095587 lr=2.188524e-06 lr_backbone=2.188524e-06 elapsed=47:44:38 avg_step=3.78s recent_step=5.33s eta=04:42:37 +2026-09-16 14:15:35 | INFO | train rank=0 | step=45520/50000 progress=91.0% loss=0.093973 lr=2.178858e-06 lr_backbone=2.178858e-06 elapsed=47:45:13 avg_step=3.78s recent_step=3.52s eta=04:41:59 +2026-09-16 14:16:08 | INFO | train rank=0 | step=45530/50000 progress=91.1% loss=0.087453 lr=2.169212e-06 lr_backbone=2.169212e-06 elapsed=47:45:46 avg_step=3.78s recent_step=3.28s eta=04:41:21 +2026-09-16 14:16:44 | INFO | train rank=0 | step=45540/50000 progress=91.1% loss=0.091603 lr=2.159588e-06 lr_backbone=2.159588e-06 elapsed=47:46:22 avg_step=3.78s recent_step=3.57s eta=04:40:43 +2026-09-16 14:17:33 | INFO | train rank=0 | step=45550/50000 progress=91.1% loss=0.084332 lr=2.149984e-06 lr_backbone=2.149984e-06 elapsed=47:47:11 avg_step=3.78s recent_step=4.93s eta=04:40:06 +2026-09-16 14:18:00 | INFO | train rank=0 | step=45560/50000 progress=91.1% loss=0.091510 lr=2.140402e-06 lr_backbone=2.140402e-06 elapsed=47:47:38 avg_step=3.78s recent_step=2.66s eta=04:39:27 +2026-09-16 14:18:38 | INFO | train rank=0 | step=45570/50000 progress=91.1% loss=0.096512 lr=2.130840e-06 lr_backbone=2.130840e-06 elapsed=47:48:16 avg_step=3.78s recent_step=3.85s eta=04:38:50 +2026-09-16 14:19:16 | INFO | train rank=0 | step=45580/50000 progress=91.2% loss=0.080068 lr=2.121300e-06 lr_backbone=2.121300e-06 elapsed=47:48:54 avg_step=3.78s recent_step=3.76s eta=04:38:12 +2026-09-16 14:20:05 | INFO | train rank=0 | step=45590/50000 progress=91.2% loss=0.072563 lr=2.111780e-06 lr_backbone=2.111780e-06 elapsed=47:49:43 avg_step=3.78s recent_step=4.90s eta=04:37:35 +2026-09-16 14:20:28 | INFO | train rank=0 | step=45600/50000 progress=91.2% loss=0.091528 lr=2.102281e-06 lr_backbone=2.102281e-06 elapsed=47:50:06 avg_step=3.78s recent_step=2.36s eta=04:36:56 +2026-09-16 14:21:05 | INFO | train rank=0 | step=45610/50000 progress=91.2% loss=0.083290 lr=2.092803e-06 lr_backbone=2.092803e-06 elapsed=47:50:43 avg_step=3.78s recent_step=3.63s eta=04:36:18 +2026-09-16 14:21:38 | INFO | train rank=0 | step=45620/50000 progress=91.2% loss=0.095023 lr=2.083346e-06 lr_backbone=2.083346e-06 elapsed=47:51:16 avg_step=3.78s recent_step=3.34s eta=04:35:40 +2026-09-16 14:22:19 | INFO | train rank=0 | step=45630/50000 progress=91.3% loss=0.094129 lr=2.073911e-06 lr_backbone=2.073911e-06 elapsed=47:51:57 avg_step=3.78s recent_step=4.06s eta=04:35:02 +2026-09-16 14:22:53 | INFO | train rank=0 | step=45640/50000 progress=91.3% loss=0.078073 lr=2.064496e-06 lr_backbone=2.064496e-06 elapsed=47:52:31 avg_step=3.78s recent_step=3.46s eta=04:34:24 +2026-09-16 14:23:43 | INFO | train rank=0 | step=45650/50000 progress=91.3% loss=0.085084 lr=2.055102e-06 lr_backbone=2.055102e-06 elapsed=47:53:21 avg_step=3.78s recent_step=5.00s eta=04:33:48 +2026-09-16 14:24:21 | INFO | train rank=0 | step=45660/50000 progress=91.3% loss=0.109902 lr=2.045729e-06 lr_backbone=2.045729e-06 elapsed=47:53:59 avg_step=3.78s recent_step=3.75s eta=04:33:10 +2026-09-16 14:24:50 | INFO | train rank=0 | step=45670/50000 progress=91.3% loss=0.092508 lr=2.036377e-06 lr_backbone=2.036377e-06 elapsed=47:54:28 avg_step=3.78s recent_step=2.90s eta=04:32:31 +2026-09-16 14:25:28 | INFO | train rank=0 | step=45680/50000 progress=91.4% loss=0.092808 lr=2.027046e-06 lr_backbone=2.027046e-06 elapsed=47:55:06 avg_step=3.78s recent_step=3.79s eta=04:31:54 +2026-09-16 14:26:23 | INFO | train rank=0 | step=45690/50000 progress=91.4% loss=0.084031 lr=2.017736e-06 lr_backbone=2.017736e-06 elapsed=47:56:01 avg_step=3.78s recent_step=5.50s eta=04:31:17 +2026-09-16 14:26:58 | INFO | train rank=0 | step=45700/50000 progress=91.4% loss=0.096153 lr=2.008447e-06 lr_backbone=2.008447e-06 elapsed=47:56:36 avg_step=3.78s recent_step=3.52s eta=04:30:39 +2026-09-16 14:27:42 | INFO | train rank=0 | step=45710/50000 progress=91.4% loss=0.079525 lr=1.999178e-06 lr_backbone=1.999178e-06 elapsed=47:57:20 avg_step=3.78s recent_step=4.39s eta=04:30:02 +2026-09-16 14:28:14 | INFO | train rank=0 | step=45720/50000 progress=91.4% loss=0.095131 lr=1.989931e-06 lr_backbone=1.989931e-06 elapsed=47:57:52 avg_step=3.78s recent_step=3.22s eta=04:29:24 +2026-09-16 14:28:56 | INFO | train rank=0 | step=45730/50000 progress=91.5% loss=0.071212 lr=1.980705e-06 lr_backbone=1.980705e-06 elapsed=47:58:34 avg_step=3.78s recent_step=4.23s eta=04:28:47 +2026-09-16 14:29:26 | INFO | train rank=0 | step=45740/50000 progress=91.5% loss=0.093303 lr=1.971500e-06 lr_backbone=1.971500e-06 elapsed=47:59:04 avg_step=3.78s recent_step=3.01s eta=04:28:08 +2026-09-16 14:30:06 | INFO | train rank=0 | step=45750/50000 progress=91.5% loss=0.075868 lr=1.962316e-06 lr_backbone=1.962316e-06 elapsed=47:59:44 avg_step=3.78s recent_step=3.97s eta=04:27:31 +2026-09-16 14:30:41 | INFO | train rank=0 | step=45760/50000 progress=91.5% loss=0.072014 lr=1.953153e-06 lr_backbone=1.953153e-06 elapsed=48:00:19 avg_step=3.78s recent_step=3.44s eta=04:26:52 +2026-09-16 14:31:33 | INFO | train rank=0 | step=45770/50000 progress=91.5% loss=0.103707 lr=1.944011e-06 lr_backbone=1.944011e-06 elapsed=48:01:11 avg_step=3.78s recent_step=5.28s eta=04:26:16 +2026-09-16 14:32:09 | INFO | train rank=0 | step=45780/50000 progress=91.6% loss=0.081699 lr=1.934890e-06 lr_backbone=1.934890e-06 elapsed=48:01:47 avg_step=3.78s recent_step=3.52s eta=04:25:38 +2026-09-16 14:32:41 | INFO | train rank=0 | step=45790/50000 progress=91.6% loss=0.075312 lr=1.925790e-06 lr_backbone=1.925790e-06 elapsed=48:02:19 avg_step=3.78s recent_step=3.23s eta=04:25:00 +2026-09-16 14:33:03 | INFO | train rank=0 | step=45800/50000 progress=91.6% loss=0.093817 lr=1.916711e-06 lr_backbone=1.916711e-06 elapsed=48:02:41 avg_step=3.78s recent_step=2.18s eta=04:24:21 +2026-09-16 14:33:47 | INFO | train rank=0 | step=45810/50000 progress=91.6% loss=0.089956 lr=1.907653e-06 lr_backbone=1.907653e-06 elapsed=48:03:25 avg_step=3.78s recent_step=4.42s eta=04:23:43 +2026-09-16 14:34:29 | INFO | train rank=0 | step=45820/50000 progress=91.6% loss=0.094194 lr=1.898616e-06 lr_backbone=1.898616e-06 elapsed=48:04:07 avg_step=3.78s recent_step=4.18s eta=04:23:06 +2026-09-16 14:35:01 | INFO | train rank=0 | step=45830/50000 progress=91.7% loss=0.075246 lr=1.889601e-06 lr_backbone=1.889601e-06 elapsed=48:04:39 avg_step=3.78s recent_step=3.21s eta=04:22:28 +2026-09-16 14:35:33 | INFO | train rank=0 | step=45840/50000 progress=91.7% loss=0.096264 lr=1.880606e-06 lr_backbone=1.880606e-06 elapsed=48:05:11 avg_step=3.78s recent_step=3.22s eta=04:21:49 +2026-09-16 14:36:19 | INFO | train rank=0 | step=45850/50000 progress=91.7% loss=0.079733 lr=1.871632e-06 lr_backbone=1.871632e-06 elapsed=48:05:57 avg_step=3.78s recent_step=4.66s eta=04:21:12 +2026-09-16 14:36:57 | INFO | train rank=0 | step=45860/50000 progress=91.7% loss=0.071361 lr=1.862679e-06 lr_backbone=1.862679e-06 elapsed=48:06:35 avg_step=3.78s recent_step=3.78s eta=04:20:35 +2026-09-16 14:37:27 | INFO | train rank=0 | step=45870/50000 progress=91.7% loss=0.083019 lr=1.853748e-06 lr_backbone=1.853748e-06 elapsed=48:07:05 avg_step=3.78s recent_step=2.93s eta=04:19:56 +2026-09-16 14:38:02 | INFO | train rank=0 | step=45880/50000 progress=91.8% loss=0.066238 lr=1.844837e-06 lr_backbone=1.844837e-06 elapsed=48:07:40 avg_step=3.78s recent_step=3.52s eta=04:19:18 +2026-09-16 14:38:44 | INFO | train rank=0 | step=45890/50000 progress=91.8% loss=0.084437 lr=1.835948e-06 lr_backbone=1.835948e-06 elapsed=48:08:22 avg_step=3.78s recent_step=4.20s eta=04:18:41 +2026-09-16 14:39:23 | INFO | train rank=0 | step=45900/50000 progress=91.8% loss=0.097575 lr=1.827079e-06 lr_backbone=1.827079e-06 elapsed=48:09:01 avg_step=3.78s recent_step=3.87s eta=04:18:03 +2026-09-16 14:39:57 | INFO | train rank=0 | step=45910/50000 progress=91.8% loss=0.062486 lr=1.818232e-06 lr_backbone=1.818232e-06 elapsed=48:09:35 avg_step=3.78s recent_step=3.47s eta=04:17:25 +2026-09-16 14:40:35 | INFO | train rank=0 | step=45920/50000 progress=91.8% loss=0.100461 lr=1.809406e-06 lr_backbone=1.809406e-06 elapsed=48:10:13 avg_step=3.78s recent_step=3.78s eta=04:16:47 +2026-09-16 14:41:14 | INFO | train rank=0 | step=45930/50000 progress=91.9% loss=0.082218 lr=1.800600e-06 lr_backbone=1.800600e-06 elapsed=48:10:52 avg_step=3.78s recent_step=3.85s eta=04:16:10 +2026-09-16 14:41:39 | INFO | train rank=0 | step=45940/50000 progress=91.9% loss=0.120533 lr=1.791816e-06 lr_backbone=1.791816e-06 elapsed=48:11:17 avg_step=3.78s recent_step=2.56s eta=04:15:31 +2026-09-16 14:42:18 | INFO | train rank=0 | step=45950/50000 progress=91.9% loss=0.088491 lr=1.783053e-06 lr_backbone=1.783053e-06 elapsed=48:11:56 avg_step=3.78s recent_step=3.86s eta=04:14:53 +2026-09-16 14:42:50 | INFO | train rank=0 | step=45960/50000 progress=91.9% loss=0.080477 lr=1.774311e-06 lr_backbone=1.774311e-06 elapsed=48:12:28 avg_step=3.78s recent_step=3.17s eta=04:14:15 +2026-09-16 14:43:45 | INFO | train rank=0 | step=45970/50000 progress=91.9% loss=0.098885 lr=1.765590e-06 lr_backbone=1.765590e-06 elapsed=48:13:23 avg_step=3.78s recent_step=5.54s eta=04:13:39 +2026-09-16 14:44:18 | INFO | train rank=0 | step=45980/50000 progress=92.0% loss=0.102185 lr=1.756891e-06 lr_backbone=1.756891e-06 elapsed=48:13:56 avg_step=3.78s recent_step=3.33s eta=04:13:00 +2026-09-16 14:44:56 | INFO | train rank=0 | step=45990/50000 progress=92.0% loss=0.077247 lr=1.748212e-06 lr_backbone=1.748212e-06 elapsed=48:14:34 avg_step=3.78s recent_step=3.78s eta=04:12:23 +2026-09-16 14:45:27 | INFO | train rank=0 | step=46000/50000 progress=92.0% loss=0.061025 lr=1.739554e-06 lr_backbone=1.739554e-06 elapsed=48:15:05 avg_step=3.78s recent_step=3.09s eta=04:11:44 +2026-09-16 14:46:11 | INFO | train rank=0 | step=46010/50000 progress=92.0% loss=0.102900 lr=1.730918e-06 lr_backbone=1.730918e-06 elapsed=48:15:49 avg_step=3.78s recent_step=4.43s eta=04:11:07 +2026-09-16 14:46:50 | INFO | train rank=0 | step=46020/50000 progress=92.0% loss=0.091142 lr=1.722303e-06 lr_backbone=1.722303e-06 elapsed=48:16:28 avg_step=3.78s recent_step=3.85s eta=04:10:29 +2026-09-16 14:47:28 | INFO | train rank=0 | step=46030/50000 progress=92.1% loss=0.080130 lr=1.713709e-06 lr_backbone=1.713709e-06 elapsed=48:17:06 avg_step=3.78s recent_step=3.79s eta=04:09:52 +2026-09-16 14:48:00 | INFO | train rank=0 | step=46040/50000 progress=92.1% loss=0.109845 lr=1.705135e-06 lr_backbone=1.705135e-06 elapsed=48:17:38 avg_step=3.78s recent_step=3.24s eta=04:09:13 +2026-09-16 14:48:56 | INFO | train rank=0 | step=46050/50000 progress=92.1% loss=0.088793 lr=1.696584e-06 lr_backbone=1.696584e-06 elapsed=48:18:34 avg_step=3.78s recent_step=5.64s eta=04:08:37 +2026-09-16 14:49:29 | INFO | train rank=0 | step=46060/50000 progress=92.1% loss=0.085253 lr=1.688053e-06 lr_backbone=1.688053e-06 elapsed=48:19:07 avg_step=3.78s recent_step=3.26s eta=04:07:59 +2026-09-16 14:50:00 | INFO | train rank=0 | step=46070/50000 progress=92.1% loss=0.090504 lr=1.679543e-06 lr_backbone=1.679543e-06 elapsed=48:19:38 avg_step=3.78s recent_step=3.14s eta=04:07:21 +2026-09-16 14:50:27 | INFO | train rank=0 | step=46080/50000 progress=92.2% loss=0.088483 lr=1.671054e-06 lr_backbone=1.671054e-06 elapsed=48:20:05 avg_step=3.78s recent_step=2.65s eta=04:06:42 +2026-09-16 14:51:16 | INFO | train rank=0 | step=46090/50000 progress=92.2% loss=0.083349 lr=1.662587e-06 lr_backbone=1.662587e-06 elapsed=48:20:54 avg_step=3.78s recent_step=4.93s eta=04:06:05 +2026-09-16 14:51:50 | INFO | train rank=0 | step=46100/50000 progress=92.2% loss=0.086312 lr=1.654141e-06 lr_backbone=1.654141e-06 elapsed=48:21:28 avg_step=3.78s recent_step=3.36s eta=04:05:27 +2026-09-16 14:52:20 | INFO | train rank=0 | step=46110/50000 progress=92.2% loss=0.076036 lr=1.645716e-06 lr_backbone=1.645716e-06 elapsed=48:21:58 avg_step=3.78s recent_step=3.06s eta=04:04:49 +2026-09-16 14:52:56 | INFO | train rank=0 | step=46120/50000 progress=92.2% loss=0.071799 lr=1.637312e-06 lr_backbone=1.637312e-06 elapsed=48:22:34 avg_step=3.78s recent_step=3.57s eta=04:04:11 +2026-09-16 14:53:44 | INFO | train rank=0 | step=46130/50000 progress=92.3% loss=0.098273 lr=1.628929e-06 lr_backbone=1.628929e-06 elapsed=48:23:22 avg_step=3.78s recent_step=4.82s eta=04:03:34 +2026-09-16 14:54:19 | INFO | train rank=0 | step=46140/50000 progress=92.3% loss=0.090541 lr=1.620567e-06 lr_backbone=1.620567e-06 elapsed=48:23:57 avg_step=3.78s recent_step=3.43s eta=04:02:56 +2026-09-16 14:54:55 | INFO | train rank=0 | step=46150/50000 progress=92.3% loss=0.089120 lr=1.612227e-06 lr_backbone=1.612227e-06 elapsed=48:24:33 avg_step=3.78s recent_step=3.67s eta=04:02:18 +2026-09-16 14:55:28 | INFO | train rank=0 | step=46160/50000 progress=92.3% loss=0.082547 lr=1.603907e-06 lr_backbone=1.603907e-06 elapsed=48:25:06 avg_step=3.78s recent_step=3.27s eta=04:01:40 +2026-09-16 14:56:22 | INFO | train rank=0 | step=46170/50000 progress=92.3% loss=0.086478 lr=1.595609e-06 lr_backbone=1.595609e-06 elapsed=48:26:00 avg_step=3.78s recent_step=5.38s eta=04:01:03 +2026-09-16 14:57:01 | INFO | train rank=0 | step=46180/50000 progress=92.4% loss=0.085356 lr=1.587332e-06 lr_backbone=1.587332e-06 elapsed=48:26:39 avg_step=3.78s recent_step=3.91s eta=04:00:26 +2026-09-16 14:57:26 | INFO | train rank=0 | step=46190/50000 progress=92.4% loss=0.075837 lr=1.579077e-06 lr_backbone=1.579077e-06 elapsed=48:27:04 avg_step=3.78s recent_step=2.46s eta=03:59:47 +2026-09-16 14:58:01 | INFO | train rank=0 | step=46200/50000 progress=92.4% loss=0.068060 lr=1.570842e-06 lr_backbone=1.570842e-06 elapsed=48:27:39 avg_step=3.78s recent_step=3.55s eta=03:59:09 +2026-09-16 14:58:49 | INFO | train rank=0 | step=46210/50000 progress=92.4% loss=0.093475 lr=1.562629e-06 lr_backbone=1.562629e-06 elapsed=48:28:27 avg_step=3.78s recent_step=4.83s eta=03:58:32 +2026-09-16 14:59:20 | INFO | train rank=0 | step=46220/50000 progress=92.4% loss=0.082570 lr=1.554436e-06 lr_backbone=1.554436e-06 elapsed=48:28:58 avg_step=3.78s recent_step=3.02s eta=03:57:54 +2026-09-16 14:59:47 | INFO | train rank=0 | step=46230/50000 progress=92.5% loss=0.087930 lr=1.546265e-06 lr_backbone=1.546265e-06 elapsed=48:29:25 avg_step=3.78s recent_step=2.69s eta=03:57:15 +2026-09-16 15:00:23 | INFO | train rank=0 | step=46240/50000 progress=92.5% loss=0.090841 lr=1.538115e-06 lr_backbone=1.538115e-06 elapsed=48:30:01 avg_step=3.78s recent_step=3.63s eta=03:56:37 +2026-09-16 15:01:16 | INFO | train rank=0 | step=46250/50000 progress=92.5% loss=0.065850 lr=1.529987e-06 lr_backbone=1.529987e-06 elapsed=48:30:54 avg_step=3.78s recent_step=5.36s eta=03:56:01 +2026-09-16 15:01:52 | INFO | train rank=0 | step=46260/50000 progress=92.5% loss=0.080588 lr=1.521879e-06 lr_backbone=1.521879e-06 elapsed=48:31:30 avg_step=3.78s recent_step=3.55s eta=03:55:23 +2026-09-16 15:02:18 | INFO | train rank=0 | step=46270/50000 progress=92.5% loss=0.084614 lr=1.513793e-06 lr_backbone=1.513793e-06 elapsed=48:31:56 avg_step=3.78s recent_step=2.62s eta=03:54:44 +2026-09-16 15:02:51 | INFO | train rank=0 | step=46280/50000 progress=92.6% loss=0.097617 lr=1.505728e-06 lr_backbone=1.505728e-06 elapsed=48:32:29 avg_step=3.78s recent_step=3.29s eta=03:54:06 +2026-09-16 15:03:29 | INFO | train rank=0 | step=46290/50000 progress=92.6% loss=0.109665 lr=1.497684e-06 lr_backbone=1.497684e-06 elapsed=48:33:07 avg_step=3.78s recent_step=3.84s eta=03:53:28 +2026-09-16 15:03:59 | INFO | train rank=0 | step=46300/50000 progress=92.6% loss=0.106328 lr=1.489662e-06 lr_backbone=1.489662e-06 elapsed=48:33:37 avg_step=3.78s recent_step=2.95s eta=03:52:50 +2026-09-16 15:04:36 | INFO | train rank=0 | step=46310/50000 progress=92.6% loss=0.081933 lr=1.481660e-06 lr_backbone=1.481660e-06 elapsed=48:34:14 avg_step=3.78s recent_step=3.68s eta=03:52:12 +2026-09-16 15:05:13 | INFO | train rank=0 | step=46320/50000 progress=92.6% loss=0.108853 lr=1.473680e-06 lr_backbone=1.473680e-06 elapsed=48:34:51 avg_step=3.78s recent_step=3.75s eta=03:51:34 +2026-09-16 15:05:57 | INFO | train rank=0 | step=46330/50000 progress=92.7% loss=0.071192 lr=1.465721e-06 lr_backbone=1.465721e-06 elapsed=48:35:35 avg_step=3.78s recent_step=4.42s eta=03:50:57 +2026-09-16 15:06:28 | INFO | train rank=0 | step=46340/50000 progress=92.7% loss=0.080262 lr=1.457783e-06 lr_backbone=1.457783e-06 elapsed=48:36:06 avg_step=3.78s recent_step=3.12s eta=03:50:19 +2026-09-16 15:07:07 | INFO | train rank=0 | step=46350/50000 progress=92.7% loss=0.071325 lr=1.449867e-06 lr_backbone=1.449867e-06 elapsed=48:36:45 avg_step=3.78s recent_step=3.82s eta=03:49:41 +2026-09-16 15:07:49 | INFO | train rank=0 | step=46360/50000 progress=92.7% loss=0.103940 lr=1.441972e-06 lr_backbone=1.441972e-06 elapsed=48:37:27 avg_step=3.78s recent_step=4.23s eta=03:49:04 +2026-09-16 15:08:26 | INFO | train rank=0 | step=46370/50000 progress=92.7% loss=0.070095 lr=1.434098e-06 lr_backbone=1.434098e-06 elapsed=48:38:04 avg_step=3.78s recent_step=3.67s eta=03:48:26 +2026-09-16 15:08:55 | INFO | train rank=0 | step=46380/50000 progress=92.8% loss=0.087499 lr=1.426245e-06 lr_backbone=1.426245e-06 elapsed=48:38:33 avg_step=3.78s recent_step=2.91s eta=03:47:47 +2026-09-16 15:09:28 | INFO | train rank=0 | step=46390/50000 progress=92.8% loss=0.115435 lr=1.418413e-06 lr_backbone=1.418413e-06 elapsed=48:39:06 avg_step=3.78s recent_step=3.31s eta=03:47:09 +2026-09-16 15:10:04 | INFO | train rank=0 | step=46400/50000 progress=92.8% loss=0.098220 lr=1.410603e-06 lr_backbone=1.410603e-06 elapsed=48:39:42 avg_step=3.78s recent_step=3.59s eta=03:46:31 +2026-09-16 15:10:47 | INFO | train rank=0 | step=46410/50000 progress=92.8% loss=0.087551 lr=1.402814e-06 lr_backbone=1.402814e-06 elapsed=48:40:25 avg_step=3.78s recent_step=4.32s eta=03:45:54 +2026-09-16 15:11:27 | INFO | train rank=0 | step=46420/50000 progress=92.8% loss=0.067685 lr=1.395046e-06 lr_backbone=1.395046e-06 elapsed=48:41:05 avg_step=3.78s recent_step=3.99s eta=03:45:16 +2026-09-16 15:11:55 | INFO | train rank=0 | step=46430/50000 progress=92.9% loss=0.084272 lr=1.387300e-06 lr_backbone=1.387300e-06 elapsed=48:41:33 avg_step=3.78s recent_step=2.80s eta=03:44:38 +2026-09-16 15:12:23 | INFO | train rank=0 | step=46440/50000 progress=92.9% loss=0.082887 lr=1.379575e-06 lr_backbone=1.379575e-06 elapsed=48:42:01 avg_step=3.78s recent_step=2.81s eta=03:43:59 +2026-09-16 15:13:04 | INFO | train rank=0 | step=46450/50000 progress=92.9% loss=0.078556 lr=1.371871e-06 lr_backbone=1.371871e-06 elapsed=48:42:42 avg_step=3.78s recent_step=4.15s eta=03:43:22 +2026-09-16 15:13:59 | INFO | train rank=0 | step=46460/50000 progress=92.9% loss=0.101194 lr=1.364188e-06 lr_backbone=1.364188e-06 elapsed=48:43:37 avg_step=3.78s recent_step=5.46s eta=03:42:45 +2026-09-16 15:14:32 | INFO | train rank=0 | step=46470/50000 progress=92.9% loss=0.083765 lr=1.356527e-06 lr_backbone=1.356527e-06 elapsed=48:44:10 avg_step=3.78s recent_step=3.27s eta=03:42:07 +2026-09-16 15:15:03 | INFO | train rank=0 | step=46480/50000 progress=93.0% loss=0.079717 lr=1.348887e-06 lr_backbone=1.348887e-06 elapsed=48:44:41 avg_step=3.78s recent_step=3.15s eta=03:41:29 +2026-09-16 15:15:41 | INFO | train rank=0 | step=46490/50000 progress=93.0% loss=0.088396 lr=1.341268e-06 lr_backbone=1.341268e-06 elapsed=48:45:19 avg_step=3.78s recent_step=3.75s eta=03:40:51 +2026-09-16 15:16:27 | INFO | train rank=0 | step=46500/50000 progress=93.0% loss=0.102467 lr=1.333670e-06 lr_backbone=1.333670e-06 elapsed=48:46:05 avg_step=3.78s recent_step=4.65s eta=03:40:14 +2026-09-16 15:16:59 | INFO | train rank=0 | step=46510/50000 progress=93.0% loss=0.066549 lr=1.326094e-06 lr_backbone=1.326094e-06 elapsed=48:46:37 avg_step=3.78s recent_step=3.16s eta=03:39:36 +2026-09-16 15:17:39 | INFO | train rank=0 | step=46520/50000 progress=93.0% loss=0.098270 lr=1.318539e-06 lr_backbone=1.318539e-06 elapsed=48:47:18 avg_step=3.78s recent_step=4.05s eta=03:38:58 +2026-09-16 15:18:19 | INFO | train rank=0 | step=46530/50000 progress=93.1% loss=0.102832 lr=1.311005e-06 lr_backbone=1.311005e-06 elapsed=48:47:57 avg_step=3.78s recent_step=3.90s eta=03:38:21 +2026-09-16 15:19:10 | INFO | train rank=0 | step=46540/50000 progress=93.1% loss=0.079744 lr=1.303493e-06 lr_backbone=1.303493e-06 elapsed=48:48:48 avg_step=3.78s recent_step=5.13s eta=03:37:44 +2026-09-16 15:19:41 | INFO | train rank=0 | step=46550/50000 progress=93.1% loss=0.089775 lr=1.296002e-06 lr_backbone=1.296002e-06 elapsed=48:49:19 avg_step=3.78s recent_step=3.10s eta=03:37:06 +2026-09-16 15:20:16 | INFO | train rank=0 | step=46560/50000 progress=93.1% loss=0.080394 lr=1.288532e-06 lr_backbone=1.288532e-06 elapsed=48:49:54 avg_step=3.78s recent_step=3.50s eta=03:36:28 +2026-09-16 15:20:47 | INFO | train rank=0 | step=46570/50000 progress=93.1% loss=0.066315 lr=1.281084e-06 lr_backbone=1.281084e-06 elapsed=48:50:25 avg_step=3.78s recent_step=3.08s eta=03:35:49 +2026-09-16 15:21:39 | INFO | train rank=0 | step=46580/50000 progress=93.2% loss=0.084685 lr=1.273656e-06 lr_backbone=1.273656e-06 elapsed=48:51:17 avg_step=3.78s recent_step=5.26s eta=03:35:13 +2026-09-16 15:22:12 | INFO | train rank=0 | step=46590/50000 progress=93.2% loss=0.076164 lr=1.266251e-06 lr_backbone=1.266251e-06 elapsed=48:51:50 avg_step=3.78s recent_step=3.30s eta=03:34:35 +2026-09-16 15:22:46 | INFO | train rank=0 | step=46600/50000 progress=93.2% loss=0.069464 lr=1.258866e-06 lr_backbone=1.258866e-06 elapsed=48:52:24 avg_step=3.78s recent_step=3.36s eta=03:33:57 +2026-09-16 15:23:19 | INFO | train rank=0 | step=46610/50000 progress=93.2% loss=0.073256 lr=1.251503e-06 lr_backbone=1.251503e-06 elapsed=48:52:57 avg_step=3.78s recent_step=3.37s eta=03:33:19 +2026-09-16 15:24:00 | INFO | train rank=0 | step=46620/50000 progress=93.2% loss=0.079471 lr=1.244161e-06 lr_backbone=1.244161e-06 elapsed=48:53:38 avg_step=3.78s recent_step=4.10s eta=03:32:41 +2026-09-16 15:24:41 | INFO | train rank=0 | step=46630/50000 progress=93.3% loss=0.074979 lr=1.236840e-06 lr_backbone=1.236840e-06 elapsed=48:54:19 avg_step=3.78s recent_step=4.03s eta=03:32:03 +2026-09-16 15:25:04 | INFO | train rank=0 | step=46640/50000 progress=93.3% loss=0.079850 lr=1.229541e-06 lr_backbone=1.229541e-06 elapsed=48:54:42 avg_step=3.78s recent_step=2.33s eta=03:31:25 +2026-09-16 15:25:37 | INFO | train rank=0 | step=46650/50000 progress=93.3% loss=0.091612 lr=1.222263e-06 lr_backbone=1.222263e-06 elapsed=48:55:15 avg_step=3.78s recent_step=3.33s eta=03:30:47 +2026-09-16 15:26:30 | INFO | train rank=0 | step=46660/50000 progress=93.3% loss=0.072869 lr=1.215007e-06 lr_backbone=1.215007e-06 elapsed=48:56:08 avg_step=3.78s recent_step=5.23s eta=03:30:10 +2026-09-16 15:27:06 | INFO | train rank=0 | step=46670/50000 progress=93.3% loss=0.102951 lr=1.207772e-06 lr_backbone=1.207772e-06 elapsed=48:56:44 avg_step=3.78s recent_step=3.61s eta=03:29:32 +2026-09-16 15:27:38 | INFO | train rank=0 | step=46680/50000 progress=93.4% loss=0.090949 lr=1.200558e-06 lr_backbone=1.200558e-06 elapsed=48:57:16 avg_step=3.78s recent_step=3.24s eta=03:28:54 +2026-09-16 15:28:16 | INFO | train rank=0 | step=46690/50000 progress=93.4% loss=0.081039 lr=1.193365e-06 lr_backbone=1.193365e-06 elapsed=48:57:54 avg_step=3.78s recent_step=3.82s eta=03:28:16 +2026-09-16 15:29:08 | INFO | train rank=0 | step=46700/50000 progress=93.4% loss=0.085269 lr=1.186194e-06 lr_backbone=1.186194e-06 elapsed=48:58:46 avg_step=3.78s recent_step=5.21s eta=03:27:39 +2026-09-16 15:29:46 | INFO | train rank=0 | step=46710/50000 progress=93.4% loss=0.073253 lr=1.179044e-06 lr_backbone=1.179044e-06 elapsed=48:59:24 avg_step=3.78s recent_step=3.78s eta=03:27:02 +2026-09-16 15:30:21 | INFO | train rank=0 | step=46720/50000 progress=93.4% loss=0.084036 lr=1.171916e-06 lr_backbone=1.171916e-06 elapsed=48:59:59 avg_step=3.78s recent_step=3.47s eta=03:26:24 +2026-09-16 15:30:58 | INFO | train rank=0 | step=46730/50000 progress=93.5% loss=0.085898 lr=1.164809e-06 lr_backbone=1.164809e-06 elapsed=49:00:36 avg_step=3.78s recent_step=3.71s eta=03:25:46 +2026-09-16 15:31:51 | INFO | train rank=0 | step=46740/50000 progress=93.5% loss=0.117100 lr=1.157723e-06 lr_backbone=1.157723e-06 elapsed=49:01:29 avg_step=3.78s recent_step=5.26s eta=03:25:09 +2026-09-16 15:32:19 | INFO | train rank=0 | step=46750/50000 progress=93.5% loss=0.080298 lr=1.150658e-06 lr_backbone=1.150658e-06 elapsed=49:01:57 avg_step=3.78s recent_step=2.86s eta=03:24:31 +2026-09-16 15:32:53 | INFO | train rank=0 | step=46760/50000 progress=93.5% loss=0.073721 lr=1.143615e-06 lr_backbone=1.143615e-06 elapsed=49:02:31 avg_step=3.78s recent_step=3.34s eta=03:23:53 +2026-09-16 15:33:28 | INFO | train rank=0 | step=46770/50000 progress=93.5% loss=0.085741 lr=1.136594e-06 lr_backbone=1.136594e-06 elapsed=49:03:06 avg_step=3.78s recent_step=3.50s eta=03:23:15 +2026-09-16 15:34:08 | INFO | train rank=0 | step=46780/50000 progress=93.6% loss=0.088816 lr=1.129594e-06 lr_backbone=1.129594e-06 elapsed=49:03:46 avg_step=3.78s recent_step=4.01s eta=03:22:37 +2026-09-16 15:34:43 | INFO | train rank=0 | step=46790/50000 progress=93.6% loss=0.112890 lr=1.122615e-06 lr_backbone=1.122615e-06 elapsed=49:04:21 avg_step=3.78s recent_step=3.48s eta=03:21:59 +2026-09-16 15:35:19 | INFO | train rank=0 | step=46800/50000 progress=93.6% loss=0.078093 lr=1.115657e-06 lr_backbone=1.115657e-06 elapsed=49:04:57 avg_step=3.78s recent_step=3.63s eta=03:21:21 +2026-09-16 15:35:58 | INFO | train rank=0 | step=46810/50000 progress=93.6% loss=0.093263 lr=1.108721e-06 lr_backbone=1.108721e-06 elapsed=49:05:36 avg_step=3.78s recent_step=3.95s eta=03:20:44 +2026-09-16 15:36:47 | INFO | train rank=0 | step=46820/50000 progress=93.6% loss=0.087366 lr=1.101806e-06 lr_backbone=1.101806e-06 elapsed=49:06:25 avg_step=3.78s recent_step=4.88s eta=03:20:07 +2026-09-16 15:37:25 | INFO | train rank=0 | step=46830/50000 progress=93.7% loss=0.063834 lr=1.094913e-06 lr_backbone=1.094913e-06 elapsed=49:07:03 avg_step=3.78s recent_step=3.76s eta=03:19:29 +2026-09-16 15:37:55 | INFO | train rank=0 | step=46840/50000 progress=93.7% loss=0.078703 lr=1.088041e-06 lr_backbone=1.088041e-06 elapsed=49:07:33 avg_step=3.78s recent_step=3.00s eta=03:18:51 +2026-09-16 15:38:31 | INFO | train rank=0 | step=46850/50000 progress=93.7% loss=0.089684 lr=1.081190e-06 lr_backbone=1.081190e-06 elapsed=49:08:09 avg_step=3.78s recent_step=3.61s eta=03:18:13 +2026-09-16 15:39:11 | INFO | train rank=0 | step=46860/50000 progress=93.7% loss=0.104043 lr=1.074361e-06 lr_backbone=1.074361e-06 elapsed=49:08:49 avg_step=3.78s recent_step=3.98s eta=03:17:35 +2026-09-16 15:39:40 | INFO | train rank=0 | step=46870/50000 progress=93.7% loss=0.101060 lr=1.067554e-06 lr_backbone=1.067554e-06 elapsed=49:09:18 avg_step=3.78s recent_step=2.92s eta=03:16:57 +2026-09-16 15:40:19 | INFO | train rank=0 | step=46880/50000 progress=93.8% loss=0.072295 lr=1.060767e-06 lr_backbone=1.060767e-06 elapsed=49:09:57 avg_step=3.78s recent_step=3.91s eta=03:16:19 +2026-09-16 15:40:48 | INFO | train rank=0 | step=46890/50000 progress=93.8% loss=0.095784 lr=1.054002e-06 lr_backbone=1.054002e-06 elapsed=49:10:26 avg_step=3.78s recent_step=2.92s eta=03:15:41 +2026-09-16 15:41:23 | INFO | train rank=0 | step=46900/50000 progress=93.8% loss=0.091502 lr=1.047259e-06 lr_backbone=1.047259e-06 elapsed=49:11:01 avg_step=3.78s recent_step=3.48s eta=03:15:03 +2026-09-16 15:41:59 | INFO | train rank=0 | step=46910/50000 progress=93.8% loss=0.088606 lr=1.040537e-06 lr_backbone=1.040537e-06 elapsed=49:11:37 avg_step=3.78s recent_step=3.64s eta=03:14:25 +2026-09-16 15:42:32 | INFO | train rank=0 | step=46920/50000 progress=93.8% loss=0.096399 lr=1.033836e-06 lr_backbone=1.033836e-06 elapsed=49:12:10 avg_step=3.78s recent_step=3.23s eta=03:13:47 +2026-09-16 15:43:06 | INFO | train rank=0 | step=46930/50000 progress=93.9% loss=0.076415 lr=1.027157e-06 lr_backbone=1.027157e-06 elapsed=49:12:44 avg_step=3.78s recent_step=3.44s eta=03:13:09 +2026-09-16 15:43:52 | INFO | train rank=0 | step=46940/50000 progress=93.9% loss=0.091856 lr=1.020499e-06 lr_backbone=1.020499e-06 elapsed=49:13:30 avg_step=3.78s recent_step=4.58s eta=03:12:32 +2026-09-16 15:44:28 | INFO | train rank=0 | step=46950/50000 progress=93.9% loss=0.070616 lr=1.013862e-06 lr_backbone=1.013862e-06 elapsed=49:14:06 avg_step=3.78s recent_step=3.63s eta=03:11:54 +2026-09-16 15:45:14 | INFO | train rank=0 | step=46960/50000 progress=93.9% loss=0.090336 lr=1.007247e-06 lr_backbone=1.007247e-06 elapsed=49:14:52 avg_step=3.78s recent_step=4.56s eta=03:11:17 +2026-09-16 15:45:53 | INFO | train rank=0 | step=46970/50000 progress=93.9% loss=0.082472 lr=1.000654e-06 lr_backbone=1.000654e-06 elapsed=49:15:31 avg_step=3.78s recent_step=3.90s eta=03:10:39 +2026-09-16 15:46:34 | INFO | train rank=0 | step=46980/50000 progress=94.0% loss=0.083721 lr=9.940817e-07 lr_backbone=9.940817e-07 elapsed=49:16:12 avg_step=3.78s recent_step=4.09s eta=03:10:01 +2026-09-16 15:47:05 | INFO | train rank=0 | step=46990/50000 progress=94.0% loss=0.077749 lr=9.875310e-07 lr_backbone=9.875310e-07 elapsed=49:16:43 avg_step=3.78s recent_step=3.07s eta=03:09:23 +2026-09-16 15:47:52 | INFO | train rank=0 | step=47000/50000 progress=94.0% loss=0.080023 lr=9.810017e-07 lr_backbone=9.810017e-07 elapsed=49:17:30 avg_step=3.78s recent_step=4.76s eta=03:08:46 +2026-09-16 15:48:23 | INFO | train rank=0 | step=47010/50000 progress=94.0% loss=0.078953 lr=9.744939e-07 lr_backbone=9.744939e-07 elapsed=49:18:01 avg_step=3.78s recent_step=3.14s eta=03:08:08 +2026-09-16 15:48:59 | INFO | train rank=0 | step=47020/50000 progress=94.0% loss=0.099365 lr=9.680075e-07 lr_backbone=9.680075e-07 elapsed=49:18:37 avg_step=3.78s recent_step=3.55s eta=03:07:30 +2026-09-16 15:49:35 | INFO | train rank=0 | step=47030/50000 progress=94.1% loss=0.098195 lr=9.615426e-07 lr_backbone=9.615426e-07 elapsed=49:19:13 avg_step=3.78s recent_step=3.57s eta=03:06:52 +2026-09-16 15:50:24 | INFO | train rank=0 | step=47040/50000 progress=94.1% loss=0.107853 lr=9.550991e-07 lr_backbone=9.550991e-07 elapsed=49:20:02 avg_step=3.78s recent_step=4.94s eta=03:06:15 +2026-09-16 15:50:59 | INFO | train rank=0 | step=47050/50000 progress=94.1% loss=0.082093 lr=9.486771e-07 lr_backbone=9.486771e-07 elapsed=49:20:37 avg_step=3.78s recent_step=3.50s eta=03:05:37 +2026-09-16 15:51:39 | INFO | train rank=0 | step=47060/50000 progress=94.1% loss=0.110392 lr=9.422765e-07 lr_backbone=9.422765e-07 elapsed=49:21:17 avg_step=3.78s recent_step=3.98s eta=03:05:00 +2026-09-16 15:52:05 | INFO | train rank=0 | step=47070/50000 progress=94.1% loss=0.081405 lr=9.358974e-07 lr_backbone=9.358974e-07 elapsed=49:21:43 avg_step=3.78s recent_step=2.60s eta=03:04:21 +2026-09-16 15:53:00 | INFO | train rank=0 | step=47080/50000 progress=94.2% loss=0.123447 lr=9.295398e-07 lr_backbone=9.295398e-07 elapsed=49:22:38 avg_step=3.78s recent_step=5.56s eta=03:03:44 +2026-09-16 15:53:29 | INFO | train rank=0 | step=47090/50000 progress=94.2% loss=0.094043 lr=9.232036e-07 lr_backbone=9.232036e-07 elapsed=49:23:07 avg_step=3.78s recent_step=2.89s eta=03:03:06 +2026-09-16 15:54:03 | INFO | train rank=0 | step=47100/50000 progress=94.2% loss=0.089294 lr=9.168889e-07 lr_backbone=9.168889e-07 elapsed=49:23:41 avg_step=3.78s recent_step=3.35s eta=03:02:28 +2026-09-16 15:54:33 | INFO | train rank=0 | step=47110/50000 progress=94.2% loss=0.086355 lr=9.105957e-07 lr_backbone=9.105957e-07 elapsed=49:24:11 avg_step=3.78s recent_step=2.97s eta=03:01:50 +2026-09-16 15:55:14 | INFO | train rank=0 | step=47120/50000 progress=94.2% loss=0.084350 lr=9.043239e-07 lr_backbone=9.043239e-07 elapsed=49:24:52 avg_step=3.78s recent_step=4.19s eta=03:01:12 +2026-09-16 15:55:55 | INFO | train rank=0 | step=47130/50000 progress=94.3% loss=0.136756 lr=8.980736e-07 lr_backbone=8.980736e-07 elapsed=49:25:33 avg_step=3.78s recent_step=4.02s eta=03:00:35 +2026-09-16 15:56:38 | INFO | train rank=0 | step=47140/50000 progress=94.3% loss=0.075196 lr=8.918448e-07 lr_backbone=8.918448e-07 elapsed=49:26:16 avg_step=3.78s recent_step=4.32s eta=02:59:57 +2026-09-16 15:57:12 | INFO | train rank=0 | step=47150/50000 progress=94.3% loss=0.081558 lr=8.856375e-07 lr_backbone=8.856375e-07 elapsed=49:26:50 avg_step=3.78s recent_step=3.46s eta=02:59:19 +2026-09-16 15:58:02 | INFO | train rank=0 | step=47160/50000 progress=94.3% loss=0.079781 lr=8.794516e-07 lr_backbone=8.794516e-07 elapsed=49:27:40 avg_step=3.78s recent_step=4.96s eta=02:58:42 +2026-09-16 15:58:36 | INFO | train rank=0 | step=47170/50000 progress=94.3% loss=0.064843 lr=8.732873e-07 lr_backbone=8.732873e-07 elapsed=49:28:14 avg_step=3.78s recent_step=3.38s eta=02:58:04 +2026-09-16 15:59:09 | INFO | train rank=0 | step=47180/50000 progress=94.4% loss=0.095520 lr=8.671444e-07 lr_backbone=8.671444e-07 elapsed=49:28:47 avg_step=3.78s recent_step=3.30s eta=02:57:26 +2026-09-16 15:59:42 | INFO | train rank=0 | step=47190/50000 progress=94.4% loss=0.084064 lr=8.610230e-07 lr_backbone=8.610230e-07 elapsed=49:29:20 avg_step=3.78s recent_step=3.29s eta=02:56:48 +2026-09-16 16:00:35 | INFO | train rank=0 | step=47200/50000 progress=94.4% loss=0.098267 lr=8.549231e-07 lr_backbone=8.549231e-07 elapsed=49:30:13 avg_step=3.78s recent_step=5.34s eta=02:56:11 +2026-09-16 16:01:07 | INFO | train rank=0 | step=47210/50000 progress=94.4% loss=0.090206 lr=8.488448e-07 lr_backbone=8.488448e-07 elapsed=49:30:45 avg_step=3.78s recent_step=3.14s eta=02:55:33 +2026-09-16 16:01:41 | INFO | train rank=0 | step=47220/50000 progress=94.4% loss=0.094029 lr=8.427879e-07 lr_backbone=8.427879e-07 elapsed=49:31:19 avg_step=3.78s recent_step=3.46s eta=02:54:55 +2026-09-16 16:02:14 | INFO | train rank=0 | step=47230/50000 progress=94.5% loss=0.105073 lr=8.367525e-07 lr_backbone=8.367525e-07 elapsed=49:31:52 avg_step=3.78s recent_step=3.30s eta=02:54:17 +2026-09-16 16:03:04 | INFO | train rank=0 | step=47240/50000 progress=94.5% loss=0.085220 lr=8.307386e-07 lr_backbone=8.307386e-07 elapsed=49:32:42 avg_step=3.78s recent_step=5.00s eta=02:53:40 +2026-09-16 16:03:40 | INFO | train rank=0 | step=47250/50000 progress=94.5% loss=0.075657 lr=8.247463e-07 lr_backbone=8.247463e-07 elapsed=49:33:18 avg_step=3.78s recent_step=3.60s eta=02:53:02 +2026-09-16 16:04:14 | INFO | train rank=0 | step=47260/50000 progress=94.5% loss=0.075982 lr=8.187754e-07 lr_backbone=8.187754e-07 elapsed=49:33:52 avg_step=3.78s recent_step=3.42s eta=02:52:25 +2026-09-16 16:04:53 | INFO | train rank=0 | step=47270/50000 progress=94.5% loss=0.077303 lr=8.128261e-07 lr_backbone=8.128261e-07 elapsed=49:34:31 avg_step=3.78s recent_step=3.83s eta=02:51:47 +2026-09-16 16:05:27 | INFO | train rank=0 | step=47280/50000 progress=94.6% loss=0.113902 lr=8.068982e-07 lr_backbone=8.068982e-07 elapsed=49:35:05 avg_step=3.78s recent_step=3.43s eta=02:51:09 +2026-09-16 16:06:03 | INFO | train rank=0 | step=47290/50000 progress=94.6% loss=0.091887 lr=8.009919e-07 lr_backbone=8.009919e-07 elapsed=49:35:41 avg_step=3.78s recent_step=3.63s eta=02:50:31 +2026-09-16 16:06:47 | INFO | train rank=0 | step=47300/50000 progress=94.6% loss=0.076416 lr=7.951071e-07 lr_backbone=7.951071e-07 elapsed=49:36:25 avg_step=3.78s recent_step=4.35s eta=02:49:54 +2026-09-16 16:07:23 | INFO | train rank=0 | step=47310/50000 progress=94.6% loss=0.087184 lr=7.892439e-07 lr_backbone=7.892439e-07 elapsed=49:37:01 avg_step=3.78s recent_step=3.61s eta=02:49:16 +2026-09-16 16:08:04 | INFO | train rank=0 | step=47320/50000 progress=94.6% loss=0.094406 lr=7.834022e-07 lr_backbone=7.834022e-07 elapsed=49:37:42 avg_step=3.78s recent_step=4.11s eta=02:48:38 +2026-09-16 16:08:34 | INFO | train rank=0 | step=47330/50000 progress=94.7% loss=0.079862 lr=7.775819e-07 lr_backbone=7.775819e-07 elapsed=49:38:12 avg_step=3.78s recent_step=3.02s eta=02:48:00 +2026-09-16 16:09:02 | INFO | train rank=0 | step=47340/50000 progress=94.7% loss=0.086811 lr=7.717833e-07 lr_backbone=7.717833e-07 elapsed=49:38:40 avg_step=3.78s recent_step=2.84s eta=02:47:22 +2026-09-16 16:09:45 | INFO | train rank=0 | step=47350/50000 progress=94.7% loss=0.087631 lr=7.660061e-07 lr_backbone=7.660061e-07 elapsed=49:39:23 avg_step=3.78s recent_step=4.28s eta=02:46:44 +2026-09-16 16:10:19 | INFO | train rank=0 | step=47360/50000 progress=94.7% loss=0.112187 lr=7.602505e-07 lr_backbone=7.602505e-07 elapsed=49:39:57 avg_step=3.78s recent_step=3.40s eta=02:46:06 +2026-09-16 16:10:57 | INFO | train rank=0 | step=47370/50000 progress=94.7% loss=0.077428 lr=7.545165e-07 lr_backbone=7.545165e-07 elapsed=49:40:35 avg_step=3.78s recent_step=3.76s eta=02:45:28 +2026-09-16 16:11:35 | INFO | train rank=0 | step=47380/50000 progress=94.8% loss=0.100126 lr=7.488039e-07 lr_backbone=7.488039e-07 elapsed=49:41:13 avg_step=3.78s recent_step=3.84s eta=02:44:51 +2026-09-16 16:12:25 | INFO | train rank=0 | step=47390/50000 progress=94.8% loss=0.062642 lr=7.431130e-07 lr_backbone=7.431130e-07 elapsed=49:42:03 avg_step=3.78s recent_step=5.04s eta=02:44:14 +2026-09-16 16:12:57 | INFO | train rank=0 | step=47400/50000 progress=94.8% loss=0.097703 lr=7.374435e-07 lr_backbone=7.374435e-07 elapsed=49:42:35 avg_step=3.78s recent_step=3.20s eta=02:43:36 +2026-09-16 16:13:33 | INFO | train rank=0 | step=47410/50000 progress=94.8% loss=0.089102 lr=7.317957e-07 lr_backbone=7.317957e-07 elapsed=49:43:11 avg_step=3.78s recent_step=3.58s eta=02:42:58 +2026-09-16 16:14:10 | INFO | train rank=0 | step=47420/50000 progress=94.8% loss=0.078494 lr=7.261693e-07 lr_backbone=7.261693e-07 elapsed=49:43:48 avg_step=3.78s recent_step=3.69s eta=02:42:20 +2026-09-16 16:15:12 | INFO | train rank=0 | step=47430/50000 progress=94.9% loss=0.087768 lr=7.205645e-07 lr_backbone=7.205645e-07 elapsed=49:44:50 avg_step=3.78s recent_step=6.19s eta=02:41:44 +2026-09-16 16:15:47 | INFO | train rank=0 | step=47440/50000 progress=94.9% loss=0.082420 lr=7.149813e-07 lr_backbone=7.149813e-07 elapsed=49:45:25 avg_step=3.78s recent_step=3.52s eta=02:41:06 +2026-09-16 16:16:21 | INFO | train rank=0 | step=47450/50000 progress=94.9% loss=0.097424 lr=7.094197e-07 lr_backbone=7.094197e-07 elapsed=49:45:59 avg_step=3.78s recent_step=3.40s eta=02:40:28 +2026-09-16 16:16:46 | INFO | train rank=0 | step=47460/50000 progress=94.9% loss=0.081795 lr=7.038796e-07 lr_backbone=7.038796e-07 elapsed=49:46:24 avg_step=3.78s recent_step=2.45s eta=02:39:49 +2026-09-16 16:17:36 | INFO | train rank=0 | step=47470/50000 progress=94.9% loss=0.106926 lr=6.983610e-07 lr_backbone=6.983610e-07 elapsed=49:47:14 avg_step=3.78s recent_step=5.03s eta=02:39:12 +2026-09-16 16:18:09 | INFO | train rank=0 | step=47480/50000 progress=95.0% loss=0.084541 lr=6.928641e-07 lr_backbone=6.928641e-07 elapsed=49:47:47 avg_step=3.78s recent_step=3.27s eta=02:38:34 +2026-09-16 16:18:42 | INFO | train rank=0 | step=47490/50000 progress=95.0% loss=0.086600 lr=6.873887e-07 lr_backbone=6.873887e-07 elapsed=49:48:20 avg_step=3.78s recent_step=3.33s eta=02:37:56 +2026-09-16 16:19:15 | INFO | train rank=0 | step=47500/50000 progress=95.0% loss=0.088403 lr=6.819348e-07 lr_backbone=6.819348e-07 elapsed=49:48:53 avg_step=3.78s recent_step=3.28s eta=02:37:18 +2026-09-16 16:20:03 | INFO | train rank=0 | step=47510/50000 progress=95.0% loss=0.082921 lr=6.765026e-07 lr_backbone=6.765026e-07 elapsed=49:49:41 avg_step=3.78s recent_step=4.87s eta=02:36:41 +2026-09-16 16:20:35 | INFO | train rank=0 | step=47520/50000 progress=95.0% loss=0.079261 lr=6.710919e-07 lr_backbone=6.710919e-07 elapsed=49:50:13 avg_step=3.78s recent_step=3.11s eta=02:36:03 +2026-09-16 16:21:12 | INFO | train rank=0 | step=47530/50000 progress=95.1% loss=0.078517 lr=6.657028e-07 lr_backbone=6.657028e-07 elapsed=49:50:50 avg_step=3.78s recent_step=3.72s eta=02:35:25 +2026-09-16 16:21:51 | INFO | train rank=0 | step=47540/50000 progress=95.1% loss=0.081216 lr=6.603353e-07 lr_backbone=6.603353e-07 elapsed=49:51:29 avg_step=3.78s recent_step=3.90s eta=02:34:47 +2026-09-16 16:22:33 | INFO | train rank=0 | step=47550/50000 progress=95.1% loss=0.062265 lr=6.549893e-07 lr_backbone=6.549893e-07 elapsed=49:52:11 avg_step=3.78s recent_step=4.26s eta=02:34:10 +2026-09-16 16:23:05 | INFO | train rank=0 | step=47560/50000 progress=95.1% loss=0.084806 lr=6.496650e-07 lr_backbone=6.496650e-07 elapsed=49:52:43 avg_step=3.78s recent_step=3.17s eta=02:33:32 +2026-09-16 16:23:39 | INFO | train rank=0 | step=47570/50000 progress=95.1% loss=0.082371 lr=6.443622e-07 lr_backbone=6.443622e-07 elapsed=49:53:17 avg_step=3.78s recent_step=3.35s eta=02:32:54 +2026-09-16 16:24:08 | INFO | train rank=0 | step=47580/50000 progress=95.2% loss=0.078933 lr=6.390810e-07 lr_backbone=6.390810e-07 elapsed=49:53:46 avg_step=3.78s recent_step=2.89s eta=02:32:16 +2026-09-16 16:24:50 | INFO | train rank=0 | step=47590/50000 progress=95.2% loss=0.082170 lr=6.338214e-07 lr_backbone=6.338214e-07 elapsed=49:54:28 avg_step=3.78s recent_step=4.26s eta=02:31:38 +2026-09-16 16:25:42 | INFO | train rank=0 | step=47600/50000 progress=95.2% loss=0.094299 lr=6.285835e-07 lr_backbone=6.285835e-07 elapsed=49:55:20 avg_step=3.78s recent_step=5.14s eta=02:31:01 +2026-09-16 16:26:16 | INFO | train rank=0 | step=47610/50000 progress=95.2% loss=0.087916 lr=6.233671e-07 lr_backbone=6.233671e-07 elapsed=49:55:54 avg_step=3.78s recent_step=3.49s eta=02:30:23 +2026-09-16 16:26:51 | INFO | train rank=0 | step=47620/50000 progress=95.2% loss=0.099896 lr=6.181723e-07 lr_backbone=6.181723e-07 elapsed=49:56:29 avg_step=3.78s recent_step=3.43s eta=02:29:45 +2026-09-16 16:27:27 | INFO | train rank=0 | step=47630/50000 progress=95.3% loss=0.071189 lr=6.129991e-07 lr_backbone=6.129991e-07 elapsed=49:57:05 avg_step=3.78s recent_step=3.62s eta=02:29:07 +2026-09-16 16:28:19 | INFO | train rank=0 | step=47640/50000 progress=95.3% loss=0.074694 lr=6.078475e-07 lr_backbone=6.078475e-07 elapsed=49:57:57 avg_step=3.78s recent_step=5.22s eta=02:28:30 +2026-09-16 16:28:50 | INFO | train rank=0 | step=47650/50000 progress=95.3% loss=0.083189 lr=6.027175e-07 lr_backbone=6.027175e-07 elapsed=49:58:28 avg_step=3.78s recent_step=3.12s eta=02:27:52 +2026-09-16 16:29:18 | INFO | train rank=0 | step=47660/50000 progress=95.3% loss=0.075817 lr=5.976091e-07 lr_backbone=5.976091e-07 elapsed=49:58:56 avg_step=3.78s recent_step=2.77s eta=02:27:14 +2026-09-16 16:29:54 | INFO | train rank=0 | step=47670/50000 progress=95.3% loss=0.093351 lr=5.925224e-07 lr_backbone=5.925224e-07 elapsed=49:59:32 avg_step=3.78s recent_step=3.63s eta=02:26:36 +2026-09-16 16:30:51 | INFO | train rank=0 | step=47680/50000 progress=95.4% loss=0.075572 lr=5.874572e-07 lr_backbone=5.874572e-07 elapsed=50:00:29 avg_step=3.78s recent_step=5.66s eta=02:25:59 +2026-09-16 16:31:26 | INFO | train rank=0 | step=47690/50000 progress=95.4% loss=0.082331 lr=5.824137e-07 lr_backbone=5.824137e-07 elapsed=50:01:04 avg_step=3.78s recent_step=3.55s eta=02:25:21 +2026-09-16 16:31:59 | INFO | train rank=0 | step=47700/50000 progress=95.4% loss=0.070478 lr=5.773917e-07 lr_backbone=5.773917e-07 elapsed=50:01:37 avg_step=3.78s recent_step=3.30s eta=02:24:43 +2026-09-16 16:32:34 | INFO | train rank=0 | step=47710/50000 progress=95.4% loss=0.072974 lr=5.723914e-07 lr_backbone=5.723914e-07 elapsed=50:02:12 avg_step=3.78s recent_step=3.42s eta=02:24:06 +2026-09-16 16:33:21 | INFO | train rank=0 | step=47720/50000 progress=95.4% loss=0.083997 lr=5.674128e-07 lr_backbone=5.674128e-07 elapsed=50:02:59 avg_step=3.78s recent_step=4.78s eta=02:23:28 +2026-09-16 16:33:54 | INFO | train rank=0 | step=47730/50000 progress=95.5% loss=0.094051 lr=5.624557e-07 lr_backbone=5.624557e-07 elapsed=50:03:32 avg_step=3.78s recent_step=3.23s eta=02:22:50 +2026-09-16 16:34:21 | INFO | train rank=0 | step=47740/50000 progress=95.5% loss=0.103300 lr=5.575203e-07 lr_backbone=5.575203e-07 elapsed=50:03:59 avg_step=3.78s recent_step=2.72s eta=02:22:12 +2026-09-16 16:34:59 | INFO | train rank=0 | step=47750/50000 progress=95.5% loss=0.078367 lr=5.526065e-07 lr_backbone=5.526065e-07 elapsed=50:04:37 avg_step=3.78s recent_step=3.83s eta=02:21:34 +2026-09-16 16:35:52 | INFO | train rank=0 | step=47760/50000 progress=95.5% loss=0.094126 lr=5.477143e-07 lr_backbone=5.477143e-07 elapsed=50:05:30 avg_step=3.78s recent_step=5.27s eta=02:20:57 +2026-09-16 16:36:27 | INFO | train rank=0 | step=47770/50000 progress=95.5% loss=0.092377 lr=5.428438e-07 lr_backbone=5.428438e-07 elapsed=50:06:05 avg_step=3.78s recent_step=3.47s eta=02:20:19 +2026-09-16 16:37:08 | INFO | train rank=0 | step=47780/50000 progress=95.6% loss=0.081064 lr=5.379948e-07 lr_backbone=5.379948e-07 elapsed=50:06:46 avg_step=3.78s recent_step=4.15s eta=02:19:42 +2026-09-16 16:37:41 | INFO | train rank=0 | step=47790/50000 progress=95.6% loss=0.107366 lr=5.331676e-07 lr_backbone=5.331676e-07 elapsed=50:07:19 avg_step=3.78s recent_step=3.31s eta=02:19:04 +2026-09-16 16:38:37 | INFO | train rank=0 | step=47800/50000 progress=95.6% loss=0.094253 lr=5.283619e-07 lr_backbone=5.283619e-07 elapsed=50:08:15 avg_step=3.78s recent_step=5.61s eta=02:18:27 +2026-09-16 16:39:07 | INFO | train rank=0 | step=47810/50000 progress=95.6% loss=0.103270 lr=5.235780e-07 lr_backbone=5.235780e-07 elapsed=50:08:45 avg_step=3.78s recent_step=2.96s eta=02:17:49 +2026-09-16 16:39:37 | INFO | train rank=0 | step=47820/50000 progress=95.6% loss=0.078784 lr=5.188156e-07 lr_backbone=5.188156e-07 elapsed=50:09:15 avg_step=3.78s recent_step=3.06s eta=02:17:11 +2026-09-16 16:40:06 | INFO | train rank=0 | step=47830/50000 progress=95.7% loss=0.095302 lr=5.140749e-07 lr_backbone=5.140749e-07 elapsed=50:09:44 avg_step=3.78s recent_step=2.84s eta=02:16:32 +2026-09-16 16:40:57 | INFO | train rank=0 | step=47840/50000 progress=95.7% loss=0.071622 lr=5.093558e-07 lr_backbone=5.093558e-07 elapsed=50:10:35 avg_step=3.78s recent_step=5.11s eta=02:15:55 +2026-09-16 16:41:24 | INFO | train rank=0 | step=47850/50000 progress=95.7% loss=0.077509 lr=5.046584e-07 lr_backbone=5.046584e-07 elapsed=50:11:02 avg_step=3.78s recent_step=2.73s eta=02:15:17 +2026-09-16 16:41:53 | INFO | train rank=0 | step=47860/50000 progress=95.7% loss=0.098688 lr=4.999827e-07 lr_backbone=4.999827e-07 elapsed=50:11:31 avg_step=3.78s recent_step=2.90s eta=02:14:39 +2026-09-16 16:42:26 | INFO | train rank=0 | step=47870/50000 progress=95.7% loss=0.087780 lr=4.953286e-07 lr_backbone=4.953286e-07 elapsed=50:12:04 avg_step=3.78s recent_step=3.30s eta=02:14:01 +2026-09-16 16:43:18 | INFO | train rank=0 | step=47880/50000 progress=95.8% loss=0.127822 lr=4.906961e-07 lr_backbone=4.906961e-07 elapsed=50:12:56 avg_step=3.78s recent_step=5.18s eta=02:13:24 +2026-09-16 16:43:48 | INFO | train rank=0 | step=47890/50000 progress=95.8% loss=0.091563 lr=4.860853e-07 lr_backbone=4.860853e-07 elapsed=50:13:26 avg_step=3.78s recent_step=2.98s eta=02:12:46 +2026-09-16 16:44:20 | INFO | train rank=0 | step=47900/50000 progress=95.8% loss=0.081369 lr=4.814962e-07 lr_backbone=4.814962e-07 elapsed=50:13:58 avg_step=3.78s recent_step=3.20s eta=02:12:08 +2026-09-16 16:44:52 | INFO | train rank=0 | step=47910/50000 progress=95.8% loss=0.103256 lr=4.769287e-07 lr_backbone=4.769287e-07 elapsed=50:14:30 avg_step=3.78s recent_step=3.17s eta=02:11:30 +2026-09-16 16:45:42 | INFO | train rank=0 | step=47920/50000 progress=95.8% loss=0.077294 lr=4.723829e-07 lr_backbone=4.723829e-07 elapsed=50:15:20 avg_step=3.78s recent_step=5.07s eta=02:10:52 +2026-09-16 16:46:13 | INFO | train rank=0 | step=47930/50000 progress=95.9% loss=0.085136 lr=4.678588e-07 lr_backbone=4.678588e-07 elapsed=50:15:51 avg_step=3.78s recent_step=3.10s eta=02:10:14 +2026-09-16 16:46:47 | INFO | train rank=0 | step=47940/50000 progress=95.9% loss=0.089650 lr=4.633563e-07 lr_backbone=4.633563e-07 elapsed=50:16:25 avg_step=3.78s recent_step=3.32s eta=02:09:36 +2026-09-16 16:47:20 | INFO | train rank=0 | step=47950/50000 progress=95.9% loss=0.100676 lr=4.588755e-07 lr_backbone=4.588755e-07 elapsed=50:16:58 avg_step=3.78s recent_step=3.38s eta=02:08:59 +2026-09-16 16:48:03 | INFO | train rank=0 | step=47960/50000 progress=95.9% loss=0.077831 lr=4.544163e-07 lr_backbone=4.544163e-07 elapsed=50:17:41 avg_step=3.78s recent_step=4.29s eta=02:08:21 +2026-09-16 16:48:38 | INFO | train rank=0 | step=47970/50000 progress=95.9% loss=0.092919 lr=4.499789e-07 lr_backbone=4.499789e-07 elapsed=50:18:16 avg_step=3.78s recent_step=3.47s eta=02:07:43 +2026-09-16 16:49:09 | INFO | train rank=0 | step=47980/50000 progress=96.0% loss=0.089257 lr=4.455631e-07 lr_backbone=4.455631e-07 elapsed=50:18:47 avg_step=3.78s recent_step=3.14s eta=02:07:05 +2026-09-16 16:49:41 | INFO | train rank=0 | step=47990/50000 progress=96.0% loss=0.088898 lr=4.411690e-07 lr_backbone=4.411690e-07 elapsed=50:19:19 avg_step=3.77s recent_step=3.14s eta=02:06:27 +2026-09-16 16:50:34 | INFO | train rank=0 | step=48000/50000 progress=96.0% loss=0.067874 lr=4.367965e-07 lr_backbone=4.367965e-07 elapsed=50:20:12 avg_step=3.78s recent_step=5.31s eta=02:05:50 +2026-09-16 16:51:06 | INFO | train rank=0 | step=48010/50000 progress=96.0% loss=0.073791 lr=4.324458e-07 lr_backbone=4.324458e-07 elapsed=50:20:44 avg_step=3.78s recent_step=3.22s eta=02:05:12 +2026-09-16 16:51:32 | INFO | train rank=0 | step=48020/50000 progress=96.0% loss=0.068822 lr=4.281167e-07 lr_backbone=4.281167e-07 elapsed=50:21:10 avg_step=3.77s recent_step=2.63s eta=02:04:34 +2026-09-16 16:52:15 | INFO | train rank=0 | step=48030/50000 progress=96.1% loss=0.095470 lr=4.238093e-07 lr_backbone=4.238093e-07 elapsed=50:21:53 avg_step=3.78s recent_step=4.27s eta=02:03:56 +2026-09-16 16:53:05 | INFO | train rank=0 | step=48040/50000 progress=96.1% loss=0.084153 lr=4.195236e-07 lr_backbone=4.195236e-07 elapsed=50:22:43 avg_step=3.78s recent_step=5.00s eta=02:03:19 +2026-09-16 16:53:37 | INFO | train rank=0 | step=48050/50000 progress=96.1% loss=0.067007 lr=4.152596e-07 lr_backbone=4.152596e-07 elapsed=50:23:16 avg_step=3.78s recent_step=3.25s eta=02:02:41 +2026-09-16 16:54:01 | INFO | train rank=0 | step=48060/50000 progress=96.1% loss=0.082280 lr=4.110173e-07 lr_backbone=4.110173e-07 elapsed=50:23:39 avg_step=3.77s recent_step=2.36s eta=02:02:03 +2026-09-16 16:54:29 | INFO | train rank=0 | step=48070/50000 progress=96.1% loss=0.077861 lr=4.067967e-07 lr_backbone=4.067967e-07 elapsed=50:24:07 avg_step=3.77s recent_step=2.77s eta=02:01:25 +2026-09-16 16:55:24 | INFO | train rank=0 | step=48080/50000 progress=96.2% loss=0.102831 lr=4.025977e-07 lr_backbone=4.025977e-07 elapsed=50:25:02 avg_step=3.78s recent_step=5.47s eta=02:00:48 +2026-09-16 16:56:02 | INFO | train rank=0 | step=48090/50000 progress=96.2% loss=0.078523 lr=3.984205e-07 lr_backbone=3.984205e-07 elapsed=50:25:40 avg_step=3.78s recent_step=3.81s eta=02:00:10 +2026-09-16 16:56:29 | INFO | train rank=0 | step=48100/50000 progress=96.2% loss=0.077609 lr=3.942649e-07 lr_backbone=3.942649e-07 elapsed=50:26:07 avg_step=3.77s recent_step=2.78s eta=01:59:32 +2026-09-16 16:57:03 | INFO | train rank=0 | step=48110/50000 progress=96.2% loss=0.074125 lr=3.901311e-07 lr_backbone=3.901311e-07 elapsed=50:26:41 avg_step=3.77s recent_step=3.33s eta=01:58:54 +2026-09-16 16:57:55 | INFO | train rank=0 | step=48120/50000 progress=96.2% loss=0.090135 lr=3.860189e-07 lr_backbone=3.860189e-07 elapsed=50:27:33 avg_step=3.78s recent_step=5.28s eta=01:58:17 +2026-09-16 16:58:31 | INFO | train rank=0 | step=48130/50000 progress=96.3% loss=0.089891 lr=3.819285e-07 lr_backbone=3.819285e-07 elapsed=50:28:09 avg_step=3.77s recent_step=3.54s eta=01:57:39 +2026-09-16 16:59:03 | INFO | train rank=0 | step=48140/50000 progress=96.3% loss=0.061591 lr=3.778598e-07 lr_backbone=3.778598e-07 elapsed=50:28:41 avg_step=3.77s recent_step=3.22s eta=01:57:01 +2026-09-16 16:59:47 | INFO | train rank=0 | step=48150/50000 progress=96.3% loss=0.084444 lr=3.738127e-07 lr_backbone=3.738127e-07 elapsed=50:29:25 avg_step=3.77s recent_step=4.41s eta=01:56:23 +2026-09-16 17:00:41 | INFO | train rank=0 | step=48160/50000 progress=96.3% loss=0.121155 lr=3.697874e-07 lr_backbone=3.697874e-07 elapsed=50:30:19 avg_step=3.78s recent_step=5.40s eta=01:55:46 +2026-09-16 17:01:10 | INFO | train rank=0 | step=48170/50000 progress=96.3% loss=0.082582 lr=3.657838e-07 lr_backbone=3.657838e-07 elapsed=50:30:48 avg_step=3.78s recent_step=2.85s eta=01:55:08 +2026-09-16 17:01:44 | INFO | train rank=0 | step=48180/50000 progress=96.4% loss=0.080107 lr=3.618019e-07 lr_backbone=3.618019e-07 elapsed=50:31:22 avg_step=3.78s recent_step=3.42s eta=01:54:30 +2026-09-16 17:02:23 | INFO | train rank=0 | step=48190/50000 progress=96.4% loss=0.070234 lr=3.578417e-07 lr_backbone=3.578417e-07 elapsed=50:32:01 avg_step=3.78s recent_step=3.90s eta=01:53:52 +2026-09-16 17:03:15 | INFO | train rank=0 | step=48200/50000 progress=96.4% loss=0.087824 lr=3.539033e-07 lr_backbone=3.539033e-07 elapsed=50:32:53 avg_step=3.78s recent_step=5.17s eta=01:53:15 +2026-09-16 17:03:43 | INFO | train rank=0 | step=48210/50000 progress=96.4% loss=0.075235 lr=3.499865e-07 lr_backbone=3.499865e-07 elapsed=50:33:21 avg_step=3.78s recent_step=2.80s eta=01:52:37 +2026-09-16 17:04:16 | INFO | train rank=0 | step=48220/50000 progress=96.4% loss=0.101329 lr=3.460915e-07 lr_backbone=3.460915e-07 elapsed=50:33:54 avg_step=3.78s recent_step=3.31s eta=01:51:59 +2026-09-16 17:04:50 | INFO | train rank=0 | step=48230/50000 progress=96.5% loss=0.080600 lr=3.422181e-07 lr_backbone=3.422181e-07 elapsed=50:34:28 avg_step=3.77s recent_step=3.40s eta=01:51:21 +2026-09-16 17:05:40 | INFO | train rank=0 | step=48240/50000 progress=96.5% loss=0.116021 lr=3.383666e-07 lr_backbone=3.383666e-07 elapsed=50:35:18 avg_step=3.78s recent_step=4.99s eta=01:50:44 +2026-09-16 17:06:05 | INFO | train rank=0 | step=48250/50000 progress=96.5% loss=0.095542 lr=3.345367e-07 lr_backbone=3.345367e-07 elapsed=50:35:43 avg_step=3.77s recent_step=2.57s eta=01:50:06 +2026-09-16 17:06:33 | INFO | train rank=0 | step=48260/50000 progress=96.5% loss=0.103117 lr=3.307285e-07 lr_backbone=3.307285e-07 elapsed=50:36:11 avg_step=3.77s recent_step=2.80s eta=01:49:28 +2026-09-16 17:07:04 | INFO | train rank=0 | step=48270/50000 progress=96.5% loss=0.076536 lr=3.269421e-07 lr_backbone=3.269421e-07 elapsed=50:36:42 avg_step=3.77s recent_step=3.04s eta=01:48:50 +2026-09-16 17:07:47 | INFO | train rank=0 | step=48280/50000 progress=96.6% loss=0.096325 lr=3.231774e-07 lr_backbone=3.231774e-07 elapsed=50:37:25 avg_step=3.77s recent_step=4.31s eta=01:48:12 +2026-09-16 17:08:19 | INFO | train rank=0 | step=48290/50000 progress=96.6% loss=0.085616 lr=3.194345e-07 lr_backbone=3.194345e-07 elapsed=50:37:57 avg_step=3.77s recent_step=3.27s eta=01:47:34 +2026-09-16 17:08:51 | INFO | train rank=0 | step=48300/50000 progress=96.6% loss=0.110736 lr=3.157132e-07 lr_backbone=3.157132e-07 elapsed=50:38:29 avg_step=3.77s recent_step=3.12s eta=01:46:56 +2026-09-16 17:09:24 | INFO | train rank=0 | step=48310/50000 progress=96.6% loss=0.074513 lr=3.120138e-07 lr_backbone=3.120138e-07 elapsed=50:39:02 avg_step=3.77s recent_step=3.37s eta=01:46:18 +2026-09-16 17:10:14 | INFO | train rank=0 | step=48320/50000 progress=96.6% loss=0.088336 lr=3.083360e-07 lr_backbone=3.083360e-07 elapsed=50:39:52 avg_step=3.77s recent_step=4.94s eta=01:45:41 +2026-09-16 17:10:47 | INFO | train rank=0 | step=48330/50000 progress=96.7% loss=0.063821 lr=3.046800e-07 lr_backbone=3.046800e-07 elapsed=50:40:25 avg_step=3.77s recent_step=3.36s eta=01:45:03 +2026-09-16 17:11:24 | INFO | train rank=0 | step=48340/50000 progress=96.7% loss=0.071493 lr=3.010457e-07 lr_backbone=3.010457e-07 elapsed=50:41:02 avg_step=3.77s recent_step=3.69s eta=01:44:25 +2026-09-16 17:12:16 | INFO | train rank=0 | step=48350/50000 progress=96.7% loss=0.086400 lr=2.974332e-07 lr_backbone=2.974332e-07 elapsed=50:41:54 avg_step=3.77s recent_step=5.16s eta=01:43:48 +2026-09-16 17:12:49 | INFO | train rank=0 | step=48360/50000 progress=96.7% loss=0.088790 lr=2.938424e-07 lr_backbone=2.938424e-07 elapsed=50:42:27 avg_step=3.77s recent_step=3.33s eta=01:43:10 +2026-09-16 17:13:23 | INFO | train rank=0 | step=48370/50000 progress=96.7% loss=0.083003 lr=2.902733e-07 lr_backbone=2.902733e-07 elapsed=50:43:01 avg_step=3.77s recent_step=3.40s eta=01:42:32 +2026-09-16 17:13:56 | INFO | train rank=0 | step=48380/50000 progress=96.8% loss=0.091581 lr=2.867260e-07 lr_backbone=2.867260e-07 elapsed=50:43:34 avg_step=3.77s recent_step=3.30s eta=01:41:54 +2026-09-16 17:14:51 | INFO | train rank=0 | step=48390/50000 progress=96.8% loss=0.068693 lr=2.832004e-07 lr_backbone=2.832004e-07 elapsed=50:44:29 avg_step=3.77s recent_step=5.46s eta=01:41:17 +2026-09-16 17:15:20 | INFO | train rank=0 | step=48400/50000 progress=96.8% loss=0.072044 lr=2.796966e-07 lr_backbone=2.796966e-07 elapsed=50:44:58 avg_step=3.77s recent_step=2.94s eta=01:40:39 +2026-09-16 17:15:56 | INFO | train rank=0 | step=48410/50000 progress=96.8% loss=0.087124 lr=2.762145e-07 lr_backbone=2.762145e-07 elapsed=50:45:34 avg_step=3.77s recent_step=3.57s eta=01:40:01 +2026-09-16 17:16:32 | INFO | train rank=0 | step=48420/50000 progress=96.8% loss=0.099681 lr=2.727542e-07 lr_backbone=2.727542e-07 elapsed=50:46:10 avg_step=3.77s recent_step=3.63s eta=01:39:24 +2026-09-16 17:17:17 | INFO | train rank=0 | step=48430/50000 progress=96.9% loss=0.069042 lr=2.693156e-07 lr_backbone=2.693156e-07 elapsed=50:46:55 avg_step=3.77s recent_step=4.50s eta=01:38:46 +2026-09-16 17:17:57 | INFO | train rank=0 | step=48440/50000 progress=96.9% loss=0.086702 lr=2.658988e-07 lr_backbone=2.658988e-07 elapsed=50:47:35 avg_step=3.77s recent_step=3.96s eta=01:38:08 +2026-09-16 17:18:36 | INFO | train rank=0 | step=48450/50000 progress=96.9% loss=0.071929 lr=2.625038e-07 lr_backbone=2.625038e-07 elapsed=50:48:14 avg_step=3.77s recent_step=3.92s eta=01:37:31 +2026-09-16 17:19:12 | INFO | train rank=0 | step=48460/50000 progress=96.9% loss=0.074375 lr=2.591305e-07 lr_backbone=2.591305e-07 elapsed=50:48:50 avg_step=3.77s recent_step=3.63s eta=01:36:53 +2026-09-16 17:19:52 | INFO | train rank=0 | step=48470/50000 progress=96.9% loss=0.078786 lr=2.557789e-07 lr_backbone=2.557789e-07 elapsed=50:49:30 avg_step=3.77s recent_step=4.00s eta=01:36:15 +2026-09-16 17:20:33 | INFO | train rank=0 | step=48480/50000 progress=97.0% loss=0.070483 lr=2.524492e-07 lr_backbone=2.524492e-07 elapsed=50:50:11 avg_step=3.77s recent_step=4.05s eta=01:35:37 +2026-09-16 17:21:08 | INFO | train rank=0 | step=48490/50000 progress=97.0% loss=0.097386 lr=2.491411e-07 lr_backbone=2.491411e-07 elapsed=50:50:46 avg_step=3.77s recent_step=3.53s eta=01:35:00 +2026-09-16 17:21:39 | INFO | train rank=0 | step=48500/50000 progress=97.0% loss=0.082692 lr=2.458549e-07 lr_backbone=2.458549e-07 elapsed=50:51:17 avg_step=3.77s recent_step=3.14s eta=01:34:22 +2026-09-16 17:22:11 | INFO | train rank=0 | step=48510/50000 progress=97.0% loss=0.098586 lr=2.425904e-07 lr_backbone=2.425904e-07 elapsed=50:51:49 avg_step=3.77s recent_step=3.12s eta=01:33:44 +2026-09-16 17:22:55 | INFO | train rank=0 | step=48520/50000 progress=97.0% loss=0.105680 lr=2.393477e-07 lr_backbone=2.393477e-07 elapsed=50:52:33 avg_step=3.77s recent_step=4.42s eta=01:33:06 +2026-09-16 17:23:31 | INFO | train rank=0 | step=48530/50000 progress=97.1% loss=0.077280 lr=2.361267e-07 lr_backbone=2.361267e-07 elapsed=50:53:09 avg_step=3.77s recent_step=3.64s eta=01:32:28 +2026-09-16 17:24:04 | INFO | train rank=0 | step=48540/50000 progress=97.1% loss=0.070521 lr=2.329275e-07 lr_backbone=2.329275e-07 elapsed=50:53:42 avg_step=3.77s recent_step=3.29s eta=01:31:51 +2026-09-16 17:24:32 | INFO | train rank=0 | step=48550/50000 progress=97.1% loss=0.075425 lr=2.297501e-07 lr_backbone=2.297501e-07 elapsed=50:54:10 avg_step=3.77s recent_step=2.80s eta=01:31:12 +2026-09-16 17:25:16 | INFO | train rank=0 | step=48560/50000 progress=97.1% loss=0.085897 lr=2.265944e-07 lr_backbone=2.265944e-07 elapsed=50:54:54 avg_step=3.77s recent_step=4.39s eta=01:30:35 +2026-09-16 17:25:48 | INFO | train rank=0 | step=48570/50000 progress=97.1% loss=0.086972 lr=2.234605e-07 lr_backbone=2.234605e-07 elapsed=50:55:26 avg_step=3.77s recent_step=3.24s eta=01:29:57 +2026-09-16 17:26:27 | INFO | train rank=0 | step=48580/50000 progress=97.2% loss=0.101239 lr=2.203484e-07 lr_backbone=2.203484e-07 elapsed=50:56:05 avg_step=3.77s recent_step=3.88s eta=01:29:19 +2026-09-16 17:27:08 | INFO | train rank=0 | step=48590/50000 progress=97.2% loss=0.089096 lr=2.172581e-07 lr_backbone=2.172581e-07 elapsed=50:56:46 avg_step=3.77s recent_step=4.06s eta=01:28:42 +2026-09-16 17:28:06 | INFO | train rank=0 | step=48600/50000 progress=97.2% loss=0.107270 lr=2.141896e-07 lr_backbone=2.141896e-07 elapsed=50:57:44 avg_step=3.77s recent_step=5.80s eta=01:28:04 +2026-09-16 17:28:38 | INFO | train rank=0 | step=48610/50000 progress=97.2% loss=0.078179 lr=2.111428e-07 lr_backbone=2.111428e-07 elapsed=50:58:16 avg_step=3.77s recent_step=3.21s eta=01:27:27 +2026-09-16 17:29:12 | INFO | train rank=0 | step=48620/50000 progress=97.2% loss=0.080228 lr=2.081178e-07 lr_backbone=2.081178e-07 elapsed=50:58:50 avg_step=3.77s recent_step=3.47s eta=01:26:49 +2026-09-16 17:29:44 | INFO | train rank=0 | step=48630/50000 progress=97.3% loss=0.078532 lr=2.051146e-07 lr_backbone=2.051146e-07 elapsed=50:59:22 avg_step=3.77s recent_step=3.11s eta=01:26:11 +2026-09-16 17:30:26 | INFO | train rank=0 | step=48640/50000 progress=97.3% loss=0.074583 lr=2.021331e-07 lr_backbone=2.021331e-07 elapsed=51:00:04 avg_step=3.77s recent_step=4.29s eta=01:25:33 +2026-09-16 17:30:54 | INFO | train rank=0 | step=48650/50000 progress=97.3% loss=0.076227 lr=1.991735e-07 lr_backbone=1.991735e-07 elapsed=51:00:32 avg_step=3.77s recent_step=2.74s eta=01:24:55 +2026-09-16 17:31:29 | INFO | train rank=0 | step=48660/50000 progress=97.3% loss=0.090693 lr=1.962356e-07 lr_backbone=1.962356e-07 elapsed=51:01:07 avg_step=3.77s recent_step=3.50s eta=01:24:17 +2026-09-16 17:32:02 | INFO | train rank=0 | step=48670/50000 progress=97.3% loss=0.071828 lr=1.933195e-07 lr_backbone=1.933195e-07 elapsed=51:01:40 avg_step=3.77s recent_step=3.32s eta=01:23:39 +2026-09-16 17:32:50 | INFO | train rank=0 | step=48680/50000 progress=97.4% loss=0.115924 lr=1.904253e-07 lr_backbone=1.904253e-07 elapsed=51:02:28 avg_step=3.77s recent_step=4.76s eta=01:23:02 +2026-09-16 17:33:17 | INFO | train rank=0 | step=48690/50000 progress=97.4% loss=0.099181 lr=1.875527e-07 lr_backbone=1.875527e-07 elapsed=51:02:55 avg_step=3.77s recent_step=2.78s eta=01:22:24 +2026-09-16 17:33:59 | INFO | train rank=0 | step=48700/50000 progress=97.4% loss=0.078412 lr=1.847020e-07 lr_backbone=1.847020e-07 elapsed=51:03:37 avg_step=3.77s recent_step=4.12s eta=01:21:46 +2026-09-16 17:34:32 | INFO | train rank=0 | step=48710/50000 progress=97.4% loss=0.082319 lr=1.818731e-07 lr_backbone=1.818731e-07 elapsed=51:04:10 avg_step=3.77s recent_step=3.31s eta=01:21:08 +2026-09-16 17:35:11 | INFO | train rank=0 | step=48720/50000 progress=97.4% loss=0.086561 lr=1.790660e-07 lr_backbone=1.790660e-07 elapsed=51:04:49 avg_step=3.77s recent_step=3.89s eta=01:20:31 +2026-09-16 17:35:47 | INFO | train rank=0 | step=48730/50000 progress=97.5% loss=0.143197 lr=1.762806e-07 lr_backbone=1.762806e-07 elapsed=51:05:25 avg_step=3.77s recent_step=3.65s eta=01:19:53 +2026-09-16 17:36:19 | INFO | train rank=0 | step=48740/50000 progress=97.5% loss=0.092894 lr=1.735171e-07 lr_backbone=1.735171e-07 elapsed=51:05:57 avg_step=3.77s recent_step=3.21s eta=01:19:15 +2026-09-16 17:36:52 | INFO | train rank=0 | step=48750/50000 progress=97.5% loss=0.096123 lr=1.707753e-07 lr_backbone=1.707753e-07 elapsed=51:06:30 avg_step=3.77s recent_step=3.27s eta=01:18:37 +2026-09-16 17:37:42 | INFO | train rank=0 | step=48760/50000 progress=97.5% loss=0.073636 lr=1.680554e-07 lr_backbone=1.680554e-07 elapsed=51:07:20 avg_step=3.77s recent_step=4.98s eta=01:18:00 +2026-09-16 17:38:19 | INFO | train rank=0 | step=48770/50000 progress=97.5% loss=0.088951 lr=1.653572e-07 lr_backbone=1.653572e-07 elapsed=51:07:57 avg_step=3.77s recent_step=3.70s eta=01:17:22 +2026-09-16 17:38:55 | INFO | train rank=0 | step=48780/50000 progress=97.6% loss=0.077640 lr=1.626809e-07 lr_backbone=1.626809e-07 elapsed=51:08:33 avg_step=3.77s recent_step=3.67s eta=01:16:44 +2026-09-16 17:39:27 | INFO | train rank=0 | step=48790/50000 progress=97.6% loss=0.129763 lr=1.600263e-07 lr_backbone=1.600263e-07 elapsed=51:09:05 avg_step=3.77s recent_step=3.17s eta=01:16:06 +2026-09-16 17:40:02 | INFO | train rank=0 | step=48800/50000 progress=97.6% loss=0.098193 lr=1.573936e-07 lr_backbone=1.573936e-07 elapsed=51:09:40 avg_step=3.77s recent_step=3.54s eta=01:15:29 +2026-09-16 17:40:29 | INFO | train rank=0 | step=48810/50000 progress=97.6% loss=0.076739 lr=1.547826e-07 lr_backbone=1.547826e-07 elapsed=51:10:07 avg_step=3.77s recent_step=2.61s eta=01:14:51 +2026-09-16 17:40:51 | INFO | train rank=0 | step=48820/50000 progress=97.6% loss=0.113709 lr=1.521935e-07 lr_backbone=1.521935e-07 elapsed=51:10:29 avg_step=3.77s recent_step=2.23s eta=01:14:12 +2026-09-16 17:41:24 | INFO | train rank=0 | step=48830/50000 progress=97.7% loss=0.081604 lr=1.496262e-07 lr_backbone=1.496262e-07 elapsed=51:11:02 avg_step=3.77s recent_step=3.34s eta=01:13:35 +2026-09-16 17:41:48 | INFO | train rank=0 | step=48840/50000 progress=97.7% loss=0.073809 lr=1.470806e-07 lr_backbone=1.470806e-07 elapsed=51:11:26 avg_step=3.77s recent_step=2.36s eta=01:12:56 +2026-09-16 17:42:13 | INFO | train rank=0 | step=48850/50000 progress=97.7% loss=0.086800 lr=1.445569e-07 lr_backbone=1.445569e-07 elapsed=51:11:51 avg_step=3.77s recent_step=2.54s eta=01:12:18 +2026-09-16 17:42:38 | INFO | train rank=0 | step=48860/50000 progress=97.7% loss=0.076614 lr=1.420550e-07 lr_backbone=1.420550e-07 elapsed=51:12:16 avg_step=3.77s recent_step=2.44s eta=01:11:40 +2026-09-16 17:43:03 | INFO | train rank=0 | step=48870/50000 progress=97.7% loss=0.075124 lr=1.395749e-07 lr_backbone=1.395749e-07 elapsed=51:12:41 avg_step=3.77s recent_step=2.53s eta=01:11:02 +2026-09-16 17:43:29 | INFO | train rank=0 | step=48880/50000 progress=97.8% loss=0.080430 lr=1.371166e-07 lr_backbone=1.371166e-07 elapsed=51:13:07 avg_step=3.77s recent_step=2.64s eta=01:10:24 +2026-09-16 17:43:58 | INFO | train rank=0 | step=48890/50000 progress=97.8% loss=0.099610 lr=1.346801e-07 lr_backbone=1.346801e-07 elapsed=51:13:36 avg_step=3.77s recent_step=2.83s eta=01:09:46 +2026-09-16 17:44:29 | INFO | train rank=0 | step=48900/50000 progress=97.8% loss=0.081278 lr=1.322654e-07 lr_backbone=1.322654e-07 elapsed=51:14:07 avg_step=3.77s recent_step=3.08s eta=01:09:09 +2026-09-16 17:45:02 | INFO | train rank=0 | step=48910/50000 progress=97.8% loss=0.091901 lr=1.298726e-07 lr_backbone=1.298726e-07 elapsed=51:14:40 avg_step=3.77s recent_step=3.39s eta=01:08:31 +2026-09-16 17:45:33 | INFO | train rank=0 | step=48920/50000 progress=97.8% loss=0.100815 lr=1.275015e-07 lr_backbone=1.275015e-07 elapsed=51:15:11 avg_step=3.77s recent_step=3.08s eta=01:07:53 +2026-09-16 17:45:53 | INFO | train rank=0 | step=48930/50000 progress=97.9% loss=0.097772 lr=1.251523e-07 lr_backbone=1.251523e-07 elapsed=51:15:31 avg_step=3.77s recent_step=1.95s eta=01:07:15 +2026-09-16 17:46:26 | INFO | train rank=0 | step=48940/50000 progress=97.9% loss=0.099266 lr=1.228249e-07 lr_backbone=1.228249e-07 elapsed=51:16:04 avg_step=3.77s recent_step=3.31s eta=01:06:37 +2026-09-16 17:47:01 | INFO | train rank=0 | step=48950/50000 progress=97.9% loss=0.083401 lr=1.205193e-07 lr_backbone=1.205193e-07 elapsed=51:16:39 avg_step=3.77s recent_step=3.49s eta=01:05:59 +2026-09-16 17:47:31 | INFO | train rank=0 | step=48960/50000 progress=97.9% loss=0.075766 lr=1.182355e-07 lr_backbone=1.182355e-07 elapsed=51:17:09 avg_step=3.77s recent_step=3.00s eta=01:05:21 +2026-09-16 17:47:56 | INFO | train rank=0 | step=48970/50000 progress=97.9% loss=0.094923 lr=1.159736e-07 lr_backbone=1.159736e-07 elapsed=51:17:34 avg_step=3.77s recent_step=2.48s eta=01:04:43 +2026-09-16 17:48:22 | INFO | train rank=0 | step=48980/50000 progress=98.0% loss=0.091333 lr=1.137334e-07 lr_backbone=1.137334e-07 elapsed=51:18:00 avg_step=3.77s recent_step=2.65s eta=01:04:05 +2026-09-16 17:48:50 | INFO | train rank=0 | step=48990/50000 progress=98.0% loss=0.113074 lr=1.115151e-07 lr_backbone=1.115151e-07 elapsed=51:18:28 avg_step=3.77s recent_step=2.82s eta=01:03:28 +2026-09-16 17:49:21 | INFO | train rank=0 | step=49000/50000 progress=98.0% loss=0.112374 lr=1.093186e-07 lr_backbone=1.093186e-07 elapsed=51:18:59 avg_step=3.77s recent_step=3.08s eta=01:02:50 +2026-09-16 17:49:47 | INFO | train rank=0 | step=49010/50000 progress=98.0% loss=0.092182 lr=1.071440e-07 lr_backbone=1.071440e-07 elapsed=51:19:25 avg_step=3.77s recent_step=2.61s eta=01:02:12 +2026-09-16 17:50:15 | INFO | train rank=0 | step=49020/50000 progress=98.0% loss=0.097284 lr=1.049911e-07 lr_backbone=1.049911e-07 elapsed=51:19:53 avg_step=3.77s recent_step=2.78s eta=01:01:34 +2026-09-16 17:50:45 | INFO | train rank=0 | step=49030/50000 progress=98.1% loss=0.086379 lr=1.028601e-07 lr_backbone=1.028601e-07 elapsed=51:20:23 avg_step=3.77s recent_step=2.95s eta=01:00:56 +2026-09-16 17:51:13 | INFO | train rank=0 | step=49040/50000 progress=98.1% loss=0.126879 lr=1.007509e-07 lr_backbone=1.007509e-07 elapsed=51:20:51 avg_step=3.77s recent_step=2.83s eta=01:00:18 +2026-09-16 17:51:41 | INFO | train rank=0 | step=49050/50000 progress=98.1% loss=0.101340 lr=9.866358e-08 lr_backbone=9.866358e-08 elapsed=51:21:19 avg_step=3.77s recent_step=2.80s eta=00:59:40 +2026-09-16 17:52:12 | INFO | train rank=0 | step=49060/50000 progress=98.1% loss=0.095355 lr=9.659805e-08 lr_backbone=9.659805e-08 elapsed=51:21:50 avg_step=3.77s recent_step=3.10s eta=00:59:02 +2026-09-16 17:52:40 | INFO | train rank=0 | step=49070/50000 progress=98.1% loss=0.097216 lr=9.455435e-08 lr_backbone=9.455435e-08 elapsed=51:22:18 avg_step=3.77s recent_step=2.85s eta=00:58:25 +2026-09-16 17:53:11 | INFO | train rank=0 | step=49080/50000 progress=98.2% loss=0.070741 lr=9.253248e-08 lr_backbone=9.253248e-08 elapsed=51:22:49 avg_step=3.77s recent_step=3.10s eta=00:57:47 +2026-09-16 17:53:37 | INFO | train rank=0 | step=49090/50000 progress=98.2% loss=0.107569 lr=9.053244e-08 lr_backbone=9.053244e-08 elapsed=51:23:16 avg_step=3.77s recent_step=2.61s eta=00:57:09 +2026-09-16 17:54:18 | INFO | train rank=0 | step=49100/50000 progress=98.2% loss=0.094895 lr=8.855423e-08 lr_backbone=8.855423e-08 elapsed=51:23:56 avg_step=3.77s recent_step=4.01s eta=00:56:31 +2026-09-16 17:54:47 | INFO | train rank=0 | step=49110/50000 progress=98.2% loss=0.078111 lr=8.659786e-08 lr_backbone=8.659786e-08 elapsed=51:24:25 avg_step=3.77s recent_step=2.91s eta=00:55:53 +2026-09-16 17:55:14 | INFO | train rank=0 | step=49120/50000 progress=98.2% loss=0.114762 lr=8.466332e-08 lr_backbone=8.466332e-08 elapsed=51:24:52 avg_step=3.77s recent_step=2.76s eta=00:55:15 +2026-09-16 17:55:39 | INFO | train rank=0 | step=49130/50000 progress=98.3% loss=0.110376 lr=8.275061e-08 lr_backbone=8.275061e-08 elapsed=51:25:17 avg_step=3.77s recent_step=2.46s eta=00:54:38 +2026-09-16 17:56:13 | INFO | train rank=0 | step=49140/50000 progress=98.3% loss=0.095550 lr=8.085974e-08 lr_backbone=8.085974e-08 elapsed=51:25:51 avg_step=3.77s recent_step=3.42s eta=00:54:00 +2026-09-16 17:56:39 | INFO | train rank=0 | step=49150/50000 progress=98.3% loss=0.089015 lr=7.899071e-08 lr_backbone=7.899071e-08 elapsed=51:26:17 avg_step=3.77s recent_step=2.57s eta=00:53:22 +2026-09-16 17:57:05 | INFO | train rank=0 | step=49160/50000 progress=98.3% loss=0.087315 lr=7.714351e-08 lr_backbone=7.714351e-08 elapsed=51:26:43 avg_step=3.77s recent_step=2.65s eta=00:52:44 +2026-09-16 17:57:34 | INFO | train rank=0 | step=49170/50000 progress=98.3% loss=0.072766 lr=7.531815e-08 lr_backbone=7.531815e-08 elapsed=51:27:12 avg_step=3.77s recent_step=2.86s eta=00:52:06 +2026-09-16 17:58:17 | INFO | train rank=0 | step=49180/50000 progress=98.4% loss=0.086585 lr=7.351463e-08 lr_backbone=7.351463e-08 elapsed=51:27:55 avg_step=3.77s recent_step=4.35s eta=00:51:29 +2026-09-16 17:58:46 | INFO | train rank=0 | step=49190/50000 progress=98.4% loss=0.089633 lr=7.173295e-08 lr_backbone=7.173295e-08 elapsed=51:28:24 avg_step=3.77s recent_step=2.89s eta=00:50:51 +2026-09-16 17:59:12 | INFO | train rank=0 | step=49200/50000 progress=98.4% loss=0.093782 lr=6.997311e-08 lr_backbone=6.997311e-08 elapsed=51:28:50 avg_step=3.77s recent_step=2.57s eta=00:50:13 +2026-09-16 17:59:30 | INFO | train rank=0 | step=49210/50000 progress=98.4% loss=0.101796 lr=6.823511e-08 lr_backbone=6.823511e-08 elapsed=51:29:08 avg_step=3.77s recent_step=1.80s eta=00:49:35 +2026-09-16 18:00:04 | INFO | train rank=0 | step=49220/50000 progress=98.4% loss=0.095322 lr=6.651896e-08 lr_backbone=6.651896e-08 elapsed=51:29:42 avg_step=3.77s recent_step=3.42s eta=00:48:57 +2026-09-16 18:00:27 | INFO | train rank=0 | step=49230/50000 progress=98.5% loss=0.088169 lr=6.482464e-08 lr_backbone=6.482464e-08 elapsed=51:30:05 avg_step=3.77s recent_step=2.32s eta=00:48:19 +2026-09-16 18:00:49 | INFO | train rank=0 | step=49240/50000 progress=98.5% loss=0.089630 lr=6.315217e-08 lr_backbone=6.315217e-08 elapsed=51:30:27 avg_step=3.77s recent_step=2.20s eta=00:47:42 +2026-09-16 18:01:17 | INFO | train rank=0 | step=49250/50000 progress=98.5% loss=0.089060 lr=6.150154e-08 lr_backbone=6.150154e-08 elapsed=51:30:55 avg_step=3.77s recent_step=2.80s eta=00:47:04 +2026-09-16 18:01:51 | INFO | train rank=0 | step=49260/50000 progress=98.5% loss=0.101968 lr=5.987276e-08 lr_backbone=5.987276e-08 elapsed=51:31:29 avg_step=3.77s recent_step=3.37s eta=00:46:26 +2026-09-16 18:02:22 | INFO | train rank=0 | step=49270/50000 progress=98.5% loss=0.114766 lr=5.826582e-08 lr_backbone=5.826582e-08 elapsed=51:32:00 avg_step=3.77s recent_step=3.12s eta=00:45:48 +2026-09-16 18:02:49 | INFO | train rank=0 | step=49280/50000 progress=98.6% loss=0.104032 lr=5.668073e-08 lr_backbone=5.668073e-08 elapsed=51:32:27 avg_step=3.77s recent_step=2.71s eta=00:45:10 +2026-09-16 18:03:13 | INFO | train rank=0 | step=49290/50000 progress=98.6% loss=0.097028 lr=5.511749e-08 lr_backbone=5.511749e-08 elapsed=51:32:51 avg_step=3.76s recent_step=2.36s eta=00:44:33 +2026-09-16 18:03:47 | INFO | train rank=0 | step=49300/50000 progress=98.6% loss=0.062456 lr=5.357609e-08 lr_backbone=5.357609e-08 elapsed=51:33:25 avg_step=3.76s recent_step=3.41s eta=00:43:55 +2026-09-16 18:04:14 | INFO | train rank=0 | step=49310/50000 progress=98.6% loss=0.064964 lr=5.205655e-08 lr_backbone=5.205655e-08 elapsed=51:33:52 avg_step=3.76s recent_step=2.66s eta=00:43:17 +2026-09-16 18:04:39 | INFO | train rank=0 | step=49320/50000 progress=98.6% loss=0.083983 lr=5.055885e-08 lr_backbone=5.055885e-08 elapsed=51:34:17 avg_step=3.76s recent_step=2.56s eta=00:42:39 +2026-09-16 18:05:01 | INFO | train rank=0 | step=49330/50000 progress=98.7% loss=0.103228 lr=4.908300e-08 lr_backbone=4.908300e-08 elapsed=51:34:39 avg_step=3.76s recent_step=2.20s eta=00:42:01 +2026-09-16 18:05:42 | INFO | train rank=0 | step=49340/50000 progress=98.7% loss=0.090485 lr=4.762900e-08 lr_backbone=4.762900e-08 elapsed=51:35:20 avg_step=3.76s recent_step=4.09s eta=00:41:24 +2026-09-16 18:06:14 | INFO | train rank=0 | step=49350/50000 progress=98.7% loss=0.082408 lr=4.619685e-08 lr_backbone=4.619685e-08 elapsed=51:35:52 avg_step=3.76s recent_step=3.15s eta=00:40:46 +2026-09-16 18:06:37 | INFO | train rank=0 | step=49360/50000 progress=98.7% loss=0.096916 lr=4.478655e-08 lr_backbone=4.478655e-08 elapsed=51:36:15 avg_step=3.76s recent_step=2.31s eta=00:40:08 +2026-09-16 18:07:06 | INFO | train rank=0 | step=49370/50000 progress=98.7% loss=0.063335 lr=4.339811e-08 lr_backbone=4.339811e-08 elapsed=51:36:44 avg_step=3.76s recent_step=2.88s eta=00:39:31 +2026-09-16 18:07:37 | INFO | train rank=0 | step=49380/50000 progress=98.8% loss=0.083981 lr=4.203152e-08 lr_backbone=4.203152e-08 elapsed=51:37:15 avg_step=3.76s recent_step=3.13s eta=00:38:53 +2026-09-16 18:08:02 | INFO | train rank=0 | step=49390/50000 progress=98.8% loss=0.091991 lr=4.068678e-08 lr_backbone=4.068678e-08 elapsed=51:37:40 avg_step=3.76s recent_step=2.50s eta=00:38:15 +2026-09-16 18:08:42 | INFO | train rank=0 | step=49400/50000 progress=98.8% loss=0.085058 lr=3.936389e-08 lr_backbone=3.936389e-08 elapsed=51:38:20 avg_step=3.76s recent_step=3.98s eta=00:37:37 +2026-09-16 18:09:07 | INFO | train rank=0 | step=49410/50000 progress=98.8% loss=0.078303 lr=3.806286e-08 lr_backbone=3.806286e-08 elapsed=51:38:45 avg_step=3.76s recent_step=2.49s eta=00:37:00 +2026-09-16 18:09:40 | INFO | train rank=0 | step=49420/50000 progress=98.8% loss=0.107213 lr=3.678369e-08 lr_backbone=3.678369e-08 elapsed=51:39:18 avg_step=3.76s recent_step=3.32s eta=00:36:22 +2026-09-16 18:10:06 | INFO | train rank=0 | step=49430/50000 progress=98.9% loss=0.072190 lr=3.552637e-08 lr_backbone=3.552637e-08 elapsed=51:39:44 avg_step=3.76s recent_step=2.63s eta=00:35:44 +2026-09-16 18:10:41 | INFO | train rank=0 | step=49440/50000 progress=98.9% loss=0.068146 lr=3.429090e-08 lr_backbone=3.429090e-08 elapsed=51:40:19 avg_step=3.76s recent_step=3.44s eta=00:35:07 +2026-09-16 18:11:06 | INFO | train rank=0 | step=49450/50000 progress=98.9% loss=0.079382 lr=3.307730e-08 lr_backbone=3.307730e-08 elapsed=51:40:44 avg_step=3.76s recent_step=2.52s eta=00:34:29 +2026-09-16 18:11:36 | INFO | train rank=0 | step=49460/50000 progress=98.9% loss=0.090713 lr=3.188555e-08 lr_backbone=3.188555e-08 elapsed=51:41:14 avg_step=3.76s recent_step=3.06s eta=00:33:51 +2026-09-16 18:11:59 | INFO | train rank=0 | step=49470/50000 progress=98.9% loss=0.108781 lr=3.071566e-08 lr_backbone=3.071566e-08 elapsed=51:41:37 avg_step=3.76s recent_step=2.31s eta=00:33:13 +2026-09-16 18:12:33 | INFO | train rank=0 | step=49480/50000 progress=99.0% loss=0.143849 lr=2.956762e-08 lr_backbone=2.956762e-08 elapsed=51:42:11 avg_step=3.76s recent_step=3.40s eta=00:32:36 +2026-09-16 18:13:04 | INFO | train rank=0 | step=49490/50000 progress=99.0% loss=0.078105 lr=2.844145e-08 lr_backbone=2.844145e-08 elapsed=51:42:42 avg_step=3.76s recent_step=3.08s eta=00:31:58 +2026-09-16 18:13:24 | INFO | train rank=0 | step=49500/50000 progress=99.0% loss=0.083685 lr=2.733713e-08 lr_backbone=2.733713e-08 elapsed=51:43:02 avg_step=3.76s recent_step=1.98s eta=00:31:20 +2026-09-16 18:13:59 | INFO | train rank=0 | step=49510/50000 progress=99.0% loss=0.086574 lr=2.625468e-08 lr_backbone=2.625468e-08 elapsed=51:43:37 avg_step=3.76s recent_step=3.52s eta=00:30:42 +2026-09-16 18:14:27 | INFO | train rank=0 | step=49520/50000 progress=99.0% loss=0.070573 lr=2.519408e-08 lr_backbone=2.519408e-08 elapsed=51:44:05 avg_step=3.76s recent_step=2.77s eta=00:30:05 +2026-09-16 18:14:57 | INFO | train rank=0 | step=49530/50000 progress=99.1% loss=0.071357 lr=2.415535e-08 lr_backbone=2.415535e-08 elapsed=51:44:35 avg_step=3.76s recent_step=2.97s eta=00:29:27 +2026-09-16 18:15:37 | INFO | train rank=0 | step=49540/50000 progress=99.1% loss=0.075508 lr=2.313847e-08 lr_backbone=2.313847e-08 elapsed=51:45:15 avg_step=3.76s recent_step=4.06s eta=00:28:50 +2026-09-16 18:16:07 | INFO | train rank=0 | step=49550/50000 progress=99.1% loss=0.105685 lr=2.214346e-08 lr_backbone=2.214346e-08 elapsed=51:45:45 avg_step=3.76s recent_step=2.95s eta=00:28:12 +2026-09-16 18:16:32 | INFO | train rank=0 | step=49560/50000 progress=99.1% loss=0.064261 lr=2.117031e-08 lr_backbone=2.117031e-08 elapsed=51:46:10 avg_step=3.76s recent_step=2.53s eta=00:27:34 +2026-09-16 18:17:02 | INFO | train rank=0 | step=49570/50000 progress=99.1% loss=0.075006 lr=2.021902e-08 lr_backbone=2.021902e-08 elapsed=51:46:40 avg_step=3.76s recent_step=3.04s eta=00:26:56 +2026-09-16 18:17:34 | INFO | train rank=0 | step=49580/50000 progress=99.2% loss=0.082340 lr=1.928960e-08 lr_backbone=1.928960e-08 elapsed=51:47:12 avg_step=3.76s recent_step=3.16s eta=00:26:19 +2026-09-16 18:18:04 | INFO | train rank=0 | step=49590/50000 progress=99.2% loss=0.091481 lr=1.838204e-08 lr_backbone=1.838204e-08 elapsed=51:47:42 avg_step=3.76s recent_step=3.01s eta=00:25:41 +2026-09-16 18:18:34 | INFO | train rank=0 | step=49600/50000 progress=99.2% loss=0.084969 lr=1.749634e-08 lr_backbone=1.749634e-08 elapsed=51:48:12 avg_step=3.76s recent_step=2.93s eta=00:25:03 +2026-09-16 18:19:01 | INFO | train rank=0 | step=49610/50000 progress=99.2% loss=0.084136 lr=1.663251e-08 lr_backbone=1.663251e-08 elapsed=51:48:39 avg_step=3.76s recent_step=2.78s eta=00:24:26 +2026-09-16 18:19:28 | INFO | train rank=0 | step=49620/50000 progress=99.2% loss=0.108279 lr=1.579054e-08 lr_backbone=1.579054e-08 elapsed=51:49:06 avg_step=3.76s recent_step=2.63s eta=00:23:48 +2026-09-16 18:20:02 | INFO | train rank=0 | step=49630/50000 progress=99.3% loss=0.068334 lr=1.497043e-08 lr_backbone=1.497043e-08 elapsed=51:49:40 avg_step=3.76s recent_step=3.47s eta=00:23:10 +2026-09-16 18:20:26 | INFO | train rank=0 | step=49640/50000 progress=99.3% loss=0.074883 lr=1.417219e-08 lr_backbone=1.417219e-08 elapsed=51:50:04 avg_step=3.76s recent_step=2.41s eta=00:22:33 +2026-09-16 18:20:54 | INFO | train rank=0 | step=49650/50000 progress=99.3% loss=0.081405 lr=1.339582e-08 lr_backbone=1.339582e-08 elapsed=51:50:32 avg_step=3.76s recent_step=2.75s eta=00:21:55 +2026-09-16 18:21:20 | INFO | train rank=0 | step=49660/50000 progress=99.3% loss=0.091787 lr=1.264131e-08 lr_backbone=1.264131e-08 elapsed=51:50:58 avg_step=3.76s recent_step=2.61s eta=00:21:17 +2026-09-16 18:21:54 | INFO | train rank=0 | step=49670/50000 progress=99.3% loss=0.117428 lr=1.190867e-08 lr_backbone=1.190867e-08 elapsed=51:51:32 avg_step=3.76s recent_step=3.36s eta=00:20:40 +2026-09-16 18:22:22 | INFO | train rank=0 | step=49680/50000 progress=99.4% loss=0.091134 lr=1.119789e-08 lr_backbone=1.119789e-08 elapsed=51:52:00 avg_step=3.76s recent_step=2.84s eta=00:20:02 +2026-09-16 18:22:49 | INFO | train rank=0 | step=49690/50000 progress=99.4% loss=0.090261 lr=1.050898e-08 lr_backbone=1.050898e-08 elapsed=51:52:27 avg_step=3.76s recent_step=2.70s eta=00:19:25 +2026-09-16 18:23:23 | INFO | train rank=0 | step=49700/50000 progress=99.4% loss=0.075158 lr=9.841942e-09 lr_backbone=9.841942e-09 elapsed=51:53:01 avg_step=3.76s recent_step=3.43s eta=00:18:47 +2026-09-16 18:23:52 | INFO | train rank=0 | step=49710/50000 progress=99.4% loss=0.075101 lr=9.196768e-09 lr_backbone=9.196768e-09 elapsed=51:53:30 avg_step=3.76s recent_step=2.84s eta=00:18:09 +2026-09-16 18:24:19 | INFO | train rank=0 | step=49720/50000 progress=99.4% loss=0.072416 lr=8.573461e-09 lr_backbone=8.573461e-09 elapsed=51:53:57 avg_step=3.76s recent_step=2.70s eta=00:17:32 +2026-09-16 18:24:44 | INFO | train rank=0 | step=49730/50000 progress=99.5% loss=0.070368 lr=7.972023e-09 lr_backbone=7.972023e-09 elapsed=51:54:22 avg_step=3.76s recent_step=2.59s eta=00:16:54 +2026-09-16 18:25:15 | INFO | train rank=0 | step=49740/50000 progress=99.5% loss=0.091847 lr=7.392452e-09 lr_backbone=7.392452e-09 elapsed=51:54:53 avg_step=3.76s recent_step=3.10s eta=00:16:16 +2026-09-16 18:25:40 | INFO | train rank=0 | step=49750/50000 progress=99.5% loss=0.109570 lr=6.834750e-09 lr_backbone=6.834750e-09 elapsed=51:55:18 avg_step=3.76s recent_step=2.46s eta=00:15:39 +2026-09-16 18:26:10 | INFO | train rank=0 | step=49760/50000 progress=99.5% loss=0.096934 lr=6.298917e-09 lr_backbone=6.298917e-09 elapsed=51:55:48 avg_step=3.76s recent_step=3.03s eta=00:15:01 +2026-09-16 18:26:41 | INFO | train rank=0 | step=49770/50000 progress=99.5% loss=0.089649 lr=5.784953e-09 lr_backbone=5.784953e-09 elapsed=51:56:19 avg_step=3.76s recent_step=3.04s eta=00:14:24 +2026-09-16 18:27:04 | INFO | train rank=0 | step=49780/50000 progress=99.6% loss=0.082417 lr=5.292858e-09 lr_backbone=5.292858e-09 elapsed=51:56:42 avg_step=3.76s recent_step=2.30s eta=00:13:46 +2026-09-16 18:27:47 | INFO | train rank=0 | step=49790/50000 progress=99.6% loss=0.084487 lr=4.822632e-09 lr_backbone=4.822632e-09 elapsed=51:57:25 avg_step=3.76s recent_step=4.30s eta=00:13:08 +2026-09-16 18:28:12 | INFO | train rank=0 | step=49800/50000 progress=99.6% loss=0.087160 lr=4.374276e-09 lr_backbone=4.374276e-09 elapsed=51:57:50 avg_step=3.76s recent_step=2.57s eta=00:12:31 +2026-09-16 18:28:38 | INFO | train rank=0 | step=49810/50000 progress=99.6% loss=0.077084 lr=3.947790e-09 lr_backbone=3.947790e-09 elapsed=51:58:16 avg_step=3.76s recent_step=2.55s eta=00:11:53 +2026-09-16 18:29:07 | INFO | train rank=0 | step=49820/50000 progress=99.6% loss=0.100540 lr=3.543173e-09 lr_backbone=3.543173e-09 elapsed=51:58:45 avg_step=3.76s recent_step=2.90s eta=00:11:16 +2026-09-16 18:29:49 | INFO | train rank=0 | step=49830/50000 progress=99.7% loss=0.077648 lr=3.160427e-09 lr_backbone=3.160427e-09 elapsed=51:59:27 avg_step=3.76s recent_step=4.19s eta=00:10:38 +2026-09-16 18:30:17 | INFO | train rank=0 | step=49840/50000 progress=99.7% loss=0.084739 lr=2.799551e-09 lr_backbone=2.799551e-09 elapsed=51:59:55 avg_step=3.76s recent_step=2.77s eta=00:10:00 +2026-09-16 18:30:41 | INFO | train rank=0 | step=49850/50000 progress=99.7% loss=0.088534 lr=2.460546e-09 lr_backbone=2.460546e-09 elapsed=52:00:19 avg_step=3.76s recent_step=2.40s eta=00:09:23 +2026-09-16 18:31:11 | INFO | train rank=0 | step=49860/50000 progress=99.7% loss=0.089569 lr=2.143411e-09 lr_backbone=2.143411e-09 elapsed=52:00:49 avg_step=3.76s recent_step=2.99s eta=00:08:45 +2026-09-16 18:31:43 | INFO | train rank=0 | step=49870/50000 progress=99.7% loss=0.080070 lr=1.848147e-09 lr_backbone=1.848147e-09 elapsed=52:01:21 avg_step=3.76s recent_step=3.28s eta=00:08:08 +2026-09-16 18:32:14 | INFO | train rank=0 | step=49880/50000 progress=99.8% loss=0.075222 lr=1.574754e-09 lr_backbone=1.574754e-09 elapsed=52:01:52 avg_step=3.76s recent_step=3.02s eta=00:07:30 +2026-09-16 18:32:44 | INFO | train rank=0 | step=49890/50000 progress=99.8% loss=0.075180 lr=1.323232e-09 lr_backbone=1.323232e-09 elapsed=52:02:22 avg_step=3.76s recent_step=3.08s eta=00:06:53 +2026-09-16 18:33:09 | INFO | train rank=0 | step=49900/50000 progress=99.8% loss=0.108326 lr=1.093581e-09 lr_backbone=1.093581e-09 elapsed=52:02:47 avg_step=3.75s recent_step=2.50s eta=00:06:15 +2026-09-16 18:33:33 | INFO | train rank=0 | step=49910/50000 progress=99.8% loss=0.066413 lr=8.858012e-10 lr_backbone=8.858012e-10 elapsed=52:03:11 avg_step=3.75s recent_step=2.36s eta=00:05:37 +2026-09-16 18:34:03 | INFO | train rank=0 | step=49920/50000 progress=99.8% loss=0.096761 lr=6.998928e-10 lr_backbone=6.998928e-10 elapsed=52:03:41 avg_step=3.75s recent_step=3.01s eta=00:05:00 +2026-09-16 18:34:30 | INFO | train rank=0 | step=49930/50000 progress=99.9% loss=0.076597 lr=5.358557e-10 lr_backbone=5.358557e-10 elapsed=52:04:08 avg_step=3.75s recent_step=2.73s eta=00:04:22 +2026-09-16 18:34:57 | INFO | train rank=0 | step=49940/50000 progress=99.9% loss=0.082024 lr=3.936901e-10 lr_backbone=3.936901e-10 elapsed=52:04:35 avg_step=3.75s recent_step=2.71s eta=00:03:45 +2026-09-16 18:35:30 | INFO | train rank=0 | step=49950/50000 progress=99.9% loss=0.110681 lr=2.733960e-10 lr_backbone=2.733960e-10 elapsed=52:05:08 avg_step=3.75s recent_step=3.21s eta=00:03:07 +2026-09-16 18:35:56 | INFO | train rank=0 | step=49960/50000 progress=99.9% loss=0.102438 lr=1.749735e-10 lr_backbone=1.749735e-10 elapsed=52:05:34 avg_step=3.75s recent_step=2.61s eta=00:02:30 +2026-09-16 18:36:23 | INFO | train rank=0 | step=49970/50000 progress=99.9% loss=0.080185 lr=9.842262e-11 lr_backbone=9.842262e-11 elapsed=52:06:01 avg_step=3.75s recent_step=2.69s eta=00:01:52 +2026-09-16 18:36:47 | INFO | train rank=0 | step=49980/50000 progress=100.0% loss=0.101819 lr=4.374339e-11 lr_backbone=4.374339e-11 elapsed=52:06:25 avg_step=3.75s recent_step=2.45s eta=00:01:15 +2026-09-16 18:37:19 | INFO | train rank=0 | step=49990/50000 progress=100.0% loss=0.075504 lr=1.093585e-11 lr_backbone=1.093585e-11 elapsed=52:06:57 avg_step=3.75s recent_step=3.15s eta=00:00:37 +2026-09-16 18:37:47 | INFO | train rank=0 | step=50000/50000 progress=100.0% loss=0.090104 lr=0.000000e+00 lr_backbone=0.000000e+00 elapsed=52:07:25 avg_step=3.75s recent_step=2.79s eta=00:00:00 +2026-09-16 18:38:29 | INFO | train rank=0 | Saved DeepSpeed checkpoint to /data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_50000 +2026-09-16 18:38:29 | INFO | train rank=0 | Finished training final_step=50000/50000 elapsed=52:08:07 avg_step=3.75s +2026-09-16 18:38:29 | INFO | train rank=0 | Training finished +2026-09-16 18:39:03 | INFO | train rank=0 | Saved final DeepSpeed checkpoint to /data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16/ds_checkpoint_final diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..c1085572878f587ae35377b1c9673cb24665dc6b --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fb57463db6681ab559a9e0b030bd157c04abf57e5bfacb192e148b78acf23199 +size 4033352816 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..19627f7bd38f8d0e4f57254a0cb15c1543864096 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:89c83da0743e8ce8f22fbed0244ad21b3dbcadda98dd3cc00f39606dcec73b89 +size 4033337776 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..7b6872b85d681f6aa9d2aa62fe4f4c5bbcc574e1 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e10e86185e718185399b9dbd6a036b336e02cab90c6acd482717d45fbe51b54 +size 4033334576 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..01e75acdb23441284c1bdf3fd45266a883ba64f4 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4511211dee6519f0c3150e5a8ea9e49a8d06ab26c52c4695bd89e67ca2678c95 +size 4033337712 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..586a62c16c13fe6ed7b72a1e32e0bc8394d0f25e --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b71e24b4619369a89a79eb6fb83a992dcf1d87dd111ececc6f4114232050f6bd +size 4033337904 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..9820740733f69fcdbe0df5be57d3c50cb1bc7094 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ff7e6813e02e60ae4b967a3a9fd846639287aa5c49109cc7c59f8e12430fc765 +size 4033338800 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..cff71bec06ba329ffb958a9fee4ef3638ab14f94 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3b6dd680795df187258a98e7aed6c12f110ce1cd5cd0a8f22e5c2863bf5e44db +size 4033338160 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..09c0d3f7dcc7e518b4de0895ea20d4d2eb113ac9 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:df0d6d89874ce952498ade8b1a15954fc5e0f52d54c780c822b85e951d333eb7 +size 4033337712 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/mp_rank_00_model_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..d9b9d9c494ce26bdd8f625c3b1934db25860eaaa --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/global_step50000/mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2e15cb0a8696a86b003201a9d5776156e3c72651e5d14a25285276fb2427beb3 +size 5378167768 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/latest b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/latest new file mode 100644 index 0000000000000000000000000000000000000000..59a558c0ee32a055bd8fc16bf0d99a5950c59ac1 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/latest @@ -0,0 +1 @@ +global_step50000 \ No newline at end of file diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/zero_to_fp32.py b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/zero_to_fp32.py new file mode 100644 index 0000000000000000000000000000000000000000..f40f1f0acbbff30e8e8d29906b56aae4e8b8a138 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000/zero_to_fp32.py @@ -0,0 +1,790 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: +# python zero_to_fp32.py . output_dir/ +# or +# python zero_to_fp32.py . output_dir/ --safe_serialization + +import argparse +import torch +import glob +import math +import os +import re +import gc +import json +import numpy as np +from tqdm import tqdm +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS, AUTOEP_LAYERS_KEY, + AUTOEP_LAYERS_KEY_LEGACY, AUTOEP_ZERO3_EXPERT_STATE_FORMAT_KEY, + AUTOEP_ZERO3_PARTITIONED_EXPERT_STATE_FORMAT) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def _has_autoep_zero3_partitioned_metadata(state_dict): + autoep_layers = state_dict.get(AUTOEP_LAYERS_KEY) + if autoep_layers is None: + autoep_layers = state_dict.get(AUTOEP_LAYERS_KEY_LEGACY) + if not isinstance(autoep_layers, list): + return False + return any( + isinstance(entry, dict) + and entry.get(AUTOEP_ZERO3_EXPERT_STATE_FORMAT_KEY) == AUTOEP_ZERO3_PARTITIONED_EXPERT_STATE_FORMAT + for entry in autoep_layers) + + +def _raise_if_autoep_zero3_partitioned_state(state_dict): + if _has_autoep_zero3_partitioned_metadata(state_dict): + raise NotImplementedError("zero_to_fp32 does not support AutoEP ZeRO-3 partition-native checkpoints. " + "AutoEP expert parameters are partitioned over expert replica groups, so " + "global data-parallel consolidation would produce incomplete expert tensors. " + "Use ds_to_universal.py for expert-aware conversion.") + + +def _raise_if_autoep_zero3_partitioned_checkpoint(model_files): + for file in model_files: + state_dict = torch.load(file, map_location=device, weights_only=False) + _raise_if_autoep_zero3_partitioned_state(state_dict) + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device, weights_only=False) + _raise_if_autoep_zero3_partitioned_state(state_dict) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + total_files = len(files) + state_dicts = [] + for f in tqdm(files, desc='Loading checkpoint shards'): + state_dict = torch.load(f, map_location=device, mmap=True, weights_only=False) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if ZERO_STAGE not in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + # parse_model_states rejects AutoEP ZeRO-3 partition-native checkpoints + # before the expensive optimizer-shard load below. + model_files = get_model_state_files(ds_checkpoint_dir) + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +class GatheredTensor: + """ + A pseudo tensor that collects partitioned weights. + It is more memory efficient when there are multiple groups. + """ + + def __init__(self, flat_groups, flat_groups_offset, offset, partitioned_numel, shape): + self.flat_groups = flat_groups + self.flat_groups_offset = flat_groups_offset + self.offset = offset + self.partitioned_numel = partitioned_numel + self.shape = shape + self.dtype = self.flat_groups[0][0].dtype + + def contiguous(self): + """ + Merge partitioned weights from flat_groups into a single tensor. + """ + end_idx = self.offset + self.partitioned_numel + world_size = len(self.flat_groups) + pad_flat_param_chunks = [] + + for rank_i in range(world_size): + # for each rank, we need to collect weights from related group/groups + flat_groups_at_rank_i = self.flat_groups[rank_i] + start_group_id = None + end_group_id = None + for group_id in range(len(self.flat_groups_offset)): + if self.flat_groups_offset[group_id] <= self.offset < self.flat_groups_offset[group_id + 1]: + start_group_id = group_id + if self.flat_groups_offset[group_id] < end_idx <= self.flat_groups_offset[group_id + 1]: + end_group_id = group_id + break + # collect weights from related group/groups + for group_id in range(start_group_id, end_group_id + 1): + flat_tensor = flat_groups_at_rank_i[group_id] + start_offset = self.offset - self.flat_groups_offset[group_id] + end_offset = min(end_idx, self.flat_groups_offset[group_id + 1]) - self.flat_groups_offset[group_id] + pad_flat_param_chunks.append(flat_tensor[start_offset:end_offset]) + + # collect weights from all ranks + pad_flat_param = torch.cat(pad_flat_param_chunks, dim=0) + param = pad_flat_param[:self.shape.numel()].view(self.shape).contiguous() + return param + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = sum([flat_group.numel() for flat_group in fp32_flat_groups[0]]) * world_size + + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + flat_groups_offset = [0] + list(np.cumsum([flat_tensor.numel() for flat_tensor in fp32_flat_groups[0]])) + for name, shape in tqdm(param_shapes.items(), desc='Gathering sharded weights'): + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # memory efficient tensor + tensor = GatheredTensor(fp32_flat_groups, flat_groups_offset, offset, partitioned_numel, shape) + state_dict[name] = tensor + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def to_torch_tensor(state_dict, return_empty_tensor=False): + """ + Convert state_dict of GatheredTensor to torch tensor + """ + torch_state_dict = {} + converted_tensors = {} + for name, tensor in state_dict.items(): + tensor_id = id(tensor) + if tensor_id in converted_tensors: # shared tensors + shared_tensor = torch_state_dict[converted_tensors[tensor_id]] + torch_state_dict[name] = shared_tensor + else: + converted_tensors[tensor_id] = name + if return_empty_tensor: + torch_state_dict[name] = torch.empty(tensor.shape, dtype=tensor.dtype) + else: + torch_state_dict[name] = tensor.contiguous() + return torch_state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, + tag=None, + exclude_frozen_parameters=False, + lazy_mode=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + - ``lazy_mode``: get state_dict in lazy mode. It returns a dict of pesduo tensor instead of torch tensor, which is more memory efficient. + Convert the pesduo tensor to torch tensor by ``.contiguous()`` + + Returns: + - pytorch ``state_dict`` + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + Note: the above usage may not work if your application doesn't have sufficient free CPU memory. + You may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. Or you can load state_dict in lazy mode :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, lazy_mode=True) # not on cpu + for name, lazy_tensor in state_dict.item(): + tensor = lazy_tensor.contiguous() # to cpu + print(name, tensor) + # del tensor to release memory if it no longer in use + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + state_dict = _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + if lazy_mode: + return state_dict + else: + return to_torch_tensor(state_dict) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, + output_dir, + max_shard_size="5GB", + safe_serialization=False, + tag=None, + exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_dir``: directory to the pytorch fp32 state_dict output files + - ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB + - ``safe_serialization``: whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`). + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + # Dependency pre-check + if safe_serialization: + try: + from safetensors.torch import save_file + except ImportError: + print('If you want to use `safe_serialization`, please `pip install safetensors`') + raise + if max_shard_size is not None: + try: + from huggingface_hub import split_torch_state_dict_into_shards + except ImportError: + print('If you want to use `max_shard_size`, please `pip install huggingface_hub`') + raise + + # Convert zero checkpoint to state_dict + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, + tag, + exclude_frozen_parameters, + lazy_mode=True) + + # Shard the model if it is too big. + weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin" + if max_shard_size is not None: + filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors") + # an memory-efficient approach for sharding + empty_state_dict = to_torch_tensor(state_dict, return_empty_tensor=True) + state_dict_split = split_torch_state_dict_into_shards(empty_state_dict, + filename_pattern=filename_pattern, + max_shard_size=max_shard_size) + else: + from collections import namedtuple + StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"]) + state_dict_split = StateDictSplit(is_sharded=False, + filename_to_tensors={weights_name: list(state_dict.keys())}) + + # Save the model by shard + os.makedirs(output_dir, exist_ok=True) + filename_to_tensors = state_dict_split.filename_to_tensors.items() + for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"): + shard_state_dict = {tensor_name: state_dict[tensor_name] for tensor_name in tensors} + shard_state_dict = to_torch_tensor(shard_state_dict) + output_path = os.path.join(output_dir, shard_file) + if safe_serialization: + save_file(shard_state_dict, output_path, metadata={"format": "pt"}) + else: + torch.save(shard_state_dict, output_path) + # release the memory of current shard + for tensor_name in list(shard_state_dict.keys()): + del state_dict[tensor_name] + del shard_state_dict[tensor_name] + del shard_state_dict + gc.collect() + + # Save index if sharded + if state_dict_split.is_sharded: + index = { + "metadata": state_dict_split.metadata, + "weight_map": state_dict_split.tensor_to_filename, + } + save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json" + save_index_file = os.path.join(output_dir, save_index_file) + with open(save_index_file, "w", encoding="utf-8") as f: + content = json.dumps(index, indent=2, sort_keys=True) + "\n" + f.write(content) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info("Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info("Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument("output_dir", + type=str, + help="directory to the pytorch fp32 state_dict output files" + "(e.g. path/checkpoint-12-output/)") + parser.add_argument( + "--max_shard_size", + type=str, + default="5GB", + help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size" + "lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`" + "We default it to 5GB in order for models to be able to run easily on free-tier google colab instances" + "without CPU OOM issues.") + parser.add_argument( + "--safe_serialization", + default=False, + action='store_true', + help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_dir, + max_shard_size=args.max_shard_size, + safe_serialization=args.safe_serialization, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..c1085572878f587ae35377b1c9673cb24665dc6b --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fb57463db6681ab559a9e0b030bd157c04abf57e5bfacb192e148b78acf23199 +size 4033352816 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..19627f7bd38f8d0e4f57254a0cb15c1543864096 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:89c83da0743e8ce8f22fbed0244ad21b3dbcadda98dd3cc00f39606dcec73b89 +size 4033337776 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..7b6872b85d681f6aa9d2aa62fe4f4c5bbcc574e1 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e10e86185e718185399b9dbd6a036b336e02cab90c6acd482717d45fbe51b54 +size 4033334576 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..01e75acdb23441284c1bdf3fd45266a883ba64f4 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4511211dee6519f0c3150e5a8ea9e49a8d06ab26c52c4695bd89e67ca2678c95 +size 4033337712 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..586a62c16c13fe6ed7b72a1e32e0bc8394d0f25e --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b71e24b4619369a89a79eb6fb83a992dcf1d87dd111ececc6f4114232050f6bd +size 4033337904 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..9820740733f69fcdbe0df5be57d3c50cb1bc7094 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ff7e6813e02e60ae4b967a3a9fd846639287aa5c49109cc7c59f8e12430fc765 +size 4033338800 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..cff71bec06ba329ffb958a9fee4ef3638ab14f94 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3b6dd680795df187258a98e7aed6c12f110ce1cd5cd0a8f22e5c2863bf5e44db +size 4033338160 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..09c0d3f7dcc7e518b4de0895ea20d4d2eb113ac9 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:df0d6d89874ce952498ade8b1a15954fc5e0f52d54c780c822b85e951d333eb7 +size 4033337712 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/mp_rank_00_model_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..d9b9d9c494ce26bdd8f625c3b1934db25860eaaa --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/global_step50000/mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2e15cb0a8696a86b003201a9d5776156e3c72651e5d14a25285276fb2427beb3 +size 5378167768 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/latest b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/latest new file mode 100644 index 0000000000000000000000000000000000000000..59a558c0ee32a055bd8fc16bf0d99a5950c59ac1 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/latest @@ -0,0 +1 @@ +global_step50000 \ No newline at end of file diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/zero_to_fp32.py b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/zero_to_fp32.py new file mode 100644 index 0000000000000000000000000000000000000000..f40f1f0acbbff30e8e8d29906b56aae4e8b8a138 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final/zero_to_fp32.py @@ -0,0 +1,790 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: +# python zero_to_fp32.py . output_dir/ +# or +# python zero_to_fp32.py . output_dir/ --safe_serialization + +import argparse +import torch +import glob +import math +import os +import re +import gc +import json +import numpy as np +from tqdm import tqdm +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS, AUTOEP_LAYERS_KEY, + AUTOEP_LAYERS_KEY_LEGACY, AUTOEP_ZERO3_EXPERT_STATE_FORMAT_KEY, + AUTOEP_ZERO3_PARTITIONED_EXPERT_STATE_FORMAT) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def _has_autoep_zero3_partitioned_metadata(state_dict): + autoep_layers = state_dict.get(AUTOEP_LAYERS_KEY) + if autoep_layers is None: + autoep_layers = state_dict.get(AUTOEP_LAYERS_KEY_LEGACY) + if not isinstance(autoep_layers, list): + return False + return any( + isinstance(entry, dict) + and entry.get(AUTOEP_ZERO3_EXPERT_STATE_FORMAT_KEY) == AUTOEP_ZERO3_PARTITIONED_EXPERT_STATE_FORMAT + for entry in autoep_layers) + + +def _raise_if_autoep_zero3_partitioned_state(state_dict): + if _has_autoep_zero3_partitioned_metadata(state_dict): + raise NotImplementedError("zero_to_fp32 does not support AutoEP ZeRO-3 partition-native checkpoints. " + "AutoEP expert parameters are partitioned over expert replica groups, so " + "global data-parallel consolidation would produce incomplete expert tensors. " + "Use ds_to_universal.py for expert-aware conversion.") + + +def _raise_if_autoep_zero3_partitioned_checkpoint(model_files): + for file in model_files: + state_dict = torch.load(file, map_location=device, weights_only=False) + _raise_if_autoep_zero3_partitioned_state(state_dict) + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device, weights_only=False) + _raise_if_autoep_zero3_partitioned_state(state_dict) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + total_files = len(files) + state_dicts = [] + for f in tqdm(files, desc='Loading checkpoint shards'): + state_dict = torch.load(f, map_location=device, mmap=True, weights_only=False) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if ZERO_STAGE not in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + # parse_model_states rejects AutoEP ZeRO-3 partition-native checkpoints + # before the expensive optimizer-shard load below. + model_files = get_model_state_files(ds_checkpoint_dir) + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +class GatheredTensor: + """ + A pseudo tensor that collects partitioned weights. + It is more memory efficient when there are multiple groups. + """ + + def __init__(self, flat_groups, flat_groups_offset, offset, partitioned_numel, shape): + self.flat_groups = flat_groups + self.flat_groups_offset = flat_groups_offset + self.offset = offset + self.partitioned_numel = partitioned_numel + self.shape = shape + self.dtype = self.flat_groups[0][0].dtype + + def contiguous(self): + """ + Merge partitioned weights from flat_groups into a single tensor. + """ + end_idx = self.offset + self.partitioned_numel + world_size = len(self.flat_groups) + pad_flat_param_chunks = [] + + for rank_i in range(world_size): + # for each rank, we need to collect weights from related group/groups + flat_groups_at_rank_i = self.flat_groups[rank_i] + start_group_id = None + end_group_id = None + for group_id in range(len(self.flat_groups_offset)): + if self.flat_groups_offset[group_id] <= self.offset < self.flat_groups_offset[group_id + 1]: + start_group_id = group_id + if self.flat_groups_offset[group_id] < end_idx <= self.flat_groups_offset[group_id + 1]: + end_group_id = group_id + break + # collect weights from related group/groups + for group_id in range(start_group_id, end_group_id + 1): + flat_tensor = flat_groups_at_rank_i[group_id] + start_offset = self.offset - self.flat_groups_offset[group_id] + end_offset = min(end_idx, self.flat_groups_offset[group_id + 1]) - self.flat_groups_offset[group_id] + pad_flat_param_chunks.append(flat_tensor[start_offset:end_offset]) + + # collect weights from all ranks + pad_flat_param = torch.cat(pad_flat_param_chunks, dim=0) + param = pad_flat_param[:self.shape.numel()].view(self.shape).contiguous() + return param + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = sum([flat_group.numel() for flat_group in fp32_flat_groups[0]]) * world_size + + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + flat_groups_offset = [0] + list(np.cumsum([flat_tensor.numel() for flat_tensor in fp32_flat_groups[0]])) + for name, shape in tqdm(param_shapes.items(), desc='Gathering sharded weights'): + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # memory efficient tensor + tensor = GatheredTensor(fp32_flat_groups, flat_groups_offset, offset, partitioned_numel, shape) + state_dict[name] = tensor + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def to_torch_tensor(state_dict, return_empty_tensor=False): + """ + Convert state_dict of GatheredTensor to torch tensor + """ + torch_state_dict = {} + converted_tensors = {} + for name, tensor in state_dict.items(): + tensor_id = id(tensor) + if tensor_id in converted_tensors: # shared tensors + shared_tensor = torch_state_dict[converted_tensors[tensor_id]] + torch_state_dict[name] = shared_tensor + else: + converted_tensors[tensor_id] = name + if return_empty_tensor: + torch_state_dict[name] = torch.empty(tensor.shape, dtype=tensor.dtype) + else: + torch_state_dict[name] = tensor.contiguous() + return torch_state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, + tag=None, + exclude_frozen_parameters=False, + lazy_mode=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + - ``lazy_mode``: get state_dict in lazy mode. It returns a dict of pesduo tensor instead of torch tensor, which is more memory efficient. + Convert the pesduo tensor to torch tensor by ``.contiguous()`` + + Returns: + - pytorch ``state_dict`` + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + Note: the above usage may not work if your application doesn't have sufficient free CPU memory. + You may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. Or you can load state_dict in lazy mode :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, lazy_mode=True) # not on cpu + for name, lazy_tensor in state_dict.item(): + tensor = lazy_tensor.contiguous() # to cpu + print(name, tensor) + # del tensor to release memory if it no longer in use + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + state_dict = _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + if lazy_mode: + return state_dict + else: + return to_torch_tensor(state_dict) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, + output_dir, + max_shard_size="5GB", + safe_serialization=False, + tag=None, + exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_dir``: directory to the pytorch fp32 state_dict output files + - ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB + - ``safe_serialization``: whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`). + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + # Dependency pre-check + if safe_serialization: + try: + from safetensors.torch import save_file + except ImportError: + print('If you want to use `safe_serialization`, please `pip install safetensors`') + raise + if max_shard_size is not None: + try: + from huggingface_hub import split_torch_state_dict_into_shards + except ImportError: + print('If you want to use `max_shard_size`, please `pip install huggingface_hub`') + raise + + # Convert zero checkpoint to state_dict + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, + tag, + exclude_frozen_parameters, + lazy_mode=True) + + # Shard the model if it is too big. + weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin" + if max_shard_size is not None: + filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors") + # an memory-efficient approach for sharding + empty_state_dict = to_torch_tensor(state_dict, return_empty_tensor=True) + state_dict_split = split_torch_state_dict_into_shards(empty_state_dict, + filename_pattern=filename_pattern, + max_shard_size=max_shard_size) + else: + from collections import namedtuple + StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"]) + state_dict_split = StateDictSplit(is_sharded=False, + filename_to_tensors={weights_name: list(state_dict.keys())}) + + # Save the model by shard + os.makedirs(output_dir, exist_ok=True) + filename_to_tensors = state_dict_split.filename_to_tensors.items() + for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"): + shard_state_dict = {tensor_name: state_dict[tensor_name] for tensor_name in tensors} + shard_state_dict = to_torch_tensor(shard_state_dict) + output_path = os.path.join(output_dir, shard_file) + if safe_serialization: + save_file(shard_state_dict, output_path, metadata={"format": "pt"}) + else: + torch.save(shard_state_dict, output_path) + # release the memory of current shard + for tensor_name in list(shard_state_dict.keys()): + del state_dict[tensor_name] + del shard_state_dict[tensor_name] + del shard_state_dict + gc.collect() + + # Save index if sharded + if state_dict_split.is_sharded: + index = { + "metadata": state_dict_split.metadata, + "weight_map": state_dict_split.tensor_to_filename, + } + save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json" + save_index_file = os.path.join(output_dir, save_index_file) + with open(save_index_file, "w", encoding="utf-8") as f: + content = json.dumps(index, indent=2, sort_keys=True) + "\n" + f.write(content) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info("Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info("Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument("output_dir", + type=str, + help="directory to the pytorch fp32 state_dict output files" + "(e.g. path/checkpoint-12-output/)") + parser.add_argument( + "--max_shard_size", + type=str, + default="5GB", + help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size" + "lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`" + "We default it to 5GB in order for models to be able to run easily on free-tier google colab instances" + "without CPU OOM issues.") + parser.add_argument( + "--safe_serialization", + default=False, + action='store_true', + help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_dir, + max_shard_size=args.max_shard_size, + safe_serialization=args.safe_serialization, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/train.log b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/train.log new file mode 100644 index 0000000000000000000000000000000000000000..73d2047b58a8e219e45de9e29306c90e8c113fcb --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/train.log @@ -0,0 +1,5014 @@ +2026-09-17 17:25:01 | INFO | train rank=0 | Starting training project=codebase_v8_droid_joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction save_dir=/data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction distributed=True deepspeed=True world_size=8 device=cuda:0 +2026-09-17 17:25:24 | INFO | train rank=0 | No pretrained checkpoint provided; training from scratch +2026-09-17 17:25:24 | INFO | train rank=0 | Initializing Droid LeRobot Dataset: path=/data/NTU_slab/draven/data/DROID fps=15.00 use_relabelled_instruction=True +2026-09-17 17:25:24 | INFO | train rank=0 | Batch config: total_batch_size=128 world_size=8 grad_acc_steps=1 per_device_batch_size=16 +2026-09-17 17:25:24 | INFO | train rank=0 | Dataset image_resize_size=320 +2026-09-17 17:25:24 | INFO | train rank=0 | Droid uses parquet-shard streaming with a per-rank shuffle buffer: batch_size=16 buffer_size=1024 num_workers=8 drop_last=True +2026-09-17 17:25:31 | INFO | train rank=0 | Droid streaming layout: parquet_shards=50 active_workers=8 worker_buffer_min=128 worker_buffer_max=128 max_active_shards=2 +2026-09-17 17:25:31 | INFO | train rank=0 | Resolved Droid joint action stats from /data/NTU_slab/draven/data/DROID/meta/joint_delta_stats_h16.json: q01=[-0.3050941824913025, -0.47471314668655396, -0.3020089268684387, -0.493502140045166, -0.4844675362110138, -0.4544465243816376, -0.550822913646698] q99=[0.30148017406463623, 0.5795503258705139, 0.3001617193222046, 0.5117637515068054, 0.4750978946685791, 0.5400477647781372, 0.5597710609436035] +2026-09-17 17:25:31 | INFO | train rank=0 | Trainable params: 2,688,879,624 / 2,688,879,624 +2026-09-17 17:28:36 | INFO | train rank=0 | Starting training start_step=0 max_steps=50000 remaining_steps=50000 log_interval=10 save_interval=50000 +2026-09-17 17:33:00 | INFO | train rank=0 | step=10/50000 progress=0.0% loss=1.152303 lr=3.600000e-07 lr_backbone=3.600000e-07 elapsed=00:04:23 avg_step=26.35s recent_step=26.35s eta=365:54:46 +2026-09-17 17:33:26 | INFO | train rank=0 | step=20/50000 progress=0.0% loss=1.212590 lr=7.600000e-07 lr_backbone=7.600000e-07 elapsed=00:04:49 avg_step=14.50s recent_step=2.65s eta=201:18:14 +2026-09-17 17:33:57 | INFO | train rank=0 | step=30/50000 progress=0.1% loss=1.202617 lr=1.160000e-06 lr_backbone=1.160000e-06 elapsed=00:05:20 avg_step=10.69s recent_step=3.07s eta=148:21:37 +2026-09-17 17:34:22 | INFO | train rank=0 | step=40/50000 progress=0.1% loss=1.182447 lr=1.560000e-06 lr_backbone=1.560000e-06 elapsed=00:05:46 avg_step=8.66s recent_step=2.56s eta=120:07:26 +2026-09-17 17:35:11 | INFO | train rank=0 | step=50/50000 progress=0.1% loss=1.161436 lr=1.960000e-06 lr_backbone=1.960000e-06 elapsed=00:06:35 avg_step=7.90s recent_step=4.89s eta=109:38:11 +2026-09-17 17:35:43 | INFO | train rank=0 | step=60/50000 progress=0.1% loss=1.147813 lr=2.360000e-06 lr_backbone=2.360000e-06 elapsed=00:07:07 avg_step=7.12s recent_step=3.20s eta=98:45:04 +2026-09-17 17:36:15 | INFO | train rank=0 | step=70/50000 progress=0.1% loss=1.162725 lr=2.760000e-06 lr_backbone=2.760000e-06 elapsed=00:07:39 avg_step=6.56s recent_step=3.22s eta=91:00:02 +2026-09-17 17:36:47 | INFO | train rank=0 | step=80/50000 progress=0.2% loss=1.171709 lr=3.160000e-06 lr_backbone=3.160000e-06 elapsed=00:08:10 avg_step=6.14s recent_step=3.15s eta=85:04:22 +2026-09-17 17:37:39 | INFO | train rank=0 | step=90/50000 progress=0.2% loss=1.157530 lr=3.560000e-06 lr_backbone=3.560000e-06 elapsed=00:09:02 avg_step=6.03s recent_step=5.21s eta=83:38:17 +2026-09-17 17:38:15 | INFO | train rank=0 | step=100/50000 progress=0.2% loss=1.128714 lr=3.960000e-06 lr_backbone=3.960000e-06 elapsed=00:09:38 avg_step=5.79s recent_step=3.58s eta=80:13:23 +2026-09-17 17:38:44 | INFO | train rank=0 | step=110/50000 progress=0.2% loss=1.114809 lr=4.360000e-06 lr_backbone=4.360000e-06 elapsed=00:10:07 avg_step=5.52s recent_step=2.88s eta=76:32:30 +2026-09-17 17:39:13 | INFO | train rank=0 | step=120/50000 progress=0.2% loss=1.136812 lr=4.760000e-06 lr_backbone=4.760000e-06 elapsed=00:10:36 avg_step=5.30s recent_step=2.89s eta=73:28:55 +2026-09-17 17:40:03 | INFO | train rank=0 | step=130/50000 progress=0.3% loss=1.013976 lr=5.160000e-06 lr_backbone=5.160000e-06 elapsed=00:11:26 avg_step=5.28s recent_step=5.04s eta=73:11:05 +2026-09-17 17:40:36 | INFO | train rank=0 | step=140/50000 progress=0.3% loss=1.045131 lr=5.560000e-06 lr_backbone=5.560000e-06 elapsed=00:11:59 avg_step=5.14s recent_step=3.29s eta=71:11:55 +2026-09-17 17:41:10 | INFO | train rank=0 | step=150/50000 progress=0.3% loss=1.039700 lr=5.960000e-06 lr_backbone=5.960000e-06 elapsed=00:12:33 avg_step=5.03s recent_step=3.42s eta=69:35:35 +2026-09-17 17:41:39 | INFO | train rank=0 | step=160/50000 progress=0.3% loss=1.004089 lr=6.360000e-06 lr_backbone=6.360000e-06 elapsed=00:13:02 avg_step=4.89s recent_step=2.85s eta=67:41:57 +2026-09-17 17:42:23 | INFO | train rank=0 | step=170/50000 progress=0.3% loss=0.967818 lr=6.760000e-06 lr_backbone=6.760000e-06 elapsed=00:13:46 avg_step=4.86s recent_step=4.44s eta=67:19:17 +2026-09-17 17:42:56 | INFO | train rank=0 | step=180/50000 progress=0.4% loss=0.910042 lr=7.160000e-06 lr_backbone=7.160000e-06 elapsed=00:14:20 avg_step=4.78s recent_step=3.33s eta=66:07:57 +2026-09-17 17:43:26 | INFO | train rank=0 | step=190/50000 progress=0.4% loss=0.892380 lr=7.560000e-06 lr_backbone=7.560000e-06 elapsed=00:14:49 avg_step=4.68s recent_step=2.97s eta=64:48:02 +2026-09-17 17:44:00 | INFO | train rank=0 | step=200/50000 progress=0.4% loss=0.949434 lr=7.960000e-06 lr_backbone=7.960000e-06 elapsed=00:15:23 avg_step=4.62s recent_step=3.39s eta=63:53:35 +2026-09-17 17:44:52 | INFO | train rank=0 | step=210/50000 progress=0.4% loss=0.907690 lr=8.360000e-06 lr_backbone=8.360000e-06 elapsed=00:16:15 avg_step=4.65s recent_step=5.17s eta=64:14:39 +2026-09-17 17:45:12 | INFO | train rank=0 | step=220/50000 progress=0.4% loss=0.721573 lr=8.760000e-06 lr_backbone=8.760000e-06 elapsed=00:16:35 avg_step=4.53s recent_step=2.05s eta=62:35:59 +2026-09-17 17:45:42 | INFO | train rank=0 | step=230/50000 progress=0.5% loss=0.762826 lr=9.160000e-06 lr_backbone=9.160000e-06 elapsed=00:17:05 avg_step=4.46s recent_step=2.95s eta=61:38:11 +2026-09-17 17:46:13 | INFO | train rank=0 | step=240/50000 progress=0.5% loss=0.660649 lr=9.560000e-06 lr_backbone=9.560000e-06 elapsed=00:17:36 avg_step=4.40s recent_step=3.10s eta=60:50:26 +2026-09-17 17:47:08 | INFO | train rank=0 | step=250/50000 progress=0.5% loss=0.589996 lr=9.960000e-06 lr_backbone=9.960000e-06 elapsed=00:18:31 avg_step=4.45s recent_step=5.54s eta=61:27:35 +2026-09-17 17:47:41 | INFO | train rank=0 | step=260/50000 progress=0.5% loss=0.504285 lr=1.036000e-05 lr_backbone=1.036000e-05 elapsed=00:19:05 avg_step=4.40s recent_step=3.34s eta=60:51:39 +2026-09-17 17:48:14 | INFO | train rank=0 | step=270/50000 progress=0.5% loss=0.521228 lr=1.076000e-05 lr_backbone=1.076000e-05 elapsed=00:19:38 avg_step=4.36s recent_step=3.30s eta=60:17:04 +2026-09-17 17:48:48 | INFO | train rank=0 | step=280/50000 progress=0.6% loss=0.388657 lr=1.116000e-05 lr_backbone=1.116000e-05 elapsed=00:20:11 avg_step=4.33s recent_step=3.32s eta=59:45:28 +2026-09-17 17:49:23 | INFO | train rank=0 | step=290/50000 progress=0.6% loss=0.372032 lr=1.156000e-05 lr_backbone=1.156000e-05 elapsed=00:20:46 avg_step=4.30s recent_step=3.54s eta=59:22:11 +2026-09-17 17:50:00 | INFO | train rank=0 | step=300/50000 progress=0.6% loss=0.362246 lr=1.196000e-05 lr_backbone=1.196000e-05 elapsed=00:21:24 avg_step=4.28s recent_step=3.72s eta=59:05:35 +2026-09-17 17:50:31 | INFO | train rank=0 | step=310/50000 progress=0.6% loss=0.346144 lr=1.236000e-05 lr_backbone=1.236000e-05 elapsed=00:21:55 avg_step=4.24s recent_step=3.09s eta=58:33:06 +2026-09-17 17:51:03 | INFO | train rank=0 | step=320/50000 progress=0.6% loss=0.412824 lr=1.276000e-05 lr_backbone=1.276000e-05 elapsed=00:22:26 avg_step=4.21s recent_step=3.19s eta=58:05:09 +2026-09-17 17:51:39 | INFO | train rank=0 | step=330/50000 progress=0.7% loss=0.334619 lr=1.316000e-05 lr_backbone=1.316000e-05 elapsed=00:23:03 avg_step=4.19s recent_step=3.61s eta=57:49:24 +2026-09-17 17:52:22 | INFO | train rank=0 | step=340/50000 progress=0.7% loss=0.381252 lr=1.356000e-05 lr_backbone=1.356000e-05 elapsed=00:23:45 avg_step=4.19s recent_step=4.27s eta=57:50:39 +2026-09-17 17:52:53 | INFO | train rank=0 | step=350/50000 progress=0.7% loss=0.260806 lr=1.396000e-05 lr_backbone=1.396000e-05 elapsed=00:24:16 avg_step=4.16s recent_step=3.08s eta=57:23:35 +2026-09-17 17:53:27 | INFO | train rank=0 | step=360/50000 progress=0.7% loss=0.292349 lr=1.436000e-05 lr_backbone=1.436000e-05 elapsed=00:24:50 avg_step=4.14s recent_step=3.42s eta=57:05:48 +2026-09-17 17:53:57 | INFO | train rank=0 | step=370/50000 progress=0.7% loss=0.331474 lr=1.476000e-05 lr_backbone=1.476000e-05 elapsed=00:25:21 avg_step=4.11s recent_step=3.04s eta=56:40:26 +2026-09-17 17:54:46 | INFO | train rank=0 | step=380/50000 progress=0.8% loss=0.267736 lr=1.516000e-05 lr_backbone=1.516000e-05 elapsed=00:26:09 avg_step=4.13s recent_step=4.86s eta=56:56:07 +2026-09-17 17:55:13 | INFO | train rank=0 | step=390/50000 progress=0.8% loss=0.325171 lr=1.556000e-05 lr_backbone=1.556000e-05 elapsed=00:26:36 avg_step=4.09s recent_step=2.71s eta=56:25:25 +2026-09-17 17:55:40 | INFO | train rank=0 | step=400/50000 progress=0.8% loss=0.275442 lr=1.596000e-05 lr_backbone=1.596000e-05 elapsed=00:27:03 avg_step=4.06s recent_step=2.65s eta=55:54:52 +2026-09-17 17:56:28 | INFO | train rank=0 | step=410/50000 progress=0.8% loss=0.301020 lr=1.636000e-05 lr_backbone=1.636000e-05 elapsed=00:27:51 avg_step=4.08s recent_step=4.83s eta=56:09:39 +2026-09-17 17:57:02 | INFO | train rank=0 | step=420/50000 progress=0.8% loss=0.280966 lr=1.676000e-05 lr_backbone=1.676000e-05 elapsed=00:28:26 avg_step=4.06s recent_step=3.45s eta=55:56:41 +2026-09-17 17:57:36 | INFO | train rank=0 | step=430/50000 progress=0.9% loss=0.280025 lr=1.716000e-05 lr_backbone=1.716000e-05 elapsed=00:28:59 avg_step=4.05s recent_step=3.33s eta=55:41:54 +2026-09-17 17:58:04 | INFO | train rank=0 | step=440/50000 progress=0.9% loss=0.228260 lr=1.756000e-05 lr_backbone=1.756000e-05 elapsed=00:29:27 avg_step=4.02s recent_step=2.80s eta=55:17:47 +2026-09-17 17:58:59 | INFO | train rank=0 | step=450/50000 progress=0.9% loss=0.200583 lr=1.796000e-05 lr_backbone=1.796000e-05 elapsed=00:30:23 avg_step=4.05s recent_step=5.59s eta=55:45:57 +2026-09-17 17:59:24 | INFO | train rank=0 | step=460/50000 progress=0.9% loss=0.176279 lr=1.836000e-05 lr_backbone=1.836000e-05 elapsed=00:30:47 avg_step=4.02s recent_step=2.46s eta=55:16:46 +2026-09-17 17:59:55 | INFO | train rank=0 | step=470/50000 progress=0.9% loss=0.229099 lr=1.876000e-05 lr_backbone=1.876000e-05 elapsed=00:31:18 avg_step=4.00s recent_step=3.08s eta=54:59:41 +2026-09-17 18:00:25 | INFO | train rank=0 | step=480/50000 progress=1.0% loss=0.217600 lr=1.916000e-05 lr_backbone=1.916000e-05 elapsed=00:31:49 avg_step=3.98s recent_step=3.03s eta=54:42:26 +2026-09-17 18:01:15 | INFO | train rank=0 | step=490/50000 progress=1.0% loss=0.239735 lr=1.956000e-05 lr_backbone=1.956000e-05 elapsed=00:32:38 avg_step=4.00s recent_step=4.98s eta=54:58:36 +2026-09-17 18:01:45 | INFO | train rank=0 | step=500/50000 progress=1.0% loss=0.186677 lr=1.996000e-05 lr_backbone=1.996000e-05 elapsed=00:33:08 avg_step=3.98s recent_step=2.98s eta=54:41:04 +2026-09-17 18:02:11 | INFO | train rank=0 | step=510/50000 progress=1.0% loss=0.176610 lr=2.036000e-05 lr_backbone=2.036000e-05 elapsed=00:33:34 avg_step=3.95s recent_step=2.59s eta=54:17:55 +2026-09-17 18:02:44 | INFO | train rank=0 | step=520/50000 progress=1.0% loss=0.242226 lr=2.076000e-05 lr_backbone=2.076000e-05 elapsed=00:34:08 avg_step=3.94s recent_step=3.39s eta=54:08:22 +2026-09-17 18:03:35 | INFO | train rank=0 | step=530/50000 progress=1.1% loss=0.195185 lr=2.116000e-05 lr_backbone=2.116000e-05 elapsed=00:34:59 avg_step=3.96s recent_step=5.08s eta=54:25:29 +2026-09-17 18:04:06 | INFO | train rank=0 | step=540/50000 progress=1.1% loss=0.186095 lr=2.156000e-05 lr_backbone=2.156000e-05 elapsed=00:35:29 avg_step=3.94s recent_step=3.08s eta=54:11:24 +2026-09-17 18:04:35 | INFO | train rank=0 | step=550/50000 progress=1.1% loss=0.195368 lr=2.196000e-05 lr_backbone=2.196000e-05 elapsed=00:35:58 avg_step=3.93s recent_step=2.89s eta=53:54:53 +2026-09-17 18:05:09 | INFO | train rank=0 | step=560/50000 progress=1.1% loss=0.235826 lr=2.236000e-05 lr_backbone=2.236000e-05 elapsed=00:36:33 avg_step=3.92s recent_step=3.43s eta=53:47:00 +2026-09-17 18:05:52 | INFO | train rank=0 | step=570/50000 progress=1.1% loss=0.212799 lr=2.276000e-05 lr_backbone=2.276000e-05 elapsed=00:37:15 avg_step=3.92s recent_step=4.26s eta=53:51:14 +2026-09-17 18:06:23 | INFO | train rank=0 | step=580/50000 progress=1.2% loss=0.198607 lr=2.316000e-05 lr_backbone=2.316000e-05 elapsed=00:37:46 avg_step=3.91s recent_step=3.06s eta=53:38:25 +2026-09-17 18:06:52 | INFO | train rank=0 | step=590/50000 progress=1.2% loss=0.242121 lr=2.356000e-05 lr_backbone=2.356000e-05 elapsed=00:38:15 avg_step=3.89s recent_step=2.92s eta=53:24:00 +2026-09-17 18:07:33 | INFO | train rank=0 | step=600/50000 progress=1.2% loss=0.184807 lr=2.396000e-05 lr_backbone=2.396000e-05 elapsed=00:38:56 avg_step=3.89s recent_step=4.09s eta=53:26:08 +2026-09-17 18:08:04 | INFO | train rank=0 | step=610/50000 progress=1.2% loss=0.178343 lr=2.436000e-05 lr_backbone=2.436000e-05 elapsed=00:39:28 avg_step=3.88s recent_step=3.18s eta=53:15:52 +2026-09-17 18:08:37 | INFO | train rank=0 | step=620/50000 progress=1.2% loss=0.188413 lr=2.476000e-05 lr_backbone=2.476000e-05 elapsed=00:40:00 avg_step=3.87s recent_step=3.22s eta=53:06:28 +2026-09-17 18:09:05 | INFO | train rank=0 | step=630/50000 progress=1.3% loss=0.198396 lr=2.516000e-05 lr_backbone=2.516000e-05 elapsed=00:40:28 avg_step=3.85s recent_step=2.80s eta=52:51:48 +2026-09-17 18:09:49 | INFO | train rank=0 | step=640/50000 progress=1.3% loss=0.250289 lr=2.556000e-05 lr_backbone=2.556000e-05 elapsed=00:41:12 avg_step=3.86s recent_step=4.40s eta=52:58:10 +2026-09-17 18:10:22 | INFO | train rank=0 | step=650/50000 progress=1.3% loss=0.183617 lr=2.596000e-05 lr_backbone=2.596000e-05 elapsed=00:41:45 avg_step=3.86s recent_step=3.34s eta=52:50:57 +2026-09-17 18:10:52 | INFO | train rank=0 | step=660/50000 progress=1.3% loss=0.184089 lr=2.636000e-05 lr_backbone=2.636000e-05 elapsed=00:42:16 avg_step=3.84s recent_step=3.02s eta=52:39:52 +2026-09-17 18:11:27 | INFO | train rank=0 | step=670/50000 progress=1.3% loss=0.179751 lr=2.676000e-05 lr_backbone=2.676000e-05 elapsed=00:42:50 avg_step=3.84s recent_step=3.44s eta=52:34:19 +2026-09-17 18:12:17 | INFO | train rank=0 | step=680/50000 progress=1.4% loss=0.314377 lr=2.716000e-05 lr_backbone=2.716000e-05 elapsed=00:43:40 avg_step=3.85s recent_step=5.01s eta=52:47:51 +2026-09-17 18:12:50 | INFO | train rank=0 | step=690/50000 progress=1.4% loss=0.192075 lr=2.756000e-05 lr_backbone=2.756000e-05 elapsed=00:44:13 avg_step=3.85s recent_step=3.31s eta=52:40:41 +2026-09-17 18:13:23 | INFO | train rank=0 | step=700/50000 progress=1.4% loss=0.184020 lr=2.796000e-05 lr_backbone=2.796000e-05 elapsed=00:44:46 avg_step=3.84s recent_step=3.27s eta=52:33:20 +2026-09-17 18:13:54 | INFO | train rank=0 | step=710/50000 progress=1.4% loss=0.181607 lr=2.836000e-05 lr_backbone=2.836000e-05 elapsed=00:45:17 avg_step=3.83s recent_step=3.13s eta=52:24:32 +2026-09-17 18:14:41 | INFO | train rank=0 | step=720/50000 progress=1.4% loss=0.210818 lr=2.876000e-05 lr_backbone=2.876000e-05 elapsed=00:46:04 avg_step=3.84s recent_step=4.69s eta=52:33:44 +2026-09-17 18:15:19 | INFO | train rank=0 | step=730/50000 progress=1.5% loss=0.215881 lr=2.916000e-05 lr_backbone=2.916000e-05 elapsed=00:46:43 avg_step=3.84s recent_step=3.86s eta=52:33:18 +2026-09-17 18:15:48 | INFO | train rank=0 | step=740/50000 progress=1.5% loss=0.174433 lr=2.956000e-05 lr_backbone=2.956000e-05 elapsed=00:47:12 avg_step=3.83s recent_step=2.89s eta=52:22:07 +2026-09-17 18:16:15 | INFO | train rank=0 | step=750/50000 progress=1.5% loss=0.197291 lr=2.996000e-05 lr_backbone=2.996000e-05 elapsed=00:47:39 avg_step=3.81s recent_step=2.71s eta=52:09:16 +2026-09-17 18:17:05 | INFO | train rank=0 | step=760/50000 progress=1.5% loss=0.187499 lr=3.036000e-05 lr_backbone=3.036000e-05 elapsed=00:48:28 avg_step=3.83s recent_step=4.95s eta=52:20:57 +2026-09-17 18:17:36 | INFO | train rank=0 | step=770/50000 progress=1.5% loss=0.178281 lr=3.076000e-05 lr_backbone=3.076000e-05 elapsed=00:48:59 avg_step=3.82s recent_step=3.08s eta=52:12:21 +2026-09-17 18:18:08 | INFO | train rank=0 | step=780/50000 progress=1.6% loss=0.182211 lr=3.116000e-05 lr_backbone=3.116000e-05 elapsed=00:49:31 avg_step=3.81s recent_step=3.21s eta=52:05:21 +2026-09-17 18:18:41 | INFO | train rank=0 | step=790/50000 progress=1.6% loss=0.180555 lr=3.156000e-05 lr_backbone=3.156000e-05 elapsed=00:50:04 avg_step=3.80s recent_step=3.32s eta=51:59:36 +2026-09-17 18:19:23 | INFO | train rank=0 | step=800/50000 progress=1.6% loss=0.191914 lr=3.196000e-05 lr_backbone=3.196000e-05 elapsed=00:50:46 avg_step=3.81s recent_step=4.15s eta=52:02:30 +2026-09-17 18:20:04 | INFO | train rank=0 | step=810/50000 progress=1.6% loss=0.205994 lr=3.236000e-05 lr_backbone=3.236000e-05 elapsed=00:51:27 avg_step=3.81s recent_step=4.11s eta=52:04:57 +2026-09-17 18:20:34 | INFO | train rank=0 | step=820/50000 progress=1.6% loss=0.222229 lr=3.276000e-05 lr_backbone=3.276000e-05 elapsed=00:51:57 avg_step=3.80s recent_step=3.04s eta=51:56:35 +2026-09-17 18:21:07 | INFO | train rank=0 | step=830/50000 progress=1.7% loss=0.185498 lr=3.316000e-05 lr_backbone=3.316000e-05 elapsed=00:52:30 avg_step=3.80s recent_step=3.26s eta=51:50:34 +2026-09-17 18:21:51 | INFO | train rank=0 | step=840/50000 progress=1.7% loss=0.148193 lr=3.356000e-05 lr_backbone=3.356000e-05 elapsed=00:53:14 avg_step=3.80s recent_step=4.42s eta=51:55:59 +2026-09-17 18:22:24 | INFO | train rank=0 | step=850/50000 progress=1.7% loss=0.191169 lr=3.396000e-05 lr_backbone=3.396000e-05 elapsed=00:53:47 avg_step=3.80s recent_step=3.28s eta=51:50:20 +2026-09-17 18:22:53 | INFO | train rank=0 | step=860/50000 progress=1.7% loss=0.165997 lr=3.436000e-05 lr_backbone=3.436000e-05 elapsed=00:54:17 avg_step=3.79s recent_step=2.98s eta=51:41:52 +2026-09-17 18:23:30 | INFO | train rank=0 | step=870/50000 progress=1.7% loss=0.206191 lr=3.476000e-05 lr_backbone=3.476000e-05 elapsed=00:54:54 avg_step=3.79s recent_step=3.70s eta=51:40:23 +2026-09-17 18:24:08 | INFO | train rank=0 | step=880/50000 progress=1.8% loss=0.221243 lr=3.516000e-05 lr_backbone=3.516000e-05 elapsed=00:55:31 avg_step=3.79s recent_step=3.74s eta=51:39:20 +2026-09-17 18:24:38 | INFO | train rank=0 | step=890/50000 progress=1.8% loss=0.168866 lr=3.556000e-05 lr_backbone=3.556000e-05 elapsed=00:56:02 avg_step=3.78s recent_step=3.07s eta=51:32:07 +2026-09-17 18:25:06 | INFO | train rank=0 | step=900/50000 progress=1.8% loss=0.202175 lr=3.596000e-05 lr_backbone=3.596000e-05 elapsed=00:56:29 avg_step=3.77s recent_step=2.76s eta=51:22:12 +2026-09-17 18:25:47 | INFO | train rank=0 | step=910/50000 progress=1.8% loss=0.210595 lr=3.636000e-05 lr_backbone=3.636000e-05 elapsed=00:57:11 avg_step=3.77s recent_step=4.14s eta=51:24:55 +2026-09-17 18:26:24 | INFO | train rank=0 | step=920/50000 progress=1.8% loss=0.164145 lr=3.676000e-05 lr_backbone=3.676000e-05 elapsed=00:57:47 avg_step=3.77s recent_step=3.68s eta=51:23:28 +2026-09-17 18:26:50 | INFO | train rank=0 | step=930/50000 progress=1.9% loss=0.197052 lr=3.716000e-05 lr_backbone=3.716000e-05 elapsed=00:58:13 avg_step=3.76s recent_step=2.59s eta=51:12:28 +2026-09-17 18:27:26 | INFO | train rank=0 | step=940/50000 progress=1.9% loss=0.203356 lr=3.756000e-05 lr_backbone=3.756000e-05 elapsed=00:58:49 avg_step=3.75s recent_step=3.54s eta=51:09:59 +2026-09-17 18:27:57 | INFO | train rank=0 | step=950/50000 progress=1.9% loss=0.184939 lr=3.796000e-05 lr_backbone=3.796000e-05 elapsed=00:59:21 avg_step=3.75s recent_step=3.18s eta=51:04:23 +2026-09-17 18:28:51 | INFO | train rank=0 | step=960/50000 progress=1.9% loss=0.207549 lr=3.836000e-05 lr_backbone=3.836000e-05 elapsed=01:00:14 avg_step=3.76s recent_step=5.33s eta=51:17:12 +2026-09-17 18:29:27 | INFO | train rank=0 | step=970/50000 progress=1.9% loss=0.204657 lr=3.876000e-05 lr_backbone=3.876000e-05 elapsed=01:00:50 avg_step=3.76s recent_step=3.66s eta=51:15:39 +2026-09-17 18:29:55 | INFO | train rank=0 | step=980/50000 progress=2.0% loss=0.169379 lr=3.916000e-05 lr_backbone=3.916000e-05 elapsed=01:01:18 avg_step=3.75s recent_step=2.77s eta=51:06:45 +2026-09-17 18:30:25 | INFO | train rank=0 | step=990/50000 progress=2.0% loss=0.196974 lr=3.956000e-05 lr_backbone=3.956000e-05 elapsed=01:01:48 avg_step=3.75s recent_step=3.03s eta=51:00:11 +2026-09-17 18:31:20 | INFO | train rank=0 | step=1000/50000 progress=2.0% loss=0.175132 lr=3.996000e-05 lr_backbone=3.996000e-05 elapsed=01:02:43 avg_step=3.76s recent_step=5.44s eta=51:13:25 +2026-09-17 18:31:47 | INFO | train rank=0 | step=1010/50000 progress=2.0% loss=0.255150 lr=4.036000e-05 lr_backbone=4.036000e-05 elapsed=01:03:11 avg_step=3.75s recent_step=2.79s eta=51:04:53 +2026-09-17 18:32:20 | INFO | train rank=0 | step=1020/50000 progress=2.0% loss=0.187830 lr=4.076000e-05 lr_backbone=4.076000e-05 elapsed=01:03:43 avg_step=3.75s recent_step=3.26s eta=51:00:21 +2026-09-17 18:32:54 | INFO | train rank=0 | step=1030/50000 progress=2.1% loss=0.197660 lr=4.116000e-05 lr_backbone=4.116000e-05 elapsed=01:04:17 avg_step=3.75s recent_step=3.38s eta=50:56:48 +2026-09-17 18:33:40 | INFO | train rank=0 | step=1040/50000 progress=2.1% loss=0.189536 lr=4.156000e-05 lr_backbone=4.156000e-05 elapsed=01:05:03 avg_step=3.75s recent_step=4.56s eta=51:02:36 +2026-09-17 18:34:12 | INFO | train rank=0 | step=1050/50000 progress=2.1% loss=0.218911 lr=4.196000e-05 lr_backbone=4.196000e-05 elapsed=01:05:35 avg_step=3.75s recent_step=3.25s eta=50:58:05 +2026-09-17 18:34:39 | INFO | train rank=0 | step=1060/50000 progress=2.1% loss=0.224542 lr=4.236000e-05 lr_backbone=4.236000e-05 elapsed=01:06:02 avg_step=3.74s recent_step=2.71s eta=50:49:30 +2026-09-17 18:35:15 | INFO | train rank=0 | step=1070/50000 progress=2.1% loss=0.189288 lr=4.276000e-05 lr_backbone=4.276000e-05 elapsed=01:06:39 avg_step=3.74s recent_step=3.62s eta=50:47:57 +2026-09-17 18:36:02 | INFO | train rank=0 | step=1080/50000 progress=2.2% loss=0.155929 lr=4.316000e-05 lr_backbone=4.316000e-05 elapsed=01:07:25 avg_step=3.75s recent_step=4.66s eta=50:54:16 +2026-09-17 18:36:39 | INFO | train rank=0 | step=1090/50000 progress=2.2% loss=0.214973 lr=4.356000e-05 lr_backbone=4.356000e-05 elapsed=01:08:02 avg_step=3.75s recent_step=3.67s eta=50:53:06 +2026-09-17 18:37:08 | INFO | train rank=0 | step=1100/50000 progress=2.2% loss=0.220303 lr=4.396000e-05 lr_backbone=4.396000e-05 elapsed=01:08:31 avg_step=3.74s recent_step=2.95s eta=50:46:34 +2026-09-17 18:37:34 | INFO | train rank=0 | step=1110/50000 progress=2.2% loss=0.165003 lr=4.436000e-05 lr_backbone=4.436000e-05 elapsed=01:08:58 avg_step=3.73s recent_step=2.63s eta=50:37:50 +2026-09-17 18:38:24 | INFO | train rank=0 | step=1120/50000 progress=2.2% loss=0.194896 lr=4.476000e-05 lr_backbone=4.476000e-05 elapsed=01:09:47 avg_step=3.74s recent_step=4.91s eta=50:45:51 +2026-09-17 18:38:52 | INFO | train rank=0 | step=1130/50000 progress=2.3% loss=0.177864 lr=4.516000e-05 lr_backbone=4.516000e-05 elapsed=01:10:15 avg_step=3.73s recent_step=2.85s eta=50:38:48 +2026-09-17 18:39:19 | INFO | train rank=0 | step=1140/50000 progress=2.3% loss=0.168514 lr=4.556000e-05 lr_backbone=4.556000e-05 elapsed=01:10:42 avg_step=3.72s recent_step=2.64s eta=50:30:23 +2026-09-17 18:39:52 | INFO | train rank=0 | step=1150/50000 progress=2.3% loss=0.217710 lr=4.596000e-05 lr_backbone=4.596000e-05 elapsed=01:11:15 avg_step=3.72s recent_step=3.33s eta=50:27:00 +2026-09-17 18:40:43 | INFO | train rank=0 | step=1160/50000 progress=2.3% loss=0.173919 lr=4.636000e-05 lr_backbone=4.636000e-05 elapsed=01:12:07 avg_step=3.73s recent_step=5.15s eta=50:36:28 +2026-09-17 18:41:15 | INFO | train rank=0 | step=1170/50000 progress=2.3% loss=0.260542 lr=4.676000e-05 lr_backbone=4.676000e-05 elapsed=01:12:38 avg_step=3.73s recent_step=3.14s eta=50:31:46 +2026-09-17 18:41:49 | INFO | train rank=0 | step=1180/50000 progress=2.4% loss=0.182552 lr=4.716000e-05 lr_backbone=4.716000e-05 elapsed=01:13:12 avg_step=3.72s recent_step=3.37s eta=50:28:42 +2026-09-17 18:42:16 | INFO | train rank=0 | step=1190/50000 progress=2.4% loss=0.178927 lr=4.756000e-05 lr_backbone=4.756000e-05 elapsed=01:13:39 avg_step=3.71s recent_step=2.76s eta=50:21:29 +2026-09-17 18:43:04 | INFO | train rank=0 | step=1200/50000 progress=2.4% loss=0.219636 lr=4.796000e-05 lr_backbone=4.796000e-05 elapsed=01:14:27 avg_step=3.72s recent_step=4.80s eta=50:28:15 +2026-09-17 18:43:33 | INFO | train rank=0 | step=1210/50000 progress=2.4% loss=0.191223 lr=4.836000e-05 lr_backbone=4.836000e-05 elapsed=01:14:56 avg_step=3.72s recent_step=2.88s eta=50:21:59 +2026-09-17 18:44:06 | INFO | train rank=0 | step=1220/50000 progress=2.4% loss=0.174697 lr=4.876000e-05 lr_backbone=4.876000e-05 elapsed=01:15:29 avg_step=3.71s recent_step=3.30s eta=50:18:33 +2026-09-17 18:44:34 | INFO | train rank=0 | step=1230/50000 progress=2.5% loss=0.157315 lr=4.916000e-05 lr_backbone=4.916000e-05 elapsed=01:15:57 avg_step=3.71s recent_step=2.79s eta=50:11:51 +2026-09-17 18:45:27 | INFO | train rank=0 | step=1240/50000 progress=2.5% loss=0.142156 lr=4.956000e-05 lr_backbone=4.956000e-05 elapsed=01:16:51 avg_step=3.72s recent_step=5.35s eta=50:22:02 +2026-09-17 18:45:59 | INFO | train rank=0 | step=1250/50000 progress=2.5% loss=0.186425 lr=4.996000e-05 lr_backbone=4.996000e-05 elapsed=01:17:23 avg_step=3.71s recent_step=3.19s eta=50:18:00 +2026-09-17 18:46:29 | INFO | train rank=0 | step=1260/50000 progress=2.5% loss=0.168673 lr=5.036000e-05 lr_backbone=5.036000e-05 elapsed=01:17:52 avg_step=3.71s recent_step=2.95s eta=50:12:28 +2026-09-17 18:46:59 | INFO | train rank=0 | step=1270/50000 progress=2.5% loss=0.155753 lr=5.076000e-05 lr_backbone=5.076000e-05 elapsed=01:18:23 avg_step=3.70s recent_step=3.04s eta=50:07:36 +2026-09-17 18:47:41 | INFO | train rank=0 | step=1280/50000 progress=2.6% loss=0.182773 lr=5.116000e-05 lr_backbone=5.116000e-05 elapsed=01:19:04 avg_step=3.71s recent_step=4.16s eta=50:09:53 +2026-09-17 18:48:08 | INFO | train rank=0 | step=1290/50000 progress=2.6% loss=0.183963 lr=5.156000e-05 lr_backbone=5.156000e-05 elapsed=01:19:32 avg_step=3.70s recent_step=2.76s eta=50:03:17 +2026-09-17 18:48:32 | INFO | train rank=0 | step=1300/50000 progress=2.6% loss=0.208807 lr=5.196000e-05 lr_backbone=5.196000e-05 elapsed=01:19:56 avg_step=3.69s recent_step=2.39s eta=49:54:29 +2026-09-17 18:49:02 | INFO | train rank=0 | step=1310/50000 progress=2.6% loss=0.151347 lr=5.236000e-05 lr_backbone=5.236000e-05 elapsed=01:20:26 avg_step=3.68s recent_step=2.99s eta=49:49:34 +2026-09-17 18:49:50 | INFO | train rank=0 | step=1320/50000 progress=2.6% loss=0.195671 lr=5.276000e-05 lr_backbone=5.276000e-05 elapsed=01:21:13 avg_step=3.69s recent_step=4.74s eta=49:55:26 +2026-09-17 18:50:20 | INFO | train rank=0 | step=1330/50000 progress=2.7% loss=0.154617 lr=5.316000e-05 lr_backbone=5.316000e-05 elapsed=01:21:44 avg_step=3.69s recent_step=3.07s eta=49:51:01 +2026-09-17 18:50:50 | INFO | train rank=0 | step=1340/50000 progress=2.7% loss=0.163690 lr=5.356000e-05 lr_backbone=5.356000e-05 elapsed=01:22:13 avg_step=3.68s recent_step=2.95s eta=49:45:58 +2026-09-17 18:51:23 | INFO | train rank=0 | step=1350/50000 progress=2.7% loss=0.134937 lr=5.396000e-05 lr_backbone=5.396000e-05 elapsed=01:22:46 avg_step=3.68s recent_step=3.31s eta=49:43:07 +2026-09-17 18:52:12 | INFO | train rank=0 | step=1360/50000 progress=2.7% loss=0.201079 lr=5.436000e-05 lr_backbone=5.436000e-05 elapsed=01:23:35 avg_step=3.69s recent_step=4.91s eta=49:49:49 +2026-09-17 18:52:42 | INFO | train rank=0 | step=1370/50000 progress=2.7% loss=0.207254 lr=5.476000e-05 lr_backbone=5.476000e-05 elapsed=01:24:06 avg_step=3.68s recent_step=3.04s eta=49:45:24 +2026-09-17 18:53:13 | INFO | train rank=0 | step=1380/50000 progress=2.8% loss=0.175349 lr=5.516000e-05 lr_backbone=5.516000e-05 elapsed=01:24:37 avg_step=3.68s recent_step=3.08s eta=49:41:13 +2026-09-17 18:53:46 | INFO | train rank=0 | step=1390/50000 progress=2.8% loss=0.241869 lr=5.556000e-05 lr_backbone=5.556000e-05 elapsed=01:25:10 avg_step=3.68s recent_step=3.30s eta=49:38:25 +2026-09-17 18:54:34 | INFO | train rank=0 | step=1400/50000 progress=2.8% loss=0.232410 lr=5.596000e-05 lr_backbone=5.596000e-05 elapsed=01:25:58 avg_step=3.68s recent_step=4.82s eta=49:44:26 +2026-09-17 18:55:03 | INFO | train rank=0 | step=1410/50000 progress=2.8% loss=0.176427 lr=5.636000e-05 lr_backbone=5.636000e-05 elapsed=01:26:26 avg_step=3.68s recent_step=2.86s eta=49:39:03 +2026-09-17 18:55:37 | INFO | train rank=0 | step=1420/50000 progress=2.8% loss=0.139517 lr=5.676000e-05 lr_backbone=5.676000e-05 elapsed=01:27:00 avg_step=3.68s recent_step=3.35s eta=49:36:36 +2026-09-17 18:56:10 | INFO | train rank=0 | step=1430/50000 progress=2.9% loss=0.170739 lr=5.716000e-05 lr_backbone=5.716000e-05 elapsed=01:27:33 avg_step=3.67s recent_step=3.32s eta=49:33:59 +2026-09-17 18:56:50 | INFO | train rank=0 | step=1440/50000 progress=2.9% loss=0.140419 lr=5.756000e-05 lr_backbone=5.756000e-05 elapsed=01:28:13 avg_step=3.68s recent_step=4.03s eta=49:35:22 +2026-09-17 18:57:30 | INFO | train rank=0 | step=1450/50000 progress=2.9% loss=0.138787 lr=5.796000e-05 lr_backbone=5.796000e-05 elapsed=01:28:54 avg_step=3.68s recent_step=4.03s eta=49:36:45 +2026-09-17 18:58:01 | INFO | train rank=0 | step=1460/50000 progress=2.9% loss=0.241015 lr=5.836000e-05 lr_backbone=5.836000e-05 elapsed=01:29:24 avg_step=3.67s recent_step=3.05s eta=49:32:39 +2026-09-17 18:58:33 | INFO | train rank=0 | step=1470/50000 progress=2.9% loss=0.167079 lr=5.876000e-05 lr_backbone=5.876000e-05 elapsed=01:29:56 avg_step=3.67s recent_step=3.21s eta=49:29:30 +2026-09-17 18:59:08 | INFO | train rank=0 | step=1480/50000 progress=3.0% loss=0.167667 lr=5.916000e-05 lr_backbone=5.916000e-05 elapsed=01:30:31 avg_step=3.67s recent_step=3.48s eta=49:27:49 +2026-09-17 18:59:51 | INFO | train rank=0 | step=1490/50000 progress=3.0% loss=0.277899 lr=5.956000e-05 lr_backbone=5.956000e-05 elapsed=01:31:14 avg_step=3.67s recent_step=4.31s eta=49:30:42 +2026-09-17 19:00:27 | INFO | train rank=0 | step=1500/50000 progress=3.0% loss=0.162270 lr=5.996000e-05 lr_backbone=5.996000e-05 elapsed=01:31:50 avg_step=3.67s recent_step=3.57s eta=49:29:31 +2026-09-17 19:00:51 | INFO | train rank=0 | step=1510/50000 progress=3.0% loss=0.145709 lr=6.036000e-05 lr_backbone=6.036000e-05 elapsed=01:32:15 avg_step=3.67s recent_step=2.48s eta=49:22:32 +2026-09-17 19:01:41 | INFO | train rank=0 | step=1520/50000 progress=3.0% loss=0.178900 lr=6.076000e-05 lr_backbone=6.076000e-05 elapsed=01:33:05 avg_step=3.67s recent_step=4.99s eta=49:28:58 +2026-09-17 19:02:11 | INFO | train rank=0 | step=1530/50000 progress=3.1% loss=0.152592 lr=6.116000e-05 lr_backbone=6.116000e-05 elapsed=01:33:34 avg_step=3.67s recent_step=2.96s eta=49:24:35 +2026-09-17 19:02:39 | INFO | train rank=0 | step=1540/50000 progress=3.1% loss=0.166420 lr=6.156000e-05 lr_backbone=6.156000e-05 elapsed=01:34:02 avg_step=3.66s recent_step=2.80s eta=49:19:25 +2026-09-17 19:03:13 | INFO | train rank=0 | step=1550/50000 progress=3.1% loss=0.159383 lr=6.196000e-05 lr_backbone=6.196000e-05 elapsed=01:34:36 avg_step=3.66s recent_step=3.41s eta=49:17:28 +2026-09-17 19:03:48 | INFO | train rank=0 | step=1560/50000 progress=3.1% loss=0.148528 lr=6.236000e-05 lr_backbone=6.236000e-05 elapsed=01:35:11 avg_step=3.66s recent_step=3.46s eta=49:15:50 +2026-09-17 19:04:35 | INFO | train rank=0 | step=1570/50000 progress=3.1% loss=0.161499 lr=6.276000e-05 lr_backbone=6.276000e-05 elapsed=01:35:58 avg_step=3.67s recent_step=4.73s eta=49:20:43 +2026-09-17 19:04:59 | INFO | train rank=0 | step=1580/50000 progress=3.2% loss=0.183694 lr=6.316000e-05 lr_backbone=6.316000e-05 elapsed=01:36:22 avg_step=3.66s recent_step=2.36s eta=49:13:24 +2026-09-17 19:05:37 | INFO | train rank=0 | step=1590/50000 progress=3.2% loss=0.216698 lr=6.356000e-05 lr_backbone=6.356000e-05 elapsed=01:37:00 avg_step=3.66s recent_step=3.79s eta=49:13:28 +2026-09-17 19:06:09 | INFO | train rank=0 | step=1600/50000 progress=3.2% loss=0.176888 lr=6.396000e-05 lr_backbone=6.396000e-05 elapsed=01:37:32 avg_step=3.66s recent_step=3.23s eta=49:10:43 +2026-09-17 19:06:47 | INFO | train rank=0 | step=1610/50000 progress=3.2% loss=0.187387 lr=6.436000e-05 lr_backbone=6.436000e-05 elapsed=01:38:11 avg_step=3.66s recent_step=3.84s eta=49:11:00 +2026-09-17 19:07:18 | INFO | train rank=0 | step=1620/50000 progress=3.2% loss=0.160426 lr=6.476000e-05 lr_backbone=6.476000e-05 elapsed=01:38:41 avg_step=3.66s recent_step=3.08s eta=49:07:30 +2026-09-17 19:08:14 | INFO | train rank=0 | step=1630/50000 progress=3.3% loss=0.155202 lr=6.516000e-05 lr_backbone=6.516000e-05 elapsed=01:39:37 avg_step=3.67s recent_step=5.60s eta=49:16:29 +2026-09-17 19:08:47 | INFO | train rank=0 | step=1640/50000 progress=3.3% loss=0.193292 lr=6.556000e-05 lr_backbone=6.556000e-05 elapsed=01:40:10 avg_step=3.67s recent_step=3.30s eta=49:14:04 +2026-09-17 19:09:18 | INFO | train rank=0 | step=1650/50000 progress=3.3% loss=0.150409 lr=6.596000e-05 lr_backbone=6.596000e-05 elapsed=01:40:41 avg_step=3.66s recent_step=3.08s eta=49:10:36 +2026-09-17 19:09:51 | INFO | train rank=0 | step=1660/50000 progress=3.3% loss=0.185943 lr=6.636000e-05 lr_backbone=6.636000e-05 elapsed=01:41:14 avg_step=3.66s recent_step=3.34s eta=49:08:26 +2026-09-17 19:10:38 | INFO | train rank=0 | step=1670/50000 progress=3.3% loss=0.176124 lr=6.676000e-05 lr_backbone=6.676000e-05 elapsed=01:42:02 avg_step=3.67s recent_step=4.72s eta=49:12:55 +2026-09-17 19:11:11 | INFO | train rank=0 | step=1680/50000 progress=3.4% loss=0.146444 lr=6.716000e-05 lr_backbone=6.716000e-05 elapsed=01:42:35 avg_step=3.66s recent_step=3.31s eta=49:10:34 +2026-09-17 19:11:40 | INFO | train rank=0 | step=1690/50000 progress=3.4% loss=0.197024 lr=6.756000e-05 lr_backbone=6.756000e-05 elapsed=01:43:03 avg_step=3.66s recent_step=2.84s eta=49:06:03 +2026-09-17 19:12:08 | INFO | train rank=0 | step=1700/50000 progress=3.4% loss=0.150759 lr=6.796000e-05 lr_backbone=6.796000e-05 elapsed=01:43:32 avg_step=3.65s recent_step=2.87s eta=49:01:42 +2026-09-17 19:13:00 | INFO | train rank=0 | step=1710/50000 progress=3.4% loss=0.232574 lr=6.836000e-05 lr_backbone=6.836000e-05 elapsed=01:44:23 avg_step=3.66s recent_step=5.14s eta=49:08:05 +2026-09-17 19:13:34 | INFO | train rank=0 | step=1720/50000 progress=3.4% loss=0.144731 lr=6.876000e-05 lr_backbone=6.876000e-05 elapsed=01:44:57 avg_step=3.66s recent_step=3.37s eta=49:06:06 +2026-09-17 19:14:09 | INFO | train rank=0 | step=1730/50000 progress=3.5% loss=0.176126 lr=6.916000e-05 lr_backbone=6.916000e-05 elapsed=01:45:33 avg_step=3.66s recent_step=3.57s eta=49:05:03 +2026-09-17 19:14:42 | INFO | train rank=0 | step=1740/50000 progress=3.5% loss=0.195916 lr=6.956000e-05 lr_backbone=6.956000e-05 elapsed=01:46:06 avg_step=3.66s recent_step=3.31s eta=49:02:49 +2026-09-17 19:15:23 | INFO | train rank=0 | step=1750/50000 progress=3.5% loss=0.141279 lr=6.996000e-05 lr_backbone=6.996000e-05 elapsed=01:46:46 avg_step=3.66s recent_step=4.06s eta=49:04:04 +2026-09-17 19:15:52 | INFO | train rank=0 | step=1760/50000 progress=3.5% loss=0.182721 lr=7.036000e-05 lr_backbone=7.036000e-05 elapsed=01:47:15 avg_step=3.66s recent_step=2.87s eta=48:59:50 +2026-09-17 19:16:22 | INFO | train rank=0 | step=1770/50000 progress=3.5% loss=0.174526 lr=7.076000e-05 lr_backbone=7.076000e-05 elapsed=01:47:45 avg_step=3.65s recent_step=3.01s eta=48:56:18 +2026-09-17 19:17:01 | INFO | train rank=0 | step=1780/50000 progress=3.6% loss=0.173984 lr=7.116000e-05 lr_backbone=7.116000e-05 elapsed=01:48:24 avg_step=3.65s recent_step=3.91s eta=48:56:51 +2026-09-17 19:17:51 | INFO | train rank=0 | step=1790/50000 progress=3.6% loss=0.154447 lr=7.156000e-05 lr_backbone=7.156000e-05 elapsed=01:49:14 avg_step=3.66s recent_step=4.98s eta=49:02:12 +2026-09-17 19:18:16 | INFO | train rank=0 | step=1800/50000 progress=3.6% loss=0.170299 lr=7.196000e-05 lr_backbone=7.196000e-05 elapsed=01:49:39 avg_step=3.66s recent_step=2.53s eta=48:56:32 +2026-09-17 19:18:51 | INFO | train rank=0 | step=1810/50000 progress=3.6% loss=0.182757 lr=7.236000e-05 lr_backbone=7.236000e-05 elapsed=01:50:15 avg_step=3.65s recent_step=3.55s eta=48:55:27 +2026-09-17 19:19:20 | INFO | train rank=0 | step=1820/50000 progress=3.6% loss=0.172360 lr=7.276000e-05 lr_backbone=7.276000e-05 elapsed=01:50:43 avg_step=3.65s recent_step=2.81s eta=48:51:07 +2026-09-17 19:20:09 | INFO | train rank=0 | step=1830/50000 progress=3.7% loss=0.160193 lr=7.316000e-05 lr_backbone=7.316000e-05 elapsed=01:51:32 avg_step=3.66s recent_step=4.92s eta=48:56:05 +2026-09-17 19:20:42 | INFO | train rank=0 | step=1840/50000 progress=3.7% loss=0.159523 lr=7.356000e-05 lr_backbone=7.356000e-05 elapsed=01:52:05 avg_step=3.66s recent_step=3.32s eta=48:54:01 +2026-09-17 19:21:17 | INFO | train rank=0 | step=1850/50000 progress=3.7% loss=0.201300 lr=7.396000e-05 lr_backbone=7.396000e-05 elapsed=01:52:40 avg_step=3.65s recent_step=3.51s eta=48:52:48 +2026-09-17 19:21:46 | INFO | train rank=0 | step=1860/50000 progress=3.7% loss=0.149077 lr=7.436000e-05 lr_backbone=7.436000e-05 elapsed=01:53:09 avg_step=3.65s recent_step=2.90s eta=48:48:55 +2026-09-17 19:22:39 | INFO | train rank=0 | step=1870/50000 progress=3.7% loss=0.163673 lr=7.476000e-05 lr_backbone=7.476000e-05 elapsed=01:54:02 avg_step=3.66s recent_step=5.27s eta=48:55:17 +2026-09-17 19:23:04 | INFO | train rank=0 | step=1880/50000 progress=3.8% loss=0.165006 lr=7.516000e-05 lr_backbone=7.516000e-05 elapsed=01:54:27 avg_step=3.65s recent_step=2.51s eta=48:49:46 +2026-09-17 19:23:35 | INFO | train rank=0 | step=1890/50000 progress=3.8% loss=0.170695 lr=7.556000e-05 lr_backbone=7.556000e-05 elapsed=01:54:59 avg_step=3.65s recent_step=3.13s eta=48:46:57 +2026-09-17 19:24:07 | INFO | train rank=0 | step=1900/50000 progress=3.8% loss=0.151075 lr=7.596000e-05 lr_backbone=7.596000e-05 elapsed=01:55:31 avg_step=3.65s recent_step=3.19s eta=48:44:25 +2026-09-17 19:24:59 | INFO | train rank=0 | step=1910/50000 progress=3.8% loss=0.217169 lr=7.636000e-05 lr_backbone=7.636000e-05 elapsed=01:56:22 avg_step=3.66s recent_step=5.14s eta=48:50:04 +2026-09-17 19:25:31 | INFO | train rank=0 | step=1920/50000 progress=3.8% loss=0.156233 lr=7.676000e-05 lr_backbone=7.676000e-05 elapsed=01:56:54 avg_step=3.65s recent_step=3.22s eta=48:47:37 +2026-09-17 19:25:53 | INFO | train rank=0 | step=1930/50000 progress=3.9% loss=0.171516 lr=7.716000e-05 lr_backbone=7.716000e-05 elapsed=01:57:16 avg_step=3.65s recent_step=2.21s eta=48:41:01 +2026-09-17 19:26:30 | INFO | train rank=0 | step=1940/50000 progress=3.9% loss=0.163064 lr=7.756000e-05 lr_backbone=7.756000e-05 elapsed=01:57:53 avg_step=3.65s recent_step=3.67s eta=48:40:30 +2026-09-17 19:27:25 | INFO | train rank=0 | step=1950/50000 progress=3.9% loss=0.174805 lr=7.796000e-05 lr_backbone=7.796000e-05 elapsed=01:58:48 avg_step=3.66s recent_step=5.52s eta=48:47:36 +2026-09-17 19:27:52 | INFO | train rank=0 | step=1960/50000 progress=3.9% loss=0.138981 lr=7.836000e-05 lr_backbone=7.836000e-05 elapsed=01:59:16 avg_step=3.65s recent_step=2.75s eta=48:43:17 +2026-09-17 19:28:24 | INFO | train rank=0 | step=1970/50000 progress=3.9% loss=0.130988 lr=7.876000e-05 lr_backbone=7.876000e-05 elapsed=01:59:47 avg_step=3.65s recent_step=3.15s eta=48:40:38 +2026-09-17 19:28:54 | INFO | train rank=0 | step=1980/50000 progress=4.0% loss=0.162343 lr=7.916000e-05 lr_backbone=7.916000e-05 elapsed=02:00:17 avg_step=3.65s recent_step=3.00s eta=48:37:24 +2026-09-17 19:29:44 | INFO | train rank=0 | step=1990/50000 progress=4.0% loss=0.210119 lr=7.956000e-05 lr_backbone=7.956000e-05 elapsed=02:01:07 avg_step=3.65s recent_step=5.00s eta=48:42:16 +2026-09-17 19:30:10 | INFO | train rank=0 | step=2000/50000 progress=4.0% loss=0.170799 lr=7.996000e-05 lr_backbone=7.996000e-05 elapsed=02:01:34 avg_step=3.65s recent_step=2.65s eta=48:37:38 +2026-09-17 19:30:42 | INFO | train rank=0 | step=2010/50000 progress=4.0% loss=0.166065 lr=8.036000e-05 lr_backbone=8.036000e-05 elapsed=02:02:05 avg_step=3.64s recent_step=3.17s eta=48:35:07 +2026-09-17 19:31:13 | INFO | train rank=0 | step=2020/50000 progress=4.0% loss=0.159667 lr=8.076000e-05 lr_backbone=8.076000e-05 elapsed=02:02:36 avg_step=3.64s recent_step=3.09s eta=48:32:17 +2026-09-17 19:32:02 | INFO | train rank=0 | step=2030/50000 progress=4.1% loss=0.190848 lr=8.116000e-05 lr_backbone=8.116000e-05 elapsed=02:03:25 avg_step=3.65s recent_step=4.94s eta=48:36:47 +2026-09-17 19:32:35 | INFO | train rank=0 | step=2040/50000 progress=4.1% loss=0.181873 lr=8.156000e-05 lr_backbone=8.156000e-05 elapsed=02:03:58 avg_step=3.65s recent_step=3.25s eta=48:34:37 +2026-09-17 19:32:57 | INFO | train rank=0 | step=2050/50000 progress=4.1% loss=0.164694 lr=8.196000e-05 lr_backbone=8.196000e-05 elapsed=02:04:21 avg_step=3.64s recent_step=2.26s eta=48:28:36 +2026-09-17 19:33:30 | INFO | train rank=0 | step=2060/50000 progress=4.1% loss=0.180614 lr=8.236000e-05 lr_backbone=8.236000e-05 elapsed=02:04:53 avg_step=3.64s recent_step=3.27s eta=48:26:32 +2026-09-17 19:34:20 | INFO | train rank=0 | step=2070/50000 progress=4.1% loss=0.173696 lr=8.276000e-05 lr_backbone=8.276000e-05 elapsed=02:05:43 avg_step=3.64s recent_step=5.01s eta=48:31:14 +2026-09-17 19:34:53 | INFO | train rank=0 | step=2080/50000 progress=4.2% loss=0.200586 lr=8.316000e-05 lr_backbone=8.316000e-05 elapsed=02:06:16 avg_step=3.64s recent_step=3.31s eta=48:29:20 +2026-09-17 19:35:28 | INFO | train rank=0 | step=2090/50000 progress=4.2% loss=0.128259 lr=8.356000e-05 lr_backbone=8.356000e-05 elapsed=02:06:51 avg_step=3.64s recent_step=3.50s eta=48:28:11 +2026-09-17 19:35:56 | INFO | train rank=0 | step=2100/50000 progress=4.2% loss=0.179970 lr=8.396000e-05 lr_backbone=8.396000e-05 elapsed=02:07:20 avg_step=3.64s recent_step=2.83s eta=48:24:30 +2026-09-17 19:36:46 | INFO | train rank=0 | step=2110/50000 progress=4.2% loss=0.180010 lr=8.436000e-05 lr_backbone=8.436000e-05 elapsed=02:08:09 avg_step=3.64s recent_step=4.91s eta=48:28:42 +2026-09-17 19:37:10 | INFO | train rank=0 | step=2120/50000 progress=4.2% loss=0.165291 lr=8.476000e-05 lr_backbone=8.476000e-05 elapsed=02:08:33 avg_step=3.64s recent_step=2.40s eta=48:23:26 +2026-09-17 19:37:39 | INFO | train rank=0 | step=2130/50000 progress=4.3% loss=0.140801 lr=8.516000e-05 lr_backbone=8.516000e-05 elapsed=02:09:02 avg_step=3.64s recent_step=2.95s eta=48:20:14 +2026-09-17 19:38:05 | INFO | train rank=0 | step=2140/50000 progress=4.3% loss=0.158775 lr=8.556000e-05 lr_backbone=8.556000e-05 elapsed=02:09:28 avg_step=3.63s recent_step=2.55s eta=48:15:35 +2026-09-17 19:38:55 | INFO | train rank=0 | step=2150/50000 progress=4.3% loss=0.175604 lr=8.596000e-05 lr_backbone=8.596000e-05 elapsed=02:10:18 avg_step=3.64s recent_step=5.03s eta=48:20:09 +2026-09-17 19:39:32 | INFO | train rank=0 | step=2160/50000 progress=4.3% loss=0.266612 lr=8.636000e-05 lr_backbone=8.636000e-05 elapsed=02:10:55 avg_step=3.64s recent_step=3.72s eta=48:19:51 +2026-09-17 19:40:02 | INFO | train rank=0 | step=2170/50000 progress=4.3% loss=0.157596 lr=8.676000e-05 lr_backbone=8.676000e-05 elapsed=02:11:26 avg_step=3.63s recent_step=3.03s eta=48:17:01 +2026-09-17 19:40:35 | INFO | train rank=0 | step=2180/50000 progress=4.4% loss=0.150604 lr=8.716000e-05 lr_backbone=8.716000e-05 elapsed=02:11:58 avg_step=3.63s recent_step=3.26s eta=48:15:02 +2026-09-17 19:41:25 | INFO | train rank=0 | step=2190/50000 progress=4.4% loss=0.141323 lr=8.756000e-05 lr_backbone=8.756000e-05 elapsed=02:12:49 avg_step=3.64s recent_step=5.06s eta=48:19:37 +2026-09-17 19:41:54 | INFO | train rank=0 | step=2200/50000 progress=4.4% loss=0.146130 lr=8.796000e-05 lr_backbone=8.796000e-05 elapsed=02:13:18 avg_step=3.64s recent_step=2.88s eta=48:16:16 +2026-09-17 19:42:27 | INFO | train rank=0 | step=2210/50000 progress=4.4% loss=0.149057 lr=8.836000e-05 lr_backbone=8.836000e-05 elapsed=02:13:50 avg_step=3.63s recent_step=3.25s eta=48:14:16 +2026-09-17 19:42:58 | INFO | train rank=0 | step=2220/50000 progress=4.4% loss=0.141886 lr=8.876000e-05 lr_backbone=8.876000e-05 elapsed=02:14:21 avg_step=3.63s recent_step=3.09s eta=48:11:43 +2026-09-17 19:43:38 | INFO | train rank=0 | step=2230/50000 progress=4.5% loss=0.173069 lr=8.916000e-05 lr_backbone=8.916000e-05 elapsed=02:15:01 avg_step=3.63s recent_step=4.03s eta=48:12:32 +2026-09-17 19:44:12 | INFO | train rank=0 | step=2240/50000 progress=4.5% loss=0.161306 lr=8.956000e-05 lr_backbone=8.956000e-05 elapsed=02:15:35 avg_step=3.63s recent_step=3.37s eta=48:11:01 +2026-09-17 19:44:48 | INFO | train rank=0 | step=2250/50000 progress=4.5% loss=0.170659 lr=8.996000e-05 lr_backbone=8.996000e-05 elapsed=02:16:11 avg_step=3.63s recent_step=3.63s eta=48:10:25 +2026-09-17 19:45:18 | INFO | train rank=0 | step=2260/50000 progress=4.5% loss=0.202362 lr=9.036000e-05 lr_backbone=9.036000e-05 elapsed=02:16:41 avg_step=3.63s recent_step=2.95s eta=48:07:23 +2026-09-17 19:46:00 | INFO | train rank=0 | step=2270/50000 progress=4.5% loss=0.173997 lr=9.076000e-05 lr_backbone=9.076000e-05 elapsed=02:17:23 avg_step=3.63s recent_step=4.22s eta=48:08:52 +2026-09-17 19:46:37 | INFO | train rank=0 | step=2280/50000 progress=4.6% loss=0.139061 lr=9.116000e-05 lr_backbone=9.116000e-05 elapsed=02:18:00 avg_step=3.63s recent_step=3.74s eta=48:08:39 +2026-09-17 19:47:11 | INFO | train rank=0 | step=2290/50000 progress=4.6% loss=0.189708 lr=9.156000e-05 lr_backbone=9.156000e-05 elapsed=02:18:35 avg_step=3.63s recent_step=3.42s eta=48:07:19 +2026-09-17 19:47:45 | INFO | train rank=0 | step=2300/50000 progress=4.6% loss=0.199820 lr=9.196000e-05 lr_backbone=9.196000e-05 elapsed=02:19:08 avg_step=3.63s recent_step=3.38s eta=48:05:50 +2026-09-17 19:48:27 | INFO | train rank=0 | step=2310/50000 progress=4.6% loss=0.219190 lr=9.236000e-05 lr_backbone=9.236000e-05 elapsed=02:19:50 avg_step=3.63s recent_step=4.19s eta=48:07:09 +2026-09-17 19:49:00 | INFO | train rank=0 | step=2320/50000 progress=4.6% loss=0.190774 lr=9.276000e-05 lr_backbone=9.276000e-05 elapsed=02:20:23 avg_step=3.63s recent_step=3.26s eta=48:05:16 +2026-09-17 19:49:28 | INFO | train rank=0 | step=2330/50000 progress=4.7% loss=0.180812 lr=9.316000e-05 lr_backbone=9.316000e-05 elapsed=02:20:52 avg_step=3.63s recent_step=2.88s eta=48:02:05 +2026-09-17 19:50:01 | INFO | train rank=0 | step=2340/50000 progress=4.7% loss=0.147729 lr=9.356000e-05 lr_backbone=9.356000e-05 elapsed=02:21:24 avg_step=3.63s recent_step=3.25s eta=48:00:11 +2026-09-17 19:50:55 | INFO | train rank=0 | step=2350/50000 progress=4.7% loss=0.224958 lr=9.396000e-05 lr_backbone=9.396000e-05 elapsed=02:22:18 avg_step=3.63s recent_step=5.41s eta=48:05:37 +2026-09-17 19:51:23 | INFO | train rank=0 | step=2360/50000 progress=4.7% loss=0.138042 lr=9.436000e-05 lr_backbone=9.436000e-05 elapsed=02:22:46 avg_step=3.63s recent_step=2.76s eta=48:02:04 +2026-09-17 19:51:45 | INFO | train rank=0 | step=2370/50000 progress=4.7% loss=0.184603 lr=9.476000e-05 lr_backbone=9.476000e-05 elapsed=02:23:09 avg_step=3.62s recent_step=2.28s eta=47:56:57 +2026-09-17 19:52:12 | INFO | train rank=0 | step=2380/50000 progress=4.8% loss=0.174140 lr=9.516000e-05 lr_backbone=9.516000e-05 elapsed=02:23:36 avg_step=3.62s recent_step=2.70s eta=47:53:17 +2026-09-17 19:52:56 | INFO | train rank=0 | step=2390/50000 progress=4.8% loss=0.190989 lr=9.556000e-05 lr_backbone=9.556000e-05 elapsed=02:24:19 avg_step=3.62s recent_step=4.36s eta=47:55:09 +2026-09-17 19:53:28 | INFO | train rank=0 | step=2400/50000 progress=4.8% loss=0.140955 lr=9.596000e-05 lr_backbone=9.596000e-05 elapsed=02:24:51 avg_step=3.62s recent_step=3.18s eta=47:53:05 +2026-09-17 19:53:59 | INFO | train rank=0 | step=2410/50000 progress=4.8% loss=0.155962 lr=9.636000e-05 lr_backbone=9.636000e-05 elapsed=02:25:22 avg_step=3.62s recent_step=3.08s eta=47:50:40 +2026-09-17 19:54:25 | INFO | train rank=0 | step=2420/50000 progress=4.8% loss=0.222314 lr=9.676000e-05 lr_backbone=9.676000e-05 elapsed=02:25:48 avg_step=3.62s recent_step=2.65s eta=47:46:54 +2026-09-17 19:55:16 | INFO | train rank=0 | step=2430/50000 progress=4.9% loss=0.145105 lr=9.716000e-05 lr_backbone=9.716000e-05 elapsed=02:26:39 avg_step=3.62s recent_step=5.05s eta=47:50:59 +2026-09-17 19:55:49 | INFO | train rank=0 | step=2440/50000 progress=4.9% loss=0.154527 lr=9.756000e-05 lr_backbone=9.756000e-05 elapsed=02:27:12 avg_step=3.62s recent_step=3.29s eta=47:49:19 +2026-09-17 19:56:23 | INFO | train rank=0 | step=2450/50000 progress=4.9% loss=0.163811 lr=9.796000e-05 lr_backbone=9.796000e-05 elapsed=02:27:46 avg_step=3.62s recent_step=3.44s eta=47:48:07 +2026-09-17 19:56:54 | INFO | train rank=0 | step=2460/50000 progress=4.9% loss=0.156252 lr=9.836000e-05 lr_backbone=9.836000e-05 elapsed=02:28:17 avg_step=3.62s recent_step=3.07s eta=47:45:45 +2026-09-17 19:57:44 | INFO | train rank=0 | step=2470/50000 progress=4.9% loss=0.151075 lr=9.876000e-05 lr_backbone=9.876000e-05 elapsed=02:29:07 avg_step=3.62s recent_step=4.99s eta=47:49:34 +2026-09-17 19:58:07 | INFO | train rank=0 | step=2480/50000 progress=5.0% loss=0.155215 lr=9.916000e-05 lr_backbone=9.916000e-05 elapsed=02:29:30 avg_step=3.62s recent_step=2.34s eta=47:44:52 +2026-09-17 19:58:37 | INFO | train rank=0 | step=2490/50000 progress=5.0% loss=0.151842 lr=9.956000e-05 lr_backbone=9.956000e-05 elapsed=02:30:01 avg_step=3.61s recent_step=3.03s eta=47:42:23 +2026-09-17 19:59:37 | INFO | train rank=0 | step=2500/50000 progress=5.0% loss=0.152900 lr=9.996000e-05 lr_backbone=9.996000e-05 elapsed=02:31:00 avg_step=3.62s recent_step=5.99s eta=47:49:18 +2026-09-17 20:00:13 | INFO | train rank=0 | step=2510/50000 progress=5.0% loss=0.162780 lr=9.999999e-05 lr_backbone=9.999999e-05 elapsed=02:31:36 avg_step=3.62s recent_step=3.54s eta=47:48:26 +2026-09-17 20:00:42 | INFO | train rank=0 | step=2520/50000 progress=5.0% loss=0.132622 lr=9.999996e-05 lr_backbone=9.999996e-05 elapsed=02:32:05 avg_step=3.62s recent_step=2.94s eta=47:45:40 +2026-09-17 20:01:14 | INFO | train rank=0 | step=2530/50000 progress=5.1% loss=0.240032 lr=9.999990e-05 lr_backbone=9.999990e-05 elapsed=02:32:37 avg_step=3.62s recent_step=3.20s eta=47:43:44 +2026-09-17 20:02:00 | INFO | train rank=0 | step=2540/50000 progress=5.1% loss=0.184366 lr=9.999983e-05 lr_backbone=9.999983e-05 elapsed=02:33:23 avg_step=3.62s recent_step=4.63s eta=47:46:16 +2026-09-17 20:02:32 | INFO | train rank=0 | step=2550/50000 progress=5.1% loss=0.140264 lr=9.999973e-05 lr_backbone=9.999973e-05 elapsed=02:33:56 avg_step=3.62s recent_step=3.21s eta=47:44:23 +2026-09-17 20:03:01 | INFO | train rank=0 | step=2560/50000 progress=5.1% loss=0.128749 lr=9.999961e-05 lr_backbone=9.999961e-05 elapsed=02:34:25 avg_step=3.62s recent_step=2.91s eta=47:41:35 +2026-09-17 20:03:36 | INFO | train rank=0 | step=2570/50000 progress=5.1% loss=0.175157 lr=9.999946e-05 lr_backbone=9.999946e-05 elapsed=02:34:59 avg_step=3.62s recent_step=3.45s eta=47:40:27 +2026-09-17 20:04:22 | INFO | train rank=0 | step=2580/50000 progress=5.2% loss=0.168868 lr=9.999930e-05 lr_backbone=9.999930e-05 elapsed=02:35:45 avg_step=3.62s recent_step=4.59s eta=47:42:49 +2026-09-17 20:04:51 | INFO | train rank=0 | step=2590/50000 progress=5.2% loss=0.199120 lr=9.999911e-05 lr_backbone=9.999911e-05 elapsed=02:36:14 avg_step=3.62s recent_step=2.92s eta=47:40:04 +2026-09-17 20:05:24 | INFO | train rank=0 | step=2600/50000 progress=5.2% loss=0.154257 lr=9.999891e-05 lr_backbone=9.999891e-05 elapsed=02:36:47 avg_step=3.62s recent_step=3.32s eta=47:38:33 +2026-09-17 20:05:53 | INFO | train rank=0 | step=2610/50000 progress=5.2% loss=0.160618 lr=9.999868e-05 lr_backbone=9.999868e-05 elapsed=02:37:16 avg_step=3.62s recent_step=2.90s eta=47:35:47 +2026-09-17 20:06:41 | INFO | train rank=0 | step=2620/50000 progress=5.2% loss=0.163862 lr=9.999843e-05 lr_backbone=9.999843e-05 elapsed=02:38:04 avg_step=3.62s recent_step=4.74s eta=47:38:35 +2026-09-17 20:07:09 | INFO | train rank=0 | step=2630/50000 progress=5.3% loss=0.200513 lr=9.999815e-05 lr_backbone=9.999815e-05 elapsed=02:38:32 avg_step=3.62s recent_step=2.81s eta=47:35:32 +2026-09-17 20:07:33 | INFO | train rank=0 | step=2640/50000 progress=5.3% loss=0.161663 lr=9.999786e-05 lr_backbone=9.999786e-05 elapsed=02:38:56 avg_step=3.61s recent_step=2.41s eta=47:31:20 +2026-09-17 20:08:14 | INFO | train rank=0 | step=2650/50000 progress=5.3% loss=0.161056 lr=9.999754e-05 lr_backbone=9.999754e-05 elapsed=02:39:37 avg_step=3.61s recent_step=4.14s eta=47:32:18 +2026-09-17 20:09:00 | INFO | train rank=0 | step=2660/50000 progress=5.3% loss=0.148506 lr=9.999720e-05 lr_backbone=9.999720e-05 elapsed=02:40:23 avg_step=3.62s recent_step=4.56s eta=47:34:30 +2026-09-17 20:09:41 | INFO | train rank=0 | step=2670/50000 progress=5.3% loss=0.150180 lr=9.999684e-05 lr_backbone=9.999684e-05 elapsed=02:41:04 avg_step=3.62s recent_step=4.10s eta=47:35:19 +2026-09-17 20:10:17 | INFO | train rank=0 | step=2680/50000 progress=5.4% loss=0.150829 lr=9.999646e-05 lr_backbone=9.999646e-05 elapsed=02:41:41 avg_step=3.62s recent_step=3.67s eta=47:34:52 +2026-09-17 20:10:55 | INFO | train rank=0 | step=2690/50000 progress=5.4% loss=0.173929 lr=9.999605e-05 lr_backbone=9.999605e-05 elapsed=02:42:19 avg_step=3.62s recent_step=3.79s eta=47:34:45 +2026-09-17 20:11:20 | INFO | train rank=0 | step=2700/50000 progress=5.4% loss=0.142168 lr=9.999563e-05 lr_backbone=9.999563e-05 elapsed=02:42:43 avg_step=3.62s recent_step=2.48s eta=47:30:49 +2026-09-17 20:12:14 | INFO | train rank=0 | step=2710/50000 progress=5.4% loss=0.159312 lr=9.999518e-05 lr_backbone=9.999518e-05 elapsed=02:43:37 avg_step=3.62s recent_step=5.40s eta=47:35:23 +2026-09-17 20:12:43 | INFO | train rank=0 | step=2720/50000 progress=5.4% loss=0.159986 lr=9.999471e-05 lr_backbone=9.999471e-05 elapsed=02:44:06 avg_step=3.62s recent_step=2.87s eta=47:32:36 +2026-09-17 20:13:13 | INFO | train rank=0 | step=2730/50000 progress=5.5% loss=0.162717 lr=9.999422e-05 lr_backbone=9.999422e-05 elapsed=02:44:36 avg_step=3.62s recent_step=3.04s eta=47:30:20 +2026-09-17 20:13:40 | INFO | train rank=0 | step=2740/50000 progress=5.5% loss=0.155833 lr=9.999370e-05 lr_backbone=9.999370e-05 elapsed=02:45:04 avg_step=3.61s recent_step=2.72s eta=47:27:08 +2026-09-17 20:14:28 | INFO | train rank=0 | step=2750/50000 progress=5.5% loss=0.182594 lr=9.999317e-05 lr_backbone=9.999317e-05 elapsed=02:45:52 avg_step=3.62s recent_step=4.79s eta=47:29:53 +2026-09-17 20:14:54 | INFO | train rank=0 | step=2760/50000 progress=5.5% loss=0.137295 lr=9.999261e-05 lr_backbone=9.999261e-05 elapsed=02:46:18 avg_step=3.62s recent_step=2.61s eta=47:26:23 +2026-09-17 20:15:30 | INFO | train rank=0 | step=2770/50000 progress=5.5% loss=0.168311 lr=9.999203e-05 lr_backbone=9.999203e-05 elapsed=02:46:53 avg_step=3.62s recent_step=3.57s eta=47:25:39 +2026-09-17 20:15:59 | INFO | train rank=0 | step=2780/50000 progress=5.6% loss=0.150664 lr=9.999143e-05 lr_backbone=9.999143e-05 elapsed=02:47:22 avg_step=3.61s recent_step=2.88s eta=47:22:59 +2026-09-17 20:16:45 | INFO | train rank=0 | step=2790/50000 progress=5.6% loss=0.135379 lr=9.999080e-05 lr_backbone=9.999080e-05 elapsed=02:48:08 avg_step=3.62s recent_step=4.58s eta=47:25:06 +2026-09-17 20:17:19 | INFO | train rank=0 | step=2800/50000 progress=5.6% loss=0.137117 lr=9.999016e-05 lr_backbone=9.999016e-05 elapsed=02:48:42 avg_step=3.62s recent_step=3.43s eta=47:23:59 +2026-09-17 20:17:48 | INFO | train rank=0 | step=2810/50000 progress=5.6% loss=0.185796 lr=9.998949e-05 lr_backbone=9.998949e-05 elapsed=02:49:12 avg_step=3.61s recent_step=2.96s eta=47:21:33 +2026-09-17 20:18:17 | INFO | train rank=0 | step=2820/50000 progress=5.6% loss=0.139800 lr=9.998880e-05 lr_backbone=9.998880e-05 elapsed=02:49:40 avg_step=3.61s recent_step=2.86s eta=47:18:51 +2026-09-17 20:19:12 | INFO | train rank=0 | step=2830/50000 progress=5.7% loss=0.152480 lr=9.998809e-05 lr_backbone=9.998809e-05 elapsed=02:50:35 avg_step=3.62s recent_step=5.45s eta=47:23:21 +2026-09-17 20:19:40 | INFO | train rank=0 | step=2840/50000 progress=5.7% loss=0.150304 lr=9.998736e-05 lr_backbone=9.998736e-05 elapsed=02:51:03 avg_step=3.61s recent_step=2.80s eta=47:20:30 +2026-09-17 20:20:05 | INFO | train rank=0 | step=2850/50000 progress=5.7% loss=0.180458 lr=9.998660e-05 lr_backbone=9.998660e-05 elapsed=02:51:28 avg_step=3.61s recent_step=2.50s eta=47:16:50 +2026-09-17 20:20:52 | INFO | train rank=0 | step=2860/50000 progress=5.7% loss=0.170792 lr=9.998583e-05 lr_backbone=9.998583e-05 elapsed=02:52:16 avg_step=3.61s recent_step=4.77s eta=47:19:24 +2026-09-17 20:21:22 | INFO | train rank=0 | step=2870/50000 progress=5.7% loss=0.141979 lr=9.998503e-05 lr_backbone=9.998503e-05 elapsed=02:52:45 avg_step=3.61s recent_step=2.92s eta=47:16:55 +2026-09-17 20:21:55 | INFO | train rank=0 | step=2880/50000 progress=5.8% loss=0.159706 lr=9.998421e-05 lr_backbone=9.998421e-05 elapsed=02:53:18 avg_step=3.61s recent_step=3.32s eta=47:15:31 +2026-09-17 20:22:30 | INFO | train rank=0 | step=2890/50000 progress=5.8% loss=0.130595 lr=9.998337e-05 lr_backbone=9.998337e-05 elapsed=02:53:54 avg_step=3.61s recent_step=3.58s eta=47:14:49 +2026-09-17 20:23:17 | INFO | train rank=0 | step=2900/50000 progress=5.8% loss=0.165930 lr=9.998250e-05 lr_backbone=9.998250e-05 elapsed=02:54:41 avg_step=3.61s recent_step=4.68s eta=47:17:06 +2026-09-17 20:23:46 | INFO | train rank=0 | step=2910/50000 progress=5.8% loss=0.126252 lr=9.998162e-05 lr_backbone=9.998162e-05 elapsed=02:55:09 avg_step=3.61s recent_step=2.83s eta=47:14:24 +2026-09-17 20:24:21 | INFO | train rank=0 | step=2920/50000 progress=5.8% loss=0.175450 lr=9.998071e-05 lr_backbone=9.998071e-05 elapsed=02:55:45 avg_step=3.61s recent_step=3.56s eta=47:13:40 +2026-09-17 20:24:50 | INFO | train rank=0 | step=2930/50000 progress=5.9% loss=0.107938 lr=9.997978e-05 lr_backbone=9.997978e-05 elapsed=02:56:13 avg_step=3.61s recent_step=2.89s eta=47:11:09 +2026-09-17 20:25:30 | INFO | train rank=0 | step=2940/50000 progress=5.9% loss=0.155598 lr=9.997883e-05 lr_backbone=9.997883e-05 elapsed=02:56:53 avg_step=3.61s recent_step=3.97s eta=47:11:31 +2026-09-17 20:26:02 | INFO | train rank=0 | step=2950/50000 progress=5.9% loss=0.136573 lr=9.997786e-05 lr_backbone=9.997786e-05 elapsed=02:57:26 avg_step=3.61s recent_step=3.25s eta=47:09:57 +2026-09-17 20:26:31 | INFO | train rank=0 | step=2960/50000 progress=5.9% loss=0.159325 lr=9.997686e-05 lr_backbone=9.997686e-05 elapsed=02:57:55 avg_step=3.61s recent_step=2.91s eta=47:07:29 +2026-09-17 20:27:06 | INFO | train rank=0 | step=2970/50000 progress=5.9% loss=0.152216 lr=9.997584e-05 lr_backbone=9.997584e-05 elapsed=02:58:29 avg_step=3.61s recent_step=3.46s eta=47:06:30 +2026-09-17 20:27:54 | INFO | train rank=0 | step=2980/50000 progress=6.0% loss=0.141277 lr=9.997481e-05 lr_backbone=9.997481e-05 elapsed=02:59:17 avg_step=3.61s recent_step=4.77s eta=47:08:57 +2026-09-17 20:28:20 | INFO | train rank=0 | step=2990/50000 progress=6.0% loss=0.136936 lr=9.997375e-05 lr_backbone=9.997375e-05 elapsed=02:59:43 avg_step=3.61s recent_step=2.64s eta=47:05:49 +2026-09-17 20:28:48 | INFO | train rank=0 | step=3000/50000 progress=6.0% loss=0.163513 lr=9.997266e-05 lr_backbone=9.997266e-05 elapsed=03:00:11 avg_step=3.60s recent_step=2.75s eta=47:02:59 +2026-09-17 20:29:25 | INFO | train rank=0 | step=3010/50000 progress=6.0% loss=0.137943 lr=9.997156e-05 lr_backbone=9.997156e-05 elapsed=03:00:48 avg_step=3.60s recent_step=3.70s eta=47:02:38 +2026-09-17 20:30:14 | INFO | train rank=0 | step=3020/50000 progress=6.0% loss=0.146125 lr=9.997043e-05 lr_backbone=9.997043e-05 elapsed=03:01:37 avg_step=3.61s recent_step=4.92s eta=47:05:26 +2026-09-17 20:30:45 | INFO | train rank=0 | step=3030/50000 progress=6.1% loss=0.131577 lr=9.996928e-05 lr_backbone=9.996928e-05 elapsed=03:02:09 avg_step=3.61s recent_step=3.17s eta=47:03:42 +2026-09-17 20:31:18 | INFO | train rank=0 | step=3040/50000 progress=6.1% loss=0.114113 lr=9.996811e-05 lr_backbone=9.996811e-05 elapsed=03:02:41 avg_step=3.61s recent_step=3.27s eta=47:02:13 +2026-09-17 20:31:49 | INFO | train rank=0 | step=3050/50000 progress=6.1% loss=0.137706 lr=9.996692e-05 lr_backbone=9.996692e-05 elapsed=03:03:12 avg_step=3.60s recent_step=3.10s eta=47:00:19 +2026-09-17 20:32:43 | INFO | train rank=0 | step=3060/50000 progress=6.1% loss=0.245631 lr=9.996571e-05 lr_backbone=9.996571e-05 elapsed=03:04:06 avg_step=3.61s recent_step=5.35s eta=47:04:11 +2026-09-17 20:33:11 | INFO | train rank=0 | step=3070/50000 progress=6.1% loss=0.166494 lr=9.996447e-05 lr_backbone=9.996447e-05 elapsed=03:04:35 avg_step=3.61s recent_step=2.88s eta=47:01:44 +2026-09-17 20:33:44 | INFO | train rank=0 | step=3080/50000 progress=6.2% loss=0.135883 lr=9.996322e-05 lr_backbone=9.996322e-05 elapsed=03:05:07 avg_step=3.61s recent_step=3.27s eta=47:00:16 +2026-09-17 20:34:18 | INFO | train rank=0 | step=3090/50000 progress=6.2% loss=0.165420 lr=9.996194e-05 lr_backbone=9.996194e-05 elapsed=03:05:41 avg_step=3.61s recent_step=3.38s eta=46:59:05 +2026-09-17 20:35:11 | INFO | train rank=0 | step=3100/50000 progress=6.2% loss=0.158598 lr=9.996064e-05 lr_backbone=9.996064e-05 elapsed=03:06:34 avg_step=3.61s recent_step=5.30s eta=47:02:45 +2026-09-17 20:35:39 | INFO | train rank=0 | step=3110/50000 progress=6.2% loss=0.117562 lr=9.995931e-05 lr_backbone=9.995931e-05 elapsed=03:07:02 avg_step=3.61s recent_step=2.82s eta=47:00:09 +2026-09-17 20:36:15 | INFO | train rank=0 | step=3120/50000 progress=6.2% loss=0.145302 lr=9.995797e-05 lr_backbone=9.995797e-05 elapsed=03:07:38 avg_step=3.61s recent_step=3.54s eta=46:59:23 +2026-09-17 20:36:48 | INFO | train rank=0 | step=3130/50000 progress=6.3% loss=0.143482 lr=9.995660e-05 lr_backbone=9.995660e-05 elapsed=03:08:11 avg_step=3.61s recent_step=3.36s eta=46:58:10 +2026-09-17 20:37:38 | INFO | train rank=0 | step=3140/50000 progress=6.3% loss=0.135364 lr=9.995521e-05 lr_backbone=9.995521e-05 elapsed=03:09:02 avg_step=3.61s recent_step=5.03s eta=47:01:07 +2026-09-17 20:38:11 | INFO | train rank=0 | step=3150/50000 progress=6.3% loss=0.150262 lr=9.995380e-05 lr_backbone=9.995380e-05 elapsed=03:09:35 avg_step=3.61s recent_step=3.30s eta=46:59:44 +2026-09-17 20:38:43 | INFO | train rank=0 | step=3160/50000 progress=6.3% loss=0.136270 lr=9.995237e-05 lr_backbone=9.995237e-05 elapsed=03:10:07 avg_step=3.61s recent_step=3.19s eta=46:58:05 +2026-09-17 20:39:13 | INFO | train rank=0 | step=3170/50000 progress=6.3% loss=0.191388 lr=9.995092e-05 lr_backbone=9.995092e-05 elapsed=03:10:36 avg_step=3.61s recent_step=2.94s eta=46:55:49 +2026-09-17 20:40:00 | INFO | train rank=0 | step=3180/50000 progress=6.4% loss=0.154490 lr=9.994944e-05 lr_backbone=9.994944e-05 elapsed=03:11:23 avg_step=3.61s recent_step=4.68s eta=46:57:52 +2026-09-17 20:40:32 | INFO | train rank=0 | step=3190/50000 progress=6.4% loss=0.133402 lr=9.994794e-05 lr_backbone=9.994794e-05 elapsed=03:11:56 avg_step=3.61s recent_step=3.28s eta=46:56:27 +2026-09-17 20:41:04 | INFO | train rank=0 | step=3200/50000 progress=6.4% loss=0.165421 lr=9.994642e-05 lr_backbone=9.994642e-05 elapsed=03:12:27 avg_step=3.61s recent_step=3.18s eta=46:54:48 +2026-09-17 20:41:32 | INFO | train rank=0 | step=3210/50000 progress=6.4% loss=0.120887 lr=9.994488e-05 lr_backbone=9.994488e-05 elapsed=03:12:55 avg_step=3.61s recent_step=2.81s eta=46:52:15 +2026-09-17 20:42:14 | INFO | train rank=0 | step=3220/50000 progress=6.4% loss=0.126137 lr=9.994332e-05 lr_backbone=9.994332e-05 elapsed=03:13:38 avg_step=3.61s recent_step=4.22s eta=46:53:08 +2026-09-17 20:42:43 | INFO | train rank=0 | step=3230/50000 progress=6.5% loss=0.135387 lr=9.994173e-05 lr_backbone=9.994173e-05 elapsed=03:14:06 avg_step=3.61s recent_step=2.87s eta=46:50:45 +2026-09-17 20:43:14 | INFO | train rank=0 | step=3240/50000 progress=6.5% loss=0.152315 lr=9.994013e-05 lr_backbone=9.994013e-05 elapsed=03:14:38 avg_step=3.60s recent_step=3.13s eta=46:49:00 +2026-09-17 20:43:47 | INFO | train rank=0 | step=3250/50000 progress=6.5% loss=0.165669 lr=9.993850e-05 lr_backbone=9.993850e-05 elapsed=03:15:10 avg_step=3.60s recent_step=3.21s eta=46:47:28 +2026-09-17 20:44:32 | INFO | train rank=0 | step=3260/50000 progress=6.5% loss=0.137082 lr=9.993685e-05 lr_backbone=9.993685e-05 elapsed=03:15:55 avg_step=3.61s recent_step=4.55s eta=46:49:07 +2026-09-17 20:45:02 | INFO | train rank=0 | step=3270/50000 progress=6.5% loss=0.122113 lr=9.993518e-05 lr_backbone=9.993518e-05 elapsed=03:16:25 avg_step=3.60s recent_step=2.95s eta=46:46:58 +2026-09-17 20:45:31 | INFO | train rank=0 | step=3280/50000 progress=6.6% loss=0.130619 lr=9.993348e-05 lr_backbone=9.993348e-05 elapsed=03:16:55 avg_step=3.60s recent_step=3.00s eta=46:44:55 +2026-09-17 20:46:05 | INFO | train rank=0 | step=3290/50000 progress=6.6% loss=0.142040 lr=9.993176e-05 lr_backbone=9.993176e-05 elapsed=03:17:28 avg_step=3.60s recent_step=3.31s eta=46:43:38 +2026-09-17 20:47:00 | INFO | train rank=0 | step=3300/50000 progress=6.6% loss=0.139410 lr=9.993003e-05 lr_backbone=9.993003e-05 elapsed=03:18:23 avg_step=3.61s recent_step=5.51s eta=46:47:33 +2026-09-17 20:47:34 | INFO | train rank=0 | step=3310/50000 progress=6.6% loss=0.145038 lr=9.992827e-05 lr_backbone=9.992827e-05 elapsed=03:18:57 avg_step=3.61s recent_step=3.42s eta=46:46:30 +2026-09-17 20:48:05 | INFO | train rank=0 | step=3320/50000 progress=6.6% loss=0.159734 lr=9.992649e-05 lr_backbone=9.992649e-05 elapsed=03:19:29 avg_step=3.61s recent_step=3.14s eta=46:44:49 +2026-09-17 20:48:34 | INFO | train rank=0 | step=3330/50000 progress=6.7% loss=0.128180 lr=9.992468e-05 lr_backbone=9.992468e-05 elapsed=03:19:57 avg_step=3.60s recent_step=2.86s eta=46:42:28 +2026-09-17 20:49:17 | INFO | train rank=0 | step=3340/50000 progress=6.7% loss=0.145147 lr=9.992286e-05 lr_backbone=9.992286e-05 elapsed=03:20:41 avg_step=3.61s recent_step=4.34s eta=46:43:35 +2026-09-17 20:49:47 | INFO | train rank=0 | step=3350/50000 progress=6.7% loss=0.096660 lr=9.992101e-05 lr_backbone=9.992101e-05 elapsed=03:21:10 avg_step=3.60s recent_step=2.98s eta=46:41:32 +2026-09-17 20:50:18 | INFO | train rank=0 | step=3360/50000 progress=6.7% loss=0.183970 lr=9.991914e-05 lr_backbone=9.991914e-05 elapsed=03:21:42 avg_step=3.60s recent_step=3.11s eta=46:39:48 +2026-09-17 20:50:47 | INFO | train rank=0 | step=3370/50000 progress=6.7% loss=0.152223 lr=9.991725e-05 lr_backbone=9.991725e-05 elapsed=03:22:10 avg_step=3.60s recent_step=2.87s eta=46:37:31 +2026-09-17 20:51:34 | INFO | train rank=0 | step=3380/50000 progress=6.8% loss=0.153400 lr=9.991534e-05 lr_backbone=9.991534e-05 elapsed=03:22:58 avg_step=3.60s recent_step=4.74s eta=46:39:32 +2026-09-17 20:52:04 | INFO | train rank=0 | step=3390/50000 progress=6.8% loss=0.157889 lr=9.991340e-05 lr_backbone=9.991340e-05 elapsed=03:23:27 avg_step=3.60s recent_step=2.94s eta=46:37:25 +2026-09-17 20:52:34 | INFO | train rank=0 | step=3400/50000 progress=6.8% loss=0.135244 lr=9.991145e-05 lr_backbone=9.991145e-05 elapsed=03:23:57 avg_step=3.60s recent_step=2.97s eta=46:35:23 +2026-09-17 20:53:04 | INFO | train rank=0 | step=3410/50000 progress=6.8% loss=0.128574 lr=9.990947e-05 lr_backbone=9.990947e-05 elapsed=03:24:27 avg_step=3.60s recent_step=3.06s eta=46:33:33 +2026-09-17 20:54:02 | INFO | train rank=0 | step=3420/50000 progress=6.8% loss=0.108899 lr=9.990747e-05 lr_backbone=9.990747e-05 elapsed=03:25:25 avg_step=3.60s recent_step=5.78s eta=46:37:54 +2026-09-17 20:54:35 | INFO | train rank=0 | step=3430/50000 progress=6.9% loss=0.121305 lr=9.990545e-05 lr_backbone=9.990545e-05 elapsed=03:25:58 avg_step=3.60s recent_step=3.32s eta=46:36:39 +2026-09-17 20:55:02 | INFO | train rank=0 | step=3440/50000 progress=6.9% loss=0.188043 lr=9.990340e-05 lr_backbone=9.990340e-05 elapsed=03:26:25 avg_step=3.60s recent_step=2.70s eta=46:34:01 +2026-09-17 20:55:33 | INFO | train rank=0 | step=3450/50000 progress=6.9% loss=0.130577 lr=9.990134e-05 lr_backbone=9.990134e-05 elapsed=03:26:56 avg_step=3.60s recent_step=3.07s eta=46:32:13 +2026-09-17 20:56:32 | INFO | train rank=0 | step=3460/50000 progress=6.9% loss=0.126930 lr=9.989925e-05 lr_backbone=9.989925e-05 elapsed=03:27:56 avg_step=3.61s recent_step=5.95s eta=46:36:54 +2026-09-17 20:56:55 | INFO | train rank=0 | step=3470/50000 progress=6.9% loss=0.173382 lr=9.989714e-05 lr_backbone=9.989714e-05 elapsed=03:28:18 avg_step=3.60s recent_step=2.25s eta=46:33:15 +2026-09-17 20:57:22 | INFO | train rank=0 | step=3480/50000 progress=7.0% loss=0.183836 lr=9.989501e-05 lr_backbone=9.989501e-05 elapsed=03:28:45 avg_step=3.60s recent_step=2.70s eta=46:30:39 +2026-09-17 20:57:58 | INFO | train rank=0 | step=3490/50000 progress=7.0% loss=0.157843 lr=9.989286e-05 lr_backbone=9.989286e-05 elapsed=03:29:21 avg_step=3.60s recent_step=3.59s eta=46:30:02 +2026-09-17 20:58:46 | INFO | train rank=0 | step=3500/50000 progress=7.0% loss=0.116877 lr=9.989068e-05 lr_backbone=9.989068e-05 elapsed=03:30:09 avg_step=3.60s recent_step=4.81s eta=46:32:06 +2026-09-17 20:59:15 | INFO | train rank=0 | step=3510/50000 progress=7.0% loss=0.159595 lr=9.988848e-05 lr_backbone=9.988848e-05 elapsed=03:30:39 avg_step=3.60s recent_step=2.97s eta=46:30:06 +2026-09-17 20:59:50 | INFO | train rank=0 | step=3520/50000 progress=7.0% loss=0.118024 lr=9.988627e-05 lr_backbone=9.988627e-05 elapsed=03:31:13 avg_step=3.60s recent_step=3.45s eta=46:29:10 +2026-09-17 21:00:16 | INFO | train rank=0 | step=3530/50000 progress=7.1% loss=0.145672 lr=9.988403e-05 lr_backbone=9.988403e-05 elapsed=03:31:39 avg_step=3.60s recent_step=2.59s eta=46:26:22 +2026-09-17 21:01:06 | INFO | train rank=0 | step=3540/50000 progress=7.1% loss=0.117020 lr=9.988176e-05 lr_backbone=9.988176e-05 elapsed=03:32:29 avg_step=3.60s recent_step=4.97s eta=46:28:46 +2026-09-17 21:01:31 | INFO | train rank=0 | step=3550/50000 progress=7.1% loss=0.154084 lr=9.987948e-05 lr_backbone=9.987948e-05 elapsed=03:32:54 avg_step=3.60s recent_step=2.55s eta=46:25:53 +2026-09-17 21:02:10 | INFO | train rank=0 | step=3560/50000 progress=7.1% loss=0.139518 lr=9.987718e-05 lr_backbone=9.987718e-05 elapsed=03:33:34 avg_step=3.60s recent_step=3.93s eta=46:26:00 +2026-09-17 21:02:47 | INFO | train rank=0 | step=3570/50000 progress=7.1% loss=0.141785 lr=9.987485e-05 lr_backbone=9.987485e-05 elapsed=03:34:11 avg_step=3.60s recent_step=3.70s eta=46:25:37 +2026-09-17 21:03:27 | INFO | train rank=0 | step=3580/50000 progress=7.2% loss=0.137230 lr=9.987250e-05 lr_backbone=9.987250e-05 elapsed=03:34:50 avg_step=3.60s recent_step=3.94s eta=46:25:45 +2026-09-17 21:03:57 | INFO | train rank=0 | step=3590/50000 progress=7.2% loss=0.144215 lr=9.987013e-05 lr_backbone=9.987013e-05 elapsed=03:35:21 avg_step=3.60s recent_step=3.05s eta=46:23:58 +2026-09-17 21:04:24 | INFO | train rank=0 | step=3600/50000 progress=7.2% loss=0.130300 lr=9.986773e-05 lr_backbone=9.986773e-05 elapsed=03:35:48 avg_step=3.60s recent_step=2.71s eta=46:21:27 +2026-09-17 21:04:59 | INFO | train rank=0 | step=3610/50000 progress=7.2% loss=0.107725 lr=9.986532e-05 lr_backbone=9.986532e-05 elapsed=03:36:22 avg_step=3.60s recent_step=3.47s eta=46:20:34 +2026-09-17 21:05:51 | INFO | train rank=0 | step=3620/50000 progress=7.2% loss=0.139199 lr=9.986288e-05 lr_backbone=9.986288e-05 elapsed=03:37:14 avg_step=3.60s recent_step=5.16s eta=46:23:19 +2026-09-17 21:06:23 | INFO | train rank=0 | step=3630/50000 progress=7.3% loss=0.122429 lr=9.986043e-05 lr_backbone=9.986043e-05 elapsed=03:37:46 avg_step=3.60s recent_step=3.20s eta=46:21:52 +2026-09-17 21:06:56 | INFO | train rank=0 | step=3640/50000 progress=7.3% loss=0.148914 lr=9.985795e-05 lr_backbone=9.985795e-05 elapsed=03:38:19 avg_step=3.60s recent_step=3.29s eta=46:20:37 +2026-09-17 21:07:25 | INFO | train rank=0 | step=3650/50000 progress=7.3% loss=0.142709 lr=9.985544e-05 lr_backbone=9.985544e-05 elapsed=03:38:49 avg_step=3.60s recent_step=2.96s eta=46:18:40 +2026-09-17 21:08:22 | INFO | train rank=0 | step=3660/50000 progress=7.3% loss=0.135236 lr=9.985292e-05 lr_backbone=9.985292e-05 elapsed=03:39:46 avg_step=3.60s recent_step=5.71s eta=46:22:32 +2026-09-17 21:08:54 | INFO | train rank=0 | step=3670/50000 progress=7.3% loss=0.117763 lr=9.985037e-05 lr_backbone=9.985037e-05 elapsed=03:40:17 avg_step=3.60s recent_step=3.14s eta=46:20:58 +2026-09-17 21:09:26 | INFO | train rank=0 | step=3680/50000 progress=7.4% loss=0.142765 lr=9.984781e-05 lr_backbone=9.984781e-05 elapsed=03:40:49 avg_step=3.60s recent_step=3.21s eta=46:19:33 +2026-09-17 21:09:59 | INFO | train rank=0 | step=3690/50000 progress=7.4% loss=0.128268 lr=9.984522e-05 lr_backbone=9.984522e-05 elapsed=03:41:22 avg_step=3.60s recent_step=3.31s eta=46:18:21 +2026-09-17 21:10:43 | INFO | train rank=0 | step=3700/50000 progress=7.4% loss=0.111739 lr=9.984261e-05 lr_backbone=9.984261e-05 elapsed=03:42:06 avg_step=3.60s recent_step=4.36s eta=46:19:19 +2026-09-17 21:11:14 | INFO | train rank=0 | step=3710/50000 progress=7.4% loss=0.169861 lr=9.983997e-05 lr_backbone=9.983997e-05 elapsed=03:42:37 avg_step=3.60s recent_step=3.15s eta=46:17:46 +2026-09-17 21:11:36 | INFO | train rank=0 | step=3720/50000 progress=7.4% loss=0.122212 lr=9.983732e-05 lr_backbone=9.983732e-05 elapsed=03:42:59 avg_step=3.60s recent_step=2.21s eta=46:14:17 +2026-09-17 21:12:12 | INFO | train rank=0 | step=3730/50000 progress=7.5% loss=0.149076 lr=9.983464e-05 lr_backbone=9.983464e-05 elapsed=03:43:35 avg_step=3.60s recent_step=3.61s eta=46:13:42 +2026-09-17 21:13:06 | INFO | train rank=0 | step=3740/50000 progress=7.5% loss=0.110566 lr=9.983194e-05 lr_backbone=9.983194e-05 elapsed=03:44:29 avg_step=3.60s recent_step=5.34s eta=46:16:42 +2026-09-17 21:13:38 | INFO | train rank=0 | step=3750/50000 progress=7.5% loss=0.119338 lr=9.982922e-05 lr_backbone=9.982922e-05 elapsed=03:45:02 avg_step=3.60s recent_step=3.27s eta=46:15:25 +2026-09-17 21:14:13 | INFO | train rank=0 | step=3760/50000 progress=7.5% loss=0.148847 lr=9.982648e-05 lr_backbone=9.982648e-05 elapsed=03:45:36 avg_step=3.60s recent_step=3.46s eta=46:14:31 +2026-09-17 21:14:38 | INFO | train rank=0 | step=3770/50000 progress=7.5% loss=0.116221 lr=9.982372e-05 lr_backbone=9.982372e-05 elapsed=03:46:01 avg_step=3.60s recent_step=2.49s eta=46:11:40 +2026-09-17 21:15:30 | INFO | train rank=0 | step=3780/50000 progress=7.6% loss=0.107820 lr=9.982093e-05 lr_backbone=9.982093e-05 elapsed=03:46:53 avg_step=3.60s recent_step=5.20s eta=46:14:19 +2026-09-17 21:16:01 | INFO | train rank=0 | step=3790/50000 progress=7.6% loss=0.125696 lr=9.981813e-05 lr_backbone=9.981813e-05 elapsed=03:47:25 avg_step=3.60s recent_step=3.16s eta=46:12:50 +2026-09-17 21:16:29 | INFO | train rank=0 | step=3800/50000 progress=7.6% loss=0.197142 lr=9.981530e-05 lr_backbone=9.981530e-05 elapsed=03:47:53 avg_step=3.60s recent_step=2.81s eta=46:10:38 +2026-09-17 21:16:58 | INFO | train rank=0 | step=3810/50000 progress=7.6% loss=0.124399 lr=9.981245e-05 lr_backbone=9.981245e-05 elapsed=03:48:21 avg_step=3.60s recent_step=2.86s eta=46:08:32 +2026-09-17 21:17:51 | INFO | train rank=0 | step=3820/50000 progress=7.6% loss=0.165641 lr=9.980957e-05 lr_backbone=9.980957e-05 elapsed=03:49:14 avg_step=3.60s recent_step=5.26s eta=46:11:17 +2026-09-17 21:18:23 | INFO | train rank=0 | step=3830/50000 progress=7.7% loss=0.143727 lr=9.980668e-05 lr_backbone=9.980668e-05 elapsed=03:49:46 avg_step=3.60s recent_step=3.20s eta=46:09:52 +2026-09-17 21:18:51 | INFO | train rank=0 | step=3840/50000 progress=7.7% loss=0.144021 lr=9.980376e-05 lr_backbone=9.980376e-05 elapsed=03:50:14 avg_step=3.60s recent_step=2.82s eta=46:07:43 +2026-09-17 21:19:15 | INFO | train rank=0 | step=3850/50000 progress=7.7% loss=0.129034 lr=9.980083e-05 lr_backbone=9.980083e-05 elapsed=03:50:38 avg_step=3.59s recent_step=2.43s eta=46:04:48 +2026-09-17 21:20:05 | INFO | train rank=0 | step=3860/50000 progress=7.7% loss=0.122359 lr=9.979787e-05 lr_backbone=9.979787e-05 elapsed=03:51:28 avg_step=3.60s recent_step=4.96s eta=46:06:55 +2026-09-17 21:20:33 | INFO | train rank=0 | step=3870/50000 progress=7.7% loss=0.118277 lr=9.979489e-05 lr_backbone=9.979489e-05 elapsed=03:51:56 avg_step=3.60s recent_step=2.83s eta=46:04:47 +2026-09-17 21:21:05 | INFO | train rank=0 | step=3880/50000 progress=7.8% loss=0.122152 lr=9.979188e-05 lr_backbone=9.979188e-05 elapsed=03:52:29 avg_step=3.60s recent_step=3.22s eta=46:03:26 +2026-09-17 21:21:36 | INFO | train rank=0 | step=3890/50000 progress=7.8% loss=0.113417 lr=9.978886e-05 lr_backbone=9.978886e-05 elapsed=03:52:59 avg_step=3.59s recent_step=3.04s eta=46:01:45 +2026-09-17 21:22:23 | INFO | train rank=0 | step=3900/50000 progress=7.8% loss=0.106298 lr=9.978581e-05 lr_backbone=9.978581e-05 elapsed=03:53:46 avg_step=3.60s recent_step=4.72s eta=46:03:22 +2026-09-17 21:22:46 | INFO | train rank=0 | step=3910/50000 progress=7.8% loss=0.093223 lr=9.978274e-05 lr_backbone=9.978274e-05 elapsed=03:54:09 avg_step=3.59s recent_step=2.31s eta=46:00:15 +2026-09-17 21:23:19 | INFO | train rank=0 | step=3920/50000 progress=7.8% loss=0.114842 lr=9.977965e-05 lr_backbone=9.977965e-05 elapsed=03:54:43 avg_step=3.59s recent_step=3.34s eta=45:59:09 +2026-09-17 21:23:50 | INFO | train rank=0 | step=3930/50000 progress=7.9% loss=0.114777 lr=9.977654e-05 lr_backbone=9.977654e-05 elapsed=03:55:13 avg_step=3.59s recent_step=3.02s eta=45:57:26 +2026-09-17 21:24:36 | INFO | train rank=0 | step=3940/50000 progress=7.9% loss=0.195807 lr=9.977341e-05 lr_backbone=9.977341e-05 elapsed=03:56:00 avg_step=3.59s recent_step=4.67s eta=45:58:56 +2026-09-17 21:25:11 | INFO | train rank=0 | step=3950/50000 progress=7.9% loss=0.132548 lr=9.977025e-05 lr_backbone=9.977025e-05 elapsed=03:56:34 avg_step=3.59s recent_step=3.43s eta=45:58:01 +2026-09-17 21:25:45 | INFO | train rank=0 | step=3960/50000 progress=7.9% loss=0.112607 lr=9.976707e-05 lr_backbone=9.976707e-05 elapsed=03:57:09 avg_step=3.59s recent_step=3.47s eta=45:57:11 +2026-09-17 21:26:23 | INFO | train rank=0 | step=3970/50000 progress=7.9% loss=0.106052 lr=9.976387e-05 lr_backbone=9.976387e-05 elapsed=03:57:46 avg_step=3.59s recent_step=3.76s eta=45:56:54 +2026-09-17 21:27:13 | INFO | train rank=0 | step=3980/50000 progress=8.0% loss=0.137587 lr=9.976065e-05 lr_backbone=9.976065e-05 elapsed=03:58:36 avg_step=3.60s recent_step=5.02s eta=45:59:03 +2026-09-17 21:27:43 | INFO | train rank=0 | step=3990/50000 progress=8.0% loss=0.112314 lr=9.975741e-05 lr_backbone=9.975741e-05 elapsed=03:59:07 avg_step=3.60s recent_step=3.02s eta=45:57:20 +2026-09-17 21:28:13 | INFO | train rank=0 | step=4000/50000 progress=8.0% loss=0.141566 lr=9.975415e-05 lr_backbone=9.975415e-05 elapsed=03:59:36 avg_step=3.59s recent_step=2.98s eta=45:55:33 +2026-09-17 21:28:44 | INFO | train rank=0 | step=4010/50000 progress=8.0% loss=0.121843 lr=9.975086e-05 lr_backbone=9.975086e-05 elapsed=04:00:07 avg_step=3.59s recent_step=3.07s eta=45:53:57 +2026-09-17 21:29:27 | INFO | train rank=0 | step=4020/50000 progress=8.0% loss=0.138236 lr=9.974755e-05 lr_backbone=9.974755e-05 elapsed=04:00:50 avg_step=3.59s recent_step=4.30s eta=45:54:42 +2026-09-17 21:29:56 | INFO | train rank=0 | step=4030/50000 progress=8.1% loss=0.106946 lr=9.974422e-05 lr_backbone=9.974422e-05 elapsed=04:01:20 avg_step=3.59s recent_step=2.96s eta=45:52:53 +2026-09-17 21:30:25 | INFO | train rank=0 | step=4040/50000 progress=8.1% loss=0.149674 lr=9.974087e-05 lr_backbone=9.974087e-05 elapsed=04:01:48 avg_step=3.59s recent_step=2.87s eta=45:50:54 +2026-09-17 21:31:00 | INFO | train rank=0 | step=4050/50000 progress=8.1% loss=0.143423 lr=9.973750e-05 lr_backbone=9.973750e-05 elapsed=04:02:23 avg_step=3.59s recent_step=3.52s eta=45:50:11 +2026-09-17 21:31:49 | INFO | train rank=0 | step=4060/50000 progress=8.1% loss=0.131339 lr=9.973410e-05 lr_backbone=9.973410e-05 elapsed=04:03:13 avg_step=3.59s recent_step=4.91s eta=45:52:04 +2026-09-17 21:32:26 | INFO | train rank=0 | step=4070/50000 progress=8.1% loss=0.114107 lr=9.973068e-05 lr_backbone=9.973068e-05 elapsed=04:03:49 avg_step=3.59s recent_step=3.65s eta=45:51:34 +2026-09-17 21:32:52 | INFO | train rank=0 | step=4080/50000 progress=8.2% loss=0.107006 lr=9.972725e-05 lr_backbone=9.972725e-05 elapsed=04:04:15 avg_step=3.59s recent_step=2.63s eta=45:49:10 +2026-09-17 21:33:29 | INFO | train rank=0 | step=4090/50000 progress=8.2% loss=0.144683 lr=9.972379e-05 lr_backbone=9.972379e-05 elapsed=04:04:53 avg_step=3.59s recent_step=3.74s eta=45:48:50 +2026-09-17 21:34:14 | INFO | train rank=0 | step=4100/50000 progress=8.2% loss=0.117976 lr=9.972030e-05 lr_backbone=9.972030e-05 elapsed=04:05:38 avg_step=3.59s recent_step=4.51s eta=45:49:57 +2026-09-17 21:34:42 | INFO | train rank=0 | step=4110/50000 progress=8.2% loss=0.113725 lr=9.971680e-05 lr_backbone=9.971680e-05 elapsed=04:06:05 avg_step=3.59s recent_step=2.77s eta=45:47:49 +2026-09-17 21:35:10 | INFO | train rank=0 | step=4120/50000 progress=8.2% loss=0.120132 lr=9.971327e-05 lr_backbone=9.971327e-05 elapsed=04:06:33 avg_step=3.59s recent_step=2.73s eta=45:45:37 +2026-09-17 21:35:53 | INFO | train rank=0 | step=4130/50000 progress=8.3% loss=0.119089 lr=9.970973e-05 lr_backbone=9.970973e-05 elapsed=04:07:17 avg_step=3.59s recent_step=4.38s eta=45:46:28 +2026-09-17 21:36:26 | INFO | train rank=0 | step=4140/50000 progress=8.3% loss=0.134794 lr=9.970616e-05 lr_backbone=9.970616e-05 elapsed=04:07:49 avg_step=3.59s recent_step=3.28s eta=45:45:18 +2026-09-17 21:37:00 | INFO | train rank=0 | step=4150/50000 progress=8.3% loss=0.112929 lr=9.970257e-05 lr_backbone=9.970257e-05 elapsed=04:08:23 avg_step=3.59s recent_step=3.40s eta=45:44:21 +2026-09-17 21:37:36 | INFO | train rank=0 | step=4160/50000 progress=8.3% loss=0.135971 lr=9.969895e-05 lr_backbone=9.969895e-05 elapsed=04:08:59 avg_step=3.59s recent_step=3.55s eta=45:43:41 +2026-09-17 21:38:19 | INFO | train rank=0 | step=4170/50000 progress=8.3% loss=0.160384 lr=9.969532e-05 lr_backbone=9.969532e-05 elapsed=04:09:42 avg_step=3.59s recent_step=4.35s eta=45:44:28 +2026-09-17 21:38:49 | INFO | train rank=0 | step=4180/50000 progress=8.4% loss=0.106137 lr=9.969166e-05 lr_backbone=9.969166e-05 elapsed=04:10:12 avg_step=3.59s recent_step=2.99s eta=45:42:46 +2026-09-17 21:39:29 | INFO | train rank=0 | step=4190/50000 progress=8.4% loss=0.138561 lr=9.968799e-05 lr_backbone=9.968799e-05 elapsed=04:10:53 avg_step=3.59s recent_step=4.03s eta=45:42:59 +2026-09-17 21:40:00 | INFO | train rank=0 | step=4200/50000 progress=8.4% loss=0.118044 lr=9.968429e-05 lr_backbone=9.968429e-05 elapsed=04:11:23 avg_step=3.59s recent_step=3.02s eta=45:41:20 +2026-09-17 21:40:48 | INFO | train rank=0 | step=4210/50000 progress=8.4% loss=0.117300 lr=9.968057e-05 lr_backbone=9.968057e-05 elapsed=04:12:12 avg_step=3.59s recent_step=4.87s eta=45:43:03 +2026-09-17 21:41:11 | INFO | train rank=0 | step=4220/50000 progress=8.4% loss=0.126079 lr=9.967682e-05 lr_backbone=9.967682e-05 elapsed=04:12:35 avg_step=3.59s recent_step=2.30s eta=45:40:06 +2026-09-17 21:41:32 | INFO | train rank=0 | step=4230/50000 progress=8.5% loss=0.144099 lr=9.967306e-05 lr_backbone=9.967306e-05 elapsed=04:12:56 avg_step=3.59s recent_step=2.10s eta=45:36:49 +2026-09-17 21:42:07 | INFO | train rank=0 | step=4240/50000 progress=8.5% loss=0.115817 lr=9.966927e-05 lr_backbone=9.966927e-05 elapsed=04:13:30 avg_step=3.59s recent_step=3.46s eta=45:36:00 +2026-09-17 21:42:58 | INFO | train rank=0 | step=4250/50000 progress=8.5% loss=0.195708 lr=9.966546e-05 lr_backbone=9.966546e-05 elapsed=04:14:21 avg_step=3.59s recent_step=5.12s eta=45:38:09 +2026-09-17 21:43:30 | INFO | train rank=0 | step=4260/50000 progress=8.5% loss=0.128916 lr=9.966163e-05 lr_backbone=9.966163e-05 elapsed=04:14:54 avg_step=3.59s recent_step=3.23s eta=45:36:54 +2026-09-17 21:44:04 | INFO | train rank=0 | step=4270/50000 progress=8.5% loss=0.130930 lr=9.965778e-05 lr_backbone=9.965778e-05 elapsed=04:15:27 avg_step=3.59s recent_step=3.38s eta=45:35:56 +2026-09-17 21:44:34 | INFO | train rank=0 | step=4280/50000 progress=8.6% loss=0.135258 lr=9.965391e-05 lr_backbone=9.965391e-05 elapsed=04:15:58 avg_step=3.59s recent_step=3.02s eta=45:34:20 +2026-09-17 21:45:27 | INFO | train rank=0 | step=4290/50000 progress=8.6% loss=0.145592 lr=9.965001e-05 lr_backbone=9.965001e-05 elapsed=04:16:50 avg_step=3.59s recent_step=5.26s eta=45:36:42 +2026-09-17 21:46:01 | INFO | train rank=0 | step=4300/50000 progress=8.6% loss=0.215596 lr=9.964610e-05 lr_backbone=9.964610e-05 elapsed=04:17:25 avg_step=3.59s recent_step=3.45s eta=45:35:51 +2026-09-17 21:46:28 | INFO | train rank=0 | step=4310/50000 progress=8.6% loss=0.129512 lr=9.964216e-05 lr_backbone=9.964216e-05 elapsed=04:17:51 avg_step=3.59s recent_step=2.61s eta=45:33:30 +2026-09-17 21:46:58 | INFO | train rank=0 | step=4320/50000 progress=8.6% loss=0.128749 lr=9.963820e-05 lr_backbone=9.963820e-05 elapsed=04:18:22 avg_step=3.59s recent_step=3.08s eta=45:32:01 +2026-09-17 21:47:43 | INFO | train rank=0 | step=4330/50000 progress=8.7% loss=0.101645 lr=9.963422e-05 lr_backbone=9.963422e-05 elapsed=04:19:07 avg_step=3.59s recent_step=4.49s eta=45:32:59 +2026-09-17 21:48:21 | INFO | train rank=0 | step=4340/50000 progress=8.7% loss=0.130861 lr=9.963021e-05 lr_backbone=9.963021e-05 elapsed=04:19:44 avg_step=3.59s recent_step=3.73s eta=45:32:38 +2026-09-17 21:48:53 | INFO | train rank=0 | step=4350/50000 progress=8.7% loss=0.119612 lr=9.962619e-05 lr_backbone=9.962619e-05 elapsed=04:20:16 avg_step=3.59s recent_step=3.20s eta=45:31:21 +2026-09-17 21:49:24 | INFO | train rank=0 | step=4360/50000 progress=8.7% loss=0.139032 lr=9.962214e-05 lr_backbone=9.962214e-05 elapsed=04:20:48 avg_step=3.59s recent_step=3.18s eta=45:30:02 +2026-09-17 21:50:17 | INFO | train rank=0 | step=4370/50000 progress=8.7% loss=0.128687 lr=9.961807e-05 lr_backbone=9.961807e-05 elapsed=04:21:40 avg_step=3.59s recent_step=5.22s eta=45:32:17 +2026-09-17 21:50:47 | INFO | train rank=0 | step=4380/50000 progress=8.8% loss=0.120298 lr=9.961398e-05 lr_backbone=9.961398e-05 elapsed=04:22:10 avg_step=3.59s recent_step=3.02s eta=45:30:42 +2026-09-17 21:51:21 | INFO | train rank=0 | step=4390/50000 progress=8.8% loss=0.119728 lr=9.960987e-05 lr_backbone=9.960987e-05 elapsed=04:22:45 avg_step=3.59s recent_step=3.45s eta=45:29:51 +2026-09-17 21:51:54 | INFO | train rank=0 | step=4400/50000 progress=8.8% loss=0.124279 lr=9.960574e-05 lr_backbone=9.960574e-05 elapsed=04:23:18 avg_step=3.59s recent_step=3.32s eta=45:28:46 +2026-09-17 21:52:45 | INFO | train rank=0 | step=4410/50000 progress=8.8% loss=0.144130 lr=9.960158e-05 lr_backbone=9.960158e-05 elapsed=04:24:08 avg_step=3.59s recent_step=5.04s eta=45:30:40 +2026-09-17 21:53:22 | INFO | train rank=0 | step=4420/50000 progress=8.8% loss=0.132519 lr=9.959740e-05 lr_backbone=9.959740e-05 elapsed=04:24:46 avg_step=3.59s recent_step=3.76s eta=45:30:22 +2026-09-17 21:53:54 | INFO | train rank=0 | step=4430/50000 progress=8.9% loss=0.136810 lr=9.959320e-05 lr_backbone=9.959320e-05 elapsed=04:25:17 avg_step=3.59s recent_step=3.14s eta=45:28:59 +2026-09-17 21:54:23 | INFO | train rank=0 | step=4440/50000 progress=8.9% loss=0.126895 lr=9.958898e-05 lr_backbone=9.958898e-05 elapsed=04:25:46 avg_step=3.59s recent_step=2.91s eta=45:27:13 +2026-09-17 21:55:04 | INFO | train rank=0 | step=4450/50000 progress=8.9% loss=0.135948 lr=9.958474e-05 lr_backbone=9.958474e-05 elapsed=04:26:27 avg_step=3.59s recent_step=4.08s eta=45:27:28 +2026-09-17 21:55:37 | INFO | train rank=0 | step=4460/50000 progress=8.9% loss=0.118412 lr=9.958048e-05 lr_backbone=9.958048e-05 elapsed=04:27:00 avg_step=3.59s recent_step=3.28s eta=45:26:20 +2026-09-17 21:56:10 | INFO | train rank=0 | step=4470/50000 progress=8.9% loss=0.153528 lr=9.957619e-05 lr_backbone=9.957619e-05 elapsed=04:27:33 avg_step=3.59s recent_step=3.30s eta=45:25:14 +2026-09-17 21:56:45 | INFO | train rank=0 | step=4480/50000 progress=9.0% loss=0.169340 lr=9.957188e-05 lr_backbone=9.957188e-05 elapsed=04:28:09 avg_step=3.59s recent_step=3.57s eta=45:24:36 +2026-09-17 21:57:28 | INFO | train rank=0 | step=4490/50000 progress=9.0% loss=0.090261 lr=9.956755e-05 lr_backbone=9.956755e-05 elapsed=04:28:51 avg_step=3.59s recent_step=4.24s eta=45:25:06 +2026-09-17 21:57:58 | INFO | train rank=0 | step=4500/50000 progress=9.0% loss=0.096536 lr=9.956320e-05 lr_backbone=9.956320e-05 elapsed=04:29:21 avg_step=3.59s recent_step=3.00s eta=45:23:31 +2026-09-17 21:58:26 | INFO | train rank=0 | step=4510/50000 progress=9.0% loss=0.157320 lr=9.955883e-05 lr_backbone=9.955883e-05 elapsed=04:29:49 avg_step=3.59s recent_step=2.84s eta=45:21:39 +2026-09-17 21:59:09 | INFO | train rank=0 | step=4520/50000 progress=9.0% loss=0.126404 lr=9.955444e-05 lr_backbone=9.955444e-05 elapsed=04:30:33 avg_step=3.59s recent_step=4.33s eta=45:22:18 +2026-09-17 21:59:38 | INFO | train rank=0 | step=4530/50000 progress=9.1% loss=0.118267 lr=9.955002e-05 lr_backbone=9.955002e-05 elapsed=04:31:01 avg_step=3.59s recent_step=2.86s eta=45:20:28 +2026-09-17 22:00:12 | INFO | train rank=0 | step=4540/50000 progress=9.1% loss=0.123178 lr=9.954558e-05 lr_backbone=9.954558e-05 elapsed=04:31:35 avg_step=3.59s recent_step=3.36s eta=45:19:30 +2026-09-17 22:00:42 | INFO | train rank=0 | step=4550/50000 progress=9.1% loss=0.152937 lr=9.954112e-05 lr_backbone=9.954112e-05 elapsed=04:32:05 avg_step=3.59s recent_step=3.01s eta=45:17:56 +2026-09-17 22:01:33 | INFO | train rank=0 | step=4560/50000 progress=9.1% loss=0.126917 lr=9.953664e-05 lr_backbone=9.953664e-05 elapsed=04:32:57 avg_step=3.59s recent_step=5.16s eta=45:19:57 +2026-09-17 22:02:05 | INFO | train rank=0 | step=4570/50000 progress=9.1% loss=0.237188 lr=9.953214e-05 lr_backbone=9.953214e-05 elapsed=04:33:28 avg_step=3.59s recent_step=3.12s eta=45:18:34 +2026-09-17 22:02:40 | INFO | train rank=0 | step=4580/50000 progress=9.2% loss=0.164585 lr=9.952762e-05 lr_backbone=9.952762e-05 elapsed=04:34:03 avg_step=3.59s recent_step=3.54s eta=45:17:53 +2026-09-17 22:03:17 | INFO | train rank=0 | step=4590/50000 progress=9.2% loss=0.097453 lr=9.952307e-05 lr_backbone=9.952307e-05 elapsed=04:34:40 avg_step=3.59s recent_step=3.67s eta=45:17:25 +2026-09-17 22:03:49 | INFO | train rank=0 | step=4600/50000 progress=9.2% loss=0.094412 lr=9.951850e-05 lr_backbone=9.951850e-05 elapsed=04:35:12 avg_step=3.59s recent_step=3.20s eta=45:16:10 +2026-09-17 22:04:17 | INFO | train rank=0 | step=4610/50000 progress=9.2% loss=0.109933 lr=9.951391e-05 lr_backbone=9.951391e-05 elapsed=04:35:41 avg_step=3.59s recent_step=2.87s eta=45:14:23 +2026-09-17 22:04:50 | INFO | train rank=0 | step=4620/50000 progress=9.2% loss=0.103491 lr=9.950930e-05 lr_backbone=9.950930e-05 elapsed=04:36:13 avg_step=3.59s recent_step=3.27s eta=45:13:16 +2026-09-17 22:05:28 | INFO | train rank=0 | step=4630/50000 progress=9.3% loss=0.131520 lr=9.950467e-05 lr_backbone=9.950467e-05 elapsed=04:36:51 avg_step=3.59s recent_step=3.81s eta=45:13:02 +2026-09-17 22:05:59 | INFO | train rank=0 | step=4640/50000 progress=9.3% loss=0.129604 lr=9.950002e-05 lr_backbone=9.950002e-05 elapsed=04:37:23 avg_step=3.59s recent_step=3.12s eta=45:11:40 +2026-09-17 22:06:34 | INFO | train rank=0 | step=4650/50000 progress=9.3% loss=0.132962 lr=9.949534e-05 lr_backbone=9.949534e-05 elapsed=04:37:57 avg_step=3.59s recent_step=3.44s eta=45:10:50 +2026-09-17 22:07:09 | INFO | train rank=0 | step=4660/50000 progress=9.3% loss=0.121386 lr=9.949064e-05 lr_backbone=9.949064e-05 elapsed=04:38:33 avg_step=3.59s recent_step=3.57s eta=45:10:12 +2026-09-17 22:07:52 | INFO | train rank=0 | step=4670/50000 progress=9.3% loss=0.100329 lr=9.948593e-05 lr_backbone=9.948593e-05 elapsed=04:39:15 avg_step=3.59s recent_step=4.27s eta=45:10:42 +2026-09-17 22:08:27 | INFO | train rank=0 | step=4680/50000 progress=9.4% loss=0.123010 lr=9.948118e-05 lr_backbone=9.948118e-05 elapsed=04:39:50 avg_step=3.59s recent_step=3.51s eta=45:09:59 +2026-09-17 22:09:12 | INFO | train rank=0 | step=4690/50000 progress=9.4% loss=0.139277 lr=9.947642e-05 lr_backbone=9.947642e-05 elapsed=04:40:35 avg_step=3.59s recent_step=4.50s eta=45:10:51 +2026-09-17 22:09:46 | INFO | train rank=0 | step=4700/50000 progress=9.4% loss=0.116432 lr=9.947164e-05 lr_backbone=9.947164e-05 elapsed=04:41:09 avg_step=3.59s recent_step=3.36s eta=45:09:53 +2026-09-17 22:10:20 | INFO | train rank=0 | step=4710/50000 progress=9.4% loss=0.123799 lr=9.946683e-05 lr_backbone=9.946683e-05 elapsed=04:41:43 avg_step=3.59s recent_step=3.43s eta=45:09:01 +2026-09-17 22:10:54 | INFO | train rank=0 | step=4720/50000 progress=9.4% loss=0.112100 lr=9.946201e-05 lr_backbone=9.946201e-05 elapsed=04:42:18 avg_step=3.59s recent_step=3.43s eta=45:08:10 +2026-09-17 22:11:42 | INFO | train rank=0 | step=4730/50000 progress=9.5% loss=0.125322 lr=9.945716e-05 lr_backbone=9.945716e-05 elapsed=04:43:05 avg_step=3.59s recent_step=4.77s eta=45:09:28 +2026-09-17 22:12:16 | INFO | train rank=0 | step=4740/50000 progress=9.5% loss=0.145139 lr=9.945229e-05 lr_backbone=9.945229e-05 elapsed=04:43:39 avg_step=3.59s recent_step=3.42s eta=45:08:35 +2026-09-17 22:12:48 | INFO | train rank=0 | step=4750/50000 progress=9.5% loss=0.129684 lr=9.944739e-05 lr_backbone=9.944739e-05 elapsed=04:44:11 avg_step=3.59s recent_step=3.14s eta=45:07:16 +2026-09-17 22:13:18 | INFO | train rank=0 | step=4760/50000 progress=9.5% loss=0.187453 lr=9.944248e-05 lr_backbone=9.944248e-05 elapsed=04:44:41 avg_step=3.59s recent_step=3.01s eta=45:05:46 +2026-09-17 22:14:07 | INFO | train rank=0 | step=4770/50000 progress=9.5% loss=0.122605 lr=9.943754e-05 lr_backbone=9.943754e-05 elapsed=04:45:30 avg_step=3.59s recent_step=4.94s eta=45:07:18 +2026-09-17 22:14:36 | INFO | train rank=0 | step=4780/50000 progress=9.6% loss=0.137371 lr=9.943259e-05 lr_backbone=9.943259e-05 elapsed=04:46:00 avg_step=3.59s recent_step=2.94s eta=45:05:40 +2026-09-17 22:15:09 | INFO | train rank=0 | step=4790/50000 progress=9.6% loss=0.136982 lr=9.942761e-05 lr_backbone=9.942761e-05 elapsed=04:46:32 avg_step=3.59s recent_step=3.24s eta=45:04:31 +2026-09-17 22:15:38 | INFO | train rank=0 | step=4800/50000 progress=9.6% loss=0.109161 lr=9.942261e-05 lr_backbone=9.942261e-05 elapsed=04:47:01 avg_step=3.59s recent_step=2.87s eta=45:02:48 +2026-09-17 22:16:33 | INFO | train rank=0 | step=4810/50000 progress=9.6% loss=0.119540 lr=9.941759e-05 lr_backbone=9.941759e-05 elapsed=04:47:56 avg_step=3.59s recent_step=5.54s eta=45:05:15 +2026-09-17 22:16:59 | INFO | train rank=0 | step=4820/50000 progress=9.6% loss=0.132010 lr=9.941254e-05 lr_backbone=9.941254e-05 elapsed=04:48:22 avg_step=3.59s recent_step=2.60s eta=45:03:06 +2026-09-17 22:17:32 | INFO | train rank=0 | step=4830/50000 progress=9.7% loss=0.122260 lr=9.940748e-05 lr_backbone=9.940748e-05 elapsed=04:48:55 avg_step=3.59s recent_step=3.28s eta=45:02:01 +2026-09-17 22:18:19 | INFO | train rank=0 | step=4840/50000 progress=9.7% loss=0.108372 lr=9.940239e-05 lr_backbone=9.940239e-05 elapsed=04:49:43 avg_step=3.59s recent_step=4.75s eta=45:03:13 +2026-09-17 22:18:45 | INFO | train rank=0 | step=4850/50000 progress=9.7% loss=0.121136 lr=9.939728e-05 lr_backbone=9.939728e-05 elapsed=04:50:08 avg_step=3.59s recent_step=2.57s eta=45:01:02 +2026-09-17 22:19:19 | INFO | train rank=0 | step=4860/50000 progress=9.7% loss=0.105571 lr=9.939215e-05 lr_backbone=9.939215e-05 elapsed=04:50:43 avg_step=3.59s recent_step=3.46s eta=45:00:13 +2026-09-17 22:19:53 | INFO | train rank=0 | step=4870/50000 progress=9.7% loss=0.129658 lr=9.938700e-05 lr_backbone=9.938700e-05 elapsed=04:51:17 avg_step=3.59s recent_step=3.39s eta=44:59:19 +2026-09-17 22:20:47 | INFO | train rank=0 | step=4880/50000 progress=9.8% loss=0.238217 lr=9.938183e-05 lr_backbone=9.938183e-05 elapsed=04:52:11 avg_step=3.59s recent_step=5.40s eta=45:01:30 +2026-09-17 22:21:13 | INFO | train rank=0 | step=4890/50000 progress=9.8% loss=0.130024 lr=9.937663e-05 lr_backbone=9.937663e-05 elapsed=04:52:36 avg_step=3.59s recent_step=2.54s eta=44:59:17 +2026-09-17 22:21:46 | INFO | train rank=0 | step=4900/50000 progress=9.8% loss=0.152036 lr=9.937142e-05 lr_backbone=9.937142e-05 elapsed=04:53:09 avg_step=3.59s recent_step=3.33s eta=44:58:18 +2026-09-17 22:22:19 | INFO | train rank=0 | step=4910/50000 progress=9.8% loss=0.115988 lr=9.936618e-05 lr_backbone=9.936618e-05 elapsed=04:53:42 avg_step=3.59s recent_step=3.29s eta=44:57:15 +2026-09-17 22:23:09 | INFO | train rank=0 | step=4920/50000 progress=9.8% loss=0.142185 lr=9.936092e-05 lr_backbone=9.936092e-05 elapsed=04:54:32 avg_step=3.59s recent_step=4.97s eta=44:58:45 +2026-09-17 22:23:45 | INFO | train rank=0 | step=4930/50000 progress=9.9% loss=0.087148 lr=9.935564e-05 lr_backbone=9.935564e-05 elapsed=04:55:08 avg_step=3.59s recent_step=3.59s eta=44:58:09 +2026-09-17 22:24:13 | INFO | train rank=0 | step=4940/50000 progress=9.9% loss=0.090094 lr=9.935034e-05 lr_backbone=9.935034e-05 elapsed=04:55:37 avg_step=3.59s recent_step=2.87s eta=44:56:28 +2026-09-17 22:24:45 | INFO | train rank=0 | step=4950/50000 progress=9.9% loss=0.131427 lr=9.934501e-05 lr_backbone=9.934501e-05 elapsed=04:56:09 avg_step=3.59s recent_step=3.21s eta=44:55:18 +2026-09-17 22:25:33 | INFO | train rank=0 | step=4960/50000 progress=9.9% loss=0.110880 lr=9.933966e-05 lr_backbone=9.933966e-05 elapsed=04:56:56 avg_step=3.59s recent_step=4.76s eta=44:56:28 +2026-09-17 22:26:04 | INFO | train rank=0 | step=4970/50000 progress=9.9% loss=0.132969 lr=9.933430e-05 lr_backbone=9.933430e-05 elapsed=04:57:27 avg_step=3.59s recent_step=3.11s eta=44:55:08 +2026-09-17 22:26:38 | INFO | train rank=0 | step=4980/50000 progress=10.0% loss=0.098015 lr=9.932891e-05 lr_backbone=9.932891e-05 elapsed=04:58:01 avg_step=3.59s recent_step=3.40s eta=44:54:15 +2026-09-17 22:27:10 | INFO | train rank=0 | step=4990/50000 progress=10.0% loss=0.115576 lr=9.932350e-05 lr_backbone=9.932350e-05 elapsed=04:58:33 avg_step=3.59s recent_step=3.17s eta=44:53:02 +2026-09-17 22:28:02 | INFO | train rank=0 | step=5000/50000 progress=10.0% loss=0.137739 lr=9.931807e-05 lr_backbone=9.931807e-05 elapsed=04:59:25 avg_step=3.59s recent_step=5.22s eta=44:54:53 +2026-09-17 22:28:32 | INFO | train rank=0 | step=5010/50000 progress=10.0% loss=0.115781 lr=9.931261e-05 lr_backbone=9.931261e-05 elapsed=04:59:56 avg_step=3.59s recent_step=3.03s eta=44:53:26 +2026-09-17 22:28:59 | INFO | train rank=0 | step=5020/50000 progress=10.0% loss=0.100251 lr=9.930714e-05 lr_backbone=9.930714e-05 elapsed=05:00:22 avg_step=3.59s recent_step=2.67s eta=44:51:28 +2026-09-17 22:29:30 | INFO | train rank=0 | step=5030/50000 progress=10.1% loss=0.127411 lr=9.930164e-05 lr_backbone=9.930164e-05 elapsed=05:00:53 avg_step=3.59s recent_step=3.06s eta=44:50:05 +2026-09-17 22:30:27 | INFO | train rank=0 | step=5040/50000 progress=10.1% loss=0.127680 lr=9.929612e-05 lr_backbone=9.929612e-05 elapsed=05:01:50 avg_step=3.59s recent_step=5.70s eta=44:52:37 +2026-09-17 22:30:59 | INFO | train rank=0 | step=5050/50000 progress=10.1% loss=0.117883 lr=9.929058e-05 lr_backbone=9.929058e-05 elapsed=05:02:22 avg_step=3.59s recent_step=3.22s eta=44:51:28 +2026-09-17 22:31:29 | INFO | train rank=0 | step=5060/50000 progress=10.1% loss=0.123718 lr=9.928502e-05 lr_backbone=9.928502e-05 elapsed=05:02:52 avg_step=3.59s recent_step=2.99s eta=44:49:58 +2026-09-17 22:32:02 | INFO | train rank=0 | step=5070/50000 progress=10.1% loss=0.111957 lr=9.927944e-05 lr_backbone=9.927944e-05 elapsed=05:03:26 avg_step=3.59s recent_step=3.35s eta=44:49:00 +2026-09-17 22:32:55 | INFO | train rank=0 | step=5080/50000 progress=10.2% loss=0.118657 lr=9.927383e-05 lr_backbone=9.927383e-05 elapsed=05:04:18 avg_step=3.59s recent_step=5.23s eta=44:50:49 +2026-09-17 22:33:19 | INFO | train rank=0 | step=5090/50000 progress=10.2% loss=0.137751 lr=9.926820e-05 lr_backbone=9.926820e-05 elapsed=05:04:42 avg_step=3.59s recent_step=2.44s eta=44:48:32 +2026-09-17 22:33:45 | INFO | train rank=0 | step=5100/50000 progress=10.2% loss=0.123344 lr=9.926256e-05 lr_backbone=9.926256e-05 elapsed=05:05:08 avg_step=3.59s recent_step=2.59s eta=44:46:27 +2026-09-17 22:34:11 | INFO | train rank=0 | step=5110/50000 progress=10.2% loss=0.175592 lr=9.925689e-05 lr_backbone=9.925689e-05 elapsed=05:05:34 avg_step=3.59s recent_step=2.59s eta=44:44:24 +2026-09-17 22:34:56 | INFO | train rank=0 | step=5120/50000 progress=10.2% loss=0.123963 lr=9.925120e-05 lr_backbone=9.925120e-05 elapsed=05:06:20 avg_step=3.59s recent_step=4.56s eta=44:45:13 +2026-09-17 22:35:29 | INFO | train rank=0 | step=5130/50000 progress=10.3% loss=0.115851 lr=9.924548e-05 lr_backbone=9.924548e-05 elapsed=05:06:53 avg_step=3.59s recent_step=3.31s eta=44:44:13 +2026-09-17 22:36:01 | INFO | train rank=0 | step=5140/50000 progress=10.3% loss=0.129655 lr=9.923975e-05 lr_backbone=9.923975e-05 elapsed=05:07:25 avg_step=3.59s recent_step=3.19s eta=44:43:02 +2026-09-17 22:36:30 | INFO | train rank=0 | step=5150/50000 progress=10.3% loss=0.121103 lr=9.923399e-05 lr_backbone=9.923399e-05 elapsed=05:07:53 avg_step=3.59s recent_step=2.86s eta=44:41:22 +2026-09-17 22:37:21 | INFO | train rank=0 | step=5160/50000 progress=10.3% loss=0.106828 lr=9.922822e-05 lr_backbone=9.922822e-05 elapsed=05:08:44 avg_step=3.59s recent_step=5.07s eta=44:42:55 +2026-09-17 22:37:54 | INFO | train rank=0 | step=5170/50000 progress=10.3% loss=0.110219 lr=9.922242e-05 lr_backbone=9.922242e-05 elapsed=05:09:17 avg_step=3.59s recent_step=3.34s eta=44:41:58 +2026-09-17 22:38:23 | INFO | train rank=0 | step=5180/50000 progress=10.4% loss=0.114630 lr=9.921660e-05 lr_backbone=9.921660e-05 elapsed=05:09:47 avg_step=3.59s recent_step=2.93s eta=44:40:25 +2026-09-17 22:39:02 | INFO | train rank=0 | step=5190/50000 progress=10.4% loss=0.135789 lr=9.921076e-05 lr_backbone=9.921076e-05 elapsed=05:10:25 avg_step=3.59s recent_step=3.88s eta=44:40:15 +2026-09-17 22:39:41 | INFO | train rank=0 | step=5200/50000 progress=10.4% loss=0.119636 lr=9.920489e-05 lr_backbone=9.920489e-05 elapsed=05:11:04 avg_step=3.59s recent_step=3.89s eta=44:40:04 +2026-09-17 22:40:10 | INFO | train rank=0 | step=5210/50000 progress=10.4% loss=0.127090 lr=9.919901e-05 lr_backbone=9.919901e-05 elapsed=05:11:34 avg_step=3.59s recent_step=2.92s eta=44:38:31 +2026-09-17 22:40:40 | INFO | train rank=0 | step=5220/50000 progress=10.4% loss=0.146051 lr=9.919310e-05 lr_backbone=9.919310e-05 elapsed=05:12:03 avg_step=3.59s recent_step=2.93s eta=44:36:59 +2026-09-17 22:41:09 | INFO | train rank=0 | step=5230/50000 progress=10.5% loss=0.097419 lr=9.918717e-05 lr_backbone=9.918717e-05 elapsed=05:12:32 avg_step=3.59s recent_step=2.95s eta=44:35:28 +2026-09-17 22:41:46 | INFO | train rank=0 | step=5240/50000 progress=10.5% loss=0.196788 lr=9.918122e-05 lr_backbone=9.918122e-05 elapsed=05:13:10 avg_step=3.59s recent_step=3.73s eta=44:35:05 +2026-09-17 22:42:18 | INFO | train rank=0 | step=5250/50000 progress=10.5% loss=0.156885 lr=9.917525e-05 lr_backbone=9.917525e-05 elapsed=05:13:42 avg_step=3.59s recent_step=3.20s eta=44:33:56 +2026-09-17 22:42:54 | INFO | train rank=0 | step=5260/50000 progress=10.5% loss=0.122706 lr=9.916926e-05 lr_backbone=9.916926e-05 elapsed=05:14:17 avg_step=3.59s recent_step=3.54s eta=44:33:16 +2026-09-17 22:43:45 | INFO | train rank=0 | step=5270/50000 progress=10.5% loss=0.108548 lr=9.916325e-05 lr_backbone=9.916325e-05 elapsed=05:15:08 avg_step=3.59s recent_step=5.14s eta=44:34:52 +2026-09-17 22:44:22 | INFO | train rank=0 | step=5280/50000 progress=10.6% loss=0.155648 lr=9.915721e-05 lr_backbone=9.915721e-05 elapsed=05:15:45 avg_step=3.59s recent_step=3.67s eta=44:34:24 +2026-09-17 22:44:57 | INFO | train rank=0 | step=5290/50000 progress=10.6% loss=0.151368 lr=9.915116e-05 lr_backbone=9.915116e-05 elapsed=05:16:20 avg_step=3.59s recent_step=3.52s eta=44:33:42 +2026-09-17 22:45:29 | INFO | train rank=0 | step=5300/50000 progress=10.6% loss=0.123556 lr=9.914508e-05 lr_backbone=9.914508e-05 elapsed=05:16:52 avg_step=3.59s recent_step=3.21s eta=44:32:34 +2026-09-17 22:46:10 | INFO | train rank=0 | step=5310/50000 progress=10.6% loss=0.127000 lr=9.913898e-05 lr_backbone=9.913898e-05 elapsed=05:17:33 avg_step=3.59s recent_step=4.10s eta=44:32:41 +2026-09-17 22:46:43 | INFO | train rank=0 | step=5320/50000 progress=10.6% loss=0.111952 lr=9.913286e-05 lr_backbone=9.913286e-05 elapsed=05:18:07 avg_step=3.59s recent_step=3.32s eta=44:31:43 +2026-09-17 22:47:15 | INFO | train rank=0 | step=5330/50000 progress=10.7% loss=0.121753 lr=9.912671e-05 lr_backbone=9.912671e-05 elapsed=05:18:39 avg_step=3.59s recent_step=3.21s eta=44:30:35 +2026-09-17 22:47:48 | INFO | train rank=0 | step=5340/50000 progress=10.7% loss=0.120724 lr=9.912055e-05 lr_backbone=9.912055e-05 elapsed=05:19:11 avg_step=3.59s recent_step=3.26s eta=44:29:32 +2026-09-17 22:48:35 | INFO | train rank=0 | step=5350/50000 progress=10.7% loss=0.109851 lr=9.911436e-05 lr_backbone=9.911436e-05 elapsed=05:19:59 avg_step=3.59s recent_step=4.73s eta=44:30:32 +2026-09-17 22:49:12 | INFO | train rank=0 | step=5360/50000 progress=10.7% loss=0.142573 lr=9.910816e-05 lr_backbone=9.910816e-05 elapsed=05:20:35 avg_step=3.59s recent_step=3.63s eta=44:29:59 +2026-09-17 22:49:48 | INFO | train rank=0 | step=5370/50000 progress=10.7% loss=0.128748 lr=9.910193e-05 lr_backbone=9.910193e-05 elapsed=05:21:11 avg_step=3.59s recent_step=3.61s eta=44:29:25 +2026-09-17 22:50:24 | INFO | train rank=0 | step=5380/50000 progress=10.8% loss=0.154587 lr=9.909568e-05 lr_backbone=9.909568e-05 elapsed=05:21:47 avg_step=3.59s recent_step=3.58s eta=44:28:48 +2026-09-17 22:51:04 | INFO | train rank=0 | step=5390/50000 progress=10.8% loss=0.102697 lr=9.908940e-05 lr_backbone=9.908940e-05 elapsed=05:22:28 avg_step=3.59s recent_step=4.08s eta=44:28:53 +2026-09-17 22:51:49 | INFO | train rank=0 | step=5400/50000 progress=10.8% loss=0.117302 lr=9.908311e-05 lr_backbone=9.908311e-05 elapsed=05:23:12 avg_step=3.59s recent_step=4.44s eta=44:29:28 +2026-09-17 22:52:18 | INFO | train rank=0 | step=5410/50000 progress=10.8% loss=0.120171 lr=9.907680e-05 lr_backbone=9.907680e-05 elapsed=05:23:41 avg_step=3.59s recent_step=2.93s eta=44:27:58 +2026-09-17 22:52:57 | INFO | train rank=0 | step=5420/50000 progress=10.8% loss=0.091535 lr=9.907046e-05 lr_backbone=9.907046e-05 elapsed=05:24:20 avg_step=3.59s recent_step=3.84s eta=44:27:43 +2026-09-17 22:53:28 | INFO | train rank=0 | step=5430/50000 progress=10.9% loss=0.112014 lr=9.906410e-05 lr_backbone=9.906410e-05 elapsed=05:24:51 avg_step=3.59s recent_step=3.10s eta=44:26:27 +2026-09-17 22:54:18 | INFO | train rank=0 | step=5440/50000 progress=10.9% loss=0.115572 lr=9.905772e-05 lr_backbone=9.905772e-05 elapsed=05:25:42 avg_step=3.59s recent_step=5.07s eta=44:27:52 +2026-09-17 22:54:45 | INFO | train rank=0 | step=5450/50000 progress=10.9% loss=0.090898 lr=9.905132e-05 lr_backbone=9.905132e-05 elapsed=05:26:09 avg_step=3.59s recent_step=2.71s eta=44:26:05 +2026-09-17 22:55:25 | INFO | train rank=0 | step=5460/50000 progress=10.9% loss=0.119255 lr=9.904490e-05 lr_backbone=9.904490e-05 elapsed=05:26:48 avg_step=3.59s recent_step=3.95s eta=44:25:58 +2026-09-17 22:55:59 | INFO | train rank=0 | step=5470/50000 progress=10.9% loss=0.124645 lr=9.903846e-05 lr_backbone=9.903846e-05 elapsed=05:27:23 avg_step=3.59s recent_step=3.43s eta=44:25:09 +2026-09-17 22:56:42 | INFO | train rank=0 | step=5480/50000 progress=11.0% loss=0.113415 lr=9.903199e-05 lr_backbone=9.903199e-05 elapsed=05:28:05 avg_step=3.59s recent_step=4.25s eta=44:25:27 +2026-09-17 22:57:14 | INFO | train rank=0 | step=5490/50000 progress=11.0% loss=0.108418 lr=9.902551e-05 lr_backbone=9.902551e-05 elapsed=05:28:37 avg_step=3.59s recent_step=3.21s eta=44:24:20 +2026-09-17 22:57:45 | INFO | train rank=0 | step=5500/50000 progress=11.0% loss=0.117708 lr=9.901900e-05 lr_backbone=9.901900e-05 elapsed=05:29:08 avg_step=3.59s recent_step=3.10s eta=44:23:04 +2026-09-17 22:58:15 | INFO | train rank=0 | step=5510/50000 progress=11.0% loss=0.112523 lr=9.901247e-05 lr_backbone=9.901247e-05 elapsed=05:29:38 avg_step=3.59s recent_step=3.00s eta=44:21:41 +2026-09-17 22:59:01 | INFO | train rank=0 | step=5520/50000 progress=11.0% loss=0.103473 lr=9.900592e-05 lr_backbone=9.900592e-05 elapsed=05:30:24 avg_step=3.59s recent_step=4.60s eta=44:22:26 +2026-09-17 22:59:25 | INFO | train rank=0 | step=5530/50000 progress=11.1% loss=0.114624 lr=9.899935e-05 lr_backbone=9.899935e-05 elapsed=05:30:48 avg_step=3.59s recent_step=2.39s eta=44:20:13 +2026-09-17 22:59:57 | INFO | train rank=0 | step=5540/50000 progress=11.1% loss=0.109799 lr=9.899275e-05 lr_backbone=9.899275e-05 elapsed=05:31:20 avg_step=3.59s recent_step=3.18s eta=44:19:05 +2026-09-17 23:00:24 | INFO | train rank=0 | step=5550/50000 progress=11.1% loss=0.107809 lr=9.898614e-05 lr_backbone=9.898614e-05 elapsed=05:31:47 avg_step=3.59s recent_step=2.77s eta=44:17:23 +2026-09-17 23:01:15 | INFO | train rank=0 | step=5560/50000 progress=11.1% loss=0.106958 lr=9.897950e-05 lr_backbone=9.897950e-05 elapsed=05:32:38 avg_step=3.59s recent_step=5.10s eta=44:18:48 +2026-09-17 23:01:47 | INFO | train rank=0 | step=5570/50000 progress=11.1% loss=0.101494 lr=9.897284e-05 lr_backbone=9.897284e-05 elapsed=05:33:11 avg_step=3.59s recent_step=3.22s eta=44:17:42 +2026-09-17 23:02:16 | INFO | train rank=0 | step=5580/50000 progress=11.2% loss=0.118388 lr=9.896616e-05 lr_backbone=9.896616e-05 elapsed=05:33:40 avg_step=3.59s recent_step=2.91s eta=44:16:12 +2026-09-17 23:02:49 | INFO | train rank=0 | step=5590/50000 progress=11.2% loss=0.111391 lr=9.895946e-05 lr_backbone=9.895946e-05 elapsed=05:34:13 avg_step=3.59s recent_step=3.30s eta=44:15:13 +2026-09-17 23:03:35 | INFO | train rank=0 | step=5600/50000 progress=11.2% loss=0.130378 lr=9.895274e-05 lr_backbone=9.895274e-05 elapsed=05:34:58 avg_step=3.59s recent_step=4.55s eta=44:15:53 +2026-09-17 23:04:00 | INFO | train rank=0 | step=5610/50000 progress=11.2% loss=0.141707 lr=9.894600e-05 lr_backbone=9.894600e-05 elapsed=05:35:23 avg_step=3.59s recent_step=2.46s eta=44:13:48 +2026-09-17 23:04:27 | INFO | train rank=0 | step=5620/50000 progress=11.2% loss=0.114161 lr=9.893923e-05 lr_backbone=9.893923e-05 elapsed=05:35:50 avg_step=3.59s recent_step=2.70s eta=44:12:03 +2026-09-17 23:04:55 | INFO | train rank=0 | step=5630/50000 progress=11.3% loss=0.119241 lr=9.893245e-05 lr_backbone=9.893245e-05 elapsed=05:36:18 avg_step=3.58s recent_step=2.86s eta=44:10:30 +2026-09-17 23:05:40 | INFO | train rank=0 | step=5640/50000 progress=11.3% loss=0.104208 lr=9.892564e-05 lr_backbone=9.892564e-05 elapsed=05:37:03 avg_step=3.59s recent_step=4.48s eta=44:11:04 +2026-09-17 23:06:12 | INFO | train rank=0 | step=5650/50000 progress=11.3% loss=0.091055 lr=9.891881e-05 lr_backbone=9.891881e-05 elapsed=05:37:36 avg_step=3.59s recent_step=3.25s eta=44:10:02 +2026-09-17 23:06:43 | INFO | train rank=0 | step=5660/50000 progress=11.3% loss=0.126116 lr=9.891196e-05 lr_backbone=9.891196e-05 elapsed=05:38:06 avg_step=3.58s recent_step=3.07s eta=44:08:45 +2026-09-17 23:07:22 | INFO | train rank=0 | step=5670/50000 progress=11.3% loss=0.156385 lr=9.890509e-05 lr_backbone=9.890509e-05 elapsed=05:38:45 avg_step=3.58s recent_step=3.90s eta=44:08:34 +2026-09-17 23:08:09 | INFO | train rank=0 | step=5680/50000 progress=11.4% loss=0.111553 lr=9.889819e-05 lr_backbone=9.889819e-05 elapsed=05:39:33 avg_step=3.59s recent_step=4.71s eta=44:09:27 +2026-09-17 23:08:42 | INFO | train rank=0 | step=5690/50000 progress=11.4% loss=0.115135 lr=9.889128e-05 lr_backbone=9.889128e-05 elapsed=05:40:06 avg_step=3.59s recent_step=3.30s eta=44:08:29 +2026-09-17 23:09:12 | INFO | train rank=0 | step=5700/50000 progress=11.4% loss=0.142340 lr=9.888434e-05 lr_backbone=9.888434e-05 elapsed=05:40:36 avg_step=3.59s recent_step=3.02s eta=44:07:09 +2026-09-17 23:09:45 | INFO | train rank=0 | step=5710/50000 progress=11.4% loss=0.114704 lr=9.887739e-05 lr_backbone=9.887739e-05 elapsed=05:41:08 avg_step=3.58s recent_step=3.24s eta=44:06:06 +2026-09-17 23:10:31 | INFO | train rank=0 | step=5720/50000 progress=11.4% loss=0.121266 lr=9.887041e-05 lr_backbone=9.887041e-05 elapsed=05:41:54 avg_step=3.59s recent_step=4.61s eta=44:06:50 +2026-09-17 23:11:06 | INFO | train rank=0 | step=5730/50000 progress=11.5% loss=0.114748 lr=9.886341e-05 lr_backbone=9.886341e-05 elapsed=05:42:29 avg_step=3.59s recent_step=3.50s eta=44:06:07 +2026-09-17 23:11:39 | INFO | train rank=0 | step=5740/50000 progress=11.5% loss=0.109064 lr=9.885638e-05 lr_backbone=9.885638e-05 elapsed=05:43:03 avg_step=3.59s recent_step=3.32s eta=44:05:11 +2026-09-17 23:12:10 | INFO | train rank=0 | step=5750/50000 progress=11.5% loss=0.114064 lr=9.884934e-05 lr_backbone=9.884934e-05 elapsed=05:43:33 avg_step=3.58s recent_step=3.06s eta=44:03:55 +2026-09-17 23:12:56 | INFO | train rank=0 | step=5760/50000 progress=11.5% loss=0.102430 lr=9.884228e-05 lr_backbone=9.884228e-05 elapsed=05:44:19 avg_step=3.59s recent_step=4.59s eta=44:04:36 +2026-09-17 23:13:25 | INFO | train rank=0 | step=5770/50000 progress=11.5% loss=0.101155 lr=9.883519e-05 lr_backbone=9.883519e-05 elapsed=05:44:48 avg_step=3.59s recent_step=2.91s eta=44:03:08 +2026-09-17 23:13:52 | INFO | train rank=0 | step=5780/50000 progress=11.6% loss=0.124805 lr=9.882808e-05 lr_backbone=9.882808e-05 elapsed=05:45:15 avg_step=3.58s recent_step=2.70s eta=44:01:25 +2026-09-17 23:14:28 | INFO | train rank=0 | step=5790/50000 progress=11.6% loss=0.132462 lr=9.882096e-05 lr_backbone=9.882096e-05 elapsed=05:45:52 avg_step=3.58s recent_step=3.65s eta=44:00:54 +2026-09-17 23:15:10 | INFO | train rank=0 | step=5800/50000 progress=11.6% loss=0.129163 lr=9.881381e-05 lr_backbone=9.881381e-05 elapsed=05:46:33 avg_step=3.59s recent_step=4.15s eta=44:01:01 +2026-09-17 23:15:47 | INFO | train rank=0 | step=5810/50000 progress=11.6% loss=0.121920 lr=9.880663e-05 lr_backbone=9.880663e-05 elapsed=05:47:11 avg_step=3.59s recent_step=3.76s eta=44:00:39 +2026-09-17 23:16:23 | INFO | train rank=0 | step=5820/50000 progress=11.6% loss=0.145472 lr=9.879944e-05 lr_backbone=9.879944e-05 elapsed=05:47:46 avg_step=3.59s recent_step=3.51s eta=43:59:57 +2026-09-17 23:16:50 | INFO | train rank=0 | step=5830/50000 progress=11.7% loss=0.089266 lr=9.879223e-05 lr_backbone=9.879223e-05 elapsed=05:48:13 avg_step=3.58s recent_step=2.75s eta=43:58:18 +2026-09-17 23:17:31 | INFO | train rank=0 | step=5840/50000 progress=11.7% loss=0.125978 lr=9.878499e-05 lr_backbone=9.878499e-05 elapsed=05:48:54 avg_step=3.58s recent_step=4.05s eta=43:58:18 +2026-09-17 23:18:12 | INFO | train rank=0 | step=5850/50000 progress=11.7% loss=0.141485 lr=9.877774e-05 lr_backbone=9.877774e-05 elapsed=05:49:35 avg_step=3.59s recent_step=4.12s eta=43:58:23 +2026-09-17 23:18:45 | INFO | train rank=0 | step=5860/50000 progress=11.7% loss=0.095041 lr=9.877046e-05 lr_backbone=9.877046e-05 elapsed=05:50:08 avg_step=3.59s recent_step=3.33s eta=43:57:27 +2026-09-17 23:19:20 | INFO | train rank=0 | step=5870/50000 progress=11.7% loss=0.129225 lr=9.876316e-05 lr_backbone=9.876316e-05 elapsed=05:50:43 avg_step=3.58s recent_step=3.47s eta=43:56:43 +2026-09-17 23:20:06 | INFO | train rank=0 | step=5880/50000 progress=11.8% loss=0.101282 lr=9.875584e-05 lr_backbone=9.875584e-05 elapsed=05:51:29 avg_step=3.59s recent_step=4.61s eta=43:57:24 +2026-09-17 23:20:42 | INFO | train rank=0 | step=5890/50000 progress=11.8% loss=0.140425 lr=9.874850e-05 lr_backbone=9.874850e-05 elapsed=05:52:05 avg_step=3.59s recent_step=3.58s eta=43:56:47 +2026-09-17 23:21:10 | INFO | train rank=0 | step=5900/50000 progress=11.8% loss=0.100471 lr=9.874113e-05 lr_backbone=9.874113e-05 elapsed=05:52:34 avg_step=3.59s recent_step=2.86s eta=43:55:17 +2026-09-17 23:21:44 | INFO | train rank=0 | step=5910/50000 progress=11.8% loss=0.129607 lr=9.873375e-05 lr_backbone=9.873375e-05 elapsed=05:53:07 avg_step=3.59s recent_step=3.38s eta=43:54:26 +2026-09-17 23:22:38 | INFO | train rank=0 | step=5920/50000 progress=11.8% loss=0.101009 lr=9.872634e-05 lr_backbone=9.872634e-05 elapsed=05:54:01 avg_step=3.59s recent_step=5.35s eta=43:56:01 +2026-09-17 23:23:09 | INFO | train rank=0 | step=5930/50000 progress=11.9% loss=0.144952 lr=9.871892e-05 lr_backbone=9.871892e-05 elapsed=05:54:32 avg_step=3.59s recent_step=3.12s eta=43:54:51 +2026-09-17 23:23:40 | INFO | train rank=0 | step=5940/50000 progress=11.9% loss=0.116564 lr=9.871147e-05 lr_backbone=9.871147e-05 elapsed=05:55:03 avg_step=3.59s recent_step=3.13s eta=43:53:41 +2026-09-17 23:24:11 | INFO | train rank=0 | step=5950/50000 progress=11.9% loss=0.118599 lr=9.870400e-05 lr_backbone=9.870400e-05 elapsed=05:55:34 avg_step=3.59s recent_step=3.11s eta=43:52:30 +2026-09-17 23:24:59 | INFO | train rank=0 | step=5960/50000 progress=11.9% loss=0.136191 lr=9.869651e-05 lr_backbone=9.869651e-05 elapsed=05:56:22 avg_step=3.59s recent_step=4.74s eta=43:53:19 +2026-09-17 23:25:29 | INFO | train rank=0 | step=5970/50000 progress=11.9% loss=0.116475 lr=9.868899e-05 lr_backbone=9.868899e-05 elapsed=05:56:52 avg_step=3.59s recent_step=3.05s eta=43:52:04 +2026-09-17 23:26:07 | INFO | train rank=0 | step=5980/50000 progress=12.0% loss=0.114523 lr=9.868146e-05 lr_backbone=9.868146e-05 elapsed=05:57:31 avg_step=3.59s recent_step=3.82s eta=43:51:45 +2026-09-17 23:26:36 | INFO | train rank=0 | step=5990/50000 progress=12.0% loss=0.093845 lr=9.867391e-05 lr_backbone=9.867391e-05 elapsed=05:57:59 avg_step=3.59s recent_step=2.83s eta=43:50:14 +2026-09-17 23:27:33 | INFO | train rank=0 | step=6000/50000 progress=12.0% loss=0.119558 lr=9.866633e-05 lr_backbone=9.866633e-05 elapsed=05:58:56 avg_step=3.59s recent_step=5.70s eta=43:52:13 +2026-09-17 23:28:03 | INFO | train rank=0 | step=6010/50000 progress=12.0% loss=0.136279 lr=9.865873e-05 lr_backbone=9.865873e-05 elapsed=05:59:27 avg_step=3.59s recent_step=3.07s eta=43:50:59 +2026-09-17 23:28:35 | INFO | train rank=0 | step=6020/50000 progress=12.0% loss=0.086084 lr=9.865111e-05 lr_backbone=9.865111e-05 elapsed=05:59:59 avg_step=3.59s recent_step=3.20s eta=43:49:55 +2026-09-17 23:29:05 | INFO | train rank=0 | step=6030/50000 progress=12.1% loss=0.133279 lr=9.864347e-05 lr_backbone=9.864347e-05 elapsed=06:00:28 avg_step=3.59s recent_step=2.95s eta=43:48:33 +2026-09-17 23:29:53 | INFO | train rank=0 | step=6040/50000 progress=12.1% loss=0.103542 lr=9.863581e-05 lr_backbone=9.863581e-05 elapsed=06:01:16 avg_step=3.59s recent_step=4.81s eta=43:49:26 +2026-09-17 23:30:28 | INFO | train rank=0 | step=6050/50000 progress=12.1% loss=0.138594 lr=9.862813e-05 lr_backbone=9.862813e-05 elapsed=06:01:51 avg_step=3.59s recent_step=3.48s eta=43:48:42 +2026-09-17 23:31:04 | INFO | train rank=0 | step=6060/50000 progress=12.1% loss=0.129061 lr=9.862043e-05 lr_backbone=9.862043e-05 elapsed=06:02:28 avg_step=3.59s recent_step=3.67s eta=43:48:13 +2026-09-17 23:31:38 | INFO | train rank=0 | step=6070/50000 progress=12.1% loss=0.087717 lr=9.861270e-05 lr_backbone=9.861270e-05 elapsed=06:03:01 avg_step=3.59s recent_step=3.34s eta=43:47:19 +2026-09-17 23:32:27 | INFO | train rank=0 | step=6080/50000 progress=12.2% loss=0.096429 lr=9.860495e-05 lr_backbone=9.860495e-05 elapsed=06:03:50 avg_step=3.59s recent_step=4.89s eta=43:48:17 +2026-09-17 23:32:54 | INFO | train rank=0 | step=6090/50000 progress=12.2% loss=0.146655 lr=9.859719e-05 lr_backbone=9.859719e-05 elapsed=06:04:17 avg_step=3.59s recent_step=2.68s eta=43:46:36 +2026-09-17 23:33:29 | INFO | train rank=0 | step=6100/50000 progress=12.2% loss=0.130036 lr=9.858940e-05 lr_backbone=9.858940e-05 elapsed=06:04:52 avg_step=3.59s recent_step=3.49s eta=43:45:53 +2026-09-17 23:33:56 | INFO | train rank=0 | step=6110/50000 progress=12.2% loss=0.086403 lr=9.858159e-05 lr_backbone=9.858159e-05 elapsed=06:05:20 avg_step=3.59s recent_step=2.78s eta=43:44:19 +2026-09-17 23:34:46 | INFO | train rank=0 | step=6120/50000 progress=12.2% loss=0.115229 lr=9.857376e-05 lr_backbone=9.857376e-05 elapsed=06:06:09 avg_step=3.59s recent_step=4.93s eta=43:45:20 +2026-09-17 23:35:17 | INFO | train rank=0 | step=6130/50000 progress=12.3% loss=0.128432 lr=9.856590e-05 lr_backbone=9.856590e-05 elapsed=06:06:41 avg_step=3.59s recent_step=3.15s eta=43:44:13 +2026-09-17 23:35:51 | INFO | train rank=0 | step=6140/50000 progress=12.3% loss=0.096680 lr=9.855803e-05 lr_backbone=9.855803e-05 elapsed=06:07:14 avg_step=3.59s recent_step=3.38s eta=43:43:22 +2026-09-17 23:36:30 | INFO | train rank=0 | step=6150/50000 progress=12.3% loss=0.143788 lr=9.855013e-05 lr_backbone=9.855013e-05 elapsed=06:07:53 avg_step=3.59s recent_step=3.88s eta=43:43:07 +2026-09-17 23:37:21 | INFO | train rank=0 | step=6160/50000 progress=12.3% loss=0.112348 lr=9.854222e-05 lr_backbone=9.854222e-05 elapsed=06:08:44 avg_step=3.59s recent_step=5.10s eta=43:44:18 +2026-09-17 23:37:51 | INFO | train rank=0 | step=6170/50000 progress=12.3% loss=0.135284 lr=9.853428e-05 lr_backbone=9.853428e-05 elapsed=06:09:14 avg_step=3.59s recent_step=2.97s eta=43:42:58 +2026-09-17 23:38:16 | INFO | train rank=0 | step=6180/50000 progress=12.4% loss=0.087087 lr=9.852632e-05 lr_backbone=9.852632e-05 elapsed=06:09:40 avg_step=3.59s recent_step=2.57s eta=43:41:10 +2026-09-17 23:38:49 | INFO | train rank=0 | step=6190/50000 progress=12.4% loss=0.147967 lr=9.851834e-05 lr_backbone=9.851834e-05 elapsed=06:10:12 avg_step=3.59s recent_step=3.24s eta=43:40:09 +2026-09-17 23:39:34 | INFO | train rank=0 | step=6200/50000 progress=12.4% loss=0.137014 lr=9.851034e-05 lr_backbone=9.851034e-05 elapsed=06:10:57 avg_step=3.59s recent_step=4.52s eta=43:40:39 +2026-09-17 23:40:06 | INFO | train rank=0 | step=6210/50000 progress=12.4% loss=0.113027 lr=9.850232e-05 lr_backbone=9.850232e-05 elapsed=06:11:29 avg_step=3.59s recent_step=3.17s eta=43:39:34 +2026-09-17 23:40:36 | INFO | train rank=0 | step=6220/50000 progress=12.4% loss=0.100236 lr=9.849427e-05 lr_backbone=9.849427e-05 elapsed=06:11:59 avg_step=3.59s recent_step=3.05s eta=43:38:20 +2026-09-17 23:41:08 | INFO | train rank=0 | step=6230/50000 progress=12.5% loss=0.157015 lr=9.848621e-05 lr_backbone=9.848621e-05 elapsed=06:12:31 avg_step=3.59s recent_step=3.18s eta=43:37:16 +2026-09-17 23:41:58 | INFO | train rank=0 | step=6240/50000 progress=12.5% loss=0.107826 lr=9.847812e-05 lr_backbone=9.847812e-05 elapsed=06:13:21 avg_step=3.59s recent_step=5.02s eta=43:38:20 +2026-09-17 23:42:32 | INFO | train rank=0 | step=6250/50000 progress=12.5% loss=0.129829 lr=9.847001e-05 lr_backbone=9.847001e-05 elapsed=06:13:55 avg_step=3.59s recent_step=3.35s eta=43:37:27 +2026-09-17 23:43:03 | INFO | train rank=0 | step=6260/50000 progress=12.5% loss=0.119432 lr=9.846188e-05 lr_backbone=9.846188e-05 elapsed=06:14:26 avg_step=3.59s recent_step=3.09s eta=43:36:17 +2026-09-17 23:43:37 | INFO | train rank=0 | step=6270/50000 progress=12.5% loss=0.123054 lr=9.845373e-05 lr_backbone=9.845373e-05 elapsed=06:15:00 avg_step=3.59s recent_step=3.40s eta=43:35:28 +2026-09-17 23:44:28 | INFO | train rank=0 | step=6280/50000 progress=12.6% loss=0.098767 lr=9.844556e-05 lr_backbone=9.844556e-05 elapsed=06:15:51 avg_step=3.59s recent_step=5.09s eta=43:36:37 +2026-09-17 23:44:59 | INFO | train rank=0 | step=6290/50000 progress=12.6% loss=0.116778 lr=9.843737e-05 lr_backbone=9.843737e-05 elapsed=06:16:22 avg_step=3.59s recent_step=3.11s eta=43:35:27 +2026-09-17 23:45:29 | INFO | train rank=0 | step=6300/50000 progress=12.6% loss=0.096801 lr=9.842916e-05 lr_backbone=9.842916e-05 elapsed=06:16:53 avg_step=3.59s recent_step=3.07s eta=43:34:15 +2026-09-17 23:45:59 | INFO | train rank=0 | step=6310/50000 progress=12.6% loss=0.109546 lr=9.842092e-05 lr_backbone=9.842092e-05 elapsed=06:17:23 avg_step=3.59s recent_step=3.01s eta=43:32:59 +2026-09-17 23:46:48 | INFO | train rank=0 | step=6320/50000 progress=12.6% loss=0.153697 lr=9.841267e-05 lr_backbone=9.841267e-05 elapsed=06:18:12 avg_step=3.59s recent_step=4.89s eta=43:33:53 +2026-09-17 23:47:18 | INFO | train rank=0 | step=6330/50000 progress=12.7% loss=0.128918 lr=9.840439e-05 lr_backbone=9.840439e-05 elapsed=06:18:41 avg_step=3.59s recent_step=2.93s eta=43:32:32 +2026-09-17 23:47:48 | INFO | train rank=0 | step=6340/50000 progress=12.7% loss=0.143251 lr=9.839609e-05 lr_backbone=9.839609e-05 elapsed=06:19:11 avg_step=3.59s recent_step=3.04s eta=43:31:18 +2026-09-17 23:48:28 | INFO | train rank=0 | step=6350/50000 progress=12.7% loss=0.158625 lr=9.838777e-05 lr_backbone=9.838777e-05 elapsed=06:19:51 avg_step=3.59s recent_step=3.99s eta=43:31:09 +2026-09-17 23:49:11 | INFO | train rank=0 | step=6360/50000 progress=12.7% loss=0.116534 lr=9.837943e-05 lr_backbone=9.837943e-05 elapsed=06:20:34 avg_step=3.59s recent_step=4.31s eta=43:31:23 +2026-09-17 23:49:45 | INFO | train rank=0 | step=6370/50000 progress=12.7% loss=0.124786 lr=9.837107e-05 lr_backbone=9.837107e-05 elapsed=06:21:08 avg_step=3.59s recent_step=3.41s eta=43:30:35 +2026-09-17 23:50:13 | INFO | train rank=0 | step=6380/50000 progress=12.8% loss=0.123584 lr=9.836269e-05 lr_backbone=9.836269e-05 elapsed=06:21:37 avg_step=3.59s recent_step=2.84s eta=43:29:07 +2026-09-17 23:50:44 | INFO | train rank=0 | step=6390/50000 progress=12.8% loss=0.116697 lr=9.835428e-05 lr_backbone=9.835428e-05 elapsed=06:22:07 avg_step=3.59s recent_step=3.02s eta=43:27:53 +2026-09-17 23:51:33 | INFO | train rank=0 | step=6400/50000 progress=12.8% loss=0.100129 lr=9.834586e-05 lr_backbone=9.834586e-05 elapsed=06:22:56 avg_step=3.59s recent_step=4.92s eta=43:28:47 +2026-09-17 23:51:59 | INFO | train rank=0 | step=6410/50000 progress=12.8% loss=0.111828 lr=9.833741e-05 lr_backbone=9.833741e-05 elapsed=06:23:22 avg_step=3.59s recent_step=2.62s eta=43:27:05 +2026-09-17 23:52:32 | INFO | train rank=0 | step=6420/50000 progress=12.8% loss=0.147943 lr=9.832895e-05 lr_backbone=9.832895e-05 elapsed=06:23:55 avg_step=3.59s recent_step=3.26s eta=43:26:07 +2026-09-17 23:53:08 | INFO | train rank=0 | step=6430/50000 progress=12.9% loss=0.113304 lr=9.832046e-05 lr_backbone=9.832046e-05 elapsed=06:24:31 avg_step=3.59s recent_step=3.63s eta=43:25:34 +2026-09-17 23:54:06 | INFO | train rank=0 | step=6440/50000 progress=12.9% loss=0.118351 lr=9.831195e-05 lr_backbone=9.831195e-05 elapsed=06:25:30 avg_step=3.59s recent_step=5.84s eta=43:27:31 +2026-09-17 23:54:42 | INFO | train rank=0 | step=6450/50000 progress=12.9% loss=0.141189 lr=9.830342e-05 lr_backbone=9.830342e-05 elapsed=06:26:05 avg_step=3.59s recent_step=3.55s eta=43:26:52 +2026-09-17 23:55:10 | INFO | train rank=0 | step=6460/50000 progress=12.9% loss=0.105129 lr=9.829486e-05 lr_backbone=9.829486e-05 elapsed=06:26:33 avg_step=3.59s recent_step=2.78s eta=43:25:21 +2026-09-17 23:55:37 | INFO | train rank=0 | step=6470/50000 progress=12.9% loss=0.125326 lr=9.828629e-05 lr_backbone=9.828629e-05 elapsed=06:27:00 avg_step=3.59s recent_step=2.72s eta=43:23:47 +2026-09-17 23:56:28 | INFO | train rank=0 | step=6480/50000 progress=13.0% loss=0.103907 lr=9.827770e-05 lr_backbone=9.827770e-05 elapsed=06:27:51 avg_step=3.59s recent_step=5.08s eta=43:24:51 +2026-09-17 23:56:59 | INFO | train rank=0 | step=6490/50000 progress=13.0% loss=0.100566 lr=9.826908e-05 lr_backbone=9.826908e-05 elapsed=06:28:22 avg_step=3.59s recent_step=3.11s eta=43:23:43 +2026-09-17 23:57:30 | INFO | train rank=0 | step=6500/50000 progress=13.0% loss=0.144382 lr=9.826045e-05 lr_backbone=9.826045e-05 elapsed=06:28:53 avg_step=3.59s recent_step=3.13s eta=43:22:36 +2026-09-17 23:57:59 | INFO | train rank=0 | step=6510/50000 progress=13.0% loss=0.100852 lr=9.825179e-05 lr_backbone=9.825179e-05 elapsed=06:29:22 avg_step=3.59s recent_step=2.89s eta=43:21:13 +2026-09-17 23:58:45 | INFO | train rank=0 | step=6520/50000 progress=13.0% loss=0.180024 lr=9.824311e-05 lr_backbone=9.824311e-05 elapsed=06:30:08 avg_step=3.59s recent_step=4.57s eta=43:21:43 +2026-09-17 23:59:19 | INFO | train rank=0 | step=6530/50000 progress=13.1% loss=0.128118 lr=9.823441e-05 lr_backbone=9.823441e-05 elapsed=06:30:43 avg_step=3.59s recent_step=3.48s eta=43:20:59 +2026-09-17 23:59:46 | INFO | train rank=0 | step=6540/50000 progress=13.1% loss=0.105183 lr=9.822569e-05 lr_backbone=9.822569e-05 elapsed=06:31:10 avg_step=3.59s recent_step=2.69s eta=43:19:24 +2026-09-18 00:00:24 | INFO | train rank=0 | step=6550/50000 progress=13.1% loss=0.109004 lr=9.821695e-05 lr_backbone=9.821695e-05 elapsed=06:31:47 avg_step=3.59s recent_step=3.78s eta=43:19:01 +2026-09-18 00:01:04 | INFO | train rank=0 | step=6560/50000 progress=13.1% loss=0.120728 lr=9.820818e-05 lr_backbone=9.820818e-05 elapsed=06:32:27 avg_step=3.59s recent_step=3.97s eta=43:18:50 +2026-09-18 00:01:37 | INFO | train rank=0 | step=6570/50000 progress=13.1% loss=0.139885 lr=9.819940e-05 lr_backbone=9.819940e-05 elapsed=06:33:00 avg_step=3.59s recent_step=3.35s eta=43:17:58 +2026-09-18 00:02:08 | INFO | train rank=0 | step=6580/50000 progress=13.2% loss=0.113270 lr=9.819059e-05 lr_backbone=9.819059e-05 elapsed=06:33:31 avg_step=3.59s recent_step=3.06s eta=43:16:47 +2026-09-18 00:02:35 | INFO | train rank=0 | step=6590/50000 progress=13.2% loss=0.135173 lr=9.818177e-05 lr_backbone=9.818177e-05 elapsed=06:33:58 avg_step=3.59s recent_step=2.73s eta=43:15:14 +2026-09-18 00:03:24 | INFO | train rank=0 | step=6600/50000 progress=13.2% loss=0.171424 lr=9.817292e-05 lr_backbone=9.817292e-05 elapsed=06:34:48 avg_step=3.59s recent_step=4.93s eta=43:16:07 +2026-09-18 00:04:02 | INFO | train rank=0 | step=6610/50000 progress=13.2% loss=0.117159 lr=9.816405e-05 lr_backbone=9.816405e-05 elapsed=06:35:25 avg_step=3.59s recent_step=3.79s eta=43:15:44 +2026-09-18 00:04:25 | INFO | train rank=0 | step=6620/50000 progress=13.2% loss=0.107637 lr=9.815516e-05 lr_backbone=9.815516e-05 elapsed=06:35:48 avg_step=3.59s recent_step=2.26s eta=43:13:41 +2026-09-18 00:04:53 | INFO | train rank=0 | step=6630/50000 progress=13.3% loss=0.112594 lr=9.814625e-05 lr_backbone=9.814625e-05 elapsed=06:36:16 avg_step=3.59s recent_step=2.82s eta=43:12:15 +2026-09-18 00:05:37 | INFO | train rank=0 | step=6640/50000 progress=13.3% loss=0.165453 lr=9.813732e-05 lr_backbone=9.813732e-05 elapsed=06:37:00 avg_step=3.59s recent_step=4.39s eta=43:12:31 +2026-09-18 00:06:09 | INFO | train rank=0 | step=6650/50000 progress=13.3% loss=0.100233 lr=9.812837e-05 lr_backbone=9.812837e-05 elapsed=06:37:33 avg_step=3.59s recent_step=3.25s eta=43:11:34 +2026-09-18 00:06:44 | INFO | train rank=0 | step=6660/50000 progress=13.3% loss=0.105233 lr=9.811939e-05 lr_backbone=9.811939e-05 elapsed=06:38:07 avg_step=3.59s recent_step=3.47s eta=43:10:50 +2026-09-18 00:07:11 | INFO | train rank=0 | step=6670/50000 progress=13.3% loss=0.119395 lr=9.811040e-05 lr_backbone=9.811040e-05 elapsed=06:38:34 avg_step=3.59s recent_step=2.64s eta=43:09:13 +2026-09-18 00:07:51 | INFO | train rank=0 | step=6680/50000 progress=13.4% loss=0.133705 lr=9.810138e-05 lr_backbone=9.810138e-05 elapsed=06:39:14 avg_step=3.59s recent_step=4.02s eta=43:09:05 +2026-09-18 00:08:32 | INFO | train rank=0 | step=6690/50000 progress=13.4% loss=0.102754 lr=9.809235e-05 lr_backbone=9.809235e-05 elapsed=06:39:55 avg_step=3.59s recent_step=4.13s eta=43:09:04 +2026-09-18 00:09:02 | INFO | train rank=0 | step=6700/50000 progress=13.4% loss=0.102781 lr=9.808329e-05 lr_backbone=9.808329e-05 elapsed=06:40:26 avg_step=3.59s recent_step=3.05s eta=43:07:54 +2026-09-18 00:09:53 | INFO | train rank=0 | step=6710/50000 progress=13.4% loss=0.097495 lr=9.807421e-05 lr_backbone=9.807421e-05 elapsed=06:41:16 avg_step=3.59s recent_step=5.05s eta=43:08:52 +2026-09-18 00:10:24 | INFO | train rank=0 | step=6720/50000 progress=13.4% loss=0.148489 lr=9.806511e-05 lr_backbone=9.806511e-05 elapsed=06:41:47 avg_step=3.59s recent_step=3.12s eta=43:07:46 +2026-09-18 00:11:03 | INFO | train rank=0 | step=6730/50000 progress=13.5% loss=0.102403 lr=9.805599e-05 lr_backbone=9.805599e-05 elapsed=06:42:27 avg_step=3.59s recent_step=3.93s eta=43:07:32 +2026-09-18 00:11:37 | INFO | train rank=0 | step=6740/50000 progress=13.5% loss=0.128666 lr=9.804685e-05 lr_backbone=9.804685e-05 elapsed=06:43:01 avg_step=3.59s recent_step=3.40s eta=43:06:44 +2026-09-18 00:12:21 | INFO | train rank=0 | step=6750/50000 progress=13.5% loss=0.094909 lr=9.803768e-05 lr_backbone=9.803768e-05 elapsed=06:43:44 avg_step=3.59s recent_step=4.35s eta=43:06:57 +2026-09-18 00:12:55 | INFO | train rank=0 | step=6760/50000 progress=13.5% loss=0.111649 lr=9.802850e-05 lr_backbone=9.802850e-05 elapsed=06:44:18 avg_step=3.59s recent_step=3.38s eta=43:06:08 +2026-09-18 00:13:24 | INFO | train rank=0 | step=6770/50000 progress=13.5% loss=0.107734 lr=9.801929e-05 lr_backbone=9.801929e-05 elapsed=06:44:47 avg_step=3.59s recent_step=2.92s eta=43:04:49 +2026-09-18 00:13:52 | INFO | train rank=0 | step=6780/50000 progress=13.6% loss=0.110421 lr=9.801007e-05 lr_backbone=9.801007e-05 elapsed=06:45:15 avg_step=3.59s recent_step=2.77s eta=43:03:21 +2026-09-18 00:14:42 | INFO | train rank=0 | step=6790/50000 progress=13.6% loss=0.093409 lr=9.800082e-05 lr_backbone=9.800082e-05 elapsed=06:46:05 avg_step=3.59s recent_step=5.05s eta=43:04:18 +2026-09-18 00:15:09 | INFO | train rank=0 | step=6800/50000 progress=13.6% loss=0.100085 lr=9.799155e-05 lr_backbone=9.799155e-05 elapsed=06:46:32 avg_step=3.59s recent_step=2.67s eta=43:02:44 +2026-09-18 00:15:49 | INFO | train rank=0 | step=6810/50000 progress=13.6% loss=0.101527 lr=9.798226e-05 lr_backbone=9.798226e-05 elapsed=06:47:12 avg_step=3.59s recent_step=3.99s eta=43:02:34 +2026-09-18 00:16:20 | INFO | train rank=0 | step=6820/50000 progress=13.6% loss=0.089837 lr=9.797295e-05 lr_backbone=9.797295e-05 elapsed=06:47:44 avg_step=3.59s recent_step=3.16s eta=43:01:31 +2026-09-18 00:16:59 | INFO | train rank=0 | step=6830/50000 progress=13.7% loss=0.110239 lr=9.796362e-05 lr_backbone=9.796362e-05 elapsed=06:48:22 avg_step=3.59s recent_step=3.85s eta=43:01:12 +2026-09-18 00:17:32 | INFO | train rank=0 | step=6840/50000 progress=13.7% loss=0.110152 lr=9.795427e-05 lr_backbone=9.795427e-05 elapsed=06:48:56 avg_step=3.59s recent_step=3.34s eta=43:00:20 +2026-09-18 00:18:12 | INFO | train rank=0 | step=6850/50000 progress=13.7% loss=0.127571 lr=9.794490e-05 lr_backbone=9.794490e-05 elapsed=06:49:35 avg_step=3.59s recent_step=3.98s eta=43:00:09 +2026-09-18 00:18:47 | INFO | train rank=0 | step=6860/50000 progress=13.7% loss=0.108524 lr=9.793550e-05 lr_backbone=9.793550e-05 elapsed=06:50:10 avg_step=3.59s recent_step=3.47s eta=42:59:26 +2026-09-18 00:19:42 | INFO | train rank=0 | step=6870/50000 progress=13.7% loss=0.105668 lr=9.792609e-05 lr_backbone=9.792609e-05 elapsed=06:51:05 avg_step=3.59s recent_step=5.54s eta=43:00:53 +2026-09-18 00:20:12 | INFO | train rank=0 | step=6880/50000 progress=13.8% loss=0.115188 lr=9.791665e-05 lr_backbone=9.791665e-05 elapsed=06:51:35 avg_step=3.59s recent_step=2.94s eta=42:59:37 +2026-09-18 00:20:45 | INFO | train rank=0 | step=6890/50000 progress=13.8% loss=0.113273 lr=9.790720e-05 lr_backbone=9.790720e-05 elapsed=06:52:08 avg_step=3.59s recent_step=3.34s eta=42:58:45 +2026-09-18 00:21:17 | INFO | train rank=0 | step=6900/50000 progress=13.8% loss=0.111200 lr=9.789772e-05 lr_backbone=9.789772e-05 elapsed=06:52:41 avg_step=3.59s recent_step=3.22s eta=42:57:46 +2026-09-18 00:22:00 | INFO | train rank=0 | step=6910/50000 progress=13.8% loss=0.110395 lr=9.788822e-05 lr_backbone=9.788822e-05 elapsed=06:53:23 avg_step=3.59s recent_step=4.29s eta=42:57:54 +2026-09-18 00:22:37 | INFO | train rank=0 | step=6920/50000 progress=13.8% loss=0.141041 lr=9.787870e-05 lr_backbone=9.787870e-05 elapsed=06:54:01 avg_step=3.59s recent_step=3.72s eta=42:57:26 +2026-09-18 00:23:06 | INFO | train rank=0 | step=6930/50000 progress=13.9% loss=0.156806 lr=9.786916e-05 lr_backbone=9.786916e-05 elapsed=06:54:29 avg_step=3.59s recent_step=2.87s eta=42:56:05 +2026-09-18 00:23:48 | INFO | train rank=0 | step=6940/50000 progress=13.9% loss=0.102588 lr=9.785960e-05 lr_backbone=9.785960e-05 elapsed=06:55:11 avg_step=3.59s recent_step=4.22s eta=42:56:08 +2026-09-18 00:24:27 | INFO | train rank=0 | step=6950/50000 progress=13.9% loss=0.120904 lr=9.785002e-05 lr_backbone=9.785002e-05 elapsed=06:55:50 avg_step=3.59s recent_step=3.85s eta=42:55:48 +2026-09-18 00:24:59 | INFO | train rank=0 | step=6960/50000 progress=13.9% loss=0.111093 lr=9.784041e-05 lr_backbone=9.784041e-05 elapsed=06:56:23 avg_step=3.59s recent_step=3.28s eta=42:54:53 +2026-09-18 00:25:34 | INFO | train rank=0 | step=6970/50000 progress=13.9% loss=0.119262 lr=9.783079e-05 lr_backbone=9.783079e-05 elapsed=06:56:57 avg_step=3.59s recent_step=3.42s eta=42:54:07 +2026-09-18 00:26:27 | INFO | train rank=0 | step=6980/50000 progress=14.0% loss=0.109061 lr=9.782114e-05 lr_backbone=9.782114e-05 elapsed=06:57:50 avg_step=3.59s recent_step=5.36s eta=42:55:20 +2026-09-18 00:27:01 | INFO | train rank=0 | step=6990/50000 progress=14.0% loss=0.114754 lr=9.781148e-05 lr_backbone=9.781148e-05 elapsed=06:58:24 avg_step=3.59s recent_step=3.38s eta=42:54:31 +2026-09-18 00:27:36 | INFO | train rank=0 | step=7000/50000 progress=14.0% loss=0.105846 lr=9.780179e-05 lr_backbone=9.780179e-05 elapsed=06:58:59 avg_step=3.59s recent_step=3.47s eta=42:53:48 +2026-09-18 00:28:13 | INFO | train rank=0 | step=7010/50000 progress=14.0% loss=0.108959 lr=9.779208e-05 lr_backbone=9.779208e-05 elapsed=06:59:36 avg_step=3.59s recent_step=3.72s eta=42:53:19 +2026-09-18 00:28:47 | INFO | train rank=0 | step=7020/50000 progress=14.0% loss=0.116405 lr=9.778235e-05 lr_backbone=9.778235e-05 elapsed=07:00:10 avg_step=3.59s recent_step=3.37s eta=42:52:30 +2026-09-18 00:29:31 | INFO | train rank=0 | step=7030/50000 progress=14.1% loss=0.098235 lr=9.777260e-05 lr_backbone=9.777260e-05 elapsed=07:00:54 avg_step=3.59s recent_step=4.44s eta=42:52:46 +2026-09-18 00:30:07 | INFO | train rank=0 | step=7040/50000 progress=14.1% loss=0.157705 lr=9.776283e-05 lr_backbone=9.776283e-05 elapsed=07:01:31 avg_step=3.59s recent_step=3.62s eta=42:52:12 +2026-09-18 00:30:41 | INFO | train rank=0 | step=7050/50000 progress=14.1% loss=0.128654 lr=9.775304e-05 lr_backbone=9.775304e-05 elapsed=07:02:04 avg_step=3.59s recent_step=3.37s eta=42:51:22 +2026-09-18 00:31:06 | INFO | train rank=0 | step=7060/50000 progress=14.1% loss=0.137664 lr=9.774323e-05 lr_backbone=9.774323e-05 elapsed=07:02:29 avg_step=3.59s recent_step=2.51s eta=42:49:40 +2026-09-18 00:32:03 | INFO | train rank=0 | step=7070/50000 progress=14.1% loss=0.105724 lr=9.773339e-05 lr_backbone=9.773339e-05 elapsed=07:03:26 avg_step=3.59s recent_step=5.70s eta=42:51:13 +2026-09-18 00:32:36 | INFO | train rank=0 | step=7080/50000 progress=14.2% loss=0.127991 lr=9.772354e-05 lr_backbone=9.772354e-05 elapsed=07:03:59 avg_step=3.59s recent_step=3.28s eta=42:50:18 +2026-09-18 00:33:11 | INFO | train rank=0 | step=7090/50000 progress=14.2% loss=0.124682 lr=9.771366e-05 lr_backbone=9.771366e-05 elapsed=07:04:35 avg_step=3.59s recent_step=3.56s eta=42:49:40 +2026-09-18 00:33:44 | INFO | train rank=0 | step=7100/50000 progress=14.2% loss=0.155286 lr=9.770377e-05 lr_backbone=9.770377e-05 elapsed=07:05:08 avg_step=3.59s recent_step=3.30s eta=42:48:46 +2026-09-18 00:34:34 | INFO | train rank=0 | step=7110/50000 progress=14.2% loss=0.101259 lr=9.769385e-05 lr_backbone=9.769385e-05 elapsed=07:05:57 avg_step=3.59s recent_step=4.92s eta=42:49:30 +2026-09-18 00:34:59 | INFO | train rank=0 | step=7120/50000 progress=14.2% loss=0.107432 lr=9.768391e-05 lr_backbone=9.768391e-05 elapsed=07:06:22 avg_step=3.59s recent_step=2.51s eta=42:47:49 +2026-09-18 00:35:30 | INFO | train rank=0 | step=7130/50000 progress=14.3% loss=0.105032 lr=9.767395e-05 lr_backbone=9.767395e-05 elapsed=07:06:53 avg_step=3.59s recent_step=3.15s eta=42:46:47 +2026-09-18 00:36:08 | INFO | train rank=0 | step=7140/50000 progress=14.3% loss=0.111989 lr=9.766398e-05 lr_backbone=9.766398e-05 elapsed=07:07:31 avg_step=3.59s recent_step=3.77s eta=42:46:21 +2026-09-18 00:37:00 | INFO | train rank=0 | step=7150/50000 progress=14.3% loss=0.108914 lr=9.765398e-05 lr_backbone=9.765398e-05 elapsed=07:08:23 avg_step=3.59s recent_step=5.18s eta=42:47:21 +2026-09-18 00:37:31 | INFO | train rank=0 | step=7160/50000 progress=14.3% loss=0.129644 lr=9.764395e-05 lr_backbone=9.764395e-05 elapsed=07:08:55 avg_step=3.59s recent_step=3.17s eta=42:46:20 +2026-09-18 00:38:02 | INFO | train rank=0 | step=7170/50000 progress=14.3% loss=0.095357 lr=9.763391e-05 lr_backbone=9.763391e-05 elapsed=07:09:26 avg_step=3.59s recent_step=3.08s eta=42:45:13 +2026-09-18 00:38:30 | INFO | train rank=0 | step=7180/50000 progress=14.4% loss=0.132727 lr=9.762385e-05 lr_backbone=9.762385e-05 elapsed=07:09:53 avg_step=3.59s recent_step=2.78s eta=42:43:48 +2026-09-18 00:39:23 | INFO | train rank=0 | step=7190/50000 progress=14.4% loss=0.135837 lr=9.761377e-05 lr_backbone=9.761377e-05 elapsed=07:10:47 avg_step=3.59s recent_step=5.33s eta=42:44:56 +2026-09-18 00:39:50 | INFO | train rank=0 | step=7200/50000 progress=14.4% loss=0.121706 lr=9.760366e-05 lr_backbone=9.760366e-05 elapsed=07:11:14 avg_step=3.59s recent_step=2.69s eta=42:43:26 +2026-09-18 00:40:24 | INFO | train rank=0 | step=7210/50000 progress=14.4% loss=0.111810 lr=9.759354e-05 lr_backbone=9.759354e-05 elapsed=07:11:48 avg_step=3.59s recent_step=3.40s eta=42:42:39 +2026-09-18 00:40:54 | INFO | train rank=0 | step=7220/50000 progress=14.4% loss=0.108854 lr=9.758339e-05 lr_backbone=9.758339e-05 elapsed=07:12:17 avg_step=3.59s recent_step=2.94s eta=42:41:24 +2026-09-18 00:41:40 | INFO | train rank=0 | step=7230/50000 progress=14.5% loss=0.142335 lr=9.757322e-05 lr_backbone=9.757322e-05 elapsed=07:13:04 avg_step=3.59s recent_step=4.67s eta=42:41:52 +2026-09-18 00:42:15 | INFO | train rank=0 | step=7240/50000 progress=14.5% loss=0.144843 lr=9.756303e-05 lr_backbone=9.756303e-05 elapsed=07:13:39 avg_step=3.59s recent_step=3.51s eta=42:41:11 +2026-09-18 00:42:45 | INFO | train rank=0 | step=7250/50000 progress=14.5% loss=0.095802 lr=9.755283e-05 lr_backbone=9.755283e-05 elapsed=07:14:08 avg_step=3.59s recent_step=2.96s eta=42:39:58 +2026-09-18 00:43:30 | INFO | train rank=0 | step=7260/50000 progress=14.5% loss=0.097485 lr=9.754260e-05 lr_backbone=9.754260e-05 elapsed=07:14:53 avg_step=3.59s recent_step=4.48s eta=42:40:14 +2026-09-18 00:44:02 | INFO | train rank=0 | step=7270/50000 progress=14.5% loss=0.095687 lr=9.753235e-05 lr_backbone=9.753235e-05 elapsed=07:15:26 avg_step=3.59s recent_step=3.24s eta=42:39:17 +2026-09-18 00:44:31 | INFO | train rank=0 | step=7280/50000 progress=14.6% loss=0.162822 lr=9.752208e-05 lr_backbone=9.752208e-05 elapsed=07:15:54 avg_step=3.59s recent_step=2.90s eta=42:38:00 +2026-09-18 00:45:03 | INFO | train rank=0 | step=7290/50000 progress=14.6% loss=0.148268 lr=9.751178e-05 lr_backbone=9.751178e-05 elapsed=07:16:26 avg_step=3.59s recent_step=3.14s eta=42:36:58 +2026-09-18 00:45:47 | INFO | train rank=0 | step=7300/50000 progress=14.6% loss=0.088535 lr=9.750147e-05 lr_backbone=9.750147e-05 elapsed=07:17:11 avg_step=3.59s recent_step=4.48s eta=42:37:14 +2026-09-18 00:46:21 | INFO | train rank=0 | step=7310/50000 progress=14.6% loss=0.130640 lr=9.749114e-05 lr_backbone=9.749114e-05 elapsed=07:17:44 avg_step=3.59s recent_step=3.32s eta=42:36:22 +2026-09-18 00:46:55 | INFO | train rank=0 | step=7320/50000 progress=14.6% loss=0.106374 lr=9.748078e-05 lr_backbone=9.748078e-05 elapsed=07:18:18 avg_step=3.59s recent_step=3.46s eta=42:35:38 +2026-09-18 00:47:34 | INFO | train rank=0 | step=7330/50000 progress=14.7% loss=0.117807 lr=9.747041e-05 lr_backbone=9.747041e-05 elapsed=07:18:57 avg_step=3.59s recent_step=3.88s eta=42:35:19 +2026-09-18 00:48:15 | INFO | train rank=0 | step=7340/50000 progress=14.7% loss=0.118311 lr=9.746001e-05 lr_backbone=9.746001e-05 elapsed=07:19:38 avg_step=3.59s recent_step=4.11s eta=42:35:13 +2026-09-18 00:48:59 | INFO | train rank=0 | step=7350/50000 progress=14.7% loss=0.113459 lr=9.744960e-05 lr_backbone=9.744960e-05 elapsed=07:20:22 avg_step=3.59s recent_step=4.37s eta=42:35:22 +2026-09-18 00:49:30 | INFO | train rank=0 | step=7360/50000 progress=14.7% loss=0.109749 lr=9.743916e-05 lr_backbone=9.743916e-05 elapsed=07:20:53 avg_step=3.59s recent_step=3.11s eta=42:34:18 +2026-09-18 00:50:05 | INFO | train rank=0 | step=7370/50000 progress=14.7% loss=0.165776 lr=9.742870e-05 lr_backbone=9.742870e-05 elapsed=07:21:28 avg_step=3.59s recent_step=3.50s eta=42:33:37 +2026-09-18 00:50:42 | INFO | train rank=0 | step=7380/50000 progress=14.8% loss=0.109968 lr=9.741822e-05 lr_backbone=9.741822e-05 elapsed=07:22:06 avg_step=3.59s recent_step=3.74s eta=42:33:09 +2026-09-18 00:51:25 | INFO | train rank=0 | step=7390/50000 progress=14.8% loss=0.107726 lr=9.740772e-05 lr_backbone=9.740772e-05 elapsed=07:22:48 avg_step=3.60s recent_step=4.26s eta=42:33:11 +2026-09-18 00:51:59 | INFO | train rank=0 | step=7400/50000 progress=14.8% loss=0.145749 lr=9.739720e-05 lr_backbone=9.739720e-05 elapsed=07:23:23 avg_step=3.60s recent_step=3.46s eta=42:32:28 +2026-09-18 00:52:28 | INFO | train rank=0 | step=7410/50000 progress=14.8% loss=0.153385 lr=9.738666e-05 lr_backbone=9.738666e-05 elapsed=07:23:52 avg_step=3.59s recent_step=2.90s eta=42:31:12 +2026-09-18 00:53:05 | INFO | train rank=0 | step=7420/50000 progress=14.8% loss=0.124898 lr=9.737610e-05 lr_backbone=9.737610e-05 elapsed=07:24:28 avg_step=3.59s recent_step=3.66s eta=42:30:40 +2026-09-18 00:53:46 | INFO | train rank=0 | step=7430/50000 progress=14.9% loss=0.121986 lr=9.736552e-05 lr_backbone=9.736552e-05 elapsed=07:25:09 avg_step=3.59s recent_step=4.12s eta=42:30:34 +2026-09-18 00:54:14 | INFO | train rank=0 | step=7440/50000 progress=14.9% loss=0.101493 lr=9.735492e-05 lr_backbone=9.735492e-05 elapsed=07:25:37 avg_step=3.59s recent_step=2.76s eta=42:29:10 +2026-09-18 00:54:58 | INFO | train rank=0 | step=7450/50000 progress=14.9% loss=0.101907 lr=9.734429e-05 lr_backbone=9.734429e-05 elapsed=07:26:21 avg_step=3.59s recent_step=4.43s eta=42:29:22 +2026-09-18 00:55:26 | INFO | train rank=0 | step=7460/50000 progress=14.9% loss=0.162176 lr=9.733365e-05 lr_backbone=9.733365e-05 elapsed=07:26:49 avg_step=3.59s recent_step=2.76s eta=42:27:58 +2026-09-18 00:56:18 | INFO | train rank=0 | step=7470/50000 progress=14.9% loss=0.111915 lr=9.732298e-05 lr_backbone=9.732298e-05 elapsed=07:27:42 avg_step=3.60s recent_step=5.26s eta=42:28:57 +2026-09-18 00:56:50 | INFO | train rank=0 | step=7480/50000 progress=15.0% loss=0.100919 lr=9.731230e-05 lr_backbone=9.731230e-05 elapsed=07:28:13 avg_step=3.60s recent_step=3.17s eta=42:27:57 +2026-09-18 00:57:21 | INFO | train rank=0 | step=7490/50000 progress=15.0% loss=0.123072 lr=9.730159e-05 lr_backbone=9.730159e-05 elapsed=07:28:45 avg_step=3.59s recent_step=3.15s eta=42:26:56 +2026-09-18 00:58:01 | INFO | train rank=0 | step=7500/50000 progress=15.0% loss=0.104327 lr=9.729086e-05 lr_backbone=9.729086e-05 elapsed=07:29:24 avg_step=3.60s recent_step=3.92s eta=42:26:38 +2026-09-18 00:58:53 | INFO | train rank=0 | step=7510/50000 progress=15.0% loss=0.130269 lr=9.728011e-05 lr_backbone=9.728011e-05 elapsed=07:30:17 avg_step=3.60s recent_step=5.26s eta=42:27:36 +2026-09-18 00:59:30 | INFO | train rank=0 | step=7520/50000 progress=15.0% loss=0.103085 lr=9.726935e-05 lr_backbone=9.726935e-05 elapsed=07:30:53 avg_step=3.60s recent_step=3.67s eta=42:27:04 +2026-09-18 00:59:57 | INFO | train rank=0 | step=7530/50000 progress=15.1% loss=0.138228 lr=9.725856e-05 lr_backbone=9.725856e-05 elapsed=07:31:20 avg_step=3.60s recent_step=2.70s eta=42:25:38 +2026-09-18 01:00:29 | INFO | train rank=0 | step=7540/50000 progress=15.1% loss=0.122661 lr=9.724775e-05 lr_backbone=9.724775e-05 elapsed=07:31:53 avg_step=3.60s recent_step=3.24s eta=42:24:41 +2026-09-18 01:01:25 | INFO | train rank=0 | step=7550/50000 progress=15.1% loss=0.139940 lr=9.723692e-05 lr_backbone=9.723692e-05 elapsed=07:32:48 avg_step=3.60s recent_step=5.56s eta=42:25:56 +2026-09-18 01:01:50 | INFO | train rank=0 | step=7560/50000 progress=15.1% loss=0.095187 lr=9.722606e-05 lr_backbone=9.722606e-05 elapsed=07:33:13 avg_step=3.60s recent_step=2.52s eta=42:24:20 +2026-09-18 01:02:24 | INFO | train rank=0 | step=7570/50000 progress=15.1% loss=0.114336 lr=9.721519e-05 lr_backbone=9.721519e-05 elapsed=07:33:47 avg_step=3.60s recent_step=3.34s eta=42:23:29 +2026-09-18 01:02:57 | INFO | train rank=0 | step=7580/50000 progress=15.2% loss=0.083883 lr=9.720430e-05 lr_backbone=9.720430e-05 elapsed=07:34:21 avg_step=3.60s recent_step=3.39s eta=42:22:42 +2026-09-18 01:03:45 | INFO | train rank=0 | step=7590/50000 progress=15.2% loss=0.084408 lr=9.719339e-05 lr_backbone=9.719339e-05 elapsed=07:35:08 avg_step=3.60s recent_step=4.75s eta=42:23:11 +2026-09-18 01:04:14 | INFO | train rank=0 | step=7600/50000 progress=15.2% loss=0.092184 lr=9.718245e-05 lr_backbone=9.718245e-05 elapsed=07:35:38 avg_step=3.60s recent_step=2.93s eta=42:21:58 +2026-09-18 01:04:53 | INFO | train rank=0 | step=7610/50000 progress=15.2% loss=0.128762 lr=9.717150e-05 lr_backbone=9.717150e-05 elapsed=07:36:16 avg_step=3.60s recent_step=3.87s eta=42:21:37 +2026-09-18 01:05:22 | INFO | train rank=0 | step=7620/50000 progress=15.2% loss=0.129441 lr=9.716052e-05 lr_backbone=9.716052e-05 elapsed=07:36:46 avg_step=3.60s recent_step=2.93s eta=42:20:24 +2026-09-18 01:06:17 | INFO | train rank=0 | step=7630/50000 progress=15.3% loss=0.155414 lr=9.714953e-05 lr_backbone=9.714953e-05 elapsed=07:37:41 avg_step=3.60s recent_step=5.49s eta=42:21:33 +2026-09-18 01:06:51 | INFO | train rank=0 | step=7640/50000 progress=15.3% loss=0.142868 lr=9.713851e-05 lr_backbone=9.713851e-05 elapsed=07:38:14 avg_step=3.60s recent_step=3.37s eta=42:20:44 +2026-09-18 01:07:18 | INFO | train rank=0 | step=7650/50000 progress=15.3% loss=0.110758 lr=9.712747e-05 lr_backbone=9.712747e-05 elapsed=07:38:41 avg_step=3.60s recent_step=2.67s eta=42:19:16 +2026-09-18 01:07:52 | INFO | train rank=0 | step=7660/50000 progress=15.3% loss=0.137975 lr=9.711642e-05 lr_backbone=9.711642e-05 elapsed=07:39:15 avg_step=3.60s recent_step=3.43s eta=42:18:31 +2026-09-18 01:08:40 | INFO | train rank=0 | step=7670/50000 progress=15.3% loss=0.105793 lr=9.710534e-05 lr_backbone=9.710534e-05 elapsed=07:40:04 avg_step=3.60s recent_step=4.85s eta=42:19:04 +2026-09-18 01:09:04 | INFO | train rank=0 | step=7680/50000 progress=15.4% loss=0.095562 lr=9.709424e-05 lr_backbone=9.709424e-05 elapsed=07:40:27 avg_step=3.60s recent_step=2.35s eta=42:17:19 +2026-09-18 01:09:34 | INFO | train rank=0 | step=7690/50000 progress=15.4% loss=0.092108 lr=9.708312e-05 lr_backbone=9.708312e-05 elapsed=07:40:57 avg_step=3.60s recent_step=3.00s eta=42:16:10 +2026-09-18 01:10:09 | INFO | train rank=0 | step=7700/50000 progress=15.4% loss=0.109576 lr=9.707198e-05 lr_backbone=9.707198e-05 elapsed=07:41:32 avg_step=3.60s recent_step=3.52s eta=42:15:30 +2026-09-18 01:11:03 | INFO | train rank=0 | step=7710/50000 progress=15.4% loss=0.123360 lr=9.706082e-05 lr_backbone=9.706082e-05 elapsed=07:42:26 avg_step=3.60s recent_step=5.37s eta=42:16:31 +2026-09-18 01:11:30 | INFO | train rank=0 | step=7720/50000 progress=15.4% loss=0.096452 lr=9.704964e-05 lr_backbone=9.704964e-05 elapsed=07:42:54 avg_step=3.60s recent_step=2.75s eta=42:15:09 +2026-09-18 01:12:02 | INFO | train rank=0 | step=7730/50000 progress=15.5% loss=0.130985 lr=9.703843e-05 lr_backbone=9.703843e-05 elapsed=07:43:25 avg_step=3.60s recent_step=3.13s eta=42:14:08 +2026-09-18 01:12:39 | INFO | train rank=0 | step=7740/50000 progress=15.5% loss=0.108650 lr=9.702721e-05 lr_backbone=9.702721e-05 elapsed=07:44:02 avg_step=3.60s recent_step=3.73s eta=42:13:39 +2026-09-18 01:13:28 | INFO | train rank=0 | step=7750/50000 progress=15.5% loss=0.128664 lr=9.701597e-05 lr_backbone=9.701597e-05 elapsed=07:44:51 avg_step=3.60s recent_step=4.90s eta=42:14:14 +2026-09-18 01:13:59 | INFO | train rank=0 | step=7760/50000 progress=15.5% loss=0.128560 lr=9.700471e-05 lr_backbone=9.700471e-05 elapsed=07:45:22 avg_step=3.60s recent_step=3.11s eta=42:13:11 +2026-09-18 01:14:28 | INFO | train rank=0 | step=7770/50000 progress=15.5% loss=0.110393 lr=9.699342e-05 lr_backbone=9.699342e-05 elapsed=07:45:51 avg_step=3.60s recent_step=2.87s eta=42:11:56 +2026-09-18 01:14:56 | INFO | train rank=0 | step=7780/50000 progress=15.6% loss=0.128991 lr=9.698212e-05 lr_backbone=9.698212e-05 elapsed=07:46:20 avg_step=3.60s recent_step=2.87s eta=42:10:40 +2026-09-18 01:15:54 | INFO | train rank=0 | step=7790/50000 progress=15.6% loss=0.112726 lr=9.697079e-05 lr_backbone=9.697079e-05 elapsed=07:47:18 avg_step=3.60s recent_step=5.79s eta=42:12:03 +2026-09-18 01:16:27 | INFO | train rank=0 | step=7800/50000 progress=15.6% loss=0.102709 lr=9.695945e-05 lr_backbone=9.695945e-05 elapsed=07:47:50 avg_step=3.60s recent_step=3.24s eta=42:11:08 +2026-09-18 01:16:57 | INFO | train rank=0 | step=7810/50000 progress=15.6% loss=0.115219 lr=9.694808e-05 lr_backbone=9.694808e-05 elapsed=07:48:20 avg_step=3.60s recent_step=3.01s eta=42:10:00 +2026-09-18 01:17:29 | INFO | train rank=0 | step=7820/50000 progress=15.6% loss=0.138402 lr=9.693669e-05 lr_backbone=9.693669e-05 elapsed=07:48:53 avg_step=3.60s recent_step=3.24s eta=42:09:05 +2026-09-18 01:18:14 | INFO | train rank=0 | step=7830/50000 progress=15.7% loss=0.110476 lr=9.692529e-05 lr_backbone=9.692529e-05 elapsed=07:49:37 avg_step=3.60s recent_step=4.47s eta=42:09:16 +2026-09-18 01:18:42 | INFO | train rank=0 | step=7840/50000 progress=15.7% loss=0.090606 lr=9.691386e-05 lr_backbone=9.691386e-05 elapsed=07:50:05 avg_step=3.60s recent_step=2.77s eta=42:07:56 +2026-09-18 01:19:07 | INFO | train rank=0 | step=7850/50000 progress=15.7% loss=0.116107 lr=9.690241e-05 lr_backbone=9.690241e-05 elapsed=07:50:30 avg_step=3.60s recent_step=2.49s eta=42:06:20 +2026-09-18 01:19:36 | INFO | train rank=0 | step=7860/50000 progress=15.7% loss=0.112315 lr=9.689094e-05 lr_backbone=9.689094e-05 elapsed=07:50:59 avg_step=3.60s recent_step=2.93s eta=42:05:09 +2026-09-18 01:20:27 | INFO | train rank=0 | step=7870/50000 progress=15.7% loss=0.111309 lr=9.687945e-05 lr_backbone=9.687945e-05 elapsed=07:51:50 avg_step=3.60s recent_step=5.09s eta=42:05:53 +2026-09-18 01:21:00 | INFO | train rank=0 | step=7880/50000 progress=15.8% loss=0.152901 lr=9.686794e-05 lr_backbone=9.686794e-05 elapsed=07:52:23 avg_step=3.60s recent_step=3.28s eta=42:05:00 +2026-09-18 01:21:29 | INFO | train rank=0 | step=7890/50000 progress=15.8% loss=0.132476 lr=9.685641e-05 lr_backbone=9.685641e-05 elapsed=07:52:52 avg_step=3.60s recent_step=2.91s eta=42:03:47 +2026-09-18 01:22:07 | INFO | train rank=0 | step=7900/50000 progress=15.8% loss=0.118848 lr=9.684486e-05 lr_backbone=9.684486e-05 elapsed=07:53:30 avg_step=3.60s recent_step=3.79s eta=42:03:21 +2026-09-18 01:22:54 | INFO | train rank=0 | step=7910/50000 progress=15.8% loss=0.173466 lr=9.683329e-05 lr_backbone=9.683329e-05 elapsed=07:54:17 avg_step=3.60s recent_step=4.72s eta=42:03:45 +2026-09-18 01:23:15 | INFO | train rank=0 | step=7920/50000 progress=15.8% loss=0.104179 lr=9.682170e-05 lr_backbone=9.682170e-05 elapsed=07:54:38 avg_step=3.60s recent_step=2.14s eta=42:01:52 +2026-09-18 01:23:42 | INFO | train rank=0 | step=7930/50000 progress=15.9% loss=0.112133 lr=9.681008e-05 lr_backbone=9.681008e-05 elapsed=07:55:05 avg_step=3.59s recent_step=2.69s eta=42:00:28 +2026-09-18 01:24:18 | INFO | train rank=0 | step=7940/50000 progress=15.9% loss=0.104645 lr=9.679845e-05 lr_backbone=9.679845e-05 elapsed=07:55:41 avg_step=3.59s recent_step=3.56s eta=41:59:50 +2026-09-18 01:24:53 | INFO | train rank=0 | step=7950/50000 progress=15.9% loss=0.107814 lr=9.678680e-05 lr_backbone=9.678680e-05 elapsed=07:56:17 avg_step=3.59s recent_step=3.59s eta=41:59:14 +2026-09-18 01:25:30 | INFO | train rank=0 | step=7960/50000 progress=15.9% loss=0.120787 lr=9.677512e-05 lr_backbone=9.677512e-05 elapsed=07:56:54 avg_step=3.59s recent_step=3.70s eta=41:58:43 +2026-09-18 01:26:08 | INFO | train rank=0 | step=7970/50000 progress=15.9% loss=0.092643 lr=9.676343e-05 lr_backbone=9.676343e-05 elapsed=07:57:31 avg_step=3.59s recent_step=3.71s eta=41:58:13 +2026-09-18 01:26:37 | INFO | train rank=0 | step=7980/50000 progress=16.0% loss=0.140057 lr=9.675171e-05 lr_backbone=9.675171e-05 elapsed=07:58:00 avg_step=3.59s recent_step=2.93s eta=41:57:02 +2026-09-18 01:27:36 | INFO | train rank=0 | step=7990/50000 progress=16.0% loss=0.111507 lr=9.673998e-05 lr_backbone=9.673998e-05 elapsed=07:58:59 avg_step=3.60s recent_step=5.88s eta=41:58:26 +2026-09-18 01:28:09 | INFO | train rank=0 | step=8000/50000 progress=16.0% loss=0.127538 lr=9.672822e-05 lr_backbone=9.672822e-05 elapsed=07:59:32 avg_step=3.60s recent_step=3.33s eta=41:57:36 +2026-09-18 01:28:41 | INFO | train rank=0 | step=8010/50000 progress=16.0% loss=0.110656 lr=9.671645e-05 lr_backbone=9.671645e-05 elapsed=08:00:04 avg_step=3.60s recent_step=3.21s eta=41:56:40 +2026-09-18 01:29:12 | INFO | train rank=0 | step=8020/50000 progress=16.0% loss=0.116406 lr=9.670465e-05 lr_backbone=9.670465e-05 elapsed=08:00:36 avg_step=3.60s recent_step=3.13s eta=41:55:40 +2026-09-18 01:29:57 | INFO | train rank=0 | step=8030/50000 progress=16.1% loss=0.098154 lr=9.669283e-05 lr_backbone=9.669283e-05 elapsed=08:01:21 avg_step=3.60s recent_step=4.49s eta=41:55:51 +2026-09-18 01:30:28 | INFO | train rank=0 | step=8040/50000 progress=16.1% loss=0.105829 lr=9.668100e-05 lr_backbone=9.668100e-05 elapsed=08:01:51 avg_step=3.60s recent_step=3.05s eta=41:54:46 +2026-09-18 01:30:57 | INFO | train rank=0 | step=8050/50000 progress=16.1% loss=0.102020 lr=9.666914e-05 lr_backbone=9.666914e-05 elapsed=08:02:20 avg_step=3.60s recent_step=2.89s eta=41:53:34 +2026-09-18 01:31:31 | INFO | train rank=0 | step=8060/50000 progress=16.1% loss=0.125980 lr=9.665726e-05 lr_backbone=9.665726e-05 elapsed=08:02:54 avg_step=3.59s recent_step=3.41s eta=41:52:48 +2026-09-18 01:32:13 | INFO | train rank=0 | step=8070/50000 progress=16.1% loss=0.100246 lr=9.664536e-05 lr_backbone=9.664536e-05 elapsed=08:03:37 avg_step=3.60s recent_step=4.26s eta=41:52:47 +2026-09-18 01:32:48 | INFO | train rank=0 | step=8080/50000 progress=16.2% loss=0.158206 lr=9.663344e-05 lr_backbone=9.663344e-05 elapsed=08:04:11 avg_step=3.60s recent_step=3.42s eta=41:52:01 +2026-09-18 01:33:19 | INFO | train rank=0 | step=8090/50000 progress=16.2% loss=0.092277 lr=9.662150e-05 lr_backbone=9.662150e-05 elapsed=08:04:43 avg_step=3.59s recent_step=3.17s eta=41:51:04 +2026-09-18 01:34:00 | INFO | train rank=0 | step=8100/50000 progress=16.2% loss=0.115906 lr=9.660954e-05 lr_backbone=9.660954e-05 elapsed=08:05:23 avg_step=3.60s recent_step=4.05s eta=41:50:51 +2026-09-18 01:34:46 | INFO | train rank=0 | step=8110/50000 progress=16.2% loss=0.117147 lr=9.659756e-05 lr_backbone=9.659756e-05 elapsed=08:06:10 avg_step=3.60s recent_step=4.64s eta=41:51:09 +2026-09-18 01:35:21 | INFO | train rank=0 | step=8120/50000 progress=16.2% loss=0.108303 lr=9.658556e-05 lr_backbone=9.658556e-05 elapsed=08:06:44 avg_step=3.60s recent_step=3.47s eta=41:50:27 +2026-09-18 01:35:55 | INFO | train rank=0 | step=8130/50000 progress=16.3% loss=0.131456 lr=9.657354e-05 lr_backbone=9.657354e-05 elapsed=08:07:18 avg_step=3.60s recent_step=3.36s eta=41:49:39 +2026-09-18 01:36:30 | INFO | train rank=0 | step=8140/50000 progress=16.3% loss=0.111713 lr=9.656150e-05 lr_backbone=9.656150e-05 elapsed=08:07:53 avg_step=3.60s recent_step=3.51s eta=41:48:59 +2026-09-18 01:37:21 | INFO | train rank=0 | step=8150/50000 progress=16.3% loss=0.106181 lr=9.654944e-05 lr_backbone=9.654944e-05 elapsed=08:08:44 avg_step=3.60s recent_step=5.11s eta=41:49:40 +2026-09-18 01:37:54 | INFO | train rank=0 | step=8160/50000 progress=16.3% loss=0.104122 lr=9.653736e-05 lr_backbone=9.653736e-05 elapsed=08:09:18 avg_step=3.60s recent_step=3.36s eta=41:48:52 +2026-09-18 01:38:28 | INFO | train rank=0 | step=8170/50000 progress=16.3% loss=0.104600 lr=9.652525e-05 lr_backbone=9.652525e-05 elapsed=08:09:51 avg_step=3.60s recent_step=3.36s eta=41:48:04 +2026-09-18 01:39:05 | INFO | train rank=0 | step=8180/50000 progress=16.4% loss=0.096848 lr=9.651313e-05 lr_backbone=9.651313e-05 elapsed=08:10:28 avg_step=3.60s recent_step=3.68s eta=41:47:32 +2026-09-18 01:39:50 | INFO | train rank=0 | step=8190/50000 progress=16.4% loss=0.120739 lr=9.650099e-05 lr_backbone=9.650099e-05 elapsed=08:11:14 avg_step=3.60s recent_step=4.57s eta=41:47:46 +2026-09-18 01:40:18 | INFO | train rank=0 | step=8200/50000 progress=16.4% loss=0.106429 lr=9.648882e-05 lr_backbone=9.648882e-05 elapsed=08:11:41 avg_step=3.60s recent_step=2.72s eta=41:46:25 +2026-09-18 01:40:55 | INFO | train rank=0 | step=8210/50000 progress=16.4% loss=0.100618 lr=9.647664e-05 lr_backbone=9.647664e-05 elapsed=08:12:18 avg_step=3.60s recent_step=3.74s eta=41:45:56 +2026-09-18 01:41:37 | INFO | train rank=0 | step=8220/50000 progress=16.4% loss=0.112045 lr=9.646444e-05 lr_backbone=9.646444e-05 elapsed=08:13:00 avg_step=3.60s recent_step=4.15s eta=41:45:48 +2026-09-18 01:42:21 | INFO | train rank=0 | step=8230/50000 progress=16.5% loss=0.110952 lr=9.645221e-05 lr_backbone=9.645221e-05 elapsed=08:13:44 avg_step=3.60s recent_step=4.45s eta=41:45:55 +2026-09-18 01:42:53 | INFO | train rank=0 | step=8240/50000 progress=16.5% loss=0.130596 lr=9.643997e-05 lr_backbone=9.643997e-05 elapsed=08:14:16 avg_step=3.60s recent_step=3.18s eta=41:44:58 +2026-09-18 01:43:27 | INFO | train rank=0 | step=8250/50000 progress=16.5% loss=0.120952 lr=9.642770e-05 lr_backbone=9.642770e-05 elapsed=08:14:50 avg_step=3.60s recent_step=3.41s eta=41:44:12 +2026-09-18 01:43:58 | INFO | train rank=0 | step=8260/50000 progress=16.5% loss=0.169687 lr=9.641542e-05 lr_backbone=9.641542e-05 elapsed=08:15:22 avg_step=3.60s recent_step=3.14s eta=41:43:13 +2026-09-18 01:44:47 | INFO | train rank=0 | step=8270/50000 progress=16.5% loss=0.127454 lr=9.640311e-05 lr_backbone=9.640311e-05 elapsed=08:16:11 avg_step=3.60s recent_step=4.89s eta=41:43:42 +2026-09-18 01:45:16 | INFO | train rank=0 | step=8280/50000 progress=16.6% loss=0.138152 lr=9.639078e-05 lr_backbone=9.639078e-05 elapsed=08:16:39 avg_step=3.60s recent_step=2.87s eta=41:42:30 +2026-09-18 01:45:55 | INFO | train rank=0 | step=8290/50000 progress=16.6% loss=0.112942 lr=9.637844e-05 lr_backbone=9.637844e-05 elapsed=08:17:18 avg_step=3.60s recent_step=3.88s eta=41:42:08 +2026-09-18 01:46:29 | INFO | train rank=0 | step=8300/50000 progress=16.6% loss=0.133395 lr=9.636607e-05 lr_backbone=9.636607e-05 elapsed=08:17:52 avg_step=3.60s recent_step=3.39s eta=41:41:21 +2026-09-18 01:47:19 | INFO | train rank=0 | step=8310/50000 progress=16.6% loss=0.114222 lr=9.635368e-05 lr_backbone=9.635368e-05 elapsed=08:18:42 avg_step=3.60s recent_step=5.01s eta=41:41:56 +2026-09-18 01:47:48 | INFO | train rank=0 | step=8320/50000 progress=16.6% loss=0.079130 lr=9.634128e-05 lr_backbone=9.634128e-05 elapsed=08:19:11 avg_step=3.60s recent_step=2.90s eta=41:40:45 +2026-09-18 01:48:18 | INFO | train rank=0 | step=8330/50000 progress=16.7% loss=0.135900 lr=9.632885e-05 lr_backbone=9.632885e-05 elapsed=08:19:41 avg_step=3.60s recent_step=3.00s eta=41:39:39 +2026-09-18 01:48:47 | INFO | train rank=0 | step=8340/50000 progress=16.7% loss=0.108941 lr=9.631640e-05 lr_backbone=9.631640e-05 elapsed=08:20:11 avg_step=3.60s recent_step=2.97s eta=41:38:31 +2026-09-18 01:49:39 | INFO | train rank=0 | step=8350/50000 progress=16.7% loss=0.096724 lr=9.630393e-05 lr_backbone=9.630393e-05 elapsed=08:21:02 avg_step=3.60s recent_step=5.17s eta=41:39:14 +2026-09-18 01:50:07 | INFO | train rank=0 | step=8360/50000 progress=16.7% loss=0.094421 lr=9.629145e-05 lr_backbone=9.629145e-05 elapsed=08:21:30 avg_step=3.60s recent_step=2.81s eta=41:37:58 +2026-09-18 01:50:37 | INFO | train rank=0 | step=8370/50000 progress=16.7% loss=0.109921 lr=9.627894e-05 lr_backbone=9.627894e-05 elapsed=08:22:00 avg_step=3.60s recent_step=2.96s eta=41:36:50 +2026-09-18 01:51:07 | INFO | train rank=0 | step=8380/50000 progress=16.8% loss=0.102534 lr=9.626641e-05 lr_backbone=9.626641e-05 elapsed=08:22:30 avg_step=3.60s recent_step=3.03s eta=41:35:46 +2026-09-18 01:52:09 | INFO | train rank=0 | step=8390/50000 progress=16.8% loss=0.119670 lr=9.625386e-05 lr_backbone=9.625386e-05 elapsed=08:23:32 avg_step=3.60s recent_step=6.22s eta=41:37:20 +2026-09-18 01:52:42 | INFO | train rank=0 | step=8400/50000 progress=16.8% loss=0.085325 lr=9.624129e-05 lr_backbone=9.624129e-05 elapsed=08:24:05 avg_step=3.60s recent_step=3.24s eta=41:36:26 +2026-09-18 01:53:15 | INFO | train rank=0 | step=8410/50000 progress=16.8% loss=0.138440 lr=9.622870e-05 lr_backbone=9.622870e-05 elapsed=08:24:38 avg_step=3.60s recent_step=3.34s eta=41:35:37 +2026-09-18 01:53:42 | INFO | train rank=0 | step=8420/50000 progress=16.8% loss=0.115358 lr=9.621609e-05 lr_backbone=9.621609e-05 elapsed=08:25:05 avg_step=3.60s recent_step=2.70s eta=41:34:17 +2026-09-18 01:54:34 | INFO | train rank=0 | step=8430/50000 progress=16.9% loss=0.125714 lr=9.620346e-05 lr_backbone=9.620346e-05 elapsed=08:25:57 avg_step=3.60s recent_step=5.20s eta=41:35:00 +2026-09-18 01:55:12 | INFO | train rank=0 | step=8440/50000 progress=16.9% loss=0.113956 lr=9.619081e-05 lr_backbone=9.619081e-05 elapsed=08:26:35 avg_step=3.60s recent_step=3.76s eta=41:34:31 +2026-09-18 01:55:39 | INFO | train rank=0 | step=8450/50000 progress=16.9% loss=0.096313 lr=9.617814e-05 lr_backbone=9.617814e-05 elapsed=08:27:02 avg_step=3.60s recent_step=2.76s eta=41:33:14 +2026-09-18 01:56:09 | INFO | train rank=0 | step=8460/50000 progress=16.9% loss=0.139532 lr=9.616545e-05 lr_backbone=9.616545e-05 elapsed=08:27:33 avg_step=3.60s recent_step=3.03s eta=41:32:10 +2026-09-18 01:56:57 | INFO | train rank=0 | step=8470/50000 progress=16.9% loss=0.087118 lr=9.615274e-05 lr_backbone=9.615274e-05 elapsed=08:28:21 avg_step=3.60s recent_step=4.81s eta=41:32:33 +2026-09-18 01:57:22 | INFO | train rank=0 | step=8480/50000 progress=17.0% loss=0.103525 lr=9.614001e-05 lr_backbone=9.614001e-05 elapsed=08:28:45 avg_step=3.60s recent_step=2.47s eta=41:31:02 +2026-09-18 01:57:57 | INFO | train rank=0 | step=8490/50000 progress=17.0% loss=0.087531 lr=9.612726e-05 lr_backbone=9.612726e-05 elapsed=08:29:20 avg_step=3.60s recent_step=3.49s eta=41:30:20 +2026-09-18 01:58:29 | INFO | train rank=0 | step=8500/50000 progress=17.0% loss=0.133620 lr=9.611449e-05 lr_backbone=9.611449e-05 elapsed=08:29:53 avg_step=3.60s recent_step=3.23s eta=41:29:26 +2026-09-18 01:59:26 | INFO | train rank=0 | step=8510/50000 progress=17.0% loss=0.118271 lr=9.610170e-05 lr_backbone=9.610170e-05 elapsed=08:30:49 avg_step=3.60s recent_step=5.66s eta=41:30:31 +2026-09-18 02:00:00 | INFO | train rank=0 | step=8520/50000 progress=17.0% loss=0.116745 lr=9.608888e-05 lr_backbone=9.608888e-05 elapsed=08:31:23 avg_step=3.60s recent_step=3.38s eta=41:29:44 +2026-09-18 02:00:24 | INFO | train rank=0 | step=8530/50000 progress=17.1% loss=0.140274 lr=9.607605e-05 lr_backbone=9.607605e-05 elapsed=08:31:48 avg_step=3.60s recent_step=2.46s eta=41:28:12 +2026-09-18 02:00:55 | INFO | train rank=0 | step=8540/50000 progress=17.1% loss=0.118469 lr=9.606320e-05 lr_backbone=9.606320e-05 elapsed=08:32:18 avg_step=3.60s recent_step=3.06s eta=41:27:10 +2026-09-18 02:01:52 | INFO | train rank=0 | step=8550/50000 progress=17.1% loss=0.114671 lr=9.605033e-05 lr_backbone=9.605033e-05 elapsed=08:33:15 avg_step=3.60s recent_step=5.69s eta=41:28:15 +2026-09-18 02:02:28 | INFO | train rank=0 | step=8560/50000 progress=17.1% loss=0.096923 lr=9.603744e-05 lr_backbone=9.603744e-05 elapsed=08:33:51 avg_step=3.60s recent_step=3.62s eta=41:27:40 +2026-09-18 02:02:53 | INFO | train rank=0 | step=8570/50000 progress=17.1% loss=0.105668 lr=9.602452e-05 lr_backbone=9.602452e-05 elapsed=08:34:16 avg_step=3.60s recent_step=2.45s eta=41:26:08 +2026-09-18 02:03:25 | INFO | train rank=0 | step=8580/50000 progress=17.2% loss=0.119607 lr=9.601159e-05 lr_backbone=9.601159e-05 elapsed=08:34:48 avg_step=3.60s recent_step=3.23s eta=41:25:15 +2026-09-18 02:04:15 | INFO | train rank=0 | step=8590/50000 progress=17.2% loss=0.081385 lr=9.599864e-05 lr_backbone=9.599864e-05 elapsed=08:35:38 avg_step=3.60s recent_step=4.97s eta=41:25:45 +2026-09-18 02:04:46 | INFO | train rank=0 | step=8600/50000 progress=17.2% loss=0.127906 lr=9.598567e-05 lr_backbone=9.598567e-05 elapsed=08:36:09 avg_step=3.60s recent_step=3.10s eta=41:24:44 +2026-09-18 02:05:16 | INFO | train rank=0 | step=8610/50000 progress=17.2% loss=0.091050 lr=9.597267e-05 lr_backbone=9.597267e-05 elapsed=08:36:40 avg_step=3.60s recent_step=3.09s eta=41:23:44 +2026-09-18 02:05:52 | INFO | train rank=0 | step=8620/50000 progress=17.2% loss=0.130633 lr=9.595966e-05 lr_backbone=9.595966e-05 elapsed=08:37:16 avg_step=3.60s recent_step=3.59s eta=41:23:07 +2026-09-18 02:06:40 | INFO | train rank=0 | step=8630/50000 progress=17.3% loss=0.115236 lr=9.594663e-05 lr_backbone=9.594663e-05 elapsed=08:38:03 avg_step=3.60s recent_step=4.74s eta=41:23:26 +2026-09-18 02:07:03 | INFO | train rank=0 | step=8640/50000 progress=17.3% loss=0.117707 lr=9.593358e-05 lr_backbone=9.593358e-05 elapsed=08:38:26 avg_step=3.60s recent_step=2.30s eta=41:21:48 +2026-09-18 02:07:39 | INFO | train rank=0 | step=8650/50000 progress=17.3% loss=0.086472 lr=9.592050e-05 lr_backbone=9.592050e-05 elapsed=08:39:02 avg_step=3.60s recent_step=3.60s eta=41:21:11 +2026-09-18 02:08:06 | INFO | train rank=0 | step=8660/50000 progress=17.3% loss=0.079740 lr=9.590741e-05 lr_backbone=9.590741e-05 elapsed=08:39:29 avg_step=3.60s recent_step=2.73s eta=41:19:54 +2026-09-18 02:08:55 | INFO | train rank=0 | step=8670/50000 progress=17.3% loss=0.122634 lr=9.589430e-05 lr_backbone=9.589430e-05 elapsed=08:40:19 avg_step=3.60s recent_step=4.94s eta=41:20:22 +2026-09-18 02:09:28 | INFO | train rank=0 | step=8680/50000 progress=17.4% loss=0.139186 lr=9.588116e-05 lr_backbone=9.588116e-05 elapsed=08:40:52 avg_step=3.60s recent_step=3.31s eta=41:19:32 +2026-09-18 02:09:59 | INFO | train rank=0 | step=8690/50000 progress=17.4% loss=0.104795 lr=9.586801e-05 lr_backbone=9.586801e-05 elapsed=08:41:22 avg_step=3.60s recent_step=3.07s eta=41:18:31 +2026-09-18 02:10:31 | INFO | train rank=0 | step=8700/50000 progress=17.4% loss=0.135568 lr=9.585483e-05 lr_backbone=9.585483e-05 elapsed=08:41:54 avg_step=3.60s recent_step=3.17s eta=41:17:34 +2026-09-18 02:11:22 | INFO | train rank=0 | step=8710/50000 progress=17.4% loss=0.102797 lr=9.584164e-05 lr_backbone=9.584164e-05 elapsed=08:42:46 avg_step=3.60s recent_step=5.16s eta=41:18:12 +2026-09-18 02:11:54 | INFO | train rank=0 | step=8720/50000 progress=17.4% loss=0.101704 lr=9.582843e-05 lr_backbone=9.582843e-05 elapsed=08:43:17 avg_step=3.60s recent_step=3.12s eta=41:17:14 +2026-09-18 02:12:24 | INFO | train rank=0 | step=8730/50000 progress=17.5% loss=0.087913 lr=9.581519e-05 lr_backbone=9.581519e-05 elapsed=08:43:47 avg_step=3.60s recent_step=2.99s eta=41:16:09 +2026-09-18 02:12:55 | INFO | train rank=0 | step=8740/50000 progress=17.5% loss=0.109927 lr=9.580194e-05 lr_backbone=9.580194e-05 elapsed=08:44:18 avg_step=3.60s recent_step=3.11s eta=41:15:10 +2026-09-18 02:13:45 | INFO | train rank=0 | step=8750/50000 progress=17.5% loss=0.147809 lr=9.578867e-05 lr_backbone=9.578867e-05 elapsed=08:45:09 avg_step=3.60s recent_step=5.08s eta=41:15:43 +2026-09-18 02:14:10 | INFO | train rank=0 | step=8760/50000 progress=17.5% loss=0.180360 lr=9.577537e-05 lr_backbone=9.577537e-05 elapsed=08:45:33 avg_step=3.60s recent_step=2.43s eta=41:14:12 +2026-09-18 02:14:39 | INFO | train rank=0 | step=8770/50000 progress=17.5% loss=0.096634 lr=9.576206e-05 lr_backbone=9.576206e-05 elapsed=08:46:03 avg_step=3.60s recent_step=2.95s eta=41:13:06 +2026-09-18 02:15:06 | INFO | train rank=0 | step=8780/50000 progress=17.6% loss=0.115954 lr=9.574872e-05 lr_backbone=9.574872e-05 elapsed=08:46:29 avg_step=3.60s recent_step=2.63s eta=41:11:44 +2026-09-18 02:15:57 | INFO | train rank=0 | step=8790/50000 progress=17.6% loss=0.113012 lr=9.573537e-05 lr_backbone=9.573537e-05 elapsed=08:47:20 avg_step=3.60s recent_step=5.12s eta=41:12:20 +2026-09-18 02:16:31 | INFO | train rank=0 | step=8800/50000 progress=17.6% loss=0.122081 lr=9.572200e-05 lr_backbone=9.572200e-05 elapsed=08:47:54 avg_step=3.60s recent_step=3.42s eta=41:11:35 +2026-09-18 02:16:58 | INFO | train rank=0 | step=8810/50000 progress=17.6% loss=0.110350 lr=9.570860e-05 lr_backbone=9.570860e-05 elapsed=08:48:21 avg_step=3.60s recent_step=2.64s eta=41:10:15 +2026-09-18 02:17:28 | INFO | train rank=0 | step=8820/50000 progress=17.6% loss=0.104323 lr=9.569519e-05 lr_backbone=9.569519e-05 elapsed=08:48:51 avg_step=3.60s recent_step=3.07s eta=41:09:14 +2026-09-18 02:18:15 | INFO | train rank=0 | step=8830/50000 progress=17.7% loss=0.109242 lr=9.568176e-05 lr_backbone=9.568176e-05 elapsed=08:49:38 avg_step=3.60s recent_step=4.70s eta=41:09:29 +2026-09-18 02:18:46 | INFO | train rank=0 | step=8840/50000 progress=17.7% loss=0.094021 lr=9.566830e-05 lr_backbone=9.566830e-05 elapsed=08:50:09 avg_step=3.60s recent_step=3.09s eta=41:08:30 +2026-09-18 02:19:17 | INFO | train rank=0 | step=8850/50000 progress=17.7% loss=0.068116 lr=9.565483e-05 lr_backbone=9.565483e-05 elapsed=08:50:40 avg_step=3.60s recent_step=3.05s eta=41:07:29 +2026-09-18 02:19:55 | INFO | train rank=0 | step=8860/50000 progress=17.7% loss=0.095795 lr=9.564133e-05 lr_backbone=9.564133e-05 elapsed=08:51:18 avg_step=3.60s recent_step=3.82s eta=41:07:03 +2026-09-18 02:20:46 | INFO | train rank=0 | step=8870/50000 progress=17.7% loss=0.172039 lr=9.562782e-05 lr_backbone=9.562782e-05 elapsed=08:52:09 avg_step=3.60s recent_step=5.13s eta=41:07:38 +2026-09-18 02:21:15 | INFO | train rank=0 | step=8880/50000 progress=17.8% loss=0.088015 lr=9.561429e-05 lr_backbone=9.561429e-05 elapsed=08:52:38 avg_step=3.60s recent_step=2.88s eta=41:06:29 +2026-09-18 02:21:45 | INFO | train rank=0 | step=8890/50000 progress=17.8% loss=0.115871 lr=9.560073e-05 lr_backbone=9.560073e-05 elapsed=08:53:08 avg_step=3.60s recent_step=3.00s eta=41:05:25 +2026-09-18 02:22:15 | INFO | train rank=0 | step=8900/50000 progress=17.8% loss=0.113747 lr=9.558716e-05 lr_backbone=9.558716e-05 elapsed=08:53:38 avg_step=3.60s recent_step=3.00s eta=41:04:22 +2026-09-18 02:23:06 | INFO | train rank=0 | step=8910/50000 progress=17.8% loss=0.103259 lr=9.557357e-05 lr_backbone=9.557357e-05 elapsed=08:54:29 avg_step=3.60s recent_step=5.10s eta=41:04:55 +2026-09-18 02:23:41 | INFO | train rank=0 | step=8920/50000 progress=17.8% loss=0.123978 lr=9.555995e-05 lr_backbone=9.555995e-05 elapsed=08:55:04 avg_step=3.60s recent_step=3.49s eta=41:04:14 +2026-09-18 02:24:17 | INFO | train rank=0 | step=8930/50000 progress=17.9% loss=0.107086 lr=9.554632e-05 lr_backbone=9.554632e-05 elapsed=08:55:40 avg_step=3.60s recent_step=3.59s eta=41:03:38 +2026-09-18 02:24:47 | INFO | train rank=0 | step=8940/50000 progress=17.9% loss=0.093897 lr=9.553267e-05 lr_backbone=9.553267e-05 elapsed=08:56:10 avg_step=3.60s recent_step=2.97s eta=41:02:33 +2026-09-18 02:25:30 | INFO | train rank=0 | step=8950/50000 progress=17.9% loss=0.091543 lr=9.551899e-05 lr_backbone=9.551899e-05 elapsed=08:56:54 avg_step=3.60s recent_step=4.37s eta=41:02:32 +2026-09-18 02:26:07 | INFO | train rank=0 | step=8960/50000 progress=17.9% loss=0.118753 lr=9.550530e-05 lr_backbone=9.550530e-05 elapsed=08:57:30 avg_step=3.60s recent_step=3.65s eta=41:01:59 +2026-09-18 02:26:36 | INFO | train rank=0 | step=8970/50000 progress=17.9% loss=0.125978 lr=9.549159e-05 lr_backbone=9.549159e-05 elapsed=08:57:59 avg_step=3.60s recent_step=2.94s eta=41:00:52 +2026-09-18 02:27:11 | INFO | train rank=0 | step=8980/50000 progress=18.0% loss=0.129278 lr=9.547785e-05 lr_backbone=9.547785e-05 elapsed=08:58:35 avg_step=3.60s recent_step=3.53s eta=41:00:13 +2026-09-18 02:28:01 | INFO | train rank=0 | step=8990/50000 progress=18.0% loss=0.108633 lr=9.546410e-05 lr_backbone=9.546410e-05 elapsed=08:59:25 avg_step=3.60s recent_step=4.98s eta=41:00:41 +2026-09-18 02:28:34 | INFO | train rank=0 | step=9000/50000 progress=18.0% loss=0.120118 lr=9.545033e-05 lr_backbone=9.545033e-05 elapsed=08:59:58 avg_step=3.60s recent_step=3.30s eta=40:59:51 +2026-09-18 02:29:05 | INFO | train rank=0 | step=9010/50000 progress=18.0% loss=0.118069 lr=9.543653e-05 lr_backbone=9.543653e-05 elapsed=09:00:28 avg_step=3.60s recent_step=3.03s eta=40:58:49 +2026-09-18 02:29:31 | INFO | train rank=0 | step=9020/50000 progress=18.0% loss=0.097339 lr=9.542272e-05 lr_backbone=9.542272e-05 elapsed=09:00:54 avg_step=3.60s recent_step=2.65s eta=40:57:30 +2026-09-18 02:30:31 | INFO | train rank=0 | step=9030/50000 progress=18.1% loss=0.114336 lr=9.540889e-05 lr_backbone=9.540889e-05 elapsed=09:01:55 avg_step=3.60s recent_step=6.02s eta=40:58:44 +2026-09-18 02:30:59 | INFO | train rank=0 | step=9040/50000 progress=18.1% loss=0.109152 lr=9.539504e-05 lr_backbone=9.539504e-05 elapsed=09:02:22 avg_step=3.60s recent_step=2.78s eta=40:57:31 +2026-09-18 02:31:26 | INFO | train rank=0 | step=9050/50000 progress=18.1% loss=0.123614 lr=9.538116e-05 lr_backbone=9.538116e-05 elapsed=09:02:50 avg_step=3.60s recent_step=2.71s eta=40:56:15 +2026-09-18 02:31:51 | INFO | train rank=0 | step=9060/50000 progress=18.1% loss=0.091734 lr=9.536727e-05 lr_backbone=9.536727e-05 elapsed=09:03:14 avg_step=3.60s recent_step=2.43s eta=40:54:46 +2026-09-18 02:32:42 | INFO | train rank=0 | step=9070/50000 progress=18.1% loss=0.176323 lr=9.535336e-05 lr_backbone=9.535336e-05 elapsed=09:04:05 avg_step=3.60s recent_step=5.15s eta=40:55:20 +2026-09-18 02:33:18 | INFO | train rank=0 | step=9080/50000 progress=18.2% loss=0.117329 lr=9.533943e-05 lr_backbone=9.533943e-05 elapsed=09:04:41 avg_step=3.60s recent_step=3.59s eta=40:54:43 +2026-09-18 02:33:44 | INFO | train rank=0 | step=9090/50000 progress=18.2% loss=0.104434 lr=9.532548e-05 lr_backbone=9.532548e-05 elapsed=09:05:07 avg_step=3.60s recent_step=2.61s eta=40:53:23 +2026-09-18 02:34:19 | INFO | train rank=0 | step=9100/50000 progress=18.2% loss=0.105616 lr=9.531151e-05 lr_backbone=9.531151e-05 elapsed=09:05:42 avg_step=3.60s recent_step=3.50s eta=40:52:43 +2026-09-18 02:35:15 | INFO | train rank=0 | step=9110/50000 progress=18.2% loss=0.121307 lr=9.529752e-05 lr_backbone=9.529752e-05 elapsed=09:06:38 avg_step=3.60s recent_step=5.55s eta=40:53:34 +2026-09-18 02:35:43 | INFO | train rank=0 | step=9120/50000 progress=18.2% loss=0.121362 lr=9.528350e-05 lr_backbone=9.528350e-05 elapsed=09:07:07 avg_step=3.60s recent_step=2.88s eta=40:52:26 +2026-09-18 02:36:09 | INFO | train rank=0 | step=9130/50000 progress=18.3% loss=0.129221 lr=9.526947e-05 lr_backbone=9.526947e-05 elapsed=09:07:32 avg_step=3.60s recent_step=2.52s eta=40:51:02 +2026-09-18 02:36:40 | INFO | train rank=0 | step=9140/50000 progress=18.3% loss=0.108092 lr=9.525542e-05 lr_backbone=9.525542e-05 elapsed=09:08:03 avg_step=3.60s recent_step=3.14s eta=40:50:05 +2026-09-18 02:37:34 | INFO | train rank=0 | step=9150/50000 progress=18.3% loss=0.108086 lr=9.524135e-05 lr_backbone=9.524135e-05 elapsed=09:08:57 avg_step=3.60s recent_step=5.41s eta=40:50:50 +2026-09-18 02:38:04 | INFO | train rank=0 | step=9160/50000 progress=18.3% loss=0.119878 lr=9.522726e-05 lr_backbone=9.522726e-05 elapsed=09:09:28 avg_step=3.60s recent_step=3.03s eta=40:49:49 +2026-09-18 02:38:43 | INFO | train rank=0 | step=9170/50000 progress=18.3% loss=0.118566 lr=9.521315e-05 lr_backbone=9.521315e-05 elapsed=09:10:07 avg_step=3.60s recent_step=3.88s eta=40:49:26 +2026-09-18 02:39:16 | INFO | train rank=0 | step=9180/50000 progress=18.4% loss=0.132993 lr=9.519902e-05 lr_backbone=9.519902e-05 elapsed=09:10:39 avg_step=3.60s recent_step=3.28s eta=40:48:35 +2026-09-18 02:39:46 | INFO | train rank=0 | step=9190/50000 progress=18.4% loss=0.120743 lr=9.518487e-05 lr_backbone=9.518487e-05 elapsed=09:11:09 avg_step=3.60s recent_step=3.01s eta=40:47:33 +2026-09-18 02:40:22 | INFO | train rank=0 | step=9200/50000 progress=18.4% loss=0.121644 lr=9.517070e-05 lr_backbone=9.517070e-05 elapsed=09:11:45 avg_step=3.60s recent_step=3.56s eta=40:46:55 +2026-09-18 02:41:03 | INFO | train rank=0 | step=9210/50000 progress=18.4% loss=0.102762 lr=9.515652e-05 lr_backbone=9.515652e-05 elapsed=09:12:26 avg_step=3.60s recent_step=4.12s eta=40:46:42 +2026-09-18 02:41:46 | INFO | train rank=0 | step=9220/50000 progress=18.4% loss=0.099430 lr=9.514231e-05 lr_backbone=9.514231e-05 elapsed=09:13:09 avg_step=3.60s recent_step=4.31s eta=40:46:38 +2026-09-18 02:42:18 | INFO | train rank=0 | step=9230/50000 progress=18.5% loss=0.131029 lr=9.512808e-05 lr_backbone=9.512808e-05 elapsed=09:13:41 avg_step=3.60s recent_step=3.19s eta=40:45:44 +2026-09-18 02:42:41 | INFO | train rank=0 | step=9240/50000 progress=18.5% loss=0.123137 lr=9.511383e-05 lr_backbone=9.511383e-05 elapsed=09:14:05 avg_step=3.60s recent_step=2.35s eta=40:44:13 +2026-09-18 02:43:29 | INFO | train rank=0 | step=9250/50000 progress=18.5% loss=0.106577 lr=9.509956e-05 lr_backbone=9.509956e-05 elapsed=09:14:53 avg_step=3.60s recent_step=4.78s eta=40:44:29 +2026-09-18 02:44:08 | INFO | train rank=0 | step=9260/50000 progress=18.5% loss=0.112663 lr=9.508527e-05 lr_backbone=9.508527e-05 elapsed=09:15:31 avg_step=3.60s recent_step=3.85s eta=40:44:04 +2026-09-18 02:44:33 | INFO | train rank=0 | step=9270/50000 progress=18.5% loss=0.093407 lr=9.507097e-05 lr_backbone=9.507097e-05 elapsed=09:15:57 avg_step=3.60s recent_step=2.58s eta=40:42:43 +2026-09-18 02:45:05 | INFO | train rank=0 | step=9280/50000 progress=18.6% loss=0.097341 lr=9.505664e-05 lr_backbone=9.505664e-05 elapsed=09:16:29 avg_step=3.60s recent_step=3.20s eta=40:41:49 +2026-09-18 02:45:54 | INFO | train rank=0 | step=9290/50000 progress=18.6% loss=0.117222 lr=9.504229e-05 lr_backbone=9.504229e-05 elapsed=09:17:17 avg_step=3.60s recent_step=4.81s eta=40:42:06 +2026-09-18 02:46:27 | INFO | train rank=0 | step=9300/50000 progress=18.6% loss=0.090088 lr=9.502793e-05 lr_backbone=9.502793e-05 elapsed=09:17:51 avg_step=3.60s recent_step=3.37s eta=40:41:20 +2026-09-18 02:47:02 | INFO | train rank=0 | step=9310/50000 progress=18.6% loss=0.105517 lr=9.501354e-05 lr_backbone=9.501354e-05 elapsed=09:18:25 avg_step=3.60s recent_step=3.43s eta=40:40:37 +2026-09-18 02:47:45 | INFO | train rank=0 | step=9320/50000 progress=18.6% loss=0.104214 lr=9.499913e-05 lr_backbone=9.499913e-05 elapsed=09:19:08 avg_step=3.60s recent_step=4.31s eta=40:40:32 +2026-09-18 02:48:34 | INFO | train rank=0 | step=9330/50000 progress=18.7% loss=0.103595 lr=9.498471e-05 lr_backbone=9.498471e-05 elapsed=09:19:58 avg_step=3.60s recent_step=4.97s eta=40:40:56 +2026-09-18 02:49:09 | INFO | train rank=0 | step=9340/50000 progress=18.7% loss=0.096857 lr=9.497026e-05 lr_backbone=9.497026e-05 elapsed=09:20:32 avg_step=3.60s recent_step=3.48s eta=40:40:14 +2026-09-18 02:49:40 | INFO | train rank=0 | step=9350/50000 progress=18.7% loss=0.099714 lr=9.495580e-05 lr_backbone=9.495580e-05 elapsed=09:21:03 avg_step=3.60s recent_step=3.08s eta=40:39:16 +2026-09-18 02:50:13 | INFO | train rank=0 | step=9360/50000 progress=18.7% loss=0.110994 lr=9.494131e-05 lr_backbone=9.494131e-05 elapsed=09:21:36 avg_step=3.60s recent_step=3.29s eta=40:38:26 +2026-09-18 02:50:55 | INFO | train rank=0 | step=9370/50000 progress=18.7% loss=0.103037 lr=9.492681e-05 lr_backbone=9.492681e-05 elapsed=09:22:18 avg_step=3.60s recent_step=4.18s eta=40:38:15 +2026-09-18 02:51:27 | INFO | train rank=0 | step=9380/50000 progress=18.8% loss=0.101224 lr=9.491228e-05 lr_backbone=9.491228e-05 elapsed=09:22:51 avg_step=3.60s recent_step=3.29s eta=40:37:25 +2026-09-18 02:52:01 | INFO | train rank=0 | step=9390/50000 progress=18.8% loss=0.120197 lr=9.489774e-05 lr_backbone=9.489774e-05 elapsed=09:23:24 avg_step=3.60s recent_step=3.35s eta=40:36:39 +2026-09-18 02:52:42 | INFO | train rank=0 | step=9400/50000 progress=18.8% loss=0.154221 lr=9.488318e-05 lr_backbone=9.488318e-05 elapsed=09:24:05 avg_step=3.60s recent_step=4.07s eta=40:36:23 +2026-09-18 02:53:20 | INFO | train rank=0 | step=9410/50000 progress=18.8% loss=0.097081 lr=9.486859e-05 lr_backbone=9.486859e-05 elapsed=09:24:43 avg_step=3.60s recent_step=3.83s eta=40:35:57 +2026-09-18 02:53:54 | INFO | train rank=0 | step=9420/50000 progress=18.8% loss=0.113272 lr=9.485399e-05 lr_backbone=9.485399e-05 elapsed=09:25:17 avg_step=3.60s recent_step=3.37s eta=40:35:11 +2026-09-18 02:54:20 | INFO | train rank=0 | step=9430/50000 progress=18.9% loss=0.109913 lr=9.483937e-05 lr_backbone=9.483937e-05 elapsed=09:25:43 avg_step=3.60s recent_step=2.65s eta=40:33:54 +2026-09-18 02:55:07 | INFO | train rank=0 | step=9440/50000 progress=18.9% loss=0.095458 lr=9.482473e-05 lr_backbone=9.482473e-05 elapsed=09:26:30 avg_step=3.60s recent_step=4.69s eta=40:34:05 +2026-09-18 02:55:50 | INFO | train rank=0 | step=9450/50000 progress=18.9% loss=0.118945 lr=9.481007e-05 lr_backbone=9.481007e-05 elapsed=09:27:13 avg_step=3.60s recent_step=4.30s eta=40:33:59 +2026-09-18 02:56:25 | INFO | train rank=0 | step=9460/50000 progress=18.9% loss=0.108775 lr=9.479539e-05 lr_backbone=9.479539e-05 elapsed=09:27:48 avg_step=3.60s recent_step=3.45s eta=40:33:17 +2026-09-18 02:56:55 | INFO | train rank=0 | step=9470/50000 progress=18.9% loss=0.126301 lr=9.478069e-05 lr_backbone=9.478069e-05 elapsed=09:28:19 avg_step=3.60s recent_step=3.08s eta=40:32:19 +2026-09-18 02:57:26 | INFO | train rank=0 | step=9480/50000 progress=19.0% loss=0.121218 lr=9.476597e-05 lr_backbone=9.476597e-05 elapsed=09:28:50 avg_step=3.60s recent_step=3.07s eta=40:31:20 +2026-09-18 02:58:21 | INFO | train rank=0 | step=9490/50000 progress=19.0% loss=0.138640 lr=9.475123e-05 lr_backbone=9.475123e-05 elapsed=09:29:44 avg_step=3.60s recent_step=5.48s eta=40:32:04 +2026-09-18 02:58:49 | INFO | train rank=0 | step=9500/50000 progress=19.0% loss=0.174709 lr=9.473647e-05 lr_backbone=9.473647e-05 elapsed=09:30:12 avg_step=3.60s recent_step=2.80s eta=40:30:54 +2026-09-18 02:59:21 | INFO | train rank=0 | step=9510/50000 progress=19.0% loss=0.151611 lr=9.472169e-05 lr_backbone=9.472169e-05 elapsed=09:30:44 avg_step=3.60s recent_step=3.18s eta=40:30:00 +2026-09-18 02:59:54 | INFO | train rank=0 | step=9520/50000 progress=19.0% loss=0.078134 lr=9.470689e-05 lr_backbone=9.470689e-05 elapsed=09:31:17 avg_step=3.60s recent_step=3.33s eta=40:29:13 +2026-09-18 03:00:32 | INFO | train rank=0 | step=9530/50000 progress=19.1% loss=0.107022 lr=9.469207e-05 lr_backbone=9.469207e-05 elapsed=09:31:55 avg_step=3.60s recent_step=3.74s eta=40:28:43 +2026-09-18 03:01:07 | INFO | train rank=0 | step=9540/50000 progress=19.1% loss=0.083497 lr=9.467723e-05 lr_backbone=9.467723e-05 elapsed=09:32:30 avg_step=3.60s recent_step=3.55s eta=40:28:04 +2026-09-18 03:02:03 | INFO | train rank=0 | step=9550/50000 progress=19.1% loss=0.095360 lr=9.466238e-05 lr_backbone=9.466238e-05 elapsed=09:33:26 avg_step=3.60s recent_step=5.60s eta=40:28:53 +2026-09-18 03:02:34 | INFO | train rank=0 | step=9560/50000 progress=19.1% loss=0.088098 lr=9.464750e-05 lr_backbone=9.464750e-05 elapsed=09:33:57 avg_step=3.60s recent_step=3.10s eta=40:27:56 +2026-09-18 03:03:03 | INFO | train rank=0 | step=9570/50000 progress=19.1% loss=0.121980 lr=9.463260e-05 lr_backbone=9.463260e-05 elapsed=09:34:26 avg_step=3.60s recent_step=2.88s eta=40:26:49 +2026-09-18 03:03:32 | INFO | train rank=0 | step=9580/50000 progress=19.2% loss=0.144233 lr=9.461769e-05 lr_backbone=9.461769e-05 elapsed=09:34:55 avg_step=3.60s recent_step=2.88s eta=40:25:43 +2026-09-18 03:04:18 | INFO | train rank=0 | step=9590/50000 progress=19.2% loss=0.116055 lr=9.460275e-05 lr_backbone=9.460275e-05 elapsed=09:35:41 avg_step=3.60s recent_step=4.61s eta=40:25:49 +2026-09-18 03:04:53 | INFO | train rank=0 | step=9600/50000 progress=19.2% loss=0.152585 lr=9.458780e-05 lr_backbone=9.458780e-05 elapsed=09:36:16 avg_step=3.60s recent_step=3.51s eta=40:25:09 +2026-09-18 03:05:23 | INFO | train rank=0 | step=9610/50000 progress=19.2% loss=0.096016 lr=9.457282e-05 lr_backbone=9.457282e-05 elapsed=09:36:46 avg_step=3.60s recent_step=2.99s eta=40:24:08 +2026-09-18 03:05:59 | INFO | train rank=0 | step=9620/50000 progress=19.2% loss=0.112182 lr=9.455783e-05 lr_backbone=9.455783e-05 elapsed=09:37:22 avg_step=3.60s recent_step=3.59s eta=40:23:31 +2026-09-18 03:06:45 | INFO | train rank=0 | step=9630/50000 progress=19.3% loss=0.111518 lr=9.454282e-05 lr_backbone=9.454282e-05 elapsed=09:38:08 avg_step=3.60s recent_step=4.62s eta=40:23:38 +2026-09-18 03:07:21 | INFO | train rank=0 | step=9640/50000 progress=19.3% loss=0.130812 lr=9.452778e-05 lr_backbone=9.452778e-05 elapsed=09:38:44 avg_step=3.60s recent_step=3.63s eta=40:23:03 +2026-09-18 03:07:55 | INFO | train rank=0 | step=9650/50000 progress=19.3% loss=0.109779 lr=9.451273e-05 lr_backbone=9.451273e-05 elapsed=09:39:19 avg_step=3.60s recent_step=3.41s eta=40:22:19 +2026-09-18 03:08:31 | INFO | train rank=0 | step=9660/50000 progress=19.3% loss=0.091000 lr=9.449766e-05 lr_backbone=9.449766e-05 elapsed=09:39:55 avg_step=3.60s recent_step=3.60s eta=40:21:43 +2026-09-18 03:09:02 | INFO | train rank=0 | step=9670/50000 progress=19.3% loss=0.167626 lr=9.448257e-05 lr_backbone=9.448257e-05 elapsed=09:40:25 avg_step=3.60s recent_step=3.04s eta=40:20:43 +2026-09-18 03:09:50 | INFO | train rank=0 | step=9680/50000 progress=19.4% loss=0.171842 lr=9.446746e-05 lr_backbone=9.446746e-05 elapsed=09:41:13 avg_step=3.60s recent_step=4.83s eta=40:20:58 +2026-09-18 03:10:23 | INFO | train rank=0 | step=9690/50000 progress=19.4% loss=0.102513 lr=9.445233e-05 lr_backbone=9.445233e-05 elapsed=09:41:46 avg_step=3.60s recent_step=3.26s eta=40:20:08 +2026-09-18 03:10:52 | INFO | train rank=0 | step=9700/50000 progress=19.4% loss=0.109339 lr=9.443718e-05 lr_backbone=9.443718e-05 elapsed=09:42:16 avg_step=3.60s recent_step=3.00s eta=40:19:07 +2026-09-18 03:11:44 | INFO | train rank=0 | step=9710/50000 progress=19.4% loss=0.150504 lr=9.442201e-05 lr_backbone=9.442201e-05 elapsed=09:43:07 avg_step=3.60s recent_step=5.11s eta=40:19:34 +2026-09-18 03:12:18 | INFO | train rank=0 | step=9720/50000 progress=19.4% loss=0.124264 lr=9.440682e-05 lr_backbone=9.440682e-05 elapsed=09:43:41 avg_step=3.60s recent_step=3.40s eta=40:18:49 +2026-09-18 03:12:51 | INFO | train rank=0 | step=9730/50000 progress=19.5% loss=0.115312 lr=9.439161e-05 lr_backbone=9.439161e-05 elapsed=09:44:14 avg_step=3.60s recent_step=3.30s eta=40:18:01 +2026-09-18 03:13:25 | INFO | train rank=0 | step=9740/50000 progress=19.5% loss=0.086588 lr=9.437639e-05 lr_backbone=9.437639e-05 elapsed=09:44:48 avg_step=3.60s recent_step=3.40s eta=40:17:16 +2026-09-18 03:14:01 | INFO | train rank=0 | step=9750/50000 progress=19.5% loss=0.098322 lr=9.436114e-05 lr_backbone=9.436114e-05 elapsed=09:45:24 avg_step=3.60s recent_step=3.66s eta=40:16:43 +2026-09-18 03:14:48 | INFO | train rank=0 | step=9760/50000 progress=19.5% loss=0.074771 lr=9.434587e-05 lr_backbone=9.434587e-05 elapsed=09:46:11 avg_step=3.60s recent_step=4.65s eta=40:16:50 +2026-09-18 03:15:13 | INFO | train rank=0 | step=9770/50000 progress=19.5% loss=0.107295 lr=9.433059e-05 lr_backbone=9.433059e-05 elapsed=09:46:37 avg_step=3.60s recent_step=2.58s eta=40:15:32 +2026-09-18 03:15:50 | INFO | train rank=0 | step=9780/50000 progress=19.6% loss=0.090621 lr=9.431528e-05 lr_backbone=9.431528e-05 elapsed=09:47:14 avg_step=3.60s recent_step=3.69s eta=40:14:59 +2026-09-18 03:16:29 | INFO | train rank=0 | step=9790/50000 progress=19.6% loss=0.097801 lr=9.429996e-05 lr_backbone=9.429996e-05 elapsed=09:47:52 avg_step=3.60s recent_step=3.87s eta=40:14:34 +2026-09-18 03:17:19 | INFO | train rank=0 | step=9800/50000 progress=19.6% loss=0.164214 lr=9.428462e-05 lr_backbone=9.428462e-05 elapsed=09:48:42 avg_step=3.60s recent_step=4.95s eta=40:14:53 +2026-09-18 03:17:44 | INFO | train rank=0 | step=9810/50000 progress=19.6% loss=0.107572 lr=9.426925e-05 lr_backbone=9.426925e-05 elapsed=09:49:07 avg_step=3.60s recent_step=2.51s eta=40:13:33 +2026-09-18 03:18:15 | INFO | train rank=0 | step=9820/50000 progress=19.6% loss=0.160420 lr=9.425387e-05 lr_backbone=9.425387e-05 elapsed=09:49:38 avg_step=3.60s recent_step=3.14s eta=40:12:38 +2026-09-18 03:18:50 | INFO | train rank=0 | step=9830/50000 progress=19.7% loss=0.078445 lr=9.423847e-05 lr_backbone=9.423847e-05 elapsed=09:50:14 avg_step=3.60s recent_step=3.53s eta=40:11:59 +2026-09-18 03:19:43 | INFO | train rank=0 | step=9840/50000 progress=19.7% loss=0.138165 lr=9.422305e-05 lr_backbone=9.422305e-05 elapsed=09:51:07 avg_step=3.60s recent_step=5.30s eta=40:12:32 +2026-09-18 03:20:15 | INFO | train rank=0 | step=9850/50000 progress=19.7% loss=0.135651 lr=9.420761e-05 lr_backbone=9.420761e-05 elapsed=09:51:38 avg_step=3.60s recent_step=3.12s eta=40:11:36 +2026-09-18 03:20:49 | INFO | train rank=0 | step=9860/50000 progress=19.7% loss=0.114649 lr=9.419215e-05 lr_backbone=9.419215e-05 elapsed=09:52:12 avg_step=3.60s recent_step=3.40s eta=40:10:52 +2026-09-18 03:21:19 | INFO | train rank=0 | step=9870/50000 progress=19.7% loss=0.118744 lr=9.417667e-05 lr_backbone=9.417667e-05 elapsed=09:52:42 avg_step=3.60s recent_step=3.03s eta=40:09:53 +2026-09-18 03:22:12 | INFO | train rank=0 | step=9880/50000 progress=19.8% loss=0.096412 lr=9.416117e-05 lr_backbone=9.416117e-05 elapsed=09:53:35 avg_step=3.60s recent_step=5.27s eta=40:10:25 +2026-09-18 03:22:43 | INFO | train rank=0 | step=9890/50000 progress=19.8% loss=0.098018 lr=9.414565e-05 lr_backbone=9.414565e-05 elapsed=09:54:06 avg_step=3.60s recent_step=3.11s eta=40:09:28 +2026-09-18 03:23:16 | INFO | train rank=0 | step=9900/50000 progress=19.8% loss=0.089726 lr=9.413012e-05 lr_backbone=9.413012e-05 elapsed=09:54:40 avg_step=3.60s recent_step=3.35s eta=40:08:42 +2026-09-18 03:23:41 | INFO | train rank=0 | step=9910/50000 progress=19.8% loss=0.087105 lr=9.411456e-05 lr_backbone=9.411456e-05 elapsed=09:55:04 avg_step=3.60s recent_step=2.44s eta=40:07:19 +2026-09-18 03:24:20 | INFO | train rank=0 | step=9920/50000 progress=19.8% loss=0.107612 lr=9.409899e-05 lr_backbone=9.409899e-05 elapsed=09:55:44 avg_step=3.60s recent_step=3.98s eta=40:06:58 +2026-09-18 03:24:50 | INFO | train rank=0 | step=9930/50000 progress=19.9% loss=0.110744 lr=9.408339e-05 lr_backbone=9.408339e-05 elapsed=09:56:14 avg_step=3.60s recent_step=2.99s eta=40:05:57 +2026-09-18 03:25:25 | INFO | train rank=0 | step=9940/50000 progress=19.9% loss=0.109150 lr=9.406778e-05 lr_backbone=9.406778e-05 elapsed=09:56:48 avg_step=3.60s recent_step=3.42s eta=40:05:14 +2026-09-18 03:25:55 | INFO | train rank=0 | step=9950/50000 progress=19.9% loss=0.116785 lr=9.405214e-05 lr_backbone=9.405214e-05 elapsed=09:57:18 avg_step=3.60s recent_step=3.02s eta=40:04:14 +2026-09-18 03:26:55 | INFO | train rank=0 | step=9960/50000 progress=19.9% loss=0.100173 lr=9.403649e-05 lr_backbone=9.403649e-05 elapsed=09:58:18 avg_step=3.60s recent_step=6.00s eta=40:05:15 +2026-09-18 03:27:24 | INFO | train rank=0 | step=9970/50000 progress=19.9% loss=0.106077 lr=9.402082e-05 lr_backbone=9.402082e-05 elapsed=09:58:47 avg_step=3.60s recent_step=2.89s eta=40:04:10 +2026-09-18 03:28:01 | INFO | train rank=0 | step=9980/50000 progress=20.0% loss=0.182578 lr=9.400513e-05 lr_backbone=9.400513e-05 elapsed=09:59:25 avg_step=3.60s recent_step=3.76s eta=40:03:40 +2026-09-18 03:28:29 | INFO | train rank=0 | step=9990/50000 progress=20.0% loss=0.110613 lr=9.398942e-05 lr_backbone=9.398942e-05 elapsed=09:59:52 avg_step=3.60s recent_step=2.73s eta=40:02:29 +2026-09-18 03:29:21 | INFO | train rank=0 | step=10000/50000 progress=20.0% loss=0.153531 lr=9.397369e-05 lr_backbone=9.397369e-05 elapsed=10:00:44 avg_step=3.60s recent_step=5.21s eta=40:02:57 +2026-09-18 03:29:50 | INFO | train rank=0 | step=10010/50000 progress=20.0% loss=0.150553 lr=9.395794e-05 lr_backbone=9.395794e-05 elapsed=10:01:13 avg_step=3.60s recent_step=2.94s eta=40:01:55 +2026-09-18 03:30:15 | INFO | train rank=0 | step=10020/50000 progress=20.0% loss=0.150168 lr=9.394217e-05 lr_backbone=9.394217e-05 elapsed=10:01:38 avg_step=3.60s recent_step=2.49s eta=40:00:34 +2026-09-18 03:30:52 | INFO | train rank=0 | step=10030/50000 progress=20.1% loss=0.097832 lr=9.392638e-05 lr_backbone=9.392638e-05 elapsed=10:02:16 avg_step=3.60s recent_step=3.74s eta=40:00:04 +2026-09-18 03:31:40 | INFO | train rank=0 | step=10040/50000 progress=20.1% loss=0.089654 lr=9.391058e-05 lr_backbone=9.391058e-05 elapsed=10:03:03 avg_step=3.60s recent_step=4.76s eta=40:00:14 +2026-09-18 03:32:09 | INFO | train rank=0 | step=10050/50000 progress=20.1% loss=0.102840 lr=9.389475e-05 lr_backbone=9.389475e-05 elapsed=10:03:33 avg_step=3.60s recent_step=2.94s eta=39:59:12 +2026-09-18 03:32:41 | INFO | train rank=0 | step=10060/50000 progress=20.1% loss=0.132515 lr=9.387891e-05 lr_backbone=9.387891e-05 elapsed=10:04:04 avg_step=3.60s recent_step=3.14s eta=39:58:17 +2026-09-18 03:33:12 | INFO | train rank=0 | step=10070/50000 progress=20.1% loss=0.085631 lr=9.386304e-05 lr_backbone=9.386304e-05 elapsed=10:04:35 avg_step=3.60s recent_step=3.08s eta=39:57:20 +2026-09-18 03:33:59 | INFO | train rank=0 | step=10080/50000 progress=20.2% loss=0.142576 lr=9.384716e-05 lr_backbone=9.384716e-05 elapsed=10:05:22 avg_step=3.60s recent_step=4.76s eta=39:57:30 +2026-09-18 03:34:33 | INFO | train rank=0 | step=10090/50000 progress=20.2% loss=0.102169 lr=9.383126e-05 lr_backbone=9.383126e-05 elapsed=10:05:56 avg_step=3.60s recent_step=3.39s eta=39:56:46 +2026-09-18 03:35:05 | INFO | train rank=0 | step=10100/50000 progress=20.2% loss=0.190290 lr=9.381533e-05 lr_backbone=9.381533e-05 elapsed=10:06:29 avg_step=3.60s recent_step=3.24s eta=39:55:55 +2026-09-18 03:35:42 | INFO | train rank=0 | step=10110/50000 progress=20.2% loss=0.110374 lr=9.379939e-05 lr_backbone=9.379939e-05 elapsed=10:07:06 avg_step=3.60s recent_step=3.68s eta=39:55:22 +2026-09-18 03:36:28 | INFO | train rank=0 | step=10120/50000 progress=20.2% loss=0.090003 lr=9.378343e-05 lr_backbone=9.378343e-05 elapsed=10:07:51 avg_step=3.60s recent_step=4.56s eta=39:55:24 +2026-09-18 03:36:58 | INFO | train rank=0 | step=10130/50000 progress=20.3% loss=0.151382 lr=9.376745e-05 lr_backbone=9.376745e-05 elapsed=10:08:21 avg_step=3.60s recent_step=2.96s eta=39:54:23 +2026-09-18 03:37:29 | INFO | train rank=0 | step=10140/50000 progress=20.3% loss=0.164296 lr=9.375146e-05 lr_backbone=9.375146e-05 elapsed=10:08:53 avg_step=3.60s recent_step=3.17s eta=39:53:30 +2026-09-18 03:38:03 | INFO | train rank=0 | step=10150/50000 progress=20.3% loss=0.118419 lr=9.373544e-05 lr_backbone=9.373544e-05 elapsed=10:09:26 avg_step=3.60s recent_step=3.35s eta=39:52:44 +2026-09-18 03:38:47 | INFO | train rank=0 | step=10160/50000 progress=20.3% loss=0.097559 lr=9.371940e-05 lr_backbone=9.371940e-05 elapsed=10:10:10 avg_step=3.60s recent_step=4.41s eta=39:52:40 +2026-09-18 03:39:20 | INFO | train rank=0 | step=10170/50000 progress=20.3% loss=0.096996 lr=9.370335e-05 lr_backbone=9.370335e-05 elapsed=10:10:44 avg_step=3.60s recent_step=3.33s eta=39:51:53 +2026-09-18 03:39:52 | INFO | train rank=0 | step=10180/50000 progress=20.4% loss=0.115360 lr=9.368727e-05 lr_backbone=9.368727e-05 elapsed=10:11:16 avg_step=3.60s recent_step=3.21s eta=39:51:02 +2026-09-18 03:40:26 | INFO | train rank=0 | step=10190/50000 progress=20.4% loss=0.121596 lr=9.367118e-05 lr_backbone=9.367118e-05 elapsed=10:11:50 avg_step=3.60s recent_step=3.39s eta=39:50:18 +2026-09-18 03:41:15 | INFO | train rank=0 | step=10200/50000 progress=20.4% loss=0.102879 lr=9.365506e-05 lr_backbone=9.365506e-05 elapsed=10:12:38 avg_step=3.60s recent_step=4.86s eta=39:50:31 +2026-09-18 03:41:50 | INFO | train rank=0 | step=10210/50000 progress=20.4% loss=0.111188 lr=9.363893e-05 lr_backbone=9.363893e-05 elapsed=10:13:13 avg_step=3.60s recent_step=3.50s eta=39:49:51 +2026-09-18 03:42:21 | INFO | train rank=0 | step=10220/50000 progress=20.4% loss=0.113732 lr=9.362278e-05 lr_backbone=9.362278e-05 elapsed=10:13:44 avg_step=3.60s recent_step=3.12s eta=39:48:56 +2026-09-18 03:42:54 | INFO | train rank=0 | step=10230/50000 progress=20.5% loss=0.103123 lr=9.360661e-05 lr_backbone=9.360661e-05 elapsed=10:14:18 avg_step=3.60s recent_step=3.30s eta=39:48:08 +2026-09-18 03:43:46 | INFO | train rank=0 | step=10240/50000 progress=20.5% loss=0.086501 lr=9.359042e-05 lr_backbone=9.359042e-05 elapsed=10:15:09 avg_step=3.60s recent_step=5.14s eta=39:48:32 +2026-09-18 03:44:16 | INFO | train rank=0 | step=10250/50000 progress=20.5% loss=0.110709 lr=9.357421e-05 lr_backbone=9.357421e-05 elapsed=10:15:39 avg_step=3.60s recent_step=3.02s eta=39:47:33 +2026-09-18 03:44:50 | INFO | train rank=0 | step=10260/50000 progress=20.5% loss=0.088195 lr=9.355798e-05 lr_backbone=9.355798e-05 elapsed=10:16:13 avg_step=3.60s recent_step=3.39s eta=39:46:49 +2026-09-18 03:45:24 | INFO | train rank=0 | step=10270/50000 progress=20.5% loss=0.105002 lr=9.354174e-05 lr_backbone=9.354174e-05 elapsed=10:16:47 avg_step=3.60s recent_step=3.43s eta=39:46:06 +2026-09-18 03:46:11 | INFO | train rank=0 | step=10280/50000 progress=20.6% loss=0.121323 lr=9.352547e-05 lr_backbone=9.352547e-05 elapsed=10:17:34 avg_step=3.60s recent_step=4.72s eta=39:46:13 +2026-09-18 03:46:42 | INFO | train rank=0 | step=10290/50000 progress=20.6% loss=0.082105 lr=9.350919e-05 lr_backbone=9.350919e-05 elapsed=10:18:06 avg_step=3.60s recent_step=3.13s eta=39:45:18 +2026-09-18 03:47:14 | INFO | train rank=0 | step=10300/50000 progress=20.6% loss=0.074905 lr=9.349288e-05 lr_backbone=9.349288e-05 elapsed=10:18:38 avg_step=3.60s recent_step=3.19s eta=39:44:26 +2026-09-18 03:47:43 | INFO | train rank=0 | step=10310/50000 progress=20.6% loss=0.141582 lr=9.347656e-05 lr_backbone=9.347656e-05 elapsed=10:19:06 avg_step=3.60s recent_step=2.87s eta=39:43:22 +2026-09-18 03:48:33 | INFO | train rank=0 | step=10320/50000 progress=20.6% loss=0.114154 lr=9.346022e-05 lr_backbone=9.346022e-05 elapsed=10:19:57 avg_step=3.60s recent_step=5.02s eta=39:43:41 +2026-09-18 03:49:03 | INFO | train rank=0 | step=10330/50000 progress=20.7% loss=0.099107 lr=9.344386e-05 lr_backbone=9.344386e-05 elapsed=10:20:26 avg_step=3.60s recent_step=2.95s eta=39:42:39 +2026-09-18 03:49:37 | INFO | train rank=0 | step=10340/50000 progress=20.7% loss=0.106279 lr=9.342748e-05 lr_backbone=9.342748e-05 elapsed=10:21:00 avg_step=3.60s recent_step=3.39s eta=39:41:55 +2026-09-18 03:50:13 | INFO | train rank=0 | step=10350/50000 progress=20.7% loss=0.108984 lr=9.341108e-05 lr_backbone=9.341108e-05 elapsed=10:21:36 avg_step=3.60s recent_step=3.64s eta=39:41:20 +2026-09-18 03:50:51 | INFO | train rank=0 | step=10360/50000 progress=20.7% loss=0.150754 lr=9.339466e-05 lr_backbone=9.339466e-05 elapsed=10:22:15 avg_step=3.60s recent_step=3.83s eta=39:40:53 +2026-09-18 03:51:24 | INFO | train rank=0 | step=10370/50000 progress=20.7% loss=0.116232 lr=9.337823e-05 lr_backbone=9.337823e-05 elapsed=10:22:48 avg_step=3.60s recent_step=3.30s eta=39:40:06 +2026-09-18 03:51:49 | INFO | train rank=0 | step=10380/50000 progress=20.8% loss=0.107849 lr=9.336177e-05 lr_backbone=9.336177e-05 elapsed=10:23:12 avg_step=3.60s recent_step=2.46s eta=39:38:46 +2026-09-18 03:52:18 | INFO | train rank=0 | step=10390/50000 progress=20.8% loss=0.108160 lr=9.334530e-05 lr_backbone=9.334530e-05 elapsed=10:23:42 avg_step=3.60s recent_step=2.94s eta=39:37:45 +2026-09-18 03:53:16 | INFO | train rank=0 | step=10400/50000 progress=20.8% loss=0.118393 lr=9.332880e-05 lr_backbone=9.332880e-05 elapsed=10:24:39 avg_step=3.60s recent_step=5.71s eta=39:38:29 +2026-09-18 03:53:45 | INFO | train rank=0 | step=10410/50000 progress=20.8% loss=0.091190 lr=9.331229e-05 lr_backbone=9.331229e-05 elapsed=10:25:09 avg_step=3.60s recent_step=2.99s eta=39:37:30 +2026-09-18 03:54:18 | INFO | train rank=0 | step=10420/50000 progress=20.8% loss=0.103883 lr=9.329576e-05 lr_backbone=9.329576e-05 elapsed=10:25:41 avg_step=3.60s recent_step=3.21s eta=39:36:39 +2026-09-18 03:54:46 | INFO | train rank=0 | step=10430/50000 progress=20.9% loss=0.123688 lr=9.327921e-05 lr_backbone=9.327921e-05 elapsed=10:26:10 avg_step=3.60s recent_step=2.90s eta=39:35:36 +2026-09-18 03:55:36 | INFO | train rank=0 | step=10440/50000 progress=20.9% loss=0.111525 lr=9.326264e-05 lr_backbone=9.326264e-05 elapsed=10:26:59 avg_step=3.60s recent_step=4.95s eta=39:35:51 +2026-09-18 03:56:07 | INFO | train rank=0 | step=10450/50000 progress=20.9% loss=0.079508 lr=9.324605e-05 lr_backbone=9.324605e-05 elapsed=10:27:30 avg_step=3.60s recent_step=3.09s eta=39:34:55 +2026-09-18 03:56:40 | INFO | train rank=0 | step=10460/50000 progress=20.9% loss=0.078454 lr=9.322944e-05 lr_backbone=9.322944e-05 elapsed=10:28:04 avg_step=3.60s recent_step=3.35s eta=39:34:10 +2026-09-18 03:57:09 | INFO | train rank=0 | step=10470/50000 progress=20.9% loss=0.108720 lr=9.321282e-05 lr_backbone=9.321282e-05 elapsed=10:28:32 avg_step=3.60s recent_step=2.82s eta=39:33:04 +2026-09-18 03:58:02 | INFO | train rank=0 | step=10480/50000 progress=21.0% loss=0.104234 lr=9.319617e-05 lr_backbone=9.319617e-05 elapsed=10:29:25 avg_step=3.60s recent_step=5.37s eta=39:33:35 +2026-09-18 03:58:36 | INFO | train rank=0 | step=10490/50000 progress=21.0% loss=0.075894 lr=9.317951e-05 lr_backbone=9.317951e-05 elapsed=10:29:59 avg_step=3.60s recent_step=3.39s eta=39:32:51 +2026-09-18 03:59:10 | INFO | train rank=0 | step=10500/50000 progress=21.0% loss=0.144789 lr=9.316282e-05 lr_backbone=9.316282e-05 elapsed=10:30:34 avg_step=3.60s recent_step=3.42s eta=39:32:08 +2026-09-18 03:59:49 | INFO | train rank=0 | step=10510/50000 progress=21.0% loss=0.117871 lr=9.314612e-05 lr_backbone=9.314612e-05 elapsed=10:31:13 avg_step=3.60s recent_step=3.89s eta=39:31:43 +2026-09-18 04:00:41 | INFO | train rank=0 | step=10520/50000 progress=21.0% loss=0.094485 lr=9.312940e-05 lr_backbone=9.312940e-05 elapsed=10:32:04 avg_step=3.60s recent_step=5.15s eta=39:32:05 +2026-09-18 04:01:10 | INFO | train rank=0 | step=10530/50000 progress=21.1% loss=0.112901 lr=9.311266e-05 lr_backbone=9.311266e-05 elapsed=10:32:34 avg_step=3.60s recent_step=2.96s eta=39:31:04 +2026-09-18 04:01:40 | INFO | train rank=0 | step=10540/50000 progress=21.1% loss=0.098203 lr=9.309590e-05 lr_backbone=9.309590e-05 elapsed=10:33:04 avg_step=3.60s recent_step=2.99s eta=39:30:05 +2026-09-18 04:02:12 | INFO | train rank=0 | step=10550/50000 progress=21.1% loss=0.096292 lr=9.307913e-05 lr_backbone=9.307913e-05 elapsed=10:33:35 avg_step=3.60s recent_step=3.14s eta=39:29:12 +2026-09-18 04:03:03 | INFO | train rank=0 | step=10560/50000 progress=21.1% loss=0.100074 lr=9.306233e-05 lr_backbone=9.306233e-05 elapsed=10:34:26 avg_step=3.60s recent_step=5.13s eta=39:29:33 +2026-09-18 04:03:34 | INFO | train rank=0 | step=10570/50000 progress=21.1% loss=0.098868 lr=9.304552e-05 lr_backbone=9.304552e-05 elapsed=10:34:57 avg_step=3.60s recent_step=3.09s eta=39:28:38 +2026-09-18 04:04:08 | INFO | train rank=0 | step=10580/50000 progress=21.2% loss=0.102692 lr=9.302868e-05 lr_backbone=9.302868e-05 elapsed=10:35:32 avg_step=3.60s recent_step=3.45s eta=39:27:56 +2026-09-18 04:04:37 | INFO | train rank=0 | step=10590/50000 progress=21.2% loss=0.107755 lr=9.301183e-05 lr_backbone=9.301183e-05 elapsed=10:36:00 avg_step=3.60s recent_step=2.87s eta=39:26:53 +2026-09-18 04:05:34 | INFO | train rank=0 | step=10600/50000 progress=21.2% loss=0.107001 lr=9.299496e-05 lr_backbone=9.299496e-05 elapsed=10:36:57 avg_step=3.61s recent_step=5.65s eta=39:27:33 +2026-09-18 04:06:00 | INFO | train rank=0 | step=10610/50000 progress=21.2% loss=0.105248 lr=9.297807e-05 lr_backbone=9.297807e-05 elapsed=10:37:23 avg_step=3.60s recent_step=2.65s eta=39:26:21 +2026-09-18 04:06:32 | INFO | train rank=0 | step=10620/50000 progress=21.2% loss=0.097757 lr=9.296116e-05 lr_backbone=9.296116e-05 elapsed=10:37:55 avg_step=3.60s recent_step=3.21s eta=39:25:30 +2026-09-18 04:07:00 | INFO | train rank=0 | step=10630/50000 progress=21.3% loss=0.094352 lr=9.294423e-05 lr_backbone=9.294423e-05 elapsed=10:38:24 avg_step=3.60s recent_step=2.83s eta=39:24:26 +2026-09-18 04:07:57 | INFO | train rank=0 | step=10640/50000 progress=21.3% loss=0.114344 lr=9.292728e-05 lr_backbone=9.292728e-05 elapsed=10:39:20 avg_step=3.61s recent_step=5.68s eta=39:25:06 +2026-09-18 04:08:26 | INFO | train rank=0 | step=10650/50000 progress=21.3% loss=0.131538 lr=9.291032e-05 lr_backbone=9.291032e-05 elapsed=10:39:50 avg_step=3.60s recent_step=2.93s eta=39:24:05 +2026-09-18 04:08:53 | INFO | train rank=0 | step=10660/50000 progress=21.3% loss=0.101772 lr=9.289334e-05 lr_backbone=9.289334e-05 elapsed=10:40:16 avg_step=3.60s recent_step=2.68s eta=39:22:55 +2026-09-18 04:09:28 | INFO | train rank=0 | step=10670/50000 progress=21.3% loss=0.102020 lr=9.287633e-05 lr_backbone=9.287633e-05 elapsed=10:40:51 avg_step=3.60s recent_step=3.45s eta=39:22:13 +2026-09-18 04:10:21 | INFO | train rank=0 | step=10680/50000 progress=21.4% loss=0.118288 lr=9.285931e-05 lr_backbone=9.285931e-05 elapsed=10:41:44 avg_step=3.61s recent_step=5.28s eta=39:22:39 +2026-09-18 04:10:44 | INFO | train rank=0 | step=10690/50000 progress=21.4% loss=0.100358 lr=9.284227e-05 lr_backbone=9.284227e-05 elapsed=10:42:07 avg_step=3.60s recent_step=2.31s eta=39:21:15 +2026-09-18 04:11:12 | INFO | train rank=0 | step=10700/50000 progress=21.4% loss=0.132367 lr=9.282521e-05 lr_backbone=9.282521e-05 elapsed=10:42:35 avg_step=3.60s recent_step=2.82s eta=39:20:10 +2026-09-18 04:11:43 | INFO | train rank=0 | step=10710/50000 progress=21.4% loss=0.104356 lr=9.280813e-05 lr_backbone=9.280813e-05 elapsed=10:43:07 avg_step=3.60s recent_step=3.15s eta=39:19:18 +2026-09-18 04:12:35 | INFO | train rank=0 | step=10720/50000 progress=21.4% loss=0.081860 lr=9.279104e-05 lr_backbone=9.279104e-05 elapsed=10:43:59 avg_step=3.60s recent_step=5.21s eta=39:19:40 +2026-09-18 04:13:08 | INFO | train rank=0 | step=10730/50000 progress=21.5% loss=0.106129 lr=9.277392e-05 lr_backbone=9.277392e-05 elapsed=10:44:32 avg_step=3.60s recent_step=3.31s eta=39:18:54 +2026-09-18 04:13:43 | INFO | train rank=0 | step=10740/50000 progress=21.5% loss=0.098185 lr=9.275679e-05 lr_backbone=9.275679e-05 elapsed=10:45:06 avg_step=3.60s recent_step=3.41s eta=39:18:10 +2026-09-18 04:14:12 | INFO | train rank=0 | step=10750/50000 progress=21.5% loss=0.090252 lr=9.273964e-05 lr_backbone=9.273964e-05 elapsed=10:45:36 avg_step=3.60s recent_step=2.97s eta=39:17:11 +2026-09-18 04:15:04 | INFO | train rank=0 | step=10760/50000 progress=21.5% loss=0.097710 lr=9.272246e-05 lr_backbone=9.272246e-05 elapsed=10:46:27 avg_step=3.60s recent_step=5.18s eta=39:17:33 +2026-09-18 04:15:37 | INFO | train rank=0 | step=10770/50000 progress=21.5% loss=0.087383 lr=9.270527e-05 lr_backbone=9.270527e-05 elapsed=10:47:01 avg_step=3.60s recent_step=3.34s eta=39:16:47 +2026-09-18 04:16:10 | INFO | train rank=0 | step=10780/50000 progress=21.6% loss=0.122331 lr=9.268807e-05 lr_backbone=9.268807e-05 elapsed=10:47:33 avg_step=3.60s recent_step=3.24s eta=39:15:58 +2026-09-18 04:16:45 | INFO | train rank=0 | step=10790/50000 progress=21.6% loss=0.120341 lr=9.267084e-05 lr_backbone=9.267084e-05 elapsed=10:48:08 avg_step=3.60s recent_step=3.50s eta=39:15:18 +2026-09-18 04:17:39 | INFO | train rank=0 | step=10800/50000 progress=21.6% loss=0.078149 lr=9.265359e-05 lr_backbone=9.265359e-05 elapsed=10:49:03 avg_step=3.61s recent_step=5.43s eta=39:15:48 +2026-09-18 04:18:10 | INFO | train rank=0 | step=10810/50000 progress=21.6% loss=0.095990 lr=9.263633e-05 lr_backbone=9.263633e-05 elapsed=10:49:33 avg_step=3.61s recent_step=3.07s eta=39:14:53 +2026-09-18 04:18:41 | INFO | train rank=0 | step=10820/50000 progress=21.6% loss=0.095722 lr=9.261904e-05 lr_backbone=9.261904e-05 elapsed=10:50:04 avg_step=3.60s recent_step=3.10s eta=39:13:59 +2026-09-18 04:19:14 | INFO | train rank=0 | step=10830/50000 progress=21.7% loss=0.132015 lr=9.260174e-05 lr_backbone=9.260174e-05 elapsed=10:50:37 avg_step=3.60s recent_step=3.28s eta=39:13:11 +2026-09-18 04:20:02 | INFO | train rank=0 | step=10840/50000 progress=21.7% loss=0.104557 lr=9.258442e-05 lr_backbone=9.258442e-05 elapsed=10:51:25 avg_step=3.61s recent_step=4.78s eta=39:13:17 +2026-09-18 04:20:28 | INFO | train rank=0 | step=10850/50000 progress=21.7% loss=0.084320 lr=9.256708e-05 lr_backbone=9.256708e-05 elapsed=10:51:51 avg_step=3.60s recent_step=2.63s eta=39:12:06 +2026-09-18 04:21:04 | INFO | train rank=0 | step=10860/50000 progress=21.7% loss=0.106617 lr=9.254972e-05 lr_backbone=9.254972e-05 elapsed=10:52:27 avg_step=3.60s recent_step=3.59s eta=39:11:29 +2026-09-18 04:21:35 | INFO | train rank=0 | step=10870/50000 progress=21.7% loss=0.085516 lr=9.253235e-05 lr_backbone=9.253235e-05 elapsed=10:52:58 avg_step=3.60s recent_step=3.11s eta=39:10:35 +2026-09-18 04:22:30 | INFO | train rank=0 | step=10880/50000 progress=21.8% loss=0.093530 lr=9.251495e-05 lr_backbone=9.251495e-05 elapsed=10:53:54 avg_step=3.61s recent_step=5.54s eta=39:11:09 +2026-09-18 04:22:59 | INFO | train rank=0 | step=10890/50000 progress=21.8% loss=0.093426 lr=9.249754e-05 lr_backbone=9.249754e-05 elapsed=10:54:22 avg_step=3.61s recent_step=2.83s eta=39:10:05 +2026-09-18 04:23:31 | INFO | train rank=0 | step=10900/50000 progress=21.8% loss=0.104528 lr=9.248011e-05 lr_backbone=9.248011e-05 elapsed=10:54:54 avg_step=3.60s recent_step=3.20s eta=39:09:14 +2026-09-18 04:24:06 | INFO | train rank=0 | step=10910/50000 progress=21.8% loss=0.138794 lr=9.246266e-05 lr_backbone=9.246266e-05 elapsed=10:55:29 avg_step=3.60s recent_step=3.56s eta=39:08:37 +2026-09-18 04:24:57 | INFO | train rank=0 | step=10920/50000 progress=21.8% loss=0.114589 lr=9.244519e-05 lr_backbone=9.244519e-05 elapsed=10:56:20 avg_step=3.61s recent_step=5.08s eta=39:08:54 +2026-09-18 04:25:26 | INFO | train rank=0 | step=10930/50000 progress=21.9% loss=0.087303 lr=9.242770e-05 lr_backbone=9.242770e-05 elapsed=10:56:49 avg_step=3.61s recent_step=2.86s eta=39:07:51 +2026-09-18 04:26:01 | INFO | train rank=0 | step=10940/50000 progress=21.9% loss=0.087940 lr=9.241019e-05 lr_backbone=9.241019e-05 elapsed=10:57:24 avg_step=3.61s recent_step=3.53s eta=39:07:12 +2026-09-18 04:26:32 | INFO | train rank=0 | step=10950/50000 progress=21.9% loss=0.111054 lr=9.239267e-05 lr_backbone=9.239267e-05 elapsed=10:57:55 avg_step=3.61s recent_step=3.10s eta=39:06:18 +2026-09-18 04:27:22 | INFO | train rank=0 | step=10960/50000 progress=21.9% loss=0.149976 lr=9.237512e-05 lr_backbone=9.237512e-05 elapsed=10:58:45 avg_step=3.61s recent_step=5.03s eta=39:06:32 +2026-09-18 04:27:52 | INFO | train rank=0 | step=10970/50000 progress=21.9% loss=0.112479 lr=9.235756e-05 lr_backbone=9.235756e-05 elapsed=10:59:15 avg_step=3.61s recent_step=2.96s eta=39:05:34 +2026-09-18 04:28:24 | INFO | train rank=0 | step=10980/50000 progress=22.0% loss=0.087963 lr=9.233998e-05 lr_backbone=9.233998e-05 elapsed=10:59:47 avg_step=3.61s recent_step=3.23s eta=39:04:44 +2026-09-18 04:29:00 | INFO | train rank=0 | step=10990/50000 progress=22.0% loss=0.093573 lr=9.232238e-05 lr_backbone=9.232238e-05 elapsed=11:00:23 avg_step=3.61s recent_step=3.62s eta=39:04:08 +2026-09-18 04:29:45 | INFO | train rank=0 | step=11000/50000 progress=22.0% loss=0.103023 lr=9.230476e-05 lr_backbone=9.230476e-05 elapsed=11:01:08 avg_step=3.61s recent_step=4.45s eta=39:04:02 +2026-09-18 04:30:15 | INFO | train rank=0 | step=11010/50000 progress=22.0% loss=0.092973 lr=9.228713e-05 lr_backbone=9.228713e-05 elapsed=11:01:38 avg_step=3.61s recent_step=3.02s eta=39:03:06 +2026-09-18 04:30:47 | INFO | train rank=0 | step=11020/50000 progress=22.0% loss=0.135324 lr=9.226947e-05 lr_backbone=9.226947e-05 elapsed=11:02:10 avg_step=3.61s recent_step=3.20s eta=39:02:15 +2026-09-18 04:31:22 | INFO | train rank=0 | step=11030/50000 progress=22.1% loss=0.107516 lr=9.225180e-05 lr_backbone=9.225180e-05 elapsed=11:02:45 avg_step=3.61s recent_step=3.47s eta=39:01:34 +2026-09-18 04:32:10 | INFO | train rank=0 | step=11040/50000 progress=22.1% loss=0.089961 lr=9.223411e-05 lr_backbone=9.223411e-05 elapsed=11:03:33 avg_step=3.61s recent_step=4.86s eta=39:01:42 +2026-09-18 04:32:43 | INFO | train rank=0 | step=11050/50000 progress=22.1% loss=0.099196 lr=9.221640e-05 lr_backbone=9.221640e-05 elapsed=11:04:06 avg_step=3.61s recent_step=3.28s eta=39:00:55 +2026-09-18 04:33:17 | INFO | train rank=0 | step=11060/50000 progress=22.1% loss=0.104168 lr=9.219867e-05 lr_backbone=9.219867e-05 elapsed=11:04:40 avg_step=3.61s recent_step=3.37s eta=39:00:10 +2026-09-18 04:33:42 | INFO | train rank=0 | step=11070/50000 progress=22.1% loss=0.104321 lr=9.218092e-05 lr_backbone=9.218092e-05 elapsed=11:05:06 avg_step=3.60s recent_step=2.58s eta=38:58:58 +2026-09-18 04:34:35 | INFO | train rank=0 | step=11080/50000 progress=22.2% loss=0.110330 lr=9.216315e-05 lr_backbone=9.216315e-05 elapsed=11:05:58 avg_step=3.61s recent_step=5.27s eta=38:59:21 +2026-09-18 04:35:07 | INFO | train rank=0 | step=11090/50000 progress=22.2% loss=0.106856 lr=9.214537e-05 lr_backbone=9.214537e-05 elapsed=11:06:30 avg_step=3.61s recent_step=3.14s eta=38:58:28 +2026-09-18 04:35:44 | INFO | train rank=0 | step=11100/50000 progress=22.2% loss=0.114932 lr=9.212757e-05 lr_backbone=9.212757e-05 elapsed=11:07:08 avg_step=3.61s recent_step=3.79s eta=38:57:59 +2026-09-18 04:36:13 | INFO | train rank=0 | step=11110/50000 progress=22.2% loss=0.126488 lr=9.210975e-05 lr_backbone=9.210975e-05 elapsed=11:07:36 avg_step=3.61s recent_step=2.84s eta=38:56:56 +2026-09-18 04:36:59 | INFO | train rank=0 | step=11120/50000 progress=22.2% loss=0.156193 lr=9.209191e-05 lr_backbone=9.209191e-05 elapsed=11:08:22 avg_step=3.61s recent_step=4.63s eta=38:56:56 +2026-09-18 04:37:32 | INFO | train rank=0 | step=11130/50000 progress=22.3% loss=0.097335 lr=9.207405e-05 lr_backbone=9.207405e-05 elapsed=11:08:55 avg_step=3.61s recent_step=3.30s eta=38:56:09 +2026-09-18 04:37:55 | INFO | train rank=0 | step=11140/50000 progress=22.3% loss=0.096055 lr=9.205617e-05 lr_backbone=9.205617e-05 elapsed=11:09:19 avg_step=3.60s recent_step=2.30s eta=38:54:47 +2026-09-18 04:38:22 | INFO | train rank=0 | step=11150/50000 progress=22.3% loss=0.092817 lr=9.203828e-05 lr_backbone=9.203828e-05 elapsed=11:09:45 avg_step=3.60s recent_step=2.68s eta=38:53:39 +2026-09-18 04:39:13 | INFO | train rank=0 | step=11160/50000 progress=22.3% loss=0.083618 lr=9.202037e-05 lr_backbone=9.202037e-05 elapsed=11:10:36 avg_step=3.61s recent_step=5.08s eta=38:53:55 +2026-09-18 04:39:43 | INFO | train rank=0 | step=11170/50000 progress=22.3% loss=0.136303 lr=9.200244e-05 lr_backbone=9.200244e-05 elapsed=11:11:06 avg_step=3.60s recent_step=2.99s eta=38:52:57 +2026-09-18 04:40:14 | INFO | train rank=0 | step=11180/50000 progress=22.4% loss=0.092752 lr=9.198449e-05 lr_backbone=9.198449e-05 elapsed=11:11:37 avg_step=3.60s recent_step=3.08s eta=38:52:03 +2026-09-18 04:40:38 | INFO | train rank=0 | step=11190/50000 progress=22.4% loss=0.106719 lr=9.196652e-05 lr_backbone=9.196652e-05 elapsed=11:12:02 avg_step=3.60s recent_step=2.46s eta=38:50:47 +2026-09-18 04:41:21 | INFO | train rank=0 | step=11200/50000 progress=22.4% loss=0.112509 lr=9.194853e-05 lr_backbone=9.194853e-05 elapsed=11:12:44 avg_step=3.60s recent_step=4.26s eta=38:50:34 +2026-09-18 04:41:46 | INFO | train rank=0 | step=11210/50000 progress=22.4% loss=0.102419 lr=9.193053e-05 lr_backbone=9.193053e-05 elapsed=11:13:09 avg_step=3.60s recent_step=2.51s eta=38:49:20 +2026-09-18 04:42:34 | INFO | train rank=0 | step=11220/50000 progress=22.4% loss=0.102375 lr=9.191250e-05 lr_backbone=9.191250e-05 elapsed=11:13:57 avg_step=3.60s recent_step=4.80s eta=38:49:26 +2026-09-18 04:43:03 | INFO | train rank=0 | step=11230/50000 progress=22.5% loss=0.122706 lr=9.189446e-05 lr_backbone=9.189446e-05 elapsed=11:14:27 avg_step=3.60s recent_step=2.94s eta=38:48:27 +2026-09-18 04:43:40 | INFO | train rank=0 | step=11240/50000 progress=22.5% loss=0.135238 lr=9.187640e-05 lr_backbone=9.187640e-05 elapsed=11:15:03 avg_step=3.60s recent_step=3.66s eta=38:47:53 +2026-09-18 04:44:12 | INFO | train rank=0 | step=11250/50000 progress=22.5% loss=0.086041 lr=9.185832e-05 lr_backbone=9.185832e-05 elapsed=11:15:35 avg_step=3.60s recent_step=3.15s eta=38:47:01 +2026-09-18 04:44:52 | INFO | train rank=0 | step=11260/50000 progress=22.5% loss=0.104878 lr=9.184023e-05 lr_backbone=9.184023e-05 elapsed=11:16:15 avg_step=3.60s recent_step=4.06s eta=38:46:41 +2026-09-18 04:45:27 | INFO | train rank=0 | step=11270/50000 progress=22.5% loss=0.093856 lr=9.182211e-05 lr_backbone=9.182211e-05 elapsed=11:16:50 avg_step=3.60s recent_step=3.47s eta=38:46:00 +2026-09-18 04:46:23 | INFO | train rank=0 | step=11280/50000 progress=22.6% loss=0.105471 lr=9.180398e-05 lr_backbone=9.180398e-05 elapsed=11:17:47 avg_step=3.61s recent_step=5.65s eta=38:46:35 +2026-09-18 04:46:55 | INFO | train rank=0 | step=11290/50000 progress=22.6% loss=0.087325 lr=9.178583e-05 lr_backbone=9.178583e-05 elapsed=11:18:18 avg_step=3.60s recent_step=3.13s eta=38:45:42 +2026-09-18 04:47:22 | INFO | train rank=0 | step=11300/50000 progress=22.6% loss=0.074036 lr=9.176766e-05 lr_backbone=9.176766e-05 elapsed=11:18:45 avg_step=3.60s recent_step=2.70s eta=38:44:35 +2026-09-18 04:47:56 | INFO | train rank=0 | step=11310/50000 progress=22.6% loss=0.100745 lr=9.174947e-05 lr_backbone=9.174947e-05 elapsed=11:19:19 avg_step=3.60s recent_step=3.40s eta=38:43:52 +2026-09-18 04:48:44 | INFO | train rank=0 | step=11320/50000 progress=22.6% loss=0.105887 lr=9.173127e-05 lr_backbone=9.173127e-05 elapsed=11:20:07 avg_step=3.60s recent_step=4.84s eta=38:43:58 +2026-09-18 04:49:21 | INFO | train rank=0 | step=11330/50000 progress=22.7% loss=0.119426 lr=9.171304e-05 lr_backbone=9.171304e-05 elapsed=11:20:44 avg_step=3.60s recent_step=3.65s eta=38:43:24 +2026-09-18 04:49:52 | INFO | train rank=0 | step=11340/50000 progress=22.7% loss=0.104398 lr=9.169480e-05 lr_backbone=9.169480e-05 elapsed=11:21:15 avg_step=3.60s recent_step=3.16s eta=38:42:33 +2026-09-18 04:50:29 | INFO | train rank=0 | step=11350/50000 progress=22.7% loss=0.151463 lr=9.167654e-05 lr_backbone=9.167654e-05 elapsed=11:21:52 avg_step=3.60s recent_step=3.67s eta=38:41:59 +2026-09-18 04:51:26 | INFO | train rank=0 | step=11360/50000 progress=22.7% loss=0.157644 lr=9.165826e-05 lr_backbone=9.165826e-05 elapsed=11:22:49 avg_step=3.61s recent_step=5.73s eta=38:42:35 +2026-09-18 04:51:55 | INFO | train rank=0 | step=11370/50000 progress=22.7% loss=0.086063 lr=9.163996e-05 lr_backbone=9.163996e-05 elapsed=11:23:18 avg_step=3.61s recent_step=2.86s eta=38:41:33 +2026-09-18 04:52:25 | INFO | train rank=0 | step=11380/50000 progress=22.8% loss=0.114189 lr=9.162165e-05 lr_backbone=9.162165e-05 elapsed=11:23:49 avg_step=3.61s recent_step=3.07s eta=38:40:39 +2026-09-18 04:53:01 | INFO | train rank=0 | step=11390/50000 progress=22.8% loss=0.125061 lr=9.160331e-05 lr_backbone=9.160331e-05 elapsed=11:24:24 avg_step=3.61s recent_step=3.52s eta=38:40:00 +2026-09-18 04:53:48 | INFO | train rank=0 | step=11400/50000 progress=22.8% loss=0.091352 lr=9.158496e-05 lr_backbone=9.158496e-05 elapsed=11:25:11 avg_step=3.61s recent_step=4.72s eta=38:40:02 +2026-09-18 04:54:20 | INFO | train rank=0 | step=11410/50000 progress=22.8% loss=0.183755 lr=9.156659e-05 lr_backbone=9.156659e-05 elapsed=11:25:43 avg_step=3.61s recent_step=3.19s eta=38:39:11 +2026-09-18 04:54:51 | INFO | train rank=0 | step=11420/50000 progress=22.8% loss=0.133803 lr=9.154820e-05 lr_backbone=9.154820e-05 elapsed=11:26:14 avg_step=3.61s recent_step=3.14s eta=38:38:20 +2026-09-18 04:55:27 | INFO | train rank=0 | step=11430/50000 progress=22.9% loss=0.122159 lr=9.152979e-05 lr_backbone=9.152979e-05 elapsed=11:26:50 avg_step=3.61s recent_step=3.56s eta=38:37:42 +2026-09-18 04:56:18 | INFO | train rank=0 | step=11440/50000 progress=22.9% loss=0.104815 lr=9.151137e-05 lr_backbone=9.151137e-05 elapsed=11:27:41 avg_step=3.61s recent_step=5.10s eta=38:37:57 +2026-09-18 04:56:44 | INFO | train rank=0 | step=11450/50000 progress=22.9% loss=0.099639 lr=9.149293e-05 lr_backbone=9.149293e-05 elapsed=11:28:07 avg_step=3.61s recent_step=2.63s eta=38:36:48 +2026-09-18 04:57:20 | INFO | train rank=0 | step=11460/50000 progress=22.9% loss=0.142687 lr=9.147447e-05 lr_backbone=9.147447e-05 elapsed=11:28:44 avg_step=3.61s recent_step=3.64s eta=38:36:13 +2026-09-18 04:57:48 | INFO | train rank=0 | step=11470/50000 progress=22.9% loss=0.111426 lr=9.145599e-05 lr_backbone=9.145599e-05 elapsed=11:29:12 avg_step=3.61s recent_step=2.79s eta=38:35:09 +2026-09-18 04:58:23 | INFO | train rank=0 | step=11480/50000 progress=23.0% loss=0.121644 lr=9.143749e-05 lr_backbone=9.143749e-05 elapsed=11:29:47 avg_step=3.61s recent_step=3.52s eta=38:34:30 +2026-09-18 04:58:57 | INFO | train rank=0 | step=11490/50000 progress=23.0% loss=0.103205 lr=9.141898e-05 lr_backbone=9.141898e-05 elapsed=11:30:21 avg_step=3.60s recent_step=3.39s eta=38:33:47 +2026-09-18 04:59:33 | INFO | train rank=0 | step=11500/50000 progress=23.0% loss=0.116173 lr=9.140044e-05 lr_backbone=9.140044e-05 elapsed=11:30:57 avg_step=3.60s recent_step=3.60s eta=38:33:11 +2026-09-18 05:00:16 | INFO | train rank=0 | step=11510/50000 progress=23.0% loss=0.093690 lr=9.138189e-05 lr_backbone=9.138189e-05 elapsed=11:31:40 avg_step=3.61s recent_step=4.28s eta=38:32:58 +2026-09-18 05:00:47 | INFO | train rank=0 | step=11520/50000 progress=23.0% loss=0.102363 lr=9.136332e-05 lr_backbone=9.136332e-05 elapsed=11:32:10 avg_step=3.61s recent_step=3.06s eta=38:32:03 +2026-09-18 05:01:21 | INFO | train rank=0 | step=11530/50000 progress=23.1% loss=0.110859 lr=9.134473e-05 lr_backbone=9.134473e-05 elapsed=11:32:44 avg_step=3.60s recent_step=3.39s eta=38:31:20 +2026-09-18 05:01:59 | INFO | train rank=0 | step=11540/50000 progress=23.1% loss=0.088920 lr=9.132613e-05 lr_backbone=9.132613e-05 elapsed=11:33:22 avg_step=3.61s recent_step=3.80s eta=38:30:51 +2026-09-18 05:02:52 | INFO | train rank=0 | step=11550/50000 progress=23.1% loss=0.096390 lr=9.130750e-05 lr_backbone=9.130750e-05 elapsed=11:34:15 avg_step=3.61s recent_step=5.32s eta=38:31:12 +2026-09-18 05:03:23 | INFO | train rank=0 | step=11560/50000 progress=23.1% loss=0.094910 lr=9.128886e-05 lr_backbone=9.128886e-05 elapsed=11:34:46 avg_step=3.61s recent_step=3.07s eta=38:30:18 +2026-09-18 05:04:02 | INFO | train rank=0 | step=11570/50000 progress=23.1% loss=0.098569 lr=9.127020e-05 lr_backbone=9.127020e-05 elapsed=11:35:25 avg_step=3.61s recent_step=3.90s eta=38:29:52 +2026-09-18 05:04:34 | INFO | train rank=0 | step=11580/50000 progress=23.2% loss=0.115698 lr=9.125152e-05 lr_backbone=9.125152e-05 elapsed=11:35:57 avg_step=3.61s recent_step=3.25s eta=38:29:04 +2026-09-18 05:05:29 | INFO | train rank=0 | step=11590/50000 progress=23.2% loss=0.095391 lr=9.123283e-05 lr_backbone=9.123283e-05 elapsed=11:36:52 avg_step=3.61s recent_step=5.47s eta=38:29:29 +2026-09-18 05:05:52 | INFO | train rank=0 | step=11600/50000 progress=23.2% loss=0.088443 lr=9.121411e-05 lr_backbone=9.121411e-05 elapsed=11:37:15 avg_step=3.61s recent_step=2.26s eta=38:28:09 +2026-09-18 05:06:20 | INFO | train rank=0 | step=11610/50000 progress=23.2% loss=0.116761 lr=9.119538e-05 lr_backbone=9.119538e-05 elapsed=11:37:44 avg_step=3.61s recent_step=2.89s eta=38:27:09 +2026-09-18 05:06:53 | INFO | train rank=0 | step=11620/50000 progress=23.2% loss=0.104802 lr=9.117663e-05 lr_backbone=9.117663e-05 elapsed=11:38:17 avg_step=3.61s recent_step=3.29s eta=38:26:23 +2026-09-18 05:07:44 | INFO | train rank=0 | step=11630/50000 progress=23.3% loss=0.082520 lr=9.115786e-05 lr_backbone=9.115786e-05 elapsed=11:39:07 avg_step=3.61s recent_step=5.09s eta=38:26:35 +2026-09-18 05:08:15 | INFO | train rank=0 | step=11640/50000 progress=23.3% loss=0.083142 lr=9.113908e-05 lr_backbone=9.113908e-05 elapsed=11:39:38 avg_step=3.61s recent_step=3.07s eta=38:25:42 +2026-09-18 05:08:50 | INFO | train rank=0 | step=11650/50000 progress=23.3% loss=0.123545 lr=9.112027e-05 lr_backbone=9.112027e-05 elapsed=11:40:13 avg_step=3.61s recent_step=3.49s eta=38:25:02 +2026-09-18 05:09:23 | INFO | train rank=0 | step=11660/50000 progress=23.3% loss=0.094709 lr=9.110145e-05 lr_backbone=9.110145e-05 elapsed=11:40:47 avg_step=3.61s recent_step=3.36s eta=38:24:18 +2026-09-18 05:10:14 | INFO | train rank=0 | step=11670/50000 progress=23.3% loss=0.187129 lr=9.108261e-05 lr_backbone=9.108261e-05 elapsed=11:41:37 avg_step=3.61s recent_step=5.03s eta=38:24:28 +2026-09-18 05:10:41 | INFO | train rank=0 | step=11680/50000 progress=23.4% loss=0.106107 lr=9.106375e-05 lr_backbone=9.106375e-05 elapsed=11:42:05 avg_step=3.61s recent_step=2.77s eta=38:23:25 +2026-09-18 05:11:17 | INFO | train rank=0 | step=11690/50000 progress=23.4% loss=0.098418 lr=9.104487e-05 lr_backbone=9.104487e-05 elapsed=11:42:40 avg_step=3.61s recent_step=3.52s eta=38:22:46 +2026-09-18 05:11:55 | INFO | train rank=0 | step=11700/50000 progress=23.4% loss=0.098803 lr=9.102598e-05 lr_backbone=9.102598e-05 elapsed=11:43:19 avg_step=3.61s recent_step=3.86s eta=38:22:18 +2026-09-18 05:12:49 | INFO | train rank=0 | step=11710/50000 progress=23.4% loss=0.085260 lr=9.100707e-05 lr_backbone=9.100707e-05 elapsed=11:44:12 avg_step=3.61s recent_step=5.39s eta=38:22:40 +2026-09-18 05:13:23 | INFO | train rank=0 | step=11720/50000 progress=23.4% loss=0.106289 lr=9.098814e-05 lr_backbone=9.098814e-05 elapsed=11:44:47 avg_step=3.61s recent_step=3.42s eta=38:21:58 +2026-09-18 05:13:48 | INFO | train rank=0 | step=11730/50000 progress=23.5% loss=0.135082 lr=9.096919e-05 lr_backbone=9.096919e-05 elapsed=11:45:11 avg_step=3.61s recent_step=2.47s eta=38:20:45 +2026-09-18 05:14:20 | INFO | train rank=0 | step=11740/50000 progress=23.5% loss=0.160071 lr=9.095022e-05 lr_backbone=9.095022e-05 elapsed=11:45:43 avg_step=3.61s recent_step=3.20s eta=38:19:56 +2026-09-18 05:15:15 | INFO | train rank=0 | step=11750/50000 progress=23.5% loss=0.119268 lr=9.093124e-05 lr_backbone=9.093124e-05 elapsed=11:46:38 avg_step=3.61s recent_step=5.45s eta=38:20:20 +2026-09-18 05:15:49 | INFO | train rank=0 | step=11760/50000 progress=23.5% loss=0.127323 lr=9.091224e-05 lr_backbone=9.091224e-05 elapsed=11:47:12 avg_step=3.61s recent_step=3.46s eta=38:19:39 +2026-09-18 05:16:21 | INFO | train rank=0 | step=11770/50000 progress=23.5% loss=0.102535 lr=9.089322e-05 lr_backbone=9.089322e-05 elapsed=11:47:45 avg_step=3.61s recent_step=3.23s eta=38:18:51 +2026-09-18 05:16:47 | INFO | train rank=0 | step=11780/50000 progress=23.6% loss=0.122502 lr=9.087418e-05 lr_backbone=9.087418e-05 elapsed=11:48:11 avg_step=3.61s recent_step=2.59s eta=38:17:41 +2026-09-18 05:17:42 | INFO | train rank=0 | step=11790/50000 progress=23.6% loss=0.101410 lr=9.085513e-05 lr_backbone=9.085513e-05 elapsed=11:49:05 avg_step=3.61s recent_step=5.46s eta=38:18:05 +2026-09-18 05:18:12 | INFO | train rank=0 | step=11800/50000 progress=23.6% loss=0.136538 lr=9.083605e-05 lr_backbone=9.083605e-05 elapsed=11:49:36 avg_step=3.61s recent_step=3.04s eta=38:17:11 +2026-09-18 05:18:38 | INFO | train rank=0 | step=11810/50000 progress=23.6% loss=0.102775 lr=9.081696e-05 lr_backbone=9.081696e-05 elapsed=11:50:02 avg_step=3.61s recent_step=2.59s eta=38:16:02 +2026-09-18 05:19:06 | INFO | train rank=0 | step=11820/50000 progress=23.6% loss=0.081200 lr=9.079785e-05 lr_backbone=9.079785e-05 elapsed=11:50:29 avg_step=3.61s recent_step=2.77s eta=38:14:59 +2026-09-18 05:19:59 | INFO | train rank=0 | step=11830/50000 progress=23.7% loss=0.113561 lr=9.077873e-05 lr_backbone=9.077873e-05 elapsed=11:51:22 avg_step=3.61s recent_step=5.29s eta=38:15:17 +2026-09-18 05:20:34 | INFO | train rank=0 | step=11840/50000 progress=23.7% loss=0.144486 lr=9.075958e-05 lr_backbone=9.075958e-05 elapsed=11:51:57 avg_step=3.61s recent_step=3.46s eta=38:14:36 +2026-09-18 05:20:58 | INFO | train rank=0 | step=11850/50000 progress=23.7% loss=0.100334 lr=9.074042e-05 lr_backbone=9.074042e-05 elapsed=11:52:22 avg_step=3.61s recent_step=2.47s eta=38:13:24 +2026-09-18 05:21:32 | INFO | train rank=0 | step=11860/50000 progress=23.7% loss=0.102504 lr=9.072124e-05 lr_backbone=9.072124e-05 elapsed=11:52:55 avg_step=3.61s recent_step=3.39s eta=38:12:41 +2026-09-18 05:22:24 | INFO | train rank=0 | step=11870/50000 progress=23.7% loss=0.112989 lr=9.070204e-05 lr_backbone=9.070204e-05 elapsed=11:53:48 avg_step=3.61s recent_step=5.21s eta=38:12:56 +2026-09-18 05:22:59 | INFO | train rank=0 | step=11880/50000 progress=23.8% loss=0.104114 lr=9.068283e-05 lr_backbone=9.068283e-05 elapsed=11:54:22 avg_step=3.61s recent_step=3.48s eta=38:12:16 +2026-09-18 05:23:36 | INFO | train rank=0 | step=11890/50000 progress=23.8% loss=0.110746 lr=9.066359e-05 lr_backbone=9.066359e-05 elapsed=11:55:00 avg_step=3.61s recent_step=3.72s eta=38:11:43 +2026-09-18 05:24:05 | INFO | train rank=0 | step=11900/50000 progress=23.8% loss=0.145837 lr=9.064434e-05 lr_backbone=9.064434e-05 elapsed=11:55:29 avg_step=3.61s recent_step=2.93s eta=38:10:45 +2026-09-18 05:24:53 | INFO | train rank=0 | step=11910/50000 progress=23.8% loss=0.078312 lr=9.062507e-05 lr_backbone=9.062507e-05 elapsed=11:56:17 avg_step=3.61s recent_step=4.79s eta=38:10:47 +2026-09-18 05:25:26 | INFO | train rank=0 | step=11920/50000 progress=23.8% loss=0.119023 lr=9.060578e-05 lr_backbone=9.060578e-05 elapsed=11:56:50 avg_step=3.61s recent_step=3.29s eta=38:10:01 +2026-09-18 05:25:54 | INFO | train rank=0 | step=11930/50000 progress=23.9% loss=0.090144 lr=9.058648e-05 lr_backbone=9.058648e-05 elapsed=11:57:17 avg_step=3.61s recent_step=2.78s eta=38:08:58 +2026-09-18 05:26:16 | INFO | train rank=0 | step=11940/50000 progress=23.9% loss=0.108967 lr=9.056716e-05 lr_backbone=9.056716e-05 elapsed=11:57:39 avg_step=3.61s recent_step=2.19s eta=38:07:37 +2026-09-18 05:27:04 | INFO | train rank=0 | step=11950/50000 progress=23.9% loss=0.074719 lr=9.054782e-05 lr_backbone=9.054782e-05 elapsed=11:58:27 avg_step=3.61s recent_step=4.79s eta=38:07:39 +2026-09-18 05:27:36 | INFO | train rank=0 | step=11960/50000 progress=23.9% loss=0.104890 lr=9.052846e-05 lr_backbone=9.052846e-05 elapsed=11:58:59 avg_step=3.61s recent_step=3.16s eta=38:06:49 +2026-09-18 05:28:09 | INFO | train rank=0 | step=11970/50000 progress=23.9% loss=0.098462 lr=9.050908e-05 lr_backbone=9.050908e-05 elapsed=11:59:32 avg_step=3.61s recent_step=3.34s eta=38:06:04 +2026-09-18 05:28:38 | INFO | train rank=0 | step=11980/50000 progress=24.0% loss=0.125739 lr=9.048969e-05 lr_backbone=9.048969e-05 elapsed=12:00:01 avg_step=3.61s recent_step=2.87s eta=38:05:05 +2026-09-18 05:29:32 | INFO | train rank=0 | step=11990/50000 progress=24.0% loss=0.098808 lr=9.047028e-05 lr_backbone=9.047028e-05 elapsed=12:00:55 avg_step=3.61s recent_step=5.39s eta=38:05:25 +2026-09-18 05:30:05 | INFO | train rank=0 | step=12000/50000 progress=24.0% loss=0.091407 lr=9.045085e-05 lr_backbone=9.045085e-05 elapsed=12:01:28 avg_step=3.61s recent_step=3.31s eta=38:04:40 +2026-09-18 05:30:36 | INFO | train rank=0 | step=12010/50000 progress=24.0% loss=0.097415 lr=9.043140e-05 lr_backbone=9.043140e-05 elapsed=12:01:59 avg_step=3.61s recent_step=3.14s eta=38:03:49 +2026-09-18 05:31:06 | INFO | train rank=0 | step=12020/50000 progress=24.0% loss=0.090976 lr=9.041194e-05 lr_backbone=9.041194e-05 elapsed=12:02:30 avg_step=3.61s recent_step=3.04s eta=38:02:55 +2026-09-18 05:31:57 | INFO | train rank=0 | step=12030/50000 progress=24.1% loss=0.160239 lr=9.039246e-05 lr_backbone=9.039246e-05 elapsed=12:03:21 avg_step=3.61s recent_step=5.10s eta=38:03:06 +2026-09-18 05:32:28 | INFO | train rank=0 | step=12040/50000 progress=24.1% loss=0.157028 lr=9.037296e-05 lr_backbone=9.037296e-05 elapsed=12:03:52 avg_step=3.61s recent_step=3.10s eta=38:02:14 +2026-09-18 05:33:03 | INFO | train rank=0 | step=12050/50000 progress=24.1% loss=0.102592 lr=9.035344e-05 lr_backbone=9.035344e-05 elapsed=12:04:26 avg_step=3.61s recent_step=3.45s eta=38:01:33 +2026-09-18 05:33:41 | INFO | train rank=0 | step=12060/50000 progress=24.1% loss=0.111679 lr=9.033391e-05 lr_backbone=9.033391e-05 elapsed=12:05:04 avg_step=3.61s recent_step=3.82s eta=38:01:03 +2026-09-18 05:34:25 | INFO | train rank=0 | step=12070/50000 progress=24.1% loss=0.090746 lr=9.031435e-05 lr_backbone=9.031435e-05 elapsed=12:05:48 avg_step=3.61s recent_step=4.37s eta=38:00:51 +2026-09-18 05:34:59 | INFO | train rank=0 | step=12080/50000 progress=24.2% loss=0.135918 lr=9.029478e-05 lr_backbone=9.029478e-05 elapsed=12:06:22 avg_step=3.61s recent_step=3.38s eta=38:00:08 +2026-09-18 05:35:38 | INFO | train rank=0 | step=12090/50000 progress=24.2% loss=0.094627 lr=9.027519e-05 lr_backbone=9.027519e-05 elapsed=12:07:01 avg_step=3.61s recent_step=3.92s eta=37:59:42 +2026-09-18 05:36:07 | INFO | train rank=0 | step=12100/50000 progress=24.2% loss=0.104874 lr=9.025559e-05 lr_backbone=9.025559e-05 elapsed=12:07:30 avg_step=3.61s recent_step=2.89s eta=37:58:43 +2026-09-18 05:37:06 | INFO | train rank=0 | step=12110/50000 progress=24.2% loss=0.119935 lr=9.023597e-05 lr_backbone=9.023597e-05 elapsed=12:08:29 avg_step=3.61s recent_step=5.90s eta=37:59:19 +2026-09-18 05:37:38 | INFO | train rank=0 | step=12120/50000 progress=24.2% loss=0.107774 lr=9.021633e-05 lr_backbone=9.021633e-05 elapsed=12:09:01 avg_step=3.61s recent_step=3.19s eta=37:58:30 +2026-09-18 05:38:06 | INFO | train rank=0 | step=12130/50000 progress=24.3% loss=0.090558 lr=9.019667e-05 lr_backbone=9.019667e-05 elapsed=12:09:29 avg_step=3.61s recent_step=2.82s eta=37:57:29 +2026-09-18 05:38:41 | INFO | train rank=0 | step=12140/50000 progress=24.3% loss=0.115177 lr=9.017699e-05 lr_backbone=9.017699e-05 elapsed=12:10:04 avg_step=3.61s recent_step=3.45s eta=37:56:48 +2026-09-18 05:39:30 | INFO | train rank=0 | step=12150/50000 progress=24.3% loss=0.081596 lr=9.015730e-05 lr_backbone=9.015730e-05 elapsed=12:10:53 avg_step=3.61s recent_step=4.96s eta=37:56:55 +2026-09-18 05:40:05 | INFO | train rank=0 | step=12160/50000 progress=24.3% loss=0.106032 lr=9.013759e-05 lr_backbone=9.013759e-05 elapsed=12:11:28 avg_step=3.61s recent_step=3.46s eta=37:56:14 +2026-09-18 05:40:32 | INFO | train rank=0 | step=12170/50000 progress=24.3% loss=0.133514 lr=9.011786e-05 lr_backbone=9.011786e-05 elapsed=12:11:55 avg_step=3.61s recent_step=2.70s eta=37:55:09 +2026-09-18 05:41:06 | INFO | train rank=0 | step=12180/50000 progress=24.4% loss=0.112468 lr=9.009811e-05 lr_backbone=9.009811e-05 elapsed=12:12:29 avg_step=3.61s recent_step=3.38s eta=37:54:26 +2026-09-18 05:41:56 | INFO | train rank=0 | step=12190/50000 progress=24.4% loss=0.089346 lr=9.007835e-05 lr_backbone=9.007835e-05 elapsed=12:13:19 avg_step=3.61s recent_step=5.00s eta=37:54:33 +2026-09-18 05:42:28 | INFO | train rank=0 | step=12200/50000 progress=24.4% loss=0.085119 lr=9.005857e-05 lr_backbone=9.005857e-05 elapsed=12:13:52 avg_step=3.61s recent_step=3.29s eta=37:53:47 +2026-09-18 05:43:04 | INFO | train rank=0 | step=12210/50000 progress=24.4% loss=0.084541 lr=9.003877e-05 lr_backbone=9.003877e-05 elapsed=12:14:27 avg_step=3.61s recent_step=3.51s eta=37:53:08 +2026-09-18 05:43:33 | INFO | train rank=0 | step=12220/50000 progress=24.4% loss=0.090858 lr=9.001895e-05 lr_backbone=9.001895e-05 elapsed=12:14:56 avg_step=3.61s recent_step=2.91s eta=37:52:10 +2026-09-18 05:44:19 | INFO | train rank=0 | step=12230/50000 progress=24.5% loss=0.109163 lr=8.999912e-05 lr_backbone=8.999912e-05 elapsed=12:15:42 avg_step=3.61s recent_step=4.60s eta=37:52:05 +2026-09-18 05:44:51 | INFO | train rank=0 | step=12240/50000 progress=24.5% loss=0.109782 lr=8.997927e-05 lr_backbone=8.997927e-05 elapsed=12:16:14 avg_step=3.61s recent_step=3.21s eta=37:51:16 +2026-09-18 05:45:22 | INFO | train rank=0 | step=12250/50000 progress=24.5% loss=0.182578 lr=8.995940e-05 lr_backbone=8.995940e-05 elapsed=12:16:45 avg_step=3.61s recent_step=3.11s eta=37:50:25 +2026-09-18 05:45:56 | INFO | train rank=0 | step=12260/50000 progress=24.5% loss=0.115520 lr=8.993951e-05 lr_backbone=8.993951e-05 elapsed=12:17:20 avg_step=3.61s recent_step=3.44s eta=37:49:44 +2026-09-18 05:46:44 | INFO | train rank=0 | step=12270/50000 progress=24.5% loss=0.087315 lr=8.991961e-05 lr_backbone=8.991961e-05 elapsed=12:18:07 avg_step=3.61s recent_step=4.74s eta=37:49:42 +2026-09-18 05:47:19 | INFO | train rank=0 | step=12280/50000 progress=24.6% loss=0.085369 lr=8.989969e-05 lr_backbone=8.989969e-05 elapsed=12:18:42 avg_step=3.61s recent_step=3.54s eta=37:49:04 +2026-09-18 05:47:51 | INFO | train rank=0 | step=12290/50000 progress=24.6% loss=0.110986 lr=8.987975e-05 lr_backbone=8.987975e-05 elapsed=12:19:15 avg_step=3.61s recent_step=3.21s eta=37:48:16 +2026-09-18 05:48:29 | INFO | train rank=0 | step=12300/50000 progress=24.6% loss=0.111031 lr=8.985979e-05 lr_backbone=8.985979e-05 elapsed=12:19:53 avg_step=3.61s recent_step=3.82s eta=37:47:46 +2026-09-18 05:49:19 | INFO | train rank=0 | step=12310/50000 progress=24.6% loss=0.085009 lr=8.983982e-05 lr_backbone=8.983982e-05 elapsed=12:20:42 avg_step=3.61s recent_step=4.98s eta=37:47:52 +2026-09-18 05:49:49 | INFO | train rank=0 | step=12320/50000 progress=24.6% loss=0.090612 lr=8.981983e-05 lr_backbone=8.981983e-05 elapsed=12:21:12 avg_step=3.61s recent_step=2.98s eta=37:46:57 +2026-09-18 05:50:21 | INFO | train rank=0 | step=12330/50000 progress=24.7% loss=0.106279 lr=8.979982e-05 lr_backbone=8.979982e-05 elapsed=12:21:44 avg_step=3.61s recent_step=3.23s eta=37:46:09 +2026-09-18 05:50:58 | INFO | train rank=0 | step=12340/50000 progress=24.7% loss=0.099137 lr=8.977980e-05 lr_backbone=8.977980e-05 elapsed=12:22:21 avg_step=3.61s recent_step=3.69s eta=37:45:35 +2026-09-18 05:51:45 | INFO | train rank=0 | step=12350/50000 progress=24.7% loss=0.082960 lr=8.975975e-05 lr_backbone=8.975975e-05 elapsed=12:23:08 avg_step=3.61s recent_step=4.69s eta=37:45:32 +2026-09-18 05:52:17 | INFO | train rank=0 | step=12360/50000 progress=24.7% loss=0.100011 lr=8.973969e-05 lr_backbone=8.973969e-05 elapsed=12:23:40 avg_step=3.61s recent_step=3.15s eta=37:44:42 +2026-09-18 05:52:44 | INFO | train rank=0 | step=12370/50000 progress=24.7% loss=0.190984 lr=8.971962e-05 lr_backbone=8.971962e-05 elapsed=12:24:07 avg_step=3.61s recent_step=2.71s eta=37:43:39 +2026-09-18 05:53:16 | INFO | train rank=0 | step=12380/50000 progress=24.8% loss=0.129081 lr=8.969952e-05 lr_backbone=8.969952e-05 elapsed=12:24:39 avg_step=3.61s recent_step=3.22s eta=37:42:51 +2026-09-18 05:54:15 | INFO | train rank=0 | step=12390/50000 progress=24.8% loss=0.112105 lr=8.967941e-05 lr_backbone=8.967941e-05 elapsed=12:25:38 avg_step=3.61s recent_step=5.91s eta=37:43:24 +2026-09-18 05:54:38 | INFO | train rank=0 | step=12400/50000 progress=24.8% loss=0.123783 lr=8.965928e-05 lr_backbone=8.965928e-05 elapsed=12:26:02 avg_step=3.61s recent_step=2.36s eta=37:42:10 +2026-09-18 05:55:09 | INFO | train rank=0 | step=12410/50000 progress=24.8% loss=0.111749 lr=8.963913e-05 lr_backbone=8.963913e-05 elapsed=12:26:32 avg_step=3.61s recent_step=3.01s eta=37:41:16 +2026-09-18 05:55:39 | INFO | train rank=0 | step=12420/50000 progress=24.8% loss=0.109275 lr=8.961897e-05 lr_backbone=8.961897e-05 elapsed=12:27:03 avg_step=3.61s recent_step=3.08s eta=37:40:24 +2026-09-18 05:56:25 | INFO | train rank=0 | step=12430/50000 progress=24.9% loss=0.089784 lr=8.959878e-05 lr_backbone=8.959878e-05 elapsed=12:27:48 avg_step=3.61s recent_step=4.58s eta=37:40:17 +2026-09-18 05:56:58 | INFO | train rank=0 | step=12440/50000 progress=24.9% loss=0.088660 lr=8.957858e-05 lr_backbone=8.957858e-05 elapsed=12:28:21 avg_step=3.61s recent_step=3.23s eta=37:39:30 +2026-09-18 05:57:31 | INFO | train rank=0 | step=12450/50000 progress=24.9% loss=0.141012 lr=8.955837e-05 lr_backbone=8.955837e-05 elapsed=12:28:55 avg_step=3.61s recent_step=3.39s eta=37:38:47 +2026-09-18 05:58:09 | INFO | train rank=0 | step=12460/50000 progress=24.9% loss=0.106016 lr=8.953813e-05 lr_backbone=8.953813e-05 elapsed=12:29:33 avg_step=3.61s recent_step=3.80s eta=37:38:16 +2026-09-18 05:58:52 | INFO | train rank=0 | step=12470/50000 progress=24.9% loss=0.086781 lr=8.951788e-05 lr_backbone=8.951788e-05 elapsed=12:30:15 avg_step=3.61s recent_step=4.25s eta=37:38:00 +2026-09-18 05:59:28 | INFO | train rank=0 | step=12480/50000 progress=25.0% loss=0.101158 lr=8.949761e-05 lr_backbone=8.949761e-05 elapsed=12:30:51 avg_step=3.61s recent_step=3.56s eta=37:37:22 +2026-09-18 05:59:54 | INFO | train rank=0 | step=12490/50000 progress=25.0% loss=0.115761 lr=8.947733e-05 lr_backbone=8.947733e-05 elapsed=12:31:18 avg_step=3.61s recent_step=2.69s eta=37:36:19 +2026-09-18 06:00:26 | INFO | train rank=0 | step=12500/50000 progress=25.0% loss=0.100546 lr=8.945703e-05 lr_backbone=8.945703e-05 elapsed=12:31:49 avg_step=3.61s recent_step=3.12s eta=37:35:28 +2026-09-18 06:01:27 | INFO | train rank=0 | step=12510/50000 progress=25.0% loss=0.117552 lr=8.943671e-05 lr_backbone=8.943671e-05 elapsed=12:32:50 avg_step=3.61s recent_step=6.16s eta=37:36:08 +2026-09-18 06:01:56 | INFO | train rank=0 | step=12520/50000 progress=25.0% loss=0.079984 lr=8.941637e-05 lr_backbone=8.941637e-05 elapsed=12:33:19 avg_step=3.61s recent_step=2.90s eta=37:35:10 +2026-09-18 06:02:28 | INFO | train rank=0 | step=12530/50000 progress=25.1% loss=0.074226 lr=8.939601e-05 lr_backbone=8.939601e-05 elapsed=12:33:51 avg_step=3.61s recent_step=3.18s eta=37:34:21 +2026-09-18 06:02:57 | INFO | train rank=0 | step=12540/50000 progress=25.1% loss=0.079043 lr=8.937564e-05 lr_backbone=8.937564e-05 elapsed=12:34:20 avg_step=3.61s recent_step=2.91s eta=37:33:24 +2026-09-18 06:03:43 | INFO | train rank=0 | step=12550/50000 progress=25.1% loss=0.084585 lr=8.935525e-05 lr_backbone=8.935525e-05 elapsed=12:35:07 avg_step=3.61s recent_step=4.64s eta=37:33:19 +2026-09-18 06:04:17 | INFO | train rank=0 | step=12560/50000 progress=25.1% loss=0.124886 lr=8.933485e-05 lr_backbone=8.933485e-05 elapsed=12:35:40 avg_step=3.61s recent_step=3.37s eta=37:32:36 +2026-09-18 06:04:46 | INFO | train rank=0 | step=12570/50000 progress=25.1% loss=0.127592 lr=8.931442e-05 lr_backbone=8.931442e-05 elapsed=12:36:10 avg_step=3.61s recent_step=2.94s eta=37:31:40 +2026-09-18 06:05:21 | INFO | train rank=0 | step=12580/50000 progress=25.2% loss=0.094074 lr=8.929398e-05 lr_backbone=8.929398e-05 elapsed=12:36:45 avg_step=3.61s recent_step=3.50s eta=37:31:00 +2026-09-18 06:06:02 | INFO | train rank=0 | step=12590/50000 progress=25.2% loss=0.095665 lr=8.927352e-05 lr_backbone=8.927352e-05 elapsed=12:37:26 avg_step=3.61s recent_step=4.10s eta=37:30:39 +2026-09-18 06:06:29 | INFO | train rank=0 | step=12600/50000 progress=25.2% loss=0.105380 lr=8.925305e-05 lr_backbone=8.925305e-05 elapsed=12:37:53 avg_step=3.61s recent_step=2.68s eta=37:29:35 +2026-09-18 06:07:05 | INFO | train rank=0 | step=12610/50000 progress=25.2% loss=0.132669 lr=8.923256e-05 lr_backbone=8.923256e-05 elapsed=12:38:29 avg_step=3.61s recent_step=3.60s eta=37:28:59 +2026-09-18 06:07:56 | INFO | train rank=0 | step=12620/50000 progress=25.2% loss=0.094552 lr=8.921205e-05 lr_backbone=8.921205e-05 elapsed=12:39:19 avg_step=3.61s recent_step=5.02s eta=37:29:05 +2026-09-18 06:08:28 | INFO | train rank=0 | step=12630/50000 progress=25.3% loss=0.108465 lr=8.919152e-05 lr_backbone=8.919152e-05 elapsed=12:39:51 avg_step=3.61s recent_step=3.22s eta=37:28:17 +2026-09-18 06:09:00 | INFO | train rank=0 | step=12640/50000 progress=25.3% loss=0.113287 lr=8.917098e-05 lr_backbone=8.917098e-05 elapsed=12:40:24 avg_step=3.61s recent_step=3.25s eta=37:27:30 +2026-09-18 06:09:31 | INFO | train rank=0 | step=12650/50000 progress=25.3% loss=0.080157 lr=8.915041e-05 lr_backbone=8.915041e-05 elapsed=12:40:54 avg_step=3.61s recent_step=3.09s eta=37:26:39 +2026-09-18 06:10:20 | INFO | train rank=0 | step=12660/50000 progress=25.3% loss=0.100290 lr=8.912984e-05 lr_backbone=8.912984e-05 elapsed=12:41:43 avg_step=3.61s recent_step=4.88s eta=37:26:40 +2026-09-18 06:10:52 | INFO | train rank=0 | step=12670/50000 progress=25.3% loss=0.100086 lr=8.910924e-05 lr_backbone=8.910924e-05 elapsed=12:42:16 avg_step=3.61s recent_step=3.24s eta=37:25:53 +2026-09-18 06:11:23 | INFO | train rank=0 | step=12680/50000 progress=25.4% loss=0.103459 lr=8.908863e-05 lr_backbone=8.908863e-05 elapsed=12:42:46 avg_step=3.61s recent_step=3.02s eta=37:25:00 +2026-09-18 06:11:46 | INFO | train rank=0 | step=12690/50000 progress=25.4% loss=0.176595 lr=8.906800e-05 lr_backbone=8.906800e-05 elapsed=12:43:10 avg_step=3.61s recent_step=2.39s eta=37:23:48 +2026-09-18 06:12:36 | INFO | train rank=0 | step=12700/50000 progress=25.4% loss=0.123017 lr=8.904735e-05 lr_backbone=8.904735e-05 elapsed=12:43:59 avg_step=3.61s recent_step=4.92s eta=37:23:50 +2026-09-18 06:13:11 | INFO | train rank=0 | step=12710/50000 progress=25.4% loss=0.075436 lr=8.902669e-05 lr_backbone=8.902669e-05 elapsed=12:44:34 avg_step=3.61s recent_step=3.48s eta=37:23:11 +2026-09-18 06:13:47 | INFO | train rank=0 | step=12720/50000 progress=25.4% loss=0.112375 lr=8.900601e-05 lr_backbone=8.900601e-05 elapsed=12:45:11 avg_step=3.61s recent_step=3.68s eta=37:22:37 +2026-09-18 06:14:12 | INFO | train rank=0 | step=12730/50000 progress=25.5% loss=0.135437 lr=8.898531e-05 lr_backbone=8.898531e-05 elapsed=12:45:35 avg_step=3.61s recent_step=2.47s eta=37:21:27 +2026-09-18 06:15:03 | INFO | train rank=0 | step=12740/50000 progress=25.5% loss=0.088125 lr=8.896460e-05 lr_backbone=8.896460e-05 elapsed=12:46:26 avg_step=3.61s recent_step=5.09s eta=37:21:34 +2026-09-18 06:15:32 | INFO | train rank=0 | step=12750/50000 progress=25.5% loss=0.095638 lr=8.894386e-05 lr_backbone=8.894386e-05 elapsed=12:46:56 avg_step=3.61s recent_step=2.93s eta=37:20:38 +2026-09-18 06:16:07 | INFO | train rank=0 | step=12760/50000 progress=25.5% loss=0.132858 lr=8.892312e-05 lr_backbone=8.892312e-05 elapsed=12:47:30 avg_step=3.61s recent_step=3.44s eta=37:19:57 +2026-09-18 06:16:35 | INFO | train rank=0 | step=12770/50000 progress=25.5% loss=0.101308 lr=8.890235e-05 lr_backbone=8.890235e-05 elapsed=12:47:58 avg_step=3.61s recent_step=2.81s eta=37:18:58 +2026-09-18 06:17:34 | INFO | train rank=0 | step=12780/50000 progress=25.6% loss=0.106382 lr=8.888157e-05 lr_backbone=8.888157e-05 elapsed=12:48:57 avg_step=3.61s recent_step=5.92s eta=37:19:29 +2026-09-18 06:18:03 | INFO | train rank=0 | step=12790/50000 progress=25.6% loss=0.093560 lr=8.886077e-05 lr_backbone=8.886077e-05 elapsed=12:49:26 avg_step=3.61s recent_step=2.90s eta=37:18:33 +2026-09-18 06:18:42 | INFO | train rank=0 | step=12800/50000 progress=25.6% loss=0.081353 lr=8.883995e-05 lr_backbone=8.883995e-05 elapsed=12:50:05 avg_step=3.61s recent_step=3.90s eta=37:18:05 +2026-09-18 06:19:11 | INFO | train rank=0 | step=12810/50000 progress=25.6% loss=0.110181 lr=8.881912e-05 lr_backbone=8.881912e-05 elapsed=12:50:35 avg_step=3.61s recent_step=2.94s eta=37:17:09 +2026-09-18 06:20:03 | INFO | train rank=0 | step=12820/50000 progress=25.6% loss=0.109683 lr=8.879826e-05 lr_backbone=8.879826e-05 elapsed=12:51:26 avg_step=3.61s recent_step=5.19s eta=37:17:19 +2026-09-18 06:20:38 | INFO | train rank=0 | step=12830/50000 progress=25.7% loss=0.102831 lr=8.877740e-05 lr_backbone=8.877740e-05 elapsed=12:52:01 avg_step=3.61s recent_step=3.48s eta=37:16:39 +2026-09-18 06:21:15 | INFO | train rank=0 | step=12840/50000 progress=25.7% loss=0.114183 lr=8.875651e-05 lr_backbone=8.875651e-05 elapsed=12:52:38 avg_step=3.61s recent_step=3.71s eta=37:16:06 +2026-09-18 06:21:40 | INFO | train rank=0 | step=12850/50000 progress=25.7% loss=0.124234 lr=8.873561e-05 lr_backbone=8.873561e-05 elapsed=12:53:03 avg_step=3.61s recent_step=2.48s eta=37:14:57 +2026-09-18 06:22:18 | INFO | train rank=0 | step=12860/50000 progress=25.7% loss=0.091228 lr=8.871469e-05 lr_backbone=8.871469e-05 elapsed=12:53:41 avg_step=3.61s recent_step=3.80s eta=37:14:26 +2026-09-18 06:23:04 | INFO | train rank=0 | step=12870/50000 progress=25.7% loss=0.115559 lr=8.869376e-05 lr_backbone=8.869376e-05 elapsed=12:54:27 avg_step=3.61s recent_step=4.57s eta=37:14:18 +2026-09-18 06:23:35 | INFO | train rank=0 | step=12880/50000 progress=25.8% loss=0.098094 lr=8.867280e-05 lr_backbone=8.867280e-05 elapsed=12:54:58 avg_step=3.61s recent_step=3.11s eta=37:13:27 +2026-09-18 06:24:00 | INFO | train rank=0 | step=12890/50000 progress=25.8% loss=0.102547 lr=8.865183e-05 lr_backbone=8.865183e-05 elapsed=12:55:23 avg_step=3.61s recent_step=2.50s eta=37:12:19 +2026-09-18 06:24:39 | INFO | train rank=0 | step=12900/50000 progress=25.8% loss=0.115332 lr=8.863085e-05 lr_backbone=8.863085e-05 elapsed=12:56:03 avg_step=3.61s recent_step=3.97s eta=37:11:54 +2026-09-18 06:25:26 | INFO | train rank=0 | step=12910/50000 progress=25.8% loss=0.094886 lr=8.860984e-05 lr_backbone=8.860984e-05 elapsed=12:56:50 avg_step=3.61s recent_step=4.69s eta=37:11:49 +2026-09-18 06:26:01 | INFO | train rank=0 | step=12920/50000 progress=25.8% loss=0.181757 lr=8.858882e-05 lr_backbone=8.858882e-05 elapsed=12:57:24 avg_step=3.61s recent_step=3.44s eta=37:11:07 +2026-09-18 06:26:36 | INFO | train rank=0 | step=12930/50000 progress=25.9% loss=0.108682 lr=8.856779e-05 lr_backbone=8.856779e-05 elapsed=12:57:59 avg_step=3.61s recent_step=3.52s eta=37:10:29 +2026-09-18 06:27:05 | INFO | train rank=0 | step=12940/50000 progress=25.9% loss=0.113680 lr=8.854673e-05 lr_backbone=8.854673e-05 elapsed=12:58:29 avg_step=3.61s recent_step=2.97s eta=37:09:34 +2026-09-18 06:27:47 | INFO | train rank=0 | step=12950/50000 progress=25.9% loss=0.091688 lr=8.852566e-05 lr_backbone=8.852566e-05 elapsed=12:59:10 avg_step=3.61s recent_step=4.17s eta=37:09:14 +2026-09-18 06:28:15 | INFO | train rank=0 | step=12960/50000 progress=25.9% loss=0.114733 lr=8.850457e-05 lr_backbone=8.850457e-05 elapsed=12:59:38 avg_step=3.61s recent_step=2.79s eta=37:08:15 +2026-09-18 06:28:48 | INFO | train rank=0 | step=12970/50000 progress=25.9% loss=0.090700 lr=8.848347e-05 lr_backbone=8.848347e-05 elapsed=13:00:12 avg_step=3.61s recent_step=3.31s eta=37:07:30 +2026-09-18 06:29:15 | INFO | train rank=0 | step=12980/50000 progress=26.0% loss=0.114449 lr=8.846235e-05 lr_backbone=8.846235e-05 elapsed=13:00:38 avg_step=3.61s recent_step=2.64s eta=37:06:26 +2026-09-18 06:29:44 | INFO | train rank=0 | step=12990/50000 progress=26.0% loss=0.100854 lr=8.844121e-05 lr_backbone=8.844121e-05 elapsed=13:01:08 avg_step=3.61s recent_step=2.97s eta=37:05:32 +2026-09-18 06:30:12 | INFO | train rank=0 | step=13000/50000 progress=26.0% loss=0.097659 lr=8.842006e-05 lr_backbone=8.842006e-05 elapsed=13:01:36 avg_step=3.61s recent_step=2.79s eta=37:04:33 +2026-09-18 06:30:46 | INFO | train rank=0 | step=13010/50000 progress=26.0% loss=0.093364 lr=8.839888e-05 lr_backbone=8.839888e-05 elapsed=13:02:10 avg_step=3.61s recent_step=3.40s eta=37:03:51 +2026-09-18 06:31:10 | INFO | train rank=0 | step=13020/50000 progress=26.0% loss=0.090385 lr=8.837770e-05 lr_backbone=8.837770e-05 elapsed=13:02:34 avg_step=3.61s recent_step=2.40s eta=37:02:40 +2026-09-18 06:31:32 | INFO | train rank=0 | step=13030/50000 progress=26.1% loss=0.127797 lr=8.835649e-05 lr_backbone=8.835649e-05 elapsed=13:02:55 avg_step=3.61s recent_step=2.14s eta=37:01:23 +2026-09-18 06:31:54 | INFO | train rank=0 | step=13040/50000 progress=26.1% loss=0.160863 lr=8.833527e-05 lr_backbone=8.833527e-05 elapsed=13:03:17 avg_step=3.60s recent_step=2.20s eta=37:00:07 +2026-09-18 06:32:19 | INFO | train rank=0 | step=13050/50000 progress=26.1% loss=0.144294 lr=8.831403e-05 lr_backbone=8.831403e-05 elapsed=13:03:43 avg_step=3.60s recent_step=2.57s eta=36:59:02 +2026-09-18 06:32:41 | INFO | train rank=0 | step=13060/50000 progress=26.1% loss=0.145161 lr=8.829277e-05 lr_backbone=8.829277e-05 elapsed=13:04:05 avg_step=3.60s recent_step=2.21s eta=36:57:46 +2026-09-18 06:33:07 | INFO | train rank=0 | step=13070/50000 progress=26.1% loss=0.103316 lr=8.827150e-05 lr_backbone=8.827150e-05 elapsed=13:04:30 avg_step=3.60s recent_step=2.51s eta=36:56:39 +2026-09-18 06:33:26 | INFO | train rank=0 | step=13080/50000 progress=26.2% loss=0.076192 lr=8.825021e-05 lr_backbone=8.825021e-05 elapsed=13:04:50 avg_step=3.60s recent_step=2.00s eta=36:55:18 +2026-09-18 06:33:49 | INFO | train rank=0 | step=13090/50000 progress=26.2% loss=0.112786 lr=8.822891e-05 lr_backbone=8.822891e-05 elapsed=13:05:12 avg_step=3.60s recent_step=2.25s eta=36:54:04 +2026-09-18 06:34:11 | INFO | train rank=0 | step=13100/50000 progress=26.2% loss=0.106016 lr=8.820758e-05 lr_backbone=8.820758e-05 elapsed=13:05:34 avg_step=3.60s recent_step=2.16s eta=36:52:47 +2026-09-18 06:34:38 | INFO | train rank=0 | step=13110/50000 progress=26.2% loss=0.108530 lr=8.818624e-05 lr_backbone=8.818624e-05 elapsed=13:06:01 avg_step=3.60s recent_step=2.70s eta=36:51:46 +2026-09-18 06:35:13 | INFO | train rank=0 | step=13120/50000 progress=26.2% loss=0.097422 lr=8.816489e-05 lr_backbone=8.816489e-05 elapsed=13:06:36 avg_step=3.60s recent_step=3.54s eta=36:51:08 +2026-09-18 06:35:43 | INFO | train rank=0 | step=13130/50000 progress=26.3% loss=0.144082 lr=8.814351e-05 lr_backbone=8.814351e-05 elapsed=13:07:06 avg_step=3.60s recent_step=2.98s eta=36:50:15 +2026-09-18 06:36:13 | INFO | train rank=0 | step=13140/50000 progress=26.3% loss=0.116050 lr=8.812213e-05 lr_backbone=8.812213e-05 elapsed=13:07:36 avg_step=3.60s recent_step=3.01s eta=36:49:23 +2026-09-18 06:36:40 | INFO | train rank=0 | step=13150/50000 progress=26.3% loss=0.082038 lr=8.810072e-05 lr_backbone=8.810072e-05 elapsed=13:08:03 avg_step=3.60s recent_step=2.73s eta=36:48:22 +2026-09-18 06:37:31 | INFO | train rank=0 | step=13160/50000 progress=26.3% loss=0.110171 lr=8.807930e-05 lr_backbone=8.807930e-05 elapsed=13:08:54 avg_step=3.60s recent_step=5.04s eta=36:48:27 +2026-09-18 06:38:26 | INFO | train rank=0 | step=13170/50000 progress=26.3% loss=0.124902 lr=8.805786e-05 lr_backbone=8.805786e-05 elapsed=13:09:49 avg_step=3.60s recent_step=5.51s eta=36:48:45 +2026-09-18 06:39:11 | INFO | train rank=0 | step=13180/50000 progress=26.4% loss=0.096881 lr=8.803640e-05 lr_backbone=8.803640e-05 elapsed=13:10:34 avg_step=3.60s recent_step=4.55s eta=36:48:35 +2026-09-18 06:39:54 | INFO | train rank=0 | step=13190/50000 progress=26.4% loss=0.110175 lr=8.801493e-05 lr_backbone=8.801493e-05 elapsed=13:11:18 avg_step=3.60s recent_step=4.31s eta=36:48:19 +2026-09-18 06:40:39 | INFO | train rank=0 | step=13200/50000 progress=26.4% loss=0.091844 lr=8.799344e-05 lr_backbone=8.799344e-05 elapsed=13:12:03 avg_step=3.60s recent_step=4.51s eta=36:48:08 +2026-09-18 06:41:24 | INFO | train rank=0 | step=13210/50000 progress=26.4% loss=0.110441 lr=8.797193e-05 lr_backbone=8.797193e-05 elapsed=13:12:48 avg_step=3.60s recent_step=4.48s eta=36:47:57 +2026-09-18 06:42:12 | INFO | train rank=0 | step=13220/50000 progress=26.4% loss=0.105876 lr=8.795041e-05 lr_backbone=8.795041e-05 elapsed=13:13:35 avg_step=3.60s recent_step=4.76s eta=36:47:53 +2026-09-18 06:42:57 | INFO | train rank=0 | step=13230/50000 progress=26.5% loss=0.087747 lr=8.792887e-05 lr_backbone=8.792887e-05 elapsed=13:14:20 avg_step=3.60s recent_step=4.49s eta=36:47:42 +2026-09-18 06:43:44 | INFO | train rank=0 | step=13240/50000 progress=26.5% loss=0.100743 lr=8.790732e-05 lr_backbone=8.790732e-05 elapsed=13:15:08 avg_step=3.60s recent_step=4.74s eta=36:47:38 +2026-09-18 06:44:32 | INFO | train rank=0 | step=13250/50000 progress=26.5% loss=0.115263 lr=8.788574e-05 lr_backbone=8.788574e-05 elapsed=13:15:55 avg_step=3.60s recent_step=4.79s eta=36:47:35 +2026-09-18 06:45:24 | INFO | train rank=0 | step=13260/50000 progress=26.5% loss=0.099030 lr=8.786415e-05 lr_backbone=8.786415e-05 elapsed=13:16:47 avg_step=3.61s recent_step=5.15s eta=36:47:41 +2026-09-18 06:46:09 | INFO | train rank=0 | step=13270/50000 progress=26.5% loss=0.104322 lr=8.784255e-05 lr_backbone=8.784255e-05 elapsed=13:17:33 avg_step=3.61s recent_step=4.56s eta=36:47:32 +2026-09-18 06:46:58 | INFO | train rank=0 | step=13280/50000 progress=26.6% loss=0.107907 lr=8.782093e-05 lr_backbone=8.782093e-05 elapsed=13:18:21 avg_step=3.61s recent_step=4.83s eta=36:47:30 +2026-09-18 06:47:50 | INFO | train rank=0 | step=13290/50000 progress=26.6% loss=0.143935 lr=8.779929e-05 lr_backbone=8.779929e-05 elapsed=13:19:13 avg_step=3.61s recent_step=5.26s eta=36:47:39 +2026-09-18 06:48:38 | INFO | train rank=0 | step=13300/50000 progress=26.6% loss=0.093683 lr=8.777763e-05 lr_backbone=8.777763e-05 elapsed=13:20:01 avg_step=3.61s recent_step=4.79s eta=36:47:36 +2026-09-18 06:49:28 | INFO | train rank=0 | step=13310/50000 progress=26.6% loss=0.085624 lr=8.775596e-05 lr_backbone=8.775596e-05 elapsed=13:20:51 avg_step=3.61s recent_step=4.97s eta=36:47:37 +2026-09-18 06:50:15 | INFO | train rank=0 | step=13320/50000 progress=26.6% loss=0.096933 lr=8.773427e-05 lr_backbone=8.773427e-05 elapsed=13:21:39 avg_step=3.61s recent_step=4.77s eta=36:47:33 +2026-09-18 06:51:04 | INFO | train rank=0 | step=13330/50000 progress=26.7% loss=0.105807 lr=8.771257e-05 lr_backbone=8.771257e-05 elapsed=13:22:27 avg_step=3.61s recent_step=4.81s eta=36:47:30 +2026-09-18 06:51:52 | INFO | train rank=0 | step=13340/50000 progress=26.7% loss=0.124004 lr=8.769085e-05 lr_backbone=8.769085e-05 elapsed=13:23:15 avg_step=3.61s recent_step=4.85s eta=36:47:28 +2026-09-18 06:52:38 | INFO | train rank=0 | step=13350/50000 progress=26.7% loss=0.153403 lr=8.766911e-05 lr_backbone=8.766911e-05 elapsed=13:24:01 avg_step=3.61s recent_step=4.59s eta=36:47:19 +2026-09-18 06:53:24 | INFO | train rank=0 | step=13360/50000 progress=26.7% loss=0.112023 lr=8.764736e-05 lr_backbone=8.764736e-05 elapsed=13:24:48 avg_step=3.61s recent_step=4.64s eta=36:47:10 +2026-09-18 06:54:07 | INFO | train rank=0 | step=13370/50000 progress=26.7% loss=0.136952 lr=8.762559e-05 lr_backbone=8.762559e-05 elapsed=13:25:31 avg_step=3.61s recent_step=4.30s eta=36:46:53 +2026-09-18 06:54:55 | INFO | train rank=0 | step=13380/50000 progress=26.8% loss=0.102169 lr=8.760380e-05 lr_backbone=8.760380e-05 elapsed=13:26:18 avg_step=3.62s recent_step=4.74s eta=36:46:48 +2026-09-18 06:55:41 | INFO | train rank=0 | step=13390/50000 progress=26.8% loss=0.097215 lr=8.758200e-05 lr_backbone=8.758200e-05 elapsed=13:27:04 avg_step=3.62s recent_step=4.61s eta=36:46:39 +2026-09-18 06:56:25 | INFO | train rank=0 | step=13400/50000 progress=26.8% loss=0.100797 lr=8.756018e-05 lr_backbone=8.756018e-05 elapsed=13:27:48 avg_step=3.62s recent_step=4.41s eta=36:46:24 +2026-09-18 06:57:15 | INFO | train rank=0 | step=13410/50000 progress=26.8% loss=0.128110 lr=8.753834e-05 lr_backbone=8.753834e-05 elapsed=13:28:38 avg_step=3.62s recent_step=5.00s eta=36:46:26 +2026-09-18 06:58:02 | INFO | train rank=0 | step=13420/50000 progress=26.8% loss=0.096084 lr=8.751649e-05 lr_backbone=8.751649e-05 elapsed=13:29:25 avg_step=3.62s recent_step=4.72s eta=36:46:19 +2026-09-18 06:58:53 | INFO | train rank=0 | step=13430/50000 progress=26.9% loss=0.081534 lr=8.749462e-05 lr_backbone=8.749462e-05 elapsed=13:30:16 avg_step=3.62s recent_step=5.06s eta=36:46:22 +2026-09-18 06:59:38 | INFO | train rank=0 | step=13440/50000 progress=26.9% loss=0.091292 lr=8.747273e-05 lr_backbone=8.747273e-05 elapsed=13:31:01 avg_step=3.62s recent_step=4.50s eta=36:46:10 +2026-09-18 07:00:25 | INFO | train rank=0 | step=13450/50000 progress=26.9% loss=0.115527 lr=8.745083e-05 lr_backbone=8.745083e-05 elapsed=13:31:48 avg_step=3.62s recent_step=4.71s eta=36:46:04 +2026-09-18 07:01:14 | INFO | train rank=0 | step=13460/50000 progress=26.9% loss=0.088298 lr=8.742891e-05 lr_backbone=8.742891e-05 elapsed=13:32:37 avg_step=3.62s recent_step=4.92s eta=36:46:03 +2026-09-18 07:02:03 | INFO | train rank=0 | step=13470/50000 progress=26.9% loss=0.118809 lr=8.740698e-05 lr_backbone=8.740698e-05 elapsed=13:33:26 avg_step=3.62s recent_step=4.91s eta=36:46:01 +2026-09-18 07:02:50 | INFO | train rank=0 | step=13480/50000 progress=27.0% loss=0.110990 lr=8.738503e-05 lr_backbone=8.738503e-05 elapsed=13:34:14 avg_step=3.62s recent_step=4.72s eta=36:45:55 +2026-09-18 07:03:43 | INFO | train rank=0 | step=13490/50000 progress=27.0% loss=0.092417 lr=8.736306e-05 lr_backbone=8.736306e-05 elapsed=13:35:06 avg_step=3.63s recent_step=5.24s eta=36:46:02 +2026-09-18 07:04:24 | INFO | train rank=0 | step=13500/50000 progress=27.0% loss=0.103656 lr=8.734107e-05 lr_backbone=8.734107e-05 elapsed=13:35:47 avg_step=3.63s recent_step=4.09s eta=36:45:39 +2026-09-18 07:05:00 | INFO | train rank=0 | step=13510/50000 progress=27.0% loss=0.100696 lr=8.731907e-05 lr_backbone=8.731907e-05 elapsed=13:36:23 avg_step=3.63s recent_step=3.61s eta=36:45:02 +2026-09-18 07:05:37 | INFO | train rank=0 | step=13520/50000 progress=27.0% loss=0.096119 lr=8.729706e-05 lr_backbone=8.729706e-05 elapsed=13:37:00 avg_step=3.63s recent_step=3.72s eta=36:44:28 +2026-09-18 07:06:36 | INFO | train rank=0 | step=13530/50000 progress=27.1% loss=0.115178 lr=8.727502e-05 lr_backbone=8.727502e-05 elapsed=13:38:00 avg_step=3.63s recent_step=5.95s eta=36:44:55 +2026-09-18 07:07:46 | INFO | train rank=0 | step=13540/50000 progress=27.1% loss=0.148984 lr=8.725298e-05 lr_backbone=8.725298e-05 elapsed=13:39:09 avg_step=3.63s recent_step=6.92s eta=36:45:47 +2026-09-18 07:09:01 | INFO | train rank=0 | step=13550/50000 progress=27.1% loss=0.111032 lr=8.723091e-05 lr_backbone=8.723091e-05 elapsed=13:40:24 avg_step=3.63s recent_step=7.54s eta=36:46:56 +2026-09-18 07:10:14 | INFO | train rank=0 | step=13560/50000 progress=27.1% loss=0.153547 lr=8.720883e-05 lr_backbone=8.720883e-05 elapsed=13:41:37 avg_step=3.64s recent_step=7.26s eta=36:47:57 +2026-09-18 07:11:28 | INFO | train rank=0 | step=13570/50000 progress=27.1% loss=0.090810 lr=8.718673e-05 lr_backbone=8.718673e-05 elapsed=13:42:51 avg_step=3.64s recent_step=7.43s eta=36:49:03 +2026-09-18 07:12:39 | INFO | train rank=0 | step=13580/50000 progress=27.2% loss=0.092419 lr=8.716462e-05 lr_backbone=8.716462e-05 elapsed=13:44:02 avg_step=3.64s recent_step=7.13s eta=36:50:00 +2026-09-18 07:13:49 | INFO | train rank=0 | step=13590/50000 progress=27.2% loss=0.111319 lr=8.714249e-05 lr_backbone=8.714249e-05 elapsed=13:45:12 avg_step=3.64s recent_step=6.96s eta=36:50:52 +2026-09-18 07:14:55 | INFO | train rank=0 | step=13600/50000 progress=27.2% loss=0.087113 lr=8.712034e-05 lr_backbone=8.712034e-05 elapsed=13:46:18 avg_step=3.65s recent_step=6.62s eta=36:51:35 +2026-09-18 07:21:11 | INFO | train rank=0 | step=13610/50000 progress=27.2% loss=0.120853 lr=8.709818e-05 lr_backbone=8.709818e-05 elapsed=13:52:34 avg_step=3.67s recent_step=37.62s eta=37:06:07 +2026-09-18 07:21:40 | INFO | train rank=0 | step=13620/50000 progress=27.2% loss=0.082314 lr=8.707600e-05 lr_backbone=8.707600e-05 elapsed=13:53:03 avg_step=3.67s recent_step=2.84s eta=37:05:09 +2026-09-18 07:22:11 | INFO | train rank=0 | step=13630/50000 progress=27.3% loss=0.107569 lr=8.705380e-05 lr_backbone=8.705380e-05 elapsed=13:53:34 avg_step=3.67s recent_step=3.15s eta=37:04:18 +2026-09-18 07:22:41 | INFO | train rank=0 | step=13640/50000 progress=27.3% loss=0.075185 lr=8.703159e-05 lr_backbone=8.703159e-05 elapsed=13:54:05 avg_step=3.67s recent_step=3.03s eta=37:03:24 +2026-09-18 07:23:38 | INFO | train rank=0 | step=13650/50000 progress=27.3% loss=0.102265 lr=8.700936e-05 lr_backbone=8.700936e-05 elapsed=13:55:01 avg_step=3.67s recent_step=5.65s eta=37:03:40 +2026-09-18 07:24:02 | INFO | train rank=0 | step=13660/50000 progress=27.3% loss=0.104456 lr=8.698712e-05 lr_backbone=8.698712e-05 elapsed=13:55:26 avg_step=3.67s recent_step=2.44s eta=37:02:31 +2026-09-18 07:24:36 | INFO | train rank=0 | step=13670/50000 progress=27.3% loss=0.108066 lr=8.696486e-05 lr_backbone=8.696486e-05 elapsed=13:55:59 avg_step=3.67s recent_step=3.35s eta=37:01:46 +2026-09-18 07:25:09 | INFO | train rank=0 | step=13680/50000 progress=27.4% loss=0.096471 lr=8.694258e-05 lr_backbone=8.694258e-05 elapsed=13:56:32 avg_step=3.67s recent_step=3.28s eta=37:00:59 +2026-09-18 07:26:02 | INFO | train rank=0 | step=13690/50000 progress=27.4% loss=0.090169 lr=8.692029e-05 lr_backbone=8.692029e-05 elapsed=13:57:25 avg_step=3.67s recent_step=5.32s eta=37:01:06 +2026-09-18 07:26:36 | INFO | train rank=0 | step=13700/50000 progress=27.4% loss=0.108350 lr=8.689798e-05 lr_backbone=8.689798e-05 elapsed=13:57:59 avg_step=3.67s recent_step=3.38s eta=37:00:22 +2026-09-18 07:27:07 | INFO | train rank=0 | step=13710/50000 progress=27.4% loss=0.114063 lr=8.687566e-05 lr_backbone=8.687566e-05 elapsed=13:58:30 avg_step=3.67s recent_step=3.12s eta=36:59:30 +2026-09-18 07:27:42 | INFO | train rank=0 | step=13720/50000 progress=27.4% loss=0.089901 lr=8.685331e-05 lr_backbone=8.685331e-05 elapsed=13:59:05 avg_step=3.67s recent_step=3.52s eta=36:58:50 +2026-09-18 07:28:34 | INFO | train rank=0 | step=13730/50000 progress=27.5% loss=0.079254 lr=8.683096e-05 lr_backbone=8.683096e-05 elapsed=13:59:57 avg_step=3.67s recent_step=5.18s eta=36:58:53 +2026-09-18 07:29:02 | INFO | train rank=0 | step=13740/50000 progress=27.5% loss=0.099148 lr=8.680858e-05 lr_backbone=8.680858e-05 elapsed=14:00:25 avg_step=3.67s recent_step=2.79s eta=36:57:53 +2026-09-18 07:29:36 | INFO | train rank=0 | step=13750/50000 progress=27.5% loss=0.118086 lr=8.678620e-05 lr_backbone=8.678620e-05 elapsed=14:00:59 avg_step=3.67s recent_step=3.43s eta=36:57:10 +2026-09-18 07:30:03 | INFO | train rank=0 | step=13760/50000 progress=27.5% loss=0.079649 lr=8.676379e-05 lr_backbone=8.676379e-05 elapsed=14:01:26 avg_step=3.67s recent_step=2.68s eta=36:56:07 +2026-09-18 07:30:52 | INFO | train rank=0 | step=13770/50000 progress=27.5% loss=0.096964 lr=8.674137e-05 lr_backbone=8.674137e-05 elapsed=14:02:15 avg_step=3.67s recent_step=4.90s eta=36:56:03 +2026-09-18 07:31:28 | INFO | train rank=0 | step=13780/50000 progress=27.6% loss=0.081549 lr=8.671893e-05 lr_backbone=8.671893e-05 elapsed=14:02:51 avg_step=3.67s recent_step=3.64s eta=36:55:25 +2026-09-18 07:31:57 | INFO | train rank=0 | step=13790/50000 progress=27.6% loss=0.110537 lr=8.669648e-05 lr_backbone=8.669648e-05 elapsed=14:03:20 avg_step=3.67s recent_step=2.89s eta=36:54:28 +2026-09-18 07:32:26 | INFO | train rank=0 | step=13800/50000 progress=27.6% loss=0.081608 lr=8.667401e-05 lr_backbone=8.667401e-05 elapsed=14:03:49 avg_step=3.67s recent_step=2.90s eta=36:53:31 +2026-09-18 07:33:22 | INFO | train rank=0 | step=13810/50000 progress=27.6% loss=0.088617 lr=8.665152e-05 lr_backbone=8.665152e-05 elapsed=14:04:45 avg_step=3.67s recent_step=5.55s eta=36:53:44 +2026-09-18 07:33:47 | INFO | train rank=0 | step=13820/50000 progress=27.6% loss=0.101263 lr=8.662902e-05 lr_backbone=8.662902e-05 elapsed=14:05:11 avg_step=3.67s recent_step=2.57s eta=36:52:38 +2026-09-18 07:34:18 | INFO | train rank=0 | step=13830/50000 progress=27.7% loss=0.128475 lr=8.660650e-05 lr_backbone=8.660650e-05 elapsed=14:05:41 avg_step=3.67s recent_step=3.07s eta=36:51:46 +2026-09-18 07:34:58 | INFO | train rank=0 | step=13840/50000 progress=27.7% loss=0.122886 lr=8.658397e-05 lr_backbone=8.658397e-05 elapsed=14:06:22 avg_step=3.67s recent_step=4.03s eta=36:51:19 +2026-09-18 07:35:51 | INFO | train rank=0 | step=13850/50000 progress=27.7% loss=0.137768 lr=8.656142e-05 lr_backbone=8.656142e-05 elapsed=14:07:14 avg_step=3.67s recent_step=5.25s eta=36:51:24 +2026-09-18 07:36:15 | INFO | train rank=0 | step=13860/50000 progress=27.7% loss=0.140910 lr=8.653885e-05 lr_backbone=8.653885e-05 elapsed=14:07:39 avg_step=3.67s recent_step=2.43s eta=36:50:15 +2026-09-18 07:36:48 | INFO | train rank=0 | step=13870/50000 progress=27.7% loss=0.174760 lr=8.651627e-05 lr_backbone=8.651627e-05 elapsed=14:08:11 avg_step=3.67s recent_step=3.25s eta=36:49:27 +2026-09-18 07:37:21 | INFO | train rank=0 | step=13880/50000 progress=27.8% loss=0.157368 lr=8.649367e-05 lr_backbone=8.649367e-05 elapsed=14:08:44 avg_step=3.67s recent_step=3.34s eta=36:48:42 +2026-09-18 07:38:10 | INFO | train rank=0 | step=13890/50000 progress=27.8% loss=0.105279 lr=8.647106e-05 lr_backbone=8.647106e-05 elapsed=14:09:33 avg_step=3.67s recent_step=4.88s eta=36:48:36 +2026-09-18 07:38:48 | INFO | train rank=0 | step=13900/50000 progress=27.8% loss=0.104749 lr=8.644843e-05 lr_backbone=8.644843e-05 elapsed=14:10:11 avg_step=3.67s recent_step=3.82s eta=36:48:04 +2026-09-18 07:39:18 | INFO | train rank=0 | step=13910/50000 progress=27.8% loss=0.105931 lr=8.642579e-05 lr_backbone=8.642579e-05 elapsed=14:10:42 avg_step=3.67s recent_step=3.03s eta=36:47:10 +2026-09-18 07:39:49 | INFO | train rank=0 | step=13920/50000 progress=27.8% loss=0.121229 lr=8.640312e-05 lr_backbone=8.640312e-05 elapsed=14:11:13 avg_step=3.67s recent_step=3.11s eta=36:46:19 +2026-09-18 07:40:36 | INFO | train rank=0 | step=13930/50000 progress=27.9% loss=0.135721 lr=8.638045e-05 lr_backbone=8.638045e-05 elapsed=14:11:59 avg_step=3.67s recent_step=4.61s eta=36:46:07 +2026-09-18 07:41:10 | INFO | train rank=0 | step=13940/50000 progress=27.9% loss=0.179520 lr=8.635775e-05 lr_backbone=8.635775e-05 elapsed=14:12:34 avg_step=3.67s recent_step=3.48s eta=36:45:25 +2026-09-18 07:41:48 | INFO | train rank=0 | step=13950/50000 progress=27.9% loss=0.077632 lr=8.633504e-05 lr_backbone=8.633504e-05 elapsed=14:13:11 avg_step=3.67s recent_step=3.76s eta=36:44:51 +2026-09-18 07:42:21 | INFO | train rank=0 | step=13960/50000 progress=27.9% loss=0.099856 lr=8.631232e-05 lr_backbone=8.631232e-05 elapsed=14:13:44 avg_step=3.67s recent_step=3.31s eta=36:44:05 +2026-09-18 07:43:12 | INFO | train rank=0 | step=13970/50000 progress=27.9% loss=0.094741 lr=8.628958e-05 lr_backbone=8.628958e-05 elapsed=14:14:35 avg_step=3.67s recent_step=5.12s eta=36:44:05 +2026-09-18 07:43:43 | INFO | train rank=0 | step=13980/50000 progress=28.0% loss=0.118874 lr=8.626682e-05 lr_backbone=8.626682e-05 elapsed=14:15:07 avg_step=3.67s recent_step=3.10s eta=36:43:14 +2026-09-18 07:44:16 | INFO | train rank=0 | step=13990/50000 progress=28.0% loss=0.093232 lr=8.624405e-05 lr_backbone=8.624405e-05 elapsed=14:15:40 avg_step=3.67s recent_step=3.30s eta=36:42:28 +2026-09-18 07:44:45 | INFO | train rank=0 | step=14000/50000 progress=28.0% loss=0.109439 lr=8.622126e-05 lr_backbone=8.622126e-05 elapsed=14:16:08 avg_step=3.67s recent_step=2.85s eta=36:41:30 +2026-09-18 07:45:35 | INFO | train rank=0 | step=14010/50000 progress=28.0% loss=0.087878 lr=8.619846e-05 lr_backbone=8.619846e-05 elapsed=14:16:58 avg_step=3.67s recent_step=5.01s eta=36:41:28 +2026-09-18 07:46:02 | INFO | train rank=0 | step=14020/50000 progress=28.0% loss=0.107527 lr=8.617564e-05 lr_backbone=8.617564e-05 elapsed=14:17:25 avg_step=3.67s recent_step=2.71s eta=36:40:26 +2026-09-18 07:46:33 | INFO | train rank=0 | step=14030/50000 progress=28.1% loss=0.125524 lr=8.615280e-05 lr_backbone=8.615280e-05 elapsed=14:17:56 avg_step=3.67s recent_step=3.10s eta=36:39:35 +2026-09-18 07:47:09 | INFO | train rank=0 | step=14040/50000 progress=28.1% loss=0.086376 lr=8.612995e-05 lr_backbone=8.612995e-05 elapsed=14:18:32 avg_step=3.67s recent_step=3.62s eta=36:38:57 +2026-09-18 07:48:01 | INFO | train rank=0 | step=14050/50000 progress=28.1% loss=0.097038 lr=8.610708e-05 lr_backbone=8.610708e-05 elapsed=14:19:24 avg_step=3.67s recent_step=5.21s eta=36:39:00 +2026-09-18 07:48:33 | INFO | train rank=0 | step=14060/50000 progress=28.1% loss=0.102944 lr=8.608420e-05 lr_backbone=8.608420e-05 elapsed=14:19:57 avg_step=3.67s recent_step=3.23s eta=36:38:12 +2026-09-18 07:49:03 | INFO | train rank=0 | step=14070/50000 progress=28.1% loss=0.094852 lr=8.606130e-05 lr_backbone=8.606130e-05 elapsed=14:20:26 avg_step=3.67s recent_step=2.91s eta=36:37:16 +2026-09-18 07:49:31 | INFO | train rank=0 | step=14080/50000 progress=28.2% loss=0.075793 lr=8.603838e-05 lr_backbone=8.603838e-05 elapsed=14:20:54 avg_step=3.67s recent_step=2.84s eta=36:36:18 +2026-09-18 07:50:26 | INFO | train rank=0 | step=14090/50000 progress=28.2% loss=0.108549 lr=8.601545e-05 lr_backbone=8.601545e-05 elapsed=14:21:49 avg_step=3.67s recent_step=5.50s eta=36:36:28 +2026-09-18 07:50:53 | INFO | train rank=0 | step=14100/50000 progress=28.2% loss=0.082370 lr=8.599250e-05 lr_backbone=8.599250e-05 elapsed=14:22:16 avg_step=3.67s recent_step=2.69s eta=36:35:26 +2026-09-18 07:51:23 | INFO | train rank=0 | step=14110/50000 progress=28.2% loss=0.126724 lr=8.596954e-05 lr_backbone=8.596954e-05 elapsed=14:22:46 avg_step=3.67s recent_step=3.00s eta=36:34:32 +2026-09-18 07:51:56 | INFO | train rank=0 | step=14120/50000 progress=28.2% loss=0.112223 lr=8.594656e-05 lr_backbone=8.594656e-05 elapsed=14:23:19 avg_step=3.67s recent_step=3.29s eta=36:33:46 +2026-09-18 07:52:51 | INFO | train rank=0 | step=14130/50000 progress=28.3% loss=0.111109 lr=8.592357e-05 lr_backbone=8.592357e-05 elapsed=14:24:14 avg_step=3.67s recent_step=5.53s eta=36:33:57 +2026-09-18 07:53:23 | INFO | train rank=0 | step=14140/50000 progress=28.3% loss=0.118188 lr=8.590056e-05 lr_backbone=8.590056e-05 elapsed=14:24:46 avg_step=3.67s recent_step=3.17s eta=36:33:07 +2026-09-18 07:53:52 | INFO | train rank=0 | step=14150/50000 progress=28.3% loss=0.122191 lr=8.587754e-05 lr_backbone=8.587754e-05 elapsed=14:25:16 avg_step=3.67s recent_step=2.95s eta=36:32:12 +2026-09-18 07:54:20 | INFO | train rank=0 | step=14160/50000 progress=28.3% loss=0.086019 lr=8.585449e-05 lr_backbone=8.585449e-05 elapsed=14:25:43 avg_step=3.67s recent_step=2.75s eta=36:31:12 +2026-09-18 07:55:16 | INFO | train rank=0 | step=14170/50000 progress=28.3% loss=0.071989 lr=8.583144e-05 lr_backbone=8.583144e-05 elapsed=14:26:40 avg_step=3.67s recent_step=5.67s eta=36:31:26 +2026-09-18 07:55:48 | INFO | train rank=0 | step=14180/50000 progress=28.4% loss=0.150405 lr=8.580837e-05 lr_backbone=8.580837e-05 elapsed=14:27:11 avg_step=3.67s recent_step=3.15s eta=36:30:36 +2026-09-18 07:56:20 | INFO | train rank=0 | step=14190/50000 progress=28.4% loss=0.081240 lr=8.578528e-05 lr_backbone=8.578528e-05 elapsed=14:27:43 avg_step=3.67s recent_step=3.21s eta=36:29:48 +2026-09-18 07:56:52 | INFO | train rank=0 | step=14200/50000 progress=28.4% loss=0.093790 lr=8.576217e-05 lr_backbone=8.576217e-05 elapsed=14:28:16 avg_step=3.67s recent_step=3.25s eta=36:29:01 +2026-09-18 07:57:40 | INFO | train rank=0 | step=14210/50000 progress=28.4% loss=0.093133 lr=8.573906e-05 lr_backbone=8.573906e-05 elapsed=14:29:03 avg_step=3.67s recent_step=4.75s eta=36:28:51 +2026-09-18 07:58:08 | INFO | train rank=0 | step=14220/50000 progress=28.4% loss=0.086277 lr=8.571592e-05 lr_backbone=8.571592e-05 elapsed=14:29:32 avg_step=3.67s recent_step=2.84s eta=36:27:54 +2026-09-18 07:58:42 | INFO | train rank=0 | step=14230/50000 progress=28.5% loss=0.077539 lr=8.569277e-05 lr_backbone=8.569277e-05 elapsed=14:30:06 avg_step=3.67s recent_step=3.40s eta=36:27:10 +2026-09-18 07:59:20 | INFO | train rank=0 | step=14240/50000 progress=28.5% loss=0.100147 lr=8.566960e-05 lr_backbone=8.566960e-05 elapsed=14:30:44 avg_step=3.67s recent_step=3.79s eta=36:26:37 +2026-09-18 08:00:11 | INFO | train rank=0 | step=14250/50000 progress=28.5% loss=0.097259 lr=8.564642e-05 lr_backbone=8.564642e-05 elapsed=14:31:34 avg_step=3.67s recent_step=5.03s eta=36:26:34 +2026-09-18 08:00:37 | INFO | train rank=0 | step=14260/50000 progress=28.5% loss=0.093186 lr=8.562323e-05 lr_backbone=8.562323e-05 elapsed=14:32:00 avg_step=3.67s recent_step=2.60s eta=36:25:31 +2026-09-18 08:01:14 | INFO | train rank=0 | step=14270/50000 progress=28.5% loss=0.094484 lr=8.560001e-05 lr_backbone=8.560001e-05 elapsed=14:32:37 avg_step=3.67s recent_step=3.69s eta=36:24:55 +2026-09-18 08:01:47 | INFO | train rank=0 | step=14280/50000 progress=28.6% loss=0.094429 lr=8.557678e-05 lr_backbone=8.557678e-05 elapsed=14:33:10 avg_step=3.67s recent_step=3.31s eta=36:24:09 +2026-09-18 08:02:34 | INFO | train rank=0 | step=14290/50000 progress=28.6% loss=0.122270 lr=8.555354e-05 lr_backbone=8.555354e-05 elapsed=14:33:57 avg_step=3.67s recent_step=4.72s eta=36:23:59 +2026-09-18 08:03:11 | INFO | train rank=0 | step=14300/50000 progress=28.6% loss=0.089760 lr=8.553028e-05 lr_backbone=8.553028e-05 elapsed=14:34:34 avg_step=3.67s recent_step=3.68s eta=36:23:22 +2026-09-18 08:03:41 | INFO | train rank=0 | step=14310/50000 progress=28.6% loss=0.102339 lr=8.550701e-05 lr_backbone=8.550701e-05 elapsed=14:35:04 avg_step=3.67s recent_step=3.05s eta=36:22:30 +2026-09-18 08:04:13 | INFO | train rank=0 | step=14320/50000 progress=28.6% loss=0.100604 lr=8.548371e-05 lr_backbone=8.548371e-05 elapsed=14:35:36 avg_step=3.67s recent_step=3.21s eta=36:21:42 +2026-09-18 08:05:06 | INFO | train rank=0 | step=14330/50000 progress=28.7% loss=0.083347 lr=8.546041e-05 lr_backbone=8.546041e-05 elapsed=14:36:29 avg_step=3.67s recent_step=5.28s eta=36:21:45 +2026-09-18 08:05:35 | INFO | train rank=0 | step=14340/50000 progress=28.7% loss=0.097834 lr=8.543709e-05 lr_backbone=8.543709e-05 elapsed=14:36:59 avg_step=3.67s recent_step=2.92s eta=36:20:50 +2026-09-18 08:06:02 | INFO | train rank=0 | step=14350/50000 progress=28.7% loss=0.099241 lr=8.541375e-05 lr_backbone=8.541375e-05 elapsed=14:37:26 avg_step=3.67s recent_step=2.73s eta=36:19:50 +2026-09-18 08:06:40 | INFO | train rank=0 | step=14360/50000 progress=28.7% loss=0.106465 lr=8.539040e-05 lr_backbone=8.539040e-05 elapsed=14:38:03 avg_step=3.67s recent_step=3.73s eta=36:19:15 +2026-09-18 08:07:24 | INFO | train rank=0 | step=14370/50000 progress=28.7% loss=0.100392 lr=8.536703e-05 lr_backbone=8.536703e-05 elapsed=14:38:48 avg_step=3.67s recent_step=4.46s eta=36:18:57 +2026-09-18 08:08:03 | INFO | train rank=0 | step=14380/50000 progress=28.8% loss=0.153537 lr=8.534365e-05 lr_backbone=8.534365e-05 elapsed=14:39:26 avg_step=3.67s recent_step=3.82s eta=36:18:25 +2026-09-18 08:08:36 | INFO | train rank=0 | step=14390/50000 progress=28.8% loss=0.122363 lr=8.532025e-05 lr_backbone=8.532025e-05 elapsed=14:39:59 avg_step=3.67s recent_step=3.34s eta=36:17:40 +2026-09-18 08:09:06 | INFO | train rank=0 | step=14400/50000 progress=28.8% loss=0.125615 lr=8.529683e-05 lr_backbone=8.529683e-05 elapsed=14:40:29 avg_step=3.67s recent_step=2.98s eta=36:16:46 +2026-09-18 08:09:58 | INFO | train rank=0 | step=14410/50000 progress=28.8% loss=0.137345 lr=8.527340e-05 lr_backbone=8.527340e-05 elapsed=14:41:22 avg_step=3.67s recent_step=5.25s eta=36:16:48 +2026-09-18 08:10:29 | INFO | train rank=0 | step=14420/50000 progress=28.8% loss=0.087750 lr=8.524996e-05 lr_backbone=8.524996e-05 elapsed=14:41:52 avg_step=3.67s recent_step=3.06s eta=36:15:57 +2026-09-18 08:11:02 | INFO | train rank=0 | step=14430/50000 progress=28.9% loss=0.094258 lr=8.522650e-05 lr_backbone=8.522650e-05 elapsed=14:42:25 avg_step=3.67s recent_step=3.29s eta=36:15:11 +2026-09-18 08:11:37 | INFO | train rank=0 | step=14440/50000 progress=28.9% loss=0.110032 lr=8.520302e-05 lr_backbone=8.520302e-05 elapsed=14:43:00 avg_step=3.67s recent_step=3.51s eta=36:14:30 +2026-09-18 08:12:24 | INFO | train rank=0 | step=14450/50000 progress=28.9% loss=0.116489 lr=8.517953e-05 lr_backbone=8.517953e-05 elapsed=14:43:47 avg_step=3.67s recent_step=4.73s eta=36:14:19 +2026-09-18 08:12:57 | INFO | train rank=0 | step=14460/50000 progress=28.9% loss=0.095964 lr=8.515602e-05 lr_backbone=8.515602e-05 elapsed=14:44:20 avg_step=3.67s recent_step=3.25s eta=36:13:32 +2026-09-18 08:13:29 | INFO | train rank=0 | step=14470/50000 progress=28.9% loss=0.078545 lr=8.513250e-05 lr_backbone=8.513250e-05 elapsed=14:44:52 avg_step=3.67s recent_step=3.21s eta=36:12:44 +2026-09-18 08:14:00 | INFO | train rank=0 | step=14480/50000 progress=29.0% loss=0.094744 lr=8.510896e-05 lr_backbone=8.510896e-05 elapsed=14:45:23 avg_step=3.67s recent_step=3.09s eta=36:11:54 +2026-09-18 08:14:58 | INFO | train rank=0 | step=14490/50000 progress=29.0% loss=0.095604 lr=8.508541e-05 lr_backbone=8.508541e-05 elapsed=14:46:21 avg_step=3.67s recent_step=5.78s eta=36:12:09 +2026-09-18 08:15:29 | INFO | train rank=0 | step=14500/50000 progress=29.0% loss=0.111272 lr=8.506184e-05 lr_backbone=8.506184e-05 elapsed=14:46:52 avg_step=3.67s recent_step=3.14s eta=36:11:19 +2026-09-18 08:16:03 | INFO | train rank=0 | step=14510/50000 progress=29.0% loss=0.120562 lr=8.503826e-05 lr_backbone=8.503826e-05 elapsed=14:47:26 avg_step=3.67s recent_step=3.41s eta=36:10:36 +2026-09-18 08:16:34 | INFO | train rank=0 | step=14520/50000 progress=29.0% loss=0.092021 lr=8.501466e-05 lr_backbone=8.501466e-05 elapsed=14:47:58 avg_step=3.67s recent_step=3.13s eta=36:09:46 +2026-09-18 08:17:19 | INFO | train rank=0 | step=14530/50000 progress=29.1% loss=0.076266 lr=8.499104e-05 lr_backbone=8.499104e-05 elapsed=14:48:43 avg_step=3.67s recent_step=4.51s eta=36:09:30 +2026-09-18 08:17:49 | INFO | train rank=0 | step=14540/50000 progress=29.1% loss=0.100197 lr=8.496741e-05 lr_backbone=8.496741e-05 elapsed=14:49:12 avg_step=3.67s recent_step=2.92s eta=36:08:35 +2026-09-18 08:18:23 | INFO | train rank=0 | step=14550/50000 progress=29.1% loss=0.093923 lr=8.494377e-05 lr_backbone=8.494377e-05 elapsed=14:49:46 avg_step=3.67s recent_step=3.39s eta=36:07:51 +2026-09-18 08:19:02 | INFO | train rank=0 | step=14560/50000 progress=29.1% loss=0.116236 lr=8.492011e-05 lr_backbone=8.492011e-05 elapsed=14:50:25 avg_step=3.67s recent_step=3.96s eta=36:07:22 +2026-09-18 08:19:50 | INFO | train rank=0 | step=14570/50000 progress=29.1% loss=0.084774 lr=8.489643e-05 lr_backbone=8.489643e-05 elapsed=14:51:13 avg_step=3.67s recent_step=4.78s eta=36:07:12 +2026-09-18 08:20:25 | INFO | train rank=0 | step=14580/50000 progress=29.2% loss=0.125254 lr=8.487274e-05 lr_backbone=8.487274e-05 elapsed=14:51:48 avg_step=3.67s recent_step=3.46s eta=36:06:30 +2026-09-18 08:20:57 | INFO | train rank=0 | step=14590/50000 progress=29.2% loss=0.116736 lr=8.484903e-05 lr_backbone=8.484903e-05 elapsed=14:52:21 avg_step=3.67s recent_step=3.30s eta=36:05:44 +2026-09-18 08:21:26 | INFO | train rank=0 | step=14600/50000 progress=29.2% loss=0.127659 lr=8.482531e-05 lr_backbone=8.482531e-05 elapsed=14:52:49 avg_step=3.67s recent_step=2.81s eta=36:04:47 +2026-09-18 08:22:07 | INFO | train rank=0 | step=14610/50000 progress=29.2% loss=0.118994 lr=8.480158e-05 lr_backbone=8.480158e-05 elapsed=14:53:30 avg_step=3.67s recent_step=4.16s eta=36:04:22 +2026-09-18 08:22:45 | INFO | train rank=0 | step=14620/50000 progress=29.2% loss=0.103251 lr=8.477782e-05 lr_backbone=8.477782e-05 elapsed=14:54:08 avg_step=3.67s recent_step=3.79s eta=36:03:48 +2026-09-18 08:23:18 | INFO | train rank=0 | step=14630/50000 progress=29.3% loss=0.089630 lr=8.475406e-05 lr_backbone=8.475406e-05 elapsed=14:54:42 avg_step=3.67s recent_step=3.32s eta=36:03:03 +2026-09-18 08:23:44 | INFO | train rank=0 | step=14640/50000 progress=29.3% loss=0.098106 lr=8.473028e-05 lr_backbone=8.473028e-05 elapsed=14:55:07 avg_step=3.67s recent_step=2.55s eta=36:01:59 +2026-09-18 08:24:38 | INFO | train rank=0 | step=14650/50000 progress=29.3% loss=0.084277 lr=8.470648e-05 lr_backbone=8.470648e-05 elapsed=14:56:02 avg_step=3.67s recent_step=5.47s eta=36:02:06 +2026-09-18 08:25:09 | INFO | train rank=0 | step=14660/50000 progress=29.3% loss=0.092230 lr=8.468267e-05 lr_backbone=8.468267e-05 elapsed=14:56:32 avg_step=3.67s recent_step=3.02s eta=36:01:14 +2026-09-18 08:25:40 | INFO | train rank=0 | step=14670/50000 progress=29.3% loss=0.108924 lr=8.465884e-05 lr_backbone=8.465884e-05 elapsed=14:57:03 avg_step=3.67s recent_step=3.14s eta=36:00:24 +2026-09-18 08:26:16 | INFO | train rank=0 | step=14680/50000 progress=29.4% loss=0.083241 lr=8.463499e-05 lr_backbone=8.463499e-05 elapsed=14:57:39 avg_step=3.67s recent_step=3.59s eta=35:59:46 +2026-09-18 08:27:08 | INFO | train rank=0 | step=14690/50000 progress=29.4% loss=0.121011 lr=8.461114e-05 lr_backbone=8.461114e-05 elapsed=14:58:32 avg_step=3.67s recent_step=5.26s eta=35:59:47 +2026-09-18 08:27:39 | INFO | train rank=0 | step=14700/50000 progress=29.4% loss=0.107022 lr=8.458726e-05 lr_backbone=8.458726e-05 elapsed=14:59:03 avg_step=3.67s recent_step=3.08s eta=35:58:56 +2026-09-18 08:28:08 | INFO | train rank=0 | step=14710/50000 progress=29.4% loss=0.086297 lr=8.456337e-05 lr_backbone=8.456337e-05 elapsed=14:59:31 avg_step=3.67s recent_step=2.86s eta=35:58:00 +2026-09-18 08:28:37 | INFO | train rank=0 | step=14720/50000 progress=29.4% loss=0.091467 lr=8.453947e-05 lr_backbone=8.453947e-05 elapsed=15:00:01 avg_step=3.67s recent_step=2.94s eta=35:57:06 +2026-09-18 08:29:17 | INFO | train rank=0 | step=14730/50000 progress=29.5% loss=0.071343 lr=8.451555e-05 lr_backbone=8.451555e-05 elapsed=15:00:40 avg_step=3.67s recent_step=3.96s eta=35:56:36 +2026-09-18 08:29:48 | INFO | train rank=0 | step=14740/50000 progress=29.5% loss=0.103591 lr=8.449162e-05 lr_backbone=8.449162e-05 elapsed=15:01:11 avg_step=3.67s recent_step=3.08s eta=35:55:45 +2026-09-18 08:30:21 | INFO | train rank=0 | step=14750/50000 progress=29.5% loss=0.090176 lr=8.446767e-05 lr_backbone=8.446767e-05 elapsed=15:01:44 avg_step=3.67s recent_step=3.31s eta=35:55:00 +2026-09-18 08:30:50 | INFO | train rank=0 | step=14760/50000 progress=29.5% loss=0.081289 lr=8.444371e-05 lr_backbone=8.444371e-05 elapsed=15:02:13 avg_step=3.67s recent_step=2.92s eta=35:54:06 +2026-09-18 08:31:42 | INFO | train rank=0 | step=14770/50000 progress=29.5% loss=0.153772 lr=8.441973e-05 lr_backbone=8.441973e-05 elapsed=15:03:05 avg_step=3.67s recent_step=5.16s eta=35:54:05 +2026-09-18 08:32:10 | INFO | train rank=0 | step=14780/50000 progress=29.6% loss=0.094468 lr=8.439573e-05 lr_backbone=8.439573e-05 elapsed=15:03:33 avg_step=3.67s recent_step=2.85s eta=35:53:08 +2026-09-18 08:32:45 | INFO | train rank=0 | step=14790/50000 progress=29.6% loss=0.092977 lr=8.437172e-05 lr_backbone=8.437172e-05 elapsed=15:04:08 avg_step=3.67s recent_step=3.46s eta=35:52:27 +2026-09-18 08:33:20 | INFO | train rank=0 | step=14800/50000 progress=29.6% loss=0.117946 lr=8.434770e-05 lr_backbone=8.434770e-05 elapsed=15:04:43 avg_step=3.67s recent_step=3.49s eta=35:51:46 +2026-09-18 08:34:11 | INFO | train rank=0 | step=14810/50000 progress=29.6% loss=0.089435 lr=8.432366e-05 lr_backbone=8.432366e-05 elapsed=15:05:34 avg_step=3.67s recent_step=5.13s eta=35:51:44 +2026-09-18 08:34:48 | INFO | train rank=0 | step=14820/50000 progress=29.6% loss=0.097981 lr=8.429961e-05 lr_backbone=8.429961e-05 elapsed=15:06:11 avg_step=3.67s recent_step=3.70s eta=35:51:08 +2026-09-18 08:35:22 | INFO | train rank=0 | step=14830/50000 progress=29.7% loss=0.110817 lr=8.427554e-05 lr_backbone=8.427554e-05 elapsed=15:06:45 avg_step=3.67s recent_step=3.42s eta=35:50:25 +2026-09-18 08:35:54 | INFO | train rank=0 | step=14840/50000 progress=29.7% loss=0.142040 lr=8.425145e-05 lr_backbone=8.425145e-05 elapsed=15:07:18 avg_step=3.67s recent_step=3.23s eta=35:49:38 +2026-09-18 08:36:49 | INFO | train rank=0 | step=14850/50000 progress=29.7% loss=0.100706 lr=8.422736e-05 lr_backbone=8.422736e-05 elapsed=15:08:12 avg_step=3.67s recent_step=5.47s eta=35:49:44 +2026-09-18 08:37:19 | INFO | train rank=0 | step=14860/50000 progress=29.7% loss=0.086459 lr=8.420324e-05 lr_backbone=8.420324e-05 elapsed=15:08:42 avg_step=3.67s recent_step=2.96s eta=35:48:51 +2026-09-18 08:37:50 | INFO | train rank=0 | step=14870/50000 progress=29.7% loss=0.082680 lr=8.417911e-05 lr_backbone=8.417911e-05 elapsed=15:09:14 avg_step=3.67s recent_step=3.18s eta=35:48:02 +2026-09-18 08:38:13 | INFO | train rank=0 | step=14880/50000 progress=29.8% loss=0.112881 lr=8.415497e-05 lr_backbone=8.415497e-05 elapsed=15:09:36 avg_step=3.67s recent_step=2.28s eta=35:46:53 +2026-09-18 08:39:02 | INFO | train rank=0 | step=14890/50000 progress=29.8% loss=0.156828 lr=8.413081e-05 lr_backbone=8.413081e-05 elapsed=15:10:26 avg_step=3.67s recent_step=4.92s eta=35:46:46 +2026-09-18 08:39:33 | INFO | train rank=0 | step=14900/50000 progress=29.8% loss=0.107673 lr=8.410664e-05 lr_backbone=8.410664e-05 elapsed=15:10:57 avg_step=3.67s recent_step=3.10s eta=35:45:56 +2026-09-18 08:40:12 | INFO | train rank=0 | step=14910/50000 progress=29.8% loss=0.094149 lr=8.408245e-05 lr_backbone=8.408245e-05 elapsed=15:11:35 avg_step=3.67s recent_step=3.88s eta=35:45:24 +2026-09-18 08:40:46 | INFO | train rank=0 | step=14920/50000 progress=29.8% loss=0.098692 lr=8.405824e-05 lr_backbone=8.405824e-05 elapsed=15:12:09 avg_step=3.67s recent_step=3.36s eta=35:44:40 +2026-09-18 08:41:40 | INFO | train rank=0 | step=14930/50000 progress=29.9% loss=0.119454 lr=8.403402e-05 lr_backbone=8.403402e-05 elapsed=15:13:04 avg_step=3.67s recent_step=5.48s eta=35:44:46 +2026-09-18 08:42:11 | INFO | train rank=0 | step=14940/50000 progress=29.9% loss=0.116304 lr=8.400979e-05 lr_backbone=8.400979e-05 elapsed=15:13:35 avg_step=3.67s recent_step=3.09s eta=35:43:55 +2026-09-18 08:42:39 | INFO | train rank=0 | step=14950/50000 progress=29.9% loss=0.090902 lr=8.398554e-05 lr_backbone=8.398554e-05 elapsed=15:14:02 avg_step=3.67s recent_step=2.73s eta=35:42:57 +2026-09-18 08:43:11 | INFO | train rank=0 | step=14960/50000 progress=29.9% loss=0.109159 lr=8.396128e-05 lr_backbone=8.396128e-05 elapsed=15:14:34 avg_step=3.67s recent_step=3.23s eta=35:42:10 +2026-09-18 08:43:53 | INFO | train rank=0 | step=14970/50000 progress=29.9% loss=0.092650 lr=8.393700e-05 lr_backbone=8.393700e-05 elapsed=15:15:16 avg_step=3.67s recent_step=4.22s eta=35:41:46 +2026-09-18 08:44:30 | INFO | train rank=0 | step=14980/50000 progress=30.0% loss=0.103271 lr=8.391271e-05 lr_backbone=8.391271e-05 elapsed=15:15:53 avg_step=3.67s recent_step=3.69s eta=35:41:10 +2026-09-18 08:45:00 | INFO | train rank=0 | step=14990/50000 progress=30.0% loss=0.088652 lr=8.388840e-05 lr_backbone=8.388840e-05 elapsed=15:16:24 avg_step=3.67s recent_step=3.03s eta=35:40:18 +2026-09-18 08:45:38 | INFO | train rank=0 | step=15000/50000 progress=30.0% loss=0.089762 lr=8.386408e-05 lr_backbone=8.386408e-05 elapsed=15:17:02 avg_step=3.67s recent_step=3.79s eta=35:39:44 +2026-09-18 08:46:42 | INFO | train rank=0 | step=15010/50000 progress=30.0% loss=0.123798 lr=8.383974e-05 lr_backbone=8.383974e-05 elapsed=15:18:05 avg_step=3.67s recent_step=6.34s eta=35:40:10 +2026-09-18 08:47:05 | INFO | train rank=0 | step=15020/50000 progress=30.0% loss=0.089419 lr=8.381539e-05 lr_backbone=8.381539e-05 elapsed=15:18:28 avg_step=3.67s recent_step=2.31s eta=35:39:02 +2026-09-18 08:47:31 | INFO | train rank=0 | step=15030/50000 progress=30.1% loss=0.117936 lr=8.379102e-05 lr_backbone=8.379102e-05 elapsed=15:18:54 avg_step=3.67s recent_step=2.58s eta=35:38:00 +2026-09-18 08:48:06 | INFO | train rank=0 | step=15040/50000 progress=30.1% loss=0.099551 lr=8.376664e-05 lr_backbone=8.376664e-05 elapsed=15:19:29 avg_step=3.67s recent_step=3.54s eta=35:37:20 +2026-09-18 08:49:00 | INFO | train rank=0 | step=15050/50000 progress=30.1% loss=0.085970 lr=8.374224e-05 lr_backbone=8.374224e-05 elapsed=15:20:23 avg_step=3.67s recent_step=5.36s eta=35:37:23 +2026-09-18 08:49:32 | INFO | train rank=0 | step=15060/50000 progress=30.1% loss=0.082445 lr=8.371783e-05 lr_backbone=8.371783e-05 elapsed=15:20:55 avg_step=3.67s recent_step=3.22s eta=35:36:36 +2026-09-18 08:50:10 | INFO | train rank=0 | step=15070/50000 progress=30.1% loss=0.108171 lr=8.369341e-05 lr_backbone=8.369341e-05 elapsed=15:21:33 avg_step=3.67s recent_step=3.80s eta=35:36:02 +2026-09-18 08:50:33 | INFO | train rank=0 | step=15080/50000 progress=30.2% loss=0.115983 lr=8.366897e-05 lr_backbone=8.366897e-05 elapsed=15:21:56 avg_step=3.67s recent_step=2.26s eta=35:34:53 +2026-09-18 08:51:21 | INFO | train rank=0 | step=15090/50000 progress=30.2% loss=0.118910 lr=8.364451e-05 lr_backbone=8.364451e-05 elapsed=15:22:44 avg_step=3.67s recent_step=4.86s eta=35:34:44 +2026-09-18 08:51:52 | INFO | train rank=0 | step=15100/50000 progress=30.2% loss=0.106303 lr=8.362004e-05 lr_backbone=8.362004e-05 elapsed=15:23:16 avg_step=3.67s recent_step=3.13s eta=35:33:54 +2026-09-18 08:52:18 | INFO | train rank=0 | step=15110/50000 progress=30.2% loss=0.089549 lr=8.359556e-05 lr_backbone=8.359556e-05 elapsed=15:23:41 avg_step=3.67s recent_step=2.58s eta=35:32:53 +2026-09-18 08:52:56 | INFO | train rank=0 | step=15120/50000 progress=30.2% loss=0.093642 lr=8.357106e-05 lr_backbone=8.357106e-05 elapsed=15:24:19 avg_step=3.67s recent_step=3.77s eta=35:32:18 +2026-09-18 08:53:47 | INFO | train rank=0 | step=15130/50000 progress=30.3% loss=0.098418 lr=8.354654e-05 lr_backbone=8.354654e-05 elapsed=15:25:10 avg_step=3.67s recent_step=5.07s eta=35:32:14 +2026-09-18 08:54:14 | INFO | train rank=0 | step=15140/50000 progress=30.3% loss=0.147309 lr=8.352201e-05 lr_backbone=8.352201e-05 elapsed=15:25:37 avg_step=3.67s recent_step=2.72s eta=35:31:15 +2026-09-18 08:54:40 | INFO | train rank=0 | step=15150/50000 progress=30.3% loss=0.078491 lr=8.349747e-05 lr_backbone=8.349747e-05 elapsed=15:26:04 avg_step=3.67s recent_step=2.65s eta=35:30:15 +2026-09-18 08:55:12 | INFO | train rank=0 | step=15160/50000 progress=30.3% loss=0.109642 lr=8.347291e-05 lr_backbone=8.347291e-05 elapsed=15:26:35 avg_step=3.67s recent_step=3.13s eta=35:29:26 +2026-09-18 08:56:03 | INFO | train rank=0 | step=15170/50000 progress=30.3% loss=0.092754 lr=8.344834e-05 lr_backbone=8.344834e-05 elapsed=15:27:27 avg_step=3.67s recent_step=5.16s eta=35:29:24 +2026-09-18 08:56:33 | INFO | train rank=0 | step=15180/50000 progress=30.4% loss=0.118792 lr=8.342375e-05 lr_backbone=8.342375e-05 elapsed=15:27:56 avg_step=3.67s recent_step=2.93s eta=35:28:30 +2026-09-18 08:57:08 | INFO | train rank=0 | step=15190/50000 progress=30.4% loss=0.143153 lr=8.339915e-05 lr_backbone=8.339915e-05 elapsed=15:28:31 avg_step=3.67s recent_step=3.53s eta=35:27:51 +2026-09-18 08:57:44 | INFO | train rank=0 | step=15200/50000 progress=30.4% loss=0.080700 lr=8.337453e-05 lr_backbone=8.337453e-05 elapsed=15:29:07 avg_step=3.67s recent_step=3.58s eta=35:27:12 +2026-09-18 08:58:36 | INFO | train rank=0 | step=15210/50000 progress=30.4% loss=0.081662 lr=8.334990e-05 lr_backbone=8.334990e-05 elapsed=15:29:59 avg_step=3.67s recent_step=5.24s eta=35:27:11 +2026-09-18 08:59:14 | INFO | train rank=0 | step=15220/50000 progress=30.4% loss=0.102828 lr=8.332525e-05 lr_backbone=8.332525e-05 elapsed=15:30:38 avg_step=3.67s recent_step=3.83s eta=35:26:38 +2026-09-18 08:59:43 | INFO | train rank=0 | step=15230/50000 progress=30.5% loss=0.088551 lr=8.330059e-05 lr_backbone=8.330059e-05 elapsed=15:31:07 avg_step=3.67s recent_step=2.88s eta=35:25:44 +2026-09-18 09:00:17 | INFO | train rank=0 | step=15240/50000 progress=30.5% loss=0.111925 lr=8.327592e-05 lr_backbone=8.327592e-05 elapsed=15:31:41 avg_step=3.67s recent_step=3.39s eta=35:25:01 +2026-09-18 09:01:08 | INFO | train rank=0 | step=15250/50000 progress=30.5% loss=0.112710 lr=8.325123e-05 lr_backbone=8.325123e-05 elapsed=15:32:31 avg_step=3.67s recent_step=5.06s eta=35:24:56 +2026-09-18 09:01:40 | INFO | train rank=0 | step=15260/50000 progress=30.5% loss=0.110594 lr=8.322652e-05 lr_backbone=8.322652e-05 elapsed=15:33:03 avg_step=3.67s recent_step=3.20s eta=35:24:08 +2026-09-18 09:02:02 | INFO | train rank=0 | step=15270/50000 progress=30.5% loss=0.094882 lr=8.320181e-05 lr_backbone=8.320181e-05 elapsed=15:33:25 avg_step=3.67s recent_step=2.23s eta=35:22:59 +2026-09-18 09:02:40 | INFO | train rank=0 | step=15280/50000 progress=30.6% loss=0.092642 lr=8.317707e-05 lr_backbone=8.317707e-05 elapsed=15:34:04 avg_step=3.67s recent_step=3.82s eta=35:22:26 +2026-09-18 09:03:36 | INFO | train rank=0 | step=15290/50000 progress=30.6% loss=0.085156 lr=8.315232e-05 lr_backbone=8.315232e-05 elapsed=15:35:00 avg_step=3.67s recent_step=5.61s eta=35:22:33 +2026-09-18 09:04:06 | INFO | train rank=0 | step=15300/50000 progress=30.6% loss=0.093042 lr=8.312756e-05 lr_backbone=8.312756e-05 elapsed=15:35:29 avg_step=3.67s recent_step=2.95s eta=35:21:40 +2026-09-18 09:04:38 | INFO | train rank=0 | step=15310/50000 progress=30.6% loss=0.103005 lr=8.310279e-05 lr_backbone=8.310279e-05 elapsed=15:36:02 avg_step=3.67s recent_step=3.22s eta=35:20:54 +2026-09-18 09:05:16 | INFO | train rank=0 | step=15320/50000 progress=30.6% loss=0.095627 lr=8.307799e-05 lr_backbone=8.307799e-05 elapsed=15:36:39 avg_step=3.67s recent_step=3.77s eta=35:20:19 +2026-09-18 09:05:56 | INFO | train rank=0 | step=15330/50000 progress=30.7% loss=0.113666 lr=8.305319e-05 lr_backbone=8.305319e-05 elapsed=15:37:20 avg_step=3.67s recent_step=4.04s eta=35:19:51 +2026-09-18 09:06:33 | INFO | train rank=0 | step=15340/50000 progress=30.7% loss=0.115238 lr=8.302837e-05 lr_backbone=8.302837e-05 elapsed=15:37:56 avg_step=3.67s recent_step=3.63s eta=35:19:14 +2026-09-18 09:07:01 | INFO | train rank=0 | step=15350/50000 progress=30.7% loss=0.106310 lr=8.300353e-05 lr_backbone=8.300353e-05 elapsed=15:38:25 avg_step=3.67s recent_step=2.87s eta=35:18:19 +2026-09-18 09:07:39 | INFO | train rank=0 | step=15360/50000 progress=30.7% loss=0.077984 lr=8.297868e-05 lr_backbone=8.297868e-05 elapsed=15:39:02 avg_step=3.67s recent_step=3.75s eta=35:17:44 +2026-09-18 09:08:16 | INFO | train rank=0 | step=15370/50000 progress=30.7% loss=0.094356 lr=8.295382e-05 lr_backbone=8.295382e-05 elapsed=15:39:39 avg_step=3.67s recent_step=3.71s eta=35:17:08 +2026-09-18 09:09:03 | INFO | train rank=0 | step=15380/50000 progress=30.8% loss=0.129489 lr=8.292894e-05 lr_backbone=8.292894e-05 elapsed=15:40:26 avg_step=3.67s recent_step=4.69s eta=35:16:54 +2026-09-18 09:09:32 | INFO | train rank=0 | step=15390/50000 progress=30.8% loss=0.100433 lr=8.290405e-05 lr_backbone=8.290405e-05 elapsed=15:40:55 avg_step=3.67s recent_step=2.88s eta=35:16:00 +2026-09-18 09:10:13 | INFO | train rank=0 | step=15400/50000 progress=30.8% loss=0.105270 lr=8.287914e-05 lr_backbone=8.287914e-05 elapsed=15:41:36 avg_step=3.67s recent_step=4.13s eta=35:15:34 +2026-09-18 09:10:47 | INFO | train rank=0 | step=15410/50000 progress=30.8% loss=0.139086 lr=8.285422e-05 lr_backbone=8.285422e-05 elapsed=15:42:11 avg_step=3.67s recent_step=3.45s eta=35:14:52 +2026-09-18 09:11:29 | INFO | train rank=0 | step=15420/50000 progress=30.8% loss=0.109906 lr=8.282929e-05 lr_backbone=8.282929e-05 elapsed=15:42:52 avg_step=3.67s recent_step=4.13s eta=35:14:26 +2026-09-18 09:11:57 | INFO | train rank=0 | step=15430/50000 progress=30.9% loss=0.090101 lr=8.280434e-05 lr_backbone=8.280434e-05 elapsed=15:43:21 avg_step=3.67s recent_step=2.86s eta=35:13:31 +2026-09-18 09:12:41 | INFO | train rank=0 | step=15440/50000 progress=30.9% loss=0.089962 lr=8.277937e-05 lr_backbone=8.277937e-05 elapsed=15:44:04 avg_step=3.67s recent_step=4.38s eta=35:13:10 +2026-09-18 09:13:21 | INFO | train rank=0 | step=15450/50000 progress=30.9% loss=0.135265 lr=8.275440e-05 lr_backbone=8.275440e-05 elapsed=15:44:45 avg_step=3.67s recent_step=4.02s eta=35:12:41 +2026-09-18 09:13:56 | INFO | train rank=0 | step=15460/50000 progress=30.9% loss=0.140190 lr=8.272940e-05 lr_backbone=8.272940e-05 elapsed=15:45:19 avg_step=3.67s recent_step=3.48s eta=35:12:01 +2026-09-18 09:14:26 | INFO | train rank=0 | step=15470/50000 progress=30.9% loss=0.086515 lr=8.270440e-05 lr_backbone=8.270440e-05 elapsed=15:45:49 avg_step=3.67s recent_step=2.95s eta=35:11:08 +2026-09-18 09:15:00 | INFO | train rank=0 | step=15480/50000 progress=31.0% loss=0.110595 lr=8.267937e-05 lr_backbone=8.267937e-05 elapsed=15:46:24 avg_step=3.67s recent_step=3.47s eta=35:10:27 +2026-09-18 09:15:49 | INFO | train rank=0 | step=15490/50000 progress=31.0% loss=0.106675 lr=8.265434e-05 lr_backbone=8.265434e-05 elapsed=15:47:13 avg_step=3.67s recent_step=4.88s eta=35:10:17 +2026-09-18 09:16:18 | INFO | train rank=0 | step=15500/50000 progress=31.0% loss=0.099911 lr=8.262929e-05 lr_backbone=8.262929e-05 elapsed=15:47:41 avg_step=3.67s recent_step=2.84s eta=35:09:22 +2026-09-18 09:16:39 | INFO | train rank=0 | step=15510/50000 progress=31.0% loss=0.103931 lr=8.260422e-05 lr_backbone=8.260422e-05 elapsed=15:48:02 avg_step=3.67s recent_step=2.09s eta=35:08:10 +2026-09-18 09:17:16 | INFO | train rank=0 | step=15520/50000 progress=31.0% loss=0.113091 lr=8.257915e-05 lr_backbone=8.257915e-05 elapsed=15:48:39 avg_step=3.67s recent_step=3.71s eta=35:07:35 +2026-09-18 09:18:02 | INFO | train rank=0 | step=15530/50000 progress=31.1% loss=0.113370 lr=8.255405e-05 lr_backbone=8.255405e-05 elapsed=15:49:25 avg_step=3.67s recent_step=4.60s eta=35:07:19 +2026-09-18 09:18:37 | INFO | train rank=0 | step=15540/50000 progress=31.1% loss=0.096320 lr=8.252895e-05 lr_backbone=8.252895e-05 elapsed=15:50:01 avg_step=3.67s recent_step=3.56s eta=35:06:40 +2026-09-18 09:19:11 | INFO | train rank=0 | step=15550/50000 progress=31.1% loss=0.108402 lr=8.250382e-05 lr_backbone=8.250382e-05 elapsed=15:50:34 avg_step=3.67s recent_step=3.35s eta=35:05:56 +2026-09-18 09:19:50 | INFO | train rank=0 | step=15560/50000 progress=31.1% loss=0.091868 lr=8.247869e-05 lr_backbone=8.247869e-05 elapsed=15:51:13 avg_step=3.67s recent_step=3.89s eta=35:05:24 +2026-09-18 09:20:39 | INFO | train rank=0 | step=15570/50000 progress=31.1% loss=0.150108 lr=8.245354e-05 lr_backbone=8.245354e-05 elapsed=15:52:02 avg_step=3.67s recent_step=4.92s eta=35:05:15 +2026-09-18 09:21:07 | INFO | train rank=0 | step=15580/50000 progress=31.2% loss=0.149175 lr=8.242837e-05 lr_backbone=8.242837e-05 elapsed=15:52:30 avg_step=3.67s recent_step=2.81s eta=35:04:20 +2026-09-18 09:21:42 | INFO | train rank=0 | step=15590/50000 progress=31.2% loss=0.100881 lr=8.240320e-05 lr_backbone=8.240320e-05 elapsed=15:53:05 avg_step=3.67s recent_step=3.50s eta=35:03:39 +2026-09-18 09:22:16 | INFO | train rank=0 | step=15600/50000 progress=31.2% loss=0.105934 lr=8.237800e-05 lr_backbone=8.237800e-05 elapsed=15:53:40 avg_step=3.67s recent_step=3.42s eta=35:02:57 +2026-09-18 09:22:57 | INFO | train rank=0 | step=15610/50000 progress=31.2% loss=0.096319 lr=8.235280e-05 lr_backbone=8.235280e-05 elapsed=15:54:20 avg_step=3.67s recent_step=4.09s eta=35:02:30 +2026-09-18 09:23:38 | INFO | train rank=0 | step=15620/50000 progress=31.2% loss=0.122049 lr=8.232758e-05 lr_backbone=8.232758e-05 elapsed=15:55:01 avg_step=3.67s recent_step=4.07s eta=35:02:02 +2026-09-18 09:24:10 | INFO | train rank=0 | step=15630/50000 progress=31.3% loss=0.097372 lr=8.230234e-05 lr_backbone=8.230234e-05 elapsed=15:55:33 avg_step=3.67s recent_step=3.18s eta=35:01:14 +2026-09-18 09:24:52 | INFO | train rank=0 | step=15640/50000 progress=31.3% loss=0.093967 lr=8.227709e-05 lr_backbone=8.227709e-05 elapsed=15:56:15 avg_step=3.67s recent_step=4.25s eta=35:00:51 +2026-09-18 09:25:27 | INFO | train rank=0 | step=15650/50000 progress=31.3% loss=0.111308 lr=8.225183e-05 lr_backbone=8.225183e-05 elapsed=15:56:51 avg_step=3.67s recent_step=3.53s eta=35:00:11 +2026-09-18 09:25:57 | INFO | train rank=0 | step=15660/50000 progress=31.3% loss=0.110073 lr=8.222655e-05 lr_backbone=8.222655e-05 elapsed=15:57:21 avg_step=3.67s recent_step=2.98s eta=34:59:19 +2026-09-18 09:26:27 | INFO | train rank=0 | step=15670/50000 progress=31.3% loss=0.110276 lr=8.220126e-05 lr_backbone=8.220126e-05 elapsed=15:57:51 avg_step=3.67s recent_step=3.01s eta=34:58:28 +2026-09-18 09:27:02 | INFO | train rank=0 | step=15680/50000 progress=31.4% loss=0.090515 lr=8.217596e-05 lr_backbone=8.217596e-05 elapsed=15:58:25 avg_step=3.67s recent_step=3.41s eta=34:57:46 +2026-09-18 09:27:48 | INFO | train rank=0 | step=15690/50000 progress=31.4% loss=0.100782 lr=8.215064e-05 lr_backbone=8.215064e-05 elapsed=15:59:12 avg_step=3.67s recent_step=4.69s eta=34:57:31 +2026-09-18 09:28:24 | INFO | train rank=0 | step=15700/50000 progress=31.4% loss=0.087590 lr=8.212530e-05 lr_backbone=8.212530e-05 elapsed=15:59:48 avg_step=3.67s recent_step=3.60s eta=34:56:53 +2026-09-18 09:28:54 | INFO | train rank=0 | step=15710/50000 progress=31.4% loss=0.095014 lr=8.209996e-05 lr_backbone=8.209996e-05 elapsed=16:00:18 avg_step=3.67s recent_step=3.00s eta=34:56:02 +2026-09-18 09:29:41 | INFO | train rank=0 | step=15720/50000 progress=31.4% loss=0.108690 lr=8.207460e-05 lr_backbone=8.207460e-05 elapsed=16:01:04 avg_step=3.67s recent_step=4.64s eta=34:55:46 +2026-09-18 09:30:18 | INFO | train rank=0 | step=15730/50000 progress=31.5% loss=0.097791 lr=8.204922e-05 lr_backbone=8.204922e-05 elapsed=16:01:42 avg_step=3.67s recent_step=3.74s eta=34:55:11 +2026-09-18 09:30:56 | INFO | train rank=0 | step=15740/50000 progress=31.5% loss=0.092556 lr=8.202383e-05 lr_backbone=8.202383e-05 elapsed=16:02:19 avg_step=3.67s recent_step=3.79s eta=34:54:37 +2026-09-18 09:31:33 | INFO | train rank=0 | step=15750/50000 progress=31.5% loss=0.163870 lr=8.199843e-05 lr_backbone=8.199843e-05 elapsed=16:02:57 avg_step=3.67s recent_step=3.73s eta=34:54:02 +2026-09-18 09:32:03 | INFO | train rank=0 | step=15760/50000 progress=31.5% loss=0.107243 lr=8.197301e-05 lr_backbone=8.197301e-05 elapsed=16:03:26 avg_step=3.67s recent_step=2.94s eta=34:53:09 +2026-09-18 09:32:53 | INFO | train rank=0 | step=15770/50000 progress=31.5% loss=0.105019 lr=8.194758e-05 lr_backbone=8.194758e-05 elapsed=16:04:17 avg_step=3.67s recent_step=5.06s eta=34:53:03 +2026-09-18 09:33:23 | INFO | train rank=0 | step=15780/50000 progress=31.6% loss=0.095808 lr=8.192213e-05 lr_backbone=8.192213e-05 elapsed=16:04:47 avg_step=3.67s recent_step=3.00s eta=34:52:12 +2026-09-18 09:33:58 | INFO | train rank=0 | step=15790/50000 progress=31.6% loss=0.091904 lr=8.189667e-05 lr_backbone=8.189667e-05 elapsed=16:05:21 avg_step=3.67s recent_step=3.47s eta=34:51:31 +2026-09-18 09:34:32 | INFO | train rank=0 | step=15800/50000 progress=31.6% loss=0.088423 lr=8.187120e-05 lr_backbone=8.187120e-05 elapsed=16:05:56 avg_step=3.67s recent_step=3.41s eta=34:50:49 +2026-09-18 09:35:22 | INFO | train rank=0 | step=15810/50000 progress=31.6% loss=0.099206 lr=8.184571e-05 lr_backbone=8.184571e-05 elapsed=16:06:45 avg_step=3.67s recent_step=4.97s eta=34:50:40 +2026-09-18 09:35:53 | INFO | train rank=0 | step=15820/50000 progress=31.6% loss=0.102710 lr=8.182021e-05 lr_backbone=8.182021e-05 elapsed=16:07:16 avg_step=3.67s recent_step=3.08s eta=34:49:51 +2026-09-18 09:36:26 | INFO | train rank=0 | step=15830/50000 progress=31.7% loss=0.074752 lr=8.179470e-05 lr_backbone=8.179470e-05 elapsed=16:07:49 avg_step=3.67s recent_step=3.28s eta=34:49:06 +2026-09-18 09:37:00 | INFO | train rank=0 | step=15840/50000 progress=31.7% loss=0.092341 lr=8.176917e-05 lr_backbone=8.176917e-05 elapsed=16:08:23 avg_step=3.67s recent_step=3.41s eta=34:48:24 +2026-09-18 09:37:43 | INFO | train rank=0 | step=15850/50000 progress=31.7% loss=0.097642 lr=8.174362e-05 lr_backbone=8.174362e-05 elapsed=16:09:06 avg_step=3.67s recent_step=4.35s eta=34:48:02 +2026-09-18 09:38:21 | INFO | train rank=0 | step=15860/50000 progress=31.7% loss=0.092749 lr=8.171807e-05 lr_backbone=8.171807e-05 elapsed=16:09:45 avg_step=3.67s recent_step=3.82s eta=34:47:28 +2026-09-18 09:38:51 | INFO | train rank=0 | step=15870/50000 progress=31.7% loss=0.111370 lr=8.169250e-05 lr_backbone=8.169250e-05 elapsed=16:10:15 avg_step=3.67s recent_step=2.99s eta=34:46:37 +2026-09-18 09:39:26 | INFO | train rank=0 | step=15880/50000 progress=31.8% loss=0.096892 lr=8.166691e-05 lr_backbone=8.166691e-05 elapsed=16:10:49 avg_step=3.67s recent_step=3.43s eta=34:45:55 +2026-09-18 09:40:18 | INFO | train rank=0 | step=15890/50000 progress=31.8% loss=0.127146 lr=8.164131e-05 lr_backbone=8.164131e-05 elapsed=16:11:42 avg_step=3.67s recent_step=5.26s eta=34:45:53 +2026-09-18 09:40:49 | INFO | train rank=0 | step=15900/50000 progress=31.8% loss=0.098167 lr=8.161570e-05 lr_backbone=8.161570e-05 elapsed=16:12:13 avg_step=3.67s recent_step=3.11s eta=34:45:04 +2026-09-18 09:41:19 | INFO | train rank=0 | step=15910/50000 progress=31.8% loss=0.133479 lr=8.159007e-05 lr_backbone=8.159007e-05 elapsed=16:12:42 avg_step=3.67s recent_step=2.97s eta=34:44:12 +2026-09-18 09:41:50 | INFO | train rank=0 | step=15920/50000 progress=31.8% loss=0.159496 lr=8.156443e-05 lr_backbone=8.156443e-05 elapsed=16:13:14 avg_step=3.67s recent_step=3.13s eta=34:43:24 +2026-09-18 09:42:38 | INFO | train rank=0 | step=15930/50000 progress=31.9% loss=0.106637 lr=8.153878e-05 lr_backbone=8.153878e-05 elapsed=16:14:02 avg_step=3.67s recent_step=4.79s eta=34:43:11 +2026-09-18 09:43:20 | INFO | train rank=0 | step=15940/50000 progress=31.9% loss=0.085998 lr=8.151311e-05 lr_backbone=8.151311e-05 elapsed=16:14:43 avg_step=3.67s recent_step=4.15s eta=34:42:45 +2026-09-18 09:43:49 | INFO | train rank=0 | step=15950/50000 progress=31.9% loss=0.088989 lr=8.148743e-05 lr_backbone=8.148743e-05 elapsed=16:15:13 avg_step=3.67s recent_step=2.96s eta=34:41:53 +2026-09-18 09:44:21 | INFO | train rank=0 | step=15960/50000 progress=31.9% loss=0.092665 lr=8.146174e-05 lr_backbone=8.146174e-05 elapsed=16:15:44 avg_step=3.67s recent_step=3.13s eta=34:41:05 +2026-09-18 09:45:02 | INFO | train rank=0 | step=15970/50000 progress=31.9% loss=0.094346 lr=8.143603e-05 lr_backbone=8.143603e-05 elapsed=16:16:25 avg_step=3.67s recent_step=4.09s eta=34:40:37 +2026-09-18 09:45:48 | INFO | train rank=0 | step=15980/50000 progress=32.0% loss=0.093537 lr=8.141030e-05 lr_backbone=8.141030e-05 elapsed=16:17:12 avg_step=3.67s recent_step=4.70s eta=34:40:22 +2026-09-18 09:46:24 | INFO | train rank=0 | step=15990/50000 progress=32.0% loss=0.095551 lr=8.138457e-05 lr_backbone=8.138457e-05 elapsed=16:17:47 avg_step=3.67s recent_step=3.51s eta=34:39:42 +2026-09-18 09:46:55 | INFO | train rank=0 | step=16000/50000 progress=32.0% loss=0.102583 lr=8.135882e-05 lr_backbone=8.135882e-05 elapsed=16:18:18 avg_step=3.67s recent_step=3.12s eta=34:38:54 +2026-09-18 09:47:30 | INFO | train rank=0 | step=16010/50000 progress=32.0% loss=0.102677 lr=8.133305e-05 lr_backbone=8.133305e-05 elapsed=16:18:53 avg_step=3.67s recent_step=3.54s eta=34:38:14 +2026-09-18 09:48:17 | INFO | train rank=0 | step=16020/50000 progress=32.0% loss=0.091098 lr=8.130728e-05 lr_backbone=8.130728e-05 elapsed=16:19:41 avg_step=3.67s recent_step=4.74s eta=34:38:00 +2026-09-18 09:48:56 | INFO | train rank=0 | step=16030/50000 progress=32.1% loss=0.085456 lr=8.128149e-05 lr_backbone=8.128149e-05 elapsed=16:20:19 avg_step=3.67s recent_step=3.85s eta=34:37:28 +2026-09-18 09:49:28 | INFO | train rank=0 | step=16040/50000 progress=32.1% loss=0.112930 lr=8.125568e-05 lr_backbone=8.125568e-05 elapsed=16:20:51 avg_step=3.67s recent_step=3.21s eta=34:36:41 +2026-09-18 09:49:58 | INFO | train rank=0 | step=16050/50000 progress=32.1% loss=0.091351 lr=8.122986e-05 lr_backbone=8.122986e-05 elapsed=16:21:21 avg_step=3.67s recent_step=2.95s eta=34:35:49 +2026-09-18 09:50:50 | INFO | train rank=0 | step=16060/50000 progress=32.1% loss=0.121880 lr=8.120403e-05 lr_backbone=8.120403e-05 elapsed=16:22:13 avg_step=3.67s recent_step=5.19s eta=34:35:45 +2026-09-18 09:51:22 | INFO | train rank=0 | step=16070/50000 progress=32.1% loss=0.149550 lr=8.117818e-05 lr_backbone=8.117818e-05 elapsed=16:22:45 avg_step=3.67s recent_step=3.26s eta=34:34:59 +2026-09-18 09:51:56 | INFO | train rank=0 | step=16080/50000 progress=32.2% loss=0.074906 lr=8.115232e-05 lr_backbone=8.115232e-05 elapsed=16:23:20 avg_step=3.67s recent_step=3.43s eta=34:34:18 +2026-09-18 09:52:23 | INFO | train rank=0 | step=16090/50000 progress=32.2% loss=0.082633 lr=8.112645e-05 lr_backbone=8.112645e-05 elapsed=16:23:47 avg_step=3.67s recent_step=2.70s eta=34:33:21 +2026-09-18 09:53:18 | INFO | train rank=0 | step=16100/50000 progress=32.2% loss=0.084135 lr=8.110056e-05 lr_backbone=8.110056e-05 elapsed=16:24:41 avg_step=3.67s recent_step=5.43s eta=34:33:21 +2026-09-18 09:53:52 | INFO | train rank=0 | step=16110/50000 progress=32.2% loss=0.134026 lr=8.107466e-05 lr_backbone=8.107466e-05 elapsed=16:25:16 avg_step=3.67s recent_step=3.46s eta=34:32:40 +2026-09-18 09:54:26 | INFO | train rank=0 | step=16120/50000 progress=32.2% loss=0.085264 lr=8.104875e-05 lr_backbone=8.104875e-05 elapsed=16:25:49 avg_step=3.67s recent_step=3.34s eta=34:31:56 +2026-09-18 09:54:56 | INFO | train rank=0 | step=16130/50000 progress=32.3% loss=0.086549 lr=8.102282e-05 lr_backbone=8.102282e-05 elapsed=16:26:19 avg_step=3.67s recent_step=2.98s eta=34:31:05 +2026-09-18 09:55:36 | INFO | train rank=0 | step=16140/50000 progress=32.3% loss=0.126865 lr=8.099688e-05 lr_backbone=8.099688e-05 elapsed=16:27:00 avg_step=3.67s recent_step=4.08s eta=34:30:37 +2026-09-18 09:56:08 | INFO | train rank=0 | step=16150/50000 progress=32.3% loss=0.136563 lr=8.097093e-05 lr_backbone=8.097093e-05 elapsed=16:27:32 avg_step=3.67s recent_step=3.20s eta=34:29:51 +2026-09-18 09:56:41 | INFO | train rank=0 | step=16160/50000 progress=32.3% loss=0.127334 lr=8.094496e-05 lr_backbone=8.094496e-05 elapsed=16:28:05 avg_step=3.67s recent_step=3.28s eta=34:29:06 +2026-09-18 09:57:12 | INFO | train rank=0 | step=16170/50000 progress=32.3% loss=0.118685 lr=8.091898e-05 lr_backbone=8.091898e-05 elapsed=16:28:35 avg_step=3.67s recent_step=3.07s eta=34:28:17 +2026-09-18 09:58:12 | INFO | train rank=0 | step=16180/50000 progress=32.4% loss=0.134278 lr=8.089298e-05 lr_backbone=8.089298e-05 elapsed=16:29:36 avg_step=3.67s recent_step=6.03s eta=34:28:29 +2026-09-18 09:58:45 | INFO | train rank=0 | step=16190/50000 progress=32.4% loss=0.124458 lr=8.086697e-05 lr_backbone=8.086697e-05 elapsed=16:30:08 avg_step=3.67s recent_step=3.28s eta=34:27:44 +2026-09-18 09:59:17 | INFO | train rank=0 | step=16200/50000 progress=32.4% loss=0.105858 lr=8.084095e-05 lr_backbone=8.084095e-05 elapsed=16:30:40 avg_step=3.67s recent_step=3.17s eta=34:26:57 +2026-09-18 09:59:48 | INFO | train rank=0 | step=16210/50000 progress=32.4% loss=0.109924 lr=8.081491e-05 lr_backbone=8.081491e-05 elapsed=16:31:11 avg_step=3.67s recent_step=3.09s eta=34:26:08 +2026-09-18 10:00:40 | INFO | train rank=0 | step=16220/50000 progress=32.4% loss=0.092628 lr=8.078886e-05 lr_backbone=8.078886e-05 elapsed=16:32:03 avg_step=3.67s recent_step=5.20s eta=34:26:04 +2026-09-18 10:01:12 | INFO | train rank=0 | step=16230/50000 progress=32.5% loss=0.100537 lr=8.076280e-05 lr_backbone=8.076280e-05 elapsed=16:32:35 avg_step=3.67s recent_step=3.22s eta=34:25:18 +2026-09-18 10:01:42 | INFO | train rank=0 | step=16240/50000 progress=32.5% loss=0.107495 lr=8.073673e-05 lr_backbone=8.073673e-05 elapsed=16:33:05 avg_step=3.67s recent_step=3.04s eta=34:24:28 +2026-09-18 10:02:13 | INFO | train rank=0 | step=16250/50000 progress=32.5% loss=0.121268 lr=8.071064e-05 lr_backbone=8.071064e-05 elapsed=16:33:37 avg_step=3.67s recent_step=3.12s eta=34:23:40 +2026-09-18 10:02:57 | INFO | train rank=0 | step=16260/50000 progress=32.5% loss=0.087716 lr=8.068453e-05 lr_backbone=8.068453e-05 elapsed=16:34:21 avg_step=3.67s recent_step=4.40s eta=34:23:18 +2026-09-18 10:03:32 | INFO | train rank=0 | step=16270/50000 progress=32.5% loss=0.116846 lr=8.065842e-05 lr_backbone=8.065842e-05 elapsed=16:34:55 avg_step=3.67s recent_step=3.48s eta=34:22:38 +2026-09-18 10:04:05 | INFO | train rank=0 | step=16280/50000 progress=32.6% loss=0.101977 lr=8.063229e-05 lr_backbone=8.063229e-05 elapsed=16:35:29 avg_step=3.67s recent_step=3.33s eta=34:21:54 +2026-09-18 10:04:38 | INFO | train rank=0 | step=16290/50000 progress=32.6% loss=0.073471 lr=8.060614e-05 lr_backbone=8.060614e-05 elapsed=16:36:01 avg_step=3.67s recent_step=3.27s eta=34:21:09 +2026-09-18 10:05:21 | INFO | train rank=0 | step=16300/50000 progress=32.6% loss=0.087582 lr=8.057999e-05 lr_backbone=8.057999e-05 elapsed=16:36:44 avg_step=3.67s recent_step=4.31s eta=34:20:45 +2026-09-18 10:05:56 | INFO | train rank=0 | step=16310/50000 progress=32.6% loss=0.131160 lr=8.055382e-05 lr_backbone=8.055382e-05 elapsed=16:37:20 avg_step=3.67s recent_step=3.52s eta=34:20:06 +2026-09-18 10:06:28 | INFO | train rank=0 | step=16320/50000 progress=32.6% loss=0.080051 lr=8.052763e-05 lr_backbone=8.052763e-05 elapsed=16:37:52 avg_step=3.67s recent_step=3.21s eta=34:19:19 +2026-09-18 10:07:03 | INFO | train rank=0 | step=16330/50000 progress=32.7% loss=0.093543 lr=8.050144e-05 lr_backbone=8.050144e-05 elapsed=16:38:26 avg_step=3.67s recent_step=3.42s eta=34:18:37 +2026-09-18 10:07:38 | INFO | train rank=0 | step=16340/50000 progress=32.7% loss=0.112678 lr=8.047523e-05 lr_backbone=8.047523e-05 elapsed=16:39:01 avg_step=3.67s recent_step=3.53s eta=34:17:58 +2026-09-18 10:08:22 | INFO | train rank=0 | step=16350/50000 progress=32.7% loss=0.103876 lr=8.044900e-05 lr_backbone=8.044900e-05 elapsed=16:39:45 avg_step=3.67s recent_step=4.41s eta=34:17:36 +2026-09-18 10:08:57 | INFO | train rank=0 | step=16360/50000 progress=32.7% loss=0.097101 lr=8.042277e-05 lr_backbone=8.042277e-05 elapsed=16:40:20 avg_step=3.67s recent_step=3.50s eta=34:16:56 +2026-09-18 10:09:32 | INFO | train rank=0 | step=16370/50000 progress=32.7% loss=0.103551 lr=8.039651e-05 lr_backbone=8.039651e-05 elapsed=16:40:55 avg_step=3.67s recent_step=3.47s eta=34:16:16 +2026-09-18 10:10:22 | INFO | train rank=0 | step=16380/50000 progress=32.8% loss=0.094083 lr=8.037025e-05 lr_backbone=8.037025e-05 elapsed=16:41:46 avg_step=3.67s recent_step=5.07s eta=34:16:08 +2026-09-18 10:10:55 | INFO | train rank=0 | step=16390/50000 progress=32.8% loss=0.090515 lr=8.034397e-05 lr_backbone=8.034397e-05 elapsed=16:42:18 avg_step=3.67s recent_step=3.22s eta=34:15:22 +2026-09-18 10:11:29 | INFO | train rank=0 | step=16400/50000 progress=32.8% loss=0.087598 lr=8.031768e-05 lr_backbone=8.031768e-05 elapsed=16:42:52 avg_step=3.67s recent_step=3.45s eta=34:14:41 +2026-09-18 10:12:18 | INFO | train rank=0 | step=16410/50000 progress=32.8% loss=0.225201 lr=8.029138e-05 lr_backbone=8.029138e-05 elapsed=16:43:42 avg_step=3.67s recent_step=4.93s eta=34:14:30 +2026-09-18 10:12:56 | INFO | train rank=0 | step=16420/50000 progress=32.8% loss=0.163923 lr=8.026507e-05 lr_backbone=8.026507e-05 elapsed=16:44:19 avg_step=3.67s recent_step=3.73s eta=34:13:54 +2026-09-18 10:13:30 | INFO | train rank=0 | step=16430/50000 progress=32.9% loss=0.095148 lr=8.023874e-05 lr_backbone=8.023874e-05 elapsed=16:44:54 avg_step=3.67s recent_step=3.47s eta=34:13:14 +2026-09-18 10:13:58 | INFO | train rank=0 | step=16440/50000 progress=32.9% loss=0.105191 lr=8.021239e-05 lr_backbone=8.021239e-05 elapsed=16:45:22 avg_step=3.67s recent_step=2.79s eta=34:12:19 +2026-09-18 10:14:43 | INFO | train rank=0 | step=16450/50000 progress=32.9% loss=0.100109 lr=8.018604e-05 lr_backbone=8.018604e-05 elapsed=16:46:07 avg_step=3.67s recent_step=4.48s eta=34:11:59 +2026-09-18 10:15:14 | INFO | train rank=0 | step=16460/50000 progress=32.9% loss=0.079452 lr=8.015967e-05 lr_backbone=8.015967e-05 elapsed=16:46:38 avg_step=3.67s recent_step=3.13s eta=34:11:11 +2026-09-18 10:15:51 | INFO | train rank=0 | step=16470/50000 progress=32.9% loss=0.103609 lr=8.013328e-05 lr_backbone=8.013328e-05 elapsed=16:47:15 avg_step=3.67s recent_step=3.68s eta=34:10:35 +2026-09-18 10:16:27 | INFO | train rank=0 | step=16480/50000 progress=33.0% loss=0.147706 lr=8.010689e-05 lr_backbone=8.010689e-05 elapsed=16:47:50 avg_step=3.67s recent_step=3.55s eta=34:09:55 +2026-09-18 10:16:57 | INFO | train rank=0 | step=16490/50000 progress=33.0% loss=0.082504 lr=8.008048e-05 lr_backbone=8.008048e-05 elapsed=16:48:20 avg_step=3.67s recent_step=3.03s eta=34:09:06 +2026-09-18 10:17:48 | INFO | train rank=0 | step=16500/50000 progress=33.0% loss=0.106048 lr=8.005406e-05 lr_backbone=8.005406e-05 elapsed=16:49:12 avg_step=3.67s recent_step=5.13s eta=34:08:59 +2026-09-18 10:18:14 | INFO | train rank=0 | step=16510/50000 progress=33.0% loss=0.143899 lr=8.002762e-05 lr_backbone=8.002762e-05 elapsed=16:49:37 avg_step=3.67s recent_step=2.52s eta=34:07:59 +2026-09-18 10:18:46 | INFO | train rank=0 | step=16520/50000 progress=33.0% loss=0.095435 lr=8.000117e-05 lr_backbone=8.000117e-05 elapsed=16:50:10 avg_step=3.67s recent_step=3.27s eta=34:07:14 +2026-09-18 10:19:24 | INFO | train rank=0 | step=16530/50000 progress=33.1% loss=0.091857 lr=7.997471e-05 lr_backbone=7.997471e-05 elapsed=16:50:47 avg_step=3.67s recent_step=3.73s eta=34:06:39 +2026-09-18 10:20:04 | INFO | train rank=0 | step=16540/50000 progress=33.1% loss=0.104676 lr=7.994824e-05 lr_backbone=7.994824e-05 elapsed=16:51:27 avg_step=3.67s recent_step=4.05s eta=34:06:09 +2026-09-18 10:20:42 | INFO | train rank=0 | step=16550/50000 progress=33.1% loss=0.102417 lr=7.992175e-05 lr_backbone=7.992175e-05 elapsed=16:52:06 avg_step=3.67s recent_step=3.83s eta=34:05:36 +2026-09-18 10:21:20 | INFO | train rank=0 | step=16560/50000 progress=33.1% loss=0.099939 lr=7.989525e-05 lr_backbone=7.989525e-05 elapsed=16:52:44 avg_step=3.67s recent_step=3.80s eta=34:05:02 +2026-09-18 10:21:55 | INFO | train rank=0 | step=16570/50000 progress=33.1% loss=0.108281 lr=7.986874e-05 lr_backbone=7.986874e-05 elapsed=16:53:18 avg_step=3.67s recent_step=3.45s eta=34:04:21 +2026-09-18 10:22:37 | INFO | train rank=0 | step=16580/50000 progress=33.2% loss=0.095509 lr=7.984221e-05 lr_backbone=7.984221e-05 elapsed=16:54:01 avg_step=3.67s recent_step=4.24s eta=34:03:56 +2026-09-18 10:23:13 | INFO | train rank=0 | step=16590/50000 progress=33.2% loss=0.135049 lr=7.981567e-05 lr_backbone=7.981567e-05 elapsed=16:54:36 avg_step=3.67s recent_step=3.58s eta=34:03:17 +2026-09-18 10:23:45 | INFO | train rank=0 | step=16600/50000 progress=33.2% loss=0.082110 lr=7.978912e-05 lr_backbone=7.978912e-05 elapsed=16:55:08 avg_step=3.67s recent_step=3.20s eta=34:02:31 +2026-09-18 10:24:21 | INFO | train rank=0 | step=16610/50000 progress=33.2% loss=0.100298 lr=7.976255e-05 lr_backbone=7.976255e-05 elapsed=16:55:44 avg_step=3.67s recent_step=3.60s eta=34:01:53 +2026-09-18 10:24:54 | INFO | train rank=0 | step=16620/50000 progress=33.2% loss=0.080143 lr=7.973597e-05 lr_backbone=7.973597e-05 elapsed=16:56:17 avg_step=3.67s recent_step=3.26s eta=34:01:08 +2026-09-18 10:25:26 | INFO | train rank=0 | step=16630/50000 progress=33.3% loss=0.130137 lr=7.970938e-05 lr_backbone=7.970938e-05 elapsed=16:56:49 avg_step=3.67s recent_step=3.19s eta=34:00:22 +2026-09-18 10:25:56 | INFO | train rank=0 | step=16640/50000 progress=33.3% loss=0.106690 lr=7.968277e-05 lr_backbone=7.968277e-05 elapsed=16:57:19 avg_step=3.67s recent_step=3.00s eta=33:59:32 +2026-09-18 10:26:47 | INFO | train rank=0 | step=16650/50000 progress=33.3% loss=0.077321 lr=7.965615e-05 lr_backbone=7.965615e-05 elapsed=16:58:10 avg_step=3.67s recent_step=5.13s eta=33:59:24 +2026-09-18 10:27:14 | INFO | train rank=0 | step=16660/50000 progress=33.3% loss=0.090881 lr=7.962952e-05 lr_backbone=7.962952e-05 elapsed=16:58:37 avg_step=3.67s recent_step=2.73s eta=33:58:29 +2026-09-18 10:27:54 | INFO | train rank=0 | step=16670/50000 progress=33.3% loss=0.114553 lr=7.960288e-05 lr_backbone=7.960288e-05 elapsed=16:59:17 avg_step=3.67s recent_step=3.98s eta=33:57:58 +2026-09-18 10:28:27 | INFO | train rank=0 | step=16680/50000 progress=33.4% loss=0.100761 lr=7.957622e-05 lr_backbone=7.957622e-05 elapsed=16:59:50 avg_step=3.67s recent_step=3.30s eta=33:57:14 +2026-09-18 10:29:08 | INFO | train rank=0 | step=16690/50000 progress=33.4% loss=0.152857 lr=7.954955e-05 lr_backbone=7.954955e-05 elapsed=17:00:31 avg_step=3.67s recent_step=4.08s eta=33:56:46 +2026-09-18 10:30:00 | INFO | train rank=0 | step=16700/50000 progress=33.4% loss=0.087834 lr=7.952287e-05 lr_backbone=7.952287e-05 elapsed=17:01:24 avg_step=3.67s recent_step=5.24s eta=33:56:41 +2026-09-18 10:30:32 | INFO | train rank=0 | step=16710/50000 progress=33.4% loss=0.155277 lr=7.949617e-05 lr_backbone=7.949617e-05 elapsed=17:01:55 avg_step=3.67s recent_step=3.15s eta=33:55:54 +2026-09-18 10:31:06 | INFO | train rank=0 | step=16720/50000 progress=33.4% loss=0.108174 lr=7.946947e-05 lr_backbone=7.946947e-05 elapsed=17:02:29 avg_step=3.67s recent_step=3.41s eta=33:55:12 +2026-09-18 10:31:36 | INFO | train rank=0 | step=16730/50000 progress=33.5% loss=0.098150 lr=7.944274e-05 lr_backbone=7.944274e-05 elapsed=17:02:59 avg_step=3.67s recent_step=3.03s eta=33:54:22 +2026-09-18 10:32:23 | INFO | train rank=0 | step=16740/50000 progress=33.5% loss=0.102635 lr=7.941601e-05 lr_backbone=7.941601e-05 elapsed=17:03:46 avg_step=3.67s recent_step=4.67s eta=33:54:06 +2026-09-18 10:32:54 | INFO | train rank=0 | step=16750/50000 progress=33.5% loss=0.093101 lr=7.938926e-05 lr_backbone=7.938926e-05 elapsed=17:04:17 avg_step=3.67s recent_step=3.13s eta=33:53:18 +2026-09-18 10:33:24 | INFO | train rank=0 | step=16760/50000 progress=33.5% loss=0.096968 lr=7.936250e-05 lr_backbone=7.936250e-05 elapsed=17:04:48 avg_step=3.67s recent_step=3.01s eta=33:52:28 +2026-09-18 10:33:56 | INFO | train rank=0 | step=16770/50000 progress=33.5% loss=0.090172 lr=7.933573e-05 lr_backbone=7.933573e-05 elapsed=17:05:20 avg_step=3.67s recent_step=3.22s eta=33:51:43 +2026-09-18 10:34:47 | INFO | train rank=0 | step=16780/50000 progress=33.6% loss=0.086711 lr=7.930894e-05 lr_backbone=7.930894e-05 elapsed=17:06:10 avg_step=3.67s recent_step=5.02s eta=33:51:33 +2026-09-18 10:35:22 | INFO | train rank=0 | step=16790/50000 progress=33.6% loss=0.090349 lr=7.928215e-05 lr_backbone=7.928215e-05 elapsed=17:06:45 avg_step=3.67s recent_step=3.52s eta=33:50:53 +2026-09-18 10:35:47 | INFO | train rank=0 | step=16800/50000 progress=33.6% loss=0.088548 lr=7.925533e-05 lr_backbone=7.925533e-05 elapsed=17:07:10 avg_step=3.67s recent_step=2.53s eta=33:49:54 +2026-09-18 10:36:25 | INFO | train rank=0 | step=16810/50000 progress=33.6% loss=0.108922 lr=7.922851e-05 lr_backbone=7.922851e-05 elapsed=17:07:49 avg_step=3.67s recent_step=3.82s eta=33:49:20 +2026-09-18 10:37:21 | INFO | train rank=0 | step=16820/50000 progress=33.6% loss=0.094798 lr=7.920167e-05 lr_backbone=7.920167e-05 elapsed=17:08:45 avg_step=3.67s recent_step=5.60s eta=33:49:22 +2026-09-18 10:37:48 | INFO | train rank=0 | step=16830/50000 progress=33.7% loss=0.097686 lr=7.917482e-05 lr_backbone=7.917482e-05 elapsed=17:09:11 avg_step=3.67s recent_step=2.66s eta=33:48:25 +2026-09-18 10:38:23 | INFO | train rank=0 | step=16840/50000 progress=33.7% loss=0.111276 lr=7.914796e-05 lr_backbone=7.914796e-05 elapsed=17:09:47 avg_step=3.67s recent_step=3.53s eta=33:47:46 +2026-09-18 10:38:55 | INFO | train rank=0 | step=16850/50000 progress=33.7% loss=0.102337 lr=7.912108e-05 lr_backbone=7.912108e-05 elapsed=17:10:19 avg_step=3.67s recent_step=3.21s eta=33:47:00 +2026-09-18 10:39:46 | INFO | train rank=0 | step=16860/50000 progress=33.7% loss=0.093155 lr=7.909420e-05 lr_backbone=7.909420e-05 elapsed=17:11:09 avg_step=3.67s recent_step=5.02s eta=33:46:50 +2026-09-18 10:40:17 | INFO | train rank=0 | step=16870/50000 progress=33.7% loss=0.096920 lr=7.906730e-05 lr_backbone=7.906730e-05 elapsed=17:11:40 avg_step=3.67s recent_step=3.15s eta=33:46:03 +2026-09-18 10:40:50 | INFO | train rank=0 | step=16880/50000 progress=33.8% loss=0.101273 lr=7.904038e-05 lr_backbone=7.904038e-05 elapsed=17:12:13 avg_step=3.67s recent_step=3.30s eta=33:45:19 +2026-09-18 10:41:23 | INFO | train rank=0 | step=16890/50000 progress=33.8% loss=0.131002 lr=7.901346e-05 lr_backbone=7.901346e-05 elapsed=17:12:47 avg_step=3.67s recent_step=3.33s eta=33:44:36 +2026-09-18 10:42:16 | INFO | train rank=0 | step=16900/50000 progress=33.8% loss=0.103886 lr=7.898652e-05 lr_backbone=7.898652e-05 elapsed=17:13:40 avg_step=3.67s recent_step=5.30s eta=33:44:31 +2026-09-18 10:42:51 | INFO | train rank=0 | step=16910/50000 progress=33.8% loss=0.123717 lr=7.895957e-05 lr_backbone=7.895957e-05 elapsed=17:14:14 avg_step=3.67s recent_step=3.44s eta=33:43:50 +2026-09-18 10:43:20 | INFO | train rank=0 | step=16920/50000 progress=33.8% loss=0.112749 lr=7.893260e-05 lr_backbone=7.893260e-05 elapsed=17:14:43 avg_step=3.67s recent_step=2.91s eta=33:42:58 +2026-09-18 10:43:50 | INFO | train rank=0 | step=16930/50000 progress=33.9% loss=0.104833 lr=7.890562e-05 lr_backbone=7.890562e-05 elapsed=17:15:13 avg_step=3.67s recent_step=2.99s eta=33:42:08 +2026-09-18 10:44:47 | INFO | train rank=0 | step=16940/50000 progress=33.9% loss=0.086592 lr=7.887864e-05 lr_backbone=7.887864e-05 elapsed=17:16:10 avg_step=3.67s recent_step=5.70s eta=33:42:11 +2026-09-18 10:45:17 | INFO | train rank=0 | step=16950/50000 progress=33.9% loss=0.113784 lr=7.885163e-05 lr_backbone=7.885163e-05 elapsed=17:16:40 avg_step=3.67s recent_step=2.99s eta=33:41:21 +2026-09-18 10:45:45 | INFO | train rank=0 | step=16960/50000 progress=33.9% loss=0.114927 lr=7.882462e-05 lr_backbone=7.882462e-05 elapsed=17:17:09 avg_step=3.67s recent_step=2.89s eta=33:40:29 +2026-09-18 10:46:16 | INFO | train rank=0 | step=16970/50000 progress=33.9% loss=0.107594 lr=7.879759e-05 lr_backbone=7.879759e-05 elapsed=17:17:39 avg_step=3.67s recent_step=3.03s eta=33:39:40 +2026-09-18 10:47:03 | INFO | train rank=0 | step=16980/50000 progress=34.0% loss=0.087243 lr=7.877055e-05 lr_backbone=7.877055e-05 elapsed=17:18:26 avg_step=3.67s recent_step=4.74s eta=33:39:24 +2026-09-18 10:47:30 | INFO | train rank=0 | step=16990/50000 progress=34.0% loss=0.103947 lr=7.874350e-05 lr_backbone=7.874350e-05 elapsed=17:18:53 avg_step=3.67s recent_step=2.66s eta=33:38:28 +2026-09-18 10:47:59 | INFO | train rank=0 | step=17000/50000 progress=34.0% loss=0.089442 lr=7.871643e-05 lr_backbone=7.871643e-05 elapsed=17:19:22 avg_step=3.67s recent_step=2.87s eta=33:37:36 +2026-09-18 10:48:24 | INFO | train rank=0 | step=17010/50000 progress=34.0% loss=0.102623 lr=7.868936e-05 lr_backbone=7.868936e-05 elapsed=17:19:47 avg_step=3.67s recent_step=2.56s eta=33:36:38 +2026-09-18 10:49:05 | INFO | train rank=0 | step=17020/50000 progress=34.0% loss=0.089626 lr=7.866227e-05 lr_backbone=7.866227e-05 elapsed=17:20:28 avg_step=3.67s recent_step=4.06s eta=33:36:08 +2026-09-18 10:49:44 | INFO | train rank=0 | step=17030/50000 progress=34.1% loss=0.084075 lr=7.863516e-05 lr_backbone=7.863516e-05 elapsed=17:21:08 avg_step=3.67s recent_step=3.96s eta=33:35:37 +2026-09-18 10:50:18 | INFO | train rank=0 | step=17040/50000 progress=34.1% loss=0.081647 lr=7.860805e-05 lr_backbone=7.860805e-05 elapsed=17:21:41 avg_step=3.67s recent_step=3.35s eta=33:34:55 +2026-09-18 10:50:53 | INFO | train rank=0 | step=17050/50000 progress=34.1% loss=0.074613 lr=7.858092e-05 lr_backbone=7.858092e-05 elapsed=17:22:17 avg_step=3.67s recent_step=3.56s eta=33:34:16 +2026-09-18 10:51:46 | INFO | train rank=0 | step=17060/50000 progress=34.1% loss=0.111923 lr=7.855378e-05 lr_backbone=7.855378e-05 elapsed=17:23:10 avg_step=3.67s recent_step=5.31s eta=33:34:11 +2026-09-18 10:52:19 | INFO | train rank=0 | step=17070/50000 progress=34.1% loss=0.100942 lr=7.852663e-05 lr_backbone=7.852663e-05 elapsed=17:23:42 avg_step=3.67s recent_step=3.24s eta=33:33:26 +2026-09-18 10:52:52 | INFO | train rank=0 | step=17080/50000 progress=34.2% loss=0.144033 lr=7.849946e-05 lr_backbone=7.849946e-05 elapsed=17:24:15 avg_step=3.67s recent_step=3.27s eta=33:32:41 +2026-09-18 10:53:20 | INFO | train rank=0 | step=17090/50000 progress=34.2% loss=0.075385 lr=7.847228e-05 lr_backbone=7.847228e-05 elapsed=17:24:43 avg_step=3.67s recent_step=2.81s eta=33:31:48 +2026-09-18 10:54:09 | INFO | train rank=0 | step=17100/50000 progress=34.2% loss=0.111013 lr=7.844509e-05 lr_backbone=7.844509e-05 elapsed=17:25:32 avg_step=3.67s recent_step=4.91s eta=33:31:35 +2026-09-18 10:54:37 | INFO | train rank=0 | step=17110/50000 progress=34.2% loss=0.100069 lr=7.841789e-05 lr_backbone=7.841789e-05 elapsed=17:26:01 avg_step=3.67s recent_step=2.88s eta=33:30:43 +2026-09-18 10:55:08 | INFO | train rank=0 | step=17120/50000 progress=34.2% loss=0.073875 lr=7.839067e-05 lr_backbone=7.839067e-05 elapsed=17:26:31 avg_step=3.67s recent_step=3.04s eta=33:29:55 +2026-09-18 10:55:39 | INFO | train rank=0 | step=17130/50000 progress=34.3% loss=0.078984 lr=7.836345e-05 lr_backbone=7.836345e-05 elapsed=17:27:03 avg_step=3.67s recent_step=3.14s eta=33:29:08 +2026-09-18 10:56:32 | INFO | train rank=0 | step=17140/50000 progress=34.3% loss=0.098328 lr=7.833621e-05 lr_backbone=7.833621e-05 elapsed=17:27:55 avg_step=3.67s recent_step=5.29s eta=33:29:02 +2026-09-18 10:56:59 | INFO | train rank=0 | step=17150/50000 progress=34.3% loss=0.115954 lr=7.830896e-05 lr_backbone=7.830896e-05 elapsed=17:28:23 avg_step=3.67s recent_step=2.71s eta=33:28:07 +2026-09-18 10:57:34 | INFO | train rank=0 | step=17160/50000 progress=34.3% loss=0.094343 lr=7.828169e-05 lr_backbone=7.828169e-05 elapsed=17:28:57 avg_step=3.67s recent_step=3.49s eta=33:27:27 +2026-09-18 10:58:11 | INFO | train rank=0 | step=17170/50000 progress=34.3% loss=0.090531 lr=7.825441e-05 lr_backbone=7.825441e-05 elapsed=17:29:35 avg_step=3.67s recent_step=3.71s eta=33:26:51 +2026-09-18 10:58:53 | INFO | train rank=0 | step=17180/50000 progress=34.4% loss=0.105305 lr=7.822712e-05 lr_backbone=7.822712e-05 elapsed=17:30:16 avg_step=3.67s recent_step=4.18s eta=33:26:25 +2026-09-18 10:59:30 | INFO | train rank=0 | step=17190/50000 progress=34.4% loss=0.100653 lr=7.819982e-05 lr_backbone=7.819982e-05 elapsed=17:30:53 avg_step=3.67s recent_step=3.65s eta=33:25:47 +2026-09-18 11:00:03 | INFO | train rank=0 | step=17200/50000 progress=34.4% loss=0.094790 lr=7.817251e-05 lr_backbone=7.817251e-05 elapsed=17:31:26 avg_step=3.67s recent_step=3.36s eta=33:25:05 +2026-09-18 11:00:38 | INFO | train rank=0 | step=17210/50000 progress=34.4% loss=0.131015 lr=7.814518e-05 lr_backbone=7.814518e-05 elapsed=17:32:02 avg_step=3.67s recent_step=3.52s eta=33:24:25 +2026-09-18 11:01:25 | INFO | train rank=0 | step=17220/50000 progress=34.4% loss=0.120576 lr=7.811784e-05 lr_backbone=7.811784e-05 elapsed=17:32:48 avg_step=3.67s recent_step=4.63s eta=33:24:07 +2026-09-18 11:01:56 | INFO | train rank=0 | step=17230/50000 progress=34.5% loss=0.128602 lr=7.809049e-05 lr_backbone=7.809049e-05 elapsed=17:33:19 avg_step=3.67s recent_step=3.15s eta=33:23:20 +2026-09-18 11:02:31 | INFO | train rank=0 | step=17240/50000 progress=34.5% loss=0.114201 lr=7.806313e-05 lr_backbone=7.806313e-05 elapsed=17:33:54 avg_step=3.67s recent_step=3.47s eta=33:22:40 +2026-09-18 11:03:09 | INFO | train rank=0 | step=17250/50000 progress=34.5% loss=0.092899 lr=7.803575e-05 lr_backbone=7.803575e-05 elapsed=17:34:32 avg_step=3.67s recent_step=3.84s eta=33:22:06 +2026-09-18 11:03:55 | INFO | train rank=0 | step=17260/50000 progress=34.5% loss=0.106993 lr=7.800836e-05 lr_backbone=7.800836e-05 elapsed=17:35:18 avg_step=3.67s recent_step=4.57s eta=33:21:47 +2026-09-18 11:04:28 | INFO | train rank=0 | step=17270/50000 progress=34.5% loss=0.097184 lr=7.798096e-05 lr_backbone=7.798096e-05 elapsed=17:35:51 avg_step=3.67s recent_step=3.31s eta=33:21:03 +2026-09-18 11:05:08 | INFO | train rank=0 | step=17280/50000 progress=34.6% loss=0.153438 lr=7.795355e-05 lr_backbone=7.795355e-05 elapsed=17:36:31 avg_step=3.67s recent_step=3.99s eta=33:20:33 +2026-09-18 11:05:41 | INFO | train rank=0 | step=17290/50000 progress=34.6% loss=0.098716 lr=7.792613e-05 lr_backbone=7.792613e-05 elapsed=17:37:04 avg_step=3.67s recent_step=3.31s eta=33:19:49 +2026-09-18 11:06:38 | INFO | train rank=0 | step=17300/50000 progress=34.6% loss=0.089940 lr=7.789869e-05 lr_backbone=7.789869e-05 elapsed=17:38:01 avg_step=3.67s recent_step=5.70s eta=33:19:51 +2026-09-18 11:07:09 | INFO | train rank=0 | step=17310/50000 progress=34.6% loss=0.099003 lr=7.787124e-05 lr_backbone=7.787124e-05 elapsed=17:38:32 avg_step=3.67s recent_step=3.13s eta=33:19:04 +2026-09-18 11:07:40 | INFO | train rank=0 | step=17320/50000 progress=34.6% loss=0.095229 lr=7.784378e-05 lr_backbone=7.784378e-05 elapsed=17:39:03 avg_step=3.67s recent_step=3.09s eta=33:18:16 +2026-09-18 11:08:09 | INFO | train rank=0 | step=17330/50000 progress=34.7% loss=0.078263 lr=7.781631e-05 lr_backbone=7.781631e-05 elapsed=17:39:33 avg_step=3.67s recent_step=2.93s eta=33:17:26 +2026-09-18 11:08:55 | INFO | train rank=0 | step=17340/50000 progress=34.7% loss=0.099515 lr=7.778882e-05 lr_backbone=7.778882e-05 elapsed=17:40:19 avg_step=3.67s recent_step=4.61s eta=33:17:07 +2026-09-18 11:09:22 | INFO | train rank=0 | step=17350/50000 progress=34.7% loss=0.095603 lr=7.776132e-05 lr_backbone=7.776132e-05 elapsed=17:40:45 avg_step=3.67s recent_step=2.66s eta=33:16:11 +2026-09-18 11:09:58 | INFO | train rank=0 | step=17360/50000 progress=34.7% loss=0.107747 lr=7.773381e-05 lr_backbone=7.773381e-05 elapsed=17:41:21 avg_step=3.67s recent_step=3.54s eta=33:15:32 +2026-09-18 11:10:33 | INFO | train rank=0 | step=17370/50000 progress=34.7% loss=0.090552 lr=7.770629e-05 lr_backbone=7.770629e-05 elapsed=17:41:56 avg_step=3.67s recent_step=3.52s eta=33:14:53 +2026-09-18 11:11:18 | INFO | train rank=0 | step=17380/50000 progress=34.8% loss=0.096576 lr=7.767876e-05 lr_backbone=7.767876e-05 elapsed=17:42:42 avg_step=3.67s recent_step=4.58s eta=33:14:33 +2026-09-18 11:11:52 | INFO | train rank=0 | step=17390/50000 progress=34.8% loss=0.135545 lr=7.765121e-05 lr_backbone=7.765121e-05 elapsed=17:43:15 avg_step=3.67s recent_step=3.35s eta=33:13:50 +2026-09-18 11:12:26 | INFO | train rank=0 | step=17400/50000 progress=34.8% loss=0.105166 lr=7.762365e-05 lr_backbone=7.762365e-05 elapsed=17:43:49 avg_step=3.67s recent_step=3.36s eta=33:13:08 +2026-09-18 11:13:10 | INFO | train rank=0 | step=17410/50000 progress=34.8% loss=0.109491 lr=7.759608e-05 lr_backbone=7.759608e-05 elapsed=17:44:33 avg_step=3.67s recent_step=4.45s eta=33:12:46 +2026-09-18 11:14:00 | INFO | train rank=0 | step=17420/50000 progress=34.8% loss=0.111051 lr=7.756850e-05 lr_backbone=7.756850e-05 elapsed=17:45:24 avg_step=3.67s recent_step=5.02s eta=33:12:34 +2026-09-18 11:14:32 | INFO | train rank=0 | step=17430/50000 progress=34.9% loss=0.100045 lr=7.754091e-05 lr_backbone=7.754091e-05 elapsed=17:45:56 avg_step=3.67s recent_step=3.21s eta=33:11:49 +2026-09-18 11:15:03 | INFO | train rank=0 | step=17440/50000 progress=34.9% loss=0.119082 lr=7.751330e-05 lr_backbone=7.751330e-05 elapsed=17:46:27 avg_step=3.67s recent_step=3.09s eta=33:11:02 +2026-09-18 11:15:44 | INFO | train rank=0 | step=17450/50000 progress=34.9% loss=0.102432 lr=7.748568e-05 lr_backbone=7.748568e-05 elapsed=17:47:08 avg_step=3.67s recent_step=4.10s eta=33:10:33 +2026-09-18 11:16:30 | INFO | train rank=0 | step=17460/50000 progress=34.9% loss=0.085544 lr=7.745805e-05 lr_backbone=7.745805e-05 elapsed=17:47:53 avg_step=3.67s recent_step=4.53s eta=33:10:12 +2026-09-18 11:17:00 | INFO | train rank=0 | step=17470/50000 progress=34.9% loss=0.122047 lr=7.743041e-05 lr_backbone=7.743041e-05 elapsed=17:48:23 avg_step=3.67s recent_step=2.99s eta=33:09:23 +2026-09-18 11:17:35 | INFO | train rank=0 | step=17480/50000 progress=35.0% loss=0.082147 lr=7.740275e-05 lr_backbone=7.740275e-05 elapsed=17:48:59 avg_step=3.67s recent_step=3.57s eta=33:08:45 +2026-09-18 11:18:10 | INFO | train rank=0 | step=17490/50000 progress=35.0% loss=0.092929 lr=7.737509e-05 lr_backbone=7.737509e-05 elapsed=17:49:33 avg_step=3.67s recent_step=3.48s eta=33:08:04 +2026-09-18 11:18:52 | INFO | train rank=0 | step=17500/50000 progress=35.0% loss=0.115698 lr=7.734741e-05 lr_backbone=7.734741e-05 elapsed=17:50:15 avg_step=3.67s recent_step=4.18s eta=33:07:37 +2026-09-18 11:19:19 | INFO | train rank=0 | step=17510/50000 progress=35.0% loss=0.081962 lr=7.731972e-05 lr_backbone=7.731972e-05 elapsed=17:50:42 avg_step=3.67s recent_step=2.67s eta=33:06:42 +2026-09-18 11:19:48 | INFO | train rank=0 | step=17520/50000 progress=35.0% loss=0.103229 lr=7.729201e-05 lr_backbone=7.729201e-05 elapsed=17:51:12 avg_step=3.67s recent_step=2.99s eta=33:05:53 +2026-09-18 11:20:38 | INFO | train rank=0 | step=17530/50000 progress=35.1% loss=0.084259 lr=7.726430e-05 lr_backbone=7.726430e-05 elapsed=17:52:01 avg_step=3.67s recent_step=4.96s eta=33:05:40 +2026-09-18 11:21:09 | INFO | train rank=0 | step=17540/50000 progress=35.1% loss=0.103796 lr=7.723657e-05 lr_backbone=7.723657e-05 elapsed=17:52:32 avg_step=3.67s recent_step=3.06s eta=33:04:52 +2026-09-18 11:21:39 | INFO | train rank=0 | step=17550/50000 progress=35.1% loss=0.080470 lr=7.720884e-05 lr_backbone=7.720884e-05 elapsed=17:53:03 avg_step=3.67s recent_step=3.08s eta=33:04:04 +2026-09-18 11:22:19 | INFO | train rank=0 | step=17560/50000 progress=35.1% loss=0.121809 lr=7.718109e-05 lr_backbone=7.718109e-05 elapsed=17:53:42 avg_step=3.67s recent_step=3.95s eta=33:03:33 +2026-09-18 11:22:53 | INFO | train rank=0 | step=17570/50000 progress=35.1% loss=0.079025 lr=7.715332e-05 lr_backbone=7.715332e-05 elapsed=17:54:16 avg_step=3.67s recent_step=3.37s eta=33:02:51 +2026-09-18 11:23:48 | INFO | train rank=0 | step=17580/50000 progress=35.2% loss=0.178685 lr=7.712555e-05 lr_backbone=7.712555e-05 elapsed=17:55:11 avg_step=3.67s recent_step=5.52s eta=33:02:48 +2026-09-18 11:24:24 | INFO | train rank=0 | step=17590/50000 progress=35.2% loss=0.096620 lr=7.709776e-05 lr_backbone=7.709776e-05 elapsed=17:55:47 avg_step=3.67s recent_step=3.63s eta=33:02:11 +2026-09-18 11:24:56 | INFO | train rank=0 | step=17600/50000 progress=35.2% loss=0.106246 lr=7.706997e-05 lr_backbone=7.706997e-05 elapsed=17:56:20 avg_step=3.67s recent_step=3.20s eta=33:01:25 +2026-09-18 11:25:30 | INFO | train rank=0 | step=17610/50000 progress=35.2% loss=0.092312 lr=7.704216e-05 lr_backbone=7.704216e-05 elapsed=17:56:53 avg_step=3.67s recent_step=3.37s eta=33:00:43 +2026-09-18 11:26:20 | INFO | train rank=0 | step=17620/50000 progress=35.2% loss=0.101494 lr=7.701434e-05 lr_backbone=7.701434e-05 elapsed=17:57:44 avg_step=3.67s recent_step=5.04s eta=33:00:32 +2026-09-18 11:26:40 | INFO | train rank=0 | step=17630/50000 progress=35.3% loss=0.105730 lr=7.698650e-05 lr_backbone=7.698650e-05 elapsed=17:58:04 avg_step=3.67s recent_step=2.01s eta=32:59:24 +2026-09-18 11:27:12 | INFO | train rank=0 | step=17640/50000 progress=35.3% loss=0.126436 lr=7.695866e-05 lr_backbone=7.695866e-05 elapsed=17:58:35 avg_step=3.67s recent_step=3.18s eta=32:58:39 +2026-09-18 11:27:44 | INFO | train rank=0 | step=17650/50000 progress=35.3% loss=0.113338 lr=7.693080e-05 lr_backbone=7.693080e-05 elapsed=17:59:07 avg_step=3.67s recent_step=3.17s eta=32:57:53 +2026-09-18 11:28:38 | INFO | train rank=0 | step=17660/50000 progress=35.3% loss=0.113483 lr=7.690293e-05 lr_backbone=7.690293e-05 elapsed=18:00:01 avg_step=3.67s recent_step=5.38s eta=32:57:48 +2026-09-18 11:29:02 | INFO | train rank=0 | step=17670/50000 progress=35.3% loss=0.097534 lr=7.687505e-05 lr_backbone=7.687505e-05 elapsed=18:00:25 avg_step=3.67s recent_step=2.39s eta=32:56:47 +2026-09-18 11:29:35 | INFO | train rank=0 | step=17680/50000 progress=35.4% loss=0.097109 lr=7.684716e-05 lr_backbone=7.684716e-05 elapsed=18:00:59 avg_step=3.67s recent_step=3.40s eta=32:56:06 +2026-09-18 11:30:28 | INFO | train rank=0 | step=17690/50000 progress=35.4% loss=0.091669 lr=7.681926e-05 lr_backbone=7.681926e-05 elapsed=18:01:51 avg_step=3.67s recent_step=5.25s eta=32:55:58 +2026-09-18 11:30:58 | INFO | train rank=0 | step=17700/50000 progress=35.4% loss=0.116149 lr=7.679134e-05 lr_backbone=7.679134e-05 elapsed=18:02:22 avg_step=3.67s recent_step=3.02s eta=32:55:10 +2026-09-18 11:31:35 | INFO | train rank=0 | step=17710/50000 progress=35.4% loss=0.119037 lr=7.676341e-05 lr_backbone=7.676341e-05 elapsed=18:02:58 avg_step=3.67s recent_step=3.66s eta=32:54:33 +2026-09-18 11:32:04 | INFO | train rank=0 | step=17720/50000 progress=35.4% loss=0.083094 lr=7.673547e-05 lr_backbone=7.673547e-05 elapsed=18:03:27 avg_step=3.67s recent_step=2.90s eta=32:53:42 +2026-09-18 11:32:45 | INFO | train rank=0 | step=17730/50000 progress=35.5% loss=0.113933 lr=7.670752e-05 lr_backbone=7.670752e-05 elapsed=18:04:08 avg_step=3.67s recent_step=4.09s eta=32:53:13 +2026-09-18 11:33:26 | INFO | train rank=0 | step=17740/50000 progress=35.5% loss=0.092841 lr=7.667956e-05 lr_backbone=7.667956e-05 elapsed=18:04:49 avg_step=3.67s recent_step=4.09s eta=32:52:44 +2026-09-18 11:33:53 | INFO | train rank=0 | step=17750/50000 progress=35.5% loss=0.102663 lr=7.665159e-05 lr_backbone=7.665159e-05 elapsed=18:05:16 avg_step=3.67s recent_step=2.70s eta=32:51:50 +2026-09-18 11:34:20 | INFO | train rank=0 | step=17760/50000 progress=35.5% loss=0.117299 lr=7.662360e-05 lr_backbone=7.662360e-05 elapsed=18:05:44 avg_step=3.67s recent_step=2.77s eta=32:50:57 +2026-09-18 11:35:15 | INFO | train rank=0 | step=17770/50000 progress=35.5% loss=0.090152 lr=7.659560e-05 lr_backbone=7.659560e-05 elapsed=18:06:38 avg_step=3.67s recent_step=5.41s eta=32:50:52 +2026-09-18 11:35:46 | INFO | train rank=0 | step=17780/50000 progress=35.6% loss=0.083041 lr=7.656759e-05 lr_backbone=7.656759e-05 elapsed=18:07:09 avg_step=3.67s recent_step=3.12s eta=32:50:05 +2026-09-18 11:36:20 | INFO | train rank=0 | step=17790/50000 progress=35.6% loss=0.091753 lr=7.653957e-05 lr_backbone=7.653957e-05 elapsed=18:07:43 avg_step=3.67s recent_step=3.38s eta=32:49:23 +2026-09-18 11:36:54 | INFO | train rank=0 | step=17800/50000 progress=35.6% loss=0.147588 lr=7.651154e-05 lr_backbone=7.651154e-05 elapsed=18:08:17 avg_step=3.67s recent_step=3.46s eta=32:48:43 +2026-09-18 11:37:47 | INFO | train rank=0 | step=17810/50000 progress=35.6% loss=0.097070 lr=7.648350e-05 lr_backbone=7.648350e-05 elapsed=18:09:10 avg_step=3.67s recent_step=5.24s eta=32:48:34 +2026-09-18 11:38:18 | INFO | train rank=0 | step=17820/50000 progress=35.6% loss=0.108401 lr=7.645544e-05 lr_backbone=7.645544e-05 elapsed=18:09:41 avg_step=3.67s recent_step=3.14s eta=32:47:48 +2026-09-18 11:38:48 | INFO | train rank=0 | step=17830/50000 progress=35.7% loss=0.110539 lr=7.642738e-05 lr_backbone=7.642738e-05 elapsed=18:10:11 avg_step=3.67s recent_step=3.00s eta=32:46:59 +2026-09-18 11:39:43 | INFO | train rank=0 | step=17840/50000 progress=35.7% loss=0.086430 lr=7.639930e-05 lr_backbone=7.639930e-05 elapsed=18:11:06 avg_step=3.67s recent_step=5.50s eta=32:46:56 +2026-09-18 11:40:10 | INFO | train rank=0 | step=17850/50000 progress=35.7% loss=0.110401 lr=7.637121e-05 lr_backbone=7.637121e-05 elapsed=18:11:33 avg_step=3.67s recent_step=2.67s eta=32:46:01 +2026-09-18 11:40:40 | INFO | train rank=0 | step=17860/50000 progress=35.7% loss=0.100340 lr=7.634311e-05 lr_backbone=7.634311e-05 elapsed=18:12:04 avg_step=3.67s recent_step=3.07s eta=32:45:14 +2026-09-18 11:41:12 | INFO | train rank=0 | step=17870/50000 progress=35.7% loss=0.098785 lr=7.631499e-05 lr_backbone=7.631499e-05 elapsed=18:12:35 avg_step=3.67s recent_step=3.13s eta=32:44:27 +2026-09-18 11:42:00 | INFO | train rank=0 | step=17880/50000 progress=35.8% loss=0.092211 lr=7.628687e-05 lr_backbone=7.628687e-05 elapsed=18:13:24 avg_step=3.67s recent_step=4.89s eta=32:44:12 +2026-09-18 11:42:40 | INFO | train rank=0 | step=17890/50000 progress=35.8% loss=0.101638 lr=7.625873e-05 lr_backbone=7.625873e-05 elapsed=18:14:04 avg_step=3.67s recent_step=4.00s eta=32:43:42 +2026-09-18 11:43:10 | INFO | train rank=0 | step=17900/50000 progress=35.8% loss=0.093773 lr=7.623058e-05 lr_backbone=7.623058e-05 elapsed=18:14:33 avg_step=3.67s recent_step=2.96s eta=32:42:52 +2026-09-18 11:43:42 | INFO | train rank=0 | step=17910/50000 progress=35.8% loss=0.092373 lr=7.620242e-05 lr_backbone=7.620242e-05 elapsed=18:15:05 avg_step=3.67s recent_step=3.14s eta=32:42:06 +2026-09-18 11:44:22 | INFO | train rank=0 | step=17920/50000 progress=35.8% loss=0.111367 lr=7.617425e-05 lr_backbone=7.617425e-05 elapsed=18:15:45 avg_step=3.67s recent_step=4.01s eta=32:41:36 +2026-09-18 11:45:13 | INFO | train rank=0 | step=17930/50000 progress=35.9% loss=0.098305 lr=7.614607e-05 lr_backbone=7.614607e-05 elapsed=18:16:37 avg_step=3.67s recent_step=5.16s eta=32:41:26 +2026-09-18 11:45:40 | INFO | train rank=0 | step=17940/50000 progress=35.9% loss=0.079096 lr=7.611788e-05 lr_backbone=7.611788e-05 elapsed=18:17:04 avg_step=3.67s recent_step=2.72s eta=32:40:32 +2026-09-18 11:46:10 | INFO | train rank=0 | step=17950/50000 progress=35.9% loss=0.113022 lr=7.608967e-05 lr_backbone=7.608967e-05 elapsed=18:17:33 avg_step=3.67s recent_step=2.97s eta=32:39:43 +2026-09-18 11:46:50 | INFO | train rank=0 | step=17960/50000 progress=35.9% loss=0.136200 lr=7.606146e-05 lr_backbone=7.606146e-05 elapsed=18:18:14 avg_step=3.67s recent_step=4.03s eta=32:39:12 +2026-09-18 11:47:24 | INFO | train rank=0 | step=17970/50000 progress=35.9% loss=0.086893 lr=7.603323e-05 lr_backbone=7.603323e-05 elapsed=18:18:47 avg_step=3.67s recent_step=3.32s eta=32:38:29 +2026-09-18 11:47:59 | INFO | train rank=0 | step=17980/50000 progress=36.0% loss=0.091774 lr=7.600499e-05 lr_backbone=7.600499e-05 elapsed=18:19:22 avg_step=3.67s recent_step=3.50s eta=32:37:50 +2026-09-18 11:48:30 | INFO | train rank=0 | step=17990/50000 progress=36.0% loss=0.086827 lr=7.597674e-05 lr_backbone=7.597674e-05 elapsed=18:19:53 avg_step=3.67s recent_step=3.11s eta=32:37:03 +2026-09-18 11:49:27 | INFO | train rank=0 | step=18000/50000 progress=36.0% loss=0.086092 lr=7.594848e-05 lr_backbone=7.594848e-05 elapsed=18:20:50 avg_step=3.67s recent_step=5.74s eta=32:37:03 +2026-09-18 11:49:56 | INFO | train rank=0 | step=18010/50000 progress=36.0% loss=0.105867 lr=7.592021e-05 lr_backbone=7.592021e-05 elapsed=18:21:19 avg_step=3.67s recent_step=2.89s eta=32:36:13 +2026-09-18 11:50:32 | INFO | train rank=0 | step=18020/50000 progress=36.0% loss=0.109327 lr=7.589192e-05 lr_backbone=7.589192e-05 elapsed=18:21:55 avg_step=3.67s recent_step=3.58s eta=32:35:35 +2026-09-18 11:51:06 | INFO | train rank=0 | step=18030/50000 progress=36.1% loss=0.088291 lr=7.586363e-05 lr_backbone=7.586363e-05 elapsed=18:22:30 avg_step=3.67s recent_step=3.46s eta=32:34:54 +2026-09-18 11:51:48 | INFO | train rank=0 | step=18040/50000 progress=36.1% loss=0.104757 lr=7.583532e-05 lr_backbone=7.583532e-05 elapsed=18:23:11 avg_step=3.67s recent_step=4.17s eta=32:34:26 +2026-09-18 11:52:25 | INFO | train rank=0 | step=18050/50000 progress=36.1% loss=0.091296 lr=7.580700e-05 lr_backbone=7.580700e-05 elapsed=18:23:49 avg_step=3.67s recent_step=3.73s eta=32:33:51 +2026-09-18 11:52:55 | INFO | train rank=0 | step=18060/50000 progress=36.1% loss=0.071688 lr=7.577867e-05 lr_backbone=7.577867e-05 elapsed=18:24:18 avg_step=3.67s recent_step=2.98s eta=32:33:02 +2026-09-18 11:53:33 | INFO | train rank=0 | step=18070/50000 progress=36.1% loss=0.106771 lr=7.575033e-05 lr_backbone=7.575033e-05 elapsed=18:24:57 avg_step=3.67s recent_step=3.81s eta=32:32:28 +2026-09-18 11:54:21 | INFO | train rank=0 | step=18080/50000 progress=36.2% loss=0.110315 lr=7.572198e-05 lr_backbone=7.572198e-05 elapsed=18:25:44 avg_step=3.67s recent_step=4.74s eta=32:32:10 +2026-09-18 11:54:57 | INFO | train rank=0 | step=18090/50000 progress=36.2% loss=0.102105 lr=7.569361e-05 lr_backbone=7.569361e-05 elapsed=18:26:20 avg_step=3.67s recent_step=3.60s eta=32:31:32 +2026-09-18 11:55:24 | INFO | train rank=0 | step=18100/50000 progress=36.2% loss=0.103125 lr=7.566524e-05 lr_backbone=7.566524e-05 elapsed=18:26:47 avg_step=3.67s recent_step=2.71s eta=32:30:38 +2026-09-18 11:55:59 | INFO | train rank=0 | step=18110/50000 progress=36.2% loss=0.092035 lr=7.563685e-05 lr_backbone=7.563685e-05 elapsed=18:27:22 avg_step=3.67s recent_step=3.47s eta=32:29:58 +2026-09-18 11:56:50 | INFO | train rank=0 | step=18120/50000 progress=36.2% loss=0.098618 lr=7.560846e-05 lr_backbone=7.560846e-05 elapsed=18:28:13 avg_step=3.67s recent_step=5.12s eta=32:29:47 +2026-09-18 11:57:19 | INFO | train rank=0 | step=18130/50000 progress=36.3% loss=0.111925 lr=7.558005e-05 lr_backbone=7.558005e-05 elapsed=18:28:42 avg_step=3.67s recent_step=2.88s eta=32:28:56 +2026-09-18 11:57:46 | INFO | train rank=0 | step=18140/50000 progress=36.3% loss=0.101083 lr=7.555163e-05 lr_backbone=7.555163e-05 elapsed=18:29:10 avg_step=3.67s recent_step=2.77s eta=32:28:04 +2026-09-18 11:58:17 | INFO | train rank=0 | step=18150/50000 progress=36.3% loss=0.085194 lr=7.552320e-05 lr_backbone=7.552320e-05 elapsed=18:29:40 avg_step=3.67s recent_step=3.04s eta=32:27:16 +2026-09-18 11:59:12 | INFO | train rank=0 | step=18160/50000 progress=36.3% loss=0.100080 lr=7.549475e-05 lr_backbone=7.549475e-05 elapsed=18:30:35 avg_step=3.67s recent_step=5.51s eta=32:27:12 +2026-09-18 11:59:45 | INFO | train rank=0 | step=18170/50000 progress=36.3% loss=0.092688 lr=7.546630e-05 lr_backbone=7.546630e-05 elapsed=18:31:08 avg_step=3.67s recent_step=3.35s eta=32:26:29 +2026-09-18 12:00:17 | INFO | train rank=0 | step=18180/50000 progress=36.4% loss=0.081666 lr=7.543784e-05 lr_backbone=7.543784e-05 elapsed=18:31:40 avg_step=3.67s recent_step=3.15s eta=32:25:44 +2026-09-18 12:00:50 | INFO | train rank=0 | step=18190/50000 progress=36.4% loss=0.129330 lr=7.540936e-05 lr_backbone=7.540936e-05 elapsed=18:32:13 avg_step=3.67s recent_step=3.35s eta=32:25:01 +2026-09-18 12:01:34 | INFO | train rank=0 | step=18200/50000 progress=36.4% loss=0.102496 lr=7.538088e-05 lr_backbone=7.538088e-05 elapsed=18:32:57 avg_step=3.67s recent_step=4.37s eta=32:24:37 +2026-09-18 12:02:08 | INFO | train rank=0 | step=18210/50000 progress=36.4% loss=0.101326 lr=7.535238e-05 lr_backbone=7.535238e-05 elapsed=18:33:31 avg_step=3.67s recent_step=3.43s eta=32:23:56 +2026-09-18 12:02:37 | INFO | train rank=0 | step=18220/50000 progress=36.4% loss=0.102726 lr=7.532387e-05 lr_backbone=7.532387e-05 elapsed=18:34:00 avg_step=3.67s recent_step=2.89s eta=32:23:06 +2026-09-18 12:03:13 | INFO | train rank=0 | step=18230/50000 progress=36.5% loss=0.088716 lr=7.529535e-05 lr_backbone=7.529535e-05 elapsed=18:34:36 avg_step=3.67s recent_step=3.61s eta=32:22:28 +2026-09-18 12:04:08 | INFO | train rank=0 | step=18240/50000 progress=36.5% loss=0.169290 lr=7.526682e-05 lr_backbone=7.526682e-05 elapsed=18:35:31 avg_step=3.67s recent_step=5.46s eta=32:22:23 +2026-09-18 12:04:42 | INFO | train rank=0 | step=18250/50000 progress=36.5% loss=0.088167 lr=7.523828e-05 lr_backbone=7.523828e-05 elapsed=18:36:06 avg_step=3.67s recent_step=3.47s eta=32:21:42 +2026-09-18 12:05:09 | INFO | train rank=0 | step=18260/50000 progress=36.5% loss=0.093877 lr=7.520972e-05 lr_backbone=7.520972e-05 elapsed=18:36:32 avg_step=3.67s recent_step=2.65s eta=32:20:48 +2026-09-18 12:05:42 | INFO | train rank=0 | step=18270/50000 progress=36.5% loss=0.092636 lr=7.518116e-05 lr_backbone=7.518116e-05 elapsed=18:37:05 avg_step=3.67s recent_step=3.28s eta=32:20:04 +2026-09-18 12:06:24 | INFO | train rank=0 | step=18280/50000 progress=36.6% loss=0.098399 lr=7.515259e-05 lr_backbone=7.515259e-05 elapsed=18:37:48 avg_step=3.67s recent_step=4.25s eta=32:19:38 +2026-09-18 12:07:07 | INFO | train rank=0 | step=18290/50000 progress=36.6% loss=0.070247 lr=7.512400e-05 lr_backbone=7.512400e-05 elapsed=18:38:30 avg_step=3.67s recent_step=4.25s eta=32:19:11 +2026-09-18 12:07:40 | INFO | train rank=0 | step=18300/50000 progress=36.6% loss=0.147245 lr=7.509540e-05 lr_backbone=7.509540e-05 elapsed=18:39:04 avg_step=3.67s recent_step=3.36s eta=32:18:29 +2026-09-18 12:08:07 | INFO | train rank=0 | step=18310/50000 progress=36.6% loss=0.101500 lr=7.506679e-05 lr_backbone=7.506679e-05 elapsed=18:39:30 avg_step=3.67s recent_step=2.68s eta=32:17:35 +2026-09-18 12:08:54 | INFO | train rank=0 | step=18320/50000 progress=36.6% loss=0.085967 lr=7.503818e-05 lr_backbone=7.503818e-05 elapsed=18:40:17 avg_step=3.67s recent_step=4.67s eta=32:17:16 +2026-09-18 12:09:35 | INFO | train rank=0 | step=18330/50000 progress=36.7% loss=0.124982 lr=7.500955e-05 lr_backbone=7.500955e-05 elapsed=18:40:59 avg_step=3.67s recent_step=4.17s eta=32:16:48 +2026-09-18 12:10:08 | INFO | train rank=0 | step=18340/50000 progress=36.7% loss=0.082662 lr=7.498090e-05 lr_backbone=7.498090e-05 elapsed=18:41:31 avg_step=3.67s recent_step=3.24s eta=32:16:04 +2026-09-18 12:10:41 | INFO | train rank=0 | step=18350/50000 progress=36.7% loss=0.104634 lr=7.495225e-05 lr_backbone=7.495225e-05 elapsed=18:42:05 avg_step=3.67s recent_step=3.34s eta=32:15:21 +2026-09-18 12:11:24 | INFO | train rank=0 | step=18360/50000 progress=36.7% loss=0.109708 lr=7.492359e-05 lr_backbone=7.492359e-05 elapsed=18:42:47 avg_step=3.67s recent_step=4.24s eta=32:14:55 +2026-09-18 12:11:58 | INFO | train rank=0 | step=18370/50000 progress=36.7% loss=0.101255 lr=7.489492e-05 lr_backbone=7.489492e-05 elapsed=18:43:21 avg_step=3.67s recent_step=3.40s eta=32:14:13 +2026-09-18 12:12:29 | INFO | train rank=0 | step=18380/50000 progress=36.8% loss=0.118292 lr=7.486623e-05 lr_backbone=7.486623e-05 elapsed=18:43:52 avg_step=3.67s recent_step=3.11s eta=32:13:27 +2026-09-18 12:13:07 | INFO | train rank=0 | step=18390/50000 progress=36.8% loss=0.157466 lr=7.483754e-05 lr_backbone=7.483754e-05 elapsed=18:44:31 avg_step=3.67s recent_step=3.85s eta=32:12:53 +2026-09-18 12:13:58 | INFO | train rank=0 | step=18400/50000 progress=36.8% loss=0.111228 lr=7.480883e-05 lr_backbone=7.480883e-05 elapsed=18:45:21 avg_step=3.67s recent_step=5.09s eta=32:12:41 +2026-09-18 12:14:28 | INFO | train rank=0 | step=18410/50000 progress=36.8% loss=0.105156 lr=7.478011e-05 lr_backbone=7.478011e-05 elapsed=18:45:52 avg_step=3.67s recent_step=3.02s eta=32:11:53 +2026-09-18 12:15:04 | INFO | train rank=0 | step=18420/50000 progress=36.8% loss=0.099490 lr=7.475139e-05 lr_backbone=7.475139e-05 elapsed=18:46:27 avg_step=3.67s recent_step=3.53s eta=32:11:14 +2026-09-18 12:15:35 | INFO | train rank=0 | step=18430/50000 progress=36.9% loss=0.081462 lr=7.472265e-05 lr_backbone=7.472265e-05 elapsed=18:46:59 avg_step=3.67s recent_step=3.15s eta=32:10:29 +2026-09-18 12:16:29 | INFO | train rank=0 | step=18440/50000 progress=36.9% loss=0.102979 lr=7.469390e-05 lr_backbone=7.469390e-05 elapsed=18:47:53 avg_step=3.67s recent_step=5.41s eta=32:10:22 +2026-09-18 12:17:03 | INFO | train rank=0 | step=18450/50000 progress=36.9% loss=0.109066 lr=7.466514e-05 lr_backbone=7.466514e-05 elapsed=18:48:26 avg_step=3.67s recent_step=3.37s eta=32:09:40 +2026-09-18 12:17:28 | INFO | train rank=0 | step=18460/50000 progress=36.9% loss=0.120091 lr=7.463637e-05 lr_backbone=7.463637e-05 elapsed=18:48:51 avg_step=3.67s recent_step=2.51s eta=32:08:44 +2026-09-18 12:17:56 | INFO | train rank=0 | step=18470/50000 progress=36.9% loss=0.107386 lr=7.460759e-05 lr_backbone=7.460759e-05 elapsed=18:49:19 avg_step=3.67s recent_step=2.74s eta=32:07:51 +2026-09-18 12:18:46 | INFO | train rank=0 | step=18480/50000 progress=37.0% loss=0.095111 lr=7.457879e-05 lr_backbone=7.457879e-05 elapsed=18:50:09 avg_step=3.67s recent_step=5.05s eta=32:07:38 +2026-09-18 12:19:18 | INFO | train rank=0 | step=18490/50000 progress=37.0% loss=0.074735 lr=7.454999e-05 lr_backbone=7.454999e-05 elapsed=18:50:41 avg_step=3.67s recent_step=3.22s eta=32:06:53 +2026-09-18 12:19:53 | INFO | train rank=0 | step=18500/50000 progress=37.0% loss=0.101026 lr=7.452118e-05 lr_backbone=7.452118e-05 elapsed=18:51:16 avg_step=3.67s recent_step=3.47s eta=32:06:13 +2026-09-18 12:20:26 | INFO | train rank=0 | step=18510/50000 progress=37.0% loss=0.094194 lr=7.449235e-05 lr_backbone=7.449235e-05 elapsed=18:51:49 avg_step=3.67s recent_step=3.28s eta=32:05:30 +2026-09-18 12:21:16 | INFO | train rank=0 | step=18520/50000 progress=37.0% loss=0.089296 lr=7.446351e-05 lr_backbone=7.446351e-05 elapsed=18:52:39 avg_step=3.67s recent_step=5.03s eta=32:05:17 +2026-09-18 12:21:52 | INFO | train rank=0 | step=18530/50000 progress=37.1% loss=0.102996 lr=7.443467e-05 lr_backbone=7.443467e-05 elapsed=18:53:15 avg_step=3.67s recent_step=3.59s eta=32:04:39 +2026-09-18 12:22:25 | INFO | train rank=0 | step=18540/50000 progress=37.1% loss=0.110031 lr=7.440581e-05 lr_backbone=7.440581e-05 elapsed=18:53:48 avg_step=3.67s recent_step=3.25s eta=32:03:55 +2026-09-18 12:22:51 | INFO | train rank=0 | step=18550/50000 progress=37.1% loss=0.085192 lr=7.437694e-05 lr_backbone=7.437694e-05 elapsed=18:54:14 avg_step=3.67s recent_step=2.62s eta=32:03:00 +2026-09-18 12:23:48 | INFO | train rank=0 | step=18560/50000 progress=37.1% loss=0.102886 lr=7.434807e-05 lr_backbone=7.434807e-05 elapsed=18:55:11 avg_step=3.67s recent_step=5.72s eta=32:02:58 +2026-09-18 12:24:14 | INFO | train rank=0 | step=18570/50000 progress=37.1% loss=0.098065 lr=7.431918e-05 lr_backbone=7.431918e-05 elapsed=18:55:38 avg_step=3.67s recent_step=2.65s eta=32:02:04 +2026-09-18 12:24:45 | INFO | train rank=0 | step=18580/50000 progress=37.2% loss=0.088837 lr=7.429028e-05 lr_backbone=7.429028e-05 elapsed=18:56:08 avg_step=3.67s recent_step=3.07s eta=32:01:17 +2026-09-18 12:25:12 | INFO | train rank=0 | step=18590/50000 progress=37.2% loss=0.120894 lr=7.426137e-05 lr_backbone=7.426137e-05 elapsed=18:56:35 avg_step=3.67s recent_step=2.69s eta=32:00:24 +2026-09-18 12:26:04 | INFO | train rank=0 | step=18600/50000 progress=37.2% loss=0.094949 lr=7.423245e-05 lr_backbone=7.423245e-05 elapsed=18:57:28 avg_step=3.67s recent_step=5.27s eta=32:00:14 +2026-09-18 12:26:36 | INFO | train rank=0 | step=18610/50000 progress=37.2% loss=0.116357 lr=7.420352e-05 lr_backbone=7.420352e-05 elapsed=18:58:00 avg_step=3.67s recent_step=3.17s eta=31:59:29 +2026-09-18 12:27:07 | INFO | train rank=0 | step=18620/50000 progress=37.2% loss=0.091276 lr=7.417457e-05 lr_backbone=7.417457e-05 elapsed=18:58:31 avg_step=3.67s recent_step=3.12s eta=31:58:43 +2026-09-18 12:27:39 | INFO | train rank=0 | step=18630/50000 progress=37.3% loss=0.087231 lr=7.414562e-05 lr_backbone=7.414562e-05 elapsed=18:59:02 avg_step=3.67s recent_step=3.13s eta=31:57:58 +2026-09-18 12:28:35 | INFO | train rank=0 | step=18640/50000 progress=37.3% loss=0.109312 lr=7.411666e-05 lr_backbone=7.411666e-05 elapsed=18:59:58 avg_step=3.67s recent_step=5.62s eta=31:57:54 +2026-09-18 12:29:05 | INFO | train rank=0 | step=18650/50000 progress=37.3% loss=0.100904 lr=7.408768e-05 lr_backbone=7.408768e-05 elapsed=19:00:28 avg_step=3.67s recent_step=2.98s eta=31:57:06 +2026-09-18 12:29:38 | INFO | train rank=0 | step=18660/50000 progress=37.3% loss=0.074848 lr=7.405870e-05 lr_backbone=7.405870e-05 elapsed=19:01:01 avg_step=3.67s recent_step=3.30s eta=31:56:23 +2026-09-18 12:30:02 | INFO | train rank=0 | step=18670/50000 progress=37.3% loss=0.097128 lr=7.402970e-05 lr_backbone=7.402970e-05 elapsed=19:01:26 avg_step=3.67s recent_step=2.46s eta=31:55:26 +2026-09-18 12:30:58 | INFO | train rank=0 | step=18680/50000 progress=37.4% loss=0.093668 lr=7.400070e-05 lr_backbone=7.400070e-05 elapsed=19:02:22 avg_step=3.67s recent_step=5.60s eta=31:55:21 +2026-09-18 12:31:33 | INFO | train rank=0 | step=18690/50000 progress=37.4% loss=0.166300 lr=7.397168e-05 lr_backbone=7.397168e-05 elapsed=19:02:56 avg_step=3.67s recent_step=3.49s eta=31:54:42 +2026-09-18 12:32:04 | INFO | train rank=0 | step=18700/50000 progress=37.4% loss=0.152825 lr=7.394266e-05 lr_backbone=7.394266e-05 elapsed=19:03:28 avg_step=3.67s recent_step=3.12s eta=31:53:56 +2026-09-18 12:32:38 | INFO | train rank=0 | step=18710/50000 progress=37.4% loss=0.093980 lr=7.391362e-05 lr_backbone=7.391362e-05 elapsed=19:04:02 avg_step=3.67s recent_step=3.40s eta=31:53:14 +2026-09-18 12:33:29 | INFO | train rank=0 | step=18720/50000 progress=37.4% loss=0.136907 lr=7.388457e-05 lr_backbone=7.388457e-05 elapsed=19:04:53 avg_step=3.67s recent_step=5.11s eta=31:53:02 +2026-09-18 12:34:04 | INFO | train rank=0 | step=18730/50000 progress=37.5% loss=0.099587 lr=7.385552e-05 lr_backbone=7.385552e-05 elapsed=19:05:27 avg_step=3.67s recent_step=3.48s eta=31:52:22 +2026-09-18 12:34:37 | INFO | train rank=0 | step=18740/50000 progress=37.5% loss=0.093297 lr=7.382645e-05 lr_backbone=7.382645e-05 elapsed=19:06:00 avg_step=3.67s recent_step=3.23s eta=31:51:38 +2026-09-18 12:35:10 | INFO | train rank=0 | step=18750/50000 progress=37.5% loss=0.104017 lr=7.379737e-05 lr_backbone=7.379737e-05 elapsed=19:06:33 avg_step=3.67s recent_step=3.36s eta=31:50:56 +2026-09-18 12:36:00 | INFO | train rank=0 | step=18760/50000 progress=37.5% loss=0.086729 lr=7.376828e-05 lr_backbone=7.376828e-05 elapsed=19:07:23 avg_step=3.67s recent_step=5.00s eta=31:50:42 +2026-09-18 12:36:30 | INFO | train rank=0 | step=18770/50000 progress=37.5% loss=0.086473 lr=7.373918e-05 lr_backbone=7.373918e-05 elapsed=19:07:53 avg_step=3.67s recent_step=2.93s eta=31:49:53 +2026-09-18 12:37:01 | INFO | train rank=0 | step=18780/50000 progress=37.6% loss=0.104809 lr=7.371007e-05 lr_backbone=7.371007e-05 elapsed=19:08:25 avg_step=3.67s recent_step=3.18s eta=31:49:08 +2026-09-18 12:37:33 | INFO | train rank=0 | step=18790/50000 progress=37.6% loss=0.132140 lr=7.368095e-05 lr_backbone=7.368095e-05 elapsed=19:08:56 avg_step=3.67s recent_step=3.18s eta=31:48:23 +2026-09-18 12:38:16 | INFO | train rank=0 | step=18800/50000 progress=37.6% loss=0.090865 lr=7.365182e-05 lr_backbone=7.365182e-05 elapsed=19:09:39 avg_step=3.67s recent_step=4.25s eta=31:47:56 +2026-09-18 12:38:47 | INFO | train rank=0 | step=18810/50000 progress=37.6% loss=0.107085 lr=7.362268e-05 lr_backbone=7.362268e-05 elapsed=19:10:10 avg_step=3.67s recent_step=3.10s eta=31:47:10 +2026-09-18 12:39:11 | INFO | train rank=0 | step=18820/50000 progress=37.6% loss=0.097774 lr=7.359353e-05 lr_backbone=7.359353e-05 elapsed=19:10:35 avg_step=3.67s recent_step=2.46s eta=31:46:13 +2026-09-18 12:39:38 | INFO | train rank=0 | step=18830/50000 progress=37.7% loss=0.088469 lr=7.356437e-05 lr_backbone=7.356437e-05 elapsed=19:11:02 avg_step=3.67s recent_step=2.71s eta=31:45:21 +2026-09-18 12:40:31 | INFO | train rank=0 | step=18840/50000 progress=37.7% loss=0.127857 lr=7.353520e-05 lr_backbone=7.353520e-05 elapsed=19:11:54 avg_step=3.67s recent_step=5.24s eta=31:45:10 +2026-09-18 12:41:00 | INFO | train rank=0 | step=18850/50000 progress=37.7% loss=0.115473 lr=7.350601e-05 lr_backbone=7.350601e-05 elapsed=19:12:24 avg_step=3.67s recent_step=2.97s eta=31:44:22 +2026-09-18 12:41:30 | INFO | train rank=0 | step=18860/50000 progress=37.7% loss=0.094280 lr=7.347682e-05 lr_backbone=7.347682e-05 elapsed=19:12:53 avg_step=3.67s recent_step=2.95s eta=31:43:33 +2026-09-18 12:42:01 | INFO | train rank=0 | step=18870/50000 progress=37.7% loss=0.131317 lr=7.344762e-05 lr_backbone=7.344762e-05 elapsed=19:13:25 avg_step=3.67s recent_step=3.15s eta=31:42:48 +2026-09-18 12:42:54 | INFO | train rank=0 | step=18880/50000 progress=37.8% loss=0.089377 lr=7.341841e-05 lr_backbone=7.341841e-05 elapsed=19:14:18 avg_step=3.67s recent_step=5.27s eta=31:42:38 +2026-09-18 12:43:23 | INFO | train rank=0 | step=18890/50000 progress=37.8% loss=0.066319 lr=7.338918e-05 lr_backbone=7.338918e-05 elapsed=19:14:46 avg_step=3.67s recent_step=2.84s eta=31:41:47 +2026-09-18 12:43:59 | INFO | train rank=0 | step=18900/50000 progress=37.8% loss=0.100764 lr=7.335995e-05 lr_backbone=7.335995e-05 elapsed=19:15:22 avg_step=3.67s recent_step=3.63s eta=31:41:10 +2026-09-18 12:44:35 | INFO | train rank=0 | step=18910/50000 progress=37.8% loss=0.113397 lr=7.333071e-05 lr_backbone=7.333071e-05 elapsed=19:15:59 avg_step=3.67s recent_step=3.66s eta=31:40:33 +2026-09-18 12:45:24 | INFO | train rank=0 | step=18920/50000 progress=37.8% loss=0.121963 lr=7.330145e-05 lr_backbone=7.330145e-05 elapsed=19:16:47 avg_step=3.67s recent_step=4.81s eta=31:40:15 +2026-09-18 12:45:59 | INFO | train rank=0 | step=18930/50000 progress=37.9% loss=0.108712 lr=7.327219e-05 lr_backbone=7.327219e-05 elapsed=19:17:22 avg_step=3.67s recent_step=3.51s eta=31:39:36 +2026-09-18 12:46:39 | INFO | train rank=0 | step=18940/50000 progress=37.9% loss=0.085881 lr=7.324292e-05 lr_backbone=7.324292e-05 elapsed=19:18:02 avg_step=3.67s recent_step=4.00s eta=31:39:05 +2026-09-18 12:47:16 | INFO | train rank=0 | step=18950/50000 progress=37.9% loss=0.160612 lr=7.321363e-05 lr_backbone=7.321363e-05 elapsed=19:18:39 avg_step=3.67s recent_step=3.74s eta=31:38:30 +2026-09-18 12:48:07 | INFO | train rank=0 | step=18960/50000 progress=37.9% loss=0.073359 lr=7.318434e-05 lr_backbone=7.318434e-05 elapsed=19:19:30 avg_step=3.67s recent_step=5.08s eta=31:38:16 +2026-09-18 12:48:43 | INFO | train rank=0 | step=18970/50000 progress=37.9% loss=0.080283 lr=7.315503e-05 lr_backbone=7.315503e-05 elapsed=19:20:06 avg_step=3.67s recent_step=3.56s eta=31:37:37 +2026-09-18 12:49:17 | INFO | train rank=0 | step=18980/50000 progress=38.0% loss=0.086210 lr=7.312572e-05 lr_backbone=7.312572e-05 elapsed=19:20:40 avg_step=3.67s recent_step=3.42s eta=31:36:57 +2026-09-18 12:49:54 | INFO | train rank=0 | step=18990/50000 progress=38.0% loss=0.105169 lr=7.309639e-05 lr_backbone=7.309639e-05 elapsed=19:21:18 avg_step=3.67s recent_step=3.77s eta=31:36:21 +2026-09-18 12:50:47 | INFO | train rank=0 | step=19000/50000 progress=38.0% loss=0.108054 lr=7.306706e-05 lr_backbone=7.306706e-05 elapsed=19:22:10 avg_step=3.67s recent_step=5.26s eta=31:36:11 +2026-09-18 12:51:18 | INFO | train rank=0 | step=19010/50000 progress=38.0% loss=0.100761 lr=7.303771e-05 lr_backbone=7.303771e-05 elapsed=19:22:41 avg_step=3.67s recent_step=3.08s eta=31:35:24 +2026-09-18 12:51:54 | INFO | train rank=0 | step=19020/50000 progress=38.0% loss=0.078390 lr=7.300836e-05 lr_backbone=7.300836e-05 elapsed=19:23:18 avg_step=3.67s recent_step=3.66s eta=31:34:48 +2026-09-18 12:52:26 | INFO | train rank=0 | step=19030/50000 progress=38.1% loss=0.127628 lr=7.297899e-05 lr_backbone=7.297899e-05 elapsed=19:23:49 avg_step=3.67s recent_step=3.17s eta=31:34:03 +2026-09-18 12:53:19 | INFO | train rank=0 | step=19040/50000 progress=38.1% loss=0.179647 lr=7.294962e-05 lr_backbone=7.294962e-05 elapsed=19:24:43 avg_step=3.67s recent_step=5.34s eta=31:33:53 +2026-09-18 12:53:48 | INFO | train rank=0 | step=19050/50000 progress=38.1% loss=0.096566 lr=7.292023e-05 lr_backbone=7.292023e-05 elapsed=19:25:11 avg_step=3.67s recent_step=2.85s eta=31:33:03 +2026-09-18 12:54:20 | INFO | train rank=0 | step=19060/50000 progress=38.1% loss=0.096863 lr=7.289084e-05 lr_backbone=7.289084e-05 elapsed=19:25:43 avg_step=3.67s recent_step=3.19s eta=31:32:19 +2026-09-18 12:54:50 | INFO | train rank=0 | step=19070/50000 progress=38.1% loss=0.099781 lr=7.286143e-05 lr_backbone=7.286143e-05 elapsed=19:26:13 avg_step=3.67s recent_step=3.02s eta=31:31:32 +2026-09-18 12:55:39 | INFO | train rank=0 | step=19080/50000 progress=38.2% loss=0.096332 lr=7.283202e-05 lr_backbone=7.283202e-05 elapsed=19:27:02 avg_step=3.67s recent_step=4.88s eta=31:31:14 +2026-09-18 12:56:08 | INFO | train rank=0 | step=19090/50000 progress=38.2% loss=0.099721 lr=7.280259e-05 lr_backbone=7.280259e-05 elapsed=19:27:31 avg_step=3.67s recent_step=2.89s eta=31:30:25 +2026-09-18 12:56:41 | INFO | train rank=0 | step=19100/50000 progress=38.2% loss=0.105211 lr=7.277316e-05 lr_backbone=7.277316e-05 elapsed=19:28:04 avg_step=3.67s recent_step=3.32s eta=31:29:43 +2026-09-18 12:57:08 | INFO | train rank=0 | step=19110/50000 progress=38.2% loss=0.113456 lr=7.274371e-05 lr_backbone=7.274371e-05 elapsed=19:28:31 avg_step=3.67s recent_step=2.69s eta=31:28:50 +2026-09-18 12:57:54 | INFO | train rank=0 | step=19120/50000 progress=38.2% loss=0.124892 lr=7.271426e-05 lr_backbone=7.271426e-05 elapsed=19:29:18 avg_step=3.67s recent_step=4.64s eta=31:28:29 +2026-09-18 12:58:27 | INFO | train rank=0 | step=19130/50000 progress=38.3% loss=0.088943 lr=7.268479e-05 lr_backbone=7.268479e-05 elapsed=19:29:51 avg_step=3.67s recent_step=3.32s eta=31:27:47 +2026-09-18 12:59:03 | INFO | train rank=0 | step=19140/50000 progress=38.3% loss=0.076978 lr=7.265532e-05 lr_backbone=7.265532e-05 elapsed=19:30:27 avg_step=3.67s recent_step=3.61s eta=31:27:09 +2026-09-18 12:59:36 | INFO | train rank=0 | step=19150/50000 progress=38.3% loss=0.084606 lr=7.262583e-05 lr_backbone=7.262583e-05 elapsed=19:30:59 avg_step=3.67s recent_step=3.20s eta=31:26:25 +2026-09-18 13:00:31 | INFO | train rank=0 | step=19160/50000 progress=38.3% loss=0.099259 lr=7.259634e-05 lr_backbone=7.259634e-05 elapsed=19:31:54 avg_step=3.67s recent_step=5.57s eta=31:26:19 +2026-09-18 13:01:00 | INFO | train rank=0 | step=19170/50000 progress=38.3% loss=0.104567 lr=7.256683e-05 lr_backbone=7.256683e-05 elapsed=19:32:24 avg_step=3.67s recent_step=2.93s eta=31:25:30 +2026-09-18 13:01:29 | INFO | train rank=0 | step=19180/50000 progress=38.4% loss=0.109674 lr=7.253732e-05 lr_backbone=7.253732e-05 elapsed=19:32:53 avg_step=3.67s recent_step=2.89s eta=31:24:41 +2026-09-18 13:02:06 | INFO | train rank=0 | step=19190/50000 progress=38.4% loss=0.104036 lr=7.250779e-05 lr_backbone=7.250779e-05 elapsed=19:33:30 avg_step=3.67s recent_step=3.71s eta=31:24:05 +2026-09-18 13:02:58 | INFO | train rank=0 | step=19200/50000 progress=38.4% loss=0.101601 lr=7.247826e-05 lr_backbone=7.247826e-05 elapsed=19:34:22 avg_step=3.67s recent_step=5.19s eta=31:23:53 +2026-09-18 13:03:25 | INFO | train rank=0 | step=19210/50000 progress=38.4% loss=0.093314 lr=7.244871e-05 lr_backbone=7.244871e-05 elapsed=19:34:49 avg_step=3.67s recent_step=2.68s eta=31:23:00 +2026-09-18 13:03:50 | INFO | train rank=0 | step=19220/50000 progress=38.4% loss=0.117667 lr=7.241916e-05 lr_backbone=7.241916e-05 elapsed=19:35:14 avg_step=3.67s recent_step=2.50s eta=31:22:05 +2026-09-18 13:04:24 | INFO | train rank=0 | step=19230/50000 progress=38.5% loss=0.086175 lr=7.238960e-05 lr_backbone=7.238960e-05 elapsed=19:35:47 avg_step=3.67s recent_step=3.38s eta=31:21:23 +2026-09-18 13:05:24 | INFO | train rank=0 | step=19240/50000 progress=38.5% loss=0.097756 lr=7.236002e-05 lr_backbone=7.236002e-05 elapsed=19:36:47 avg_step=3.67s recent_step=6.01s eta=31:21:24 +2026-09-18 13:05:57 | INFO | train rank=0 | step=19250/50000 progress=38.5% loss=0.091089 lr=7.233044e-05 lr_backbone=7.233044e-05 elapsed=19:37:20 avg_step=3.67s recent_step=3.25s eta=31:20:41 +2026-09-18 13:06:32 | INFO | train rank=0 | step=19260/50000 progress=38.5% loss=0.068775 lr=7.230085e-05 lr_backbone=7.230085e-05 elapsed=19:37:55 avg_step=3.67s recent_step=3.55s eta=31:20:02 +2026-09-18 13:07:07 | INFO | train rank=0 | step=19270/50000 progress=38.5% loss=0.120850 lr=7.227124e-05 lr_backbone=7.227124e-05 elapsed=19:38:30 avg_step=3.67s recent_step=3.44s eta=31:19:22 +2026-09-18 13:08:03 | INFO | train rank=0 | step=19280/50000 progress=38.6% loss=0.099249 lr=7.224163e-05 lr_backbone=7.224163e-05 elapsed=19:39:27 avg_step=3.67s recent_step=5.68s eta=31:19:17 +2026-09-18 13:08:32 | INFO | train rank=0 | step=19290/50000 progress=38.6% loss=0.089582 lr=7.221201e-05 lr_backbone=7.221201e-05 elapsed=19:39:56 avg_step=3.67s recent_step=2.92s eta=31:18:28 +2026-09-18 13:09:02 | INFO | train rank=0 | step=19300/50000 progress=38.6% loss=0.098250 lr=7.218238e-05 lr_backbone=7.218238e-05 elapsed=19:40:25 avg_step=3.67s recent_step=2.96s eta=31:17:40 +2026-09-18 13:09:32 | INFO | train rank=0 | step=19310/50000 progress=38.6% loss=0.078713 lr=7.215274e-05 lr_backbone=7.215274e-05 elapsed=19:40:56 avg_step=3.67s recent_step=3.04s eta=31:16:54 +2026-09-18 13:10:19 | INFO | train rank=0 | step=19320/50000 progress=38.6% loss=0.097131 lr=7.212308e-05 lr_backbone=7.212308e-05 elapsed=19:41:43 avg_step=3.67s recent_step=4.70s eta=31:16:33 +2026-09-18 13:10:53 | INFO | train rank=0 | step=19330/50000 progress=38.7% loss=0.095663 lr=7.209342e-05 lr_backbone=7.209342e-05 elapsed=19:42:17 avg_step=3.67s recent_step=3.41s eta=31:15:52 +2026-09-18 13:11:23 | INFO | train rank=0 | step=19340/50000 progress=38.7% loss=0.106024 lr=7.206375e-05 lr_backbone=7.206375e-05 elapsed=19:42:46 avg_step=3.67s recent_step=2.95s eta=31:15:04 +2026-09-18 13:11:53 | INFO | train rank=0 | step=19350/50000 progress=38.7% loss=0.092555 lr=7.203407e-05 lr_backbone=7.203407e-05 elapsed=19:43:16 avg_step=3.67s recent_step=2.98s eta=31:14:17 +2026-09-18 13:12:47 | INFO | train rank=0 | step=19360/50000 progress=38.7% loss=0.119637 lr=7.200438e-05 lr_backbone=7.200438e-05 elapsed=19:44:10 avg_step=3.67s recent_step=5.42s eta=31:14:08 +2026-09-18 13:13:22 | INFO | train rank=0 | step=19370/50000 progress=38.7% loss=0.131419 lr=7.197468e-05 lr_backbone=7.197468e-05 elapsed=19:44:45 avg_step=3.67s recent_step=3.48s eta=31:13:28 +2026-09-18 13:13:56 | INFO | train rank=0 | step=19380/50000 progress=38.8% loss=0.073100 lr=7.194497e-05 lr_backbone=7.194497e-05 elapsed=19:45:20 avg_step=3.67s recent_step=3.46s eta=31:12:48 +2026-09-18 13:14:29 | INFO | train rank=0 | step=19390/50000 progress=38.8% loss=0.085308 lr=7.191525e-05 lr_backbone=7.191525e-05 elapsed=19:45:52 avg_step=3.67s recent_step=3.22s eta=31:12:04 +2026-09-18 13:15:20 | INFO | train rank=0 | step=19400/50000 progress=38.8% loss=0.088664 lr=7.188553e-05 lr_backbone=7.188553e-05 elapsed=19:46:43 avg_step=3.67s recent_step=5.10s eta=31:11:50 +2026-09-18 13:15:48 | INFO | train rank=0 | step=19410/50000 progress=38.8% loss=0.103513 lr=7.185579e-05 lr_backbone=7.185579e-05 elapsed=19:47:11 avg_step=3.67s recent_step=2.83s eta=31:11:00 +2026-09-18 13:16:18 | INFO | train rank=0 | step=19420/50000 progress=38.8% loss=0.091698 lr=7.182604e-05 lr_backbone=7.182604e-05 elapsed=19:47:42 avg_step=3.67s recent_step=3.06s eta=31:10:14 +2026-09-18 13:16:49 | INFO | train rank=0 | step=19430/50000 progress=38.9% loss=0.134328 lr=7.179628e-05 lr_backbone=7.179628e-05 elapsed=19:48:12 avg_step=3.67s recent_step=3.04s eta=31:09:27 +2026-09-18 13:17:35 | INFO | train rank=0 | step=19440/50000 progress=38.9% loss=0.132002 lr=7.176652e-05 lr_backbone=7.176652e-05 elapsed=19:48:58 avg_step=3.67s recent_step=4.58s eta=31:09:05 +2026-09-18 13:18:12 | INFO | train rank=0 | step=19450/50000 progress=38.9% loss=0.098383 lr=7.173674e-05 lr_backbone=7.173674e-05 elapsed=19:49:35 avg_step=3.67s recent_step=3.72s eta=31:08:29 +2026-09-18 13:18:39 | INFO | train rank=0 | step=19460/50000 progress=38.9% loss=0.101338 lr=7.170696e-05 lr_backbone=7.170696e-05 elapsed=19:50:02 avg_step=3.67s recent_step=2.70s eta=31:07:37 +2026-09-18 13:19:11 | INFO | train rank=0 | step=19470/50000 progress=38.9% loss=0.110119 lr=7.167716e-05 lr_backbone=7.167716e-05 elapsed=19:50:34 avg_step=3.67s recent_step=3.18s eta=31:06:53 +2026-09-18 13:19:52 | INFO | train rank=0 | step=19480/50000 progress=39.0% loss=0.085311 lr=7.164736e-05 lr_backbone=7.164736e-05 elapsed=19:51:16 avg_step=3.67s recent_step=4.17s eta=31:06:24 +2026-09-18 13:20:22 | INFO | train rank=0 | step=19490/50000 progress=39.0% loss=0.081144 lr=7.161754e-05 lr_backbone=7.161754e-05 elapsed=19:51:45 avg_step=3.67s recent_step=2.98s eta=31:05:36 +2026-09-18 13:20:47 | INFO | train rank=0 | step=19500/50000 progress=39.0% loss=0.101348 lr=7.158772e-05 lr_backbone=7.158772e-05 elapsed=19:52:11 avg_step=3.67s recent_step=2.52s eta=31:04:42 +2026-09-18 13:21:26 | INFO | train rank=0 | step=19510/50000 progress=39.0% loss=0.081386 lr=7.155788e-05 lr_backbone=7.155788e-05 elapsed=19:52:50 avg_step=3.67s recent_step=3.90s eta=31:04:09 +2026-09-18 13:22:17 | INFO | train rank=0 | step=19520/50000 progress=39.0% loss=0.108902 lr=7.152804e-05 lr_backbone=7.152804e-05 elapsed=19:53:41 avg_step=3.67s recent_step=5.11s eta=31:03:54 +2026-09-18 13:22:54 | INFO | train rank=0 | step=19530/50000 progress=39.1% loss=0.086038 lr=7.149819e-05 lr_backbone=7.149819e-05 elapsed=19:54:17 avg_step=3.67s recent_step=3.65s eta=31:03:17 +2026-09-18 13:23:32 | INFO | train rank=0 | step=19540/50000 progress=39.1% loss=0.074050 lr=7.146833e-05 lr_backbone=7.146833e-05 elapsed=19:54:56 avg_step=3.67s recent_step=3.85s eta=31:02:43 +2026-09-18 13:24:02 | INFO | train rank=0 | step=19550/50000 progress=39.1% loss=0.113609 lr=7.143846e-05 lr_backbone=7.143846e-05 elapsed=19:55:25 avg_step=3.67s recent_step=2.92s eta=31:01:55 +2026-09-18 13:24:52 | INFO | train rank=0 | step=19560/50000 progress=39.1% loss=0.099571 lr=7.140858e-05 lr_backbone=7.140858e-05 elapsed=19:56:15 avg_step=3.67s recent_step=5.05s eta=31:01:40 +2026-09-18 13:25:20 | INFO | train rank=0 | step=19570/50000 progress=39.1% loss=0.094531 lr=7.137869e-05 lr_backbone=7.137869e-05 elapsed=19:56:44 avg_step=3.67s recent_step=2.80s eta=31:00:50 +2026-09-18 13:25:53 | INFO | train rank=0 | step=19580/50000 progress=39.2% loss=0.090450 lr=7.134879e-05 lr_backbone=7.134879e-05 elapsed=19:57:17 avg_step=3.67s recent_step=3.33s eta=31:00:08 +2026-09-18 13:26:22 | INFO | train rank=0 | step=19590/50000 progress=39.2% loss=0.114306 lr=7.131888e-05 lr_backbone=7.131888e-05 elapsed=19:57:45 avg_step=3.67s recent_step=2.86s eta=30:59:19 +2026-09-18 13:27:06 | INFO | train rank=0 | step=19600/50000 progress=39.2% loss=0.093642 lr=7.128896e-05 lr_backbone=7.128896e-05 elapsed=19:58:30 avg_step=3.67s recent_step=4.42s eta=30:58:54 +2026-09-18 13:27:33 | INFO | train rank=0 | step=19610/50000 progress=39.2% loss=0.082407 lr=7.125904e-05 lr_backbone=7.125904e-05 elapsed=19:58:56 avg_step=3.67s recent_step=2.67s eta=30:58:01 +2026-09-18 13:28:09 | INFO | train rank=0 | step=19620/50000 progress=39.2% loss=0.074732 lr=7.122910e-05 lr_backbone=7.122910e-05 elapsed=19:59:32 avg_step=3.67s recent_step=3.57s eta=30:57:23 +2026-09-18 13:28:42 | INFO | train rank=0 | step=19630/50000 progress=39.3% loss=0.090297 lr=7.119916e-05 lr_backbone=7.119916e-05 elapsed=20:00:05 avg_step=3.67s recent_step=3.30s eta=30:56:41 +2026-09-18 13:29:23 | INFO | train rank=0 | step=19640/50000 progress=39.3% loss=0.159801 lr=7.116920e-05 lr_backbone=7.116920e-05 elapsed=20:00:46 avg_step=3.67s recent_step=4.11s eta=30:56:11 +2026-09-18 13:30:06 | INFO | train rank=0 | step=19650/50000 progress=39.3% loss=0.137696 lr=7.113924e-05 lr_backbone=7.113924e-05 elapsed=20:01:29 avg_step=3.67s recent_step=4.32s eta=30:55:44 +2026-09-18 13:30:36 | INFO | train rank=0 | step=19660/50000 progress=39.3% loss=0.092337 lr=7.110926e-05 lr_backbone=7.110926e-05 elapsed=20:01:59 avg_step=3.67s recent_step=2.99s eta=30:54:57 +2026-09-18 13:31:12 | INFO | train rank=0 | step=19670/50000 progress=39.3% loss=0.098195 lr=7.107928e-05 lr_backbone=7.107928e-05 elapsed=20:02:36 avg_step=3.67s recent_step=3.65s eta=30:54:20 +2026-09-18 13:31:46 | INFO | train rank=0 | step=19680/50000 progress=39.4% loss=0.078492 lr=7.104929e-05 lr_backbone=7.104929e-05 elapsed=20:03:10 avg_step=3.67s recent_step=3.40s eta=30:53:39 +2026-09-18 13:32:27 | INFO | train rank=0 | step=19690/50000 progress=39.4% loss=0.126257 lr=7.101929e-05 lr_backbone=7.101929e-05 elapsed=20:03:50 avg_step=3.67s recent_step=4.04s eta=30:53:09 +2026-09-18 13:33:00 | INFO | train rank=0 | step=19700/50000 progress=39.4% loss=0.103947 lr=7.098928e-05 lr_backbone=7.098928e-05 elapsed=20:04:24 avg_step=3.67s recent_step=3.35s eta=30:52:27 +2026-09-18 13:33:47 | INFO | train rank=0 | step=19710/50000 progress=39.4% loss=0.075839 lr=7.095926e-05 lr_backbone=7.095926e-05 elapsed=20:05:10 avg_step=3.67s recent_step=4.62s eta=30:52:05 +2026-09-18 13:34:16 | INFO | train rank=0 | step=19720/50000 progress=39.4% loss=0.093403 lr=7.092923e-05 lr_backbone=7.092923e-05 elapsed=20:05:39 avg_step=3.67s recent_step=2.93s eta=30:51:17 +2026-09-18 13:34:55 | INFO | train rank=0 | step=19730/50000 progress=39.5% loss=0.126536 lr=7.089920e-05 lr_backbone=7.089920e-05 elapsed=20:06:18 avg_step=3.67s recent_step=3.86s eta=30:50:43 +2026-09-18 13:35:32 | INFO | train rank=0 | step=19740/50000 progress=39.5% loss=0.102241 lr=7.086915e-05 lr_backbone=7.086915e-05 elapsed=20:06:56 avg_step=3.67s recent_step=3.76s eta=30:50:08 +2026-09-18 13:36:10 | INFO | train rank=0 | step=19750/50000 progress=39.5% loss=0.130374 lr=7.083909e-05 lr_backbone=7.083909e-05 elapsed=20:07:34 avg_step=3.67s recent_step=3.82s eta=30:49:34 +2026-09-18 13:36:44 | INFO | train rank=0 | step=19760/50000 progress=39.5% loss=0.101430 lr=7.080903e-05 lr_backbone=7.080903e-05 elapsed=20:08:07 avg_step=3.67s recent_step=3.33s eta=30:48:52 +2026-09-18 13:37:35 | INFO | train rank=0 | step=19770/50000 progress=39.5% loss=0.106672 lr=7.077895e-05 lr_backbone=7.077895e-05 elapsed=20:08:58 avg_step=3.67s recent_step=5.10s eta=30:48:37 +2026-09-18 13:38:02 | INFO | train rank=0 | step=19780/50000 progress=39.6% loss=0.094733 lr=7.074887e-05 lr_backbone=7.074887e-05 elapsed=20:09:25 avg_step=3.67s recent_step=2.74s eta=30:47:46 +2026-09-18 13:38:37 | INFO | train rank=0 | step=19790/50000 progress=39.6% loss=0.086586 lr=7.071878e-05 lr_backbone=7.071878e-05 elapsed=20:10:00 avg_step=3.67s recent_step=3.51s eta=30:47:07 +2026-09-18 13:39:08 | INFO | train rank=0 | step=19800/50000 progress=39.6% loss=0.087313 lr=7.068868e-05 lr_backbone=7.068868e-05 elapsed=20:10:31 avg_step=3.67s recent_step=3.06s eta=30:46:21 +2026-09-18 13:39:49 | INFO | train rank=0 | step=19810/50000 progress=39.6% loss=0.097813 lr=7.065857e-05 lr_backbone=7.065857e-05 elapsed=20:11:12 avg_step=3.67s recent_step=4.09s eta=30:45:51 +2026-09-18 13:40:32 | INFO | train rank=0 | step=19820/50000 progress=39.6% loss=0.080516 lr=7.062845e-05 lr_backbone=7.062845e-05 elapsed=20:11:55 avg_step=3.67s recent_step=4.34s eta=30:45:24 +2026-09-18 13:41:09 | INFO | train rank=0 | step=19830/50000 progress=39.7% loss=0.093524 lr=7.059832e-05 lr_backbone=7.059832e-05 elapsed=20:12:32 avg_step=3.67s recent_step=3.64s eta=30:44:47 +2026-09-18 13:41:52 | INFO | train rank=0 | step=19840/50000 progress=39.7% loss=0.099899 lr=7.056818e-05 lr_backbone=7.056818e-05 elapsed=20:13:16 avg_step=3.67s recent_step=4.39s eta=30:44:22 +2026-09-18 13:42:25 | INFO | train rank=0 | step=19850/50000 progress=39.7% loss=0.127865 lr=7.053804e-05 lr_backbone=7.053804e-05 elapsed=20:13:48 avg_step=3.67s recent_step=3.22s eta=30:43:38 +2026-09-18 13:42:56 | INFO | train rank=0 | step=19860/50000 progress=39.7% loss=0.090928 lr=7.050788e-05 lr_backbone=7.050788e-05 elapsed=20:14:19 avg_step=3.67s recent_step=3.14s eta=30:42:53 +2026-09-18 13:43:32 | INFO | train rank=0 | step=19870/50000 progress=39.7% loss=0.103291 lr=7.047772e-05 lr_backbone=7.047772e-05 elapsed=20:14:56 avg_step=3.67s recent_step=3.63s eta=30:42:16 +2026-09-18 13:44:25 | INFO | train rank=0 | step=19880/50000 progress=39.8% loss=0.109110 lr=7.044754e-05 lr_backbone=7.044754e-05 elapsed=20:15:49 avg_step=3.67s recent_step=5.31s eta=30:42:04 +2026-09-18 13:44:52 | INFO | train rank=0 | step=19890/50000 progress=39.8% loss=0.086592 lr=7.041736e-05 lr_backbone=7.041736e-05 elapsed=20:16:16 avg_step=3.67s recent_step=2.69s eta=30:41:13 +2026-09-18 13:45:36 | INFO | train rank=0 | step=19900/50000 progress=39.8% loss=0.112532 lr=7.038717e-05 lr_backbone=7.038717e-05 elapsed=20:16:59 avg_step=3.67s recent_step=4.37s eta=30:40:47 +2026-09-18 13:46:11 | INFO | train rank=0 | step=19910/50000 progress=39.8% loss=0.073392 lr=7.035697e-05 lr_backbone=7.035697e-05 elapsed=20:17:34 avg_step=3.67s recent_step=3.51s eta=30:40:08 +2026-09-18 13:46:44 | INFO | train rank=0 | step=19920/50000 progress=39.8% loss=0.108752 lr=7.032676e-05 lr_backbone=7.032676e-05 elapsed=20:18:07 avg_step=3.67s recent_step=3.24s eta=30:39:24 +2026-09-18 13:47:12 | INFO | train rank=0 | step=19930/50000 progress=39.9% loss=0.081200 lr=7.029654e-05 lr_backbone=7.029654e-05 elapsed=20:18:35 avg_step=3.67s recent_step=2.86s eta=30:38:35 +2026-09-18 13:48:08 | INFO | train rank=0 | step=19940/50000 progress=39.9% loss=0.092615 lr=7.026632e-05 lr_backbone=7.026632e-05 elapsed=20:19:31 avg_step=3.67s recent_step=5.60s eta=30:38:28 +2026-09-18 13:48:46 | INFO | train rank=0 | step=19950/50000 progress=39.9% loss=0.079998 lr=7.023608e-05 lr_backbone=7.023608e-05 elapsed=20:20:09 avg_step=3.67s recent_step=3.75s eta=30:37:52 +2026-09-18 13:49:19 | INFO | train rank=0 | step=19960/50000 progress=39.9% loss=0.091783 lr=7.020584e-05 lr_backbone=7.020584e-05 elapsed=20:20:42 avg_step=3.67s recent_step=3.34s eta=30:37:11 +2026-09-18 13:49:50 | INFO | train rank=0 | step=19970/50000 progress=39.9% loss=0.073489 lr=7.017558e-05 lr_backbone=7.017558e-05 elapsed=20:21:14 avg_step=3.67s recent_step=3.12s eta=30:36:26 +2026-09-18 13:50:36 | INFO | train rank=0 | step=19980/50000 progress=40.0% loss=0.103912 lr=7.014532e-05 lr_backbone=7.014532e-05 elapsed=20:21:59 avg_step=3.67s recent_step=4.56s eta=30:36:03 +2026-09-18 13:51:14 | INFO | train rank=0 | step=19990/50000 progress=40.0% loss=0.068813 lr=7.011505e-05 lr_backbone=7.011505e-05 elapsed=20:22:37 avg_step=3.67s recent_step=3.82s eta=30:35:28 +2026-09-18 13:51:49 | INFO | train rank=0 | step=20000/50000 progress=40.0% loss=0.095738 lr=7.008477e-05 lr_backbone=7.008477e-05 elapsed=20:23:12 avg_step=3.67s recent_step=3.45s eta=30:34:48 +2026-09-18 13:52:18 | INFO | train rank=0 | step=20010/50000 progress=40.0% loss=0.143751 lr=7.005448e-05 lr_backbone=7.005448e-05 elapsed=20:23:41 avg_step=3.67s recent_step=2.90s eta=30:34:00 +2026-09-18 13:53:08 | INFO | train rank=0 | step=20020/50000 progress=40.0% loss=0.133950 lr=7.002419e-05 lr_backbone=7.002419e-05 elapsed=20:24:31 avg_step=3.67s recent_step=5.00s eta=30:33:43 +2026-09-18 13:53:39 | INFO | train rank=0 | step=20030/50000 progress=40.1% loss=0.107659 lr=6.999388e-05 lr_backbone=6.999388e-05 elapsed=20:25:02 avg_step=3.67s recent_step=3.13s eta=30:32:58 +2026-09-18 13:54:06 | INFO | train rank=0 | step=20040/50000 progress=40.1% loss=0.122373 lr=6.996356e-05 lr_backbone=6.996356e-05 elapsed=20:25:29 avg_step=3.67s recent_step=2.67s eta=30:32:07 +2026-09-18 13:54:43 | INFO | train rank=0 | step=20050/50000 progress=40.1% loss=0.106761 lr=6.993324e-05 lr_backbone=6.993324e-05 elapsed=20:26:06 avg_step=3.67s recent_step=3.70s eta=30:31:31 +2026-09-18 13:55:35 | INFO | train rank=0 | step=20060/50000 progress=40.1% loss=0.134330 lr=6.990291e-05 lr_backbone=6.990291e-05 elapsed=20:26:58 avg_step=3.67s recent_step=5.24s eta=30:31:17 +2026-09-18 13:56:08 | INFO | train rank=0 | step=20070/50000 progress=40.1% loss=0.098409 lr=6.987257e-05 lr_backbone=6.987257e-05 elapsed=20:27:31 avg_step=3.67s recent_step=3.27s eta=30:30:35 +2026-09-18 13:56:36 | INFO | train rank=0 | step=20080/50000 progress=40.2% loss=0.082604 lr=6.984222e-05 lr_backbone=6.984222e-05 elapsed=20:27:59 avg_step=3.67s recent_step=2.83s eta=30:29:45 +2026-09-18 13:57:04 | INFO | train rank=0 | step=20090/50000 progress=40.2% loss=0.101937 lr=6.981186e-05 lr_backbone=6.981186e-05 elapsed=20:28:27 avg_step=3.67s recent_step=2.75s eta=30:28:55 +2026-09-18 13:57:53 | INFO | train rank=0 | step=20100/50000 progress=40.2% loss=0.108264 lr=6.978149e-05 lr_backbone=6.978149e-05 elapsed=20:29:16 avg_step=3.67s recent_step=4.89s eta=30:28:37 +2026-09-18 13:58:28 | INFO | train rank=0 | step=20110/50000 progress=40.2% loss=0.090199 lr=6.975112e-05 lr_backbone=6.975112e-05 elapsed=20:29:51 avg_step=3.67s recent_step=3.53s eta=30:27:58 +2026-09-18 13:58:59 | INFO | train rank=0 | step=20120/50000 progress=40.2% loss=0.110573 lr=6.972073e-05 lr_backbone=6.972073e-05 elapsed=20:30:22 avg_step=3.67s recent_step=3.11s eta=30:27:13 +2026-09-18 13:59:35 | INFO | train rank=0 | step=20130/50000 progress=40.3% loss=0.169368 lr=6.969034e-05 lr_backbone=6.969034e-05 elapsed=20:30:58 avg_step=3.67s recent_step=3.55s eta=30:26:35 +2026-09-18 14:00:19 | INFO | train rank=0 | step=20140/50000 progress=40.3% loss=0.094677 lr=6.965994e-05 lr_backbone=6.965994e-05 elapsed=20:31:42 avg_step=3.67s recent_step=4.40s eta=30:26:09 +2026-09-18 14:00:59 | INFO | train rank=0 | step=20150/50000 progress=40.3% loss=0.100922 lr=6.962953e-05 lr_backbone=6.962953e-05 elapsed=20:32:22 avg_step=3.67s recent_step=4.06s eta=30:25:38 +2026-09-18 14:01:29 | INFO | train rank=0 | step=20160/50000 progress=40.3% loss=0.088586 lr=6.959911e-05 lr_backbone=6.959911e-05 elapsed=20:32:52 avg_step=3.67s recent_step=2.98s eta=30:24:51 +2026-09-18 14:02:01 | INFO | train rank=0 | step=20170/50000 progress=40.3% loss=0.086994 lr=6.956868e-05 lr_backbone=6.956868e-05 elapsed=20:33:24 avg_step=3.67s recent_step=3.22s eta=30:24:08 +2026-09-18 14:02:37 | INFO | train rank=0 | step=20180/50000 progress=40.4% loss=0.117411 lr=6.953825e-05 lr_backbone=6.953825e-05 elapsed=20:34:01 avg_step=3.67s recent_step=3.63s eta=30:23:30 +2026-09-18 14:03:23 | INFO | train rank=0 | step=20190/50000 progress=40.4% loss=0.098946 lr=6.950780e-05 lr_backbone=6.950780e-05 elapsed=20:34:47 avg_step=3.67s recent_step=4.60s eta=30:23:07 +2026-09-18 14:04:00 | INFO | train rank=0 | step=20200/50000 progress=40.4% loss=0.096914 lr=6.947735e-05 lr_backbone=6.947735e-05 elapsed=20:35:24 avg_step=3.67s recent_step=3.69s eta=30:22:31 +2026-09-18 14:04:38 | INFO | train rank=0 | step=20210/50000 progress=40.4% loss=0.090814 lr=6.944689e-05 lr_backbone=6.944689e-05 elapsed=20:36:01 avg_step=3.67s recent_step=3.77s eta=30:21:56 +2026-09-18 14:05:02 | INFO | train rank=0 | step=20220/50000 progress=40.4% loss=0.097278 lr=6.941642e-05 lr_backbone=6.941642e-05 elapsed=20:36:25 avg_step=3.67s recent_step=2.40s eta=30:21:00 +2026-09-18 14:05:55 | INFO | train rank=0 | step=20230/50000 progress=40.5% loss=0.108032 lr=6.938594e-05 lr_backbone=6.938594e-05 elapsed=20:37:18 avg_step=3.67s recent_step=5.28s eta=30:20:47 +2026-09-18 14:06:27 | INFO | train rank=0 | step=20240/50000 progress=40.5% loss=0.083492 lr=6.935545e-05 lr_backbone=6.935545e-05 elapsed=20:37:50 avg_step=3.67s recent_step=3.22s eta=30:20:04 +2026-09-18 14:07:07 | INFO | train rank=0 | step=20250/50000 progress=40.5% loss=0.084668 lr=6.932496e-05 lr_backbone=6.932496e-05 elapsed=20:38:30 avg_step=3.67s recent_step=3.99s eta=30:19:32 +2026-09-18 14:07:39 | INFO | train rank=0 | step=20260/50000 progress=40.5% loss=0.101369 lr=6.929445e-05 lr_backbone=6.929445e-05 elapsed=20:39:03 avg_step=3.67s recent_step=3.23s eta=30:18:49 +2026-09-18 14:08:28 | INFO | train rank=0 | step=20270/50000 progress=40.5% loss=0.075106 lr=6.926394e-05 lr_backbone=6.926394e-05 elapsed=20:39:52 avg_step=3.67s recent_step=4.89s eta=30:18:30 +2026-09-18 14:08:58 | INFO | train rank=0 | step=20280/50000 progress=40.6% loss=0.086556 lr=6.923342e-05 lr_backbone=6.923342e-05 elapsed=20:40:22 avg_step=3.67s recent_step=3.02s eta=30:17:44 +2026-09-18 14:09:30 | INFO | train rank=0 | step=20290/50000 progress=40.6% loss=0.097387 lr=6.920289e-05 lr_backbone=6.920289e-05 elapsed=20:40:53 avg_step=3.67s recent_step=3.16s eta=30:17:00 +2026-09-18 14:10:02 | INFO | train rank=0 | step=20300/50000 progress=40.6% loss=0.096200 lr=6.917236e-05 lr_backbone=6.917236e-05 elapsed=20:41:25 avg_step=3.67s recent_step=3.20s eta=30:16:16 +2026-09-18 14:10:52 | INFO | train rank=0 | step=20310/50000 progress=40.6% loss=0.080659 lr=6.914181e-05 lr_backbone=6.914181e-05 elapsed=20:42:15 avg_step=3.67s recent_step=4.96s eta=30:15:59 +2026-09-18 14:11:13 | INFO | train rank=0 | step=20320/50000 progress=40.6% loss=0.081851 lr=6.911126e-05 lr_backbone=6.911126e-05 elapsed=20:42:37 avg_step=3.67s recent_step=2.18s eta=30:15:00 +2026-09-18 14:11:45 | INFO | train rank=0 | step=20330/50000 progress=40.7% loss=0.102957 lr=6.908069e-05 lr_backbone=6.908069e-05 elapsed=20:43:08 avg_step=3.67s recent_step=3.13s eta=30:14:15 +2026-09-18 14:12:21 | INFO | train rank=0 | step=20340/50000 progress=40.7% loss=0.112866 lr=6.905012e-05 lr_backbone=6.905012e-05 elapsed=20:43:44 avg_step=3.67s recent_step=3.61s eta=30:13:38 +2026-09-18 14:13:08 | INFO | train rank=0 | step=20350/50000 progress=40.7% loss=0.094858 lr=6.901954e-05 lr_backbone=6.901954e-05 elapsed=20:44:31 avg_step=3.67s recent_step=4.72s eta=30:13:17 +2026-09-18 14:13:39 | INFO | train rank=0 | step=20360/50000 progress=40.7% loss=0.097839 lr=6.898895e-05 lr_backbone=6.898895e-05 elapsed=20:45:02 avg_step=3.67s recent_step=3.09s eta=30:12:31 +2026-09-18 14:14:12 | INFO | train rank=0 | step=20370/50000 progress=40.7% loss=0.069084 lr=6.895836e-05 lr_backbone=6.895836e-05 elapsed=20:45:35 avg_step=3.67s recent_step=3.32s eta=30:11:50 +2026-09-18 14:14:57 | INFO | train rank=0 | step=20380/50000 progress=40.8% loss=0.074806 lr=6.892775e-05 lr_backbone=6.892775e-05 elapsed=20:46:20 avg_step=3.67s recent_step=4.48s eta=30:11:25 +2026-09-18 14:15:34 | INFO | train rank=0 | step=20390/50000 progress=40.8% loss=0.115693 lr=6.889714e-05 lr_backbone=6.889714e-05 elapsed=20:46:58 avg_step=3.67s recent_step=3.74s eta=30:10:49 +2026-09-18 14:16:05 | INFO | train rank=0 | step=20400/50000 progress=40.8% loss=0.147304 lr=6.886652e-05 lr_backbone=6.886652e-05 elapsed=20:47:28 avg_step=3.67s recent_step=3.05s eta=30:10:03 +2026-09-18 14:16:37 | INFO | train rank=0 | step=20410/50000 progress=40.8% loss=0.112415 lr=6.883589e-05 lr_backbone=6.883589e-05 elapsed=20:48:00 avg_step=3.67s recent_step=3.23s eta=30:09:20 +2026-09-18 14:17:31 | INFO | train rank=0 | step=20420/50000 progress=40.8% loss=0.095571 lr=6.880526e-05 lr_backbone=6.880526e-05 elapsed=20:48:54 avg_step=3.67s recent_step=5.40s eta=30:09:09 +2026-09-18 14:18:09 | INFO | train rank=0 | step=20430/50000 progress=40.9% loss=0.080564 lr=6.877461e-05 lr_backbone=6.877461e-05 elapsed=20:49:32 avg_step=3.67s recent_step=3.78s eta=30:08:34 +2026-09-18 14:18:34 | INFO | train rank=0 | step=20440/50000 progress=40.9% loss=0.096403 lr=6.874396e-05 lr_backbone=6.874396e-05 elapsed=20:49:57 avg_step=3.67s recent_step=2.48s eta=30:07:40 +2026-09-18 14:19:05 | INFO | train rank=0 | step=20450/50000 progress=40.9% loss=0.148061 lr=6.871329e-05 lr_backbone=6.871329e-05 elapsed=20:50:28 avg_step=3.67s recent_step=3.14s eta=30:06:55 +2026-09-18 14:20:06 | INFO | train rank=0 | step=20460/50000 progress=40.9% loss=0.097710 lr=6.868262e-05 lr_backbone=6.868262e-05 elapsed=20:51:29 avg_step=3.67s recent_step=6.10s eta=30:06:54 +2026-09-18 14:20:38 | INFO | train rank=0 | step=20470/50000 progress=40.9% loss=0.096343 lr=6.865195e-05 lr_backbone=6.865195e-05 elapsed=20:52:02 avg_step=3.67s recent_step=3.23s eta=30:06:11 +2026-09-18 14:21:12 | INFO | train rank=0 | step=20480/50000 progress=41.0% loss=0.117772 lr=6.862126e-05 lr_backbone=6.862126e-05 elapsed=20:52:36 avg_step=3.67s recent_step=3.39s eta=30:05:30 +2026-09-18 14:21:33 | INFO | train rank=0 | step=20490/50000 progress=41.0% loss=0.112135 lr=6.859057e-05 lr_backbone=6.859057e-05 elapsed=20:52:57 avg_step=3.67s recent_step=2.09s eta=30:04:31 +2026-09-18 14:22:28 | INFO | train rank=0 | step=20500/50000 progress=41.0% loss=0.081468 lr=6.855986e-05 lr_backbone=6.855986e-05 elapsed=20:53:52 avg_step=3.67s recent_step=5.49s eta=30:04:20 +2026-09-18 14:22:58 | INFO | train rank=0 | step=20510/50000 progress=41.0% loss=0.132004 lr=6.852915e-05 lr_backbone=6.852915e-05 elapsed=20:54:22 avg_step=3.67s recent_step=3.02s eta=30:03:34 +2026-09-18 14:23:33 | INFO | train rank=0 | step=20520/50000 progress=41.0% loss=0.081688 lr=6.849843e-05 lr_backbone=6.849843e-05 elapsed=20:54:56 avg_step=3.67s recent_step=3.44s eta=30:02:54 +2026-09-18 14:24:02 | INFO | train rank=0 | step=20530/50000 progress=41.1% loss=0.082936 lr=6.846771e-05 lr_backbone=6.846771e-05 elapsed=20:55:25 avg_step=3.67s recent_step=2.92s eta=30:02:07 +2026-09-18 14:24:55 | INFO | train rank=0 | step=20540/50000 progress=41.1% loss=0.116391 lr=6.843697e-05 lr_backbone=6.843697e-05 elapsed=20:56:18 avg_step=3.67s recent_step=5.30s eta=30:01:53 +2026-09-18 14:25:27 | INFO | train rank=0 | step=20550/50000 progress=41.1% loss=0.097215 lr=6.840623e-05 lr_backbone=6.840623e-05 elapsed=20:56:50 avg_step=3.67s recent_step=3.18s eta=30:01:10 +2026-09-18 14:26:02 | INFO | train rank=0 | step=20560/50000 progress=41.1% loss=0.099742 lr=6.837548e-05 lr_backbone=6.837548e-05 elapsed=20:57:25 avg_step=3.67s recent_step=3.50s eta=30:00:31 +2026-09-18 14:26:29 | INFO | train rank=0 | step=20570/50000 progress=41.1% loss=0.120020 lr=6.834472e-05 lr_backbone=6.834472e-05 elapsed=20:57:52 avg_step=3.67s recent_step=2.68s eta=29:59:40 +2026-09-18 14:27:19 | INFO | train rank=0 | step=20580/50000 progress=41.2% loss=0.090675 lr=6.831395e-05 lr_backbone=6.831395e-05 elapsed=20:58:43 avg_step=3.67s recent_step=5.07s eta=29:59:23 +2026-09-18 14:27:53 | INFO | train rank=0 | step=20590/50000 progress=41.2% loss=0.074393 lr=6.828317e-05 lr_backbone=6.828317e-05 elapsed=20:59:16 avg_step=3.67s recent_step=3.35s eta=29:58:42 +2026-09-18 14:28:26 | INFO | train rank=0 | step=20600/50000 progress=41.2% loss=0.101269 lr=6.825239e-05 lr_backbone=6.825239e-05 elapsed=20:59:49 avg_step=3.67s recent_step=3.29s eta=29:58:00 +2026-09-18 14:29:02 | INFO | train rank=0 | step=20610/50000 progress=41.2% loss=0.128584 lr=6.822160e-05 lr_backbone=6.822160e-05 elapsed=21:00:25 avg_step=3.67s recent_step=3.61s eta=29:57:22 +2026-09-18 14:29:41 | INFO | train rank=0 | step=20620/50000 progress=41.2% loss=0.089083 lr=6.819080e-05 lr_backbone=6.819080e-05 elapsed=21:01:04 avg_step=3.67s recent_step=3.93s eta=29:56:49 +2026-09-18 14:30:24 | INFO | train rank=0 | step=20630/50000 progress=41.3% loss=0.082465 lr=6.815999e-05 lr_backbone=6.815999e-05 elapsed=21:01:47 avg_step=3.67s recent_step=4.30s eta=29:56:21 +2026-09-18 14:30:58 | INFO | train rank=0 | step=20640/50000 progress=41.3% loss=0.087805 lr=6.812918e-05 lr_backbone=6.812918e-05 elapsed=21:02:21 avg_step=3.67s recent_step=3.40s eta=29:55:41 +2026-09-18 14:31:35 | INFO | train rank=0 | step=20650/50000 progress=41.3% loss=0.151190 lr=6.809836e-05 lr_backbone=6.809836e-05 elapsed=21:02:58 avg_step=3.67s recent_step=3.69s eta=29:55:05 +2026-09-18 14:32:10 | INFO | train rank=0 | step=20660/50000 progress=41.3% loss=0.076626 lr=6.806752e-05 lr_backbone=6.806752e-05 elapsed=21:03:33 avg_step=3.67s recent_step=3.47s eta=29:54:25 +2026-09-18 14:32:55 | INFO | train rank=0 | step=20670/50000 progress=41.3% loss=0.112186 lr=6.803669e-05 lr_backbone=6.803669e-05 elapsed=21:04:19 avg_step=3.67s recent_step=4.56s eta=29:54:01 +2026-09-18 14:33:32 | INFO | train rank=0 | step=20680/50000 progress=41.4% loss=0.078477 lr=6.800584e-05 lr_backbone=6.800584e-05 elapsed=21:04:55 avg_step=3.67s recent_step=3.67s eta=29:53:24 +2026-09-18 14:34:01 | INFO | train rank=0 | step=20690/50000 progress=41.4% loss=0.094067 lr=6.797498e-05 lr_backbone=6.797498e-05 elapsed=21:05:25 avg_step=3.67s recent_step=2.93s eta=29:52:37 +2026-09-18 14:34:32 | INFO | train rank=0 | step=20700/50000 progress=41.4% loss=0.106820 lr=6.794412e-05 lr_backbone=6.794412e-05 elapsed=21:05:55 avg_step=3.67s recent_step=3.05s eta=29:51:52 +2026-09-18 14:35:13 | INFO | train rank=0 | step=20710/50000 progress=41.4% loss=0.098553 lr=6.791325e-05 lr_backbone=6.791325e-05 elapsed=21:06:36 avg_step=3.67s recent_step=4.06s eta=29:51:21 +2026-09-18 14:35:48 | INFO | train rank=0 | step=20720/50000 progress=41.4% loss=0.082276 lr=6.788237e-05 lr_backbone=6.788237e-05 elapsed=21:07:12 avg_step=3.67s recent_step=3.57s eta=29:50:42 +2026-09-18 14:36:21 | INFO | train rank=0 | step=20730/50000 progress=41.5% loss=0.143277 lr=6.785149e-05 lr_backbone=6.785149e-05 elapsed=21:07:44 avg_step=3.67s recent_step=3.27s eta=29:50:00 +2026-09-18 14:36:48 | INFO | train rank=0 | step=20740/50000 progress=41.5% loss=0.114418 lr=6.782059e-05 lr_backbone=6.782059e-05 elapsed=21:08:12 avg_step=3.67s recent_step=2.74s eta=29:49:10 +2026-09-18 14:37:40 | INFO | train rank=0 | step=20750/50000 progress=41.5% loss=0.104020 lr=6.778969e-05 lr_backbone=6.778969e-05 elapsed=21:09:03 avg_step=3.67s recent_step=5.19s eta=29:48:55 +2026-09-18 14:38:08 | INFO | train rank=0 | step=20760/50000 progress=41.5% loss=0.106499 lr=6.775878e-05 lr_backbone=6.775878e-05 elapsed=21:09:31 avg_step=3.67s recent_step=2.78s eta=29:48:06 +2026-09-18 14:38:40 | INFO | train rank=0 | step=20770/50000 progress=41.5% loss=0.091532 lr=6.772787e-05 lr_backbone=6.772787e-05 elapsed=21:10:03 avg_step=3.67s recent_step=3.22s eta=29:47:23 +2026-09-18 14:39:11 | INFO | train rank=0 | step=20780/50000 progress=41.6% loss=0.070785 lr=6.769694e-05 lr_backbone=6.769694e-05 elapsed=21:10:35 avg_step=3.67s recent_step=3.12s eta=29:46:38 +2026-09-18 14:39:50 | INFO | train rank=0 | step=20790/50000 progress=41.6% loss=0.093882 lr=6.766601e-05 lr_backbone=6.766601e-05 elapsed=21:11:13 avg_step=3.67s recent_step=3.83s eta=29:46:04 +2026-09-18 14:40:29 | INFO | train rank=0 | step=20800/50000 progress=41.6% loss=0.084090 lr=6.763507e-05 lr_backbone=6.763507e-05 elapsed=21:11:52 avg_step=3.67s recent_step=3.95s eta=29:45:31 +2026-09-18 14:41:03 | INFO | train rank=0 | step=20810/50000 progress=41.6% loss=0.100704 lr=6.760412e-05 lr_backbone=6.760412e-05 elapsed=21:12:26 avg_step=3.67s recent_step=3.37s eta=29:44:50 +2026-09-18 14:41:48 | INFO | train rank=0 | step=20820/50000 progress=41.6% loss=0.095754 lr=6.757316e-05 lr_backbone=6.757316e-05 elapsed=21:13:11 avg_step=3.67s recent_step=4.51s eta=29:44:25 +2026-09-18 14:42:22 | INFO | train rank=0 | step=20830/50000 progress=41.7% loss=0.121846 lr=6.754220e-05 lr_backbone=6.754220e-05 elapsed=21:13:46 avg_step=3.67s recent_step=3.45s eta=29:43:46 +2026-09-18 14:42:56 | INFO | train rank=0 | step=20840/50000 progress=41.7% loss=0.074470 lr=6.751123e-05 lr_backbone=6.751123e-05 elapsed=21:14:19 avg_step=3.67s recent_step=3.32s eta=29:43:04 +2026-09-18 14:43:30 | INFO | train rank=0 | step=20850/50000 progress=41.7% loss=0.092363 lr=6.748025e-05 lr_backbone=6.748025e-05 elapsed=21:14:53 avg_step=3.67s recent_step=3.42s eta=29:42:24 +2026-09-18 14:44:19 | INFO | train rank=0 | step=20860/50000 progress=41.7% loss=0.080514 lr=6.744926e-05 lr_backbone=6.744926e-05 elapsed=21:15:43 avg_step=3.67s recent_step=4.97s eta=29:42:05 +2026-09-18 14:44:53 | INFO | train rank=0 | step=20870/50000 progress=41.7% loss=0.078045 lr=6.741827e-05 lr_backbone=6.741827e-05 elapsed=21:16:17 avg_step=3.67s recent_step=3.39s eta=29:41:25 +2026-09-18 14:45:23 | INFO | train rank=0 | step=20880/50000 progress=41.8% loss=0.073601 lr=6.738727e-05 lr_backbone=6.738727e-05 elapsed=21:16:47 avg_step=3.67s recent_step=2.99s eta=29:40:39 +2026-09-18 14:46:04 | INFO | train rank=0 | step=20890/50000 progress=41.8% loss=0.098771 lr=6.735626e-05 lr_backbone=6.735626e-05 elapsed=21:17:27 avg_step=3.67s recent_step=4.07s eta=29:40:07 +2026-09-18 14:46:46 | INFO | train rank=0 | step=20900/50000 progress=41.8% loss=0.103385 lr=6.732524e-05 lr_backbone=6.732524e-05 elapsed=21:18:09 avg_step=3.67s recent_step=4.17s eta=29:39:38 +2026-09-18 14:47:17 | INFO | train rank=0 | step=20910/50000 progress=41.8% loss=0.133667 lr=6.729422e-05 lr_backbone=6.729422e-05 elapsed=21:18:41 avg_step=3.67s recent_step=3.16s eta=29:38:54 +2026-09-18 14:47:52 | INFO | train rank=0 | step=20920/50000 progress=41.8% loss=0.092062 lr=6.726319e-05 lr_backbone=6.726319e-05 elapsed=21:19:15 avg_step=3.67s recent_step=3.46s eta=29:38:14 +2026-09-18 14:48:40 | INFO | train rank=0 | step=20930/50000 progress=41.9% loss=0.084330 lr=6.723215e-05 lr_backbone=6.723215e-05 elapsed=21:20:03 avg_step=3.67s recent_step=4.80s eta=29:37:53 +2026-09-18 14:49:16 | INFO | train rank=0 | step=20940/50000 progress=41.9% loss=0.085333 lr=6.720110e-05 lr_backbone=6.720110e-05 elapsed=21:20:39 avg_step=3.67s recent_step=3.57s eta=29:37:15 +2026-09-18 14:49:43 | INFO | train rank=0 | step=20950/50000 progress=41.9% loss=0.091807 lr=6.717004e-05 lr_backbone=6.717004e-05 elapsed=21:21:06 avg_step=3.67s recent_step=2.73s eta=29:36:26 +2026-09-18 14:50:15 | INFO | train rank=0 | step=20960/50000 progress=41.9% loss=0.114175 lr=6.713898e-05 lr_backbone=6.713898e-05 elapsed=21:21:38 avg_step=3.67s recent_step=3.22s eta=29:35:43 +2026-09-18 14:51:09 | INFO | train rank=0 | step=20970/50000 progress=41.9% loss=0.119260 lr=6.710791e-05 lr_backbone=6.710791e-05 elapsed=21:22:33 avg_step=3.67s recent_step=5.42s eta=29:35:30 +2026-09-18 14:51:47 | INFO | train rank=0 | step=20980/50000 progress=42.0% loss=0.067382 lr=6.707684e-05 lr_backbone=6.707684e-05 elapsed=21:23:11 avg_step=3.67s recent_step=3.80s eta=29:34:55 +2026-09-18 14:52:16 | INFO | train rank=0 | step=20990/50000 progress=42.0% loss=0.120919 lr=6.704575e-05 lr_backbone=6.704575e-05 elapsed=21:23:40 avg_step=3.67s recent_step=2.91s eta=29:34:08 +2026-09-18 14:52:56 | INFO | train rank=0 | step=21000/50000 progress=42.0% loss=0.079810 lr=6.701466e-05 lr_backbone=6.701466e-05 elapsed=21:24:19 avg_step=3.67s recent_step=3.96s eta=29:33:35 +2026-09-18 14:53:35 | INFO | train rank=0 | step=21010/50000 progress=42.0% loss=0.091369 lr=6.698356e-05 lr_backbone=6.698356e-05 elapsed=21:24:59 avg_step=3.67s recent_step=3.94s eta=29:33:03 +2026-09-18 14:54:22 | INFO | train rank=0 | step=21020/50000 progress=42.0% loss=0.146982 lr=6.695245e-05 lr_backbone=6.695245e-05 elapsed=21:25:45 avg_step=3.67s recent_step=4.67s eta=29:32:40 +2026-09-18 14:54:59 | INFO | train rank=0 | step=21030/50000 progress=42.1% loss=0.088382 lr=6.692134e-05 lr_backbone=6.692134e-05 elapsed=21:26:22 avg_step=3.67s recent_step=3.66s eta=29:32:03 +2026-09-18 14:55:33 | INFO | train rank=0 | step=21040/50000 progress=42.1% loss=0.093181 lr=6.689022e-05 lr_backbone=6.689022e-05 elapsed=21:26:56 avg_step=3.67s recent_step=3.44s eta=29:31:23 +2026-09-18 14:56:06 | INFO | train rank=0 | step=21050/50000 progress=42.1% loss=0.083234 lr=6.685909e-05 lr_backbone=6.685909e-05 elapsed=21:27:30 avg_step=3.67s recent_step=3.34s eta=29:30:42 +2026-09-18 14:56:56 | INFO | train rank=0 | step=21060/50000 progress=42.1% loss=0.105066 lr=6.682795e-05 lr_backbone=6.682795e-05 elapsed=21:28:19 avg_step=3.67s recent_step=4.91s eta=29:30:22 +2026-09-18 14:57:23 | INFO | train rank=0 | step=21070/50000 progress=42.1% loss=0.119700 lr=6.679681e-05 lr_backbone=6.679681e-05 elapsed=21:28:46 avg_step=3.67s recent_step=2.75s eta=29:29:33 +2026-09-18 14:57:58 | INFO | train rank=0 | step=21080/50000 progress=42.2% loss=0.085678 lr=6.676566e-05 lr_backbone=6.676566e-05 elapsed=21:29:22 avg_step=3.67s recent_step=3.53s eta=29:28:54 +2026-09-18 14:58:37 | INFO | train rank=0 | step=21090/50000 progress=42.2% loss=0.111360 lr=6.673450e-05 lr_backbone=6.673450e-05 elapsed=21:30:00 avg_step=3.67s recent_step=3.85s eta=29:28:20 +2026-09-18 14:59:19 | INFO | train rank=0 | step=21100/50000 progress=42.2% loss=0.126248 lr=6.670333e-05 lr_backbone=6.670333e-05 elapsed=21:30:43 avg_step=3.67s recent_step=4.25s eta=29:27:51 +2026-09-18 14:59:46 | INFO | train rank=0 | step=21110/50000 progress=42.2% loss=0.101861 lr=6.667216e-05 lr_backbone=6.667216e-05 elapsed=21:31:09 avg_step=3.67s recent_step=2.66s eta=29:27:00 +2026-09-18 15:00:19 | INFO | train rank=0 | step=21120/50000 progress=42.2% loss=0.081586 lr=6.664098e-05 lr_backbone=6.664098e-05 elapsed=21:31:42 avg_step=3.67s recent_step=3.27s eta=29:26:18 +2026-09-18 15:01:11 | INFO | train rank=0 | step=21130/50000 progress=42.3% loss=0.094562 lr=6.660979e-05 lr_backbone=6.660979e-05 elapsed=21:32:34 avg_step=3.67s recent_step=5.25s eta=29:26:03 +2026-09-18 15:01:46 | INFO | train rank=0 | step=21140/50000 progress=42.3% loss=0.084528 lr=6.657859e-05 lr_backbone=6.657859e-05 elapsed=21:33:09 avg_step=3.67s recent_step=3.47s eta=29:25:24 +2026-09-18 15:02:13 | INFO | train rank=0 | step=21150/50000 progress=42.3% loss=0.083004 lr=6.654739e-05 lr_backbone=6.654739e-05 elapsed=21:33:36 avg_step=3.67s recent_step=2.68s eta=29:24:33 +2026-09-18 15:02:46 | INFO | train rank=0 | step=21160/50000 progress=42.3% loss=0.109218 lr=6.651618e-05 lr_backbone=6.651618e-05 elapsed=21:34:09 avg_step=3.67s recent_step=3.30s eta=29:23:52 +2026-09-18 15:03:34 | INFO | train rank=0 | step=21170/50000 progress=42.3% loss=0.103869 lr=6.648497e-05 lr_backbone=6.648497e-05 elapsed=21:34:58 avg_step=3.67s recent_step=4.88s eta=29:23:31 +2026-09-18 15:04:14 | INFO | train rank=0 | step=21180/50000 progress=42.4% loss=0.078660 lr=6.645374e-05 lr_backbone=6.645374e-05 elapsed=21:35:37 avg_step=3.67s recent_step=3.95s eta=29:22:58 +2026-09-18 15:04:45 | INFO | train rank=0 | step=21190/50000 progress=42.4% loss=0.087265 lr=6.642251e-05 lr_backbone=6.642251e-05 elapsed=21:36:08 avg_step=3.67s recent_step=3.12s eta=29:22:14 +2026-09-18 15:05:16 | INFO | train rank=0 | step=21200/50000 progress=42.4% loss=0.097852 lr=6.639127e-05 lr_backbone=6.639127e-05 elapsed=21:36:39 avg_step=3.67s recent_step=3.07s eta=29:21:29 +2026-09-18 15:06:03 | INFO | train rank=0 | step=21210/50000 progress=42.4% loss=0.120682 lr=6.636003e-05 lr_backbone=6.636003e-05 elapsed=21:37:27 avg_step=3.67s recent_step=4.75s eta=29:21:07 +2026-09-18 15:06:42 | INFO | train rank=0 | step=21220/50000 progress=42.4% loss=0.073035 lr=6.632877e-05 lr_backbone=6.632877e-05 elapsed=21:38:06 avg_step=3.67s recent_step=3.92s eta=29:20:34 +2026-09-18 15:07:13 | INFO | train rank=0 | step=21230/50000 progress=42.5% loss=0.084569 lr=6.629751e-05 lr_backbone=6.629751e-05 elapsed=21:38:37 avg_step=3.67s recent_step=3.09s eta=29:19:49 +2026-09-18 15:07:43 | INFO | train rank=0 | step=21240/50000 progress=42.5% loss=0.104492 lr=6.626625e-05 lr_backbone=6.626625e-05 elapsed=21:39:06 avg_step=3.67s recent_step=2.96s eta=29:19:03 +2026-09-18 15:08:22 | INFO | train rank=0 | step=21250/50000 progress=42.5% loss=0.088931 lr=6.623497e-05 lr_backbone=6.623497e-05 elapsed=21:39:45 avg_step=3.67s recent_step=3.89s eta=29:18:29 +2026-09-18 15:08:49 | INFO | train rank=0 | step=21260/50000 progress=42.5% loss=0.157896 lr=6.620369e-05 lr_backbone=6.620369e-05 elapsed=21:40:12 avg_step=3.67s recent_step=2.73s eta=29:17:40 +2026-09-18 15:09:22 | INFO | train rank=0 | step=21270/50000 progress=42.5% loss=0.091778 lr=6.617240e-05 lr_backbone=6.617240e-05 elapsed=21:40:45 avg_step=3.67s recent_step=3.30s eta=29:16:58 +2026-09-18 15:09:54 | INFO | train rank=0 | step=21280/50000 progress=42.6% loss=0.093455 lr=6.614111e-05 lr_backbone=6.614111e-05 elapsed=21:41:18 avg_step=3.67s recent_step=3.21s eta=29:16:15 +2026-09-18 15:10:52 | INFO | train rank=0 | step=21290/50000 progress=42.6% loss=0.082639 lr=6.610981e-05 lr_backbone=6.610981e-05 elapsed=21:42:15 avg_step=3.67s recent_step=5.74s eta=29:16:07 +2026-09-18 15:11:24 | INFO | train rank=0 | step=21300/50000 progress=42.6% loss=0.074673 lr=6.607850e-05 lr_backbone=6.607850e-05 elapsed=21:42:48 avg_step=3.67s recent_step=3.26s eta=29:15:24 +2026-09-18 15:12:02 | INFO | train rank=0 | step=21310/50000 progress=42.6% loss=0.093832 lr=6.604718e-05 lr_backbone=6.604718e-05 elapsed=21:43:25 avg_step=3.67s recent_step=3.74s eta=29:14:49 +2026-09-18 15:12:33 | INFO | train rank=0 | step=21320/50000 progress=42.6% loss=0.110609 lr=6.601586e-05 lr_backbone=6.601586e-05 elapsed=21:43:57 avg_step=3.67s recent_step=3.17s eta=29:14:05 +2026-09-18 15:13:19 | INFO | train rank=0 | step=21330/50000 progress=42.7% loss=0.109179 lr=6.598453e-05 lr_backbone=6.598453e-05 elapsed=21:44:42 avg_step=3.67s recent_step=4.58s eta=29:13:41 +2026-09-18 15:13:52 | INFO | train rank=0 | step=21340/50000 progress=42.7% loss=0.084916 lr=6.595319e-05 lr_backbone=6.595319e-05 elapsed=21:45:16 avg_step=3.67s recent_step=3.34s eta=29:13:00 +2026-09-18 15:14:23 | INFO | train rank=0 | step=21350/50000 progress=42.7% loss=0.117051 lr=6.592184e-05 lr_backbone=6.592184e-05 elapsed=21:45:46 avg_step=3.67s recent_step=3.05s eta=29:12:15 +2026-09-18 15:14:55 | INFO | train rank=0 | step=21360/50000 progress=42.7% loss=0.092324 lr=6.589049e-05 lr_backbone=6.589049e-05 elapsed=21:46:18 avg_step=3.67s recent_step=3.18s eta=29:11:31 +2026-09-18 15:15:37 | INFO | train rank=0 | step=21370/50000 progress=42.7% loss=0.104109 lr=6.585913e-05 lr_backbone=6.585913e-05 elapsed=21:47:00 avg_step=3.67s recent_step=4.20s eta=29:11:02 +2026-09-18 15:16:07 | INFO | train rank=0 | step=21380/50000 progress=42.8% loss=0.102875 lr=6.582777e-05 lr_backbone=6.582777e-05 elapsed=21:47:31 avg_step=3.67s recent_step=3.06s eta=29:10:17 +2026-09-18 15:16:44 | INFO | train rank=0 | step=21390/50000 progress=42.8% loss=0.105810 lr=6.579640e-05 lr_backbone=6.579640e-05 elapsed=21:48:07 avg_step=3.67s recent_step=3.66s eta=29:09:40 +2026-09-18 15:17:20 | INFO | train rank=0 | step=21400/50000 progress=42.8% loss=0.087829 lr=6.576502e-05 lr_backbone=6.576502e-05 elapsed=21:48:43 avg_step=3.67s recent_step=3.59s eta=29:09:02 +2026-09-18 15:18:17 | INFO | train rank=0 | step=21410/50000 progress=42.8% loss=0.091512 lr=6.573363e-05 lr_backbone=6.573363e-05 elapsed=21:49:40 avg_step=3.67s recent_step=5.71s eta=29:08:53 +2026-09-18 15:18:53 | INFO | train rank=0 | step=21420/50000 progress=42.8% loss=0.110567 lr=6.570224e-05 lr_backbone=6.570224e-05 elapsed=21:50:17 avg_step=3.67s recent_step=3.63s eta=29:08:16 +2026-09-18 15:19:24 | INFO | train rank=0 | step=21430/50000 progress=42.9% loss=0.150326 lr=6.567084e-05 lr_backbone=6.567084e-05 elapsed=21:50:47 avg_step=3.67s recent_step=3.04s eta=29:07:31 +2026-09-18 15:19:57 | INFO | train rank=0 | step=21440/50000 progress=42.9% loss=0.091796 lr=6.563943e-05 lr_backbone=6.563943e-05 elapsed=21:51:20 avg_step=3.67s recent_step=3.33s eta=29:06:49 +2026-09-18 15:20:40 | INFO | train rank=0 | step=21450/50000 progress=42.9% loss=0.083986 lr=6.560802e-05 lr_backbone=6.560802e-05 elapsed=21:52:03 avg_step=3.67s recent_step=4.27s eta=29:06:21 +2026-09-18 15:21:12 | INFO | train rank=0 | step=21460/50000 progress=42.9% loss=0.078455 lr=6.557660e-05 lr_backbone=6.557660e-05 elapsed=21:52:35 avg_step=3.67s recent_step=3.21s eta=29:05:38 +2026-09-18 15:21:42 | INFO | train rank=0 | step=21470/50000 progress=42.9% loss=0.079214 lr=6.554517e-05 lr_backbone=6.554517e-05 elapsed=21:53:06 avg_step=3.67s recent_step=3.07s eta=29:04:53 +2026-09-18 15:22:17 | INFO | train rank=0 | step=21480/50000 progress=43.0% loss=0.106181 lr=6.551374e-05 lr_backbone=6.551374e-05 elapsed=21:53:41 avg_step=3.67s recent_step=3.49s eta=29:04:14 +2026-09-18 15:23:16 | INFO | train rank=0 | step=21490/50000 progress=43.0% loss=0.151013 lr=6.548230e-05 lr_backbone=6.548230e-05 elapsed=21:54:39 avg_step=3.67s recent_step=5.82s eta=29:04:06 +2026-09-18 15:23:48 | INFO | train rank=0 | step=21500/50000 progress=43.0% loss=0.089791 lr=6.545085e-05 lr_backbone=6.545085e-05 elapsed=21:55:11 avg_step=3.67s recent_step=3.22s eta=29:03:23 +2026-09-18 15:24:14 | INFO | train rank=0 | step=21510/50000 progress=43.0% loss=0.085946 lr=6.541940e-05 lr_backbone=6.541940e-05 elapsed=21:55:38 avg_step=3.67s recent_step=2.67s eta=29:02:33 +2026-09-18 15:24:49 | INFO | train rank=0 | step=21520/50000 progress=43.0% loss=0.078973 lr=6.538793e-05 lr_backbone=6.538793e-05 elapsed=21:56:13 avg_step=3.67s recent_step=3.49s eta=29:01:54 +2026-09-18 15:25:40 | INFO | train rank=0 | step=21530/50000 progress=43.1% loss=0.094446 lr=6.535647e-05 lr_backbone=6.535647e-05 elapsed=21:57:03 avg_step=3.67s recent_step=5.03s eta=29:01:35 +2026-09-18 15:26:10 | INFO | train rank=0 | step=21540/50000 progress=43.1% loss=0.084009 lr=6.532499e-05 lr_backbone=6.532499e-05 elapsed=21:57:33 avg_step=3.67s recent_step=3.03s eta=29:00:50 +2026-09-18 15:26:40 | INFO | train rank=0 | step=21550/50000 progress=43.1% loss=0.096103 lr=6.529351e-05 lr_backbone=6.529351e-05 elapsed=21:58:03 avg_step=3.67s recent_step=3.01s eta=29:00:05 +2026-09-18 15:27:12 | INFO | train rank=0 | step=21560/50000 progress=43.1% loss=0.075489 lr=6.526202e-05 lr_backbone=6.526202e-05 elapsed=21:58:35 avg_step=3.67s recent_step=3.19s eta=28:59:22 +2026-09-18 15:27:59 | INFO | train rank=0 | step=21570/50000 progress=43.1% loss=0.082006 lr=6.523053e-05 lr_backbone=6.523053e-05 elapsed=21:59:22 avg_step=3.67s recent_step=4.69s eta=28:58:59 +2026-09-18 15:28:39 | INFO | train rank=0 | step=21580/50000 progress=43.2% loss=0.107476 lr=6.519903e-05 lr_backbone=6.519903e-05 elapsed=22:00:03 avg_step=3.67s recent_step=4.06s eta=28:58:27 +2026-09-18 15:29:09 | INFO | train rank=0 | step=21590/50000 progress=43.2% loss=0.099996 lr=6.516752e-05 lr_backbone=6.516752e-05 elapsed=22:00:33 avg_step=3.67s recent_step=3.00s eta=28:57:42 +2026-09-18 15:29:41 | INFO | train rank=0 | step=21600/50000 progress=43.2% loss=0.078649 lr=6.513601e-05 lr_backbone=6.513601e-05 elapsed=22:01:04 avg_step=3.67s recent_step=3.11s eta=28:56:58 +2026-09-18 15:30:37 | INFO | train rank=0 | step=21610/50000 progress=43.2% loss=0.083553 lr=6.510448e-05 lr_backbone=6.510448e-05 elapsed=22:02:01 avg_step=3.67s recent_step=5.69s eta=28:56:47 +2026-09-18 15:31:03 | INFO | train rank=0 | step=21620/50000 progress=43.2% loss=0.120198 lr=6.507296e-05 lr_backbone=6.507296e-05 elapsed=22:02:26 avg_step=3.67s recent_step=2.56s eta=28:55:56 +2026-09-18 15:31:41 | INFO | train rank=0 | step=21630/50000 progress=43.3% loss=0.094079 lr=6.504142e-05 lr_backbone=6.504142e-05 elapsed=22:03:04 avg_step=3.67s recent_step=3.76s eta=28:55:21 +2026-09-18 15:32:10 | INFO | train rank=0 | step=21640/50000 progress=43.3% loss=0.099493 lr=6.500988e-05 lr_backbone=6.500988e-05 elapsed=22:03:33 avg_step=3.67s recent_step=2.93s eta=28:54:34 +2026-09-18 15:33:03 | INFO | train rank=0 | step=21650/50000 progress=43.3% loss=0.101822 lr=6.497833e-05 lr_backbone=6.497833e-05 elapsed=22:04:27 avg_step=3.67s recent_step=5.34s eta=28:54:19 +2026-09-18 15:33:39 | INFO | train rank=0 | step=21660/50000 progress=43.3% loss=0.078391 lr=6.494678e-05 lr_backbone=6.494678e-05 elapsed=22:05:02 avg_step=3.67s recent_step=3.55s eta=28:53:41 +2026-09-18 15:34:12 | INFO | train rank=0 | step=21670/50000 progress=43.3% loss=0.088960 lr=6.491522e-05 lr_backbone=6.491522e-05 elapsed=22:05:35 avg_step=3.67s recent_step=3.26s eta=28:52:59 +2026-09-18 15:34:46 | INFO | train rank=0 | step=21680/50000 progress=43.4% loss=0.100785 lr=6.488365e-05 lr_backbone=6.488365e-05 elapsed=22:06:09 avg_step=3.67s recent_step=3.45s eta=28:52:20 +2026-09-18 15:35:35 | INFO | train rank=0 | step=21690/50000 progress=43.4% loss=0.091172 lr=6.485208e-05 lr_backbone=6.485208e-05 elapsed=22:06:58 avg_step=3.67s recent_step=4.84s eta=28:51:58 +2026-09-18 15:36:09 | INFO | train rank=0 | step=21700/50000 progress=43.4% loss=0.118550 lr=6.482050e-05 lr_backbone=6.482050e-05 elapsed=22:07:32 avg_step=3.67s recent_step=3.43s eta=28:51:18 +2026-09-18 15:36:37 | INFO | train rank=0 | step=21710/50000 progress=43.4% loss=0.142157 lr=6.478891e-05 lr_backbone=6.478891e-05 elapsed=22:08:00 avg_step=3.67s recent_step=2.79s eta=28:50:30 +2026-09-18 15:37:08 | INFO | train rank=0 | step=21720/50000 progress=43.4% loss=0.084190 lr=6.475732e-05 lr_backbone=6.475732e-05 elapsed=22:08:32 avg_step=3.67s recent_step=3.18s eta=28:49:47 +2026-09-18 15:37:53 | INFO | train rank=0 | step=21730/50000 progress=43.5% loss=0.095794 lr=6.472572e-05 lr_backbone=6.472572e-05 elapsed=22:09:16 avg_step=3.67s recent_step=4.46s eta=28:49:21 +2026-09-18 15:38:23 | INFO | train rank=0 | step=21740/50000 progress=43.5% loss=0.123940 lr=6.469411e-05 lr_backbone=6.469411e-05 elapsed=22:09:47 avg_step=3.67s recent_step=3.04s eta=28:48:36 +2026-09-18 15:38:59 | INFO | train rank=0 | step=21750/50000 progress=43.5% loss=0.150998 lr=6.466250e-05 lr_backbone=6.466250e-05 elapsed=22:10:22 avg_step=3.67s recent_step=3.55s eta=28:47:57 +2026-09-18 15:39:26 | INFO | train rank=0 | step=21760/50000 progress=43.5% loss=0.088693 lr=6.463088e-05 lr_backbone=6.463088e-05 elapsed=22:10:50 avg_step=3.67s recent_step=2.73s eta=28:47:08 +2026-09-18 15:40:19 | INFO | train rank=0 | step=21770/50000 progress=43.5% loss=0.090795 lr=6.459926e-05 lr_backbone=6.459926e-05 elapsed=22:11:42 avg_step=3.67s recent_step=5.23s eta=28:46:52 +2026-09-18 15:40:52 | INFO | train rank=0 | step=21780/50000 progress=43.6% loss=0.124786 lr=6.456763e-05 lr_backbone=6.456763e-05 elapsed=22:12:15 avg_step=3.67s recent_step=3.31s eta=28:46:11 +2026-09-18 15:41:29 | INFO | train rank=0 | step=21790/50000 progress=43.6% loss=0.086227 lr=6.453599e-05 lr_backbone=6.453599e-05 elapsed=22:12:52 avg_step=3.67s recent_step=3.73s eta=28:45:35 +2026-09-18 15:42:00 | INFO | train rank=0 | step=21800/50000 progress=43.6% loss=0.113733 lr=6.450434e-05 lr_backbone=6.450434e-05 elapsed=22:13:23 avg_step=3.67s recent_step=3.11s eta=28:44:51 +2026-09-18 15:42:56 | INFO | train rank=0 | step=21810/50000 progress=43.6% loss=0.091527 lr=6.447269e-05 lr_backbone=6.447269e-05 elapsed=22:14:19 avg_step=3.67s recent_step=5.58s eta=28:44:39 +2026-09-18 15:43:21 | INFO | train rank=0 | step=21820/50000 progress=43.6% loss=0.081116 lr=6.444104e-05 lr_backbone=6.444104e-05 elapsed=22:14:45 avg_step=3.67s recent_step=2.54s eta=28:43:47 +2026-09-18 15:43:54 | INFO | train rank=0 | step=21830/50000 progress=43.7% loss=0.091352 lr=6.440937e-05 lr_backbone=6.440937e-05 elapsed=22:15:17 avg_step=3.67s recent_step=3.27s eta=28:43:06 +2026-09-18 15:44:28 | INFO | train rank=0 | step=21840/50000 progress=43.7% loss=0.067037 lr=6.437770e-05 lr_backbone=6.437770e-05 elapsed=22:15:52 avg_step=3.67s recent_step=3.45s eta=28:42:26 +2026-09-18 15:45:14 | INFO | train rank=0 | step=21850/50000 progress=43.7% loss=0.086472 lr=6.434603e-05 lr_backbone=6.434603e-05 elapsed=22:16:37 avg_step=3.67s recent_step=4.55s eta=28:42:01 +2026-09-18 15:45:44 | INFO | train rank=0 | step=21860/50000 progress=43.7% loss=0.073677 lr=6.431435e-05 lr_backbone=6.431435e-05 elapsed=22:17:07 avg_step=3.67s recent_step=2.98s eta=28:41:15 +2026-09-18 15:46:13 | INFO | train rank=0 | step=21870/50000 progress=43.7% loss=0.081208 lr=6.428266e-05 lr_backbone=6.428266e-05 elapsed=22:17:36 avg_step=3.67s recent_step=2.92s eta=28:40:29 +2026-09-18 15:46:40 | INFO | train rank=0 | step=21880/50000 progress=43.8% loss=0.128710 lr=6.425096e-05 lr_backbone=6.425096e-05 elapsed=22:18:04 avg_step=3.67s recent_step=2.72s eta=28:39:40 +2026-09-18 15:47:40 | INFO | train rank=0 | step=21890/50000 progress=43.8% loss=0.089598 lr=6.421926e-05 lr_backbone=6.421926e-05 elapsed=22:19:03 avg_step=3.67s recent_step=5.97s eta=28:39:33 +2026-09-18 15:48:06 | INFO | train rank=0 | step=21900/50000 progress=43.8% loss=0.075050 lr=6.418756e-05 lr_backbone=6.418756e-05 elapsed=22:19:29 avg_step=3.67s recent_step=2.63s eta=28:38:43 +2026-09-18 15:48:37 | INFO | train rank=0 | step=21910/50000 progress=43.8% loss=0.090459 lr=6.415584e-05 lr_backbone=6.415584e-05 elapsed=22:20:00 avg_step=3.67s recent_step=3.08s eta=28:37:58 +2026-09-18 15:49:07 | INFO | train rank=0 | step=21920/50000 progress=43.8% loss=0.112251 lr=6.412412e-05 lr_backbone=6.412412e-05 elapsed=22:20:30 avg_step=3.67s recent_step=3.02s eta=28:37:13 +2026-09-18 15:50:03 | INFO | train rank=0 | step=21930/50000 progress=43.9% loss=0.095016 lr=6.409240e-05 lr_backbone=6.409240e-05 elapsed=22:21:26 avg_step=3.67s recent_step=5.55s eta=28:37:01 +2026-09-18 15:50:35 | INFO | train rank=0 | step=21940/50000 progress=43.9% loss=0.111244 lr=6.406066e-05 lr_backbone=6.406066e-05 elapsed=22:21:58 avg_step=3.67s recent_step=3.23s eta=28:36:18 +2026-09-18 15:51:06 | INFO | train rank=0 | step=21950/50000 progress=43.9% loss=0.088914 lr=6.402893e-05 lr_backbone=6.402893e-05 elapsed=22:22:29 avg_step=3.67s recent_step=3.13s eta=28:35:35 +2026-09-18 15:51:43 | INFO | train rank=0 | step=21960/50000 progress=43.9% loss=0.101388 lr=6.399718e-05 lr_backbone=6.399718e-05 elapsed=22:23:06 avg_step=3.67s recent_step=3.69s eta=28:34:58 +2026-09-18 15:52:29 | INFO | train rank=0 | step=21970/50000 progress=43.9% loss=0.096837 lr=6.396543e-05 lr_backbone=6.396543e-05 elapsed=22:23:53 avg_step=3.67s recent_step=4.62s eta=28:34:34 +2026-09-18 15:52:55 | INFO | train rank=0 | step=21980/50000 progress=44.0% loss=0.072161 lr=6.393368e-05 lr_backbone=6.393368e-05 elapsed=22:24:18 avg_step=3.67s recent_step=2.54s eta=28:33:42 +2026-09-18 15:53:32 | INFO | train rank=0 | step=21990/50000 progress=44.0% loss=0.073976 lr=6.390191e-05 lr_backbone=6.390191e-05 elapsed=22:24:55 avg_step=3.67s recent_step=3.74s eta=28:33:07 +2026-09-18 15:54:10 | INFO | train rank=0 | step=22000/50000 progress=44.0% loss=0.106346 lr=6.387015e-05 lr_backbone=6.387015e-05 elapsed=22:25:34 avg_step=3.67s recent_step=3.85s eta=28:32:32 +2026-09-18 15:55:07 | INFO | train rank=0 | step=22010/50000 progress=44.0% loss=0.090533 lr=6.383837e-05 lr_backbone=6.383837e-05 elapsed=22:26:31 avg_step=3.67s recent_step=5.67s eta=28:32:21 +2026-09-18 15:55:33 | INFO | train rank=0 | step=22020/50000 progress=44.0% loss=0.072050 lr=6.380659e-05 lr_backbone=6.380659e-05 elapsed=22:26:57 avg_step=3.67s recent_step=2.61s eta=28:31:31 +2026-09-18 15:56:03 | INFO | train rank=0 | step=22030/50000 progress=44.1% loss=0.096771 lr=6.377480e-05 lr_backbone=6.377480e-05 elapsed=22:27:27 avg_step=3.67s recent_step=3.01s eta=28:30:46 +2026-09-18 15:56:42 | INFO | train rank=0 | step=22040/50000 progress=44.1% loss=0.083917 lr=6.374301e-05 lr_backbone=6.374301e-05 elapsed=22:28:05 avg_step=3.67s recent_step=3.83s eta=28:30:11 +2026-09-18 15:57:34 | INFO | train rank=0 | step=22050/50000 progress=44.1% loss=0.100745 lr=6.371121e-05 lr_backbone=6.371121e-05 elapsed=22:28:57 avg_step=3.67s recent_step=5.19s eta=28:29:54 +2026-09-18 15:58:10 | INFO | train rank=0 | step=22060/50000 progress=44.1% loss=0.078439 lr=6.367941e-05 lr_backbone=6.367941e-05 elapsed=22:29:33 avg_step=3.67s recent_step=3.61s eta=28:29:16 +2026-09-18 15:58:42 | INFO | train rank=0 | step=22070/50000 progress=44.1% loss=0.074919 lr=6.364760e-05 lr_backbone=6.364760e-05 elapsed=22:30:06 avg_step=3.67s recent_step=3.28s eta=28:28:34 +2026-09-18 15:59:14 | INFO | train rank=0 | step=22080/50000 progress=44.2% loss=0.087081 lr=6.361578e-05 lr_backbone=6.361578e-05 elapsed=22:30:38 avg_step=3.67s recent_step=3.19s eta=28:27:52 +2026-09-18 16:00:05 | INFO | train rank=0 | step=22090/50000 progress=44.2% loss=0.121703 lr=6.358396e-05 lr_backbone=6.358396e-05 elapsed=22:31:28 avg_step=3.67s recent_step=5.03s eta=28:27:32 +2026-09-18 16:00:40 | INFO | train rank=0 | step=22100/50000 progress=44.2% loss=0.084466 lr=6.355213e-05 lr_backbone=6.355213e-05 elapsed=22:32:03 avg_step=3.67s recent_step=3.54s eta=28:26:54 +2026-09-18 16:01:10 | INFO | train rank=0 | step=22110/50000 progress=44.2% loss=0.087656 lr=6.352029e-05 lr_backbone=6.352029e-05 elapsed=22:32:33 avg_step=3.67s recent_step=2.96s eta=28:26:08 +2026-09-18 16:01:41 | INFO | train rank=0 | step=22120/50000 progress=44.2% loss=0.102118 lr=6.348845e-05 lr_backbone=6.348845e-05 elapsed=22:33:04 avg_step=3.67s recent_step=3.08s eta=28:25:24 +2026-09-18 16:02:21 | INFO | train rank=0 | step=22130/50000 progress=44.3% loss=0.097221 lr=6.345661e-05 lr_backbone=6.345661e-05 elapsed=22:33:44 avg_step=3.67s recent_step=4.04s eta=28:24:52 +2026-09-18 16:03:00 | INFO | train rank=0 | step=22140/50000 progress=44.3% loss=0.122584 lr=6.342476e-05 lr_backbone=6.342476e-05 elapsed=22:34:23 avg_step=3.67s recent_step=3.89s eta=28:24:18 +2026-09-18 16:03:30 | INFO | train rank=0 | step=22150/50000 progress=44.3% loss=0.103176 lr=6.339290e-05 lr_backbone=6.339290e-05 elapsed=22:34:54 avg_step=3.67s recent_step=3.06s eta=28:23:34 +2026-09-18 16:03:59 | INFO | train rank=0 | step=22160/50000 progress=44.3% loss=0.114852 lr=6.336103e-05 lr_backbone=6.336103e-05 elapsed=22:35:22 avg_step=3.67s recent_step=2.83s eta=28:22:46 +2026-09-18 16:04:39 | INFO | train rank=0 | step=22170/50000 progress=44.3% loss=0.086179 lr=6.332916e-05 lr_backbone=6.332916e-05 elapsed=22:36:03 avg_step=3.67s recent_step=4.06s eta=28:22:15 +2026-09-18 16:05:13 | INFO | train rank=0 | step=22180/50000 progress=44.4% loss=0.086073 lr=6.329729e-05 lr_backbone=6.329729e-05 elapsed=22:36:37 avg_step=3.67s recent_step=3.41s eta=28:21:35 +2026-09-18 16:05:48 | INFO | train rank=0 | step=22190/50000 progress=44.4% loss=0.107588 lr=6.326541e-05 lr_backbone=6.326541e-05 elapsed=22:37:12 avg_step=3.67s recent_step=3.48s eta=28:20:56 +2026-09-18 16:06:23 | INFO | train rank=0 | step=22200/50000 progress=44.4% loss=0.079730 lr=6.323352e-05 lr_backbone=6.323352e-05 elapsed=22:37:46 avg_step=3.67s recent_step=3.44s eta=28:20:16 +2026-09-18 16:07:06 | INFO | train rank=0 | step=22210/50000 progress=44.4% loss=0.094045 lr=6.320163e-05 lr_backbone=6.320163e-05 elapsed=22:38:29 avg_step=3.67s recent_step=4.35s eta=28:19:48 +2026-09-18 16:07:44 | INFO | train rank=0 | step=22220/50000 progress=44.4% loss=0.103127 lr=6.316973e-05 lr_backbone=6.316973e-05 elapsed=22:39:07 avg_step=3.67s recent_step=3.77s eta=28:19:12 +2026-09-18 16:08:19 | INFO | train rank=0 | step=22230/50000 progress=44.5% loss=0.092860 lr=6.313782e-05 lr_backbone=6.313782e-05 elapsed=22:39:42 avg_step=3.67s recent_step=3.50s eta=28:18:34 +2026-09-18 16:08:53 | INFO | train rank=0 | step=22240/50000 progress=44.5% loss=0.095003 lr=6.310591e-05 lr_backbone=6.310591e-05 elapsed=22:40:16 avg_step=3.67s recent_step=3.42s eta=28:17:54 +2026-09-18 16:09:45 | INFO | train rank=0 | step=22250/50000 progress=44.5% loss=0.106949 lr=6.307400e-05 lr_backbone=6.307400e-05 elapsed=22:41:08 avg_step=3.67s recent_step=5.16s eta=28:17:36 +2026-09-18 16:10:14 | INFO | train rank=0 | step=22260/50000 progress=44.5% loss=0.135789 lr=6.304208e-05 lr_backbone=6.304208e-05 elapsed=22:41:38 avg_step=3.67s recent_step=2.97s eta=28:16:50 +2026-09-18 16:10:45 | INFO | train rank=0 | step=22270/50000 progress=44.5% loss=0.086537 lr=6.301015e-05 lr_backbone=6.301015e-05 elapsed=22:42:08 avg_step=3.67s recent_step=3.03s eta=28:16:06 +2026-09-18 16:11:19 | INFO | train rank=0 | step=22280/50000 progress=44.6% loss=0.094239 lr=6.297821e-05 lr_backbone=6.297821e-05 elapsed=22:42:42 avg_step=3.67s recent_step=3.39s eta=28:15:25 +2026-09-18 16:12:08 | INFO | train rank=0 | step=22290/50000 progress=44.6% loss=0.084650 lr=6.294628e-05 lr_backbone=6.294628e-05 elapsed=22:43:31 avg_step=3.67s recent_step=4.94s eta=28:15:05 +2026-09-18 16:12:37 | INFO | train rank=0 | step=22300/50000 progress=44.6% loss=0.128317 lr=6.291433e-05 lr_backbone=6.291433e-05 elapsed=22:44:00 avg_step=3.67s recent_step=2.88s eta=28:14:18 +2026-09-18 16:13:05 | INFO | train rank=0 | step=22310/50000 progress=44.6% loss=0.082118 lr=6.288238e-05 lr_backbone=6.288238e-05 elapsed=22:44:28 avg_step=3.67s recent_step=2.83s eta=28:13:31 +2026-09-18 16:13:37 | INFO | train rank=0 | step=22320/50000 progress=44.6% loss=0.108844 lr=6.285043e-05 lr_backbone=6.285043e-05 elapsed=22:45:00 avg_step=3.67s recent_step=3.20s eta=28:12:48 +2026-09-18 16:14:29 | INFO | train rank=0 | step=22330/50000 progress=44.7% loss=0.128303 lr=6.281846e-05 lr_backbone=6.281846e-05 elapsed=22:45:52 avg_step=3.67s recent_step=5.15s eta=28:12:30 +2026-09-18 16:14:56 | INFO | train rank=0 | step=22340/50000 progress=44.7% loss=0.104415 lr=6.278650e-05 lr_backbone=6.278650e-05 elapsed=22:46:19 avg_step=3.67s recent_step=2.75s eta=28:11:42 +2026-09-18 16:15:29 | INFO | train rank=0 | step=22350/50000 progress=44.7% loss=0.133140 lr=6.275452e-05 lr_backbone=6.275452e-05 elapsed=22:46:52 avg_step=3.67s recent_step=3.27s eta=28:11:00 +2026-09-18 16:16:00 | INFO | train rank=0 | step=22360/50000 progress=44.7% loss=0.113675 lr=6.272255e-05 lr_backbone=6.272255e-05 elapsed=22:47:23 avg_step=3.67s recent_step=3.07s eta=28:10:16 +2026-09-18 16:16:51 | INFO | train rank=0 | step=22370/50000 progress=44.7% loss=0.117559 lr=6.269056e-05 lr_backbone=6.269056e-05 elapsed=22:48:14 avg_step=3.67s recent_step=5.13s eta=28:09:58 +2026-09-18 16:17:24 | INFO | train rank=0 | step=22380/50000 progress=44.8% loss=0.118288 lr=6.265857e-05 lr_backbone=6.265857e-05 elapsed=22:48:47 avg_step=3.67s recent_step=3.32s eta=28:09:17 +2026-09-18 16:17:56 | INFO | train rank=0 | step=22390/50000 progress=44.8% loss=0.118720 lr=6.262658e-05 lr_backbone=6.262658e-05 elapsed=22:49:19 avg_step=3.67s recent_step=3.19s eta=28:08:34 +2026-09-18 16:18:29 | INFO | train rank=0 | step=22400/50000 progress=44.8% loss=0.082351 lr=6.259458e-05 lr_backbone=6.259458e-05 elapsed=22:49:52 avg_step=3.67s recent_step=3.26s eta=28:07:52 +2026-09-18 16:19:14 | INFO | train rank=0 | step=22410/50000 progress=44.8% loss=0.124910 lr=6.256257e-05 lr_backbone=6.256257e-05 elapsed=22:50:38 avg_step=3.67s recent_step=4.56s eta=28:07:27 +2026-09-18 16:19:54 | INFO | train rank=0 | step=22420/50000 progress=44.8% loss=0.129448 lr=6.253056e-05 lr_backbone=6.253056e-05 elapsed=22:51:18 avg_step=3.67s recent_step=4.01s eta=28:06:54 +2026-09-18 16:20:17 | INFO | train rank=0 | step=22430/50000 progress=44.9% loss=0.082488 lr=6.249854e-05 lr_backbone=6.249854e-05 elapsed=22:51:40 avg_step=3.67s recent_step=2.28s eta=28:06:00 +2026-09-18 16:20:52 | INFO | train rank=0 | step=22440/50000 progress=44.9% loss=0.130415 lr=6.246652e-05 lr_backbone=6.246652e-05 elapsed=22:52:15 avg_step=3.67s recent_step=3.48s eta=28:05:21 +2026-09-18 16:21:31 | INFO | train rank=0 | step=22450/50000 progress=44.9% loss=0.133052 lr=6.243449e-05 lr_backbone=6.243449e-05 elapsed=22:52:54 avg_step=3.67s recent_step=3.91s eta=28:04:48 +2026-09-18 16:22:17 | INFO | train rank=0 | step=22460/50000 progress=44.9% loss=0.096399 lr=6.240246e-05 lr_backbone=6.240246e-05 elapsed=22:53:41 avg_step=3.67s recent_step=4.64s eta=28:04:23 +2026-09-18 16:22:54 | INFO | train rank=0 | step=22470/50000 progress=44.9% loss=0.094670 lr=6.237042e-05 lr_backbone=6.237042e-05 elapsed=22:54:17 avg_step=3.67s recent_step=3.66s eta=28:03:46 +2026-09-18 16:23:26 | INFO | train rank=0 | step=22480/50000 progress=45.0% loss=0.091914 lr=6.233838e-05 lr_backbone=6.233838e-05 elapsed=22:54:49 avg_step=3.67s recent_step=3.22s eta=28:03:04 +2026-09-18 16:23:59 | INFO | train rank=0 | step=22490/50000 progress=45.0% loss=0.112730 lr=6.230633e-05 lr_backbone=6.230633e-05 elapsed=22:55:22 avg_step=3.67s recent_step=3.30s eta=28:02:22 +2026-09-18 16:24:43 | INFO | train rank=0 | step=22500/50000 progress=45.0% loss=0.087376 lr=6.227427e-05 lr_backbone=6.227427e-05 elapsed=22:56:06 avg_step=3.67s recent_step=4.35s eta=28:01:54 +2026-09-18 16:25:19 | INFO | train rank=0 | step=22510/50000 progress=45.0% loss=0.089851 lr=6.224221e-05 lr_backbone=6.224221e-05 elapsed=22:56:42 avg_step=3.67s recent_step=3.61s eta=28:01:17 +2026-09-18 16:25:51 | INFO | train rank=0 | step=22520/50000 progress=45.0% loss=0.100201 lr=6.221015e-05 lr_backbone=6.221015e-05 elapsed=22:57:14 avg_step=3.67s recent_step=3.20s eta=28:00:34 +2026-09-18 16:26:27 | INFO | train rank=0 | step=22530/50000 progress=45.1% loss=0.075309 lr=6.217808e-05 lr_backbone=6.217808e-05 elapsed=22:57:50 avg_step=3.67s recent_step=3.62s eta=27:59:57 +2026-09-18 16:27:16 | INFO | train rank=0 | step=22540/50000 progress=45.1% loss=0.069046 lr=6.214600e-05 lr_backbone=6.214600e-05 elapsed=22:58:39 avg_step=3.67s recent_step=4.89s eta=27:59:35 +2026-09-18 16:27:50 | INFO | train rank=0 | step=22550/50000 progress=45.1% loss=0.092165 lr=6.211392e-05 lr_backbone=6.211392e-05 elapsed=22:59:14 avg_step=3.67s recent_step=3.44s eta=27:58:56 +2026-09-18 16:28:25 | INFO | train rank=0 | step=22560/50000 progress=45.1% loss=0.122383 lr=6.208183e-05 lr_backbone=6.208183e-05 elapsed=22:59:49 avg_step=3.67s recent_step=3.51s eta=27:58:17 +2026-09-18 16:29:00 | INFO | train rank=0 | step=22570/50000 progress=45.1% loss=0.156676 lr=6.204974e-05 lr_backbone=6.204974e-05 elapsed=23:00:23 avg_step=3.67s recent_step=3.44s eta=27:57:37 +2026-09-18 16:29:41 | INFO | train rank=0 | step=22580/50000 progress=45.2% loss=0.101547 lr=6.201764e-05 lr_backbone=6.201764e-05 elapsed=23:01:04 avg_step=3.67s recent_step=4.11s eta=27:57:06 +2026-09-18 16:30:12 | INFO | train rank=0 | step=22590/50000 progress=45.2% loss=0.085673 lr=6.198554e-05 lr_backbone=6.198554e-05 elapsed=23:01:35 avg_step=3.67s recent_step=3.08s eta=27:56:22 +2026-09-18 16:30:52 | INFO | train rank=0 | step=22600/50000 progress=45.2% loss=0.101546 lr=6.195343e-05 lr_backbone=6.195343e-05 elapsed=23:02:15 avg_step=3.67s recent_step=4.00s eta=27:55:50 +2026-09-18 16:31:18 | INFO | train rank=0 | step=22610/50000 progress=45.2% loss=0.089367 lr=6.192132e-05 lr_backbone=6.192132e-05 elapsed=23:02:42 avg_step=3.67s recent_step=2.68s eta=27:55:01 +2026-09-18 16:32:03 | INFO | train rank=0 | step=22620/50000 progress=45.2% loss=0.092732 lr=6.188920e-05 lr_backbone=6.188920e-05 elapsed=23:03:26 avg_step=3.67s recent_step=4.44s eta=27:54:34 +2026-09-18 16:32:34 | INFO | train rank=0 | step=22630/50000 progress=45.3% loss=0.137198 lr=6.185708e-05 lr_backbone=6.185708e-05 elapsed=23:03:58 avg_step=3.67s recent_step=3.14s eta=27:53:50 +2026-09-18 16:33:08 | INFO | train rank=0 | step=22640/50000 progress=45.3% loss=0.116258 lr=6.182495e-05 lr_backbone=6.182495e-05 elapsed=23:04:31 avg_step=3.67s recent_step=3.32s eta=27:53:10 +2026-09-18 16:34:07 | INFO | train rank=0 | step=22650/50000 progress=45.3% loss=0.097259 lr=6.179282e-05 lr_backbone=6.179282e-05 elapsed=23:05:30 avg_step=3.67s recent_step=5.94s eta=27:53:00 +2026-09-18 16:34:43 | INFO | train rank=0 | step=22660/50000 progress=45.3% loss=0.094566 lr=6.176068e-05 lr_backbone=6.176068e-05 elapsed=23:06:07 avg_step=3.67s recent_step=3.64s eta=27:52:23 +2026-09-18 16:35:12 | INFO | train rank=0 | step=22670/50000 progress=45.3% loss=0.069827 lr=6.172853e-05 lr_backbone=6.172853e-05 elapsed=23:06:35 avg_step=3.67s recent_step=2.83s eta=27:51:36 +2026-09-18 16:35:52 | INFO | train rank=0 | step=22680/50000 progress=45.4% loss=0.112716 lr=6.169638e-05 lr_backbone=6.169638e-05 elapsed=23:07:15 avg_step=3.67s recent_step=4.05s eta=27:51:04 +2026-09-18 16:36:43 | INFO | train rank=0 | step=22690/50000 progress=45.4% loss=0.107197 lr=6.166423e-05 lr_backbone=6.166423e-05 elapsed=23:08:06 avg_step=3.67s recent_step=5.07s eta=27:50:44 +2026-09-18 16:37:12 | INFO | train rank=0 | step=22700/50000 progress=45.4% loss=0.078800 lr=6.163207e-05 lr_backbone=6.163207e-05 elapsed=23:08:35 avg_step=3.67s recent_step=2.90s eta=27:49:58 +2026-09-18 16:37:45 | INFO | train rank=0 | step=22710/50000 progress=45.4% loss=0.096937 lr=6.159991e-05 lr_backbone=6.159991e-05 elapsed=23:09:08 avg_step=3.67s recent_step=3.33s eta=27:49:18 +2026-09-18 16:38:18 | INFO | train rank=0 | step=22720/50000 progress=45.4% loss=0.144353 lr=6.156774e-05 lr_backbone=6.156774e-05 elapsed=23:09:42 avg_step=3.67s recent_step=3.32s eta=27:48:37 +2026-09-18 16:39:10 | INFO | train rank=0 | step=22730/50000 progress=45.5% loss=0.110697 lr=6.153556e-05 lr_backbone=6.153556e-05 elapsed=23:10:34 avg_step=3.67s recent_step=5.18s eta=27:48:18 +2026-09-18 16:39:40 | INFO | train rank=0 | step=22740/50000 progress=45.5% loss=0.089227 lr=6.150338e-05 lr_backbone=6.150338e-05 elapsed=23:11:04 avg_step=3.67s recent_step=3.00s eta=27:47:34 +2026-09-18 16:40:10 | INFO | train rank=0 | step=22750/50000 progress=45.5% loss=0.073939 lr=6.147120e-05 lr_backbone=6.147120e-05 elapsed=23:11:34 avg_step=3.67s recent_step=3.00s eta=27:46:49 +2026-09-18 16:40:46 | INFO | train rank=0 | step=22760/50000 progress=45.5% loss=0.070302 lr=6.143901e-05 lr_backbone=6.143901e-05 elapsed=23:12:09 avg_step=3.67s recent_step=3.55s eta=27:46:11 +2026-09-18 16:41:35 | INFO | train rank=0 | step=22770/50000 progress=45.5% loss=0.108404 lr=6.140681e-05 lr_backbone=6.140681e-05 elapsed=23:12:58 avg_step=3.67s recent_step=4.93s eta=27:45:49 +2026-09-18 16:42:07 | INFO | train rank=0 | step=22780/50000 progress=45.6% loss=0.118428 lr=6.137461e-05 lr_backbone=6.137461e-05 elapsed=23:13:30 avg_step=3.67s recent_step=3.20s eta=27:45:07 +2026-09-18 16:42:39 | INFO | train rank=0 | step=22790/50000 progress=45.6% loss=0.096976 lr=6.134241e-05 lr_backbone=6.134241e-05 elapsed=23:14:03 avg_step=3.67s recent_step=3.24s eta=27:44:25 +2026-09-18 16:43:09 | INFO | train rank=0 | step=22800/50000 progress=45.6% loss=0.184441 lr=6.131020e-05 lr_backbone=6.131020e-05 elapsed=23:14:33 avg_step=3.67s recent_step=3.00s eta=27:43:40 +2026-09-18 16:43:58 | INFO | train rank=0 | step=22810/50000 progress=45.6% loss=0.089428 lr=6.127798e-05 lr_backbone=6.127798e-05 elapsed=23:15:22 avg_step=3.67s recent_step=4.90s eta=27:43:18 +2026-09-18 16:44:35 | INFO | train rank=0 | step=22820/50000 progress=45.6% loss=0.146975 lr=6.124576e-05 lr_backbone=6.124576e-05 elapsed=23:15:58 avg_step=3.67s recent_step=3.62s eta=27:42:41 +2026-09-18 16:45:03 | INFO | train rank=0 | step=22830/50000 progress=45.7% loss=0.080814 lr=6.121354e-05 lr_backbone=6.121354e-05 elapsed=23:16:27 avg_step=3.67s recent_step=2.88s eta=27:41:55 +2026-09-18 16:45:40 | INFO | train rank=0 | step=22840/50000 progress=45.7% loss=0.091214 lr=6.118131e-05 lr_backbone=6.118131e-05 elapsed=23:17:03 avg_step=3.67s recent_step=3.64s eta=27:41:18 +2026-09-18 16:46:31 | INFO | train rank=0 | step=22850/50000 progress=45.7% loss=0.118085 lr=6.114907e-05 lr_backbone=6.114907e-05 elapsed=23:17:54 avg_step=3.67s recent_step=5.09s eta=27:40:58 +2026-09-18 16:47:04 | INFO | train rank=0 | step=22860/50000 progress=45.7% loss=0.131503 lr=6.111683e-05 lr_backbone=6.111683e-05 elapsed=23:18:27 avg_step=3.67s recent_step=3.34s eta=27:40:17 +2026-09-18 16:47:36 | INFO | train rank=0 | step=22870/50000 progress=45.7% loss=0.122012 lr=6.108459e-05 lr_backbone=6.108459e-05 elapsed=23:19:00 avg_step=3.67s recent_step=3.22s eta=27:39:35 +2026-09-18 16:48:10 | INFO | train rank=0 | step=22880/50000 progress=45.8% loss=0.084913 lr=6.105234e-05 lr_backbone=6.105234e-05 elapsed=23:19:34 avg_step=3.67s recent_step=3.40s eta=27:38:55 +2026-09-18 16:49:03 | INFO | train rank=0 | step=22890/50000 progress=45.8% loss=0.083429 lr=6.102009e-05 lr_backbone=6.102009e-05 elapsed=23:20:26 avg_step=3.67s recent_step=5.24s eta=27:38:37 +2026-09-18 16:49:32 | INFO | train rank=0 | step=22900/50000 progress=45.8% loss=0.082052 lr=6.098783e-05 lr_backbone=6.098783e-05 elapsed=23:20:55 avg_step=3.67s recent_step=2.91s eta=27:37:51 +2026-09-18 16:50:06 | INFO | train rank=0 | step=22910/50000 progress=45.8% loss=0.094883 lr=6.095557e-05 lr_backbone=6.095557e-05 elapsed=23:21:29 avg_step=3.67s recent_step=3.40s eta=27:37:12 +2026-09-18 16:50:37 | INFO | train rank=0 | step=22920/50000 progress=45.8% loss=0.083731 lr=6.092330e-05 lr_backbone=6.092330e-05 elapsed=23:22:00 avg_step=3.67s recent_step=3.10s eta=27:36:28 +2026-09-18 16:51:25 | INFO | train rank=0 | step=22930/50000 progress=45.9% loss=0.084799 lr=6.089102e-05 lr_backbone=6.089102e-05 elapsed=23:22:48 avg_step=3.67s recent_step=4.82s eta=27:36:05 +2026-09-18 16:51:57 | INFO | train rank=0 | step=22940/50000 progress=45.9% loss=0.083542 lr=6.085875e-05 lr_backbone=6.085875e-05 elapsed=23:23:20 avg_step=3.67s recent_step=3.16s eta=27:35:22 +2026-09-18 16:52:28 | INFO | train rank=0 | step=22950/50000 progress=45.9% loss=0.101601 lr=6.082646e-05 lr_backbone=6.082646e-05 elapsed=23:23:51 avg_step=3.67s recent_step=3.13s eta=27:34:39 +2026-09-18 16:52:58 | INFO | train rank=0 | step=22960/50000 progress=45.9% loss=0.108510 lr=6.079418e-05 lr_backbone=6.079418e-05 elapsed=23:24:21 avg_step=3.67s recent_step=2.97s eta=27:33:54 +2026-09-18 16:53:47 | INFO | train rank=0 | step=22970/50000 progress=45.9% loss=0.144161 lr=6.076189e-05 lr_backbone=6.076189e-05 elapsed=23:25:10 avg_step=3.67s recent_step=4.93s eta=27:33:32 +2026-09-18 16:54:17 | INFO | train rank=0 | step=22980/50000 progress=46.0% loss=0.108860 lr=6.072959e-05 lr_backbone=6.072959e-05 elapsed=23:25:41 avg_step=3.67s recent_step=3.03s eta=27:32:48 +2026-09-18 16:54:50 | INFO | train rank=0 | step=22990/50000 progress=46.0% loss=0.075615 lr=6.069729e-05 lr_backbone=6.069729e-05 elapsed=23:26:13 avg_step=3.67s recent_step=3.28s eta=27:32:07 +2026-09-18 16:55:25 | INFO | train rank=0 | step=23000/50000 progress=46.0% loss=0.096663 lr=6.066498e-05 lr_backbone=6.066498e-05 elapsed=23:26:48 avg_step=3.67s recent_step=3.48s eta=27:31:28 +2026-09-18 16:56:01 | INFO | train rank=0 | step=23010/50000 progress=46.0% loss=0.102456 lr=6.063267e-05 lr_backbone=6.063267e-05 elapsed=23:27:25 avg_step=3.67s recent_step=3.64s eta=27:30:51 +2026-09-18 16:56:30 | INFO | train rank=0 | step=23020/50000 progress=46.0% loss=0.088148 lr=6.060036e-05 lr_backbone=6.060036e-05 elapsed=23:27:53 avg_step=3.67s recent_step=2.86s eta=27:30:05 +2026-09-18 16:57:00 | INFO | train rank=0 | step=23030/50000 progress=46.1% loss=0.097055 lr=6.056804e-05 lr_backbone=6.056804e-05 elapsed=23:28:24 avg_step=3.67s recent_step=3.07s eta=27:29:21 +2026-09-18 16:57:55 | INFO | train rank=0 | step=23040/50000 progress=46.1% loss=0.109354 lr=6.053571e-05 lr_backbone=6.053571e-05 elapsed=23:29:18 avg_step=3.67s recent_step=5.44s eta=27:29:05 +2026-09-18 16:58:33 | INFO | train rank=0 | step=23050/50000 progress=46.1% loss=0.103899 lr=6.050338e-05 lr_backbone=6.050338e-05 elapsed=23:29:57 avg_step=3.67s recent_step=3.86s eta=27:28:30 +2026-09-18 16:59:05 | INFO | train rank=0 | step=23060/50000 progress=46.1% loss=0.085983 lr=6.047105e-05 lr_backbone=6.047105e-05 elapsed=23:30:29 avg_step=3.67s recent_step=3.19s eta=27:27:48 +2026-09-18 16:59:40 | INFO | train rank=0 | step=23070/50000 progress=46.1% loss=0.130740 lr=6.043871e-05 lr_backbone=6.043871e-05 elapsed=23:31:04 avg_step=3.67s recent_step=3.50s eta=27:27:09 +2026-09-18 17:00:26 | INFO | train rank=0 | step=23080/50000 progress=46.2% loss=0.151374 lr=6.040637e-05 lr_backbone=6.040637e-05 elapsed=23:31:49 avg_step=3.67s recent_step=4.52s eta=27:26:43 +2026-09-18 17:00:56 | INFO | train rank=0 | step=23090/50000 progress=46.2% loss=0.114435 lr=6.037402e-05 lr_backbone=6.037402e-05 elapsed=23:32:19 avg_step=3.67s recent_step=3.05s eta=27:25:59 +2026-09-18 17:01:23 | INFO | train rank=0 | step=23100/50000 progress=46.2% loss=0.107933 lr=6.034167e-05 lr_backbone=6.034167e-05 elapsed=23:32:46 avg_step=3.67s recent_step=2.68s eta=27:25:10 +2026-09-18 17:01:57 | INFO | train rank=0 | step=23110/50000 progress=46.2% loss=0.103058 lr=6.030931e-05 lr_backbone=6.030931e-05 elapsed=23:33:20 avg_step=3.67s recent_step=3.38s eta=27:24:30 +2026-09-18 17:02:50 | INFO | train rank=0 | step=23120/50000 progress=46.2% loss=0.104205 lr=6.027695e-05 lr_backbone=6.027695e-05 elapsed=23:34:13 avg_step=3.67s recent_step=5.35s eta=27:24:13 +2026-09-18 17:03:26 | INFO | train rank=0 | step=23130/50000 progress=46.3% loss=0.107957 lr=6.024458e-05 lr_backbone=6.024458e-05 elapsed=23:34:49 avg_step=3.67s recent_step=3.56s eta=27:23:35 +2026-09-18 17:04:04 | INFO | train rank=0 | step=23140/50000 progress=46.3% loss=0.085498 lr=6.021221e-05 lr_backbone=6.021221e-05 elapsed=23:35:28 avg_step=3.67s recent_step=3.84s eta=27:23:01 +2026-09-18 17:04:33 | INFO | train rank=0 | step=23150/50000 progress=46.3% loss=0.085733 lr=6.017984e-05 lr_backbone=6.017984e-05 elapsed=23:35:56 avg_step=3.67s recent_step=2.87s eta=27:22:15 +2026-09-18 17:05:13 | INFO | train rank=0 | step=23160/50000 progress=46.3% loss=0.119810 lr=6.014746e-05 lr_backbone=6.014746e-05 elapsed=23:36:36 avg_step=3.67s recent_step=3.98s eta=27:21:42 +2026-09-18 17:06:00 | INFO | train rank=0 | step=23170/50000 progress=46.3% loss=0.092334 lr=6.011508e-05 lr_backbone=6.011508e-05 elapsed=23:37:24 avg_step=3.67s recent_step=4.76s eta=27:21:17 +2026-09-18 17:06:34 | INFO | train rank=0 | step=23180/50000 progress=46.4% loss=0.078705 lr=6.008269e-05 lr_backbone=6.008269e-05 elapsed=23:37:57 avg_step=3.67s recent_step=3.32s eta=27:20:37 +2026-09-18 17:07:09 | INFO | train rank=0 | step=23190/50000 progress=46.4% loss=0.082676 lr=6.005030e-05 lr_backbone=6.005030e-05 elapsed=23:38:32 avg_step=3.67s recent_step=3.51s eta=27:19:58 +2026-09-18 17:07:37 | INFO | train rank=0 | step=23200/50000 progress=46.4% loss=0.117885 lr=6.001790e-05 lr_backbone=6.001790e-05 elapsed=23:39:00 avg_step=3.67s recent_step=2.84s eta=27:19:12 +2026-09-18 17:08:28 | INFO | train rank=0 | step=23210/50000 progress=46.4% loss=0.091570 lr=5.998550e-05 lr_backbone=5.998550e-05 elapsed=23:39:51 avg_step=3.67s recent_step=5.10s eta=27:18:52 +2026-09-18 17:09:05 | INFO | train rank=0 | step=23220/50000 progress=46.4% loss=0.087728 lr=5.995309e-05 lr_backbone=5.995309e-05 elapsed=23:40:29 avg_step=3.67s recent_step=3.71s eta=27:18:16 +2026-09-18 17:09:37 | INFO | train rank=0 | step=23230/50000 progress=46.5% loss=0.109523 lr=5.992068e-05 lr_backbone=5.992068e-05 elapsed=23:41:00 avg_step=3.67s recent_step=3.19s eta=27:17:33 +2026-09-18 17:10:10 | INFO | train rank=0 | step=23240/50000 progress=46.5% loss=0.100166 lr=5.988827e-05 lr_backbone=5.988827e-05 elapsed=23:41:33 avg_step=3.67s recent_step=3.29s eta=27:16:52 +2026-09-18 17:10:58 | INFO | train rank=0 | step=23250/50000 progress=46.5% loss=0.108831 lr=5.985585e-05 lr_backbone=5.985585e-05 elapsed=23:42:21 avg_step=3.67s recent_step=4.79s eta=27:16:28 +2026-09-18 17:11:21 | INFO | train rank=0 | step=23260/50000 progress=46.5% loss=0.102379 lr=5.982343e-05 lr_backbone=5.982343e-05 elapsed=23:42:45 avg_step=3.67s recent_step=2.33s eta=27:15:36 +2026-09-18 17:11:52 | INFO | train rank=0 | step=23270/50000 progress=46.5% loss=0.100101 lr=5.979100e-05 lr_backbone=5.979100e-05 elapsed=23:43:15 avg_step=3.67s recent_step=3.03s eta=27:14:52 +2026-09-18 17:12:21 | INFO | train rank=0 | step=23280/50000 progress=46.6% loss=0.089614 lr=5.975857e-05 lr_backbone=5.975857e-05 elapsed=23:43:44 avg_step=3.67s recent_step=2.94s eta=27:14:07 +2026-09-18 17:13:08 | INFO | train rank=0 | step=23290/50000 progress=46.6% loss=0.099374 lr=5.972613e-05 lr_backbone=5.972613e-05 elapsed=23:44:31 avg_step=3.67s recent_step=4.71s eta=27:13:42 +2026-09-18 17:13:49 | INFO | train rank=0 | step=23300/50000 progress=46.6% loss=0.125113 lr=5.969369e-05 lr_backbone=5.969369e-05 elapsed=23:45:12 avg_step=3.67s recent_step=4.09s eta=27:13:10 +2026-09-18 17:14:21 | INFO | train rank=0 | step=23310/50000 progress=46.6% loss=0.112829 lr=5.966125e-05 lr_backbone=5.966125e-05 elapsed=23:45:44 avg_step=3.67s recent_step=3.18s eta=27:12:28 +2026-09-18 17:14:55 | INFO | train rank=0 | step=23320/50000 progress=46.6% loss=0.096013 lr=5.962880e-05 lr_backbone=5.962880e-05 elapsed=23:46:19 avg_step=3.67s recent_step=3.46s eta=27:11:49 +2026-09-18 17:15:32 | INFO | train rank=0 | step=23330/50000 progress=46.7% loss=0.102781 lr=5.959635e-05 lr_backbone=5.959635e-05 elapsed=23:46:55 avg_step=3.67s recent_step=3.63s eta=27:11:12 +2026-09-18 17:16:04 | INFO | train rank=0 | step=23340/50000 progress=46.7% loss=0.108104 lr=5.956389e-05 lr_backbone=5.956389e-05 elapsed=23:47:27 avg_step=3.67s recent_step=3.26s eta=27:10:30 +2026-09-18 17:16:58 | INFO | train rank=0 | step=23350/50000 progress=46.7% loss=0.077651 lr=5.953143e-05 lr_backbone=5.953143e-05 elapsed=23:48:21 avg_step=3.67s recent_step=5.35s eta=27:10:13 +2026-09-18 17:17:27 | INFO | train rank=0 | step=23360/50000 progress=46.7% loss=0.096493 lr=5.949897e-05 lr_backbone=5.949897e-05 elapsed=23:48:51 avg_step=3.67s recent_step=2.95s eta=27:09:28 +2026-09-18 17:18:09 | INFO | train rank=0 | step=23370/50000 progress=46.7% loss=0.106796 lr=5.946650e-05 lr_backbone=5.946650e-05 elapsed=23:49:32 avg_step=3.67s recent_step=4.18s eta=27:08:57 +2026-09-18 17:18:42 | INFO | train rank=0 | step=23380/50000 progress=46.8% loss=0.098115 lr=5.943402e-05 lr_backbone=5.943402e-05 elapsed=23:50:05 avg_step=3.67s recent_step=3.28s eta=27:08:16 +2026-09-18 17:19:15 | INFO | train rank=0 | step=23390/50000 progress=46.8% loss=0.095262 lr=5.940155e-05 lr_backbone=5.940155e-05 elapsed=23:50:39 avg_step=3.67s recent_step=3.36s eta=27:07:36 +2026-09-18 17:19:49 | INFO | train rank=0 | step=23400/50000 progress=46.8% loss=0.086364 lr=5.936907e-05 lr_backbone=5.936907e-05 elapsed=23:51:13 avg_step=3.67s recent_step=3.40s eta=27:06:56 +2026-09-18 17:20:35 | INFO | train rank=0 | step=23410/50000 progress=46.8% loss=0.079310 lr=5.933658e-05 lr_backbone=5.933658e-05 elapsed=23:51:58 avg_step=3.67s recent_step=4.56s eta=27:06:30 +2026-09-18 17:21:10 | INFO | train rank=0 | step=23420/50000 progress=46.8% loss=0.103024 lr=5.930409e-05 lr_backbone=5.930409e-05 elapsed=23:52:33 avg_step=3.67s recent_step=3.47s eta=27:05:51 +2026-09-18 17:21:48 | INFO | train rank=0 | step=23430/50000 progress=46.9% loss=0.106921 lr=5.927160e-05 lr_backbone=5.927160e-05 elapsed=23:53:12 avg_step=3.67s recent_step=3.85s eta=27:05:16 +2026-09-18 17:22:23 | INFO | train rank=0 | step=23440/50000 progress=46.9% loss=0.092812 lr=5.923910e-05 lr_backbone=5.923910e-05 elapsed=23:53:46 avg_step=3.67s recent_step=3.42s eta=27:04:36 +2026-09-18 17:23:02 | INFO | train rank=0 | step=23450/50000 progress=46.9% loss=0.109941 lr=5.920660e-05 lr_backbone=5.920660e-05 elapsed=23:54:25 avg_step=3.67s recent_step=3.93s eta=27:04:03 +2026-09-18 17:23:29 | INFO | train rank=0 | step=23460/50000 progress=46.9% loss=0.110855 lr=5.917409e-05 lr_backbone=5.917409e-05 elapsed=23:54:52 avg_step=3.67s recent_step=2.73s eta=27:03:15 +2026-09-18 17:24:22 | INFO | train rank=0 | step=23470/50000 progress=46.9% loss=0.089650 lr=5.914158e-05 lr_backbone=5.914158e-05 elapsed=23:55:46 avg_step=3.67s recent_step=5.31s eta=27:02:57 +2026-09-18 17:24:55 | INFO | train rank=0 | step=23480/50000 progress=47.0% loss=0.076605 lr=5.910907e-05 lr_backbone=5.910907e-05 elapsed=23:56:18 avg_step=3.67s recent_step=3.25s eta=27:02:16 +2026-09-18 17:25:24 | INFO | train rank=0 | step=23490/50000 progress=47.0% loss=0.091843 lr=5.907655e-05 lr_backbone=5.907655e-05 elapsed=23:56:47 avg_step=3.67s recent_step=2.91s eta=27:01:30 +2026-09-18 17:25:57 | INFO | train rank=0 | step=23500/50000 progress=47.0% loss=0.071431 lr=5.904403e-05 lr_backbone=5.904403e-05 elapsed=23:57:20 avg_step=3.67s recent_step=3.31s eta=27:00:50 +2026-09-18 17:26:43 | INFO | train rank=0 | step=23510/50000 progress=47.0% loss=0.122669 lr=5.901150e-05 lr_backbone=5.901150e-05 elapsed=23:58:06 avg_step=3.67s recent_step=4.61s eta=27:00:24 +2026-09-18 17:27:21 | INFO | train rank=0 | step=23520/50000 progress=47.0% loss=0.092181 lr=5.897897e-05 lr_backbone=5.897897e-05 elapsed=23:58:44 avg_step=3.67s recent_step=3.80s eta=26:59:48 +2026-09-18 17:28:00 | INFO | train rank=0 | step=23530/50000 progress=47.1% loss=0.089985 lr=5.894644e-05 lr_backbone=5.894644e-05 elapsed=23:59:23 avg_step=3.67s recent_step=3.87s eta=26:59:14 +2026-09-18 17:28:26 | INFO | train rank=0 | step=23540/50000 progress=47.1% loss=0.096333 lr=5.891390e-05 lr_backbone=5.891390e-05 elapsed=23:59:49 avg_step=3.67s recent_step=2.63s eta=26:58:25 +2026-09-18 17:29:21 | INFO | train rank=0 | step=23550/50000 progress=47.1% loss=0.077295 lr=5.888136e-05 lr_backbone=5.888136e-05 elapsed=24:00:44 avg_step=3.67s recent_step=5.50s eta=26:58:09 +2026-09-18 17:29:55 | INFO | train rank=0 | step=23560/50000 progress=47.1% loss=0.116265 lr=5.884881e-05 lr_backbone=5.884881e-05 elapsed=24:01:19 avg_step=3.67s recent_step=3.43s eta=26:57:30 +2026-09-18 17:30:25 | INFO | train rank=0 | step=23570/50000 progress=47.1% loss=0.078727 lr=5.881626e-05 lr_backbone=5.881626e-05 elapsed=24:01:48 avg_step=3.67s recent_step=2.94s eta=26:56:45 +2026-09-18 17:30:59 | INFO | train rank=0 | step=23580/50000 progress=47.2% loss=0.105154 lr=5.878371e-05 lr_backbone=5.878371e-05 elapsed=24:02:23 avg_step=3.67s recent_step=3.48s eta=26:56:06 +2026-09-18 17:31:43 | INFO | train rank=0 | step=23590/50000 progress=47.2% loss=0.076124 lr=5.875115e-05 lr_backbone=5.875115e-05 elapsed=24:03:07 avg_step=3.67s recent_step=4.38s eta=26:55:37 +2026-09-18 17:32:21 | INFO | train rank=0 | step=23600/50000 progress=47.2% loss=0.079375 lr=5.871859e-05 lr_backbone=5.871859e-05 elapsed=24:03:44 avg_step=3.67s recent_step=3.77s eta=26:55:02 +2026-09-18 17:32:50 | INFO | train rank=0 | step=23610/50000 progress=47.2% loss=0.104043 lr=5.868603e-05 lr_backbone=5.868603e-05 elapsed=24:04:13 avg_step=3.67s recent_step=2.91s eta=26:54:17 +2026-09-18 17:33:24 | INFO | train rank=0 | step=23620/50000 progress=47.2% loss=0.104591 lr=5.865346e-05 lr_backbone=5.865346e-05 elapsed=24:04:47 avg_step=3.67s recent_step=3.41s eta=26:53:37 +2026-09-18 17:34:00 | INFO | train rank=0 | step=23630/50000 progress=47.3% loss=0.086952 lr=5.862089e-05 lr_backbone=5.862089e-05 elapsed=24:05:23 avg_step=3.67s recent_step=3.61s eta=26:52:59 +2026-09-18 17:34:32 | INFO | train rank=0 | step=23640/50000 progress=47.3% loss=0.084784 lr=5.858831e-05 lr_backbone=5.858831e-05 elapsed=24:05:56 avg_step=3.67s recent_step=3.23s eta=26:52:18 +2026-09-18 17:35:13 | INFO | train rank=0 | step=23650/50000 progress=47.3% loss=0.093217 lr=5.855573e-05 lr_backbone=5.855573e-05 elapsed=24:06:36 avg_step=3.67s recent_step=4.01s eta=26:51:45 +2026-09-18 17:35:42 | INFO | train rank=0 | step=23660/50000 progress=47.3% loss=0.111720 lr=5.852315e-05 lr_backbone=5.852315e-05 elapsed=24:07:05 avg_step=3.67s recent_step=2.93s eta=26:51:00 +2026-09-18 17:36:18 | INFO | train rank=0 | step=23670/50000 progress=47.3% loss=0.082404 lr=5.849056e-05 lr_backbone=5.849056e-05 elapsed=24:07:41 avg_step=3.67s recent_step=3.58s eta=26:50:22 +2026-09-18 17:37:21 | INFO | train rank=0 | step=23680/50000 progress=47.4% loss=0.093771 lr=5.845797e-05 lr_backbone=5.845797e-05 elapsed=24:08:44 avg_step=3.67s recent_step=6.34s eta=26:50:15 +2026-09-18 17:37:53 | INFO | train rank=0 | step=23690/50000 progress=47.4% loss=0.103700 lr=5.842537e-05 lr_backbone=5.842537e-05 elapsed=24:09:16 avg_step=3.67s recent_step=3.20s eta=26:49:33 +2026-09-18 17:38:26 | INFO | train rank=0 | step=23700/50000 progress=47.4% loss=0.090010 lr=5.839278e-05 lr_backbone=5.839278e-05 elapsed=24:09:50 avg_step=3.67s recent_step=3.35s eta=26:48:53 +2026-09-18 17:38:56 | INFO | train rank=0 | step=23710/50000 progress=47.4% loss=0.078321 lr=5.836017e-05 lr_backbone=5.836017e-05 elapsed=24:10:20 avg_step=3.67s recent_step=2.97s eta=26:48:09 +2026-09-18 17:39:47 | INFO | train rank=0 | step=23720/50000 progress=47.4% loss=0.095563 lr=5.832757e-05 lr_backbone=5.832757e-05 elapsed=24:11:10 avg_step=3.67s recent_step=5.07s eta=26:47:47 +2026-09-18 17:40:17 | INFO | train rank=0 | step=23730/50000 progress=47.5% loss=0.105246 lr=5.829496e-05 lr_backbone=5.829496e-05 elapsed=24:11:41 avg_step=3.67s recent_step=3.03s eta=26:47:04 +2026-09-18 17:40:53 | INFO | train rank=0 | step=23740/50000 progress=47.5% loss=0.111398 lr=5.826235e-05 lr_backbone=5.826235e-05 elapsed=24:12:16 avg_step=3.67s recent_step=3.55s eta=26:46:26 +2026-09-18 17:41:28 | INFO | train rank=0 | step=23750/50000 progress=47.5% loss=0.077255 lr=5.822973e-05 lr_backbone=5.822973e-05 elapsed=24:12:51 avg_step=3.67s recent_step=3.49s eta=26:45:47 +2026-09-18 17:42:12 | INFO | train rank=0 | step=23760/50000 progress=47.5% loss=0.085314 lr=5.819711e-05 lr_backbone=5.819711e-05 elapsed=24:13:35 avg_step=3.67s recent_step=4.41s eta=26:45:18 +2026-09-18 17:42:40 | INFO | train rank=0 | step=23770/50000 progress=47.5% loss=0.081074 lr=5.816449e-05 lr_backbone=5.816449e-05 elapsed=24:14:03 avg_step=3.67s recent_step=2.83s eta=26:44:32 +2026-09-18 17:43:14 | INFO | train rank=0 | step=23780/50000 progress=47.6% loss=0.112105 lr=5.813186e-05 lr_backbone=5.813186e-05 elapsed=24:14:37 avg_step=3.67s recent_step=3.40s eta=26:43:53 +2026-09-18 17:43:50 | INFO | train rank=0 | step=23790/50000 progress=47.6% loss=0.077519 lr=5.809923e-05 lr_backbone=5.809923e-05 elapsed=24:15:13 avg_step=3.67s recent_step=3.58s eta=26:43:15 +2026-09-18 17:44:41 | INFO | train rank=0 | step=23800/50000 progress=47.6% loss=0.091455 lr=5.806659e-05 lr_backbone=5.806659e-05 elapsed=24:16:05 avg_step=3.67s recent_step=5.16s eta=26:42:55 +2026-09-18 17:45:13 | INFO | train rank=0 | step=23810/50000 progress=47.6% loss=0.076146 lr=5.803395e-05 lr_backbone=5.803395e-05 elapsed=24:16:36 avg_step=3.67s recent_step=3.13s eta=26:42:12 +2026-09-18 17:45:44 | INFO | train rank=0 | step=23820/50000 progress=47.6% loss=0.075926 lr=5.800131e-05 lr_backbone=5.800131e-05 elapsed=24:17:07 avg_step=3.67s recent_step=3.11s eta=26:41:29 +2026-09-18 17:46:14 | INFO | train rank=0 | step=23830/50000 progress=47.7% loss=0.098680 lr=5.796867e-05 lr_backbone=5.796867e-05 elapsed=24:17:37 avg_step=3.67s recent_step=3.04s eta=26:40:45 +2026-09-18 17:47:04 | INFO | train rank=0 | step=23840/50000 progress=47.7% loss=0.096261 lr=5.793602e-05 lr_backbone=5.793602e-05 elapsed=24:18:28 avg_step=3.67s recent_step=5.02s eta=26:40:24 +2026-09-18 17:47:34 | INFO | train rank=0 | step=23850/50000 progress=47.7% loss=0.116102 lr=5.790337e-05 lr_backbone=5.790337e-05 elapsed=24:18:57 avg_step=3.67s recent_step=2.98s eta=26:39:39 +2026-09-18 17:48:04 | INFO | train rank=0 | step=23860/50000 progress=47.7% loss=0.106128 lr=5.787071e-05 lr_backbone=5.787071e-05 elapsed=24:19:27 avg_step=3.67s recent_step=3.00s eta=26:38:55 +2026-09-18 17:48:38 | INFO | train rank=0 | step=23870/50000 progress=47.7% loss=0.100296 lr=5.783805e-05 lr_backbone=5.783805e-05 elapsed=24:20:02 avg_step=3.67s recent_step=3.41s eta=26:38:16 +2026-09-18 17:49:31 | INFO | train rank=0 | step=23880/50000 progress=47.8% loss=0.128696 lr=5.780539e-05 lr_backbone=5.780539e-05 elapsed=24:20:54 avg_step=3.67s recent_step=5.27s eta=26:37:56 +2026-09-18 17:50:04 | INFO | train rank=0 | step=23890/50000 progress=47.8% loss=0.107954 lr=5.777273e-05 lr_backbone=5.777273e-05 elapsed=24:21:27 avg_step=3.67s recent_step=3.26s eta=26:37:15 +2026-09-18 17:50:39 | INFO | train rank=0 | step=23900/50000 progress=47.8% loss=0.084564 lr=5.774006e-05 lr_backbone=5.774006e-05 elapsed=24:22:02 avg_step=3.67s recent_step=3.53s eta=26:36:37 +2026-09-18 17:51:08 | INFO | train rank=0 | step=23910/50000 progress=47.8% loss=0.094366 lr=5.770738e-05 lr_backbone=5.770738e-05 elapsed=24:22:32 avg_step=3.67s recent_step=2.95s eta=26:35:52 +2026-09-18 17:51:48 | INFO | train rank=0 | step=23920/50000 progress=47.8% loss=0.107503 lr=5.767471e-05 lr_backbone=5.767471e-05 elapsed=24:23:11 avg_step=3.67s recent_step=3.95s eta=26:35:19 +2026-09-18 17:52:18 | INFO | train rank=0 | step=23930/50000 progress=47.9% loss=0.122173 lr=5.764203e-05 lr_backbone=5.764203e-05 elapsed=24:23:41 avg_step=3.67s recent_step=2.97s eta=26:34:34 +2026-09-18 17:52:57 | INFO | train rank=0 | step=23940/50000 progress=47.9% loss=0.142622 lr=5.760935e-05 lr_backbone=5.760935e-05 elapsed=24:24:20 avg_step=3.67s recent_step=3.93s eta=26:34:01 +2026-09-18 17:53:41 | INFO | train rank=0 | step=23950/50000 progress=47.9% loss=0.073678 lr=5.757666e-05 lr_backbone=5.757666e-05 elapsed=24:25:04 avg_step=3.67s recent_step=4.39s eta=26:33:32 +2026-09-18 17:54:14 | INFO | train rank=0 | step=23960/50000 progress=47.9% loss=0.107249 lr=5.754397e-05 lr_backbone=5.754397e-05 elapsed=24:25:37 avg_step=3.67s recent_step=3.33s eta=26:32:51 +2026-09-18 17:54:50 | INFO | train rank=0 | step=23970/50000 progress=47.9% loss=0.074374 lr=5.751128e-05 lr_backbone=5.751128e-05 elapsed=24:26:13 avg_step=3.67s recent_step=3.57s eta=26:32:13 +2026-09-18 17:55:21 | INFO | train rank=0 | step=23980/50000 progress=48.0% loss=0.094794 lr=5.747858e-05 lr_backbone=5.747858e-05 elapsed=24:26:44 avg_step=3.67s recent_step=3.13s eta=26:31:31 +2026-09-18 17:56:13 | INFO | train rank=0 | step=23990/50000 progress=48.0% loss=0.087055 lr=5.744588e-05 lr_backbone=5.744588e-05 elapsed=24:27:36 avg_step=3.67s recent_step=5.23s eta=26:31:11 +2026-09-18 17:56:51 | INFO | train rank=0 | step=24000/50000 progress=48.0% loss=0.078446 lr=5.741318e-05 lr_backbone=5.741318e-05 elapsed=24:28:14 avg_step=3.67s recent_step=3.74s eta=26:30:35 +2026-09-18 17:57:27 | INFO | train rank=0 | step=24010/50000 progress=48.0% loss=0.083476 lr=5.738048e-05 lr_backbone=5.738048e-05 elapsed=24:28:51 avg_step=3.67s recent_step=3.67s eta=26:29:58 +2026-09-18 17:57:58 | INFO | train rank=0 | step=24020/50000 progress=48.0% loss=0.116857 lr=5.734777e-05 lr_backbone=5.734777e-05 elapsed=24:29:22 avg_step=3.67s recent_step=3.11s eta=26:29:15 +2026-09-18 17:58:50 | INFO | train rank=0 | step=24030/50000 progress=48.1% loss=0.079634 lr=5.731506e-05 lr_backbone=5.731506e-05 elapsed=24:30:13 avg_step=3.67s recent_step=5.19s eta=26:28:55 +2026-09-18 17:59:19 | INFO | train rank=0 | step=24040/50000 progress=48.1% loss=0.098744 lr=5.728234e-05 lr_backbone=5.728234e-05 elapsed=24:30:42 avg_step=3.67s recent_step=2.88s eta=26:28:10 +2026-09-18 17:59:52 | INFO | train rank=0 | step=24050/50000 progress=48.1% loss=0.077514 lr=5.724962e-05 lr_backbone=5.724962e-05 elapsed=24:31:16 avg_step=3.67s recent_step=3.34s eta=26:27:30 +2026-09-18 18:00:21 | INFO | train rank=0 | step=24060/50000 progress=48.1% loss=0.107540 lr=5.721690e-05 lr_backbone=5.721690e-05 elapsed=24:31:44 avg_step=3.67s recent_step=2.87s eta=26:26:44 +2026-09-18 18:01:12 | INFO | train rank=0 | step=24070/50000 progress=48.1% loss=0.095801 lr=5.718418e-05 lr_backbone=5.718418e-05 elapsed=24:32:36 avg_step=3.67s recent_step=5.11s eta=26:26:23 +2026-09-18 18:01:50 | INFO | train rank=0 | step=24080/50000 progress=48.2% loss=0.091295 lr=5.715145e-05 lr_backbone=5.715145e-05 elapsed=24:33:14 avg_step=3.67s recent_step=3.83s eta=26:25:48 +2026-09-18 18:02:23 | INFO | train rank=0 | step=24090/50000 progress=48.2% loss=0.098130 lr=5.711872e-05 lr_backbone=5.711872e-05 elapsed=24:33:46 avg_step=3.67s recent_step=3.23s eta=26:25:07 +2026-09-18 18:03:01 | INFO | train rank=0 | step=24100/50000 progress=48.2% loss=0.076809 lr=5.708598e-05 lr_backbone=5.708598e-05 elapsed=24:34:24 avg_step=3.67s recent_step=3.83s eta=26:24:32 +2026-09-18 18:03:33 | INFO | train rank=0 | step=24110/50000 progress=48.2% loss=0.076216 lr=5.705325e-05 lr_backbone=5.705325e-05 elapsed=24:34:56 avg_step=3.67s recent_step=3.20s eta=26:23:50 +2026-09-18 18:04:22 | INFO | train rank=0 | step=24120/50000 progress=48.2% loss=0.115222 lr=5.702051e-05 lr_backbone=5.702051e-05 elapsed=24:35:45 avg_step=3.67s recent_step=4.86s eta=26:23:26 +2026-09-18 18:04:54 | INFO | train rank=0 | step=24130/50000 progress=48.3% loss=0.111332 lr=5.698776e-05 lr_backbone=5.698776e-05 elapsed=24:36:17 avg_step=3.67s recent_step=3.23s eta=26:22:44 +2026-09-18 18:05:26 | INFO | train rank=0 | step=24140/50000 progress=48.3% loss=0.086095 lr=5.695502e-05 lr_backbone=5.695502e-05 elapsed=24:36:50 avg_step=3.67s recent_step=3.25s eta=26:22:03 +2026-09-18 18:06:04 | INFO | train rank=0 | step=24150/50000 progress=48.3% loss=0.094251 lr=5.692227e-05 lr_backbone=5.692227e-05 elapsed=24:37:28 avg_step=3.67s recent_step=3.80s eta=26:21:28 +2026-09-18 18:06:47 | INFO | train rank=0 | step=24160/50000 progress=48.3% loss=0.070523 lr=5.688951e-05 lr_backbone=5.688951e-05 elapsed=24:38:10 avg_step=3.67s recent_step=4.24s eta=26:20:57 +2026-09-18 18:07:24 | INFO | train rank=0 | step=24170/50000 progress=48.3% loss=0.115867 lr=5.685676e-05 lr_backbone=5.685676e-05 elapsed=24:38:47 avg_step=3.67s recent_step=3.69s eta=26:20:21 +2026-09-18 18:07:50 | INFO | train rank=0 | step=24180/50000 progress=48.4% loss=0.098523 lr=5.682400e-05 lr_backbone=5.682400e-05 elapsed=24:39:13 avg_step=3.67s recent_step=2.59s eta=26:19:33 +2026-09-18 18:08:30 | INFO | train rank=0 | step=24190/50000 progress=48.4% loss=0.087956 lr=5.679124e-05 lr_backbone=5.679124e-05 elapsed=24:39:53 avg_step=3.67s recent_step=4.04s eta=26:19:00 +2026-09-18 18:09:10 | INFO | train rank=0 | step=24200/50000 progress=48.4% loss=0.090529 lr=5.675847e-05 lr_backbone=5.675847e-05 elapsed=24:40:33 avg_step=3.67s recent_step=3.97s eta=26:18:26 +2026-09-18 18:09:42 | INFO | train rank=0 | step=24210/50000 progress=48.4% loss=0.114543 lr=5.672571e-05 lr_backbone=5.672571e-05 elapsed=24:41:05 avg_step=3.67s recent_step=3.21s eta=26:17:45 +2026-09-18 18:10:12 | INFO | train rank=0 | step=24220/50000 progress=48.4% loss=0.090470 lr=5.669294e-05 lr_backbone=5.669294e-05 elapsed=24:41:35 avg_step=3.67s recent_step=2.99s eta=26:17:01 +2026-09-18 18:10:53 | INFO | train rank=0 | step=24230/50000 progress=48.5% loss=0.091288 lr=5.666016e-05 lr_backbone=5.666016e-05 elapsed=24:42:17 avg_step=3.67s recent_step=4.16s eta=26:16:29 +2026-09-18 18:11:35 | INFO | train rank=0 | step=24240/50000 progress=48.5% loss=0.070587 lr=5.662739e-05 lr_backbone=5.662739e-05 elapsed=24:42:59 avg_step=3.67s recent_step=4.20s eta=26:15:58 +2026-09-18 18:12:10 | INFO | train rank=0 | step=24250/50000 progress=48.5% loss=0.097792 lr=5.659461e-05 lr_backbone=5.659461e-05 elapsed=24:43:34 avg_step=3.67s recent_step=3.49s eta=26:15:20 +2026-09-18 18:12:46 | INFO | train rank=0 | step=24260/50000 progress=48.5% loss=0.110599 lr=5.656183e-05 lr_backbone=5.656183e-05 elapsed=24:44:09 avg_step=3.67s recent_step=3.55s eta=26:14:42 +2026-09-18 18:13:22 | INFO | train rank=0 | step=24270/50000 progress=48.5% loss=0.082713 lr=5.652904e-05 lr_backbone=5.652904e-05 elapsed=24:44:45 avg_step=3.67s recent_step=3.59s eta=26:14:04 +2026-09-18 18:14:16 | INFO | train rank=0 | step=24280/50000 progress=48.6% loss=0.075283 lr=5.649625e-05 lr_backbone=5.649625e-05 elapsed=24:45:39 avg_step=3.67s recent_step=5.41s eta=26:13:46 +2026-09-18 18:14:47 | INFO | train rank=0 | step=24290/50000 progress=48.6% loss=0.074617 lr=5.646346e-05 lr_backbone=5.646346e-05 elapsed=24:46:11 avg_step=3.67s recent_step=3.15s eta=26:13:04 +2026-09-18 18:15:22 | INFO | train rank=0 | step=24300/50000 progress=48.6% loss=0.106293 lr=5.643067e-05 lr_backbone=5.643067e-05 elapsed=24:46:45 avg_step=3.67s recent_step=3.46s eta=26:12:25 +2026-09-18 18:15:52 | INFO | train rank=0 | step=24310/50000 progress=48.6% loss=0.123346 lr=5.639787e-05 lr_backbone=5.639787e-05 elapsed=24:47:15 avg_step=3.67s recent_step=3.00s eta=26:11:41 +2026-09-18 18:16:45 | INFO | train rank=0 | step=24320/50000 progress=48.6% loss=0.107059 lr=5.636508e-05 lr_backbone=5.636508e-05 elapsed=24:48:08 avg_step=3.67s recent_step=5.31s eta=26:11:21 +2026-09-18 18:17:18 | INFO | train rank=0 | step=24330/50000 progress=48.7% loss=0.090332 lr=5.633227e-05 lr_backbone=5.633227e-05 elapsed=24:48:41 avg_step=3.67s recent_step=3.28s eta=26:10:41 +2026-09-18 18:17:53 | INFO | train rank=0 | step=24340/50000 progress=48.7% loss=0.086080 lr=5.629947e-05 lr_backbone=5.629947e-05 elapsed=24:49:17 avg_step=3.67s recent_step=3.56s eta=26:10:03 +2026-09-18 18:18:25 | INFO | train rank=0 | step=24350/50000 progress=48.7% loss=0.081448 lr=5.626666e-05 lr_backbone=5.626666e-05 elapsed=24:49:48 avg_step=3.67s recent_step=3.13s eta=26:09:20 +2026-09-18 18:19:09 | INFO | train rank=0 | step=24360/50000 progress=48.7% loss=0.070151 lr=5.623385e-05 lr_backbone=5.623385e-05 elapsed=24:50:33 avg_step=3.67s recent_step=4.45s eta=26:08:52 +2026-09-18 18:19:40 | INFO | train rank=0 | step=24370/50000 progress=48.7% loss=0.092025 lr=5.620104e-05 lr_backbone=5.620104e-05 elapsed=24:51:03 avg_step=3.67s recent_step=3.07s eta=26:08:09 +2026-09-18 18:20:11 | INFO | train rank=0 | step=24380/50000 progress=48.8% loss=0.101243 lr=5.616822e-05 lr_backbone=5.616822e-05 elapsed=24:51:35 avg_step=3.67s recent_step=3.15s eta=26:07:27 +2026-09-18 18:20:45 | INFO | train rank=0 | step=24390/50000 progress=48.8% loss=0.091629 lr=5.613541e-05 lr_backbone=5.613541e-05 elapsed=24:52:09 avg_step=3.67s recent_step=3.38s eta=26:06:47 +2026-09-18 18:21:42 | INFO | train rank=0 | step=24400/50000 progress=48.8% loss=0.121093 lr=5.610258e-05 lr_backbone=5.610258e-05 elapsed=24:53:05 avg_step=3.67s recent_step=5.65s eta=26:06:31 +2026-09-18 18:22:09 | INFO | train rank=0 | step=24410/50000 progress=48.8% loss=0.099393 lr=5.606976e-05 lr_backbone=5.606976e-05 elapsed=24:53:32 avg_step=3.67s recent_step=2.74s eta=26:05:44 +2026-09-18 18:22:41 | INFO | train rank=0 | step=24420/50000 progress=48.8% loss=0.079138 lr=5.603693e-05 lr_backbone=5.603693e-05 elapsed=24:54:04 avg_step=3.67s recent_step=3.20s eta=26:05:03 +2026-09-18 18:23:13 | INFO | train rank=0 | step=24430/50000 progress=48.9% loss=0.104326 lr=5.600411e-05 lr_backbone=5.600411e-05 elapsed=24:54:37 avg_step=3.67s recent_step=3.22s eta=26:04:21 +2026-09-18 18:23:59 | INFO | train rank=0 | step=24440/50000 progress=48.9% loss=0.099011 lr=5.597127e-05 lr_backbone=5.597127e-05 elapsed=24:55:23 avg_step=3.67s recent_step=4.61s eta=26:03:54 +2026-09-18 18:24:35 | INFO | train rank=0 | step=24450/50000 progress=48.9% loss=0.091447 lr=5.593844e-05 lr_backbone=5.593844e-05 elapsed=24:55:58 avg_step=3.67s recent_step=3.51s eta=26:03:16 +2026-09-18 18:25:10 | INFO | train rank=0 | step=24460/50000 progress=48.9% loss=0.166283 lr=5.590560e-05 lr_backbone=5.590560e-05 elapsed=24:56:34 avg_step=3.67s recent_step=3.59s eta=26:02:38 +2026-09-18 18:25:43 | INFO | train rank=0 | step=24470/50000 progress=48.9% loss=0.118406 lr=5.587276e-05 lr_backbone=5.587276e-05 elapsed=24:57:06 avg_step=3.67s recent_step=3.27s eta=26:01:57 +2026-09-18 18:26:33 | INFO | train rank=0 | step=24480/50000 progress=49.0% loss=0.095487 lr=5.583992e-05 lr_backbone=5.583992e-05 elapsed=24:57:56 avg_step=3.67s recent_step=4.99s eta=26:01:35 +2026-09-18 18:27:08 | INFO | train rank=0 | step=24490/50000 progress=49.0% loss=0.087817 lr=5.580708e-05 lr_backbone=5.580708e-05 elapsed=24:58:31 avg_step=3.67s recent_step=3.47s eta=26:00:56 +2026-09-18 18:27:36 | INFO | train rank=0 | step=24500/50000 progress=49.0% loss=0.110193 lr=5.577423e-05 lr_backbone=5.577423e-05 elapsed=24:59:00 avg_step=3.67s recent_step=2.87s eta=26:00:11 +2026-09-18 18:28:15 | INFO | train rank=0 | step=24510/50000 progress=49.0% loss=0.084235 lr=5.574138e-05 lr_backbone=5.574138e-05 elapsed=24:59:39 avg_step=3.67s recent_step=3.91s eta=25:59:36 +2026-09-18 18:28:57 | INFO | train rank=0 | step=24520/50000 progress=49.0% loss=0.093855 lr=5.570853e-05 lr_backbone=5.570853e-05 elapsed=25:00:20 avg_step=3.67s recent_step=4.13s eta=25:59:05 +2026-09-18 18:29:32 | INFO | train rank=0 | step=24530/50000 progress=49.1% loss=0.101089 lr=5.567568e-05 lr_backbone=5.567568e-05 elapsed=25:00:56 avg_step=3.67s recent_step=3.57s eta=25:58:27 +2026-09-18 18:30:07 | INFO | train rank=0 | step=24540/50000 progress=49.1% loss=0.100363 lr=5.564282e-05 lr_backbone=5.564282e-05 elapsed=25:01:30 avg_step=3.67s recent_step=3.43s eta=25:57:48 +2026-09-18 18:30:47 | INFO | train rank=0 | step=24550/50000 progress=49.1% loss=0.087044 lr=5.560996e-05 lr_backbone=5.560996e-05 elapsed=25:02:10 avg_step=3.67s recent_step=4.03s eta=25:57:15 +2026-09-18 18:31:23 | INFO | train rank=0 | step=24560/50000 progress=49.1% loss=0.087313 lr=5.557710e-05 lr_backbone=5.557710e-05 elapsed=25:02:47 avg_step=3.67s recent_step=3.63s eta=25:56:37 +2026-09-18 18:31:56 | INFO | train rank=0 | step=24570/50000 progress=49.1% loss=0.098903 lr=5.554423e-05 lr_backbone=5.554423e-05 elapsed=25:03:19 avg_step=3.67s recent_step=3.24s eta=25:55:56 +2026-09-18 18:32:29 | INFO | train rank=0 | step=24580/50000 progress=49.2% loss=0.099000 lr=5.551137e-05 lr_backbone=5.551137e-05 elapsed=25:03:53 avg_step=3.67s recent_step=3.37s eta=25:55:16 +2026-09-18 18:33:15 | INFO | train rank=0 | step=24590/50000 progress=49.2% loss=0.083746 lr=5.547850e-05 lr_backbone=5.547850e-05 elapsed=25:04:38 avg_step=3.67s recent_step=4.52s eta=25:54:49 +2026-09-18 18:33:47 | INFO | train rank=0 | step=24600/50000 progress=49.2% loss=0.067709 lr=5.544563e-05 lr_backbone=5.544563e-05 elapsed=25:05:10 avg_step=3.67s recent_step=3.20s eta=25:54:07 +2026-09-18 18:34:17 | INFO | train rank=0 | step=24610/50000 progress=49.2% loss=0.116735 lr=5.541275e-05 lr_backbone=5.541275e-05 elapsed=25:05:40 avg_step=3.67s recent_step=3.01s eta=25:53:24 +2026-09-18 18:34:40 | INFO | train rank=0 | step=24620/50000 progress=49.2% loss=0.103583 lr=5.537988e-05 lr_backbone=5.537988e-05 elapsed=25:06:03 avg_step=3.67s recent_step=2.32s eta=25:52:33 +2026-09-18 18:35:31 | INFO | train rank=0 | step=24630/50000 progress=49.3% loss=0.133676 lr=5.534700e-05 lr_backbone=5.534700e-05 elapsed=25:06:54 avg_step=3.67s recent_step=5.09s eta=25:52:11 +2026-09-18 18:36:21 | INFO | train rank=0 | step=24640/50000 progress=49.3% loss=0.095246 lr=5.531412e-05 lr_backbone=5.531412e-05 elapsed=25:07:44 avg_step=3.67s recent_step=4.99s eta=25:51:48 +2026-09-18 18:36:57 | INFO | train rank=0 | step=24650/50000 progress=49.3% loss=0.092272 lr=5.528123e-05 lr_backbone=5.528123e-05 elapsed=25:08:20 avg_step=3.67s recent_step=3.60s eta=25:51:10 +2026-09-18 18:37:26 | INFO | train rank=0 | step=24660/50000 progress=49.3% loss=0.118440 lr=5.524835e-05 lr_backbone=5.524835e-05 elapsed=25:08:49 avg_step=3.67s recent_step=2.89s eta=25:50:25 +2026-09-18 18:38:02 | INFO | train rank=0 | step=24670/50000 progress=49.3% loss=0.084712 lr=5.521546e-05 lr_backbone=5.521546e-05 elapsed=25:09:25 avg_step=3.67s recent_step=3.62s eta=25:49:48 +2026-09-18 18:38:51 | INFO | train rank=0 | step=24680/50000 progress=49.4% loss=0.099944 lr=5.518257e-05 lr_backbone=5.518257e-05 elapsed=25:10:14 avg_step=3.67s recent_step=4.89s eta=25:49:24 +2026-09-18 18:39:21 | INFO | train rank=0 | step=24690/50000 progress=49.4% loss=0.102387 lr=5.514968e-05 lr_backbone=5.514968e-05 elapsed=25:10:44 avg_step=3.67s recent_step=3.01s eta=25:48:40 +2026-09-18 18:39:58 | INFO | train rank=0 | step=24700/50000 progress=49.4% loss=0.134232 lr=5.511678e-05 lr_backbone=5.511678e-05 elapsed=25:11:22 avg_step=3.67s recent_step=3.73s eta=25:48:04 +2026-09-18 18:40:34 | INFO | train rank=0 | step=24710/50000 progress=49.4% loss=0.089004 lr=5.508389e-05 lr_backbone=5.508389e-05 elapsed=25:11:57 avg_step=3.67s recent_step=3.54s eta=25:47:26 +2026-09-18 18:41:12 | INFO | train rank=0 | step=24720/50000 progress=49.4% loss=0.079326 lr=5.505099e-05 lr_backbone=5.505099e-05 elapsed=25:12:35 avg_step=3.67s recent_step=3.82s eta=25:46:51 +2026-09-18 18:41:51 | INFO | train rank=0 | step=24730/50000 progress=49.5% loss=0.076276 lr=5.501809e-05 lr_backbone=5.501809e-05 elapsed=25:13:15 avg_step=3.67s recent_step=3.95s eta=25:46:17 +2026-09-18 18:42:25 | INFO | train rank=0 | step=24740/50000 progress=49.5% loss=0.081944 lr=5.498518e-05 lr_backbone=5.498518e-05 elapsed=25:13:49 avg_step=3.67s recent_step=3.39s eta=25:45:38 +2026-09-18 18:42:54 | INFO | train rank=0 | step=24750/50000 progress=49.5% loss=0.087402 lr=5.495228e-05 lr_backbone=5.495228e-05 elapsed=25:14:18 avg_step=3.67s recent_step=2.91s eta=25:44:53 +2026-09-18 18:43:33 | INFO | train rank=0 | step=24760/50000 progress=49.5% loss=0.081569 lr=5.491937e-05 lr_backbone=5.491937e-05 elapsed=25:14:56 avg_step=3.67s recent_step=3.86s eta=25:44:18 +2026-09-18 18:44:03 | INFO | train rank=0 | step=24770/50000 progress=49.5% loss=0.081759 lr=5.488646e-05 lr_backbone=5.488646e-05 elapsed=25:15:26 avg_step=3.67s recent_step=3.00s eta=25:43:35 +2026-09-18 18:44:36 | INFO | train rank=0 | step=24780/50000 progress=49.6% loss=0.104531 lr=5.485355e-05 lr_backbone=5.485355e-05 elapsed=25:15:59 avg_step=3.67s recent_step=3.31s eta=25:42:55 +2026-09-18 18:45:03 | INFO | train rank=0 | step=24790/50000 progress=49.6% loss=0.091296 lr=5.482063e-05 lr_backbone=5.482063e-05 elapsed=25:16:27 avg_step=3.67s recent_step=2.72s eta=25:42:08 +2026-09-18 18:45:58 | INFO | train rank=0 | step=24800/50000 progress=49.6% loss=0.107917 lr=5.478772e-05 lr_backbone=5.478772e-05 elapsed=25:17:21 avg_step=3.67s recent_step=5.47s eta=25:41:50 +2026-09-18 18:46:29 | INFO | train rank=0 | step=24810/50000 progress=49.6% loss=0.096589 lr=5.475480e-05 lr_backbone=5.475480e-05 elapsed=25:17:52 avg_step=3.67s recent_step=3.08s eta=25:41:07 +2026-09-18 18:46:58 | INFO | train rank=0 | step=24820/50000 progress=49.6% loss=0.092640 lr=5.472188e-05 lr_backbone=5.472188e-05 elapsed=25:18:22 avg_step=3.67s recent_step=2.96s eta=25:40:23 +2026-09-18 18:47:21 | INFO | train rank=0 | step=24830/50000 progress=49.7% loss=0.093488 lr=5.468895e-05 lr_backbone=5.468895e-05 elapsed=25:18:45 avg_step=3.67s recent_step=2.29s eta=25:39:32 +2026-09-18 18:48:04 | INFO | train rank=0 | step=24840/50000 progress=49.7% loss=0.075614 lr=5.465603e-05 lr_backbone=5.465603e-05 elapsed=25:19:27 avg_step=3.67s recent_step=4.28s eta=25:39:02 +2026-09-18 18:48:42 | INFO | train rank=0 | step=24850/50000 progress=49.7% loss=0.092072 lr=5.462310e-05 lr_backbone=5.462310e-05 elapsed=25:20:05 avg_step=3.67s recent_step=3.76s eta=25:38:26 +2026-09-18 18:49:12 | INFO | train rank=0 | step=24860/50000 progress=49.7% loss=0.092585 lr=5.459017e-05 lr_backbone=5.459017e-05 elapsed=25:20:35 avg_step=3.67s recent_step=2.98s eta=25:37:42 +2026-09-18 18:49:47 | INFO | train rank=0 | step=24870/50000 progress=49.7% loss=0.078229 lr=5.455724e-05 lr_backbone=5.455724e-05 elapsed=25:21:10 avg_step=3.67s recent_step=3.53s eta=25:37:04 +2026-09-18 18:50:40 | INFO | train rank=0 | step=24880/50000 progress=49.8% loss=0.078485 lr=5.452431e-05 lr_backbone=5.452431e-05 elapsed=25:22:04 avg_step=3.67s recent_step=5.33s eta=25:36:44 +2026-09-18 18:51:11 | INFO | train rank=0 | step=24890/50000 progress=49.8% loss=0.079261 lr=5.449138e-05 lr_backbone=5.449138e-05 elapsed=25:22:34 avg_step=3.67s recent_step=3.08s eta=25:36:02 +2026-09-18 18:51:43 | INFO | train rank=0 | step=24900/50000 progress=49.8% loss=0.081104 lr=5.445844e-05 lr_backbone=5.445844e-05 elapsed=25:23:06 avg_step=3.67s recent_step=3.17s eta=25:35:20 +2026-09-18 18:52:08 | INFO | train rank=0 | step=24910/50000 progress=49.8% loss=0.093853 lr=5.442550e-05 lr_backbone=5.442550e-05 elapsed=25:23:32 avg_step=3.67s recent_step=2.55s eta=25:34:32 +2026-09-18 18:52:54 | INFO | train rank=0 | step=24920/50000 progress=49.8% loss=0.090096 lr=5.439256e-05 lr_backbone=5.439256e-05 elapsed=25:24:18 avg_step=3.67s recent_step=4.62s eta=25:34:05 +2026-09-18 18:53:28 | INFO | train rank=0 | step=24930/50000 progress=49.9% loss=0.121338 lr=5.435962e-05 lr_backbone=5.435962e-05 elapsed=25:24:52 avg_step=3.67s recent_step=3.37s eta=25:33:25 +2026-09-18 18:54:04 | INFO | train rank=0 | step=24940/50000 progress=49.9% loss=0.094024 lr=5.432667e-05 lr_backbone=5.432667e-05 elapsed=25:25:27 avg_step=3.67s recent_step=3.56s eta=25:32:47 +2026-09-18 18:54:37 | INFO | train rank=0 | step=24950/50000 progress=49.9% loss=0.096681 lr=5.429373e-05 lr_backbone=5.429373e-05 elapsed=25:26:01 avg_step=3.67s recent_step=3.36s eta=25:32:08 +2026-09-18 18:55:27 | INFO | train rank=0 | step=24960/50000 progress=49.9% loss=0.100077 lr=5.426078e-05 lr_backbone=5.426078e-05 elapsed=25:26:50 avg_step=3.67s recent_step=4.92s eta=25:31:43 +2026-09-18 18:56:01 | INFO | train rank=0 | step=24970/50000 progress=49.9% loss=0.094001 lr=5.422783e-05 lr_backbone=5.422783e-05 elapsed=25:27:24 avg_step=3.67s recent_step=3.41s eta=25:31:04 +2026-09-18 18:56:35 | INFO | train rank=0 | step=24980/50000 progress=50.0% loss=0.106650 lr=5.419488e-05 lr_backbone=5.419488e-05 elapsed=25:27:58 avg_step=3.67s recent_step=3.39s eta=25:30:25 +2026-09-18 18:57:09 | INFO | train rank=0 | step=24990/50000 progress=50.0% loss=0.094209 lr=5.416192e-05 lr_backbone=5.416192e-05 elapsed=25:28:33 avg_step=3.67s recent_step=3.47s eta=25:29:46 +2026-09-18 18:58:13 | INFO | train rank=0 | step=25000/50000 progress=50.0% loss=0.112242 lr=5.412897e-05 lr_backbone=5.412897e-05 elapsed=25:29:36 avg_step=3.67s recent_step=6.36s eta=25:29:36 +2026-09-18 18:58:51 | INFO | train rank=0 | step=25010/50000 progress=50.0% loss=0.119971 lr=5.409601e-05 lr_backbone=5.409601e-05 elapsed=25:30:14 avg_step=3.67s recent_step=3.79s eta=25:29:01 +2026-09-18 18:59:22 | INFO | train rank=0 | step=25020/50000 progress=50.0% loss=0.092118 lr=5.406305e-05 lr_backbone=5.406305e-05 elapsed=25:30:45 avg_step=3.67s recent_step=3.12s eta=25:28:19 +2026-09-18 18:59:53 | INFO | train rank=0 | step=25030/50000 progress=50.1% loss=0.098199 lr=5.403009e-05 lr_backbone=5.403009e-05 elapsed=25:31:16 avg_step=3.67s recent_step=3.08s eta=25:27:36 +2026-09-18 19:00:39 | INFO | train rank=0 | step=25040/50000 progress=50.1% loss=0.094993 lr=5.399713e-05 lr_backbone=5.399713e-05 elapsed=25:32:03 avg_step=3.67s recent_step=4.66s eta=25:27:09 +2026-09-18 19:01:13 | INFO | train rank=0 | step=25050/50000 progress=50.1% loss=0.134458 lr=5.396416e-05 lr_backbone=5.396416e-05 elapsed=25:32:36 avg_step=3.67s recent_step=3.32s eta=25:26:29 +2026-09-18 19:01:47 | INFO | train rank=0 | step=25060/50000 progress=50.1% loss=0.100040 lr=5.393120e-05 lr_backbone=5.393120e-05 elapsed=25:33:10 avg_step=3.67s recent_step=3.45s eta=25:25:50 +2026-09-18 19:02:20 | INFO | train rank=0 | step=25070/50000 progress=50.1% loss=0.118044 lr=5.389823e-05 lr_backbone=5.389823e-05 elapsed=25:33:43 avg_step=3.67s recent_step=3.30s eta=25:25:10 +2026-09-18 19:03:02 | INFO | train rank=0 | step=25080/50000 progress=50.2% loss=0.132757 lr=5.386526e-05 lr_backbone=5.386526e-05 elapsed=25:34:26 avg_step=3.67s recent_step=4.23s eta=25:24:38 +2026-09-18 19:03:35 | INFO | train rank=0 | step=25090/50000 progress=50.2% loss=0.091424 lr=5.383229e-05 lr_backbone=5.383229e-05 elapsed=25:34:59 avg_step=3.67s recent_step=3.30s eta=25:23:58 +2026-09-18 19:04:10 | INFO | train rank=0 | step=25100/50000 progress=50.2% loss=0.084117 lr=5.379931e-05 lr_backbone=5.379931e-05 elapsed=25:35:33 avg_step=3.67s recent_step=3.47s eta=25:23:19 +2026-09-18 19:04:38 | INFO | train rank=0 | step=25110/50000 progress=50.2% loss=0.093104 lr=5.376634e-05 lr_backbone=5.376634e-05 elapsed=25:36:01 avg_step=3.67s recent_step=2.77s eta=25:22:34 +2026-09-18 19:05:26 | INFO | train rank=0 | step=25120/50000 progress=50.2% loss=0.091494 lr=5.373336e-05 lr_backbone=5.373336e-05 elapsed=25:36:50 avg_step=3.67s recent_step=4.85s eta=25:22:09 +2026-09-18 19:05:57 | INFO | train rank=0 | step=25130/50000 progress=50.3% loss=0.119513 lr=5.370039e-05 lr_backbone=5.370039e-05 elapsed=25:37:20 avg_step=3.67s recent_step=3.07s eta=25:21:26 +2026-09-18 19:06:41 | INFO | train rank=0 | step=25140/50000 progress=50.3% loss=0.114758 lr=5.366741e-05 lr_backbone=5.366741e-05 elapsed=25:38:04 avg_step=3.67s recent_step=4.35s eta=25:20:56 +2026-09-18 19:07:10 | INFO | train rank=0 | step=25150/50000 progress=50.3% loss=0.081325 lr=5.363443e-05 lr_backbone=5.363443e-05 elapsed=25:38:34 avg_step=3.67s recent_step=2.99s eta=25:20:13 +2026-09-18 19:08:03 | INFO | train rank=0 | step=25160/50000 progress=50.3% loss=0.078509 lr=5.360144e-05 lr_backbone=5.360144e-05 elapsed=25:39:27 avg_step=3.67s recent_step=5.29s eta=25:19:52 +2026-09-18 19:08:41 | INFO | train rank=0 | step=25170/50000 progress=50.3% loss=0.094881 lr=5.356846e-05 lr_backbone=5.356846e-05 elapsed=25:40:04 avg_step=3.67s recent_step=3.72s eta=25:19:16 +2026-09-18 19:09:13 | INFO | train rank=0 | step=25180/50000 progress=50.4% loss=0.087109 lr=5.353547e-05 lr_backbone=5.353547e-05 elapsed=25:40:36 avg_step=3.67s recent_step=3.20s eta=25:18:34 +2026-09-18 19:09:43 | INFO | train rank=0 | step=25190/50000 progress=50.4% loss=0.092438 lr=5.350249e-05 lr_backbone=5.350249e-05 elapsed=25:41:07 avg_step=3.67s recent_step=3.06s eta=25:17:52 +2026-09-18 19:10:33 | INFO | train rank=0 | step=25200/50000 progress=50.4% loss=0.091364 lr=5.346950e-05 lr_backbone=5.346950e-05 elapsed=25:41:57 avg_step=3.67s recent_step=5.03s eta=25:17:28 +2026-09-18 19:11:08 | INFO | train rank=0 | step=25210/50000 progress=50.4% loss=0.083133 lr=5.343651e-05 lr_backbone=5.343651e-05 elapsed=25:42:31 avg_step=3.67s recent_step=3.43s eta=25:16:49 +2026-09-18 19:11:41 | INFO | train rank=0 | step=25220/50000 progress=50.4% loss=0.083756 lr=5.340351e-05 lr_backbone=5.340351e-05 elapsed=25:43:04 avg_step=3.67s recent_step=3.28s eta=25:16:09 +2026-09-18 19:12:18 | INFO | train rank=0 | step=25230/50000 progress=50.5% loss=0.097515 lr=5.337052e-05 lr_backbone=5.337052e-05 elapsed=25:43:41 avg_step=3.67s recent_step=3.69s eta=25:15:32 +2026-09-18 19:13:04 | INFO | train rank=0 | step=25240/50000 progress=50.5% loss=0.112307 lr=5.333753e-05 lr_backbone=5.333753e-05 elapsed=25:44:28 avg_step=3.67s recent_step=4.69s eta=25:15:05 +2026-09-18 19:13:32 | INFO | train rank=0 | step=25250/50000 progress=50.5% loss=0.093919 lr=5.330453e-05 lr_backbone=5.330453e-05 elapsed=25:44:55 avg_step=3.67s recent_step=2.71s eta=25:14:19 +2026-09-18 19:14:03 | INFO | train rank=0 | step=25260/50000 progress=50.5% loss=0.082002 lr=5.327153e-05 lr_backbone=5.327153e-05 elapsed=25:45:27 avg_step=3.67s recent_step=3.20s eta=25:13:38 +2026-09-18 19:14:33 | INFO | train rank=0 | step=25270/50000 progress=50.5% loss=0.083143 lr=5.323853e-05 lr_backbone=5.323853e-05 elapsed=25:45:57 avg_step=3.67s recent_step=3.00s eta=25:12:55 +2026-09-18 19:15:31 | INFO | train rank=0 | step=25280/50000 progress=50.6% loss=0.124159 lr=5.320553e-05 lr_backbone=5.320553e-05 elapsed=25:46:55 avg_step=3.67s recent_step=5.78s eta=25:12:39 +2026-09-18 19:15:59 | INFO | train rank=0 | step=25290/50000 progress=50.6% loss=0.080219 lr=5.317253e-05 lr_backbone=5.317253e-05 elapsed=25:47:22 avg_step=3.67s recent_step=2.79s eta=25:11:53 +2026-09-18 19:16:35 | INFO | train rank=0 | step=25300/50000 progress=50.6% loss=0.093396 lr=5.313953e-05 lr_backbone=5.313953e-05 elapsed=25:47:58 avg_step=3.67s recent_step=3.60s eta=25:11:16 +2026-09-18 19:16:59 | INFO | train rank=0 | step=25310/50000 progress=50.6% loss=0.091865 lr=5.310652e-05 lr_backbone=5.310652e-05 elapsed=25:48:22 avg_step=3.67s recent_step=2.40s eta=25:10:27 +2026-09-18 19:17:47 | INFO | train rank=0 | step=25320/50000 progress=50.6% loss=0.079660 lr=5.307351e-05 lr_backbone=5.307351e-05 elapsed=25:49:10 avg_step=3.67s recent_step=4.78s eta=25:10:01 +2026-09-18 19:18:24 | INFO | train rank=0 | step=25330/50000 progress=50.7% loss=0.085420 lr=5.304051e-05 lr_backbone=5.304051e-05 elapsed=25:49:47 avg_step=3.67s recent_step=3.66s eta=25:09:24 +2026-09-18 19:18:57 | INFO | train rank=0 | step=25340/50000 progress=50.7% loss=0.075527 lr=5.300750e-05 lr_backbone=5.300750e-05 elapsed=25:50:20 avg_step=3.67s recent_step=3.32s eta=25:08:44 +2026-09-18 19:19:29 | INFO | train rank=0 | step=25350/50000 progress=50.7% loss=0.086602 lr=5.297449e-05 lr_backbone=5.297449e-05 elapsed=25:50:53 avg_step=3.67s recent_step=3.24s eta=25:08:03 +2026-09-18 19:20:15 | INFO | train rank=0 | step=25360/50000 progress=50.7% loss=0.086752 lr=5.294148e-05 lr_backbone=5.294148e-05 elapsed=25:51:38 avg_step=3.67s recent_step=4.59s eta=25:07:35 +2026-09-18 19:20:48 | INFO | train rank=0 | step=25370/50000 progress=50.7% loss=0.103466 lr=5.290846e-05 lr_backbone=5.290846e-05 elapsed=25:52:12 avg_step=3.67s recent_step=3.33s eta=25:06:55 +2026-09-18 19:21:21 | INFO | train rank=0 | step=25380/50000 progress=50.8% loss=0.099389 lr=5.287545e-05 lr_backbone=5.287545e-05 elapsed=25:52:44 avg_step=3.67s recent_step=3.24s eta=25:06:14 +2026-09-18 19:21:51 | INFO | train rank=0 | step=25390/50000 progress=50.8% loss=0.139149 lr=5.284243e-05 lr_backbone=5.284243e-05 elapsed=25:53:14 avg_step=3.67s recent_step=3.03s eta=25:05:31 +2026-09-18 19:22:44 | INFO | train rank=0 | step=25400/50000 progress=50.8% loss=0.076497 lr=5.280942e-05 lr_backbone=5.280942e-05 elapsed=25:54:07 avg_step=3.67s recent_step=5.29s eta=25:05:10 +2026-09-18 19:23:18 | INFO | train rank=0 | step=25410/50000 progress=50.8% loss=0.086055 lr=5.277640e-05 lr_backbone=5.277640e-05 elapsed=25:54:41 avg_step=3.67s recent_step=3.40s eta=25:04:31 +2026-09-18 19:23:52 | INFO | train rank=0 | step=25420/50000 progress=50.8% loss=0.104975 lr=5.274338e-05 lr_backbone=5.274338e-05 elapsed=25:55:15 avg_step=3.67s recent_step=3.40s eta=25:03:52 +2026-09-18 19:24:23 | INFO | train rank=0 | step=25430/50000 progress=50.9% loss=0.102002 lr=5.271036e-05 lr_backbone=5.271036e-05 elapsed=25:55:47 avg_step=3.67s recent_step=3.14s eta=25:03:10 +2026-09-18 19:25:14 | INFO | train rank=0 | step=25440/50000 progress=50.9% loss=0.094641 lr=5.267734e-05 lr_backbone=5.267734e-05 elapsed=25:56:38 avg_step=3.67s recent_step=5.08s eta=25:02:47 +2026-09-18 19:25:47 | INFO | train rank=0 | step=25450/50000 progress=50.9% loss=0.075914 lr=5.264432e-05 lr_backbone=5.264432e-05 elapsed=25:57:11 avg_step=3.67s recent_step=3.31s eta=25:02:07 +2026-09-18 19:26:09 | INFO | train rank=0 | step=25460/50000 progress=50.9% loss=0.129269 lr=5.261129e-05 lr_backbone=5.261129e-05 elapsed=25:57:33 avg_step=3.67s recent_step=2.20s eta=25:01:16 +2026-09-18 19:26:43 | INFO | train rank=0 | step=25470/50000 progress=50.9% loss=0.072891 lr=5.257827e-05 lr_backbone=5.257827e-05 elapsed=25:58:07 avg_step=3.67s recent_step=3.39s eta=25:00:36 +2026-09-18 19:27:33 | INFO | train rank=0 | step=25480/50000 progress=51.0% loss=0.118777 lr=5.254524e-05 lr_backbone=5.254524e-05 elapsed=25:58:57 avg_step=3.67s recent_step=5.00s eta=25:00:12 +2026-09-18 19:28:09 | INFO | train rank=0 | step=25490/50000 progress=51.0% loss=0.087084 lr=5.251222e-05 lr_backbone=5.251222e-05 elapsed=25:59:33 avg_step=3.67s recent_step=3.59s eta=24:59:35 +2026-09-18 19:28:46 | INFO | train rank=0 | step=25500/50000 progress=51.0% loss=0.088785 lr=5.247919e-05 lr_backbone=5.247919e-05 elapsed=26:00:09 avg_step=3.67s recent_step=3.68s eta=24:58:58 +2026-09-18 19:29:17 | INFO | train rank=0 | step=25510/50000 progress=51.0% loss=0.079218 lr=5.244616e-05 lr_backbone=5.244616e-05 elapsed=26:00:40 avg_step=3.67s recent_step=3.11s eta=24:58:16 +2026-09-18 19:30:15 | INFO | train rank=0 | step=25520/50000 progress=51.0% loss=0.096510 lr=5.241313e-05 lr_backbone=5.241313e-05 elapsed=26:01:39 avg_step=3.67s recent_step=5.83s eta=24:58:00 +2026-09-18 19:30:40 | INFO | train rank=0 | step=25530/50000 progress=51.1% loss=0.089873 lr=5.238010e-05 lr_backbone=5.238010e-05 elapsed=26:02:04 avg_step=3.67s recent_step=2.50s eta=24:57:12 +2026-09-18 19:31:16 | INFO | train rank=0 | step=25540/50000 progress=51.1% loss=0.089597 lr=5.234706e-05 lr_backbone=5.234706e-05 elapsed=26:02:39 avg_step=3.67s recent_step=3.53s eta=24:56:34 +2026-09-18 19:31:45 | INFO | train rank=0 | step=25550/50000 progress=51.1% loss=0.072636 lr=5.231403e-05 lr_backbone=5.231403e-05 elapsed=26:03:08 avg_step=3.67s recent_step=2.92s eta=24:55:50 +2026-09-18 19:32:37 | INFO | train rank=0 | step=25560/50000 progress=51.1% loss=0.072349 lr=5.228100e-05 lr_backbone=5.228100e-05 elapsed=26:04:00 avg_step=3.67s recent_step=5.22s eta=24:55:28 +2026-09-18 19:33:07 | INFO | train rank=0 | step=25570/50000 progress=51.1% loss=0.114634 lr=5.224796e-05 lr_backbone=5.224796e-05 elapsed=26:04:31 avg_step=3.67s recent_step=3.03s eta=24:54:45 +2026-09-18 19:33:41 | INFO | train rank=0 | step=25580/50000 progress=51.2% loss=0.105597 lr=5.221492e-05 lr_backbone=5.221492e-05 elapsed=26:05:04 avg_step=3.67s recent_step=3.32s eta=24:54:05 +2026-09-18 19:34:11 | INFO | train rank=0 | step=25590/50000 progress=51.2% loss=0.125612 lr=5.218189e-05 lr_backbone=5.218189e-05 elapsed=26:05:34 avg_step=3.67s recent_step=3.03s eta=24:53:23 +2026-09-18 19:35:10 | INFO | train rank=0 | step=25600/50000 progress=51.2% loss=0.085577 lr=5.214885e-05 lr_backbone=5.214885e-05 elapsed=26:06:33 avg_step=3.67s recent_step=5.89s eta=24:53:07 +2026-09-18 19:35:40 | INFO | train rank=0 | step=25610/50000 progress=51.2% loss=0.098735 lr=5.211581e-05 lr_backbone=5.211581e-05 elapsed=26:07:03 avg_step=3.67s recent_step=3.00s eta=24:52:24 +2026-09-18 19:36:19 | INFO | train rank=0 | step=25620/50000 progress=51.2% loss=0.067661 lr=5.208277e-05 lr_backbone=5.208277e-05 elapsed=26:07:42 avg_step=3.67s recent_step=3.90s eta=24:51:49 +2026-09-18 19:36:50 | INFO | train rank=0 | step=25630/50000 progress=51.3% loss=0.116743 lr=5.204973e-05 lr_backbone=5.204973e-05 elapsed=26:08:13 avg_step=3.67s recent_step=3.13s eta=24:51:08 +2026-09-18 19:37:37 | INFO | train rank=0 | step=25640/50000 progress=51.3% loss=0.087481 lr=5.201669e-05 lr_backbone=5.201669e-05 elapsed=26:09:00 avg_step=3.67s recent_step=4.68s eta=24:50:41 +2026-09-18 19:38:13 | INFO | train rank=0 | step=25650/50000 progress=51.3% loss=0.091438 lr=5.198364e-05 lr_backbone=5.198364e-05 elapsed=26:09:37 avg_step=3.67s recent_step=3.65s eta=24:50:04 +2026-09-18 19:38:37 | INFO | train rank=0 | step=25660/50000 progress=51.3% loss=0.091297 lr=5.195060e-05 lr_backbone=5.195060e-05 elapsed=26:10:01 avg_step=3.67s recent_step=2.40s eta=24:49:15 +2026-09-18 19:39:08 | INFO | train rank=0 | step=25670/50000 progress=51.3% loss=0.085414 lr=5.191755e-05 lr_backbone=5.191755e-05 elapsed=26:10:31 avg_step=3.67s recent_step=3.03s eta=24:48:32 +2026-09-18 19:40:01 | INFO | train rank=0 | step=25680/50000 progress=51.4% loss=0.080695 lr=5.188451e-05 lr_backbone=5.188451e-05 elapsed=26:11:25 avg_step=3.67s recent_step=5.38s eta=24:48:11 +2026-09-18 19:40:32 | INFO | train rank=0 | step=25690/50000 progress=51.4% loss=0.100188 lr=5.185146e-05 lr_backbone=5.185146e-05 elapsed=26:11:55 avg_step=3.67s recent_step=3.07s eta=24:47:29 +2026-09-18 19:41:04 | INFO | train rank=0 | step=25700/50000 progress=51.4% loss=0.085880 lr=5.181842e-05 lr_backbone=5.181842e-05 elapsed=26:12:27 avg_step=3.67s recent_step=3.14s eta=24:46:47 +2026-09-18 19:41:28 | INFO | train rank=0 | step=25710/50000 progress=51.4% loss=0.077020 lr=5.178537e-05 lr_backbone=5.178537e-05 elapsed=26:12:51 avg_step=3.67s recent_step=2.45s eta=24:45:59 +2026-09-18 19:42:24 | INFO | train rank=0 | step=25720/50000 progress=51.4% loss=0.083931 lr=5.175232e-05 lr_backbone=5.175232e-05 elapsed=26:13:47 avg_step=3.67s recent_step=5.59s eta=24:45:41 +2026-09-18 19:42:59 | INFO | train rank=0 | step=25730/50000 progress=51.5% loss=0.086087 lr=5.171927e-05 lr_backbone=5.171927e-05 elapsed=26:14:23 avg_step=3.67s recent_step=3.52s eta=24:45:02 +2026-09-18 19:43:30 | INFO | train rank=0 | step=25740/50000 progress=51.5% loss=0.084803 lr=5.168622e-05 lr_backbone=5.168622e-05 elapsed=26:14:53 avg_step=3.67s recent_step=3.08s eta=24:44:20 +2026-09-18 19:44:01 | INFO | train rank=0 | step=25750/50000 progress=51.5% loss=0.086777 lr=5.165317e-05 lr_backbone=5.165317e-05 elapsed=26:15:24 avg_step=3.67s recent_step=3.07s eta=24:43:38 +2026-09-18 19:44:58 | INFO | train rank=0 | step=25760/50000 progress=51.5% loss=0.135273 lr=5.162012e-05 lr_backbone=5.162012e-05 elapsed=26:16:21 avg_step=3.67s recent_step=5.74s eta=24:43:20 +2026-09-18 19:45:32 | INFO | train rank=0 | step=25770/50000 progress=51.5% loss=0.097028 lr=5.158706e-05 lr_backbone=5.158706e-05 elapsed=26:16:56 avg_step=3.67s recent_step=3.42s eta=24:42:41 +2026-09-18 19:46:01 | INFO | train rank=0 | step=25780/50000 progress=51.6% loss=0.200826 lr=5.155401e-05 lr_backbone=5.155401e-05 elapsed=26:17:25 avg_step=3.67s recent_step=2.92s eta=24:41:58 +2026-09-18 19:46:32 | INFO | train rank=0 | step=25790/50000 progress=51.6% loss=0.133711 lr=5.152096e-05 lr_backbone=5.152096e-05 elapsed=26:17:55 avg_step=3.67s recent_step=3.02s eta=24:41:15 +2026-09-18 19:47:23 | INFO | train rank=0 | step=25800/50000 progress=51.6% loss=0.088227 lr=5.148790e-05 lr_backbone=5.148790e-05 elapsed=26:18:47 avg_step=3.67s recent_step=5.16s eta=24:40:52 +2026-09-18 19:47:55 | INFO | train rank=0 | step=25810/50000 progress=51.6% loss=0.126814 lr=5.145485e-05 lr_backbone=5.145485e-05 elapsed=26:19:18 avg_step=3.67s recent_step=3.15s eta=24:40:10 +2026-09-18 19:48:24 | INFO | train rank=0 | step=25820/50000 progress=51.6% loss=0.087019 lr=5.142179e-05 lr_backbone=5.142179e-05 elapsed=26:19:48 avg_step=3.67s recent_step=2.95s eta=24:39:27 +2026-09-18 19:48:56 | INFO | train rank=0 | step=25830/50000 progress=51.7% loss=0.101019 lr=5.138874e-05 lr_backbone=5.138874e-05 elapsed=26:20:19 avg_step=3.67s recent_step=3.13s eta=24:38:45 +2026-09-18 19:49:51 | INFO | train rank=0 | step=25840/50000 progress=51.7% loss=0.076035 lr=5.135568e-05 lr_backbone=5.135568e-05 elapsed=26:21:15 avg_step=3.67s recent_step=5.56s eta=24:38:26 +2026-09-18 19:50:22 | INFO | train rank=0 | step=25850/50000 progress=51.7% loss=0.084698 lr=5.132262e-05 lr_backbone=5.132262e-05 elapsed=26:21:45 avg_step=3.67s recent_step=3.06s eta=24:37:44 +2026-09-18 19:50:48 | INFO | train rank=0 | step=25860/50000 progress=51.7% loss=0.088717 lr=5.128956e-05 lr_backbone=5.128956e-05 elapsed=26:22:11 avg_step=3.67s recent_step=2.63s eta=24:36:57 +2026-09-18 19:51:23 | INFO | train rank=0 | step=25870/50000 progress=51.7% loss=0.102992 lr=5.125650e-05 lr_backbone=5.125650e-05 elapsed=26:22:46 avg_step=3.67s recent_step=3.50s eta=24:36:19 +2026-09-18 19:52:14 | INFO | train rank=0 | step=25880/50000 progress=51.8% loss=0.104027 lr=5.122345e-05 lr_backbone=5.122345e-05 elapsed=26:23:37 avg_step=3.67s recent_step=5.07s eta=24:35:55 +2026-09-18 19:52:39 | INFO | train rank=0 | step=25890/50000 progress=51.8% loss=0.093949 lr=5.119039e-05 lr_backbone=5.119039e-05 elapsed=26:24:02 avg_step=3.67s recent_step=2.52s eta=24:35:08 +2026-09-18 19:53:09 | INFO | train rank=0 | step=25900/50000 progress=51.8% loss=0.077911 lr=5.115733e-05 lr_backbone=5.115733e-05 elapsed=26:24:32 avg_step=3.67s recent_step=2.98s eta=24:34:25 +2026-09-18 19:53:40 | INFO | train rank=0 | step=25910/50000 progress=51.8% loss=0.076379 lr=5.112426e-05 lr_backbone=5.112426e-05 elapsed=26:25:04 avg_step=3.67s recent_step=3.16s eta=24:33:43 +2026-09-18 19:54:37 | INFO | train rank=0 | step=25920/50000 progress=51.8% loss=0.114038 lr=5.109120e-05 lr_backbone=5.109120e-05 elapsed=26:26:00 avg_step=3.67s recent_step=5.63s eta=24:33:25 +2026-09-18 19:55:10 | INFO | train rank=0 | step=25930/50000 progress=51.9% loss=0.075914 lr=5.105814e-05 lr_backbone=5.105814e-05 elapsed=26:26:33 avg_step=3.67s recent_step=3.33s eta=24:32:45 +2026-09-18 19:55:51 | INFO | train rank=0 | step=25940/50000 progress=51.9% loss=0.107288 lr=5.102508e-05 lr_backbone=5.102508e-05 elapsed=26:27:14 avg_step=3.67s recent_step=4.12s eta=24:32:12 +2026-09-18 19:56:20 | INFO | train rank=0 | step=25950/50000 progress=51.9% loss=0.075711 lr=5.099202e-05 lr_backbone=5.099202e-05 elapsed=26:27:43 avg_step=3.67s recent_step=2.88s eta=24:31:28 +2026-09-18 19:57:11 | INFO | train rank=0 | step=25960/50000 progress=51.9% loss=0.064077 lr=5.095895e-05 lr_backbone=5.095895e-05 elapsed=26:28:34 avg_step=3.67s recent_step=5.08s eta=24:31:05 +2026-09-18 19:57:44 | INFO | train rank=0 | step=25970/50000 progress=51.9% loss=0.086786 lr=5.092589e-05 lr_backbone=5.092589e-05 elapsed=26:29:07 avg_step=3.67s recent_step=3.31s eta=24:30:24 +2026-09-18 19:58:18 | INFO | train rank=0 | step=25980/50000 progress=52.0% loss=0.075359 lr=5.089283e-05 lr_backbone=5.089283e-05 elapsed=26:29:42 avg_step=3.67s recent_step=3.44s eta=24:29:46 +2026-09-18 19:58:56 | INFO | train rank=0 | step=25990/50000 progress=52.0% loss=0.079623 lr=5.085976e-05 lr_backbone=5.085976e-05 elapsed=26:30:19 avg_step=3.67s recent_step=3.74s eta=24:29:10 +2026-09-18 19:59:45 | INFO | train rank=0 | step=26000/50000 progress=52.0% loss=0.113680 lr=5.082670e-05 lr_backbone=5.082670e-05 elapsed=26:31:09 avg_step=3.67s recent_step=4.98s eta=24:28:45 +2026-09-18 20:00:15 | INFO | train rank=0 | step=26010/50000 progress=52.0% loss=0.102074 lr=5.079363e-05 lr_backbone=5.079363e-05 elapsed=26:31:38 avg_step=3.67s recent_step=2.93s eta=24:28:01 +2026-09-18 20:00:46 | INFO | train rank=0 | step=26020/50000 progress=52.0% loss=0.106920 lr=5.076057e-05 lr_backbone=5.076057e-05 elapsed=26:32:09 avg_step=3.67s recent_step=3.11s eta=24:27:19 +2026-09-18 20:01:22 | INFO | train rank=0 | step=26030/50000 progress=52.1% loss=0.091675 lr=5.072750e-05 lr_backbone=5.072750e-05 elapsed=26:32:45 avg_step=3.67s recent_step=3.57s eta=24:26:42 +2026-09-18 20:02:08 | INFO | train rank=0 | step=26040/50000 progress=52.1% loss=0.110984 lr=5.069443e-05 lr_backbone=5.069443e-05 elapsed=26:33:32 avg_step=3.67s recent_step=4.69s eta=24:26:14 +2026-09-18 20:02:46 | INFO | train rank=0 | step=26050/50000 progress=52.1% loss=0.095644 lr=5.066137e-05 lr_backbone=5.066137e-05 elapsed=26:34:10 avg_step=3.67s recent_step=3.79s eta=24:25:39 +2026-09-18 20:03:20 | INFO | train rank=0 | step=26060/50000 progress=52.1% loss=0.093897 lr=5.062830e-05 lr_backbone=5.062830e-05 elapsed=26:34:43 avg_step=3.67s recent_step=3.37s eta=24:24:59 +2026-09-18 20:03:46 | INFO | train rank=0 | step=26070/50000 progress=52.1% loss=0.074395 lr=5.059524e-05 lr_backbone=5.059524e-05 elapsed=26:35:09 avg_step=3.67s recent_step=2.59s eta=24:24:13 +2026-09-18 20:04:39 | INFO | train rank=0 | step=26080/50000 progress=52.2% loss=0.086163 lr=5.056217e-05 lr_backbone=5.056217e-05 elapsed=26:36:03 avg_step=3.67s recent_step=5.35s eta=24:23:51 +2026-09-18 20:05:08 | INFO | train rank=0 | step=26090/50000 progress=52.2% loss=0.094920 lr=5.052910e-05 lr_backbone=5.052910e-05 elapsed=26:36:31 avg_step=3.67s recent_step=2.81s eta=24:23:07 +2026-09-18 20:05:35 | INFO | train rank=0 | step=26100/50000 progress=52.2% loss=0.104927 lr=5.049603e-05 lr_backbone=5.049603e-05 elapsed=26:36:59 avg_step=3.67s recent_step=2.79s eta=24:22:22 +2026-09-18 20:06:08 | INFO | train rank=0 | step=26110/50000 progress=52.2% loss=0.134360 lr=5.046296e-05 lr_backbone=5.046296e-05 elapsed=26:37:31 avg_step=3.67s recent_step=3.22s eta=24:21:41 +2026-09-18 20:07:00 | INFO | train rank=0 | step=26120/50000 progress=52.2% loss=0.093671 lr=5.042990e-05 lr_backbone=5.042990e-05 elapsed=26:38:24 avg_step=3.67s recent_step=5.28s eta=24:21:19 +2026-09-18 20:07:37 | INFO | train rank=0 | step=26130/50000 progress=52.3% loss=0.116695 lr=5.039683e-05 lr_backbone=5.039683e-05 elapsed=26:39:00 avg_step=3.67s recent_step=3.65s eta=24:20:42 +2026-09-18 20:08:14 | INFO | train rank=0 | step=26140/50000 progress=52.3% loss=0.108849 lr=5.036376e-05 lr_backbone=5.036376e-05 elapsed=26:39:38 avg_step=3.67s recent_step=3.74s eta=24:20:06 +2026-09-18 20:08:45 | INFO | train rank=0 | step=26150/50000 progress=52.3% loss=0.102996 lr=5.033069e-05 lr_backbone=5.033069e-05 elapsed=26:40:08 avg_step=3.67s recent_step=3.05s eta=24:19:24 +2026-09-18 20:09:36 | INFO | train rank=0 | step=26160/50000 progress=52.3% loss=0.104031 lr=5.029762e-05 lr_backbone=5.029762e-05 elapsed=26:40:59 avg_step=3.67s recent_step=5.13s eta=24:19:00 +2026-09-18 20:10:14 | INFO | train rank=0 | step=26170/50000 progress=52.3% loss=0.100873 lr=5.026455e-05 lr_backbone=5.026455e-05 elapsed=26:41:37 avg_step=3.67s recent_step=3.80s eta=24:18:25 +2026-09-18 20:10:47 | INFO | train rank=0 | step=26180/50000 progress=52.4% loss=0.094480 lr=5.023148e-05 lr_backbone=5.023148e-05 elapsed=26:42:10 avg_step=3.67s recent_step=3.25s eta=24:17:44 +2026-09-18 20:11:16 | INFO | train rank=0 | step=26190/50000 progress=52.4% loss=0.089899 lr=5.019842e-05 lr_backbone=5.019842e-05 elapsed=26:42:39 avg_step=3.67s recent_step=2.94s eta=24:17:01 +2026-09-18 20:12:12 | INFO | train rank=0 | step=26200/50000 progress=52.4% loss=0.143197 lr=5.016535e-05 lr_backbone=5.016535e-05 elapsed=26:43:36 avg_step=3.67s recent_step=5.65s eta=24:16:42 +2026-09-18 20:12:45 | INFO | train rank=0 | step=26210/50000 progress=52.4% loss=0.104746 lr=5.013228e-05 lr_backbone=5.013228e-05 elapsed=26:44:08 avg_step=3.67s recent_step=3.24s eta=24:16:01 +2026-09-18 20:13:14 | INFO | train rank=0 | step=26220/50000 progress=52.4% loss=0.074300 lr=5.009921e-05 lr_backbone=5.009921e-05 elapsed=26:44:38 avg_step=3.67s recent_step=2.94s eta=24:15:18 +2026-09-18 20:13:40 | INFO | train rank=0 | step=26230/50000 progress=52.5% loss=0.098481 lr=5.006614e-05 lr_backbone=5.006614e-05 elapsed=26:45:04 avg_step=3.67s recent_step=2.61s eta=24:14:32 +2026-09-18 20:14:33 | INFO | train rank=0 | step=26240/50000 progress=52.5% loss=0.093711 lr=5.003307e-05 lr_backbone=5.003307e-05 elapsed=26:45:56 avg_step=3.67s recent_step=5.23s eta=24:14:09 +2026-09-18 20:15:03 | INFO | train rank=0 | step=26250/50000 progress=52.5% loss=0.086189 lr=5.000000e-05 lr_backbone=5.000000e-05 elapsed=26:46:27 avg_step=3.67s recent_step=3.07s eta=24:13:27 +2026-09-18 20:15:42 | INFO | train rank=0 | step=26260/50000 progress=52.5% loss=0.101183 lr=4.996693e-05 lr_backbone=4.996693e-05 elapsed=26:47:05 avg_step=3.67s recent_step=3.85s eta=24:12:52 +2026-09-18 20:16:15 | INFO | train rank=0 | step=26270/50000 progress=52.5% loss=0.084276 lr=4.993386e-05 lr_backbone=4.993386e-05 elapsed=26:47:38 avg_step=3.67s recent_step=3.33s eta=24:12:12 +2026-09-18 20:17:06 | INFO | train rank=0 | step=26280/50000 progress=52.6% loss=0.109726 lr=4.990079e-05 lr_backbone=4.990079e-05 elapsed=26:48:29 avg_step=3.67s recent_step=5.05s eta=24:11:48 +2026-09-18 20:17:39 | INFO | train rank=0 | step=26290/50000 progress=52.6% loss=0.091809 lr=4.986772e-05 lr_backbone=4.986772e-05 elapsed=26:49:02 avg_step=3.67s recent_step=3.34s eta=24:11:08 +2026-09-18 20:18:10 | INFO | train rank=0 | step=26300/50000 progress=52.6% loss=0.091391 lr=4.983465e-05 lr_backbone=4.983465e-05 elapsed=26:49:33 avg_step=3.67s recent_step=3.05s eta=24:10:26 +2026-09-18 20:18:41 | INFO | train rank=0 | step=26310/50000 progress=52.6% loss=0.103466 lr=4.980158e-05 lr_backbone=4.980158e-05 elapsed=26:50:04 avg_step=3.67s recent_step=3.10s eta=24:09:44 +2026-09-18 20:19:23 | INFO | train rank=0 | step=26320/50000 progress=52.6% loss=0.077819 lr=4.976852e-05 lr_backbone=4.976852e-05 elapsed=26:50:47 avg_step=3.67s recent_step=4.28s eta=24:09:13 +2026-09-18 20:19:50 | INFO | train rank=0 | step=26330/50000 progress=52.7% loss=0.157522 lr=4.973545e-05 lr_backbone=4.973545e-05 elapsed=26:51:13 avg_step=3.67s recent_step=2.62s eta=24:08:26 +2026-09-18 20:20:26 | INFO | train rank=0 | step=26340/50000 progress=52.7% loss=0.111279 lr=4.970238e-05 lr_backbone=4.970238e-05 elapsed=26:51:49 avg_step=3.67s recent_step=3.63s eta=24:07:49 +2026-09-18 20:20:57 | INFO | train rank=0 | step=26350/50000 progress=52.7% loss=0.074829 lr=4.966931e-05 lr_backbone=4.966931e-05 elapsed=26:52:20 avg_step=3.67s recent_step=3.13s eta=24:07:08 +2026-09-18 20:21:51 | INFO | train rank=0 | step=26360/50000 progress=52.7% loss=0.078667 lr=4.963624e-05 lr_backbone=4.963624e-05 elapsed=26:53:14 avg_step=3.67s recent_step=5.39s eta=24:06:46 +2026-09-18 20:22:27 | INFO | train rank=0 | step=26370/50000 progress=52.7% loss=0.093596 lr=4.960317e-05 lr_backbone=4.960317e-05 elapsed=26:53:50 avg_step=3.67s recent_step=3.59s eta=24:06:09 +2026-09-18 20:22:58 | INFO | train rank=0 | step=26380/50000 progress=52.8% loss=0.080637 lr=4.957010e-05 lr_backbone=4.957010e-05 elapsed=26:54:21 avg_step=3.67s recent_step=3.07s eta=24:05:27 +2026-09-18 20:23:34 | INFO | train rank=0 | step=26390/50000 progress=52.8% loss=0.094641 lr=4.953704e-05 lr_backbone=4.953704e-05 elapsed=26:54:57 avg_step=3.67s recent_step=3.64s eta=24:04:50 +2026-09-18 20:24:21 | INFO | train rank=0 | step=26400/50000 progress=52.8% loss=0.102535 lr=4.950397e-05 lr_backbone=4.950397e-05 elapsed=26:55:44 avg_step=3.67s recent_step=4.65s eta=24:04:22 +2026-09-18 20:24:54 | INFO | train rank=0 | step=26410/50000 progress=52.8% loss=0.095340 lr=4.947090e-05 lr_backbone=4.947090e-05 elapsed=26:56:17 avg_step=3.67s recent_step=3.32s eta=24:03:42 +2026-09-18 20:25:28 | INFO | train rank=0 | step=26420/50000 progress=52.8% loss=0.103514 lr=4.943783e-05 lr_backbone=4.943783e-05 elapsed=26:56:51 avg_step=3.67s recent_step=3.43s eta=24:03:03 +2026-09-18 20:26:00 | INFO | train rank=0 | step=26430/50000 progress=52.9% loss=0.084561 lr=4.940476e-05 lr_backbone=4.940476e-05 elapsed=26:57:24 avg_step=3.67s recent_step=3.22s eta=24:02:22 +2026-09-18 20:26:53 | INFO | train rank=0 | step=26440/50000 progress=52.9% loss=0.131962 lr=4.937170e-05 lr_backbone=4.937170e-05 elapsed=26:58:17 avg_step=3.67s recent_step=5.31s eta=24:02:00 +2026-09-18 20:27:28 | INFO | train rank=0 | step=26450/50000 progress=52.9% loss=0.080650 lr=4.933863e-05 lr_backbone=4.933863e-05 elapsed=26:58:51 avg_step=3.67s recent_step=3.43s eta=24:01:21 +2026-09-18 20:27:57 | INFO | train rank=0 | step=26460/50000 progress=52.9% loss=0.070708 lr=4.930557e-05 lr_backbone=4.930557e-05 elapsed=26:59:20 avg_step=3.67s recent_step=2.90s eta=24:00:38 +2026-09-18 20:28:22 | INFO | train rank=0 | step=26470/50000 progress=52.9% loss=0.083248 lr=4.927250e-05 lr_backbone=4.927250e-05 elapsed=26:59:45 avg_step=3.67s recent_step=2.55s eta=23:59:51 +2026-09-18 20:29:17 | INFO | train rank=0 | step=26480/50000 progress=53.0% loss=0.073095 lr=4.923943e-05 lr_backbone=4.923943e-05 elapsed=27:00:40 avg_step=3.67s recent_step=5.46s eta=23:59:30 +2026-09-18 20:29:52 | INFO | train rank=0 | step=26490/50000 progress=53.0% loss=0.104622 lr=4.920637e-05 lr_backbone=4.920637e-05 elapsed=27:01:15 avg_step=3.67s recent_step=3.47s eta=23:58:52 +2026-09-18 20:30:23 | INFO | train rank=0 | step=26500/50000 progress=53.0% loss=0.091803 lr=4.917330e-05 lr_backbone=4.917330e-05 elapsed=27:01:46 avg_step=3.67s recent_step=3.15s eta=23:58:11 +2026-09-18 20:30:59 | INFO | train rank=0 | step=26510/50000 progress=53.0% loss=0.069117 lr=4.914024e-05 lr_backbone=4.914024e-05 elapsed=27:02:23 avg_step=3.67s recent_step=3.63s eta=23:57:33 +2026-09-18 20:31:36 | INFO | train rank=0 | step=26520/50000 progress=53.0% loss=0.078917 lr=4.910717e-05 lr_backbone=4.910717e-05 elapsed=27:02:59 avg_step=3.67s recent_step=3.62s eta=23:56:56 +2026-09-18 20:32:10 | INFO | train rank=0 | step=26530/50000 progress=53.1% loss=0.094837 lr=4.907411e-05 lr_backbone=4.907411e-05 elapsed=27:03:34 avg_step=3.67s recent_step=3.49s eta=23:56:18 +2026-09-18 20:32:43 | INFO | train rank=0 | step=26540/50000 progress=53.1% loss=0.105439 lr=4.904105e-05 lr_backbone=4.904105e-05 elapsed=27:04:06 avg_step=3.67s recent_step=3.22s eta=23:55:37 +2026-09-18 20:33:15 | INFO | train rank=0 | step=26550/50000 progress=53.1% loss=0.089325 lr=4.900798e-05 lr_backbone=4.900798e-05 elapsed=27:04:38 avg_step=3.67s recent_step=3.25s eta=23:54:57 +2026-09-18 20:34:08 | INFO | train rank=0 | step=26560/50000 progress=53.1% loss=0.092364 lr=4.897492e-05 lr_backbone=4.897492e-05 elapsed=27:05:31 avg_step=3.67s recent_step=5.28s eta=23:54:34 +2026-09-18 20:34:36 | INFO | train rank=0 | step=26570/50000 progress=53.1% loss=0.076959 lr=4.894186e-05 lr_backbone=4.894186e-05 elapsed=27:06:00 avg_step=3.67s recent_step=2.84s eta=23:53:50 +2026-09-18 20:35:09 | INFO | train rank=0 | step=26580/50000 progress=53.2% loss=0.106936 lr=4.890880e-05 lr_backbone=4.890880e-05 elapsed=27:06:32 avg_step=3.67s recent_step=3.27s eta=23:53:10 +2026-09-18 20:35:38 | INFO | train rank=0 | step=26590/50000 progress=53.2% loss=0.085976 lr=4.887574e-05 lr_backbone=4.887574e-05 elapsed=27:07:01 avg_step=3.67s recent_step=2.85s eta=23:52:26 +2026-09-18 20:36:13 | INFO | train rank=0 | step=26600/50000 progress=53.2% loss=0.087499 lr=4.884267e-05 lr_backbone=4.884267e-05 elapsed=27:07:36 avg_step=3.67s recent_step=3.54s eta=23:51:48 +2026-09-18 20:36:39 | INFO | train rank=0 | step=26610/50000 progress=53.2% loss=0.097247 lr=4.880961e-05 lr_backbone=4.880961e-05 elapsed=27:08:02 avg_step=3.67s recent_step=2.60s eta=23:51:02 +2026-09-18 20:36:57 | INFO | train rank=0 | step=26620/50000 progress=53.2% loss=0.079027 lr=4.877655e-05 lr_backbone=4.877655e-05 elapsed=27:08:21 avg_step=3.67s recent_step=1.83s eta=23:50:09 +2026-09-18 20:37:19 | INFO | train rank=0 | step=26630/50000 progress=53.3% loss=0.105578 lr=4.874350e-05 lr_backbone=4.874350e-05 elapsed=27:08:43 avg_step=3.67s recent_step=2.19s eta=23:49:19 +2026-09-18 20:37:47 | INFO | train rank=0 | step=26640/50000 progress=53.3% loss=0.084401 lr=4.871044e-05 lr_backbone=4.871044e-05 elapsed=27:09:10 avg_step=3.67s recent_step=2.73s eta=23:48:35 +2026-09-18 20:38:08 | INFO | train rank=0 | step=26650/50000 progress=53.3% loss=0.102826 lr=4.867738e-05 lr_backbone=4.867738e-05 elapsed=27:09:31 avg_step=3.67s recent_step=2.10s eta=23:47:44 +2026-09-18 20:38:27 | INFO | train rank=0 | step=26660/50000 progress=53.3% loss=0.114751 lr=4.864432e-05 lr_backbone=4.864432e-05 elapsed=27:09:50 avg_step=3.67s recent_step=1.93s eta=23:46:52 +2026-09-18 20:38:51 | INFO | train rank=0 | step=26670/50000 progress=53.3% loss=0.098101 lr=4.861126e-05 lr_backbone=4.861126e-05 elapsed=27:10:15 avg_step=3.67s recent_step=2.45s eta=23:46:05 +2026-09-18 20:39:12 | INFO | train rank=0 | step=26680/50000 progress=53.4% loss=0.087802 lr=4.857821e-05 lr_backbone=4.857821e-05 elapsed=27:10:35 avg_step=3.67s recent_step=2.05s eta=23:45:14 +2026-09-18 20:39:35 | INFO | train rank=0 | step=26690/50000 progress=53.4% loss=0.102204 lr=4.854515e-05 lr_backbone=4.854515e-05 elapsed=27:10:58 avg_step=3.67s recent_step=2.33s eta=23:44:26 +2026-09-18 20:39:52 | INFO | train rank=0 | step=26700/50000 progress=53.4% loss=0.087975 lr=4.851210e-05 lr_backbone=4.851210e-05 elapsed=27:11:15 avg_step=3.67s recent_step=1.69s eta=23:43:32 +2026-09-18 20:40:05 | INFO | train rank=0 | step=26710/50000 progress=53.4% loss=0.069530 lr=4.847904e-05 lr_backbone=4.847904e-05 elapsed=27:11:28 avg_step=3.66s recent_step=1.26s eta=23:42:34 +2026-09-18 20:40:29 | INFO | train rank=0 | step=26720/50000 progress=53.4% loss=0.105455 lr=4.844599e-05 lr_backbone=4.844599e-05 elapsed=27:11:53 avg_step=3.66s recent_step=2.47s eta=23:41:47 +2026-09-18 20:41:00 | INFO | train rank=0 | step=26730/50000 progress=53.5% loss=0.091968 lr=4.841294e-05 lr_backbone=4.841294e-05 elapsed=27:12:23 avg_step=3.66s recent_step=3.05s eta=23:41:05 +2026-09-18 20:41:33 | INFO | train rank=0 | step=26740/50000 progress=53.5% loss=0.098838 lr=4.837988e-05 lr_backbone=4.837988e-05 elapsed=27:12:56 avg_step=3.66s recent_step=3.29s eta=23:40:25 +2026-09-18 20:42:09 | INFO | train rank=0 | step=26750/50000 progress=53.5% loss=0.083382 lr=4.834683e-05 lr_backbone=4.834683e-05 elapsed=27:13:32 avg_step=3.66s recent_step=3.65s eta=23:39:48 +2026-09-18 20:42:43 | INFO | train rank=0 | step=26760/50000 progress=53.5% loss=0.096445 lr=4.831378e-05 lr_backbone=4.831378e-05 elapsed=27:14:06 avg_step=3.66s recent_step=3.41s eta=23:39:09 +2026-09-18 20:43:31 | INFO | train rank=0 | step=26770/50000 progress=53.5% loss=0.062924 lr=4.828073e-05 lr_backbone=4.828073e-05 elapsed=27:14:54 avg_step=3.66s recent_step=4.77s eta=23:38:42 +2026-09-18 20:44:21 | INFO | train rank=0 | step=26780/50000 progress=53.6% loss=0.121253 lr=4.824768e-05 lr_backbone=4.824768e-05 elapsed=27:15:44 avg_step=3.66s recent_step=5.01s eta=23:38:17 +2026-09-18 20:45:11 | INFO | train rank=0 | step=26790/50000 progress=53.6% loss=0.177657 lr=4.821463e-05 lr_backbone=4.821463e-05 elapsed=27:16:34 avg_step=3.67s recent_step=4.99s eta=23:37:52 +2026-09-18 20:45:59 | INFO | train rank=0 | step=26800/50000 progress=53.6% loss=0.108327 lr=4.818158e-05 lr_backbone=4.818158e-05 elapsed=27:17:23 avg_step=3.67s recent_step=4.86s eta=23:37:26 +2026-09-18 20:46:48 | INFO | train rank=0 | step=26810/50000 progress=53.6% loss=0.134440 lr=4.814854e-05 lr_backbone=4.814854e-05 elapsed=27:18:12 avg_step=3.67s recent_step=4.89s eta=23:37:00 +2026-09-18 20:47:30 | INFO | train rank=0 | step=26820/50000 progress=53.6% loss=0.088832 lr=4.811549e-05 lr_backbone=4.811549e-05 elapsed=27:18:53 avg_step=3.67s recent_step=4.15s eta=23:36:27 +2026-09-18 20:48:20 | INFO | train rank=0 | step=26830/50000 progress=53.7% loss=0.089006 lr=4.808245e-05 lr_backbone=4.808245e-05 elapsed=27:19:44 avg_step=3.67s recent_step=5.05s eta=23:36:03 +2026-09-18 20:49:07 | INFO | train rank=0 | step=26840/50000 progress=53.7% loss=0.071726 lr=4.804940e-05 lr_backbone=4.804940e-05 elapsed=27:20:30 avg_step=3.67s recent_step=4.69s eta=23:35:35 +2026-09-18 20:49:53 | INFO | train rank=0 | step=26850/50000 progress=53.7% loss=0.084543 lr=4.801636e-05 lr_backbone=4.801636e-05 elapsed=27:21:16 avg_step=3.67s recent_step=4.58s eta=23:35:06 +2026-09-18 20:50:43 | INFO | train rank=0 | step=26860/50000 progress=53.7% loss=0.092180 lr=4.798331e-05 lr_backbone=4.798331e-05 elapsed=27:22:07 avg_step=3.67s recent_step=5.03s eta=23:34:41 +2026-09-18 20:51:35 | INFO | train rank=0 | step=26870/50000 progress=53.7% loss=0.094893 lr=4.795027e-05 lr_backbone=4.795027e-05 elapsed=27:22:58 avg_step=3.67s recent_step=5.13s eta=23:34:17 +2026-09-18 20:52:22 | INFO | train rank=0 | step=26880/50000 progress=53.8% loss=0.084486 lr=4.791723e-05 lr_backbone=4.791723e-05 elapsed=27:23:45 avg_step=3.67s recent_step=4.74s eta=23:33:49 +2026-09-18 20:53:14 | INFO | train rank=0 | step=26890/50000 progress=53.8% loss=0.082377 lr=4.788419e-05 lr_backbone=4.788419e-05 elapsed=27:24:38 avg_step=3.67s recent_step=5.23s eta=23:33:26 +2026-09-18 20:54:03 | INFO | train rank=0 | step=26900/50000 progress=53.8% loss=0.077080 lr=4.785115e-05 lr_backbone=4.785115e-05 elapsed=27:25:26 avg_step=3.67s recent_step=4.88s eta=23:33:00 +2026-09-18 20:54:49 | INFO | train rank=0 | step=26910/50000 progress=53.8% loss=0.092884 lr=4.781811e-05 lr_backbone=4.781811e-05 elapsed=27:26:12 avg_step=3.67s recent_step=4.57s eta=23:32:31 +2026-09-18 20:55:38 | INFO | train rank=0 | step=26920/50000 progress=53.8% loss=0.133153 lr=4.778508e-05 lr_backbone=4.778508e-05 elapsed=27:27:02 avg_step=3.67s recent_step=4.96s eta=23:32:05 +2026-09-18 20:56:20 | INFO | train rank=0 | step=26930/50000 progress=53.9% loss=0.105142 lr=4.775204e-05 lr_backbone=4.775204e-05 elapsed=27:27:43 avg_step=3.67s recent_step=4.17s eta=23:31:33 +2026-09-18 20:57:08 | INFO | train rank=0 | step=26940/50000 progress=53.9% loss=0.106524 lr=4.771900e-05 lr_backbone=4.771900e-05 elapsed=27:28:31 avg_step=3.67s recent_step=4.77s eta=23:31:05 +2026-09-18 20:57:53 | INFO | train rank=0 | step=26950/50000 progress=53.9% loss=0.100284 lr=4.768597e-05 lr_backbone=4.768597e-05 elapsed=27:29:17 avg_step=3.67s recent_step=4.55s eta=23:30:36 +2026-09-18 20:58:43 | INFO | train rank=0 | step=26960/50000 progress=53.9% loss=0.108291 lr=4.765294e-05 lr_backbone=4.765294e-05 elapsed=27:30:06 avg_step=3.67s recent_step=4.94s eta=23:30:10 +2026-09-18 20:59:29 | INFO | train rank=0 | step=26970/50000 progress=53.9% loss=0.103895 lr=4.761990e-05 lr_backbone=4.761990e-05 elapsed=27:30:53 avg_step=3.67s recent_step=4.68s eta=23:29:42 +2026-09-18 21:00:15 | INFO | train rank=0 | step=26980/50000 progress=54.0% loss=0.093215 lr=4.758687e-05 lr_backbone=4.758687e-05 elapsed=27:31:39 avg_step=3.67s recent_step=4.60s eta=23:29:13 +2026-09-18 21:01:06 | INFO | train rank=0 | step=26990/50000 progress=54.0% loss=0.069953 lr=4.755384e-05 lr_backbone=4.755384e-05 elapsed=27:32:29 avg_step=3.67s recent_step=5.01s eta=23:28:48 +2026-09-18 21:01:50 | INFO | train rank=0 | step=27000/50000 progress=54.0% loss=0.097020 lr=4.752081e-05 lr_backbone=4.752081e-05 elapsed=27:33:13 avg_step=3.67s recent_step=4.46s eta=23:28:18 +2026-09-18 21:02:44 | INFO | train rank=0 | step=27010/50000 progress=54.0% loss=0.148911 lr=4.748778e-05 lr_backbone=4.748778e-05 elapsed=27:34:07 avg_step=3.67s recent_step=5.40s eta=23:27:56 +2026-09-18 21:03:34 | INFO | train rank=0 | step=27020/50000 progress=54.0% loss=0.064000 lr=4.745476e-05 lr_backbone=4.745476e-05 elapsed=27:34:57 avg_step=3.67s recent_step=4.96s eta=23:27:30 +2026-09-18 21:04:21 | INFO | train rank=0 | step=27030/50000 progress=54.1% loss=0.100850 lr=4.742173e-05 lr_backbone=4.742173e-05 elapsed=27:35:45 avg_step=3.68s recent_step=4.76s eta=23:27:03 +2026-09-18 21:05:04 | INFO | train rank=0 | step=27040/50000 progress=54.1% loss=0.073993 lr=4.738871e-05 lr_backbone=4.738871e-05 elapsed=27:36:28 avg_step=3.68s recent_step=4.30s eta=23:26:31 +2026-09-18 21:05:49 | INFO | train rank=0 | step=27050/50000 progress=54.1% loss=0.101069 lr=4.735568e-05 lr_backbone=4.735568e-05 elapsed=27:37:12 avg_step=3.68s recent_step=4.47s eta=23:26:01 +2026-09-18 21:06:35 | INFO | train rank=0 | step=27060/50000 progress=54.1% loss=0.116067 lr=4.732266e-05 lr_backbone=4.732266e-05 elapsed=27:37:58 avg_step=3.68s recent_step=4.57s eta=23:25:32 +2026-09-18 21:07:23 | INFO | train rank=0 | step=27070/50000 progress=54.1% loss=0.110056 lr=4.728964e-05 lr_backbone=4.728964e-05 elapsed=27:38:46 avg_step=3.68s recent_step=4.79s eta=23:25:05 +2026-09-18 21:08:08 | INFO | train rank=0 | step=27080/50000 progress=54.2% loss=0.115785 lr=4.725662e-05 lr_backbone=4.725662e-05 elapsed=27:39:31 avg_step=3.68s recent_step=4.51s eta=23:24:35 +2026-09-18 21:08:55 | INFO | train rank=0 | step=27090/50000 progress=54.2% loss=0.096426 lr=4.722360e-05 lr_backbone=4.722360e-05 elapsed=27:40:18 avg_step=3.68s recent_step=4.69s eta=23:24:07 +2026-09-18 21:09:39 | INFO | train rank=0 | step=27100/50000 progress=54.2% loss=0.096850 lr=4.719058e-05 lr_backbone=4.719058e-05 elapsed=27:41:02 avg_step=3.68s recent_step=4.39s eta=23:23:36 +2026-09-18 21:10:17 | INFO | train rank=0 | step=27110/50000 progress=54.2% loss=0.099155 lr=4.715757e-05 lr_backbone=4.715757e-05 elapsed=27:41:40 avg_step=3.68s recent_step=3.78s eta=23:23:00 +2026-09-18 21:10:55 | INFO | train rank=0 | step=27120/50000 progress=54.2% loss=0.095547 lr=4.712455e-05 lr_backbone=4.712455e-05 elapsed=27:42:18 avg_step=3.68s recent_step=3.82s eta=23:22:25 +2026-09-18 21:11:41 | INFO | train rank=0 | step=27130/50000 progress=54.3% loss=0.081515 lr=4.709154e-05 lr_backbone=4.709154e-05 elapsed=27:43:04 avg_step=3.68s recent_step=4.64s eta=23:21:56 +2026-09-18 21:12:54 | INFO | train rank=0 | step=27140/50000 progress=54.3% loss=0.113331 lr=4.705852e-05 lr_backbone=4.705852e-05 elapsed=27:44:18 avg_step=3.68s recent_step=7.31s eta=23:21:50 +2026-09-18 21:14:12 | INFO | train rank=0 | step=27150/50000 progress=54.3% loss=0.087857 lr=4.702551e-05 lr_backbone=4.702551e-05 elapsed=27:45:36 avg_step=3.68s recent_step=7.82s eta=23:21:48 +2026-09-18 21:15:28 | INFO | train rank=0 | step=27160/50000 progress=54.3% loss=0.086707 lr=4.699250e-05 lr_backbone=4.699250e-05 elapsed=27:46:51 avg_step=3.68s recent_step=7.51s eta=23:21:43 +2026-09-18 21:16:38 | INFO | train rank=0 | step=27170/50000 progress=54.3% loss=0.097051 lr=4.695949e-05 lr_backbone=4.695949e-05 elapsed=27:48:01 avg_step=3.68s recent_step=7.04s eta=23:21:35 +2026-09-18 21:17:57 | INFO | train rank=0 | step=27180/50000 progress=54.4% loss=0.080059 lr=4.692649e-05 lr_backbone=4.692649e-05 elapsed=27:49:21 avg_step=3.69s recent_step=7.93s eta=23:21:34 +2026-09-18 21:19:11 | INFO | train rank=0 | step=27190/50000 progress=54.4% loss=0.084994 lr=4.689348e-05 lr_backbone=4.689348e-05 elapsed=27:50:34 avg_step=3.69s recent_step=7.32s eta=23:21:27 +2026-09-18 21:20:23 | INFO | train rank=0 | step=27200/50000 progress=54.4% loss=0.092576 lr=4.686047e-05 lr_backbone=4.686047e-05 elapsed=27:51:47 avg_step=3.69s recent_step=7.27s eta=23:21:20 +2026-09-18 21:21:09 | INFO | train rank=0 | step=27210/50000 progress=54.4% loss=0.106038 lr=4.682747e-05 lr_backbone=4.682747e-05 elapsed=27:52:32 avg_step=3.69s recent_step=4.58s eta=23:20:51 +2026-09-18 21:27:28 | INFO | train rank=0 | step=27220/50000 progress=54.4% loss=0.094106 lr=4.679447e-05 lr_backbone=4.679447e-05 elapsed=27:58:51 avg_step=3.70s recent_step=37.91s eta=23:25:01 +2026-09-18 21:27:55 | INFO | train rank=0 | step=27230/50000 progress=54.5% loss=0.091195 lr=4.676147e-05 lr_backbone=4.676147e-05 elapsed=27:59:19 avg_step=3.70s recent_step=2.71s eta=23:24:15 +2026-09-18 21:28:29 | INFO | train rank=0 | step=27240/50000 progress=54.5% loss=0.128766 lr=4.672847e-05 lr_backbone=4.672847e-05 elapsed=27:59:52 avg_step=3.70s recent_step=3.35s eta=23:23:35 +2026-09-18 21:29:11 | INFO | train rank=0 | step=27250/50000 progress=54.5% loss=0.092781 lr=4.669547e-05 lr_backbone=4.669547e-05 elapsed=28:00:34 avg_step=3.70s recent_step=4.20s eta=23:23:03 +2026-09-18 21:29:53 | INFO | train rank=0 | step=27260/50000 progress=54.5% loss=0.103922 lr=4.666247e-05 lr_backbone=4.666247e-05 elapsed=28:01:16 avg_step=3.70s recent_step=4.21s eta=23:22:30 +2026-09-18 21:30:23 | INFO | train rank=0 | step=27270/50000 progress=54.5% loss=0.078702 lr=4.662948e-05 lr_backbone=4.662948e-05 elapsed=28:01:47 avg_step=3.70s recent_step=3.04s eta=23:21:47 +2026-09-18 21:30:54 | INFO | train rank=0 | step=27280/50000 progress=54.6% loss=0.085028 lr=4.659649e-05 lr_backbone=4.659649e-05 elapsed=28:02:18 avg_step=3.70s recent_step=3.11s eta=23:21:05 +2026-09-18 21:31:46 | INFO | train rank=0 | step=27290/50000 progress=54.6% loss=0.095324 lr=4.656349e-05 lr_backbone=4.656349e-05 elapsed=28:03:10 avg_step=3.70s recent_step=5.19s eta=23:20:41 +2026-09-18 21:32:17 | INFO | train rank=0 | step=27300/50000 progress=54.6% loss=0.091061 lr=4.653050e-05 lr_backbone=4.653050e-05 elapsed=28:03:40 avg_step=3.70s recent_step=3.06s eta=23:19:58 +2026-09-18 21:32:51 | INFO | train rank=0 | step=27310/50000 progress=54.6% loss=0.112211 lr=4.649751e-05 lr_backbone=4.649751e-05 elapsed=28:04:14 avg_step=3.70s recent_step=3.40s eta=23:19:19 +2026-09-18 21:33:18 | INFO | train rank=0 | step=27320/50000 progress=54.6% loss=0.090157 lr=4.646453e-05 lr_backbone=4.646453e-05 elapsed=28:04:42 avg_step=3.70s recent_step=2.75s eta=23:18:34 +2026-09-18 21:34:09 | INFO | train rank=0 | step=27330/50000 progress=54.7% loss=0.132893 lr=4.643154e-05 lr_backbone=4.643154e-05 elapsed=28:05:32 avg_step=3.70s recent_step=5.05s eta=23:18:08 +2026-09-18 21:34:44 | INFO | train rank=0 | step=27340/50000 progress=54.7% loss=0.085195 lr=4.639856e-05 lr_backbone=4.639856e-05 elapsed=28:06:08 avg_step=3.70s recent_step=3.53s eta=23:17:30 +2026-09-18 21:35:24 | INFO | train rank=0 | step=27350/50000 progress=54.7% loss=0.094715 lr=4.636557e-05 lr_backbone=4.636557e-05 elapsed=28:06:48 avg_step=3.70s recent_step=4.01s eta=23:16:55 +2026-09-18 21:35:49 | INFO | train rank=0 | step=27360/50000 progress=54.7% loss=0.089940 lr=4.633259e-05 lr_backbone=4.633259e-05 elapsed=28:07:12 avg_step=3.70s recent_step=2.48s eta=23:16:08 +2026-09-18 21:36:42 | INFO | train rank=0 | step=27370/50000 progress=54.7% loss=0.084468 lr=4.629961e-05 lr_backbone=4.629961e-05 elapsed=28:08:05 avg_step=3.70s recent_step=5.25s eta=23:15:44 +2026-09-18 21:37:14 | INFO | train rank=0 | step=27380/50000 progress=54.8% loss=0.135065 lr=4.626664e-05 lr_backbone=4.626664e-05 elapsed=28:08:37 avg_step=3.70s recent_step=3.24s eta=23:15:03 +2026-09-18 21:37:46 | INFO | train rank=0 | step=27390/50000 progress=54.8% loss=0.091693 lr=4.623366e-05 lr_backbone=4.623366e-05 elapsed=28:09:09 avg_step=3.70s recent_step=3.18s eta=23:14:22 +2026-09-18 21:38:19 | INFO | train rank=0 | step=27400/50000 progress=54.8% loss=0.088451 lr=4.620069e-05 lr_backbone=4.620069e-05 elapsed=28:09:42 avg_step=3.70s recent_step=3.33s eta=23:13:42 +2026-09-18 21:39:04 | INFO | train rank=0 | step=27410/50000 progress=54.8% loss=0.071800 lr=4.616771e-05 lr_backbone=4.616771e-05 elapsed=28:10:27 avg_step=3.70s recent_step=4.46s eta=23:13:11 +2026-09-18 21:39:35 | INFO | train rank=0 | step=27420/50000 progress=54.8% loss=0.084453 lr=4.613474e-05 lr_backbone=4.613474e-05 elapsed=28:10:58 avg_step=3.70s recent_step=3.13s eta=23:12:30 +2026-09-18 21:40:15 | INFO | train rank=0 | step=27430/50000 progress=54.9% loss=0.109788 lr=4.610177e-05 lr_backbone=4.610177e-05 elapsed=28:11:38 avg_step=3.70s recent_step=3.98s eta=23:11:55 +2026-09-18 21:40:51 | INFO | train rank=0 | step=27440/50000 progress=54.9% loss=0.104821 lr=4.606880e-05 lr_backbone=4.606880e-05 elapsed=28:12:15 avg_step=3.70s recent_step=3.64s eta=23:11:17 +2026-09-18 21:41:30 | INFO | train rank=0 | step=27450/50000 progress=54.9% loss=0.105183 lr=4.603584e-05 lr_backbone=4.603584e-05 elapsed=28:12:54 avg_step=3.70s recent_step=3.91s eta=23:10:42 +2026-09-18 21:42:05 | INFO | train rank=0 | step=27460/50000 progress=54.9% loss=0.083105 lr=4.600287e-05 lr_backbone=4.600287e-05 elapsed=28:13:28 avg_step=3.70s recent_step=3.43s eta=23:10:03 +2026-09-18 21:42:41 | INFO | train rank=0 | step=27470/50000 progress=54.9% loss=0.082814 lr=4.596991e-05 lr_backbone=4.596991e-05 elapsed=28:14:04 avg_step=3.70s recent_step=3.61s eta=23:09:25 +2026-09-18 21:43:15 | INFO | train rank=0 | step=27480/50000 progress=55.0% loss=0.090235 lr=4.593695e-05 lr_backbone=4.593695e-05 elapsed=28:14:38 avg_step=3.70s recent_step=3.40s eta=23:08:46 +2026-09-18 21:43:55 | INFO | train rank=0 | step=27490/50000 progress=55.0% loss=0.079105 lr=4.590399e-05 lr_backbone=4.590399e-05 elapsed=28:15:19 avg_step=3.70s recent_step=4.05s eta=23:08:11 +2026-09-18 21:44:41 | INFO | train rank=0 | step=27500/50000 progress=55.0% loss=0.106631 lr=4.587103e-05 lr_backbone=4.587103e-05 elapsed=28:16:04 avg_step=3.70s recent_step=4.56s eta=23:07:41 +2026-09-18 21:45:12 | INFO | train rank=0 | step=27510/50000 progress=55.0% loss=0.104725 lr=4.583808e-05 lr_backbone=4.583808e-05 elapsed=28:16:36 avg_step=3.70s recent_step=3.14s eta=23:07:00 +2026-09-18 21:45:47 | INFO | train rank=0 | step=27520/50000 progress=55.0% loss=0.092317 lr=4.580512e-05 lr_backbone=4.580512e-05 elapsed=28:17:10 avg_step=3.70s recent_step=3.43s eta=23:06:21 +2026-09-18 21:46:28 | INFO | train rank=0 | step=27530/50000 progress=55.1% loss=0.068941 lr=4.577217e-05 lr_backbone=4.577217e-05 elapsed=28:17:52 avg_step=3.70s recent_step=4.17s eta=23:05:48 +2026-09-18 21:46:58 | INFO | train rank=0 | step=27540/50000 progress=55.1% loss=0.106008 lr=4.573922e-05 lr_backbone=4.573922e-05 elapsed=28:18:21 avg_step=3.70s recent_step=2.98s eta=23:05:05 +2026-09-18 21:47:38 | INFO | train rank=0 | step=27550/50000 progress=55.1% loss=0.091623 lr=4.570627e-05 lr_backbone=4.570627e-05 elapsed=28:19:01 avg_step=3.70s recent_step=4.00s eta=23:04:30 +2026-09-18 21:48:14 | INFO | train rank=0 | step=27560/50000 progress=55.1% loss=0.081079 lr=4.567333e-05 lr_backbone=4.567333e-05 elapsed=28:19:37 avg_step=3.70s recent_step=3.60s eta=23:03:52 +2026-09-18 21:48:48 | INFO | train rank=0 | step=27570/50000 progress=55.1% loss=0.094404 lr=4.564038e-05 lr_backbone=4.564038e-05 elapsed=28:20:11 avg_step=3.70s recent_step=3.34s eta=23:03:12 +2026-09-18 21:49:23 | INFO | train rank=0 | step=27580/50000 progress=55.2% loss=0.104587 lr=4.560744e-05 lr_backbone=4.560744e-05 elapsed=28:20:46 avg_step=3.70s recent_step=3.55s eta=23:02:34 +2026-09-18 21:50:04 | INFO | train rank=0 | step=27590/50000 progress=55.2% loss=0.095614 lr=4.557450e-05 lr_backbone=4.557450e-05 elapsed=28:21:28 avg_step=3.70s recent_step=4.13s eta=23:02:01 +2026-09-18 21:50:44 | INFO | train rank=0 | step=27600/50000 progress=55.2% loss=0.083145 lr=4.554156e-05 lr_backbone=4.554156e-05 elapsed=28:22:07 avg_step=3.70s recent_step=3.94s eta=23:01:26 +2026-09-18 21:51:20 | INFO | train rank=0 | step=27610/50000 progress=55.2% loss=0.097768 lr=4.550862e-05 lr_backbone=4.550862e-05 elapsed=28:22:43 avg_step=3.70s recent_step=3.60s eta=23:00:48 +2026-09-18 21:51:49 | INFO | train rank=0 | step=27620/50000 progress=55.2% loss=0.080607 lr=4.547569e-05 lr_backbone=4.547569e-05 elapsed=28:23:12 avg_step=3.70s recent_step=2.90s eta=23:00:04 +2026-09-18 21:52:41 | INFO | train rank=0 | step=27630/50000 progress=55.3% loss=0.076229 lr=4.544276e-05 lr_backbone=4.544276e-05 elapsed=28:24:04 avg_step=3.70s recent_step=5.19s eta=22:59:39 +2026-09-18 21:53:18 | INFO | train rank=0 | step=27640/50000 progress=55.3% loss=0.100230 lr=4.540983e-05 lr_backbone=4.540983e-05 elapsed=28:24:41 avg_step=3.70s recent_step=3.70s eta=22:59:02 +2026-09-18 21:53:51 | INFO | train rank=0 | step=27650/50000 progress=55.3% loss=0.088151 lr=4.537690e-05 lr_backbone=4.537690e-05 elapsed=28:25:14 avg_step=3.70s recent_step=3.29s eta=22:58:22 +2026-09-18 21:54:23 | INFO | train rank=0 | step=27660/50000 progress=55.3% loss=0.132237 lr=4.534397e-05 lr_backbone=4.534397e-05 elapsed=28:25:46 avg_step=3.70s recent_step=3.20s eta=22:57:41 +2026-09-18 21:55:00 | INFO | train rank=0 | step=27670/50000 progress=55.3% loss=0.098249 lr=4.531105e-05 lr_backbone=4.531105e-05 elapsed=28:26:23 avg_step=3.70s recent_step=3.74s eta=22:57:04 +2026-09-18 21:55:28 | INFO | train rank=0 | step=27680/50000 progress=55.4% loss=0.084977 lr=4.527812e-05 lr_backbone=4.527812e-05 elapsed=28:26:52 avg_step=3.70s recent_step=2.83s eta=22:56:20 +2026-09-18 21:56:19 | INFO | train rank=0 | step=27690/50000 progress=55.4% loss=0.075328 lr=4.524520e-05 lr_backbone=4.524520e-05 elapsed=28:27:43 avg_step=3.70s recent_step=5.10s eta=22:55:55 +2026-09-18 21:56:52 | INFO | train rank=0 | step=27700/50000 progress=55.4% loss=0.074931 lr=4.521228e-05 lr_backbone=4.521228e-05 elapsed=28:28:15 avg_step=3.70s recent_step=3.27s eta=22:55:14 +2026-09-18 21:57:23 | INFO | train rank=0 | step=27710/50000 progress=55.4% loss=0.082472 lr=4.517937e-05 lr_backbone=4.517937e-05 elapsed=28:28:46 avg_step=3.70s recent_step=3.08s eta=22:54:32 +2026-09-18 21:58:02 | INFO | train rank=0 | step=27720/50000 progress=55.4% loss=0.085749 lr=4.514645e-05 lr_backbone=4.514645e-05 elapsed=28:29:25 avg_step=3.70s recent_step=3.92s eta=22:53:57 +2026-09-18 21:58:56 | INFO | train rank=0 | step=27730/50000 progress=55.5% loss=0.087566 lr=4.511354e-05 lr_backbone=4.511354e-05 elapsed=28:30:19 avg_step=3.70s recent_step=5.36s eta=22:53:33 +2026-09-18 21:59:25 | INFO | train rank=0 | step=27740/50000 progress=55.5% loss=0.097088 lr=4.508063e-05 lr_backbone=4.508063e-05 elapsed=28:30:48 avg_step=3.70s recent_step=2.94s eta=22:52:50 +2026-09-18 21:59:56 | INFO | train rank=0 | step=27750/50000 progress=55.5% loss=0.094536 lr=4.504772e-05 lr_backbone=4.504772e-05 elapsed=28:31:19 avg_step=3.70s recent_step=3.11s eta=22:52:08 +2026-09-18 22:00:38 | INFO | train rank=0 | step=27760/50000 progress=55.5% loss=0.093058 lr=4.501482e-05 lr_backbone=4.501482e-05 elapsed=28:32:01 avg_step=3.70s recent_step=4.17s eta=22:51:35 +2026-09-18 22:01:22 | INFO | train rank=0 | step=27770/50000 progress=55.5% loss=0.096085 lr=4.498191e-05 lr_backbone=4.498191e-05 elapsed=28:32:46 avg_step=3.70s recent_step=4.46s eta=22:51:04 +2026-09-18 22:01:56 | INFO | train rank=0 | step=27780/50000 progress=55.6% loss=0.088637 lr=4.494901e-05 lr_backbone=4.494901e-05 elapsed=28:33:19 avg_step=3.70s recent_step=3.35s eta=22:50:24 +2026-09-18 22:02:26 | INFO | train rank=0 | step=27790/50000 progress=55.6% loss=0.116497 lr=4.491611e-05 lr_backbone=4.491611e-05 elapsed=28:33:49 avg_step=3.70s recent_step=3.00s eta=22:49:42 +2026-09-18 22:03:01 | INFO | train rank=0 | step=27800/50000 progress=55.6% loss=0.161275 lr=4.488322e-05 lr_backbone=4.488322e-05 elapsed=28:34:24 avg_step=3.70s recent_step=3.50s eta=22:49:03 +2026-09-18 22:03:48 | INFO | train rank=0 | step=27810/50000 progress=55.6% loss=0.091147 lr=4.485032e-05 lr_backbone=4.485032e-05 elapsed=28:35:12 avg_step=3.70s recent_step=4.76s eta=22:48:35 +2026-09-18 22:04:17 | INFO | train rank=0 | step=27820/50000 progress=55.6% loss=0.082496 lr=4.481743e-05 lr_backbone=4.481743e-05 elapsed=28:35:40 avg_step=3.70s recent_step=2.82s eta=22:47:51 +2026-09-18 22:04:51 | INFO | train rank=0 | step=27830/50000 progress=55.7% loss=0.098944 lr=4.478454e-05 lr_backbone=4.478454e-05 elapsed=28:36:14 avg_step=3.70s recent_step=3.43s eta=22:47:11 +2026-09-18 22:05:27 | INFO | train rank=0 | step=27840/50000 progress=55.7% loss=0.074043 lr=4.475165e-05 lr_backbone=4.475165e-05 elapsed=28:36:50 avg_step=3.70s recent_step=3.60s eta=22:46:34 +2026-09-18 22:06:15 | INFO | train rank=0 | step=27850/50000 progress=55.7% loss=0.077333 lr=4.471877e-05 lr_backbone=4.471877e-05 elapsed=28:37:38 avg_step=3.70s recent_step=4.81s eta=22:46:06 +2026-09-18 22:06:39 | INFO | train rank=0 | step=27860/50000 progress=55.7% loss=0.133842 lr=4.468588e-05 lr_backbone=4.468588e-05 elapsed=28:38:02 avg_step=3.70s recent_step=2.40s eta=22:45:18 +2026-09-18 22:07:08 | INFO | train rank=0 | step=27870/50000 progress=55.7% loss=0.111679 lr=4.465300e-05 lr_backbone=4.465300e-05 elapsed=28:38:31 avg_step=3.70s recent_step=2.90s eta=22:44:35 +2026-09-18 22:07:49 | INFO | train rank=0 | step=27880/50000 progress=55.8% loss=0.092962 lr=4.462012e-05 lr_backbone=4.462012e-05 elapsed=28:39:12 avg_step=3.70s recent_step=4.05s eta=22:44:01 +2026-09-18 22:08:39 | INFO | train rank=0 | step=27890/50000 progress=55.8% loss=0.093118 lr=4.458725e-05 lr_backbone=4.458725e-05 elapsed=28:40:03 avg_step=3.70s recent_step=5.07s eta=22:43:34 +2026-09-18 22:09:08 | INFO | train rank=0 | step=27900/50000 progress=55.8% loss=0.107308 lr=4.455437e-05 lr_backbone=4.455437e-05 elapsed=28:40:32 avg_step=3.70s recent_step=2.91s eta=22:42:51 +2026-09-18 22:09:37 | INFO | train rank=0 | step=27910/50000 progress=55.8% loss=0.087244 lr=4.452150e-05 lr_backbone=4.452150e-05 elapsed=28:41:01 avg_step=3.70s recent_step=2.91s eta=22:42:08 +2026-09-18 22:10:12 | INFO | train rank=0 | step=27920/50000 progress=55.8% loss=0.081457 lr=4.448863e-05 lr_backbone=4.448863e-05 elapsed=28:41:35 avg_step=3.70s recent_step=3.44s eta=22:41:29 +2026-09-18 22:11:11 | INFO | train rank=0 | step=27930/50000 progress=55.9% loss=0.088269 lr=4.445577e-05 lr_backbone=4.445577e-05 elapsed=28:42:34 avg_step=3.70s recent_step=5.90s eta=22:41:09 +2026-09-18 22:11:46 | INFO | train rank=0 | step=27940/50000 progress=55.9% loss=0.093288 lr=4.442290e-05 lr_backbone=4.442290e-05 elapsed=28:43:10 avg_step=3.70s recent_step=3.54s eta=22:40:31 +2026-09-18 22:12:14 | INFO | train rank=0 | step=27950/50000 progress=55.9% loss=0.150860 lr=4.439004e-05 lr_backbone=4.439004e-05 elapsed=28:43:37 avg_step=3.70s recent_step=2.73s eta=22:39:46 +2026-09-18 22:12:43 | INFO | train rank=0 | step=27960/50000 progress=55.9% loss=0.138151 lr=4.435718e-05 lr_backbone=4.435718e-05 elapsed=28:44:07 avg_step=3.70s recent_step=2.98s eta=22:39:04 +2026-09-18 22:13:35 | INFO | train rank=0 | step=27970/50000 progress=55.9% loss=0.100439 lr=4.432432e-05 lr_backbone=4.432432e-05 elapsed=28:44:58 avg_step=3.70s recent_step=5.13s eta=22:38:38 +2026-09-18 22:14:09 | INFO | train rank=0 | step=27980/50000 progress=56.0% loss=0.082241 lr=4.429147e-05 lr_backbone=4.429147e-05 elapsed=28:45:32 avg_step=3.70s recent_step=3.38s eta=22:37:59 +2026-09-18 22:14:38 | INFO | train rank=0 | step=27990/50000 progress=56.0% loss=0.108814 lr=4.425862e-05 lr_backbone=4.425862e-05 elapsed=28:46:02 avg_step=3.70s recent_step=2.98s eta=22:37:16 +2026-09-18 22:15:13 | INFO | train rank=0 | step=28000/50000 progress=56.0% loss=0.100196 lr=4.422577e-05 lr_backbone=4.422577e-05 elapsed=28:46:37 avg_step=3.70s recent_step=3.49s eta=22:36:37 +2026-09-18 22:16:06 | INFO | train rank=0 | step=28010/50000 progress=56.0% loss=0.073517 lr=4.419292e-05 lr_backbone=4.419292e-05 elapsed=28:47:29 avg_step=3.70s recent_step=5.24s eta=22:36:12 +2026-09-18 22:16:38 | INFO | train rank=0 | step=28020/50000 progress=56.0% loss=0.084756 lr=4.416008e-05 lr_backbone=4.416008e-05 elapsed=28:48:02 avg_step=3.70s recent_step=3.28s eta=22:35:32 +2026-09-18 22:17:14 | INFO | train rank=0 | step=28030/50000 progress=56.1% loss=0.085776 lr=4.412724e-05 lr_backbone=4.412724e-05 elapsed=28:48:37 avg_step=3.70s recent_step=3.57s eta=22:34:54 +2026-09-18 22:17:46 | INFO | train rank=0 | step=28040/50000 progress=56.1% loss=0.099396 lr=4.409440e-05 lr_backbone=4.409440e-05 elapsed=28:49:09 avg_step=3.70s recent_step=3.20s eta=22:34:13 +2026-09-18 22:18:31 | INFO | train rank=0 | step=28050/50000 progress=56.1% loss=0.085997 lr=4.406156e-05 lr_backbone=4.406156e-05 elapsed=28:49:55 avg_step=3.70s recent_step=4.53s eta=22:33:43 +2026-09-18 22:19:00 | INFO | train rank=0 | step=28060/50000 progress=56.1% loss=0.100214 lr=4.402873e-05 lr_backbone=4.402873e-05 elapsed=28:50:23 avg_step=3.70s recent_step=2.82s eta=22:32:59 +2026-09-18 22:19:30 | INFO | train rank=0 | step=28070/50000 progress=56.1% loss=0.070009 lr=4.399589e-05 lr_backbone=4.399589e-05 elapsed=28:50:53 avg_step=3.70s recent_step=3.02s eta=22:32:16 +2026-09-18 22:20:03 | INFO | train rank=0 | step=28080/50000 progress=56.2% loss=0.114866 lr=4.396307e-05 lr_backbone=4.396307e-05 elapsed=28:51:26 avg_step=3.70s recent_step=3.33s eta=22:31:36 +2026-09-18 22:20:48 | INFO | train rank=0 | step=28090/50000 progress=56.2% loss=0.097196 lr=4.393024e-05 lr_backbone=4.393024e-05 elapsed=28:52:11 avg_step=3.70s recent_step=4.44s eta=22:31:05 +2026-09-18 22:21:27 | INFO | train rank=0 | step=28100/50000 progress=56.2% loss=0.080629 lr=4.389742e-05 lr_backbone=4.389742e-05 elapsed=28:52:50 avg_step=3.70s recent_step=3.90s eta=22:30:30 +2026-09-18 22:22:01 | INFO | train rank=0 | step=28110/50000 progress=56.2% loss=0.093726 lr=4.386459e-05 lr_backbone=4.386459e-05 elapsed=28:53:24 avg_step=3.70s recent_step=3.43s eta=22:29:51 +2026-09-18 22:22:40 | INFO | train rank=0 | step=28120/50000 progress=56.2% loss=0.086820 lr=4.383178e-05 lr_backbone=4.383178e-05 elapsed=28:54:03 avg_step=3.70s recent_step=3.93s eta=22:29:15 +2026-09-18 22:23:37 | INFO | train rank=0 | step=28130/50000 progress=56.3% loss=0.084802 lr=4.379896e-05 lr_backbone=4.379896e-05 elapsed=28:55:00 avg_step=3.70s recent_step=5.66s eta=22:28:54 +2026-09-18 22:24:04 | INFO | train rank=0 | step=28140/50000 progress=56.3% loss=0.088862 lr=4.376615e-05 lr_backbone=4.376615e-05 elapsed=28:55:27 avg_step=3.70s recent_step=2.70s eta=22:28:09 +2026-09-18 22:24:39 | INFO | train rank=0 | step=28150/50000 progress=56.3% loss=0.122485 lr=4.373334e-05 lr_backbone=4.373334e-05 elapsed=28:56:02 avg_step=3.70s recent_step=3.53s eta=22:27:31 +2026-09-18 22:25:14 | INFO | train rank=0 | step=28160/50000 progress=56.3% loss=0.078260 lr=4.370053e-05 lr_backbone=4.370053e-05 elapsed=28:56:37 avg_step=3.70s recent_step=3.50s eta=22:26:52 +2026-09-18 22:26:06 | INFO | train rank=0 | step=28170/50000 progress=56.3% loss=0.099050 lr=4.366773e-05 lr_backbone=4.366773e-05 elapsed=28:57:29 avg_step=3.70s recent_step=5.15s eta=22:26:26 +2026-09-18 22:26:37 | INFO | train rank=0 | step=28180/50000 progress=56.4% loss=0.084322 lr=4.363492e-05 lr_backbone=4.363492e-05 elapsed=28:58:00 avg_step=3.70s recent_step=3.13s eta=22:25:45 +2026-09-18 22:27:02 | INFO | train rank=0 | step=28190/50000 progress=56.4% loss=0.101269 lr=4.360213e-05 lr_backbone=4.360213e-05 elapsed=28:58:25 avg_step=3.70s recent_step=2.47s eta=22:24:58 +2026-09-18 22:27:30 | INFO | train rank=0 | step=28200/50000 progress=56.4% loss=0.102578 lr=4.356933e-05 lr_backbone=4.356933e-05 elapsed=28:58:54 avg_step=3.70s recent_step=2.88s eta=22:24:15 +2026-09-18 22:28:23 | INFO | train rank=0 | step=28210/50000 progress=56.4% loss=0.069663 lr=4.353654e-05 lr_backbone=4.353654e-05 elapsed=28:59:47 avg_step=3.70s recent_step=5.28s eta=22:23:50 +2026-09-18 22:28:56 | INFO | train rank=0 | step=28220/50000 progress=56.4% loss=0.082650 lr=4.350375e-05 lr_backbone=4.350375e-05 elapsed=29:00:19 avg_step=3.70s recent_step=3.24s eta=22:23:10 +2026-09-18 22:29:27 | INFO | train rank=0 | step=28230/50000 progress=56.5% loss=0.097853 lr=4.347096e-05 lr_backbone=4.347096e-05 elapsed=29:00:50 avg_step=3.70s recent_step=3.16s eta=22:22:28 +2026-09-18 22:29:54 | INFO | train rank=0 | step=28240/50000 progress=56.5% loss=0.145811 lr=4.343817e-05 lr_backbone=4.343817e-05 elapsed=29:01:17 avg_step=3.70s recent_step=2.69s eta=22:21:44 +2026-09-18 22:30:41 | INFO | train rank=0 | step=28250/50000 progress=56.5% loss=0.091786 lr=4.340539e-05 lr_backbone=4.340539e-05 elapsed=29:02:04 avg_step=3.70s recent_step=4.70s eta=22:21:14 +2026-09-18 22:31:19 | INFO | train rank=0 | step=28260/50000 progress=56.5% loss=0.068882 lr=4.337261e-05 lr_backbone=4.337261e-05 elapsed=29:02:42 avg_step=3.70s recent_step=3.74s eta=22:20:38 +2026-09-18 22:31:49 | INFO | train rank=0 | step=28270/50000 progress=56.5% loss=0.087906 lr=4.333984e-05 lr_backbone=4.333984e-05 elapsed=29:03:13 avg_step=3.70s recent_step=3.08s eta=22:19:56 +2026-09-18 22:32:20 | INFO | train rank=0 | step=28280/50000 progress=56.6% loss=0.072757 lr=4.330706e-05 lr_backbone=4.330706e-05 elapsed=29:03:44 avg_step=3.70s recent_step=3.10s eta=22:19:14 +2026-09-18 22:33:15 | INFO | train rank=0 | step=28290/50000 progress=56.6% loss=0.077752 lr=4.327429e-05 lr_backbone=4.327429e-05 elapsed=29:04:38 avg_step=3.70s recent_step=5.41s eta=22:18:51 +2026-09-18 22:33:45 | INFO | train rank=0 | step=28300/50000 progress=56.6% loss=0.097761 lr=4.324153e-05 lr_backbone=4.324153e-05 elapsed=29:05:08 avg_step=3.70s recent_step=3.05s eta=22:18:09 +2026-09-18 22:34:18 | INFO | train rank=0 | step=28310/50000 progress=56.6% loss=0.073506 lr=4.320876e-05 lr_backbone=4.320876e-05 elapsed=29:05:41 avg_step=3.70s recent_step=3.25s eta=22:17:28 +2026-09-18 22:34:49 | INFO | train rank=0 | step=28320/50000 progress=56.6% loss=0.089451 lr=4.317600e-05 lr_backbone=4.317600e-05 elapsed=29:06:13 avg_step=3.70s recent_step=3.18s eta=22:16:47 +2026-09-18 22:35:42 | INFO | train rank=0 | step=28330/50000 progress=56.7% loss=0.087136 lr=4.314324e-05 lr_backbone=4.314324e-05 elapsed=29:07:05 avg_step=3.70s recent_step=5.23s eta=22:16:22 +2026-09-18 22:36:15 | INFO | train rank=0 | step=28340/50000 progress=56.7% loss=0.095198 lr=4.311049e-05 lr_backbone=4.311049e-05 elapsed=29:07:38 avg_step=3.70s recent_step=3.31s eta=22:15:42 +2026-09-18 22:36:40 | INFO | train rank=0 | step=28350/50000 progress=56.7% loss=0.075375 lr=4.307773e-05 lr_backbone=4.307773e-05 elapsed=29:08:03 avg_step=3.70s recent_step=2.51s eta=22:14:56 +2026-09-18 22:37:12 | INFO | train rank=0 | step=28360/50000 progress=56.7% loss=0.103092 lr=4.304498e-05 lr_backbone=4.304498e-05 elapsed=29:08:36 avg_step=3.70s recent_step=3.24s eta=22:14:15 +2026-09-18 22:38:10 | INFO | train rank=0 | step=28370/50000 progress=56.7% loss=0.082612 lr=4.301224e-05 lr_backbone=4.301224e-05 elapsed=29:09:34 avg_step=3.70s recent_step=5.80s eta=22:13:54 +2026-09-18 22:38:46 | INFO | train rank=0 | step=28380/50000 progress=56.8% loss=0.141949 lr=4.297949e-05 lr_backbone=4.297949e-05 elapsed=29:10:09 avg_step=3.70s recent_step=3.58s eta=22:13:16 +2026-09-18 22:39:21 | INFO | train rank=0 | step=28390/50000 progress=56.8% loss=0.124907 lr=4.294675e-05 lr_backbone=4.294675e-05 elapsed=29:10:44 avg_step=3.70s recent_step=3.45s eta=22:12:38 +2026-09-18 22:39:51 | INFO | train rank=0 | step=28400/50000 progress=56.8% loss=0.088859 lr=4.291402e-05 lr_backbone=4.291402e-05 elapsed=29:11:14 avg_step=3.70s recent_step=3.01s eta=22:11:55 +2026-09-18 22:40:44 | INFO | train rank=0 | step=28410/50000 progress=56.8% loss=0.100092 lr=4.288128e-05 lr_backbone=4.288128e-05 elapsed=29:12:07 avg_step=3.70s recent_step=5.29s eta=22:11:30 +2026-09-18 22:41:18 | INFO | train rank=0 | step=28420/50000 progress=56.8% loss=0.086887 lr=4.284855e-05 lr_backbone=4.284855e-05 elapsed=29:12:41 avg_step=3.70s recent_step=3.42s eta=22:10:51 +2026-09-18 22:41:50 | INFO | train rank=0 | step=28430/50000 progress=56.9% loss=0.083654 lr=4.281582e-05 lr_backbone=4.281582e-05 elapsed=29:13:13 avg_step=3.70s recent_step=3.23s eta=22:10:11 +2026-09-18 22:42:26 | INFO | train rank=0 | step=28440/50000 progress=56.9% loss=0.091309 lr=4.278310e-05 lr_backbone=4.278310e-05 elapsed=29:13:49 avg_step=3.70s recent_step=3.62s eta=22:09:33 +2026-09-18 22:43:14 | INFO | train rank=0 | step=28450/50000 progress=56.9% loss=0.076856 lr=4.275038e-05 lr_backbone=4.275038e-05 elapsed=29:14:37 avg_step=3.70s recent_step=4.79s eta=22:09:04 +2026-09-18 22:43:49 | INFO | train rank=0 | step=28460/50000 progress=56.9% loss=0.107201 lr=4.271766e-05 lr_backbone=4.271766e-05 elapsed=29:15:13 avg_step=3.70s recent_step=3.54s eta=22:08:26 +2026-09-18 22:44:10 | INFO | train rank=0 | step=28470/50000 progress=56.9% loss=0.089213 lr=4.268494e-05 lr_backbone=4.268494e-05 elapsed=29:15:33 avg_step=3.70s recent_step=2.04s eta=22:07:36 +2026-09-18 22:44:42 | INFO | train rank=0 | step=28480/50000 progress=57.0% loss=0.092108 lr=4.265223e-05 lr_backbone=4.265223e-05 elapsed=29:16:05 avg_step=3.70s recent_step=3.21s eta=22:06:56 +2026-09-18 22:45:36 | INFO | train rank=0 | step=28490/50000 progress=57.0% loss=0.111662 lr=4.261952e-05 lr_backbone=4.261952e-05 elapsed=29:16:59 avg_step=3.70s recent_step=5.39s eta=22:06:32 +2026-09-18 22:46:10 | INFO | train rank=0 | step=28500/50000 progress=57.0% loss=0.128585 lr=4.258682e-05 lr_backbone=4.258682e-05 elapsed=29:17:33 avg_step=3.70s recent_step=3.43s eta=22:05:52 +2026-09-18 22:46:43 | INFO | train rank=0 | step=28510/50000 progress=57.0% loss=0.073071 lr=4.255412e-05 lr_backbone=4.255412e-05 elapsed=29:18:06 avg_step=3.70s recent_step=3.30s eta=22:05:12 +2026-09-18 22:47:08 | INFO | train rank=0 | step=28520/50000 progress=57.0% loss=0.094159 lr=4.252142e-05 lr_backbone=4.252142e-05 elapsed=29:18:31 avg_step=3.70s recent_step=2.45s eta=22:04:26 +2026-09-18 22:48:07 | INFO | train rank=0 | step=28530/50000 progress=57.1% loss=0.072616 lr=4.248872e-05 lr_backbone=4.248872e-05 elapsed=29:19:30 avg_step=3.70s recent_step=5.90s eta=22:04:06 +2026-09-18 22:48:32 | INFO | train rank=0 | step=28540/50000 progress=57.1% loss=0.072315 lr=4.245603e-05 lr_backbone=4.245603e-05 elapsed=29:19:55 avg_step=3.70s recent_step=2.54s eta=22:03:20 +2026-09-18 22:49:08 | INFO | train rank=0 | step=28550/50000 progress=57.1% loss=0.084960 lr=4.242334e-05 lr_backbone=4.242334e-05 elapsed=29:20:31 avg_step=3.70s recent_step=3.59s eta=22:02:42 +2026-09-18 22:49:41 | INFO | train rank=0 | step=28560/50000 progress=57.1% loss=0.087141 lr=4.239065e-05 lr_backbone=4.239065e-05 elapsed=29:21:04 avg_step=3.70s recent_step=3.30s eta=22:02:02 +2026-09-18 22:50:35 | INFO | train rank=0 | step=28570/50000 progress=57.1% loss=0.113218 lr=4.235797e-05 lr_backbone=4.235797e-05 elapsed=29:21:59 avg_step=3.70s recent_step=5.46s eta=22:01:38 +2026-09-18 22:51:10 | INFO | train rank=0 | step=28580/50000 progress=57.2% loss=0.139071 lr=4.232529e-05 lr_backbone=4.232529e-05 elapsed=29:22:33 avg_step=3.70s recent_step=3.41s eta=22:00:59 +2026-09-18 22:51:42 | INFO | train rank=0 | step=28590/50000 progress=57.2% loss=0.085722 lr=4.229262e-05 lr_backbone=4.229262e-05 elapsed=29:23:05 avg_step=3.70s recent_step=3.23s eta=22:00:19 +2026-09-18 22:52:07 | INFO | train rank=0 | step=28600/50000 progress=57.2% loss=0.083292 lr=4.225994e-05 lr_backbone=4.225994e-05 elapsed=29:23:30 avg_step=3.70s recent_step=2.47s eta=21:59:32 +2026-09-18 22:53:00 | INFO | train rank=0 | step=28610/50000 progress=57.2% loss=0.081721 lr=4.222727e-05 lr_backbone=4.222727e-05 elapsed=29:24:23 avg_step=3.70s recent_step=5.30s eta=21:59:07 +2026-09-18 22:53:30 | INFO | train rank=0 | step=28620/50000 progress=57.2% loss=0.086444 lr=4.219461e-05 lr_backbone=4.219461e-05 elapsed=29:24:54 avg_step=3.70s recent_step=3.06s eta=21:58:26 +2026-09-18 22:54:04 | INFO | train rank=0 | step=28630/50000 progress=57.3% loss=0.090849 lr=4.216195e-05 lr_backbone=4.216195e-05 elapsed=29:25:27 avg_step=3.70s recent_step=3.38s eta=21:57:46 +2026-09-18 22:54:41 | INFO | train rank=0 | step=28640/50000 progress=57.3% loss=0.087086 lr=4.212929e-05 lr_backbone=4.212929e-05 elapsed=29:26:04 avg_step=3.70s recent_step=3.67s eta=21:57:09 +2026-09-18 22:55:28 | INFO | train rank=0 | step=28650/50000 progress=57.3% loss=0.092614 lr=4.209663e-05 lr_backbone=4.209663e-05 elapsed=29:26:52 avg_step=3.70s recent_step=4.76s eta=21:56:40 +2026-09-18 22:56:00 | INFO | train rank=0 | step=28660/50000 progress=57.3% loss=0.094223 lr=4.206398e-05 lr_backbone=4.206398e-05 elapsed=29:27:23 avg_step=3.70s recent_step=3.14s eta=21:55:59 +2026-09-18 22:56:35 | INFO | train rank=0 | step=28670/50000 progress=57.3% loss=0.126841 lr=4.203133e-05 lr_backbone=4.203133e-05 elapsed=29:27:58 avg_step=3.70s recent_step=3.51s eta=21:55:20 +2026-09-18 22:57:05 | INFO | train rank=0 | step=28680/50000 progress=57.4% loss=0.074445 lr=4.199869e-05 lr_backbone=4.199869e-05 elapsed=29:28:28 avg_step=3.70s recent_step=3.03s eta=21:54:38 +2026-09-18 22:57:46 | INFO | train rank=0 | step=28690/50000 progress=57.4% loss=0.098465 lr=4.196605e-05 lr_backbone=4.196605e-05 elapsed=29:29:10 avg_step=3.70s recent_step=4.12s eta=21:54:04 +2026-09-18 22:58:26 | INFO | train rank=0 | step=28700/50000 progress=57.4% loss=0.079053 lr=4.193341e-05 lr_backbone=4.193341e-05 elapsed=29:29:49 avg_step=3.70s recent_step=3.94s eta=21:53:29 +2026-09-18 22:59:00 | INFO | train rank=0 | step=28710/50000 progress=57.4% loss=0.064402 lr=4.190077e-05 lr_backbone=4.190077e-05 elapsed=29:30:23 avg_step=3.70s recent_step=3.39s eta=21:52:50 +2026-09-18 22:59:28 | INFO | train rank=0 | step=28720/50000 progress=57.4% loss=0.100957 lr=4.186814e-05 lr_backbone=4.186814e-05 elapsed=29:30:52 avg_step=3.70s recent_step=2.88s eta=21:52:07 +2026-09-18 23:00:12 | INFO | train rank=0 | step=28730/50000 progress=57.5% loss=0.115238 lr=4.183551e-05 lr_backbone=4.183551e-05 elapsed=29:31:36 avg_step=3.70s recent_step=4.39s eta=21:51:35 +2026-09-18 23:00:42 | INFO | train rank=0 | step=28740/50000 progress=57.5% loss=0.062214 lr=4.180289e-05 lr_backbone=4.180289e-05 elapsed=29:32:05 avg_step=3.70s recent_step=2.98s eta=21:50:53 +2026-09-18 23:01:18 | INFO | train rank=0 | step=28750/50000 progress=57.5% loss=0.099035 lr=4.177027e-05 lr_backbone=4.177027e-05 elapsed=29:32:41 avg_step=3.70s recent_step=3.59s eta=21:50:15 +2026-09-18 23:01:50 | INFO | train rank=0 | step=28760/50000 progress=57.5% loss=0.113046 lr=4.173765e-05 lr_backbone=4.173765e-05 elapsed=29:33:13 avg_step=3.70s recent_step=3.21s eta=21:49:34 +2026-09-18 23:02:44 | INFO | train rank=0 | step=28770/50000 progress=57.5% loss=0.079326 lr=4.170504e-05 lr_backbone=4.170504e-05 elapsed=29:34:08 avg_step=3.70s recent_step=5.41s eta=21:49:10 +2026-09-18 23:03:16 | INFO | train rank=0 | step=28780/50000 progress=57.6% loss=0.065235 lr=4.167243e-05 lr_backbone=4.167243e-05 elapsed=29:34:40 avg_step=3.70s recent_step=3.22s eta=21:48:29 +2026-09-18 23:03:48 | INFO | train rank=0 | step=28790/50000 progress=57.6% loss=0.109878 lr=4.163983e-05 lr_backbone=4.163983e-05 elapsed=29:35:11 avg_step=3.70s recent_step=3.14s eta=21:47:48 +2026-09-18 23:04:16 | INFO | train rank=0 | step=28800/50000 progress=57.6% loss=0.075830 lr=4.160722e-05 lr_backbone=4.160722e-05 elapsed=29:35:39 avg_step=3.70s recent_step=2.83s eta=21:47:05 +2026-09-18 23:05:13 | INFO | train rank=0 | step=28810/50000 progress=57.6% loss=0.091449 lr=4.157463e-05 lr_backbone=4.157463e-05 elapsed=29:36:36 avg_step=3.70s recent_step=5.66s eta=21:46:42 +2026-09-18 23:05:43 | INFO | train rank=0 | step=28820/50000 progress=57.6% loss=0.085274 lr=4.154203e-05 lr_backbone=4.154203e-05 elapsed=29:37:06 avg_step=3.70s recent_step=3.00s eta=21:46:00 +2026-09-18 23:06:12 | INFO | train rank=0 | step=28830/50000 progress=57.7% loss=0.092817 lr=4.150944e-05 lr_backbone=4.150944e-05 elapsed=29:37:35 avg_step=3.70s recent_step=2.94s eta=21:45:17 +2026-09-18 23:06:45 | INFO | train rank=0 | step=28840/50000 progress=57.7% loss=0.113564 lr=4.147685e-05 lr_backbone=4.147685e-05 elapsed=29:38:08 avg_step=3.70s recent_step=3.24s eta=21:44:37 +2026-09-18 23:07:39 | INFO | train rank=0 | step=28850/50000 progress=57.7% loss=0.081929 lr=4.144427e-05 lr_backbone=4.144427e-05 elapsed=29:39:02 avg_step=3.70s recent_step=5.44s eta=21:44:13 +2026-09-18 23:08:12 | INFO | train rank=0 | step=28860/50000 progress=57.7% loss=0.087657 lr=4.141169e-05 lr_backbone=4.141169e-05 elapsed=29:39:35 avg_step=3.70s recent_step=3.31s eta=21:43:33 +2026-09-18 23:08:41 | INFO | train rank=0 | step=28870/50000 progress=57.7% loss=0.097267 lr=4.137911e-05 lr_backbone=4.137911e-05 elapsed=29:40:05 avg_step=3.70s recent_step=2.94s eta=21:42:50 +2026-09-18 23:09:19 | INFO | train rank=0 | step=28880/50000 progress=57.8% loss=0.087373 lr=4.134654e-05 lr_backbone=4.134654e-05 elapsed=29:40:42 avg_step=3.70s recent_step=3.77s eta=21:42:14 +2026-09-18 23:10:18 | INFO | train rank=0 | step=28890/50000 progress=57.8% loss=0.067577 lr=4.131397e-05 lr_backbone=4.131397e-05 elapsed=29:41:42 avg_step=3.70s recent_step=5.93s eta=21:41:53 +2026-09-18 23:10:43 | INFO | train rank=0 | step=28900/50000 progress=57.8% loss=0.086988 lr=4.128141e-05 lr_backbone=4.128141e-05 elapsed=29:42:06 avg_step=3.70s recent_step=2.46s eta=21:41:07 +2026-09-18 23:11:18 | INFO | train rank=0 | step=28910/50000 progress=57.8% loss=0.069989 lr=4.124885e-05 lr_backbone=4.124885e-05 elapsed=29:42:41 avg_step=3.70s recent_step=3.51s eta=21:40:29 +2026-09-18 23:11:48 | INFO | train rank=0 | step=28920/50000 progress=57.8% loss=0.135285 lr=4.121629e-05 lr_backbone=4.121629e-05 elapsed=29:43:11 avg_step=3.70s recent_step=2.95s eta=21:39:46 +2026-09-18 23:12:44 | INFO | train rank=0 | step=28930/50000 progress=57.9% loss=0.089163 lr=4.118374e-05 lr_backbone=4.118374e-05 elapsed=29:44:07 avg_step=3.70s recent_step=5.61s eta=21:39:23 +2026-09-18 23:13:17 | INFO | train rank=0 | step=28940/50000 progress=57.9% loss=0.124328 lr=4.115119e-05 lr_backbone=4.115119e-05 elapsed=29:44:41 avg_step=3.70s recent_step=3.36s eta=21:38:44 +2026-09-18 23:13:35 | INFO | train rank=0 | step=28950/50000 progress=57.9% loss=0.089450 lr=4.111864e-05 lr_backbone=4.111864e-05 elapsed=29:44:59 avg_step=3.70s recent_step=1.78s eta=21:37:53 +2026-09-18 23:14:02 | INFO | train rank=0 | step=28960/50000 progress=57.9% loss=0.099670 lr=4.108610e-05 lr_backbone=4.108610e-05 elapsed=29:45:26 avg_step=3.70s recent_step=2.72s eta=21:37:09 +2026-09-18 23:14:55 | INFO | train rank=0 | step=28970/50000 progress=57.9% loss=0.102718 lr=4.105356e-05 lr_backbone=4.105356e-05 elapsed=29:46:19 avg_step=3.70s recent_step=5.28s eta=21:36:43 +2026-09-18 23:15:26 | INFO | train rank=0 | step=28980/50000 progress=58.0% loss=0.105248 lr=4.102103e-05 lr_backbone=4.102103e-05 elapsed=29:46:50 avg_step=3.70s recent_step=3.12s eta=21:36:02 +2026-09-18 23:16:01 | INFO | train rank=0 | step=28990/50000 progress=58.0% loss=0.080328 lr=4.098850e-05 lr_backbone=4.098850e-05 elapsed=29:47:24 avg_step=3.70s recent_step=3.47s eta=21:35:23 +2026-09-18 23:16:26 | INFO | train rank=0 | step=29000/50000 progress=58.0% loss=0.070440 lr=4.095597e-05 lr_backbone=4.095597e-05 elapsed=29:47:50 avg_step=3.70s recent_step=2.53s eta=21:34:38 +2026-09-18 23:17:15 | INFO | train rank=0 | step=29010/50000 progress=58.0% loss=0.090321 lr=4.092345e-05 lr_backbone=4.092345e-05 elapsed=29:48:38 avg_step=3.70s recent_step=4.87s eta=21:34:09 +2026-09-18 23:17:50 | INFO | train rank=0 | step=29020/50000 progress=58.0% loss=0.083391 lr=4.089093e-05 lr_backbone=4.089093e-05 elapsed=29:49:14 avg_step=3.70s recent_step=3.52s eta=21:33:31 +2026-09-18 23:18:17 | INFO | train rank=0 | step=29030/50000 progress=58.1% loss=0.155494 lr=4.085842e-05 lr_backbone=4.085842e-05 elapsed=29:49:40 avg_step=3.70s recent_step=2.64s eta=21:32:46 +2026-09-18 23:18:52 | INFO | train rank=0 | step=29040/50000 progress=58.1% loss=0.141503 lr=4.082591e-05 lr_backbone=4.082591e-05 elapsed=29:50:15 avg_step=3.70s recent_step=3.49s eta=21:32:08 +2026-09-18 23:19:35 | INFO | train rank=0 | step=29050/50000 progress=58.1% loss=0.089038 lr=4.079340e-05 lr_backbone=4.079340e-05 elapsed=29:50:58 avg_step=3.70s recent_step=4.31s eta=21:31:35 +2026-09-18 23:20:06 | INFO | train rank=0 | step=29060/50000 progress=58.1% loss=0.119273 lr=4.076090e-05 lr_backbone=4.076090e-05 elapsed=29:51:29 avg_step=3.70s recent_step=3.08s eta=21:30:54 +2026-09-18 23:20:40 | INFO | train rank=0 | step=29070/50000 progress=58.1% loss=0.084819 lr=4.072840e-05 lr_backbone=4.072840e-05 elapsed=29:52:04 avg_step=3.70s recent_step=3.47s eta=21:30:15 +2026-09-18 23:21:24 | INFO | train rank=0 | step=29080/50000 progress=58.2% loss=0.160154 lr=4.069591e-05 lr_backbone=4.069591e-05 elapsed=29:52:47 avg_step=3.70s recent_step=4.37s eta=21:29:43 +2026-09-18 23:22:07 | INFO | train rank=0 | step=29090/50000 progress=58.2% loss=0.082871 lr=4.066342e-05 lr_backbone=4.066342e-05 elapsed=29:53:30 avg_step=3.70s recent_step=4.33s eta=21:29:11 +2026-09-18 23:22:43 | INFO | train rank=0 | step=29100/50000 progress=58.2% loss=0.097083 lr=4.063093e-05 lr_backbone=4.063093e-05 elapsed=29:54:07 avg_step=3.70s recent_step=3.62s eta=21:28:33 +2026-09-18 23:23:17 | INFO | train rank=0 | step=29110/50000 progress=58.2% loss=0.103286 lr=4.059845e-05 lr_backbone=4.059845e-05 elapsed=29:54:41 avg_step=3.70s recent_step=3.41s eta=21:27:54 +2026-09-18 23:24:03 | INFO | train rank=0 | step=29120/50000 progress=58.2% loss=0.106170 lr=4.056598e-05 lr_backbone=4.056598e-05 elapsed=29:55:26 avg_step=3.70s recent_step=4.52s eta=21:27:23 +2026-09-18 23:24:44 | INFO | train rank=0 | step=29130/50000 progress=58.3% loss=0.087869 lr=4.053350e-05 lr_backbone=4.053350e-05 elapsed=29:56:07 avg_step=3.70s recent_step=4.11s eta=21:26:49 +2026-09-18 23:25:17 | INFO | train rank=0 | step=29140/50000 progress=58.3% loss=0.104574 lr=4.050103e-05 lr_backbone=4.050103e-05 elapsed=29:56:41 avg_step=3.70s recent_step=3.37s eta=21:26:09 +2026-09-18 23:25:48 | INFO | train rank=0 | step=29150/50000 progress=58.3% loss=0.067865 lr=4.046857e-05 lr_backbone=4.046857e-05 elapsed=29:57:11 avg_step=3.70s recent_step=3.02s eta=21:25:28 +2026-09-18 23:26:35 | INFO | train rank=0 | step=29160/50000 progress=58.3% loss=0.194030 lr=4.043611e-05 lr_backbone=4.043611e-05 elapsed=29:57:58 avg_step=3.70s recent_step=4.73s eta=21:24:58 +2026-09-18 23:27:14 | INFO | train rank=0 | step=29170/50000 progress=58.3% loss=0.068877 lr=4.040365e-05 lr_backbone=4.040365e-05 elapsed=29:58:37 avg_step=3.70s recent_step=3.86s eta=21:24:22 +2026-09-18 23:27:43 | INFO | train rank=0 | step=29180/50000 progress=58.4% loss=0.099657 lr=4.037120e-05 lr_backbone=4.037120e-05 elapsed=29:59:06 avg_step=3.70s recent_step=2.95s eta=21:23:40 +2026-09-18 23:28:17 | INFO | train rank=0 | step=29190/50000 progress=58.4% loss=0.093433 lr=4.033875e-05 lr_backbone=4.033875e-05 elapsed=29:59:40 avg_step=3.70s recent_step=3.39s eta=21:23:01 +2026-09-18 23:29:09 | INFO | train rank=0 | step=29200/50000 progress=58.4% loss=0.094321 lr=4.030631e-05 lr_backbone=4.030631e-05 elapsed=30:00:32 avg_step=3.70s recent_step=5.18s eta=21:22:34 +2026-09-18 23:29:40 | INFO | train rank=0 | step=29210/50000 progress=58.4% loss=0.073448 lr=4.027387e-05 lr_backbone=4.027387e-05 elapsed=30:01:03 avg_step=3.70s recent_step=3.08s eta=21:21:53 +2026-09-18 23:30:16 | INFO | train rank=0 | step=29220/50000 progress=58.4% loss=0.101231 lr=4.024143e-05 lr_backbone=4.024143e-05 elapsed=30:01:40 avg_step=3.70s recent_step=3.67s eta=21:21:16 +2026-09-18 23:30:47 | INFO | train rank=0 | step=29230/50000 progress=58.5% loss=0.086849 lr=4.020900e-05 lr_backbone=4.020900e-05 elapsed=30:02:10 avg_step=3.70s recent_step=3.03s eta=21:20:34 +2026-09-18 23:31:39 | INFO | train rank=0 | step=29240/50000 progress=58.5% loss=0.135563 lr=4.017657e-05 lr_backbone=4.017657e-05 elapsed=30:03:02 avg_step=3.70s recent_step=5.23s eta=21:20:08 +2026-09-18 23:32:10 | INFO | train rank=0 | step=29250/50000 progress=58.5% loss=0.108057 lr=4.014415e-05 lr_backbone=4.014415e-05 elapsed=30:03:34 avg_step=3.70s recent_step=3.14s eta=21:19:27 +2026-09-18 23:32:39 | INFO | train rank=0 | step=29260/50000 progress=58.5% loss=0.123298 lr=4.011173e-05 lr_backbone=4.011173e-05 elapsed=30:04:03 avg_step=3.70s recent_step=2.90s eta=21:18:44 +2026-09-18 23:33:15 | INFO | train rank=0 | step=29270/50000 progress=58.5% loss=0.100757 lr=4.007932e-05 lr_backbone=4.007932e-05 elapsed=30:04:38 avg_step=3.70s recent_step=3.52s eta=21:18:06 +2026-09-18 23:34:11 | INFO | train rank=0 | step=29280/50000 progress=58.6% loss=0.101006 lr=4.004691e-05 lr_backbone=4.004691e-05 elapsed=30:05:35 avg_step=3.70s recent_step=5.69s eta=21:17:43 +2026-09-18 23:34:39 | INFO | train rank=0 | step=29290/50000 progress=58.6% loss=0.077662 lr=4.001450e-05 lr_backbone=4.001450e-05 elapsed=30:06:02 avg_step=3.70s recent_step=2.78s eta=21:16:59 +2026-09-18 23:35:12 | INFO | train rank=0 | step=29300/50000 progress=58.6% loss=0.100317 lr=3.998210e-05 lr_backbone=3.998210e-05 elapsed=30:06:36 avg_step=3.70s recent_step=3.31s eta=21:16:20 +2026-09-18 23:35:42 | INFO | train rank=0 | step=29310/50000 progress=58.6% loss=0.121129 lr=3.994970e-05 lr_backbone=3.994970e-05 elapsed=30:07:05 avg_step=3.70s recent_step=2.96s eta=21:15:37 +2026-09-18 23:36:39 | INFO | train rank=0 | step=29320/50000 progress=58.6% loss=0.110332 lr=3.991731e-05 lr_backbone=3.991731e-05 elapsed=30:08:02 avg_step=3.70s recent_step=5.68s eta=21:15:14 +2026-09-18 23:37:10 | INFO | train rank=0 | step=29330/50000 progress=58.7% loss=0.139821 lr=3.988492e-05 lr_backbone=3.988492e-05 elapsed=30:08:33 avg_step=3.70s recent_step=3.13s eta=21:14:33 +2026-09-18 23:37:41 | INFO | train rank=0 | step=29340/50000 progress=58.7% loss=0.129099 lr=3.985254e-05 lr_backbone=3.985254e-05 elapsed=30:09:04 avg_step=3.70s recent_step=3.09s eta=21:13:52 +2026-09-18 23:38:15 | INFO | train rank=0 | step=29350/50000 progress=58.7% loss=0.090139 lr=3.982016e-05 lr_backbone=3.982016e-05 elapsed=30:09:38 avg_step=3.70s recent_step=3.37s eta=21:13:13 +2026-09-18 23:39:05 | INFO | train rank=0 | step=29360/50000 progress=58.7% loss=0.080833 lr=3.978779e-05 lr_backbone=3.978779e-05 elapsed=30:10:28 avg_step=3.70s recent_step=5.00s eta=21:12:45 +2026-09-18 23:39:31 | INFO | train rank=0 | step=29370/50000 progress=58.7% loss=0.095562 lr=3.975542e-05 lr_backbone=3.975542e-05 elapsed=30:10:54 avg_step=3.70s recent_step=2.65s eta=21:12:01 +2026-09-18 23:40:03 | INFO | train rank=0 | step=29380/50000 progress=58.8% loss=0.116317 lr=3.972305e-05 lr_backbone=3.972305e-05 elapsed=30:11:26 avg_step=3.70s recent_step=3.21s eta=21:11:20 +2026-09-18 23:40:38 | INFO | train rank=0 | step=29390/50000 progress=58.8% loss=0.085165 lr=3.969069e-05 lr_backbone=3.969069e-05 elapsed=30:12:01 avg_step=3.70s recent_step=3.49s eta=21:10:42 +2026-09-18 23:41:30 | INFO | train rank=0 | step=29400/50000 progress=58.8% loss=0.081851 lr=3.965833e-05 lr_backbone=3.965833e-05 elapsed=30:12:53 avg_step=3.70s recent_step=5.18s eta=21:10:15 +2026-09-18 23:42:00 | INFO | train rank=0 | step=29410/50000 progress=58.8% loss=0.079024 lr=3.962598e-05 lr_backbone=3.962598e-05 elapsed=30:13:23 avg_step=3.70s recent_step=2.99s eta=21:09:33 +2026-09-18 23:42:35 | INFO | train rank=0 | step=29420/50000 progress=58.8% loss=0.165043 lr=3.959363e-05 lr_backbone=3.959363e-05 elapsed=30:13:58 avg_step=3.70s recent_step=3.54s eta=21:08:55 +2026-09-18 23:43:05 | INFO | train rank=0 | step=29430/50000 progress=58.9% loss=0.081481 lr=3.956129e-05 lr_backbone=3.956129e-05 elapsed=30:14:28 avg_step=3.70s recent_step=3.00s eta=21:08:13 +2026-09-18 23:43:47 | INFO | train rank=0 | step=29440/50000 progress=58.9% loss=0.101327 lr=3.952895e-05 lr_backbone=3.952895e-05 elapsed=30:15:11 avg_step=3.70s recent_step=4.23s eta=21:07:40 +2026-09-18 23:44:17 | INFO | train rank=0 | step=29450/50000 progress=58.9% loss=0.087352 lr=3.949662e-05 lr_backbone=3.949662e-05 elapsed=30:15:41 avg_step=3.70s recent_step=2.99s eta=21:06:58 +2026-09-18 23:44:44 | INFO | train rank=0 | step=29460/50000 progress=58.9% loss=0.080214 lr=3.946429e-05 lr_backbone=3.946429e-05 elapsed=30:16:07 avg_step=3.70s recent_step=2.62s eta=21:06:13 +2026-09-18 23:45:19 | INFO | train rank=0 | step=29470/50000 progress=58.9% loss=0.094875 lr=3.943196e-05 lr_backbone=3.943196e-05 elapsed=30:16:42 avg_step=3.70s recent_step=3.52s eta=21:05:35 +2026-09-18 23:46:07 | INFO | train rank=0 | step=29480/50000 progress=59.0% loss=0.102309 lr=3.939964e-05 lr_backbone=3.939964e-05 elapsed=30:17:30 avg_step=3.70s recent_step=4.84s eta=21:05:06 +2026-09-18 23:46:37 | INFO | train rank=0 | step=29490/50000 progress=59.0% loss=0.093858 lr=3.936733e-05 lr_backbone=3.936733e-05 elapsed=30:18:01 avg_step=3.70s recent_step=3.01s eta=21:04:24 +2026-09-18 23:47:20 | INFO | train rank=0 | step=29500/50000 progress=59.0% loss=0.078462 lr=3.933502e-05 lr_backbone=3.933502e-05 elapsed=30:18:43 avg_step=3.70s recent_step=4.25s eta=21:03:51 +2026-09-18 23:47:56 | INFO | train rank=0 | step=29510/50000 progress=59.0% loss=0.088684 lr=3.930271e-05 lr_backbone=3.930271e-05 elapsed=30:19:20 avg_step=3.70s recent_step=3.65s eta=21:03:14 +2026-09-18 23:48:26 | INFO | train rank=0 | step=29520/50000 progress=59.0% loss=0.099547 lr=3.927041e-05 lr_backbone=3.927041e-05 elapsed=30:19:49 avg_step=3.70s recent_step=2.97s eta=21:02:32 +2026-09-18 23:49:06 | INFO | train rank=0 | step=29530/50000 progress=59.1% loss=0.094440 lr=3.923811e-05 lr_backbone=3.923811e-05 elapsed=30:20:30 avg_step=3.70s recent_step=4.03s eta=21:01:57 +2026-09-18 23:49:48 | INFO | train rank=0 | step=29540/50000 progress=59.1% loss=0.099920 lr=3.920582e-05 lr_backbone=3.920582e-05 elapsed=30:21:12 avg_step=3.70s recent_step=4.22s eta=21:01:24 +2026-09-18 23:50:21 | INFO | train rank=0 | step=29550/50000 progress=59.1% loss=0.168914 lr=3.917354e-05 lr_backbone=3.917354e-05 elapsed=30:21:45 avg_step=3.70s recent_step=3.28s eta=21:00:44 +2026-09-18 23:51:00 | INFO | train rank=0 | step=29560/50000 progress=59.1% loss=0.089554 lr=3.914125e-05 lr_backbone=3.914125e-05 elapsed=30:22:23 avg_step=3.70s recent_step=3.89s eta=21:00:08 +2026-09-18 23:51:41 | INFO | train rank=0 | step=29570/50000 progress=59.1% loss=0.103768 lr=3.910898e-05 lr_backbone=3.910898e-05 elapsed=30:23:05 avg_step=3.70s recent_step=4.10s eta=20:59:34 +2026-09-18 23:52:15 | INFO | train rank=0 | step=29580/50000 progress=59.2% loss=0.100674 lr=3.907670e-05 lr_backbone=3.907670e-05 elapsed=30:23:38 avg_step=3.70s recent_step=3.34s eta=20:58:54 +2026-09-18 23:52:40 | INFO | train rank=0 | step=29590/50000 progress=59.2% loss=0.101798 lr=3.904443e-05 lr_backbone=3.904443e-05 elapsed=30:24:04 avg_step=3.70s recent_step=2.58s eta=20:58:10 +2026-09-18 23:53:10 | INFO | train rank=0 | step=29600/50000 progress=59.2% loss=0.081641 lr=3.901217e-05 lr_backbone=3.901217e-05 elapsed=30:24:33 avg_step=3.70s recent_step=2.94s eta=20:57:28 +2026-09-18 23:54:02 | INFO | train rank=0 | step=29610/50000 progress=59.2% loss=0.102380 lr=3.897991e-05 lr_backbone=3.897991e-05 elapsed=30:25:25 avg_step=3.70s recent_step=5.23s eta=20:57:01 +2026-09-18 23:54:50 | INFO | train rank=0 | step=29620/50000 progress=59.2% loss=0.072276 lr=3.894766e-05 lr_backbone=3.894766e-05 elapsed=30:26:13 avg_step=3.70s recent_step=4.77s eta=20:56:32 +2026-09-18 23:55:26 | INFO | train rank=0 | step=29630/50000 progress=59.3% loss=0.061894 lr=3.891541e-05 lr_backbone=3.891541e-05 elapsed=30:26:49 avg_step=3.70s recent_step=3.59s eta=20:55:54 +2026-09-18 23:55:55 | INFO | train rank=0 | step=29640/50000 progress=59.3% loss=0.086641 lr=3.888317e-05 lr_backbone=3.888317e-05 elapsed=30:27:18 avg_step=3.70s recent_step=2.93s eta=20:55:12 +2026-09-18 23:56:32 | INFO | train rank=0 | step=29650/50000 progress=59.3% loss=0.074092 lr=3.885093e-05 lr_backbone=3.885093e-05 elapsed=30:27:55 avg_step=3.70s recent_step=3.70s eta=20:54:35 +2026-09-18 23:57:13 | INFO | train rank=0 | step=29660/50000 progress=59.3% loss=0.081433 lr=3.881869e-05 lr_backbone=3.881869e-05 elapsed=30:28:37 avg_step=3.70s recent_step=4.12s eta=20:54:00 +2026-09-18 23:57:42 | INFO | train rank=0 | step=29670/50000 progress=59.3% loss=0.063413 lr=3.878646e-05 lr_backbone=3.878646e-05 elapsed=30:29:06 avg_step=3.70s recent_step=2.90s eta=20:53:18 +2026-09-18 23:58:11 | INFO | train rank=0 | step=29680/50000 progress=59.4% loss=0.080659 lr=3.875424e-05 lr_backbone=3.875424e-05 elapsed=30:29:34 avg_step=3.70s recent_step=2.88s eta=20:52:35 +2026-09-18 23:59:02 | INFO | train rank=0 | step=29690/50000 progress=59.4% loss=0.096743 lr=3.872202e-05 lr_backbone=3.872202e-05 elapsed=30:30:25 avg_step=3.70s recent_step=5.08s eta=20:52:08 +2026-09-18 23:59:40 | INFO | train rank=0 | step=29700/50000 progress=59.4% loss=0.089387 lr=3.868980e-05 lr_backbone=3.868980e-05 elapsed=30:31:03 avg_step=3.70s recent_step=3.80s eta=20:51:32 +2026-09-19 00:00:19 | INFO | train rank=0 | step=29710/50000 progress=59.4% loss=0.067704 lr=3.865759e-05 lr_backbone=3.865759e-05 elapsed=30:31:42 avg_step=3.70s recent_step=3.91s eta=20:50:56 +2026-09-19 00:00:52 | INFO | train rank=0 | step=29720/50000 progress=59.4% loss=0.083338 lr=3.862539e-05 lr_backbone=3.862539e-05 elapsed=30:32:15 avg_step=3.70s recent_step=3.30s eta=20:50:16 +2026-09-19 00:01:37 | INFO | train rank=0 | step=29730/50000 progress=59.5% loss=0.120269 lr=3.859319e-05 lr_backbone=3.859319e-05 elapsed=30:33:00 avg_step=3.70s recent_step=4.49s eta=20:49:45 +2026-09-19 00:02:19 | INFO | train rank=0 | step=29740/50000 progress=59.5% loss=0.097314 lr=3.856099e-05 lr_backbone=3.856099e-05 elapsed=30:33:43 avg_step=3.70s recent_step=4.24s eta=20:49:11 +2026-09-19 00:02:50 | INFO | train rank=0 | step=29750/50000 progress=59.5% loss=0.137477 lr=3.852880e-05 lr_backbone=3.852880e-05 elapsed=30:34:14 avg_step=3.70s recent_step=3.09s eta=20:48:30 +2026-09-19 00:03:25 | INFO | train rank=0 | step=29760/50000 progress=59.5% loss=0.083983 lr=3.849662e-05 lr_backbone=3.849662e-05 elapsed=30:34:48 avg_step=3.70s recent_step=3.44s eta=20:47:52 +2026-09-19 00:04:16 | INFO | train rank=0 | step=29770/50000 progress=59.5% loss=0.089474 lr=3.846444e-05 lr_backbone=3.846444e-05 elapsed=30:35:39 avg_step=3.70s recent_step=5.10s eta=20:47:24 +2026-09-19 00:04:52 | INFO | train rank=0 | step=29780/50000 progress=59.6% loss=0.083574 lr=3.843226e-05 lr_backbone=3.843226e-05 elapsed=30:36:15 avg_step=3.70s recent_step=3.58s eta=20:46:46 +2026-09-19 00:05:28 | INFO | train rank=0 | step=29790/50000 progress=59.6% loss=0.083558 lr=3.840009e-05 lr_backbone=3.840009e-05 elapsed=30:36:51 avg_step=3.70s recent_step=3.61s eta=20:46:09 +2026-09-19 00:05:53 | INFO | train rank=0 | step=29800/50000 progress=59.6% loss=0.086751 lr=3.836793e-05 lr_backbone=3.836793e-05 elapsed=30:37:16 avg_step=3.70s recent_step=2.53s eta=20:45:24 +2026-09-19 00:06:47 | INFO | train rank=0 | step=29810/50000 progress=59.6% loss=0.120531 lr=3.833577e-05 lr_backbone=3.833577e-05 elapsed=30:38:10 avg_step=3.70s recent_step=5.43s eta=20:44:58 +2026-09-19 00:07:22 | INFO | train rank=0 | step=29820/50000 progress=59.6% loss=0.092787 lr=3.830362e-05 lr_backbone=3.830362e-05 elapsed=30:38:46 avg_step=3.70s recent_step=3.52s eta=20:44:20 +2026-09-19 00:07:59 | INFO | train rank=0 | step=29830/50000 progress=59.7% loss=0.087907 lr=3.827147e-05 lr_backbone=3.827147e-05 elapsed=30:39:22 avg_step=3.70s recent_step=3.68s eta=20:43:43 +2026-09-19 00:08:23 | INFO | train rank=0 | step=29840/50000 progress=59.7% loss=0.081977 lr=3.823932e-05 lr_backbone=3.823932e-05 elapsed=30:39:46 avg_step=3.70s recent_step=2.39s eta=20:42:57 +2026-09-19 00:09:16 | INFO | train rank=0 | step=29850/50000 progress=59.7% loss=0.094549 lr=3.820718e-05 lr_backbone=3.820718e-05 elapsed=30:40:39 avg_step=3.70s recent_step=5.31s eta=20:42:31 +2026-09-19 00:09:44 | INFO | train rank=0 | step=29860/50000 progress=59.7% loss=0.073832 lr=3.817505e-05 lr_backbone=3.817505e-05 elapsed=30:41:08 avg_step=3.70s recent_step=2.83s eta=20:41:48 +2026-09-19 00:10:18 | INFO | train rank=0 | step=29870/50000 progress=59.7% loss=0.074866 lr=3.814292e-05 lr_backbone=3.814292e-05 elapsed=30:41:41 avg_step=3.70s recent_step=3.36s eta=20:41:09 +2026-09-19 00:10:52 | INFO | train rank=0 | step=29880/50000 progress=59.8% loss=0.106861 lr=3.811080e-05 lr_backbone=3.811080e-05 elapsed=30:42:15 avg_step=3.70s recent_step=3.34s eta=20:40:30 +2026-09-19 00:11:45 | INFO | train rank=0 | step=29890/50000 progress=59.8% loss=0.075069 lr=3.807868e-05 lr_backbone=3.807868e-05 elapsed=30:43:08 avg_step=3.70s recent_step=5.34s eta=20:40:04 +2026-09-19 00:12:20 | INFO | train rank=0 | step=29900/50000 progress=59.8% loss=0.117635 lr=3.804657e-05 lr_backbone=3.804657e-05 elapsed=30:43:43 avg_step=3.70s recent_step=3.50s eta=20:39:25 +2026-09-19 00:12:49 | INFO | train rank=0 | step=29910/50000 progress=59.8% loss=0.097655 lr=3.801446e-05 lr_backbone=3.801446e-05 elapsed=30:44:12 avg_step=3.70s recent_step=2.92s eta=20:38:43 +2026-09-19 00:13:29 | INFO | train rank=0 | step=29920/50000 progress=59.8% loss=0.090529 lr=3.798236e-05 lr_backbone=3.798236e-05 elapsed=30:44:53 avg_step=3.70s recent_step=4.02s eta=20:38:08 +2026-09-19 00:14:15 | INFO | train rank=0 | step=29930/50000 progress=59.9% loss=0.076500 lr=3.795026e-05 lr_backbone=3.795026e-05 elapsed=30:45:39 avg_step=3.70s recent_step=4.61s eta=20:37:37 +2026-09-19 00:14:48 | INFO | train rank=0 | step=29940/50000 progress=59.9% loss=0.064272 lr=3.791817e-05 lr_backbone=3.791817e-05 elapsed=30:46:11 avg_step=3.70s recent_step=3.26s eta=20:36:57 +2026-09-19 00:15:28 | INFO | train rank=0 | step=29950/50000 progress=59.9% loss=0.095780 lr=3.788608e-05 lr_backbone=3.788608e-05 elapsed=30:46:51 avg_step=3.70s recent_step=3.96s eta=20:36:22 +2026-09-19 00:15:57 | INFO | train rank=0 | step=29960/50000 progress=59.9% loss=0.079249 lr=3.785400e-05 lr_backbone=3.785400e-05 elapsed=30:47:20 avg_step=3.70s recent_step=2.93s eta=20:35:40 +2026-09-19 00:16:51 | INFO | train rank=0 | step=29970/50000 progress=59.9% loss=0.078250 lr=3.782192e-05 lr_backbone=3.782192e-05 elapsed=30:48:14 avg_step=3.70s recent_step=5.38s eta=20:35:14 +2026-09-19 00:17:21 | INFO | train rank=0 | step=29980/50000 progress=60.0% loss=0.081336 lr=3.778985e-05 lr_backbone=3.778985e-05 elapsed=30:48:44 avg_step=3.70s recent_step=3.00s eta=20:34:32 +2026-09-19 00:17:52 | INFO | train rank=0 | step=29990/50000 progress=60.0% loss=0.080125 lr=3.775779e-05 lr_backbone=3.775779e-05 elapsed=30:49:15 avg_step=3.70s recent_step=3.09s eta=20:33:51 +2026-09-19 00:18:17 | INFO | train rank=0 | step=30000/50000 progress=60.0% loss=0.082619 lr=3.772573e-05 lr_backbone=3.772573e-05 elapsed=30:49:40 avg_step=3.70s recent_step=2.52s eta=20:33:07 +2026-09-19 00:19:11 | INFO | train rank=0 | step=30010/50000 progress=60.0% loss=0.157589 lr=3.769367e-05 lr_backbone=3.769367e-05 elapsed=30:50:34 avg_step=3.70s recent_step=5.43s eta=20:32:41 +2026-09-19 00:19:43 | INFO | train rank=0 | step=30020/50000 progress=60.0% loss=0.075674 lr=3.766162e-05 lr_backbone=3.766162e-05 elapsed=30:51:06 avg_step=3.70s recent_step=3.19s eta=20:32:01 +2026-09-19 00:20:13 | INFO | train rank=0 | step=30030/50000 progress=60.1% loss=0.097249 lr=3.762958e-05 lr_backbone=3.762958e-05 elapsed=30:51:37 avg_step=3.70s recent_step=3.04s eta=20:31:19 +2026-09-19 00:20:49 | INFO | train rank=0 | step=30040/50000 progress=60.1% loss=0.059904 lr=3.759754e-05 lr_backbone=3.759754e-05 elapsed=30:52:13 avg_step=3.70s recent_step=3.61s eta=20:30:42 +2026-09-19 00:21:46 | INFO | train rank=0 | step=30050/50000 progress=60.1% loss=0.103965 lr=3.756551e-05 lr_backbone=3.756551e-05 elapsed=30:53:09 avg_step=3.70s recent_step=5.66s eta=20:30:18 +2026-09-19 00:22:15 | INFO | train rank=0 | step=30060/50000 progress=60.1% loss=0.081766 lr=3.753348e-05 lr_backbone=3.753348e-05 elapsed=30:53:39 avg_step=3.70s recent_step=2.93s eta=20:29:36 +2026-09-19 00:22:43 | INFO | train rank=0 | step=30070/50000 progress=60.1% loss=0.124141 lr=3.750146e-05 lr_backbone=3.750146e-05 elapsed=30:54:06 avg_step=3.70s recent_step=2.79s eta=20:28:53 +2026-09-19 00:23:08 | INFO | train rank=0 | step=30080/50000 progress=60.2% loss=0.100988 lr=3.746944e-05 lr_backbone=3.746944e-05 elapsed=30:54:31 avg_step=3.70s recent_step=2.50s eta=20:28:08 +2026-09-19 00:23:55 | INFO | train rank=0 | step=30090/50000 progress=60.2% loss=0.088293 lr=3.743743e-05 lr_backbone=3.743743e-05 elapsed=30:55:18 avg_step=3.70s recent_step=4.67s eta=20:27:37 +2026-09-19 00:24:25 | INFO | train rank=0 | step=30100/50000 progress=60.2% loss=0.080950 lr=3.740542e-05 lr_backbone=3.740542e-05 elapsed=30:55:49 avg_step=3.70s recent_step=3.05s eta=20:26:56 +2026-09-19 00:25:06 | INFO | train rank=0 | step=30110/50000 progress=60.2% loss=0.080589 lr=3.737342e-05 lr_backbone=3.737342e-05 elapsed=30:56:29 avg_step=3.70s recent_step=4.05s eta=20:26:21 +2026-09-19 00:25:39 | INFO | train rank=0 | step=30120/50000 progress=60.2% loss=0.143741 lr=3.734143e-05 lr_backbone=3.734143e-05 elapsed=30:57:03 avg_step=3.70s recent_step=3.36s eta=20:25:42 +2026-09-19 00:26:42 | INFO | train rank=0 | step=30130/50000 progress=60.3% loss=0.067403 lr=3.730944e-05 lr_backbone=3.730944e-05 elapsed=30:58:05 avg_step=3.70s recent_step=6.23s eta=20:25:21 +2026-09-19 00:27:03 | INFO | train rank=0 | step=30140/50000 progress=60.3% loss=0.082924 lr=3.727745e-05 lr_backbone=3.727745e-05 elapsed=30:58:26 avg_step=3.70s recent_step=2.12s eta=20:24:34 +2026-09-19 00:27:33 | INFO | train rank=0 | step=30150/50000 progress=60.3% loss=0.077853 lr=3.724548e-05 lr_backbone=3.724548e-05 elapsed=30:58:56 avg_step=3.70s recent_step=3.01s eta=20:23:52 +2026-09-19 00:28:01 | INFO | train rank=0 | step=30160/50000 progress=60.3% loss=0.107609 lr=3.721350e-05 lr_backbone=3.721350e-05 elapsed=30:59:24 avg_step=3.70s recent_step=2.77s eta=20:23:09 +2026-09-19 00:28:42 | INFO | train rank=0 | step=30170/50000 progress=60.3% loss=0.088549 lr=3.718154e-05 lr_backbone=3.718154e-05 elapsed=31:00:05 avg_step=3.70s recent_step=4.13s eta=20:22:35 +2026-09-19 00:29:10 | INFO | train rank=0 | step=30180/50000 progress=60.4% loss=0.137745 lr=3.714957e-05 lr_backbone=3.714957e-05 elapsed=31:00:34 avg_step=3.70s recent_step=2.85s eta=20:21:53 +2026-09-19 00:29:45 | INFO | train rank=0 | step=30190/50000 progress=60.4% loss=0.075775 lr=3.711762e-05 lr_backbone=3.711762e-05 elapsed=31:01:08 avg_step=3.70s recent_step=3.47s eta=20:21:14 +2026-09-19 00:30:16 | INFO | train rank=0 | step=30200/50000 progress=60.4% loss=0.078208 lr=3.708567e-05 lr_backbone=3.708567e-05 elapsed=31:01:39 avg_step=3.70s recent_step=3.07s eta=20:20:33 +2026-09-19 00:31:15 | INFO | train rank=0 | step=30210/50000 progress=60.4% loss=0.076431 lr=3.705372e-05 lr_backbone=3.705372e-05 elapsed=31:02:38 avg_step=3.70s recent_step=5.88s eta=20:20:10 +2026-09-19 00:31:53 | INFO | train rank=0 | step=30220/50000 progress=60.4% loss=0.101064 lr=3.702179e-05 lr_backbone=3.702179e-05 elapsed=31:03:16 avg_step=3.70s recent_step=3.84s eta=20:19:34 +2026-09-19 00:32:28 | INFO | train rank=0 | step=30230/50000 progress=60.5% loss=0.077826 lr=3.698985e-05 lr_backbone=3.698985e-05 elapsed=31:03:51 avg_step=3.70s recent_step=3.50s eta=20:18:56 +2026-09-19 00:33:04 | INFO | train rank=0 | step=30240/50000 progress=60.5% loss=0.091590 lr=3.695792e-05 lr_backbone=3.695792e-05 elapsed=31:04:27 avg_step=3.70s recent_step=3.59s eta=20:18:18 +2026-09-19 00:33:55 | INFO | train rank=0 | step=30250/50000 progress=60.5% loss=0.088562 lr=3.692600e-05 lr_backbone=3.692600e-05 elapsed=31:05:18 avg_step=3.70s recent_step=5.12s eta=20:17:50 +2026-09-19 00:34:27 | INFO | train rank=0 | step=30260/50000 progress=60.5% loss=0.090356 lr=3.689409e-05 lr_backbone=3.689409e-05 elapsed=31:05:50 avg_step=3.70s recent_step=3.17s eta=20:17:10 +2026-09-19 00:34:54 | INFO | train rank=0 | step=30270/50000 progress=60.5% loss=0.113226 lr=3.686218e-05 lr_backbone=3.686218e-05 elapsed=31:06:17 avg_step=3.70s recent_step=2.74s eta=20:16:27 +2026-09-19 00:35:22 | INFO | train rank=0 | step=30280/50000 progress=60.6% loss=0.093192 lr=3.683027e-05 lr_backbone=3.683027e-05 elapsed=31:06:45 avg_step=3.70s recent_step=2.75s eta=20:15:44 +2026-09-19 00:36:17 | INFO | train rank=0 | step=30290/50000 progress=60.6% loss=0.060148 lr=3.679837e-05 lr_backbone=3.679837e-05 elapsed=31:07:40 avg_step=3.70s recent_step=5.52s eta=20:15:18 +2026-09-19 00:36:45 | INFO | train rank=0 | step=30300/50000 progress=60.6% loss=0.075035 lr=3.676648e-05 lr_backbone=3.676648e-05 elapsed=31:08:08 avg_step=3.70s recent_step=2.82s eta=20:14:36 +2026-09-19 00:37:25 | INFO | train rank=0 | step=30310/50000 progress=60.6% loss=0.097017 lr=3.673459e-05 lr_backbone=3.673459e-05 elapsed=31:08:49 avg_step=3.70s recent_step=4.03s eta=20:14:01 +2026-09-19 00:37:48 | INFO | train rank=0 | step=30320/50000 progress=60.6% loss=0.093372 lr=3.670271e-05 lr_backbone=3.670271e-05 elapsed=31:09:11 avg_step=3.70s recent_step=2.27s eta=20:13:15 +2026-09-19 00:38:45 | INFO | train rank=0 | step=30330/50000 progress=60.7% loss=0.102124 lr=3.667084e-05 lr_backbone=3.667084e-05 elapsed=31:10:09 avg_step=3.70s recent_step=5.74s eta=20:12:51 +2026-09-19 00:39:20 | INFO | train rank=0 | step=30340/50000 progress=60.7% loss=0.084879 lr=3.663897e-05 lr_backbone=3.663897e-05 elapsed=31:10:43 avg_step=3.70s recent_step=3.43s eta=20:12:12 +2026-09-19 00:39:53 | INFO | train rank=0 | step=30350/50000 progress=60.7% loss=0.085323 lr=3.660710e-05 lr_backbone=3.660710e-05 elapsed=31:11:17 avg_step=3.70s recent_step=3.36s eta=20:11:33 +2026-09-19 00:40:27 | INFO | train rank=0 | step=30360/50000 progress=60.7% loss=0.090677 lr=3.657524e-05 lr_backbone=3.657524e-05 elapsed=31:11:50 avg_step=3.70s recent_step=3.32s eta=20:10:53 +2026-09-19 00:41:24 | INFO | train rank=0 | step=30370/50000 progress=60.7% loss=0.085628 lr=3.654339e-05 lr_backbone=3.654339e-05 elapsed=31:12:47 avg_step=3.70s recent_step=5.74s eta=20:10:30 +2026-09-19 00:41:47 | INFO | train rank=0 | step=30380/50000 progress=60.8% loss=0.095277 lr=3.651155e-05 lr_backbone=3.651155e-05 elapsed=31:13:11 avg_step=3.70s recent_step=2.35s eta=20:09:44 +2026-09-19 00:42:18 | INFO | train rank=0 | step=30390/50000 progress=60.8% loss=0.083818 lr=3.647971e-05 lr_backbone=3.647971e-05 elapsed=31:13:41 avg_step=3.70s recent_step=3.07s eta=20:09:03 +2026-09-19 00:42:48 | INFO | train rank=0 | step=30400/50000 progress=60.8% loss=0.106573 lr=3.644787e-05 lr_backbone=3.644787e-05 elapsed=31:14:11 avg_step=3.70s recent_step=2.97s eta=20:08:21 +2026-09-19 00:43:37 | INFO | train rank=0 | step=30410/50000 progress=60.8% loss=0.081161 lr=3.641604e-05 lr_backbone=3.641604e-05 elapsed=31:15:01 avg_step=3.70s recent_step=4.95s eta=20:07:52 +2026-09-19 00:44:05 | INFO | train rank=0 | step=30420/50000 progress=60.8% loss=0.095083 lr=3.638422e-05 lr_backbone=3.638422e-05 elapsed=31:15:29 avg_step=3.70s recent_step=2.82s eta=20:07:10 +2026-09-19 00:44:43 | INFO | train rank=0 | step=30430/50000 progress=60.9% loss=0.075261 lr=3.635240e-05 lr_backbone=3.635240e-05 elapsed=31:16:06 avg_step=3.70s recent_step=3.76s eta=20:06:33 +2026-09-19 00:45:15 | INFO | train rank=0 | step=30440/50000 progress=60.9% loss=0.082118 lr=3.632059e-05 lr_backbone=3.632059e-05 elapsed=31:16:39 avg_step=3.70s recent_step=3.22s eta=20:05:53 +2026-09-19 00:46:09 | INFO | train rank=0 | step=30450/50000 progress=60.9% loss=0.068520 lr=3.628879e-05 lr_backbone=3.628879e-05 elapsed=31:17:32 avg_step=3.70s recent_step=5.37s eta=20:05:27 +2026-09-19 00:46:42 | INFO | train rank=0 | step=30460/50000 progress=60.9% loss=0.111966 lr=3.625699e-05 lr_backbone=3.625699e-05 elapsed=31:18:06 avg_step=3.70s recent_step=3.35s eta=20:04:47 +2026-09-19 00:47:19 | INFO | train rank=0 | step=30470/50000 progress=60.9% loss=0.089021 lr=3.622520e-05 lr_backbone=3.622520e-05 elapsed=31:18:43 avg_step=3.70s recent_step=3.68s eta=20:04:10 +2026-09-19 00:47:49 | INFO | train rank=0 | step=30480/50000 progress=61.0% loss=0.104057 lr=3.619341e-05 lr_backbone=3.619341e-05 elapsed=31:19:12 avg_step=3.70s recent_step=2.98s eta=20:03:29 +2026-09-19 00:48:33 | INFO | train rank=0 | step=30490/50000 progress=61.0% loss=0.084340 lr=3.616163e-05 lr_backbone=3.616163e-05 elapsed=31:19:56 avg_step=3.70s recent_step=4.36s eta=20:02:56 +2026-09-19 00:49:06 | INFO | train rank=0 | step=30500/50000 progress=61.0% loss=0.074676 lr=3.612985e-05 lr_backbone=3.612985e-05 elapsed=31:20:30 avg_step=3.70s recent_step=3.35s eta=20:02:17 +2026-09-19 00:49:35 | INFO | train rank=0 | step=30510/50000 progress=61.0% loss=0.086673 lr=3.609809e-05 lr_backbone=3.609809e-05 elapsed=31:20:58 avg_step=3.70s recent_step=2.84s eta=20:01:34 +2026-09-19 00:50:14 | INFO | train rank=0 | step=30520/50000 progress=61.0% loss=0.095249 lr=3.606632e-05 lr_backbone=3.606632e-05 elapsed=31:21:37 avg_step=3.70s recent_step=3.91s eta=20:00:59 +2026-09-19 00:51:02 | INFO | train rank=0 | step=30530/50000 progress=61.1% loss=0.130121 lr=3.603457e-05 lr_backbone=3.603457e-05 elapsed=31:22:26 avg_step=3.70s recent_step=4.85s eta=20:00:29 +2026-09-19 00:51:38 | INFO | train rank=0 | step=30540/50000 progress=61.1% loss=0.094671 lr=3.600282e-05 lr_backbone=3.600282e-05 elapsed=31:23:01 avg_step=3.70s recent_step=3.53s eta=19:59:51 +2026-09-19 00:52:14 | INFO | train rank=0 | step=30550/50000 progress=61.1% loss=0.078736 lr=3.597107e-05 lr_backbone=3.597107e-05 elapsed=31:23:37 avg_step=3.70s recent_step=3.65s eta=19:59:14 +2026-09-19 00:52:46 | INFO | train rank=0 | step=30560/50000 progress=61.1% loss=0.098616 lr=3.593934e-05 lr_backbone=3.593934e-05 elapsed=31:24:10 avg_step=3.70s recent_step=3.23s eta=19:58:34 +2026-09-19 00:53:25 | INFO | train rank=0 | step=30570/50000 progress=61.1% loss=0.078656 lr=3.590760e-05 lr_backbone=3.590760e-05 elapsed=31:24:48 avg_step=3.70s recent_step=3.82s eta=19:57:57 +2026-09-19 00:54:14 | INFO | train rank=0 | step=30580/50000 progress=61.2% loss=0.072498 lr=3.587588e-05 lr_backbone=3.587588e-05 elapsed=31:25:37 avg_step=3.70s recent_step=4.90s eta=19:57:28 +2026-09-19 00:54:46 | INFO | train rank=0 | step=30590/50000 progress=61.2% loss=0.093292 lr=3.584416e-05 lr_backbone=3.584416e-05 elapsed=31:26:09 avg_step=3.70s recent_step=3.23s eta=19:56:48 +2026-09-19 00:55:15 | INFO | train rank=0 | step=30600/50000 progress=61.2% loss=0.063124 lr=3.581244e-05 lr_backbone=3.581244e-05 elapsed=31:26:38 avg_step=3.70s recent_step=2.91s eta=19:56:06 +2026-09-19 00:56:03 | INFO | train rank=0 | step=30610/50000 progress=61.2% loss=0.075002 lr=3.578074e-05 lr_backbone=3.578074e-05 elapsed=31:27:26 avg_step=3.70s recent_step=4.80s eta=19:55:36 +2026-09-19 00:56:47 | INFO | train rank=0 | step=30620/50000 progress=61.2% loss=0.107117 lr=3.574904e-05 lr_backbone=3.574904e-05 elapsed=31:28:10 avg_step=3.70s recent_step=4.38s eta=19:55:03 +2026-09-19 00:57:13 | INFO | train rank=0 | step=30630/50000 progress=61.3% loss=0.106085 lr=3.571734e-05 lr_backbone=3.571734e-05 elapsed=31:28:37 avg_step=3.70s recent_step=2.65s eta=19:54:20 +2026-09-19 00:57:50 | INFO | train rank=0 | step=30640/50000 progress=61.3% loss=0.086451 lr=3.568565e-05 lr_backbone=3.568565e-05 elapsed=31:29:13 avg_step=3.70s recent_step=3.61s eta=19:53:42 +2026-09-19 00:58:28 | INFO | train rank=0 | step=30650/50000 progress=61.3% loss=0.101156 lr=3.565397e-05 lr_backbone=3.565397e-05 elapsed=31:29:51 avg_step=3.70s recent_step=3.84s eta=19:53:06 +2026-09-19 00:59:20 | INFO | train rank=0 | step=30660/50000 progress=61.3% loss=0.088765 lr=3.562230e-05 lr_backbone=3.562230e-05 elapsed=31:30:43 avg_step=3.70s recent_step=5.20s eta=19:52:39 +2026-09-19 00:59:52 | INFO | train rank=0 | step=30670/50000 progress=61.3% loss=0.102829 lr=3.559063e-05 lr_backbone=3.559063e-05 elapsed=31:31:16 avg_step=3.70s recent_step=3.26s eta=19:51:59 +2026-09-19 01:00:20 | INFO | train rank=0 | step=30680/50000 progress=61.4% loss=0.085003 lr=3.555896e-05 lr_backbone=3.555896e-05 elapsed=31:31:43 avg_step=3.70s recent_step=2.73s eta=19:51:16 +2026-09-19 01:00:51 | INFO | train rank=0 | step=30690/50000 progress=61.4% loss=0.101882 lr=3.552731e-05 lr_backbone=3.552731e-05 elapsed=31:32:14 avg_step=3.70s recent_step=3.11s eta=19:50:35 +2026-09-19 01:01:46 | INFO | train rank=0 | step=30700/50000 progress=61.4% loss=0.091417 lr=3.549566e-05 lr_backbone=3.549566e-05 elapsed=31:33:09 avg_step=3.70s recent_step=5.49s eta=19:50:09 +2026-09-19 01:02:20 | INFO | train rank=0 | step=30710/50000 progress=61.4% loss=0.082852 lr=3.546401e-05 lr_backbone=3.546401e-05 elapsed=31:33:43 avg_step=3.70s recent_step=3.39s eta=19:49:30 +2026-09-19 01:02:50 | INFO | train rank=0 | step=30720/50000 progress=61.4% loss=0.076653 lr=3.543237e-05 lr_backbone=3.543237e-05 elapsed=31:34:13 avg_step=3.70s recent_step=3.04s eta=19:48:49 +2026-09-19 01:03:25 | INFO | train rank=0 | step=30730/50000 progress=61.5% loss=0.090751 lr=3.540074e-05 lr_backbone=3.540074e-05 elapsed=31:34:48 avg_step=3.70s recent_step=3.50s eta=19:48:11 +2026-09-19 01:04:20 | INFO | train rank=0 | step=30740/50000 progress=61.5% loss=0.095009 lr=3.536912e-05 lr_backbone=3.536912e-05 elapsed=31:35:44 avg_step=3.70s recent_step=5.54s eta=19:47:45 +2026-09-19 01:04:47 | INFO | train rank=0 | step=30750/50000 progress=61.5% loss=0.120893 lr=3.533750e-05 lr_backbone=3.533750e-05 elapsed=31:36:10 avg_step=3.70s recent_step=2.66s eta=19:47:02 +2026-09-19 01:05:24 | INFO | train rank=0 | step=30760/50000 progress=61.5% loss=0.091132 lr=3.530589e-05 lr_backbone=3.530589e-05 elapsed=31:36:47 avg_step=3.70s recent_step=3.70s eta=19:46:25 +2026-09-19 01:05:55 | INFO | train rank=0 | step=30770/50000 progress=61.5% loss=0.111939 lr=3.527428e-05 lr_backbone=3.527428e-05 elapsed=31:37:18 avg_step=3.70s recent_step=3.06s eta=19:45:44 +2026-09-19 01:06:45 | INFO | train rank=0 | step=30780/50000 progress=61.6% loss=0.095250 lr=3.524268e-05 lr_backbone=3.524268e-05 elapsed=31:38:09 avg_step=3.70s recent_step=5.08s eta=19:45:15 +2026-09-19 01:07:15 | INFO | train rank=0 | step=30790/50000 progress=61.6% loss=0.075665 lr=3.521109e-05 lr_backbone=3.521109e-05 elapsed=31:38:38 avg_step=3.70s recent_step=2.93s eta=19:44:34 +2026-09-19 01:07:44 | INFO | train rank=0 | step=30800/50000 progress=61.6% loss=0.109075 lr=3.517950e-05 lr_backbone=3.517950e-05 elapsed=31:39:07 avg_step=3.70s recent_step=2.95s eta=19:43:52 +2026-09-19 01:08:21 | INFO | train rank=0 | step=30810/50000 progress=61.6% loss=0.093857 lr=3.514792e-05 lr_backbone=3.514792e-05 elapsed=31:39:44 avg_step=3.70s recent_step=3.64s eta=19:43:15 +2026-09-19 01:09:10 | INFO | train rank=0 | step=30820/50000 progress=61.6% loss=0.113439 lr=3.511635e-05 lr_backbone=3.511635e-05 elapsed=31:40:34 avg_step=3.70s recent_step=4.99s eta=19:42:46 +2026-09-19 01:09:48 | INFO | train rank=0 | step=30830/50000 progress=61.7% loss=0.100233 lr=3.508478e-05 lr_backbone=3.508478e-05 elapsed=31:41:11 avg_step=3.70s recent_step=3.76s eta=19:42:09 +2026-09-19 01:10:17 | INFO | train rank=0 | step=30840/50000 progress=61.7% loss=0.092277 lr=3.505322e-05 lr_backbone=3.505322e-05 elapsed=31:41:40 avg_step=3.70s recent_step=2.91s eta=19:41:27 +2026-09-19 01:10:40 | INFO | train rank=0 | step=30850/50000 progress=61.7% loss=0.084807 lr=3.502167e-05 lr_backbone=3.502167e-05 elapsed=31:42:04 avg_step=3.70s recent_step=2.34s eta=19:40:42 +2026-09-19 01:11:32 | INFO | train rank=0 | step=30860/50000 progress=61.7% loss=0.082628 lr=3.499012e-05 lr_backbone=3.499012e-05 elapsed=31:42:55 avg_step=3.70s recent_step=5.15s eta=19:40:14 +2026-09-19 01:12:11 | INFO | train rank=0 | step=30870/50000 progress=61.7% loss=0.107747 lr=3.495858e-05 lr_backbone=3.495858e-05 elapsed=31:43:34 avg_step=3.70s recent_step=3.89s eta=19:39:38 +2026-09-19 01:12:46 | INFO | train rank=0 | step=30880/50000 progress=61.8% loss=0.093546 lr=3.492704e-05 lr_backbone=3.492704e-05 elapsed=31:44:09 avg_step=3.70s recent_step=3.49s eta=19:38:59 +2026-09-19 01:13:22 | INFO | train rank=0 | step=30890/50000 progress=61.8% loss=0.104427 lr=3.489552e-05 lr_backbone=3.489552e-05 elapsed=31:44:46 avg_step=3.70s recent_step=3.65s eta=19:38:22 +2026-09-19 01:14:12 | INFO | train rank=0 | step=30900/50000 progress=61.8% loss=0.091812 lr=3.486399e-05 lr_backbone=3.486399e-05 elapsed=31:45:35 avg_step=3.70s recent_step=4.98s eta=19:37:53 +2026-09-19 01:14:41 | INFO | train rank=0 | step=30910/50000 progress=61.8% loss=0.092291 lr=3.483248e-05 lr_backbone=3.483248e-05 elapsed=31:46:04 avg_step=3.70s recent_step=2.89s eta=19:37:11 +2026-09-19 01:15:13 | INFO | train rank=0 | step=30920/50000 progress=61.8% loss=0.086620 lr=3.480097e-05 lr_backbone=3.480097e-05 elapsed=31:46:36 avg_step=3.70s recent_step=3.23s eta=19:36:31 +2026-09-19 01:15:37 | INFO | train rank=0 | step=30930/50000 progress=61.9% loss=0.097548 lr=3.476947e-05 lr_backbone=3.476947e-05 elapsed=31:47:00 avg_step=3.70s recent_step=2.38s eta=19:35:46 +2026-09-19 01:16:24 | INFO | train rank=0 | step=30940/50000 progress=61.9% loss=0.099635 lr=3.473798e-05 lr_backbone=3.473798e-05 elapsed=31:47:48 avg_step=3.70s recent_step=4.74s eta=19:35:15 +2026-09-19 01:16:52 | INFO | train rank=0 | step=30950/50000 progress=61.9% loss=0.067624 lr=3.470649e-05 lr_backbone=3.470649e-05 elapsed=31:48:15 avg_step=3.70s recent_step=2.74s eta=19:34:33 +2026-09-19 01:17:24 | INFO | train rank=0 | step=30960/50000 progress=61.9% loss=0.078475 lr=3.467501e-05 lr_backbone=3.467501e-05 elapsed=31:48:47 avg_step=3.70s recent_step=3.19s eta=19:33:52 +2026-09-19 01:18:00 | INFO | train rank=0 | step=30970/50000 progress=61.9% loss=0.078137 lr=3.464353e-05 lr_backbone=3.464353e-05 elapsed=31:49:24 avg_step=3.70s recent_step=3.65s eta=19:33:15 +2026-09-19 01:18:42 | INFO | train rank=0 | step=30980/50000 progress=62.0% loss=0.089183 lr=3.461207e-05 lr_backbone=3.461207e-05 elapsed=31:50:05 avg_step=3.70s recent_step=4.14s eta=19:32:41 +2026-09-19 01:19:09 | INFO | train rank=0 | step=30990/50000 progress=62.0% loss=0.093196 lr=3.458060e-05 lr_backbone=3.458060e-05 elapsed=31:50:32 avg_step=3.70s recent_step=2.74s eta=19:31:58 +2026-09-19 01:19:40 | INFO | train rank=0 | step=31000/50000 progress=62.0% loss=0.092192 lr=3.454915e-05 lr_backbone=3.454915e-05 elapsed=31:51:03 avg_step=3.70s recent_step=3.05s eta=19:31:17 +2026-09-19 01:20:27 | INFO | train rank=0 | step=31010/50000 progress=62.0% loss=0.095660 lr=3.451770e-05 lr_backbone=3.451770e-05 elapsed=31:51:51 avg_step=3.70s recent_step=4.79s eta=19:30:47 +2026-09-19 01:21:05 | INFO | train rank=0 | step=31020/50000 progress=62.0% loss=0.094207 lr=3.448626e-05 lr_backbone=3.448626e-05 elapsed=31:52:29 avg_step=3.70s recent_step=3.80s eta=19:30:10 +2026-09-19 01:21:32 | INFO | train rank=0 | step=31030/50000 progress=62.1% loss=0.080963 lr=3.445483e-05 lr_backbone=3.445483e-05 elapsed=31:52:55 avg_step=3.70s recent_step=2.64s eta=19:29:27 +2026-09-19 01:22:05 | INFO | train rank=0 | step=31040/50000 progress=62.1% loss=0.087774 lr=3.442340e-05 lr_backbone=3.442340e-05 elapsed=31:53:28 avg_step=3.70s recent_step=3.30s eta=19:28:47 +2026-09-19 01:22:52 | INFO | train rank=0 | step=31050/50000 progress=62.1% loss=0.114312 lr=3.439198e-05 lr_backbone=3.439198e-05 elapsed=31:54:15 avg_step=3.70s recent_step=4.74s eta=19:28:17 +2026-09-19 01:23:29 | INFO | train rank=0 | step=31060/50000 progress=62.1% loss=0.068953 lr=3.436057e-05 lr_backbone=3.436057e-05 elapsed=31:54:52 avg_step=3.70s recent_step=3.68s eta=19:27:40 +2026-09-19 01:24:09 | INFO | train rank=0 | step=31070/50000 progress=62.1% loss=0.082840 lr=3.432916e-05 lr_backbone=3.432916e-05 elapsed=31:55:33 avg_step=3.70s recent_step=4.04s eta=19:27:05 +2026-09-19 01:24:44 | INFO | train rank=0 | step=31080/50000 progress=62.2% loss=0.106456 lr=3.429776e-05 lr_backbone=3.429776e-05 elapsed=31:56:08 avg_step=3.70s recent_step=3.50s eta=19:26:27 +2026-09-19 01:25:31 | INFO | train rank=0 | step=31090/50000 progress=62.2% loss=0.092337 lr=3.426637e-05 lr_backbone=3.426637e-05 elapsed=31:56:54 avg_step=3.70s recent_step=4.68s eta=19:25:55 +2026-09-19 01:26:10 | INFO | train rank=0 | step=31100/50000 progress=62.2% loss=0.102937 lr=3.423498e-05 lr_backbone=3.423498e-05 elapsed=31:57:34 avg_step=3.70s recent_step=3.94s eta=19:25:20 +2026-09-19 01:26:42 | INFO | train rank=0 | step=31110/50000 progress=62.2% loss=0.082319 lr=3.420360e-05 lr_backbone=3.420360e-05 elapsed=31:58:05 avg_step=3.70s recent_step=3.16s eta=19:24:40 +2026-09-19 01:27:20 | INFO | train rank=0 | step=31120/50000 progress=62.2% loss=0.080314 lr=3.417223e-05 lr_backbone=3.417223e-05 elapsed=31:58:43 avg_step=3.70s recent_step=3.81s eta=19:24:03 +2026-09-19 01:27:56 | INFO | train rank=0 | step=31130/50000 progress=62.3% loss=0.075900 lr=3.414087e-05 lr_backbone=3.414087e-05 elapsed=31:59:19 avg_step=3.70s recent_step=3.59s eta=19:23:26 +2026-09-19 01:28:26 | INFO | train rank=0 | step=31140/50000 progress=62.3% loss=0.075276 lr=3.410951e-05 lr_backbone=3.410951e-05 elapsed=31:59:50 avg_step=3.70s recent_step=3.04s eta=19:22:45 +2026-09-19 01:28:58 | INFO | train rank=0 | step=31150/50000 progress=62.3% loss=0.087099 lr=3.407816e-05 lr_backbone=3.407816e-05 elapsed=32:00:21 avg_step=3.70s recent_step=3.12s eta=19:22:04 +2026-09-19 01:29:58 | INFO | train rank=0 | step=31160/50000 progress=62.3% loss=0.084228 lr=3.404681e-05 lr_backbone=3.404681e-05 elapsed=32:01:21 avg_step=3.70s recent_step=6.06s eta=19:21:41 +2026-09-19 01:30:29 | INFO | train rank=0 | step=31170/50000 progress=62.3% loss=0.079074 lr=3.401547e-05 lr_backbone=3.401547e-05 elapsed=32:01:52 avg_step=3.70s recent_step=3.06s eta=19:21:01 +2026-09-19 01:31:04 | INFO | train rank=0 | step=31180/50000 progress=62.4% loss=0.080446 lr=3.398414e-05 lr_backbone=3.398414e-05 elapsed=32:02:28 avg_step=3.70s recent_step=3.55s eta=19:20:23 +2026-09-19 01:31:39 | INFO | train rank=0 | step=31190/50000 progress=62.4% loss=0.077900 lr=3.395282e-05 lr_backbone=3.395282e-05 elapsed=32:03:02 avg_step=3.70s recent_step=3.49s eta=19:19:44 +2026-09-19 01:32:31 | INFO | train rank=0 | step=31200/50000 progress=62.4% loss=0.100585 lr=3.392150e-05 lr_backbone=3.392150e-05 elapsed=32:03:54 avg_step=3.70s recent_step=5.15s eta=19:19:16 +2026-09-19 01:33:02 | INFO | train rank=0 | step=31210/50000 progress=62.4% loss=0.081509 lr=3.389019e-05 lr_backbone=3.389019e-05 elapsed=32:04:25 avg_step=3.70s recent_step=3.15s eta=19:18:36 +2026-09-19 01:33:33 | INFO | train rank=0 | step=31220/50000 progress=62.4% loss=0.082272 lr=3.385889e-05 lr_backbone=3.385889e-05 elapsed=32:04:56 avg_step=3.70s recent_step=3.08s eta=19:17:55 +2026-09-19 01:34:09 | INFO | train rank=0 | step=31230/50000 progress=62.5% loss=0.071471 lr=3.382760e-05 lr_backbone=3.382760e-05 elapsed=32:05:32 avg_step=3.70s recent_step=3.61s eta=19:17:18 +2026-09-19 01:34:58 | INFO | train rank=0 | step=31240/50000 progress=62.5% loss=0.088165 lr=3.379631e-05 lr_backbone=3.379631e-05 elapsed=32:06:22 avg_step=3.70s recent_step=4.93s eta=19:16:48 +2026-09-19 01:35:24 | INFO | train rank=0 | step=31250/50000 progress=62.5% loss=0.088530 lr=3.376503e-05 lr_backbone=3.376503e-05 elapsed=32:06:48 avg_step=3.70s recent_step=2.59s eta=19:16:04 +2026-09-19 01:35:57 | INFO | train rank=0 | step=31260/50000 progress=62.5% loss=0.066754 lr=3.373375e-05 lr_backbone=3.373375e-05 elapsed=32:07:20 avg_step=3.70s recent_step=3.25s eta=19:15:25 +2026-09-19 01:36:30 | INFO | train rank=0 | step=31270/50000 progress=62.5% loss=0.090390 lr=3.370249e-05 lr_backbone=3.370249e-05 elapsed=32:07:53 avg_step=3.70s recent_step=3.30s eta=19:14:45 +2026-09-19 01:37:25 | INFO | train rank=0 | step=31280/50000 progress=62.6% loss=0.092462 lr=3.367123e-05 lr_backbone=3.367123e-05 elapsed=32:08:48 avg_step=3.70s recent_step=5.49s eta=19:14:19 +2026-09-19 01:37:59 | INFO | train rank=0 | step=31290/50000 progress=62.6% loss=0.081114 lr=3.363997e-05 lr_backbone=3.363997e-05 elapsed=32:09:22 avg_step=3.70s recent_step=3.39s eta=19:13:40 +2026-09-19 01:38:34 | INFO | train rank=0 | step=31300/50000 progress=62.6% loss=0.096918 lr=3.360873e-05 lr_backbone=3.360873e-05 elapsed=32:09:57 avg_step=3.70s recent_step=3.55s eta=19:13:02 +2026-09-19 01:39:14 | INFO | train rank=0 | step=31310/50000 progress=62.6% loss=0.063668 lr=3.357749e-05 lr_backbone=3.357749e-05 elapsed=32:10:37 avg_step=3.70s recent_step=4.00s eta=19:12:27 +2026-09-19 01:39:57 | INFO | train rank=0 | step=31320/50000 progress=62.6% loss=0.084011 lr=3.354626e-05 lr_backbone=3.354626e-05 elapsed=32:11:21 avg_step=3.70s recent_step=4.32s eta=19:11:54 +2026-09-19 01:40:29 | INFO | train rank=0 | step=31330/50000 progress=62.7% loss=0.120467 lr=3.351503e-05 lr_backbone=3.351503e-05 elapsed=32:11:53 avg_step=3.70s recent_step=3.21s eta=19:11:14 +2026-09-19 01:41:03 | INFO | train rank=0 | step=31340/50000 progress=62.7% loss=0.105046 lr=3.348382e-05 lr_backbone=3.348382e-05 elapsed=32:12:26 avg_step=3.70s recent_step=3.38s eta=19:10:35 +2026-09-19 01:41:45 | INFO | train rank=0 | step=31350/50000 progress=62.7% loss=0.108557 lr=3.345261e-05 lr_backbone=3.345261e-05 elapsed=32:13:09 avg_step=3.70s recent_step=4.21s eta=19:10:01 +2026-09-19 01:42:30 | INFO | train rank=0 | step=31360/50000 progress=62.7% loss=0.112955 lr=3.342141e-05 lr_backbone=3.342141e-05 elapsed=32:13:53 avg_step=3.70s recent_step=4.45s eta=19:09:28 +2026-09-19 01:42:52 | INFO | train rank=0 | step=31370/50000 progress=62.7% loss=0.088696 lr=3.339021e-05 lr_backbone=3.339021e-05 elapsed=32:14:15 avg_step=3.70s recent_step=2.23s eta=19:08:43 +2026-09-19 01:43:28 | INFO | train rank=0 | step=31380/50000 progress=62.8% loss=0.097285 lr=3.335902e-05 lr_backbone=3.335902e-05 elapsed=32:14:51 avg_step=3.70s recent_step=3.56s eta=19:08:05 +2026-09-19 01:44:10 | INFO | train rank=0 | step=31390/50000 progress=62.8% loss=0.094362 lr=3.332784e-05 lr_backbone=3.332784e-05 elapsed=32:15:33 avg_step=3.70s recent_step=4.20s eta=19:07:31 +2026-09-19 01:44:56 | INFO | train rank=0 | step=31400/50000 progress=62.8% loss=0.110631 lr=3.329667e-05 lr_backbone=3.329667e-05 elapsed=32:16:19 avg_step=3.70s recent_step=4.58s eta=19:06:59 +2026-09-19 01:45:27 | INFO | train rank=0 | step=31410/50000 progress=62.8% loss=0.070111 lr=3.326550e-05 lr_backbone=3.326550e-05 elapsed=32:16:50 avg_step=3.70s recent_step=3.14s eta=19:06:19 +2026-09-19 01:46:04 | INFO | train rank=0 | step=31420/50000 progress=62.8% loss=0.125199 lr=3.323434e-05 lr_backbone=3.323434e-05 elapsed=32:17:28 avg_step=3.70s recent_step=3.75s eta=19:05:42 +2026-09-19 01:46:38 | INFO | train rank=0 | step=31430/50000 progress=62.9% loss=0.092380 lr=3.320319e-05 lr_backbone=3.320319e-05 elapsed=32:18:02 avg_step=3.70s recent_step=3.38s eta=19:05:03 +2026-09-19 01:47:29 | INFO | train rank=0 | step=31440/50000 progress=62.9% loss=0.078133 lr=3.317205e-05 lr_backbone=3.317205e-05 elapsed=32:18:52 avg_step=3.70s recent_step=5.08s eta=19:04:34 +2026-09-19 01:48:03 | INFO | train rank=0 | step=31450/50000 progress=62.9% loss=0.080182 lr=3.314091e-05 lr_backbone=3.314091e-05 elapsed=32:19:26 avg_step=3.70s recent_step=3.35s eta=19:03:55 +2026-09-19 01:48:34 | INFO | train rank=0 | step=31460/50000 progress=62.9% loss=0.073793 lr=3.310978e-05 lr_backbone=3.310978e-05 elapsed=32:19:58 avg_step=3.70s recent_step=3.18s eta=19:03:15 +2026-09-19 01:49:08 | INFO | train rank=0 | step=31470/50000 progress=62.9% loss=0.105540 lr=3.307866e-05 lr_backbone=3.307866e-05 elapsed=32:20:31 avg_step=3.70s recent_step=3.33s eta=19:02:36 +2026-09-19 01:49:55 | INFO | train rank=0 | step=31480/50000 progress=63.0% loss=0.111370 lr=3.304755e-05 lr_backbone=3.304755e-05 elapsed=32:21:19 avg_step=3.70s recent_step=4.76s eta=19:02:05 +2026-09-19 01:50:28 | INFO | train rank=0 | step=31490/50000 progress=63.0% loss=0.115021 lr=3.301644e-05 lr_backbone=3.301644e-05 elapsed=32:21:51 avg_step=3.70s recent_step=3.27s eta=19:01:26 +2026-09-19 01:50:58 | INFO | train rank=0 | step=31500/50000 progress=63.0% loss=0.087257 lr=3.298534e-05 lr_backbone=3.298534e-05 elapsed=32:22:22 avg_step=3.70s recent_step=3.04s eta=19:00:45 +2026-09-19 01:51:37 | INFO | train rank=0 | step=31510/50000 progress=63.0% loss=0.083161 lr=3.295425e-05 lr_backbone=3.295425e-05 elapsed=32:23:00 avg_step=3.70s recent_step=3.86s eta=19:00:09 +2026-09-19 01:52:14 | INFO | train rank=0 | step=31520/50000 progress=63.0% loss=0.068370 lr=3.292316e-05 lr_backbone=3.292316e-05 elapsed=32:23:37 avg_step=3.70s recent_step=3.70s eta=18:59:32 +2026-09-19 01:52:53 | INFO | train rank=0 | step=31530/50000 progress=63.1% loss=0.096689 lr=3.289209e-05 lr_backbone=3.289209e-05 elapsed=32:24:16 avg_step=3.70s recent_step=3.88s eta=18:58:56 +2026-09-19 01:53:38 | INFO | train rank=0 | step=31540/50000 progress=63.1% loss=0.110464 lr=3.286102e-05 lr_backbone=3.286102e-05 elapsed=32:25:02 avg_step=3.70s recent_step=4.55s eta=18:58:24 +2026-09-19 01:54:02 | INFO | train rank=0 | step=31550/50000 progress=63.1% loss=0.081961 lr=3.282996e-05 lr_backbone=3.282996e-05 elapsed=32:25:26 avg_step=3.70s recent_step=2.40s eta=18:57:39 +2026-09-19 01:54:55 | INFO | train rank=0 | step=31560/50000 progress=63.1% loss=0.076326 lr=3.279890e-05 lr_backbone=3.279890e-05 elapsed=32:26:18 avg_step=3.70s recent_step=5.22s eta=18:57:11 +2026-09-19 01:55:26 | INFO | train rank=0 | step=31570/50000 progress=63.1% loss=0.071459 lr=3.276785e-05 lr_backbone=3.276785e-05 elapsed=32:26:50 avg_step=3.70s recent_step=3.19s eta=18:56:31 +2026-09-19 01:56:04 | INFO | train rank=0 | step=31580/50000 progress=63.2% loss=0.071779 lr=3.273681e-05 lr_backbone=3.273681e-05 elapsed=32:27:27 avg_step=3.70s recent_step=3.77s eta=18:55:55 +2026-09-19 01:56:44 | INFO | train rank=0 | step=31590/50000 progress=63.2% loss=0.162535 lr=3.270578e-05 lr_backbone=3.270578e-05 elapsed=32:28:07 avg_step=3.70s recent_step=3.95s eta=18:55:19 +2026-09-19 01:57:15 | INFO | train rank=0 | step=31600/50000 progress=63.2% loss=0.076032 lr=3.267476e-05 lr_backbone=3.267476e-05 elapsed=32:28:38 avg_step=3.70s recent_step=3.10s eta=18:54:39 +2026-09-19 01:57:49 | INFO | train rank=0 | step=31610/50000 progress=63.2% loss=0.096069 lr=3.264374e-05 lr_backbone=3.264374e-05 elapsed=32:29:13 avg_step=3.70s recent_step=3.46s eta=18:54:00 +2026-09-19 01:58:29 | INFO | train rank=0 | step=31620/50000 progress=63.2% loss=0.111617 lr=3.261273e-05 lr_backbone=3.261273e-05 elapsed=32:29:52 avg_step=3.70s recent_step=3.94s eta=18:53:25 +2026-09-19 01:59:10 | INFO | train rank=0 | step=31630/50000 progress=63.3% loss=0.081887 lr=3.258173e-05 lr_backbone=3.258173e-05 elapsed=32:30:33 avg_step=3.70s recent_step=4.10s eta=18:52:50 +2026-09-19 01:59:42 | INFO | train rank=0 | step=31640/50000 progress=63.3% loss=0.137351 lr=3.255074e-05 lr_backbone=3.255074e-05 elapsed=32:31:06 avg_step=3.70s recent_step=3.25s eta=18:52:10 +2026-09-19 02:00:21 | INFO | train rank=0 | step=31650/50000 progress=63.3% loss=0.088312 lr=3.251975e-05 lr_backbone=3.251975e-05 elapsed=32:31:44 avg_step=3.70s recent_step=3.84s eta=18:51:34 +2026-09-19 02:01:07 | INFO | train rank=0 | step=31660/50000 progress=63.3% loss=0.098190 lr=3.248877e-05 lr_backbone=3.248877e-05 elapsed=32:32:30 avg_step=3.70s recent_step=4.63s eta=18:51:03 +2026-09-19 02:01:35 | INFO | train rank=0 | step=31670/50000 progress=63.3% loss=0.085942 lr=3.245780e-05 lr_backbone=3.245780e-05 elapsed=32:32:58 avg_step=3.70s recent_step=2.79s eta=18:50:20 +2026-09-19 02:02:06 | INFO | train rank=0 | step=31680/50000 progress=63.4% loss=0.083489 lr=3.242684e-05 lr_backbone=3.242684e-05 elapsed=32:33:29 avg_step=3.70s recent_step=3.12s eta=18:49:40 +2026-09-19 02:02:43 | INFO | train rank=0 | step=31690/50000 progress=63.4% loss=0.081163 lr=3.239588e-05 lr_backbone=3.239588e-05 elapsed=32:34:07 avg_step=3.70s recent_step=3.74s eta=18:49:03 +2026-09-19 02:03:37 | INFO | train rank=0 | step=31700/50000 progress=63.4% loss=0.085984 lr=3.236493e-05 lr_backbone=3.236493e-05 elapsed=32:35:01 avg_step=3.70s recent_step=5.40s eta=18:48:36 +2026-09-19 02:04:05 | INFO | train rank=0 | step=31710/50000 progress=63.4% loss=0.123581 lr=3.233399e-05 lr_backbone=3.233399e-05 elapsed=32:35:29 avg_step=3.70s recent_step=2.79s eta=18:47:54 +2026-09-19 02:04:41 | INFO | train rank=0 | step=31720/50000 progress=63.4% loss=0.140230 lr=3.230306e-05 lr_backbone=3.230306e-05 elapsed=32:36:04 avg_step=3.70s recent_step=3.59s eta=18:47:16 +2026-09-19 02:05:12 | INFO | train rank=0 | step=31730/50000 progress=63.5% loss=0.091009 lr=3.227213e-05 lr_backbone=3.227213e-05 elapsed=32:36:35 avg_step=3.70s recent_step=3.06s eta=18:46:35 +2026-09-19 02:06:11 | INFO | train rank=0 | step=31740/50000 progress=63.5% loss=0.079413 lr=3.224122e-05 lr_backbone=3.224122e-05 elapsed=32:37:34 avg_step=3.70s recent_step=5.90s eta=18:46:11 +2026-09-19 02:06:39 | INFO | train rank=0 | step=31750/50000 progress=63.5% loss=0.079120 lr=3.221031e-05 lr_backbone=3.221031e-05 elapsed=32:38:02 avg_step=3.70s recent_step=2.78s eta=18:45:29 +2026-09-19 02:07:06 | INFO | train rank=0 | step=31760/50000 progress=63.5% loss=0.081195 lr=3.217941e-05 lr_backbone=3.217941e-05 elapsed=32:38:29 avg_step=3.70s recent_step=2.71s eta=18:44:46 +2026-09-19 02:07:52 | INFO | train rank=0 | step=31770/50000 progress=63.5% loss=0.079954 lr=3.214851e-05 lr_backbone=3.214851e-05 elapsed=32:39:15 avg_step=3.70s recent_step=4.59s eta=18:44:14 +2026-09-19 02:08:29 | INFO | train rank=0 | step=31780/50000 progress=63.6% loss=0.090039 lr=3.211763e-05 lr_backbone=3.211763e-05 elapsed=32:39:52 avg_step=3.70s recent_step=3.70s eta=18:43:37 +2026-09-19 02:09:07 | INFO | train rank=0 | step=31790/50000 progress=63.6% loss=0.093766 lr=3.208675e-05 lr_backbone=3.208675e-05 elapsed=32:40:30 avg_step=3.70s recent_step=3.83s eta=18:43:01 +2026-09-19 02:09:43 | INFO | train rank=0 | step=31800/50000 progress=63.6% loss=0.090718 lr=3.205588e-05 lr_backbone=3.205588e-05 elapsed=32:41:06 avg_step=3.70s recent_step=3.62s eta=18:42:23 +2026-09-19 02:10:20 | INFO | train rank=0 | step=31810/50000 progress=63.6% loss=0.109621 lr=3.202502e-05 lr_backbone=3.202502e-05 elapsed=32:41:43 avg_step=3.70s recent_step=3.70s eta=18:41:46 +2026-09-19 02:10:49 | INFO | train rank=0 | step=31820/50000 progress=63.6% loss=0.085369 lr=3.199416e-05 lr_backbone=3.199416e-05 elapsed=32:42:12 avg_step=3.70s recent_step=2.90s eta=18:41:05 +2026-09-19 02:11:44 | INFO | train rank=0 | step=31830/50000 progress=63.7% loss=0.079460 lr=3.196331e-05 lr_backbone=3.196331e-05 elapsed=32:43:07 avg_step=3.70s recent_step=5.47s eta=18:40:38 +2026-09-19 02:12:21 | INFO | train rank=0 | step=31840/50000 progress=63.7% loss=0.089219 lr=3.193248e-05 lr_backbone=3.193248e-05 elapsed=32:43:45 avg_step=3.70s recent_step=3.74s eta=18:40:01 +2026-09-19 02:12:50 | INFO | train rank=0 | step=31850/50000 progress=63.7% loss=0.092410 lr=3.190164e-05 lr_backbone=3.190164e-05 elapsed=32:44:13 avg_step=3.70s recent_step=2.88s eta=18:39:20 +2026-09-19 02:13:18 | INFO | train rank=0 | step=31860/50000 progress=63.7% loss=0.099503 lr=3.187082e-05 lr_backbone=3.187082e-05 elapsed=32:44:41 avg_step=3.70s recent_step=2.77s eta=18:38:37 +2026-09-19 02:14:11 | INFO | train rank=0 | step=31870/50000 progress=63.7% loss=0.094550 lr=3.184001e-05 lr_backbone=3.184001e-05 elapsed=32:45:34 avg_step=3.70s recent_step=5.32s eta=18:38:09 +2026-09-19 02:14:50 | INFO | train rank=0 | step=31880/50000 progress=63.8% loss=0.096393 lr=3.180920e-05 lr_backbone=3.180920e-05 elapsed=32:46:13 avg_step=3.70s recent_step=3.89s eta=18:37:34 +2026-09-19 02:15:18 | INFO | train rank=0 | step=31890/50000 progress=63.8% loss=0.133997 lr=3.177840e-05 lr_backbone=3.177840e-05 elapsed=32:46:41 avg_step=3.70s recent_step=2.83s eta=18:36:52 +2026-09-19 02:15:54 | INFO | train rank=0 | step=31900/50000 progress=63.8% loss=0.090189 lr=3.174761e-05 lr_backbone=3.174761e-05 elapsed=32:47:17 avg_step=3.70s recent_step=3.54s eta=18:36:14 +2026-09-19 02:16:43 | INFO | train rank=0 | step=31910/50000 progress=63.8% loss=0.074344 lr=3.171683e-05 lr_backbone=3.171683e-05 elapsed=32:48:07 avg_step=3.70s recent_step=4.98s eta=18:35:44 +2026-09-19 02:17:14 | INFO | train rank=0 | step=31920/50000 progress=63.8% loss=0.111988 lr=3.168605e-05 lr_backbone=3.168605e-05 elapsed=32:48:37 avg_step=3.70s recent_step=3.04s eta=18:35:03 +2026-09-19 02:17:54 | INFO | train rank=0 | step=31930/50000 progress=63.9% loss=0.078544 lr=3.165528e-05 lr_backbone=3.165528e-05 elapsed=32:49:17 avg_step=3.70s recent_step=4.03s eta=18:34:28 +2026-09-19 02:18:21 | INFO | train rank=0 | step=31940/50000 progress=63.9% loss=0.091740 lr=3.162452e-05 lr_backbone=3.162452e-05 elapsed=32:49:44 avg_step=3.70s recent_step=2.67s eta=18:33:45 +2026-09-19 02:19:16 | INFO | train rank=0 | step=31950/50000 progress=63.9% loss=0.110035 lr=3.159377e-05 lr_backbone=3.159377e-05 elapsed=32:50:39 avg_step=3.70s recent_step=5.56s eta=18:33:19 +2026-09-19 02:19:43 | INFO | train rank=0 | step=31960/50000 progress=63.9% loss=0.115807 lr=3.156303e-05 lr_backbone=3.156303e-05 elapsed=32:51:06 avg_step=3.70s recent_step=2.66s eta=18:32:36 +2026-09-19 02:20:08 | INFO | train rank=0 | step=31970/50000 progress=63.9% loss=0.074840 lr=3.153229e-05 lr_backbone=3.153229e-05 elapsed=32:51:32 avg_step=3.70s recent_step=2.56s eta=18:31:52 +2026-09-19 02:20:41 | INFO | train rank=0 | step=31980/50000 progress=64.0% loss=0.086693 lr=3.150157e-05 lr_backbone=3.150157e-05 elapsed=32:52:05 avg_step=3.70s recent_step=3.31s eta=18:31:13 +2026-09-19 02:21:22 | INFO | train rank=0 | step=31990/50000 progress=64.0% loss=0.098815 lr=3.147085e-05 lr_backbone=3.147085e-05 elapsed=32:52:45 avg_step=3.70s recent_step=4.01s eta=18:30:38 +2026-09-19 02:21:54 | INFO | train rank=0 | step=32000/50000 progress=64.0% loss=0.071917 lr=3.144014e-05 lr_backbone=3.144014e-05 elapsed=32:53:18 avg_step=3.70s recent_step=3.28s eta=18:29:58 +2026-09-19 02:22:27 | INFO | train rank=0 | step=32010/50000 progress=64.0% loss=0.087437 lr=3.140943e-05 lr_backbone=3.140943e-05 elapsed=32:53:51 avg_step=3.70s recent_step=3.30s eta=18:29:19 +2026-09-19 02:22:58 | INFO | train rank=0 | step=32020/50000 progress=64.0% loss=0.083204 lr=3.137874e-05 lr_backbone=3.137874e-05 elapsed=32:54:22 avg_step=3.70s recent_step=3.10s eta=18:28:39 +2026-09-19 02:23:51 | INFO | train rank=0 | step=32030/50000 progress=64.1% loss=0.087941 lr=3.134805e-05 lr_backbone=3.134805e-05 elapsed=32:55:15 avg_step=3.70s recent_step=5.31s eta=18:28:11 +2026-09-19 02:24:16 | INFO | train rank=0 | step=32040/50000 progress=64.1% loss=0.083005 lr=3.131738e-05 lr_backbone=3.131738e-05 elapsed=32:55:39 avg_step=3.70s recent_step=2.41s eta=18:27:27 +2026-09-19 02:24:53 | INFO | train rank=0 | step=32050/50000 progress=64.1% loss=0.071492 lr=3.128671e-05 lr_backbone=3.128671e-05 elapsed=32:56:16 avg_step=3.70s recent_step=3.72s eta=18:26:50 +2026-09-19 02:25:24 | INFO | train rank=0 | step=32060/50000 progress=64.1% loss=0.081690 lr=3.125604e-05 lr_backbone=3.125604e-05 elapsed=32:56:47 avg_step=3.70s recent_step=3.09s eta=18:26:09 +2026-09-19 02:26:09 | INFO | train rank=0 | step=32070/50000 progress=64.1% loss=0.110677 lr=3.122539e-05 lr_backbone=3.122539e-05 elapsed=32:57:33 avg_step=3.70s recent_step=4.57s eta=18:25:37 +2026-09-19 02:26:47 | INFO | train rank=0 | step=32080/50000 progress=64.2% loss=0.063402 lr=3.119474e-05 lr_backbone=3.119474e-05 elapsed=32:58:10 avg_step=3.70s recent_step=3.74s eta=18:25:00 +2026-09-19 02:27:22 | INFO | train rank=0 | step=32090/50000 progress=64.2% loss=0.100368 lr=3.116411e-05 lr_backbone=3.116411e-05 elapsed=32:58:45 avg_step=3.70s recent_step=3.49s eta=18:24:22 +2026-09-19 02:27:52 | INFO | train rank=0 | step=32100/50000 progress=64.2% loss=0.087683 lr=3.113348e-05 lr_backbone=3.113348e-05 elapsed=32:59:15 avg_step=3.70s recent_step=3.03s eta=18:23:42 +2026-09-19 02:28:46 | INFO | train rank=0 | step=32110/50000 progress=64.2% loss=0.086736 lr=3.110286e-05 lr_backbone=3.110286e-05 elapsed=33:00:10 avg_step=3.70s recent_step=5.44s eta=18:23:14 +2026-09-19 02:29:18 | INFO | train rank=0 | step=32120/50000 progress=64.2% loss=0.115870 lr=3.107225e-05 lr_backbone=3.107225e-05 elapsed=33:00:42 avg_step=3.70s recent_step=3.19s eta=18:22:34 +2026-09-19 02:29:55 | INFO | train rank=0 | step=32130/50000 progress=64.3% loss=0.108634 lr=3.104164e-05 lr_backbone=3.104164e-05 elapsed=33:01:18 avg_step=3.70s recent_step=3.63s eta=18:21:57 +2026-09-19 02:30:25 | INFO | train rank=0 | step=32140/50000 progress=64.3% loss=0.093609 lr=3.101105e-05 lr_backbone=3.101105e-05 elapsed=33:01:48 avg_step=3.70s recent_step=3.06s eta=18:21:16 +2026-09-19 02:31:15 | INFO | train rank=0 | step=32150/50000 progress=64.3% loss=0.097556 lr=3.098046e-05 lr_backbone=3.098046e-05 elapsed=33:02:39 avg_step=3.70s recent_step=5.03s eta=18:20:47 +2026-09-19 02:31:46 | INFO | train rank=0 | step=32160/50000 progress=64.3% loss=0.075259 lr=3.094988e-05 lr_backbone=3.094988e-05 elapsed=33:03:09 avg_step=3.70s recent_step=3.04s eta=18:20:06 +2026-09-19 02:32:22 | INFO | train rank=0 | step=32170/50000 progress=64.3% loss=0.096053 lr=3.091931e-05 lr_backbone=3.091931e-05 elapsed=33:03:45 avg_step=3.70s recent_step=3.61s eta=18:19:29 +2026-09-19 02:33:01 | INFO | train rank=0 | step=32180/50000 progress=64.4% loss=0.098577 lr=3.088874e-05 lr_backbone=3.088874e-05 elapsed=33:04:24 avg_step=3.70s recent_step=3.87s eta=18:18:53 +2026-09-19 02:33:57 | INFO | train rank=0 | step=32190/50000 progress=64.4% loss=0.088637 lr=3.085819e-05 lr_backbone=3.085819e-05 elapsed=33:05:20 avg_step=3.70s recent_step=5.61s eta=18:18:26 +2026-09-19 02:34:26 | INFO | train rank=0 | step=32200/50000 progress=64.4% loss=0.098991 lr=3.082764e-05 lr_backbone=3.082764e-05 elapsed=33:05:50 avg_step=3.70s recent_step=2.97s eta=18:17:45 +2026-09-19 02:35:03 | INFO | train rank=0 | step=32210/50000 progress=64.4% loss=0.129885 lr=3.079711e-05 lr_backbone=3.079711e-05 elapsed=33:06:27 avg_step=3.70s recent_step=3.69s eta=18:17:08 +2026-09-19 02:35:35 | INFO | train rank=0 | step=32220/50000 progress=64.4% loss=0.088768 lr=3.076658e-05 lr_backbone=3.076658e-05 elapsed=33:06:58 avg_step=3.70s recent_step=3.18s eta=18:16:28 +2026-09-19 02:36:20 | INFO | train rank=0 | step=32230/50000 progress=64.5% loss=0.088811 lr=3.073606e-05 lr_backbone=3.073606e-05 elapsed=33:07:43 avg_step=3.70s recent_step=4.49s eta=18:15:56 +2026-09-19 02:36:55 | INFO | train rank=0 | step=32240/50000 progress=64.5% loss=0.086979 lr=3.070555e-05 lr_backbone=3.070555e-05 elapsed=33:08:19 avg_step=3.70s recent_step=3.55s eta=18:15:18 +2026-09-19 02:37:27 | INFO | train rank=0 | step=32250/50000 progress=64.5% loss=0.079363 lr=3.067504e-05 lr_backbone=3.067504e-05 elapsed=33:08:50 avg_step=3.70s recent_step=3.13s eta=18:14:38 +2026-09-19 02:37:59 | INFO | train rank=0 | step=32260/50000 progress=64.5% loss=0.072912 lr=3.064455e-05 lr_backbone=3.064455e-05 elapsed=33:09:23 avg_step=3.70s recent_step=3.27s eta=18:13:58 +2026-09-19 02:38:42 | INFO | train rank=0 | step=32270/50000 progress=64.5% loss=0.104235 lr=3.061406e-05 lr_backbone=3.061406e-05 elapsed=33:10:06 avg_step=3.70s recent_step=4.30s eta=18:13:24 +2026-09-19 02:39:17 | INFO | train rank=0 | step=32280/50000 progress=64.6% loss=0.084197 lr=3.058358e-05 lr_backbone=3.058358e-05 elapsed=33:10:41 avg_step=3.70s recent_step=3.48s eta=18:12:46 +2026-09-19 02:39:49 | INFO | train rank=0 | step=32290/50000 progress=64.6% loss=0.094186 lr=3.055311e-05 lr_backbone=3.055311e-05 elapsed=33:11:12 avg_step=3.70s recent_step=3.13s eta=18:12:06 +2026-09-19 02:40:24 | INFO | train rank=0 | step=32300/50000 progress=64.6% loss=0.076852 lr=3.052265e-05 lr_backbone=3.052265e-05 elapsed=33:11:47 avg_step=3.70s recent_step=3.55s eta=18:11:28 +2026-09-19 02:41:04 | INFO | train rank=0 | step=32310/50000 progress=64.6% loss=0.079862 lr=3.049220e-05 lr_backbone=3.049220e-05 elapsed=33:12:27 avg_step=3.70s recent_step=3.98s eta=18:10:53 +2026-09-19 02:41:58 | INFO | train rank=0 | step=32320/50000 progress=64.6% loss=0.086987 lr=3.046175e-05 lr_backbone=3.046175e-05 elapsed=33:13:21 avg_step=3.70s recent_step=5.42s eta=18:10:25 +2026-09-19 02:42:33 | INFO | train rank=0 | step=32330/50000 progress=64.7% loss=0.080872 lr=3.043132e-05 lr_backbone=3.043132e-05 elapsed=33:13:56 avg_step=3.70s recent_step=3.50s eta=18:09:47 +2026-09-19 02:43:06 | INFO | train rank=0 | step=32340/50000 progress=64.7% loss=0.066828 lr=3.040089e-05 lr_backbone=3.040089e-05 elapsed=33:14:29 avg_step=3.70s recent_step=3.29s eta=18:09:08 +2026-09-19 02:43:38 | INFO | train rank=0 | step=32350/50000 progress=64.7% loss=0.092315 lr=3.037047e-05 lr_backbone=3.037047e-05 elapsed=33:15:01 avg_step=3.70s recent_step=3.19s eta=18:08:28 +2026-09-19 02:44:23 | INFO | train rank=0 | step=32360/50000 progress=64.7% loss=0.100326 lr=3.034006e-05 lr_backbone=3.034006e-05 elapsed=33:15:46 avg_step=3.70s recent_step=4.54s eta=18:07:56 +2026-09-19 02:44:49 | INFO | train rank=0 | step=32370/50000 progress=64.7% loss=0.109187 lr=3.030966e-05 lr_backbone=3.030966e-05 elapsed=33:16:13 avg_step=3.70s recent_step=2.62s eta=18:07:13 +2026-09-19 02:45:20 | INFO | train rank=0 | step=32380/50000 progress=64.8% loss=0.086301 lr=3.027927e-05 lr_backbone=3.027927e-05 elapsed=33:16:43 avg_step=3.70s recent_step=3.05s eta=18:06:32 +2026-09-19 02:45:50 | INFO | train rank=0 | step=32390/50000 progress=64.8% loss=0.077350 lr=3.024888e-05 lr_backbone=3.024888e-05 elapsed=33:17:14 avg_step=3.70s recent_step=3.05s eta=18:05:52 +2026-09-19 02:46:42 | INFO | train rank=0 | step=32400/50000 progress=64.8% loss=0.081756 lr=3.021851e-05 lr_backbone=3.021851e-05 elapsed=33:18:05 avg_step=3.70s recent_step=5.13s eta=18:05:22 +2026-09-19 02:47:14 | INFO | train rank=0 | step=32410/50000 progress=64.8% loss=0.085687 lr=3.018814e-05 lr_backbone=3.018814e-05 elapsed=33:18:37 avg_step=3.70s recent_step=3.22s eta=18:04:43 +2026-09-19 02:47:47 | INFO | train rank=0 | step=32420/50000 progress=64.8% loss=0.108017 lr=3.015778e-05 lr_backbone=3.015778e-05 elapsed=33:19:11 avg_step=3.70s recent_step=3.36s eta=18:04:04 +2026-09-19 02:48:31 | INFO | train rank=0 | step=32430/50000 progress=64.9% loss=0.085725 lr=3.012743e-05 lr_backbone=3.012743e-05 elapsed=33:19:54 avg_step=3.70s recent_step=4.36s eta=18:03:31 +2026-09-19 02:49:11 | INFO | train rank=0 | step=32440/50000 progress=64.9% loss=0.080716 lr=3.009709e-05 lr_backbone=3.009709e-05 elapsed=33:20:34 avg_step=3.70s recent_step=4.00s eta=18:02:55 +2026-09-19 02:49:43 | INFO | train rank=0 | step=32450/50000 progress=64.9% loss=0.081652 lr=3.006676e-05 lr_backbone=3.006676e-05 elapsed=33:21:06 avg_step=3.70s recent_step=3.16s eta=18:02:15 +2026-09-19 02:50:17 | INFO | train rank=0 | step=32460/50000 progress=64.9% loss=0.078576 lr=3.003644e-05 lr_backbone=3.003644e-05 elapsed=33:21:40 avg_step=3.70s recent_step=3.41s eta=18:01:37 +2026-09-19 02:50:54 | INFO | train rank=0 | step=32470/50000 progress=64.9% loss=0.124721 lr=3.000612e-05 lr_backbone=3.000612e-05 elapsed=33:22:18 avg_step=3.70s recent_step=3.75s eta=18:01:00 +2026-09-19 02:51:43 | INFO | train rank=0 | step=32480/50000 progress=65.0% loss=0.072604 lr=2.997581e-05 lr_backbone=2.997581e-05 elapsed=33:23:06 avg_step=3.70s recent_step=4.85s eta=18:00:29 +2026-09-19 02:52:06 | INFO | train rank=0 | step=32490/50000 progress=65.0% loss=0.076761 lr=2.994552e-05 lr_backbone=2.994552e-05 elapsed=33:23:30 avg_step=3.70s recent_step=2.36s eta=17:59:45 +2026-09-19 02:52:35 | INFO | train rank=0 | step=32500/50000 progress=65.0% loss=0.071988 lr=2.991523e-05 lr_backbone=2.991523e-05 elapsed=33:23:58 avg_step=3.70s recent_step=2.86s eta=17:59:03 +2026-09-19 02:53:19 | INFO | train rank=0 | step=32510/50000 progress=65.0% loss=0.084630 lr=2.988495e-05 lr_backbone=2.988495e-05 elapsed=33:24:42 avg_step=3.70s recent_step=4.37s eta=17:58:30 +2026-09-19 02:53:55 | INFO | train rank=0 | step=32520/50000 progress=65.0% loss=0.072846 lr=2.985468e-05 lr_backbone=2.985468e-05 elapsed=33:25:18 avg_step=3.70s recent_step=3.61s eta=17:57:53 +2026-09-19 02:54:26 | INFO | train rank=0 | step=32530/50000 progress=65.1% loss=0.122143 lr=2.982442e-05 lr_backbone=2.982442e-05 elapsed=33:25:49 avg_step=3.70s recent_step=3.12s eta=17:57:12 +2026-09-19 02:55:06 | INFO | train rank=0 | step=32540/50000 progress=65.1% loss=0.105658 lr=2.979416e-05 lr_backbone=2.979416e-05 elapsed=33:26:29 avg_step=3.70s recent_step=3.98s eta=17:56:37 +2026-09-19 02:55:51 | INFO | train rank=0 | step=32550/50000 progress=65.1% loss=0.084828 lr=2.976392e-05 lr_backbone=2.976392e-05 elapsed=33:27:14 avg_step=3.70s recent_step=4.51s eta=17:56:04 +2026-09-19 02:56:40 | INFO | train rank=0 | step=32560/50000 progress=65.1% loss=0.080067 lr=2.973368e-05 lr_backbone=2.973368e-05 elapsed=33:28:03 avg_step=3.70s recent_step=4.92s eta=17:55:34 +2026-09-19 02:57:17 | INFO | train rank=0 | step=32570/50000 progress=65.1% loss=0.098862 lr=2.970346e-05 lr_backbone=2.970346e-05 elapsed=33:28:40 avg_step=3.70s recent_step=3.65s eta=17:54:57 +2026-09-19 02:57:49 | INFO | train rank=0 | step=32580/50000 progress=65.2% loss=0.085536 lr=2.967324e-05 lr_backbone=2.967324e-05 elapsed=33:29:13 avg_step=3.70s recent_step=3.27s eta=17:54:17 +2026-09-19 02:58:23 | INFO | train rank=0 | step=32590/50000 progress=65.2% loss=0.125765 lr=2.964303e-05 lr_backbone=2.964303e-05 elapsed=33:29:46 avg_step=3.70s recent_step=3.36s eta=17:53:38 +2026-09-19 02:59:14 | INFO | train rank=0 | step=32600/50000 progress=65.2% loss=0.101102 lr=2.961283e-05 lr_backbone=2.961283e-05 elapsed=33:30:38 avg_step=3.70s recent_step=5.16s eta=17:53:09 +2026-09-19 02:59:46 | INFO | train rank=0 | step=32610/50000 progress=65.2% loss=0.092372 lr=2.958264e-05 lr_backbone=2.958264e-05 elapsed=33:31:10 avg_step=3.70s recent_step=3.20s eta=17:52:30 +2026-09-19 03:00:15 | INFO | train rank=0 | step=32620/50000 progress=65.2% loss=0.077050 lr=2.955246e-05 lr_backbone=2.955246e-05 elapsed=33:31:38 avg_step=3.70s recent_step=2.84s eta=17:51:48 +2026-09-19 03:00:44 | INFO | train rank=0 | step=32630/50000 progress=65.3% loss=0.082100 lr=2.952228e-05 lr_backbone=2.952228e-05 elapsed=33:32:08 avg_step=3.70s recent_step=2.97s eta=17:51:07 +2026-09-19 03:01:31 | INFO | train rank=0 | step=32640/50000 progress=65.3% loss=0.105516 lr=2.949212e-05 lr_backbone=2.949212e-05 elapsed=33:32:54 avg_step=3.70s recent_step=4.67s eta=17:50:35 +2026-09-19 03:02:02 | INFO | train rank=0 | step=32650/50000 progress=65.3% loss=0.083633 lr=2.946196e-05 lr_backbone=2.946196e-05 elapsed=33:33:25 avg_step=3.70s recent_step=3.09s eta=17:49:55 +2026-09-19 03:02:34 | INFO | train rank=0 | step=32660/50000 progress=65.3% loss=0.094742 lr=2.943182e-05 lr_backbone=2.943182e-05 elapsed=33:33:58 avg_step=3.70s recent_step=3.23s eta=17:49:15 +2026-09-19 03:03:14 | INFO | train rank=0 | step=32670/50000 progress=65.3% loss=0.091986 lr=2.940168e-05 lr_backbone=2.940168e-05 elapsed=33:34:37 avg_step=3.70s recent_step=3.94s eta=17:48:40 +2026-09-19 03:03:51 | INFO | train rank=0 | step=32680/50000 progress=65.4% loss=0.110508 lr=2.937155e-05 lr_backbone=2.937155e-05 elapsed=33:35:15 avg_step=3.70s recent_step=3.76s eta=17:48:03 +2026-09-19 03:04:40 | INFO | train rank=0 | step=32690/50000 progress=65.4% loss=0.102424 lr=2.934143e-05 lr_backbone=2.934143e-05 elapsed=33:36:04 avg_step=3.70s recent_step=4.92s eta=17:47:32 +2026-09-19 03:05:02 | INFO | train rank=0 | step=32700/50000 progress=65.4% loss=0.112400 lr=2.931132e-05 lr_backbone=2.931132e-05 elapsed=33:36:26 avg_step=3.70s recent_step=2.17s eta=17:46:47 +2026-09-19 03:05:45 | INFO | train rank=0 | step=32710/50000 progress=65.4% loss=0.080321 lr=2.928122e-05 lr_backbone=2.928122e-05 elapsed=33:37:08 avg_step=3.70s recent_step=4.29s eta=17:46:14 +2026-09-19 03:06:16 | INFO | train rank=0 | step=32720/50000 progress=65.4% loss=0.070867 lr=2.925113e-05 lr_backbone=2.925113e-05 elapsed=33:37:39 avg_step=3.70s recent_step=3.07s eta=17:45:33 +2026-09-19 03:06:52 | INFO | train rank=0 | step=32730/50000 progress=65.5% loss=0.089450 lr=2.922105e-05 lr_backbone=2.922105e-05 elapsed=33:38:15 avg_step=3.70s recent_step=3.58s eta=17:44:56 +2026-09-19 03:07:27 | INFO | train rank=0 | step=32740/50000 progress=65.5% loss=0.092824 lr=2.919097e-05 lr_backbone=2.919097e-05 elapsed=33:38:51 avg_step=3.70s recent_step=3.57s eta=17:44:18 +2026-09-19 03:08:28 | INFO | train rank=0 | step=32750/50000 progress=65.5% loss=0.089476 lr=2.916091e-05 lr_backbone=2.916091e-05 elapsed=33:39:51 avg_step=3.70s recent_step=6.02s eta=17:43:53 +2026-09-19 03:08:58 | INFO | train rank=0 | step=32760/50000 progress=65.5% loss=0.139481 lr=2.913085e-05 lr_backbone=2.913085e-05 elapsed=33:40:21 avg_step=3.70s recent_step=3.02s eta=17:43:13 +2026-09-19 03:09:33 | INFO | train rank=0 | step=32770/50000 progress=65.5% loss=0.087064 lr=2.910080e-05 lr_backbone=2.910080e-05 elapsed=33:40:56 avg_step=3.70s recent_step=3.50s eta=17:42:34 +2026-09-19 03:10:04 | INFO | train rank=0 | step=32780/50000 progress=65.6% loss=0.091432 lr=2.907077e-05 lr_backbone=2.907077e-05 elapsed=33:41:27 avg_step=3.70s recent_step=3.09s eta=17:41:54 +2026-09-19 03:10:43 | INFO | train rank=0 | step=32790/50000 progress=65.6% loss=0.080083 lr=2.904074e-05 lr_backbone=2.904074e-05 elapsed=33:42:06 avg_step=3.70s recent_step=3.89s eta=17:41:18 +2026-09-19 03:11:35 | INFO | train rank=0 | step=32800/50000 progress=65.6% loss=0.088530 lr=2.901072e-05 lr_backbone=2.901072e-05 elapsed=33:42:59 avg_step=3.70s recent_step=5.28s eta=17:40:50 +2026-09-19 03:12:08 | INFO | train rank=0 | step=32810/50000 progress=65.6% loss=0.098360 lr=2.898071e-05 lr_backbone=2.898071e-05 elapsed=33:43:32 avg_step=3.70s recent_step=3.30s eta=17:40:10 +2026-09-19 03:12:44 | INFO | train rank=0 | step=32820/50000 progress=65.6% loss=0.081681 lr=2.895071e-05 lr_backbone=2.895071e-05 elapsed=33:44:07 avg_step=3.70s recent_step=3.54s eta=17:39:33 +2026-09-19 03:13:19 | INFO | train rank=0 | step=32830/50000 progress=65.7% loss=0.074387 lr=2.892072e-05 lr_backbone=2.892072e-05 elapsed=33:44:43 avg_step=3.70s recent_step=3.58s eta=17:38:55 +2026-09-19 03:14:11 | INFO | train rank=0 | step=32840/50000 progress=65.7% loss=0.060485 lr=2.889074e-05 lr_backbone=2.889074e-05 elapsed=33:45:34 avg_step=3.70s recent_step=5.16s eta=17:38:26 +2026-09-19 03:14:44 | INFO | train rank=0 | step=32850/50000 progress=65.7% loss=0.066839 lr=2.886076e-05 lr_backbone=2.886076e-05 elapsed=33:46:07 avg_step=3.70s recent_step=3.25s eta=17:37:46 +2026-09-19 03:15:16 | INFO | train rank=0 | step=32860/50000 progress=65.7% loss=0.070296 lr=2.883080e-05 lr_backbone=2.883080e-05 elapsed=33:46:39 avg_step=3.70s recent_step=3.24s eta=17:37:07 +2026-09-19 03:15:46 | INFO | train rank=0 | step=32870/50000 progress=65.7% loss=0.078177 lr=2.880084e-05 lr_backbone=2.880084e-05 elapsed=33:47:09 avg_step=3.70s recent_step=2.99s eta=17:36:26 +2026-09-19 03:16:26 | INFO | train rank=0 | step=32880/50000 progress=65.8% loss=0.098662 lr=2.877090e-05 lr_backbone=2.877090e-05 elapsed=33:47:50 avg_step=3.70s recent_step=4.05s eta=17:35:51 +2026-09-19 03:17:03 | INFO | train rank=0 | step=32890/50000 progress=65.8% loss=0.069870 lr=2.874096e-05 lr_backbone=2.874096e-05 elapsed=33:48:26 avg_step=3.70s recent_step=3.62s eta=17:35:13 +2026-09-19 03:17:31 | INFO | train rank=0 | step=32900/50000 progress=65.8% loss=0.082547 lr=2.871104e-05 lr_backbone=2.871104e-05 elapsed=33:48:54 avg_step=3.70s recent_step=2.83s eta=17:34:32 +2026-09-19 03:18:02 | INFO | train rank=0 | step=32910/50000 progress=65.8% loss=0.123117 lr=2.868112e-05 lr_backbone=2.868112e-05 elapsed=33:49:25 avg_step=3.70s recent_step=3.09s eta=17:33:52 +2026-09-19 03:19:00 | INFO | train rank=0 | step=32920/50000 progress=65.8% loss=0.105250 lr=2.865121e-05 lr_backbone=2.865121e-05 elapsed=33:50:24 avg_step=3.70s recent_step=5.85s eta=17:33:26 +2026-09-19 03:19:33 | INFO | train rank=0 | step=32930/50000 progress=65.9% loss=0.076315 lr=2.862131e-05 lr_backbone=2.862131e-05 elapsed=33:50:56 avg_step=3.70s recent_step=3.23s eta=17:32:47 +2026-09-19 03:20:11 | INFO | train rank=0 | step=32940/50000 progress=65.9% loss=0.079425 lr=2.859142e-05 lr_backbone=2.859142e-05 elapsed=33:51:34 avg_step=3.70s recent_step=3.83s eta=17:32:10 +2026-09-19 03:20:43 | INFO | train rank=0 | step=32950/50000 progress=65.9% loss=0.126833 lr=2.856154e-05 lr_backbone=2.856154e-05 elapsed=33:52:07 avg_step=3.70s recent_step=3.22s eta=17:31:31 +2026-09-19 03:21:36 | INFO | train rank=0 | step=32960/50000 progress=65.9% loss=0.100437 lr=2.853167e-05 lr_backbone=2.853167e-05 elapsed=33:53:00 avg_step=3.70s recent_step=5.32s eta=17:31:02 +2026-09-19 03:22:08 | INFO | train rank=0 | step=32970/50000 progress=65.9% loss=0.085790 lr=2.850181e-05 lr_backbone=2.850181e-05 elapsed=33:53:32 avg_step=3.70s recent_step=3.18s eta=17:30:22 +2026-09-19 03:22:45 | INFO | train rank=0 | step=32980/50000 progress=66.0% loss=0.077229 lr=2.847196e-05 lr_backbone=2.847196e-05 elapsed=33:54:08 avg_step=3.70s recent_step=3.69s eta=17:29:45 +2026-09-19 03:23:13 | INFO | train rank=0 | step=32990/50000 progress=66.0% loss=0.119406 lr=2.844212e-05 lr_backbone=2.844212e-05 elapsed=33:54:37 avg_step=3.70s recent_step=2.81s eta=17:29:04 +2026-09-19 03:24:09 | INFO | train rank=0 | step=33000/50000 progress=66.0% loss=0.111749 lr=2.841228e-05 lr_backbone=2.841228e-05 elapsed=33:55:33 avg_step=3.70s recent_step=5.62s eta=17:28:37 +2026-09-19 03:24:41 | INFO | train rank=0 | step=33010/50000 progress=66.0% loss=0.070006 lr=2.838246e-05 lr_backbone=2.838246e-05 elapsed=33:56:04 avg_step=3.70s recent_step=3.16s eta=17:27:57 +2026-09-19 03:25:06 | INFO | train rank=0 | step=33020/50000 progress=66.0% loss=0.090296 lr=2.835264e-05 lr_backbone=2.835264e-05 elapsed=33:56:29 avg_step=3.70s recent_step=2.44s eta=17:27:13 +2026-09-19 03:25:37 | INFO | train rank=0 | step=33030/50000 progress=66.1% loss=0.073046 lr=2.832284e-05 lr_backbone=2.832284e-05 elapsed=33:57:00 avg_step=3.70s recent_step=3.16s eta=17:26:34 +2026-09-19 03:26:32 | INFO | train rank=0 | step=33040/50000 progress=66.1% loss=0.096737 lr=2.829304e-05 lr_backbone=2.829304e-05 elapsed=33:57:55 avg_step=3.70s recent_step=5.44s eta=17:26:06 +2026-09-19 03:26:56 | INFO | train rank=0 | step=33050/50000 progress=66.1% loss=0.094216 lr=2.826326e-05 lr_backbone=2.826326e-05 elapsed=33:58:19 avg_step=3.70s recent_step=2.40s eta=17:25:22 +2026-09-19 03:27:24 | INFO | train rank=0 | step=33060/50000 progress=66.1% loss=0.106317 lr=2.823348e-05 lr_backbone=2.823348e-05 elapsed=33:58:48 avg_step=3.70s recent_step=2.87s eta=17:24:41 +2026-09-19 03:27:59 | INFO | train rank=0 | step=33070/50000 progress=66.1% loss=0.089304 lr=2.820372e-05 lr_backbone=2.820372e-05 elapsed=33:59:22 avg_step=3.70s recent_step=3.47s eta=17:24:02 +2026-09-19 03:28:48 | INFO | train rank=0 | step=33080/50000 progress=66.2% loss=0.090694 lr=2.817396e-05 lr_backbone=2.817396e-05 elapsed=34:00:12 avg_step=3.70s recent_step=4.95s eta=17:23:32 +2026-09-19 03:29:22 | INFO | train rank=0 | step=33090/50000 progress=66.2% loss=0.095374 lr=2.814421e-05 lr_backbone=2.814421e-05 elapsed=34:00:45 avg_step=3.70s recent_step=3.37s eta=17:22:53 +2026-09-19 03:29:58 | INFO | train rank=0 | step=33100/50000 progress=66.2% loss=0.103537 lr=2.811447e-05 lr_backbone=2.811447e-05 elapsed=34:01:21 avg_step=3.70s recent_step=3.61s eta=17:22:16 +2026-09-19 03:30:34 | INFO | train rank=0 | step=33110/50000 progress=66.2% loss=0.080120 lr=2.808475e-05 lr_backbone=2.808475e-05 elapsed=34:01:57 avg_step=3.70s recent_step=3.59s eta=17:21:38 +2026-09-19 03:31:26 | INFO | train rank=0 | step=33120/50000 progress=66.2% loss=0.079669 lr=2.805503e-05 lr_backbone=2.805503e-05 elapsed=34:02:50 avg_step=3.70s recent_step=5.22s eta=17:21:09 +2026-09-19 03:31:52 | INFO | train rank=0 | step=33130/50000 progress=66.3% loss=0.097788 lr=2.802532e-05 lr_backbone=2.802532e-05 elapsed=34:03:15 avg_step=3.70s recent_step=2.58s eta=17:20:26 +2026-09-19 03:32:20 | INFO | train rank=0 | step=33140/50000 progress=66.3% loss=0.082306 lr=2.799562e-05 lr_backbone=2.799562e-05 elapsed=34:03:43 avg_step=3.70s recent_step=2.80s eta=17:19:44 +2026-09-19 03:32:56 | INFO | train rank=0 | step=33150/50000 progress=66.3% loss=0.080433 lr=2.796593e-05 lr_backbone=2.796593e-05 elapsed=34:04:20 avg_step=3.70s recent_step=3.64s eta=17:19:07 +2026-09-19 03:33:43 | INFO | train rank=0 | step=33160/50000 progress=66.3% loss=0.101172 lr=2.793625e-05 lr_backbone=2.793625e-05 elapsed=34:05:07 avg_step=3.70s recent_step=4.69s eta=17:18:35 +2026-09-19 03:34:12 | INFO | train rank=0 | step=33170/50000 progress=66.3% loss=0.070133 lr=2.790658e-05 lr_backbone=2.790658e-05 elapsed=34:05:35 avg_step=3.70s recent_step=2.82s eta=17:17:54 +2026-09-19 03:34:44 | INFO | train rank=0 | step=33180/50000 progress=66.4% loss=0.124560 lr=2.787692e-05 lr_backbone=2.787692e-05 elapsed=34:06:08 avg_step=3.70s recent_step=3.28s eta=17:17:15 +2026-09-19 03:35:20 | INFO | train rank=0 | step=33190/50000 progress=66.4% loss=0.099366 lr=2.784726e-05 lr_backbone=2.784726e-05 elapsed=34:06:44 avg_step=3.70s recent_step=3.60s eta=17:16:37 +2026-09-19 03:36:15 | INFO | train rank=0 | step=33200/50000 progress=66.4% loss=0.083119 lr=2.781762e-05 lr_backbone=2.781762e-05 elapsed=34:07:38 avg_step=3.70s recent_step=5.48s eta=17:16:09 +2026-09-19 03:36:52 | INFO | train rank=0 | step=33210/50000 progress=66.4% loss=0.101131 lr=2.778799e-05 lr_backbone=2.778799e-05 elapsed=34:08:15 avg_step=3.70s recent_step=3.70s eta=17:15:32 +2026-09-19 03:37:31 | INFO | train rank=0 | step=33220/50000 progress=66.4% loss=0.088994 lr=2.775837e-05 lr_backbone=2.775837e-05 elapsed=34:08:54 avg_step=3.70s recent_step=3.87s eta=17:14:56 +2026-09-19 03:37:58 | INFO | train rank=0 | step=33230/50000 progress=66.5% loss=0.096275 lr=2.772876e-05 lr_backbone=2.772876e-05 elapsed=34:09:21 avg_step=3.70s recent_step=2.70s eta=17:14:14 +2026-09-19 03:38:47 | INFO | train rank=0 | step=33240/50000 progress=66.5% loss=0.089136 lr=2.769915e-05 lr_backbone=2.769915e-05 elapsed=34:10:10 avg_step=3.70s recent_step=4.87s eta=17:13:43 +2026-09-19 03:39:22 | INFO | train rank=0 | step=33250/50000 progress=66.5% loss=0.087031 lr=2.766956e-05 lr_backbone=2.766956e-05 elapsed=34:10:46 avg_step=3.70s recent_step=3.58s eta=17:13:05 +2026-09-19 03:39:56 | INFO | train rank=0 | step=33260/50000 progress=66.5% loss=0.114055 lr=2.763998e-05 lr_backbone=2.763998e-05 elapsed=34:11:19 avg_step=3.70s recent_step=3.38s eta=17:12:27 +2026-09-19 03:40:24 | INFO | train rank=0 | step=33270/50000 progress=66.5% loss=0.101267 lr=2.761040e-05 lr_backbone=2.761040e-05 elapsed=34:11:47 avg_step=3.70s recent_step=2.74s eta=17:11:45 +2026-09-19 03:41:20 | INFO | train rank=0 | step=33280/50000 progress=66.6% loss=0.099498 lr=2.758084e-05 lr_backbone=2.758084e-05 elapsed=34:12:44 avg_step=3.70s recent_step=5.67s eta=17:11:18 +2026-09-19 03:41:53 | INFO | train rank=0 | step=33290/50000 progress=66.6% loss=0.088180 lr=2.755129e-05 lr_backbone=2.755129e-05 elapsed=34:13:16 avg_step=3.70s recent_step=3.26s eta=17:10:38 +2026-09-19 03:42:23 | INFO | train rank=0 | step=33300/50000 progress=66.6% loss=0.108754 lr=2.752174e-05 lr_backbone=2.752174e-05 elapsed=34:13:46 avg_step=3.70s recent_step=3.00s eta=17:09:58 +2026-09-19 03:42:53 | INFO | train rank=0 | step=33310/50000 progress=66.6% loss=0.099741 lr=2.749221e-05 lr_backbone=2.749221e-05 elapsed=34:14:16 avg_step=3.70s recent_step=3.01s eta=17:09:17 +2026-09-19 03:43:49 | INFO | train rank=0 | step=33320/50000 progress=66.6% loss=0.076705 lr=2.746268e-05 lr_backbone=2.746268e-05 elapsed=34:15:12 avg_step=3.70s recent_step=5.60s eta=17:08:50 +2026-09-19 03:44:24 | INFO | train rank=0 | step=33330/50000 progress=66.7% loss=0.109789 lr=2.743317e-05 lr_backbone=2.743317e-05 elapsed=34:15:47 avg_step=3.70s recent_step=3.49s eta=17:08:12 +2026-09-19 03:44:56 | INFO | train rank=0 | step=33340/50000 progress=66.7% loss=0.091303 lr=2.740366e-05 lr_backbone=2.740366e-05 elapsed=34:16:19 avg_step=3.70s recent_step=3.16s eta=17:07:32 +2026-09-19 03:45:32 | INFO | train rank=0 | step=33350/50000 progress=66.7% loss=0.125788 lr=2.737417e-05 lr_backbone=2.737417e-05 elapsed=34:16:56 avg_step=3.70s recent_step=3.66s eta=17:06:55 +2026-09-19 03:46:11 | INFO | train rank=0 | step=33360/50000 progress=66.7% loss=0.096621 lr=2.734468e-05 lr_backbone=2.734468e-05 elapsed=34:17:34 avg_step=3.70s recent_step=3.86s eta=17:06:19 +2026-09-19 03:46:41 | INFO | train rank=0 | step=33370/50000 progress=66.7% loss=0.096101 lr=2.731521e-05 lr_backbone=2.731521e-05 elapsed=34:18:04 avg_step=3.70s recent_step=2.97s eta=17:05:38 +2026-09-19 03:47:12 | INFO | train rank=0 | step=33380/50000 progress=66.8% loss=0.076081 lr=2.728574e-05 lr_backbone=2.728574e-05 elapsed=34:18:35 avg_step=3.70s recent_step=3.17s eta=17:04:58 +2026-09-19 03:47:46 | INFO | train rank=0 | step=33390/50000 progress=66.8% loss=0.095744 lr=2.725629e-05 lr_backbone=2.725629e-05 elapsed=34:19:09 avg_step=3.70s recent_step=3.36s eta=17:04:20 +2026-09-19 03:48:31 | INFO | train rank=0 | step=33400/50000 progress=66.8% loss=0.092294 lr=2.722684e-05 lr_backbone=2.722684e-05 elapsed=34:19:54 avg_step=3.70s recent_step=4.54s eta=17:03:47 +2026-09-19 03:48:58 | INFO | train rank=0 | step=33410/50000 progress=66.8% loss=0.100873 lr=2.719741e-05 lr_backbone=2.719741e-05 elapsed=34:20:21 avg_step=3.70s recent_step=2.64s eta=17:03:05 +2026-09-19 03:49:35 | INFO | train rank=0 | step=33420/50000 progress=66.8% loss=0.084339 lr=2.716798e-05 lr_backbone=2.716798e-05 elapsed=34:20:58 avg_step=3.70s recent_step=3.74s eta=17:02:28 +2026-09-19 03:50:18 | INFO | train rank=0 | step=33430/50000 progress=66.9% loss=0.080892 lr=2.713857e-05 lr_backbone=2.713857e-05 elapsed=34:21:42 avg_step=3.70s recent_step=4.32s eta=17:01:54 +2026-09-19 03:51:03 | INFO | train rank=0 | step=33440/50000 progress=66.9% loss=0.079832 lr=2.710916e-05 lr_backbone=2.710916e-05 elapsed=34:22:26 avg_step=3.70s recent_step=4.47s eta=17:01:21 +2026-09-19 03:51:33 | INFO | train rank=0 | step=33450/50000 progress=66.9% loss=0.100885 lr=2.707977e-05 lr_backbone=2.707977e-05 elapsed=34:22:56 avg_step=3.70s recent_step=3.01s eta=17:00:40 +2026-09-19 03:52:08 | INFO | train rank=0 | step=33460/50000 progress=66.9% loss=0.098663 lr=2.705038e-05 lr_backbone=2.705038e-05 elapsed=34:23:31 avg_step=3.70s recent_step=3.47s eta=17:00:02 +2026-09-19 03:52:46 | INFO | train rank=0 | step=33470/50000 progress=66.9% loss=0.085737 lr=2.702101e-05 lr_backbone=2.702101e-05 elapsed=34:24:09 avg_step=3.70s recent_step=3.84s eta=16:59:26 +2026-09-19 03:53:46 | INFO | train rank=0 | step=33480/50000 progress=67.0% loss=0.075934 lr=2.699164e-05 lr_backbone=2.699164e-05 elapsed=34:25:09 avg_step=3.70s recent_step=6.00s eta=16:59:00 +2026-09-19 03:54:13 | INFO | train rank=0 | step=33490/50000 progress=67.0% loss=0.086033 lr=2.696229e-05 lr_backbone=2.696229e-05 elapsed=34:25:36 avg_step=3.70s recent_step=2.65s eta=16:58:18 +2026-09-19 03:54:43 | INFO | train rank=0 | step=33500/50000 progress=67.0% loss=0.108502 lr=2.693294e-05 lr_backbone=2.693294e-05 elapsed=34:26:06 avg_step=3.70s recent_step=3.01s eta=16:57:38 +2026-09-19 03:55:21 | INFO | train rank=0 | step=33510/50000 progress=67.0% loss=0.083236 lr=2.690361e-05 lr_backbone=2.690361e-05 elapsed=34:26:44 avg_step=3.70s recent_step=3.79s eta=16:57:01 +2026-09-19 03:56:18 | INFO | train rank=0 | step=33520/50000 progress=67.0% loss=0.083818 lr=2.687428e-05 lr_backbone=2.687428e-05 elapsed=34:27:42 avg_step=3.70s recent_step=5.77s eta=16:56:34 +2026-09-19 03:56:50 | INFO | train rank=0 | step=33530/50000 progress=67.1% loss=0.080820 lr=2.684497e-05 lr_backbone=2.684497e-05 elapsed=34:28:14 avg_step=3.70s recent_step=3.21s eta=16:55:55 +2026-09-19 03:57:22 | INFO | train rank=0 | step=33540/50000 progress=67.1% loss=0.122629 lr=2.681566e-05 lr_backbone=2.681566e-05 elapsed=34:28:45 avg_step=3.70s recent_step=3.16s eta=16:55:15 +2026-09-19 03:57:56 | INFO | train rank=0 | step=33550/50000 progress=67.1% loss=0.095528 lr=2.678637e-05 lr_backbone=2.678637e-05 elapsed=34:29:19 avg_step=3.70s recent_step=3.40s eta=16:54:37 +2026-09-19 03:58:49 | INFO | train rank=0 | step=33560/50000 progress=67.1% loss=0.084094 lr=2.675708e-05 lr_backbone=2.675708e-05 elapsed=34:30:12 avg_step=3.70s recent_step=5.31s eta=16:54:08 +2026-09-19 03:59:19 | INFO | train rank=0 | step=33570/50000 progress=67.1% loss=0.093757 lr=2.672781e-05 lr_backbone=2.672781e-05 elapsed=34:30:43 avg_step=3.70s recent_step=3.03s eta=16:53:27 +2026-09-19 03:59:34 | INFO | train rank=0 | step=33580/50000 progress=67.2% loss=0.076544 lr=2.669855e-05 lr_backbone=2.669855e-05 elapsed=34:30:57 avg_step=3.70s recent_step=1.47s eta=16:52:39 +2026-09-19 04:00:05 | INFO | train rank=0 | step=33590/50000 progress=67.2% loss=0.128381 lr=2.666929e-05 lr_backbone=2.666929e-05 elapsed=34:31:28 avg_step=3.70s recent_step=3.06s eta=16:51:59 +2026-09-19 04:00:58 | INFO | train rank=0 | step=33600/50000 progress=67.2% loss=0.114353 lr=2.664005e-05 lr_backbone=2.664005e-05 elapsed=34:32:22 avg_step=3.70s recent_step=5.39s eta=16:51:30 +2026-09-19 04:01:34 | INFO | train rank=0 | step=33610/50000 progress=67.2% loss=0.079661 lr=2.661082e-05 lr_backbone=2.661082e-05 elapsed=34:32:58 avg_step=3.70s recent_step=3.57s eta=16:50:53 +2026-09-19 04:02:04 | INFO | train rank=0 | step=33620/50000 progress=67.2% loss=0.108384 lr=2.658159e-05 lr_backbone=2.658159e-05 elapsed=34:33:27 avg_step=3.70s recent_step=2.99s eta=16:50:12 +2026-09-19 04:02:35 | INFO | train rank=0 | step=33630/50000 progress=67.3% loss=0.126955 lr=2.655238e-05 lr_backbone=2.655238e-05 elapsed=34:33:58 avg_step=3.70s recent_step=3.10s eta=16:49:32 +2026-09-19 04:03:23 | INFO | train rank=0 | step=33640/50000 progress=67.3% loss=0.090213 lr=2.652318e-05 lr_backbone=2.652318e-05 elapsed=34:34:46 avg_step=3.70s recent_step=4.79s eta=16:49:01 +2026-09-19 04:03:57 | INFO | train rank=0 | step=33650/50000 progress=67.3% loss=0.065108 lr=2.649399e-05 lr_backbone=2.649399e-05 elapsed=34:35:20 avg_step=3.70s recent_step=3.36s eta=16:48:22 +2026-09-19 04:04:25 | INFO | train rank=0 | step=33660/50000 progress=67.3% loss=0.073499 lr=2.646480e-05 lr_backbone=2.646480e-05 elapsed=34:35:48 avg_step=3.70s recent_step=2.85s eta=16:47:41 +2026-09-19 04:04:58 | INFO | train rank=0 | step=33670/50000 progress=67.3% loss=0.075048 lr=2.643563e-05 lr_backbone=2.643563e-05 elapsed=34:36:21 avg_step=3.70s recent_step=3.23s eta=16:47:02 +2026-09-19 04:05:59 | INFO | train rank=0 | step=33680/50000 progress=67.4% loss=0.077044 lr=2.640647e-05 lr_backbone=2.640647e-05 elapsed=34:37:23 avg_step=3.70s recent_step=6.19s eta=16:46:37 +2026-09-19 04:06:35 | INFO | train rank=0 | step=33690/50000 progress=67.4% loss=0.081322 lr=2.637732e-05 lr_backbone=2.637732e-05 elapsed=34:37:59 avg_step=3.70s recent_step=3.60s eta=16:45:59 +2026-09-19 04:07:07 | INFO | train rank=0 | step=33700/50000 progress=67.4% loss=0.076404 lr=2.634818e-05 lr_backbone=2.634818e-05 elapsed=34:38:30 avg_step=3.70s recent_step=3.12s eta=16:45:19 +2026-09-19 04:07:34 | INFO | train rank=0 | step=33710/50000 progress=67.4% loss=0.098452 lr=2.631905e-05 lr_backbone=2.631905e-05 elapsed=34:38:58 avg_step=3.70s recent_step=2.76s eta=16:44:38 +2026-09-19 04:08:36 | INFO | train rank=0 | step=33720/50000 progress=67.4% loss=0.105524 lr=2.628993e-05 lr_backbone=2.628993e-05 elapsed=34:39:59 avg_step=3.70s recent_step=6.17s eta=16:44:13 +2026-09-19 04:09:06 | INFO | train rank=0 | step=33730/50000 progress=67.5% loss=0.096831 lr=2.626082e-05 lr_backbone=2.626082e-05 elapsed=34:40:30 avg_step=3.70s recent_step=3.04s eta=16:43:33 +2026-09-19 04:09:40 | INFO | train rank=0 | step=33740/50000 progress=67.5% loss=0.089473 lr=2.623172e-05 lr_backbone=2.623172e-05 elapsed=34:41:03 avg_step=3.70s recent_step=3.34s eta=16:42:54 +2026-09-19 04:10:14 | INFO | train rank=0 | step=33750/50000 progress=67.5% loss=0.077400 lr=2.620263e-05 lr_backbone=2.620263e-05 elapsed=34:41:38 avg_step=3.70s recent_step=3.47s eta=16:42:16 +2026-09-19 04:11:06 | INFO | train rank=0 | step=33760/50000 progress=67.5% loss=0.069308 lr=2.617355e-05 lr_backbone=2.617355e-05 elapsed=34:42:30 avg_step=3.70s recent_step=5.19s eta=16:41:46 +2026-09-19 04:11:42 | INFO | train rank=0 | step=33770/50000 progress=67.5% loss=0.143162 lr=2.614448e-05 lr_backbone=2.614448e-05 elapsed=34:43:05 avg_step=3.70s recent_step=3.55s eta=16:41:08 +2026-09-19 04:12:14 | INFO | train rank=0 | step=33780/50000 progress=67.6% loss=0.087085 lr=2.611543e-05 lr_backbone=2.611543e-05 elapsed=34:43:37 avg_step=3.70s recent_step=3.20s eta=16:40:29 +2026-09-19 04:12:44 | INFO | train rank=0 | step=33790/50000 progress=67.6% loss=0.117202 lr=2.608638e-05 lr_backbone=2.608638e-05 elapsed=34:44:07 avg_step=3.70s recent_step=3.02s eta=16:39:48 +2026-09-19 04:13:30 | INFO | train rank=0 | step=33800/50000 progress=67.6% loss=0.121493 lr=2.605734e-05 lr_backbone=2.605734e-05 elapsed=34:44:53 avg_step=3.70s recent_step=4.57s eta=16:39:16 +2026-09-19 04:13:58 | INFO | train rank=0 | step=33810/50000 progress=67.6% loss=0.074142 lr=2.602832e-05 lr_backbone=2.602832e-05 elapsed=34:45:21 avg_step=3.70s recent_step=2.78s eta=16:38:34 +2026-09-19 04:14:28 | INFO | train rank=0 | step=33820/50000 progress=67.6% loss=0.081641 lr=2.599930e-05 lr_backbone=2.599930e-05 elapsed=34:45:51 avg_step=3.70s recent_step=3.06s eta=16:37:54 +2026-09-19 04:15:04 | INFO | train rank=0 | step=33830/50000 progress=67.7% loss=0.136316 lr=2.597030e-05 lr_backbone=2.597030e-05 elapsed=34:46:27 avg_step=3.70s recent_step=3.57s eta=16:37:16 +2026-09-19 04:15:53 | INFO | train rank=0 | step=33840/50000 progress=67.7% loss=0.153778 lr=2.594130e-05 lr_backbone=2.594130e-05 elapsed=34:47:17 avg_step=3.70s recent_step=4.95s eta=16:36:45 +2026-09-19 04:16:26 | INFO | train rank=0 | step=33850/50000 progress=67.7% loss=0.092111 lr=2.591232e-05 lr_backbone=2.591232e-05 elapsed=34:47:50 avg_step=3.70s recent_step=3.30s eta=16:36:06 +2026-09-19 04:17:06 | INFO | train rank=0 | step=33860/50000 progress=67.7% loss=0.105379 lr=2.588334e-05 lr_backbone=2.588334e-05 elapsed=34:48:30 avg_step=3.70s recent_step=4.00s eta=16:35:31 +2026-09-19 04:17:42 | INFO | train rank=0 | step=33870/50000 progress=67.7% loss=0.067521 lr=2.585438e-05 lr_backbone=2.585438e-05 elapsed=34:49:06 avg_step=3.70s recent_step=3.60s eta=16:34:53 +2026-09-19 04:18:39 | INFO | train rank=0 | step=33880/50000 progress=67.8% loss=0.090144 lr=2.582543e-05 lr_backbone=2.582543e-05 elapsed=34:50:02 avg_step=3.70s recent_step=5.66s eta=16:34:26 +2026-09-19 04:19:13 | INFO | train rank=0 | step=33890/50000 progress=67.8% loss=0.090279 lr=2.579648e-05 lr_backbone=2.579648e-05 elapsed=34:50:36 avg_step=3.70s recent_step=3.36s eta=16:33:47 +2026-09-19 04:19:44 | INFO | train rank=0 | step=33900/50000 progress=67.8% loss=0.079327 lr=2.576755e-05 lr_backbone=2.576755e-05 elapsed=34:51:07 avg_step=3.70s recent_step=3.13s eta=16:33:07 +2026-09-19 04:20:13 | INFO | train rank=0 | step=33910/50000 progress=67.8% loss=0.119816 lr=2.573863e-05 lr_backbone=2.573863e-05 elapsed=34:51:36 avg_step=3.70s recent_step=2.93s eta=16:32:27 +2026-09-19 04:21:02 | INFO | train rank=0 | step=33920/50000 progress=67.8% loss=0.091390 lr=2.570972e-05 lr_backbone=2.570972e-05 elapsed=34:52:25 avg_step=3.70s recent_step=4.84s eta=16:31:55 +2026-09-19 04:21:31 | INFO | train rank=0 | step=33930/50000 progress=67.9% loss=0.093877 lr=2.568082e-05 lr_backbone=2.568082e-05 elapsed=34:52:54 avg_step=3.70s recent_step=2.89s eta=16:31:14 +2026-09-19 04:22:00 | INFO | train rank=0 | step=33940/50000 progress=67.9% loss=0.062506 lr=2.565193e-05 lr_backbone=2.565193e-05 elapsed=34:53:23 avg_step=3.70s recent_step=2.96s eta=16:30:34 +2026-09-19 04:22:30 | INFO | train rank=0 | step=33950/50000 progress=67.9% loss=0.085553 lr=2.562306e-05 lr_backbone=2.562306e-05 elapsed=34:53:54 avg_step=3.70s recent_step=3.04s eta=16:29:54 +2026-09-19 04:23:27 | INFO | train rank=0 | step=33960/50000 progress=67.9% loss=0.075970 lr=2.559419e-05 lr_backbone=2.559419e-05 elapsed=34:54:50 avg_step=3.70s recent_step=5.65s eta=16:29:26 +2026-09-19 04:24:01 | INFO | train rank=0 | step=33970/50000 progress=67.9% loss=0.091739 lr=2.556533e-05 lr_backbone=2.556533e-05 elapsed=34:55:24 avg_step=3.70s recent_step=3.37s eta=16:28:47 +2026-09-19 04:24:26 | INFO | train rank=0 | step=33980/50000 progress=68.0% loss=0.095841 lr=2.553649e-05 lr_backbone=2.553649e-05 elapsed=34:55:50 avg_step=3.70s recent_step=2.57s eta=16:28:05 +2026-09-19 04:25:02 | INFO | train rank=0 | step=33990/50000 progress=68.0% loss=0.081146 lr=2.550765e-05 lr_backbone=2.550765e-05 elapsed=34:56:25 avg_step=3.70s recent_step=3.55s eta=16:27:27 +2026-09-19 04:26:00 | INFO | train rank=0 | step=34000/50000 progress=68.0% loss=0.073233 lr=2.547882e-05 lr_backbone=2.547882e-05 elapsed=34:57:23 avg_step=3.70s recent_step=5.78s eta=16:27:00 +2026-09-19 04:26:35 | INFO | train rank=0 | step=34010/50000 progress=68.0% loss=0.085408 lr=2.545001e-05 lr_backbone=2.545001e-05 elapsed=34:57:59 avg_step=3.70s recent_step=3.55s eta=16:26:22 +2026-09-19 04:27:06 | INFO | train rank=0 | step=34020/50000 progress=68.0% loss=0.090531 lr=2.542121e-05 lr_backbone=2.542121e-05 elapsed=34:58:30 avg_step=3.70s recent_step=3.12s eta=16:25:43 +2026-09-19 04:27:39 | INFO | train rank=0 | step=34030/50000 progress=68.1% loss=0.083802 lr=2.539241e-05 lr_backbone=2.539241e-05 elapsed=34:59:02 avg_step=3.70s recent_step=3.24s eta=16:25:03 +2026-09-19 04:28:30 | INFO | train rank=0 | step=34040/50000 progress=68.1% loss=0.118709 lr=2.536363e-05 lr_backbone=2.536363e-05 elapsed=34:59:53 avg_step=3.70s recent_step=5.08s eta=16:24:33 +2026-09-19 04:29:01 | INFO | train rank=0 | step=34050/50000 progress=68.1% loss=0.107457 lr=2.533486e-05 lr_backbone=2.533486e-05 elapsed=35:00:24 avg_step=3.70s recent_step=3.16s eta=16:23:53 +2026-09-19 04:29:36 | INFO | train rank=0 | step=34060/50000 progress=68.1% loss=0.084890 lr=2.530610e-05 lr_backbone=2.530610e-05 elapsed=35:00:59 avg_step=3.70s recent_step=3.45s eta=16:23:15 +2026-09-19 04:30:04 | INFO | train rank=0 | step=34070/50000 progress=68.1% loss=0.063082 lr=2.527735e-05 lr_backbone=2.527735e-05 elapsed=35:01:27 avg_step=3.70s recent_step=2.85s eta=16:22:34 +2026-09-19 04:31:02 | INFO | train rank=0 | step=34080/50000 progress=68.2% loss=0.072608 lr=2.524861e-05 lr_backbone=2.524861e-05 elapsed=35:02:25 avg_step=3.70s recent_step=5.77s eta=16:22:07 +2026-09-19 04:31:32 | INFO | train rank=0 | step=34090/50000 progress=68.2% loss=0.085981 lr=2.521989e-05 lr_backbone=2.521989e-05 elapsed=35:02:55 avg_step=3.70s recent_step=3.00s eta=16:21:26 +2026-09-19 04:32:08 | INFO | train rank=0 | step=34100/50000 progress=68.2% loss=0.080640 lr=2.519117e-05 lr_backbone=2.519117e-05 elapsed=35:03:31 avg_step=3.70s recent_step=3.59s eta=16:20:49 +2026-09-19 04:32:43 | INFO | train rank=0 | step=34110/50000 progress=68.2% loss=0.084660 lr=2.516246e-05 lr_backbone=2.516246e-05 elapsed=35:04:06 avg_step=3.70s recent_step=3.55s eta=16:20:11 +2026-09-19 04:33:37 | INFO | train rank=0 | step=34120/50000 progress=68.2% loss=0.090540 lr=2.513377e-05 lr_backbone=2.513377e-05 elapsed=35:05:00 avg_step=3.70s recent_step=5.40s eta=16:19:42 +2026-09-19 04:34:04 | INFO | train rank=0 | step=34130/50000 progress=68.3% loss=0.075034 lr=2.510508e-05 lr_backbone=2.510508e-05 elapsed=35:05:27 avg_step=3.70s recent_step=2.65s eta=16:19:00 +2026-09-19 04:34:37 | INFO | train rank=0 | step=34140/50000 progress=68.3% loss=0.078395 lr=2.507641e-05 lr_backbone=2.507641e-05 elapsed=35:06:00 avg_step=3.70s recent_step=3.33s eta=16:18:21 +2026-09-19 04:35:11 | INFO | train rank=0 | step=34150/50000 progress=68.3% loss=0.097970 lr=2.504775e-05 lr_backbone=2.504775e-05 elapsed=35:06:35 avg_step=3.70s recent_step=3.43s eta=16:17:43 +2026-09-19 04:36:04 | INFO | train rank=0 | step=34160/50000 progress=68.3% loss=0.074133 lr=2.501910e-05 lr_backbone=2.501910e-05 elapsed=35:07:27 avg_step=3.70s recent_step=5.27s eta=16:17:13 +2026-09-19 04:36:33 | INFO | train rank=0 | step=34170/50000 progress=68.3% loss=0.098921 lr=2.499045e-05 lr_backbone=2.499045e-05 elapsed=35:07:56 avg_step=3.70s recent_step=2.90s eta=16:16:33 +2026-09-19 04:37:06 | INFO | train rank=0 | step=34180/50000 progress=68.4% loss=0.085459 lr=2.496182e-05 lr_backbone=2.496182e-05 elapsed=35:08:29 avg_step=3.70s recent_step=3.30s eta=16:15:54 +2026-09-19 04:37:42 | INFO | train rank=0 | step=34190/50000 progress=68.4% loss=0.113445 lr=2.493321e-05 lr_backbone=2.493321e-05 elapsed=35:09:06 avg_step=3.70s recent_step=3.65s eta=16:15:16 +2026-09-19 04:38:35 | INFO | train rank=0 | step=34200/50000 progress=68.4% loss=0.079217 lr=2.490460e-05 lr_backbone=2.490460e-05 elapsed=35:09:58 avg_step=3.70s recent_step=5.22s eta=16:14:46 +2026-09-19 04:39:06 | INFO | train rank=0 | step=34210/50000 progress=68.4% loss=0.091070 lr=2.487600e-05 lr_backbone=2.487600e-05 elapsed=35:10:29 avg_step=3.70s recent_step=3.14s eta=16:14:07 +2026-09-19 04:39:34 | INFO | train rank=0 | step=34220/50000 progress=68.4% loss=0.080729 lr=2.484741e-05 lr_backbone=2.484741e-05 elapsed=35:10:57 avg_step=3.70s recent_step=2.78s eta=16:13:26 +2026-09-19 04:40:00 | INFO | train rank=0 | step=34230/50000 progress=68.5% loss=0.075203 lr=2.481884e-05 lr_backbone=2.481884e-05 elapsed=35:11:24 avg_step=3.70s recent_step=2.66s eta=16:12:44 +2026-09-19 04:40:55 | INFO | train rank=0 | step=34240/50000 progress=68.5% loss=0.121140 lr=2.479028e-05 lr_backbone=2.479028e-05 elapsed=35:12:19 avg_step=3.70s recent_step=5.49s eta=16:12:15 +2026-09-19 04:41:24 | INFO | train rank=0 | step=34250/50000 progress=68.5% loss=0.093383 lr=2.476172e-05 lr_backbone=2.476172e-05 elapsed=35:12:47 avg_step=3.70s recent_step=2.88s eta=16:11:34 +2026-09-19 04:41:59 | INFO | train rank=0 | step=34260/50000 progress=68.5% loss=0.083659 lr=2.473318e-05 lr_backbone=2.473318e-05 elapsed=35:13:22 avg_step=3.70s recent_step=3.45s eta=16:10:56 +2026-09-19 04:42:20 | INFO | train rank=0 | step=34270/50000 progress=68.5% loss=0.078482 lr=2.470465e-05 lr_backbone=2.470465e-05 elapsed=35:13:43 avg_step=3.70s recent_step=2.10s eta=16:10:12 +2026-09-19 04:43:19 | INFO | train rank=0 | step=34280/50000 progress=68.6% loss=0.120557 lr=2.467613e-05 lr_backbone=2.467613e-05 elapsed=35:14:43 avg_step=3.70s recent_step=5.98s eta=16:09:45 +2026-09-19 04:43:56 | INFO | train rank=0 | step=34290/50000 progress=68.6% loss=0.075133 lr=2.464762e-05 lr_backbone=2.464762e-05 elapsed=35:15:19 avg_step=3.70s recent_step=3.62s eta=16:09:08 +2026-09-19 04:44:29 | INFO | train rank=0 | step=34300/50000 progress=68.6% loss=0.089457 lr=2.461912e-05 lr_backbone=2.461912e-05 elapsed=35:15:53 avg_step=3.70s recent_step=3.37s eta=16:08:29 +2026-09-19 04:45:01 | INFO | train rank=0 | step=34310/50000 progress=68.6% loss=0.094787 lr=2.459064e-05 lr_backbone=2.459064e-05 elapsed=35:16:25 avg_step=3.70s recent_step=3.19s eta=16:07:50 +2026-09-19 04:45:56 | INFO | train rank=0 | step=34320/50000 progress=68.6% loss=0.113246 lr=2.456216e-05 lr_backbone=2.456216e-05 elapsed=35:17:19 avg_step=3.70s recent_step=5.46s eta=16:07:21 +2026-09-19 04:46:26 | INFO | train rank=0 | step=34330/50000 progress=68.7% loss=0.084237 lr=2.453370e-05 lr_backbone=2.453370e-05 elapsed=35:17:49 avg_step=3.70s recent_step=3.03s eta=16:06:41 +2026-09-19 04:47:01 | INFO | train rank=0 | step=34340/50000 progress=68.7% loss=0.068501 lr=2.450525e-05 lr_backbone=2.450525e-05 elapsed=35:18:24 avg_step=3.70s recent_step=3.47s eta=16:06:03 +2026-09-19 04:47:32 | INFO | train rank=0 | step=34350/50000 progress=68.7% loss=0.076896 lr=2.447680e-05 lr_backbone=2.447680e-05 elapsed=35:18:55 avg_step=3.70s recent_step=3.07s eta=16:05:23 +2026-09-19 04:48:21 | INFO | train rank=0 | step=34360/50000 progress=68.7% loss=0.070160 lr=2.444837e-05 lr_backbone=2.444837e-05 elapsed=35:19:45 avg_step=3.70s recent_step=4.97s eta=16:04:52 +2026-09-19 04:48:51 | INFO | train rank=0 | step=34370/50000 progress=68.7% loss=0.077954 lr=2.441995e-05 lr_backbone=2.441995e-05 elapsed=35:20:14 avg_step=3.70s recent_step=2.93s eta=16:04:11 +2026-09-19 04:49:26 | INFO | train rank=0 | step=34380/50000 progress=68.8% loss=0.093856 lr=2.439154e-05 lr_backbone=2.439154e-05 elapsed=35:20:50 avg_step=3.70s recent_step=3.57s eta=16:03:34 +2026-09-19 04:50:03 | INFO | train rank=0 | step=34390/50000 progress=68.8% loss=0.118074 lr=2.436315e-05 lr_backbone=2.436315e-05 elapsed=35:21:26 avg_step=3.70s recent_step=3.65s eta=16:02:56 +2026-09-19 04:50:49 | INFO | train rank=0 | step=34400/50000 progress=68.8% loss=0.090984 lr=2.433476e-05 lr_backbone=2.433476e-05 elapsed=35:22:13 avg_step=3.70s recent_step=4.65s eta=16:02:24 +2026-09-19 04:51:29 | INFO | train rank=0 | step=34410/50000 progress=68.8% loss=0.104758 lr=2.430639e-05 lr_backbone=2.430639e-05 elapsed=35:22:53 avg_step=3.70s recent_step=4.02s eta=16:01:48 +2026-09-19 04:51:53 | INFO | train rank=0 | step=34420/50000 progress=68.8% loss=0.081549 lr=2.427802e-05 lr_backbone=2.427802e-05 elapsed=35:23:16 avg_step=3.70s recent_step=2.36s eta=16:01:05 +2026-09-19 04:52:29 | INFO | train rank=0 | step=34430/50000 progress=68.9% loss=0.075298 lr=2.424967e-05 lr_backbone=2.424967e-05 elapsed=35:23:53 avg_step=3.70s recent_step=3.63s eta=16:00:28 +2026-09-19 04:53:25 | INFO | train rank=0 | step=34440/50000 progress=68.9% loss=0.111489 lr=2.422133e-05 lr_backbone=2.422133e-05 elapsed=35:24:48 avg_step=3.70s recent_step=5.52s eta=15:59:59 +2026-09-19 04:53:59 | INFO | train rank=0 | step=34450/50000 progress=68.9% loss=0.130558 lr=2.419300e-05 lr_backbone=2.419300e-05 elapsed=35:25:22 avg_step=3.70s recent_step=3.40s eta=15:59:20 +2026-09-19 04:54:23 | INFO | train rank=0 | step=34460/50000 progress=68.9% loss=0.097569 lr=2.416468e-05 lr_backbone=2.416468e-05 elapsed=35:25:46 avg_step=3.70s recent_step=2.45s eta=15:58:38 +2026-09-19 04:54:58 | INFO | train rank=0 | step=34470/50000 progress=68.9% loss=0.086850 lr=2.413637e-05 lr_backbone=2.413637e-05 elapsed=35:26:22 avg_step=3.70s recent_step=3.53s eta=15:58:00 +2026-09-19 04:55:50 | INFO | train rank=0 | step=34480/50000 progress=69.0% loss=0.111983 lr=2.410808e-05 lr_backbone=2.410808e-05 elapsed=35:27:13 avg_step=3.70s recent_step=5.17s eta=15:57:30 +2026-09-19 04:56:22 | INFO | train rank=0 | step=34490/50000 progress=69.0% loss=0.105016 lr=2.407979e-05 lr_backbone=2.407979e-05 elapsed=35:27:45 avg_step=3.70s recent_step=3.16s eta=15:56:50 +2026-09-19 04:56:52 | INFO | train rank=0 | step=34500/50000 progress=69.0% loss=0.096549 lr=2.405152e-05 lr_backbone=2.405152e-05 elapsed=35:28:15 avg_step=3.70s recent_step=3.01s eta=15:56:10 +2026-09-19 04:57:24 | INFO | train rank=0 | step=34510/50000 progress=69.0% loss=0.095513 lr=2.402326e-05 lr_backbone=2.402326e-05 elapsed=35:28:47 avg_step=3.70s recent_step=3.17s eta=15:55:31 +2026-09-19 04:58:08 | INFO | train rank=0 | step=34520/50000 progress=69.0% loss=0.088026 lr=2.399501e-05 lr_backbone=2.399501e-05 elapsed=35:29:31 avg_step=3.70s recent_step=4.46s eta=15:54:57 +2026-09-19 04:58:47 | INFO | train rank=0 | step=34530/50000 progress=69.1% loss=0.079439 lr=2.396677e-05 lr_backbone=2.396677e-05 elapsed=35:30:11 avg_step=3.70s recent_step=3.91s eta=15:54:21 +2026-09-19 04:59:18 | INFO | train rank=0 | step=34540/50000 progress=69.1% loss=0.104883 lr=2.393854e-05 lr_backbone=2.393854e-05 elapsed=35:30:42 avg_step=3.70s recent_step=3.11s eta=15:53:41 +2026-09-19 04:59:41 | INFO | train rank=0 | step=34550/50000 progress=69.1% loss=0.084380 lr=2.391033e-05 lr_backbone=2.391033e-05 elapsed=35:31:05 avg_step=3.70s recent_step=2.31s eta=15:52:58 +2026-09-19 05:00:37 | INFO | train rank=0 | step=34560/50000 progress=69.1% loss=0.139667 lr=2.388212e-05 lr_backbone=2.388212e-05 elapsed=35:32:00 avg_step=3.70s recent_step=5.53s eta=15:52:29 +2026-09-19 05:01:09 | INFO | train rank=0 | step=34570/50000 progress=69.1% loss=0.093382 lr=2.385393e-05 lr_backbone=2.385393e-05 elapsed=35:32:32 avg_step=3.70s recent_step=3.22s eta=15:51:50 +2026-09-19 05:01:40 | INFO | train rank=0 | step=34580/50000 progress=69.2% loss=0.096995 lr=2.382575e-05 lr_backbone=2.382575e-05 elapsed=35:33:03 avg_step=3.70s recent_step=3.12s eta=15:51:10 +2026-09-19 05:02:14 | INFO | train rank=0 | step=34590/50000 progress=69.2% loss=0.092213 lr=2.379758e-05 lr_backbone=2.379758e-05 elapsed=35:33:37 avg_step=3.70s recent_step=3.36s eta=15:50:32 +2026-09-19 05:03:09 | INFO | train rank=0 | step=34600/50000 progress=69.2% loss=0.076567 lr=2.376942e-05 lr_backbone=2.376942e-05 elapsed=35:34:33 avg_step=3.70s recent_step=5.58s eta=15:50:03 +2026-09-19 05:03:45 | INFO | train rank=0 | step=34610/50000 progress=69.2% loss=0.106326 lr=2.374127e-05 lr_backbone=2.374127e-05 elapsed=35:35:08 avg_step=3.70s recent_step=3.51s eta=15:49:25 +2026-09-19 05:04:13 | INFO | train rank=0 | step=34620/50000 progress=69.2% loss=0.097046 lr=2.371313e-05 lr_backbone=2.371313e-05 elapsed=35:35:36 avg_step=3.70s recent_step=2.81s eta=15:48:44 +2026-09-19 05:04:49 | INFO | train rank=0 | step=34630/50000 progress=69.3% loss=0.093358 lr=2.368501e-05 lr_backbone=2.368501e-05 elapsed=35:36:12 avg_step=3.70s recent_step=3.61s eta=15:48:07 +2026-09-19 05:05:41 | INFO | train rank=0 | step=34640/50000 progress=69.3% loss=0.119095 lr=2.365689e-05 lr_backbone=2.365689e-05 elapsed=35:37:04 avg_step=3.70s recent_step=5.19s eta=15:47:37 +2026-09-19 05:06:10 | INFO | train rank=0 | step=34650/50000 progress=69.3% loss=0.077771 lr=2.362879e-05 lr_backbone=2.362879e-05 elapsed=35:37:33 avg_step=3.70s recent_step=2.91s eta=15:46:56 +2026-09-19 05:06:36 | INFO | train rank=0 | step=34660/50000 progress=69.3% loss=0.085415 lr=2.360070e-05 lr_backbone=2.360070e-05 elapsed=35:38:00 avg_step=3.70s recent_step=2.66s eta=15:46:14 +2026-09-19 05:07:07 | INFO | train rank=0 | step=34670/50000 progress=69.3% loss=0.089574 lr=2.357262e-05 lr_backbone=2.357262e-05 elapsed=35:38:31 avg_step=3.70s recent_step=3.12s eta=15:45:35 +2026-09-19 05:07:59 | INFO | train rank=0 | step=34680/50000 progress=69.4% loss=0.127944 lr=2.354456e-05 lr_backbone=2.354456e-05 elapsed=35:39:23 avg_step=3.70s recent_step=5.21s eta=15:45:04 +2026-09-19 05:08:29 | INFO | train rank=0 | step=34690/50000 progress=69.4% loss=0.088120 lr=2.351650e-05 lr_backbone=2.351650e-05 elapsed=35:39:52 avg_step=3.70s recent_step=2.94s eta=15:44:24 +2026-09-19 05:09:03 | INFO | train rank=0 | step=34700/50000 progress=69.4% loss=0.094255 lr=2.348846e-05 lr_backbone=2.348846e-05 elapsed=35:40:26 avg_step=3.70s recent_step=3.39s eta=15:43:46 +2026-09-19 05:09:38 | INFO | train rank=0 | step=34710/50000 progress=69.4% loss=0.085974 lr=2.346043e-05 lr_backbone=2.346043e-05 elapsed=35:41:01 avg_step=3.70s recent_step=3.48s eta=15:43:08 +2026-09-19 05:10:22 | INFO | train rank=0 | step=34720/50000 progress=69.4% loss=0.081090 lr=2.343241e-05 lr_backbone=2.343241e-05 elapsed=35:41:45 avg_step=3.70s recent_step=4.45s eta=15:42:34 +2026-09-19 05:10:45 | INFO | train rank=0 | step=34730/50000 progress=69.5% loss=0.086504 lr=2.340440e-05 lr_backbone=2.340440e-05 elapsed=35:42:08 avg_step=3.70s recent_step=2.25s eta=15:41:51 +2026-09-19 05:11:16 | INFO | train rank=0 | step=34740/50000 progress=69.5% loss=0.071323 lr=2.337640e-05 lr_backbone=2.337640e-05 elapsed=35:42:40 avg_step=3.70s recent_step=3.18s eta=15:41:11 +2026-09-19 05:11:55 | INFO | train rank=0 | step=34750/50000 progress=69.5% loss=0.110981 lr=2.334841e-05 lr_backbone=2.334841e-05 elapsed=35:43:18 avg_step=3.70s recent_step=3.86s eta=15:40:35 +2026-09-19 05:12:39 | INFO | train rank=0 | step=34760/50000 progress=69.5% loss=0.101791 lr=2.332044e-05 lr_backbone=2.332044e-05 elapsed=35:44:03 avg_step=3.70s recent_step=4.42s eta=15:40:01 +2026-09-19 05:13:22 | INFO | train rank=0 | step=34770/50000 progress=69.5% loss=0.089204 lr=2.329248e-05 lr_backbone=2.329248e-05 elapsed=35:44:46 avg_step=3.70s recent_step=4.31s eta=15:39:27 +2026-09-19 05:13:58 | INFO | train rank=0 | step=34780/50000 progress=69.6% loss=0.117662 lr=2.326453e-05 lr_backbone=2.326453e-05 elapsed=35:45:21 avg_step=3.70s recent_step=3.54s eta=15:38:49 +2026-09-19 05:14:30 | INFO | train rank=0 | step=34790/50000 progress=69.6% loss=0.072976 lr=2.323659e-05 lr_backbone=2.323659e-05 elapsed=35:45:53 avg_step=3.70s recent_step=3.22s eta=15:38:10 +2026-09-19 05:15:12 | INFO | train rank=0 | step=34800/50000 progress=69.6% loss=0.085463 lr=2.320866e-05 lr_backbone=2.320866e-05 elapsed=35:46:36 avg_step=3.70s recent_step=4.23s eta=15:37:35 +2026-09-19 05:16:06 | INFO | train rank=0 | step=34810/50000 progress=69.6% loss=0.089411 lr=2.318074e-05 lr_backbone=2.318074e-05 elapsed=35:47:30 avg_step=3.70s recent_step=5.42s eta=15:37:06 +2026-09-19 05:16:38 | INFO | train rank=0 | step=34820/50000 progress=69.6% loss=0.087187 lr=2.315284e-05 lr_backbone=2.315284e-05 elapsed=35:48:01 avg_step=3.70s recent_step=3.15s eta=15:36:26 +2026-09-19 05:17:12 | INFO | train rank=0 | step=34830/50000 progress=69.7% loss=0.083701 lr=2.312495e-05 lr_backbone=2.312495e-05 elapsed=35:48:35 avg_step=3.70s recent_step=3.40s eta=15:35:48 +2026-09-19 05:17:43 | INFO | train rank=0 | step=34840/50000 progress=69.7% loss=0.096640 lr=2.309707e-05 lr_backbone=2.309707e-05 elapsed=35:49:06 avg_step=3.70s recent_step=3.09s eta=15:35:08 +2026-09-19 05:18:30 | INFO | train rank=0 | step=34850/50000 progress=69.7% loss=0.090578 lr=2.306920e-05 lr_backbone=2.306920e-05 elapsed=35:49:53 avg_step=3.70s recent_step=4.68s eta=15:34:36 +2026-09-19 05:19:02 | INFO | train rank=0 | step=34860/50000 progress=69.7% loss=0.071274 lr=2.304134e-05 lr_backbone=2.304134e-05 elapsed=35:50:25 avg_step=3.70s recent_step=3.23s eta=15:33:57 +2026-09-19 05:19:37 | INFO | train rank=0 | step=34870/50000 progress=69.7% loss=0.110440 lr=2.301350e-05 lr_backbone=2.301350e-05 elapsed=35:51:01 avg_step=3.70s recent_step=3.54s eta=15:33:19 +2026-09-19 05:20:12 | INFO | train rank=0 | step=34880/50000 progress=69.8% loss=0.120383 lr=2.298566e-05 lr_backbone=2.298566e-05 elapsed=35:51:35 avg_step=3.70s recent_step=3.42s eta=15:32:41 +2026-09-19 05:20:57 | INFO | train rank=0 | step=34890/50000 progress=69.8% loss=0.135024 lr=2.295784e-05 lr_backbone=2.295784e-05 elapsed=35:52:20 avg_step=3.70s recent_step=4.52s eta=15:32:07 +2026-09-19 05:21:24 | INFO | train rank=0 | step=34900/50000 progress=69.8% loss=0.088803 lr=2.293003e-05 lr_backbone=2.293003e-05 elapsed=35:52:48 avg_step=3.70s recent_step=2.76s eta=15:31:26 +2026-09-19 05:21:59 | INFO | train rank=0 | step=34910/50000 progress=69.8% loss=0.130185 lr=2.290224e-05 lr_backbone=2.290224e-05 elapsed=35:53:22 avg_step=3.70s recent_step=3.43s eta=15:30:48 +2026-09-19 05:22:34 | INFO | train rank=0 | step=34920/50000 progress=69.8% loss=0.080159 lr=2.287445e-05 lr_backbone=2.287445e-05 elapsed=35:53:58 avg_step=3.70s recent_step=3.58s eta=15:30:10 +2026-09-19 05:23:20 | INFO | train rank=0 | step=34930/50000 progress=69.9% loss=0.142868 lr=2.284668e-05 lr_backbone=2.284668e-05 elapsed=35:54:43 avg_step=3.70s recent_step=4.58s eta=15:29:37 +2026-09-19 05:24:01 | INFO | train rank=0 | step=34940/50000 progress=69.9% loss=0.106134 lr=2.281891e-05 lr_backbone=2.281891e-05 elapsed=35:55:25 avg_step=3.70s recent_step=4.11s eta=15:29:02 +2026-09-19 05:24:34 | INFO | train rank=0 | step=34950/50000 progress=69.9% loss=0.077518 lr=2.279116e-05 lr_backbone=2.279116e-05 elapsed=35:55:57 avg_step=3.70s recent_step=3.25s eta=15:28:23 +2026-09-19 05:25:09 | INFO | train rank=0 | step=34960/50000 progress=69.9% loss=0.102300 lr=2.276343e-05 lr_backbone=2.276343e-05 elapsed=35:56:33 avg_step=3.70s recent_step=3.56s eta=15:27:45 +2026-09-19 05:26:03 | INFO | train rank=0 | step=34970/50000 progress=69.9% loss=0.087624 lr=2.273570e-05 lr_backbone=2.273570e-05 elapsed=35:57:26 avg_step=3.70s recent_step=5.34s eta=15:27:15 +2026-09-19 05:26:38 | INFO | train rank=0 | step=34980/50000 progress=70.0% loss=0.075364 lr=2.270799e-05 lr_backbone=2.270799e-05 elapsed=35:58:01 avg_step=3.70s recent_step=3.51s eta=15:26:37 +2026-09-19 05:26:59 | INFO | train rank=0 | step=34990/50000 progress=70.0% loss=0.089426 lr=2.268028e-05 lr_backbone=2.268028e-05 elapsed=35:58:23 avg_step=3.70s recent_step=2.14s eta=15:25:54 +2026-09-19 05:27:32 | INFO | train rank=0 | step=35000/50000 progress=70.0% loss=0.091584 lr=2.265259e-05 lr_backbone=2.265259e-05 elapsed=35:58:55 avg_step=3.70s recent_step=3.24s eta=15:25:15 +2026-09-19 05:28:25 | INFO | train rank=0 | step=35010/50000 progress=70.0% loss=0.086577 lr=2.262491e-05 lr_backbone=2.262491e-05 elapsed=35:59:48 avg_step=3.70s recent_step=5.32s eta=15:24:45 +2026-09-19 05:29:03 | INFO | train rank=0 | step=35020/50000 progress=70.0% loss=0.073896 lr=2.259725e-05 lr_backbone=2.259725e-05 elapsed=36:00:26 avg_step=3.70s recent_step=3.84s eta=15:24:08 +2026-09-19 05:29:29 | INFO | train rank=0 | step=35030/50000 progress=70.1% loss=0.081090 lr=2.256959e-05 lr_backbone=2.256959e-05 elapsed=36:00:52 avg_step=3.70s recent_step=2.56s eta=15:23:26 +2026-09-19 05:30:00 | INFO | train rank=0 | step=35040/50000 progress=70.1% loss=0.079350 lr=2.254195e-05 lr_backbone=2.254195e-05 elapsed=36:01:23 avg_step=3.70s recent_step=3.07s eta=15:22:47 +2026-09-19 05:30:53 | INFO | train rank=0 | step=35050/50000 progress=70.1% loss=0.098118 lr=2.251432e-05 lr_backbone=2.251432e-05 elapsed=36:02:17 avg_step=3.70s recent_step=5.37s eta=15:22:17 +2026-09-19 05:31:28 | INFO | train rank=0 | step=35060/50000 progress=70.1% loss=0.104536 lr=2.248670e-05 lr_backbone=2.248670e-05 elapsed=36:02:51 avg_step=3.70s recent_step=3.47s eta=15:21:39 +2026-09-19 05:31:55 | INFO | train rank=0 | step=35070/50000 progress=70.1% loss=0.096242 lr=2.245909e-05 lr_backbone=2.245909e-05 elapsed=36:03:18 avg_step=3.70s recent_step=2.67s eta=15:20:57 +2026-09-19 05:32:28 | INFO | train rank=0 | step=35080/50000 progress=70.2% loss=0.085810 lr=2.243150e-05 lr_backbone=2.243150e-05 elapsed=36:03:52 avg_step=3.70s recent_step=3.37s eta=15:20:19 +2026-09-19 05:33:24 | INFO | train rank=0 | step=35090/50000 progress=70.2% loss=0.087102 lr=2.240392e-05 lr_backbone=2.240392e-05 elapsed=36:04:47 avg_step=3.70s recent_step=5.54s eta=15:19:50 +2026-09-19 05:34:01 | INFO | train rank=0 | step=35100/50000 progress=70.2% loss=0.098209 lr=2.237635e-05 lr_backbone=2.237635e-05 elapsed=36:05:25 avg_step=3.70s recent_step=3.76s eta=15:19:13 +2026-09-19 05:34:30 | INFO | train rank=0 | step=35110/50000 progress=70.2% loss=0.068689 lr=2.234879e-05 lr_backbone=2.234879e-05 elapsed=36:05:53 avg_step=3.70s recent_step=2.85s eta=15:18:32 +2026-09-19 05:35:04 | INFO | train rank=0 | step=35120/50000 progress=70.2% loss=0.081105 lr=2.232124e-05 lr_backbone=2.232124e-05 elapsed=36:06:27 avg_step=3.70s recent_step=3.45s eta=15:17:54 +2026-09-19 05:35:56 | INFO | train rank=0 | step=35130/50000 progress=70.3% loss=0.087320 lr=2.229371e-05 lr_backbone=2.229371e-05 elapsed=36:07:19 avg_step=3.70s recent_step=5.18s eta=15:17:23 +2026-09-19 05:36:25 | INFO | train rank=0 | step=35140/50000 progress=70.3% loss=0.069709 lr=2.226619e-05 lr_backbone=2.226619e-05 elapsed=36:07:48 avg_step=3.70s recent_step=2.88s eta=15:16:43 +2026-09-19 05:37:03 | INFO | train rank=0 | step=35150/50000 progress=70.3% loss=0.122310 lr=2.223868e-05 lr_backbone=2.223868e-05 elapsed=36:08:26 avg_step=3.70s recent_step=3.79s eta=15:16:06 +2026-09-19 05:37:30 | INFO | train rank=0 | step=35160/50000 progress=70.3% loss=0.095154 lr=2.221118e-05 lr_backbone=2.221118e-05 elapsed=36:08:53 avg_step=3.70s recent_step=2.70s eta=15:15:25 +2026-09-19 05:38:16 | INFO | train rank=0 | step=35170/50000 progress=70.3% loss=0.111960 lr=2.218369e-05 lr_backbone=2.218369e-05 elapsed=36:09:39 avg_step=3.70s recent_step=4.63s eta=15:14:52 +2026-09-19 05:38:41 | INFO | train rank=0 | step=35180/50000 progress=70.4% loss=0.099163 lr=2.215622e-05 lr_backbone=2.215622e-05 elapsed=36:10:04 avg_step=3.70s recent_step=2.50s eta=15:14:10 +2026-09-19 05:39:34 | INFO | train rank=0 | step=35190/50000 progress=70.4% loss=0.112381 lr=2.212876e-05 lr_backbone=2.212876e-05 elapsed=36:10:57 avg_step=3.70s recent_step=5.29s eta=15:13:40 +2026-09-19 05:40:06 | INFO | train rank=0 | step=35200/50000 progress=70.4% loss=0.102310 lr=2.210131e-05 lr_backbone=2.210131e-05 elapsed=36:11:29 avg_step=3.70s recent_step=3.22s eta=15:13:01 +2026-09-19 05:40:38 | INFO | train rank=0 | step=35210/50000 progress=70.4% loss=0.083620 lr=2.207387e-05 lr_backbone=2.207387e-05 elapsed=36:12:01 avg_step=3.70s recent_step=3.16s eta=15:12:21 +2026-09-19 05:41:11 | INFO | train rank=0 | step=35220/50000 progress=70.4% loss=0.082791 lr=2.204645e-05 lr_backbone=2.204645e-05 elapsed=36:12:34 avg_step=3.70s recent_step=3.33s eta=15:11:43 +2026-09-19 05:42:04 | INFO | train rank=0 | step=35230/50000 progress=70.5% loss=0.075220 lr=2.201904e-05 lr_backbone=2.201904e-05 elapsed=36:13:27 avg_step=3.70s recent_step=5.25s eta=15:11:12 +2026-09-19 05:42:36 | INFO | train rank=0 | step=35240/50000 progress=70.5% loss=0.116677 lr=2.199164e-05 lr_backbone=2.199164e-05 elapsed=36:13:59 avg_step=3.70s recent_step=3.24s eta=15:10:33 +2026-09-19 05:43:05 | INFO | train rank=0 | step=35250/50000 progress=70.5% loss=0.072391 lr=2.196425e-05 lr_backbone=2.196425e-05 elapsed=36:14:28 avg_step=3.70s recent_step=2.89s eta=15:09:53 +2026-09-19 05:43:52 | INFO | train rank=0 | step=35260/50000 progress=70.5% loss=0.089136 lr=2.193687e-05 lr_backbone=2.193687e-05 elapsed=36:15:15 avg_step=3.70s recent_step=4.72s eta=15:09:20 +2026-09-19 05:44:27 | INFO | train rank=0 | step=35270/50000 progress=70.5% loss=0.071887 lr=2.190951e-05 lr_backbone=2.190951e-05 elapsed=36:15:50 avg_step=3.70s recent_step=3.50s eta=15:08:42 +2026-09-19 05:44:54 | INFO | train rank=0 | step=35280/50000 progress=70.6% loss=0.101730 lr=2.188216e-05 lr_backbone=2.188216e-05 elapsed=36:16:17 avg_step=3.70s recent_step=2.66s eta=15:08:01 +2026-09-19 05:45:29 | INFO | train rank=0 | step=35290/50000 progress=70.6% loss=0.098569 lr=2.185482e-05 lr_backbone=2.185482e-05 elapsed=36:16:52 avg_step=3.70s recent_step=3.54s eta=15:07:23 +2026-09-19 05:46:27 | INFO | train rank=0 | step=35300/50000 progress=70.6% loss=0.076180 lr=2.182749e-05 lr_backbone=2.182749e-05 elapsed=36:17:51 avg_step=3.70s recent_step=5.82s eta=15:06:55 +2026-09-19 05:46:56 | INFO | train rank=0 | step=35310/50000 progress=70.6% loss=0.126787 lr=2.180018e-05 lr_backbone=2.180018e-05 elapsed=36:18:19 avg_step=3.70s recent_step=2.88s eta=15:06:15 +2026-09-19 05:47:32 | INFO | train rank=0 | step=35320/50000 progress=70.6% loss=0.094691 lr=2.177288e-05 lr_backbone=2.177288e-05 elapsed=36:18:56 avg_step=3.70s recent_step=3.63s eta=15:05:37 +2026-09-19 05:48:09 | INFO | train rank=0 | step=35330/50000 progress=70.7% loss=0.108347 lr=2.174559e-05 lr_backbone=2.174559e-05 elapsed=36:19:32 avg_step=3.70s recent_step=3.61s eta=15:05:00 +2026-09-19 05:48:56 | INFO | train rank=0 | step=35340/50000 progress=70.7% loss=0.090490 lr=2.171831e-05 lr_backbone=2.171831e-05 elapsed=36:20:20 avg_step=3.70s recent_step=4.78s eta=15:04:27 +2026-09-19 05:49:28 | INFO | train rank=0 | step=35350/50000 progress=70.7% loss=0.101016 lr=2.169104e-05 lr_backbone=2.169104e-05 elapsed=36:20:51 avg_step=3.70s recent_step=3.14s eta=15:03:48 +2026-09-19 05:50:01 | INFO | train rank=0 | step=35360/50000 progress=70.7% loss=0.179499 lr=2.166379e-05 lr_backbone=2.166379e-05 elapsed=36:21:24 avg_step=3.70s recent_step=3.28s eta=15:03:09 +2026-09-19 05:50:40 | INFO | train rank=0 | step=35370/50000 progress=70.7% loss=0.118299 lr=2.163655e-05 lr_backbone=2.163655e-05 elapsed=36:22:03 avg_step=3.70s recent_step=3.93s eta=15:02:33 +2026-09-19 05:51:28 | INFO | train rank=0 | step=35380/50000 progress=70.8% loss=0.065970 lr=2.160933e-05 lr_backbone=2.160933e-05 elapsed=36:22:51 avg_step=3.70s recent_step=4.80s eta=15:02:01 +2026-09-19 05:52:03 | INFO | train rank=0 | step=35390/50000 progress=70.8% loss=0.083103 lr=2.158211e-05 lr_backbone=2.158211e-05 elapsed=36:23:26 avg_step=3.70s recent_step=3.51s eta=15:01:23 +2026-09-19 05:52:44 | INFO | train rank=0 | step=35400/50000 progress=70.8% loss=0.082792 lr=2.155491e-05 lr_backbone=2.155491e-05 elapsed=36:24:07 avg_step=3.70s recent_step=4.11s eta=15:00:47 +2026-09-19 05:53:18 | INFO | train rank=0 | step=35410/50000 progress=70.8% loss=0.075613 lr=2.152772e-05 lr_backbone=2.152772e-05 elapsed=36:24:41 avg_step=3.70s recent_step=3.36s eta=15:00:09 +2026-09-19 05:54:07 | INFO | train rank=0 | step=35420/50000 progress=70.8% loss=0.078914 lr=2.150054e-05 lr_backbone=2.150054e-05 elapsed=36:25:30 avg_step=3.70s recent_step=4.91s eta=14:59:37 +2026-09-19 05:54:37 | INFO | train rank=0 | step=35430/50000 progress=70.9% loss=0.093142 lr=2.147337e-05 lr_backbone=2.147337e-05 elapsed=36:26:00 avg_step=3.70s recent_step=3.00s eta=14:58:57 +2026-09-19 05:55:10 | INFO | train rank=0 | step=35440/50000 progress=70.9% loss=0.081686 lr=2.144622e-05 lr_backbone=2.144622e-05 elapsed=36:26:33 avg_step=3.70s recent_step=3.28s eta=14:58:18 +2026-09-19 05:55:35 | INFO | train rank=0 | step=35450/50000 progress=70.9% loss=0.060049 lr=2.141908e-05 lr_backbone=2.141908e-05 elapsed=36:26:58 avg_step=3.70s recent_step=2.54s eta=14:57:37 +2026-09-19 05:56:22 | INFO | train rank=0 | step=35460/50000 progress=70.9% loss=0.098992 lr=2.139195e-05 lr_backbone=2.139195e-05 elapsed=36:27:45 avg_step=3.70s recent_step=4.70s eta=14:57:04 +2026-09-19 05:56:58 | INFO | train rank=0 | step=35470/50000 progress=70.9% loss=0.086669 lr=2.136484e-05 lr_backbone=2.136484e-05 elapsed=36:28:22 avg_step=3.70s recent_step=3.63s eta=14:56:26 +2026-09-19 05:57:36 | INFO | train rank=0 | step=35480/50000 progress=71.0% loss=0.105669 lr=2.133773e-05 lr_backbone=2.133773e-05 elapsed=36:29:00 avg_step=3.70s recent_step=3.80s eta=14:55:50 +2026-09-19 05:58:12 | INFO | train rank=0 | step=35490/50000 progress=71.0% loss=0.114649 lr=2.131064e-05 lr_backbone=2.131064e-05 elapsed=36:29:35 avg_step=3.70s recent_step=3.57s eta=14:55:12 +2026-09-19 05:59:03 | INFO | train rank=0 | step=35500/50000 progress=71.0% loss=0.088352 lr=2.128357e-05 lr_backbone=2.128357e-05 elapsed=36:30:26 avg_step=3.70s recent_step=5.08s eta=14:54:41 +2026-09-19 05:59:36 | INFO | train rank=0 | step=35510/50000 progress=71.0% loss=0.081558 lr=2.125650e-05 lr_backbone=2.125650e-05 elapsed=36:30:59 avg_step=3.70s recent_step=3.28s eta=14:54:02 +2026-09-19 06:00:09 | INFO | train rank=0 | step=35520/50000 progress=71.0% loss=0.065244 lr=2.122945e-05 lr_backbone=2.122945e-05 elapsed=36:31:32 avg_step=3.70s recent_step=3.35s eta=14:53:24 +2026-09-19 06:00:44 | INFO | train rank=0 | step=35530/50000 progress=71.1% loss=0.088123 lr=2.120241e-05 lr_backbone=2.120241e-05 elapsed=36:32:07 avg_step=3.70s recent_step=3.44s eta=14:52:46 +2026-09-19 06:01:35 | INFO | train rank=0 | step=35540/50000 progress=71.1% loss=0.072838 lr=2.117538e-05 lr_backbone=2.117538e-05 elapsed=36:32:58 avg_step=3.70s recent_step=5.13s eta=14:52:14 +2026-09-19 06:02:10 | INFO | train rank=0 | step=35550/50000 progress=71.1% loss=0.090171 lr=2.114837e-05 lr_backbone=2.114837e-05 elapsed=36:33:33 avg_step=3.70s recent_step=3.50s eta=14:51:36 +2026-09-19 06:02:35 | INFO | train rank=0 | step=35560/50000 progress=71.1% loss=0.083432 lr=2.112136e-05 lr_backbone=2.112136e-05 elapsed=36:33:58 avg_step=3.70s recent_step=2.54s eta=14:50:55 +2026-09-19 06:03:10 | INFO | train rank=0 | step=35570/50000 progress=71.1% loss=0.097538 lr=2.109438e-05 lr_backbone=2.109438e-05 elapsed=36:34:34 avg_step=3.70s recent_step=3.51s eta=14:50:17 +2026-09-19 06:03:57 | INFO | train rank=0 | step=35580/50000 progress=71.2% loss=0.089042 lr=2.106740e-05 lr_backbone=2.106740e-05 elapsed=36:35:21 avg_step=3.70s recent_step=4.71s eta=14:49:44 +2026-09-19 06:04:31 | INFO | train rank=0 | step=35590/50000 progress=71.2% loss=0.100024 lr=2.104043e-05 lr_backbone=2.104043e-05 elapsed=36:35:55 avg_step=3.70s recent_step=3.38s eta=14:49:06 +2026-09-19 06:05:06 | INFO | train rank=0 | step=35600/50000 progress=71.2% loss=0.079177 lr=2.101348e-05 lr_backbone=2.101348e-05 elapsed=36:36:29 avg_step=3.70s recent_step=3.43s eta=14:48:28 +2026-09-19 06:05:39 | INFO | train rank=0 | step=35610/50000 progress=71.2% loss=0.090275 lr=2.098654e-05 lr_backbone=2.098654e-05 elapsed=36:37:02 avg_step=3.70s recent_step=3.32s eta=14:47:49 +2026-09-19 06:06:31 | INFO | train rank=0 | step=35620/50000 progress=71.2% loss=0.112124 lr=2.095962e-05 lr_backbone=2.095962e-05 elapsed=36:37:54 avg_step=3.70s recent_step=5.21s eta=14:47:18 +2026-09-19 06:07:07 | INFO | train rank=0 | step=35630/50000 progress=71.3% loss=0.085392 lr=2.093270e-05 lr_backbone=2.093270e-05 elapsed=36:38:30 avg_step=3.70s recent_step=3.58s eta=14:46:41 +2026-09-19 06:07:37 | INFO | train rank=0 | step=35640/50000 progress=71.3% loss=0.072025 lr=2.090580e-05 lr_backbone=2.090580e-05 elapsed=36:39:01 avg_step=3.70s recent_step=3.07s eta=14:46:01 +2026-09-19 06:08:09 | INFO | train rank=0 | step=35650/50000 progress=71.3% loss=0.095082 lr=2.087892e-05 lr_backbone=2.087892e-05 elapsed=36:39:32 avg_step=3.70s recent_step=3.14s eta=14:45:22 +2026-09-19 06:09:02 | INFO | train rank=0 | step=35660/50000 progress=71.3% loss=0.069539 lr=2.085204e-05 lr_backbone=2.085204e-05 elapsed=36:40:25 avg_step=3.70s recent_step=5.31s eta=14:44:51 +2026-09-19 06:09:31 | INFO | train rank=0 | step=35670/50000 progress=71.3% loss=0.127841 lr=2.082518e-05 lr_backbone=2.082518e-05 elapsed=36:40:54 avg_step=3.70s recent_step=2.93s eta=14:44:11 +2026-09-19 06:10:01 | INFO | train rank=0 | step=35680/50000 progress=71.4% loss=0.084159 lr=2.079833e-05 lr_backbone=2.079833e-05 elapsed=36:41:24 avg_step=3.70s recent_step=2.99s eta=14:43:31 +2026-09-19 06:10:33 | INFO | train rank=0 | step=35690/50000 progress=71.4% loss=0.081735 lr=2.077149e-05 lr_backbone=2.077149e-05 elapsed=36:41:57 avg_step=3.70s recent_step=3.25s eta=14:42:52 +2026-09-19 06:11:26 | INFO | train rank=0 | step=35700/50000 progress=71.4% loss=0.087633 lr=2.074467e-05 lr_backbone=2.074467e-05 elapsed=36:42:49 avg_step=3.70s recent_step=5.27s eta=14:42:21 +2026-09-19 06:12:02 | INFO | train rank=0 | step=35710/50000 progress=71.4% loss=0.086486 lr=2.071785e-05 lr_backbone=2.071785e-05 elapsed=36:43:25 avg_step=3.70s recent_step=3.56s eta=14:41:44 +2026-09-19 06:12:38 | INFO | train rank=0 | step=35720/50000 progress=71.4% loss=0.128936 lr=2.069106e-05 lr_backbone=2.069106e-05 elapsed=36:44:02 avg_step=3.70s recent_step=3.67s eta=14:41:07 +2026-09-19 06:13:09 | INFO | train rank=0 | step=35730/50000 progress=71.5% loss=0.076659 lr=2.066427e-05 lr_backbone=2.066427e-05 elapsed=36:44:33 avg_step=3.70s recent_step=3.10s eta=14:40:27 +2026-09-19 06:13:56 | INFO | train rank=0 | step=35740/50000 progress=71.5% loss=0.077083 lr=2.063750e-05 lr_backbone=2.063750e-05 elapsed=36:45:20 avg_step=3.70s recent_step=4.72s eta=14:39:54 +2026-09-19 06:14:29 | INFO | train rank=0 | step=35750/50000 progress=71.5% loss=0.075007 lr=2.061074e-05 lr_backbone=2.061074e-05 elapsed=36:45:52 avg_step=3.70s recent_step=3.21s eta=14:39:15 +2026-09-19 06:15:05 | INFO | train rank=0 | step=35760/50000 progress=71.5% loss=0.093033 lr=2.058399e-05 lr_backbone=2.058399e-05 elapsed=36:46:28 avg_step=3.70s recent_step=3.66s eta=14:38:38 +2026-09-19 06:15:36 | INFO | train rank=0 | step=35770/50000 progress=71.5% loss=0.089958 lr=2.055726e-05 lr_backbone=2.055726e-05 elapsed=36:46:59 avg_step=3.70s recent_step=3.10s eta=14:37:59 +2026-09-19 06:16:24 | INFO | train rank=0 | step=35780/50000 progress=71.6% loss=0.081308 lr=2.053053e-05 lr_backbone=2.053053e-05 elapsed=36:47:47 avg_step=3.70s recent_step=4.76s eta=14:37:26 +2026-09-19 06:16:51 | INFO | train rank=0 | step=35790/50000 progress=71.6% loss=0.082059 lr=2.050383e-05 lr_backbone=2.050383e-05 elapsed=36:48:14 avg_step=3.70s recent_step=2.74s eta=14:36:45 +2026-09-19 06:17:23 | INFO | train rank=0 | step=35800/50000 progress=71.6% loss=0.097129 lr=2.047713e-05 lr_backbone=2.047713e-05 elapsed=36:48:46 avg_step=3.70s recent_step=3.20s eta=14:36:06 +2026-09-19 06:18:02 | INFO | train rank=0 | step=35810/50000 progress=71.6% loss=0.085522 lr=2.045045e-05 lr_backbone=2.045045e-05 elapsed=36:49:25 avg_step=3.70s recent_step=3.85s eta=14:35:30 +2026-09-19 06:18:53 | INFO | train rank=0 | step=35820/50000 progress=71.6% loss=0.071429 lr=2.042378e-05 lr_backbone=2.042378e-05 elapsed=36:50:16 avg_step=3.70s recent_step=5.11s eta=14:34:58 +2026-09-19 06:19:24 | INFO | train rank=0 | step=35830/50000 progress=71.7% loss=0.074247 lr=2.039712e-05 lr_backbone=2.039712e-05 elapsed=36:50:48 avg_step=3.70s recent_step=3.15s eta=14:34:19 +2026-09-19 06:20:01 | INFO | train rank=0 | step=35840/50000 progress=71.7% loss=0.122242 lr=2.037048e-05 lr_backbone=2.037048e-05 elapsed=36:51:25 avg_step=3.70s recent_step=3.71s eta=14:33:42 +2026-09-19 06:20:24 | INFO | train rank=0 | step=35850/50000 progress=71.7% loss=0.084321 lr=2.034385e-05 lr_backbone=2.034385e-05 elapsed=36:51:48 avg_step=3.70s recent_step=2.29s eta=14:32:59 +2026-09-19 06:21:06 | INFO | train rank=0 | step=35860/50000 progress=71.7% loss=0.082598 lr=2.031723e-05 lr_backbone=2.031723e-05 elapsed=36:52:30 avg_step=3.70s recent_step=4.19s eta=14:32:24 +2026-09-19 06:21:34 | INFO | train rank=0 | step=35870/50000 progress=71.7% loss=0.079541 lr=2.029062e-05 lr_backbone=2.029062e-05 elapsed=36:52:58 avg_step=3.70s recent_step=2.81s eta=14:31:44 +2026-09-19 06:22:07 | INFO | train rank=0 | step=35880/50000 progress=71.8% loss=0.117493 lr=2.026403e-05 lr_backbone=2.026403e-05 elapsed=36:53:30 avg_step=3.70s recent_step=3.27s eta=14:31:05 +2026-09-19 06:22:52 | INFO | train rank=0 | step=35890/50000 progress=71.8% loss=0.105624 lr=2.023745e-05 lr_backbone=2.023745e-05 elapsed=36:54:15 avg_step=3.70s recent_step=4.49s eta=14:30:31 +2026-09-19 06:23:43 | INFO | train rank=0 | step=35900/50000 progress=71.8% loss=0.074056 lr=2.021088e-05 lr_backbone=2.021088e-05 elapsed=36:55:07 avg_step=3.70s recent_step=5.15s eta=14:30:00 +2026-09-19 06:24:14 | INFO | train rank=0 | step=35910/50000 progress=71.8% loss=0.095125 lr=2.018433e-05 lr_backbone=2.018433e-05 elapsed=36:55:38 avg_step=3.70s recent_step=3.09s eta=14:29:20 +2026-09-19 06:24:42 | INFO | train rank=0 | step=35920/50000 progress=71.8% loss=0.084758 lr=2.015779e-05 lr_backbone=2.015779e-05 elapsed=36:56:06 avg_step=3.70s recent_step=2.81s eta=14:28:40 +2026-09-19 06:25:11 | INFO | train rank=0 | step=35930/50000 progress=71.9% loss=0.085284 lr=2.013126e-05 lr_backbone=2.013126e-05 elapsed=36:56:34 avg_step=3.70s recent_step=2.86s eta=14:28:00 +2026-09-19 06:25:57 | INFO | train rank=0 | step=35940/50000 progress=71.9% loss=0.096480 lr=2.010475e-05 lr_backbone=2.010475e-05 elapsed=36:57:21 avg_step=3.70s recent_step=4.66s eta=14:27:26 +2026-09-19 06:26:27 | INFO | train rank=0 | step=35950/50000 progress=71.9% loss=0.087578 lr=2.007825e-05 lr_backbone=2.007825e-05 elapsed=36:57:50 avg_step=3.70s recent_step=2.96s eta=14:26:46 +2026-09-19 06:27:01 | INFO | train rank=0 | step=35960/50000 progress=71.9% loss=0.112369 lr=2.005176e-05 lr_backbone=2.005176e-05 elapsed=36:58:25 avg_step=3.70s recent_step=3.43s eta=14:26:08 +2026-09-19 06:27:44 | INFO | train rank=0 | step=35970/50000 progress=71.9% loss=0.145845 lr=2.002529e-05 lr_backbone=2.002529e-05 elapsed=36:59:07 avg_step=3.70s recent_step=4.27s eta=14:25:33 +2026-09-19 06:28:32 | INFO | train rank=0 | step=35980/50000 progress=72.0% loss=0.085169 lr=1.999883e-05 lr_backbone=1.999883e-05 elapsed=36:59:56 avg_step=3.70s recent_step=4.83s eta=14:25:01 +2026-09-19 06:28:57 | INFO | train rank=0 | step=35990/50000 progress=72.0% loss=0.089033 lr=1.997238e-05 lr_backbone=1.997238e-05 elapsed=37:00:21 avg_step=3.70s recent_step=2.48s eta=14:24:19 +2026-09-19 06:29:32 | INFO | train rank=0 | step=36000/50000 progress=72.0% loss=0.067396 lr=1.994594e-05 lr_backbone=1.994594e-05 elapsed=37:00:56 avg_step=3.70s recent_step=3.50s eta=14:23:41 +2026-09-19 06:30:08 | INFO | train rank=0 | step=36010/50000 progress=72.0% loss=0.085025 lr=1.991952e-05 lr_backbone=1.991952e-05 elapsed=37:01:32 avg_step=3.70s recent_step=3.60s eta=14:23:04 +2026-09-19 06:30:53 | INFO | train rank=0 | step=36020/50000 progress=72.0% loss=0.102674 lr=1.989311e-05 lr_backbone=1.989311e-05 elapsed=37:02:16 avg_step=3.70s recent_step=4.46s eta=14:22:30 +2026-09-19 06:31:25 | INFO | train rank=0 | step=36030/50000 progress=72.1% loss=0.101988 lr=1.986672e-05 lr_backbone=1.986672e-05 elapsed=37:02:48 avg_step=3.70s recent_step=3.19s eta=14:21:51 +2026-09-19 06:32:01 | INFO | train rank=0 | step=36040/50000 progress=72.1% loss=0.075042 lr=1.984033e-05 lr_backbone=1.984033e-05 elapsed=37:03:24 avg_step=3.70s recent_step=3.62s eta=14:21:13 +2026-09-19 06:32:39 | INFO | train rank=0 | step=36050/50000 progress=72.1% loss=0.105749 lr=1.981396e-05 lr_backbone=1.981396e-05 elapsed=37:04:03 avg_step=3.70s recent_step=3.84s eta=14:20:37 +2026-09-19 06:33:28 | INFO | train rank=0 | step=36060/50000 progress=72.1% loss=0.084272 lr=1.978761e-05 lr_backbone=1.978761e-05 elapsed=37:04:52 avg_step=3.70s recent_step=4.91s eta=14:20:05 +2026-09-19 06:34:13 | INFO | train rank=0 | step=36070/50000 progress=72.1% loss=0.097041 lr=1.976126e-05 lr_backbone=1.976126e-05 elapsed=37:05:36 avg_step=3.70s recent_step=4.46s eta=14:19:31 +2026-09-19 06:34:43 | INFO | train rank=0 | step=36080/50000 progress=72.2% loss=0.112491 lr=1.973493e-05 lr_backbone=1.973493e-05 elapsed=37:06:06 avg_step=3.70s recent_step=2.99s eta=14:18:51 +2026-09-19 06:35:15 | INFO | train rank=0 | step=36090/50000 progress=72.2% loss=0.086213 lr=1.970862e-05 lr_backbone=1.970862e-05 elapsed=37:06:39 avg_step=3.70s recent_step=3.24s eta=14:18:12 +2026-09-19 06:36:08 | INFO | train rank=0 | step=36100/50000 progress=72.2% loss=0.077571 lr=1.968232e-05 lr_backbone=1.968232e-05 elapsed=37:07:31 avg_step=3.70s recent_step=5.27s eta=14:17:41 +2026-09-19 06:36:39 | INFO | train rank=0 | step=36110/50000 progress=72.2% loss=0.082086 lr=1.965603e-05 lr_backbone=1.965603e-05 elapsed=37:08:02 avg_step=3.70s recent_step=3.11s eta=14:17:02 +2026-09-19 06:37:09 | INFO | train rank=0 | step=36120/50000 progress=72.2% loss=0.074037 lr=1.962975e-05 lr_backbone=1.962975e-05 elapsed=37:08:32 avg_step=3.70s recent_step=2.95s eta=14:16:22 +2026-09-19 06:37:38 | INFO | train rank=0 | step=36130/50000 progress=72.3% loss=0.109987 lr=1.960349e-05 lr_backbone=1.960349e-05 elapsed=37:09:01 avg_step=3.70s recent_step=2.90s eta=14:15:42 +2026-09-19 06:38:24 | INFO | train rank=0 | step=36140/50000 progress=72.3% loss=0.078635 lr=1.957723e-05 lr_backbone=1.957723e-05 elapsed=37:09:48 avg_step=3.70s recent_step=4.68s eta=14:15:08 +2026-09-19 06:39:04 | INFO | train rank=0 | step=36150/50000 progress=72.3% loss=0.092600 lr=1.955100e-05 lr_backbone=1.955100e-05 elapsed=37:10:28 avg_step=3.70s recent_step=4.00s eta=14:14:33 +2026-09-19 06:39:44 | INFO | train rank=0 | step=36160/50000 progress=72.3% loss=0.089270 lr=1.952477e-05 lr_backbone=1.952477e-05 elapsed=37:11:07 avg_step=3.70s recent_step=3.92s eta=14:13:56 +2026-09-19 06:40:15 | INFO | train rank=0 | step=36170/50000 progress=72.3% loss=0.082929 lr=1.949856e-05 lr_backbone=1.949856e-05 elapsed=37:11:39 avg_step=3.70s recent_step=3.17s eta=14:13:17 +2026-09-19 06:41:04 | INFO | train rank=0 | step=36180/50000 progress=72.4% loss=0.084577 lr=1.947237e-05 lr_backbone=1.947237e-05 elapsed=37:12:27 avg_step=3.70s recent_step=4.87s eta=14:12:45 +2026-09-19 06:41:32 | INFO | train rank=0 | step=36190/50000 progress=72.4% loss=0.083269 lr=1.944618e-05 lr_backbone=1.944618e-05 elapsed=37:12:56 avg_step=3.70s recent_step=2.82s eta=14:12:04 +2026-09-19 06:42:00 | INFO | train rank=0 | step=36200/50000 progress=72.4% loss=0.105305 lr=1.942001e-05 lr_backbone=1.942001e-05 elapsed=37:13:24 avg_step=3.70s recent_step=2.82s eta=14:11:24 +2026-09-19 06:42:33 | INFO | train rank=0 | step=36210/50000 progress=72.4% loss=0.080725 lr=1.939386e-05 lr_backbone=1.939386e-05 elapsed=37:13:57 avg_step=3.70s recent_step=3.27s eta=14:10:45 +2026-09-19 06:43:28 | INFO | train rank=0 | step=36220/50000 progress=72.4% loss=0.082011 lr=1.936771e-05 lr_backbone=1.936771e-05 elapsed=37:14:52 avg_step=3.70s recent_step=5.52s eta=14:10:15 +2026-09-19 06:44:02 | INFO | train rank=0 | step=36230/50000 progress=72.5% loss=0.100123 lr=1.934158e-05 lr_backbone=1.934158e-05 elapsed=37:15:25 avg_step=3.70s recent_step=3.37s eta=14:09:37 +2026-09-19 06:44:31 | INFO | train rank=0 | step=36240/50000 progress=72.5% loss=0.070771 lr=1.931547e-05 lr_backbone=1.931547e-05 elapsed=37:15:54 avg_step=3.70s recent_step=2.91s eta=14:08:57 +2026-09-19 06:45:05 | INFO | train rank=0 | step=36250/50000 progress=72.5% loss=0.137458 lr=1.928936e-05 lr_backbone=1.928936e-05 elapsed=37:16:29 avg_step=3.70s recent_step=3.42s eta=14:08:19 +2026-09-19 06:45:52 | INFO | train rank=0 | step=36260/50000 progress=72.5% loss=0.091036 lr=1.926327e-05 lr_backbone=1.926327e-05 elapsed=37:17:15 avg_step=3.70s recent_step=4.67s eta=14:07:45 +2026-09-19 06:46:25 | INFO | train rank=0 | step=36270/50000 progress=72.5% loss=0.075396 lr=1.923720e-05 lr_backbone=1.923720e-05 elapsed=37:17:48 avg_step=3.70s recent_step=3.24s eta=14:07:07 +2026-09-19 06:47:03 | INFO | train rank=0 | step=36280/50000 progress=72.6% loss=0.087644 lr=1.921114e-05 lr_backbone=1.921114e-05 elapsed=37:18:27 avg_step=3.70s recent_step=3.87s eta=14:06:30 +2026-09-19 06:47:36 | INFO | train rank=0 | step=36290/50000 progress=72.6% loss=0.081409 lr=1.918509e-05 lr_backbone=1.918509e-05 elapsed=37:18:59 avg_step=3.70s recent_step=3.24s eta=14:05:52 +2026-09-19 06:48:10 | INFO | train rank=0 | step=36300/50000 progress=72.6% loss=0.082771 lr=1.915905e-05 lr_backbone=1.915905e-05 elapsed=37:19:33 avg_step=3.70s recent_step=3.42s eta=14:05:14 +2026-09-19 06:48:56 | INFO | train rank=0 | step=36310/50000 progress=72.6% loss=0.115778 lr=1.913303e-05 lr_backbone=1.913303e-05 elapsed=37:20:19 avg_step=3.70s recent_step=4.63s eta=14:04:40 +2026-09-19 06:49:32 | INFO | train rank=0 | step=36320/50000 progress=72.6% loss=0.094433 lr=1.910702e-05 lr_backbone=1.910702e-05 elapsed=37:20:55 avg_step=3.70s recent_step=3.57s eta=14:04:02 +2026-09-19 06:49:59 | INFO | train rank=0 | step=36330/50000 progress=72.7% loss=0.094615 lr=1.908102e-05 lr_backbone=1.908102e-05 elapsed=37:21:22 avg_step=3.70s recent_step=2.68s eta=14:03:22 +2026-09-19 06:50:44 | INFO | train rank=0 | step=36340/50000 progress=72.7% loss=0.083013 lr=1.905504e-05 lr_backbone=1.905504e-05 elapsed=37:22:07 avg_step=3.70s recent_step=4.49s eta=14:02:48 +2026-09-19 06:51:35 | INFO | train rank=0 | step=36350/50000 progress=72.7% loss=0.082985 lr=1.902907e-05 lr_backbone=1.902907e-05 elapsed=37:22:59 avg_step=3.70s recent_step=5.17s eta=14:02:16 +2026-09-19 06:52:02 | INFO | train rank=0 | step=36360/50000 progress=72.7% loss=0.106228 lr=1.900312e-05 lr_backbone=1.900312e-05 elapsed=37:23:25 avg_step=3.70s recent_step=2.66s eta=14:01:35 +2026-09-19 06:52:43 | INFO | train rank=0 | step=36370/50000 progress=72.7% loss=0.096453 lr=1.897718e-05 lr_backbone=1.897718e-05 elapsed=37:24:07 avg_step=3.70s recent_step=4.15s eta=14:01:00 +2026-09-19 06:53:14 | INFO | train rank=0 | step=36380/50000 progress=72.8% loss=0.090337 lr=1.895125e-05 lr_backbone=1.895125e-05 elapsed=37:24:37 avg_step=3.70s recent_step=3.07s eta=14:00:20 +2026-09-19 06:54:09 | INFO | train rank=0 | step=36390/50000 progress=72.8% loss=0.098904 lr=1.892534e-05 lr_backbone=1.892534e-05 elapsed=37:25:32 avg_step=3.70s recent_step=5.49s eta=13:59:50 +2026-09-19 06:54:44 | INFO | train rank=0 | step=36400/50000 progress=72.8% loss=0.091249 lr=1.889944e-05 lr_backbone=1.889944e-05 elapsed=37:26:07 avg_step=3.70s recent_step=3.46s eta=13:59:12 +2026-09-19 06:55:10 | INFO | train rank=0 | step=36410/50000 progress=72.8% loss=0.089270 lr=1.887355e-05 lr_backbone=1.887355e-05 elapsed=37:26:33 avg_step=3.70s recent_step=2.62s eta=13:58:31 +2026-09-19 06:55:39 | INFO | train rank=0 | step=36420/50000 progress=72.8% loss=0.092545 lr=1.884768e-05 lr_backbone=1.884768e-05 elapsed=37:27:02 avg_step=3.70s recent_step=2.92s eta=13:57:51 +2026-09-19 06:56:29 | INFO | train rank=0 | step=36430/50000 progress=72.9% loss=0.082638 lr=1.882182e-05 lr_backbone=1.882182e-05 elapsed=37:27:52 avg_step=3.70s recent_step=4.97s eta=13:57:19 +2026-09-19 06:57:04 | INFO | train rank=0 | step=36440/50000 progress=72.9% loss=0.083375 lr=1.879597e-05 lr_backbone=1.879597e-05 elapsed=37:28:27 avg_step=3.70s recent_step=3.54s eta=13:56:41 +2026-09-19 06:57:35 | INFO | train rank=0 | step=36450/50000 progress=72.9% loss=0.089464 lr=1.877014e-05 lr_backbone=1.877014e-05 elapsed=37:28:58 avg_step=3.70s recent_step=3.09s eta=13:56:02 +2026-09-19 06:58:06 | INFO | train rank=0 | step=36460/50000 progress=72.9% loss=0.093047 lr=1.874432e-05 lr_backbone=1.874432e-05 elapsed=37:29:30 avg_step=3.70s recent_step=3.15s eta=13:55:23 +2026-09-19 06:58:51 | INFO | train rank=0 | step=36470/50000 progress=72.9% loss=0.074576 lr=1.871851e-05 lr_backbone=1.871851e-05 elapsed=37:30:14 avg_step=3.70s recent_step=4.46s eta=13:54:49 +2026-09-19 06:59:30 | INFO | train rank=0 | step=36480/50000 progress=73.0% loss=0.126911 lr=1.869272e-05 lr_backbone=1.869272e-05 elapsed=37:30:53 avg_step=3.70s recent_step=3.88s eta=13:54:12 +2026-09-19 07:00:00 | INFO | train rank=0 | step=36490/50000 progress=73.0% loss=0.078771 lr=1.866695e-05 lr_backbone=1.866695e-05 elapsed=37:31:23 avg_step=3.70s recent_step=2.99s eta=13:53:33 +2026-09-19 07:00:36 | INFO | train rank=0 | step=36500/50000 progress=73.0% loss=0.088961 lr=1.864118e-05 lr_backbone=1.864118e-05 elapsed=37:31:59 avg_step=3.70s recent_step=3.64s eta=13:52:55 +2026-09-19 07:01:09 | INFO | train rank=0 | step=36510/50000 progress=73.0% loss=0.127339 lr=1.861543e-05 lr_backbone=1.861543e-05 elapsed=37:32:32 avg_step=3.70s recent_step=3.27s eta=13:52:17 +2026-09-19 07:01:38 | INFO | train rank=0 | step=36520/50000 progress=73.0% loss=0.078797 lr=1.858970e-05 lr_backbone=1.858970e-05 elapsed=37:33:01 avg_step=3.70s recent_step=2.90s eta=13:51:37 +2026-09-19 07:02:23 | INFO | train rank=0 | step=36530/50000 progress=73.1% loss=0.090074 lr=1.856397e-05 lr_backbone=1.856397e-05 elapsed=37:33:46 avg_step=3.70s recent_step=4.47s eta=13:51:03 +2026-09-19 07:03:01 | INFO | train rank=0 | step=36540/50000 progress=73.1% loss=0.090018 lr=1.853826e-05 lr_backbone=1.853826e-05 elapsed=37:34:24 avg_step=3.70s recent_step=3.81s eta=13:50:26 +2026-09-19 07:03:46 | INFO | train rank=0 | step=36550/50000 progress=73.1% loss=0.079020 lr=1.851257e-05 lr_backbone=1.851257e-05 elapsed=37:35:09 avg_step=3.70s recent_step=4.54s eta=13:49:52 +2026-09-19 07:04:19 | INFO | train rank=0 | step=36560/50000 progress=73.1% loss=0.075886 lr=1.848689e-05 lr_backbone=1.848689e-05 elapsed=37:35:43 avg_step=3.70s recent_step=3.34s eta=13:49:14 +2026-09-19 07:04:57 | INFO | train rank=0 | step=36570/50000 progress=73.1% loss=0.083326 lr=1.846122e-05 lr_backbone=1.846122e-05 elapsed=37:36:21 avg_step=3.70s recent_step=3.80s eta=13:48:37 +2026-09-19 07:05:48 | INFO | train rank=0 | step=36580/50000 progress=73.2% loss=0.084561 lr=1.843557e-05 lr_backbone=1.843557e-05 elapsed=37:37:12 avg_step=3.70s recent_step=5.11s eta=13:48:05 +2026-09-19 07:06:15 | INFO | train rank=0 | step=36590/50000 progress=73.2% loss=0.097904 lr=1.840993e-05 lr_backbone=1.840993e-05 elapsed=37:37:38 avg_step=3.70s recent_step=2.61s eta=13:47:24 +2026-09-19 07:06:52 | INFO | train rank=0 | step=36600/50000 progress=73.2% loss=0.092451 lr=1.838430e-05 lr_backbone=1.838430e-05 elapsed=37:38:15 avg_step=3.70s recent_step=3.73s eta=13:46:47 +2026-09-19 07:07:27 | INFO | train rank=0 | step=36610/50000 progress=73.2% loss=0.094878 lr=1.835869e-05 lr_backbone=1.835869e-05 elapsed=37:38:50 avg_step=3.70s recent_step=3.48s eta=13:46:09 +2026-09-19 07:08:09 | INFO | train rank=0 | step=36620/50000 progress=73.2% loss=0.082382 lr=1.833309e-05 lr_backbone=1.833309e-05 elapsed=37:39:33 avg_step=3.70s recent_step=4.26s eta=13:45:34 +2026-09-19 07:08:52 | INFO | train rank=0 | step=36630/50000 progress=73.3% loss=0.096964 lr=1.830750e-05 lr_backbone=1.830750e-05 elapsed=37:40:15 avg_step=3.70s recent_step=4.23s eta=13:44:59 +2026-09-19 07:09:22 | INFO | train rank=0 | step=36640/50000 progress=73.3% loss=0.118056 lr=1.828193e-05 lr_backbone=1.828193e-05 elapsed=37:40:46 avg_step=3.70s recent_step=3.06s eta=13:44:20 +2026-09-19 07:09:59 | INFO | train rank=0 | step=36650/50000 progress=73.3% loss=0.081383 lr=1.825638e-05 lr_backbone=1.825638e-05 elapsed=37:41:23 avg_step=3.70s recent_step=3.71s eta=13:43:43 +2026-09-19 07:10:56 | INFO | train rank=0 | step=36660/50000 progress=73.3% loss=0.097699 lr=1.823083e-05 lr_backbone=1.823083e-05 elapsed=37:42:19 avg_step=3.70s recent_step=5.67s eta=13:43:13 +2026-09-19 07:11:29 | INFO | train rank=0 | step=36670/50000 progress=73.3% loss=0.082578 lr=1.820530e-05 lr_backbone=1.820530e-05 elapsed=37:42:52 avg_step=3.70s recent_step=3.31s eta=13:42:35 +2026-09-19 07:12:02 | INFO | train rank=0 | step=36680/50000 progress=73.4% loss=0.085317 lr=1.817979e-05 lr_backbone=1.817979e-05 elapsed=37:43:25 avg_step=3.70s recent_step=3.25s eta=13:41:56 +2026-09-19 07:12:24 | INFO | train rank=0 | step=36690/50000 progress=73.4% loss=0.111058 lr=1.815429e-05 lr_backbone=1.815429e-05 elapsed=37:43:47 avg_step=3.70s recent_step=2.24s eta=13:41:14 +2026-09-19 07:13:22 | INFO | train rank=0 | step=36700/50000 progress=73.4% loss=0.086410 lr=1.812880e-05 lr_backbone=1.812880e-05 elapsed=37:44:46 avg_step=3.70s recent_step=5.83s eta=13:40:44 +2026-09-19 07:13:51 | INFO | train rank=0 | step=36710/50000 progress=73.4% loss=0.086340 lr=1.810333e-05 lr_backbone=1.810333e-05 elapsed=37:45:15 avg_step=3.70s recent_step=2.92s eta=13:40:04 +2026-09-19 07:14:25 | INFO | train rank=0 | step=36720/50000 progress=73.4% loss=0.073068 lr=1.807787e-05 lr_backbone=1.807787e-05 elapsed=37:45:48 avg_step=3.70s recent_step=3.32s eta=13:39:26 +2026-09-19 07:14:53 | INFO | train rank=0 | step=36730/50000 progress=73.5% loss=0.075656 lr=1.805242e-05 lr_backbone=1.805242e-05 elapsed=37:46:16 avg_step=3.70s recent_step=2.79s eta=13:38:46 +2026-09-19 07:15:40 | INFO | train rank=0 | step=36740/50000 progress=73.5% loss=0.108530 lr=1.802699e-05 lr_backbone=1.802699e-05 elapsed=37:47:03 avg_step=3.70s recent_step=4.70s eta=13:38:12 +2026-09-19 07:16:19 | INFO | train rank=0 | step=36750/50000 progress=73.5% loss=0.070929 lr=1.800157e-05 lr_backbone=1.800157e-05 elapsed=37:47:42 avg_step=3.70s recent_step=3.91s eta=13:37:36 +2026-09-19 07:16:52 | INFO | train rank=0 | step=36760/50000 progress=73.5% loss=0.084267 lr=1.797617e-05 lr_backbone=1.797617e-05 elapsed=37:48:15 avg_step=3.70s recent_step=3.33s eta=13:36:58 +2026-09-19 07:17:24 | INFO | train rank=0 | step=36770/50000 progress=73.5% loss=0.089955 lr=1.795078e-05 lr_backbone=1.795078e-05 elapsed=37:48:47 avg_step=3.70s recent_step=3.18s eta=13:36:19 +2026-09-19 07:17:54 | INFO | train rank=0 | step=36780/50000 progress=73.6% loss=0.066665 lr=1.792540e-05 lr_backbone=1.792540e-05 elapsed=37:49:17 avg_step=3.70s recent_step=2.99s eta=13:35:39 +2026-09-19 07:18:48 | INFO | train rank=0 | step=36790/50000 progress=73.6% loss=0.092927 lr=1.790004e-05 lr_backbone=1.790004e-05 elapsed=37:50:12 avg_step=3.70s recent_step=5.45s eta=13:35:08 +2026-09-19 07:19:17 | INFO | train rank=0 | step=36800/50000 progress=73.6% loss=0.102710 lr=1.787470e-05 lr_backbone=1.787470e-05 elapsed=37:50:41 avg_step=3.70s recent_step=2.90s eta=13:34:29 +2026-09-19 07:19:46 | INFO | train rank=0 | step=36810/50000 progress=73.6% loss=0.107607 lr=1.784936e-05 lr_backbone=1.784936e-05 elapsed=37:51:10 avg_step=3.70s recent_step=2.92s eta=13:33:49 +2026-09-19 07:20:21 | INFO | train rank=0 | step=36820/50000 progress=73.6% loss=0.113642 lr=1.782404e-05 lr_backbone=1.782404e-05 elapsed=37:51:44 avg_step=3.70s recent_step=3.46s eta=13:33:11 +2026-09-19 07:21:14 | INFO | train rank=0 | step=36830/50000 progress=73.7% loss=0.081401 lr=1.779874e-05 lr_backbone=1.779874e-05 elapsed=37:52:37 avg_step=3.70s recent_step=5.28s eta=13:32:39 +2026-09-19 07:21:52 | INFO | train rank=0 | step=36840/50000 progress=73.7% loss=0.093230 lr=1.777345e-05 lr_backbone=1.777345e-05 elapsed=37:53:16 avg_step=3.70s recent_step=3.85s eta=13:32:03 +2026-09-19 07:22:19 | INFO | train rank=0 | step=36850/50000 progress=73.7% loss=0.150749 lr=1.774817e-05 lr_backbone=1.774817e-05 elapsed=37:53:42 avg_step=3.70s recent_step=2.64s eta=13:31:22 +2026-09-19 07:22:55 | INFO | train rank=0 | step=36860/50000 progress=73.7% loss=0.067372 lr=1.772291e-05 lr_backbone=1.772291e-05 elapsed=37:54:18 avg_step=3.70s recent_step=3.65s eta=13:30:45 +2026-09-19 07:23:36 | INFO | train rank=0 | step=36870/50000 progress=73.7% loss=0.072088 lr=1.769766e-05 lr_backbone=1.769766e-05 elapsed=37:55:00 avg_step=3.70s recent_step=4.12s eta=13:30:09 +2026-09-19 07:24:04 | INFO | train rank=0 | step=36880/50000 progress=73.8% loss=0.081191 lr=1.767242e-05 lr_backbone=1.767242e-05 elapsed=37:55:28 avg_step=3.70s recent_step=2.80s eta=13:29:29 +2026-09-19 07:24:40 | INFO | train rank=0 | step=36890/50000 progress=73.8% loss=0.117362 lr=1.764720e-05 lr_backbone=1.764720e-05 elapsed=37:56:03 avg_step=3.70s recent_step=3.57s eta=13:28:52 +2026-09-19 07:25:17 | INFO | train rank=0 | step=36900/50000 progress=73.8% loss=0.122139 lr=1.762200e-05 lr_backbone=1.762200e-05 elapsed=37:56:40 avg_step=3.70s recent_step=3.70s eta=13:28:15 +2026-09-19 07:25:59 | INFO | train rank=0 | step=36910/50000 progress=73.8% loss=0.091116 lr=1.759680e-05 lr_backbone=1.759680e-05 elapsed=37:57:22 avg_step=3.70s recent_step=4.18s eta=13:27:39 +2026-09-19 07:26:38 | INFO | train rank=0 | step=36920/50000 progress=73.8% loss=0.090194 lr=1.757163e-05 lr_backbone=1.757163e-05 elapsed=37:58:01 avg_step=3.70s recent_step=3.91s eta=13:27:03 +2026-09-19 07:27:12 | INFO | train rank=0 | step=36930/50000 progress=73.9% loss=0.078186 lr=1.754646e-05 lr_backbone=1.754646e-05 elapsed=37:58:36 avg_step=3.70s recent_step=3.43s eta=13:26:25 +2026-09-19 07:27:53 | INFO | train rank=0 | step=36940/50000 progress=73.9% loss=0.133372 lr=1.752131e-05 lr_backbone=1.752131e-05 elapsed=37:59:16 avg_step=3.70s recent_step=4.07s eta=13:25:49 +2026-09-19 07:28:42 | INFO | train rank=0 | step=36950/50000 progress=73.9% loss=0.080093 lr=1.749618e-05 lr_backbone=1.749618e-05 elapsed=38:00:06 avg_step=3.70s recent_step=4.95s eta=13:25:17 +2026-09-19 07:29:17 | INFO | train rank=0 | step=36960/50000 progress=73.9% loss=0.091175 lr=1.747105e-05 lr_backbone=1.747105e-05 elapsed=38:00:41 avg_step=3.70s recent_step=3.50s eta=13:24:39 +2026-09-19 07:29:47 | INFO | train rank=0 | step=36970/50000 progress=73.9% loss=0.076784 lr=1.744595e-05 lr_backbone=1.744595e-05 elapsed=38:01:10 avg_step=3.70s recent_step=2.97s eta=13:23:59 +2026-09-19 07:30:21 | INFO | train rank=0 | step=36980/50000 progress=74.0% loss=0.094079 lr=1.742085e-05 lr_backbone=1.742085e-05 elapsed=38:01:44 avg_step=3.70s recent_step=3.40s eta=13:23:21 +2026-09-19 07:31:09 | INFO | train rank=0 | step=36990/50000 progress=74.0% loss=0.102839 lr=1.739578e-05 lr_backbone=1.739578e-05 elapsed=38:02:32 avg_step=3.70s recent_step=4.80s eta=13:22:48 +2026-09-19 07:31:44 | INFO | train rank=0 | step=37000/50000 progress=74.0% loss=0.086063 lr=1.737071e-05 lr_backbone=1.737071e-05 elapsed=38:03:07 avg_step=3.70s recent_step=3.46s eta=13:22:10 +2026-09-19 07:32:15 | INFO | train rank=0 | step=37010/50000 progress=74.0% loss=0.080504 lr=1.734566e-05 lr_backbone=1.734566e-05 elapsed=38:03:38 avg_step=3.70s recent_step=3.14s eta=13:21:31 +2026-09-19 07:32:57 | INFO | train rank=0 | step=37020/50000 progress=74.0% loss=0.076162 lr=1.732063e-05 lr_backbone=1.732063e-05 elapsed=38:04:20 avg_step=3.70s recent_step=4.16s eta=13:20:56 +2026-09-19 07:33:29 | INFO | train rank=0 | step=37030/50000 progress=74.1% loss=0.077435 lr=1.729560e-05 lr_backbone=1.729560e-05 elapsed=38:04:52 avg_step=3.70s recent_step=3.25s eta=13:20:17 +2026-09-19 07:34:01 | INFO | train rank=0 | step=37040/50000 progress=74.1% loss=0.071666 lr=1.727060e-05 lr_backbone=1.727060e-05 elapsed=38:05:24 avg_step=3.70s recent_step=3.14s eta=13:19:38 +2026-09-19 07:34:34 | INFO | train rank=0 | step=37050/50000 progress=74.1% loss=0.089207 lr=1.724560e-05 lr_backbone=1.724560e-05 elapsed=38:05:58 avg_step=3.70s recent_step=3.39s eta=13:19:00 +2026-09-19 07:35:27 | INFO | train rank=0 | step=37060/50000 progress=74.1% loss=0.083275 lr=1.722063e-05 lr_backbone=1.722063e-05 elapsed=38:06:50 avg_step=3.70s recent_step=5.25s eta=13:18:28 +2026-09-19 07:36:08 | INFO | train rank=0 | step=37070/50000 progress=74.1% loss=0.108781 lr=1.719566e-05 lr_backbone=1.719566e-05 elapsed=38:07:31 avg_step=3.70s recent_step=4.05s eta=13:17:53 +2026-09-19 07:36:39 | INFO | train rank=0 | step=37080/50000 progress=74.2% loss=0.118440 lr=1.717071e-05 lr_backbone=1.717071e-05 elapsed=38:08:02 avg_step=3.70s recent_step=3.16s eta=13:17:14 +2026-09-19 07:37:17 | INFO | train rank=0 | step=37090/50000 progress=74.2% loss=0.072551 lr=1.714578e-05 lr_backbone=1.714578e-05 elapsed=38:08:40 avg_step=3.70s recent_step=3.76s eta=13:16:37 +2026-09-19 07:38:02 | INFO | train rank=0 | step=37100/50000 progress=74.2% loss=0.075681 lr=1.712086e-05 lr_backbone=1.712086e-05 elapsed=38:09:25 avg_step=3.70s recent_step=4.51s eta=13:16:03 +2026-09-19 07:38:36 | INFO | train rank=0 | step=37110/50000 progress=74.2% loss=0.102649 lr=1.709595e-05 lr_backbone=1.709595e-05 elapsed=38:09:59 avg_step=3.70s recent_step=3.40s eta=13:15:25 +2026-09-19 07:39:08 | INFO | train rank=0 | step=37120/50000 progress=74.2% loss=0.078017 lr=1.707106e-05 lr_backbone=1.707106e-05 elapsed=38:10:32 avg_step=3.70s recent_step=3.26s eta=13:14:46 +2026-09-19 07:39:44 | INFO | train rank=0 | step=37130/50000 progress=74.3% loss=0.092079 lr=1.704618e-05 lr_backbone=1.704618e-05 elapsed=38:11:07 avg_step=3.70s recent_step=3.56s eta=13:14:09 +2026-09-19 07:40:37 | INFO | train rank=0 | step=37140/50000 progress=74.3% loss=0.078659 lr=1.702132e-05 lr_backbone=1.702132e-05 elapsed=38:12:00 avg_step=3.70s recent_step=5.31s eta=13:13:37 +2026-09-19 07:41:12 | INFO | train rank=0 | step=37150/50000 progress=74.3% loss=0.098799 lr=1.699647e-05 lr_backbone=1.699647e-05 elapsed=38:12:36 avg_step=3.70s recent_step=3.53s eta=13:13:00 +2026-09-19 07:41:45 | INFO | train rank=0 | step=37160/50000 progress=74.3% loss=0.095015 lr=1.697163e-05 lr_backbone=1.697163e-05 elapsed=38:13:08 avg_step=3.70s recent_step=3.26s eta=13:12:21 +2026-09-19 07:42:21 | INFO | train rank=0 | step=37170/50000 progress=74.3% loss=0.109536 lr=1.694681e-05 lr_backbone=1.694681e-05 elapsed=38:13:44 avg_step=3.70s recent_step=3.60s eta=13:11:44 +2026-09-19 07:43:02 | INFO | train rank=0 | step=37180/50000 progress=74.4% loss=0.087381 lr=1.692201e-05 lr_backbone=1.692201e-05 elapsed=38:14:25 avg_step=3.70s recent_step=4.06s eta=13:11:08 +2026-09-19 07:43:30 | INFO | train rank=0 | step=37190/50000 progress=74.4% loss=0.148556 lr=1.689721e-05 lr_backbone=1.689721e-05 elapsed=38:14:53 avg_step=3.70s recent_step=2.84s eta=13:10:28 +2026-09-19 07:44:04 | INFO | train rank=0 | step=37200/50000 progress=74.4% loss=0.078150 lr=1.687244e-05 lr_backbone=1.687244e-05 elapsed=38:15:28 avg_step=3.70s recent_step=3.42s eta=13:09:50 +2026-09-19 07:44:44 | INFO | train rank=0 | step=37210/50000 progress=74.4% loss=0.077381 lr=1.684768e-05 lr_backbone=1.684768e-05 elapsed=38:16:07 avg_step=3.70s recent_step=3.95s eta=13:09:14 +2026-09-19 07:45:24 | INFO | train rank=0 | step=37220/50000 progress=74.4% loss=0.072614 lr=1.682293e-05 lr_backbone=1.682293e-05 elapsed=38:16:47 avg_step=3.70s recent_step=4.02s eta=13:08:38 +2026-09-19 07:46:00 | INFO | train rank=0 | step=37230/50000 progress=74.5% loss=0.083632 lr=1.679819e-05 lr_backbone=1.679819e-05 elapsed=38:17:24 avg_step=3.70s recent_step=3.63s eta=13:08:00 +2026-09-19 07:46:45 | INFO | train rank=0 | step=37240/50000 progress=74.5% loss=0.072795 lr=1.677348e-05 lr_backbone=1.677348e-05 elapsed=38:18:08 avg_step=3.70s recent_step=4.45s eta=13:07:26 +2026-09-19 07:47:22 | INFO | train rank=0 | step=37250/50000 progress=74.5% loss=0.087235 lr=1.674877e-05 lr_backbone=1.674877e-05 elapsed=38:18:46 avg_step=3.70s recent_step=3.77s eta=13:06:49 +2026-09-19 07:47:57 | INFO | train rank=0 | step=37260/50000 progress=74.5% loss=0.108623 lr=1.672408e-05 lr_backbone=1.672408e-05 elapsed=38:19:21 avg_step=3.70s recent_step=3.49s eta=13:06:11 +2026-09-19 07:48:33 | INFO | train rank=0 | step=37270/50000 progress=74.5% loss=0.068247 lr=1.669941e-05 lr_backbone=1.669941e-05 elapsed=38:19:56 avg_step=3.70s recent_step=3.53s eta=13:05:34 +2026-09-19 07:49:10 | INFO | train rank=0 | step=37280/50000 progress=74.6% loss=0.101465 lr=1.667475e-05 lr_backbone=1.667475e-05 elapsed=38:20:34 avg_step=3.70s recent_step=3.79s eta=13:04:57 +2026-09-19 07:49:52 | INFO | train rank=0 | step=37290/50000 progress=74.6% loss=0.074939 lr=1.665010e-05 lr_backbone=1.665010e-05 elapsed=38:21:15 avg_step=3.70s recent_step=4.13s eta=13:04:21 +2026-09-19 07:50:22 | INFO | train rank=0 | step=37300/50000 progress=74.6% loss=0.080784 lr=1.662547e-05 lr_backbone=1.662547e-05 elapsed=38:21:45 avg_step=3.70s recent_step=3.01s eta=13:03:42 +2026-09-19 07:50:55 | INFO | train rank=0 | step=37310/50000 progress=74.6% loss=0.085777 lr=1.660085e-05 lr_backbone=1.660085e-05 elapsed=38:22:18 avg_step=3.70s recent_step=3.27s eta=13:03:04 +2026-09-19 07:51:35 | INFO | train rank=0 | step=37320/50000 progress=74.6% loss=0.096357 lr=1.657625e-05 lr_backbone=1.657625e-05 elapsed=38:22:58 avg_step=3.70s recent_step=4.02s eta=13:02:28 +2026-09-19 07:52:17 | INFO | train rank=0 | step=37330/50000 progress=74.7% loss=0.084736 lr=1.655166e-05 lr_backbone=1.655166e-05 elapsed=38:23:40 avg_step=3.70s recent_step=4.22s eta=13:01:52 +2026-09-19 07:52:57 | INFO | train rank=0 | step=37340/50000 progress=74.7% loss=0.071095 lr=1.652709e-05 lr_backbone=1.652709e-05 elapsed=38:24:20 avg_step=3.70s recent_step=3.97s eta=13:01:16 +2026-09-19 07:53:30 | INFO | train rank=0 | step=37350/50000 progress=74.7% loss=0.093314 lr=1.650253e-05 lr_backbone=1.650253e-05 elapsed=38:24:53 avg_step=3.70s recent_step=3.35s eta=13:00:38 +2026-09-19 07:54:04 | INFO | train rank=0 | step=37360/50000 progress=74.7% loss=0.084430 lr=1.647799e-05 lr_backbone=1.647799e-05 elapsed=38:25:27 avg_step=3.70s recent_step=3.40s eta=13:00:00 +2026-09-19 07:54:50 | INFO | train rank=0 | step=37370/50000 progress=74.7% loss=0.085565 lr=1.645346e-05 lr_backbone=1.645346e-05 elapsed=38:26:13 avg_step=3.70s recent_step=4.54s eta=12:59:26 +2026-09-19 07:55:36 | INFO | train rank=0 | step=37380/50000 progress=74.8% loss=0.086705 lr=1.642894e-05 lr_backbone=1.642894e-05 elapsed=38:26:59 avg_step=3.70s recent_step=4.65s eta=12:58:52 +2026-09-19 07:56:15 | INFO | train rank=0 | step=37390/50000 progress=74.8% loss=0.128005 lr=1.640444e-05 lr_backbone=1.640444e-05 elapsed=38:27:38 avg_step=3.70s recent_step=3.90s eta=12:58:16 +2026-09-19 07:56:53 | INFO | train rank=0 | step=37400/50000 progress=74.8% loss=0.080888 lr=1.637996e-05 lr_backbone=1.637996e-05 elapsed=38:28:17 avg_step=3.70s recent_step=3.84s eta=12:57:39 +2026-09-19 07:57:22 | INFO | train rank=0 | step=37410/50000 progress=74.8% loss=0.096037 lr=1.635549e-05 lr_backbone=1.635549e-05 elapsed=38:28:45 avg_step=3.70s recent_step=2.87s eta=12:56:59 +2026-09-19 07:58:18 | INFO | train rank=0 | step=37420/50000 progress=74.8% loss=0.075906 lr=1.633103e-05 lr_backbone=1.633103e-05 elapsed=38:29:41 avg_step=3.70s recent_step=5.59s eta=12:56:28 +2026-09-19 07:58:54 | INFO | train rank=0 | step=37430/50000 progress=74.9% loss=0.115537 lr=1.630659e-05 lr_backbone=1.630659e-05 elapsed=38:30:17 avg_step=3.70s recent_step=3.57s eta=12:55:51 +2026-09-19 07:59:15 | INFO | train rank=0 | step=37440/50000 progress=74.9% loss=0.067562 lr=1.628217e-05 lr_backbone=1.628217e-05 elapsed=38:30:39 avg_step=3.70s recent_step=2.16s eta=12:55:09 +2026-09-19 07:59:49 | INFO | train rank=0 | step=37450/50000 progress=74.9% loss=0.086370 lr=1.625776e-05 lr_backbone=1.625776e-05 elapsed=38:31:13 avg_step=3.70s recent_step=3.40s eta=12:54:31 +2026-09-19 08:00:37 | INFO | train rank=0 | step=37460/50000 progress=74.9% loss=0.080811 lr=1.623336e-05 lr_backbone=1.623336e-05 elapsed=38:32:00 avg_step=3.70s recent_step=4.78s eta=12:53:57 +2026-09-19 08:01:18 | INFO | train rank=0 | step=37470/50000 progress=74.9% loss=0.081787 lr=1.620898e-05 lr_backbone=1.620898e-05 elapsed=38:32:41 avg_step=3.70s recent_step=4.06s eta=12:53:21 +2026-09-19 08:01:55 | INFO | train rank=0 | step=37480/50000 progress=75.0% loss=0.066464 lr=1.618461e-05 lr_backbone=1.618461e-05 elapsed=38:33:18 avg_step=3.70s recent_step=3.74s eta=12:52:45 +2026-09-19 08:02:29 | INFO | train rank=0 | step=37490/50000 progress=75.0% loss=0.078601 lr=1.616026e-05 lr_backbone=1.616026e-05 elapsed=38:33:53 avg_step=3.70s recent_step=3.42s eta=12:52:07 +2026-09-19 08:02:58 | INFO | train rank=0 | step=37500/50000 progress=75.0% loss=0.087012 lr=1.613592e-05 lr_backbone=1.613592e-05 elapsed=38:34:21 avg_step=3.70s recent_step=2.87s eta=12:51:27 +2026-09-19 08:03:49 | INFO | train rank=0 | step=37510/50000 progress=75.0% loss=0.088483 lr=1.611160e-05 lr_backbone=1.611160e-05 elapsed=38:35:12 avg_step=3.70s recent_step=5.09s eta=12:50:54 +2026-09-19 08:04:16 | INFO | train rank=0 | step=37520/50000 progress=75.0% loss=0.087796 lr=1.608729e-05 lr_backbone=1.608729e-05 elapsed=38:35:39 avg_step=3.70s recent_step=2.67s eta=12:50:14 +2026-09-19 08:04:46 | INFO | train rank=0 | step=37530/50000 progress=75.1% loss=0.091092 lr=1.606300e-05 lr_backbone=1.606300e-05 elapsed=38:36:09 avg_step=3.70s recent_step=3.01s eta=12:49:35 +2026-09-19 08:05:14 | INFO | train rank=0 | step=37540/50000 progress=75.1% loss=0.073396 lr=1.603872e-05 lr_backbone=1.603872e-05 elapsed=38:36:37 avg_step=3.70s recent_step=2.82s eta=12:48:55 +2026-09-19 08:05:50 | INFO | train rank=0 | step=37550/50000 progress=75.1% loss=0.088006 lr=1.601446e-05 lr_backbone=1.601446e-05 elapsed=38:37:13 avg_step=3.70s recent_step=3.55s eta=12:48:17 +2026-09-19 08:06:20 | INFO | train rank=0 | step=37560/50000 progress=75.1% loss=0.083838 lr=1.599021e-05 lr_backbone=1.599021e-05 elapsed=38:37:43 avg_step=3.70s recent_step=3.01s eta=12:47:38 +2026-09-19 08:06:55 | INFO | train rank=0 | step=37570/50000 progress=75.1% loss=0.093063 lr=1.596598e-05 lr_backbone=1.596598e-05 elapsed=38:38:18 avg_step=3.70s recent_step=3.49s eta=12:47:00 +2026-09-19 08:07:37 | INFO | train rank=0 | step=37580/50000 progress=75.2% loss=0.087466 lr=1.594176e-05 lr_backbone=1.594176e-05 elapsed=38:39:00 avg_step=3.70s recent_step=4.24s eta=12:46:25 +2026-09-19 08:08:22 | INFO | train rank=0 | step=37590/50000 progress=75.2% loss=0.085340 lr=1.591755e-05 lr_backbone=1.591755e-05 elapsed=38:39:46 avg_step=3.70s recent_step=4.55s eta=12:45:51 +2026-09-19 08:08:55 | INFO | train rank=0 | step=37600/50000 progress=75.2% loss=0.112155 lr=1.589336e-05 lr_backbone=1.589336e-05 elapsed=38:40:18 avg_step=3.70s recent_step=3.27s eta=12:45:12 +2026-09-19 08:09:33 | INFO | train rank=0 | step=37610/50000 progress=75.2% loss=0.096606 lr=1.586919e-05 lr_backbone=1.586919e-05 elapsed=38:40:57 avg_step=3.70s recent_step=3.82s eta=12:44:35 +2026-09-19 08:10:07 | INFO | train rank=0 | step=37620/50000 progress=75.2% loss=0.079480 lr=1.584503e-05 lr_backbone=1.584503e-05 elapsed=38:41:31 avg_step=3.70s recent_step=3.40s eta=12:43:57 +2026-09-19 08:10:43 | INFO | train rank=0 | step=37630/50000 progress=75.3% loss=0.076294 lr=1.582089e-05 lr_backbone=1.582089e-05 elapsed=38:42:07 avg_step=3.70s recent_step=3.60s eta=12:43:20 +2026-09-19 08:11:38 | INFO | train rank=0 | step=37640/50000 progress=75.3% loss=0.070411 lr=1.579676e-05 lr_backbone=1.579676e-05 elapsed=38:43:01 avg_step=3.70s recent_step=5.43s eta=12:42:49 +2026-09-19 08:12:04 | INFO | train rank=0 | step=37650/50000 progress=75.3% loss=0.092204 lr=1.577264e-05 lr_backbone=1.577264e-05 elapsed=38:43:28 avg_step=3.70s recent_step=2.68s eta=12:42:08 +2026-09-19 08:12:44 | INFO | train rank=0 | step=37660/50000 progress=75.3% loss=0.077736 lr=1.574855e-05 lr_backbone=1.574855e-05 elapsed=38:44:08 avg_step=3.70s recent_step=4.02s eta=12:41:32 +2026-09-19 08:13:13 | INFO | train rank=0 | step=37670/50000 progress=75.3% loss=0.079451 lr=1.572446e-05 lr_backbone=1.572446e-05 elapsed=38:44:36 avg_step=3.70s recent_step=2.83s eta=12:40:52 +2026-09-19 08:14:06 | INFO | train rank=0 | step=37680/50000 progress=75.4% loss=0.100931 lr=1.570039e-05 lr_backbone=1.570039e-05 elapsed=38:45:29 avg_step=3.70s recent_step=5.31s eta=12:40:21 +2026-09-19 08:14:37 | INFO | train rank=0 | step=37690/50000 progress=75.4% loss=0.084695 lr=1.567634e-05 lr_backbone=1.567634e-05 elapsed=38:46:00 avg_step=3.70s recent_step=3.13s eta=12:39:42 +2026-09-19 08:15:21 | INFO | train rank=0 | step=37700/50000 progress=75.4% loss=0.086566 lr=1.565230e-05 lr_backbone=1.565230e-05 elapsed=38:46:44 avg_step=3.70s recent_step=4.34s eta=12:39:07 +2026-09-19 08:15:51 | INFO | train rank=0 | step=37710/50000 progress=75.4% loss=0.114291 lr=1.562828e-05 lr_backbone=1.562828e-05 elapsed=38:47:14 avg_step=3.70s recent_step=3.05s eta=12:38:28 +2026-09-19 08:16:32 | INFO | train rank=0 | step=37720/50000 progress=75.4% loss=0.085164 lr=1.560427e-05 lr_backbone=1.560427e-05 elapsed=38:47:55 avg_step=3.70s recent_step=4.10s eta=12:37:52 +2026-09-19 08:17:07 | INFO | train rank=0 | step=37730/50000 progress=75.5% loss=0.095870 lr=1.558027e-05 lr_backbone=1.558027e-05 elapsed=38:48:31 avg_step=3.70s recent_step=3.54s eta=12:37:14 +2026-09-19 08:17:43 | INFO | train rank=0 | step=37740/50000 progress=75.5% loss=0.070278 lr=1.555629e-05 lr_backbone=1.555629e-05 elapsed=38:49:07 avg_step=3.70s recent_step=3.59s eta=12:36:37 +2026-09-19 08:18:17 | INFO | train rank=0 | step=37750/50000 progress=75.5% loss=0.103669 lr=1.553233e-05 lr_backbone=1.553233e-05 elapsed=38:49:41 avg_step=3.70s recent_step=3.40s eta=12:35:59 +2026-09-19 08:19:00 | INFO | train rank=0 | step=37760/50000 progress=75.5% loss=0.087993 lr=1.550838e-05 lr_backbone=1.550838e-05 elapsed=38:50:23 avg_step=3.70s recent_step=4.23s eta=12:35:24 +2026-09-19 08:19:42 | INFO | train rank=0 | step=37770/50000 progress=75.5% loss=0.123374 lr=1.548445e-05 lr_backbone=1.548445e-05 elapsed=38:51:06 avg_step=3.70s recent_step=4.27s eta=12:34:48 +2026-09-19 08:20:21 | INFO | train rank=0 | step=37780/50000 progress=75.6% loss=0.079165 lr=1.546053e-05 lr_backbone=1.546053e-05 elapsed=38:51:44 avg_step=3.70s recent_step=3.85s eta=12:34:12 +2026-09-19 08:20:54 | INFO | train rank=0 | step=37790/50000 progress=75.6% loss=0.097739 lr=1.543663e-05 lr_backbone=1.543663e-05 elapsed=38:52:17 avg_step=3.70s recent_step=3.31s eta=12:33:34 +2026-09-19 08:21:31 | INFO | train rank=0 | step=37800/50000 progress=75.6% loss=0.078480 lr=1.541274e-05 lr_backbone=1.541274e-05 elapsed=38:52:55 avg_step=3.70s recent_step=3.73s eta=12:32:57 +2026-09-19 08:22:14 | INFO | train rank=0 | step=37810/50000 progress=75.6% loss=0.079748 lr=1.538886e-05 lr_backbone=1.538886e-05 elapsed=38:53:37 avg_step=3.70s recent_step=4.24s eta=12:32:21 +2026-09-19 08:22:50 | INFO | train rank=0 | step=37820/50000 progress=75.6% loss=0.088073 lr=1.536501e-05 lr_backbone=1.536501e-05 elapsed=38:54:14 avg_step=3.70s recent_step=3.66s eta=12:31:44 +2026-09-19 08:23:21 | INFO | train rank=0 | step=37830/50000 progress=75.7% loss=0.081416 lr=1.534116e-05 lr_backbone=1.534116e-05 elapsed=38:54:44 avg_step=3.70s recent_step=3.06s eta=12:31:05 +2026-09-19 08:23:55 | INFO | train rank=0 | step=37840/50000 progress=75.7% loss=0.067483 lr=1.531733e-05 lr_backbone=1.531733e-05 elapsed=38:55:18 avg_step=3.70s recent_step=3.38s eta=12:30:27 +2026-09-19 08:24:46 | INFO | train rank=0 | step=37850/50000 progress=75.7% loss=0.101177 lr=1.529352e-05 lr_backbone=1.529352e-05 elapsed=38:56:09 avg_step=3.70s recent_step=5.14s eta=12:29:55 +2026-09-19 08:25:12 | INFO | train rank=0 | step=37860/50000 progress=75.7% loss=0.066074 lr=1.526972e-05 lr_backbone=1.526972e-05 elapsed=38:56:36 avg_step=3.70s recent_step=2.63s eta=12:29:14 +2026-09-19 08:25:53 | INFO | train rank=0 | step=37870/50000 progress=75.7% loss=0.091027 lr=1.524594e-05 lr_backbone=1.524594e-05 elapsed=38:57:17 avg_step=3.70s recent_step=4.10s eta=12:28:38 +2026-09-19 08:26:32 | INFO | train rank=0 | step=37880/50000 progress=75.8% loss=0.077586 lr=1.522218e-05 lr_backbone=1.522218e-05 elapsed=38:57:56 avg_step=3.70s recent_step=3.91s eta=12:28:02 +2026-09-19 08:27:09 | INFO | train rank=0 | step=37890/50000 progress=75.8% loss=0.071185 lr=1.519842e-05 lr_backbone=1.519842e-05 elapsed=38:58:32 avg_step=3.70s recent_step=3.64s eta=12:27:25 +2026-09-19 08:27:49 | INFO | train rank=0 | step=37900/50000 progress=75.8% loss=0.082459 lr=1.517469e-05 lr_backbone=1.517469e-05 elapsed=38:59:12 avg_step=3.70s recent_step=4.01s eta=12:26:49 +2026-09-19 08:28:25 | INFO | train rank=0 | step=37910/50000 progress=75.8% loss=0.081973 lr=1.515097e-05 lr_backbone=1.515097e-05 elapsed=38:59:48 avg_step=3.70s recent_step=3.58s eta=12:26:11 +2026-09-19 08:28:57 | INFO | train rank=0 | step=37920/50000 progress=75.8% loss=0.098706 lr=1.512726e-05 lr_backbone=1.512726e-05 elapsed=39:00:21 avg_step=3.70s recent_step=3.26s eta=12:25:33 +2026-09-19 08:29:35 | INFO | train rank=0 | step=37930/50000 progress=75.9% loss=0.079257 lr=1.510357e-05 lr_backbone=1.510357e-05 elapsed=39:00:58 avg_step=3.70s recent_step=3.78s eta=12:24:56 +2026-09-19 08:30:22 | INFO | train rank=0 | step=37940/50000 progress=75.9% loss=0.066911 lr=1.507989e-05 lr_backbone=1.507989e-05 elapsed=39:01:45 avg_step=3.70s recent_step=4.70s eta=12:24:22 +2026-09-19 08:30:55 | INFO | train rank=0 | step=37950/50000 progress=75.9% loss=0.070100 lr=1.505623e-05 lr_backbone=1.505623e-05 elapsed=39:02:18 avg_step=3.70s recent_step=3.26s eta=12:23:44 +2026-09-19 08:31:33 | INFO | train rank=0 | step=37960/50000 progress=75.9% loss=0.074197 lr=1.503259e-05 lr_backbone=1.503259e-05 elapsed=39:02:56 avg_step=3.70s recent_step=3.84s eta=12:23:07 +2026-09-19 08:32:08 | INFO | train rank=0 | step=37970/50000 progress=75.9% loss=0.079591 lr=1.500896e-05 lr_backbone=1.500896e-05 elapsed=39:03:31 avg_step=3.70s recent_step=3.46s eta=12:22:29 +2026-09-19 08:32:32 | INFO | train rank=0 | step=37980/50000 progress=76.0% loss=0.080308 lr=1.498534e-05 lr_backbone=1.498534e-05 elapsed=39:03:56 avg_step=3.70s recent_step=2.46s eta=12:21:48 +2026-09-19 08:33:14 | INFO | train rank=0 | step=37990/50000 progress=76.0% loss=0.090018 lr=1.496174e-05 lr_backbone=1.496174e-05 elapsed=39:04:37 avg_step=3.70s recent_step=4.16s eta=12:21:13 +2026-09-19 08:33:48 | INFO | train rank=0 | step=38000/50000 progress=76.0% loss=0.086440 lr=1.493816e-05 lr_backbone=1.493816e-05 elapsed=39:05:11 avg_step=3.70s recent_step=3.36s eta=12:20:35 +2026-09-19 08:34:40 | INFO | train rank=0 | step=38010/50000 progress=76.0% loss=0.093455 lr=1.491459e-05 lr_backbone=1.491459e-05 elapsed=39:06:03 avg_step=3.70s recent_step=5.24s eta=12:20:02 +2026-09-19 08:35:16 | INFO | train rank=0 | step=38020/50000 progress=76.0% loss=0.079355 lr=1.489104e-05 lr_backbone=1.489104e-05 elapsed=39:06:40 avg_step=3.70s recent_step=3.63s eta=12:19:25 +2026-09-19 08:35:48 | INFO | train rank=0 | step=38030/50000 progress=76.1% loss=0.085356 lr=1.486750e-05 lr_backbone=1.486750e-05 elapsed=39:07:11 avg_step=3.70s recent_step=3.13s eta=12:18:46 +2026-09-19 08:36:15 | INFO | train rank=0 | step=38040/50000 progress=76.1% loss=0.105089 lr=1.484398e-05 lr_backbone=1.484398e-05 elapsed=39:07:38 avg_step=3.70s recent_step=2.75s eta=12:18:06 +2026-09-19 08:37:09 | INFO | train rank=0 | step=38050/50000 progress=76.1% loss=0.111529 lr=1.482047e-05 lr_backbone=1.482047e-05 elapsed=39:08:32 avg_step=3.70s recent_step=5.35s eta=12:17:35 +2026-09-19 08:37:39 | INFO | train rank=0 | step=38060/50000 progress=76.1% loss=0.086769 lr=1.479698e-05 lr_backbone=1.479698e-05 elapsed=39:09:03 avg_step=3.70s recent_step=3.07s eta=12:16:55 +2026-09-19 08:38:14 | INFO | train rank=0 | step=38070/50000 progress=76.1% loss=0.073206 lr=1.477350e-05 lr_backbone=1.477350e-05 elapsed=39:09:37 avg_step=3.70s recent_step=3.43s eta=12:16:18 +2026-09-19 08:38:51 | INFO | train rank=0 | step=38080/50000 progress=76.2% loss=0.117456 lr=1.475004e-05 lr_backbone=1.475004e-05 elapsed=39:10:14 avg_step=3.70s recent_step=3.72s eta=12:15:41 +2026-09-19 08:39:44 | INFO | train rank=0 | step=38090/50000 progress=76.2% loss=0.105654 lr=1.472660e-05 lr_backbone=1.472660e-05 elapsed=39:11:07 avg_step=3.70s recent_step=5.33s eta=12:15:09 +2026-09-19 08:40:12 | INFO | train rank=0 | step=38100/50000 progress=76.2% loss=0.077887 lr=1.470317e-05 lr_backbone=1.470317e-05 elapsed=39:11:35 avg_step=3.70s recent_step=2.80s eta=12:14:29 +2026-09-19 08:40:45 | INFO | train rank=0 | step=38110/50000 progress=76.2% loss=0.095213 lr=1.467975e-05 lr_backbone=1.467975e-05 elapsed=39:12:08 avg_step=3.70s recent_step=3.28s eta=12:13:50 +2026-09-19 08:41:17 | INFO | train rank=0 | step=38120/50000 progress=76.2% loss=0.116772 lr=1.465635e-05 lr_backbone=1.465635e-05 elapsed=39:12:40 avg_step=3.70s recent_step=3.21s eta=12:13:12 +2026-09-19 08:42:11 | INFO | train rank=0 | step=38130/50000 progress=76.3% loss=0.097959 lr=1.463297e-05 lr_backbone=1.463297e-05 elapsed=39:13:35 avg_step=3.70s recent_step=5.44s eta=12:12:40 +2026-09-19 08:42:42 | INFO | train rank=0 | step=38140/50000 progress=76.3% loss=0.099350 lr=1.460960e-05 lr_backbone=1.460960e-05 elapsed=39:14:06 avg_step=3.70s recent_step=3.10s eta=12:12:01 +2026-09-19 08:43:14 | INFO | train rank=0 | step=38150/50000 progress=76.3% loss=0.079694 lr=1.458625e-05 lr_backbone=1.458625e-05 elapsed=39:14:37 avg_step=3.70s recent_step=3.17s eta=12:11:23 +2026-09-19 08:43:50 | INFO | train rank=0 | step=38160/50000 progress=76.3% loss=0.077654 lr=1.456291e-05 lr_backbone=1.456291e-05 elapsed=39:15:14 avg_step=3.70s recent_step=3.64s eta=12:10:45 +2026-09-19 08:44:46 | INFO | train rank=0 | step=38170/50000 progress=76.3% loss=0.092433 lr=1.453959e-05 lr_backbone=1.453959e-05 elapsed=39:16:09 avg_step=3.70s recent_step=5.55s eta=12:10:14 +2026-09-19 08:45:19 | INFO | train rank=0 | step=38180/50000 progress=76.4% loss=0.093484 lr=1.451629e-05 lr_backbone=1.451629e-05 elapsed=39:16:43 avg_step=3.70s recent_step=3.35s eta=12:09:36 +2026-09-19 08:45:48 | INFO | train rank=0 | step=38190/50000 progress=76.4% loss=0.074014 lr=1.449299e-05 lr_backbone=1.449299e-05 elapsed=39:17:11 avg_step=3.70s recent_step=2.85s eta=12:08:56 +2026-09-19 08:46:17 | INFO | train rank=0 | step=38200/50000 progress=76.4% loss=0.097280 lr=1.446972e-05 lr_backbone=1.446972e-05 elapsed=39:17:40 avg_step=3.70s recent_step=2.89s eta=12:08:17 +2026-09-19 08:47:09 | INFO | train rank=0 | step=38210/50000 progress=76.4% loss=0.118636 lr=1.444646e-05 lr_backbone=1.444646e-05 elapsed=39:18:33 avg_step=3.70s recent_step=5.24s eta=12:07:44 +2026-09-19 08:47:45 | INFO | train rank=0 | step=38220/50000 progress=76.4% loss=0.086359 lr=1.442322e-05 lr_backbone=1.442322e-05 elapsed=39:19:08 avg_step=3.70s recent_step=3.59s eta=12:07:07 +2026-09-19 08:48:13 | INFO | train rank=0 | step=38230/50000 progress=76.5% loss=0.077966 lr=1.439999e-05 lr_backbone=1.439999e-05 elapsed=39:19:37 avg_step=3.70s recent_step=2.83s eta=12:06:27 +2026-09-19 08:48:47 | INFO | train rank=0 | step=38240/50000 progress=76.5% loss=0.078370 lr=1.437677e-05 lr_backbone=1.437677e-05 elapsed=39:20:10 avg_step=3.70s recent_step=3.37s eta=12:05:49 +2026-09-19 08:49:33 | INFO | train rank=0 | step=38250/50000 progress=76.5% loss=0.091188 lr=1.435358e-05 lr_backbone=1.435358e-05 elapsed=39:20:56 avg_step=3.70s recent_step=4.57s eta=12:05:15 +2026-09-19 08:50:07 | INFO | train rank=0 | step=38260/50000 progress=76.5% loss=0.077338 lr=1.433040e-05 lr_backbone=1.433040e-05 elapsed=39:21:30 avg_step=3.70s recent_step=3.44s eta=12:04:37 +2026-09-19 08:50:54 | INFO | train rank=0 | step=38270/50000 progress=76.5% loss=0.097435 lr=1.430723e-05 lr_backbone=1.430723e-05 elapsed=39:22:17 avg_step=3.70s recent_step=4.65s eta=12:04:03 +2026-09-19 08:51:22 | INFO | train rank=0 | step=38280/50000 progress=76.6% loss=0.076353 lr=1.428408e-05 lr_backbone=1.428408e-05 elapsed=39:22:46 avg_step=3.70s recent_step=2.89s eta=12:03:23 +2026-09-19 08:52:06 | INFO | train rank=0 | step=38290/50000 progress=76.6% loss=0.068630 lr=1.426094e-05 lr_backbone=1.426094e-05 elapsed=39:23:29 avg_step=3.70s recent_step=4.35s eta=12:02:48 +2026-09-19 08:52:39 | INFO | train rank=0 | step=38300/50000 progress=76.6% loss=0.083332 lr=1.423783e-05 lr_backbone=1.423783e-05 elapsed=39:24:02 avg_step=3.70s recent_step=3.31s eta=12:02:10 +2026-09-19 08:53:19 | INFO | train rank=0 | step=38310/50000 progress=76.6% loss=0.070805 lr=1.421472e-05 lr_backbone=1.421472e-05 elapsed=39:24:42 avg_step=3.70s recent_step=3.95s eta=12:01:34 +2026-09-19 08:53:49 | INFO | train rank=0 | step=38320/50000 progress=76.6% loss=0.071500 lr=1.419163e-05 lr_backbone=1.419163e-05 elapsed=39:25:12 avg_step=3.70s recent_step=3.02s eta=12:00:55 +2026-09-19 08:54:28 | INFO | train rank=0 | step=38330/50000 progress=76.7% loss=0.075150 lr=1.416856e-05 lr_backbone=1.416856e-05 elapsed=39:25:51 avg_step=3.70s recent_step=3.88s eta=12:00:18 +2026-09-19 08:54:59 | INFO | train rank=0 | step=38340/50000 progress=76.7% loss=0.076349 lr=1.414551e-05 lr_backbone=1.414551e-05 elapsed=39:26:22 avg_step=3.70s recent_step=3.14s eta=11:59:39 +2026-09-19 08:55:53 | INFO | train rank=0 | step=38350/50000 progress=76.7% loss=0.074510 lr=1.412246e-05 lr_backbone=1.412246e-05 elapsed=39:27:16 avg_step=3.70s recent_step=5.41s eta=11:59:08 +2026-09-19 08:56:20 | INFO | train rank=0 | step=38360/50000 progress=76.7% loss=0.086314 lr=1.409944e-05 lr_backbone=1.409944e-05 elapsed=39:27:44 avg_step=3.70s recent_step=2.72s eta=11:58:28 +2026-09-19 08:56:59 | INFO | train rank=0 | step=38370/50000 progress=76.7% loss=0.089574 lr=1.407643e-05 lr_backbone=1.407643e-05 elapsed=39:28:22 avg_step=3.70s recent_step=3.88s eta=11:57:51 +2026-09-19 08:57:23 | INFO | train rank=0 | step=38380/50000 progress=76.8% loss=0.096913 lr=1.405344e-05 lr_backbone=1.405344e-05 elapsed=39:28:47 avg_step=3.70s recent_step=2.43s eta=11:57:10 +2026-09-19 08:58:12 | INFO | train rank=0 | step=38390/50000 progress=76.8% loss=0.091119 lr=1.403046e-05 lr_backbone=1.403046e-05 elapsed=39:29:35 avg_step=3.70s recent_step=4.84s eta=11:56:37 +2026-09-19 08:58:39 | INFO | train rank=0 | step=38400/50000 progress=76.8% loss=0.075947 lr=1.400750e-05 lr_backbone=1.400750e-05 elapsed=39:30:02 avg_step=3.70s recent_step=2.73s eta=11:55:57 +2026-09-19 08:59:14 | INFO | train rank=0 | step=38410/50000 progress=76.8% loss=0.113142 lr=1.398455e-05 lr_backbone=1.398455e-05 elapsed=39:30:38 avg_step=3.70s recent_step=3.52s eta=11:55:19 +2026-09-19 08:59:49 | INFO | train rank=0 | step=38420/50000 progress=76.8% loss=0.092616 lr=1.396162e-05 lr_backbone=1.396162e-05 elapsed=39:31:13 avg_step=3.70s recent_step=3.51s eta=11:54:41 +2026-09-19 09:00:28 | INFO | train rank=0 | step=38430/50000 progress=76.9% loss=0.106089 lr=1.393870e-05 lr_backbone=1.393870e-05 elapsed=39:31:51 avg_step=3.70s recent_step=3.82s eta=11:54:05 +2026-09-19 09:01:06 | INFO | train rank=0 | step=38440/50000 progress=76.9% loss=0.072317 lr=1.391580e-05 lr_backbone=1.391580e-05 elapsed=39:32:29 avg_step=3.70s recent_step=3.87s eta=11:53:28 +2026-09-19 09:01:43 | INFO | train rank=0 | step=38450/50000 progress=76.9% loss=0.071866 lr=1.389292e-05 lr_backbone=1.389292e-05 elapsed=39:33:07 avg_step=3.70s recent_step=3.71s eta=11:52:51 +2026-09-19 09:02:19 | INFO | train rank=0 | step=38460/50000 progress=76.9% loss=0.070295 lr=1.387005e-05 lr_backbone=1.387005e-05 elapsed=39:33:43 avg_step=3.70s recent_step=3.60s eta=11:52:14 +2026-09-19 09:03:09 | INFO | train rank=0 | step=38470/50000 progress=76.9% loss=0.092677 lr=1.384720e-05 lr_backbone=1.384720e-05 elapsed=39:34:32 avg_step=3.70s recent_step=4.97s eta=11:51:41 +2026-09-19 09:03:41 | INFO | train rank=0 | step=38480/50000 progress=77.0% loss=0.090804 lr=1.382436e-05 lr_backbone=1.382436e-05 elapsed=39:35:04 avg_step=3.70s recent_step=3.20s eta=11:51:02 +2026-09-19 09:04:17 | INFO | train rank=0 | step=38490/50000 progress=77.0% loss=0.112839 lr=1.380154e-05 lr_backbone=1.380154e-05 elapsed=39:35:41 avg_step=3.70s recent_step=3.63s eta=11:50:25 +2026-09-19 09:04:55 | INFO | train rank=0 | step=38500/50000 progress=77.0% loss=0.089002 lr=1.377874e-05 lr_backbone=1.377874e-05 elapsed=39:36:19 avg_step=3.70s recent_step=3.80s eta=11:49:48 +2026-09-19 09:05:41 | INFO | train rank=0 | step=38510/50000 progress=77.0% loss=0.074457 lr=1.375595e-05 lr_backbone=1.375595e-05 elapsed=39:37:04 avg_step=3.70s recent_step=4.53s eta=11:49:14 +2026-09-19 09:06:15 | INFO | train rank=0 | step=38520/50000 progress=77.0% loss=0.079281 lr=1.373318e-05 lr_backbone=1.373318e-05 elapsed=39:37:38 avg_step=3.70s recent_step=3.41s eta=11:48:36 +2026-09-19 09:06:58 | INFO | train rank=0 | step=38530/50000 progress=77.1% loss=0.091110 lr=1.371042e-05 lr_backbone=1.371042e-05 elapsed=39:38:21 avg_step=3.70s recent_step=4.33s eta=11:48:00 +2026-09-19 09:07:29 | INFO | train rank=0 | step=38540/50000 progress=77.1% loss=0.113635 lr=1.368768e-05 lr_backbone=1.368768e-05 elapsed=39:38:53 avg_step=3.70s recent_step=3.13s eta=11:47:22 +2026-09-19 09:08:23 | INFO | train rank=0 | step=38550/50000 progress=77.1% loss=0.081663 lr=1.366496e-05 lr_backbone=1.366496e-05 elapsed=39:39:46 avg_step=3.70s recent_step=5.32s eta=11:46:49 +2026-09-19 09:08:56 | INFO | train rank=0 | step=38560/50000 progress=77.1% loss=0.099712 lr=1.364225e-05 lr_backbone=1.364225e-05 elapsed=39:40:20 avg_step=3.70s recent_step=3.38s eta=11:46:11 +2026-09-19 09:09:31 | INFO | train rank=0 | step=38570/50000 progress=77.1% loss=0.085062 lr=1.361955e-05 lr_backbone=1.361955e-05 elapsed=39:40:54 avg_step=3.70s recent_step=3.49s eta=11:45:34 +2026-09-19 09:10:06 | INFO | train rank=0 | step=38580/50000 progress=77.2% loss=0.109496 lr=1.359688e-05 lr_backbone=1.359688e-05 elapsed=39:41:29 avg_step=3.70s recent_step=3.46s eta=11:44:56 +2026-09-19 09:11:00 | INFO | train rank=0 | step=38590/50000 progress=77.2% loss=0.132134 lr=1.357421e-05 lr_backbone=1.357421e-05 elapsed=39:42:23 avg_step=3.70s recent_step=5.38s eta=11:44:24 +2026-09-19 09:11:34 | INFO | train rank=0 | step=38600/50000 progress=77.2% loss=0.099154 lr=1.355157e-05 lr_backbone=1.355157e-05 elapsed=39:42:57 avg_step=3.70s recent_step=3.40s eta=11:43:46 +2026-09-19 09:11:58 | INFO | train rank=0 | step=38610/50000 progress=77.2% loss=0.074086 lr=1.352894e-05 lr_backbone=1.352894e-05 elapsed=39:43:21 avg_step=3.70s recent_step=2.44s eta=11:43:05 +2026-09-19 09:12:31 | INFO | train rank=0 | step=38620/50000 progress=77.2% loss=0.114985 lr=1.350633e-05 lr_backbone=1.350633e-05 elapsed=39:43:55 avg_step=3.70s recent_step=3.33s eta=11:42:27 +2026-09-19 09:13:30 | INFO | train rank=0 | step=38630/50000 progress=77.3% loss=0.074089 lr=1.348373e-05 lr_backbone=1.348373e-05 elapsed=39:44:53 avg_step=3.70s recent_step=5.83s eta=11:41:56 +2026-09-19 09:14:03 | INFO | train rank=0 | step=38640/50000 progress=77.3% loss=0.123632 lr=1.346115e-05 lr_backbone=1.346115e-05 elapsed=39:45:26 avg_step=3.70s recent_step=3.32s eta=11:41:18 +2026-09-19 09:14:40 | INFO | train rank=0 | step=38650/50000 progress=77.3% loss=0.080326 lr=1.343858e-05 lr_backbone=1.343858e-05 elapsed=39:46:03 avg_step=3.70s recent_step=3.68s eta=11:40:41 +2026-09-19 09:15:09 | INFO | train rank=0 | step=38660/50000 progress=77.3% loss=0.078630 lr=1.341603e-05 lr_backbone=1.341603e-05 elapsed=39:46:33 avg_step=3.70s recent_step=2.97s eta=11:40:02 +2026-09-19 09:16:00 | INFO | train rank=0 | step=38670/50000 progress=77.3% loss=0.078086 lr=1.339350e-05 lr_backbone=1.339350e-05 elapsed=39:47:24 avg_step=3.70s recent_step=5.10s eta=11:39:29 +2026-09-19 09:16:37 | INFO | train rank=0 | step=38680/50000 progress=77.4% loss=0.099266 lr=1.337098e-05 lr_backbone=1.337098e-05 elapsed=39:48:00 avg_step=3.70s recent_step=3.67s eta=11:38:52 +2026-09-19 09:17:01 | INFO | train rank=0 | step=38690/50000 progress=77.4% loss=0.080855 lr=1.334848e-05 lr_backbone=1.334848e-05 elapsed=39:48:24 avg_step=3.70s recent_step=2.42s eta=11:38:11 +2026-09-19 09:17:35 | INFO | train rank=0 | step=38700/50000 progress=77.4% loss=0.083713 lr=1.332599e-05 lr_backbone=1.332599e-05 elapsed=39:48:58 avg_step=3.70s recent_step=3.35s eta=11:37:33 +2026-09-19 09:18:32 | INFO | train rank=0 | step=38710/50000 progress=77.4% loss=0.087532 lr=1.330352e-05 lr_backbone=1.330352e-05 elapsed=39:49:56 avg_step=3.70s recent_step=5.76s eta=11:37:02 +2026-09-19 09:19:01 | INFO | train rank=0 | step=38720/50000 progress=77.4% loss=0.090809 lr=1.328107e-05 lr_backbone=1.328107e-05 elapsed=39:50:24 avg_step=3.70s recent_step=2.87s eta=11:36:22 +2026-09-19 09:19:33 | INFO | train rank=0 | step=38730/50000 progress=77.5% loss=0.092297 lr=1.325863e-05 lr_backbone=1.325863e-05 elapsed=39:50:57 avg_step=3.70s recent_step=3.22s eta=11:35:44 +2026-09-19 09:20:02 | INFO | train rank=0 | step=38740/50000 progress=77.5% loss=0.092553 lr=1.323621e-05 lr_backbone=1.323621e-05 elapsed=39:51:25 avg_step=3.70s recent_step=2.86s eta=11:35:04 +2026-09-19 09:20:48 | INFO | train rank=0 | step=38750/50000 progress=77.5% loss=0.081761 lr=1.321380e-05 lr_backbone=1.321380e-05 elapsed=39:52:11 avg_step=3.70s recent_step=4.64s eta=11:34:30 +2026-09-19 09:21:21 | INFO | train rank=0 | step=38760/50000 progress=77.5% loss=0.081192 lr=1.319142e-05 lr_backbone=1.319142e-05 elapsed=39:52:44 avg_step=3.70s recent_step=3.24s eta=11:33:52 +2026-09-19 09:21:54 | INFO | train rank=0 | step=38770/50000 progress=77.5% loss=0.081539 lr=1.316904e-05 lr_backbone=1.316904e-05 elapsed=39:53:17 avg_step=3.70s recent_step=3.30s eta=11:33:13 +2026-09-19 09:22:20 | INFO | train rank=0 | step=38780/50000 progress=77.6% loss=0.091883 lr=1.314669e-05 lr_backbone=1.314669e-05 elapsed=39:53:44 avg_step=3.70s recent_step=2.67s eta=11:32:33 +2026-09-19 09:23:15 | INFO | train rank=0 | step=38790/50000 progress=77.6% loss=0.094852 lr=1.312434e-05 lr_backbone=1.312434e-05 elapsed=39:54:39 avg_step=3.70s recent_step=5.50s eta=11:32:02 +2026-09-19 09:23:48 | INFO | train rank=0 | step=38800/50000 progress=77.6% loss=0.089631 lr=1.310202e-05 lr_backbone=1.310202e-05 elapsed=39:55:12 avg_step=3.70s recent_step=3.30s eta=11:31:23 +2026-09-19 09:24:24 | INFO | train rank=0 | step=38810/50000 progress=77.6% loss=0.100957 lr=1.307971e-05 lr_backbone=1.307971e-05 elapsed=39:55:47 avg_step=3.70s recent_step=3.55s eta=11:30:46 +2026-09-19 09:24:55 | INFO | train rank=0 | step=38820/50000 progress=77.6% loss=0.098758 lr=1.305742e-05 lr_backbone=1.305742e-05 elapsed=39:56:18 avg_step=3.70s recent_step=3.07s eta=11:30:07 +2026-09-19 09:25:48 | INFO | train rank=0 | step=38830/50000 progress=77.7% loss=0.088978 lr=1.303514e-05 lr_backbone=1.303514e-05 elapsed=39:57:12 avg_step=3.70s recent_step=5.38s eta=11:29:35 +2026-09-19 09:26:22 | INFO | train rank=0 | step=38840/50000 progress=77.7% loss=0.128314 lr=1.301288e-05 lr_backbone=1.301288e-05 elapsed=39:57:45 avg_step=3.70s recent_step=3.38s eta=11:28:57 +2026-09-19 09:26:54 | INFO | train rank=0 | step=38850/50000 progress=77.7% loss=0.067826 lr=1.299064e-05 lr_backbone=1.299064e-05 elapsed=39:58:17 avg_step=3.70s recent_step=3.17s eta=11:28:18 +2026-09-19 09:27:28 | INFO | train rank=0 | step=38860/50000 progress=77.7% loss=0.159449 lr=1.296841e-05 lr_backbone=1.296841e-05 elapsed=39:58:51 avg_step=3.70s recent_step=3.43s eta=11:27:40 +2026-09-19 09:28:28 | INFO | train rank=0 | step=38870/50000 progress=77.7% loss=0.073616 lr=1.294620e-05 lr_backbone=1.294620e-05 elapsed=39:59:52 avg_step=3.70s recent_step=6.03s eta=11:27:10 +2026-09-19 09:28:56 | INFO | train rank=0 | step=38880/50000 progress=77.8% loss=0.121138 lr=1.292400e-05 lr_backbone=1.292400e-05 elapsed=40:00:19 avg_step=3.70s recent_step=2.76s eta=11:26:30 +2026-09-19 09:29:34 | INFO | train rank=0 | step=38890/50000 progress=77.8% loss=0.083956 lr=1.290182e-05 lr_backbone=1.290182e-05 elapsed=40:00:57 avg_step=3.70s recent_step=3.80s eta=11:25:54 +2026-09-19 09:30:00 | INFO | train rank=0 | step=38900/50000 progress=77.8% loss=0.099160 lr=1.287966e-05 lr_backbone=1.287966e-05 elapsed=40:01:24 avg_step=3.70s recent_step=2.63s eta=11:25:13 +2026-09-19 09:30:53 | INFO | train rank=0 | step=38910/50000 progress=77.8% loss=0.092364 lr=1.285751e-05 lr_backbone=1.285751e-05 elapsed=40:02:16 avg_step=3.70s recent_step=5.26s eta=11:24:41 +2026-09-19 09:31:27 | INFO | train rank=0 | step=38920/50000 progress=77.8% loss=0.100545 lr=1.283538e-05 lr_backbone=1.283538e-05 elapsed=40:02:50 avg_step=3.70s recent_step=3.39s eta=11:24:03 +2026-09-19 09:32:06 | INFO | train rank=0 | step=38930/50000 progress=77.9% loss=0.093382 lr=1.281327e-05 lr_backbone=1.281327e-05 elapsed=40:03:29 avg_step=3.70s recent_step=3.94s eta=11:23:27 +2026-09-19 09:32:33 | INFO | train rank=0 | step=38940/50000 progress=77.9% loss=0.093263 lr=1.279117e-05 lr_backbone=1.279117e-05 elapsed=40:03:56 avg_step=3.70s recent_step=2.66s eta=11:22:47 +2026-09-19 09:33:31 | INFO | train rank=0 | step=38950/50000 progress=77.9% loss=0.068637 lr=1.276909e-05 lr_backbone=1.276909e-05 elapsed=40:04:54 avg_step=3.70s recent_step=5.80s eta=11:22:15 +2026-09-19 09:33:53 | INFO | train rank=0 | step=38960/50000 progress=77.9% loss=0.078287 lr=1.274702e-05 lr_backbone=1.274702e-05 elapsed=40:05:16 avg_step=3.70s recent_step=2.22s eta=11:21:34 +2026-09-19 09:34:26 | INFO | train rank=0 | step=38970/50000 progress=77.9% loss=0.076080 lr=1.272498e-05 lr_backbone=1.272498e-05 elapsed=40:05:49 avg_step=3.70s recent_step=3.29s eta=11:20:56 +2026-09-19 09:35:03 | INFO | train rank=0 | step=38980/50000 progress=78.0% loss=0.075333 lr=1.270294e-05 lr_backbone=1.270294e-05 elapsed=40:06:27 avg_step=3.70s recent_step=3.76s eta=11:20:19 +2026-09-19 09:35:53 | INFO | train rank=0 | step=38990/50000 progress=78.0% loss=0.086099 lr=1.268093e-05 lr_backbone=1.268093e-05 elapsed=40:07:16 avg_step=3.70s recent_step=4.95s eta=11:19:46 +2026-09-19 09:36:21 | INFO | train rank=0 | step=39000/50000 progress=78.0% loss=0.076259 lr=1.265893e-05 lr_backbone=1.265893e-05 elapsed=40:07:44 avg_step=3.70s recent_step=2.78s eta=11:19:06 +2026-09-19 09:36:54 | INFO | train rank=0 | step=39010/50000 progress=78.0% loss=0.122976 lr=1.263694e-05 lr_backbone=1.263694e-05 elapsed=40:08:17 avg_step=3.70s recent_step=3.30s eta=11:18:28 +2026-09-19 09:37:29 | INFO | train rank=0 | step=39020/50000 progress=78.0% loss=0.089486 lr=1.261497e-05 lr_backbone=1.261497e-05 elapsed=40:08:53 avg_step=3.70s recent_step=3.56s eta=11:17:50 +2026-09-19 09:38:16 | INFO | train rank=0 | step=39030/50000 progress=78.1% loss=0.079152 lr=1.259302e-05 lr_backbone=1.259302e-05 elapsed=40:09:39 avg_step=3.70s recent_step=4.64s eta=11:17:16 +2026-09-19 09:38:49 | INFO | train rank=0 | step=39040/50000 progress=78.1% loss=0.076253 lr=1.257109e-05 lr_backbone=1.257109e-05 elapsed=40:10:12 avg_step=3.70s recent_step=3.29s eta=11:16:38 +2026-09-19 09:39:18 | INFO | train rank=0 | step=39050/50000 progress=78.1% loss=0.112735 lr=1.254917e-05 lr_backbone=1.254917e-05 elapsed=40:10:42 avg_step=3.70s recent_step=2.95s eta=11:15:59 +2026-09-19 09:39:44 | INFO | train rank=0 | step=39060/50000 progress=78.1% loss=0.079715 lr=1.252727e-05 lr_backbone=1.252727e-05 elapsed=40:11:08 avg_step=3.70s recent_step=2.63s eta=11:15:18 +2026-09-19 09:40:38 | INFO | train rank=0 | step=39070/50000 progress=78.1% loss=0.081232 lr=1.250538e-05 lr_backbone=1.250538e-05 elapsed=40:12:01 avg_step=3.70s recent_step=5.33s eta=11:14:46 +2026-09-19 09:41:16 | INFO | train rank=0 | step=39080/50000 progress=78.2% loss=0.111860 lr=1.248351e-05 lr_backbone=1.248351e-05 elapsed=40:12:39 avg_step=3.70s recent_step=3.81s eta=11:14:09 +2026-09-19 09:41:50 | INFO | train rank=0 | step=39090/50000 progress=78.2% loss=0.099008 lr=1.246166e-05 lr_backbone=1.246166e-05 elapsed=40:13:13 avg_step=3.70s recent_step=3.39s eta=11:13:31 +2026-09-19 09:42:25 | INFO | train rank=0 | step=39100/50000 progress=78.2% loss=0.080202 lr=1.243982e-05 lr_backbone=1.243982e-05 elapsed=40:13:48 avg_step=3.70s recent_step=3.51s eta=11:12:54 +2026-09-19 09:43:15 | INFO | train rank=0 | step=39110/50000 progress=78.2% loss=0.093270 lr=1.241800e-05 lr_backbone=1.241800e-05 elapsed=40:14:38 avg_step=3.70s recent_step=5.00s eta=11:12:20 +2026-09-19 09:43:56 | INFO | train rank=0 | step=39120/50000 progress=78.2% loss=0.140505 lr=1.239620e-05 lr_backbone=1.239620e-05 elapsed=40:15:19 avg_step=3.70s recent_step=4.09s eta=11:11:44 +2026-09-19 09:44:27 | INFO | train rank=0 | step=39130/50000 progress=78.3% loss=0.093774 lr=1.237441e-05 lr_backbone=1.237441e-05 elapsed=40:15:50 avg_step=3.70s recent_step=3.06s eta=11:11:06 +2026-09-19 09:44:58 | INFO | train rank=0 | step=39140/50000 progress=78.3% loss=0.099505 lr=1.235264e-05 lr_backbone=1.235264e-05 elapsed=40:16:21 avg_step=3.70s recent_step=3.14s eta=11:10:27 +2026-09-19 09:45:51 | INFO | train rank=0 | step=39150/50000 progress=78.3% loss=0.101475 lr=1.233089e-05 lr_backbone=1.233089e-05 elapsed=40:17:15 avg_step=3.70s recent_step=5.36s eta=11:09:54 +2026-09-19 09:46:23 | INFO | train rank=0 | step=39160/50000 progress=78.3% loss=0.135200 lr=1.230915e-05 lr_backbone=1.230915e-05 elapsed=40:17:47 avg_step=3.70s recent_step=3.18s eta=11:09:16 +2026-09-19 09:46:50 | INFO | train rank=0 | step=39170/50000 progress=78.3% loss=0.111837 lr=1.228743e-05 lr_backbone=1.228743e-05 elapsed=40:18:13 avg_step=3.70s recent_step=2.65s eta=11:08:36 +2026-09-19 09:47:28 | INFO | train rank=0 | step=39180/50000 progress=78.4% loss=0.092822 lr=1.226573e-05 lr_backbone=1.226573e-05 elapsed=40:18:52 avg_step=3.70s recent_step=3.86s eta=11:07:59 +2026-09-19 09:48:19 | INFO | train rank=0 | step=39190/50000 progress=78.4% loss=0.099945 lr=1.224404e-05 lr_backbone=1.224404e-05 elapsed=40:19:42 avg_step=3.70s recent_step=5.02s eta=11:07:26 +2026-09-19 09:48:54 | INFO | train rank=0 | step=39200/50000 progress=78.4% loss=0.104572 lr=1.222237e-05 lr_backbone=1.222237e-05 elapsed=40:20:17 avg_step=3.70s recent_step=3.49s eta=11:06:48 +2026-09-19 09:49:23 | INFO | train rank=0 | step=39210/50000 progress=78.4% loss=0.066409 lr=1.220071e-05 lr_backbone=1.220071e-05 elapsed=40:20:46 avg_step=3.70s recent_step=2.96s eta=11:06:09 +2026-09-19 09:49:56 | INFO | train rank=0 | step=39220/50000 progress=78.4% loss=0.089905 lr=1.217907e-05 lr_backbone=1.217907e-05 elapsed=40:21:20 avg_step=3.70s recent_step=3.33s eta=11:05:31 +2026-09-19 09:50:43 | INFO | train rank=0 | step=39230/50000 progress=78.5% loss=0.079745 lr=1.215745e-05 lr_backbone=1.215745e-05 elapsed=40:22:06 avg_step=3.70s recent_step=4.65s eta=11:04:57 +2026-09-19 09:51:13 | INFO | train rank=0 | step=39240/50000 progress=78.5% loss=0.108625 lr=1.213585e-05 lr_backbone=1.213585e-05 elapsed=40:22:36 avg_step=3.70s recent_step=3.00s eta=11:04:18 +2026-09-19 09:51:44 | INFO | train rank=0 | step=39250/50000 progress=78.5% loss=0.069287 lr=1.211426e-05 lr_backbone=1.211426e-05 elapsed=40:23:08 avg_step=3.70s recent_step=3.14s eta=11:03:39 +2026-09-19 09:52:18 | INFO | train rank=0 | step=39260/50000 progress=78.5% loss=0.093049 lr=1.209268e-05 lr_backbone=1.209268e-05 elapsed=40:23:41 avg_step=3.70s recent_step=3.36s eta=11:03:01 +2026-09-19 09:53:02 | INFO | train rank=0 | step=39270/50000 progress=78.5% loss=0.113407 lr=1.207113e-05 lr_backbone=1.207113e-05 elapsed=40:24:25 avg_step=3.70s recent_step=4.36s eta=11:02:26 +2026-09-19 09:53:39 | INFO | train rank=0 | step=39280/50000 progress=78.6% loss=0.073376 lr=1.204959e-05 lr_backbone=1.204959e-05 elapsed=40:25:02 avg_step=3.70s recent_step=3.72s eta=11:01:49 +2026-09-19 09:54:10 | INFO | train rank=0 | step=39290/50000 progress=78.6% loss=0.076409 lr=1.202807e-05 lr_backbone=1.202807e-05 elapsed=40:25:34 avg_step=3.70s recent_step=3.16s eta=11:01:10 +2026-09-19 09:54:46 | INFO | train rank=0 | step=39300/50000 progress=78.6% loss=0.079686 lr=1.200656e-05 lr_backbone=1.200656e-05 elapsed=40:26:09 avg_step=3.70s recent_step=3.57s eta=11:00:33 +2026-09-19 09:55:21 | INFO | train rank=0 | step=39310/50000 progress=78.6% loss=0.132776 lr=1.198507e-05 lr_backbone=1.198507e-05 elapsed=40:26:45 avg_step=3.70s recent_step=3.54s eta=10:59:56 +2026-09-19 09:56:05 | INFO | train rank=0 | step=39320/50000 progress=78.6% loss=0.104660 lr=1.196360e-05 lr_backbone=1.196360e-05 elapsed=40:27:28 avg_step=3.70s recent_step=4.36s eta=10:59:20 +2026-09-19 09:56:40 | INFO | train rank=0 | step=39330/50000 progress=78.7% loss=0.074520 lr=1.194214e-05 lr_backbone=1.194214e-05 elapsed=40:28:03 avg_step=3.70s recent_step=3.46s eta=10:58:43 +2026-09-19 09:57:09 | INFO | train rank=0 | step=39340/50000 progress=78.7% loss=0.111817 lr=1.192070e-05 lr_backbone=1.192070e-05 elapsed=40:28:32 avg_step=3.70s recent_step=2.92s eta=10:58:03 +2026-09-19 09:58:02 | INFO | train rank=0 | step=39350/50000 progress=78.7% loss=0.092060 lr=1.189928e-05 lr_backbone=1.189928e-05 elapsed=40:29:25 avg_step=3.70s recent_step=5.27s eta=10:57:31 +2026-09-19 09:58:32 | INFO | train rank=0 | step=39360/50000 progress=78.7% loss=0.121688 lr=1.187787e-05 lr_backbone=1.187787e-05 elapsed=40:29:55 avg_step=3.70s recent_step=3.00s eta=10:56:52 +2026-09-19 09:59:10 | INFO | train rank=0 | step=39370/50000 progress=78.7% loss=0.082239 lr=1.185649e-05 lr_backbone=1.185649e-05 elapsed=40:30:34 avg_step=3.70s recent_step=3.88s eta=10:56:15 +2026-09-19 09:59:42 | INFO | train rank=0 | step=39380/50000 progress=78.8% loss=0.065232 lr=1.183511e-05 lr_backbone=1.183511e-05 elapsed=40:31:05 avg_step=3.70s recent_step=3.19s eta=10:55:37 +2026-09-19 10:00:38 | INFO | train rank=0 | step=39390/50000 progress=78.8% loss=0.090180 lr=1.181376e-05 lr_backbone=1.181376e-05 elapsed=40:32:01 avg_step=3.70s recent_step=5.54s eta=10:55:05 +2026-09-19 10:01:09 | INFO | train rank=0 | step=39400/50000 progress=78.8% loss=0.093799 lr=1.179242e-05 lr_backbone=1.179242e-05 elapsed=40:32:32 avg_step=3.70s recent_step=3.12s eta=10:54:26 +2026-09-19 10:01:45 | INFO | train rank=0 | step=39410/50000 progress=78.8% loss=0.091889 lr=1.177109e-05 lr_backbone=1.177109e-05 elapsed=40:33:08 avg_step=3.70s recent_step=3.58s eta=10:53:49 +2026-09-19 10:02:11 | INFO | train rank=0 | step=39420/50000 progress=78.8% loss=0.066511 lr=1.174979e-05 lr_backbone=1.174979e-05 elapsed=40:33:35 avg_step=3.70s recent_step=2.67s eta=10:53:09 +2026-09-19 10:03:15 | INFO | train rank=0 | step=39430/50000 progress=78.9% loss=0.074655 lr=1.172850e-05 lr_backbone=1.172850e-05 elapsed=40:34:39 avg_step=3.70s recent_step=6.40s eta=10:52:39 +2026-09-19 10:03:46 | INFO | train rank=0 | step=39440/50000 progress=78.9% loss=0.099283 lr=1.170723e-05 lr_backbone=1.170723e-05 elapsed=40:35:09 avg_step=3.70s recent_step=3.05s eta=10:52:00 +2026-09-19 10:04:20 | INFO | train rank=0 | step=39450/50000 progress=78.9% loss=0.095911 lr=1.168597e-05 lr_backbone=1.168597e-05 elapsed=40:35:43 avg_step=3.70s recent_step=3.43s eta=10:51:22 +2026-09-19 10:04:46 | INFO | train rank=0 | step=39460/50000 progress=78.9% loss=0.091742 lr=1.166473e-05 lr_backbone=1.166473e-05 elapsed=40:36:09 avg_step=3.70s recent_step=2.58s eta=10:50:42 +2026-09-19 10:05:43 | INFO | train rank=0 | step=39470/50000 progress=78.9% loss=0.080542 lr=1.164351e-05 lr_backbone=1.164351e-05 elapsed=40:37:06 avg_step=3.70s recent_step=5.72s eta=10:50:11 +2026-09-19 10:06:18 | INFO | train rank=0 | step=39480/50000 progress=79.0% loss=0.099185 lr=1.162230e-05 lr_backbone=1.162230e-05 elapsed=40:37:42 avg_step=3.70s recent_step=3.52s eta=10:49:33 +2026-09-19 10:06:47 | INFO | train rank=0 | step=39490/50000 progress=79.0% loss=0.077614 lr=1.160112e-05 lr_backbone=1.160112e-05 elapsed=40:38:10 avg_step=3.70s recent_step=2.83s eta=10:48:54 +2026-09-19 10:07:19 | INFO | train rank=0 | step=39500/50000 progress=79.0% loss=0.081012 lr=1.157994e-05 lr_backbone=1.157994e-05 elapsed=40:38:42 avg_step=3.70s recent_step=3.19s eta=10:48:15 +2026-09-19 10:08:03 | INFO | train rank=0 | step=39510/50000 progress=79.0% loss=0.090635 lr=1.155879e-05 lr_backbone=1.155879e-05 elapsed=40:39:26 avg_step=3.70s recent_step=4.44s eta=10:47:40 +2026-09-19 10:08:43 | INFO | train rank=0 | step=39520/50000 progress=79.0% loss=0.072823 lr=1.153765e-05 lr_backbone=1.153765e-05 elapsed=40:40:07 avg_step=3.70s recent_step=4.04s eta=10:47:04 +2026-09-19 10:09:13 | INFO | train rank=0 | step=39530/50000 progress=79.1% loss=0.091241 lr=1.151653e-05 lr_backbone=1.151653e-05 elapsed=40:40:36 avg_step=3.70s recent_step=2.95s eta=10:46:25 +2026-09-19 10:09:47 | INFO | train rank=0 | step=39540/50000 progress=79.1% loss=0.079756 lr=1.149543e-05 lr_backbone=1.149543e-05 elapsed=40:41:10 avg_step=3.70s recent_step=3.43s eta=10:45:47 +2026-09-19 10:10:33 | INFO | train rank=0 | step=39550/50000 progress=79.1% loss=0.126393 lr=1.147434e-05 lr_backbone=1.147434e-05 elapsed=40:41:56 avg_step=3.70s recent_step=4.58s eta=10:45:13 +2026-09-19 10:11:05 | INFO | train rank=0 | step=39560/50000 progress=79.1% loss=0.096859 lr=1.145327e-05 lr_backbone=1.145327e-05 elapsed=40:42:28 avg_step=3.70s recent_step=3.17s eta=10:44:34 +2026-09-19 10:11:32 | INFO | train rank=0 | step=39570/50000 progress=79.1% loss=0.080834 lr=1.143221e-05 lr_backbone=1.143221e-05 elapsed=40:42:55 avg_step=3.70s recent_step=2.72s eta=10:43:54 +2026-09-19 10:12:06 | INFO | train rank=0 | step=39580/50000 progress=79.2% loss=0.071283 lr=1.141118e-05 lr_backbone=1.141118e-05 elapsed=40:43:29 avg_step=3.70s recent_step=3.41s eta=10:43:17 +2026-09-19 10:12:39 | INFO | train rank=0 | step=39590/50000 progress=79.2% loss=0.080453 lr=1.139016e-05 lr_backbone=1.139016e-05 elapsed=40:44:02 avg_step=3.70s recent_step=3.27s eta=10:42:38 +2026-09-19 10:13:08 | INFO | train rank=0 | step=39600/50000 progress=79.2% loss=0.082918 lr=1.136915e-05 lr_backbone=1.136915e-05 elapsed=40:44:31 avg_step=3.70s recent_step=2.95s eta=10:41:59 +2026-09-19 10:13:35 | INFO | train rank=0 | step=39610/50000 progress=79.2% loss=0.074837 lr=1.134817e-05 lr_backbone=1.134817e-05 elapsed=40:44:59 avg_step=3.70s recent_step=2.72s eta=10:41:20 +2026-09-19 10:14:10 | INFO | train rank=0 | step=39620/50000 progress=79.2% loss=0.112147 lr=1.132720e-05 lr_backbone=1.132720e-05 elapsed=40:45:33 avg_step=3.70s recent_step=3.45s eta=10:40:42 +2026-09-19 10:15:04 | INFO | train rank=0 | step=39630/50000 progress=79.3% loss=0.075224 lr=1.130624e-05 lr_backbone=1.130624e-05 elapsed=40:46:28 avg_step=3.70s recent_step=5.47s eta=10:40:10 +2026-09-19 10:15:37 | INFO | train rank=0 | step=39640/50000 progress=79.3% loss=0.087456 lr=1.128531e-05 lr_backbone=1.128531e-05 elapsed=40:47:00 avg_step=3.70s recent_step=3.23s eta=10:39:31 +2026-09-19 10:16:07 | INFO | train rank=0 | step=39650/50000 progress=79.3% loss=0.079563 lr=1.126439e-05 lr_backbone=1.126439e-05 elapsed=40:47:30 avg_step=3.70s recent_step=3.00s eta=10:38:52 +2026-09-19 10:16:37 | INFO | train rank=0 | step=39660/50000 progress=79.3% loss=0.088347 lr=1.124349e-05 lr_backbone=1.124349e-05 elapsed=40:48:01 avg_step=3.70s recent_step=3.07s eta=10:38:14 +2026-09-19 10:17:31 | INFO | train rank=0 | step=39670/50000 progress=79.3% loss=0.089869 lr=1.122260e-05 lr_backbone=1.122260e-05 elapsed=40:48:54 avg_step=3.70s recent_step=5.36s eta=10:37:41 +2026-09-19 10:18:01 | INFO | train rank=0 | step=39680/50000 progress=79.4% loss=0.060240 lr=1.120174e-05 lr_backbone=1.120174e-05 elapsed=40:49:24 avg_step=3.70s recent_step=2.98s eta=10:37:02 +2026-09-19 10:18:35 | INFO | train rank=0 | step=39690/50000 progress=79.4% loss=0.087576 lr=1.118088e-05 lr_backbone=1.118088e-05 elapsed=40:49:59 avg_step=3.70s recent_step=3.46s eta=10:36:24 +2026-09-19 10:19:06 | INFO | train rank=0 | step=39700/50000 progress=79.4% loss=0.081480 lr=1.116005e-05 lr_backbone=1.116005e-05 elapsed=40:50:29 avg_step=3.70s recent_step=3.05s eta=10:35:46 +2026-09-19 10:20:00 | INFO | train rank=0 | step=39710/50000 progress=79.4% loss=0.084020 lr=1.113923e-05 lr_backbone=1.113923e-05 elapsed=40:51:24 avg_step=3.70s recent_step=5.43s eta=10:35:13 +2026-09-19 10:20:35 | INFO | train rank=0 | step=39720/50000 progress=79.4% loss=0.087631 lr=1.111843e-05 lr_backbone=1.111843e-05 elapsed=40:51:59 avg_step=3.70s recent_step=3.50s eta=10:34:36 +2026-09-19 10:21:10 | INFO | train rank=0 | step=39730/50000 progress=79.5% loss=0.064595 lr=1.109765e-05 lr_backbone=1.109765e-05 elapsed=40:52:33 avg_step=3.70s recent_step=3.44s eta=10:33:58 +2026-09-19 10:21:38 | INFO | train rank=0 | step=39740/50000 progress=79.5% loss=0.094620 lr=1.107688e-05 lr_backbone=1.107688e-05 elapsed=40:53:02 avg_step=3.70s recent_step=2.88s eta=10:33:19 +2026-09-19 10:22:31 | INFO | train rank=0 | step=39750/50000 progress=79.5% loss=0.097537 lr=1.105614e-05 lr_backbone=1.105614e-05 elapsed=40:53:54 avg_step=3.70s recent_step=5.25s eta=10:32:46 +2026-09-19 10:23:10 | INFO | train rank=0 | step=39760/50000 progress=79.5% loss=0.129805 lr=1.103540e-05 lr_backbone=1.103540e-05 elapsed=40:54:33 avg_step=3.70s recent_step=3.90s eta=10:32:09 +2026-09-19 10:23:36 | INFO | train rank=0 | step=39770/50000 progress=79.5% loss=0.083732 lr=1.101469e-05 lr_backbone=1.101469e-05 elapsed=40:54:59 avg_step=3.70s recent_step=2.61s eta=10:31:29 +2026-09-19 10:24:12 | INFO | train rank=0 | step=39780/50000 progress=79.6% loss=0.076057 lr=1.099399e-05 lr_backbone=1.099399e-05 elapsed=40:55:35 avg_step=3.70s recent_step=3.54s eta=10:30:52 +2026-09-19 10:24:56 | INFO | train rank=0 | step=39790/50000 progress=79.6% loss=0.100476 lr=1.097331e-05 lr_backbone=1.097331e-05 elapsed=40:56:20 avg_step=3.70s recent_step=4.47s eta=10:30:17 +2026-09-19 10:25:29 | INFO | train rank=0 | step=39800/50000 progress=79.6% loss=0.094955 lr=1.095265e-05 lr_backbone=1.095265e-05 elapsed=40:56:52 avg_step=3.70s recent_step=3.24s eta=10:29:39 +2026-09-19 10:26:06 | INFO | train rank=0 | step=39810/50000 progress=79.6% loss=0.119568 lr=1.093200e-05 lr_backbone=1.093200e-05 elapsed=40:57:29 avg_step=3.70s recent_step=3.76s eta=10:29:02 +2026-09-19 10:26:38 | INFO | train rank=0 | step=39820/50000 progress=79.6% loss=0.085035 lr=1.091137e-05 lr_backbone=1.091137e-05 elapsed=40:58:01 avg_step=3.70s recent_step=3.17s eta=10:28:23 +2026-09-19 10:27:35 | INFO | train rank=0 | step=39830/50000 progress=79.7% loss=0.073645 lr=1.089076e-05 lr_backbone=1.089076e-05 elapsed=40:58:58 avg_step=3.70s recent_step=5.67s eta=10:27:51 +2026-09-19 10:28:09 | INFO | train rank=0 | step=39840/50000 progress=79.7% loss=0.068929 lr=1.087016e-05 lr_backbone=1.087016e-05 elapsed=40:59:32 avg_step=3.70s recent_step=3.46s eta=10:27:14 +2026-09-19 10:28:38 | INFO | train rank=0 | step=39850/50000 progress=79.7% loss=0.082068 lr=1.084959e-05 lr_backbone=1.084959e-05 elapsed=41:00:01 avg_step=3.70s recent_step=2.90s eta=10:26:34 +2026-09-19 10:29:15 | INFO | train rank=0 | step=39860/50000 progress=79.7% loss=0.084990 lr=1.082902e-05 lr_backbone=1.082902e-05 elapsed=41:00:39 avg_step=3.70s recent_step=3.71s eta=10:25:57 +2026-09-19 10:30:06 | INFO | train rank=0 | step=39870/50000 progress=79.7% loss=0.104201 lr=1.080848e-05 lr_backbone=1.080848e-05 elapsed=41:01:30 avg_step=3.70s recent_step=5.11s eta=10:25:24 +2026-09-19 10:30:38 | INFO | train rank=0 | step=39880/50000 progress=79.8% loss=0.074672 lr=1.078795e-05 lr_backbone=1.078795e-05 elapsed=41:02:01 avg_step=3.70s recent_step=3.12s eta=10:24:45 +2026-09-19 10:31:17 | INFO | train rank=0 | step=39890/50000 progress=79.8% loss=0.075810 lr=1.076744e-05 lr_backbone=1.076744e-05 elapsed=41:02:41 avg_step=3.70s recent_step=3.96s eta=10:24:09 +2026-09-19 10:31:37 | INFO | train rank=0 | step=39900/50000 progress=79.8% loss=0.094917 lr=1.074695e-05 lr_backbone=1.074695e-05 elapsed=41:03:00 avg_step=3.70s recent_step=2.00s eta=10:23:28 +2026-09-19 10:32:31 | INFO | train rank=0 | step=39910/50000 progress=79.8% loss=0.081885 lr=1.072648e-05 lr_backbone=1.072648e-05 elapsed=41:03:55 avg_step=3.70s recent_step=5.41s eta=10:22:55 +2026-09-19 10:33:03 | INFO | train rank=0 | step=39920/50000 progress=79.8% loss=0.070809 lr=1.070602e-05 lr_backbone=1.070602e-05 elapsed=41:04:26 avg_step=3.70s recent_step=3.15s eta=10:22:17 +2026-09-19 10:33:43 | INFO | train rank=0 | step=39930/50000 progress=79.9% loss=0.072174 lr=1.068558e-05 lr_backbone=1.068558e-05 elapsed=41:05:06 avg_step=3.70s recent_step=4.03s eta=10:21:40 +2026-09-19 10:34:10 | INFO | train rank=0 | step=39940/50000 progress=79.9% loss=0.091026 lr=1.066515e-05 lr_backbone=1.066515e-05 elapsed=41:05:34 avg_step=3.70s recent_step=2.74s eta=10:21:01 +2026-09-19 10:34:58 | INFO | train rank=0 | step=39950/50000 progress=79.9% loss=0.086474 lr=1.064475e-05 lr_backbone=1.064475e-05 elapsed=41:06:21 avg_step=3.70s recent_step=4.71s eta=10:20:26 +2026-09-19 10:35:32 | INFO | train rank=0 | step=39960/50000 progress=79.9% loss=0.094653 lr=1.062436e-05 lr_backbone=1.062436e-05 elapsed=41:06:55 avg_step=3.70s recent_step=3.46s eta=10:19:49 +2026-09-19 10:36:12 | INFO | train rank=0 | step=39970/50000 progress=79.9% loss=0.084101 lr=1.060399e-05 lr_backbone=1.060399e-05 elapsed=41:07:35 avg_step=3.70s recent_step=4.00s eta=10:19:12 +2026-09-19 10:36:45 | INFO | train rank=0 | step=39980/50000 progress=80.0% loss=0.090845 lr=1.058363e-05 lr_backbone=1.058363e-05 elapsed=41:08:09 avg_step=3.70s recent_step=3.33s eta=10:18:34 +2026-09-19 10:37:20 | INFO | train rank=0 | step=39990/50000 progress=80.0% loss=0.092339 lr=1.056329e-05 lr_backbone=1.056329e-05 elapsed=41:08:44 avg_step=3.70s recent_step=3.50s eta=10:17:57 +2026-09-19 10:38:27 | INFO | train rank=0 | step=40000/50000 progress=80.0% loss=0.072002 lr=1.054297e-05 lr_backbone=1.054297e-05 elapsed=41:09:50 avg_step=3.70s recent_step=6.63s eta=10:17:27 +2026-09-19 10:39:05 | INFO | train rank=0 | step=40010/50000 progress=80.0% loss=0.089948 lr=1.052267e-05 lr_backbone=1.052267e-05 elapsed=41:10:28 avg_step=3.70s recent_step=3.83s eta=10:16:50 +2026-09-19 10:39:31 | INFO | train rank=0 | step=40020/50000 progress=80.0% loss=0.097275 lr=1.050239e-05 lr_backbone=1.050239e-05 elapsed=41:10:54 avg_step=3.70s recent_step=2.62s eta=10:16:11 +2026-09-19 10:40:03 | INFO | train rank=0 | step=40030/50000 progress=80.1% loss=0.097408 lr=1.048212e-05 lr_backbone=1.048212e-05 elapsed=41:11:26 avg_step=3.70s recent_step=3.17s eta=10:15:32 +2026-09-19 10:40:53 | INFO | train rank=0 | step=40040/50000 progress=80.1% loss=0.082968 lr=1.046187e-05 lr_backbone=1.046187e-05 elapsed=41:12:16 avg_step=3.70s recent_step=4.97s eta=10:14:58 +2026-09-19 10:41:27 | INFO | train rank=0 | step=40050/50000 progress=80.1% loss=0.070921 lr=1.044163e-05 lr_backbone=1.044163e-05 elapsed=41:12:50 avg_step=3.70s recent_step=3.43s eta=10:14:21 +2026-09-19 10:42:01 | INFO | train rank=0 | step=40060/50000 progress=80.1% loss=0.132546 lr=1.042142e-05 lr_backbone=1.042142e-05 elapsed=41:13:25 avg_step=3.70s recent_step=3.44s eta=10:13:43 +2026-09-19 10:42:35 | INFO | train rank=0 | step=40070/50000 progress=80.1% loss=0.070607 lr=1.040122e-05 lr_backbone=1.040122e-05 elapsed=41:13:59 avg_step=3.70s recent_step=3.43s eta=10:13:05 +2026-09-19 10:43:25 | INFO | train rank=0 | step=40080/50000 progress=80.2% loss=0.086652 lr=1.038103e-05 lr_backbone=1.038103e-05 elapsed=41:14:48 avg_step=3.70s recent_step=4.93s eta=10:12:31 +2026-09-19 10:43:49 | INFO | train rank=0 | step=40090/50000 progress=80.2% loss=0.088589 lr=1.036087e-05 lr_backbone=1.036087e-05 elapsed=41:15:12 avg_step=3.70s recent_step=2.39s eta=10:11:51 +2026-09-19 10:44:21 | INFO | train rank=0 | step=40100/50000 progress=80.2% loss=0.119404 lr=1.034072e-05 lr_backbone=1.034072e-05 elapsed=41:15:45 avg_step=3.70s recent_step=3.25s eta=10:11:13 +2026-09-19 10:44:52 | INFO | train rank=0 | step=40110/50000 progress=80.2% loss=0.074231 lr=1.032059e-05 lr_backbone=1.032059e-05 elapsed=41:16:15 avg_step=3.70s recent_step=3.06s eta=10:10:34 +2026-09-19 10:45:43 | INFO | train rank=0 | step=40120/50000 progress=80.2% loss=0.069931 lr=1.030048e-05 lr_backbone=1.030048e-05 elapsed=41:17:06 avg_step=3.70s recent_step=5.09s eta=10:10:00 +2026-09-19 10:46:14 | INFO | train rank=0 | step=40130/50000 progress=80.3% loss=0.083216 lr=1.028038e-05 lr_backbone=1.028038e-05 elapsed=41:17:37 avg_step=3.70s recent_step=3.08s eta=10:09:22 +2026-09-19 10:46:47 | INFO | train rank=0 | step=40140/50000 progress=80.3% loss=0.094736 lr=1.026031e-05 lr_backbone=1.026031e-05 elapsed=41:18:10 avg_step=3.70s recent_step=3.36s eta=10:08:44 +2026-09-19 10:47:25 | INFO | train rank=0 | step=40150/50000 progress=80.3% loss=0.090825 lr=1.024025e-05 lr_backbone=1.024025e-05 elapsed=41:18:48 avg_step=3.70s recent_step=3.75s eta=10:08:07 +2026-09-19 10:48:12 | INFO | train rank=0 | step=40160/50000 progress=80.3% loss=0.076608 lr=1.022020e-05 lr_backbone=1.022020e-05 elapsed=41:19:35 avg_step=3.70s recent_step=4.75s eta=10:07:33 +2026-09-19 10:48:44 | INFO | train rank=0 | step=40170/50000 progress=80.3% loss=0.117176 lr=1.020018e-05 lr_backbone=1.020018e-05 elapsed=41:20:07 avg_step=3.70s recent_step=3.17s eta=10:06:54 +2026-09-19 10:49:16 | INFO | train rank=0 | step=40180/50000 progress=80.4% loss=0.107835 lr=1.018017e-05 lr_backbone=1.018017e-05 elapsed=41:20:39 avg_step=3.70s recent_step=3.20s eta=10:06:16 +2026-09-19 10:49:47 | INFO | train rank=0 | step=40190/50000 progress=80.4% loss=0.123394 lr=1.016018e-05 lr_backbone=1.016018e-05 elapsed=41:21:10 avg_step=3.70s recent_step=3.11s eta=10:05:37 +2026-09-19 10:50:32 | INFO | train rank=0 | step=40200/50000 progress=80.4% loss=0.097702 lr=1.014021e-05 lr_backbone=1.014021e-05 elapsed=41:21:55 avg_step=3.70s recent_step=4.46s eta=10:05:02 +2026-09-19 10:50:57 | INFO | train rank=0 | step=40210/50000 progress=80.4% loss=0.106269 lr=1.012025e-05 lr_backbone=1.012025e-05 elapsed=41:22:21 avg_step=3.70s recent_step=2.57s eta=10:04:22 +2026-09-19 10:51:21 | INFO | train rank=0 | step=40220/50000 progress=80.4% loss=0.081221 lr=1.010031e-05 lr_backbone=1.010031e-05 elapsed=41:22:44 avg_step=3.70s recent_step=2.39s eta=10:03:42 +2026-09-19 10:51:42 | INFO | train rank=0 | step=40230/50000 progress=80.5% loss=0.071051 lr=1.008039e-05 lr_backbone=1.008039e-05 elapsed=41:23:05 avg_step=3.70s recent_step=2.07s eta=10:03:01 +2026-09-19 10:52:11 | INFO | train rank=0 | step=40240/50000 progress=80.5% loss=0.100417 lr=1.006049e-05 lr_backbone=1.006049e-05 elapsed=41:23:34 avg_step=3.70s recent_step=2.93s eta=10:02:22 +2026-09-19 10:52:31 | INFO | train rank=0 | step=40250/50000 progress=80.5% loss=0.065853 lr=1.004060e-05 lr_backbone=1.004060e-05 elapsed=41:23:54 avg_step=3.70s recent_step=2.00s eta=10:01:41 +2026-09-19 10:52:50 | INFO | train rank=0 | step=40260/50000 progress=80.5% loss=0.086904 lr=1.002073e-05 lr_backbone=1.002073e-05 elapsed=41:24:13 avg_step=3.70s recent_step=1.84s eta=10:01:00 +2026-09-19 10:53:15 | INFO | train rank=0 | step=40270/50000 progress=80.5% loss=0.079677 lr=1.000088e-05 lr_backbone=1.000088e-05 elapsed=41:24:38 avg_step=3.70s recent_step=2.52s eta=10:00:20 +2026-09-19 10:53:33 | INFO | train rank=0 | step=40280/50000 progress=80.6% loss=0.075015 lr=9.981047e-06 lr_backbone=9.981047e-06 elapsed=41:24:56 avg_step=3.70s recent_step=1.82s eta=09:59:38 +2026-09-19 10:53:56 | INFO | train rank=0 | step=40290/50000 progress=80.6% loss=0.087095 lr=9.961231e-06 lr_backbone=9.961231e-06 elapsed=41:25:20 avg_step=3.70s recent_step=2.34s eta=09:58:58 +2026-09-19 10:54:15 | INFO | train rank=0 | step=40300/50000 progress=80.6% loss=0.083294 lr=9.941432e-06 lr_backbone=9.941432e-06 elapsed=41:25:38 avg_step=3.70s recent_step=1.81s eta=09:58:16 +2026-09-19 10:54:38 | INFO | train rank=0 | step=40310/50000 progress=80.6% loss=0.074172 lr=9.921651e-06 lr_backbone=9.921651e-06 elapsed=41:26:01 avg_step=3.70s recent_step=2.35s eta=09:57:36 +2026-09-19 10:55:04 | INFO | train rank=0 | step=40320/50000 progress=80.6% loss=0.081769 lr=9.901887e-06 lr_backbone=9.901887e-06 elapsed=41:26:27 avg_step=3.70s recent_step=2.60s eta=09:56:56 +2026-09-19 10:55:36 | INFO | train rank=0 | step=40330/50000 progress=80.7% loss=0.080208 lr=9.882141e-06 lr_backbone=9.882141e-06 elapsed=41:27:00 avg_step=3.70s recent_step=3.23s eta=09:56:18 +2026-09-19 10:56:00 | INFO | train rank=0 | step=40340/50000 progress=80.7% loss=0.089295 lr=9.862413e-06 lr_backbone=9.862413e-06 elapsed=41:27:23 avg_step=3.70s recent_step=2.39s eta=09:55:38 +2026-09-19 10:56:42 | INFO | train rank=0 | step=40350/50000 progress=80.7% loss=0.073590 lr=9.842702e-06 lr_backbone=9.842702e-06 elapsed=41:28:05 avg_step=3.70s recent_step=4.15s eta=09:55:02 +2026-09-19 10:57:08 | INFO | train rank=0 | step=40360/50000 progress=80.7% loss=0.093305 lr=9.823008e-06 lr_backbone=9.823008e-06 elapsed=41:28:31 avg_step=3.70s recent_step=2.64s eta=09:54:23 +2026-09-19 10:57:50 | INFO | train rank=0 | step=40370/50000 progress=80.7% loss=0.121491 lr=9.803332e-06 lr_backbone=9.803332e-06 elapsed=41:29:13 avg_step=3.70s recent_step=4.15s eta=09:53:47 +2026-09-19 10:58:38 | INFO | train rank=0 | step=40380/50000 progress=80.8% loss=0.070981 lr=9.783674e-06 lr_backbone=9.783674e-06 elapsed=41:30:02 avg_step=3.70s recent_step=4.87s eta=09:53:13 +2026-09-19 10:59:33 | INFO | train rank=0 | step=40390/50000 progress=80.8% loss=0.096532 lr=9.764034e-06 lr_backbone=9.764034e-06 elapsed=41:30:56 avg_step=3.70s recent_step=5.47s eta=09:52:40 +2026-09-19 11:00:22 | INFO | train rank=0 | step=40400/50000 progress=80.8% loss=0.081671 lr=9.744411e-06 lr_backbone=9.744411e-06 elapsed=41:31:46 avg_step=3.70s recent_step=4.94s eta=09:52:06 +2026-09-19 11:01:13 | INFO | train rank=0 | step=40410/50000 progress=80.8% loss=0.076843 lr=9.724805e-06 lr_backbone=9.724805e-06 elapsed=41:32:36 avg_step=3.70s recent_step=5.08s eta=09:51:32 +2026-09-19 11:02:01 | INFO | train rank=0 | step=40420/50000 progress=80.8% loss=0.066243 lr=9.705217e-06 lr_backbone=9.705217e-06 elapsed=41:33:24 avg_step=3.70s recent_step=4.77s eta=09:50:57 +2026-09-19 11:02:48 | INFO | train rank=0 | step=40430/50000 progress=80.9% loss=0.099841 lr=9.685647e-06 lr_backbone=9.685647e-06 elapsed=41:34:11 avg_step=3.70s recent_step=4.69s eta=09:50:23 +2026-09-19 11:03:38 | INFO | train rank=0 | step=40440/50000 progress=80.9% loss=0.066760 lr=9.666095e-06 lr_backbone=9.666095e-06 elapsed=41:35:01 avg_step=3.70s recent_step=4.99s eta=09:49:49 +2026-09-19 11:04:23 | INFO | train rank=0 | step=40450/50000 progress=80.9% loss=0.067128 lr=9.646560e-06 lr_backbone=9.646560e-06 elapsed=41:35:47 avg_step=3.70s recent_step=4.58s eta=09:49:14 +2026-09-19 11:05:07 | INFO | train rank=0 | step=40460/50000 progress=80.9% loss=0.082782 lr=9.627043e-06 lr_backbone=9.627043e-06 elapsed=41:36:30 avg_step=3.70s recent_step=4.36s eta=09:48:38 +2026-09-19 11:05:55 | INFO | train rank=0 | step=40470/50000 progress=80.9% loss=0.090523 lr=9.607543e-06 lr_backbone=9.607543e-06 elapsed=41:37:18 avg_step=3.70s recent_step=4.80s eta=09:48:04 +2026-09-19 11:06:45 | INFO | train rank=0 | step=40480/50000 progress=81.0% loss=0.077293 lr=9.588061e-06 lr_backbone=9.588061e-06 elapsed=41:38:09 avg_step=3.70s recent_step=5.02s eta=09:47:30 +2026-09-19 11:07:36 | INFO | train rank=0 | step=40490/50000 progress=81.0% loss=0.096852 lr=9.568597e-06 lr_backbone=9.568597e-06 elapsed=41:38:59 avg_step=3.70s recent_step=5.09s eta=09:46:56 +2026-09-19 11:08:30 | INFO | train rank=0 | step=40500/50000 progress=81.0% loss=0.063412 lr=9.549150e-06 lr_backbone=9.549150e-06 elapsed=41:39:53 avg_step=3.70s recent_step=5.34s eta=09:46:23 +2026-09-19 11:09:20 | INFO | train rank=0 | step=40510/50000 progress=81.0% loss=0.090972 lr=9.529721e-06 lr_backbone=9.529721e-06 elapsed=41:40:43 avg_step=3.70s recent_step=5.01s eta=09:45:49 +2026-09-19 11:10:10 | INFO | train rank=0 | step=40520/50000 progress=81.0% loss=0.092447 lr=9.510310e-06 lr_backbone=9.510310e-06 elapsed=41:41:33 avg_step=3.70s recent_step=5.04s eta=09:45:15 +2026-09-19 11:11:01 | INFO | train rank=0 | step=40530/50000 progress=81.1% loss=0.125217 lr=9.490917e-06 lr_backbone=9.490917e-06 elapsed=41:42:25 avg_step=3.70s recent_step=5.13s eta=09:44:42 +2026-09-19 11:11:47 | INFO | train rank=0 | step=40540/50000 progress=81.1% loss=0.063087 lr=9.471541e-06 lr_backbone=9.471541e-06 elapsed=41:43:10 avg_step=3.70s recent_step=4.54s eta=09:44:06 +2026-09-19 11:12:38 | INFO | train rank=0 | step=40550/50000 progress=81.1% loss=0.061061 lr=9.452183e-06 lr_backbone=9.452183e-06 elapsed=41:44:01 avg_step=3.71s recent_step=5.09s eta=09:43:33 +2026-09-19 11:13:21 | INFO | train rank=0 | step=40560/50000 progress=81.1% loss=0.126408 lr=9.432843e-06 lr_backbone=9.432843e-06 elapsed=41:44:44 avg_step=3.71s recent_step=4.34s eta=09:42:57 +2026-09-19 11:14:10 | INFO | train rank=0 | step=40570/50000 progress=81.1% loss=0.082411 lr=9.413520e-06 lr_backbone=9.413520e-06 elapsed=41:45:34 avg_step=3.71s recent_step=4.94s eta=09:42:23 +2026-09-19 11:14:55 | INFO | train rank=0 | step=40580/50000 progress=81.2% loss=0.090153 lr=9.394216e-06 lr_backbone=9.394216e-06 elapsed=41:46:19 avg_step=3.71s recent_step=4.51s eta=09:41:48 +2026-09-19 11:15:42 | INFO | train rank=0 | step=40590/50000 progress=81.2% loss=0.087703 lr=9.374929e-06 lr_backbone=9.374929e-06 elapsed=41:47:05 avg_step=3.71s recent_step=4.66s eta=09:41:13 +2026-09-19 11:16:33 | INFO | train rank=0 | step=40600/50000 progress=81.2% loss=0.098751 lr=9.355659e-06 lr_backbone=9.355659e-06 elapsed=41:47:56 avg_step=3.71s recent_step=5.08s eta=09:40:39 +2026-09-19 11:17:20 | INFO | train rank=0 | step=40610/50000 progress=81.2% loss=0.107953 lr=9.336408e-06 lr_backbone=9.336408e-06 elapsed=41:48:43 avg_step=3.71s recent_step=4.67s eta=09:40:04 +2026-09-19 11:18:13 | INFO | train rank=0 | step=40620/50000 progress=81.2% loss=0.081478 lr=9.317174e-06 lr_backbone=9.317174e-06 elapsed=41:49:36 avg_step=3.71s recent_step=5.35s eta=09:39:31 +2026-09-19 11:19:02 | INFO | train rank=0 | step=40630/50000 progress=81.3% loss=0.073578 lr=9.297958e-06 lr_backbone=9.297958e-06 elapsed=41:50:25 avg_step=3.71s recent_step=4.90s eta=09:38:57 +2026-09-19 11:19:55 | INFO | train rank=0 | step=40640/50000 progress=81.3% loss=0.098020 lr=9.278760e-06 lr_backbone=9.278760e-06 elapsed=41:51:18 avg_step=3.71s recent_step=5.26s eta=09:38:23 +2026-09-19 11:20:38 | INFO | train rank=0 | step=40650/50000 progress=81.3% loss=0.082454 lr=9.259580e-06 lr_backbone=9.259580e-06 elapsed=41:52:02 avg_step=3.71s recent_step=4.36s eta=09:37:47 +2026-09-19 11:21:27 | INFO | train rank=0 | step=40660/50000 progress=81.3% loss=0.097722 lr=9.240418e-06 lr_backbone=9.240418e-06 elapsed=41:52:50 avg_step=3.71s recent_step=4.88s eta=09:37:13 +2026-09-19 11:22:09 | INFO | train rank=0 | step=40670/50000 progress=81.3% loss=0.077985 lr=9.221273e-06 lr_backbone=9.221273e-06 elapsed=41:53:32 avg_step=3.71s recent_step=4.20s eta=09:36:37 +2026-09-19 11:22:57 | INFO | train rank=0 | step=40680/50000 progress=81.4% loss=0.079164 lr=9.202146e-06 lr_backbone=9.202146e-06 elapsed=41:54:20 avg_step=3.71s recent_step=4.75s eta=09:36:02 +2026-09-19 11:23:46 | INFO | train rank=0 | step=40690/50000 progress=81.4% loss=0.064026 lr=9.183037e-06 lr_backbone=9.183037e-06 elapsed=41:55:09 avg_step=3.71s recent_step=4.91s eta=09:35:28 +2026-09-19 11:24:36 | INFO | train rank=0 | step=40700/50000 progress=81.4% loss=0.093440 lr=9.163946e-06 lr_backbone=9.163946e-06 elapsed=41:55:59 avg_step=3.71s recent_step=5.03s eta=09:34:54 +2026-09-19 11:25:18 | INFO | train rank=0 | step=40710/50000 progress=81.4% loss=0.109725 lr=9.144873e-06 lr_backbone=9.144873e-06 elapsed=41:56:41 avg_step=3.71s recent_step=4.21s eta=09:34:18 +2026-09-19 11:25:53 | INFO | train rank=0 | step=40720/50000 progress=81.4% loss=0.090100 lr=9.125818e-06 lr_backbone=9.125818e-06 elapsed=41:57:16 avg_step=3.71s recent_step=3.47s eta=09:33:40 +2026-09-19 11:26:35 | INFO | train rank=0 | step=40730/50000 progress=81.5% loss=0.163566 lr=9.106780e-06 lr_backbone=9.106780e-06 elapsed=41:57:58 avg_step=3.71s recent_step=4.22s eta=09:33:04 +2026-09-19 11:27:27 | INFO | train rank=0 | step=40740/50000 progress=81.5% loss=0.067861 lr=9.087761e-06 lr_backbone=9.087761e-06 elapsed=41:58:50 avg_step=3.71s recent_step=5.22s eta=09:32:31 +2026-09-19 11:28:42 | INFO | train rank=0 | step=40750/50000 progress=81.5% loss=0.077847 lr=9.068759e-06 lr_backbone=9.068759e-06 elapsed=42:00:05 avg_step=3.71s recent_step=7.47s eta=09:32:02 +2026-09-19 11:30:02 | INFO | train rank=0 | step=40760/50000 progress=81.5% loss=0.071636 lr=9.049776e-06 lr_backbone=9.049776e-06 elapsed=42:01:25 avg_step=3.71s recent_step=8.00s eta=09:31:35 +2026-09-19 11:31:19 | INFO | train rank=0 | step=40770/50000 progress=81.5% loss=0.097936 lr=9.030810e-06 lr_backbone=9.030810e-06 elapsed=42:02:43 avg_step=3.71s recent_step=7.76s eta=09:31:07 +2026-09-19 11:32:31 | INFO | train rank=0 | step=40780/50000 progress=81.6% loss=0.082647 lr=9.011862e-06 lr_backbone=9.011862e-06 elapsed=42:03:54 avg_step=3.71s recent_step=7.18s eta=09:30:38 +2026-09-19 11:33:53 | INFO | train rank=0 | step=40790/50000 progress=81.6% loss=0.079729 lr=8.992932e-06 lr_backbone=8.992932e-06 elapsed=42:05:16 avg_step=3.71s recent_step=8.15s eta=09:30:10 +2026-09-19 11:35:07 | INFO | train rank=0 | step=40800/50000 progress=81.6% loss=0.091422 lr=8.974020e-06 lr_backbone=8.974020e-06 elapsed=42:06:31 avg_step=3.72s recent_step=7.47s eta=09:29:42 +2026-09-19 11:36:21 | INFO | train rank=0 | step=40810/50000 progress=81.6% loss=0.083391 lr=8.955126e-06 lr_backbone=8.955126e-06 elapsed=42:07:45 avg_step=3.72s recent_step=7.39s eta=09:29:13 +2026-09-19 11:39:46 | INFO | train rank=0 | step=40820/50000 progress=81.6% loss=0.100290 lr=8.936249e-06 lr_backbone=8.936249e-06 elapsed=42:11:09 avg_step=3.72s recent_step=20.47s eta=09:29:13 +2026-09-19 11:43:22 | INFO | train rank=0 | step=40830/50000 progress=81.7% loss=0.100108 lr=8.917391e-06 lr_backbone=8.917391e-06 elapsed=42:14:45 avg_step=3.72s recent_step=21.62s eta=09:29:16 +2026-09-19 11:43:48 | INFO | train rank=0 | step=40840/50000 progress=81.7% loss=0.082263 lr=8.898551e-06 lr_backbone=8.898551e-06 elapsed=42:15:11 avg_step=3.72s recent_step=2.56s eta=09:28:37 +2026-09-19 11:44:42 | INFO | train rank=0 | step=40850/50000 progress=81.7% loss=0.092594 lr=8.879729e-06 lr_backbone=8.879729e-06 elapsed=42:16:06 avg_step=3.72s recent_step=5.46s eta=09:28:03 +2026-09-19 11:45:10 | INFO | train rank=0 | step=40860/50000 progress=81.7% loss=0.091996 lr=8.860925e-06 lr_backbone=8.860925e-06 elapsed=42:16:33 avg_step=3.72s recent_step=2.72s eta=09:27:24 +2026-09-19 11:45:47 | INFO | train rank=0 | step=40870/50000 progress=81.7% loss=0.078211 lr=8.842138e-06 lr_backbone=8.842138e-06 elapsed=42:17:10 avg_step=3.72s recent_step=3.73s eta=09:26:46 +2026-09-19 11:46:16 | INFO | train rank=0 | step=40880/50000 progress=81.8% loss=0.095722 lr=8.823370e-06 lr_backbone=8.823370e-06 elapsed=42:17:40 avg_step=3.72s recent_step=2.95s eta=09:26:08 +2026-09-19 11:47:07 | INFO | train rank=0 | step=40890/50000 progress=81.8% loss=0.081553 lr=8.804620e-06 lr_backbone=8.804620e-06 elapsed=42:18:31 avg_step=3.72s recent_step=5.12s eta=09:25:33 +2026-09-19 11:47:44 | INFO | train rank=0 | step=40900/50000 progress=81.8% loss=0.091252 lr=8.785888e-06 lr_backbone=8.785888e-06 elapsed=42:19:07 avg_step=3.72s recent_step=3.63s eta=09:24:56 +2026-09-19 11:48:19 | INFO | train rank=0 | step=40910/50000 progress=81.8% loss=0.104396 lr=8.767173e-06 lr_backbone=8.767173e-06 elapsed=42:19:42 avg_step=3.72s recent_step=3.49s eta=09:24:18 +2026-09-19 11:48:53 | INFO | train rank=0 | step=40920/50000 progress=81.8% loss=0.074236 lr=8.748477e-06 lr_backbone=8.748477e-06 elapsed=42:20:17 avg_step=3.72s recent_step=3.46s eta=09:23:40 +2026-09-19 11:49:36 | INFO | train rank=0 | step=40930/50000 progress=81.9% loss=0.077337 lr=8.729799e-06 lr_backbone=8.729799e-06 elapsed=42:20:59 avg_step=3.72s recent_step=4.24s eta=09:23:04 +2026-09-19 11:50:20 | INFO | train rank=0 | step=40940/50000 progress=81.9% loss=0.079363 lr=8.711139e-06 lr_backbone=8.711139e-06 elapsed=42:21:43 avg_step=3.73s recent_step=4.39s eta=09:22:28 +2026-09-19 11:50:54 | INFO | train rank=0 | step=40950/50000 progress=81.9% loss=0.086845 lr=8.692497e-06 lr_backbone=8.692497e-06 elapsed=42:22:18 avg_step=3.72s recent_step=3.48s eta=09:21:51 +2026-09-19 11:51:24 | INFO | train rank=0 | step=40960/50000 progress=81.9% loss=0.078363 lr=8.673873e-06 lr_backbone=8.673873e-06 elapsed=42:22:47 avg_step=3.72s recent_step=2.98s eta=09:21:12 +2026-09-19 11:52:00 | INFO | train rank=0 | step=40970/50000 progress=81.9% loss=0.073362 lr=8.655267e-06 lr_backbone=8.655267e-06 elapsed=42:23:23 avg_step=3.72s recent_step=3.60s eta=09:20:34 +2026-09-19 11:52:46 | INFO | train rank=0 | step=40980/50000 progress=82.0% loss=0.094079 lr=8.636680e-06 lr_backbone=8.636680e-06 elapsed=42:24:09 avg_step=3.72s recent_step=4.57s eta=09:19:59 +2026-09-19 11:53:17 | INFO | train rank=0 | step=40990/50000 progress=82.0% loss=0.079454 lr=8.618110e-06 lr_backbone=8.618110e-06 elapsed=42:24:41 avg_step=3.72s recent_step=3.15s eta=09:19:20 +2026-09-19 11:53:52 | INFO | train rank=0 | step=41000/50000 progress=82.0% loss=0.078815 lr=8.599558e-06 lr_backbone=8.599558e-06 elapsed=42:25:15 avg_step=3.72s recent_step=3.48s eta=09:18:43 +2026-09-19 11:54:26 | INFO | train rank=0 | step=41010/50000 progress=82.0% loss=0.067928 lr=8.581025e-06 lr_backbone=8.581025e-06 elapsed=42:25:49 avg_step=3.72s recent_step=3.38s eta=09:18:05 +2026-09-19 11:55:21 | INFO | train rank=0 | step=41020/50000 progress=82.0% loss=0.075104 lr=8.562510e-06 lr_backbone=8.562510e-06 elapsed=42:26:45 avg_step=3.73s recent_step=5.52s eta=09:17:31 +2026-09-19 11:55:45 | INFO | train rank=0 | step=41030/50000 progress=82.1% loss=0.079853 lr=8.544012e-06 lr_backbone=8.544012e-06 elapsed=42:27:09 avg_step=3.72s recent_step=2.42s eta=09:16:51 +2026-09-19 11:56:19 | INFO | train rank=0 | step=41040/50000 progress=82.1% loss=0.087132 lr=8.525533e-06 lr_backbone=8.525533e-06 elapsed=42:27:42 avg_step=3.72s recent_step=3.35s eta=09:16:13 +2026-09-19 11:56:57 | INFO | train rank=0 | step=41050/50000 progress=82.1% loss=0.079942 lr=8.507072e-06 lr_backbone=8.507072e-06 elapsed=42:28:20 avg_step=3.72s recent_step=3.82s eta=09:15:36 +2026-09-19 11:57:49 | INFO | train rank=0 | step=41060/50000 progress=82.1% loss=0.095487 lr=8.488630e-06 lr_backbone=8.488630e-06 elapsed=42:29:12 avg_step=3.73s recent_step=5.15s eta=09:15:02 +2026-09-19 11:58:22 | INFO | train rank=0 | step=41070/50000 progress=82.1% loss=0.081893 lr=8.470205e-06 lr_backbone=8.470205e-06 elapsed=42:29:45 avg_step=3.72s recent_step=3.30s eta=09:14:24 +2026-09-19 11:58:53 | INFO | train rank=0 | step=41080/50000 progress=82.2% loss=0.073808 lr=8.451799e-06 lr_backbone=8.451799e-06 elapsed=42:30:17 avg_step=3.72s recent_step=3.19s eta=09:13:45 +2026-09-19 11:59:19 | INFO | train rank=0 | step=41090/50000 progress=82.2% loss=0.083331 lr=8.433410e-06 lr_backbone=8.433410e-06 elapsed=42:30:42 avg_step=3.72s recent_step=2.57s eta=09:13:05 +2026-09-19 11:59:56 | INFO | train rank=0 | step=41100/50000 progress=82.2% loss=0.092288 lr=8.415040e-06 lr_backbone=8.415040e-06 elapsed=42:31:19 avg_step=3.72s recent_step=3.65s eta=09:12:28 +2026-09-19 12:00:26 | INFO | train rank=0 | step=41110/50000 progress=82.2% loss=0.120847 lr=8.396688e-06 lr_backbone=8.396688e-06 elapsed=42:31:49 avg_step=3.72s recent_step=3.03s eta=09:11:49 +2026-09-19 12:01:07 | INFO | train rank=0 | step=41120/50000 progress=82.2% loss=0.078988 lr=8.378355e-06 lr_backbone=8.378355e-06 elapsed=42:32:30 avg_step=3.72s recent_step=4.10s eta=09:11:13 +2026-09-19 12:01:42 | INFO | train rank=0 | step=41130/50000 progress=82.3% loss=0.088185 lr=8.360039e-06 lr_backbone=8.360039e-06 elapsed=42:33:05 avg_step=3.72s recent_step=3.48s eta=09:10:35 +2026-09-19 12:02:16 | INFO | train rank=0 | step=41140/50000 progress=82.3% loss=0.091562 lr=8.341742e-06 lr_backbone=8.341742e-06 elapsed=42:33:39 avg_step=3.72s recent_step=3.43s eta=09:09:57 +2026-09-19 12:02:51 | INFO | train rank=0 | step=41150/50000 progress=82.3% loss=0.080480 lr=8.323463e-06 lr_backbone=8.323463e-06 elapsed=42:34:14 avg_step=3.72s recent_step=3.47s eta=09:09:19 +2026-09-19 12:03:30 | INFO | train rank=0 | step=41160/50000 progress=82.3% loss=0.081184 lr=8.305202e-06 lr_backbone=8.305202e-06 elapsed=42:34:53 avg_step=3.72s recent_step=3.90s eta=09:08:43 +2026-09-19 12:04:08 | INFO | train rank=0 | step=41170/50000 progress=82.3% loss=0.094790 lr=8.286959e-06 lr_backbone=8.286959e-06 elapsed=42:35:31 avg_step=3.72s recent_step=3.81s eta=09:08:06 +2026-09-19 12:04:44 | INFO | train rank=0 | step=41180/50000 progress=82.4% loss=0.081633 lr=8.268735e-06 lr_backbone=8.268735e-06 elapsed=42:36:08 avg_step=3.72s recent_step=3.63s eta=09:07:28 +2026-09-19 12:05:31 | INFO | train rank=0 | step=41190/50000 progress=82.4% loss=0.144844 lr=8.250529e-06 lr_backbone=8.250529e-06 elapsed=42:36:54 avg_step=3.72s recent_step=4.68s eta=09:06:53 +2026-09-19 12:06:05 | INFO | train rank=0 | step=41200/50000 progress=82.4% loss=0.103020 lr=8.232341e-06 lr_backbone=8.232341e-06 elapsed=42:37:28 avg_step=3.72s recent_step=3.36s eta=09:06:15 +2026-09-19 12:06:41 | INFO | train rank=0 | step=41210/50000 progress=82.4% loss=0.078039 lr=8.214171e-06 lr_backbone=8.214171e-06 elapsed=42:38:05 avg_step=3.72s recent_step=3.67s eta=09:05:38 +2026-09-19 12:07:12 | INFO | train rank=0 | step=41220/50000 progress=82.4% loss=0.147462 lr=8.196020e-06 lr_backbone=8.196020e-06 elapsed=42:38:35 avg_step=3.72s recent_step=3.03s eta=09:04:59 +2026-09-19 12:07:54 | INFO | train rank=0 | step=41230/50000 progress=82.5% loss=0.072282 lr=8.177887e-06 lr_backbone=8.177887e-06 elapsed=42:39:17 avg_step=3.72s recent_step=4.21s eta=09:04:23 +2026-09-19 12:08:33 | INFO | train rank=0 | step=41240/50000 progress=82.5% loss=0.164762 lr=8.159772e-06 lr_backbone=8.159772e-06 elapsed=42:39:56 avg_step=3.72s recent_step=3.91s eta=09:03:46 +2026-09-19 12:09:20 | INFO | train rank=0 | step=41250/50000 progress=82.5% loss=0.081092 lr=8.141676e-06 lr_backbone=8.141676e-06 elapsed=42:40:44 avg_step=3.72s recent_step=4.75s eta=09:03:11 +2026-09-19 12:09:55 | INFO | train rank=0 | step=41260/50000 progress=82.5% loss=0.104358 lr=8.123598e-06 lr_backbone=8.123598e-06 elapsed=42:41:18 avg_step=3.72s recent_step=3.45s eta=09:02:33 +2026-09-19 12:10:36 | INFO | train rank=0 | step=41270/50000 progress=82.5% loss=0.088194 lr=8.105538e-06 lr_backbone=8.105538e-06 elapsed=42:42:00 avg_step=3.72s recent_step=4.16s eta=09:01:57 +2026-09-19 12:11:11 | INFO | train rank=0 | step=41280/50000 progress=82.6% loss=0.096227 lr=8.087497e-06 lr_backbone=8.087497e-06 elapsed=42:42:34 avg_step=3.72s recent_step=3.46s eta=09:01:19 +2026-09-19 12:11:53 | INFO | train rank=0 | step=41290/50000 progress=82.6% loss=0.089545 lr=8.069474e-06 lr_backbone=8.069474e-06 elapsed=42:43:17 avg_step=3.72s recent_step=4.22s eta=09:00:43 +2026-09-19 12:12:26 | INFO | train rank=0 | step=41300/50000 progress=82.6% loss=0.078785 lr=8.051469e-06 lr_backbone=8.051469e-06 elapsed=42:43:49 avg_step=3.72s recent_step=3.25s eta=09:00:04 +2026-09-19 12:13:01 | INFO | train rank=0 | step=41310/50000 progress=82.6% loss=0.088354 lr=8.033482e-06 lr_backbone=8.033482e-06 elapsed=42:44:25 avg_step=3.72s recent_step=3.57s eta=08:59:27 +2026-09-19 12:13:50 | INFO | train rank=0 | step=41320/50000 progress=82.6% loss=0.125354 lr=8.015514e-06 lr_backbone=8.015514e-06 elapsed=42:45:14 avg_step=3.72s recent_step=4.90s eta=08:58:52 +2026-09-19 12:14:20 | INFO | train rank=0 | step=41330/50000 progress=82.7% loss=0.078635 lr=7.997565e-06 lr_backbone=7.997565e-06 elapsed=42:45:43 avg_step=3.72s recent_step=2.97s eta=08:58:13 +2026-09-19 12:14:52 | INFO | train rank=0 | step=41340/50000 progress=82.7% loss=0.096530 lr=7.979633e-06 lr_backbone=7.979633e-06 elapsed=42:46:15 avg_step=3.72s recent_step=3.18s eta=08:57:35 +2026-09-19 12:15:27 | INFO | train rank=0 | step=41350/50000 progress=82.7% loss=0.142232 lr=7.961720e-06 lr_backbone=7.961720e-06 elapsed=42:46:51 avg_step=3.72s recent_step=3.55s eta=08:56:57 +2026-09-19 12:16:19 | INFO | train rank=0 | step=41360/50000 progress=82.7% loss=0.086066 lr=7.943826e-06 lr_backbone=7.943826e-06 elapsed=42:47:43 avg_step=3.72s recent_step=5.18s eta=08:56:23 +2026-09-19 12:16:52 | INFO | train rank=0 | step=41370/50000 progress=82.7% loss=0.110465 lr=7.925950e-06 lr_backbone=7.925950e-06 elapsed=42:48:15 avg_step=3.72s recent_step=3.26s eta=08:55:45 +2026-09-19 12:17:27 | INFO | train rank=0 | step=41380/50000 progress=82.8% loss=0.086753 lr=7.908092e-06 lr_backbone=7.908092e-06 elapsed=42:48:50 avg_step=3.72s recent_step=3.53s eta=08:55:07 +2026-09-19 12:18:02 | INFO | train rank=0 | step=41390/50000 progress=82.8% loss=0.113642 lr=7.890253e-06 lr_backbone=7.890253e-06 elapsed=42:49:26 avg_step=3.72s recent_step=3.51s eta=08:54:29 +2026-09-19 12:18:48 | INFO | train rank=0 | step=41400/50000 progress=82.8% loss=0.085409 lr=7.872432e-06 lr_backbone=7.872432e-06 elapsed=42:50:11 avg_step=3.72s recent_step=4.55s eta=08:53:54 +2026-09-19 12:19:19 | INFO | train rank=0 | step=41410/50000 progress=82.8% loss=0.102574 lr=7.854629e-06 lr_backbone=7.854629e-06 elapsed=42:50:42 avg_step=3.72s recent_step=3.10s eta=08:53:15 +2026-09-19 12:19:53 | INFO | train rank=0 | step=41420/50000 progress=82.8% loss=0.110512 lr=7.836845e-06 lr_backbone=7.836845e-06 elapsed=42:51:16 avg_step=3.72s recent_step=3.43s eta=08:52:37 +2026-09-19 12:20:28 | INFO | train rank=0 | step=41430/50000 progress=82.9% loss=0.074322 lr=7.819080e-06 lr_backbone=7.819080e-06 elapsed=42:51:51 avg_step=3.72s recent_step=3.49s eta=08:52:00 +2026-09-19 12:21:04 | INFO | train rank=0 | step=41440/50000 progress=82.9% loss=0.076392 lr=7.801332e-06 lr_backbone=7.801332e-06 elapsed=42:52:27 avg_step=3.72s recent_step=3.58s eta=08:51:22 +2026-09-19 12:21:47 | INFO | train rank=0 | step=41450/50000 progress=82.9% loss=0.073354 lr=7.783604e-06 lr_backbone=7.783604e-06 elapsed=42:53:10 avg_step=3.72s recent_step=4.34s eta=08:50:46 +2026-09-19 12:22:25 | INFO | train rank=0 | step=41460/50000 progress=82.9% loss=0.098954 lr=7.765893e-06 lr_backbone=7.765893e-06 elapsed=42:53:48 avg_step=3.72s recent_step=3.75s eta=08:50:09 +2026-09-19 12:22:54 | INFO | train rank=0 | step=41470/50000 progress=82.9% loss=0.084030 lr=7.748202e-06 lr_backbone=7.748202e-06 elapsed=42:54:17 avg_step=3.72s recent_step=2.89s eta=08:49:30 +2026-09-19 12:23:37 | INFO | train rank=0 | step=41480/50000 progress=83.0% loss=0.077287 lr=7.730528e-06 lr_backbone=7.730528e-06 elapsed=42:55:00 avg_step=3.72s recent_step=4.31s eta=08:48:54 +2026-09-19 12:24:28 | INFO | train rank=0 | step=41490/50000 progress=83.0% loss=0.093919 lr=7.712874e-06 lr_backbone=7.712874e-06 elapsed=42:55:51 avg_step=3.73s recent_step=5.14s eta=08:48:20 +2026-09-19 12:24:58 | INFO | train rank=0 | step=41500/50000 progress=83.0% loss=0.086842 lr=7.695237e-06 lr_backbone=7.695237e-06 elapsed=42:56:21 avg_step=3.72s recent_step=2.97s eta=08:47:41 +2026-09-19 12:25:22 | INFO | train rank=0 | step=41510/50000 progress=83.0% loss=0.108186 lr=7.677620e-06 lr_backbone=7.677620e-06 elapsed=42:56:45 avg_step=3.72s recent_step=2.42s eta=08:47:01 +2026-09-19 12:26:02 | INFO | train rank=0 | step=41520/50000 progress=83.0% loss=0.079189 lr=7.660020e-06 lr_backbone=7.660020e-06 elapsed=42:57:25 avg_step=3.72s recent_step=4.00s eta=08:46:24 +2026-09-19 12:26:59 | INFO | train rank=0 | step=41530/50000 progress=83.1% loss=0.100658 lr=7.642440e-06 lr_backbone=7.642440e-06 elapsed=42:58:22 avg_step=3.73s recent_step=5.70s eta=08:45:51 +2026-09-19 12:27:32 | INFO | train rank=0 | step=41540/50000 progress=83.1% loss=0.104502 lr=7.624877e-06 lr_backbone=7.624877e-06 elapsed=42:58:55 avg_step=3.72s recent_step=3.28s eta=08:45:13 +2026-09-19 12:27:58 | INFO | train rank=0 | step=41550/50000 progress=83.1% loss=0.107045 lr=7.607334e-06 lr_backbone=7.607334e-06 elapsed=42:59:21 avg_step=3.72s recent_step=2.61s eta=08:44:33 +2026-09-19 12:28:25 | INFO | train rank=0 | step=41560/50000 progress=83.1% loss=0.073359 lr=7.589809e-06 lr_backbone=7.589809e-06 elapsed=42:59:48 avg_step=3.72s recent_step=2.68s eta=08:43:54 +2026-09-19 12:29:17 | INFO | train rank=0 | step=41570/50000 progress=83.1% loss=0.084685 lr=7.572302e-06 lr_backbone=7.572302e-06 elapsed=43:00:41 avg_step=3.72s recent_step=5.27s eta=08:43:20 +2026-09-19 12:29:51 | INFO | train rank=0 | step=41580/50000 progress=83.2% loss=0.085380 lr=7.554814e-06 lr_backbone=7.554814e-06 elapsed=43:01:15 avg_step=3.72s recent_step=3.39s eta=08:42:42 +2026-09-19 12:30:27 | INFO | train rank=0 | step=41590/50000 progress=83.2% loss=0.082846 lr=7.537344e-06 lr_backbone=7.537344e-06 elapsed=43:01:51 avg_step=3.72s recent_step=3.60s eta=08:42:04 +2026-09-19 12:31:01 | INFO | train rank=0 | step=41600/50000 progress=83.2% loss=0.103477 lr=7.519894e-06 lr_backbone=7.519894e-06 elapsed=43:02:24 avg_step=3.72s recent_step=3.36s eta=08:41:26 +2026-09-19 12:31:47 | INFO | train rank=0 | step=41610/50000 progress=83.2% loss=0.101512 lr=7.502461e-06 lr_backbone=7.502461e-06 elapsed=43:03:11 avg_step=3.72s recent_step=4.64s eta=08:40:51 +2026-09-19 12:32:24 | INFO | train rank=0 | step=41620/50000 progress=83.2% loss=0.077769 lr=7.485048e-06 lr_backbone=7.485048e-06 elapsed=43:03:47 avg_step=3.72s recent_step=3.64s eta=08:40:14 +2026-09-19 12:33:02 | INFO | train rank=0 | step=41630/50000 progress=83.3% loss=0.082213 lr=7.467652e-06 lr_backbone=7.467652e-06 elapsed=43:04:25 avg_step=3.72s recent_step=3.84s eta=08:39:37 +2026-09-19 12:33:24 | INFO | train rank=0 | step=41640/50000 progress=83.3% loss=0.085359 lr=7.450276e-06 lr_backbone=7.450276e-06 elapsed=43:04:47 avg_step=3.72s recent_step=2.20s eta=08:38:56 +2026-09-19 12:34:12 | INFO | train rank=0 | step=41650/50000 progress=83.3% loss=0.102481 lr=7.432918e-06 lr_backbone=7.432918e-06 elapsed=43:05:35 avg_step=3.72s recent_step=4.76s eta=08:38:21 +2026-09-19 12:34:46 | INFO | train rank=0 | step=41660/50000 progress=83.3% loss=0.075074 lr=7.415579e-06 lr_backbone=7.415579e-06 elapsed=43:06:10 avg_step=3.72s recent_step=3.49s eta=08:37:43 +2026-09-19 12:35:24 | INFO | train rank=0 | step=41670/50000 progress=83.3% loss=0.116879 lr=7.398258e-06 lr_backbone=7.398258e-06 elapsed=43:06:47 avg_step=3.72s recent_step=3.73s eta=08:37:06 +2026-09-19 12:36:02 | INFO | train rank=0 | step=41680/50000 progress=83.4% loss=0.073444 lr=7.380956e-06 lr_backbone=7.380956e-06 elapsed=43:07:26 avg_step=3.72s recent_step=3.85s eta=08:36:29 +2026-09-19 12:36:52 | INFO | train rank=0 | step=41690/50000 progress=83.4% loss=0.106964 lr=7.363673e-06 lr_backbone=7.363673e-06 elapsed=43:08:15 avg_step=3.73s recent_step=4.92s eta=08:35:54 +2026-09-19 12:37:26 | INFO | train rank=0 | step=41700/50000 progress=83.4% loss=0.102492 lr=7.346408e-06 lr_backbone=7.346408e-06 elapsed=43:08:49 avg_step=3.72s recent_step=3.45s eta=08:35:16 +2026-09-19 12:38:00 | INFO | train rank=0 | step=41710/50000 progress=83.4% loss=0.077771 lr=7.329162e-06 lr_backbone=7.329162e-06 elapsed=43:09:24 avg_step=3.72s recent_step=3.42s eta=08:34:39 +2026-09-19 12:38:32 | INFO | train rank=0 | step=41720/50000 progress=83.4% loss=0.063835 lr=7.311934e-06 lr_backbone=7.311934e-06 elapsed=43:09:56 avg_step=3.72s recent_step=3.21s eta=08:34:00 +2026-09-19 12:39:19 | INFO | train rank=0 | step=41730/50000 progress=83.5% loss=0.174369 lr=7.294726e-06 lr_backbone=7.294726e-06 elapsed=43:10:42 avg_step=3.72s recent_step=4.68s eta=08:33:25 +2026-09-19 12:39:49 | INFO | train rank=0 | step=41740/50000 progress=83.5% loss=0.102164 lr=7.277536e-06 lr_backbone=7.277536e-06 elapsed=43:11:12 avg_step=3.72s recent_step=2.99s eta=08:32:46 +2026-09-19 12:40:30 | INFO | train rank=0 | step=41750/50000 progress=83.5% loss=0.154737 lr=7.260364e-06 lr_backbone=7.260364e-06 elapsed=43:11:53 avg_step=3.72s recent_step=4.10s eta=08:32:10 +2026-09-19 12:41:04 | INFO | train rank=0 | step=41760/50000 progress=83.5% loss=0.080030 lr=7.243212e-06 lr_backbone=7.243212e-06 elapsed=43:12:27 avg_step=3.72s recent_step=3.36s eta=08:31:32 +2026-09-19 12:42:00 | INFO | train rank=0 | step=41770/50000 progress=83.5% loss=0.124445 lr=7.226078e-06 lr_backbone=7.226078e-06 elapsed=43:13:23 avg_step=3.73s recent_step=5.66s eta=08:30:58 +2026-09-19 12:42:32 | INFO | train rank=0 | step=41780/50000 progress=83.6% loss=0.089201 lr=7.208963e-06 lr_backbone=7.208963e-06 elapsed=43:13:55 avg_step=3.73s recent_step=3.14s eta=08:30:20 +2026-09-19 12:43:09 | INFO | train rank=0 | step=41790/50000 progress=83.6% loss=0.091074 lr=7.191866e-06 lr_backbone=7.191866e-06 elapsed=43:14:32 avg_step=3.73s recent_step=3.71s eta=08:29:43 +2026-09-19 12:43:34 | INFO | train rank=0 | step=41800/50000 progress=83.6% loss=0.101871 lr=7.174788e-06 lr_backbone=7.174788e-06 elapsed=43:14:57 avg_step=3.72s recent_step=2.49s eta=08:29:03 +2026-09-19 12:44:23 | INFO | train rank=0 | step=41810/50000 progress=83.6% loss=0.162807 lr=7.157729e-06 lr_backbone=7.157729e-06 elapsed=43:15:46 avg_step=3.73s recent_step=4.92s eta=08:28:28 +2026-09-19 12:44:53 | INFO | train rank=0 | step=41820/50000 progress=83.6% loss=0.103240 lr=7.140689e-06 lr_backbone=7.140689e-06 elapsed=43:16:17 avg_step=3.72s recent_step=3.07s eta=08:27:50 +2026-09-19 12:45:27 | INFO | train rank=0 | step=41830/50000 progress=83.7% loss=0.106900 lr=7.123667e-06 lr_backbone=7.123667e-06 elapsed=43:16:50 avg_step=3.72s recent_step=3.33s eta=08:27:12 +2026-09-19 12:45:57 | INFO | train rank=0 | step=41840/50000 progress=83.7% loss=0.085512 lr=7.106664e-06 lr_backbone=7.106664e-06 elapsed=43:17:20 avg_step=3.72s recent_step=3.01s eta=08:26:33 +2026-09-19 12:46:48 | INFO | train rank=0 | step=41850/50000 progress=83.7% loss=0.075726 lr=7.089680e-06 lr_backbone=7.089680e-06 elapsed=43:18:11 avg_step=3.73s recent_step=5.10s eta=08:25:58 +2026-09-19 12:47:22 | INFO | train rank=0 | step=41860/50000 progress=83.7% loss=0.081595 lr=7.072715e-06 lr_backbone=7.072715e-06 elapsed=43:18:45 avg_step=3.72s recent_step=3.39s eta=08:25:20 +2026-09-19 12:47:55 | INFO | train rank=0 | step=41870/50000 progress=83.7% loss=0.077203 lr=7.055768e-06 lr_backbone=7.055768e-06 elapsed=43:19:18 avg_step=3.72s recent_step=3.29s eta=08:24:42 +2026-09-19 12:48:27 | INFO | train rank=0 | step=41880/50000 progress=83.8% loss=0.090713 lr=7.038841e-06 lr_backbone=7.038841e-06 elapsed=43:19:51 avg_step=3.72s recent_step=3.28s eta=08:24:04 +2026-09-19 12:49:19 | INFO | train rank=0 | step=41890/50000 progress=83.8% loss=0.105569 lr=7.021932e-06 lr_backbone=7.021932e-06 elapsed=43:20:43 avg_step=3.73s recent_step=5.18s eta=08:23:30 +2026-09-19 12:49:47 | INFO | train rank=0 | step=41900/50000 progress=83.8% loss=0.106685 lr=7.005042e-06 lr_backbone=7.005042e-06 elapsed=43:21:10 avg_step=3.72s recent_step=2.75s eta=08:22:51 +2026-09-19 12:50:21 | INFO | train rank=0 | step=41910/50000 progress=83.8% loss=0.080242 lr=6.988170e-06 lr_backbone=6.988170e-06 elapsed=43:21:44 avg_step=3.72s recent_step=3.44s eta=08:22:13 +2026-09-19 12:50:53 | INFO | train rank=0 | step=41920/50000 progress=83.8% loss=0.066716 lr=6.971318e-06 lr_backbone=6.971318e-06 elapsed=43:22:16 avg_step=3.72s recent_step=3.14s eta=08:21:34 +2026-09-19 12:51:38 | INFO | train rank=0 | step=41930/50000 progress=83.9% loss=0.098284 lr=6.954484e-06 lr_backbone=6.954484e-06 elapsed=43:23:01 avg_step=3.72s recent_step=4.52s eta=08:20:59 +2026-09-19 12:52:13 | INFO | train rank=0 | step=41940/50000 progress=83.9% loss=0.065413 lr=6.937669e-06 lr_backbone=6.937669e-06 elapsed=43:23:36 avg_step=3.72s recent_step=3.48s eta=08:20:21 +2026-09-19 12:52:42 | INFO | train rank=0 | step=41950/50000 progress=83.9% loss=0.081765 lr=6.920873e-06 lr_backbone=6.920873e-06 elapsed=43:24:06 avg_step=3.72s recent_step=2.98s eta=08:19:42 +2026-09-19 12:53:22 | INFO | train rank=0 | step=41960/50000 progress=83.9% loss=0.117116 lr=6.904096e-06 lr_backbone=6.904096e-06 elapsed=43:24:45 avg_step=3.72s recent_step=3.96s eta=08:19:06 +2026-09-19 12:54:00 | INFO | train rank=0 | step=41970/50000 progress=83.9% loss=0.080415 lr=6.887338e-06 lr_backbone=6.887338e-06 elapsed=43:25:24 avg_step=3.72s recent_step=3.84s eta=08:18:29 +2026-09-19 12:54:34 | INFO | train rank=0 | step=41980/50000 progress=84.0% loss=0.087806 lr=6.870598e-06 lr_backbone=6.870598e-06 elapsed=43:25:57 avg_step=3.72s recent_step=3.35s eta=08:17:51 +2026-09-19 12:55:08 | INFO | train rank=0 | step=41990/50000 progress=84.0% loss=0.062789 lr=6.853878e-06 lr_backbone=6.853878e-06 elapsed=43:26:32 avg_step=3.72s recent_step=3.45s eta=08:17:13 +2026-09-19 12:55:49 | INFO | train rank=0 | step=42000/50000 progress=84.0% loss=0.123571 lr=6.837176e-06 lr_backbone=6.837176e-06 elapsed=43:27:12 avg_step=3.72s recent_step=4.08s eta=08:16:36 +2026-09-19 12:56:39 | INFO | train rank=0 | step=42010/50000 progress=84.0% loss=0.081352 lr=6.820493e-06 lr_backbone=6.820493e-06 elapsed=43:28:02 avg_step=3.72s recent_step=4.96s eta=08:16:01 +2026-09-19 12:57:08 | INFO | train rank=0 | step=42020/50000 progress=84.0% loss=0.098541 lr=6.803829e-06 lr_backbone=6.803829e-06 elapsed=43:28:31 avg_step=3.72s recent_step=2.93s eta=08:15:23 +2026-09-19 12:57:42 | INFO | train rank=0 | step=42030/50000 progress=84.1% loss=0.095880 lr=6.787184e-06 lr_backbone=6.787184e-06 elapsed=43:29:06 avg_step=3.72s recent_step=3.43s eta=08:14:45 +2026-09-19 12:58:28 | INFO | train rank=0 | step=42040/50000 progress=84.1% loss=0.088538 lr=6.770558e-06 lr_backbone=6.770558e-06 elapsed=43:29:51 avg_step=3.72s recent_step=4.55s eta=08:14:09 +2026-09-19 12:59:15 | INFO | train rank=0 | step=42050/50000 progress=84.1% loss=0.085090 lr=6.753951e-06 lr_backbone=6.753951e-06 elapsed=43:30:39 avg_step=3.73s recent_step=4.75s eta=08:13:34 +2026-09-19 12:59:48 | INFO | train rank=0 | step=42060/50000 progress=84.1% loss=0.078389 lr=6.737362e-06 lr_backbone=6.737362e-06 elapsed=43:31:11 avg_step=3.72s recent_step=3.22s eta=08:12:56 +2026-09-19 13:00:16 | INFO | train rank=0 | step=42070/50000 progress=84.1% loss=0.061495 lr=6.720793e-06 lr_backbone=6.720793e-06 elapsed=43:31:39 avg_step=3.72s recent_step=2.84s eta=08:12:17 +2026-09-19 13:00:48 | INFO | train rank=0 | step=42080/50000 progress=84.2% loss=0.100147 lr=6.704242e-06 lr_backbone=6.704242e-06 elapsed=43:32:11 avg_step=3.72s recent_step=3.20s eta=08:11:38 +2026-09-19 13:01:40 | INFO | train rank=0 | step=42090/50000 progress=84.2% loss=0.059579 lr=6.687711e-06 lr_backbone=6.687711e-06 elapsed=43:33:03 avg_step=3.72s recent_step=5.22s eta=08:11:04 +2026-09-19 13:02:12 | INFO | train rank=0 | step=42100/50000 progress=84.2% loss=0.111887 lr=6.671198e-06 lr_backbone=6.671198e-06 elapsed=43:33:36 avg_step=3.72s recent_step=3.23s eta=08:10:26 +2026-09-19 13:02:44 | INFO | train rank=0 | step=42110/50000 progress=84.2% loss=0.070789 lr=6.654705e-06 lr_backbone=6.654705e-06 elapsed=43:34:07 avg_step=3.72s recent_step=3.17s eta=08:09:48 +2026-09-19 13:03:20 | INFO | train rank=0 | step=42120/50000 progress=84.2% loss=0.080933 lr=6.638230e-06 lr_backbone=6.638230e-06 elapsed=43:34:43 avg_step=3.72s recent_step=3.53s eta=08:09:10 +2026-09-19 13:04:07 | INFO | train rank=0 | step=42130/50000 progress=84.3% loss=0.074146 lr=6.621774e-06 lr_backbone=6.621774e-06 elapsed=43:35:30 avg_step=3.72s recent_step=4.70s eta=08:08:35 +2026-09-19 13:04:41 | INFO | train rank=0 | step=42140/50000 progress=84.3% loss=0.084367 lr=6.605338e-06 lr_backbone=6.605338e-06 elapsed=43:36:04 avg_step=3.72s recent_step=3.42s eta=08:07:57 +2026-09-19 13:05:18 | INFO | train rank=0 | step=42150/50000 progress=84.3% loss=0.062588 lr=6.588920e-06 lr_backbone=6.588920e-06 elapsed=43:36:41 avg_step=3.72s recent_step=3.71s eta=08:07:19 +2026-09-19 13:05:47 | INFO | train rank=0 | step=42160/50000 progress=84.3% loss=0.058966 lr=6.572521e-06 lr_backbone=6.572521e-06 elapsed=43:37:10 avg_step=3.72s recent_step=2.93s eta=08:06:41 +2026-09-19 13:06:32 | INFO | train rank=0 | step=42170/50000 progress=84.3% loss=0.086552 lr=6.556141e-06 lr_backbone=6.556141e-06 elapsed=43:37:55 avg_step=3.72s recent_step=4.49s eta=08:06:05 +2026-09-19 13:07:00 | INFO | train rank=0 | step=42180/50000 progress=84.4% loss=0.069232 lr=6.539780e-06 lr_backbone=6.539780e-06 elapsed=43:38:23 avg_step=3.72s recent_step=2.76s eta=08:05:26 +2026-09-19 13:07:31 | INFO | train rank=0 | step=42190/50000 progress=84.4% loss=0.087148 lr=6.523439e-06 lr_backbone=6.523439e-06 elapsed=43:38:54 avg_step=3.72s recent_step=3.11s eta=08:04:47 +2026-09-19 13:08:19 | INFO | train rank=0 | step=42200/50000 progress=84.4% loss=0.055868 lr=6.507116e-06 lr_backbone=6.507116e-06 elapsed=43:39:43 avg_step=3.72s recent_step=4.85s eta=08:04:12 +2026-09-19 13:09:10 | INFO | train rank=0 | step=42210/50000 progress=84.4% loss=0.078190 lr=6.490812e-06 lr_backbone=6.490812e-06 elapsed=43:40:33 avg_step=3.73s recent_step=5.06s eta=08:03:38 +2026-09-19 13:09:39 | INFO | train rank=0 | step=42220/50000 progress=84.4% loss=0.110328 lr=6.474528e-06 lr_backbone=6.474528e-06 elapsed=43:41:02 avg_step=3.72s recent_step=2.89s eta=08:02:59 +2026-09-19 13:10:17 | INFO | train rank=0 | step=42230/50000 progress=84.5% loss=0.089594 lr=6.458262e-06 lr_backbone=6.458262e-06 elapsed=43:41:41 avg_step=3.72s recent_step=3.86s eta=08:02:22 +2026-09-19 13:10:47 | INFO | train rank=0 | step=42240/50000 progress=84.5% loss=0.070616 lr=6.442015e-06 lr_backbone=6.442015e-06 elapsed=43:42:10 avg_step=3.72s recent_step=2.98s eta=08:01:43 +2026-09-19 13:11:45 | INFO | train rank=0 | step=42250/50000 progress=84.5% loss=0.094580 lr=6.425788e-06 lr_backbone=6.425788e-06 elapsed=43:43:09 avg_step=3.73s recent_step=5.82s eta=08:01:10 +2026-09-19 13:12:13 | INFO | train rank=0 | step=42260/50000 progress=84.5% loss=0.103348 lr=6.409579e-06 lr_backbone=6.409579e-06 elapsed=43:43:36 avg_step=3.72s recent_step=2.72s eta=08:00:31 +2026-09-19 13:12:48 | INFO | train rank=0 | step=42270/50000 progress=84.5% loss=0.077727 lr=6.393390e-06 lr_backbone=6.393390e-06 elapsed=43:44:11 avg_step=3.72s recent_step=3.53s eta=07:59:53 +2026-09-19 13:13:21 | INFO | train rank=0 | step=42280/50000 progress=84.6% loss=0.080439 lr=6.377220e-06 lr_backbone=6.377220e-06 elapsed=43:44:45 avg_step=3.72s recent_step=3.35s eta=07:59:15 +2026-09-19 13:14:06 | INFO | train rank=0 | step=42290/50000 progress=84.6% loss=0.096934 lr=6.361068e-06 lr_backbone=6.361068e-06 elapsed=43:45:29 avg_step=3.72s recent_step=4.43s eta=07:58:39 +2026-09-19 13:14:36 | INFO | train rank=0 | step=42300/50000 progress=84.6% loss=0.072922 lr=6.344936e-06 lr_backbone=6.344936e-06 elapsed=43:45:59 avg_step=3.72s recent_step=3.01s eta=07:58:01 +2026-09-19 13:15:10 | INFO | train rank=0 | step=42310/50000 progress=84.6% loss=0.088008 lr=6.328823e-06 lr_backbone=6.328823e-06 elapsed=43:46:33 avg_step=3.72s recent_step=3.39s eta=07:57:23 +2026-09-19 13:15:45 | INFO | train rank=0 | step=42320/50000 progress=84.6% loss=0.099629 lr=6.312729e-06 lr_backbone=6.312729e-06 elapsed=43:47:08 avg_step=3.72s recent_step=3.54s eta=07:56:45 +2026-09-19 13:16:34 | INFO | train rank=0 | step=42330/50000 progress=84.7% loss=0.078582 lr=6.296654e-06 lr_backbone=6.296654e-06 elapsed=43:47:58 avg_step=3.72s recent_step=4.95s eta=07:56:10 +2026-09-19 13:17:05 | INFO | train rank=0 | step=42340/50000 progress=84.7% loss=0.081720 lr=6.280599e-06 lr_backbone=6.280599e-06 elapsed=43:48:29 avg_step=3.72s recent_step=3.09s eta=07:55:32 +2026-09-19 13:17:39 | INFO | train rank=0 | step=42350/50000 progress=84.7% loss=0.075159 lr=6.264562e-06 lr_backbone=6.264562e-06 elapsed=43:49:03 avg_step=3.72s recent_step=3.39s eta=07:54:54 +2026-09-19 13:18:14 | INFO | train rank=0 | step=42360/50000 progress=84.7% loss=0.087447 lr=6.248544e-06 lr_backbone=6.248544e-06 elapsed=43:49:37 avg_step=3.72s recent_step=3.45s eta=07:54:16 +2026-09-19 13:19:03 | INFO | train rank=0 | step=42370/50000 progress=84.7% loss=0.092824 lr=6.232546e-06 lr_backbone=6.232546e-06 elapsed=43:50:26 avg_step=3.72s recent_step=4.88s eta=07:53:41 +2026-09-19 13:19:33 | INFO | train rank=0 | step=42380/50000 progress=84.8% loss=0.062912 lr=6.216567e-06 lr_backbone=6.216567e-06 elapsed=43:50:56 avg_step=3.72s recent_step=3.03s eta=07:53:02 +2026-09-19 13:20:02 | INFO | train rank=0 | step=42390/50000 progress=84.8% loss=0.083632 lr=6.200607e-06 lr_backbone=6.200607e-06 elapsed=43:51:25 avg_step=3.72s recent_step=2.88s eta=07:52:24 +2026-09-19 13:20:32 | INFO | train rank=0 | step=42400/50000 progress=84.8% loss=0.090195 lr=6.184666e-06 lr_backbone=6.184666e-06 elapsed=43:51:55 avg_step=3.72s recent_step=3.03s eta=07:51:45 +2026-09-19 13:21:21 | INFO | train rank=0 | step=42410/50000 progress=84.8% loss=0.073415 lr=6.168744e-06 lr_backbone=6.168744e-06 elapsed=43:52:45 avg_step=3.72s recent_step=4.92s eta=07:51:10 +2026-09-19 13:21:47 | INFO | train rank=0 | step=42420/50000 progress=84.8% loss=0.098626 lr=6.152842e-06 lr_backbone=6.152842e-06 elapsed=43:53:10 avg_step=3.72s recent_step=2.56s eta=07:50:31 +2026-09-19 13:22:19 | INFO | train rank=0 | step=42430/50000 progress=84.9% loss=0.081701 lr=6.136958e-06 lr_backbone=6.136958e-06 elapsed=43:53:42 avg_step=3.72s recent_step=3.18s eta=07:49:53 +2026-09-19 13:22:55 | INFO | train rank=0 | step=42440/50000 progress=84.9% loss=0.100277 lr=6.121094e-06 lr_backbone=6.121094e-06 elapsed=43:54:18 avg_step=3.72s recent_step=3.63s eta=07:49:15 +2026-09-19 13:23:42 | INFO | train rank=0 | step=42450/50000 progress=84.9% loss=0.097417 lr=6.105249e-06 lr_backbone=6.105249e-06 elapsed=43:55:05 avg_step=3.72s recent_step=4.72s eta=07:48:40 +2026-09-19 13:24:09 | INFO | train rank=0 | step=42460/50000 progress=84.9% loss=0.080857 lr=6.089423e-06 lr_backbone=6.089423e-06 elapsed=43:55:33 avg_step=3.72s recent_step=2.72s eta=07:48:01 +2026-09-19 13:24:40 | INFO | train rank=0 | step=42470/50000 progress=84.9% loss=0.091227 lr=6.073617e-06 lr_backbone=6.073617e-06 elapsed=43:56:03 avg_step=3.72s recent_step=3.02s eta=07:47:22 +2026-09-19 13:25:31 | INFO | train rank=0 | step=42480/50000 progress=85.0% loss=0.092182 lr=6.057829e-06 lr_backbone=6.057829e-06 elapsed=43:56:55 avg_step=3.72s recent_step=5.18s eta=07:46:47 +2026-09-19 13:26:09 | INFO | train rank=0 | step=42490/50000 progress=85.0% loss=0.078782 lr=6.042061e-06 lr_backbone=6.042061e-06 elapsed=43:57:32 avg_step=3.72s recent_step=3.73s eta=07:46:10 +2026-09-19 13:26:41 | INFO | train rank=0 | step=42500/50000 progress=85.0% loss=0.087378 lr=6.026312e-06 lr_backbone=6.026312e-06 elapsed=43:58:05 avg_step=3.72s recent_step=3.28s eta=07:45:32 +2026-09-19 13:27:16 | INFO | train rank=0 | step=42510/50000 progress=85.0% loss=0.095273 lr=6.010583e-06 lr_backbone=6.010583e-06 elapsed=43:58:39 avg_step=3.72s recent_step=3.45s eta=07:44:54 +2026-09-19 13:28:11 | INFO | train rank=0 | step=42520/50000 progress=85.0% loss=0.075482 lr=5.994872e-06 lr_backbone=5.994872e-06 elapsed=43:59:35 avg_step=3.72s recent_step=5.53s eta=07:44:20 +2026-09-19 13:28:45 | INFO | train rank=0 | step=42530/50000 progress=85.1% loss=0.100790 lr=5.979181e-06 lr_backbone=5.979181e-06 elapsed=44:00:08 avg_step=3.72s recent_step=3.38s eta=07:43:43 +2026-09-19 13:29:15 | INFO | train rank=0 | step=42540/50000 progress=85.1% loss=0.074603 lr=5.963509e-06 lr_backbone=5.963509e-06 elapsed=44:00:39 avg_step=3.72s recent_step=3.03s eta=07:43:04 +2026-09-19 13:29:52 | INFO | train rank=0 | step=42550/50000 progress=85.1% loss=0.081932 lr=5.947857e-06 lr_backbone=5.947857e-06 elapsed=44:01:15 avg_step=3.72s recent_step=3.66s eta=07:42:27 +2026-09-19 13:30:49 | INFO | train rank=0 | step=42560/50000 progress=85.1% loss=0.081918 lr=5.932223e-06 lr_backbone=5.932223e-06 elapsed=44:02:12 avg_step=3.72s recent_step=5.70s eta=07:41:53 +2026-09-19 13:31:19 | INFO | train rank=0 | step=42570/50000 progress=85.1% loss=0.074770 lr=5.916609e-06 lr_backbone=5.916609e-06 elapsed=44:02:42 avg_step=3.72s recent_step=2.95s eta=07:41:14 +2026-09-19 13:31:57 | INFO | train rank=0 | step=42580/50000 progress=85.2% loss=0.076683 lr=5.901014e-06 lr_backbone=5.901014e-06 elapsed=44:03:20 avg_step=3.72s recent_step=3.83s eta=07:40:37 +2026-09-19 13:32:29 | INFO | train rank=0 | step=42590/50000 progress=85.2% loss=0.088756 lr=5.885439e-06 lr_backbone=5.885439e-06 elapsed=44:03:53 avg_step=3.72s recent_step=3.24s eta=07:39:59 +2026-09-19 13:33:17 | INFO | train rank=0 | step=42600/50000 progress=85.2% loss=0.078315 lr=5.869882e-06 lr_backbone=5.869882e-06 elapsed=44:04:40 avg_step=3.72s recent_step=4.72s eta=07:39:24 +2026-09-19 13:33:53 | INFO | train rank=0 | step=42610/50000 progress=85.2% loss=0.077722 lr=5.854345e-06 lr_backbone=5.854345e-06 elapsed=44:05:17 avg_step=3.72s recent_step=3.69s eta=07:38:46 +2026-09-19 13:34:28 | INFO | train rank=0 | step=42620/50000 progress=85.2% loss=0.097738 lr=5.838828e-06 lr_backbone=5.838828e-06 elapsed=44:05:51 avg_step=3.72s recent_step=3.42s eta=07:38:09 +2026-09-19 13:34:58 | INFO | train rank=0 | step=42630/50000 progress=85.3% loss=0.066452 lr=5.823330e-06 lr_backbone=5.823330e-06 elapsed=44:06:21 avg_step=3.72s recent_step=3.05s eta=07:37:30 +2026-09-19 13:35:49 | INFO | train rank=0 | step=42640/50000 progress=85.3% loss=0.112293 lr=5.807851e-06 lr_backbone=5.807851e-06 elapsed=44:07:12 avg_step=3.72s recent_step=5.08s eta=07:36:55 +2026-09-19 13:36:19 | INFO | train rank=0 | step=42650/50000 progress=85.3% loss=0.117962 lr=5.792391e-06 lr_backbone=5.792391e-06 elapsed=44:07:42 avg_step=3.72s recent_step=3.02s eta=07:36:17 +2026-09-19 13:36:58 | INFO | train rank=0 | step=42660/50000 progress=85.3% loss=0.099813 lr=5.776951e-06 lr_backbone=5.776951e-06 elapsed=44:08:21 avg_step=3.72s recent_step=3.88s eta=07:35:40 +2026-09-19 13:37:29 | INFO | train rank=0 | step=42670/50000 progress=85.3% loss=0.071541 lr=5.761530e-06 lr_backbone=5.761530e-06 elapsed=44:08:52 avg_step=3.72s recent_step=3.12s eta=07:35:02 +2026-09-19 13:38:18 | INFO | train rank=0 | step=42680/50000 progress=85.4% loss=0.082789 lr=5.746128e-06 lr_backbone=5.746128e-06 elapsed=44:09:42 avg_step=3.72s recent_step=4.93s eta=07:34:26 +2026-09-19 13:38:53 | INFO | train rank=0 | step=42690/50000 progress=85.4% loss=0.113084 lr=5.730746e-06 lr_backbone=5.730746e-06 elapsed=44:10:16 avg_step=3.72s recent_step=3.41s eta=07:33:49 +2026-09-19 13:39:33 | INFO | train rank=0 | step=42700/50000 progress=85.4% loss=0.080409 lr=5.715383e-06 lr_backbone=5.715383e-06 elapsed=44:10:56 avg_step=3.72s recent_step=4.03s eta=07:33:12 +2026-09-19 13:40:08 | INFO | train rank=0 | step=42710/50000 progress=85.4% loss=0.100263 lr=5.700039e-06 lr_backbone=5.700039e-06 elapsed=44:11:32 avg_step=3.72s recent_step=3.56s eta=07:32:34 +2026-09-19 13:40:53 | INFO | train rank=0 | step=42720/50000 progress=85.4% loss=0.165530 lr=5.684715e-06 lr_backbone=5.684715e-06 elapsed=44:12:17 avg_step=3.73s recent_step=4.49s eta=07:31:58 +2026-09-19 13:41:26 | INFO | train rank=0 | step=42730/50000 progress=85.5% loss=0.077834 lr=5.669410e-06 lr_backbone=5.669410e-06 elapsed=44:12:49 avg_step=3.73s recent_step=3.24s eta=07:31:20 +2026-09-19 13:41:59 | INFO | train rank=0 | step=42740/50000 progress=85.5% loss=0.070929 lr=5.654125e-06 lr_backbone=5.654125e-06 elapsed=44:13:22 avg_step=3.72s recent_step=3.32s eta=07:30:42 +2026-09-19 13:42:23 | INFO | train rank=0 | step=42750/50000 progress=85.5% loss=0.080835 lr=5.638859e-06 lr_backbone=5.638859e-06 elapsed=44:13:46 avg_step=3.72s recent_step=2.42s eta=07:30:03 +2026-09-19 13:43:14 | INFO | train rank=0 | step=42760/50000 progress=85.5% loss=0.093984 lr=5.623612e-06 lr_backbone=5.623612e-06 elapsed=44:14:37 avg_step=3.72s recent_step=5.10s eta=07:29:28 +2026-09-19 13:43:44 | INFO | train rank=0 | step=42770/50000 progress=85.5% loss=0.081207 lr=5.608385e-06 lr_backbone=5.608385e-06 elapsed=44:15:07 avg_step=3.72s recent_step=2.96s eta=07:28:49 +2026-09-19 13:44:14 | INFO | train rank=0 | step=42780/50000 progress=85.6% loss=0.107579 lr=5.593178e-06 lr_backbone=5.593178e-06 elapsed=44:15:38 avg_step=3.72s recent_step=3.05s eta=07:28:11 +2026-09-19 13:44:49 | INFO | train rank=0 | step=42790/50000 progress=85.6% loss=0.062846 lr=5.577989e-06 lr_backbone=5.577989e-06 elapsed=44:16:13 avg_step=3.72s recent_step=3.49s eta=07:27:33 +2026-09-19 13:45:41 | INFO | train rank=0 | step=42800/50000 progress=85.6% loss=0.100233 lr=5.562820e-06 lr_backbone=5.562820e-06 elapsed=44:17:05 avg_step=3.72s recent_step=5.22s eta=07:26:59 +2026-09-19 13:46:17 | INFO | train rank=0 | step=42810/50000 progress=85.6% loss=0.097935 lr=5.547671e-06 lr_backbone=5.547671e-06 elapsed=44:17:40 avg_step=3.72s recent_step=3.52s eta=07:26:21 +2026-09-19 13:46:52 | INFO | train rank=0 | step=42820/50000 progress=85.6% loss=0.079205 lr=5.532541e-06 lr_backbone=5.532541e-06 elapsed=44:18:15 avg_step=3.72s recent_step=3.50s eta=07:25:43 +2026-09-19 13:47:25 | INFO | train rank=0 | step=42830/50000 progress=85.7% loss=0.085259 lr=5.517430e-06 lr_backbone=5.517430e-06 elapsed=44:18:48 avg_step=3.72s recent_step=3.30s eta=07:25:06 +2026-09-19 13:48:20 | INFO | train rank=0 | step=42840/50000 progress=85.7% loss=0.067475 lr=5.502339e-06 lr_backbone=5.502339e-06 elapsed=44:19:43 avg_step=3.73s recent_step=5.55s eta=07:24:31 +2026-09-19 13:48:49 | INFO | train rank=0 | step=42850/50000 progress=85.7% loss=0.093173 lr=5.487268e-06 lr_backbone=5.487268e-06 elapsed=44:20:12 avg_step=3.72s recent_step=2.86s eta=07:23:53 +2026-09-19 13:49:13 | INFO | train rank=0 | step=42860/50000 progress=85.7% loss=0.065733 lr=5.472215e-06 lr_backbone=5.472215e-06 elapsed=44:20:37 avg_step=3.72s recent_step=2.47s eta=07:23:13 +2026-09-19 13:49:45 | INFO | train rank=0 | step=42870/50000 progress=85.7% loss=0.073664 lr=5.457183e-06 lr_backbone=5.457183e-06 elapsed=44:21:08 avg_step=3.72s recent_step=3.16s eta=07:22:35 +2026-09-19 13:50:39 | INFO | train rank=0 | step=42880/50000 progress=85.8% loss=0.108882 lr=5.442170e-06 lr_backbone=5.442170e-06 elapsed=44:22:03 avg_step=3.72s recent_step=5.44s eta=07:22:01 +2026-09-19 13:51:12 | INFO | train rank=0 | step=42890/50000 progress=85.8% loss=0.093999 lr=5.427176e-06 lr_backbone=5.427176e-06 elapsed=44:22:36 avg_step=3.72s recent_step=3.29s eta=07:21:23 +2026-09-19 13:51:40 | INFO | train rank=0 | step=42900/50000 progress=85.8% loss=0.064908 lr=5.412202e-06 lr_backbone=5.412202e-06 elapsed=44:23:03 avg_step=3.72s recent_step=2.77s eta=07:20:44 +2026-09-19 13:52:17 | INFO | train rank=0 | step=42910/50000 progress=85.8% loss=0.074759 lr=5.397247e-06 lr_backbone=5.397247e-06 elapsed=44:23:40 avg_step=3.72s recent_step=3.70s eta=07:20:07 +2026-09-19 13:53:10 | INFO | train rank=0 | step=42920/50000 progress=85.8% loss=0.079083 lr=5.382312e-06 lr_backbone=5.382312e-06 elapsed=44:24:34 avg_step=3.72s recent_step=5.35s eta=07:19:32 +2026-09-19 13:53:46 | INFO | train rank=0 | step=42930/50000 progress=85.9% loss=0.094843 lr=5.367396e-06 lr_backbone=5.367396e-06 elapsed=44:25:09 avg_step=3.72s recent_step=3.51s eta=07:18:54 +2026-09-19 13:54:12 | INFO | train rank=0 | step=42940/50000 progress=85.9% loss=0.094371 lr=5.352500e-06 lr_backbone=5.352500e-06 elapsed=44:25:35 avg_step=3.72s recent_step=2.61s eta=07:18:15 +2026-09-19 13:54:44 | INFO | train rank=0 | step=42950/50000 progress=85.9% loss=0.104088 lr=5.337623e-06 lr_backbone=5.337623e-06 elapsed=44:26:07 avg_step=3.72s recent_step=3.24s eta=07:17:37 +2026-09-19 13:55:34 | INFO | train rank=0 | step=42960/50000 progress=85.9% loss=0.103849 lr=5.322766e-06 lr_backbone=5.322766e-06 elapsed=44:26:58 avg_step=3.72s recent_step=5.03s eta=07:17:02 +2026-09-19 13:56:08 | INFO | train rank=0 | step=42970/50000 progress=85.9% loss=0.104803 lr=5.307929e-06 lr_backbone=5.307929e-06 elapsed=44:27:31 avg_step=3.72s recent_step=3.37s eta=07:16:24 +2026-09-19 13:56:45 | INFO | train rank=0 | step=42980/50000 progress=86.0% loss=0.083578 lr=5.293111e-06 lr_backbone=5.293111e-06 elapsed=44:28:09 avg_step=3.72s recent_step=3.73s eta=07:15:47 +2026-09-19 13:57:20 | INFO | train rank=0 | step=42990/50000 progress=86.0% loss=0.093754 lr=5.278312e-06 lr_backbone=5.278312e-06 elapsed=44:28:43 avg_step=3.72s recent_step=3.45s eta=07:15:09 +2026-09-19 13:58:10 | INFO | train rank=0 | step=43000/50000 progress=86.0% loss=0.141151 lr=5.263534e-06 lr_backbone=5.263534e-06 elapsed=44:29:34 avg_step=3.72s recent_step=5.05s eta=07:14:34 +2026-09-19 13:58:39 | INFO | train rank=0 | step=43010/50000 progress=86.0% loss=0.071536 lr=5.248774e-06 lr_backbone=5.248774e-06 elapsed=44:30:02 avg_step=3.72s recent_step=2.88s eta=07:13:56 +2026-09-19 13:59:14 | INFO | train rank=0 | step=43020/50000 progress=86.0% loss=0.125425 lr=5.234034e-06 lr_backbone=5.234034e-06 elapsed=44:30:37 avg_step=3.72s recent_step=3.49s eta=07:13:18 +2026-09-19 13:59:45 | INFO | train rank=0 | step=43030/50000 progress=86.1% loss=0.078008 lr=5.219314e-06 lr_backbone=5.219314e-06 elapsed=44:31:08 avg_step=3.72s recent_step=3.05s eta=07:12:40 +2026-09-19 14:00:41 | INFO | train rank=0 | step=43040/50000 progress=86.1% loss=0.074546 lr=5.204614e-06 lr_backbone=5.204614e-06 elapsed=44:32:04 avg_step=3.73s recent_step=5.63s eta=07:12:06 +2026-09-19 14:01:17 | INFO | train rank=0 | step=43050/50000 progress=86.1% loss=0.067699 lr=5.189933e-06 lr_backbone=5.189933e-06 elapsed=44:32:41 avg_step=3.73s recent_step=3.67s eta=07:11:28 +2026-09-19 14:01:50 | INFO | train rank=0 | step=43060/50000 progress=86.1% loss=0.088927 lr=5.175271e-06 lr_backbone=5.175271e-06 elapsed=44:33:14 avg_step=3.72s recent_step=3.29s eta=07:10:50 +2026-09-19 14:02:16 | INFO | train rank=0 | step=43070/50000 progress=86.1% loss=0.125422 lr=5.160630e-06 lr_backbone=5.160630e-06 elapsed=44:33:39 avg_step=3.72s recent_step=2.57s eta=07:10:11 +2026-09-19 14:03:15 | INFO | train rank=0 | step=43080/50000 progress=86.2% loss=0.115015 lr=5.146008e-06 lr_backbone=5.146008e-06 elapsed=44:34:38 avg_step=3.73s recent_step=5.86s eta=07:09:37 +2026-09-19 14:03:48 | INFO | train rank=0 | step=43090/50000 progress=86.2% loss=0.099521 lr=5.131405e-06 lr_backbone=5.131405e-06 elapsed=44:35:11 avg_step=3.73s recent_step=3.31s eta=07:08:59 +2026-09-19 14:04:24 | INFO | train rank=0 | step=43100/50000 progress=86.2% loss=0.106453 lr=5.116822e-06 lr_backbone=5.116822e-06 elapsed=44:35:47 avg_step=3.72s recent_step=3.58s eta=07:08:22 +2026-09-19 14:04:55 | INFO | train rank=0 | step=43110/50000 progress=86.2% loss=0.081325 lr=5.102259e-06 lr_backbone=5.102259e-06 elapsed=44:36:18 avg_step=3.72s recent_step=3.15s eta=07:07:44 +2026-09-19 14:05:44 | INFO | train rank=0 | step=43120/50000 progress=86.2% loss=0.082432 lr=5.087715e-06 lr_backbone=5.087715e-06 elapsed=44:37:07 avg_step=3.73s recent_step=4.86s eta=07:07:08 +2026-09-19 14:06:20 | INFO | train rank=0 | step=43130/50000 progress=86.3% loss=0.088890 lr=5.073191e-06 lr_backbone=5.073191e-06 elapsed=44:37:44 avg_step=3.73s recent_step=3.67s eta=07:06:31 +2026-09-19 14:06:56 | INFO | train rank=0 | step=43140/50000 progress=86.3% loss=0.102542 lr=5.058687e-06 lr_backbone=5.058687e-06 elapsed=44:38:20 avg_step=3.73s recent_step=3.60s eta=07:05:54 +2026-09-19 14:07:28 | INFO | train rank=0 | step=43150/50000 progress=86.3% loss=0.064047 lr=5.044202e-06 lr_backbone=5.044202e-06 elapsed=44:38:51 avg_step=3.72s recent_step=3.18s eta=07:05:15 +2026-09-19 14:08:23 | INFO | train rank=0 | step=43160/50000 progress=86.3% loss=0.063222 lr=5.029737e-06 lr_backbone=5.029737e-06 elapsed=44:39:46 avg_step=3.73s recent_step=5.46s eta=07:04:41 +2026-09-19 14:08:56 | INFO | train rank=0 | step=43170/50000 progress=86.3% loss=0.105201 lr=5.015292e-06 lr_backbone=5.015292e-06 elapsed=44:40:19 avg_step=3.73s recent_step=3.33s eta=07:04:03 +2026-09-19 14:09:27 | INFO | train rank=0 | step=43180/50000 progress=86.4% loss=0.094152 lr=5.000866e-06 lr_backbone=5.000866e-06 elapsed=44:40:51 avg_step=3.73s recent_step=3.15s eta=07:03:25 +2026-09-19 14:09:56 | INFO | train rank=0 | step=43190/50000 progress=86.4% loss=0.098536 lr=4.986461e-06 lr_backbone=4.986461e-06 elapsed=44:41:19 avg_step=3.72s recent_step=2.84s eta=07:02:46 +2026-09-19 14:10:44 | INFO | train rank=0 | step=43200/50000 progress=86.4% loss=0.139610 lr=4.972074e-06 lr_backbone=4.972074e-06 elapsed=44:42:07 avg_step=3.73s recent_step=4.81s eta=07:02:11 +2026-09-19 14:11:14 | INFO | train rank=0 | step=43210/50000 progress=86.4% loss=0.079489 lr=4.957708e-06 lr_backbone=4.957708e-06 elapsed=44:42:37 avg_step=3.73s recent_step=2.99s eta=07:01:32 +2026-09-19 14:11:47 | INFO | train rank=0 | step=43220/50000 progress=86.4% loss=0.127238 lr=4.943361e-06 lr_backbone=4.943361e-06 elapsed=44:43:10 avg_step=3.72s recent_step=3.28s eta=07:00:54 +2026-09-19 14:12:19 | INFO | train rank=0 | step=43230/50000 progress=86.5% loss=0.066035 lr=4.929034e-06 lr_backbone=4.929034e-06 elapsed=44:43:43 avg_step=3.72s recent_step=3.27s eta=07:00:16 +2026-09-19 14:13:18 | INFO | train rank=0 | step=43240/50000 progress=86.5% loss=0.106471 lr=4.914726e-06 lr_backbone=4.914726e-06 elapsed=44:44:42 avg_step=3.73s recent_step=5.89s eta=06:59:43 +2026-09-19 14:13:56 | INFO | train rank=0 | step=43250/50000 progress=86.5% loss=0.066944 lr=4.900438e-06 lr_backbone=4.900438e-06 elapsed=44:45:19 avg_step=3.73s recent_step=3.76s eta=06:59:05 +2026-09-19 14:14:19 | INFO | train rank=0 | step=43260/50000 progress=86.5% loss=0.101866 lr=4.886171e-06 lr_backbone=4.886171e-06 elapsed=44:45:42 avg_step=3.72s recent_step=2.32s eta=06:58:26 +2026-09-19 14:14:54 | INFO | train rank=0 | step=43270/50000 progress=86.5% loss=0.098550 lr=4.871922e-06 lr_backbone=4.871922e-06 elapsed=44:46:17 avg_step=3.72s recent_step=3.52s eta=06:57:48 +2026-09-19 14:15:44 | INFO | train rank=0 | step=43280/50000 progress=86.6% loss=0.072802 lr=4.857694e-06 lr_backbone=4.857694e-06 elapsed=44:47:07 avg_step=3.73s recent_step=4.95s eta=06:57:13 +2026-09-19 14:16:14 | INFO | train rank=0 | step=43290/50000 progress=86.6% loss=0.076812 lr=4.843485e-06 lr_backbone=4.843485e-06 elapsed=44:47:37 avg_step=3.73s recent_step=2.99s eta=06:56:35 +2026-09-19 14:16:42 | INFO | train rank=0 | step=43300/50000 progress=86.6% loss=0.089840 lr=4.829296e-06 lr_backbone=4.829296e-06 elapsed=44:48:05 avg_step=3.72s recent_step=2.86s eta=06:55:56 +2026-09-19 14:17:21 | INFO | train rank=0 | step=43310/50000 progress=86.6% loss=0.072856 lr=4.815127e-06 lr_backbone=4.815127e-06 elapsed=44:48:45 avg_step=3.72s recent_step=3.93s eta=06:55:19 +2026-09-19 14:18:15 | INFO | train rank=0 | step=43320/50000 progress=86.6% loss=0.074344 lr=4.800977e-06 lr_backbone=4.800977e-06 elapsed=44:49:38 avg_step=3.73s recent_step=5.35s eta=06:54:44 +2026-09-19 14:18:49 | INFO | train rank=0 | step=43330/50000 progress=86.7% loss=0.096852 lr=4.786847e-06 lr_backbone=4.786847e-06 elapsed=44:50:12 avg_step=3.73s recent_step=3.39s eta=06:54:07 +2026-09-19 14:19:17 | INFO | train rank=0 | step=43340/50000 progress=86.7% loss=0.083672 lr=4.772738e-06 lr_backbone=4.772738e-06 elapsed=44:50:40 avg_step=3.72s recent_step=2.77s eta=06:53:28 +2026-09-19 14:19:53 | INFO | train rank=0 | step=43350/50000 progress=86.7% loss=0.089355 lr=4.758647e-06 lr_backbone=4.758647e-06 elapsed=44:51:16 avg_step=3.72s recent_step=3.61s eta=06:52:50 +2026-09-19 14:20:40 | INFO | train rank=0 | step=43360/50000 progress=86.7% loss=0.094259 lr=4.744577e-06 lr_backbone=4.744577e-06 elapsed=44:52:04 avg_step=3.73s recent_step=4.75s eta=06:52:15 +2026-09-19 14:21:15 | INFO | train rank=0 | step=43370/50000 progress=86.7% loss=0.136589 lr=4.730526e-06 lr_backbone=4.730526e-06 elapsed=44:52:38 avg_step=3.73s recent_step=3.48s eta=06:51:37 +2026-09-19 14:21:51 | INFO | train rank=0 | step=43380/50000 progress=86.8% loss=0.090206 lr=4.716496e-06 lr_backbone=4.716496e-06 elapsed=44:53:14 avg_step=3.73s recent_step=3.58s eta=06:51:00 +2026-09-19 14:22:23 | INFO | train rank=0 | step=43390/50000 progress=86.8% loss=0.072215 lr=4.702485e-06 lr_backbone=4.702485e-06 elapsed=44:53:47 avg_step=3.72s recent_step=3.25s eta=06:50:22 +2026-09-19 14:23:03 | INFO | train rank=0 | step=43400/50000 progress=86.8% loss=0.089264 lr=4.688494e-06 lr_backbone=4.688494e-06 elapsed=44:54:26 avg_step=3.73s recent_step=3.98s eta=06:49:45 +2026-09-19 14:23:43 | INFO | train rank=0 | step=43410/50000 progress=86.8% loss=0.093999 lr=4.674522e-06 lr_backbone=4.674522e-06 elapsed=44:55:07 avg_step=3.73s recent_step=4.01s eta=06:49:08 +2026-09-19 14:24:17 | INFO | train rank=0 | step=43420/50000 progress=86.8% loss=0.073020 lr=4.660571e-06 lr_backbone=4.660571e-06 elapsed=44:55:41 avg_step=3.73s recent_step=3.41s eta=06:48:30 +2026-09-19 14:24:46 | INFO | train rank=0 | step=43430/50000 progress=86.9% loss=0.082631 lr=4.646639e-06 lr_backbone=4.646639e-06 elapsed=44:56:09 avg_step=3.72s recent_step=2.83s eta=06:47:52 +2026-09-19 14:25:39 | INFO | train rank=0 | step=43440/50000 progress=86.9% loss=0.088733 lr=4.632727e-06 lr_backbone=4.632727e-06 elapsed=44:57:02 avg_step=3.73s recent_step=5.34s eta=06:47:17 +2026-09-19 14:26:03 | INFO | train rank=0 | step=43450/50000 progress=86.9% loss=0.099591 lr=4.618835e-06 lr_backbone=4.618835e-06 elapsed=44:57:27 avg_step=3.72s recent_step=2.44s eta=06:46:38 +2026-09-19 14:26:28 | INFO | train rank=0 | step=43460/50000 progress=86.9% loss=0.078800 lr=4.604963e-06 lr_backbone=4.604963e-06 elapsed=44:57:52 avg_step=3.72s recent_step=2.51s eta=06:45:59 +2026-09-19 14:27:00 | INFO | train rank=0 | step=43470/50000 progress=86.9% loss=0.075473 lr=4.591111e-06 lr_backbone=4.591111e-06 elapsed=44:58:24 avg_step=3.72s recent_step=3.18s eta=06:45:20 +2026-09-19 14:28:05 | INFO | train rank=0 | step=43480/50000 progress=87.0% loss=0.095847 lr=4.577278e-06 lr_backbone=4.577278e-06 elapsed=44:59:28 avg_step=3.73s recent_step=6.46s eta=06:44:47 +2026-09-19 14:28:39 | INFO | train rank=0 | step=43490/50000 progress=87.0% loss=0.162052 lr=4.563466e-06 lr_backbone=4.563466e-06 elapsed=45:00:02 avg_step=3.73s recent_step=3.38s eta=06:44:10 +2026-09-19 14:29:11 | INFO | train rank=0 | step=43500/50000 progress=87.0% loss=0.081377 lr=4.549673e-06 lr_backbone=4.549673e-06 elapsed=45:00:35 avg_step=3.72s recent_step=3.28s eta=06:43:32 +2026-09-19 14:29:45 | INFO | train rank=0 | step=43510/50000 progress=87.0% loss=0.097003 lr=4.535900e-06 lr_backbone=4.535900e-06 elapsed=45:01:08 avg_step=3.72s recent_step=3.35s eta=06:42:54 +2026-09-19 14:30:34 | INFO | train rank=0 | step=43520/50000 progress=87.0% loss=0.089513 lr=4.522148e-06 lr_backbone=4.522148e-06 elapsed=45:01:57 avg_step=3.73s recent_step=4.88s eta=06:42:18 +2026-09-19 14:31:05 | INFO | train rank=0 | step=43530/50000 progress=87.1% loss=0.083456 lr=4.508415e-06 lr_backbone=4.508415e-06 elapsed=45:02:28 avg_step=3.72s recent_step=3.14s eta=06:41:40 +2026-09-19 14:31:30 | INFO | train rank=0 | step=43540/50000 progress=87.1% loss=0.076909 lr=4.494701e-06 lr_backbone=4.494701e-06 elapsed=45:02:54 avg_step=3.72s recent_step=2.53s eta=06:41:01 +2026-09-19 14:32:05 | INFO | train rank=0 | step=43550/50000 progress=87.1% loss=0.068257 lr=4.481008e-06 lr_backbone=4.481008e-06 elapsed=45:03:28 avg_step=3.72s recent_step=3.45s eta=06:40:24 +2026-09-19 14:33:04 | INFO | train rank=0 | step=43560/50000 progress=87.1% loss=0.071706 lr=4.467335e-06 lr_backbone=4.467335e-06 elapsed=45:04:27 avg_step=3.73s recent_step=5.92s eta=06:39:50 +2026-09-19 14:33:39 | INFO | train rank=0 | step=43570/50000 progress=87.1% loss=0.075021 lr=4.453682e-06 lr_backbone=4.453682e-06 elapsed=45:05:02 avg_step=3.73s recent_step=3.47s eta=06:39:12 +2026-09-19 14:34:13 | INFO | train rank=0 | step=43580/50000 progress=87.2% loss=0.103370 lr=4.440048e-06 lr_backbone=4.440048e-06 elapsed=45:05:36 avg_step=3.73s recent_step=3.41s eta=06:38:34 +2026-09-19 14:34:45 | INFO | train rank=0 | step=43590/50000 progress=87.2% loss=0.095116 lr=4.426435e-06 lr_backbone=4.426435e-06 elapsed=45:06:09 avg_step=3.72s recent_step=3.24s eta=06:37:56 +2026-09-19 14:35:36 | INFO | train rank=0 | step=43600/50000 progress=87.2% loss=0.071354 lr=4.412841e-06 lr_backbone=4.412841e-06 elapsed=45:07:00 avg_step=3.73s recent_step=5.11s eta=06:37:21 +2026-09-19 14:36:09 | INFO | train rank=0 | step=43610/50000 progress=87.2% loss=0.086651 lr=4.399267e-06 lr_backbone=4.399267e-06 elapsed=45:07:32 avg_step=3.73s recent_step=3.27s eta=06:36:43 +2026-09-19 14:36:43 | INFO | train rank=0 | step=43620/50000 progress=87.2% loss=0.087884 lr=4.385714e-06 lr_backbone=4.385714e-06 elapsed=45:08:06 avg_step=3.73s recent_step=3.37s eta=06:36:05 +2026-09-19 14:37:15 | INFO | train rank=0 | step=43630/50000 progress=87.3% loss=0.100763 lr=4.372180e-06 lr_backbone=4.372180e-06 elapsed=45:08:39 avg_step=3.72s recent_step=3.26s eta=06:35:27 +2026-09-19 14:38:06 | INFO | train rank=0 | step=43640/50000 progress=87.3% loss=0.072611 lr=4.358666e-06 lr_backbone=4.358666e-06 elapsed=45:09:30 avg_step=3.73s recent_step=5.10s eta=06:34:52 +2026-09-19 14:38:31 | INFO | train rank=0 | step=43650/50000 progress=87.3% loss=0.091816 lr=4.345172e-06 lr_backbone=4.345172e-06 elapsed=45:09:55 avg_step=3.72s recent_step=2.49s eta=06:34:13 +2026-09-19 14:39:08 | INFO | train rank=0 | step=43660/50000 progress=87.3% loss=0.120765 lr=4.331699e-06 lr_backbone=4.331699e-06 elapsed=45:10:31 avg_step=3.72s recent_step=3.67s eta=06:33:36 +2026-09-19 14:39:40 | INFO | train rank=0 | step=43670/50000 progress=87.3% loss=0.093977 lr=4.318245e-06 lr_backbone=4.318245e-06 elapsed=45:11:03 avg_step=3.72s recent_step=3.20s eta=06:32:58 +2026-09-19 14:40:34 | INFO | train rank=0 | step=43680/50000 progress=87.4% loss=0.084897 lr=4.304811e-06 lr_backbone=4.304811e-06 elapsed=45:11:58 avg_step=3.73s recent_step=5.43s eta=06:32:23 +2026-09-19 14:41:12 | INFO | train rank=0 | step=43690/50000 progress=87.4% loss=0.082947 lr=4.291397e-06 lr_backbone=4.291397e-06 elapsed=45:12:35 avg_step=3.73s recent_step=3.74s eta=06:31:46 +2026-09-19 14:41:47 | INFO | train rank=0 | step=43700/50000 progress=87.4% loss=0.105896 lr=4.278003e-06 lr_backbone=4.278003e-06 elapsed=45:13:10 avg_step=3.73s recent_step=3.48s eta=06:31:08 +2026-09-19 14:42:22 | INFO | train rank=0 | step=43710/50000 progress=87.4% loss=0.100856 lr=4.264629e-06 lr_backbone=4.264629e-06 elapsed=45:13:46 avg_step=3.73s recent_step=3.56s eta=06:30:31 +2026-09-19 14:43:09 | INFO | train rank=0 | step=43720/50000 progress=87.4% loss=0.085457 lr=4.251275e-06 lr_backbone=4.251275e-06 elapsed=45:14:32 avg_step=3.73s recent_step=4.63s eta=06:29:55 +2026-09-19 14:43:35 | INFO | train rank=0 | step=43730/50000 progress=87.5% loss=0.133844 lr=4.237941e-06 lr_backbone=4.237941e-06 elapsed=45:14:59 avg_step=3.73s recent_step=2.68s eta=06:29:16 +2026-09-19 14:44:03 | INFO | train rank=0 | step=43740/50000 progress=87.5% loss=0.087935 lr=4.224628e-06 lr_backbone=4.224628e-06 elapsed=45:15:26 avg_step=3.72s recent_step=2.74s eta=06:28:37 +2026-09-19 14:44:26 | INFO | train rank=0 | step=43750/50000 progress=87.5% loss=0.052609 lr=4.211334e-06 lr_backbone=4.211334e-06 elapsed=45:15:49 avg_step=3.72s recent_step=2.32s eta=06:27:58 +2026-09-19 14:45:23 | INFO | train rank=0 | step=43760/50000 progress=87.5% loss=0.061227 lr=4.198060e-06 lr_backbone=4.198060e-06 elapsed=45:16:46 avg_step=3.73s recent_step=5.68s eta=06:27:24 +2026-09-19 14:45:58 | INFO | train rank=0 | step=43770/50000 progress=87.5% loss=0.081902 lr=4.184806e-06 lr_backbone=4.184806e-06 elapsed=45:17:21 avg_step=3.72s recent_step=3.52s eta=06:26:46 +2026-09-19 14:46:30 | INFO | train rank=0 | step=43780/50000 progress=87.6% loss=0.134525 lr=4.171572e-06 lr_backbone=4.171572e-06 elapsed=45:17:53 avg_step=3.72s recent_step=3.20s eta=06:26:08 +2026-09-19 14:47:07 | INFO | train rank=0 | step=43790/50000 progress=87.6% loss=0.091023 lr=4.158359e-06 lr_backbone=4.158359e-06 elapsed=45:18:30 avg_step=3.72s recent_step=3.69s eta=06:25:31 +2026-09-19 14:48:06 | INFO | train rank=0 | step=43800/50000 progress=87.6% loss=0.090696 lr=4.145165e-06 lr_backbone=4.145165e-06 elapsed=45:19:29 avg_step=3.73s recent_step=5.93s eta=06:24:57 +2026-09-19 14:48:36 | INFO | train rank=0 | step=43810/50000 progress=87.6% loss=0.073621 lr=4.131991e-06 lr_backbone=4.131991e-06 elapsed=45:20:00 avg_step=3.73s recent_step=3.04s eta=06:24:18 +2026-09-19 14:49:17 | INFO | train rank=0 | step=43820/50000 progress=87.6% loss=0.104858 lr=4.118838e-06 lr_backbone=4.118838e-06 elapsed=45:20:40 avg_step=3.73s recent_step=4.05s eta=06:23:42 +2026-09-19 14:49:52 | INFO | train rank=0 | step=43830/50000 progress=87.7% loss=0.083667 lr=4.105705e-06 lr_backbone=4.105705e-06 elapsed=45:21:15 avg_step=3.73s recent_step=3.52s eta=06:23:04 +2026-09-19 14:50:38 | INFO | train rank=0 | step=43840/50000 progress=87.7% loss=0.089974 lr=4.092591e-06 lr_backbone=4.092591e-06 elapsed=45:22:01 avg_step=3.73s recent_step=4.60s eta=06:22:28 +2026-09-19 14:51:03 | INFO | train rank=0 | step=43850/50000 progress=87.7% loss=0.100671 lr=4.079498e-06 lr_backbone=4.079498e-06 elapsed=45:22:26 avg_step=3.73s recent_step=2.45s eta=06:21:49 +2026-09-19 14:51:53 | INFO | train rank=0 | step=43860/50000 progress=87.7% loss=0.072984 lr=4.066425e-06 lr_backbone=4.066425e-06 elapsed=45:23:16 avg_step=3.73s recent_step=5.03s eta=06:21:14 +2026-09-19 14:52:27 | INFO | train rank=0 | step=43870/50000 progress=87.7% loss=0.097947 lr=4.053372e-06 lr_backbone=4.053372e-06 elapsed=45:23:51 avg_step=3.73s recent_step=3.44s eta=06:20:36 +2026-09-19 14:53:03 | INFO | train rank=0 | step=43880/50000 progress=87.8% loss=0.079483 lr=4.040339e-06 lr_backbone=4.040339e-06 elapsed=45:24:26 avg_step=3.73s recent_step=3.56s eta=06:19:58 +2026-09-19 14:53:26 | INFO | train rank=0 | step=43890/50000 progress=87.8% loss=0.101495 lr=4.027326e-06 lr_backbone=4.027326e-06 elapsed=45:24:50 avg_step=3.72s recent_step=2.34s eta=06:19:19 +2026-09-19 14:54:27 | INFO | train rank=0 | step=43900/50000 progress=87.8% loss=0.101576 lr=4.014333e-06 lr_backbone=4.014333e-06 elapsed=45:25:50 avg_step=3.73s recent_step=6.04s eta=06:18:45 +2026-09-19 14:54:53 | INFO | train rank=0 | step=43910/50000 progress=87.8% loss=0.114457 lr=4.001360e-06 lr_backbone=4.001360e-06 elapsed=45:26:17 avg_step=3.73s recent_step=2.66s eta=06:18:06 +2026-09-19 14:55:23 | INFO | train rank=0 | step=43920/50000 progress=87.8% loss=0.105346 lr=3.988408e-06 lr_backbone=3.988408e-06 elapsed=45:26:46 avg_step=3.73s recent_step=2.92s eta=06:17:28 +2026-09-19 14:56:00 | INFO | train rank=0 | step=43930/50000 progress=87.9% loss=0.106059 lr=3.975475e-06 lr_backbone=3.975475e-06 elapsed=45:27:23 avg_step=3.73s recent_step=3.74s eta=06:16:51 +2026-09-19 14:56:53 | INFO | train rank=0 | step=43940/50000 progress=87.9% loss=0.077582 lr=3.962563e-06 lr_backbone=3.962563e-06 elapsed=45:28:17 avg_step=3.73s recent_step=5.35s eta=06:16:16 +2026-09-19 14:57:24 | INFO | train rank=0 | step=43950/50000 progress=87.9% loss=0.075572 lr=3.949671e-06 lr_backbone=3.949671e-06 elapsed=45:28:48 avg_step=3.73s recent_step=3.09s eta=06:15:38 +2026-09-19 14:58:00 | INFO | train rank=0 | step=43960/50000 progress=87.9% loss=0.080610 lr=3.936799e-06 lr_backbone=3.936799e-06 elapsed=45:29:23 avg_step=3.73s recent_step=3.53s eta=06:15:00 +2026-09-19 14:58:31 | INFO | train rank=0 | step=43970/50000 progress=87.9% loss=0.091733 lr=3.923947e-06 lr_backbone=3.923947e-06 elapsed=45:29:55 avg_step=3.73s recent_step=3.17s eta=06:14:22 +2026-09-19 14:59:26 | INFO | train rank=0 | step=43980/50000 progress=88.0% loss=0.099823 lr=3.911115e-06 lr_backbone=3.911115e-06 elapsed=45:30:49 avg_step=3.73s recent_step=5.46s eta=06:13:47 +2026-09-19 14:59:58 | INFO | train rank=0 | step=43990/50000 progress=88.0% loss=0.091430 lr=3.898304e-06 lr_backbone=3.898304e-06 elapsed=45:31:21 avg_step=3.73s recent_step=3.19s eta=06:13:09 +2026-09-19 15:00:30 | INFO | train rank=0 | step=44000/50000 progress=88.0% loss=0.076837 lr=3.885512e-06 lr_backbone=3.885512e-06 elapsed=45:31:54 avg_step=3.73s recent_step=3.25s eta=06:12:31 +2026-09-19 15:01:05 | INFO | train rank=0 | step=44010/50000 progress=88.0% loss=0.078361 lr=3.872741e-06 lr_backbone=3.872741e-06 elapsed=45:32:28 avg_step=3.73s recent_step=3.48s eta=06:11:54 +2026-09-19 15:01:50 | INFO | train rank=0 | step=44020/50000 progress=88.0% loss=0.088094 lr=3.859990e-06 lr_backbone=3.859990e-06 elapsed=45:33:13 avg_step=3.73s recent_step=4.48s eta=06:11:18 +2026-09-19 15:02:15 | INFO | train rank=0 | step=44030/50000 progress=88.1% loss=0.078371 lr=3.847259e-06 lr_backbone=3.847259e-06 elapsed=45:33:38 avg_step=3.73s recent_step=2.50s eta=06:10:39 +2026-09-19 15:02:49 | INFO | train rank=0 | step=44040/50000 progress=88.1% loss=0.111515 lr=3.834549e-06 lr_backbone=3.834549e-06 elapsed=45:34:13 avg_step=3.73s recent_step=3.43s eta=06:10:01 +2026-09-19 15:03:23 | INFO | train rank=0 | step=44050/50000 progress=88.1% loss=0.095324 lr=3.821858e-06 lr_backbone=3.821858e-06 elapsed=45:34:46 avg_step=3.73s recent_step=3.32s eta=06:09:23 +2026-09-19 15:04:07 | INFO | train rank=0 | step=44060/50000 progress=88.1% loss=0.097333 lr=3.809188e-06 lr_backbone=3.809188e-06 elapsed=45:35:30 avg_step=3.73s recent_step=4.42s eta=06:08:47 +2026-09-19 15:04:31 | INFO | train rank=0 | step=44070/50000 progress=88.1% loss=0.077863 lr=3.796538e-06 lr_backbone=3.796538e-06 elapsed=45:35:54 avg_step=3.72s recent_step=2.41s eta=06:08:08 +2026-09-19 15:05:14 | INFO | train rank=0 | step=44080/50000 progress=88.2% loss=0.085499 lr=3.783908e-06 lr_backbone=3.783908e-06 elapsed=45:36:37 avg_step=3.72s recent_step=4.28s eta=06:07:31 +2026-09-19 15:05:45 | INFO | train rank=0 | step=44090/50000 progress=88.2% loss=0.077098 lr=3.771298e-06 lr_backbone=3.771298e-06 elapsed=45:37:08 avg_step=3.72s recent_step=3.09s eta=06:06:53 +2026-09-19 15:06:29 | INFO | train rank=0 | step=44100/50000 progress=88.2% loss=0.082474 lr=3.758709e-06 lr_backbone=3.758709e-06 elapsed=45:37:53 avg_step=3.73s recent_step=4.48s eta=06:06:17 +2026-09-19 15:07:11 | INFO | train rank=0 | step=44110/50000 progress=88.2% loss=0.071746 lr=3.746140e-06 lr_backbone=3.746140e-06 elapsed=45:38:34 avg_step=3.73s recent_step=4.15s eta=06:05:40 +2026-09-19 15:07:51 | INFO | train rank=0 | step=44120/50000 progress=88.2% loss=0.065230 lr=3.733591e-06 lr_backbone=3.733591e-06 elapsed=45:39:14 avg_step=3.73s recent_step=3.99s eta=06:05:04 +2026-09-19 15:08:20 | INFO | train rank=0 | step=44130/50000 progress=88.3% loss=0.084478 lr=3.721062e-06 lr_backbone=3.721062e-06 elapsed=45:39:43 avg_step=3.72s recent_step=2.90s eta=06:04:25 +2026-09-19 15:09:12 | INFO | train rank=0 | step=44140/50000 progress=88.3% loss=0.092082 lr=3.708554e-06 lr_backbone=3.708554e-06 elapsed=45:40:35 avg_step=3.73s recent_step=5.23s eta=06:03:50 +2026-09-19 15:09:36 | INFO | train rank=0 | step=44150/50000 progress=88.3% loss=0.077969 lr=3.696065e-06 lr_backbone=3.696065e-06 elapsed=45:41:00 avg_step=3.73s recent_step=2.41s eta=06:03:11 +2026-09-19 15:10:19 | INFO | train rank=0 | step=44160/50000 progress=88.3% loss=0.069332 lr=3.683597e-06 lr_backbone=3.683597e-06 elapsed=45:41:43 avg_step=3.73s recent_step=4.31s eta=06:02:34 +2026-09-19 15:10:51 | INFO | train rank=0 | step=44170/50000 progress=88.3% loss=0.084165 lr=3.671150e-06 lr_backbone=3.671150e-06 elapsed=45:42:14 avg_step=3.73s recent_step=3.18s eta=06:01:56 +2026-09-19 15:11:27 | INFO | train rank=0 | step=44180/50000 progress=88.4% loss=0.076077 lr=3.658722e-06 lr_backbone=3.658722e-06 elapsed=45:42:50 avg_step=3.73s recent_step=3.60s eta=06:01:19 +2026-09-19 15:11:57 | INFO | train rank=0 | step=44190/50000 progress=88.4% loss=0.078761 lr=3.646315e-06 lr_backbone=3.646315e-06 elapsed=45:43:21 avg_step=3.72s recent_step=3.01s eta=06:00:41 +2026-09-19 15:12:45 | INFO | train rank=0 | step=44200/50000 progress=88.4% loss=0.076442 lr=3.633928e-06 lr_backbone=3.633928e-06 elapsed=45:44:08 avg_step=3.73s recent_step=4.79s eta=06:00:05 +2026-09-19 15:13:26 | INFO | train rank=0 | step=44210/50000 progress=88.4% loss=0.085732 lr=3.621562e-06 lr_backbone=3.621562e-06 elapsed=45:44:49 avg_step=3.73s recent_step=4.09s eta=05:59:28 +2026-09-19 15:14:04 | INFO | train rank=0 | step=44220/50000 progress=88.4% loss=0.087757 lr=3.609215e-06 lr_backbone=3.609215e-06 elapsed=45:45:27 avg_step=3.73s recent_step=3.82s eta=05:58:51 +2026-09-19 15:14:41 | INFO | train rank=0 | step=44230/50000 progress=88.5% loss=0.088865 lr=3.596889e-06 lr_backbone=3.596889e-06 elapsed=45:46:04 avg_step=3.73s recent_step=3.67s eta=05:58:14 +2026-09-19 15:15:31 | INFO | train rank=0 | step=44240/50000 progress=88.5% loss=0.061201 lr=3.584583e-06 lr_backbone=3.584583e-06 elapsed=45:46:54 avg_step=3.73s recent_step=5.00s eta=05:57:38 +2026-09-19 15:16:05 | INFO | train rank=0 | step=44250/50000 progress=88.5% loss=0.076728 lr=3.572298e-06 lr_backbone=3.572298e-06 elapsed=45:47:28 avg_step=3.73s recent_step=3.37s eta=05:57:00 +2026-09-19 15:16:34 | INFO | train rank=0 | step=44260/50000 progress=88.5% loss=0.080808 lr=3.560033e-06 lr_backbone=3.560033e-06 elapsed=45:47:58 avg_step=3.73s recent_step=2.96s eta=05:56:22 +2026-09-19 15:17:07 | INFO | train rank=0 | step=44270/50000 progress=88.5% loss=0.083721 lr=3.547788e-06 lr_backbone=3.547788e-06 elapsed=45:48:31 avg_step=3.73s recent_step=3.31s eta=05:55:44 +2026-09-19 15:18:00 | INFO | train rank=0 | step=44280/50000 progress=88.6% loss=0.097590 lr=3.535564e-06 lr_backbone=3.535564e-06 elapsed=45:49:24 avg_step=3.73s recent_step=5.31s eta=05:55:09 +2026-09-19 15:18:32 | INFO | train rank=0 | step=44290/50000 progress=88.6% loss=0.063854 lr=3.523360e-06 lr_backbone=3.523360e-06 elapsed=45:49:55 avg_step=3.73s recent_step=3.12s eta=05:54:31 +2026-09-19 15:19:05 | INFO | train rank=0 | step=44300/50000 progress=88.6% loss=0.077348 lr=3.511176e-06 lr_backbone=3.511176e-06 elapsed=45:50:28 avg_step=3.73s recent_step=3.29s eta=05:53:53 +2026-09-19 15:19:44 | INFO | train rank=0 | step=44310/50000 progress=88.6% loss=0.105105 lr=3.499012e-06 lr_backbone=3.499012e-06 elapsed=45:51:07 avg_step=3.73s recent_step=3.91s eta=05:53:16 +2026-09-19 15:20:29 | INFO | train rank=0 | step=44320/50000 progress=88.6% loss=0.079723 lr=3.486869e-06 lr_backbone=3.486869e-06 elapsed=45:51:52 avg_step=3.73s recent_step=4.53s eta=05:52:40 +2026-09-19 15:21:00 | INFO | train rank=0 | step=44330/50000 progress=88.7% loss=0.085225 lr=3.474746e-06 lr_backbone=3.474746e-06 elapsed=45:52:23 avg_step=3.73s recent_step=3.08s eta=05:52:02 +2026-09-19 15:21:31 | INFO | train rank=0 | step=44340/50000 progress=88.7% loss=0.095202 lr=3.462644e-06 lr_backbone=3.462644e-06 elapsed=45:52:54 avg_step=3.73s recent_step=3.13s eta=05:51:24 +2026-09-19 15:22:07 | INFO | train rank=0 | step=44350/50000 progress=88.7% loss=0.063441 lr=3.450562e-06 lr_backbone=3.450562e-06 elapsed=45:53:30 avg_step=3.73s recent_step=3.54s eta=05:50:47 +2026-09-19 15:22:46 | INFO | train rank=0 | step=44360/50000 progress=88.7% loss=0.158423 lr=3.438500e-06 lr_backbone=3.438500e-06 elapsed=45:54:10 avg_step=3.73s recent_step=3.97s eta=05:50:10 +2026-09-19 15:23:18 | INFO | train rank=0 | step=44370/50000 progress=88.7% loss=0.091707 lr=3.426459e-06 lr_backbone=3.426459e-06 elapsed=45:54:42 avg_step=3.73s recent_step=3.20s eta=05:49:32 +2026-09-19 15:24:00 | INFO | train rank=0 | step=44380/50000 progress=88.8% loss=0.081796 lr=3.414438e-06 lr_backbone=3.414438e-06 elapsed=45:55:23 avg_step=3.73s recent_step=4.16s eta=05:48:55 +2026-09-19 15:24:36 | INFO | train rank=0 | step=44390/50000 progress=88.8% loss=0.073333 lr=3.402437e-06 lr_backbone=3.402437e-06 elapsed=45:55:59 avg_step=3.73s recent_step=3.62s eta=05:48:18 +2026-09-19 15:25:26 | INFO | train rank=0 | step=44400/50000 progress=88.8% loss=0.105425 lr=3.390457e-06 lr_backbone=3.390457e-06 elapsed=45:56:49 avg_step=3.73s recent_step=5.00s eta=05:47:42 +2026-09-19 15:26:01 | INFO | train rank=0 | step=44410/50000 progress=88.8% loss=0.083018 lr=3.378497e-06 lr_backbone=3.378497e-06 elapsed=45:57:24 avg_step=3.73s recent_step=3.46s eta=05:47:04 +2026-09-19 15:26:36 | INFO | train rank=0 | step=44420/50000 progress=88.8% loss=0.081308 lr=3.366558e-06 lr_backbone=3.366558e-06 elapsed=45:58:00 avg_step=3.73s recent_step=3.57s eta=05:46:27 +2026-09-19 15:27:10 | INFO | train rank=0 | step=44430/50000 progress=88.9% loss=0.071970 lr=3.354638e-06 lr_backbone=3.354638e-06 elapsed=45:58:33 avg_step=3.73s recent_step=3.35s eta=05:45:49 +2026-09-19 15:28:00 | INFO | train rank=0 | step=44440/50000 progress=88.9% loss=0.108983 lr=3.342740e-06 lr_backbone=3.342740e-06 elapsed=45:59:23 avg_step=3.73s recent_step=5.00s eta=05:45:14 +2026-09-19 15:28:32 | INFO | train rank=0 | step=44450/50000 progress=88.9% loss=0.091587 lr=3.330862e-06 lr_backbone=3.330862e-06 elapsed=45:59:56 avg_step=3.73s recent_step=3.25s eta=05:44:36 +2026-09-19 15:29:00 | INFO | train rank=0 | step=44460/50000 progress=88.9% loss=0.077215 lr=3.319004e-06 lr_backbone=3.319004e-06 elapsed=46:00:24 avg_step=3.73s recent_step=2.80s eta=05:43:57 +2026-09-19 15:29:33 | INFO | train rank=0 | step=44470/50000 progress=88.9% loss=0.115082 lr=3.307166e-06 lr_backbone=3.307166e-06 elapsed=46:00:56 avg_step=3.73s recent_step=3.25s eta=05:43:20 +2026-09-19 15:30:19 | INFO | train rank=0 | step=44480/50000 progress=89.0% loss=0.094150 lr=3.295349e-06 lr_backbone=3.295349e-06 elapsed=46:01:43 avg_step=3.73s recent_step=4.63s eta=05:42:43 +2026-09-19 15:30:47 | INFO | train rank=0 | step=44490/50000 progress=89.0% loss=0.104147 lr=3.283553e-06 lr_backbone=3.283553e-06 elapsed=46:02:10 avg_step=3.73s recent_step=2.75s eta=05:42:05 +2026-09-19 15:31:26 | INFO | train rank=0 | step=44500/50000 progress=89.0% loss=0.084018 lr=3.271777e-06 lr_backbone=3.271777e-06 elapsed=46:02:49 avg_step=3.73s recent_step=3.94s eta=05:41:28 +2026-09-19 15:32:00 | INFO | train rank=0 | step=44510/50000 progress=89.0% loss=0.114521 lr=3.260021e-06 lr_backbone=3.260021e-06 elapsed=46:03:23 avg_step=3.73s recent_step=3.34s eta=05:40:50 +2026-09-19 15:32:59 | INFO | train rank=0 | step=44520/50000 progress=89.0% loss=0.091705 lr=3.248286e-06 lr_backbone=3.248286e-06 elapsed=46:04:22 avg_step=3.73s recent_step=5.92s eta=05:40:16 +2026-09-19 15:33:28 | INFO | train rank=0 | step=44530/50000 progress=89.1% loss=0.083463 lr=3.236571e-06 lr_backbone=3.236571e-06 elapsed=46:04:51 avg_step=3.73s recent_step=2.87s eta=05:39:37 +2026-09-19 15:33:58 | INFO | train rank=0 | step=44540/50000 progress=89.1% loss=0.083368 lr=3.224877e-06 lr_backbone=3.224877e-06 elapsed=46:05:21 avg_step=3.73s recent_step=3.00s eta=05:38:59 +2026-09-19 15:34:28 | INFO | train rank=0 | step=44550/50000 progress=89.1% loss=0.082006 lr=3.213203e-06 lr_backbone=3.213203e-06 elapsed=46:05:51 avg_step=3.73s recent_step=3.01s eta=05:38:21 +2026-09-19 15:35:15 | INFO | train rank=0 | step=44560/50000 progress=89.1% loss=0.068592 lr=3.201550e-06 lr_backbone=3.201550e-06 elapsed=46:06:38 avg_step=3.73s recent_step=4.74s eta=05:37:45 +2026-09-19 15:35:41 | INFO | train rank=0 | step=44570/50000 progress=89.1% loss=0.069523 lr=3.189917e-06 lr_backbone=3.189917e-06 elapsed=46:07:05 avg_step=3.73s recent_step=2.63s eta=05:37:06 +2026-09-19 15:36:34 | INFO | train rank=0 | step=44580/50000 progress=89.2% loss=0.073666 lr=3.178304e-06 lr_backbone=3.178304e-06 elapsed=46:07:57 avg_step=3.73s recent_step=5.24s eta=05:36:31 +2026-09-19 15:37:03 | INFO | train rank=0 | step=44590/50000 progress=89.2% loss=0.081957 lr=3.166712e-06 lr_backbone=3.166712e-06 elapsed=46:08:27 avg_step=3.73s recent_step=2.97s eta=05:35:53 +2026-09-19 15:37:50 | INFO | train rank=0 | step=44600/50000 progress=89.2% loss=0.091426 lr=3.155141e-06 lr_backbone=3.155141e-06 elapsed=46:09:13 avg_step=3.73s recent_step=4.66s eta=05:35:17 +2026-09-19 15:38:21 | INFO | train rank=0 | step=44610/50000 progress=89.2% loss=0.079716 lr=3.143590e-06 lr_backbone=3.143590e-06 elapsed=46:09:45 avg_step=3.73s recent_step=3.13s eta=05:34:39 +2026-09-19 15:38:56 | INFO | train rank=0 | step=44620/50000 progress=89.2% loss=0.065599 lr=3.132059e-06 lr_backbone=3.132059e-06 elapsed=46:10:19 avg_step=3.73s recent_step=3.46s eta=05:34:01 +2026-09-19 15:39:24 | INFO | train rank=0 | step=44630/50000 progress=89.3% loss=0.087948 lr=3.120549e-06 lr_backbone=3.120549e-06 elapsed=46:10:48 avg_step=3.73s recent_step=2.84s eta=05:33:23 +2026-09-19 15:40:11 | INFO | train rank=0 | step=44640/50000 progress=89.3% loss=0.110915 lr=3.109060e-06 lr_backbone=3.109060e-06 elapsed=46:11:34 avg_step=3.73s recent_step=4.63s eta=05:32:47 +2026-09-19 15:40:38 | INFO | train rank=0 | step=44650/50000 progress=89.3% loss=0.086836 lr=3.097591e-06 lr_backbone=3.097591e-06 elapsed=46:12:01 avg_step=3.73s recent_step=2.76s eta=05:32:08 +2026-09-19 15:41:25 | INFO | train rank=0 | step=44660/50000 progress=89.3% loss=0.101632 lr=3.086142e-06 lr_backbone=3.086142e-06 elapsed=46:12:49 avg_step=3.73s recent_step=4.71s eta=05:31:32 +2026-09-19 15:41:59 | INFO | train rank=0 | step=44670/50000 progress=89.3% loss=0.091676 lr=3.074715e-06 lr_backbone=3.074715e-06 elapsed=46:13:22 avg_step=3.73s recent_step=3.39s eta=05:30:55 +2026-09-19 15:42:39 | INFO | train rank=0 | step=44680/50000 progress=89.4% loss=0.080303 lr=3.063307e-06 lr_backbone=3.063307e-06 elapsed=46:14:02 avg_step=3.73s recent_step=3.94s eta=05:30:18 +2026-09-19 15:43:03 | INFO | train rank=0 | step=44690/50000 progress=89.4% loss=0.072491 lr=3.051920e-06 lr_backbone=3.051920e-06 elapsed=46:14:27 avg_step=3.72s recent_step=2.47s eta=05:29:39 +2026-09-19 15:43:56 | INFO | train rank=0 | step=44700/50000 progress=89.4% loss=0.084429 lr=3.040554e-06 lr_backbone=3.040554e-06 elapsed=46:15:19 avg_step=3.73s recent_step=5.24s eta=05:29:03 +2026-09-19 15:44:34 | INFO | train rank=0 | step=44710/50000 progress=89.4% loss=0.088538 lr=3.029208e-06 lr_backbone=3.029208e-06 elapsed=46:15:58 avg_step=3.73s recent_step=3.87s eta=05:28:26 +2026-09-19 15:45:09 | INFO | train rank=0 | step=44720/50000 progress=89.4% loss=0.088180 lr=3.017883e-06 lr_backbone=3.017883e-06 elapsed=46:16:32 avg_step=3.73s recent_step=3.43s eta=05:27:49 +2026-09-19 15:45:40 | INFO | train rank=0 | step=44730/50000 progress=89.5% loss=0.071166 lr=3.006578e-06 lr_backbone=3.006578e-06 elapsed=46:17:03 avg_step=3.73s recent_step=3.13s eta=05:27:11 +2026-09-19 15:46:31 | INFO | train rank=0 | step=44740/50000 progress=89.5% loss=0.117088 lr=2.995294e-06 lr_backbone=2.995294e-06 elapsed=46:17:54 avg_step=3.73s recent_step=5.10s eta=05:26:35 +2026-09-19 15:47:04 | INFO | train rank=0 | step=44750/50000 progress=89.5% loss=0.076094 lr=2.984030e-06 lr_backbone=2.984030e-06 elapsed=46:18:28 avg_step=3.73s recent_step=3.36s eta=05:25:57 +2026-09-19 15:47:34 | INFO | train rank=0 | step=44760/50000 progress=89.5% loss=0.074722 lr=2.972787e-06 lr_backbone=2.972787e-06 elapsed=46:18:57 avg_step=3.73s recent_step=2.95s eta=05:25:19 +2026-09-19 15:48:08 | INFO | train rank=0 | step=44770/50000 progress=89.5% loss=0.119231 lr=2.961565e-06 lr_backbone=2.961565e-06 elapsed=46:19:31 avg_step=3.73s recent_step=3.36s eta=05:24:42 +2026-09-19 15:48:58 | INFO | train rank=0 | step=44780/50000 progress=89.6% loss=0.112602 lr=2.950363e-06 lr_backbone=2.950363e-06 elapsed=46:20:21 avg_step=3.73s recent_step=5.01s eta=05:24:06 +2026-09-19 15:49:32 | INFO | train rank=0 | step=44790/50000 progress=89.6% loss=0.058350 lr=2.939182e-06 lr_backbone=2.939182e-06 elapsed=46:20:55 avg_step=3.73s recent_step=3.42s eta=05:23:28 +2026-09-19 15:50:01 | INFO | train rank=0 | step=44800/50000 progress=89.6% loss=0.139655 lr=2.928021e-06 lr_backbone=2.928021e-06 elapsed=46:21:24 avg_step=3.73s recent_step=2.89s eta=05:22:50 +2026-09-19 15:50:38 | INFO | train rank=0 | step=44810/50000 progress=89.6% loss=0.069664 lr=2.916881e-06 lr_backbone=2.916881e-06 elapsed=46:22:02 avg_step=3.73s recent_step=3.76s eta=05:22:13 +2026-09-19 15:51:16 | INFO | train rank=0 | step=44820/50000 progress=89.6% loss=0.076826 lr=2.905762e-06 lr_backbone=2.905762e-06 elapsed=46:22:39 avg_step=3.73s recent_step=3.76s eta=05:21:36 +2026-09-19 15:51:51 | INFO | train rank=0 | step=44830/50000 progress=89.7% loss=0.087695 lr=2.894663e-06 lr_backbone=2.894663e-06 elapsed=46:23:15 avg_step=3.73s recent_step=3.55s eta=05:20:58 +2026-09-19 15:52:30 | INFO | train rank=0 | step=44840/50000 progress=89.7% loss=0.066412 lr=2.883584e-06 lr_backbone=2.883584e-06 elapsed=46:23:53 avg_step=3.73s recent_step=3.86s eta=05:20:21 +2026-09-19 15:53:06 | INFO | train rank=0 | step=44850/50000 progress=89.7% loss=0.084874 lr=2.872527e-06 lr_backbone=2.872527e-06 elapsed=46:24:29 avg_step=3.73s recent_step=3.59s eta=05:19:44 +2026-09-19 15:53:58 | INFO | train rank=0 | step=44860/50000 progress=89.7% loss=0.064832 lr=2.861490e-06 lr_backbone=2.861490e-06 elapsed=46:25:22 avg_step=3.73s recent_step=5.27s eta=05:19:08 +2026-09-19 15:54:28 | INFO | train rank=0 | step=44870/50000 progress=89.7% loss=0.078339 lr=2.850473e-06 lr_backbone=2.850473e-06 elapsed=46:25:51 avg_step=3.73s recent_step=2.96s eta=05:18:30 +2026-09-19 15:55:00 | INFO | train rank=0 | step=44880/50000 progress=89.8% loss=0.084005 lr=2.839477e-06 lr_backbone=2.839477e-06 elapsed=46:26:23 avg_step=3.73s recent_step=3.21s eta=05:17:52 +2026-09-19 15:55:31 | INFO | train rank=0 | step=44890/50000 progress=89.8% loss=0.102466 lr=2.828502e-06 lr_backbone=2.828502e-06 elapsed=46:26:54 avg_step=3.72s recent_step=3.07s eta=05:17:14 +2026-09-19 15:56:33 | INFO | train rank=0 | step=44900/50000 progress=89.8% loss=0.088704 lr=2.817548e-06 lr_backbone=2.817548e-06 elapsed=46:27:56 avg_step=3.73s recent_step=6.18s eta=05:16:40 +2026-09-19 15:56:57 | INFO | train rank=0 | step=44910/50000 progress=89.8% loss=0.068828 lr=2.806614e-06 lr_backbone=2.806614e-06 elapsed=46:28:20 avg_step=3.73s recent_step=2.44s eta=05:16:01 +2026-09-19 15:57:26 | INFO | train rank=0 | step=44920/50000 progress=89.8% loss=0.062140 lr=2.795700e-06 lr_backbone=2.795700e-06 elapsed=46:28:49 avg_step=3.73s recent_step=2.89s eta=05:15:23 +2026-09-19 15:57:57 | INFO | train rank=0 | step=44930/50000 progress=89.9% loss=0.082270 lr=2.784808e-06 lr_backbone=2.784808e-06 elapsed=46:29:20 avg_step=3.72s recent_step=3.09s eta=05:14:45 +2026-09-19 15:58:51 | INFO | train rank=0 | step=44940/50000 progress=89.9% loss=0.084940 lr=2.773936e-06 lr_backbone=2.773936e-06 elapsed=46:30:14 avg_step=3.73s recent_step=5.41s eta=05:14:10 +2026-09-19 15:59:22 | INFO | train rank=0 | step=44950/50000 progress=89.9% loss=0.081615 lr=2.763084e-06 lr_backbone=2.763084e-06 elapsed=46:30:46 avg_step=3.73s recent_step=3.14s eta=05:13:32 +2026-09-19 15:59:54 | INFO | train rank=0 | step=44960/50000 progress=89.9% loss=0.087988 lr=2.752254e-06 lr_backbone=2.752254e-06 elapsed=46:31:17 avg_step=3.73s recent_step=3.17s eta=05:12:54 +2026-09-19 16:00:35 | INFO | train rank=0 | step=44970/50000 progress=89.9% loss=0.072281 lr=2.741444e-06 lr_backbone=2.741444e-06 elapsed=46:31:59 avg_step=3.73s recent_step=4.12s eta=05:12:17 +2026-09-19 16:01:14 | INFO | train rank=0 | step=44980/50000 progress=90.0% loss=0.080943 lr=2.730655e-06 lr_backbone=2.730655e-06 elapsed=46:32:37 avg_step=3.73s recent_step=3.86s eta=05:11:40 +2026-09-19 16:01:43 | INFO | train rank=0 | step=44990/50000 progress=90.0% loss=0.102355 lr=2.719886e-06 lr_backbone=2.719886e-06 elapsed=46:33:06 avg_step=3.72s recent_step=2.93s eta=05:11:02 +2026-09-19 16:02:35 | INFO | train rank=0 | step=45000/50000 progress=90.0% loss=0.066795 lr=2.709138e-06 lr_backbone=2.709138e-06 elapsed=46:33:58 avg_step=3.73s recent_step=5.16s eta=05:10:26 +2026-09-19 16:03:02 | INFO | train rank=0 | step=45010/50000 progress=90.0% loss=0.082162 lr=2.698411e-06 lr_backbone=2.698411e-06 elapsed=46:34:26 avg_step=3.73s recent_step=2.77s eta=05:09:48 +2026-09-19 16:03:44 | INFO | train rank=0 | step=45020/50000 progress=90.0% loss=0.082232 lr=2.687704e-06 lr_backbone=2.687704e-06 elapsed=46:35:07 avg_step=3.73s recent_step=4.17s eta=05:09:11 +2026-09-19 16:04:17 | INFO | train rank=0 | step=45030/50000 progress=90.1% loss=0.083477 lr=2.677018e-06 lr_backbone=2.677018e-06 elapsed=46:35:40 avg_step=3.73s recent_step=3.28s eta=05:08:33 +2026-09-19 16:05:03 | INFO | train rank=0 | step=45040/50000 progress=90.1% loss=0.098751 lr=2.666353e-06 lr_backbone=2.666353e-06 elapsed=46:36:26 avg_step=3.73s recent_step=4.56s eta=05:07:57 +2026-09-19 16:05:39 | INFO | train rank=0 | step=45050/50000 progress=90.1% loss=0.072956 lr=2.655709e-06 lr_backbone=2.655709e-06 elapsed=46:37:03 avg_step=3.73s recent_step=3.67s eta=05:07:20 +2026-09-19 16:06:15 | INFO | train rank=0 | step=45060/50000 progress=90.1% loss=0.080125 lr=2.645085e-06 lr_backbone=2.645085e-06 elapsed=46:37:38 avg_step=3.73s recent_step=3.58s eta=05:06:42 +2026-09-19 16:06:46 | INFO | train rank=0 | step=45070/50000 progress=90.1% loss=0.113440 lr=2.634482e-06 lr_backbone=2.634482e-06 elapsed=46:38:09 avg_step=3.73s recent_step=3.06s eta=05:06:04 +2026-09-19 16:07:20 | INFO | train rank=0 | step=45080/50000 progress=90.2% loss=0.078608 lr=2.623899e-06 lr_backbone=2.623899e-06 elapsed=46:38:43 avg_step=3.73s recent_step=3.39s eta=05:05:27 +2026-09-19 16:07:50 | INFO | train rank=0 | step=45090/50000 progress=90.2% loss=0.073584 lr=2.613338e-06 lr_backbone=2.613338e-06 elapsed=46:39:13 avg_step=3.72s recent_step=3.05s eta=05:04:49 +2026-09-19 16:08:44 | INFO | train rank=0 | step=45100/50000 progress=90.2% loss=0.094462 lr=2.602797e-06 lr_backbone=2.602797e-06 elapsed=46:40:07 avg_step=3.73s recent_step=5.39s eta=05:04:13 +2026-09-19 16:09:19 | INFO | train rank=0 | step=45110/50000 progress=90.2% loss=0.079070 lr=2.592277e-06 lr_backbone=2.592277e-06 elapsed=46:40:42 avg_step=3.73s recent_step=3.45s eta=05:03:36 +2026-09-19 16:09:47 | INFO | train rank=0 | step=45120/50000 progress=90.2% loss=0.122090 lr=2.581777e-06 lr_backbone=2.581777e-06 elapsed=46:41:10 avg_step=3.72s recent_step=2.83s eta=05:02:57 +2026-09-19 16:10:15 | INFO | train rank=0 | step=45130/50000 progress=90.3% loss=0.083546 lr=2.571299e-06 lr_backbone=2.571299e-06 elapsed=46:41:38 avg_step=3.72s recent_step=2.77s eta=05:02:19 +2026-09-19 16:11:16 | INFO | train rank=0 | step=45140/50000 progress=90.3% loss=0.068298 lr=2.560841e-06 lr_backbone=2.560841e-06 elapsed=46:42:40 avg_step=3.73s recent_step=6.17s eta=05:01:44 +2026-09-19 16:11:47 | INFO | train rank=0 | step=45150/50000 progress=90.3% loss=0.076662 lr=2.550404e-06 lr_backbone=2.550404e-06 elapsed=46:43:10 avg_step=3.73s recent_step=3.09s eta=05:01:07 +2026-09-19 16:12:24 | INFO | train rank=0 | step=45160/50000 progress=90.3% loss=0.071986 lr=2.539987e-06 lr_backbone=2.539987e-06 elapsed=46:43:47 avg_step=3.73s recent_step=3.66s eta=05:00:29 +2026-09-19 16:12:55 | INFO | train rank=0 | step=45170/50000 progress=90.3% loss=0.074211 lr=2.529591e-06 lr_backbone=2.529591e-06 elapsed=46:44:18 avg_step=3.73s recent_step=3.08s eta=04:59:51 +2026-09-19 16:13:50 | INFO | train rank=0 | step=45180/50000 progress=90.4% loss=0.088785 lr=2.519217e-06 lr_backbone=2.519217e-06 elapsed=46:45:13 avg_step=3.73s recent_step=5.51s eta=04:59:16 +2026-09-19 16:14:19 | INFO | train rank=0 | step=45190/50000 progress=90.4% loss=0.081691 lr=2.508862e-06 lr_backbone=2.508862e-06 elapsed=46:45:42 avg_step=3.73s recent_step=2.92s eta=04:58:38 +2026-09-19 16:14:48 | INFO | train rank=0 | step=45200/50000 progress=90.4% loss=0.077512 lr=2.498529e-06 lr_backbone=2.498529e-06 elapsed=46:46:12 avg_step=3.73s recent_step=2.95s eta=04:58:00 +2026-09-19 16:15:21 | INFO | train rank=0 | step=45210/50000 progress=90.4% loss=0.087433 lr=2.488217e-06 lr_backbone=2.488217e-06 elapsed=46:46:45 avg_step=3.72s recent_step=3.30s eta=04:57:22 +2026-09-19 16:16:12 | INFO | train rank=0 | step=45220/50000 progress=90.4% loss=0.088896 lr=2.477925e-06 lr_backbone=2.477925e-06 elapsed=46:47:35 avg_step=3.73s recent_step=5.04s eta=04:56:46 +2026-09-19 16:16:39 | INFO | train rank=0 | step=45230/50000 progress=90.5% loss=0.072656 lr=2.467654e-06 lr_backbone=2.467654e-06 elapsed=46:48:02 avg_step=3.73s recent_step=2.67s eta=04:56:08 +2026-09-19 16:17:16 | INFO | train rank=0 | step=45240/50000 progress=90.5% loss=0.090073 lr=2.457404e-06 lr_backbone=2.457404e-06 elapsed=46:48:39 avg_step=3.73s recent_step=3.71s eta=04:55:31 +2026-09-19 16:17:42 | INFO | train rank=0 | step=45250/50000 progress=90.5% loss=0.101082 lr=2.447174e-06 lr_backbone=2.447174e-06 elapsed=46:49:05 avg_step=3.72s recent_step=2.61s eta=04:54:52 +2026-09-19 16:18:28 | INFO | train rank=0 | step=45260/50000 progress=90.5% loss=0.091048 lr=2.436966e-06 lr_backbone=2.436966e-06 elapsed=46:49:51 avg_step=3.72s recent_step=4.64s eta=04:54:16 +2026-09-19 16:18:58 | INFO | train rank=0 | step=45270/50000 progress=90.5% loss=0.088993 lr=2.426778e-06 lr_backbone=2.426778e-06 elapsed=46:50:21 avg_step=3.72s recent_step=3.00s eta=04:53:38 +2026-09-19 16:19:31 | INFO | train rank=0 | step=45280/50000 progress=90.6% loss=0.096721 lr=2.416611e-06 lr_backbone=2.416611e-06 elapsed=46:50:55 avg_step=3.72s recent_step=3.32s eta=04:53:00 +2026-09-19 16:20:06 | INFO | train rank=0 | step=45290/50000 progress=90.6% loss=0.074111 lr=2.406465e-06 lr_backbone=2.406465e-06 elapsed=46:51:30 avg_step=3.72s recent_step=3.51s eta=04:52:23 +2026-09-19 16:21:03 | INFO | train rank=0 | step=45300/50000 progress=90.6% loss=0.069005 lr=2.396339e-06 lr_backbone=2.396339e-06 elapsed=46:52:26 avg_step=3.73s recent_step=5.66s eta=04:51:47 +2026-09-19 16:21:37 | INFO | train rank=0 | step=45310/50000 progress=90.6% loss=0.078399 lr=2.386235e-06 lr_backbone=2.386235e-06 elapsed=46:53:00 avg_step=3.73s recent_step=3.40s eta=04:51:10 +2026-09-19 16:22:09 | INFO | train rank=0 | step=45320/50000 progress=90.6% loss=0.101621 lr=2.376151e-06 lr_backbone=2.376151e-06 elapsed=46:53:32 avg_step=3.72s recent_step=3.21s eta=04:50:32 +2026-09-19 16:22:44 | INFO | train rank=0 | step=45330/50000 progress=90.7% loss=0.097474 lr=2.366088e-06 lr_backbone=2.366088e-06 elapsed=46:54:08 avg_step=3.72s recent_step=3.52s eta=04:49:55 +2026-09-19 16:23:25 | INFO | train rank=0 | step=45340/50000 progress=90.7% loss=0.058566 lr=2.356046e-06 lr_backbone=2.356046e-06 elapsed=46:54:48 avg_step=3.72s recent_step=4.06s eta=04:49:18 +2026-09-19 16:24:03 | INFO | train rank=0 | step=45350/50000 progress=90.7% loss=0.077851 lr=2.346025e-06 lr_backbone=2.346025e-06 elapsed=46:55:26 avg_step=3.72s recent_step=3.78s eta=04:48:41 +2026-09-19 16:24:42 | INFO | train rank=0 | step=45360/50000 progress=90.7% loss=0.113759 lr=2.336025e-06 lr_backbone=2.336025e-06 elapsed=46:56:05 avg_step=3.72s recent_step=3.89s eta=04:48:03 +2026-09-19 16:25:11 | INFO | train rank=0 | step=45370/50000 progress=90.7% loss=0.075697 lr=2.326045e-06 lr_backbone=2.326045e-06 elapsed=46:56:34 avg_step=3.72s recent_step=2.94s eta=04:47:25 +2026-09-19 16:26:06 | INFO | train rank=0 | step=45380/50000 progress=90.8% loss=0.069521 lr=2.316086e-06 lr_backbone=2.316086e-06 elapsed=46:57:29 avg_step=3.73s recent_step=5.47s eta=04:46:50 +2026-09-19 16:26:33 | INFO | train rank=0 | step=45390/50000 progress=90.8% loss=0.076611 lr=2.306149e-06 lr_backbone=2.306149e-06 elapsed=46:57:57 avg_step=3.72s recent_step=2.76s eta=04:46:12 +2026-09-19 16:27:11 | INFO | train rank=0 | step=45400/50000 progress=90.8% loss=0.077612 lr=2.296232e-06 lr_backbone=2.296232e-06 elapsed=46:58:35 avg_step=3.73s recent_step=3.79s eta=04:45:35 +2026-09-19 16:27:43 | INFO | train rank=0 | step=45410/50000 progress=90.8% loss=0.075877 lr=2.286336e-06 lr_backbone=2.286336e-06 elapsed=46:59:06 avg_step=3.72s recent_step=3.13s eta=04:44:57 +2026-09-19 16:28:40 | INFO | train rank=0 | step=45420/50000 progress=90.8% loss=0.092263 lr=2.276460e-06 lr_backbone=2.276460e-06 elapsed=47:00:03 avg_step=3.73s recent_step=5.75s eta=04:44:21 +2026-09-19 16:29:14 | INFO | train rank=0 | step=45430/50000 progress=90.9% loss=0.079070 lr=2.266606e-06 lr_backbone=2.266606e-06 elapsed=47:00:37 avg_step=3.73s recent_step=3.34s eta=04:43:44 +2026-09-19 16:29:43 | INFO | train rank=0 | step=45440/50000 progress=90.9% loss=0.076708 lr=2.256773e-06 lr_backbone=2.256773e-06 elapsed=47:01:06 avg_step=3.73s recent_step=2.95s eta=04:43:06 +2026-09-19 16:30:20 | INFO | train rank=0 | step=45450/50000 progress=90.9% loss=0.093917 lr=2.246960e-06 lr_backbone=2.246960e-06 elapsed=47:01:44 avg_step=3.73s recent_step=3.73s eta=04:42:29 +2026-09-19 16:31:10 | INFO | train rank=0 | step=45460/50000 progress=90.9% loss=0.095190 lr=2.237169e-06 lr_backbone=2.237169e-06 elapsed=47:02:33 avg_step=3.73s recent_step=4.96s eta=04:41:53 +2026-09-19 16:31:37 | INFO | train rank=0 | step=45470/50000 progress=90.9% loss=0.080184 lr=2.227398e-06 lr_backbone=2.227398e-06 elapsed=47:03:00 avg_step=3.73s recent_step=2.69s eta=04:41:14 +2026-09-19 16:32:11 | INFO | train rank=0 | step=45480/50000 progress=91.0% loss=0.101307 lr=2.217648e-06 lr_backbone=2.217648e-06 elapsed=47:03:34 avg_step=3.73s recent_step=3.41s eta=04:40:37 +2026-09-19 16:32:41 | INFO | train rank=0 | step=45490/50000 progress=91.0% loss=0.081114 lr=2.207919e-06 lr_backbone=2.207919e-06 elapsed=47:04:05 avg_step=3.72s recent_step=3.05s eta=04:39:59 +2026-09-19 16:33:28 | INFO | train rank=0 | step=45500/50000 progress=91.0% loss=0.101883 lr=2.198211e-06 lr_backbone=2.198211e-06 elapsed=47:04:52 avg_step=3.73s recent_step=4.69s eta=04:39:22 +2026-09-19 16:33:54 | INFO | train rank=0 | step=45510/50000 progress=91.0% loss=0.089010 lr=2.188524e-06 lr_backbone=2.188524e-06 elapsed=47:05:18 avg_step=3.72s recent_step=2.60s eta=04:38:44 +2026-09-19 16:34:27 | INFO | train rank=0 | step=45520/50000 progress=91.0% loss=0.082450 lr=2.178858e-06 lr_backbone=2.178858e-06 elapsed=47:05:51 avg_step=3.72s recent_step=3.30s eta=04:38:06 +2026-09-19 16:35:00 | INFO | train rank=0 | step=45530/50000 progress=91.1% loss=0.090153 lr=2.169212e-06 lr_backbone=2.169212e-06 elapsed=47:06:23 avg_step=3.72s recent_step=3.22s eta=04:37:29 +2026-09-19 16:35:57 | INFO | train rank=0 | step=45540/50000 progress=91.1% loss=0.090516 lr=2.159588e-06 lr_backbone=2.159588e-06 elapsed=47:07:20 avg_step=3.73s recent_step=5.73s eta=04:36:53 +2026-09-19 16:36:31 | INFO | train rank=0 | step=45550/50000 progress=91.1% loss=0.084712 lr=2.149984e-06 lr_backbone=2.149984e-06 elapsed=47:07:55 avg_step=3.73s recent_step=3.44s eta=04:36:16 +2026-09-19 16:37:10 | INFO | train rank=0 | step=45560/50000 progress=91.1% loss=0.081013 lr=2.140402e-06 lr_backbone=2.140402e-06 elapsed=47:08:33 avg_step=3.73s recent_step=3.81s eta=04:35:39 +2026-09-19 16:37:47 | INFO | train rank=0 | step=45570/50000 progress=91.1% loss=0.090124 lr=2.130840e-06 lr_backbone=2.130840e-06 elapsed=47:09:10 avg_step=3.73s recent_step=3.72s eta=04:35:01 +2026-09-19 16:38:37 | INFO | train rank=0 | step=45580/50000 progress=91.2% loss=0.089645 lr=2.121300e-06 lr_backbone=2.121300e-06 elapsed=47:10:00 avg_step=3.73s recent_step=5.00s eta=04:34:25 +2026-09-19 16:39:08 | INFO | train rank=0 | step=45590/50000 progress=91.2% loss=0.066708 lr=2.111780e-06 lr_backbone=2.111780e-06 elapsed=47:10:31 avg_step=3.73s recent_step=3.13s eta=04:33:48 +2026-09-19 16:39:45 | INFO | train rank=0 | step=45600/50000 progress=91.2% loss=0.076420 lr=2.102281e-06 lr_backbone=2.102281e-06 elapsed=47:11:08 avg_step=3.73s recent_step=3.66s eta=04:33:10 +2026-09-19 16:40:17 | INFO | train rank=0 | step=45610/50000 progress=91.2% loss=0.108442 lr=2.092803e-06 lr_backbone=2.092803e-06 elapsed=47:11:41 avg_step=3.73s recent_step=3.28s eta=04:32:33 +2026-09-19 16:41:06 | INFO | train rank=0 | step=45620/50000 progress=91.2% loss=0.085773 lr=2.083346e-06 lr_backbone=2.083346e-06 elapsed=47:12:29 avg_step=3.73s recent_step=4.85s eta=04:31:56 +2026-09-19 16:41:31 | INFO | train rank=0 | step=45630/50000 progress=91.3% loss=0.066278 lr=2.073911e-06 lr_backbone=2.073911e-06 elapsed=47:12:55 avg_step=3.73s recent_step=2.54s eta=04:31:18 +2026-09-19 16:41:57 | INFO | train rank=0 | step=45640/50000 progress=91.3% loss=0.090957 lr=2.064496e-06 lr_backbone=2.064496e-06 elapsed=47:13:20 avg_step=3.72s recent_step=2.57s eta=04:30:40 +2026-09-19 16:42:29 | INFO | train rank=0 | step=45650/50000 progress=91.3% loss=0.065038 lr=2.055102e-06 lr_backbone=2.055102e-06 elapsed=47:13:52 avg_step=3.72s recent_step=3.17s eta=04:30:02 +2026-09-19 16:43:21 | INFO | train rank=0 | step=45660/50000 progress=91.3% loss=0.169615 lr=2.045729e-06 lr_backbone=2.045729e-06 elapsed=47:14:45 avg_step=3.73s recent_step=5.26s eta=04:29:26 +2026-09-19 16:43:59 | INFO | train rank=0 | step=45670/50000 progress=91.3% loss=0.104318 lr=2.036377e-06 lr_backbone=2.036377e-06 elapsed=47:15:23 avg_step=3.73s recent_step=3.81s eta=04:28:49 +2026-09-19 16:44:34 | INFO | train rank=0 | step=45680/50000 progress=91.4% loss=0.083874 lr=2.027046e-06 lr_backbone=2.027046e-06 elapsed=47:15:57 avg_step=3.72s recent_step=3.45s eta=04:28:11 +2026-09-19 16:45:03 | INFO | train rank=0 | step=45690/50000 progress=91.4% loss=0.072940 lr=2.017736e-06 lr_backbone=2.017736e-06 elapsed=47:16:27 avg_step=3.72s recent_step=2.97s eta=04:27:33 +2026-09-19 16:46:01 | INFO | train rank=0 | step=45700/50000 progress=91.4% loss=0.083061 lr=2.008447e-06 lr_backbone=2.008447e-06 elapsed=47:17:25 avg_step=3.73s recent_step=5.80s eta=04:26:58 +2026-09-19 16:46:39 | INFO | train rank=0 | step=45710/50000 progress=91.4% loss=0.079404 lr=1.999178e-06 lr_backbone=1.999178e-06 elapsed=47:18:02 avg_step=3.73s recent_step=3.74s eta=04:26:21 +2026-09-19 16:47:08 | INFO | train rank=0 | step=45720/50000 progress=91.4% loss=0.075684 lr=1.989931e-06 lr_backbone=1.989931e-06 elapsed=47:18:31 avg_step=3.73s recent_step=2.87s eta=04:25:43 +2026-09-19 16:47:41 | INFO | train rank=0 | step=45730/50000 progress=91.5% loss=0.069808 lr=1.980705e-06 lr_backbone=1.980705e-06 elapsed=47:19:05 avg_step=3.73s recent_step=3.39s eta=04:25:05 +2026-09-19 16:48:34 | INFO | train rank=0 | step=45740/50000 progress=91.5% loss=0.091767 lr=1.971500e-06 lr_backbone=1.971500e-06 elapsed=47:19:57 avg_step=3.73s recent_step=5.24s eta=04:24:30 +2026-09-19 16:49:08 | INFO | train rank=0 | step=45750/50000 progress=91.5% loss=0.082305 lr=1.962316e-06 lr_backbone=1.962316e-06 elapsed=47:20:32 avg_step=3.73s recent_step=3.45s eta=04:23:52 +2026-09-19 16:49:46 | INFO | train rank=0 | step=45760/50000 progress=91.5% loss=0.065679 lr=1.953153e-06 lr_backbone=1.953153e-06 elapsed=47:21:09 avg_step=3.73s recent_step=3.77s eta=04:23:15 +2026-09-19 16:50:18 | INFO | train rank=0 | step=45770/50000 progress=91.5% loss=0.105889 lr=1.944011e-06 lr_backbone=1.944011e-06 elapsed=47:21:41 avg_step=3.73s recent_step=3.20s eta=04:22:37 +2026-09-19 16:51:06 | INFO | train rank=0 | step=45780/50000 progress=91.6% loss=0.066172 lr=1.934890e-06 lr_backbone=1.934890e-06 elapsed=47:22:29 avg_step=3.73s recent_step=4.79s eta=04:22:01 +2026-09-19 16:51:34 | INFO | train rank=0 | step=45790/50000 progress=91.6% loss=0.071829 lr=1.925790e-06 lr_backbone=1.925790e-06 elapsed=47:22:57 avg_step=3.73s recent_step=2.81s eta=04:21:23 +2026-09-19 16:52:11 | INFO | train rank=0 | step=45800/50000 progress=91.6% loss=0.103559 lr=1.916711e-06 lr_backbone=1.916711e-06 elapsed=47:23:34 avg_step=3.73s recent_step=3.66s eta=04:20:45 +2026-09-19 16:52:42 | INFO | train rank=0 | step=45810/50000 progress=91.6% loss=0.072713 lr=1.907653e-06 lr_backbone=1.907653e-06 elapsed=47:24:05 avg_step=3.73s recent_step=3.08s eta=04:20:08 +2026-09-19 16:53:36 | INFO | train rank=0 | step=45820/50000 progress=91.6% loss=0.109187 lr=1.898616e-06 lr_backbone=1.898616e-06 elapsed=47:25:00 avg_step=3.73s recent_step=5.47s eta=04:19:32 +2026-09-19 16:54:09 | INFO | train rank=0 | step=45830/50000 progress=91.7% loss=0.082093 lr=1.889601e-06 lr_backbone=1.889601e-06 elapsed=47:25:32 avg_step=3.73s recent_step=3.24s eta=04:18:54 +2026-09-19 16:54:39 | INFO | train rank=0 | step=45840/50000 progress=91.7% loss=0.120216 lr=1.880606e-06 lr_backbone=1.880606e-06 elapsed=47:26:03 avg_step=3.73s recent_step=3.06s eta=04:18:16 +2026-09-19 16:55:14 | INFO | train rank=0 | step=45850/50000 progress=91.7% loss=0.096013 lr=1.871632e-06 lr_backbone=1.871632e-06 elapsed=47:26:37 avg_step=3.73s recent_step=3.49s eta=04:17:39 +2026-09-19 16:56:03 | INFO | train rank=0 | step=45860/50000 progress=91.7% loss=0.099769 lr=1.862679e-06 lr_backbone=1.862679e-06 elapsed=47:27:27 avg_step=3.73s recent_step=4.93s eta=04:17:03 +2026-09-19 16:56:25 | INFO | train rank=0 | step=45870/50000 progress=91.7% loss=0.085914 lr=1.853748e-06 lr_backbone=1.853748e-06 elapsed=47:27:48 avg_step=3.73s recent_step=2.18s eta=04:16:24 +2026-09-19 16:56:53 | INFO | train rank=0 | step=45880/50000 progress=91.8% loss=0.079055 lr=1.844837e-06 lr_backbone=1.844837e-06 elapsed=47:28:17 avg_step=3.72s recent_step=2.82s eta=04:15:46 +2026-09-19 16:57:23 | INFO | train rank=0 | step=45890/50000 progress=91.8% loss=0.103733 lr=1.835948e-06 lr_backbone=1.835948e-06 elapsed=47:28:46 avg_step=3.72s recent_step=2.93s eta=04:15:08 +2026-09-19 16:58:09 | INFO | train rank=0 | step=45900/50000 progress=91.8% loss=0.082802 lr=1.827079e-06 lr_backbone=1.827079e-06 elapsed=47:29:32 avg_step=3.72s recent_step=4.60s eta=04:14:32 +2026-09-19 16:58:34 | INFO | train rank=0 | step=45910/50000 progress=91.8% loss=0.074709 lr=1.818232e-06 lr_backbone=1.818232e-06 elapsed=47:29:57 avg_step=3.72s recent_step=2.53s eta=04:13:53 +2026-09-19 16:59:08 | INFO | train rank=0 | step=45920/50000 progress=91.8% loss=0.079263 lr=1.809406e-06 lr_backbone=1.809406e-06 elapsed=47:30:31 avg_step=3.72s recent_step=3.40s eta=04:13:16 +2026-09-19 16:59:40 | INFO | train rank=0 | step=45930/50000 progress=91.9% loss=0.081492 lr=1.800600e-06 lr_backbone=1.800600e-06 elapsed=47:31:03 avg_step=3.72s recent_step=3.21s eta=04:12:38 +2026-09-19 17:00:35 | INFO | train rank=0 | step=45940/50000 progress=91.9% loss=0.104350 lr=1.791816e-06 lr_backbone=1.791816e-06 elapsed=47:31:58 avg_step=3.72s recent_step=5.46s eta=04:12:02 +2026-09-19 17:01:10 | INFO | train rank=0 | step=45950/50000 progress=91.9% loss=0.087660 lr=1.783053e-06 lr_backbone=1.783053e-06 elapsed=47:32:33 avg_step=3.72s recent_step=3.54s eta=04:11:25 +2026-09-19 17:01:47 | INFO | train rank=0 | step=45960/50000 progress=91.9% loss=0.082178 lr=1.774311e-06 lr_backbone=1.774311e-06 elapsed=47:33:10 avg_step=3.72s recent_step=3.67s eta=04:10:48 +2026-09-19 17:02:25 | INFO | train rank=0 | step=45970/50000 progress=91.9% loss=0.083129 lr=1.765590e-06 lr_backbone=1.765590e-06 elapsed=47:33:48 avg_step=3.72s recent_step=3.85s eta=04:10:10 +2026-09-19 17:03:08 | INFO | train rank=0 | step=45980/50000 progress=92.0% loss=0.081488 lr=1.756891e-06 lr_backbone=1.756891e-06 elapsed=47:34:31 avg_step=3.72s recent_step=4.27s eta=04:09:34 +2026-09-19 17:03:50 | INFO | train rank=0 | step=45990/50000 progress=92.0% loss=0.076131 lr=1.748212e-06 lr_backbone=1.748212e-06 elapsed=47:35:13 avg_step=3.73s recent_step=4.21s eta=04:08:57 +2026-09-19 17:04:22 | INFO | train rank=0 | step=46000/50000 progress=92.0% loss=0.055840 lr=1.739554e-06 lr_backbone=1.739554e-06 elapsed=47:35:46 avg_step=3.72s recent_step=3.25s eta=04:08:19 +2026-09-19 17:04:48 | INFO | train rank=0 | step=46010/50000 progress=92.0% loss=0.106126 lr=1.730918e-06 lr_backbone=1.730918e-06 elapsed=47:36:11 avg_step=3.72s recent_step=2.55s eta=04:07:41 +2026-09-19 17:05:42 | INFO | train rank=0 | step=46020/50000 progress=92.0% loss=0.101814 lr=1.722303e-06 lr_backbone=1.722303e-06 elapsed=47:37:05 avg_step=3.73s recent_step=5.41s eta=04:07:05 +2026-09-19 17:06:15 | INFO | train rank=0 | step=46030/50000 progress=92.1% loss=0.084909 lr=1.713709e-06 lr_backbone=1.713709e-06 elapsed=47:37:39 avg_step=3.72s recent_step=3.33s eta=04:06:28 +2026-09-19 17:06:51 | INFO | train rank=0 | step=46040/50000 progress=92.1% loss=0.086342 lr=1.705135e-06 lr_backbone=1.705135e-06 elapsed=47:38:14 avg_step=3.72s recent_step=3.55s eta=04:05:50 +2026-09-19 17:07:18 | INFO | train rank=0 | step=46050/50000 progress=92.1% loss=0.076739 lr=1.696584e-06 lr_backbone=1.696584e-06 elapsed=47:38:41 avg_step=3.72s recent_step=2.69s eta=04:05:12 +2026-09-19 17:08:18 | INFO | train rank=0 | step=46060/50000 progress=92.1% loss=0.084322 lr=1.688053e-06 lr_backbone=1.688053e-06 elapsed=47:39:42 avg_step=3.73s recent_step=6.05s eta=04:04:37 +2026-09-19 17:08:39 | INFO | train rank=0 | step=46070/50000 progress=92.1% loss=0.083205 lr=1.679543e-06 lr_backbone=1.679543e-06 elapsed=47:40:02 avg_step=3.72s recent_step=2.02s eta=04:03:58 +2026-09-19 17:09:09 | INFO | train rank=0 | step=46080/50000 progress=92.2% loss=0.086911 lr=1.671054e-06 lr_backbone=1.671054e-06 elapsed=47:40:32 avg_step=3.72s recent_step=3.02s eta=04:03:20 +2026-09-19 17:09:41 | INFO | train rank=0 | step=46090/50000 progress=92.2% loss=0.087508 lr=1.662587e-06 lr_backbone=1.662587e-06 elapsed=47:41:05 avg_step=3.72s recent_step=3.26s eta=04:02:43 +2026-09-19 17:10:39 | INFO | train rank=0 | step=46100/50000 progress=92.2% loss=0.096624 lr=1.654141e-06 lr_backbone=1.654141e-06 elapsed=47:42:02 avg_step=3.72s recent_step=5.71s eta=04:02:07 +2026-09-19 17:11:11 | INFO | train rank=0 | step=46110/50000 progress=92.2% loss=0.070078 lr=1.645716e-06 lr_backbone=1.645716e-06 elapsed=47:42:34 avg_step=3.72s recent_step=3.25s eta=04:01:29 +2026-09-19 17:11:37 | INFO | train rank=0 | step=46120/50000 progress=92.2% loss=0.095964 lr=1.637312e-06 lr_backbone=1.637312e-06 elapsed=47:43:01 avg_step=3.72s recent_step=2.63s eta=04:00:51 +2026-09-19 17:12:07 | INFO | train rank=0 | step=46130/50000 progress=92.3% loss=0.087531 lr=1.628929e-06 lr_backbone=1.628929e-06 elapsed=47:43:31 avg_step=3.72s recent_step=3.01s eta=04:00:13 +2026-09-19 17:13:10 | INFO | train rank=0 | step=46140/50000 progress=92.3% loss=0.074747 lr=1.620567e-06 lr_backbone=1.620567e-06 elapsed=47:44:33 avg_step=3.73s recent_step=6.21s eta=03:59:38 +2026-09-19 17:13:36 | INFO | train rank=0 | step=46150/50000 progress=92.3% loss=0.092171 lr=1.612227e-06 lr_backbone=1.612227e-06 elapsed=47:45:00 avg_step=3.72s recent_step=2.68s eta=03:59:00 +2026-09-19 17:14:13 | INFO | train rank=0 | step=46160/50000 progress=92.3% loss=0.077729 lr=1.603907e-06 lr_backbone=1.603907e-06 elapsed=47:45:36 avg_step=3.72s recent_step=3.65s eta=03:58:23 +2026-09-19 17:14:48 | INFO | train rank=0 | step=46170/50000 progress=92.3% loss=0.087614 lr=1.595609e-06 lr_backbone=1.595609e-06 elapsed=47:46:12 avg_step=3.72s recent_step=3.57s eta=03:57:45 +2026-09-19 17:15:34 | INFO | train rank=0 | step=46180/50000 progress=92.4% loss=0.105476 lr=1.587332e-06 lr_backbone=1.587332e-06 elapsed=47:46:57 avg_step=3.72s recent_step=4.54s eta=03:57:09 +2026-09-19 17:16:07 | INFO | train rank=0 | step=46190/50000 progress=92.4% loss=0.072600 lr=1.579077e-06 lr_backbone=1.579077e-06 elapsed=47:47:30 avg_step=3.72s recent_step=3.28s eta=03:56:31 +2026-09-19 17:16:40 | INFO | train rank=0 | step=46200/50000 progress=92.4% loss=0.101873 lr=1.570842e-06 lr_backbone=1.570842e-06 elapsed=47:48:03 avg_step=3.72s recent_step=3.32s eta=03:55:54 +2026-09-19 17:17:15 | INFO | train rank=0 | step=46210/50000 progress=92.4% loss=0.098708 lr=1.562629e-06 lr_backbone=1.562629e-06 elapsed=47:48:39 avg_step=3.72s recent_step=3.56s eta=03:55:16 +2026-09-19 17:18:16 | INFO | train rank=0 | step=46220/50000 progress=92.4% loss=0.090040 lr=1.554436e-06 lr_backbone=1.554436e-06 elapsed=47:49:39 avg_step=3.73s recent_step=6.01s eta=03:54:41 +2026-09-19 17:18:53 | INFO | train rank=0 | step=46230/50000 progress=92.5% loss=0.105550 lr=1.546265e-06 lr_backbone=1.546265e-06 elapsed=47:50:16 avg_step=3.73s recent_step=3.71s eta=03:54:04 +2026-09-19 17:19:22 | INFO | train rank=0 | step=46240/50000 progress=92.5% loss=0.096600 lr=1.538115e-06 lr_backbone=1.538115e-06 elapsed=47:50:45 avg_step=3.73s recent_step=2.92s eta=03:53:26 +2026-09-19 17:19:57 | INFO | train rank=0 | step=46250/50000 progress=92.5% loss=0.067395 lr=1.529987e-06 lr_backbone=1.529987e-06 elapsed=47:51:21 avg_step=3.72s recent_step=3.55s eta=03:52:48 +2026-09-19 17:20:44 | INFO | train rank=0 | step=46260/50000 progress=92.5% loss=0.086672 lr=1.521879e-06 lr_backbone=1.521879e-06 elapsed=47:52:08 avg_step=3.73s recent_step=4.70s eta=03:52:12 +2026-09-19 17:21:16 | INFO | train rank=0 | step=46270/50000 progress=92.5% loss=0.089932 lr=1.513793e-06 lr_backbone=1.513793e-06 elapsed=47:52:40 avg_step=3.73s recent_step=3.21s eta=03:51:34 +2026-09-19 17:21:50 | INFO | train rank=0 | step=46280/50000 progress=92.6% loss=0.088340 lr=1.505728e-06 lr_backbone=1.505728e-06 elapsed=47:53:14 avg_step=3.73s recent_step=3.40s eta=03:50:57 +2026-09-19 17:22:28 | INFO | train rank=0 | step=46290/50000 progress=92.6% loss=0.091085 lr=1.497684e-06 lr_backbone=1.497684e-06 elapsed=47:53:51 avg_step=3.73s recent_step=3.76s eta=03:50:19 +2026-09-19 17:23:19 | INFO | train rank=0 | step=46300/50000 progress=92.6% loss=0.142101 lr=1.489662e-06 lr_backbone=1.489662e-06 elapsed=47:54:43 avg_step=3.73s recent_step=5.12s eta=03:49:43 +2026-09-19 17:23:53 | INFO | train rank=0 | step=46310/50000 progress=92.6% loss=0.076647 lr=1.481660e-06 lr_backbone=1.481660e-06 elapsed=47:55:16 avg_step=3.73s recent_step=3.37s eta=03:49:06 +2026-09-19 17:24:28 | INFO | train rank=0 | step=46320/50000 progress=92.6% loss=0.092052 lr=1.473680e-06 lr_backbone=1.473680e-06 elapsed=47:55:51 avg_step=3.73s recent_step=3.47s eta=03:48:28 +2026-09-19 17:24:56 | INFO | train rank=0 | step=46330/50000 progress=92.7% loss=0.070279 lr=1.465721e-06 lr_backbone=1.465721e-06 elapsed=47:56:19 avg_step=3.73s recent_step=2.80s eta=03:47:50 +2026-09-19 17:25:44 | INFO | train rank=0 | step=46340/50000 progress=92.7% loss=0.074413 lr=1.457783e-06 lr_backbone=1.457783e-06 elapsed=47:57:07 avg_step=3.73s recent_step=4.84s eta=03:47:14 +2026-09-19 17:26:05 | INFO | train rank=0 | step=46350/50000 progress=92.7% loss=0.087949 lr=1.449867e-06 lr_backbone=1.449867e-06 elapsed=47:57:28 avg_step=3.72s recent_step=2.09s eta=03:46:35 +2026-09-19 17:26:31 | INFO | train rank=0 | step=46360/50000 progress=92.7% loss=0.085704 lr=1.441972e-06 lr_backbone=1.441972e-06 elapsed=47:57:54 avg_step=3.72s recent_step=2.59s eta=03:45:57 +2026-09-19 17:27:02 | INFO | train rank=0 | step=46370/50000 progress=92.7% loss=0.075944 lr=1.434098e-06 lr_backbone=1.434098e-06 elapsed=47:58:25 avg_step=3.72s recent_step=3.10s eta=03:45:19 +2026-09-19 17:27:59 | INFO | train rank=0 | step=46380/50000 progress=92.8% loss=0.084901 lr=1.426245e-06 lr_backbone=1.426245e-06 elapsed=47:59:22 avg_step=3.72s recent_step=5.68s eta=03:44:44 +2026-09-19 17:28:33 | INFO | train rank=0 | step=46390/50000 progress=92.8% loss=0.110768 lr=1.418413e-06 lr_backbone=1.418413e-06 elapsed=47:59:56 avg_step=3.72s recent_step=3.42s eta=03:44:06 +2026-09-19 17:29:08 | INFO | train rank=0 | step=46400/50000 progress=92.8% loss=0.096516 lr=1.410603e-06 lr_backbone=1.410603e-06 elapsed=48:00:32 avg_step=3.72s recent_step=3.55s eta=03:43:29 +2026-09-19 17:29:42 | INFO | train rank=0 | step=46410/50000 progress=92.8% loss=0.063942 lr=1.402814e-06 lr_backbone=1.402814e-06 elapsed=48:01:05 avg_step=3.72s recent_step=3.33s eta=03:42:51 +2026-09-19 17:30:28 | INFO | train rank=0 | step=46420/50000 progress=92.8% loss=0.081448 lr=1.395046e-06 lr_backbone=1.395046e-06 elapsed=48:01:51 avg_step=3.72s recent_step=4.65s eta=03:42:15 +2026-09-19 17:31:02 | INFO | train rank=0 | step=46430/50000 progress=92.9% loss=0.081906 lr=1.387300e-06 lr_backbone=1.387300e-06 elapsed=48:02:25 avg_step=3.72s recent_step=3.36s eta=03:41:37 +2026-09-19 17:31:32 | INFO | train rank=0 | step=46440/50000 progress=92.9% loss=0.098462 lr=1.379575e-06 lr_backbone=1.379575e-06 elapsed=48:02:56 avg_step=3.72s recent_step=3.06s eta=03:41:00 +2026-09-19 17:32:06 | INFO | train rank=0 | step=46450/50000 progress=92.9% loss=0.088108 lr=1.371871e-06 lr_backbone=1.371871e-06 elapsed=48:03:30 avg_step=3.72s recent_step=3.40s eta=03:40:22 +2026-09-19 17:32:57 | INFO | train rank=0 | step=46460/50000 progress=92.9% loss=0.131228 lr=1.364188e-06 lr_backbone=1.364188e-06 elapsed=48:04:20 avg_step=3.72s recent_step=5.06s eta=03:39:46 +2026-09-19 17:33:34 | INFO | train rank=0 | step=46470/50000 progress=92.9% loss=0.080970 lr=1.356527e-06 lr_backbone=1.356527e-06 elapsed=48:04:58 avg_step=3.72s recent_step=3.74s eta=03:39:09 +2026-09-19 17:34:07 | INFO | train rank=0 | step=46480/50000 progress=93.0% loss=0.096184 lr=1.348887e-06 lr_backbone=1.348887e-06 elapsed=48:05:31 avg_step=3.72s recent_step=3.31s eta=03:38:31 +2026-09-19 17:34:40 | INFO | train rank=0 | step=46490/50000 progress=93.0% loss=0.103171 lr=1.341268e-06 lr_backbone=1.341268e-06 elapsed=48:06:03 avg_step=3.72s recent_step=3.23s eta=03:37:53 +2026-09-19 17:35:31 | INFO | train rank=0 | step=46500/50000 progress=93.0% loss=0.123584 lr=1.333670e-06 lr_backbone=1.333670e-06 elapsed=48:06:54 avg_step=3.73s recent_step=5.10s eta=03:37:17 +2026-09-19 17:36:06 | INFO | train rank=0 | step=46510/50000 progress=93.0% loss=0.076212 lr=1.326094e-06 lr_backbone=1.326094e-06 elapsed=48:07:29 avg_step=3.73s recent_step=3.52s eta=03:36:40 +2026-09-19 17:36:37 | INFO | train rank=0 | step=46520/50000 progress=93.0% loss=0.124703 lr=1.318539e-06 lr_backbone=1.318539e-06 elapsed=48:08:00 avg_step=3.72s recent_step=3.09s eta=03:36:02 +2026-09-19 17:37:10 | INFO | train rank=0 | step=46530/50000 progress=93.1% loss=0.098696 lr=1.311005e-06 lr_backbone=1.311005e-06 elapsed=48:08:33 avg_step=3.72s recent_step=3.31s eta=03:35:24 +2026-09-19 17:38:04 | INFO | train rank=0 | step=46540/50000 progress=93.1% loss=0.090939 lr=1.303493e-06 lr_backbone=1.303493e-06 elapsed=48:09:27 avg_step=3.73s recent_step=5.41s eta=03:34:48 +2026-09-19 17:38:43 | INFO | train rank=0 | step=46550/50000 progress=93.1% loss=0.099174 lr=1.296002e-06 lr_backbone=1.296002e-06 elapsed=48:10:07 avg_step=3.73s recent_step=3.92s eta=03:34:11 +2026-09-19 17:39:15 | INFO | train rank=0 | step=46560/50000 progress=93.1% loss=0.076391 lr=1.288532e-06 lr_backbone=1.288532e-06 elapsed=48:10:38 avg_step=3.73s recent_step=3.14s eta=03:33:34 +2026-09-19 17:39:48 | INFO | train rank=0 | step=46570/50000 progress=93.1% loss=0.083071 lr=1.281084e-06 lr_backbone=1.281084e-06 elapsed=48:11:11 avg_step=3.72s recent_step=3.32s eta=03:32:56 +2026-09-19 17:40:36 | INFO | train rank=0 | step=46580/50000 progress=93.2% loss=0.091765 lr=1.273656e-06 lr_backbone=1.273656e-06 elapsed=48:11:59 avg_step=3.73s recent_step=4.78s eta=03:32:20 +2026-09-19 17:41:08 | INFO | train rank=0 | step=46590/50000 progress=93.2% loss=0.070442 lr=1.266251e-06 lr_backbone=1.266251e-06 elapsed=48:12:31 avg_step=3.73s recent_step=3.26s eta=03:31:42 +2026-09-19 17:41:34 | INFO | train rank=0 | step=46600/50000 progress=93.2% loss=0.068614 lr=1.258866e-06 lr_backbone=1.258866e-06 elapsed=48:12:58 avg_step=3.72s recent_step=2.61s eta=03:31:04 +2026-09-19 17:42:05 | INFO | train rank=0 | step=46610/50000 progress=93.2% loss=0.063665 lr=1.251503e-06 lr_backbone=1.251503e-06 elapsed=48:13:28 avg_step=3.72s recent_step=3.06s eta=03:30:26 +2026-09-19 17:42:53 | INFO | train rank=0 | step=46620/50000 progress=93.2% loss=0.074522 lr=1.244161e-06 lr_backbone=1.244161e-06 elapsed=48:14:17 avg_step=3.72s recent_step=4.85s eta=03:29:50 +2026-09-19 17:43:31 | INFO | train rank=0 | step=46630/50000 progress=93.3% loss=0.079931 lr=1.236840e-06 lr_backbone=1.236840e-06 elapsed=48:14:54 avg_step=3.72s recent_step=3.76s eta=03:29:13 +2026-09-19 17:43:58 | INFO | train rank=0 | step=46640/50000 progress=93.3% loss=0.093369 lr=1.229541e-06 lr_backbone=1.229541e-06 elapsed=48:15:22 avg_step=3.72s recent_step=2.72s eta=03:28:35 +2026-09-19 17:44:36 | INFO | train rank=0 | step=46650/50000 progress=93.3% loss=0.103233 lr=1.222263e-06 lr_backbone=1.222263e-06 elapsed=48:15:59 avg_step=3.72s recent_step=3.76s eta=03:27:57 +2026-09-19 17:45:33 | INFO | train rank=0 | step=46660/50000 progress=93.3% loss=0.070435 lr=1.215007e-06 lr_backbone=1.215007e-06 elapsed=48:16:56 avg_step=3.73s recent_step=5.71s eta=03:27:22 +2026-09-19 17:46:03 | INFO | train rank=0 | step=46670/50000 progress=93.3% loss=0.095014 lr=1.207772e-06 lr_backbone=1.207772e-06 elapsed=48:17:27 avg_step=3.73s recent_step=3.03s eta=03:26:44 +2026-09-19 17:46:32 | INFO | train rank=0 | step=46680/50000 progress=93.4% loss=0.079465 lr=1.200558e-06 lr_backbone=1.200558e-06 elapsed=48:17:55 avg_step=3.72s recent_step=2.83s eta=03:26:06 +2026-09-19 17:47:05 | INFO | train rank=0 | step=46690/50000 progress=93.4% loss=0.080168 lr=1.193365e-06 lr_backbone=1.193365e-06 elapsed=48:18:28 avg_step=3.72s recent_step=3.33s eta=03:25:28 +2026-09-19 17:48:07 | INFO | train rank=0 | step=46700/50000 progress=93.4% loss=0.066839 lr=1.186194e-06 lr_backbone=1.186194e-06 elapsed=48:19:30 avg_step=3.73s recent_step=6.19s eta=03:24:53 +2026-09-19 17:48:34 | INFO | train rank=0 | step=46710/50000 progress=93.4% loss=0.071133 lr=1.179044e-06 lr_backbone=1.179044e-06 elapsed=48:19:58 avg_step=3.73s recent_step=2.76s eta=03:24:15 +2026-09-19 17:49:10 | INFO | train rank=0 | step=46720/50000 progress=93.4% loss=0.083261 lr=1.171916e-06 lr_backbone=1.171916e-06 elapsed=48:20:33 avg_step=3.73s recent_step=3.56s eta=03:23:38 +2026-09-19 17:49:38 | INFO | train rank=0 | step=46730/50000 progress=93.5% loss=0.076151 lr=1.164809e-06 lr_backbone=1.164809e-06 elapsed=48:21:01 avg_step=3.72s recent_step=2.78s eta=03:23:00 +2026-09-19 17:50:27 | INFO | train rank=0 | step=46740/50000 progress=93.5% loss=0.083014 lr=1.157723e-06 lr_backbone=1.157723e-06 elapsed=48:21:50 avg_step=3.73s recent_step=4.91s eta=03:22:23 +2026-09-19 17:51:02 | INFO | train rank=0 | step=46750/50000 progress=93.5% loss=0.078350 lr=1.150658e-06 lr_backbone=1.150658e-06 elapsed=48:22:25 avg_step=3.73s recent_step=3.51s eta=03:21:46 +2026-09-19 17:51:33 | INFO | train rank=0 | step=46760/50000 progress=93.5% loss=0.064830 lr=1.143615e-06 lr_backbone=1.143615e-06 elapsed=48:22:57 avg_step=3.72s recent_step=3.15s eta=03:21:08 +2026-09-19 17:52:11 | INFO | train rank=0 | step=46770/50000 progress=93.5% loss=0.083105 lr=1.136594e-06 lr_backbone=1.136594e-06 elapsed=48:23:34 avg_step=3.72s recent_step=3.75s eta=03:20:31 +2026-09-19 17:52:58 | INFO | train rank=0 | step=46780/50000 progress=93.6% loss=0.084081 lr=1.129594e-06 lr_backbone=1.129594e-06 elapsed=48:24:21 avg_step=3.73s recent_step=4.67s eta=03:19:54 +2026-09-19 17:53:29 | INFO | train rank=0 | step=46790/50000 progress=93.6% loss=0.146578 lr=1.122615e-06 lr_backbone=1.122615e-06 elapsed=48:24:52 avg_step=3.73s recent_step=3.14s eta=03:19:17 +2026-09-19 17:54:04 | INFO | train rank=0 | step=46800/50000 progress=93.6% loss=0.069052 lr=1.115657e-06 lr_backbone=1.115657e-06 elapsed=48:25:27 avg_step=3.72s recent_step=3.47s eta=03:18:39 +2026-09-19 17:54:36 | INFO | train rank=0 | step=46810/50000 progress=93.6% loss=0.084280 lr=1.108721e-06 lr_backbone=1.108721e-06 elapsed=48:25:59 avg_step=3.72s recent_step=3.23s eta=03:18:02 +2026-09-19 17:55:24 | INFO | train rank=0 | step=46820/50000 progress=93.6% loss=0.072231 lr=1.101806e-06 lr_backbone=1.101806e-06 elapsed=48:26:47 avg_step=3.73s recent_step=4.75s eta=03:17:25 +2026-09-19 17:55:57 | INFO | train rank=0 | step=46830/50000 progress=93.7% loss=0.066460 lr=1.094913e-06 lr_backbone=1.094913e-06 elapsed=48:27:20 avg_step=3.72s recent_step=3.32s eta=03:16:48 +2026-09-19 17:56:24 | INFO | train rank=0 | step=46840/50000 progress=93.7% loss=0.088833 lr=1.088041e-06 lr_backbone=1.088041e-06 elapsed=48:27:47 avg_step=3.72s recent_step=2.74s eta=03:16:10 +2026-09-19 17:56:57 | INFO | train rank=0 | step=46850/50000 progress=93.7% loss=0.086565 lr=1.081190e-06 lr_backbone=1.081190e-06 elapsed=48:28:20 avg_step=3.72s recent_step=3.29s eta=03:15:32 +2026-09-19 17:57:47 | INFO | train rank=0 | step=46860/50000 progress=93.7% loss=0.066780 lr=1.074361e-06 lr_backbone=1.074361e-06 elapsed=48:29:10 avg_step=3.72s recent_step=4.96s eta=03:14:56 +2026-09-19 17:58:23 | INFO | train rank=0 | step=46870/50000 progress=93.7% loss=0.103200 lr=1.067554e-06 lr_backbone=1.067554e-06 elapsed=48:29:46 avg_step=3.72s recent_step=3.60s eta=03:14:18 +2026-09-19 17:58:56 | INFO | train rank=0 | step=46880/50000 progress=93.8% loss=0.084212 lr=1.060767e-06 lr_backbone=1.060767e-06 elapsed=48:30:20 avg_step=3.72s recent_step=3.38s eta=03:13:41 +2026-09-19 17:59:28 | INFO | train rank=0 | step=46890/50000 progress=93.8% loss=0.106749 lr=1.054002e-06 lr_backbone=1.054002e-06 elapsed=48:30:51 avg_step=3.72s recent_step=3.12s eta=03:13:03 +2026-09-19 18:00:10 | INFO | train rank=0 | step=46900/50000 progress=93.8% loss=0.084962 lr=1.047259e-06 lr_backbone=1.047259e-06 elapsed=48:31:33 avg_step=3.72s recent_step=4.20s eta=03:12:26 +2026-09-19 18:00:43 | INFO | train rank=0 | step=46910/50000 progress=93.8% loss=0.138450 lr=1.040537e-06 lr_backbone=1.040537e-06 elapsed=48:32:07 avg_step=3.72s recent_step=3.36s eta=03:11:49 +2026-09-19 18:01:13 | INFO | train rank=0 | step=46920/50000 progress=93.8% loss=0.098459 lr=1.033836e-06 lr_backbone=1.033836e-06 elapsed=48:32:37 avg_step=3.72s recent_step=3.00s eta=03:11:11 +2026-09-19 18:01:47 | INFO | train rank=0 | step=46930/50000 progress=93.9% loss=0.091128 lr=1.027157e-06 lr_backbone=1.027157e-06 elapsed=48:33:10 avg_step=3.72s recent_step=3.39s eta=03:10:34 +2026-09-19 18:02:42 | INFO | train rank=0 | step=46940/50000 progress=93.9% loss=0.083537 lr=1.020499e-06 lr_backbone=1.020499e-06 elapsed=48:34:05 avg_step=3.72s recent_step=5.49s eta=03:09:58 +2026-09-19 18:03:16 | INFO | train rank=0 | step=46950/50000 progress=93.9% loss=0.074884 lr=1.013862e-06 lr_backbone=1.013862e-06 elapsed=48:34:39 avg_step=3.72s recent_step=3.38s eta=03:09:20 +2026-09-19 18:03:47 | INFO | train rank=0 | step=46960/50000 progress=93.9% loss=0.085182 lr=1.007247e-06 lr_backbone=1.007247e-06 elapsed=48:35:10 avg_step=3.72s recent_step=3.12s eta=03:08:43 +2026-09-19 18:04:18 | INFO | train rank=0 | step=46970/50000 progress=93.9% loss=0.066608 lr=1.000654e-06 lr_backbone=1.000654e-06 elapsed=48:35:41 avg_step=3.72s recent_step=3.08s eta=03:08:05 +2026-09-19 18:05:15 | INFO | train rank=0 | step=46980/50000 progress=94.0% loss=0.080517 lr=9.940817e-07 lr_backbone=9.940817e-07 elapsed=48:36:39 avg_step=3.72s recent_step=5.75s eta=03:07:29 +2026-09-19 18:05:51 | INFO | train rank=0 | step=46990/50000 progress=94.0% loss=0.070525 lr=9.875310e-07 lr_backbone=9.875310e-07 elapsed=48:37:14 avg_step=3.72s recent_step=3.57s eta=03:06:52 +2026-09-19 18:06:21 | INFO | train rank=0 | step=47000/50000 progress=94.0% loss=0.083612 lr=9.810017e-07 lr_backbone=9.810017e-07 elapsed=48:37:44 avg_step=3.72s recent_step=2.94s eta=03:06:14 +2026-09-19 18:07:07 | INFO | train rank=0 | step=47010/50000 progress=94.0% loss=0.077656 lr=9.744939e-07 lr_backbone=9.744939e-07 elapsed=48:38:30 avg_step=3.72s recent_step=4.64s eta=03:05:37 +2026-09-19 18:07:47 | INFO | train rank=0 | step=47020/50000 progress=94.0% loss=0.091915 lr=9.680075e-07 lr_backbone=9.680075e-07 elapsed=48:39:10 avg_step=3.73s recent_step=4.00s eta=03:05:00 +2026-09-19 18:08:14 | INFO | train rank=0 | step=47030/50000 progress=94.1% loss=0.086482 lr=9.615426e-07 lr_backbone=9.615426e-07 elapsed=48:39:37 avg_step=3.72s recent_step=2.68s eta=03:04:22 +2026-09-19 18:08:43 | INFO | train rank=0 | step=47040/50000 progress=94.1% loss=0.078273 lr=9.550991e-07 lr_backbone=9.550991e-07 elapsed=48:40:06 avg_step=3.72s recent_step=2.92s eta=03:03:44 +2026-09-19 18:09:34 | INFO | train rank=0 | step=47050/50000 progress=94.1% loss=0.081219 lr=9.486771e-07 lr_backbone=9.486771e-07 elapsed=48:40:57 avg_step=3.72s recent_step=5.10s eta=03:03:08 +2026-09-19 18:10:04 | INFO | train rank=0 | step=47060/50000 progress=94.1% loss=0.131233 lr=9.422765e-07 lr_backbone=9.422765e-07 elapsed=48:41:27 avg_step=3.72s recent_step=3.01s eta=03:02:30 +2026-09-19 18:10:39 | INFO | train rank=0 | step=47070/50000 progress=94.1% loss=0.107602 lr=9.358974e-07 lr_backbone=9.358974e-07 elapsed=48:42:02 avg_step=3.72s recent_step=3.51s eta=03:01:53 +2026-09-19 18:11:09 | INFO | train rank=0 | step=47080/50000 progress=94.2% loss=0.152300 lr=9.295398e-07 lr_backbone=9.295398e-07 elapsed=48:42:33 avg_step=3.72s recent_step=3.04s eta=03:01:15 +2026-09-19 18:11:59 | INFO | train rank=0 | step=47090/50000 progress=94.2% loss=0.087310 lr=9.232036e-07 lr_backbone=9.232036e-07 elapsed=48:43:22 avg_step=3.72s recent_step=4.95s eta=03:00:39 +2026-09-19 18:12:45 | INFO | train rank=0 | step=47100/50000 progress=94.2% loss=0.083591 lr=9.168889e-07 lr_backbone=9.168889e-07 elapsed=48:44:08 avg_step=3.73s recent_step=4.61s eta=03:00:02 +2026-09-19 18:13:15 | INFO | train rank=0 | step=47110/50000 progress=94.2% loss=0.098782 lr=9.105957e-07 lr_backbone=9.105957e-07 elapsed=48:44:38 avg_step=3.72s recent_step=3.01s eta=02:59:24 +2026-09-19 18:13:46 | INFO | train rank=0 | step=47120/50000 progress=94.2% loss=0.117076 lr=9.043239e-07 lr_backbone=9.043239e-07 elapsed=48:45:09 avg_step=3.72s recent_step=3.05s eta=02:58:47 +2026-09-19 18:14:22 | INFO | train rank=0 | step=47130/50000 progress=94.3% loss=0.179449 lr=8.980736e-07 lr_backbone=8.980736e-07 elapsed=48:45:45 avg_step=3.72s recent_step=3.60s eta=02:58:09 +2026-09-19 18:15:11 | INFO | train rank=0 | step=47140/50000 progress=94.3% loss=0.076381 lr=8.918448e-07 lr_backbone=8.918448e-07 elapsed=48:46:34 avg_step=3.72s recent_step=4.93s eta=02:57:33 +2026-09-19 18:15:46 | INFO | train rank=0 | step=47150/50000 progress=94.3% loss=0.083222 lr=8.856375e-07 lr_backbone=8.856375e-07 elapsed=48:47:10 avg_step=3.72s recent_step=3.55s eta=02:56:56 +2026-09-19 18:16:22 | INFO | train rank=0 | step=47160/50000 progress=94.3% loss=0.093466 lr=8.794516e-07 lr_backbone=8.794516e-07 elapsed=48:47:45 avg_step=3.72s recent_step=3.52s eta=02:56:18 +2026-09-19 18:16:55 | INFO | train rank=0 | step=47170/50000 progress=94.3% loss=0.077705 lr=8.732873e-07 lr_backbone=8.732873e-07 elapsed=48:48:19 avg_step=3.72s recent_step=3.37s eta=02:55:41 +2026-09-19 18:17:45 | INFO | train rank=0 | step=47180/50000 progress=94.4% loss=0.098413 lr=8.671444e-07 lr_backbone=8.671444e-07 elapsed=48:49:08 avg_step=3.73s recent_step=4.93s eta=02:55:04 +2026-09-19 18:18:22 | INFO | train rank=0 | step=47190/50000 progress=94.4% loss=0.075271 lr=8.610230e-07 lr_backbone=8.610230e-07 elapsed=48:49:45 avg_step=3.73s recent_step=3.71s eta=02:54:27 +2026-09-19 18:18:53 | INFO | train rank=0 | step=47200/50000 progress=94.4% loss=0.110245 lr=8.549231e-07 lr_backbone=8.549231e-07 elapsed=48:50:16 avg_step=3.72s recent_step=3.11s eta=02:53:49 +2026-09-19 18:19:30 | INFO | train rank=0 | step=47210/50000 progress=94.4% loss=0.089212 lr=8.488448e-07 lr_backbone=8.488448e-07 elapsed=48:50:54 avg_step=3.72s recent_step=3.74s eta=02:53:12 +2026-09-19 18:20:20 | INFO | train rank=0 | step=47220/50000 progress=94.4% loss=0.075584 lr=8.427879e-07 lr_backbone=8.427879e-07 elapsed=48:51:43 avg_step=3.73s recent_step=4.96s eta=02:52:36 +2026-09-19 18:20:56 | INFO | train rank=0 | step=47230/50000 progress=94.5% loss=0.087927 lr=8.367525e-07 lr_backbone=8.367525e-07 elapsed=48:52:20 avg_step=3.73s recent_step=3.66s eta=02:51:58 +2026-09-19 18:21:26 | INFO | train rank=0 | step=47240/50000 progress=94.5% loss=0.093766 lr=8.307386e-07 lr_backbone=8.307386e-07 elapsed=48:52:50 avg_step=3.73s recent_step=2.98s eta=02:51:21 +2026-09-19 18:21:56 | INFO | train rank=0 | step=47250/50000 progress=94.5% loss=0.071002 lr=8.247463e-07 lr_backbone=8.247463e-07 elapsed=48:53:20 avg_step=3.72s recent_step=2.99s eta=02:50:43 +2026-09-19 18:22:49 | INFO | train rank=0 | step=47260/50000 progress=94.5% loss=0.092532 lr=8.187754e-07 lr_backbone=8.187754e-07 elapsed=48:54:12 avg_step=3.73s recent_step=5.25s eta=02:50:07 +2026-09-19 18:23:24 | INFO | train rank=0 | step=47270/50000 progress=94.5% loss=0.098657 lr=8.128261e-07 lr_backbone=8.128261e-07 elapsed=48:54:47 avg_step=3.73s recent_step=3.50s eta=02:49:29 +2026-09-19 18:23:59 | INFO | train rank=0 | step=47280/50000 progress=94.6% loss=0.121196 lr=8.068982e-07 lr_backbone=8.068982e-07 elapsed=48:55:22 avg_step=3.73s recent_step=3.48s eta=02:48:52 +2026-09-19 18:24:36 | INFO | train rank=0 | step=47290/50000 progress=94.6% loss=0.073414 lr=8.009919e-07 lr_backbone=8.009919e-07 elapsed=48:56:00 avg_step=3.73s recent_step=3.77s eta=02:48:15 +2026-09-19 18:25:34 | INFO | train rank=0 | step=47300/50000 progress=94.6% loss=0.070084 lr=7.951071e-07 lr_backbone=7.951071e-07 elapsed=48:56:57 avg_step=3.73s recent_step=5.73s eta=02:47:38 +2026-09-19 18:26:03 | INFO | train rank=0 | step=47310/50000 progress=94.6% loss=0.094520 lr=7.892439e-07 lr_backbone=7.892439e-07 elapsed=48:57:27 avg_step=3.73s recent_step=2.97s eta=02:47:01 +2026-09-19 18:26:28 | INFO | train rank=0 | step=47320/50000 progress=94.6% loss=0.073457 lr=7.834022e-07 lr_backbone=7.834022e-07 elapsed=48:57:51 avg_step=3.73s recent_step=2.45s eta=02:46:23 +2026-09-19 18:26:59 | INFO | train rank=0 | step=47330/50000 progress=94.7% loss=0.087524 lr=7.775819e-07 lr_backbone=7.775819e-07 elapsed=48:58:23 avg_step=3.72s recent_step=3.16s eta=02:45:45 +2026-09-19 18:27:53 | INFO | train rank=0 | step=47340/50000 progress=94.7% loss=0.084969 lr=7.717833e-07 lr_backbone=7.717833e-07 elapsed=48:59:16 avg_step=3.73s recent_step=5.34s eta=02:45:09 +2026-09-19 18:28:23 | INFO | train rank=0 | step=47350/50000 progress=94.7% loss=0.087799 lr=7.660061e-07 lr_backbone=7.660061e-07 elapsed=48:59:46 avg_step=3.73s recent_step=2.98s eta=02:44:31 +2026-09-19 18:28:57 | INFO | train rank=0 | step=47360/50000 progress=94.7% loss=0.091984 lr=7.602505e-07 lr_backbone=7.602505e-07 elapsed=49:00:21 avg_step=3.73s recent_step=3.48s eta=02:43:54 +2026-09-19 18:29:32 | INFO | train rank=0 | step=47370/50000 progress=94.7% loss=0.079266 lr=7.545165e-07 lr_backbone=7.545165e-07 elapsed=49:00:55 avg_step=3.73s recent_step=3.41s eta=02:43:16 +2026-09-19 18:30:23 | INFO | train rank=0 | step=47380/50000 progress=94.8% loss=0.090766 lr=7.488039e-07 lr_backbone=7.488039e-07 elapsed=49:01:46 avg_step=3.73s recent_step=5.13s eta=02:42:40 +2026-09-19 18:31:00 | INFO | train rank=0 | step=47390/50000 progress=94.8% loss=0.078617 lr=7.431130e-07 lr_backbone=7.431130e-07 elapsed=49:02:23 avg_step=3.73s recent_step=3.72s eta=02:42:03 +2026-09-19 18:31:30 | INFO | train rank=0 | step=47400/50000 progress=94.8% loss=0.059674 lr=7.374435e-07 lr_backbone=7.374435e-07 elapsed=49:02:54 avg_step=3.73s recent_step=3.04s eta=02:41:25 +2026-09-19 18:31:58 | INFO | train rank=0 | step=47410/50000 progress=94.8% loss=0.096181 lr=7.317957e-07 lr_backbone=7.317957e-07 elapsed=49:03:21 avg_step=3.72s recent_step=2.76s eta=02:40:47 +2026-09-19 18:32:56 | INFO | train rank=0 | step=47420/50000 progress=94.8% loss=0.079662 lr=7.261693e-07 lr_backbone=7.261693e-07 elapsed=49:04:19 avg_step=3.73s recent_step=5.75s eta=02:40:11 +2026-09-19 18:33:29 | INFO | train rank=0 | step=47430/50000 progress=94.9% loss=0.096188 lr=7.205645e-07 lr_backbone=7.205645e-07 elapsed=49:04:52 avg_step=3.73s recent_step=3.36s eta=02:39:34 +2026-09-19 18:33:58 | INFO | train rank=0 | step=47440/50000 progress=94.9% loss=0.075249 lr=7.149813e-07 lr_backbone=7.149813e-07 elapsed=49:05:21 avg_step=3.73s recent_step=2.85s eta=02:38:56 +2026-09-19 18:34:26 | INFO | train rank=0 | step=47450/50000 progress=94.9% loss=0.103721 lr=7.094197e-07 lr_backbone=7.094197e-07 elapsed=49:05:49 avg_step=3.72s recent_step=2.83s eta=02:38:18 +2026-09-19 18:35:19 | INFO | train rank=0 | step=47460/50000 progress=94.9% loss=0.081293 lr=7.038796e-07 lr_backbone=7.038796e-07 elapsed=49:06:43 avg_step=3.73s recent_step=5.34s eta=02:37:42 +2026-09-19 18:35:51 | INFO | train rank=0 | step=47470/50000 progress=94.9% loss=0.092260 lr=6.983610e-07 lr_backbone=6.983610e-07 elapsed=49:07:14 avg_step=3.73s recent_step=3.19s eta=02:37:04 +2026-09-19 18:36:27 | INFO | train rank=0 | step=47480/50000 progress=95.0% loss=0.088761 lr=6.928641e-07 lr_backbone=6.928641e-07 elapsed=49:07:50 avg_step=3.73s recent_step=3.60s eta=02:36:27 +2026-09-19 18:37:04 | INFO | train rank=0 | step=47490/50000 progress=95.0% loss=0.078952 lr=6.873887e-07 lr_backbone=6.873887e-07 elapsed=49:08:27 avg_step=3.73s recent_step=3.64s eta=02:35:50 +2026-09-19 18:37:54 | INFO | train rank=0 | step=47500/50000 progress=95.0% loss=0.088374 lr=6.819348e-07 lr_backbone=6.819348e-07 elapsed=49:09:17 avg_step=3.73s recent_step=5.03s eta=02:35:13 +2026-09-19 18:38:26 | INFO | train rank=0 | step=47510/50000 progress=95.0% loss=0.070924 lr=6.765026e-07 lr_backbone=6.765026e-07 elapsed=49:09:49 avg_step=3.73s recent_step=3.19s eta=02:34:36 +2026-09-19 18:39:01 | INFO | train rank=0 | step=47520/50000 progress=95.0% loss=0.108494 lr=6.710919e-07 lr_backbone=6.710919e-07 elapsed=49:10:24 avg_step=3.73s recent_step=3.54s eta=02:33:58 +2026-09-19 18:39:23 | INFO | train rank=0 | step=47530/50000 progress=95.1% loss=0.071751 lr=6.657028e-07 lr_backbone=6.657028e-07 elapsed=49:10:46 avg_step=3.72s recent_step=2.19s eta=02:33:20 +2026-09-19 18:40:22 | INFO | train rank=0 | step=47540/50000 progress=95.1% loss=0.090940 lr=6.603353e-07 lr_backbone=6.603353e-07 elapsed=49:11:46 avg_step=3.73s recent_step=5.92s eta=02:32:44 +2026-09-19 18:40:54 | INFO | train rank=0 | step=47550/50000 progress=95.1% loss=0.070581 lr=6.549893e-07 lr_backbone=6.549893e-07 elapsed=49:12:17 avg_step=3.73s recent_step=3.14s eta=02:32:06 +2026-09-19 18:41:30 | INFO | train rank=0 | step=47560/50000 progress=95.1% loss=0.067279 lr=6.496650e-07 lr_backbone=6.496650e-07 elapsed=49:12:53 avg_step=3.73s recent_step=3.59s eta=02:31:29 +2026-09-19 18:41:58 | INFO | train rank=0 | step=47570/50000 progress=95.1% loss=0.092645 lr=6.443622e-07 lr_backbone=6.443622e-07 elapsed=49:13:22 avg_step=3.73s recent_step=2.88s eta=02:30:51 +2026-09-19 18:42:56 | INFO | train rank=0 | step=47580/50000 progress=95.2% loss=0.096216 lr=6.390810e-07 lr_backbone=6.390810e-07 elapsed=49:14:19 avg_step=3.73s recent_step=5.77s eta=02:30:15 +2026-09-19 18:43:28 | INFO | train rank=0 | step=47590/50000 progress=95.2% loss=0.088109 lr=6.338214e-07 lr_backbone=6.338214e-07 elapsed=49:14:51 avg_step=3.73s recent_step=3.15s eta=02:29:38 +2026-09-19 18:43:55 | INFO | train rank=0 | step=47600/50000 progress=95.2% loss=0.082631 lr=6.285835e-07 lr_backbone=6.285835e-07 elapsed=49:15:19 avg_step=3.73s recent_step=2.77s eta=02:29:00 +2026-09-19 18:44:23 | INFO | train rank=0 | step=47610/50000 progress=95.2% loss=0.085232 lr=6.233671e-07 lr_backbone=6.233671e-07 elapsed=49:15:46 avg_step=3.72s recent_step=2.76s eta=02:28:22 +2026-09-19 18:45:21 | INFO | train rank=0 | step=47620/50000 progress=95.2% loss=0.092655 lr=6.181723e-07 lr_backbone=6.181723e-07 elapsed=49:16:45 avg_step=3.73s recent_step=5.83s eta=02:27:46 +2026-09-19 18:45:49 | INFO | train rank=0 | step=47630/50000 progress=95.3% loss=0.077836 lr=6.129991e-07 lr_backbone=6.129991e-07 elapsed=49:17:12 avg_step=3.73s recent_step=2.74s eta=02:27:08 +2026-09-19 18:46:25 | INFO | train rank=0 | step=47640/50000 progress=95.3% loss=0.079241 lr=6.078475e-07 lr_backbone=6.078475e-07 elapsed=49:17:49 avg_step=3.73s recent_step=3.67s eta=02:26:31 +2026-09-19 18:46:54 | INFO | train rank=0 | step=47650/50000 progress=95.3% loss=0.087030 lr=6.027175e-07 lr_backbone=6.027175e-07 elapsed=49:18:17 avg_step=3.73s recent_step=2.81s eta=02:25:53 +2026-09-19 18:47:46 | INFO | train rank=0 | step=47660/50000 progress=95.3% loss=0.076148 lr=5.976091e-07 lr_backbone=5.976091e-07 elapsed=49:19:09 avg_step=3.73s recent_step=5.20s eta=02:25:17 +2026-09-19 18:48:15 | INFO | train rank=0 | step=47670/50000 progress=95.3% loss=0.081476 lr=5.925224e-07 lr_backbone=5.925224e-07 elapsed=49:19:38 avg_step=3.73s recent_step=2.92s eta=02:24:39 +2026-09-19 18:48:50 | INFO | train rank=0 | step=47680/50000 progress=95.4% loss=0.074816 lr=5.874572e-07 lr_backbone=5.874572e-07 elapsed=49:20:13 avg_step=3.73s recent_step=3.55s eta=02:24:02 +2026-09-19 18:49:29 | INFO | train rank=0 | step=47690/50000 progress=95.4% loss=0.067607 lr=5.824137e-07 lr_backbone=5.824137e-07 elapsed=49:20:52 avg_step=3.73s recent_step=3.85s eta=02:23:25 +2026-09-19 18:50:17 | INFO | train rank=0 | step=47700/50000 progress=95.4% loss=0.073792 lr=5.773917e-07 lr_backbone=5.773917e-07 elapsed=49:21:41 avg_step=3.73s recent_step=4.88s eta=02:22:48 +2026-09-19 18:50:49 | INFO | train rank=0 | step=47710/50000 progress=95.4% loss=0.070739 lr=5.723914e-07 lr_backbone=5.723914e-07 elapsed=49:22:12 avg_step=3.73s recent_step=3.15s eta=02:22:10 +2026-09-19 18:51:23 | INFO | train rank=0 | step=47720/50000 progress=95.4% loss=0.078501 lr=5.674128e-07 lr_backbone=5.674128e-07 elapsed=49:22:46 avg_step=3.73s recent_step=3.39s eta=02:21:33 +2026-09-19 18:51:56 | INFO | train rank=0 | step=47730/50000 progress=95.5% loss=0.081385 lr=5.624557e-07 lr_backbone=5.624557e-07 elapsed=49:23:19 avg_step=3.73s recent_step=3.33s eta=02:20:56 +2026-09-19 18:52:50 | INFO | train rank=0 | step=47740/50000 progress=95.5% loss=0.106292 lr=5.575203e-07 lr_backbone=5.575203e-07 elapsed=49:24:13 avg_step=3.73s recent_step=5.41s eta=02:20:19 +2026-09-19 18:53:16 | INFO | train rank=0 | step=47750/50000 progress=95.5% loss=0.071076 lr=5.526065e-07 lr_backbone=5.526065e-07 elapsed=49:24:40 avg_step=3.73s recent_step=2.61s eta=02:19:41 +2026-09-19 18:53:50 | INFO | train rank=0 | step=47760/50000 progress=95.5% loss=0.098840 lr=5.477143e-07 lr_backbone=5.477143e-07 elapsed=49:25:13 avg_step=3.73s recent_step=3.34s eta=02:19:04 +2026-09-19 18:54:28 | INFO | train rank=0 | step=47770/50000 progress=95.5% loss=0.097090 lr=5.428438e-07 lr_backbone=5.428438e-07 elapsed=49:25:51 avg_step=3.73s recent_step=3.83s eta=02:18:27 +2026-09-19 18:55:16 | INFO | train rank=0 | step=47780/50000 progress=95.6% loss=0.080447 lr=5.379948e-07 lr_backbone=5.379948e-07 elapsed=49:26:39 avg_step=3.73s recent_step=4.81s eta=02:17:50 +2026-09-19 18:55:48 | INFO | train rank=0 | step=47790/50000 progress=95.6% loss=0.122714 lr=5.331676e-07 lr_backbone=5.331676e-07 elapsed=49:27:12 avg_step=3.73s recent_step=3.23s eta=02:17:12 +2026-09-19 18:56:15 | INFO | train rank=0 | step=47800/50000 progress=95.6% loss=0.069525 lr=5.283619e-07 lr_backbone=5.283619e-07 elapsed=49:27:39 avg_step=3.73s recent_step=2.69s eta=02:16:35 +2026-09-19 18:56:44 | INFO | train rank=0 | step=47810/50000 progress=95.6% loss=0.079952 lr=5.235780e-07 lr_backbone=5.235780e-07 elapsed=49:28:08 avg_step=3.72s recent_step=2.91s eta=02:15:57 +2026-09-19 18:57:45 | INFO | train rank=0 | step=47820/50000 progress=95.6% loss=0.073530 lr=5.188156e-07 lr_backbone=5.188156e-07 elapsed=49:29:09 avg_step=3.73s recent_step=6.09s eta=02:15:21 +2026-09-19 18:58:19 | INFO | train rank=0 | step=47830/50000 progress=95.7% loss=0.076014 lr=5.140749e-07 lr_backbone=5.140749e-07 elapsed=49:29:42 avg_step=3.73s recent_step=3.32s eta=02:14:43 +2026-09-19 18:58:55 | INFO | train rank=0 | step=47840/50000 progress=95.7% loss=0.091560 lr=5.093558e-07 lr_backbone=5.093558e-07 elapsed=49:30:18 avg_step=3.73s recent_step=3.62s eta=02:14:06 +2026-09-19 18:59:23 | INFO | train rank=0 | step=47850/50000 progress=95.7% loss=0.085444 lr=5.046584e-07 lr_backbone=5.046584e-07 elapsed=49:30:46 avg_step=3.73s recent_step=2.82s eta=02:13:28 +2026-09-19 19:00:07 | INFO | train rank=0 | step=47860/50000 progress=95.7% loss=0.091163 lr=4.999827e-07 lr_backbone=4.999827e-07 elapsed=49:31:30 avg_step=3.73s recent_step=4.39s eta=02:12:52 +2026-09-19 19:00:44 | INFO | train rank=0 | step=47870/50000 progress=95.7% loss=0.084933 lr=4.953286e-07 lr_backbone=4.953286e-07 elapsed=49:32:07 avg_step=3.73s recent_step=3.69s eta=02:12:14 +2026-09-19 19:01:18 | INFO | train rank=0 | step=47880/50000 progress=95.8% loss=0.101039 lr=4.906961e-07 lr_backbone=4.906961e-07 elapsed=49:32:41 avg_step=3.73s recent_step=3.44s eta=02:11:37 +2026-09-19 19:01:51 | INFO | train rank=0 | step=47890/50000 progress=95.8% loss=0.097863 lr=4.860853e-07 lr_backbone=4.860853e-07 elapsed=49:33:14 avg_step=3.73s recent_step=3.30s eta=02:10:59 +2026-09-19 19:02:42 | INFO | train rank=0 | step=47900/50000 progress=95.8% loss=0.072387 lr=4.814962e-07 lr_backbone=4.814962e-07 elapsed=49:34:05 avg_step=3.73s recent_step=5.12s eta=02:10:23 +2026-09-19 19:03:12 | INFO | train rank=0 | step=47910/50000 progress=95.8% loss=0.132177 lr=4.769287e-07 lr_backbone=4.769287e-07 elapsed=49:34:35 avg_step=3.73s recent_step=2.98s eta=02:09:45 +2026-09-19 19:03:49 | INFO | train rank=0 | step=47920/50000 progress=95.8% loss=0.070727 lr=4.723829e-07 lr_backbone=4.723829e-07 elapsed=49:35:12 avg_step=3.73s recent_step=3.66s eta=02:09:08 +2026-09-19 19:04:21 | INFO | train rank=0 | step=47930/50000 progress=95.9% loss=0.077853 lr=4.678588e-07 lr_backbone=4.678588e-07 elapsed=49:35:44 avg_step=3.73s recent_step=3.21s eta=02:08:30 +2026-09-19 19:05:11 | INFO | train rank=0 | step=47940/50000 progress=95.9% loss=0.118379 lr=4.633563e-07 lr_backbone=4.633563e-07 elapsed=49:36:35 avg_step=3.73s recent_step=5.06s eta=02:07:54 +2026-09-19 19:05:43 | INFO | train rank=0 | step=47950/50000 progress=95.9% loss=0.087117 lr=4.588755e-07 lr_backbone=4.588755e-07 elapsed=49:37:06 avg_step=3.73s recent_step=3.19s eta=02:07:16 +2026-09-19 19:06:15 | INFO | train rank=0 | step=47960/50000 progress=95.9% loss=0.072999 lr=4.544163e-07 lr_backbone=4.544163e-07 elapsed=49:37:38 avg_step=3.73s recent_step=3.15s eta=02:06:39 +2026-09-19 19:06:53 | INFO | train rank=0 | step=47970/50000 progress=95.9% loss=0.084063 lr=4.499789e-07 lr_backbone=4.499789e-07 elapsed=49:38:16 avg_step=3.73s recent_step=3.81s eta=02:06:02 +2026-09-19 19:07:45 | INFO | train rank=0 | step=47980/50000 progress=96.0% loss=0.080604 lr=4.455631e-07 lr_backbone=4.455631e-07 elapsed=49:39:08 avg_step=3.73s recent_step=5.24s eta=02:05:25 +2026-09-19 19:08:08 | INFO | train rank=0 | step=47990/50000 progress=96.0% loss=0.103510 lr=4.411690e-07 lr_backbone=4.411690e-07 elapsed=49:39:32 avg_step=3.73s recent_step=2.31s eta=02:04:47 +2026-09-19 19:08:36 | INFO | train rank=0 | step=48000/50000 progress=96.0% loss=0.068963 lr=4.367965e-07 lr_backbone=4.367965e-07 elapsed=49:40:00 avg_step=3.73s recent_step=2.81s eta=02:04:10 +2026-09-19 19:09:10 | INFO | train rank=0 | step=48010/50000 progress=96.0% loss=0.087120 lr=4.324458e-07 lr_backbone=4.324458e-07 elapsed=49:40:33 avg_step=3.72s recent_step=3.36s eta=02:03:32 +2026-09-19 19:10:04 | INFO | train rank=0 | step=48020/50000 progress=96.0% loss=0.063465 lr=4.281167e-07 lr_backbone=4.281167e-07 elapsed=49:41:27 avg_step=3.73s recent_step=5.37s eta=02:02:56 +2026-09-19 19:10:32 | INFO | train rank=0 | step=48030/50000 progress=96.1% loss=0.080773 lr=4.238093e-07 lr_backbone=4.238093e-07 elapsed=49:41:55 avg_step=3.73s recent_step=2.79s eta=02:02:18 +2026-09-19 19:11:07 | INFO | train rank=0 | step=48040/50000 progress=96.1% loss=0.079545 lr=4.195236e-07 lr_backbone=4.195236e-07 elapsed=49:42:30 avg_step=3.73s recent_step=3.51s eta=02:01:41 +2026-09-19 19:11:39 | INFO | train rank=0 | step=48050/50000 progress=96.1% loss=0.073160 lr=4.152596e-07 lr_backbone=4.152596e-07 elapsed=49:43:02 avg_step=3.72s recent_step=3.24s eta=02:01:03 +2026-09-19 19:12:30 | INFO | train rank=0 | step=48060/50000 progress=96.1% loss=0.075624 lr=4.110173e-07 lr_backbone=4.110173e-07 elapsed=49:43:53 avg_step=3.73s recent_step=5.08s eta=02:00:26 +2026-09-19 19:12:51 | INFO | train rank=0 | step=48070/50000 progress=96.1% loss=0.075693 lr=4.067967e-07 lr_backbone=4.067967e-07 elapsed=49:44:14 avg_step=3.72s recent_step=2.09s eta=01:59:49 +2026-09-19 19:13:21 | INFO | train rank=0 | step=48080/50000 progress=96.2% loss=0.111650 lr=4.025977e-07 lr_backbone=4.025977e-07 elapsed=49:44:45 avg_step=3.72s recent_step=3.04s eta=01:59:11 +2026-09-19 19:13:55 | INFO | train rank=0 | step=48090/50000 progress=96.2% loss=0.104313 lr=3.984205e-07 lr_backbone=3.984205e-07 elapsed=49:45:18 avg_step=3.72s recent_step=3.37s eta=01:58:34 +2026-09-19 19:14:42 | INFO | train rank=0 | step=48100/50000 progress=96.2% loss=0.101363 lr=3.942649e-07 lr_backbone=3.942649e-07 elapsed=49:46:05 avg_step=3.72s recent_step=4.68s eta=01:57:57 +2026-09-19 19:15:12 | INFO | train rank=0 | step=48110/50000 progress=96.2% loss=0.065231 lr=3.901311e-07 lr_backbone=3.901311e-07 elapsed=49:46:35 avg_step=3.72s recent_step=3.01s eta=01:57:19 +2026-09-19 19:15:45 | INFO | train rank=0 | step=48120/50000 progress=96.2% loss=0.065855 lr=3.860189e-07 lr_backbone=3.860189e-07 elapsed=49:47:09 avg_step=3.72s recent_step=3.36s eta=01:56:42 +2026-09-19 19:16:19 | INFO | train rank=0 | step=48130/50000 progress=96.3% loss=0.073740 lr=3.819285e-07 lr_backbone=3.819285e-07 elapsed=49:47:43 avg_step=3.72s recent_step=3.39s eta=01:56:04 +2026-09-19 19:17:09 | INFO | train rank=0 | step=48140/50000 progress=96.3% loss=0.064281 lr=3.778598e-07 lr_backbone=3.778598e-07 elapsed=49:48:32 avg_step=3.72s recent_step=4.96s eta=01:55:28 +2026-09-19 19:17:42 | INFO | train rank=0 | step=48150/50000 progress=96.3% loss=0.074890 lr=3.738127e-07 lr_backbone=3.738127e-07 elapsed=49:49:05 avg_step=3.72s recent_step=3.33s eta=01:54:50 +2026-09-19 19:18:19 | INFO | train rank=0 | step=48160/50000 progress=96.3% loss=0.087150 lr=3.697874e-07 lr_backbone=3.697874e-07 elapsed=49:49:42 avg_step=3.72s recent_step=3.68s eta=01:54:13 +2026-09-19 19:18:53 | INFO | train rank=0 | step=48170/50000 progress=96.3% loss=0.075355 lr=3.657838e-07 lr_backbone=3.657838e-07 elapsed=49:50:17 avg_step=3.72s recent_step=3.45s eta=01:53:36 +2026-09-19 19:19:37 | INFO | train rank=0 | step=48180/50000 progress=96.4% loss=0.086325 lr=3.618019e-07 lr_backbone=3.618019e-07 elapsed=49:51:01 avg_step=3.72s recent_step=4.40s eta=01:52:59 +2026-09-19 19:20:05 | INFO | train rank=0 | step=48190/50000 progress=96.4% loss=0.074164 lr=3.578417e-07 lr_backbone=3.578417e-07 elapsed=49:51:29 avg_step=3.72s recent_step=2.79s eta=01:52:21 +2026-09-19 19:20:36 | INFO | train rank=0 | step=48200/50000 progress=96.4% loss=0.093987 lr=3.539033e-07 lr_backbone=3.539033e-07 elapsed=49:51:59 avg_step=3.72s recent_step=3.04s eta=01:51:44 +2026-09-19 19:21:05 | INFO | train rank=0 | step=48210/50000 progress=96.4% loss=0.079967 lr=3.499865e-07 lr_backbone=3.499865e-07 elapsed=49:52:29 avg_step=3.72s recent_step=2.95s eta=01:51:06 +2026-09-19 19:21:47 | INFO | train rank=0 | step=48220/50000 progress=96.4% loss=0.102734 lr=3.460915e-07 lr_backbone=3.460915e-07 elapsed=49:53:10 avg_step=3.72s recent_step=4.13s eta=01:50:29 +2026-09-19 19:22:25 | INFO | train rank=0 | step=48230/50000 progress=96.5% loss=0.075174 lr=3.422181e-07 lr_backbone=3.422181e-07 elapsed=49:53:49 avg_step=3.72s recent_step=3.88s eta=01:49:52 +2026-09-19 19:22:55 | INFO | train rank=0 | step=48240/50000 progress=96.5% loss=0.137553 lr=3.383666e-07 lr_backbone=3.383666e-07 elapsed=49:54:19 avg_step=3.72s recent_step=3.01s eta=01:49:14 +2026-09-19 19:23:26 | INFO | train rank=0 | step=48250/50000 progress=96.5% loss=0.100832 lr=3.345367e-07 lr_backbone=3.345367e-07 elapsed=49:54:49 avg_step=3.72s recent_step=3.04s eta=01:48:37 +2026-09-19 19:24:19 | INFO | train rank=0 | step=48260/50000 progress=96.5% loss=0.073016 lr=3.307285e-07 lr_backbone=3.307285e-07 elapsed=49:55:42 avg_step=3.72s recent_step=5.33s eta=01:48:00 +2026-09-19 19:24:51 | INFO | train rank=0 | step=48270/50000 progress=96.5% loss=0.087326 lr=3.269421e-07 lr_backbone=3.269421e-07 elapsed=49:56:15 avg_step=3.72s recent_step=3.24s eta=01:47:23 +2026-09-19 19:25:18 | INFO | train rank=0 | step=48280/50000 progress=96.6% loss=0.109323 lr=3.231774e-07 lr_backbone=3.231774e-07 elapsed=49:56:42 avg_step=3.72s recent_step=2.70s eta=01:46:45 +2026-09-19 19:26:07 | INFO | train rank=0 | step=48290/50000 progress=96.6% loss=0.097457 lr=3.194345e-07 lr_backbone=3.194345e-07 elapsed=49:57:30 avg_step=3.72s recent_step=4.85s eta=01:46:08 +2026-09-19 19:26:53 | INFO | train rank=0 | step=48300/50000 progress=96.6% loss=0.134570 lr=3.157132e-07 lr_backbone=3.157132e-07 elapsed=49:58:16 avg_step=3.72s recent_step=4.61s eta=01:45:31 +2026-09-19 19:27:31 | INFO | train rank=0 | step=48310/50000 progress=96.6% loss=0.071690 lr=3.120138e-07 lr_backbone=3.120138e-07 elapsed=49:58:54 avg_step=3.72s recent_step=3.78s eta=01:44:54 +2026-09-19 19:28:02 | INFO | train rank=0 | step=48320/50000 progress=96.6% loss=0.090372 lr=3.083360e-07 lr_backbone=3.083360e-07 elapsed=49:59:25 avg_step=3.72s recent_step=3.08s eta=01:44:17 +2026-09-19 19:28:39 | INFO | train rank=0 | step=48330/50000 progress=96.7% loss=0.066384 lr=3.046800e-07 lr_backbone=3.046800e-07 elapsed=50:00:03 avg_step=3.72s recent_step=3.78s eta=01:43:39 +2026-09-19 19:29:29 | INFO | train rank=0 | step=48340/50000 progress=96.7% loss=0.063163 lr=3.010457e-07 lr_backbone=3.010457e-07 elapsed=50:00:52 avg_step=3.72s recent_step=4.97s eta=01:43:03 +2026-09-19 19:29:59 | INFO | train rank=0 | step=48350/50000 progress=96.7% loss=0.068850 lr=2.974332e-07 lr_backbone=2.974332e-07 elapsed=50:01:22 avg_step=3.72s recent_step=2.99s eta=01:42:25 +2026-09-19 19:30:29 | INFO | train rank=0 | step=48360/50000 progress=96.7% loss=0.072794 lr=2.938424e-07 lr_backbone=2.938424e-07 elapsed=50:01:52 avg_step=3.72s recent_step=3.00s eta=01:41:48 +2026-09-19 19:31:12 | INFO | train rank=0 | step=48370/50000 progress=96.7% loss=0.077190 lr=2.902733e-07 lr_backbone=2.902733e-07 elapsed=50:02:36 avg_step=3.72s recent_step=4.33s eta=01:41:11 +2026-09-19 19:31:54 | INFO | train rank=0 | step=48380/50000 progress=96.8% loss=0.077488 lr=2.867260e-07 lr_backbone=2.867260e-07 elapsed=50:03:17 avg_step=3.72s recent_step=4.16s eta=01:40:33 +2026-09-19 19:32:28 | INFO | train rank=0 | step=48390/50000 progress=96.8% loss=0.073896 lr=2.832004e-07 lr_backbone=2.832004e-07 elapsed=50:03:52 avg_step=3.72s recent_step=3.45s eta=01:39:56 +2026-09-19 19:33:01 | INFO | train rank=0 | step=48400/50000 progress=96.8% loss=0.067817 lr=2.796966e-07 lr_backbone=2.796966e-07 elapsed=50:04:25 avg_step=3.72s recent_step=3.30s eta=01:39:19 +2026-09-19 19:33:40 | INFO | train rank=0 | step=48410/50000 progress=96.8% loss=0.065174 lr=2.762145e-07 lr_backbone=2.762145e-07 elapsed=50:05:03 avg_step=3.72s recent_step=3.85s eta=01:38:41 +2026-09-19 19:34:30 | INFO | train rank=0 | step=48420/50000 progress=96.8% loss=0.096312 lr=2.727542e-07 lr_backbone=2.727542e-07 elapsed=50:05:53 avg_step=3.72s recent_step=4.98s eta=01:38:05 +2026-09-19 19:34:56 | INFO | train rank=0 | step=48430/50000 progress=96.9% loss=0.091861 lr=2.693156e-07 lr_backbone=2.693156e-07 elapsed=50:06:19 avg_step=3.72s recent_step=2.64s eta=01:37:27 +2026-09-19 19:35:25 | INFO | train rank=0 | step=48440/50000 progress=96.9% loss=0.077764 lr=2.658988e-07 lr_backbone=2.658988e-07 elapsed=50:06:48 avg_step=3.72s recent_step=2.88s eta=01:36:50 +2026-09-19 19:36:03 | INFO | train rank=0 | step=48450/50000 progress=96.9% loss=0.085834 lr=2.625038e-07 lr_backbone=2.625038e-07 elapsed=50:07:26 avg_step=3.72s recent_step=3.79s eta=01:36:12 +2026-09-19 19:36:46 | INFO | train rank=0 | step=48460/50000 progress=96.9% loss=0.068027 lr=2.591305e-07 lr_backbone=2.591305e-07 elapsed=50:08:09 avg_step=3.72s recent_step=4.34s eta=01:35:35 +2026-09-19 19:37:19 | INFO | train rank=0 | step=48470/50000 progress=96.9% loss=0.092363 lr=2.557789e-07 lr_backbone=2.557789e-07 elapsed=50:08:42 avg_step=3.72s recent_step=3.29s eta=01:34:58 +2026-09-19 19:37:52 | INFO | train rank=0 | step=48480/50000 progress=97.0% loss=0.087760 lr=2.524492e-07 lr_backbone=2.524492e-07 elapsed=50:09:15 avg_step=3.72s recent_step=3.32s eta=01:34:20 +2026-09-19 19:38:43 | INFO | train rank=0 | step=48490/50000 progress=97.0% loss=0.071488 lr=2.491411e-07 lr_backbone=2.491411e-07 elapsed=50:10:06 avg_step=3.72s recent_step=5.08s eta=01:33:44 +2026-09-19 19:39:26 | INFO | train rank=0 | step=48500/50000 progress=97.0% loss=0.074136 lr=2.458549e-07 lr_backbone=2.458549e-07 elapsed=50:10:49 avg_step=3.72s recent_step=4.28s eta=01:33:07 +2026-09-19 19:39:53 | INFO | train rank=0 | step=48510/50000 progress=97.0% loss=0.102793 lr=2.425904e-07 lr_backbone=2.425904e-07 elapsed=50:11:17 avg_step=3.72s recent_step=2.77s eta=01:32:29 +2026-09-19 19:40:27 | INFO | train rank=0 | step=48520/50000 progress=97.0% loss=0.088262 lr=2.393477e-07 lr_backbone=2.393477e-07 elapsed=50:11:51 avg_step=3.72s recent_step=3.38s eta=01:31:52 +2026-09-19 19:40:57 | INFO | train rank=0 | step=48530/50000 progress=97.1% loss=0.068596 lr=2.361267e-07 lr_backbone=2.361267e-07 elapsed=50:12:20 avg_step=3.72s recent_step=2.94s eta=01:31:14 +2026-09-19 19:41:51 | INFO | train rank=0 | step=48540/50000 progress=97.1% loss=0.086907 lr=2.329275e-07 lr_backbone=2.329275e-07 elapsed=50:13:14 avg_step=3.72s recent_step=5.39s eta=01:30:37 +2026-09-19 19:42:17 | INFO | train rank=0 | step=48550/50000 progress=97.1% loss=0.064855 lr=2.297501e-07 lr_backbone=2.297501e-07 elapsed=50:13:40 avg_step=3.72s recent_step=2.63s eta=01:30:00 +2026-09-19 19:42:51 | INFO | train rank=0 | step=48560/50000 progress=97.1% loss=0.092311 lr=2.265944e-07 lr_backbone=2.265944e-07 elapsed=50:14:14 avg_step=3.72s recent_step=3.37s eta=01:29:23 +2026-09-19 19:43:24 | INFO | train rank=0 | step=48570/50000 progress=97.1% loss=0.104342 lr=2.234605e-07 lr_backbone=2.234605e-07 elapsed=50:14:47 avg_step=3.72s recent_step=3.32s eta=01:28:45 +2026-09-19 19:44:22 | INFO | train rank=0 | step=48580/50000 progress=97.2% loss=0.105059 lr=2.203484e-07 lr_backbone=2.203484e-07 elapsed=50:15:46 avg_step=3.72s recent_step=5.86s eta=01:28:09 +2026-09-19 19:44:54 | INFO | train rank=0 | step=48590/50000 progress=97.2% loss=0.099272 lr=2.172581e-07 lr_backbone=2.172581e-07 elapsed=50:16:18 avg_step=3.72s recent_step=3.19s eta=01:27:31 +2026-09-19 19:45:30 | INFO | train rank=0 | step=48600/50000 progress=97.2% loss=0.089153 lr=2.141896e-07 lr_backbone=2.141896e-07 elapsed=50:16:53 avg_step=3.72s recent_step=3.54s eta=01:26:54 +2026-09-19 19:45:57 | INFO | train rank=0 | step=48610/50000 progress=97.2% loss=0.073229 lr=2.111428e-07 lr_backbone=2.111428e-07 elapsed=50:17:20 avg_step=3.72s recent_step=2.72s eta=01:26:16 +2026-09-19 19:46:50 | INFO | train rank=0 | step=48620/50000 progress=97.2% loss=0.054415 lr=2.081178e-07 lr_backbone=2.081178e-07 elapsed=50:18:13 avg_step=3.72s recent_step=5.31s eta=01:25:40 +2026-09-19 19:47:26 | INFO | train rank=0 | step=48630/50000 progress=97.3% loss=0.090636 lr=2.051146e-07 lr_backbone=2.051146e-07 elapsed=50:18:49 avg_step=3.72s recent_step=3.60s eta=01:25:02 +2026-09-19 19:47:58 | INFO | train rank=0 | step=48640/50000 progress=97.3% loss=0.082038 lr=2.021331e-07 lr_backbone=2.021331e-07 elapsed=50:19:21 avg_step=3.72s recent_step=3.14s eta=01:24:25 +2026-09-19 19:48:38 | INFO | train rank=0 | step=48650/50000 progress=97.3% loss=0.098034 lr=1.991735e-07 lr_backbone=1.991735e-07 elapsed=50:20:01 avg_step=3.72s recent_step=4.06s eta=01:23:48 +2026-09-19 19:49:31 | INFO | train rank=0 | step=48660/50000 progress=97.3% loss=0.086377 lr=1.962356e-07 lr_backbone=1.962356e-07 elapsed=50:20:54 avg_step=3.72s recent_step=5.30s eta=01:23:11 +2026-09-19 19:50:11 | INFO | train rank=0 | step=48670/50000 progress=97.3% loss=0.095284 lr=1.933195e-07 lr_backbone=1.933195e-07 elapsed=50:21:34 avg_step=3.72s recent_step=3.99s eta=01:22:34 +2026-09-19 19:50:40 | INFO | train rank=0 | step=48680/50000 progress=97.4% loss=0.117000 lr=1.904253e-07 lr_backbone=1.904253e-07 elapsed=50:22:04 avg_step=3.72s recent_step=2.91s eta=01:21:56 +2026-09-19 19:51:13 | INFO | train rank=0 | step=48690/50000 progress=97.4% loss=0.130846 lr=1.875527e-07 lr_backbone=1.875527e-07 elapsed=50:22:36 avg_step=3.72s recent_step=3.23s eta=01:21:19 +2026-09-19 19:52:08 | INFO | train rank=0 | step=48700/50000 progress=97.4% loss=0.060598 lr=1.847020e-07 lr_backbone=1.847020e-07 elapsed=50:23:31 avg_step=3.73s recent_step=5.55s eta=01:20:42 +2026-09-19 19:52:43 | INFO | train rank=0 | step=48710/50000 progress=97.4% loss=0.072757 lr=1.818731e-07 lr_backbone=1.818731e-07 elapsed=50:24:07 avg_step=3.73s recent_step=3.51s eta=01:20:05 +2026-09-19 19:53:14 | INFO | train rank=0 | step=48720/50000 progress=97.4% loss=0.080814 lr=1.790660e-07 lr_backbone=1.790660e-07 elapsed=50:24:37 avg_step=3.72s recent_step=3.04s eta=01:19:27 +2026-09-19 19:53:43 | INFO | train rank=0 | step=48730/50000 progress=97.5% loss=0.151822 lr=1.762806e-07 lr_backbone=1.762806e-07 elapsed=50:25:06 avg_step=3.72s recent_step=2.92s eta=01:18:50 +2026-09-19 19:54:30 | INFO | train rank=0 | step=48740/50000 progress=97.5% loss=0.106096 lr=1.735171e-07 lr_backbone=1.735171e-07 elapsed=50:25:53 avg_step=3.72s recent_step=4.70s eta=01:18:13 +2026-09-19 19:55:08 | INFO | train rank=0 | step=48750/50000 progress=97.5% loss=0.103679 lr=1.707753e-07 lr_backbone=1.707753e-07 elapsed=50:26:31 avg_step=3.72s recent_step=3.79s eta=01:17:36 +2026-09-19 19:55:35 | INFO | train rank=0 | step=48760/50000 progress=97.5% loss=0.072230 lr=1.680554e-07 lr_backbone=1.680554e-07 elapsed=50:26:58 avg_step=3.72s recent_step=2.72s eta=01:16:58 +2026-09-19 19:56:12 | INFO | train rank=0 | step=48770/50000 progress=97.5% loss=0.081101 lr=1.653572e-07 lr_backbone=1.653572e-07 elapsed=50:27:35 avg_step=3.72s recent_step=3.69s eta=01:16:21 +2026-09-19 19:57:09 | INFO | train rank=0 | step=48780/50000 progress=97.6% loss=0.095015 lr=1.626809e-07 lr_backbone=1.626809e-07 elapsed=50:28:32 avg_step=3.73s recent_step=5.69s eta=01:15:44 +2026-09-19 19:57:43 | INFO | train rank=0 | step=48790/50000 progress=97.6% loss=0.106770 lr=1.600263e-07 lr_backbone=1.600263e-07 elapsed=50:29:06 avg_step=3.73s recent_step=3.38s eta=01:15:07 +2026-09-19 19:58:19 | INFO | train rank=0 | step=48800/50000 progress=97.6% loss=0.097504 lr=1.573936e-07 lr_backbone=1.573936e-07 elapsed=50:29:42 avg_step=3.73s recent_step=3.60s eta=01:14:30 +2026-09-19 19:58:55 | INFO | train rank=0 | step=48810/50000 progress=97.6% loss=0.072404 lr=1.547826e-07 lr_backbone=1.547826e-07 elapsed=50:30:18 avg_step=3.73s recent_step=3.64s eta=01:13:52 +2026-09-19 19:59:44 | INFO | train rank=0 | step=48820/50000 progress=97.6% loss=0.070881 lr=1.521935e-07 lr_backbone=1.521935e-07 elapsed=50:31:07 avg_step=3.73s recent_step=4.89s eta=01:13:15 +2026-09-19 20:00:13 | INFO | train rank=0 | step=48830/50000 progress=97.7% loss=0.089971 lr=1.496262e-07 lr_backbone=1.496262e-07 elapsed=50:31:37 avg_step=3.73s recent_step=2.96s eta=01:12:38 +2026-09-19 20:00:47 | INFO | train rank=0 | step=48840/50000 progress=97.7% loss=0.092283 lr=1.470806e-07 lr_backbone=1.470806e-07 elapsed=50:32:10 avg_step=3.73s recent_step=3.36s eta=01:12:01 +2026-09-19 20:01:22 | INFO | train rank=0 | step=48850/50000 progress=97.7% loss=0.072922 lr=1.445569e-07 lr_backbone=1.445569e-07 elapsed=50:32:45 avg_step=3.72s recent_step=3.51s eta=01:11:23 +2026-09-19 20:02:16 | INFO | train rank=0 | step=48860/50000 progress=97.7% loss=0.079092 lr=1.420550e-07 lr_backbone=1.420550e-07 elapsed=50:33:39 avg_step=3.73s recent_step=5.35s eta=01:10:46 +2026-09-19 20:02:51 | INFO | train rank=0 | step=48870/50000 progress=97.7% loss=0.079048 lr=1.395749e-07 lr_backbone=1.395749e-07 elapsed=50:34:14 avg_step=3.73s recent_step=3.51s eta=01:10:09 +2026-09-19 20:03:26 | INFO | train rank=0 | step=48880/50000 progress=97.8% loss=0.099217 lr=1.371166e-07 lr_backbone=1.371166e-07 elapsed=50:34:49 avg_step=3.73s recent_step=3.53s eta=01:09:32 +2026-09-19 20:03:55 | INFO | train rank=0 | step=48890/50000 progress=97.8% loss=0.112697 lr=1.346801e-07 lr_backbone=1.346801e-07 elapsed=50:35:18 avg_step=3.73s recent_step=2.89s eta=01:08:54 +2026-09-19 20:04:44 | INFO | train rank=0 | step=48900/50000 progress=97.8% loss=0.095052 lr=1.322654e-07 lr_backbone=1.322654e-07 elapsed=50:36:08 avg_step=3.73s recent_step=4.95s eta=01:08:17 +2026-09-19 20:05:22 | INFO | train rank=0 | step=48910/50000 progress=97.8% loss=0.069965 lr=1.298726e-07 lr_backbone=1.298726e-07 elapsed=50:36:46 avg_step=3.73s recent_step=3.80s eta=01:07:40 +2026-09-19 20:05:48 | INFO | train rank=0 | step=48920/50000 progress=97.8% loss=0.105184 lr=1.275015e-07 lr_backbone=1.275015e-07 elapsed=50:37:11 avg_step=3.73s recent_step=2.56s eta=01:07:03 +2026-09-19 20:06:29 | INFO | train rank=0 | step=48930/50000 progress=97.9% loss=0.128492 lr=1.251523e-07 lr_backbone=1.251523e-07 elapsed=50:37:52 avg_step=3.73s recent_step=4.10s eta=01:06:25 +2026-09-19 20:07:17 | INFO | train rank=0 | step=48940/50000 progress=97.9% loss=0.099956 lr=1.228249e-07 lr_backbone=1.228249e-07 elapsed=50:38:41 avg_step=3.73s recent_step=4.84s eta=01:05:48 +2026-09-19 20:07:41 | INFO | train rank=0 | step=48950/50000 progress=97.9% loss=0.097240 lr=1.205193e-07 lr_backbone=1.205193e-07 elapsed=50:39:05 avg_step=3.73s recent_step=2.39s eta=01:05:11 +2026-09-19 20:08:10 | INFO | train rank=0 | step=48960/50000 progress=97.9% loss=0.091878 lr=1.182355e-07 lr_backbone=1.182355e-07 elapsed=50:39:33 avg_step=3.72s recent_step=2.88s eta=01:04:33 +2026-09-19 20:08:44 | INFO | train rank=0 | step=48970/50000 progress=97.9% loss=0.107199 lr=1.159736e-07 lr_backbone=1.159736e-07 elapsed=50:40:08 avg_step=3.72s recent_step=3.41s eta=01:03:56 +2026-09-19 20:09:41 | INFO | train rank=0 | step=48980/50000 progress=98.0% loss=0.081593 lr=1.137334e-07 lr_backbone=1.137334e-07 elapsed=50:41:04 avg_step=3.73s recent_step=5.65s eta=01:03:19 +2026-09-19 20:10:12 | INFO | train rank=0 | step=48990/50000 progress=98.0% loss=0.168845 lr=1.115151e-07 lr_backbone=1.115151e-07 elapsed=50:41:35 avg_step=3.73s recent_step=3.09s eta=01:02:42 +2026-09-19 20:10:40 | INFO | train rank=0 | step=49000/50000 progress=98.0% loss=0.099848 lr=1.093186e-07 lr_backbone=1.093186e-07 elapsed=50:42:04 avg_step=3.72s recent_step=2.87s eta=01:02:04 +2026-09-19 20:11:13 | INFO | train rank=0 | step=49010/50000 progress=98.0% loss=0.091076 lr=1.071440e-07 lr_backbone=1.071440e-07 elapsed=50:42:36 avg_step=3.72s recent_step=3.27s eta=01:01:27 +2026-09-19 20:12:09 | INFO | train rank=0 | step=49020/50000 progress=98.0% loss=0.082935 lr=1.049911e-07 lr_backbone=1.049911e-07 elapsed=50:43:32 avg_step=3.73s recent_step=5.59s eta=01:00:50 +2026-09-19 20:12:43 | INFO | train rank=0 | step=49030/50000 progress=98.1% loss=0.072909 lr=1.028601e-07 lr_backbone=1.028601e-07 elapsed=50:44:06 avg_step=3.73s recent_step=3.36s eta=01:00:13 +2026-09-19 20:13:12 | INFO | train rank=0 | step=49040/50000 progress=98.1% loss=0.123485 lr=1.007509e-07 lr_backbone=1.007509e-07 elapsed=50:44:35 avg_step=3.73s recent_step=2.93s eta=00:59:36 +2026-09-19 20:13:47 | INFO | train rank=0 | step=49050/50000 progress=98.1% loss=0.118396 lr=9.866358e-08 lr_backbone=9.866358e-08 elapsed=50:45:10 avg_step=3.72s recent_step=3.53s eta=00:58:58 +2026-09-19 20:14:31 | INFO | train rank=0 | step=49060/50000 progress=98.1% loss=0.088859 lr=9.659805e-08 lr_backbone=9.659805e-08 elapsed=50:45:54 avg_step=3.73s recent_step=4.40s eta=00:58:21 +2026-09-19 20:15:19 | INFO | train rank=0 | step=49070/50000 progress=98.1% loss=0.113837 lr=9.455435e-08 lr_backbone=9.455435e-08 elapsed=50:46:42 avg_step=3.73s recent_step=4.80s eta=00:57:44 +2026-09-19 20:15:47 | INFO | train rank=0 | step=49080/50000 progress=98.2% loss=0.081981 lr=9.253248e-08 lr_backbone=9.253248e-08 elapsed=50:47:10 avg_step=3.73s recent_step=2.79s eta=00:57:07 +2026-09-19 20:16:17 | INFO | train rank=0 | step=49090/50000 progress=98.2% loss=0.090027 lr=9.053244e-08 lr_backbone=9.053244e-08 elapsed=50:47:40 avg_step=3.73s recent_step=2.95s eta=00:56:29 +2026-09-19 20:17:02 | INFO | train rank=0 | step=49100/50000 progress=98.2% loss=0.107577 lr=8.855423e-08 lr_backbone=8.855423e-08 elapsed=50:48:25 avg_step=3.73s recent_step=4.56s eta=00:55:52 +2026-09-19 20:17:38 | INFO | train rank=0 | step=49110/50000 progress=98.2% loss=0.069306 lr=8.659786e-08 lr_backbone=8.659786e-08 elapsed=50:49:02 avg_step=3.73s recent_step=3.62s eta=00:55:15 +2026-09-19 20:18:02 | INFO | train rank=0 | step=49120/50000 progress=98.2% loss=0.097162 lr=8.466332e-08 lr_backbone=8.466332e-08 elapsed=50:49:26 avg_step=3.72s recent_step=2.41s eta=00:54:37 +2026-09-19 20:18:34 | INFO | train rank=0 | step=49130/50000 progress=98.3% loss=0.118224 lr=8.275061e-08 lr_backbone=8.275061e-08 elapsed=50:49:57 avg_step=3.72s recent_step=3.12s eta=00:54:00 +2026-09-19 20:19:15 | INFO | train rank=0 | step=49140/50000 progress=98.3% loss=0.086718 lr=8.085974e-08 lr_backbone=8.085974e-08 elapsed=50:50:39 avg_step=3.72s recent_step=4.16s eta=00:53:23 +2026-09-19 20:19:54 | INFO | train rank=0 | step=49150/50000 progress=98.3% loss=0.080329 lr=7.899071e-08 lr_backbone=7.899071e-08 elapsed=50:51:17 avg_step=3.72s recent_step=3.86s eta=00:52:46 +2026-09-19 20:20:26 | INFO | train rank=0 | step=49160/50000 progress=98.3% loss=0.071844 lr=7.714351e-08 lr_backbone=7.714351e-08 elapsed=50:51:50 avg_step=3.72s recent_step=3.24s eta=00:52:08 +2026-09-19 20:20:52 | INFO | train rank=0 | step=49170/50000 progress=98.3% loss=0.066325 lr=7.531815e-08 lr_backbone=7.531815e-08 elapsed=50:52:16 avg_step=3.72s recent_step=2.62s eta=00:51:31 +2026-09-19 20:21:54 | INFO | train rank=0 | step=49180/50000 progress=98.4% loss=0.052231 lr=7.351463e-08 lr_backbone=7.351463e-08 elapsed=50:53:17 avg_step=3.73s recent_step=6.14s eta=00:50:54 +2026-09-19 20:22:26 | INFO | train rank=0 | step=49190/50000 progress=98.4% loss=0.072795 lr=7.173295e-08 lr_backbone=7.173295e-08 elapsed=50:53:49 avg_step=3.72s recent_step=3.23s eta=00:50:17 +2026-09-19 20:23:00 | INFO | train rank=0 | step=49200/50000 progress=98.4% loss=0.077774 lr=6.997311e-08 lr_backbone=6.997311e-08 elapsed=50:54:23 avg_step=3.72s recent_step=3.40s eta=00:49:39 +2026-09-19 20:23:32 | INFO | train rank=0 | step=49210/50000 progress=98.4% loss=0.076930 lr=6.823511e-08 lr_backbone=6.823511e-08 elapsed=50:54:55 avg_step=3.72s recent_step=3.17s eta=00:49:02 +2026-09-19 20:24:28 | INFO | train rank=0 | step=49220/50000 progress=98.4% loss=0.096674 lr=6.651896e-08 lr_backbone=6.651896e-08 elapsed=50:55:51 avg_step=3.73s recent_step=5.59s eta=00:48:25 +2026-09-19 20:24:59 | INFO | train rank=0 | step=49230/50000 progress=98.5% loss=0.076829 lr=6.482464e-08 lr_backbone=6.482464e-08 elapsed=50:56:22 avg_step=3.73s recent_step=3.12s eta=00:47:48 +2026-09-19 20:25:39 | INFO | train rank=0 | step=49240/50000 progress=98.5% loss=0.110206 lr=6.315217e-08 lr_backbone=6.315217e-08 elapsed=50:57:02 avg_step=3.73s recent_step=3.96s eta=00:47:11 +2026-09-19 20:26:11 | INFO | train rank=0 | step=49250/50000 progress=98.5% loss=0.072731 lr=6.150154e-08 lr_backbone=6.150154e-08 elapsed=50:57:34 avg_step=3.72s recent_step=3.24s eta=00:46:33 +2026-09-19 20:27:09 | INFO | train rank=0 | step=49260/50000 progress=98.5% loss=0.087634 lr=5.987276e-08 lr_backbone=5.987276e-08 elapsed=50:58:32 avg_step=3.73s recent_step=5.78s eta=00:45:56 +2026-09-19 20:27:41 | INFO | train rank=0 | step=49270/50000 progress=98.5% loss=0.147108 lr=5.826582e-08 lr_backbone=5.826582e-08 elapsed=50:59:04 avg_step=3.73s recent_step=3.22s eta=00:45:19 +2026-09-19 20:28:17 | INFO | train rank=0 | step=49280/50000 progress=98.6% loss=0.110898 lr=5.668073e-08 lr_backbone=5.668073e-08 elapsed=50:59:40 avg_step=3.73s recent_step=3.54s eta=00:44:42 +2026-09-19 20:28:39 | INFO | train rank=0 | step=49290/50000 progress=98.6% loss=0.119510 lr=5.511749e-08 lr_backbone=5.511749e-08 elapsed=51:00:02 avg_step=3.72s recent_step=2.24s eta=00:44:04 +2026-09-19 20:29:31 | INFO | train rank=0 | step=49300/50000 progress=98.6% loss=0.074369 lr=5.357609e-08 lr_backbone=5.357609e-08 elapsed=51:00:55 avg_step=3.73s recent_step=5.23s eta=00:43:27 +2026-09-19 20:30:02 | INFO | train rank=0 | step=49310/50000 progress=98.6% loss=0.095037 lr=5.205655e-08 lr_backbone=5.205655e-08 elapsed=51:01:26 avg_step=3.73s recent_step=3.12s eta=00:42:50 +2026-09-19 20:30:32 | INFO | train rank=0 | step=49320/50000 progress=98.6% loss=0.097192 lr=5.055885e-08 lr_backbone=5.055885e-08 elapsed=51:01:56 avg_step=3.72s recent_step=2.99s eta=00:42:12 +2026-09-19 20:31:05 | INFO | train rank=0 | step=49330/50000 progress=98.7% loss=0.089563 lr=4.908300e-08 lr_backbone=4.908300e-08 elapsed=51:02:29 avg_step=3.72s recent_step=3.30s eta=00:41:35 +2026-09-19 20:32:01 | INFO | train rank=0 | step=49340/50000 progress=98.7% loss=0.099300 lr=4.762900e-08 lr_backbone=4.762900e-08 elapsed=51:03:25 avg_step=3.73s recent_step=5.60s eta=00:40:58 +2026-09-19 20:32:40 | INFO | train rank=0 | step=49350/50000 progress=98.7% loss=0.089718 lr=4.619685e-08 lr_backbone=4.619685e-08 elapsed=51:04:03 avg_step=3.73s recent_step=3.81s eta=00:40:21 +2026-09-19 20:33:08 | INFO | train rank=0 | step=49360/50000 progress=98.7% loss=0.071376 lr=4.478655e-08 lr_backbone=4.478655e-08 elapsed=51:04:31 avg_step=3.73s recent_step=2.83s eta=00:39:44 +2026-09-19 20:33:43 | INFO | train rank=0 | step=49370/50000 progress=98.7% loss=0.082185 lr=4.339811e-08 lr_backbone=4.339811e-08 elapsed=51:05:06 avg_step=3.73s recent_step=3.52s eta=00:39:06 +2026-09-19 20:34:39 | INFO | train rank=0 | step=49380/50000 progress=98.8% loss=0.081749 lr=4.203152e-08 lr_backbone=4.203152e-08 elapsed=51:06:02 avg_step=3.73s recent_step=5.60s eta=00:38:29 +2026-09-19 20:35:15 | INFO | train rank=0 | step=49390/50000 progress=98.8% loss=0.095040 lr=4.068678e-08 lr_backbone=4.068678e-08 elapsed=51:06:38 avg_step=3.73s recent_step=3.61s eta=00:37:52 +2026-09-19 20:35:48 | INFO | train rank=0 | step=49400/50000 progress=98.8% loss=0.086811 lr=3.936389e-08 lr_backbone=3.936389e-08 elapsed=51:07:11 avg_step=3.73s recent_step=3.25s eta=00:37:15 +2026-09-19 20:36:17 | INFO | train rank=0 | step=49410/50000 progress=98.8% loss=0.075615 lr=3.806286e-08 lr_backbone=3.806286e-08 elapsed=51:07:40 avg_step=3.73s recent_step=2.93s eta=00:36:37 +2026-09-19 20:37:15 | INFO | train rank=0 | step=49420/50000 progress=98.8% loss=0.083369 lr=3.678369e-08 lr_backbone=3.678369e-08 elapsed=51:08:38 avg_step=3.73s recent_step=5.77s eta=00:36:00 +2026-09-19 20:37:44 | INFO | train rank=0 | step=49430/50000 progress=98.9% loss=0.083840 lr=3.552637e-08 lr_backbone=3.552637e-08 elapsed=51:09:08 avg_step=3.73s recent_step=2.97s eta=00:35:23 +2026-09-19 20:38:20 | INFO | train rank=0 | step=49440/50000 progress=98.9% loss=0.083494 lr=3.429090e-08 lr_backbone=3.429090e-08 elapsed=51:09:44 avg_step=3.73s recent_step=3.59s eta=00:34:46 +2026-09-19 20:38:52 | INFO | train rank=0 | step=49450/50000 progress=98.9% loss=0.088369 lr=3.307730e-08 lr_backbone=3.307730e-08 elapsed=51:10:16 avg_step=3.73s recent_step=3.23s eta=00:34:08 +2026-09-19 20:39:38 | INFO | train rank=0 | step=49460/50000 progress=98.9% loss=0.097683 lr=3.188555e-08 lr_backbone=3.188555e-08 elapsed=51:11:02 avg_step=3.73s recent_step=4.58s eta=00:33:31 +2026-09-19 20:40:18 | INFO | train rank=0 | step=49470/50000 progress=98.9% loss=0.083227 lr=3.071566e-08 lr_backbone=3.071566e-08 elapsed=51:11:41 avg_step=3.73s recent_step=3.95s eta=00:32:54 +2026-09-19 20:40:46 | INFO | train rank=0 | step=49480/50000 progress=99.0% loss=0.081283 lr=2.956762e-08 lr_backbone=2.956762e-08 elapsed=51:12:09 avg_step=3.73s recent_step=2.79s eta=00:32:17 +2026-09-19 20:41:13 | INFO | train rank=0 | step=49490/50000 progress=99.0% loss=0.115408 lr=2.844145e-08 lr_backbone=2.844145e-08 elapsed=51:12:36 avg_step=3.73s recent_step=2.72s eta=00:31:39 +2026-09-19 20:41:55 | INFO | train rank=0 | step=49500/50000 progress=99.0% loss=0.067183 lr=2.733713e-08 lr_backbone=2.733713e-08 elapsed=51:13:18 avg_step=3.73s recent_step=4.21s eta=00:31:02 +2026-09-19 20:42:32 | INFO | train rank=0 | step=49510/50000 progress=99.0% loss=0.064667 lr=2.625468e-08 lr_backbone=2.625468e-08 elapsed=51:13:55 avg_step=3.73s recent_step=3.68s eta=00:30:25 +2026-09-19 20:43:10 | INFO | train rank=0 | step=49520/50000 progress=99.0% loss=0.077749 lr=2.519408e-08 lr_backbone=2.519408e-08 elapsed=51:14:33 avg_step=3.73s recent_step=3.84s eta=00:29:48 +2026-09-19 20:43:41 | INFO | train rank=0 | step=49530/50000 progress=99.1% loss=0.074127 lr=2.415535e-08 lr_backbone=2.415535e-08 elapsed=51:15:04 avg_step=3.73s recent_step=3.11s eta=00:29:10 +2026-09-19 20:44:35 | INFO | train rank=0 | step=49540/50000 progress=99.1% loss=0.084024 lr=2.313847e-08 lr_backbone=2.313847e-08 elapsed=51:15:59 avg_step=3.73s recent_step=5.43s eta=00:28:33 +2026-09-19 20:45:03 | INFO | train rank=0 | step=49550/50000 progress=99.1% loss=0.116514 lr=2.214346e-08 lr_backbone=2.214346e-08 elapsed=51:16:26 avg_step=3.73s recent_step=2.73s eta=00:27:56 +2026-09-19 20:45:30 | INFO | train rank=0 | step=49560/50000 progress=99.1% loss=0.068049 lr=2.117031e-08 lr_backbone=2.117031e-08 elapsed=51:16:53 avg_step=3.73s recent_step=2.69s eta=00:27:19 +2026-09-19 20:46:04 | INFO | train rank=0 | step=49570/50000 progress=99.1% loss=0.064813 lr=2.021902e-08 lr_backbone=2.021902e-08 elapsed=51:17:27 avg_step=3.72s recent_step=3.40s eta=00:26:41 +2026-09-19 20:46:51 | INFO | train rank=0 | step=49580/50000 progress=99.2% loss=0.089134 lr=1.928960e-08 lr_backbone=1.928960e-08 elapsed=51:18:14 avg_step=3.73s recent_step=4.72s eta=00:26:04 +2026-09-19 20:47:27 | INFO | train rank=0 | step=49590/50000 progress=99.2% loss=0.103606 lr=1.838204e-08 lr_backbone=1.838204e-08 elapsed=51:18:51 avg_step=3.73s recent_step=3.65s eta=00:25:27 +2026-09-19 20:47:50 | INFO | train rank=0 | step=49600/50000 progress=99.2% loss=0.107366 lr=1.749634e-08 lr_backbone=1.749634e-08 elapsed=51:19:13 avg_step=3.72s recent_step=2.22s eta=00:24:49 +2026-09-19 20:48:26 | INFO | train rank=0 | step=49610/50000 progress=99.2% loss=0.074888 lr=1.663251e-08 lr_backbone=1.663251e-08 elapsed=51:19:49 avg_step=3.72s recent_step=3.59s eta=00:24:12 +2026-09-19 20:49:19 | INFO | train rank=0 | step=49620/50000 progress=99.2% loss=0.075559 lr=1.579054e-08 lr_backbone=1.579054e-08 elapsed=51:20:42 avg_step=3.73s recent_step=5.30s eta=00:23:35 +2026-09-19 20:49:58 | INFO | train rank=0 | step=49630/50000 progress=99.3% loss=0.074816 lr=1.497043e-08 lr_backbone=1.497043e-08 elapsed=51:21:21 avg_step=3.73s recent_step=3.92s eta=00:22:58 +2026-09-19 20:50:28 | INFO | train rank=0 | step=49640/50000 progress=99.3% loss=0.070877 lr=1.417219e-08 lr_backbone=1.417219e-08 elapsed=51:21:52 avg_step=3.73s recent_step=3.06s eta=00:22:21 +2026-09-19 20:51:03 | INFO | train rank=0 | step=49650/50000 progress=99.3% loss=0.066005 lr=1.339582e-08 lr_backbone=1.339582e-08 elapsed=51:22:26 avg_step=3.73s recent_step=3.46s eta=00:21:43 +2026-09-19 20:51:48 | INFO | train rank=0 | step=49660/50000 progress=99.3% loss=0.070400 lr=1.264131e-08 lr_backbone=1.264131e-08 elapsed=51:23:11 avg_step=3.73s recent_step=4.46s eta=00:21:06 +2026-09-19 20:52:32 | INFO | train rank=0 | step=49670/50000 progress=99.3% loss=0.143470 lr=1.190867e-08 lr_backbone=1.190867e-08 elapsed=51:23:55 avg_step=3.73s recent_step=4.40s eta=00:20:29 +2026-09-19 20:53:02 | INFO | train rank=0 | step=49680/50000 progress=99.4% loss=0.074150 lr=1.119789e-08 lr_backbone=1.119789e-08 elapsed=51:24:26 avg_step=3.73s recent_step=3.07s eta=00:19:52 +2026-09-19 20:53:33 | INFO | train rank=0 | step=49690/50000 progress=99.4% loss=0.084155 lr=1.050898e-08 lr_backbone=1.050898e-08 elapsed=51:24:56 avg_step=3.73s recent_step=3.09s eta=00:19:14 +2026-09-19 20:54:17 | INFO | train rank=0 | step=49700/50000 progress=99.4% loss=0.085941 lr=9.841942e-09 lr_backbone=9.841942e-09 elapsed=51:25:41 avg_step=3.73s recent_step=4.42s eta=00:18:37 +2026-09-19 20:55:03 | INFO | train rank=0 | step=49710/50000 progress=99.4% loss=0.076448 lr=9.196768e-09 lr_backbone=9.196768e-09 elapsed=51:26:26 avg_step=3.73s recent_step=4.58s eta=00:18:00 +2026-09-19 20:55:39 | INFO | train rank=0 | step=49720/50000 progress=99.4% loss=0.062278 lr=8.573461e-09 lr_backbone=8.573461e-09 elapsed=51:27:02 avg_step=3.73s recent_step=3.59s eta=00:17:23 +2026-09-19 20:56:12 | INFO | train rank=0 | step=49730/50000 progress=99.5% loss=0.090106 lr=7.972023e-09 lr_backbone=7.972023e-09 elapsed=51:27:35 avg_step=3.73s recent_step=3.26s eta=00:16:45 +2026-09-19 20:56:54 | INFO | train rank=0 | step=49740/50000 progress=99.5% loss=0.086197 lr=7.392452e-09 lr_backbone=7.392452e-09 elapsed=51:28:17 avg_step=3.73s recent_step=4.20s eta=00:16:08 +2026-09-19 20:57:40 | INFO | train rank=0 | step=49750/50000 progress=99.5% loss=0.113733 lr=6.834750e-09 lr_backbone=6.834750e-09 elapsed=51:29:03 avg_step=3.73s recent_step=4.58s eta=00:15:31 +2026-09-19 20:58:13 | INFO | train rank=0 | step=49760/50000 progress=99.5% loss=0.104339 lr=6.298917e-09 lr_backbone=6.298917e-09 elapsed=51:29:36 avg_step=3.73s recent_step=3.30s eta=00:14:54 +2026-09-19 20:58:45 | INFO | train rank=0 | step=49770/50000 progress=99.5% loss=0.101522 lr=5.784953e-09 lr_backbone=5.784953e-09 elapsed=51:30:08 avg_step=3.73s recent_step=3.20s eta=00:14:16 +2026-09-19 20:59:36 | INFO | train rank=0 | step=49780/50000 progress=99.6% loss=0.113204 lr=5.292858e-09 lr_backbone=5.292858e-09 elapsed=51:30:59 avg_step=3.73s recent_step=5.14s eta=00:13:39 +2026-09-19 21:00:12 | INFO | train rank=0 | step=49790/50000 progress=99.6% loss=0.052749 lr=4.822632e-09 lr_backbone=4.822632e-09 elapsed=51:31:35 avg_step=3.73s recent_step=3.58s eta=00:13:02 +2026-09-19 21:00:42 | INFO | train rank=0 | step=49800/50000 progress=99.6% loss=0.080292 lr=4.374276e-09 lr_backbone=4.374276e-09 elapsed=51:32:05 avg_step=3.73s recent_step=3.05s eta=00:12:25 +2026-09-19 21:01:10 | INFO | train rank=0 | step=49810/50000 progress=99.6% loss=0.084293 lr=3.947790e-09 lr_backbone=3.947790e-09 elapsed=51:32:34 avg_step=3.73s recent_step=2.83s eta=00:11:47 +2026-09-19 21:02:06 | INFO | train rank=0 | step=49820/50000 progress=99.6% loss=0.082514 lr=3.543173e-09 lr_backbone=3.543173e-09 elapsed=51:33:29 avg_step=3.73s recent_step=5.55s eta=00:11:10 +2026-09-19 21:02:33 | INFO | train rank=0 | step=49830/50000 progress=99.7% loss=0.076644 lr=3.160427e-09 lr_backbone=3.160427e-09 elapsed=51:33:56 avg_step=3.73s recent_step=2.71s eta=00:10:33 +2026-09-19 21:03:07 | INFO | train rank=0 | step=49840/50000 progress=99.7% loss=0.085023 lr=2.799551e-09 lr_backbone=2.799551e-09 elapsed=51:34:30 avg_step=3.73s recent_step=3.39s eta=00:09:56 +2026-09-19 21:03:40 | INFO | train rank=0 | step=49850/50000 progress=99.7% loss=0.081776 lr=2.460546e-09 lr_backbone=2.460546e-09 elapsed=51:35:03 avg_step=3.73s recent_step=3.30s eta=00:09:18 +2026-09-19 21:04:25 | INFO | train rank=0 | step=49860/50000 progress=99.7% loss=0.101053 lr=2.143411e-09 lr_backbone=2.143411e-09 elapsed=51:35:48 avg_step=3.73s recent_step=4.47s eta=00:08:41 +2026-09-19 21:05:04 | INFO | train rank=0 | step=49870/50000 progress=99.7% loss=0.077253 lr=1.848147e-09 lr_backbone=1.848147e-09 elapsed=51:36:28 avg_step=3.73s recent_step=3.97s eta=00:08:04 +2026-09-19 21:05:39 | INFO | train rank=0 | step=49880/50000 progress=99.8% loss=0.086026 lr=1.574754e-09 lr_backbone=1.574754e-09 elapsed=51:37:02 avg_step=3.73s recent_step=3.45s eta=00:07:27 +2026-09-19 21:06:07 | INFO | train rank=0 | step=49890/50000 progress=99.8% loss=0.068129 lr=1.323232e-09 lr_backbone=1.323232e-09 elapsed=51:37:30 avg_step=3.73s recent_step=2.77s eta=00:06:49 +2026-09-19 21:06:49 | INFO | train rank=0 | step=49900/50000 progress=99.8% loss=0.107507 lr=1.093581e-09 lr_backbone=1.093581e-09 elapsed=51:38:12 avg_step=3.73s recent_step=4.26s eta=00:06:12 +2026-09-19 21:07:33 | INFO | train rank=0 | step=49910/50000 progress=99.8% loss=0.066782 lr=8.858012e-10 lr_backbone=8.858012e-10 elapsed=51:38:57 avg_step=3.73s recent_step=4.40s eta=00:05:35 +2026-09-19 21:08:04 | INFO | train rank=0 | step=49920/50000 progress=99.8% loss=0.091739 lr=6.998928e-10 lr_backbone=6.998928e-10 elapsed=51:39:27 avg_step=3.73s recent_step=3.05s eta=00:04:58 +2026-09-19 21:08:35 | INFO | train rank=0 | step=49930/50000 progress=99.9% loss=0.073409 lr=5.358557e-10 lr_backbone=5.358557e-10 elapsed=51:39:58 avg_step=3.73s recent_step=3.14s eta=00:04:20 +2026-09-19 21:09:26 | INFO | train rank=0 | step=49940/50000 progress=99.9% loss=0.076117 lr=3.936901e-10 lr_backbone=3.936901e-10 elapsed=51:40:49 avg_step=3.73s recent_step=5.09s eta=00:03:43 +2026-09-19 21:10:04 | INFO | train rank=0 | step=49950/50000 progress=99.9% loss=0.086401 lr=2.733960e-10 lr_backbone=2.733960e-10 elapsed=51:41:27 avg_step=3.73s recent_step=3.76s eta=00:03:06 +2026-09-19 21:10:38 | INFO | train rank=0 | step=49960/50000 progress=99.9% loss=0.066509 lr=1.749735e-10 lr_backbone=1.749735e-10 elapsed=51:42:01 avg_step=3.73s recent_step=3.45s eta=00:02:29 +2026-09-19 21:11:09 | INFO | train rank=0 | step=49970/50000 progress=99.9% loss=0.077455 lr=9.842262e-11 lr_backbone=9.842262e-11 elapsed=51:42:33 avg_step=3.73s recent_step=3.11s eta=00:01:51 +2026-09-19 21:12:01 | INFO | train rank=0 | step=49980/50000 progress=100.0% loss=0.087184 lr=4.374339e-11 lr_backbone=4.374339e-11 elapsed=51:43:24 avg_step=3.73s recent_step=5.19s eta=00:01:14 +2026-09-19 21:12:34 | INFO | train rank=0 | step=49990/50000 progress=100.0% loss=0.071303 lr=1.093585e-11 lr_backbone=1.093585e-11 elapsed=51:43:57 avg_step=3.73s recent_step=3.26s eta=00:00:37 +2026-09-19 21:13:08 | INFO | train rank=0 | step=50000/50000 progress=100.0% loss=0.069159 lr=0.000000e+00 lr_backbone=0.000000e+00 elapsed=51:44:31 avg_step=3.73s recent_step=3.38s eta=00:00:00 +2026-09-19 21:13:34 | INFO | train rank=0 | Saved DeepSpeed checkpoint to /data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_50000 +2026-09-19 21:13:34 | INFO | train rank=0 | Finished training final_step=50000/50000 elapsed=51:44:57 avg_step=3.73s +2026-09-19 21:13:34 | INFO | train rank=0 | Training finished +2026-09-19 21:13:54 | INFO | train rank=0 | Saved final DeepSpeed checkpoint to /data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_new_instruction/ds_checkpoint_final diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..865538cf7e88d1bcf1a8b3a89e07f57d84b9f8e1 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e87d11c81d18c4e78991c5dff9b78a25fe8974eeab9c283b6e4d0929d554a331 +size 4033352816 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..e8b5045ca912eaf505134a2fee1df200066976fd --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2d4f164cdebb069d27a7f854cd6c9a4748bb33c71b3de2e0ca0bbfb6c8f975c3 +size 4033337776 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..16ab75f268fe8c0770e0f3d6ef0ced9aa0655afa --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ca5c21b8fabeaa095ac6712484a98cf900fd4acca0e356a62aff787b4cb2cf60 +size 4033334576 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..51161144bb43dec5e5fee1a57474dc9a2da0702d --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:38ce052a74b0f1ea76f96cc5110560122e900dc267f3308d24b5572306e186ff +size 4033337712 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..fdf2b883986bfe70c0272b2b1c154720830d2617 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:942f70caa6d8f2f1f3c85146b5df177d83001a4df0e23898194655ac5c00c7fd +size 4033337904 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..9c5c612f9bea4de29a2691bf03b470b91a3de8d6 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1d0973a31068e1288f7b4a76898879a44471d2bcaae080a57ea1b71fb9f87652 +size 4033338800 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..3e6a72275afa178ecc83b823d4382ec632b25279 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e8c2ea3b9109044ae45fddf035a4c231622ec626002cb7b6fa5172f229ce53fc +size 4033338160 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..f2b32b124484daf1a3d8993d871c5d484597f6bc --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d2916c687b512eaf7ebba6e8bd88bbe285d9579c94cd1e7bdd3aa511d7954a70 +size 4033337712 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/mp_rank_00_model_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..8b2f83897249332f321c95b85811ec95e4521661 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/global_step50000/mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b6a76a3a4eb76a0228e956f75589e4b6fe93e736d09ffcf4f4cf309dacf3699c +size 5378167832 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/latest b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/latest new file mode 100644 index 0000000000000000000000000000000000000000..59a558c0ee32a055bd8fc16bf0d99a5950c59ac1 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/latest @@ -0,0 +1 @@ +global_step50000 \ No newline at end of file diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/zero_to_fp32.py b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/zero_to_fp32.py new file mode 100644 index 0000000000000000000000000000000000000000..f40f1f0acbbff30e8e8d29906b56aae4e8b8a138 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000/zero_to_fp32.py @@ -0,0 +1,790 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: +# python zero_to_fp32.py . output_dir/ +# or +# python zero_to_fp32.py . output_dir/ --safe_serialization + +import argparse +import torch +import glob +import math +import os +import re +import gc +import json +import numpy as np +from tqdm import tqdm +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS, AUTOEP_LAYERS_KEY, + AUTOEP_LAYERS_KEY_LEGACY, AUTOEP_ZERO3_EXPERT_STATE_FORMAT_KEY, + AUTOEP_ZERO3_PARTITIONED_EXPERT_STATE_FORMAT) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def _has_autoep_zero3_partitioned_metadata(state_dict): + autoep_layers = state_dict.get(AUTOEP_LAYERS_KEY) + if autoep_layers is None: + autoep_layers = state_dict.get(AUTOEP_LAYERS_KEY_LEGACY) + if not isinstance(autoep_layers, list): + return False + return any( + isinstance(entry, dict) + and entry.get(AUTOEP_ZERO3_EXPERT_STATE_FORMAT_KEY) == AUTOEP_ZERO3_PARTITIONED_EXPERT_STATE_FORMAT + for entry in autoep_layers) + + +def _raise_if_autoep_zero3_partitioned_state(state_dict): + if _has_autoep_zero3_partitioned_metadata(state_dict): + raise NotImplementedError("zero_to_fp32 does not support AutoEP ZeRO-3 partition-native checkpoints. " + "AutoEP expert parameters are partitioned over expert replica groups, so " + "global data-parallel consolidation would produce incomplete expert tensors. " + "Use ds_to_universal.py for expert-aware conversion.") + + +def _raise_if_autoep_zero3_partitioned_checkpoint(model_files): + for file in model_files: + state_dict = torch.load(file, map_location=device, weights_only=False) + _raise_if_autoep_zero3_partitioned_state(state_dict) + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device, weights_only=False) + _raise_if_autoep_zero3_partitioned_state(state_dict) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + total_files = len(files) + state_dicts = [] + for f in tqdm(files, desc='Loading checkpoint shards'): + state_dict = torch.load(f, map_location=device, mmap=True, weights_only=False) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if ZERO_STAGE not in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + # parse_model_states rejects AutoEP ZeRO-3 partition-native checkpoints + # before the expensive optimizer-shard load below. + model_files = get_model_state_files(ds_checkpoint_dir) + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +class GatheredTensor: + """ + A pseudo tensor that collects partitioned weights. + It is more memory efficient when there are multiple groups. + """ + + def __init__(self, flat_groups, flat_groups_offset, offset, partitioned_numel, shape): + self.flat_groups = flat_groups + self.flat_groups_offset = flat_groups_offset + self.offset = offset + self.partitioned_numel = partitioned_numel + self.shape = shape + self.dtype = self.flat_groups[0][0].dtype + + def contiguous(self): + """ + Merge partitioned weights from flat_groups into a single tensor. + """ + end_idx = self.offset + self.partitioned_numel + world_size = len(self.flat_groups) + pad_flat_param_chunks = [] + + for rank_i in range(world_size): + # for each rank, we need to collect weights from related group/groups + flat_groups_at_rank_i = self.flat_groups[rank_i] + start_group_id = None + end_group_id = None + for group_id in range(len(self.flat_groups_offset)): + if self.flat_groups_offset[group_id] <= self.offset < self.flat_groups_offset[group_id + 1]: + start_group_id = group_id + if self.flat_groups_offset[group_id] < end_idx <= self.flat_groups_offset[group_id + 1]: + end_group_id = group_id + break + # collect weights from related group/groups + for group_id in range(start_group_id, end_group_id + 1): + flat_tensor = flat_groups_at_rank_i[group_id] + start_offset = self.offset - self.flat_groups_offset[group_id] + end_offset = min(end_idx, self.flat_groups_offset[group_id + 1]) - self.flat_groups_offset[group_id] + pad_flat_param_chunks.append(flat_tensor[start_offset:end_offset]) + + # collect weights from all ranks + pad_flat_param = torch.cat(pad_flat_param_chunks, dim=0) + param = pad_flat_param[:self.shape.numel()].view(self.shape).contiguous() + return param + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = sum([flat_group.numel() for flat_group in fp32_flat_groups[0]]) * world_size + + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + flat_groups_offset = [0] + list(np.cumsum([flat_tensor.numel() for flat_tensor in fp32_flat_groups[0]])) + for name, shape in tqdm(param_shapes.items(), desc='Gathering sharded weights'): + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # memory efficient tensor + tensor = GatheredTensor(fp32_flat_groups, flat_groups_offset, offset, partitioned_numel, shape) + state_dict[name] = tensor + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def to_torch_tensor(state_dict, return_empty_tensor=False): + """ + Convert state_dict of GatheredTensor to torch tensor + """ + torch_state_dict = {} + converted_tensors = {} + for name, tensor in state_dict.items(): + tensor_id = id(tensor) + if tensor_id in converted_tensors: # shared tensors + shared_tensor = torch_state_dict[converted_tensors[tensor_id]] + torch_state_dict[name] = shared_tensor + else: + converted_tensors[tensor_id] = name + if return_empty_tensor: + torch_state_dict[name] = torch.empty(tensor.shape, dtype=tensor.dtype) + else: + torch_state_dict[name] = tensor.contiguous() + return torch_state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, + tag=None, + exclude_frozen_parameters=False, + lazy_mode=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + - ``lazy_mode``: get state_dict in lazy mode. It returns a dict of pesduo tensor instead of torch tensor, which is more memory efficient. + Convert the pesduo tensor to torch tensor by ``.contiguous()`` + + Returns: + - pytorch ``state_dict`` + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + Note: the above usage may not work if your application doesn't have sufficient free CPU memory. + You may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. Or you can load state_dict in lazy mode :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, lazy_mode=True) # not on cpu + for name, lazy_tensor in state_dict.item(): + tensor = lazy_tensor.contiguous() # to cpu + print(name, tensor) + # del tensor to release memory if it no longer in use + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + state_dict = _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + if lazy_mode: + return state_dict + else: + return to_torch_tensor(state_dict) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, + output_dir, + max_shard_size="5GB", + safe_serialization=False, + tag=None, + exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_dir``: directory to the pytorch fp32 state_dict output files + - ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB + - ``safe_serialization``: whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`). + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + # Dependency pre-check + if safe_serialization: + try: + from safetensors.torch import save_file + except ImportError: + print('If you want to use `safe_serialization`, please `pip install safetensors`') + raise + if max_shard_size is not None: + try: + from huggingface_hub import split_torch_state_dict_into_shards + except ImportError: + print('If you want to use `max_shard_size`, please `pip install huggingface_hub`') + raise + + # Convert zero checkpoint to state_dict + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, + tag, + exclude_frozen_parameters, + lazy_mode=True) + + # Shard the model if it is too big. + weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin" + if max_shard_size is not None: + filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors") + # an memory-efficient approach for sharding + empty_state_dict = to_torch_tensor(state_dict, return_empty_tensor=True) + state_dict_split = split_torch_state_dict_into_shards(empty_state_dict, + filename_pattern=filename_pattern, + max_shard_size=max_shard_size) + else: + from collections import namedtuple + StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"]) + state_dict_split = StateDictSplit(is_sharded=False, + filename_to_tensors={weights_name: list(state_dict.keys())}) + + # Save the model by shard + os.makedirs(output_dir, exist_ok=True) + filename_to_tensors = state_dict_split.filename_to_tensors.items() + for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"): + shard_state_dict = {tensor_name: state_dict[tensor_name] for tensor_name in tensors} + shard_state_dict = to_torch_tensor(shard_state_dict) + output_path = os.path.join(output_dir, shard_file) + if safe_serialization: + save_file(shard_state_dict, output_path, metadata={"format": "pt"}) + else: + torch.save(shard_state_dict, output_path) + # release the memory of current shard + for tensor_name in list(shard_state_dict.keys()): + del state_dict[tensor_name] + del shard_state_dict[tensor_name] + del shard_state_dict + gc.collect() + + # Save index if sharded + if state_dict_split.is_sharded: + index = { + "metadata": state_dict_split.metadata, + "weight_map": state_dict_split.tensor_to_filename, + } + save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json" + save_index_file = os.path.join(output_dir, save_index_file) + with open(save_index_file, "w", encoding="utf-8") as f: + content = json.dumps(index, indent=2, sort_keys=True) + "\n" + f.write(content) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info("Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info("Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument("output_dir", + type=str, + help="directory to the pytorch fp32 state_dict output files" + "(e.g. path/checkpoint-12-output/)") + parser.add_argument( + "--max_shard_size", + type=str, + default="5GB", + help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size" + "lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`" + "We default it to 5GB in order for models to be able to run easily on free-tier google colab instances" + "without CPU OOM issues.") + parser.add_argument( + "--safe_serialization", + default=False, + action='store_true', + help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_dir, + max_shard_size=args.max_shard_size, + safe_serialization=args.safe_serialization, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..865538cf7e88d1bcf1a8b3a89e07f57d84b9f8e1 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e87d11c81d18c4e78991c5dff9b78a25fe8974eeab9c283b6e4d0929d554a331 +size 4033352816 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..e8b5045ca912eaf505134a2fee1df200066976fd --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2d4f164cdebb069d27a7f854cd6c9a4748bb33c71b3de2e0ca0bbfb6c8f975c3 +size 4033337776 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..16ab75f268fe8c0770e0f3d6ef0ced9aa0655afa --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ca5c21b8fabeaa095ac6712484a98cf900fd4acca0e356a62aff787b4cb2cf60 +size 4033334576 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..51161144bb43dec5e5fee1a57474dc9a2da0702d --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:38ce052a74b0f1ea76f96cc5110560122e900dc267f3308d24b5572306e186ff +size 4033337712 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..fdf2b883986bfe70c0272b2b1c154720830d2617 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:942f70caa6d8f2f1f3c85146b5df177d83001a4df0e23898194655ac5c00c7fd +size 4033337904 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..9c5c612f9bea4de29a2691bf03b470b91a3de8d6 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1d0973a31068e1288f7b4a76898879a44471d2bcaae080a57ea1b71fb9f87652 +size 4033338800 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..3e6a72275afa178ecc83b823d4382ec632b25279 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e8c2ea3b9109044ae45fddf035a4c231622ec626002cb7b6fa5172f229ce53fc +size 4033338160 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..f2b32b124484daf1a3d8993d871c5d484597f6bc --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d2916c687b512eaf7ebba6e8bd88bbe285d9579c94cd1e7bdd3aa511d7954a70 +size 4033337712 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/mp_rank_00_model_states.pt b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..8b2f83897249332f321c95b85811ec95e4521661 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/global_step50000/mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b6a76a3a4eb76a0228e956f75589e4b6fe93e736d09ffcf4f4cf309dacf3699c +size 5378167832 diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/latest b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/latest new file mode 100644 index 0000000000000000000000000000000000000000..59a558c0ee32a055bd8fc16bf0d99a5950c59ac1 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/latest @@ -0,0 +1 @@ +global_step50000 \ No newline at end of file diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/zero_to_fp32.py b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/zero_to_fp32.py new file mode 100644 index 0000000000000000000000000000000000000000..f40f1f0acbbff30e8e8d29906b56aae4e8b8a138 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final/zero_to_fp32.py @@ -0,0 +1,790 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: +# python zero_to_fp32.py . output_dir/ +# or +# python zero_to_fp32.py . output_dir/ --safe_serialization + +import argparse +import torch +import glob +import math +import os +import re +import gc +import json +import numpy as np +from tqdm import tqdm +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS, AUTOEP_LAYERS_KEY, + AUTOEP_LAYERS_KEY_LEGACY, AUTOEP_ZERO3_EXPERT_STATE_FORMAT_KEY, + AUTOEP_ZERO3_PARTITIONED_EXPERT_STATE_FORMAT) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def _has_autoep_zero3_partitioned_metadata(state_dict): + autoep_layers = state_dict.get(AUTOEP_LAYERS_KEY) + if autoep_layers is None: + autoep_layers = state_dict.get(AUTOEP_LAYERS_KEY_LEGACY) + if not isinstance(autoep_layers, list): + return False + return any( + isinstance(entry, dict) + and entry.get(AUTOEP_ZERO3_EXPERT_STATE_FORMAT_KEY) == AUTOEP_ZERO3_PARTITIONED_EXPERT_STATE_FORMAT + for entry in autoep_layers) + + +def _raise_if_autoep_zero3_partitioned_state(state_dict): + if _has_autoep_zero3_partitioned_metadata(state_dict): + raise NotImplementedError("zero_to_fp32 does not support AutoEP ZeRO-3 partition-native checkpoints. " + "AutoEP expert parameters are partitioned over expert replica groups, so " + "global data-parallel consolidation would produce incomplete expert tensors. " + "Use ds_to_universal.py for expert-aware conversion.") + + +def _raise_if_autoep_zero3_partitioned_checkpoint(model_files): + for file in model_files: + state_dict = torch.load(file, map_location=device, weights_only=False) + _raise_if_autoep_zero3_partitioned_state(state_dict) + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device, weights_only=False) + _raise_if_autoep_zero3_partitioned_state(state_dict) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + total_files = len(files) + state_dicts = [] + for f in tqdm(files, desc='Loading checkpoint shards'): + state_dict = torch.load(f, map_location=device, mmap=True, weights_only=False) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if ZERO_STAGE not in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + # parse_model_states rejects AutoEP ZeRO-3 partition-native checkpoints + # before the expensive optimizer-shard load below. + model_files = get_model_state_files(ds_checkpoint_dir) + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +class GatheredTensor: + """ + A pseudo tensor that collects partitioned weights. + It is more memory efficient when there are multiple groups. + """ + + def __init__(self, flat_groups, flat_groups_offset, offset, partitioned_numel, shape): + self.flat_groups = flat_groups + self.flat_groups_offset = flat_groups_offset + self.offset = offset + self.partitioned_numel = partitioned_numel + self.shape = shape + self.dtype = self.flat_groups[0][0].dtype + + def contiguous(self): + """ + Merge partitioned weights from flat_groups into a single tensor. + """ + end_idx = self.offset + self.partitioned_numel + world_size = len(self.flat_groups) + pad_flat_param_chunks = [] + + for rank_i in range(world_size): + # for each rank, we need to collect weights from related group/groups + flat_groups_at_rank_i = self.flat_groups[rank_i] + start_group_id = None + end_group_id = None + for group_id in range(len(self.flat_groups_offset)): + if self.flat_groups_offset[group_id] <= self.offset < self.flat_groups_offset[group_id + 1]: + start_group_id = group_id + if self.flat_groups_offset[group_id] < end_idx <= self.flat_groups_offset[group_id + 1]: + end_group_id = group_id + break + # collect weights from related group/groups + for group_id in range(start_group_id, end_group_id + 1): + flat_tensor = flat_groups_at_rank_i[group_id] + start_offset = self.offset - self.flat_groups_offset[group_id] + end_offset = min(end_idx, self.flat_groups_offset[group_id + 1]) - self.flat_groups_offset[group_id] + pad_flat_param_chunks.append(flat_tensor[start_offset:end_offset]) + + # collect weights from all ranks + pad_flat_param = torch.cat(pad_flat_param_chunks, dim=0) + param = pad_flat_param[:self.shape.numel()].view(self.shape).contiguous() + return param + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = sum([flat_group.numel() for flat_group in fp32_flat_groups[0]]) * world_size + + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + flat_groups_offset = [0] + list(np.cumsum([flat_tensor.numel() for flat_tensor in fp32_flat_groups[0]])) + for name, shape in tqdm(param_shapes.items(), desc='Gathering sharded weights'): + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # memory efficient tensor + tensor = GatheredTensor(fp32_flat_groups, flat_groups_offset, offset, partitioned_numel, shape) + state_dict[name] = tensor + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def to_torch_tensor(state_dict, return_empty_tensor=False): + """ + Convert state_dict of GatheredTensor to torch tensor + """ + torch_state_dict = {} + converted_tensors = {} + for name, tensor in state_dict.items(): + tensor_id = id(tensor) + if tensor_id in converted_tensors: # shared tensors + shared_tensor = torch_state_dict[converted_tensors[tensor_id]] + torch_state_dict[name] = shared_tensor + else: + converted_tensors[tensor_id] = name + if return_empty_tensor: + torch_state_dict[name] = torch.empty(tensor.shape, dtype=tensor.dtype) + else: + torch_state_dict[name] = tensor.contiguous() + return torch_state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, + tag=None, + exclude_frozen_parameters=False, + lazy_mode=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + - ``lazy_mode``: get state_dict in lazy mode. It returns a dict of pesduo tensor instead of torch tensor, which is more memory efficient. + Convert the pesduo tensor to torch tensor by ``.contiguous()`` + + Returns: + - pytorch ``state_dict`` + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + Note: the above usage may not work if your application doesn't have sufficient free CPU memory. + You may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. Or you can load state_dict in lazy mode :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, lazy_mode=True) # not on cpu + for name, lazy_tensor in state_dict.item(): + tensor = lazy_tensor.contiguous() # to cpu + print(name, tensor) + # del tensor to release memory if it no longer in use + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + state_dict = _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + if lazy_mode: + return state_dict + else: + return to_torch_tensor(state_dict) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, + output_dir, + max_shard_size="5GB", + safe_serialization=False, + tag=None, + exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_dir``: directory to the pytorch fp32 state_dict output files + - ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB + - ``safe_serialization``: whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`). + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + # Dependency pre-check + if safe_serialization: + try: + from safetensors.torch import save_file + except ImportError: + print('If you want to use `safe_serialization`, please `pip install safetensors`') + raise + if max_shard_size is not None: + try: + from huggingface_hub import split_torch_state_dict_into_shards + except ImportError: + print('If you want to use `max_shard_size`, please `pip install huggingface_hub`') + raise + + # Convert zero checkpoint to state_dict + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, + tag, + exclude_frozen_parameters, + lazy_mode=True) + + # Shard the model if it is too big. + weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin" + if max_shard_size is not None: + filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors") + # an memory-efficient approach for sharding + empty_state_dict = to_torch_tensor(state_dict, return_empty_tensor=True) + state_dict_split = split_torch_state_dict_into_shards(empty_state_dict, + filename_pattern=filename_pattern, + max_shard_size=max_shard_size) + else: + from collections import namedtuple + StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"]) + state_dict_split = StateDictSplit(is_sharded=False, + filename_to_tensors={weights_name: list(state_dict.keys())}) + + # Save the model by shard + os.makedirs(output_dir, exist_ok=True) + filename_to_tensors = state_dict_split.filename_to_tensors.items() + for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"): + shard_state_dict = {tensor_name: state_dict[tensor_name] for tensor_name in tensors} + shard_state_dict = to_torch_tensor(shard_state_dict) + output_path = os.path.join(output_dir, shard_file) + if safe_serialization: + save_file(shard_state_dict, output_path, metadata={"format": "pt"}) + else: + torch.save(shard_state_dict, output_path) + # release the memory of current shard + for tensor_name in list(shard_state_dict.keys()): + del state_dict[tensor_name] + del shard_state_dict[tensor_name] + del shard_state_dict + gc.collect() + + # Save index if sharded + if state_dict_split.is_sharded: + index = { + "metadata": state_dict_split.metadata, + "weight_map": state_dict_split.tensor_to_filename, + } + save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json" + save_index_file = os.path.join(output_dir, save_index_file) + with open(save_index_file, "w", encoding="utf-8") as f: + content = json.dumps(index, indent=2, sort_keys=True) + "\n" + f.write(content) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info("Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info("Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument("output_dir", + type=str, + help="directory to the pytorch fp32 state_dict output files" + "(e.g. path/checkpoint-12-output/)") + parser.add_argument( + "--max_shard_size", + type=str, + default="5GB", + help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size" + "lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`" + "We default it to 5GB in order for models to be able to run easily on free-tier google colab instances" + "without CPU OOM issues.") + parser.add_argument( + "--safe_serialization", + default=False, + action='store_true', + help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_dir, + max_shard_size=args.max_shard_size, + safe_serialization=args.safe_serialization, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/train.log b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/train.log new file mode 100644 index 0000000000000000000000000000000000000000..c198b85ce31a0333a8ac465d45a68c8328ca7bb5 --- /dev/null +++ b/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/train.log @@ -0,0 +1,5033 @@ +2026-09-20 05:54:07 | INFO | train rank=0 | Starting training project=codebase_v8_droid_joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset save_dir=/data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset distributed=True deepspeed=True world_size=8 device=cuda:0 +2026-09-20 05:54:30 | INFO | train rank=0 | No pretrained checkpoint provided; training from scratch +2026-09-20 05:54:30 | INFO | train rank=0 | Initializing Droid LeRobot Dataset: path=/data/NTU_slab/draven/data/DROID_dreamzero_train fps=15.00 use_relabelled_instruction=False +2026-09-20 05:54:30 | INFO | train rank=0 | Batch config: total_batch_size=128 world_size=8 grad_acc_steps=1 per_device_batch_size=16 +2026-09-20 05:54:30 | INFO | train rank=0 | Dataset image_resize_size=320 +2026-09-20 05:54:30 | INFO | train rank=0 | Droid uses parquet-shard streaming with a per-rank shuffle buffer: batch_size=16 buffer_size=1024 num_workers=8 drop_last=True +2026-09-20 05:59:04 | INFO | train rank=0 | Starting training project=codebase_v8_droid_joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset save_dir=/data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset distributed=True deepspeed=True world_size=8 device=cuda:0 +2026-09-20 05:59:27 | INFO | train rank=0 | No pretrained checkpoint provided; training from scratch +2026-09-20 05:59:27 | INFO | train rank=0 | Initializing Droid LeRobot Dataset: path=/data/NTU_slab/draven/data/DROID_dreamzero_train fps=15.00 use_relabelled_instruction=False +2026-09-20 05:59:27 | INFO | train rank=0 | Batch config: total_batch_size=128 world_size=8 grad_acc_steps=1 per_device_batch_size=16 +2026-09-20 05:59:27 | INFO | train rank=0 | Dataset image_resize_size=320 +2026-09-20 05:59:27 | INFO | train rank=0 | Droid uses parquet-shard streaming with a per-rank shuffle buffer: batch_size=16 buffer_size=1024 num_workers=8 drop_last=True +2026-09-20 06:06:18 | INFO | train rank=0 | Starting training project=codebase_v8_droid_joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset save_dir=/data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset distributed=True deepspeed=True world_size=8 device=cuda:0 +2026-09-20 06:06:41 | INFO | train rank=0 | No pretrained checkpoint provided; training from scratch +2026-09-20 06:06:41 | INFO | train rank=0 | Initializing Droid LeRobot Dataset: path=/data/NTU_slab/draven/data/DROID_dreamzero_train fps=15.00 use_relabelled_instruction=False +2026-09-20 06:06:41 | INFO | train rank=0 | Batch config: total_batch_size=128 world_size=8 grad_acc_steps=1 per_device_batch_size=16 +2026-09-20 06:06:41 | INFO | train rank=0 | Dataset image_resize_size=320 +2026-09-20 06:06:41 | INFO | train rank=0 | Droid uses parquet-shard streaming with a per-rank shuffle buffer: batch_size=16 buffer_size=1024 num_workers=8 drop_last=True +2026-09-20 06:09:33 | INFO | train rank=0 | Starting training project=codebase_v8_droid_joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset save_dir=/data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset distributed=True deepspeed=True world_size=8 device=cuda:0 +2026-09-20 06:31:51 | INFO | train rank=0 | Starting training project=codebase_v8_droid_joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset save_dir=/data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset distributed=True deepspeed=True world_size=8 device=cuda:0 +2026-09-20 06:32:14 | INFO | train rank=0 | No pretrained checkpoint provided; training from scratch +2026-09-20 06:32:14 | INFO | train rank=0 | Initializing Droid LeRobot Dataset: path=/data/NTU_slab/draven/data/DROID_dreamzero_train fps=15.00 use_relabelled_instruction=False +2026-09-20 06:32:14 | INFO | train rank=0 | Batch config: total_batch_size=128 world_size=8 grad_acc_steps=1 per_device_batch_size=16 +2026-09-20 06:32:14 | INFO | train rank=0 | Dataset image_resize_size=320 +2026-09-20 06:32:14 | INFO | train rank=0 | Droid uses parquet-shard streaming with a per-rank shuffle buffer: batch_size=16 buffer_size=1024 num_workers=8 drop_last=True +2026-09-20 06:35:06 | INFO | train rank=0 | Droid streaming layout: parquet_shards=57774 active_workers=8 worker_buffer_min=128 worker_buffer_max=128 max_active_shards=2 +2026-09-20 06:35:06 | INFO | train rank=0 | Resolved Droid joint action stats from /data/NTU_slab/draven/data/DROID_dreamzero_train/meta/joint_delta_stats_h16.json: q01=[-0.30741822719573975, -0.4808177649974823, -0.3037826418876648, -0.4987107813358307, -0.4846664369106293, -0.4581000804901123, -0.5548821091651917] q99=[0.302254855632782, 0.5667306780815125, 0.3019354045391083, 0.5085083246231079, 0.47911858558654785, 0.533471405506134, 0.5613947510719299] +2026-09-20 06:35:06 | INFO | train rank=0 | Trainable params: 2,688,879,624 / 2,688,879,624 +2026-09-20 06:38:00 | INFO | train rank=0 | Starting training start_step=0 max_steps=50000 remaining_steps=50000 log_interval=10 save_interval=50000 +2026-09-20 06:40:03 | INFO | train rank=0 | step=10/50000 progress=0.0% loss=1.127002 lr=3.600000e-07 lr_backbone=3.600000e-07 elapsed=00:02:02 avg_step=12.29s recent_step=12.29s eta=170:42:57 +2026-09-20 06:40:14 | INFO | train rank=0 | step=20/50000 progress=0.0% loss=1.219113 lr=7.600000e-07 lr_backbone=7.600000e-07 elapsed=00:02:14 avg_step=6.72s recent_step=1.15s eta=93:18:32 +2026-09-20 06:40:30 | INFO | train rank=0 | step=30/50000 progress=0.1% loss=1.207310 lr=1.160000e-06 lr_backbone=1.160000e-06 elapsed=00:02:29 avg_step=5.00s recent_step=1.55s eta=69:20:57 +2026-09-20 06:40:45 | INFO | train rank=0 | step=40/50000 progress=0.1% loss=1.186673 lr=1.560000e-06 lr_backbone=1.560000e-06 elapsed=00:02:44 avg_step=4.12s recent_step=1.49s eta=57:10:46 +2026-09-20 06:41:08 | INFO | train rank=0 | step=50/50000 progress=0.1% loss=1.186347 lr=1.960000e-06 lr_backbone=1.960000e-06 elapsed=00:03:07 avg_step=3.76s recent_step=2.31s eta=52:09:13 +2026-09-20 06:41:19 | INFO | train rank=0 | step=60/50000 progress=0.1% loss=1.192854 lr=2.360000e-06 lr_backbone=2.360000e-06 elapsed=00:03:18 avg_step=3.31s recent_step=1.08s eta=45:56:27 +2026-09-20 06:41:34 | INFO | train rank=0 | step=70/50000 progress=0.1% loss=1.165909 lr=2.760000e-06 lr_backbone=2.760000e-06 elapsed=00:03:34 avg_step=3.06s recent_step=1.54s eta=42:25:27 +2026-09-20 06:41:46 | INFO | train rank=0 | step=80/50000 progress=0.2% loss=1.158024 lr=3.160000e-06 lr_backbone=3.160000e-06 elapsed=00:03:46 avg_step=2.83s recent_step=1.21s eta=39:12:56 +2026-09-20 06:42:06 | INFO | train rank=0 | step=90/50000 progress=0.2% loss=1.136001 lr=3.560000e-06 lr_backbone=3.560000e-06 elapsed=00:04:06 avg_step=2.74s recent_step=2.03s eta=37:58:35 +2026-09-20 06:42:18 | INFO | train rank=0 | step=100/50000 progress=0.2% loss=1.126919 lr=3.960000e-06 lr_backbone=3.960000e-06 elapsed=00:04:18 avg_step=2.58s recent_step=1.15s eta=35:45:52 +2026-09-20 06:42:30 | INFO | train rank=0 | step=110/50000 progress=0.2% loss=1.065790 lr=4.360000e-06 lr_backbone=4.360000e-06 elapsed=00:04:30 avg_step=2.46s recent_step=1.22s eta=34:02:56 +2026-09-20 06:42:42 | INFO | train rank=0 | step=120/50000 progress=0.2% loss=1.122360 lr=4.760000e-06 lr_backbone=4.760000e-06 elapsed=00:04:42 avg_step=2.35s recent_step=1.20s eta=32:35:10 +2026-09-20 06:43:01 | INFO | train rank=0 | step=130/50000 progress=0.3% loss=1.013085 lr=5.160000e-06 lr_backbone=5.160000e-06 elapsed=00:05:01 avg_step=2.32s recent_step=1.91s eta=32:06:40 +2026-09-20 06:43:21 | INFO | train rank=0 | step=140/50000 progress=0.3% loss=1.059067 lr=5.560000e-06 lr_backbone=5.560000e-06 elapsed=00:05:20 avg_step=2.29s recent_step=1.94s eta=31:44:03 +2026-09-20 06:43:32 | INFO | train rank=0 | step=150/50000 progress=0.3% loss=1.028393 lr=5.960000e-06 lr_backbone=5.960000e-06 elapsed=00:05:32 avg_step=2.22s recent_step=1.17s eta=30:41:24 +2026-09-20 06:43:46 | INFO | train rank=0 | step=160/50000 progress=0.3% loss=0.992907 lr=6.360000e-06 lr_backbone=6.360000e-06 elapsed=00:05:45 avg_step=2.16s recent_step=1.33s eta=29:55:08 +2026-09-20 06:44:04 | INFO | train rank=0 | step=170/50000 progress=0.3% loss=1.039169 lr=6.760000e-06 lr_backbone=6.760000e-06 elapsed=00:06:03 avg_step=2.14s recent_step=1.79s eta=29:36:43 +2026-09-20 06:44:17 | INFO | train rank=0 | step=180/50000 progress=0.4% loss=0.924833 lr=7.160000e-06 lr_backbone=7.160000e-06 elapsed=00:06:17 avg_step=2.10s recent_step=1.36s eta=29:00:23 +2026-09-20 06:44:33 | INFO | train rank=0 | step=190/50000 progress=0.4% loss=0.902046 lr=7.560000e-06 lr_backbone=7.560000e-06 elapsed=00:06:32 avg_step=2.07s recent_step=1.56s eta=28:36:45 +2026-09-20 06:44:47 | INFO | train rank=0 | step=200/50000 progress=0.4% loss=0.905121 lr=7.960000e-06 lr_backbone=7.960000e-06 elapsed=00:06:46 avg_step=2.03s recent_step=1.39s eta=28:08:10 +2026-09-20 06:45:03 | INFO | train rank=0 | step=210/50000 progress=0.4% loss=0.832396 lr=8.360000e-06 lr_backbone=8.360000e-06 elapsed=00:07:02 avg_step=2.01s recent_step=1.61s eta=27:51:11 +2026-09-20 06:45:15 | INFO | train rank=0 | step=220/50000 progress=0.4% loss=0.719027 lr=8.760000e-06 lr_backbone=8.760000e-06 elapsed=00:07:15 avg_step=1.98s recent_step=1.22s eta=27:20:44 +2026-09-20 06:45:26 | INFO | train rank=0 | step=230/50000 progress=0.5% loss=0.720426 lr=9.160000e-06 lr_backbone=9.160000e-06 elapsed=00:07:26 avg_step=1.94s recent_step=1.15s eta=26:50:32 +2026-09-20 06:45:46 | INFO | train rank=0 | step=240/50000 progress=0.5% loss=0.657993 lr=9.560000e-06 lr_backbone=9.560000e-06 elapsed=00:07:45 avg_step=1.94s recent_step=1.91s eta=26:49:10 +2026-09-20 06:45:57 | INFO | train rank=0 | step=250/50000 progress=0.5% loss=0.562300 lr=9.960000e-06 lr_backbone=9.960000e-06 elapsed=00:07:57 avg_step=1.91s recent_step=1.18s eta=26:23:47 +2026-09-20 06:46:15 | INFO | train rank=0 | step=260/50000 progress=0.5% loss=0.488119 lr=1.036000e-05 lr_backbone=1.036000e-05 elapsed=00:08:14 avg_step=1.90s recent_step=1.72s eta=26:17:24 +2026-09-20 06:46:39 | INFO | train rank=0 | step=270/50000 progress=0.5% loss=0.494312 lr=1.076000e-05 lr_backbone=1.076000e-05 elapsed=00:08:38 avg_step=1.92s recent_step=2.42s eta=26:32:52 +2026-09-20 06:46:49 | INFO | train rank=0 | step=280/50000 progress=0.6% loss=0.402690 lr=1.116000e-05 lr_backbone=1.116000e-05 elapsed=00:08:49 avg_step=1.89s recent_step=1.06s eta=26:07:02 +2026-09-20 06:47:02 | INFO | train rank=0 | step=290/50000 progress=0.6% loss=0.362849 lr=1.156000e-05 lr_backbone=1.156000e-05 elapsed=00:09:01 avg_step=1.87s recent_step=1.23s eta=25:47:50 +2026-09-20 06:47:13 | INFO | train rank=0 | step=300/50000 progress=0.6% loss=0.384340 lr=1.196000e-05 lr_backbone=1.196000e-05 elapsed=00:09:13 avg_step=1.84s recent_step=1.15s eta=25:27:45 +2026-09-20 06:47:29 | INFO | train rank=0 | step=310/50000 progress=0.6% loss=0.322729 lr=1.236000e-05 lr_backbone=1.236000e-05 elapsed=00:09:29 avg_step=1.84s recent_step=1.59s eta=25:20:38 +2026-09-20 06:47:42 | INFO | train rank=0 | step=320/50000 progress=0.6% loss=0.451901 lr=1.276000e-05 lr_backbone=1.276000e-05 elapsed=00:09:42 avg_step=1.82s recent_step=1.28s eta=25:06:03 +2026-09-20 06:47:57 | INFO | train rank=0 | step=330/50000 progress=0.7% loss=0.314606 lr=1.316000e-05 lr_backbone=1.316000e-05 elapsed=00:09:57 avg_step=1.81s recent_step=1.51s eta=24:58:06 +2026-09-20 06:48:18 | INFO | train rank=0 | step=340/50000 progress=0.7% loss=0.323738 lr=1.356000e-05 lr_backbone=1.356000e-05 elapsed=00:10:18 avg_step=1.82s recent_step=2.14s eta=25:05:51 +2026-09-20 06:48:32 | INFO | train rank=0 | step=350/50000 progress=0.7% loss=0.244703 lr=1.396000e-05 lr_backbone=1.396000e-05 elapsed=00:10:32 avg_step=1.81s recent_step=1.36s eta=24:54:39 +2026-09-20 06:48:48 | INFO | train rank=0 | step=360/50000 progress=0.7% loss=0.256034 lr=1.436000e-05 lr_backbone=1.436000e-05 elapsed=00:10:48 avg_step=1.80s recent_step=1.62s eta=24:50:06 +2026-09-20 06:49:00 | INFO | train rank=0 | step=370/50000 progress=0.7% loss=0.356262 lr=1.476000e-05 lr_backbone=1.476000e-05 elapsed=00:10:59 avg_step=1.78s recent_step=1.13s eta=24:34:50 +2026-09-20 06:49:18 | INFO | train rank=0 | step=380/50000 progress=0.8% loss=0.262284 lr=1.516000e-05 lr_backbone=1.516000e-05 elapsed=00:11:18 avg_step=1.79s recent_step=1.89s eta=24:36:50 +2026-09-20 06:49:37 | INFO | train rank=0 | step=390/50000 progress=0.8% loss=0.327675 lr=1.556000e-05 lr_backbone=1.556000e-05 elapsed=00:11:36 avg_step=1.79s recent_step=1.80s eta=24:36:54 +2026-09-20 06:49:49 | INFO | train rank=0 | step=400/50000 progress=0.8% loss=0.243571 lr=1.596000e-05 lr_backbone=1.596000e-05 elapsed=00:11:48 avg_step=1.77s recent_step=1.22s eta=24:24:56 +2026-09-20 06:50:01 | INFO | train rank=0 | step=410/50000 progress=0.8% loss=0.269397 lr=1.636000e-05 lr_backbone=1.636000e-05 elapsed=00:12:00 avg_step=1.76s recent_step=1.20s eta=24:13:02 +2026-09-20 06:50:19 | INFO | train rank=0 | step=420/50000 progress=0.8% loss=0.277618 lr=1.676000e-05 lr_backbone=1.676000e-05 elapsed=00:12:18 avg_step=1.76s recent_step=1.81s eta=24:13:48 +2026-09-20 06:50:36 | INFO | train rank=0 | step=430/50000 progress=0.9% loss=0.287273 lr=1.716000e-05 lr_backbone=1.716000e-05 elapsed=00:12:36 avg_step=1.76s recent_step=1.73s eta=24:12:59 +2026-09-20 06:50:49 | INFO | train rank=0 | step=440/50000 progress=0.9% loss=0.236167 lr=1.756000e-05 lr_backbone=1.756000e-05 elapsed=00:12:49 avg_step=1.75s recent_step=1.33s eta=24:04:40 +2026-09-20 06:51:04 | INFO | train rank=0 | step=450/50000 progress=0.9% loss=0.201279 lr=1.796000e-05 lr_backbone=1.796000e-05 elapsed=00:13:03 avg_step=1.74s recent_step=1.43s eta=23:58:36 +2026-09-20 06:51:23 | INFO | train rank=0 | step=460/50000 progress=0.9% loss=0.197422 lr=1.836000e-05 lr_backbone=1.836000e-05 elapsed=00:13:23 avg_step=1.75s recent_step=1.93s eta=24:01:39 +2026-09-20 06:51:36 | INFO | train rank=0 | step=470/50000 progress=0.9% loss=0.227404 lr=1.876000e-05 lr_backbone=1.876000e-05 elapsed=00:13:36 avg_step=1.74s recent_step=1.29s eta=23:53:25 +2026-09-20 06:51:47 | INFO | train rank=0 | step=480/50000 progress=1.0% loss=0.200343 lr=1.916000e-05 lr_backbone=1.916000e-05 elapsed=00:13:47 avg_step=1.72s recent_step=1.11s eta=23:42:25 +2026-09-20 06:52:02 | INFO | train rank=0 | step=490/50000 progress=1.0% loss=0.218990 lr=1.956000e-05 lr_backbone=1.956000e-05 elapsed=00:14:01 avg_step=1.72s recent_step=1.46s eta=23:37:38 +2026-09-20 06:52:16 | INFO | train rank=0 | step=500/50000 progress=1.0% loss=0.202294 lr=1.996000e-05 lr_backbone=1.996000e-05 elapsed=00:14:16 avg_step=1.71s recent_step=1.46s eta=23:33:10 +2026-09-20 06:52:34 | INFO | train rank=0 | step=510/50000 progress=1.0% loss=0.223754 lr=2.036000e-05 lr_backbone=2.036000e-05 elapsed=00:14:33 avg_step=1.71s recent_step=1.73s eta=23:33:13 +2026-09-20 06:52:50 | INFO | train rank=0 | step=520/50000 progress=1.0% loss=0.245861 lr=2.076000e-05 lr_backbone=2.076000e-05 elapsed=00:14:50 avg_step=1.71s recent_step=1.65s eta=23:31:52 +2026-09-20 06:53:04 | INFO | train rank=0 | step=530/50000 progress=1.1% loss=0.217831 lr=2.116000e-05 lr_backbone=2.116000e-05 elapsed=00:15:04 avg_step=1.71s recent_step=1.39s eta=23:26:39 +2026-09-20 06:53:16 | INFO | train rank=0 | step=540/50000 progress=1.1% loss=0.214818 lr=2.156000e-05 lr_backbone=2.156000e-05 elapsed=00:15:15 avg_step=1.70s recent_step=1.18s eta=23:18:17 +2026-09-20 06:53:38 | INFO | train rank=0 | step=550/50000 progress=1.1% loss=0.199835 lr=2.196000e-05 lr_backbone=2.196000e-05 elapsed=00:15:38 avg_step=1.71s recent_step=2.23s eta=23:26:03 +2026-09-20 06:53:50 | INFO | train rank=0 | step=560/50000 progress=1.1% loss=0.235483 lr=2.236000e-05 lr_backbone=2.236000e-05 elapsed=00:15:49 avg_step=1.70s recent_step=1.14s eta=23:17:24 +2026-09-20 06:54:02 | INFO | train rank=0 | step=570/50000 progress=1.1% loss=0.200940 lr=2.276000e-05 lr_backbone=2.276000e-05 elapsed=00:16:01 avg_step=1.69s recent_step=1.20s eta=23:09:59 +2026-09-20 06:54:21 | INFO | train rank=0 | step=580/50000 progress=1.2% loss=0.190732 lr=2.316000e-05 lr_backbone=2.316000e-05 elapsed=00:16:21 avg_step=1.69s recent_step=1.93s eta=23:13:09 +2026-09-20 06:54:34 | INFO | train rank=0 | step=590/50000 progress=1.2% loss=0.232949 lr=2.356000e-05 lr_backbone=2.356000e-05 elapsed=00:16:34 avg_step=1.68s recent_step=1.31s eta=23:07:34 +2026-09-20 06:54:53 | INFO | train rank=0 | step=600/50000 progress=1.2% loss=0.212342 lr=2.396000e-05 lr_backbone=2.396000e-05 elapsed=00:16:52 avg_step=1.69s recent_step=1.85s eta=23:09:33 +2026-09-20 06:55:06 | INFO | train rank=0 | step=610/50000 progress=1.2% loss=0.204488 lr=2.436000e-05 lr_backbone=2.436000e-05 elapsed=00:17:06 avg_step=1.68s recent_step=1.35s eta=23:04:39 +2026-09-20 06:55:20 | INFO | train rank=0 | step=620/50000 progress=1.2% loss=0.187001 lr=2.476000e-05 lr_backbone=2.476000e-05 elapsed=00:17:19 avg_step=1.68s recent_step=1.37s eta=23:00:16 +2026-09-20 06:55:40 | INFO | train rank=0 | step=630/50000 progress=1.3% loss=0.193425 lr=2.516000e-05 lr_backbone=2.516000e-05 elapsed=00:17:40 avg_step=1.68s recent_step=2.04s eta=23:04:46 +2026-09-20 06:55:52 | INFO | train rank=0 | step=640/50000 progress=1.3% loss=0.248708 lr=2.556000e-05 lr_backbone=2.556000e-05 elapsed=00:17:52 avg_step=1.68s recent_step=1.20s eta=22:58:17 +2026-09-20 06:56:04 | INFO | train rank=0 | step=650/50000 progress=1.3% loss=0.186248 lr=2.596000e-05 lr_backbone=2.596000e-05 elapsed=00:18:03 avg_step=1.67s recent_step=1.16s eta=22:51:26 +2026-09-20 06:56:16 | INFO | train rank=0 | step=660/50000 progress=1.3% loss=0.168351 lr=2.636000e-05 lr_backbone=2.636000e-05 elapsed=00:18:16 avg_step=1.66s recent_step=1.25s eta=22:45:55 +2026-09-20 06:56:29 | INFO | train rank=0 | step=670/50000 progress=1.3% loss=0.185081 lr=2.676000e-05 lr_backbone=2.676000e-05 elapsed=00:18:29 avg_step=1.66s recent_step=1.32s eta=22:41:28 +2026-09-20 06:56:47 | INFO | train rank=0 | step=680/50000 progress=1.4% loss=0.310955 lr=2.716000e-05 lr_backbone=2.716000e-05 elapsed=00:18:47 avg_step=1.66s recent_step=1.77s eta=22:42:37 +2026-09-20 06:57:00 | INFO | train rank=0 | step=690/50000 progress=1.4% loss=0.185853 lr=2.756000e-05 lr_backbone=2.756000e-05 elapsed=00:18:59 avg_step=1.65s recent_step=1.25s eta=22:37:27 +2026-09-20 06:57:18 | INFO | train rank=0 | step=700/50000 progress=1.4% loss=0.208000 lr=2.796000e-05 lr_backbone=2.796000e-05 elapsed=00:19:18 avg_step=1.65s recent_step=1.85s eta=22:39:30 +2026-09-20 06:57:33 | INFO | train rank=0 | step=710/50000 progress=1.4% loss=0.171967 lr=2.836000e-05 lr_backbone=2.836000e-05 elapsed=00:19:32 avg_step=1.65s recent_step=1.45s eta=22:36:54 +2026-09-20 06:57:53 | INFO | train rank=0 | step=720/50000 progress=1.4% loss=0.189355 lr=2.876000e-05 lr_backbone=2.876000e-05 elapsed=00:19:53 avg_step=1.66s recent_step=2.04s eta=22:41:02 +2026-09-20 06:58:04 | INFO | train rank=0 | step=730/50000 progress=1.5% loss=0.207370 lr=2.916000e-05 lr_backbone=2.916000e-05 elapsed=00:20:04 avg_step=1.65s recent_step=1.11s eta=22:34:33 +2026-09-20 06:58:18 | INFO | train rank=0 | step=740/50000 progress=1.5% loss=0.179436 lr=2.956000e-05 lr_backbone=2.956000e-05 elapsed=00:20:18 avg_step=1.65s recent_step=1.40s eta=22:31:28 +2026-09-20 06:58:36 | INFO | train rank=0 | step=750/50000 progress=1.5% loss=0.169939 lr=2.996000e-05 lr_backbone=2.996000e-05 elapsed=00:20:36 avg_step=1.65s recent_step=1.82s eta=22:33:07 +2026-09-20 06:58:51 | INFO | train rank=0 | step=760/50000 progress=1.5% loss=0.192439 lr=3.036000e-05 lr_backbone=3.036000e-05 elapsed=00:20:50 avg_step=1.65s recent_step=1.43s eta=22:30:28 +2026-09-20 06:59:05 | INFO | train rank=0 | step=770/50000 progress=1.5% loss=0.202243 lr=3.076000e-05 lr_backbone=3.076000e-05 elapsed=00:21:04 avg_step=1.64s recent_step=1.43s eta=22:27:52 +2026-09-20 06:59:19 | INFO | train rank=0 | step=780/50000 progress=1.6% loss=0.179174 lr=3.116000e-05 lr_backbone=3.116000e-05 elapsed=00:21:19 avg_step=1.64s recent_step=1.44s eta=22:25:30 +2026-09-20 06:59:37 | INFO | train rank=0 | step=790/50000 progress=1.6% loss=0.169179 lr=3.156000e-05 lr_backbone=3.156000e-05 elapsed=00:21:37 avg_step=1.64s recent_step=1.79s eta=22:26:45 +2026-09-20 06:59:49 | INFO | train rank=0 | step=800/50000 progress=1.6% loss=0.181542 lr=3.196000e-05 lr_backbone=3.196000e-05 elapsed=00:21:49 avg_step=1.64s recent_step=1.23s eta=22:22:12 +2026-09-20 07:00:08 | INFO | train rank=0 | step=810/50000 progress=1.6% loss=0.176399 lr=3.236000e-05 lr_backbone=3.236000e-05 elapsed=00:22:08 avg_step=1.64s recent_step=1.86s eta=22:24:14 +2026-09-20 07:00:20 | INFO | train rank=0 | step=820/50000 progress=1.6% loss=0.203331 lr=3.276000e-05 lr_backbone=3.276000e-05 elapsed=00:22:20 avg_step=1.63s recent_step=1.21s eta=22:19:42 +2026-09-20 07:00:37 | INFO | train rank=0 | step=830/50000 progress=1.7% loss=0.183814 lr=3.316000e-05 lr_backbone=3.316000e-05 elapsed=00:22:37 avg_step=1.64s recent_step=1.69s eta=22:19:59 +2026-09-20 07:00:52 | INFO | train rank=0 | step=840/50000 progress=1.7% loss=0.180866 lr=3.356000e-05 lr_backbone=3.356000e-05 elapsed=00:22:51 avg_step=1.63s recent_step=1.45s eta=22:17:53 +2026-09-20 07:01:08 | INFO | train rank=0 | step=850/50000 progress=1.7% loss=0.186890 lr=3.396000e-05 lr_backbone=3.396000e-05 elapsed=00:23:08 avg_step=1.63s recent_step=1.69s eta=22:18:12 +2026-09-20 07:01:21 | INFO | train rank=0 | step=860/50000 progress=1.7% loss=0.182755 lr=3.436000e-05 lr_backbone=3.436000e-05 elapsed=00:23:20 avg_step=1.63s recent_step=1.24s eta=22:14:11 +2026-09-20 07:01:33 | INFO | train rank=0 | step=870/50000 progress=1.7% loss=0.199562 lr=3.476000e-05 lr_backbone=3.476000e-05 elapsed=00:23:32 avg_step=1.62s recent_step=1.17s eta=22:09:38 +2026-09-20 07:01:49 | INFO | train rank=0 | step=880/50000 progress=1.8% loss=0.186229 lr=3.516000e-05 lr_backbone=3.516000e-05 elapsed=00:23:49 avg_step=1.62s recent_step=1.65s eta=22:09:37 +2026-09-20 07:02:10 | INFO | train rank=0 | step=890/50000 progress=1.8% loss=0.203867 lr=3.556000e-05 lr_backbone=3.556000e-05 elapsed=00:24:10 avg_step=1.63s recent_step=2.09s eta=22:13:36 +2026-09-20 07:02:21 | INFO | train rank=0 | step=900/50000 progress=1.8% loss=0.222786 lr=3.596000e-05 lr_backbone=3.596000e-05 elapsed=00:24:21 avg_step=1.62s recent_step=1.14s eta=22:08:52 +2026-09-20 07:02:36 | INFO | train rank=0 | step=910/50000 progress=1.8% loss=0.216772 lr=3.636000e-05 lr_backbone=3.636000e-05 elapsed=00:24:36 avg_step=1.62s recent_step=1.49s eta=22:07:22 +2026-09-20 07:02:50 | INFO | train rank=0 | step=920/50000 progress=1.8% loss=0.166899 lr=3.676000e-05 lr_backbone=3.676000e-05 elapsed=00:24:49 avg_step=1.62s recent_step=1.34s eta=22:04:36 +2026-09-20 07:03:07 | INFO | train rank=0 | step=930/50000 progress=1.9% loss=0.201794 lr=3.716000e-05 lr_backbone=3.716000e-05 elapsed=00:25:07 avg_step=1.62s recent_step=1.74s eta=22:05:25 +2026-09-20 07:03:19 | INFO | train rank=0 | step=940/50000 progress=1.9% loss=0.187279 lr=3.756000e-05 lr_backbone=3.756000e-05 elapsed=00:25:19 avg_step=1.62s recent_step=1.21s eta=22:01:35 +2026-09-20 07:03:35 | INFO | train rank=0 | step=950/50000 progress=1.9% loss=0.181075 lr=3.796000e-05 lr_backbone=3.796000e-05 elapsed=00:25:35 avg_step=1.62s recent_step=1.61s eta=22:01:16 +2026-09-20 07:03:52 | INFO | train rank=0 | step=960/50000 progress=1.9% loss=0.227841 lr=3.836000e-05 lr_backbone=3.836000e-05 elapsed=00:25:52 avg_step=1.62s recent_step=1.69s eta=22:01:39 +2026-09-20 07:04:10 | INFO | train rank=0 | step=970/50000 progress=1.9% loss=0.185783 lr=3.876000e-05 lr_backbone=3.876000e-05 elapsed=00:26:09 avg_step=1.62s recent_step=1.76s eta=22:02:35 +2026-09-20 07:04:26 | INFO | train rank=0 | step=980/50000 progress=2.0% loss=0.181688 lr=3.916000e-05 lr_backbone=3.916000e-05 elapsed=00:26:26 avg_step=1.62s recent_step=1.63s eta=22:02:27 +2026-09-20 07:04:39 | INFO | train rank=0 | step=990/50000 progress=2.0% loss=0.176759 lr=3.956000e-05 lr_backbone=3.956000e-05 elapsed=00:26:39 avg_step=1.62s recent_step=1.28s eta=21:59:21 +2026-09-20 07:04:53 | INFO | train rank=0 | step=1000/50000 progress=2.0% loss=0.159786 lr=3.996000e-05 lr_backbone=3.996000e-05 elapsed=00:26:53 avg_step=1.61s recent_step=1.40s eta=21:57:21 +2026-09-20 07:05:12 | INFO | train rank=0 | step=1010/50000 progress=2.0% loss=0.234427 lr=4.036000e-05 lr_backbone=4.036000e-05 elapsed=00:27:12 avg_step=1.62s recent_step=1.93s eta=21:59:37 +2026-09-20 07:05:25 | INFO | train rank=0 | step=1020/50000 progress=2.0% loss=0.171368 lr=4.076000e-05 lr_backbone=4.076000e-05 elapsed=00:27:25 avg_step=1.61s recent_step=1.30s eta=21:56:47 +2026-09-20 07:05:38 | INFO | train rank=0 | step=1030/50000 progress=2.1% loss=0.180328 lr=4.116000e-05 lr_backbone=4.116000e-05 elapsed=00:27:38 avg_step=1.61s recent_step=1.28s eta=21:53:52 +2026-09-20 07:05:58 | INFO | train rank=0 | step=1040/50000 progress=2.1% loss=0.203100 lr=4.156000e-05 lr_backbone=4.156000e-05 elapsed=00:27:58 avg_step=1.61s recent_step=2.01s eta=21:56:45 +2026-09-20 07:06:10 | INFO | train rank=0 | step=1050/50000 progress=2.1% loss=0.241387 lr=4.196000e-05 lr_backbone=4.196000e-05 elapsed=00:28:10 avg_step=1.61s recent_step=1.22s eta=21:53:25 +2026-09-20 07:06:23 | INFO | train rank=0 | step=1060/50000 progress=2.1% loss=0.215397 lr=4.236000e-05 lr_backbone=4.236000e-05 elapsed=00:28:22 avg_step=1.61s recent_step=1.25s eta=21:50:21 +2026-09-20 07:06:37 | INFO | train rank=0 | step=1070/50000 progress=2.1% loss=0.203908 lr=4.276000e-05 lr_backbone=4.276000e-05 elapsed=00:28:37 avg_step=1.61s recent_step=1.45s eta=21:48:54 +2026-09-20 07:06:51 | INFO | train rank=0 | step=1080/50000 progress=2.2% loss=0.161151 lr=4.316000e-05 lr_backbone=4.316000e-05 elapsed=00:28:51 avg_step=1.60s recent_step=1.38s eta=21:46:55 +2026-09-20 07:07:10 | INFO | train rank=0 | step=1090/50000 progress=2.2% loss=0.198828 lr=4.356000e-05 lr_backbone=4.356000e-05 elapsed=00:29:10 avg_step=1.61s recent_step=1.90s eta=21:48:52 +2026-09-20 07:07:26 | INFO | train rank=0 | step=1100/50000 progress=2.2% loss=0.185647 lr=4.396000e-05 lr_backbone=4.396000e-05 elapsed=00:29:26 avg_step=1.61s recent_step=1.60s eta=21:48:34 +2026-09-20 07:07:42 | INFO | train rank=0 | step=1110/50000 progress=2.2% loss=0.180509 lr=4.436000e-05 lr_backbone=4.436000e-05 elapsed=00:29:42 avg_step=1.61s recent_step=1.59s eta=21:48:10 +2026-09-20 07:07:55 | INFO | train rank=0 | step=1120/50000 progress=2.2% loss=0.203935 lr=4.476000e-05 lr_backbone=4.476000e-05 elapsed=00:29:55 avg_step=1.60s recent_step=1.34s eta=21:46:00 +2026-09-20 07:08:10 | INFO | train rank=0 | step=1130/50000 progress=2.3% loss=0.183846 lr=4.516000e-05 lr_backbone=4.516000e-05 elapsed=00:30:10 avg_step=1.60s recent_step=1.46s eta=21:44:40 +2026-09-20 07:08:25 | INFO | train rank=0 | step=1140/50000 progress=2.3% loss=0.189963 lr=4.556000e-05 lr_backbone=4.556000e-05 elapsed=00:30:25 avg_step=1.60s recent_step=1.54s eta=21:43:58 +2026-09-20 07:08:40 | INFO | train rank=0 | step=1150/50000 progress=2.3% loss=0.192400 lr=4.596000e-05 lr_backbone=4.596000e-05 elapsed=00:30:40 avg_step=1.60s recent_step=1.49s eta=21:42:54 +2026-09-20 07:08:54 | INFO | train rank=0 | step=1160/50000 progress=2.3% loss=0.186784 lr=4.636000e-05 lr_backbone=4.636000e-05 elapsed=00:30:54 avg_step=1.60s recent_step=1.38s eta=21:41:06 +2026-09-20 07:09:10 | INFO | train rank=0 | step=1170/50000 progress=2.3% loss=0.233936 lr=4.676000e-05 lr_backbone=4.676000e-05 elapsed=00:31:10 avg_step=1.60s recent_step=1.62s eta=21:41:01 +2026-09-20 07:09:28 | INFO | train rank=0 | step=1180/50000 progress=2.4% loss=0.185876 lr=4.716000e-05 lr_backbone=4.716000e-05 elapsed=00:31:28 avg_step=1.60s recent_step=1.82s eta=21:42:15 +2026-09-20 07:09:41 | INFO | train rank=0 | step=1190/50000 progress=2.4% loss=0.173934 lr=4.756000e-05 lr_backbone=4.756000e-05 elapsed=00:31:41 avg_step=1.60s recent_step=1.25s eta=21:39:34 +2026-09-20 07:09:56 | INFO | train rank=0 | step=1200/50000 progress=2.4% loss=0.233315 lr=4.796000e-05 lr_backbone=4.796000e-05 elapsed=00:31:56 avg_step=1.60s recent_step=1.54s eta=21:38:55 +2026-09-20 07:10:16 | INFO | train rank=0 | step=1210/50000 progress=2.4% loss=0.159252 lr=4.836000e-05 lr_backbone=4.836000e-05 elapsed=00:32:16 avg_step=1.60s recent_step=1.98s eta=21:41:13 +2026-09-20 07:10:28 | INFO | train rank=0 | step=1220/50000 progress=2.4% loss=0.185380 lr=4.876000e-05 lr_backbone=4.876000e-05 elapsed=00:32:28 avg_step=1.60s recent_step=1.21s eta=21:38:21 +2026-09-20 07:10:43 | INFO | train rank=0 | step=1230/50000 progress=2.5% loss=0.194260 lr=4.916000e-05 lr_backbone=4.916000e-05 elapsed=00:32:43 avg_step=1.60s recent_step=1.51s eta=21:37:30 +2026-09-20 07:10:58 | INFO | train rank=0 | step=1240/50000 progress=2.5% loss=0.167557 lr=4.956000e-05 lr_backbone=4.956000e-05 elapsed=00:32:58 avg_step=1.60s recent_step=1.51s eta=21:36:41 +2026-09-20 07:11:11 | INFO | train rank=0 | step=1250/50000 progress=2.5% loss=0.164969 lr=4.996000e-05 lr_backbone=4.996000e-05 elapsed=00:33:10 avg_step=1.59s recent_step=1.24s eta=21:34:07 +2026-09-20 07:11:22 | INFO | train rank=0 | step=1260/50000 progress=2.5% loss=0.142605 lr=5.036000e-05 lr_backbone=5.036000e-05 elapsed=00:33:22 avg_step=1.59s recent_step=1.16s eta=21:31:04 +2026-09-20 07:11:36 | INFO | train rank=0 | step=1270/50000 progress=2.5% loss=0.178933 lr=5.076000e-05 lr_backbone=5.076000e-05 elapsed=00:33:35 avg_step=1.59s recent_step=1.31s eta=21:29:02 +2026-09-20 07:11:55 | INFO | train rank=0 | step=1280/50000 progress=2.6% loss=0.191357 lr=5.116000e-05 lr_backbone=5.116000e-05 elapsed=00:33:55 avg_step=1.59s recent_step=1.96s eta=21:31:07 +2026-09-20 07:12:11 | INFO | train rank=0 | step=1290/50000 progress=2.6% loss=0.187598 lr=5.156000e-05 lr_backbone=5.156000e-05 elapsed=00:34:11 avg_step=1.59s recent_step=1.62s eta=21:31:04 +2026-09-20 07:12:26 | INFO | train rank=0 | step=1300/50000 progress=2.6% loss=0.224580 lr=5.196000e-05 lr_backbone=5.196000e-05 elapsed=00:34:26 avg_step=1.59s recent_step=1.50s eta=21:30:14 +2026-09-20 07:12:41 | INFO | train rank=0 | step=1310/50000 progress=2.6% loss=0.167672 lr=5.236000e-05 lr_backbone=5.236000e-05 elapsed=00:34:41 avg_step=1.59s recent_step=1.51s eta=21:29:28 +2026-09-20 07:12:59 | INFO | train rank=0 | step=1320/50000 progress=2.6% loss=0.165328 lr=5.276000e-05 lr_backbone=5.276000e-05 elapsed=00:34:59 avg_step=1.59s recent_step=1.77s eta=21:30:20 +2026-09-20 07:13:16 | INFO | train rank=0 | step=1330/50000 progress=2.7% loss=0.162462 lr=5.316000e-05 lr_backbone=5.316000e-05 elapsed=00:35:15 avg_step=1.59s recent_step=1.66s eta=21:30:30 +2026-09-20 07:13:31 | INFO | train rank=0 | step=1340/50000 progress=2.7% loss=0.164461 lr=5.356000e-05 lr_backbone=5.356000e-05 elapsed=00:35:31 avg_step=1.59s recent_step=1.54s eta=21:29:55 +2026-09-20 07:13:45 | INFO | train rank=0 | step=1350/50000 progress=2.7% loss=0.155049 lr=5.396000e-05 lr_backbone=5.396000e-05 elapsed=00:35:44 avg_step=1.59s recent_step=1.33s eta=21:28:06 +2026-09-20 07:14:01 | INFO | train rank=0 | step=1360/50000 progress=2.7% loss=0.159877 lr=5.436000e-05 lr_backbone=5.436000e-05 elapsed=00:36:00 avg_step=1.59s recent_step=1.62s eta=21:28:02 +2026-09-20 07:14:13 | INFO | train rank=0 | step=1370/50000 progress=2.7% loss=0.193834 lr=5.476000e-05 lr_backbone=5.476000e-05 elapsed=00:36:13 avg_step=1.59s recent_step=1.23s eta=21:25:37 +2026-09-20 07:14:33 | INFO | train rank=0 | step=1380/50000 progress=2.8% loss=0.178422 lr=5.516000e-05 lr_backbone=5.516000e-05 elapsed=00:36:32 avg_step=1.59s recent_step=1.96s eta=21:27:33 +2026-09-20 07:14:48 | INFO | train rank=0 | step=1390/50000 progress=2.8% loss=0.272095 lr=5.556000e-05 lr_backbone=5.556000e-05 elapsed=00:36:47 avg_step=1.59s recent_step=1.49s eta=21:26:43 +2026-09-20 07:14:59 | INFO | train rank=0 | step=1400/50000 progress=2.8% loss=0.248893 lr=5.596000e-05 lr_backbone=5.596000e-05 elapsed=00:36:59 avg_step=1.59s recent_step=1.18s eta=21:24:04 +2026-09-20 07:15:15 | INFO | train rank=0 | step=1410/50000 progress=2.8% loss=0.186881 lr=5.636000e-05 lr_backbone=5.636000e-05 elapsed=00:37:14 avg_step=1.58s recent_step=1.53s eta=21:23:29 +2026-09-20 07:15:30 | INFO | train rank=0 | step=1420/50000 progress=2.8% loss=0.154738 lr=5.676000e-05 lr_backbone=5.676000e-05 elapsed=00:37:30 avg_step=1.58s recent_step=1.58s eta=21:23:12 +2026-09-20 07:15:50 | INFO | train rank=0 | step=1430/50000 progress=2.9% loss=0.166913 lr=5.716000e-05 lr_backbone=5.716000e-05 elapsed=00:37:49 avg_step=1.59s recent_step=1.94s eta=21:24:55 +2026-09-20 07:16:02 | INFO | train rank=0 | step=1440/50000 progress=2.9% loss=0.135641 lr=5.756000e-05 lr_backbone=5.756000e-05 elapsed=00:38:02 avg_step=1.58s recent_step=1.23s eta=21:22:39 +2026-09-20 07:16:14 | INFO | train rank=0 | step=1450/50000 progress=2.9% loss=0.131409 lr=5.796000e-05 lr_backbone=5.796000e-05 elapsed=00:38:14 avg_step=1.58s recent_step=1.24s eta=21:20:28 +2026-09-20 07:16:29 | INFO | train rank=0 | step=1460/50000 progress=2.9% loss=0.228531 lr=5.836000e-05 lr_backbone=5.836000e-05 elapsed=00:38:28 avg_step=1.58s recent_step=1.44s eta=21:19:24 +2026-09-20 07:16:45 | INFO | train rank=0 | step=1470/50000 progress=2.9% loss=0.190648 lr=5.876000e-05 lr_backbone=5.876000e-05 elapsed=00:38:45 avg_step=1.58s recent_step=1.64s eta=21:19:28 +2026-09-20 07:17:00 | INFO | train rank=0 | step=1480/50000 progress=3.0% loss=0.179475 lr=5.916000e-05 lr_backbone=5.916000e-05 elapsed=00:39:00 avg_step=1.58s recent_step=1.51s eta=21:18:50 +2026-09-20 07:17:19 | INFO | train rank=0 | step=1490/50000 progress=3.0% loss=0.246948 lr=5.956000e-05 lr_backbone=5.956000e-05 elapsed=00:39:19 avg_step=1.58s recent_step=1.85s eta=21:20:03 +2026-09-20 07:17:32 | INFO | train rank=0 | step=1500/50000 progress=3.0% loss=0.165627 lr=5.996000e-05 lr_backbone=5.996000e-05 elapsed=00:39:31 avg_step=1.58s recent_step=1.26s eta=21:18:02 +2026-09-20 07:17:48 | INFO | train rank=0 | step=1510/50000 progress=3.0% loss=0.138217 lr=6.036000e-05 lr_backbone=6.036000e-05 elapsed=00:39:47 avg_step=1.58s recent_step=1.61s eta=21:17:54 +2026-09-20 07:18:03 | INFO | train rank=0 | step=1520/50000 progress=3.0% loss=0.220449 lr=6.076000e-05 lr_backbone=6.076000e-05 elapsed=00:40:02 avg_step=1.58s recent_step=1.53s eta=21:17:23 +2026-09-20 07:18:18 | INFO | train rank=0 | step=1530/50000 progress=3.1% loss=0.168322 lr=6.116000e-05 lr_backbone=6.116000e-05 elapsed=00:40:17 avg_step=1.58s recent_step=1.48s eta=21:16:36 +2026-09-20 07:18:37 | INFO | train rank=0 | step=1540/50000 progress=3.1% loss=0.166664 lr=6.156000e-05 lr_backbone=6.156000e-05 elapsed=00:40:36 avg_step=1.58s recent_step=1.89s eta=21:17:57 +2026-09-20 07:18:54 | INFO | train rank=0 | step=1550/50000 progress=3.1% loss=0.169502 lr=6.196000e-05 lr_backbone=6.196000e-05 elapsed=00:40:53 avg_step=1.58s recent_step=1.70s eta=21:18:19 +2026-09-20 07:19:06 | INFO | train rank=0 | step=1560/50000 progress=3.1% loss=0.141259 lr=6.236000e-05 lr_backbone=6.236000e-05 elapsed=00:41:06 avg_step=1.58s recent_step=1.27s eta=21:16:26 +2026-09-20 07:19:20 | INFO | train rank=0 | step=1570/50000 progress=3.1% loss=0.166288 lr=6.276000e-05 lr_backbone=6.276000e-05 elapsed=00:41:20 avg_step=1.58s recent_step=1.37s eta=21:15:05 +2026-09-20 07:19:39 | INFO | train rank=0 | step=1580/50000 progress=3.2% loss=0.186862 lr=6.316000e-05 lr_backbone=6.316000e-05 elapsed=00:41:38 avg_step=1.58s recent_step=1.86s eta=21:16:16 +2026-09-20 07:19:52 | INFO | train rank=0 | step=1590/50000 progress=3.2% loss=0.218428 lr=6.356000e-05 lr_backbone=6.356000e-05 elapsed=00:41:51 avg_step=1.58s recent_step=1.29s eta=21:14:32 +2026-09-20 07:20:06 | INFO | train rank=0 | step=1600/50000 progress=3.2% loss=0.181196 lr=6.396000e-05 lr_backbone=6.396000e-05 elapsed=00:42:05 avg_step=1.58s recent_step=1.43s eta=21:13:30 +2026-09-20 07:20:20 | INFO | train rank=0 | step=1610/50000 progress=3.2% loss=0.179848 lr=6.436000e-05 lr_backbone=6.436000e-05 elapsed=00:42:20 avg_step=1.58s recent_step=1.44s eta=21:12:32 +2026-09-20 07:20:35 | INFO | train rank=0 | step=1620/50000 progress=3.2% loss=0.164560 lr=6.476000e-05 lr_backbone=6.476000e-05 elapsed=00:42:35 avg_step=1.58s recent_step=1.52s eta=21:12:00 +2026-09-20 07:20:53 | INFO | train rank=0 | step=1630/50000 progress=3.3% loss=0.151481 lr=6.516000e-05 lr_backbone=6.516000e-05 elapsed=00:42:53 avg_step=1.58s recent_step=1.76s eta=21:12:40 +2026-09-20 07:21:06 | INFO | train rank=0 | step=1640/50000 progress=3.3% loss=0.206418 lr=6.556000e-05 lr_backbone=6.556000e-05 elapsed=00:43:05 avg_step=1.58s recent_step=1.26s eta=21:10:51 +2026-09-20 07:21:21 | INFO | train rank=0 | step=1650/50000 progress=3.3% loss=0.152767 lr=6.596000e-05 lr_backbone=6.596000e-05 elapsed=00:43:20 avg_step=1.58s recent_step=1.50s eta=21:10:13 +2026-09-20 07:21:36 | INFO | train rank=0 | step=1660/50000 progress=3.3% loss=0.157561 lr=6.636000e-05 lr_backbone=6.636000e-05 elapsed=00:43:35 avg_step=1.58s recent_step=1.50s eta=21:09:34 +2026-09-20 07:21:52 | INFO | train rank=0 | step=1670/50000 progress=3.3% loss=0.177855 lr=6.676000e-05 lr_backbone=6.676000e-05 elapsed=00:43:52 avg_step=1.58s recent_step=1.62s eta=21:09:31 +2026-09-20 07:22:09 | INFO | train rank=0 | step=1680/50000 progress=3.4% loss=0.162593 lr=6.716000e-05 lr_backbone=6.716000e-05 elapsed=00:44:09 avg_step=1.58s recent_step=1.73s eta=21:10:01 +2026-09-20 07:22:21 | INFO | train rank=0 | step=1690/50000 progress=3.4% loss=0.183934 lr=6.756000e-05 lr_backbone=6.756000e-05 elapsed=00:44:20 avg_step=1.57s recent_step=1.15s eta=21:07:43 +2026-09-20 07:22:34 | INFO | train rank=0 | step=1700/50000 progress=3.4% loss=0.163503 lr=6.796000e-05 lr_backbone=6.796000e-05 elapsed=00:44:34 avg_step=1.57s recent_step=1.32s eta=21:06:14 +2026-09-20 07:22:52 | INFO | train rank=0 | step=1710/50000 progress=3.4% loss=0.263147 lr=6.836000e-05 lr_backbone=6.836000e-05 elapsed=00:44:52 avg_step=1.57s recent_step=1.84s eta=21:07:13 +2026-09-20 07:23:04 | INFO | train rank=0 | step=1720/50000 progress=3.4% loss=0.152354 lr=6.876000e-05 lr_backbone=6.876000e-05 elapsed=00:45:04 avg_step=1.57s recent_step=1.16s eta=21:05:01 +2026-09-20 07:23:22 | INFO | train rank=0 | step=1730/50000 progress=3.5% loss=0.166180 lr=6.916000e-05 lr_backbone=6.916000e-05 elapsed=00:45:22 avg_step=1.57s recent_step=1.81s eta=21:05:51 +2026-09-20 07:23:34 | INFO | train rank=0 | step=1740/50000 progress=3.5% loss=0.189057 lr=6.956000e-05 lr_backbone=6.956000e-05 elapsed=00:45:34 avg_step=1.57s recent_step=1.25s eta=21:04:05 +2026-09-20 07:23:56 | INFO | train rank=0 | step=1750/50000 progress=3.5% loss=0.143381 lr=6.996000e-05 lr_backbone=6.996000e-05 elapsed=00:45:55 avg_step=1.57s recent_step=2.14s eta=21:06:25 +2026-09-20 07:24:08 | INFO | train rank=0 | step=1760/50000 progress=3.5% loss=0.169724 lr=7.036000e-05 lr_backbone=7.036000e-05 elapsed=00:46:08 avg_step=1.57s recent_step=1.22s eta=21:04:33 +2026-09-20 07:24:20 | INFO | train rank=0 | step=1770/50000 progress=3.5% loss=0.194473 lr=7.076000e-05 lr_backbone=7.076000e-05 elapsed=00:46:20 avg_step=1.57s recent_step=1.24s eta=21:02:46 +2026-09-20 07:24:33 | INFO | train rank=0 | step=1780/50000 progress=3.6% loss=0.175049 lr=7.116000e-05 lr_backbone=7.116000e-05 elapsed=00:46:33 avg_step=1.57s recent_step=1.26s eta=21:01:07 +2026-09-20 07:24:55 | INFO | train rank=0 | step=1790/50000 progress=3.6% loss=0.166159 lr=7.156000e-05 lr_backbone=7.156000e-05 elapsed=00:46:54 avg_step=1.57s recent_step=2.17s eta=21:03:34 +2026-09-20 07:25:06 | INFO | train rank=0 | step=1800/50000 progress=3.6% loss=0.168026 lr=7.196000e-05 lr_backbone=7.196000e-05 elapsed=00:47:06 avg_step=1.57s recent_step=1.16s eta=21:01:29 +2026-09-20 07:25:26 | INFO | train rank=0 | step=1810/50000 progress=3.6% loss=0.169387 lr=7.236000e-05 lr_backbone=7.236000e-05 elapsed=00:47:26 avg_step=1.57s recent_step=1.95s eta=21:02:55 +2026-09-20 07:25:40 | INFO | train rank=0 | step=1820/50000 progress=3.6% loss=0.158242 lr=7.276000e-05 lr_backbone=7.276000e-05 elapsed=00:47:39 avg_step=1.57s recent_step=1.36s eta=21:01:42 +2026-09-20 07:25:54 | INFO | train rank=0 | step=1830/50000 progress=3.7% loss=0.156341 lr=7.316000e-05 lr_backbone=7.316000e-05 elapsed=00:47:54 avg_step=1.57s recent_step=1.45s eta=21:00:55 +2026-09-20 07:26:11 | INFO | train rank=0 | step=1840/50000 progress=3.7% loss=0.166812 lr=7.356000e-05 lr_backbone=7.356000e-05 elapsed=00:48:11 avg_step=1.57s recent_step=1.74s eta=21:01:24 +2026-09-20 07:26:24 | INFO | train rank=0 | step=1850/50000 progress=3.7% loss=0.204736 lr=7.396000e-05 lr_backbone=7.396000e-05 elapsed=00:48:23 avg_step=1.57s recent_step=1.20s eta=20:59:32 +2026-09-20 07:26:40 | INFO | train rank=0 | step=1860/50000 progress=3.7% loss=0.162040 lr=7.436000e-05 lr_backbone=7.436000e-05 elapsed=00:48:40 avg_step=1.57s recent_step=1.64s eta=20:59:36 +2026-09-20 07:26:52 | INFO | train rank=0 | step=1870/50000 progress=3.7% loss=0.144254 lr=7.476000e-05 lr_backbone=7.476000e-05 elapsed=00:48:52 avg_step=1.57s recent_step=1.23s eta=20:57:52 +2026-09-20 07:27:05 | INFO | train rank=0 | step=1880/50000 progress=3.8% loss=0.182652 lr=7.516000e-05 lr_backbone=7.516000e-05 elapsed=00:49:05 avg_step=1.57s recent_step=1.28s eta=20:56:23 +2026-09-20 07:27:23 | INFO | train rank=0 | step=1890/50000 progress=3.8% loss=0.163214 lr=7.556000e-05 lr_backbone=7.556000e-05 elapsed=00:49:23 avg_step=1.57s recent_step=1.79s eta=20:57:03 +2026-09-20 07:27:40 | INFO | train rank=0 | step=1900/50000 progress=3.8% loss=0.178973 lr=7.596000e-05 lr_backbone=7.596000e-05 elapsed=00:49:40 avg_step=1.57s recent_step=1.73s eta=20:57:30 +2026-09-20 07:27:54 | INFO | train rank=0 | step=1910/50000 progress=3.8% loss=0.202769 lr=7.636000e-05 lr_backbone=7.636000e-05 elapsed=00:49:54 avg_step=1.57s recent_step=1.41s eta=20:56:33 +2026-09-20 07:28:10 | INFO | train rank=0 | step=1920/50000 progress=3.8% loss=0.157313 lr=7.676000e-05 lr_backbone=7.676000e-05 elapsed=00:50:10 avg_step=1.57s recent_step=1.56s eta=20:56:16 +2026-09-20 07:28:27 | INFO | train rank=0 | step=1930/50000 progress=3.9% loss=0.160469 lr=7.716000e-05 lr_backbone=7.716000e-05 elapsed=00:50:27 avg_step=1.57s recent_step=1.71s eta=20:56:35 +2026-09-20 07:28:38 | INFO | train rank=0 | step=1940/50000 progress=3.9% loss=0.175706 lr=7.756000e-05 lr_backbone=7.756000e-05 elapsed=00:50:38 avg_step=1.57s recent_step=1.12s eta=20:54:29 +2026-09-20 07:28:53 | INFO | train rank=0 | step=1950/50000 progress=3.9% loss=0.162276 lr=7.796000e-05 lr_backbone=7.796000e-05 elapsed=00:50:52 avg_step=1.57s recent_step=1.45s eta=20:53:45 +2026-09-20 07:29:12 | INFO | train rank=0 | step=1960/50000 progress=3.9% loss=0.171948 lr=7.836000e-05 lr_backbone=7.836000e-05 elapsed=00:51:12 avg_step=1.57s recent_step=1.92s eta=20:54:56 +2026-09-20 07:29:26 | INFO | train rank=0 | step=1970/50000 progress=3.9% loss=0.122387 lr=7.876000e-05 lr_backbone=7.876000e-05 elapsed=00:51:25 avg_step=1.57s recent_step=1.37s eta=20:53:53 +2026-09-20 07:29:43 | INFO | train rank=0 | step=1980/50000 progress=4.0% loss=0.164277 lr=7.916000e-05 lr_backbone=7.916000e-05 elapsed=00:51:42 avg_step=1.57s recent_step=1.71s eta=20:54:13 +2026-09-20 07:29:55 | INFO | train rank=0 | step=1990/50000 progress=4.0% loss=0.206249 lr=7.956000e-05 lr_backbone=7.956000e-05 elapsed=00:51:54 avg_step=1.57s recent_step=1.19s eta=20:52:27 +2026-09-20 07:30:12 | INFO | train rank=0 | step=2000/50000 progress=4.0% loss=0.175457 lr=7.996000e-05 lr_backbone=7.996000e-05 elapsed=00:52:12 avg_step=1.57s recent_step=1.72s eta=20:52:49 +2026-09-20 07:30:27 | INFO | train rank=0 | step=2010/50000 progress=4.0% loss=0.158221 lr=8.036000e-05 lr_backbone=8.036000e-05 elapsed=00:52:26 avg_step=1.57s recent_step=1.46s eta=20:52:08 +2026-09-20 07:30:43 | INFO | train rank=0 | step=2020/50000 progress=4.0% loss=0.149275 lr=8.076000e-05 lr_backbone=8.076000e-05 elapsed=00:52:42 avg_step=1.57s recent_step=1.63s eta=20:52:08 +2026-09-20 07:30:55 | INFO | train rank=0 | step=2030/50000 progress=4.1% loss=0.178933 lr=8.116000e-05 lr_backbone=8.116000e-05 elapsed=00:52:54 avg_step=1.56s recent_step=1.18s eta=20:50:20 +2026-09-20 07:31:14 | INFO | train rank=0 | step=2040/50000 progress=4.1% loss=0.194774 lr=8.156000e-05 lr_backbone=8.156000e-05 elapsed=00:53:13 avg_step=1.57s recent_step=1.91s eta=20:51:26 +2026-09-20 07:31:33 | INFO | train rank=0 | step=2050/50000 progress=4.1% loss=0.169626 lr=8.196000e-05 lr_backbone=8.196000e-05 elapsed=00:53:32 avg_step=1.57s recent_step=1.88s eta=20:52:24 +2026-09-20 07:31:44 | INFO | train rank=0 | step=2060/50000 progress=4.1% loss=0.177483 lr=8.236000e-05 lr_backbone=8.236000e-05 elapsed=00:53:43 avg_step=1.57s recent_step=1.14s eta=20:50:28 +2026-09-20 07:31:59 | INFO | train rank=0 | step=2070/50000 progress=4.1% loss=0.178408 lr=8.276000e-05 lr_backbone=8.276000e-05 elapsed=00:53:59 avg_step=1.56s recent_step=1.53s eta=20:50:03 +2026-09-20 07:32:17 | INFO | train rank=0 | step=2080/50000 progress=4.2% loss=0.193783 lr=8.316000e-05 lr_backbone=8.316000e-05 elapsed=00:54:17 avg_step=1.57s recent_step=1.80s eta=20:50:41 +2026-09-20 07:32:29 | INFO | train rank=0 | step=2090/50000 progress=4.2% loss=0.155830 lr=8.356000e-05 lr_backbone=8.356000e-05 elapsed=00:54:28 avg_step=1.56s recent_step=1.17s eta=20:48:54 +2026-09-20 07:32:45 | INFO | train rank=0 | step=2100/50000 progress=4.2% loss=0.186844 lr=8.396000e-05 lr_backbone=8.396000e-05 elapsed=00:54:45 avg_step=1.56s recent_step=1.66s eta=20:49:00 +2026-09-20 07:33:01 | INFO | train rank=0 | step=2110/50000 progress=4.2% loss=0.190510 lr=8.436000e-05 lr_backbone=8.436000e-05 elapsed=00:55:01 avg_step=1.56s recent_step=1.57s eta=20:48:45 +2026-09-20 07:33:17 | INFO | train rank=0 | step=2120/50000 progress=4.2% loss=0.191547 lr=8.476000e-05 lr_backbone=8.476000e-05 elapsed=00:55:17 avg_step=1.56s recent_step=1.59s eta=20:48:36 +2026-09-20 07:33:29 | INFO | train rank=0 | step=2130/50000 progress=4.3% loss=0.161037 lr=8.516000e-05 lr_backbone=8.516000e-05 elapsed=00:55:29 avg_step=1.56s recent_step=1.24s eta=20:47:08 +2026-09-20 07:33:41 | INFO | train rank=0 | step=2140/50000 progress=4.3% loss=0.157641 lr=8.556000e-05 lr_backbone=8.556000e-05 elapsed=00:55:41 avg_step=1.56s recent_step=1.17s eta=20:45:24 +2026-09-20 07:33:57 | INFO | train rank=0 | step=2150/50000 progress=4.3% loss=0.189743 lr=8.596000e-05 lr_backbone=8.596000e-05 elapsed=00:55:56 avg_step=1.56s recent_step=1.57s eta=20:45:11 +2026-09-20 07:34:15 | INFO | train rank=0 | step=2160/50000 progress=4.3% loss=0.248421 lr=8.636000e-05 lr_backbone=8.636000e-05 elapsed=00:56:14 avg_step=1.56s recent_step=1.77s eta=20:45:42 +2026-09-20 07:34:28 | INFO | train rank=0 | step=2170/50000 progress=4.3% loss=0.170531 lr=8.676000e-05 lr_backbone=8.676000e-05 elapsed=00:56:28 avg_step=1.56s recent_step=1.38s eta=20:44:47 +2026-09-20 07:34:44 | INFO | train rank=0 | step=2180/50000 progress=4.4% loss=0.152726 lr=8.716000e-05 lr_backbone=8.716000e-05 elapsed=00:56:43 avg_step=1.56s recent_step=1.51s eta=20:44:21 +2026-09-20 07:35:00 | INFO | train rank=0 | step=2190/50000 progress=4.4% loss=0.164421 lr=8.756000e-05 lr_backbone=8.756000e-05 elapsed=00:56:59 avg_step=1.56s recent_step=1.62s eta=20:44:18 +2026-09-20 07:35:14 | INFO | train rank=0 | step=2200/50000 progress=4.4% loss=0.157205 lr=8.796000e-05 lr_backbone=8.796000e-05 elapsed=00:57:13 avg_step=1.56s recent_step=1.39s eta=20:43:24 +2026-09-20 07:35:34 | INFO | train rank=0 | step=2210/50000 progress=4.4% loss=0.139760 lr=8.836000e-05 lr_backbone=8.836000e-05 elapsed=00:57:33 avg_step=1.56s recent_step=2.02s eta=20:44:48 +2026-09-20 07:35:47 | INFO | train rank=0 | step=2220/50000 progress=4.4% loss=0.127099 lr=8.876000e-05 lr_backbone=8.876000e-05 elapsed=00:57:46 avg_step=1.56s recent_step=1.28s eta=20:43:32 +2026-09-20 07:36:01 | INFO | train rank=0 | step=2230/50000 progress=4.5% loss=0.167419 lr=8.916000e-05 lr_backbone=8.916000e-05 elapsed=00:58:01 avg_step=1.56s recent_step=1.46s eta=20:42:54 +2026-09-20 07:36:17 | INFO | train rank=0 | step=2240/50000 progress=4.5% loss=0.184657 lr=8.956000e-05 lr_backbone=8.956000e-05 elapsed=00:58:16 avg_step=1.56s recent_step=1.55s eta=20:42:35 +2026-09-20 07:36:32 | INFO | train rank=0 | step=2250/50000 progress=4.5% loss=0.172561 lr=8.996000e-05 lr_backbone=8.996000e-05 elapsed=00:58:31 avg_step=1.56s recent_step=1.50s eta=20:42:08 +2026-09-20 07:36:46 | INFO | train rank=0 | step=2260/50000 progress=4.5% loss=0.186454 lr=9.036000e-05 lr_backbone=9.036000e-05 elapsed=00:58:45 avg_step=1.56s recent_step=1.42s eta=20:41:22 +2026-09-20 07:37:00 | INFO | train rank=0 | step=2270/50000 progress=4.5% loss=0.149555 lr=9.076000e-05 lr_backbone=9.076000e-05 elapsed=00:59:00 avg_step=1.56s recent_step=1.46s eta=20:40:44 +2026-09-20 07:37:21 | INFO | train rank=0 | step=2280/50000 progress=4.6% loss=0.129001 lr=9.116000e-05 lr_backbone=9.116000e-05 elapsed=00:59:20 avg_step=1.56s recent_step=2.03s eta=20:42:07 +2026-09-20 07:37:33 | INFO | train rank=0 | step=2290/50000 progress=4.6% loss=0.183371 lr=9.156000e-05 lr_backbone=9.156000e-05 elapsed=00:59:33 avg_step=1.56s recent_step=1.22s eta=20:40:40 +2026-09-20 07:37:47 | INFO | train rank=0 | step=2300/50000 progress=4.6% loss=0.189873 lr=9.196000e-05 lr_backbone=9.196000e-05 elapsed=00:59:47 avg_step=1.56s recent_step=1.40s eta=20:39:52 +2026-09-20 07:38:01 | INFO | train rank=0 | step=2310/50000 progress=4.6% loss=0.219830 lr=9.236000e-05 lr_backbone=9.236000e-05 elapsed=01:00:01 avg_step=1.56s recent_step=1.41s eta=20:39:05 +2026-09-20 07:38:19 | INFO | train rank=0 | step=2320/50000 progress=4.6% loss=0.208629 lr=9.276000e-05 lr_backbone=9.276000e-05 elapsed=01:00:19 avg_step=1.56s recent_step=1.85s eta=20:39:49 +2026-09-20 07:38:35 | INFO | train rank=0 | step=2330/50000 progress=4.7% loss=0.168763 lr=9.316000e-05 lr_backbone=9.316000e-05 elapsed=01:00:35 avg_step=1.56s recent_step=1.56s eta=20:39:34 +2026-09-20 07:38:50 | INFO | train rank=0 | step=2340/50000 progress=4.7% loss=0.135727 lr=9.356000e-05 lr_backbone=9.356000e-05 elapsed=01:00:49 avg_step=1.56s recent_step=1.45s eta=20:38:55 +2026-09-20 07:39:05 | INFO | train rank=0 | step=2350/50000 progress=4.7% loss=0.206708 lr=9.396000e-05 lr_backbone=9.396000e-05 elapsed=01:01:04 avg_step=1.56s recent_step=1.53s eta=20:38:33 +2026-09-20 07:39:20 | INFO | train rank=0 | step=2360/50000 progress=4.7% loss=0.123987 lr=9.436000e-05 lr_backbone=9.436000e-05 elapsed=01:01:20 avg_step=1.56s recent_step=1.56s eta=20:38:18 +2026-09-20 07:39:34 | INFO | train rank=0 | step=2370/50000 progress=4.7% loss=0.178162 lr=9.476000e-05 lr_backbone=9.476000e-05 elapsed=01:01:34 avg_step=1.56s recent_step=1.35s eta=20:37:20 +2026-09-20 07:39:53 | INFO | train rank=0 | step=2380/50000 progress=4.8% loss=0.135486 lr=9.516000e-05 lr_backbone=9.516000e-05 elapsed=01:01:53 avg_step=1.56s recent_step=1.93s eta=20:38:18 +2026-09-20 07:40:04 | INFO | train rank=0 | step=2390/50000 progress=4.8% loss=0.172910 lr=9.556000e-05 lr_backbone=9.556000e-05 elapsed=01:02:04 avg_step=1.56s recent_step=1.08s eta=20:36:27 +2026-09-20 07:40:18 | INFO | train rank=0 | step=2400/50000 progress=4.8% loss=0.150586 lr=9.596000e-05 lr_backbone=9.596000e-05 elapsed=01:02:18 avg_step=1.56s recent_step=1.43s eta=20:35:45 +2026-09-20 07:40:38 | INFO | train rank=0 | step=2410/50000 progress=4.8% loss=0.149310 lr=9.636000e-05 lr_backbone=9.636000e-05 elapsed=01:02:37 avg_step=1.56s recent_step=1.95s eta=20:36:47 +2026-09-20 07:40:52 | INFO | train rank=0 | step=2420/50000 progress=4.8% loss=0.243797 lr=9.676000e-05 lr_backbone=9.676000e-05 elapsed=01:02:51 avg_step=1.56s recent_step=1.39s eta=20:35:58 +2026-09-20 07:41:06 | INFO | train rank=0 | step=2430/50000 progress=4.9% loss=0.146807 lr=9.716000e-05 lr_backbone=9.716000e-05 elapsed=01:03:06 avg_step=1.56s recent_step=1.43s eta=20:35:18 +2026-09-20 07:41:19 | INFO | train rank=0 | step=2440/50000 progress=4.9% loss=0.128707 lr=9.756000e-05 lr_backbone=9.756000e-05 elapsed=01:03:18 avg_step=1.56s recent_step=1.26s eta=20:34:04 +2026-09-20 07:41:35 | INFO | train rank=0 | step=2450/50000 progress=4.9% loss=0.139792 lr=9.796000e-05 lr_backbone=9.796000e-05 elapsed=01:03:35 avg_step=1.56s recent_step=1.64s eta=20:34:05 +2026-09-20 07:41:52 | INFO | train rank=0 | step=2460/50000 progress=4.9% loss=0.163022 lr=9.836000e-05 lr_backbone=9.836000e-05 elapsed=01:03:51 avg_step=1.56s recent_step=1.65s eta=20:34:08 +2026-09-20 07:42:06 | INFO | train rank=0 | step=2470/50000 progress=4.9% loss=0.124401 lr=9.876000e-05 lr_backbone=9.876000e-05 elapsed=01:04:06 avg_step=1.56s recent_step=1.46s eta=20:33:34 +2026-09-20 07:42:25 | INFO | train rank=0 | step=2480/50000 progress=5.0% loss=0.134794 lr=9.916000e-05 lr_backbone=9.916000e-05 elapsed=01:04:24 avg_step=1.56s recent_step=1.83s eta=20:34:11 +2026-09-20 07:42:37 | INFO | train rank=0 | step=2490/50000 progress=5.0% loss=0.129895 lr=9.956000e-05 lr_backbone=9.956000e-05 elapsed=01:04:36 avg_step=1.56s recent_step=1.21s eta=20:32:49 +2026-09-20 07:42:49 | INFO | train rank=0 | step=2500/50000 progress=5.0% loss=0.133630 lr=9.996000e-05 lr_backbone=9.996000e-05 elapsed=01:04:49 avg_step=1.56s recent_step=1.28s eta=20:31:42 +2026-09-20 07:43:09 | INFO | train rank=0 | step=2510/50000 progress=5.0% loss=0.123645 lr=9.999999e-05 lr_backbone=9.999999e-05 elapsed=01:05:08 avg_step=1.56s recent_step=1.91s eta=20:32:34 +2026-09-20 07:43:25 | INFO | train rank=0 | step=2520/50000 progress=5.0% loss=0.129628 lr=9.999996e-05 lr_backbone=9.999996e-05 elapsed=01:05:25 avg_step=1.56s recent_step=1.65s eta=20:32:36 +2026-09-20 07:43:42 | INFO | train rank=0 | step=2530/50000 progress=5.1% loss=0.223630 lr=9.999990e-05 lr_backbone=9.999990e-05 elapsed=01:05:42 avg_step=1.56s recent_step=1.73s eta=20:32:53 +2026-09-20 07:43:55 | INFO | train rank=0 | step=2540/50000 progress=5.1% loss=0.185141 lr=9.999983e-05 lr_backbone=9.999983e-05 elapsed=01:05:55 avg_step=1.56s recent_step=1.25s eta=20:31:41 +2026-09-20 07:44:12 | INFO | train rank=0 | step=2550/50000 progress=5.1% loss=0.137457 lr=9.999973e-05 lr_backbone=9.999973e-05 elapsed=01:06:12 avg_step=1.56s recent_step=1.70s eta=20:31:52 +2026-09-20 07:44:29 | INFO | train rank=0 | step=2560/50000 progress=5.1% loss=0.109210 lr=9.999961e-05 lr_backbone=9.999961e-05 elapsed=01:06:28 avg_step=1.56s recent_step=1.66s eta=20:31:55 +2026-09-20 07:44:41 | INFO | train rank=0 | step=2570/50000 progress=5.1% loss=0.137959 lr=9.999946e-05 lr_backbone=9.999946e-05 elapsed=01:06:40 avg_step=1.56s recent_step=1.22s eta=20:30:37 +2026-09-20 07:44:57 | INFO | train rank=0 | step=2580/50000 progress=5.2% loss=0.150583 lr=9.999930e-05 lr_backbone=9.999930e-05 elapsed=01:06:56 avg_step=1.56s recent_step=1.58s eta=20:30:27 +2026-09-20 07:45:12 | INFO | train rank=0 | step=2590/50000 progress=5.2% loss=0.148639 lr=9.999911e-05 lr_backbone=9.999911e-05 elapsed=01:07:12 avg_step=1.56s recent_step=1.56s eta=20:30:11 +2026-09-20 07:45:30 | INFO | train rank=0 | step=2600/50000 progress=5.2% loss=0.155523 lr=9.999891e-05 lr_backbone=9.999891e-05 elapsed=01:07:30 avg_step=1.56s recent_step=1.79s eta=20:30:39 +2026-09-20 07:45:42 | INFO | train rank=0 | step=2610/50000 progress=5.2% loss=0.147228 lr=9.999868e-05 lr_backbone=9.999868e-05 elapsed=01:07:41 avg_step=1.56s recent_step=1.15s eta=20:29:08 +2026-09-20 07:46:00 | INFO | train rank=0 | step=2620/50000 progress=5.2% loss=0.150266 lr=9.999843e-05 lr_backbone=9.999843e-05 elapsed=01:07:59 avg_step=1.56s recent_step=1.81s eta=20:29:38 +2026-09-20 07:46:16 | INFO | train rank=0 | step=2630/50000 progress=5.3% loss=0.175361 lr=9.999815e-05 lr_backbone=9.999815e-05 elapsed=01:08:16 avg_step=1.56s recent_step=1.66s eta=20:29:42 +2026-09-20 07:46:30 | INFO | train rank=0 | step=2640/50000 progress=5.3% loss=0.145791 lr=9.999786e-05 lr_backbone=9.999786e-05 elapsed=01:08:30 avg_step=1.56s recent_step=1.40s eta=20:28:57 +2026-09-20 07:46:45 | INFO | train rank=0 | step=2650/50000 progress=5.3% loss=0.157410 lr=9.999754e-05 lr_backbone=9.999754e-05 elapsed=01:08:45 avg_step=1.56s recent_step=1.48s eta=20:28:28 +2026-09-20 07:46:58 | INFO | train rank=0 | step=2660/50000 progress=5.3% loss=0.160015 lr=9.999720e-05 lr_backbone=9.999720e-05 elapsed=01:08:57 avg_step=1.56s recent_step=1.25s eta=20:27:18 +2026-09-20 07:47:19 | INFO | train rank=0 | step=2670/50000 progress=5.3% loss=0.139270 lr=9.999684e-05 lr_backbone=9.999684e-05 elapsed=01:09:19 avg_step=1.56s recent_step=2.16s eta=20:28:50 +2026-09-20 07:47:31 | INFO | train rank=0 | step=2680/50000 progress=5.4% loss=0.149344 lr=9.999646e-05 lr_backbone=9.999646e-05 elapsed=01:09:31 avg_step=1.56s recent_step=1.21s eta=20:27:34 +2026-09-20 07:47:45 | INFO | train rank=0 | step=2690/50000 progress=5.4% loss=0.152807 lr=9.999605e-05 lr_backbone=9.999605e-05 elapsed=01:09:45 avg_step=1.56s recent_step=1.41s eta=20:26:53 +2026-09-20 07:48:01 | INFO | train rank=0 | step=2700/50000 progress=5.4% loss=0.127873 lr=9.999563e-05 lr_backbone=9.999563e-05 elapsed=01:10:00 avg_step=1.56s recent_step=1.51s eta=20:26:29 +2026-09-20 07:48:18 | INFO | train rank=0 | step=2710/50000 progress=5.4% loss=0.164935 lr=9.999518e-05 lr_backbone=9.999518e-05 elapsed=01:10:17 avg_step=1.56s recent_step=1.73s eta=20:26:44 +2026-09-20 07:48:35 | INFO | train rank=0 | step=2720/50000 progress=5.4% loss=0.128317 lr=9.999471e-05 lr_backbone=9.999471e-05 elapsed=01:10:35 avg_step=1.56s recent_step=1.75s eta=20:27:03 +2026-09-20 07:48:47 | INFO | train rank=0 | step=2730/50000 progress=5.5% loss=0.143362 lr=9.999422e-05 lr_backbone=9.999422e-05 elapsed=01:10:47 avg_step=1.56s recent_step=1.16s eta=20:25:39 +2026-09-20 07:49:08 | INFO | train rank=0 | step=2740/50000 progress=5.5% loss=0.138223 lr=9.999370e-05 lr_backbone=9.999370e-05 elapsed=01:11:08 avg_step=1.56s recent_step=2.11s eta=20:26:59 +2026-09-20 07:49:20 | INFO | train rank=0 | step=2750/50000 progress=5.5% loss=0.179906 lr=9.999317e-05 lr_backbone=9.999317e-05 elapsed=01:11:19 avg_step=1.56s recent_step=1.15s eta=20:25:33 +2026-09-20 07:49:31 | INFO | train rank=0 | step=2760/50000 progress=5.5% loss=0.118774 lr=9.999261e-05 lr_backbone=9.999261e-05 elapsed=01:11:31 avg_step=1.55s recent_step=1.19s eta=20:24:14 +2026-09-20 07:49:49 | INFO | train rank=0 | step=2770/50000 progress=5.5% loss=0.147788 lr=9.999203e-05 lr_backbone=9.999203e-05 elapsed=01:11:49 avg_step=1.56s recent_step=1.74s eta=20:24:31 +2026-09-20 07:50:00 | INFO | train rank=0 | step=2780/50000 progress=5.6% loss=0.144303 lr=9.999143e-05 lr_backbone=9.999143e-05 elapsed=01:12:00 avg_step=1.55s recent_step=1.14s eta=20:23:04 +2026-09-20 07:50:16 | INFO | train rank=0 | step=2790/50000 progress=5.6% loss=0.144186 lr=9.999080e-05 lr_backbone=9.999080e-05 elapsed=01:12:16 avg_step=1.55s recent_step=1.59s eta=20:22:54 +2026-09-20 07:50:36 | INFO | train rank=0 | step=2800/50000 progress=5.6% loss=0.126971 lr=9.999016e-05 lr_backbone=9.999016e-05 elapsed=01:12:36 avg_step=1.56s recent_step=2.00s eta=20:23:54 +2026-09-20 07:50:51 | INFO | train rank=0 | step=2810/50000 progress=5.6% loss=0.162363 lr=9.998949e-05 lr_backbone=9.998949e-05 elapsed=01:12:51 avg_step=1.56s recent_step=1.52s eta=20:23:33 +2026-09-20 07:51:04 | INFO | train rank=0 | step=2820/50000 progress=5.6% loss=0.143877 lr=9.998880e-05 lr_backbone=9.998880e-05 elapsed=01:13:04 avg_step=1.55s recent_step=1.31s eta=20:22:36 +2026-09-20 07:51:20 | INFO | train rank=0 | step=2830/50000 progress=5.7% loss=0.157893 lr=9.998809e-05 lr_backbone=9.998809e-05 elapsed=01:13:19 avg_step=1.55s recent_step=1.54s eta=20:22:17 +2026-09-20 07:51:37 | INFO | train rank=0 | step=2840/50000 progress=5.7% loss=0.128544 lr=9.998736e-05 lr_backbone=9.998736e-05 elapsed=01:13:37 avg_step=1.56s recent_step=1.71s eta=20:22:28 +2026-09-20 07:51:49 | INFO | train rank=0 | step=2850/50000 progress=5.7% loss=0.156917 lr=9.998660e-05 lr_backbone=9.998660e-05 elapsed=01:13:49 avg_step=1.55s recent_step=1.21s eta=20:21:16 +2026-09-20 07:52:01 | INFO | train rank=0 | step=2860/50000 progress=5.7% loss=0.151317 lr=9.998583e-05 lr_backbone=9.998583e-05 elapsed=01:14:00 avg_step=1.55s recent_step=1.15s eta=20:19:54 +2026-09-20 07:52:19 | INFO | train rank=0 | step=2870/50000 progress=5.7% loss=0.135441 lr=9.998503e-05 lr_backbone=9.998503e-05 elapsed=01:14:18 avg_step=1.55s recent_step=1.80s eta=20:20:18 +2026-09-20 07:52:36 | INFO | train rank=0 | step=2880/50000 progress=5.8% loss=0.153644 lr=9.998421e-05 lr_backbone=9.998421e-05 elapsed=01:14:36 avg_step=1.55s recent_step=1.76s eta=20:20:36 +2026-09-20 07:52:52 | INFO | train rank=0 | step=2890/50000 progress=5.8% loss=0.114329 lr=9.998337e-05 lr_backbone=9.998337e-05 elapsed=01:14:51 avg_step=1.55s recent_step=1.57s eta=20:20:23 +2026-09-20 07:53:04 | INFO | train rank=0 | step=2900/50000 progress=5.8% loss=0.162602 lr=9.998250e-05 lr_backbone=9.998250e-05 elapsed=01:15:04 avg_step=1.55s recent_step=1.21s eta=20:19:11 +2026-09-20 07:53:25 | INFO | train rank=0 | step=2910/50000 progress=5.8% loss=0.140649 lr=9.998162e-05 lr_backbone=9.998162e-05 elapsed=01:15:25 avg_step=1.56s recent_step=2.16s eta=20:20:33 +2026-09-20 07:53:39 | INFO | train rank=0 | step=2920/50000 progress=5.8% loss=0.154239 lr=9.998071e-05 lr_backbone=9.998071e-05 elapsed=01:15:39 avg_step=1.55s recent_step=1.38s eta=20:19:50 +2026-09-20 07:53:52 | INFO | train rank=0 | step=2930/50000 progress=5.9% loss=0.095012 lr=9.997978e-05 lr_backbone=9.997978e-05 elapsed=01:15:51 avg_step=1.55s recent_step=1.22s eta=20:18:41 +2026-09-20 07:54:06 | INFO | train rank=0 | step=2940/50000 progress=5.9% loss=0.152787 lr=9.997883e-05 lr_backbone=9.997883e-05 elapsed=01:16:06 avg_step=1.55s recent_step=1.44s eta=20:18:08 +2026-09-20 07:54:25 | INFO | train rank=0 | step=2950/50000 progress=5.9% loss=0.122654 lr=9.997786e-05 lr_backbone=9.997786e-05 elapsed=01:16:25 avg_step=1.55s recent_step=1.92s eta=20:18:50 +2026-09-20 07:54:41 | INFO | train rank=0 | step=2960/50000 progress=5.9% loss=0.162516 lr=9.997686e-05 lr_backbone=9.997686e-05 elapsed=01:16:40 avg_step=1.55s recent_step=1.56s eta=20:18:35 +2026-09-20 07:54:54 | INFO | train rank=0 | step=2970/50000 progress=5.9% loss=0.145841 lr=9.997584e-05 lr_backbone=9.997584e-05 elapsed=01:16:54 avg_step=1.55s recent_step=1.36s eta=20:17:49 +2026-09-20 07:55:10 | INFO | train rank=0 | step=2980/50000 progress=6.0% loss=0.137548 lr=9.997481e-05 lr_backbone=9.997481e-05 elapsed=01:17:09 avg_step=1.55s recent_step=1.53s eta=20:17:31 +2026-09-20 07:55:26 | INFO | train rank=0 | step=2990/50000 progress=6.0% loss=0.105949 lr=9.997375e-05 lr_backbone=9.997375e-05 elapsed=01:17:26 avg_step=1.55s recent_step=1.65s eta=20:17:30 +2026-09-20 07:55:41 | INFO | train rank=0 | step=3000/50000 progress=6.0% loss=0.133492 lr=9.997266e-05 lr_backbone=9.997266e-05 elapsed=01:17:40 avg_step=1.55s recent_step=1.44s eta=20:16:57 +2026-09-20 07:55:55 | INFO | train rank=0 | step=3010/50000 progress=6.0% loss=0.122354 lr=9.997156e-05 lr_backbone=9.997156e-05 elapsed=01:17:55 avg_step=1.55s recent_step=1.43s eta=20:16:22 +2026-09-20 07:56:16 | INFO | train rank=0 | step=3020/50000 progress=6.0% loss=0.129702 lr=9.997043e-05 lr_backbone=9.997043e-05 elapsed=01:18:15 avg_step=1.55s recent_step=2.07s eta=20:17:27 +2026-09-20 07:56:27 | INFO | train rank=0 | step=3030/50000 progress=6.1% loss=0.107075 lr=9.996928e-05 lr_backbone=9.996928e-05 elapsed=01:18:26 avg_step=1.55s recent_step=1.11s eta=20:16:03 +2026-09-20 07:56:39 | INFO | train rank=0 | step=3040/50000 progress=6.1% loss=0.129192 lr=9.996811e-05 lr_backbone=9.996811e-05 elapsed=01:18:38 avg_step=1.55s recent_step=1.19s eta=20:14:51 +2026-09-20 07:56:56 | INFO | train rank=0 | step=3050/50000 progress=6.1% loss=0.124624 lr=9.996692e-05 lr_backbone=9.996692e-05 elapsed=01:18:55 avg_step=1.55s recent_step=1.70s eta=20:14:58 +2026-09-20 07:57:11 | INFO | train rank=0 | step=3060/50000 progress=6.1% loss=0.265909 lr=9.996571e-05 lr_backbone=9.996571e-05 elapsed=01:19:10 avg_step=1.55s recent_step=1.50s eta=20:14:34 +2026-09-20 07:57:27 | INFO | train rank=0 | step=3070/50000 progress=6.1% loss=0.158704 lr=9.996447e-05 lr_backbone=9.996447e-05 elapsed=01:19:26 avg_step=1.55s recent_step=1.59s eta=20:14:25 +2026-09-20 07:57:44 | INFO | train rank=0 | step=3080/50000 progress=6.2% loss=0.133471 lr=9.996322e-05 lr_backbone=9.996322e-05 elapsed=01:19:43 avg_step=1.55s recent_step=1.73s eta=20:14:37 +2026-09-20 07:57:58 | INFO | train rank=0 | step=3090/50000 progress=6.2% loss=0.147378 lr=9.996194e-05 lr_backbone=9.996194e-05 elapsed=01:19:58 avg_step=1.55s recent_step=1.46s eta=20:14:07 +2026-09-20 07:58:16 | INFO | train rank=0 | step=3100/50000 progress=6.2% loss=0.141916 lr=9.996064e-05 lr_backbone=9.996064e-05 elapsed=01:20:15 avg_step=1.55s recent_step=1.74s eta=20:14:19 +2026-09-20 07:58:28 | INFO | train rank=0 | step=3110/50000 progress=6.2% loss=0.110185 lr=9.995931e-05 lr_backbone=9.995931e-05 elapsed=01:20:27 avg_step=1.55s recent_step=1.21s eta=20:13:12 +2026-09-20 07:58:45 | INFO | train rank=0 | step=3120/50000 progress=6.2% loss=0.127930 lr=9.995797e-05 lr_backbone=9.995797e-05 elapsed=01:20:44 avg_step=1.55s recent_step=1.69s eta=20:13:18 +2026-09-20 07:59:00 | INFO | train rank=0 | step=3130/50000 progress=6.3% loss=0.145124 lr=9.995660e-05 lr_backbone=9.995660e-05 elapsed=01:21:00 avg_step=1.55s recent_step=1.56s eta=20:13:03 +2026-09-20 07:59:15 | INFO | train rank=0 | step=3140/50000 progress=6.3% loss=0.129571 lr=9.995521e-05 lr_backbone=9.995521e-05 elapsed=01:21:15 avg_step=1.55s recent_step=1.48s eta=20:12:36 +2026-09-20 07:59:30 | INFO | train rank=0 | step=3150/50000 progress=6.3% loss=0.140574 lr=9.995380e-05 lr_backbone=9.995380e-05 elapsed=01:21:30 avg_step=1.55s recent_step=1.50s eta=20:12:13 +2026-09-20 07:59:47 | INFO | train rank=0 | step=3160/50000 progress=6.3% loss=0.148298 lr=9.995237e-05 lr_backbone=9.995237e-05 elapsed=01:21:46 avg_step=1.55s recent_step=1.64s eta=20:12:10 +2026-09-20 08:00:03 | INFO | train rank=0 | step=3170/50000 progress=6.3% loss=0.166269 lr=9.995092e-05 lr_backbone=9.995092e-05 elapsed=01:22:03 avg_step=1.55s recent_step=1.68s eta=20:12:14 +2026-09-20 08:00:17 | INFO | train rank=0 | step=3180/50000 progress=6.4% loss=0.133184 lr=9.994944e-05 lr_backbone=9.994944e-05 elapsed=01:22:17 avg_step=1.55s recent_step=1.36s eta=20:11:30 +2026-09-20 08:00:32 | INFO | train rank=0 | step=3190/50000 progress=6.4% loss=0.126956 lr=9.994794e-05 lr_backbone=9.994794e-05 elapsed=01:22:31 avg_step=1.55s recent_step=1.47s eta=20:11:02 +2026-09-20 08:00:49 | INFO | train rank=0 | step=3200/50000 progress=6.4% loss=0.156407 lr=9.994642e-05 lr_backbone=9.994642e-05 elapsed=01:22:49 avg_step=1.55s recent_step=1.72s eta=20:11:11 +2026-09-20 08:01:06 | INFO | train rank=0 | step=3210/50000 progress=6.4% loss=0.135422 lr=9.994488e-05 lr_backbone=9.994488e-05 elapsed=01:23:05 avg_step=1.55s recent_step=1.69s eta=20:11:15 +2026-09-20 08:01:18 | INFO | train rank=0 | step=3220/50000 progress=6.4% loss=0.098228 lr=9.994332e-05 lr_backbone=9.994332e-05 elapsed=01:23:18 avg_step=1.55s recent_step=1.23s eta=20:10:13 +2026-09-20 08:01:30 | INFO | train rank=0 | step=3230/50000 progress=6.5% loss=0.107511 lr=9.994173e-05 lr_backbone=9.994173e-05 elapsed=01:23:30 avg_step=1.55s recent_step=1.20s eta=20:09:07 +2026-09-20 08:01:46 | INFO | train rank=0 | step=3240/50000 progress=6.5% loss=0.130927 lr=9.994013e-05 lr_backbone=9.994013e-05 elapsed=01:23:46 avg_step=1.55s recent_step=1.62s eta=20:09:01 +2026-09-20 08:02:06 | INFO | train rank=0 | step=3250/50000 progress=6.5% loss=0.137395 lr=9.993850e-05 lr_backbone=9.993850e-05 elapsed=01:24:06 avg_step=1.55s recent_step=1.98s eta=20:09:46 +2026-09-20 08:02:18 | INFO | train rank=0 | step=3260/50000 progress=6.5% loss=0.120086 lr=9.993685e-05 lr_backbone=9.993685e-05 elapsed=01:24:18 avg_step=1.55s recent_step=1.24s eta=20:08:46 +2026-09-20 08:02:36 | INFO | train rank=0 | step=3270/50000 progress=6.5% loss=0.142153 lr=9.993518e-05 lr_backbone=9.993518e-05 elapsed=01:24:36 avg_step=1.55s recent_step=1.77s eta=20:09:02 +2026-09-20 08:02:53 | INFO | train rank=0 | step=3280/50000 progress=6.6% loss=0.135541 lr=9.993348e-05 lr_backbone=9.993348e-05 elapsed=01:24:53 avg_step=1.55s recent_step=1.73s eta=20:09:12 +2026-09-20 08:03:10 | INFO | train rank=0 | step=3290/50000 progress=6.6% loss=0.125056 lr=9.993176e-05 lr_backbone=9.993176e-05 elapsed=01:25:09 avg_step=1.55s recent_step=1.61s eta=20:09:05 +2026-09-20 08:03:22 | INFO | train rank=0 | step=3300/50000 progress=6.6% loss=0.131787 lr=9.993003e-05 lr_backbone=9.993003e-05 elapsed=01:25:21 avg_step=1.55s recent_step=1.23s eta=20:08:03 +2026-09-20 08:03:40 | INFO | train rank=0 | step=3310/50000 progress=6.6% loss=0.119867 lr=9.992827e-05 lr_backbone=9.992827e-05 elapsed=01:25:39 avg_step=1.55s recent_step=1.80s eta=20:08:23 +2026-09-20 08:03:52 | INFO | train rank=0 | step=3320/50000 progress=6.6% loss=0.136163 lr=9.992649e-05 lr_backbone=9.992649e-05 elapsed=01:25:52 avg_step=1.55s recent_step=1.21s eta=20:07:19 +2026-09-20 08:04:12 | INFO | train rank=0 | step=3330/50000 progress=6.7% loss=0.134783 lr=9.992468e-05 lr_backbone=9.992468e-05 elapsed=01:26:12 avg_step=1.55s recent_step=2.02s eta=20:08:09 +2026-09-20 08:04:30 | INFO | train rank=0 | step=3340/50000 progress=6.7% loss=0.138571 lr=9.992286e-05 lr_backbone=9.992286e-05 elapsed=01:26:30 avg_step=1.55s recent_step=1.82s eta=20:08:31 +2026-09-20 08:04:43 | INFO | train rank=0 | step=3350/50000 progress=6.7% loss=0.093602 lr=9.992101e-05 lr_backbone=9.992101e-05 elapsed=01:26:42 avg_step=1.55s recent_step=1.22s eta=20:07:29 +2026-09-20 08:04:54 | INFO | train rank=0 | step=3360/50000 progress=6.7% loss=0.167774 lr=9.991914e-05 lr_backbone=9.991914e-05 elapsed=01:26:54 avg_step=1.55s recent_step=1.19s eta=20:06:23 +2026-09-20 08:05:06 | INFO | train rank=0 | step=3370/50000 progress=6.7% loss=0.140445 lr=9.991725e-05 lr_backbone=9.991725e-05 elapsed=01:27:06 avg_step=1.55s recent_step=1.19s eta=20:05:17 +2026-09-20 08:05:20 | INFO | train rank=0 | step=3380/50000 progress=6.8% loss=0.143296 lr=9.991534e-05 lr_backbone=9.991534e-05 elapsed=01:27:20 avg_step=1.55s recent_step=1.38s eta=20:04:39 +2026-09-20 08:05:42 | INFO | train rank=0 | step=3390/50000 progress=6.8% loss=0.164771 lr=9.991340e-05 lr_backbone=9.991340e-05 elapsed=01:27:42 avg_step=1.55s recent_step=2.17s eta=20:05:48 +2026-09-20 08:05:54 | INFO | train rank=0 | step=3400/50000 progress=6.8% loss=0.117464 lr=9.991145e-05 lr_backbone=9.991145e-05 elapsed=01:27:54 avg_step=1.55s recent_step=1.22s eta=20:04:48 +2026-09-20 08:06:20 | INFO | train rank=0 | step=3410/50000 progress=6.8% loss=0.113827 lr=9.990947e-05 lr_backbone=9.990947e-05 elapsed=01:28:20 avg_step=1.55s recent_step=2.60s eta=20:06:56 +2026-09-20 08:06:32 | INFO | train rank=0 | step=3420/50000 progress=6.8% loss=0.124655 lr=9.990747e-05 lr_backbone=9.990747e-05 elapsed=01:28:31 avg_step=1.55s recent_step=1.15s eta=20:05:45 +2026-09-20 08:06:44 | INFO | train rank=0 | step=3430/50000 progress=6.9% loss=0.116481 lr=9.990545e-05 lr_backbone=9.990545e-05 elapsed=01:28:43 avg_step=1.55s recent_step=1.22s eta=20:04:44 +2026-09-20 08:06:56 | INFO | train rank=0 | step=3440/50000 progress=6.9% loss=0.199625 lr=9.990340e-05 lr_backbone=9.990340e-05 elapsed=01:28:56 avg_step=1.55s recent_step=1.22s eta=20:03:43 +2026-09-20 08:07:16 | INFO | train rank=0 | step=3450/50000 progress=6.9% loss=0.130956 lr=9.990134e-05 lr_backbone=9.990134e-05 elapsed=01:29:15 avg_step=1.55s recent_step=1.96s eta=20:04:23 +2026-09-20 08:07:30 | INFO | train rank=0 | step=3460/50000 progress=6.9% loss=0.134579 lr=9.989925e-05 lr_backbone=9.989925e-05 elapsed=01:29:29 avg_step=1.55s recent_step=1.42s eta=20:03:49 +2026-09-20 08:07:42 | INFO | train rank=0 | step=3470/50000 progress=6.9% loss=0.137113 lr=9.989714e-05 lr_backbone=9.989714e-05 elapsed=01:29:42 avg_step=1.55s recent_step=1.24s eta=20:02:52 +2026-09-20 08:07:56 | INFO | train rank=0 | step=3480/50000 progress=7.0% loss=0.168043 lr=9.989501e-05 lr_backbone=9.989501e-05 elapsed=01:29:56 avg_step=1.55s recent_step=1.40s eta=20:02:16 +2026-09-20 08:08:19 | INFO | train rank=0 | step=3490/50000 progress=7.0% loss=0.154032 lr=9.989286e-05 lr_backbone=9.989286e-05 elapsed=01:30:19 avg_step=1.55s recent_step=2.31s eta=20:03:41 +2026-09-20 08:08:31 | INFO | train rank=0 | step=3500/50000 progress=7.0% loss=0.114797 lr=9.989068e-05 lr_backbone=9.989068e-05 elapsed=01:30:31 avg_step=1.55s recent_step=1.21s eta=20:02:40 +2026-09-20 08:08:46 | INFO | train rank=0 | step=3510/50000 progress=7.0% loss=0.120358 lr=9.988848e-05 lr_backbone=9.988848e-05 elapsed=01:30:46 avg_step=1.55s recent_step=1.48s eta=20:02:15 +2026-09-20 08:09:03 | INFO | train rank=0 | step=3520/50000 progress=7.0% loss=0.115054 lr=9.988627e-05 lr_backbone=9.988627e-05 elapsed=01:31:02 avg_step=1.55s recent_step=1.66s eta=20:02:13 +2026-09-20 08:09:21 | INFO | train rank=0 | step=3530/50000 progress=7.1% loss=0.139770 lr=9.988403e-05 lr_backbone=9.988403e-05 elapsed=01:31:21 avg_step=1.55s recent_step=1.86s eta=20:02:39 +2026-09-20 08:09:33 | INFO | train rank=0 | step=3540/50000 progress=7.1% loss=0.124542 lr=9.988176e-05 lr_backbone=9.988176e-05 elapsed=01:31:33 avg_step=1.55s recent_step=1.17s eta=20:01:33 +2026-09-20 08:09:47 | INFO | train rank=0 | step=3550/50000 progress=7.1% loss=0.149291 lr=9.987948e-05 lr_backbone=9.987948e-05 elapsed=01:31:47 avg_step=1.55s recent_step=1.43s eta=20:01:02 +2026-09-20 08:10:05 | INFO | train rank=0 | step=3560/50000 progress=7.1% loss=0.129426 lr=9.987718e-05 lr_backbone=9.987718e-05 elapsed=01:32:04 avg_step=1.55s recent_step=1.73s eta=20:01:10 +2026-09-20 08:10:18 | INFO | train rank=0 | step=3570/50000 progress=7.1% loss=0.147849 lr=9.987485e-05 lr_backbone=9.987485e-05 elapsed=01:32:18 avg_step=1.55s recent_step=1.37s eta=20:00:30 +2026-09-20 08:10:35 | INFO | train rank=0 | step=3580/50000 progress=7.2% loss=0.138122 lr=9.987250e-05 lr_backbone=9.987250e-05 elapsed=01:32:35 avg_step=1.55s recent_step=1.67s eta=20:00:30 +2026-09-20 08:10:51 | INFO | train rank=0 | step=3590/50000 progress=7.2% loss=0.161289 lr=9.987013e-05 lr_backbone=9.987013e-05 elapsed=01:32:51 avg_step=1.55s recent_step=1.63s eta=20:00:25 +2026-09-20 08:11:04 | INFO | train rank=0 | step=3600/50000 progress=7.2% loss=0.119283 lr=9.986773e-05 lr_backbone=9.986773e-05 elapsed=01:33:03 avg_step=1.55s recent_step=1.22s eta=19:59:27 +2026-09-20 08:11:20 | INFO | train rank=0 | step=3610/50000 progress=7.2% loss=0.119812 lr=9.986532e-05 lr_backbone=9.986532e-05 elapsed=01:33:20 avg_step=1.55s recent_step=1.67s eta=19:59:26 +2026-09-20 08:11:37 | INFO | train rank=0 | step=3620/50000 progress=7.2% loss=0.128046 lr=9.986288e-05 lr_backbone=9.986288e-05 elapsed=01:33:37 avg_step=1.55s recent_step=1.69s eta=19:59:28 +2026-09-20 08:11:50 | INFO | train rank=0 | step=3630/50000 progress=7.3% loss=0.119092 lr=9.986043e-05 lr_backbone=9.986043e-05 elapsed=01:33:50 avg_step=1.55s recent_step=1.29s eta=19:58:40 +2026-09-20 08:12:05 | INFO | train rank=0 | step=3640/50000 progress=7.3% loss=0.133010 lr=9.985795e-05 lr_backbone=9.985795e-05 elapsed=01:34:04 avg_step=1.55s recent_step=1.47s eta=19:58:14 +2026-09-20 08:12:24 | INFO | train rank=0 | step=3650/50000 progress=7.3% loss=0.133420 lr=9.985544e-05 lr_backbone=9.985544e-05 elapsed=01:34:24 avg_step=1.55s recent_step=1.92s eta=19:58:45 +2026-09-20 08:12:37 | INFO | train rank=0 | step=3660/50000 progress=7.3% loss=0.133707 lr=9.985292e-05 lr_backbone=9.985292e-05 elapsed=01:34:36 avg_step=1.55s recent_step=1.26s eta=19:57:53 +2026-09-20 08:12:53 | INFO | train rank=0 | step=3670/50000 progress=7.3% loss=0.105358 lr=9.985037e-05 lr_backbone=9.985037e-05 elapsed=01:34:53 avg_step=1.55s recent_step=1.68s eta=19:57:53 +2026-09-20 08:13:09 | INFO | train rank=0 | step=3680/50000 progress=7.4% loss=0.105710 lr=9.984781e-05 lr_backbone=9.984781e-05 elapsed=01:35:09 avg_step=1.55s recent_step=1.61s eta=19:57:45 +2026-09-20 08:13:29 | INFO | train rank=0 | step=3690/50000 progress=7.4% loss=0.130765 lr=9.984522e-05 lr_backbone=9.984522e-05 elapsed=01:35:29 avg_step=1.55s recent_step=1.98s eta=19:58:23 +2026-09-20 08:13:41 | INFO | train rank=0 | step=3700/50000 progress=7.4% loss=0.115874 lr=9.984261e-05 lr_backbone=9.984261e-05 elapsed=01:35:41 avg_step=1.55s recent_step=1.19s eta=19:57:23 +2026-09-20 08:13:53 | INFO | train rank=0 | step=3710/50000 progress=7.4% loss=0.157849 lr=9.983997e-05 lr_backbone=9.983997e-05 elapsed=01:35:53 avg_step=1.55s recent_step=1.21s eta=19:56:25 +2026-09-20 08:14:06 | INFO | train rank=0 | step=3720/50000 progress=7.4% loss=0.123918 lr=9.983732e-05 lr_backbone=9.983732e-05 elapsed=01:36:06 avg_step=1.55s recent_step=1.28s eta=19:55:36 +2026-09-20 08:14:24 | INFO | train rank=0 | step=3730/50000 progress=7.5% loss=0.141504 lr=9.983464e-05 lr_backbone=9.983464e-05 elapsed=01:36:23 avg_step=1.55s recent_step=1.78s eta=19:55:49 +2026-09-20 08:14:36 | INFO | train rank=0 | step=3740/50000 progress=7.5% loss=0.097554 lr=9.983194e-05 lr_backbone=9.983194e-05 elapsed=01:36:36 avg_step=1.55s recent_step=1.23s eta=19:54:54 +2026-09-20 08:15:00 | INFO | train rank=0 | step=3750/50000 progress=7.5% loss=0.110884 lr=9.982922e-05 lr_backbone=9.982922e-05 elapsed=01:36:59 avg_step=1.55s recent_step=2.37s eta=19:56:19 +2026-09-20 08:15:12 | INFO | train rank=0 | step=3760/50000 progress=7.5% loss=0.133084 lr=9.982648e-05 lr_backbone=9.982648e-05 elapsed=01:37:11 avg_step=1.55s recent_step=1.18s eta=19:55:17 +2026-09-20 08:15:26 | INFO | train rank=0 | step=3770/50000 progress=7.5% loss=0.127424 lr=9.982372e-05 lr_backbone=9.982372e-05 elapsed=01:37:26 avg_step=1.55s recent_step=1.46s eta=19:54:50 +2026-09-20 08:15:46 | INFO | train rank=0 | step=3780/50000 progress=7.6% loss=0.111700 lr=9.982093e-05 lr_backbone=9.982093e-05 elapsed=01:37:46 avg_step=1.55s recent_step=1.97s eta=19:55:26 +2026-09-20 08:15:58 | INFO | train rank=0 | step=3790/50000 progress=7.6% loss=0.141296 lr=9.981813e-05 lr_backbone=9.981813e-05 elapsed=01:37:57 avg_step=1.55s recent_step=1.16s eta=19:54:23 +2026-09-20 08:16:14 | INFO | train rank=0 | step=3800/50000 progress=7.6% loss=0.164692 lr=9.981530e-05 lr_backbone=9.981530e-05 elapsed=01:38:14 avg_step=1.55s recent_step=1.64s eta=19:54:19 +2026-09-20 08:16:31 | INFO | train rank=0 | step=3810/50000 progress=7.6% loss=0.119354 lr=9.981245e-05 lr_backbone=9.981245e-05 elapsed=01:38:30 avg_step=1.55s recent_step=1.68s eta=19:54:19 +2026-09-20 08:16:43 | INFO | train rank=0 | step=3820/50000 progress=7.6% loss=0.170660 lr=9.980957e-05 lr_backbone=9.980957e-05 elapsed=01:38:43 avg_step=1.55s recent_step=1.25s eta=19:53:27 +2026-09-20 08:16:57 | INFO | train rank=0 | step=3830/50000 progress=7.7% loss=0.143462 lr=9.980668e-05 lr_backbone=9.980668e-05 elapsed=01:38:56 avg_step=1.55s recent_step=1.37s eta=19:52:49 +2026-09-20 08:17:15 | INFO | train rank=0 | step=3840/50000 progress=7.7% loss=0.119920 lr=9.980376e-05 lr_backbone=9.980376e-05 elapsed=01:39:15 avg_step=1.55s recent_step=1.83s eta=19:53:07 +2026-09-20 08:17:29 | INFO | train rank=0 | step=3850/50000 progress=7.7% loss=0.124325 lr=9.980083e-05 lr_backbone=9.980083e-05 elapsed=01:39:28 avg_step=1.55s recent_step=1.34s eta=19:52:26 +2026-09-20 08:17:47 | INFO | train rank=0 | step=3860/50000 progress=7.7% loss=0.137613 lr=9.979787e-05 lr_backbone=9.979787e-05 elapsed=01:39:47 avg_step=1.55s recent_step=1.84s eta=19:52:46 +2026-09-20 08:18:07 | INFO | train rank=0 | step=3870/50000 progress=7.7% loss=0.116627 lr=9.979489e-05 lr_backbone=9.979489e-05 elapsed=01:40:06 avg_step=1.55s recent_step=1.96s eta=19:53:19 +2026-09-20 08:18:19 | INFO | train rank=0 | step=3880/50000 progress=7.8% loss=0.130658 lr=9.979188e-05 lr_backbone=9.979188e-05 elapsed=01:40:18 avg_step=1.55s recent_step=1.20s eta=19:52:21 +2026-09-20 08:18:34 | INFO | train rank=0 | step=3890/50000 progress=7.8% loss=0.104251 lr=9.978886e-05 lr_backbone=9.978886e-05 elapsed=01:40:34 avg_step=1.55s recent_step=1.58s eta=19:52:09 +2026-09-20 08:18:52 | INFO | train rank=0 | step=3900/50000 progress=7.8% loss=0.112055 lr=9.978581e-05 lr_backbone=9.978581e-05 elapsed=01:40:51 avg_step=1.55s recent_step=1.75s eta=19:52:17 +2026-09-20 08:19:04 | INFO | train rank=0 | step=3910/50000 progress=7.8% loss=0.089501 lr=9.978274e-05 lr_backbone=9.978274e-05 elapsed=01:41:04 avg_step=1.55s recent_step=1.23s eta=19:51:24 +2026-09-20 08:19:23 | INFO | train rank=0 | step=3920/50000 progress=7.8% loss=0.112074 lr=9.977965e-05 lr_backbone=9.977965e-05 elapsed=01:41:23 avg_step=1.55s recent_step=1.89s eta=19:51:48 +2026-09-20 08:19:35 | INFO | train rank=0 | step=3930/50000 progress=7.9% loss=0.126345 lr=9.977654e-05 lr_backbone=9.977654e-05 elapsed=01:41:34 avg_step=1.55s recent_step=1.16s eta=19:50:46 +2026-09-20 08:19:49 | INFO | train rank=0 | step=3940/50000 progress=7.9% loss=0.188862 lr=9.977341e-05 lr_backbone=9.977341e-05 elapsed=01:41:49 avg_step=1.55s recent_step=1.44s eta=19:50:18 +2026-09-20 08:20:06 | INFO | train rank=0 | step=3950/50000 progress=7.9% loss=0.128687 lr=9.977025e-05 lr_backbone=9.977025e-05 elapsed=01:42:06 avg_step=1.55s recent_step=1.72s eta=19:50:22 +2026-09-20 08:20:19 | INFO | train rank=0 | step=3960/50000 progress=7.9% loss=0.130865 lr=9.976707e-05 lr_backbone=9.976707e-05 elapsed=01:42:19 avg_step=1.55s recent_step=1.30s eta=19:49:38 +2026-09-20 08:20:35 | INFO | train rank=0 | step=3970/50000 progress=7.9% loss=0.099283 lr=9.976387e-05 lr_backbone=9.976387e-05 elapsed=01:42:35 avg_step=1.55s recent_step=1.56s eta=19:49:24 +2026-09-20 08:20:55 | INFO | train rank=0 | step=3980/50000 progress=8.0% loss=0.174179 lr=9.976065e-05 lr_backbone=9.976065e-05 elapsed=01:42:54 avg_step=1.55s recent_step=1.98s eta=19:49:58 +2026-09-20 08:21:07 | INFO | train rank=0 | step=3990/50000 progress=8.0% loss=0.122001 lr=9.975741e-05 lr_backbone=9.975741e-05 elapsed=01:43:06 avg_step=1.55s recent_step=1.19s eta=19:49:01 +2026-09-20 08:21:19 | INFO | train rank=0 | step=4000/50000 progress=8.0% loss=0.154909 lr=9.975415e-05 lr_backbone=9.975415e-05 elapsed=01:43:19 avg_step=1.55s recent_step=1.25s eta=19:48:11 +2026-09-20 08:21:31 | INFO | train rank=0 | step=4010/50000 progress=8.0% loss=0.122829 lr=9.975086e-05 lr_backbone=9.975086e-05 elapsed=01:43:31 avg_step=1.55s recent_step=1.18s eta=19:47:14 +2026-09-20 08:21:46 | INFO | train rank=0 | step=4020/50000 progress=8.0% loss=0.129223 lr=9.974755e-05 lr_backbone=9.974755e-05 elapsed=01:43:46 avg_step=1.55s recent_step=1.53s eta=19:46:56 +2026-09-20 08:22:05 | INFO | train rank=0 | step=4030/50000 progress=8.1% loss=0.108554 lr=9.974422e-05 lr_backbone=9.974422e-05 elapsed=01:44:05 avg_step=1.55s recent_step=1.87s eta=19:47:17 +2026-09-20 08:22:19 | INFO | train rank=0 | step=4040/50000 progress=8.1% loss=0.154445 lr=9.974087e-05 lr_backbone=9.974087e-05 elapsed=01:44:19 avg_step=1.55s recent_step=1.39s eta=19:46:44 +2026-09-20 08:22:32 | INFO | train rank=0 | step=4050/50000 progress=8.1% loss=0.158873 lr=9.973750e-05 lr_backbone=9.973750e-05 elapsed=01:44:32 avg_step=1.55s recent_step=1.31s eta=19:46:01 +2026-09-20 08:22:45 | INFO | train rank=0 | step=4060/50000 progress=8.1% loss=0.120548 lr=9.973410e-05 lr_backbone=9.973410e-05 elapsed=01:44:45 avg_step=1.55s recent_step=1.35s eta=19:45:23 +2026-09-20 08:23:07 | INFO | train rank=0 | step=4070/50000 progress=8.1% loss=0.108498 lr=9.973068e-05 lr_backbone=9.973068e-05 elapsed=01:45:07 avg_step=1.55s recent_step=2.17s eta=19:46:17 +2026-09-20 08:23:19 | INFO | train rank=0 | step=4080/50000 progress=8.2% loss=0.099833 lr=9.972725e-05 lr_backbone=9.972725e-05 elapsed=01:45:19 avg_step=1.55s recent_step=1.24s eta=19:45:26 +2026-09-20 08:23:39 | INFO | train rank=0 | step=4090/50000 progress=8.2% loss=0.130230 lr=9.972379e-05 lr_backbone=9.972379e-05 elapsed=01:45:39 avg_step=1.55s recent_step=1.95s eta=19:45:55 +2026-09-20 08:23:55 | INFO | train rank=0 | step=4100/50000 progress=8.2% loss=0.112362 lr=9.972030e-05 lr_backbone=9.972030e-05 elapsed=01:45:54 avg_step=1.55s recent_step=1.57s eta=19:45:42 +2026-09-20 08:24:07 | INFO | train rank=0 | step=4110/50000 progress=8.2% loss=0.114406 lr=9.971680e-05 lr_backbone=9.971680e-05 elapsed=01:46:06 avg_step=1.55s recent_step=1.22s eta=19:44:49 +2026-09-20 08:24:23 | INFO | train rank=0 | step=4120/50000 progress=8.2% loss=0.150755 lr=9.971327e-05 lr_backbone=9.971327e-05 elapsed=01:46:22 avg_step=1.55s recent_step=1.60s eta=19:44:40 +2026-09-20 08:24:38 | INFO | train rank=0 | step=4130/50000 progress=8.3% loss=0.114202 lr=9.970973e-05 lr_backbone=9.970973e-05 elapsed=01:46:37 avg_step=1.55s recent_step=1.48s eta=19:44:17 +2026-09-20 08:24:53 | INFO | train rank=0 | step=4140/50000 progress=8.3% loss=0.121581 lr=9.970616e-05 lr_backbone=9.970616e-05 elapsed=01:46:52 avg_step=1.55s recent_step=1.50s eta=19:43:56 +2026-09-20 08:25:08 | INFO | train rank=0 | step=4150/50000 progress=8.3% loss=0.115469 lr=9.970257e-05 lr_backbone=9.970257e-05 elapsed=01:47:08 avg_step=1.55s recent_step=1.55s eta=19:43:41 +2026-09-20 08:25:25 | INFO | train rank=0 | step=4160/50000 progress=8.3% loss=0.131059 lr=9.969895e-05 lr_backbone=9.969895e-05 elapsed=01:47:24 avg_step=1.55s recent_step=1.66s eta=19:43:38 +2026-09-20 08:25:41 | INFO | train rank=0 | step=4170/50000 progress=8.3% loss=0.168625 lr=9.969532e-05 lr_backbone=9.969532e-05 elapsed=01:47:41 avg_step=1.55s recent_step=1.66s eta=19:43:35 +2026-09-20 08:25:59 | INFO | train rank=0 | step=4180/50000 progress=8.4% loss=0.101824 lr=9.969166e-05 lr_backbone=9.969166e-05 elapsed=01:47:58 avg_step=1.55s recent_step=1.71s eta=19:43:37 +2026-09-20 08:26:11 | INFO | train rank=0 | step=4190/50000 progress=8.4% loss=0.131184 lr=9.968799e-05 lr_backbone=9.968799e-05 elapsed=01:48:10 avg_step=1.55s recent_step=1.22s eta=19:42:45 +2026-09-20 08:26:23 | INFO | train rank=0 | step=4200/50000 progress=8.4% loss=0.115225 lr=9.968429e-05 lr_backbone=9.968429e-05 elapsed=01:48:23 avg_step=1.55s recent_step=1.25s eta=19:41:57 +2026-09-20 08:26:36 | INFO | train rank=0 | step=4210/50000 progress=8.4% loss=0.103632 lr=9.968057e-05 lr_backbone=9.968057e-05 elapsed=01:48:36 avg_step=1.55s recent_step=1.28s eta=19:41:12 +2026-09-20 08:26:55 | INFO | train rank=0 | step=4220/50000 progress=8.4% loss=0.115657 lr=9.967682e-05 lr_backbone=9.967682e-05 elapsed=01:48:55 avg_step=1.55s recent_step=1.94s eta=19:41:40 +2026-09-20 08:27:12 | INFO | train rank=0 | step=4230/50000 progress=8.5% loss=0.136272 lr=9.967306e-05 lr_backbone=9.967306e-05 elapsed=01:49:11 avg_step=1.55s recent_step=1.62s eta=19:41:32 +2026-09-20 08:27:26 | INFO | train rank=0 | step=4240/50000 progress=8.5% loss=0.108546 lr=9.966927e-05 lr_backbone=9.966927e-05 elapsed=01:49:26 avg_step=1.55s recent_step=1.48s eta=19:41:08 +2026-09-20 08:27:43 | INFO | train rank=0 | step=4250/50000 progress=8.5% loss=0.216854 lr=9.966546e-05 lr_backbone=9.966546e-05 elapsed=01:49:42 avg_step=1.55s recent_step=1.65s eta=19:41:03 +2026-09-20 08:28:00 | INFO | train rank=0 | step=4260/50000 progress=8.5% loss=0.128865 lr=9.966163e-05 lr_backbone=9.966163e-05 elapsed=01:50:00 avg_step=1.55s recent_step=1.73s eta=19:41:08 +2026-09-20 08:28:16 | INFO | train rank=0 | step=4270/50000 progress=8.5% loss=0.132634 lr=9.965778e-05 lr_backbone=9.965778e-05 elapsed=01:50:16 avg_step=1.55s recent_step=1.57s eta=19:40:54 +2026-09-20 08:28:32 | INFO | train rank=0 | step=4280/50000 progress=8.6% loss=0.130402 lr=9.965391e-05 lr_backbone=9.965391e-05 elapsed=01:50:32 avg_step=1.55s recent_step=1.65s eta=19:40:50 +2026-09-20 08:28:48 | INFO | train rank=0 | step=4290/50000 progress=8.6% loss=0.148901 lr=9.965001e-05 lr_backbone=9.965001e-05 elapsed=01:50:47 avg_step=1.55s recent_step=1.52s eta=19:40:31 +2026-09-20 08:28:59 | INFO | train rank=0 | step=4300/50000 progress=8.6% loss=0.231193 lr=9.964610e-05 lr_backbone=9.964610e-05 elapsed=01:50:59 avg_step=1.55s recent_step=1.18s eta=19:39:37 +2026-09-20 08:29:20 | INFO | train rank=0 | step=4310/50000 progress=8.6% loss=0.127941 lr=9.964216e-05 lr_backbone=9.964216e-05 elapsed=01:51:19 avg_step=1.55s recent_step=2.04s eta=19:40:13 +2026-09-20 08:29:33 | INFO | train rank=0 | step=4320/50000 progress=8.6% loss=0.120896 lr=9.963820e-05 lr_backbone=9.963820e-05 elapsed=01:51:32 avg_step=1.55s recent_step=1.27s eta=19:39:29 +2026-09-20 08:29:45 | INFO | train rank=0 | step=4330/50000 progress=8.7% loss=0.138946 lr=9.963422e-05 lr_backbone=9.963422e-05 elapsed=01:51:45 avg_step=1.55s recent_step=1.28s eta=19:38:45 +2026-09-20 08:30:06 | INFO | train rank=0 | step=4340/50000 progress=8.7% loss=0.129339 lr=9.963021e-05 lr_backbone=9.963021e-05 elapsed=01:52:06 avg_step=1.55s recent_step=2.06s eta=19:39:23 +2026-09-20 08:30:20 | INFO | train rank=0 | step=4350/50000 progress=8.7% loss=0.147907 lr=9.962619e-05 lr_backbone=9.962619e-05 elapsed=01:52:19 avg_step=1.55s recent_step=1.38s eta=19:38:49 +2026-09-20 08:30:32 | INFO | train rank=0 | step=4360/50000 progress=8.7% loss=0.150787 lr=9.962214e-05 lr_backbone=9.962214e-05 elapsed=01:52:32 avg_step=1.55s recent_step=1.25s eta=19:38:02 +2026-09-20 08:30:50 | INFO | train rank=0 | step=4370/50000 progress=8.7% loss=0.121206 lr=9.961807e-05 lr_backbone=9.961807e-05 elapsed=01:52:50 avg_step=1.55s recent_step=1.79s eta=19:38:12 +2026-09-20 08:31:02 | INFO | train rank=0 | step=4380/50000 progress=8.8% loss=0.120734 lr=9.961398e-05 lr_backbone=9.961398e-05 elapsed=01:53:02 avg_step=1.55s recent_step=1.23s eta=19:37:23 +2026-09-20 08:31:25 | INFO | train rank=0 | step=4390/50000 progress=8.8% loss=0.099150 lr=9.960987e-05 lr_backbone=9.960987e-05 elapsed=01:53:24 avg_step=1.55s recent_step=2.22s eta=19:38:17 +2026-09-20 08:31:37 | INFO | train rank=0 | step=4400/50000 progress=8.8% loss=0.138385 lr=9.960574e-05 lr_backbone=9.960574e-05 elapsed=01:53:36 avg_step=1.55s recent_step=1.19s eta=19:37:25 +2026-09-20 08:31:48 | INFO | train rank=0 | step=4410/50000 progress=8.8% loss=0.142736 lr=9.960158e-05 lr_backbone=9.960158e-05 elapsed=01:53:48 avg_step=1.55s recent_step=1.18s eta=19:36:31 +2026-09-20 08:32:05 | INFO | train rank=0 | step=4420/50000 progress=8.8% loss=0.140467 lr=9.959740e-05 lr_backbone=9.959740e-05 elapsed=01:54:05 avg_step=1.55s recent_step=1.71s eta=19:36:33 +2026-09-20 08:32:22 | INFO | train rank=0 | step=4430/50000 progress=8.9% loss=0.110106 lr=9.959320e-05 lr_backbone=9.959320e-05 elapsed=01:54:21 avg_step=1.55s recent_step=1.61s eta=19:36:24 +2026-09-20 08:32:34 | INFO | train rank=0 | step=4440/50000 progress=8.9% loss=0.129259 lr=9.958898e-05 lr_backbone=9.958898e-05 elapsed=01:54:34 avg_step=1.55s recent_step=1.27s eta=19:35:39 +2026-09-20 08:32:51 | INFO | train rank=0 | step=4450/50000 progress=8.9% loss=0.129228 lr=9.958474e-05 lr_backbone=9.958474e-05 elapsed=01:54:51 avg_step=1.55s recent_step=1.72s eta=19:35:42 +2026-09-20 08:33:04 | INFO | train rank=0 | step=4460/50000 progress=8.9% loss=0.114288 lr=9.958048e-05 lr_backbone=9.958048e-05 elapsed=01:55:04 avg_step=1.55s recent_step=1.28s eta=19:34:59 +2026-09-20 08:33:24 | INFO | train rank=0 | step=4470/50000 progress=8.9% loss=0.181417 lr=9.957619e-05 lr_backbone=9.957619e-05 elapsed=01:55:23 avg_step=1.55s recent_step=1.94s eta=19:35:23 +2026-09-20 08:33:38 | INFO | train rank=0 | step=4480/50000 progress=9.0% loss=0.154383 lr=9.957188e-05 lr_backbone=9.957188e-05 elapsed=01:55:38 avg_step=1.55s recent_step=1.45s eta=19:34:58 +2026-09-20 08:33:56 | INFO | train rank=0 | step=4490/50000 progress=9.0% loss=0.107177 lr=9.956755e-05 lr_backbone=9.956755e-05 elapsed=01:55:56 avg_step=1.55s recent_step=1.80s eta=19:35:08 +2026-09-20 08:34:10 | INFO | train rank=0 | step=4500/50000 progress=9.0% loss=0.110014 lr=9.956320e-05 lr_backbone=9.956320e-05 elapsed=01:56:09 avg_step=1.55s recent_step=1.36s eta=19:34:34 +2026-09-20 08:34:32 | INFO | train rank=0 | step=4510/50000 progress=9.0% loss=0.166583 lr=9.955883e-05 lr_backbone=9.955883e-05 elapsed=01:56:31 avg_step=1.55s recent_step=2.17s eta=19:35:21 +2026-09-20 08:34:43 | INFO | train rank=0 | step=4520/50000 progress=9.0% loss=0.119786 lr=9.955444e-05 lr_backbone=9.955444e-05 elapsed=01:56:43 avg_step=1.55s recent_step=1.16s eta=19:34:26 +2026-09-20 08:34:56 | INFO | train rank=0 | step=4530/50000 progress=9.1% loss=0.113852 lr=9.955002e-05 lr_backbone=9.955002e-05 elapsed=01:56:56 avg_step=1.55s recent_step=1.30s eta=19:33:45 +2026-09-20 08:35:12 | INFO | train rank=0 | step=4540/50000 progress=9.1% loss=0.118129 lr=9.954558e-05 lr_backbone=9.954558e-05 elapsed=01:57:12 avg_step=1.55s recent_step=1.61s eta=19:33:35 +2026-09-20 08:35:27 | INFO | train rank=0 | step=4550/50000 progress=9.1% loss=0.167733 lr=9.954112e-05 lr_backbone=9.954112e-05 elapsed=01:57:27 avg_step=1.55s recent_step=1.49s eta=19:33:13 +2026-09-20 08:35:44 | INFO | train rank=0 | step=4560/50000 progress=9.1% loss=0.121752 lr=9.953664e-05 lr_backbone=9.953664e-05 elapsed=01:57:44 avg_step=1.55s recent_step=1.70s eta=19:33:13 +2026-09-20 08:36:00 | INFO | train rank=0 | step=4570/50000 progress=9.1% loss=0.228206 lr=9.953214e-05 lr_backbone=9.953214e-05 elapsed=01:58:00 avg_step=1.55s recent_step=1.61s eta=19:33:04 +2026-09-20 08:36:13 | INFO | train rank=0 | step=4580/50000 progress=9.2% loss=0.183244 lr=9.952762e-05 lr_backbone=9.952762e-05 elapsed=01:58:12 avg_step=1.55s recent_step=1.25s eta=19:32:19 +2026-09-20 08:36:30 | INFO | train rank=0 | step=4590/50000 progress=9.2% loss=0.108041 lr=9.952307e-05 lr_backbone=9.952307e-05 elapsed=01:58:29 avg_step=1.55s recent_step=1.69s eta=19:32:18 +2026-09-20 08:36:45 | INFO | train rank=0 | step=4600/50000 progress=9.2% loss=0.087614 lr=9.951850e-05 lr_backbone=9.951850e-05 elapsed=01:58:45 avg_step=1.55s recent_step=1.55s eta=19:32:02 +2026-09-20 08:36:58 | INFO | train rank=0 | step=4610/50000 progress=9.2% loss=0.105914 lr=9.951391e-05 lr_backbone=9.951391e-05 elapsed=01:58:57 avg_step=1.55s recent_step=1.26s eta=19:31:18 +2026-09-20 08:37:14 | INFO | train rank=0 | step=4620/50000 progress=9.2% loss=0.105278 lr=9.950930e-05 lr_backbone=9.950930e-05 elapsed=01:59:13 avg_step=1.55s recent_step=1.61s eta=19:31:08 +2026-09-20 08:37:32 | INFO | train rank=0 | step=4630/50000 progress=9.3% loss=0.138751 lr=9.950467e-05 lr_backbone=9.950467e-05 elapsed=01:59:32 avg_step=1.55s recent_step=1.83s eta=19:31:20 +2026-09-20 08:37:47 | INFO | train rank=0 | step=4640/50000 progress=9.3% loss=0.111808 lr=9.950002e-05 lr_backbone=9.950002e-05 elapsed=01:59:47 avg_step=1.55s recent_step=1.52s eta=19:31:01 +2026-09-20 08:38:01 | INFO | train rank=0 | step=4650/50000 progress=9.3% loss=0.117362 lr=9.949534e-05 lr_backbone=9.949534e-05 elapsed=02:00:01 avg_step=1.55s recent_step=1.43s eta=19:30:34 +2026-09-20 08:38:19 | INFO | train rank=0 | step=4660/50000 progress=9.3% loss=0.123764 lr=9.949064e-05 lr_backbone=9.949064e-05 elapsed=02:00:18 avg_step=1.55s recent_step=1.72s eta=19:30:35 +2026-09-20 08:38:37 | INFO | train rank=0 | step=4670/50000 progress=9.3% loss=0.104355 lr=9.948593e-05 lr_backbone=9.948593e-05 elapsed=02:00:37 avg_step=1.55s recent_step=1.87s eta=19:30:50 +2026-09-20 08:38:50 | INFO | train rank=0 | step=4680/50000 progress=9.4% loss=0.122618 lr=9.948118e-05 lr_backbone=9.948118e-05 elapsed=02:00:50 avg_step=1.55s recent_step=1.27s eta=19:30:08 +2026-09-20 08:39:03 | INFO | train rank=0 | step=4690/50000 progress=9.4% loss=0.112573 lr=9.947642e-05 lr_backbone=9.947642e-05 elapsed=02:01:02 avg_step=1.55s recent_step=1.25s eta=19:29:24 +2026-09-20 08:39:18 | INFO | train rank=0 | step=4700/50000 progress=9.4% loss=0.105329 lr=9.947164e-05 lr_backbone=9.947164e-05 elapsed=02:01:18 avg_step=1.55s recent_step=1.59s eta=19:29:13 +2026-09-20 08:39:37 | INFO | train rank=0 | step=4710/50000 progress=9.4% loss=0.122305 lr=9.946683e-05 lr_backbone=9.946683e-05 elapsed=02:01:36 avg_step=1.55s recent_step=1.82s eta=19:29:24 +2026-09-20 08:39:53 | INFO | train rank=0 | step=4720/50000 progress=9.4% loss=0.122165 lr=9.946201e-05 lr_backbone=9.946201e-05 elapsed=02:01:53 avg_step=1.55s recent_step=1.63s eta=19:29:16 +2026-09-20 08:40:06 | INFO | train rank=0 | step=4730/50000 progress=9.5% loss=0.135003 lr=9.945716e-05 lr_backbone=9.945716e-05 elapsed=02:02:05 avg_step=1.55s recent_step=1.28s eta=19:28:35 +2026-09-20 08:40:20 | INFO | train rank=0 | step=4740/50000 progress=9.5% loss=0.130773 lr=9.945229e-05 lr_backbone=9.945229e-05 elapsed=02:02:20 avg_step=1.55s recent_step=1.45s eta=19:28:09 +2026-09-20 08:40:42 | INFO | train rank=0 | step=4750/50000 progress=9.5% loss=0.133561 lr=9.944739e-05 lr_backbone=9.944739e-05 elapsed=02:02:42 avg_step=1.55s recent_step=2.22s eta=19:28:57 +2026-09-20 08:40:54 | INFO | train rank=0 | step=4760/50000 progress=9.5% loss=0.176328 lr=9.944248e-05 lr_backbone=9.944248e-05 elapsed=02:02:54 avg_step=1.55s recent_step=1.20s eta=19:28:09 +2026-09-20 08:41:06 | INFO | train rank=0 | step=4770/50000 progress=9.5% loss=0.123800 lr=9.943754e-05 lr_backbone=9.943754e-05 elapsed=02:03:06 avg_step=1.55s recent_step=1.18s eta=19:27:19 +2026-09-20 08:41:19 | INFO | train rank=0 | step=4780/50000 progress=9.6% loss=0.130518 lr=9.943259e-05 lr_backbone=9.943259e-05 elapsed=02:03:18 avg_step=1.55s recent_step=1.26s eta=19:26:36 +2026-09-20 08:41:43 | INFO | train rank=0 | step=4790/50000 progress=9.6% loss=0.129666 lr=9.942761e-05 lr_backbone=9.942761e-05 elapsed=02:03:43 avg_step=1.55s recent_step=2.41s eta=19:27:41 +2026-09-20 08:41:56 | INFO | train rank=0 | step=4800/50000 progress=9.6% loss=0.144100 lr=9.942261e-05 lr_backbone=9.942261e-05 elapsed=02:03:56 avg_step=1.55s recent_step=1.30s eta=19:27:02 +2026-09-20 08:42:09 | INFO | train rank=0 | step=4810/50000 progress=9.6% loss=0.104425 lr=9.941759e-05 lr_backbone=9.941759e-05 elapsed=02:04:08 avg_step=1.55s recent_step=1.29s eta=19:26:22 +2026-09-20 08:42:26 | INFO | train rank=0 | step=4820/50000 progress=9.6% loss=0.113695 lr=9.941254e-05 lr_backbone=9.941254e-05 elapsed=02:04:26 avg_step=1.55s recent_step=1.76s eta=19:26:27 +2026-09-20 08:42:43 | INFO | train rank=0 | step=4830/50000 progress=9.7% loss=0.123833 lr=9.940748e-05 lr_backbone=9.940748e-05 elapsed=02:04:43 avg_step=1.55s recent_step=1.71s eta=19:26:26 +2026-09-20 08:43:00 | INFO | train rank=0 | step=4840/50000 progress=9.7% loss=0.133003 lr=9.940239e-05 lr_backbone=9.940239e-05 elapsed=02:04:59 avg_step=1.55s recent_step=1.63s eta=19:26:18 +2026-09-20 08:43:12 | INFO | train rank=0 | step=4850/50000 progress=9.7% loss=0.113306 lr=9.939728e-05 lr_backbone=9.939728e-05 elapsed=02:05:12 avg_step=1.55s recent_step=1.21s eta=19:25:31 +2026-09-20 08:43:26 | INFO | train rank=0 | step=4860/50000 progress=9.7% loss=0.125769 lr=9.939215e-05 lr_backbone=9.939215e-05 elapsed=02:05:26 avg_step=1.55s recent_step=1.43s eta=19:25:04 +2026-09-20 08:43:42 | INFO | train rank=0 | step=4870/50000 progress=9.7% loss=0.132658 lr=9.938700e-05 lr_backbone=9.938700e-05 elapsed=02:05:42 avg_step=1.55s recent_step=1.61s eta=19:24:55 +2026-09-20 08:43:58 | INFO | train rank=0 | step=4880/50000 progress=9.8% loss=0.221412 lr=9.938183e-05 lr_backbone=9.938183e-05 elapsed=02:05:58 avg_step=1.55s recent_step=1.57s eta=19:24:42 +2026-09-20 08:44:15 | INFO | train rank=0 | step=4890/50000 progress=9.8% loss=0.144260 lr=9.937663e-05 lr_backbone=9.937663e-05 elapsed=02:06:15 avg_step=1.55s recent_step=1.69s eta=19:24:39 +2026-09-20 08:44:30 | INFO | train rank=0 | step=4900/50000 progress=9.8% loss=0.164523 lr=9.937142e-05 lr_backbone=9.937142e-05 elapsed=02:06:29 avg_step=1.55s recent_step=1.48s eta=19:24:17 +2026-09-20 08:44:50 | INFO | train rank=0 | step=4910/50000 progress=9.8% loss=0.143567 lr=9.936618e-05 lr_backbone=9.936618e-05 elapsed=02:06:49 avg_step=1.55s recent_step=1.98s eta=19:24:42 +2026-09-20 08:45:04 | INFO | train rank=0 | step=4920/50000 progress=9.8% loss=0.140651 lr=9.936092e-05 lr_backbone=9.936092e-05 elapsed=02:07:04 avg_step=1.55s recent_step=1.44s eta=19:24:16 +2026-09-20 08:45:24 | INFO | train rank=0 | step=4930/50000 progress=9.9% loss=0.110291 lr=9.935564e-05 lr_backbone=9.935564e-05 elapsed=02:07:23 avg_step=1.55s recent_step=1.98s eta=19:24:40 +2026-09-20 08:45:35 | INFO | train rank=0 | step=4940/50000 progress=9.9% loss=0.119023 lr=9.935034e-05 lr_backbone=9.935034e-05 elapsed=02:07:35 avg_step=1.55s recent_step=1.16s eta=19:23:49 +2026-09-20 08:45:48 | INFO | train rank=0 | step=4950/50000 progress=9.9% loss=0.119144 lr=9.934501e-05 lr_backbone=9.934501e-05 elapsed=02:07:48 avg_step=1.55s recent_step=1.27s eta=19:23:08 +2026-09-20 08:46:03 | INFO | train rank=0 | step=4960/50000 progress=9.9% loss=0.108491 lr=9.933966e-05 lr_backbone=9.933966e-05 elapsed=02:08:02 avg_step=1.55s recent_step=1.46s eta=19:22:44 +2026-09-20 08:46:18 | INFO | train rank=0 | step=4970/50000 progress=9.9% loss=0.137304 lr=9.933430e-05 lr_backbone=9.933430e-05 elapsed=02:08:18 avg_step=1.55s recent_step=1.53s eta=19:22:26 +2026-09-20 08:46:34 | INFO | train rank=0 | step=4980/50000 progress=10.0% loss=0.115187 lr=9.932891e-05 lr_backbone=9.932891e-05 elapsed=02:08:34 avg_step=1.55s recent_step=1.65s eta=19:22:20 +2026-09-20 08:46:53 | INFO | train rank=0 | step=4990/50000 progress=10.0% loss=0.136027 lr=9.932350e-05 lr_backbone=9.932350e-05 elapsed=02:08:53 avg_step=1.55s recent_step=1.89s eta=19:22:35 +2026-09-20 08:47:07 | INFO | train rank=0 | step=5000/50000 progress=10.0% loss=0.142565 lr=9.931807e-05 lr_backbone=9.931807e-05 elapsed=02:09:07 avg_step=1.55s recent_step=1.41s eta=19:22:07 +2026-09-20 08:47:24 | INFO | train rank=0 | step=5010/50000 progress=10.0% loss=0.118495 lr=9.931261e-05 lr_backbone=9.931261e-05 elapsed=02:09:24 avg_step=1.55s recent_step=1.69s eta=19:22:05 +2026-09-20 08:47:40 | INFO | train rank=0 | step=5020/50000 progress=10.0% loss=0.108626 lr=9.930714e-05 lr_backbone=9.930714e-05 elapsed=02:09:40 avg_step=1.55s recent_step=1.58s eta=19:21:52 +2026-09-20 08:48:00 | INFO | train rank=0 | step=5030/50000 progress=10.1% loss=0.151391 lr=9.930164e-05 lr_backbone=9.930164e-05 elapsed=02:10:00 avg_step=1.55s recent_step=1.99s eta=19:22:16 +2026-09-20 08:48:12 | INFO | train rank=0 | step=5040/50000 progress=10.1% loss=0.121491 lr=9.929612e-05 lr_backbone=9.929612e-05 elapsed=02:10:12 avg_step=1.55s recent_step=1.22s eta=19:21:31 +2026-09-20 08:48:24 | INFO | train rank=0 | step=5050/50000 progress=10.1% loss=0.122702 lr=9.929058e-05 lr_backbone=9.929058e-05 elapsed=02:10:23 avg_step=1.55s recent_step=1.15s eta=19:20:40 +2026-09-20 08:48:36 | INFO | train rank=0 | step=5060/50000 progress=10.1% loss=0.108377 lr=9.928502e-05 lr_backbone=9.928502e-05 elapsed=02:10:36 avg_step=1.55s recent_step=1.26s eta=19:19:59 +2026-09-20 08:48:57 | INFO | train rank=0 | step=5070/50000 progress=10.1% loss=0.094564 lr=9.927944e-05 lr_backbone=9.927944e-05 elapsed=02:10:57 avg_step=1.55s recent_step=2.05s eta=19:20:28 +2026-09-20 08:49:10 | INFO | train rank=0 | step=5080/50000 progress=10.2% loss=0.120076 lr=9.927383e-05 lr_backbone=9.927383e-05 elapsed=02:11:10 avg_step=1.55s recent_step=1.30s eta=19:19:51 +2026-09-20 08:49:23 | INFO | train rank=0 | step=5090/50000 progress=10.2% loss=0.142873 lr=9.926820e-05 lr_backbone=9.926820e-05 elapsed=02:11:22 avg_step=1.55s recent_step=1.29s eta=19:19:12 +2026-09-20 08:49:42 | INFO | train rank=0 | step=5100/50000 progress=10.2% loss=0.117233 lr=9.926256e-05 lr_backbone=9.926256e-05 elapsed=02:11:42 avg_step=1.55s recent_step=1.95s eta=19:19:32 +2026-09-20 08:50:01 | INFO | train rank=0 | step=5110/50000 progress=10.2% loss=0.167051 lr=9.925689e-05 lr_backbone=9.925689e-05 elapsed=02:12:00 avg_step=1.55s recent_step=1.83s eta=19:19:41 +2026-09-20 08:50:14 | INFO | train rank=0 | step=5120/50000 progress=10.2% loss=0.118732 lr=9.925120e-05 lr_backbone=9.925120e-05 elapsed=02:12:14 avg_step=1.55s recent_step=1.37s eta=19:19:10 +2026-09-20 08:50:31 | INFO | train rank=0 | step=5130/50000 progress=10.3% loss=0.099311 lr=9.924548e-05 lr_backbone=9.924548e-05 elapsed=02:12:30 avg_step=1.55s recent_step=1.65s eta=19:19:03 +2026-09-20 08:50:43 | INFO | train rank=0 | step=5140/50000 progress=10.3% loss=0.130590 lr=9.923975e-05 lr_backbone=9.923975e-05 elapsed=02:12:42 avg_step=1.55s recent_step=1.18s eta=19:18:15 +2026-09-20 08:51:02 | INFO | train rank=0 | step=5150/50000 progress=10.3% loss=0.112507 lr=9.923399e-05 lr_backbone=9.923399e-05 elapsed=02:13:02 avg_step=1.55s recent_step=1.93s eta=19:18:33 +2026-09-20 08:51:19 | INFO | train rank=0 | step=5160/50000 progress=10.3% loss=0.103744 lr=9.922822e-05 lr_backbone=9.922822e-05 elapsed=02:13:19 avg_step=1.55s recent_step=1.73s eta=19:18:33 +2026-09-20 08:51:31 | INFO | train rank=0 | step=5170/50000 progress=10.3% loss=0.121564 lr=9.922242e-05 lr_backbone=9.922242e-05 elapsed=02:13:31 avg_step=1.55s recent_step=1.18s eta=19:17:45 +2026-09-20 08:51:43 | INFO | train rank=0 | step=5180/50000 progress=10.4% loss=0.124576 lr=9.921660e-05 lr_backbone=9.921660e-05 elapsed=02:13:43 avg_step=1.55s recent_step=1.22s eta=19:17:01 +2026-09-20 08:52:00 | INFO | train rank=0 | step=5190/50000 progress=10.4% loss=0.120594 lr=9.921076e-05 lr_backbone=9.921076e-05 elapsed=02:13:59 avg_step=1.55s recent_step=1.67s eta=19:16:56 +2026-09-20 08:52:12 | INFO | train rank=0 | step=5200/50000 progress=10.4% loss=0.107102 lr=9.920489e-05 lr_backbone=9.920489e-05 elapsed=02:14:11 avg_step=1.55s recent_step=1.20s eta=19:16:11 +2026-09-20 08:52:27 | INFO | train rank=0 | step=5210/50000 progress=10.4% loss=0.114270 lr=9.919901e-05 lr_backbone=9.919901e-05 elapsed=02:14:27 avg_step=1.55s recent_step=1.54s eta=19:15:54 +2026-09-20 08:52:47 | INFO | train rank=0 | step=5220/50000 progress=10.4% loss=0.125510 lr=9.919310e-05 lr_backbone=9.919310e-05 elapsed=02:14:46 avg_step=1.55s recent_step=1.94s eta=19:16:12 +2026-09-20 08:53:05 | INFO | train rank=0 | step=5230/50000 progress=10.5% loss=0.088567 lr=9.918717e-05 lr_backbone=9.918717e-05 elapsed=02:15:04 avg_step=1.55s recent_step=1.79s eta=19:16:17 +2026-09-20 08:53:21 | INFO | train rank=0 | step=5240/50000 progress=10.5% loss=0.151282 lr=9.918122e-05 lr_backbone=9.918122e-05 elapsed=02:15:20 avg_step=1.55s recent_step=1.61s eta=19:16:07 +2026-09-20 08:53:33 | INFO | train rank=0 | step=5250/50000 progress=10.5% loss=0.157822 lr=9.917525e-05 lr_backbone=9.917525e-05 elapsed=02:15:33 avg_step=1.55s recent_step=1.25s eta=19:15:26 +2026-09-20 08:53:48 | INFO | train rank=0 | step=5260/50000 progress=10.5% loss=0.111188 lr=9.916926e-05 lr_backbone=9.916926e-05 elapsed=02:15:48 avg_step=1.55s recent_step=1.52s eta=19:15:08 +2026-09-20 08:54:06 | INFO | train rank=0 | step=5270/50000 progress=10.5% loss=0.105983 lr=9.916325e-05 lr_backbone=9.916325e-05 elapsed=02:16:05 avg_step=1.55s recent_step=1.71s eta=19:15:07 +2026-09-20 08:54:24 | INFO | train rank=0 | step=5280/50000 progress=10.6% loss=0.130982 lr=9.915721e-05 lr_backbone=9.915721e-05 elapsed=02:16:24 avg_step=1.55s recent_step=1.87s eta=19:15:19 +2026-09-20 08:54:36 | INFO | train rank=0 | step=5290/50000 progress=10.6% loss=0.142349 lr=9.915116e-05 lr_backbone=9.915116e-05 elapsed=02:16:36 avg_step=1.55s recent_step=1.21s eta=19:14:34 +2026-09-20 08:54:50 | INFO | train rank=0 | step=5300/50000 progress=10.6% loss=0.136743 lr=9.914508e-05 lr_backbone=9.914508e-05 elapsed=02:16:50 avg_step=1.55s recent_step=1.36s eta=19:14:03 +2026-09-20 08:55:07 | INFO | train rank=0 | step=5310/50000 progress=10.6% loss=0.127612 lr=9.913898e-05 lr_backbone=9.913898e-05 elapsed=02:17:07 avg_step=1.55s recent_step=1.71s eta=19:14:01 +2026-09-20 08:55:24 | INFO | train rank=0 | step=5320/50000 progress=10.6% loss=0.109054 lr=9.913286e-05 lr_backbone=9.913286e-05 elapsed=02:17:24 avg_step=1.55s recent_step=1.70s eta=19:13:58 +2026-09-20 08:55:39 | INFO | train rank=0 | step=5330/50000 progress=10.7% loss=0.128743 lr=9.912671e-05 lr_backbone=9.912671e-05 elapsed=02:17:39 avg_step=1.55s recent_step=1.51s eta=19:13:39 +2026-09-20 08:55:54 | INFO | train rank=0 | step=5340/50000 progress=10.7% loss=0.101178 lr=9.912055e-05 lr_backbone=9.912055e-05 elapsed=02:17:54 avg_step=1.55s recent_step=1.47s eta=19:13:17 +2026-09-20 08:56:08 | INFO | train rank=0 | step=5350/50000 progress=10.7% loss=0.103089 lr=9.911436e-05 lr_backbone=9.911436e-05 elapsed=02:18:08 avg_step=1.55s recent_step=1.42s eta=19:12:51 +2026-09-20 08:56:26 | INFO | train rank=0 | step=5360/50000 progress=10.7% loss=0.149903 lr=9.910816e-05 lr_backbone=9.910816e-05 elapsed=02:18:25 avg_step=1.55s recent_step=1.76s eta=19:12:53 +2026-09-20 08:56:38 | INFO | train rank=0 | step=5370/50000 progress=10.7% loss=0.120781 lr=9.910193e-05 lr_backbone=9.910193e-05 elapsed=02:18:38 avg_step=1.55s recent_step=1.27s eta=19:12:14 +2026-09-20 08:56:55 | INFO | train rank=0 | step=5380/50000 progress=10.8% loss=0.150911 lr=9.909568e-05 lr_backbone=9.909568e-05 elapsed=02:18:55 avg_step=1.55s recent_step=1.67s eta=19:12:08 +2026-09-20 08:57:12 | INFO | train rank=0 | step=5390/50000 progress=10.8% loss=0.102221 lr=9.908940e-05 lr_backbone=9.908940e-05 elapsed=02:19:11 avg_step=1.55s recent_step=1.66s eta=19:12:02 +2026-09-20 08:57:30 | INFO | train rank=0 | step=5400/50000 progress=10.8% loss=0.126150 lr=9.908311e-05 lr_backbone=9.908311e-05 elapsed=02:19:29 avg_step=1.55s recent_step=1.81s eta=19:12:08 +2026-09-20 08:57:42 | INFO | train rank=0 | step=5410/50000 progress=10.8% loss=0.117131 lr=9.907680e-05 lr_backbone=9.907680e-05 elapsed=02:19:42 avg_step=1.55s recent_step=1.27s eta=19:11:30 +2026-09-20 08:57:55 | INFO | train rank=0 | step=5420/50000 progress=10.8% loss=0.086032 lr=9.907046e-05 lr_backbone=9.907046e-05 elapsed=02:19:55 avg_step=1.55s recent_step=1.29s eta=19:10:53 +2026-09-20 08:58:16 | INFO | train rank=0 | step=5430/50000 progress=10.9% loss=0.101945 lr=9.906410e-05 lr_backbone=9.906410e-05 elapsed=02:20:16 avg_step=1.55s recent_step=2.10s eta=19:11:23 +2026-09-20 08:58:33 | INFO | train rank=0 | step=5440/50000 progress=10.9% loss=0.110443 lr=9.905772e-05 lr_backbone=9.905772e-05 elapsed=02:20:32 avg_step=1.55s recent_step=1.62s eta=19:11:13 +2026-09-20 08:58:44 | INFO | train rank=0 | step=5450/50000 progress=10.9% loss=0.109869 lr=9.905132e-05 lr_backbone=9.905132e-05 elapsed=02:20:44 avg_step=1.55s recent_step=1.18s eta=19:10:27 +2026-09-20 08:59:02 | INFO | train rank=0 | step=5460/50000 progress=10.9% loss=0.117394 lr=9.904490e-05 lr_backbone=9.904490e-05 elapsed=02:21:01 avg_step=1.55s recent_step=1.76s eta=19:10:28 +2026-09-20 08:59:20 | INFO | train rank=0 | step=5470/50000 progress=10.9% loss=0.109099 lr=9.903846e-05 lr_backbone=9.903846e-05 elapsed=02:21:20 avg_step=1.55s recent_step=1.82s eta=19:10:35 +2026-09-20 08:59:34 | INFO | train rank=0 | step=5480/50000 progress=11.0% loss=0.121498 lr=9.903199e-05 lr_backbone=9.903199e-05 elapsed=02:21:34 avg_step=1.55s recent_step=1.39s eta=19:10:06 +2026-09-20 08:59:51 | INFO | train rank=0 | step=5490/50000 progress=11.0% loss=0.118899 lr=9.902551e-05 lr_backbone=9.902551e-05 elapsed=02:21:51 avg_step=1.55s recent_step=1.70s eta=19:10:03 +2026-09-20 09:00:06 | INFO | train rank=0 | step=5500/50000 progress=11.0% loss=0.116649 lr=9.901900e-05 lr_backbone=9.901900e-05 elapsed=02:22:06 avg_step=1.55s recent_step=1.52s eta=19:09:45 +2026-09-20 09:00:19 | INFO | train rank=0 | step=5510/50000 progress=11.0% loss=0.122706 lr=9.901247e-05 lr_backbone=9.901247e-05 elapsed=02:22:18 avg_step=1.55s recent_step=1.26s eta=19:09:06 +2026-09-20 09:00:41 | INFO | train rank=0 | step=5520/50000 progress=11.0% loss=0.099676 lr=9.900592e-05 lr_backbone=9.900592e-05 elapsed=02:22:40 avg_step=1.55s recent_step=2.18s eta=19:09:42 +2026-09-20 09:00:52 | INFO | train rank=0 | step=5530/50000 progress=11.1% loss=0.128370 lr=9.899935e-05 lr_backbone=9.899935e-05 elapsed=02:22:52 avg_step=1.55s recent_step=1.15s eta=19:08:54 +2026-09-20 09:01:07 | INFO | train rank=0 | step=5540/50000 progress=11.1% loss=0.114874 lr=9.899275e-05 lr_backbone=9.899275e-05 elapsed=02:23:07 avg_step=1.55s recent_step=1.50s eta=19:08:35 +2026-09-20 09:01:26 | INFO | train rank=0 | step=5550/50000 progress=11.1% loss=0.104297 lr=9.898614e-05 lr_backbone=9.898614e-05 elapsed=02:23:26 avg_step=1.55s recent_step=1.89s eta=19:08:47 +2026-09-20 09:01:38 | INFO | train rank=0 | step=5560/50000 progress=11.1% loss=0.109391 lr=9.897950e-05 lr_backbone=9.897950e-05 elapsed=02:23:38 avg_step=1.55s recent_step=1.19s eta=19:08:02 +2026-09-20 09:01:57 | INFO | train rank=0 | step=5570/50000 progress=11.1% loss=0.124858 lr=9.897284e-05 lr_backbone=9.897284e-05 elapsed=02:23:56 avg_step=1.55s recent_step=1.85s eta=19:08:11 +2026-09-20 09:02:08 | INFO | train rank=0 | step=5580/50000 progress=11.2% loss=0.110833 lr=9.896616e-05 lr_backbone=9.896616e-05 elapsed=02:24:08 avg_step=1.55s recent_step=1.15s eta=19:07:24 +2026-09-20 09:02:29 | INFO | train rank=0 | step=5590/50000 progress=11.2% loss=0.095788 lr=9.895946e-05 lr_backbone=9.895946e-05 elapsed=02:24:29 avg_step=1.55s recent_step=2.13s eta=19:07:54 +2026-09-20 09:02:41 | INFO | train rank=0 | step=5600/50000 progress=11.2% loss=0.138974 lr=9.895274e-05 lr_backbone=9.895274e-05 elapsed=02:24:41 avg_step=1.55s recent_step=1.18s eta=19:07:09 +2026-09-20 09:02:53 | INFO | train rank=0 | step=5610/50000 progress=11.2% loss=0.144783 lr=9.894600e-05 lr_backbone=9.894600e-05 elapsed=02:24:53 avg_step=1.55s recent_step=1.19s eta=19:06:25 +2026-09-20 09:03:10 | INFO | train rank=0 | step=5620/50000 progress=11.2% loss=0.103345 lr=9.893923e-05 lr_backbone=9.893923e-05 elapsed=02:25:10 avg_step=1.55s recent_step=1.74s eta=19:06:25 +2026-09-20 09:03:29 | INFO | train rank=0 | step=5630/50000 progress=11.3% loss=0.107525 lr=9.893245e-05 lr_backbone=9.893245e-05 elapsed=02:25:28 avg_step=1.55s recent_step=1.83s eta=19:06:32 +2026-09-20 09:03:40 | INFO | train rank=0 | step=5640/50000 progress=11.3% loss=0.123470 lr=9.892564e-05 lr_backbone=9.892564e-05 elapsed=02:25:40 avg_step=1.55s recent_step=1.15s eta=19:05:44 +2026-09-20 09:03:57 | INFO | train rank=0 | step=5650/50000 progress=11.3% loss=0.096556 lr=9.891881e-05 lr_backbone=9.891881e-05 elapsed=02:25:57 avg_step=1.55s recent_step=1.68s eta=19:05:39 +2026-09-20 09:04:17 | INFO | train rank=0 | step=5660/50000 progress=11.3% loss=0.105875 lr=9.891196e-05 lr_backbone=9.891196e-05 elapsed=02:26:17 avg_step=1.55s recent_step=1.99s eta=19:05:58 +2026-09-20 09:04:30 | INFO | train rank=0 | step=5670/50000 progress=11.3% loss=0.137353 lr=9.890509e-05 lr_backbone=9.890509e-05 elapsed=02:26:29 avg_step=1.55s recent_step=1.30s eta=19:05:23 +2026-09-20 09:04:43 | INFO | train rank=0 | step=5680/50000 progress=11.4% loss=0.117630 lr=9.889819e-05 lr_backbone=9.889819e-05 elapsed=02:26:43 avg_step=1.55s recent_step=1.34s eta=19:04:50 +2026-09-20 09:05:01 | INFO | train rank=0 | step=5690/50000 progress=11.4% loss=0.120176 lr=9.889128e-05 lr_backbone=9.889128e-05 elapsed=02:27:00 avg_step=1.55s recent_step=1.75s eta=19:04:51 +2026-09-20 09:05:14 | INFO | train rank=0 | step=5700/50000 progress=11.4% loss=0.157857 lr=9.888434e-05 lr_backbone=9.888434e-05 elapsed=02:27:14 avg_step=1.55s recent_step=1.37s eta=19:04:21 +2026-09-20 09:05:35 | INFO | train rank=0 | step=5710/50000 progress=11.4% loss=0.122796 lr=9.887739e-05 lr_backbone=9.887739e-05 elapsed=02:27:35 avg_step=1.55s recent_step=2.04s eta=19:04:44 +2026-09-20 09:05:52 | INFO | train rank=0 | step=5720/50000 progress=11.4% loss=0.121561 lr=9.887041e-05 lr_backbone=9.887041e-05 elapsed=02:27:52 avg_step=1.55s recent_step=1.72s eta=19:04:41 +2026-09-20 09:06:04 | INFO | train rank=0 | step=5730/50000 progress=11.5% loss=0.116768 lr=9.886341e-05 lr_backbone=9.886341e-05 elapsed=02:28:04 avg_step=1.55s recent_step=1.21s eta=19:04:00 +2026-09-20 09:06:17 | INFO | train rank=0 | step=5740/50000 progress=11.5% loss=0.115639 lr=9.885638e-05 lr_backbone=9.885638e-05 elapsed=02:28:17 avg_step=1.55s recent_step=1.30s eta=19:03:25 +2026-09-20 09:06:39 | INFO | train rank=0 | step=5750/50000 progress=11.5% loss=0.104874 lr=9.884934e-05 lr_backbone=9.884934e-05 elapsed=02:28:39 avg_step=1.55s recent_step=2.18s eta=19:03:58 +2026-09-20 09:06:51 | INFO | train rank=0 | step=5760/50000 progress=11.5% loss=0.113484 lr=9.884228e-05 lr_backbone=9.884228e-05 elapsed=02:28:50 avg_step=1.55s recent_step=1.18s eta=19:03:14 +2026-09-20 09:07:03 | INFO | train rank=0 | step=5770/50000 progress=11.5% loss=0.109629 lr=9.883519e-05 lr_backbone=9.883519e-05 elapsed=02:29:03 avg_step=1.55s recent_step=1.22s eta=19:02:33 +2026-09-20 09:07:15 | INFO | train rank=0 | step=5780/50000 progress=11.6% loss=0.129076 lr=9.882808e-05 lr_backbone=9.882808e-05 elapsed=02:29:15 avg_step=1.55s recent_step=1.22s eta=19:01:52 +2026-09-20 09:07:34 | INFO | train rank=0 | step=5790/50000 progress=11.6% loss=0.125510 lr=9.882096e-05 lr_backbone=9.882096e-05 elapsed=02:29:34 avg_step=1.55s recent_step=1.88s eta=19:02:03 +2026-09-20 09:07:51 | INFO | train rank=0 | step=5800/50000 progress=11.6% loss=0.133310 lr=9.881381e-05 lr_backbone=9.881381e-05 elapsed=02:29:51 avg_step=1.55s recent_step=1.69s eta=19:01:58 +2026-09-20 09:08:05 | INFO | train rank=0 | step=5810/50000 progress=11.6% loss=0.120574 lr=9.880663e-05 lr_backbone=9.880663e-05 elapsed=02:30:05 avg_step=1.55s recent_step=1.39s eta=19:01:30 +2026-09-20 09:08:21 | INFO | train rank=0 | step=5820/50000 progress=11.6% loss=0.139771 lr=9.879944e-05 lr_backbone=9.879944e-05 elapsed=02:30:20 avg_step=1.55s recent_step=1.58s eta=19:01:17 +2026-09-20 09:08:41 | INFO | train rank=0 | step=5830/50000 progress=11.7% loss=0.090162 lr=9.879223e-05 lr_backbone=9.879223e-05 elapsed=02:30:40 avg_step=1.55s recent_step=1.98s eta=19:01:34 +2026-09-20 09:08:54 | INFO | train rank=0 | step=5840/50000 progress=11.7% loss=0.102034 lr=9.878499e-05 lr_backbone=9.878499e-05 elapsed=02:30:53 avg_step=1.55s recent_step=1.32s eta=19:01:01 +2026-09-20 09:09:13 | INFO | train rank=0 | step=5850/50000 progress=11.7% loss=0.140433 lr=9.877774e-05 lr_backbone=9.877774e-05 elapsed=02:31:13 avg_step=1.55s recent_step=1.92s eta=19:01:14 +2026-09-20 09:09:25 | INFO | train rank=0 | step=5860/50000 progress=11.7% loss=0.101951 lr=9.877046e-05 lr_backbone=9.877046e-05 elapsed=02:31:25 avg_step=1.55s recent_step=1.24s eta=19:00:35 +2026-09-20 09:09:40 | INFO | train rank=0 | step=5870/50000 progress=11.7% loss=0.132054 lr=9.876316e-05 lr_backbone=9.876316e-05 elapsed=02:31:40 avg_step=1.55s recent_step=1.50s eta=19:00:16 +2026-09-20 09:10:01 | INFO | train rank=0 | step=5880/50000 progress=11.8% loss=0.114118 lr=9.875584e-05 lr_backbone=9.875584e-05 elapsed=02:32:01 avg_step=1.55s recent_step=2.07s eta=19:00:40 +2026-09-20 09:10:13 | INFO | train rank=0 | step=5890/50000 progress=11.8% loss=0.126626 lr=9.874850e-05 lr_backbone=9.874850e-05 elapsed=02:32:13 avg_step=1.55s recent_step=1.20s eta=18:59:58 +2026-09-20 09:10:27 | INFO | train rank=0 | step=5900/50000 progress=11.8% loss=0.092555 lr=9.874113e-05 lr_backbone=9.874113e-05 elapsed=02:32:27 avg_step=1.55s recent_step=1.42s eta=18:59:33 +2026-09-20 09:10:45 | INFO | train rank=0 | step=5910/50000 progress=11.8% loss=0.123933 lr=9.873375e-05 lr_backbone=9.873375e-05 elapsed=02:32:45 avg_step=1.55s recent_step=1.77s eta=18:59:34 +2026-09-20 09:11:00 | INFO | train rank=0 | step=5920/50000 progress=11.8% loss=0.102789 lr=9.872634e-05 lr_backbone=9.872634e-05 elapsed=02:33:00 avg_step=1.55s recent_step=1.51s eta=18:59:15 +2026-09-20 09:11:14 | INFO | train rank=0 | step=5930/50000 progress=11.9% loss=0.141784 lr=9.871892e-05 lr_backbone=9.871892e-05 elapsed=02:33:13 avg_step=1.55s recent_step=1.35s eta=18:58:45 +2026-09-20 09:11:29 | INFO | train rank=0 | step=5940/50000 progress=11.9% loss=0.114497 lr=9.871147e-05 lr_backbone=9.871147e-05 elapsed=02:33:29 avg_step=1.55s recent_step=1.57s eta=18:58:31 +2026-09-20 09:11:45 | INFO | train rank=0 | step=5950/50000 progress=11.9% loss=0.118114 lr=9.870400e-05 lr_backbone=9.870400e-05 elapsed=02:33:44 avg_step=1.55s recent_step=1.54s eta=18:58:14 +2026-09-20 09:11:59 | INFO | train rank=0 | step=5960/50000 progress=11.9% loss=0.134865 lr=9.869651e-05 lr_backbone=9.869651e-05 elapsed=02:33:59 avg_step=1.55s recent_step=1.43s eta=18:57:50 +2026-09-20 09:12:11 | INFO | train rank=0 | step=5970/50000 progress=11.9% loss=0.110929 lr=9.868899e-05 lr_backbone=9.868899e-05 elapsed=02:34:10 avg_step=1.55s recent_step=1.17s eta=18:57:06 +2026-09-20 09:12:26 | INFO | train rank=0 | step=5980/50000 progress=12.0% loss=0.108209 lr=9.868146e-05 lr_backbone=9.868146e-05 elapsed=02:34:26 avg_step=1.55s recent_step=1.52s eta=18:56:49 +2026-09-20 09:12:51 | INFO | train rank=0 | step=5990/50000 progress=12.0% loss=0.094472 lr=9.867391e-05 lr_backbone=9.867391e-05 elapsed=02:34:50 avg_step=1.55s recent_step=2.48s eta=18:57:41 +2026-09-20 09:13:02 | INFO | train rank=0 | step=6000/50000 progress=12.0% loss=0.127616 lr=9.866633e-05 lr_backbone=9.866633e-05 elapsed=02:35:02 avg_step=1.55s recent_step=1.16s eta=18:56:57 +2026-09-20 09:13:14 | INFO | train rank=0 | step=6010/50000 progress=12.0% loss=0.144045 lr=9.865873e-05 lr_backbone=9.865873e-05 elapsed=02:35:14 avg_step=1.55s recent_step=1.22s eta=18:56:17 +2026-09-20 09:13:30 | INFO | train rank=0 | step=6020/50000 progress=12.0% loss=0.087079 lr=9.865111e-05 lr_backbone=9.865111e-05 elapsed=02:35:30 avg_step=1.55s recent_step=1.60s eta=18:56:06 +2026-09-20 09:13:50 | INFO | train rank=0 | step=6030/50000 progress=12.1% loss=0.139027 lr=9.864347e-05 lr_backbone=9.864347e-05 elapsed=02:35:50 avg_step=1.55s recent_step=1.95s eta=18:56:19 +2026-09-20 09:14:02 | INFO | train rank=0 | step=6040/50000 progress=12.1% loss=0.117157 lr=9.863581e-05 lr_backbone=9.863581e-05 elapsed=02:36:02 avg_step=1.55s recent_step=1.24s eta=18:55:42 +2026-09-20 09:14:21 | INFO | train rank=0 | step=6050/50000 progress=12.1% loss=0.141619 lr=9.862813e-05 lr_backbone=9.862813e-05 elapsed=02:36:20 avg_step=1.55s recent_step=1.81s eta=18:55:45 +2026-09-20 09:14:37 | INFO | train rank=0 | step=6060/50000 progress=12.1% loss=0.130095 lr=9.862043e-05 lr_backbone=9.862043e-05 elapsed=02:36:36 avg_step=1.55s recent_step=1.61s eta=18:55:34 +2026-09-20 09:14:53 | INFO | train rank=0 | step=6070/50000 progress=12.1% loss=0.101590 lr=9.861270e-05 lr_backbone=9.861270e-05 elapsed=02:36:53 avg_step=1.55s recent_step=1.64s eta=18:55:24 +2026-09-20 09:15:10 | INFO | train rank=0 | step=6080/50000 progress=12.2% loss=0.090879 lr=9.860495e-05 lr_backbone=9.860495e-05 elapsed=02:37:10 avg_step=1.55s recent_step=1.74s eta=18:55:23 +2026-09-20 09:15:22 | INFO | train rank=0 | step=6090/50000 progress=12.2% loss=0.143797 lr=9.859719e-05 lr_backbone=9.859719e-05 elapsed=02:37:22 avg_step=1.55s recent_step=1.20s eta=18:54:42 +2026-09-20 09:15:35 | INFO | train rank=0 | step=6100/50000 progress=12.2% loss=0.125803 lr=9.858940e-05 lr_backbone=9.858940e-05 elapsed=02:37:34 avg_step=1.55s recent_step=1.24s eta=18:54:04 +2026-09-20 09:15:54 | INFO | train rank=0 | step=6110/50000 progress=12.2% loss=0.103570 lr=9.858159e-05 lr_backbone=9.858159e-05 elapsed=02:37:54 avg_step=1.55s recent_step=1.96s eta=18:54:18 +2026-09-20 09:16:13 | INFO | train rank=0 | step=6120/50000 progress=12.2% loss=0.111214 lr=9.857376e-05 lr_backbone=9.857376e-05 elapsed=02:38:12 avg_step=1.55s recent_step=1.84s eta=18:54:23 +2026-09-20 09:16:27 | INFO | train rank=0 | step=6130/50000 progress=12.3% loss=0.129798 lr=9.856590e-05 lr_backbone=9.856590e-05 elapsed=02:38:27 avg_step=1.55s recent_step=1.47s eta=18:54:02 +2026-09-20 09:16:39 | INFO | train rank=0 | step=6140/50000 progress=12.3% loss=0.090433 lr=9.855803e-05 lr_backbone=9.855803e-05 elapsed=02:38:39 avg_step=1.55s recent_step=1.19s eta=18:53:21 +2026-09-20 09:16:52 | INFO | train rank=0 | step=6150/50000 progress=12.3% loss=0.147562 lr=9.855013e-05 lr_backbone=9.855013e-05 elapsed=02:38:51 avg_step=1.55s recent_step=1.24s eta=18:52:43 +2026-09-20 09:17:12 | INFO | train rank=0 | step=6160/50000 progress=12.3% loss=0.116300 lr=9.854222e-05 lr_backbone=9.854222e-05 elapsed=02:39:12 avg_step=1.55s recent_step=2.05s eta=18:53:03 +2026-09-20 09:17:24 | INFO | train rank=0 | step=6170/50000 progress=12.3% loss=0.122887 lr=9.853428e-05 lr_backbone=9.853428e-05 elapsed=02:39:24 avg_step=1.55s recent_step=1.18s eta=18:52:21 +2026-09-20 09:17:43 | INFO | train rank=0 | step=6180/50000 progress=12.4% loss=0.096650 lr=9.852632e-05 lr_backbone=9.852632e-05 elapsed=02:39:42 avg_step=1.55s recent_step=1.88s eta=18:52:29 +2026-09-20 09:17:58 | INFO | train rank=0 | step=6190/50000 progress=12.4% loss=0.171407 lr=9.851834e-05 lr_backbone=9.851834e-05 elapsed=02:39:58 avg_step=1.55s recent_step=1.56s eta=18:52:14 +2026-09-20 09:18:10 | INFO | train rank=0 | step=6200/50000 progress=12.4% loss=0.117926 lr=9.851034e-05 lr_backbone=9.851034e-05 elapsed=02:40:10 avg_step=1.55s recent_step=1.18s eta=18:51:32 +2026-09-20 09:18:30 | INFO | train rank=0 | step=6210/50000 progress=12.4% loss=0.114907 lr=9.850232e-05 lr_backbone=9.850232e-05 elapsed=02:40:29 avg_step=1.55s recent_step=1.93s eta=18:51:43 +2026-09-20 09:18:42 | INFO | train rank=0 | step=6220/50000 progress=12.4% loss=0.115363 lr=9.849427e-05 lr_backbone=9.849427e-05 elapsed=02:40:42 avg_step=1.55s recent_step=1.25s eta=18:51:07 +2026-09-20 09:18:58 | INFO | train rank=0 | step=6230/50000 progress=12.5% loss=0.151392 lr=9.848621e-05 lr_backbone=9.848621e-05 elapsed=02:40:57 avg_step=1.55s recent_step=1.58s eta=18:50:53 +2026-09-20 09:19:18 | INFO | train rank=0 | step=6240/50000 progress=12.5% loss=0.119903 lr=9.847812e-05 lr_backbone=9.847812e-05 elapsed=02:41:17 avg_step=1.55s recent_step=2.00s eta=18:51:09 +2026-09-20 09:19:32 | INFO | train rank=0 | step=6250/50000 progress=12.5% loss=0.142138 lr=9.847001e-05 lr_backbone=9.847001e-05 elapsed=02:41:32 avg_step=1.55s recent_step=1.45s eta=18:50:47 +2026-09-20 09:19:46 | INFO | train rank=0 | step=6260/50000 progress=12.5% loss=0.130812 lr=9.846188e-05 lr_backbone=9.846188e-05 elapsed=02:41:46 avg_step=1.55s recent_step=1.38s eta=18:50:19 +2026-09-20 09:20:05 | INFO | train rank=0 | step=6270/50000 progress=12.5% loss=0.117162 lr=9.845373e-05 lr_backbone=9.845373e-05 elapsed=02:42:05 avg_step=1.55s recent_step=1.89s eta=18:50:28 +2026-09-20 09:20:17 | INFO | train rank=0 | step=6280/50000 progress=12.6% loss=0.121996 lr=9.844556e-05 lr_backbone=9.844556e-05 elapsed=02:42:16 avg_step=1.55s recent_step=1.17s eta=18:49:45 +2026-09-20 09:20:33 | INFO | train rank=0 | step=6290/50000 progress=12.6% loss=0.126040 lr=9.843737e-05 lr_backbone=9.843737e-05 elapsed=02:42:33 avg_step=1.55s recent_step=1.65s eta=18:49:37 +2026-09-20 09:20:48 | INFO | train rank=0 | step=6300/50000 progress=12.6% loss=0.118893 lr=9.842916e-05 lr_backbone=9.842916e-05 elapsed=02:42:47 avg_step=1.55s recent_step=1.42s eta=18:49:13 +2026-09-20 09:21:02 | INFO | train rank=0 | step=6310/50000 progress=12.6% loss=0.105530 lr=9.842092e-05 lr_backbone=9.842092e-05 elapsed=02:43:01 avg_step=1.55s recent_step=1.43s eta=18:48:49 +2026-09-20 09:21:22 | INFO | train rank=0 | step=6320/50000 progress=12.6% loss=0.150792 lr=9.841267e-05 lr_backbone=9.841267e-05 elapsed=02:43:22 avg_step=1.55s recent_step=2.04s eta=18:49:07 +2026-09-20 09:21:34 | INFO | train rank=0 | step=6330/50000 progress=12.7% loss=0.102989 lr=9.840439e-05 lr_backbone=9.840439e-05 elapsed=02:43:33 avg_step=1.55s recent_step=1.17s eta=18:48:25 +2026-09-20 09:21:49 | INFO | train rank=0 | step=6340/50000 progress=12.7% loss=0.135356 lr=9.839609e-05 lr_backbone=9.839609e-05 elapsed=02:43:49 avg_step=1.55s recent_step=1.56s eta=18:48:10 +2026-09-20 09:22:06 | INFO | train rank=0 | step=6350/50000 progress=12.7% loss=0.152966 lr=9.838777e-05 lr_backbone=9.838777e-05 elapsed=02:44:05 avg_step=1.55s recent_step=1.64s eta=18:48:00 +2026-09-20 09:22:24 | INFO | train rank=0 | step=6360/50000 progress=12.7% loss=0.110252 lr=9.837943e-05 lr_backbone=9.837943e-05 elapsed=02:44:23 avg_step=1.55s recent_step=1.78s eta=18:48:00 +2026-09-20 09:22:36 | INFO | train rank=0 | step=6370/50000 progress=12.7% loss=0.123737 lr=9.837107e-05 lr_backbone=9.837107e-05 elapsed=02:44:35 avg_step=1.55s recent_step=1.20s eta=18:47:21 +2026-09-20 09:22:51 | INFO | train rank=0 | step=6380/50000 progress=12.8% loss=0.106477 lr=9.836269e-05 lr_backbone=9.836269e-05 elapsed=02:44:51 avg_step=1.55s recent_step=1.54s eta=18:47:05 +2026-09-20 09:23:04 | INFO | train rank=0 | step=6390/50000 progress=12.8% loss=0.105847 lr=9.835428e-05 lr_backbone=9.835428e-05 elapsed=02:45:03 avg_step=1.55s recent_step=1.26s eta=18:46:30 +2026-09-20 09:23:24 | INFO | train rank=0 | step=6400/50000 progress=12.8% loss=0.100530 lr=9.834586e-05 lr_backbone=9.834586e-05 elapsed=02:45:24 avg_step=1.55s recent_step=2.04s eta=18:46:48 +2026-09-20 09:23:36 | INFO | train rank=0 | step=6410/50000 progress=12.8% loss=0.111898 lr=9.833741e-05 lr_backbone=9.833741e-05 elapsed=02:45:36 avg_step=1.55s recent_step=1.23s eta=18:46:10 +2026-09-20 09:23:50 | INFO | train rank=0 | step=6420/50000 progress=12.8% loss=0.140319 lr=9.832895e-05 lr_backbone=9.832895e-05 elapsed=02:45:50 avg_step=1.55s recent_step=1.42s eta=18:45:46 +2026-09-20 09:24:08 | INFO | train rank=0 | step=6430/50000 progress=12.9% loss=0.101308 lr=9.832046e-05 lr_backbone=9.832046e-05 elapsed=02:46:08 avg_step=1.55s recent_step=1.78s eta=18:45:45 +2026-09-20 09:24:28 | INFO | train rank=0 | step=6440/50000 progress=12.9% loss=0.130484 lr=9.831195e-05 lr_backbone=9.831195e-05 elapsed=02:46:27 avg_step=1.55s recent_step=1.93s eta=18:45:56 +2026-09-20 09:24:40 | INFO | train rank=0 | step=6450/50000 progress=12.9% loss=0.142972 lr=9.830342e-05 lr_backbone=9.830342e-05 elapsed=02:46:39 avg_step=1.55s recent_step=1.22s eta=18:45:18 +2026-09-20 09:24:53 | INFO | train rank=0 | step=6460/50000 progress=12.9% loss=0.111774 lr=9.829486e-05 lr_backbone=9.829486e-05 elapsed=02:46:52 avg_step=1.55s recent_step=1.30s eta=18:44:45 +2026-09-20 09:25:05 | INFO | train rank=0 | step=6470/50000 progress=12.9% loss=0.134868 lr=9.828629e-05 lr_backbone=9.828629e-05 elapsed=02:47:05 avg_step=1.55s recent_step=1.28s eta=18:44:12 +2026-09-20 09:25:23 | INFO | train rank=0 | step=6480/50000 progress=13.0% loss=0.096595 lr=9.827770e-05 lr_backbone=9.827770e-05 elapsed=02:47:23 avg_step=1.55s recent_step=1.74s eta=18:44:09 +2026-09-20 09:25:39 | INFO | train rank=0 | step=6490/50000 progress=13.0% loss=0.114430 lr=9.826908e-05 lr_backbone=9.826908e-05 elapsed=02:47:38 avg_step=1.55s recent_step=1.56s eta=18:43:54 +2026-09-20 09:25:55 | INFO | train rank=0 | step=6500/50000 progress=13.0% loss=0.146092 lr=9.826045e-05 lr_backbone=9.826045e-05 elapsed=02:47:54 avg_step=1.55s recent_step=1.63s eta=18:43:44 +2026-09-20 09:26:11 | INFO | train rank=0 | step=6510/50000 progress=13.0% loss=0.115912 lr=9.825179e-05 lr_backbone=9.825179e-05 elapsed=02:48:11 avg_step=1.55s recent_step=1.61s eta=18:43:33 +2026-09-20 09:26:27 | INFO | train rank=0 | step=6520/50000 progress=13.0% loss=0.175637 lr=9.824311e-05 lr_backbone=9.824311e-05 elapsed=02:48:27 avg_step=1.55s recent_step=1.63s eta=18:43:23 +2026-09-20 09:26:44 | INFO | train rank=0 | step=6530/50000 progress=13.1% loss=0.131912 lr=9.823441e-05 lr_backbone=9.823441e-05 elapsed=02:48:44 avg_step=1.55s recent_step=1.70s eta=18:43:18 +2026-09-20 09:26:56 | INFO | train rank=0 | step=6540/50000 progress=13.1% loss=0.109781 lr=9.822569e-05 lr_backbone=9.822569e-05 elapsed=02:48:56 avg_step=1.55s recent_step=1.17s eta=18:42:37 +2026-09-20 09:27:10 | INFO | train rank=0 | step=6550/50000 progress=13.1% loss=0.112978 lr=9.821695e-05 lr_backbone=9.821695e-05 elapsed=02:49:09 avg_step=1.55s recent_step=1.36s eta=18:42:09 +2026-09-20 09:27:25 | INFO | train rank=0 | step=6560/50000 progress=13.1% loss=0.111448 lr=9.820818e-05 lr_backbone=9.820818e-05 elapsed=02:49:25 avg_step=1.55s recent_step=1.57s eta=18:41:55 +2026-09-20 09:27:41 | INFO | train rank=0 | step=6570/50000 progress=13.1% loss=0.135655 lr=9.819940e-05 lr_backbone=9.819940e-05 elapsed=02:49:41 avg_step=1.55s recent_step=1.57s eta=18:41:41 +2026-09-20 09:27:55 | INFO | train rank=0 | step=6580/50000 progress=13.2% loss=0.093809 lr=9.819059e-05 lr_backbone=9.819059e-05 elapsed=02:49:54 avg_step=1.55s recent_step=1.37s eta=18:41:13 +2026-09-20 09:28:15 | INFO | train rank=0 | step=6590/50000 progress=13.2% loss=0.137553 lr=9.818177e-05 lr_backbone=9.818177e-05 elapsed=02:50:15 avg_step=1.55s recent_step=2.04s eta=18:41:30 +2026-09-20 09:28:28 | INFO | train rank=0 | step=6600/50000 progress=13.2% loss=0.179344 lr=9.817292e-05 lr_backbone=9.817292e-05 elapsed=02:50:27 avg_step=1.55s recent_step=1.25s eta=18:40:55 +2026-09-20 09:28:45 | INFO | train rank=0 | step=6610/50000 progress=13.2% loss=0.123739 lr=9.816405e-05 lr_backbone=9.816405e-05 elapsed=02:50:44 avg_step=1.55s recent_step=1.68s eta=18:40:48 +2026-09-20 09:29:02 | INFO | train rank=0 | step=6620/50000 progress=13.2% loss=0.114049 lr=9.815516e-05 lr_backbone=9.815516e-05 elapsed=02:51:02 avg_step=1.55s recent_step=1.74s eta=18:40:45 +2026-09-20 09:29:16 | INFO | train rank=0 | step=6630/50000 progress=13.3% loss=0.113316 lr=9.814625e-05 lr_backbone=9.814625e-05 elapsed=02:51:16 avg_step=1.55s recent_step=1.41s eta=18:40:21 +2026-09-20 09:29:32 | INFO | train rank=0 | step=6640/50000 progress=13.3% loss=0.158826 lr=9.813732e-05 lr_backbone=9.813732e-05 elapsed=02:51:32 avg_step=1.55s recent_step=1.63s eta=18:40:10 +2026-09-20 09:29:46 | INFO | train rank=0 | step=6650/50000 progress=13.3% loss=0.098145 lr=9.812837e-05 lr_backbone=9.812837e-05 elapsed=02:51:45 avg_step=1.55s recent_step=1.35s eta=18:39:42 +2026-09-20 09:30:06 | INFO | train rank=0 | step=6660/50000 progress=13.3% loss=0.102431 lr=9.811939e-05 lr_backbone=9.811939e-05 elapsed=02:52:06 avg_step=1.55s recent_step=2.03s eta=18:39:58 +2026-09-20 09:30:18 | INFO | train rank=0 | step=6670/50000 progress=13.3% loss=0.114799 lr=9.811040e-05 lr_backbone=9.811040e-05 elapsed=02:52:17 avg_step=1.55s recent_step=1.16s eta=18:39:17 +2026-09-20 09:30:31 | INFO | train rank=0 | step=6680/50000 progress=13.4% loss=0.130517 lr=9.810138e-05 lr_backbone=9.810138e-05 elapsed=02:52:31 avg_step=1.55s recent_step=1.35s eta=18:38:49 +2026-09-20 09:30:48 | INFO | train rank=0 | step=6690/50000 progress=13.4% loss=0.119093 lr=9.809235e-05 lr_backbone=9.809235e-05 elapsed=02:52:47 avg_step=1.55s recent_step=1.64s eta=18:38:39 +2026-09-20 09:31:01 | INFO | train rank=0 | step=6700/50000 progress=13.4% loss=0.097464 lr=9.808329e-05 lr_backbone=9.808329e-05 elapsed=02:53:01 avg_step=1.55s recent_step=1.39s eta=18:38:13 +2026-09-20 09:31:21 | INFO | train rank=0 | step=6710/50000 progress=13.4% loss=0.086659 lr=9.807421e-05 lr_backbone=9.807421e-05 elapsed=02:53:21 avg_step=1.55s recent_step=1.99s eta=18:38:26 +2026-09-20 09:31:39 | INFO | train rank=0 | step=6720/50000 progress=13.4% loss=0.140498 lr=9.806511e-05 lr_backbone=9.806511e-05 elapsed=02:53:39 avg_step=1.55s recent_step=1.75s eta=18:38:23 +2026-09-20 09:31:56 | INFO | train rank=0 | step=6730/50000 progress=13.5% loss=0.106231 lr=9.805599e-05 lr_backbone=9.805599e-05 elapsed=02:53:56 avg_step=1.55s recent_step=1.74s eta=18:38:20 +2026-09-20 09:32:08 | INFO | train rank=0 | step=6740/50000 progress=13.5% loss=0.121086 lr=9.804685e-05 lr_backbone=9.804685e-05 elapsed=02:54:08 avg_step=1.55s recent_step=1.18s eta=18:37:40 +2026-09-20 09:32:20 | INFO | train rank=0 | step=6750/50000 progress=13.5% loss=0.104561 lr=9.803768e-05 lr_backbone=9.803768e-05 elapsed=02:54:20 avg_step=1.55s recent_step=1.18s eta=18:37:01 +2026-09-20 09:32:45 | INFO | train rank=0 | step=6760/50000 progress=13.5% loss=0.126250 lr=9.802850e-05 lr_backbone=9.802850e-05 elapsed=02:54:45 avg_step=1.55s recent_step=2.52s eta=18:37:47 +2026-09-20 09:32:57 | INFO | train rank=0 | step=6770/50000 progress=13.5% loss=0.126798 lr=9.801929e-05 lr_backbone=9.801929e-05 elapsed=02:54:57 avg_step=1.55s recent_step=1.22s eta=18:37:11 +2026-09-20 09:33:15 | INFO | train rank=0 | step=6780/50000 progress=13.6% loss=0.107592 lr=9.801007e-05 lr_backbone=9.801007e-05 elapsed=02:55:14 avg_step=1.55s recent_step=1.75s eta=18:37:08 +2026-09-20 09:33:27 | INFO | train rank=0 | step=6790/50000 progress=13.6% loss=0.105404 lr=9.800082e-05 lr_backbone=9.800082e-05 elapsed=02:55:26 avg_step=1.55s recent_step=1.21s eta=18:36:31 +2026-09-20 09:33:43 | INFO | train rank=0 | step=6800/50000 progress=13.6% loss=0.112188 lr=9.799155e-05 lr_backbone=9.799155e-05 elapsed=02:55:43 avg_step=1.55s recent_step=1.62s eta=18:36:20 +2026-09-20 09:33:59 | INFO | train rank=0 | step=6810/50000 progress=13.6% loss=0.104360 lr=9.798226e-05 lr_backbone=9.798226e-05 elapsed=02:55:59 avg_step=1.55s recent_step=1.63s eta=18:36:09 +2026-09-20 09:34:15 | INFO | train rank=0 | step=6820/50000 progress=13.6% loss=0.098559 lr=9.797295e-05 lr_backbone=9.797295e-05 elapsed=02:56:15 avg_step=1.55s recent_step=1.58s eta=18:35:56 +2026-09-20 09:34:27 | INFO | train rank=0 | step=6830/50000 progress=13.7% loss=0.100730 lr=9.796362e-05 lr_backbone=9.796362e-05 elapsed=02:56:27 avg_step=1.55s recent_step=1.19s eta=18:35:17 +2026-09-20 09:34:47 | INFO | train rank=0 | step=6840/50000 progress=13.7% loss=0.109096 lr=9.795427e-05 lr_backbone=9.795427e-05 elapsed=02:56:46 avg_step=1.55s recent_step=1.98s eta=18:35:29 +2026-09-20 09:35:00 | INFO | train rank=0 | step=6850/50000 progress=13.7% loss=0.121777 lr=9.794490e-05 lr_backbone=9.794490e-05 elapsed=02:56:59 avg_step=1.55s recent_step=1.29s eta=18:34:57 +2026-09-20 09:35:19 | INFO | train rank=0 | step=6860/50000 progress=13.7% loss=0.127543 lr=9.793550e-05 lr_backbone=9.793550e-05 elapsed=02:57:19 avg_step=1.55s recent_step=1.96s eta=18:35:07 +2026-09-20 09:35:35 | INFO | train rank=0 | step=6870/50000 progress=13.7% loss=0.108138 lr=9.792609e-05 lr_backbone=9.792609e-05 elapsed=02:57:35 avg_step=1.55s recent_step=1.56s eta=18:34:53 +2026-09-20 09:35:50 | INFO | train rank=0 | step=6880/50000 progress=13.8% loss=0.111706 lr=9.791665e-05 lr_backbone=9.791665e-05 elapsed=02:57:49 avg_step=1.55s recent_step=1.46s eta=18:34:32 +2026-09-20 09:36:07 | INFO | train rank=0 | step=6890/50000 progress=13.8% loss=0.117026 lr=9.790720e-05 lr_backbone=9.790720e-05 elapsed=02:58:07 avg_step=1.55s recent_step=1.76s eta=18:34:29 +2026-09-20 09:36:19 | INFO | train rank=0 | step=6900/50000 progress=13.8% loss=0.090766 lr=9.789772e-05 lr_backbone=9.789772e-05 elapsed=02:58:19 avg_step=1.55s recent_step=1.22s eta=18:33:53 +2026-09-20 09:36:32 | INFO | train rank=0 | step=6910/50000 progress=13.8% loss=0.107410 lr=9.788822e-05 lr_backbone=9.788822e-05 elapsed=02:58:32 avg_step=1.55s recent_step=1.27s eta=18:33:20 +2026-09-20 09:36:52 | INFO | train rank=0 | step=6920/50000 progress=13.8% loss=0.150519 lr=9.787870e-05 lr_backbone=9.787870e-05 elapsed=02:58:52 avg_step=1.55s recent_step=2.02s eta=18:33:34 +2026-09-20 09:37:09 | INFO | train rank=0 | step=6930/50000 progress=13.9% loss=0.142873 lr=9.786916e-05 lr_backbone=9.786916e-05 elapsed=02:59:09 avg_step=1.55s recent_step=1.70s eta=18:33:28 +2026-09-20 09:37:24 | INFO | train rank=0 | step=6940/50000 progress=13.9% loss=0.110538 lr=9.785960e-05 lr_backbone=9.785960e-05 elapsed=02:59:23 avg_step=1.55s recent_step=1.42s eta=18:33:04 +2026-09-20 09:37:37 | INFO | train rank=0 | step=6950/50000 progress=13.9% loss=0.122216 lr=9.785002e-05 lr_backbone=9.785002e-05 elapsed=02:59:37 avg_step=1.55s recent_step=1.36s eta=18:32:36 +2026-09-20 09:37:55 | INFO | train rank=0 | step=6960/50000 progress=13.9% loss=0.103462 lr=9.784041e-05 lr_backbone=9.784041e-05 elapsed=02:59:55 avg_step=1.55s recent_step=1.84s eta=18:32:38 +2026-09-20 09:38:08 | INFO | train rank=0 | step=6970/50000 progress=13.9% loss=0.109341 lr=9.783079e-05 lr_backbone=9.783079e-05 elapsed=03:00:07 avg_step=1.55s recent_step=1.23s eta=18:32:03 +2026-09-20 09:38:22 | INFO | train rank=0 | step=6980/50000 progress=14.0% loss=0.110254 lr=9.782114e-05 lr_backbone=9.782114e-05 elapsed=03:00:21 avg_step=1.55s recent_step=1.39s eta=18:31:38 +2026-09-20 09:38:40 | INFO | train rank=0 | step=6990/50000 progress=14.0% loss=0.098991 lr=9.781148e-05 lr_backbone=9.781148e-05 elapsed=03:00:39 avg_step=1.55s recent_step=1.79s eta=18:31:37 +2026-09-20 09:39:00 | INFO | train rank=0 | step=7000/50000 progress=14.0% loss=0.122823 lr=9.780179e-05 lr_backbone=9.780179e-05 elapsed=03:00:59 avg_step=1.55s recent_step=2.00s eta=18:31:49 +2026-09-20 09:39:12 | INFO | train rank=0 | step=7010/50000 progress=14.0% loss=0.090252 lr=9.779208e-05 lr_backbone=9.779208e-05 elapsed=03:01:11 avg_step=1.55s recent_step=1.21s eta=18:31:12 +2026-09-20 09:39:29 | INFO | train rank=0 | step=7020/50000 progress=14.0% loss=0.113991 lr=9.778235e-05 lr_backbone=9.778235e-05 elapsed=03:01:29 avg_step=1.55s recent_step=1.77s eta=18:31:10 +2026-09-20 09:39:46 | INFO | train rank=0 | step=7030/50000 progress=14.1% loss=0.102157 lr=9.777260e-05 lr_backbone=9.777260e-05 elapsed=03:01:46 avg_step=1.55s recent_step=1.68s eta=18:31:03 +2026-09-20 09:39:58 | INFO | train rank=0 | step=7040/50000 progress=14.1% loss=0.145708 lr=9.776283e-05 lr_backbone=9.776283e-05 elapsed=03:01:57 avg_step=1.55s recent_step=1.17s eta=18:30:24 +2026-09-20 09:40:14 | INFO | train rank=0 | step=7050/50000 progress=14.1% loss=0.127587 lr=9.775304e-05 lr_backbone=9.775304e-05 elapsed=03:02:13 avg_step=1.55s recent_step=1.57s eta=18:30:10 +2026-09-20 09:40:25 | INFO | train rank=0 | step=7060/50000 progress=14.1% loss=0.123694 lr=9.774323e-05 lr_backbone=9.774323e-05 elapsed=03:02:25 avg_step=1.55s recent_step=1.19s eta=18:29:32 +2026-09-20 09:40:38 | INFO | train rank=0 | step=7070/50000 progress=14.1% loss=0.096978 lr=9.773339e-05 lr_backbone=9.773339e-05 elapsed=03:02:38 avg_step=1.55s recent_step=1.28s eta=18:29:00 +2026-09-20 09:41:03 | INFO | train rank=0 | step=7080/50000 progress=14.2% loss=0.135351 lr=9.772354e-05 lr_backbone=9.772354e-05 elapsed=03:03:03 avg_step=1.55s recent_step=2.47s eta=18:29:41 +2026-09-20 09:41:17 | INFO | train rank=0 | step=7090/50000 progress=14.2% loss=0.135883 lr=9.771366e-05 lr_backbone=9.771366e-05 elapsed=03:03:16 avg_step=1.55s recent_step=1.37s eta=18:29:14 +2026-09-20 09:41:30 | INFO | train rank=0 | step=7100/50000 progress=14.2% loss=0.147747 lr=9.770377e-05 lr_backbone=9.770377e-05 elapsed=03:03:29 avg_step=1.55s recent_step=1.32s eta=18:28:45 +2026-09-20 09:41:49 | INFO | train rank=0 | step=7110/50000 progress=14.2% loss=0.110491 lr=9.769385e-05 lr_backbone=9.769385e-05 elapsed=03:03:49 avg_step=1.55s recent_step=1.96s eta=18:28:54 +2026-09-20 09:42:01 | INFO | train rank=0 | step=7120/50000 progress=14.2% loss=0.120172 lr=9.768391e-05 lr_backbone=9.768391e-05 elapsed=03:04:00 avg_step=1.55s recent_step=1.14s eta=18:28:13 +2026-09-20 09:42:15 | INFO | train rank=0 | step=7130/50000 progress=14.3% loss=0.119019 lr=9.767395e-05 lr_backbone=9.767395e-05 elapsed=03:04:14 avg_step=1.55s recent_step=1.38s eta=18:27:48 +2026-09-20 09:42:27 | INFO | train rank=0 | step=7140/50000 progress=14.3% loss=0.120120 lr=9.766398e-05 lr_backbone=9.766398e-05 elapsed=03:04:26 avg_step=1.55s recent_step=1.21s eta=18:27:12 +2026-09-20 09:42:40 | INFO | train rank=0 | step=7150/50000 progress=14.3% loss=0.110630 lr=9.765398e-05 lr_backbone=9.765398e-05 elapsed=03:04:39 avg_step=1.55s recent_step=1.31s eta=18:26:42 +2026-09-20 09:43:00 | INFO | train rank=0 | step=7160/50000 progress=14.3% loss=0.117769 lr=9.764395e-05 lr_backbone=9.764395e-05 elapsed=03:05:00 avg_step=1.55s recent_step=2.05s eta=18:26:56 +2026-09-20 09:43:18 | INFO | train rank=0 | step=7170/50000 progress=14.3% loss=0.105267 lr=9.763391e-05 lr_backbone=9.763391e-05 elapsed=03:05:18 avg_step=1.55s recent_step=1.80s eta=18:26:56 +2026-09-20 09:43:31 | INFO | train rank=0 | step=7180/50000 progress=14.4% loss=0.104067 lr=9.762385e-05 lr_backbone=9.762385e-05 elapsed=03:05:31 avg_step=1.55s recent_step=1.26s eta=18:26:23 +2026-09-20 09:43:49 | INFO | train rank=0 | step=7190/50000 progress=14.4% loss=0.120062 lr=9.761377e-05 lr_backbone=9.761377e-05 elapsed=03:05:49 avg_step=1.55s recent_step=1.84s eta=18:26:24 +2026-09-20 09:44:06 | INFO | train rank=0 | step=7200/50000 progress=14.4% loss=0.121665 lr=9.760366e-05 lr_backbone=9.760366e-05 elapsed=03:06:06 avg_step=1.55s recent_step=1.71s eta=18:26:18 +2026-09-20 09:44:20 | INFO | train rank=0 | step=7210/50000 progress=14.4% loss=0.129948 lr=9.759354e-05 lr_backbone=9.759354e-05 elapsed=03:06:20 avg_step=1.55s recent_step=1.38s eta=18:25:52 +2026-09-20 09:44:32 | INFO | train rank=0 | step=7220/50000 progress=14.4% loss=0.127560 lr=9.758339e-05 lr_backbone=9.758339e-05 elapsed=03:06:32 avg_step=1.55s recent_step=1.21s eta=18:25:16 +2026-09-20 09:44:50 | INFO | train rank=0 | step=7230/50000 progress=14.5% loss=0.131649 lr=9.757322e-05 lr_backbone=9.757322e-05 elapsed=03:06:50 avg_step=1.55s recent_step=1.80s eta=18:25:16 +2026-09-20 09:45:08 | INFO | train rank=0 | step=7240/50000 progress=14.5% loss=0.131775 lr=9.756303e-05 lr_backbone=9.756303e-05 elapsed=03:07:07 avg_step=1.55s recent_step=1.75s eta=18:25:12 +2026-09-20 09:45:20 | INFO | train rank=0 | step=7250/50000 progress=14.5% loss=0.096375 lr=9.755283e-05 lr_backbone=9.755283e-05 elapsed=03:07:19 avg_step=1.55s recent_step=1.20s eta=18:24:36 +2026-09-20 09:45:31 | INFO | train rank=0 | step=7260/50000 progress=14.5% loss=0.110259 lr=9.754260e-05 lr_backbone=9.754260e-05 elapsed=03:07:31 avg_step=1.55s recent_step=1.14s eta=18:23:57 +2026-09-20 09:45:52 | INFO | train rank=0 | step=7270/50000 progress=14.5% loss=0.112569 lr=9.753235e-05 lr_backbone=9.753235e-05 elapsed=03:07:51 avg_step=1.55s recent_step=2.06s eta=18:24:11 +2026-09-20 09:46:08 | INFO | train rank=0 | step=7280/50000 progress=14.6% loss=0.198025 lr=9.752208e-05 lr_backbone=9.752208e-05 elapsed=03:08:08 avg_step=1.55s recent_step=1.64s eta=18:24:01 +2026-09-20 09:46:22 | INFO | train rank=0 | step=7290/50000 progress=14.6% loss=0.141329 lr=9.751178e-05 lr_backbone=9.751178e-05 elapsed=03:08:22 avg_step=1.55s recent_step=1.42s eta=18:23:38 +2026-09-20 09:46:38 | INFO | train rank=0 | step=7300/50000 progress=14.6% loss=0.089589 lr=9.750147e-05 lr_backbone=9.750147e-05 elapsed=03:08:38 avg_step=1.55s recent_step=1.60s eta=18:23:25 +2026-09-20 09:46:56 | INFO | train rank=0 | step=7310/50000 progress=14.6% loss=0.125218 lr=9.749114e-05 lr_backbone=9.749114e-05 elapsed=03:08:56 avg_step=1.55s recent_step=1.78s eta=18:23:23 +2026-09-20 09:47:12 | INFO | train rank=0 | step=7320/50000 progress=14.6% loss=0.082038 lr=9.748078e-05 lr_backbone=9.748078e-05 elapsed=03:09:12 avg_step=1.55s recent_step=1.58s eta=18:23:09 +2026-09-20 09:47:26 | INFO | train rank=0 | step=7330/50000 progress=14.7% loss=0.115969 lr=9.747041e-05 lr_backbone=9.747041e-05 elapsed=03:09:26 avg_step=1.55s recent_step=1.40s eta=18:22:45 +2026-09-20 09:47:38 | INFO | train rank=0 | step=7340/50000 progress=14.7% loss=0.117547 lr=9.746001e-05 lr_backbone=9.746001e-05 elapsed=03:09:37 avg_step=1.55s recent_step=1.18s eta=18:22:08 +2026-09-20 09:47:53 | INFO | train rank=0 | step=7350/50000 progress=14.7% loss=0.107025 lr=9.744960e-05 lr_backbone=9.744960e-05 elapsed=03:09:53 avg_step=1.55s recent_step=1.54s eta=18:21:52 +2026-09-20 09:48:11 | INFO | train rank=0 | step=7360/50000 progress=14.7% loss=0.113438 lr=9.743916e-05 lr_backbone=9.743916e-05 elapsed=03:10:11 avg_step=1.55s recent_step=1.78s eta=18:21:50 +2026-09-20 09:48:28 | INFO | train rank=0 | step=7370/50000 progress=14.7% loss=0.177903 lr=9.742870e-05 lr_backbone=9.742870e-05 elapsed=03:10:28 avg_step=1.55s recent_step=1.73s eta=18:21:45 +2026-09-20 09:48:44 | INFO | train rank=0 | step=7380/50000 progress=14.8% loss=0.111457 lr=9.741822e-05 lr_backbone=9.741822e-05 elapsed=03:10:43 avg_step=1.55s recent_step=1.52s eta=18:21:27 +2026-09-20 09:48:59 | INFO | train rank=0 | step=7390/50000 progress=14.8% loss=0.122008 lr=9.740772e-05 lr_backbone=9.740772e-05 elapsed=03:10:58 avg_step=1.55s recent_step=1.51s eta=18:21:09 +2026-09-20 09:49:17 | INFO | train rank=0 | step=7400/50000 progress=14.8% loss=0.147585 lr=9.739720e-05 lr_backbone=9.739720e-05 elapsed=03:11:17 avg_step=1.55s recent_step=1.86s eta=18:21:11 +2026-09-20 09:49:32 | INFO | train rank=0 | step=7410/50000 progress=14.8% loss=0.144962 lr=9.738666e-05 lr_backbone=9.738666e-05 elapsed=03:11:31 avg_step=1.55s recent_step=1.45s eta=18:20:50 +2026-09-20 09:49:50 | INFO | train rank=0 | step=7420/50000 progress=14.8% loss=0.134717 lr=9.737610e-05 lr_backbone=9.737610e-05 elapsed=03:11:49 avg_step=1.55s recent_step=1.80s eta=18:20:49 +2026-09-20 09:50:04 | INFO | train rank=0 | step=7430/50000 progress=14.9% loss=0.135251 lr=9.736552e-05 lr_backbone=9.736552e-05 elapsed=03:12:03 avg_step=1.55s recent_step=1.40s eta=18:20:25 +2026-09-20 09:50:18 | INFO | train rank=0 | step=7440/50000 progress=14.9% loss=0.099604 lr=9.735492e-05 lr_backbone=9.735492e-05 elapsed=03:12:17 avg_step=1.55s recent_step=1.41s eta=18:20:01 +2026-09-20 09:50:34 | INFO | train rank=0 | step=7450/50000 progress=14.9% loss=0.115386 lr=9.734429e-05 lr_backbone=9.734429e-05 elapsed=03:12:33 avg_step=1.55s recent_step=1.59s eta=18:19:48 +2026-09-20 09:50:51 | INFO | train rank=0 | step=7460/50000 progress=14.9% loss=0.164884 lr=9.733365e-05 lr_backbone=9.733365e-05 elapsed=03:12:51 avg_step=1.55s recent_step=1.74s eta=18:19:43 +2026-09-20 09:51:03 | INFO | train rank=0 | step=7470/50000 progress=14.9% loss=0.111328 lr=9.732298e-05 lr_backbone=9.732298e-05 elapsed=03:13:03 avg_step=1.55s recent_step=1.22s eta=18:19:08 +2026-09-20 09:51:22 | INFO | train rank=0 | step=7480/50000 progress=15.0% loss=0.102703 lr=9.731230e-05 lr_backbone=9.731230e-05 elapsed=03:13:22 avg_step=1.55s recent_step=1.92s eta=18:19:14 +2026-09-20 09:51:37 | INFO | train rank=0 | step=7490/50000 progress=15.0% loss=0.117630 lr=9.730159e-05 lr_backbone=9.730159e-05 elapsed=03:13:37 avg_step=1.55s recent_step=1.49s eta=18:18:55 +2026-09-20 09:51:53 | INFO | train rank=0 | step=7500/50000 progress=15.0% loss=0.094161 lr=9.729086e-05 lr_backbone=9.729086e-05 elapsed=03:13:52 avg_step=1.55s recent_step=1.56s eta=18:18:40 +2026-09-20 09:52:04 | INFO | train rank=0 | step=7510/50000 progress=15.0% loss=0.132880 lr=9.728011e-05 lr_backbone=9.728011e-05 elapsed=03:14:04 avg_step=1.55s recent_step=1.16s eta=18:18:02 +2026-09-20 09:52:24 | INFO | train rank=0 | step=7520/50000 progress=15.0% loss=0.114852 lr=9.726935e-05 lr_backbone=9.726935e-05 elapsed=03:14:23 avg_step=1.55s recent_step=1.94s eta=18:18:08 +2026-09-20 09:52:38 | INFO | train rank=0 | step=7530/50000 progress=15.1% loss=0.151045 lr=9.725856e-05 lr_backbone=9.725856e-05 elapsed=03:14:38 avg_step=1.55s recent_step=1.41s eta=18:17:45 +2026-09-20 09:52:54 | INFO | train rank=0 | step=7540/50000 progress=15.1% loss=0.120891 lr=9.724775e-05 lr_backbone=9.724775e-05 elapsed=03:14:53 avg_step=1.55s recent_step=1.58s eta=18:17:31 +2026-09-20 09:53:06 | INFO | train rank=0 | step=7550/50000 progress=15.1% loss=0.136373 lr=9.723692e-05 lr_backbone=9.723692e-05 elapsed=03:15:06 avg_step=1.55s recent_step=1.23s eta=18:16:57 +2026-09-20 09:53:20 | INFO | train rank=0 | step=7560/50000 progress=15.1% loss=0.097304 lr=9.722606e-05 lr_backbone=9.722606e-05 elapsed=03:15:20 avg_step=1.55s recent_step=1.44s eta=18:16:36 +2026-09-20 09:53:35 | INFO | train rank=0 | step=7570/50000 progress=15.1% loss=0.106951 lr=9.721519e-05 lr_backbone=9.721519e-05 elapsed=03:15:35 avg_step=1.55s recent_step=1.47s eta=18:16:16 +2026-09-20 09:53:56 | INFO | train rank=0 | step=7580/50000 progress=15.2% loss=0.087098 lr=9.720430e-05 lr_backbone=9.720430e-05 elapsed=03:15:55 avg_step=1.55s recent_step=2.04s eta=18:16:28 +2026-09-20 09:54:07 | INFO | train rank=0 | step=7590/50000 progress=15.2% loss=0.099476 lr=9.719339e-05 lr_backbone=9.719339e-05 elapsed=03:16:07 avg_step=1.55s recent_step=1.18s eta=18:15:51 +2026-09-20 09:54:25 | INFO | train rank=0 | step=7600/50000 progress=15.2% loss=0.082216 lr=9.718245e-05 lr_backbone=9.718245e-05 elapsed=03:16:25 avg_step=1.55s recent_step=1.79s eta=18:15:50 +2026-09-20 09:54:39 | INFO | train rank=0 | step=7610/50000 progress=15.2% loss=0.100279 lr=9.717150e-05 lr_backbone=9.717150e-05 elapsed=03:16:39 avg_step=1.55s recent_step=1.42s eta=18:15:27 +2026-09-20 09:55:00 | INFO | train rank=0 | step=7620/50000 progress=15.2% loss=0.119960 lr=9.716052e-05 lr_backbone=9.716052e-05 elapsed=03:16:59 avg_step=1.55s recent_step=2.02s eta=18:15:38 +2026-09-20 09:55:12 | INFO | train rank=0 | step=7630/50000 progress=15.3% loss=0.142792 lr=9.714953e-05 lr_backbone=9.714953e-05 elapsed=03:17:12 avg_step=1.55s recent_step=1.25s eta=18:15:05 +2026-09-20 09:55:27 | INFO | train rank=0 | step=7640/50000 progress=15.3% loss=0.128235 lr=9.713851e-05 lr_backbone=9.713851e-05 elapsed=03:17:26 avg_step=1.55s recent_step=1.46s eta=18:14:45 +2026-09-20 09:55:44 | INFO | train rank=0 | step=7650/50000 progress=15.3% loss=0.123023 lr=9.712747e-05 lr_backbone=9.712747e-05 elapsed=03:17:44 avg_step=1.55s recent_step=1.74s eta=18:14:40 +2026-09-20 09:55:57 | INFO | train rank=0 | step=7660/50000 progress=15.3% loss=0.123306 lr=9.711642e-05 lr_backbone=9.711642e-05 elapsed=03:17:57 avg_step=1.55s recent_step=1.33s eta=18:14:12 +2026-09-20 09:56:12 | INFO | train rank=0 | step=7670/50000 progress=15.3% loss=0.115775 lr=9.710534e-05 lr_backbone=9.710534e-05 elapsed=03:18:11 avg_step=1.55s recent_step=1.41s eta=18:13:48 +2026-09-20 09:56:32 | INFO | train rank=0 | step=7680/50000 progress=15.4% loss=0.104466 lr=9.709424e-05 lr_backbone=9.709424e-05 elapsed=03:18:31 avg_step=1.55s recent_step=2.01s eta=18:13:58 +2026-09-20 09:56:45 | INFO | train rank=0 | step=7690/50000 progress=15.4% loss=0.092537 lr=9.708312e-05 lr_backbone=9.708312e-05 elapsed=03:18:45 avg_step=1.55s recent_step=1.36s eta=18:13:32 +2026-09-20 09:56:58 | INFO | train rank=0 | step=7700/50000 progress=15.4% loss=0.100610 lr=9.707198e-05 lr_backbone=9.707198e-05 elapsed=03:18:58 avg_step=1.55s recent_step=1.30s eta=18:13:03 +2026-09-20 09:57:18 | INFO | train rank=0 | step=7710/50000 progress=15.4% loss=0.115286 lr=9.706082e-05 lr_backbone=9.706082e-05 elapsed=03:19:17 avg_step=1.55s recent_step=1.96s eta=18:13:09 +2026-09-20 09:57:30 | INFO | train rank=0 | step=7720/50000 progress=15.4% loss=0.084307 lr=9.704964e-05 lr_backbone=9.704964e-05 elapsed=03:19:30 avg_step=1.55s recent_step=1.23s eta=18:12:36 +2026-09-20 09:57:43 | INFO | train rank=0 | step=7730/50000 progress=15.5% loss=0.138599 lr=9.703843e-05 lr_backbone=9.703843e-05 elapsed=03:19:43 avg_step=1.55s recent_step=1.31s eta=18:12:07 +2026-09-20 09:58:06 | INFO | train rank=0 | step=7740/50000 progress=15.5% loss=0.110574 lr=9.702721e-05 lr_backbone=9.702721e-05 elapsed=03:20:06 avg_step=1.55s recent_step=2.31s eta=18:12:33 +2026-09-20 09:58:18 | INFO | train rank=0 | step=7750/50000 progress=15.5% loss=0.125342 lr=9.701597e-05 lr_backbone=9.701597e-05 elapsed=03:20:18 avg_step=1.55s recent_step=1.21s eta=18:11:59 +2026-09-20 09:58:32 | INFO | train rank=0 | step=7760/50000 progress=15.5% loss=0.129067 lr=9.700471e-05 lr_backbone=9.700471e-05 elapsed=03:20:31 avg_step=1.55s recent_step=1.33s eta=18:11:32 +2026-09-20 09:58:44 | INFO | train rank=0 | step=7770/50000 progress=15.5% loss=0.098030 lr=9.699342e-05 lr_backbone=9.699342e-05 elapsed=03:20:44 avg_step=1.55s recent_step=1.24s eta=18:11:00 +2026-09-20 09:58:56 | INFO | train rank=0 | step=7780/50000 progress=15.6% loss=0.140987 lr=9.698212e-05 lr_backbone=9.698212e-05 elapsed=03:20:56 avg_step=1.55s recent_step=1.21s eta=18:10:26 +2026-09-20 09:59:16 | INFO | train rank=0 | step=7790/50000 progress=15.6% loss=0.106777 lr=9.697079e-05 lr_backbone=9.697079e-05 elapsed=03:21:16 avg_step=1.55s recent_step=1.98s eta=18:10:34 +2026-09-20 09:59:28 | INFO | train rank=0 | step=7800/50000 progress=15.6% loss=0.093268 lr=9.695945e-05 lr_backbone=9.695945e-05 elapsed=03:21:28 avg_step=1.55s recent_step=1.24s eta=18:10:01 +2026-09-20 09:59:50 | INFO | train rank=0 | step=7810/50000 progress=15.6% loss=0.106533 lr=9.694808e-05 lr_backbone=9.694808e-05 elapsed=03:21:49 avg_step=1.55s recent_step=2.13s eta=18:10:18 +2026-09-20 10:00:01 | INFO | train rank=0 | step=7820/50000 progress=15.6% loss=0.121259 lr=9.693669e-05 lr_backbone=9.693669e-05 elapsed=03:22:01 avg_step=1.55s recent_step=1.15s eta=18:09:41 +2026-09-20 10:00:14 | INFO | train rank=0 | step=7830/50000 progress=15.7% loss=0.097567 lr=9.692529e-05 lr_backbone=9.692529e-05 elapsed=03:22:14 avg_step=1.55s recent_step=1.27s eta=18:09:10 +2026-09-20 10:00:33 | INFO | train rank=0 | step=7840/50000 progress=15.7% loss=0.108856 lr=9.691386e-05 lr_backbone=9.691386e-05 elapsed=03:22:33 avg_step=1.55s recent_step=1.95s eta=18:09:16 +2026-09-20 10:00:47 | INFO | train rank=0 | step=7850/50000 progress=15.7% loss=0.095520 lr=9.690241e-05 lr_backbone=9.690241e-05 elapsed=03:22:47 avg_step=1.55s recent_step=1.34s eta=18:08:50 +2026-09-20 10:01:06 | INFO | train rank=0 | step=7860/50000 progress=15.7% loss=0.094445 lr=9.689094e-05 lr_backbone=9.689094e-05 elapsed=03:23:05 avg_step=1.55s recent_step=1.88s eta=18:08:52 +2026-09-20 10:01:18 | INFO | train rank=0 | step=7870/50000 progress=15.7% loss=0.110446 lr=9.687945e-05 lr_backbone=9.687945e-05 elapsed=03:23:17 avg_step=1.55s recent_step=1.21s eta=18:08:18 +2026-09-20 10:01:39 | INFO | train rank=0 | step=7880/50000 progress=15.8% loss=0.158561 lr=9.686794e-05 lr_backbone=9.686794e-05 elapsed=03:23:39 avg_step=1.55s recent_step=2.17s eta=18:08:35 +2026-09-20 10:01:51 | INFO | train rank=0 | step=7890/50000 progress=15.8% loss=0.148152 lr=9.685641e-05 lr_backbone=9.685641e-05 elapsed=03:23:51 avg_step=1.55s recent_step=1.19s eta=18:08:01 +2026-09-20 10:02:04 | INFO | train rank=0 | step=7900/50000 progress=15.8% loss=0.127675 lr=9.684486e-05 lr_backbone=9.684486e-05 elapsed=03:24:03 avg_step=1.55s recent_step=1.24s eta=18:07:28 +2026-09-20 10:02:24 | INFO | train rank=0 | step=7910/50000 progress=15.8% loss=0.164815 lr=9.683329e-05 lr_backbone=9.683329e-05 elapsed=03:24:24 avg_step=1.55s recent_step=2.01s eta=18:07:38 +2026-09-20 10:02:43 | INFO | train rank=0 | step=7920/50000 progress=15.8% loss=0.102905 lr=9.682170e-05 lr_backbone=9.682170e-05 elapsed=03:24:42 avg_step=1.55s recent_step=1.89s eta=18:07:40 +2026-09-20 10:02:58 | INFO | train rank=0 | step=7930/50000 progress=15.9% loss=0.107683 lr=9.681008e-05 lr_backbone=9.681008e-05 elapsed=03:24:58 avg_step=1.55s recent_step=1.52s eta=18:07:23 +2026-09-20 10:03:10 | INFO | train rank=0 | step=7940/50000 progress=15.9% loss=0.103391 lr=9.679845e-05 lr_backbone=9.679845e-05 elapsed=03:25:09 avg_step=1.55s recent_step=1.17s eta=18:06:48 +2026-09-20 10:03:24 | INFO | train rank=0 | step=7950/50000 progress=15.9% loss=0.117954 lr=9.678680e-05 lr_backbone=9.678680e-05 elapsed=03:25:23 avg_step=1.55s recent_step=1.40s eta=18:06:24 +2026-09-20 10:03:42 | INFO | train rank=0 | step=7960/50000 progress=15.9% loss=0.111425 lr=9.677512e-05 lr_backbone=9.677512e-05 elapsed=03:25:42 avg_step=1.55s recent_step=1.82s eta=18:06:23 +2026-09-20 10:03:54 | INFO | train rank=0 | step=7970/50000 progress=15.9% loss=0.106656 lr=9.676343e-05 lr_backbone=9.676343e-05 elapsed=03:25:54 avg_step=1.55s recent_step=1.23s eta=18:05:50 +2026-09-20 10:04:07 | INFO | train rank=0 | step=7980/50000 progress=16.0% loss=0.140656 lr=9.675171e-05 lr_backbone=9.675171e-05 elapsed=03:26:06 avg_step=1.55s recent_step=1.24s eta=18:05:19 +2026-09-20 10:04:25 | INFO | train rank=0 | step=7990/50000 progress=16.0% loss=0.126218 lr=9.673998e-05 lr_backbone=9.673998e-05 elapsed=03:26:25 avg_step=1.55s recent_step=1.86s eta=18:05:20 +2026-09-20 10:04:41 | INFO | train rank=0 | step=8000/50000 progress=16.0% loss=0.121118 lr=9.672822e-05 lr_backbone=9.672822e-05 elapsed=03:26:41 avg_step=1.55s recent_step=1.60s eta=18:05:07 +2026-09-20 10:04:57 | INFO | train rank=0 | step=8010/50000 progress=16.0% loss=0.127974 lr=9.671645e-05 lr_backbone=9.671645e-05 elapsed=03:26:57 avg_step=1.55s recent_step=1.58s eta=18:04:53 +2026-09-20 10:05:18 | INFO | train rank=0 | step=8020/50000 progress=16.0% loss=0.102380 lr=9.670465e-05 lr_backbone=9.670465e-05 elapsed=03:27:18 avg_step=1.55s recent_step=2.11s eta=18:05:07 +2026-09-20 10:05:30 | INFO | train rank=0 | step=8030/50000 progress=16.1% loss=0.098862 lr=9.669283e-05 lr_backbone=9.669283e-05 elapsed=03:27:30 avg_step=1.55s recent_step=1.21s eta=18:04:34 +2026-09-20 10:05:44 | INFO | train rank=0 | step=8040/50000 progress=16.1% loss=0.098253 lr=9.668100e-05 lr_backbone=9.668100e-05 elapsed=03:27:43 avg_step=1.55s recent_step=1.34s eta=18:04:07 +2026-09-20 10:05:57 | INFO | train rank=0 | step=8050/50000 progress=16.1% loss=0.115829 lr=9.666914e-05 lr_backbone=9.666914e-05 elapsed=03:27:57 avg_step=1.55s recent_step=1.32s eta=18:03:40 +2026-09-20 10:06:14 | INFO | train rank=0 | step=8060/50000 progress=16.1% loss=0.119625 lr=9.665726e-05 lr_backbone=9.665726e-05 elapsed=03:28:14 avg_step=1.55s recent_step=1.73s eta=18:03:33 +2026-09-20 10:06:28 | INFO | train rank=0 | step=8070/50000 progress=16.1% loss=0.094491 lr=9.664536e-05 lr_backbone=9.664536e-05 elapsed=03:28:28 avg_step=1.55s recent_step=1.41s eta=18:03:10 +2026-09-20 10:06:46 | INFO | train rank=0 | step=8080/50000 progress=16.2% loss=0.148892 lr=9.663344e-05 lr_backbone=9.663344e-05 elapsed=03:28:46 avg_step=1.55s recent_step=1.81s eta=18:03:08 +2026-09-20 10:07:04 | INFO | train rank=0 | step=8090/50000 progress=16.2% loss=0.099878 lr=9.662150e-05 lr_backbone=9.662150e-05 elapsed=03:29:03 avg_step=1.55s recent_step=1.72s eta=18:03:02 +2026-09-20 10:07:19 | INFO | train rank=0 | step=8100/50000 progress=16.2% loss=0.113138 lr=9.660954e-05 lr_backbone=9.660954e-05 elapsed=03:29:19 avg_step=1.55s recent_step=1.56s eta=18:02:46 +2026-09-20 10:07:33 | INFO | train rank=0 | step=8110/50000 progress=16.2% loss=0.135403 lr=9.659756e-05 lr_backbone=9.659756e-05 elapsed=03:29:32 avg_step=1.55s recent_step=1.36s eta=18:02:21 +2026-09-20 10:07:52 | INFO | train rank=0 | step=8120/50000 progress=16.2% loss=0.115545 lr=9.658556e-05 lr_backbone=9.658556e-05 elapsed=03:29:52 avg_step=1.55s recent_step=1.91s eta=18:02:25 +2026-09-20 10:08:06 | INFO | train rank=0 | step=8130/50000 progress=16.3% loss=0.144106 lr=9.657354e-05 lr_backbone=9.657354e-05 elapsed=03:30:05 avg_step=1.55s recent_step=1.39s eta=18:02:01 +2026-09-20 10:08:24 | INFO | train rank=0 | step=8140/50000 progress=16.3% loss=0.102969 lr=9.656150e-05 lr_backbone=9.656150e-05 elapsed=03:30:24 avg_step=1.55s recent_step=1.87s eta=18:02:02 +2026-09-20 10:08:37 | INFO | train rank=0 | step=8150/50000 progress=16.3% loss=0.103435 lr=9.654944e-05 lr_backbone=9.654944e-05 elapsed=03:30:37 avg_step=1.55s recent_step=1.25s eta=18:01:31 +2026-09-20 10:08:49 | INFO | train rank=0 | step=8160/50000 progress=16.3% loss=0.122577 lr=9.653736e-05 lr_backbone=9.653736e-05 elapsed=03:30:49 avg_step=1.55s recent_step=1.19s eta=18:00:57 +2026-09-20 10:09:05 | INFO | train rank=0 | step=8170/50000 progress=16.3% loss=0.103925 lr=9.652525e-05 lr_backbone=9.652525e-05 elapsed=03:31:05 avg_step=1.55s recent_step=1.64s eta=18:00:46 +2026-09-20 10:09:23 | INFO | train rank=0 | step=8180/50000 progress=16.4% loss=0.091966 lr=9.651313e-05 lr_backbone=9.651313e-05 elapsed=03:31:23 avg_step=1.55s recent_step=1.77s eta=18:00:42 +2026-09-20 10:09:37 | INFO | train rank=0 | step=8190/50000 progress=16.4% loss=0.120073 lr=9.650099e-05 lr_backbone=9.650099e-05 elapsed=03:31:36 avg_step=1.55s recent_step=1.38s eta=18:00:18 +2026-09-20 10:09:59 | INFO | train rank=0 | step=8200/50000 progress=16.4% loss=0.096513 lr=9.648882e-05 lr_backbone=9.648882e-05 elapsed=03:31:58 avg_step=1.55s recent_step=2.18s eta=18:00:34 +2026-09-20 10:10:11 | INFO | train rank=0 | step=8210/50000 progress=16.4% loss=0.108066 lr=9.647664e-05 lr_backbone=9.647664e-05 elapsed=03:32:10 avg_step=1.55s recent_step=1.22s eta=18:00:02 +2026-09-20 10:10:23 | INFO | train rank=0 | step=8220/50000 progress=16.4% loss=0.117481 lr=9.646444e-05 lr_backbone=9.646444e-05 elapsed=03:32:22 avg_step=1.55s recent_step=1.17s eta=17:59:27 +2026-09-20 10:10:39 | INFO | train rank=0 | step=8230/50000 progress=16.5% loss=0.118907 lr=9.645221e-05 lr_backbone=9.645221e-05 elapsed=03:32:39 avg_step=1.55s recent_step=1.63s eta=17:59:16 +2026-09-20 10:11:00 | INFO | train rank=0 | step=8240/50000 progress=16.5% loss=0.123852 lr=9.643997e-05 lr_backbone=9.643997e-05 elapsed=03:33:00 avg_step=1.55s recent_step=2.14s eta=17:59:30 +2026-09-20 10:11:14 | INFO | train rank=0 | step=8250/50000 progress=16.5% loss=0.111271 lr=9.642770e-05 lr_backbone=9.642770e-05 elapsed=03:33:13 avg_step=1.55s recent_step=1.36s eta=17:59:05 +2026-09-20 10:11:32 | INFO | train rank=0 | step=8260/50000 progress=16.5% loss=0.150154 lr=9.641542e-05 lr_backbone=9.641542e-05 elapsed=03:33:31 avg_step=1.55s recent_step=1.77s eta=17:59:00 +2026-09-20 10:11:43 | INFO | train rank=0 | step=8270/50000 progress=16.5% loss=0.117007 lr=9.640311e-05 lr_backbone=9.640311e-05 elapsed=03:33:43 avg_step=1.55s recent_step=1.17s eta=17:58:25 +2026-09-20 10:12:00 | INFO | train rank=0 | step=8280/50000 progress=16.6% loss=0.132813 lr=9.639078e-05 lr_backbone=9.639078e-05 elapsed=03:34:00 avg_step=1.55s recent_step=1.68s eta=17:58:17 +2026-09-20 10:12:13 | INFO | train rank=0 | step=8290/50000 progress=16.6% loss=0.103410 lr=9.637844e-05 lr_backbone=9.637844e-05 elapsed=03:34:12 avg_step=1.55s recent_step=1.26s eta=17:57:47 +2026-09-20 10:12:28 | INFO | train rank=0 | step=8300/50000 progress=16.6% loss=0.132801 lr=9.636607e-05 lr_backbone=9.636607e-05 elapsed=03:34:28 avg_step=1.55s recent_step=1.52s eta=17:57:30 +2026-09-20 10:12:46 | INFO | train rank=0 | step=8310/50000 progress=16.6% loss=0.113909 lr=9.635368e-05 lr_backbone=9.635368e-05 elapsed=03:34:46 avg_step=1.55s recent_step=1.81s eta=17:57:27 +2026-09-20 10:13:02 | INFO | train rank=0 | step=8320/50000 progress=16.6% loss=0.097686 lr=9.634128e-05 lr_backbone=9.634128e-05 elapsed=03:35:01 avg_step=1.55s recent_step=1.57s eta=17:57:12 +2026-09-20 10:13:13 | INFO | train rank=0 | step=8330/50000 progress=16.7% loss=0.139604 lr=9.632885e-05 lr_backbone=9.632885e-05 elapsed=03:35:13 avg_step=1.55s recent_step=1.17s eta=17:56:38 +2026-09-20 10:13:31 | INFO | train rank=0 | step=8340/50000 progress=16.7% loss=0.101844 lr=9.631640e-05 lr_backbone=9.631640e-05 elapsed=03:35:30 avg_step=1.55s recent_step=1.75s eta=17:56:32 +2026-09-20 10:13:46 | INFO | train rank=0 | step=8350/50000 progress=16.7% loss=0.094941 lr=9.630393e-05 lr_backbone=9.630393e-05 elapsed=03:35:46 avg_step=1.55s recent_step=1.51s eta=17:56:15 +2026-09-20 10:14:02 | INFO | train rank=0 | step=8360/50000 progress=16.7% loss=0.094065 lr=9.629145e-05 lr_backbone=9.629145e-05 elapsed=03:36:01 avg_step=1.55s recent_step=1.58s eta=17:56:01 +2026-09-20 10:14:14 | INFO | train rank=0 | step=8370/50000 progress=16.7% loss=0.120774 lr=9.627894e-05 lr_backbone=9.627894e-05 elapsed=03:36:14 avg_step=1.55s recent_step=1.23s eta=17:55:29 +2026-09-20 10:14:26 | INFO | train rank=0 | step=8380/50000 progress=16.8% loss=0.112330 lr=9.626641e-05 lr_backbone=9.626641e-05 elapsed=03:36:25 avg_step=1.55s recent_step=1.16s eta=17:54:55 +2026-09-20 10:14:40 | INFO | train rank=0 | step=8390/50000 progress=16.8% loss=0.116659 lr=9.625386e-05 lr_backbone=9.625386e-05 elapsed=03:36:39 avg_step=1.55s recent_step=1.41s eta=17:54:32 +2026-09-20 10:15:00 | INFO | train rank=0 | step=8400/50000 progress=16.8% loss=0.104726 lr=9.624129e-05 lr_backbone=9.624129e-05 elapsed=03:37:00 avg_step=1.55s recent_step=2.07s eta=17:54:42 +2026-09-20 10:15:17 | INFO | train rank=0 | step=8410/50000 progress=16.8% loss=0.145522 lr=9.622870e-05 lr_backbone=9.622870e-05 elapsed=03:37:17 avg_step=1.55s recent_step=1.66s eta=17:54:32 +2026-09-20 10:15:36 | INFO | train rank=0 | step=8420/50000 progress=16.8% loss=0.100584 lr=9.621609e-05 lr_backbone=9.621609e-05 elapsed=03:37:35 avg_step=1.55s recent_step=1.86s eta=17:54:32 +2026-09-20 10:15:48 | INFO | train rank=0 | step=8430/50000 progress=16.9% loss=0.121165 lr=9.620346e-05 lr_backbone=9.620346e-05 elapsed=03:37:48 avg_step=1.55s recent_step=1.23s eta=17:54:00 +2026-09-20 10:16:01 | INFO | train rank=0 | step=8440/50000 progress=16.9% loss=0.134186 lr=9.619081e-05 lr_backbone=9.619081e-05 elapsed=03:38:01 avg_step=1.55s recent_step=1.33s eta=17:53:34 +2026-09-20 10:16:19 | INFO | train rank=0 | step=8450/50000 progress=16.9% loss=0.099788 lr=9.617814e-05 lr_backbone=9.617814e-05 elapsed=03:38:19 avg_step=1.55s recent_step=1.81s eta=17:53:31 +2026-09-20 10:16:37 | INFO | train rank=0 | step=8460/50000 progress=16.9% loss=0.141589 lr=9.616545e-05 lr_backbone=9.616545e-05 elapsed=03:38:37 avg_step=1.55s recent_step=1.82s eta=17:53:29 +2026-09-20 10:16:49 | INFO | train rank=0 | step=8470/50000 progress=16.9% loss=0.088711 lr=9.615274e-05 lr_backbone=9.615274e-05 elapsed=03:38:49 avg_step=1.55s recent_step=1.19s eta=17:52:56 +2026-09-20 10:17:02 | INFO | train rank=0 | step=8480/50000 progress=17.0% loss=0.110257 lr=9.614001e-05 lr_backbone=9.614001e-05 elapsed=03:39:01 avg_step=1.55s recent_step=1.24s eta=17:52:25 +2026-09-20 10:17:16 | INFO | train rank=0 | step=8490/50000 progress=17.0% loss=0.097910 lr=9.612726e-05 lr_backbone=9.612726e-05 elapsed=03:39:16 avg_step=1.55s recent_step=1.43s eta=17:52:04 +2026-09-20 10:17:32 | INFO | train rank=0 | step=8500/50000 progress=17.0% loss=0.139948 lr=9.611449e-05 lr_backbone=9.611449e-05 elapsed=03:39:32 avg_step=1.55s recent_step=1.59s eta=17:51:50 +2026-09-20 10:17:51 | INFO | train rank=0 | step=8510/50000 progress=17.0% loss=0.115903 lr=9.610170e-05 lr_backbone=9.610170e-05 elapsed=03:39:51 avg_step=1.55s recent_step=1.93s eta=17:51:53 +2026-09-20 10:18:06 | INFO | train rank=0 | step=8520/50000 progress=17.0% loss=0.111446 lr=9.608888e-05 lr_backbone=9.608888e-05 elapsed=03:40:05 avg_step=1.55s recent_step=1.43s eta=17:51:32 +2026-09-20 10:18:21 | INFO | train rank=0 | step=8530/50000 progress=17.1% loss=0.130433 lr=9.607605e-05 lr_backbone=9.607605e-05 elapsed=03:40:21 avg_step=1.55s recent_step=1.58s eta=17:51:18 +2026-09-20 10:18:38 | INFO | train rank=0 | step=8540/50000 progress=17.1% loss=0.120901 lr=9.606320e-05 lr_backbone=9.606320e-05 elapsed=03:40:38 avg_step=1.55s recent_step=1.68s eta=17:51:09 +2026-09-20 10:18:51 | INFO | train rank=0 | step=8550/50000 progress=17.1% loss=0.120133 lr=9.605033e-05 lr_backbone=9.605033e-05 elapsed=03:40:50 avg_step=1.55s recent_step=1.26s eta=17:50:39 +2026-09-20 10:19:12 | INFO | train rank=0 | step=8560/50000 progress=17.1% loss=0.107348 lr=9.603744e-05 lr_backbone=9.603744e-05 elapsed=03:41:12 avg_step=1.55s recent_step=2.12s eta=17:50:51 +2026-09-20 10:19:24 | INFO | train rank=0 | step=8570/50000 progress=17.1% loss=0.101093 lr=9.602452e-05 lr_backbone=9.602452e-05 elapsed=03:41:24 avg_step=1.55s recent_step=1.19s eta=17:50:19 +2026-09-20 10:19:36 | INFO | train rank=0 | step=8580/50000 progress=17.2% loss=0.121282 lr=9.601159e-05 lr_backbone=9.601159e-05 elapsed=03:41:36 avg_step=1.55s recent_step=1.24s eta=17:49:48 +2026-09-20 10:19:55 | INFO | train rank=0 | step=8590/50000 progress=17.2% loss=0.077175 lr=9.599864e-05 lr_backbone=9.599864e-05 elapsed=03:41:54 avg_step=1.55s recent_step=1.84s eta=17:49:46 +2026-09-20 10:20:12 | INFO | train rank=0 | step=8600/50000 progress=17.2% loss=0.128966 lr=9.598567e-05 lr_backbone=9.598567e-05 elapsed=03:42:11 avg_step=1.55s recent_step=1.70s eta=17:49:38 +2026-09-20 10:20:27 | INFO | train rank=0 | step=8610/50000 progress=17.2% loss=0.096383 lr=9.597267e-05 lr_backbone=9.597267e-05 elapsed=03:42:27 avg_step=1.55s recent_step=1.53s eta=17:49:22 +2026-09-20 10:20:45 | INFO | train rank=0 | step=8620/50000 progress=17.2% loss=0.125593 lr=9.595966e-05 lr_backbone=9.595966e-05 elapsed=03:42:45 avg_step=1.55s recent_step=1.82s eta=17:49:19 +2026-09-20 10:20:59 | INFO | train rank=0 | step=8630/50000 progress=17.3% loss=0.112541 lr=9.594663e-05 lr_backbone=9.594663e-05 elapsed=03:42:58 avg_step=1.55s recent_step=1.36s eta=17:48:55 +2026-09-20 10:21:15 | INFO | train rank=0 | step=8640/50000 progress=17.3% loss=0.118869 lr=9.593358e-05 lr_backbone=9.593358e-05 elapsed=03:43:15 avg_step=1.55s recent_step=1.62s eta=17:48:42 +2026-09-20 10:21:33 | INFO | train rank=0 | step=8650/50000 progress=17.3% loss=0.092413 lr=9.592050e-05 lr_backbone=9.592050e-05 elapsed=03:43:33 avg_step=1.55s recent_step=1.83s eta=17:48:40 +2026-09-20 10:21:45 | INFO | train rank=0 | step=8660/50000 progress=17.3% loss=0.090034 lr=9.590741e-05 lr_backbone=9.590741e-05 elapsed=03:43:44 avg_step=1.55s recent_step=1.14s eta=17:48:05 +2026-09-20 10:22:02 | INFO | train rank=0 | step=8670/50000 progress=17.3% loss=0.124699 lr=9.589430e-05 lr_backbone=9.589430e-05 elapsed=03:44:02 avg_step=1.55s recent_step=1.75s eta=17:47:59 +2026-09-20 10:22:19 | INFO | train rank=0 | step=8680/50000 progress=17.4% loss=0.144390 lr=9.588116e-05 lr_backbone=9.588116e-05 elapsed=03:44:19 avg_step=1.55s recent_step=1.72s eta=17:47:52 +2026-09-20 10:22:31 | INFO | train rank=0 | step=8690/50000 progress=17.4% loss=0.103204 lr=9.586801e-05 lr_backbone=9.586801e-05 elapsed=03:44:31 avg_step=1.55s recent_step=1.19s eta=17:47:19 +2026-09-20 10:22:44 | INFO | train rank=0 | step=8700/50000 progress=17.4% loss=0.122706 lr=9.585483e-05 lr_backbone=9.585483e-05 elapsed=03:44:43 avg_step=1.55s recent_step=1.24s eta=17:46:49 +2026-09-20 10:22:59 | INFO | train rank=0 | step=8710/50000 progress=17.4% loss=0.097581 lr=9.584164e-05 lr_backbone=9.584164e-05 elapsed=03:44:58 avg_step=1.55s recent_step=1.49s eta=17:46:30 +2026-09-20 10:23:17 | INFO | train rank=0 | step=8720/50000 progress=17.4% loss=0.119984 lr=9.582843e-05 lr_backbone=9.582843e-05 elapsed=03:45:17 avg_step=1.55s recent_step=1.86s eta=17:46:30 +2026-09-20 10:23:39 | INFO | train rank=0 | step=8730/50000 progress=17.5% loss=0.092474 lr=9.581519e-05 lr_backbone=9.581519e-05 elapsed=03:45:38 avg_step=1.55s recent_step=2.14s eta=17:46:42 +2026-09-20 10:23:50 | INFO | train rank=0 | step=8740/50000 progress=17.5% loss=0.118187 lr=9.580194e-05 lr_backbone=9.580194e-05 elapsed=03:45:50 avg_step=1.55s recent_step=1.18s eta=17:46:09 +2026-09-20 10:24:04 | INFO | train rank=0 | step=8750/50000 progress=17.5% loss=0.128885 lr=9.578867e-05 lr_backbone=9.578867e-05 elapsed=03:46:03 avg_step=1.55s recent_step=1.32s eta=17:45:43 +2026-09-20 10:24:20 | INFO | train rank=0 | step=8760/50000 progress=17.5% loss=0.157904 lr=9.577537e-05 lr_backbone=9.577537e-05 elapsed=03:46:19 avg_step=1.55s recent_step=1.63s eta=17:45:31 +2026-09-20 10:24:36 | INFO | train rank=0 | step=8770/50000 progress=17.5% loss=0.080302 lr=9.576206e-05 lr_backbone=9.576206e-05 elapsed=03:46:35 avg_step=1.55s recent_step=1.57s eta=17:45:16 +2026-09-20 10:24:54 | INFO | train rank=0 | step=8780/50000 progress=17.6% loss=0.109910 lr=9.574872e-05 lr_backbone=9.574872e-05 elapsed=03:46:53 avg_step=1.55s recent_step=1.79s eta=17:45:12 +2026-09-20 10:25:07 | INFO | train rank=0 | step=8790/50000 progress=17.6% loss=0.114731 lr=9.573537e-05 lr_backbone=9.573537e-05 elapsed=03:47:07 avg_step=1.55s recent_step=1.39s eta=17:44:49 +2026-09-20 10:25:26 | INFO | train rank=0 | step=8800/50000 progress=17.6% loss=0.105886 lr=9.572200e-05 lr_backbone=9.572200e-05 elapsed=03:47:26 avg_step=1.55s recent_step=1.85s eta=17:44:48 +2026-09-20 10:25:38 | INFO | train rank=0 | step=8810/50000 progress=17.6% loss=0.116356 lr=9.570860e-05 lr_backbone=9.570860e-05 elapsed=03:47:38 avg_step=1.55s recent_step=1.23s eta=17:44:17 +2026-09-20 10:25:55 | INFO | train rank=0 | step=8820/50000 progress=17.6% loss=0.096987 lr=9.569519e-05 lr_backbone=9.569519e-05 elapsed=03:47:55 avg_step=1.55s recent_step=1.68s eta=17:44:08 +2026-09-20 10:26:08 | INFO | train rank=0 | step=8830/50000 progress=17.7% loss=0.103363 lr=9.568176e-05 lr_backbone=9.568176e-05 elapsed=03:48:08 avg_step=1.55s recent_step=1.34s eta=17:43:43 +2026-09-20 10:26:28 | INFO | train rank=0 | step=8840/50000 progress=17.7% loss=0.109761 lr=9.566830e-05 lr_backbone=9.566830e-05 elapsed=03:48:28 avg_step=1.55s recent_step=1.99s eta=17:43:48 +2026-09-20 10:26:42 | INFO | train rank=0 | step=8850/50000 progress=17.7% loss=0.077972 lr=9.565483e-05 lr_backbone=9.565483e-05 elapsed=03:48:42 avg_step=1.55s recent_step=1.36s eta=17:43:23 +2026-09-20 10:26:54 | INFO | train rank=0 | step=8860/50000 progress=17.7% loss=0.088961 lr=9.564133e-05 lr_backbone=9.564133e-05 elapsed=03:48:54 avg_step=1.55s recent_step=1.20s eta=17:42:52 +2026-09-20 10:27:09 | INFO | train rank=0 | step=8870/50000 progress=17.7% loss=0.195749 lr=9.562782e-05 lr_backbone=9.562782e-05 elapsed=03:49:09 avg_step=1.55s recent_step=1.53s eta=17:42:35 +2026-09-20 10:27:26 | INFO | train rank=0 | step=8880/50000 progress=17.8% loss=0.095615 lr=9.561429e-05 lr_backbone=9.561429e-05 elapsed=03:49:26 avg_step=1.55s recent_step=1.67s eta=17:42:25 +2026-09-20 10:27:39 | INFO | train rank=0 | step=8890/50000 progress=17.8% loss=0.115231 lr=9.560073e-05 lr_backbone=9.560073e-05 elapsed=03:49:38 avg_step=1.55s recent_step=1.26s eta=17:41:56 +2026-09-20 10:27:59 | INFO | train rank=0 | step=8900/50000 progress=17.8% loss=0.123006 lr=9.558716e-05 lr_backbone=9.558716e-05 elapsed=03:49:58 avg_step=1.55s recent_step=2.01s eta=17:42:02 +2026-09-20 10:28:11 | INFO | train rank=0 | step=8910/50000 progress=17.8% loss=0.104167 lr=9.557357e-05 lr_backbone=9.557357e-05 elapsed=03:50:10 avg_step=1.55s recent_step=1.21s eta=17:41:31 +2026-09-20 10:28:34 | INFO | train rank=0 | step=8920/50000 progress=17.8% loss=0.129773 lr=9.555995e-05 lr_backbone=9.555995e-05 elapsed=03:50:34 avg_step=1.55s recent_step=2.37s eta=17:41:53 +2026-09-20 10:28:46 | INFO | train rank=0 | step=8930/50000 progress=17.9% loss=0.106829 lr=9.554632e-05 lr_backbone=9.554632e-05 elapsed=03:50:46 avg_step=1.55s recent_step=1.17s eta=17:41:20 +2026-09-20 10:28:58 | INFO | train rank=0 | step=8940/50000 progress=17.9% loss=0.104953 lr=9.553267e-05 lr_backbone=9.553267e-05 elapsed=03:50:58 avg_step=1.55s recent_step=1.20s eta=17:40:48 +2026-09-20 10:29:11 | INFO | train rank=0 | step=8950/50000 progress=17.9% loss=0.098254 lr=9.551899e-05 lr_backbone=9.551899e-05 elapsed=03:51:10 avg_step=1.55s recent_step=1.27s eta=17:40:20 +2026-09-20 10:29:31 | INFO | train rank=0 | step=8960/50000 progress=17.9% loss=0.110057 lr=9.550530e-05 lr_backbone=9.550530e-05 elapsed=03:51:30 avg_step=1.55s recent_step=1.99s eta=17:40:24 +2026-09-20 10:29:48 | INFO | train rank=0 | step=8970/50000 progress=17.9% loss=0.122239 lr=9.549159e-05 lr_backbone=9.549159e-05 elapsed=03:51:48 avg_step=1.55s recent_step=1.73s eta=17:40:17 +2026-09-20 10:30:00 | INFO | train rank=0 | step=8980/50000 progress=18.0% loss=0.133812 lr=9.547785e-05 lr_backbone=9.547785e-05 elapsed=03:52:00 avg_step=1.55s recent_step=1.23s eta=17:39:47 +2026-09-20 10:30:14 | INFO | train rank=0 | step=8990/50000 progress=18.0% loss=0.111849 lr=9.546410e-05 lr_backbone=9.546410e-05 elapsed=03:52:13 avg_step=1.55s recent_step=1.35s eta=17:39:22 +2026-09-20 10:30:35 | INFO | train rank=0 | step=9000/50000 progress=18.0% loss=0.119867 lr=9.545033e-05 lr_backbone=9.545033e-05 elapsed=03:52:35 avg_step=1.55s recent_step=2.11s eta=17:39:33 +2026-09-20 10:30:47 | INFO | train rank=0 | step=9010/50000 progress=18.0% loss=0.127811 lr=9.543653e-05 lr_backbone=9.543653e-05 elapsed=03:52:47 avg_step=1.55s recent_step=1.23s eta=17:39:02 +2026-09-20 10:31:02 | INFO | train rank=0 | step=9020/50000 progress=18.0% loss=0.107827 lr=9.542272e-05 lr_backbone=9.542272e-05 elapsed=03:53:02 avg_step=1.55s recent_step=1.52s eta=17:38:45 +2026-09-20 10:31:16 | INFO | train rank=0 | step=9030/50000 progress=18.1% loss=0.110748 lr=9.540889e-05 lr_backbone=9.540889e-05 elapsed=03:53:16 avg_step=1.55s recent_step=1.39s eta=17:38:23 +2026-09-20 10:31:36 | INFO | train rank=0 | step=9040/50000 progress=18.1% loss=0.089921 lr=9.539504e-05 lr_backbone=9.539504e-05 elapsed=03:53:36 avg_step=1.55s recent_step=1.97s eta=17:38:26 +2026-09-20 10:31:57 | INFO | train rank=0 | step=9050/50000 progress=18.1% loss=0.123849 lr=9.538116e-05 lr_backbone=9.538116e-05 elapsed=03:53:57 avg_step=1.55s recent_step=2.13s eta=17:38:37 +2026-09-20 10:32:10 | INFO | train rank=0 | step=9060/50000 progress=18.1% loss=0.091723 lr=9.536727e-05 lr_backbone=9.536727e-05 elapsed=03:54:09 avg_step=1.55s recent_step=1.23s eta=17:38:07 +2026-09-20 10:32:22 | INFO | train rank=0 | step=9070/50000 progress=18.1% loss=0.154364 lr=9.535336e-05 lr_backbone=9.535336e-05 elapsed=03:54:21 avg_step=1.55s recent_step=1.19s eta=17:37:35 +2026-09-20 10:32:34 | INFO | train rank=0 | step=9080/50000 progress=18.2% loss=0.113300 lr=9.533943e-05 lr_backbone=9.533943e-05 elapsed=03:54:33 avg_step=1.55s recent_step=1.22s eta=17:37:05 +2026-09-20 10:32:49 | INFO | train rank=0 | step=9090/50000 progress=18.2% loss=0.096665 lr=9.532548e-05 lr_backbone=9.532548e-05 elapsed=03:54:48 avg_step=1.55s recent_step=1.51s eta=17:36:48 +2026-09-20 10:33:06 | INFO | train rank=0 | step=9100/50000 progress=18.2% loss=0.098505 lr=9.531151e-05 lr_backbone=9.531151e-05 elapsed=03:55:06 avg_step=1.55s recent_step=1.72s eta=17:36:40 +2026-09-20 10:33:19 | INFO | train rank=0 | step=9110/50000 progress=18.2% loss=0.102448 lr=9.529752e-05 lr_backbone=9.529752e-05 elapsed=03:55:18 avg_step=1.55s recent_step=1.24s eta=17:36:11 +2026-09-20 10:33:40 | INFO | train rank=0 | step=9120/50000 progress=18.2% loss=0.103469 lr=9.528350e-05 lr_backbone=9.528350e-05 elapsed=03:55:39 avg_step=1.55s recent_step=2.11s eta=17:36:20 +2026-09-20 10:33:54 | INFO | train rank=0 | step=9130/50000 progress=18.3% loss=0.146749 lr=9.526947e-05 lr_backbone=9.526947e-05 elapsed=03:55:53 avg_step=1.55s recent_step=1.40s eta=17:35:58 +2026-09-20 10:34:11 | INFO | train rank=0 | step=9140/50000 progress=18.3% loss=0.094428 lr=9.525542e-05 lr_backbone=9.525542e-05 elapsed=03:56:11 avg_step=1.55s recent_step=1.74s eta=17:35:51 +2026-09-20 10:34:24 | INFO | train rank=0 | step=9150/50000 progress=18.3% loss=0.106488 lr=9.524135e-05 lr_backbone=9.524135e-05 elapsed=03:56:23 avg_step=1.55s recent_step=1.26s eta=17:35:23 +2026-09-20 10:34:40 | INFO | train rank=0 | step=9160/50000 progress=18.3% loss=0.126502 lr=9.522726e-05 lr_backbone=9.522726e-05 elapsed=03:56:39 avg_step=1.55s recent_step=1.59s eta=17:35:09 +2026-09-20 10:34:54 | INFO | train rank=0 | step=9170/50000 progress=18.3% loss=0.109864 lr=9.521315e-05 lr_backbone=9.521315e-05 elapsed=03:56:54 avg_step=1.55s recent_step=1.43s eta=17:34:48 +2026-09-20 10:35:13 | INFO | train rank=0 | step=9180/50000 progress=18.4% loss=0.128207 lr=9.519902e-05 lr_backbone=9.519902e-05 elapsed=03:57:12 avg_step=1.55s recent_step=1.86s eta=17:34:47 +2026-09-20 10:35:24 | INFO | train rank=0 | step=9190/50000 progress=18.4% loss=0.124399 lr=9.518487e-05 lr_backbone=9.518487e-05 elapsed=03:57:24 avg_step=1.55s recent_step=1.17s eta=17:34:15 +2026-09-20 10:35:43 | INFO | train rank=0 | step=9200/50000 progress=18.4% loss=0.136607 lr=9.517070e-05 lr_backbone=9.517070e-05 elapsed=03:57:43 avg_step=1.55s recent_step=1.91s eta=17:34:15 +2026-09-20 10:36:01 | INFO | train rank=0 | step=9210/50000 progress=18.4% loss=0.100300 lr=9.515652e-05 lr_backbone=9.515652e-05 elapsed=03:58:00 avg_step=1.55s recent_step=1.71s eta=17:34:07 +2026-09-20 10:36:12 | INFO | train rank=0 | step=9220/50000 progress=18.4% loss=0.106686 lr=9.514231e-05 lr_backbone=9.514231e-05 elapsed=03:58:12 avg_step=1.55s recent_step=1.17s eta=17:33:34 +2026-09-20 10:36:29 | INFO | train rank=0 | step=9230/50000 progress=18.5% loss=0.126466 lr=9.512808e-05 lr_backbone=9.512808e-05 elapsed=03:58:28 avg_step=1.55s recent_step=1.63s eta=17:33:22 +2026-09-20 10:36:48 | INFO | train rank=0 | step=9240/50000 progress=18.5% loss=0.128831 lr=9.511383e-05 lr_backbone=9.511383e-05 elapsed=03:58:48 avg_step=1.55s recent_step=1.99s eta=17:33:26 +2026-09-20 10:37:05 | INFO | train rank=0 | step=9250/50000 progress=18.5% loss=0.099088 lr=9.509956e-05 lr_backbone=9.509956e-05 elapsed=03:59:05 avg_step=1.55s recent_step=1.71s eta=17:33:18 +2026-09-20 10:37:19 | INFO | train rank=0 | step=9260/50000 progress=18.5% loss=0.098153 lr=9.508527e-05 lr_backbone=9.508527e-05 elapsed=03:59:19 avg_step=1.55s recent_step=1.36s eta=17:32:54 +2026-09-20 10:37:33 | INFO | train rank=0 | step=9270/50000 progress=18.5% loss=0.093590 lr=9.507097e-05 lr_backbone=9.507097e-05 elapsed=03:59:32 avg_step=1.55s recent_step=1.37s eta=17:32:30 +2026-09-20 10:37:48 | INFO | train rank=0 | step=9280/50000 progress=18.6% loss=0.106737 lr=9.505664e-05 lr_backbone=9.505664e-05 elapsed=03:59:48 avg_step=1.55s recent_step=1.56s eta=17:32:15 +2026-09-20 10:38:00 | INFO | train rank=0 | step=9290/50000 progress=18.6% loss=0.115115 lr=9.504229e-05 lr_backbone=9.504229e-05 elapsed=04:00:00 avg_step=1.55s recent_step=1.19s eta=17:31:44 +2026-09-20 10:38:20 | INFO | train rank=0 | step=9300/50000 progress=18.6% loss=0.086348 lr=9.502793e-05 lr_backbone=9.502793e-05 elapsed=04:00:20 avg_step=1.55s recent_step=2.02s eta=17:31:49 +2026-09-20 10:38:35 | INFO | train rank=0 | step=9310/50000 progress=18.6% loss=0.108888 lr=9.501354e-05 lr_backbone=9.501354e-05 elapsed=04:00:34 avg_step=1.55s recent_step=1.42s eta=17:31:28 +2026-09-20 10:38:51 | INFO | train rank=0 | step=9320/50000 progress=18.6% loss=0.106063 lr=9.499913e-05 lr_backbone=9.499913e-05 elapsed=04:00:51 avg_step=1.55s recent_step=1.66s eta=17:31:17 +2026-09-20 10:39:09 | INFO | train rank=0 | step=9330/50000 progress=18.7% loss=0.113595 lr=9.498471e-05 lr_backbone=9.498471e-05 elapsed=04:01:09 avg_step=1.55s recent_step=1.78s eta=17:31:12 +2026-09-20 10:39:23 | INFO | train rank=0 | step=9340/50000 progress=18.7% loss=0.095004 lr=9.497026e-05 lr_backbone=9.497026e-05 elapsed=04:01:22 avg_step=1.55s recent_step=1.37s eta=17:30:48 +2026-09-20 10:39:39 | INFO | train rank=0 | step=9350/50000 progress=18.7% loss=0.104598 lr=9.495580e-05 lr_backbone=9.495580e-05 elapsed=04:01:39 avg_step=1.55s recent_step=1.61s eta=17:30:35 +2026-09-20 10:39:55 | INFO | train rank=0 | step=9360/50000 progress=18.7% loss=0.119488 lr=9.494131e-05 lr_backbone=9.494131e-05 elapsed=04:01:54 avg_step=1.55s recent_step=1.59s eta=17:30:22 +2026-09-20 10:40:07 | INFO | train rank=0 | step=9370/50000 progress=18.7% loss=0.121287 lr=9.492681e-05 lr_backbone=9.492681e-05 elapsed=04:02:07 avg_step=1.55s recent_step=1.26s eta=17:29:54 +2026-09-20 10:40:20 | INFO | train rank=0 | step=9380/50000 progress=18.8% loss=0.104528 lr=9.491228e-05 lr_backbone=9.491228e-05 elapsed=04:02:19 avg_step=1.55s recent_step=1.23s eta=17:29:24 +2026-09-20 10:40:43 | INFO | train rank=0 | step=9390/50000 progress=18.8% loss=0.137025 lr=9.489774e-05 lr_backbone=9.489774e-05 elapsed=04:02:43 avg_step=1.55s recent_step=2.35s eta=17:29:43 +2026-09-20 10:40:56 | INFO | train rank=0 | step=9400/50000 progress=18.8% loss=0.142684 lr=9.488318e-05 lr_backbone=9.488318e-05 elapsed=04:02:55 avg_step=1.55s recent_step=1.23s eta=17:29:14 +2026-09-20 10:41:14 | INFO | train rank=0 | step=9410/50000 progress=18.8% loss=0.094382 lr=9.486859e-05 lr_backbone=9.486859e-05 elapsed=04:03:13 avg_step=1.55s recent_step=1.83s eta=17:29:10 +2026-09-20 10:41:26 | INFO | train rank=0 | step=9420/50000 progress=18.8% loss=0.123806 lr=9.485399e-05 lr_backbone=9.485399e-05 elapsed=04:03:26 avg_step=1.55s recent_step=1.25s eta=17:28:42 +2026-09-20 10:41:40 | INFO | train rank=0 | step=9430/50000 progress=18.9% loss=0.103450 lr=9.483937e-05 lr_backbone=9.483937e-05 elapsed=04:03:40 avg_step=1.55s recent_step=1.35s eta=17:28:18 +2026-09-20 10:41:54 | INFO | train rank=0 | step=9440/50000 progress=18.9% loss=0.101639 lr=9.482473e-05 lr_backbone=9.482473e-05 elapsed=04:03:54 avg_step=1.55s recent_step=1.44s eta=17:27:58 +2026-09-20 10:42:12 | INFO | train rank=0 | step=9450/50000 progress=18.9% loss=0.112593 lr=9.481007e-05 lr_backbone=9.481007e-05 elapsed=04:04:12 avg_step=1.55s recent_step=1.79s eta=17:27:53 +2026-09-20 10:42:24 | INFO | train rank=0 | step=9460/50000 progress=18.9% loss=0.100187 lr=9.479539e-05 lr_backbone=9.479539e-05 elapsed=04:04:24 avg_step=1.55s recent_step=1.19s eta=17:27:22 +2026-09-20 10:42:36 | INFO | train rank=0 | step=9470/50000 progress=18.9% loss=0.136641 lr=9.478069e-05 lr_backbone=9.478069e-05 elapsed=04:04:36 avg_step=1.55s recent_step=1.21s eta=17:26:51 +2026-09-20 10:42:57 | INFO | train rank=0 | step=9480/50000 progress=19.0% loss=0.132839 lr=9.476597e-05 lr_backbone=9.476597e-05 elapsed=04:04:57 avg_step=1.55s recent_step=2.12s eta=17:27:00 +2026-09-20 10:43:12 | INFO | train rank=0 | step=9490/50000 progress=19.0% loss=0.143494 lr=9.475123e-05 lr_backbone=9.475123e-05 elapsed=04:05:12 avg_step=1.55s recent_step=1.52s eta=17:26:43 +2026-09-20 10:43:29 | INFO | train rank=0 | step=9500/50000 progress=19.0% loss=0.176282 lr=9.473647e-05 lr_backbone=9.473647e-05 elapsed=04:05:28 avg_step=1.55s recent_step=1.63s eta=17:26:31 +2026-09-20 10:43:41 | INFO | train rank=0 | step=9510/50000 progress=19.0% loss=0.167832 lr=9.472169e-05 lr_backbone=9.472169e-05 elapsed=04:05:41 avg_step=1.55s recent_step=1.21s eta=17:26:01 +2026-09-20 10:44:01 | INFO | train rank=0 | step=9520/50000 progress=19.0% loss=0.089623 lr=9.470689e-05 lr_backbone=9.470689e-05 elapsed=04:06:00 avg_step=1.55s recent_step=1.99s eta=17:26:04 +2026-09-20 10:44:15 | INFO | train rank=0 | step=9530/50000 progress=19.1% loss=0.113105 lr=9.469207e-05 lr_backbone=9.469207e-05 elapsed=04:06:15 avg_step=1.55s recent_step=1.42s eta=17:25:43 +2026-09-20 10:44:27 | INFO | train rank=0 | step=9540/50000 progress=19.1% loss=0.092752 lr=9.467723e-05 lr_backbone=9.467723e-05 elapsed=04:06:27 avg_step=1.55s recent_step=1.20s eta=17:25:13 +2026-09-20 10:44:44 | INFO | train rank=0 | step=9550/50000 progress=19.1% loss=0.098248 lr=9.466238e-05 lr_backbone=9.466238e-05 elapsed=04:06:43 avg_step=1.55s recent_step=1.66s eta=17:25:02 +2026-09-20 10:45:01 | INFO | train rank=0 | step=9560/50000 progress=19.1% loss=0.103042 lr=9.464750e-05 lr_backbone=9.464750e-05 elapsed=04:07:01 avg_step=1.55s recent_step=1.75s eta=17:24:55 +2026-09-20 10:45:19 | INFO | train rank=0 | step=9570/50000 progress=19.1% loss=0.115656 lr=9.463260e-05 lr_backbone=9.463260e-05 elapsed=04:07:18 avg_step=1.55s recent_step=1.75s eta=17:24:48 +2026-09-20 10:45:35 | INFO | train rank=0 | step=9580/50000 progress=19.2% loss=0.147793 lr=9.461769e-05 lr_backbone=9.461769e-05 elapsed=04:07:35 avg_step=1.55s recent_step=1.64s eta=17:24:37 +2026-09-20 10:45:47 | INFO | train rank=0 | step=9590/50000 progress=19.2% loss=0.122599 lr=9.460275e-05 lr_backbone=9.460275e-05 elapsed=04:07:47 avg_step=1.55s recent_step=1.21s eta=17:24:07 +2026-09-20 10:46:01 | INFO | train rank=0 | step=9600/50000 progress=19.2% loss=0.155703 lr=9.458780e-05 lr_backbone=9.458780e-05 elapsed=04:08:00 avg_step=1.55s recent_step=1.36s eta=17:23:43 +2026-09-20 10:46:13 | INFO | train rank=0 | step=9610/50000 progress=19.2% loss=0.106808 lr=9.457282e-05 lr_backbone=9.457282e-05 elapsed=04:08:13 avg_step=1.55s recent_step=1.22s eta=17:23:14 +2026-09-20 10:46:26 | INFO | train rank=0 | step=9620/50000 progress=19.2% loss=0.125855 lr=9.455783e-05 lr_backbone=9.455783e-05 elapsed=04:08:25 avg_step=1.55s recent_step=1.25s eta=17:22:46 +2026-09-20 10:46:38 | INFO | train rank=0 | step=9630/50000 progress=19.3% loss=0.105299 lr=9.454282e-05 lr_backbone=9.454282e-05 elapsed=04:08:37 avg_step=1.55s recent_step=1.24s eta=17:22:17 +2026-09-20 10:46:54 | INFO | train rank=0 | step=9640/50000 progress=19.3% loss=0.137271 lr=9.452778e-05 lr_backbone=9.452778e-05 elapsed=04:08:53 avg_step=1.55s recent_step=1.57s eta=17:22:03 +2026-09-20 10:47:15 | INFO | train rank=0 | step=9650/50000 progress=19.3% loss=0.110805 lr=9.451273e-05 lr_backbone=9.451273e-05 elapsed=04:09:14 avg_step=1.55s recent_step=2.11s eta=17:22:10 +2026-09-20 10:47:28 | INFO | train rank=0 | step=9660/50000 progress=19.3% loss=0.110184 lr=9.449766e-05 lr_backbone=9.449766e-05 elapsed=04:09:28 avg_step=1.55s recent_step=1.34s eta=17:21:46 +2026-09-20 10:47:50 | INFO | train rank=0 | step=9670/50000 progress=19.3% loss=0.175039 lr=9.448257e-05 lr_backbone=9.448257e-05 elapsed=04:09:50 avg_step=1.55s recent_step=2.19s eta=17:21:57 +2026-09-20 10:48:08 | INFO | train rank=0 | step=9680/50000 progress=19.4% loss=0.163136 lr=9.446746e-05 lr_backbone=9.446746e-05 elapsed=04:10:08 avg_step=1.55s recent_step=1.82s eta=17:21:53 +2026-09-20 10:48:24 | INFO | train rank=0 | step=9690/50000 progress=19.4% loss=0.091129 lr=9.445233e-05 lr_backbone=9.445233e-05 elapsed=04:10:24 avg_step=1.55s recent_step=1.58s eta=17:21:39 +2026-09-20 10:48:43 | INFO | train rank=0 | step=9700/50000 progress=19.4% loss=0.107189 lr=9.443718e-05 lr_backbone=9.443718e-05 elapsed=04:10:42 avg_step=1.55s recent_step=1.89s eta=17:21:37 +2026-09-20 10:48:59 | INFO | train rank=0 | step=9710/50000 progress=19.4% loss=0.140345 lr=9.442201e-05 lr_backbone=9.442201e-05 elapsed=04:10:59 avg_step=1.55s recent_step=1.63s eta=17:21:25 +2026-09-20 10:49:11 | INFO | train rank=0 | step=9720/50000 progress=19.4% loss=0.120701 lr=9.440682e-05 lr_backbone=9.440682e-05 elapsed=04:11:11 avg_step=1.55s recent_step=1.19s eta=17:20:55 +2026-09-20 10:49:28 | INFO | train rank=0 | step=9730/50000 progress=19.5% loss=0.109652 lr=9.439161e-05 lr_backbone=9.439161e-05 elapsed=04:11:27 avg_step=1.55s recent_step=1.68s eta=17:20:44 +2026-09-20 10:49:43 | INFO | train rank=0 | step=9740/50000 progress=19.5% loss=0.096767 lr=9.437639e-05 lr_backbone=9.437639e-05 elapsed=04:11:43 avg_step=1.55s recent_step=1.57s eta=17:20:30 +2026-09-20 10:49:55 | INFO | train rank=0 | step=9750/50000 progress=19.5% loss=0.120408 lr=9.436114e-05 lr_backbone=9.436114e-05 elapsed=04:11:55 avg_step=1.55s recent_step=1.18s eta=17:19:59 +2026-09-20 10:50:56 | INFO | train rank=0 | step=9760/50000 progress=19.5% loss=0.084060 lr=9.434587e-05 lr_backbone=9.434587e-05 elapsed=04:12:55 avg_step=1.55s recent_step=6.05s eta=17:22:49 +2026-09-20 10:51:08 | INFO | train rank=0 | step=9770/50000 progress=19.5% loss=0.115748 lr=9.433059e-05 lr_backbone=9.433059e-05 elapsed=04:13:07 avg_step=1.55s recent_step=1.21s eta=17:22:19 +2026-09-20 10:51:22 | INFO | train rank=0 | step=9780/50000 progress=19.6% loss=0.099714 lr=9.431528e-05 lr_backbone=9.431528e-05 elapsed=04:13:22 avg_step=1.55s recent_step=1.43s eta=17:21:58 +2026-09-20 10:51:35 | INFO | train rank=0 | step=9790/50000 progress=19.6% loss=0.095832 lr=9.429996e-05 lr_backbone=9.429996e-05 elapsed=04:13:35 avg_step=1.55s recent_step=1.32s eta=17:21:33 +2026-09-20 10:51:57 | INFO | train rank=0 | step=9800/50000 progress=19.6% loss=0.162341 lr=9.428462e-05 lr_backbone=9.428462e-05 elapsed=04:13:56 avg_step=1.55s recent_step=2.12s eta=17:21:41 +2026-09-20 10:52:09 | INFO | train rank=0 | step=9810/50000 progress=19.6% loss=0.105144 lr=9.426925e-05 lr_backbone=9.426925e-05 elapsed=04:14:08 avg_step=1.55s recent_step=1.21s eta=17:21:11 +2026-09-20 10:52:25 | INFO | train rank=0 | step=9820/50000 progress=19.6% loss=0.168484 lr=9.425387e-05 lr_backbone=9.425387e-05 elapsed=04:14:25 avg_step=1.55s recent_step=1.63s eta=17:20:59 +2026-09-20 10:52:44 | INFO | train rank=0 | step=9830/50000 progress=19.7% loss=0.077239 lr=9.423847e-05 lr_backbone=9.423847e-05 elapsed=04:14:43 avg_step=1.55s recent_step=1.88s eta=17:20:56 +2026-09-20 10:53:00 | INFO | train rank=0 | step=9840/50000 progress=19.7% loss=0.136572 lr=9.422305e-05 lr_backbone=9.422305e-05 elapsed=04:15:00 avg_step=1.55s recent_step=1.65s eta=17:20:45 +2026-09-20 10:53:12 | INFO | train rank=0 | step=9850/50000 progress=19.7% loss=0.136666 lr=9.420761e-05 lr_backbone=9.420761e-05 elapsed=04:15:11 avg_step=1.55s recent_step=1.16s eta=17:20:13 +2026-09-20 10:53:27 | INFO | train rank=0 | step=9860/50000 progress=19.7% loss=0.119429 lr=9.419215e-05 lr_backbone=9.419215e-05 elapsed=04:15:27 avg_step=1.55s recent_step=1.51s eta=17:19:56 +2026-09-20 10:53:45 | INFO | train rank=0 | step=9870/50000 progress=19.7% loss=0.119141 lr=9.417667e-05 lr_backbone=9.417667e-05 elapsed=04:15:44 avg_step=1.55s recent_step=1.77s eta=17:19:49 +2026-09-20 10:53:59 | INFO | train rank=0 | step=9880/50000 progress=19.8% loss=0.101367 lr=9.416117e-05 lr_backbone=9.416117e-05 elapsed=04:15:59 avg_step=1.55s recent_step=1.48s eta=17:19:31 +2026-09-20 10:54:15 | INFO | train rank=0 | step=9890/50000 progress=19.8% loss=0.097018 lr=9.414565e-05 lr_backbone=9.414565e-05 elapsed=04:16:15 avg_step=1.55s recent_step=1.58s eta=17:19:16 +2026-09-20 10:54:31 | INFO | train rank=0 | step=9900/50000 progress=19.8% loss=0.100902 lr=9.413012e-05 lr_backbone=9.413012e-05 elapsed=04:16:31 avg_step=1.55s recent_step=1.59s eta=17:19:02 +2026-09-20 10:54:45 | INFO | train rank=0 | step=9910/50000 progress=19.8% loss=0.092285 lr=9.411456e-05 lr_backbone=9.411456e-05 elapsed=04:16:45 avg_step=1.55s recent_step=1.41s eta=17:18:40 +2026-09-20 10:55:03 | INFO | train rank=0 | step=9920/50000 progress=19.8% loss=0.123498 lr=9.409899e-05 lr_backbone=9.409899e-05 elapsed=04:17:03 avg_step=1.55s recent_step=1.78s eta=17:18:34 +2026-09-20 10:55:19 | INFO | train rank=0 | step=9930/50000 progress=19.9% loss=0.121335 lr=9.408339e-05 lr_backbone=9.408339e-05 elapsed=04:17:19 avg_step=1.55s recent_step=1.64s eta=17:18:22 +2026-09-20 10:55:35 | INFO | train rank=0 | step=9940/50000 progress=19.9% loss=0.108800 lr=9.406778e-05 lr_backbone=9.406778e-05 elapsed=04:17:34 avg_step=1.55s recent_step=1.51s eta=17:18:04 +2026-09-20 10:55:53 | INFO | train rank=0 | step=9950/50000 progress=19.9% loss=0.107567 lr=9.405214e-05 lr_backbone=9.405214e-05 elapsed=04:17:52 avg_step=1.56s recent_step=1.81s eta=17:17:59 +2026-09-20 10:56:11 | INFO | train rank=0 | step=9960/50000 progress=19.9% loss=0.107560 lr=9.403649e-05 lr_backbone=9.403649e-05 elapsed=04:18:10 avg_step=1.56s recent_step=1.79s eta=17:17:53 +2026-09-20 10:56:27 | INFO | train rank=0 | step=9970/50000 progress=19.9% loss=0.109778 lr=9.402082e-05 lr_backbone=9.402082e-05 elapsed=04:18:27 avg_step=1.56s recent_step=1.65s eta=17:17:41 +2026-09-20 10:56:39 | INFO | train rank=0 | step=9980/50000 progress=20.0% loss=0.175662 lr=9.400513e-05 lr_backbone=9.400513e-05 elapsed=04:18:39 avg_step=1.56s recent_step=1.19s eta=17:17:11 +2026-09-20 10:56:56 | INFO | train rank=0 | step=9990/50000 progress=20.0% loss=0.113288 lr=9.398942e-05 lr_backbone=9.398942e-05 elapsed=04:18:56 avg_step=1.56s recent_step=1.75s eta=17:17:03 +2026-09-20 10:57:09 | INFO | train rank=0 | step=10000/50000 progress=20.0% loss=0.148923 lr=9.397369e-05 lr_backbone=9.397369e-05 elapsed=04:19:09 avg_step=1.55s recent_step=1.26s eta=17:16:36 +2026-09-20 10:57:21 | INFO | train rank=0 | step=10010/50000 progress=20.0% loss=0.148704 lr=9.395794e-05 lr_backbone=9.395794e-05 elapsed=04:19:21 avg_step=1.55s recent_step=1.21s eta=17:16:06 +2026-09-20 10:57:38 | INFO | train rank=0 | step=10020/50000 progress=20.0% loss=0.150737 lr=9.394217e-05 lr_backbone=9.394217e-05 elapsed=04:19:37 avg_step=1.55s recent_step=1.66s eta=17:15:55 +2026-09-20 10:57:50 | INFO | train rank=0 | step=10030/50000 progress=20.1% loss=0.111902 lr=9.392638e-05 lr_backbone=9.392638e-05 elapsed=04:19:50 avg_step=1.55s recent_step=1.24s eta=17:15:27 +2026-09-20 10:58:12 | INFO | train rank=0 | step=10040/50000 progress=20.1% loss=0.099013 lr=9.391058e-05 lr_backbone=9.391058e-05 elapsed=04:20:11 avg_step=1.55s recent_step=2.15s eta=17:15:35 +2026-09-20 10:58:24 | INFO | train rank=0 | step=10050/50000 progress=20.1% loss=0.110483 lr=9.389475e-05 lr_backbone=9.389475e-05 elapsed=04:20:23 avg_step=1.55s recent_step=1.20s eta=17:15:06 +2026-09-20 10:58:37 | INFO | train rank=0 | step=10060/50000 progress=20.1% loss=0.144018 lr=9.387891e-05 lr_backbone=9.387891e-05 elapsed=04:20:37 avg_step=1.55s recent_step=1.35s eta=17:14:42 +2026-09-20 10:58:53 | INFO | train rank=0 | step=10070/50000 progress=20.1% loss=0.088841 lr=9.386304e-05 lr_backbone=9.386304e-05 elapsed=04:20:52 avg_step=1.55s recent_step=1.54s eta=17:14:26 +2026-09-20 10:59:12 | INFO | train rank=0 | step=10080/50000 progress=20.2% loss=0.133355 lr=9.384716e-05 lr_backbone=9.384716e-05 elapsed=04:21:12 avg_step=1.55s recent_step=1.95s eta=17:14:26 +2026-09-20 10:59:24 | INFO | train rank=0 | step=10090/50000 progress=20.2% loss=0.115041 lr=9.383126e-05 lr_backbone=9.383126e-05 elapsed=04:21:24 avg_step=1.55s recent_step=1.23s eta=17:13:58 +2026-09-20 10:59:40 | INFO | train rank=0 | step=10100/50000 progress=20.2% loss=0.184154 lr=9.381533e-05 lr_backbone=9.381533e-05 elapsed=04:21:40 avg_step=1.55s recent_step=1.56s eta=17:13:43 +2026-09-20 10:59:52 | INFO | train rank=0 | step=10110/50000 progress=20.2% loss=0.114291 lr=9.379939e-05 lr_backbone=9.379939e-05 elapsed=04:21:52 avg_step=1.55s recent_step=1.20s eta=17:13:13 +2026-09-20 11:00:10 | INFO | train rank=0 | step=10120/50000 progress=20.2% loss=0.088954 lr=9.378343e-05 lr_backbone=9.378343e-05 elapsed=04:22:10 avg_step=1.55s recent_step=1.81s eta=17:13:08 +2026-09-20 11:00:26 | INFO | train rank=0 | step=10130/50000 progress=20.3% loss=0.171188 lr=9.376745e-05 lr_backbone=9.376745e-05 elapsed=04:22:25 avg_step=1.55s recent_step=1.55s eta=17:12:52 +2026-09-20 11:00:39 | INFO | train rank=0 | step=10140/50000 progress=20.3% loss=0.172883 lr=9.375146e-05 lr_backbone=9.375146e-05 elapsed=04:22:39 avg_step=1.55s recent_step=1.38s eta=17:12:29 +2026-09-20 11:01:00 | INFO | train rank=0 | step=10150/50000 progress=20.3% loss=0.102080 lr=9.373544e-05 lr_backbone=9.373544e-05 elapsed=04:23:00 avg_step=1.55s recent_step=2.07s eta=17:12:34 +2026-09-20 11:01:14 | INFO | train rank=0 | step=10160/50000 progress=20.3% loss=0.098692 lr=9.371940e-05 lr_backbone=9.371940e-05 elapsed=04:23:14 avg_step=1.55s recent_step=1.40s eta=17:12:13 +2026-09-20 11:01:35 | INFO | train rank=0 | step=10170/50000 progress=20.3% loss=0.104331 lr=9.370335e-05 lr_backbone=9.370335e-05 elapsed=04:23:34 avg_step=1.56s recent_step=2.06s eta=17:12:17 +2026-09-20 11:01:50 | INFO | train rank=0 | step=10180/50000 progress=20.4% loss=0.120051 lr=9.368727e-05 lr_backbone=9.368727e-05 elapsed=04:23:49 avg_step=1.56s recent_step=1.52s eta=17:12:00 +2026-09-20 11:02:47 | INFO | train rank=0 | step=10190/50000 progress=20.4% loss=0.100569 lr=9.367118e-05 lr_backbone=9.367118e-05 elapsed=04:24:47 avg_step=1.56s recent_step=5.72s eta=17:14:27 +2026-09-20 11:03:01 | INFO | train rank=0 | step=10200/50000 progress=20.4% loss=0.102416 lr=9.365506e-05 lr_backbone=9.365506e-05 elapsed=04:25:00 avg_step=1.56s recent_step=1.39s eta=17:14:04 +2026-09-20 11:03:13 | INFO | train rank=0 | step=10210/50000 progress=20.4% loss=0.111028 lr=9.363893e-05 lr_backbone=9.363893e-05 elapsed=04:25:12 avg_step=1.56s recent_step=1.19s eta=17:13:35 +2026-09-20 11:03:26 | INFO | train rank=0 | step=10220/50000 progress=20.4% loss=0.109253 lr=9.362278e-05 lr_backbone=9.362278e-05 elapsed=04:25:26 avg_step=1.56s recent_step=1.34s eta=17:13:10 +2026-09-20 11:03:47 | INFO | train rank=0 | step=10230/50000 progress=20.5% loss=0.079851 lr=9.360661e-05 lr_backbone=9.360661e-05 elapsed=04:25:46 avg_step=1.56s recent_step=2.04s eta=17:13:14 +2026-09-20 11:03:59 | INFO | train rank=0 | step=10240/50000 progress=20.5% loss=0.111636 lr=9.359042e-05 lr_backbone=9.359042e-05 elapsed=04:25:58 avg_step=1.56s recent_step=1.22s eta=17:12:45 +2026-09-20 11:04:16 | INFO | train rank=0 | step=10250/50000 progress=20.5% loss=0.098269 lr=9.357421e-05 lr_backbone=9.357421e-05 elapsed=04:26:16 avg_step=1.56s recent_step=1.71s eta=17:12:35 +2026-09-20 11:04:29 | INFO | train rank=0 | step=10260/50000 progress=20.5% loss=0.094217 lr=9.355798e-05 lr_backbone=9.355798e-05 elapsed=04:26:29 avg_step=1.56s recent_step=1.36s eta=17:12:12 +2026-09-20 11:04:48 | INFO | train rank=0 | step=10270/50000 progress=20.5% loss=0.106571 lr=9.354174e-05 lr_backbone=9.354174e-05 elapsed=04:26:48 avg_step=1.56s recent_step=1.88s eta=17:12:09 +2026-09-20 11:05:01 | INFO | train rank=0 | step=10280/50000 progress=20.6% loss=0.126480 lr=9.352547e-05 lr_backbone=9.352547e-05 elapsed=04:27:00 avg_step=1.56s recent_step=1.25s eta=17:11:41 +2026-09-20 11:05:16 | INFO | train rank=0 | step=10290/50000 progress=20.6% loss=0.096984 lr=9.350919e-05 lr_backbone=9.350919e-05 elapsed=04:27:16 avg_step=1.56s recent_step=1.51s eta=17:11:24 +2026-09-20 11:05:31 | INFO | train rank=0 | step=10300/50000 progress=20.6% loss=0.076614 lr=9.349288e-05 lr_backbone=9.349288e-05 elapsed=04:27:31 avg_step=1.56s recent_step=1.55s eta=17:11:08 +2026-09-20 11:05:45 | INFO | train rank=0 | step=10310/50000 progress=20.6% loss=0.153502 lr=9.347656e-05 lr_backbone=9.347656e-05 elapsed=04:27:44 avg_step=1.56s recent_step=1.34s eta=17:10:44 +2026-09-20 11:06:01 | INFO | train rank=0 | step=10320/50000 progress=20.6% loss=0.130750 lr=9.346022e-05 lr_backbone=9.346022e-05 elapsed=04:28:01 avg_step=1.56s recent_step=1.63s eta=17:10:31 +2026-09-20 11:06:16 | INFO | train rank=0 | step=10330/50000 progress=20.7% loss=0.094590 lr=9.344386e-05 lr_backbone=9.344386e-05 elapsed=04:28:15 avg_step=1.56s recent_step=1.47s eta=17:10:12 +2026-09-20 11:06:30 | INFO | train rank=0 | step=10340/50000 progress=20.7% loss=0.111039 lr=9.342748e-05 lr_backbone=9.342748e-05 elapsed=04:28:30 avg_step=1.56s recent_step=1.46s eta=17:09:52 +2026-09-20 11:06:47 | INFO | train rank=0 | step=10350/50000 progress=20.7% loss=0.106363 lr=9.341108e-05 lr_backbone=9.341108e-05 elapsed=04:28:46 avg_step=1.56s recent_step=1.63s eta=17:09:40 +2026-09-20 11:07:04 | INFO | train rank=0 | step=10360/50000 progress=20.7% loss=0.162532 lr=9.339466e-05 lr_backbone=9.339466e-05 elapsed=04:29:04 avg_step=1.56s recent_step=1.76s eta=17:09:32 +2026-09-20 11:07:20 | INFO | train rank=0 | step=10370/50000 progress=20.7% loss=0.107227 lr=9.337823e-05 lr_backbone=9.337823e-05 elapsed=04:29:20 avg_step=1.56s recent_step=1.62s eta=17:09:19 +2026-09-20 11:07:33 | INFO | train rank=0 | step=10380/50000 progress=20.8% loss=0.099621 lr=9.336177e-05 lr_backbone=9.336177e-05 elapsed=04:29:32 avg_step=1.56s recent_step=1.23s eta=17:08:51 +2026-09-20 11:07:49 | INFO | train rank=0 | step=10390/50000 progress=20.8% loss=0.105982 lr=9.334530e-05 lr_backbone=9.334530e-05 elapsed=04:29:49 avg_step=1.56s recent_step=1.66s eta=17:08:39 +2026-09-20 11:08:01 | INFO | train rank=0 | step=10400/50000 progress=20.8% loss=0.100838 lr=9.332880e-05 lr_backbone=9.332880e-05 elapsed=04:30:01 avg_step=1.56s recent_step=1.20s eta=17:08:10 +2026-09-20 11:08:17 | INFO | train rank=0 | step=10410/50000 progress=20.8% loss=0.091186 lr=9.331229e-05 lr_backbone=9.331229e-05 elapsed=04:30:17 avg_step=1.56s recent_step=1.58s eta=17:07:55 +2026-09-20 11:08:34 | INFO | train rank=0 | step=10420/50000 progress=20.8% loss=0.121672 lr=9.329576e-05 lr_backbone=9.329576e-05 elapsed=04:30:34 avg_step=1.56s recent_step=1.70s eta=17:07:45 +2026-09-20 11:08:54 | INFO | train rank=0 | step=10430/50000 progress=20.9% loss=0.116779 lr=9.327921e-05 lr_backbone=9.327921e-05 elapsed=04:30:54 avg_step=1.56s recent_step=1.99s eta=17:07:46 +2026-09-20 11:09:06 | INFO | train rank=0 | step=10440/50000 progress=20.9% loss=0.108387 lr=9.326264e-05 lr_backbone=9.326264e-05 elapsed=04:31:06 avg_step=1.56s recent_step=1.18s eta=17:07:16 +2026-09-20 11:09:23 | INFO | train rank=0 | step=10450/50000 progress=20.9% loss=0.091476 lr=9.324605e-05 lr_backbone=9.324605e-05 elapsed=04:31:23 avg_step=1.56s recent_step=1.70s eta=17:07:06 +2026-09-20 11:09:37 | INFO | train rank=0 | step=10460/50000 progress=20.9% loss=0.088762 lr=9.322944e-05 lr_backbone=9.322944e-05 elapsed=04:31:37 avg_step=1.56s recent_step=1.43s eta=17:06:45 +2026-09-20 11:09:52 | INFO | train rank=0 | step=10470/50000 progress=20.9% loss=0.118694 lr=9.321282e-05 lr_backbone=9.321282e-05 elapsed=04:31:52 avg_step=1.56s recent_step=1.52s eta=17:06:28 +2026-09-20 11:10:09 | INFO | train rank=0 | step=10480/50000 progress=21.0% loss=0.102549 lr=9.319617e-05 lr_backbone=9.319617e-05 elapsed=04:32:08 avg_step=1.56s recent_step=1.62s eta=17:06:15 +2026-09-20 11:10:23 | INFO | train rank=0 | step=10490/50000 progress=21.0% loss=0.088375 lr=9.317951e-05 lr_backbone=9.317951e-05 elapsed=04:32:23 avg_step=1.56s recent_step=1.46s eta=17:05:56 +2026-09-20 11:10:39 | INFO | train rank=0 | step=10500/50000 progress=21.0% loss=0.131370 lr=9.316282e-05 lr_backbone=9.316282e-05 elapsed=04:32:39 avg_step=1.56s recent_step=1.57s eta=17:05:41 +2026-09-20 11:10:55 | INFO | train rank=0 | step=10510/50000 progress=21.0% loss=0.137510 lr=9.314612e-05 lr_backbone=9.314612e-05 elapsed=04:32:55 avg_step=1.56s recent_step=1.61s eta=17:05:27 +2026-09-20 11:11:07 | INFO | train rank=0 | step=10520/50000 progress=21.0% loss=0.089275 lr=9.312940e-05 lr_backbone=9.312940e-05 elapsed=04:33:07 avg_step=1.56s recent_step=1.21s eta=17:04:58 +2026-09-20 11:11:23 | INFO | train rank=0 | step=10530/50000 progress=21.1% loss=0.111938 lr=9.311266e-05 lr_backbone=9.311266e-05 elapsed=04:33:23 avg_step=1.56s recent_step=1.59s eta=17:04:44 +2026-09-20 11:11:35 | INFO | train rank=0 | step=10540/50000 progress=21.1% loss=0.109223 lr=9.309590e-05 lr_backbone=9.309590e-05 elapsed=04:33:35 avg_step=1.56s recent_step=1.21s eta=17:04:15 +2026-09-20 11:11:54 | INFO | train rank=0 | step=10550/50000 progress=21.1% loss=0.107571 lr=9.307913e-05 lr_backbone=9.307913e-05 elapsed=04:33:53 avg_step=1.56s recent_step=1.85s eta=17:04:11 +2026-09-20 11:12:11 | INFO | train rank=0 | step=10560/50000 progress=21.1% loss=0.076081 lr=9.306233e-05 lr_backbone=9.306233e-05 elapsed=04:34:11 avg_step=1.56s recent_step=1.75s eta=17:04:02 +2026-09-20 11:12:23 | INFO | train rank=0 | step=10570/50000 progress=21.1% loss=0.097978 lr=9.304552e-05 lr_backbone=9.304552e-05 elapsed=04:34:23 avg_step=1.56s recent_step=1.22s eta=17:03:34 +2026-09-20 11:12:40 | INFO | train rank=0 | step=10580/50000 progress=21.2% loss=0.090039 lr=9.302868e-05 lr_backbone=9.302868e-05 elapsed=04:34:39 avg_step=1.56s recent_step=1.65s eta=17:03:22 +2026-09-20 11:12:52 | INFO | train rank=0 | step=10590/50000 progress=21.2% loss=0.108387 lr=9.301183e-05 lr_backbone=9.301183e-05 elapsed=04:34:51 avg_step=1.56s recent_step=1.20s eta=17:02:53 +2026-09-20 11:13:13 | INFO | train rank=0 | step=10600/50000 progress=21.2% loss=0.090620 lr=9.299496e-05 lr_backbone=9.299496e-05 elapsed=04:35:13 avg_step=1.56s recent_step=2.12s eta=17:02:58 +2026-09-20 11:13:31 | INFO | train rank=0 | step=10610/50000 progress=21.2% loss=0.112743 lr=9.297807e-05 lr_backbone=9.297807e-05 elapsed=04:35:31 avg_step=1.56s recent_step=1.85s eta=17:02:54 +2026-09-20 11:13:48 | INFO | train rank=0 | step=10620/50000 progress=21.2% loss=0.100694 lr=9.296116e-05 lr_backbone=9.296116e-05 elapsed=04:35:47 avg_step=1.56s recent_step=1.64s eta=17:02:41 +2026-09-20 11:14:44 | INFO | train rank=0 | step=10630/50000 progress=21.3% loss=0.107690 lr=9.294423e-05 lr_backbone=9.294423e-05 elapsed=04:36:44 avg_step=1.56s recent_step=5.61s eta=17:04:55 +2026-09-20 11:14:55 | INFO | train rank=0 | step=10640/50000 progress=21.3% loss=0.115454 lr=9.292728e-05 lr_backbone=9.292728e-05 elapsed=04:36:55 avg_step=1.56s recent_step=1.14s eta=17:04:24 +2026-09-20 11:15:07 | INFO | train rank=0 | step=10650/50000 progress=21.3% loss=0.137405 lr=9.291032e-05 lr_backbone=9.291032e-05 elapsed=04:37:07 avg_step=1.56s recent_step=1.21s eta=17:03:55 +2026-09-20 11:15:23 | INFO | train rank=0 | step=10660/50000 progress=21.3% loss=0.100081 lr=9.289334e-05 lr_backbone=9.289334e-05 elapsed=04:37:23 avg_step=1.56s recent_step=1.59s eta=17:03:41 +2026-09-20 11:15:40 | INFO | train rank=0 | step=10670/50000 progress=21.3% loss=0.110004 lr=9.287633e-05 lr_backbone=9.287633e-05 elapsed=04:37:40 avg_step=1.56s recent_step=1.69s eta=17:03:30 +2026-09-20 11:15:58 | INFO | train rank=0 | step=10680/50000 progress=21.4% loss=0.120416 lr=9.285931e-05 lr_backbone=9.285931e-05 elapsed=04:37:58 avg_step=1.56s recent_step=1.81s eta=17:03:23 +2026-09-20 11:16:10 | INFO | train rank=0 | step=10690/50000 progress=21.4% loss=0.102312 lr=9.284227e-05 lr_backbone=9.284227e-05 elapsed=04:38:10 avg_step=1.56s recent_step=1.21s eta=17:02:55 +2026-09-20 11:16:26 | INFO | train rank=0 | step=10700/50000 progress=21.4% loss=0.122450 lr=9.282521e-05 lr_backbone=9.282521e-05 elapsed=04:38:26 avg_step=1.56s recent_step=1.56s eta=17:02:39 +2026-09-20 11:16:43 | INFO | train rank=0 | step=10710/50000 progress=21.4% loss=0.098155 lr=9.280813e-05 lr_backbone=9.280813e-05 elapsed=04:38:43 avg_step=1.56s recent_step=1.73s eta=17:02:30 +2026-09-20 11:16:56 | INFO | train rank=0 | step=10720/50000 progress=21.4% loss=0.093254 lr=9.279104e-05 lr_backbone=9.279104e-05 elapsed=04:38:55 avg_step=1.56s recent_step=1.23s eta=17:02:02 +2026-09-20 11:17:12 | INFO | train rank=0 | step=10730/50000 progress=21.5% loss=0.110523 lr=9.277392e-05 lr_backbone=9.277392e-05 elapsed=04:39:11 avg_step=1.56s recent_step=1.63s eta=17:01:49 +2026-09-20 11:17:27 | INFO | train rank=0 | step=10740/50000 progress=21.5% loss=0.091290 lr=9.275679e-05 lr_backbone=9.275679e-05 elapsed=04:39:27 avg_step=1.56s recent_step=1.52s eta=17:01:32 +2026-09-20 11:17:46 | INFO | train rank=0 | step=10750/50000 progress=21.5% loss=0.102307 lr=9.273964e-05 lr_backbone=9.273964e-05 elapsed=04:39:46 avg_step=1.56s recent_step=1.91s eta=17:01:29 +2026-09-20 11:17:59 | INFO | train rank=0 | step=10760/50000 progress=21.5% loss=0.117936 lr=9.272246e-05 lr_backbone=9.272246e-05 elapsed=04:39:58 avg_step=1.56s recent_step=1.27s eta=17:01:02 +2026-09-20 11:18:14 | INFO | train rank=0 | step=10770/50000 progress=21.5% loss=0.089486 lr=9.270527e-05 lr_backbone=9.270527e-05 elapsed=04:40:14 avg_step=1.56s recent_step=1.52s eta=17:00:45 +2026-09-20 11:18:31 | INFO | train rank=0 | step=10780/50000 progress=21.6% loss=0.120805 lr=9.268807e-05 lr_backbone=9.268807e-05 elapsed=04:40:30 avg_step=1.56s recent_step=1.66s eta=17:00:33 +2026-09-20 11:18:51 | INFO | train rank=0 | step=10790/50000 progress=21.6% loss=0.126140 lr=9.267084e-05 lr_backbone=9.267084e-05 elapsed=04:40:50 avg_step=1.56s recent_step=2.00s eta=17:00:33 +2026-09-20 11:19:06 | INFO | train rank=0 | step=10800/50000 progress=21.6% loss=0.101676 lr=9.265359e-05 lr_backbone=9.265359e-05 elapsed=04:41:05 avg_step=1.56s recent_step=1.53s eta=17:00:17 +2026-09-20 11:19:20 | INFO | train rank=0 | step=10810/50000 progress=21.6% loss=0.093176 lr=9.263633e-05 lr_backbone=9.263633e-05 elapsed=04:41:19 avg_step=1.56s recent_step=1.39s eta=16:59:55 +2026-09-20 11:19:34 | INFO | train rank=0 | step=10820/50000 progress=21.6% loss=0.096350 lr=9.261904e-05 lr_backbone=9.261904e-05 elapsed=04:41:34 avg_step=1.56s recent_step=1.45s eta=16:59:35 +2026-09-20 11:19:54 | INFO | train rank=0 | step=10830/50000 progress=21.7% loss=0.133423 lr=9.260174e-05 lr_backbone=9.260174e-05 elapsed=04:41:54 avg_step=1.56s recent_step=1.98s eta=16:59:35 +2026-09-20 11:20:06 | INFO | train rank=0 | step=10840/50000 progress=21.7% loss=0.090446 lr=9.258442e-05 lr_backbone=9.258442e-05 elapsed=04:42:06 avg_step=1.56s recent_step=1.23s eta=16:59:07 +2026-09-20 11:20:24 | INFO | train rank=0 | step=10850/50000 progress=21.7% loss=0.094894 lr=9.256708e-05 lr_backbone=9.256708e-05 elapsed=04:42:23 avg_step=1.56s recent_step=1.74s eta=16:58:58 +2026-09-20 11:20:38 | INFO | train rank=0 | step=10860/50000 progress=21.7% loss=0.098780 lr=9.254972e-05 lr_backbone=9.254972e-05 elapsed=04:42:37 avg_step=1.56s recent_step=1.40s eta=16:58:37 +2026-09-20 11:20:52 | INFO | train rank=0 | step=10870/50000 progress=21.7% loss=0.073527 lr=9.253235e-05 lr_backbone=9.253235e-05 elapsed=04:42:52 avg_step=1.56s recent_step=1.43s eta=16:58:16 +2026-09-20 11:21:11 | INFO | train rank=0 | step=10880/50000 progress=21.8% loss=0.095652 lr=9.251495e-05 lr_backbone=9.251495e-05 elapsed=04:43:10 avg_step=1.56s recent_step=1.85s eta=16:58:11 +2026-09-20 11:21:25 | INFO | train rank=0 | step=10890/50000 progress=21.8% loss=0.093797 lr=9.249754e-05 lr_backbone=9.249754e-05 elapsed=04:43:25 avg_step=1.56s recent_step=1.45s eta=16:57:51 +2026-09-20 11:21:37 | INFO | train rank=0 | step=10900/50000 progress=21.8% loss=0.116416 lr=9.248011e-05 lr_backbone=9.248011e-05 elapsed=04:43:37 avg_step=1.56s recent_step=1.21s eta=16:57:23 +2026-09-20 11:21:49 | INFO | train rank=0 | step=10910/50000 progress=21.8% loss=0.128374 lr=9.246266e-05 lr_backbone=9.246266e-05 elapsed=04:43:48 avg_step=1.56s recent_step=1.17s eta=16:56:53 +2026-09-20 11:22:09 | INFO | train rank=0 | step=10920/50000 progress=21.8% loss=0.114417 lr=9.244519e-05 lr_backbone=9.244519e-05 elapsed=04:44:08 avg_step=1.56s recent_step=1.98s eta=16:56:53 +2026-09-20 11:22:23 | INFO | train rank=0 | step=10930/50000 progress=21.9% loss=0.083639 lr=9.242770e-05 lr_backbone=9.242770e-05 elapsed=04:44:23 avg_step=1.56s recent_step=1.43s eta=16:56:33 +2026-09-20 11:22:36 | INFO | train rank=0 | step=10940/50000 progress=21.9% loss=0.087285 lr=9.241019e-05 lr_backbone=9.241019e-05 elapsed=04:44:36 avg_step=1.56s recent_step=1.33s eta=16:56:09 +2026-09-20 11:22:55 | INFO | train rank=0 | step=10950/50000 progress=21.9% loss=0.099230 lr=9.239267e-05 lr_backbone=9.239267e-05 elapsed=04:44:55 avg_step=1.56s recent_step=1.90s eta=16:56:05 +2026-09-20 11:23:09 | INFO | train rank=0 | step=10960/50000 progress=21.9% loss=0.154212 lr=9.237512e-05 lr_backbone=9.237512e-05 elapsed=04:45:08 avg_step=1.56s recent_step=1.36s eta=16:55:42 +2026-09-20 11:23:21 | INFO | train rank=0 | step=10970/50000 progress=21.9% loss=0.100177 lr=9.235756e-05 lr_backbone=9.235756e-05 elapsed=04:45:21 avg_step=1.56s recent_step=1.23s eta=16:55:15 +2026-09-20 11:23:41 | INFO | train rank=0 | step=10980/50000 progress=22.0% loss=0.081930 lr=9.233998e-05 lr_backbone=9.233998e-05 elapsed=04:45:41 avg_step=1.56s recent_step=2.02s eta=16:55:16 +2026-09-20 11:23:53 | INFO | train rank=0 | step=10990/50000 progress=22.0% loss=0.091576 lr=9.232238e-05 lr_backbone=9.232238e-05 elapsed=04:45:53 avg_step=1.56s recent_step=1.17s eta=16:54:46 +2026-09-20 11:24:05 | INFO | train rank=0 | step=11000/50000 progress=22.0% loss=0.110182 lr=9.230476e-05 lr_backbone=9.230476e-05 elapsed=04:46:05 avg_step=1.56s recent_step=1.23s eta=16:54:19 +2026-09-20 11:24:19 | INFO | train rank=0 | step=11010/50000 progress=22.0% loss=0.101157 lr=9.228713e-05 lr_backbone=9.228713e-05 elapsed=04:46:19 avg_step=1.56s recent_step=1.36s eta=16:53:56 +2026-09-20 11:24:38 | INFO | train rank=0 | step=11020/50000 progress=22.0% loss=0.131401 lr=9.226947e-05 lr_backbone=9.226947e-05 elapsed=04:46:37 avg_step=1.56s recent_step=1.87s eta=16:53:52 +2026-09-20 11:24:53 | INFO | train rank=0 | step=11030/50000 progress=22.1% loss=0.097500 lr=9.225180e-05 lr_backbone=9.225180e-05 elapsed=04:46:53 avg_step=1.56s recent_step=1.57s eta=16:53:36 +2026-09-20 11:25:08 | INFO | train rank=0 | step=11040/50000 progress=22.1% loss=0.104727 lr=9.223411e-05 lr_backbone=9.223411e-05 elapsed=04:47:08 avg_step=1.56s recent_step=1.50s eta=16:53:19 +2026-09-20 11:25:26 | INFO | train rank=0 | step=11050/50000 progress=22.1% loss=0.106942 lr=9.221640e-05 lr_backbone=9.221640e-05 elapsed=04:47:25 avg_step=1.56s recent_step=1.73s eta=16:53:09 +2026-09-20 11:25:47 | INFO | train rank=0 | step=11060/50000 progress=22.1% loss=0.114195 lr=9.219867e-05 lr_backbone=9.219867e-05 elapsed=04:47:47 avg_step=1.56s recent_step=2.12s eta=16:53:13 +2026-09-20 11:26:08 | INFO | train rank=0 | step=11070/50000 progress=22.1% loss=0.116911 lr=9.218092e-05 lr_backbone=9.218092e-05 elapsed=04:48:08 avg_step=1.56s recent_step=2.12s eta=16:53:17 +2026-09-20 11:26:22 | INFO | train rank=0 | step=11080/50000 progress=22.2% loss=0.120863 lr=9.216315e-05 lr_backbone=9.216315e-05 elapsed=04:48:22 avg_step=1.56s recent_step=1.40s eta=16:52:56 +2026-09-20 11:27:18 | INFO | train rank=0 | step=11090/50000 progress=22.2% loss=0.097385 lr=9.214537e-05 lr_backbone=9.214537e-05 elapsed=04:49:18 avg_step=1.57s recent_step=5.63s eta=16:55:03 +2026-09-20 11:27:31 | INFO | train rank=0 | step=11100/50000 progress=22.2% loss=0.108459 lr=9.212757e-05 lr_backbone=9.212757e-05 elapsed=04:49:30 avg_step=1.56s recent_step=1.21s eta=16:54:35 +2026-09-20 11:27:42 | INFO | train rank=0 | step=11110/50000 progress=22.2% loss=0.110193 lr=9.210975e-05 lr_backbone=9.210975e-05 elapsed=04:49:42 avg_step=1.56s recent_step=1.19s eta=16:54:06 +2026-09-20 11:27:55 | INFO | train rank=0 | step=11120/50000 progress=22.2% loss=0.159652 lr=9.209191e-05 lr_backbone=9.209191e-05 elapsed=04:49:55 avg_step=1.56s recent_step=1.29s eta=16:53:41 +2026-09-20 11:28:16 | INFO | train rank=0 | step=11130/50000 progress=22.3% loss=0.117978 lr=9.207405e-05 lr_backbone=9.207405e-05 elapsed=04:50:16 avg_step=1.56s recent_step=2.08s eta=16:53:44 +2026-09-20 11:28:28 | INFO | train rank=0 | step=11140/50000 progress=22.3% loss=0.095639 lr=9.205617e-05 lr_backbone=9.205617e-05 elapsed=04:50:28 avg_step=1.56s recent_step=1.22s eta=16:53:16 +2026-09-20 11:28:42 | INFO | train rank=0 | step=11150/50000 progress=22.3% loss=0.105393 lr=9.203828e-05 lr_backbone=9.203828e-05 elapsed=04:50:41 avg_step=1.56s recent_step=1.32s eta=16:52:52 +2026-09-20 11:28:57 | INFO | train rank=0 | step=11160/50000 progress=22.3% loss=0.088138 lr=9.202037e-05 lr_backbone=9.202037e-05 elapsed=04:50:56 avg_step=1.56s recent_step=1.51s eta=16:52:34 +2026-09-20 11:29:16 | INFO | train rank=0 | step=11170/50000 progress=22.3% loss=0.119116 lr=9.200244e-05 lr_backbone=9.200244e-05 elapsed=04:51:15 avg_step=1.56s recent_step=1.90s eta=16:52:30 +2026-09-20 11:29:28 | INFO | train rank=0 | step=11180/50000 progress=22.4% loss=0.103619 lr=9.198449e-05 lr_backbone=9.198449e-05 elapsed=04:51:27 avg_step=1.56s recent_step=1.21s eta=16:52:02 +2026-09-20 11:29:42 | INFO | train rank=0 | step=11190/50000 progress=22.4% loss=0.103179 lr=9.196652e-05 lr_backbone=9.196652e-05 elapsed=04:51:42 avg_step=1.56s recent_step=1.42s eta=16:51:41 +2026-09-20 11:29:59 | INFO | train rank=0 | step=11200/50000 progress=22.4% loss=0.124909 lr=9.194853e-05 lr_backbone=9.194853e-05 elapsed=04:51:59 avg_step=1.56s recent_step=1.71s eta=16:51:31 +2026-09-20 11:30:19 | INFO | train rank=0 | step=11210/50000 progress=22.4% loss=0.090933 lr=9.193053e-05 lr_backbone=9.193053e-05 elapsed=04:52:18 avg_step=1.56s recent_step=1.95s eta=16:51:29 +2026-09-20 11:30:31 | INFO | train rank=0 | step=11220/50000 progress=22.4% loss=0.104456 lr=9.191250e-05 lr_backbone=9.191250e-05 elapsed=04:52:31 avg_step=1.56s recent_step=1.29s eta=16:51:04 +2026-09-20 11:30:49 | INFO | train rank=0 | step=11230/50000 progress=22.5% loss=0.110530 lr=9.189446e-05 lr_backbone=9.189446e-05 elapsed=04:52:49 avg_step=1.56s recent_step=1.78s eta=16:50:55 +2026-09-20 11:31:01 | INFO | train rank=0 | step=11240/50000 progress=22.5% loss=0.141410 lr=9.187640e-05 lr_backbone=9.187640e-05 elapsed=04:53:01 avg_step=1.56s recent_step=1.18s eta=16:50:26 +2026-09-20 11:31:19 | INFO | train rank=0 | step=11250/50000 progress=22.5% loss=0.102418 lr=9.185832e-05 lr_backbone=9.185832e-05 elapsed=04:53:19 avg_step=1.56s recent_step=1.85s eta=16:50:20 +2026-09-20 11:31:37 | INFO | train rank=0 | step=11260/50000 progress=22.5% loss=0.108197 lr=9.184023e-05 lr_backbone=9.184023e-05 elapsed=04:53:37 avg_step=1.56s recent_step=1.77s eta=16:50:12 +2026-09-20 11:31:51 | INFO | train rank=0 | step=11270/50000 progress=22.5% loss=0.082482 lr=9.182211e-05 lr_backbone=9.182211e-05 elapsed=04:53:50 avg_step=1.56s recent_step=1.35s eta=16:49:49 +2026-09-20 11:32:11 | INFO | train rank=0 | step=11280/50000 progress=22.6% loss=0.088592 lr=9.180398e-05 lr_backbone=9.180398e-05 elapsed=04:54:11 avg_step=1.56s recent_step=2.07s eta=16:49:51 +2026-09-20 11:32:29 | INFO | train rank=0 | step=11290/50000 progress=22.6% loss=0.088839 lr=9.178583e-05 lr_backbone=9.178583e-05 elapsed=04:54:29 avg_step=1.57s recent_step=1.79s eta=16:49:43 +2026-09-20 11:32:42 | INFO | train rank=0 | step=11300/50000 progress=22.6% loss=0.092170 lr=9.176766e-05 lr_backbone=9.176766e-05 elapsed=04:54:42 avg_step=1.56s recent_step=1.29s eta=16:49:18 +2026-09-20 11:32:58 | INFO | train rank=0 | step=11310/50000 progress=22.6% loss=0.115276 lr=9.174947e-05 lr_backbone=9.174947e-05 elapsed=04:54:58 avg_step=1.56s recent_step=1.58s eta=16:49:02 +2026-09-20 11:33:20 | INFO | train rank=0 | step=11320/50000 progress=22.6% loss=0.108556 lr=9.173127e-05 lr_backbone=9.173127e-05 elapsed=04:55:19 avg_step=1.57s recent_step=2.18s eta=16:49:07 +2026-09-20 11:33:31 | INFO | train rank=0 | step=11330/50000 progress=22.7% loss=0.091738 lr=9.171304e-05 lr_backbone=9.171304e-05 elapsed=04:55:31 avg_step=1.57s recent_step=1.16s eta=16:48:38 +2026-09-20 11:33:43 | INFO | train rank=0 | step=11340/50000 progress=22.7% loss=0.113907 lr=9.169480e-05 lr_backbone=9.169480e-05 elapsed=04:55:43 avg_step=1.56s recent_step=1.21s eta=16:48:10 +2026-09-20 11:33:58 | INFO | train rank=0 | step=11350/50000 progress=22.7% loss=0.148607 lr=9.167654e-05 lr_backbone=9.167654e-05 elapsed=04:55:58 avg_step=1.56s recent_step=1.45s eta=16:47:51 +2026-09-20 11:34:21 | INFO | train rank=0 | step=11360/50000 progress=22.7% loss=0.141865 lr=9.165826e-05 lr_backbone=9.165826e-05 elapsed=04:56:20 avg_step=1.57s recent_step=2.26s eta=16:47:59 +2026-09-20 11:34:33 | INFO | train rank=0 | step=11370/50000 progress=22.7% loss=0.092411 lr=9.163996e-05 lr_backbone=9.163996e-05 elapsed=04:56:33 avg_step=1.56s recent_step=1.27s eta=16:47:33 +2026-09-20 11:34:45 | INFO | train rank=0 | step=11380/50000 progress=22.8% loss=0.108221 lr=9.162165e-05 lr_backbone=9.162165e-05 elapsed=04:56:45 avg_step=1.56s recent_step=1.18s eta=16:47:04 +2026-09-20 11:34:59 | INFO | train rank=0 | step=11390/50000 progress=22.8% loss=0.112635 lr=9.160331e-05 lr_backbone=9.160331e-05 elapsed=04:56:59 avg_step=1.56s recent_step=1.42s eta=16:46:44 +2026-09-20 11:35:18 | INFO | train rank=0 | step=11400/50000 progress=22.8% loss=0.078119 lr=9.158496e-05 lr_backbone=9.158496e-05 elapsed=04:57:18 avg_step=1.56s recent_step=1.88s eta=16:46:39 +2026-09-20 11:35:30 | INFO | train rank=0 | step=11410/50000 progress=22.8% loss=0.195255 lr=9.156659e-05 lr_backbone=9.156659e-05 elapsed=04:57:30 avg_step=1.56s recent_step=1.22s eta=16:46:12 +2026-09-20 11:35:46 | INFO | train rank=0 | step=11420/50000 progress=22.8% loss=0.125148 lr=9.154820e-05 lr_backbone=9.154820e-05 elapsed=04:57:45 avg_step=1.56s recent_step=1.54s eta=16:45:55 +2026-09-20 11:36:03 | INFO | train rank=0 | step=11430/50000 progress=22.9% loss=0.120766 lr=9.152979e-05 lr_backbone=9.152979e-05 elapsed=04:58:03 avg_step=1.56s recent_step=1.75s eta=16:45:46 +2026-09-20 11:36:17 | INFO | train rank=0 | step=11440/50000 progress=22.9% loss=0.104123 lr=9.151137e-05 lr_backbone=9.151137e-05 elapsed=04:58:17 avg_step=1.56s recent_step=1.42s eta=16:45:25 +2026-09-20 11:36:33 | INFO | train rank=0 | step=11450/50000 progress=22.9% loss=0.093161 lr=9.149293e-05 lr_backbone=9.149293e-05 elapsed=04:58:33 avg_step=1.56s recent_step=1.60s eta=16:45:11 +2026-09-20 11:36:47 | INFO | train rank=0 | step=11460/50000 progress=22.9% loss=0.135352 lr=9.147447e-05 lr_backbone=9.147447e-05 elapsed=04:58:47 avg_step=1.56s recent_step=1.36s eta=16:44:49 +2026-09-20 11:37:00 | INFO | train rank=0 | step=11470/50000 progress=22.9% loss=0.122591 lr=9.145599e-05 lr_backbone=9.145599e-05 elapsed=04:59:00 avg_step=1.56s recent_step=1.30s eta=16:44:24 +2026-09-20 11:37:18 | INFO | train rank=0 | step=11480/50000 progress=23.0% loss=0.123096 lr=9.143749e-05 lr_backbone=9.143749e-05 elapsed=04:59:18 avg_step=1.56s recent_step=1.83s eta=16:44:17 +2026-09-20 11:37:36 | INFO | train rank=0 | step=11490/50000 progress=23.0% loss=0.111321 lr=9.141898e-05 lr_backbone=9.141898e-05 elapsed=04:59:35 avg_step=1.56s recent_step=1.73s eta=16:44:07 +2026-09-20 11:37:47 | INFO | train rank=0 | step=11500/50000 progress=23.0% loss=0.108376 lr=9.140044e-05 lr_backbone=9.140044e-05 elapsed=04:59:47 avg_step=1.56s recent_step=1.14s eta=16:43:37 +2026-09-20 11:38:09 | INFO | train rank=0 | step=11510/50000 progress=23.0% loss=0.101022 lr=9.138189e-05 lr_backbone=9.138189e-05 elapsed=05:00:08 avg_step=1.56s recent_step=2.17s eta=16:43:42 +2026-09-20 11:38:36 | INFO | train rank=0 | step=11520/50000 progress=23.0% loss=0.102738 lr=9.136332e-05 lr_backbone=9.136332e-05 elapsed=05:00:36 avg_step=1.57s recent_step=2.72s eta=16:44:05 +2026-09-20 11:38:48 | INFO | train rank=0 | step=11530/50000 progress=23.1% loss=0.098454 lr=9.134473e-05 lr_backbone=9.134473e-05 elapsed=05:00:48 avg_step=1.57s recent_step=1.20s eta=16:43:37 +2026-09-20 11:39:40 | INFO | train rank=0 | step=11540/50000 progress=23.1% loss=0.091292 lr=9.132613e-05 lr_backbone=9.132613e-05 elapsed=05:01:39 avg_step=1.57s recent_step=5.19s eta=16:45:22 +2026-09-20 11:39:51 | INFO | train rank=0 | step=11550/50000 progress=23.1% loss=0.092157 lr=9.130750e-05 lr_backbone=9.130750e-05 elapsed=05:01:51 avg_step=1.57s recent_step=1.16s eta=16:44:53 +2026-09-20 11:40:07 | INFO | train rank=0 | step=11560/50000 progress=23.1% loss=0.106147 lr=9.128886e-05 lr_backbone=9.128886e-05 elapsed=05:02:06 avg_step=1.57s recent_step=1.52s eta=16:44:36 +2026-09-20 11:40:20 | INFO | train rank=0 | step=11570/50000 progress=23.1% loss=0.105466 lr=9.127020e-05 lr_backbone=9.127020e-05 elapsed=05:02:20 avg_step=1.57s recent_step=1.36s eta=16:44:13 +2026-09-20 11:40:38 | INFO | train rank=0 | step=11580/50000 progress=23.2% loss=0.116224 lr=9.125152e-05 lr_backbone=9.125152e-05 elapsed=05:02:37 avg_step=1.57s recent_step=1.73s eta=16:44:03 +2026-09-20 11:40:50 | INFO | train rank=0 | step=11590/50000 progress=23.2% loss=0.098159 lr=9.123283e-05 lr_backbone=9.123283e-05 elapsed=05:02:49 avg_step=1.57s recent_step=1.21s eta=16:43:35 +2026-09-20 11:41:04 | INFO | train rank=0 | step=11600/50000 progress=23.2% loss=0.082909 lr=9.121411e-05 lr_backbone=9.121411e-05 elapsed=05:03:03 avg_step=1.57s recent_step=1.42s eta=16:43:15 +2026-09-20 11:41:24 | INFO | train rank=0 | step=11610/50000 progress=23.2% loss=0.118836 lr=9.119538e-05 lr_backbone=9.119538e-05 elapsed=05:03:24 avg_step=1.57s recent_step=2.02s eta=16:43:14 +2026-09-20 11:41:37 | INFO | train rank=0 | step=11620/50000 progress=23.2% loss=0.109151 lr=9.117663e-05 lr_backbone=9.117663e-05 elapsed=05:03:37 avg_step=1.57s recent_step=1.33s eta=16:42:51 +2026-09-20 11:41:51 | INFO | train rank=0 | step=11630/50000 progress=23.3% loss=0.093521 lr=9.115786e-05 lr_backbone=9.115786e-05 elapsed=05:03:51 avg_step=1.57s recent_step=1.40s eta=16:42:29 +2026-09-20 11:42:07 | INFO | train rank=0 | step=11640/50000 progress=23.3% loss=0.081234 lr=9.113908e-05 lr_backbone=9.113908e-05 elapsed=05:04:06 avg_step=1.57s recent_step=1.54s eta=16:42:13 +2026-09-20 11:42:23 | INFO | train rank=0 | step=11650/50000 progress=23.3% loss=0.101190 lr=9.112027e-05 lr_backbone=9.112027e-05 elapsed=05:04:23 avg_step=1.57s recent_step=1.64s eta=16:41:59 +2026-09-20 11:42:36 | INFO | train rank=0 | step=11660/50000 progress=23.3% loss=0.101893 lr=9.110145e-05 lr_backbone=9.110145e-05 elapsed=05:04:35 avg_step=1.57s recent_step=1.27s eta=16:41:34 +2026-09-20 11:42:48 | INFO | train rank=0 | step=11670/50000 progress=23.3% loss=0.177253 lr=9.108261e-05 lr_backbone=9.108261e-05 elapsed=05:04:48 avg_step=1.57s recent_step=1.26s eta=16:41:08 +2026-09-20 11:43:03 | INFO | train rank=0 | step=11680/50000 progress=23.4% loss=0.107306 lr=9.106375e-05 lr_backbone=9.106375e-05 elapsed=05:05:02 avg_step=1.57s recent_step=1.41s eta=16:40:47 +2026-09-20 11:43:22 | INFO | train rank=0 | step=11690/50000 progress=23.4% loss=0.106087 lr=9.104487e-05 lr_backbone=9.104487e-05 elapsed=05:05:22 avg_step=1.57s recent_step=1.98s eta=16:40:45 +2026-09-20 11:43:35 | INFO | train rank=0 | step=11700/50000 progress=23.4% loss=0.110263 lr=9.102598e-05 lr_backbone=9.102598e-05 elapsed=05:05:34 avg_step=1.57s recent_step=1.23s eta=16:40:18 +2026-09-20 11:43:51 | INFO | train rank=0 | step=11710/50000 progress=23.4% loss=0.096734 lr=9.100707e-05 lr_backbone=9.100707e-05 elapsed=05:05:50 avg_step=1.57s recent_step=1.60s eta=16:40:04 +2026-09-20 11:44:06 | INFO | train rank=0 | step=11720/50000 progress=23.4% loss=0.101124 lr=9.098814e-05 lr_backbone=9.098814e-05 elapsed=05:06:06 avg_step=1.57s recent_step=1.57s eta=16:39:48 +2026-09-20 11:44:26 | INFO | train rank=0 | step=11730/50000 progress=23.5% loss=0.137486 lr=9.096919e-05 lr_backbone=9.096919e-05 elapsed=05:06:26 avg_step=1.57s recent_step=1.98s eta=16:39:46 +2026-09-20 11:44:38 | INFO | train rank=0 | step=11740/50000 progress=23.5% loss=0.182350 lr=9.095022e-05 lr_backbone=9.095022e-05 elapsed=05:06:37 avg_step=1.57s recent_step=1.17s eta=16:39:17 +2026-09-20 11:45:00 | INFO | train rank=0 | step=11750/50000 progress=23.5% loss=0.107668 lr=9.093124e-05 lr_backbone=9.093124e-05 elapsed=05:06:59 avg_step=1.57s recent_step=2.21s eta=16:39:22 +2026-09-20 11:45:12 | INFO | train rank=0 | step=11760/50000 progress=23.5% loss=0.128088 lr=9.091224e-05 lr_backbone=9.091224e-05 elapsed=05:07:12 avg_step=1.57s recent_step=1.24s eta=16:38:56 +2026-09-20 11:45:23 | INFO | train rank=0 | step=11770/50000 progress=23.5% loss=0.092036 lr=9.089322e-05 lr_backbone=9.089322e-05 elapsed=05:07:23 avg_step=1.57s recent_step=1.11s eta=16:38:26 +2026-09-20 11:45:44 | INFO | train rank=0 | step=11780/50000 progress=23.6% loss=0.148997 lr=9.087418e-05 lr_backbone=9.087418e-05 elapsed=05:07:43 avg_step=1.57s recent_step=2.04s eta=16:38:25 +2026-09-20 11:46:02 | INFO | train rank=0 | step=11790/50000 progress=23.6% loss=0.101746 lr=9.085513e-05 lr_backbone=9.085513e-05 elapsed=05:08:01 avg_step=1.57s recent_step=1.79s eta=16:38:17 +2026-09-20 11:46:18 | INFO | train rank=0 | step=11800/50000 progress=23.6% loss=0.131407 lr=9.083605e-05 lr_backbone=9.083605e-05 elapsed=05:08:18 avg_step=1.57s recent_step=1.64s eta=16:38:04 +2026-09-20 11:46:34 | INFO | train rank=0 | step=11810/50000 progress=23.6% loss=0.090251 lr=9.081696e-05 lr_backbone=9.081696e-05 elapsed=05:08:33 avg_step=1.57s recent_step=1.56s eta=16:37:48 +2026-09-20 11:46:46 | INFO | train rank=0 | step=11820/50000 progress=23.6% loss=0.092878 lr=9.079785e-05 lr_backbone=9.079785e-05 elapsed=05:08:45 avg_step=1.57s recent_step=1.20s eta=16:37:20 +2026-09-20 11:47:07 | INFO | train rank=0 | step=11830/50000 progress=23.7% loss=0.115577 lr=9.077873e-05 lr_backbone=9.077873e-05 elapsed=05:09:06 avg_step=1.57s recent_step=2.09s eta=16:37:21 +2026-09-20 11:47:19 | INFO | train rank=0 | step=11840/50000 progress=23.7% loss=0.151276 lr=9.075958e-05 lr_backbone=9.075958e-05 elapsed=05:09:18 avg_step=1.57s recent_step=1.21s eta=16:36:54 +2026-09-20 11:47:33 | INFO | train rank=0 | step=11850/50000 progress=23.7% loss=0.088888 lr=9.074042e-05 lr_backbone=9.074042e-05 elapsed=05:09:32 avg_step=1.57s recent_step=1.39s eta=16:36:33 +2026-09-20 11:47:50 | INFO | train rank=0 | step=11860/50000 progress=23.7% loss=0.089618 lr=9.072124e-05 lr_backbone=9.072124e-05 elapsed=05:09:50 avg_step=1.57s recent_step=1.78s eta=16:36:24 +2026-09-20 11:48:05 | INFO | train rank=0 | step=11870/50000 progress=23.7% loss=0.107579 lr=9.070204e-05 lr_backbone=9.070204e-05 elapsed=05:10:05 avg_step=1.57s recent_step=1.46s eta=16:36:05 +2026-09-20 11:48:20 | INFO | train rank=0 | step=11880/50000 progress=23.8% loss=0.097808 lr=9.068283e-05 lr_backbone=9.068283e-05 elapsed=05:10:19 avg_step=1.57s recent_step=1.46s eta=16:35:46 +2026-09-20 11:48:36 | INFO | train rank=0 | step=11890/50000 progress=23.8% loss=0.102271 lr=9.066359e-05 lr_backbone=9.066359e-05 elapsed=05:10:36 avg_step=1.57s recent_step=1.67s eta=16:35:34 +2026-09-20 11:48:53 | INFO | train rank=0 | step=11900/50000 progress=23.8% loss=0.142331 lr=9.064434e-05 lr_backbone=9.064434e-05 elapsed=05:10:53 avg_step=1.57s recent_step=1.70s eta=16:35:22 +2026-09-20 11:49:09 | INFO | train rank=0 | step=11910/50000 progress=23.8% loss=0.087427 lr=9.062507e-05 lr_backbone=9.062507e-05 elapsed=05:11:09 avg_step=1.57s recent_step=1.60s eta=16:35:08 +2026-09-20 11:49:23 | INFO | train rank=0 | step=11920/50000 progress=23.8% loss=0.120842 lr=9.060578e-05 lr_backbone=9.060578e-05 elapsed=05:11:23 avg_step=1.57s recent_step=1.40s eta=16:34:46 +2026-09-20 11:49:39 | INFO | train rank=0 | step=11930/50000 progress=23.9% loss=0.110535 lr=9.058648e-05 lr_backbone=9.058648e-05 elapsed=05:11:38 avg_step=1.57s recent_step=1.53s eta=16:34:30 +2026-09-20 11:49:55 | INFO | train rank=0 | step=11940/50000 progress=23.9% loss=0.114205 lr=9.056716e-05 lr_backbone=9.056716e-05 elapsed=05:11:55 avg_step=1.57s recent_step=1.63s eta=16:34:16 +2026-09-20 11:50:08 | INFO | train rank=0 | step=11950/50000 progress=23.9% loss=0.070987 lr=9.054782e-05 lr_backbone=9.054782e-05 elapsed=05:12:07 avg_step=1.57s recent_step=1.27s eta=16:33:51 +2026-09-20 11:50:28 | INFO | train rank=0 | step=11960/50000 progress=23.9% loss=0.117505 lr=9.052846e-05 lr_backbone=9.052846e-05 elapsed=05:12:28 avg_step=1.57s recent_step=2.03s eta=16:33:50 +2026-09-20 11:50:46 | INFO | train rank=0 | step=11970/50000 progress=23.9% loss=0.090775 lr=9.050908e-05 lr_backbone=9.050908e-05 elapsed=05:12:46 avg_step=1.57s recent_step=1.83s eta=16:33:43 +2026-09-20 11:51:05 | INFO | train rank=0 | step=11980/50000 progress=24.0% loss=0.127780 lr=9.048969e-05 lr_backbone=9.048969e-05 elapsed=05:13:05 avg_step=1.57s recent_step=1.91s eta=16:33:38 +2026-09-20 11:51:59 | INFO | train rank=0 | step=11990/50000 progress=24.0% loss=0.107711 lr=9.047028e-05 lr_backbone=9.047028e-05 elapsed=05:13:58 avg_step=1.57s recent_step=5.32s eta=16:35:21 +2026-09-20 11:52:11 | INFO | train rank=0 | step=12000/50000 progress=24.0% loss=0.097803 lr=9.045085e-05 lr_backbone=9.045085e-05 elapsed=05:14:10 avg_step=1.57s recent_step=1.21s eta=16:34:54 +2026-09-20 11:52:22 | INFO | train rank=0 | step=12010/50000 progress=24.0% loss=0.108324 lr=9.043140e-05 lr_backbone=9.043140e-05 elapsed=05:14:22 avg_step=1.57s recent_step=1.17s eta=16:34:25 +2026-09-20 11:52:37 | INFO | train rank=0 | step=12020/50000 progress=24.0% loss=0.092584 lr=9.041194e-05 lr_backbone=9.041194e-05 elapsed=05:14:37 avg_step=1.57s recent_step=1.48s eta=16:34:07 +2026-09-20 11:52:51 | INFO | train rank=0 | step=12030/50000 progress=24.1% loss=0.172942 lr=9.039246e-05 lr_backbone=9.039246e-05 elapsed=05:14:51 avg_step=1.57s recent_step=1.41s eta=16:33:46 +2026-09-20 11:53:09 | INFO | train rank=0 | step=12040/50000 progress=24.1% loss=0.166289 lr=9.037296e-05 lr_backbone=9.037296e-05 elapsed=05:15:09 avg_step=1.57s recent_step=1.80s eta=16:33:38 +2026-09-20 11:53:22 | INFO | train rank=0 | step=12050/50000 progress=24.1% loss=0.104216 lr=9.035344e-05 lr_backbone=9.035344e-05 elapsed=05:15:21 avg_step=1.57s recent_step=1.23s eta=16:33:11 +2026-09-20 11:53:39 | INFO | train rank=0 | step=12060/50000 progress=24.1% loss=0.099914 lr=9.033391e-05 lr_backbone=9.033391e-05 elapsed=05:15:39 avg_step=1.57s recent_step=1.74s eta=16:33:01 +2026-09-20 11:53:52 | INFO | train rank=0 | step=12070/50000 progress=24.1% loss=0.095511 lr=9.031435e-05 lr_backbone=9.031435e-05 elapsed=05:15:52 avg_step=1.57s recent_step=1.34s eta=16:32:38 +2026-09-20 11:54:04 | INFO | train rank=0 | step=12080/50000 progress=24.2% loss=0.137275 lr=9.029478e-05 lr_backbone=9.029478e-05 elapsed=05:16:04 avg_step=1.57s recent_step=1.19s eta=16:32:10 +2026-09-20 11:54:19 | INFO | train rank=0 | step=12090/50000 progress=24.2% loss=0.108677 lr=9.027519e-05 lr_backbone=9.027519e-05 elapsed=05:16:18 avg_step=1.57s recent_step=1.44s eta=16:31:50 +2026-09-20 11:54:36 | INFO | train rank=0 | step=12100/50000 progress=24.2% loss=0.087304 lr=9.025559e-05 lr_backbone=9.025559e-05 elapsed=05:16:36 avg_step=1.57s recent_step=1.74s eta=16:31:40 +2026-09-20 11:54:51 | INFO | train rank=0 | step=12110/50000 progress=24.2% loss=0.125650 lr=9.023597e-05 lr_backbone=9.023597e-05 elapsed=05:16:51 avg_step=1.57s recent_step=1.50s eta=16:31:22 +2026-09-20 11:55:07 | INFO | train rank=0 | step=12120/50000 progress=24.2% loss=0.115870 lr=9.021633e-05 lr_backbone=9.021633e-05 elapsed=05:17:07 avg_step=1.57s recent_step=1.59s eta=16:31:07 +2026-09-20 11:55:21 | INFO | train rank=0 | step=12130/50000 progress=24.3% loss=0.087083 lr=9.019667e-05 lr_backbone=9.019667e-05 elapsed=05:17:21 avg_step=1.57s recent_step=1.45s eta=16:30:48 +2026-09-20 11:55:41 | INFO | train rank=0 | step=12140/50000 progress=24.3% loss=0.103470 lr=9.017699e-05 lr_backbone=9.017699e-05 elapsed=05:17:40 avg_step=1.57s recent_step=1.91s eta=16:30:42 +2026-09-20 11:55:56 | INFO | train rank=0 | step=12150/50000 progress=24.3% loss=0.091918 lr=9.015730e-05 lr_backbone=9.015730e-05 elapsed=05:17:56 avg_step=1.57s recent_step=1.55s eta=16:30:26 +2026-09-20 11:56:12 | INFO | train rank=0 | step=12160/50000 progress=24.3% loss=0.113611 lr=9.013759e-05 lr_backbone=9.013759e-05 elapsed=05:18:11 avg_step=1.57s recent_step=1.57s eta=16:30:10 +2026-09-20 11:56:28 | INFO | train rank=0 | step=12170/50000 progress=24.3% loss=0.127809 lr=9.011786e-05 lr_backbone=9.011786e-05 elapsed=05:18:28 avg_step=1.57s recent_step=1.62s eta=16:29:56 +2026-09-20 11:56:45 | INFO | train rank=0 | step=12180/50000 progress=24.4% loss=0.117423 lr=9.009811e-05 lr_backbone=9.009811e-05 elapsed=05:18:45 avg_step=1.57s recent_step=1.73s eta=16:29:46 +2026-09-20 11:56:58 | INFO | train rank=0 | step=12190/50000 progress=24.4% loss=0.086408 lr=9.007835e-05 lr_backbone=9.007835e-05 elapsed=05:18:58 avg_step=1.57s recent_step=1.29s eta=16:29:21 +2026-09-20 11:57:15 | INFO | train rank=0 | step=12200/50000 progress=24.4% loss=0.094055 lr=9.005857e-05 lr_backbone=9.005857e-05 elapsed=05:19:15 avg_step=1.57s recent_step=1.73s eta=16:29:10 +2026-09-20 11:57:32 | INFO | train rank=0 | step=12210/50000 progress=24.4% loss=0.091748 lr=9.003877e-05 lr_backbone=9.003877e-05 elapsed=05:19:32 avg_step=1.57s recent_step=1.66s eta=16:28:57 +2026-09-20 11:57:49 | INFO | train rank=0 | step=12220/50000 progress=24.4% loss=0.091938 lr=9.001895e-05 lr_backbone=9.001895e-05 elapsed=05:19:49 avg_step=1.57s recent_step=1.74s eta=16:28:47 +2026-09-20 11:58:06 | INFO | train rank=0 | step=12230/50000 progress=24.5% loss=0.102185 lr=8.999912e-05 lr_backbone=8.999912e-05 elapsed=05:20:06 avg_step=1.57s recent_step=1.69s eta=16:28:35 +2026-09-20 11:58:18 | INFO | train rank=0 | step=12240/50000 progress=24.5% loss=0.116647 lr=8.997927e-05 lr_backbone=8.997927e-05 elapsed=05:20:18 avg_step=1.57s recent_step=1.18s eta=16:28:07 +2026-09-20 11:58:38 | INFO | train rank=0 | step=12250/50000 progress=24.5% loss=0.169571 lr=8.995940e-05 lr_backbone=8.995940e-05 elapsed=05:20:38 avg_step=1.57s recent_step=2.02s eta=16:28:05 +2026-09-20 11:58:50 | INFO | train rank=0 | step=12260/50000 progress=24.5% loss=0.121601 lr=8.993951e-05 lr_backbone=8.993951e-05 elapsed=05:20:50 avg_step=1.57s recent_step=1.17s eta=16:27:37 +2026-09-20 11:59:09 | INFO | train rank=0 | step=12270/50000 progress=24.5% loss=0.098396 lr=8.991961e-05 lr_backbone=8.991961e-05 elapsed=05:21:09 avg_step=1.57s recent_step=1.89s eta=16:27:31 +2026-09-20 11:59:21 | INFO | train rank=0 | step=12280/50000 progress=24.6% loss=0.118708 lr=8.989969e-05 lr_backbone=8.989969e-05 elapsed=05:21:21 avg_step=1.57s recent_step=1.23s eta=16:27:05 +2026-09-20 11:59:36 | INFO | train rank=0 | step=12290/50000 progress=24.6% loss=0.095822 lr=8.987975e-05 lr_backbone=8.987975e-05 elapsed=05:21:36 avg_step=1.57s recent_step=1.52s eta=16:26:48 +2026-09-20 11:59:53 | INFO | train rank=0 | step=12300/50000 progress=24.6% loss=0.104758 lr=8.985979e-05 lr_backbone=8.985979e-05 elapsed=05:21:53 avg_step=1.57s recent_step=1.64s eta=16:26:35 +2026-09-20 12:00:10 | INFO | train rank=0 | step=12310/50000 progress=24.6% loss=0.078697 lr=8.983982e-05 lr_backbone=8.983982e-05 elapsed=05:22:10 avg_step=1.57s recent_step=1.70s eta=16:26:23 +2026-09-20 12:00:29 | INFO | train rank=0 | step=12320/50000 progress=24.6% loss=0.104944 lr=8.981983e-05 lr_backbone=8.981983e-05 elapsed=05:22:29 avg_step=1.57s recent_step=1.94s eta=16:26:19 +2026-09-20 12:00:41 | INFO | train rank=0 | step=12330/50000 progress=24.7% loss=0.121704 lr=8.979982e-05 lr_backbone=8.979982e-05 elapsed=05:22:40 avg_step=1.57s recent_step=1.14s eta=16:25:50 +2026-09-20 12:00:53 | INFO | train rank=0 | step=12340/50000 progress=24.7% loss=0.099153 lr=8.977980e-05 lr_backbone=8.977980e-05 elapsed=05:22:52 avg_step=1.57s recent_step=1.21s eta=16:25:23 +2026-09-20 12:01:06 | INFO | train rank=0 | step=12350/50000 progress=24.7% loss=0.097683 lr=8.975975e-05 lr_backbone=8.975975e-05 elapsed=05:23:05 avg_step=1.57s recent_step=1.29s eta=16:24:59 +2026-09-20 12:01:27 | INFO | train rank=0 | step=12360/50000 progress=24.7% loss=0.108036 lr=8.973969e-05 lr_backbone=8.973969e-05 elapsed=05:23:26 avg_step=1.57s recent_step=2.10s eta=16:24:59 +2026-09-20 12:01:39 | INFO | train rank=0 | step=12370/50000 progress=24.7% loss=0.197399 lr=8.971962e-05 lr_backbone=8.971962e-05 elapsed=05:23:39 avg_step=1.57s recent_step=1.23s eta=16:24:34 +2026-09-20 12:01:53 | INFO | train rank=0 | step=12380/50000 progress=24.8% loss=0.136286 lr=8.969952e-05 lr_backbone=8.969952e-05 elapsed=05:23:52 avg_step=1.57s recent_step=1.37s eta=16:24:12 +2026-09-20 12:02:07 | INFO | train rank=0 | step=12390/50000 progress=24.8% loss=0.103940 lr=8.967941e-05 lr_backbone=8.967941e-05 elapsed=05:24:07 avg_step=1.57s recent_step=1.43s eta=16:23:52 +2026-09-20 12:02:24 | INFO | train rank=0 | step=12400/50000 progress=24.8% loss=0.126504 lr=8.965928e-05 lr_backbone=8.965928e-05 elapsed=05:24:23 avg_step=1.57s recent_step=1.68s eta=16:23:39 +2026-09-20 12:02:40 | INFO | train rank=0 | step=12410/50000 progress=24.8% loss=0.118295 lr=8.963913e-05 lr_backbone=8.963913e-05 elapsed=05:24:39 avg_step=1.57s recent_step=1.59s eta=16:23:24 +2026-09-20 12:02:56 | INFO | train rank=0 | step=12420/50000 progress=24.8% loss=0.103374 lr=8.961897e-05 lr_backbone=8.961897e-05 elapsed=05:24:56 avg_step=1.57s recent_step=1.61s eta=16:23:10 +2026-09-20 12:03:09 | INFO | train rank=0 | step=12430/50000 progress=24.9% loss=0.099799 lr=8.959878e-05 lr_backbone=8.959878e-05 elapsed=05:25:09 avg_step=1.57s recent_step=1.35s eta=16:22:48 +2026-09-20 12:03:30 | INFO | train rank=0 | step=12440/50000 progress=24.9% loss=0.091006 lr=8.957858e-05 lr_backbone=8.957858e-05 elapsed=05:25:30 avg_step=1.57s recent_step=2.09s eta=16:22:48 +2026-09-20 12:03:47 | INFO | train rank=0 | step=12450/50000 progress=24.9% loss=0.141421 lr=8.955837e-05 lr_backbone=8.955837e-05 elapsed=05:25:47 avg_step=1.57s recent_step=1.71s eta=16:22:36 +2026-09-20 12:04:38 | INFO | train rank=0 | step=12460/50000 progress=24.9% loss=0.107239 lr=8.953813e-05 lr_backbone=8.953813e-05 elapsed=05:26:37 avg_step=1.57s recent_step=5.03s eta=16:24:05 +2026-09-20 12:04:49 | INFO | train rank=0 | step=12470/50000 progress=24.9% loss=0.083842 lr=8.951788e-05 lr_backbone=8.951788e-05 elapsed=05:26:49 avg_step=1.57s recent_step=1.11s eta=16:23:35 +2026-09-20 12:05:05 | INFO | train rank=0 | step=12480/50000 progress=25.0% loss=0.094374 lr=8.949761e-05 lr_backbone=8.949761e-05 elapsed=05:27:05 avg_step=1.57s recent_step=1.62s eta=16:23:21 +2026-09-20 12:05:17 | INFO | train rank=0 | step=12490/50000 progress=25.0% loss=0.119632 lr=8.947733e-05 lr_backbone=8.947733e-05 elapsed=05:27:17 avg_step=1.57s recent_step=1.20s eta=16:22:54 +2026-09-20 12:05:31 | INFO | train rank=0 | step=12500/50000 progress=25.0% loss=0.110041 lr=8.945703e-05 lr_backbone=8.945703e-05 elapsed=05:27:30 avg_step=1.57s recent_step=1.35s eta=16:22:31 +2026-09-20 12:05:43 | INFO | train rank=0 | step=12510/50000 progress=25.0% loss=0.123984 lr=8.943671e-05 lr_backbone=8.943671e-05 elapsed=05:27:43 avg_step=1.57s recent_step=1.28s eta=16:22:07 +2026-09-20 12:05:56 | INFO | train rank=0 | step=12520/50000 progress=25.0% loss=0.092186 lr=8.941637e-05 lr_backbone=8.941637e-05 elapsed=05:27:56 avg_step=1.57s recent_step=1.29s eta=16:21:43 +2026-09-20 12:06:08 | INFO | train rank=0 | step=12530/50000 progress=25.1% loss=0.090957 lr=8.939601e-05 lr_backbone=8.939601e-05 elapsed=05:28:08 avg_step=1.57s recent_step=1.19s eta=16:21:16 +2026-09-20 12:06:24 | INFO | train rank=0 | step=12540/50000 progress=25.1% loss=0.090665 lr=8.937564e-05 lr_backbone=8.937564e-05 elapsed=05:28:24 avg_step=1.57s recent_step=1.61s eta=16:21:01 +2026-09-20 12:06:43 | INFO | train rank=0 | step=12550/50000 progress=25.1% loss=0.102452 lr=8.935525e-05 lr_backbone=8.935525e-05 elapsed=05:28:42 avg_step=1.57s recent_step=1.83s eta=16:20:53 +2026-09-20 12:06:56 | INFO | train rank=0 | step=12560/50000 progress=25.1% loss=0.123514 lr=8.933485e-05 lr_backbone=8.933485e-05 elapsed=05:28:55 avg_step=1.57s recent_step=1.31s eta=16:20:30 +2026-09-20 12:07:14 | INFO | train rank=0 | step=12570/50000 progress=25.1% loss=0.120298 lr=8.931442e-05 lr_backbone=8.931442e-05 elapsed=05:29:13 avg_step=1.57s recent_step=1.78s eta=16:20:20 +2026-09-20 12:07:26 | INFO | train rank=0 | step=12580/50000 progress=25.2% loss=0.090520 lr=8.929398e-05 lr_backbone=8.929398e-05 elapsed=05:29:26 avg_step=1.57s recent_step=1.28s eta=16:19:56 +2026-09-20 12:07:44 | INFO | train rank=0 | step=12590/50000 progress=25.2% loss=0.089655 lr=8.927352e-05 lr_backbone=8.927352e-05 elapsed=05:29:43 avg_step=1.57s recent_step=1.74s eta=16:19:45 +2026-09-20 12:08:02 | INFO | train rank=0 | step=12600/50000 progress=25.2% loss=0.113013 lr=8.925305e-05 lr_backbone=8.925305e-05 elapsed=05:30:01 avg_step=1.57s recent_step=1.78s eta=16:19:36 +2026-09-20 12:08:18 | INFO | train rank=0 | step=12610/50000 progress=25.2% loss=0.145102 lr=8.923256e-05 lr_backbone=8.923256e-05 elapsed=05:30:18 avg_step=1.57s recent_step=1.64s eta=16:19:22 +2026-09-20 12:08:34 | INFO | train rank=0 | step=12620/50000 progress=25.2% loss=0.113208 lr=8.921205e-05 lr_backbone=8.921205e-05 elapsed=05:30:34 avg_step=1.57s recent_step=1.65s eta=16:19:09 +2026-09-20 12:08:56 | INFO | train rank=0 | step=12630/50000 progress=25.3% loss=0.087227 lr=8.919152e-05 lr_backbone=8.919152e-05 elapsed=05:30:56 avg_step=1.57s recent_step=2.19s eta=16:19:11 +2026-09-20 12:09:08 | INFO | train rank=0 | step=12640/50000 progress=25.3% loss=0.105430 lr=8.917098e-05 lr_backbone=8.917098e-05 elapsed=05:31:08 avg_step=1.57s recent_step=1.16s eta=16:18:44 +2026-09-20 12:09:25 | INFO | train rank=0 | step=12650/50000 progress=25.3% loss=0.085501 lr=8.915041e-05 lr_backbone=8.915041e-05 elapsed=05:31:25 avg_step=1.57s recent_step=1.72s eta=16:18:32 +2026-09-20 12:09:41 | INFO | train rank=0 | step=12660/50000 progress=25.3% loss=0.119143 lr=8.912984e-05 lr_backbone=8.912984e-05 elapsed=05:31:41 avg_step=1.57s recent_step=1.58s eta=16:18:17 +2026-09-20 12:09:56 | INFO | train rank=0 | step=12670/50000 progress=25.3% loss=0.099814 lr=8.910924e-05 lr_backbone=8.910924e-05 elapsed=05:31:56 avg_step=1.57s recent_step=1.54s eta=16:18:00 +2026-09-20 12:10:13 | INFO | train rank=0 | step=12680/50000 progress=25.4% loss=0.108096 lr=8.908863e-05 lr_backbone=8.908863e-05 elapsed=05:32:13 avg_step=1.57s recent_step=1.69s eta=16:17:48 +2026-09-20 12:10:32 | INFO | train rank=0 | step=12690/50000 progress=25.4% loss=0.173887 lr=8.906800e-05 lr_backbone=8.906800e-05 elapsed=05:32:31 avg_step=1.57s recent_step=1.84s eta=16:17:40 +2026-09-20 12:10:45 | INFO | train rank=0 | step=12700/50000 progress=25.4% loss=0.114839 lr=8.904735e-05 lr_backbone=8.904735e-05 elapsed=05:32:44 avg_step=1.57s recent_step=1.32s eta=16:17:17 +2026-09-20 12:11:03 | INFO | train rank=0 | step=12710/50000 progress=25.4% loss=0.075274 lr=8.902669e-05 lr_backbone=8.902669e-05 elapsed=05:33:02 avg_step=1.57s recent_step=1.79s eta=16:17:07 +2026-09-20 12:11:17 | INFO | train rank=0 | step=12720/50000 progress=25.4% loss=0.116151 lr=8.900601e-05 lr_backbone=8.900601e-05 elapsed=05:33:16 avg_step=1.57s recent_step=1.40s eta=16:16:47 +2026-09-20 12:11:32 | INFO | train rank=0 | step=12730/50000 progress=25.5% loss=0.122702 lr=8.898531e-05 lr_backbone=8.898531e-05 elapsed=05:33:31 avg_step=1.57s recent_step=1.49s eta=16:16:28 +2026-09-20 12:11:49 | INFO | train rank=0 | step=12740/50000 progress=25.5% loss=0.111610 lr=8.896460e-05 lr_backbone=8.896460e-05 elapsed=05:33:48 avg_step=1.57s recent_step=1.69s eta=16:16:16 +2026-09-20 12:12:00 | INFO | train rank=0 | step=12750/50000 progress=25.5% loss=0.096988 lr=8.894386e-05 lr_backbone=8.894386e-05 elapsed=05:34:00 avg_step=1.57s recent_step=1.16s eta=16:15:48 +2026-09-20 12:12:12 | INFO | train rank=0 | step=12760/50000 progress=25.5% loss=0.112746 lr=8.892312e-05 lr_backbone=8.892312e-05 elapsed=05:34:12 avg_step=1.57s recent_step=1.23s eta=16:15:23 +2026-09-20 12:12:30 | INFO | train rank=0 | step=12770/50000 progress=25.5% loss=0.090478 lr=8.890235e-05 lr_backbone=8.890235e-05 elapsed=05:34:30 avg_step=1.57s recent_step=1.76s eta=16:15:12 +2026-09-20 12:12:48 | INFO | train rank=0 | step=12780/50000 progress=25.6% loss=0.104806 lr=8.888157e-05 lr_backbone=8.888157e-05 elapsed=05:34:48 avg_step=1.57s recent_step=1.81s eta=16:15:04 +2026-09-20 12:13:01 | INFO | train rank=0 | step=12790/50000 progress=25.6% loss=0.095515 lr=8.886077e-05 lr_backbone=8.886077e-05 elapsed=05:35:01 avg_step=1.57s recent_step=1.31s eta=16:14:40 +2026-09-20 12:13:17 | INFO | train rank=0 | step=12800/50000 progress=25.6% loss=0.082242 lr=8.883995e-05 lr_backbone=8.883995e-05 elapsed=05:35:17 avg_step=1.57s recent_step=1.57s eta=16:14:25 +2026-09-20 12:13:35 | INFO | train rank=0 | step=12810/50000 progress=25.6% loss=0.110034 lr=8.881912e-05 lr_backbone=8.881912e-05 elapsed=05:35:34 avg_step=1.57s recent_step=1.79s eta=16:14:15 +2026-09-20 12:13:52 | INFO | train rank=0 | step=12820/50000 progress=25.6% loss=0.112195 lr=8.879826e-05 lr_backbone=8.879826e-05 elapsed=05:35:52 avg_step=1.57s recent_step=1.71s eta=16:14:04 +2026-09-20 12:14:05 | INFO | train rank=0 | step=12830/50000 progress=25.7% loss=0.120309 lr=8.877740e-05 lr_backbone=8.877740e-05 elapsed=05:36:05 avg_step=1.57s recent_step=1.31s eta=16:13:40 +2026-09-20 12:14:19 | INFO | train rank=0 | step=12840/50000 progress=25.7% loss=0.129225 lr=8.875651e-05 lr_backbone=8.875651e-05 elapsed=05:36:19 avg_step=1.57s recent_step=1.40s eta=16:13:20 +2026-09-20 12:14:34 | INFO | train rank=0 | step=12850/50000 progress=25.7% loss=0.111325 lr=8.873561e-05 lr_backbone=8.873561e-05 elapsed=05:36:34 avg_step=1.57s recent_step=1.50s eta=16:13:02 +2026-09-20 12:14:46 | INFO | train rank=0 | step=12860/50000 progress=25.7% loss=0.102476 lr=8.871469e-05 lr_backbone=8.871469e-05 elapsed=05:36:45 avg_step=1.57s recent_step=1.16s eta=16:12:34 +2026-09-20 12:15:02 | INFO | train rank=0 | step=12870/50000 progress=25.7% loss=0.111650 lr=8.869376e-05 lr_backbone=8.869376e-05 elapsed=05:37:02 avg_step=1.57s recent_step=1.69s eta=16:12:22 +2026-09-20 12:15:19 | INFO | train rank=0 | step=12880/50000 progress=25.8% loss=0.083760 lr=8.867280e-05 lr_backbone=8.867280e-05 elapsed=05:37:18 avg_step=1.57s recent_step=1.62s eta=16:12:07 +2026-09-20 12:15:37 | INFO | train rank=0 | step=12890/50000 progress=25.8% loss=0.089618 lr=8.865183e-05 lr_backbone=8.865183e-05 elapsed=05:37:37 avg_step=1.57s recent_step=1.85s eta=16:12:00 +2026-09-20 12:16:00 | INFO | train rank=0 | step=12900/50000 progress=25.8% loss=0.116904 lr=8.863085e-05 lr_backbone=8.863085e-05 elapsed=05:38:00 avg_step=1.57s recent_step=2.29s eta=16:12:04 +2026-09-20 12:16:54 | INFO | train rank=0 | step=12910/50000 progress=25.8% loss=0.096145 lr=8.860984e-05 lr_backbone=8.860984e-05 elapsed=05:38:53 avg_step=1.58s recent_step=5.37s eta=16:13:38 +2026-09-20 12:17:05 | INFO | train rank=0 | step=12920/50000 progress=25.8% loss=0.188472 lr=8.858882e-05 lr_backbone=8.858882e-05 elapsed=05:39:05 avg_step=1.57s recent_step=1.14s eta=16:13:10 +2026-09-20 12:17:17 | INFO | train rank=0 | step=12930/50000 progress=25.9% loss=0.108826 lr=8.856779e-05 lr_backbone=8.856779e-05 elapsed=05:39:17 avg_step=1.57s recent_step=1.19s eta=16:12:43 +2026-09-20 12:17:29 | INFO | train rank=0 | step=12940/50000 progress=25.9% loss=0.113981 lr=8.854673e-05 lr_backbone=8.854673e-05 elapsed=05:39:29 avg_step=1.57s recent_step=1.19s eta=16:12:16 +2026-09-20 12:17:44 | INFO | train rank=0 | step=12950/50000 progress=25.9% loss=0.090632 lr=8.852566e-05 lr_backbone=8.852566e-05 elapsed=05:39:44 avg_step=1.57s recent_step=1.52s eta=16:11:59 +2026-09-20 12:17:57 | INFO | train rank=0 | step=12960/50000 progress=25.9% loss=0.116155 lr=8.850457e-05 lr_backbone=8.850457e-05 elapsed=05:39:57 avg_step=1.57s recent_step=1.32s eta=16:11:36 +2026-09-20 12:18:10 | INFO | train rank=0 | step=12970/50000 progress=25.9% loss=0.102376 lr=8.848347e-05 lr_backbone=8.848347e-05 elapsed=05:40:10 avg_step=1.57s recent_step=1.30s eta=16:11:13 +2026-09-20 12:18:26 | INFO | train rank=0 | step=12980/50000 progress=26.0% loss=0.124083 lr=8.846235e-05 lr_backbone=8.846235e-05 elapsed=05:40:26 avg_step=1.57s recent_step=1.55s eta=16:10:56 +2026-09-20 12:18:39 | INFO | train rank=0 | step=12990/50000 progress=26.0% loss=0.092392 lr=8.844121e-05 lr_backbone=8.844121e-05 elapsed=05:40:39 avg_step=1.57s recent_step=1.33s eta=16:10:34 +2026-09-20 12:18:52 | INFO | train rank=0 | step=13000/50000 progress=26.0% loss=0.079085 lr=8.842006e-05 lr_backbone=8.842006e-05 elapsed=05:40:51 avg_step=1.57s recent_step=1.25s eta=16:10:09 +2026-09-20 12:19:07 | INFO | train rank=0 | step=13010/50000 progress=26.0% loss=0.076555 lr=8.839888e-05 lr_backbone=8.839888e-05 elapsed=05:41:07 avg_step=1.57s recent_step=1.55s eta=16:09:52 +2026-09-20 12:19:24 | INFO | train rank=0 | step=13020/50000 progress=26.0% loss=0.082000 lr=8.837770e-05 lr_backbone=8.837770e-05 elapsed=05:41:24 avg_step=1.57s recent_step=1.70s eta=16:09:40 +2026-09-20 12:19:38 | INFO | train rank=0 | step=13030/50000 progress=26.1% loss=0.123887 lr=8.835649e-05 lr_backbone=8.835649e-05 elapsed=05:41:37 avg_step=1.57s recent_step=1.36s eta=16:09:18 +2026-09-20 12:19:56 | INFO | train rank=0 | step=13040/50000 progress=26.1% loss=0.168840 lr=8.833527e-05 lr_backbone=8.833527e-05 elapsed=05:41:56 avg_step=1.57s recent_step=1.83s eta=16:09:10 +2026-09-20 12:20:15 | INFO | train rank=0 | step=13050/50000 progress=26.1% loss=0.144505 lr=8.831403e-05 lr_backbone=8.831403e-05 elapsed=05:42:14 avg_step=1.57s recent_step=1.85s eta=16:09:02 +2026-09-20 12:20:29 | INFO | train rank=0 | step=13060/50000 progress=26.1% loss=0.155822 lr=8.829277e-05 lr_backbone=8.829277e-05 elapsed=05:42:28 avg_step=1.57s recent_step=1.40s eta=16:08:41 +2026-09-20 12:20:48 | INFO | train rank=0 | step=13070/50000 progress=26.1% loss=0.120143 lr=8.827150e-05 lr_backbone=8.827150e-05 elapsed=05:42:48 avg_step=1.57s recent_step=1.98s eta=16:08:37 +2026-09-20 12:21:05 | INFO | train rank=0 | step=13080/50000 progress=26.2% loss=0.102400 lr=8.825021e-05 lr_backbone=8.825021e-05 elapsed=05:43:04 avg_step=1.57s recent_step=1.61s eta=16:08:22 +2026-09-20 12:21:24 | INFO | train rank=0 | step=13090/50000 progress=26.2% loss=0.121113 lr=8.822891e-05 lr_backbone=8.822891e-05 elapsed=05:43:24 avg_step=1.57s recent_step=1.97s eta=16:08:18 +2026-09-20 12:21:39 | INFO | train rank=0 | step=13100/50000 progress=26.2% loss=0.114598 lr=8.820758e-05 lr_backbone=8.820758e-05 elapsed=05:43:39 avg_step=1.57s recent_step=1.48s eta=16:07:59 +2026-09-20 12:21:53 | INFO | train rank=0 | step=13110/50000 progress=26.2% loss=0.114547 lr=8.818624e-05 lr_backbone=8.818624e-05 elapsed=05:43:53 avg_step=1.57s recent_step=1.40s eta=16:07:39 +2026-09-20 12:22:12 | INFO | train rank=0 | step=13120/50000 progress=26.2% loss=0.091721 lr=8.816489e-05 lr_backbone=8.816489e-05 elapsed=05:44:11 avg_step=1.57s recent_step=1.86s eta=16:07:31 +2026-09-20 12:22:27 | INFO | train rank=0 | step=13130/50000 progress=26.3% loss=0.136430 lr=8.814351e-05 lr_backbone=8.814351e-05 elapsed=05:44:26 avg_step=1.57s recent_step=1.51s eta=16:07:14 +2026-09-20 12:22:40 | INFO | train rank=0 | step=13140/50000 progress=26.3% loss=0.112810 lr=8.812213e-05 lr_backbone=8.812213e-05 elapsed=05:44:39 avg_step=1.57s recent_step=1.31s eta=16:06:50 +2026-09-20 12:22:58 | INFO | train rank=0 | step=13150/50000 progress=26.3% loss=0.080867 lr=8.810072e-05 lr_backbone=8.810072e-05 elapsed=05:44:58 avg_step=1.57s recent_step=1.82s eta=16:06:42 +2026-09-20 12:23:16 | INFO | train rank=0 | step=13160/50000 progress=26.3% loss=0.108039 lr=8.807930e-05 lr_backbone=8.807930e-05 elapsed=05:45:16 avg_step=1.57s recent_step=1.82s eta=16:06:33 +2026-09-20 12:23:28 | INFO | train rank=0 | step=13170/50000 progress=26.3% loss=0.118062 lr=8.805786e-05 lr_backbone=8.805786e-05 elapsed=05:45:28 avg_step=1.57s recent_step=1.21s eta=16:06:07 +2026-09-20 12:23:41 | INFO | train rank=0 | step=13180/50000 progress=26.4% loss=0.098675 lr=8.803640e-05 lr_backbone=8.803640e-05 elapsed=05:45:41 avg_step=1.57s recent_step=1.27s eta=16:05:43 +2026-09-20 12:24:02 | INFO | train rank=0 | step=13190/50000 progress=26.4% loss=0.113942 lr=8.801493e-05 lr_backbone=8.801493e-05 elapsed=05:46:01 avg_step=1.57s recent_step=2.08s eta=16:05:41 +2026-09-20 12:24:15 | INFO | train rank=0 | step=13200/50000 progress=26.4% loss=0.092487 lr=8.799344e-05 lr_backbone=8.799344e-05 elapsed=05:46:14 avg_step=1.57s recent_step=1.27s eta=16:05:17 +2026-09-20 12:24:32 | INFO | train rank=0 | step=13210/50000 progress=26.4% loss=0.126065 lr=8.797193e-05 lr_backbone=8.797193e-05 elapsed=05:46:32 avg_step=1.57s recent_step=1.77s eta=16:05:07 +2026-09-20 12:24:45 | INFO | train rank=0 | step=13220/50000 progress=26.4% loss=0.110200 lr=8.795041e-05 lr_backbone=8.795041e-05 elapsed=05:46:45 avg_step=1.57s recent_step=1.30s eta=16:04:43 +2026-09-20 12:25:04 | INFO | train rank=0 | step=13230/50000 progress=26.5% loss=0.104090 lr=8.792887e-05 lr_backbone=8.792887e-05 elapsed=05:47:04 avg_step=1.57s recent_step=1.86s eta=16:04:36 +2026-09-20 12:25:23 | INFO | train rank=0 | step=13240/50000 progress=26.5% loss=0.095050 lr=8.790732e-05 lr_backbone=8.790732e-05 elapsed=05:47:23 avg_step=1.57s recent_step=1.94s eta=16:04:30 +2026-09-20 12:25:35 | INFO | train rank=0 | step=13250/50000 progress=26.5% loss=0.110016 lr=8.788574e-05 lr_backbone=8.788574e-05 elapsed=05:47:35 avg_step=1.57s recent_step=1.17s eta=16:04:03 +2026-09-20 12:25:48 | INFO | train rank=0 | step=13260/50000 progress=26.5% loss=0.090458 lr=8.786415e-05 lr_backbone=8.786415e-05 elapsed=05:47:47 avg_step=1.57s recent_step=1.27s eta=16:03:39 +2026-09-20 12:26:01 | INFO | train rank=0 | step=13270/50000 progress=26.5% loss=0.108189 lr=8.784255e-05 lr_backbone=8.784255e-05 elapsed=05:48:01 avg_step=1.57s recent_step=1.35s eta=16:03:17 +2026-09-20 12:26:21 | INFO | train rank=0 | step=13280/50000 progress=26.6% loss=0.104380 lr=8.782093e-05 lr_backbone=8.782093e-05 elapsed=05:48:20 avg_step=1.57s recent_step=1.95s eta=16:03:12 +2026-09-20 12:26:33 | INFO | train rank=0 | step=13290/50000 progress=26.6% loss=0.128877 lr=8.779929e-05 lr_backbone=8.779929e-05 elapsed=05:48:33 avg_step=1.57s recent_step=1.27s eta=16:02:48 +2026-09-20 12:26:48 | INFO | train rank=0 | step=13300/50000 progress=26.6% loss=0.099322 lr=8.777763e-05 lr_backbone=8.777763e-05 elapsed=05:48:48 avg_step=1.57s recent_step=1.48s eta=16:02:29 +2026-09-20 12:27:10 | INFO | train rank=0 | step=13310/50000 progress=26.6% loss=0.097734 lr=8.775596e-05 lr_backbone=8.775596e-05 elapsed=05:49:09 avg_step=1.57s recent_step=2.16s eta=16:02:30 +2026-09-20 12:27:25 | INFO | train rank=0 | step=13320/50000 progress=26.6% loss=0.102472 lr=8.773427e-05 lr_backbone=8.773427e-05 elapsed=05:49:24 avg_step=1.57s recent_step=1.48s eta=16:02:11 +2026-09-20 12:27:38 | INFO | train rank=0 | step=13330/50000 progress=26.7% loss=0.095273 lr=8.771257e-05 lr_backbone=8.771257e-05 elapsed=05:49:37 avg_step=1.57s recent_step=1.31s eta=16:01:48 +2026-09-20 12:27:54 | INFO | train rank=0 | step=13340/50000 progress=26.7% loss=0.130565 lr=8.769085e-05 lr_backbone=8.769085e-05 elapsed=05:49:54 avg_step=1.57s recent_step=1.66s eta=16:01:35 +2026-09-20 12:28:13 | INFO | train rank=0 | step=13350/50000 progress=26.7% loss=0.133754 lr=8.766911e-05 lr_backbone=8.766911e-05 elapsed=05:50:12 avg_step=1.57s recent_step=1.83s eta=16:01:26 +2026-09-20 12:28:25 | INFO | train rank=0 | step=13360/50000 progress=26.7% loss=0.114639 lr=8.764736e-05 lr_backbone=8.764736e-05 elapsed=05:50:24 avg_step=1.57s recent_step=1.21s eta=16:01:00 +2026-09-20 12:28:43 | INFO | train rank=0 | step=13370/50000 progress=26.7% loss=0.131021 lr=8.762559e-05 lr_backbone=8.762559e-05 elapsed=05:50:42 avg_step=1.57s recent_step=1.83s eta=16:00:51 +2026-09-20 12:28:55 | INFO | train rank=0 | step=13380/50000 progress=26.8% loss=0.103614 lr=8.760380e-05 lr_backbone=8.760380e-05 elapsed=05:50:55 avg_step=1.57s recent_step=1.22s eta=16:00:26 +2026-09-20 12:29:15 | INFO | train rank=0 | step=13390/50000 progress=26.8% loss=0.083930 lr=8.758200e-05 lr_backbone=8.758200e-05 elapsed=05:51:15 avg_step=1.57s recent_step=2.03s eta=16:00:23 +2026-09-20 12:29:33 | INFO | train rank=0 | step=13400/50000 progress=26.8% loss=0.102005 lr=8.756018e-05 lr_backbone=8.756018e-05 elapsed=05:51:33 avg_step=1.57s recent_step=1.76s eta=16:00:12 +2026-09-20 12:29:45 | INFO | train rank=0 | step=13410/50000 progress=26.8% loss=0.117193 lr=8.753834e-05 lr_backbone=8.753834e-05 elapsed=05:51:44 avg_step=1.57s recent_step=1.18s eta=15:59:45 +2026-09-20 12:29:57 | INFO | train rank=0 | step=13420/50000 progress=26.8% loss=0.097355 lr=8.751649e-05 lr_backbone=8.751649e-05 elapsed=05:51:56 avg_step=1.57s recent_step=1.21s eta=15:59:20 +2026-09-20 12:30:17 | INFO | train rank=0 | step=13430/50000 progress=26.9% loss=0.084586 lr=8.749462e-05 lr_backbone=8.749462e-05 elapsed=05:52:17 avg_step=1.57s recent_step=2.03s eta=15:59:16 +2026-09-20 12:30:31 | INFO | train rank=0 | step=13440/50000 progress=26.9% loss=0.101479 lr=8.747273e-05 lr_backbone=8.747273e-05 elapsed=05:52:31 avg_step=1.57s recent_step=1.38s eta=15:58:56 +2026-09-20 12:30:46 | INFO | train rank=0 | step=13450/50000 progress=26.9% loss=0.105448 lr=8.745083e-05 lr_backbone=8.745083e-05 elapsed=05:52:45 avg_step=1.57s recent_step=1.47s eta=15:58:37 +2026-09-20 12:31:03 | INFO | train rank=0 | step=13460/50000 progress=26.9% loss=0.082075 lr=8.742891e-05 lr_backbone=8.742891e-05 elapsed=05:53:03 avg_step=1.57s recent_step=1.72s eta=15:58:25 +2026-09-20 12:31:15 | INFO | train rank=0 | step=13470/50000 progress=26.9% loss=0.118010 lr=8.740698e-05 lr_backbone=8.740698e-05 elapsed=05:53:14 avg_step=1.57s recent_step=1.19s eta=15:57:59 +2026-09-20 12:31:34 | INFO | train rank=0 | step=13480/50000 progress=27.0% loss=0.104150 lr=8.738503e-05 lr_backbone=8.738503e-05 elapsed=05:53:34 avg_step=1.57s recent_step=1.92s eta=15:57:53 +2026-09-20 12:31:51 | INFO | train rank=0 | step=13490/50000 progress=27.0% loss=0.091147 lr=8.736306e-05 lr_backbone=8.736306e-05 elapsed=05:53:51 avg_step=1.57s recent_step=1.73s eta=15:57:42 +2026-09-20 12:32:03 | INFO | train rank=0 | step=13500/50000 progress=27.0% loss=0.101480 lr=8.734107e-05 lr_backbone=8.734107e-05 elapsed=05:54:03 avg_step=1.57s recent_step=1.18s eta=15:57:15 +2026-09-20 12:32:18 | INFO | train rank=0 | step=13510/50000 progress=27.0% loss=0.102451 lr=8.731907e-05 lr_backbone=8.731907e-05 elapsed=05:54:17 avg_step=1.57s recent_step=1.46s eta=15:56:57 +2026-09-20 12:32:42 | INFO | train rank=0 | step=13520/50000 progress=27.0% loss=0.102473 lr=8.729706e-05 lr_backbone=8.729706e-05 elapsed=05:54:42 avg_step=1.57s recent_step=2.43s eta=15:57:04 +2026-09-20 12:32:54 | INFO | train rank=0 | step=13530/50000 progress=27.1% loss=0.122018 lr=8.727502e-05 lr_backbone=8.727502e-05 elapsed=05:54:53 avg_step=1.57s recent_step=1.17s eta=15:56:37 +2026-09-20 12:33:06 | INFO | train rank=0 | step=13540/50000 progress=27.1% loss=0.152039 lr=8.725298e-05 lr_backbone=8.725298e-05 elapsed=05:55:06 avg_step=1.57s recent_step=1.23s eta=15:56:12 +2026-09-20 12:33:23 | INFO | train rank=0 | step=13550/50000 progress=27.1% loss=0.126259 lr=8.723091e-05 lr_backbone=8.723091e-05 elapsed=05:55:22 avg_step=1.57s recent_step=1.65s eta=15:55:58 +2026-09-20 12:33:39 | INFO | train rank=0 | step=13560/50000 progress=27.1% loss=0.155017 lr=8.720883e-05 lr_backbone=8.720883e-05 elapsed=05:55:39 avg_step=1.57s recent_step=1.67s eta=15:55:45 +2026-09-20 12:33:57 | INFO | train rank=0 | step=13570/50000 progress=27.1% loss=0.105335 lr=8.718673e-05 lr_backbone=8.718673e-05 elapsed=05:55:56 avg_step=1.57s recent_step=1.75s eta=15:55:34 +2026-09-20 12:34:09 | INFO | train rank=0 | step=13580/50000 progress=27.2% loss=0.100339 lr=8.716462e-05 lr_backbone=8.716462e-05 elapsed=05:56:09 avg_step=1.57s recent_step=1.26s eta=15:55:10 +2026-09-20 12:34:23 | INFO | train rank=0 | step=13590/50000 progress=27.2% loss=0.101074 lr=8.714249e-05 lr_backbone=8.714249e-05 elapsed=05:56:23 avg_step=1.57s recent_step=1.38s eta=15:54:49 +2026-09-20 12:34:40 | INFO | train rank=0 | step=13600/50000 progress=27.2% loss=0.096248 lr=8.712034e-05 lr_backbone=8.712034e-05 elapsed=05:56:39 avg_step=1.57s recent_step=1.65s eta=15:54:35 +2026-09-20 12:34:59 | INFO | train rank=0 | step=13610/50000 progress=27.2% loss=0.113335 lr=8.709818e-05 lr_backbone=8.709818e-05 elapsed=05:56:59 avg_step=1.57s recent_step=1.95s eta=15:54:30 +2026-09-20 12:35:18 | INFO | train rank=0 | step=13620/50000 progress=27.2% loss=0.092857 lr=8.707600e-05 lr_backbone=8.707600e-05 elapsed=05:57:18 avg_step=1.57s recent_step=1.88s eta=15:54:22 +2026-09-20 12:35:30 | INFO | train rank=0 | step=13630/50000 progress=27.3% loss=0.112324 lr=8.705380e-05 lr_backbone=8.705380e-05 elapsed=05:57:29 avg_step=1.57s recent_step=1.16s eta=15:53:55 +2026-09-20 12:35:44 | INFO | train rank=0 | step=13640/50000 progress=27.3% loss=0.084226 lr=8.703159e-05 lr_backbone=8.703159e-05 elapsed=05:57:44 avg_step=1.57s recent_step=1.48s eta=15:53:37 +2026-09-20 12:36:04 | INFO | train rank=0 | step=13650/50000 progress=27.3% loss=0.106156 lr=8.700936e-05 lr_backbone=8.700936e-05 elapsed=05:58:04 avg_step=1.57s recent_step=1.96s eta=15:53:32 +2026-09-20 12:36:16 | INFO | train rank=0 | step=13660/50000 progress=27.3% loss=0.089641 lr=8.698712e-05 lr_backbone=8.698712e-05 elapsed=05:58:16 avg_step=1.57s recent_step=1.20s eta=15:53:06 +2026-09-20 12:36:33 | INFO | train rank=0 | step=13670/50000 progress=27.3% loss=0.108528 lr=8.696486e-05 lr_backbone=8.696486e-05 elapsed=05:58:32 avg_step=1.57s recent_step=1.69s eta=15:52:53 +2026-09-20 12:36:51 | INFO | train rank=0 | step=13680/50000 progress=27.4% loss=0.098910 lr=8.694258e-05 lr_backbone=8.694258e-05 elapsed=05:58:51 avg_step=1.57s recent_step=1.83s eta=15:52:44 +2026-09-20 12:37:06 | INFO | train rank=0 | step=13690/50000 progress=27.4% loss=0.097199 lr=8.692029e-05 lr_backbone=8.692029e-05 elapsed=05:59:06 avg_step=1.57s recent_step=1.50s eta=15:52:27 +2026-09-20 12:37:22 | INFO | train rank=0 | step=13700/50000 progress=27.4% loss=0.102651 lr=8.689798e-05 lr_backbone=8.689798e-05 elapsed=05:59:21 avg_step=1.57s recent_step=1.56s eta=15:52:11 +2026-09-20 12:37:34 | INFO | train rank=0 | step=13710/50000 progress=27.4% loss=0.099875 lr=8.687566e-05 lr_backbone=8.687566e-05 elapsed=05:59:34 avg_step=1.57s recent_step=1.23s eta=15:51:46 +2026-09-20 12:37:52 | INFO | train rank=0 | step=13720/50000 progress=27.4% loss=0.076009 lr=8.685331e-05 lr_backbone=8.685331e-05 elapsed=05:59:51 avg_step=1.57s recent_step=1.75s eta=15:51:35 +2026-09-20 12:38:08 | INFO | train rank=0 | step=13730/50000 progress=27.5% loss=0.077312 lr=8.683096e-05 lr_backbone=8.683096e-05 elapsed=06:00:08 avg_step=1.57s recent_step=1.66s eta=15:51:21 +2026-09-20 12:38:20 | INFO | train rank=0 | step=13740/50000 progress=27.5% loss=0.100568 lr=8.680858e-05 lr_backbone=8.680858e-05 elapsed=06:00:20 avg_step=1.57s recent_step=1.22s eta=15:50:56 +2026-09-20 12:38:39 | INFO | train rank=0 | step=13750/50000 progress=27.5% loss=0.114700 lr=8.678620e-05 lr_backbone=8.678620e-05 elapsed=06:00:39 avg_step=1.57s recent_step=1.91s eta=15:50:49 +2026-09-20 12:38:54 | INFO | train rank=0 | step=13760/50000 progress=27.5% loss=0.093866 lr=8.676379e-05 lr_backbone=8.676379e-05 elapsed=06:00:53 avg_step=1.57s recent_step=1.43s eta=15:50:30 +2026-09-20 12:39:06 | INFO | train rank=0 | step=13770/50000 progress=27.5% loss=0.103400 lr=8.674137e-05 lr_backbone=8.674137e-05 elapsed=06:01:05 avg_step=1.57s recent_step=1.19s eta=15:50:04 +2026-09-20 12:39:24 | INFO | train rank=0 | step=13780/50000 progress=27.6% loss=0.094822 lr=8.671893e-05 lr_backbone=8.671893e-05 elapsed=06:01:24 avg_step=1.57s recent_step=1.85s eta=15:49:56 +2026-09-20 12:39:41 | INFO | train rank=0 | step=13790/50000 progress=27.6% loss=0.103853 lr=8.669648e-05 lr_backbone=8.669648e-05 elapsed=06:01:41 avg_step=1.57s recent_step=1.69s eta=15:49:43 +2026-09-20 12:39:58 | INFO | train rank=0 | step=13800/50000 progress=27.6% loss=0.083996 lr=8.667401e-05 lr_backbone=8.667401e-05 elapsed=06:01:58 avg_step=1.57s recent_step=1.69s eta=15:49:30 +2026-09-20 12:40:12 | INFO | train rank=0 | step=13810/50000 progress=27.6% loss=0.090123 lr=8.665152e-05 lr_backbone=8.665152e-05 elapsed=06:02:12 avg_step=1.57s recent_step=1.42s eta=15:49:11 +2026-09-20 12:40:24 | INFO | train rank=0 | step=13820/50000 progress=27.6% loss=0.097825 lr=8.662902e-05 lr_backbone=8.662902e-05 elapsed=06:02:24 avg_step=1.57s recent_step=1.20s eta=15:48:45 +2026-09-20 12:40:43 | INFO | train rank=0 | step=13830/50000 progress=27.7% loss=0.123894 lr=8.660650e-05 lr_backbone=8.660650e-05 elapsed=06:02:42 avg_step=1.57s recent_step=1.85s eta=15:48:37 +2026-09-20 12:40:59 | INFO | train rank=0 | step=13840/50000 progress=27.7% loss=0.133377 lr=8.658397e-05 lr_backbone=8.658397e-05 elapsed=06:02:59 avg_step=1.57s recent_step=1.63s eta=15:48:22 +2026-09-20 12:41:17 | INFO | train rank=0 | step=13850/50000 progress=27.7% loss=0.129641 lr=8.656142e-05 lr_backbone=8.656142e-05 elapsed=06:03:17 avg_step=1.57s recent_step=1.79s eta=15:48:12 +2026-09-20 12:41:30 | INFO | train rank=0 | step=13860/50000 progress=27.7% loss=0.147406 lr=8.653885e-05 lr_backbone=8.653885e-05 elapsed=06:03:30 avg_step=1.57s recent_step=1.34s eta=15:47:51 +2026-09-20 12:41:44 | INFO | train rank=0 | step=13870/50000 progress=27.7% loss=0.171073 lr=8.651627e-05 lr_backbone=8.651627e-05 elapsed=06:03:44 avg_step=1.57s recent_step=1.40s eta=15:47:30 +2026-09-20 12:42:06 | INFO | train rank=0 | step=13880/50000 progress=27.8% loss=0.133203 lr=8.649367e-05 lr_backbone=8.649367e-05 elapsed=06:04:05 avg_step=1.57s recent_step=2.14s eta=15:47:29 +2026-09-20 12:42:18 | INFO | train rank=0 | step=13890/50000 progress=27.8% loss=0.099792 lr=8.647106e-05 lr_backbone=8.647106e-05 elapsed=06:04:17 avg_step=1.57s recent_step=1.18s eta=15:47:03 +2026-09-20 12:42:30 | INFO | train rank=0 | step=13900/50000 progress=27.8% loss=0.112365 lr=8.644843e-05 lr_backbone=8.644843e-05 elapsed=06:04:30 avg_step=1.57s recent_step=1.23s eta=15:46:39 +2026-09-20 12:42:48 | INFO | train rank=0 | step=13910/50000 progress=27.8% loss=0.108035 lr=8.642579e-05 lr_backbone=8.642579e-05 elapsed=06:04:47 avg_step=1.57s recent_step=1.78s eta=15:46:28 +2026-09-20 12:43:04 | INFO | train rank=0 | step=13920/50000 progress=27.8% loss=0.119520 lr=8.640312e-05 lr_backbone=8.640312e-05 elapsed=06:05:04 avg_step=1.57s recent_step=1.65s eta=15:46:14 +2026-09-20 12:43:19 | INFO | train rank=0 | step=13930/50000 progress=27.9% loss=0.139405 lr=8.638045e-05 lr_backbone=8.638045e-05 elapsed=06:05:19 avg_step=1.57s recent_step=1.51s eta=15:45:57 +2026-09-20 12:43:36 | INFO | train rank=0 | step=13940/50000 progress=27.9% loss=0.158275 lr=8.635775e-05 lr_backbone=8.635775e-05 elapsed=06:05:36 avg_step=1.57s recent_step=1.66s eta=15:45:44 +2026-09-20 12:43:55 | INFO | train rank=0 | step=13950/50000 progress=27.9% loss=0.083109 lr=8.633504e-05 lr_backbone=8.633504e-05 elapsed=06:05:55 avg_step=1.57s recent_step=1.91s eta=15:45:37 +2026-09-20 12:44:07 | INFO | train rank=0 | step=13960/50000 progress=27.9% loss=0.103565 lr=8.631232e-05 lr_backbone=8.631232e-05 elapsed=06:06:06 avg_step=1.57s recent_step=1.18s eta=15:45:11 +2026-09-20 12:44:21 | INFO | train rank=0 | step=13970/50000 progress=27.9% loss=0.099054 lr=8.628958e-05 lr_backbone=8.628958e-05 elapsed=06:06:20 avg_step=1.57s recent_step=1.40s eta=15:44:50 +2026-09-20 12:44:36 | INFO | train rank=0 | step=13980/50000 progress=28.0% loss=0.111418 lr=8.626682e-05 lr_backbone=8.626682e-05 elapsed=06:06:36 avg_step=1.57s recent_step=1.52s eta=15:44:33 +2026-09-20 12:44:56 | INFO | train rank=0 | step=13990/50000 progress=28.0% loss=0.096116 lr=8.624405e-05 lr_backbone=8.624405e-05 elapsed=06:06:56 avg_step=1.57s recent_step=2.04s eta=15:44:30 +2026-09-20 12:45:10 | INFO | train rank=0 | step=14000/50000 progress=28.0% loss=0.103940 lr=8.622126e-05 lr_backbone=8.622126e-05 elapsed=06:07:09 avg_step=1.57s recent_step=1.34s eta=15:44:08 +2026-09-20 12:45:23 | INFO | train rank=0 | step=14010/50000 progress=28.0% loss=0.090142 lr=8.619846e-05 lr_backbone=8.619846e-05 elapsed=06:07:23 avg_step=1.57s recent_step=1.33s eta=15:43:46 +2026-09-20 12:45:38 | INFO | train rank=0 | step=14020/50000 progress=28.0% loss=0.102408 lr=8.617564e-05 lr_backbone=8.617564e-05 elapsed=06:07:38 avg_step=1.57s recent_step=1.52s eta=15:43:29 +2026-09-20 12:45:57 | INFO | train rank=0 | step=14030/50000 progress=28.1% loss=0.117638 lr=8.615280e-05 lr_backbone=8.615280e-05 elapsed=06:07:57 avg_step=1.57s recent_step=1.92s eta=15:43:22 +2026-09-20 12:46:15 | INFO | train rank=0 | step=14040/50000 progress=28.1% loss=0.090854 lr=8.612995e-05 lr_backbone=8.612995e-05 elapsed=06:08:14 avg_step=1.57s recent_step=1.71s eta=15:43:10 +2026-09-20 12:46:29 | INFO | train rank=0 | step=14050/50000 progress=28.1% loss=0.099058 lr=8.610708e-05 lr_backbone=8.610708e-05 elapsed=06:08:29 avg_step=1.57s recent_step=1.47s eta=15:42:51 +2026-09-20 12:46:46 | INFO | train rank=0 | step=14060/50000 progress=28.1% loss=0.089611 lr=8.608420e-05 lr_backbone=8.608420e-05 elapsed=06:08:45 avg_step=1.57s recent_step=1.63s eta=15:42:37 +2026-09-20 12:46:59 | INFO | train rank=0 | step=14070/50000 progress=28.1% loss=0.085188 lr=8.606130e-05 lr_backbone=8.606130e-05 elapsed=06:08:59 avg_step=1.57s recent_step=1.34s eta=15:42:15 +2026-09-20 12:47:14 | INFO | train rank=0 | step=14080/50000 progress=28.2% loss=0.077508 lr=8.603838e-05 lr_backbone=8.603838e-05 elapsed=06:09:14 avg_step=1.57s recent_step=1.53s eta=15:41:58 +2026-09-20 12:47:27 | INFO | train rank=0 | step=14090/50000 progress=28.2% loss=0.103659 lr=8.601545e-05 lr_backbone=8.601545e-05 elapsed=06:09:26 avg_step=1.57s recent_step=1.23s eta=15:41:34 +2026-09-20 12:47:43 | INFO | train rank=0 | step=14100/50000 progress=28.2% loss=0.079459 lr=8.599250e-05 lr_backbone=8.599250e-05 elapsed=06:09:43 avg_step=1.57s recent_step=1.64s eta=15:41:20 +2026-09-20 12:48:01 | INFO | train rank=0 | step=14110/50000 progress=28.2% loss=0.121048 lr=8.596954e-05 lr_backbone=8.596954e-05 elapsed=06:10:01 avg_step=1.57s recent_step=1.84s eta=15:41:11 +2026-09-20 12:48:14 | INFO | train rank=0 | step=14120/50000 progress=28.2% loss=0.109369 lr=8.594656e-05 lr_backbone=8.594656e-05 elapsed=06:10:14 avg_step=1.57s recent_step=1.27s eta=15:40:47 +2026-09-20 12:48:30 | INFO | train rank=0 | step=14130/50000 progress=28.3% loss=0.114065 lr=8.592357e-05 lr_backbone=8.592357e-05 elapsed=06:10:29 avg_step=1.57s recent_step=1.58s eta=15:40:32 +2026-09-20 12:48:43 | INFO | train rank=0 | step=14140/50000 progress=28.3% loss=0.131347 lr=8.590056e-05 lr_backbone=8.590056e-05 elapsed=06:10:43 avg_step=1.57s recent_step=1.35s eta=15:40:10 +2026-09-20 12:49:00 | INFO | train rank=0 | step=14150/50000 progress=28.3% loss=0.144028 lr=8.587754e-05 lr_backbone=8.587754e-05 elapsed=06:11:00 avg_step=1.57s recent_step=1.68s eta=15:39:57 +2026-09-20 12:49:15 | INFO | train rank=0 | step=14160/50000 progress=28.3% loss=0.100567 lr=8.585449e-05 lr_backbone=8.585449e-05 elapsed=06:11:15 avg_step=1.57s recent_step=1.53s eta=15:39:41 +2026-09-20 12:49:32 | INFO | train rank=0 | step=14170/50000 progress=28.3% loss=0.092611 lr=8.583144e-05 lr_backbone=8.583144e-05 elapsed=06:11:31 avg_step=1.57s recent_step=1.62s eta=15:39:26 +2026-09-20 12:49:44 | INFO | train rank=0 | step=14180/50000 progress=28.4% loss=0.150882 lr=8.580837e-05 lr_backbone=8.580837e-05 elapsed=06:11:43 avg_step=1.57s recent_step=1.22s eta=15:39:01 +2026-09-20 12:50:04 | INFO | train rank=0 | step=14190/50000 progress=28.4% loss=0.089339 lr=8.578528e-05 lr_backbone=8.578528e-05 elapsed=06:12:04 avg_step=1.57s recent_step=2.05s eta=15:38:58 +2026-09-20 12:50:21 | INFO | train rank=0 | step=14200/50000 progress=28.4% loss=0.103627 lr=8.576217e-05 lr_backbone=8.576217e-05 elapsed=06:12:20 avg_step=1.57s recent_step=1.62s eta=15:38:43 +2026-09-20 12:50:33 | INFO | train rank=0 | step=14210/50000 progress=28.4% loss=0.088346 lr=8.573906e-05 lr_backbone=8.573906e-05 elapsed=06:12:33 avg_step=1.57s recent_step=1.24s eta=15:38:19 +2026-09-20 12:50:50 | INFO | train rank=0 | step=14220/50000 progress=28.4% loss=0.097493 lr=8.571592e-05 lr_backbone=8.571592e-05 elapsed=06:12:50 avg_step=1.57s recent_step=1.74s eta=15:38:08 +2026-09-20 12:51:07 | INFO | train rank=0 | step=14230/50000 progress=28.5% loss=0.083018 lr=8.569277e-05 lr_backbone=8.569277e-05 elapsed=06:13:07 avg_step=1.57s recent_step=1.69s eta=15:37:55 +2026-09-20 12:51:21 | INFO | train rank=0 | step=14240/50000 progress=28.5% loss=0.091613 lr=8.566960e-05 lr_backbone=8.566960e-05 elapsed=06:13:21 avg_step=1.57s recent_step=1.41s eta=15:37:35 +2026-09-20 12:51:37 | INFO | train rank=0 | step=14250/50000 progress=28.5% loss=0.099210 lr=8.564642e-05 lr_backbone=8.564642e-05 elapsed=06:13:37 avg_step=1.57s recent_step=1.56s eta=15:37:19 +2026-09-20 12:51:52 | INFO | train rank=0 | step=14260/50000 progress=28.5% loss=0.092145 lr=8.562323e-05 lr_backbone=8.562323e-05 elapsed=06:13:51 avg_step=1.57s recent_step=1.46s eta=15:37:00 +2026-09-20 12:52:08 | INFO | train rank=0 | step=14270/50000 progress=28.5% loss=0.087221 lr=8.560001e-05 lr_backbone=8.560001e-05 elapsed=06:14:07 avg_step=1.57s recent_step=1.60s eta=15:36:45 +2026-09-20 12:52:26 | INFO | train rank=0 | step=14280/50000 progress=28.6% loss=0.106180 lr=8.557678e-05 lr_backbone=8.557678e-05 elapsed=06:14:26 avg_step=1.57s recent_step=1.87s eta=15:36:37 +2026-09-20 12:52:43 | INFO | train rank=0 | step=14290/50000 progress=28.6% loss=0.113421 lr=8.555354e-05 lr_backbone=8.555354e-05 elapsed=06:14:43 avg_step=1.57s recent_step=1.67s eta=15:36:24 +2026-09-20 12:52:55 | INFO | train rank=0 | step=14300/50000 progress=28.6% loss=0.092346 lr=8.553028e-05 lr_backbone=8.553028e-05 elapsed=06:14:55 avg_step=1.57s recent_step=1.20s eta=15:35:58 +2026-09-20 12:53:08 | INFO | train rank=0 | step=14310/50000 progress=28.6% loss=0.102881 lr=8.550701e-05 lr_backbone=8.550701e-05 elapsed=06:15:07 avg_step=1.57s recent_step=1.27s eta=15:35:35 +2026-09-20 12:53:24 | INFO | train rank=0 | step=14320/50000 progress=28.6% loss=0.100465 lr=8.548371e-05 lr_backbone=8.548371e-05 elapsed=06:15:24 avg_step=1.57s recent_step=1.65s eta=15:35:21 +2026-09-20 12:53:42 | INFO | train rank=0 | step=14330/50000 progress=28.7% loss=0.078592 lr=8.546041e-05 lr_backbone=8.546041e-05 elapsed=06:15:41 avg_step=1.57s recent_step=1.76s eta=15:35:10 +2026-09-20 12:54:02 | INFO | train rank=0 | step=14340/50000 progress=28.7% loss=0.094142 lr=8.543709e-05 lr_backbone=8.543709e-05 elapsed=06:16:02 avg_step=1.57s recent_step=2.04s eta=15:35:06 +2026-09-20 12:54:14 | INFO | train rank=0 | step=14350/50000 progress=28.7% loss=0.110121 lr=8.541375e-05 lr_backbone=8.541375e-05 elapsed=06:16:14 avg_step=1.57s recent_step=1.20s eta=15:34:41 +2026-09-20 12:54:26 | INFO | train rank=0 | step=14360/50000 progress=28.7% loss=0.114497 lr=8.539040e-05 lr_backbone=8.539040e-05 elapsed=06:16:26 avg_step=1.57s recent_step=1.20s eta=15:34:16 +2026-09-20 12:54:45 | INFO | train rank=0 | step=14370/50000 progress=28.7% loss=0.106305 lr=8.536703e-05 lr_backbone=8.536703e-05 elapsed=06:16:45 avg_step=1.57s recent_step=1.93s eta=15:34:09 +2026-09-20 12:55:00 | INFO | train rank=0 | step=14380/50000 progress=28.8% loss=0.142766 lr=8.534365e-05 lr_backbone=8.534365e-05 elapsed=06:16:59 avg_step=1.57s recent_step=1.42s eta=15:33:50 +2026-09-20 12:55:18 | INFO | train rank=0 | step=14390/50000 progress=28.8% loss=0.133003 lr=8.532025e-05 lr_backbone=8.532025e-05 elapsed=06:17:18 avg_step=1.57s recent_step=1.83s eta=15:33:41 +2026-09-20 12:55:30 | INFO | train rank=0 | step=14400/50000 progress=28.8% loss=0.109894 lr=8.529683e-05 lr_backbone=8.529683e-05 elapsed=06:17:29 avg_step=1.57s recent_step=1.19s eta=15:33:15 +2026-09-20 12:55:53 | INFO | train rank=0 | step=14410/50000 progress=28.8% loss=0.146766 lr=8.527340e-05 lr_backbone=8.527340e-05 elapsed=06:17:53 avg_step=1.57s recent_step=2.31s eta=15:33:18 +2026-09-20 12:56:05 | INFO | train rank=0 | step=14420/50000 progress=28.8% loss=0.089558 lr=8.524996e-05 lr_backbone=8.524996e-05 elapsed=06:18:04 avg_step=1.57s recent_step=1.19s eta=15:32:52 +2026-09-20 12:56:16 | INFO | train rank=0 | step=14430/50000 progress=28.9% loss=0.077369 lr=8.522650e-05 lr_backbone=8.522650e-05 elapsed=06:18:16 avg_step=1.57s recent_step=1.16s eta=15:32:27 +2026-09-20 12:56:34 | INFO | train rank=0 | step=14440/50000 progress=28.9% loss=0.110357 lr=8.520302e-05 lr_backbone=8.520302e-05 elapsed=06:18:34 avg_step=1.57s recent_step=1.79s eta=15:32:16 +2026-09-20 12:56:52 | INFO | train rank=0 | step=14450/50000 progress=28.9% loss=0.128253 lr=8.517953e-05 lr_backbone=8.517953e-05 elapsed=06:18:52 avg_step=1.57s recent_step=1.81s eta=15:32:06 +2026-09-20 12:57:11 | INFO | train rank=0 | step=14460/50000 progress=28.9% loss=0.088909 lr=8.515602e-05 lr_backbone=8.515602e-05 elapsed=06:19:11 avg_step=1.57s recent_step=1.90s eta=15:31:59 +2026-09-20 12:57:23 | INFO | train rank=0 | step=14470/50000 progress=28.9% loss=0.079083 lr=8.513250e-05 lr_backbone=8.513250e-05 elapsed=06:19:23 avg_step=1.57s recent_step=1.18s eta=15:31:33 +2026-09-20 12:57:35 | INFO | train rank=0 | step=14480/50000 progress=29.0% loss=0.101811 lr=8.510896e-05 lr_backbone=8.510896e-05 elapsed=06:19:35 avg_step=1.57s recent_step=1.19s eta=15:31:08 +2026-09-20 12:57:56 | INFO | train rank=0 | step=14490/50000 progress=29.0% loss=0.093647 lr=8.508541e-05 lr_backbone=8.508541e-05 elapsed=06:19:56 avg_step=1.57s recent_step=2.13s eta=15:31:06 +2026-09-20 12:58:09 | INFO | train rank=0 | step=14500/50000 progress=29.0% loss=0.124506 lr=8.506184e-05 lr_backbone=8.506184e-05 elapsed=06:20:08 avg_step=1.57s recent_step=1.22s eta=15:30:42 +2026-09-20 12:58:25 | INFO | train rank=0 | step=14510/50000 progress=29.0% loss=0.127606 lr=8.503826e-05 lr_backbone=8.503826e-05 elapsed=06:20:25 avg_step=1.57s recent_step=1.64s eta=15:30:28 +2026-09-20 12:58:37 | INFO | train rank=0 | step=14520/50000 progress=29.0% loss=0.110736 lr=8.501466e-05 lr_backbone=8.501466e-05 elapsed=06:20:37 avg_step=1.57s recent_step=1.24s eta=15:30:04 +2026-09-20 12:58:53 | INFO | train rank=0 | step=14530/50000 progress=29.1% loss=0.077418 lr=8.499104e-05 lr_backbone=8.499104e-05 elapsed=06:20:53 avg_step=1.57s recent_step=1.59s eta=15:29:48 +2026-09-20 12:59:07 | INFO | train rank=0 | step=14540/50000 progress=29.1% loss=0.109774 lr=8.496741e-05 lr_backbone=8.496741e-05 elapsed=06:21:07 avg_step=1.57s recent_step=1.36s eta=15:29:28 +2026-09-20 12:59:23 | INFO | train rank=0 | step=14550/50000 progress=29.1% loss=0.100426 lr=8.494377e-05 lr_backbone=8.494377e-05 elapsed=06:21:23 avg_step=1.57s recent_step=1.60s eta=15:29:13 +2026-09-20 12:59:40 | INFO | train rank=0 | step=14560/50000 progress=29.1% loss=0.114235 lr=8.492011e-05 lr_backbone=8.492011e-05 elapsed=06:21:39 avg_step=1.57s recent_step=1.68s eta=15:28:59 +2026-09-20 12:59:53 | INFO | train rank=0 | step=14570/50000 progress=29.1% loss=0.081328 lr=8.489643e-05 lr_backbone=8.489643e-05 elapsed=06:21:52 avg_step=1.57s recent_step=1.28s eta=15:28:37 +2026-09-20 13:00:14 | INFO | train rank=0 | step=14580/50000 progress=29.2% loss=0.133222 lr=8.487274e-05 lr_backbone=8.487274e-05 elapsed=06:22:14 avg_step=1.57s recent_step=2.16s eta=15:28:35 +2026-09-20 13:00:31 | INFO | train rank=0 | step=14590/50000 progress=29.2% loss=0.114961 lr=8.484903e-05 lr_backbone=8.484903e-05 elapsed=06:22:31 avg_step=1.57s recent_step=1.67s eta=15:28:22 +2026-09-20 13:00:43 | INFO | train rank=0 | step=14600/50000 progress=29.2% loss=0.135199 lr=8.482531e-05 lr_backbone=8.482531e-05 elapsed=06:22:43 avg_step=1.57s recent_step=1.19s eta=15:27:57 +2026-09-20 13:00:56 | INFO | train rank=0 | step=14610/50000 progress=29.2% loss=0.116922 lr=8.480158e-05 lr_backbone=8.480158e-05 elapsed=06:22:56 avg_step=1.57s recent_step=1.32s eta=15:27:35 +2026-09-20 13:01:11 | INFO | train rank=0 | step=14620/50000 progress=29.2% loss=0.093249 lr=8.477782e-05 lr_backbone=8.477782e-05 elapsed=06:23:10 avg_step=1.57s recent_step=1.45s eta=15:27:16 +2026-09-20 13:01:30 | INFO | train rank=0 | step=14630/50000 progress=29.3% loss=0.104284 lr=8.475406e-05 lr_backbone=8.475406e-05 elapsed=06:23:30 avg_step=1.57s recent_step=1.98s eta=15:27:11 +2026-09-20 13:01:42 | INFO | train rank=0 | step=14640/50000 progress=29.3% loss=0.100843 lr=8.473028e-05 lr_backbone=8.473028e-05 elapsed=06:23:42 avg_step=1.57s recent_step=1.19s eta=15:26:46 +2026-09-20 13:02:01 | INFO | train rank=0 | step=14650/50000 progress=29.3% loss=0.079164 lr=8.470648e-05 lr_backbone=8.470648e-05 elapsed=06:24:01 avg_step=1.57s recent_step=1.90s eta=15:26:38 +2026-09-20 13:02:14 | INFO | train rank=0 | step=14660/50000 progress=29.3% loss=0.084700 lr=8.468267e-05 lr_backbone=8.468267e-05 elapsed=06:24:14 avg_step=1.57s recent_step=1.30s eta=15:26:16 +2026-09-20 13:02:32 | INFO | train rank=0 | step=14670/50000 progress=29.3% loss=0.101059 lr=8.465884e-05 lr_backbone=8.465884e-05 elapsed=06:24:32 avg_step=1.57s recent_step=1.78s eta=15:26:05 +2026-09-20 13:02:51 | INFO | train rank=0 | step=14680/50000 progress=29.4% loss=0.079629 lr=8.463499e-05 lr_backbone=8.463499e-05 elapsed=06:24:51 avg_step=1.57s recent_step=1.90s eta=15:25:57 +2026-09-20 13:03:06 | INFO | train rank=0 | step=14690/50000 progress=29.4% loss=0.094223 lr=8.461114e-05 lr_backbone=8.461114e-05 elapsed=06:25:05 avg_step=1.57s recent_step=1.46s eta=15:25:39 +2026-09-20 13:03:18 | INFO | train rank=0 | step=14700/50000 progress=29.4% loss=0.104587 lr=8.458726e-05 lr_backbone=8.458726e-05 elapsed=06:25:17 avg_step=1.57s recent_step=1.18s eta=15:25:13 +2026-09-20 13:03:34 | INFO | train rank=0 | step=14710/50000 progress=29.4% loss=0.076520 lr=8.456337e-05 lr_backbone=8.456337e-05 elapsed=06:25:34 avg_step=1.57s recent_step=1.67s eta=15:25:00 +2026-09-20 13:03:54 | INFO | train rank=0 | step=14720/50000 progress=29.4% loss=0.078403 lr=8.453947e-05 lr_backbone=8.453947e-05 elapsed=06:25:54 avg_step=1.57s recent_step=2.02s eta=15:24:55 +2026-09-20 13:04:06 | INFO | train rank=0 | step=14730/50000 progress=29.5% loss=0.103338 lr=8.451555e-05 lr_backbone=8.451555e-05 elapsed=06:26:06 avg_step=1.57s recent_step=1.14s eta=15:24:29 +2026-09-20 13:04:19 | INFO | train rank=0 | step=14740/50000 progress=29.5% loss=0.091759 lr=8.449162e-05 lr_backbone=8.449162e-05 elapsed=06:26:18 avg_step=1.57s recent_step=1.26s eta=15:24:06 +2026-09-20 13:04:33 | INFO | train rank=0 | step=14750/50000 progress=29.5% loss=0.097031 lr=8.446767e-05 lr_backbone=8.446767e-05 elapsed=06:26:32 avg_step=1.57s recent_step=1.41s eta=15:23:46 +2026-09-20 13:04:50 | INFO | train rank=0 | step=14760/50000 progress=29.5% loss=0.087125 lr=8.444371e-05 lr_backbone=8.444371e-05 elapsed=06:26:49 avg_step=1.57s recent_step=1.72s eta=15:23:34 +2026-09-20 13:05:07 | INFO | train rank=0 | step=14770/50000 progress=29.5% loss=0.158974 lr=8.441973e-05 lr_backbone=8.441973e-05 elapsed=06:27:07 avg_step=1.57s recent_step=1.73s eta=15:23:22 +2026-09-20 13:05:26 | INFO | train rank=0 | step=14780/50000 progress=29.6% loss=0.096068 lr=8.439573e-05 lr_backbone=8.439573e-05 elapsed=06:27:25 avg_step=1.57s recent_step=1.87s eta=15:23:13 +2026-09-20 13:05:43 | INFO | train rank=0 | step=14790/50000 progress=29.6% loss=0.105666 lr=8.437172e-05 lr_backbone=8.437172e-05 elapsed=06:27:42 avg_step=1.57s recent_step=1.69s eta=15:23:00 +2026-09-20 13:05:57 | INFO | train rank=0 | step=14800/50000 progress=29.6% loss=0.091480 lr=8.434770e-05 lr_backbone=8.434770e-05 elapsed=06:27:57 avg_step=1.57s recent_step=1.43s eta=15:22:41 +2026-09-20 13:06:13 | INFO | train rank=0 | step=14810/50000 progress=29.6% loss=0.094421 lr=8.432366e-05 lr_backbone=8.432366e-05 elapsed=06:28:12 avg_step=1.57s recent_step=1.59s eta=15:22:26 +2026-09-20 13:06:26 | INFO | train rank=0 | step=14820/50000 progress=29.6% loss=0.098544 lr=8.429961e-05 lr_backbone=8.429961e-05 elapsed=06:28:25 avg_step=1.57s recent_step=1.29s eta=15:22:03 +2026-09-20 13:06:40 | INFO | train rank=0 | step=14830/50000 progress=29.7% loss=0.114267 lr=8.427554e-05 lr_backbone=8.427554e-05 elapsed=06:28:40 avg_step=1.57s recent_step=1.43s eta=15:21:44 +2026-09-20 13:07:00 | INFO | train rank=0 | step=14840/50000 progress=29.7% loss=0.132988 lr=8.425145e-05 lr_backbone=8.425145e-05 elapsed=06:28:59 avg_step=1.57s recent_step=1.97s eta=15:21:38 +2026-09-20 13:07:13 | INFO | train rank=0 | step=14850/50000 progress=29.7% loss=0.093478 lr=8.422736e-05 lr_backbone=8.422736e-05 elapsed=06:29:12 avg_step=1.57s recent_step=1.29s eta=15:21:15 +2026-09-20 13:07:29 | INFO | train rank=0 | step=14860/50000 progress=29.7% loss=0.083080 lr=8.420324e-05 lr_backbone=8.420324e-05 elapsed=06:29:28 avg_step=1.57s recent_step=1.59s eta=15:21:00 +2026-09-20 13:07:47 | INFO | train rank=0 | step=14870/50000 progress=29.7% loss=0.082647 lr=8.417911e-05 lr_backbone=8.417911e-05 elapsed=06:29:47 avg_step=1.57s recent_step=1.84s eta=15:20:51 +2026-09-20 13:08:02 | INFO | train rank=0 | step=14880/50000 progress=29.8% loss=0.111789 lr=8.415497e-05 lr_backbone=8.415497e-05 elapsed=06:30:02 avg_step=1.57s recent_step=1.55s eta=15:20:35 +2026-09-20 13:08:16 | INFO | train rank=0 | step=14890/50000 progress=29.8% loss=0.166852 lr=8.413081e-05 lr_backbone=8.413081e-05 elapsed=06:30:16 avg_step=1.57s recent_step=1.35s eta=15:20:14 +2026-09-20 13:08:32 | INFO | train rank=0 | step=14900/50000 progress=29.8% loss=0.096195 lr=8.410664e-05 lr_backbone=8.410664e-05 elapsed=06:30:32 avg_step=1.57s recent_step=1.62s eta=15:19:59 +2026-09-20 13:08:51 | INFO | train rank=0 | step=14910/50000 progress=29.8% loss=0.098546 lr=8.408245e-05 lr_backbone=8.408245e-05 elapsed=06:30:50 avg_step=1.57s recent_step=1.87s eta=15:19:50 +2026-09-20 13:09:03 | INFO | train rank=0 | step=14920/50000 progress=29.8% loss=0.106876 lr=8.405824e-05 lr_backbone=8.405824e-05 elapsed=06:31:03 avg_step=1.57s recent_step=1.23s eta=15:19:26 +2026-09-20 13:09:17 | INFO | train rank=0 | step=14930/50000 progress=29.9% loss=0.139395 lr=8.403402e-05 lr_backbone=8.403402e-05 elapsed=06:31:16 avg_step=1.57s recent_step=1.36s eta=15:19:06 +2026-09-20 13:09:36 | INFO | train rank=0 | step=14940/50000 progress=29.9% loss=0.107678 lr=8.400979e-05 lr_backbone=8.400979e-05 elapsed=06:31:36 avg_step=1.57s recent_step=1.94s eta=15:18:58 +2026-09-20 13:09:51 | INFO | train rank=0 | step=14950/50000 progress=29.9% loss=0.103676 lr=8.398554e-05 lr_backbone=8.398554e-05 elapsed=06:31:50 avg_step=1.57s recent_step=1.47s eta=15:18:40 +2026-09-20 13:10:06 | INFO | train rank=0 | step=14960/50000 progress=29.9% loss=0.130897 lr=8.396128e-05 lr_backbone=8.396128e-05 elapsed=06:32:06 avg_step=1.57s recent_step=1.56s eta=15:18:24 +2026-09-20 13:10:20 | INFO | train rank=0 | step=14970/50000 progress=29.9% loss=0.099780 lr=8.393700e-05 lr_backbone=8.393700e-05 elapsed=06:32:20 avg_step=1.57s recent_step=1.38s eta=15:18:04 +2026-09-20 13:10:37 | INFO | train rank=0 | step=14980/50000 progress=30.0% loss=0.098228 lr=8.391271e-05 lr_backbone=8.391271e-05 elapsed=06:32:37 avg_step=1.57s recent_step=1.73s eta=15:17:52 +2026-09-20 13:10:55 | INFO | train rank=0 | step=14990/50000 progress=30.0% loss=0.104600 lr=8.388840e-05 lr_backbone=8.388840e-05 elapsed=06:32:54 avg_step=1.57s recent_step=1.73s eta=15:17:40 +2026-09-20 13:11:08 | INFO | train rank=0 | step=15000/50000 progress=30.0% loss=0.088862 lr=8.386408e-05 lr_backbone=8.386408e-05 elapsed=06:33:07 avg_step=1.57s recent_step=1.30s eta=15:17:18 +2026-09-20 13:11:23 | INFO | train rank=0 | step=15010/50000 progress=30.0% loss=0.127620 lr=8.383974e-05 lr_backbone=8.383974e-05 elapsed=06:33:22 avg_step=1.57s recent_step=1.50s eta=15:17:01 +2026-09-20 13:11:41 | INFO | train rank=0 | step=15020/50000 progress=30.0% loss=0.092046 lr=8.381539e-05 lr_backbone=8.381539e-05 elapsed=06:33:41 avg_step=1.57s recent_step=1.86s eta=15:16:51 +2026-09-20 13:11:57 | INFO | train rank=0 | step=15030/50000 progress=30.1% loss=0.114106 lr=8.379102e-05 lr_backbone=8.379102e-05 elapsed=06:33:56 avg_step=1.57s recent_step=1.52s eta=15:16:34 +2026-09-20 13:12:13 | INFO | train rank=0 | step=15040/50000 progress=30.1% loss=0.098182 lr=8.376664e-05 lr_backbone=8.376664e-05 elapsed=06:34:13 avg_step=1.57s recent_step=1.66s eta=15:16:21 +2026-09-20 13:12:26 | INFO | train rank=0 | step=15050/50000 progress=30.1% loss=0.089947 lr=8.374224e-05 lr_backbone=8.374224e-05 elapsed=06:34:26 avg_step=1.57s recent_step=1.28s eta=15:15:58 +2026-09-20 13:12:47 | INFO | train rank=0 | step=15060/50000 progress=30.1% loss=0.092317 lr=8.371783e-05 lr_backbone=8.371783e-05 elapsed=06:34:46 avg_step=1.57s recent_step=2.07s eta=15:15:54 +2026-09-20 13:12:59 | INFO | train rank=0 | step=15070/50000 progress=30.1% loss=0.108082 lr=8.369341e-05 lr_backbone=8.369341e-05 elapsed=06:34:59 avg_step=1.57s recent_step=1.25s eta=15:15:31 +2026-09-20 13:13:17 | INFO | train rank=0 | step=15080/50000 progress=30.2% loss=0.117143 lr=8.366897e-05 lr_backbone=8.366897e-05 elapsed=06:35:17 avg_step=1.57s recent_step=1.78s eta=15:15:20 +2026-09-20 13:13:29 | INFO | train rank=0 | step=15090/50000 progress=30.2% loss=0.121489 lr=8.364451e-05 lr_backbone=8.364451e-05 elapsed=06:35:29 avg_step=1.57s recent_step=1.20s eta=15:14:56 +2026-09-20 13:13:41 | INFO | train rank=0 | step=15100/50000 progress=30.2% loss=0.090209 lr=8.362004e-05 lr_backbone=8.362004e-05 elapsed=06:35:41 avg_step=1.57s recent_step=1.24s eta=15:14:32 +2026-09-20 13:14:04 | INFO | train rank=0 | step=15110/50000 progress=30.2% loss=0.095509 lr=8.359556e-05 lr_backbone=8.359556e-05 elapsed=06:36:04 avg_step=1.57s recent_step=2.31s eta=15:14:34 +2026-09-20 13:14:16 | INFO | train rank=0 | step=15120/50000 progress=30.2% loss=0.099960 lr=8.357106e-05 lr_backbone=8.357106e-05 elapsed=06:36:16 avg_step=1.57s recent_step=1.19s eta=15:14:09 +2026-09-20 13:14:29 | INFO | train rank=0 | step=15130/50000 progress=30.3% loss=0.094835 lr=8.354654e-05 lr_backbone=8.354654e-05 elapsed=06:36:29 avg_step=1.57s recent_step=1.28s eta=15:13:46 +2026-09-20 13:14:46 | INFO | train rank=0 | step=15140/50000 progress=30.3% loss=0.146070 lr=8.352201e-05 lr_backbone=8.352201e-05 elapsed=06:36:46 avg_step=1.57s recent_step=1.69s eta=15:13:33 +2026-09-20 13:15:03 | INFO | train rank=0 | step=15150/50000 progress=30.3% loss=0.096287 lr=8.349747e-05 lr_backbone=8.349747e-05 elapsed=06:37:03 avg_step=1.57s recent_step=1.75s eta=15:13:22 +2026-09-20 13:15:18 | INFO | train rank=0 | step=15160/50000 progress=30.3% loss=0.104594 lr=8.347291e-05 lr_backbone=8.347291e-05 elapsed=06:37:18 avg_step=1.57s recent_step=1.49s eta=15:13:04 +2026-09-20 13:15:37 | INFO | train rank=0 | step=15170/50000 progress=30.3% loss=0.094569 lr=8.344834e-05 lr_backbone=8.344834e-05 elapsed=06:37:36 avg_step=1.57s recent_step=1.82s eta=15:12:54 +2026-09-20 13:15:52 | INFO | train rank=0 | step=15180/50000 progress=30.4% loss=0.119281 lr=8.342375e-05 lr_backbone=8.342375e-05 elapsed=06:37:52 avg_step=1.57s recent_step=1.57s eta=15:12:38 +2026-09-20 13:16:05 | INFO | train rank=0 | step=15190/50000 progress=30.4% loss=0.129203 lr=8.339915e-05 lr_backbone=8.339915e-05 elapsed=06:38:05 avg_step=1.57s recent_step=1.32s eta=15:12:17 +2026-09-20 13:16:17 | INFO | train rank=0 | step=15200/50000 progress=30.4% loss=0.095410 lr=8.337453e-05 lr_backbone=8.337453e-05 elapsed=06:38:17 avg_step=1.57s recent_step=1.19s eta=15:11:52 +2026-09-20 13:16:40 | INFO | train rank=0 | step=15210/50000 progress=30.4% loss=0.084721 lr=8.334990e-05 lr_backbone=8.334990e-05 elapsed=06:38:40 avg_step=1.57s recent_step=2.25s eta=15:11:52 +2026-09-20 13:16:52 | INFO | train rank=0 | step=15220/50000 progress=30.4% loss=0.110147 lr=8.332525e-05 lr_backbone=8.332525e-05 elapsed=06:38:52 avg_step=1.57s recent_step=1.22s eta=15:11:28 +2026-09-20 13:17:05 | INFO | train rank=0 | step=15230/50000 progress=30.5% loss=0.092154 lr=8.330059e-05 lr_backbone=8.330059e-05 elapsed=06:39:04 avg_step=1.57s recent_step=1.26s eta=15:11:05 +2026-09-20 13:17:21 | INFO | train rank=0 | step=15240/50000 progress=30.5% loss=0.098989 lr=8.327592e-05 lr_backbone=8.327592e-05 elapsed=06:39:21 avg_step=1.57s recent_step=1.65s eta=15:10:51 +2026-09-20 13:17:38 | INFO | train rank=0 | step=15250/50000 progress=30.5% loss=0.105639 lr=8.325123e-05 lr_backbone=8.325123e-05 elapsed=06:39:38 avg_step=1.57s recent_step=1.70s eta=15:10:38 +2026-09-20 13:17:57 | INFO | train rank=0 | step=15260/50000 progress=30.5% loss=0.113394 lr=8.322652e-05 lr_backbone=8.322652e-05 elapsed=06:39:56 avg_step=1.57s recent_step=1.86s eta=15:10:29 +2026-09-20 13:18:12 | INFO | train rank=0 | step=15270/50000 progress=30.5% loss=0.105091 lr=8.320181e-05 lr_backbone=8.320181e-05 elapsed=06:40:12 avg_step=1.57s recent_step=1.52s eta=15:10:12 +2026-09-20 13:18:25 | INFO | train rank=0 | step=15280/50000 progress=30.6% loss=0.078631 lr=8.317707e-05 lr_backbone=8.317707e-05 elapsed=06:40:25 avg_step=1.57s recent_step=1.30s eta=15:09:50 +2026-09-20 13:18:41 | INFO | train rank=0 | step=15290/50000 progress=30.6% loss=0.078656 lr=8.315232e-05 lr_backbone=8.315232e-05 elapsed=06:40:41 avg_step=1.57s recent_step=1.63s eta=15:09:36 +2026-09-20 13:18:55 | INFO | train rank=0 | step=15300/50000 progress=30.6% loss=0.095488 lr=8.312756e-05 lr_backbone=8.312756e-05 elapsed=06:40:55 avg_step=1.57s recent_step=1.37s eta=15:09:16 +2026-09-20 13:19:13 | INFO | train rank=0 | step=15310/50000 progress=30.6% loss=0.099423 lr=8.310279e-05 lr_backbone=8.310279e-05 elapsed=06:41:13 avg_step=1.57s recent_step=1.81s eta=15:09:05 +2026-09-20 13:19:31 | INFO | train rank=0 | step=15320/50000 progress=30.6% loss=0.098818 lr=8.307799e-05 lr_backbone=8.307799e-05 elapsed=06:41:30 avg_step=1.57s recent_step=1.77s eta=15:08:54 +2026-09-20 13:19:42 | INFO | train rank=0 | step=15330/50000 progress=30.7% loss=0.111289 lr=8.305319e-05 lr_backbone=8.305319e-05 elapsed=06:41:42 avg_step=1.57s recent_step=1.18s eta=15:08:29 +2026-09-20 13:19:57 | INFO | train rank=0 | step=15340/50000 progress=30.7% loss=0.113974 lr=8.302837e-05 lr_backbone=8.302837e-05 elapsed=06:41:56 avg_step=1.57s recent_step=1.41s eta=15:08:10 +2026-09-20 13:20:16 | INFO | train rank=0 | step=15350/50000 progress=30.7% loss=0.097478 lr=8.300353e-05 lr_backbone=8.300353e-05 elapsed=06:42:16 avg_step=1.57s recent_step=1.98s eta=15:08:04 +2026-09-20 13:20:30 | INFO | train rank=0 | step=15360/50000 progress=30.7% loss=0.088678 lr=8.297868e-05 lr_backbone=8.297868e-05 elapsed=06:42:30 avg_step=1.57s recent_step=1.39s eta=15:07:44 +2026-09-20 13:20:50 | INFO | train rank=0 | step=15370/50000 progress=30.7% loss=0.099821 lr=8.295382e-05 lr_backbone=8.295382e-05 elapsed=06:42:49 avg_step=1.57s recent_step=1.93s eta=15:07:36 +2026-09-20 13:21:01 | INFO | train rank=0 | step=15380/50000 progress=30.8% loss=0.123752 lr=8.292894e-05 lr_backbone=8.292894e-05 elapsed=06:43:01 avg_step=1.57s recent_step=1.18s eta=15:07:12 +2026-09-20 13:21:15 | INFO | train rank=0 | step=15390/50000 progress=30.8% loss=0.084564 lr=8.290405e-05 lr_backbone=8.290405e-05 elapsed=06:43:15 avg_step=1.57s recent_step=1.39s eta=15:06:52 +2026-09-20 13:21:33 | INFO | train rank=0 | step=15400/50000 progress=30.8% loss=0.115150 lr=8.287914e-05 lr_backbone=8.287914e-05 elapsed=06:43:32 avg_step=1.57s recent_step=1.73s eta=15:06:40 +2026-09-20 13:21:51 | INFO | train rank=0 | step=15410/50000 progress=30.8% loss=0.134182 lr=8.285422e-05 lr_backbone=8.285422e-05 elapsed=06:43:50 avg_step=1.57s recent_step=1.80s eta=15:06:29 +2026-09-20 13:22:08 | INFO | train rank=0 | step=15420/50000 progress=30.8% loss=0.101904 lr=8.282929e-05 lr_backbone=8.282929e-05 elapsed=06:44:08 avg_step=1.57s recent_step=1.77s eta=15:06:18 +2026-09-20 13:22:21 | INFO | train rank=0 | step=15430/50000 progress=30.9% loss=0.094213 lr=8.280434e-05 lr_backbone=8.280434e-05 elapsed=06:44:20 avg_step=1.57s recent_step=1.22s eta=15:05:54 +2026-09-20 13:22:37 | INFO | train rank=0 | step=15440/50000 progress=30.9% loss=0.087341 lr=8.277937e-05 lr_backbone=8.277937e-05 elapsed=06:44:37 avg_step=1.57s recent_step=1.67s eta=15:05:41 +2026-09-20 13:22:50 | INFO | train rank=0 | step=15450/50000 progress=30.9% loss=0.151311 lr=8.275440e-05 lr_backbone=8.275440e-05 elapsed=06:44:49 avg_step=1.57s recent_step=1.24s eta=15:05:17 +2026-09-20 13:23:12 | INFO | train rank=0 | step=15460/50000 progress=30.9% loss=0.140351 lr=8.272940e-05 lr_backbone=8.272940e-05 elapsed=06:45:11 avg_step=1.57s recent_step=2.20s eta=15:05:16 +2026-09-20 13:23:24 | INFO | train rank=0 | step=15470/50000 progress=30.9% loss=0.090078 lr=8.270440e-05 lr_backbone=8.270440e-05 elapsed=06:45:23 avg_step=1.57s recent_step=1.19s eta=15:04:51 +2026-09-20 13:23:45 | INFO | train rank=0 | step=15480/50000 progress=31.0% loss=0.120203 lr=8.267937e-05 lr_backbone=8.267937e-05 elapsed=06:45:44 avg_step=1.57s recent_step=2.11s eta=15:04:48 +2026-09-20 13:23:56 | INFO | train rank=0 | step=15490/50000 progress=31.0% loss=0.113659 lr=8.265434e-05 lr_backbone=8.265434e-05 elapsed=06:45:56 avg_step=1.57s recent_step=1.17s eta=15:04:23 +2026-09-20 13:24:14 | INFO | train rank=0 | step=15500/50000 progress=31.0% loss=0.094452 lr=8.262929e-05 lr_backbone=8.262929e-05 elapsed=06:46:14 avg_step=1.57s recent_step=1.78s eta=15:04:12 +2026-09-20 13:24:27 | INFO | train rank=0 | step=15510/50000 progress=31.0% loss=0.091227 lr=8.260422e-05 lr_backbone=8.260422e-05 elapsed=06:46:27 avg_step=1.57s recent_step=1.28s eta=15:03:50 +2026-09-20 13:24:39 | INFO | train rank=0 | step=15520/50000 progress=31.0% loss=0.096592 lr=8.257915e-05 lr_backbone=8.257915e-05 elapsed=06:46:39 avg_step=1.57s recent_step=1.21s eta=15:03:26 +2026-09-20 13:25:00 | INFO | train rank=0 | step=15530/50000 progress=31.1% loss=0.109274 lr=8.255405e-05 lr_backbone=8.255405e-05 elapsed=06:46:59 avg_step=1.57s recent_step=2.05s eta=15:03:21 +2026-09-20 13:25:17 | INFO | train rank=0 | step=15540/50000 progress=31.1% loss=0.097502 lr=8.252895e-05 lr_backbone=8.252895e-05 elapsed=06:47:17 avg_step=1.57s recent_step=1.77s eta=15:03:10 +2026-09-20 13:25:29 | INFO | train rank=0 | step=15550/50000 progress=31.1% loss=0.124166 lr=8.250382e-05 lr_backbone=8.250382e-05 elapsed=06:47:29 avg_step=1.57s recent_step=1.20s eta=15:02:46 +2026-09-20 13:25:47 | INFO | train rank=0 | step=15560/50000 progress=31.1% loss=0.093669 lr=8.247869e-05 lr_backbone=8.247869e-05 elapsed=06:47:47 avg_step=1.57s recent_step=1.77s eta=15:02:34 +2026-09-20 13:26:01 | INFO | train rank=0 | step=15570/50000 progress=31.1% loss=0.139824 lr=8.245354e-05 lr_backbone=8.245354e-05 elapsed=06:48:00 avg_step=1.57s recent_step=1.38s eta=15:02:14 +2026-09-20 13:26:18 | INFO | train rank=0 | step=15580/50000 progress=31.2% loss=0.135535 lr=8.242837e-05 lr_backbone=8.242837e-05 elapsed=06:48:18 avg_step=1.57s recent_step=1.72s eta=15:02:02 +2026-09-20 13:26:32 | INFO | train rank=0 | step=15590/50000 progress=31.2% loss=0.102929 lr=8.240320e-05 lr_backbone=8.240320e-05 elapsed=06:48:32 avg_step=1.57s recent_step=1.43s eta=15:01:43 +2026-09-20 13:26:50 | INFO | train rank=0 | step=15600/50000 progress=31.2% loss=0.108939 lr=8.237800e-05 lr_backbone=8.237800e-05 elapsed=06:48:50 avg_step=1.57s recent_step=1.80s eta=15:01:32 +2026-09-20 13:27:03 | INFO | train rank=0 | step=15610/50000 progress=31.2% loss=0.084379 lr=8.235280e-05 lr_backbone=8.235280e-05 elapsed=06:49:03 avg_step=1.57s recent_step=1.30s eta=15:01:11 +2026-09-20 13:27:23 | INFO | train rank=0 | step=15620/50000 progress=31.2% loss=0.118598 lr=8.232758e-05 lr_backbone=8.232758e-05 elapsed=06:49:23 avg_step=1.57s recent_step=1.98s eta=15:01:04 +2026-09-20 13:27:41 | INFO | train rank=0 | step=15630/50000 progress=31.3% loss=0.079847 lr=8.230234e-05 lr_backbone=8.230234e-05 elapsed=06:49:41 avg_step=1.57s recent_step=1.77s eta=15:00:53 +2026-09-20 13:27:56 | INFO | train rank=0 | step=15640/50000 progress=31.3% loss=0.097634 lr=8.227709e-05 lr_backbone=8.227709e-05 elapsed=06:49:56 avg_step=1.57s recent_step=1.50s eta=15:00:35 +2026-09-20 13:28:08 | INFO | train rank=0 | step=15650/50000 progress=31.3% loss=0.122965 lr=8.225183e-05 lr_backbone=8.225183e-05 elapsed=06:50:08 avg_step=1.57s recent_step=1.23s eta=15:00:12 +2026-09-20 13:28:26 | INFO | train rank=0 | step=15660/50000 progress=31.3% loss=0.107887 lr=8.222655e-05 lr_backbone=8.222655e-05 elapsed=06:50:26 avg_step=1.57s recent_step=1.78s eta=15:00:01 +2026-09-20 13:28:38 | INFO | train rank=0 | step=15670/50000 progress=31.3% loss=0.098966 lr=8.220126e-05 lr_backbone=8.220126e-05 elapsed=06:50:38 avg_step=1.57s recent_step=1.23s eta=14:59:38 +2026-09-20 13:28:52 | INFO | train rank=0 | step=15680/50000 progress=31.4% loss=0.096187 lr=8.217596e-05 lr_backbone=8.217596e-05 elapsed=06:50:52 avg_step=1.57s recent_step=1.37s eta=14:59:18 +2026-09-20 13:29:08 | INFO | train rank=0 | step=15690/50000 progress=31.4% loss=0.102657 lr=8.215064e-05 lr_backbone=8.215064e-05 elapsed=06:51:07 avg_step=1.57s recent_step=1.57s eta=14:59:02 +2026-09-20 13:29:25 | INFO | train rank=0 | step=15700/50000 progress=31.4% loss=0.104519 lr=8.212530e-05 lr_backbone=8.212530e-05 elapsed=06:51:25 avg_step=1.57s recent_step=1.75s eta=14:58:50 +2026-09-20 13:29:39 | INFO | train rank=0 | step=15710/50000 progress=31.4% loss=0.085479 lr=8.209996e-05 lr_backbone=8.209996e-05 elapsed=06:51:38 avg_step=1.57s recent_step=1.36s eta=14:58:29 +2026-09-20 13:29:54 | INFO | train rank=0 | step=15720/50000 progress=31.4% loss=0.104844 lr=8.207460e-05 lr_backbone=8.207460e-05 elapsed=06:51:53 avg_step=1.57s recent_step=1.47s eta=14:58:12 +2026-09-20 13:30:07 | INFO | train rank=0 | step=15730/50000 progress=31.5% loss=0.101867 lr=8.204922e-05 lr_backbone=8.204922e-05 elapsed=06:52:07 avg_step=1.57s recent_step=1.35s eta=14:57:51 +2026-09-20 13:30:28 | INFO | train rank=0 | step=15740/50000 progress=31.5% loss=0.100280 lr=8.202383e-05 lr_backbone=8.202383e-05 elapsed=06:52:28 avg_step=1.57s recent_step=2.11s eta=14:57:47 +2026-09-20 13:30:47 | INFO | train rank=0 | step=15750/50000 progress=31.5% loss=0.174725 lr=8.199843e-05 lr_backbone=8.199843e-05 elapsed=06:52:47 avg_step=1.57s recent_step=1.94s eta=14:57:39 +2026-09-20 13:31:00 | INFO | train rank=0 | step=15760/50000 progress=31.5% loss=0.115992 lr=8.197301e-05 lr_backbone=8.197301e-05 elapsed=06:52:59 avg_step=1.57s recent_step=1.20s eta=14:57:15 +2026-09-20 13:31:11 | INFO | train rank=0 | step=15770/50000 progress=31.5% loss=0.087773 lr=8.194758e-05 lr_backbone=8.194758e-05 elapsed=06:53:11 avg_step=1.57s recent_step=1.15s eta=14:56:51 +2026-09-20 13:31:26 | INFO | train rank=0 | step=15780/50000 progress=31.6% loss=0.082332 lr=8.192213e-05 lr_backbone=8.192213e-05 elapsed=06:53:26 avg_step=1.57s recent_step=1.54s eta=14:56:34 +2026-09-20 13:31:43 | INFO | train rank=0 | step=15790/50000 progress=31.6% loss=0.084410 lr=8.189667e-05 lr_backbone=8.189667e-05 elapsed=06:53:43 avg_step=1.57s recent_step=1.68s eta=14:56:21 +2026-09-20 13:31:58 | INFO | train rank=0 | step=15800/50000 progress=31.6% loss=0.098393 lr=8.187120e-05 lr_backbone=8.187120e-05 elapsed=06:53:58 avg_step=1.57s recent_step=1.46s eta=14:56:03 +2026-09-20 13:32:14 | INFO | train rank=0 | step=15810/50000 progress=31.6% loss=0.089382 lr=8.184571e-05 lr_backbone=8.184571e-05 elapsed=06:54:14 avg_step=1.57s recent_step=1.63s eta=14:55:48 +2026-09-20 13:32:32 | INFO | train rank=0 | step=15820/50000 progress=31.6% loss=0.112832 lr=8.182021e-05 lr_backbone=8.182021e-05 elapsed=06:54:32 avg_step=1.57s recent_step=1.81s eta=14:55:38 +2026-09-20 13:32:48 | INFO | train rank=0 | step=15830/50000 progress=31.7% loss=0.085686 lr=8.179470e-05 lr_backbone=8.179470e-05 elapsed=06:54:48 avg_step=1.57s recent_step=1.61s eta=14:55:23 +2026-09-20 13:33:01 | INFO | train rank=0 | step=15840/50000 progress=31.7% loss=0.098521 lr=8.176917e-05 lr_backbone=8.176917e-05 elapsed=06:55:00 avg_step=1.57s recent_step=1.24s eta=14:55:00 +2026-09-20 13:33:23 | INFO | train rank=0 | step=15850/50000 progress=31.7% loss=0.098957 lr=8.174362e-05 lr_backbone=8.174362e-05 elapsed=06:55:22 avg_step=1.57s recent_step=2.21s eta=14:54:58 +2026-09-20 13:33:35 | INFO | train rank=0 | step=15860/50000 progress=31.7% loss=0.082070 lr=8.171807e-05 lr_backbone=8.171807e-05 elapsed=06:55:34 avg_step=1.57s recent_step=1.19s eta=14:54:34 +2026-09-20 13:33:53 | INFO | train rank=0 | step=15870/50000 progress=31.7% loss=0.110463 lr=8.169250e-05 lr_backbone=8.169250e-05 elapsed=06:55:52 avg_step=1.57s recent_step=1.81s eta=14:54:23 +2026-09-20 13:34:05 | INFO | train rank=0 | step=15880/50000 progress=31.8% loss=0.100047 lr=8.166691e-05 lr_backbone=8.166691e-05 elapsed=06:56:05 avg_step=1.57s recent_step=1.25s eta=14:54:01 +2026-09-20 13:34:19 | INFO | train rank=0 | step=15890/50000 progress=31.8% loss=0.117255 lr=8.164131e-05 lr_backbone=8.164131e-05 elapsed=06:56:18 avg_step=1.57s recent_step=1.35s eta=14:53:40 +2026-09-20 13:34:37 | INFO | train rank=0 | step=15900/50000 progress=31.8% loss=0.103495 lr=8.161570e-05 lr_backbone=8.161570e-05 elapsed=06:56:36 avg_step=1.57s recent_step=1.79s eta=14:53:29 +2026-09-20 13:34:59 | INFO | train rank=0 | step=15910/50000 progress=31.8% loss=0.130645 lr=8.159007e-05 lr_backbone=8.159007e-05 elapsed=06:56:58 avg_step=1.57s recent_step=2.20s eta=14:53:27 +2026-09-20 13:35:11 | INFO | train rank=0 | step=15920/50000 progress=31.8% loss=0.172969 lr=8.156443e-05 lr_backbone=8.156443e-05 elapsed=06:57:10 avg_step=1.57s recent_step=1.20s eta=14:53:03 +2026-09-20 13:35:23 | INFO | train rank=0 | step=15930/50000 progress=31.9% loss=0.097714 lr=8.153878e-05 lr_backbone=8.153878e-05 elapsed=06:57:23 avg_step=1.57s recent_step=1.28s eta=14:52:41 +2026-09-20 13:35:39 | INFO | train rank=0 | step=15940/50000 progress=31.9% loss=0.089754 lr=8.151311e-05 lr_backbone=8.151311e-05 elapsed=06:57:38 avg_step=1.57s recent_step=1.53s eta=14:52:24 +2026-09-20 13:36:02 | INFO | train rank=0 | step=15950/50000 progress=31.9% loss=0.113071 lr=8.148743e-05 lr_backbone=8.148743e-05 elapsed=06:58:02 avg_step=1.57s recent_step=2.33s eta=14:52:25 +2026-09-20 13:36:14 | INFO | train rank=0 | step=15960/50000 progress=31.9% loss=0.104515 lr=8.146174e-05 lr_backbone=8.146174e-05 elapsed=06:58:13 avg_step=1.57s recent_step=1.17s eta=14:52:01 +2026-09-20 13:36:26 | INFO | train rank=0 | step=15970/50000 progress=31.9% loss=0.082094 lr=8.143603e-05 lr_backbone=8.143603e-05 elapsed=06:58:26 avg_step=1.57s recent_step=1.22s eta=14:51:38 +2026-09-20 13:36:42 | INFO | train rank=0 | step=15980/50000 progress=32.0% loss=0.112963 lr=8.141030e-05 lr_backbone=8.141030e-05 elapsed=06:58:41 avg_step=1.57s recent_step=1.54s eta=14:51:21 +2026-09-20 13:37:04 | INFO | train rank=0 | step=15990/50000 progress=32.0% loss=0.101808 lr=8.138457e-05 lr_backbone=8.138457e-05 elapsed=06:59:04 avg_step=1.57s recent_step=2.29s eta=14:51:21 +2026-09-20 13:37:16 | INFO | train rank=0 | step=16000/50000 progress=32.0% loss=0.097046 lr=8.135882e-05 lr_backbone=8.135882e-05 elapsed=06:59:16 avg_step=1.57s recent_step=1.16s eta=14:50:56 +2026-09-20 13:37:29 | INFO | train rank=0 | step=16010/50000 progress=32.0% loss=0.087038 lr=8.133305e-05 lr_backbone=8.133305e-05 elapsed=06:59:28 avg_step=1.57s recent_step=1.28s eta=14:50:34 +2026-09-20 13:37:44 | INFO | train rank=0 | step=16020/50000 progress=32.0% loss=0.095981 lr=8.130728e-05 lr_backbone=8.130728e-05 elapsed=06:59:44 avg_step=1.57s recent_step=1.53s eta=14:50:18 +2026-09-20 13:38:04 | INFO | train rank=0 | step=16030/50000 progress=32.1% loss=0.089228 lr=8.128149e-05 lr_backbone=8.128149e-05 elapsed=07:00:03 avg_step=1.57s recent_step=1.94s eta=14:50:10 +2026-09-20 13:38:16 | INFO | train rank=0 | step=16040/50000 progress=32.1% loss=0.119753 lr=8.125568e-05 lr_backbone=8.125568e-05 elapsed=07:00:15 avg_step=1.57s recent_step=1.22s eta=14:49:47 +2026-09-20 13:38:31 | INFO | train rank=0 | step=16050/50000 progress=32.1% loss=0.093484 lr=8.122986e-05 lr_backbone=8.122986e-05 elapsed=07:00:30 avg_step=1.57s recent_step=1.50s eta=14:49:30 +2026-09-20 13:38:46 | INFO | train rank=0 | step=16060/50000 progress=32.1% loss=0.091846 lr=8.120403e-05 lr_backbone=8.120403e-05 elapsed=07:00:45 avg_step=1.57s recent_step=1.50s eta=14:49:12 +2026-09-20 13:39:04 | INFO | train rank=0 | step=16070/50000 progress=32.1% loss=0.147088 lr=8.117818e-05 lr_backbone=8.117818e-05 elapsed=07:01:03 avg_step=1.57s recent_step=1.79s eta=14:49:01 +2026-09-20 13:39:16 | INFO | train rank=0 | step=16080/50000 progress=32.2% loss=0.079703 lr=8.115232e-05 lr_backbone=8.115232e-05 elapsed=07:01:15 avg_step=1.57s recent_step=1.19s eta=14:48:37 +2026-09-20 13:39:31 | INFO | train rank=0 | step=16090/50000 progress=32.2% loss=0.074542 lr=8.112645e-05 lr_backbone=8.112645e-05 elapsed=07:01:30 avg_step=1.57s recent_step=1.53s eta=14:48:21 +2026-09-20 13:39:42 | INFO | train rank=0 | step=16100/50000 progress=32.2% loss=0.083569 lr=8.110056e-05 lr_backbone=8.110056e-05 elapsed=07:01:42 avg_step=1.57s recent_step=1.16s eta=14:47:56 +2026-09-20 13:40:01 | INFO | train rank=0 | step=16110/50000 progress=32.2% loss=0.127374 lr=8.107466e-05 lr_backbone=8.107466e-05 elapsed=07:02:01 avg_step=1.57s recent_step=1.87s eta=14:47:47 +2026-09-20 13:40:16 | INFO | train rank=0 | step=16120/50000 progress=32.2% loss=0.081921 lr=8.104875e-05 lr_backbone=8.104875e-05 elapsed=07:02:16 avg_step=1.57s recent_step=1.49s eta=14:47:29 +2026-09-20 13:40:38 | INFO | train rank=0 | step=16130/50000 progress=32.3% loss=0.087827 lr=8.102282e-05 lr_backbone=8.102282e-05 elapsed=07:02:37 avg_step=1.57s recent_step=2.16s eta=14:47:26 +2026-09-20 13:40:50 | INFO | train rank=0 | step=16140/50000 progress=32.3% loss=0.125496 lr=8.099688e-05 lr_backbone=8.099688e-05 elapsed=07:02:49 avg_step=1.57s recent_step=1.22s eta=14:47:03 +2026-09-20 13:41:05 | INFO | train rank=0 | step=16150/50000 progress=32.3% loss=0.131988 lr=8.097093e-05 lr_backbone=8.097093e-05 elapsed=07:03:05 avg_step=1.57s recent_step=1.53s eta=14:46:46 +2026-09-20 13:41:17 | INFO | train rank=0 | step=16160/50000 progress=32.3% loss=0.113815 lr=8.094496e-05 lr_backbone=8.094496e-05 elapsed=07:03:17 avg_step=1.57s recent_step=1.20s eta=14:46:23 +2026-09-20 13:41:33 | INFO | train rank=0 | step=16170/50000 progress=32.3% loss=0.124536 lr=8.091898e-05 lr_backbone=8.091898e-05 elapsed=07:03:32 avg_step=1.57s recent_step=1.54s eta=14:46:06 +2026-09-20 13:41:48 | INFO | train rank=0 | step=16180/50000 progress=32.4% loss=0.117135 lr=8.089298e-05 lr_backbone=8.089298e-05 elapsed=07:03:48 avg_step=1.57s recent_step=1.55s eta=14:45:50 +2026-09-20 13:42:05 | INFO | train rank=0 | step=16190/50000 progress=32.4% loss=0.132132 lr=8.086697e-05 lr_backbone=8.086697e-05 elapsed=07:04:04 avg_step=1.57s recent_step=1.68s eta=14:45:37 +2026-09-20 13:42:23 | INFO | train rank=0 | step=16200/50000 progress=32.4% loss=0.113563 lr=8.084095e-05 lr_backbone=8.084095e-05 elapsed=07:04:23 avg_step=1.57s recent_step=1.87s eta=14:45:27 +2026-09-20 13:42:39 | INFO | train rank=0 | step=16210/50000 progress=32.4% loss=0.107295 lr=8.081491e-05 lr_backbone=8.081491e-05 elapsed=07:04:39 avg_step=1.57s recent_step=1.56s eta=14:45:11 +2026-09-20 13:42:57 | INFO | train rank=0 | step=16220/50000 progress=32.4% loss=0.088044 lr=8.078886e-05 lr_backbone=8.078886e-05 elapsed=07:04:56 avg_step=1.57s recent_step=1.78s eta=14:45:00 +2026-09-20 13:43:09 | INFO | train rank=0 | step=16230/50000 progress=32.5% loss=0.083077 lr=8.076280e-05 lr_backbone=8.076280e-05 elapsed=07:05:09 avg_step=1.57s recent_step=1.23s eta=14:44:37 +2026-09-20 13:43:21 | INFO | train rank=0 | step=16240/50000 progress=32.5% loss=0.098722 lr=8.073673e-05 lr_backbone=8.073673e-05 elapsed=07:05:21 avg_step=1.57s recent_step=1.20s eta=14:44:14 +2026-09-20 13:43:38 | INFO | train rank=0 | step=16250/50000 progress=32.5% loss=0.115888 lr=8.071064e-05 lr_backbone=8.071064e-05 elapsed=07:05:37 avg_step=1.57s recent_step=1.65s eta=14:44:00 +2026-09-20 13:43:53 | INFO | train rank=0 | step=16260/50000 progress=32.5% loss=0.071155 lr=8.068453e-05 lr_backbone=8.068453e-05 elapsed=07:05:53 avg_step=1.57s recent_step=1.52s eta=14:43:43 +2026-09-20 13:44:12 | INFO | train rank=0 | step=16270/50000 progress=32.5% loss=0.127575 lr=8.065842e-05 lr_backbone=8.065842e-05 elapsed=07:06:11 avg_step=1.57s recent_step=1.87s eta=14:43:33 +2026-09-20 13:44:27 | INFO | train rank=0 | step=16280/50000 progress=32.6% loss=0.099595 lr=8.063229e-05 lr_backbone=8.063229e-05 elapsed=07:06:26 avg_step=1.57s recent_step=1.52s eta=14:43:16 +2026-09-20 13:44:45 | INFO | train rank=0 | step=16290/50000 progress=32.6% loss=0.081883 lr=8.060614e-05 lr_backbone=8.060614e-05 elapsed=07:06:44 avg_step=1.57s recent_step=1.81s eta=14:43:06 +2026-09-20 13:44:57 | INFO | train rank=0 | step=16300/50000 progress=32.6% loss=0.093638 lr=8.057999e-05 lr_backbone=8.057999e-05 elapsed=07:06:57 avg_step=1.57s recent_step=1.21s eta=14:42:42 +2026-09-20 13:45:17 | INFO | train rank=0 | step=16310/50000 progress=32.6% loss=0.139664 lr=8.055382e-05 lr_backbone=8.055382e-05 elapsed=07:07:16 avg_step=1.57s recent_step=1.99s eta=14:42:35 +2026-09-20 13:45:30 | INFO | train rank=0 | step=16320/50000 progress=32.6% loss=0.090531 lr=8.052763e-05 lr_backbone=8.052763e-05 elapsed=07:07:30 avg_step=1.57s recent_step=1.34s eta=14:42:15 +2026-09-20 13:45:50 | INFO | train rank=0 | step=16330/50000 progress=32.7% loss=0.099918 lr=8.050144e-05 lr_backbone=8.050144e-05 elapsed=07:07:50 avg_step=1.57s recent_step=1.98s eta=14:42:08 +2026-09-20 13:46:02 | INFO | train rank=0 | step=16340/50000 progress=32.7% loss=0.114231 lr=8.047523e-05 lr_backbone=8.047523e-05 elapsed=07:08:01 avg_step=1.57s recent_step=1.18s eta=14:41:44 +2026-09-20 13:46:19 | INFO | train rank=0 | step=16350/50000 progress=32.7% loss=0.102936 lr=8.044900e-05 lr_backbone=8.044900e-05 elapsed=07:08:19 avg_step=1.57s recent_step=1.71s eta=14:41:31 +2026-09-20 13:46:32 | INFO | train rank=0 | step=16360/50000 progress=32.7% loss=0.100243 lr=8.042277e-05 lr_backbone=8.042277e-05 elapsed=07:08:31 avg_step=1.57s recent_step=1.27s eta=14:41:09 +2026-09-20 13:46:49 | INFO | train rank=0 | step=16370/50000 progress=32.7% loss=0.100822 lr=8.039651e-05 lr_backbone=8.039651e-05 elapsed=07:08:48 avg_step=1.57s recent_step=1.70s eta=14:40:56 +2026-09-20 13:47:03 | INFO | train rank=0 | step=16380/50000 progress=32.8% loss=0.104518 lr=8.037025e-05 lr_backbone=8.037025e-05 elapsed=07:09:02 avg_step=1.57s recent_step=1.41s eta=14:40:37 +2026-09-20 13:47:24 | INFO | train rank=0 | step=16390/50000 progress=32.8% loss=0.097860 lr=8.034397e-05 lr_backbone=8.034397e-05 elapsed=07:09:23 avg_step=1.57s recent_step=2.09s eta=14:40:32 +2026-09-20 13:47:36 | INFO | train rank=0 | step=16400/50000 progress=32.8% loss=0.096910 lr=8.031768e-05 lr_backbone=8.031768e-05 elapsed=07:09:35 avg_step=1.57s recent_step=1.18s eta=14:40:08 +2026-09-20 13:47:51 | INFO | train rank=0 | step=16410/50000 progress=32.8% loss=0.234530 lr=8.029138e-05 lr_backbone=8.029138e-05 elapsed=07:09:51 avg_step=1.57s recent_step=1.58s eta=14:39:52 +2026-09-20 13:48:06 | INFO | train rank=0 | step=16420/50000 progress=32.8% loss=0.165047 lr=8.026507e-05 lr_backbone=8.026507e-05 elapsed=07:10:06 avg_step=1.57s recent_step=1.52s eta=14:39:36 +2026-09-20 13:48:28 | INFO | train rank=0 | step=16430/50000 progress=32.9% loss=0.087999 lr=8.023874e-05 lr_backbone=8.023874e-05 elapsed=07:10:27 avg_step=1.57s recent_step=2.11s eta=14:39:31 +2026-09-20 13:48:39 | INFO | train rank=0 | step=16440/50000 progress=32.9% loss=0.105286 lr=8.021239e-05 lr_backbone=8.021239e-05 elapsed=07:10:39 avg_step=1.57s recent_step=1.18s eta=14:39:07 +2026-09-20 13:48:53 | INFO | train rank=0 | step=16450/50000 progress=32.9% loss=0.105810 lr=8.018604e-05 lr_backbone=8.018604e-05 elapsed=07:10:52 avg_step=1.57s recent_step=1.33s eta=14:38:47 +2026-09-20 13:49:04 | INFO | train rank=0 | step=16460/50000 progress=32.9% loss=0.092667 lr=8.015967e-05 lr_backbone=8.015967e-05 elapsed=07:11:04 avg_step=1.57s recent_step=1.17s eta=14:38:23 +2026-09-20 13:49:27 | INFO | train rank=0 | step=16470/50000 progress=32.9% loss=0.098421 lr=8.013328e-05 lr_backbone=8.013328e-05 elapsed=07:11:27 avg_step=1.57s recent_step=2.31s eta=14:38:22 +2026-09-20 13:49:45 | INFO | train rank=0 | step=16480/50000 progress=33.0% loss=0.142373 lr=8.010689e-05 lr_backbone=8.010689e-05 elapsed=07:11:44 avg_step=1.57s recent_step=1.72s eta=14:38:09 +2026-09-20 13:49:56 | INFO | train rank=0 | step=16490/50000 progress=33.0% loss=0.077295 lr=8.008048e-05 lr_backbone=8.008048e-05 elapsed=07:11:56 avg_step=1.57s recent_step=1.15s eta=14:37:45 +2026-09-20 13:50:08 | INFO | train rank=0 | step=16500/50000 progress=33.0% loss=0.127498 lr=8.005406e-05 lr_backbone=8.005406e-05 elapsed=07:12:08 avg_step=1.57s recent_step=1.22s eta=14:37:22 +2026-09-20 13:50:27 | INFO | train rank=0 | step=16510/50000 progress=33.0% loss=0.146616 lr=8.002762e-05 lr_backbone=8.002762e-05 elapsed=07:12:27 avg_step=1.57s recent_step=1.91s eta=14:37:13 +2026-09-20 13:50:42 | INFO | train rank=0 | step=16520/50000 progress=33.0% loss=0.103994 lr=8.000117e-05 lr_backbone=8.000117e-05 elapsed=07:12:42 avg_step=1.57s recent_step=1.50s eta=14:36:56 +2026-09-20 13:51:00 | INFO | train rank=0 | step=16530/50000 progress=33.1% loss=0.083107 lr=7.997471e-05 lr_backbone=7.997471e-05 elapsed=07:13:00 avg_step=1.57s recent_step=1.75s eta=14:36:44 +2026-09-20 13:51:13 | INFO | train rank=0 | step=16540/50000 progress=33.1% loss=0.102345 lr=7.994824e-05 lr_backbone=7.994824e-05 elapsed=07:13:13 avg_step=1.57s recent_step=1.32s eta=14:36:23 +2026-09-20 13:51:35 | INFO | train rank=0 | step=16550/50000 progress=33.1% loss=0.100084 lr=7.992175e-05 lr_backbone=7.992175e-05 elapsed=07:13:35 avg_step=1.57s recent_step=2.20s eta=14:36:20 +2026-09-20 13:51:52 | INFO | train rank=0 | step=16560/50000 progress=33.1% loss=0.108531 lr=7.989525e-05 lr_backbone=7.989525e-05 elapsed=07:13:52 avg_step=1.57s recent_step=1.72s eta=14:36:07 +2026-09-20 13:52:04 | INFO | train rank=0 | step=16570/50000 progress=33.1% loss=0.107406 lr=7.986874e-05 lr_backbone=7.986874e-05 elapsed=07:14:04 avg_step=1.57s recent_step=1.21s eta=14:35:44 +2026-09-20 13:52:16 | INFO | train rank=0 | step=16580/50000 progress=33.2% loss=0.093898 lr=7.984221e-05 lr_backbone=7.984221e-05 elapsed=07:14:16 avg_step=1.57s recent_step=1.20s eta=14:35:21 +2026-09-20 13:52:36 | INFO | train rank=0 | step=16590/50000 progress=33.2% loss=0.120886 lr=7.981567e-05 lr_backbone=7.981567e-05 elapsed=07:14:35 avg_step=1.57s recent_step=1.90s eta=14:35:12 +2026-09-20 13:52:52 | INFO | train rank=0 | step=16600/50000 progress=33.2% loss=0.093807 lr=7.978912e-05 lr_backbone=7.978912e-05 elapsed=07:14:51 avg_step=1.57s recent_step=1.64s eta=14:34:58 +2026-09-20 13:53:03 | INFO | train rank=0 | step=16610/50000 progress=33.2% loss=0.106190 lr=7.976255e-05 lr_backbone=7.976255e-05 elapsed=07:15:03 avg_step=1.57s recent_step=1.16s eta=14:34:34 +2026-09-20 13:53:20 | INFO | train rank=0 | step=16620/50000 progress=33.2% loss=0.076015 lr=7.973597e-05 lr_backbone=7.973597e-05 elapsed=07:15:19 avg_step=1.57s recent_step=1.62s eta=14:34:19 +2026-09-20 13:53:42 | INFO | train rank=0 | step=16630/50000 progress=33.3% loss=0.136847 lr=7.970938e-05 lr_backbone=7.970938e-05 elapsed=07:15:41 avg_step=1.57s recent_step=2.22s eta=14:34:16 +2026-09-20 13:53:54 | INFO | train rank=0 | step=16640/50000 progress=33.3% loss=0.116030 lr=7.968277e-05 lr_backbone=7.968277e-05 elapsed=07:15:53 avg_step=1.57s recent_step=1.18s eta=14:33:53 +2026-09-20 13:54:05 | INFO | train rank=0 | step=16650/50000 progress=33.3% loss=0.092221 lr=7.965615e-05 lr_backbone=7.965615e-05 elapsed=07:16:05 avg_step=1.57s recent_step=1.17s eta=14:33:29 +2026-09-20 13:54:17 | INFO | train rank=0 | step=16660/50000 progress=33.3% loss=0.094293 lr=7.962952e-05 lr_backbone=7.962952e-05 elapsed=07:16:17 avg_step=1.57s recent_step=1.20s eta=14:33:06 +2026-09-20 13:54:34 | INFO | train rank=0 | step=16670/50000 progress=33.3% loss=0.122322 lr=7.960288e-05 lr_backbone=7.960288e-05 elapsed=07:16:34 avg_step=1.57s recent_step=1.67s eta=14:32:52 +2026-09-20 13:54:53 | INFO | train rank=0 | step=16680/50000 progress=33.4% loss=0.099282 lr=7.957622e-05 lr_backbone=7.957622e-05 elapsed=07:16:53 avg_step=1.57s recent_step=1.93s eta=14:32:44 +2026-09-20 13:55:08 | INFO | train rank=0 | step=16690/50000 progress=33.4% loss=0.134138 lr=7.954955e-05 lr_backbone=7.954955e-05 elapsed=07:17:07 avg_step=1.57s recent_step=1.44s eta=14:32:25 +2026-09-20 13:55:25 | INFO | train rank=0 | step=16700/50000 progress=33.4% loss=0.084500 lr=7.952287e-05 lr_backbone=7.952287e-05 elapsed=07:17:25 avg_step=1.57s recent_step=1.71s eta=14:32:12 +2026-09-20 13:55:40 | INFO | train rank=0 | step=16710/50000 progress=33.4% loss=0.134294 lr=7.949617e-05 lr_backbone=7.949617e-05 elapsed=07:17:39 avg_step=1.57s recent_step=1.48s eta=14:31:55 +2026-09-20 13:55:53 | INFO | train rank=0 | step=16720/50000 progress=33.4% loss=0.093526 lr=7.946947e-05 lr_backbone=7.946947e-05 elapsed=07:17:53 avg_step=1.57s recent_step=1.36s eta=14:31:35 +2026-09-20 13:56:14 | INFO | train rank=0 | step=16730/50000 progress=33.5% loss=0.086148 lr=7.944274e-05 lr_backbone=7.944274e-05 elapsed=07:18:14 avg_step=1.57s recent_step=2.09s eta=14:31:29 +2026-09-20 13:56:26 | INFO | train rank=0 | step=16740/50000 progress=33.5% loss=0.113296 lr=7.941601e-05 lr_backbone=7.941601e-05 elapsed=07:18:25 avg_step=1.57s recent_step=1.16s eta=14:31:05 +2026-09-20 13:56:42 | INFO | train rank=0 | step=16750/50000 progress=33.5% loss=0.101636 lr=7.938926e-05 lr_backbone=7.938926e-05 elapsed=07:18:42 avg_step=1.57s recent_step=1.66s eta=14:30:51 +2026-09-20 13:56:54 | INFO | train rank=0 | step=16760/50000 progress=33.5% loss=0.100091 lr=7.936250e-05 lr_backbone=7.936250e-05 elapsed=07:18:54 avg_step=1.57s recent_step=1.20s eta=14:30:28 +2026-09-20 13:57:11 | INFO | train rank=0 | step=16770/50000 progress=33.5% loss=0.094041 lr=7.933573e-05 lr_backbone=7.933573e-05 elapsed=07:19:10 avg_step=1.57s recent_step=1.62s eta=14:30:14 +2026-09-20 13:57:23 | INFO | train rank=0 | step=16780/50000 progress=33.6% loss=0.096054 lr=7.930894e-05 lr_backbone=7.930894e-05 elapsed=07:19:22 avg_step=1.57s recent_step=1.22s eta=14:29:51 +2026-09-20 13:57:40 | INFO | train rank=0 | step=16790/50000 progress=33.6% loss=0.079118 lr=7.928215e-05 lr_backbone=7.928215e-05 elapsed=07:19:40 avg_step=1.57s recent_step=1.72s eta=14:29:38 +2026-09-20 13:57:56 | INFO | train rank=0 | step=16800/50000 progress=33.6% loss=0.096923 lr=7.925533e-05 lr_backbone=7.925533e-05 elapsed=07:19:55 avg_step=1.57s recent_step=1.57s eta=14:29:23 +2026-09-20 13:58:13 | INFO | train rank=0 | step=16810/50000 progress=33.6% loss=0.102100 lr=7.922851e-05 lr_backbone=7.922851e-05 elapsed=07:20:13 avg_step=1.57s recent_step=1.76s eta=14:29:11 +2026-09-20 13:58:27 | INFO | train rank=0 | step=16820/50000 progress=33.6% loss=0.102050 lr=7.920167e-05 lr_backbone=7.920167e-05 elapsed=07:20:26 avg_step=1.57s recent_step=1.34s eta=14:28:50 +2026-09-20 13:58:43 | INFO | train rank=0 | step=16830/50000 progress=33.7% loss=0.089267 lr=7.917482e-05 lr_backbone=7.917482e-05 elapsed=07:20:42 avg_step=1.57s recent_step=1.59s eta=14:28:35 +2026-09-20 13:59:07 | INFO | train rank=0 | step=16840/50000 progress=33.7% loss=0.117316 lr=7.914796e-05 lr_backbone=7.914796e-05 elapsed=07:21:06 avg_step=1.57s recent_step=2.40s eta=14:28:36 +2026-09-20 13:59:18 | INFO | train rank=0 | step=16850/50000 progress=33.7% loss=0.120985 lr=7.912108e-05 lr_backbone=7.912108e-05 elapsed=07:21:18 avg_step=1.57s recent_step=1.15s eta=14:28:12 +2026-09-20 13:59:30 | INFO | train rank=0 | step=16860/50000 progress=33.7% loss=0.092708 lr=7.909420e-05 lr_backbone=7.909420e-05 elapsed=07:21:30 avg_step=1.57s recent_step=1.24s eta=14:27:49 +2026-09-20 13:59:47 | INFO | train rank=0 | step=16870/50000 progress=33.7% loss=0.087909 lr=7.906730e-05 lr_backbone=7.906730e-05 elapsed=07:21:46 avg_step=1.57s recent_step=1.63s eta=14:27:35 +2026-09-20 14:00:02 | INFO | train rank=0 | step=16880/50000 progress=33.8% loss=0.101090 lr=7.904038e-05 lr_backbone=7.904038e-05 elapsed=07:22:02 avg_step=1.57s recent_step=1.52s eta=14:27:18 +2026-09-20 14:00:17 | INFO | train rank=0 | step=16890/50000 progress=33.8% loss=0.119124 lr=7.901346e-05 lr_backbone=7.901346e-05 elapsed=07:22:17 avg_step=1.57s recent_step=1.52s eta=14:27:01 +2026-09-20 14:00:34 | INFO | train rank=0 | step=16900/50000 progress=33.8% loss=0.089636 lr=7.898652e-05 lr_backbone=7.898652e-05 elapsed=07:22:33 avg_step=1.57s recent_step=1.64s eta=14:26:47 +2026-09-20 14:00:51 | INFO | train rank=0 | step=16910/50000 progress=33.8% loss=0.132441 lr=7.895957e-05 lr_backbone=7.895957e-05 elapsed=07:22:51 avg_step=1.57s recent_step=1.78s eta=14:26:35 +2026-09-20 14:01:06 | INFO | train rank=0 | step=16920/50000 progress=33.8% loss=0.108714 lr=7.893260e-05 lr_backbone=7.893260e-05 elapsed=07:23:06 avg_step=1.57s recent_step=1.51s eta=14:26:18 +2026-09-20 14:01:22 | INFO | train rank=0 | step=16930/50000 progress=33.9% loss=0.105934 lr=7.890562e-05 lr_backbone=7.890562e-05 elapsed=07:23:22 avg_step=1.57s recent_step=1.58s eta=14:26:03 +2026-09-20 14:01:44 | INFO | train rank=0 | step=16940/50000 progress=33.9% loss=0.088239 lr=7.887864e-05 lr_backbone=7.887864e-05 elapsed=07:23:44 avg_step=1.57s recent_step=2.19s eta=14:25:59 +2026-09-20 14:01:56 | INFO | train rank=0 | step=16950/50000 progress=33.9% loss=0.128447 lr=7.885163e-05 lr_backbone=7.885163e-05 elapsed=07:23:55 avg_step=1.57s recent_step=1.18s eta=14:25:36 +2026-09-20 14:02:08 | INFO | train rank=0 | step=16960/50000 progress=33.9% loss=0.115433 lr=7.882462e-05 lr_backbone=7.882462e-05 elapsed=07:24:08 avg_step=1.57s recent_step=1.21s eta=14:25:13 +2026-09-20 14:02:22 | INFO | train rank=0 | step=16970/50000 progress=33.9% loss=0.107900 lr=7.879759e-05 lr_backbone=7.879759e-05 elapsed=07:24:21 avg_step=1.57s recent_step=1.38s eta=14:24:54 +2026-09-20 14:02:42 | INFO | train rank=0 | step=16980/50000 progress=34.0% loss=0.080754 lr=7.877055e-05 lr_backbone=7.877055e-05 elapsed=07:24:42 avg_step=1.57s recent_step=2.04s eta=14:24:47 +2026-09-20 14:02:58 | INFO | train rank=0 | step=16990/50000 progress=34.0% loss=0.109188 lr=7.874350e-05 lr_backbone=7.874350e-05 elapsed=07:24:58 avg_step=1.57s recent_step=1.57s eta=14:24:31 +2026-09-20 14:03:10 | INFO | train rank=0 | step=17000/50000 progress=34.0% loss=0.078889 lr=7.871643e-05 lr_backbone=7.871643e-05 elapsed=07:25:10 avg_step=1.57s recent_step=1.22s eta=14:24:09 +2026-09-20 14:03:26 | INFO | train rank=0 | step=17010/50000 progress=34.0% loss=0.074110 lr=7.868936e-05 lr_backbone=7.868936e-05 elapsed=07:25:26 avg_step=1.57s recent_step=1.64s eta=14:23:54 +2026-09-20 14:03:50 | INFO | train rank=0 | step=17020/50000 progress=34.0% loss=0.092191 lr=7.866227e-05 lr_backbone=7.866227e-05 elapsed=07:25:50 avg_step=1.57s recent_step=2.38s eta=14:23:54 +2026-09-20 14:04:02 | INFO | train rank=0 | step=17030/50000 progress=34.1% loss=0.071021 lr=7.863516e-05 lr_backbone=7.863516e-05 elapsed=07:26:02 avg_step=1.57s recent_step=1.17s eta=14:23:31 +2026-09-20 14:04:16 | INFO | train rank=0 | step=17040/50000 progress=34.1% loss=0.090671 lr=7.860805e-05 lr_backbone=7.860805e-05 elapsed=07:26:16 avg_step=1.57s recent_step=1.41s eta=14:23:12 +2026-09-20 14:04:31 | INFO | train rank=0 | step=17050/50000 progress=34.1% loss=0.079093 lr=7.858092e-05 lr_backbone=7.858092e-05 elapsed=07:26:30 avg_step=1.57s recent_step=1.47s eta=14:22:54 +2026-09-20 14:04:47 | INFO | train rank=0 | step=17060/50000 progress=34.1% loss=0.116694 lr=7.855378e-05 lr_backbone=7.855378e-05 elapsed=07:26:47 avg_step=1.57s recent_step=1.60s eta=14:22:39 +2026-09-20 14:05:05 | INFO | train rank=0 | step=17070/50000 progress=34.1% loss=0.101093 lr=7.852663e-05 lr_backbone=7.852663e-05 elapsed=07:27:04 avg_step=1.57s recent_step=1.79s eta=14:22:28 +2026-09-20 14:05:22 | INFO | train rank=0 | step=17080/50000 progress=34.2% loss=0.135353 lr=7.849946e-05 lr_backbone=7.849946e-05 elapsed=07:27:21 avg_step=1.57s recent_step=1.68s eta=14:22:14 +2026-09-20 14:05:34 | INFO | train rank=0 | step=17090/50000 progress=34.2% loss=0.086253 lr=7.847228e-05 lr_backbone=7.847228e-05 elapsed=07:27:34 avg_step=1.57s recent_step=1.25s eta=14:21:52 +2026-09-20 14:05:54 | INFO | train rank=0 | step=17100/50000 progress=34.2% loss=0.098184 lr=7.844509e-05 lr_backbone=7.844509e-05 elapsed=07:27:53 avg_step=1.57s recent_step=1.96s eta=14:21:44 +2026-09-20 14:06:06 | INFO | train rank=0 | step=17110/50000 progress=34.2% loss=0.094600 lr=7.841789e-05 lr_backbone=7.841789e-05 elapsed=07:28:06 avg_step=1.57s recent_step=1.25s eta=14:21:22 +2026-09-20 14:06:22 | INFO | train rank=0 | step=17120/50000 progress=34.2% loss=0.081370 lr=7.839067e-05 lr_backbone=7.839067e-05 elapsed=07:28:21 avg_step=1.57s recent_step=1.56s eta=14:21:06 +2026-09-20 14:06:36 | INFO | train rank=0 | step=17130/50000 progress=34.3% loss=0.091053 lr=7.836345e-05 lr_backbone=7.836345e-05 elapsed=07:28:36 avg_step=1.57s recent_step=1.43s eta=14:20:48 +2026-09-20 14:06:56 | INFO | train rank=0 | step=17140/50000 progress=34.3% loss=0.104740 lr=7.833621e-05 lr_backbone=7.833621e-05 elapsed=07:28:55 avg_step=1.57s recent_step=1.95s eta=14:20:39 +2026-09-20 14:07:07 | INFO | train rank=0 | step=17150/50000 progress=34.3% loss=0.113812 lr=7.830896e-05 lr_backbone=7.830896e-05 elapsed=07:29:07 avg_step=1.57s recent_step=1.19s eta=14:20:16 +2026-09-20 14:07:28 | INFO | train rank=0 | step=17160/50000 progress=34.3% loss=0.094698 lr=7.828169e-05 lr_backbone=7.828169e-05 elapsed=07:29:27 avg_step=1.57s recent_step=2.03s eta=14:20:09 +2026-09-20 14:07:40 | INFO | train rank=0 | step=17170/50000 progress=34.3% loss=0.096513 lr=7.825441e-05 lr_backbone=7.825441e-05 elapsed=07:29:39 avg_step=1.57s recent_step=1.21s eta=14:19:47 +2026-09-20 14:07:57 | INFO | train rank=0 | step=17180/50000 progress=34.4% loss=0.098815 lr=7.822712e-05 lr_backbone=7.822712e-05 elapsed=07:29:56 avg_step=1.57s recent_step=1.70s eta=14:19:33 +2026-09-20 14:08:11 | INFO | train rank=0 | step=17190/50000 progress=34.4% loss=0.106890 lr=7.819982e-05 lr_backbone=7.819982e-05 elapsed=07:30:10 avg_step=1.57s recent_step=1.40s eta=14:19:14 +2026-09-20 14:08:31 | INFO | train rank=0 | step=17200/50000 progress=34.4% loss=0.085382 lr=7.817251e-05 lr_backbone=7.817251e-05 elapsed=07:30:31 avg_step=1.57s recent_step=2.03s eta=14:19:07 +2026-09-20 14:08:43 | INFO | train rank=0 | step=17210/50000 progress=34.4% loss=0.126287 lr=7.814518e-05 lr_backbone=7.814518e-05 elapsed=07:30:43 avg_step=1.57s recent_step=1.19s eta=14:18:44 +2026-09-20 14:08:55 | INFO | train rank=0 | step=17220/50000 progress=34.4% loss=0.108177 lr=7.811784e-05 lr_backbone=7.811784e-05 elapsed=07:30:55 avg_step=1.57s recent_step=1.20s eta=14:18:22 +2026-09-20 14:09:13 | INFO | train rank=0 | step=17230/50000 progress=34.5% loss=0.107058 lr=7.809049e-05 lr_backbone=7.809049e-05 elapsed=07:31:13 avg_step=1.57s recent_step=1.82s eta=14:18:11 +2026-09-20 14:09:28 | INFO | train rank=0 | step=17240/50000 progress=34.5% loss=0.127449 lr=7.806313e-05 lr_backbone=7.806313e-05 elapsed=07:31:27 avg_step=1.57s recent_step=1.45s eta=14:17:53 +2026-09-20 14:09:45 | INFO | train rank=0 | step=17250/50000 progress=34.5% loss=0.076760 lr=7.803575e-05 lr_backbone=7.803575e-05 elapsed=07:31:44 avg_step=1.57s recent_step=1.70s eta=14:17:39 +2026-09-20 14:10:01 | INFO | train rank=0 | step=17260/50000 progress=34.5% loss=0.112156 lr=7.800836e-05 lr_backbone=7.800836e-05 elapsed=07:32:01 avg_step=1.57s recent_step=1.65s eta=14:17:25 +2026-09-20 14:10:15 | INFO | train rank=0 | step=17270/50000 progress=34.5% loss=0.108989 lr=7.798096e-05 lr_backbone=7.798096e-05 elapsed=07:32:14 avg_step=1.57s recent_step=1.36s eta=14:17:06 +2026-09-20 14:10:33 | INFO | train rank=0 | step=17280/50000 progress=34.6% loss=0.143969 lr=7.795355e-05 lr_backbone=7.795355e-05 elapsed=07:32:32 avg_step=1.57s recent_step=1.80s eta=14:16:54 +2026-09-20 14:10:46 | INFO | train rank=0 | step=17290/50000 progress=34.6% loss=0.090829 lr=7.792613e-05 lr_backbone=7.792613e-05 elapsed=07:32:46 avg_step=1.57s recent_step=1.30s eta=14:16:33 +2026-09-20 14:11:04 | INFO | train rank=0 | step=17300/50000 progress=34.6% loss=0.096895 lr=7.789869e-05 lr_backbone=7.789869e-05 elapsed=07:33:04 avg_step=1.57s recent_step=1.83s eta=14:16:23 +2026-09-20 14:11:16 | INFO | train rank=0 | step=17310/50000 progress=34.6% loss=0.105356 lr=7.787124e-05 lr_backbone=7.787124e-05 elapsed=07:33:16 avg_step=1.57s recent_step=1.23s eta=14:16:00 +2026-09-20 14:11:32 | INFO | train rank=0 | step=17320/50000 progress=34.6% loss=0.103650 lr=7.784378e-05 lr_backbone=7.784378e-05 elapsed=07:33:31 avg_step=1.57s recent_step=1.55s eta=14:15:44 +2026-09-20 14:11:50 | INFO | train rank=0 | step=17330/50000 progress=34.7% loss=0.083993 lr=7.781631e-05 lr_backbone=7.781631e-05 elapsed=07:33:50 avg_step=1.57s recent_step=1.81s eta=14:15:33 +2026-09-20 14:12:04 | INFO | train rank=0 | step=17340/50000 progress=34.7% loss=0.098597 lr=7.778882e-05 lr_backbone=7.778882e-05 elapsed=07:34:03 avg_step=1.57s recent_step=1.37s eta=14:15:13 +2026-09-20 14:12:24 | INFO | train rank=0 | step=17350/50000 progress=34.7% loss=0.087794 lr=7.776132e-05 lr_backbone=7.776132e-05 elapsed=07:34:24 avg_step=1.57s recent_step=2.03s eta=14:15:06 +2026-09-20 14:12:37 | INFO | train rank=0 | step=17360/50000 progress=34.7% loss=0.107079 lr=7.773381e-05 lr_backbone=7.773381e-05 elapsed=07:34:37 avg_step=1.57s recent_step=1.34s eta=14:14:46 +2026-09-20 14:12:51 | INFO | train rank=0 | step=17370/50000 progress=34.7% loss=0.101256 lr=7.770629e-05 lr_backbone=7.770629e-05 elapsed=07:34:51 avg_step=1.57s recent_step=1.39s eta=14:14:27 +2026-09-20 14:13:10 | INFO | train rank=0 | step=17380/50000 progress=34.8% loss=0.107261 lr=7.767876e-05 lr_backbone=7.767876e-05 elapsed=07:35:09 avg_step=1.57s recent_step=1.86s eta=14:14:17 +2026-09-20 14:13:27 | INFO | train rank=0 | step=17390/50000 progress=34.8% loss=0.147126 lr=7.765121e-05 lr_backbone=7.765121e-05 elapsed=07:35:27 avg_step=1.57s recent_step=1.76s eta=14:14:04 +2026-09-20 14:13:39 | INFO | train rank=0 | step=17400/50000 progress=34.8% loss=0.086793 lr=7.762365e-05 lr_backbone=7.762365e-05 elapsed=07:35:39 avg_step=1.57s recent_step=1.17s eta=14:13:41 +2026-09-20 14:13:52 | INFO | train rank=0 | step=17410/50000 progress=34.8% loss=0.103041 lr=7.759608e-05 lr_backbone=7.759608e-05 elapsed=07:35:52 avg_step=1.57s recent_step=1.30s eta=14:13:20 +2026-09-20 14:14:18 | INFO | train rank=0 | step=17420/50000 progress=34.8% loss=0.111792 lr=7.756850e-05 lr_backbone=7.756850e-05 elapsed=07:36:18 avg_step=1.57s recent_step=2.62s eta=14:13:24 +2026-09-20 14:14:30 | INFO | train rank=0 | step=17430/50000 progress=34.9% loss=0.081275 lr=7.754091e-05 lr_backbone=7.754091e-05 elapsed=07:36:29 avg_step=1.57s recent_step=1.16s eta=14:13:01 +2026-09-20 14:14:41 | INFO | train rank=0 | step=17440/50000 progress=34.9% loss=0.113678 lr=7.751330e-05 lr_backbone=7.751330e-05 elapsed=07:36:41 avg_step=1.57s recent_step=1.16s eta=14:12:37 +2026-09-20 14:14:54 | INFO | train rank=0 | step=17450/50000 progress=34.9% loss=0.089613 lr=7.748568e-05 lr_backbone=7.748568e-05 elapsed=07:36:53 avg_step=1.57s recent_step=1.22s eta=14:12:15 +2026-09-20 14:15:14 | INFO | train rank=0 | step=17460/50000 progress=34.9% loss=0.098788 lr=7.745805e-05 lr_backbone=7.745805e-05 elapsed=07:37:13 avg_step=1.57s recent_step=1.99s eta=14:12:07 +2026-09-20 14:15:26 | INFO | train rank=0 | step=17470/50000 progress=34.9% loss=0.112384 lr=7.743041e-05 lr_backbone=7.743041e-05 elapsed=07:37:25 avg_step=1.57s recent_step=1.20s eta=14:11:45 +2026-09-20 14:15:45 | INFO | train rank=0 | step=17480/50000 progress=35.0% loss=0.078548 lr=7.740275e-05 lr_backbone=7.740275e-05 elapsed=07:37:45 avg_step=1.57s recent_step=1.95s eta=14:11:36 +2026-09-20 14:15:58 | INFO | train rank=0 | step=17490/50000 progress=35.0% loss=0.085982 lr=7.737509e-05 lr_backbone=7.737509e-05 elapsed=07:37:58 avg_step=1.57s recent_step=1.34s eta=14:11:16 +2026-09-20 14:16:17 | INFO | train rank=0 | step=17500/50000 progress=35.0% loss=0.114198 lr=7.734741e-05 lr_backbone=7.734741e-05 elapsed=07:38:17 avg_step=1.57s recent_step=1.88s eta=14:11:06 +2026-09-20 14:16:31 | INFO | train rank=0 | step=17510/50000 progress=35.0% loss=0.073021 lr=7.731972e-05 lr_backbone=7.731972e-05 elapsed=07:38:30 avg_step=1.57s recent_step=1.35s eta=14:10:46 +2026-09-20 14:16:46 | INFO | train rank=0 | step=17520/50000 progress=35.0% loss=0.100345 lr=7.729201e-05 lr_backbone=7.729201e-05 elapsed=07:38:45 avg_step=1.57s recent_step=1.52s eta=14:10:29 +2026-09-20 14:17:06 | INFO | train rank=0 | step=17530/50000 progress=35.1% loss=0.076972 lr=7.726430e-05 lr_backbone=7.726430e-05 elapsed=07:39:05 avg_step=1.57s recent_step=1.98s eta=14:10:21 +2026-09-20 14:17:23 | INFO | train rank=0 | step=17540/50000 progress=35.1% loss=0.099117 lr=7.723657e-05 lr_backbone=7.723657e-05 elapsed=07:39:22 avg_step=1.57s recent_step=1.69s eta=14:10:08 +2026-09-20 14:17:39 | INFO | train rank=0 | step=17550/50000 progress=35.1% loss=0.080280 lr=7.720884e-05 lr_backbone=7.720884e-05 elapsed=07:39:38 avg_step=1.57s recent_step=1.63s eta=14:09:53 +2026-09-20 14:17:52 | INFO | train rank=0 | step=17560/50000 progress=35.1% loss=0.119101 lr=7.718109e-05 lr_backbone=7.718109e-05 elapsed=07:39:51 avg_step=1.57s recent_step=1.28s eta=14:09:32 +2026-09-20 14:18:08 | INFO | train rank=0 | step=17570/50000 progress=35.1% loss=0.085666 lr=7.715332e-05 lr_backbone=7.715332e-05 elapsed=07:40:07 avg_step=1.57s recent_step=1.62s eta=14:09:17 +2026-09-20 14:18:21 | INFO | train rank=0 | step=17580/50000 progress=35.2% loss=0.181771 lr=7.712555e-05 lr_backbone=7.712555e-05 elapsed=07:40:20 avg_step=1.57s recent_step=1.30s eta=14:08:56 +2026-09-20 14:18:37 | INFO | train rank=0 | step=17590/50000 progress=35.2% loss=0.108046 lr=7.709776e-05 lr_backbone=7.709776e-05 elapsed=07:40:36 avg_step=1.57s recent_step=1.59s eta=14:08:41 +2026-09-20 14:18:53 | INFO | train rank=0 | step=17600/50000 progress=35.2% loss=0.092310 lr=7.706997e-05 lr_backbone=7.706997e-05 elapsed=07:40:53 avg_step=1.57s recent_step=1.63s eta=14:08:26 +2026-09-20 14:19:10 | INFO | train rank=0 | step=17610/50000 progress=35.2% loss=0.102546 lr=7.704216e-05 lr_backbone=7.704216e-05 elapsed=07:41:10 avg_step=1.57s recent_step=1.71s eta=14:08:13 +2026-09-20 14:19:22 | INFO | train rank=0 | step=17620/50000 progress=35.2% loss=0.088861 lr=7.701434e-05 lr_backbone=7.701434e-05 elapsed=07:41:22 avg_step=1.57s recent_step=1.21s eta=14:07:51 +2026-09-20 14:19:34 | INFO | train rank=0 | step=17630/50000 progress=35.3% loss=0.113656 lr=7.698650e-05 lr_backbone=7.698650e-05 elapsed=07:41:34 avg_step=1.57s recent_step=1.20s eta=14:07:28 +2026-09-20 14:19:52 | INFO | train rank=0 | step=17640/50000 progress=35.3% loss=0.133902 lr=7.695866e-05 lr_backbone=7.695866e-05 elapsed=07:41:51 avg_step=1.57s recent_step=1.74s eta=14:07:16 +2026-09-20 14:20:08 | INFO | train rank=0 | step=17650/50000 progress=35.3% loss=0.119147 lr=7.693080e-05 lr_backbone=7.693080e-05 elapsed=07:42:08 avg_step=1.57s recent_step=1.65s eta=14:07:02 +2026-09-20 14:20:26 | INFO | train rank=0 | step=17660/50000 progress=35.3% loss=0.104011 lr=7.690293e-05 lr_backbone=7.690293e-05 elapsed=07:42:26 avg_step=1.57s recent_step=1.79s eta=14:06:50 +2026-09-20 14:20:44 | INFO | train rank=0 | step=17670/50000 progress=35.3% loss=0.096935 lr=7.687505e-05 lr_backbone=7.687505e-05 elapsed=07:42:43 avg_step=1.57s recent_step=1.74s eta=14:06:37 +2026-09-20 14:20:55 | INFO | train rank=0 | step=17680/50000 progress=35.4% loss=0.087166 lr=7.684716e-05 lr_backbone=7.684716e-05 elapsed=07:42:55 avg_step=1.57s recent_step=1.15s eta=14:06:14 +2026-09-20 14:21:15 | INFO | train rank=0 | step=17690/50000 progress=35.4% loss=0.092922 lr=7.681926e-05 lr_backbone=7.681926e-05 elapsed=07:43:14 avg_step=1.57s recent_step=1.96s eta=14:06:05 +2026-09-20 14:21:30 | INFO | train rank=0 | step=17700/50000 progress=35.4% loss=0.106887 lr=7.679134e-05 lr_backbone=7.679134e-05 elapsed=07:43:30 avg_step=1.57s recent_step=1.58s eta=14:05:50 +2026-09-20 14:21:44 | INFO | train rank=0 | step=17710/50000 progress=35.4% loss=0.118702 lr=7.676341e-05 lr_backbone=7.676341e-05 elapsed=07:43:43 avg_step=1.57s recent_step=1.33s eta=14:05:30 +2026-09-20 14:22:01 | INFO | train rank=0 | step=17720/50000 progress=35.4% loss=0.084761 lr=7.673547e-05 lr_backbone=7.673547e-05 elapsed=07:44:01 avg_step=1.57s recent_step=1.76s eta=14:05:17 +2026-09-20 14:22:19 | INFO | train rank=0 | step=17730/50000 progress=35.5% loss=0.108512 lr=7.670752e-05 lr_backbone=7.670752e-05 elapsed=07:44:19 avg_step=1.57s recent_step=1.77s eta=14:05:05 +2026-09-20 14:22:33 | INFO | train rank=0 | step=17740/50000 progress=35.5% loss=0.101379 lr=7.667956e-05 lr_backbone=7.667956e-05 elapsed=07:44:33 avg_step=1.57s recent_step=1.43s eta=14:04:47 +2026-09-20 14:22:52 | INFO | train rank=0 | step=17750/50000 progress=35.5% loss=0.100880 lr=7.665159e-05 lr_backbone=7.665159e-05 elapsed=07:44:51 avg_step=1.57s recent_step=1.85s eta=14:04:36 +2026-09-20 14:23:06 | INFO | train rank=0 | step=17760/50000 progress=35.5% loss=0.114867 lr=7.662360e-05 lr_backbone=7.662360e-05 elapsed=07:45:06 avg_step=1.57s recent_step=1.45s eta=14:04:18 +2026-09-20 14:23:19 | INFO | train rank=0 | step=17770/50000 progress=35.5% loss=0.078022 lr=7.659560e-05 lr_backbone=7.659560e-05 elapsed=07:45:19 avg_step=1.57s recent_step=1.27s eta=14:03:57 +2026-09-20 14:23:34 | INFO | train rank=0 | step=17780/50000 progress=35.6% loss=0.088155 lr=7.656759e-05 lr_backbone=7.656759e-05 elapsed=07:45:34 avg_step=1.57s recent_step=1.50s eta=14:03:40 +2026-09-20 14:23:49 | INFO | train rank=0 | step=17790/50000 progress=35.6% loss=0.089137 lr=7.653957e-05 lr_backbone=7.653957e-05 elapsed=07:45:49 avg_step=1.57s recent_step=1.54s eta=14:03:24 +2026-09-20 14:24:12 | INFO | train rank=0 | step=17800/50000 progress=35.6% loss=0.135805 lr=7.651154e-05 lr_backbone=7.651154e-05 elapsed=07:46:11 avg_step=1.57s recent_step=2.21s eta=14:03:20 +2026-09-20 14:24:23 | INFO | train rank=0 | step=17810/50000 progress=35.6% loss=0.090604 lr=7.648350e-05 lr_backbone=7.648350e-05 elapsed=07:46:22 avg_step=1.57s recent_step=1.13s eta=14:02:56 +2026-09-20 14:24:36 | INFO | train rank=0 | step=17820/50000 progress=35.6% loss=0.104740 lr=7.645544e-05 lr_backbone=7.645544e-05 elapsed=07:46:36 avg_step=1.57s recent_step=1.33s eta=14:02:36 +2026-09-20 14:24:49 | INFO | train rank=0 | step=17830/50000 progress=35.7% loss=0.124097 lr=7.642738e-05 lr_backbone=7.642738e-05 elapsed=07:46:49 avg_step=1.57s recent_step=1.31s eta=14:02:16 +2026-09-20 14:25:04 | INFO | train rank=0 | step=17840/50000 progress=35.7% loss=0.098195 lr=7.639930e-05 lr_backbone=7.639930e-05 elapsed=07:47:03 avg_step=1.57s recent_step=1.45s eta=14:01:58 +2026-09-20 14:25:19 | INFO | train rank=0 | step=17850/50000 progress=35.7% loss=0.112714 lr=7.637121e-05 lr_backbone=7.637121e-05 elapsed=07:47:18 avg_step=1.57s recent_step=1.51s eta=14:01:41 +2026-09-20 14:25:38 | INFO | train rank=0 | step=17860/50000 progress=35.7% loss=0.076572 lr=7.634311e-05 lr_backbone=7.634311e-05 elapsed=07:47:38 avg_step=1.57s recent_step=1.95s eta=14:01:32 +2026-09-20 14:25:53 | INFO | train rank=0 | step=17870/50000 progress=35.7% loss=0.099917 lr=7.631499e-05 lr_backbone=7.631499e-05 elapsed=07:47:52 avg_step=1.57s recent_step=1.46s eta=14:01:14 +2026-09-20 14:26:11 | INFO | train rank=0 | step=17880/50000 progress=35.8% loss=0.080691 lr=7.628687e-05 lr_backbone=7.628687e-05 elapsed=07:48:11 avg_step=1.57s recent_step=1.82s eta=14:01:03 +2026-09-20 14:26:26 | INFO | train rank=0 | step=17890/50000 progress=35.8% loss=0.079495 lr=7.625873e-05 lr_backbone=7.625873e-05 elapsed=07:48:25 avg_step=1.57s recent_step=1.48s eta=14:00:46 +2026-09-20 14:26:41 | INFO | train rank=0 | step=17900/50000 progress=35.8% loss=0.093863 lr=7.623058e-05 lr_backbone=7.623058e-05 elapsed=07:48:41 avg_step=1.57s recent_step=1.55s eta=14:00:30 +2026-09-20 14:26:57 | INFO | train rank=0 | step=17910/50000 progress=35.8% loss=0.072772 lr=7.620242e-05 lr_backbone=7.620242e-05 elapsed=07:48:57 avg_step=1.57s recent_step=1.61s eta=14:00:15 +2026-09-20 14:27:14 | INFO | train rank=0 | step=17920/50000 progress=35.8% loss=0.108099 lr=7.617425e-05 lr_backbone=7.617425e-05 elapsed=07:49:14 avg_step=1.57s recent_step=1.69s eta=14:00:01 +2026-09-20 14:27:28 | INFO | train rank=0 | step=17930/50000 progress=35.9% loss=0.110128 lr=7.614607e-05 lr_backbone=7.614607e-05 elapsed=07:49:28 avg_step=1.57s recent_step=1.42s eta=13:59:42 +2026-09-20 14:27:47 | INFO | train rank=0 | step=17940/50000 progress=35.9% loss=0.081858 lr=7.611788e-05 lr_backbone=7.611788e-05 elapsed=07:49:47 avg_step=1.57s recent_step=1.87s eta=13:59:32 +2026-09-20 14:28:00 | INFO | train rank=0 | step=17950/50000 progress=35.9% loss=0.102135 lr=7.608967e-05 lr_backbone=7.608967e-05 elapsed=07:49:59 avg_step=1.57s recent_step=1.27s eta=13:59:11 +2026-09-20 14:28:21 | INFO | train rank=0 | step=17960/50000 progress=35.9% loss=0.129198 lr=7.606146e-05 lr_backbone=7.606146e-05 elapsed=07:50:21 avg_step=1.57s recent_step=2.14s eta=13:59:05 +2026-09-20 14:28:33 | INFO | train rank=0 | step=17970/50000 progress=35.9% loss=0.089584 lr=7.603323e-05 lr_backbone=7.603323e-05 elapsed=07:50:33 avg_step=1.57s recent_step=1.17s eta=13:58:43 +2026-09-20 14:28:52 | INFO | train rank=0 | step=17980/50000 progress=36.0% loss=0.096184 lr=7.600499e-05 lr_backbone=7.600499e-05 elapsed=07:50:52 avg_step=1.57s recent_step=1.93s eta=13:58:33 +2026-09-20 14:29:06 | INFO | train rank=0 | step=17990/50000 progress=36.0% loss=0.093792 lr=7.597674e-05 lr_backbone=7.597674e-05 elapsed=07:51:06 avg_step=1.57s recent_step=1.41s eta=13:58:15 +2026-09-20 14:29:19 | INFO | train rank=0 | step=18000/50000 progress=36.0% loss=0.078037 lr=7.594848e-05 lr_backbone=7.594848e-05 elapsed=07:51:18 avg_step=1.57s recent_step=1.24s eta=13:57:53 +2026-09-20 14:29:33 | INFO | train rank=0 | step=18010/50000 progress=36.0% loss=0.111096 lr=7.592021e-05 lr_backbone=7.592021e-05 elapsed=07:51:33 avg_step=1.57s recent_step=1.45s eta=13:57:35 +2026-09-20 14:29:52 | INFO | train rank=0 | step=18020/50000 progress=36.0% loss=0.111251 lr=7.589192e-05 lr_backbone=7.589192e-05 elapsed=07:51:51 avg_step=1.57s recent_step=1.83s eta=13:57:24 +2026-09-20 14:30:10 | INFO | train rank=0 | step=18030/50000 progress=36.1% loss=0.106760 lr=7.586363e-05 lr_backbone=7.586363e-05 elapsed=07:52:10 avg_step=1.57s recent_step=1.87s eta=13:57:14 +2026-09-20 14:30:22 | INFO | train rank=0 | step=18040/50000 progress=36.1% loss=0.101389 lr=7.583532e-05 lr_backbone=7.583532e-05 elapsed=07:52:21 avg_step=1.57s recent_step=1.15s eta=13:56:51 +2026-09-20 14:30:36 | INFO | train rank=0 | step=18050/50000 progress=36.1% loss=0.093354 lr=7.580700e-05 lr_backbone=7.580700e-05 elapsed=07:52:36 avg_step=1.57s recent_step=1.41s eta=13:56:32 +2026-09-20 14:30:55 | INFO | train rank=0 | step=18060/50000 progress=36.1% loss=0.093010 lr=7.577867e-05 lr_backbone=7.577867e-05 elapsed=07:52:54 avg_step=1.57s recent_step=1.89s eta=13:56:22 +2026-09-20 14:31:13 | INFO | train rank=0 | step=18070/50000 progress=36.1% loss=0.093376 lr=7.575033e-05 lr_backbone=7.575033e-05 elapsed=07:53:13 avg_step=1.57s recent_step=1.85s eta=13:56:11 +2026-09-20 14:31:28 | INFO | train rank=0 | step=18080/50000 progress=36.2% loss=0.097181 lr=7.572198e-05 lr_backbone=7.572198e-05 elapsed=07:53:28 avg_step=1.57s recent_step=1.50s eta=13:55:54 +2026-09-20 14:31:42 | INFO | train rank=0 | step=18090/50000 progress=36.2% loss=0.106993 lr=7.569361e-05 lr_backbone=7.569361e-05 elapsed=07:53:41 avg_step=1.57s recent_step=1.31s eta=13:55:34 +2026-09-20 14:32:00 | INFO | train rank=0 | step=18100/50000 progress=36.2% loss=0.096763 lr=7.566524e-05 lr_backbone=7.566524e-05 elapsed=07:53:59 avg_step=1.57s recent_step=1.83s eta=13:55:23 +2026-09-20 14:32:17 | INFO | train rank=0 | step=18110/50000 progress=36.2% loss=0.089031 lr=7.563685e-05 lr_backbone=7.563685e-05 elapsed=07:54:17 avg_step=1.57s recent_step=1.77s eta=13:55:11 +2026-09-20 14:32:29 | INFO | train rank=0 | step=18120/50000 progress=36.2% loss=0.097292 lr=7.560846e-05 lr_backbone=7.560846e-05 elapsed=07:54:29 avg_step=1.57s recent_step=1.17s eta=13:54:48 +2026-09-20 14:32:45 | INFO | train rank=0 | step=18130/50000 progress=36.3% loss=0.108116 lr=7.558005e-05 lr_backbone=7.558005e-05 elapsed=07:54:44 avg_step=1.57s recent_step=1.58s eta=13:54:32 +2026-09-20 14:33:02 | INFO | train rank=0 | step=18140/50000 progress=36.3% loss=0.097072 lr=7.555163e-05 lr_backbone=7.555163e-05 elapsed=07:55:01 avg_step=1.57s recent_step=1.68s eta=13:54:18 +2026-09-20 14:33:14 | INFO | train rank=0 | step=18150/50000 progress=36.3% loss=0.081551 lr=7.552320e-05 lr_backbone=7.552320e-05 elapsed=07:55:13 avg_step=1.57s recent_step=1.19s eta=13:53:56 +2026-09-20 14:33:32 | INFO | train rank=0 | step=18160/50000 progress=36.3% loss=0.102027 lr=7.549475e-05 lr_backbone=7.549475e-05 elapsed=07:55:31 avg_step=1.57s recent_step=1.80s eta=13:53:44 +2026-09-20 14:33:47 | INFO | train rank=0 | step=18170/50000 progress=36.3% loss=0.106285 lr=7.546630e-05 lr_backbone=7.546630e-05 elapsed=07:55:46 avg_step=1.57s recent_step=1.50s eta=13:53:27 +2026-09-20 14:34:05 | INFO | train rank=0 | step=18180/50000 progress=36.4% loss=0.091322 lr=7.543784e-05 lr_backbone=7.543784e-05 elapsed=07:56:05 avg_step=1.57s recent_step=1.84s eta=13:53:16 +2026-09-20 14:34:17 | INFO | train rank=0 | step=18190/50000 progress=36.4% loss=0.107120 lr=7.540936e-05 lr_backbone=7.540936e-05 elapsed=07:56:17 avg_step=1.57s recent_step=1.22s eta=13:52:55 +2026-09-20 14:34:29 | INFO | train rank=0 | step=18200/50000 progress=36.4% loss=0.106033 lr=7.538088e-05 lr_backbone=7.538088e-05 elapsed=07:56:29 avg_step=1.57s recent_step=1.17s eta=13:52:32 +2026-09-20 14:34:42 | INFO | train rank=0 | step=18210/50000 progress=36.4% loss=0.106438 lr=7.535238e-05 lr_backbone=7.535238e-05 elapsed=07:56:42 avg_step=1.57s recent_step=1.30s eta=13:52:11 +2026-09-20 14:35:01 | INFO | train rank=0 | step=18220/50000 progress=36.4% loss=0.105965 lr=7.532387e-05 lr_backbone=7.532387e-05 elapsed=07:57:01 avg_step=1.57s recent_step=1.91s eta=13:52:02 +2026-09-20 14:35:20 | INFO | train rank=0 | step=18230/50000 progress=36.5% loss=0.103032 lr=7.529535e-05 lr_backbone=7.529535e-05 elapsed=07:57:19 avg_step=1.57s recent_step=1.85s eta=13:51:51 +2026-09-20 14:35:33 | INFO | train rank=0 | step=18240/50000 progress=36.5% loss=0.159723 lr=7.526682e-05 lr_backbone=7.526682e-05 elapsed=07:57:32 avg_step=1.57s recent_step=1.29s eta=13:51:30 +2026-09-20 14:35:49 | INFO | train rank=0 | step=18250/50000 progress=36.5% loss=0.106048 lr=7.523828e-05 lr_backbone=7.523828e-05 elapsed=07:57:48 avg_step=1.57s recent_step=1.61s eta=13:51:15 +2026-09-20 14:36:08 | INFO | train rank=0 | step=18260/50000 progress=36.5% loss=0.091838 lr=7.520972e-05 lr_backbone=7.520972e-05 elapsed=07:58:08 avg_step=1.57s recent_step=1.99s eta=13:51:07 +2026-09-20 14:36:21 | INFO | train rank=0 | step=18270/50000 progress=36.5% loss=0.086936 lr=7.518116e-05 lr_backbone=7.518116e-05 elapsed=07:58:21 avg_step=1.57s recent_step=1.27s eta=13:50:46 +2026-09-20 14:36:40 | INFO | train rank=0 | step=18280/50000 progress=36.6% loss=0.096189 lr=7.515259e-05 lr_backbone=7.515259e-05 elapsed=07:58:40 avg_step=1.57s recent_step=1.90s eta=13:50:36 +2026-09-20 14:36:51 | INFO | train rank=0 | step=18290/50000 progress=36.6% loss=0.074975 lr=7.512400e-05 lr_backbone=7.512400e-05 elapsed=07:58:51 avg_step=1.57s recent_step=1.13s eta=13:50:12 +2026-09-20 14:37:11 | INFO | train rank=0 | step=18300/50000 progress=36.6% loss=0.125201 lr=7.509540e-05 lr_backbone=7.509540e-05 elapsed=07:59:10 avg_step=1.57s recent_step=1.94s eta=13:50:03 +2026-09-20 14:37:23 | INFO | train rank=0 | step=18310/50000 progress=36.6% loss=0.097239 lr=7.506679e-05 lr_backbone=7.506679e-05 elapsed=07:59:22 avg_step=1.57s recent_step=1.20s eta=13:49:41 +2026-09-20 14:37:40 | INFO | train rank=0 | step=18320/50000 progress=36.6% loss=0.087665 lr=7.503818e-05 lr_backbone=7.503818e-05 elapsed=07:59:39 avg_step=1.57s recent_step=1.67s eta=13:49:27 +2026-09-20 14:37:57 | INFO | train rank=0 | step=18330/50000 progress=36.7% loss=0.119158 lr=7.500955e-05 lr_backbone=7.500955e-05 elapsed=07:59:57 avg_step=1.57s recent_step=1.76s eta=13:49:15 +2026-09-20 14:38:16 | INFO | train rank=0 | step=18340/50000 progress=36.7% loss=0.086206 lr=7.498090e-05 lr_backbone=7.498090e-05 elapsed=08:00:16 avg_step=1.57s recent_step=1.88s eta=13:49:04 +2026-09-20 14:38:28 | INFO | train rank=0 | step=18350/50000 progress=36.7% loss=0.115227 lr=7.495225e-05 lr_backbone=7.495225e-05 elapsed=08:00:27 avg_step=1.57s recent_step=1.16s eta=13:48:41 +2026-09-20 14:38:40 | INFO | train rank=0 | step=18360/50000 progress=36.7% loss=0.122017 lr=7.492359e-05 lr_backbone=7.492359e-05 elapsed=08:00:40 avg_step=1.57s recent_step=1.25s eta=13:48:20 +2026-09-20 14:38:59 | INFO | train rank=0 | step=18370/50000 progress=36.7% loss=0.092433 lr=7.489492e-05 lr_backbone=7.489492e-05 elapsed=08:00:59 avg_step=1.57s recent_step=1.93s eta=13:48:11 +2026-09-20 14:39:16 | INFO | train rank=0 | step=18380/50000 progress=36.8% loss=0.119044 lr=7.486623e-05 lr_backbone=7.486623e-05 elapsed=08:01:16 avg_step=1.57s recent_step=1.66s eta=13:47:56 +2026-09-20 14:39:32 | INFO | train rank=0 | step=18390/50000 progress=36.8% loss=0.156766 lr=7.483754e-05 lr_backbone=7.483754e-05 elapsed=08:01:32 avg_step=1.57s recent_step=1.60s eta=13:47:41 +2026-09-20 14:39:51 | INFO | train rank=0 | step=18400/50000 progress=36.8% loss=0.109226 lr=7.480883e-05 lr_backbone=7.480883e-05 elapsed=08:01:51 avg_step=1.57s recent_step=1.92s eta=13:47:32 +2026-09-20 14:40:04 | INFO | train rank=0 | step=18410/50000 progress=36.8% loss=0.107885 lr=7.478011e-05 lr_backbone=7.478011e-05 elapsed=08:02:03 avg_step=1.57s recent_step=1.23s eta=13:47:10 +2026-09-20 14:40:21 | INFO | train rank=0 | step=18420/50000 progress=36.8% loss=0.107661 lr=7.475139e-05 lr_backbone=7.475139e-05 elapsed=08:02:21 avg_step=1.57s recent_step=1.76s eta=13:46:58 +2026-09-20 14:40:40 | INFO | train rank=0 | step=18430/50000 progress=36.9% loss=0.086669 lr=7.472265e-05 lr_backbone=7.472265e-05 elapsed=08:02:39 avg_step=1.57s recent_step=1.86s eta=13:46:47 +2026-09-20 14:40:52 | INFO | train rank=0 | step=18440/50000 progress=36.9% loss=0.102764 lr=7.469390e-05 lr_backbone=7.469390e-05 elapsed=08:02:52 avg_step=1.57s recent_step=1.25s eta=13:46:26 +2026-09-20 14:41:04 | INFO | train rank=0 | step=18450/50000 progress=36.9% loss=0.106056 lr=7.466514e-05 lr_backbone=7.466514e-05 elapsed=08:03:03 avg_step=1.57s recent_step=1.15s eta=13:46:03 +2026-09-20 14:41:26 | INFO | train rank=0 | step=18460/50000 progress=36.9% loss=0.109381 lr=7.463637e-05 lr_backbone=7.463637e-05 elapsed=08:03:25 avg_step=1.57s recent_step=2.20s eta=13:45:58 +2026-09-20 14:41:40 | INFO | train rank=0 | step=18470/50000 progress=36.9% loss=0.094526 lr=7.460759e-05 lr_backbone=7.460759e-05 elapsed=08:03:40 avg_step=1.57s recent_step=1.45s eta=13:45:40 +2026-09-20 14:41:55 | INFO | train rank=0 | step=18480/50000 progress=37.0% loss=0.097555 lr=7.457879e-05 lr_backbone=7.457879e-05 elapsed=08:03:55 avg_step=1.57s recent_step=1.46s eta=13:45:22 +2026-09-20 14:42:10 | INFO | train rank=0 | step=18490/50000 progress=37.0% loss=0.075385 lr=7.454999e-05 lr_backbone=7.454999e-05 elapsed=08:04:10 avg_step=1.57s recent_step=1.54s eta=13:45:06 +2026-09-20 14:42:31 | INFO | train rank=0 | step=18500/50000 progress=37.0% loss=0.097075 lr=7.452118e-05 lr_backbone=7.452118e-05 elapsed=08:04:31 avg_step=1.57s recent_step=2.11s eta=13:45:00 +2026-09-20 14:42:43 | INFO | train rank=0 | step=18510/50000 progress=37.0% loss=0.103085 lr=7.449235e-05 lr_backbone=7.449235e-05 elapsed=08:04:42 avg_step=1.57s recent_step=1.14s eta=13:44:37 +2026-09-20 14:42:55 | INFO | train rank=0 | step=18520/50000 progress=37.0% loss=0.097230 lr=7.446351e-05 lr_backbone=7.446351e-05 elapsed=08:04:55 avg_step=1.57s recent_step=1.23s eta=13:44:15 +2026-09-20 14:43:12 | INFO | train rank=0 | step=18530/50000 progress=37.1% loss=0.114867 lr=7.443467e-05 lr_backbone=7.443467e-05 elapsed=08:05:11 avg_step=1.57s recent_step=1.66s eta=13:44:01 +2026-09-20 14:43:33 | INFO | train rank=0 | step=18540/50000 progress=37.1% loss=0.104047 lr=7.440581e-05 lr_backbone=7.440581e-05 elapsed=08:05:33 avg_step=1.57s recent_step=2.17s eta=13:43:55 +2026-09-20 14:43:46 | INFO | train rank=0 | step=18550/50000 progress=37.1% loss=0.095552 lr=7.437694e-05 lr_backbone=7.437694e-05 elapsed=08:05:45 avg_step=1.57s recent_step=1.23s eta=13:43:34 +2026-09-20 14:44:01 | INFO | train rank=0 | step=18560/50000 progress=37.1% loss=0.109514 lr=7.434807e-05 lr_backbone=7.434807e-05 elapsed=08:06:01 avg_step=1.57s recent_step=1.57s eta=13:43:18 +2026-09-20 14:44:16 | INFO | train rank=0 | step=18570/50000 progress=37.1% loss=0.102160 lr=7.431918e-05 lr_backbone=7.431918e-05 elapsed=08:06:15 avg_step=1.57s recent_step=1.42s eta=13:43:00 +2026-09-20 14:44:35 | INFO | train rank=0 | step=18580/50000 progress=37.2% loss=0.084205 lr=7.429028e-05 lr_backbone=7.429028e-05 elapsed=08:06:34 avg_step=1.57s recent_step=1.92s eta=13:42:50 +2026-09-20 14:44:49 | INFO | train rank=0 | step=18590/50000 progress=37.2% loss=0.133256 lr=7.426137e-05 lr_backbone=7.426137e-05 elapsed=08:06:48 avg_step=1.57s recent_step=1.39s eta=13:42:31 +2026-09-20 14:45:07 | INFO | train rank=0 | step=18600/50000 progress=37.2% loss=0.097936 lr=7.423245e-05 lr_backbone=7.423245e-05 elapsed=08:07:06 avg_step=1.57s recent_step=1.79s eta=13:42:19 +2026-09-20 14:45:19 | INFO | train rank=0 | step=18610/50000 progress=37.2% loss=0.122028 lr=7.420352e-05 lr_backbone=7.420352e-05 elapsed=08:07:18 avg_step=1.57s recent_step=1.22s eta=13:41:58 +2026-09-20 14:45:37 | INFO | train rank=0 | step=18620/50000 progress=37.2% loss=0.093982 lr=7.417457e-05 lr_backbone=7.417457e-05 elapsed=08:07:36 avg_step=1.57s recent_step=1.77s eta=13:41:45 +2026-09-20 14:45:49 | INFO | train rank=0 | step=18630/50000 progress=37.3% loss=0.107890 lr=7.414562e-05 lr_backbone=7.414562e-05 elapsed=08:07:49 avg_step=1.57s recent_step=1.29s eta=13:41:25 +2026-09-20 14:46:04 | INFO | train rank=0 | step=18640/50000 progress=37.3% loss=0.105507 lr=7.411666e-05 lr_backbone=7.411666e-05 elapsed=08:08:04 avg_step=1.57s recent_step=1.46s eta=13:41:07 +2026-09-20 14:46:21 | INFO | train rank=0 | step=18650/50000 progress=37.3% loss=0.093790 lr=7.408768e-05 lr_backbone=7.408768e-05 elapsed=08:08:21 avg_step=1.57s recent_step=1.74s eta=13:40:54 +2026-09-20 14:46:41 | INFO | train rank=0 | step=18660/50000 progress=37.3% loss=0.076283 lr=7.405870e-05 lr_backbone=7.405870e-05 elapsed=08:08:41 avg_step=1.57s recent_step=1.95s eta=13:40:45 +2026-09-20 14:46:57 | INFO | train rank=0 | step=18670/50000 progress=37.3% loss=0.081481 lr=7.402970e-05 lr_backbone=7.402970e-05 elapsed=08:08:56 avg_step=1.57s recent_step=1.57s eta=13:40:29 +2026-09-20 14:47:08 | INFO | train rank=0 | step=18680/50000 progress=37.4% loss=0.117405 lr=7.400070e-05 lr_backbone=7.400070e-05 elapsed=08:09:08 avg_step=1.57s recent_step=1.17s eta=13:40:07 +2026-09-20 14:47:24 | INFO | train rank=0 | step=18690/50000 progress=37.4% loss=0.152821 lr=7.397168e-05 lr_backbone=7.397168e-05 elapsed=08:09:24 avg_step=1.57s recent_step=1.59s eta=13:39:52 +2026-09-20 14:47:42 | INFO | train rank=0 | step=18700/50000 progress=37.4% loss=0.144819 lr=7.394266e-05 lr_backbone=7.394266e-05 elapsed=08:09:41 avg_step=1.57s recent_step=1.74s eta=13:39:39 +2026-09-20 14:47:55 | INFO | train rank=0 | step=18710/50000 progress=37.4% loss=0.093693 lr=7.391362e-05 lr_backbone=7.391362e-05 elapsed=08:09:55 avg_step=1.57s recent_step=1.35s eta=13:39:20 +2026-09-20 14:48:12 | INFO | train rank=0 | step=18720/50000 progress=37.4% loss=0.136525 lr=7.388457e-05 lr_backbone=7.388457e-05 elapsed=08:10:11 avg_step=1.57s recent_step=1.65s eta=13:39:05 +2026-09-20 14:48:30 | INFO | train rank=0 | step=18730/50000 progress=37.5% loss=0.106818 lr=7.385552e-05 lr_backbone=7.385552e-05 elapsed=08:10:29 avg_step=1.57s recent_step=1.78s eta=13:38:53 +2026-09-20 14:48:42 | INFO | train rank=0 | step=18740/50000 progress=37.5% loss=0.104502 lr=7.382645e-05 lr_backbone=7.382645e-05 elapsed=08:10:42 avg_step=1.57s recent_step=1.26s eta=13:38:32 +2026-09-20 14:48:58 | INFO | train rank=0 | step=18750/50000 progress=37.5% loss=0.098955 lr=7.379737e-05 lr_backbone=7.379737e-05 elapsed=08:10:58 avg_step=1.57s recent_step=1.60s eta=13:38:17 +2026-09-20 14:49:19 | INFO | train rank=0 | step=18760/50000 progress=37.5% loss=0.083868 lr=7.376828e-05 lr_backbone=7.376828e-05 elapsed=08:11:19 avg_step=1.57s recent_step=2.07s eta=13:38:09 +2026-09-20 14:49:31 | INFO | train rank=0 | step=18770/50000 progress=37.5% loss=0.091274 lr=7.373918e-05 lr_backbone=7.373918e-05 elapsed=08:11:31 avg_step=1.57s recent_step=1.20s eta=13:37:47 +2026-09-20 14:49:51 | INFO | train rank=0 | step=18780/50000 progress=37.6% loss=0.104990 lr=7.371007e-05 lr_backbone=7.371007e-05 elapsed=08:11:50 avg_step=1.57s recent_step=1.98s eta=13:37:39 +2026-09-20 14:50:08 | INFO | train rank=0 | step=18790/50000 progress=37.6% loss=0.128018 lr=7.368095e-05 lr_backbone=7.368095e-05 elapsed=08:12:08 avg_step=1.57s recent_step=1.75s eta=13:37:26 +2026-09-20 14:50:20 | INFO | train rank=0 | step=18800/50000 progress=37.6% loss=0.078314 lr=7.365182e-05 lr_backbone=7.365182e-05 elapsed=08:12:20 avg_step=1.57s recent_step=1.20s eta=13:37:04 +2026-09-20 14:50:33 | INFO | train rank=0 | step=18810/50000 progress=37.6% loss=0.093452 lr=7.362268e-05 lr_backbone=7.362268e-05 elapsed=08:12:33 avg_step=1.57s recent_step=1.27s eta=13:36:43 +2026-09-20 14:50:52 | INFO | train rank=0 | step=18820/50000 progress=37.6% loss=0.084902 lr=7.359353e-05 lr_backbone=7.359353e-05 elapsed=08:12:52 avg_step=1.57s recent_step=1.93s eta=13:36:33 +2026-09-20 14:51:04 | INFO | train rank=0 | step=18830/50000 progress=37.7% loss=0.077701 lr=7.356437e-05 lr_backbone=7.356437e-05 elapsed=08:13:04 avg_step=1.57s recent_step=1.22s eta=13:36:12 +2026-09-20 14:51:18 | INFO | train rank=0 | step=18840/50000 progress=37.7% loss=0.118041 lr=7.353520e-05 lr_backbone=7.353520e-05 elapsed=08:13:17 avg_step=1.57s recent_step=1.32s eta=13:35:52 +2026-09-20 14:51:34 | INFO | train rank=0 | step=18850/50000 progress=37.7% loss=0.110763 lr=7.350601e-05 lr_backbone=7.350601e-05 elapsed=08:13:34 avg_step=1.57s recent_step=1.65s eta=13:35:38 +2026-09-20 14:51:53 | INFO | train rank=0 | step=18860/50000 progress=37.7% loss=0.098323 lr=7.347682e-05 lr_backbone=7.347682e-05 elapsed=08:13:53 avg_step=1.57s recent_step=1.92s eta=13:35:28 +2026-09-20 14:52:07 | INFO | train rank=0 | step=18870/50000 progress=37.7% loss=0.130245 lr=7.344762e-05 lr_backbone=7.344762e-05 elapsed=08:14:07 avg_step=1.57s recent_step=1.39s eta=13:35:09 +2026-09-20 14:52:24 | INFO | train rank=0 | step=18880/50000 progress=37.8% loss=0.092897 lr=7.341841e-05 lr_backbone=7.341841e-05 elapsed=08:14:24 avg_step=1.57s recent_step=1.71s eta=13:34:56 +2026-09-20 14:52:39 | INFO | train rank=0 | step=18890/50000 progress=37.8% loss=0.082166 lr=7.338918e-05 lr_backbone=7.338918e-05 elapsed=08:14:38 avg_step=1.57s recent_step=1.43s eta=13:34:38 +2026-09-20 14:52:50 | INFO | train rank=0 | step=18900/50000 progress=37.8% loss=0.103619 lr=7.335995e-05 lr_backbone=7.335995e-05 elapsed=08:14:50 avg_step=1.57s recent_step=1.15s eta=13:34:15 +2026-09-20 14:53:07 | INFO | train rank=0 | step=18910/50000 progress=37.8% loss=0.111259 lr=7.333071e-05 lr_backbone=7.333071e-05 elapsed=08:15:07 avg_step=1.57s recent_step=1.70s eta=13:34:02 +2026-09-20 14:53:27 | INFO | train rank=0 | step=18920/50000 progress=37.8% loss=0.134107 lr=7.330145e-05 lr_backbone=7.330145e-05 elapsed=08:15:27 avg_step=1.57s recent_step=1.99s eta=13:33:53 +2026-09-20 14:53:39 | INFO | train rank=0 | step=18930/50000 progress=37.9% loss=0.091469 lr=7.327219e-05 lr_backbone=7.327219e-05 elapsed=08:15:38 avg_step=1.57s recent_step=1.17s eta=13:33:30 +2026-09-20 14:53:54 | INFO | train rank=0 | step=18940/50000 progress=37.9% loss=0.091953 lr=7.324292e-05 lr_backbone=7.324292e-05 elapsed=08:15:54 avg_step=1.57s recent_step=1.52s eta=13:33:14 +2026-09-20 14:54:09 | INFO | train rank=0 | step=18950/50000 progress=37.9% loss=0.181236 lr=7.321363e-05 lr_backbone=7.321363e-05 elapsed=08:16:09 avg_step=1.57s recent_step=1.50s eta=13:32:57 +2026-09-20 14:54:26 | INFO | train rank=0 | step=18960/50000 progress=37.9% loss=0.076954 lr=7.318434e-05 lr_backbone=7.318434e-05 elapsed=08:16:25 avg_step=1.57s recent_step=1.67s eta=13:32:43 +2026-09-20 14:54:44 | INFO | train rank=0 | step=18970/50000 progress=37.9% loss=0.083747 lr=7.315503e-05 lr_backbone=7.315503e-05 elapsed=08:16:44 avg_step=1.57s recent_step=1.83s eta=13:32:31 +2026-09-20 14:54:59 | INFO | train rank=0 | step=18980/50000 progress=38.0% loss=0.108429 lr=7.312572e-05 lr_backbone=7.312572e-05 elapsed=08:16:59 avg_step=1.57s recent_step=1.55s eta=13:32:15 +2026-09-20 14:55:18 | INFO | train rank=0 | step=18990/50000 progress=38.0% loss=0.100696 lr=7.309639e-05 lr_backbone=7.309639e-05 elapsed=08:17:17 avg_step=1.57s recent_step=1.81s eta=13:32:03 +2026-09-20 14:55:32 | INFO | train rank=0 | step=19000/50000 progress=38.0% loss=0.100176 lr=7.306706e-05 lr_backbone=7.306706e-05 elapsed=08:17:32 avg_step=1.57s recent_step=1.49s eta=13:31:46 +2026-09-20 14:55:46 | INFO | train rank=0 | step=19010/50000 progress=38.0% loss=0.107909 lr=7.303771e-05 lr_backbone=7.303771e-05 elapsed=08:17:45 avg_step=1.57s recent_step=1.31s eta=13:31:26 +2026-09-20 14:56:02 | INFO | train rank=0 | step=19020/50000 progress=38.0% loss=0.080247 lr=7.300836e-05 lr_backbone=7.300836e-05 elapsed=08:18:02 avg_step=1.57s recent_step=1.68s eta=13:31:12 +2026-09-20 14:56:16 | INFO | train rank=0 | step=19030/50000 progress=38.1% loss=0.140984 lr=7.297899e-05 lr_backbone=7.297899e-05 elapsed=08:18:15 avg_step=1.57s recent_step=1.34s eta=13:30:53 +2026-09-20 14:56:31 | INFO | train rank=0 | step=19040/50000 progress=38.1% loss=0.170658 lr=7.294962e-05 lr_backbone=7.294962e-05 elapsed=08:18:30 avg_step=1.57s recent_step=1.51s eta=13:30:36 +2026-09-20 14:56:49 | INFO | train rank=0 | step=19050/50000 progress=38.1% loss=0.103314 lr=7.292023e-05 lr_backbone=7.292023e-05 elapsed=08:18:49 avg_step=1.57s recent_step=1.83s eta=13:30:25 +2026-09-20 14:57:08 | INFO | train rank=0 | step=19060/50000 progress=38.1% loss=0.094075 lr=7.289084e-05 lr_backbone=7.289084e-05 elapsed=08:19:07 avg_step=1.57s recent_step=1.88s eta=13:30:14 +2026-09-20 14:57:22 | INFO | train rank=0 | step=19070/50000 progress=38.1% loss=0.087329 lr=7.286143e-05 lr_backbone=7.286143e-05 elapsed=08:19:21 avg_step=1.57s recent_step=1.37s eta=13:29:55 +2026-09-20 14:57:37 | INFO | train rank=0 | step=19080/50000 progress=38.2% loss=0.088564 lr=7.283202e-05 lr_backbone=7.283202e-05 elapsed=08:19:36 avg_step=1.57s recent_step=1.53s eta=13:29:38 +2026-09-20 14:57:50 | INFO | train rank=0 | step=19090/50000 progress=38.2% loss=0.111648 lr=7.280259e-05 lr_backbone=7.280259e-05 elapsed=08:19:49 avg_step=1.57s recent_step=1.29s eta=13:29:18 +2026-09-20 14:58:08 | INFO | train rank=0 | step=19100/50000 progress=38.2% loss=0.105874 lr=7.277316e-05 lr_backbone=7.277316e-05 elapsed=08:20:08 avg_step=1.57s recent_step=1.84s eta=13:29:07 +2026-09-20 14:58:27 | INFO | train rank=0 | step=19110/50000 progress=38.2% loss=0.103118 lr=7.274371e-05 lr_backbone=7.274371e-05 elapsed=08:20:26 avg_step=1.57s recent_step=1.86s eta=13:28:56 +2026-09-20 14:58:39 | INFO | train rank=0 | step=19120/50000 progress=38.2% loss=0.135049 lr=7.271426e-05 lr_backbone=7.271426e-05 elapsed=08:20:38 avg_step=1.57s recent_step=1.20s eta=13:28:34 +2026-09-20 14:58:56 | INFO | train rank=0 | step=19130/50000 progress=38.3% loss=0.097271 lr=7.268479e-05 lr_backbone=7.268479e-05 elapsed=08:20:55 avg_step=1.57s recent_step=1.72s eta=13:28:21 +2026-09-20 14:59:08 | INFO | train rank=0 | step=19140/50000 progress=38.3% loss=0.075050 lr=7.265532e-05 lr_backbone=7.265532e-05 elapsed=08:21:08 avg_step=1.57s recent_step=1.24s eta=13:28:00 +2026-09-20 14:59:20 | INFO | train rank=0 | step=19150/50000 progress=38.3% loss=0.088517 lr=7.262583e-05 lr_backbone=7.262583e-05 elapsed=08:21:20 avg_step=1.57s recent_step=1.19s eta=13:27:38 +2026-09-20 14:59:33 | INFO | train rank=0 | step=19160/50000 progress=38.3% loss=0.112332 lr=7.259634e-05 lr_backbone=7.259634e-05 elapsed=08:21:32 avg_step=1.57s recent_step=1.24s eta=13:27:17 +2026-09-20 14:59:47 | INFO | train rank=0 | step=19170/50000 progress=38.3% loss=0.114679 lr=7.256683e-05 lr_backbone=7.256683e-05 elapsed=08:21:47 avg_step=1.57s recent_step=1.44s eta=13:26:59 +2026-09-20 15:00:09 | INFO | train rank=0 | step=19180/50000 progress=38.4% loss=0.105840 lr=7.253732e-05 lr_backbone=7.253732e-05 elapsed=08:22:09 avg_step=1.57s recent_step=2.22s eta=13:26:54 +2026-09-20 15:00:22 | INFO | train rank=0 | step=19190/50000 progress=38.4% loss=0.119248 lr=7.250779e-05 lr_backbone=7.250779e-05 elapsed=08:22:22 avg_step=1.57s recent_step=1.33s eta=13:26:34 +2026-09-20 15:00:37 | INFO | train rank=0 | step=19200/50000 progress=38.4% loss=0.103588 lr=7.247826e-05 lr_backbone=7.247826e-05 elapsed=08:22:37 avg_step=1.57s recent_step=1.47s eta=13:26:17 +2026-09-20 15:00:53 | INFO | train rank=0 | step=19210/50000 progress=38.4% loss=0.098643 lr=7.244871e-05 lr_backbone=7.244871e-05 elapsed=08:22:52 avg_step=1.57s recent_step=1.54s eta=13:26:01 +2026-09-20 15:01:15 | INFO | train rank=0 | step=19220/50000 progress=38.4% loss=0.123838 lr=7.241916e-05 lr_backbone=7.241916e-05 elapsed=08:23:15 avg_step=1.57s recent_step=2.29s eta=13:25:56 +2026-09-20 15:01:27 | INFO | train rank=0 | step=19230/50000 progress=38.5% loss=0.090300 lr=7.238960e-05 lr_backbone=7.238960e-05 elapsed=08:23:27 avg_step=1.57s recent_step=1.18s eta=13:25:35 +2026-09-20 15:01:40 | INFO | train rank=0 | step=19240/50000 progress=38.5% loss=0.098295 lr=7.236002e-05 lr_backbone=7.236002e-05 elapsed=08:23:39 avg_step=1.57s recent_step=1.23s eta=13:25:13 +2026-09-20 15:01:57 | INFO | train rank=0 | step=19250/50000 progress=38.5% loss=0.083497 lr=7.233044e-05 lr_backbone=7.233044e-05 elapsed=08:23:56 avg_step=1.57s recent_step=1.71s eta=13:25:00 +2026-09-20 15:02:11 | INFO | train rank=0 | step=19260/50000 progress=38.5% loss=0.080866 lr=7.230085e-05 lr_backbone=7.230085e-05 elapsed=08:24:10 avg_step=1.57s recent_step=1.41s eta=13:24:42 +2026-09-20 15:02:29 | INFO | train rank=0 | step=19270/50000 progress=38.5% loss=0.123580 lr=7.227124e-05 lr_backbone=7.227124e-05 elapsed=08:24:29 avg_step=1.57s recent_step=1.86s eta=13:24:31 +2026-09-20 15:02:43 | INFO | train rank=0 | step=19280/50000 progress=38.6% loss=0.089524 lr=7.224163e-05 lr_backbone=7.224163e-05 elapsed=08:24:42 avg_step=1.57s recent_step=1.30s eta=13:24:11 +2026-09-20 15:02:55 | INFO | train rank=0 | step=19290/50000 progress=38.6% loss=0.095445 lr=7.221201e-05 lr_backbone=7.221201e-05 elapsed=08:24:55 avg_step=1.57s recent_step=1.27s eta=13:23:50 +2026-09-20 15:03:15 | INFO | train rank=0 | step=19300/50000 progress=38.6% loss=0.101800 lr=7.218238e-05 lr_backbone=7.218238e-05 elapsed=08:25:14 avg_step=1.57s recent_step=1.96s eta=13:23:41 +2026-09-20 15:03:30 | INFO | train rank=0 | step=19310/50000 progress=38.6% loss=0.091626 lr=7.215274e-05 lr_backbone=7.215274e-05 elapsed=08:25:30 avg_step=1.57s recent_step=1.52s eta=13:23:24 +2026-09-20 15:03:45 | INFO | train rank=0 | step=19320/50000 progress=38.6% loss=0.096516 lr=7.212308e-05 lr_backbone=7.212308e-05 elapsed=08:25:45 avg_step=1.57s recent_step=1.55s eta=13:23:08 +2026-09-20 15:04:00 | INFO | train rank=0 | step=19330/50000 progress=38.7% loss=0.092474 lr=7.209342e-05 lr_backbone=7.209342e-05 elapsed=08:25:59 avg_step=1.57s recent_step=1.42s eta=13:22:50 +2026-09-20 15:04:17 | INFO | train rank=0 | step=19340/50000 progress=38.7% loss=0.094816 lr=7.206375e-05 lr_backbone=7.206375e-05 elapsed=08:26:17 avg_step=1.57s recent_step=1.78s eta=13:22:37 +2026-09-20 15:04:34 | INFO | train rank=0 | step=19350/50000 progress=38.7% loss=0.106026 lr=7.203407e-05 lr_backbone=7.203407e-05 elapsed=08:26:33 avg_step=1.57s recent_step=1.63s eta=13:22:23 +2026-09-20 15:04:46 | INFO | train rank=0 | step=19360/50000 progress=38.7% loss=0.117248 lr=7.200438e-05 lr_backbone=7.200438e-05 elapsed=08:26:46 avg_step=1.57s recent_step=1.24s eta=13:22:02 +2026-09-20 15:04:59 | INFO | train rank=0 | step=19370/50000 progress=38.7% loss=0.131034 lr=7.197468e-05 lr_backbone=7.197468e-05 elapsed=08:26:59 avg_step=1.57s recent_step=1.32s eta=13:21:42 +2026-09-20 15:05:18 | INFO | train rank=0 | step=19380/50000 progress=38.8% loss=0.081824 lr=7.194497e-05 lr_backbone=7.194497e-05 elapsed=08:27:18 avg_step=1.57s recent_step=1.88s eta=13:21:31 +2026-09-20 15:05:34 | INFO | train rank=0 | step=19390/50000 progress=38.8% loss=0.088534 lr=7.191525e-05 lr_backbone=7.191525e-05 elapsed=08:27:33 avg_step=1.57s recent_step=1.53s eta=13:21:15 +2026-09-20 15:05:53 | INFO | train rank=0 | step=19400/50000 progress=38.8% loss=0.096820 lr=7.188553e-05 lr_backbone=7.188553e-05 elapsed=08:27:53 avg_step=1.57s recent_step=1.96s eta=13:21:06 +2026-09-20 15:06:05 | INFO | train rank=0 | step=19410/50000 progress=38.8% loss=0.104343 lr=7.185579e-05 lr_backbone=7.185579e-05 elapsed=08:28:05 avg_step=1.57s recent_step=1.20s eta=13:20:44 +2026-09-20 15:06:25 | INFO | train rank=0 | step=19420/50000 progress=38.8% loss=0.099369 lr=7.182604e-05 lr_backbone=7.182604e-05 elapsed=08:28:24 avg_step=1.57s recent_step=1.95s eta=13:20:34 +2026-09-20 15:06:41 | INFO | train rank=0 | step=19430/50000 progress=38.9% loss=0.125584 lr=7.179628e-05 lr_backbone=7.179628e-05 elapsed=08:28:41 avg_step=1.57s recent_step=1.65s eta=13:20:20 +2026-09-20 15:06:58 | INFO | train rank=0 | step=19440/50000 progress=38.9% loss=0.138128 lr=7.176652e-05 lr_backbone=7.176652e-05 elapsed=08:28:58 avg_step=1.57s recent_step=1.69s eta=13:20:06 +2026-09-20 15:07:16 | INFO | train rank=0 | step=19450/50000 progress=38.9% loss=0.098139 lr=7.173674e-05 lr_backbone=7.173674e-05 elapsed=08:29:16 avg_step=1.57s recent_step=1.80s eta=13:19:54 +2026-09-20 15:07:28 | INFO | train rank=0 | step=19460/50000 progress=38.9% loss=0.092081 lr=7.170696e-05 lr_backbone=7.170696e-05 elapsed=08:29:27 avg_step=1.57s recent_step=1.19s eta=13:19:32 +2026-09-20 15:07:40 | INFO | train rank=0 | step=19470/50000 progress=38.9% loss=0.100203 lr=7.167716e-05 lr_backbone=7.167716e-05 elapsed=08:29:40 avg_step=1.57s recent_step=1.25s eta=13:19:11 +2026-09-20 15:07:56 | INFO | train rank=0 | step=19480/50000 progress=39.0% loss=0.091492 lr=7.164736e-05 lr_backbone=7.164736e-05 elapsed=08:29:55 avg_step=1.57s recent_step=1.55s eta=13:18:55 +2026-09-20 15:08:19 | INFO | train rank=0 | step=19490/50000 progress=39.0% loss=0.079284 lr=7.161754e-05 lr_backbone=7.161754e-05 elapsed=08:30:18 avg_step=1.57s recent_step=2.28s eta=13:18:51 +2026-09-20 15:08:31 | INFO | train rank=0 | step=19500/50000 progress=39.0% loss=0.111408 lr=7.158772e-05 lr_backbone=7.158772e-05 elapsed=08:30:30 avg_step=1.57s recent_step=1.19s eta=13:18:29 +2026-09-20 15:09:30 | INFO | train rank=0 | step=19510/50000 progress=39.0% loss=0.105822 lr=7.155788e-05 lr_backbone=7.155788e-05 elapsed=08:31:30 avg_step=1.57s recent_step=5.99s eta=13:19:22 +2026-09-20 15:09:43 | INFO | train rank=0 | step=19520/50000 progress=39.0% loss=0.106929 lr=7.152804e-05 lr_backbone=7.152804e-05 elapsed=08:31:43 avg_step=1.57s recent_step=1.28s eta=13:19:02 +2026-09-20 15:10:03 | INFO | train rank=0 | step=19530/50000 progress=39.1% loss=0.102599 lr=7.149819e-05 lr_backbone=7.149819e-05 elapsed=08:32:03 avg_step=1.57s recent_step=1.98s eta=13:18:53 +2026-09-20 15:10:15 | INFO | train rank=0 | step=19540/50000 progress=39.1% loss=0.076069 lr=7.146833e-05 lr_backbone=7.146833e-05 elapsed=08:32:15 avg_step=1.57s recent_step=1.19s eta=13:18:31 +2026-09-20 15:10:27 | INFO | train rank=0 | step=19550/50000 progress=39.1% loss=0.116746 lr=7.143846e-05 lr_backbone=7.143846e-05 elapsed=08:32:27 avg_step=1.57s recent_step=1.20s eta=13:18:09 +2026-09-20 15:10:46 | INFO | train rank=0 | step=19560/50000 progress=39.1% loss=0.096482 lr=7.140858e-05 lr_backbone=7.140858e-05 elapsed=08:32:46 avg_step=1.57s recent_step=1.91s eta=13:17:59 +2026-09-20 15:11:00 | INFO | train rank=0 | step=19570/50000 progress=39.1% loss=0.086677 lr=7.137869e-05 lr_backbone=7.137869e-05 elapsed=08:32:59 avg_step=1.57s recent_step=1.38s eta=13:17:40 +2026-09-20 15:11:16 | INFO | train rank=0 | step=19580/50000 progress=39.2% loss=0.094671 lr=7.134879e-05 lr_backbone=7.134879e-05 elapsed=08:33:15 avg_step=1.57s recent_step=1.57s eta=13:17:24 +2026-09-20 15:11:31 | INFO | train rank=0 | step=19590/50000 progress=39.2% loss=0.100032 lr=7.131888e-05 lr_backbone=7.131888e-05 elapsed=08:33:31 avg_step=1.57s recent_step=1.55s eta=13:17:08 +2026-09-20 15:11:45 | INFO | train rank=0 | step=19600/50000 progress=39.2% loss=0.084272 lr=7.128896e-05 lr_backbone=7.128896e-05 elapsed=08:33:45 avg_step=1.57s recent_step=1.40s eta=13:16:50 +2026-09-20 15:12:03 | INFO | train rank=0 | step=19610/50000 progress=39.2% loss=0.087714 lr=7.125904e-05 lr_backbone=7.125904e-05 elapsed=08:34:02 avg_step=1.57s recent_step=1.74s eta=13:16:37 +2026-09-20 15:12:20 | INFO | train rank=0 | step=19620/50000 progress=39.2% loss=0.091650 lr=7.122910e-05 lr_backbone=7.122910e-05 elapsed=08:34:19 avg_step=1.57s recent_step=1.71s eta=13:16:23 +2026-09-20 15:12:35 | INFO | train rank=0 | step=19630/50000 progress=39.3% loss=0.090021 lr=7.119916e-05 lr_backbone=7.119916e-05 elapsed=08:34:34 avg_step=1.57s recent_step=1.50s eta=13:16:07 +2026-09-20 15:12:50 | INFO | train rank=0 | step=19640/50000 progress=39.3% loss=0.152715 lr=7.116920e-05 lr_backbone=7.116920e-05 elapsed=08:34:50 avg_step=1.57s recent_step=1.53s eta=13:15:50 +2026-09-20 15:13:06 | INFO | train rank=0 | step=19650/50000 progress=39.3% loss=0.140392 lr=7.113924e-05 lr_backbone=7.113924e-05 elapsed=08:35:06 avg_step=1.57s recent_step=1.61s eta=13:15:35 +2026-09-20 15:13:22 | INFO | train rank=0 | step=19660/50000 progress=39.3% loss=0.093704 lr=7.110926e-05 lr_backbone=7.110926e-05 elapsed=08:35:22 avg_step=1.57s recent_step=1.59s eta=13:15:20 +2026-09-20 15:13:41 | INFO | train rank=0 | step=19670/50000 progress=39.3% loss=0.095687 lr=7.107928e-05 lr_backbone=7.107928e-05 elapsed=08:35:41 avg_step=1.57s recent_step=1.92s eta=13:15:09 +2026-09-20 15:13:53 | INFO | train rank=0 | step=19680/50000 progress=39.4% loss=0.093137 lr=7.104929e-05 lr_backbone=7.104929e-05 elapsed=08:35:53 avg_step=1.57s recent_step=1.20s eta=13:14:48 +2026-09-20 15:14:05 | INFO | train rank=0 | step=19690/50000 progress=39.4% loss=0.133745 lr=7.101929e-05 lr_backbone=7.101929e-05 elapsed=08:36:05 avg_step=1.57s recent_step=1.18s eta=13:14:26 +2026-09-20 15:14:20 | INFO | train rank=0 | step=19700/50000 progress=39.4% loss=0.088205 lr=7.098928e-05 lr_backbone=7.098928e-05 elapsed=08:36:20 avg_step=1.57s recent_step=1.50s eta=13:14:09 +2026-09-20 15:14:33 | INFO | train rank=0 | step=19710/50000 progress=39.4% loss=0.085726 lr=7.095926e-05 lr_backbone=7.095926e-05 elapsed=08:36:32 avg_step=1.57s recent_step=1.25s eta=13:13:49 +2026-09-20 15:14:48 | INFO | train rank=0 | step=19720/50000 progress=39.4% loss=0.092472 lr=7.092923e-05 lr_backbone=7.092923e-05 elapsed=08:36:48 avg_step=1.57s recent_step=1.59s eta=13:13:33 +2026-09-20 15:15:09 | INFO | train rank=0 | step=19730/50000 progress=39.5% loss=0.148455 lr=7.089920e-05 lr_backbone=7.089920e-05 elapsed=08:37:08 avg_step=1.57s recent_step=2.00s eta=13:13:24 +2026-09-20 15:15:27 | INFO | train rank=0 | step=19740/50000 progress=39.5% loss=0.119610 lr=7.086915e-05 lr_backbone=7.086915e-05 elapsed=08:37:27 avg_step=1.57s recent_step=1.90s eta=13:13:13 +2026-09-20 15:15:40 | INFO | train rank=0 | step=19750/50000 progress=39.5% loss=0.129347 lr=7.083909e-05 lr_backbone=7.083909e-05 elapsed=08:37:39 avg_step=1.57s recent_step=1.22s eta=13:12:52 +2026-09-20 15:15:56 | INFO | train rank=0 | step=19760/50000 progress=39.5% loss=0.097944 lr=7.080903e-05 lr_backbone=7.080903e-05 elapsed=08:37:56 avg_step=1.57s recent_step=1.64s eta=13:12:37 +2026-09-20 15:16:10 | INFO | train rank=0 | step=19770/50000 progress=39.5% loss=0.090450 lr=7.077895e-05 lr_backbone=7.077895e-05 elapsed=08:38:09 avg_step=1.57s recent_step=1.37s eta=13:12:19 +2026-09-20 15:16:33 | INFO | train rank=0 | step=19780/50000 progress=39.6% loss=0.093145 lr=7.074887e-05 lr_backbone=7.074887e-05 elapsed=08:38:32 avg_step=1.57s recent_step=2.29s eta=13:12:14 +2026-09-20 15:16:44 | INFO | train rank=0 | step=19790/50000 progress=39.6% loss=0.091856 lr=7.071878e-05 lr_backbone=7.071878e-05 elapsed=08:38:44 avg_step=1.57s recent_step=1.19s eta=13:11:52 +2026-09-20 15:17:02 | INFO | train rank=0 | step=19800/50000 progress=39.6% loss=0.081320 lr=7.068868e-05 lr_backbone=7.068868e-05 elapsed=08:39:01 avg_step=1.57s recent_step=1.73s eta=13:11:39 +2026-09-20 15:17:14 | INFO | train rank=0 | step=19810/50000 progress=39.6% loss=0.104028 lr=7.065857e-05 lr_backbone=7.065857e-05 elapsed=08:39:14 avg_step=1.57s recent_step=1.23s eta=13:11:18 +2026-09-20 15:17:35 | INFO | train rank=0 | step=19820/50000 progress=39.6% loss=0.090832 lr=7.062845e-05 lr_backbone=7.062845e-05 elapsed=08:39:35 avg_step=1.57s recent_step=2.08s eta=13:11:10 +2026-09-20 15:17:53 | INFO | train rank=0 | step=19830/50000 progress=39.7% loss=0.104235 lr=7.059832e-05 lr_backbone=7.059832e-05 elapsed=08:39:52 avg_step=1.57s recent_step=1.78s eta=13:10:57 +2026-09-20 15:18:05 | INFO | train rank=0 | step=19840/50000 progress=39.7% loss=0.095297 lr=7.056818e-05 lr_backbone=7.056818e-05 elapsed=08:40:04 avg_step=1.57s recent_step=1.20s eta=13:10:36 +2026-09-20 15:18:23 | INFO | train rank=0 | step=19850/50000 progress=39.7% loss=0.136204 lr=7.053804e-05 lr_backbone=7.053804e-05 elapsed=08:40:23 avg_step=1.57s recent_step=1.83s eta=13:10:24 +2026-09-20 15:18:36 | INFO | train rank=0 | step=19860/50000 progress=39.7% loss=0.098822 lr=7.050788e-05 lr_backbone=7.050788e-05 elapsed=08:40:35 avg_step=1.57s recent_step=1.27s eta=13:10:04 +2026-09-20 15:18:54 | INFO | train rank=0 | step=19870/50000 progress=39.7% loss=0.093381 lr=7.047772e-05 lr_backbone=7.047772e-05 elapsed=08:40:54 avg_step=1.57s recent_step=1.84s eta=13:09:52 +2026-09-20 15:19:14 | INFO | train rank=0 | step=19880/50000 progress=39.8% loss=0.122296 lr=7.044754e-05 lr_backbone=7.044754e-05 elapsed=08:41:13 avg_step=1.57s recent_step=1.96s eta=13:09:42 +2026-09-20 15:19:26 | INFO | train rank=0 | step=19890/50000 progress=39.8% loss=0.082358 lr=7.041736e-05 lr_backbone=7.041736e-05 elapsed=08:41:25 avg_step=1.57s recent_step=1.22s eta=13:09:21 +2026-09-20 15:19:38 | INFO | train rank=0 | step=19900/50000 progress=39.8% loss=0.100320 lr=7.038717e-05 lr_backbone=7.038717e-05 elapsed=08:41:38 avg_step=1.57s recent_step=1.24s eta=13:09:00 +2026-09-20 15:20:00 | INFO | train rank=0 | step=19910/50000 progress=39.8% loss=0.074011 lr=7.035697e-05 lr_backbone=7.035697e-05 elapsed=08:42:00 avg_step=1.57s recent_step=2.19s eta=13:08:54 +2026-09-20 15:20:12 | INFO | train rank=0 | step=19920/50000 progress=39.8% loss=0.121458 lr=7.032676e-05 lr_backbone=7.032676e-05 elapsed=08:42:12 avg_step=1.57s recent_step=1.23s eta=13:08:33 +2026-09-20 15:20:25 | INFO | train rank=0 | step=19930/50000 progress=39.9% loss=0.087812 lr=7.029654e-05 lr_backbone=7.029654e-05 elapsed=08:42:24 avg_step=1.57s recent_step=1.22s eta=13:08:12 +2026-09-20 15:20:45 | INFO | train rank=0 | step=19940/50000 progress=39.9% loss=0.086096 lr=7.026632e-05 lr_backbone=7.026632e-05 elapsed=08:42:45 avg_step=1.57s recent_step=2.04s eta=13:08:03 +2026-09-20 15:21:01 | INFO | train rank=0 | step=19950/50000 progress=39.9% loss=0.092253 lr=7.023608e-05 lr_backbone=7.023608e-05 elapsed=08:43:01 avg_step=1.57s recent_step=1.64s eta=13:07:48 +2026-09-20 15:21:13 | INFO | train rank=0 | step=19960/50000 progress=39.9% loss=0.097433 lr=7.020584e-05 lr_backbone=7.020584e-05 elapsed=08:43:13 avg_step=1.57s recent_step=1.18s eta=13:07:27 +2026-09-20 15:21:26 | INFO | train rank=0 | step=19970/50000 progress=39.9% loss=0.077721 lr=7.017558e-05 lr_backbone=7.017558e-05 elapsed=08:43:26 avg_step=1.57s recent_step=1.33s eta=13:07:07 +2026-09-20 15:21:46 | INFO | train rank=0 | step=19980/50000 progress=40.0% loss=0.107777 lr=7.014532e-05 lr_backbone=7.014532e-05 elapsed=08:43:46 avg_step=1.57s recent_step=1.96s eta=13:06:57 +2026-09-20 15:22:00 | INFO | train rank=0 | step=19990/50000 progress=40.0% loss=0.075622 lr=7.011505e-05 lr_backbone=7.011505e-05 elapsed=08:44:00 avg_step=1.57s recent_step=1.39s eta=13:06:39 +2026-09-20 15:22:20 | INFO | train rank=0 | step=20000/50000 progress=40.0% loss=0.097580 lr=7.008477e-05 lr_backbone=7.008477e-05 elapsed=08:44:20 avg_step=1.57s recent_step=1.99s eta=13:06:30 +2026-09-20 15:22:32 | INFO | train rank=0 | step=20010/50000 progress=40.0% loss=0.125020 lr=7.005448e-05 lr_backbone=7.005448e-05 elapsed=08:44:31 avg_step=1.57s recent_step=1.19s eta=13:06:08 +2026-09-20 15:22:49 | INFO | train rank=0 | step=20020/50000 progress=40.0% loss=0.130200 lr=7.002419e-05 lr_backbone=7.002419e-05 elapsed=08:44:49 avg_step=1.57s recent_step=1.76s eta=13:05:55 +2026-09-20 15:23:05 | INFO | train rank=0 | step=20030/50000 progress=40.1% loss=0.093342 lr=6.999388e-05 lr_backbone=6.999388e-05 elapsed=08:45:05 avg_step=1.57s recent_step=1.56s eta=13:05:39 +2026-09-20 15:23:19 | INFO | train rank=0 | step=20040/50000 progress=40.1% loss=0.140939 lr=6.996356e-05 lr_backbone=6.996356e-05 elapsed=08:45:19 avg_step=1.57s recent_step=1.39s eta=13:05:21 +2026-09-20 15:23:35 | INFO | train rank=0 | step=20050/50000 progress=40.1% loss=0.107276 lr=6.993324e-05 lr_backbone=6.993324e-05 elapsed=08:45:35 avg_step=1.57s recent_step=1.61s eta=13:05:06 +2026-09-20 15:23:53 | INFO | train rank=0 | step=20060/50000 progress=40.1% loss=0.126713 lr=6.990291e-05 lr_backbone=6.990291e-05 elapsed=08:45:53 avg_step=1.57s recent_step=1.79s eta=13:04:53 +2026-09-20 15:24:10 | INFO | train rank=0 | step=20070/50000 progress=40.1% loss=0.080726 lr=6.987257e-05 lr_backbone=6.987257e-05 elapsed=08:46:09 avg_step=1.57s recent_step=1.69s eta=13:04:39 +2026-09-20 15:24:22 | INFO | train rank=0 | step=20080/50000 progress=40.2% loss=0.086958 lr=6.984222e-05 lr_backbone=6.984222e-05 elapsed=08:46:22 avg_step=1.57s recent_step=1.24s eta=13:04:19 +2026-09-20 15:24:37 | INFO | train rank=0 | step=20090/50000 progress=40.2% loss=0.110976 lr=6.981186e-05 lr_backbone=6.981186e-05 elapsed=08:46:36 avg_step=1.57s recent_step=1.45s eta=13:04:01 +2026-09-20 15:24:51 | INFO | train rank=0 | step=20100/50000 progress=40.2% loss=0.111853 lr=6.978149e-05 lr_backbone=6.978149e-05 elapsed=08:46:50 avg_step=1.57s recent_step=1.38s eta=13:03:42 +2026-09-20 15:25:11 | INFO | train rank=0 | step=20110/50000 progress=40.2% loss=0.098716 lr=6.975112e-05 lr_backbone=6.975112e-05 elapsed=08:47:11 avg_step=1.57s recent_step=2.05s eta=13:03:34 +2026-09-20 15:25:23 | INFO | train rank=0 | step=20120/50000 progress=40.2% loss=0.090657 lr=6.972073e-05 lr_backbone=6.972073e-05 elapsed=08:47:23 avg_step=1.57s recent_step=1.20s eta=13:03:12 +2026-09-20 15:25:40 | INFO | train rank=0 | step=20130/50000 progress=40.3% loss=0.154439 lr=6.969034e-05 lr_backbone=6.969034e-05 elapsed=08:47:40 avg_step=1.57s recent_step=1.74s eta=13:02:59 +2026-09-20 15:25:59 | INFO | train rank=0 | step=20140/50000 progress=40.3% loss=0.088069 lr=6.965994e-05 lr_backbone=6.965994e-05 elapsed=08:47:59 avg_step=1.57s recent_step=1.87s eta=13:02:48 +2026-09-20 15:26:11 | INFO | train rank=0 | step=20150/50000 progress=40.3% loss=0.112156 lr=6.962953e-05 lr_backbone=6.962953e-05 elapsed=08:48:11 avg_step=1.57s recent_step=1.19s eta=13:02:26 +2026-09-20 15:26:25 | INFO | train rank=0 | step=20160/50000 progress=40.3% loss=0.080076 lr=6.959911e-05 lr_backbone=6.959911e-05 elapsed=08:48:24 avg_step=1.57s recent_step=1.38s eta=13:02:08 +2026-09-20 15:26:41 | INFO | train rank=0 | step=20170/50000 progress=40.3% loss=0.090725 lr=6.956868e-05 lr_backbone=6.956868e-05 elapsed=08:48:41 avg_step=1.57s recent_step=1.65s eta=13:01:53 +2026-09-20 15:26:59 | INFO | train rank=0 | step=20180/50000 progress=40.4% loss=0.121153 lr=6.953825e-05 lr_backbone=6.953825e-05 elapsed=08:48:59 avg_step=1.57s recent_step=1.79s eta=13:01:41 +2026-09-20 15:27:15 | INFO | train rank=0 | step=20190/50000 progress=40.4% loss=0.109946 lr=6.950780e-05 lr_backbone=6.950780e-05 elapsed=08:49:15 avg_step=1.57s recent_step=1.58s eta=13:01:25 +2026-09-20 15:27:34 | INFO | train rank=0 | step=20200/50000 progress=40.4% loss=0.093562 lr=6.947735e-05 lr_backbone=6.947735e-05 elapsed=08:49:33 avg_step=1.57s recent_step=1.89s eta=13:01:14 +2026-09-20 15:27:46 | INFO | train rank=0 | step=20210/50000 progress=40.4% loss=0.077863 lr=6.944689e-05 lr_backbone=6.944689e-05 elapsed=08:49:45 avg_step=1.57s recent_step=1.17s eta=13:00:52 +2026-09-20 15:27:58 | INFO | train rank=0 | step=20220/50000 progress=40.4% loss=0.086987 lr=6.941642e-05 lr_backbone=6.941642e-05 elapsed=08:49:57 avg_step=1.57s recent_step=1.22s eta=13:00:31 +2026-09-20 15:28:10 | INFO | train rank=0 | step=20230/50000 progress=40.5% loss=0.101733 lr=6.938594e-05 lr_backbone=6.938594e-05 elapsed=08:50:10 avg_step=1.57s recent_step=1.26s eta=13:00:11 +2026-09-20 15:28:31 | INFO | train rank=0 | step=20240/50000 progress=40.5% loss=0.088494 lr=6.935545e-05 lr_backbone=6.935545e-05 elapsed=08:50:30 avg_step=1.57s recent_step=2.03s eta=13:00:02 +2026-09-20 15:28:45 | INFO | train rank=0 | step=20250/50000 progress=40.5% loss=0.086167 lr=6.932496e-05 lr_backbone=6.932496e-05 elapsed=08:50:44 avg_step=1.57s recent_step=1.40s eta=12:59:44 +2026-09-20 15:29:00 | INFO | train rank=0 | step=20260/50000 progress=40.5% loss=0.102608 lr=6.929445e-05 lr_backbone=6.929445e-05 elapsed=08:50:59 avg_step=1.57s recent_step=1.50s eta=12:59:27 +2026-09-20 15:29:19 | INFO | train rank=0 | step=20270/50000 progress=40.5% loss=0.065582 lr=6.926394e-05 lr_backbone=6.926394e-05 elapsed=08:51:19 avg_step=1.57s recent_step=1.94s eta=12:59:17 +2026-09-20 15:29:31 | INFO | train rank=0 | step=20280/50000 progress=40.6% loss=0.080785 lr=6.923342e-05 lr_backbone=6.923342e-05 elapsed=08:51:31 avg_step=1.57s recent_step=1.22s eta=12:58:56 +2026-09-20 15:29:46 | INFO | train rank=0 | step=20290/50000 progress=40.6% loss=0.098362 lr=6.920289e-05 lr_backbone=6.920289e-05 elapsed=08:51:45 avg_step=1.57s recent_step=1.45s eta=12:58:38 +2026-09-20 15:30:03 | INFO | train rank=0 | step=20300/50000 progress=40.6% loss=0.103260 lr=6.917236e-05 lr_backbone=6.917236e-05 elapsed=08:52:03 avg_step=1.57s recent_step=1.77s eta=12:58:25 +2026-09-20 15:30:17 | INFO | train rank=0 | step=20310/50000 progress=40.6% loss=0.097954 lr=6.914181e-05 lr_backbone=6.914181e-05 elapsed=08:52:16 avg_step=1.57s recent_step=1.33s eta=12:58:06 +2026-09-20 15:30:31 | INFO | train rank=0 | step=20320/50000 progress=40.6% loss=0.090483 lr=6.911126e-05 lr_backbone=6.911126e-05 elapsed=08:52:31 avg_step=1.57s recent_step=1.43s eta=12:57:48 +2026-09-20 15:30:53 | INFO | train rank=0 | step=20330/50000 progress=40.7% loss=0.101362 lr=6.908069e-05 lr_backbone=6.908069e-05 elapsed=08:52:52 avg_step=1.57s recent_step=2.17s eta=12:57:41 +2026-09-20 15:31:08 | INFO | train rank=0 | step=20340/50000 progress=40.7% loss=0.102643 lr=6.905012e-05 lr_backbone=6.905012e-05 elapsed=08:53:08 avg_step=1.57s recent_step=1.55s eta=12:57:25 +2026-09-20 15:31:25 | INFO | train rank=0 | step=20350/50000 progress=40.7% loss=0.108234 lr=6.901954e-05 lr_backbone=6.901954e-05 elapsed=08:53:25 avg_step=1.57s recent_step=1.68s eta=12:57:11 +2026-09-20 15:31:40 | INFO | train rank=0 | step=20360/50000 progress=40.7% loss=0.105786 lr=6.898895e-05 lr_backbone=6.898895e-05 elapsed=08:53:40 avg_step=1.57s recent_step=1.49s eta=12:56:54 +2026-09-20 15:31:56 | INFO | train rank=0 | step=20370/50000 progress=40.7% loss=0.064028 lr=6.895836e-05 lr_backbone=6.895836e-05 elapsed=08:53:56 avg_step=1.57s recent_step=1.62s eta=12:56:39 +2026-09-20 15:32:53 | INFO | train rank=0 | step=20380/50000 progress=40.8% loss=0.088690 lr=6.892775e-05 lr_backbone=6.892775e-05 elapsed=08:54:52 avg_step=1.57s recent_step=5.65s eta=12:57:23 +2026-09-20 15:33:06 | INFO | train rank=0 | step=20390/50000 progress=40.8% loss=0.116339 lr=6.889714e-05 lr_backbone=6.889714e-05 elapsed=08:55:05 avg_step=1.57s recent_step=1.30s eta=12:57:03 +2026-09-20 15:33:22 | INFO | train rank=0 | step=20400/50000 progress=40.8% loss=0.153562 lr=6.886652e-05 lr_backbone=6.886652e-05 elapsed=08:55:22 avg_step=1.57s recent_step=1.67s eta=12:56:49 +2026-09-20 15:33:35 | INFO | train rank=0 | step=20410/50000 progress=40.8% loss=0.132922 lr=6.883589e-05 lr_backbone=6.883589e-05 elapsed=08:55:35 avg_step=1.57s recent_step=1.30s eta=12:56:29 +2026-09-20 15:33:53 | INFO | train rank=0 | step=20420/50000 progress=40.8% loss=0.092310 lr=6.880526e-05 lr_backbone=6.880526e-05 elapsed=08:55:52 avg_step=1.57s recent_step=1.74s eta=12:56:16 +2026-09-20 15:34:06 | INFO | train rank=0 | step=20430/50000 progress=40.9% loss=0.087068 lr=6.877461e-05 lr_backbone=6.877461e-05 elapsed=08:56:06 avg_step=1.57s recent_step=1.30s eta=12:55:56 +2026-09-20 15:34:18 | INFO | train rank=0 | step=20440/50000 progress=40.9% loss=0.095732 lr=6.874396e-05 lr_backbone=6.874396e-05 elapsed=08:56:18 avg_step=1.57s recent_step=1.22s eta=12:55:35 +2026-09-20 15:34:37 | INFO | train rank=0 | step=20450/50000 progress=40.9% loss=0.140935 lr=6.871329e-05 lr_backbone=6.871329e-05 elapsed=08:56:37 avg_step=1.57s recent_step=1.89s eta=12:55:24 +2026-09-20 15:34:52 | INFO | train rank=0 | step=20460/50000 progress=40.9% loss=0.109747 lr=6.868262e-05 lr_backbone=6.868262e-05 elapsed=08:56:52 avg_step=1.57s recent_step=1.54s eta=12:55:08 +2026-09-20 15:35:05 | INFO | train rank=0 | step=20470/50000 progress=40.9% loss=0.108864 lr=6.865195e-05 lr_backbone=6.865195e-05 elapsed=08:57:05 avg_step=1.57s recent_step=1.28s eta=12:54:48 +2026-09-20 15:35:25 | INFO | train rank=0 | step=20480/50000 progress=41.0% loss=0.114295 lr=6.862126e-05 lr_backbone=6.862126e-05 elapsed=08:57:25 avg_step=1.57s recent_step=2.02s eta=12:54:38 +2026-09-20 15:35:39 | INFO | train rank=0 | step=20490/50000 progress=41.0% loss=0.114453 lr=6.859057e-05 lr_backbone=6.859057e-05 elapsed=08:57:39 avg_step=1.57s recent_step=1.38s eta=12:54:20 +2026-09-20 15:35:55 | INFO | train rank=0 | step=20500/50000 progress=41.0% loss=0.079783 lr=6.855986e-05 lr_backbone=6.855986e-05 elapsed=08:57:55 avg_step=1.57s recent_step=1.61s eta=12:54:05 +2026-09-20 15:36:13 | INFO | train rank=0 | step=20510/50000 progress=41.0% loss=0.143287 lr=6.852915e-05 lr_backbone=6.852915e-05 elapsed=08:58:13 avg_step=1.57s recent_step=1.79s eta=12:53:52 +2026-09-20 15:36:25 | INFO | train rank=0 | step=20520/50000 progress=41.0% loss=0.091883 lr=6.849843e-05 lr_backbone=6.849843e-05 elapsed=08:58:24 avg_step=1.57s recent_step=1.16s eta=12:53:30 +2026-09-20 15:36:42 | INFO | train rank=0 | step=20530/50000 progress=41.1% loss=0.084297 lr=6.846771e-05 lr_backbone=6.846771e-05 elapsed=08:58:41 avg_step=1.57s recent_step=1.69s eta=12:53:16 +2026-09-20 15:37:02 | INFO | train rank=0 | step=20540/50000 progress=41.1% loss=0.105616 lr=6.843697e-05 lr_backbone=6.843697e-05 elapsed=08:59:01 avg_step=1.57s recent_step=2.00s eta=12:53:07 +2026-09-20 15:37:14 | INFO | train rank=0 | step=20550/50000 progress=41.1% loss=0.088212 lr=6.840623e-05 lr_backbone=6.840623e-05 elapsed=08:59:13 avg_step=1.57s recent_step=1.18s eta=12:52:45 +2026-09-20 15:37:33 | INFO | train rank=0 | step=20560/50000 progress=41.1% loss=0.098881 lr=6.837548e-05 lr_backbone=6.837548e-05 elapsed=08:59:32 avg_step=1.57s recent_step=1.92s eta=12:52:34 +2026-09-20 15:37:47 | INFO | train rank=0 | step=20570/50000 progress=41.1% loss=0.100381 lr=6.834472e-05 lr_backbone=6.834472e-05 elapsed=08:59:46 avg_step=1.57s recent_step=1.40s eta=12:52:16 +2026-09-20 15:38:03 | INFO | train rank=0 | step=20580/50000 progress=41.2% loss=0.089250 lr=6.831395e-05 lr_backbone=6.831395e-05 elapsed=09:00:02 avg_step=1.57s recent_step=1.60s eta=12:52:01 +2026-09-20 15:38:15 | INFO | train rank=0 | step=20590/50000 progress=41.2% loss=0.083449 lr=6.828317e-05 lr_backbone=6.828317e-05 elapsed=09:00:15 avg_step=1.57s recent_step=1.22s eta=12:51:40 +2026-09-20 15:38:30 | INFO | train rank=0 | step=20600/50000 progress=41.2% loss=0.099961 lr=6.825239e-05 lr_backbone=6.825239e-05 elapsed=09:00:30 avg_step=1.57s recent_step=1.51s eta=12:51:23 +2026-09-20 15:38:48 | INFO | train rank=0 | step=20610/50000 progress=41.2% loss=0.139163 lr=6.822160e-05 lr_backbone=6.822160e-05 elapsed=09:00:48 avg_step=1.57s recent_step=1.83s eta=12:51:11 +2026-09-20 15:39:01 | INFO | train rank=0 | step=20620/50000 progress=41.2% loss=0.091699 lr=6.819080e-05 lr_backbone=6.819080e-05 elapsed=09:01:01 avg_step=1.57s recent_step=1.30s eta=12:50:52 +2026-09-20 15:39:16 | INFO | train rank=0 | step=20630/50000 progress=41.3% loss=0.092409 lr=6.815999e-05 lr_backbone=6.815999e-05 elapsed=09:01:16 avg_step=1.57s recent_step=1.47s eta=12:50:34 +2026-09-20 15:39:34 | INFO | train rank=0 | step=20640/50000 progress=41.3% loss=0.105784 lr=6.812918e-05 lr_backbone=6.812918e-05 elapsed=09:01:34 avg_step=1.57s recent_step=1.79s eta=12:50:22 +2026-09-20 15:39:51 | INFO | train rank=0 | step=20650/50000 progress=41.3% loss=0.150739 lr=6.809836e-05 lr_backbone=6.809836e-05 elapsed=09:01:50 avg_step=1.57s recent_step=1.68s eta=12:50:08 +2026-09-20 15:40:02 | INFO | train rank=0 | step=20660/50000 progress=41.3% loss=0.079581 lr=6.806752e-05 lr_backbone=6.806752e-05 elapsed=09:02:02 avg_step=1.57s recent_step=1.13s eta=12:49:45 +2026-09-20 15:40:17 | INFO | train rank=0 | step=20670/50000 progress=41.3% loss=0.115183 lr=6.803669e-05 lr_backbone=6.803669e-05 elapsed=09:02:17 avg_step=1.57s recent_step=1.48s eta=12:49:28 +2026-09-20 15:40:33 | INFO | train rank=0 | step=20680/50000 progress=41.4% loss=0.069331 lr=6.800584e-05 lr_backbone=6.800584e-05 elapsed=09:02:33 avg_step=1.57s recent_step=1.64s eta=12:49:14 +2026-09-20 15:40:46 | INFO | train rank=0 | step=20690/50000 progress=41.4% loss=0.102944 lr=6.797498e-05 lr_backbone=6.797498e-05 elapsed=09:02:46 avg_step=1.57s recent_step=1.28s eta=12:48:54 +2026-09-20 15:41:07 | INFO | train rank=0 | step=20700/50000 progress=41.4% loss=0.103455 lr=6.794412e-05 lr_backbone=6.794412e-05 elapsed=09:03:06 avg_step=1.57s recent_step=2.05s eta=12:48:45 +2026-09-20 15:41:18 | INFO | train rank=0 | step=20710/50000 progress=41.4% loss=0.109210 lr=6.791325e-05 lr_backbone=6.791325e-05 elapsed=09:03:18 avg_step=1.57s recent_step=1.14s eta=12:48:23 +2026-09-20 15:41:33 | INFO | train rank=0 | step=20720/50000 progress=41.4% loss=0.085862 lr=6.788237e-05 lr_backbone=6.788237e-05 elapsed=09:03:33 avg_step=1.57s recent_step=1.49s eta=12:48:06 +2026-09-20 15:41:51 | INFO | train rank=0 | step=20730/50000 progress=41.5% loss=0.129036 lr=6.785149e-05 lr_backbone=6.785149e-05 elapsed=09:03:51 avg_step=1.57s recent_step=1.83s eta=12:47:54 +2026-09-20 15:42:06 | INFO | train rank=0 | step=20740/50000 progress=41.5% loss=0.110017 lr=6.782059e-05 lr_backbone=6.782059e-05 elapsed=09:04:06 avg_step=1.57s recent_step=1.50s eta=12:47:37 +2026-09-20 15:42:23 | INFO | train rank=0 | step=20750/50000 progress=41.5% loss=0.111395 lr=6.778969e-05 lr_backbone=6.778969e-05 elapsed=09:04:22 avg_step=1.57s recent_step=1.63s eta=12:47:22 +2026-09-20 15:42:41 | INFO | train rank=0 | step=20760/50000 progress=41.5% loss=0.106603 lr=6.775878e-05 lr_backbone=6.775878e-05 elapsed=09:04:41 avg_step=1.57s recent_step=1.85s eta=12:47:10 +2026-09-20 15:42:55 | INFO | train rank=0 | step=20770/50000 progress=41.5% loss=0.090277 lr=6.772787e-05 lr_backbone=6.772787e-05 elapsed=09:04:54 avg_step=1.57s recent_step=1.36s eta=12:46:52 +2026-09-20 15:43:12 | INFO | train rank=0 | step=20780/50000 progress=41.6% loss=0.077284 lr=6.769694e-05 lr_backbone=6.769694e-05 elapsed=09:05:12 avg_step=1.57s recent_step=1.73s eta=12:46:38 +2026-09-20 15:43:27 | INFO | train rank=0 | step=20790/50000 progress=41.6% loss=0.094105 lr=6.766601e-05 lr_backbone=6.766601e-05 elapsed=09:05:26 avg_step=1.57s recent_step=1.47s eta=12:46:21 +2026-09-20 15:43:46 | INFO | train rank=0 | step=20800/50000 progress=41.6% loss=0.084963 lr=6.763507e-05 lr_backbone=6.763507e-05 elapsed=09:05:45 avg_step=1.57s recent_step=1.91s eta=12:46:10 +2026-09-20 15:43:57 | INFO | train rank=0 | step=20810/50000 progress=41.6% loss=0.100877 lr=6.760412e-05 lr_backbone=6.760412e-05 elapsed=09:05:57 avg_step=1.57s recent_step=1.17s eta=12:45:48 +2026-09-20 15:44:19 | INFO | train rank=0 | step=20820/50000 progress=41.6% loss=0.090574 lr=6.757316e-05 lr_backbone=6.757316e-05 elapsed=09:06:18 avg_step=1.57s recent_step=2.10s eta=12:45:40 +2026-09-20 15:44:30 | INFO | train rank=0 | step=20830/50000 progress=41.7% loss=0.114034 lr=6.754220e-05 lr_backbone=6.754220e-05 elapsed=09:06:30 avg_step=1.57s recent_step=1.15s eta=12:45:18 +2026-09-20 15:44:42 | INFO | train rank=0 | step=20840/50000 progress=41.7% loss=0.086104 lr=6.751123e-05 lr_backbone=6.751123e-05 elapsed=09:06:42 avg_step=1.57s recent_step=1.20s eta=12:44:57 +2026-09-20 15:45:00 | INFO | train rank=0 | step=20850/50000 progress=41.7% loss=0.088064 lr=6.748025e-05 lr_backbone=6.748025e-05 elapsed=09:07:00 avg_step=1.57s recent_step=1.83s eta=12:44:45 +2026-09-20 15:45:19 | INFO | train rank=0 | step=20860/50000 progress=41.7% loss=0.085989 lr=6.744926e-05 lr_backbone=6.744926e-05 elapsed=09:07:18 avg_step=1.57s recent_step=1.82s eta=12:44:33 +2026-09-20 15:45:30 | INFO | train rank=0 | step=20870/50000 progress=41.7% loss=0.085842 lr=6.741827e-05 lr_backbone=6.741827e-05 elapsed=09:07:30 avg_step=1.57s recent_step=1.19s eta=12:44:12 +2026-09-20 15:45:45 | INFO | train rank=0 | step=20880/50000 progress=41.8% loss=0.084049 lr=6.738727e-05 lr_backbone=6.738727e-05 elapsed=09:07:45 avg_step=1.57s recent_step=1.49s eta=12:43:55 +2026-09-20 15:46:01 | INFO | train rank=0 | step=20890/50000 progress=41.8% loss=0.087404 lr=6.735626e-05 lr_backbone=6.735626e-05 elapsed=09:08:01 avg_step=1.57s recent_step=1.56s eta=12:43:39 +2026-09-20 15:46:16 | INFO | train rank=0 | step=20900/50000 progress=41.8% loss=0.097248 lr=6.732524e-05 lr_backbone=6.732524e-05 elapsed=09:08:15 avg_step=1.57s recent_step=1.49s eta=12:43:22 +2026-09-20 15:46:31 | INFO | train rank=0 | step=20910/50000 progress=41.8% loss=0.123375 lr=6.729422e-05 lr_backbone=6.729422e-05 elapsed=09:08:30 avg_step=1.57s recent_step=1.50s eta=12:43:05 +2026-09-20 15:46:49 | INFO | train rank=0 | step=20920/50000 progress=41.8% loss=0.087907 lr=6.726319e-05 lr_backbone=6.726319e-05 elapsed=09:08:49 avg_step=1.57s recent_step=1.86s eta=12:42:54 +2026-09-20 15:47:04 | INFO | train rank=0 | step=20930/50000 progress=41.9% loss=0.104822 lr=6.723215e-05 lr_backbone=6.723215e-05 elapsed=09:09:04 avg_step=1.57s recent_step=1.48s eta=12:42:36 +2026-09-20 15:47:23 | INFO | train rank=0 | step=20940/50000 progress=41.9% loss=0.084385 lr=6.720110e-05 lr_backbone=6.720110e-05 elapsed=09:09:22 avg_step=1.57s recent_step=1.83s eta=12:42:24 +2026-09-20 15:47:37 | INFO | train rank=0 | step=20950/50000 progress=41.9% loss=0.097801 lr=6.717004e-05 lr_backbone=6.717004e-05 elapsed=09:09:36 avg_step=1.57s recent_step=1.42s eta=12:42:06 +2026-09-20 15:47:49 | INFO | train rank=0 | step=20960/50000 progress=41.9% loss=0.110897 lr=6.713898e-05 lr_backbone=6.713898e-05 elapsed=09:09:48 avg_step=1.57s recent_step=1.20s eta=12:41:45 +2026-09-20 15:48:06 | INFO | train rank=0 | step=20970/50000 progress=41.9% loss=0.102549 lr=6.710791e-05 lr_backbone=6.710791e-05 elapsed=09:10:05 avg_step=1.57s recent_step=1.70s eta=12:41:31 +2026-09-20 15:48:21 | INFO | train rank=0 | step=20980/50000 progress=42.0% loss=0.074410 lr=6.707684e-05 lr_backbone=6.707684e-05 elapsed=09:10:21 avg_step=1.57s recent_step=1.56s eta=12:41:15 +2026-09-20 15:48:39 | INFO | train rank=0 | step=20990/50000 progress=42.0% loss=0.133771 lr=6.704575e-05 lr_backbone=6.704575e-05 elapsed=09:10:38 avg_step=1.57s recent_step=1.72s eta=12:41:02 +2026-09-20 15:48:57 | INFO | train rank=0 | step=21000/50000 progress=42.0% loss=0.091072 lr=6.701466e-05 lr_backbone=6.701466e-05 elapsed=09:10:56 avg_step=1.57s recent_step=1.82s eta=12:40:49 +2026-09-20 15:49:09 | INFO | train rank=0 | step=21010/50000 progress=42.0% loss=0.103355 lr=6.698356e-05 lr_backbone=6.698356e-05 elapsed=09:11:08 avg_step=1.57s recent_step=1.19s eta=12:40:28 +2026-09-20 15:49:21 | INFO | train rank=0 | step=21020/50000 progress=42.0% loss=0.144812 lr=6.695245e-05 lr_backbone=6.695245e-05 elapsed=09:11:20 avg_step=1.57s recent_step=1.20s eta=12:40:07 +2026-09-20 15:49:40 | INFO | train rank=0 | step=21030/50000 progress=42.1% loss=0.079770 lr=6.692134e-05 lr_backbone=6.692134e-05 elapsed=09:11:39 avg_step=1.57s recent_step=1.90s eta=12:39:56 +2026-09-20 15:50:00 | INFO | train rank=0 | step=21040/50000 progress=42.1% loss=0.097611 lr=6.689022e-05 lr_backbone=6.689022e-05 elapsed=09:12:00 avg_step=1.57s recent_step=2.04s eta=12:39:47 +2026-09-20 15:50:12 | INFO | train rank=0 | step=21050/50000 progress=42.1% loss=0.089320 lr=6.685909e-05 lr_backbone=6.685909e-05 elapsed=09:12:11 avg_step=1.57s recent_step=1.15s eta=12:39:25 +2026-09-20 15:50:31 | INFO | train rank=0 | step=21060/50000 progress=42.1% loss=0.102544 lr=6.682795e-05 lr_backbone=6.682795e-05 elapsed=09:12:31 avg_step=1.57s recent_step=1.95s eta=12:39:15 +2026-09-20 15:50:44 | INFO | train rank=0 | step=21070/50000 progress=42.1% loss=0.117859 lr=6.679681e-05 lr_backbone=6.679681e-05 elapsed=09:12:44 avg_step=1.57s recent_step=1.29s eta=12:38:55 +2026-09-20 15:50:56 | INFO | train rank=0 | step=21080/50000 progress=42.2% loss=0.091745 lr=6.676566e-05 lr_backbone=6.676566e-05 elapsed=09:12:55 avg_step=1.57s recent_step=1.18s eta=12:38:34 +2026-09-20 15:51:11 | INFO | train rank=0 | step=21090/50000 progress=42.2% loss=0.095369 lr=6.673450e-05 lr_backbone=6.673450e-05 elapsed=09:13:11 avg_step=1.57s recent_step=1.56s eta=12:38:18 +2026-09-20 15:51:30 | INFO | train rank=0 | step=21100/50000 progress=42.2% loss=0.123477 lr=6.670333e-05 lr_backbone=6.670333e-05 elapsed=09:13:30 avg_step=1.57s recent_step=1.88s eta=12:38:06 +2026-09-20 15:51:46 | INFO | train rank=0 | step=21110/50000 progress=42.2% loss=0.113940 lr=6.667216e-05 lr_backbone=6.667216e-05 elapsed=09:13:46 avg_step=1.57s recent_step=1.62s eta=12:37:51 +2026-09-20 15:51:59 | INFO | train rank=0 | step=21120/50000 progress=42.2% loss=0.076266 lr=6.664098e-05 lr_backbone=6.664098e-05 elapsed=09:13:58 avg_step=1.57s recent_step=1.22s eta=12:37:31 +2026-09-20 15:52:15 | INFO | train rank=0 | step=21130/50000 progress=42.3% loss=0.087444 lr=6.660979e-05 lr_backbone=6.660979e-05 elapsed=09:14:14 avg_step=1.57s recent_step=1.60s eta=12:37:16 +2026-09-20 15:52:32 | INFO | train rank=0 | step=21140/50000 progress=42.3% loss=0.089117 lr=6.657859e-05 lr_backbone=6.657859e-05 elapsed=09:14:31 avg_step=1.57s recent_step=1.69s eta=12:37:01 +2026-09-20 15:52:43 | INFO | train rank=0 | step=21150/50000 progress=42.3% loss=0.067334 lr=6.654739e-05 lr_backbone=6.654739e-05 elapsed=09:14:43 avg_step=1.57s recent_step=1.17s eta=12:36:40 +2026-09-20 15:52:59 | INFO | train rank=0 | step=21160/50000 progress=42.3% loss=0.098511 lr=6.651618e-05 lr_backbone=6.651618e-05 elapsed=09:14:58 avg_step=1.57s recent_step=1.53s eta=12:36:24 +2026-09-20 15:53:13 | INFO | train rank=0 | step=21170/50000 progress=42.3% loss=0.110696 lr=6.648497e-05 lr_backbone=6.648497e-05 elapsed=09:15:13 avg_step=1.57s recent_step=1.49s eta=12:36:07 +2026-09-20 15:53:31 | INFO | train rank=0 | step=21180/50000 progress=42.4% loss=0.089376 lr=6.645374e-05 lr_backbone=6.645374e-05 elapsed=09:15:30 avg_step=1.57s recent_step=1.75s eta=12:35:54 +2026-09-20 15:53:43 | INFO | train rank=0 | step=21190/50000 progress=42.4% loss=0.091436 lr=6.642251e-05 lr_backbone=6.642251e-05 elapsed=09:15:42 avg_step=1.57s recent_step=1.17s eta=12:35:32 +2026-09-20 15:53:57 | INFO | train rank=0 | step=21200/50000 progress=42.4% loss=0.086982 lr=6.639127e-05 lr_backbone=6.639127e-05 elapsed=09:15:57 avg_step=1.57s recent_step=1.47s eta=12:35:15 +2026-09-20 15:54:13 | INFO | train rank=0 | step=21210/50000 progress=42.4% loss=0.118551 lr=6.636003e-05 lr_backbone=6.636003e-05 elapsed=09:16:13 avg_step=1.57s recent_step=1.60s eta=12:35:00 +2026-09-20 15:54:31 | INFO | train rank=0 | step=21220/50000 progress=42.4% loss=0.077668 lr=6.632877e-05 lr_backbone=6.632877e-05 elapsed=09:16:30 avg_step=1.57s recent_step=1.74s eta=12:34:46 +2026-09-20 15:54:46 | INFO | train rank=0 | step=21230/50000 progress=42.5% loss=0.086404 lr=6.629751e-05 lr_backbone=6.629751e-05 elapsed=09:16:46 avg_step=1.57s recent_step=1.53s eta=12:34:30 +2026-09-20 15:55:09 | INFO | train rank=0 | step=21240/50000 progress=42.5% loss=0.109156 lr=6.626625e-05 lr_backbone=6.626625e-05 elapsed=09:17:08 avg_step=1.57s recent_step=2.27s eta=12:34:24 +2026-09-20 15:55:20 | INFO | train rank=0 | step=21250/50000 progress=42.5% loss=0.081401 lr=6.623497e-05 lr_backbone=6.623497e-05 elapsed=09:17:20 avg_step=1.57s recent_step=1.18s eta=12:34:02 +2026-09-20 15:56:17 | INFO | train rank=0 | step=21260/50000 progress=42.5% loss=0.158028 lr=6.620369e-05 lr_backbone=6.620369e-05 elapsed=09:18:16 avg_step=1.58s recent_step=5.65s eta=12:34:42 +2026-09-20 15:56:32 | INFO | train rank=0 | step=21270/50000 progress=42.5% loss=0.097144 lr=6.617240e-05 lr_backbone=6.617240e-05 elapsed=09:18:31 avg_step=1.58s recent_step=1.47s eta=12:34:25 +2026-09-20 15:56:46 | INFO | train rank=0 | step=21280/50000 progress=42.6% loss=0.104120 lr=6.614111e-05 lr_backbone=6.614111e-05 elapsed=09:18:46 avg_step=1.58s recent_step=1.46s eta=12:34:07 +2026-09-20 15:57:02 | INFO | train rank=0 | step=21290/50000 progress=42.6% loss=0.091025 lr=6.610981e-05 lr_backbone=6.610981e-05 elapsed=09:19:02 avg_step=1.58s recent_step=1.62s eta=12:33:52 +2026-09-20 15:57:16 | INFO | train rank=0 | step=21300/50000 progress=42.6% loss=0.082093 lr=6.607850e-05 lr_backbone=6.607850e-05 elapsed=09:19:16 avg_step=1.58s recent_step=1.38s eta=12:33:34 +2026-09-20 15:57:31 | INFO | train rank=0 | step=21310/50000 progress=42.6% loss=0.100793 lr=6.604718e-05 lr_backbone=6.604718e-05 elapsed=09:19:31 avg_step=1.58s recent_step=1.51s eta=12:33:17 +2026-09-20 15:57:52 | INFO | train rank=0 | step=21320/50000 progress=42.6% loss=0.094680 lr=6.601586e-05 lr_backbone=6.601586e-05 elapsed=09:19:52 avg_step=1.58s recent_step=2.11s eta=12:33:09 +2026-09-20 15:58:04 | INFO | train rank=0 | step=21330/50000 progress=42.7% loss=0.117212 lr=6.598453e-05 lr_backbone=6.598453e-05 elapsed=09:20:03 avg_step=1.58s recent_step=1.14s eta=12:32:47 +2026-09-20 15:58:16 | INFO | train rank=0 | step=21340/50000 progress=42.7% loss=0.095939 lr=6.595319e-05 lr_backbone=6.595319e-05 elapsed=09:20:15 avg_step=1.58s recent_step=1.21s eta=12:32:26 +2026-09-20 15:58:32 | INFO | train rank=0 | step=21350/50000 progress=42.7% loss=0.107174 lr=6.592184e-05 lr_backbone=6.592184e-05 elapsed=09:20:31 avg_step=1.58s recent_step=1.58s eta=12:32:11 +2026-09-20 15:58:49 | INFO | train rank=0 | step=21360/50000 progress=42.7% loss=0.089896 lr=6.589049e-05 lr_backbone=6.589049e-05 elapsed=09:20:48 avg_step=1.58s recent_step=1.69s eta=12:31:56 +2026-09-20 15:59:01 | INFO | train rank=0 | step=21370/50000 progress=42.7% loss=0.095667 lr=6.585913e-05 lr_backbone=6.585913e-05 elapsed=09:21:00 avg_step=1.58s recent_step=1.23s eta=12:31:36 +2026-09-20 15:59:19 | INFO | train rank=0 | step=21380/50000 progress=42.8% loss=0.094283 lr=6.582777e-05 lr_backbone=6.582777e-05 elapsed=09:21:19 avg_step=1.58s recent_step=1.85s eta=12:31:24 +2026-09-20 15:59:31 | INFO | train rank=0 | step=21390/50000 progress=42.8% loss=0.104637 lr=6.579640e-05 lr_backbone=6.579640e-05 elapsed=09:21:31 avg_step=1.58s recent_step=1.20s eta=12:31:03 +2026-09-20 15:59:48 | INFO | train rank=0 | step=21400/50000 progress=42.8% loss=0.094653 lr=6.576502e-05 lr_backbone=6.576502e-05 elapsed=09:21:47 avg_step=1.58s recent_step=1.65s eta=12:30:48 +2026-09-20 16:00:07 | INFO | train rank=0 | step=21410/50000 progress=42.8% loss=0.091005 lr=6.573363e-05 lr_backbone=6.573363e-05 elapsed=09:22:06 avg_step=1.58s recent_step=1.88s eta=12:30:37 +2026-09-20 16:00:19 | INFO | train rank=0 | step=21420/50000 progress=42.8% loss=0.124180 lr=6.570224e-05 lr_backbone=6.570224e-05 elapsed=09:22:18 avg_step=1.58s recent_step=1.22s eta=12:30:16 +2026-09-20 16:00:38 | INFO | train rank=0 | step=21430/50000 progress=42.9% loss=0.155504 lr=6.567084e-05 lr_backbone=6.567084e-05 elapsed=09:22:38 avg_step=1.58s recent_step=1.93s eta=12:30:05 +2026-09-20 16:00:55 | INFO | train rank=0 | step=21440/50000 progress=42.9% loss=0.092880 lr=6.563943e-05 lr_backbone=6.563943e-05 elapsed=09:22:55 avg_step=1.58s recent_step=1.73s eta=12:29:52 +2026-09-20 16:01:10 | INFO | train rank=0 | step=21450/50000 progress=42.9% loss=0.086493 lr=6.560802e-05 lr_backbone=6.560802e-05 elapsed=09:23:09 avg_step=1.58s recent_step=1.43s eta=12:29:34 +2026-09-20 16:01:28 | INFO | train rank=0 | step=21460/50000 progress=42.9% loss=0.085210 lr=6.557660e-05 lr_backbone=6.557660e-05 elapsed=09:23:28 avg_step=1.58s recent_step=1.83s eta=12:29:21 +2026-09-20 16:01:42 | INFO | train rank=0 | step=21470/50000 progress=42.9% loss=0.098167 lr=6.554517e-05 lr_backbone=6.554517e-05 elapsed=09:23:41 avg_step=1.58s recent_step=1.35s eta=12:29:03 +2026-09-20 16:01:54 | INFO | train rank=0 | step=21480/50000 progress=43.0% loss=0.105358 lr=6.551374e-05 lr_backbone=6.551374e-05 elapsed=09:23:54 avg_step=1.58s recent_step=1.29s eta=12:28:43 +2026-09-20 16:02:09 | INFO | train rank=0 | step=21490/50000 progress=43.0% loss=0.147756 lr=6.548230e-05 lr_backbone=6.548230e-05 elapsed=09:24:09 avg_step=1.58s recent_step=1.48s eta=12:28:26 +2026-09-20 16:02:30 | INFO | train rank=0 | step=21500/50000 progress=43.0% loss=0.075232 lr=6.545085e-05 lr_backbone=6.545085e-05 elapsed=09:24:29 avg_step=1.58s recent_step=2.07s eta=12:28:17 +2026-09-20 16:02:42 | INFO | train rank=0 | step=21510/50000 progress=43.0% loss=0.090898 lr=6.541940e-05 lr_backbone=6.541940e-05 elapsed=09:24:42 avg_step=1.58s recent_step=1.22s eta=12:27:57 +2026-09-20 16:02:55 | INFO | train rank=0 | step=21520/50000 progress=43.0% loss=0.064820 lr=6.538793e-05 lr_backbone=6.538793e-05 elapsed=09:24:54 avg_step=1.58s recent_step=1.24s eta=12:27:36 +2026-09-20 16:03:06 | INFO | train rank=0 | step=21530/50000 progress=43.1% loss=0.114146 lr=6.535647e-05 lr_backbone=6.535647e-05 elapsed=09:25:06 avg_step=1.57s recent_step=1.19s eta=12:27:16 +2026-09-20 16:03:27 | INFO | train rank=0 | step=21540/50000 progress=43.1% loss=0.089376 lr=6.532499e-05 lr_backbone=6.532499e-05 elapsed=09:25:27 avg_step=1.58s recent_step=2.07s eta=12:27:06 +2026-09-20 16:03:39 | INFO | train rank=0 | step=21550/50000 progress=43.1% loss=0.089638 lr=6.529351e-05 lr_backbone=6.529351e-05 elapsed=09:25:39 avg_step=1.57s recent_step=1.20s eta=12:26:46 +2026-09-20 16:03:56 | INFO | train rank=0 | step=21560/50000 progress=43.1% loss=0.078912 lr=6.526202e-05 lr_backbone=6.526202e-05 elapsed=09:25:55 avg_step=1.57s recent_step=1.67s eta=12:26:31 +2026-09-20 16:04:14 | INFO | train rank=0 | step=21570/50000 progress=43.1% loss=0.088830 lr=6.523053e-05 lr_backbone=6.523053e-05 elapsed=09:26:14 avg_step=1.58s recent_step=1.84s eta=12:26:19 +2026-09-20 16:04:29 | INFO | train rank=0 | step=21580/50000 progress=43.2% loss=0.109932 lr=6.519903e-05 lr_backbone=6.519903e-05 elapsed=09:26:29 avg_step=1.58s recent_step=1.51s eta=12:26:02 +2026-09-20 16:04:48 | INFO | train rank=0 | step=21590/50000 progress=43.2% loss=0.099743 lr=6.516752e-05 lr_backbone=6.516752e-05 elapsed=09:26:48 avg_step=1.58s recent_step=1.89s eta=12:25:51 +2026-09-20 16:05:03 | INFO | train rank=0 | step=21600/50000 progress=43.2% loss=0.083109 lr=6.513601e-05 lr_backbone=6.513601e-05 elapsed=09:27:02 avg_step=1.58s recent_step=1.46s eta=12:25:33 +2026-09-20 16:05:22 | INFO | train rank=0 | step=21610/50000 progress=43.2% loss=0.081751 lr=6.510448e-05 lr_backbone=6.510448e-05 elapsed=09:27:21 avg_step=1.58s recent_step=1.87s eta=12:25:21 +2026-09-20 16:05:34 | INFO | train rank=0 | step=21620/50000 progress=43.2% loss=0.117216 lr=6.507296e-05 lr_backbone=6.507296e-05 elapsed=09:27:33 avg_step=1.58s recent_step=1.21s eta=12:25:01 +2026-09-20 16:05:49 | INFO | train rank=0 | step=21630/50000 progress=43.3% loss=0.073171 lr=6.504142e-05 lr_backbone=6.504142e-05 elapsed=09:27:49 avg_step=1.58s recent_step=1.55s eta=12:24:45 +2026-09-20 16:06:08 | INFO | train rank=0 | step=21640/50000 progress=43.3% loss=0.092246 lr=6.500988e-05 lr_backbone=6.500988e-05 elapsed=09:28:08 avg_step=1.58s recent_step=1.90s eta=12:24:33 +2026-09-20 16:06:20 | INFO | train rank=0 | step=21650/50000 progress=43.3% loss=0.107846 lr=6.497833e-05 lr_backbone=6.497833e-05 elapsed=09:28:20 avg_step=1.58s recent_step=1.22s eta=12:24:13 +2026-09-20 16:06:40 | INFO | train rank=0 | step=21660/50000 progress=43.3% loss=0.087842 lr=6.494678e-05 lr_backbone=6.494678e-05 elapsed=09:28:39 avg_step=1.58s recent_step=1.92s eta=12:24:02 +2026-09-20 16:06:52 | INFO | train rank=0 | step=21670/50000 progress=43.3% loss=0.086248 lr=6.491522e-05 lr_backbone=6.491522e-05 elapsed=09:28:51 avg_step=1.58s recent_step=1.19s eta=12:23:41 +2026-09-20 16:07:09 | INFO | train rank=0 | step=21680/50000 progress=43.4% loss=0.086105 lr=6.488365e-05 lr_backbone=6.488365e-05 elapsed=09:29:08 avg_step=1.58s recent_step=1.73s eta=12:23:27 +2026-09-20 16:07:21 | INFO | train rank=0 | step=21690/50000 progress=43.4% loss=0.097017 lr=6.485208e-05 lr_backbone=6.485208e-05 elapsed=09:29:20 avg_step=1.57s recent_step=1.19s eta=12:23:07 +2026-09-20 16:07:38 | INFO | train rank=0 | step=21700/50000 progress=43.4% loss=0.110638 lr=6.482050e-05 lr_backbone=6.482050e-05 elapsed=09:29:38 avg_step=1.58s recent_step=1.75s eta=12:22:53 +2026-09-20 16:07:54 | INFO | train rank=0 | step=21710/50000 progress=43.4% loss=0.122546 lr=6.478891e-05 lr_backbone=6.478891e-05 elapsed=09:29:53 avg_step=1.58s recent_step=1.54s eta=12:22:37 +2026-09-20 16:08:11 | INFO | train rank=0 | step=21720/50000 progress=43.4% loss=0.083012 lr=6.475732e-05 lr_backbone=6.475732e-05 elapsed=09:30:11 avg_step=1.58s recent_step=1.76s eta=12:22:24 +2026-09-20 16:08:26 | INFO | train rank=0 | step=21730/50000 progress=43.5% loss=0.082863 lr=6.472572e-05 lr_backbone=6.472572e-05 elapsed=09:30:25 avg_step=1.58s recent_step=1.44s eta=12:22:06 +2026-09-20 16:08:41 | INFO | train rank=0 | step=21740/50000 progress=43.5% loss=0.105268 lr=6.469411e-05 lr_backbone=6.469411e-05 elapsed=09:30:40 avg_step=1.58s recent_step=1.50s eta=12:21:49 +2026-09-20 16:08:55 | INFO | train rank=0 | step=21750/50000 progress=43.5% loss=0.146281 lr=6.466250e-05 lr_backbone=6.466250e-05 elapsed=09:30:54 avg_step=1.57s recent_step=1.42s eta=12:21:32 +2026-09-20 16:09:13 | INFO | train rank=0 | step=21760/50000 progress=43.5% loss=0.084959 lr=6.463088e-05 lr_backbone=6.463088e-05 elapsed=09:31:13 avg_step=1.58s recent_step=1.85s eta=12:21:19 +2026-09-20 16:09:25 | INFO | train rank=0 | step=21770/50000 progress=43.5% loss=0.076854 lr=6.459926e-05 lr_backbone=6.459926e-05 elapsed=09:31:25 avg_step=1.57s recent_step=1.22s eta=12:20:59 +2026-09-20 16:09:43 | INFO | train rank=0 | step=21780/50000 progress=43.6% loss=0.122170 lr=6.456763e-05 lr_backbone=6.456763e-05 elapsed=09:31:42 avg_step=1.57s recent_step=1.73s eta=12:20:45 +2026-09-20 16:09:58 | INFO | train rank=0 | step=21790/50000 progress=43.6% loss=0.088075 lr=6.453599e-05 lr_backbone=6.453599e-05 elapsed=09:31:58 avg_step=1.57s recent_step=1.56s eta=12:20:29 +2026-09-20 16:10:19 | INFO | train rank=0 | step=21800/50000 progress=43.6% loss=0.107840 lr=6.450434e-05 lr_backbone=6.450434e-05 elapsed=09:32:18 avg_step=1.58s recent_step=2.03s eta=12:20:20 +2026-09-20 16:10:31 | INFO | train rank=0 | step=21810/50000 progress=43.6% loss=0.100062 lr=6.447269e-05 lr_backbone=6.447269e-05 elapsed=09:32:30 avg_step=1.58s recent_step=1.21s eta=12:19:59 +2026-09-20 16:10:43 | INFO | train rank=0 | step=21820/50000 progress=43.6% loss=0.082532 lr=6.444104e-05 lr_backbone=6.444104e-05 elapsed=09:32:42 avg_step=1.57s recent_step=1.20s eta=12:19:38 +2026-09-20 16:11:01 | INFO | train rank=0 | step=21830/50000 progress=43.7% loss=0.083730 lr=6.440937e-05 lr_backbone=6.440937e-05 elapsed=09:33:00 avg_step=1.57s recent_step=1.79s eta=12:19:25 +2026-09-20 16:11:17 | INFO | train rank=0 | step=21840/50000 progress=43.7% loss=0.066961 lr=6.437770e-05 lr_backbone=6.437770e-05 elapsed=09:33:17 avg_step=1.57s recent_step=1.65s eta=12:19:11 +2026-09-20 16:11:34 | INFO | train rank=0 | step=21850/50000 progress=43.7% loss=0.092535 lr=6.434603e-05 lr_backbone=6.434603e-05 elapsed=09:33:33 avg_step=1.57s recent_step=1.64s eta=12:18:56 +2026-09-20 16:11:49 | INFO | train rank=0 | step=21860/50000 progress=43.7% loss=0.075867 lr=6.431435e-05 lr_backbone=6.431435e-05 elapsed=09:33:49 avg_step=1.57s recent_step=1.55s eta=12:18:40 +2026-09-20 16:12:06 | INFO | train rank=0 | step=21870/50000 progress=43.7% loss=0.089682 lr=6.428266e-05 lr_backbone=6.428266e-05 elapsed=09:34:05 avg_step=1.58s recent_step=1.67s eta=12:18:25 +2026-09-20 16:12:24 | INFO | train rank=0 | step=21880/50000 progress=43.8% loss=0.112309 lr=6.425096e-05 lr_backbone=6.425096e-05 elapsed=09:34:24 avg_step=1.58s recent_step=1.87s eta=12:18:13 +2026-09-20 16:12:36 | INFO | train rank=0 | step=21890/50000 progress=43.8% loss=0.095829 lr=6.421926e-05 lr_backbone=6.421926e-05 elapsed=09:34:36 avg_step=1.57s recent_step=1.18s eta=12:17:52 +2026-09-20 16:12:51 | INFO | train rank=0 | step=21900/50000 progress=43.8% loss=0.080083 lr=6.418756e-05 lr_backbone=6.418756e-05 elapsed=09:34:51 avg_step=1.57s recent_step=1.48s eta=12:17:35 +2026-09-20 16:13:12 | INFO | train rank=0 | step=21910/50000 progress=43.8% loss=0.088061 lr=6.415584e-05 lr_backbone=6.415584e-05 elapsed=09:35:12 avg_step=1.58s recent_step=2.12s eta=12:17:27 +2026-09-20 16:13:27 | INFO | train rank=0 | step=21920/50000 progress=43.8% loss=0.110244 lr=6.412412e-05 lr_backbone=6.412412e-05 elapsed=09:35:27 avg_step=1.58s recent_step=1.46s eta=12:17:09 +2026-09-20 16:13:39 | INFO | train rank=0 | step=21930/50000 progress=43.9% loss=0.085936 lr=6.409240e-05 lr_backbone=6.409240e-05 elapsed=09:35:39 avg_step=1.57s recent_step=1.21s eta=12:16:49 +2026-09-20 16:13:56 | INFO | train rank=0 | step=21940/50000 progress=43.9% loss=0.111027 lr=6.406066e-05 lr_backbone=6.406066e-05 elapsed=09:35:56 avg_step=1.58s recent_step=1.73s eta=12:16:35 +2026-09-20 16:14:13 | INFO | train rank=0 | step=21950/50000 progress=43.9% loss=0.085254 lr=6.402893e-05 lr_backbone=6.402893e-05 elapsed=09:36:12 avg_step=1.58s recent_step=1.64s eta=12:16:20 +2026-09-20 16:14:29 | INFO | train rank=0 | step=21960/50000 progress=43.9% loss=0.105422 lr=6.399718e-05 lr_backbone=6.399718e-05 elapsed=09:36:29 avg_step=1.58s recent_step=1.67s eta=12:16:06 +2026-09-20 16:14:46 | INFO | train rank=0 | step=21970/50000 progress=43.9% loss=0.092676 lr=6.396543e-05 lr_backbone=6.396543e-05 elapsed=09:36:45 avg_step=1.58s recent_step=1.63s eta=12:15:51 +2026-09-20 16:14:59 | INFO | train rank=0 | step=21980/50000 progress=44.0% loss=0.084896 lr=6.393368e-05 lr_backbone=6.393368e-05 elapsed=09:36:59 avg_step=1.58s recent_step=1.32s eta=12:15:32 +2026-09-20 16:15:20 | INFO | train rank=0 | step=21990/50000 progress=44.0% loss=0.085320 lr=6.390191e-05 lr_backbone=6.390191e-05 elapsed=09:37:20 avg_step=1.58s recent_step=2.16s eta=12:15:23 +2026-09-20 16:15:32 | INFO | train rank=0 | step=22000/50000 progress=44.0% loss=0.099301 lr=6.387015e-05 lr_backbone=6.387015e-05 elapsed=09:37:32 avg_step=1.58s recent_step=1.17s eta=12:15:02 +2026-09-20 16:15:49 | INFO | train rank=0 | step=22010/50000 progress=44.0% loss=0.083981 lr=6.383837e-05 lr_backbone=6.383837e-05 elapsed=09:37:49 avg_step=1.58s recent_step=1.72s eta=12:14:48 +2026-09-20 16:16:01 | INFO | train rank=0 | step=22020/50000 progress=44.0% loss=0.083268 lr=6.380659e-05 lr_backbone=6.380659e-05 elapsed=09:38:01 avg_step=1.58s recent_step=1.21s eta=12:14:28 +2026-09-20 16:16:14 | INFO | train rank=0 | step=22030/50000 progress=44.1% loss=0.093554 lr=6.377480e-05 lr_backbone=6.377480e-05 elapsed=09:38:14 avg_step=1.57s recent_step=1.26s eta=12:14:08 +2026-09-20 16:16:26 | INFO | train rank=0 | step=22040/50000 progress=44.1% loss=0.091736 lr=6.374301e-05 lr_backbone=6.374301e-05 elapsed=09:38:26 avg_step=1.57s recent_step=1.23s eta=12:13:48 +2026-09-20 16:16:47 | INFO | train rank=0 | step=22050/50000 progress=44.1% loss=0.106271 lr=6.371121e-05 lr_backbone=6.371121e-05 elapsed=09:38:47 avg_step=1.57s recent_step=2.09s eta=12:13:39 +2026-09-20 16:17:01 | INFO | train rank=0 | step=22060/50000 progress=44.1% loss=0.086057 lr=6.367941e-05 lr_backbone=6.367941e-05 elapsed=09:39:00 avg_step=1.57s recent_step=1.34s eta=12:13:20 +2026-09-20 16:17:18 | INFO | train rank=0 | step=22070/50000 progress=44.1% loss=0.081238 lr=6.364760e-05 lr_backbone=6.364760e-05 elapsed=09:39:18 avg_step=1.57s recent_step=1.76s eta=12:13:07 +2026-09-20 16:17:30 | INFO | train rank=0 | step=22080/50000 progress=44.2% loss=0.094438 lr=6.361578e-05 lr_backbone=6.361578e-05 elapsed=09:39:30 avg_step=1.57s recent_step=1.20s eta=12:12:46 +2026-09-20 16:17:45 | INFO | train rank=0 | step=22090/50000 progress=44.2% loss=0.117412 lr=6.358396e-05 lr_backbone=6.358396e-05 elapsed=09:39:44 avg_step=1.57s recent_step=1.43s eta=12:12:29 +2026-09-20 16:18:03 | INFO | train rank=0 | step=22100/50000 progress=44.2% loss=0.099744 lr=6.355213e-05 lr_backbone=6.355213e-05 elapsed=09:40:02 avg_step=1.57s recent_step=1.81s eta=12:12:16 +2026-09-20 16:18:18 | INFO | train rank=0 | step=22110/50000 progress=44.2% loss=0.076333 lr=6.352029e-05 lr_backbone=6.352029e-05 elapsed=09:40:18 avg_step=1.57s recent_step=1.55s eta=12:12:00 +2026-09-20 16:18:36 | INFO | train rank=0 | step=22120/50000 progress=44.2% loss=0.097970 lr=6.348845e-05 lr_backbone=6.348845e-05 elapsed=09:40:35 avg_step=1.57s recent_step=1.76s eta=12:11:46 +2026-09-20 16:18:50 | INFO | train rank=0 | step=22130/50000 progress=44.3% loss=0.096077 lr=6.345661e-05 lr_backbone=6.345661e-05 elapsed=09:40:50 avg_step=1.57s recent_step=1.45s eta=12:11:29 +2026-09-20 16:19:07 | INFO | train rank=0 | step=22140/50000 progress=44.3% loss=0.134141 lr=6.342476e-05 lr_backbone=6.342476e-05 elapsed=09:41:06 avg_step=1.57s recent_step=1.65s eta=12:11:14 +2026-09-20 16:19:29 | INFO | train rank=0 | step=22150/50000 progress=44.3% loss=0.102520 lr=6.339290e-05 lr_backbone=6.339290e-05 elapsed=09:41:29 avg_step=1.58s recent_step=2.26s eta=12:11:07 +2026-09-20 16:19:45 | INFO | train rank=0 | step=22160/50000 progress=44.3% loss=0.125181 lr=6.336103e-05 lr_backbone=6.336103e-05 elapsed=09:41:45 avg_step=1.58s recent_step=1.61s eta=12:10:52 +2026-09-20 16:20:41 | INFO | train rank=0 | step=22170/50000 progress=44.3% loss=0.105878 lr=6.332916e-05 lr_backbone=6.332916e-05 elapsed=09:42:40 avg_step=1.58s recent_step=5.52s eta=12:11:26 +2026-09-20 16:20:52 | INFO | train rank=0 | step=22180/50000 progress=44.4% loss=0.098091 lr=6.329729e-05 lr_backbone=6.329729e-05 elapsed=09:42:52 avg_step=1.58s recent_step=1.19s eta=12:11:05 +2026-09-20 16:21:04 | INFO | train rank=0 | step=22190/50000 progress=44.4% loss=0.100445 lr=6.326541e-05 lr_backbone=6.326541e-05 elapsed=09:43:04 avg_step=1.58s recent_step=1.19s eta=12:10:44 +2026-09-20 16:21:19 | INFO | train rank=0 | step=22200/50000 progress=44.4% loss=0.079945 lr=6.323352e-05 lr_backbone=6.323352e-05 elapsed=09:43:19 avg_step=1.58s recent_step=1.50s eta=12:10:28 +2026-09-20 16:21:33 | INFO | train rank=0 | step=22210/50000 progress=44.4% loss=0.104671 lr=6.320163e-05 lr_backbone=6.320163e-05 elapsed=09:43:33 avg_step=1.58s recent_step=1.41s eta=12:10:10 +2026-09-20 16:21:53 | INFO | train rank=0 | step=22220/50000 progress=44.4% loss=0.089191 lr=6.316973e-05 lr_backbone=6.316973e-05 elapsed=09:43:53 avg_step=1.58s recent_step=1.98s eta=12:09:59 +2026-09-20 16:22:05 | INFO | train rank=0 | step=22230/50000 progress=44.5% loss=0.092857 lr=6.313782e-05 lr_backbone=6.313782e-05 elapsed=09:44:05 avg_step=1.58s recent_step=1.20s eta=12:09:39 +2026-09-20 16:22:17 | INFO | train rank=0 | step=22240/50000 progress=44.5% loss=0.107524 lr=6.310591e-05 lr_backbone=6.310591e-05 elapsed=09:44:17 avg_step=1.58s recent_step=1.21s eta=12:09:18 +2026-09-20 16:22:32 | INFO | train rank=0 | step=22250/50000 progress=44.5% loss=0.093611 lr=6.307400e-05 lr_backbone=6.307400e-05 elapsed=09:44:32 avg_step=1.58s recent_step=1.48s eta=12:09:01 +2026-09-20 16:22:49 | INFO | train rank=0 | step=22260/50000 progress=44.5% loss=0.118380 lr=6.304208e-05 lr_backbone=6.304208e-05 elapsed=09:44:48 avg_step=1.58s recent_step=1.64s eta=12:08:47 +2026-09-20 16:23:07 | INFO | train rank=0 | step=22270/50000 progress=44.5% loss=0.089580 lr=6.301015e-05 lr_backbone=6.301015e-05 elapsed=09:45:06 avg_step=1.58s recent_step=1.82s eta=12:08:34 +2026-09-20 16:23:21 | INFO | train rank=0 | step=22280/50000 progress=44.6% loss=0.092291 lr=6.297821e-05 lr_backbone=6.297821e-05 elapsed=09:45:20 avg_step=1.58s recent_step=1.40s eta=12:08:16 +2026-09-20 16:23:36 | INFO | train rank=0 | step=22290/50000 progress=44.6% loss=0.090158 lr=6.294628e-05 lr_backbone=6.294628e-05 elapsed=09:45:35 avg_step=1.58s recent_step=1.50s eta=12:07:59 +2026-09-20 16:23:49 | INFO | train rank=0 | step=22300/50000 progress=44.6% loss=0.122875 lr=6.291433e-05 lr_backbone=6.291433e-05 elapsed=09:45:49 avg_step=1.58s recent_step=1.33s eta=12:07:40 +2026-09-20 16:24:04 | INFO | train rank=0 | step=22310/50000 progress=44.6% loss=0.084440 lr=6.288238e-05 lr_backbone=6.288238e-05 elapsed=09:46:03 avg_step=1.58s recent_step=1.46s eta=12:07:23 +2026-09-20 16:24:24 | INFO | train rank=0 | step=22320/50000 progress=44.6% loss=0.094800 lr=6.285043e-05 lr_backbone=6.285043e-05 elapsed=09:46:24 avg_step=1.58s recent_step=2.04s eta=12:07:13 +2026-09-20 16:24:36 | INFO | train rank=0 | step=22330/50000 progress=44.7% loss=0.123620 lr=6.281846e-05 lr_backbone=6.281846e-05 elapsed=09:46:36 avg_step=1.58s recent_step=1.17s eta=12:06:52 +2026-09-20 16:24:56 | INFO | train rank=0 | step=22340/50000 progress=44.7% loss=0.104088 lr=6.278650e-05 lr_backbone=6.278650e-05 elapsed=09:46:56 avg_step=1.58s recent_step=2.03s eta=12:06:42 +2026-09-20 16:25:11 | INFO | train rank=0 | step=22350/50000 progress=44.7% loss=0.126695 lr=6.275452e-05 lr_backbone=6.275452e-05 elapsed=09:47:11 avg_step=1.58s recent_step=1.51s eta=12:06:26 +2026-09-20 16:25:26 | INFO | train rank=0 | step=22360/50000 progress=44.7% loss=0.098921 lr=6.272255e-05 lr_backbone=6.272255e-05 elapsed=09:47:26 avg_step=1.58s recent_step=1.48s eta=12:06:09 +2026-09-20 16:25:43 | INFO | train rank=0 | step=22370/50000 progress=44.7% loss=0.116657 lr=6.269056e-05 lr_backbone=6.269056e-05 elapsed=09:47:43 avg_step=1.58s recent_step=1.69s eta=12:05:54 +2026-09-20 16:26:04 | INFO | train rank=0 | step=22380/50000 progress=44.8% loss=0.127796 lr=6.265857e-05 lr_backbone=6.265857e-05 elapsed=09:48:04 avg_step=1.58s recent_step=2.10s eta=12:05:45 +2026-09-20 16:26:16 | INFO | train rank=0 | step=22390/50000 progress=44.8% loss=0.119239 lr=6.262658e-05 lr_backbone=6.262658e-05 elapsed=09:48:16 avg_step=1.58s recent_step=1.21s eta=12:05:25 +2026-09-20 16:26:36 | INFO | train rank=0 | step=22400/50000 progress=44.8% loss=0.089518 lr=6.259458e-05 lr_backbone=6.259458e-05 elapsed=09:48:35 avg_step=1.58s recent_step=1.96s eta=12:05:14 +2026-09-20 16:26:48 | INFO | train rank=0 | step=22410/50000 progress=44.8% loss=0.128181 lr=6.256257e-05 lr_backbone=6.256257e-05 elapsed=09:48:47 avg_step=1.58s recent_step=1.19s eta=12:04:53 +2026-09-20 16:27:05 | INFO | train rank=0 | step=22420/50000 progress=44.8% loss=0.124988 lr=6.253056e-05 lr_backbone=6.253056e-05 elapsed=09:49:04 avg_step=1.58s recent_step=1.70s eta=12:04:39 +2026-09-20 16:27:22 | INFO | train rank=0 | step=22430/50000 progress=44.9% loss=0.083337 lr=6.249854e-05 lr_backbone=6.249854e-05 elapsed=09:49:22 avg_step=1.58s recent_step=1.74s eta=12:04:25 +2026-09-20 16:27:36 | INFO | train rank=0 | step=22440/50000 progress=44.9% loss=0.132969 lr=6.246652e-05 lr_backbone=6.246652e-05 elapsed=09:49:36 avg_step=1.58s recent_step=1.40s eta=12:04:07 +2026-09-20 16:27:50 | INFO | train rank=0 | step=22450/50000 progress=44.9% loss=0.136700 lr=6.243449e-05 lr_backbone=6.243449e-05 elapsed=09:49:50 avg_step=1.58s recent_step=1.41s eta=12:03:50 +2026-09-20 16:28:07 | INFO | train rank=0 | step=22460/50000 progress=44.9% loss=0.111640 lr=6.240246e-05 lr_backbone=6.240246e-05 elapsed=09:50:07 avg_step=1.58s recent_step=1.69s eta=12:03:35 +2026-09-20 16:28:26 | INFO | train rank=0 | step=22470/50000 progress=44.9% loss=0.090719 lr=6.237042e-05 lr_backbone=6.237042e-05 elapsed=09:50:26 avg_step=1.58s recent_step=1.90s eta=12:03:23 +2026-09-20 16:28:44 | INFO | train rank=0 | step=22480/50000 progress=45.0% loss=0.093707 lr=6.233838e-05 lr_backbone=6.233838e-05 elapsed=09:50:43 avg_step=1.58s recent_step=1.75s eta=12:03:10 +2026-09-20 16:28:55 | INFO | train rank=0 | step=22490/50000 progress=45.0% loss=0.106554 lr=6.230633e-05 lr_backbone=6.230633e-05 elapsed=09:50:55 avg_step=1.58s recent_step=1.18s eta=12:02:49 +2026-09-20 16:29:10 | INFO | train rank=0 | step=22500/50000 progress=45.0% loss=0.094434 lr=6.227427e-05 lr_backbone=6.227427e-05 elapsed=09:51:09 avg_step=1.58s recent_step=1.42s eta=12:02:31 +2026-09-20 16:29:30 | INFO | train rank=0 | step=22510/50000 progress=45.0% loss=0.085361 lr=6.224221e-05 lr_backbone=6.224221e-05 elapsed=09:51:30 avg_step=1.58s recent_step=2.07s eta=12:02:22 +2026-09-20 16:29:42 | INFO | train rank=0 | step=22520/50000 progress=45.0% loss=0.102895 lr=6.221015e-05 lr_backbone=6.221015e-05 elapsed=09:51:42 avg_step=1.58s recent_step=1.17s eta=12:02:01 +2026-09-20 16:29:56 | INFO | train rank=0 | step=22530/50000 progress=45.1% loss=0.074023 lr=6.217808e-05 lr_backbone=6.217808e-05 elapsed=09:51:56 avg_step=1.58s recent_step=1.40s eta=12:01:43 +2026-09-20 16:30:15 | INFO | train rank=0 | step=22540/50000 progress=45.1% loss=0.070200 lr=6.214600e-05 lr_backbone=6.214600e-05 elapsed=09:52:15 avg_step=1.58s recent_step=1.89s eta=12:01:31 +2026-09-20 16:30:32 | INFO | train rank=0 | step=22550/50000 progress=45.1% loss=0.088649 lr=6.211392e-05 lr_backbone=6.211392e-05 elapsed=09:52:32 avg_step=1.58s recent_step=1.73s eta=12:01:17 +2026-09-20 16:30:48 | INFO | train rank=0 | step=22560/50000 progress=45.1% loss=0.103962 lr=6.208183e-05 lr_backbone=6.208183e-05 elapsed=09:52:48 avg_step=1.58s recent_step=1.57s eta=12:01:01 +2026-09-20 16:31:06 | INFO | train rank=0 | step=22570/50000 progress=45.1% loss=0.153968 lr=6.204974e-05 lr_backbone=6.204974e-05 elapsed=09:53:05 avg_step=1.58s recent_step=1.76s eta=12:00:48 +2026-09-20 16:31:21 | INFO | train rank=0 | step=22580/50000 progress=45.2% loss=0.098229 lr=6.201764e-05 lr_backbone=6.201764e-05 elapsed=09:53:20 avg_step=1.58s recent_step=1.51s eta=12:00:31 +2026-09-20 16:31:33 | INFO | train rank=0 | step=22590/50000 progress=45.2% loss=0.092174 lr=6.198554e-05 lr_backbone=6.198554e-05 elapsed=09:53:33 avg_step=1.58s recent_step=1.23s eta=12:00:11 +2026-09-20 16:31:52 | INFO | train rank=0 | step=22600/50000 progress=45.2% loss=0.091574 lr=6.195343e-05 lr_backbone=6.195343e-05 elapsed=09:53:52 avg_step=1.58s recent_step=1.96s eta=12:00:00 +2026-09-20 16:32:05 | INFO | train rank=0 | step=22610/50000 progress=45.2% loss=0.081859 lr=6.192132e-05 lr_backbone=6.192132e-05 elapsed=09:54:05 avg_step=1.58s recent_step=1.29s eta=11:59:41 +2026-09-20 16:32:22 | INFO | train rank=0 | step=22620/50000 progress=45.2% loss=0.094415 lr=6.188920e-05 lr_backbone=6.188920e-05 elapsed=09:54:21 avg_step=1.58s recent_step=1.62s eta=11:59:26 +2026-09-20 16:32:38 | INFO | train rank=0 | step=22630/50000 progress=45.3% loss=0.123266 lr=6.185708e-05 lr_backbone=6.185708e-05 elapsed=09:54:37 avg_step=1.58s recent_step=1.63s eta=11:59:10 +2026-09-20 16:32:53 | INFO | train rank=0 | step=22640/50000 progress=45.3% loss=0.113680 lr=6.182495e-05 lr_backbone=6.182495e-05 elapsed=09:54:52 avg_step=1.58s recent_step=1.50s eta=11:58:54 +2026-09-20 16:33:08 | INFO | train rank=0 | step=22650/50000 progress=45.3% loss=0.089896 lr=6.179282e-05 lr_backbone=6.179282e-05 elapsed=09:55:08 avg_step=1.58s recent_step=1.52s eta=11:58:37 +2026-09-20 16:33:23 | INFO | train rank=0 | step=22660/50000 progress=45.3% loss=0.089453 lr=6.176068e-05 lr_backbone=6.176068e-05 elapsed=09:55:23 avg_step=1.58s recent_step=1.50s eta=11:58:21 +2026-09-20 16:33:37 | INFO | train rank=0 | step=22670/50000 progress=45.3% loss=0.069256 lr=6.172853e-05 lr_backbone=6.172853e-05 elapsed=09:55:37 avg_step=1.58s recent_step=1.39s eta=11:58:03 +2026-09-20 16:33:59 | INFO | train rank=0 | step=22680/50000 progress=45.4% loss=0.115521 lr=6.169638e-05 lr_backbone=6.169638e-05 elapsed=09:55:59 avg_step=1.58s recent_step=2.24s eta=11:57:55 +2026-09-20 16:34:11 | INFO | train rank=0 | step=22690/50000 progress=45.4% loss=0.100787 lr=6.166423e-05 lr_backbone=6.166423e-05 elapsed=09:56:11 avg_step=1.58s recent_step=1.16s eta=11:57:34 +2026-09-20 16:34:29 | INFO | train rank=0 | step=22700/50000 progress=45.4% loss=0.097217 lr=6.163207e-05 lr_backbone=6.163207e-05 elapsed=09:56:29 avg_step=1.58s recent_step=1.81s eta=11:57:21 +2026-09-20 16:34:41 | INFO | train rank=0 | step=22710/50000 progress=45.4% loss=0.101777 lr=6.159991e-05 lr_backbone=6.159991e-05 elapsed=09:56:40 avg_step=1.58s recent_step=1.18s eta=11:57:01 +2026-09-20 16:34:56 | INFO | train rank=0 | step=22720/50000 progress=45.4% loss=0.141114 lr=6.156774e-05 lr_backbone=6.156774e-05 elapsed=09:56:56 avg_step=1.58s recent_step=1.51s eta=11:56:44 +2026-09-20 16:35:11 | INFO | train rank=0 | step=22730/50000 progress=45.5% loss=0.113736 lr=6.153556e-05 lr_backbone=6.153556e-05 elapsed=09:57:11 avg_step=1.58s recent_step=1.53s eta=11:56:28 +2026-09-20 16:35:30 | INFO | train rank=0 | step=22740/50000 progress=45.5% loss=0.092972 lr=6.150338e-05 lr_backbone=6.150338e-05 elapsed=09:57:30 avg_step=1.58s recent_step=1.93s eta=11:56:16 +2026-09-20 16:35:42 | INFO | train rank=0 | step=22750/50000 progress=45.5% loss=0.077075 lr=6.147120e-05 lr_backbone=6.147120e-05 elapsed=09:57:42 avg_step=1.58s recent_step=1.20s eta=11:55:56 +2026-09-20 16:35:57 | INFO | train rank=0 | step=22760/50000 progress=45.5% loss=0.079573 lr=6.143901e-05 lr_backbone=6.143901e-05 elapsed=09:57:57 avg_step=1.58s recent_step=1.47s eta=11:55:39 +2026-09-20 16:36:17 | INFO | train rank=0 | step=22770/50000 progress=45.5% loss=0.107127 lr=6.140681e-05 lr_backbone=6.140681e-05 elapsed=09:58:16 avg_step=1.58s recent_step=1.94s eta=11:55:27 +2026-09-20 16:36:30 | INFO | train rank=0 | step=22780/50000 progress=45.6% loss=0.131912 lr=6.137461e-05 lr_backbone=6.137461e-05 elapsed=09:58:29 avg_step=1.58s recent_step=1.31s eta=11:55:08 +2026-09-20 16:36:45 | INFO | train rank=0 | step=22790/50000 progress=45.6% loss=0.093253 lr=6.134241e-05 lr_backbone=6.134241e-05 elapsed=09:58:44 avg_step=1.58s recent_step=1.52s eta=11:54:52 +2026-09-20 16:37:04 | INFO | train rank=0 | step=22800/50000 progress=45.6% loss=0.178876 lr=6.131020e-05 lr_backbone=6.131020e-05 elapsed=09:59:03 avg_step=1.58s recent_step=1.89s eta=11:54:40 +2026-09-20 16:37:16 | INFO | train rank=0 | step=22810/50000 progress=45.6% loss=0.084212 lr=6.127798e-05 lr_backbone=6.127798e-05 elapsed=09:59:15 avg_step=1.58s recent_step=1.20s eta=11:54:20 +2026-09-20 16:37:30 | INFO | train rank=0 | step=22820/50000 progress=45.6% loss=0.165249 lr=6.124576e-05 lr_backbone=6.124576e-05 elapsed=09:59:29 avg_step=1.58s recent_step=1.38s eta=11:54:02 +2026-09-20 16:37:45 | INFO | train rank=0 | step=22830/50000 progress=45.7% loss=0.098041 lr=6.121354e-05 lr_backbone=6.121354e-05 elapsed=09:59:45 avg_step=1.58s recent_step=1.54s eta=11:53:45 +2026-09-20 16:38:07 | INFO | train rank=0 | step=22840/50000 progress=45.7% loss=0.105446 lr=6.118131e-05 lr_backbone=6.118131e-05 elapsed=10:00:06 avg_step=1.58s recent_step=2.17s eta=11:53:37 +2026-09-20 16:38:19 | INFO | train rank=0 | step=22850/50000 progress=45.7% loss=0.133644 lr=6.114907e-05 lr_backbone=6.114907e-05 elapsed=10:00:18 avg_step=1.58s recent_step=1.18s eta=11:53:16 +2026-09-20 16:38:35 | INFO | train rank=0 | step=22860/50000 progress=45.7% loss=0.122775 lr=6.111683e-05 lr_backbone=6.111683e-05 elapsed=10:00:35 avg_step=1.58s recent_step=1.65s eta=11:53:01 +2026-09-20 16:38:49 | INFO | train rank=0 | step=22870/50000 progress=45.7% loss=0.125683 lr=6.108459e-05 lr_backbone=6.108459e-05 elapsed=10:00:49 avg_step=1.58s recent_step=1.42s eta=11:52:44 +2026-09-20 16:39:09 | INFO | train rank=0 | step=22880/50000 progress=45.8% loss=0.080546 lr=6.105234e-05 lr_backbone=6.105234e-05 elapsed=10:01:09 avg_step=1.58s recent_step=2.01s eta=11:52:33 +2026-09-20 16:39:26 | INFO | train rank=0 | step=22890/50000 progress=45.8% loss=0.084961 lr=6.102009e-05 lr_backbone=6.102009e-05 elapsed=10:01:25 avg_step=1.58s recent_step=1.64s eta=11:52:18 +2026-09-20 16:39:40 | INFO | train rank=0 | step=22900/50000 progress=45.8% loss=0.090160 lr=6.098783e-05 lr_backbone=6.098783e-05 elapsed=10:01:39 avg_step=1.58s recent_step=1.41s eta=11:52:00 +2026-09-20 16:39:53 | INFO | train rank=0 | step=22910/50000 progress=45.8% loss=0.102923 lr=6.095557e-05 lr_backbone=6.095557e-05 elapsed=10:01:53 avg_step=1.58s recent_step=1.36s eta=11:51:42 +2026-09-20 16:40:09 | INFO | train rank=0 | step=22920/50000 progress=45.8% loss=0.086505 lr=6.092330e-05 lr_backbone=6.092330e-05 elapsed=10:02:09 avg_step=1.58s recent_step=1.58s eta=11:51:26 +2026-09-20 16:40:26 | INFO | train rank=0 | step=22930/50000 progress=45.9% loss=0.102306 lr=6.089102e-05 lr_backbone=6.089102e-05 elapsed=10:02:25 avg_step=1.58s recent_step=1.65s eta=11:51:11 +2026-09-20 16:40:38 | INFO | train rank=0 | step=22940/50000 progress=45.9% loss=0.078855 lr=6.085875e-05 lr_backbone=6.085875e-05 elapsed=10:02:38 avg_step=1.58s recent_step=1.23s eta=11:50:52 +2026-09-20 16:40:52 | INFO | train rank=0 | step=22950/50000 progress=45.9% loss=0.094810 lr=6.082646e-05 lr_backbone=6.082646e-05 elapsed=10:02:51 avg_step=1.58s recent_step=1.35s eta=11:50:33 +2026-09-20 16:41:10 | INFO | train rank=0 | step=22960/50000 progress=45.9% loss=0.110089 lr=6.079418e-05 lr_backbone=6.079418e-05 elapsed=10:03:09 avg_step=1.58s recent_step=1.82s eta=11:50:20 +2026-09-20 16:41:27 | INFO | train rank=0 | step=22970/50000 progress=45.9% loss=0.158369 lr=6.076189e-05 lr_backbone=6.076189e-05 elapsed=10:03:27 avg_step=1.58s recent_step=1.76s eta=11:50:07 +2026-09-20 16:41:39 | INFO | train rank=0 | step=22980/50000 progress=46.0% loss=0.109697 lr=6.072959e-05 lr_backbone=6.072959e-05 elapsed=10:03:39 avg_step=1.58s recent_step=1.17s eta=11:49:46 +2026-09-20 16:41:52 | INFO | train rank=0 | step=22990/50000 progress=46.0% loss=0.082315 lr=6.069729e-05 lr_backbone=6.069729e-05 elapsed=10:03:51 avg_step=1.58s recent_step=1.25s eta=11:49:27 +2026-09-20 16:42:05 | INFO | train rank=0 | step=23000/50000 progress=46.0% loss=0.086897 lr=6.066498e-05 lr_backbone=6.066498e-05 elapsed=10:04:04 avg_step=1.58s recent_step=1.32s eta=11:49:08 +2026-09-20 16:42:20 | INFO | train rank=0 | step=23010/50000 progress=46.0% loss=0.090076 lr=6.063267e-05 lr_backbone=6.063267e-05 elapsed=10:04:20 avg_step=1.58s recent_step=1.57s eta=11:48:52 +2026-09-20 16:42:36 | INFO | train rank=0 | step=23020/50000 progress=46.0% loss=0.094125 lr=6.060036e-05 lr_backbone=6.060036e-05 elapsed=10:04:35 avg_step=1.58s recent_step=1.53s eta=11:48:36 +2026-09-20 16:42:53 | INFO | train rank=0 | step=23030/50000 progress=46.1% loss=0.092784 lr=6.056804e-05 lr_backbone=6.056804e-05 elapsed=10:04:52 avg_step=1.58s recent_step=1.69s eta=11:48:21 +2026-09-20 16:43:12 | INFO | train rank=0 | step=23040/50000 progress=46.1% loss=0.112850 lr=6.053571e-05 lr_backbone=6.053571e-05 elapsed=10:05:12 avg_step=1.58s recent_step=1.98s eta=11:48:10 +2026-09-20 16:43:35 | INFO | train rank=0 | step=23050/50000 progress=46.1% loss=0.076861 lr=6.050338e-05 lr_backbone=6.050338e-05 elapsed=10:05:35 avg_step=1.58s recent_step=2.25s eta=11:48:02 +2026-09-20 16:43:54 | INFO | train rank=0 | step=23060/50000 progress=46.1% loss=0.086370 lr=6.047105e-05 lr_backbone=6.047105e-05 elapsed=10:05:54 avg_step=1.58s recent_step=1.95s eta=11:47:51 +2026-09-20 16:44:47 | INFO | train rank=0 | step=23070/50000 progress=46.1% loss=0.130709 lr=6.043871e-05 lr_backbone=6.043871e-05 elapsed=10:06:47 avg_step=1.58s recent_step=5.26s eta=11:48:18 +2026-09-20 16:44:59 | INFO | train rank=0 | step=23080/50000 progress=46.2% loss=0.152635 lr=6.040637e-05 lr_backbone=6.040637e-05 elapsed=10:06:59 avg_step=1.58s recent_step=1.23s eta=11:47:58 +2026-09-20 16:45:11 | INFO | train rank=0 | step=23090/50000 progress=46.2% loss=0.116373 lr=6.037402e-05 lr_backbone=6.037402e-05 elapsed=10:07:11 avg_step=1.58s recent_step=1.20s eta=11:47:38 +2026-09-20 16:45:24 | INFO | train rank=0 | step=23100/50000 progress=46.2% loss=0.115622 lr=6.034167e-05 lr_backbone=6.034167e-05 elapsed=10:07:24 avg_step=1.58s recent_step=1.25s eta=11:47:19 +2026-09-20 16:45:38 | INFO | train rank=0 | step=23110/50000 progress=46.2% loss=0.102483 lr=6.030931e-05 lr_backbone=6.030931e-05 elapsed=10:07:37 avg_step=1.58s recent_step=1.36s eta=11:47:00 +2026-09-20 16:45:53 | INFO | train rank=0 | step=23120/50000 progress=46.2% loss=0.089753 lr=6.027695e-05 lr_backbone=6.027695e-05 elapsed=10:07:53 avg_step=1.58s recent_step=1.56s eta=11:46:44 +2026-09-20 16:46:08 | INFO | train rank=0 | step=23130/50000 progress=46.3% loss=0.126361 lr=6.024458e-05 lr_backbone=6.024458e-05 elapsed=10:08:08 avg_step=1.58s recent_step=1.49s eta=11:46:28 +2026-09-20 16:46:23 | INFO | train rank=0 | step=23140/50000 progress=46.3% loss=0.093950 lr=6.021221e-05 lr_backbone=6.021221e-05 elapsed=10:08:23 avg_step=1.58s recent_step=1.50s eta=11:46:11 +2026-09-20 16:46:36 | INFO | train rank=0 | step=23150/50000 progress=46.3% loss=0.081576 lr=6.017984e-05 lr_backbone=6.017984e-05 elapsed=10:08:35 avg_step=1.58s recent_step=1.26s eta=11:45:51 +2026-09-20 16:46:49 | INFO | train rank=0 | step=23160/50000 progress=46.3% loss=0.110501 lr=6.014746e-05 lr_backbone=6.014746e-05 elapsed=10:08:49 avg_step=1.58s recent_step=1.38s eta=11:45:33 +2026-09-20 16:47:04 | INFO | train rank=0 | step=23170/50000 progress=46.3% loss=0.115626 lr=6.011508e-05 lr_backbone=6.011508e-05 elapsed=10:09:03 avg_step=1.58s recent_step=1.43s eta=11:45:16 +2026-09-20 16:47:24 | INFO | train rank=0 | step=23180/50000 progress=46.4% loss=0.089911 lr=6.008269e-05 lr_backbone=6.008269e-05 elapsed=10:09:23 avg_step=1.58s recent_step=1.99s eta=11:45:05 +2026-09-20 16:47:39 | INFO | train rank=0 | step=23190/50000 progress=46.4% loss=0.084367 lr=6.005030e-05 lr_backbone=6.005030e-05 elapsed=10:09:39 avg_step=1.58s recent_step=1.54s eta=11:44:49 +2026-09-20 16:47:56 | INFO | train rank=0 | step=23200/50000 progress=46.4% loss=0.120091 lr=6.001790e-05 lr_backbone=6.001790e-05 elapsed=10:09:56 avg_step=1.58s recent_step=1.74s eta=11:44:35 +2026-09-20 16:48:10 | INFO | train rank=0 | step=23210/50000 progress=46.4% loss=0.088865 lr=5.998550e-05 lr_backbone=5.998550e-05 elapsed=10:10:10 avg_step=1.58s recent_step=1.38s eta=11:44:17 +2026-09-20 16:48:27 | INFO | train rank=0 | step=23220/50000 progress=46.4% loss=0.098986 lr=5.995309e-05 lr_backbone=5.995309e-05 elapsed=10:10:27 avg_step=1.58s recent_step=1.68s eta=11:44:02 +2026-09-20 16:48:40 | INFO | train rank=0 | step=23230/50000 progress=46.5% loss=0.105402 lr=5.992068e-05 lr_backbone=5.992068e-05 elapsed=10:10:40 avg_step=1.58s recent_step=1.34s eta=11:43:44 +2026-09-20 16:48:59 | INFO | train rank=0 | step=23240/50000 progress=46.5% loss=0.094444 lr=5.988827e-05 lr_backbone=5.988827e-05 elapsed=10:10:58 avg_step=1.58s recent_step=1.81s eta=11:43:31 +2026-09-20 16:49:18 | INFO | train rank=0 | step=23250/50000 progress=46.5% loss=0.109702 lr=5.985585e-05 lr_backbone=5.985585e-05 elapsed=10:11:17 avg_step=1.58s recent_step=1.93s eta=11:43:19 +2026-09-20 16:49:34 | INFO | train rank=0 | step=23260/50000 progress=46.5% loss=0.088043 lr=5.982343e-05 lr_backbone=5.982343e-05 elapsed=10:11:33 avg_step=1.58s recent_step=1.58s eta=11:43:03 +2026-09-20 16:49:47 | INFO | train rank=0 | step=23270/50000 progress=46.5% loss=0.117600 lr=5.979100e-05 lr_backbone=5.979100e-05 elapsed=10:11:47 avg_step=1.58s recent_step=1.37s eta=11:42:45 +2026-09-20 16:50:06 | INFO | train rank=0 | step=23280/50000 progress=46.6% loss=0.087257 lr=5.975857e-05 lr_backbone=5.975857e-05 elapsed=10:12:05 avg_step=1.58s recent_step=1.83s eta=11:42:32 +2026-09-20 16:50:21 | INFO | train rank=0 | step=23290/50000 progress=46.6% loss=0.094017 lr=5.972613e-05 lr_backbone=5.972613e-05 elapsed=10:12:20 avg_step=1.58s recent_step=1.49s eta=11:42:15 +2026-09-20 16:50:39 | INFO | train rank=0 | step=23300/50000 progress=46.6% loss=0.115633 lr=5.969369e-05 lr_backbone=5.969369e-05 elapsed=10:12:38 avg_step=1.58s recent_step=1.83s eta=11:42:02 +2026-09-20 16:50:56 | INFO | train rank=0 | step=23310/50000 progress=46.6% loss=0.113701 lr=5.966125e-05 lr_backbone=5.966125e-05 elapsed=10:12:56 avg_step=1.58s recent_step=1.72s eta=11:41:48 +2026-09-20 16:51:08 | INFO | train rank=0 | step=23320/50000 progress=46.6% loss=0.091532 lr=5.962880e-05 lr_backbone=5.962880e-05 elapsed=10:13:08 avg_step=1.58s recent_step=1.23s eta=11:41:28 +2026-09-20 16:51:23 | INFO | train rank=0 | step=23330/50000 progress=46.7% loss=0.097766 lr=5.959635e-05 lr_backbone=5.959635e-05 elapsed=10:13:22 avg_step=1.58s recent_step=1.43s eta=11:41:11 +2026-09-20 16:51:36 | INFO | train rank=0 | step=23340/50000 progress=46.7% loss=0.102412 lr=5.956389e-05 lr_backbone=5.956389e-05 elapsed=10:13:36 avg_step=1.58s recent_step=1.33s eta=11:40:52 +2026-09-20 16:51:54 | INFO | train rank=0 | step=23350/50000 progress=46.7% loss=0.072925 lr=5.953143e-05 lr_backbone=5.953143e-05 elapsed=10:13:53 avg_step=1.58s recent_step=1.79s eta=11:40:39 +2026-09-20 16:52:06 | INFO | train rank=0 | step=23360/50000 progress=46.7% loss=0.089631 lr=5.949897e-05 lr_backbone=5.949897e-05 elapsed=10:14:06 avg_step=1.58s recent_step=1.22s eta=11:40:19 +2026-09-20 16:52:23 | INFO | train rank=0 | step=23370/50000 progress=46.7% loss=0.105357 lr=5.946650e-05 lr_backbone=5.946650e-05 elapsed=10:14:23 avg_step=1.58s recent_step=1.73s eta=11:40:05 +2026-09-20 16:52:42 | INFO | train rank=0 | step=23380/50000 progress=46.8% loss=0.105266 lr=5.943402e-05 lr_backbone=5.943402e-05 elapsed=10:14:42 avg_step=1.58s recent_step=1.91s eta=11:39:53 +2026-09-20 16:52:58 | INFO | train rank=0 | step=23390/50000 progress=46.8% loss=0.096833 lr=5.940155e-05 lr_backbone=5.940155e-05 elapsed=10:14:58 avg_step=1.58s recent_step=1.59s eta=11:39:37 +2026-09-20 16:53:17 | INFO | train rank=0 | step=23400/50000 progress=46.8% loss=0.086994 lr=5.936907e-05 lr_backbone=5.936907e-05 elapsed=10:15:16 avg_step=1.58s recent_step=1.84s eta=11:39:25 +2026-09-20 16:53:29 | INFO | train rank=0 | step=23410/50000 progress=46.8% loss=0.087400 lr=5.933658e-05 lr_backbone=5.933658e-05 elapsed=10:15:28 avg_step=1.58s recent_step=1.19s eta=11:39:05 +2026-09-20 16:53:40 | INFO | train rank=0 | step=23420/50000 progress=46.8% loss=0.099233 lr=5.930409e-05 lr_backbone=5.930409e-05 elapsed=10:15:40 avg_step=1.58s recent_step=1.17s eta=11:38:44 +2026-09-20 16:54:00 | INFO | train rank=0 | step=23430/50000 progress=46.9% loss=0.107752 lr=5.927160e-05 lr_backbone=5.927160e-05 elapsed=10:16:00 avg_step=1.58s recent_step=2.02s eta=11:38:33 +2026-09-20 16:54:14 | INFO | train rank=0 | step=23440/50000 progress=46.9% loss=0.095655 lr=5.923910e-05 lr_backbone=5.923910e-05 elapsed=10:16:13 avg_step=1.58s recent_step=1.32s eta=11:38:15 +2026-09-20 16:54:30 | INFO | train rank=0 | step=23450/50000 progress=46.9% loss=0.109864 lr=5.920660e-05 lr_backbone=5.920660e-05 elapsed=10:16:30 avg_step=1.58s recent_step=1.64s eta=11:38:00 +2026-09-20 16:54:46 | INFO | train rank=0 | step=23460/50000 progress=46.9% loss=0.113006 lr=5.917409e-05 lr_backbone=5.917409e-05 elapsed=10:16:46 avg_step=1.58s recent_step=1.60s eta=11:37:44 +2026-09-20 16:55:01 | INFO | train rank=0 | step=23470/50000 progress=46.9% loss=0.096787 lr=5.914158e-05 lr_backbone=5.914158e-05 elapsed=10:17:00 avg_step=1.58s recent_step=1.47s eta=11:37:27 +2026-09-20 16:55:18 | INFO | train rank=0 | step=23480/50000 progress=47.0% loss=0.072124 lr=5.910907e-05 lr_backbone=5.910907e-05 elapsed=10:17:18 avg_step=1.58s recent_step=1.76s eta=11:37:13 +2026-09-20 16:55:34 | INFO | train rank=0 | step=23490/50000 progress=47.0% loss=0.097954 lr=5.907655e-05 lr_backbone=5.907655e-05 elapsed=10:17:34 avg_step=1.58s recent_step=1.56s eta=11:36:57 +2026-09-20 16:55:48 | INFO | train rank=0 | step=23500/50000 progress=47.0% loss=0.072144 lr=5.904403e-05 lr_backbone=5.904403e-05 elapsed=10:17:48 avg_step=1.58s recent_step=1.40s eta=11:36:40 +2026-09-20 16:56:06 | INFO | train rank=0 | step=23510/50000 progress=47.0% loss=0.131876 lr=5.901150e-05 lr_backbone=5.901150e-05 elapsed=10:18:06 avg_step=1.58s recent_step=1.83s eta=11:36:27 +2026-09-20 16:56:18 | INFO | train rank=0 | step=23520/50000 progress=47.0% loss=0.085120 lr=5.897897e-05 lr_backbone=5.897897e-05 elapsed=10:18:18 avg_step=1.58s recent_step=1.19s eta=11:36:07 +2026-09-20 16:56:36 | INFO | train rank=0 | step=23530/50000 progress=47.1% loss=0.089591 lr=5.894644e-05 lr_backbone=5.894644e-05 elapsed=10:18:35 avg_step=1.58s recent_step=1.75s eta=11:35:53 +2026-09-20 16:56:50 | INFO | train rank=0 | step=23540/50000 progress=47.1% loss=0.084309 lr=5.891390e-05 lr_backbone=5.891390e-05 elapsed=10:18:50 avg_step=1.58s recent_step=1.46s eta=11:35:36 +2026-09-20 16:57:05 | INFO | train rank=0 | step=23550/50000 progress=47.1% loss=0.083397 lr=5.888136e-05 lr_backbone=5.888136e-05 elapsed=10:19:05 avg_step=1.58s recent_step=1.52s eta=11:35:19 +2026-09-20 16:57:22 | INFO | train rank=0 | step=23560/50000 progress=47.1% loss=0.121471 lr=5.884881e-05 lr_backbone=5.884881e-05 elapsed=10:19:22 avg_step=1.58s recent_step=1.70s eta=11:35:05 +2026-09-20 16:57:42 | INFO | train rank=0 | step=23570/50000 progress=47.1% loss=0.077062 lr=5.881626e-05 lr_backbone=5.881626e-05 elapsed=10:19:42 avg_step=1.58s recent_step=1.96s eta=11:34:53 +2026-09-20 16:57:54 | INFO | train rank=0 | step=23580/50000 progress=47.2% loss=0.110318 lr=5.878371e-05 lr_backbone=5.878371e-05 elapsed=10:19:54 avg_step=1.58s recent_step=1.22s eta=11:34:34 +2026-09-20 16:58:07 | INFO | train rank=0 | step=23590/50000 progress=47.2% loss=0.086538 lr=5.875115e-05 lr_backbone=5.875115e-05 elapsed=10:20:07 avg_step=1.58s recent_step=1.27s eta=11:34:14 +2026-09-20 16:58:23 | INFO | train rank=0 | step=23600/50000 progress=47.2% loss=0.078101 lr=5.871859e-05 lr_backbone=5.871859e-05 elapsed=10:20:23 avg_step=1.58s recent_step=1.62s eta=11:33:59 +2026-09-20 16:58:35 | INFO | train rank=0 | step=23610/50000 progress=47.2% loss=0.085631 lr=5.868603e-05 lr_backbone=5.868603e-05 elapsed=10:20:35 avg_step=1.58s recent_step=1.21s eta=11:33:39 +2026-09-20 16:58:58 | INFO | train rank=0 | step=23620/50000 progress=47.2% loss=0.107552 lr=5.865346e-05 lr_backbone=5.865346e-05 elapsed=10:20:57 avg_step=1.58s recent_step=2.26s eta=11:33:31 +2026-09-20 16:59:14 | INFO | train rank=0 | step=23630/50000 progress=47.3% loss=0.113959 lr=5.862089e-05 lr_backbone=5.862089e-05 elapsed=10:21:14 avg_step=1.58s recent_step=1.62s eta=11:33:16 +2026-09-20 16:59:27 | INFO | train rank=0 | step=23640/50000 progress=47.3% loss=0.103019 lr=5.858831e-05 lr_backbone=5.858831e-05 elapsed=10:21:27 avg_step=1.58s recent_step=1.32s eta=11:32:57 +2026-09-20 16:59:45 | INFO | train rank=0 | step=23650/50000 progress=47.3% loss=0.084976 lr=5.855573e-05 lr_backbone=5.855573e-05 elapsed=10:21:45 avg_step=1.58s recent_step=1.81s eta=11:32:44 +2026-09-20 17:00:02 | INFO | train rank=0 | step=23660/50000 progress=47.3% loss=0.100734 lr=5.852315e-05 lr_backbone=5.852315e-05 elapsed=10:22:02 avg_step=1.58s recent_step=1.73s eta=11:32:30 +2026-09-20 17:00:14 | INFO | train rank=0 | step=23670/50000 progress=47.3% loss=0.087397 lr=5.849056e-05 lr_backbone=5.849056e-05 elapsed=10:22:13 avg_step=1.58s recent_step=1.13s eta=11:32:09 +2026-09-20 17:00:31 | INFO | train rank=0 | step=23680/50000 progress=47.4% loss=0.073662 lr=5.845797e-05 lr_backbone=5.845797e-05 elapsed=10:22:30 avg_step=1.58s recent_step=1.68s eta=11:31:54 +2026-09-20 17:00:48 | INFO | train rank=0 | step=23690/50000 progress=47.4% loss=0.110109 lr=5.842537e-05 lr_backbone=5.842537e-05 elapsed=10:22:48 avg_step=1.58s recent_step=1.74s eta=11:31:40 +2026-09-20 17:01:00 | INFO | train rank=0 | step=23700/50000 progress=47.4% loss=0.102770 lr=5.839278e-05 lr_backbone=5.839278e-05 elapsed=10:22:59 avg_step=1.58s recent_step=1.17s eta=11:31:20 +2026-09-20 17:01:25 | INFO | train rank=0 | step=23710/50000 progress=47.4% loss=0.067586 lr=5.836017e-05 lr_backbone=5.836017e-05 elapsed=10:23:25 avg_step=1.58s recent_step=2.53s eta=11:31:15 +2026-09-20 17:01:37 | INFO | train rank=0 | step=23720/50000 progress=47.4% loss=0.092729 lr=5.832757e-05 lr_backbone=5.832757e-05 elapsed=10:23:37 avg_step=1.58s recent_step=1.20s eta=11:30:55 +2026-09-20 17:01:49 | INFO | train rank=0 | step=23730/50000 progress=47.5% loss=0.100425 lr=5.829496e-05 lr_backbone=5.829496e-05 elapsed=10:23:48 avg_step=1.58s recent_step=1.17s eta=11:30:35 +2026-09-20 17:02:01 | INFO | train rank=0 | step=23740/50000 progress=47.5% loss=0.105114 lr=5.826235e-05 lr_backbone=5.826235e-05 elapsed=10:24:00 avg_step=1.58s recent_step=1.21s eta=11:30:15 +2026-09-20 17:02:15 | INFO | train rank=0 | step=23750/50000 progress=47.5% loss=0.075023 lr=5.822973e-05 lr_backbone=5.822973e-05 elapsed=10:24:15 avg_step=1.58s recent_step=1.46s eta=11:29:58 +2026-09-20 17:02:30 | INFO | train rank=0 | step=23760/50000 progress=47.5% loss=0.090005 lr=5.819711e-05 lr_backbone=5.819711e-05 elapsed=10:24:30 avg_step=1.58s recent_step=1.47s eta=11:29:41 +2026-09-20 17:02:52 | INFO | train rank=0 | step=23770/50000 progress=47.5% loss=0.096873 lr=5.816449e-05 lr_backbone=5.816449e-05 elapsed=10:24:51 avg_step=1.58s recent_step=2.13s eta=11:29:31 +2026-09-20 17:03:03 | INFO | train rank=0 | step=23780/50000 progress=47.6% loss=0.122809 lr=5.813186e-05 lr_backbone=5.813186e-05 elapsed=10:25:03 avg_step=1.58s recent_step=1.18s eta=11:29:11 +2026-09-20 17:03:23 | INFO | train rank=0 | step=23790/50000 progress=47.6% loss=0.073905 lr=5.809923e-05 lr_backbone=5.809923e-05 elapsed=10:25:23 avg_step=1.58s recent_step=2.00s eta=11:29:00 +2026-09-20 17:03:39 | INFO | train rank=0 | step=23800/50000 progress=47.6% loss=0.079691 lr=5.806659e-05 lr_backbone=5.806659e-05 elapsed=10:25:39 avg_step=1.58s recent_step=1.59s eta=11:28:44 +2026-09-20 17:03:54 | INFO | train rank=0 | step=23810/50000 progress=47.6% loss=0.086171 lr=5.803395e-05 lr_backbone=5.803395e-05 elapsed=10:25:54 avg_step=1.58s recent_step=1.52s eta=11:28:28 +2026-09-20 17:04:09 | INFO | train rank=0 | step=23820/50000 progress=47.6% loss=0.079920 lr=5.800131e-05 lr_backbone=5.800131e-05 elapsed=10:26:09 avg_step=1.58s recent_step=1.50s eta=11:28:11 +2026-09-20 17:04:25 | INFO | train rank=0 | step=23830/50000 progress=47.7% loss=0.107816 lr=5.796867e-05 lr_backbone=5.796867e-05 elapsed=10:26:24 avg_step=1.58s recent_step=1.54s eta=11:27:55 +2026-09-20 17:04:39 | INFO | train rank=0 | step=23840/50000 progress=47.7% loss=0.086636 lr=5.793602e-05 lr_backbone=5.793602e-05 elapsed=10:26:39 avg_step=1.58s recent_step=1.45s eta=11:27:38 +2026-09-20 17:04:53 | INFO | train rank=0 | step=23850/50000 progress=47.7% loss=0.107329 lr=5.790337e-05 lr_backbone=5.790337e-05 elapsed=10:26:53 avg_step=1.58s recent_step=1.42s eta=11:27:20 +2026-09-20 17:05:05 | INFO | train rank=0 | step=23860/50000 progress=47.7% loss=0.105417 lr=5.787071e-05 lr_backbone=5.787071e-05 elapsed=10:27:04 avg_step=1.58s recent_step=1.14s eta=11:27:00 +2026-09-20 17:05:22 | INFO | train rank=0 | step=23870/50000 progress=47.7% loss=0.107926 lr=5.783805e-05 lr_backbone=5.783805e-05 elapsed=10:27:22 avg_step=1.58s recent_step=1.76s eta=11:26:46 +2026-09-20 17:05:41 | INFO | train rank=0 | step=23880/50000 progress=47.8% loss=0.110026 lr=5.780539e-05 lr_backbone=5.780539e-05 elapsed=10:27:41 avg_step=1.58s recent_step=1.89s eta=11:26:34 +2026-09-20 17:05:53 | INFO | train rank=0 | step=23890/50000 progress=47.8% loss=0.111754 lr=5.777273e-05 lr_backbone=5.777273e-05 elapsed=10:27:53 avg_step=1.58s recent_step=1.18s eta=11:26:14 +2026-09-20 17:06:08 | INFO | train rank=0 | step=23900/50000 progress=47.8% loss=0.087582 lr=5.774006e-05 lr_backbone=5.774006e-05 elapsed=10:28:08 avg_step=1.58s recent_step=1.47s eta=11:25:57 +2026-09-20 17:06:25 | INFO | train rank=0 | step=23910/50000 progress=47.8% loss=0.087898 lr=5.770738e-05 lr_backbone=5.770738e-05 elapsed=10:28:25 avg_step=1.58s recent_step=1.69s eta=11:25:42 +2026-09-20 17:06:47 | INFO | train rank=0 | step=23920/50000 progress=47.8% loss=0.109766 lr=5.767471e-05 lr_backbone=5.767471e-05 elapsed=10:28:46 avg_step=1.58s recent_step=2.17s eta=11:25:33 +2026-09-20 17:06:58 | INFO | train rank=0 | step=23930/50000 progress=47.9% loss=0.108280 lr=5.764203e-05 lr_backbone=5.764203e-05 elapsed=10:28:58 avg_step=1.58s recent_step=1.17s eta=11:25:13 +2026-09-20 17:07:10 | INFO | train rank=0 | step=23940/50000 progress=47.9% loss=0.145243 lr=5.760935e-05 lr_backbone=5.760935e-05 elapsed=10:29:10 avg_step=1.58s recent_step=1.17s eta=11:24:53 +2026-09-20 17:07:22 | INFO | train rank=0 | step=23950/50000 progress=47.9% loss=0.075606 lr=5.757666e-05 lr_backbone=5.757666e-05 elapsed=10:29:21 avg_step=1.58s recent_step=1.17s eta=11:24:32 +2026-09-20 17:07:48 | INFO | train rank=0 | step=23960/50000 progress=47.9% loss=0.103190 lr=5.754397e-05 lr_backbone=5.754397e-05 elapsed=10:29:48 avg_step=1.58s recent_step=2.63s eta=11:24:28 +2026-09-20 17:08:03 | INFO | train rank=0 | step=23970/50000 progress=47.9% loss=0.080812 lr=5.751128e-05 lr_backbone=5.751128e-05 elapsed=10:30:02 avg_step=1.58s recent_step=1.46s eta=11:24:11 +2026-09-20 17:08:56 | INFO | train rank=0 | step=23980/50000 progress=48.0% loss=0.099993 lr=5.747858e-05 lr_backbone=5.747858e-05 elapsed=10:30:55 avg_step=1.58s recent_step=5.31s eta=11:24:36 +2026-09-20 17:09:08 | INFO | train rank=0 | step=23990/50000 progress=48.0% loss=0.083837 lr=5.744588e-05 lr_backbone=5.744588e-05 elapsed=10:31:08 avg_step=1.58s recent_step=1.23s eta=11:24:16 +2026-09-20 17:09:20 | INFO | train rank=0 | step=24000/50000 progress=48.0% loss=0.071526 lr=5.741318e-05 lr_backbone=5.741318e-05 elapsed=10:31:19 avg_step=1.58s recent_step=1.15s eta=11:23:56 +2026-09-20 17:09:34 | INFO | train rank=0 | step=24010/50000 progress=48.0% loss=0.073240 lr=5.738048e-05 lr_backbone=5.738048e-05 elapsed=10:31:33 avg_step=1.58s recent_step=1.40s eta=11:23:38 +2026-09-20 17:09:47 | INFO | train rank=0 | step=24020/50000 progress=48.0% loss=0.103225 lr=5.734777e-05 lr_backbone=5.734777e-05 elapsed=10:31:46 avg_step=1.58s recent_step=1.31s eta=11:23:19 +2026-09-20 17:10:05 | INFO | train rank=0 | step=24030/50000 progress=48.1% loss=0.072074 lr=5.731506e-05 lr_backbone=5.731506e-05 elapsed=10:32:05 avg_step=1.58s recent_step=1.88s eta=11:23:07 +2026-09-20 17:10:21 | INFO | train rank=0 | step=24040/50000 progress=48.1% loss=0.096206 lr=5.728234e-05 lr_backbone=5.728234e-05 elapsed=10:32:21 avg_step=1.58s recent_step=1.58s eta=11:22:51 +2026-09-20 17:10:33 | INFO | train rank=0 | step=24050/50000 progress=48.1% loss=0.074396 lr=5.724962e-05 lr_backbone=5.724962e-05 elapsed=10:32:33 avg_step=1.58s recent_step=1.17s eta=11:22:31 +2026-09-20 17:10:48 | INFO | train rank=0 | step=24060/50000 progress=48.1% loss=0.122536 lr=5.721690e-05 lr_backbone=5.721690e-05 elapsed=10:32:48 avg_step=1.58s recent_step=1.53s eta=11:22:15 +2026-09-20 17:11:05 | INFO | train rank=0 | step=24070/50000 progress=48.1% loss=0.087047 lr=5.718418e-05 lr_backbone=5.718418e-05 elapsed=10:33:04 avg_step=1.58s recent_step=1.65s eta=11:22:00 +2026-09-20 17:11:18 | INFO | train rank=0 | step=24080/50000 progress=48.2% loss=0.098344 lr=5.715145e-05 lr_backbone=5.715145e-05 elapsed=10:33:17 avg_step=1.58s recent_step=1.30s eta=11:21:41 +2026-09-20 17:11:33 | INFO | train rank=0 | step=24090/50000 progress=48.2% loss=0.100760 lr=5.711872e-05 lr_backbone=5.711872e-05 elapsed=10:33:33 avg_step=1.58s recent_step=1.56s eta=11:21:25 +2026-09-20 17:11:50 | INFO | train rank=0 | step=24100/50000 progress=48.2% loss=0.093364 lr=5.708598e-05 lr_backbone=5.708598e-05 elapsed=10:33:50 avg_step=1.58s recent_step=1.68s eta=11:21:10 +2026-09-20 17:12:06 | INFO | train rank=0 | step=24110/50000 progress=48.2% loss=0.075673 lr=5.705325e-05 lr_backbone=5.705325e-05 elapsed=10:34:06 avg_step=1.58s recent_step=1.60s eta=11:20:55 +2026-09-20 17:12:20 | INFO | train rank=0 | step=24120/50000 progress=48.2% loss=0.109040 lr=5.702051e-05 lr_backbone=5.702051e-05 elapsed=10:34:19 avg_step=1.58s recent_step=1.33s eta=11:20:36 +2026-09-20 17:12:39 | INFO | train rank=0 | step=24130/50000 progress=48.3% loss=0.098396 lr=5.698776e-05 lr_backbone=5.698776e-05 elapsed=10:34:38 avg_step=1.58s recent_step=1.94s eta=11:20:24 +2026-09-20 17:12:56 | INFO | train rank=0 | step=24140/50000 progress=48.3% loss=0.097208 lr=5.695502e-05 lr_backbone=5.695502e-05 elapsed=10:34:56 avg_step=1.58s recent_step=1.74s eta=11:20:10 +2026-09-20 17:13:13 | INFO | train rank=0 | step=24150/50000 progress=48.3% loss=0.085836 lr=5.692227e-05 lr_backbone=5.692227e-05 elapsed=10:35:13 avg_step=1.58s recent_step=1.71s eta=11:19:56 +2026-09-20 17:13:30 | INFO | train rank=0 | step=24160/50000 progress=48.3% loss=0.081614 lr=5.688951e-05 lr_backbone=5.688951e-05 elapsed=10:35:29 avg_step=1.58s recent_step=1.64s eta=11:19:41 +2026-09-20 17:13:49 | INFO | train rank=0 | step=24170/50000 progress=48.3% loss=0.117277 lr=5.685676e-05 lr_backbone=5.685676e-05 elapsed=10:35:49 avg_step=1.58s recent_step=1.91s eta=11:19:29 +2026-09-20 17:14:04 | INFO | train rank=0 | step=24180/50000 progress=48.4% loss=0.105951 lr=5.682400e-05 lr_backbone=5.682400e-05 elapsed=10:36:04 avg_step=1.58s recent_step=1.51s eta=11:19:12 +2026-09-20 17:14:18 | INFO | train rank=0 | step=24190/50000 progress=48.4% loss=0.084931 lr=5.679124e-05 lr_backbone=5.679124e-05 elapsed=10:36:18 avg_step=1.58s recent_step=1.39s eta=11:18:54 +2026-09-20 17:14:33 | INFO | train rank=0 | step=24200/50000 progress=48.4% loss=0.084899 lr=5.675847e-05 lr_backbone=5.675847e-05 elapsed=10:36:32 avg_step=1.58s recent_step=1.47s eta=11:18:37 +2026-09-20 17:14:50 | INFO | train rank=0 | step=24210/50000 progress=48.4% loss=0.123496 lr=5.672571e-05 lr_backbone=5.672571e-05 elapsed=10:36:50 avg_step=1.58s recent_step=1.79s eta=11:18:24 +2026-09-20 17:15:06 | INFO | train rank=0 | step=24220/50000 progress=48.4% loss=0.096523 lr=5.669294e-05 lr_backbone=5.669294e-05 elapsed=10:37:06 avg_step=1.58s recent_step=1.58s eta=11:18:08 +2026-09-20 17:15:20 | INFO | train rank=0 | step=24230/50000 progress=48.5% loss=0.090295 lr=5.666016e-05 lr_backbone=5.666016e-05 elapsed=10:37:20 avg_step=1.58s recent_step=1.39s eta=11:17:50 +2026-09-20 17:15:33 | INFO | train rank=0 | step=24240/50000 progress=48.5% loss=0.078614 lr=5.662739e-05 lr_backbone=5.662739e-05 elapsed=10:37:32 avg_step=1.58s recent_step=1.27s eta=11:17:31 +2026-09-20 17:15:52 | INFO | train rank=0 | step=24250/50000 progress=48.5% loss=0.095793 lr=5.659461e-05 lr_backbone=5.659461e-05 elapsed=10:37:52 avg_step=1.58s recent_step=1.95s eta=11:17:19 +2026-09-20 17:16:11 | INFO | train rank=0 | step=24260/50000 progress=48.5% loss=0.097758 lr=5.656183e-05 lr_backbone=5.656183e-05 elapsed=10:38:11 avg_step=1.58s recent_step=1.90s eta=11:17:07 +2026-09-20 17:16:23 | INFO | train rank=0 | step=24270/50000 progress=48.5% loss=0.099441 lr=5.652904e-05 lr_backbone=5.652904e-05 elapsed=10:38:23 avg_step=1.58s recent_step=1.20s eta=11:16:47 +2026-09-20 17:16:42 | INFO | train rank=0 | step=24280/50000 progress=48.6% loss=0.093819 lr=5.649625e-05 lr_backbone=5.649625e-05 elapsed=10:38:42 avg_step=1.58s recent_step=1.86s eta=11:16:34 +2026-09-20 17:16:55 | INFO | train rank=0 | step=24290/50000 progress=48.6% loss=0.082971 lr=5.646346e-05 lr_backbone=5.646346e-05 elapsed=10:38:54 avg_step=1.58s recent_step=1.26s eta=11:16:15 +2026-09-20 17:17:18 | INFO | train rank=0 | step=24300/50000 progress=48.6% loss=0.106155 lr=5.643067e-05 lr_backbone=5.643067e-05 elapsed=10:39:17 avg_step=1.58s recent_step=2.32s eta=11:16:07 +2026-09-20 17:17:30 | INFO | train rank=0 | step=24310/50000 progress=48.6% loss=0.131369 lr=5.639787e-05 lr_backbone=5.639787e-05 elapsed=10:39:30 avg_step=1.58s recent_step=1.22s eta=11:15:48 +2026-09-20 17:17:44 | INFO | train rank=0 | step=24320/50000 progress=48.6% loss=0.097517 lr=5.636508e-05 lr_backbone=5.636508e-05 elapsed=10:39:43 avg_step=1.58s recent_step=1.37s eta=11:15:30 +2026-09-20 17:17:58 | INFO | train rank=0 | step=24330/50000 progress=48.7% loss=0.097134 lr=5.633227e-05 lr_backbone=5.633227e-05 elapsed=10:39:58 avg_step=1.58s recent_step=1.48s eta=11:15:13 +2026-09-20 17:18:18 | INFO | train rank=0 | step=24340/50000 progress=48.7% loss=0.092091 lr=5.629947e-05 lr_backbone=5.629947e-05 elapsed=10:40:18 avg_step=1.58s recent_step=1.96s eta=11:15:01 +2026-09-20 17:18:36 | INFO | train rank=0 | step=24350/50000 progress=48.7% loss=0.076740 lr=5.626666e-05 lr_backbone=5.626666e-05 elapsed=10:40:35 avg_step=1.58s recent_step=1.76s eta=11:14:47 +2026-09-20 17:18:48 | INFO | train rank=0 | step=24360/50000 progress=48.7% loss=0.069874 lr=5.623385e-05 lr_backbone=5.623385e-05 elapsed=10:40:48 avg_step=1.58s recent_step=1.27s eta=11:14:28 +2026-09-20 17:19:02 | INFO | train rank=0 | step=24370/50000 progress=48.7% loss=0.104812 lr=5.620104e-05 lr_backbone=5.620104e-05 elapsed=10:41:01 avg_step=1.58s recent_step=1.36s eta=11:14:10 +2026-09-20 17:19:21 | INFO | train rank=0 | step=24380/50000 progress=48.8% loss=0.105110 lr=5.616822e-05 lr_backbone=5.616822e-05 elapsed=10:41:21 avg_step=1.58s recent_step=1.92s eta=11:13:58 +2026-09-20 17:19:36 | INFO | train rank=0 | step=24390/50000 progress=48.8% loss=0.099192 lr=5.613541e-05 lr_backbone=5.613541e-05 elapsed=10:41:36 avg_step=1.58s recent_step=1.54s eta=11:13:42 +2026-09-20 17:19:51 | INFO | train rank=0 | step=24400/50000 progress=48.8% loss=0.110177 lr=5.610258e-05 lr_backbone=5.610258e-05 elapsed=10:41:51 avg_step=1.58s recent_step=1.49s eta=11:13:25 +2026-09-20 17:20:07 | INFO | train rank=0 | step=24410/50000 progress=48.8% loss=0.096587 lr=5.606976e-05 lr_backbone=5.606976e-05 elapsed=10:42:06 avg_step=1.58s recent_step=1.53s eta=11:13:09 +2026-09-20 17:20:29 | INFO | train rank=0 | step=24420/50000 progress=48.8% loss=0.081682 lr=5.603693e-05 lr_backbone=5.603693e-05 elapsed=10:42:29 avg_step=1.58s recent_step=2.27s eta=11:13:00 +2026-09-20 17:20:41 | INFO | train rank=0 | step=24430/50000 progress=48.9% loss=0.105544 lr=5.600411e-05 lr_backbone=5.600411e-05 elapsed=10:42:40 avg_step=1.58s recent_step=1.13s eta=11:12:40 +2026-09-20 17:20:57 | INFO | train rank=0 | step=24440/50000 progress=48.9% loss=0.097532 lr=5.597127e-05 lr_backbone=5.597127e-05 elapsed=10:42:57 avg_step=1.58s recent_step=1.66s eta=11:12:25 +2026-09-20 17:21:09 | INFO | train rank=0 | step=24450/50000 progress=48.9% loss=0.103173 lr=5.593844e-05 lr_backbone=5.593844e-05 elapsed=10:43:09 avg_step=1.58s recent_step=1.20s eta=11:12:05 +2026-09-20 17:21:23 | INFO | train rank=0 | step=24460/50000 progress=48.9% loss=0.140626 lr=5.590560e-05 lr_backbone=5.590560e-05 elapsed=10:43:22 avg_step=1.58s recent_step=1.34s eta=11:11:47 +2026-09-20 17:21:40 | INFO | train rank=0 | step=24470/50000 progress=48.9% loss=0.111087 lr=5.587276e-05 lr_backbone=5.587276e-05 elapsed=10:43:40 avg_step=1.58s recent_step=1.73s eta=11:11:33 +2026-09-20 17:21:52 | INFO | train rank=0 | step=24480/50000 progress=49.0% loss=0.097096 lr=5.583992e-05 lr_backbone=5.583992e-05 elapsed=10:43:52 avg_step=1.58s recent_step=1.25s eta=11:11:13 +2026-09-20 17:22:05 | INFO | train rank=0 | step=24490/50000 progress=49.0% loss=0.089123 lr=5.580708e-05 lr_backbone=5.580708e-05 elapsed=10:44:04 avg_step=1.58s recent_step=1.22s eta=11:10:54 +2026-09-20 17:22:23 | INFO | train rank=0 | step=24500/50000 progress=49.0% loss=0.113381 lr=5.577423e-05 lr_backbone=5.577423e-05 elapsed=10:44:23 avg_step=1.58s recent_step=1.84s eta=11:10:41 +2026-09-20 17:22:38 | INFO | train rank=0 | step=24510/50000 progress=49.0% loss=0.099170 lr=5.574138e-05 lr_backbone=5.574138e-05 elapsed=10:44:38 avg_step=1.58s recent_step=1.52s eta=11:10:24 +2026-09-20 17:22:56 | INFO | train rank=0 | step=24520/50000 progress=49.0% loss=0.096664 lr=5.570853e-05 lr_backbone=5.570853e-05 elapsed=10:44:56 avg_step=1.58s recent_step=1.80s eta=11:10:11 +2026-09-20 17:23:10 | INFO | train rank=0 | step=24530/50000 progress=49.1% loss=0.095182 lr=5.567568e-05 lr_backbone=5.567568e-05 elapsed=10:45:10 avg_step=1.58s recent_step=1.37s eta=11:09:53 +2026-09-20 17:23:30 | INFO | train rank=0 | step=24540/50000 progress=49.1% loss=0.103318 lr=5.564282e-05 lr_backbone=5.564282e-05 elapsed=10:45:29 avg_step=1.58s recent_step=1.96s eta=11:09:41 +2026-09-20 17:23:45 | INFO | train rank=0 | step=24550/50000 progress=49.1% loss=0.095941 lr=5.560996e-05 lr_backbone=5.560996e-05 elapsed=10:45:45 avg_step=1.58s recent_step=1.53s eta=11:09:25 +2026-09-20 17:24:00 | INFO | train rank=0 | step=24560/50000 progress=49.1% loss=0.073703 lr=5.557710e-05 lr_backbone=5.557710e-05 elapsed=10:45:59 avg_step=1.58s recent_step=1.47s eta=11:09:08 +2026-09-20 17:24:17 | INFO | train rank=0 | step=24570/50000 progress=49.1% loss=0.082658 lr=5.554423e-05 lr_backbone=5.554423e-05 elapsed=10:46:16 avg_step=1.58s recent_step=1.71s eta=11:08:54 +2026-09-20 17:24:30 | INFO | train rank=0 | step=24580/50000 progress=49.2% loss=0.100367 lr=5.551137e-05 lr_backbone=5.551137e-05 elapsed=10:46:29 avg_step=1.58s recent_step=1.30s eta=11:08:35 +2026-09-20 17:24:49 | INFO | train rank=0 | step=24590/50000 progress=49.2% loss=0.089963 lr=5.547850e-05 lr_backbone=5.547850e-05 elapsed=10:46:49 avg_step=1.58s recent_step=1.96s eta=11:08:23 +2026-09-20 17:25:04 | INFO | train rank=0 | step=24600/50000 progress=49.2% loss=0.081574 lr=5.544563e-05 lr_backbone=5.544563e-05 elapsed=10:47:04 avg_step=1.58s recent_step=1.51s eta=11:08:07 +2026-09-20 17:25:19 | INFO | train rank=0 | step=24610/50000 progress=49.2% loss=0.114627 lr=5.541275e-05 lr_backbone=5.541275e-05 elapsed=10:47:18 avg_step=1.58s recent_step=1.41s eta=11:07:49 +2026-09-20 17:25:41 | INFO | train rank=0 | step=24620/50000 progress=49.2% loss=0.098474 lr=5.537988e-05 lr_backbone=5.537988e-05 elapsed=10:47:40 avg_step=1.58s recent_step=2.23s eta=11:07:40 +2026-09-20 17:25:55 | INFO | train rank=0 | step=24630/50000 progress=49.3% loss=0.126030 lr=5.534700e-05 lr_backbone=5.534700e-05 elapsed=10:47:55 avg_step=1.58s recent_step=1.43s eta=11:07:23 +2026-09-20 17:26:07 | INFO | train rank=0 | step=24640/50000 progress=49.3% loss=0.088768 lr=5.531412e-05 lr_backbone=5.531412e-05 elapsed=10:48:07 avg_step=1.58s recent_step=1.22s eta=11:07:03 +2026-09-20 17:26:23 | INFO | train rank=0 | step=24650/50000 progress=49.3% loss=0.089186 lr=5.528123e-05 lr_backbone=5.528123e-05 elapsed=10:48:22 avg_step=1.58s recent_step=1.54s eta=11:06:47 +2026-09-20 17:26:42 | INFO | train rank=0 | step=24660/50000 progress=49.3% loss=0.094273 lr=5.524835e-05 lr_backbone=5.524835e-05 elapsed=10:48:41 avg_step=1.58s recent_step=1.90s eta=11:06:35 +2026-09-20 17:26:56 | INFO | train rank=0 | step=24670/50000 progress=49.3% loss=0.084649 lr=5.521546e-05 lr_backbone=5.521546e-05 elapsed=10:48:56 avg_step=1.58s recent_step=1.42s eta=11:06:17 +2026-09-20 17:27:14 | INFO | train rank=0 | step=24680/50000 progress=49.4% loss=0.099221 lr=5.518257e-05 lr_backbone=5.518257e-05 elapsed=10:49:13 avg_step=1.58s recent_step=1.75s eta=11:06:03 +2026-09-20 17:27:26 | INFO | train rank=0 | step=24690/50000 progress=49.4% loss=0.089878 lr=5.514968e-05 lr_backbone=5.514968e-05 elapsed=10:49:26 avg_step=1.58s recent_step=1.24s eta=11:05:44 +2026-09-20 17:27:48 | INFO | train rank=0 | step=24700/50000 progress=49.4% loss=0.129318 lr=5.511678e-05 lr_backbone=5.511678e-05 elapsed=10:49:48 avg_step=1.58s recent_step=2.25s eta=11:05:35 +2026-09-20 17:28:00 | INFO | train rank=0 | step=24710/50000 progress=49.4% loss=0.080005 lr=5.508389e-05 lr_backbone=5.508389e-05 elapsed=10:50:00 avg_step=1.58s recent_step=1.16s eta=11:05:15 +2026-09-20 17:28:12 | INFO | train rank=0 | step=24720/50000 progress=49.4% loss=0.083052 lr=5.505099e-05 lr_backbone=5.505099e-05 elapsed=10:50:12 avg_step=1.58s recent_step=1.22s eta=11:04:56 +2026-09-20 17:28:24 | INFO | train rank=0 | step=24730/50000 progress=49.5% loss=0.085354 lr=5.501809e-05 lr_backbone=5.501809e-05 elapsed=10:50:24 avg_step=1.58s recent_step=1.21s eta=11:04:36 +2026-09-20 17:28:37 | INFO | train rank=0 | step=24740/50000 progress=49.5% loss=0.074869 lr=5.498518e-05 lr_backbone=5.498518e-05 elapsed=10:50:36 avg_step=1.58s recent_step=1.25s eta=11:04:17 +2026-09-20 17:28:54 | INFO | train rank=0 | step=24750/50000 progress=49.5% loss=0.095344 lr=5.495228e-05 lr_backbone=5.495228e-05 elapsed=10:50:53 avg_step=1.58s recent_step=1.71s eta=11:04:02 +2026-09-20 17:29:08 | INFO | train rank=0 | step=24760/50000 progress=49.5% loss=0.086670 lr=5.491937e-05 lr_backbone=5.491937e-05 elapsed=10:51:08 avg_step=1.58s recent_step=1.44s eta=11:03:45 +2026-09-20 17:29:29 | INFO | train rank=0 | step=24770/50000 progress=49.5% loss=0.095815 lr=5.488646e-05 lr_backbone=5.488646e-05 elapsed=10:51:29 avg_step=1.58s recent_step=2.07s eta=11:03:35 +2026-09-20 17:29:46 | INFO | train rank=0 | step=24780/50000 progress=49.6% loss=0.099254 lr=5.485355e-05 lr_backbone=5.485355e-05 elapsed=10:51:46 avg_step=1.58s recent_step=1.73s eta=11:03:20 +2026-09-20 17:30:00 | INFO | train rank=0 | step=24790/50000 progress=49.6% loss=0.087044 lr=5.482063e-05 lr_backbone=5.482063e-05 elapsed=10:51:59 avg_step=1.58s recent_step=1.35s eta=11:03:02 +2026-09-20 17:30:14 | INFO | train rank=0 | step=24800/50000 progress=49.6% loss=0.105411 lr=5.478772e-05 lr_backbone=5.478772e-05 elapsed=10:52:14 avg_step=1.58s recent_step=1.47s eta=11:02:45 +2026-09-20 17:30:27 | INFO | train rank=0 | step=24810/50000 progress=49.6% loss=0.095959 lr=5.475480e-05 lr_backbone=5.475480e-05 elapsed=10:52:27 avg_step=1.58s recent_step=1.28s eta=11:02:26 +2026-09-20 17:30:47 | INFO | train rank=0 | step=24820/50000 progress=49.6% loss=0.095678 lr=5.472188e-05 lr_backbone=5.472188e-05 elapsed=10:52:47 avg_step=1.58s recent_step=2.00s eta=11:02:15 +2026-09-20 17:31:00 | INFO | train rank=0 | step=24830/50000 progress=49.7% loss=0.093610 lr=5.468895e-05 lr_backbone=5.468895e-05 elapsed=10:52:59 avg_step=1.58s recent_step=1.25s eta=11:01:56 +2026-09-20 17:31:16 | INFO | train rank=0 | step=24840/50000 progress=49.7% loss=0.081486 lr=5.465603e-05 lr_backbone=5.465603e-05 elapsed=10:53:16 avg_step=1.58s recent_step=1.64s eta=11:01:41 +2026-09-20 17:31:32 | INFO | train rank=0 | step=24850/50000 progress=49.7% loss=0.087481 lr=5.462310e-05 lr_backbone=5.462310e-05 elapsed=10:53:32 avg_step=1.58s recent_step=1.61s eta=11:01:25 +2026-09-20 17:31:50 | INFO | train rank=0 | step=24860/50000 progress=49.7% loss=0.084664 lr=5.459017e-05 lr_backbone=5.459017e-05 elapsed=10:53:49 avg_step=1.58s recent_step=1.73s eta=11:01:11 +2026-09-20 17:32:01 | INFO | train rank=0 | step=24870/50000 progress=49.7% loss=0.086835 lr=5.455724e-05 lr_backbone=5.455724e-05 elapsed=10:54:01 avg_step=1.58s recent_step=1.18s eta=11:00:51 +2026-09-20 17:32:14 | INFO | train rank=0 | step=24880/50000 progress=49.8% loss=0.064380 lr=5.452431e-05 lr_backbone=5.452431e-05 elapsed=10:54:13 avg_step=1.58s recent_step=1.22s eta=11:00:32 +2026-09-20 17:32:37 | INFO | train rank=0 | step=24890/50000 progress=49.8% loss=0.081296 lr=5.449138e-05 lr_backbone=5.449138e-05 elapsed=10:54:37 avg_step=1.58s recent_step=2.33s eta=11:00:24 +2026-09-20 17:32:56 | INFO | train rank=0 | step=24900/50000 progress=49.8% loss=0.083516 lr=5.445844e-05 lr_backbone=5.445844e-05 elapsed=10:54:55 avg_step=1.58s recent_step=1.87s eta=11:00:11 +2026-09-20 17:33:46 | INFO | train rank=0 | step=24910/50000 progress=49.8% loss=0.082163 lr=5.442550e-05 lr_backbone=5.442550e-05 elapsed=10:55:46 avg_step=1.58s recent_step=5.06s eta=11:00:30 +2026-09-20 17:33:58 | INFO | train rank=0 | step=24920/50000 progress=49.8% loss=0.102444 lr=5.439256e-05 lr_backbone=5.439256e-05 elapsed=10:55:58 avg_step=1.58s recent_step=1.23s eta=11:00:11 +2026-09-20 17:34:15 | INFO | train rank=0 | step=24930/50000 progress=49.9% loss=0.125351 lr=5.435962e-05 lr_backbone=5.435962e-05 elapsed=10:56:14 avg_step=1.58s recent_step=1.64s eta=10:59:56 +2026-09-20 17:34:28 | INFO | train rank=0 | step=24940/50000 progress=49.9% loss=0.092341 lr=5.432667e-05 lr_backbone=5.432667e-05 elapsed=10:56:28 avg_step=1.58s recent_step=1.35s eta=10:59:37 +2026-09-20 17:34:43 | INFO | train rank=0 | step=24950/50000 progress=49.9% loss=0.082526 lr=5.429373e-05 lr_backbone=5.429373e-05 elapsed=10:56:42 avg_step=1.58s recent_step=1.44s eta=10:59:20 +2026-09-20 17:34:59 | INFO | train rank=0 | step=24960/50000 progress=49.9% loss=0.089001 lr=5.426078e-05 lr_backbone=5.426078e-05 elapsed=10:56:59 avg_step=1.58s recent_step=1.68s eta=10:59:05 +2026-09-20 17:35:13 | INFO | train rank=0 | step=24970/50000 progress=49.9% loss=0.097656 lr=5.422783e-05 lr_backbone=5.422783e-05 elapsed=10:57:13 avg_step=1.58s recent_step=1.40s eta=10:58:48 +2026-09-20 17:35:26 | INFO | train rank=0 | step=24980/50000 progress=50.0% loss=0.110373 lr=5.419488e-05 lr_backbone=5.419488e-05 elapsed=10:57:26 avg_step=1.58s recent_step=1.26s eta=10:58:29 +2026-09-20 17:35:39 | INFO | train rank=0 | step=24990/50000 progress=50.0% loss=0.076382 lr=5.416192e-05 lr_backbone=5.416192e-05 elapsed=10:57:38 avg_step=1.58s recent_step=1.27s eta=10:58:10 +2026-09-20 17:35:53 | INFO | train rank=0 | step=25000/50000 progress=50.0% loss=0.113593 lr=5.412897e-05 lr_backbone=5.412897e-05 elapsed=10:57:52 avg_step=1.58s recent_step=1.41s eta=10:57:52 +2026-09-20 17:36:12 | INFO | train rank=0 | step=25010/50000 progress=50.0% loss=0.123644 lr=5.409601e-05 lr_backbone=5.409601e-05 elapsed=10:58:11 avg_step=1.58s recent_step=1.89s eta=10:57:40 +2026-09-20 17:36:24 | INFO | train rank=0 | step=25020/50000 progress=50.0% loss=0.078645 lr=5.406305e-05 lr_backbone=5.406305e-05 elapsed=10:58:23 avg_step=1.58s recent_step=1.22s eta=10:57:20 +2026-09-20 17:36:39 | INFO | train rank=0 | step=25030/50000 progress=50.1% loss=0.091600 lr=5.403009e-05 lr_backbone=5.403009e-05 elapsed=10:58:39 avg_step=1.58s recent_step=1.52s eta=10:57:04 +2026-09-20 17:36:52 | INFO | train rank=0 | step=25040/50000 progress=50.1% loss=0.091233 lr=5.399713e-05 lr_backbone=5.399713e-05 elapsed=10:58:52 avg_step=1.58s recent_step=1.34s eta=10:56:46 +2026-09-20 17:37:08 | INFO | train rank=0 | step=25050/50000 progress=50.1% loss=0.143351 lr=5.396416e-05 lr_backbone=5.396416e-05 elapsed=10:59:08 avg_step=1.58s recent_step=1.59s eta=10:56:30 +2026-09-20 17:37:22 | INFO | train rank=0 | step=25060/50000 progress=50.1% loss=0.103874 lr=5.393120e-05 lr_backbone=5.393120e-05 elapsed=10:59:22 avg_step=1.58s recent_step=1.36s eta=10:56:12 +2026-09-20 17:37:42 | INFO | train rank=0 | step=25070/50000 progress=50.1% loss=0.115884 lr=5.389823e-05 lr_backbone=5.389823e-05 elapsed=10:59:42 avg_step=1.58s recent_step=2.04s eta=10:56:01 +2026-09-20 17:38:02 | INFO | train rank=0 | step=25080/50000 progress=50.2% loss=0.136107 lr=5.386526e-05 lr_backbone=5.386526e-05 elapsed=11:00:01 avg_step=1.58s recent_step=1.94s eta=10:55:49 +2026-09-20 17:38:14 | INFO | train rank=0 | step=25090/50000 progress=50.2% loss=0.089618 lr=5.383229e-05 lr_backbone=5.383229e-05 elapsed=11:00:13 avg_step=1.58s recent_step=1.21s eta=10:55:29 +2026-09-20 17:38:27 | INFO | train rank=0 | step=25100/50000 progress=50.2% loss=0.096555 lr=5.379931e-05 lr_backbone=5.379931e-05 elapsed=11:00:27 avg_step=1.58s recent_step=1.31s eta=10:55:11 +2026-09-20 17:38:47 | INFO | train rank=0 | step=25110/50000 progress=50.2% loss=0.083861 lr=5.376634e-05 lr_backbone=5.376634e-05 elapsed=11:00:46 avg_step=1.58s recent_step=1.97s eta=10:54:59 +2026-09-20 17:39:09 | INFO | train rank=0 | step=25120/50000 progress=50.2% loss=0.084483 lr=5.373336e-05 lr_backbone=5.373336e-05 elapsed=11:01:09 avg_step=1.58s recent_step=2.24s eta=10:54:50 +2026-09-20 17:39:22 | INFO | train rank=0 | step=25130/50000 progress=50.3% loss=0.112699 lr=5.370039e-05 lr_backbone=5.370039e-05 elapsed=11:01:21 avg_step=1.58s recent_step=1.24s eta=10:54:31 +2026-09-20 17:39:34 | INFO | train rank=0 | step=25140/50000 progress=50.3% loss=0.116104 lr=5.366741e-05 lr_backbone=5.366741e-05 elapsed=11:01:33 avg_step=1.58s recent_step=1.22s eta=10:54:11 +2026-09-20 17:39:49 | INFO | train rank=0 | step=25150/50000 progress=50.3% loss=0.083582 lr=5.363443e-05 lr_backbone=5.363443e-05 elapsed=11:01:48 avg_step=1.58s recent_step=1.48s eta=10:53:55 +2026-09-20 17:40:07 | INFO | train rank=0 | step=25160/50000 progress=50.3% loss=0.078516 lr=5.360144e-05 lr_backbone=5.360144e-05 elapsed=11:02:07 avg_step=1.58s recent_step=1.84s eta=10:53:41 +2026-09-20 17:40:23 | INFO | train rank=0 | step=25170/50000 progress=50.3% loss=0.076713 lr=5.356846e-05 lr_backbone=5.356846e-05 elapsed=11:02:22 avg_step=1.58s recent_step=1.57s eta=10:53:25 +2026-09-20 17:40:41 | INFO | train rank=0 | step=25180/50000 progress=50.4% loss=0.079272 lr=5.353547e-05 lr_backbone=5.353547e-05 elapsed=11:02:41 avg_step=1.58s recent_step=1.85s eta=10:53:12 +2026-09-20 17:40:54 | INFO | train rank=0 | step=25190/50000 progress=50.4% loss=0.086580 lr=5.350249e-05 lr_backbone=5.350249e-05 elapsed=11:02:53 avg_step=1.58s recent_step=1.24s eta=10:52:53 +2026-09-20 17:41:07 | INFO | train rank=0 | step=25200/50000 progress=50.4% loss=0.076969 lr=5.346950e-05 lr_backbone=5.346950e-05 elapsed=11:03:07 avg_step=1.58s recent_step=1.37s eta=10:52:35 +2026-09-20 17:41:28 | INFO | train rank=0 | step=25210/50000 progress=50.4% loss=0.105416 lr=5.343651e-05 lr_backbone=5.343651e-05 elapsed=11:03:28 avg_step=1.58s recent_step=2.08s eta=10:52:24 +2026-09-20 17:41:43 | INFO | train rank=0 | step=25220/50000 progress=50.4% loss=0.089162 lr=5.340351e-05 lr_backbone=5.340351e-05 elapsed=11:03:42 avg_step=1.58s recent_step=1.45s eta=10:52:07 +2026-09-20 17:41:58 | INFO | train rank=0 | step=25230/50000 progress=50.5% loss=0.097040 lr=5.337052e-05 lr_backbone=5.337052e-05 elapsed=11:03:57 avg_step=1.58s recent_step=1.51s eta=10:51:51 +2026-09-20 17:42:20 | INFO | train rank=0 | step=25240/50000 progress=50.5% loss=0.117085 lr=5.333753e-05 lr_backbone=5.333753e-05 elapsed=11:04:19 avg_step=1.58s recent_step=2.19s eta=10:51:41 +2026-09-20 17:42:32 | INFO | train rank=0 | step=25250/50000 progress=50.5% loss=0.078586 lr=5.330453e-05 lr_backbone=5.330453e-05 elapsed=11:04:31 avg_step=1.58s recent_step=1.20s eta=10:51:22 +2026-09-20 17:42:47 | INFO | train rank=0 | step=25260/50000 progress=50.5% loss=0.078097 lr=5.327153e-05 lr_backbone=5.327153e-05 elapsed=11:04:47 avg_step=1.58s recent_step=1.53s eta=10:51:05 +2026-09-20 17:43:03 | INFO | train rank=0 | step=25270/50000 progress=50.5% loss=0.100841 lr=5.323853e-05 lr_backbone=5.323853e-05 elapsed=11:05:02 avg_step=1.58s recent_step=1.57s eta=10:50:50 +2026-09-20 17:43:17 | INFO | train rank=0 | step=25280/50000 progress=50.6% loss=0.114793 lr=5.320553e-05 lr_backbone=5.320553e-05 elapsed=11:05:17 avg_step=1.58s recent_step=1.42s eta=10:50:32 +2026-09-20 17:43:34 | INFO | train rank=0 | step=25290/50000 progress=50.6% loss=0.088547 lr=5.317253e-05 lr_backbone=5.317253e-05 elapsed=11:05:33 avg_step=1.58s recent_step=1.67s eta=10:50:17 +2026-09-20 17:43:51 | INFO | train rank=0 | step=25300/50000 progress=50.6% loss=0.072208 lr=5.313953e-05 lr_backbone=5.313953e-05 elapsed=11:05:51 avg_step=1.58s recent_step=1.79s eta=10:50:04 +2026-09-20 17:44:06 | INFO | train rank=0 | step=25310/50000 progress=50.6% loss=0.087179 lr=5.310652e-05 lr_backbone=5.310652e-05 elapsed=11:06:06 avg_step=1.58s recent_step=1.46s eta=10:49:47 +2026-09-20 17:44:21 | INFO | train rank=0 | step=25320/50000 progress=50.6% loss=0.069146 lr=5.307351e-05 lr_backbone=5.307351e-05 elapsed=11:06:21 avg_step=1.58s recent_step=1.50s eta=10:49:30 +2026-09-20 17:44:38 | INFO | train rank=0 | step=25330/50000 progress=50.7% loss=0.081537 lr=5.304051e-05 lr_backbone=5.304051e-05 elapsed=11:06:38 avg_step=1.58s recent_step=1.70s eta=10:49:16 +2026-09-20 17:44:52 | INFO | train rank=0 | step=25340/50000 progress=50.7% loss=0.077238 lr=5.300750e-05 lr_backbone=5.300750e-05 elapsed=11:06:51 avg_step=1.58s recent_step=1.34s eta=10:48:57 +2026-09-20 17:45:13 | INFO | train rank=0 | step=25350/50000 progress=50.7% loss=0.084461 lr=5.297449e-05 lr_backbone=5.297449e-05 elapsed=11:07:13 avg_step=1.58s recent_step=2.19s eta=10:48:48 +2026-09-20 17:45:25 | INFO | train rank=0 | step=25360/50000 progress=50.7% loss=0.080370 lr=5.294148e-05 lr_backbone=5.294148e-05 elapsed=11:07:25 avg_step=1.58s recent_step=1.19s eta=10:48:28 +2026-09-20 17:45:38 | INFO | train rank=0 | step=25370/50000 progress=50.7% loss=0.097209 lr=5.290846e-05 lr_backbone=5.290846e-05 elapsed=11:07:38 avg_step=1.58s recent_step=1.25s eta=10:48:09 +2026-09-20 17:45:58 | INFO | train rank=0 | step=25380/50000 progress=50.8% loss=0.099967 lr=5.287545e-05 lr_backbone=5.287545e-05 elapsed=11:07:58 avg_step=1.58s recent_step=2.02s eta=10:47:58 +2026-09-20 17:46:12 | INFO | train rank=0 | step=25390/50000 progress=50.8% loss=0.141601 lr=5.284243e-05 lr_backbone=5.284243e-05 elapsed=11:08:12 avg_step=1.58s recent_step=1.40s eta=10:47:40 +2026-09-20 17:46:27 | INFO | train rank=0 | step=25400/50000 progress=50.8% loss=0.074539 lr=5.280942e-05 lr_backbone=5.280942e-05 elapsed=11:08:26 avg_step=1.58s recent_step=1.45s eta=10:47:23 +2026-09-20 17:46:44 | INFO | train rank=0 | step=25410/50000 progress=50.8% loss=0.080598 lr=5.277640e-05 lr_backbone=5.277640e-05 elapsed=11:08:43 avg_step=1.58s recent_step=1.72s eta=10:47:09 +2026-09-20 17:47:02 | INFO | train rank=0 | step=25420/50000 progress=50.8% loss=0.102061 lr=5.274338e-05 lr_backbone=5.274338e-05 elapsed=11:09:02 avg_step=1.58s recent_step=1.84s eta=10:46:55 +2026-09-20 17:47:14 | INFO | train rank=0 | step=25430/50000 progress=50.9% loss=0.099247 lr=5.271036e-05 lr_backbone=5.271036e-05 elapsed=11:09:13 avg_step=1.58s recent_step=1.15s eta=10:46:35 +2026-09-20 17:47:32 | INFO | train rank=0 | step=25440/50000 progress=50.9% loss=0.089207 lr=5.267734e-05 lr_backbone=5.267734e-05 elapsed=11:09:32 avg_step=1.58s recent_step=1.83s eta=10:46:22 +2026-09-20 17:47:46 | INFO | train rank=0 | step=25450/50000 progress=50.9% loss=0.082203 lr=5.264432e-05 lr_backbone=5.264432e-05 elapsed=11:09:46 avg_step=1.58s recent_step=1.45s eta=10:46:05 +2026-09-20 17:48:06 | INFO | train rank=0 | step=25460/50000 progress=50.9% loss=0.122982 lr=5.261129e-05 lr_backbone=5.261129e-05 elapsed=11:10:06 avg_step=1.58s recent_step=1.97s eta=10:45:53 +2026-09-20 17:48:18 | INFO | train rank=0 | step=25470/50000 progress=50.9% loss=0.069649 lr=5.257827e-05 lr_backbone=5.257827e-05 elapsed=11:10:18 avg_step=1.58s recent_step=1.22s eta=10:45:34 +2026-09-20 17:48:33 | INFO | train rank=0 | step=25480/50000 progress=51.0% loss=0.112597 lr=5.254524e-05 lr_backbone=5.254524e-05 elapsed=11:10:33 avg_step=1.58s recent_step=1.47s eta=10:45:17 +2026-09-20 17:48:46 | INFO | train rank=0 | step=25490/50000 progress=51.0% loss=0.076579 lr=5.251222e-05 lr_backbone=5.251222e-05 elapsed=11:10:46 avg_step=1.58s recent_step=1.32s eta=10:44:59 +2026-09-20 17:49:09 | INFO | train rank=0 | step=25500/50000 progress=51.0% loss=0.078009 lr=5.247919e-05 lr_backbone=5.247919e-05 elapsed=11:11:08 avg_step=1.58s recent_step=2.24s eta=10:44:49 +2026-09-20 17:49:21 | INFO | train rank=0 | step=25510/50000 progress=51.0% loss=0.073477 lr=5.244616e-05 lr_backbone=5.244616e-05 elapsed=11:11:20 avg_step=1.58s recent_step=1.18s eta=10:44:30 +2026-09-20 17:49:33 | INFO | train rank=0 | step=25520/50000 progress=51.0% loss=0.090532 lr=5.241313e-05 lr_backbone=5.241313e-05 elapsed=11:11:32 avg_step=1.58s recent_step=1.21s eta=10:44:10 +2026-09-20 17:49:47 | INFO | train rank=0 | step=25530/50000 progress=51.1% loss=0.089511 lr=5.238010e-05 lr_backbone=5.238010e-05 elapsed=11:11:47 avg_step=1.58s recent_step=1.45s eta=10:43:53 +2026-09-20 17:50:04 | INFO | train rank=0 | step=25540/50000 progress=51.1% loss=0.083449 lr=5.234706e-05 lr_backbone=5.234706e-05 elapsed=11:12:04 avg_step=1.58s recent_step=1.69s eta=10:43:38 +2026-09-20 17:50:21 | INFO | train rank=0 | step=25550/50000 progress=51.1% loss=0.071863 lr=5.231403e-05 lr_backbone=5.231403e-05 elapsed=11:12:21 avg_step=1.58s recent_step=1.72s eta=10:43:24 +2026-09-20 17:50:38 | INFO | train rank=0 | step=25560/50000 progress=51.1% loss=0.072634 lr=5.228100e-05 lr_backbone=5.228100e-05 elapsed=11:12:38 avg_step=1.58s recent_step=1.67s eta=10:43:09 +2026-09-20 17:50:54 | INFO | train rank=0 | step=25570/50000 progress=51.1% loss=0.117957 lr=5.224796e-05 lr_backbone=5.224796e-05 elapsed=11:12:53 avg_step=1.58s recent_step=1.60s eta=10:42:53 +2026-09-20 17:51:13 | INFO | train rank=0 | step=25580/50000 progress=51.2% loss=0.101286 lr=5.221492e-05 lr_backbone=5.221492e-05 elapsed=11:13:13 avg_step=1.58s recent_step=1.96s eta=10:42:41 +2026-09-20 17:51:25 | INFO | train rank=0 | step=25590/50000 progress=51.2% loss=0.149847 lr=5.218189e-05 lr_backbone=5.218189e-05 elapsed=11:13:25 avg_step=1.58s recent_step=1.18s eta=10:42:22 +2026-09-20 17:51:40 | INFO | train rank=0 | step=25600/50000 progress=51.2% loss=0.080459 lr=5.214885e-05 lr_backbone=5.214885e-05 elapsed=11:13:40 avg_step=1.58s recent_step=1.51s eta=10:42:05 +2026-09-20 17:51:58 | INFO | train rank=0 | step=25610/50000 progress=51.2% loss=0.090252 lr=5.211581e-05 lr_backbone=5.211581e-05 elapsed=11:13:57 avg_step=1.58s recent_step=1.72s eta=10:41:51 +2026-09-20 17:52:15 | INFO | train rank=0 | step=25620/50000 progress=51.2% loss=0.074640 lr=5.208277e-05 lr_backbone=5.208277e-05 elapsed=11:14:15 avg_step=1.58s recent_step=1.78s eta=10:41:37 +2026-09-20 17:52:29 | INFO | train rank=0 | step=25630/50000 progress=51.3% loss=0.101500 lr=5.204973e-05 lr_backbone=5.204973e-05 elapsed=11:14:28 avg_step=1.58s recent_step=1.34s eta=10:41:19 +2026-09-20 17:52:43 | INFO | train rank=0 | step=25640/50000 progress=51.3% loss=0.090698 lr=5.201669e-05 lr_backbone=5.201669e-05 elapsed=11:14:43 avg_step=1.58s recent_step=1.44s eta=10:41:02 +2026-09-20 17:52:57 | INFO | train rank=0 | step=25650/50000 progress=51.3% loss=0.114133 lr=5.198364e-05 lr_backbone=5.198364e-05 elapsed=11:14:56 avg_step=1.58s recent_step=1.34s eta=10:40:44 +2026-09-20 17:53:11 | INFO | train rank=0 | step=25660/50000 progress=51.3% loss=0.076236 lr=5.195060e-05 lr_backbone=5.195060e-05 elapsed=11:15:11 avg_step=1.58s recent_step=1.45s eta=10:40:27 +2026-09-20 17:53:29 | INFO | train rank=0 | step=25670/50000 progress=51.3% loss=0.092070 lr=5.191755e-05 lr_backbone=5.191755e-05 elapsed=11:15:28 avg_step=1.58s recent_step=1.76s eta=10:40:13 +2026-09-20 17:53:42 | INFO | train rank=0 | step=25680/50000 progress=51.4% loss=0.100043 lr=5.188451e-05 lr_backbone=5.188451e-05 elapsed=11:15:42 avg_step=1.58s recent_step=1.33s eta=10:39:55 +2026-09-20 17:53:58 | INFO | train rank=0 | step=25690/50000 progress=51.4% loss=0.098272 lr=5.185146e-05 lr_backbone=5.185146e-05 elapsed=11:15:58 avg_step=1.58s recent_step=1.62s eta=10:39:39 +2026-09-20 17:54:13 | INFO | train rank=0 | step=25700/50000 progress=51.4% loss=0.082581 lr=5.181842e-05 lr_backbone=5.181842e-05 elapsed=11:16:13 avg_step=1.58s recent_step=1.48s eta=10:39:22 +2026-09-20 17:54:27 | INFO | train rank=0 | step=25710/50000 progress=51.4% loss=0.076797 lr=5.178537e-05 lr_backbone=5.178537e-05 elapsed=11:16:26 avg_step=1.58s recent_step=1.38s eta=10:39:05 +2026-09-20 17:54:46 | INFO | train rank=0 | step=25720/50000 progress=51.4% loss=0.081527 lr=5.175232e-05 lr_backbone=5.175232e-05 elapsed=11:16:46 avg_step=1.58s recent_step=1.97s eta=10:38:53 +2026-09-20 17:55:05 | INFO | train rank=0 | step=25730/50000 progress=51.5% loss=0.092392 lr=5.171927e-05 lr_backbone=5.171927e-05 elapsed=11:17:05 avg_step=1.58s recent_step=1.89s eta=10:38:40 +2026-09-20 17:55:17 | INFO | train rank=0 | step=25740/50000 progress=51.5% loss=0.086219 lr=5.168622e-05 lr_backbone=5.168622e-05 elapsed=11:17:17 avg_step=1.58s recent_step=1.17s eta=10:38:20 +2026-09-20 17:55:33 | INFO | train rank=0 | step=25750/50000 progress=51.5% loss=0.096021 lr=5.165317e-05 lr_backbone=5.165317e-05 elapsed=11:17:33 avg_step=1.58s recent_step=1.63s eta=10:38:05 +2026-09-20 17:55:49 | INFO | train rank=0 | step=25760/50000 progress=51.5% loss=0.133955 lr=5.162012e-05 lr_backbone=5.162012e-05 elapsed=11:17:49 avg_step=1.58s recent_step=1.57s eta=10:37:49 +2026-09-20 17:56:02 | INFO | train rank=0 | step=25770/50000 progress=51.5% loss=0.086152 lr=5.158706e-05 lr_backbone=5.158706e-05 elapsed=11:18:02 avg_step=1.58s recent_step=1.33s eta=10:37:31 +2026-09-20 17:56:18 | INFO | train rank=0 | step=25780/50000 progress=51.6% loss=0.202932 lr=5.155401e-05 lr_backbone=5.155401e-05 elapsed=11:18:17 avg_step=1.58s recent_step=1.54s eta=10:37:15 +2026-09-20 17:56:30 | INFO | train rank=0 | step=25790/50000 progress=51.6% loss=0.121312 lr=5.152096e-05 lr_backbone=5.152096e-05 elapsed=11:18:29 avg_step=1.58s recent_step=1.19s eta=10:36:55 +2026-09-20 17:56:48 | INFO | train rank=0 | step=25800/50000 progress=51.6% loss=0.087129 lr=5.148790e-05 lr_backbone=5.148790e-05 elapsed=11:18:47 avg_step=1.58s recent_step=1.80s eta=10:36:42 +2026-09-20 17:57:12 | INFO | train rank=0 | step=25810/50000 progress=51.6% loss=0.112657 lr=5.145485e-05 lr_backbone=5.145485e-05 elapsed=11:19:11 avg_step=1.58s recent_step=2.42s eta=10:36:34 +2026-09-20 17:58:00 | INFO | train rank=0 | step=25820/50000 progress=51.6% loss=0.084082 lr=5.142179e-05 lr_backbone=5.142179e-05 elapsed=11:20:00 avg_step=1.58s recent_step=4.86s eta=10:36:49 +2026-09-20 17:58:13 | INFO | train rank=0 | step=25830/50000 progress=51.7% loss=0.104170 lr=5.138874e-05 lr_backbone=5.138874e-05 elapsed=11:20:13 avg_step=1.58s recent_step=1.25s eta=10:36:30 +2026-09-20 17:58:26 | INFO | train rank=0 | step=25840/50000 progress=51.7% loss=0.076398 lr=5.135568e-05 lr_backbone=5.135568e-05 elapsed=11:20:25 avg_step=1.58s recent_step=1.27s eta=10:36:11 +2026-09-20 17:58:39 | INFO | train rank=0 | step=25850/50000 progress=51.7% loss=0.077883 lr=5.132262e-05 lr_backbone=5.132262e-05 elapsed=11:20:39 avg_step=1.58s recent_step=1.35s eta=10:35:53 +2026-09-20 17:58:53 | INFO | train rank=0 | step=25860/50000 progress=51.7% loss=0.089527 lr=5.128956e-05 lr_backbone=5.128956e-05 elapsed=11:20:53 avg_step=1.58s recent_step=1.39s eta=10:35:35 +2026-09-20 17:59:11 | INFO | train rank=0 | step=25870/50000 progress=51.7% loss=0.110493 lr=5.125650e-05 lr_backbone=5.125650e-05 elapsed=11:21:11 avg_step=1.58s recent_step=1.79s eta=10:35:22 +2026-09-20 17:59:25 | INFO | train rank=0 | step=25880/50000 progress=51.8% loss=0.088131 lr=5.122345e-05 lr_backbone=5.122345e-05 elapsed=11:21:24 avg_step=1.58s recent_step=1.36s eta=10:35:04 +2026-09-20 17:59:41 | INFO | train rank=0 | step=25890/50000 progress=51.8% loss=0.105028 lr=5.119039e-05 lr_backbone=5.119039e-05 elapsed=11:21:41 avg_step=1.58s recent_step=1.65s eta=10:34:49 +2026-09-20 17:59:53 | INFO | train rank=0 | step=25900/50000 progress=51.8% loss=0.092955 lr=5.115733e-05 lr_backbone=5.115733e-05 elapsed=11:21:53 avg_step=1.58s recent_step=1.24s eta=10:34:30 +2026-09-20 18:00:14 | INFO | train rank=0 | step=25910/50000 progress=51.8% loss=0.084201 lr=5.112426e-05 lr_backbone=5.112426e-05 elapsed=11:22:13 avg_step=1.58s recent_step=2.02s eta=10:34:18 +2026-09-20 18:00:25 | INFO | train rank=0 | step=25920/50000 progress=51.8% loss=0.099400 lr=5.109120e-05 lr_backbone=5.109120e-05 elapsed=11:22:25 avg_step=1.58s recent_step=1.17s eta=10:33:58 +2026-09-20 18:00:37 | INFO | train rank=0 | step=25930/50000 progress=51.9% loss=0.086394 lr=5.105814e-05 lr_backbone=5.105814e-05 elapsed=11:22:37 avg_step=1.58s recent_step=1.20s eta=10:33:39 +2026-09-20 18:00:52 | INFO | train rank=0 | step=25940/50000 progress=51.9% loss=0.087255 lr=5.102508e-05 lr_backbone=5.102508e-05 elapsed=11:22:52 avg_step=1.58s recent_step=1.45s eta=10:33:22 +2026-09-20 18:01:08 | INFO | train rank=0 | step=25950/50000 progress=51.9% loss=0.085804 lr=5.099202e-05 lr_backbone=5.099202e-05 elapsed=11:23:08 avg_step=1.58s recent_step=1.63s eta=10:33:07 +2026-09-20 18:01:25 | INFO | train rank=0 | step=25960/50000 progress=51.9% loss=0.077292 lr=5.095895e-05 lr_backbone=5.095895e-05 elapsed=11:23:24 avg_step=1.58s recent_step=1.63s eta=10:32:51 +2026-09-20 18:01:40 | INFO | train rank=0 | step=25970/50000 progress=51.9% loss=0.086030 lr=5.092589e-05 lr_backbone=5.092589e-05 elapsed=11:23:40 avg_step=1.58s recent_step=1.54s eta=10:32:35 +2026-09-20 18:01:58 | INFO | train rank=0 | step=25980/50000 progress=52.0% loss=0.072729 lr=5.089283e-05 lr_backbone=5.089283e-05 elapsed=11:23:58 avg_step=1.58s recent_step=1.84s eta=10:32:22 +2026-09-20 18:02:17 | INFO | train rank=0 | step=25990/50000 progress=52.0% loss=0.088772 lr=5.085976e-05 lr_backbone=5.085976e-05 elapsed=11:24:16 avg_step=1.58s recent_step=1.86s eta=10:32:09 +2026-09-20 18:02:34 | INFO | train rank=0 | step=26000/50000 progress=52.0% loss=0.120832 lr=5.082670e-05 lr_backbone=5.082670e-05 elapsed=11:24:34 avg_step=1.58s recent_step=1.73s eta=10:31:54 +2026-09-20 18:02:51 | INFO | train rank=0 | step=26010/50000 progress=52.0% loss=0.100055 lr=5.079363e-05 lr_backbone=5.079363e-05 elapsed=11:24:50 avg_step=1.58s recent_step=1.67s eta=10:31:39 +2026-09-20 18:03:06 | INFO | train rank=0 | step=26020/50000 progress=52.0% loss=0.112582 lr=5.076057e-05 lr_backbone=5.076057e-05 elapsed=11:25:05 avg_step=1.58s recent_step=1.47s eta=10:31:22 +2026-09-20 18:03:25 | INFO | train rank=0 | step=26030/50000 progress=52.1% loss=0.095101 lr=5.072750e-05 lr_backbone=5.072750e-05 elapsed=11:25:25 avg_step=1.58s recent_step=1.96s eta=10:31:10 +2026-09-20 18:03:37 | INFO | train rank=0 | step=26040/50000 progress=52.1% loss=0.114031 lr=5.069443e-05 lr_backbone=5.069443e-05 elapsed=11:25:37 avg_step=1.58s recent_step=1.17s eta=10:30:51 +2026-09-20 18:03:53 | INFO | train rank=0 | step=26050/50000 progress=52.1% loss=0.092520 lr=5.066137e-05 lr_backbone=5.066137e-05 elapsed=11:25:52 avg_step=1.58s recent_step=1.58s eta=10:30:35 +2026-09-20 18:04:07 | INFO | train rank=0 | step=26060/50000 progress=52.1% loss=0.088467 lr=5.062830e-05 lr_backbone=5.062830e-05 elapsed=11:26:07 avg_step=1.58s recent_step=1.46s eta=10:30:18 +2026-09-20 18:04:23 | INFO | train rank=0 | step=26070/50000 progress=52.1% loss=0.074667 lr=5.059524e-05 lr_backbone=5.059524e-05 elapsed=11:26:23 avg_step=1.58s recent_step=1.63s eta=10:30:02 +2026-09-20 18:04:44 | INFO | train rank=0 | step=26080/50000 progress=52.2% loss=0.091555 lr=5.056217e-05 lr_backbone=5.056217e-05 elapsed=11:26:43 avg_step=1.58s recent_step=2.02s eta=10:29:51 +2026-09-20 18:04:56 | INFO | train rank=0 | step=26090/50000 progress=52.2% loss=0.086008 lr=5.052910e-05 lr_backbone=5.052910e-05 elapsed=11:26:56 avg_step=1.58s recent_step=1.22s eta=10:29:32 +2026-09-20 18:05:12 | INFO | train rank=0 | step=26100/50000 progress=52.2% loss=0.098697 lr=5.049603e-05 lr_backbone=5.049603e-05 elapsed=11:27:12 avg_step=1.58s recent_step=1.60s eta=10:29:16 +2026-09-20 18:05:30 | INFO | train rank=0 | step=26110/50000 progress=52.2% loss=0.140684 lr=5.046296e-05 lr_backbone=5.046296e-05 elapsed=11:27:29 avg_step=1.58s recent_step=1.79s eta=10:29:02 +2026-09-20 18:05:42 | INFO | train rank=0 | step=26120/50000 progress=52.2% loss=0.116048 lr=5.042990e-05 lr_backbone=5.042990e-05 elapsed=11:27:41 avg_step=1.58s recent_step=1.21s eta=10:28:43 +2026-09-20 18:06:01 | INFO | train rank=0 | step=26130/50000 progress=52.3% loss=0.108025 lr=5.039683e-05 lr_backbone=5.039683e-05 elapsed=11:28:01 avg_step=1.58s recent_step=1.95s eta=10:28:31 +2026-09-20 18:06:14 | INFO | train rank=0 | step=26140/50000 progress=52.3% loss=0.111172 lr=5.036376e-05 lr_backbone=5.036376e-05 elapsed=11:28:13 avg_step=1.58s recent_step=1.23s eta=10:28:12 +2026-09-20 18:06:30 | INFO | train rank=0 | step=26150/50000 progress=52.3% loss=0.113447 lr=5.033069e-05 lr_backbone=5.033069e-05 elapsed=11:28:30 avg_step=1.58s recent_step=1.64s eta=10:27:56 +2026-09-20 18:06:45 | INFO | train rank=0 | step=26160/50000 progress=52.3% loss=0.093865 lr=5.029762e-05 lr_backbone=5.029762e-05 elapsed=11:28:45 avg_step=1.58s recent_step=1.49s eta=10:27:40 +2026-09-20 18:07:02 | INFO | train rank=0 | step=26170/50000 progress=52.3% loss=0.098278 lr=5.026455e-05 lr_backbone=5.026455e-05 elapsed=11:29:01 avg_step=1.58s recent_step=1.66s eta=10:27:25 +2026-09-20 18:07:16 | INFO | train rank=0 | step=26180/50000 progress=52.4% loss=0.075717 lr=5.023148e-05 lr_backbone=5.023148e-05 elapsed=11:29:15 avg_step=1.58s recent_step=1.43s eta=10:27:07 +2026-09-20 18:07:34 | INFO | train rank=0 | step=26190/50000 progress=52.4% loss=0.101152 lr=5.019842e-05 lr_backbone=5.019842e-05 elapsed=11:29:34 avg_step=1.58s recent_step=1.84s eta=10:26:54 +2026-09-20 18:07:48 | INFO | train rank=0 | step=26200/50000 progress=52.4% loss=0.145833 lr=5.016535e-05 lr_backbone=5.016535e-05 elapsed=11:29:48 avg_step=1.58s recent_step=1.40s eta=10:26:37 +2026-09-20 18:08:06 | INFO | train rank=0 | step=26210/50000 progress=52.4% loss=0.107498 lr=5.013228e-05 lr_backbone=5.013228e-05 elapsed=11:30:06 avg_step=1.58s recent_step=1.80s eta=10:26:23 +2026-09-20 18:08:20 | INFO | train rank=0 | step=26220/50000 progress=52.4% loss=0.069606 lr=5.009921e-05 lr_backbone=5.009921e-05 elapsed=11:30:19 avg_step=1.58s recent_step=1.36s eta=10:26:05 +2026-09-20 18:08:36 | INFO | train rank=0 | step=26230/50000 progress=52.5% loss=0.091799 lr=5.006614e-05 lr_backbone=5.006614e-05 elapsed=11:30:36 avg_step=1.58s recent_step=1.65s eta=10:25:50 +2026-09-20 18:08:53 | INFO | train rank=0 | step=26240/50000 progress=52.5% loss=0.095337 lr=5.003307e-05 lr_backbone=5.003307e-05 elapsed=11:30:53 avg_step=1.58s recent_step=1.68s eta=10:25:35 +2026-09-20 18:09:06 | INFO | train rank=0 | step=26250/50000 progress=52.5% loss=0.079693 lr=5.000000e-05 lr_backbone=5.000000e-05 elapsed=11:31:06 avg_step=1.58s recent_step=1.31s eta=10:25:17 +2026-09-20 18:09:24 | INFO | train rank=0 | step=26260/50000 progress=52.5% loss=0.089910 lr=4.996693e-05 lr_backbone=4.996693e-05 elapsed=11:31:24 avg_step=1.58s recent_step=1.79s eta=10:25:03 +2026-09-20 18:09:38 | INFO | train rank=0 | step=26270/50000 progress=52.5% loss=0.078133 lr=4.993386e-05 lr_backbone=4.993386e-05 elapsed=11:31:37 avg_step=1.58s recent_step=1.35s eta=10:24:45 +2026-09-20 18:09:57 | INFO | train rank=0 | step=26280/50000 progress=52.6% loss=0.112778 lr=4.990079e-05 lr_backbone=4.990079e-05 elapsed=11:31:57 avg_step=1.58s recent_step=1.94s eta=10:24:32 +2026-09-20 18:10:11 | INFO | train rank=0 | step=26290/50000 progress=52.6% loss=0.091874 lr=4.986772e-05 lr_backbone=4.986772e-05 elapsed=11:32:11 avg_step=1.58s recent_step=1.41s eta=10:24:15 +2026-09-20 18:10:28 | INFO | train rank=0 | step=26300/50000 progress=52.6% loss=0.094900 lr=4.983465e-05 lr_backbone=4.983465e-05 elapsed=11:32:28 avg_step=1.58s recent_step=1.74s eta=10:24:01 +2026-09-20 18:10:46 | INFO | train rank=0 | step=26310/50000 progress=52.6% loss=0.096408 lr=4.980158e-05 lr_backbone=4.980158e-05 elapsed=11:32:46 avg_step=1.58s recent_step=1.79s eta=10:23:47 +2026-09-20 18:10:58 | INFO | train rank=0 | step=26320/50000 progress=52.6% loss=0.078897 lr=4.976852e-05 lr_backbone=4.976852e-05 elapsed=11:32:58 avg_step=1.58s recent_step=1.20s eta=10:23:28 +2026-09-20 18:11:11 | INFO | train rank=0 | step=26330/50000 progress=52.7% loss=0.159831 lr=4.973545e-05 lr_backbone=4.973545e-05 elapsed=11:33:10 avg_step=1.58s recent_step=1.21s eta=10:23:08 +2026-09-20 18:11:35 | INFO | train rank=0 | step=26340/50000 progress=52.7% loss=0.103769 lr=4.970238e-05 lr_backbone=4.970238e-05 elapsed=11:33:35 avg_step=1.58s recent_step=2.46s eta=10:23:01 +2026-09-20 18:11:48 | INFO | train rank=0 | step=26350/50000 progress=52.7% loss=0.074433 lr=4.966931e-05 lr_backbone=4.966931e-05 elapsed=11:33:48 avg_step=1.58s recent_step=1.32s eta=10:22:42 +2026-09-20 18:12:00 | INFO | train rank=0 | step=26360/50000 progress=52.7% loss=0.081515 lr=4.963624e-05 lr_backbone=4.963624e-05 elapsed=11:34:00 avg_step=1.58s recent_step=1.21s eta=10:22:23 +2026-09-20 18:12:12 | INFO | train rank=0 | step=26370/50000 progress=52.7% loss=0.088326 lr=4.960317e-05 lr_backbone=4.960317e-05 elapsed=11:34:12 avg_step=1.58s recent_step=1.20s eta=10:22:04 +2026-09-20 18:12:30 | INFO | train rank=0 | step=26380/50000 progress=52.8% loss=0.095069 lr=4.957010e-05 lr_backbone=4.957010e-05 elapsed=11:34:30 avg_step=1.58s recent_step=1.76s eta=10:21:50 +2026-09-20 18:12:51 | INFO | train rank=0 | step=26390/50000 progress=52.8% loss=0.090216 lr=4.953704e-05 lr_backbone=4.953704e-05 elapsed=11:34:50 avg_step=1.58s recent_step=2.06s eta=10:21:38 +2026-09-20 18:13:04 | INFO | train rank=0 | step=26400/50000 progress=52.8% loss=0.115175 lr=4.950397e-05 lr_backbone=4.950397e-05 elapsed=11:35:04 avg_step=1.58s recent_step=1.33s eta=10:21:20 +2026-09-20 18:13:21 | INFO | train rank=0 | step=26410/50000 progress=52.8% loss=0.092946 lr=4.947090e-05 lr_backbone=4.947090e-05 elapsed=11:35:21 avg_step=1.58s recent_step=1.72s eta=10:21:06 +2026-09-20 18:13:33 | INFO | train rank=0 | step=26420/50000 progress=52.8% loss=0.115725 lr=4.943783e-05 lr_backbone=4.943783e-05 elapsed=11:35:33 avg_step=1.58s recent_step=1.23s eta=10:20:47 +2026-09-20 18:13:50 | INFO | train rank=0 | step=26430/50000 progress=52.9% loss=0.071016 lr=4.940476e-05 lr_backbone=4.940476e-05 elapsed=11:35:49 avg_step=1.58s recent_step=1.64s eta=10:20:32 +2026-09-20 18:14:06 | INFO | train rank=0 | step=26440/50000 progress=52.9% loss=0.109702 lr=4.937170e-05 lr_backbone=4.937170e-05 elapsed=11:36:06 avg_step=1.58s recent_step=1.62s eta=10:20:16 +2026-09-20 18:14:23 | INFO | train rank=0 | step=26450/50000 progress=52.9% loss=0.068777 lr=4.933863e-05 lr_backbone=4.933863e-05 elapsed=11:36:23 avg_step=1.58s recent_step=1.70s eta=10:20:01 +2026-09-20 18:14:36 | INFO | train rank=0 | step=26460/50000 progress=52.9% loss=0.084181 lr=4.930557e-05 lr_backbone=4.930557e-05 elapsed=11:36:35 avg_step=1.58s recent_step=1.27s eta=10:19:43 +2026-09-20 18:14:54 | INFO | train rank=0 | step=26470/50000 progress=52.9% loss=0.104282 lr=4.927250e-05 lr_backbone=4.927250e-05 elapsed=11:36:54 avg_step=1.58s recent_step=1.88s eta=10:19:30 +2026-09-20 18:15:07 | INFO | train rank=0 | step=26480/50000 progress=53.0% loss=0.087172 lr=4.923943e-05 lr_backbone=4.923943e-05 elapsed=11:37:06 avg_step=1.58s recent_step=1.22s eta=10:19:11 +2026-09-20 18:15:23 | INFO | train rank=0 | step=26490/50000 progress=53.0% loss=0.105299 lr=4.920637e-05 lr_backbone=4.920637e-05 elapsed=11:37:22 avg_step=1.58s recent_step=1.58s eta=10:18:55 +2026-09-20 18:15:42 | INFO | train rank=0 | step=26500/50000 progress=53.0% loss=0.083336 lr=4.917330e-05 lr_backbone=4.917330e-05 elapsed=11:37:42 avg_step=1.58s recent_step=1.97s eta=10:18:43 +2026-09-20 18:15:54 | INFO | train rank=0 | step=26510/50000 progress=53.0% loss=0.074972 lr=4.914024e-05 lr_backbone=4.914024e-05 elapsed=11:37:54 avg_step=1.58s recent_step=1.21s eta=10:18:24 +2026-09-20 18:16:11 | INFO | train rank=0 | step=26520/50000 progress=53.0% loss=0.093699 lr=4.910717e-05 lr_backbone=4.910717e-05 elapsed=11:38:10 avg_step=1.58s recent_step=1.65s eta=10:18:08 +2026-09-20 18:16:30 | INFO | train rank=0 | step=26530/50000 progress=53.1% loss=0.103319 lr=4.907411e-05 lr_backbone=4.907411e-05 elapsed=11:38:29 avg_step=1.58s recent_step=1.91s eta=10:17:55 +2026-09-20 18:16:43 | INFO | train rank=0 | step=26540/50000 progress=53.1% loss=0.104885 lr=4.904105e-05 lr_backbone=4.904105e-05 elapsed=11:38:42 avg_step=1.58s recent_step=1.29s eta=10:17:37 +2026-09-20 18:17:03 | INFO | train rank=0 | step=26550/50000 progress=53.1% loss=0.086062 lr=4.900798e-05 lr_backbone=4.900798e-05 elapsed=11:39:02 avg_step=1.58s recent_step=2.00s eta=10:17:25 +2026-09-20 18:17:15 | INFO | train rank=0 | step=26560/50000 progress=53.1% loss=0.107427 lr=4.897492e-05 lr_backbone=4.897492e-05 elapsed=11:39:14 avg_step=1.58s recent_step=1.20s eta=10:17:06 +2026-09-20 18:17:31 | INFO | train rank=0 | step=26570/50000 progress=53.1% loss=0.092709 lr=4.894186e-05 lr_backbone=4.894186e-05 elapsed=11:39:31 avg_step=1.58s recent_step=1.64s eta=10:16:51 +2026-09-20 18:17:48 | INFO | train rank=0 | step=26580/50000 progress=53.2% loss=0.090009 lr=4.890880e-05 lr_backbone=4.890880e-05 elapsed=11:39:48 avg_step=1.58s recent_step=1.71s eta=10:16:36 +2026-09-20 18:18:08 | INFO | train rank=0 | step=26590/50000 progress=53.2% loss=0.079448 lr=4.887574e-05 lr_backbone=4.887574e-05 elapsed=11:40:08 avg_step=1.58s recent_step=2.01s eta=10:16:24 +2026-09-20 18:18:21 | INFO | train rank=0 | step=26600/50000 progress=53.2% loss=0.101220 lr=4.884267e-05 lr_backbone=4.884267e-05 elapsed=11:40:20 avg_step=1.58s recent_step=1.23s eta=10:16:05 +2026-09-20 18:18:32 | INFO | train rank=0 | step=26610/50000 progress=53.2% loss=0.097971 lr=4.880961e-05 lr_backbone=4.880961e-05 elapsed=11:40:32 avg_step=1.58s recent_step=1.18s eta=10:15:46 +2026-09-20 18:18:45 | INFO | train rank=0 | step=26620/50000 progress=53.2% loss=0.078786 lr=4.877655e-05 lr_backbone=4.877655e-05 elapsed=11:40:45 avg_step=1.58s recent_step=1.26s eta=10:15:27 +2026-09-20 18:19:01 | INFO | train rank=0 | step=26630/50000 progress=53.3% loss=0.107314 lr=4.874350e-05 lr_backbone=4.874350e-05 elapsed=11:41:00 avg_step=1.58s recent_step=1.58s eta=10:15:11 +2026-09-20 18:19:18 | INFO | train rank=0 | step=26640/50000 progress=53.3% loss=0.080611 lr=4.871044e-05 lr_backbone=4.871044e-05 elapsed=11:41:17 avg_step=1.58s recent_step=1.68s eta=10:14:56 +2026-09-20 18:19:39 | INFO | train rank=0 | step=26650/50000 progress=53.3% loss=0.109651 lr=4.867738e-05 lr_backbone=4.867738e-05 elapsed=11:41:39 avg_step=1.58s recent_step=2.17s eta=10:14:46 +2026-09-20 18:19:51 | INFO | train rank=0 | step=26660/50000 progress=53.3% loss=0.096987 lr=4.864432e-05 lr_backbone=4.864432e-05 elapsed=11:41:51 avg_step=1.58s recent_step=1.18s eta=10:14:27 +2026-09-20 18:20:09 | INFO | train rank=0 | step=26670/50000 progress=53.3% loss=0.099369 lr=4.861126e-05 lr_backbone=4.861126e-05 elapsed=11:42:09 avg_step=1.58s recent_step=1.84s eta=10:14:13 +2026-09-20 18:20:22 | INFO | train rank=0 | step=26680/50000 progress=53.4% loss=0.079628 lr=4.857821e-05 lr_backbone=4.857821e-05 elapsed=11:42:22 avg_step=1.58s recent_step=1.26s eta=10:13:54 +2026-09-20 18:20:37 | INFO | train rank=0 | step=26690/50000 progress=53.4% loss=0.113695 lr=4.854515e-05 lr_backbone=4.854515e-05 elapsed=11:42:37 avg_step=1.58s recent_step=1.49s eta=10:13:38 +2026-09-20 18:20:56 | INFO | train rank=0 | step=26700/50000 progress=53.4% loss=0.077960 lr=4.851210e-05 lr_backbone=4.851210e-05 elapsed=11:42:56 avg_step=1.58s recent_step=1.95s eta=10:13:25 +2026-09-20 18:21:11 | INFO | train rank=0 | step=26710/50000 progress=53.4% loss=0.085866 lr=4.847904e-05 lr_backbone=4.847904e-05 elapsed=11:43:11 avg_step=1.58s recent_step=1.49s eta=10:13:09 +2026-09-20 18:21:24 | INFO | train rank=0 | step=26720/50000 progress=53.4% loss=0.099923 lr=4.844599e-05 lr_backbone=4.844599e-05 elapsed=11:43:24 avg_step=1.58s recent_step=1.28s eta=10:12:50 +2026-09-20 18:21:42 | INFO | train rank=0 | step=26730/50000 progress=53.5% loss=0.084255 lr=4.841294e-05 lr_backbone=4.841294e-05 elapsed=11:43:42 avg_step=1.58s recent_step=1.82s eta=10:12:37 +2026-09-20 18:21:59 | INFO | train rank=0 | step=26740/50000 progress=53.5% loss=0.106130 lr=4.837988e-05 lr_backbone=4.837988e-05 elapsed=11:43:59 avg_step=1.58s recent_step=1.69s eta=10:12:22 +2026-09-20 18:22:16 | INFO | train rank=0 | step=26750/50000 progress=53.5% loss=0.086432 lr=4.834683e-05 lr_backbone=4.834683e-05 elapsed=11:44:15 avg_step=1.58s recent_step=1.62s eta=10:12:06 +2026-09-20 18:22:28 | INFO | train rank=0 | step=26760/50000 progress=53.5% loss=0.105098 lr=4.831378e-05 lr_backbone=4.831378e-05 elapsed=11:44:28 avg_step=1.58s recent_step=1.25s eta=10:11:48 +2026-09-20 18:22:41 | INFO | train rank=0 | step=26770/50000 progress=53.5% loss=0.072853 lr=4.828073e-05 lr_backbone=4.828073e-05 elapsed=11:44:41 avg_step=1.58s recent_step=1.29s eta=10:11:29 +2026-09-20 18:22:59 | INFO | train rank=0 | step=26780/50000 progress=53.6% loss=0.115102 lr=4.824768e-05 lr_backbone=4.824768e-05 elapsed=11:44:59 avg_step=1.58s recent_step=1.85s eta=10:11:16 +2026-09-20 18:23:13 | INFO | train rank=0 | step=26790/50000 progress=53.6% loss=0.168453 lr=4.821463e-05 lr_backbone=4.821463e-05 elapsed=11:45:13 avg_step=1.58s recent_step=1.37s eta=10:10:58 +2026-09-20 18:23:35 | INFO | train rank=0 | step=26800/50000 progress=53.6% loss=0.123044 lr=4.818158e-05 lr_backbone=4.818158e-05 elapsed=11:45:34 avg_step=1.58s recent_step=2.14s eta=10:10:47 +2026-09-20 18:23:46 | INFO | train rank=0 | step=26810/50000 progress=53.6% loss=0.140919 lr=4.814854e-05 lr_backbone=4.814854e-05 elapsed=11:45:46 avg_step=1.58s recent_step=1.18s eta=10:10:28 +2026-09-20 18:23:59 | INFO | train rank=0 | step=26820/50000 progress=53.6% loss=0.085545 lr=4.811549e-05 lr_backbone=4.811549e-05 elapsed=11:45:58 avg_step=1.58s recent_step=1.24s eta=10:10:09 +2026-09-20 18:24:23 | INFO | train rank=0 | step=26830/50000 progress=53.7% loss=0.090435 lr=4.808245e-05 lr_backbone=4.808245e-05 elapsed=11:46:23 avg_step=1.58s recent_step=2.44s eta=10:10:01 +2026-09-20 18:24:35 | INFO | train rank=0 | step=26840/50000 progress=53.7% loss=0.079439 lr=4.804940e-05 lr_backbone=4.804940e-05 elapsed=11:46:35 avg_step=1.58s recent_step=1.21s eta=10:09:42 +2026-09-20 18:24:55 | INFO | train rank=0 | step=26850/50000 progress=53.7% loss=0.090157 lr=4.801636e-05 lr_backbone=4.801636e-05 elapsed=11:46:55 avg_step=1.58s recent_step=2.01s eta=10:09:30 +2026-09-20 18:25:07 | INFO | train rank=0 | step=26860/50000 progress=53.7% loss=0.096999 lr=4.798331e-05 lr_backbone=4.798331e-05 elapsed=11:47:06 avg_step=1.58s recent_step=1.14s eta=10:09:10 +2026-09-20 18:25:19 | INFO | train rank=0 | step=26870/50000 progress=53.7% loss=0.087433 lr=4.795027e-05 lr_backbone=4.795027e-05 elapsed=11:47:19 avg_step=1.58s recent_step=1.24s eta=10:08:52 +2026-09-20 18:25:37 | INFO | train rank=0 | step=26880/50000 progress=53.8% loss=0.079274 lr=4.791723e-05 lr_backbone=4.791723e-05 elapsed=11:47:37 avg_step=1.58s recent_step=1.78s eta=10:08:38 +2026-09-20 18:25:49 | INFO | train rank=0 | step=26890/50000 progress=53.8% loss=0.072801 lr=4.788419e-05 lr_backbone=4.788419e-05 elapsed=11:47:49 avg_step=1.58s recent_step=1.21s eta=10:08:19 +2026-09-20 18:26:05 | INFO | train rank=0 | step=26900/50000 progress=53.8% loss=0.084642 lr=4.785115e-05 lr_backbone=4.785115e-05 elapsed=11:48:05 avg_step=1.58s recent_step=1.63s eta=10:08:03 +2026-09-20 18:26:28 | INFO | train rank=0 | step=26910/50000 progress=53.8% loss=0.083634 lr=4.781811e-05 lr_backbone=4.781811e-05 elapsed=11:48:28 avg_step=1.58s recent_step=2.27s eta=10:07:53 +2026-09-20 18:26:40 | INFO | train rank=0 | step=26920/50000 progress=53.8% loss=0.123866 lr=4.778508e-05 lr_backbone=4.778508e-05 elapsed=11:48:40 avg_step=1.58s recent_step=1.21s eta=10:07:34 +2026-09-20 18:26:54 | INFO | train rank=0 | step=26930/50000 progress=53.9% loss=0.090392 lr=4.775204e-05 lr_backbone=4.775204e-05 elapsed=11:48:54 avg_step=1.58s recent_step=1.40s eta=10:07:17 +2026-09-20 18:27:08 | INFO | train rank=0 | step=26940/50000 progress=53.9% loss=0.097691 lr=4.771900e-05 lr_backbone=4.771900e-05 elapsed=11:49:08 avg_step=1.58s recent_step=1.42s eta=10:07:00 +2026-09-20 18:27:29 | INFO | train rank=0 | step=26950/50000 progress=53.9% loss=0.100964 lr=4.768597e-05 lr_backbone=4.768597e-05 elapsed=11:49:28 avg_step=1.58s recent_step=2.03s eta=10:06:48 +2026-09-20 18:27:41 | INFO | train rank=0 | step=26960/50000 progress=53.9% loss=0.117682 lr=4.765294e-05 lr_backbone=4.765294e-05 elapsed=11:49:41 avg_step=1.58s recent_step=1.28s eta=10:06:30 +2026-09-20 18:27:57 | INFO | train rank=0 | step=26970/50000 progress=53.9% loss=0.092030 lr=4.761990e-05 lr_backbone=4.761990e-05 elapsed=11:49:57 avg_step=1.58s recent_step=1.60s eta=10:06:14 +2026-09-20 18:28:12 | INFO | train rank=0 | step=26980/50000 progress=54.0% loss=0.091688 lr=4.758687e-05 lr_backbone=4.758687e-05 elapsed=11:50:11 avg_step=1.58s recent_step=1.42s eta=10:05:57 +2026-09-20 18:28:28 | INFO | train rank=0 | step=26990/50000 progress=54.0% loss=0.070281 lr=4.755384e-05 lr_backbone=4.755384e-05 elapsed=11:50:28 avg_step=1.58s recent_step=1.63s eta=10:05:42 +2026-09-20 18:28:42 | INFO | train rank=0 | step=27000/50000 progress=54.0% loss=0.096695 lr=4.752081e-05 lr_backbone=4.752081e-05 elapsed=11:50:41 avg_step=1.58s recent_step=1.36s eta=10:05:24 +2026-09-20 18:29:00 | INFO | train rank=0 | step=27010/50000 progress=54.0% loss=0.141322 lr=4.748778e-05 lr_backbone=4.748778e-05 elapsed=11:50:59 avg_step=1.58s recent_step=1.82s eta=10:05:10 +2026-09-20 18:29:19 | INFO | train rank=0 | step=27020/50000 progress=54.0% loss=0.066578 lr=4.745476e-05 lr_backbone=4.745476e-05 elapsed=11:51:18 avg_step=1.58s recent_step=1.88s eta=10:04:57 +2026-09-20 18:29:31 | INFO | train rank=0 | step=27030/50000 progress=54.1% loss=0.105319 lr=4.742173e-05 lr_backbone=4.742173e-05 elapsed=11:51:31 avg_step=1.58s recent_step=1.29s eta=10:04:39 +2026-09-20 18:29:51 | INFO | train rank=0 | step=27040/50000 progress=54.1% loss=0.072935 lr=4.738871e-05 lr_backbone=4.738871e-05 elapsed=11:51:50 avg_step=1.58s recent_step=1.91s eta=10:04:26 +2026-09-20 18:30:02 | INFO | train rank=0 | step=27050/50000 progress=54.1% loss=0.096399 lr=4.735568e-05 lr_backbone=4.735568e-05 elapsed=11:52:02 avg_step=1.58s recent_step=1.16s eta=10:04:06 +2026-09-20 18:30:22 | INFO | train rank=0 | step=27060/50000 progress=54.1% loss=0.101723 lr=4.732266e-05 lr_backbone=4.732266e-05 elapsed=11:52:21 avg_step=1.58s recent_step=1.94s eta=10:03:54 +2026-09-20 18:30:38 | INFO | train rank=0 | step=27070/50000 progress=54.1% loss=0.094375 lr=4.728964e-05 lr_backbone=4.728964e-05 elapsed=11:52:38 avg_step=1.58s recent_step=1.66s eta=10:03:38 +2026-09-20 18:30:51 | INFO | train rank=0 | step=27080/50000 progress=54.2% loss=0.105566 lr=4.725662e-05 lr_backbone=4.725662e-05 elapsed=11:52:50 avg_step=1.58s recent_step=1.27s eta=10:03:20 +2026-09-20 18:31:09 | INFO | train rank=0 | step=27090/50000 progress=54.2% loss=0.090716 lr=4.722360e-05 lr_backbone=4.722360e-05 elapsed=11:53:09 avg_step=1.58s recent_step=1.86s eta=10:03:07 +2026-09-20 18:31:21 | INFO | train rank=0 | step=27100/50000 progress=54.2% loss=0.105440 lr=4.719058e-05 lr_backbone=4.719058e-05 elapsed=11:53:21 avg_step=1.58s recent_step=1.21s eta=10:02:48 +2026-09-20 18:31:42 | INFO | train rank=0 | step=27110/50000 progress=54.2% loss=0.089105 lr=4.715757e-05 lr_backbone=4.715757e-05 elapsed=11:53:41 avg_step=1.58s recent_step=2.02s eta=10:02:36 +2026-09-20 18:31:54 | INFO | train rank=0 | step=27120/50000 progress=54.2% loss=0.096247 lr=4.712455e-05 lr_backbone=4.712455e-05 elapsed=11:53:53 avg_step=1.58s recent_step=1.20s eta=10:02:17 +2026-09-20 18:32:06 | INFO | train rank=0 | step=27130/50000 progress=54.3% loss=0.093509 lr=4.709154e-05 lr_backbone=4.709154e-05 elapsed=11:54:06 avg_step=1.58s recent_step=1.23s eta=10:01:58 +2026-09-20 18:32:25 | INFO | train rank=0 | step=27140/50000 progress=54.3% loss=0.092391 lr=4.705852e-05 lr_backbone=4.705852e-05 elapsed=11:54:25 avg_step=1.58s recent_step=1.94s eta=10:01:45 +2026-09-20 18:32:43 | INFO | train rank=0 | step=27150/50000 progress=54.3% loss=0.072102 lr=4.702551e-05 lr_backbone=4.702551e-05 elapsed=11:54:42 avg_step=1.58s recent_step=1.72s eta=10:01:30 +2026-09-20 18:32:54 | INFO | train rank=0 | step=27160/50000 progress=54.3% loss=0.095176 lr=4.699250e-05 lr_backbone=4.699250e-05 elapsed=11:54:54 avg_step=1.58s recent_step=1.17s eta=10:01:11 +2026-09-20 18:33:12 | INFO | train rank=0 | step=27170/50000 progress=54.3% loss=0.083740 lr=4.695949e-05 lr_backbone=4.695949e-05 elapsed=11:55:11 avg_step=1.58s recent_step=1.73s eta=10:00:57 +2026-09-20 18:33:26 | INFO | train rank=0 | step=27180/50000 progress=54.4% loss=0.087887 lr=4.692649e-05 lr_backbone=4.692649e-05 elapsed=11:55:25 avg_step=1.58s recent_step=1.40s eta=10:00:39 +2026-09-20 18:33:42 | INFO | train rank=0 | step=27190/50000 progress=54.4% loss=0.084269 lr=4.689348e-05 lr_backbone=4.689348e-05 elapsed=11:55:42 avg_step=1.58s recent_step=1.67s eta=10:00:24 +2026-09-20 18:33:56 | INFO | train rank=0 | step=27200/50000 progress=54.4% loss=0.082117 lr=4.686047e-05 lr_backbone=4.686047e-05 elapsed=11:55:56 avg_step=1.58s recent_step=1.40s eta=10:00:07 +2026-09-20 18:34:16 | INFO | train rank=0 | step=27210/50000 progress=54.4% loss=0.101943 lr=4.682747e-05 lr_backbone=4.682747e-05 elapsed=11:56:15 avg_step=1.58s recent_step=1.94s eta=09:59:54 +2026-09-20 18:34:28 | INFO | train rank=0 | step=27220/50000 progress=54.4% loss=0.099283 lr=4.679447e-05 lr_backbone=4.679447e-05 elapsed=11:56:28 avg_step=1.58s recent_step=1.28s eta=09:59:36 +2026-09-20 18:34:43 | INFO | train rank=0 | step=27230/50000 progress=54.5% loss=0.099205 lr=4.676147e-05 lr_backbone=4.676147e-05 elapsed=11:56:42 avg_step=1.58s recent_step=1.42s eta=09:59:19 +2026-09-20 18:35:01 | INFO | train rank=0 | step=27240/50000 progress=54.5% loss=0.130340 lr=4.672847e-05 lr_backbone=4.672847e-05 elapsed=11:57:00 avg_step=1.58s recent_step=1.80s eta=09:59:05 +2026-09-20 18:35:18 | INFO | train rank=0 | step=27250/50000 progress=54.5% loss=0.088288 lr=4.669547e-05 lr_backbone=4.669547e-05 elapsed=11:57:17 avg_step=1.58s recent_step=1.69s eta=09:58:50 +2026-09-20 18:35:31 | INFO | train rank=0 | step=27260/50000 progress=54.5% loss=0.093129 lr=4.666247e-05 lr_backbone=4.666247e-05 elapsed=11:57:30 avg_step=1.58s recent_step=1.31s eta=09:58:32 +2026-09-20 18:35:46 | INFO | train rank=0 | step=27270/50000 progress=54.5% loss=0.073994 lr=4.662948e-05 lr_backbone=4.662948e-05 elapsed=11:57:46 avg_step=1.58s recent_step=1.54s eta=09:58:16 +2026-09-20 18:36:04 | INFO | train rank=0 | step=27280/50000 progress=54.6% loss=0.102783 lr=4.659649e-05 lr_backbone=4.659649e-05 elapsed=11:58:04 avg_step=1.58s recent_step=1.82s eta=09:58:02 +2026-09-20 18:36:17 | INFO | train rank=0 | step=27290/50000 progress=54.6% loss=0.091960 lr=4.656349e-05 lr_backbone=4.656349e-05 elapsed=11:58:16 avg_step=1.58s recent_step=1.23s eta=09:57:43 +2026-09-20 18:36:29 | INFO | train rank=0 | step=27300/50000 progress=54.6% loss=0.091667 lr=4.653050e-05 lr_backbone=4.653050e-05 elapsed=11:58:29 avg_step=1.58s recent_step=1.30s eta=09:57:25 +2026-09-20 18:36:43 | INFO | train rank=0 | step=27310/50000 progress=54.6% loss=0.122510 lr=4.649751e-05 lr_backbone=4.649751e-05 elapsed=11:58:42 avg_step=1.58s recent_step=1.32s eta=09:57:07 +2026-09-20 18:37:03 | INFO | train rank=0 | step=27320/50000 progress=54.6% loss=0.099461 lr=4.646453e-05 lr_backbone=4.646453e-05 elapsed=11:59:02 avg_step=1.58s recent_step=2.00s eta=09:56:55 +2026-09-20 18:37:27 | INFO | train rank=0 | step=27330/50000 progress=54.7% loss=0.136466 lr=4.643154e-05 lr_backbone=4.643154e-05 elapsed=11:59:27 avg_step=1.58s recent_step=2.42s eta=09:56:46 +2026-09-20 18:37:39 | INFO | train rank=0 | step=27340/50000 progress=54.7% loss=0.078261 lr=4.639856e-05 lr_backbone=4.639856e-05 elapsed=11:59:39 avg_step=1.58s recent_step=1.21s eta=09:56:27 +2026-09-20 18:37:51 | INFO | train rank=0 | step=27350/50000 progress=54.7% loss=0.097026 lr=4.636557e-05 lr_backbone=4.636557e-05 elapsed=11:59:51 avg_step=1.58s recent_step=1.20s eta=09:56:08 +2026-09-20 18:38:07 | INFO | train rank=0 | step=27360/50000 progress=54.7% loss=0.079124 lr=4.633259e-05 lr_backbone=4.633259e-05 elapsed=12:00:06 avg_step=1.58s recent_step=1.57s eta=09:55:53 +2026-09-20 18:38:23 | INFO | train rank=0 | step=27370/50000 progress=54.7% loss=0.088573 lr=4.629961e-05 lr_backbone=4.629961e-05 elapsed=12:00:23 avg_step=1.58s recent_step=1.63s eta=09:55:37 +2026-09-20 18:38:38 | INFO | train rank=0 | step=27380/50000 progress=54.8% loss=0.134939 lr=4.626664e-05 lr_backbone=4.626664e-05 elapsed=12:00:38 avg_step=1.58s recent_step=1.49s eta=09:55:21 +2026-09-20 18:38:54 | INFO | train rank=0 | step=27390/50000 progress=54.8% loss=0.079366 lr=4.623366e-05 lr_backbone=4.623366e-05 elapsed=12:00:54 avg_step=1.58s recent_step=1.62s eta=09:55:05 +2026-09-20 18:39:15 | INFO | train rank=0 | step=27400/50000 progress=54.8% loss=0.085671 lr=4.620069e-05 lr_backbone=4.620069e-05 elapsed=12:01:15 avg_step=1.58s recent_step=2.08s eta=09:54:54 +2026-09-20 18:39:27 | INFO | train rank=0 | step=27410/50000 progress=54.8% loss=0.084178 lr=4.616771e-05 lr_backbone=4.616771e-05 elapsed=12:01:27 avg_step=1.58s recent_step=1.22s eta=09:54:35 +2026-09-20 18:39:42 | INFO | train rank=0 | step=27420/50000 progress=54.8% loss=0.083732 lr=4.613474e-05 lr_backbone=4.613474e-05 elapsed=12:01:41 avg_step=1.58s recent_step=1.45s eta=09:54:18 +2026-09-20 18:39:57 | INFO | train rank=0 | step=27430/50000 progress=54.9% loss=0.097056 lr=4.610177e-05 lr_backbone=4.610177e-05 elapsed=12:01:57 avg_step=1.58s recent_step=1.58s eta=09:54:02 +2026-09-20 18:40:16 | INFO | train rank=0 | step=27440/50000 progress=54.9% loss=0.100140 lr=4.606880e-05 lr_backbone=4.606880e-05 elapsed=12:02:16 avg_step=1.58s recent_step=1.86s eta=09:53:49 +2026-09-20 18:40:30 | INFO | train rank=0 | step=27450/50000 progress=54.9% loss=0.107497 lr=4.603584e-05 lr_backbone=4.603584e-05 elapsed=12:02:30 avg_step=1.58s recent_step=1.42s eta=09:53:31 +2026-09-20 18:40:43 | INFO | train rank=0 | step=27460/50000 progress=54.9% loss=0.089082 lr=4.600287e-05 lr_backbone=4.600287e-05 elapsed=12:02:43 avg_step=1.58s recent_step=1.31s eta=09:53:13 +2026-09-20 18:41:00 | INFO | train rank=0 | step=27470/50000 progress=54.9% loss=0.081821 lr=4.596991e-05 lr_backbone=4.596991e-05 elapsed=12:03:00 avg_step=1.58s recent_step=1.69s eta=09:52:59 +2026-09-20 18:41:16 | INFO | train rank=0 | step=27480/50000 progress=55.0% loss=0.076886 lr=4.593695e-05 lr_backbone=4.593695e-05 elapsed=12:03:16 avg_step=1.58s recent_step=1.59s eta=09:52:43 +2026-09-20 18:41:30 | INFO | train rank=0 | step=27490/50000 progress=55.0% loss=0.066399 lr=4.590399e-05 lr_backbone=4.590399e-05 elapsed=12:03:29 avg_step=1.58s recent_step=1.35s eta=09:52:25 +2026-09-20 18:41:53 | INFO | train rank=0 | step=27500/50000 progress=55.0% loss=0.109403 lr=4.587103e-05 lr_backbone=4.587103e-05 elapsed=12:03:53 avg_step=1.58s recent_step=2.37s eta=09:52:16 +2026-09-20 18:42:05 | INFO | train rank=0 | step=27510/50000 progress=55.0% loss=0.086335 lr=4.583808e-05 lr_backbone=4.583808e-05 elapsed=12:04:05 avg_step=1.58s recent_step=1.16s eta=09:51:57 +2026-09-20 18:42:17 | INFO | train rank=0 | step=27520/50000 progress=55.0% loss=0.094942 lr=4.580512e-05 lr_backbone=4.580512e-05 elapsed=12:04:17 avg_step=1.58s recent_step=1.23s eta=09:51:38 +2026-09-20 18:42:34 | INFO | train rank=0 | step=27530/50000 progress=55.1% loss=0.077538 lr=4.577217e-05 lr_backbone=4.577217e-05 elapsed=12:04:33 avg_step=1.58s recent_step=1.63s eta=09:51:23 +2026-09-20 18:42:51 | INFO | train rank=0 | step=27540/50000 progress=55.1% loss=0.106763 lr=4.573922e-05 lr_backbone=4.573922e-05 elapsed=12:04:50 avg_step=1.58s recent_step=1.71s eta=09:51:08 +2026-09-20 18:43:10 | INFO | train rank=0 | step=27550/50000 progress=55.1% loss=0.090357 lr=4.570627e-05 lr_backbone=4.570627e-05 elapsed=12:05:10 avg_step=1.58s recent_step=1.93s eta=09:50:55 +2026-09-20 18:43:27 | INFO | train rank=0 | step=27560/50000 progress=55.1% loss=0.084170 lr=4.567333e-05 lr_backbone=4.567333e-05 elapsed=12:05:26 avg_step=1.58s recent_step=1.65s eta=09:50:40 +2026-09-20 18:43:39 | INFO | train rank=0 | step=27570/50000 progress=55.1% loss=0.098121 lr=4.564038e-05 lr_backbone=4.564038e-05 elapsed=12:05:38 avg_step=1.58s recent_step=1.24s eta=09:50:21 +2026-09-20 18:43:53 | INFO | train rank=0 | step=27580/50000 progress=55.2% loss=0.104393 lr=4.560744e-05 lr_backbone=4.560744e-05 elapsed=12:05:53 avg_step=1.58s recent_step=1.45s eta=09:50:05 +2026-09-20 18:44:12 | INFO | train rank=0 | step=27590/50000 progress=55.2% loss=0.087208 lr=4.557450e-05 lr_backbone=4.557450e-05 elapsed=12:06:11 avg_step=1.58s recent_step=1.82s eta=09:49:51 +2026-09-20 18:44:26 | INFO | train rank=0 | step=27600/50000 progress=55.2% loss=0.094456 lr=4.554156e-05 lr_backbone=4.554156e-05 elapsed=12:06:26 avg_step=1.58s recent_step=1.47s eta=09:49:34 +2026-09-20 18:44:45 | INFO | train rank=0 | step=27610/50000 progress=55.2% loss=0.074867 lr=4.550862e-05 lr_backbone=4.550862e-05 elapsed=12:06:44 avg_step=1.58s recent_step=1.82s eta=09:49:20 +2026-09-20 18:45:01 | INFO | train rank=0 | step=27620/50000 progress=55.2% loss=0.085534 lr=4.547569e-05 lr_backbone=4.547569e-05 elapsed=12:07:01 avg_step=1.58s recent_step=1.67s eta=09:49:05 +2026-09-20 18:45:14 | INFO | train rank=0 | step=27630/50000 progress=55.3% loss=0.075160 lr=4.544276e-05 lr_backbone=4.544276e-05 elapsed=12:07:13 avg_step=1.58s recent_step=1.26s eta=09:48:47 +2026-09-20 18:45:28 | INFO | train rank=0 | step=27640/50000 progress=55.3% loss=0.096145 lr=4.540983e-05 lr_backbone=4.540983e-05 elapsed=12:07:27 avg_step=1.58s recent_step=1.39s eta=09:48:29 +2026-09-20 18:45:49 | INFO | train rank=0 | step=27650/50000 progress=55.3% loss=0.083181 lr=4.537690e-05 lr_backbone=4.537690e-05 elapsed=12:07:48 avg_step=1.58s recent_step=2.09s eta=09:48:18 +2026-09-20 18:46:01 | INFO | train rank=0 | step=27660/50000 progress=55.3% loss=0.114900 lr=4.534397e-05 lr_backbone=4.534397e-05 elapsed=12:08:00 avg_step=1.58s recent_step=1.19s eta=09:47:59 +2026-09-20 18:46:14 | INFO | train rank=0 | step=27670/50000 progress=55.3% loss=0.082058 lr=4.531105e-05 lr_backbone=4.531105e-05 elapsed=12:08:14 avg_step=1.58s recent_step=1.37s eta=09:47:41 +2026-09-20 18:46:34 | INFO | train rank=0 | step=27680/50000 progress=55.4% loss=0.099122 lr=4.527812e-05 lr_backbone=4.527812e-05 elapsed=12:08:33 avg_step=1.58s recent_step=1.94s eta=09:47:29 +2026-09-20 18:46:50 | INFO | train rank=0 | step=27690/50000 progress=55.4% loss=0.083002 lr=4.524520e-05 lr_backbone=4.524520e-05 elapsed=12:08:49 avg_step=1.58s recent_step=1.61s eta=09:47:13 +2026-09-20 18:47:01 | INFO | train rank=0 | step=27700/50000 progress=55.4% loss=0.081134 lr=4.521228e-05 lr_backbone=4.521228e-05 elapsed=12:09:01 avg_step=1.58s recent_step=1.16s eta=09:46:54 +2026-09-20 18:47:20 | INFO | train rank=0 | step=27710/50000 progress=55.4% loss=0.076678 lr=4.517937e-05 lr_backbone=4.517937e-05 elapsed=12:09:19 avg_step=1.58s recent_step=1.84s eta=09:46:40 +2026-09-20 18:47:35 | INFO | train rank=0 | step=27720/50000 progress=55.4% loss=0.080197 lr=4.514645e-05 lr_backbone=4.514645e-05 elapsed=12:09:35 avg_step=1.58s recent_step=1.53s eta=09:46:24 +2026-09-20 18:47:54 | INFO | train rank=0 | step=27730/50000 progress=55.5% loss=0.082139 lr=4.511354e-05 lr_backbone=4.511354e-05 elapsed=12:09:53 avg_step=1.58s recent_step=1.85s eta=09:46:10 +2026-09-20 18:48:06 | INFO | train rank=0 | step=27740/50000 progress=55.5% loss=0.102483 lr=4.508063e-05 lr_backbone=4.508063e-05 elapsed=12:10:06 avg_step=1.58s recent_step=1.27s eta=09:45:52 +2026-09-20 18:48:21 | INFO | train rank=0 | step=27750/50000 progress=55.5% loss=0.079629 lr=4.504772e-05 lr_backbone=4.504772e-05 elapsed=12:10:20 avg_step=1.58s recent_step=1.45s eta=09:45:35 +2026-09-20 18:48:39 | INFO | train rank=0 | step=27760/50000 progress=55.5% loss=0.086289 lr=4.501482e-05 lr_backbone=4.501482e-05 elapsed=12:10:39 avg_step=1.58s recent_step=1.81s eta=09:45:21 +2026-09-20 18:48:54 | INFO | train rank=0 | step=27770/50000 progress=55.5% loss=0.102980 lr=4.498191e-05 lr_backbone=4.498191e-05 elapsed=12:10:54 avg_step=1.58s recent_step=1.53s eta=09:45:05 +2026-09-20 18:49:13 | INFO | train rank=0 | step=27780/50000 progress=55.6% loss=0.081691 lr=4.494901e-05 lr_backbone=4.494901e-05 elapsed=12:11:12 avg_step=1.58s recent_step=1.86s eta=09:44:51 +2026-09-20 18:49:25 | INFO | train rank=0 | step=27790/50000 progress=55.6% loss=0.126399 lr=4.491611e-05 lr_backbone=4.491611e-05 elapsed=12:11:24 avg_step=1.58s recent_step=1.19s eta=09:44:33 +2026-09-20 18:49:38 | INFO | train rank=0 | step=27800/50000 progress=55.6% loss=0.163295 lr=4.488322e-05 lr_backbone=4.488322e-05 elapsed=12:11:37 avg_step=1.58s recent_step=1.29s eta=09:44:15 +2026-09-20 18:49:55 | INFO | train rank=0 | step=27810/50000 progress=55.6% loss=0.106909 lr=4.485032e-05 lr_backbone=4.485032e-05 elapsed=12:11:54 avg_step=1.58s recent_step=1.71s eta=09:44:00 +2026-09-20 18:50:08 | INFO | train rank=0 | step=27820/50000 progress=55.6% loss=0.085411 lr=4.481743e-05 lr_backbone=4.481743e-05 elapsed=12:12:08 avg_step=1.58s recent_step=1.34s eta=09:43:42 +2026-09-20 18:50:25 | INFO | train rank=0 | step=27830/50000 progress=55.7% loss=0.092140 lr=4.478454e-05 lr_backbone=4.478454e-05 elapsed=12:12:25 avg_step=1.58s recent_step=1.72s eta=09:43:28 +2026-09-20 18:50:42 | INFO | train rank=0 | step=27840/50000 progress=55.7% loss=0.085653 lr=4.475165e-05 lr_backbone=4.475165e-05 elapsed=12:12:41 avg_step=1.58s recent_step=1.62s eta=09:43:12 +2026-09-20 18:50:54 | INFO | train rank=0 | step=27850/50000 progress=55.7% loss=0.071246 lr=4.471877e-05 lr_backbone=4.471877e-05 elapsed=12:12:54 avg_step=1.58s recent_step=1.24s eta=09:42:54 +2026-09-20 18:51:11 | INFO | train rank=0 | step=27860/50000 progress=55.7% loss=0.121398 lr=4.468588e-05 lr_backbone=4.468588e-05 elapsed=12:13:11 avg_step=1.58s recent_step=1.74s eta=09:42:39 +2026-09-20 18:51:23 | INFO | train rank=0 | step=27870/50000 progress=55.7% loss=0.107773 lr=4.465300e-05 lr_backbone=4.465300e-05 elapsed=12:13:23 avg_step=1.58s recent_step=1.20s eta=09:42:20 +2026-09-20 18:51:44 | INFO | train rank=0 | step=27880/50000 progress=55.8% loss=0.088707 lr=4.462012e-05 lr_backbone=4.462012e-05 elapsed=12:13:44 avg_step=1.58s recent_step=2.06s eta=09:42:08 +2026-09-20 18:51:55 | INFO | train rank=0 | step=27890/50000 progress=55.8% loss=0.085623 lr=4.458725e-05 lr_backbone=4.458725e-05 elapsed=12:13:55 avg_step=1.58s recent_step=1.13s eta=09:41:49 +2026-09-20 18:52:08 | INFO | train rank=0 | step=27900/50000 progress=55.8% loss=0.104725 lr=4.455437e-05 lr_backbone=4.455437e-05 elapsed=12:14:07 avg_step=1.58s recent_step=1.23s eta=09:41:30 +2026-09-20 18:52:25 | INFO | train rank=0 | step=27910/50000 progress=55.8% loss=0.082676 lr=4.452150e-05 lr_backbone=4.452150e-05 elapsed=12:14:25 avg_step=1.58s recent_step=1.78s eta=09:41:16 +2026-09-20 18:52:45 | INFO | train rank=0 | step=27920/50000 progress=55.8% loss=0.081251 lr=4.448863e-05 lr_backbone=4.448863e-05 elapsed=12:14:44 avg_step=1.58s recent_step=1.94s eta=09:41:03 +2026-09-20 18:53:00 | INFO | train rank=0 | step=27930/50000 progress=55.9% loss=0.101803 lr=4.445577e-05 lr_backbone=4.445577e-05 elapsed=12:15:00 avg_step=1.58s recent_step=1.55s eta=09:40:47 +2026-09-20 18:53:16 | INFO | train rank=0 | step=27940/50000 progress=55.9% loss=0.084008 lr=4.442290e-05 lr_backbone=4.442290e-05 elapsed=12:15:16 avg_step=1.58s recent_step=1.60s eta=09:40:32 +2026-09-20 18:53:29 | INFO | train rank=0 | step=27950/50000 progress=55.9% loss=0.169041 lr=4.439004e-05 lr_backbone=4.439004e-05 elapsed=12:15:29 avg_step=1.58s recent_step=1.30s eta=09:40:14 +2026-09-20 18:53:49 | INFO | train rank=0 | step=27960/50000 progress=55.9% loss=0.110441 lr=4.435718e-05 lr_backbone=4.435718e-05 elapsed=12:15:49 avg_step=1.58s recent_step=1.96s eta=09:40:01 +2026-09-20 18:54:01 | INFO | train rank=0 | step=27970/50000 progress=55.9% loss=0.111922 lr=4.432432e-05 lr_backbone=4.432432e-05 elapsed=12:16:00 avg_step=1.58s recent_step=1.17s eta=09:39:42 +2026-09-20 18:54:13 | INFO | train rank=0 | step=27980/50000 progress=56.0% loss=0.074323 lr=4.429147e-05 lr_backbone=4.429147e-05 elapsed=12:16:12 avg_step=1.58s recent_step=1.22s eta=09:39:23 +2026-09-20 18:54:31 | INFO | train rank=0 | step=27990/50000 progress=56.0% loss=0.121172 lr=4.425862e-05 lr_backbone=4.425862e-05 elapsed=12:16:31 avg_step=1.58s recent_step=1.83s eta=09:39:09 +2026-09-20 18:54:49 | INFO | train rank=0 | step=28000/50000 progress=56.0% loss=0.102025 lr=4.422577e-05 lr_backbone=4.422577e-05 elapsed=12:16:49 avg_step=1.58s recent_step=1.83s eta=09:38:56 +2026-09-20 18:55:01 | INFO | train rank=0 | step=28010/50000 progress=56.0% loss=0.064620 lr=4.419292e-05 lr_backbone=4.419292e-05 elapsed=12:17:01 avg_step=1.58s recent_step=1.18s eta=09:38:37 +2026-09-20 18:55:18 | INFO | train rank=0 | step=28020/50000 progress=56.0% loss=0.065288 lr=4.416008e-05 lr_backbone=4.416008e-05 elapsed=12:17:18 avg_step=1.58s recent_step=1.73s eta=09:38:22 +2026-09-20 18:55:30 | INFO | train rank=0 | step=28030/50000 progress=56.1% loss=0.096208 lr=4.412724e-05 lr_backbone=4.412724e-05 elapsed=12:17:29 avg_step=1.58s recent_step=1.11s eta=09:38:03 +2026-09-20 18:55:46 | INFO | train rank=0 | step=28040/50000 progress=56.1% loss=0.095741 lr=4.409440e-05 lr_backbone=4.409440e-05 elapsed=12:17:45 avg_step=1.58s recent_step=1.60s eta=09:37:47 +2026-09-20 18:56:02 | INFO | train rank=0 | step=28050/50000 progress=56.1% loss=0.079448 lr=4.406156e-05 lr_backbone=4.406156e-05 elapsed=12:18:02 avg_step=1.58s recent_step=1.67s eta=09:37:32 +2026-09-20 18:56:20 | INFO | train rank=0 | step=28060/50000 progress=56.1% loss=0.093234 lr=4.402873e-05 lr_backbone=4.402873e-05 elapsed=12:18:20 avg_step=1.58s recent_step=1.77s eta=09:37:18 +2026-09-20 18:56:36 | INFO | train rank=0 | step=28070/50000 progress=56.1% loss=0.071127 lr=4.399589e-05 lr_backbone=4.399589e-05 elapsed=12:18:36 avg_step=1.58s recent_step=1.62s eta=09:37:02 +2026-09-20 18:56:56 | INFO | train rank=0 | step=28080/50000 progress=56.2% loss=0.111042 lr=4.396307e-05 lr_backbone=4.396307e-05 elapsed=12:18:56 avg_step=1.58s recent_step=2.01s eta=09:36:50 +2026-09-20 18:57:12 | INFO | train rank=0 | step=28090/50000 progress=56.2% loss=0.094949 lr=4.393024e-05 lr_backbone=4.393024e-05 elapsed=12:19:11 avg_step=1.58s recent_step=1.55s eta=09:36:34 +2026-09-20 18:57:24 | INFO | train rank=0 | step=28100/50000 progress=56.2% loss=0.073631 lr=4.389742e-05 lr_backbone=4.389742e-05 elapsed=12:19:23 avg_step=1.58s recent_step=1.22s eta=09:36:15 +2026-09-20 18:57:41 | INFO | train rank=0 | step=28110/50000 progress=56.2% loss=0.103402 lr=4.386459e-05 lr_backbone=4.386459e-05 elapsed=12:19:41 avg_step=1.58s recent_step=1.70s eta=09:36:00 +2026-09-20 18:57:59 | INFO | train rank=0 | step=28120/50000 progress=56.2% loss=0.092264 lr=4.383178e-05 lr_backbone=4.383178e-05 elapsed=12:19:59 avg_step=1.58s recent_step=1.86s eta=09:35:47 +2026-09-20 18:58:17 | INFO | train rank=0 | step=28130/50000 progress=56.3% loss=0.094580 lr=4.379896e-05 lr_backbone=4.379896e-05 elapsed=12:20:17 avg_step=1.58s recent_step=1.76s eta=09:35:32 +2026-09-20 18:58:28 | INFO | train rank=0 | step=28140/50000 progress=56.3% loss=0.076416 lr=4.376615e-05 lr_backbone=4.376615e-05 elapsed=12:20:28 avg_step=1.58s recent_step=1.10s eta=09:35:13 +2026-09-20 18:58:42 | INFO | train rank=0 | step=28150/50000 progress=56.3% loss=0.125849 lr=4.373334e-05 lr_backbone=4.373334e-05 elapsed=12:20:42 avg_step=1.58s recent_step=1.43s eta=09:34:56 +2026-09-20 18:59:03 | INFO | train rank=0 | step=28160/50000 progress=56.3% loss=0.089042 lr=4.370053e-05 lr_backbone=4.370053e-05 elapsed=12:21:02 avg_step=1.58s recent_step=2.05s eta=09:34:44 +2026-09-20 18:59:16 | INFO | train rank=0 | step=28170/50000 progress=56.3% loss=0.098045 lr=4.366773e-05 lr_backbone=4.366773e-05 elapsed=12:21:15 avg_step=1.58s recent_step=1.27s eta=09:34:25 +2026-09-20 18:59:29 | INFO | train rank=0 | step=28180/50000 progress=56.4% loss=0.088238 lr=4.363492e-05 lr_backbone=4.363492e-05 elapsed=12:21:29 avg_step=1.58s recent_step=1.37s eta=09:34:08 +2026-09-20 18:59:44 | INFO | train rank=0 | step=28190/50000 progress=56.4% loss=0.089194 lr=4.360213e-05 lr_backbone=4.360213e-05 elapsed=12:21:44 avg_step=1.58s recent_step=1.53s eta=09:33:52 +2026-09-20 19:00:03 | INFO | train rank=0 | step=28200/50000 progress=56.4% loss=0.098959 lr=4.356933e-05 lr_backbone=4.356933e-05 elapsed=12:22:03 avg_step=1.58s recent_step=1.85s eta=09:33:38 +2026-09-20 19:00:19 | INFO | train rank=0 | step=28210/50000 progress=56.4% loss=0.090695 lr=4.353654e-05 lr_backbone=4.353654e-05 elapsed=12:22:18 avg_step=1.58s recent_step=1.57s eta=09:33:22 +2026-09-20 19:00:37 | INFO | train rank=0 | step=28220/50000 progress=56.4% loss=0.085231 lr=4.350375e-05 lr_backbone=4.350375e-05 elapsed=12:22:37 avg_step=1.58s recent_step=1.86s eta=09:33:09 +2026-09-20 19:00:49 | INFO | train rank=0 | step=28230/50000 progress=56.5% loss=0.085955 lr=4.347096e-05 lr_backbone=4.347096e-05 elapsed=12:22:49 avg_step=1.58s recent_step=1.20s eta=09:32:50 +2026-09-20 19:01:02 | INFO | train rank=0 | step=28240/50000 progress=56.5% loss=0.146196 lr=4.343817e-05 lr_backbone=4.343817e-05 elapsed=12:23:02 avg_step=1.58s recent_step=1.30s eta=09:32:32 +2026-09-20 19:01:18 | INFO | train rank=0 | step=28250/50000 progress=56.5% loss=0.089143 lr=4.340539e-05 lr_backbone=4.340539e-05 elapsed=12:23:18 avg_step=1.58s recent_step=1.61s eta=09:32:16 +2026-09-20 19:01:42 | INFO | train rank=0 | step=28260/50000 progress=56.5% loss=0.080303 lr=4.337261e-05 lr_backbone=4.337261e-05 elapsed=12:23:41 avg_step=1.58s recent_step=2.35s eta=09:32:06 +2026-09-20 19:01:54 | INFO | train rank=0 | step=28270/50000 progress=56.5% loss=0.080487 lr=4.333984e-05 lr_backbone=4.333984e-05 elapsed=12:23:53 avg_step=1.58s recent_step=1.20s eta=09:31:48 +2026-09-20 19:02:05 | INFO | train rank=0 | step=28280/50000 progress=56.6% loss=0.073767 lr=4.330706e-05 lr_backbone=4.330706e-05 elapsed=12:24:05 avg_step=1.58s recent_step=1.15s eta=09:31:29 +2026-09-20 19:02:20 | INFO | train rank=0 | step=28290/50000 progress=56.6% loss=0.080266 lr=4.327429e-05 lr_backbone=4.327429e-05 elapsed=12:24:19 avg_step=1.58s recent_step=1.44s eta=09:31:12 +2026-09-20 19:02:39 | INFO | train rank=0 | step=28300/50000 progress=56.6% loss=0.090533 lr=4.324153e-05 lr_backbone=4.324153e-05 elapsed=12:24:39 avg_step=1.58s recent_step=1.95s eta=09:30:59 +2026-09-20 19:02:56 | INFO | train rank=0 | step=28310/50000 progress=56.6% loss=0.082098 lr=4.320876e-05 lr_backbone=4.320876e-05 elapsed=12:24:55 avg_step=1.58s recent_step=1.66s eta=09:30:44 +2026-09-20 19:03:14 | INFO | train rank=0 | step=28320/50000 progress=56.6% loss=0.081175 lr=4.317600e-05 lr_backbone=4.317600e-05 elapsed=12:25:13 avg_step=1.58s recent_step=1.79s eta=09:30:30 +2026-09-20 19:03:26 | INFO | train rank=0 | step=28330/50000 progress=56.7% loss=0.098230 lr=4.314324e-05 lr_backbone=4.314324e-05 elapsed=12:25:26 avg_step=1.58s recent_step=1.28s eta=09:30:11 +2026-09-20 19:03:41 | INFO | train rank=0 | step=28340/50000 progress=56.7% loss=0.090622 lr=4.311049e-05 lr_backbone=4.311049e-05 elapsed=12:25:41 avg_step=1.58s recent_step=1.45s eta=09:29:55 +2026-09-20 19:03:54 | INFO | train rank=0 | step=28350/50000 progress=56.7% loss=0.076420 lr=4.307773e-05 lr_backbone=4.307773e-05 elapsed=12:25:53 avg_step=1.58s recent_step=1.28s eta=09:29:37 +2026-09-20 19:04:11 | INFO | train rank=0 | step=28360/50000 progress=56.7% loss=0.079001 lr=4.304498e-05 lr_backbone=4.304498e-05 elapsed=12:26:11 avg_step=1.58s recent_step=1.77s eta=09:29:22 +2026-09-20 19:04:30 | INFO | train rank=0 | step=28370/50000 progress=56.7% loss=0.072770 lr=4.301224e-05 lr_backbone=4.301224e-05 elapsed=12:26:30 avg_step=1.58s recent_step=1.85s eta=09:29:09 +2026-09-20 19:04:50 | INFO | train rank=0 | step=28380/50000 progress=56.8% loss=0.153403 lr=4.297949e-05 lr_backbone=4.297949e-05 elapsed=12:26:50 avg_step=1.58s recent_step=2.05s eta=09:28:56 +2026-09-20 19:05:02 | INFO | train rank=0 | step=28390/50000 progress=56.8% loss=0.125900 lr=4.294675e-05 lr_backbone=4.294675e-05 elapsed=12:27:02 avg_step=1.58s recent_step=1.16s eta=09:28:37 +2026-09-20 19:05:18 | INFO | train rank=0 | step=28400/50000 progress=56.8% loss=0.090499 lr=4.291402e-05 lr_backbone=4.291402e-05 elapsed=12:27:17 avg_step=1.58s recent_step=1.57s eta=09:28:22 +2026-09-20 19:05:36 | INFO | train rank=0 | step=28410/50000 progress=56.8% loss=0.097864 lr=4.288128e-05 lr_backbone=4.288128e-05 elapsed=12:27:36 avg_step=1.58s recent_step=1.86s eta=09:28:08 +2026-09-20 19:05:48 | INFO | train rank=0 | step=28420/50000 progress=56.8% loss=0.089403 lr=4.284855e-05 lr_backbone=4.284855e-05 elapsed=12:27:47 avg_step=1.58s recent_step=1.16s eta=09:27:49 +2026-09-20 19:06:05 | INFO | train rank=0 | step=28430/50000 progress=56.9% loss=0.089117 lr=4.281582e-05 lr_backbone=4.281582e-05 elapsed=12:28:04 avg_step=1.58s recent_step=1.69s eta=09:27:34 +2026-09-20 19:06:24 | INFO | train rank=0 | step=28440/50000 progress=56.9% loss=0.091420 lr=4.278310e-05 lr_backbone=4.278310e-05 elapsed=12:28:24 avg_step=1.58s recent_step=1.96s eta=09:27:21 +2026-09-20 19:06:38 | INFO | train rank=0 | step=28450/50000 progress=56.9% loss=0.070553 lr=4.275038e-05 lr_backbone=4.275038e-05 elapsed=12:28:38 avg_step=1.58s recent_step=1.35s eta=09:27:04 +2026-09-20 19:06:51 | INFO | train rank=0 | step=28460/50000 progress=56.9% loss=0.111291 lr=4.271766e-05 lr_backbone=4.271766e-05 elapsed=12:28:51 avg_step=1.58s recent_step=1.33s eta=09:26:46 +2026-09-20 19:07:03 | INFO | train rank=0 | step=28470/50000 progress=56.9% loss=0.084115 lr=4.268494e-05 lr_backbone=4.268494e-05 elapsed=12:29:03 avg_step=1.58s recent_step=1.22s eta=09:26:27 +2026-09-20 19:07:21 | INFO | train rank=0 | step=28480/50000 progress=57.0% loss=0.088899 lr=4.265223e-05 lr_backbone=4.265223e-05 elapsed=12:29:21 avg_step=1.58s recent_step=1.77s eta=09:26:13 +2026-09-20 19:07:38 | INFO | train rank=0 | step=28490/50000 progress=57.0% loss=0.092929 lr=4.261952e-05 lr_backbone=4.261952e-05 elapsed=12:29:38 avg_step=1.58s recent_step=1.74s eta=09:25:58 +2026-09-20 19:07:53 | INFO | train rank=0 | step=28500/50000 progress=57.0% loss=0.125269 lr=4.258682e-05 lr_backbone=4.258682e-05 elapsed=12:29:53 avg_step=1.58s recent_step=1.46s eta=09:25:42 +2026-09-20 19:08:11 | INFO | train rank=0 | step=28510/50000 progress=57.0% loss=0.079100 lr=4.255412e-05 lr_backbone=4.255412e-05 elapsed=12:30:10 avg_step=1.58s recent_step=1.76s eta=09:25:27 +2026-09-20 19:08:28 | INFO | train rank=0 | step=28520/50000 progress=57.0% loss=0.098752 lr=4.252142e-05 lr_backbone=4.252142e-05 elapsed=12:30:28 avg_step=1.58s recent_step=1.76s eta=09:25:13 +2026-09-20 19:08:40 | INFO | train rank=0 | step=28530/50000 progress=57.1% loss=0.080465 lr=4.248872e-05 lr_backbone=4.248872e-05 elapsed=12:30:39 avg_step=1.58s recent_step=1.16s eta=09:24:54 +2026-09-20 19:08:57 | INFO | train rank=0 | step=28540/50000 progress=57.1% loss=0.072363 lr=4.245603e-05 lr_backbone=4.245603e-05 elapsed=12:30:57 avg_step=1.58s recent_step=1.76s eta=09:24:39 +2026-09-20 19:09:19 | INFO | train rank=0 | step=28550/50000 progress=57.1% loss=0.082831 lr=4.242334e-05 lr_backbone=4.242334e-05 elapsed=12:31:18 avg_step=1.58s recent_step=2.12s eta=09:24:28 +2026-09-20 19:09:30 | INFO | train rank=0 | step=28560/50000 progress=57.1% loss=0.087165 lr=4.239065e-05 lr_backbone=4.239065e-05 elapsed=12:31:30 avg_step=1.58s recent_step=1.18s eta=09:24:09 +2026-09-20 19:09:43 | INFO | train rank=0 | step=28570/50000 progress=57.1% loss=0.108477 lr=4.235797e-05 lr_backbone=4.235797e-05 elapsed=12:31:42 avg_step=1.58s recent_step=1.21s eta=09:23:50 +2026-09-20 19:09:55 | INFO | train rank=0 | step=28580/50000 progress=57.2% loss=0.143570 lr=4.232529e-05 lr_backbone=4.232529e-05 elapsed=12:31:54 avg_step=1.58s recent_step=1.21s eta=09:23:32 +2026-09-20 19:10:13 | INFO | train rank=0 | step=28590/50000 progress=57.2% loss=0.089013 lr=4.229262e-05 lr_backbone=4.229262e-05 elapsed=12:32:13 avg_step=1.58s recent_step=1.82s eta=09:23:18 +2026-09-20 19:10:30 | INFO | train rank=0 | step=28600/50000 progress=57.2% loss=0.087228 lr=4.225994e-05 lr_backbone=4.225994e-05 elapsed=12:32:29 avg_step=1.58s recent_step=1.69s eta=09:23:03 +2026-09-20 19:10:42 | INFO | train rank=0 | step=28610/50000 progress=57.2% loss=0.089124 lr=4.222727e-05 lr_backbone=4.222727e-05 elapsed=12:32:42 avg_step=1.58s recent_step=1.22s eta=09:22:45 +2026-09-20 19:11:06 | INFO | train rank=0 | step=28620/50000 progress=57.2% loss=0.092904 lr=4.219461e-05 lr_backbone=4.219461e-05 elapsed=12:33:05 avg_step=1.58s recent_step=2.36s eta=09:22:35 +2026-09-20 19:11:17 | INFO | train rank=0 | step=28630/50000 progress=57.3% loss=0.092782 lr=4.216195e-05 lr_backbone=4.216195e-05 elapsed=12:33:17 avg_step=1.58s recent_step=1.16s eta=09:22:16 +2026-09-20 19:11:34 | INFO | train rank=0 | step=28640/50000 progress=57.3% loss=0.104760 lr=4.212929e-05 lr_backbone=4.212929e-05 elapsed=12:33:33 avg_step=1.58s recent_step=1.65s eta=09:22:00 +2026-09-20 19:11:46 | INFO | train rank=0 | step=28650/50000 progress=57.3% loss=0.105699 lr=4.209663e-05 lr_backbone=4.209663e-05 elapsed=12:33:46 avg_step=1.58s recent_step=1.25s eta=09:21:42 +2026-09-20 19:12:06 | INFO | train rank=0 | step=28660/50000 progress=57.3% loss=0.090281 lr=4.206398e-05 lr_backbone=4.206398e-05 elapsed=12:34:06 avg_step=1.58s recent_step=1.99s eta=09:21:29 +2026-09-20 19:12:21 | INFO | train rank=0 | step=28670/50000 progress=57.3% loss=0.131406 lr=4.203133e-05 lr_backbone=4.203133e-05 elapsed=12:34:21 avg_step=1.58s recent_step=1.49s eta=09:21:13 +2026-09-20 19:12:41 | INFO | train rank=0 | step=28680/50000 progress=57.4% loss=0.085080 lr=4.199869e-05 lr_backbone=4.199869e-05 elapsed=12:34:40 avg_step=1.58s recent_step=1.98s eta=09:21:00 +2026-09-20 19:12:54 | INFO | train rank=0 | step=28690/50000 progress=57.4% loss=0.097723 lr=4.196605e-05 lr_backbone=4.196605e-05 elapsed=12:34:54 avg_step=1.58s recent_step=1.37s eta=09:20:43 +2026-09-20 19:13:07 | INFO | train rank=0 | step=28700/50000 progress=57.4% loss=0.076208 lr=4.193341e-05 lr_backbone=4.193341e-05 elapsed=12:35:06 avg_step=1.58s recent_step=1.24s eta=09:20:25 +2026-09-20 19:13:25 | INFO | train rank=0 | step=28710/50000 progress=57.4% loss=0.070724 lr=4.190077e-05 lr_backbone=4.190077e-05 elapsed=12:35:25 avg_step=1.58s recent_step=1.81s eta=09:20:10 +2026-09-20 19:13:40 | INFO | train rank=0 | step=28720/50000 progress=57.4% loss=0.118374 lr=4.186814e-05 lr_backbone=4.186814e-05 elapsed=12:35:40 avg_step=1.58s recent_step=1.49s eta=09:19:54 +2026-09-20 19:13:53 | INFO | train rank=0 | step=28730/50000 progress=57.5% loss=0.117040 lr=4.183551e-05 lr_backbone=4.183551e-05 elapsed=12:35:52 avg_step=1.58s recent_step=1.28s eta=09:19:36 +2026-09-20 19:14:12 | INFO | train rank=0 | step=28740/50000 progress=57.5% loss=0.065016 lr=4.180289e-05 lr_backbone=4.180289e-05 elapsed=12:36:12 avg_step=1.58s recent_step=1.94s eta=09:19:23 +2026-09-20 19:14:27 | INFO | train rank=0 | step=28750/50000 progress=57.5% loss=0.096863 lr=4.177027e-05 lr_backbone=4.177027e-05 elapsed=12:36:27 avg_step=1.58s recent_step=1.54s eta=09:19:07 +2026-09-20 19:14:43 | INFO | train rank=0 | step=28760/50000 progress=57.5% loss=0.109089 lr=4.173765e-05 lr_backbone=4.173765e-05 elapsed=12:36:43 avg_step=1.58s recent_step=1.58s eta=09:18:51 +2026-09-20 19:15:02 | INFO | train rank=0 | step=28770/50000 progress=57.5% loss=0.064701 lr=4.170504e-05 lr_backbone=4.170504e-05 elapsed=12:37:01 avg_step=1.58s recent_step=1.86s eta=09:18:37 +2026-09-20 19:15:18 | INFO | train rank=0 | step=28780/50000 progress=57.6% loss=0.075061 lr=4.167243e-05 lr_backbone=4.167243e-05 elapsed=12:37:18 avg_step=1.58s recent_step=1.66s eta=09:18:22 +2026-09-20 19:15:32 | INFO | train rank=0 | step=28790/50000 progress=57.6% loss=0.118471 lr=4.163983e-05 lr_backbone=4.163983e-05 elapsed=12:37:31 avg_step=1.58s recent_step=1.34s eta=09:18:05 +2026-09-20 19:15:46 | INFO | train rank=0 | step=28800/50000 progress=57.6% loss=0.069277 lr=4.160722e-05 lr_backbone=4.160722e-05 elapsed=12:37:46 avg_step=1.58s recent_step=1.41s eta=09:17:48 +2026-09-20 19:15:58 | INFO | train rank=0 | step=28810/50000 progress=57.6% loss=0.093776 lr=4.157463e-05 lr_backbone=4.157463e-05 elapsed=12:37:57 avg_step=1.58s recent_step=1.19s eta=09:17:29 +2026-09-20 19:16:14 | INFO | train rank=0 | step=28820/50000 progress=57.6% loss=0.082974 lr=4.154203e-05 lr_backbone=4.154203e-05 elapsed=12:38:14 avg_step=1.58s recent_step=1.65s eta=09:17:14 +2026-09-20 19:16:30 | INFO | train rank=0 | step=28830/50000 progress=57.7% loss=0.097570 lr=4.150944e-05 lr_backbone=4.150944e-05 elapsed=12:38:30 avg_step=1.58s recent_step=1.60s eta=09:16:58 +2026-09-20 19:16:46 | INFO | train rank=0 | step=28840/50000 progress=57.7% loss=0.113997 lr=4.147685e-05 lr_backbone=4.147685e-05 elapsed=12:38:46 avg_step=1.58s recent_step=1.59s eta=09:16:42 +2026-09-20 19:17:05 | INFO | train rank=0 | step=28850/50000 progress=57.7% loss=0.074556 lr=4.144427e-05 lr_backbone=4.144427e-05 elapsed=12:39:05 avg_step=1.58s recent_step=1.90s eta=09:16:29 +2026-09-20 19:17:21 | INFO | train rank=0 | step=28860/50000 progress=57.7% loss=0.077113 lr=4.141169e-05 lr_backbone=4.141169e-05 elapsed=12:39:21 avg_step=1.58s recent_step=1.59s eta=09:16:13 +2026-09-20 19:17:35 | INFO | train rank=0 | step=28870/50000 progress=57.7% loss=0.090698 lr=4.137911e-05 lr_backbone=4.137911e-05 elapsed=12:39:34 avg_step=1.58s recent_step=1.34s eta=09:15:56 +2026-09-20 19:17:54 | INFO | train rank=0 | step=28880/50000 progress=57.8% loss=0.093361 lr=4.134654e-05 lr_backbone=4.134654e-05 elapsed=12:39:53 avg_step=1.58s recent_step=1.90s eta=09:15:42 +2026-09-20 19:18:06 | INFO | train rank=0 | step=28890/50000 progress=57.8% loss=0.099165 lr=4.131397e-05 lr_backbone=4.131397e-05 elapsed=12:40:05 avg_step=1.58s recent_step=1.19s eta=09:15:24 +2026-09-20 19:18:22 | INFO | train rank=0 | step=28900/50000 progress=57.8% loss=0.092155 lr=4.128141e-05 lr_backbone=4.128141e-05 elapsed=12:40:21 avg_step=1.58s recent_step=1.63s eta=09:15:08 +2026-09-20 19:18:40 | INFO | train rank=0 | step=28910/50000 progress=57.8% loss=0.072968 lr=4.124885e-05 lr_backbone=4.124885e-05 elapsed=12:40:40 avg_step=1.58s recent_step=1.82s eta=09:14:54 +2026-09-20 19:18:53 | INFO | train rank=0 | step=28920/50000 progress=57.8% loss=0.143465 lr=4.121629e-05 lr_backbone=4.121629e-05 elapsed=12:40:53 avg_step=1.58s recent_step=1.32s eta=09:14:36 +2026-09-20 19:19:07 | INFO | train rank=0 | step=28930/50000 progress=57.9% loss=0.107909 lr=4.118374e-05 lr_backbone=4.118374e-05 elapsed=12:41:07 avg_step=1.58s recent_step=1.39s eta=09:14:19 +2026-09-20 19:19:30 | INFO | train rank=0 | step=28940/50000 progress=57.9% loss=0.095938 lr=4.115119e-05 lr_backbone=4.115119e-05 elapsed=12:41:29 avg_step=1.58s recent_step=2.28s eta=09:14:09 +2026-09-20 19:19:42 | INFO | train rank=0 | step=28950/50000 progress=57.9% loss=0.081083 lr=4.111864e-05 lr_backbone=4.111864e-05 elapsed=12:41:41 avg_step=1.58s recent_step=1.20s eta=09:13:50 +2026-09-20 19:19:54 | INFO | train rank=0 | step=28960/50000 progress=57.9% loss=0.102281 lr=4.108610e-05 lr_backbone=4.108610e-05 elapsed=12:41:54 avg_step=1.58s recent_step=1.22s eta=09:13:32 +2026-09-20 19:20:06 | INFO | train rank=0 | step=28970/50000 progress=57.9% loss=0.096817 lr=4.105356e-05 lr_backbone=4.105356e-05 elapsed=12:42:06 avg_step=1.58s recent_step=1.24s eta=09:13:13 +2026-09-20 19:20:31 | INFO | train rank=0 | step=28980/50000 progress=58.0% loss=0.095674 lr=4.102103e-05 lr_backbone=4.102103e-05 elapsed=12:42:30 avg_step=1.58s recent_step=2.43s eta=09:13:04 +2026-09-20 19:20:49 | INFO | train rank=0 | step=28990/50000 progress=58.0% loss=0.075581 lr=4.098850e-05 lr_backbone=4.098850e-05 elapsed=12:42:48 avg_step=1.58s recent_step=1.79s eta=09:12:50 +2026-09-20 19:21:04 | INFO | train rank=0 | step=29000/50000 progress=58.0% loss=0.080830 lr=4.095597e-05 lr_backbone=4.095597e-05 elapsed=12:43:04 avg_step=1.58s recent_step=1.57s eta=09:12:34 +2026-09-20 19:21:16 | INFO | train rank=0 | step=29010/50000 progress=58.0% loss=0.095696 lr=4.092345e-05 lr_backbone=4.092345e-05 elapsed=12:43:16 avg_step=1.58s recent_step=1.17s eta=09:12:15 +2026-09-20 19:21:34 | INFO | train rank=0 | step=29020/50000 progress=58.0% loss=0.081886 lr=4.089093e-05 lr_backbone=4.089093e-05 elapsed=12:43:33 avg_step=1.58s recent_step=1.79s eta=09:12:01 +2026-09-20 19:21:50 | INFO | train rank=0 | step=29030/50000 progress=58.1% loss=0.130928 lr=4.085842e-05 lr_backbone=4.085842e-05 elapsed=12:43:50 avg_step=1.58s recent_step=1.63s eta=09:11:45 +2026-09-20 19:22:04 | INFO | train rank=0 | step=29040/50000 progress=58.1% loss=0.143707 lr=4.082591e-05 lr_backbone=4.082591e-05 elapsed=12:44:04 avg_step=1.58s recent_step=1.41s eta=09:11:28 +2026-09-20 19:22:22 | INFO | train rank=0 | step=29050/50000 progress=58.1% loss=0.101753 lr=4.079340e-05 lr_backbone=4.079340e-05 elapsed=12:44:22 avg_step=1.58s recent_step=1.78s eta=09:11:14 +2026-09-20 19:22:34 | INFO | train rank=0 | step=29060/50000 progress=58.1% loss=0.119715 lr=4.076090e-05 lr_backbone=4.076090e-05 elapsed=12:44:33 avg_step=1.58s recent_step=1.17s eta=09:10:55 +2026-09-20 19:22:46 | INFO | train rank=0 | step=29070/50000 progress=58.1% loss=0.076770 lr=4.072840e-05 lr_backbone=4.072840e-05 elapsed=12:44:45 avg_step=1.58s recent_step=1.21s eta=09:10:37 +2026-09-20 19:22:58 | INFO | train rank=0 | step=29080/50000 progress=58.2% loss=0.171377 lr=4.069591e-05 lr_backbone=4.069591e-05 elapsed=12:44:57 avg_step=1.58s recent_step=1.21s eta=09:10:18 +2026-09-20 19:23:09 | INFO | train rank=0 | step=29090/50000 progress=58.2% loss=0.072010 lr=4.066342e-05 lr_backbone=4.066342e-05 elapsed=12:45:09 avg_step=1.58s recent_step=1.15s eta=09:09:59 +2026-09-20 19:23:23 | INFO | train rank=0 | step=29100/50000 progress=58.2% loss=0.115407 lr=4.063093e-05 lr_backbone=4.063093e-05 elapsed=12:45:22 avg_step=1.58s recent_step=1.31s eta=09:09:42 +2026-09-20 19:23:39 | INFO | train rank=0 | step=29110/50000 progress=58.2% loss=0.078954 lr=4.059845e-05 lr_backbone=4.059845e-05 elapsed=12:45:38 avg_step=1.58s recent_step=1.60s eta=09:09:26 +2026-09-20 19:23:58 | INFO | train rank=0 | step=29120/50000 progress=58.2% loss=0.110119 lr=4.056598e-05 lr_backbone=4.056598e-05 elapsed=12:45:58 avg_step=1.58s recent_step=1.96s eta=09:09:13 +2026-09-20 19:24:10 | INFO | train rank=0 | step=29130/50000 progress=58.3% loss=0.093273 lr=4.053350e-05 lr_backbone=4.053350e-05 elapsed=12:46:10 avg_step=1.58s recent_step=1.22s eta=09:08:55 +2026-09-20 19:24:31 | INFO | train rank=0 | step=29140/50000 progress=58.3% loss=0.102508 lr=4.050103e-05 lr_backbone=4.050103e-05 elapsed=12:46:31 avg_step=1.58s recent_step=2.11s eta=09:08:43 +2026-09-20 19:24:45 | INFO | train rank=0 | step=29150/50000 progress=58.3% loss=0.082658 lr=4.046857e-05 lr_backbone=4.046857e-05 elapsed=12:46:44 avg_step=1.58s recent_step=1.33s eta=09:08:25 +2026-09-20 19:25:03 | INFO | train rank=0 | step=29160/50000 progress=58.3% loss=0.161317 lr=4.043611e-05 lr_backbone=4.043611e-05 elapsed=12:47:02 avg_step=1.58s recent_step=1.80s eta=09:08:11 +2026-09-20 19:25:22 | INFO | train rank=0 | step=29170/50000 progress=58.3% loss=0.071720 lr=4.040365e-05 lr_backbone=4.040365e-05 elapsed=12:47:22 avg_step=1.58s recent_step=1.98s eta=09:07:58 +2026-09-20 19:25:43 | INFO | train rank=0 | step=29180/50000 progress=58.4% loss=0.096076 lr=4.037120e-05 lr_backbone=4.037120e-05 elapsed=12:47:43 avg_step=1.58s recent_step=2.07s eta=09:07:46 +2026-09-20 19:26:00 | INFO | train rank=0 | step=29190/50000 progress=58.4% loss=0.094750 lr=4.033875e-05 lr_backbone=4.033875e-05 elapsed=12:47:59 avg_step=1.58s recent_step=1.66s eta=09:07:31 +2026-09-20 19:26:18 | INFO | train rank=0 | step=29200/50000 progress=58.4% loss=0.107183 lr=4.030631e-05 lr_backbone=4.030631e-05 elapsed=12:48:18 avg_step=1.58s recent_step=1.84s eta=09:07:17 +2026-09-20 19:26:34 | INFO | train rank=0 | step=29210/50000 progress=58.4% loss=0.075920 lr=4.027387e-05 lr_backbone=4.027387e-05 elapsed=12:48:33 avg_step=1.58s recent_step=1.55s eta=09:07:01 +2026-09-20 19:26:48 | INFO | train rank=0 | step=29220/50000 progress=58.4% loss=0.100571 lr=4.024143e-05 lr_backbone=4.024143e-05 elapsed=12:48:48 avg_step=1.58s recent_step=1.45s eta=09:06:44 +2026-09-20 19:27:03 | INFO | train rank=0 | step=29230/50000 progress=58.5% loss=0.083844 lr=4.020900e-05 lr_backbone=4.020900e-05 elapsed=12:49:02 avg_step=1.58s recent_step=1.45s eta=09:06:27 +2026-09-20 19:27:21 | INFO | train rank=0 | step=29240/50000 progress=58.5% loss=0.124417 lr=4.017657e-05 lr_backbone=4.017657e-05 elapsed=12:49:20 avg_step=1.58s recent_step=1.81s eta=09:06:13 +2026-09-20 19:27:34 | INFO | train rank=0 | step=29250/50000 progress=58.5% loss=0.121962 lr=4.014415e-05 lr_backbone=4.014415e-05 elapsed=12:49:34 avg_step=1.58s recent_step=1.36s eta=09:05:56 +2026-09-20 19:28:33 | INFO | train rank=0 | step=29260/50000 progress=58.5% loss=0.113779 lr=4.011173e-05 lr_backbone=4.011173e-05 elapsed=12:50:33 avg_step=1.58s recent_step=5.88s eta=09:06:10 +2026-09-20 19:28:48 | INFO | train rank=0 | step=29270/50000 progress=58.5% loss=0.096818 lr=4.007932e-05 lr_backbone=4.007932e-05 elapsed=12:50:48 avg_step=1.58s recent_step=1.48s eta=09:05:54 +2026-09-20 19:29:05 | INFO | train rank=0 | step=29280/50000 progress=58.6% loss=0.091429 lr=4.004691e-05 lr_backbone=4.004691e-05 elapsed=12:51:04 avg_step=1.58s recent_step=1.69s eta=09:05:39 +2026-09-20 19:29:20 | INFO | train rank=0 | step=29290/50000 progress=58.6% loss=0.085722 lr=4.001450e-05 lr_backbone=4.001450e-05 elapsed=12:51:20 avg_step=1.58s recent_step=1.53s eta=09:05:23 +2026-09-20 19:29:34 | INFO | train rank=0 | step=29300/50000 progress=58.6% loss=0.090490 lr=3.998210e-05 lr_backbone=3.998210e-05 elapsed=12:51:34 avg_step=1.58s recent_step=1.38s eta=09:05:06 +2026-09-20 19:29:53 | INFO | train rank=0 | step=29310/50000 progress=58.6% loss=0.127169 lr=3.994970e-05 lr_backbone=3.994970e-05 elapsed=12:51:52 avg_step=1.58s recent_step=1.87s eta=09:04:52 +2026-09-20 19:30:05 | INFO | train rank=0 | step=29320/50000 progress=58.6% loss=0.087260 lr=3.991731e-05 lr_backbone=3.991731e-05 elapsed=12:52:05 avg_step=1.58s recent_step=1.22s eta=09:04:33 +2026-09-20 19:30:22 | INFO | train rank=0 | step=29330/50000 progress=58.7% loss=0.147566 lr=3.988492e-05 lr_backbone=3.988492e-05 elapsed=12:52:22 avg_step=1.58s recent_step=1.72s eta=09:04:19 +2026-09-20 19:30:35 | INFO | train rank=0 | step=29340/50000 progress=58.7% loss=0.126640 lr=3.985254e-05 lr_backbone=3.985254e-05 elapsed=12:52:35 avg_step=1.58s recent_step=1.30s eta=09:04:01 +2026-09-20 19:30:55 | INFO | train rank=0 | step=29350/50000 progress=58.7% loss=0.080420 lr=3.982016e-05 lr_backbone=3.982016e-05 elapsed=12:52:55 avg_step=1.58s recent_step=1.99s eta=09:03:48 +2026-09-20 19:31:11 | INFO | train rank=0 | step=29360/50000 progress=58.7% loss=0.082260 lr=3.978779e-05 lr_backbone=3.978779e-05 elapsed=12:53:11 avg_step=1.58s recent_step=1.62s eta=09:03:32 +2026-09-20 19:31:27 | INFO | train rank=0 | step=29370/50000 progress=58.7% loss=0.089396 lr=3.975542e-05 lr_backbone=3.975542e-05 elapsed=12:53:26 avg_step=1.58s recent_step=1.55s eta=09:03:16 +2026-09-20 19:31:39 | INFO | train rank=0 | step=29380/50000 progress=58.8% loss=0.126267 lr=3.972305e-05 lr_backbone=3.972305e-05 elapsed=12:53:39 avg_step=1.58s recent_step=1.26s eta=09:02:58 +2026-09-20 19:31:59 | INFO | train rank=0 | step=29390/50000 progress=58.8% loss=0.084804 lr=3.969069e-05 lr_backbone=3.969069e-05 elapsed=12:53:58 avg_step=1.58s recent_step=1.95s eta=09:02:45 +2026-09-20 19:32:16 | INFO | train rank=0 | step=29400/50000 progress=58.8% loss=0.090179 lr=3.965833e-05 lr_backbone=3.965833e-05 elapsed=12:54:16 avg_step=1.58s recent_step=1.72s eta=09:02:30 +2026-09-20 19:32:28 | INFO | train rank=0 | step=29410/50000 progress=58.8% loss=0.095961 lr=3.962598e-05 lr_backbone=3.962598e-05 elapsed=12:54:27 avg_step=1.58s recent_step=1.15s eta=09:02:12 +2026-09-20 19:32:39 | INFO | train rank=0 | step=29420/50000 progress=58.8% loss=0.167130 lr=3.959363e-05 lr_backbone=3.959363e-05 elapsed=12:54:39 avg_step=1.58s recent_step=1.19s eta=09:01:53 +2026-09-20 19:33:01 | INFO | train rank=0 | step=29430/50000 progress=58.9% loss=0.085600 lr=3.956129e-05 lr_backbone=3.956129e-05 elapsed=12:55:01 avg_step=1.58s recent_step=2.19s eta=09:01:41 +2026-09-20 19:33:15 | INFO | train rank=0 | step=29440/50000 progress=58.9% loss=0.105956 lr=3.952895e-05 lr_backbone=3.952895e-05 elapsed=12:55:14 avg_step=1.58s recent_step=1.32s eta=09:01:24 +2026-09-20 19:33:30 | INFO | train rank=0 | step=29450/50000 progress=58.9% loss=0.095474 lr=3.949662e-05 lr_backbone=3.949662e-05 elapsed=12:55:30 avg_step=1.58s recent_step=1.55s eta=09:01:08 +2026-09-20 19:33:46 | INFO | train rank=0 | step=29460/50000 progress=58.9% loss=0.089353 lr=3.946429e-05 lr_backbone=3.946429e-05 elapsed=12:55:45 avg_step=1.58s recent_step=1.55s eta=09:00:52 +2026-09-20 19:34:01 | INFO | train rank=0 | step=29470/50000 progress=58.9% loss=0.095927 lr=3.943196e-05 lr_backbone=3.943196e-05 elapsed=12:56:01 avg_step=1.58s recent_step=1.54s eta=09:00:36 +2026-09-20 19:34:17 | INFO | train rank=0 | step=29480/50000 progress=59.0% loss=0.094177 lr=3.939964e-05 lr_backbone=3.939964e-05 elapsed=12:56:16 avg_step=1.58s recent_step=1.59s eta=09:00:20 +2026-09-20 19:34:32 | INFO | train rank=0 | step=29490/50000 progress=59.0% loss=0.092500 lr=3.936733e-05 lr_backbone=3.936733e-05 elapsed=12:56:31 avg_step=1.58s recent_step=1.48s eta=09:00:04 +2026-09-20 19:34:46 | INFO | train rank=0 | step=29500/50000 progress=59.0% loss=0.070859 lr=3.933502e-05 lr_backbone=3.933502e-05 elapsed=12:56:46 avg_step=1.58s recent_step=1.49s eta=08:59:47 +2026-09-20 19:35:06 | INFO | train rank=0 | step=29510/50000 progress=59.0% loss=0.094645 lr=3.930271e-05 lr_backbone=3.930271e-05 elapsed=12:57:06 avg_step=1.58s recent_step=1.96s eta=08:59:34 +2026-09-20 19:35:18 | INFO | train rank=0 | step=29520/50000 progress=59.0% loss=0.094208 lr=3.927041e-05 lr_backbone=3.927041e-05 elapsed=12:57:18 avg_step=1.58s recent_step=1.22s eta=08:59:16 +2026-09-20 19:35:37 | INFO | train rank=0 | step=29530/50000 progress=59.1% loss=0.099935 lr=3.923811e-05 lr_backbone=3.923811e-05 elapsed=12:57:36 avg_step=1.58s recent_step=1.87s eta=08:59:02 +2026-09-20 19:35:50 | INFO | train rank=0 | step=29540/50000 progress=59.1% loss=0.106301 lr=3.920582e-05 lr_backbone=3.920582e-05 elapsed=12:57:50 avg_step=1.58s recent_step=1.31s eta=08:58:44 +2026-09-20 19:36:11 | INFO | train rank=0 | step=29550/50000 progress=59.1% loss=0.179890 lr=3.917354e-05 lr_backbone=3.917354e-05 elapsed=12:58:10 avg_step=1.58s recent_step=2.09s eta=08:58:32 +2026-09-20 19:36:22 | INFO | train rank=0 | step=29560/50000 progress=59.1% loss=0.092389 lr=3.914125e-05 lr_backbone=3.914125e-05 elapsed=12:58:22 avg_step=1.58s recent_step=1.15s eta=08:58:13 +2026-09-20 19:36:35 | INFO | train rank=0 | step=29570/50000 progress=59.1% loss=0.111675 lr=3.910898e-05 lr_backbone=3.910898e-05 elapsed=12:58:34 avg_step=1.58s recent_step=1.25s eta=08:57:55 +2026-09-20 19:36:54 | INFO | train rank=0 | step=29580/50000 progress=59.2% loss=0.099326 lr=3.907670e-05 lr_backbone=3.907670e-05 elapsed=12:58:54 avg_step=1.58s recent_step=1.96s eta=08:57:42 +2026-09-20 19:37:14 | INFO | train rank=0 | step=29590/50000 progress=59.2% loss=0.099589 lr=3.904443e-05 lr_backbone=3.904443e-05 elapsed=12:59:13 avg_step=1.58s recent_step=1.91s eta=08:57:28 +2026-09-20 19:37:25 | INFO | train rank=0 | step=29600/50000 progress=59.2% loss=0.100297 lr=3.901217e-05 lr_backbone=3.901217e-05 elapsed=12:59:25 avg_step=1.58s recent_step=1.17s eta=08:57:10 +2026-09-20 19:37:38 | INFO | train rank=0 | step=29610/50000 progress=59.2% loss=0.119224 lr=3.897991e-05 lr_backbone=3.897991e-05 elapsed=12:59:37 avg_step=1.58s recent_step=1.25s eta=08:56:52 +2026-09-20 19:37:57 | INFO | train rank=0 | step=29620/50000 progress=59.2% loss=0.075775 lr=3.894766e-05 lr_backbone=3.894766e-05 elapsed=12:59:57 avg_step=1.58s recent_step=1.97s eta=08:56:39 +2026-09-20 19:38:10 | INFO | train rank=0 | step=29630/50000 progress=59.3% loss=0.074264 lr=3.891541e-05 lr_backbone=3.891541e-05 elapsed=13:00:09 avg_step=1.58s recent_step=1.24s eta=08:56:20 +2026-09-20 19:38:26 | INFO | train rank=0 | step=29640/50000 progress=59.3% loss=0.091854 lr=3.888317e-05 lr_backbone=3.888317e-05 elapsed=13:00:25 avg_step=1.58s recent_step=1.58s eta=08:56:05 +2026-09-20 19:38:39 | INFO | train rank=0 | step=29650/50000 progress=59.3% loss=0.088443 lr=3.885093e-05 lr_backbone=3.885093e-05 elapsed=13:00:39 avg_step=1.58s recent_step=1.38s eta=08:55:47 +2026-09-20 19:38:57 | INFO | train rank=0 | step=29660/50000 progress=59.3% loss=0.065963 lr=3.881869e-05 lr_backbone=3.881869e-05 elapsed=13:00:56 avg_step=1.58s recent_step=1.71s eta=08:55:33 +2026-09-20 19:39:12 | INFO | train rank=0 | step=29670/50000 progress=59.3% loss=0.070268 lr=3.878646e-05 lr_backbone=3.878646e-05 elapsed=13:01:12 avg_step=1.58s recent_step=1.57s eta=08:55:17 +2026-09-20 19:39:34 | INFO | train rank=0 | step=29680/50000 progress=59.4% loss=0.063649 lr=3.875424e-05 lr_backbone=3.875424e-05 elapsed=13:01:34 avg_step=1.58s recent_step=2.18s eta=08:55:05 +2026-09-20 19:39:46 | INFO | train rank=0 | step=29690/50000 progress=59.4% loss=0.083032 lr=3.872202e-05 lr_backbone=3.872202e-05 elapsed=13:01:46 avg_step=1.58s recent_step=1.23s eta=08:54:47 +2026-09-20 19:39:58 | INFO | train rank=0 | step=29700/50000 progress=59.4% loss=0.095306 lr=3.868980e-05 lr_backbone=3.868980e-05 elapsed=13:01:58 avg_step=1.58s recent_step=1.20s eta=08:54:28 +2026-09-20 19:40:17 | INFO | train rank=0 | step=29710/50000 progress=59.4% loss=0.076848 lr=3.865759e-05 lr_backbone=3.865759e-05 elapsed=13:02:17 avg_step=1.58s recent_step=1.84s eta=08:54:14 +2026-09-20 19:40:32 | INFO | train rank=0 | step=29720/50000 progress=59.4% loss=0.088616 lr=3.862539e-05 lr_backbone=3.862539e-05 elapsed=13:02:32 avg_step=1.58s recent_step=1.53s eta=08:53:58 +2026-09-20 19:40:48 | INFO | train rank=0 | step=29730/50000 progress=59.5% loss=0.122320 lr=3.859319e-05 lr_backbone=3.859319e-05 elapsed=13:02:48 avg_step=1.58s recent_step=1.58s eta=08:53:43 +2026-09-20 19:41:03 | INFO | train rank=0 | step=29740/50000 progress=59.5% loss=0.096081 lr=3.856099e-05 lr_backbone=3.856099e-05 elapsed=13:03:03 avg_step=1.58s recent_step=1.49s eta=08:53:26 +2026-09-20 19:41:18 | INFO | train rank=0 | step=29750/50000 progress=59.5% loss=0.134257 lr=3.852880e-05 lr_backbone=3.852880e-05 elapsed=13:03:18 avg_step=1.58s recent_step=1.52s eta=08:53:10 +2026-09-20 19:41:35 | INFO | train rank=0 | step=29760/50000 progress=59.5% loss=0.096015 lr=3.849662e-05 lr_backbone=3.849662e-05 elapsed=13:03:35 avg_step=1.58s recent_step=1.73s eta=08:52:55 +2026-09-20 19:41:53 | INFO | train rank=0 | step=29770/50000 progress=59.5% loss=0.079973 lr=3.846444e-05 lr_backbone=3.846444e-05 elapsed=13:03:53 avg_step=1.58s recent_step=1.75s eta=08:52:41 +2026-09-20 19:42:07 | INFO | train rank=0 | step=29780/50000 progress=59.6% loss=0.085525 lr=3.843226e-05 lr_backbone=3.843226e-05 elapsed=13:04:06 avg_step=1.58s recent_step=1.38s eta=08:52:23 +2026-09-20 19:42:20 | INFO | train rank=0 | step=29790/50000 progress=59.6% loss=0.089536 lr=3.840009e-05 lr_backbone=3.840009e-05 elapsed=13:04:20 avg_step=1.58s recent_step=1.32s eta=08:52:06 +2026-09-20 19:42:39 | INFO | train rank=0 | step=29800/50000 progress=59.6% loss=0.080166 lr=3.836793e-05 lr_backbone=3.836793e-05 elapsed=13:04:39 avg_step=1.58s recent_step=1.91s eta=08:51:52 +2026-09-20 19:42:55 | INFO | train rank=0 | step=29810/50000 progress=59.6% loss=0.106214 lr=3.833577e-05 lr_backbone=3.833577e-05 elapsed=13:04:55 avg_step=1.58s recent_step=1.62s eta=08:51:37 +2026-09-20 19:43:11 | INFO | train rank=0 | step=29820/50000 progress=59.6% loss=0.100547 lr=3.830362e-05 lr_backbone=3.830362e-05 elapsed=13:05:11 avg_step=1.58s recent_step=1.61s eta=08:51:21 +2026-09-20 19:43:28 | INFO | train rank=0 | step=29830/50000 progress=59.7% loss=0.089008 lr=3.827147e-05 lr_backbone=3.827147e-05 elapsed=13:05:28 avg_step=1.58s recent_step=1.70s eta=08:51:06 +2026-09-20 19:43:40 | INFO | train rank=0 | step=29840/50000 progress=59.7% loss=0.073602 lr=3.823932e-05 lr_backbone=3.823932e-05 elapsed=13:05:40 avg_step=1.58s recent_step=1.17s eta=08:50:48 +2026-09-20 19:43:57 | INFO | train rank=0 | step=29850/50000 progress=59.7% loss=0.069791 lr=3.820718e-05 lr_backbone=3.820718e-05 elapsed=13:05:57 avg_step=1.58s recent_step=1.72s eta=08:50:33 +2026-09-20 19:44:11 | INFO | train rank=0 | step=29860/50000 progress=59.7% loss=0.078038 lr=3.817505e-05 lr_backbone=3.817505e-05 elapsed=13:06:11 avg_step=1.58s recent_step=1.41s eta=08:50:16 +2026-09-20 19:44:28 | INFO | train rank=0 | step=29870/50000 progress=59.7% loss=0.089832 lr=3.814292e-05 lr_backbone=3.814292e-05 elapsed=13:06:28 avg_step=1.58s recent_step=1.69s eta=08:50:01 +2026-09-20 19:44:46 | INFO | train rank=0 | step=29880/50000 progress=59.8% loss=0.107317 lr=3.811080e-05 lr_backbone=3.811080e-05 elapsed=13:06:46 avg_step=1.58s recent_step=1.79s eta=08:49:46 +2026-09-20 19:44:58 | INFO | train rank=0 | step=29890/50000 progress=59.8% loss=0.074081 lr=3.807868e-05 lr_backbone=3.807868e-05 elapsed=13:06:58 avg_step=1.58s recent_step=1.20s eta=08:49:28 +2026-09-20 19:45:16 | INFO | train rank=0 | step=29900/50000 progress=59.8% loss=0.119409 lr=3.804657e-05 lr_backbone=3.804657e-05 elapsed=13:07:16 avg_step=1.58s recent_step=1.80s eta=08:49:14 +2026-09-20 19:45:30 | INFO | train rank=0 | step=29910/50000 progress=59.8% loss=0.087974 lr=3.801446e-05 lr_backbone=3.801446e-05 elapsed=13:07:30 avg_step=1.58s recent_step=1.40s eta=08:48:57 +2026-09-20 19:45:51 | INFO | train rank=0 | step=29920/50000 progress=59.8% loss=0.100756 lr=3.798236e-05 lr_backbone=3.798236e-05 elapsed=13:07:51 avg_step=1.58s recent_step=2.12s eta=08:48:44 +2026-09-20 19:46:03 | INFO | train rank=0 | step=29930/50000 progress=59.9% loss=0.077561 lr=3.795026e-05 lr_backbone=3.795026e-05 elapsed=13:08:03 avg_step=1.58s recent_step=1.16s eta=08:48:26 +2026-09-20 19:46:15 | INFO | train rank=0 | step=29940/50000 progress=59.9% loss=0.070384 lr=3.791817e-05 lr_backbone=3.791817e-05 elapsed=13:08:15 avg_step=1.58s recent_step=1.21s eta=08:48:08 +2026-09-20 19:46:30 | INFO | train rank=0 | step=29950/50000 progress=59.9% loss=0.082437 lr=3.788608e-05 lr_backbone=3.788608e-05 elapsed=13:08:29 avg_step=1.58s recent_step=1.46s eta=08:47:51 +2026-09-20 19:46:55 | INFO | train rank=0 | step=29960/50000 progress=59.9% loss=0.076347 lr=3.785400e-05 lr_backbone=3.785400e-05 elapsed=13:08:54 avg_step=1.58s recent_step=2.50s eta=08:47:41 +2026-09-20 19:47:06 | INFO | train rank=0 | step=29970/50000 progress=59.9% loss=0.073461 lr=3.782192e-05 lr_backbone=3.782192e-05 elapsed=13:09:06 avg_step=1.58s recent_step=1.19s eta=08:47:23 +2026-09-20 19:47:18 | INFO | train rank=0 | step=29980/50000 progress=60.0% loss=0.082608 lr=3.778985e-05 lr_backbone=3.778985e-05 elapsed=13:09:18 avg_step=1.58s recent_step=1.17s eta=08:47:04 +2026-09-20 19:47:35 | INFO | train rank=0 | step=29990/50000 progress=60.0% loss=0.082195 lr=3.775779e-05 lr_backbone=3.775779e-05 elapsed=13:09:34 avg_step=1.58s recent_step=1.65s eta=08:46:49 +2026-09-20 19:47:48 | INFO | train rank=0 | step=30000/50000 progress=60.0% loss=0.089332 lr=3.772573e-05 lr_backbone=3.772573e-05 elapsed=13:09:47 avg_step=1.58s recent_step=1.31s eta=08:46:31 +2026-09-20 19:48:02 | INFO | train rank=0 | step=30010/50000 progress=60.0% loss=0.164078 lr=3.769367e-05 lr_backbone=3.769367e-05 elapsed=13:10:02 avg_step=1.58s recent_step=1.44s eta=08:46:15 +2026-09-20 19:48:23 | INFO | train rank=0 | step=30020/50000 progress=60.0% loss=0.077254 lr=3.766162e-05 lr_backbone=3.766162e-05 elapsed=13:10:23 avg_step=1.58s recent_step=2.09s eta=08:46:02 +2026-09-20 19:48:35 | INFO | train rank=0 | step=30030/50000 progress=60.1% loss=0.103534 lr=3.762958e-05 lr_backbone=3.762958e-05 elapsed=13:10:35 avg_step=1.58s recent_step=1.20s eta=08:45:44 +2026-09-20 19:48:55 | INFO | train rank=0 | step=30040/50000 progress=60.1% loss=0.064095 lr=3.759754e-05 lr_backbone=3.759754e-05 elapsed=13:10:54 avg_step=1.58s recent_step=1.98s eta=08:45:31 +2026-09-20 19:49:10 | INFO | train rank=0 | step=30050/50000 progress=60.1% loss=0.099819 lr=3.756551e-05 lr_backbone=3.756551e-05 elapsed=13:11:10 avg_step=1.58s recent_step=1.55s eta=08:45:15 +2026-09-20 19:49:24 | INFO | train rank=0 | step=30060/50000 progress=60.1% loss=0.095213 lr=3.753348e-05 lr_backbone=3.753348e-05 elapsed=13:11:23 avg_step=1.58s recent_step=1.35s eta=08:44:57 +2026-09-20 19:49:36 | INFO | train rank=0 | step=30070/50000 progress=60.1% loss=0.128830 lr=3.750146e-05 lr_backbone=3.750146e-05 elapsed=13:11:36 avg_step=1.58s recent_step=1.24s eta=08:44:39 +2026-09-20 19:49:53 | INFO | train rank=0 | step=30080/50000 progress=60.2% loss=0.110288 lr=3.746944e-05 lr_backbone=3.746944e-05 elapsed=13:11:52 avg_step=1.58s recent_step=1.65s eta=08:44:24 +2026-09-20 19:50:14 | INFO | train rank=0 | step=30090/50000 progress=60.2% loss=0.080582 lr=3.743743e-05 lr_backbone=3.743743e-05 elapsed=13:12:14 avg_step=1.58s recent_step=2.18s eta=08:44:12 +2026-09-20 19:50:26 | INFO | train rank=0 | step=30100/50000 progress=60.2% loss=0.080287 lr=3.740542e-05 lr_backbone=3.740542e-05 elapsed=13:12:26 avg_step=1.58s recent_step=1.16s eta=08:43:54 +2026-09-20 19:50:38 | INFO | train rank=0 | step=30110/50000 progress=60.2% loss=0.088253 lr=3.737342e-05 lr_backbone=3.737342e-05 elapsed=13:12:37 avg_step=1.58s recent_step=1.19s eta=08:43:35 +2026-09-20 19:51:02 | INFO | train rank=0 | step=30120/50000 progress=60.2% loss=0.131754 lr=3.734143e-05 lr_backbone=3.734143e-05 elapsed=13:13:01 avg_step=1.58s recent_step=2.39s eta=08:43:25 +2026-09-20 19:51:17 | INFO | train rank=0 | step=30130/50000 progress=60.3% loss=0.074352 lr=3.730944e-05 lr_backbone=3.730944e-05 elapsed=13:13:17 avg_step=1.58s recent_step=1.56s eta=08:43:09 +2026-09-20 19:51:36 | INFO | train rank=0 | step=30140/50000 progress=60.3% loss=0.075007 lr=3.727745e-05 lr_backbone=3.727745e-05 elapsed=13:13:36 avg_step=1.58s recent_step=1.89s eta=08:42:55 +2026-09-20 19:51:48 | INFO | train rank=0 | step=30150/50000 progress=60.3% loss=0.082836 lr=3.724548e-05 lr_backbone=3.724548e-05 elapsed=13:13:47 avg_step=1.58s recent_step=1.15s eta=08:42:36 +2026-09-20 19:52:03 | INFO | train rank=0 | step=30160/50000 progress=60.3% loss=0.116518 lr=3.721350e-05 lr_backbone=3.721350e-05 elapsed=13:14:03 avg_step=1.58s recent_step=1.57s eta=08:42:21 +2026-09-20 19:52:21 | INFO | train rank=0 | step=30170/50000 progress=60.3% loss=0.098285 lr=3.718154e-05 lr_backbone=3.718154e-05 elapsed=13:14:20 avg_step=1.58s recent_step=1.71s eta=08:42:06 +2026-09-20 19:52:34 | INFO | train rank=0 | step=30180/50000 progress=60.4% loss=0.105704 lr=3.714957e-05 lr_backbone=3.714957e-05 elapsed=13:14:33 avg_step=1.58s recent_step=1.30s eta=08:41:48 +2026-09-20 19:52:49 | INFO | train rank=0 | step=30190/50000 progress=60.4% loss=0.069698 lr=3.711762e-05 lr_backbone=3.711762e-05 elapsed=13:14:49 avg_step=1.58s recent_step=1.54s eta=08:41:32 +2026-09-20 19:53:07 | INFO | train rank=0 | step=30200/50000 progress=60.4% loss=0.074468 lr=3.708567e-05 lr_backbone=3.708567e-05 elapsed=13:15:07 avg_step=1.58s recent_step=1.79s eta=08:41:18 +2026-09-20 19:53:23 | INFO | train rank=0 | step=30210/50000 progress=60.4% loss=0.085252 lr=3.705372e-05 lr_backbone=3.705372e-05 elapsed=13:15:23 avg_step=1.58s recent_step=1.60s eta=08:41:02 +2026-09-20 19:53:40 | INFO | train rank=0 | step=30220/50000 progress=60.4% loss=0.096099 lr=3.702179e-05 lr_backbone=3.702179e-05 elapsed=13:15:40 avg_step=1.58s recent_step=1.72s eta=08:40:47 +2026-09-20 19:53:55 | INFO | train rank=0 | step=30230/50000 progress=60.5% loss=0.073215 lr=3.698985e-05 lr_backbone=3.698985e-05 elapsed=13:15:54 avg_step=1.58s recent_step=1.46s eta=08:40:31 +2026-09-20 19:54:10 | INFO | train rank=0 | step=30240/50000 progress=60.5% loss=0.082137 lr=3.695792e-05 lr_backbone=3.695792e-05 elapsed=13:16:10 avg_step=1.58s recent_step=1.52s eta=08:40:14 +2026-09-20 19:54:27 | INFO | train rank=0 | step=30250/50000 progress=60.5% loss=0.069947 lr=3.692600e-05 lr_backbone=3.692600e-05 elapsed=13:16:26 avg_step=1.58s recent_step=1.69s eta=08:39:59 +2026-09-20 19:54:45 | INFO | train rank=0 | step=30260/50000 progress=60.5% loss=0.077940 lr=3.689409e-05 lr_backbone=3.689409e-05 elapsed=13:16:44 avg_step=1.58s recent_step=1.78s eta=08:39:45 +2026-09-20 19:54:56 | INFO | train rank=0 | step=30270/50000 progress=60.5% loss=0.097491 lr=3.686218e-05 lr_backbone=3.686218e-05 elapsed=13:16:56 avg_step=1.58s recent_step=1.17s eta=08:39:26 +2026-09-20 19:55:16 | INFO | train rank=0 | step=30280/50000 progress=60.6% loss=0.091420 lr=3.683027e-05 lr_backbone=3.683027e-05 elapsed=13:17:16 avg_step=1.58s recent_step=1.99s eta=08:39:13 +2026-09-20 19:55:28 | INFO | train rank=0 | step=30290/50000 progress=60.6% loss=0.078627 lr=3.679837e-05 lr_backbone=3.679837e-05 elapsed=13:17:28 avg_step=1.58s recent_step=1.16s eta=08:38:55 +2026-09-20 19:55:44 | INFO | train rank=0 | step=30300/50000 progress=60.6% loss=0.078305 lr=3.676648e-05 lr_backbone=3.676648e-05 elapsed=13:17:43 avg_step=1.58s recent_step=1.58s eta=08:38:39 +2026-09-20 19:55:59 | INFO | train rank=0 | step=30310/50000 progress=60.6% loss=0.094018 lr=3.673459e-05 lr_backbone=3.673459e-05 elapsed=13:17:59 avg_step=1.58s recent_step=1.55s eta=08:38:23 +2026-09-20 19:56:18 | INFO | train rank=0 | step=30320/50000 progress=60.6% loss=0.074283 lr=3.670271e-05 lr_backbone=3.670271e-05 elapsed=13:18:17 avg_step=1.58s recent_step=1.87s eta=08:38:09 +2026-09-20 19:56:30 | INFO | train rank=0 | step=30330/50000 progress=60.7% loss=0.097265 lr=3.667084e-05 lr_backbone=3.667084e-05 elapsed=13:18:30 avg_step=1.58s recent_step=1.20s eta=08:37:51 +2026-09-20 19:56:43 | INFO | train rank=0 | step=30340/50000 progress=60.7% loss=0.077690 lr=3.663897e-05 lr_backbone=3.663897e-05 elapsed=13:18:43 avg_step=1.58s recent_step=1.33s eta=08:37:33 +2026-09-20 19:56:59 | INFO | train rank=0 | step=30350/50000 progress=60.7% loss=0.089082 lr=3.660710e-05 lr_backbone=3.660710e-05 elapsed=13:18:58 avg_step=1.58s recent_step=1.55s eta=08:37:17 +2026-09-20 19:57:21 | INFO | train rank=0 | step=30360/50000 progress=60.7% loss=0.088934 lr=3.657524e-05 lr_backbone=3.657524e-05 elapsed=13:19:20 avg_step=1.58s recent_step=2.21s eta=08:37:06 +2026-09-20 19:57:33 | INFO | train rank=0 | step=30370/50000 progress=60.7% loss=0.091384 lr=3.654339e-05 lr_backbone=3.654339e-05 elapsed=13:19:32 avg_step=1.58s recent_step=1.19s eta=08:36:47 +2026-09-20 19:57:48 | INFO | train rank=0 | step=30380/50000 progress=60.8% loss=0.081762 lr=3.651155e-05 lr_backbone=3.651155e-05 elapsed=13:19:48 avg_step=1.58s recent_step=1.54s eta=08:36:31 +2026-09-20 19:58:05 | INFO | train rank=0 | step=30390/50000 progress=60.8% loss=0.091607 lr=3.647971e-05 lr_backbone=3.647971e-05 elapsed=13:20:04 avg_step=1.58s recent_step=1.68s eta=08:36:16 +2026-09-20 19:58:22 | INFO | train rank=0 | step=30400/50000 progress=60.8% loss=0.101370 lr=3.644787e-05 lr_backbone=3.644787e-05 elapsed=13:20:21 avg_step=1.58s recent_step=1.70s eta=08:36:01 +2026-09-20 19:58:36 | INFO | train rank=0 | step=30410/50000 progress=60.8% loss=0.077432 lr=3.641604e-05 lr_backbone=3.641604e-05 elapsed=13:20:35 avg_step=1.58s recent_step=1.37s eta=08:35:44 +2026-09-20 19:58:54 | INFO | train rank=0 | step=30420/50000 progress=60.8% loss=0.108830 lr=3.638422e-05 lr_backbone=3.638422e-05 elapsed=13:20:53 avg_step=1.58s recent_step=1.81s eta=08:35:30 +2026-09-20 19:59:05 | INFO | train rank=0 | step=30430/50000 progress=60.9% loss=0.089495 lr=3.635240e-05 lr_backbone=3.635240e-05 elapsed=13:21:05 avg_step=1.58s recent_step=1.17s eta=08:35:11 +2026-09-20 19:59:24 | INFO | train rank=0 | step=30440/50000 progress=60.9% loss=0.083507 lr=3.632059e-05 lr_backbone=3.632059e-05 elapsed=13:21:24 avg_step=1.58s recent_step=1.90s eta=08:34:57 +2026-09-20 19:59:40 | INFO | train rank=0 | step=30450/50000 progress=60.9% loss=0.075998 lr=3.628879e-05 lr_backbone=3.628879e-05 elapsed=13:21:40 avg_step=1.58s recent_step=1.57s eta=08:34:41 +2026-09-20 19:59:52 | INFO | train rank=0 | step=30460/50000 progress=60.9% loss=0.110297 lr=3.625699e-05 lr_backbone=3.625699e-05 elapsed=13:21:51 avg_step=1.58s recent_step=1.16s eta=08:34:23 +2026-09-20 20:00:04 | INFO | train rank=0 | step=30470/50000 progress=60.9% loss=0.087578 lr=3.622520e-05 lr_backbone=3.622520e-05 elapsed=13:22:03 avg_step=1.58s recent_step=1.22s eta=08:34:05 +2026-09-20 20:00:22 | INFO | train rank=0 | step=30480/50000 progress=61.0% loss=0.105683 lr=3.619341e-05 lr_backbone=3.619341e-05 elapsed=13:22:22 avg_step=1.58s recent_step=1.83s eta=08:33:51 +2026-09-20 20:00:34 | INFO | train rank=0 | step=30490/50000 progress=61.0% loss=0.098284 lr=3.616163e-05 lr_backbone=3.616163e-05 elapsed=13:22:33 avg_step=1.58s recent_step=1.17s eta=08:33:32 +2026-09-20 20:00:53 | INFO | train rank=0 | step=30500/50000 progress=61.0% loss=0.081764 lr=3.612985e-05 lr_backbone=3.612985e-05 elapsed=13:22:52 avg_step=1.58s recent_step=1.91s eta=08:33:19 +2026-09-20 20:01:06 | INFO | train rank=0 | step=30510/50000 progress=61.0% loss=0.088329 lr=3.609809e-05 lr_backbone=3.609809e-05 elapsed=13:23:05 avg_step=1.58s recent_step=1.29s eta=08:33:01 +2026-09-20 20:01:24 | INFO | train rank=0 | step=30520/50000 progress=61.0% loss=0.101955 lr=3.606632e-05 lr_backbone=3.606632e-05 elapsed=13:23:24 avg_step=1.58s recent_step=1.85s eta=08:32:47 +2026-09-20 20:01:46 | INFO | train rank=0 | step=30530/50000 progress=61.1% loss=0.139206 lr=3.603457e-05 lr_backbone=3.603457e-05 elapsed=13:23:45 avg_step=1.58s recent_step=2.15s eta=08:32:35 +2026-09-20 20:02:00 | INFO | train rank=0 | step=30540/50000 progress=61.1% loss=0.076760 lr=3.600282e-05 lr_backbone=3.600282e-05 elapsed=13:23:59 avg_step=1.58s recent_step=1.39s eta=08:32:18 +2026-09-20 20:02:23 | INFO | train rank=0 | step=30550/50000 progress=61.1% loss=0.067525 lr=3.597107e-05 lr_backbone=3.597107e-05 elapsed=13:24:22 avg_step=1.58s recent_step=2.31s eta=08:32:07 +2026-09-20 20:02:34 | INFO | train rank=0 | step=30560/50000 progress=61.1% loss=0.092412 lr=3.593934e-05 lr_backbone=3.593934e-05 elapsed=13:24:34 avg_step=1.58s recent_step=1.17s eta=08:31:48 +2026-09-20 20:03:32 | INFO | train rank=0 | step=30570/50000 progress=61.1% loss=0.084123 lr=3.590760e-05 lr_backbone=3.590760e-05 elapsed=13:25:32 avg_step=1.58s recent_step=5.77s eta=08:31:59 +2026-09-20 20:03:44 | INFO | train rank=0 | step=30580/50000 progress=61.2% loss=0.075659 lr=3.587588e-05 lr_backbone=3.587588e-05 elapsed=13:25:44 avg_step=1.58s recent_step=1.22s eta=08:31:41 +2026-09-20 20:03:58 | INFO | train rank=0 | step=30590/50000 progress=61.2% loss=0.081256 lr=3.584416e-05 lr_backbone=3.584416e-05 elapsed=13:25:57 avg_step=1.58s recent_step=1.34s eta=08:31:24 +2026-09-20 20:04:16 | INFO | train rank=0 | step=30600/50000 progress=61.2% loss=0.068801 lr=3.581244e-05 lr_backbone=3.581244e-05 elapsed=13:26:16 avg_step=1.58s recent_step=1.85s eta=08:31:10 +2026-09-20 20:04:29 | INFO | train rank=0 | step=30610/50000 progress=61.2% loss=0.083743 lr=3.578074e-05 lr_backbone=3.578074e-05 elapsed=13:26:28 avg_step=1.58s recent_step=1.25s eta=08:30:52 +2026-09-20 20:04:47 | INFO | train rank=0 | step=30620/50000 progress=61.2% loss=0.094954 lr=3.574904e-05 lr_backbone=3.574904e-05 elapsed=13:26:47 avg_step=1.58s recent_step=1.82s eta=08:30:37 +2026-09-20 20:05:03 | INFO | train rank=0 | step=30630/50000 progress=61.3% loss=0.094081 lr=3.571734e-05 lr_backbone=3.571734e-05 elapsed=13:27:03 avg_step=1.58s recent_step=1.62s eta=08:30:22 +2026-09-20 20:05:20 | INFO | train rank=0 | step=30640/50000 progress=61.3% loss=0.105170 lr=3.568565e-05 lr_backbone=3.568565e-05 elapsed=13:27:19 avg_step=1.58s recent_step=1.64s eta=08:30:06 +2026-09-20 20:05:35 | INFO | train rank=0 | step=30650/50000 progress=61.3% loss=0.108194 lr=3.565397e-05 lr_backbone=3.565397e-05 elapsed=13:27:35 avg_step=1.58s recent_step=1.52s eta=08:29:50 +2026-09-20 20:05:50 | INFO | train rank=0 | step=30660/50000 progress=61.3% loss=0.079146 lr=3.562230e-05 lr_backbone=3.562230e-05 elapsed=13:27:49 avg_step=1.58s recent_step=1.47s eta=08:29:34 +2026-09-20 20:06:04 | INFO | train rank=0 | step=30670/50000 progress=61.3% loss=0.099002 lr=3.559063e-05 lr_backbone=3.559063e-05 elapsed=13:28:03 avg_step=1.58s recent_step=1.42s eta=08:29:17 +2026-09-20 20:06:22 | INFO | train rank=0 | step=30680/50000 progress=61.4% loss=0.089571 lr=3.555896e-05 lr_backbone=3.555896e-05 elapsed=13:28:22 avg_step=1.58s recent_step=1.82s eta=08:29:03 +2026-09-20 20:06:34 | INFO | train rank=0 | step=30690/50000 progress=61.4% loss=0.098508 lr=3.552731e-05 lr_backbone=3.552731e-05 elapsed=13:28:34 avg_step=1.58s recent_step=1.22s eta=08:28:45 +2026-09-20 20:06:52 | INFO | train rank=0 | step=30700/50000 progress=61.4% loss=0.091777 lr=3.549566e-05 lr_backbone=3.549566e-05 elapsed=13:28:52 avg_step=1.58s recent_step=1.79s eta=08:28:30 +2026-09-20 20:07:04 | INFO | train rank=0 | step=30710/50000 progress=61.4% loss=0.082760 lr=3.546401e-05 lr_backbone=3.546401e-05 elapsed=13:29:04 avg_step=1.58s recent_step=1.21s eta=08:28:12 +2026-09-20 20:07:22 | INFO | train rank=0 | step=30720/50000 progress=61.4% loss=0.079298 lr=3.543237e-05 lr_backbone=3.543237e-05 elapsed=13:29:22 avg_step=1.58s recent_step=1.82s eta=08:27:58 +2026-09-20 20:07:39 | INFO | train rank=0 | step=30730/50000 progress=61.5% loss=0.082422 lr=3.540074e-05 lr_backbone=3.540074e-05 elapsed=13:29:39 avg_step=1.58s recent_step=1.69s eta=08:27:42 +2026-09-20 20:07:54 | INFO | train rank=0 | step=30740/50000 progress=61.5% loss=0.087401 lr=3.536912e-05 lr_backbone=3.536912e-05 elapsed=13:29:54 avg_step=1.58s recent_step=1.48s eta=08:27:26 +2026-09-20 20:08:14 | INFO | train rank=0 | step=30750/50000 progress=61.5% loss=0.106176 lr=3.533750e-05 lr_backbone=3.533750e-05 elapsed=13:30:13 avg_step=1.58s recent_step=1.96s eta=08:27:13 +2026-09-20 20:08:26 | INFO | train rank=0 | step=30760/50000 progress=61.5% loss=0.083373 lr=3.530589e-05 lr_backbone=3.530589e-05 elapsed=13:30:26 avg_step=1.58s recent_step=1.22s eta=08:26:55 +2026-09-20 20:08:37 | INFO | train rank=0 | step=30770/50000 progress=61.5% loss=0.112218 lr=3.527428e-05 lr_backbone=3.527428e-05 elapsed=13:30:37 avg_step=1.58s recent_step=1.15s eta=08:26:36 +2026-09-20 20:08:50 | INFO | train rank=0 | step=30780/50000 progress=61.6% loss=0.075426 lr=3.524268e-05 lr_backbone=3.524268e-05 elapsed=13:30:50 avg_step=1.58s recent_step=1.24s eta=08:26:18 +2026-09-20 20:09:05 | INFO | train rank=0 | step=30790/50000 progress=61.6% loss=0.063919 lr=3.521109e-05 lr_backbone=3.521109e-05 elapsed=13:31:05 avg_step=1.58s recent_step=1.52s eta=08:26:02 +2026-09-20 20:09:22 | INFO | train rank=0 | step=30800/50000 progress=61.6% loss=0.108745 lr=3.517950e-05 lr_backbone=3.517950e-05 elapsed=13:31:22 avg_step=1.58s recent_step=1.71s eta=08:25:47 +2026-09-20 20:09:37 | INFO | train rank=0 | step=30810/50000 progress=61.6% loss=0.084886 lr=3.514792e-05 lr_backbone=3.514792e-05 elapsed=13:31:36 avg_step=1.58s recent_step=1.47s eta=08:25:30 +2026-09-20 20:09:54 | INFO | train rank=0 | step=30820/50000 progress=61.6% loss=0.113880 lr=3.511635e-05 lr_backbone=3.511635e-05 elapsed=13:31:54 avg_step=1.58s recent_step=1.72s eta=08:25:15 +2026-09-20 20:10:11 | INFO | train rank=0 | step=30830/50000 progress=61.7% loss=0.101168 lr=3.508478e-05 lr_backbone=3.508478e-05 elapsed=13:32:11 avg_step=1.58s recent_step=1.72s eta=08:25:01 +2026-09-20 20:10:27 | INFO | train rank=0 | step=30840/50000 progress=61.7% loss=0.084326 lr=3.505322e-05 lr_backbone=3.505322e-05 elapsed=13:32:27 avg_step=1.58s recent_step=1.56s eta=08:24:45 +2026-09-20 20:10:43 | INFO | train rank=0 | step=30850/50000 progress=61.7% loss=0.083969 lr=3.502167e-05 lr_backbone=3.502167e-05 elapsed=13:32:42 avg_step=1.58s recent_step=1.57s eta=08:24:29 +2026-09-20 20:10:55 | INFO | train rank=0 | step=30860/50000 progress=61.7% loss=0.093231 lr=3.499012e-05 lr_backbone=3.499012e-05 elapsed=13:32:55 avg_step=1.58s recent_step=1.27s eta=08:24:11 +2026-09-20 20:11:19 | INFO | train rank=0 | step=30870/50000 progress=61.7% loss=0.094932 lr=3.495858e-05 lr_backbone=3.495858e-05 elapsed=13:33:18 avg_step=1.58s recent_step=2.35s eta=08:24:00 +2026-09-20 20:11:30 | INFO | train rank=0 | step=30880/50000 progress=61.8% loss=0.086451 lr=3.492704e-05 lr_backbone=3.492704e-05 elapsed=13:33:30 avg_step=1.58s recent_step=1.14s eta=08:23:41 +2026-09-20 20:11:44 | INFO | train rank=0 | step=30890/50000 progress=61.8% loss=0.095160 lr=3.489552e-05 lr_backbone=3.489552e-05 elapsed=13:33:43 avg_step=1.58s recent_step=1.35s eta=08:23:24 +2026-09-20 20:11:57 | INFO | train rank=0 | step=30900/50000 progress=61.8% loss=0.080561 lr=3.486399e-05 lr_backbone=3.486399e-05 elapsed=13:33:57 avg_step=1.58s recent_step=1.36s eta=08:23:07 +2026-09-20 20:12:13 | INFO | train rank=0 | step=30910/50000 progress=61.8% loss=0.092090 lr=3.483248e-05 lr_backbone=3.483248e-05 elapsed=13:34:13 avg_step=1.58s recent_step=1.59s eta=08:22:51 +2026-09-20 20:12:32 | INFO | train rank=0 | step=30920/50000 progress=61.8% loss=0.081306 lr=3.480097e-05 lr_backbone=3.480097e-05 elapsed=13:34:32 avg_step=1.58s recent_step=1.86s eta=08:22:37 +2026-09-20 20:12:44 | INFO | train rank=0 | step=30930/50000 progress=61.9% loss=0.096487 lr=3.476947e-05 lr_backbone=3.476947e-05 elapsed=13:34:44 avg_step=1.58s recent_step=1.26s eta=08:22:19 +2026-09-20 20:13:07 | INFO | train rank=0 | step=30940/50000 progress=61.9% loss=0.087059 lr=3.473798e-05 lr_backbone=3.473798e-05 elapsed=13:35:06 avg_step=1.58s recent_step=2.20s eta=08:22:08 +2026-09-20 20:13:18 | INFO | train rank=0 | step=30950/50000 progress=61.9% loss=0.061447 lr=3.470649e-05 lr_backbone=3.470649e-05 elapsed=13:35:18 avg_step=1.58s recent_step=1.19s eta=08:21:49 +2026-09-20 20:13:37 | INFO | train rank=0 | step=30960/50000 progress=61.9% loss=0.084849 lr=3.467501e-05 lr_backbone=3.467501e-05 elapsed=13:35:37 avg_step=1.58s recent_step=1.85s eta=08:21:35 +2026-09-20 20:13:53 | INFO | train rank=0 | step=30970/50000 progress=61.9% loss=0.093882 lr=3.464353e-05 lr_backbone=3.464353e-05 elapsed=13:35:52 avg_step=1.58s recent_step=1.57s eta=08:21:19 +2026-09-20 20:14:04 | INFO | train rank=0 | step=30980/50000 progress=62.0% loss=0.077929 lr=3.461207e-05 lr_backbone=3.461207e-05 elapsed=13:36:04 avg_step=1.58s recent_step=1.16s eta=08:21:01 +2026-09-20 20:14:19 | INFO | train rank=0 | step=30990/50000 progress=62.0% loss=0.100620 lr=3.458060e-05 lr_backbone=3.458060e-05 elapsed=13:36:19 avg_step=1.58s recent_step=1.50s eta=08:20:45 +2026-09-20 20:14:37 | INFO | train rank=0 | step=31000/50000 progress=62.0% loss=0.095550 lr=3.454915e-05 lr_backbone=3.454915e-05 elapsed=13:36:37 avg_step=1.58s recent_step=1.81s eta=08:20:30 +2026-09-20 20:14:58 | INFO | train rank=0 | step=31010/50000 progress=62.0% loss=0.079652 lr=3.451770e-05 lr_backbone=3.451770e-05 elapsed=13:36:58 avg_step=1.58s recent_step=2.10s eta=08:20:18 +2026-09-20 20:15:10 | INFO | train rank=0 | step=31020/50000 progress=62.0% loss=0.097297 lr=3.448626e-05 lr_backbone=3.448626e-05 elapsed=13:37:09 avg_step=1.58s recent_step=1.12s eta=08:19:59 +2026-09-20 20:15:22 | INFO | train rank=0 | step=31030/50000 progress=62.1% loss=0.084377 lr=3.445483e-05 lr_backbone=3.445483e-05 elapsed=13:37:21 avg_step=1.58s recent_step=1.22s eta=08:19:41 +2026-09-20 20:15:36 | INFO | train rank=0 | step=31040/50000 progress=62.1% loss=0.087021 lr=3.442340e-05 lr_backbone=3.442340e-05 elapsed=13:37:36 avg_step=1.58s recent_step=1.42s eta=08:19:24 +2026-09-20 20:15:54 | INFO | train rank=0 | step=31050/50000 progress=62.1% loss=0.108348 lr=3.439198e-05 lr_backbone=3.439198e-05 elapsed=13:37:53 avg_step=1.58s recent_step=1.78s eta=08:19:10 +2026-09-20 20:16:09 | INFO | train rank=0 | step=31060/50000 progress=62.1% loss=0.072712 lr=3.436057e-05 lr_backbone=3.436057e-05 elapsed=13:38:08 avg_step=1.58s recent_step=1.48s eta=08:18:53 +2026-09-20 20:16:25 | INFO | train rank=0 | step=31070/50000 progress=62.1% loss=0.091130 lr=3.432916e-05 lr_backbone=3.432916e-05 elapsed=13:38:24 avg_step=1.58s recent_step=1.62s eta=08:18:38 +2026-09-20 20:16:39 | INFO | train rank=0 | step=31080/50000 progress=62.2% loss=0.101355 lr=3.429776e-05 lr_backbone=3.429776e-05 elapsed=13:38:39 avg_step=1.58s recent_step=1.42s eta=08:18:21 +2026-09-20 20:16:55 | INFO | train rank=0 | step=31090/50000 progress=62.2% loss=0.083659 lr=3.426637e-05 lr_backbone=3.426637e-05 elapsed=13:38:55 avg_step=1.58s recent_step=1.63s eta=08:18:05 +2026-09-20 20:17:14 | INFO | train rank=0 | step=31100/50000 progress=62.2% loss=0.104752 lr=3.423498e-05 lr_backbone=3.423498e-05 elapsed=13:39:14 avg_step=1.58s recent_step=1.92s eta=08:17:52 +2026-09-20 20:17:28 | INFO | train rank=0 | step=31110/50000 progress=62.2% loss=0.066446 lr=3.420360e-05 lr_backbone=3.420360e-05 elapsed=13:39:28 avg_step=1.58s recent_step=1.41s eta=08:17:35 +2026-09-20 20:17:48 | INFO | train rank=0 | step=31120/50000 progress=62.2% loss=0.082952 lr=3.417223e-05 lr_backbone=3.417223e-05 elapsed=13:39:47 avg_step=1.58s recent_step=1.93s eta=08:17:21 +2026-09-20 20:17:59 | INFO | train rank=0 | step=31130/50000 progress=62.3% loss=0.096710 lr=3.414087e-05 lr_backbone=3.414087e-05 elapsed=13:39:58 avg_step=1.58s recent_step=1.10s eta=08:17:02 +2026-09-20 20:18:17 | INFO | train rank=0 | step=31140/50000 progress=62.3% loss=0.079318 lr=3.410951e-05 lr_backbone=3.410951e-05 elapsed=13:40:17 avg_step=1.58s recent_step=1.83s eta=08:16:48 +2026-09-20 20:18:29 | INFO | train rank=0 | step=31150/50000 progress=62.3% loss=0.081483 lr=3.407816e-05 lr_backbone=3.407816e-05 elapsed=13:40:29 avg_step=1.58s recent_step=1.20s eta=08:16:30 +2026-09-20 20:18:43 | INFO | train rank=0 | step=31160/50000 progress=62.3% loss=0.086634 lr=3.404681e-05 lr_backbone=3.404681e-05 elapsed=13:40:43 avg_step=1.58s recent_step=1.38s eta=08:16:13 +2026-09-20 20:19:01 | INFO | train rank=0 | step=31170/50000 progress=62.3% loss=0.080463 lr=3.401547e-05 lr_backbone=3.401547e-05 elapsed=13:41:01 avg_step=1.58s recent_step=1.85s eta=08:15:59 +2026-09-20 20:19:21 | INFO | train rank=0 | step=31180/50000 progress=62.4% loss=0.070666 lr=3.398414e-05 lr_backbone=3.398414e-05 elapsed=13:41:21 avg_step=1.58s recent_step=1.97s eta=08:15:45 +2026-09-20 20:19:33 | INFO | train rank=0 | step=31190/50000 progress=62.4% loss=0.076922 lr=3.395282e-05 lr_backbone=3.395282e-05 elapsed=13:41:32 avg_step=1.58s recent_step=1.15s eta=08:15:27 +2026-09-20 20:19:46 | INFO | train rank=0 | step=31200/50000 progress=62.4% loss=0.091021 lr=3.392150e-05 lr_backbone=3.392150e-05 elapsed=13:41:46 avg_step=1.58s recent_step=1.32s eta=08:15:10 +2026-09-20 20:20:03 | INFO | train rank=0 | step=31210/50000 progress=62.4% loss=0.085900 lr=3.389019e-05 lr_backbone=3.389019e-05 elapsed=13:42:03 avg_step=1.58s recent_step=1.76s eta=08:14:55 +2026-09-20 20:20:23 | INFO | train rank=0 | step=31220/50000 progress=62.4% loss=0.079144 lr=3.385889e-05 lr_backbone=3.385889e-05 elapsed=13:42:22 avg_step=1.58s recent_step=1.94s eta=08:14:41 +2026-09-20 20:20:40 | INFO | train rank=0 | step=31230/50000 progress=62.5% loss=0.079242 lr=3.382760e-05 lr_backbone=3.382760e-05 elapsed=13:42:39 avg_step=1.58s recent_step=1.68s eta=08:14:26 +2026-09-20 20:20:51 | INFO | train rank=0 | step=31240/50000 progress=62.5% loss=0.081425 lr=3.379631e-05 lr_backbone=3.379631e-05 elapsed=13:42:51 avg_step=1.58s recent_step=1.13s eta=08:14:07 +2026-09-20 20:21:09 | INFO | train rank=0 | step=31250/50000 progress=62.5% loss=0.078834 lr=3.376503e-05 lr_backbone=3.376503e-05 elapsed=13:43:08 avg_step=1.58s recent_step=1.77s eta=08:13:53 +2026-09-20 20:21:21 | INFO | train rank=0 | step=31260/50000 progress=62.5% loss=0.078262 lr=3.373375e-05 lr_backbone=3.373375e-05 elapsed=13:43:21 avg_step=1.58s recent_step=1.26s eta=08:13:35 +2026-09-20 20:21:37 | INFO | train rank=0 | step=31270/50000 progress=62.5% loss=0.094983 lr=3.370249e-05 lr_backbone=3.370249e-05 elapsed=13:43:37 avg_step=1.58s recent_step=1.59s eta=08:13:19 +2026-09-20 20:21:56 | INFO | train rank=0 | step=31280/50000 progress=62.6% loss=0.083020 lr=3.367123e-05 lr_backbone=3.367123e-05 elapsed=13:43:55 avg_step=1.58s recent_step=1.86s eta=08:13:05 +2026-09-20 20:22:13 | INFO | train rank=0 | step=31290/50000 progress=62.6% loss=0.087184 lr=3.363997e-05 lr_backbone=3.363997e-05 elapsed=13:44:13 avg_step=1.58s recent_step=1.74s eta=08:12:50 +2026-09-20 20:22:25 | INFO | train rank=0 | step=31300/50000 progress=62.6% loss=0.087527 lr=3.360873e-05 lr_backbone=3.360873e-05 elapsed=13:44:24 avg_step=1.58s recent_step=1.17s eta=08:12:32 +2026-09-20 20:22:38 | INFO | train rank=0 | step=31310/50000 progress=62.6% loss=0.069134 lr=3.357749e-05 lr_backbone=3.357749e-05 elapsed=13:44:37 avg_step=1.58s recent_step=1.28s eta=08:12:14 +2026-09-20 20:22:57 | INFO | train rank=0 | step=31320/50000 progress=62.6% loss=0.076651 lr=3.354626e-05 lr_backbone=3.354626e-05 elapsed=13:44:57 avg_step=1.58s recent_step=1.93s eta=08:12:01 +2026-09-20 20:23:11 | INFO | train rank=0 | step=31330/50000 progress=62.7% loss=0.114694 lr=3.351503e-05 lr_backbone=3.351503e-05 elapsed=13:45:11 avg_step=1.58s recent_step=1.43s eta=08:11:44 +2026-09-20 20:23:30 | INFO | train rank=0 | step=31340/50000 progress=62.7% loss=0.111968 lr=3.348382e-05 lr_backbone=3.348382e-05 elapsed=13:45:29 avg_step=1.58s recent_step=1.86s eta=08:11:30 +2026-09-20 20:23:42 | INFO | train rank=0 | step=31350/50000 progress=62.7% loss=0.093541 lr=3.345261e-05 lr_backbone=3.345261e-05 elapsed=13:45:42 avg_step=1.58s recent_step=1.26s eta=08:11:12 +2026-09-20 20:23:57 | INFO | train rank=0 | step=31360/50000 progress=62.7% loss=0.107575 lr=3.342141e-05 lr_backbone=3.342141e-05 elapsed=13:45:57 avg_step=1.58s recent_step=1.46s eta=08:10:56 +2026-09-20 20:24:12 | INFO | train rank=0 | step=31370/50000 progress=62.7% loss=0.079869 lr=3.339021e-05 lr_backbone=3.339021e-05 elapsed=13:46:12 avg_step=1.58s recent_step=1.50s eta=08:10:39 +2026-09-20 20:24:31 | INFO | train rank=0 | step=31380/50000 progress=62.8% loss=0.090881 lr=3.335902e-05 lr_backbone=3.335902e-05 elapsed=13:46:30 avg_step=1.58s recent_step=1.89s eta=08:10:25 +2026-09-20 20:24:47 | INFO | train rank=0 | step=31390/50000 progress=62.8% loss=0.096256 lr=3.332784e-05 lr_backbone=3.332784e-05 elapsed=13:46:47 avg_step=1.58s recent_step=1.66s eta=08:10:10 +2026-09-20 20:25:02 | INFO | train rank=0 | step=31400/50000 progress=62.8% loss=0.103292 lr=3.329667e-05 lr_backbone=3.329667e-05 elapsed=13:47:02 avg_step=1.58s recent_step=1.50s eta=08:09:54 +2026-09-20 20:25:17 | INFO | train rank=0 | step=31410/50000 progress=62.8% loss=0.069120 lr=3.326550e-05 lr_backbone=3.326550e-05 elapsed=13:47:17 avg_step=1.58s recent_step=1.49s eta=08:09:37 +2026-09-20 20:25:36 | INFO | train rank=0 | step=31420/50000 progress=62.8% loss=0.120549 lr=3.323434e-05 lr_backbone=3.323434e-05 elapsed=13:47:36 avg_step=1.58s recent_step=1.86s eta=08:09:23 +2026-09-20 20:25:50 | INFO | train rank=0 | step=31430/50000 progress=62.9% loss=0.076707 lr=3.320319e-05 lr_backbone=3.320319e-05 elapsed=13:47:49 avg_step=1.58s recent_step=1.36s eta=08:09:06 +2026-09-20 20:26:05 | INFO | train rank=0 | step=31440/50000 progress=62.9% loss=0.094315 lr=3.317205e-05 lr_backbone=3.317205e-05 elapsed=13:48:04 avg_step=1.58s recent_step=1.51s eta=08:08:50 +2026-09-20 20:26:17 | INFO | train rank=0 | step=31450/50000 progress=62.9% loss=0.091191 lr=3.314091e-05 lr_backbone=3.314091e-05 elapsed=13:48:16 avg_step=1.58s recent_step=1.23s eta=08:08:32 +2026-09-20 20:26:37 | INFO | train rank=0 | step=31460/50000 progress=62.9% loss=0.072400 lr=3.310978e-05 lr_backbone=3.310978e-05 elapsed=13:48:36 avg_step=1.58s recent_step=1.97s eta=08:08:18 +2026-09-20 20:26:54 | INFO | train rank=0 | step=31470/50000 progress=62.9% loss=0.094769 lr=3.307866e-05 lr_backbone=3.307866e-05 elapsed=13:48:53 avg_step=1.58s recent_step=1.73s eta=08:08:04 +2026-09-20 20:27:07 | INFO | train rank=0 | step=31480/50000 progress=63.0% loss=0.111683 lr=3.304755e-05 lr_backbone=3.304755e-05 elapsed=13:49:07 avg_step=1.58s recent_step=1.33s eta=08:07:46 +2026-09-20 20:27:19 | INFO | train rank=0 | step=31490/50000 progress=63.0% loss=0.125485 lr=3.301644e-05 lr_backbone=3.301644e-05 elapsed=13:49:18 avg_step=1.58s recent_step=1.15s eta=08:07:28 +2026-09-20 20:27:38 | INFO | train rank=0 | step=31500/50000 progress=63.0% loss=0.098723 lr=3.298534e-05 lr_backbone=3.298534e-05 elapsed=13:49:37 avg_step=1.58s recent_step=1.90s eta=08:07:14 +2026-09-20 20:27:54 | INFO | train rank=0 | step=31510/50000 progress=63.0% loss=0.085691 lr=3.295425e-05 lr_backbone=3.295425e-05 elapsed=13:49:53 avg_step=1.58s recent_step=1.61s eta=08:06:58 +2026-09-20 20:28:11 | INFO | train rank=0 | step=31520/50000 progress=63.0% loss=0.070363 lr=3.292316e-05 lr_backbone=3.292316e-05 elapsed=13:50:11 avg_step=1.58s recent_step=1.76s eta=08:06:44 +2026-09-20 20:28:24 | INFO | train rank=0 | step=31530/50000 progress=63.1% loss=0.101124 lr=3.289209e-05 lr_backbone=3.289209e-05 elapsed=13:50:23 avg_step=1.58s recent_step=1.24s eta=08:06:26 +2026-09-20 20:28:41 | INFO | train rank=0 | step=31540/50000 progress=63.1% loss=0.101328 lr=3.286102e-05 lr_backbone=3.286102e-05 elapsed=13:50:40 avg_step=1.58s recent_step=1.71s eta=08:06:11 +2026-09-20 20:28:57 | INFO | train rank=0 | step=31550/50000 progress=63.1% loss=0.080903 lr=3.282996e-05 lr_backbone=3.282996e-05 elapsed=13:50:57 avg_step=1.58s recent_step=1.62s eta=08:05:55 +2026-09-20 20:29:10 | INFO | train rank=0 | step=31560/50000 progress=63.1% loss=0.074165 lr=3.279890e-05 lr_backbone=3.279890e-05 elapsed=13:51:09 avg_step=1.58s recent_step=1.29s eta=08:05:38 +2026-09-20 20:29:29 | INFO | train rank=0 | step=31570/50000 progress=63.1% loss=0.078828 lr=3.276785e-05 lr_backbone=3.276785e-05 elapsed=13:51:28 avg_step=1.58s recent_step=1.91s eta=08:05:24 +2026-09-20 20:29:50 | INFO | train rank=0 | step=31580/50000 progress=63.2% loss=0.080140 lr=3.273681e-05 lr_backbone=3.273681e-05 elapsed=13:51:50 avg_step=1.58s recent_step=2.13s eta=08:05:11 +2026-09-20 20:30:02 | INFO | train rank=0 | step=31590/50000 progress=63.2% loss=0.151560 lr=3.270578e-05 lr_backbone=3.270578e-05 elapsed=13:52:02 avg_step=1.58s recent_step=1.19s eta=08:04:53 +2026-09-20 20:30:13 | INFO | train rank=0 | step=31600/50000 progress=63.2% loss=0.075253 lr=3.267476e-05 lr_backbone=3.267476e-05 elapsed=13:52:13 avg_step=1.58s recent_step=1.10s eta=08:04:35 +2026-09-20 20:30:27 | INFO | train rank=0 | step=31610/50000 progress=63.2% loss=0.101727 lr=3.264374e-05 lr_backbone=3.264374e-05 elapsed=13:52:27 avg_step=1.58s recent_step=1.41s eta=08:04:18 +2026-09-20 20:30:48 | INFO | train rank=0 | step=31620/50000 progress=63.2% loss=0.111895 lr=3.261273e-05 lr_backbone=3.261273e-05 elapsed=13:52:48 avg_step=1.58s recent_step=2.12s eta=08:04:05 +2026-09-20 20:31:12 | INFO | train rank=0 | step=31630/50000 progress=63.3% loss=0.084844 lr=3.258173e-05 lr_backbone=3.258173e-05 elapsed=13:53:11 avg_step=1.58s recent_step=2.31s eta=08:03:54 +2026-09-20 20:31:24 | INFO | train rank=0 | step=31640/50000 progress=63.3% loss=0.136907 lr=3.255074e-05 lr_backbone=3.255074e-05 elapsed=13:53:23 avg_step=1.58s recent_step=1.22s eta=08:03:36 +2026-09-20 20:31:38 | INFO | train rank=0 | step=31650/50000 progress=63.3% loss=0.072489 lr=3.251975e-05 lr_backbone=3.251975e-05 elapsed=13:53:37 avg_step=1.58s recent_step=1.40s eta=08:03:19 +2026-09-20 20:31:49 | INFO | train rank=0 | step=31660/50000 progress=63.3% loss=0.104749 lr=3.248877e-05 lr_backbone=3.248877e-05 elapsed=13:53:49 avg_step=1.58s recent_step=1.16s eta=08:03:01 +2026-09-20 20:32:05 | INFO | train rank=0 | step=31670/50000 progress=63.3% loss=0.091949 lr=3.245780e-05 lr_backbone=3.245780e-05 elapsed=13:54:04 avg_step=1.58s recent_step=1.54s eta=08:02:45 +2026-09-20 20:32:21 | INFO | train rank=0 | step=31680/50000 progress=63.4% loss=0.085040 lr=3.242684e-05 lr_backbone=3.242684e-05 elapsed=13:54:21 avg_step=1.58s recent_step=1.65s eta=08:02:29 +2026-09-20 20:32:38 | INFO | train rank=0 | step=31690/50000 progress=63.4% loss=0.093164 lr=3.239588e-05 lr_backbone=3.239588e-05 elapsed=13:54:38 avg_step=1.58s recent_step=1.71s eta=08:02:14 +2026-09-20 20:32:55 | INFO | train rank=0 | step=31700/50000 progress=63.4% loss=0.079586 lr=3.236493e-05 lr_backbone=3.236493e-05 elapsed=13:54:54 avg_step=1.58s recent_step=1.63s eta=08:01:59 +2026-09-20 20:33:06 | INFO | train rank=0 | step=31710/50000 progress=63.4% loss=0.132360 lr=3.233399e-05 lr_backbone=3.233399e-05 elapsed=13:55:06 avg_step=1.58s recent_step=1.15s eta=08:01:40 +2026-09-20 20:33:24 | INFO | train rank=0 | step=31720/50000 progress=63.4% loss=0.154103 lr=3.230306e-05 lr_backbone=3.230306e-05 elapsed=13:55:24 avg_step=1.58s recent_step=1.78s eta=08:01:26 +2026-09-20 20:33:43 | INFO | train rank=0 | step=31730/50000 progress=63.5% loss=0.093559 lr=3.227213e-05 lr_backbone=3.227213e-05 elapsed=13:55:42 avg_step=1.58s recent_step=1.89s eta=08:01:12 +2026-09-20 20:33:54 | INFO | train rank=0 | step=31740/50000 progress=63.5% loss=0.073789 lr=3.224122e-05 lr_backbone=3.224122e-05 elapsed=13:55:54 avg_step=1.58s recent_step=1.13s eta=08:00:53 +2026-09-20 20:34:12 | INFO | train rank=0 | step=31750/50000 progress=63.5% loss=0.084442 lr=3.221031e-05 lr_backbone=3.221031e-05 elapsed=13:56:12 avg_step=1.58s recent_step=1.78s eta=08:00:39 +2026-09-20 20:34:27 | INFO | train rank=0 | step=31760/50000 progress=63.5% loss=0.075752 lr=3.217941e-05 lr_backbone=3.217941e-05 elapsed=13:56:27 avg_step=1.58s recent_step=1.52s eta=08:00:22 +2026-09-20 20:34:42 | INFO | train rank=0 | step=31770/50000 progress=63.5% loss=0.096205 lr=3.214851e-05 lr_backbone=3.214851e-05 elapsed=13:56:41 avg_step=1.58s recent_step=1.45s eta=08:00:06 +2026-09-20 20:34:57 | INFO | train rank=0 | step=31780/50000 progress=63.6% loss=0.097944 lr=3.211763e-05 lr_backbone=3.211763e-05 elapsed=13:56:56 avg_step=1.58s recent_step=1.49s eta=07:59:50 +2026-09-20 20:35:12 | INFO | train rank=0 | step=31790/50000 progress=63.6% loss=0.111991 lr=3.208675e-05 lr_backbone=3.208675e-05 elapsed=13:57:12 avg_step=1.58s recent_step=1.59s eta=07:59:34 +2026-09-20 20:35:26 | INFO | train rank=0 | step=31800/50000 progress=63.6% loss=0.087578 lr=3.205588e-05 lr_backbone=3.205588e-05 elapsed=13:57:25 avg_step=1.58s recent_step=1.31s eta=07:59:16 +2026-09-20 20:35:43 | INFO | train rank=0 | step=31810/50000 progress=63.6% loss=0.111328 lr=3.202502e-05 lr_backbone=3.202502e-05 elapsed=13:57:42 avg_step=1.58s recent_step=1.72s eta=07:59:01 +2026-09-20 20:35:55 | INFO | train rank=0 | step=31820/50000 progress=63.6% loss=0.085734 lr=3.199416e-05 lr_backbone=3.199416e-05 elapsed=13:57:54 avg_step=1.58s recent_step=1.19s eta=07:58:43 +2026-09-20 20:36:10 | INFO | train rank=0 | step=31830/50000 progress=63.7% loss=0.081447 lr=3.196331e-05 lr_backbone=3.196331e-05 elapsed=13:58:10 avg_step=1.58s recent_step=1.58s eta=07:58:28 +2026-09-20 20:36:25 | INFO | train rank=0 | step=31840/50000 progress=63.7% loss=0.087283 lr=3.193248e-05 lr_backbone=3.193248e-05 elapsed=13:58:25 avg_step=1.58s recent_step=1.46s eta=07:58:11 +2026-09-20 20:36:45 | INFO | train rank=0 | step=31850/50000 progress=63.7% loss=0.081865 lr=3.190164e-05 lr_backbone=3.190164e-05 elapsed=13:58:45 avg_step=1.58s recent_step=2.05s eta=07:57:58 +2026-09-20 20:37:05 | INFO | train rank=0 | step=31860/50000 progress=63.7% loss=0.103692 lr=3.187082e-05 lr_backbone=3.187082e-05 elapsed=13:59:05 avg_step=1.58s recent_step=1.98s eta=07:57:44 +2026-09-20 20:37:22 | INFO | train rank=0 | step=31870/50000 progress=63.7% loss=0.092806 lr=3.184001e-05 lr_backbone=3.184001e-05 elapsed=13:59:22 avg_step=1.58s recent_step=1.71s eta=07:57:29 +2026-09-20 20:37:33 | INFO | train rank=0 | step=31880/50000 progress=63.8% loss=0.092575 lr=3.180920e-05 lr_backbone=3.180920e-05 elapsed=13:59:33 avg_step=1.58s recent_step=1.07s eta=07:57:11 +2026-09-20 20:38:32 | INFO | train rank=0 | step=31890/50000 progress=63.8% loss=0.134323 lr=3.177840e-05 lr_backbone=3.177840e-05 elapsed=14:00:31 avg_step=1.58s recent_step=5.89s eta=07:57:19 +2026-09-20 20:38:44 | INFO | train rank=0 | step=31900/50000 progress=63.8% loss=0.105259 lr=3.174761e-05 lr_backbone=3.174761e-05 elapsed=14:00:43 avg_step=1.58s recent_step=1.17s eta=07:57:01 +2026-09-20 20:38:57 | INFO | train rank=0 | step=31910/50000 progress=63.8% loss=0.083829 lr=3.171683e-05 lr_backbone=3.171683e-05 elapsed=14:00:57 avg_step=1.58s recent_step=1.35s eta=07:56:44 +2026-09-20 20:39:12 | INFO | train rank=0 | step=31920/50000 progress=63.8% loss=0.084614 lr=3.168605e-05 lr_backbone=3.168605e-05 elapsed=14:01:12 avg_step=1.58s recent_step=1.50s eta=07:56:28 +2026-09-20 20:39:28 | INFO | train rank=0 | step=31930/50000 progress=63.9% loss=0.087194 lr=3.165528e-05 lr_backbone=3.165528e-05 elapsed=14:01:27 avg_step=1.58s recent_step=1.58s eta=07:56:12 +2026-09-20 20:39:43 | INFO | train rank=0 | step=31940/50000 progress=63.9% loss=0.092660 lr=3.162452e-05 lr_backbone=3.162452e-05 elapsed=14:01:43 avg_step=1.58s recent_step=1.53s eta=07:55:56 +2026-09-20 20:40:00 | INFO | train rank=0 | step=31950/50000 progress=63.9% loss=0.106963 lr=3.159377e-05 lr_backbone=3.159377e-05 elapsed=14:01:59 avg_step=1.58s recent_step=1.66s eta=07:55:40 +2026-09-20 20:40:16 | INFO | train rank=0 | step=31960/50000 progress=63.9% loss=0.117663 lr=3.156303e-05 lr_backbone=3.156303e-05 elapsed=14:02:16 avg_step=1.58s recent_step=1.62s eta=07:55:25 +2026-09-20 20:40:33 | INFO | train rank=0 | step=31970/50000 progress=63.9% loss=0.077801 lr=3.153229e-05 lr_backbone=3.153229e-05 elapsed=14:02:33 avg_step=1.58s recent_step=1.75s eta=07:55:10 +2026-09-20 20:40:47 | INFO | train rank=0 | step=31980/50000 progress=64.0% loss=0.104959 lr=3.150157e-05 lr_backbone=3.150157e-05 elapsed=14:02:47 avg_step=1.58s recent_step=1.39s eta=07:54:53 +2026-09-20 20:40:59 | INFO | train rank=0 | step=31990/50000 progress=64.0% loss=0.100618 lr=3.147085e-05 lr_backbone=3.147085e-05 elapsed=14:02:59 avg_step=1.58s recent_step=1.20s eta=07:54:35 +2026-09-20 20:41:14 | INFO | train rank=0 | step=32000/50000 progress=64.0% loss=0.073694 lr=3.144014e-05 lr_backbone=3.144014e-05 elapsed=14:03:14 avg_step=1.58s recent_step=1.48s eta=07:54:19 +2026-09-20 20:41:34 | INFO | train rank=0 | step=32010/50000 progress=64.0% loss=0.090543 lr=3.140943e-05 lr_backbone=3.140943e-05 elapsed=14:03:34 avg_step=1.58s recent_step=2.03s eta=07:54:05 +2026-09-20 20:41:49 | INFO | train rank=0 | step=32020/50000 progress=64.0% loss=0.085761 lr=3.137874e-05 lr_backbone=3.137874e-05 elapsed=14:03:49 avg_step=1.58s recent_step=1.45s eta=07:53:49 +2026-09-20 20:42:01 | INFO | train rank=0 | step=32030/50000 progress=64.1% loss=0.080061 lr=3.134805e-05 lr_backbone=3.134805e-05 elapsed=14:04:01 avg_step=1.58s recent_step=1.24s eta=07:53:31 +2026-09-20 20:42:16 | INFO | train rank=0 | step=32040/50000 progress=64.1% loss=0.077960 lr=3.131738e-05 lr_backbone=3.131738e-05 elapsed=14:04:16 avg_step=1.58s recent_step=1.47s eta=07:53:15 +2026-09-20 20:42:34 | INFO | train rank=0 | step=32050/50000 progress=64.1% loss=0.076574 lr=3.128671e-05 lr_backbone=3.128671e-05 elapsed=14:04:34 avg_step=1.58s recent_step=1.80s eta=07:53:00 +2026-09-20 20:42:47 | INFO | train rank=0 | step=32060/50000 progress=64.1% loss=0.085942 lr=3.125604e-05 lr_backbone=3.125604e-05 elapsed=14:04:46 avg_step=1.58s recent_step=1.24s eta=07:52:43 +2026-09-20 20:43:05 | INFO | train rank=0 | step=32070/50000 progress=64.1% loss=0.094821 lr=3.122539e-05 lr_backbone=3.122539e-05 elapsed=14:05:04 avg_step=1.58s recent_step=1.81s eta=07:52:28 +2026-09-20 20:43:19 | INFO | train rank=0 | step=32080/50000 progress=64.2% loss=0.063554 lr=3.119474e-05 lr_backbone=3.119474e-05 elapsed=14:05:19 avg_step=1.58s recent_step=1.43s eta=07:52:11 +2026-09-20 20:43:38 | INFO | train rank=0 | step=32090/50000 progress=64.2% loss=0.088537 lr=3.116411e-05 lr_backbone=3.116411e-05 elapsed=14:05:38 avg_step=1.58s recent_step=1.94s eta=07:51:58 +2026-09-20 20:43:50 | INFO | train rank=0 | step=32100/50000 progress=64.2% loss=0.100662 lr=3.113348e-05 lr_backbone=3.113348e-05 elapsed=14:05:50 avg_step=1.58s recent_step=1.20s eta=07:51:40 +2026-09-20 20:44:07 | INFO | train rank=0 | step=32110/50000 progress=64.2% loss=0.075811 lr=3.110286e-05 lr_backbone=3.110286e-05 elapsed=14:06:06 avg_step=1.58s recent_step=1.63s eta=07:51:24 +2026-09-20 20:44:22 | INFO | train rank=0 | step=32120/50000 progress=64.2% loss=0.122397 lr=3.107225e-05 lr_backbone=3.107225e-05 elapsed=14:06:21 avg_step=1.58s recent_step=1.49s eta=07:51:08 +2026-09-20 20:44:39 | INFO | train rank=0 | step=32130/50000 progress=64.3% loss=0.107629 lr=3.104164e-05 lr_backbone=3.104164e-05 elapsed=14:06:38 avg_step=1.58s recent_step=1.70s eta=07:50:53 +2026-09-20 20:44:56 | INFO | train rank=0 | step=32140/50000 progress=64.3% loss=0.099477 lr=3.101105e-05 lr_backbone=3.101105e-05 elapsed=14:06:56 avg_step=1.58s recent_step=1.78s eta=07:50:38 +2026-09-20 20:45:09 | INFO | train rank=0 | step=32150/50000 progress=64.3% loss=0.097276 lr=3.098046e-05 lr_backbone=3.098046e-05 elapsed=14:07:08 avg_step=1.58s recent_step=1.23s eta=07:50:20 +2026-09-20 20:45:27 | INFO | train rank=0 | step=32160/50000 progress=64.3% loss=0.092092 lr=3.094988e-05 lr_backbone=3.094988e-05 elapsed=14:07:26 avg_step=1.58s recent_step=1.78s eta=07:50:05 +2026-09-20 20:45:43 | INFO | train rank=0 | step=32170/50000 progress=64.3% loss=0.094004 lr=3.091931e-05 lr_backbone=3.091931e-05 elapsed=14:07:43 avg_step=1.58s recent_step=1.68s eta=07:49:50 +2026-09-20 20:45:59 | INFO | train rank=0 | step=32180/50000 progress=64.4% loss=0.093601 lr=3.088874e-05 lr_backbone=3.088874e-05 elapsed=14:07:59 avg_step=1.58s recent_step=1.58s eta=07:49:34 +2026-09-20 20:46:14 | INFO | train rank=0 | step=32190/50000 progress=64.4% loss=0.074510 lr=3.085819e-05 lr_backbone=3.085819e-05 elapsed=14:08:13 avg_step=1.58s recent_step=1.46s eta=07:49:18 +2026-09-20 20:46:33 | INFO | train rank=0 | step=32200/50000 progress=64.4% loss=0.094722 lr=3.082764e-05 lr_backbone=3.082764e-05 elapsed=14:08:32 avg_step=1.58s recent_step=1.91s eta=07:49:04 +2026-09-20 20:46:45 | INFO | train rank=0 | step=32210/50000 progress=64.4% loss=0.123678 lr=3.079711e-05 lr_backbone=3.079711e-05 elapsed=14:08:45 avg_step=1.58s recent_step=1.24s eta=07:48:46 +2026-09-20 20:47:07 | INFO | train rank=0 | step=32220/50000 progress=64.4% loss=0.091595 lr=3.076658e-05 lr_backbone=3.076658e-05 elapsed=14:09:07 avg_step=1.58s recent_step=2.18s eta=07:48:34 +2026-09-20 20:47:19 | INFO | train rank=0 | step=32230/50000 progress=64.5% loss=0.074721 lr=3.073606e-05 lr_backbone=3.073606e-05 elapsed=14:09:19 avg_step=1.58s recent_step=1.23s eta=07:48:16 +2026-09-20 20:47:31 | INFO | train rank=0 | step=32240/50000 progress=64.5% loss=0.091402 lr=3.070555e-05 lr_backbone=3.070555e-05 elapsed=14:09:31 avg_step=1.58s recent_step=1.22s eta=07:47:58 +2026-09-20 20:47:43 | INFO | train rank=0 | step=32250/50000 progress=64.5% loss=0.084561 lr=3.067504e-05 lr_backbone=3.067504e-05 elapsed=14:09:43 avg_step=1.58s recent_step=1.17s eta=07:47:40 +2026-09-20 20:47:59 | INFO | train rank=0 | step=32260/50000 progress=64.5% loss=0.082901 lr=3.064455e-05 lr_backbone=3.064455e-05 elapsed=14:09:59 avg_step=1.58s recent_step=1.63s eta=07:47:24 +2026-09-20 20:48:17 | INFO | train rank=0 | step=32270/50000 progress=64.5% loss=0.087867 lr=3.061406e-05 lr_backbone=3.061406e-05 elapsed=14:10:17 avg_step=1.58s recent_step=1.76s eta=07:47:10 +2026-09-20 20:48:34 | INFO | train rank=0 | step=32280/50000 progress=64.6% loss=0.083634 lr=3.058358e-05 lr_backbone=3.058358e-05 elapsed=14:10:33 avg_step=1.58s recent_step=1.69s eta=07:46:54 +2026-09-20 20:48:47 | INFO | train rank=0 | step=32290/50000 progress=64.6% loss=0.111708 lr=3.055311e-05 lr_backbone=3.055311e-05 elapsed=14:10:47 avg_step=1.58s recent_step=1.36s eta=07:46:37 +2026-09-20 20:49:08 | INFO | train rank=0 | step=32300/50000 progress=64.6% loss=0.090059 lr=3.052265e-05 lr_backbone=3.052265e-05 elapsed=14:11:08 avg_step=1.58s recent_step=2.10s eta=07:46:24 +2026-09-20 20:49:22 | INFO | train rank=0 | step=32310/50000 progress=64.6% loss=0.068622 lr=3.049220e-05 lr_backbone=3.049220e-05 elapsed=14:11:22 avg_step=1.58s recent_step=1.37s eta=07:46:07 +2026-09-20 20:49:35 | INFO | train rank=0 | step=32320/50000 progress=64.6% loss=0.084329 lr=3.046175e-05 lr_backbone=3.046175e-05 elapsed=14:11:34 avg_step=1.58s recent_step=1.26s eta=07:45:50 +2026-09-20 20:49:50 | INFO | train rank=0 | step=32330/50000 progress=64.7% loss=0.084515 lr=3.043132e-05 lr_backbone=3.043132e-05 elapsed=14:11:49 avg_step=1.58s recent_step=1.49s eta=07:45:34 +2026-09-20 20:50:07 | INFO | train rank=0 | step=32340/50000 progress=64.7% loss=0.069626 lr=3.040089e-05 lr_backbone=3.040089e-05 elapsed=14:12:06 avg_step=1.58s recent_step=1.72s eta=07:45:19 +2026-09-20 20:50:23 | INFO | train rank=0 | step=32350/50000 progress=64.7% loss=0.093570 lr=3.037047e-05 lr_backbone=3.037047e-05 elapsed=14:12:22 avg_step=1.58s recent_step=1.60s eta=07:45:03 +2026-09-20 20:50:38 | INFO | train rank=0 | step=32360/50000 progress=64.7% loss=0.096723 lr=3.034006e-05 lr_backbone=3.034006e-05 elapsed=14:12:38 avg_step=1.58s recent_step=1.55s eta=07:44:47 +2026-09-20 20:50:56 | INFO | train rank=0 | step=32370/50000 progress=64.7% loss=0.113347 lr=3.030966e-05 lr_backbone=3.030966e-05 elapsed=14:12:55 avg_step=1.58s recent_step=1.72s eta=07:44:32 +2026-09-20 20:51:16 | INFO | train rank=0 | step=32380/50000 progress=64.8% loss=0.085897 lr=3.027927e-05 lr_backbone=3.027927e-05 elapsed=14:13:15 avg_step=1.58s recent_step=2.02s eta=07:44:18 +2026-09-20 20:51:29 | INFO | train rank=0 | step=32390/50000 progress=64.8% loss=0.080349 lr=3.024888e-05 lr_backbone=3.024888e-05 elapsed=14:13:28 avg_step=1.58s recent_step=1.30s eta=07:44:01 +2026-09-20 20:51:49 | INFO | train rank=0 | step=32400/50000 progress=64.8% loss=0.078312 lr=3.021851e-05 lr_backbone=3.021851e-05 elapsed=14:13:48 avg_step=1.58s recent_step=1.98s eta=07:43:47 +2026-09-20 20:52:00 | INFO | train rank=0 | step=32410/50000 progress=64.8% loss=0.092191 lr=3.018814e-05 lr_backbone=3.018814e-05 elapsed=14:14:00 avg_step=1.58s recent_step=1.14s eta=07:43:29 +2026-09-20 20:52:15 | INFO | train rank=0 | step=32420/50000 progress=64.8% loss=0.110909 lr=3.015778e-05 lr_backbone=3.015778e-05 elapsed=14:14:14 avg_step=1.58s recent_step=1.47s eta=07:43:13 +2026-09-20 20:52:30 | INFO | train rank=0 | step=32430/50000 progress=64.9% loss=0.092835 lr=3.012743e-05 lr_backbone=3.012743e-05 elapsed=14:14:30 avg_step=1.58s recent_step=1.55s eta=07:42:57 +2026-09-20 20:52:49 | INFO | train rank=0 | step=32440/50000 progress=64.9% loss=0.083243 lr=3.009709e-05 lr_backbone=3.009709e-05 elapsed=14:14:48 avg_step=1.58s recent_step=1.86s eta=07:42:42 +2026-09-20 20:53:01 | INFO | train rank=0 | step=32450/50000 progress=64.9% loss=0.090615 lr=3.006676e-05 lr_backbone=3.006676e-05 elapsed=14:15:00 avg_step=1.58s recent_step=1.20s eta=07:42:25 +2026-09-20 20:53:20 | INFO | train rank=0 | step=32460/50000 progress=64.9% loss=0.085392 lr=3.003644e-05 lr_backbone=3.003644e-05 elapsed=14:15:19 avg_step=1.58s recent_step=1.89s eta=07:42:10 +2026-09-20 20:53:31 | INFO | train rank=0 | step=32470/50000 progress=64.9% loss=0.115181 lr=3.000612e-05 lr_backbone=3.000612e-05 elapsed=14:15:30 avg_step=1.58s recent_step=1.11s eta=07:41:52 +2026-09-20 20:53:48 | INFO | train rank=0 | step=32480/50000 progress=65.0% loss=0.079058 lr=2.997581e-05 lr_backbone=2.997581e-05 elapsed=14:15:48 avg_step=1.58s recent_step=1.76s eta=07:41:37 +2026-09-20 20:54:04 | INFO | train rank=0 | step=32490/50000 progress=65.0% loss=0.078251 lr=2.994552e-05 lr_backbone=2.994552e-05 elapsed=14:16:03 avg_step=1.58s recent_step=1.54s eta=07:41:21 +2026-09-20 20:54:17 | INFO | train rank=0 | step=32500/50000 progress=65.0% loss=0.076277 lr=2.991523e-05 lr_backbone=2.991523e-05 elapsed=14:16:17 avg_step=1.58s recent_step=1.35s eta=07:41:04 +2026-09-20 20:54:36 | INFO | train rank=0 | step=32510/50000 progress=65.0% loss=0.085362 lr=2.988495e-05 lr_backbone=2.988495e-05 elapsed=14:16:36 avg_step=1.58s recent_step=1.86s eta=07:40:50 +2026-09-20 20:54:48 | INFO | train rank=0 | step=32520/50000 progress=65.0% loss=0.078830 lr=2.985468e-05 lr_backbone=2.985468e-05 elapsed=14:16:47 avg_step=1.58s recent_step=1.19s eta=07:40:32 +2026-09-20 20:55:03 | INFO | train rank=0 | step=32530/50000 progress=65.1% loss=0.121739 lr=2.982442e-05 lr_backbone=2.982442e-05 elapsed=14:17:03 avg_step=1.58s recent_step=1.52s eta=07:40:16 +2026-09-20 20:55:21 | INFO | train rank=0 | step=32540/50000 progress=65.1% loss=0.106451 lr=2.979416e-05 lr_backbone=2.979416e-05 elapsed=14:17:20 avg_step=1.58s recent_step=1.77s eta=07:40:01 +2026-09-20 20:55:35 | INFO | train rank=0 | step=32550/50000 progress=65.1% loss=0.076454 lr=2.976392e-05 lr_backbone=2.976392e-05 elapsed=14:17:35 avg_step=1.58s recent_step=1.47s eta=07:39:45 +2026-09-20 20:55:55 | INFO | train rank=0 | step=32560/50000 progress=65.1% loss=0.077755 lr=2.973368e-05 lr_backbone=2.973368e-05 elapsed=14:17:54 avg_step=1.58s recent_step=1.91s eta=07:39:31 +2026-09-20 20:56:08 | INFO | train rank=0 | step=32570/50000 progress=65.1% loss=0.117314 lr=2.970346e-05 lr_backbone=2.970346e-05 elapsed=14:18:08 avg_step=1.58s recent_step=1.36s eta=07:39:14 +2026-09-20 20:56:21 | INFO | train rank=0 | step=32580/50000 progress=65.2% loss=0.076761 lr=2.967324e-05 lr_backbone=2.967324e-05 elapsed=14:18:21 avg_step=1.58s recent_step=1.32s eta=07:38:57 +2026-09-20 20:56:38 | INFO | train rank=0 | step=32590/50000 progress=65.2% loss=0.121242 lr=2.964303e-05 lr_backbone=2.964303e-05 elapsed=14:18:37 avg_step=1.58s recent_step=1.62s eta=07:38:41 +2026-09-20 20:56:54 | INFO | train rank=0 | step=32600/50000 progress=65.2% loss=0.104483 lr=2.961283e-05 lr_backbone=2.961283e-05 elapsed=14:18:53 avg_step=1.58s recent_step=1.59s eta=07:38:25 +2026-09-20 20:57:10 | INFO | train rank=0 | step=32610/50000 progress=65.2% loss=0.079095 lr=2.958264e-05 lr_backbone=2.958264e-05 elapsed=14:19:09 avg_step=1.58s recent_step=1.60s eta=07:38:09 +2026-09-20 20:57:26 | INFO | train rank=0 | step=32620/50000 progress=65.2% loss=0.075004 lr=2.955246e-05 lr_backbone=2.955246e-05 elapsed=14:19:26 avg_step=1.58s recent_step=1.64s eta=07:37:54 +2026-09-20 20:57:40 | INFO | train rank=0 | step=32630/50000 progress=65.3% loss=0.085737 lr=2.952228e-05 lr_backbone=2.952228e-05 elapsed=14:19:39 avg_step=1.58s recent_step=1.40s eta=07:37:37 +2026-09-20 20:57:57 | INFO | train rank=0 | step=32640/50000 progress=65.3% loss=0.103373 lr=2.949212e-05 lr_backbone=2.949212e-05 elapsed=14:19:57 avg_step=1.58s recent_step=1.73s eta=07:37:22 +2026-09-20 20:58:15 | INFO | train rank=0 | step=32650/50000 progress=65.3% loss=0.092763 lr=2.946196e-05 lr_backbone=2.946196e-05 elapsed=14:20:15 avg_step=1.58s recent_step=1.79s eta=07:37:08 +2026-09-20 20:58:34 | INFO | train rank=0 | step=32660/50000 progress=65.3% loss=0.092820 lr=2.943182e-05 lr_backbone=2.943182e-05 elapsed=14:20:33 avg_step=1.58s recent_step=1.85s eta=07:36:53 +2026-09-20 20:58:45 | INFO | train rank=0 | step=32670/50000 progress=65.3% loss=0.094832 lr=2.940168e-05 lr_backbone=2.940168e-05 elapsed=14:20:45 avg_step=1.58s recent_step=1.18s eta=07:36:35 +2026-09-20 20:59:00 | INFO | train rank=0 | step=32680/50000 progress=65.4% loss=0.107728 lr=2.937155e-05 lr_backbone=2.937155e-05 elapsed=14:21:00 avg_step=1.58s recent_step=1.46s eta=07:36:19 +2026-09-20 20:59:18 | INFO | train rank=0 | step=32690/50000 progress=65.4% loss=0.111698 lr=2.934143e-05 lr_backbone=2.934143e-05 elapsed=14:21:18 avg_step=1.58s recent_step=1.82s eta=07:36:04 +2026-09-20 20:59:37 | INFO | train rank=0 | step=32700/50000 progress=65.4% loss=0.123944 lr=2.931132e-05 lr_backbone=2.931132e-05 elapsed=14:21:37 avg_step=1.58s recent_step=1.90s eta=07:35:50 +2026-09-20 20:59:50 | INFO | train rank=0 | step=32710/50000 progress=65.4% loss=0.072232 lr=2.928122e-05 lr_backbone=2.928122e-05 elapsed=14:21:49 avg_step=1.58s recent_step=1.23s eta=07:35:32 +2026-09-20 21:00:01 | INFO | train rank=0 | step=32720/50000 progress=65.4% loss=0.077768 lr=2.925113e-05 lr_backbone=2.925113e-05 elapsed=14:22:00 avg_step=1.58s recent_step=1.12s eta=07:35:14 +2026-09-20 21:00:20 | INFO | train rank=0 | step=32730/50000 progress=65.5% loss=0.087951 lr=2.922105e-05 lr_backbone=2.922105e-05 elapsed=14:22:20 avg_step=1.58s recent_step=1.95s eta=07:35:00 +2026-09-20 21:00:38 | INFO | train rank=0 | step=32740/50000 progress=65.5% loss=0.085270 lr=2.919097e-05 lr_backbone=2.919097e-05 elapsed=14:22:38 avg_step=1.58s recent_step=1.77s eta=07:34:46 +2026-09-20 21:00:52 | INFO | train rank=0 | step=32750/50000 progress=65.5% loss=0.088883 lr=2.916091e-05 lr_backbone=2.916091e-05 elapsed=14:22:52 avg_step=1.58s recent_step=1.40s eta=07:34:29 +2026-09-20 21:01:06 | INFO | train rank=0 | step=32760/50000 progress=65.5% loss=0.153693 lr=2.913085e-05 lr_backbone=2.913085e-05 elapsed=14:23:06 avg_step=1.58s recent_step=1.41s eta=07:34:12 +2026-09-20 21:01:24 | INFO | train rank=0 | step=32770/50000 progress=65.5% loss=0.095083 lr=2.910080e-05 lr_backbone=2.910080e-05 elapsed=14:23:24 avg_step=1.58s recent_step=1.82s eta=07:33:58 +2026-09-20 21:01:41 | INFO | train rank=0 | step=32780/50000 progress=65.6% loss=0.091951 lr=2.907077e-05 lr_backbone=2.907077e-05 elapsed=14:23:41 avg_step=1.58s recent_step=1.68s eta=07:33:42 +2026-09-20 21:01:57 | INFO | train rank=0 | step=32790/50000 progress=65.6% loss=0.084349 lr=2.904074e-05 lr_backbone=2.904074e-05 elapsed=14:23:57 avg_step=1.58s recent_step=1.60s eta=07:33:27 +2026-09-20 21:02:13 | INFO | train rank=0 | step=32800/50000 progress=65.6% loss=0.082628 lr=2.901072e-05 lr_backbone=2.901072e-05 elapsed=14:24:12 avg_step=1.58s recent_step=1.56s eta=07:33:11 +2026-09-20 21:02:30 | INFO | train rank=0 | step=32810/50000 progress=65.6% loss=0.089433 lr=2.898071e-05 lr_backbone=2.898071e-05 elapsed=14:24:30 avg_step=1.58s recent_step=1.72s eta=07:32:56 +2026-09-20 21:02:44 | INFO | train rank=0 | step=32820/50000 progress=65.6% loss=0.079214 lr=2.895071e-05 lr_backbone=2.895071e-05 elapsed=14:24:44 avg_step=1.58s recent_step=1.39s eta=07:32:39 +2026-09-20 21:02:55 | INFO | train rank=0 | step=32830/50000 progress=65.7% loss=0.082815 lr=2.892072e-05 lr_backbone=2.892072e-05 elapsed=14:24:55 avg_step=1.58s recent_step=1.13s eta=07:32:21 +2026-09-20 21:03:14 | INFO | train rank=0 | step=32840/50000 progress=65.7% loss=0.077520 lr=2.889074e-05 lr_backbone=2.889074e-05 elapsed=14:25:13 avg_step=1.58s recent_step=1.84s eta=07:32:06 +2026-09-20 21:03:33 | INFO | train rank=0 | step=32850/50000 progress=65.7% loss=0.085772 lr=2.886076e-05 lr_backbone=2.886076e-05 elapsed=14:25:33 avg_step=1.58s recent_step=1.95s eta=07:31:52 +2026-09-20 21:03:45 | INFO | train rank=0 | step=32860/50000 progress=65.7% loss=0.084598 lr=2.883080e-05 lr_backbone=2.883080e-05 elapsed=14:25:44 avg_step=1.58s recent_step=1.17s eta=07:31:34 +2026-09-20 21:04:03 | INFO | train rank=0 | step=32870/50000 progress=65.7% loss=0.080591 lr=2.880084e-05 lr_backbone=2.880084e-05 elapsed=14:26:03 avg_step=1.58s recent_step=1.85s eta=07:31:20 +2026-09-20 21:04:16 | INFO | train rank=0 | step=32880/50000 progress=65.8% loss=0.092863 lr=2.877090e-05 lr_backbone=2.877090e-05 elapsed=14:26:15 avg_step=1.58s recent_step=1.23s eta=07:31:02 +2026-09-20 21:04:38 | INFO | train rank=0 | step=32890/50000 progress=65.8% loss=0.065187 lr=2.874096e-05 lr_backbone=2.874096e-05 elapsed=14:26:38 avg_step=1.58s recent_step=2.24s eta=07:30:50 +2026-09-20 21:04:50 | INFO | train rank=0 | step=32900/50000 progress=65.8% loss=0.081155 lr=2.871104e-05 lr_backbone=2.871104e-05 elapsed=14:26:49 avg_step=1.58s recent_step=1.18s eta=07:30:32 +2026-09-20 21:05:07 | INFO | train rank=0 | step=32910/50000 progress=65.8% loss=0.120719 lr=2.868112e-05 lr_backbone=2.868112e-05 elapsed=14:27:07 avg_step=1.58s recent_step=1.73s eta=07:30:17 +2026-09-20 21:05:19 | INFO | train rank=0 | step=32920/50000 progress=65.8% loss=0.108218 lr=2.865121e-05 lr_backbone=2.865121e-05 elapsed=14:27:18 avg_step=1.58s recent_step=1.15s eta=07:29:59 +2026-09-20 21:05:39 | INFO | train rank=0 | step=32930/50000 progress=65.9% loss=0.067975 lr=2.862131e-05 lr_backbone=2.862131e-05 elapsed=14:27:39 avg_step=1.58s recent_step=2.03s eta=07:29:45 +2026-09-20 21:05:53 | INFO | train rank=0 | step=32940/50000 progress=65.9% loss=0.072604 lr=2.859142e-05 lr_backbone=2.859142e-05 elapsed=14:27:53 avg_step=1.58s recent_step=1.46s eta=07:29:29 +2026-09-20 21:06:09 | INFO | train rank=0 | step=32950/50000 progress=65.9% loss=0.131988 lr=2.856154e-05 lr_backbone=2.856154e-05 elapsed=14:28:08 avg_step=1.58s recent_step=1.52s eta=07:29:13 +2026-09-20 21:06:22 | INFO | train rank=0 | step=32960/50000 progress=65.9% loss=0.107041 lr=2.853167e-05 lr_backbone=2.853167e-05 elapsed=14:28:21 avg_step=1.58s recent_step=1.29s eta=07:28:56 +2026-09-20 21:06:39 | INFO | train rank=0 | step=32970/50000 progress=65.9% loss=0.077902 lr=2.850181e-05 lr_backbone=2.850181e-05 elapsed=14:28:39 avg_step=1.58s recent_step=1.78s eta=07:28:41 +2026-09-20 21:06:55 | INFO | train rank=0 | step=32980/50000 progress=66.0% loss=0.082103 lr=2.847196e-05 lr_backbone=2.847196e-05 elapsed=14:28:55 avg_step=1.58s recent_step=1.60s eta=07:28:25 +2026-09-20 21:07:14 | INFO | train rank=0 | step=32990/50000 progress=66.0% loss=0.112441 lr=2.844212e-05 lr_backbone=2.844212e-05 elapsed=14:29:14 avg_step=1.58s recent_step=1.90s eta=07:28:11 +2026-09-20 21:07:26 | INFO | train rank=0 | step=33000/50000 progress=66.0% loss=0.112917 lr=2.841228e-05 lr_backbone=2.841228e-05 elapsed=14:29:25 avg_step=1.58s recent_step=1.13s eta=07:27:53 +2026-09-20 21:07:40 | INFO | train rank=0 | step=33010/50000 progress=66.0% loss=0.079249 lr=2.838246e-05 lr_backbone=2.838246e-05 elapsed=14:29:40 avg_step=1.58s recent_step=1.44s eta=07:27:36 +2026-09-20 21:08:00 | INFO | train rank=0 | step=33020/50000 progress=66.0% loss=0.076322 lr=2.835264e-05 lr_backbone=2.835264e-05 elapsed=14:30:00 avg_step=1.58s recent_step=2.01s eta=07:27:23 +2026-09-20 21:08:14 | INFO | train rank=0 | step=33030/50000 progress=66.1% loss=0.074590 lr=2.832284e-05 lr_backbone=2.832284e-05 elapsed=14:30:13 avg_step=1.58s recent_step=1.36s eta=07:27:06 +2026-09-20 21:08:27 | INFO | train rank=0 | step=33040/50000 progress=66.1% loss=0.104148 lr=2.829304e-05 lr_backbone=2.829304e-05 elapsed=14:30:27 avg_step=1.58s recent_step=1.32s eta=07:26:49 +2026-09-20 21:08:46 | INFO | train rank=0 | step=33050/50000 progress=66.1% loss=0.088496 lr=2.826326e-05 lr_backbone=2.826326e-05 elapsed=14:30:46 avg_step=1.58s recent_step=1.95s eta=07:26:35 +2026-09-20 21:09:00 | INFO | train rank=0 | step=33060/50000 progress=66.1% loss=0.114776 lr=2.823348e-05 lr_backbone=2.823348e-05 elapsed=14:31:00 avg_step=1.58s recent_step=1.40s eta=07:26:18 +2026-09-20 21:09:18 | INFO | train rank=0 | step=33070/50000 progress=66.1% loss=0.095092 lr=2.820372e-05 lr_backbone=2.820372e-05 elapsed=14:31:17 avg_step=1.58s recent_step=1.72s eta=07:26:03 +2026-09-20 21:09:33 | INFO | train rank=0 | step=33080/50000 progress=66.2% loss=0.096024 lr=2.817396e-05 lr_backbone=2.817396e-05 elapsed=14:31:33 avg_step=1.58s recent_step=1.59s eta=07:25:47 +2026-09-20 21:09:45 | INFO | train rank=0 | step=33090/50000 progress=66.2% loss=0.094796 lr=2.814421e-05 lr_backbone=2.814421e-05 elapsed=14:31:45 avg_step=1.58s recent_step=1.19s eta=07:25:29 +2026-09-20 21:10:03 | INFO | train rank=0 | step=33100/50000 progress=66.2% loss=0.092312 lr=2.811447e-05 lr_backbone=2.811447e-05 elapsed=14:32:03 avg_step=1.58s recent_step=1.81s eta=07:25:15 +2026-09-20 21:10:15 | INFO | train rank=0 | step=33110/50000 progress=66.2% loss=0.091871 lr=2.808475e-05 lr_backbone=2.808475e-05 elapsed=14:32:15 avg_step=1.58s recent_step=1.20s eta=07:24:57 +2026-09-20 21:10:28 | INFO | train rank=0 | step=33120/50000 progress=66.2% loss=0.078682 lr=2.805503e-05 lr_backbone=2.805503e-05 elapsed=14:32:27 avg_step=1.58s recent_step=1.21s eta=07:24:39 +2026-09-20 21:10:45 | INFO | train rank=0 | step=33130/50000 progress=66.3% loss=0.096895 lr=2.802532e-05 lr_backbone=2.802532e-05 elapsed=14:32:45 avg_step=1.58s recent_step=1.78s eta=07:24:24 +2026-09-20 21:10:57 | INFO | train rank=0 | step=33140/50000 progress=66.3% loss=0.088175 lr=2.799562e-05 lr_backbone=2.799562e-05 elapsed=14:32:57 avg_step=1.58s recent_step=1.18s eta=07:24:06 +2026-09-20 21:11:13 | INFO | train rank=0 | step=33150/50000 progress=66.3% loss=0.084700 lr=2.796593e-05 lr_backbone=2.796593e-05 elapsed=14:33:12 avg_step=1.58s recent_step=1.56s eta=07:23:51 +2026-09-20 21:11:26 | INFO | train rank=0 | step=33160/50000 progress=66.3% loss=0.080891 lr=2.793625e-05 lr_backbone=2.793625e-05 elapsed=14:33:26 avg_step=1.58s recent_step=1.35s eta=07:23:34 +2026-09-20 21:11:43 | INFO | train rank=0 | step=33170/50000 progress=66.3% loss=0.064282 lr=2.790658e-05 lr_backbone=2.790658e-05 elapsed=14:33:43 avg_step=1.58s recent_step=1.69s eta=07:23:18 +2026-09-20 21:11:59 | INFO | train rank=0 | step=33180/50000 progress=66.4% loss=0.121496 lr=2.787692e-05 lr_backbone=2.787692e-05 elapsed=14:33:59 avg_step=1.58s recent_step=1.63s eta=07:23:03 +2026-09-20 21:12:14 | INFO | train rank=0 | step=33190/50000 progress=66.4% loss=0.114325 lr=2.784726e-05 lr_backbone=2.784726e-05 elapsed=14:34:14 avg_step=1.58s recent_step=1.50s eta=07:22:47 +2026-09-20 21:12:34 | INFO | train rank=0 | step=33200/50000 progress=66.4% loss=0.094077 lr=2.781762e-05 lr_backbone=2.781762e-05 elapsed=14:34:33 avg_step=1.58s recent_step=1.92s eta=07:22:32 +2026-09-20 21:12:54 | INFO | train rank=0 | step=33210/50000 progress=66.4% loss=0.115940 lr=2.778799e-05 lr_backbone=2.778799e-05 elapsed=14:34:53 avg_step=1.58s recent_step=2.03s eta=07:22:19 +2026-09-20 21:13:12 | INFO | train rank=0 | step=33220/50000 progress=66.4% loss=0.090356 lr=2.775837e-05 lr_backbone=2.775837e-05 elapsed=14:35:11 avg_step=1.58s recent_step=1.77s eta=07:22:04 +2026-09-20 21:13:31 | INFO | train rank=0 | step=33230/50000 progress=66.5% loss=0.099797 lr=2.772876e-05 lr_backbone=2.772876e-05 elapsed=14:35:31 avg_step=1.58s recent_step=1.99s eta=07:21:50 +2026-09-20 21:13:46 | INFO | train rank=0 | step=33240/50000 progress=66.5% loss=0.085433 lr=2.769915e-05 lr_backbone=2.769915e-05 elapsed=14:35:46 avg_step=1.58s recent_step=1.48s eta=07:21:34 +2026-09-20 21:14:45 | INFO | train rank=0 | step=33250/50000 progress=66.5% loss=0.083093 lr=2.766956e-05 lr_backbone=2.766956e-05 elapsed=14:36:44 avg_step=1.58s recent_step=5.84s eta=07:21:40 +2026-09-20 21:14:57 | INFO | train rank=0 | step=33260/50000 progress=66.5% loss=0.109541 lr=2.763998e-05 lr_backbone=2.763998e-05 elapsed=14:36:56 avg_step=1.58s recent_step=1.22s eta=07:21:22 +2026-09-20 21:15:09 | INFO | train rank=0 | step=33270/50000 progress=66.5% loss=0.096859 lr=2.761040e-05 lr_backbone=2.761040e-05 elapsed=14:37:08 avg_step=1.58s recent_step=1.19s eta=07:21:04 +2026-09-20 21:15:24 | INFO | train rank=0 | step=33280/50000 progress=66.6% loss=0.098621 lr=2.758084e-05 lr_backbone=2.758084e-05 elapsed=14:37:23 avg_step=1.58s recent_step=1.49s eta=07:20:48 +2026-09-20 21:15:40 | INFO | train rank=0 | step=33290/50000 progress=66.6% loss=0.098970 lr=2.755129e-05 lr_backbone=2.755129e-05 elapsed=14:37:39 avg_step=1.58s recent_step=1.61s eta=07:20:32 +2026-09-20 21:15:58 | INFO | train rank=0 | step=33300/50000 progress=66.6% loss=0.111299 lr=2.752174e-05 lr_backbone=2.752174e-05 elapsed=14:37:57 avg_step=1.58s recent_step=1.79s eta=07:20:18 +2026-09-20 21:16:10 | INFO | train rank=0 | step=33310/50000 progress=66.6% loss=0.102207 lr=2.749221e-05 lr_backbone=2.749221e-05 elapsed=14:38:10 avg_step=1.58s recent_step=1.22s eta=07:20:00 +2026-09-20 21:16:23 | INFO | train rank=0 | step=33320/50000 progress=66.6% loss=0.082878 lr=2.746268e-05 lr_backbone=2.746268e-05 elapsed=14:38:23 avg_step=1.58s recent_step=1.35s eta=07:19:43 +2026-09-20 21:16:46 | INFO | train rank=0 | step=33330/50000 progress=66.7% loss=0.094444 lr=2.743317e-05 lr_backbone=2.743317e-05 elapsed=14:38:45 avg_step=1.58s recent_step=2.21s eta=07:19:30 +2026-09-20 21:16:57 | INFO | train rank=0 | step=33340/50000 progress=66.7% loss=0.083881 lr=2.740366e-05 lr_backbone=2.740366e-05 elapsed=14:38:57 avg_step=1.58s recent_step=1.17s eta=07:19:12 +2026-09-20 21:17:10 | INFO | train rank=0 | step=33350/50000 progress=66.7% loss=0.127306 lr=2.737417e-05 lr_backbone=2.737417e-05 elapsed=14:39:09 avg_step=1.58s recent_step=1.22s eta=07:18:55 +2026-09-20 21:17:29 | INFO | train rank=0 | step=33360/50000 progress=66.7% loss=0.083039 lr=2.734468e-05 lr_backbone=2.734468e-05 elapsed=14:39:29 avg_step=1.58s recent_step=1.94s eta=07:18:41 +2026-09-20 21:17:43 | INFO | train rank=0 | step=33370/50000 progress=66.7% loss=0.102059 lr=2.731521e-05 lr_backbone=2.731521e-05 elapsed=14:39:42 avg_step=1.58s recent_step=1.37s eta=07:18:24 +2026-09-20 21:17:58 | INFO | train rank=0 | step=33380/50000 progress=66.8% loss=0.079771 lr=2.728574e-05 lr_backbone=2.728574e-05 elapsed=14:39:57 avg_step=1.58s recent_step=1.49s eta=07:18:08 +2026-09-20 21:18:14 | INFO | train rank=0 | step=33390/50000 progress=66.8% loss=0.073349 lr=2.725629e-05 lr_backbone=2.725629e-05 elapsed=14:40:13 avg_step=1.58s recent_step=1.60s eta=07:17:52 +2026-09-20 21:18:26 | INFO | train rank=0 | step=33400/50000 progress=66.8% loss=0.103097 lr=2.722684e-05 lr_backbone=2.722684e-05 elapsed=14:40:26 avg_step=1.58s recent_step=1.29s eta=07:17:35 +2026-09-20 21:18:41 | INFO | train rank=0 | step=33410/50000 progress=66.8% loss=0.091625 lr=2.719741e-05 lr_backbone=2.719741e-05 elapsed=14:40:41 avg_step=1.58s recent_step=1.47s eta=07:17:18 +2026-09-20 21:18:59 | INFO | train rank=0 | step=33420/50000 progress=66.8% loss=0.079819 lr=2.716798e-05 lr_backbone=2.716798e-05 elapsed=14:40:58 avg_step=1.58s recent_step=1.76s eta=07:17:03 +2026-09-20 21:19:18 | INFO | train rank=0 | step=33430/50000 progress=66.9% loss=0.072500 lr=2.713857e-05 lr_backbone=2.713857e-05 elapsed=14:41:18 avg_step=1.58s recent_step=1.96s eta=07:16:49 +2026-09-20 21:19:32 | INFO | train rank=0 | step=33440/50000 progress=66.9% loss=0.085623 lr=2.710916e-05 lr_backbone=2.710916e-05 elapsed=14:41:32 avg_step=1.58s recent_step=1.36s eta=07:16:32 +2026-09-20 21:19:52 | INFO | train rank=0 | step=33450/50000 progress=66.9% loss=0.095175 lr=2.707977e-05 lr_backbone=2.707977e-05 elapsed=14:41:52 avg_step=1.58s recent_step=2.00s eta=07:16:19 +2026-09-20 21:20:05 | INFO | train rank=0 | step=33460/50000 progress=66.9% loss=0.095294 lr=2.705038e-05 lr_backbone=2.705038e-05 elapsed=14:42:05 avg_step=1.58s recent_step=1.33s eta=07:16:02 +2026-09-20 21:20:21 | INFO | train rank=0 | step=33470/50000 progress=66.9% loss=0.076935 lr=2.702101e-05 lr_backbone=2.702101e-05 elapsed=14:42:21 avg_step=1.58s recent_step=1.62s eta=07:15:46 +2026-09-20 21:20:38 | INFO | train rank=0 | step=33480/50000 progress=67.0% loss=0.073435 lr=2.699164e-05 lr_backbone=2.699164e-05 elapsed=14:42:38 avg_step=1.58s recent_step=1.69s eta=07:15:31 +2026-09-20 21:20:57 | INFO | train rank=0 | step=33490/50000 progress=67.0% loss=0.086266 lr=2.696229e-05 lr_backbone=2.696229e-05 elapsed=14:42:57 avg_step=1.58s recent_step=1.89s eta=07:15:16 +2026-09-20 21:21:09 | INFO | train rank=0 | step=33500/50000 progress=67.0% loss=0.106767 lr=2.693294e-05 lr_backbone=2.693294e-05 elapsed=14:43:09 avg_step=1.58s recent_step=1.17s eta=07:14:59 +2026-09-20 21:21:28 | INFO | train rank=0 | step=33510/50000 progress=67.0% loss=0.081509 lr=2.690361e-05 lr_backbone=2.690361e-05 elapsed=14:43:28 avg_step=1.58s recent_step=1.90s eta=07:14:44 +2026-09-20 21:21:40 | INFO | train rank=0 | step=33520/50000 progress=67.0% loss=0.090601 lr=2.687428e-05 lr_backbone=2.687428e-05 elapsed=14:43:40 avg_step=1.58s recent_step=1.24s eta=07:14:27 +2026-09-20 21:21:58 | INFO | train rank=0 | step=33530/50000 progress=67.1% loss=0.073907 lr=2.684497e-05 lr_backbone=2.684497e-05 elapsed=14:43:57 avg_step=1.58s recent_step=1.73s eta=07:14:12 +2026-09-20 21:22:10 | INFO | train rank=0 | step=33540/50000 progress=67.1% loss=0.134184 lr=2.681566e-05 lr_backbone=2.681566e-05 elapsed=14:44:09 avg_step=1.58s recent_step=1.21s eta=07:13:54 +2026-09-20 21:22:28 | INFO | train rank=0 | step=33550/50000 progress=67.1% loss=0.087117 lr=2.678637e-05 lr_backbone=2.678637e-05 elapsed=14:44:28 avg_step=1.58s recent_step=1.86s eta=07:13:40 +2026-09-20 21:22:45 | INFO | train rank=0 | step=33560/50000 progress=67.1% loss=0.088166 lr=2.675708e-05 lr_backbone=2.675708e-05 elapsed=14:44:45 avg_step=1.58s recent_step=1.66s eta=07:13:24 +2026-09-20 21:22:59 | INFO | train rank=0 | step=33570/50000 progress=67.1% loss=0.082002 lr=2.672781e-05 lr_backbone=2.672781e-05 elapsed=14:44:58 avg_step=1.58s recent_step=1.38s eta=07:13:07 +2026-09-20 21:23:13 | INFO | train rank=0 | step=33580/50000 progress=67.2% loss=0.069483 lr=2.669855e-05 lr_backbone=2.669855e-05 elapsed=14:45:13 avg_step=1.58s recent_step=1.43s eta=07:12:51 +2026-09-20 21:23:34 | INFO | train rank=0 | step=33590/50000 progress=67.2% loss=0.133598 lr=2.666929e-05 lr_backbone=2.666929e-05 elapsed=14:45:33 avg_step=1.58s recent_step=2.07s eta=07:12:37 +2026-09-20 21:23:50 | INFO | train rank=0 | step=33600/50000 progress=67.2% loss=0.113120 lr=2.664005e-05 lr_backbone=2.664005e-05 elapsed=14:45:50 avg_step=1.58s recent_step=1.65s eta=07:12:22 +2026-09-20 21:24:04 | INFO | train rank=0 | step=33610/50000 progress=67.2% loss=0.090642 lr=2.661082e-05 lr_backbone=2.661082e-05 elapsed=14:46:04 avg_step=1.58s recent_step=1.39s eta=07:12:05 +2026-09-20 21:24:18 | INFO | train rank=0 | step=33620/50000 progress=67.2% loss=0.102970 lr=2.658159e-05 lr_backbone=2.658159e-05 elapsed=14:46:18 avg_step=1.58s recent_step=1.37s eta=07:11:48 +2026-09-20 21:24:37 | INFO | train rank=0 | step=33630/50000 progress=67.3% loss=0.125242 lr=2.655238e-05 lr_backbone=2.655238e-05 elapsed=14:46:36 avg_step=1.58s recent_step=1.86s eta=07:11:34 +2026-09-20 21:24:49 | INFO | train rank=0 | step=33640/50000 progress=67.3% loss=0.080057 lr=2.652318e-05 lr_backbone=2.652318e-05 elapsed=14:46:49 avg_step=1.58s recent_step=1.28s eta=07:11:17 +2026-09-20 21:25:06 | INFO | train rank=0 | step=33650/50000 progress=67.3% loss=0.082849 lr=2.649399e-05 lr_backbone=2.649399e-05 elapsed=14:47:05 avg_step=1.58s recent_step=1.64s eta=07:11:01 +2026-09-20 21:25:19 | INFO | train rank=0 | step=33660/50000 progress=67.3% loss=0.080123 lr=2.646480e-05 lr_backbone=2.646480e-05 elapsed=14:47:19 avg_step=1.58s recent_step=1.33s eta=07:10:44 +2026-09-20 21:25:39 | INFO | train rank=0 | step=33670/50000 progress=67.3% loss=0.084146 lr=2.643563e-05 lr_backbone=2.643563e-05 elapsed=14:47:39 avg_step=1.58s recent_step=2.03s eta=07:10:30 +2026-09-20 21:25:51 | INFO | train rank=0 | step=33680/50000 progress=67.4% loss=0.082826 lr=2.640647e-05 lr_backbone=2.640647e-05 elapsed=14:47:51 avg_step=1.58s recent_step=1.19s eta=07:10:13 +2026-09-20 21:26:10 | INFO | train rank=0 | step=33690/50000 progress=67.4% loss=0.084145 lr=2.637732e-05 lr_backbone=2.637732e-05 elapsed=14:48:09 avg_step=1.58s recent_step=1.85s eta=07:09:58 +2026-09-20 21:26:22 | INFO | train rank=0 | step=33700/50000 progress=67.4% loss=0.095730 lr=2.634818e-05 lr_backbone=2.634818e-05 elapsed=14:48:22 avg_step=1.58s recent_step=1.24s eta=07:09:41 +2026-09-20 21:26:40 | INFO | train rank=0 | step=33710/50000 progress=67.4% loss=0.102324 lr=2.631905e-05 lr_backbone=2.631905e-05 elapsed=14:48:39 avg_step=1.58s recent_step=1.75s eta=07:09:26 +2026-09-20 21:26:55 | INFO | train rank=0 | step=33720/50000 progress=67.4% loss=0.082641 lr=2.628993e-05 lr_backbone=2.628993e-05 elapsed=14:48:55 avg_step=1.58s recent_step=1.54s eta=07:09:10 +2026-09-20 21:27:13 | INFO | train rank=0 | step=33730/50000 progress=67.5% loss=0.084906 lr=2.626082e-05 lr_backbone=2.626082e-05 elapsed=14:49:12 avg_step=1.58s recent_step=1.77s eta=07:08:55 +2026-09-20 21:27:27 | INFO | train rank=0 | step=33740/50000 progress=67.5% loss=0.079793 lr=2.623172e-05 lr_backbone=2.623172e-05 elapsed=14:49:27 avg_step=1.58s recent_step=1.43s eta=07:08:38 +2026-09-20 21:27:41 | INFO | train rank=0 | step=33750/50000 progress=67.5% loss=0.076288 lr=2.620263e-05 lr_backbone=2.620263e-05 elapsed=14:49:41 avg_step=1.58s recent_step=1.39s eta=07:08:22 +2026-09-20 21:27:57 | INFO | train rank=0 | step=33760/50000 progress=67.5% loss=0.062481 lr=2.617355e-05 lr_backbone=2.617355e-05 elapsed=14:49:56 avg_step=1.58s recent_step=1.55s eta=07:08:06 +2026-09-20 21:28:12 | INFO | train rank=0 | step=33770/50000 progress=67.5% loss=0.149089 lr=2.614448e-05 lr_backbone=2.614448e-05 elapsed=14:50:12 avg_step=1.58s recent_step=1.59s eta=07:07:50 +2026-09-20 21:28:26 | INFO | train rank=0 | step=33780/50000 progress=67.6% loss=0.089749 lr=2.611543e-05 lr_backbone=2.611543e-05 elapsed=14:50:25 avg_step=1.58s recent_step=1.32s eta=07:07:33 +2026-09-20 21:28:45 | INFO | train rank=0 | step=33790/50000 progress=67.6% loss=0.130707 lr=2.608638e-05 lr_backbone=2.608638e-05 elapsed=14:50:45 avg_step=1.58s recent_step=1.97s eta=07:07:19 +2026-09-20 21:28:58 | INFO | train rank=0 | step=33800/50000 progress=67.6% loss=0.119835 lr=2.605734e-05 lr_backbone=2.605734e-05 elapsed=14:50:58 avg_step=1.58s recent_step=1.30s eta=07:07:02 +2026-09-20 21:29:18 | INFO | train rank=0 | step=33810/50000 progress=67.6% loss=0.077464 lr=2.602832e-05 lr_backbone=2.602832e-05 elapsed=14:51:18 avg_step=1.58s recent_step=1.96s eta=07:06:48 +2026-09-20 21:29:30 | INFO | train rank=0 | step=33820/50000 progress=67.6% loss=0.080107 lr=2.599930e-05 lr_backbone=2.599930e-05 elapsed=14:51:30 avg_step=1.58s recent_step=1.22s eta=07:06:30 +2026-09-20 21:29:46 | INFO | train rank=0 | step=33830/50000 progress=67.7% loss=0.126647 lr=2.597030e-05 lr_backbone=2.597030e-05 elapsed=14:51:45 avg_step=1.58s recent_step=1.55s eta=07:06:14 +2026-09-20 21:30:03 | INFO | train rank=0 | step=33840/50000 progress=67.7% loss=0.137234 lr=2.594130e-05 lr_backbone=2.594130e-05 elapsed=14:52:03 avg_step=1.58s recent_step=1.73s eta=07:05:59 +2026-09-20 21:30:18 | INFO | train rank=0 | step=33850/50000 progress=67.7% loss=0.084323 lr=2.591232e-05 lr_backbone=2.591232e-05 elapsed=14:52:18 avg_step=1.58s recent_step=1.53s eta=07:05:43 +2026-09-20 21:30:33 | INFO | train rank=0 | step=33860/50000 progress=67.7% loss=0.104997 lr=2.588334e-05 lr_backbone=2.588334e-05 elapsed=14:52:32 avg_step=1.58s recent_step=1.47s eta=07:05:27 +2026-09-20 21:30:51 | INFO | train rank=0 | step=33870/50000 progress=67.7% loss=0.085676 lr=2.585438e-05 lr_backbone=2.585438e-05 elapsed=14:52:51 avg_step=1.58s recent_step=1.84s eta=07:05:12 +2026-09-20 21:31:04 | INFO | train rank=0 | step=33880/50000 progress=67.8% loss=0.080883 lr=2.582543e-05 lr_backbone=2.582543e-05 elapsed=14:53:03 avg_step=1.58s recent_step=1.23s eta=07:04:54 +2026-09-20 21:31:16 | INFO | train rank=0 | step=33890/50000 progress=67.8% loss=0.082573 lr=2.579648e-05 lr_backbone=2.579648e-05 elapsed=14:53:15 avg_step=1.58s recent_step=1.23s eta=07:04:37 +2026-09-20 21:31:32 | INFO | train rank=0 | step=33900/50000 progress=67.8% loss=0.088218 lr=2.576755e-05 lr_backbone=2.576755e-05 elapsed=14:53:32 avg_step=1.58s recent_step=1.62s eta=07:04:21 +2026-09-20 21:31:49 | INFO | train rank=0 | step=33910/50000 progress=67.8% loss=0.093641 lr=2.573863e-05 lr_backbone=2.573863e-05 elapsed=14:53:48 avg_step=1.58s recent_step=1.66s eta=07:04:06 +2026-09-20 21:32:07 | INFO | train rank=0 | step=33920/50000 progress=67.8% loss=0.083535 lr=2.570972e-05 lr_backbone=2.570972e-05 elapsed=14:54:06 avg_step=1.58s recent_step=1.82s eta=07:03:51 +2026-09-20 21:32:24 | INFO | train rank=0 | step=33930/50000 progress=67.9% loss=0.093313 lr=2.568082e-05 lr_backbone=2.568082e-05 elapsed=14:54:24 avg_step=1.58s recent_step=1.73s eta=07:03:36 +2026-09-20 21:32:38 | INFO | train rank=0 | step=33940/50000 progress=67.9% loss=0.064994 lr=2.565193e-05 lr_backbone=2.565193e-05 elapsed=14:54:38 avg_step=1.58s recent_step=1.38s eta=07:03:19 +2026-09-20 21:32:57 | INFO | train rank=0 | step=33950/50000 progress=67.9% loss=0.071418 lr=2.562306e-05 lr_backbone=2.562306e-05 elapsed=14:54:56 avg_step=1.58s recent_step=1.87s eta=07:03:05 +2026-09-20 21:33:09 | INFO | train rank=0 | step=33960/50000 progress=67.9% loss=0.082409 lr=2.559419e-05 lr_backbone=2.559419e-05 elapsed=14:55:08 avg_step=1.58s recent_step=1.19s eta=07:02:47 +2026-09-20 21:33:21 | INFO | train rank=0 | step=33970/50000 progress=67.9% loss=0.113697 lr=2.556533e-05 lr_backbone=2.556533e-05 elapsed=14:55:20 avg_step=1.58s recent_step=1.20s eta=07:02:30 +2026-09-20 21:33:38 | INFO | train rank=0 | step=33980/50000 progress=68.0% loss=0.089724 lr=2.553649e-05 lr_backbone=2.553649e-05 elapsed=14:55:38 avg_step=1.58s recent_step=1.75s eta=07:02:15 +2026-09-20 21:33:54 | INFO | train rank=0 | step=33990/50000 progress=68.0% loss=0.083370 lr=2.550765e-05 lr_backbone=2.550765e-05 elapsed=14:55:54 avg_step=1.58s recent_step=1.62s eta=07:01:59 +2026-09-20 21:34:12 | INFO | train rank=0 | step=34000/50000 progress=68.0% loss=0.080635 lr=2.547882e-05 lr_backbone=2.547882e-05 elapsed=14:56:11 avg_step=1.58s recent_step=1.73s eta=07:01:44 +2026-09-20 21:34:24 | INFO | train rank=0 | step=34010/50000 progress=68.0% loss=0.079191 lr=2.545001e-05 lr_backbone=2.545001e-05 elapsed=14:56:24 avg_step=1.58s recent_step=1.26s eta=07:01:26 +2026-09-20 21:34:43 | INFO | train rank=0 | step=34020/50000 progress=68.0% loss=0.084450 lr=2.542121e-05 lr_backbone=2.542121e-05 elapsed=14:56:42 avg_step=1.58s recent_step=1.86s eta=07:01:12 +2026-09-20 21:35:00 | INFO | train rank=0 | step=34030/50000 progress=68.1% loss=0.093504 lr=2.539241e-05 lr_backbone=2.539241e-05 elapsed=14:56:59 avg_step=1.58s recent_step=1.69s eta=07:00:57 +2026-09-20 21:35:17 | INFO | train rank=0 | step=34040/50000 progress=68.1% loss=0.104318 lr=2.536363e-05 lr_backbone=2.536363e-05 elapsed=14:57:16 avg_step=1.58s recent_step=1.72s eta=07:00:41 +2026-09-20 21:35:29 | INFO | train rank=0 | step=34050/50000 progress=68.1% loss=0.110112 lr=2.533486e-05 lr_backbone=2.533486e-05 elapsed=14:57:29 avg_step=1.58s recent_step=1.21s eta=07:00:24 +2026-09-20 21:35:43 | INFO | train rank=0 | step=34060/50000 progress=68.1% loss=0.082002 lr=2.530610e-05 lr_backbone=2.530610e-05 elapsed=14:57:43 avg_step=1.58s recent_step=1.40s eta=07:00:07 +2026-09-20 21:36:04 | INFO | train rank=0 | step=34070/50000 progress=68.1% loss=0.077492 lr=2.527735e-05 lr_backbone=2.527735e-05 elapsed=14:58:03 avg_step=1.58s recent_step=2.07s eta=06:59:54 +2026-09-20 21:36:15 | INFO | train rank=0 | step=34080/50000 progress=68.2% loss=0.078603 lr=2.524861e-05 lr_backbone=2.524861e-05 elapsed=14:58:15 avg_step=1.58s recent_step=1.17s eta=06:59:36 +2026-09-20 21:36:34 | INFO | train rank=0 | step=34090/50000 progress=68.2% loss=0.081748 lr=2.521989e-05 lr_backbone=2.521989e-05 elapsed=14:58:33 avg_step=1.58s recent_step=1.82s eta=06:59:21 +2026-09-20 21:36:46 | INFO | train rank=0 | step=34100/50000 progress=68.2% loss=0.074889 lr=2.519117e-05 lr_backbone=2.519117e-05 elapsed=14:58:45 avg_step=1.58s recent_step=1.23s eta=06:59:04 +2026-09-20 21:37:03 | INFO | train rank=0 | step=34110/50000 progress=68.2% loss=0.101381 lr=2.516246e-05 lr_backbone=2.516246e-05 elapsed=14:59:03 avg_step=1.58s recent_step=1.72s eta=06:58:49 +2026-09-20 21:37:18 | INFO | train rank=0 | step=34120/50000 progress=68.2% loss=0.095121 lr=2.513377e-05 lr_backbone=2.513377e-05 elapsed=14:59:18 avg_step=1.58s recent_step=1.49s eta=06:58:32 +2026-09-20 21:37:33 | INFO | train rank=0 | step=34130/50000 progress=68.3% loss=0.071766 lr=2.510508e-05 lr_backbone=2.510508e-05 elapsed=14:59:33 avg_step=1.58s recent_step=1.50s eta=06:58:16 +2026-09-20 21:37:46 | INFO | train rank=0 | step=34140/50000 progress=68.3% loss=0.081134 lr=2.507641e-05 lr_backbone=2.507641e-05 elapsed=14:59:45 avg_step=1.58s recent_step=1.28s eta=06:57:59 +2026-09-20 21:38:06 | INFO | train rank=0 | step=34150/50000 progress=68.3% loss=0.089757 lr=2.504775e-05 lr_backbone=2.504775e-05 elapsed=15:00:05 avg_step=1.58s recent_step=2.00s eta=06:57:45 +2026-09-20 21:38:22 | INFO | train rank=0 | step=34160/50000 progress=68.3% loss=0.079267 lr=2.501910e-05 lr_backbone=2.501910e-05 elapsed=15:00:21 avg_step=1.58s recent_step=1.60s eta=06:57:29 +2026-09-20 21:38:34 | INFO | train rank=0 | step=34170/50000 progress=68.3% loss=0.101310 lr=2.499045e-05 lr_backbone=2.499045e-05 elapsed=15:00:33 avg_step=1.58s recent_step=1.18s eta=06:57:12 +2026-09-20 21:38:51 | INFO | train rank=0 | step=34180/50000 progress=68.4% loss=0.079869 lr=2.496182e-05 lr_backbone=2.496182e-05 elapsed=15:00:51 avg_step=1.58s recent_step=1.78s eta=06:56:57 +2026-09-20 21:39:11 | INFO | train rank=0 | step=34190/50000 progress=68.4% loss=0.110160 lr=2.493321e-05 lr_backbone=2.493321e-05 elapsed=15:01:10 avg_step=1.58s recent_step=1.92s eta=06:56:43 +2026-09-20 21:39:22 | INFO | train rank=0 | step=34200/50000 progress=68.4% loss=0.085457 lr=2.490460e-05 lr_backbone=2.490460e-05 elapsed=15:01:22 avg_step=1.58s recent_step=1.18s eta=06:56:25 +2026-09-20 21:39:35 | INFO | train rank=0 | step=34210/50000 progress=68.4% loss=0.078453 lr=2.487600e-05 lr_backbone=2.487600e-05 elapsed=15:01:34 avg_step=1.58s recent_step=1.23s eta=06:56:08 +2026-09-20 21:39:51 | INFO | train rank=0 | step=34220/50000 progress=68.4% loss=0.074190 lr=2.484741e-05 lr_backbone=2.484741e-05 elapsed=15:01:51 avg_step=1.58s recent_step=1.62s eta=06:55:52 +2026-09-20 21:40:12 | INFO | train rank=0 | step=34230/50000 progress=68.5% loss=0.065515 lr=2.481884e-05 lr_backbone=2.481884e-05 elapsed=15:02:11 avg_step=1.58s recent_step=2.09s eta=06:55:38 +2026-09-20 21:40:28 | INFO | train rank=0 | step=34240/50000 progress=68.5% loss=0.100939 lr=2.479028e-05 lr_backbone=2.479028e-05 elapsed=15:02:28 avg_step=1.58s recent_step=1.65s eta=06:55:23 +2026-09-20 21:40:40 | INFO | train rank=0 | step=34250/50000 progress=68.5% loss=0.087978 lr=2.476172e-05 lr_backbone=2.476172e-05 elapsed=15:02:40 avg_step=1.58s recent_step=1.19s eta=06:55:05 +2026-09-20 21:40:56 | INFO | train rank=0 | step=34260/50000 progress=68.5% loss=0.078288 lr=2.473318e-05 lr_backbone=2.473318e-05 elapsed=15:02:56 avg_step=1.58s recent_step=1.62s eta=06:54:50 +2026-09-20 21:41:14 | INFO | train rank=0 | step=34270/50000 progress=68.5% loss=0.094784 lr=2.470465e-05 lr_backbone=2.470465e-05 elapsed=15:03:14 avg_step=1.58s recent_step=1.75s eta=06:54:35 +2026-09-20 21:41:29 | INFO | train rank=0 | step=34280/50000 progress=68.6% loss=0.113402 lr=2.467613e-05 lr_backbone=2.467613e-05 elapsed=15:03:29 avg_step=1.58s recent_step=1.52s eta=06:54:19 +2026-09-20 21:41:46 | INFO | train rank=0 | step=34290/50000 progress=68.6% loss=0.071076 lr=2.464762e-05 lr_backbone=2.464762e-05 elapsed=15:03:45 avg_step=1.58s recent_step=1.64s eta=06:54:03 +2026-09-20 21:42:01 | INFO | train rank=0 | step=34300/50000 progress=68.6% loss=0.094267 lr=2.461912e-05 lr_backbone=2.461912e-05 elapsed=15:04:01 avg_step=1.58s recent_step=1.58s eta=06:53:47 +2026-09-20 21:42:14 | INFO | train rank=0 | step=34310/50000 progress=68.6% loss=0.106038 lr=2.459064e-05 lr_backbone=2.459064e-05 elapsed=15:04:13 avg_step=1.58s recent_step=1.25s eta=06:53:30 +2026-09-20 21:42:28 | INFO | train rank=0 | step=34320/50000 progress=68.6% loss=0.116463 lr=2.456216e-05 lr_backbone=2.456216e-05 elapsed=15:04:28 avg_step=1.58s recent_step=1.42s eta=06:53:13 +2026-09-20 21:42:48 | INFO | train rank=0 | step=34330/50000 progress=68.7% loss=0.085096 lr=2.453370e-05 lr_backbone=2.453370e-05 elapsed=15:04:47 avg_step=1.58s recent_step=1.98s eta=06:52:59 +2026-09-20 21:43:01 | INFO | train rank=0 | step=34340/50000 progress=68.7% loss=0.083831 lr=2.450525e-05 lr_backbone=2.450525e-05 elapsed=15:05:00 avg_step=1.58s recent_step=1.28s eta=06:52:42 +2026-09-20 21:43:19 | INFO | train rank=0 | step=34350/50000 progress=68.7% loss=0.086367 lr=2.447680e-05 lr_backbone=2.447680e-05 elapsed=15:05:19 avg_step=1.58s recent_step=1.88s eta=06:52:28 +2026-09-20 21:43:31 | INFO | train rank=0 | step=34360/50000 progress=68.7% loss=0.089922 lr=2.444837e-05 lr_backbone=2.444837e-05 elapsed=15:05:31 avg_step=1.58s recent_step=1.18s eta=06:52:10 +2026-09-20 21:43:46 | INFO | train rank=0 | step=34370/50000 progress=68.7% loss=0.069017 lr=2.441995e-05 lr_backbone=2.441995e-05 elapsed=15:05:45 avg_step=1.58s recent_step=1.45s eta=06:51:54 +2026-09-20 21:44:04 | INFO | train rank=0 | step=34380/50000 progress=68.8% loss=0.100680 lr=2.439154e-05 lr_backbone=2.439154e-05 elapsed=15:06:04 avg_step=1.58s recent_step=1.86s eta=06:51:39 +2026-09-20 21:44:21 | INFO | train rank=0 | step=34390/50000 progress=68.8% loss=0.126985 lr=2.436315e-05 lr_backbone=2.436315e-05 elapsed=15:06:21 avg_step=1.58s recent_step=1.70s eta=06:51:24 +2026-09-20 21:44:37 | INFO | train rank=0 | step=34400/50000 progress=68.8% loss=0.076712 lr=2.433476e-05 lr_backbone=2.433476e-05 elapsed=15:06:36 avg_step=1.58s recent_step=1.54s eta=06:51:08 +2026-09-20 21:44:50 | INFO | train rank=0 | step=34410/50000 progress=68.8% loss=0.127224 lr=2.430639e-05 lr_backbone=2.430639e-05 elapsed=15:06:50 avg_step=1.58s recent_step=1.34s eta=06:50:51 +2026-09-20 21:45:10 | INFO | train rank=0 | step=34420/50000 progress=68.8% loss=0.074396 lr=2.427802e-05 lr_backbone=2.427802e-05 elapsed=15:07:09 avg_step=1.58s recent_step=1.93s eta=06:50:37 +2026-09-20 21:45:21 | INFO | train rank=0 | step=34430/50000 progress=68.9% loss=0.067481 lr=2.424967e-05 lr_backbone=2.424967e-05 elapsed=15:07:21 avg_step=1.58s recent_step=1.20s eta=06:50:19 +2026-09-20 21:45:33 | INFO | train rank=0 | step=34440/50000 progress=68.9% loss=0.124611 lr=2.422133e-05 lr_backbone=2.422133e-05 elapsed=15:07:33 avg_step=1.58s recent_step=1.18s eta=06:50:02 +2026-09-20 21:45:46 | INFO | train rank=0 | step=34450/50000 progress=68.9% loss=0.118635 lr=2.419300e-05 lr_backbone=2.419300e-05 elapsed=15:07:46 avg_step=1.58s recent_step=1.28s eta=06:49:44 +2026-09-20 21:46:05 | INFO | train rank=0 | step=34460/50000 progress=68.9% loss=0.110963 lr=2.416468e-05 lr_backbone=2.416468e-05 elapsed=15:08:05 avg_step=1.58s recent_step=1.93s eta=06:49:30 +2026-09-20 21:46:18 | INFO | train rank=0 | step=34470/50000 progress=68.9% loss=0.075946 lr=2.413637e-05 lr_backbone=2.413637e-05 elapsed=15:08:18 avg_step=1.58s recent_step=1.30s eta=06:49:13 +2026-09-20 21:46:35 | INFO | train rank=0 | step=34480/50000 progress=69.0% loss=0.122847 lr=2.410808e-05 lr_backbone=2.410808e-05 elapsed=15:08:34 avg_step=1.58s recent_step=1.63s eta=06:48:57 +2026-09-20 21:46:50 | INFO | train rank=0 | step=34490/50000 progress=69.0% loss=0.094576 lr=2.407979e-05 lr_backbone=2.407979e-05 elapsed=15:08:49 avg_step=1.58s recent_step=1.48s eta=06:48:41 +2026-09-20 21:47:06 | INFO | train rank=0 | step=34500/50000 progress=69.0% loss=0.099701 lr=2.405152e-05 lr_backbone=2.405152e-05 elapsed=15:09:05 avg_step=1.58s recent_step=1.62s eta=06:48:26 +2026-09-20 21:47:25 | INFO | train rank=0 | step=34510/50000 progress=69.0% loss=0.092703 lr=2.402326e-05 lr_backbone=2.402326e-05 elapsed=15:09:24 avg_step=1.58s recent_step=1.88s eta=06:48:11 +2026-09-20 21:47:36 | INFO | train rank=0 | step=34520/50000 progress=69.0% loss=0.086351 lr=2.399501e-05 lr_backbone=2.399501e-05 elapsed=15:09:36 avg_step=1.58s recent_step=1.19s eta=06:47:54 +2026-09-20 21:47:50 | INFO | train rank=0 | step=34530/50000 progress=69.1% loss=0.078819 lr=2.396677e-05 lr_backbone=2.396677e-05 elapsed=15:09:50 avg_step=1.58s recent_step=1.39s eta=06:47:37 +2026-09-20 21:48:10 | INFO | train rank=0 | step=34540/50000 progress=69.1% loss=0.118724 lr=2.393854e-05 lr_backbone=2.393854e-05 elapsed=15:10:09 avg_step=1.58s recent_step=1.95s eta=06:47:23 +2026-09-20 21:48:22 | INFO | train rank=0 | step=34550/50000 progress=69.1% loss=0.097552 lr=2.391033e-05 lr_backbone=2.391033e-05 elapsed=15:10:21 avg_step=1.58s recent_step=1.18s eta=06:47:05 +2026-09-20 21:48:35 | INFO | train rank=0 | step=34560/50000 progress=69.1% loss=0.135267 lr=2.388212e-05 lr_backbone=2.388212e-05 elapsed=15:10:35 avg_step=1.58s recent_step=1.36s eta=06:46:48 +2026-09-20 21:48:53 | INFO | train rank=0 | step=34570/50000 progress=69.1% loss=0.084539 lr=2.385393e-05 lr_backbone=2.385393e-05 elapsed=15:10:53 avg_step=1.58s recent_step=1.78s eta=06:46:33 +2026-09-20 21:49:19 | INFO | train rank=0 | step=34580/50000 progress=69.2% loss=0.098940 lr=2.382575e-05 lr_backbone=2.382575e-05 elapsed=15:11:19 avg_step=1.58s recent_step=2.63s eta=06:46:22 +2026-09-20 21:49:42 | INFO | train rank=0 | step=34590/50000 progress=69.2% loss=0.083777 lr=2.379758e-05 lr_backbone=2.379758e-05 elapsed=15:11:42 avg_step=1.58s recent_step=2.29s eta=06:46:10 +2026-09-20 21:50:37 | INFO | train rank=0 | step=34600/50000 progress=69.2% loss=0.077508 lr=2.376942e-05 lr_backbone=2.376942e-05 elapsed=15:12:37 avg_step=1.58s recent_step=5.48s eta=06:46:11 +2026-09-20 21:50:48 | INFO | train rank=0 | step=34610/50000 progress=69.2% loss=0.088321 lr=2.374127e-05 lr_backbone=2.374127e-05 elapsed=15:12:48 avg_step=1.58s recent_step=1.12s eta=06:45:53 +2026-09-20 21:51:00 | INFO | train rank=0 | step=34620/50000 progress=69.2% loss=0.093973 lr=2.371313e-05 lr_backbone=2.371313e-05 elapsed=15:13:00 avg_step=1.58s recent_step=1.22s eta=06:45:36 +2026-09-20 21:51:13 | INFO | train rank=0 | step=34630/50000 progress=69.3% loss=0.085780 lr=2.368501e-05 lr_backbone=2.368501e-05 elapsed=15:13:12 avg_step=1.58s recent_step=1.24s eta=06:45:19 +2026-09-20 21:51:24 | INFO | train rank=0 | step=34640/50000 progress=69.3% loss=0.124436 lr=2.365689e-05 lr_backbone=2.365689e-05 elapsed=15:13:24 avg_step=1.58s recent_step=1.16s eta=06:45:01 +2026-09-20 21:51:41 | INFO | train rank=0 | step=34650/50000 progress=69.3% loss=0.086466 lr=2.362879e-05 lr_backbone=2.362879e-05 elapsed=15:13:40 avg_step=1.58s recent_step=1.64s eta=06:44:45 +2026-09-20 21:51:57 | INFO | train rank=0 | step=34660/50000 progress=69.3% loss=0.078242 lr=2.360070e-05 lr_backbone=2.360070e-05 elapsed=15:13:57 avg_step=1.58s recent_step=1.63s eta=06:44:30 +2026-09-20 21:52:15 | INFO | train rank=0 | step=34670/50000 progress=69.3% loss=0.094840 lr=2.357262e-05 lr_backbone=2.357262e-05 elapsed=15:14:14 avg_step=1.58s recent_step=1.75s eta=06:44:15 +2026-09-20 21:52:28 | INFO | train rank=0 | step=34680/50000 progress=69.4% loss=0.127849 lr=2.354456e-05 lr_backbone=2.354456e-05 elapsed=15:14:27 avg_step=1.58s recent_step=1.30s eta=06:43:57 +2026-09-20 21:52:43 | INFO | train rank=0 | step=34690/50000 progress=69.4% loss=0.102332 lr=2.351650e-05 lr_backbone=2.351650e-05 elapsed=15:14:43 avg_step=1.58s recent_step=1.53s eta=06:43:41 +2026-09-20 21:52:59 | INFO | train rank=0 | step=34700/50000 progress=69.4% loss=0.092428 lr=2.348846e-05 lr_backbone=2.348846e-05 elapsed=15:14:59 avg_step=1.58s recent_step=1.63s eta=06:43:26 +2026-09-20 21:53:12 | INFO | train rank=0 | step=34710/50000 progress=69.4% loss=0.077990 lr=2.346043e-05 lr_backbone=2.346043e-05 elapsed=15:15:11 avg_step=1.58s recent_step=1.22s eta=06:43:08 +2026-09-20 21:53:29 | INFO | train rank=0 | step=34720/50000 progress=69.4% loss=0.089374 lr=2.343241e-05 lr_backbone=2.343241e-05 elapsed=15:15:29 avg_step=1.58s recent_step=1.77s eta=06:42:53 +2026-09-20 21:53:45 | INFO | train rank=0 | step=34730/50000 progress=69.5% loss=0.096551 lr=2.340440e-05 lr_backbone=2.340440e-05 elapsed=15:15:45 avg_step=1.58s recent_step=1.59s eta=06:42:38 +2026-09-20 21:53:58 | INFO | train rank=0 | step=34740/50000 progress=69.5% loss=0.084392 lr=2.337640e-05 lr_backbone=2.337640e-05 elapsed=15:15:57 avg_step=1.58s recent_step=1.26s eta=06:42:20 +2026-09-20 21:54:14 | INFO | train rank=0 | step=34750/50000 progress=69.5% loss=0.115360 lr=2.334841e-05 lr_backbone=2.334841e-05 elapsed=15:16:14 avg_step=1.58s recent_step=1.67s eta=06:42:05 +2026-09-20 21:54:28 | INFO | train rank=0 | step=34760/50000 progress=69.5% loss=0.093329 lr=2.332044e-05 lr_backbone=2.332044e-05 elapsed=15:16:28 avg_step=1.58s recent_step=1.38s eta=06:41:48 +2026-09-20 21:54:45 | INFO | train rank=0 | step=34770/50000 progress=69.5% loss=0.078784 lr=2.329248e-05 lr_backbone=2.329248e-05 elapsed=15:16:45 avg_step=1.58s recent_step=1.67s eta=06:41:33 +2026-09-20 21:55:00 | INFO | train rank=0 | step=34780/50000 progress=69.6% loss=0.116148 lr=2.326453e-05 lr_backbone=2.326453e-05 elapsed=15:17:00 avg_step=1.58s recent_step=1.55s eta=06:41:17 +2026-09-20 21:55:22 | INFO | train rank=0 | step=34790/50000 progress=69.6% loss=0.075995 lr=2.323659e-05 lr_backbone=2.323659e-05 elapsed=15:17:21 avg_step=1.58s recent_step=2.11s eta=06:41:03 +2026-09-20 21:55:37 | INFO | train rank=0 | step=34800/50000 progress=69.6% loss=0.087334 lr=2.320866e-05 lr_backbone=2.320866e-05 elapsed=15:17:37 avg_step=1.58s recent_step=1.59s eta=06:40:48 +2026-09-20 21:55:51 | INFO | train rank=0 | step=34810/50000 progress=69.6% loss=0.097662 lr=2.318074e-05 lr_backbone=2.318074e-05 elapsed=15:17:51 avg_step=1.58s recent_step=1.39s eta=06:40:31 +2026-09-20 21:56:07 | INFO | train rank=0 | step=34820/50000 progress=69.6% loss=0.086280 lr=2.315284e-05 lr_backbone=2.315284e-05 elapsed=15:18:07 avg_step=1.58s recent_step=1.59s eta=06:40:15 +2026-09-20 21:56:28 | INFO | train rank=0 | step=34830/50000 progress=69.7% loss=0.092579 lr=2.312495e-05 lr_backbone=2.312495e-05 elapsed=15:18:27 avg_step=1.58s recent_step=2.02s eta=06:40:01 +2026-09-20 21:56:41 | INFO | train rank=0 | step=34840/50000 progress=69.7% loss=0.093766 lr=2.309707e-05 lr_backbone=2.309707e-05 elapsed=15:18:41 avg_step=1.58s recent_step=1.39s eta=06:39:45 +2026-09-20 21:56:59 | INFO | train rank=0 | step=34850/50000 progress=69.7% loss=0.093992 lr=2.306920e-05 lr_backbone=2.306920e-05 elapsed=15:18:59 avg_step=1.58s recent_step=1.80s eta=06:39:30 +2026-09-20 21:57:12 | INFO | train rank=0 | step=34860/50000 progress=69.7% loss=0.084473 lr=2.304134e-05 lr_backbone=2.304134e-05 elapsed=15:19:12 avg_step=1.58s recent_step=1.29s eta=06:39:13 +2026-09-20 21:57:32 | INFO | train rank=0 | step=34870/50000 progress=69.7% loss=0.108514 lr=2.301350e-05 lr_backbone=2.301350e-05 elapsed=15:19:32 avg_step=1.58s recent_step=1.99s eta=06:38:59 +2026-09-20 21:57:47 | INFO | train rank=0 | step=34880/50000 progress=69.8% loss=0.117354 lr=2.298566e-05 lr_backbone=2.298566e-05 elapsed=15:19:47 avg_step=1.58s recent_step=1.49s eta=06:38:42 +2026-09-20 21:58:00 | INFO | train rank=0 | step=34890/50000 progress=69.8% loss=0.155809 lr=2.295784e-05 lr_backbone=2.295784e-05 elapsed=15:20:00 avg_step=1.58s recent_step=1.30s eta=06:38:25 +2026-09-20 21:58:13 | INFO | train rank=0 | step=34900/50000 progress=69.8% loss=0.088000 lr=2.293003e-05 lr_backbone=2.293003e-05 elapsed=15:20:12 avg_step=1.58s recent_step=1.27s eta=06:38:08 +2026-09-20 21:58:36 | INFO | train rank=0 | step=34910/50000 progress=69.8% loss=0.119965 lr=2.290224e-05 lr_backbone=2.290224e-05 elapsed=15:20:36 avg_step=1.58s recent_step=2.31s eta=06:37:56 +2026-09-20 21:58:49 | INFO | train rank=0 | step=34920/50000 progress=69.8% loss=0.090593 lr=2.287445e-05 lr_backbone=2.287445e-05 elapsed=15:20:48 avg_step=1.58s recent_step=1.26s eta=06:37:38 +2026-09-20 21:59:05 | INFO | train rank=0 | step=34930/50000 progress=69.9% loss=0.139907 lr=2.284668e-05 lr_backbone=2.284668e-05 elapsed=15:21:05 avg_step=1.58s recent_step=1.66s eta=06:37:23 +2026-09-20 21:59:20 | INFO | train rank=0 | step=34940/50000 progress=69.9% loss=0.101285 lr=2.281891e-05 lr_backbone=2.281891e-05 elapsed=15:21:19 avg_step=1.58s recent_step=1.45s eta=06:37:06 +2026-09-20 21:59:37 | INFO | train rank=0 | step=34950/50000 progress=69.9% loss=0.075698 lr=2.279116e-05 lr_backbone=2.279116e-05 elapsed=15:21:37 avg_step=1.58s recent_step=1.73s eta=06:36:51 +2026-09-20 21:59:49 | INFO | train rank=0 | step=34960/50000 progress=69.9% loss=0.101498 lr=2.276343e-05 lr_backbone=2.276343e-05 elapsed=15:21:48 avg_step=1.58s recent_step=1.19s eta=06:36:34 +2026-09-20 22:00:04 | INFO | train rank=0 | step=34970/50000 progress=69.9% loss=0.091294 lr=2.273570e-05 lr_backbone=2.273570e-05 elapsed=15:22:03 avg_step=1.58s recent_step=1.49s eta=06:36:18 +2026-09-20 22:00:19 | INFO | train rank=0 | step=34980/50000 progress=70.0% loss=0.090412 lr=2.270799e-05 lr_backbone=2.270799e-05 elapsed=15:22:18 avg_step=1.58s recent_step=1.51s eta=06:36:01 +2026-09-20 22:00:39 | INFO | train rank=0 | step=34990/50000 progress=70.0% loss=0.077323 lr=2.268028e-05 lr_backbone=2.268028e-05 elapsed=15:22:39 avg_step=1.58s recent_step=2.06s eta=06:35:48 +2026-09-20 22:00:53 | INFO | train rank=0 | step=35000/50000 progress=70.0% loss=0.101652 lr=2.265259e-05 lr_backbone=2.265259e-05 elapsed=15:22:53 avg_step=1.58s recent_step=1.35s eta=06:35:31 +2026-09-20 22:01:09 | INFO | train rank=0 | step=35010/50000 progress=70.0% loss=0.091808 lr=2.262491e-05 lr_backbone=2.262491e-05 elapsed=15:23:09 avg_step=1.58s recent_step=1.60s eta=06:35:15 +2026-09-20 22:01:26 | INFO | train rank=0 | step=35020/50000 progress=70.0% loss=0.072804 lr=2.259725e-05 lr_backbone=2.259725e-05 elapsed=15:23:26 avg_step=1.58s recent_step=1.73s eta=06:35:00 +2026-09-20 22:01:43 | INFO | train rank=0 | step=35030/50000 progress=70.1% loss=0.084989 lr=2.256959e-05 lr_backbone=2.256959e-05 elapsed=15:23:42 avg_step=1.58s recent_step=1.66s eta=06:34:44 +2026-09-20 22:01:58 | INFO | train rank=0 | step=35040/50000 progress=70.1% loss=0.095324 lr=2.254195e-05 lr_backbone=2.254195e-05 elapsed=15:23:57 avg_step=1.58s recent_step=1.47s eta=06:34:28 +2026-09-20 22:02:11 | INFO | train rank=0 | step=35050/50000 progress=70.1% loss=0.082741 lr=2.251432e-05 lr_backbone=2.251432e-05 elapsed=15:24:11 avg_step=1.58s recent_step=1.38s eta=06:34:11 +2026-09-20 22:02:31 | INFO | train rank=0 | step=35060/50000 progress=70.1% loss=0.104327 lr=2.248670e-05 lr_backbone=2.248670e-05 elapsed=15:24:31 avg_step=1.58s recent_step=1.97s eta=06:33:57 +2026-09-20 22:02:43 | INFO | train rank=0 | step=35070/50000 progress=70.1% loss=0.112842 lr=2.245909e-05 lr_backbone=2.245909e-05 elapsed=15:24:43 avg_step=1.58s recent_step=1.21s eta=06:33:40 +2026-09-20 22:03:00 | INFO | train rank=0 | step=35080/50000 progress=70.2% loss=0.095456 lr=2.243150e-05 lr_backbone=2.243150e-05 elapsed=15:25:00 avg_step=1.58s recent_step=1.72s eta=06:33:25 +2026-09-20 22:03:16 | INFO | train rank=0 | step=35090/50000 progress=70.2% loss=0.083153 lr=2.240392e-05 lr_backbone=2.240392e-05 elapsed=15:25:16 avg_step=1.58s recent_step=1.56s eta=06:33:09 +2026-09-20 22:03:30 | INFO | train rank=0 | step=35100/50000 progress=70.2% loss=0.087469 lr=2.237635e-05 lr_backbone=2.237635e-05 elapsed=15:25:30 avg_step=1.58s recent_step=1.40s eta=06:32:52 +2026-09-20 22:03:42 | INFO | train rank=0 | step=35110/50000 progress=70.2% loss=0.067696 lr=2.234879e-05 lr_backbone=2.234879e-05 elapsed=15:25:42 avg_step=1.58s recent_step=1.24s eta=06:32:35 +2026-09-20 22:04:03 | INFO | train rank=0 | step=35120/50000 progress=70.2% loss=0.081415 lr=2.232124e-05 lr_backbone=2.232124e-05 elapsed=15:26:02 avg_step=1.58s recent_step=2.01s eta=06:32:21 +2026-09-20 22:04:14 | INFO | train rank=0 | step=35130/50000 progress=70.3% loss=0.079090 lr=2.229371e-05 lr_backbone=2.229371e-05 elapsed=15:26:14 avg_step=1.58s recent_step=1.18s eta=06:32:03 +2026-09-20 22:04:36 | INFO | train rank=0 | step=35140/50000 progress=70.3% loss=0.084938 lr=2.226619e-05 lr_backbone=2.226619e-05 elapsed=15:26:36 avg_step=1.58s recent_step=2.16s eta=06:31:50 +2026-09-20 22:04:49 | INFO | train rank=0 | step=35150/50000 progress=70.3% loss=0.124288 lr=2.223868e-05 lr_backbone=2.223868e-05 elapsed=15:26:48 avg_step=1.58s recent_step=1.29s eta=06:31:33 +2026-09-20 22:05:05 | INFO | train rank=0 | step=35160/50000 progress=70.3% loss=0.082188 lr=2.221118e-05 lr_backbone=2.221118e-05 elapsed=15:27:04 avg_step=1.58s recent_step=1.57s eta=06:31:17 +2026-09-20 22:05:20 | INFO | train rank=0 | step=35170/50000 progress=70.3% loss=0.096617 lr=2.218369e-05 lr_backbone=2.218369e-05 elapsed=15:27:20 avg_step=1.58s recent_step=1.56s eta=06:31:01 +2026-09-20 22:05:41 | INFO | train rank=0 | step=35180/50000 progress=70.4% loss=0.104594 lr=2.215622e-05 lr_backbone=2.215622e-05 elapsed=15:27:40 avg_step=1.58s recent_step=2.06s eta=06:30:47 +2026-09-20 22:05:53 | INFO | train rank=0 | step=35190/50000 progress=70.4% loss=0.124009 lr=2.212876e-05 lr_backbone=2.212876e-05 elapsed=15:27:53 avg_step=1.58s recent_step=1.23s eta=06:30:30 +2026-09-20 22:06:05 | INFO | train rank=0 | step=35200/50000 progress=70.4% loss=0.088449 lr=2.210131e-05 lr_backbone=2.210131e-05 elapsed=15:28:05 avg_step=1.58s recent_step=1.21s eta=06:30:13 +2026-09-20 22:06:20 | INFO | train rank=0 | step=35210/50000 progress=70.4% loss=0.077581 lr=2.207387e-05 lr_backbone=2.207387e-05 elapsed=15:28:19 avg_step=1.58s recent_step=1.46s eta=06:29:56 +2026-09-20 22:06:33 | INFO | train rank=0 | step=35220/50000 progress=70.4% loss=0.075456 lr=2.204645e-05 lr_backbone=2.204645e-05 elapsed=15:28:33 avg_step=1.58s recent_step=1.35s eta=06:29:39 +2026-09-20 22:06:49 | INFO | train rank=0 | step=35230/50000 progress=70.5% loss=0.061624 lr=2.201904e-05 lr_backbone=2.201904e-05 elapsed=15:28:48 avg_step=1.58s recent_step=1.56s eta=06:29:24 +2026-09-20 22:07:05 | INFO | train rank=0 | step=35240/50000 progress=70.5% loss=0.113726 lr=2.199164e-05 lr_backbone=2.199164e-05 elapsed=15:29:05 avg_step=1.58s recent_step=1.64s eta=06:29:08 +2026-09-20 22:07:25 | INFO | train rank=0 | step=35250/50000 progress=70.5% loss=0.073956 lr=2.196425e-05 lr_backbone=2.196425e-05 elapsed=15:29:25 avg_step=1.58s recent_step=1.97s eta=06:28:54 +2026-09-20 22:07:39 | INFO | train rank=0 | step=35260/50000 progress=70.5% loss=0.081695 lr=2.193687e-05 lr_backbone=2.193687e-05 elapsed=15:29:39 avg_step=1.58s recent_step=1.42s eta=06:28:37 +2026-09-20 22:07:54 | INFO | train rank=0 | step=35270/50000 progress=70.5% loss=0.089196 lr=2.190951e-05 lr_backbone=2.190951e-05 elapsed=15:29:54 avg_step=1.58s recent_step=1.49s eta=06:28:21 +2026-09-20 22:08:11 | INFO | train rank=0 | step=35280/50000 progress=70.6% loss=0.096870 lr=2.188216e-05 lr_backbone=2.188216e-05 elapsed=15:30:10 avg_step=1.58s recent_step=1.65s eta=06:28:06 +2026-09-20 22:08:28 | INFO | train rank=0 | step=35290/50000 progress=70.6% loss=0.096075 lr=2.185482e-05 lr_backbone=2.185482e-05 elapsed=15:30:27 avg_step=1.58s recent_step=1.73s eta=06:27:50 +2026-09-20 22:08:49 | INFO | train rank=0 | step=35300/50000 progress=70.6% loss=0.068560 lr=2.182749e-05 lr_backbone=2.182749e-05 elapsed=15:30:48 avg_step=1.58s recent_step=2.09s eta=06:27:37 +2026-09-20 22:09:03 | INFO | train rank=0 | step=35310/50000 progress=70.6% loss=0.098979 lr=2.180018e-05 lr_backbone=2.180018e-05 elapsed=15:31:02 avg_step=1.58s recent_step=1.39s eta=06:27:20 +2026-09-20 22:09:17 | INFO | train rank=0 | step=35320/50000 progress=70.6% loss=0.087300 lr=2.177288e-05 lr_backbone=2.177288e-05 elapsed=15:31:17 avg_step=1.58s recent_step=1.47s eta=06:27:04 +2026-09-20 22:09:36 | INFO | train rank=0 | step=35330/50000 progress=70.7% loss=0.124307 lr=2.174559e-05 lr_backbone=2.174559e-05 elapsed=15:31:35 avg_step=1.58s recent_step=1.84s eta=06:26:49 +2026-09-20 22:09:51 | INFO | train rank=0 | step=35340/50000 progress=70.7% loss=0.080271 lr=2.171831e-05 lr_backbone=2.171831e-05 elapsed=15:31:51 avg_step=1.58s recent_step=1.54s eta=06:26:33 +2026-09-20 22:10:08 | INFO | train rank=0 | step=35350/50000 progress=70.7% loss=0.102039 lr=2.169104e-05 lr_backbone=2.169104e-05 elapsed=15:32:07 avg_step=1.58s recent_step=1.65s eta=06:26:17 +2026-09-20 22:10:23 | INFO | train rank=0 | step=35360/50000 progress=70.7% loss=0.180718 lr=2.166379e-05 lr_backbone=2.166379e-05 elapsed=15:32:23 avg_step=1.58s recent_step=1.55s eta=06:26:01 +2026-09-20 22:10:35 | INFO | train rank=0 | step=35370/50000 progress=70.7% loss=0.123853 lr=2.163655e-05 lr_backbone=2.163655e-05 elapsed=15:32:35 avg_step=1.58s recent_step=1.19s eta=06:25:44 +2026-09-20 22:10:49 | INFO | train rank=0 | step=35380/50000 progress=70.8% loss=0.077397 lr=2.160933e-05 lr_backbone=2.160933e-05 elapsed=15:32:48 avg_step=1.58s recent_step=1.38s eta=06:25:27 +2026-09-20 22:11:05 | INFO | train rank=0 | step=35390/50000 progress=70.8% loss=0.094352 lr=2.158211e-05 lr_backbone=2.158211e-05 elapsed=15:33:05 avg_step=1.58s recent_step=1.62s eta=06:25:12 +2026-09-20 22:11:24 | INFO | train rank=0 | step=35400/50000 progress=70.8% loss=0.092896 lr=2.155491e-05 lr_backbone=2.155491e-05 elapsed=15:33:24 avg_step=1.58s recent_step=1.95s eta=06:24:57 +2026-09-20 22:11:37 | INFO | train rank=0 | step=35410/50000 progress=70.8% loss=0.089965 lr=2.152772e-05 lr_backbone=2.152772e-05 elapsed=15:33:36 avg_step=1.58s recent_step=1.23s eta=06:24:40 +2026-09-20 22:11:55 | INFO | train rank=0 | step=35420/50000 progress=70.8% loss=0.085948 lr=2.150054e-05 lr_backbone=2.150054e-05 elapsed=15:33:54 avg_step=1.58s recent_step=1.78s eta=06:24:25 +2026-09-20 22:12:10 | INFO | train rank=0 | step=35430/50000 progress=70.9% loss=0.099037 lr=2.147337e-05 lr_backbone=2.147337e-05 elapsed=15:34:10 avg_step=1.58s recent_step=1.54s eta=06:24:09 +2026-09-20 22:12:28 | INFO | train rank=0 | step=35440/50000 progress=70.9% loss=0.076752 lr=2.144622e-05 lr_backbone=2.144622e-05 elapsed=15:34:27 avg_step=1.58s recent_step=1.75s eta=06:23:54 +2026-09-20 22:12:42 | INFO | train rank=0 | step=35450/50000 progress=70.9% loss=0.067182 lr=2.141908e-05 lr_backbone=2.141908e-05 elapsed=15:34:42 avg_step=1.58s recent_step=1.49s eta=06:23:38 +2026-09-20 22:13:04 | INFO | train rank=0 | step=35460/50000 progress=70.9% loss=0.086980 lr=2.139195e-05 lr_backbone=2.139195e-05 elapsed=15:35:04 avg_step=1.58s recent_step=2.18s eta=06:23:24 +2026-09-20 22:13:17 | INFO | train rank=0 | step=35470/50000 progress=70.9% loss=0.091303 lr=2.136484e-05 lr_backbone=2.136484e-05 elapsed=15:35:16 avg_step=1.58s recent_step=1.24s eta=06:23:07 +2026-09-20 22:13:28 | INFO | train rank=0 | step=35480/50000 progress=71.0% loss=0.087706 lr=2.133773e-05 lr_backbone=2.133773e-05 elapsed=15:35:28 avg_step=1.58s recent_step=1.17s eta=06:22:50 +2026-09-20 22:13:43 | INFO | train rank=0 | step=35490/50000 progress=71.0% loss=0.106921 lr=2.131064e-05 lr_backbone=2.131064e-05 elapsed=15:35:43 avg_step=1.58s recent_step=1.51s eta=06:22:34 +2026-09-20 22:14:02 | INFO | train rank=0 | step=35500/50000 progress=71.0% loss=0.078447 lr=2.128357e-05 lr_backbone=2.128357e-05 elapsed=15:36:01 avg_step=1.58s recent_step=1.85s eta=06:22:19 +2026-09-20 22:14:19 | INFO | train rank=0 | step=35510/50000 progress=71.0% loss=0.102867 lr=2.125650e-05 lr_backbone=2.125650e-05 elapsed=15:36:19 avg_step=1.58s recent_step=1.76s eta=06:22:04 +2026-09-20 22:14:31 | INFO | train rank=0 | step=35520/50000 progress=71.0% loss=0.074659 lr=2.122945e-05 lr_backbone=2.122945e-05 elapsed=15:36:31 avg_step=1.58s recent_step=1.21s eta=06:21:46 +2026-09-20 22:14:48 | INFO | train rank=0 | step=35530/50000 progress=71.1% loss=0.075366 lr=2.120241e-05 lr_backbone=2.120241e-05 elapsed=15:36:48 avg_step=1.58s recent_step=1.70s eta=06:21:31 +2026-09-20 22:15:04 | INFO | train rank=0 | step=35540/50000 progress=71.1% loss=0.068347 lr=2.117538e-05 lr_backbone=2.117538e-05 elapsed=15:37:04 avg_step=1.58s recent_step=1.60s eta=06:21:15 +2026-09-20 22:15:20 | INFO | train rank=0 | step=35550/50000 progress=71.1% loss=0.102569 lr=2.114837e-05 lr_backbone=2.114837e-05 elapsed=15:37:19 avg_step=1.58s recent_step=1.52s eta=06:20:59 +2026-09-20 22:15:38 | INFO | train rank=0 | step=35560/50000 progress=71.1% loss=0.087944 lr=2.112136e-05 lr_backbone=2.112136e-05 elapsed=15:37:37 avg_step=1.58s recent_step=1.79s eta=06:20:44 +2026-09-20 22:15:49 | INFO | train rank=0 | step=35570/50000 progress=71.1% loss=0.099414 lr=2.109438e-05 lr_backbone=2.109438e-05 elapsed=15:37:49 avg_step=1.58s recent_step=1.16s eta=06:20:27 +2026-09-20 22:16:08 | INFO | train rank=0 | step=35580/50000 progress=71.2% loss=0.083838 lr=2.106740e-05 lr_backbone=2.106740e-05 elapsed=15:38:07 avg_step=1.58s recent_step=1.83s eta=06:20:12 +2026-09-20 22:16:23 | INFO | train rank=0 | step=35590/50000 progress=71.2% loss=0.095157 lr=2.104043e-05 lr_backbone=2.104043e-05 elapsed=15:38:23 avg_step=1.58s recent_step=1.54s eta=06:19:56 +2026-09-20 22:16:39 | INFO | train rank=0 | step=35600/50000 progress=71.2% loss=0.077575 lr=2.101348e-05 lr_backbone=2.101348e-05 elapsed=15:38:38 avg_step=1.58s recent_step=1.58s eta=06:19:40 +2026-09-20 22:16:55 | INFO | train rank=0 | step=35610/50000 progress=71.2% loss=0.097238 lr=2.098654e-05 lr_backbone=2.098654e-05 elapsed=15:38:54 avg_step=1.58s recent_step=1.61s eta=06:19:24 +2026-09-20 22:17:11 | INFO | train rank=0 | step=35620/50000 progress=71.2% loss=0.100527 lr=2.095962e-05 lr_backbone=2.095962e-05 elapsed=15:39:11 avg_step=1.58s recent_step=1.66s eta=06:19:09 +2026-09-20 22:17:23 | INFO | train rank=0 | step=35630/50000 progress=71.3% loss=0.078288 lr=2.093270e-05 lr_backbone=2.093270e-05 elapsed=15:39:23 avg_step=1.58s recent_step=1.18s eta=06:18:51 +2026-09-20 22:17:41 | INFO | train rank=0 | step=35640/50000 progress=71.3% loss=0.066504 lr=2.090580e-05 lr_backbone=2.090580e-05 elapsed=15:39:40 avg_step=1.58s recent_step=1.73s eta=06:18:36 +2026-09-20 22:17:58 | INFO | train rank=0 | step=35650/50000 progress=71.3% loss=0.089683 lr=2.087892e-05 lr_backbone=2.087892e-05 elapsed=15:39:58 avg_step=1.58s recent_step=1.78s eta=06:18:21 +2026-09-20 22:18:10 | INFO | train rank=0 | step=35660/50000 progress=71.3% loss=0.072974 lr=2.085204e-05 lr_backbone=2.085204e-05 elapsed=15:40:10 avg_step=1.58s recent_step=1.20s eta=06:18:04 +2026-09-20 22:18:25 | INFO | train rank=0 | step=35670/50000 progress=71.3% loss=0.117393 lr=2.082518e-05 lr_backbone=2.082518e-05 elapsed=15:40:24 avg_step=1.58s recent_step=1.42s eta=06:17:47 +2026-09-20 22:18:42 | INFO | train rank=0 | step=35680/50000 progress=71.4% loss=0.090517 lr=2.079833e-05 lr_backbone=2.079833e-05 elapsed=15:40:42 avg_step=1.58s recent_step=1.78s eta=06:17:32 +2026-09-20 22:18:55 | INFO | train rank=0 | step=35690/50000 progress=71.4% loss=0.092887 lr=2.077149e-05 lr_backbone=2.077149e-05 elapsed=15:40:55 avg_step=1.58s recent_step=1.27s eta=06:17:15 +2026-09-20 22:19:11 | INFO | train rank=0 | step=35700/50000 progress=71.4% loss=0.095978 lr=2.074467e-05 lr_backbone=2.074467e-05 elapsed=15:41:11 avg_step=1.58s recent_step=1.60s eta=06:17:00 +2026-09-20 22:19:24 | INFO | train rank=0 | step=35710/50000 progress=71.4% loss=0.091612 lr=2.071785e-05 lr_backbone=2.071785e-05 elapsed=15:41:24 avg_step=1.58s recent_step=1.35s eta=06:16:43 +2026-09-20 22:19:41 | INFO | train rank=0 | step=35720/50000 progress=71.4% loss=0.126742 lr=2.069106e-05 lr_backbone=2.069106e-05 elapsed=15:41:40 avg_step=1.58s recent_step=1.63s eta=06:16:27 +2026-09-20 22:19:59 | INFO | train rank=0 | step=35730/50000 progress=71.5% loss=0.072275 lr=2.066427e-05 lr_backbone=2.066427e-05 elapsed=15:41:59 avg_step=1.58s recent_step=1.83s eta=06:16:12 +2026-09-20 22:20:16 | INFO | train rank=0 | step=35740/50000 progress=71.5% loss=0.073167 lr=2.063750e-05 lr_backbone=2.063750e-05 elapsed=15:42:15 avg_step=1.58s recent_step=1.67s eta=06:15:57 +2026-09-20 22:20:31 | INFO | train rank=0 | step=35750/50000 progress=71.5% loss=0.071845 lr=2.061074e-05 lr_backbone=2.061074e-05 elapsed=15:42:30 avg_step=1.58s recent_step=1.49s eta=06:15:41 +2026-09-20 22:20:47 | INFO | train rank=0 | step=35760/50000 progress=71.5% loss=0.094532 lr=2.058399e-05 lr_backbone=2.058399e-05 elapsed=15:42:46 avg_step=1.58s recent_step=1.61s eta=06:15:25 +2026-09-20 22:21:01 | INFO | train rank=0 | step=35770/50000 progress=71.5% loss=0.092292 lr=2.055726e-05 lr_backbone=2.055726e-05 elapsed=15:43:01 avg_step=1.58s recent_step=1.45s eta=06:15:09 +2026-09-20 22:21:22 | INFO | train rank=0 | step=35780/50000 progress=71.6% loss=0.077110 lr=2.053053e-05 lr_backbone=2.053053e-05 elapsed=15:43:22 avg_step=1.58s recent_step=2.07s eta=06:14:55 +2026-09-20 22:21:34 | INFO | train rank=0 | step=35790/50000 progress=71.6% loss=0.071761 lr=2.050383e-05 lr_backbone=2.050383e-05 elapsed=15:43:33 avg_step=1.58s recent_step=1.18s eta=06:14:37 +2026-09-20 22:21:49 | INFO | train rank=0 | step=35800/50000 progress=71.6% loss=0.104654 lr=2.047713e-05 lr_backbone=2.047713e-05 elapsed=15:43:48 avg_step=1.58s recent_step=1.48s eta=06:14:21 +2026-09-20 22:22:01 | INFO | train rank=0 | step=35810/50000 progress=71.6% loss=0.109306 lr=2.045045e-05 lr_backbone=2.045045e-05 elapsed=15:44:01 avg_step=1.58s recent_step=1.29s eta=06:14:04 +2026-09-20 22:22:19 | INFO | train rank=0 | step=35820/50000 progress=71.6% loss=0.069348 lr=2.042378e-05 lr_backbone=2.042378e-05 elapsed=15:44:18 avg_step=1.58s recent_step=1.74s eta=06:13:49 +2026-09-20 22:22:33 | INFO | train rank=0 | step=35830/50000 progress=71.7% loss=0.065037 lr=2.039712e-05 lr_backbone=2.039712e-05 elapsed=15:44:33 avg_step=1.58s recent_step=1.45s eta=06:13:33 +2026-09-20 22:22:50 | INFO | train rank=0 | step=35840/50000 progress=71.7% loss=0.123923 lr=2.037048e-05 lr_backbone=2.037048e-05 elapsed=15:44:50 avg_step=1.58s recent_step=1.67s eta=06:13:17 +2026-09-20 22:23:04 | INFO | train rank=0 | step=35850/50000 progress=71.7% loss=0.078515 lr=2.034385e-05 lr_backbone=2.034385e-05 elapsed=15:45:03 avg_step=1.58s recent_step=1.36s eta=06:13:00 +2026-09-20 22:23:18 | INFO | train rank=0 | step=35860/50000 progress=71.7% loss=0.088122 lr=2.031723e-05 lr_backbone=2.031723e-05 elapsed=15:45:18 avg_step=1.58s recent_step=1.45s eta=06:12:44 +2026-09-20 22:23:29 | INFO | train rank=0 | step=35870/50000 progress=71.7% loss=0.081676 lr=2.029062e-05 lr_backbone=2.029062e-05 elapsed=15:45:29 avg_step=1.58s recent_step=1.13s eta=06:12:27 +2026-09-20 22:23:42 | INFO | train rank=0 | step=35880/50000 progress=71.8% loss=0.124269 lr=2.026403e-05 lr_backbone=2.026403e-05 elapsed=15:45:42 avg_step=1.58s recent_step=1.28s eta=06:12:10 +2026-09-20 22:24:02 | INFO | train rank=0 | step=35890/50000 progress=71.8% loss=0.105476 lr=2.023745e-05 lr_backbone=2.023745e-05 elapsed=15:46:02 avg_step=1.58s recent_step=2.02s eta=06:11:55 +2026-09-20 22:24:21 | INFO | train rank=0 | step=35900/50000 progress=71.8% loss=0.067307 lr=2.021088e-05 lr_backbone=2.021088e-05 elapsed=15:46:20 avg_step=1.58s recent_step=1.83s eta=06:11:41 +2026-09-20 22:24:38 | INFO | train rank=0 | step=35910/50000 progress=71.8% loss=0.099308 lr=2.018433e-05 lr_backbone=2.018433e-05 elapsed=15:46:38 avg_step=1.58s recent_step=1.75s eta=06:11:25 +2026-09-20 22:24:50 | INFO | train rank=0 | step=35920/50000 progress=71.8% loss=0.089691 lr=2.015779e-05 lr_backbone=2.015779e-05 elapsed=15:46:49 avg_step=1.58s recent_step=1.16s eta=06:11:08 +2026-09-20 22:25:04 | INFO | train rank=0 | step=35930/50000 progress=71.9% loss=0.084209 lr=2.013126e-05 lr_backbone=2.013126e-05 elapsed=15:47:04 avg_step=1.58s recent_step=1.45s eta=06:10:52 +2026-09-20 22:25:25 | INFO | train rank=0 | step=35940/50000 progress=71.9% loss=0.110211 lr=2.010475e-05 lr_backbone=2.010475e-05 elapsed=15:47:25 avg_step=1.58s recent_step=2.10s eta=06:10:38 +2026-09-20 22:25:44 | INFO | train rank=0 | step=35950/50000 progress=71.9% loss=0.078201 lr=2.007825e-05 lr_backbone=2.007825e-05 elapsed=15:47:44 avg_step=1.58s recent_step=1.91s eta=06:10:23 +2026-09-20 22:26:39 | INFO | train rank=0 | step=35960/50000 progress=71.9% loss=0.098081 lr=2.005176e-05 lr_backbone=2.005176e-05 elapsed=15:48:39 avg_step=1.58s recent_step=5.46s eta=06:10:23 +2026-09-20 22:26:51 | INFO | train rank=0 | step=35970/50000 progress=71.9% loss=0.144205 lr=2.002529e-05 lr_backbone=2.002529e-05 elapsed=15:48:50 avg_step=1.58s recent_step=1.18s eta=06:10:05 +2026-09-20 22:27:02 | INFO | train rank=0 | step=35980/50000 progress=72.0% loss=0.081275 lr=1.999883e-05 lr_backbone=1.999883e-05 elapsed=15:49:02 avg_step=1.58s recent_step=1.11s eta=06:09:48 +2026-09-20 22:27:14 | INFO | train rank=0 | step=35990/50000 progress=72.0% loss=0.091770 lr=1.997238e-05 lr_backbone=1.997238e-05 elapsed=15:49:13 avg_step=1.58s recent_step=1.17s eta=06:09:30 +2026-09-20 22:27:26 | INFO | train rank=0 | step=36000/50000 progress=72.0% loss=0.077416 lr=1.994594e-05 lr_backbone=1.994594e-05 elapsed=15:49:25 avg_step=1.58s recent_step=1.19s eta=06:09:13 +2026-09-20 22:27:43 | INFO | train rank=0 | step=36010/50000 progress=72.0% loss=0.079778 lr=1.991952e-05 lr_backbone=1.991952e-05 elapsed=15:49:43 avg_step=1.58s recent_step=1.75s eta=06:08:58 +2026-09-20 22:27:57 | INFO | train rank=0 | step=36020/50000 progress=72.0% loss=0.097411 lr=1.989311e-05 lr_backbone=1.989311e-05 elapsed=15:49:56 avg_step=1.58s recent_step=1.37s eta=06:08:41 +2026-09-20 22:28:13 | INFO | train rank=0 | step=36030/50000 progress=72.1% loss=0.112864 lr=1.986672e-05 lr_backbone=1.986672e-05 elapsed=15:50:13 avg_step=1.58s recent_step=1.62s eta=06:08:25 +2026-09-20 22:28:31 | INFO | train rank=0 | step=36040/50000 progress=72.1% loss=0.071089 lr=1.984033e-05 lr_backbone=1.984033e-05 elapsed=15:50:31 avg_step=1.58s recent_step=1.80s eta=06:08:10 +2026-09-20 22:28:48 | INFO | train rank=0 | step=36050/50000 progress=72.1% loss=0.111996 lr=1.981396e-05 lr_backbone=1.981396e-05 elapsed=15:50:48 avg_step=1.58s recent_step=1.73s eta=06:07:55 +2026-09-20 22:29:01 | INFO | train rank=0 | step=36060/50000 progress=72.1% loss=0.082864 lr=1.978761e-05 lr_backbone=1.978761e-05 elapsed=15:51:01 avg_step=1.58s recent_step=1.31s eta=06:07:38 +2026-09-20 22:29:15 | INFO | train rank=0 | step=36070/50000 progress=72.1% loss=0.100160 lr=1.976126e-05 lr_backbone=1.976126e-05 elapsed=15:51:15 avg_step=1.58s recent_step=1.38s eta=06:07:22 +2026-09-20 22:29:30 | INFO | train rank=0 | step=36080/50000 progress=72.2% loss=0.098417 lr=1.973493e-05 lr_backbone=1.973493e-05 elapsed=15:51:29 avg_step=1.58s recent_step=1.45s eta=06:07:05 +2026-09-20 22:29:50 | INFO | train rank=0 | step=36090/50000 progress=72.2% loss=0.068528 lr=1.970862e-05 lr_backbone=1.970862e-05 elapsed=15:51:50 avg_step=1.58s recent_step=2.02s eta=06:06:51 +2026-09-20 22:30:02 | INFO | train rank=0 | step=36100/50000 progress=72.2% loss=0.072539 lr=1.968232e-05 lr_backbone=1.968232e-05 elapsed=15:52:02 avg_step=1.58s recent_step=1.22s eta=06:06:34 +2026-09-20 22:30:26 | INFO | train rank=0 | step=36110/50000 progress=72.2% loss=0.077780 lr=1.965603e-05 lr_backbone=1.965603e-05 elapsed=15:52:26 avg_step=1.58s recent_step=2.42s eta=06:06:21 +2026-09-20 22:30:41 | INFO | train rank=0 | step=36120/50000 progress=72.2% loss=0.085934 lr=1.962975e-05 lr_backbone=1.962975e-05 elapsed=15:52:40 avg_step=1.58s recent_step=1.44s eta=06:06:05 +2026-09-20 22:30:54 | INFO | train rank=0 | step=36130/50000 progress=72.3% loss=0.106933 lr=1.960349e-05 lr_backbone=1.960349e-05 elapsed=15:52:53 avg_step=1.58s recent_step=1.28s eta=06:05:48 +2026-09-20 22:31:09 | INFO | train rank=0 | step=36140/50000 progress=72.3% loss=0.076492 lr=1.957723e-05 lr_backbone=1.957723e-05 elapsed=15:53:08 avg_step=1.58s recent_step=1.52s eta=06:05:32 +2026-09-20 22:31:28 | INFO | train rank=0 | step=36150/50000 progress=72.3% loss=0.095376 lr=1.955100e-05 lr_backbone=1.955100e-05 elapsed=15:53:28 avg_step=1.58s recent_step=1.92s eta=06:05:17 +2026-09-20 22:31:46 | INFO | train rank=0 | step=36160/50000 progress=72.3% loss=0.094616 lr=1.952477e-05 lr_backbone=1.952477e-05 elapsed=15:53:46 avg_step=1.58s recent_step=1.83s eta=06:05:03 +2026-09-20 22:32:00 | INFO | train rank=0 | step=36170/50000 progress=72.3% loss=0.079496 lr=1.949856e-05 lr_backbone=1.949856e-05 elapsed=15:54:00 avg_step=1.58s recent_step=1.41s eta=06:04:46 +2026-09-20 22:32:15 | INFO | train rank=0 | step=36180/50000 progress=72.4% loss=0.071076 lr=1.947237e-05 lr_backbone=1.947237e-05 elapsed=15:54:15 avg_step=1.58s recent_step=1.50s eta=06:04:30 +2026-09-20 22:32:31 | INFO | train rank=0 | step=36190/50000 progress=72.4% loss=0.081739 lr=1.944618e-05 lr_backbone=1.944618e-05 elapsed=15:54:31 avg_step=1.58s recent_step=1.58s eta=06:04:14 +2026-09-20 22:32:49 | INFO | train rank=0 | step=36200/50000 progress=72.4% loss=0.116460 lr=1.942001e-05 lr_backbone=1.942001e-05 elapsed=15:54:48 avg_step=1.58s recent_step=1.77s eta=06:03:59 +2026-09-20 22:33:00 | INFO | train rank=0 | step=36210/50000 progress=72.4% loss=0.091625 lr=1.939386e-05 lr_backbone=1.939386e-05 elapsed=15:55:00 avg_step=1.58s recent_step=1.13s eta=06:03:41 +2026-09-20 22:33:17 | INFO | train rank=0 | step=36220/50000 progress=72.4% loss=0.064634 lr=1.936771e-05 lr_backbone=1.936771e-05 elapsed=15:55:16 avg_step=1.58s recent_step=1.66s eta=06:03:26 +2026-09-20 22:33:35 | INFO | train rank=0 | step=36230/50000 progress=72.5% loss=0.087223 lr=1.934158e-05 lr_backbone=1.934158e-05 elapsed=15:55:34 avg_step=1.58s recent_step=1.80s eta=06:03:11 +2026-09-20 22:33:47 | INFO | train rank=0 | step=36240/50000 progress=72.5% loss=0.069305 lr=1.931547e-05 lr_backbone=1.931547e-05 elapsed=15:55:46 avg_step=1.58s recent_step=1.18s eta=06:02:54 +2026-09-20 22:34:06 | INFO | train rank=0 | step=36250/50000 progress=72.5% loss=0.143057 lr=1.928936e-05 lr_backbone=1.928936e-05 elapsed=15:56:06 avg_step=1.58s recent_step=1.96s eta=06:02:39 +2026-09-20 22:34:21 | INFO | train rank=0 | step=36260/50000 progress=72.5% loss=0.089464 lr=1.926327e-05 lr_backbone=1.926327e-05 elapsed=15:56:21 avg_step=1.58s recent_step=1.47s eta=06:02:23 +2026-09-20 22:34:40 | INFO | train rank=0 | step=36270/50000 progress=72.5% loss=0.090091 lr=1.923720e-05 lr_backbone=1.923720e-05 elapsed=15:56:39 avg_step=1.58s recent_step=1.89s eta=06:02:08 +2026-09-20 22:34:53 | INFO | train rank=0 | step=36280/50000 progress=72.6% loss=0.067128 lr=1.921114e-05 lr_backbone=1.921114e-05 elapsed=15:56:53 avg_step=1.58s recent_step=1.33s eta=06:01:51 +2026-09-20 22:35:12 | INFO | train rank=0 | step=36290/50000 progress=72.6% loss=0.077013 lr=1.918509e-05 lr_backbone=1.918509e-05 elapsed=15:57:11 avg_step=1.58s recent_step=1.87s eta=06:01:37 +2026-09-20 22:35:24 | INFO | train rank=0 | step=36300/50000 progress=72.6% loss=0.085432 lr=1.915905e-05 lr_backbone=1.915905e-05 elapsed=15:57:24 avg_step=1.58s recent_step=1.24s eta=06:01:20 +2026-09-20 22:35:41 | INFO | train rank=0 | step=36310/50000 progress=72.6% loss=0.113674 lr=1.913303e-05 lr_backbone=1.913303e-05 elapsed=15:57:41 avg_step=1.58s recent_step=1.72s eta=06:01:04 +2026-09-20 22:35:55 | INFO | train rank=0 | step=36320/50000 progress=72.6% loss=0.088736 lr=1.910702e-05 lr_backbone=1.910702e-05 elapsed=15:57:55 avg_step=1.58s recent_step=1.40s eta=06:00:48 +2026-09-20 22:36:11 | INFO | train rank=0 | step=36330/50000 progress=72.7% loss=0.087818 lr=1.908102e-05 lr_backbone=1.908102e-05 elapsed=15:58:11 avg_step=1.58s recent_step=1.59s eta=06:00:32 +2026-09-20 22:36:27 | INFO | train rank=0 | step=36340/50000 progress=72.7% loss=0.087664 lr=1.905504e-05 lr_backbone=1.905504e-05 elapsed=15:58:27 avg_step=1.58s recent_step=1.56s eta=06:00:16 +2026-09-20 22:36:43 | INFO | train rank=0 | step=36350/50000 progress=72.7% loss=0.082695 lr=1.902907e-05 lr_backbone=1.902907e-05 elapsed=15:58:43 avg_step=1.58s recent_step=1.61s eta=06:00:00 +2026-09-20 22:37:00 | INFO | train rank=0 | step=36360/50000 progress=72.7% loss=0.104806 lr=1.900312e-05 lr_backbone=1.900312e-05 elapsed=15:58:59 avg_step=1.58s recent_step=1.69s eta=05:59:45 +2026-09-20 22:37:16 | INFO | train rank=0 | step=36370/50000 progress=72.7% loss=0.096153 lr=1.897718e-05 lr_backbone=1.897718e-05 elapsed=15:59:15 avg_step=1.58s recent_step=1.60s eta=05:59:29 +2026-09-20 22:37:33 | INFO | train rank=0 | step=36380/50000 progress=72.8% loss=0.086379 lr=1.895125e-05 lr_backbone=1.895125e-05 elapsed=15:59:33 avg_step=1.58s recent_step=1.71s eta=05:59:14 +2026-09-20 22:37:47 | INFO | train rank=0 | step=36390/50000 progress=72.8% loss=0.107078 lr=1.892534e-05 lr_backbone=1.892534e-05 elapsed=15:59:47 avg_step=1.58s recent_step=1.45s eta=05:58:57 +2026-09-20 22:37:59 | INFO | train rank=0 | step=36400/50000 progress=72.8% loss=0.085017 lr=1.889944e-05 lr_backbone=1.889944e-05 elapsed=15:59:59 avg_step=1.58s recent_step=1.16s eta=05:58:40 +2026-09-20 22:38:20 | INFO | train rank=0 | step=36410/50000 progress=72.8% loss=0.082575 lr=1.887355e-05 lr_backbone=1.887355e-05 elapsed=16:00:19 avg_step=1.58s recent_step=2.09s eta=05:58:26 +2026-09-20 22:38:32 | INFO | train rank=0 | step=36420/50000 progress=72.8% loss=0.091300 lr=1.884768e-05 lr_backbone=1.884768e-05 elapsed=16:00:31 avg_step=1.58s recent_step=1.18s eta=05:58:09 +2026-09-20 22:38:49 | INFO | train rank=0 | step=36430/50000 progress=72.9% loss=0.083165 lr=1.882182e-05 lr_backbone=1.882182e-05 elapsed=16:00:49 avg_step=1.58s recent_step=1.73s eta=05:57:53 +2026-09-20 22:39:07 | INFO | train rank=0 | step=36440/50000 progress=72.9% loss=0.080848 lr=1.879597e-05 lr_backbone=1.879597e-05 elapsed=16:01:07 avg_step=1.58s recent_step=1.81s eta=05:57:39 +2026-09-20 22:39:19 | INFO | train rank=0 | step=36450/50000 progress=72.9% loss=0.078360 lr=1.877014e-05 lr_backbone=1.877014e-05 elapsed=16:01:19 avg_step=1.58s recent_step=1.21s eta=05:57:21 +2026-09-20 22:39:31 | INFO | train rank=0 | step=36460/50000 progress=72.9% loss=0.086451 lr=1.874432e-05 lr_backbone=1.874432e-05 elapsed=16:01:30 avg_step=1.58s recent_step=1.17s eta=05:57:04 +2026-09-20 22:39:54 | INFO | train rank=0 | step=36470/50000 progress=72.9% loss=0.092615 lr=1.871851e-05 lr_backbone=1.871851e-05 elapsed=16:01:53 avg_step=1.58s recent_step=2.27s eta=05:56:51 +2026-09-20 22:40:08 | INFO | train rank=0 | step=36480/50000 progress=73.0% loss=0.144304 lr=1.869272e-05 lr_backbone=1.869272e-05 elapsed=16:02:08 avg_step=1.58s recent_step=1.47s eta=05:56:34 +2026-09-20 22:40:20 | INFO | train rank=0 | step=36490/50000 progress=73.0% loss=0.088405 lr=1.866695e-05 lr_backbone=1.866695e-05 elapsed=16:02:19 avg_step=1.58s recent_step=1.15s eta=05:56:17 +2026-09-20 22:40:39 | INFO | train rank=0 | step=36500/50000 progress=73.0% loss=0.083444 lr=1.864118e-05 lr_backbone=1.864118e-05 elapsed=16:02:39 avg_step=1.58s recent_step=1.96s eta=05:56:03 +2026-09-20 22:40:55 | INFO | train rank=0 | step=36510/50000 progress=73.0% loss=0.132451 lr=1.861543e-05 lr_backbone=1.861543e-05 elapsed=16:02:55 avg_step=1.58s recent_step=1.60s eta=05:55:47 +2026-09-20 22:41:10 | INFO | train rank=0 | step=36520/50000 progress=73.0% loss=0.076313 lr=1.858970e-05 lr_backbone=1.858970e-05 elapsed=16:03:09 avg_step=1.58s recent_step=1.43s eta=05:55:30 +2026-09-20 22:41:30 | INFO | train rank=0 | step=36530/50000 progress=73.1% loss=0.086661 lr=1.856397e-05 lr_backbone=1.856397e-05 elapsed=16:03:30 avg_step=1.58s recent_step=2.04s eta=05:55:16 +2026-09-20 22:41:42 | INFO | train rank=0 | step=36540/50000 progress=73.1% loss=0.085732 lr=1.853826e-05 lr_backbone=1.853826e-05 elapsed=16:03:42 avg_step=1.58s recent_step=1.19s eta=05:54:59 +2026-09-20 22:41:57 | INFO | train rank=0 | step=36550/50000 progress=73.1% loss=0.096527 lr=1.851257e-05 lr_backbone=1.851257e-05 elapsed=16:03:56 avg_step=1.58s recent_step=1.45s eta=05:54:43 +2026-09-20 22:42:15 | INFO | train rank=0 | step=36560/50000 progress=73.1% loss=0.080079 lr=1.848689e-05 lr_backbone=1.848689e-05 elapsed=16:04:14 avg_step=1.58s recent_step=1.81s eta=05:54:28 +2026-09-20 22:42:31 | INFO | train rank=0 | step=36570/50000 progress=73.1% loss=0.090317 lr=1.846122e-05 lr_backbone=1.846122e-05 elapsed=16:04:30 avg_step=1.58s recent_step=1.60s eta=05:54:12 +2026-09-20 22:42:47 | INFO | train rank=0 | step=36580/50000 progress=73.2% loss=0.102055 lr=1.843557e-05 lr_backbone=1.843557e-05 elapsed=16:04:47 avg_step=1.58s recent_step=1.66s eta=05:53:56 +2026-09-20 22:43:03 | INFO | train rank=0 | step=36590/50000 progress=73.2% loss=0.109913 lr=1.840993e-05 lr_backbone=1.840993e-05 elapsed=16:05:03 avg_step=1.58s recent_step=1.60s eta=05:53:41 +2026-09-20 22:43:16 | INFO | train rank=0 | step=36600/50000 progress=73.2% loss=0.096020 lr=1.838430e-05 lr_backbone=1.838430e-05 elapsed=16:05:16 avg_step=1.58s recent_step=1.34s eta=05:53:24 +2026-09-20 22:43:37 | INFO | train rank=0 | step=36610/50000 progress=73.2% loss=0.097107 lr=1.835869e-05 lr_backbone=1.835869e-05 elapsed=16:05:37 avg_step=1.58s recent_step=2.10s eta=05:53:10 +2026-09-20 22:43:50 | INFO | train rank=0 | step=36620/50000 progress=73.2% loss=0.083804 lr=1.833309e-05 lr_backbone=1.833309e-05 elapsed=16:05:49 avg_step=1.58s recent_step=1.22s eta=05:52:53 +2026-09-20 22:44:02 | INFO | train rank=0 | step=36630/50000 progress=73.3% loss=0.090987 lr=1.830750e-05 lr_backbone=1.830750e-05 elapsed=16:06:02 avg_step=1.58s recent_step=1.24s eta=05:52:36 +2026-09-20 22:44:15 | INFO | train rank=0 | step=36640/50000 progress=73.3% loss=0.110875 lr=1.828193e-05 lr_backbone=1.828193e-05 elapsed=16:06:14 avg_step=1.58s recent_step=1.29s eta=05:52:19 +2026-09-20 22:44:32 | INFO | train rank=0 | step=36650/50000 progress=73.3% loss=0.083593 lr=1.825638e-05 lr_backbone=1.825638e-05 elapsed=16:06:32 avg_step=1.58s recent_step=1.71s eta=05:52:03 +2026-09-20 22:44:45 | INFO | train rank=0 | step=36660/50000 progress=73.3% loss=0.090971 lr=1.823083e-05 lr_backbone=1.823083e-05 elapsed=16:06:44 avg_step=1.58s recent_step=1.30s eta=05:51:47 +2026-09-20 22:45:04 | INFO | train rank=0 | step=36670/50000 progress=73.3% loss=0.097143 lr=1.820530e-05 lr_backbone=1.820530e-05 elapsed=16:07:04 avg_step=1.58s recent_step=1.90s eta=05:51:32 +2026-09-20 22:45:16 | INFO | train rank=0 | step=36680/50000 progress=73.4% loss=0.078206 lr=1.817979e-05 lr_backbone=1.817979e-05 elapsed=16:07:16 avg_step=1.58s recent_step=1.24s eta=05:51:15 +2026-09-20 22:45:34 | INFO | train rank=0 | step=36690/50000 progress=73.4% loss=0.107366 lr=1.815429e-05 lr_backbone=1.815429e-05 elapsed=16:07:34 avg_step=1.58s recent_step=1.81s eta=05:51:00 +2026-09-20 22:45:50 | INFO | train rank=0 | step=36700/50000 progress=73.4% loss=0.088664 lr=1.812880e-05 lr_backbone=1.812880e-05 elapsed=16:07:49 avg_step=1.58s recent_step=1.50s eta=05:50:44 +2026-09-20 22:46:09 | INFO | train rank=0 | step=36710/50000 progress=73.4% loss=0.084417 lr=1.810333e-05 lr_backbone=1.810333e-05 elapsed=16:08:08 avg_step=1.58s recent_step=1.93s eta=05:50:29 +2026-09-20 22:46:22 | INFO | train rank=0 | step=36720/50000 progress=73.4% loss=0.073954 lr=1.807787e-05 lr_backbone=1.807787e-05 elapsed=16:08:21 avg_step=1.58s recent_step=1.28s eta=05:50:12 +2026-09-20 22:46:42 | INFO | train rank=0 | step=36730/50000 progress=73.5% loss=0.068403 lr=1.805242e-05 lr_backbone=1.805242e-05 elapsed=16:08:41 avg_step=1.58s recent_step=2.00s eta=05:49:58 +2026-09-20 22:46:53 | INFO | train rank=0 | step=36740/50000 progress=73.5% loss=0.094371 lr=1.802699e-05 lr_backbone=1.802699e-05 elapsed=16:08:53 avg_step=1.58s recent_step=1.13s eta=05:49:41 +2026-09-20 22:47:10 | INFO | train rank=0 | step=36750/50000 progress=73.5% loss=0.071358 lr=1.800157e-05 lr_backbone=1.800157e-05 elapsed=16:09:10 avg_step=1.58s recent_step=1.71s eta=05:49:25 +2026-09-20 22:47:26 | INFO | train rank=0 | step=36760/50000 progress=73.5% loss=0.085578 lr=1.797617e-05 lr_backbone=1.797617e-05 elapsed=16:09:25 avg_step=1.58s recent_step=1.58s eta=05:49:09 +2026-09-20 22:47:45 | INFO | train rank=0 | step=36770/50000 progress=73.5% loss=0.079280 lr=1.795078e-05 lr_backbone=1.795078e-05 elapsed=16:09:44 avg_step=1.58s recent_step=1.90s eta=05:48:55 +2026-09-20 22:47:57 | INFO | train rank=0 | step=36780/50000 progress=73.6% loss=0.060822 lr=1.792540e-05 lr_backbone=1.792540e-05 elapsed=16:09:57 avg_step=1.58s recent_step=1.24s eta=05:48:38 +2026-09-20 22:48:14 | INFO | train rank=0 | step=36790/50000 progress=73.6% loss=0.098162 lr=1.790004e-05 lr_backbone=1.790004e-05 elapsed=16:10:14 avg_step=1.58s recent_step=1.71s eta=05:48:22 +2026-09-20 22:48:26 | INFO | train rank=0 | step=36800/50000 progress=73.6% loss=0.101302 lr=1.787470e-05 lr_backbone=1.787470e-05 elapsed=16:10:26 avg_step=1.58s recent_step=1.19s eta=05:48:05 +2026-09-20 22:48:45 | INFO | train rank=0 | step=36810/50000 progress=73.6% loss=0.102007 lr=1.784936e-05 lr_backbone=1.784936e-05 elapsed=16:10:45 avg_step=1.58s recent_step=1.92s eta=05:47:50 +2026-09-20 22:49:00 | INFO | train rank=0 | step=36820/50000 progress=73.6% loss=0.115123 lr=1.782404e-05 lr_backbone=1.782404e-05 elapsed=16:10:59 avg_step=1.58s recent_step=1.45s eta=05:47:34 +2026-09-20 22:49:20 | INFO | train rank=0 | step=36830/50000 progress=73.7% loss=0.077805 lr=1.779874e-05 lr_backbone=1.779874e-05 elapsed=16:11:19 avg_step=1.58s recent_step=1.99s eta=05:47:20 +2026-09-20 22:49:32 | INFO | train rank=0 | step=36840/50000 progress=73.7% loss=0.103603 lr=1.777345e-05 lr_backbone=1.777345e-05 elapsed=16:11:32 avg_step=1.58s recent_step=1.26s eta=05:47:03 +2026-09-20 22:49:53 | INFO | train rank=0 | step=36850/50000 progress=73.7% loss=0.146392 lr=1.774817e-05 lr_backbone=1.774817e-05 elapsed=16:11:53 avg_step=1.58s recent_step=2.08s eta=05:46:49 +2026-09-20 22:50:09 | INFO | train rank=0 | step=36860/50000 progress=73.7% loss=0.074000 lr=1.772291e-05 lr_backbone=1.772291e-05 elapsed=16:12:09 avg_step=1.58s recent_step=1.59s eta=05:46:33 +2026-09-20 22:50:22 | INFO | train rank=0 | step=36870/50000 progress=73.7% loss=0.079596 lr=1.769766e-05 lr_backbone=1.769766e-05 elapsed=16:12:22 avg_step=1.58s recent_step=1.29s eta=05:46:16 +2026-09-20 22:50:37 | INFO | train rank=0 | step=36880/50000 progress=73.8% loss=0.082396 lr=1.767242e-05 lr_backbone=1.767242e-05 elapsed=16:12:36 avg_step=1.58s recent_step=1.50s eta=05:46:00 +2026-09-20 22:51:00 | INFO | train rank=0 | step=36890/50000 progress=73.8% loss=0.117713 lr=1.764720e-05 lr_backbone=1.764720e-05 elapsed=16:13:00 avg_step=1.58s recent_step=2.34s eta=05:45:47 +2026-09-20 22:51:12 | INFO | train rank=0 | step=36900/50000 progress=73.8% loss=0.129569 lr=1.762200e-05 lr_backbone=1.762200e-05 elapsed=16:13:12 avg_step=1.58s recent_step=1.19s eta=05:45:30 +2026-09-20 22:51:24 | INFO | train rank=0 | step=36910/50000 progress=73.8% loss=0.093906 lr=1.759680e-05 lr_backbone=1.759680e-05 elapsed=16:13:23 avg_step=1.58s recent_step=1.17s eta=05:45:12 +2026-09-20 22:51:39 | INFO | train rank=0 | step=36920/50000 progress=73.8% loss=0.090173 lr=1.757163e-05 lr_backbone=1.757163e-05 elapsed=16:13:38 avg_step=1.58s recent_step=1.49s eta=05:44:56 +2026-09-20 22:51:57 | INFO | train rank=0 | step=36930/50000 progress=73.9% loss=0.080106 lr=1.754646e-05 lr_backbone=1.754646e-05 elapsed=16:13:57 avg_step=1.58s recent_step=1.83s eta=05:44:41 +2026-09-20 22:52:13 | INFO | train rank=0 | step=36940/50000 progress=73.9% loss=0.145142 lr=1.752131e-05 lr_backbone=1.752131e-05 elapsed=16:14:13 avg_step=1.58s recent_step=1.61s eta=05:44:25 +2026-09-20 22:52:32 | INFO | train rank=0 | step=36950/50000 progress=73.9% loss=0.070714 lr=1.749618e-05 lr_backbone=1.749618e-05 elapsed=16:14:32 avg_step=1.58s recent_step=1.90s eta=05:44:11 +2026-09-20 22:52:44 | INFO | train rank=0 | step=36960/50000 progress=73.9% loss=0.091958 lr=1.747105e-05 lr_backbone=1.747105e-05 elapsed=16:14:44 avg_step=1.58s recent_step=1.20s eta=05:43:53 +2026-09-20 22:53:03 | INFO | train rank=0 | step=36970/50000 progress=73.9% loss=0.094744 lr=1.744595e-05 lr_backbone=1.744595e-05 elapsed=16:15:03 avg_step=1.58s recent_step=1.92s eta=05:43:39 +2026-09-20 22:53:16 | INFO | train rank=0 | step=36980/50000 progress=74.0% loss=0.083556 lr=1.742085e-05 lr_backbone=1.742085e-05 elapsed=16:15:15 avg_step=1.58s recent_step=1.24s eta=05:43:22 +2026-09-20 22:53:34 | INFO | train rank=0 | step=36990/50000 progress=74.0% loss=0.098798 lr=1.739578e-05 lr_backbone=1.739578e-05 elapsed=16:15:33 avg_step=1.58s recent_step=1.82s eta=05:43:07 +2026-09-20 22:53:46 | INFO | train rank=0 | step=37000/50000 progress=74.0% loss=0.071733 lr=1.737071e-05 lr_backbone=1.737071e-05 elapsed=16:15:45 avg_step=1.58s recent_step=1.20s eta=05:42:50 +2026-09-20 22:54:03 | INFO | train rank=0 | step=37010/50000 progress=74.0% loss=0.077421 lr=1.734566e-05 lr_backbone=1.734566e-05 elapsed=16:16:03 avg_step=1.58s recent_step=1.73s eta=05:42:34 +2026-09-20 22:54:19 | INFO | train rank=0 | step=37020/50000 progress=74.0% loss=0.081615 lr=1.732063e-05 lr_backbone=1.732063e-05 elapsed=16:16:19 avg_step=1.58s recent_step=1.65s eta=05:42:19 +2026-09-20 22:54:34 | INFO | train rank=0 | step=37030/50000 progress=74.1% loss=0.075965 lr=1.729560e-05 lr_backbone=1.729560e-05 elapsed=16:16:34 avg_step=1.58s recent_step=1.49s eta=05:42:03 +2026-09-20 22:54:53 | INFO | train rank=0 | step=37040/50000 progress=74.1% loss=0.071309 lr=1.727060e-05 lr_backbone=1.727060e-05 elapsed=16:16:53 avg_step=1.58s recent_step=1.86s eta=05:41:48 +2026-09-20 22:55:06 | INFO | train rank=0 | step=37050/50000 progress=74.1% loss=0.077749 lr=1.724560e-05 lr_backbone=1.724560e-05 elapsed=16:17:05 avg_step=1.58s recent_step=1.28s eta=05:41:31 +2026-09-20 22:55:23 | INFO | train rank=0 | step=37060/50000 progress=74.1% loss=0.078460 lr=1.722063e-05 lr_backbone=1.722063e-05 elapsed=16:17:23 avg_step=1.58s recent_step=1.73s eta=05:41:16 +2026-09-20 22:55:35 | INFO | train rank=0 | step=37070/50000 progress=74.1% loss=0.097804 lr=1.719566e-05 lr_backbone=1.719566e-05 elapsed=16:17:35 avg_step=1.58s recent_step=1.18s eta=05:40:58 +2026-09-20 22:55:54 | INFO | train rank=0 | step=37080/50000 progress=74.2% loss=0.121909 lr=1.717071e-05 lr_backbone=1.717071e-05 elapsed=16:17:54 avg_step=1.58s recent_step=1.94s eta=05:40:44 +2026-09-20 22:56:16 | INFO | train rank=0 | step=37090/50000 progress=74.2% loss=0.072702 lr=1.714578e-05 lr_backbone=1.714578e-05 elapsed=16:18:15 avg_step=1.58s recent_step=2.14s eta=05:40:30 +2026-09-20 22:56:27 | INFO | train rank=0 | step=37100/50000 progress=74.2% loss=0.079539 lr=1.712086e-05 lr_backbone=1.712086e-05 elapsed=16:18:27 avg_step=1.58s recent_step=1.13s eta=05:40:13 +2026-09-20 22:56:41 | INFO | train rank=0 | step=37110/50000 progress=74.2% loss=0.106189 lr=1.709595e-05 lr_backbone=1.709595e-05 elapsed=16:18:40 avg_step=1.58s recent_step=1.34s eta=05:39:56 +2026-09-20 22:57:00 | INFO | train rank=0 | step=37120/50000 progress=74.2% loss=0.083453 lr=1.707106e-05 lr_backbone=1.707106e-05 elapsed=16:19:00 avg_step=1.58s recent_step=1.97s eta=05:39:41 +2026-09-20 22:57:12 | INFO | train rank=0 | step=37130/50000 progress=74.3% loss=0.092430 lr=1.704618e-05 lr_backbone=1.704618e-05 elapsed=16:19:11 avg_step=1.58s recent_step=1.16s eta=05:39:24 +2026-09-20 22:57:25 | INFO | train rank=0 | step=37140/50000 progress=74.3% loss=0.077966 lr=1.702132e-05 lr_backbone=1.702132e-05 elapsed=16:19:25 avg_step=1.58s recent_step=1.34s eta=05:39:07 +2026-09-20 22:57:43 | INFO | train rank=0 | step=37150/50000 progress=74.3% loss=0.081587 lr=1.699647e-05 lr_backbone=1.699647e-05 elapsed=16:19:43 avg_step=1.58s recent_step=1.80s eta=05:38:52 +2026-09-20 22:57:55 | INFO | train rank=0 | step=37160/50000 progress=74.3% loss=0.094539 lr=1.697163e-05 lr_backbone=1.697163e-05 elapsed=16:19:54 avg_step=1.58s recent_step=1.14s eta=05:38:35 +2026-09-20 22:58:14 | INFO | train rank=0 | step=37170/50000 progress=74.3% loss=0.103940 lr=1.694681e-05 lr_backbone=1.694681e-05 elapsed=16:20:14 avg_step=1.58s recent_step=1.95s eta=05:38:20 +2026-09-20 22:58:31 | INFO | train rank=0 | step=37180/50000 progress=74.4% loss=0.082217 lr=1.692201e-05 lr_backbone=1.692201e-05 elapsed=16:20:30 avg_step=1.58s recent_step=1.64s eta=05:38:05 +2026-09-20 22:58:47 | INFO | train rank=0 | step=37190/50000 progress=74.4% loss=0.120791 lr=1.689721e-05 lr_backbone=1.689721e-05 elapsed=16:20:46 avg_step=1.58s recent_step=1.60s eta=05:37:49 +2026-09-20 22:59:00 | INFO | train rank=0 | step=37200/50000 progress=74.4% loss=0.074226 lr=1.687244e-05 lr_backbone=1.687244e-05 elapsed=16:21:00 avg_step=1.58s recent_step=1.40s eta=05:37:33 +2026-09-20 22:59:15 | INFO | train rank=0 | step=37210/50000 progress=74.4% loss=0.083331 lr=1.684768e-05 lr_backbone=1.684768e-05 elapsed=16:21:14 avg_step=1.58s recent_step=1.44s eta=05:37:16 +2026-09-20 22:59:30 | INFO | train rank=0 | step=37220/50000 progress=74.4% loss=0.058427 lr=1.682293e-05 lr_backbone=1.682293e-05 elapsed=16:21:29 avg_step=1.58s recent_step=1.47s eta=05:37:00 +2026-09-20 22:59:47 | INFO | train rank=0 | step=37230/50000 progress=74.5% loss=0.072949 lr=1.679819e-05 lr_backbone=1.679819e-05 elapsed=16:21:46 avg_step=1.58s recent_step=1.71s eta=05:36:45 +2026-09-20 22:59:59 | INFO | train rank=0 | step=37240/50000 progress=74.5% loss=0.074324 lr=1.677348e-05 lr_backbone=1.677348e-05 elapsed=16:21:59 avg_step=1.58s recent_step=1.28s eta=05:36:28 +2026-09-20 23:00:18 | INFO | train rank=0 | step=37250/50000 progress=74.5% loss=0.077563 lr=1.674877e-05 lr_backbone=1.674877e-05 elapsed=16:22:17 avg_step=1.58s recent_step=1.84s eta=05:36:13 +2026-09-20 23:00:30 | INFO | train rank=0 | step=37260/50000 progress=74.5% loss=0.121107 lr=1.672408e-05 lr_backbone=1.672408e-05 elapsed=16:22:30 avg_step=1.58s recent_step=1.24s eta=05:35:56 +2026-09-20 23:00:44 | INFO | train rank=0 | step=37270/50000 progress=74.5% loss=0.062040 lr=1.669941e-05 lr_backbone=1.669941e-05 elapsed=16:22:44 avg_step=1.58s recent_step=1.39s eta=05:35:39 +2026-09-20 23:00:58 | INFO | train rank=0 | step=37280/50000 progress=74.6% loss=0.093657 lr=1.667475e-05 lr_backbone=1.667475e-05 elapsed=16:22:57 avg_step=1.58s recent_step=1.36s eta=05:35:23 +2026-09-20 23:01:17 | INFO | train rank=0 | step=37290/50000 progress=74.6% loss=0.077879 lr=1.665010e-05 lr_backbone=1.665010e-05 elapsed=16:23:17 avg_step=1.58s recent_step=1.97s eta=05:35:08 +2026-09-20 23:01:37 | INFO | train rank=0 | step=37300/50000 progress=74.6% loss=0.077950 lr=1.662547e-05 lr_backbone=1.662547e-05 elapsed=16:23:36 avg_step=1.58s recent_step=1.93s eta=05:34:54 +2026-09-20 23:01:48 | INFO | train rank=0 | step=37310/50000 progress=74.6% loss=0.093483 lr=1.660085e-05 lr_backbone=1.660085e-05 elapsed=16:23:48 avg_step=1.58s recent_step=1.17s eta=05:34:36 +2026-09-20 23:02:00 | INFO | train rank=0 | step=37320/50000 progress=74.6% loss=0.082659 lr=1.657625e-05 lr_backbone=1.657625e-05 elapsed=16:24:00 avg_step=1.58s recent_step=1.20s eta=05:34:19 +2026-09-20 23:02:19 | INFO | train rank=0 | step=37330/50000 progress=74.7% loss=0.073227 lr=1.655166e-05 lr_backbone=1.655166e-05 elapsed=16:24:19 avg_step=1.58s recent_step=1.88s eta=05:34:05 +2026-09-20 23:02:37 | INFO | train rank=0 | step=37340/50000 progress=74.7% loss=0.073159 lr=1.652709e-05 lr_backbone=1.652709e-05 elapsed=16:24:36 avg_step=1.58s recent_step=1.78s eta=05:33:49 +2026-09-20 23:03:02 | INFO | train rank=0 | step=37350/50000 progress=74.7% loss=0.097879 lr=1.650253e-05 lr_backbone=1.650253e-05 elapsed=16:25:01 avg_step=1.58s recent_step=2.48s eta=05:33:37 +2026-09-20 23:03:53 | INFO | train rank=0 | step=37360/50000 progress=74.7% loss=0.089946 lr=1.647799e-05 lr_backbone=1.647799e-05 elapsed=16:25:53 avg_step=1.58s recent_step=5.16s eta=05:33:33 +2026-09-20 23:04:05 | INFO | train rank=0 | step=37370/50000 progress=74.7% loss=0.094189 lr=1.645346e-05 lr_backbone=1.645346e-05 elapsed=16:26:05 avg_step=1.58s recent_step=1.22s eta=05:33:16 +2026-09-20 23:04:17 | INFO | train rank=0 | step=37380/50000 progress=74.8% loss=0.077005 lr=1.642894e-05 lr_backbone=1.642894e-05 elapsed=16:26:16 avg_step=1.58s recent_step=1.13s eta=05:32:58 +2026-09-20 23:04:29 | INFO | train rank=0 | step=37390/50000 progress=74.8% loss=0.128082 lr=1.640444e-05 lr_backbone=1.640444e-05 elapsed=16:26:29 avg_step=1.58s recent_step=1.27s eta=05:32:42 +2026-09-20 23:04:43 | INFO | train rank=0 | step=37400/50000 progress=74.8% loss=0.084275 lr=1.637996e-05 lr_backbone=1.637996e-05 elapsed=16:26:43 avg_step=1.58s recent_step=1.35s eta=05:32:25 +2026-09-20 23:04:58 | INFO | train rank=0 | step=37410/50000 progress=74.8% loss=0.098649 lr=1.635549e-05 lr_backbone=1.635549e-05 elapsed=16:26:58 avg_step=1.58s recent_step=1.55s eta=05:32:09 +2026-09-20 23:05:12 | INFO | train rank=0 | step=37420/50000 progress=74.8% loss=0.083238 lr=1.633103e-05 lr_backbone=1.633103e-05 elapsed=16:27:12 avg_step=1.58s recent_step=1.38s eta=05:31:52 +2026-09-20 23:05:30 | INFO | train rank=0 | step=37430/50000 progress=74.9% loss=0.125435 lr=1.630659e-05 lr_backbone=1.630659e-05 elapsed=16:27:29 avg_step=1.58s recent_step=1.74s eta=05:31:37 +2026-09-20 23:05:42 | INFO | train rank=0 | step=37440/50000 progress=74.9% loss=0.069651 lr=1.628217e-05 lr_backbone=1.628217e-05 elapsed=16:27:41 avg_step=1.58s recent_step=1.22s eta=05:31:20 +2026-09-20 23:05:55 | INFO | train rank=0 | step=37450/50000 progress=74.9% loss=0.103742 lr=1.625776e-05 lr_backbone=1.625776e-05 elapsed=16:27:55 avg_step=1.58s recent_step=1.34s eta=05:31:03 +2026-09-20 23:06:09 | INFO | train rank=0 | step=37460/50000 progress=74.9% loss=0.077641 lr=1.623336e-05 lr_backbone=1.623336e-05 elapsed=16:28:09 avg_step=1.58s recent_step=1.41s eta=05:30:47 +2026-09-20 23:06:29 | INFO | train rank=0 | step=37470/50000 progress=74.9% loss=0.095809 lr=1.620898e-05 lr_backbone=1.620898e-05 elapsed=16:28:29 avg_step=1.58s recent_step=1.96s eta=05:30:32 +2026-09-20 23:06:43 | INFO | train rank=0 | step=37480/50000 progress=75.0% loss=0.067030 lr=1.618461e-05 lr_backbone=1.618461e-05 elapsed=16:28:43 avg_step=1.58s recent_step=1.44s eta=05:30:16 +2026-09-20 23:07:00 | INFO | train rank=0 | step=37490/50000 progress=75.0% loss=0.076306 lr=1.616026e-05 lr_backbone=1.616026e-05 elapsed=16:29:00 avg_step=1.58s recent_step=1.69s eta=05:30:01 +2026-09-20 23:07:22 | INFO | train rank=0 | step=37500/50000 progress=75.0% loss=0.086842 lr=1.613592e-05 lr_backbone=1.613592e-05 elapsed=16:29:21 avg_step=1.58s recent_step=2.13s eta=05:29:47 +2026-09-20 23:07:36 | INFO | train rank=0 | step=37510/50000 progress=75.0% loss=0.085412 lr=1.611160e-05 lr_backbone=1.611160e-05 elapsed=16:29:36 avg_step=1.58s recent_step=1.48s eta=05:29:31 +2026-09-20 23:07:54 | INFO | train rank=0 | step=37520/50000 progress=75.0% loss=0.082537 lr=1.608729e-05 lr_backbone=1.608729e-05 elapsed=16:29:54 avg_step=1.58s recent_step=1.77s eta=05:29:15 +2026-09-20 23:08:08 | INFO | train rank=0 | step=37530/50000 progress=75.1% loss=0.085856 lr=1.606300e-05 lr_backbone=1.606300e-05 elapsed=16:30:07 avg_step=1.58s recent_step=1.36s eta=05:28:59 +2026-09-20 23:08:25 | INFO | train rank=0 | step=37540/50000 progress=75.1% loss=0.085872 lr=1.603872e-05 lr_backbone=1.603872e-05 elapsed=16:30:24 avg_step=1.58s recent_step=1.70s eta=05:28:43 +2026-09-20 23:08:39 | INFO | train rank=0 | step=37550/50000 progress=75.1% loss=0.091209 lr=1.601446e-05 lr_backbone=1.601446e-05 elapsed=16:30:39 avg_step=1.58s recent_step=1.43s eta=05:28:27 +2026-09-20 23:08:56 | INFO | train rank=0 | step=37560/50000 progress=75.1% loss=0.088434 lr=1.599021e-05 lr_backbone=1.599021e-05 elapsed=16:30:55 avg_step=1.58s recent_step=1.68s eta=05:28:11 +2026-09-20 23:09:11 | INFO | train rank=0 | step=37570/50000 progress=75.1% loss=0.080812 lr=1.596598e-05 lr_backbone=1.596598e-05 elapsed=16:31:11 avg_step=1.58s recent_step=1.51s eta=05:27:55 +2026-09-20 23:09:33 | INFO | train rank=0 | step=37580/50000 progress=75.2% loss=0.093829 lr=1.594176e-05 lr_backbone=1.594176e-05 elapsed=16:31:32 avg_step=1.58s recent_step=2.19s eta=05:27:42 +2026-09-20 23:09:45 | INFO | train rank=0 | step=37590/50000 progress=75.2% loss=0.085715 lr=1.591755e-05 lr_backbone=1.591755e-05 elapsed=16:31:45 avg_step=1.58s recent_step=1.22s eta=05:27:25 +2026-09-20 23:09:58 | INFO | train rank=0 | step=37600/50000 progress=75.2% loss=0.116275 lr=1.589336e-05 lr_backbone=1.589336e-05 elapsed=16:31:57 avg_step=1.58s recent_step=1.25s eta=05:27:08 +2026-09-20 23:10:12 | INFO | train rank=0 | step=37610/50000 progress=75.2% loss=0.109619 lr=1.586919e-05 lr_backbone=1.586919e-05 elapsed=16:32:12 avg_step=1.58s recent_step=1.45s eta=05:26:51 +2026-09-20 23:10:32 | INFO | train rank=0 | step=37620/50000 progress=75.2% loss=0.075534 lr=1.584503e-05 lr_backbone=1.584503e-05 elapsed=16:32:32 avg_step=1.58s recent_step=2.02s eta=05:26:37 +2026-09-20 23:10:50 | INFO | train rank=0 | step=37630/50000 progress=75.3% loss=0.091405 lr=1.582089e-05 lr_backbone=1.582089e-05 elapsed=16:32:49 avg_step=1.58s recent_step=1.75s eta=05:26:22 +2026-09-20 23:11:01 | INFO | train rank=0 | step=37640/50000 progress=75.3% loss=0.084157 lr=1.579676e-05 lr_backbone=1.579676e-05 elapsed=16:33:01 avg_step=1.58s recent_step=1.18s eta=05:26:05 +2026-09-20 23:11:18 | INFO | train rank=0 | step=37650/50000 progress=75.3% loss=0.102470 lr=1.577264e-05 lr_backbone=1.577264e-05 elapsed=16:33:17 avg_step=1.58s recent_step=1.62s eta=05:25:49 +2026-09-20 23:11:35 | INFO | train rank=0 | step=37660/50000 progress=75.3% loss=0.075010 lr=1.574855e-05 lr_backbone=1.574855e-05 elapsed=16:33:34 avg_step=1.58s recent_step=1.70s eta=05:25:33 +2026-09-20 23:11:48 | INFO | train rank=0 | step=37670/50000 progress=75.3% loss=0.081889 lr=1.572446e-05 lr_backbone=1.572446e-05 elapsed=16:33:48 avg_step=1.58s recent_step=1.36s eta=05:25:17 +2026-09-20 23:12:07 | INFO | train rank=0 | step=37680/50000 progress=75.4% loss=0.094202 lr=1.570039e-05 lr_backbone=1.570039e-05 elapsed=16:34:07 avg_step=1.58s recent_step=1.90s eta=05:25:02 +2026-09-20 23:12:26 | INFO | train rank=0 | step=37690/50000 progress=75.4% loss=0.078780 lr=1.567634e-05 lr_backbone=1.567634e-05 elapsed=16:34:26 avg_step=1.58s recent_step=1.91s eta=05:24:47 +2026-09-20 23:12:39 | INFO | train rank=0 | step=37700/50000 progress=75.4% loss=0.084238 lr=1.565230e-05 lr_backbone=1.565230e-05 elapsed=16:34:38 avg_step=1.58s recent_step=1.22s eta=05:24:30 +2026-09-20 23:12:51 | INFO | train rank=0 | step=37710/50000 progress=75.4% loss=0.095596 lr=1.562828e-05 lr_backbone=1.562828e-05 elapsed=16:34:50 avg_step=1.58s recent_step=1.23s eta=05:24:13 +2026-09-20 23:13:03 | INFO | train rank=0 | step=37720/50000 progress=75.4% loss=0.095455 lr=1.560427e-05 lr_backbone=1.560427e-05 elapsed=16:35:03 avg_step=1.58s recent_step=1.21s eta=05:23:56 +2026-09-20 23:13:24 | INFO | train rank=0 | step=37730/50000 progress=75.5% loss=0.094281 lr=1.558027e-05 lr_backbone=1.558027e-05 elapsed=16:35:23 avg_step=1.58s recent_step=2.07s eta=05:23:42 +2026-09-20 23:13:37 | INFO | train rank=0 | step=37740/50000 progress=75.5% loss=0.068412 lr=1.555629e-05 lr_backbone=1.555629e-05 elapsed=16:35:36 avg_step=1.58s recent_step=1.30s eta=05:23:25 +2026-09-20 23:13:58 | INFO | train rank=0 | step=37750/50000 progress=75.5% loss=0.102450 lr=1.553233e-05 lr_backbone=1.553233e-05 elapsed=16:35:58 avg_step=1.58s recent_step=2.15s eta=05:23:11 +2026-09-20 23:14:11 | INFO | train rank=0 | step=37760/50000 progress=75.5% loss=0.074351 lr=1.550838e-05 lr_backbone=1.550838e-05 elapsed=16:36:10 avg_step=1.58s recent_step=1.26s eta=05:22:54 +2026-09-20 23:14:28 | INFO | train rank=0 | step=37770/50000 progress=75.5% loss=0.108161 lr=1.548445e-05 lr_backbone=1.548445e-05 elapsed=16:36:28 avg_step=1.58s recent_step=1.72s eta=05:22:39 +2026-09-20 23:14:42 | INFO | train rank=0 | step=37780/50000 progress=75.6% loss=0.084461 lr=1.546053e-05 lr_backbone=1.546053e-05 elapsed=16:36:41 avg_step=1.58s recent_step=1.38s eta=05:22:23 +2026-09-20 23:14:59 | INFO | train rank=0 | step=37790/50000 progress=75.6% loss=0.098102 lr=1.543663e-05 lr_backbone=1.543663e-05 elapsed=16:36:59 avg_step=1.58s recent_step=1.73s eta=05:22:07 +2026-09-20 23:15:14 | INFO | train rank=0 | step=37800/50000 progress=75.6% loss=0.087487 lr=1.541274e-05 lr_backbone=1.541274e-05 elapsed=16:37:13 avg_step=1.58s recent_step=1.44s eta=05:21:51 +2026-09-20 23:15:29 | INFO | train rank=0 | step=37810/50000 progress=75.6% loss=0.071973 lr=1.538886e-05 lr_backbone=1.538886e-05 elapsed=16:37:29 avg_step=1.58s recent_step=1.54s eta=05:21:35 +2026-09-20 23:15:48 | INFO | train rank=0 | step=37820/50000 progress=75.6% loss=0.093875 lr=1.536501e-05 lr_backbone=1.536501e-05 elapsed=16:37:48 avg_step=1.58s recent_step=1.91s eta=05:21:20 +2026-09-20 23:16:04 | INFO | train rank=0 | step=37830/50000 progress=75.7% loss=0.074477 lr=1.534116e-05 lr_backbone=1.534116e-05 elapsed=16:38:04 avg_step=1.58s recent_step=1.61s eta=05:21:04 +2026-09-20 23:16:16 | INFO | train rank=0 | step=37840/50000 progress=75.7% loss=0.072707 lr=1.531733e-05 lr_backbone=1.531733e-05 elapsed=16:38:16 avg_step=1.58s recent_step=1.21s eta=05:20:47 +2026-09-20 23:16:31 | INFO | train rank=0 | step=37850/50000 progress=75.7% loss=0.095226 lr=1.529352e-05 lr_backbone=1.529352e-05 elapsed=16:38:31 avg_step=1.58s recent_step=1.48s eta=05:20:31 +2026-09-20 23:16:51 | INFO | train rank=0 | step=37860/50000 progress=75.7% loss=0.064591 lr=1.526972e-05 lr_backbone=1.526972e-05 elapsed=16:38:50 avg_step=1.58s recent_step=1.97s eta=05:20:17 +2026-09-20 23:17:08 | INFO | train rank=0 | step=37870/50000 progress=75.7% loss=0.086203 lr=1.524594e-05 lr_backbone=1.524594e-05 elapsed=16:39:08 avg_step=1.58s recent_step=1.78s eta=05:20:01 +2026-09-20 23:17:20 | INFO | train rank=0 | step=37880/50000 progress=75.8% loss=0.083664 lr=1.522218e-05 lr_backbone=1.522218e-05 elapsed=16:39:19 avg_step=1.58s recent_step=1.14s eta=05:19:44 +2026-09-20 23:17:37 | INFO | train rank=0 | step=37890/50000 progress=75.8% loss=0.074673 lr=1.519842e-05 lr_backbone=1.519842e-05 elapsed=16:39:36 avg_step=1.58s recent_step=1.69s eta=05:19:29 +2026-09-20 23:17:50 | INFO | train rank=0 | step=37900/50000 progress=75.8% loss=0.102441 lr=1.517469e-05 lr_backbone=1.517469e-05 elapsed=16:39:50 avg_step=1.58s recent_step=1.35s eta=05:19:12 +2026-09-20 23:18:07 | INFO | train rank=0 | step=37910/50000 progress=75.8% loss=0.083224 lr=1.515097e-05 lr_backbone=1.515097e-05 elapsed=16:40:07 avg_step=1.58s recent_step=1.71s eta=05:18:57 +2026-09-20 23:18:24 | INFO | train rank=0 | step=37920/50000 progress=75.8% loss=0.089463 lr=1.512726e-05 lr_backbone=1.512726e-05 elapsed=16:40:24 avg_step=1.58s recent_step=1.65s eta=05:18:41 +2026-09-20 23:18:42 | INFO | train rank=0 | step=37930/50000 progress=75.9% loss=0.090613 lr=1.510357e-05 lr_backbone=1.510357e-05 elapsed=16:40:41 avg_step=1.58s recent_step=1.77s eta=05:18:26 +2026-09-20 23:18:53 | INFO | train rank=0 | step=37940/50000 progress=75.9% loss=0.069979 lr=1.507989e-05 lr_backbone=1.507989e-05 elapsed=16:40:53 avg_step=1.58s recent_step=1.18s eta=05:18:09 +2026-09-20 23:19:14 | INFO | train rank=0 | step=37950/50000 progress=75.9% loss=0.078951 lr=1.505623e-05 lr_backbone=1.505623e-05 elapsed=16:41:13 avg_step=1.58s recent_step=2.03s eta=05:17:54 +2026-09-20 23:19:26 | INFO | train rank=0 | step=37960/50000 progress=75.9% loss=0.077526 lr=1.503259e-05 lr_backbone=1.503259e-05 elapsed=16:41:25 avg_step=1.58s recent_step=1.20s eta=05:17:37 +2026-09-20 23:19:42 | INFO | train rank=0 | step=37970/50000 progress=75.9% loss=0.080340 lr=1.500896e-05 lr_backbone=1.500896e-05 elapsed=16:41:42 avg_step=1.58s recent_step=1.65s eta=05:17:22 +2026-09-20 23:20:00 | INFO | train rank=0 | step=37980/50000 progress=76.0% loss=0.086232 lr=1.498534e-05 lr_backbone=1.498534e-05 elapsed=16:41:59 avg_step=1.58s recent_step=1.75s eta=05:17:06 +2026-09-20 23:20:15 | INFO | train rank=0 | step=37990/50000 progress=76.0% loss=0.096066 lr=1.496174e-05 lr_backbone=1.496174e-05 elapsed=16:42:15 avg_step=1.58s recent_step=1.57s eta=05:16:51 +2026-09-20 23:20:29 | INFO | train rank=0 | step=38000/50000 progress=76.0% loss=0.080951 lr=1.493816e-05 lr_backbone=1.493816e-05 elapsed=16:42:29 avg_step=1.58s recent_step=1.37s eta=05:16:34 +2026-09-20 23:20:43 | INFO | train rank=0 | step=38010/50000 progress=76.0% loss=0.088904 lr=1.491459e-05 lr_backbone=1.491459e-05 elapsed=16:42:43 avg_step=1.58s recent_step=1.43s eta=05:16:18 +2026-09-20 23:21:02 | INFO | train rank=0 | step=38020/50000 progress=76.0% loss=0.078496 lr=1.489104e-05 lr_backbone=1.489104e-05 elapsed=16:43:02 avg_step=1.58s recent_step=1.87s eta=05:16:03 +2026-09-20 23:21:17 | INFO | train rank=0 | step=38030/50000 progress=76.1% loss=0.092648 lr=1.486750e-05 lr_backbone=1.486750e-05 elapsed=16:43:16 avg_step=1.58s recent_step=1.45s eta=05:15:47 +2026-09-20 23:21:31 | INFO | train rank=0 | step=38040/50000 progress=76.1% loss=0.094356 lr=1.484398e-05 lr_backbone=1.484398e-05 elapsed=16:43:31 avg_step=1.58s recent_step=1.48s eta=05:15:30 +2026-09-20 23:21:46 | INFO | train rank=0 | step=38050/50000 progress=76.1% loss=0.116793 lr=1.482047e-05 lr_backbone=1.482047e-05 elapsed=16:43:45 avg_step=1.58s recent_step=1.45s eta=05:15:14 +2026-09-20 23:22:05 | INFO | train rank=0 | step=38060/50000 progress=76.1% loss=0.088788 lr=1.479698e-05 lr_backbone=1.479698e-05 elapsed=16:44:04 avg_step=1.58s recent_step=1.90s eta=05:14:59 +2026-09-20 23:22:17 | INFO | train rank=0 | step=38070/50000 progress=76.1% loss=0.063496 lr=1.477350e-05 lr_backbone=1.477350e-05 elapsed=16:44:17 avg_step=1.58s recent_step=1.23s eta=05:14:42 +2026-09-20 23:22:33 | INFO | train rank=0 | step=38080/50000 progress=76.2% loss=0.122642 lr=1.475004e-05 lr_backbone=1.475004e-05 elapsed=16:44:32 avg_step=1.58s recent_step=1.55s eta=05:14:26 +2026-09-20 23:22:49 | INFO | train rank=0 | step=38090/50000 progress=76.2% loss=0.115516 lr=1.472660e-05 lr_backbone=1.472660e-05 elapsed=16:44:49 avg_step=1.58s recent_step=1.66s eta=05:14:11 +2026-09-20 23:23:06 | INFO | train rank=0 | step=38100/50000 progress=76.2% loss=0.072926 lr=1.470317e-05 lr_backbone=1.470317e-05 elapsed=16:45:05 avg_step=1.58s recent_step=1.65s eta=05:13:55 +2026-09-20 23:23:21 | INFO | train rank=0 | step=38110/50000 progress=76.2% loss=0.077549 lr=1.467975e-05 lr_backbone=1.467975e-05 elapsed=16:45:20 avg_step=1.58s recent_step=1.51s eta=05:13:39 +2026-09-20 23:23:37 | INFO | train rank=0 | step=38120/50000 progress=76.2% loss=0.130061 lr=1.465635e-05 lr_backbone=1.465635e-05 elapsed=16:45:37 avg_step=1.58s recent_step=1.61s eta=05:13:23 +2026-09-20 23:23:49 | INFO | train rank=0 | step=38130/50000 progress=76.3% loss=0.082465 lr=1.463297e-05 lr_backbone=1.463297e-05 elapsed=16:45:48 avg_step=1.58s recent_step=1.19s eta=05:13:06 +2026-09-20 23:24:01 | INFO | train rank=0 | step=38140/50000 progress=76.3% loss=0.092950 lr=1.460960e-05 lr_backbone=1.460960e-05 elapsed=16:46:01 avg_step=1.58s recent_step=1.23s eta=05:12:49 +2026-09-20 23:24:18 | INFO | train rank=0 | step=38150/50000 progress=76.3% loss=0.079119 lr=1.458625e-05 lr_backbone=1.458625e-05 elapsed=16:46:17 avg_step=1.58s recent_step=1.65s eta=05:12:34 +2026-09-20 23:24:37 | INFO | train rank=0 | step=38160/50000 progress=76.3% loss=0.088395 lr=1.456291e-05 lr_backbone=1.456291e-05 elapsed=16:46:37 avg_step=1.58s recent_step=1.98s eta=05:12:19 +2026-09-20 23:24:50 | INFO | train rank=0 | step=38170/50000 progress=76.3% loss=0.082975 lr=1.453959e-05 lr_backbone=1.453959e-05 elapsed=16:46:49 avg_step=1.58s recent_step=1.22s eta=05:12:02 +2026-09-20 23:25:10 | INFO | train rank=0 | step=38180/50000 progress=76.4% loss=0.106011 lr=1.451629e-05 lr_backbone=1.451629e-05 elapsed=16:47:09 avg_step=1.58s recent_step=2.01s eta=05:11:48 +2026-09-20 23:25:29 | INFO | train rank=0 | step=38190/50000 progress=76.4% loss=0.080712 lr=1.449299e-05 lr_backbone=1.449299e-05 elapsed=16:47:29 avg_step=1.58s recent_step=1.94s eta=05:11:33 +2026-09-20 23:25:41 | INFO | train rank=0 | step=38200/50000 progress=76.4% loss=0.111174 lr=1.446972e-05 lr_backbone=1.446972e-05 elapsed=16:47:40 avg_step=1.58s recent_step=1.17s eta=05:11:16 +2026-09-20 23:25:58 | INFO | train rank=0 | step=38210/50000 progress=76.4% loss=0.121362 lr=1.444646e-05 lr_backbone=1.444646e-05 elapsed=16:47:57 avg_step=1.58s recent_step=1.70s eta=05:11:00 +2026-09-20 23:26:10 | INFO | train rank=0 | step=38220/50000 progress=76.4% loss=0.084657 lr=1.442322e-05 lr_backbone=1.442322e-05 elapsed=16:48:09 avg_step=1.58s recent_step=1.19s eta=05:10:43 +2026-09-20 23:26:26 | INFO | train rank=0 | step=38230/50000 progress=76.5% loss=0.076959 lr=1.439999e-05 lr_backbone=1.439999e-05 elapsed=16:48:25 avg_step=1.58s recent_step=1.60s eta=05:10:28 +2026-09-20 23:26:42 | INFO | train rank=0 | step=38240/50000 progress=76.5% loss=0.072402 lr=1.437677e-05 lr_backbone=1.437677e-05 elapsed=16:48:42 avg_step=1.58s recent_step=1.64s eta=05:10:12 +2026-09-20 23:26:56 | INFO | train rank=0 | step=38250/50000 progress=76.5% loss=0.073521 lr=1.435358e-05 lr_backbone=1.435358e-05 elapsed=16:48:55 avg_step=1.58s recent_step=1.36s eta=05:09:55 +2026-09-20 23:27:14 | INFO | train rank=0 | step=38260/50000 progress=76.5% loss=0.073703 lr=1.433040e-05 lr_backbone=1.433040e-05 elapsed=16:49:13 avg_step=1.58s recent_step=1.79s eta=05:09:40 +2026-09-20 23:27:33 | INFO | train rank=0 | step=38270/50000 progress=76.5% loss=0.089114 lr=1.430723e-05 lr_backbone=1.430723e-05 elapsed=16:49:32 avg_step=1.58s recent_step=1.92s eta=05:09:25 +2026-09-20 23:27:45 | INFO | train rank=0 | step=38280/50000 progress=76.6% loss=0.084242 lr=1.428408e-05 lr_backbone=1.428408e-05 elapsed=16:49:44 avg_step=1.58s recent_step=1.21s eta=05:09:09 +2026-09-20 23:27:59 | INFO | train rank=0 | step=38290/50000 progress=76.6% loss=0.059072 lr=1.426094e-05 lr_backbone=1.426094e-05 elapsed=16:49:59 avg_step=1.58s recent_step=1.45s eta=05:08:52 +2026-09-20 23:28:17 | INFO | train rank=0 | step=38300/50000 progress=76.6% loss=0.080852 lr=1.423783e-05 lr_backbone=1.423783e-05 elapsed=16:50:16 avg_step=1.58s recent_step=1.72s eta=05:08:37 +2026-09-20 23:28:33 | INFO | train rank=0 | step=38310/50000 progress=76.6% loss=0.072068 lr=1.421472e-05 lr_backbone=1.421472e-05 elapsed=16:50:32 avg_step=1.58s recent_step=1.62s eta=05:08:21 +2026-09-20 23:28:51 | INFO | train rank=0 | step=38320/50000 progress=76.6% loss=0.070767 lr=1.419163e-05 lr_backbone=1.419163e-05 elapsed=16:50:51 avg_step=1.58s recent_step=1.84s eta=05:08:06 +2026-09-20 23:29:08 | INFO | train rank=0 | step=38330/50000 progress=76.7% loss=0.069839 lr=1.416856e-05 lr_backbone=1.416856e-05 elapsed=16:51:07 avg_step=1.58s recent_step=1.67s eta=05:07:51 +2026-09-20 23:29:20 | INFO | train rank=0 | step=38340/50000 progress=76.7% loss=0.090706 lr=1.414551e-05 lr_backbone=1.414551e-05 elapsed=16:51:20 avg_step=1.58s recent_step=1.27s eta=05:07:34 +2026-09-20 23:29:36 | INFO | train rank=0 | step=38350/50000 progress=76.7% loss=0.070470 lr=1.412246e-05 lr_backbone=1.412246e-05 elapsed=16:51:36 avg_step=1.58s recent_step=1.59s eta=05:07:18 +2026-09-20 23:29:51 | INFO | train rank=0 | step=38360/50000 progress=76.7% loss=0.098814 lr=1.409944e-05 lr_backbone=1.409944e-05 elapsed=16:51:50 avg_step=1.58s recent_step=1.41s eta=05:07:02 +2026-09-20 23:30:05 | INFO | train rank=0 | step=38370/50000 progress=76.7% loss=0.089108 lr=1.407643e-05 lr_backbone=1.407643e-05 elapsed=16:52:04 avg_step=1.58s recent_step=1.42s eta=05:06:45 +2026-09-20 23:30:22 | INFO | train rank=0 | step=38380/50000 progress=76.8% loss=0.092793 lr=1.405344e-05 lr_backbone=1.405344e-05 elapsed=16:52:22 avg_step=1.58s recent_step=1.76s eta=05:06:30 +2026-09-20 23:30:40 | INFO | train rank=0 | step=38390/50000 progress=76.8% loss=0.075737 lr=1.403046e-05 lr_backbone=1.403046e-05 elapsed=16:52:40 avg_step=1.58s recent_step=1.81s eta=05:06:15 +2026-09-20 23:30:53 | INFO | train rank=0 | step=38400/50000 progress=76.8% loss=0.083126 lr=1.400750e-05 lr_backbone=1.400750e-05 elapsed=16:52:52 avg_step=1.58s recent_step=1.22s eta=05:05:58 +2026-09-20 23:31:07 | INFO | train rank=0 | step=38410/50000 progress=76.8% loss=0.100349 lr=1.398455e-05 lr_backbone=1.398455e-05 elapsed=16:53:06 avg_step=1.58s recent_step=1.42s eta=05:05:42 +2026-09-20 23:31:20 | INFO | train rank=0 | step=38420/50000 progress=76.8% loss=0.087515 lr=1.396162e-05 lr_backbone=1.396162e-05 elapsed=16:53:20 avg_step=1.58s recent_step=1.36s eta=05:05:25 +2026-09-20 23:31:37 | INFO | train rank=0 | step=38430/50000 progress=76.9% loss=0.098870 lr=1.393870e-05 lr_backbone=1.393870e-05 elapsed=16:53:37 avg_step=1.58s recent_step=1.67s eta=05:05:10 +2026-09-20 23:31:56 | INFO | train rank=0 | step=38440/50000 progress=76.9% loss=0.076693 lr=1.391580e-05 lr_backbone=1.391580e-05 elapsed=16:53:55 avg_step=1.58s recent_step=1.85s eta=05:04:55 +2026-09-20 23:32:09 | INFO | train rank=0 | step=38450/50000 progress=76.9% loss=0.073654 lr=1.389292e-05 lr_backbone=1.389292e-05 elapsed=16:54:09 avg_step=1.58s recent_step=1.32s eta=05:04:38 +2026-09-20 23:32:28 | INFO | train rank=0 | step=38460/50000 progress=76.9% loss=0.071854 lr=1.387005e-05 lr_backbone=1.387005e-05 elapsed=16:54:28 avg_step=1.58s recent_step=1.96s eta=05:04:23 +2026-09-20 23:32:40 | INFO | train rank=0 | step=38470/50000 progress=76.9% loss=0.107484 lr=1.384720e-05 lr_backbone=1.384720e-05 elapsed=16:54:40 avg_step=1.58s recent_step=1.18s eta=05:04:06 +2026-09-20 23:32:58 | INFO | train rank=0 | step=38480/50000 progress=77.0% loss=0.074629 lr=1.382436e-05 lr_backbone=1.382436e-05 elapsed=16:54:58 avg_step=1.58s recent_step=1.82s eta=05:03:51 +2026-09-20 23:33:16 | INFO | train rank=0 | step=38490/50000 progress=77.0% loss=0.122276 lr=1.380154e-05 lr_backbone=1.380154e-05 elapsed=16:55:16 avg_step=1.58s recent_step=1.79s eta=05:03:36 +2026-09-20 23:33:34 | INFO | train rank=0 | step=38500/50000 progress=77.0% loss=0.072451 lr=1.377874e-05 lr_backbone=1.377874e-05 elapsed=16:55:34 avg_step=1.58s recent_step=1.75s eta=05:03:21 +2026-09-20 23:33:46 | INFO | train rank=0 | step=38510/50000 progress=77.0% loss=0.074814 lr=1.375595e-05 lr_backbone=1.375595e-05 elapsed=16:55:46 avg_step=1.58s recent_step=1.22s eta=05:03:04 +2026-09-20 23:34:05 | INFO | train rank=0 | step=38520/50000 progress=77.0% loss=0.097280 lr=1.373318e-05 lr_backbone=1.373318e-05 elapsed=16:56:04 avg_step=1.58s recent_step=1.87s eta=05:02:49 +2026-09-20 23:34:16 | INFO | train rank=0 | step=38530/50000 progress=77.1% loss=0.078784 lr=1.371042e-05 lr_backbone=1.371042e-05 elapsed=16:56:16 avg_step=1.58s recent_step=1.16s eta=05:02:32 +2026-09-20 23:34:35 | INFO | train rank=0 | step=38540/50000 progress=77.1% loss=0.094435 lr=1.368768e-05 lr_backbone=1.368768e-05 elapsed=16:56:35 avg_step=1.58s recent_step=1.90s eta=05:02:17 +2026-09-20 23:34:48 | INFO | train rank=0 | step=38550/50000 progress=77.1% loss=0.080004 lr=1.366496e-05 lr_backbone=1.366496e-05 elapsed=16:56:47 avg_step=1.58s recent_step=1.23s eta=05:02:00 +2026-09-20 23:35:02 | INFO | train rank=0 | step=38560/50000 progress=77.1% loss=0.081452 lr=1.364225e-05 lr_backbone=1.364225e-05 elapsed=16:57:02 avg_step=1.58s recent_step=1.43s eta=05:01:44 +2026-09-20 23:35:18 | INFO | train rank=0 | step=38570/50000 progress=77.1% loss=0.088874 lr=1.361955e-05 lr_backbone=1.361955e-05 elapsed=16:57:18 avg_step=1.58s recent_step=1.60s eta=05:01:28 +2026-09-20 23:35:37 | INFO | train rank=0 | step=38580/50000 progress=77.2% loss=0.112091 lr=1.359688e-05 lr_backbone=1.359688e-05 elapsed=16:57:37 avg_step=1.58s recent_step=1.89s eta=05:01:13 +2026-09-20 23:35:49 | INFO | train rank=0 | step=38590/50000 progress=77.2% loss=0.134241 lr=1.357421e-05 lr_backbone=1.357421e-05 elapsed=16:57:48 avg_step=1.58s recent_step=1.16s eta=05:00:56 +2026-09-20 23:36:02 | INFO | train rank=0 | step=38600/50000 progress=77.2% loss=0.107357 lr=1.355157e-05 lr_backbone=1.355157e-05 elapsed=16:58:02 avg_step=1.58s recent_step=1.35s eta=05:00:39 +2026-09-20 23:36:14 | INFO | train rank=0 | step=38610/50000 progress=77.2% loss=0.080655 lr=1.352894e-05 lr_backbone=1.352894e-05 elapsed=16:58:14 avg_step=1.58s recent_step=1.22s eta=05:00:22 +2026-09-20 23:36:31 | INFO | train rank=0 | step=38620/50000 progress=77.2% loss=0.119456 lr=1.350633e-05 lr_backbone=1.350633e-05 elapsed=16:58:31 avg_step=1.58s recent_step=1.70s eta=05:00:07 +2026-09-20 23:36:44 | INFO | train rank=0 | step=38630/50000 progress=77.3% loss=0.059189 lr=1.348373e-05 lr_backbone=1.348373e-05 elapsed=16:58:44 avg_step=1.58s recent_step=1.31s eta=04:59:50 +2026-09-20 23:37:05 | INFO | train rank=0 | step=38640/50000 progress=77.3% loss=0.111918 lr=1.346115e-05 lr_backbone=1.346115e-05 elapsed=16:59:04 avg_step=1.58s recent_step=2.03s eta=04:59:36 +2026-09-20 23:37:21 | INFO | train rank=0 | step=38650/50000 progress=77.3% loss=0.079113 lr=1.343858e-05 lr_backbone=1.343858e-05 elapsed=16:59:21 avg_step=1.58s recent_step=1.66s eta=04:59:20 +2026-09-20 23:37:38 | INFO | train rank=0 | step=38660/50000 progress=77.3% loss=0.079167 lr=1.341603e-05 lr_backbone=1.341603e-05 elapsed=16:59:37 avg_step=1.58s recent_step=1.64s eta=04:59:05 +2026-09-20 23:37:50 | INFO | train rank=0 | step=38670/50000 progress=77.3% loss=0.082357 lr=1.339350e-05 lr_backbone=1.339350e-05 elapsed=16:59:49 avg_step=1.58s recent_step=1.18s eta=04:58:48 +2026-09-20 23:38:05 | INFO | train rank=0 | step=38680/50000 progress=77.4% loss=0.082624 lr=1.337098e-05 lr_backbone=1.337098e-05 elapsed=17:00:04 avg_step=1.58s recent_step=1.53s eta=04:58:32 +2026-09-20 23:38:17 | INFO | train rank=0 | step=38690/50000 progress=77.4% loss=0.072578 lr=1.334848e-05 lr_backbone=1.334848e-05 elapsed=17:00:16 avg_step=1.58s recent_step=1.17s eta=04:58:15 +2026-09-20 23:38:35 | INFO | train rank=0 | step=38700/50000 progress=77.4% loss=0.088348 lr=1.332599e-05 lr_backbone=1.332599e-05 elapsed=17:00:35 avg_step=1.58s recent_step=1.87s eta=04:58:00 +2026-09-20 23:38:50 | INFO | train rank=0 | step=38710/50000 progress=77.4% loss=0.084907 lr=1.330352e-05 lr_backbone=1.330352e-05 elapsed=17:00:50 avg_step=1.58s recent_step=1.50s eta=04:57:44 +2026-09-20 23:39:16 | INFO | train rank=0 | step=38720/50000 progress=77.4% loss=0.095764 lr=1.328107e-05 lr_backbone=1.328107e-05 elapsed=17:01:15 avg_step=1.58s recent_step=2.55s eta=04:57:31 +2026-09-20 23:40:06 | INFO | train rank=0 | step=38730/50000 progress=77.5% loss=0.096455 lr=1.325863e-05 lr_backbone=1.325863e-05 elapsed=17:02:06 avg_step=1.58s recent_step=5.03s eta=04:57:25 +2026-09-20 23:40:19 | INFO | train rank=0 | step=38740/50000 progress=77.5% loss=0.077078 lr=1.323621e-05 lr_backbone=1.323621e-05 elapsed=17:02:18 avg_step=1.58s recent_step=1.27s eta=04:57:08 +2026-09-20 23:40:31 | INFO | train rank=0 | step=38750/50000 progress=77.5% loss=0.086162 lr=1.321380e-05 lr_backbone=1.321380e-05 elapsed=17:02:30 avg_step=1.58s recent_step=1.20s eta=04:56:51 +2026-09-20 23:40:43 | INFO | train rank=0 | step=38760/50000 progress=77.5% loss=0.075111 lr=1.319142e-05 lr_backbone=1.319142e-05 elapsed=17:02:43 avg_step=1.58s recent_step=1.25s eta=04:56:34 +2026-09-20 23:40:59 | INFO | train rank=0 | step=38770/50000 progress=77.5% loss=0.079034 lr=1.316904e-05 lr_backbone=1.316904e-05 elapsed=17:02:59 avg_step=1.58s recent_step=1.62s eta=04:56:19 +2026-09-20 23:41:11 | INFO | train rank=0 | step=38780/50000 progress=77.6% loss=0.098284 lr=1.314669e-05 lr_backbone=1.314669e-05 elapsed=17:03:11 avg_step=1.58s recent_step=1.16s eta=04:56:01 +2026-09-20 23:41:27 | INFO | train rank=0 | step=38790/50000 progress=77.6% loss=0.078943 lr=1.312434e-05 lr_backbone=1.312434e-05 elapsed=17:03:26 avg_step=1.58s recent_step=1.56s eta=04:55:46 +2026-09-20 23:41:41 | INFO | train rank=0 | step=38800/50000 progress=77.6% loss=0.078695 lr=1.310202e-05 lr_backbone=1.310202e-05 elapsed=17:03:41 avg_step=1.58s recent_step=1.47s eta=04:55:29 +2026-09-20 23:41:56 | INFO | train rank=0 | step=38810/50000 progress=77.6% loss=0.104792 lr=1.307971e-05 lr_backbone=1.307971e-05 elapsed=17:03:56 avg_step=1.58s recent_step=1.51s eta=04:55:13 +2026-09-20 23:42:10 | INFO | train rank=0 | step=38820/50000 progress=77.6% loss=0.103626 lr=1.305742e-05 lr_backbone=1.305742e-05 elapsed=17:04:09 avg_step=1.58s recent_step=1.32s eta=04:54:57 +2026-09-20 23:42:23 | INFO | train rank=0 | step=38830/50000 progress=77.7% loss=0.085137 lr=1.303514e-05 lr_backbone=1.303514e-05 elapsed=17:04:22 avg_step=1.58s recent_step=1.31s eta=04:54:40 +2026-09-20 23:42:36 | INFO | train rank=0 | step=38840/50000 progress=77.7% loss=0.147180 lr=1.301288e-05 lr_backbone=1.301288e-05 elapsed=17:04:35 avg_step=1.58s recent_step=1.28s eta=04:54:23 +2026-09-20 23:42:53 | INFO | train rank=0 | step=38850/50000 progress=77.7% loss=0.072781 lr=1.299064e-05 lr_backbone=1.299064e-05 elapsed=17:04:53 avg_step=1.58s recent_step=1.75s eta=04:54:08 +2026-09-20 23:43:08 | INFO | train rank=0 | step=38860/50000 progress=77.7% loss=0.150856 lr=1.296841e-05 lr_backbone=1.296841e-05 elapsed=17:05:07 avg_step=1.58s recent_step=1.47s eta=04:53:52 +2026-09-20 23:43:32 | INFO | train rank=0 | step=38870/50000 progress=77.7% loss=0.078461 lr=1.294620e-05 lr_backbone=1.294620e-05 elapsed=17:05:31 avg_step=1.58s recent_step=2.40s eta=04:53:38 +2026-09-20 23:43:44 | INFO | train rank=0 | step=38880/50000 progress=77.8% loss=0.115167 lr=1.292400e-05 lr_backbone=1.292400e-05 elapsed=17:05:44 avg_step=1.58s recent_step=1.23s eta=04:53:22 +2026-09-20 23:44:04 | INFO | train rank=0 | step=38890/50000 progress=77.8% loss=0.079315 lr=1.290182e-05 lr_backbone=1.290182e-05 elapsed=17:06:04 avg_step=1.58s recent_step=2.01s eta=04:53:07 +2026-09-20 23:44:19 | INFO | train rank=0 | step=38900/50000 progress=77.8% loss=0.104512 lr=1.287966e-05 lr_backbone=1.287966e-05 elapsed=17:06:19 avg_step=1.58s recent_step=1.53s eta=04:52:51 +2026-09-20 23:44:35 | INFO | train rank=0 | step=38910/50000 progress=77.8% loss=0.083412 lr=1.285751e-05 lr_backbone=1.285751e-05 elapsed=17:06:35 avg_step=1.58s recent_step=1.58s eta=04:52:35 +2026-09-20 23:44:55 | INFO | train rank=0 | step=38920/50000 progress=77.8% loss=0.101458 lr=1.283538e-05 lr_backbone=1.283538e-05 elapsed=17:06:55 avg_step=1.58s recent_step=2.01s eta=04:52:21 +2026-09-20 23:45:15 | INFO | train rank=0 | step=38930/50000 progress=77.9% loss=0.095051 lr=1.281327e-05 lr_backbone=1.281327e-05 elapsed=17:07:14 avg_step=1.58s recent_step=1.92s eta=04:52:06 +2026-09-20 23:45:34 | INFO | train rank=0 | step=38940/50000 progress=77.9% loss=0.105971 lr=1.279117e-05 lr_backbone=1.279117e-05 elapsed=17:07:33 avg_step=1.58s recent_step=1.89s eta=04:51:51 +2026-09-20 23:45:50 | INFO | train rank=0 | step=38950/50000 progress=77.9% loss=0.065514 lr=1.276909e-05 lr_backbone=1.276909e-05 elapsed=17:07:50 avg_step=1.58s recent_step=1.69s eta=04:51:35 +2026-09-20 23:46:04 | INFO | train rank=0 | step=38960/50000 progress=77.9% loss=0.071351 lr=1.274702e-05 lr_backbone=1.274702e-05 elapsed=17:08:03 avg_step=1.58s recent_step=1.34s eta=04:51:19 +2026-09-20 23:46:22 | INFO | train rank=0 | step=38970/50000 progress=77.9% loss=0.083336 lr=1.272498e-05 lr_backbone=1.272498e-05 elapsed=17:08:22 avg_step=1.58s recent_step=1.82s eta=04:51:04 +2026-09-20 23:46:37 | INFO | train rank=0 | step=38980/50000 progress=78.0% loss=0.075795 lr=1.270294e-05 lr_backbone=1.270294e-05 elapsed=17:08:37 avg_step=1.58s recent_step=1.53s eta=04:50:48 +2026-09-20 23:46:53 | INFO | train rank=0 | step=38990/50000 progress=78.0% loss=0.100306 lr=1.268093e-05 lr_backbone=1.268093e-05 elapsed=17:08:53 avg_step=1.58s recent_step=1.60s eta=04:50:32 +2026-09-20 23:47:09 | INFO | train rank=0 | step=39000/50000 progress=78.0% loss=0.082715 lr=1.265893e-05 lr_backbone=1.265893e-05 elapsed=17:09:08 avg_step=1.58s recent_step=1.52s eta=04:50:16 +2026-09-20 23:48:08 | INFO | train rank=0 | step=39010/50000 progress=78.0% loss=0.123757 lr=1.263694e-05 lr_backbone=1.263694e-05 elapsed=17:10:08 avg_step=1.58s recent_step=5.96s eta=04:50:12 +2026-09-20 23:48:24 | INFO | train rank=0 | step=39020/50000 progress=78.0% loss=0.080920 lr=1.261497e-05 lr_backbone=1.261497e-05 elapsed=17:10:23 avg_step=1.58s recent_step=1.54s eta=04:49:56 +2026-09-20 23:48:37 | INFO | train rank=0 | step=39030/50000 progress=78.1% loss=0.088092 lr=1.259302e-05 lr_backbone=1.259302e-05 elapsed=17:10:37 avg_step=1.58s recent_step=1.35s eta=04:49:40 +2026-09-20 23:48:54 | INFO | train rank=0 | step=39040/50000 progress=78.1% loss=0.076974 lr=1.257109e-05 lr_backbone=1.257109e-05 elapsed=17:10:53 avg_step=1.58s recent_step=1.68s eta=04:49:24 +2026-09-20 23:49:09 | INFO | train rank=0 | step=39050/50000 progress=78.1% loss=0.081204 lr=1.254917e-05 lr_backbone=1.254917e-05 elapsed=17:11:08 avg_step=1.58s recent_step=1.48s eta=04:49:08 +2026-09-20 23:49:23 | INFO | train rank=0 | step=39060/50000 progress=78.1% loss=0.079260 lr=1.252727e-05 lr_backbone=1.252727e-05 elapsed=17:11:23 avg_step=1.58s recent_step=1.43s eta=04:48:52 +2026-09-20 23:49:37 | INFO | train rank=0 | step=39070/50000 progress=78.1% loss=0.090868 lr=1.250538e-05 lr_backbone=1.250538e-05 elapsed=17:11:37 avg_step=1.58s recent_step=1.45s eta=04:48:36 +2026-09-20 23:49:55 | INFO | train rank=0 | step=39080/50000 progress=78.2% loss=0.103509 lr=1.248351e-05 lr_backbone=1.248351e-05 elapsed=17:11:55 avg_step=1.58s recent_step=1.75s eta=04:48:20 +2026-09-20 23:50:10 | INFO | train rank=0 | step=39090/50000 progress=78.2% loss=0.090036 lr=1.246166e-05 lr_backbone=1.246166e-05 elapsed=17:12:09 avg_step=1.58s recent_step=1.46s eta=04:48:04 +2026-09-20 23:50:25 | INFO | train rank=0 | step=39100/50000 progress=78.2% loss=0.078284 lr=1.243982e-05 lr_backbone=1.243982e-05 elapsed=17:12:25 avg_step=1.58s recent_step=1.57s eta=04:47:48 +2026-09-20 23:50:44 | INFO | train rank=0 | step=39110/50000 progress=78.2% loss=0.097161 lr=1.241800e-05 lr_backbone=1.241800e-05 elapsed=17:12:43 avg_step=1.58s recent_step=1.84s eta=04:47:33 +2026-09-20 23:51:02 | INFO | train rank=0 | step=39120/50000 progress=78.2% loss=0.125411 lr=1.239620e-05 lr_backbone=1.239620e-05 elapsed=17:13:02 avg_step=1.58s recent_step=1.83s eta=04:47:18 +2026-09-20 23:51:15 | INFO | train rank=0 | step=39130/50000 progress=78.3% loss=0.094998 lr=1.237441e-05 lr_backbone=1.237441e-05 elapsed=17:13:15 avg_step=1.58s recent_step=1.35s eta=04:47:01 +2026-09-20 23:51:27 | INFO | train rank=0 | step=39140/50000 progress=78.3% loss=0.089165 lr=1.235264e-05 lr_backbone=1.235264e-05 elapsed=17:13:27 avg_step=1.58s recent_step=1.19s eta=04:46:44 +2026-09-20 23:51:47 | INFO | train rank=0 | step=39150/50000 progress=78.3% loss=0.095118 lr=1.233089e-05 lr_backbone=1.233089e-05 elapsed=17:13:47 avg_step=1.58s recent_step=1.97s eta=04:46:30 +2026-09-20 23:52:00 | INFO | train rank=0 | step=39160/50000 progress=78.3% loss=0.137199 lr=1.230915e-05 lr_backbone=1.230915e-05 elapsed=17:13:59 avg_step=1.58s recent_step=1.25s eta=04:46:13 +2026-09-20 23:52:16 | INFO | train rank=0 | step=39170/50000 progress=78.3% loss=0.114557 lr=1.228743e-05 lr_backbone=1.228743e-05 elapsed=17:14:16 avg_step=1.58s recent_step=1.65s eta=04:45:57 +2026-09-20 23:52:29 | INFO | train rank=0 | step=39180/50000 progress=78.4% loss=0.094041 lr=1.226573e-05 lr_backbone=1.226573e-05 elapsed=17:14:29 avg_step=1.58s recent_step=1.29s eta=04:45:41 +2026-09-20 23:52:47 | INFO | train rank=0 | step=39190/50000 progress=78.4% loss=0.095765 lr=1.224404e-05 lr_backbone=1.224404e-05 elapsed=17:14:47 avg_step=1.58s recent_step=1.82s eta=04:45:25 +2026-09-20 23:53:03 | INFO | train rank=0 | step=39200/50000 progress=78.4% loss=0.103596 lr=1.222237e-05 lr_backbone=1.222237e-05 elapsed=17:15:03 avg_step=1.58s recent_step=1.57s eta=04:45:10 +2026-09-20 23:53:20 | INFO | train rank=0 | step=39210/50000 progress=78.4% loss=0.077143 lr=1.220071e-05 lr_backbone=1.220071e-05 elapsed=17:15:20 avg_step=1.58s recent_step=1.73s eta=04:44:54 +2026-09-20 23:53:32 | INFO | train rank=0 | step=39220/50000 progress=78.4% loss=0.084465 lr=1.217907e-05 lr_backbone=1.217907e-05 elapsed=17:15:32 avg_step=1.58s recent_step=1.20s eta=04:44:37 +2026-09-20 23:53:51 | INFO | train rank=0 | step=39230/50000 progress=78.5% loss=0.068939 lr=1.215745e-05 lr_backbone=1.215745e-05 elapsed=17:15:50 avg_step=1.58s recent_step=1.83s eta=04:44:22 +2026-09-20 23:54:08 | INFO | train rank=0 | step=39240/50000 progress=78.5% loss=0.117933 lr=1.213585e-05 lr_backbone=1.213585e-05 elapsed=17:16:08 avg_step=1.58s recent_step=1.75s eta=04:44:07 +2026-09-20 23:54:22 | INFO | train rank=0 | step=39250/50000 progress=78.5% loss=0.064958 lr=1.211426e-05 lr_backbone=1.211426e-05 elapsed=17:16:21 avg_step=1.58s recent_step=1.37s eta=04:43:50 +2026-09-20 23:54:40 | INFO | train rank=0 | step=39260/50000 progress=78.5% loss=0.092987 lr=1.209268e-05 lr_backbone=1.209268e-05 elapsed=17:16:40 avg_step=1.58s recent_step=1.86s eta=04:43:35 +2026-09-20 23:54:56 | INFO | train rank=0 | step=39270/50000 progress=78.5% loss=0.119403 lr=1.207113e-05 lr_backbone=1.207113e-05 elapsed=17:16:56 avg_step=1.58s recent_step=1.57s eta=04:43:19 +2026-09-20 23:55:13 | INFO | train rank=0 | step=39280/50000 progress=78.6% loss=0.070714 lr=1.204959e-05 lr_backbone=1.204959e-05 elapsed=17:17:13 avg_step=1.58s recent_step=1.68s eta=04:43:04 +2026-09-20 23:55:28 | INFO | train rank=0 | step=39290/50000 progress=78.6% loss=0.070803 lr=1.202807e-05 lr_backbone=1.202807e-05 elapsed=17:17:27 avg_step=1.58s recent_step=1.49s eta=04:42:48 +2026-09-20 23:55:41 | INFO | train rank=0 | step=39300/50000 progress=78.6% loss=0.083752 lr=1.200656e-05 lr_backbone=1.200656e-05 elapsed=17:17:41 avg_step=1.58s recent_step=1.32s eta=04:42:31 +2026-09-20 23:56:02 | INFO | train rank=0 | step=39310/50000 progress=78.6% loss=0.124327 lr=1.198507e-05 lr_backbone=1.198507e-05 elapsed=17:18:02 avg_step=1.58s recent_step=2.11s eta=04:42:17 +2026-09-20 23:56:20 | INFO | train rank=0 | step=39320/50000 progress=78.6% loss=0.093797 lr=1.196360e-05 lr_backbone=1.196360e-05 elapsed=17:18:19 avg_step=1.58s recent_step=1.76s eta=04:42:01 +2026-09-20 23:56:31 | INFO | train rank=0 | step=39330/50000 progress=78.7% loss=0.082455 lr=1.194214e-05 lr_backbone=1.194214e-05 elapsed=17:18:31 avg_step=1.58s recent_step=1.15s eta=04:41:44 +2026-09-20 23:56:43 | INFO | train rank=0 | step=39340/50000 progress=78.7% loss=0.121627 lr=1.192070e-05 lr_backbone=1.192070e-05 elapsed=17:18:43 avg_step=1.58s recent_step=1.21s eta=04:41:27 +2026-09-20 23:57:01 | INFO | train rank=0 | step=39350/50000 progress=78.7% loss=0.071459 lr=1.189928e-05 lr_backbone=1.189928e-05 elapsed=17:19:01 avg_step=1.58s recent_step=1.76s eta=04:41:12 +2026-09-20 23:57:21 | INFO | train rank=0 | step=39360/50000 progress=78.7% loss=0.113663 lr=1.187787e-05 lr_backbone=1.187787e-05 elapsed=17:19:21 avg_step=1.58s recent_step=2.02s eta=04:40:57 +2026-09-20 23:57:33 | INFO | train rank=0 | step=39370/50000 progress=78.7% loss=0.068165 lr=1.185649e-05 lr_backbone=1.185649e-05 elapsed=17:19:32 avg_step=1.58s recent_step=1.15s eta=04:40:40 +2026-09-20 23:57:45 | INFO | train rank=0 | step=39380/50000 progress=78.8% loss=0.068037 lr=1.183511e-05 lr_backbone=1.183511e-05 elapsed=17:19:44 avg_step=1.58s recent_step=1.20s eta=04:40:23 +2026-09-20 23:58:01 | INFO | train rank=0 | step=39390/50000 progress=78.8% loss=0.095512 lr=1.181376e-05 lr_backbone=1.181376e-05 elapsed=17:20:00 avg_step=1.58s recent_step=1.59s eta=04:40:08 +2026-09-20 23:58:20 | INFO | train rank=0 | step=39400/50000 progress=78.8% loss=0.094154 lr=1.179242e-05 lr_backbone=1.179242e-05 elapsed=17:20:19 avg_step=1.58s recent_step=1.90s eta=04:39:53 +2026-09-20 23:58:38 | INFO | train rank=0 | step=39410/50000 progress=78.8% loss=0.086761 lr=1.177109e-05 lr_backbone=1.177109e-05 elapsed=17:20:38 avg_step=1.58s recent_step=1.84s eta=04:39:37 +2026-09-20 23:58:51 | INFO | train rank=0 | step=39420/50000 progress=78.8% loss=0.073342 lr=1.174979e-05 lr_backbone=1.174979e-05 elapsed=17:20:51 avg_step=1.58s recent_step=1.33s eta=04:39:21 +2026-09-20 23:59:11 | INFO | train rank=0 | step=39430/50000 progress=78.9% loss=0.086604 lr=1.172850e-05 lr_backbone=1.172850e-05 elapsed=17:21:11 avg_step=1.58s recent_step=1.99s eta=04:39:06 +2026-09-20 23:59:24 | INFO | train rank=0 | step=39440/50000 progress=78.9% loss=0.098240 lr=1.170723e-05 lr_backbone=1.170723e-05 elapsed=17:21:24 avg_step=1.58s recent_step=1.29s eta=04:38:50 +2026-09-20 23:59:43 | INFO | train rank=0 | step=39450/50000 progress=78.9% loss=0.089747 lr=1.168597e-05 lr_backbone=1.168597e-05 elapsed=17:21:42 avg_step=1.58s recent_step=1.85s eta=04:38:34 +2026-09-20 23:59:55 | INFO | train rank=0 | step=39460/50000 progress=78.9% loss=0.084099 lr=1.166473e-05 lr_backbone=1.166473e-05 elapsed=17:21:55 avg_step=1.58s recent_step=1.24s eta=04:38:18 +2026-09-21 00:00:11 | INFO | train rank=0 | step=39470/50000 progress=78.9% loss=0.076675 lr=1.164351e-05 lr_backbone=1.164351e-05 elapsed=17:22:11 avg_step=1.58s recent_step=1.64s eta=04:38:02 +2026-09-21 00:00:27 | INFO | train rank=0 | step=39480/50000 progress=79.0% loss=0.091399 lr=1.162230e-05 lr_backbone=1.162230e-05 elapsed=17:22:26 avg_step=1.58s recent_step=1.52s eta=04:37:46 +2026-09-21 00:00:44 | INFO | train rank=0 | step=39490/50000 progress=79.0% loss=0.078630 lr=1.160112e-05 lr_backbone=1.160112e-05 elapsed=17:22:44 avg_step=1.58s recent_step=1.77s eta=04:37:31 +2026-09-21 00:00:57 | INFO | train rank=0 | step=39500/50000 progress=79.0% loss=0.073120 lr=1.157994e-05 lr_backbone=1.157994e-05 elapsed=17:22:56 avg_step=1.58s recent_step=1.21s eta=04:37:14 +2026-09-21 00:01:13 | INFO | train rank=0 | step=39510/50000 progress=79.0% loss=0.077257 lr=1.155879e-05 lr_backbone=1.155879e-05 elapsed=17:23:12 avg_step=1.58s recent_step=1.63s eta=04:36:58 +2026-09-21 00:01:31 | INFO | train rank=0 | step=39520/50000 progress=79.0% loss=0.089380 lr=1.153765e-05 lr_backbone=1.153765e-05 elapsed=17:23:31 avg_step=1.58s recent_step=1.84s eta=04:36:43 +2026-09-21 00:01:47 | INFO | train rank=0 | step=39530/50000 progress=79.1% loss=0.111992 lr=1.151653e-05 lr_backbone=1.151653e-05 elapsed=17:23:46 avg_step=1.58s recent_step=1.55s eta=04:36:27 +2026-09-21 00:01:59 | INFO | train rank=0 | step=39540/50000 progress=79.1% loss=0.076210 lr=1.149543e-05 lr_backbone=1.149543e-05 elapsed=17:23:58 avg_step=1.58s recent_step=1.19s eta=04:36:10 +2026-09-21 00:02:14 | INFO | train rank=0 | step=39550/50000 progress=79.1% loss=0.118967 lr=1.147434e-05 lr_backbone=1.147434e-05 elapsed=17:24:14 avg_step=1.58s recent_step=1.57s eta=04:35:54 +2026-09-21 00:02:28 | INFO | train rank=0 | step=39560/50000 progress=79.1% loss=0.088800 lr=1.145327e-05 lr_backbone=1.145327e-05 elapsed=17:24:28 avg_step=1.58s recent_step=1.42s eta=04:35:38 +2026-09-21 00:02:46 | INFO | train rank=0 | step=39570/50000 progress=79.1% loss=0.091221 lr=1.143221e-05 lr_backbone=1.143221e-05 elapsed=17:24:45 avg_step=1.58s recent_step=1.71s eta=04:35:22 +2026-09-21 00:03:01 | INFO | train rank=0 | step=39580/50000 progress=79.2% loss=0.074354 lr=1.141118e-05 lr_backbone=1.141118e-05 elapsed=17:25:00 avg_step=1.58s recent_step=1.51s eta=04:35:06 +2026-09-21 00:03:19 | INFO | train rank=0 | step=39590/50000 progress=79.2% loss=0.076260 lr=1.139016e-05 lr_backbone=1.139016e-05 elapsed=17:25:19 avg_step=1.58s recent_step=1.87s eta=04:34:51 +2026-09-21 00:03:33 | INFO | train rank=0 | step=39600/50000 progress=79.2% loss=0.095825 lr=1.136915e-05 lr_backbone=1.136915e-05 elapsed=17:25:32 avg_step=1.58s recent_step=1.32s eta=04:34:35 +2026-09-21 00:03:49 | INFO | train rank=0 | step=39610/50000 progress=79.2% loss=0.077533 lr=1.134817e-05 lr_backbone=1.134817e-05 elapsed=17:25:49 avg_step=1.58s recent_step=1.70s eta=04:34:19 +2026-09-21 00:04:01 | INFO | train rank=0 | step=39620/50000 progress=79.2% loss=0.105883 lr=1.132720e-05 lr_backbone=1.132720e-05 elapsed=17:26:01 avg_step=1.58s recent_step=1.19s eta=04:34:02 +2026-09-21 00:04:25 | INFO | train rank=0 | step=39630/50000 progress=79.3% loss=0.079019 lr=1.130624e-05 lr_backbone=1.130624e-05 elapsed=17:26:24 avg_step=1.58s recent_step=2.34s eta=04:33:48 +2026-09-21 00:04:37 | INFO | train rank=0 | step=39640/50000 progress=79.3% loss=0.081937 lr=1.128531e-05 lr_backbone=1.128531e-05 elapsed=17:26:37 avg_step=1.58s recent_step=1.24s eta=04:33:32 +2026-09-21 00:04:51 | INFO | train rank=0 | step=39650/50000 progress=79.3% loss=0.091066 lr=1.126439e-05 lr_backbone=1.126439e-05 elapsed=17:26:51 avg_step=1.58s recent_step=1.41s eta=04:33:15 +2026-09-21 00:05:09 | INFO | train rank=0 | step=39660/50000 progress=79.3% loss=0.079677 lr=1.124349e-05 lr_backbone=1.124349e-05 elapsed=17:27:09 avg_step=1.58s recent_step=1.76s eta=04:33:00 +2026-09-21 00:05:27 | INFO | train rank=0 | step=39670/50000 progress=79.3% loss=0.069987 lr=1.122260e-05 lr_backbone=1.122260e-05 elapsed=17:27:26 avg_step=1.58s recent_step=1.80s eta=04:32:45 +2026-09-21 00:05:43 | INFO | train rank=0 | step=39680/50000 progress=79.4% loss=0.079904 lr=1.120174e-05 lr_backbone=1.120174e-05 elapsed=17:27:43 avg_step=1.58s recent_step=1.62s eta=04:32:29 +2026-09-21 00:05:59 | INFO | train rank=0 | step=39690/50000 progress=79.4% loss=0.102305 lr=1.118088e-05 lr_backbone=1.118088e-05 elapsed=17:27:58 avg_step=1.58s recent_step=1.57s eta=04:32:13 +2026-09-21 00:06:14 | INFO | train rank=0 | step=39700/50000 progress=79.4% loss=0.067889 lr=1.116005e-05 lr_backbone=1.116005e-05 elapsed=17:28:14 avg_step=1.58s recent_step=1.51s eta=04:31:57 +2026-09-21 00:06:31 | INFO | train rank=0 | step=39710/50000 progress=79.4% loss=0.079022 lr=1.113923e-05 lr_backbone=1.113923e-05 elapsed=17:28:30 avg_step=1.58s recent_step=1.69s eta=04:31:42 +2026-09-21 00:06:50 | INFO | train rank=0 | step=39720/50000 progress=79.4% loss=0.082078 lr=1.111843e-05 lr_backbone=1.111843e-05 elapsed=17:28:50 avg_step=1.58s recent_step=1.93s eta=04:31:27 +2026-09-21 00:07:02 | INFO | train rank=0 | step=39730/50000 progress=79.5% loss=0.063476 lr=1.109765e-05 lr_backbone=1.109765e-05 elapsed=17:29:01 avg_step=1.58s recent_step=1.17s eta=04:31:10 +2026-09-21 00:07:14 | INFO | train rank=0 | step=39740/50000 progress=79.5% loss=0.092727 lr=1.107688e-05 lr_backbone=1.107688e-05 elapsed=17:29:13 avg_step=1.58s recent_step=1.20s eta=04:30:53 +2026-09-21 00:07:35 | INFO | train rank=0 | step=39750/50000 progress=79.5% loss=0.109541 lr=1.105614e-05 lr_backbone=1.105614e-05 elapsed=17:29:34 avg_step=1.58s recent_step=2.08s eta=04:30:38 +2026-09-21 00:07:46 | INFO | train rank=0 | step=39760/50000 progress=79.5% loss=0.140188 lr=1.103540e-05 lr_backbone=1.103540e-05 elapsed=17:29:46 avg_step=1.58s recent_step=1.19s eta=04:30:21 +2026-09-21 00:07:58 | INFO | train rank=0 | step=39770/50000 progress=79.5% loss=0.075424 lr=1.101469e-05 lr_backbone=1.101469e-05 elapsed=17:29:58 avg_step=1.58s recent_step=1.17s eta=04:30:04 +2026-09-21 00:08:11 | INFO | train rank=0 | step=39780/50000 progress=79.6% loss=0.074363 lr=1.099399e-05 lr_backbone=1.099399e-05 elapsed=17:30:10 avg_step=1.58s recent_step=1.24s eta=04:29:48 +2026-09-21 00:08:35 | INFO | train rank=0 | step=39790/50000 progress=79.6% loss=0.083203 lr=1.097331e-05 lr_backbone=1.097331e-05 elapsed=17:30:35 avg_step=1.58s recent_step=2.50s eta=04:29:34 +2026-09-21 00:08:48 | INFO | train rank=0 | step=39800/50000 progress=79.6% loss=0.108920 lr=1.095265e-05 lr_backbone=1.095265e-05 elapsed=17:30:48 avg_step=1.58s recent_step=1.24s eta=04:29:18 +2026-09-21 00:09:01 | INFO | train rank=0 | step=39810/50000 progress=79.6% loss=0.129951 lr=1.093200e-05 lr_backbone=1.093200e-05 elapsed=17:31:00 avg_step=1.58s recent_step=1.29s eta=04:29:01 +2026-09-21 00:09:20 | INFO | train rank=0 | step=39820/50000 progress=79.6% loss=0.077841 lr=1.091137e-05 lr_backbone=1.091137e-05 elapsed=17:31:20 avg_step=1.58s recent_step=1.95s eta=04:28:46 +2026-09-21 00:09:34 | INFO | train rank=0 | step=39830/50000 progress=79.7% loss=0.078941 lr=1.089076e-05 lr_backbone=1.089076e-05 elapsed=17:31:34 avg_step=1.58s recent_step=1.36s eta=04:28:30 +2026-09-21 00:09:50 | INFO | train rank=0 | step=39840/50000 progress=79.7% loss=0.079764 lr=1.087016e-05 lr_backbone=1.087016e-05 elapsed=17:31:50 avg_step=1.58s recent_step=1.60s eta=04:28:14 +2026-09-21 00:10:07 | INFO | train rank=0 | step=39850/50000 progress=79.7% loss=0.085362 lr=1.084959e-05 lr_backbone=1.084959e-05 elapsed=17:32:06 avg_step=1.58s recent_step=1.69s eta=04:27:58 +2026-09-21 00:10:19 | INFO | train rank=0 | step=39860/50000 progress=79.7% loss=0.080926 lr=1.082902e-05 lr_backbone=1.082902e-05 elapsed=17:32:19 avg_step=1.58s recent_step=1.26s eta=04:27:42 +2026-09-21 00:10:40 | INFO | train rank=0 | step=39870/50000 progress=79.7% loss=0.110264 lr=1.080848e-05 lr_backbone=1.080848e-05 elapsed=17:32:40 avg_step=1.58s recent_step=2.09s eta=04:27:27 +2026-09-21 00:10:52 | INFO | train rank=0 | step=39880/50000 progress=79.8% loss=0.075053 lr=1.078795e-05 lr_backbone=1.078795e-05 elapsed=17:32:52 avg_step=1.58s recent_step=1.20s eta=04:27:10 +2026-09-21 00:11:08 | INFO | train rank=0 | step=39890/50000 progress=79.8% loss=0.078727 lr=1.076744e-05 lr_backbone=1.076744e-05 elapsed=17:33:08 avg_step=1.58s recent_step=1.57s eta=04:26:54 +2026-09-21 00:11:24 | INFO | train rank=0 | step=39900/50000 progress=79.8% loss=0.083391 lr=1.074695e-05 lr_backbone=1.074695e-05 elapsed=17:33:24 avg_step=1.58s recent_step=1.58s eta=04:26:39 +2026-09-21 00:11:41 | INFO | train rank=0 | step=39910/50000 progress=79.8% loss=0.083768 lr=1.072648e-05 lr_backbone=1.072648e-05 elapsed=17:33:41 avg_step=1.58s recent_step=1.75s eta=04:26:23 +2026-09-21 00:11:54 | INFO | train rank=0 | step=39920/50000 progress=79.8% loss=0.062817 lr=1.070602e-05 lr_backbone=1.070602e-05 elapsed=17:33:54 avg_step=1.58s recent_step=1.27s eta=04:26:06 +2026-09-21 00:12:11 | INFO | train rank=0 | step=39930/50000 progress=79.9% loss=0.075464 lr=1.068558e-05 lr_backbone=1.068558e-05 elapsed=17:34:11 avg_step=1.58s recent_step=1.69s eta=04:25:51 +2026-09-21 00:12:24 | INFO | train rank=0 | step=39940/50000 progress=79.9% loss=0.101690 lr=1.066515e-05 lr_backbone=1.066515e-05 elapsed=17:34:24 avg_step=1.58s recent_step=1.33s eta=04:25:34 +2026-09-21 00:12:45 | INFO | train rank=0 | step=39950/50000 progress=79.9% loss=0.092398 lr=1.064475e-05 lr_backbone=1.064475e-05 elapsed=17:34:44 avg_step=1.58s recent_step=2.03s eta=04:25:20 +2026-09-21 00:12:58 | INFO | train rank=0 | step=39960/50000 progress=79.9% loss=0.083735 lr=1.062436e-05 lr_backbone=1.062436e-05 elapsed=17:34:58 avg_step=1.58s recent_step=1.35s eta=04:25:03 +2026-09-21 00:13:11 | INFO | train rank=0 | step=39970/50000 progress=79.9% loss=0.078354 lr=1.060399e-05 lr_backbone=1.060399e-05 elapsed=17:35:11 avg_step=1.58s recent_step=1.31s eta=04:24:47 +2026-09-21 00:13:30 | INFO | train rank=0 | step=39980/50000 progress=80.0% loss=0.091366 lr=1.058363e-05 lr_backbone=1.058363e-05 elapsed=17:35:29 avg_step=1.58s recent_step=1.83s eta=04:24:32 +2026-09-21 00:13:48 | INFO | train rank=0 | step=39990/50000 progress=80.0% loss=0.079112 lr=1.056329e-05 lr_backbone=1.056329e-05 elapsed=17:35:48 avg_step=1.58s recent_step=1.88s eta=04:24:16 +2026-09-21 00:14:01 | INFO | train rank=0 | step=40000/50000 progress=80.0% loss=0.078597 lr=1.054297e-05 lr_backbone=1.054297e-05 elapsed=17:36:01 avg_step=1.58s recent_step=1.30s eta=04:24:00 +2026-09-21 00:14:17 | INFO | train rank=0 | step=40010/50000 progress=80.0% loss=0.077796 lr=1.052267e-05 lr_backbone=1.052267e-05 elapsed=17:36:16 avg_step=1.58s recent_step=1.52s eta=04:23:44 +2026-09-21 00:14:34 | INFO | train rank=0 | step=40020/50000 progress=80.0% loss=0.095193 lr=1.050239e-05 lr_backbone=1.050239e-05 elapsed=17:36:34 avg_step=1.58s recent_step=1.77s eta=04:23:28 +2026-09-21 00:14:53 | INFO | train rank=0 | step=40030/50000 progress=80.1% loss=0.100468 lr=1.048212e-05 lr_backbone=1.048212e-05 elapsed=17:36:52 avg_step=1.58s recent_step=1.85s eta=04:23:13 +2026-09-21 00:15:04 | INFO | train rank=0 | step=40040/50000 progress=80.1% loss=0.077572 lr=1.046187e-05 lr_backbone=1.046187e-05 elapsed=17:37:04 avg_step=1.58s recent_step=1.17s eta=04:22:56 +2026-09-21 00:15:17 | INFO | train rank=0 | step=40050/50000 progress=80.1% loss=0.069644 lr=1.044163e-05 lr_backbone=1.044163e-05 elapsed=17:37:16 avg_step=1.58s recent_step=1.23s eta=04:22:40 +2026-09-21 00:15:33 | INFO | train rank=0 | step=40060/50000 progress=80.1% loss=0.139573 lr=1.042142e-05 lr_backbone=1.042142e-05 elapsed=17:37:33 avg_step=1.58s recent_step=1.64s eta=04:22:24 +2026-09-21 00:15:50 | INFO | train rank=0 | step=40070/50000 progress=80.1% loss=0.088727 lr=1.040122e-05 lr_backbone=1.040122e-05 elapsed=17:37:49 avg_step=1.58s recent_step=1.66s eta=04:22:08 +2026-09-21 00:16:04 | INFO | train rank=0 | step=40080/50000 progress=80.2% loss=0.088888 lr=1.038103e-05 lr_backbone=1.038103e-05 elapsed=17:38:04 avg_step=1.58s recent_step=1.45s eta=04:21:52 +2026-09-21 00:16:23 | INFO | train rank=0 | step=40090/50000 progress=80.2% loss=0.079044 lr=1.036087e-05 lr_backbone=1.036087e-05 elapsed=17:38:23 avg_step=1.58s recent_step=1.89s eta=04:21:37 +2026-09-21 00:16:34 | INFO | train rank=0 | step=40100/50000 progress=80.2% loss=0.112348 lr=1.034072e-05 lr_backbone=1.034072e-05 elapsed=17:38:34 avg_step=1.58s recent_step=1.13s eta=04:21:20 +2026-09-21 00:16:57 | INFO | train rank=0 | step=40110/50000 progress=80.2% loss=0.070011 lr=1.032059e-05 lr_backbone=1.032059e-05 elapsed=17:38:56 avg_step=1.58s recent_step=2.21s eta=04:21:06 +2026-09-21 00:17:08 | INFO | train rank=0 | step=40120/50000 progress=80.2% loss=0.091341 lr=1.030048e-05 lr_backbone=1.030048e-05 elapsed=17:39:08 avg_step=1.58s recent_step=1.14s eta=04:20:49 +2026-09-21 00:17:25 | INFO | train rank=0 | step=40130/50000 progress=80.3% loss=0.077895 lr=1.028038e-05 lr_backbone=1.028038e-05 elapsed=17:39:24 avg_step=1.58s recent_step=1.69s eta=04:20:33 +2026-09-21 00:17:45 | INFO | train rank=0 | step=40140/50000 progress=80.3% loss=0.091306 lr=1.026031e-05 lr_backbone=1.026031e-05 elapsed=17:39:45 avg_step=1.58s recent_step=2.04s eta=04:20:19 +2026-09-21 00:17:57 | INFO | train rank=0 | step=40150/50000 progress=80.3% loss=0.084578 lr=1.024025e-05 lr_backbone=1.024025e-05 elapsed=17:39:56 avg_step=1.58s recent_step=1.13s eta=04:20:02 +2026-09-21 00:18:09 | INFO | train rank=0 | step=40160/50000 progress=80.3% loss=0.087553 lr=1.022020e-05 lr_backbone=1.022020e-05 elapsed=17:40:08 avg_step=1.58s recent_step=1.20s eta=04:19:45 +2026-09-21 00:18:21 | INFO | train rank=0 | step=40170/50000 progress=80.3% loss=0.123288 lr=1.020018e-05 lr_backbone=1.020018e-05 elapsed=17:40:20 avg_step=1.58s recent_step=1.22s eta=04:19:28 +2026-09-21 00:18:41 | INFO | train rank=0 | step=40180/50000 progress=80.4% loss=0.108594 lr=1.018017e-05 lr_backbone=1.018017e-05 elapsed=17:40:40 avg_step=1.58s recent_step=2.01s eta=04:19:13 +2026-09-21 00:19:00 | INFO | train rank=0 | step=40190/50000 progress=80.4% loss=0.126312 lr=1.016018e-05 lr_backbone=1.016018e-05 elapsed=17:40:59 avg_step=1.58s recent_step=1.90s eta=04:18:58 +2026-09-21 00:19:13 | INFO | train rank=0 | step=40200/50000 progress=80.4% loss=0.098229 lr=1.014021e-05 lr_backbone=1.014021e-05 elapsed=17:41:13 avg_step=1.58s recent_step=1.33s eta=04:18:42 +2026-09-21 00:19:29 | INFO | train rank=0 | step=40210/50000 progress=80.4% loss=0.095725 lr=1.012025e-05 lr_backbone=1.012025e-05 elapsed=17:41:28 avg_step=1.58s recent_step=1.56s eta=04:18:26 +2026-09-21 00:19:45 | INFO | train rank=0 | step=40220/50000 progress=80.4% loss=0.074894 lr=1.010031e-05 lr_backbone=1.010031e-05 elapsed=17:41:44 avg_step=1.58s recent_step=1.60s eta=04:18:10 +2026-09-21 00:20:06 | INFO | train rank=0 | step=40230/50000 progress=80.5% loss=0.076115 lr=1.008039e-05 lr_backbone=1.008039e-05 elapsed=17:42:05 avg_step=1.58s recent_step=2.11s eta=04:17:56 +2026-09-21 00:20:17 | INFO | train rank=0 | step=40240/50000 progress=80.5% loss=0.101263 lr=1.006049e-05 lr_backbone=1.006049e-05 elapsed=17:42:17 avg_step=1.58s recent_step=1.15s eta=04:17:39 +2026-09-21 00:20:30 | INFO | train rank=0 | step=40250/50000 progress=80.5% loss=0.077534 lr=1.004060e-05 lr_backbone=1.004060e-05 elapsed=17:42:29 avg_step=1.58s recent_step=1.22s eta=04:17:22 +2026-09-21 00:20:48 | INFO | train rank=0 | step=40260/50000 progress=80.5% loss=0.091851 lr=1.002073e-05 lr_backbone=1.002073e-05 elapsed=17:42:47 avg_step=1.58s recent_step=1.80s eta=04:17:07 +2026-09-21 00:21:09 | INFO | train rank=0 | step=40270/50000 progress=80.5% loss=0.085105 lr=1.000088e-05 lr_backbone=1.000088e-05 elapsed=17:43:09 avg_step=1.58s recent_step=2.18s eta=04:16:52 +2026-09-21 00:21:25 | INFO | train rank=0 | step=40280/50000 progress=80.6% loss=0.086935 lr=9.981047e-06 lr_backbone=9.981047e-06 elapsed=17:43:25 avg_step=1.58s recent_step=1.57s eta=04:16:36 +2026-09-21 00:21:38 | INFO | train rank=0 | step=40290/50000 progress=80.6% loss=0.081646 lr=9.961231e-06 lr_backbone=9.961231e-06 elapsed=17:43:37 avg_step=1.58s recent_step=1.28s eta=04:16:20 +2026-09-21 00:21:53 | INFO | train rank=0 | step=40300/50000 progress=80.6% loss=0.088404 lr=9.941432e-06 lr_backbone=9.941432e-06 elapsed=17:43:53 avg_step=1.58s recent_step=1.54s eta=04:16:04 +2026-09-21 00:22:10 | INFO | train rank=0 | step=40310/50000 progress=80.6% loss=0.072349 lr=9.921651e-06 lr_backbone=9.921651e-06 elapsed=17:44:09 avg_step=1.58s recent_step=1.66s eta=04:15:48 +2026-09-21 00:22:23 | INFO | train rank=0 | step=40320/50000 progress=80.6% loss=0.078668 lr=9.901887e-06 lr_backbone=9.901887e-06 elapsed=17:44:22 avg_step=1.58s recent_step=1.27s eta=04:15:32 +2026-09-21 00:22:40 | INFO | train rank=0 | step=40330/50000 progress=80.7% loss=0.076620 lr=9.882141e-06 lr_backbone=9.882141e-06 elapsed=17:44:40 avg_step=1.58s recent_step=1.75s eta=04:15:16 +2026-09-21 00:22:55 | INFO | train rank=0 | step=40340/50000 progress=80.7% loss=0.078005 lr=9.862413e-06 lr_backbone=9.862413e-06 elapsed=17:44:54 avg_step=1.58s recent_step=1.48s eta=04:15:00 +2026-09-21 00:23:14 | INFO | train rank=0 | step=40350/50000 progress=80.7% loss=0.076365 lr=9.842702e-06 lr_backbone=9.842702e-06 elapsed=17:45:14 avg_step=1.58s recent_step=1.91s eta=04:14:45 +2026-09-21 00:23:31 | INFO | train rank=0 | step=40360/50000 progress=80.7% loss=0.094453 lr=9.823008e-06 lr_backbone=9.823008e-06 elapsed=17:45:31 avg_step=1.58s recent_step=1.74s eta=04:14:30 +2026-09-21 00:23:46 | INFO | train rank=0 | step=40370/50000 progress=80.7% loss=0.110937 lr=9.803332e-06 lr_backbone=9.803332e-06 elapsed=17:45:46 avg_step=1.58s recent_step=1.48s eta=04:14:13 +2026-09-21 00:24:02 | INFO | train rank=0 | step=40380/50000 progress=80.8% loss=0.065613 lr=9.783674e-06 lr_backbone=9.783674e-06 elapsed=17:46:02 avg_step=1.58s recent_step=1.61s eta=04:13:58 +2026-09-21 00:24:19 | INFO | train rank=0 | step=40390/50000 progress=80.8% loss=0.098208 lr=9.764034e-06 lr_backbone=9.764034e-06 elapsed=17:46:19 avg_step=1.58s recent_step=1.67s eta=04:13:42 +2026-09-21 00:24:36 | INFO | train rank=0 | step=40400/50000 progress=80.8% loss=0.081381 lr=9.744411e-06 lr_backbone=9.744411e-06 elapsed=17:46:36 avg_step=1.58s recent_step=1.70s eta=04:13:26 +2026-09-21 00:24:55 | INFO | train rank=0 | step=40410/50000 progress=80.8% loss=0.076834 lr=9.724805e-06 lr_backbone=9.724805e-06 elapsed=17:46:54 avg_step=1.58s recent_step=1.87s eta=04:13:11 +2026-09-21 00:25:07 | INFO | train rank=0 | step=40420/50000 progress=80.8% loss=0.073874 lr=9.705217e-06 lr_backbone=9.705217e-06 elapsed=17:47:06 avg_step=1.58s recent_step=1.21s eta=04:12:55 +2026-09-21 00:25:26 | INFO | train rank=0 | step=40430/50000 progress=80.9% loss=0.089404 lr=9.685647e-06 lr_backbone=9.685647e-06 elapsed=17:47:25 avg_step=1.58s recent_step=1.88s eta=04:12:39 +2026-09-21 00:25:43 | INFO | train rank=0 | step=40440/50000 progress=80.9% loss=0.076441 lr=9.666095e-06 lr_backbone=9.666095e-06 elapsed=17:47:42 avg_step=1.58s recent_step=1.70s eta=04:12:24 +2026-09-21 00:25:57 | INFO | train rank=0 | step=40450/50000 progress=80.9% loss=0.069407 lr=9.646560e-06 lr_backbone=9.646560e-06 elapsed=17:47:56 avg_step=1.58s recent_step=1.41s eta=04:12:08 +2026-09-21 00:26:08 | INFO | train rank=0 | step=40460/50000 progress=80.9% loss=0.077171 lr=9.627043e-06 lr_backbone=9.627043e-06 elapsed=17:48:08 avg_step=1.58s recent_step=1.17s eta=04:11:51 +2026-09-21 00:26:27 | INFO | train rank=0 | step=40470/50000 progress=80.9% loss=0.082005 lr=9.607543e-06 lr_backbone=9.607543e-06 elapsed=17:48:27 avg_step=1.58s recent_step=1.91s eta=04:11:36 +2026-09-21 00:26:43 | INFO | train rank=0 | step=40480/50000 progress=81.0% loss=0.067377 lr=9.588061e-06 lr_backbone=9.588061e-06 elapsed=17:48:42 avg_step=1.58s recent_step=1.54s eta=04:11:20 +2026-09-21 00:26:59 | INFO | train rank=0 | step=40490/50000 progress=81.0% loss=0.097427 lr=9.568597e-06 lr_backbone=9.568597e-06 elapsed=17:48:59 avg_step=1.58s recent_step=1.61s eta=04:11:04 +2026-09-21 00:27:13 | INFO | train rank=0 | step=40500/50000 progress=81.0% loss=0.092333 lr=9.549150e-06 lr_backbone=9.549150e-06 elapsed=17:49:13 avg_step=1.58s recent_step=1.42s eta=04:10:48 +2026-09-21 00:27:35 | INFO | train rank=0 | step=40510/50000 progress=81.0% loss=0.101674 lr=9.529721e-06 lr_backbone=9.529721e-06 elapsed=17:49:34 avg_step=1.58s recent_step=2.15s eta=04:10:33 +2026-09-21 00:27:46 | INFO | train rank=0 | step=40520/50000 progress=81.0% loss=0.083672 lr=9.510310e-06 lr_backbone=9.510310e-06 elapsed=17:49:46 avg_step=1.58s recent_step=1.13s eta=04:10:16 +2026-09-21 00:27:58 | INFO | train rank=0 | step=40530/50000 progress=81.1% loss=0.104910 lr=9.490917e-06 lr_backbone=9.490917e-06 elapsed=17:49:58 avg_step=1.58s recent_step=1.19s eta=04:10:00 +2026-09-21 00:28:16 | INFO | train rank=0 | step=40540/50000 progress=81.1% loss=0.060523 lr=9.471541e-06 lr_backbone=9.471541e-06 elapsed=17:50:16 avg_step=1.58s recent_step=1.80s eta=04:09:44 +2026-09-21 00:28:30 | INFO | train rank=0 | step=40550/50000 progress=81.1% loss=0.065248 lr=9.452183e-06 lr_backbone=9.452183e-06 elapsed=17:50:30 avg_step=1.58s recent_step=1.44s eta=04:09:28 +2026-09-21 00:28:50 | INFO | train rank=0 | step=40560/50000 progress=81.1% loss=0.116993 lr=9.432843e-06 lr_backbone=9.432843e-06 elapsed=17:50:50 avg_step=1.58s recent_step=2.02s eta=04:09:13 +2026-09-21 00:29:03 | INFO | train rank=0 | step=40570/50000 progress=81.1% loss=0.082115 lr=9.413520e-06 lr_backbone=9.413520e-06 elapsed=17:51:02 avg_step=1.58s recent_step=1.22s eta=04:08:57 +2026-09-21 00:29:20 | INFO | train rank=0 | step=40580/50000 progress=81.2% loss=0.075608 lr=9.394216e-06 lr_backbone=9.394216e-06 elapsed=17:51:20 avg_step=1.58s recent_step=1.75s eta=04:08:41 +2026-09-21 00:29:32 | INFO | train rank=0 | step=40590/50000 progress=81.2% loss=0.082432 lr=9.374929e-06 lr_backbone=9.374929e-06 elapsed=17:51:31 avg_step=1.58s recent_step=1.15s eta=04:08:24 +2026-09-21 00:29:50 | INFO | train rank=0 | step=40600/50000 progress=81.2% loss=0.098253 lr=9.355659e-06 lr_backbone=9.355659e-06 elapsed=17:51:50 avg_step=1.58s recent_step=1.85s eta=04:08:09 +2026-09-21 00:30:08 | INFO | train rank=0 | step=40610/50000 progress=81.2% loss=0.090815 lr=9.336408e-06 lr_backbone=9.336408e-06 elapsed=17:52:08 avg_step=1.58s recent_step=1.80s eta=04:07:54 +2026-09-21 00:30:20 | INFO | train rank=0 | step=40620/50000 progress=81.2% loss=0.075721 lr=9.317174e-06 lr_backbone=9.317174e-06 elapsed=17:52:20 avg_step=1.58s recent_step=1.18s eta=04:07:37 +2026-09-21 00:30:39 | INFO | train rank=0 | step=40630/50000 progress=81.3% loss=0.078448 lr=9.297958e-06 lr_backbone=9.297958e-06 elapsed=17:52:39 avg_step=1.58s recent_step=1.89s eta=04:07:22 +2026-09-21 00:30:55 | INFO | train rank=0 | step=40640/50000 progress=81.3% loss=0.094808 lr=9.278760e-06 lr_backbone=9.278760e-06 elapsed=17:52:54 avg_step=1.58s recent_step=1.57s eta=04:07:06 +2026-09-21 00:31:07 | INFO | train rank=0 | step=40650/50000 progress=81.3% loss=0.072166 lr=9.259580e-06 lr_backbone=9.259580e-06 elapsed=17:53:06 avg_step=1.58s recent_step=1.21s eta=04:06:49 +2026-09-21 00:31:19 | INFO | train rank=0 | step=40660/50000 progress=81.3% loss=0.099214 lr=9.240418e-06 lr_backbone=9.240418e-06 elapsed=17:53:18 avg_step=1.58s recent_step=1.18s eta=04:06:33 +2026-09-21 00:31:36 | INFO | train rank=0 | step=40670/50000 progress=81.3% loss=0.075313 lr=9.221273e-06 lr_backbone=9.221273e-06 elapsed=17:53:36 avg_step=1.58s recent_step=1.78s eta=04:06:17 +2026-09-21 00:31:54 | INFO | train rank=0 | step=40680/50000 progress=81.4% loss=0.069937 lr=9.202146e-06 lr_backbone=9.202146e-06 elapsed=17:53:54 avg_step=1.58s recent_step=1.82s eta=04:06:02 +2026-09-21 00:32:07 | INFO | train rank=0 | step=40690/50000 progress=81.4% loss=0.066837 lr=9.183037e-06 lr_backbone=9.183037e-06 elapsed=17:54:06 avg_step=1.58s recent_step=1.23s eta=04:05:45 +2026-09-21 00:32:24 | INFO | train rank=0 | step=40700/50000 progress=81.4% loss=0.094393 lr=9.163946e-06 lr_backbone=9.163946e-06 elapsed=17:54:24 avg_step=1.58s recent_step=1.77s eta=04:05:30 +2026-09-21 00:32:43 | INFO | train rank=0 | step=40710/50000 progress=81.4% loss=0.125363 lr=9.144873e-06 lr_backbone=9.144873e-06 elapsed=17:54:42 avg_step=1.58s recent_step=1.85s eta=04:05:14 +2026-09-21 00:32:57 | INFO | train rank=0 | step=40720/50000 progress=81.4% loss=0.091007 lr=9.125818e-06 lr_backbone=9.125818e-06 elapsed=17:54:57 avg_step=1.58s recent_step=1.45s eta=04:04:58 +2026-09-21 00:33:18 | INFO | train rank=0 | step=40730/50000 progress=81.5% loss=0.163338 lr=9.106780e-06 lr_backbone=9.106780e-06 elapsed=17:55:17 avg_step=1.58s recent_step=2.05s eta=04:04:44 +2026-09-21 00:33:35 | INFO | train rank=0 | step=40740/50000 progress=81.5% loss=0.078621 lr=9.087761e-06 lr_backbone=9.087761e-06 elapsed=17:55:34 avg_step=1.58s recent_step=1.70s eta=04:04:28 +2026-09-21 00:33:48 | INFO | train rank=0 | step=40750/50000 progress=81.5% loss=0.079390 lr=9.068759e-06 lr_backbone=9.068759e-06 elapsed=17:55:48 avg_step=1.58s recent_step=1.33s eta=04:04:12 +2026-09-21 00:34:45 | INFO | train rank=0 | step=40760/50000 progress=81.5% loss=0.080655 lr=9.049776e-06 lr_backbone=9.049776e-06 elapsed=17:56:44 avg_step=1.59s recent_step=5.66s eta=04:04:05 +2026-09-21 00:35:00 | INFO | train rank=0 | step=40770/50000 progress=81.5% loss=0.093682 lr=9.030810e-06 lr_backbone=9.030810e-06 elapsed=17:56:59 avg_step=1.58s recent_step=1.50s eta=04:03:49 +2026-09-21 00:35:13 | INFO | train rank=0 | step=40780/50000 progress=81.6% loss=0.072339 lr=9.011862e-06 lr_backbone=9.011862e-06 elapsed=17:57:12 avg_step=1.58s recent_step=1.29s eta=04:03:32 +2026-09-21 00:35:32 | INFO | train rank=0 | step=40790/50000 progress=81.6% loss=0.075496 lr=8.992932e-06 lr_backbone=8.992932e-06 elapsed=17:57:32 avg_step=1.58s recent_step=1.94s eta=04:03:17 +2026-09-21 00:35:43 | INFO | train rank=0 | step=40800/50000 progress=81.6% loss=0.094988 lr=8.974020e-06 lr_backbone=8.974020e-06 elapsed=17:57:43 avg_step=1.58s recent_step=1.15s eta=04:03:00 +2026-09-21 00:36:00 | INFO | train rank=0 | step=40810/50000 progress=81.6% loss=0.082163 lr=8.955126e-06 lr_backbone=8.955126e-06 elapsed=17:58:00 avg_step=1.58s recent_step=1.65s eta=04:02:45 +2026-09-21 00:36:14 | INFO | train rank=0 | step=40820/50000 progress=81.6% loss=0.081432 lr=8.936249e-06 lr_backbone=8.936249e-06 elapsed=17:58:14 avg_step=1.58s recent_step=1.42s eta=04:02:29 +2026-09-21 00:36:29 | INFO | train rank=0 | step=40830/50000 progress=81.7% loss=0.113221 lr=8.917391e-06 lr_backbone=8.917391e-06 elapsed=17:58:29 avg_step=1.58s recent_step=1.50s eta=04:02:13 +2026-09-21 00:36:46 | INFO | train rank=0 | step=40840/50000 progress=81.7% loss=0.074954 lr=8.898551e-06 lr_backbone=8.898551e-06 elapsed=17:58:45 avg_step=1.58s recent_step=1.65s eta=04:01:57 +2026-09-21 00:37:04 | INFO | train rank=0 | step=40850/50000 progress=81.7% loss=0.080622 lr=8.879729e-06 lr_backbone=8.879729e-06 elapsed=17:59:04 avg_step=1.58s recent_step=1.87s eta=04:01:42 +2026-09-21 00:37:20 | INFO | train rank=0 | step=40860/50000 progress=81.7% loss=0.078577 lr=8.860925e-06 lr_backbone=8.860925e-06 elapsed=17:59:19 avg_step=1.58s recent_step=1.52s eta=04:01:26 +2026-09-21 00:37:35 | INFO | train rank=0 | step=40870/50000 progress=81.7% loss=0.088498 lr=8.842138e-06 lr_backbone=8.842138e-06 elapsed=17:59:35 avg_step=1.58s recent_step=1.59s eta=04:01:10 +2026-09-21 00:37:50 | INFO | train rank=0 | step=40880/50000 progress=81.8% loss=0.096028 lr=8.823370e-06 lr_backbone=8.823370e-06 elapsed=17:59:50 avg_step=1.58s recent_step=1.50s eta=04:00:54 +2026-09-21 00:38:06 | INFO | train rank=0 | step=40890/50000 progress=81.8% loss=0.084458 lr=8.804620e-06 lr_backbone=8.804620e-06 elapsed=18:00:06 avg_step=1.58s recent_step=1.58s eta=04:00:38 +2026-09-21 00:38:23 | INFO | train rank=0 | step=40900/50000 progress=81.8% loss=0.088433 lr=8.785888e-06 lr_backbone=8.785888e-06 elapsed=18:00:22 avg_step=1.58s recent_step=1.66s eta=04:00:22 +2026-09-21 00:38:35 | INFO | train rank=0 | step=40910/50000 progress=81.8% loss=0.092379 lr=8.767173e-06 lr_backbone=8.767173e-06 elapsed=18:00:35 avg_step=1.58s recent_step=1.25s eta=04:00:06 +2026-09-21 00:38:49 | INFO | train rank=0 | step=40920/50000 progress=81.8% loss=0.078423 lr=8.748477e-06 lr_backbone=8.748477e-06 elapsed=18:00:49 avg_step=1.58s recent_step=1.40s eta=03:59:49 +2026-09-21 00:39:09 | INFO | train rank=0 | step=40930/50000 progress=81.9% loss=0.081151 lr=8.729799e-06 lr_backbone=8.729799e-06 elapsed=18:01:09 avg_step=1.58s recent_step=1.96s eta=03:59:34 +2026-09-21 00:39:22 | INFO | train rank=0 | step=40940/50000 progress=81.9% loss=0.094602 lr=8.711139e-06 lr_backbone=8.711139e-06 elapsed=18:01:22 avg_step=1.58s recent_step=1.34s eta=03:59:18 +2026-09-21 00:39:39 | INFO | train rank=0 | step=40950/50000 progress=81.9% loss=0.072471 lr=8.692497e-06 lr_backbone=8.692497e-06 elapsed=18:01:39 avg_step=1.58s recent_step=1.67s eta=03:59:02 +2026-09-21 00:39:57 | INFO | train rank=0 | step=40960/50000 progress=81.9% loss=0.070245 lr=8.673873e-06 lr_backbone=8.673873e-06 elapsed=18:01:56 avg_step=1.58s recent_step=1.77s eta=03:58:47 +2026-09-21 00:40:13 | INFO | train rank=0 | step=40970/50000 progress=81.9% loss=0.064477 lr=8.655267e-06 lr_backbone=8.655267e-06 elapsed=18:02:12 avg_step=1.58s recent_step=1.58s eta=03:58:31 +2026-09-21 00:40:29 | INFO | train rank=0 | step=40980/50000 progress=82.0% loss=0.092946 lr=8.636680e-06 lr_backbone=8.636680e-06 elapsed=18:02:28 avg_step=1.58s recent_step=1.60s eta=03:58:15 +2026-09-21 00:40:41 | INFO | train rank=0 | step=40990/50000 progress=82.0% loss=0.070878 lr=8.618110e-06 lr_backbone=8.618110e-06 elapsed=18:02:41 avg_step=1.58s recent_step=1.23s eta=03:57:59 +2026-09-21 00:40:57 | INFO | train rank=0 | step=41000/50000 progress=82.0% loss=0.073253 lr=8.599558e-06 lr_backbone=8.599558e-06 elapsed=18:02:56 avg_step=1.58s recent_step=1.59s eta=03:57:43 +2026-09-21 00:41:15 | INFO | train rank=0 | step=41010/50000 progress=82.0% loss=0.071185 lr=8.581025e-06 lr_backbone=8.581025e-06 elapsed=18:03:14 avg_step=1.58s recent_step=1.78s eta=03:57:27 +2026-09-21 00:41:31 | INFO | train rank=0 | step=41020/50000 progress=82.0% loss=0.073287 lr=8.562510e-06 lr_backbone=8.562510e-06 elapsed=18:03:31 avg_step=1.58s recent_step=1.68s eta=03:57:12 +2026-09-21 00:41:48 | INFO | train rank=0 | step=41030/50000 progress=82.1% loss=0.088472 lr=8.544012e-06 lr_backbone=8.544012e-06 elapsed=18:03:48 avg_step=1.58s recent_step=1.67s eta=03:56:56 +2026-09-21 00:42:02 | INFO | train rank=0 | step=41040/50000 progress=82.1% loss=0.093135 lr=8.525533e-06 lr_backbone=8.525533e-06 elapsed=18:04:02 avg_step=1.58s recent_step=1.43s eta=03:56:40 +2026-09-21 00:42:20 | INFO | train rank=0 | step=41050/50000 progress=82.1% loss=0.098354 lr=8.507072e-06 lr_backbone=8.507072e-06 elapsed=18:04:20 avg_step=1.58s recent_step=1.78s eta=03:56:24 +2026-09-21 00:42:33 | INFO | train rank=0 | step=41060/50000 progress=82.1% loss=0.082377 lr=8.488630e-06 lr_backbone=8.488630e-06 elapsed=18:04:32 avg_step=1.58s recent_step=1.25s eta=03:56:08 +2026-09-21 00:42:45 | INFO | train rank=0 | step=41070/50000 progress=82.1% loss=0.082766 lr=8.470205e-06 lr_backbone=8.470205e-06 elapsed=18:04:45 avg_step=1.58s recent_step=1.27s eta=03:55:51 +2026-09-21 00:43:02 | INFO | train rank=0 | step=41080/50000 progress=82.2% loss=0.091023 lr=8.451799e-06 lr_backbone=8.451799e-06 elapsed=18:05:01 avg_step=1.58s recent_step=1.64s eta=03:55:36 +2026-09-21 00:43:24 | INFO | train rank=0 | step=41090/50000 progress=82.2% loss=0.082969 lr=8.433410e-06 lr_backbone=8.433410e-06 elapsed=18:05:24 avg_step=1.58s recent_step=2.24s eta=03:55:21 +2026-09-21 00:43:36 | INFO | train rank=0 | step=41100/50000 progress=82.2% loss=0.077846 lr=8.415040e-06 lr_backbone=8.415040e-06 elapsed=18:05:36 avg_step=1.58s recent_step=1.22s eta=03:55:04 +2026-09-21 00:43:48 | INFO | train rank=0 | step=41110/50000 progress=82.2% loss=0.113213 lr=8.396688e-06 lr_backbone=8.396688e-06 elapsed=18:05:48 avg_step=1.58s recent_step=1.18s eta=03:54:48 +2026-09-21 00:44:02 | INFO | train rank=0 | step=41120/50000 progress=82.2% loss=0.079513 lr=8.378355e-06 lr_backbone=8.378355e-06 elapsed=18:06:01 avg_step=1.58s recent_step=1.35s eta=03:54:31 +2026-09-21 00:44:22 | INFO | train rank=0 | step=41130/50000 progress=82.3% loss=0.077981 lr=8.360039e-06 lr_backbone=8.360039e-06 elapsed=18:06:21 avg_step=1.58s recent_step=1.98s eta=03:54:16 +2026-09-21 00:44:38 | INFO | train rank=0 | step=41140/50000 progress=82.3% loss=0.075109 lr=8.341742e-06 lr_backbone=8.341742e-06 elapsed=18:06:37 avg_step=1.58s recent_step=1.62s eta=03:54:01 +2026-09-21 00:44:55 | INFO | train rank=0 | step=41150/50000 progress=82.3% loss=0.088194 lr=8.323463e-06 lr_backbone=8.323463e-06 elapsed=18:06:55 avg_step=1.58s recent_step=1.73s eta=03:53:45 +2026-09-21 00:45:10 | INFO | train rank=0 | step=41160/50000 progress=82.3% loss=0.080043 lr=8.305202e-06 lr_backbone=8.305202e-06 elapsed=18:07:10 avg_step=1.58s recent_step=1.53s eta=03:53:29 +2026-09-21 00:45:24 | INFO | train rank=0 | step=41170/50000 progress=82.3% loss=0.092060 lr=8.286959e-06 lr_backbone=8.286959e-06 elapsed=18:07:24 avg_step=1.58s recent_step=1.36s eta=03:53:13 +2026-09-21 00:45:45 | INFO | train rank=0 | step=41180/50000 progress=82.4% loss=0.070959 lr=8.268735e-06 lr_backbone=8.268735e-06 elapsed=18:07:45 avg_step=1.58s recent_step=2.11s eta=03:52:58 +2026-09-21 00:45:56 | INFO | train rank=0 | step=41190/50000 progress=82.4% loss=0.140380 lr=8.250529e-06 lr_backbone=8.250529e-06 elapsed=18:07:56 avg_step=1.58s recent_step=1.10s eta=03:52:41 +2026-09-21 00:46:11 | INFO | train rank=0 | step=41200/50000 progress=82.4% loss=0.097563 lr=8.232341e-06 lr_backbone=8.232341e-06 elapsed=18:08:10 avg_step=1.58s recent_step=1.48s eta=03:52:25 +2026-09-21 00:46:32 | INFO | train rank=0 | step=41210/50000 progress=82.4% loss=0.072631 lr=8.214171e-06 lr_backbone=8.214171e-06 elapsed=18:08:31 avg_step=1.58s recent_step=2.09s eta=03:52:10 +2026-09-21 00:46:46 | INFO | train rank=0 | step=41220/50000 progress=82.4% loss=0.135276 lr=8.196020e-06 lr_backbone=8.196020e-06 elapsed=18:08:45 avg_step=1.58s recent_step=1.39s eta=03:51:54 +2026-09-21 00:47:06 | INFO | train rank=0 | step=41230/50000 progress=82.5% loss=0.084249 lr=8.177887e-06 lr_backbone=8.177887e-06 elapsed=18:09:06 avg_step=1.58s recent_step=2.04s eta=03:51:39 +2026-09-21 00:47:18 | INFO | train rank=0 | step=41240/50000 progress=82.5% loss=0.143963 lr=8.159772e-06 lr_backbone=8.159772e-06 elapsed=18:09:17 avg_step=1.58s recent_step=1.16s eta=03:51:22 +2026-09-21 00:47:31 | INFO | train rank=0 | step=41250/50000 progress=82.5% loss=0.102889 lr=8.141676e-06 lr_backbone=8.141676e-06 elapsed=18:09:31 avg_step=1.58s recent_step=1.35s eta=03:51:06 +2026-09-21 00:47:51 | INFO | train rank=0 | step=41260/50000 progress=82.5% loss=0.115089 lr=8.123598e-06 lr_backbone=8.123598e-06 elapsed=18:09:50 avg_step=1.58s recent_step=1.96s eta=03:50:51 +2026-09-21 00:48:06 | INFO | train rank=0 | step=41270/50000 progress=82.5% loss=0.082661 lr=8.105538e-06 lr_backbone=8.105538e-06 elapsed=18:10:05 avg_step=1.58s recent_step=1.47s eta=03:50:35 +2026-09-21 00:48:20 | INFO | train rank=0 | step=41280/50000 progress=82.6% loss=0.106936 lr=8.087497e-06 lr_backbone=8.087497e-06 elapsed=18:10:19 avg_step=1.58s recent_step=1.40s eta=03:50:19 +2026-09-21 00:48:35 | INFO | train rank=0 | step=41290/50000 progress=82.6% loss=0.089548 lr=8.069474e-06 lr_backbone=8.069474e-06 elapsed=18:10:34 avg_step=1.58s recent_step=1.50s eta=03:50:03 +2026-09-21 00:48:53 | INFO | train rank=0 | step=41300/50000 progress=82.6% loss=0.078492 lr=8.051469e-06 lr_backbone=8.051469e-06 elapsed=18:10:53 avg_step=1.58s recent_step=1.86s eta=03:49:47 +2026-09-21 00:49:10 | INFO | train rank=0 | step=41310/50000 progress=82.6% loss=0.089494 lr=8.033482e-06 lr_backbone=8.033482e-06 elapsed=18:11:10 avg_step=1.58s recent_step=1.69s eta=03:49:32 +2026-09-21 00:49:29 | INFO | train rank=0 | step=41320/50000 progress=82.6% loss=0.116429 lr=8.015514e-06 lr_backbone=8.015514e-06 elapsed=18:11:28 avg_step=1.58s recent_step=1.85s eta=03:49:17 +2026-09-21 00:49:41 | INFO | train rank=0 | step=41330/50000 progress=82.7% loss=0.083345 lr=7.997565e-06 lr_backbone=7.997565e-06 elapsed=18:11:40 avg_step=1.58s recent_step=1.21s eta=03:49:00 +2026-09-21 00:49:56 | INFO | train rank=0 | step=41340/50000 progress=82.7% loss=0.099591 lr=7.979633e-06 lr_backbone=7.979633e-06 elapsed=18:11:56 avg_step=1.58s recent_step=1.53s eta=03:48:44 +2026-09-21 00:50:11 | INFO | train rank=0 | step=41350/50000 progress=82.7% loss=0.152754 lr=7.961720e-06 lr_backbone=7.961720e-06 elapsed=18:12:11 avg_step=1.58s recent_step=1.50s eta=03:48:28 +2026-09-21 00:50:28 | INFO | train rank=0 | step=41360/50000 progress=82.7% loss=0.078162 lr=7.943826e-06 lr_backbone=7.943826e-06 elapsed=18:12:27 avg_step=1.58s recent_step=1.67s eta=03:48:12 +2026-09-21 00:50:42 | INFO | train rank=0 | step=41370/50000 progress=82.7% loss=0.100168 lr=7.925950e-06 lr_backbone=7.925950e-06 elapsed=18:12:42 avg_step=1.58s recent_step=1.46s eta=03:47:56 +2026-09-21 00:51:03 | INFO | train rank=0 | step=41380/50000 progress=82.8% loss=0.088801 lr=7.908092e-06 lr_backbone=7.908092e-06 elapsed=18:13:03 avg_step=1.58s recent_step=2.08s eta=03:47:41 +2026-09-21 00:51:15 | INFO | train rank=0 | step=41390/50000 progress=82.8% loss=0.111448 lr=7.890253e-06 lr_backbone=7.890253e-06 elapsed=18:13:15 avg_step=1.58s recent_step=1.18s eta=03:47:25 +2026-09-21 00:51:32 | INFO | train rank=0 | step=41400/50000 progress=82.8% loss=0.086049 lr=7.872432e-06 lr_backbone=7.872432e-06 elapsed=18:13:31 avg_step=1.58s recent_step=1.66s eta=03:47:09 +2026-09-21 00:51:52 | INFO | train rank=0 | step=41410/50000 progress=82.8% loss=0.097639 lr=7.854629e-06 lr_backbone=7.854629e-06 elapsed=18:13:52 avg_step=1.58s recent_step=2.06s eta=03:46:54 +2026-09-21 00:52:06 | INFO | train rank=0 | step=41420/50000 progress=82.8% loss=0.112946 lr=7.836845e-06 lr_backbone=7.836845e-06 elapsed=18:14:06 avg_step=1.58s recent_step=1.38s eta=03:46:38 +2026-09-21 00:52:18 | INFO | train rank=0 | step=41430/50000 progress=82.9% loss=0.073011 lr=7.819080e-06 lr_backbone=7.819080e-06 elapsed=18:14:18 avg_step=1.58s recent_step=1.20s eta=03:46:21 +2026-09-21 00:52:36 | INFO | train rank=0 | step=41440/50000 progress=82.9% loss=0.079734 lr=7.801332e-06 lr_backbone=7.801332e-06 elapsed=18:14:35 avg_step=1.58s recent_step=1.78s eta=03:46:06 +2026-09-21 00:52:57 | INFO | train rank=0 | step=41450/50000 progress=82.9% loss=0.070849 lr=7.783604e-06 lr_backbone=7.783604e-06 elapsed=18:14:57 avg_step=1.58s recent_step=2.18s eta=03:45:51 +2026-09-21 00:53:09 | INFO | train rank=0 | step=41460/50000 progress=82.9% loss=0.098947 lr=7.765893e-06 lr_backbone=7.765893e-06 elapsed=18:15:09 avg_step=1.58s recent_step=1.14s eta=03:45:34 +2026-09-21 00:53:21 | INFO | train rank=0 | step=41470/50000 progress=82.9% loss=0.081456 lr=7.748202e-06 lr_backbone=7.748202e-06 elapsed=18:15:21 avg_step=1.58s recent_step=1.23s eta=03:45:18 +2026-09-21 00:53:38 | INFO | train rank=0 | step=41480/50000 progress=83.0% loss=0.084648 lr=7.730528e-06 lr_backbone=7.730528e-06 elapsed=18:15:38 avg_step=1.58s recent_step=1.73s eta=03:45:02 +2026-09-21 00:53:55 | INFO | train rank=0 | step=41490/50000 progress=83.0% loss=0.100503 lr=7.712874e-06 lr_backbone=7.712874e-06 elapsed=18:15:55 avg_step=1.58s recent_step=1.65s eta=03:44:47 +2026-09-21 00:54:12 | INFO | train rank=0 | step=41500/50000 progress=83.0% loss=0.079701 lr=7.695237e-06 lr_backbone=7.695237e-06 elapsed=18:16:12 avg_step=1.58s recent_step=1.70s eta=03:44:31 +2026-09-21 00:54:30 | INFO | train rank=0 | step=41510/50000 progress=83.0% loss=0.105605 lr=7.677620e-06 lr_backbone=7.677620e-06 elapsed=18:16:29 avg_step=1.58s recent_step=1.78s eta=03:44:15 +2026-09-21 00:54:41 | INFO | train rank=0 | step=41520/50000 progress=83.0% loss=0.077411 lr=7.660020e-06 lr_backbone=7.660020e-06 elapsed=18:16:41 avg_step=1.58s recent_step=1.17s eta=03:43:59 +2026-09-21 00:54:58 | INFO | train rank=0 | step=41530/50000 progress=83.1% loss=0.098308 lr=7.642440e-06 lr_backbone=7.642440e-06 elapsed=18:16:57 avg_step=1.58s recent_step=1.62s eta=03:43:43 +2026-09-21 00:55:13 | INFO | train rank=0 | step=41540/50000 progress=83.1% loss=0.092260 lr=7.624877e-06 lr_backbone=7.624877e-06 elapsed=18:17:13 avg_step=1.58s recent_step=1.56s eta=03:43:27 +2026-09-21 00:55:29 | INFO | train rank=0 | step=41550/50000 progress=83.1% loss=0.116636 lr=7.607334e-06 lr_backbone=7.607334e-06 elapsed=18:17:29 avg_step=1.58s recent_step=1.62s eta=03:43:11 +2026-09-21 00:55:45 | INFO | train rank=0 | step=41560/50000 progress=83.1% loss=0.071435 lr=7.589809e-06 lr_backbone=7.589809e-06 elapsed=18:17:45 avg_step=1.58s recent_step=1.60s eta=03:42:55 +2026-09-21 00:56:03 | INFO | train rank=0 | step=41570/50000 progress=83.1% loss=0.097592 lr=7.572302e-06 lr_backbone=7.572302e-06 elapsed=18:18:02 avg_step=1.58s recent_step=1.72s eta=03:42:40 +2026-09-21 00:56:16 | INFO | train rank=0 | step=41580/50000 progress=83.2% loss=0.084528 lr=7.554814e-06 lr_backbone=7.554814e-06 elapsed=18:18:16 avg_step=1.58s recent_step=1.38s eta=03:42:24 +2026-09-21 00:56:36 | INFO | train rank=0 | step=41590/50000 progress=83.2% loss=0.075300 lr=7.537344e-06 lr_backbone=7.537344e-06 elapsed=18:18:36 avg_step=1.58s recent_step=1.97s eta=03:42:09 +2026-09-21 00:56:48 | INFO | train rank=0 | step=41600/50000 progress=83.2% loss=0.106403 lr=7.519894e-06 lr_backbone=7.519894e-06 elapsed=18:18:47 avg_step=1.58s recent_step=1.15s eta=03:41:52 +2026-09-21 00:57:07 | INFO | train rank=0 | step=41610/50000 progress=83.2% loss=0.088137 lr=7.502461e-06 lr_backbone=7.502461e-06 elapsed=18:19:06 avg_step=1.58s recent_step=1.93s eta=03:41:37 +2026-09-21 00:57:19 | INFO | train rank=0 | step=41620/50000 progress=83.2% loss=0.085610 lr=7.485048e-06 lr_backbone=7.485048e-06 elapsed=18:19:19 avg_step=1.58s recent_step=1.22s eta=03:41:20 +2026-09-21 00:57:37 | INFO | train rank=0 | step=41630/50000 progress=83.3% loss=0.081440 lr=7.467652e-06 lr_backbone=7.467652e-06 elapsed=18:19:36 avg_step=1.58s recent_step=1.75s eta=03:41:05 +2026-09-21 00:57:53 | INFO | train rank=0 | step=41640/50000 progress=83.3% loss=0.091745 lr=7.450276e-06 lr_backbone=7.450276e-06 elapsed=18:19:53 avg_step=1.58s recent_step=1.68s eta=03:40:49 +2026-09-21 00:58:10 | INFO | train rank=0 | step=41650/50000 progress=83.3% loss=0.102409 lr=7.432918e-06 lr_backbone=7.432918e-06 elapsed=18:20:09 avg_step=1.58s recent_step=1.63s eta=03:40:33 +2026-09-21 00:58:21 | INFO | train rank=0 | step=41660/50000 progress=83.3% loss=0.078139 lr=7.415579e-06 lr_backbone=7.415579e-06 elapsed=18:20:21 avg_step=1.58s recent_step=1.14s eta=03:40:16 +2026-09-21 00:58:40 | INFO | train rank=0 | step=41670/50000 progress=83.3% loss=0.107157 lr=7.398258e-06 lr_backbone=7.398258e-06 elapsed=18:20:40 avg_step=1.58s recent_step=1.90s eta=03:40:01 +2026-09-21 00:58:54 | INFO | train rank=0 | step=41680/50000 progress=83.4% loss=0.071806 lr=7.380956e-06 lr_backbone=7.380956e-06 elapsed=18:20:54 avg_step=1.58s recent_step=1.41s eta=03:39:45 +2026-09-21 00:59:12 | INFO | train rank=0 | step=41690/50000 progress=83.4% loss=0.116253 lr=7.363673e-06 lr_backbone=7.363673e-06 elapsed=18:21:12 avg_step=1.58s recent_step=1.82s eta=03:39:30 +2026-09-21 00:59:27 | INFO | train rank=0 | step=41700/50000 progress=83.4% loss=0.091815 lr=7.346408e-06 lr_backbone=7.346408e-06 elapsed=18:21:26 avg_step=1.58s recent_step=1.43s eta=03:39:13 +2026-09-21 00:59:43 | INFO | train rank=0 | step=41710/50000 progress=83.4% loss=0.080619 lr=7.329162e-06 lr_backbone=7.329162e-06 elapsed=18:21:43 avg_step=1.58s recent_step=1.62s eta=03:38:58 +2026-09-21 00:59:57 | INFO | train rank=0 | step=41720/50000 progress=83.4% loss=0.062658 lr=7.311934e-06 lr_backbone=7.311934e-06 elapsed=18:21:56 avg_step=1.58s recent_step=1.38s eta=03:38:41 +2026-09-21 01:00:13 | INFO | train rank=0 | step=41730/50000 progress=83.5% loss=0.160592 lr=7.294726e-06 lr_backbone=7.294726e-06 elapsed=18:22:13 avg_step=1.58s recent_step=1.61s eta=03:38:26 +2026-09-21 01:00:32 | INFO | train rank=0 | step=41740/50000 progress=83.5% loss=0.099110 lr=7.277536e-06 lr_backbone=7.277536e-06 elapsed=18:22:31 avg_step=1.58s recent_step=1.88s eta=03:38:10 +2026-09-21 01:00:45 | INFO | train rank=0 | step=41750/50000 progress=83.5% loss=0.161657 lr=7.260364e-06 lr_backbone=7.260364e-06 elapsed=18:22:44 avg_step=1.58s recent_step=1.30s eta=03:37:54 +2026-09-21 01:01:02 | INFO | train rank=0 | step=41760/50000 progress=83.5% loss=0.082231 lr=7.243212e-06 lr_backbone=7.243212e-06 elapsed=18:23:02 avg_step=1.58s recent_step=1.75s eta=03:37:38 +2026-09-21 01:01:17 | INFO | train rank=0 | step=41770/50000 progress=83.5% loss=0.145633 lr=7.226078e-06 lr_backbone=7.226078e-06 elapsed=18:23:17 avg_step=1.58s recent_step=1.52s eta=03:37:22 +2026-09-21 01:01:32 | INFO | train rank=0 | step=41780/50000 progress=83.6% loss=0.069588 lr=7.208963e-06 lr_backbone=7.208963e-06 elapsed=18:23:32 avg_step=1.58s recent_step=1.46s eta=03:37:06 +2026-09-21 01:01:50 | INFO | train rank=0 | step=41790/50000 progress=83.6% loss=0.099829 lr=7.191866e-06 lr_backbone=7.191866e-06 elapsed=18:23:49 avg_step=1.58s recent_step=1.78s eta=03:36:51 +2026-09-21 01:02:08 | INFO | train rank=0 | step=41800/50000 progress=83.6% loss=0.114364 lr=7.174788e-06 lr_backbone=7.174788e-06 elapsed=18:24:08 avg_step=1.58s recent_step=1.81s eta=03:36:36 +2026-09-21 01:02:20 | INFO | train rank=0 | step=41810/50000 progress=83.6% loss=0.132839 lr=7.157729e-06 lr_backbone=7.157729e-06 elapsed=18:24:20 avg_step=1.58s recent_step=1.24s eta=03:36:19 +2026-09-21 01:02:40 | INFO | train rank=0 | step=41820/50000 progress=83.6% loss=0.100571 lr=7.140689e-06 lr_backbone=7.140689e-06 elapsed=18:24:39 avg_step=1.58s recent_step=1.91s eta=03:36:04 +2026-09-21 01:02:52 | INFO | train rank=0 | step=41830/50000 progress=83.7% loss=0.103422 lr=7.123667e-06 lr_backbone=7.123667e-06 elapsed=18:24:51 avg_step=1.58s recent_step=1.23s eta=03:35:47 +2026-09-21 01:03:11 | INFO | train rank=0 | step=41840/50000 progress=83.7% loss=0.101406 lr=7.106664e-06 lr_backbone=7.106664e-06 elapsed=18:25:11 avg_step=1.58s recent_step=1.93s eta=03:35:32 +2026-09-21 01:03:23 | INFO | train rank=0 | step=41850/50000 progress=83.7% loss=0.093977 lr=7.089680e-06 lr_backbone=7.089680e-06 elapsed=18:25:23 avg_step=1.58s recent_step=1.21s eta=03:35:16 +2026-09-21 01:03:41 | INFO | train rank=0 | step=41860/50000 progress=83.7% loss=0.085363 lr=7.072715e-06 lr_backbone=7.072715e-06 elapsed=18:25:41 avg_step=1.58s recent_step=1.81s eta=03:35:00 +2026-09-21 01:03:56 | INFO | train rank=0 | step=41870/50000 progress=83.7% loss=0.079693 lr=7.055768e-06 lr_backbone=7.055768e-06 elapsed=18:25:56 avg_step=1.58s recent_step=1.49s eta=03:34:44 +2026-09-21 01:04:09 | INFO | train rank=0 | step=41880/50000 progress=83.8% loss=0.103523 lr=7.038841e-06 lr_backbone=7.038841e-06 elapsed=18:26:09 avg_step=1.58s recent_step=1.29s eta=03:34:28 +2026-09-21 01:04:24 | INFO | train rank=0 | step=41890/50000 progress=83.8% loss=0.104361 lr=7.021932e-06 lr_backbone=7.021932e-06 elapsed=18:26:24 avg_step=1.58s recent_step=1.52s eta=03:34:12 +2026-09-21 01:04:41 | INFO | train rank=0 | step=41900/50000 progress=83.8% loss=0.101532 lr=7.005042e-06 lr_backbone=7.005042e-06 elapsed=18:26:41 avg_step=1.58s recent_step=1.69s eta=03:33:56 +2026-09-21 01:04:56 | INFO | train rank=0 | step=41910/50000 progress=83.8% loss=0.080019 lr=6.988170e-06 lr_backbone=6.988170e-06 elapsed=18:26:55 avg_step=1.58s recent_step=1.43s eta=03:33:40 +2026-09-21 01:05:12 | INFO | train rank=0 | step=41920/50000 progress=83.8% loss=0.073021 lr=6.971318e-06 lr_backbone=6.971318e-06 elapsed=18:27:12 avg_step=1.58s recent_step=1.67s eta=03:33:24 +2026-09-21 01:05:29 | INFO | train rank=0 | step=41930/50000 progress=83.9% loss=0.106390 lr=6.954484e-06 lr_backbone=6.954484e-06 elapsed=18:27:28 avg_step=1.58s recent_step=1.63s eta=03:33:08 +2026-09-21 01:05:44 | INFO | train rank=0 | step=41940/50000 progress=83.9% loss=0.076872 lr=6.937669e-06 lr_backbone=6.937669e-06 elapsed=18:27:43 avg_step=1.58s recent_step=1.51s eta=03:32:52 +2026-09-21 01:06:02 | INFO | train rank=0 | step=41950/50000 progress=83.9% loss=0.083344 lr=6.920873e-06 lr_backbone=6.920873e-06 elapsed=18:28:01 avg_step=1.58s recent_step=1.82s eta=03:32:37 +2026-09-21 01:06:18 | INFO | train rank=0 | step=41960/50000 progress=83.9% loss=0.120782 lr=6.904096e-06 lr_backbone=6.904096e-06 elapsed=18:28:18 avg_step=1.58s recent_step=1.62s eta=03:32:21 +2026-09-21 01:06:30 | INFO | train rank=0 | step=41970/50000 progress=83.9% loss=0.091726 lr=6.887338e-06 lr_backbone=6.887338e-06 elapsed=18:28:30 avg_step=1.58s recent_step=1.22s eta=03:32:05 +2026-09-21 01:06:56 | INFO | train rank=0 | step=41980/50000 progress=84.0% loss=0.097545 lr=6.870598e-06 lr_backbone=6.870598e-06 elapsed=18:28:56 avg_step=1.58s recent_step=2.62s eta=03:31:51 +2026-09-21 01:07:08 | INFO | train rank=0 | step=41990/50000 progress=84.0% loss=0.054318 lr=6.853878e-06 lr_backbone=6.853878e-06 elapsed=18:29:08 avg_step=1.58s recent_step=1.17s eta=03:31:34 +2026-09-21 01:07:20 | INFO | train rank=0 | step=42000/50000 progress=84.0% loss=0.111308 lr=6.837176e-06 lr_backbone=6.837176e-06 elapsed=18:29:20 avg_step=1.58s recent_step=1.21s eta=03:31:18 +2026-09-21 01:07:33 | INFO | train rank=0 | step=42010/50000 progress=84.0% loss=0.094105 lr=6.820493e-06 lr_backbone=6.820493e-06 elapsed=18:29:32 avg_step=1.58s recent_step=1.25s eta=03:31:01 +2026-09-21 01:07:49 | INFO | train rank=0 | step=42020/50000 progress=84.0% loss=0.085569 lr=6.803829e-06 lr_backbone=6.803829e-06 elapsed=18:29:49 avg_step=1.58s recent_step=1.67s eta=03:30:45 +2026-09-21 01:08:05 | INFO | train rank=0 | step=42030/50000 progress=84.1% loss=0.107417 lr=6.787184e-06 lr_backbone=6.787184e-06 elapsed=18:30:05 avg_step=1.58s recent_step=1.59s eta=03:30:30 +2026-09-21 01:08:22 | INFO | train rank=0 | step=42040/50000 progress=84.1% loss=0.107514 lr=6.770558e-06 lr_backbone=6.770558e-06 elapsed=18:30:21 avg_step=1.58s recent_step=1.64s eta=03:30:14 +2026-09-21 01:08:38 | INFO | train rank=0 | step=42050/50000 progress=84.1% loss=0.087120 lr=6.753951e-06 lr_backbone=6.753951e-06 elapsed=18:30:37 avg_step=1.58s recent_step=1.60s eta=03:29:58 +2026-09-21 01:08:58 | INFO | train rank=0 | step=42060/50000 progress=84.1% loss=0.077094 lr=6.737362e-06 lr_backbone=6.737362e-06 elapsed=18:30:58 avg_step=1.58s recent_step=2.06s eta=03:29:43 +2026-09-21 01:09:11 | INFO | train rank=0 | step=42070/50000 progress=84.1% loss=0.075818 lr=6.720793e-06 lr_backbone=6.720793e-06 elapsed=18:31:11 avg_step=1.58s recent_step=1.28s eta=03:29:27 +2026-09-21 01:09:23 | INFO | train rank=0 | step=42080/50000 progress=84.2% loss=0.110117 lr=6.704242e-06 lr_backbone=6.704242e-06 elapsed=18:31:23 avg_step=1.58s recent_step=1.23s eta=03:29:10 +2026-09-21 01:09:38 | INFO | train rank=0 | step=42090/50000 progress=84.2% loss=0.062342 lr=6.687711e-06 lr_backbone=6.687711e-06 elapsed=18:31:37 avg_step=1.58s recent_step=1.44s eta=03:28:54 +2026-09-21 01:10:00 | INFO | train rank=0 | step=42100/50000 progress=84.2% loss=0.116740 lr=6.671198e-06 lr_backbone=6.671198e-06 elapsed=18:32:00 avg_step=1.58s recent_step=2.24s eta=03:28:39 +2026-09-21 01:10:12 | INFO | train rank=0 | step=42110/50000 progress=84.2% loss=0.060962 lr=6.654705e-06 lr_backbone=6.654705e-06 elapsed=18:32:12 avg_step=1.58s recent_step=1.20s eta=03:28:23 +2026-09-21 01:10:24 | INFO | train rank=0 | step=42120/50000 progress=84.2% loss=0.099953 lr=6.638230e-06 lr_backbone=6.638230e-06 elapsed=18:32:24 avg_step=1.58s recent_step=1.18s eta=03:28:06 +2026-09-21 01:10:45 | INFO | train rank=0 | step=42130/50000 progress=84.3% loss=0.065914 lr=6.621774e-06 lr_backbone=6.621774e-06 elapsed=18:32:45 avg_step=1.58s recent_step=2.14s eta=03:27:51 +2026-09-21 01:11:02 | INFO | train rank=0 | step=42140/50000 progress=84.3% loss=0.085045 lr=6.605338e-06 lr_backbone=6.605338e-06 elapsed=18:33:02 avg_step=1.58s recent_step=1.66s eta=03:27:36 +2026-09-21 01:11:19 | INFO | train rank=0 | step=42150/50000 progress=84.3% loss=0.072682 lr=6.588920e-06 lr_backbone=6.588920e-06 elapsed=18:33:19 avg_step=1.58s recent_step=1.72s eta=03:27:20 +2026-09-21 01:11:34 | INFO | train rank=0 | step=42160/50000 progress=84.3% loss=0.072845 lr=6.572521e-06 lr_backbone=6.572521e-06 elapsed=18:33:33 avg_step=1.58s recent_step=1.44s eta=03:27:04 +2026-09-21 01:11:46 | INFO | train rank=0 | step=42170/50000 progress=84.3% loss=0.076505 lr=6.556141e-06 lr_backbone=6.556141e-06 elapsed=18:33:46 avg_step=1.58s recent_step=1.25s eta=03:26:48 +2026-09-21 01:12:08 | INFO | train rank=0 | step=42180/50000 progress=84.4% loss=0.065727 lr=6.539780e-06 lr_backbone=6.539780e-06 elapsed=18:34:08 avg_step=1.58s recent_step=2.21s eta=03:26:33 +2026-09-21 01:12:20 | INFO | train rank=0 | step=42190/50000 progress=84.4% loss=0.077784 lr=6.523439e-06 lr_backbone=6.523439e-06 elapsed=18:34:19 avg_step=1.58s recent_step=1.15s eta=03:26:16 +2026-09-21 01:12:36 | INFO | train rank=0 | step=42200/50000 progress=84.4% loss=0.072730 lr=6.507116e-06 lr_backbone=6.507116e-06 elapsed=18:34:36 avg_step=1.58s recent_step=1.64s eta=03:26:00 +2026-09-21 01:12:49 | INFO | train rank=0 | step=42210/50000 progress=84.4% loss=0.077097 lr=6.490812e-06 lr_backbone=6.490812e-06 elapsed=18:34:49 avg_step=1.58s recent_step=1.31s eta=03:25:44 +2026-09-21 01:13:12 | INFO | train rank=0 | step=42220/50000 progress=84.4% loss=0.100857 lr=6.474528e-06 lr_backbone=6.474528e-06 elapsed=18:35:12 avg_step=1.58s recent_step=2.30s eta=03:25:30 +2026-09-21 01:13:24 | INFO | train rank=0 | step=42230/50000 progress=84.5% loss=0.081374 lr=6.458262e-06 lr_backbone=6.458262e-06 elapsed=18:35:23 avg_step=1.58s recent_step=1.17s eta=03:25:13 +2026-09-21 01:13:40 | INFO | train rank=0 | step=42240/50000 progress=84.5% loss=0.071293 lr=6.442015e-06 lr_backbone=6.442015e-06 elapsed=18:35:40 avg_step=1.58s recent_step=1.65s eta=03:24:57 +2026-09-21 01:13:53 | INFO | train rank=0 | step=42250/50000 progress=84.5% loss=0.073389 lr=6.425788e-06 lr_backbone=6.425788e-06 elapsed=18:35:53 avg_step=1.58s recent_step=1.26s eta=03:24:41 +2026-09-21 01:14:06 | INFO | train rank=0 | step=42260/50000 progress=84.5% loss=0.088025 lr=6.409579e-06 lr_backbone=6.409579e-06 elapsed=18:36:06 avg_step=1.58s recent_step=1.34s eta=03:24:25 +2026-09-21 01:14:29 | INFO | train rank=0 | step=42270/50000 progress=84.5% loss=0.087659 lr=6.393390e-06 lr_backbone=6.393390e-06 elapsed=18:36:29 avg_step=1.58s recent_step=2.29s eta=03:24:10 +2026-09-21 01:14:44 | INFO | train rank=0 | step=42280/50000 progress=84.6% loss=0.081594 lr=6.377220e-06 lr_backbone=6.377220e-06 elapsed=18:36:44 avg_step=1.58s recent_step=1.49s eta=03:23:54 +2026-09-21 01:15:00 | INFO | train rank=0 | step=42290/50000 progress=84.6% loss=0.094176 lr=6.361068e-06 lr_backbone=6.361068e-06 elapsed=18:36:59 avg_step=1.58s recent_step=1.54s eta=03:23:38 +2026-09-21 01:15:13 | INFO | train rank=0 | step=42300/50000 progress=84.6% loss=0.069191 lr=6.344936e-06 lr_backbone=6.344936e-06 elapsed=18:37:12 avg_step=1.58s recent_step=1.30s eta=03:23:22 +2026-09-21 01:15:32 | INFO | train rank=0 | step=42310/50000 progress=84.6% loss=0.082170 lr=6.328823e-06 lr_backbone=6.328823e-06 elapsed=18:37:32 avg_step=1.58s recent_step=1.94s eta=03:23:06 +2026-09-21 01:15:44 | INFO | train rank=0 | step=42320/50000 progress=84.6% loss=0.091291 lr=6.312729e-06 lr_backbone=6.312729e-06 elapsed=18:37:43 avg_step=1.58s recent_step=1.19s eta=03:22:50 +2026-09-21 01:15:59 | INFO | train rank=0 | step=42330/50000 progress=84.7% loss=0.076487 lr=6.296654e-06 lr_backbone=6.296654e-06 elapsed=18:37:58 avg_step=1.58s recent_step=1.50s eta=03:22:34 +2026-09-21 01:16:17 | INFO | train rank=0 | step=42340/50000 progress=84.7% loss=0.072191 lr=6.280599e-06 lr_backbone=6.280599e-06 elapsed=18:38:16 avg_step=1.58s recent_step=1.79s eta=03:22:18 +2026-09-21 01:16:33 | INFO | train rank=0 | step=42350/50000 progress=84.7% loss=0.095200 lr=6.264562e-06 lr_backbone=6.264562e-06 elapsed=18:38:33 avg_step=1.58s recent_step=1.62s eta=03:22:03 +2026-09-21 01:16:49 | INFO | train rank=0 | step=42360/50000 progress=84.7% loss=0.091364 lr=6.248544e-06 lr_backbone=6.248544e-06 elapsed=18:38:49 avg_step=1.58s recent_step=1.60s eta=03:21:47 +2026-09-21 01:17:06 | INFO | train rank=0 | step=42370/50000 progress=84.7% loss=0.093472 lr=6.232546e-06 lr_backbone=6.232546e-06 elapsed=18:39:05 avg_step=1.58s recent_step=1.66s eta=03:21:31 +2026-09-21 01:17:17 | INFO | train rank=0 | step=42380/50000 progress=84.8% loss=0.056833 lr=6.216567e-06 lr_backbone=6.216567e-06 elapsed=18:39:17 avg_step=1.58s recent_step=1.19s eta=03:21:15 +2026-09-21 01:17:35 | INFO | train rank=0 | step=42390/50000 progress=84.8% loss=0.094461 lr=6.200607e-06 lr_backbone=6.200607e-06 elapsed=18:39:34 avg_step=1.58s recent_step=1.74s eta=03:20:59 +2026-09-21 01:17:53 | INFO | train rank=0 | step=42400/50000 progress=84.8% loss=0.100135 lr=6.184666e-06 lr_backbone=6.184666e-06 elapsed=18:39:52 avg_step=1.58s recent_step=1.79s eta=03:20:43 +2026-09-21 01:18:04 | INFO | train rank=0 | step=42410/50000 progress=84.8% loss=0.073422 lr=6.168744e-06 lr_backbone=6.168744e-06 elapsed=18:40:04 avg_step=1.58s recent_step=1.17s eta=03:20:27 +2026-09-21 01:18:21 | INFO | train rank=0 | step=42420/50000 progress=84.8% loss=0.101634 lr=6.152842e-06 lr_backbone=6.152842e-06 elapsed=18:40:21 avg_step=1.58s recent_step=1.67s eta=03:20:11 +2026-09-21 01:18:38 | INFO | train rank=0 | step=42430/50000 progress=84.9% loss=0.088435 lr=6.136958e-06 lr_backbone=6.136958e-06 elapsed=18:40:37 avg_step=1.58s recent_step=1.67s eta=03:19:56 +2026-09-21 01:18:49 | INFO | train rank=0 | step=42440/50000 progress=84.9% loss=0.084900 lr=6.121094e-06 lr_backbone=6.121094e-06 elapsed=18:40:49 avg_step=1.58s recent_step=1.16s eta=03:19:39 +2026-09-21 01:19:02 | INFO | train rank=0 | step=42450/50000 progress=84.9% loss=0.103335 lr=6.105249e-06 lr_backbone=6.105249e-06 elapsed=18:41:01 avg_step=1.58s recent_step=1.22s eta=03:19:22 +2026-09-21 01:19:14 | INFO | train rank=0 | step=42460/50000 progress=84.9% loss=0.097509 lr=6.089423e-06 lr_backbone=6.089423e-06 elapsed=18:41:14 avg_step=1.58s recent_step=1.22s eta=03:19:06 +2026-09-21 01:19:38 | INFO | train rank=0 | step=42470/50000 progress=84.9% loss=0.089752 lr=6.073617e-06 lr_backbone=6.073617e-06 elapsed=18:41:38 avg_step=1.58s recent_step=2.41s eta=03:18:52 +2026-09-21 01:19:49 | INFO | train rank=0 | step=42480/50000 progress=85.0% loss=0.089394 lr=6.057829e-06 lr_backbone=6.057829e-06 elapsed=18:41:49 avg_step=1.58s recent_step=1.15s eta=03:18:35 +2026-09-21 01:20:14 | INFO | train rank=0 | step=42490/50000 progress=85.0% loss=0.077287 lr=6.042061e-06 lr_backbone=6.042061e-06 elapsed=18:42:13 avg_step=1.58s recent_step=2.44s eta=03:18:21 +2026-09-21 01:20:27 | INFO | train rank=0 | step=42500/50000 progress=85.0% loss=0.075125 lr=6.026312e-06 lr_backbone=6.026312e-06 elapsed=18:42:27 avg_step=1.58s recent_step=1.31s eta=03:18:04 +2026-09-21 01:20:39 | INFO | train rank=0 | step=42510/50000 progress=85.0% loss=0.084292 lr=6.010583e-06 lr_backbone=6.010583e-06 elapsed=18:42:39 avg_step=1.58s recent_step=1.21s eta=03:17:48 +2026-09-21 01:21:36 | INFO | train rank=0 | step=42520/50000 progress=85.0% loss=0.102019 lr=5.994872e-06 lr_backbone=5.994872e-06 elapsed=18:43:36 avg_step=1.59s recent_step=5.71s eta=03:17:39 +2026-09-21 01:21:49 | INFO | train rank=0 | step=42530/50000 progress=85.1% loss=0.099058 lr=5.979181e-06 lr_backbone=5.979181e-06 elapsed=18:43:48 avg_step=1.59s recent_step=1.23s eta=03:17:23 +2026-09-21 01:22:03 | INFO | train rank=0 | step=42540/50000 progress=85.1% loss=0.083054 lr=5.963509e-06 lr_backbone=5.963509e-06 elapsed=18:44:03 avg_step=1.59s recent_step=1.50s eta=03:17:07 +2026-09-21 01:22:17 | INFO | train rank=0 | step=42550/50000 progress=85.1% loss=0.089666 lr=5.947857e-06 lr_backbone=5.947857e-06 elapsed=18:44:17 avg_step=1.59s recent_step=1.36s eta=03:16:50 +2026-09-21 01:22:32 | INFO | train rank=0 | step=42560/50000 progress=85.1% loss=0.077093 lr=5.932223e-06 lr_backbone=5.932223e-06 elapsed=18:44:32 avg_step=1.59s recent_step=1.53s eta=03:16:35 +2026-09-21 01:22:50 | INFO | train rank=0 | step=42570/50000 progress=85.1% loss=0.080275 lr=5.916609e-06 lr_backbone=5.916609e-06 elapsed=18:44:50 avg_step=1.59s recent_step=1.75s eta=03:16:19 +2026-09-21 01:23:06 | INFO | train rank=0 | step=42580/50000 progress=85.2% loss=0.090979 lr=5.901014e-06 lr_backbone=5.901014e-06 elapsed=18:45:05 avg_step=1.59s recent_step=1.58s eta=03:16:03 +2026-09-21 01:23:21 | INFO | train rank=0 | step=42590/50000 progress=85.2% loss=0.094316 lr=5.885439e-06 lr_backbone=5.885439e-06 elapsed=18:45:21 avg_step=1.59s recent_step=1.55s eta=03:15:47 +2026-09-21 01:23:35 | INFO | train rank=0 | step=42600/50000 progress=85.2% loss=0.082879 lr=5.869882e-06 lr_backbone=5.869882e-06 elapsed=18:45:35 avg_step=1.59s recent_step=1.40s eta=03:15:31 +2026-09-21 01:23:48 | INFO | train rank=0 | step=42610/50000 progress=85.2% loss=0.089441 lr=5.854345e-06 lr_backbone=5.854345e-06 elapsed=18:45:48 avg_step=1.59s recent_step=1.27s eta=03:15:15 +2026-09-21 01:24:08 | INFO | train rank=0 | step=42620/50000 progress=85.2% loss=0.080232 lr=5.838828e-06 lr_backbone=5.838828e-06 elapsed=18:46:07 avg_step=1.59s recent_step=1.98s eta=03:14:59 +2026-09-21 01:24:20 | INFO | train rank=0 | step=42630/50000 progress=85.3% loss=0.073399 lr=5.823330e-06 lr_backbone=5.823330e-06 elapsed=18:46:19 avg_step=1.59s recent_step=1.18s eta=03:14:43 +2026-09-21 01:24:37 | INFO | train rank=0 | step=42640/50000 progress=85.3% loss=0.109953 lr=5.807851e-06 lr_backbone=5.807851e-06 elapsed=18:46:36 avg_step=1.59s recent_step=1.72s eta=03:14:27 +2026-09-21 01:24:54 | INFO | train rank=0 | step=42650/50000 progress=85.3% loss=0.115357 lr=5.792391e-06 lr_backbone=5.792391e-06 elapsed=18:46:53 avg_step=1.59s recent_step=1.68s eta=03:14:12 +2026-09-21 01:25:06 | INFO | train rank=0 | step=42660/50000 progress=85.3% loss=0.092603 lr=5.776951e-06 lr_backbone=5.776951e-06 elapsed=18:47:05 avg_step=1.59s recent_step=1.20s eta=03:13:55 +2026-09-21 01:25:18 | INFO | train rank=0 | step=42670/50000 progress=85.3% loss=0.082658 lr=5.761530e-06 lr_backbone=5.761530e-06 elapsed=18:47:18 avg_step=1.59s recent_step=1.25s eta=03:13:39 +2026-09-21 01:25:34 | INFO | train rank=0 | step=42680/50000 progress=85.4% loss=0.094496 lr=5.746128e-06 lr_backbone=5.746128e-06 elapsed=18:47:33 avg_step=1.59s recent_step=1.56s eta=03:13:23 +2026-09-21 01:25:53 | INFO | train rank=0 | step=42690/50000 progress=85.4% loss=0.095496 lr=5.730746e-06 lr_backbone=5.730746e-06 elapsed=18:47:53 avg_step=1.59s recent_step=1.98s eta=03:13:08 +2026-09-21 01:26:06 | INFO | train rank=0 | step=42700/50000 progress=85.4% loss=0.081869 lr=5.715383e-06 lr_backbone=5.715383e-06 elapsed=18:48:05 avg_step=1.59s recent_step=1.23s eta=03:12:51 +2026-09-21 01:26:25 | INFO | train rank=0 | step=42710/50000 progress=85.4% loss=0.133829 lr=5.700039e-06 lr_backbone=5.700039e-06 elapsed=18:48:25 avg_step=1.59s recent_step=1.95s eta=03:12:36 +2026-09-21 01:26:40 | INFO | train rank=0 | step=42720/50000 progress=85.4% loss=0.159252 lr=5.684715e-06 lr_backbone=5.684715e-06 elapsed=18:48:40 avg_step=1.59s recent_step=1.50s eta=03:12:20 +2026-09-21 01:26:54 | INFO | train rank=0 | step=42730/50000 progress=85.5% loss=0.076914 lr=5.669410e-06 lr_backbone=5.669410e-06 elapsed=18:48:54 avg_step=1.59s recent_step=1.42s eta=03:12:04 +2026-09-21 01:27:10 | INFO | train rank=0 | step=42740/50000 progress=85.5% loss=0.066181 lr=5.654125e-06 lr_backbone=5.654125e-06 elapsed=18:49:09 avg_step=1.59s recent_step=1.52s eta=03:11:48 +2026-09-21 01:27:28 | INFO | train rank=0 | step=42750/50000 progress=85.5% loss=0.081226 lr=5.638859e-06 lr_backbone=5.638859e-06 elapsed=18:49:28 avg_step=1.59s recent_step=1.86s eta=03:11:32 +2026-09-21 01:27:44 | INFO | train rank=0 | step=42760/50000 progress=85.5% loss=0.091440 lr=5.623612e-06 lr_backbone=5.623612e-06 elapsed=18:49:43 avg_step=1.59s recent_step=1.55s eta=03:11:16 +2026-09-21 01:28:01 | INFO | train rank=0 | step=42770/50000 progress=85.5% loss=0.081142 lr=5.608385e-06 lr_backbone=5.608385e-06 elapsed=18:50:01 avg_step=1.59s recent_step=1.72s eta=03:11:01 +2026-09-21 01:28:19 | INFO | train rank=0 | step=42780/50000 progress=85.6% loss=0.117751 lr=5.593178e-06 lr_backbone=5.593178e-06 elapsed=18:50:18 avg_step=1.59s recent_step=1.79s eta=03:10:45 +2026-09-21 01:28:33 | INFO | train rank=0 | step=42790/50000 progress=85.6% loss=0.065954 lr=5.577989e-06 lr_backbone=5.577989e-06 elapsed=18:50:33 avg_step=1.59s recent_step=1.46s eta=03:10:29 +2026-09-21 01:28:49 | INFO | train rank=0 | step=42800/50000 progress=85.6% loss=0.092326 lr=5.562820e-06 lr_backbone=5.562820e-06 elapsed=18:50:49 avg_step=1.59s recent_step=1.59s eta=03:10:13 +2026-09-21 01:29:04 | INFO | train rank=0 | step=42810/50000 progress=85.6% loss=0.108951 lr=5.547671e-06 lr_backbone=5.547671e-06 elapsed=18:51:04 avg_step=1.59s recent_step=1.52s eta=03:09:57 +2026-09-21 01:29:23 | INFO | train rank=0 | step=42820/50000 progress=85.6% loss=0.080648 lr=5.532541e-06 lr_backbone=5.532541e-06 elapsed=18:51:23 avg_step=1.59s recent_step=1.88s eta=03:09:42 +2026-09-21 01:29:42 | INFO | train rank=0 | step=42830/50000 progress=85.7% loss=0.087420 lr=5.517430e-06 lr_backbone=5.517430e-06 elapsed=18:51:42 avg_step=1.59s recent_step=1.91s eta=03:09:27 +2026-09-21 01:29:54 | INFO | train rank=0 | step=42840/50000 progress=85.7% loss=0.067378 lr=5.502339e-06 lr_backbone=5.502339e-06 elapsed=18:51:54 avg_step=1.59s recent_step=1.20s eta=03:09:10 +2026-09-21 01:30:09 | INFO | train rank=0 | step=42850/50000 progress=85.7% loss=0.084442 lr=5.487268e-06 lr_backbone=5.487268e-06 elapsed=18:52:08 avg_step=1.59s recent_step=1.44s eta=03:08:54 +2026-09-21 01:30:26 | INFO | train rank=0 | step=42860/50000 progress=85.7% loss=0.080534 lr=5.472215e-06 lr_backbone=5.472215e-06 elapsed=18:52:25 avg_step=1.59s recent_step=1.69s eta=03:08:39 +2026-09-21 01:30:42 | INFO | train rank=0 | step=42870/50000 progress=85.7% loss=0.067809 lr=5.457183e-06 lr_backbone=5.457183e-06 elapsed=18:52:41 avg_step=1.59s recent_step=1.62s eta=03:08:23 +2026-09-21 01:30:55 | INFO | train rank=0 | step=42880/50000 progress=85.8% loss=0.089150 lr=5.442170e-06 lr_backbone=5.442170e-06 elapsed=18:52:54 avg_step=1.59s recent_step=1.29s eta=03:08:06 +2026-09-21 01:31:12 | INFO | train rank=0 | step=42890/50000 progress=85.8% loss=0.091832 lr=5.427176e-06 lr_backbone=5.427176e-06 elapsed=18:53:12 avg_step=1.59s recent_step=1.76s eta=03:07:51 +2026-09-21 01:31:29 | INFO | train rank=0 | step=42900/50000 progress=85.8% loss=0.066190 lr=5.412202e-06 lr_backbone=5.412202e-06 elapsed=18:53:29 avg_step=1.59s recent_step=1.69s eta=03:07:35 +2026-09-21 01:31:45 | INFO | train rank=0 | step=42910/50000 progress=85.8% loss=0.067696 lr=5.397247e-06 lr_backbone=5.397247e-06 elapsed=18:53:45 avg_step=1.59s recent_step=1.59s eta=03:07:19 +2026-09-21 01:32:01 | INFO | train rank=0 | step=42920/50000 progress=85.8% loss=0.081369 lr=5.382312e-06 lr_backbone=5.382312e-06 elapsed=18:54:00 avg_step=1.59s recent_step=1.56s eta=03:07:03 +2026-09-21 01:32:16 | INFO | train rank=0 | step=42930/50000 progress=85.9% loss=0.095710 lr=5.367396e-06 lr_backbone=5.367396e-06 elapsed=18:54:16 avg_step=1.59s recent_step=1.52s eta=03:06:47 +2026-09-21 01:32:35 | INFO | train rank=0 | step=42940/50000 progress=85.9% loss=0.104800 lr=5.352500e-06 lr_backbone=5.352500e-06 elapsed=18:54:34 avg_step=1.59s recent_step=1.87s eta=03:06:32 +2026-09-21 01:32:52 | INFO | train rank=0 | step=42950/50000 progress=85.9% loss=0.113777 lr=5.337623e-06 lr_backbone=5.337623e-06 elapsed=18:54:52 avg_step=1.59s recent_step=1.76s eta=03:06:16 +2026-09-21 01:33:09 | INFO | train rank=0 | step=42960/50000 progress=85.9% loss=0.107133 lr=5.322766e-06 lr_backbone=5.322766e-06 elapsed=18:55:08 avg_step=1.59s recent_step=1.64s eta=03:06:01 +2026-09-21 01:33:23 | INFO | train rank=0 | step=42970/50000 progress=85.9% loss=0.099438 lr=5.307929e-06 lr_backbone=5.307929e-06 elapsed=18:55:23 avg_step=1.59s recent_step=1.45s eta=03:05:45 +2026-09-21 01:33:42 | INFO | train rank=0 | step=42980/50000 progress=86.0% loss=0.074677 lr=5.293111e-06 lr_backbone=5.293111e-06 elapsed=18:55:42 avg_step=1.59s recent_step=1.94s eta=03:05:29 +2026-09-21 01:33:54 | INFO | train rank=0 | step=42990/50000 progress=86.0% loss=0.084886 lr=5.278312e-06 lr_backbone=5.278312e-06 elapsed=18:55:54 avg_step=1.59s recent_step=1.19s eta=03:05:13 +2026-09-21 01:34:07 | INFO | train rank=0 | step=43000/50000 progress=86.0% loss=0.163198 lr=5.263534e-06 lr_backbone=5.263534e-06 elapsed=18:56:06 avg_step=1.59s recent_step=1.22s eta=03:04:56 +2026-09-21 01:34:22 | INFO | train rank=0 | step=43010/50000 progress=86.0% loss=0.066372 lr=5.248774e-06 lr_backbone=5.248774e-06 elapsed=18:56:22 avg_step=1.59s recent_step=1.58s eta=03:04:41 +2026-09-21 01:34:44 | INFO | train rank=0 | step=43020/50000 progress=86.0% loss=0.128612 lr=5.234034e-06 lr_backbone=5.234034e-06 elapsed=18:56:44 avg_step=1.59s recent_step=2.17s eta=03:04:26 +2026-09-21 01:34:59 | INFO | train rank=0 | step=43030/50000 progress=86.1% loss=0.081711 lr=5.219314e-06 lr_backbone=5.219314e-06 elapsed=18:56:59 avg_step=1.59s recent_step=1.51s eta=03:04:10 +2026-09-21 01:35:15 | INFO | train rank=0 | step=43040/50000 progress=86.1% loss=0.079567 lr=5.204614e-06 lr_backbone=5.204614e-06 elapsed=18:57:14 avg_step=1.59s recent_step=1.55s eta=03:03:54 +2026-09-21 01:35:26 | INFO | train rank=0 | step=43050/50000 progress=86.1% loss=0.063882 lr=5.189933e-06 lr_backbone=5.189933e-06 elapsed=18:57:26 avg_step=1.59s recent_step=1.17s eta=03:03:37 +2026-09-21 01:35:43 | INFO | train rank=0 | step=43060/50000 progress=86.1% loss=0.083089 lr=5.175271e-06 lr_backbone=5.175271e-06 elapsed=18:57:43 avg_step=1.59s recent_step=1.66s eta=03:03:22 +2026-09-21 01:35:59 | INFO | train rank=0 | step=43070/50000 progress=86.1% loss=0.107693 lr=5.160630e-06 lr_backbone=5.160630e-06 elapsed=18:57:58 avg_step=1.59s recent_step=1.56s eta=03:03:06 +2026-09-21 01:36:11 | INFO | train rank=0 | step=43080/50000 progress=86.2% loss=0.101290 lr=5.146008e-06 lr_backbone=5.146008e-06 elapsed=18:58:11 avg_step=1.59s recent_step=1.24s eta=03:02:49 +2026-09-21 01:36:28 | INFO | train rank=0 | step=43090/50000 progress=86.2% loss=0.102622 lr=5.131405e-06 lr_backbone=5.131405e-06 elapsed=18:58:28 avg_step=1.59s recent_step=1.71s eta=03:02:34 +2026-09-21 01:36:47 | INFO | train rank=0 | step=43100/50000 progress=86.2% loss=0.101476 lr=5.116822e-06 lr_backbone=5.116822e-06 elapsed=18:58:46 avg_step=1.59s recent_step=1.87s eta=03:02:18 +2026-09-21 01:37:00 | INFO | train rank=0 | step=43110/50000 progress=86.2% loss=0.078404 lr=5.102259e-06 lr_backbone=5.102259e-06 elapsed=18:58:59 avg_step=1.59s recent_step=1.29s eta=03:02:02 +2026-09-21 01:37:15 | INFO | train rank=0 | step=43120/50000 progress=86.2% loss=0.074858 lr=5.087715e-06 lr_backbone=5.087715e-06 elapsed=18:59:15 avg_step=1.59s recent_step=1.55s eta=03:01:46 +2026-09-21 01:37:31 | INFO | train rank=0 | step=43130/50000 progress=86.3% loss=0.067581 lr=5.073191e-06 lr_backbone=5.073191e-06 elapsed=18:59:30 avg_step=1.59s recent_step=1.53s eta=03:01:30 +2026-09-21 01:37:48 | INFO | train rank=0 | step=43140/50000 progress=86.3% loss=0.104086 lr=5.058687e-06 lr_backbone=5.058687e-06 elapsed=18:59:48 avg_step=1.59s recent_step=1.75s eta=03:01:14 +2026-09-21 01:38:07 | INFO | train rank=0 | step=43150/50000 progress=86.3% loss=0.060717 lr=5.044202e-06 lr_backbone=5.044202e-06 elapsed=19:00:07 avg_step=1.59s recent_step=1.91s eta=03:00:59 +2026-09-21 01:38:19 | INFO | train rank=0 | step=43160/50000 progress=86.3% loss=0.065470 lr=5.029737e-06 lr_backbone=5.029737e-06 elapsed=19:00:19 avg_step=1.59s recent_step=1.18s eta=03:00:43 +2026-09-21 01:38:33 | INFO | train rank=0 | step=43170/50000 progress=86.3% loss=0.089099 lr=5.015292e-06 lr_backbone=5.015292e-06 elapsed=19:00:33 avg_step=1.59s recent_step=1.41s eta=03:00:26 +2026-09-21 01:38:45 | INFO | train rank=0 | step=43180/50000 progress=86.4% loss=0.092309 lr=5.000866e-06 lr_backbone=5.000866e-06 elapsed=19:00:45 avg_step=1.59s recent_step=1.23s eta=03:00:10 +2026-09-21 01:39:02 | INFO | train rank=0 | step=43190/50000 progress=86.4% loss=0.090695 lr=4.986461e-06 lr_backbone=4.986461e-06 elapsed=19:01:02 avg_step=1.59s recent_step=1.71s eta=02:59:54 +2026-09-21 01:39:18 | INFO | train rank=0 | step=43200/50000 progress=86.4% loss=0.153788 lr=4.972074e-06 lr_backbone=4.972074e-06 elapsed=19:01:18 avg_step=1.59s recent_step=1.58s eta=02:59:39 +2026-09-21 01:39:37 | INFO | train rank=0 | step=43210/50000 progress=86.4% loss=0.089282 lr=4.957708e-06 lr_backbone=4.957708e-06 elapsed=19:01:36 avg_step=1.59s recent_step=1.86s eta=02:59:23 +2026-09-21 01:39:49 | INFO | train rank=0 | step=43220/50000 progress=86.4% loss=0.125056 lr=4.943361e-06 lr_backbone=4.943361e-06 elapsed=19:01:49 avg_step=1.59s recent_step=1.24s eta=02:59:07 +2026-09-21 01:40:05 | INFO | train rank=0 | step=43230/50000 progress=86.5% loss=0.077886 lr=4.929034e-06 lr_backbone=4.929034e-06 elapsed=19:02:05 avg_step=1.59s recent_step=1.58s eta=02:58:51 +2026-09-21 01:40:22 | INFO | train rank=0 | step=43240/50000 progress=86.5% loss=0.119144 lr=4.914726e-06 lr_backbone=4.914726e-06 elapsed=19:02:21 avg_step=1.59s recent_step=1.66s eta=02:58:35 +2026-09-21 01:40:41 | INFO | train rank=0 | step=43250/50000 progress=86.5% loss=0.062980 lr=4.900438e-06 lr_backbone=4.900438e-06 elapsed=19:02:41 avg_step=1.59s recent_step=1.94s eta=02:58:20 +2026-09-21 01:40:54 | INFO | train rank=0 | step=43260/50000 progress=86.5% loss=0.100908 lr=4.886171e-06 lr_backbone=4.886171e-06 elapsed=19:02:53 avg_step=1.59s recent_step=1.26s eta=02:58:03 +2026-09-21 01:41:11 | INFO | train rank=0 | step=43270/50000 progress=86.5% loss=0.091192 lr=4.871922e-06 lr_backbone=4.871922e-06 elapsed=19:03:11 avg_step=1.59s recent_step=1.74s eta=02:57:48 +2026-09-21 01:41:23 | INFO | train rank=0 | step=43280/50000 progress=86.6% loss=0.075092 lr=4.857694e-06 lr_backbone=4.857694e-06 elapsed=19:03:23 avg_step=1.59s recent_step=1.22s eta=02:57:31 +2026-09-21 01:41:44 | INFO | train rank=0 | step=43290/50000 progress=86.6% loss=0.085822 lr=4.843485e-06 lr_backbone=4.843485e-06 elapsed=19:03:44 avg_step=1.59s recent_step=2.07s eta=02:57:16 +2026-09-21 01:41:58 | INFO | train rank=0 | step=43300/50000 progress=86.6% loss=0.090457 lr=4.829296e-06 lr_backbone=4.829296e-06 elapsed=19:03:58 avg_step=1.59s recent_step=1.45s eta=02:57:00 +2026-09-21 01:42:11 | INFO | train rank=0 | step=43310/50000 progress=86.6% loss=0.069057 lr=4.815127e-06 lr_backbone=4.815127e-06 elapsed=19:04:11 avg_step=1.59s recent_step=1.26s eta=02:56:44 +2026-09-21 01:42:23 | INFO | train rank=0 | step=43320/50000 progress=86.6% loss=0.084115 lr=4.800977e-06 lr_backbone=4.800977e-06 elapsed=19:04:23 avg_step=1.59s recent_step=1.20s eta=02:56:27 +2026-09-21 01:42:45 | INFO | train rank=0 | step=43330/50000 progress=86.7% loss=0.109690 lr=4.786847e-06 lr_backbone=4.786847e-06 elapsed=19:04:45 avg_step=1.59s recent_step=2.21s eta=02:56:13 +2026-09-21 01:43:01 | INFO | train rank=0 | step=43340/50000 progress=86.7% loss=0.079051 lr=4.772738e-06 lr_backbone=4.772738e-06 elapsed=19:05:00 avg_step=1.59s recent_step=1.55s eta=02:55:57 +2026-09-21 01:43:12 | INFO | train rank=0 | step=43350/50000 progress=86.7% loss=0.096978 lr=4.758647e-06 lr_backbone=4.758647e-06 elapsed=19:05:12 avg_step=1.59s recent_step=1.17s eta=02:55:40 +2026-09-21 01:43:32 | INFO | train rank=0 | step=43360/50000 progress=86.7% loss=0.096305 lr=4.744577e-06 lr_backbone=4.744577e-06 elapsed=19:05:31 avg_step=1.59s recent_step=1.95s eta=02:55:25 +2026-09-21 01:43:50 | INFO | train rank=0 | step=43370/50000 progress=86.7% loss=0.132140 lr=4.730526e-06 lr_backbone=4.730526e-06 elapsed=19:05:50 avg_step=1.59s recent_step=1.82s eta=02:55:09 +2026-09-21 01:44:05 | INFO | train rank=0 | step=43380/50000 progress=86.8% loss=0.082968 lr=4.716496e-06 lr_backbone=4.716496e-06 elapsed=19:06:05 avg_step=1.59s recent_step=1.53s eta=02:54:53 +2026-09-21 01:44:20 | INFO | train rank=0 | step=43390/50000 progress=86.8% loss=0.072726 lr=4.702485e-06 lr_backbone=4.702485e-06 elapsed=19:06:20 avg_step=1.59s recent_step=1.49s eta=02:54:37 +2026-09-21 01:44:36 | INFO | train rank=0 | step=43400/50000 progress=86.8% loss=0.083267 lr=4.688494e-06 lr_backbone=4.688494e-06 elapsed=19:06:35 avg_step=1.59s recent_step=1.56s eta=02:54:22 +2026-09-21 01:44:48 | INFO | train rank=0 | step=43410/50000 progress=86.8% loss=0.107250 lr=4.674522e-06 lr_backbone=4.674522e-06 elapsed=19:06:48 avg_step=1.59s recent_step=1.28s eta=02:54:05 +2026-09-21 01:45:03 | INFO | train rank=0 | step=43420/50000 progress=86.8% loss=0.074673 lr=4.660571e-06 lr_backbone=4.660571e-06 elapsed=19:07:03 avg_step=1.59s recent_step=1.47s eta=02:53:49 +2026-09-21 01:45:20 | INFO | train rank=0 | step=43430/50000 progress=86.9% loss=0.080149 lr=4.646639e-06 lr_backbone=4.646639e-06 elapsed=19:07:19 avg_step=1.59s recent_step=1.67s eta=02:53:33 +2026-09-21 01:45:36 | INFO | train rank=0 | step=43440/50000 progress=86.9% loss=0.075930 lr=4.632727e-06 lr_backbone=4.632727e-06 elapsed=19:07:36 avg_step=1.59s recent_step=1.67s eta=02:53:18 +2026-09-21 01:45:52 | INFO | train rank=0 | step=43450/50000 progress=86.9% loss=0.102241 lr=4.618835e-06 lr_backbone=4.618835e-06 elapsed=19:07:51 avg_step=1.59s recent_step=1.51s eta=02:53:02 +2026-09-21 01:46:07 | INFO | train rank=0 | step=43460/50000 progress=86.9% loss=0.066101 lr=4.604963e-06 lr_backbone=4.604963e-06 elapsed=19:08:06 avg_step=1.59s recent_step=1.52s eta=02:52:46 +2026-09-21 01:46:23 | INFO | train rank=0 | step=43470/50000 progress=86.9% loss=0.079436 lr=4.591111e-06 lr_backbone=4.591111e-06 elapsed=19:08:23 avg_step=1.59s recent_step=1.68s eta=02:52:30 +2026-09-21 01:46:44 | INFO | train rank=0 | step=43480/50000 progress=87.0% loss=0.083033 lr=4.577278e-06 lr_backbone=4.577278e-06 elapsed=19:08:43 avg_step=1.59s recent_step=2.01s eta=02:52:15 +2026-09-21 01:46:59 | INFO | train rank=0 | step=43490/50000 progress=87.0% loss=0.161557 lr=4.563466e-06 lr_backbone=4.563466e-06 elapsed=19:08:59 avg_step=1.59s recent_step=1.58s eta=02:51:59 +2026-09-21 01:47:12 | INFO | train rank=0 | step=43500/50000 progress=87.0% loss=0.081040 lr=4.549673e-06 lr_backbone=4.549673e-06 elapsed=19:09:11 avg_step=1.59s recent_step=1.22s eta=02:51:43 +2026-09-21 01:47:26 | INFO | train rank=0 | step=43510/50000 progress=87.0% loss=0.076394 lr=4.535900e-06 lr_backbone=4.535900e-06 elapsed=19:09:25 avg_step=1.59s recent_step=1.42s eta=02:51:27 +2026-09-21 01:47:43 | INFO | train rank=0 | step=43520/50000 progress=87.0% loss=0.076888 lr=4.522148e-06 lr_backbone=4.522148e-06 elapsed=19:09:43 avg_step=1.59s recent_step=1.74s eta=02:51:11 +2026-09-21 01:48:00 | INFO | train rank=0 | step=43530/50000 progress=87.1% loss=0.096424 lr=4.508415e-06 lr_backbone=4.508415e-06 elapsed=19:10:00 avg_step=1.59s recent_step=1.71s eta=02:50:55 +2026-09-21 01:48:14 | INFO | train rank=0 | step=43540/50000 progress=87.1% loss=0.069988 lr=4.494701e-06 lr_backbone=4.494701e-06 elapsed=19:10:14 avg_step=1.59s recent_step=1.42s eta=02:50:39 +2026-09-21 01:48:33 | INFO | train rank=0 | step=43550/50000 progress=87.1% loss=0.062173 lr=4.481008e-06 lr_backbone=4.481008e-06 elapsed=19:10:33 avg_step=1.59s recent_step=1.87s eta=02:50:24 +2026-09-21 01:48:47 | INFO | train rank=0 | step=43560/50000 progress=87.1% loss=0.074104 lr=4.467335e-06 lr_backbone=4.467335e-06 elapsed=19:10:46 avg_step=1.59s recent_step=1.35s eta=02:50:08 +2026-09-21 01:49:05 | INFO | train rank=0 | step=43570/50000 progress=87.1% loss=0.083921 lr=4.453682e-06 lr_backbone=4.453682e-06 elapsed=19:11:04 avg_step=1.59s recent_step=1.83s eta=02:49:52 +2026-09-21 01:49:23 | INFO | train rank=0 | step=43580/50000 progress=87.2% loss=0.105861 lr=4.440048e-06 lr_backbone=4.440048e-06 elapsed=19:11:23 avg_step=1.59s recent_step=1.80s eta=02:49:36 +2026-09-21 01:49:41 | INFO | train rank=0 | step=43590/50000 progress=87.2% loss=0.104638 lr=4.426435e-06 lr_backbone=4.426435e-06 elapsed=19:11:40 avg_step=1.59s recent_step=1.78s eta=02:49:21 +2026-09-21 01:49:52 | INFO | train rank=0 | step=43600/50000 progress=87.2% loss=0.076764 lr=4.412841e-06 lr_backbone=4.412841e-06 elapsed=19:11:52 avg_step=1.59s recent_step=1.15s eta=02:49:04 +2026-09-21 01:50:08 | INFO | train rank=0 | step=43610/50000 progress=87.2% loss=0.080137 lr=4.399267e-06 lr_backbone=4.399267e-06 elapsed=19:12:08 avg_step=1.59s recent_step=1.57s eta=02:48:49 +2026-09-21 01:50:22 | INFO | train rank=0 | step=43620/50000 progress=87.2% loss=0.100033 lr=4.385714e-06 lr_backbone=4.385714e-06 elapsed=19:12:22 avg_step=1.59s recent_step=1.42s eta=02:48:32 +2026-09-21 01:50:36 | INFO | train rank=0 | step=43630/50000 progress=87.3% loss=0.094303 lr=4.372180e-06 lr_backbone=4.372180e-06 elapsed=19:12:36 avg_step=1.59s recent_step=1.42s eta=02:48:16 +2026-09-21 01:50:51 | INFO | train rank=0 | step=43640/50000 progress=87.3% loss=0.086861 lr=4.358666e-06 lr_backbone=4.358666e-06 elapsed=19:12:51 avg_step=1.59s recent_step=1.50s eta=02:48:00 +2026-09-21 01:51:10 | INFO | train rank=0 | step=43650/50000 progress=87.3% loss=0.090714 lr=4.345172e-06 lr_backbone=4.345172e-06 elapsed=19:13:10 avg_step=1.59s recent_step=1.88s eta=02:47:45 +2026-09-21 01:51:27 | INFO | train rank=0 | step=43660/50000 progress=87.3% loss=0.112558 lr=4.331699e-06 lr_backbone=4.331699e-06 elapsed=19:13:26 avg_step=1.59s recent_step=1.66s eta=02:47:29 +2026-09-21 01:51:40 | INFO | train rank=0 | step=43670/50000 progress=87.3% loss=0.086778 lr=4.318245e-06 lr_backbone=4.318245e-06 elapsed=19:13:40 avg_step=1.59s recent_step=1.34s eta=02:47:13 +2026-09-21 01:52:01 | INFO | train rank=0 | step=43680/50000 progress=87.4% loss=0.077810 lr=4.304811e-06 lr_backbone=4.304811e-06 elapsed=19:14:00 avg_step=1.59s recent_step=2.02s eta=02:46:58 +2026-09-21 01:52:14 | INFO | train rank=0 | step=43690/50000 progress=87.4% loss=0.080677 lr=4.291397e-06 lr_backbone=4.291397e-06 elapsed=19:14:14 avg_step=1.59s recent_step=1.37s eta=02:46:42 +2026-09-21 01:52:30 | INFO | train rank=0 | step=43700/50000 progress=87.4% loss=0.107673 lr=4.278003e-06 lr_backbone=4.278003e-06 elapsed=19:14:30 avg_step=1.59s recent_step=1.61s eta=02:46:26 +2026-09-21 01:52:42 | INFO | train rank=0 | step=43710/50000 progress=87.4% loss=0.107674 lr=4.264629e-06 lr_backbone=4.264629e-06 elapsed=19:14:42 avg_step=1.59s recent_step=1.16s eta=02:46:09 +2026-09-21 01:53:01 | INFO | train rank=0 | step=43720/50000 progress=87.4% loss=0.080378 lr=4.251275e-06 lr_backbone=4.251275e-06 elapsed=19:15:01 avg_step=1.59s recent_step=1.91s eta=02:45:54 +2026-09-21 01:53:16 | INFO | train rank=0 | step=43730/50000 progress=87.5% loss=0.126487 lr=4.237941e-06 lr_backbone=4.237941e-06 elapsed=19:15:16 avg_step=1.59s recent_step=1.53s eta=02:45:38 +2026-09-21 01:53:29 | INFO | train rank=0 | step=43740/50000 progress=87.5% loss=0.078807 lr=4.224628e-06 lr_backbone=4.224628e-06 elapsed=19:15:29 avg_step=1.59s recent_step=1.28s eta=02:45:22 +2026-09-21 01:53:47 | INFO | train rank=0 | step=43750/50000 progress=87.5% loss=0.056433 lr=4.211334e-06 lr_backbone=4.211334e-06 elapsed=19:15:46 avg_step=1.59s recent_step=1.76s eta=02:45:06 +2026-09-21 01:54:03 | INFO | train rank=0 | step=43760/50000 progress=87.5% loss=0.051857 lr=4.198060e-06 lr_backbone=4.198060e-06 elapsed=19:16:03 avg_step=1.59s recent_step=1.67s eta=02:44:50 +2026-09-21 01:54:21 | INFO | train rank=0 | step=43770/50000 progress=87.5% loss=0.080449 lr=4.184806e-06 lr_backbone=4.184806e-06 elapsed=19:16:21 avg_step=1.59s recent_step=1.80s eta=02:44:35 +2026-09-21 01:54:34 | INFO | train rank=0 | step=43780/50000 progress=87.6% loss=0.128421 lr=4.171572e-06 lr_backbone=4.171572e-06 elapsed=19:16:34 avg_step=1.59s recent_step=1.24s eta=02:44:19 +2026-09-21 01:54:50 | INFO | train rank=0 | step=43790/50000 progress=87.6% loss=0.085869 lr=4.158359e-06 lr_backbone=4.158359e-06 elapsed=19:16:49 avg_step=1.59s recent_step=1.60s eta=02:44:03 +2026-09-21 01:55:04 | INFO | train rank=0 | step=43800/50000 progress=87.6% loss=0.077753 lr=4.145165e-06 lr_backbone=4.145165e-06 elapsed=19:17:04 avg_step=1.59s recent_step=1.45s eta=02:43:47 +2026-09-21 01:55:22 | INFO | train rank=0 | step=43810/50000 progress=87.6% loss=0.077234 lr=4.131991e-06 lr_backbone=4.131991e-06 elapsed=19:17:22 avg_step=1.59s recent_step=1.78s eta=02:43:31 +2026-09-21 01:55:41 | INFO | train rank=0 | step=43820/50000 progress=87.6% loss=0.114690 lr=4.118838e-06 lr_backbone=4.118838e-06 elapsed=19:17:41 avg_step=1.59s recent_step=1.90s eta=02:43:16 +2026-09-21 01:55:55 | INFO | train rank=0 | step=43830/50000 progress=87.7% loss=0.101338 lr=4.105705e-06 lr_backbone=4.105705e-06 elapsed=19:17:55 avg_step=1.59s recent_step=1.40s eta=02:43:00 +2026-09-21 01:56:07 | INFO | train rank=0 | step=43840/50000 progress=87.7% loss=0.091051 lr=4.092591e-06 lr_backbone=4.092591e-06 elapsed=19:18:07 avg_step=1.59s recent_step=1.18s eta=02:42:43 +2026-09-21 01:56:30 | INFO | train rank=0 | step=43850/50000 progress=87.7% loss=0.093367 lr=4.079498e-06 lr_backbone=4.079498e-06 elapsed=19:18:30 avg_step=1.59s recent_step=2.32s eta=02:42:28 +2026-09-21 01:56:42 | INFO | train rank=0 | step=43860/50000 progress=87.7% loss=0.074488 lr=4.066425e-06 lr_backbone=4.066425e-06 elapsed=19:18:42 avg_step=1.59s recent_step=1.22s eta=02:42:12 +2026-09-21 01:56:55 | INFO | train rank=0 | step=43870/50000 progress=87.7% loss=0.096386 lr=4.053372e-06 lr_backbone=4.053372e-06 elapsed=19:18:55 avg_step=1.59s recent_step=1.28s eta=02:41:56 +2026-09-21 01:57:12 | INFO | train rank=0 | step=43880/50000 progress=87.8% loss=0.082583 lr=4.040339e-06 lr_backbone=4.040339e-06 elapsed=19:19:12 avg_step=1.59s recent_step=1.67s eta=02:41:40 +2026-09-21 01:57:34 | INFO | train rank=0 | step=43890/50000 progress=87.8% loss=0.099308 lr=4.027326e-06 lr_backbone=4.027326e-06 elapsed=19:19:34 avg_step=1.59s recent_step=2.21s eta=02:41:25 +2026-09-21 01:57:47 | INFO | train rank=0 | step=43900/50000 progress=87.8% loss=0.112399 lr=4.014333e-06 lr_backbone=4.014333e-06 elapsed=19:19:47 avg_step=1.59s recent_step=1.34s eta=02:41:09 +2026-09-21 01:58:00 | INFO | train rank=0 | step=43910/50000 progress=87.8% loss=0.108484 lr=4.001360e-06 lr_backbone=4.001360e-06 elapsed=19:20:00 avg_step=1.59s recent_step=1.26s eta=02:40:53 +2026-09-21 01:58:20 | INFO | train rank=0 | step=43920/50000 progress=87.8% loss=0.091767 lr=3.988408e-06 lr_backbone=3.988408e-06 elapsed=19:20:20 avg_step=1.59s recent_step=2.00s eta=02:40:37 +2026-09-21 01:58:35 | INFO | train rank=0 | step=43930/50000 progress=87.9% loss=0.089069 lr=3.975475e-06 lr_backbone=3.975475e-06 elapsed=19:20:35 avg_step=1.59s recent_step=1.50s eta=02:40:21 +2026-09-21 01:58:51 | INFO | train rank=0 | step=43940/50000 progress=87.9% loss=0.070096 lr=3.962563e-06 lr_backbone=3.962563e-06 elapsed=19:20:51 avg_step=1.59s recent_step=1.63s eta=02:40:06 +2026-09-21 01:59:03 | INFO | train rank=0 | step=43950/50000 progress=87.9% loss=0.075664 lr=3.949671e-06 lr_backbone=3.949671e-06 elapsed=19:21:03 avg_step=1.59s recent_step=1.20s eta=02:39:49 +2026-09-21 01:59:16 | INFO | train rank=0 | step=43960/50000 progress=87.9% loss=0.060753 lr=3.936799e-06 lr_backbone=3.936799e-06 elapsed=19:21:16 avg_step=1.58s recent_step=1.31s eta=02:39:33 +2026-09-21 01:59:39 | INFO | train rank=0 | step=43970/50000 progress=87.9% loss=0.090239 lr=3.923947e-06 lr_backbone=3.923947e-06 elapsed=19:21:39 avg_step=1.59s recent_step=2.26s eta=02:39:18 +2026-09-21 01:59:51 | INFO | train rank=0 | step=43980/50000 progress=88.0% loss=0.093988 lr=3.911115e-06 lr_backbone=3.911115e-06 elapsed=19:21:51 avg_step=1.59s recent_step=1.24s eta=02:39:02 +2026-09-21 02:00:04 | INFO | train rank=0 | step=43990/50000 progress=88.0% loss=0.084556 lr=3.898304e-06 lr_backbone=3.898304e-06 elapsed=19:22:03 avg_step=1.58s recent_step=1.21s eta=02:38:45 +2026-09-21 02:00:16 | INFO | train rank=0 | step=44000/50000 progress=88.0% loss=0.081518 lr=3.885512e-06 lr_backbone=3.885512e-06 elapsed=19:22:16 avg_step=1.58s recent_step=1.24s eta=02:38:29 +2026-09-21 02:00:34 | INFO | train rank=0 | step=44010/50000 progress=88.0% loss=0.087272 lr=3.872741e-06 lr_backbone=3.872741e-06 elapsed=19:22:34 avg_step=1.58s recent_step=1.82s eta=02:38:13 +2026-09-21 02:00:51 | INFO | train rank=0 | step=44020/50000 progress=88.0% loss=0.075109 lr=3.859990e-06 lr_backbone=3.859990e-06 elapsed=19:22:51 avg_step=1.58s recent_step=1.69s eta=02:37:58 +2026-09-21 02:01:09 | INFO | train rank=0 | step=44030/50000 progress=88.1% loss=0.085891 lr=3.847259e-06 lr_backbone=3.847259e-06 elapsed=19:23:09 avg_step=1.59s recent_step=1.82s eta=02:37:42 +2026-09-21 02:01:23 | INFO | train rank=0 | step=44040/50000 progress=88.1% loss=0.102062 lr=3.834549e-06 lr_backbone=3.834549e-06 elapsed=19:23:22 avg_step=1.58s recent_step=1.33s eta=02:37:26 +2026-09-21 02:01:40 | INFO | train rank=0 | step=44050/50000 progress=88.1% loss=0.084198 lr=3.821858e-06 lr_backbone=3.821858e-06 elapsed=19:23:39 avg_step=1.59s recent_step=1.73s eta=02:37:10 +2026-09-21 02:01:55 | INFO | train rank=0 | step=44060/50000 progress=88.1% loss=0.099926 lr=3.809188e-06 lr_backbone=3.809188e-06 elapsed=19:23:54 avg_step=1.58s recent_step=1.47s eta=02:36:54 +2026-09-21 02:02:10 | INFO | train rank=0 | step=44070/50000 progress=88.1% loss=0.070572 lr=3.796538e-06 lr_backbone=3.796538e-06 elapsed=19:24:10 avg_step=1.58s recent_step=1.55s eta=02:36:38 +2026-09-21 02:02:23 | INFO | train rank=0 | step=44080/50000 progress=88.2% loss=0.076288 lr=3.783908e-06 lr_backbone=3.783908e-06 elapsed=19:24:23 avg_step=1.58s recent_step=1.31s eta=02:36:22 +2026-09-21 02:02:45 | INFO | train rank=0 | step=44090/50000 progress=88.2% loss=0.067185 lr=3.771298e-06 lr_backbone=3.771298e-06 elapsed=19:24:44 avg_step=1.59s recent_step=2.13s eta=02:36:07 +2026-09-21 02:02:57 | INFO | train rank=0 | step=44100/50000 progress=88.2% loss=0.070145 lr=3.758709e-06 lr_backbone=3.758709e-06 elapsed=19:24:56 avg_step=1.58s recent_step=1.23s eta=02:35:51 +2026-09-21 02:03:09 | INFO | train rank=0 | step=44110/50000 progress=88.2% loss=0.075584 lr=3.746140e-06 lr_backbone=3.746140e-06 elapsed=19:25:09 avg_step=1.58s recent_step=1.22s eta=02:35:34 +2026-09-21 02:03:21 | INFO | train rank=0 | step=44120/50000 progress=88.2% loss=0.073729 lr=3.733591e-06 lr_backbone=3.733591e-06 elapsed=19:25:21 avg_step=1.58s recent_step=1.23s eta=02:35:18 +2026-09-21 02:03:40 | INFO | train rank=0 | step=44130/50000 progress=88.3% loss=0.089116 lr=3.721062e-06 lr_backbone=3.721062e-06 elapsed=19:25:39 avg_step=1.58s recent_step=1.82s eta=02:35:03 +2026-09-21 02:03:57 | INFO | train rank=0 | step=44140/50000 progress=88.3% loss=0.091083 lr=3.708554e-06 lr_backbone=3.708554e-06 elapsed=19:25:57 avg_step=1.58s recent_step=1.75s eta=02:34:47 +2026-09-21 02:04:13 | INFO | train rank=0 | step=44150/50000 progress=88.3% loss=0.083250 lr=3.696065e-06 lr_backbone=3.696065e-06 elapsed=19:26:13 avg_step=1.58s recent_step=1.59s eta=02:34:31 +2026-09-21 02:04:32 | INFO | train rank=0 | step=44160/50000 progress=88.3% loss=0.091363 lr=3.683597e-06 lr_backbone=3.683597e-06 elapsed=19:26:32 avg_step=1.58s recent_step=1.93s eta=02:34:16 +2026-09-21 02:04:53 | INFO | train rank=0 | step=44170/50000 progress=88.3% loss=0.085436 lr=3.671150e-06 lr_backbone=3.671150e-06 elapsed=19:26:53 avg_step=1.59s recent_step=2.09s eta=02:34:01 +2026-09-21 02:05:05 | INFO | train rank=0 | step=44180/50000 progress=88.4% loss=0.085194 lr=3.658722e-06 lr_backbone=3.658722e-06 elapsed=19:27:05 avg_step=1.59s recent_step=1.20s eta=02:33:44 +2026-09-21 02:05:18 | INFO | train rank=0 | step=44190/50000 progress=88.4% loss=0.068675 lr=3.646315e-06 lr_backbone=3.646315e-06 elapsed=19:27:18 avg_step=1.58s recent_step=1.30s eta=02:33:28 +2026-09-21 02:05:35 | INFO | train rank=0 | step=44200/50000 progress=88.4% loss=0.070959 lr=3.633928e-06 lr_backbone=3.633928e-06 elapsed=19:27:35 avg_step=1.58s recent_step=1.70s eta=02:33:12 +2026-09-21 02:05:48 | INFO | train rank=0 | step=44210/50000 progress=88.4% loss=0.080159 lr=3.621562e-06 lr_backbone=3.621562e-06 elapsed=19:27:47 avg_step=1.58s recent_step=1.24s eta=02:32:56 +2026-09-21 02:06:02 | INFO | train rank=0 | step=44220/50000 progress=88.4% loss=0.087149 lr=3.609215e-06 lr_backbone=3.609215e-06 elapsed=19:28:01 avg_step=1.58s recent_step=1.39s eta=02:32:40 +2026-09-21 02:06:21 | INFO | train rank=0 | step=44230/50000 progress=88.5% loss=0.087083 lr=3.596889e-06 lr_backbone=3.596889e-06 elapsed=19:28:21 avg_step=1.58s recent_step=1.95s eta=02:32:25 +2026-09-21 02:06:34 | INFO | train rank=0 | step=44240/50000 progress=88.5% loss=0.069913 lr=3.584583e-06 lr_backbone=3.584583e-06 elapsed=19:28:33 avg_step=1.58s recent_step=1.25s eta=02:32:08 +2026-09-21 02:06:49 | INFO | train rank=0 | step=44250/50000 progress=88.5% loss=0.075383 lr=3.572298e-06 lr_backbone=3.572298e-06 elapsed=19:28:49 avg_step=1.58s recent_step=1.57s eta=02:31:52 +2026-09-21 02:07:05 | INFO | train rank=0 | step=44260/50000 progress=88.5% loss=0.073388 lr=3.560033e-06 lr_backbone=3.560033e-06 elapsed=19:29:04 avg_step=1.58s recent_step=1.53s eta=02:31:36 +2026-09-21 02:07:17 | INFO | train rank=0 | step=44270/50000 progress=88.5% loss=0.079903 lr=3.547788e-06 lr_backbone=3.547788e-06 elapsed=19:29:16 avg_step=1.58s recent_step=1.21s eta=02:31:20 +2026-09-21 02:07:35 | INFO | train rank=0 | step=44280/50000 progress=88.6% loss=0.100243 lr=3.535564e-06 lr_backbone=3.535564e-06 elapsed=19:29:35 avg_step=1.58s recent_step=1.85s eta=02:31:05 +2026-09-21 02:07:51 | INFO | train rank=0 | step=44290/50000 progress=88.6% loss=0.064194 lr=3.523360e-06 lr_backbone=3.523360e-06 elapsed=19:29:51 avg_step=1.58s recent_step=1.60s eta=02:30:49 +2026-09-21 02:08:08 | INFO | train rank=0 | step=44300/50000 progress=88.6% loss=0.084687 lr=3.511176e-06 lr_backbone=3.511176e-06 elapsed=19:30:08 avg_step=1.58s recent_step=1.71s eta=02:30:33 +2026-09-21 02:08:27 | INFO | train rank=0 | step=44310/50000 progress=88.6% loss=0.091134 lr=3.499012e-06 lr_backbone=3.499012e-06 elapsed=19:30:27 avg_step=1.58s recent_step=1.91s eta=02:30:18 +2026-09-21 02:08:41 | INFO | train rank=0 | step=44320/50000 progress=88.6% loss=0.072693 lr=3.486869e-06 lr_backbone=3.486869e-06 elapsed=19:30:41 avg_step=1.58s recent_step=1.38s eta=02:30:02 +2026-09-21 02:09:37 | INFO | train rank=0 | step=44330/50000 progress=88.7% loss=0.082412 lr=3.474746e-06 lr_backbone=3.474746e-06 elapsed=19:31:36 avg_step=1.59s recent_step=5.55s eta=02:29:51 +2026-09-21 02:09:49 | INFO | train rank=0 | step=44340/50000 progress=88.7% loss=0.098378 lr=3.462644e-06 lr_backbone=3.462644e-06 elapsed=19:31:48 avg_step=1.59s recent_step=1.21s eta=02:29:34 +2026-09-21 02:10:02 | INFO | train rank=0 | step=44350/50000 progress=88.7% loss=0.081052 lr=3.450562e-06 lr_backbone=3.450562e-06 elapsed=19:32:01 avg_step=1.59s recent_step=1.30s eta=02:29:18 +2026-09-21 02:10:20 | INFO | train rank=0 | step=44360/50000 progress=88.7% loss=0.156839 lr=3.438500e-06 lr_backbone=3.438500e-06 elapsed=19:32:19 avg_step=1.59s recent_step=1.79s eta=02:29:03 +2026-09-21 02:10:32 | INFO | train rank=0 | step=44370/50000 progress=88.7% loss=0.082841 lr=3.426459e-06 lr_backbone=3.426459e-06 elapsed=19:32:32 avg_step=1.59s recent_step=1.25s eta=02:28:46 +2026-09-21 02:10:50 | INFO | train rank=0 | step=44380/50000 progress=88.8% loss=0.081676 lr=3.414438e-06 lr_backbone=3.414438e-06 elapsed=19:32:50 avg_step=1.59s recent_step=1.80s eta=02:28:31 +2026-09-21 02:11:04 | INFO | train rank=0 | step=44390/50000 progress=88.8% loss=0.086756 lr=3.402437e-06 lr_backbone=3.402437e-06 elapsed=19:33:03 avg_step=1.59s recent_step=1.34s eta=02:28:15 +2026-09-21 02:11:16 | INFO | train rank=0 | step=44400/50000 progress=88.8% loss=0.099450 lr=3.390457e-06 lr_backbone=3.390457e-06 elapsed=19:33:16 avg_step=1.59s recent_step=1.24s eta=02:27:58 +2026-09-21 02:11:32 | INFO | train rank=0 | step=44410/50000 progress=88.8% loss=0.086707 lr=3.378497e-06 lr_backbone=3.378497e-06 elapsed=19:33:31 avg_step=1.59s recent_step=1.59s eta=02:27:42 +2026-09-21 02:11:51 | INFO | train rank=0 | step=44420/50000 progress=88.8% loss=0.078369 lr=3.366558e-06 lr_backbone=3.366558e-06 elapsed=19:33:50 avg_step=1.59s recent_step=1.89s eta=02:27:27 +2026-09-21 02:12:07 | INFO | train rank=0 | step=44430/50000 progress=88.9% loss=0.069313 lr=3.354638e-06 lr_backbone=3.354638e-06 elapsed=19:34:07 avg_step=1.59s recent_step=1.66s eta=02:27:11 +2026-09-21 02:12:20 | INFO | train rank=0 | step=44440/50000 progress=88.9% loss=0.106334 lr=3.342740e-06 lr_backbone=3.342740e-06 elapsed=19:34:19 avg_step=1.59s recent_step=1.23s eta=02:26:55 +2026-09-21 02:12:35 | INFO | train rank=0 | step=44450/50000 progress=88.9% loss=0.069531 lr=3.330862e-06 lr_backbone=3.330862e-06 elapsed=19:34:34 avg_step=1.59s recent_step=1.51s eta=02:26:39 +2026-09-21 02:12:50 | INFO | train rank=0 | step=44460/50000 progress=88.9% loss=0.073435 lr=3.319004e-06 lr_backbone=3.319004e-06 elapsed=19:34:49 avg_step=1.59s recent_step=1.52s eta=02:26:23 +2026-09-21 02:13:08 | INFO | train rank=0 | step=44470/50000 progress=88.9% loss=0.110276 lr=3.307166e-06 lr_backbone=3.307166e-06 elapsed=19:35:07 avg_step=1.59s recent_step=1.79s eta=02:26:07 +2026-09-21 02:13:23 | INFO | train rank=0 | step=44480/50000 progress=89.0% loss=0.101106 lr=3.295349e-06 lr_backbone=3.295349e-06 elapsed=19:35:23 avg_step=1.59s recent_step=1.53s eta=02:25:51 +2026-09-21 02:13:37 | INFO | train rank=0 | step=44490/50000 progress=89.0% loss=0.092790 lr=3.283553e-06 lr_backbone=3.283553e-06 elapsed=19:35:36 avg_step=1.59s recent_step=1.38s eta=02:25:35 +2026-09-21 02:13:55 | INFO | train rank=0 | step=44500/50000 progress=89.0% loss=0.081245 lr=3.271777e-06 lr_backbone=3.271777e-06 elapsed=19:35:55 avg_step=1.59s recent_step=1.83s eta=02:25:20 +2026-09-21 02:14:07 | INFO | train rank=0 | step=44510/50000 progress=89.0% loss=0.113472 lr=3.260021e-06 lr_backbone=3.260021e-06 elapsed=19:36:07 avg_step=1.59s recent_step=1.19s eta=02:25:03 +2026-09-21 02:14:25 | INFO | train rank=0 | step=44520/50000 progress=89.0% loss=0.093662 lr=3.248286e-06 lr_backbone=3.248286e-06 elapsed=19:36:24 avg_step=1.59s recent_step=1.74s eta=02:24:48 +2026-09-21 02:14:40 | INFO | train rank=0 | step=44530/50000 progress=89.1% loss=0.086241 lr=3.236571e-06 lr_backbone=3.236571e-06 elapsed=19:36:40 avg_step=1.59s recent_step=1.56s eta=02:24:32 +2026-09-21 02:14:57 | INFO | train rank=0 | step=44540/50000 progress=89.1% loss=0.089196 lr=3.224877e-06 lr_backbone=3.224877e-06 elapsed=19:36:57 avg_step=1.59s recent_step=1.69s eta=02:24:16 +2026-09-21 02:15:13 | INFO | train rank=0 | step=44550/50000 progress=89.1% loss=0.071289 lr=3.213203e-06 lr_backbone=3.213203e-06 elapsed=19:37:13 avg_step=1.59s recent_step=1.60s eta=02:24:00 +2026-09-21 02:15:28 | INFO | train rank=0 | step=44560/50000 progress=89.1% loss=0.069835 lr=3.201550e-06 lr_backbone=3.201550e-06 elapsed=19:37:27 avg_step=1.59s recent_step=1.46s eta=02:23:44 +2026-09-21 02:15:44 | INFO | train rank=0 | step=44570/50000 progress=89.1% loss=0.069903 lr=3.189917e-06 lr_backbone=3.189917e-06 elapsed=19:37:44 avg_step=1.59s recent_step=1.69s eta=02:23:29 +2026-09-21 02:15:56 | INFO | train rank=0 | step=44580/50000 progress=89.2% loss=0.080662 lr=3.178304e-06 lr_backbone=3.178304e-06 elapsed=19:37:56 avg_step=1.59s recent_step=1.18s eta=02:23:12 +2026-09-21 02:16:16 | INFO | train rank=0 | step=44590/50000 progress=89.2% loss=0.073609 lr=3.166712e-06 lr_backbone=3.166712e-06 elapsed=19:38:15 avg_step=1.59s recent_step=1.93s eta=02:22:57 +2026-09-21 02:16:28 | INFO | train rank=0 | step=44600/50000 progress=89.2% loss=0.095225 lr=3.155141e-06 lr_backbone=3.155141e-06 elapsed=19:38:28 avg_step=1.59s recent_step=1.24s eta=02:22:41 +2026-09-21 02:16:42 | INFO | train rank=0 | step=44610/50000 progress=89.2% loss=0.067887 lr=3.143590e-06 lr_backbone=3.143590e-06 elapsed=19:38:42 avg_step=1.59s recent_step=1.45s eta=02:22:25 +2026-09-21 02:17:01 | INFO | train rank=0 | step=44620/50000 progress=89.2% loss=0.082102 lr=3.132059e-06 lr_backbone=3.132059e-06 elapsed=19:39:00 avg_step=1.59s recent_step=1.82s eta=02:22:09 +2026-09-21 02:17:18 | INFO | train rank=0 | step=44630/50000 progress=89.3% loss=0.085566 lr=3.120549e-06 lr_backbone=3.120549e-06 elapsed=19:39:18 avg_step=1.59s recent_step=1.78s eta=02:21:53 +2026-09-21 02:17:33 | INFO | train rank=0 | step=44640/50000 progress=89.3% loss=0.104305 lr=3.109060e-06 lr_backbone=3.109060e-06 elapsed=19:39:33 avg_step=1.59s recent_step=1.46s eta=02:21:37 +2026-09-21 02:17:51 | INFO | train rank=0 | step=44650/50000 progress=89.3% loss=0.081644 lr=3.097591e-06 lr_backbone=3.097591e-06 elapsed=19:39:51 avg_step=1.59s recent_step=1.82s eta=02:21:22 +2026-09-21 02:18:04 | INFO | train rank=0 | step=44660/50000 progress=89.3% loss=0.098715 lr=3.086142e-06 lr_backbone=3.086142e-06 elapsed=19:40:03 avg_step=1.59s recent_step=1.27s eta=02:21:06 +2026-09-21 02:18:22 | INFO | train rank=0 | step=44670/50000 progress=89.3% loss=0.078378 lr=3.074715e-06 lr_backbone=3.074715e-06 elapsed=19:40:22 avg_step=1.59s recent_step=1.80s eta=02:20:50 +2026-09-21 02:18:40 | INFO | train rank=0 | step=44680/50000 progress=89.4% loss=0.074501 lr=3.063307e-06 lr_backbone=3.063307e-06 elapsed=19:40:40 avg_step=1.59s recent_step=1.86s eta=02:20:34 +2026-09-21 02:18:52 | INFO | train rank=0 | step=44690/50000 progress=89.4% loss=0.068889 lr=3.051920e-06 lr_backbone=3.051920e-06 elapsed=19:40:52 avg_step=1.59s recent_step=1.20s eta=02:20:18 +2026-09-21 02:19:06 | INFO | train rank=0 | step=44700/50000 progress=89.4% loss=0.077903 lr=3.040554e-06 lr_backbone=3.040554e-06 elapsed=19:41:06 avg_step=1.59s recent_step=1.40s eta=02:20:02 +2026-09-21 02:19:25 | INFO | train rank=0 | step=44710/50000 progress=89.4% loss=0.086647 lr=3.029208e-06 lr_backbone=3.029208e-06 elapsed=19:41:25 avg_step=1.59s recent_step=1.90s eta=02:19:47 +2026-09-21 02:19:38 | INFO | train rank=0 | step=44720/50000 progress=89.4% loss=0.074842 lr=3.017883e-06 lr_backbone=3.017883e-06 elapsed=19:41:37 avg_step=1.59s recent_step=1.23s eta=02:19:30 +2026-09-21 02:19:57 | INFO | train rank=0 | step=44730/50000 progress=89.5% loss=0.072045 lr=3.006578e-06 lr_backbone=3.006578e-06 elapsed=19:41:57 avg_step=1.59s recent_step=1.93s eta=02:19:15 +2026-09-21 02:20:13 | INFO | train rank=0 | step=44740/50000 progress=89.5% loss=0.119029 lr=2.995294e-06 lr_backbone=2.995294e-06 elapsed=19:42:12 avg_step=1.59s recent_step=1.56s eta=02:18:59 +2026-09-21 02:20:27 | INFO | train rank=0 | step=44750/50000 progress=89.5% loss=0.070888 lr=2.984030e-06 lr_backbone=2.984030e-06 elapsed=19:42:26 avg_step=1.59s recent_step=1.40s eta=02:18:43 +2026-09-21 02:20:42 | INFO | train rank=0 | step=44760/50000 progress=89.5% loss=0.067955 lr=2.972787e-06 lr_backbone=2.972787e-06 elapsed=19:42:42 avg_step=1.59s recent_step=1.56s eta=02:18:27 +2026-09-21 02:20:57 | INFO | train rank=0 | step=44770/50000 progress=89.5% loss=0.114460 lr=2.961565e-06 lr_backbone=2.961565e-06 elapsed=19:42:57 avg_step=1.59s recent_step=1.50s eta=02:18:11 +2026-09-21 02:21:15 | INFO | train rank=0 | step=44780/50000 progress=89.6% loss=0.108052 lr=2.950363e-06 lr_backbone=2.950363e-06 elapsed=19:43:14 avg_step=1.59s recent_step=1.77s eta=02:17:55 +2026-09-21 02:21:31 | INFO | train rank=0 | step=44790/50000 progress=89.6% loss=0.083869 lr=2.939182e-06 lr_backbone=2.939182e-06 elapsed=19:43:31 avg_step=1.59s recent_step=1.65s eta=02:17:40 +2026-09-21 02:21:44 | INFO | train rank=0 | step=44800/50000 progress=89.6% loss=0.130976 lr=2.928021e-06 lr_backbone=2.928021e-06 elapsed=19:43:44 avg_step=1.59s recent_step=1.31s eta=02:17:23 +2026-09-21 02:22:06 | INFO | train rank=0 | step=44810/50000 progress=89.6% loss=0.073135 lr=2.916881e-06 lr_backbone=2.916881e-06 elapsed=19:44:05 avg_step=1.59s recent_step=2.13s eta=02:17:08 +2026-09-21 02:22:18 | INFO | train rank=0 | step=44820/50000 progress=89.6% loss=0.079441 lr=2.905762e-06 lr_backbone=2.905762e-06 elapsed=19:44:18 avg_step=1.59s recent_step=1.22s eta=02:16:52 +2026-09-21 02:22:30 | INFO | train rank=0 | step=44830/50000 progress=89.7% loss=0.079890 lr=2.894663e-06 lr_backbone=2.894663e-06 elapsed=19:44:30 avg_step=1.59s recent_step=1.23s eta=02:16:36 +2026-09-21 02:22:50 | INFO | train rank=0 | step=44840/50000 progress=89.7% loss=0.073081 lr=2.883584e-06 lr_backbone=2.883584e-06 elapsed=19:44:50 avg_step=1.59s recent_step=2.00s eta=02:16:20 +2026-09-21 02:23:06 | INFO | train rank=0 | step=44850/50000 progress=89.7% loss=0.081975 lr=2.872527e-06 lr_backbone=2.872527e-06 elapsed=19:45:06 avg_step=1.59s recent_step=1.57s eta=02:16:04 +2026-09-21 02:23:26 | INFO | train rank=0 | step=44860/50000 progress=89.7% loss=0.075593 lr=2.861490e-06 lr_backbone=2.861490e-06 elapsed=19:45:26 avg_step=1.59s recent_step=2.03s eta=02:15:49 +2026-09-21 02:23:38 | INFO | train rank=0 | step=44870/50000 progress=89.7% loss=0.091118 lr=2.850473e-06 lr_backbone=2.850473e-06 elapsed=19:45:38 avg_step=1.59s recent_step=1.20s eta=02:15:33 +2026-09-21 02:23:51 | INFO | train rank=0 | step=44880/50000 progress=89.8% loss=0.088457 lr=2.839477e-06 lr_backbone=2.839477e-06 elapsed=19:45:50 avg_step=1.59s recent_step=1.23s eta=02:15:17 +2026-09-21 02:24:11 | INFO | train rank=0 | step=44890/50000 progress=89.8% loss=0.104184 lr=2.828502e-06 lr_backbone=2.828502e-06 elapsed=19:46:11 avg_step=1.59s recent_step=2.04s eta=02:15:01 +2026-09-21 02:24:27 | INFO | train rank=0 | step=44900/50000 progress=89.8% loss=0.079839 lr=2.817548e-06 lr_backbone=2.817548e-06 elapsed=19:46:27 avg_step=1.59s recent_step=1.65s eta=02:14:45 +2026-09-21 02:24:40 | INFO | train rank=0 | step=44910/50000 progress=89.8% loss=0.066456 lr=2.806614e-06 lr_backbone=2.806614e-06 elapsed=19:46:39 avg_step=1.59s recent_step=1.23s eta=02:14:29 +2026-09-21 02:24:54 | INFO | train rank=0 | step=44920/50000 progress=89.8% loss=0.066326 lr=2.795700e-06 lr_backbone=2.795700e-06 elapsed=19:46:54 avg_step=1.59s recent_step=1.47s eta=02:14:13 +2026-09-21 02:25:15 | INFO | train rank=0 | step=44930/50000 progress=89.9% loss=0.069351 lr=2.784808e-06 lr_backbone=2.784808e-06 elapsed=19:47:15 avg_step=1.59s recent_step=2.05s eta=02:13:58 +2026-09-21 02:25:28 | INFO | train rank=0 | step=44940/50000 progress=89.9% loss=0.092483 lr=2.773936e-06 lr_backbone=2.773936e-06 elapsed=19:47:28 avg_step=1.59s recent_step=1.30s eta=02:13:42 +2026-09-21 02:25:46 | INFO | train rank=0 | step=44950/50000 progress=89.9% loss=0.083178 lr=2.763084e-06 lr_backbone=2.763084e-06 elapsed=19:47:46 avg_step=1.59s recent_step=1.82s eta=02:13:26 +2026-09-21 02:26:02 | INFO | train rank=0 | step=44960/50000 progress=89.9% loss=0.090394 lr=2.752254e-06 lr_backbone=2.752254e-06 elapsed=19:48:02 avg_step=1.59s recent_step=1.59s eta=02:13:10 +2026-09-21 02:26:16 | INFO | train rank=0 | step=44970/50000 progress=89.9% loss=0.081334 lr=2.741444e-06 lr_backbone=2.741444e-06 elapsed=19:48:16 avg_step=1.59s recent_step=1.38s eta=02:12:54 +2026-09-21 02:26:32 | INFO | train rank=0 | step=44980/50000 progress=90.0% loss=0.084430 lr=2.730655e-06 lr_backbone=2.730655e-06 elapsed=19:48:32 avg_step=1.59s recent_step=1.66s eta=02:12:38 +2026-09-21 02:26:45 | INFO | train rank=0 | step=44990/50000 progress=90.0% loss=0.083536 lr=2.719886e-06 lr_backbone=2.719886e-06 elapsed=19:48:44 avg_step=1.59s recent_step=1.21s eta=02:12:22 +2026-09-21 02:26:58 | INFO | train rank=0 | step=45000/50000 progress=90.0% loss=0.074771 lr=2.709138e-06 lr_backbone=2.709138e-06 elapsed=19:48:58 avg_step=1.59s recent_step=1.37s eta=02:12:06 +2026-09-21 02:27:19 | INFO | train rank=0 | step=45010/50000 progress=90.0% loss=0.093910 lr=2.698411e-06 lr_backbone=2.698411e-06 elapsed=19:49:19 avg_step=1.59s recent_step=2.08s eta=02:11:51 +2026-09-21 02:27:37 | INFO | train rank=0 | step=45020/50000 progress=90.0% loss=0.088127 lr=2.687704e-06 lr_backbone=2.687704e-06 elapsed=19:49:37 avg_step=1.59s recent_step=1.81s eta=02:11:35 +2026-09-21 02:27:49 | INFO | train rank=0 | step=45030/50000 progress=90.1% loss=0.083328 lr=2.677018e-06 lr_backbone=2.677018e-06 elapsed=19:49:49 avg_step=1.59s recent_step=1.19s eta=02:11:19 +2026-09-21 02:28:02 | INFO | train rank=0 | step=45040/50000 progress=90.1% loss=0.106152 lr=2.666353e-06 lr_backbone=2.666353e-06 elapsed=19:50:02 avg_step=1.59s recent_step=1.33s eta=02:11:03 +2026-09-21 02:28:18 | INFO | train rank=0 | step=45050/50000 progress=90.1% loss=0.078890 lr=2.655709e-06 lr_backbone=2.655709e-06 elapsed=19:50:18 avg_step=1.59s recent_step=1.62s eta=02:10:47 +2026-09-21 02:28:35 | INFO | train rank=0 | step=45060/50000 progress=90.1% loss=0.074052 lr=2.645085e-06 lr_backbone=2.645085e-06 elapsed=19:50:35 avg_step=1.59s recent_step=1.69s eta=02:10:31 +2026-09-21 02:28:54 | INFO | train rank=0 | step=45070/50000 progress=90.1% loss=0.106072 lr=2.634482e-06 lr_backbone=2.634482e-06 elapsed=19:50:54 avg_step=1.59s recent_step=1.91s eta=02:10:16 +2026-09-21 02:29:10 | INFO | train rank=0 | step=45080/50000 progress=90.2% loss=0.086360 lr=2.623899e-06 lr_backbone=2.623899e-06 elapsed=19:51:10 avg_step=1.59s recent_step=1.61s eta=02:10:00 +2026-09-21 02:29:22 | INFO | train rank=0 | step=45090/50000 progress=90.2% loss=0.076549 lr=2.613338e-06 lr_backbone=2.613338e-06 elapsed=19:51:22 avg_step=1.59s recent_step=1.17s eta=02:09:43 +2026-09-21 02:29:41 | INFO | train rank=0 | step=45100/50000 progress=90.2% loss=0.099917 lr=2.602797e-06 lr_backbone=2.602797e-06 elapsed=19:51:41 avg_step=1.59s recent_step=1.89s eta=02:09:28 +2026-09-21 02:29:59 | INFO | train rank=0 | step=45110/50000 progress=90.2% loss=0.082986 lr=2.592277e-06 lr_backbone=2.592277e-06 elapsed=19:51:58 avg_step=1.59s recent_step=1.75s eta=02:09:12 +2026-09-21 02:30:10 | INFO | train rank=0 | step=45120/50000 progress=90.2% loss=0.123504 lr=2.581777e-06 lr_backbone=2.581777e-06 elapsed=19:52:10 avg_step=1.59s recent_step=1.17s eta=02:08:56 +2026-09-21 02:30:22 | INFO | train rank=0 | step=45130/50000 progress=90.3% loss=0.085142 lr=2.571299e-06 lr_backbone=2.571299e-06 elapsed=19:52:22 avg_step=1.59s recent_step=1.20s eta=02:08:40 +2026-09-21 02:30:42 | INFO | train rank=0 | step=45140/50000 progress=90.3% loss=0.067919 lr=2.560841e-06 lr_backbone=2.560841e-06 elapsed=19:52:41 avg_step=1.59s recent_step=1.93s eta=02:08:24 +2026-09-21 02:30:54 | INFO | train rank=0 | step=45150/50000 progress=90.3% loss=0.078080 lr=2.550404e-06 lr_backbone=2.550404e-06 elapsed=19:52:54 avg_step=1.59s recent_step=1.23s eta=02:08:08 +2026-09-21 02:31:08 | INFO | train rank=0 | step=45160/50000 progress=90.3% loss=0.071447 lr=2.539987e-06 lr_backbone=2.539987e-06 elapsed=19:53:07 avg_step=1.59s recent_step=1.37s eta=02:07:52 +2026-09-21 02:31:26 | INFO | train rank=0 | step=45170/50000 progress=90.3% loss=0.081361 lr=2.529591e-06 lr_backbone=2.529591e-06 elapsed=19:53:26 avg_step=1.59s recent_step=1.86s eta=02:07:36 +2026-09-21 02:31:46 | INFO | train rank=0 | step=45180/50000 progress=90.4% loss=0.091164 lr=2.519217e-06 lr_backbone=2.519217e-06 elapsed=19:53:46 avg_step=1.59s recent_step=1.98s eta=02:07:21 +2026-09-21 02:31:58 | INFO | train rank=0 | step=45190/50000 progress=90.4% loss=0.067147 lr=2.508862e-06 lr_backbone=2.508862e-06 elapsed=19:53:58 avg_step=1.59s recent_step=1.19s eta=02:07:05 +2026-09-21 02:32:17 | INFO | train rank=0 | step=45200/50000 progress=90.4% loss=0.077501 lr=2.498529e-06 lr_backbone=2.498529e-06 elapsed=19:54:16 avg_step=1.59s recent_step=1.87s eta=02:06:49 +2026-09-21 02:32:33 | INFO | train rank=0 | step=45210/50000 progress=90.4% loss=0.080974 lr=2.488217e-06 lr_backbone=2.488217e-06 elapsed=19:54:33 avg_step=1.59s recent_step=1.63s eta=02:06:33 +2026-09-21 02:32:50 | INFO | train rank=0 | step=45220/50000 progress=90.4% loss=0.075709 lr=2.477925e-06 lr_backbone=2.477925e-06 elapsed=19:54:50 avg_step=1.59s recent_step=1.72s eta=02:06:18 +2026-09-21 02:33:05 | INFO | train rank=0 | step=45230/50000 progress=90.5% loss=0.077036 lr=2.467654e-06 lr_backbone=2.467654e-06 elapsed=19:55:05 avg_step=1.59s recent_step=1.51s eta=02:06:02 +2026-09-21 02:33:18 | INFO | train rank=0 | step=45240/50000 progress=90.5% loss=0.085889 lr=2.457404e-06 lr_backbone=2.457404e-06 elapsed=19:55:18 avg_step=1.59s recent_step=1.29s eta=02:05:45 +2026-09-21 02:33:37 | INFO | train rank=0 | step=45250/50000 progress=90.5% loss=0.101944 lr=2.447174e-06 lr_backbone=2.447174e-06 elapsed=19:55:37 avg_step=1.59s recent_step=1.92s eta=02:05:30 +2026-09-21 02:33:54 | INFO | train rank=0 | step=45260/50000 progress=90.5% loss=0.092302 lr=2.436966e-06 lr_backbone=2.436966e-06 elapsed=19:55:54 avg_step=1.59s recent_step=1.67s eta=02:05:14 +2026-09-21 02:34:07 | INFO | train rank=0 | step=45270/50000 progress=90.5% loss=0.074020 lr=2.426778e-06 lr_backbone=2.426778e-06 elapsed=19:56:06 avg_step=1.59s recent_step=1.25s eta=02:04:58 +2026-09-21 02:34:24 | INFO | train rank=0 | step=45280/50000 progress=90.6% loss=0.094662 lr=2.416611e-06 lr_backbone=2.416611e-06 elapsed=19:56:24 avg_step=1.59s recent_step=1.74s eta=02:04:42 +2026-09-21 02:34:42 | INFO | train rank=0 | step=45290/50000 progress=90.6% loss=0.072731 lr=2.406465e-06 lr_backbone=2.406465e-06 elapsed=19:56:41 avg_step=1.59s recent_step=1.76s eta=02:04:27 +2026-09-21 02:34:55 | INFO | train rank=0 | step=45300/50000 progress=90.6% loss=0.083034 lr=2.396339e-06 lr_backbone=2.396339e-06 elapsed=19:56:55 avg_step=1.59s recent_step=1.36s eta=02:04:11 +2026-09-21 02:35:13 | INFO | train rank=0 | step=45310/50000 progress=90.6% loss=0.079359 lr=2.386235e-06 lr_backbone=2.386235e-06 elapsed=19:57:13 avg_step=1.59s recent_step=1.83s eta=02:03:55 +2026-09-21 02:35:26 | INFO | train rank=0 | step=45320/50000 progress=90.6% loss=0.094091 lr=2.376151e-06 lr_backbone=2.376151e-06 elapsed=19:57:26 avg_step=1.59s recent_step=1.26s eta=02:03:39 +2026-09-21 02:35:44 | INFO | train rank=0 | step=45330/50000 progress=90.7% loss=0.092434 lr=2.366088e-06 lr_backbone=2.366088e-06 elapsed=19:57:43 avg_step=1.59s recent_step=1.78s eta=02:03:23 +2026-09-21 02:35:55 | INFO | train rank=0 | step=45340/50000 progress=90.7% loss=0.063660 lr=2.356046e-06 lr_backbone=2.356046e-06 elapsed=19:57:55 avg_step=1.59s recent_step=1.17s eta=02:03:07 +2026-09-21 02:36:13 | INFO | train rank=0 | step=45350/50000 progress=90.7% loss=0.069862 lr=2.346025e-06 lr_backbone=2.346025e-06 elapsed=19:58:13 avg_step=1.59s recent_step=1.75s eta=02:02:51 +2026-09-21 02:36:32 | INFO | train rank=0 | step=45360/50000 progress=90.7% loss=0.111070 lr=2.336025e-06 lr_backbone=2.336025e-06 elapsed=19:58:31 avg_step=1.59s recent_step=1.88s eta=02:02:36 +2026-09-21 02:36:44 | INFO | train rank=0 | step=45370/50000 progress=90.7% loss=0.069438 lr=2.326045e-06 lr_backbone=2.326045e-06 elapsed=19:58:44 avg_step=1.59s recent_step=1.26s eta=02:02:19 +2026-09-21 02:36:57 | INFO | train rank=0 | step=45380/50000 progress=90.8% loss=0.080743 lr=2.316086e-06 lr_backbone=2.316086e-06 elapsed=19:58:57 avg_step=1.59s recent_step=1.25s eta=02:02:03 +2026-09-21 02:37:12 | INFO | train rank=0 | step=45390/50000 progress=90.8% loss=0.092445 lr=2.306149e-06 lr_backbone=2.306149e-06 elapsed=19:59:12 avg_step=1.59s recent_step=1.55s eta=02:01:47 +2026-09-21 02:37:29 | INFO | train rank=0 | step=45400/50000 progress=90.8% loss=0.098176 lr=2.296232e-06 lr_backbone=2.296232e-06 elapsed=19:59:29 avg_step=1.59s recent_step=1.70s eta=02:01:32 +2026-09-21 02:37:49 | INFO | train rank=0 | step=45410/50000 progress=90.8% loss=0.069200 lr=2.286336e-06 lr_backbone=2.286336e-06 elapsed=19:59:48 avg_step=1.59s recent_step=1.93s eta=02:01:16 +2026-09-21 02:38:00 | INFO | train rank=0 | step=45420/50000 progress=90.8% loss=0.088882 lr=2.276460e-06 lr_backbone=2.276460e-06 elapsed=20:00:00 avg_step=1.59s recent_step=1.18s eta=02:01:00 +2026-09-21 02:38:15 | INFO | train rank=0 | step=45430/50000 progress=90.9% loss=0.073921 lr=2.266606e-06 lr_backbone=2.266606e-06 elapsed=20:00:15 avg_step=1.59s recent_step=1.44s eta=02:00:44 +2026-09-21 02:38:30 | INFO | train rank=0 | step=45440/50000 progress=90.9% loss=0.081223 lr=2.256773e-06 lr_backbone=2.256773e-06 elapsed=20:00:30 avg_step=1.59s recent_step=1.56s eta=02:00:28 +2026-09-21 02:38:48 | INFO | train rank=0 | step=45450/50000 progress=90.9% loss=0.112328 lr=2.246960e-06 lr_backbone=2.246960e-06 elapsed=20:00:48 avg_step=1.59s recent_step=1.76s eta=02:00:12 +2026-09-21 02:39:07 | INFO | train rank=0 | step=45460/50000 progress=90.9% loss=0.089602 lr=2.237169e-06 lr_backbone=2.237169e-06 elapsed=20:01:06 avg_step=1.59s recent_step=1.85s eta=01:59:57 +2026-09-21 02:39:21 | INFO | train rank=0 | step=45470/50000 progress=90.9% loss=0.073656 lr=2.227398e-06 lr_backbone=2.227398e-06 elapsed=20:01:20 avg_step=1.59s recent_step=1.41s eta=01:59:41 +2026-09-21 02:39:36 | INFO | train rank=0 | step=45480/50000 progress=91.0% loss=0.106700 lr=2.217648e-06 lr_backbone=2.217648e-06 elapsed=20:01:36 avg_step=1.59s recent_step=1.58s eta=01:59:25 +2026-09-21 02:39:54 | INFO | train rank=0 | step=45490/50000 progress=91.0% loss=0.079150 lr=2.207919e-06 lr_backbone=2.207919e-06 elapsed=20:01:53 avg_step=1.59s recent_step=1.70s eta=01:59:09 +2026-09-21 02:40:13 | INFO | train rank=0 | step=45500/50000 progress=91.0% loss=0.088197 lr=2.198211e-06 lr_backbone=2.198211e-06 elapsed=20:02:12 avg_step=1.59s recent_step=1.92s eta=01:58:54 +2026-09-21 02:40:26 | INFO | train rank=0 | step=45510/50000 progress=91.0% loss=0.096362 lr=2.188524e-06 lr_backbone=2.188524e-06 elapsed=20:02:25 avg_step=1.59s recent_step=1.29s eta=01:58:37 +2026-09-21 02:40:38 | INFO | train rank=0 | step=45520/50000 progress=91.0% loss=0.079839 lr=2.178858e-06 lr_backbone=2.178858e-06 elapsed=20:02:37 avg_step=1.59s recent_step=1.20s eta=01:58:21 +2026-09-21 02:41:03 | INFO | train rank=0 | step=45530/50000 progress=91.1% loss=0.063025 lr=2.169212e-06 lr_backbone=2.169212e-06 elapsed=20:03:02 avg_step=1.59s recent_step=2.53s eta=01:58:06 +2026-09-21 02:41:14 | INFO | train rank=0 | step=45540/50000 progress=91.1% loss=0.101742 lr=2.159588e-06 lr_backbone=2.159588e-06 elapsed=20:03:14 avg_step=1.59s recent_step=1.13s eta=01:57:50 +2026-09-21 02:41:27 | INFO | train rank=0 | step=45550/50000 progress=91.1% loss=0.088118 lr=2.149984e-06 lr_backbone=2.149984e-06 elapsed=20:03:26 avg_step=1.59s recent_step=1.26s eta=01:57:34 +2026-09-21 02:41:38 | INFO | train rank=0 | step=45560/50000 progress=91.1% loss=0.079583 lr=2.140402e-06 lr_backbone=2.140402e-06 elapsed=20:03:38 avg_step=1.59s recent_step=1.16s eta=01:57:17 +2026-09-21 02:41:59 | INFO | train rank=0 | step=45570/50000 progress=91.1% loss=0.082828 lr=2.130840e-06 lr_backbone=2.130840e-06 elapsed=20:03:58 avg_step=1.59s recent_step=2.05s eta=01:57:02 +2026-09-21 02:42:13 | INFO | train rank=0 | step=45580/50000 progress=91.2% loss=0.082316 lr=2.121300e-06 lr_backbone=2.121300e-06 elapsed=20:04:13 avg_step=1.59s recent_step=1.44s eta=01:56:46 +2026-09-21 02:42:32 | INFO | train rank=0 | step=45590/50000 progress=91.2% loss=0.059905 lr=2.111780e-06 lr_backbone=2.111780e-06 elapsed=20:04:32 avg_step=1.59s recent_step=1.88s eta=01:56:31 +2026-09-21 02:42:44 | INFO | train rank=0 | step=45600/50000 progress=91.2% loss=0.068417 lr=2.102281e-06 lr_backbone=2.102281e-06 elapsed=20:04:43 avg_step=1.59s recent_step=1.18s eta=01:56:14 +2026-09-21 02:43:01 | INFO | train rank=0 | step=45610/50000 progress=91.2% loss=0.100249 lr=2.092803e-06 lr_backbone=2.092803e-06 elapsed=20:05:01 avg_step=1.59s recent_step=1.76s eta=01:55:59 +2026-09-21 02:43:16 | INFO | train rank=0 | step=45620/50000 progress=91.2% loss=0.088176 lr=2.083346e-06 lr_backbone=2.083346e-06 elapsed=20:05:15 avg_step=1.59s recent_step=1.41s eta=01:55:43 +2026-09-21 02:43:30 | INFO | train rank=0 | step=45630/50000 progress=91.3% loss=0.069557 lr=2.073911e-06 lr_backbone=2.073911e-06 elapsed=20:05:30 avg_step=1.59s recent_step=1.44s eta=01:55:27 +2026-09-21 02:43:50 | INFO | train rank=0 | step=45640/50000 progress=91.3% loss=0.095548 lr=2.064496e-06 lr_backbone=2.064496e-06 elapsed=20:05:50 avg_step=1.59s recent_step=2.02s eta=01:55:11 +2026-09-21 02:44:04 | INFO | train rank=0 | step=45650/50000 progress=91.3% loss=0.075076 lr=2.055102e-06 lr_backbone=2.055102e-06 elapsed=20:06:03 avg_step=1.59s recent_step=1.35s eta=01:54:55 +2026-09-21 02:44:16 | INFO | train rank=0 | step=45660/50000 progress=91.3% loss=0.154677 lr=2.045729e-06 lr_backbone=2.045729e-06 elapsed=20:06:15 avg_step=1.59s recent_step=1.21s eta=01:54:39 +2026-09-21 02:44:33 | INFO | train rank=0 | step=45670/50000 progress=91.3% loss=0.090362 lr=2.036377e-06 lr_backbone=2.036377e-06 elapsed=20:06:32 avg_step=1.59s recent_step=1.71s eta=01:54:23 +2026-09-21 02:44:46 | INFO | train rank=0 | step=45680/50000 progress=91.4% loss=0.085222 lr=2.027046e-06 lr_backbone=2.027046e-06 elapsed=20:06:45 avg_step=1.59s recent_step=1.29s eta=01:54:07 +2026-09-21 02:45:11 | INFO | train rank=0 | step=45690/50000 progress=91.4% loss=0.083954 lr=2.017736e-06 lr_backbone=2.017736e-06 elapsed=20:07:10 avg_step=1.59s recent_step=2.51s eta=01:53:52 +2026-09-21 02:45:23 | INFO | train rank=0 | step=45700/50000 progress=91.4% loss=0.087956 lr=2.008447e-06 lr_backbone=2.008447e-06 elapsed=20:07:22 avg_step=1.59s recent_step=1.20s eta=01:53:36 +2026-09-21 02:45:40 | INFO | train rank=0 | step=45710/50000 progress=91.4% loss=0.066890 lr=1.999178e-06 lr_backbone=1.999178e-06 elapsed=20:07:40 avg_step=1.59s recent_step=1.74s eta=01:53:20 +2026-09-21 02:45:53 | INFO | train rank=0 | step=45720/50000 progress=91.4% loss=0.074249 lr=1.989931e-06 lr_backbone=1.989931e-06 elapsed=20:07:52 avg_step=1.59s recent_step=1.24s eta=01:53:04 +2026-09-21 02:46:10 | INFO | train rank=0 | step=45730/50000 progress=91.5% loss=0.079601 lr=1.980705e-06 lr_backbone=1.980705e-06 elapsed=20:08:09 avg_step=1.59s recent_step=1.70s eta=01:52:48 +2026-09-21 02:46:28 | INFO | train rank=0 | step=45740/50000 progress=91.5% loss=0.106056 lr=1.971500e-06 lr_backbone=1.971500e-06 elapsed=20:08:28 avg_step=1.59s recent_step=1.82s eta=01:52:33 +2026-09-21 02:46:41 | INFO | train rank=0 | step=45750/50000 progress=91.5% loss=0.078730 lr=1.962316e-06 lr_backbone=1.962316e-06 elapsed=20:08:40 avg_step=1.59s recent_step=1.30s eta=01:52:16 +2026-09-21 02:47:00 | INFO | train rank=0 | step=45760/50000 progress=91.5% loss=0.062886 lr=1.953153e-06 lr_backbone=1.953153e-06 elapsed=20:09:00 avg_step=1.59s recent_step=1.91s eta=01:52:01 +2026-09-21 02:47:13 | INFO | train rank=0 | step=45770/50000 progress=91.5% loss=0.095502 lr=1.944011e-06 lr_backbone=1.944011e-06 elapsed=20:09:13 avg_step=1.59s recent_step=1.33s eta=01:51:45 +2026-09-21 02:47:29 | INFO | train rank=0 | step=45780/50000 progress=91.6% loss=0.066395 lr=1.934890e-06 lr_backbone=1.934890e-06 elapsed=20:09:28 avg_step=1.59s recent_step=1.55s eta=01:51:29 +2026-09-21 02:47:46 | INFO | train rank=0 | step=45790/50000 progress=91.6% loss=0.080493 lr=1.925790e-06 lr_backbone=1.925790e-06 elapsed=20:09:46 avg_step=1.59s recent_step=1.73s eta=01:51:13 +2026-09-21 02:47:59 | INFO | train rank=0 | step=45800/50000 progress=91.6% loss=0.080678 lr=1.916711e-06 lr_backbone=1.916711e-06 elapsed=20:09:58 avg_step=1.59s recent_step=1.25s eta=01:50:57 +2026-09-21 02:48:12 | INFO | train rank=0 | step=45810/50000 progress=91.6% loss=0.073559 lr=1.907653e-06 lr_backbone=1.907653e-06 elapsed=20:10:11 avg_step=1.59s recent_step=1.29s eta=01:50:41 +2026-09-21 02:48:33 | INFO | train rank=0 | step=45820/50000 progress=91.6% loss=0.096903 lr=1.898616e-06 lr_backbone=1.898616e-06 elapsed=20:10:33 avg_step=1.59s recent_step=2.15s eta=01:50:26 +2026-09-21 02:48:45 | INFO | train rank=0 | step=45830/50000 progress=91.7% loss=0.063285 lr=1.889601e-06 lr_backbone=1.889601e-06 elapsed=20:10:45 avg_step=1.59s recent_step=1.22s eta=01:50:09 +2026-09-21 02:49:02 | INFO | train rank=0 | step=45840/50000 progress=91.7% loss=0.125041 lr=1.880606e-06 lr_backbone=1.880606e-06 elapsed=20:11:01 avg_step=1.59s recent_step=1.62s eta=01:49:54 +2026-09-21 02:49:19 | INFO | train rank=0 | step=45850/50000 progress=91.7% loss=0.091548 lr=1.871632e-06 lr_backbone=1.871632e-06 elapsed=20:11:18 avg_step=1.59s recent_step=1.72s eta=01:49:38 +2026-09-21 02:49:35 | INFO | train rank=0 | step=45860/50000 progress=91.7% loss=0.092330 lr=1.862679e-06 lr_backbone=1.862679e-06 elapsed=20:11:35 avg_step=1.59s recent_step=1.61s eta=01:49:22 +2026-09-21 02:49:50 | INFO | train rank=0 | step=45870/50000 progress=91.7% loss=0.079635 lr=1.853748e-06 lr_backbone=1.853748e-06 elapsed=20:11:50 avg_step=1.59s recent_step=1.51s eta=01:49:06 +2026-09-21 02:50:03 | INFO | train rank=0 | step=45880/50000 progress=91.8% loss=0.080608 lr=1.844837e-06 lr_backbone=1.844837e-06 elapsed=20:12:03 avg_step=1.59s recent_step=1.32s eta=01:48:50 +2026-09-21 02:50:23 | INFO | train rank=0 | step=45890/50000 progress=91.8% loss=0.096362 lr=1.835948e-06 lr_backbone=1.835948e-06 elapsed=20:12:23 avg_step=1.59s recent_step=2.00s eta=01:48:35 +2026-09-21 02:50:38 | INFO | train rank=0 | step=45900/50000 progress=91.8% loss=0.077541 lr=1.827079e-06 lr_backbone=1.827079e-06 elapsed=20:12:38 avg_step=1.59s recent_step=1.53s eta=01:48:19 +2026-09-21 02:50:51 | INFO | train rank=0 | step=45910/50000 progress=91.8% loss=0.063209 lr=1.818232e-06 lr_backbone=1.818232e-06 elapsed=20:12:50 avg_step=1.59s recent_step=1.23s eta=01:48:02 +2026-09-21 02:51:03 | INFO | train rank=0 | step=45920/50000 progress=91.8% loss=0.082530 lr=1.809406e-06 lr_backbone=1.809406e-06 elapsed=20:13:02 avg_step=1.58s recent_step=1.20s eta=01:47:46 +2026-09-21 02:51:22 | INFO | train rank=0 | step=45930/50000 progress=91.9% loss=0.088770 lr=1.800600e-06 lr_backbone=1.800600e-06 elapsed=20:13:22 avg_step=1.59s recent_step=1.94s eta=01:47:31 +2026-09-21 02:51:36 | INFO | train rank=0 | step=45940/50000 progress=91.9% loss=0.087005 lr=1.791816e-06 lr_backbone=1.791816e-06 elapsed=20:13:36 avg_step=1.59s recent_step=1.40s eta=01:47:15 +2026-09-21 02:51:53 | INFO | train rank=0 | step=45950/50000 progress=91.9% loss=0.092950 lr=1.783053e-06 lr_backbone=1.783053e-06 elapsed=20:13:53 avg_step=1.59s recent_step=1.72s eta=01:46:59 +2026-09-21 02:52:05 | INFO | train rank=0 | step=45960/50000 progress=91.9% loss=0.076859 lr=1.774311e-06 lr_backbone=1.774311e-06 elapsed=20:14:05 avg_step=1.58s recent_step=1.20s eta=01:46:43 +2026-09-21 02:52:26 | INFO | train rank=0 | step=45970/50000 progress=91.9% loss=0.084690 lr=1.765590e-06 lr_backbone=1.765590e-06 elapsed=20:14:26 avg_step=1.59s recent_step=2.10s eta=01:46:27 +2026-09-21 02:52:38 | INFO | train rank=0 | step=45980/50000 progress=92.0% loss=0.085130 lr=1.756891e-06 lr_backbone=1.756891e-06 elapsed=20:14:38 avg_step=1.59s recent_step=1.21s eta=01:46:11 +2026-09-21 02:52:57 | INFO | train rank=0 | step=45990/50000 progress=92.0% loss=0.067369 lr=1.748212e-06 lr_backbone=1.748212e-06 elapsed=20:14:57 avg_step=1.59s recent_step=1.88s eta=01:45:56 +2026-09-21 02:53:10 | INFO | train rank=0 | step=46000/50000 progress=92.0% loss=0.055510 lr=1.739554e-06 lr_backbone=1.739554e-06 elapsed=20:15:09 avg_step=1.58s recent_step=1.24s eta=01:45:39 +2026-09-21 02:53:25 | INFO | train rank=0 | step=46010/50000 progress=92.0% loss=0.109495 lr=1.730918e-06 lr_backbone=1.730918e-06 elapsed=20:15:25 avg_step=1.58s recent_step=1.57s eta=01:45:24 +2026-09-21 02:53:38 | INFO | train rank=0 | step=46020/50000 progress=92.0% loss=0.099463 lr=1.722303e-06 lr_backbone=1.722303e-06 elapsed=20:15:37 avg_step=1.58s recent_step=1.23s eta=01:45:07 +2026-09-21 02:53:52 | INFO | train rank=0 | step=46030/50000 progress=92.1% loss=0.091642 lr=1.713709e-06 lr_backbone=1.713709e-06 elapsed=20:15:52 avg_step=1.58s recent_step=1.45s eta=01:44:51 +2026-09-21 02:54:12 | INFO | train rank=0 | step=46040/50000 progress=92.1% loss=0.094537 lr=1.705135e-06 lr_backbone=1.705135e-06 elapsed=20:16:11 avg_step=1.58s recent_step=1.94s eta=01:44:36 +2026-09-21 02:54:24 | INFO | train rank=0 | step=46050/50000 progress=92.1% loss=0.085305 lr=1.696584e-06 lr_backbone=1.696584e-06 elapsed=20:16:24 avg_step=1.58s recent_step=1.25s eta=01:44:20 +2026-09-21 02:54:41 | INFO | train rank=0 | step=46060/50000 progress=92.1% loss=0.071566 lr=1.688053e-06 lr_backbone=1.688053e-06 elapsed=20:16:41 avg_step=1.58s recent_step=1.71s eta=01:44:04 +2026-09-21 02:54:53 | INFO | train rank=0 | step=46070/50000 progress=92.1% loss=0.077910 lr=1.679543e-06 lr_backbone=1.679543e-06 elapsed=20:16:53 avg_step=1.58s recent_step=1.19s eta=01:43:48 +2026-09-21 02:55:09 | INFO | train rank=0 | step=46080/50000 progress=92.2% loss=0.072148 lr=1.671054e-06 lr_backbone=1.671054e-06 elapsed=20:17:09 avg_step=1.58s recent_step=1.65s eta=01:43:32 +2026-09-21 02:55:23 | INFO | train rank=0 | step=46090/50000 progress=92.2% loss=0.088440 lr=1.662587e-06 lr_backbone=1.662587e-06 elapsed=20:17:23 avg_step=1.58s recent_step=1.40s eta=01:43:16 +2026-09-21 02:55:36 | INFO | train rank=0 | step=46100/50000 progress=92.2% loss=0.063573 lr=1.654141e-06 lr_backbone=1.654141e-06 elapsed=20:17:35 avg_step=1.58s recent_step=1.21s eta=01:43:00 +2026-09-21 02:56:00 | INFO | train rank=0 | step=46110/50000 progress=92.2% loss=0.082740 lr=1.645716e-06 lr_backbone=1.645716e-06 elapsed=20:18:00 avg_step=1.58s recent_step=2.44s eta=01:42:45 +2026-09-21 02:56:28 | INFO | train rank=0 | step=46120/50000 progress=92.2% loss=0.080766 lr=1.637312e-06 lr_backbone=1.637312e-06 elapsed=20:18:27 avg_step=1.59s recent_step=2.75s eta=01:42:30 +2026-09-21 02:56:41 | INFO | train rank=0 | step=46130/50000 progress=92.3% loss=0.091956 lr=1.628929e-06 lr_backbone=1.628929e-06 elapsed=20:18:40 avg_step=1.59s recent_step=1.31s eta=01:42:14 +2026-09-21 02:57:35 | INFO | train rank=0 | step=46140/50000 progress=92.3% loss=0.083012 lr=1.620567e-06 lr_backbone=1.620567e-06 elapsed=20:19:34 avg_step=1.59s recent_step=5.39s eta=01:42:01 +2026-09-21 02:57:46 | INFO | train rank=0 | step=46150/50000 progress=92.3% loss=0.100659 lr=1.612227e-06 lr_backbone=1.612227e-06 elapsed=20:19:46 avg_step=1.59s recent_step=1.14s eta=01:41:45 +2026-09-21 02:57:59 | INFO | train rank=0 | step=46160/50000 progress=92.3% loss=0.084005 lr=1.603907e-06 lr_backbone=1.603907e-06 elapsed=20:19:58 avg_step=1.59s recent_step=1.26s eta=01:41:29 +2026-09-21 02:58:13 | INFO | train rank=0 | step=46170/50000 progress=92.3% loss=0.098023 lr=1.595609e-06 lr_backbone=1.595609e-06 elapsed=20:20:12 avg_step=1.59s recent_step=1.42s eta=01:41:13 +2026-09-21 02:58:29 | INFO | train rank=0 | step=46180/50000 progress=92.4% loss=0.100711 lr=1.587332e-06 lr_backbone=1.587332e-06 elapsed=20:20:29 avg_step=1.59s recent_step=1.62s eta=01:40:57 +2026-09-21 02:58:45 | INFO | train rank=0 | step=46190/50000 progress=92.4% loss=0.084582 lr=1.579077e-06 lr_backbone=1.579077e-06 elapsed=20:20:45 avg_step=1.59s recent_step=1.59s eta=01:40:41 +2026-09-21 02:59:00 | INFO | train rank=0 | step=46200/50000 progress=92.4% loss=0.095705 lr=1.570842e-06 lr_backbone=1.570842e-06 elapsed=20:21:00 avg_step=1.59s recent_step=1.55s eta=01:40:25 +2026-09-21 02:59:14 | INFO | train rank=0 | step=46210/50000 progress=92.4% loss=0.088796 lr=1.562629e-06 lr_backbone=1.562629e-06 elapsed=20:21:14 avg_step=1.59s recent_step=1.36s eta=01:40:09 +2026-09-21 02:59:30 | INFO | train rank=0 | step=46220/50000 progress=92.4% loss=0.085166 lr=1.554436e-06 lr_backbone=1.554436e-06 elapsed=20:21:30 avg_step=1.59s recent_step=1.58s eta=01:39:53 +2026-09-21 02:59:45 | INFO | train rank=0 | step=46230/50000 progress=92.5% loss=0.109822 lr=1.546265e-06 lr_backbone=1.546265e-06 elapsed=20:21:44 avg_step=1.59s recent_step=1.48s eta=01:39:37 +2026-09-21 03:00:03 | INFO | train rank=0 | step=46240/50000 progress=92.5% loss=0.090104 lr=1.538115e-06 lr_backbone=1.538115e-06 elapsed=20:22:02 avg_step=1.59s recent_step=1.78s eta=01:39:22 +2026-09-21 03:00:14 | INFO | train rank=0 | step=46250/50000 progress=92.5% loss=0.070820 lr=1.529987e-06 lr_backbone=1.529987e-06 elapsed=20:22:14 avg_step=1.59s recent_step=1.19s eta=01:39:06 +2026-09-21 03:00:34 | INFO | train rank=0 | step=46260/50000 progress=92.5% loss=0.086616 lr=1.521879e-06 lr_backbone=1.521879e-06 elapsed=20:22:34 avg_step=1.59s recent_step=1.99s eta=01:38:50 +2026-09-21 03:00:49 | INFO | train rank=0 | step=46270/50000 progress=92.5% loss=0.087294 lr=1.513793e-06 lr_backbone=1.513793e-06 elapsed=20:22:48 avg_step=1.59s recent_step=1.44s eta=01:38:34 +2026-09-21 03:01:03 | INFO | train rank=0 | step=46280/50000 progress=92.6% loss=0.087147 lr=1.505728e-06 lr_backbone=1.505728e-06 elapsed=20:23:02 avg_step=1.59s recent_step=1.42s eta=01:38:18 +2026-09-21 03:01:16 | INFO | train rank=0 | step=46290/50000 progress=92.6% loss=0.093549 lr=1.497684e-06 lr_backbone=1.497684e-06 elapsed=20:23:15 avg_step=1.59s recent_step=1.28s eta=01:38:02 +2026-09-21 03:01:33 | INFO | train rank=0 | step=46300/50000 progress=92.6% loss=0.133592 lr=1.489662e-06 lr_backbone=1.489662e-06 elapsed=20:23:33 avg_step=1.59s recent_step=1.77s eta=01:37:46 +2026-09-21 03:01:54 | INFO | train rank=0 | step=46310/50000 progress=92.6% loss=0.074790 lr=1.481660e-06 lr_backbone=1.481660e-06 elapsed=20:23:54 avg_step=1.59s recent_step=2.06s eta=01:37:31 +2026-09-21 03:02:06 | INFO | train rank=0 | step=46320/50000 progress=92.6% loss=0.090798 lr=1.473680e-06 lr_backbone=1.473680e-06 elapsed=20:24:06 avg_step=1.59s recent_step=1.19s eta=01:37:15 +2026-09-21 03:02:23 | INFO | train rank=0 | step=46330/50000 progress=92.7% loss=0.071015 lr=1.465721e-06 lr_backbone=1.465721e-06 elapsed=20:24:23 avg_step=1.59s recent_step=1.72s eta=01:36:59 +2026-09-21 03:02:37 | INFO | train rank=0 | step=46340/50000 progress=92.7% loss=0.075052 lr=1.457783e-06 lr_backbone=1.457783e-06 elapsed=20:24:36 avg_step=1.59s recent_step=1.37s eta=01:36:43 +2026-09-21 03:02:54 | INFO | train rank=0 | step=46350/50000 progress=92.7% loss=0.081221 lr=1.449867e-06 lr_backbone=1.449867e-06 elapsed=20:24:53 avg_step=1.59s recent_step=1.70s eta=01:36:27 +2026-09-21 03:03:09 | INFO | train rank=0 | step=46360/50000 progress=92.7% loss=0.076883 lr=1.441972e-06 lr_backbone=1.441972e-06 elapsed=20:25:09 avg_step=1.59s recent_step=1.54s eta=01:36:11 +2026-09-21 03:03:25 | INFO | train rank=0 | step=46370/50000 progress=92.7% loss=0.082553 lr=1.434098e-06 lr_backbone=1.434098e-06 elapsed=20:25:25 avg_step=1.59s recent_step=1.60s eta=01:35:55 +2026-09-21 03:03:42 | INFO | train rank=0 | step=46380/50000 progress=92.8% loss=0.073633 lr=1.426245e-06 lr_backbone=1.426245e-06 elapsed=20:25:42 avg_step=1.59s recent_step=1.68s eta=01:35:40 +2026-09-21 03:03:56 | INFO | train rank=0 | step=46390/50000 progress=92.8% loss=0.114229 lr=1.418413e-06 lr_backbone=1.418413e-06 elapsed=20:25:55 avg_step=1.59s recent_step=1.38s eta=01:35:24 +2026-09-21 03:04:18 | INFO | train rank=0 | step=46400/50000 progress=92.8% loss=0.104744 lr=1.410603e-06 lr_backbone=1.410603e-06 elapsed=20:26:17 avg_step=1.59s recent_step=2.20s eta=01:35:08 +2026-09-21 03:04:30 | INFO | train rank=0 | step=46410/50000 progress=92.8% loss=0.084006 lr=1.402814e-06 lr_backbone=1.402814e-06 elapsed=20:26:29 avg_step=1.59s recent_step=1.20s eta=01:34:52 +2026-09-21 03:04:46 | INFO | train rank=0 | step=46420/50000 progress=92.8% loss=0.085434 lr=1.395046e-06 lr_backbone=1.395046e-06 elapsed=20:26:46 avg_step=1.59s recent_step=1.67s eta=01:34:36 +2026-09-21 03:05:03 | INFO | train rank=0 | step=46430/50000 progress=92.9% loss=0.085432 lr=1.387300e-06 lr_backbone=1.387300e-06 elapsed=20:27:03 avg_step=1.59s recent_step=1.64s eta=01:34:20 +2026-09-21 03:05:19 | INFO | train rank=0 | step=46440/50000 progress=92.9% loss=0.086790 lr=1.379575e-06 lr_backbone=1.379575e-06 elapsed=20:27:19 avg_step=1.59s recent_step=1.64s eta=01:34:05 +2026-09-21 03:05:34 | INFO | train rank=0 | step=46450/50000 progress=92.9% loss=0.083657 lr=1.371871e-06 lr_backbone=1.371871e-06 elapsed=20:27:33 avg_step=1.59s recent_step=1.44s eta=01:33:49 +2026-09-21 03:05:46 | INFO | train rank=0 | step=46460/50000 progress=92.9% loss=0.122396 lr=1.364188e-06 lr_backbone=1.364188e-06 elapsed=20:27:45 avg_step=1.59s recent_step=1.21s eta=01:33:32 +2026-09-21 03:06:02 | INFO | train rank=0 | step=46470/50000 progress=92.9% loss=0.080767 lr=1.356527e-06 lr_backbone=1.356527e-06 elapsed=20:28:02 avg_step=1.59s recent_step=1.63s eta=01:33:17 +2026-09-21 03:06:25 | INFO | train rank=0 | step=46480/50000 progress=93.0% loss=0.090482 lr=1.348887e-06 lr_backbone=1.348887e-06 elapsed=20:28:24 avg_step=1.59s recent_step=2.27s eta=01:33:01 +2026-09-21 03:06:36 | INFO | train rank=0 | step=46490/50000 progress=93.0% loss=0.101166 lr=1.341268e-06 lr_backbone=1.341268e-06 elapsed=20:28:36 avg_step=1.59s recent_step=1.16s eta=01:32:45 +2026-09-21 03:06:54 | INFO | train rank=0 | step=46500/50000 progress=93.0% loss=0.116078 lr=1.333670e-06 lr_backbone=1.333670e-06 elapsed=20:28:54 avg_step=1.59s recent_step=1.77s eta=01:32:29 +2026-09-21 03:07:08 | INFO | train rank=0 | step=46510/50000 progress=93.0% loss=0.074763 lr=1.326094e-06 lr_backbone=1.326094e-06 elapsed=20:29:07 avg_step=1.59s recent_step=1.37s eta=01:32:13 +2026-09-21 03:07:27 | INFO | train rank=0 | step=46520/50000 progress=93.0% loss=0.124380 lr=1.318539e-06 lr_backbone=1.318539e-06 elapsed=20:29:26 avg_step=1.59s recent_step=1.88s eta=01:31:58 +2026-09-21 03:07:39 | INFO | train rank=0 | step=46530/50000 progress=93.1% loss=0.102851 lr=1.311005e-06 lr_backbone=1.311005e-06 elapsed=20:29:39 avg_step=1.59s recent_step=1.24s eta=01:31:42 +2026-09-21 03:07:51 | INFO | train rank=0 | step=46540/50000 progress=93.1% loss=0.084270 lr=1.303493e-06 lr_backbone=1.303493e-06 elapsed=20:29:50 avg_step=1.59s recent_step=1.15s eta=01:31:25 +2026-09-21 03:08:13 | INFO | train rank=0 | step=46550/50000 progress=93.1% loss=0.097150 lr=1.296002e-06 lr_backbone=1.296002e-06 elapsed=20:30:12 avg_step=1.59s recent_step=2.20s eta=01:31:10 +2026-09-21 03:08:26 | INFO | train rank=0 | step=46560/50000 progress=93.1% loss=0.080465 lr=1.288532e-06 lr_backbone=1.288532e-06 elapsed=20:30:26 avg_step=1.59s recent_step=1.38s eta=01:30:54 +2026-09-21 03:08:41 | INFO | train rank=0 | step=46570/50000 progress=93.1% loss=0.079834 lr=1.281084e-06 lr_backbone=1.281084e-06 elapsed=20:30:41 avg_step=1.59s recent_step=1.50s eta=01:30:38 +2026-09-21 03:08:59 | INFO | train rank=0 | step=46580/50000 progress=93.2% loss=0.093338 lr=1.273656e-06 lr_backbone=1.273656e-06 elapsed=20:30:58 avg_step=1.59s recent_step=1.72s eta=01:30:22 +2026-09-21 03:09:12 | INFO | train rank=0 | step=46590/50000 progress=93.2% loss=0.073456 lr=1.266251e-06 lr_backbone=1.266251e-06 elapsed=20:31:11 avg_step=1.59s recent_step=1.31s eta=01:30:06 +2026-09-21 03:09:30 | INFO | train rank=0 | step=46600/50000 progress=93.2% loss=0.082484 lr=1.258866e-06 lr_backbone=1.258866e-06 elapsed=20:31:30 avg_step=1.59s recent_step=1.87s eta=01:29:51 +2026-09-21 03:09:44 | INFO | train rank=0 | step=46610/50000 progress=93.2% loss=0.073331 lr=1.251503e-06 lr_backbone=1.251503e-06 elapsed=20:31:44 avg_step=1.59s recent_step=1.36s eta=01:29:35 +2026-09-21 03:09:59 | INFO | train rank=0 | step=46620/50000 progress=93.2% loss=0.068060 lr=1.244161e-06 lr_backbone=1.244161e-06 elapsed=20:31:58 avg_step=1.59s recent_step=1.46s eta=01:29:19 +2026-09-21 03:10:17 | INFO | train rank=0 | step=46630/50000 progress=93.3% loss=0.085600 lr=1.236840e-06 lr_backbone=1.236840e-06 elapsed=20:32:16 avg_step=1.59s recent_step=1.82s eta=01:29:03 +2026-09-21 03:10:33 | INFO | train rank=0 | step=46640/50000 progress=93.3% loss=0.091743 lr=1.229541e-06 lr_backbone=1.229541e-06 elapsed=20:32:33 avg_step=1.59s recent_step=1.66s eta=01:28:47 +2026-09-21 03:10:45 | INFO | train rank=0 | step=46650/50000 progress=93.3% loss=0.103060 lr=1.222263e-06 lr_backbone=1.222263e-06 elapsed=20:32:45 avg_step=1.59s recent_step=1.19s eta=01:28:31 +2026-09-21 03:11:05 | INFO | train rank=0 | step=46660/50000 progress=93.3% loss=0.073499 lr=1.215007e-06 lr_backbone=1.215007e-06 elapsed=20:33:05 avg_step=1.59s recent_step=2.02s eta=01:28:16 +2026-09-21 03:11:18 | INFO | train rank=0 | step=46670/50000 progress=93.3% loss=0.093574 lr=1.207772e-06 lr_backbone=1.207772e-06 elapsed=20:33:18 avg_step=1.59s recent_step=1.27s eta=01:27:59 +2026-09-21 03:11:35 | INFO | train rank=0 | step=46680/50000 progress=93.4% loss=0.080327 lr=1.200558e-06 lr_backbone=1.200558e-06 elapsed=20:33:34 avg_step=1.59s recent_step=1.65s eta=01:27:44 +2026-09-21 03:11:53 | INFO | train rank=0 | step=46690/50000 progress=93.4% loss=0.079676 lr=1.193365e-06 lr_backbone=1.193365e-06 elapsed=20:33:53 avg_step=1.59s recent_step=1.88s eta=01:27:28 +2026-09-21 03:12:05 | INFO | train rank=0 | step=46700/50000 progress=93.4% loss=0.075230 lr=1.186194e-06 lr_backbone=1.186194e-06 elapsed=20:34:05 avg_step=1.59s recent_step=1.19s eta=01:27:12 +2026-09-21 03:12:20 | INFO | train rank=0 | step=46710/50000 progress=93.4% loss=0.081181 lr=1.179044e-06 lr_backbone=1.179044e-06 elapsed=20:34:20 avg_step=1.59s recent_step=1.47s eta=01:26:56 +2026-09-21 03:12:41 | INFO | train rank=0 | step=46720/50000 progress=93.4% loss=0.086109 lr=1.171916e-06 lr_backbone=1.171916e-06 elapsed=20:34:41 avg_step=1.59s recent_step=2.10s eta=01:26:40 +2026-09-21 03:12:56 | INFO | train rank=0 | step=46730/50000 progress=93.5% loss=0.074630 lr=1.164809e-06 lr_backbone=1.164809e-06 elapsed=20:34:56 avg_step=1.59s recent_step=1.52s eta=01:26:24 +2026-09-21 03:13:11 | INFO | train rank=0 | step=46740/50000 progress=93.5% loss=0.081029 lr=1.157723e-06 lr_backbone=1.157723e-06 elapsed=20:35:10 avg_step=1.59s recent_step=1.44s eta=01:26:09 +2026-09-21 03:13:27 | INFO | train rank=0 | step=46750/50000 progress=93.5% loss=0.093191 lr=1.150658e-06 lr_backbone=1.150658e-06 elapsed=20:35:27 avg_step=1.59s recent_step=1.67s eta=01:25:53 +2026-09-21 03:13:43 | INFO | train rank=0 | step=46760/50000 progress=93.5% loss=0.063172 lr=1.143615e-06 lr_backbone=1.143615e-06 elapsed=20:35:43 avg_step=1.59s recent_step=1.58s eta=01:25:37 +2026-09-21 03:14:00 | INFO | train rank=0 | step=46770/50000 progress=93.5% loss=0.084990 lr=1.136594e-06 lr_backbone=1.136594e-06 elapsed=20:36:00 avg_step=1.59s recent_step=1.72s eta=01:25:21 +2026-09-21 03:14:15 | INFO | train rank=0 | step=46780/50000 progress=93.6% loss=0.093886 lr=1.129594e-06 lr_backbone=1.129594e-06 elapsed=20:36:14 avg_step=1.59s recent_step=1.43s eta=01:25:05 +2026-09-21 03:14:27 | INFO | train rank=0 | step=46790/50000 progress=93.6% loss=0.142376 lr=1.122615e-06 lr_backbone=1.122615e-06 elapsed=20:36:27 avg_step=1.59s recent_step=1.27s eta=01:24:49 +2026-09-21 03:14:45 | INFO | train rank=0 | step=46800/50000 progress=93.6% loss=0.080435 lr=1.115657e-06 lr_backbone=1.115657e-06 elapsed=20:36:45 avg_step=1.59s recent_step=1.81s eta=01:24:33 +2026-09-21 03:14:57 | INFO | train rank=0 | step=46810/50000 progress=93.6% loss=0.081140 lr=1.108721e-06 lr_backbone=1.108721e-06 elapsed=20:36:57 avg_step=1.59s recent_step=1.21s eta=01:24:17 +2026-09-21 03:15:15 | INFO | train rank=0 | step=46820/50000 progress=93.6% loss=0.067038 lr=1.101806e-06 lr_backbone=1.101806e-06 elapsed=20:37:15 avg_step=1.59s recent_step=1.77s eta=01:24:02 +2026-09-21 03:15:34 | INFO | train rank=0 | step=46830/50000 progress=93.7% loss=0.061539 lr=1.094913e-06 lr_backbone=1.094913e-06 elapsed=20:37:34 avg_step=1.59s recent_step=1.89s eta=01:23:46 +2026-09-21 03:15:49 | INFO | train rank=0 | step=46840/50000 progress=93.7% loss=0.102485 lr=1.088041e-06 lr_backbone=1.088041e-06 elapsed=20:37:48 avg_step=1.59s recent_step=1.44s eta=01:23:30 +2026-09-21 03:16:03 | INFO | train rank=0 | step=46850/50000 progress=93.7% loss=0.085186 lr=1.081190e-06 lr_backbone=1.081190e-06 elapsed=20:38:03 avg_step=1.59s recent_step=1.49s eta=01:23:14 +2026-09-21 03:16:20 | INFO | train rank=0 | step=46860/50000 progress=93.7% loss=0.069715 lr=1.074361e-06 lr_backbone=1.074361e-06 elapsed=20:38:20 avg_step=1.59s recent_step=1.69s eta=01:22:58 +2026-09-21 03:16:35 | INFO | train rank=0 | step=46870/50000 progress=93.7% loss=0.116985 lr=1.067554e-06 lr_backbone=1.067554e-06 elapsed=20:38:35 avg_step=1.59s recent_step=1.46s eta=01:22:42 +2026-09-21 03:16:53 | INFO | train rank=0 | step=46880/50000 progress=93.8% loss=0.087396 lr=1.060767e-06 lr_backbone=1.060767e-06 elapsed=20:38:53 avg_step=1.59s recent_step=1.82s eta=01:22:27 +2026-09-21 03:17:05 | INFO | train rank=0 | step=46890/50000 progress=93.8% loss=0.098499 lr=1.054002e-06 lr_backbone=1.054002e-06 elapsed=20:39:05 avg_step=1.59s recent_step=1.22s eta=01:22:10 +2026-09-21 03:17:19 | INFO | train rank=0 | step=46900/50000 progress=93.8% loss=0.093566 lr=1.047259e-06 lr_backbone=1.047259e-06 elapsed=20:39:19 avg_step=1.59s recent_step=1.35s eta=01:21:54 +2026-09-21 03:17:40 | INFO | train rank=0 | step=46910/50000 progress=93.8% loss=0.135734 lr=1.040537e-06 lr_backbone=1.040537e-06 elapsed=20:39:39 avg_step=1.59s recent_step=2.06s eta=01:21:39 +2026-09-21 03:17:52 | INFO | train rank=0 | step=46920/50000 progress=93.8% loss=0.103243 lr=1.033836e-06 lr_backbone=1.033836e-06 elapsed=20:39:52 avg_step=1.59s recent_step=1.25s eta=01:21:23 +2026-09-21 03:18:11 | INFO | train rank=0 | step=46930/50000 progress=93.9% loss=0.092238 lr=1.027157e-06 lr_backbone=1.027157e-06 elapsed=20:40:11 avg_step=1.59s recent_step=1.95s eta=01:21:07 +2026-09-21 03:18:24 | INFO | train rank=0 | step=46940/50000 progress=93.9% loss=0.080724 lr=1.020499e-06 lr_backbone=1.020499e-06 elapsed=20:40:24 avg_step=1.59s recent_step=1.24s eta=01:20:51 +2026-09-21 03:18:42 | INFO | train rank=0 | step=46950/50000 progress=93.9% loss=0.067136 lr=1.013862e-06 lr_backbone=1.013862e-06 elapsed=20:40:41 avg_step=1.59s recent_step=1.79s eta=01:20:35 +2026-09-21 03:18:56 | INFO | train rank=0 | step=46960/50000 progress=93.9% loss=0.082165 lr=1.007247e-06 lr_backbone=1.007247e-06 elapsed=20:40:55 avg_step=1.59s recent_step=1.39s eta=01:20:19 +2026-09-21 03:19:13 | INFO | train rank=0 | step=46970/50000 progress=93.9% loss=0.071849 lr=1.000654e-06 lr_backbone=1.000654e-06 elapsed=20:41:13 avg_step=1.59s recent_step=1.74s eta=01:20:04 +2026-09-21 03:19:26 | INFO | train rank=0 | step=46980/50000 progress=94.0% loss=0.084141 lr=9.940817e-07 lr_backbone=9.940817e-07 elapsed=20:41:26 avg_step=1.59s recent_step=1.28s eta=01:19:48 +2026-09-21 03:19:44 | INFO | train rank=0 | step=46990/50000 progress=94.0% loss=0.094412 lr=9.875310e-07 lr_backbone=9.875310e-07 elapsed=20:41:43 avg_step=1.59s recent_step=1.78s eta=01:19:32 +2026-09-21 03:19:58 | INFO | train rank=0 | step=47000/50000 progress=94.0% loss=0.069960 lr=9.810017e-07 lr_backbone=9.810017e-07 elapsed=20:41:58 avg_step=1.59s recent_step=1.46s eta=01:19:16 +2026-09-21 03:20:14 | INFO | train rank=0 | step=47010/50000 progress=94.0% loss=0.079950 lr=9.744939e-07 lr_backbone=9.744939e-07 elapsed=20:42:14 avg_step=1.59s recent_step=1.55s eta=01:19:00 +2026-09-21 03:20:30 | INFO | train rank=0 | step=47020/50000 progress=94.0% loss=0.092161 lr=9.680075e-07 lr_backbone=9.680075e-07 elapsed=20:42:30 avg_step=1.59s recent_step=1.65s eta=01:18:44 +2026-09-21 03:20:43 | INFO | train rank=0 | step=47030/50000 progress=94.1% loss=0.098344 lr=9.615426e-07 lr_backbone=9.615426e-07 elapsed=20:42:43 avg_step=1.59s recent_step=1.31s eta=01:18:28 +2026-09-21 03:21:05 | INFO | train rank=0 | step=47040/50000 progress=94.1% loss=0.080546 lr=9.550991e-07 lr_backbone=9.550991e-07 elapsed=20:43:04 avg_step=1.59s recent_step=2.13s eta=01:18:13 +2026-09-21 03:21:17 | INFO | train rank=0 | step=47050/50000 progress=94.1% loss=0.089464 lr=9.486771e-07 lr_backbone=9.486771e-07 elapsed=20:43:16 avg_step=1.59s recent_step=1.19s eta=01:17:57 +2026-09-21 03:21:29 | INFO | train rank=0 | step=47060/50000 progress=94.1% loss=0.126151 lr=9.422765e-07 lr_backbone=9.422765e-07 elapsed=20:43:29 avg_step=1.59s recent_step=1.23s eta=01:17:41 +2026-09-21 03:21:45 | INFO | train rank=0 | step=47070/50000 progress=94.1% loss=0.095546 lr=9.358974e-07 lr_backbone=9.358974e-07 elapsed=20:43:45 avg_step=1.59s recent_step=1.60s eta=01:17:25 +2026-09-21 03:21:59 | INFO | train rank=0 | step=47080/50000 progress=94.2% loss=0.127427 lr=9.295398e-07 lr_backbone=9.295398e-07 elapsed=20:43:58 avg_step=1.59s recent_step=1.37s eta=01:17:09 +2026-09-21 03:22:15 | INFO | train rank=0 | step=47090/50000 progress=94.2% loss=0.092535 lr=9.232036e-07 lr_backbone=9.232036e-07 elapsed=20:44:14 avg_step=1.59s recent_step=1.62s eta=01:16:53 +2026-09-21 03:22:31 | INFO | train rank=0 | step=47100/50000 progress=94.2% loss=0.083039 lr=9.168889e-07 lr_backbone=9.168889e-07 elapsed=20:44:31 avg_step=1.59s recent_step=1.63s eta=01:16:37 +2026-09-21 03:22:46 | INFO | train rank=0 | step=47110/50000 progress=94.2% loss=0.093207 lr=9.105957e-07 lr_backbone=9.105957e-07 elapsed=20:44:45 avg_step=1.59s recent_step=1.44s eta=01:16:21 +2026-09-21 03:23:06 | INFO | train rank=0 | step=47120/50000 progress=94.2% loss=0.119720 lr=9.043239e-07 lr_backbone=9.043239e-07 elapsed=20:45:05 avg_step=1.59s recent_step=2.00s eta=01:16:06 +2026-09-21 03:23:22 | INFO | train rank=0 | step=47130/50000 progress=94.3% loss=0.197785 lr=8.980736e-07 lr_backbone=8.980736e-07 elapsed=20:45:22 avg_step=1.59s recent_step=1.67s eta=01:15:50 +2026-09-21 03:23:39 | INFO | train rank=0 | step=47140/50000 progress=94.3% loss=0.065072 lr=8.918448e-07 lr_backbone=8.918448e-07 elapsed=20:45:38 avg_step=1.59s recent_step=1.62s eta=01:15:34 +2026-09-21 03:23:52 | INFO | train rank=0 | step=47150/50000 progress=94.3% loss=0.087451 lr=8.856375e-07 lr_backbone=8.856375e-07 elapsed=20:45:52 avg_step=1.59s recent_step=1.39s eta=01:15:18 +2026-09-21 03:24:11 | INFO | train rank=0 | step=47160/50000 progress=94.3% loss=0.084685 lr=8.794516e-07 lr_backbone=8.794516e-07 elapsed=20:46:10 avg_step=1.59s recent_step=1.81s eta=01:15:02 +2026-09-21 03:24:26 | INFO | train rank=0 | step=47170/50000 progress=94.3% loss=0.078693 lr=8.732873e-07 lr_backbone=8.732873e-07 elapsed=20:46:25 avg_step=1.59s recent_step=1.51s eta=01:14:46 +2026-09-21 03:24:41 | INFO | train rank=0 | step=47180/50000 progress=94.4% loss=0.079219 lr=8.671444e-07 lr_backbone=8.671444e-07 elapsed=20:46:40 avg_step=1.59s recent_step=1.52s eta=01:14:30 +2026-09-21 03:24:54 | INFO | train rank=0 | step=47190/50000 progress=94.4% loss=0.067802 lr=8.610230e-07 lr_backbone=8.610230e-07 elapsed=20:46:54 avg_step=1.59s recent_step=1.32s eta=01:14:14 +2026-09-21 03:25:13 | INFO | train rank=0 | step=47200/50000 progress=94.4% loss=0.088688 lr=8.549231e-07 lr_backbone=8.549231e-07 elapsed=20:47:12 avg_step=1.59s recent_step=1.88s eta=01:13:59 +2026-09-21 03:25:25 | INFO | train rank=0 | step=47210/50000 progress=94.4% loss=0.079091 lr=8.488448e-07 lr_backbone=8.488448e-07 elapsed=20:47:25 avg_step=1.59s recent_step=1.24s eta=01:13:43 +2026-09-21 03:25:42 | INFO | train rank=0 | step=47220/50000 progress=94.4% loss=0.066140 lr=8.427879e-07 lr_backbone=8.427879e-07 elapsed=20:47:42 avg_step=1.59s recent_step=1.72s eta=01:13:27 +2026-09-21 03:25:54 | INFO | train rank=0 | step=47230/50000 progress=94.5% loss=0.100368 lr=8.367525e-07 lr_backbone=8.367525e-07 elapsed=20:47:54 avg_step=1.59s recent_step=1.19s eta=01:13:11 +2026-09-21 03:26:12 | INFO | train rank=0 | step=47240/50000 progress=94.5% loss=0.078912 lr=8.307386e-07 lr_backbone=8.307386e-07 elapsed=20:48:12 avg_step=1.59s recent_step=1.78s eta=01:12:55 +2026-09-21 03:26:28 | INFO | train rank=0 | step=47250/50000 progress=94.5% loss=0.062825 lr=8.247463e-07 lr_backbone=8.247463e-07 elapsed=20:48:28 avg_step=1.59s recent_step=1.60s eta=01:12:39 +2026-09-21 03:26:46 | INFO | train rank=0 | step=47260/50000 progress=94.5% loss=0.089628 lr=8.187754e-07 lr_backbone=8.187754e-07 elapsed=20:48:46 avg_step=1.59s recent_step=1.82s eta=01:12:24 +2026-09-21 03:26:58 | INFO | train rank=0 | step=47270/50000 progress=94.5% loss=0.107029 lr=8.128261e-07 lr_backbone=8.128261e-07 elapsed=20:48:57 avg_step=1.59s recent_step=1.13s eta=01:12:07 +2026-09-21 03:27:15 | INFO | train rank=0 | step=47280/50000 progress=94.6% loss=0.136572 lr=8.068982e-07 lr_backbone=8.068982e-07 elapsed=20:49:14 avg_step=1.59s recent_step=1.70s eta=01:11:52 +2026-09-21 03:27:31 | INFO | train rank=0 | step=47290/50000 progress=94.6% loss=0.067285 lr=8.009919e-07 lr_backbone=8.009919e-07 elapsed=20:49:30 avg_step=1.59s recent_step=1.60s eta=01:11:36 +2026-09-21 03:27:49 | INFO | train rank=0 | step=47300/50000 progress=94.6% loss=0.068405 lr=7.951071e-07 lr_backbone=7.951071e-07 elapsed=20:49:48 avg_step=1.59s recent_step=1.82s eta=01:11:20 +2026-09-21 03:28:04 | INFO | train rank=0 | step=47310/50000 progress=94.6% loss=0.084494 lr=7.892439e-07 lr_backbone=7.892439e-07 elapsed=20:50:04 avg_step=1.59s recent_step=1.57s eta=01:11:04 +2026-09-21 03:28:21 | INFO | train rank=0 | step=47320/50000 progress=94.6% loss=0.082983 lr=7.834022e-07 lr_backbone=7.834022e-07 elapsed=20:50:21 avg_step=1.59s recent_step=1.68s eta=01:10:48 +2026-09-21 03:28:34 | INFO | train rank=0 | step=47330/50000 progress=94.7% loss=0.094004 lr=7.775819e-07 lr_backbone=7.775819e-07 elapsed=20:50:33 avg_step=1.59s recent_step=1.26s eta=01:10:32 +2026-09-21 03:28:53 | INFO | train rank=0 | step=47340/50000 progress=94.7% loss=0.097191 lr=7.717833e-07 lr_backbone=7.717833e-07 elapsed=20:50:53 avg_step=1.59s recent_step=1.97s eta=01:10:17 +2026-09-21 03:29:06 | INFO | train rank=0 | step=47350/50000 progress=94.7% loss=0.094639 lr=7.660061e-07 lr_backbone=7.660061e-07 elapsed=20:51:05 avg_step=1.59s recent_step=1.22s eta=01:10:01 +2026-09-21 03:29:19 | INFO | train rank=0 | step=47360/50000 progress=94.7% loss=0.087661 lr=7.602505e-07 lr_backbone=7.602505e-07 elapsed=20:51:19 avg_step=1.59s recent_step=1.37s eta=01:09:45 +2026-09-21 03:29:37 | INFO | train rank=0 | step=47370/50000 progress=94.7% loss=0.078116 lr=7.545165e-07 lr_backbone=7.545165e-07 elapsed=20:51:36 avg_step=1.59s recent_step=1.73s eta=01:09:29 +2026-09-21 03:29:48 | INFO | train rank=0 | step=47380/50000 progress=94.8% loss=0.086960 lr=7.488039e-07 lr_backbone=7.488039e-07 elapsed=20:51:48 avg_step=1.59s recent_step=1.16s eta=01:09:13 +2026-09-21 03:30:03 | INFO | train rank=0 | step=47390/50000 progress=94.8% loss=0.082055 lr=7.431130e-07 lr_backbone=7.431130e-07 elapsed=20:52:02 avg_step=1.59s recent_step=1.46s eta=01:08:57 +2026-09-21 03:30:23 | INFO | train rank=0 | step=47400/50000 progress=94.8% loss=0.066503 lr=7.374435e-07 lr_backbone=7.374435e-07 elapsed=20:52:23 avg_step=1.59s recent_step=2.03s eta=01:08:41 +2026-09-21 03:30:40 | INFO | train rank=0 | step=47410/50000 progress=94.8% loss=0.102621 lr=7.317957e-07 lr_backbone=7.317957e-07 elapsed=20:52:40 avg_step=1.59s recent_step=1.73s eta=01:08:26 +2026-09-21 03:30:54 | INFO | train rank=0 | step=47420/50000 progress=94.8% loss=0.078626 lr=7.261693e-07 lr_backbone=7.261693e-07 elapsed=20:52:53 avg_step=1.59s recent_step=1.32s eta=01:08:10 +2026-09-21 03:31:07 | INFO | train rank=0 | step=47430/50000 progress=94.9% loss=0.100168 lr=7.205645e-07 lr_backbone=7.205645e-07 elapsed=20:53:06 avg_step=1.59s recent_step=1.29s eta=01:07:53 +2026-09-21 03:31:26 | INFO | train rank=0 | step=47440/50000 progress=94.9% loss=0.101579 lr=7.149813e-07 lr_backbone=7.149813e-07 elapsed=20:53:26 avg_step=1.59s recent_step=1.99s eta=01:07:38 +2026-09-21 03:31:41 | INFO | train rank=0 | step=47450/50000 progress=94.9% loss=0.109338 lr=7.094197e-07 lr_backbone=7.094197e-07 elapsed=20:53:41 avg_step=1.59s recent_step=1.46s eta=01:07:22 +2026-09-21 03:31:55 | INFO | train rank=0 | step=47460/50000 progress=94.9% loss=0.078502 lr=7.038796e-07 lr_backbone=7.038796e-07 elapsed=20:53:55 avg_step=1.59s recent_step=1.44s eta=01:07:06 +2026-09-21 03:32:13 | INFO | train rank=0 | step=47470/50000 progress=94.9% loss=0.084474 lr=6.983610e-07 lr_backbone=6.983610e-07 elapsed=20:54:12 avg_step=1.59s recent_step=1.73s eta=01:06:50 +2026-09-21 03:32:29 | INFO | train rank=0 | step=47480/50000 progress=95.0% loss=0.075918 lr=6.928641e-07 lr_backbone=6.928641e-07 elapsed=20:54:29 avg_step=1.59s recent_step=1.67s eta=01:06:34 +2026-09-21 03:32:42 | INFO | train rank=0 | step=47490/50000 progress=95.0% loss=0.095700 lr=6.873887e-07 lr_backbone=6.873887e-07 elapsed=20:54:42 avg_step=1.59s recent_step=1.25s eta=01:06:18 +2026-09-21 03:32:57 | INFO | train rank=0 | step=47500/50000 progress=95.0% loss=0.091491 lr=6.819348e-07 lr_backbone=6.819348e-07 elapsed=20:54:57 avg_step=1.59s recent_step=1.50s eta=01:06:03 +2026-09-21 03:33:15 | INFO | train rank=0 | step=47510/50000 progress=95.0% loss=0.067936 lr=6.765026e-07 lr_backbone=6.765026e-07 elapsed=20:55:15 avg_step=1.59s recent_step=1.85s eta=01:05:47 +2026-09-21 03:33:30 | INFO | train rank=0 | step=47520/50000 progress=95.0% loss=0.094206 lr=6.710919e-07 lr_backbone=6.710919e-07 elapsed=20:55:29 avg_step=1.59s recent_step=1.42s eta=01:05:31 +2026-09-21 03:33:45 | INFO | train rank=0 | step=47530/50000 progress=95.1% loss=0.061826 lr=6.657028e-07 lr_backbone=6.657028e-07 elapsed=20:55:44 avg_step=1.59s recent_step=1.50s eta=01:05:15 +2026-09-21 03:34:01 | INFO | train rank=0 | step=47540/50000 progress=95.1% loss=0.104187 lr=6.603353e-07 lr_backbone=6.603353e-07 elapsed=20:56:01 avg_step=1.59s recent_step=1.66s eta=01:04:59 +2026-09-21 03:34:16 | INFO | train rank=0 | step=47550/50000 progress=95.1% loss=0.084867 lr=6.549893e-07 lr_backbone=6.549893e-07 elapsed=20:56:16 avg_step=1.59s recent_step=1.47s eta=01:04:43 +2026-09-21 03:34:35 | INFO | train rank=0 | step=47560/50000 progress=95.1% loss=0.069676 lr=6.496650e-07 lr_backbone=6.496650e-07 elapsed=20:56:35 avg_step=1.59s recent_step=1.90s eta=01:04:28 +2026-09-21 03:34:50 | INFO | train rank=0 | step=47570/50000 progress=95.1% loss=0.104143 lr=6.443622e-07 lr_backbone=6.443622e-07 elapsed=20:56:50 avg_step=1.59s recent_step=1.52s eta=01:04:12 +2026-09-21 03:35:04 | INFO | train rank=0 | step=47580/50000 progress=95.2% loss=0.098925 lr=6.390810e-07 lr_backbone=6.390810e-07 elapsed=20:57:04 avg_step=1.59s recent_step=1.43s eta=01:03:56 +2026-09-21 03:35:18 | INFO | train rank=0 | step=47590/50000 progress=95.2% loss=0.085829 lr=6.338214e-07 lr_backbone=6.338214e-07 elapsed=20:57:18 avg_step=1.59s recent_step=1.35s eta=01:03:40 +2026-09-21 03:35:39 | INFO | train rank=0 | step=47600/50000 progress=95.2% loss=0.092057 lr=6.285835e-07 lr_backbone=6.285835e-07 elapsed=20:57:39 avg_step=1.59s recent_step=2.15s eta=01:03:24 +2026-09-21 03:35:51 | INFO | train rank=0 | step=47610/50000 progress=95.2% loss=0.087733 lr=6.233671e-07 lr_backbone=6.233671e-07 elapsed=20:57:51 avg_step=1.59s recent_step=1.20s eta=01:03:08 +2026-09-21 03:36:10 | INFO | train rank=0 | step=47620/50000 progress=95.2% loss=0.103236 lr=6.181723e-07 lr_backbone=6.181723e-07 elapsed=20:58:10 avg_step=1.59s recent_step=1.85s eta=01:02:52 +2026-09-21 03:36:22 | INFO | train rank=0 | step=47630/50000 progress=95.3% loss=0.085552 lr=6.129991e-07 lr_backbone=6.129991e-07 elapsed=20:58:22 avg_step=1.59s recent_step=1.23s eta=01:02:36 +2026-09-21 03:36:35 | INFO | train rank=0 | step=47640/50000 progress=95.3% loss=0.081752 lr=6.078475e-07 lr_backbone=6.078475e-07 elapsed=20:58:35 avg_step=1.59s recent_step=1.32s eta=01:02:20 +2026-09-21 03:36:54 | INFO | train rank=0 | step=47650/50000 progress=95.3% loss=0.080233 lr=6.027175e-07 lr_backbone=6.027175e-07 elapsed=20:58:53 avg_step=1.59s recent_step=1.84s eta=01:02:05 +2026-09-21 03:37:11 | INFO | train rank=0 | step=47660/50000 progress=95.3% loss=0.084428 lr=5.976091e-07 lr_backbone=5.976091e-07 elapsed=20:59:10 avg_step=1.59s recent_step=1.68s eta=01:01:49 +2026-09-21 03:37:26 | INFO | train rank=0 | step=47670/50000 progress=95.3% loss=0.077113 lr=5.925224e-07 lr_backbone=5.925224e-07 elapsed=20:59:26 avg_step=1.59s recent_step=1.57s eta=01:01:33 +2026-09-21 03:37:46 | INFO | train rank=0 | step=47680/50000 progress=95.4% loss=0.079540 lr=5.874572e-07 lr_backbone=5.874572e-07 elapsed=20:59:45 avg_step=1.59s recent_step=1.95s eta=01:01:17 +2026-09-21 03:37:58 | INFO | train rank=0 | step=47690/50000 progress=95.4% loss=0.062852 lr=5.824137e-07 lr_backbone=5.824137e-07 elapsed=20:59:58 avg_step=1.59s recent_step=1.20s eta=01:01:01 +2026-09-21 03:38:10 | INFO | train rank=0 | step=47700/50000 progress=95.4% loss=0.087869 lr=5.773917e-07 lr_backbone=5.773917e-07 elapsed=21:00:10 avg_step=1.59s recent_step=1.21s eta=01:00:45 +2026-09-21 03:38:24 | INFO | train rank=0 | step=47710/50000 progress=95.4% loss=0.068917 lr=5.723914e-07 lr_backbone=5.723914e-07 elapsed=21:00:23 avg_step=1.59s recent_step=1.39s eta=01:00:29 +2026-09-21 03:38:38 | INFO | train rank=0 | step=47720/50000 progress=95.4% loss=0.069950 lr=5.674128e-07 lr_backbone=5.674128e-07 elapsed=21:00:37 avg_step=1.59s recent_step=1.40s eta=01:00:13 +2026-09-21 03:38:56 | INFO | train rank=0 | step=47730/50000 progress=95.5% loss=0.082164 lr=5.624557e-07 lr_backbone=5.624557e-07 elapsed=21:00:55 avg_step=1.59s recent_step=1.77s eta=00:59:58 +2026-09-21 03:39:18 | INFO | train rank=0 | step=47740/50000 progress=95.5% loss=0.107648 lr=5.575203e-07 lr_backbone=5.575203e-07 elapsed=21:01:18 avg_step=1.59s recent_step=2.24s eta=00:59:42 +2026-09-21 03:39:30 | INFO | train rank=0 | step=47750/50000 progress=95.5% loss=0.071879 lr=5.526065e-07 lr_backbone=5.526065e-07 elapsed=21:01:30 avg_step=1.59s recent_step=1.22s eta=00:59:26 +2026-09-21 03:39:44 | INFO | train rank=0 | step=47760/50000 progress=95.5% loss=0.089352 lr=5.477143e-07 lr_backbone=5.477143e-07 elapsed=21:01:44 avg_step=1.59s recent_step=1.38s eta=00:59:10 +2026-09-21 03:40:01 | INFO | train rank=0 | step=47770/50000 progress=95.5% loss=0.082743 lr=5.428438e-07 lr_backbone=5.428438e-07 elapsed=21:02:00 avg_step=1.59s recent_step=1.66s eta=00:58:54 +2026-09-21 03:40:17 | INFO | train rank=0 | step=47780/50000 progress=95.6% loss=0.081284 lr=5.379948e-07 lr_backbone=5.379948e-07 elapsed=21:02:16 avg_step=1.59s recent_step=1.60s eta=00:58:38 +2026-09-21 03:40:30 | INFO | train rank=0 | step=47790/50000 progress=95.6% loss=0.122455 lr=5.331676e-07 lr_backbone=5.331676e-07 elapsed=21:02:30 avg_step=1.59s recent_step=1.38s eta=00:58:23 +2026-09-21 03:40:43 | INFO | train rank=0 | step=47800/50000 progress=95.6% loss=0.079758 lr=5.283619e-07 lr_backbone=5.283619e-07 elapsed=21:02:43 avg_step=1.59s recent_step=1.29s eta=00:58:07 +2026-09-21 03:41:03 | INFO | train rank=0 | step=47810/50000 progress=95.6% loss=0.082393 lr=5.235780e-07 lr_backbone=5.235780e-07 elapsed=21:03:02 avg_step=1.59s recent_step=1.95s eta=00:57:51 +2026-09-21 03:41:15 | INFO | train rank=0 | step=47820/50000 progress=95.6% loss=0.073017 lr=5.188156e-07 lr_backbone=5.188156e-07 elapsed=21:03:14 avg_step=1.59s recent_step=1.18s eta=00:57:35 +2026-09-21 03:41:33 | INFO | train rank=0 | step=47830/50000 progress=95.7% loss=0.079305 lr=5.140749e-07 lr_backbone=5.140749e-07 elapsed=21:03:33 avg_step=1.59s recent_step=1.85s eta=00:57:19 +2026-09-21 03:41:51 | INFO | train rank=0 | step=47840/50000 progress=95.7% loss=0.070393 lr=5.093558e-07 lr_backbone=5.093558e-07 elapsed=21:03:50 avg_step=1.59s recent_step=1.74s eta=00:57:03 +2026-09-21 03:42:02 | INFO | train rank=0 | step=47850/50000 progress=95.7% loss=0.089344 lr=5.046584e-07 lr_backbone=5.046584e-07 elapsed=21:04:02 avg_step=1.59s recent_step=1.19s eta=00:56:47 +2026-09-21 03:42:18 | INFO | train rank=0 | step=47860/50000 progress=95.7% loss=0.094106 lr=4.999827e-07 lr_backbone=4.999827e-07 elapsed=21:04:18 avg_step=1.59s recent_step=1.58s eta=00:56:31 +2026-09-21 03:42:30 | INFO | train rank=0 | step=47870/50000 progress=95.7% loss=0.077876 lr=4.953286e-07 lr_backbone=4.953286e-07 elapsed=21:04:30 avg_step=1.58s recent_step=1.21s eta=00:56:15 +2026-09-21 03:42:47 | INFO | train rank=0 | step=47880/50000 progress=95.8% loss=0.100665 lr=4.906961e-07 lr_backbone=4.906961e-07 elapsed=21:04:47 avg_step=1.58s recent_step=1.69s eta=00:56:00 +2026-09-21 03:43:00 | INFO | train rank=0 | step=47890/50000 progress=95.8% loss=0.114298 lr=4.860853e-07 lr_backbone=4.860853e-07 elapsed=21:04:59 avg_step=1.58s recent_step=1.26s eta=00:55:44 +2026-09-21 03:43:15 | INFO | train rank=0 | step=47900/50000 progress=95.8% loss=0.084446 lr=4.814962e-07 lr_backbone=4.814962e-07 elapsed=21:05:15 avg_step=1.58s recent_step=1.53s eta=00:55:28 +2026-09-21 03:43:32 | INFO | train rank=0 | step=47910/50000 progress=95.8% loss=0.112358 lr=4.769287e-07 lr_backbone=4.769287e-07 elapsed=21:05:32 avg_step=1.58s recent_step=1.70s eta=00:55:12 +2026-09-21 03:43:51 | INFO | train rank=0 | step=47920/50000 progress=95.8% loss=0.077643 lr=4.723829e-07 lr_backbone=4.723829e-07 elapsed=21:05:51 avg_step=1.58s recent_step=1.89s eta=00:54:56 +2026-09-21 03:44:12 | INFO | train rank=0 | step=47930/50000 progress=95.9% loss=0.073634 lr=4.678588e-07 lr_backbone=4.678588e-07 elapsed=21:06:11 avg_step=1.59s recent_step=2.05s eta=00:54:41 +2026-09-21 03:44:29 | INFO | train rank=0 | step=47940/50000 progress=95.9% loss=0.114119 lr=4.633563e-07 lr_backbone=4.633563e-07 elapsed=21:06:28 avg_step=1.59s recent_step=1.72s eta=00:54:25 +2026-09-21 03:45:22 | INFO | train rank=0 | step=47950/50000 progress=95.9% loss=0.079795 lr=4.588755e-07 lr_backbone=4.588755e-07 elapsed=21:07:22 avg_step=1.59s recent_step=5.35s eta=00:54:11 +2026-09-21 03:45:34 | INFO | train rank=0 | step=47960/50000 progress=95.9% loss=0.077421 lr=4.544163e-07 lr_backbone=4.544163e-07 elapsed=21:07:34 avg_step=1.59s recent_step=1.20s eta=00:53:55 +2026-09-21 03:45:49 | INFO | train rank=0 | step=47970/50000 progress=95.9% loss=0.085956 lr=4.499789e-07 lr_backbone=4.499789e-07 elapsed=21:07:48 avg_step=1.59s recent_step=1.42s eta=00:53:39 +2026-09-21 03:46:01 | INFO | train rank=0 | step=47980/50000 progress=96.0% loss=0.074061 lr=4.455631e-07 lr_backbone=4.455631e-07 elapsed=21:08:00 avg_step=1.59s recent_step=1.22s eta=00:53:23 +2026-09-21 03:46:14 | INFO | train rank=0 | step=47990/50000 progress=96.0% loss=0.111911 lr=4.411690e-07 lr_backbone=4.411690e-07 elapsed=21:08:14 avg_step=1.59s recent_step=1.35s eta=00:53:07 +2026-09-21 03:46:28 | INFO | train rank=0 | step=48000/50000 progress=96.0% loss=0.077599 lr=4.367965e-07 lr_backbone=4.367965e-07 elapsed=21:08:28 avg_step=1.59s recent_step=1.37s eta=00:52:51 +2026-09-21 03:46:45 | INFO | train rank=0 | step=48010/50000 progress=96.0% loss=0.092966 lr=4.324458e-07 lr_backbone=4.324458e-07 elapsed=21:08:44 avg_step=1.59s recent_step=1.67s eta=00:52:35 +2026-09-21 03:47:00 | INFO | train rank=0 | step=48020/50000 progress=96.0% loss=0.068613 lr=4.281167e-07 lr_backbone=4.281167e-07 elapsed=21:09:00 avg_step=1.59s recent_step=1.54s eta=00:52:19 +2026-09-21 03:47:14 | INFO | train rank=0 | step=48030/50000 progress=96.1% loss=0.091806 lr=4.238093e-07 lr_backbone=4.238093e-07 elapsed=21:09:14 avg_step=1.59s recent_step=1.40s eta=00:52:03 +2026-09-21 03:47:27 | INFO | train rank=0 | step=48040/50000 progress=96.1% loss=0.092548 lr=4.195236e-07 lr_backbone=4.195236e-07 elapsed=21:09:27 avg_step=1.59s recent_step=1.30s eta=00:51:47 +2026-09-21 03:47:46 | INFO | train rank=0 | step=48050/50000 progress=96.1% loss=0.081657 lr=4.152596e-07 lr_backbone=4.152596e-07 elapsed=21:09:45 avg_step=1.59s recent_step=1.88s eta=00:51:31 +2026-09-21 03:47:59 | INFO | train rank=0 | step=48060/50000 progress=96.1% loss=0.079363 lr=4.110173e-07 lr_backbone=4.110173e-07 elapsed=21:09:58 avg_step=1.59s recent_step=1.29s eta=00:51:15 +2026-09-21 03:48:18 | INFO | train rank=0 | step=48070/50000 progress=96.1% loss=0.090508 lr=4.067967e-07 lr_backbone=4.067967e-07 elapsed=21:10:18 avg_step=1.59s recent_step=1.91s eta=00:51:00 +2026-09-21 03:48:35 | INFO | train rank=0 | step=48080/50000 progress=96.2% loss=0.101117 lr=4.025977e-07 lr_backbone=4.025977e-07 elapsed=21:10:35 avg_step=1.59s recent_step=1.74s eta=00:50:44 +2026-09-21 03:48:48 | INFO | train rank=0 | step=48090/50000 progress=96.2% loss=0.102694 lr=3.984205e-07 lr_backbone=3.984205e-07 elapsed=21:10:48 avg_step=1.59s recent_step=1.31s eta=00:50:28 +2026-09-21 03:49:09 | INFO | train rank=0 | step=48100/50000 progress=96.2% loss=0.089620 lr=3.942649e-07 lr_backbone=3.942649e-07 elapsed=21:11:08 avg_step=1.59s recent_step=2.02s eta=00:50:12 +2026-09-21 03:49:23 | INFO | train rank=0 | step=48110/50000 progress=96.2% loss=0.074275 lr=3.901311e-07 lr_backbone=3.901311e-07 elapsed=21:11:23 avg_step=1.59s recent_step=1.46s eta=00:49:56 +2026-09-21 03:49:37 | INFO | train rank=0 | step=48120/50000 progress=96.2% loss=0.081264 lr=3.860189e-07 lr_backbone=3.860189e-07 elapsed=21:11:36 avg_step=1.59s recent_step=1.33s eta=00:49:40 +2026-09-21 03:49:49 | INFO | train rank=0 | step=48130/50000 progress=96.3% loss=0.082836 lr=3.819285e-07 lr_backbone=3.819285e-07 elapsed=21:11:49 avg_step=1.59s recent_step=1.30s eta=00:49:24 +2026-09-21 03:50:07 | INFO | train rank=0 | step=48140/50000 progress=96.3% loss=0.070577 lr=3.778598e-07 lr_backbone=3.778598e-07 elapsed=21:12:07 avg_step=1.59s recent_step=1.78s eta=00:49:09 +2026-09-21 03:50:24 | INFO | train rank=0 | step=48150/50000 progress=96.3% loss=0.070029 lr=3.738127e-07 lr_backbone=3.738127e-07 elapsed=21:12:24 avg_step=1.59s recent_step=1.71s eta=00:48:53 +2026-09-21 03:50:48 | INFO | train rank=0 | step=48160/50000 progress=96.3% loss=0.093391 lr=3.697874e-07 lr_backbone=3.697874e-07 elapsed=21:12:48 avg_step=1.59s recent_step=2.39s eta=00:48:37 +2026-09-21 03:51:00 | INFO | train rank=0 | step=48170/50000 progress=96.3% loss=0.076228 lr=3.657838e-07 lr_backbone=3.657838e-07 elapsed=21:13:00 avg_step=1.59s recent_step=1.20s eta=00:48:21 +2026-09-21 03:51:12 | INFO | train rank=0 | step=48180/50000 progress=96.4% loss=0.093694 lr=3.618019e-07 lr_backbone=3.618019e-07 elapsed=21:13:12 avg_step=1.59s recent_step=1.18s eta=00:48:05 +2026-09-21 03:51:25 | INFO | train rank=0 | step=48190/50000 progress=96.4% loss=0.064917 lr=3.578417e-07 lr_backbone=3.578417e-07 elapsed=21:13:25 avg_step=1.59s recent_step=1.31s eta=00:47:49 +2026-09-21 03:51:46 | INFO | train rank=0 | step=48200/50000 progress=96.4% loss=0.088282 lr=3.539033e-07 lr_backbone=3.539033e-07 elapsed=21:13:46 avg_step=1.59s recent_step=2.10s eta=00:47:34 +2026-09-21 03:52:01 | INFO | train rank=0 | step=48210/50000 progress=96.4% loss=0.084601 lr=3.499865e-07 lr_backbone=3.499865e-07 elapsed=21:14:01 avg_step=1.59s recent_step=1.54s eta=00:47:18 +2026-09-21 03:52:16 | INFO | train rank=0 | step=48220/50000 progress=96.4% loss=0.102387 lr=3.460915e-07 lr_backbone=3.460915e-07 elapsed=21:14:15 avg_step=1.59s recent_step=1.43s eta=00:47:02 +2026-09-21 03:52:33 | INFO | train rank=0 | step=48230/50000 progress=96.5% loss=0.078546 lr=3.422181e-07 lr_backbone=3.422181e-07 elapsed=21:14:33 avg_step=1.59s recent_step=1.77s eta=00:46:46 +2026-09-21 03:52:48 | INFO | train rank=0 | step=48240/50000 progress=96.5% loss=0.133160 lr=3.383666e-07 lr_backbone=3.383666e-07 elapsed=21:14:48 avg_step=1.59s recent_step=1.49s eta=00:46:30 +2026-09-21 03:53:06 | INFO | train rank=0 | step=48250/50000 progress=96.5% loss=0.096542 lr=3.345367e-07 lr_backbone=3.345367e-07 elapsed=21:15:05 avg_step=1.59s recent_step=1.74s eta=00:46:14 +2026-09-21 03:53:24 | INFO | train rank=0 | step=48260/50000 progress=96.5% loss=0.065893 lr=3.307285e-07 lr_backbone=3.307285e-07 elapsed=21:15:23 avg_step=1.59s recent_step=1.77s eta=00:45:59 +2026-09-21 03:53:39 | INFO | train rank=0 | step=48270/50000 progress=96.5% loss=0.084654 lr=3.269421e-07 lr_backbone=3.269421e-07 elapsed=21:15:39 avg_step=1.59s recent_step=1.54s eta=00:45:43 +2026-09-21 03:53:52 | INFO | train rank=0 | step=48280/50000 progress=96.6% loss=0.095692 lr=3.231774e-07 lr_backbone=3.231774e-07 elapsed=21:15:51 avg_step=1.59s recent_step=1.27s eta=00:45:27 +2026-09-21 03:54:11 | INFO | train rank=0 | step=48290/50000 progress=96.6% loss=0.084720 lr=3.194345e-07 lr_backbone=3.194345e-07 elapsed=21:16:10 avg_step=1.59s recent_step=1.89s eta=00:45:11 +2026-09-21 03:54:29 | INFO | train rank=0 | step=48300/50000 progress=96.6% loss=0.126798 lr=3.157132e-07 lr_backbone=3.157132e-07 elapsed=21:16:29 avg_step=1.59s recent_step=1.84s eta=00:44:55 +2026-09-21 03:54:45 | INFO | train rank=0 | step=48310/50000 progress=96.6% loss=0.083878 lr=3.120138e-07 lr_backbone=3.120138e-07 elapsed=21:16:45 avg_step=1.59s recent_step=1.64s eta=00:44:39 +2026-09-21 03:55:00 | INFO | train rank=0 | step=48320/50000 progress=96.6% loss=0.081975 lr=3.083360e-07 lr_backbone=3.083360e-07 elapsed=21:16:59 avg_step=1.59s recent_step=1.43s eta=00:44:23 +2026-09-21 03:55:16 | INFO | train rank=0 | step=48330/50000 progress=96.7% loss=0.068364 lr=3.046800e-07 lr_backbone=3.046800e-07 elapsed=21:17:16 avg_step=1.59s recent_step=1.68s eta=00:44:08 +2026-09-21 03:55:29 | INFO | train rank=0 | step=48340/50000 progress=96.7% loss=0.065714 lr=3.010457e-07 lr_backbone=3.010457e-07 elapsed=21:17:28 avg_step=1.59s recent_step=1.22s eta=00:43:52 +2026-09-21 03:55:44 | INFO | train rank=0 | step=48350/50000 progress=96.7% loss=0.071679 lr=2.974332e-07 lr_backbone=2.974332e-07 elapsed=21:17:44 avg_step=1.59s recent_step=1.56s eta=00:43:36 +2026-09-21 03:56:00 | INFO | train rank=0 | step=48360/50000 progress=96.7% loss=0.078677 lr=2.938424e-07 lr_backbone=2.938424e-07 elapsed=21:18:00 avg_step=1.59s recent_step=1.63s eta=00:43:20 +2026-09-21 03:56:16 | INFO | train rank=0 | step=48370/50000 progress=96.7% loss=0.083036 lr=2.902733e-07 lr_backbone=2.902733e-07 elapsed=21:18:16 avg_step=1.59s recent_step=1.58s eta=00:43:04 +2026-09-21 03:56:29 | INFO | train rank=0 | step=48380/50000 progress=96.8% loss=0.077339 lr=2.867260e-07 lr_backbone=2.867260e-07 elapsed=21:18:29 avg_step=1.59s recent_step=1.30s eta=00:42:48 +2026-09-21 03:56:47 | INFO | train rank=0 | step=48390/50000 progress=96.8% loss=0.067411 lr=2.832004e-07 lr_backbone=2.832004e-07 elapsed=21:18:46 avg_step=1.59s recent_step=1.75s eta=00:42:32 +2026-09-21 03:57:08 | INFO | train rank=0 | step=48400/50000 progress=96.8% loss=0.069399 lr=2.796966e-07 lr_backbone=2.796966e-07 elapsed=21:19:08 avg_step=1.59s recent_step=2.12s eta=00:42:17 +2026-09-21 03:57:20 | INFO | train rank=0 | step=48410/50000 progress=96.8% loss=0.077233 lr=2.762145e-07 lr_backbone=2.762145e-07 elapsed=21:19:20 avg_step=1.59s recent_step=1.21s eta=00:42:01 +2026-09-21 03:57:34 | INFO | train rank=0 | step=48420/50000 progress=96.8% loss=0.090526 lr=2.727542e-07 lr_backbone=2.727542e-07 elapsed=21:19:34 avg_step=1.59s recent_step=1.39s eta=00:41:45 +2026-09-21 03:57:53 | INFO | train rank=0 | step=48430/50000 progress=96.9% loss=0.075988 lr=2.693156e-07 lr_backbone=2.693156e-07 elapsed=21:19:52 avg_step=1.59s recent_step=1.86s eta=00:41:29 +2026-09-21 03:58:08 | INFO | train rank=0 | step=48440/50000 progress=96.9% loss=0.087372 lr=2.658988e-07 lr_backbone=2.658988e-07 elapsed=21:20:07 avg_step=1.59s recent_step=1.52s eta=00:41:13 +2026-09-21 03:58:22 | INFO | train rank=0 | step=48450/50000 progress=96.9% loss=0.097597 lr=2.625038e-07 lr_backbone=2.625038e-07 elapsed=21:20:22 avg_step=1.59s recent_step=1.43s eta=00:40:57 +2026-09-21 03:58:36 | INFO | train rank=0 | step=48460/50000 progress=96.9% loss=0.066181 lr=2.591305e-07 lr_backbone=2.591305e-07 elapsed=21:20:36 avg_step=1.59s recent_step=1.44s eta=00:40:41 +2026-09-21 03:58:53 | INFO | train rank=0 | step=48470/50000 progress=96.9% loss=0.085762 lr=2.557789e-07 lr_backbone=2.557789e-07 elapsed=21:20:52 avg_step=1.59s recent_step=1.63s eta=00:40:25 +2026-09-21 03:59:08 | INFO | train rank=0 | step=48480/50000 progress=97.0% loss=0.084385 lr=2.524492e-07 lr_backbone=2.524492e-07 elapsed=21:21:08 avg_step=1.59s recent_step=1.54s eta=00:40:10 +2026-09-21 03:59:30 | INFO | train rank=0 | step=48490/50000 progress=97.0% loss=0.077505 lr=2.491411e-07 lr_backbone=2.491411e-07 elapsed=21:21:30 avg_step=1.59s recent_step=2.19s eta=00:39:54 +2026-09-21 03:59:46 | INFO | train rank=0 | step=48500/50000 progress=97.0% loss=0.070301 lr=2.458549e-07 lr_backbone=2.458549e-07 elapsed=21:21:46 avg_step=1.59s recent_step=1.60s eta=00:39:38 +2026-09-21 03:59:58 | INFO | train rank=0 | step=48510/50000 progress=97.0% loss=0.091487 lr=2.425904e-07 lr_backbone=2.425904e-07 elapsed=21:21:57 avg_step=1.59s recent_step=1.19s eta=00:39:22 +2026-09-21 04:00:13 | INFO | train rank=0 | step=48520/50000 progress=97.0% loss=0.095842 lr=2.393477e-07 lr_backbone=2.393477e-07 elapsed=21:22:13 avg_step=1.59s recent_step=1.53s eta=00:39:06 +2026-09-21 04:00:29 | INFO | train rank=0 | step=48530/50000 progress=97.1% loss=0.072290 lr=2.361267e-07 lr_backbone=2.361267e-07 elapsed=21:22:28 avg_step=1.59s recent_step=1.56s eta=00:38:50 +2026-09-21 04:00:41 | INFO | train rank=0 | step=48540/50000 progress=97.1% loss=0.093264 lr=2.329275e-07 lr_backbone=2.329275e-07 elapsed=21:22:41 avg_step=1.59s recent_step=1.25s eta=00:38:34 +2026-09-21 04:00:55 | INFO | train rank=0 | step=48550/50000 progress=97.1% loss=0.063672 lr=2.297501e-07 lr_backbone=2.297501e-07 elapsed=21:22:55 avg_step=1.59s recent_step=1.42s eta=00:38:18 +2026-09-21 04:01:14 | INFO | train rank=0 | step=48560/50000 progress=97.1% loss=0.087433 lr=2.265944e-07 lr_backbone=2.265944e-07 elapsed=21:23:13 avg_step=1.59s recent_step=1.84s eta=00:38:03 +2026-09-21 04:01:29 | INFO | train rank=0 | step=48570/50000 progress=97.1% loss=0.093422 lr=2.234605e-07 lr_backbone=2.234605e-07 elapsed=21:23:28 avg_step=1.59s recent_step=1.49s eta=00:37:47 +2026-09-21 04:01:43 | INFO | train rank=0 | step=48580/50000 progress=97.2% loss=0.113689 lr=2.203484e-07 lr_backbone=2.203484e-07 elapsed=21:23:43 avg_step=1.59s recent_step=1.42s eta=00:37:31 +2026-09-21 04:01:56 | INFO | train rank=0 | step=48590/50000 progress=97.2% loss=0.097464 lr=2.172581e-07 lr_backbone=2.172581e-07 elapsed=21:23:56 avg_step=1.59s recent_step=1.33s eta=00:37:15 +2026-09-21 04:02:14 | INFO | train rank=0 | step=48600/50000 progress=97.2% loss=0.095410 lr=2.141896e-07 lr_backbone=2.141896e-07 elapsed=21:24:13 avg_step=1.59s recent_step=1.74s eta=00:36:59 +2026-09-21 04:02:26 | INFO | train rank=0 | step=48610/50000 progress=97.2% loss=0.072232 lr=2.111428e-07 lr_backbone=2.111428e-07 elapsed=21:24:26 avg_step=1.59s recent_step=1.27s eta=00:36:43 +2026-09-21 04:02:42 | INFO | train rank=0 | step=48620/50000 progress=97.2% loss=0.067402 lr=2.081178e-07 lr_backbone=2.081178e-07 elapsed=21:24:42 avg_step=1.59s recent_step=1.58s eta=00:36:27 +2026-09-21 04:02:55 | INFO | train rank=0 | step=48630/50000 progress=97.3% loss=0.087725 lr=2.051146e-07 lr_backbone=2.051146e-07 elapsed=21:24:55 avg_step=1.59s recent_step=1.31s eta=00:36:11 +2026-09-21 04:03:19 | INFO | train rank=0 | step=48640/50000 progress=97.3% loss=0.080987 lr=2.021331e-07 lr_backbone=2.021331e-07 elapsed=21:25:18 avg_step=1.59s recent_step=2.36s eta=00:35:56 +2026-09-21 04:03:32 | INFO | train rank=0 | step=48650/50000 progress=97.3% loss=0.087178 lr=1.991735e-07 lr_backbone=1.991735e-07 elapsed=21:25:32 avg_step=1.59s recent_step=1.34s eta=00:35:40 +2026-09-21 04:03:49 | INFO | train rank=0 | step=48660/50000 progress=97.3% loss=0.087235 lr=1.962356e-07 lr_backbone=1.962356e-07 elapsed=21:25:48 avg_step=1.59s recent_step=1.65s eta=00:35:24 +2026-09-21 04:04:09 | INFO | train rank=0 | step=48670/50000 progress=97.3% loss=0.080120 lr=1.933195e-07 lr_backbone=1.933195e-07 elapsed=21:26:09 avg_step=1.59s recent_step=2.06s eta=00:35:08 +2026-09-21 04:04:24 | INFO | train rank=0 | step=48680/50000 progress=97.4% loss=0.095231 lr=1.904253e-07 lr_backbone=1.904253e-07 elapsed=21:26:23 avg_step=1.59s recent_step=1.43s eta=00:34:52 +2026-09-21 04:04:42 | INFO | train rank=0 | step=48690/50000 progress=97.4% loss=0.133047 lr=1.875527e-07 lr_backbone=1.875527e-07 elapsed=21:26:41 avg_step=1.59s recent_step=1.80s eta=00:34:37 +2026-09-21 04:04:59 | INFO | train rank=0 | step=48700/50000 progress=97.4% loss=0.064829 lr=1.847020e-07 lr_backbone=1.847020e-07 elapsed=21:26:59 avg_step=1.59s recent_step=1.77s eta=00:34:21 +2026-09-21 04:05:15 | INFO | train rank=0 | step=48710/50000 progress=97.4% loss=0.081635 lr=1.818731e-07 lr_backbone=1.818731e-07 elapsed=21:27:15 avg_step=1.59s recent_step=1.59s eta=00:34:05 +2026-09-21 04:05:27 | INFO | train rank=0 | step=48720/50000 progress=97.4% loss=0.077306 lr=1.790660e-07 lr_backbone=1.790660e-07 elapsed=21:27:27 avg_step=1.59s recent_step=1.20s eta=00:33:49 +2026-09-21 04:05:47 | INFO | train rank=0 | step=48730/50000 progress=97.5% loss=0.155922 lr=1.762806e-07 lr_backbone=1.762806e-07 elapsed=21:27:47 avg_step=1.59s recent_step=2.00s eta=00:33:33 +2026-09-21 04:06:01 | INFO | train rank=0 | step=48740/50000 progress=97.5% loss=0.098555 lr=1.735171e-07 lr_backbone=1.735171e-07 elapsed=21:28:00 avg_step=1.59s recent_step=1.37s eta=00:33:17 +2026-09-21 04:06:17 | INFO | train rank=0 | step=48750/50000 progress=97.5% loss=0.087829 lr=1.707753e-07 lr_backbone=1.707753e-07 elapsed=21:28:17 avg_step=1.59s recent_step=1.61s eta=00:33:01 +2026-09-21 04:07:18 | INFO | train rank=0 | step=48760/50000 progress=97.5% loss=0.072813 lr=1.680554e-07 lr_backbone=1.680554e-07 elapsed=21:29:17 avg_step=1.59s recent_step=6.08s eta=00:32:47 +2026-09-21 04:07:30 | INFO | train rank=0 | step=48770/50000 progress=97.5% loss=0.062596 lr=1.653572e-07 lr_backbone=1.653572e-07 elapsed=21:29:29 avg_step=1.59s recent_step=1.19s eta=00:32:31 +2026-09-21 04:07:42 | INFO | train rank=0 | step=48780/50000 progress=97.6% loss=0.078425 lr=1.626809e-07 lr_backbone=1.626809e-07 elapsed=21:29:42 avg_step=1.59s recent_step=1.23s eta=00:32:15 +2026-09-21 04:08:01 | INFO | train rank=0 | step=48790/50000 progress=97.6% loss=0.102937 lr=1.600263e-07 lr_backbone=1.600263e-07 elapsed=21:30:00 avg_step=1.59s recent_step=1.87s eta=00:31:59 +2026-09-21 04:08:17 | INFO | train rank=0 | step=48800/50000 progress=97.6% loss=0.105424 lr=1.573936e-07 lr_backbone=1.573936e-07 elapsed=21:30:17 avg_step=1.59s recent_step=1.63s eta=00:31:43 +2026-09-21 04:08:32 | INFO | train rank=0 | step=48810/50000 progress=97.6% loss=0.066725 lr=1.547826e-07 lr_backbone=1.547826e-07 elapsed=21:30:31 avg_step=1.59s recent_step=1.45s eta=00:31:27 +2026-09-21 04:08:49 | INFO | train rank=0 | step=48820/50000 progress=97.6% loss=0.074342 lr=1.521935e-07 lr_backbone=1.521935e-07 elapsed=21:30:49 avg_step=1.59s recent_step=1.78s eta=00:31:11 +2026-09-21 04:09:05 | INFO | train rank=0 | step=48830/50000 progress=97.7% loss=0.091135 lr=1.496262e-07 lr_backbone=1.496262e-07 elapsed=21:31:04 avg_step=1.59s recent_step=1.51s eta=00:30:56 +2026-09-21 04:09:19 | INFO | train rank=0 | step=48840/50000 progress=97.7% loss=0.097513 lr=1.470806e-07 lr_backbone=1.470806e-07 elapsed=21:31:19 avg_step=1.59s recent_step=1.47s eta=00:30:40 +2026-09-21 04:09:34 | INFO | train rank=0 | step=48850/50000 progress=97.7% loss=0.072879 lr=1.445569e-07 lr_backbone=1.445569e-07 elapsed=21:31:34 avg_step=1.59s recent_step=1.53s eta=00:30:24 +2026-09-21 04:09:51 | INFO | train rank=0 | step=48860/50000 progress=97.7% loss=0.075495 lr=1.420550e-07 lr_backbone=1.420550e-07 elapsed=21:31:50 avg_step=1.59s recent_step=1.63s eta=00:30:08 +2026-09-21 04:10:06 | INFO | train rank=0 | step=48870/50000 progress=97.7% loss=0.073237 lr=1.395749e-07 lr_backbone=1.395749e-07 elapsed=21:32:05 avg_step=1.59s recent_step=1.49s eta=00:29:52 +2026-09-21 04:10:22 | INFO | train rank=0 | step=48880/50000 progress=97.8% loss=0.093733 lr=1.371166e-07 lr_backbone=1.371166e-07 elapsed=21:32:22 avg_step=1.59s recent_step=1.63s eta=00:29:36 +2026-09-21 04:10:35 | INFO | train rank=0 | step=48890/50000 progress=97.8% loss=0.125689 lr=1.346801e-07 lr_backbone=1.346801e-07 elapsed=21:32:35 avg_step=1.59s recent_step=1.31s eta=00:29:20 +2026-09-21 04:10:55 | INFO | train rank=0 | step=48900/50000 progress=97.8% loss=0.083904 lr=1.322654e-07 lr_backbone=1.322654e-07 elapsed=21:32:54 avg_step=1.59s recent_step=1.94s eta=00:29:05 +2026-09-21 04:11:10 | INFO | train rank=0 | step=48910/50000 progress=97.8% loss=0.073811 lr=1.298726e-07 lr_backbone=1.298726e-07 elapsed=21:33:10 avg_step=1.59s recent_step=1.57s eta=00:28:49 +2026-09-21 04:11:28 | INFO | train rank=0 | step=48920/50000 progress=97.8% loss=0.111974 lr=1.275015e-07 lr_backbone=1.275015e-07 elapsed=21:33:27 avg_step=1.59s recent_step=1.74s eta=00:28:33 +2026-09-21 04:11:39 | INFO | train rank=0 | step=48930/50000 progress=97.9% loss=0.124431 lr=1.251523e-07 lr_backbone=1.251523e-07 elapsed=21:33:39 avg_step=1.59s recent_step=1.18s eta=00:28:17 +2026-09-21 04:11:51 | INFO | train rank=0 | step=48940/50000 progress=97.9% loss=0.080606 lr=1.228249e-07 lr_backbone=1.228249e-07 elapsed=21:33:51 avg_step=1.59s recent_step=1.17s eta=00:28:01 +2026-09-21 04:12:16 | INFO | train rank=0 | step=48950/50000 progress=97.9% loss=0.085261 lr=1.205193e-07 lr_backbone=1.205193e-07 elapsed=21:34:15 avg_step=1.59s recent_step=2.44s eta=00:27:45 +2026-09-21 04:12:28 | INFO | train rank=0 | step=48960/50000 progress=97.9% loss=0.102467 lr=1.182355e-07 lr_backbone=1.182355e-07 elapsed=21:34:27 avg_step=1.59s recent_step=1.22s eta=00:27:29 +2026-09-21 04:12:45 | INFO | train rank=0 | step=48970/50000 progress=97.9% loss=0.105962 lr=1.159736e-07 lr_backbone=1.159736e-07 elapsed=21:34:44 avg_step=1.59s recent_step=1.68s eta=00:27:13 +2026-09-21 04:12:57 | INFO | train rank=0 | step=48980/50000 progress=98.0% loss=0.097693 lr=1.137334e-07 lr_backbone=1.137334e-07 elapsed=21:34:56 avg_step=1.59s recent_step=1.21s eta=00:26:58 +2026-09-21 04:13:21 | INFO | train rank=0 | step=48990/50000 progress=98.0% loss=0.177130 lr=1.115151e-07 lr_backbone=1.115151e-07 elapsed=21:35:21 avg_step=1.59s recent_step=2.45s eta=00:26:42 +2026-09-21 04:13:32 | INFO | train rank=0 | step=49000/50000 progress=98.0% loss=0.094255 lr=1.093186e-07 lr_backbone=1.093186e-07 elapsed=21:35:32 avg_step=1.59s recent_step=1.11s eta=00:26:26 +2026-09-21 04:13:45 | INFO | train rank=0 | step=49010/50000 progress=98.0% loss=0.096836 lr=1.071440e-07 lr_backbone=1.071440e-07 elapsed=21:35:44 avg_step=1.59s recent_step=1.23s eta=00:26:10 +2026-09-21 04:13:56 | INFO | train rank=0 | step=49020/50000 progress=98.0% loss=0.086503 lr=1.049911e-07 lr_backbone=1.049911e-07 elapsed=21:35:56 avg_step=1.59s recent_step=1.17s eta=00:25:54 +2026-09-21 04:14:20 | INFO | train rank=0 | step=49030/50000 progress=98.1% loss=0.072501 lr=1.028601e-07 lr_backbone=1.028601e-07 elapsed=21:36:20 avg_step=1.59s recent_step=2.39s eta=00:25:38 +2026-09-21 04:14:35 | INFO | train rank=0 | step=49040/50000 progress=98.1% loss=0.110325 lr=1.007509e-07 lr_backbone=1.007509e-07 elapsed=21:36:35 avg_step=1.59s recent_step=1.49s eta=00:25:22 +2026-09-21 04:14:49 | INFO | train rank=0 | step=49050/50000 progress=98.1% loss=0.130403 lr=9.866358e-08 lr_backbone=9.866358e-08 elapsed=21:36:49 avg_step=1.59s recent_step=1.43s eta=00:25:07 +2026-09-21 04:15:01 | INFO | train rank=0 | step=49060/50000 progress=98.1% loss=0.093473 lr=9.659805e-08 lr_backbone=9.659805e-08 elapsed=21:37:01 avg_step=1.59s recent_step=1.17s eta=00:24:51 +2026-09-21 04:15:20 | INFO | train rank=0 | step=49070/50000 progress=98.1% loss=0.095754 lr=9.455435e-08 lr_backbone=9.455435e-08 elapsed=21:37:20 avg_step=1.59s recent_step=1.88s eta=00:24:35 +2026-09-21 04:15:37 | INFO | train rank=0 | step=49080/50000 progress=98.2% loss=0.081907 lr=9.253248e-08 lr_backbone=9.253248e-08 elapsed=21:37:36 avg_step=1.59s recent_step=1.66s eta=00:24:19 +2026-09-21 04:15:51 | INFO | train rank=0 | step=49090/50000 progress=98.2% loss=0.072413 lr=9.053244e-08 lr_backbone=9.053244e-08 elapsed=21:37:50 avg_step=1.59s recent_step=1.42s eta=00:24:03 +2026-09-21 04:16:05 | INFO | train rank=0 | step=49100/50000 progress=98.2% loss=0.111402 lr=8.855423e-08 lr_backbone=8.855423e-08 elapsed=21:38:04 avg_step=1.59s recent_step=1.39s eta=00:23:47 +2026-09-21 04:16:25 | INFO | train rank=0 | step=49110/50000 progress=98.2% loss=0.064296 lr=8.659786e-08 lr_backbone=8.659786e-08 elapsed=21:38:25 avg_step=1.59s recent_step=2.05s eta=00:23:31 +2026-09-21 04:16:38 | INFO | train rank=0 | step=49120/50000 progress=98.2% loss=0.099396 lr=8.466332e-08 lr_backbone=8.466332e-08 elapsed=21:38:37 avg_step=1.59s recent_step=1.26s eta=00:23:15 +2026-09-21 04:16:58 | INFO | train rank=0 | step=49130/50000 progress=98.3% loss=0.114357 lr=8.275061e-08 lr_backbone=8.275061e-08 elapsed=21:38:57 avg_step=1.59s recent_step=2.00s eta=00:23:00 +2026-09-21 04:17:09 | INFO | train rank=0 | step=49140/50000 progress=98.3% loss=0.087847 lr=8.085974e-08 lr_backbone=8.085974e-08 elapsed=21:39:09 avg_step=1.59s recent_step=1.16s eta=00:22:44 +2026-09-21 04:17:34 | INFO | train rank=0 | step=49150/50000 progress=98.3% loss=0.087653 lr=7.899071e-08 lr_backbone=7.899071e-08 elapsed=21:39:33 avg_step=1.59s recent_step=2.45s eta=00:22:28 +2026-09-21 04:17:45 | INFO | train rank=0 | step=49160/50000 progress=98.3% loss=0.073303 lr=7.714351e-08 lr_backbone=7.714351e-08 elapsed=21:39:45 avg_step=1.59s recent_step=1.12s eta=00:22:12 +2026-09-21 04:17:57 | INFO | train rank=0 | step=49170/50000 progress=98.3% loss=0.070367 lr=7.531815e-08 lr_backbone=7.531815e-08 elapsed=21:39:57 avg_step=1.59s recent_step=1.20s eta=00:21:56 +2026-09-21 04:18:16 | INFO | train rank=0 | step=49180/50000 progress=98.4% loss=0.074984 lr=7.351463e-08 lr_backbone=7.351463e-08 elapsed=21:40:16 avg_step=1.59s recent_step=1.91s eta=00:21:40 +2026-09-21 04:18:29 | INFO | train rank=0 | step=49190/50000 progress=98.4% loss=0.074370 lr=7.173295e-08 lr_backbone=7.173295e-08 elapsed=21:40:28 avg_step=1.59s recent_step=1.27s eta=00:21:24 +2026-09-21 04:18:44 | INFO | train rank=0 | step=49200/50000 progress=98.4% loss=0.076460 lr=6.997311e-08 lr_backbone=6.997311e-08 elapsed=21:40:44 avg_step=1.59s recent_step=1.54s eta=00:21:09 +2026-09-21 04:19:02 | INFO | train rank=0 | step=49210/50000 progress=98.4% loss=0.080653 lr=6.823511e-08 lr_backbone=6.823511e-08 elapsed=21:41:01 avg_step=1.59s recent_step=1.77s eta=00:20:53 +2026-09-21 04:19:13 | INFO | train rank=0 | step=49220/50000 progress=98.4% loss=0.082260 lr=6.651896e-08 lr_backbone=6.651896e-08 elapsed=21:41:13 avg_step=1.59s recent_step=1.14s eta=00:20:37 +2026-09-21 04:19:29 | INFO | train rank=0 | step=49230/50000 progress=98.5% loss=0.083807 lr=6.482464e-08 lr_backbone=6.482464e-08 elapsed=21:41:28 avg_step=1.59s recent_step=1.57s eta=00:20:21 +2026-09-21 04:19:47 | INFO | train rank=0 | step=49240/50000 progress=98.5% loss=0.096317 lr=6.315217e-08 lr_backbone=6.315217e-08 elapsed=21:41:46 avg_step=1.59s recent_step=1.78s eta=00:20:05 +2026-09-21 04:20:00 | INFO | train rank=0 | step=49250/50000 progress=98.5% loss=0.071305 lr=6.150154e-08 lr_backbone=6.150154e-08 elapsed=21:42:00 avg_step=1.59s recent_step=1.33s eta=00:19:49 +2026-09-21 04:20:23 | INFO | train rank=0 | step=49260/50000 progress=98.5% loss=0.096728 lr=5.987276e-08 lr_backbone=5.987276e-08 elapsed=21:42:22 avg_step=1.59s recent_step=2.26s eta=00:19:33 +2026-09-21 04:20:35 | INFO | train rank=0 | step=49270/50000 progress=98.5% loss=0.164621 lr=5.826582e-08 lr_backbone=5.826582e-08 elapsed=21:42:34 avg_step=1.59s recent_step=1.21s eta=00:19:17 +2026-09-21 04:20:48 | INFO | train rank=0 | step=49280/50000 progress=98.6% loss=0.114641 lr=5.668073e-08 lr_backbone=5.668073e-08 elapsed=21:42:48 avg_step=1.59s recent_step=1.35s eta=00:19:02 +2026-09-21 04:21:07 | INFO | train rank=0 | step=49290/50000 progress=98.6% loss=0.108922 lr=5.511749e-08 lr_backbone=5.511749e-08 elapsed=21:43:06 avg_step=1.59s recent_step=1.87s eta=00:18:46 +2026-09-21 04:21:23 | INFO | train rank=0 | step=49300/50000 progress=98.6% loss=0.071683 lr=5.357609e-08 lr_backbone=5.357609e-08 elapsed=21:43:22 avg_step=1.59s recent_step=1.57s eta=00:18:30 +2026-09-21 04:21:39 | INFO | train rank=0 | step=49310/50000 progress=98.6% loss=0.085634 lr=5.205655e-08 lr_backbone=5.205655e-08 elapsed=21:43:39 avg_step=1.59s recent_step=1.65s eta=00:18:14 +2026-09-21 04:21:57 | INFO | train rank=0 | step=49320/50000 progress=98.6% loss=0.098800 lr=5.055885e-08 lr_backbone=5.055885e-08 elapsed=21:43:57 avg_step=1.59s recent_step=1.80s eta=00:17:58 +2026-09-21 04:22:12 | INFO | train rank=0 | step=49330/50000 progress=98.7% loss=0.088209 lr=4.908300e-08 lr_backbone=4.908300e-08 elapsed=21:44:11 avg_step=1.59s recent_step=1.46s eta=00:17:42 +2026-09-21 04:22:23 | INFO | train rank=0 | step=49340/50000 progress=98.7% loss=0.105253 lr=4.762900e-08 lr_backbone=4.762900e-08 elapsed=21:44:23 avg_step=1.59s recent_step=1.17s eta=00:17:26 +2026-09-21 04:22:45 | INFO | train rank=0 | step=49350/50000 progress=98.7% loss=0.091623 lr=4.619685e-08 lr_backbone=4.619685e-08 elapsed=21:44:44 avg_step=1.59s recent_step=2.12s eta=00:17:11 +2026-09-21 04:22:57 | INFO | train rank=0 | step=49360/50000 progress=98.7% loss=0.078906 lr=4.478655e-08 lr_backbone=4.478655e-08 elapsed=21:44:56 avg_step=1.59s recent_step=1.23s eta=00:16:55 +2026-09-21 04:23:14 | INFO | train rank=0 | step=49370/50000 progress=98.7% loss=0.061359 lr=4.339811e-08 lr_backbone=4.339811e-08 elapsed=21:45:14 avg_step=1.59s recent_step=1.76s eta=00:16:39 +2026-09-21 04:23:29 | INFO | train rank=0 | step=49380/50000 progress=98.8% loss=0.074193 lr=4.203152e-08 lr_backbone=4.203152e-08 elapsed=21:45:29 avg_step=1.59s recent_step=1.49s eta=00:16:23 +2026-09-21 04:23:45 | INFO | train rank=0 | step=49390/50000 progress=98.8% loss=0.078785 lr=4.068678e-08 lr_backbone=4.068678e-08 elapsed=21:45:45 avg_step=1.59s recent_step=1.61s eta=00:16:07 +2026-09-21 04:23:58 | INFO | train rank=0 | step=49400/50000 progress=98.8% loss=0.096793 lr=3.936389e-08 lr_backbone=3.936389e-08 elapsed=21:45:58 avg_step=1.59s recent_step=1.26s eta=00:15:51 +2026-09-21 04:24:16 | INFO | train rank=0 | step=49410/50000 progress=98.8% loss=0.066603 lr=3.806286e-08 lr_backbone=3.806286e-08 elapsed=21:46:16 avg_step=1.59s recent_step=1.81s eta=00:15:35 +2026-09-21 04:24:32 | INFO | train rank=0 | step=49420/50000 progress=98.8% loss=0.068812 lr=3.678369e-08 lr_backbone=3.678369e-08 elapsed=21:46:31 avg_step=1.59s recent_step=1.54s eta=00:15:20 +2026-09-21 04:24:45 | INFO | train rank=0 | step=49430/50000 progress=98.9% loss=0.077920 lr=3.552637e-08 lr_backbone=3.552637e-08 elapsed=21:46:45 avg_step=1.59s recent_step=1.38s eta=00:15:04 +2026-09-21 04:25:01 | INFO | train rank=0 | step=49440/50000 progress=98.9% loss=0.093604 lr=3.429090e-08 lr_backbone=3.429090e-08 elapsed=21:47:00 avg_step=1.59s recent_step=1.53s eta=00:14:48 +2026-09-21 04:25:17 | INFO | train rank=0 | step=49450/50000 progress=98.9% loss=0.080546 lr=3.307730e-08 lr_backbone=3.307730e-08 elapsed=21:47:17 avg_step=1.59s recent_step=1.65s eta=00:14:32 +2026-09-21 04:25:31 | INFO | train rank=0 | step=49460/50000 progress=98.9% loss=0.084459 lr=3.188555e-08 lr_backbone=3.188555e-08 elapsed=21:47:30 avg_step=1.59s recent_step=1.36s eta=00:14:16 +2026-09-21 04:25:52 | INFO | train rank=0 | step=49470/50000 progress=98.9% loss=0.069974 lr=3.071566e-08 lr_backbone=3.071566e-08 elapsed=21:47:52 avg_step=1.59s recent_step=2.14s eta=00:14:00 +2026-09-21 04:26:04 | INFO | train rank=0 | step=49480/50000 progress=99.0% loss=0.083149 lr=2.956762e-08 lr_backbone=2.956762e-08 elapsed=21:48:04 avg_step=1.59s recent_step=1.21s eta=00:13:44 +2026-09-21 04:26:18 | INFO | train rank=0 | step=49490/50000 progress=99.0% loss=0.099772 lr=2.844145e-08 lr_backbone=2.844145e-08 elapsed=21:48:18 avg_step=1.59s recent_step=1.38s eta=00:13:28 +2026-09-21 04:26:38 | INFO | train rank=0 | step=49500/50000 progress=99.0% loss=0.067559 lr=2.733713e-08 lr_backbone=2.733713e-08 elapsed=21:48:38 avg_step=1.59s recent_step=2.02s eta=00:13:13 +2026-09-21 04:26:51 | INFO | train rank=0 | step=49510/50000 progress=99.0% loss=0.073680 lr=2.625468e-08 lr_backbone=2.625468e-08 elapsed=21:48:50 avg_step=1.59s recent_step=1.25s eta=00:12:57 +2026-09-21 04:27:03 | INFO | train rank=0 | step=49520/50000 progress=99.0% loss=0.069266 lr=2.519408e-08 lr_backbone=2.519408e-08 elapsed=21:49:03 avg_step=1.59s recent_step=1.24s eta=00:12:41 +2026-09-21 04:27:17 | INFO | train rank=0 | step=49530/50000 progress=99.1% loss=0.067336 lr=2.415535e-08 lr_backbone=2.415535e-08 elapsed=21:49:16 avg_step=1.59s recent_step=1.35s eta=00:12:25 +2026-09-21 04:27:38 | INFO | train rank=0 | step=49540/50000 progress=99.1% loss=0.087269 lr=2.313847e-08 lr_backbone=2.313847e-08 elapsed=21:49:38 avg_step=1.59s recent_step=2.16s eta=00:12:09 +2026-09-21 04:27:53 | INFO | train rank=0 | step=49550/50000 progress=99.1% loss=0.105277 lr=2.214346e-08 lr_backbone=2.214346e-08 elapsed=21:49:53 avg_step=1.59s recent_step=1.47s eta=00:11:53 +2026-09-21 04:28:07 | INFO | train rank=0 | step=49560/50000 progress=99.1% loss=0.071759 lr=2.117031e-08 lr_backbone=2.117031e-08 elapsed=21:50:07 avg_step=1.59s recent_step=1.44s eta=00:11:37 +2026-09-21 04:28:21 | INFO | train rank=0 | step=49570/50000 progress=99.1% loss=0.071872 lr=2.021902e-08 lr_backbone=2.021902e-08 elapsed=21:50:20 avg_step=1.59s recent_step=1.34s eta=00:11:22 +2026-09-21 04:28:39 | INFO | train rank=0 | step=49580/50000 progress=99.2% loss=0.078071 lr=1.928960e-08 lr_backbone=1.928960e-08 elapsed=21:50:39 avg_step=1.59s recent_step=1.86s eta=00:11:06 +2026-09-21 04:28:58 | INFO | train rank=0 | step=49590/50000 progress=99.2% loss=0.091965 lr=1.838204e-08 lr_backbone=1.838204e-08 elapsed=21:50:58 avg_step=1.59s recent_step=1.89s eta=00:10:50 +2026-09-21 04:29:11 | INFO | train rank=0 | step=49600/50000 progress=99.2% loss=0.100266 lr=1.749634e-08 lr_backbone=1.749634e-08 elapsed=21:51:11 avg_step=1.59s recent_step=1.28s eta=00:10:34 +2026-09-21 04:29:24 | INFO | train rank=0 | step=49610/50000 progress=99.2% loss=0.079951 lr=1.663251e-08 lr_backbone=1.663251e-08 elapsed=21:51:24 avg_step=1.59s recent_step=1.34s eta=00:10:18 +2026-09-21 04:29:43 | INFO | train rank=0 | step=49620/50000 progress=99.2% loss=0.080466 lr=1.579054e-08 lr_backbone=1.579054e-08 elapsed=21:51:43 avg_step=1.59s recent_step=1.90s eta=00:10:02 +2026-09-21 04:29:56 | INFO | train rank=0 | step=49630/50000 progress=99.3% loss=0.076601 lr=1.497043e-08 lr_backbone=1.497043e-08 elapsed=21:51:55 avg_step=1.59s recent_step=1.21s eta=00:09:46 +2026-09-21 04:30:12 | INFO | train rank=0 | step=49640/50000 progress=99.3% loss=0.074637 lr=1.417219e-08 lr_backbone=1.417219e-08 elapsed=21:52:12 avg_step=1.59s recent_step=1.63s eta=00:09:30 +2026-09-21 04:30:27 | INFO | train rank=0 | step=49650/50000 progress=99.3% loss=0.061479 lr=1.339582e-08 lr_backbone=1.339582e-08 elapsed=21:52:27 avg_step=1.59s recent_step=1.51s eta=00:09:15 +2026-09-21 04:30:42 | INFO | train rank=0 | step=49660/50000 progress=99.3% loss=0.065554 lr=1.264131e-08 lr_backbone=1.264131e-08 elapsed=21:52:41 avg_step=1.59s recent_step=1.48s eta=00:08:59 +2026-09-21 04:31:03 | INFO | train rank=0 | step=49670/50000 progress=99.3% loss=0.156069 lr=1.190867e-08 lr_backbone=1.190867e-08 elapsed=21:53:02 avg_step=1.59s recent_step=2.11s eta=00:08:43 +2026-09-21 04:31:14 | INFO | train rank=0 | step=49680/50000 progress=99.4% loss=0.078930 lr=1.119789e-08 lr_backbone=1.119789e-08 elapsed=21:53:14 avg_step=1.59s recent_step=1.17s eta=00:08:27 +2026-09-21 04:31:26 | INFO | train rank=0 | step=49690/50000 progress=99.4% loss=0.081980 lr=1.050898e-08 lr_backbone=1.050898e-08 elapsed=21:53:26 avg_step=1.59s recent_step=1.19s eta=00:08:11 +2026-09-21 04:31:39 | INFO | train rank=0 | step=49700/50000 progress=99.4% loss=0.071802 lr=9.841942e-09 lr_backbone=9.841942e-09 elapsed=21:53:39 avg_step=1.59s recent_step=1.31s eta=00:07:55 +2026-09-21 04:31:58 | INFO | train rank=0 | step=49710/50000 progress=99.4% loss=0.080395 lr=9.196768e-09 lr_backbone=9.196768e-09 elapsed=21:53:58 avg_step=1.59s recent_step=1.87s eta=00:07:39 +2026-09-21 04:32:15 | INFO | train rank=0 | step=49720/50000 progress=99.4% loss=0.073069 lr=8.573461e-09 lr_backbone=8.573461e-09 elapsed=21:54:15 avg_step=1.59s recent_step=1.74s eta=00:07:24 +2026-09-21 04:32:34 | INFO | train rank=0 | step=49730/50000 progress=99.5% loss=0.082774 lr=7.972023e-09 lr_backbone=7.972023e-09 elapsed=21:54:34 avg_step=1.59s recent_step=1.87s eta=00:07:08 +2026-09-21 04:32:46 | INFO | train rank=0 | step=49740/50000 progress=99.5% loss=0.086669 lr=7.392452e-09 lr_backbone=7.392452e-09 elapsed=21:54:46 avg_step=1.59s recent_step=1.18s eta=00:06:52 +2026-09-21 04:32:58 | INFO | train rank=0 | step=49750/50000 progress=99.5% loss=0.114353 lr=6.834750e-09 lr_backbone=6.834750e-09 elapsed=21:54:57 avg_step=1.59s recent_step=1.19s eta=00:06:36 +2026-09-21 04:33:11 | INFO | train rank=0 | step=49760/50000 progress=99.5% loss=0.102164 lr=6.298917e-09 lr_backbone=6.298917e-09 elapsed=21:55:11 avg_step=1.59s recent_step=1.33s eta=00:06:20 +2026-09-21 04:33:28 | INFO | train rank=0 | step=49770/50000 progress=99.5% loss=0.092517 lr=5.784953e-09 lr_backbone=5.784953e-09 elapsed=21:55:27 avg_step=1.59s recent_step=1.65s eta=00:06:04 +2026-09-21 04:33:42 | INFO | train rank=0 | step=49780/50000 progress=99.6% loss=0.101579 lr=5.292858e-09 lr_backbone=5.292858e-09 elapsed=21:55:41 avg_step=1.59s recent_step=1.41s eta=00:05:48 +2026-09-21 04:33:59 | INFO | train rank=0 | step=49790/50000 progress=99.6% loss=0.068302 lr=4.822632e-09 lr_backbone=4.822632e-09 elapsed=21:55:59 avg_step=1.59s recent_step=1.76s eta=00:05:33 +2026-09-21 04:34:27 | INFO | train rank=0 | step=49800/50000 progress=99.6% loss=0.101513 lr=4.374276e-09 lr_backbone=4.374276e-09 elapsed=21:56:27 avg_step=1.59s recent_step=2.77s eta=00:05:17 +2026-09-21 04:35:16 | INFO | train rank=0 | step=49810/50000 progress=99.6% loss=0.090607 lr=3.947790e-09 lr_backbone=3.947790e-09 elapsed=21:57:16 avg_step=1.59s recent_step=4.92s eta=00:05:01 +2026-09-21 04:35:28 | INFO | train rank=0 | step=49820/50000 progress=99.6% loss=0.071899 lr=3.543173e-09 lr_backbone=3.543173e-09 elapsed=21:57:28 avg_step=1.59s recent_step=1.16s eta=00:04:45 +2026-09-21 04:35:42 | INFO | train rank=0 | step=49830/50000 progress=99.7% loss=0.072440 lr=3.160427e-09 lr_backbone=3.160427e-09 elapsed=21:57:41 avg_step=1.59s recent_step=1.39s eta=00:04:29 +2026-09-21 04:35:58 | INFO | train rank=0 | step=49840/50000 progress=99.7% loss=0.075543 lr=2.799551e-09 lr_backbone=2.799551e-09 elapsed=21:57:58 avg_step=1.59s recent_step=1.66s eta=00:04:13 +2026-09-21 04:36:10 | INFO | train rank=0 | step=49850/50000 progress=99.7% loss=0.083310 lr=2.460546e-09 lr_backbone=2.460546e-09 elapsed=21:58:10 avg_step=1.59s recent_step=1.18s eta=00:03:57 +2026-09-21 04:36:25 | INFO | train rank=0 | step=49860/50000 progress=99.7% loss=0.105071 lr=2.143411e-09 lr_backbone=2.143411e-09 elapsed=21:58:25 avg_step=1.59s recent_step=1.50s eta=00:03:42 +2026-09-21 04:36:46 | INFO | train rank=0 | step=49870/50000 progress=99.7% loss=0.086533 lr=1.848147e-09 lr_backbone=1.848147e-09 elapsed=21:58:45 avg_step=1.59s recent_step=2.04s eta=00:03:26 +2026-09-21 04:36:57 | INFO | train rank=0 | step=49880/50000 progress=99.8% loss=0.077352 lr=1.574754e-09 lr_backbone=1.574754e-09 elapsed=21:58:57 avg_step=1.59s recent_step=1.19s eta=00:03:10 +2026-09-21 04:37:09 | INFO | train rank=0 | step=49890/50000 progress=99.8% loss=0.065187 lr=1.323232e-09 lr_backbone=1.323232e-09 elapsed=21:59:09 avg_step=1.59s recent_step=1.21s eta=00:02:54 +2026-09-21 04:37:25 | INFO | train rank=0 | step=49900/50000 progress=99.8% loss=0.114528 lr=1.093581e-09 lr_backbone=1.093581e-09 elapsed=21:59:24 avg_step=1.59s recent_step=1.54s eta=00:02:38 +2026-09-21 04:37:44 | INFO | train rank=0 | step=49910/50000 progress=99.8% loss=0.068169 lr=8.858012e-10 lr_backbone=8.858012e-10 elapsed=21:59:44 avg_step=1.59s recent_step=1.92s eta=00:02:22 +2026-09-21 04:37:56 | INFO | train rank=0 | step=49920/50000 progress=99.8% loss=0.086217 lr=6.998928e-10 lr_backbone=6.998928e-10 elapsed=21:59:55 avg_step=1.59s recent_step=1.15s eta=00:02:06 +2026-09-21 04:38:11 | INFO | train rank=0 | step=49930/50000 progress=99.9% loss=0.081629 lr=5.358557e-10 lr_backbone=5.358557e-10 elapsed=22:00:11 avg_step=1.59s recent_step=1.54s eta=00:01:51 +2026-09-21 04:38:27 | INFO | train rank=0 | step=49940/50000 progress=99.9% loss=0.072919 lr=3.936901e-10 lr_backbone=3.936901e-10 elapsed=22:00:27 avg_step=1.59s recent_step=1.64s eta=00:01:35 +2026-09-21 04:38:40 | INFO | train rank=0 | step=49950/50000 progress=99.9% loss=0.076126 lr=2.733960e-10 lr_backbone=2.733960e-10 elapsed=22:00:40 avg_step=1.59s recent_step=1.25s eta=00:01:19 +2026-09-21 04:38:59 | INFO | train rank=0 | step=49960/50000 progress=99.9% loss=0.070639 lr=1.749735e-10 lr_backbone=1.749735e-10 elapsed=22:00:59 avg_step=1.59s recent_step=1.92s eta=00:01:03 +2026-09-21 04:39:15 | INFO | train rank=0 | step=49970/50000 progress=99.9% loss=0.088223 lr=9.842262e-11 lr_backbone=9.842262e-11 elapsed=22:01:14 avg_step=1.59s recent_step=1.57s eta=00:00:47 +2026-09-21 04:39:27 | INFO | train rank=0 | step=49980/50000 progress=100.0% loss=0.082739 lr=4.374339e-11 lr_backbone=4.374339e-11 elapsed=22:01:26 avg_step=1.59s recent_step=1.19s eta=00:00:31 +2026-09-21 04:39:45 | INFO | train rank=0 | step=49990/50000 progress=100.0% loss=0.074667 lr=1.093585e-11 lr_backbone=1.093585e-11 elapsed=22:01:45 avg_step=1.59s recent_step=1.88s eta=00:00:15 +2026-09-21 04:40:00 | INFO | train rank=0 | step=50000/50000 progress=100.0% loss=0.064867 lr=0.000000e+00 lr_backbone=0.000000e+00 elapsed=22:02:00 avg_step=1.59s recent_step=1.47s eta=00:00:00 +2026-09-21 04:40:24 | INFO | train rank=0 | Saved DeepSpeed checkpoint to /data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_50000 +2026-09-21 04:40:24 | INFO | train rank=0 | Finished training final_step=50000/50000 elapsed=22:02:23 avg_step=1.59s +2026-09-21 04:40:24 | INFO | train rank=0 | Training finished +2026-09-21 04:40:43 | INFO | train rank=0 | Saved final DeepSpeed checkpoint to /data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_qwen3vl2b_end_padding_mask_chunk16_using_dreamzero_dataset/ds_checkpoint_final diff --git a/joint_50ksteps_bs128_wan5b_end_padding_mask_chunk16_downsample2_new_instruction/train.log b/joint_50ksteps_bs128_wan5b_end_padding_mask_chunk16_downsample2_new_instruction/train.log new file mode 100644 index 0000000000000000000000000000000000000000..e1a2b41b227ac91b5a6f50d20ed2aa62d8891c99 --- /dev/null +++ b/joint_50ksteps_bs128_wan5b_end_padding_mask_chunk16_downsample2_new_instruction/train.log @@ -0,0 +1,37 @@ +2026-09-17 17:45:01 | INFO | train rank=0 | Starting training project=codebase_v8_droid_joint_50ksteps_bs128_wan5b_end_padding_mask_chunk16_downsample2_new_instruction save_dir=/data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_wan5b_end_padding_mask_chunk16_downsample2_new_instruction distributed=True deepspeed=True world_size=8 device=cuda:0 +2026-09-18 01:54:04 | INFO | train rank=0 | Starting training project=codebase_v8_droid_joint_50ksteps_bs128_wan5b_end_padding_mask_chunk16_downsample2_new_instruction save_dir=/data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_wan5b_end_padding_mask_chunk16_downsample2_new_instruction distributed=True deepspeed=True world_size=8 device=cuda:0 +2026-09-18 02:00:10 | INFO | train rank=0 | No pretrained checkpoint provided; training from scratch +2026-09-18 02:00:10 | INFO | train rank=0 | Initializing Droid LeRobot Dataset: path=/data/NTU_slab/draven/data/DROID fps=15.00 use_relabelled_instruction=True +2026-09-18 02:00:10 | INFO | train rank=0 | Batch config: total_batch_size=128 world_size=8 grad_acc_steps=1 per_device_batch_size=16 +2026-09-18 02:00:10 | INFO | train rank=0 | Dataset image_resize_size=320 +2026-09-18 02:00:10 | INFO | train rank=0 | Droid uses parquet-shard streaming with a per-rank shuffle buffer: batch_size=16 buffer_size=1024 num_workers=8 drop_last=True +2026-09-18 02:00:18 | INFO | train rank=0 | Droid streaming layout: parquet_shards=50 active_workers=8 worker_buffer_min=128 worker_buffer_max=128 max_active_shards=2 +2026-09-18 02:00:18 | INFO | train rank=0 | Resolved Droid joint action stats from /data/NTU_slab/draven/data/DROID/meta/joint_delta_stats_h16.json: q01=[-0.3050941824913025, -0.47471314668655396, -0.3020089268684387, -0.493502140045166, -0.4844675362110138, -0.4544465243816376, -0.550822913646698] q99=[0.30148017406463623, 0.5795503258705139, 0.3001617193222046, 0.5117637515068054, 0.4750978946685791, 0.5400477647781372, 0.5597710609436035] +2026-09-18 02:00:18 | INFO | train rank=0 | Trainable params: 5,624,093,896 / 12,009,742,052 +2026-09-18 02:03:12 | INFO | train rank=0 | Starting training start_step=0 max_steps=50000 remaining_steps=50000 log_interval=10 save_interval=50000 +2026-09-20 05:45:21 | INFO | train rank=0 | Starting training project=codebase_v8_droid_joint_50ksteps_bs128_wan5b_end_padding_mask_chunk16_downsample2_new_instruction save_dir=/data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_wan5b_end_padding_mask_chunk16_downsample2_new_instruction distributed=True deepspeed=True world_size=8 device=cuda:0 +2026-09-20 05:51:25 | INFO | train rank=0 | No pretrained checkpoint provided; training from scratch +2026-09-20 05:51:25 | INFO | train rank=0 | Initializing Droid LeRobot Dataset: path=/data/NTU_slab/draven/data/DROID fps=15.00 use_relabelled_instruction=True +2026-09-20 05:51:25 | INFO | train rank=0 | Batch config: total_batch_size=128 world_size=8 grad_acc_steps=1 per_device_batch_size=16 +2026-09-20 05:51:25 | INFO | train rank=0 | Dataset image_resize_size=320 +2026-09-20 05:51:25 | INFO | train rank=0 | Droid uses parquet-shard streaming with a per-rank shuffle buffer: batch_size=16 buffer_size=1024 num_workers=8 drop_last=True +2026-09-20 05:51:32 | INFO | train rank=0 | Droid streaming layout: parquet_shards=50 active_workers=8 worker_buffer_min=128 worker_buffer_max=128 max_active_shards=2 +2026-09-20 05:51:32 | INFO | train rank=0 | Resolved Droid joint action stats from /data/NTU_slab/draven/data/DROID/meta/joint_delta_stats_h16.json: q01=[-0.3050941824913025, -0.47471314668655396, -0.3020089268684387, -0.493502140045166, -0.4844675362110138, -0.4544465243816376, -0.550822913646698] q99=[0.30148017406463623, 0.5795503258705139, 0.3001617193222046, 0.5117637515068054, 0.4750978946685791, 0.5400477647781372, 0.5597710609436035] +2026-09-20 05:51:33 | INFO | train rank=0 | Trainable params: 5,624,093,896 / 12,009,742,052 +2026-09-20 05:54:41 | INFO | train rank=0 | Starting training start_step=0 max_steps=50000 remaining_steps=50000 log_interval=10 save_interval=50000 +2026-09-20 05:59:44 | INFO | train rank=0 | step=10/50000 progress=0.0% loss=1.228930 lr=3.600000e-07 lr_backbone=3.600000e-07 elapsed=00:05:02 avg_step=30.26s recent_step=30.26s eta=420:09:53 +2026-09-20 06:00:28 | INFO | train rank=0 | step=20/50000 progress=0.0% loss=1.166950 lr=7.600000e-07 lr_backbone=7.600000e-07 elapsed=00:05:47 avg_step=17.36s recent_step=4.46s eta=240:59:28 +2026-09-20 06:01:12 | INFO | train rank=0 | step=30/50000 progress=0.1% loss=1.228521 lr=1.160000e-06 lr_backbone=1.160000e-06 elapsed=00:06:30 avg_step=13.03s recent_step=4.37s eta=180:51:57 +2026-09-20 06:01:55 | INFO | train rank=0 | step=40/50000 progress=0.1% loss=1.176345 lr=1.560000e-06 lr_backbone=1.560000e-06 elapsed=00:07:14 avg_step=10.86s recent_step=4.34s eta=150:41:07 +2026-09-20 06:02:38 | INFO | train rank=0 | step=50/50000 progress=0.1% loss=1.151849 lr=1.960000e-06 lr_backbone=1.960000e-06 elapsed=00:07:57 avg_step=9.55s recent_step=4.30s eta=132:27:12 +2026-09-20 06:03:22 | INFO | train rank=0 | step=60/50000 progress=0.1% loss=1.175675 lr=2.360000e-06 lr_backbone=2.360000e-06 elapsed=00:08:40 avg_step=8.68s recent_step=4.33s eta=120:22:29 +2026-09-20 06:04:05 | INFO | train rank=0 | step=70/50000 progress=0.1% loss=1.227936 lr=2.760000e-06 lr_backbone=2.760000e-06 elapsed=00:09:23 avg_step=8.05s recent_step=4.32s eta=111:42:28 +2026-09-20 06:04:48 | INFO | train rank=0 | step=80/50000 progress=0.2% loss=1.194378 lr=3.160000e-06 lr_backbone=3.160000e-06 elapsed=00:10:07 avg_step=7.59s recent_step=4.33s eta=105:14:15 +2026-09-20 06:05:31 | INFO | train rank=0 | step=90/50000 progress=0.2% loss=1.212808 lr=3.560000e-06 lr_backbone=3.560000e-06 elapsed=00:10:50 avg_step=7.23s recent_step=4.34s eta=100:12:21 +2026-09-20 06:06:14 | INFO | train rank=0 | step=100/50000 progress=0.2% loss=1.163769 lr=3.960000e-06 lr_backbone=3.960000e-06 elapsed=00:11:33 avg_step=6.93s recent_step=4.28s eta=96:06:02 +2026-09-20 06:06:56 | INFO | train rank=0 | step=110/50000 progress=0.2% loss=1.167954 lr=4.360000e-06 lr_backbone=4.360000e-06 elapsed=00:12:15 avg_step=6.68s recent_step=4.18s eta=92:36:45 +2026-09-20 06:07:38 | INFO | train rank=0 | step=120/50000 progress=0.2% loss=1.150114 lr=4.760000e-06 lr_backbone=4.760000e-06 elapsed=00:12:56 avg_step=6.47s recent_step=4.17s eta=89:41:29 +2026-09-20 06:08:20 | INFO | train rank=0 | step=130/50000 progress=0.3% loss=1.135506 lr=5.160000e-06 lr_backbone=5.160000e-06 elapsed=00:13:39 avg_step=6.30s recent_step=4.23s eta=87:17:00 +2026-09-20 06:09:03 | INFO | train rank=0 | step=140/50000 progress=0.3% loss=1.071019 lr=5.560000e-06 lr_backbone=5.560000e-06 elapsed=00:14:21 avg_step=6.16s recent_step=4.26s eta=85:15:01 +2026-09-20 06:09:45 | INFO | train rank=0 | step=150/50000 progress=0.3% loss=1.083817 lr=5.960000e-06 lr_backbone=5.960000e-06 elapsed=00:15:04 avg_step=6.03s recent_step=4.24s eta=83:28:08 +2026-09-20 06:10:28 | INFO | train rank=0 | step=160/50000 progress=0.3% loss=0.975133 lr=6.360000e-06 lr_backbone=6.360000e-06 elapsed=00:15:46 avg_step=5.92s recent_step=4.25s eta=81:54:48 diff --git a/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..24cc8ef44d1979ee11af1fa385e74694eae653fc --- /dev/null +++ b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:31c53a34601e8e86e57c75563ab01d29123ba8d97a0dbd5c027588c96c5330cc +size 8436160936 diff --git a/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..6922b3df314b03c8ea6f00c6a251938430bce024 --- /dev/null +++ b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7e65655fd2d65a592d130ccd23cab293484577db9e4ce8124de9d27a78232a9e +size 8436161896 diff --git a/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..5c3e5c78ffc9aa3d8040cba1770a3201f14b2e6d --- /dev/null +++ b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d6c379a78edbac459cb50e12f6b40627718d791ff0ca05ac413910d18c505539 +size 8436162152 diff --git a/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..3ce8cd95917ac6d81c80dcc84d678d0367cd0b9c --- /dev/null +++ b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b9898da15044559f9bf0d75a6a0e22a8241359db6662705bf2c76b8016f25442 +size 8436162728 diff --git a/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..44a04bdf5eaab1484a92adba9065a71749dbea97 --- /dev/null +++ b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:550e461ff0f47450ba079025947b55d3630ac921fbd24781739519a9d688d166 +size 8436162472 diff --git a/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..2677c1cde5d4e8c0be1b5d37f922de986520c435 --- /dev/null +++ b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:343cc814af24edfbd31646c6528de9af160f242c4ce39861eada27d764c0556a +size 8436162024 diff --git a/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..a596cce7ccc47db2c73c574be700039dfa993aac --- /dev/null +++ b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:00d4061dbdb93d1f4dfe10260bf7c8135dfca91c3575bdca83004ff0c9086128 +size 8436161576 diff --git a/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..35c864071e397fffeddca2a40d0f0c1f83ea293c --- /dev/null +++ b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5e8bccc651a4f187f010e4ab643f2ae38de262f21621190a2b6191e0d9a1d197 +size 8436162088 diff --git a/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/mp_rank_00_model_states.pt b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..debf6ca49db8fba7569e9c58ace096ce4c5882eb --- /dev/null +++ b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/global_step50000/mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ef160439571108be3d91dd60feeede9500d5cd7e7613c40829fd58c72e8c1f09 +size 24020131416 diff --git a/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/latest b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/latest new file mode 100644 index 0000000000000000000000000000000000000000..59a558c0ee32a055bd8fc16bf0d99a5950c59ac1 --- /dev/null +++ b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/latest @@ -0,0 +1 @@ +global_step50000 \ No newline at end of file diff --git a/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/zero_to_fp32.py b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/zero_to_fp32.py new file mode 100644 index 0000000000000000000000000000000000000000..f40f1f0acbbff30e8e8d29906b56aae4e8b8a138 --- /dev/null +++ b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000/zero_to_fp32.py @@ -0,0 +1,790 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: +# python zero_to_fp32.py . output_dir/ +# or +# python zero_to_fp32.py . output_dir/ --safe_serialization + +import argparse +import torch +import glob +import math +import os +import re +import gc +import json +import numpy as np +from tqdm import tqdm +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS, AUTOEP_LAYERS_KEY, + AUTOEP_LAYERS_KEY_LEGACY, AUTOEP_ZERO3_EXPERT_STATE_FORMAT_KEY, + AUTOEP_ZERO3_PARTITIONED_EXPERT_STATE_FORMAT) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def _has_autoep_zero3_partitioned_metadata(state_dict): + autoep_layers = state_dict.get(AUTOEP_LAYERS_KEY) + if autoep_layers is None: + autoep_layers = state_dict.get(AUTOEP_LAYERS_KEY_LEGACY) + if not isinstance(autoep_layers, list): + return False + return any( + isinstance(entry, dict) + and entry.get(AUTOEP_ZERO3_EXPERT_STATE_FORMAT_KEY) == AUTOEP_ZERO3_PARTITIONED_EXPERT_STATE_FORMAT + for entry in autoep_layers) + + +def _raise_if_autoep_zero3_partitioned_state(state_dict): + if _has_autoep_zero3_partitioned_metadata(state_dict): + raise NotImplementedError("zero_to_fp32 does not support AutoEP ZeRO-3 partition-native checkpoints. " + "AutoEP expert parameters are partitioned over expert replica groups, so " + "global data-parallel consolidation would produce incomplete expert tensors. " + "Use ds_to_universal.py for expert-aware conversion.") + + +def _raise_if_autoep_zero3_partitioned_checkpoint(model_files): + for file in model_files: + state_dict = torch.load(file, map_location=device, weights_only=False) + _raise_if_autoep_zero3_partitioned_state(state_dict) + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device, weights_only=False) + _raise_if_autoep_zero3_partitioned_state(state_dict) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + total_files = len(files) + state_dicts = [] + for f in tqdm(files, desc='Loading checkpoint shards'): + state_dict = torch.load(f, map_location=device, mmap=True, weights_only=False) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if ZERO_STAGE not in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + # parse_model_states rejects AutoEP ZeRO-3 partition-native checkpoints + # before the expensive optimizer-shard load below. + model_files = get_model_state_files(ds_checkpoint_dir) + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +class GatheredTensor: + """ + A pseudo tensor that collects partitioned weights. + It is more memory efficient when there are multiple groups. + """ + + def __init__(self, flat_groups, flat_groups_offset, offset, partitioned_numel, shape): + self.flat_groups = flat_groups + self.flat_groups_offset = flat_groups_offset + self.offset = offset + self.partitioned_numel = partitioned_numel + self.shape = shape + self.dtype = self.flat_groups[0][0].dtype + + def contiguous(self): + """ + Merge partitioned weights from flat_groups into a single tensor. + """ + end_idx = self.offset + self.partitioned_numel + world_size = len(self.flat_groups) + pad_flat_param_chunks = [] + + for rank_i in range(world_size): + # for each rank, we need to collect weights from related group/groups + flat_groups_at_rank_i = self.flat_groups[rank_i] + start_group_id = None + end_group_id = None + for group_id in range(len(self.flat_groups_offset)): + if self.flat_groups_offset[group_id] <= self.offset < self.flat_groups_offset[group_id + 1]: + start_group_id = group_id + if self.flat_groups_offset[group_id] < end_idx <= self.flat_groups_offset[group_id + 1]: + end_group_id = group_id + break + # collect weights from related group/groups + for group_id in range(start_group_id, end_group_id + 1): + flat_tensor = flat_groups_at_rank_i[group_id] + start_offset = self.offset - self.flat_groups_offset[group_id] + end_offset = min(end_idx, self.flat_groups_offset[group_id + 1]) - self.flat_groups_offset[group_id] + pad_flat_param_chunks.append(flat_tensor[start_offset:end_offset]) + + # collect weights from all ranks + pad_flat_param = torch.cat(pad_flat_param_chunks, dim=0) + param = pad_flat_param[:self.shape.numel()].view(self.shape).contiguous() + return param + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = sum([flat_group.numel() for flat_group in fp32_flat_groups[0]]) * world_size + + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + flat_groups_offset = [0] + list(np.cumsum([flat_tensor.numel() for flat_tensor in fp32_flat_groups[0]])) + for name, shape in tqdm(param_shapes.items(), desc='Gathering sharded weights'): + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # memory efficient tensor + tensor = GatheredTensor(fp32_flat_groups, flat_groups_offset, offset, partitioned_numel, shape) + state_dict[name] = tensor + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def to_torch_tensor(state_dict, return_empty_tensor=False): + """ + Convert state_dict of GatheredTensor to torch tensor + """ + torch_state_dict = {} + converted_tensors = {} + for name, tensor in state_dict.items(): + tensor_id = id(tensor) + if tensor_id in converted_tensors: # shared tensors + shared_tensor = torch_state_dict[converted_tensors[tensor_id]] + torch_state_dict[name] = shared_tensor + else: + converted_tensors[tensor_id] = name + if return_empty_tensor: + torch_state_dict[name] = torch.empty(tensor.shape, dtype=tensor.dtype) + else: + torch_state_dict[name] = tensor.contiguous() + return torch_state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, + tag=None, + exclude_frozen_parameters=False, + lazy_mode=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + - ``lazy_mode``: get state_dict in lazy mode. It returns a dict of pesduo tensor instead of torch tensor, which is more memory efficient. + Convert the pesduo tensor to torch tensor by ``.contiguous()`` + + Returns: + - pytorch ``state_dict`` + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + Note: the above usage may not work if your application doesn't have sufficient free CPU memory. + You may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. Or you can load state_dict in lazy mode :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, lazy_mode=True) # not on cpu + for name, lazy_tensor in state_dict.item(): + tensor = lazy_tensor.contiguous() # to cpu + print(name, tensor) + # del tensor to release memory if it no longer in use + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + state_dict = _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + if lazy_mode: + return state_dict + else: + return to_torch_tensor(state_dict) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, + output_dir, + max_shard_size="5GB", + safe_serialization=False, + tag=None, + exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_dir``: directory to the pytorch fp32 state_dict output files + - ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB + - ``safe_serialization``: whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`). + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + # Dependency pre-check + if safe_serialization: + try: + from safetensors.torch import save_file + except ImportError: + print('If you want to use `safe_serialization`, please `pip install safetensors`') + raise + if max_shard_size is not None: + try: + from huggingface_hub import split_torch_state_dict_into_shards + except ImportError: + print('If you want to use `max_shard_size`, please `pip install huggingface_hub`') + raise + + # Convert zero checkpoint to state_dict + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, + tag, + exclude_frozen_parameters, + lazy_mode=True) + + # Shard the model if it is too big. + weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin" + if max_shard_size is not None: + filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors") + # an memory-efficient approach for sharding + empty_state_dict = to_torch_tensor(state_dict, return_empty_tensor=True) + state_dict_split = split_torch_state_dict_into_shards(empty_state_dict, + filename_pattern=filename_pattern, + max_shard_size=max_shard_size) + else: + from collections import namedtuple + StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"]) + state_dict_split = StateDictSplit(is_sharded=False, + filename_to_tensors={weights_name: list(state_dict.keys())}) + + # Save the model by shard + os.makedirs(output_dir, exist_ok=True) + filename_to_tensors = state_dict_split.filename_to_tensors.items() + for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"): + shard_state_dict = {tensor_name: state_dict[tensor_name] for tensor_name in tensors} + shard_state_dict = to_torch_tensor(shard_state_dict) + output_path = os.path.join(output_dir, shard_file) + if safe_serialization: + save_file(shard_state_dict, output_path, metadata={"format": "pt"}) + else: + torch.save(shard_state_dict, output_path) + # release the memory of current shard + for tensor_name in list(shard_state_dict.keys()): + del state_dict[tensor_name] + del shard_state_dict[tensor_name] + del shard_state_dict + gc.collect() + + # Save index if sharded + if state_dict_split.is_sharded: + index = { + "metadata": state_dict_split.metadata, + "weight_map": state_dict_split.tensor_to_filename, + } + save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json" + save_index_file = os.path.join(output_dir, save_index_file) + with open(save_index_file, "w", encoding="utf-8") as f: + content = json.dumps(index, indent=2, sort_keys=True) + "\n" + f.write(content) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info("Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info("Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument("output_dir", + type=str, + help="directory to the pytorch fp32 state_dict output files" + "(e.g. path/checkpoint-12-output/)") + parser.add_argument( + "--max_shard_size", + type=str, + default="5GB", + help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size" + "lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`" + "We default it to 5GB in order for models to be able to run easily on free-tier google colab instances" + "without CPU OOM issues.") + parser.add_argument( + "--safe_serialization", + default=False, + action='store_true', + help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_dir, + max_shard_size=args.max_shard_size, + safe_serialization=args.safe_serialization, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..24cc8ef44d1979ee11af1fa385e74694eae653fc --- /dev/null +++ b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:31c53a34601e8e86e57c75563ab01d29123ba8d97a0dbd5c027588c96c5330cc +size 8436160936 diff --git a/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..6922b3df314b03c8ea6f00c6a251938430bce024 --- /dev/null +++ b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7e65655fd2d65a592d130ccd23cab293484577db9e4ce8124de9d27a78232a9e +size 8436161896 diff --git a/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..5c3e5c78ffc9aa3d8040cba1770a3201f14b2e6d --- /dev/null +++ b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d6c379a78edbac459cb50e12f6b40627718d791ff0ca05ac413910d18c505539 +size 8436162152 diff --git a/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..3ce8cd95917ac6d81c80dcc84d678d0367cd0b9c --- /dev/null +++ b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b9898da15044559f9bf0d75a6a0e22a8241359db6662705bf2c76b8016f25442 +size 8436162728 diff --git a/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..44a04bdf5eaab1484a92adba9065a71749dbea97 --- /dev/null +++ b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:550e461ff0f47450ba079025947b55d3630ac921fbd24781739519a9d688d166 +size 8436162472 diff --git a/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..2677c1cde5d4e8c0be1b5d37f922de986520c435 --- /dev/null +++ b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:343cc814af24edfbd31646c6528de9af160f242c4ce39861eada27d764c0556a +size 8436162024 diff --git a/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..a596cce7ccc47db2c73c574be700039dfa993aac --- /dev/null +++ b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:00d4061dbdb93d1f4dfe10260bf7c8135dfca91c3575bdca83004ff0c9086128 +size 8436161576 diff --git a/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..35c864071e397fffeddca2a40d0f0c1f83ea293c --- /dev/null +++ b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5e8bccc651a4f187f010e4ab643f2ae38de262f21621190a2b6191e0d9a1d197 +size 8436162088 diff --git a/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/mp_rank_00_model_states.pt b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..debf6ca49db8fba7569e9c58ace096ce4c5882eb --- /dev/null +++ b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/global_step50000/mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ef160439571108be3d91dd60feeede9500d5cd7e7613c40829fd58c72e8c1f09 +size 24020131416 diff --git a/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/latest b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/latest new file mode 100644 index 0000000000000000000000000000000000000000..59a558c0ee32a055bd8fc16bf0d99a5950c59ac1 --- /dev/null +++ b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/latest @@ -0,0 +1 @@ +global_step50000 \ No newline at end of file diff --git a/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/zero_to_fp32.py b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/zero_to_fp32.py new file mode 100644 index 0000000000000000000000000000000000000000..f40f1f0acbbff30e8e8d29906b56aae4e8b8a138 --- /dev/null +++ b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final/zero_to_fp32.py @@ -0,0 +1,790 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: +# python zero_to_fp32.py . output_dir/ +# or +# python zero_to_fp32.py . output_dir/ --safe_serialization + +import argparse +import torch +import glob +import math +import os +import re +import gc +import json +import numpy as np +from tqdm import tqdm +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS, AUTOEP_LAYERS_KEY, + AUTOEP_LAYERS_KEY_LEGACY, AUTOEP_ZERO3_EXPERT_STATE_FORMAT_KEY, + AUTOEP_ZERO3_PARTITIONED_EXPERT_STATE_FORMAT) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def _has_autoep_zero3_partitioned_metadata(state_dict): + autoep_layers = state_dict.get(AUTOEP_LAYERS_KEY) + if autoep_layers is None: + autoep_layers = state_dict.get(AUTOEP_LAYERS_KEY_LEGACY) + if not isinstance(autoep_layers, list): + return False + return any( + isinstance(entry, dict) + and entry.get(AUTOEP_ZERO3_EXPERT_STATE_FORMAT_KEY) == AUTOEP_ZERO3_PARTITIONED_EXPERT_STATE_FORMAT + for entry in autoep_layers) + + +def _raise_if_autoep_zero3_partitioned_state(state_dict): + if _has_autoep_zero3_partitioned_metadata(state_dict): + raise NotImplementedError("zero_to_fp32 does not support AutoEP ZeRO-3 partition-native checkpoints. " + "AutoEP expert parameters are partitioned over expert replica groups, so " + "global data-parallel consolidation would produce incomplete expert tensors. " + "Use ds_to_universal.py for expert-aware conversion.") + + +def _raise_if_autoep_zero3_partitioned_checkpoint(model_files): + for file in model_files: + state_dict = torch.load(file, map_location=device, weights_only=False) + _raise_if_autoep_zero3_partitioned_state(state_dict) + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device, weights_only=False) + _raise_if_autoep_zero3_partitioned_state(state_dict) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + total_files = len(files) + state_dicts = [] + for f in tqdm(files, desc='Loading checkpoint shards'): + state_dict = torch.load(f, map_location=device, mmap=True, weights_only=False) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if ZERO_STAGE not in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + # parse_model_states rejects AutoEP ZeRO-3 partition-native checkpoints + # before the expensive optimizer-shard load below. + model_files = get_model_state_files(ds_checkpoint_dir) + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +class GatheredTensor: + """ + A pseudo tensor that collects partitioned weights. + It is more memory efficient when there are multiple groups. + """ + + def __init__(self, flat_groups, flat_groups_offset, offset, partitioned_numel, shape): + self.flat_groups = flat_groups + self.flat_groups_offset = flat_groups_offset + self.offset = offset + self.partitioned_numel = partitioned_numel + self.shape = shape + self.dtype = self.flat_groups[0][0].dtype + + def contiguous(self): + """ + Merge partitioned weights from flat_groups into a single tensor. + """ + end_idx = self.offset + self.partitioned_numel + world_size = len(self.flat_groups) + pad_flat_param_chunks = [] + + for rank_i in range(world_size): + # for each rank, we need to collect weights from related group/groups + flat_groups_at_rank_i = self.flat_groups[rank_i] + start_group_id = None + end_group_id = None + for group_id in range(len(self.flat_groups_offset)): + if self.flat_groups_offset[group_id] <= self.offset < self.flat_groups_offset[group_id + 1]: + start_group_id = group_id + if self.flat_groups_offset[group_id] < end_idx <= self.flat_groups_offset[group_id + 1]: + end_group_id = group_id + break + # collect weights from related group/groups + for group_id in range(start_group_id, end_group_id + 1): + flat_tensor = flat_groups_at_rank_i[group_id] + start_offset = self.offset - self.flat_groups_offset[group_id] + end_offset = min(end_idx, self.flat_groups_offset[group_id + 1]) - self.flat_groups_offset[group_id] + pad_flat_param_chunks.append(flat_tensor[start_offset:end_offset]) + + # collect weights from all ranks + pad_flat_param = torch.cat(pad_flat_param_chunks, dim=0) + param = pad_flat_param[:self.shape.numel()].view(self.shape).contiguous() + return param + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = sum([flat_group.numel() for flat_group in fp32_flat_groups[0]]) * world_size + + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + flat_groups_offset = [0] + list(np.cumsum([flat_tensor.numel() for flat_tensor in fp32_flat_groups[0]])) + for name, shape in tqdm(param_shapes.items(), desc='Gathering sharded weights'): + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # memory efficient tensor + tensor = GatheredTensor(fp32_flat_groups, flat_groups_offset, offset, partitioned_numel, shape) + state_dict[name] = tensor + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def to_torch_tensor(state_dict, return_empty_tensor=False): + """ + Convert state_dict of GatheredTensor to torch tensor + """ + torch_state_dict = {} + converted_tensors = {} + for name, tensor in state_dict.items(): + tensor_id = id(tensor) + if tensor_id in converted_tensors: # shared tensors + shared_tensor = torch_state_dict[converted_tensors[tensor_id]] + torch_state_dict[name] = shared_tensor + else: + converted_tensors[tensor_id] = name + if return_empty_tensor: + torch_state_dict[name] = torch.empty(tensor.shape, dtype=tensor.dtype) + else: + torch_state_dict[name] = tensor.contiguous() + return torch_state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, + tag=None, + exclude_frozen_parameters=False, + lazy_mode=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + - ``lazy_mode``: get state_dict in lazy mode. It returns a dict of pesduo tensor instead of torch tensor, which is more memory efficient. + Convert the pesduo tensor to torch tensor by ``.contiguous()`` + + Returns: + - pytorch ``state_dict`` + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + Note: the above usage may not work if your application doesn't have sufficient free CPU memory. + You may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. Or you can load state_dict in lazy mode :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, lazy_mode=True) # not on cpu + for name, lazy_tensor in state_dict.item(): + tensor = lazy_tensor.contiguous() # to cpu + print(name, tensor) + # del tensor to release memory if it no longer in use + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + state_dict = _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + if lazy_mode: + return state_dict + else: + return to_torch_tensor(state_dict) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, + output_dir, + max_shard_size="5GB", + safe_serialization=False, + tag=None, + exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_dir``: directory to the pytorch fp32 state_dict output files + - ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB + - ``safe_serialization``: whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`). + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + # Dependency pre-check + if safe_serialization: + try: + from safetensors.torch import save_file + except ImportError: + print('If you want to use `safe_serialization`, please `pip install safetensors`') + raise + if max_shard_size is not None: + try: + from huggingface_hub import split_torch_state_dict_into_shards + except ImportError: + print('If you want to use `max_shard_size`, please `pip install huggingface_hub`') + raise + + # Convert zero checkpoint to state_dict + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, + tag, + exclude_frozen_parameters, + lazy_mode=True) + + # Shard the model if it is too big. + weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin" + if max_shard_size is not None: + filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors") + # an memory-efficient approach for sharding + empty_state_dict = to_torch_tensor(state_dict, return_empty_tensor=True) + state_dict_split = split_torch_state_dict_into_shards(empty_state_dict, + filename_pattern=filename_pattern, + max_shard_size=max_shard_size) + else: + from collections import namedtuple + StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"]) + state_dict_split = StateDictSplit(is_sharded=False, + filename_to_tensors={weights_name: list(state_dict.keys())}) + + # Save the model by shard + os.makedirs(output_dir, exist_ok=True) + filename_to_tensors = state_dict_split.filename_to_tensors.items() + for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"): + shard_state_dict = {tensor_name: state_dict[tensor_name] for tensor_name in tensors} + shard_state_dict = to_torch_tensor(shard_state_dict) + output_path = os.path.join(output_dir, shard_file) + if safe_serialization: + save_file(shard_state_dict, output_path, metadata={"format": "pt"}) + else: + torch.save(shard_state_dict, output_path) + # release the memory of current shard + for tensor_name in list(shard_state_dict.keys()): + del state_dict[tensor_name] + del shard_state_dict[tensor_name] + del shard_state_dict + gc.collect() + + # Save index if sharded + if state_dict_split.is_sharded: + index = { + "metadata": state_dict_split.metadata, + "weight_map": state_dict_split.tensor_to_filename, + } + save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json" + save_index_file = os.path.join(output_dir, save_index_file) + with open(save_index_file, "w", encoding="utf-8") as f: + content = json.dumps(index, indent=2, sort_keys=True) + "\n" + f.write(content) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info("Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info("Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument("output_dir", + type=str, + help="directory to the pytorch fp32 state_dict output files" + "(e.g. path/checkpoint-12-output/)") + parser.add_argument( + "--max_shard_size", + type=str, + default="5GB", + help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size" + "lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`" + "We default it to 5GB in order for models to be able to run easily on free-tier google colab instances" + "without CPU OOM issues.") + parser.add_argument( + "--safe_serialization", + default=False, + action='store_true', + help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_dir, + max_shard_size=args.max_shard_size, + safe_serialization=args.safe_serialization, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/train.log b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/train.log new file mode 100644 index 0000000000000000000000000000000000000000..f196744631480550bdcedfbe9470da4716e1ffe4 --- /dev/null +++ b/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/train.log @@ -0,0 +1,5014 @@ +2026-09-20 06:18:03 | INFO | train rank=0 | Starting training project=codebase_v8_droid_joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2 save_dir=/data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2 distributed=True deepspeed=True world_size=8 device=cuda:0 +2026-09-20 06:23:37 | INFO | train rank=0 | No pretrained checkpoint provided; training from scratch +2026-09-20 06:23:37 | INFO | train rank=0 | Initializing Droid LeRobot Dataset: path=/data/NTU_slab/draven/data/DROID fps=15.00 use_relabelled_instruction=False +2026-09-20 06:23:37 | INFO | train rank=0 | Batch config: total_batch_size=128 world_size=8 grad_acc_steps=1 per_device_batch_size=16 +2026-09-20 06:23:37 | INFO | train rank=0 | Dataset image_resize_size=320 +2026-09-20 06:23:37 | INFO | train rank=0 | Droid uses parquet-shard streaming with a per-rank shuffle buffer: batch_size=16 buffer_size=1024 num_workers=8 drop_last=True +2026-09-20 06:23:43 | INFO | train rank=0 | Droid streaming layout: parquet_shards=50 active_workers=8 worker_buffer_min=128 worker_buffer_max=128 max_active_shards=2 +2026-09-20 06:23:43 | INFO | train rank=0 | Resolved Droid joint action stats from /data/NTU_slab/draven/data/DROID/meta/joint_delta_stats_h16.json: q01=[-0.3050941824913025, -0.47471314668655396, -0.3020089268684387, -0.493502140045166, -0.4844675362110138, -0.4544465243816376, -0.550822913646698] q99=[0.30148017406463623, 0.5795503258705139, 0.3001617193222046, 0.5117637515068054, 0.4750978946685791, 0.5400477647781372, 0.5597710609436035] +2026-09-20 06:23:43 | INFO | train rank=0 | Trainable params: 5,624,093,896 / 12,009,742,052 +2026-09-20 06:26:36 | INFO | train rank=0 | Starting training start_step=0 max_steps=50000 remaining_steps=50000 log_interval=10 save_interval=50000 +2026-09-20 06:31:21 | INFO | train rank=0 | step=10/50000 progress=0.0% loss=1.853691 lr=3.600000e-07 lr_backbone=3.600000e-07 elapsed=00:04:45 avg_step=28.51s recent_step=28.51s eta=395:54:58 +2026-09-20 06:32:05 | INFO | train rank=0 | step=20/50000 progress=0.0% loss=1.781186 lr=7.600000e-07 lr_backbone=7.600000e-07 elapsed=00:05:28 avg_step=16.43s recent_step=4.35s eta=228:08:01 +2026-09-20 06:32:48 | INFO | train rank=0 | step=30/50000 progress=0.1% loss=1.640488 lr=1.160000e-06 lr_backbone=1.160000e-06 elapsed=00:06:11 avg_step=12.38s recent_step=4.28s eta=171:51:35 +2026-09-20 06:33:31 | INFO | train rank=0 | step=40/50000 progress=0.1% loss=1.506737 lr=1.560000e-06 lr_backbone=1.560000e-06 elapsed=00:06:54 avg_step=10.36s recent_step=4.32s eta=143:50:32 +2026-09-20 06:34:14 | INFO | train rank=0 | step=50/50000 progress=0.1% loss=1.496737 lr=1.960000e-06 lr_backbone=1.960000e-06 elapsed=00:07:37 avg_step=9.15s recent_step=4.28s eta=126:56:14 +2026-09-20 06:34:56 | INFO | train rank=0 | step=60/50000 progress=0.1% loss=1.467937 lr=2.360000e-06 lr_backbone=2.360000e-06 elapsed=00:08:19 avg_step=8.32s recent_step=4.20s eta=115:28:04 +2026-09-20 06:35:37 | INFO | train rank=0 | step=70/50000 progress=0.1% loss=1.485005 lr=2.760000e-06 lr_backbone=2.760000e-06 elapsed=00:09:00 avg_step=7.72s recent_step=4.13s eta=107:08:28 +2026-09-20 06:36:18 | INFO | train rank=0 | step=80/50000 progress=0.2% loss=1.476600 lr=3.160000e-06 lr_backbone=3.160000e-06 elapsed=00:09:42 avg_step=7.28s recent_step=4.14s eta=100:54:31 +2026-09-20 06:37:00 | INFO | train rank=0 | step=90/50000 progress=0.2% loss=1.523742 lr=3.560000e-06 lr_backbone=3.560000e-06 elapsed=00:10:23 avg_step=6.93s recent_step=4.14s eta=96:03:06 +2026-09-20 06:37:41 | INFO | train rank=0 | step=100/50000 progress=0.2% loss=1.458957 lr=3.960000e-06 lr_backbone=3.960000e-06 elapsed=00:11:04 avg_step=6.65s recent_step=4.13s eta=92:08:54 +2026-09-20 06:38:22 | INFO | train rank=0 | step=110/50000 progress=0.2% loss=1.461770 lr=4.360000e-06 lr_backbone=4.360000e-06 elapsed=00:11:45 avg_step=6.42s recent_step=4.10s eta=88:55:13 +2026-09-20 06:39:03 | INFO | train rank=0 | step=120/50000 progress=0.2% loss=1.461744 lr=4.760000e-06 lr_backbone=4.760000e-06 elapsed=00:12:26 avg_step=6.22s recent_step=4.10s eta=86:13:46 +2026-09-20 06:39:44 | INFO | train rank=0 | step=130/50000 progress=0.3% loss=1.378855 lr=5.160000e-06 lr_backbone=5.160000e-06 elapsed=00:13:08 avg_step=6.06s recent_step=4.13s eta=83:58:54 +2026-09-20 06:40:26 | INFO | train rank=0 | step=140/50000 progress=0.3% loss=1.338836 lr=5.560000e-06 lr_backbone=5.560000e-06 elapsed=00:13:49 avg_step=5.92s recent_step=4.12s eta=82:02:50 +2026-09-20 06:41:07 | INFO | train rank=0 | step=150/50000 progress=0.3% loss=1.352456 lr=5.960000e-06 lr_backbone=5.960000e-06 elapsed=00:14:30 avg_step=5.80s recent_step=4.12s eta=80:22:10 +2026-09-20 06:41:48 | INFO | train rank=0 | step=160/50000 progress=0.3% loss=1.231435 lr=6.360000e-06 lr_backbone=6.360000e-06 elapsed=00:15:11 avg_step=5.70s recent_step=4.13s eta=78:54:11 +2026-09-20 06:42:29 | INFO | train rank=0 | step=170/50000 progress=0.3% loss=1.247363 lr=6.760000e-06 lr_backbone=6.760000e-06 elapsed=00:15:52 avg_step=5.61s recent_step=4.11s eta=77:35:24 +2026-09-20 06:43:10 | INFO | train rank=0 | step=180/50000 progress=0.4% loss=1.114667 lr=7.160000e-06 lr_backbone=7.160000e-06 elapsed=00:16:34 avg_step=5.52s recent_step=4.13s eta=76:26:27 +2026-09-20 06:43:52 | INFO | train rank=0 | step=190/50000 progress=0.4% loss=1.045280 lr=7.560000e-06 lr_backbone=7.560000e-06 elapsed=00:17:15 avg_step=5.45s recent_step=4.15s eta=75:25:33 +2026-09-20 06:44:33 | INFO | train rank=0 | step=200/50000 progress=0.4% loss=0.980774 lr=7.960000e-06 lr_backbone=7.960000e-06 elapsed=00:17:57 avg_step=5.39s recent_step=4.14s eta=74:30:06 +2026-09-20 06:45:15 | INFO | train rank=0 | step=210/50000 progress=0.4% loss=1.024153 lr=8.360000e-06 lr_backbone=8.360000e-06 elapsed=00:18:38 avg_step=5.33s recent_step=4.15s eta=73:40:23 +2026-09-20 06:45:56 | INFO | train rank=0 | step=220/50000 progress=0.4% loss=0.925906 lr=8.760000e-06 lr_backbone=8.760000e-06 elapsed=00:19:20 avg_step=5.27s recent_step=4.15s eta=72:55:13 +2026-09-20 06:46:37 | INFO | train rank=0 | step=230/50000 progress=0.5% loss=0.836113 lr=9.160000e-06 lr_backbone=9.160000e-06 elapsed=00:20:00 avg_step=5.22s recent_step=4.07s eta=72:11:02 +2026-09-20 06:47:19 | INFO | train rank=0 | step=240/50000 progress=0.5% loss=0.769162 lr=9.560000e-06 lr_backbone=9.560000e-06 elapsed=00:20:42 avg_step=5.18s recent_step=4.15s eta=71:33:11 +2026-09-20 06:48:00 | INFO | train rank=0 | step=250/50000 progress=0.5% loss=0.694512 lr=9.960000e-06 lr_backbone=9.960000e-06 elapsed=00:21:23 avg_step=5.13s recent_step=4.13s eta=70:57:28 +2026-09-20 06:48:41 | INFO | train rank=0 | step=260/50000 progress=0.5% loss=0.667187 lr=1.036000e-05 lr_backbone=1.036000e-05 elapsed=00:22:04 avg_step=5.10s recent_step=4.13s eta=70:24:26 +2026-09-20 06:49:22 | INFO | train rank=0 | step=270/50000 progress=0.5% loss=0.645457 lr=1.076000e-05 lr_backbone=1.076000e-05 elapsed=00:22:46 avg_step=5.06s recent_step=4.12s eta=69:53:38 +2026-09-20 06:50:04 | INFO | train rank=0 | step=280/50000 progress=0.6% loss=0.673773 lr=1.116000e-05 lr_backbone=1.116000e-05 elapsed=00:23:27 avg_step=5.03s recent_step=4.13s eta=69:25:13 +2026-09-20 06:50:45 | INFO | train rank=0 | step=290/50000 progress=0.6% loss=0.611383 lr=1.156000e-05 lr_backbone=1.156000e-05 elapsed=00:24:08 avg_step=4.99s recent_step=4.11s eta=68:58:08 +2026-09-20 06:51:26 | INFO | train rank=0 | step=300/50000 progress=0.6% loss=0.617745 lr=1.196000e-05 lr_backbone=1.196000e-05 elapsed=00:24:49 avg_step=4.97s recent_step=4.10s eta=68:32:41 +2026-09-20 06:52:07 | INFO | train rank=0 | step=310/50000 progress=0.6% loss=0.645761 lr=1.236000e-05 lr_backbone=1.236000e-05 elapsed=00:25:30 avg_step=4.94s recent_step=4.14s eta=68:09:48 +2026-09-20 06:52:48 | INFO | train rank=0 | step=320/50000 progress=0.6% loss=0.521389 lr=1.276000e-05 lr_backbone=1.276000e-05 elapsed=00:26:12 avg_step=4.91s recent_step=4.14s eta=67:48:23 +2026-09-20 06:53:30 | INFO | train rank=0 | step=330/50000 progress=0.7% loss=0.596575 lr=1.316000e-05 lr_backbone=1.316000e-05 elapsed=00:26:53 avg_step=4.89s recent_step=4.14s eta=67:28:10 +2026-09-20 06:54:11 | INFO | train rank=0 | step=340/50000 progress=0.7% loss=0.571438 lr=1.356000e-05 lr_backbone=1.356000e-05 elapsed=00:27:35 avg_step=4.87s recent_step=4.14s eta=67:09:11 +2026-09-20 06:54:52 | INFO | train rank=0 | step=350/50000 progress=0.7% loss=0.549368 lr=1.396000e-05 lr_backbone=1.396000e-05 elapsed=00:28:16 avg_step=4.85s recent_step=4.10s eta=66:50:12 +2026-09-20 06:55:34 | INFO | train rank=0 | step=360/50000 progress=0.7% loss=0.556392 lr=1.436000e-05 lr_backbone=1.436000e-05 elapsed=00:28:57 avg_step=4.83s recent_step=4.15s eta=66:33:29 +2026-09-20 06:56:15 | INFO | train rank=0 | step=370/50000 progress=0.7% loss=0.537916 lr=1.476000e-05 lr_backbone=1.476000e-05 elapsed=00:29:38 avg_step=4.81s recent_step=4.11s eta=66:16:38 +2026-09-20 06:56:56 | INFO | train rank=0 | step=380/50000 progress=0.8% loss=0.558400 lr=1.516000e-05 lr_backbone=1.516000e-05 elapsed=00:30:20 avg_step=4.79s recent_step=4.14s eta=66:01:15 +2026-09-20 06:57:38 | INFO | train rank=0 | step=390/50000 progress=0.8% loss=0.598571 lr=1.556000e-05 lr_backbone=1.556000e-05 elapsed=00:31:01 avg_step=4.77s recent_step=4.15s eta=65:46:47 +2026-09-20 06:58:19 | INFO | train rank=0 | step=400/50000 progress=0.8% loss=0.490158 lr=1.596000e-05 lr_backbone=1.596000e-05 elapsed=00:31:43 avg_step=4.76s recent_step=4.16s eta=65:33:14 +2026-09-20 06:59:00 | INFO | train rank=0 | step=410/50000 progress=0.8% loss=0.543594 lr=1.636000e-05 lr_backbone=1.636000e-05 elapsed=00:32:24 avg_step=4.74s recent_step=4.11s eta=65:19:19 +2026-09-20 06:59:42 | INFO | train rank=0 | step=420/50000 progress=0.8% loss=0.508080 lr=1.676000e-05 lr_backbone=1.676000e-05 elapsed=00:33:05 avg_step=4.73s recent_step=4.13s eta=65:06:31 +2026-09-20 07:00:23 | INFO | train rank=0 | step=430/50000 progress=0.9% loss=0.578401 lr=1.716000e-05 lr_backbone=1.716000e-05 elapsed=00:33:46 avg_step=4.71s recent_step=4.11s eta=64:53:54 +2026-09-20 07:01:04 | INFO | train rank=0 | step=440/50000 progress=0.9% loss=0.515183 lr=1.756000e-05 lr_backbone=1.756000e-05 elapsed=00:34:28 avg_step=4.70s recent_step=4.13s eta=64:42:15 +2026-09-20 07:01:46 | INFO | train rank=0 | step=450/50000 progress=0.9% loss=0.483606 lr=1.796000e-05 lr_backbone=1.796000e-05 elapsed=00:35:09 avg_step=4.69s recent_step=4.13s eta=64:31:02 +2026-09-20 07:02:27 | INFO | train rank=0 | step=460/50000 progress=0.9% loss=0.513573 lr=1.836000e-05 lr_backbone=1.836000e-05 elapsed=00:35:50 avg_step=4.68s recent_step=4.13s eta=64:20:12 +2026-09-20 07:03:08 | INFO | train rank=0 | step=470/50000 progress=0.9% loss=0.527036 lr=1.876000e-05 lr_backbone=1.876000e-05 elapsed=00:36:31 avg_step=4.66s recent_step=4.09s eta=64:09:10 +2026-09-20 07:03:49 | INFO | train rank=0 | step=480/50000 progress=1.0% loss=0.550929 lr=1.916000e-05 lr_backbone=1.916000e-05 elapsed=00:37:12 avg_step=4.65s recent_step=4.14s eta=63:59:20 +2026-09-20 07:04:30 | INFO | train rank=0 | step=490/50000 progress=1.0% loss=0.477840 lr=1.956000e-05 lr_backbone=1.956000e-05 elapsed=00:37:54 avg_step=4.64s recent_step=4.11s eta=63:49:27 +2026-09-20 07:05:12 | INFO | train rank=0 | step=500/50000 progress=1.0% loss=0.529581 lr=1.996000e-05 lr_backbone=1.996000e-05 elapsed=00:38:35 avg_step=4.63s recent_step=4.15s eta=63:40:32 +2026-09-20 07:05:53 | INFO | train rank=0 | step=510/50000 progress=1.0% loss=0.469864 lr=2.036000e-05 lr_backbone=2.036000e-05 elapsed=00:39:16 avg_step=4.62s recent_step=4.13s eta=63:31:40 +2026-09-20 07:06:34 | INFO | train rank=0 | step=520/50000 progress=1.0% loss=0.456457 lr=2.076000e-05 lr_backbone=2.076000e-05 elapsed=00:39:58 avg_step=4.61s recent_step=4.15s eta=63:23:29 +2026-09-20 07:07:15 | INFO | train rank=0 | step=530/50000 progress=1.1% loss=0.452002 lr=2.116000e-05 lr_backbone=2.116000e-05 elapsed=00:40:39 avg_step=4.60s recent_step=4.10s eta=63:14:45 +2026-09-20 07:07:57 | INFO | train rank=0 | step=540/50000 progress=1.1% loss=0.414246 lr=2.156000e-05 lr_backbone=2.156000e-05 elapsed=00:41:20 avg_step=4.59s recent_step=4.15s eta=63:07:00 +2026-09-20 07:08:39 | INFO | train rank=0 | step=550/50000 progress=1.1% loss=0.435968 lr=2.196000e-05 lr_backbone=2.196000e-05 elapsed=00:42:02 avg_step=4.59s recent_step=4.17s eta=62:59:50 +2026-09-20 07:09:20 | INFO | train rank=0 | step=560/50000 progress=1.1% loss=0.437507 lr=2.236000e-05 lr_backbone=2.236000e-05 elapsed=00:42:43 avg_step=4.58s recent_step=4.13s eta=62:52:19 +2026-09-20 07:10:01 | INFO | train rank=0 | step=570/50000 progress=1.1% loss=0.403414 lr=2.276000e-05 lr_backbone=2.276000e-05 elapsed=00:43:25 avg_step=4.57s recent_step=4.14s eta=62:45:11 +2026-09-20 07:10:43 | INFO | train rank=0 | step=580/50000 progress=1.2% loss=0.494365 lr=2.316000e-05 lr_backbone=2.316000e-05 elapsed=00:44:06 avg_step=4.56s recent_step=4.16s eta=62:38:35 +2026-09-20 07:11:24 | INFO | train rank=0 | step=590/50000 progress=1.2% loss=0.462918 lr=2.356000e-05 lr_backbone=2.356000e-05 elapsed=00:44:47 avg_step=4.56s recent_step=4.08s eta=62:31:08 +2026-09-20 07:12:05 | INFO | train rank=0 | step=600/50000 progress=1.2% loss=0.462079 lr=2.396000e-05 lr_backbone=2.396000e-05 elapsed=00:45:28 avg_step=4.55s recent_step=4.15s eta=62:24:47 +2026-09-20 07:12:46 | INFO | train rank=0 | step=610/50000 progress=1.2% loss=0.433341 lr=2.436000e-05 lr_backbone=2.436000e-05 elapsed=00:46:10 avg_step=4.54s recent_step=4.13s eta=62:18:23 +2026-09-20 07:13:28 | INFO | train rank=0 | step=620/50000 progress=1.2% loss=0.488325 lr=2.476000e-05 lr_backbone=2.476000e-05 elapsed=00:46:51 avg_step=4.53s recent_step=4.11s eta=62:11:51 +2026-09-20 07:14:09 | INFO | train rank=0 | step=630/50000 progress=1.3% loss=0.461653 lr=2.516000e-05 lr_backbone=2.516000e-05 elapsed=00:47:32 avg_step=4.53s recent_step=4.14s eta=62:05:54 +2026-09-20 07:14:50 | INFO | train rank=0 | step=640/50000 progress=1.3% loss=0.477958 lr=2.556000e-05 lr_backbone=2.556000e-05 elapsed=00:48:14 avg_step=4.52s recent_step=4.14s eta=62:00:08 +2026-09-20 07:15:31 | INFO | train rank=0 | step=650/50000 progress=1.3% loss=0.386365 lr=2.596000e-05 lr_backbone=2.596000e-05 elapsed=00:48:55 avg_step=4.52s recent_step=4.11s eta=61:54:07 +2026-09-20 07:16:13 | INFO | train rank=0 | step=660/50000 progress=1.3% loss=0.420818 lr=2.636000e-05 lr_backbone=2.636000e-05 elapsed=00:49:36 avg_step=4.51s recent_step=4.13s eta=61:48:36 +2026-09-20 07:16:54 | INFO | train rank=0 | step=670/50000 progress=1.3% loss=0.488350 lr=2.676000e-05 lr_backbone=2.676000e-05 elapsed=00:50:17 avg_step=4.50s recent_step=4.11s eta=61:42:57 +2026-09-20 07:17:35 | INFO | train rank=0 | step=680/50000 progress=1.4% loss=0.499636 lr=2.716000e-05 lr_backbone=2.716000e-05 elapsed=00:50:58 avg_step=4.50s recent_step=4.12s eta=61:37:37 +2026-09-20 07:18:16 | INFO | train rank=0 | step=690/50000 progress=1.4% loss=0.464873 lr=2.756000e-05 lr_backbone=2.756000e-05 elapsed=00:51:40 avg_step=4.49s recent_step=4.14s eta=61:32:38 +2026-09-20 07:18:58 | INFO | train rank=0 | step=700/50000 progress=1.4% loss=0.477476 lr=2.796000e-05 lr_backbone=2.796000e-05 elapsed=00:52:21 avg_step=4.49s recent_step=4.12s eta=61:27:30 +2026-09-20 07:19:39 | INFO | train rank=0 | step=710/50000 progress=1.4% loss=0.459560 lr=2.836000e-05 lr_backbone=2.836000e-05 elapsed=00:53:02 avg_step=4.48s recent_step=4.11s eta=61:22:24 +2026-09-20 07:20:20 | INFO | train rank=0 | step=720/50000 progress=1.4% loss=0.471917 lr=2.876000e-05 lr_backbone=2.876000e-05 elapsed=00:53:44 avg_step=4.48s recent_step=4.15s eta=61:17:51 +2026-09-20 07:21:02 | INFO | train rank=0 | step=730/50000 progress=1.5% loss=0.482004 lr=2.916000e-05 lr_backbone=2.916000e-05 elapsed=00:54:25 avg_step=4.47s recent_step=4.12s eta=61:13:08 +2026-09-20 07:21:43 | INFO | train rank=0 | step=740/50000 progress=1.5% loss=0.453175 lr=2.956000e-05 lr_backbone=2.956000e-05 elapsed=00:55:06 avg_step=4.47s recent_step=4.16s eta=61:08:52 +2026-09-20 07:22:24 | INFO | train rank=0 | step=750/50000 progress=1.5% loss=0.456458 lr=2.996000e-05 lr_backbone=2.996000e-05 elapsed=00:55:47 avg_step=4.46s recent_step=4.11s eta=61:04:11 +2026-09-20 07:23:06 | INFO | train rank=0 | step=760/50000 progress=1.5% loss=0.389179 lr=3.036000e-05 lr_backbone=3.036000e-05 elapsed=00:56:29 avg_step=4.46s recent_step=4.15s eta=61:00:04 +2026-09-20 07:23:47 | INFO | train rank=0 | step=770/50000 progress=1.5% loss=0.391428 lr=3.076000e-05 lr_backbone=3.076000e-05 elapsed=00:57:10 avg_step=4.46s recent_step=4.09s eta=60:55:24 +2026-09-20 07:24:28 | INFO | train rank=0 | step=780/50000 progress=1.6% loss=0.402925 lr=3.116000e-05 lr_backbone=3.116000e-05 elapsed=00:57:51 avg_step=4.45s recent_step=4.13s eta=60:51:17 +2026-09-20 07:25:09 | INFO | train rank=0 | step=790/50000 progress=1.6% loss=0.426754 lr=3.156000e-05 lr_backbone=3.156000e-05 elapsed=00:58:33 avg_step=4.45s recent_step=4.13s eta=60:47:11 +2026-09-20 07:25:51 | INFO | train rank=0 | step=800/50000 progress=1.6% loss=0.423344 lr=3.196000e-05 lr_backbone=3.196000e-05 elapsed=00:59:14 avg_step=4.44s recent_step=4.16s eta=60:43:29 +2026-09-20 07:26:32 | INFO | train rank=0 | step=810/50000 progress=1.6% loss=0.469811 lr=3.236000e-05 lr_backbone=3.236000e-05 elapsed=00:59:55 avg_step=4.44s recent_step=4.10s eta=60:39:14 +2026-09-20 07:27:13 | INFO | train rank=0 | step=820/50000 progress=1.6% loss=0.478239 lr=3.276000e-05 lr_backbone=3.276000e-05 elapsed=01:00:36 avg_step=4.44s recent_step=4.12s eta=60:35:20 +2026-09-20 07:27:54 | INFO | train rank=0 | step=830/50000 progress=1.7% loss=0.443162 lr=3.316000e-05 lr_backbone=3.316000e-05 elapsed=01:01:17 avg_step=4.43s recent_step=4.11s eta=60:31:23 +2026-09-20 07:28:35 | INFO | train rank=0 | step=840/50000 progress=1.7% loss=0.458390 lr=3.356000e-05 lr_backbone=3.356000e-05 elapsed=01:01:59 avg_step=4.43s recent_step=4.13s eta=60:27:44 +2026-09-20 07:29:17 | INFO | train rank=0 | step=850/50000 progress=1.7% loss=0.437175 lr=3.396000e-05 lr_backbone=3.396000e-05 elapsed=01:02:40 avg_step=4.42s recent_step=4.16s eta=60:24:25 +2026-09-20 07:29:58 | INFO | train rank=0 | step=860/50000 progress=1.7% loss=0.426868 lr=3.436000e-05 lr_backbone=3.436000e-05 elapsed=01:03:22 avg_step=4.42s recent_step=4.14s eta=60:20:56 +2026-09-20 07:30:39 | INFO | train rank=0 | step=870/50000 progress=1.7% loss=0.460692 lr=3.476000e-05 lr_backbone=3.476000e-05 elapsed=01:04:03 avg_step=4.42s recent_step=4.11s eta=60:17:16 +2026-09-20 07:31:21 | INFO | train rank=0 | step=880/50000 progress=1.8% loss=0.423377 lr=3.516000e-05 lr_backbone=3.516000e-05 elapsed=01:04:44 avg_step=4.41s recent_step=4.14s eta=60:14:00 +2026-09-20 07:32:02 | INFO | train rank=0 | step=890/50000 progress=1.8% loss=0.439866 lr=3.556000e-05 lr_backbone=3.556000e-05 elapsed=01:05:25 avg_step=4.41s recent_step=4.10s eta=60:10:24 +2026-09-20 07:32:43 | INFO | train rank=0 | step=900/50000 progress=1.8% loss=0.419453 lr=3.596000e-05 lr_backbone=3.596000e-05 elapsed=01:06:07 avg_step=4.41s recent_step=4.14s eta=60:07:10 +2026-09-20 07:33:25 | INFO | train rank=0 | step=910/50000 progress=1.8% loss=0.375312 lr=3.636000e-05 lr_backbone=3.636000e-05 elapsed=01:06:48 avg_step=4.41s recent_step=4.15s eta=60:04:07 +2026-09-20 07:34:06 | INFO | train rank=0 | step=920/50000 progress=1.8% loss=0.460797 lr=3.676000e-05 lr_backbone=3.676000e-05 elapsed=01:07:30 avg_step=4.40s recent_step=4.14s eta=60:01:02 +2026-09-20 07:34:47 | INFO | train rank=0 | step=930/50000 progress=1.9% loss=0.432108 lr=3.716000e-05 lr_backbone=3.716000e-05 elapsed=01:08:11 avg_step=4.40s recent_step=4.10s eta=59:57:40 +2026-09-20 07:35:29 | INFO | train rank=0 | step=940/50000 progress=1.9% loss=0.534540 lr=3.756000e-05 lr_backbone=3.756000e-05 elapsed=01:08:52 avg_step=4.40s recent_step=4.13s eta=59:54:36 +2026-09-20 07:36:10 | INFO | train rank=0 | step=950/50000 progress=1.9% loss=0.455735 lr=3.796000e-05 lr_backbone=3.796000e-05 elapsed=01:09:33 avg_step=4.39s recent_step=4.10s eta=59:51:20 +2026-09-20 07:36:51 | INFO | train rank=0 | step=960/50000 progress=1.9% loss=0.431417 lr=3.836000e-05 lr_backbone=3.836000e-05 elapsed=01:10:14 avg_step=4.39s recent_step=4.16s eta=59:48:35 +2026-09-20 07:37:33 | INFO | train rank=0 | step=970/50000 progress=1.9% loss=0.448021 lr=3.876000e-05 lr_backbone=3.876000e-05 elapsed=01:10:56 avg_step=4.39s recent_step=4.15s eta=59:45:49 +2026-09-20 07:38:14 | INFO | train rank=0 | step=980/50000 progress=2.0% loss=0.373319 lr=3.916000e-05 lr_backbone=3.916000e-05 elapsed=01:11:37 avg_step=4.39s recent_step=4.11s eta=59:42:46 +2026-09-20 07:38:55 | INFO | train rank=0 | step=990/50000 progress=2.0% loss=0.462979 lr=3.956000e-05 lr_backbone=3.956000e-05 elapsed=01:12:18 avg_step=4.38s recent_step=4.14s eta=59:40:01 +2026-09-20 07:39:37 | INFO | train rank=0 | step=1000/50000 progress=2.0% loss=0.424200 lr=3.996000e-05 lr_backbone=3.996000e-05 elapsed=01:13:00 avg_step=4.38s recent_step=4.14s eta=59:37:20 +2026-09-20 07:40:18 | INFO | train rank=0 | step=1010/50000 progress=2.0% loss=0.421857 lr=4.036000e-05 lr_backbone=4.036000e-05 elapsed=01:13:41 avg_step=4.38s recent_step=4.13s eta=59:34:34 +2026-09-20 07:40:59 | INFO | train rank=0 | step=1020/50000 progress=2.0% loss=0.402908 lr=4.076000e-05 lr_backbone=4.076000e-05 elapsed=01:14:23 avg_step=4.38s recent_step=4.15s eta=59:32:03 +2026-09-20 07:41:41 | INFO | train rank=0 | step=1030/50000 progress=2.1% loss=0.478976 lr=4.116000e-05 lr_backbone=4.116000e-05 elapsed=01:15:04 avg_step=4.37s recent_step=4.13s eta=59:29:23 +2026-09-20 07:42:22 | INFO | train rank=0 | step=1040/50000 progress=2.1% loss=0.512137 lr=4.156000e-05 lr_backbone=4.156000e-05 elapsed=01:15:45 avg_step=4.37s recent_step=4.13s eta=59:26:45 +2026-09-20 07:43:03 | INFO | train rank=0 | step=1050/50000 progress=2.1% loss=0.395459 lr=4.196000e-05 lr_backbone=4.196000e-05 elapsed=01:16:26 avg_step=4.37s recent_step=4.11s eta=59:23:58 +2026-09-20 07:43:45 | INFO | train rank=0 | step=1060/50000 progress=2.1% loss=0.432065 lr=4.236000e-05 lr_backbone=4.236000e-05 elapsed=01:17:08 avg_step=4.37s recent_step=4.15s eta=59:21:31 +2026-09-20 07:44:25 | INFO | train rank=0 | step=1070/50000 progress=2.1% loss=0.415040 lr=4.276000e-05 lr_backbone=4.276000e-05 elapsed=01:17:49 avg_step=4.36s recent_step=4.09s eta=59:18:41 +2026-09-20 07:45:07 | INFO | train rank=0 | step=1080/50000 progress=2.2% loss=0.370111 lr=4.316000e-05 lr_backbone=4.316000e-05 elapsed=01:18:30 avg_step=4.36s recent_step=4.16s eta=59:16:26 +2026-09-20 07:45:48 | INFO | train rank=0 | step=1090/50000 progress=2.2% loss=0.459373 lr=4.356000e-05 lr_backbone=4.356000e-05 elapsed=01:19:12 avg_step=4.36s recent_step=4.13s eta=59:13:56 +2026-09-20 07:46:30 | INFO | train rank=0 | step=1100/50000 progress=2.2% loss=0.507528 lr=4.396000e-05 lr_backbone=4.396000e-05 elapsed=01:19:53 avg_step=4.36s recent_step=4.12s eta=59:11:27 +2026-09-20 07:47:11 | INFO | train rank=0 | step=1110/50000 progress=2.2% loss=0.374656 lr=4.436000e-05 lr_backbone=4.436000e-05 elapsed=01:20:34 avg_step=4.36s recent_step=4.16s eta=59:09:14 +2026-09-20 07:47:52 | INFO | train rank=0 | step=1120/50000 progress=2.2% loss=0.538260 lr=4.476000e-05 lr_backbone=4.476000e-05 elapsed=01:21:16 avg_step=4.35s recent_step=4.13s eta=59:06:50 +2026-09-20 07:48:33 | INFO | train rank=0 | step=1130/50000 progress=2.3% loss=0.423759 lr=4.516000e-05 lr_backbone=4.516000e-05 elapsed=01:21:57 avg_step=4.35s recent_step=4.09s eta=59:04:14 +2026-09-20 07:49:15 | INFO | train rank=0 | step=1140/50000 progress=2.3% loss=0.415554 lr=4.556000e-05 lr_backbone=4.556000e-05 elapsed=01:22:38 avg_step=4.35s recent_step=4.13s eta=59:01:56 +2026-09-20 07:49:56 | INFO | train rank=0 | step=1150/50000 progress=2.3% loss=0.418605 lr=4.596000e-05 lr_backbone=4.596000e-05 elapsed=01:23:19 avg_step=4.35s recent_step=4.13s eta=58:59:40 +2026-09-20 07:50:37 | INFO | train rank=0 | step=1160/50000 progress=2.3% loss=0.423656 lr=4.636000e-05 lr_backbone=4.636000e-05 elapsed=01:24:01 avg_step=4.35s recent_step=4.14s eta=58:57:29 +2026-09-20 07:51:19 | INFO | train rank=0 | step=1170/50000 progress=2.3% loss=0.380731 lr=4.676000e-05 lr_backbone=4.676000e-05 elapsed=01:24:42 avg_step=4.34s recent_step=4.14s eta=58:55:20 +2026-09-20 07:52:00 | INFO | train rank=0 | step=1180/50000 progress=2.4% loss=0.435996 lr=4.716000e-05 lr_backbone=4.716000e-05 elapsed=01:25:23 avg_step=4.34s recent_step=4.13s eta=58:53:08 +2026-09-20 07:52:41 | INFO | train rank=0 | step=1190/50000 progress=2.4% loss=0.415131 lr=4.756000e-05 lr_backbone=4.756000e-05 elapsed=01:26:04 avg_step=4.34s recent_step=4.11s eta=58:50:51 +2026-09-20 07:53:23 | INFO | train rank=0 | step=1200/50000 progress=2.4% loss=0.412513 lr=4.796000e-05 lr_backbone=4.796000e-05 elapsed=01:26:46 avg_step=4.34s recent_step=4.15s eta=58:48:49 +2026-09-20 07:54:04 | INFO | train rank=0 | step=1210/50000 progress=2.4% loss=0.445469 lr=4.836000e-05 lr_backbone=4.836000e-05 elapsed=01:27:27 avg_step=4.34s recent_step=4.11s eta=58:46:34 +2026-09-20 07:54:45 | INFO | train rank=0 | step=1220/50000 progress=2.4% loss=0.448053 lr=4.876000e-05 lr_backbone=4.876000e-05 elapsed=01:28:08 avg_step=4.34s recent_step=4.12s eta=58:44:26 +2026-09-20 07:55:26 | INFO | train rank=0 | step=1230/50000 progress=2.5% loss=0.374403 lr=4.916000e-05 lr_backbone=4.916000e-05 elapsed=01:28:50 avg_step=4.33s recent_step=4.13s eta=58:42:22 +2026-09-20 07:56:07 | INFO | train rank=0 | step=1240/50000 progress=2.5% loss=0.402032 lr=4.956000e-05 lr_backbone=4.956000e-05 elapsed=01:29:31 avg_step=4.33s recent_step=4.12s eta=58:40:14 +2026-09-20 07:56:48 | INFO | train rank=0 | step=1250/50000 progress=2.5% loss=0.461245 lr=4.996000e-05 lr_backbone=4.996000e-05 elapsed=01:30:12 avg_step=4.33s recent_step=4.07s eta=58:37:50 +2026-09-20 07:57:29 | INFO | train rank=0 | step=1260/50000 progress=2.5% loss=0.509142 lr=5.036000e-05 lr_backbone=5.036000e-05 elapsed=01:30:53 avg_step=4.33s recent_step=4.12s eta=58:35:45 +2026-09-20 07:58:11 | INFO | train rank=0 | step=1270/50000 progress=2.5% loss=0.390707 lr=5.076000e-05 lr_backbone=5.076000e-05 elapsed=01:31:34 avg_step=4.33s recent_step=4.12s eta=58:33:41 +2026-09-20 07:58:52 | INFO | train rank=0 | step=1280/50000 progress=2.6% loss=0.445906 lr=5.116000e-05 lr_backbone=5.116000e-05 elapsed=01:32:15 avg_step=4.32s recent_step=4.12s eta=58:31:37 +2026-09-20 07:59:33 | INFO | train rank=0 | step=1290/50000 progress=2.6% loss=0.387460 lr=5.156000e-05 lr_backbone=5.156000e-05 elapsed=01:32:56 avg_step=4.32s recent_step=4.13s eta=58:29:39 +2026-09-20 08:00:14 | INFO | train rank=0 | step=1300/50000 progress=2.6% loss=0.439394 lr=5.196000e-05 lr_backbone=5.196000e-05 elapsed=01:33:38 avg_step=4.32s recent_step=4.15s eta=58:27:51 +2026-09-20 08:00:55 | INFO | train rank=0 | step=1310/50000 progress=2.6% loss=0.451240 lr=5.236000e-05 lr_backbone=5.236000e-05 elapsed=01:34:19 avg_step=4.32s recent_step=4.07s eta=58:25:34 +2026-09-20 08:01:36 | INFO | train rank=0 | step=1320/50000 progress=2.6% loss=0.447730 lr=5.276000e-05 lr_backbone=5.276000e-05 elapsed=01:35:00 avg_step=4.32s recent_step=4.13s eta=58:23:39 +2026-09-20 08:02:18 | INFO | train rank=0 | step=1330/50000 progress=2.7% loss=0.417382 lr=5.316000e-05 lr_backbone=5.316000e-05 elapsed=01:35:41 avg_step=4.32s recent_step=4.12s eta=58:21:45 +2026-09-20 08:02:59 | INFO | train rank=0 | step=1340/50000 progress=2.7% loss=0.515148 lr=5.356000e-05 lr_backbone=5.356000e-05 elapsed=01:36:23 avg_step=4.32s recent_step=4.15s eta=58:20:03 +2026-09-20 08:03:40 | INFO | train rank=0 | step=1350/50000 progress=2.7% loss=0.424658 lr=5.396000e-05 lr_backbone=5.396000e-05 elapsed=01:37:04 avg_step=4.31s recent_step=4.13s eta=58:18:11 +2026-09-20 08:04:22 | INFO | train rank=0 | step=1360/50000 progress=2.7% loss=0.442690 lr=5.436000e-05 lr_backbone=5.436000e-05 elapsed=01:37:45 avg_step=4.31s recent_step=4.12s eta=58:16:18 +2026-09-20 08:05:03 | INFO | train rank=0 | step=1370/50000 progress=2.7% loss=0.499698 lr=5.476000e-05 lr_backbone=5.476000e-05 elapsed=01:38:26 avg_step=4.31s recent_step=4.11s eta=58:14:22 +2026-09-20 08:05:44 | INFO | train rank=0 | step=1380/50000 progress=2.8% loss=0.440497 lr=5.516000e-05 lr_backbone=5.516000e-05 elapsed=01:39:07 avg_step=4.31s recent_step=4.11s eta=58:12:27 +2026-09-20 08:06:25 | INFO | train rank=0 | step=1390/50000 progress=2.8% loss=0.471275 lr=5.556000e-05 lr_backbone=5.556000e-05 elapsed=01:39:49 avg_step=4.31s recent_step=4.13s eta=58:10:43 +2026-09-20 08:07:07 | INFO | train rank=0 | step=1400/50000 progress=2.8% loss=0.413566 lr=5.596000e-05 lr_backbone=5.596000e-05 elapsed=01:40:30 avg_step=4.31s recent_step=4.14s eta=58:09:01 +2026-09-20 08:07:48 | INFO | train rank=0 | step=1410/50000 progress=2.8% loss=0.401797 lr=5.636000e-05 lr_backbone=5.636000e-05 elapsed=01:41:11 avg_step=4.31s recent_step=4.14s eta=58:07:19 +2026-09-20 08:08:29 | INFO | train rank=0 | step=1420/50000 progress=2.8% loss=0.407247 lr=5.676000e-05 lr_backbone=5.676000e-05 elapsed=01:41:52 avg_step=4.30s recent_step=4.12s eta=58:05:33 +2026-09-20 08:09:10 | INFO | train rank=0 | step=1430/50000 progress=2.9% loss=0.430753 lr=5.716000e-05 lr_backbone=5.716000e-05 elapsed=01:42:33 avg_step=4.30s recent_step=4.10s eta=58:03:40 +2026-09-20 08:09:51 | INFO | train rank=0 | step=1440/50000 progress=2.9% loss=0.441160 lr=5.756000e-05 lr_backbone=5.756000e-05 elapsed=01:43:14 avg_step=4.30s recent_step=4.10s eta=58:01:48 +2026-09-20 08:10:33 | INFO | train rank=0 | step=1450/50000 progress=2.9% loss=0.401137 lr=5.796000e-05 lr_backbone=5.796000e-05 elapsed=01:43:56 avg_step=4.30s recent_step=4.16s eta=58:00:17 +2026-09-20 08:11:14 | INFO | train rank=0 | step=1460/50000 progress=2.9% loss=0.383210 lr=5.836000e-05 lr_backbone=5.836000e-05 elapsed=01:44:37 avg_step=4.30s recent_step=4.14s eta=57:58:41 +2026-09-20 08:11:55 | INFO | train rank=0 | step=1470/50000 progress=2.9% loss=0.385337 lr=5.876000e-05 lr_backbone=5.876000e-05 elapsed=01:45:19 avg_step=4.30s recent_step=4.12s eta=57:56:59 +2026-09-20 08:12:37 | INFO | train rank=0 | step=1480/50000 progress=3.0% loss=0.439420 lr=5.916000e-05 lr_backbone=5.916000e-05 elapsed=01:46:00 avg_step=4.30s recent_step=4.13s eta=57:55:20 +2026-09-20 08:13:18 | INFO | train rank=0 | step=1490/50000 progress=3.0% loss=0.412028 lr=5.956000e-05 lr_backbone=5.956000e-05 elapsed=01:46:41 avg_step=4.30s recent_step=4.09s eta=57:53:29 +2026-09-20 08:13:59 | INFO | train rank=0 | step=1500/50000 progress=3.0% loss=0.395324 lr=5.996000e-05 lr_backbone=5.996000e-05 elapsed=01:47:22 avg_step=4.30s recent_step=4.12s eta=57:51:50 +2026-09-20 08:14:40 | INFO | train rank=0 | step=1510/50000 progress=3.0% loss=0.413132 lr=6.036000e-05 lr_backbone=6.036000e-05 elapsed=01:48:03 avg_step=4.29s recent_step=4.12s eta=57:50:10 +2026-09-20 08:15:21 | INFO | train rank=0 | step=1520/50000 progress=3.0% loss=0.422739 lr=6.076000e-05 lr_backbone=6.076000e-05 elapsed=01:48:45 avg_step=4.29s recent_step=4.13s eta=57:48:33 +2026-09-20 08:16:02 | INFO | train rank=0 | step=1530/50000 progress=3.1% loss=0.465532 lr=6.116000e-05 lr_backbone=6.116000e-05 elapsed=01:49:26 avg_step=4.29s recent_step=4.12s eta=57:46:56 +2026-09-20 08:16:44 | INFO | train rank=0 | step=1540/50000 progress=3.1% loss=0.398235 lr=6.156000e-05 lr_backbone=6.156000e-05 elapsed=01:50:07 avg_step=4.29s recent_step=4.14s eta=57:45:24 +2026-09-20 08:17:25 | INFO | train rank=0 | step=1550/50000 progress=3.1% loss=0.404742 lr=6.196000e-05 lr_backbone=6.196000e-05 elapsed=01:50:48 avg_step=4.29s recent_step=4.09s eta=57:43:39 +2026-09-20 08:18:06 | INFO | train rank=0 | step=1560/50000 progress=3.1% loss=0.407457 lr=6.236000e-05 lr_backbone=6.236000e-05 elapsed=01:51:29 avg_step=4.29s recent_step=4.14s eta=57:42:09 +2026-09-20 08:18:47 | INFO | train rank=0 | step=1570/50000 progress=3.1% loss=0.420281 lr=6.276000e-05 lr_backbone=6.276000e-05 elapsed=01:52:11 avg_step=4.29s recent_step=4.13s eta=57:40:38 +2026-09-20 08:19:29 | INFO | train rank=0 | step=1580/50000 progress=3.2% loss=0.491909 lr=6.316000e-05 lr_backbone=6.316000e-05 elapsed=01:52:52 avg_step=4.29s recent_step=4.14s eta=57:39:11 +2026-09-20 08:20:10 | INFO | train rank=0 | step=1590/50000 progress=3.2% loss=0.507534 lr=6.356000e-05 lr_backbone=6.356000e-05 elapsed=01:53:33 avg_step=4.29s recent_step=4.12s eta=57:37:37 +2026-09-20 08:20:51 | INFO | train rank=0 | step=1600/50000 progress=3.2% loss=0.444845 lr=6.396000e-05 lr_backbone=6.396000e-05 elapsed=01:54:15 avg_step=4.28s recent_step=4.14s eta=57:36:12 +2026-09-20 08:21:33 | INFO | train rank=0 | step=1610/50000 progress=3.2% loss=0.445177 lr=6.436000e-05 lr_backbone=6.436000e-05 elapsed=01:54:56 avg_step=4.28s recent_step=4.12s eta=57:34:39 +2026-09-20 08:22:14 | INFO | train rank=0 | step=1620/50000 progress=3.2% loss=0.448998 lr=6.476000e-05 lr_backbone=6.476000e-05 elapsed=01:55:37 avg_step=4.28s recent_step=4.15s eta=57:33:17 +2026-09-20 08:22:56 | INFO | train rank=0 | step=1630/50000 progress=3.3% loss=0.407628 lr=6.516000e-05 lr_backbone=6.516000e-05 elapsed=01:56:19 avg_step=4.28s recent_step=4.16s eta=57:31:57 +2026-09-20 08:23:37 | INFO | train rank=0 | step=1640/50000 progress=3.3% loss=0.432723 lr=6.556000e-05 lr_backbone=6.556000e-05 elapsed=01:57:00 avg_step=4.28s recent_step=4.10s eta=57:30:20 +2026-09-20 08:24:18 | INFO | train rank=0 | step=1650/50000 progress=3.3% loss=0.435830 lr=6.596000e-05 lr_backbone=6.596000e-05 elapsed=01:57:41 avg_step=4.28s recent_step=4.13s eta=57:28:54 +2026-09-20 08:24:59 | INFO | train rank=0 | step=1660/50000 progress=3.3% loss=0.523102 lr=6.636000e-05 lr_backbone=6.636000e-05 elapsed=01:58:23 avg_step=4.28s recent_step=4.13s eta=57:27:26 +2026-09-20 08:25:40 | INFO | train rank=0 | step=1670/50000 progress=3.3% loss=0.384098 lr=6.676000e-05 lr_backbone=6.676000e-05 elapsed=01:59:04 avg_step=4.28s recent_step=4.09s eta=57:25:48 +2026-09-20 08:26:22 | INFO | train rank=0 | step=1680/50000 progress=3.4% loss=0.381851 lr=6.716000e-05 lr_backbone=6.716000e-05 elapsed=01:59:45 avg_step=4.28s recent_step=4.14s eta=57:24:27 +2026-09-20 08:27:03 | INFO | train rank=0 | step=1690/50000 progress=3.4% loss=0.419297 lr=6.756000e-05 lr_backbone=6.756000e-05 elapsed=02:00:27 avg_step=4.28s recent_step=4.16s eta=57:23:11 +2026-09-20 08:27:44 | INFO | train rank=0 | step=1700/50000 progress=3.4% loss=0.391601 lr=6.796000e-05 lr_backbone=6.796000e-05 elapsed=02:01:08 avg_step=4.28s recent_step=4.11s eta=57:21:41 +2026-09-20 08:28:26 | INFO | train rank=0 | step=1710/50000 progress=3.4% loss=0.360910 lr=6.836000e-05 lr_backbone=6.836000e-05 elapsed=02:01:49 avg_step=4.27s recent_step=4.13s eta=57:20:16 +2026-09-20 08:29:07 | INFO | train rank=0 | step=1720/50000 progress=3.4% loss=0.454583 lr=6.876000e-05 lr_backbone=6.876000e-05 elapsed=02:02:30 avg_step=4.27s recent_step=4.12s eta=57:18:50 +2026-09-20 08:29:48 | INFO | train rank=0 | step=1730/50000 progress=3.5% loss=0.405739 lr=6.916000e-05 lr_backbone=6.916000e-05 elapsed=02:03:11 avg_step=4.27s recent_step=4.09s eta=57:17:16 +2026-09-20 08:30:29 | INFO | train rank=0 | step=1740/50000 progress=3.5% loss=0.443737 lr=6.956000e-05 lr_backbone=6.956000e-05 elapsed=02:03:53 avg_step=4.27s recent_step=4.15s eta=57:15:59 +2026-09-20 08:31:11 | INFO | train rank=0 | step=1750/50000 progress=3.5% loss=0.377757 lr=6.996000e-05 lr_backbone=6.996000e-05 elapsed=02:04:34 avg_step=4.27s recent_step=4.15s eta=57:14:43 +2026-09-20 08:31:52 | INFO | train rank=0 | step=1760/50000 progress=3.5% loss=0.516514 lr=7.036000e-05 lr_backbone=7.036000e-05 elapsed=02:05:15 avg_step=4.27s recent_step=4.14s eta=57:13:23 +2026-09-20 08:32:33 | INFO | train rank=0 | step=1770/50000 progress=3.5% loss=0.483181 lr=7.076000e-05 lr_backbone=7.076000e-05 elapsed=02:05:57 avg_step=4.27s recent_step=4.13s eta=57:12:02 +2026-09-20 08:33:15 | INFO | train rank=0 | step=1780/50000 progress=3.6% loss=0.417994 lr=7.116000e-05 lr_backbone=7.116000e-05 elapsed=02:06:38 avg_step=4.27s recent_step=4.15s eta=57:10:46 +2026-09-20 08:33:56 | INFO | train rank=0 | step=1790/50000 progress=3.6% loss=0.410699 lr=7.156000e-05 lr_backbone=7.156000e-05 elapsed=02:07:19 avg_step=4.27s recent_step=4.09s eta=57:09:16 +2026-09-20 08:34:37 | INFO | train rank=0 | step=1800/50000 progress=3.6% loss=0.485467 lr=7.196000e-05 lr_backbone=7.196000e-05 elapsed=02:08:00 avg_step=4.27s recent_step=4.12s eta=57:07:53 +2026-09-20 08:35:19 | INFO | train rank=0 | step=1810/50000 progress=3.6% loss=0.467955 lr=7.236000e-05 lr_backbone=7.236000e-05 elapsed=02:08:42 avg_step=4.27s recent_step=4.16s eta=57:06:42 +2026-09-20 08:36:00 | INFO | train rank=0 | step=1820/50000 progress=3.6% loss=0.492683 lr=7.276000e-05 lr_backbone=7.276000e-05 elapsed=02:09:23 avg_step=4.27s recent_step=4.12s eta=57:05:20 +2026-09-20 08:36:41 | INFO | train rank=0 | step=1830/50000 progress=3.7% loss=0.398125 lr=7.316000e-05 lr_backbone=7.316000e-05 elapsed=02:10:04 avg_step=4.26s recent_step=4.10s eta=57:03:54 +2026-09-20 08:37:22 | INFO | train rank=0 | step=1840/50000 progress=3.7% loss=0.415893 lr=7.356000e-05 lr_backbone=7.356000e-05 elapsed=02:10:45 avg_step=4.26s recent_step=4.15s eta=57:02:40 +2026-09-20 08:38:03 | INFO | train rank=0 | step=1850/50000 progress=3.7% loss=0.422894 lr=7.396000e-05 lr_backbone=7.396000e-05 elapsed=02:11:26 avg_step=4.26s recent_step=4.09s eta=57:01:11 +2026-09-20 08:38:44 | INFO | train rank=0 | step=1860/50000 progress=3.7% loss=0.430681 lr=7.436000e-05 lr_backbone=7.436000e-05 elapsed=02:12:07 avg_step=4.26s recent_step=4.11s eta=56:59:50 +2026-09-20 08:39:25 | INFO | train rank=0 | step=1870/50000 progress=3.7% loss=0.404461 lr=7.476000e-05 lr_backbone=7.476000e-05 elapsed=02:12:49 avg_step=4.26s recent_step=4.13s eta=56:58:33 +2026-09-20 08:40:07 | INFO | train rank=0 | step=1880/50000 progress=3.8% loss=0.426501 lr=7.516000e-05 lr_backbone=7.516000e-05 elapsed=02:13:30 avg_step=4.26s recent_step=4.13s eta=56:57:16 +2026-09-20 08:40:48 | INFO | train rank=0 | step=1890/50000 progress=3.8% loss=0.426993 lr=7.556000e-05 lr_backbone=7.556000e-05 elapsed=02:14:11 avg_step=4.26s recent_step=4.13s eta=56:56:01 +2026-09-20 08:41:29 | INFO | train rank=0 | step=1900/50000 progress=3.8% loss=0.485717 lr=7.596000e-05 lr_backbone=7.596000e-05 elapsed=02:14:52 avg_step=4.26s recent_step=4.10s eta=56:54:38 +2026-09-20 08:42:10 | INFO | train rank=0 | step=1910/50000 progress=3.8% loss=0.434088 lr=7.636000e-05 lr_backbone=7.636000e-05 elapsed=02:15:33 avg_step=4.26s recent_step=4.11s eta=56:53:17 +2026-09-20 08:42:51 | INFO | train rank=0 | step=1920/50000 progress=3.8% loss=0.444011 lr=7.676000e-05 lr_backbone=7.676000e-05 elapsed=02:16:15 avg_step=4.26s recent_step=4.14s eta=56:52:04 +2026-09-20 08:43:33 | INFO | train rank=0 | step=1930/50000 progress=3.9% loss=0.433621 lr=7.716000e-05 lr_backbone=7.716000e-05 elapsed=02:16:56 avg_step=4.26s recent_step=4.14s eta=56:50:51 +2026-09-20 08:44:14 | INFO | train rank=0 | step=1940/50000 progress=3.9% loss=0.396573 lr=7.756000e-05 lr_backbone=7.756000e-05 elapsed=02:17:38 avg_step=4.26s recent_step=4.14s eta=56:49:38 +2026-09-20 08:44:56 | INFO | train rank=0 | step=1950/50000 progress=3.9% loss=0.462558 lr=7.796000e-05 lr_backbone=7.796000e-05 elapsed=02:18:19 avg_step=4.26s recent_step=4.13s eta=56:48:25 +2026-09-20 08:45:37 | INFO | train rank=0 | step=1960/50000 progress=3.9% loss=0.393580 lr=7.836000e-05 lr_backbone=7.836000e-05 elapsed=02:19:00 avg_step=4.26s recent_step=4.14s eta=56:47:14 +2026-09-20 08:46:18 | INFO | train rank=0 | step=1970/50000 progress=3.9% loss=0.596106 lr=7.876000e-05 lr_backbone=7.876000e-05 elapsed=02:19:41 avg_step=4.25s recent_step=4.09s eta=56:45:50 +2026-09-20 08:46:59 | INFO | train rank=0 | step=1980/50000 progress=4.0% loss=0.437651 lr=7.916000e-05 lr_backbone=7.916000e-05 elapsed=02:20:22 avg_step=4.25s recent_step=4.12s eta=56:44:34 +2026-09-20 08:47:40 | INFO | train rank=0 | step=1990/50000 progress=4.0% loss=0.410248 lr=7.956000e-05 lr_backbone=7.956000e-05 elapsed=02:21:04 avg_step=4.25s recent_step=4.13s eta=56:43:21 +2026-09-20 08:48:22 | INFO | train rank=0 | step=2000/50000 progress=4.0% loss=0.355929 lr=7.996000e-05 lr_backbone=7.996000e-05 elapsed=02:21:45 avg_step=4.25s recent_step=4.14s eta=56:42:11 +2026-09-20 08:49:03 | INFO | train rank=0 | step=2010/50000 progress=4.0% loss=0.423237 lr=8.036000e-05 lr_backbone=8.036000e-05 elapsed=02:22:26 avg_step=4.25s recent_step=4.14s eta=56:41:02 +2026-09-20 08:49:44 | INFO | train rank=0 | step=2020/50000 progress=4.0% loss=0.433277 lr=8.076000e-05 lr_backbone=8.076000e-05 elapsed=02:23:08 avg_step=4.25s recent_step=4.12s eta=56:39:48 +2026-09-20 08:50:25 | INFO | train rank=0 | step=2030/50000 progress=4.1% loss=0.446377 lr=8.116000e-05 lr_backbone=8.116000e-05 elapsed=02:23:48 avg_step=4.25s recent_step=4.09s eta=56:38:27 +2026-09-20 08:51:06 | INFO | train rank=0 | step=2040/50000 progress=4.1% loss=0.413359 lr=8.156000e-05 lr_backbone=8.156000e-05 elapsed=02:24:30 avg_step=4.25s recent_step=4.13s eta=56:37:16 +2026-09-20 08:51:48 | INFO | train rank=0 | step=2050/50000 progress=4.1% loss=0.463046 lr=8.196000e-05 lr_backbone=8.196000e-05 elapsed=02:25:11 avg_step=4.25s recent_step=4.12s eta=56:36:03 +2026-09-20 08:52:29 | INFO | train rank=0 | step=2060/50000 progress=4.1% loss=0.414989 lr=8.236000e-05 lr_backbone=8.236000e-05 elapsed=02:25:52 avg_step=4.25s recent_step=4.13s eta=56:34:52 +2026-09-20 08:53:10 | INFO | train rank=0 | step=2070/50000 progress=4.1% loss=0.389767 lr=8.276000e-05 lr_backbone=8.276000e-05 elapsed=02:26:33 avg_step=4.25s recent_step=4.11s eta=56:33:38 +2026-09-20 08:53:51 | INFO | train rank=0 | step=2080/50000 progress=4.2% loss=0.481108 lr=8.316000e-05 lr_backbone=8.316000e-05 elapsed=02:27:15 avg_step=4.25s recent_step=4.13s eta=56:32:28 +2026-09-20 08:54:32 | INFO | train rank=0 | step=2090/50000 progress=4.2% loss=0.404274 lr=8.356000e-05 lr_backbone=8.356000e-05 elapsed=02:27:56 avg_step=4.25s recent_step=4.10s eta=56:31:12 +2026-09-20 08:55:14 | INFO | train rank=0 | step=2100/50000 progress=4.2% loss=0.449158 lr=8.396000e-05 lr_backbone=8.396000e-05 elapsed=02:28:37 avg_step=4.25s recent_step=4.15s eta=56:30:06 +2026-09-20 08:55:55 | INFO | train rank=0 | step=2110/50000 progress=4.2% loss=0.434127 lr=8.436000e-05 lr_backbone=8.436000e-05 elapsed=02:29:18 avg_step=4.25s recent_step=4.13s eta=56:28:57 +2026-09-20 08:56:36 | INFO | train rank=0 | step=2120/50000 progress=4.2% loss=0.465189 lr=8.476000e-05 lr_backbone=8.476000e-05 elapsed=02:30:00 avg_step=4.25s recent_step=4.12s eta=56:27:46 +2026-09-20 08:57:18 | INFO | train rank=0 | step=2130/50000 progress=4.3% loss=0.465366 lr=8.516000e-05 lr_backbone=8.516000e-05 elapsed=02:30:41 avg_step=4.24s recent_step=4.13s eta=56:26:37 +2026-09-20 08:57:59 | INFO | train rank=0 | step=2140/50000 progress=4.3% loss=0.436929 lr=8.556000e-05 lr_backbone=8.556000e-05 elapsed=02:31:22 avg_step=4.24s recent_step=4.11s eta=56:25:24 +2026-09-20 08:58:40 | INFO | train rank=0 | step=2150/50000 progress=4.3% loss=0.459865 lr=8.596000e-05 lr_backbone=8.596000e-05 elapsed=02:32:03 avg_step=4.24s recent_step=4.11s eta=56:24:11 +2026-09-20 08:59:21 | INFO | train rank=0 | step=2160/50000 progress=4.3% loss=0.469018 lr=8.636000e-05 lr_backbone=8.636000e-05 elapsed=02:32:44 avg_step=4.24s recent_step=4.13s eta=56:23:03 +2026-09-20 09:00:02 | INFO | train rank=0 | step=2170/50000 progress=4.3% loss=0.427893 lr=8.676000e-05 lr_backbone=8.676000e-05 elapsed=02:33:26 avg_step=4.24s recent_step=4.12s eta=56:21:54 +2026-09-20 09:00:44 | INFO | train rank=0 | step=2180/50000 progress=4.4% loss=0.522387 lr=8.716000e-05 lr_backbone=8.716000e-05 elapsed=02:34:07 avg_step=4.24s recent_step=4.13s eta=56:20:48 +2026-09-20 09:01:25 | INFO | train rank=0 | step=2190/50000 progress=4.4% loss=0.395981 lr=8.756000e-05 lr_backbone=8.756000e-05 elapsed=02:34:48 avg_step=4.24s recent_step=4.13s eta=56:19:42 +2026-09-20 09:02:06 | INFO | train rank=0 | step=2200/50000 progress=4.4% loss=0.386795 lr=8.796000e-05 lr_backbone=8.796000e-05 elapsed=02:35:29 avg_step=4.24s recent_step=4.10s eta=56:18:28 +2026-09-20 09:02:47 | INFO | train rank=0 | step=2210/50000 progress=4.4% loss=0.439178 lr=8.836000e-05 lr_backbone=8.836000e-05 elapsed=02:36:10 avg_step=4.24s recent_step=4.10s eta=56:17:16 +2026-09-20 09:03:28 | INFO | train rank=0 | step=2220/50000 progress=4.4% loss=0.483105 lr=8.876000e-05 lr_backbone=8.876000e-05 elapsed=02:36:51 avg_step=4.24s recent_step=4.12s eta=56:16:07 +2026-09-20 09:04:09 | INFO | train rank=0 | step=2230/50000 progress=4.5% loss=0.408720 lr=8.916000e-05 lr_backbone=8.916000e-05 elapsed=02:37:33 avg_step=4.24s recent_step=4.13s eta=56:15:02 +2026-09-20 09:04:51 | INFO | train rank=0 | step=2240/50000 progress=4.5% loss=0.396049 lr=8.956000e-05 lr_backbone=8.956000e-05 elapsed=02:38:14 avg_step=4.24s recent_step=4.14s eta=56:13:59 +2026-09-20 09:05:32 | INFO | train rank=0 | step=2250/50000 progress=4.5% loss=0.423230 lr=8.996000e-05 lr_backbone=8.996000e-05 elapsed=02:38:55 avg_step=4.24s recent_step=4.11s eta=56:12:50 +2026-09-20 09:06:13 | INFO | train rank=0 | step=2260/50000 progress=4.5% loss=0.381196 lr=9.036000e-05 lr_backbone=9.036000e-05 elapsed=02:39:37 avg_step=4.24s recent_step=4.13s eta=56:11:46 +2026-09-20 09:06:54 | INFO | train rank=0 | step=2270/50000 progress=4.5% loss=0.390001 lr=9.076000e-05 lr_backbone=9.076000e-05 elapsed=02:40:18 avg_step=4.24s recent_step=4.10s eta=56:10:35 +2026-09-20 09:07:36 | INFO | train rank=0 | step=2280/50000 progress=4.6% loss=0.388464 lr=9.116000e-05 lr_backbone=9.116000e-05 elapsed=02:40:59 avg_step=4.24s recent_step=4.13s eta=56:09:29 +2026-09-20 09:08:17 | INFO | train rank=0 | step=2290/50000 progress=4.6% loss=0.377645 lr=9.156000e-05 lr_backbone=9.156000e-05 elapsed=02:41:40 avg_step=4.24s recent_step=4.14s eta=56:08:27 +2026-09-20 09:08:58 | INFO | train rank=0 | step=2300/50000 progress=4.6% loss=0.415423 lr=9.196000e-05 lr_backbone=9.196000e-05 elapsed=02:42:21 avg_step=4.24s recent_step=4.12s eta=56:07:20 +2026-09-20 09:09:39 | INFO | train rank=0 | step=2310/50000 progress=4.6% loss=0.458442 lr=9.236000e-05 lr_backbone=9.236000e-05 elapsed=02:43:03 avg_step=4.24s recent_step=4.13s eta=56:06:16 +2026-09-20 09:10:20 | INFO | train rank=0 | step=2320/50000 progress=4.6% loss=0.366163 lr=9.276000e-05 lr_backbone=9.276000e-05 elapsed=02:43:44 avg_step=4.23s recent_step=4.09s eta=56:05:05 +2026-09-20 09:11:01 | INFO | train rank=0 | step=2330/50000 progress=4.7% loss=0.456349 lr=9.316000e-05 lr_backbone=9.316000e-05 elapsed=02:44:25 avg_step=4.23s recent_step=4.08s eta=56:03:51 +2026-09-20 09:11:42 | INFO | train rank=0 | step=2340/50000 progress=4.7% loss=0.449025 lr=9.356000e-05 lr_backbone=9.356000e-05 elapsed=02:45:06 avg_step=4.23s recent_step=4.11s eta=56:02:43 +2026-09-20 09:12:24 | INFO | train rank=0 | step=2350/50000 progress=4.7% loss=0.447608 lr=9.396000e-05 lr_backbone=9.396000e-05 elapsed=02:45:47 avg_step=4.23s recent_step=4.12s eta=56:01:38 +2026-09-20 09:13:05 | INFO | train rank=0 | step=2360/50000 progress=4.7% loss=0.497834 lr=9.436000e-05 lr_backbone=9.436000e-05 elapsed=02:46:28 avg_step=4.23s recent_step=4.12s eta=56:00:32 +2026-09-20 09:13:46 | INFO | train rank=0 | step=2370/50000 progress=4.7% loss=0.432554 lr=9.476000e-05 lr_backbone=9.476000e-05 elapsed=02:47:09 avg_step=4.23s recent_step=4.10s eta=55:59:23 +2026-09-20 09:14:27 | INFO | train rank=0 | step=2380/50000 progress=4.8% loss=0.479320 lr=9.516000e-05 lr_backbone=9.516000e-05 elapsed=02:47:50 avg_step=4.23s recent_step=4.11s eta=55:58:16 +2026-09-20 09:15:08 | INFO | train rank=0 | step=2390/50000 progress=4.8% loss=0.386504 lr=9.556000e-05 lr_backbone=9.556000e-05 elapsed=02:48:31 avg_step=4.23s recent_step=4.08s eta=55:57:03 +2026-09-20 09:15:49 | INFO | train rank=0 | step=2400/50000 progress=4.8% loss=0.423083 lr=9.596000e-05 lr_backbone=9.596000e-05 elapsed=02:49:12 avg_step=4.23s recent_step=4.11s eta=55:55:56 +2026-09-20 09:16:30 | INFO | train rank=0 | step=2410/50000 progress=4.8% loss=0.481200 lr=9.636000e-05 lr_backbone=9.636000e-05 elapsed=02:49:53 avg_step=4.23s recent_step=4.12s eta=55:54:52 +2026-09-20 09:17:11 | INFO | train rank=0 | step=2420/50000 progress=4.8% loss=0.374400 lr=9.676000e-05 lr_backbone=9.676000e-05 elapsed=02:50:34 avg_step=4.23s recent_step=4.11s eta=55:53:47 +2026-09-20 09:17:52 | INFO | train rank=0 | step=2430/50000 progress=4.9% loss=0.443066 lr=9.716000e-05 lr_backbone=9.716000e-05 elapsed=02:51:15 avg_step=4.23s recent_step=4.10s eta=55:52:40 +2026-09-20 09:18:33 | INFO | train rank=0 | step=2440/50000 progress=4.9% loss=0.415536 lr=9.756000e-05 lr_backbone=9.756000e-05 elapsed=02:51:56 avg_step=4.23s recent_step=4.12s eta=55:51:36 +2026-09-20 09:19:14 | INFO | train rank=0 | step=2450/50000 progress=4.9% loss=0.402112 lr=9.796000e-05 lr_backbone=9.796000e-05 elapsed=02:52:37 avg_step=4.23s recent_step=4.08s eta=55:50:24 +2026-09-20 09:19:55 | INFO | train rank=0 | step=2460/50000 progress=4.9% loss=0.405987 lr=9.836000e-05 lr_backbone=9.836000e-05 elapsed=02:53:19 avg_step=4.23s recent_step=4.14s eta=55:49:25 +2026-09-20 09:20:37 | INFO | train rank=0 | step=2470/50000 progress=4.9% loss=0.475700 lr=9.876000e-05 lr_backbone=9.876000e-05 elapsed=02:54:00 avg_step=4.23s recent_step=4.14s eta=55:48:26 +2026-09-20 09:21:18 | INFO | train rank=0 | step=2480/50000 progress=5.0% loss=0.438949 lr=9.916000e-05 lr_backbone=9.916000e-05 elapsed=02:54:41 avg_step=4.23s recent_step=4.12s eta=55:47:23 +2026-09-20 09:21:59 | INFO | train rank=0 | step=2490/50000 progress=5.0% loss=0.440796 lr=9.956000e-05 lr_backbone=9.956000e-05 elapsed=02:55:23 avg_step=4.23s recent_step=4.13s eta=55:46:23 +2026-09-20 09:22:41 | INFO | train rank=0 | step=2500/50000 progress=5.0% loss=0.380231 lr=9.996000e-05 lr_backbone=9.996000e-05 elapsed=02:56:04 avg_step=4.23s recent_step=4.14s eta=55:45:24 +2026-09-20 09:23:22 | INFO | train rank=0 | step=2510/50000 progress=5.0% loss=0.494097 lr=9.999999e-05 lr_backbone=9.999999e-05 elapsed=02:56:45 avg_step=4.23s recent_step=4.09s eta=55:44:17 +2026-09-20 09:24:03 | INFO | train rank=0 | step=2520/50000 progress=5.0% loss=0.441931 lr=9.999996e-05 lr_backbone=9.999996e-05 elapsed=02:57:26 avg_step=4.22s recent_step=4.11s eta=55:43:13 +2026-09-20 09:24:44 | INFO | train rank=0 | step=2530/50000 progress=5.1% loss=0.424304 lr=9.999990e-05 lr_backbone=9.999990e-05 elapsed=02:58:07 avg_step=4.22s recent_step=4.14s eta=55:42:15 +2026-09-20 09:25:26 | INFO | train rank=0 | step=2540/50000 progress=5.1% loss=0.443587 lr=9.999983e-05 lr_backbone=9.999983e-05 elapsed=02:58:49 avg_step=4.22s recent_step=4.15s eta=55:41:19 +2026-09-20 09:26:07 | INFO | train rank=0 | step=2550/50000 progress=5.1% loss=0.417189 lr=9.999973e-05 lr_backbone=9.999973e-05 elapsed=02:59:30 avg_step=4.22s recent_step=4.12s eta=55:40:16 +2026-09-20 09:26:48 | INFO | train rank=0 | step=2560/50000 progress=5.1% loss=0.376738 lr=9.999961e-05 lr_backbone=9.999961e-05 elapsed=03:00:11 avg_step=4.22s recent_step=4.12s eta=55:39:16 +2026-09-20 09:27:29 | INFO | train rank=0 | step=2570/50000 progress=5.1% loss=0.424361 lr=9.999946e-05 lr_backbone=9.999946e-05 elapsed=03:00:52 avg_step=4.22s recent_step=4.10s eta=55:38:10 +2026-09-20 09:28:10 | INFO | train rank=0 | step=2580/50000 progress=5.2% loss=0.457970 lr=9.999930e-05 lr_backbone=9.999930e-05 elapsed=03:01:33 avg_step=4.22s recent_step=4.11s eta=55:37:07 +2026-09-20 09:28:52 | INFO | train rank=0 | step=2590/50000 progress=5.2% loss=0.462565 lr=9.999911e-05 lr_backbone=9.999911e-05 elapsed=03:02:15 avg_step=4.22s recent_step=4.15s eta=55:36:11 +2026-09-20 09:29:33 | INFO | train rank=0 | step=2600/50000 progress=5.2% loss=0.367938 lr=9.999891e-05 lr_backbone=9.999891e-05 elapsed=03:02:56 avg_step=4.22s recent_step=4.14s eta=55:35:15 +2026-09-20 09:30:14 | INFO | train rank=0 | step=2610/50000 progress=5.2% loss=0.434513 lr=9.999868e-05 lr_backbone=9.999868e-05 elapsed=03:03:38 avg_step=4.22s recent_step=4.13s eta=55:34:15 +2026-09-20 09:30:56 | INFO | train rank=0 | step=2620/50000 progress=5.2% loss=0.421257 lr=9.999843e-05 lr_backbone=9.999843e-05 elapsed=03:04:19 avg_step=4.22s recent_step=4.16s eta=55:33:21 +2026-09-20 09:31:37 | INFO | train rank=0 | step=2630/50000 progress=5.3% loss=0.384586 lr=9.999815e-05 lr_backbone=9.999815e-05 elapsed=03:05:00 avg_step=4.22s recent_step=4.10s eta=55:32:17 +2026-09-20 09:32:18 | INFO | train rank=0 | step=2640/50000 progress=5.3% loss=0.413107 lr=9.999786e-05 lr_backbone=9.999786e-05 elapsed=03:05:41 avg_step=4.22s recent_step=4.13s eta=55:31:18 +2026-09-20 09:32:59 | INFO | train rank=0 | step=2650/50000 progress=5.3% loss=0.414780 lr=9.999754e-05 lr_backbone=9.999754e-05 elapsed=03:06:23 avg_step=4.22s recent_step=4.11s eta=55:30:17 +2026-09-20 09:33:41 | INFO | train rank=0 | step=2660/50000 progress=5.3% loss=0.380409 lr=9.999720e-05 lr_backbone=9.999720e-05 elapsed=03:07:04 avg_step=4.22s recent_step=4.15s eta=55:29:22 +2026-09-20 09:34:22 | INFO | train rank=0 | step=2670/50000 progress=5.3% loss=0.432054 lr=9.999684e-05 lr_backbone=9.999684e-05 elapsed=03:07:45 avg_step=4.22s recent_step=4.12s eta=55:28:23 +2026-09-20 09:35:03 | INFO | train rank=0 | step=2680/50000 progress=5.4% loss=0.389927 lr=9.999646e-05 lr_backbone=9.999646e-05 elapsed=03:08:26 avg_step=4.22s recent_step=4.12s eta=55:27:24 +2026-09-20 09:35:44 | INFO | train rank=0 | step=2690/50000 progress=5.4% loss=0.437641 lr=9.999605e-05 lr_backbone=9.999605e-05 elapsed=03:09:07 avg_step=4.22s recent_step=4.10s eta=55:26:21 +2026-09-20 09:36:26 | INFO | train rank=0 | step=2700/50000 progress=5.4% loss=0.405676 lr=9.999563e-05 lr_backbone=9.999563e-05 elapsed=03:09:49 avg_step=4.22s recent_step=4.14s eta=55:25:24 +2026-09-20 09:37:07 | INFO | train rank=0 | step=2710/50000 progress=5.4% loss=0.455253 lr=9.999518e-05 lr_backbone=9.999518e-05 elapsed=03:10:30 avg_step=4.22s recent_step=4.11s eta=55:24:23 +2026-09-20 09:37:48 | INFO | train rank=0 | step=2720/50000 progress=5.4% loss=0.473551 lr=9.999471e-05 lr_backbone=9.999471e-05 elapsed=03:11:11 avg_step=4.22s recent_step=4.14s eta=55:23:27 +2026-09-20 09:38:29 | INFO | train rank=0 | step=2730/50000 progress=5.5% loss=0.416580 lr=9.999422e-05 lr_backbone=9.999422e-05 elapsed=03:11:53 avg_step=4.22s recent_step=4.13s eta=55:22:30 +2026-09-20 09:39:11 | INFO | train rank=0 | step=2740/50000 progress=5.5% loss=0.511091 lr=9.999370e-05 lr_backbone=9.999370e-05 elapsed=03:12:34 avg_step=4.22s recent_step=4.13s eta=55:21:32 +2026-09-20 09:39:51 | INFO | train rank=0 | step=2750/50000 progress=5.5% loss=0.435355 lr=9.999317e-05 lr_backbone=9.999317e-05 elapsed=03:13:15 avg_step=4.22s recent_step=4.09s eta=55:20:28 +2026-09-20 09:40:33 | INFO | train rank=0 | step=2760/50000 progress=5.5% loss=0.452459 lr=9.999261e-05 lr_backbone=9.999261e-05 elapsed=03:13:56 avg_step=4.22s recent_step=4.14s eta=55:19:33 +2026-09-20 09:41:14 | INFO | train rank=0 | step=2770/50000 progress=5.5% loss=0.512813 lr=9.999203e-05 lr_backbone=9.999203e-05 elapsed=03:14:38 avg_step=4.22s recent_step=4.13s eta=55:18:36 +2026-09-20 09:41:55 | INFO | train rank=0 | step=2780/50000 progress=5.6% loss=0.396029 lr=9.999143e-05 lr_backbone=9.999143e-05 elapsed=03:15:19 avg_step=4.22s recent_step=4.11s eta=55:17:35 +2026-09-20 09:42:37 | INFO | train rank=0 | step=2790/50000 progress=5.6% loss=0.465126 lr=9.999080e-05 lr_backbone=9.999080e-05 elapsed=03:16:00 avg_step=4.22s recent_step=4.15s eta=55:16:41 +2026-09-20 09:43:18 | INFO | train rank=0 | step=2800/50000 progress=5.6% loss=0.444215 lr=9.999016e-05 lr_backbone=9.999016e-05 elapsed=03:16:41 avg_step=4.21s recent_step=4.13s eta=55:15:45 +2026-09-20 09:43:59 | INFO | train rank=0 | step=2810/50000 progress=5.6% loss=0.455183 lr=9.998949e-05 lr_backbone=9.998949e-05 elapsed=03:17:22 avg_step=4.21s recent_step=4.09s eta=55:14:42 +2026-09-20 09:44:40 | INFO | train rank=0 | step=2820/50000 progress=5.6% loss=0.491904 lr=9.998880e-05 lr_backbone=9.998880e-05 elapsed=03:18:04 avg_step=4.21s recent_step=4.13s eta=55:13:46 +2026-09-20 09:45:22 | INFO | train rank=0 | step=2830/50000 progress=5.7% loss=0.410127 lr=9.998809e-05 lr_backbone=9.998809e-05 elapsed=03:18:45 avg_step=4.21s recent_step=4.14s eta=55:12:51 +2026-09-20 09:46:03 | INFO | train rank=0 | step=2840/50000 progress=5.7% loss=0.408979 lr=9.998736e-05 lr_backbone=9.998736e-05 elapsed=03:19:26 avg_step=4.21s recent_step=4.10s eta=55:11:50 +2026-09-20 09:46:44 | INFO | train rank=0 | step=2850/50000 progress=5.7% loss=0.414879 lr=9.998660e-05 lr_backbone=9.998660e-05 elapsed=03:20:07 avg_step=4.21s recent_step=4.12s eta=55:10:53 +2026-09-20 09:47:25 | INFO | train rank=0 | step=2860/50000 progress=5.7% loss=0.459304 lr=9.998583e-05 lr_backbone=9.998583e-05 elapsed=03:20:49 avg_step=4.21s recent_step=4.15s eta=55:10:00 +2026-09-20 09:48:06 | INFO | train rank=0 | step=2870/50000 progress=5.7% loss=0.421857 lr=9.998503e-05 lr_backbone=9.998503e-05 elapsed=03:21:30 avg_step=4.21s recent_step=4.09s eta=55:08:57 +2026-09-20 09:48:48 | INFO | train rank=0 | step=2880/50000 progress=5.8% loss=0.454419 lr=9.998421e-05 lr_backbone=9.998421e-05 elapsed=03:22:11 avg_step=4.21s recent_step=4.14s eta=55:08:03 +2026-09-20 09:49:29 | INFO | train rank=0 | step=2890/50000 progress=5.8% loss=0.469844 lr=9.998337e-05 lr_backbone=9.998337e-05 elapsed=03:22:52 avg_step=4.21s recent_step=4.13s eta=55:07:07 +2026-09-20 09:50:10 | INFO | train rank=0 | step=2900/50000 progress=5.8% loss=0.471470 lr=9.998250e-05 lr_backbone=9.998250e-05 elapsed=03:23:34 avg_step=4.21s recent_step=4.14s eta=55:06:13 +2026-09-20 09:50:51 | INFO | train rank=0 | step=2910/50000 progress=5.8% loss=0.382336 lr=9.998162e-05 lr_backbone=9.998162e-05 elapsed=03:24:15 avg_step=4.21s recent_step=4.11s eta=55:05:13 +2026-09-20 09:51:33 | INFO | train rank=0 | step=2920/50000 progress=5.8% loss=0.422466 lr=9.998071e-05 lr_backbone=9.998071e-05 elapsed=03:24:56 avg_step=4.21s recent_step=4.13s eta=55:04:18 +2026-09-20 09:52:14 | INFO | train rank=0 | step=2930/50000 progress=5.9% loss=0.474261 lr=9.997978e-05 lr_backbone=9.997978e-05 elapsed=03:25:37 avg_step=4.21s recent_step=4.10s eta=55:03:18 +2026-09-20 09:52:55 | INFO | train rank=0 | step=2940/50000 progress=5.9% loss=0.522612 lr=9.997883e-05 lr_backbone=9.997883e-05 elapsed=03:26:18 avg_step=4.21s recent_step=4.13s eta=55:02:24 +2026-09-20 09:53:36 | INFO | train rank=0 | step=2950/50000 progress=5.9% loss=0.438900 lr=9.997786e-05 lr_backbone=9.997786e-05 elapsed=03:27:00 avg_step=4.21s recent_step=4.13s eta=55:01:29 +2026-09-20 09:54:18 | INFO | train rank=0 | step=2960/50000 progress=5.9% loss=0.440493 lr=9.997686e-05 lr_backbone=9.997686e-05 elapsed=03:27:41 avg_step=4.21s recent_step=4.13s eta=55:00:34 +2026-09-20 09:54:59 | INFO | train rank=0 | step=2970/50000 progress=5.9% loss=0.450147 lr=9.997584e-05 lr_backbone=9.997584e-05 elapsed=03:28:22 avg_step=4.21s recent_step=4.13s eta=54:59:40 +2026-09-20 09:55:40 | INFO | train rank=0 | step=2980/50000 progress=6.0% loss=0.567843 lr=9.997481e-05 lr_backbone=9.997481e-05 elapsed=03:29:04 avg_step=4.21s recent_step=4.16s eta=54:58:50 +2026-09-20 09:56:22 | INFO | train rank=0 | step=2990/50000 progress=6.0% loss=0.395827 lr=9.997375e-05 lr_backbone=9.997375e-05 elapsed=03:29:45 avg_step=4.21s recent_step=4.11s eta=54:57:53 +2026-09-20 09:57:03 | INFO | train rank=0 | step=3000/50000 progress=6.0% loss=0.431454 lr=9.997266e-05 lr_backbone=9.997266e-05 elapsed=03:30:26 avg_step=4.21s recent_step=4.14s eta=54:57:00 +2026-09-20 09:57:44 | INFO | train rank=0 | step=3010/50000 progress=6.0% loss=0.408544 lr=9.997156e-05 lr_backbone=9.997156e-05 elapsed=03:31:08 avg_step=4.21s recent_step=4.15s eta=54:56:08 +2026-09-20 09:58:26 | INFO | train rank=0 | step=3020/50000 progress=6.0% loss=0.403391 lr=9.997043e-05 lr_backbone=9.997043e-05 elapsed=03:31:49 avg_step=4.21s recent_step=4.15s eta=54:55:17 +2026-09-20 09:59:07 | INFO | train rank=0 | step=3030/50000 progress=6.1% loss=0.469510 lr=9.996928e-05 lr_backbone=9.996928e-05 elapsed=03:32:31 avg_step=4.21s recent_step=4.12s eta=54:54:22 +2026-09-20 09:59:48 | INFO | train rank=0 | step=3040/50000 progress=6.1% loss=0.527039 lr=9.996811e-05 lr_backbone=9.996811e-05 elapsed=03:33:12 avg_step=4.21s recent_step=4.10s eta=54:53:24 +2026-09-20 10:00:29 | INFO | train rank=0 | step=3050/50000 progress=6.1% loss=0.397785 lr=9.996692e-05 lr_backbone=9.996692e-05 elapsed=03:33:53 avg_step=4.21s recent_step=4.10s eta=54:52:25 +2026-09-20 10:01:11 | INFO | train rank=0 | step=3060/50000 progress=6.1% loss=0.410853 lr=9.996571e-05 lr_backbone=9.996571e-05 elapsed=03:34:34 avg_step=4.21s recent_step=4.15s eta=54:51:34 +2026-09-20 10:01:52 | INFO | train rank=0 | step=3070/50000 progress=6.1% loss=0.378128 lr=9.996447e-05 lr_backbone=9.996447e-05 elapsed=03:35:15 avg_step=4.21s recent_step=4.13s eta=54:50:40 +2026-09-20 10:02:33 | INFO | train rank=0 | step=3080/50000 progress=6.2% loss=0.391641 lr=9.996322e-05 lr_backbone=9.996322e-05 elapsed=03:35:57 avg_step=4.21s recent_step=4.14s eta=54:49:47 +2026-09-20 10:03:15 | INFO | train rank=0 | step=3090/50000 progress=6.2% loss=0.395289 lr=9.996194e-05 lr_backbone=9.996194e-05 elapsed=03:36:38 avg_step=4.21s recent_step=4.14s eta=54:48:54 +2026-09-20 10:03:56 | INFO | train rank=0 | step=3100/50000 progress=6.2% loss=0.437585 lr=9.996064e-05 lr_backbone=9.996064e-05 elapsed=03:37:19 avg_step=4.21s recent_step=4.10s eta=54:47:57 +2026-09-20 10:04:37 | INFO | train rank=0 | step=3110/50000 progress=6.2% loss=0.401749 lr=9.995931e-05 lr_backbone=9.995931e-05 elapsed=03:38:00 avg_step=4.21s recent_step=4.10s eta=54:46:58 +2026-09-20 10:05:18 | INFO | train rank=0 | step=3120/50000 progress=6.2% loss=0.421362 lr=9.995797e-05 lr_backbone=9.995797e-05 elapsed=03:38:41 avg_step=4.21s recent_step=4.13s eta=54:46:04 +2026-09-20 10:05:59 | INFO | train rank=0 | step=3130/50000 progress=6.3% loss=0.435137 lr=9.995660e-05 lr_backbone=9.995660e-05 elapsed=03:39:23 avg_step=4.21s recent_step=4.12s eta=54:45:09 +2026-09-20 10:06:41 | INFO | train rank=0 | step=3140/50000 progress=6.3% loss=0.389738 lr=9.995521e-05 lr_backbone=9.995521e-05 elapsed=03:40:04 avg_step=4.21s recent_step=4.14s eta=54:44:17 +2026-09-20 10:07:22 | INFO | train rank=0 | step=3150/50000 progress=6.3% loss=0.381768 lr=9.995380e-05 lr_backbone=9.995380e-05 elapsed=03:40:45 avg_step=4.21s recent_step=4.14s eta=54:43:25 +2026-09-20 10:08:03 | INFO | train rank=0 | step=3160/50000 progress=6.3% loss=0.417129 lr=9.995237e-05 lr_backbone=9.995237e-05 elapsed=03:41:26 avg_step=4.20s recent_step=4.10s eta=54:42:27 +2026-09-20 10:08:44 | INFO | train rank=0 | step=3170/50000 progress=6.3% loss=0.451785 lr=9.995092e-05 lr_backbone=9.995092e-05 elapsed=03:42:07 avg_step=4.20s recent_step=4.08s eta=54:41:26 +2026-09-20 10:09:25 | INFO | train rank=0 | step=3180/50000 progress=6.4% loss=0.416049 lr=9.994944e-05 lr_backbone=9.994944e-05 elapsed=03:42:48 avg_step=4.20s recent_step=4.13s eta=54:40:33 +2026-09-20 10:10:06 | INFO | train rank=0 | step=3190/50000 progress=6.4% loss=0.449126 lr=9.994794e-05 lr_backbone=9.994794e-05 elapsed=03:43:29 avg_step=4.20s recent_step=4.11s eta=54:39:37 +2026-09-20 10:10:47 | INFO | train rank=0 | step=3200/50000 progress=6.4% loss=0.444091 lr=9.994642e-05 lr_backbone=9.994642e-05 elapsed=03:44:11 avg_step=4.20s recent_step=4.13s eta=54:38:44 +2026-09-20 10:11:28 | INFO | train rank=0 | step=3210/50000 progress=6.4% loss=0.470897 lr=9.994488e-05 lr_backbone=9.994488e-05 elapsed=03:44:52 avg_step=4.20s recent_step=4.09s eta=54:37:45 +2026-09-20 10:12:09 | INFO | train rank=0 | step=3220/50000 progress=6.4% loss=0.417239 lr=9.994332e-05 lr_backbone=9.994332e-05 elapsed=03:45:33 avg_step=4.20s recent_step=4.10s eta=54:36:48 +2026-09-20 10:12:50 | INFO | train rank=0 | step=3230/50000 progress=6.5% loss=0.454185 lr=9.994173e-05 lr_backbone=9.994173e-05 elapsed=03:46:13 avg_step=4.20s recent_step=4.07s eta=54:35:47 +2026-09-20 10:13:31 | INFO | train rank=0 | step=3240/50000 progress=6.5% loss=0.454925 lr=9.994013e-05 lr_backbone=9.994013e-05 elapsed=03:46:55 avg_step=4.20s recent_step=4.12s eta=54:34:53 +2026-09-20 10:14:13 | INFO | train rank=0 | step=3250/50000 progress=6.5% loss=0.502820 lr=9.993850e-05 lr_backbone=9.993850e-05 elapsed=03:47:36 avg_step=4.20s recent_step=4.14s eta=54:34:02 +2026-09-20 10:14:54 | INFO | train rank=0 | step=3260/50000 progress=6.5% loss=0.420907 lr=9.993685e-05 lr_backbone=9.993685e-05 elapsed=03:48:17 avg_step=4.20s recent_step=4.10s eta=54:33:05 +2026-09-20 10:15:35 | INFO | train rank=0 | step=3270/50000 progress=6.5% loss=0.394297 lr=9.993518e-05 lr_backbone=9.993518e-05 elapsed=03:48:58 avg_step=4.20s recent_step=4.13s eta=54:32:12 +2026-09-20 10:16:16 | INFO | train rank=0 | step=3280/50000 progress=6.6% loss=0.459613 lr=9.993348e-05 lr_backbone=9.993348e-05 elapsed=03:49:40 avg_step=4.20s recent_step=4.14s eta=54:31:21 +2026-09-20 10:16:57 | INFO | train rank=0 | step=3290/50000 progress=6.6% loss=0.405763 lr=9.993176e-05 lr_backbone=9.993176e-05 elapsed=03:50:21 avg_step=4.20s recent_step=4.10s eta=54:30:24 +2026-09-20 10:17:38 | INFO | train rank=0 | step=3300/50000 progress=6.6% loss=0.417575 lr=9.993003e-05 lr_backbone=9.993003e-05 elapsed=03:51:02 avg_step=4.20s recent_step=4.11s eta=54:29:29 +2026-09-20 10:18:19 | INFO | train rank=0 | step=3310/50000 progress=6.6% loss=0.370617 lr=9.992827e-05 lr_backbone=9.992827e-05 elapsed=03:51:43 avg_step=4.20s recent_step=4.12s eta=54:28:36 +2026-09-20 10:19:01 | INFO | train rank=0 | step=3320/50000 progress=6.6% loss=0.420768 lr=9.992649e-05 lr_backbone=9.992649e-05 elapsed=03:52:24 avg_step=4.20s recent_step=4.13s eta=54:27:43 +2026-09-20 10:19:42 | INFO | train rank=0 | step=3330/50000 progress=6.7% loss=0.435777 lr=9.992468e-05 lr_backbone=9.992468e-05 elapsed=03:53:05 avg_step=4.20s recent_step=4.14s eta=54:26:53 +2026-09-20 10:20:24 | INFO | train rank=0 | step=3340/50000 progress=6.7% loss=0.394882 lr=9.992286e-05 lr_backbone=9.992286e-05 elapsed=03:53:47 avg_step=4.20s recent_step=4.15s eta=54:26:04 +2026-09-20 10:21:04 | INFO | train rank=0 | step=3350/50000 progress=6.7% loss=0.439141 lr=9.992101e-05 lr_backbone=9.992101e-05 elapsed=03:54:28 avg_step=4.20s recent_step=4.08s eta=54:25:06 +2026-09-20 10:21:46 | INFO | train rank=0 | step=3360/50000 progress=6.7% loss=0.411622 lr=9.991914e-05 lr_backbone=9.991914e-05 elapsed=03:55:09 avg_step=4.20s recent_step=4.11s eta=54:24:11 +2026-09-20 10:22:27 | INFO | train rank=0 | step=3370/50000 progress=6.7% loss=0.388178 lr=9.991725e-05 lr_backbone=9.991725e-05 elapsed=03:55:50 avg_step=4.20s recent_step=4.14s eta=54:23:22 +2026-09-20 10:23:08 | INFO | train rank=0 | step=3380/50000 progress=6.8% loss=0.461730 lr=9.991534e-05 lr_backbone=9.991534e-05 elapsed=03:56:32 avg_step=4.20s recent_step=4.14s eta=54:22:31 +2026-09-20 10:23:49 | INFO | train rank=0 | step=3390/50000 progress=6.8% loss=0.415235 lr=9.991340e-05 lr_backbone=9.991340e-05 elapsed=03:57:13 avg_step=4.20s recent_step=4.11s eta=54:21:37 +2026-09-20 10:24:31 | INFO | train rank=0 | step=3400/50000 progress=6.8% loss=0.368589 lr=9.991145e-05 lr_backbone=9.991145e-05 elapsed=03:57:54 avg_step=4.20s recent_step=4.13s eta=54:20:46 +2026-09-20 10:25:12 | INFO | train rank=0 | step=3410/50000 progress=6.8% loss=0.392901 lr=9.990947e-05 lr_backbone=9.990947e-05 elapsed=03:58:35 avg_step=4.20s recent_step=4.10s eta=54:19:50 +2026-09-20 10:25:53 | INFO | train rank=0 | step=3420/50000 progress=6.8% loss=0.399037 lr=9.990747e-05 lr_backbone=9.990747e-05 elapsed=03:59:17 avg_step=4.20s recent_step=4.15s eta=54:19:01 +2026-09-20 10:26:34 | INFO | train rank=0 | step=3430/50000 progress=6.9% loss=0.375268 lr=9.990545e-05 lr_backbone=9.990545e-05 elapsed=03:59:58 avg_step=4.20s recent_step=4.10s eta=54:18:06 +2026-09-20 10:27:16 | INFO | train rank=0 | step=3440/50000 progress=6.9% loss=0.373695 lr=9.990340e-05 lr_backbone=9.990340e-05 elapsed=04:00:39 avg_step=4.20s recent_step=4.13s eta=54:17:15 +2026-09-20 10:27:57 | INFO | train rank=0 | step=3450/50000 progress=6.9% loss=0.447194 lr=9.990134e-05 lr_backbone=9.990134e-05 elapsed=04:01:20 avg_step=4.20s recent_step=4.13s eta=54:16:24 +2026-09-20 10:28:38 | INFO | train rank=0 | step=3460/50000 progress=6.9% loss=0.378397 lr=9.989925e-05 lr_backbone=9.989925e-05 elapsed=04:02:02 avg_step=4.20s recent_step=4.14s eta=54:15:34 +2026-09-20 10:29:19 | INFO | train rank=0 | step=3470/50000 progress=6.9% loss=0.484150 lr=9.989714e-05 lr_backbone=9.989714e-05 elapsed=04:02:43 avg_step=4.20s recent_step=4.10s eta=54:14:39 +2026-09-20 10:30:01 | INFO | train rank=0 | step=3480/50000 progress=7.0% loss=0.414713 lr=9.989501e-05 lr_backbone=9.989501e-05 elapsed=04:03:24 avg_step=4.20s recent_step=4.14s eta=54:13:49 +2026-09-20 10:30:42 | INFO | train rank=0 | step=3490/50000 progress=7.0% loss=0.400479 lr=9.989286e-05 lr_backbone=9.989286e-05 elapsed=04:04:05 avg_step=4.20s recent_step=4.11s eta=54:12:55 +2026-09-20 10:31:23 | INFO | train rank=0 | step=3500/50000 progress=7.0% loss=0.456415 lr=9.989068e-05 lr_backbone=9.989068e-05 elapsed=04:04:46 avg_step=4.20s recent_step=4.12s eta=54:12:03 +2026-09-20 10:32:04 | INFO | train rank=0 | step=3510/50000 progress=7.0% loss=0.412476 lr=9.988848e-05 lr_backbone=9.988848e-05 elapsed=04:05:28 avg_step=4.20s recent_step=4.15s eta=54:11:15 +2026-09-20 10:32:46 | INFO | train rank=0 | step=3520/50000 progress=7.0% loss=0.488189 lr=9.988627e-05 lr_backbone=9.988627e-05 elapsed=04:06:09 avg_step=4.20s recent_step=4.14s eta=54:10:25 +2026-09-20 10:33:27 | INFO | train rank=0 | step=3530/50000 progress=7.1% loss=0.419092 lr=9.988403e-05 lr_backbone=9.988403e-05 elapsed=04:06:50 avg_step=4.20s recent_step=4.10s eta=54:09:30 +2026-09-20 10:34:08 | INFO | train rank=0 | step=3540/50000 progress=7.1% loss=0.410567 lr=9.988176e-05 lr_backbone=9.988176e-05 elapsed=04:07:32 avg_step=4.20s recent_step=4.15s eta=54:08:42 +2026-09-20 10:34:49 | INFO | train rank=0 | step=3550/50000 progress=7.1% loss=0.410296 lr=9.987948e-05 lr_backbone=9.987948e-05 elapsed=04:08:13 avg_step=4.20s recent_step=4.10s eta=54:07:48 +2026-09-20 10:35:30 | INFO | train rank=0 | step=3560/50000 progress=7.1% loss=0.398286 lr=9.987718e-05 lr_backbone=9.987718e-05 elapsed=04:08:54 avg_step=4.20s recent_step=4.12s eta=54:06:57 +2026-09-20 10:36:12 | INFO | train rank=0 | step=3570/50000 progress=7.1% loss=0.378430 lr=9.987485e-05 lr_backbone=9.987485e-05 elapsed=04:09:35 avg_step=4.19s recent_step=4.13s eta=54:06:06 +2026-09-20 10:36:53 | INFO | train rank=0 | step=3580/50000 progress=7.2% loss=0.405150 lr=9.987250e-05 lr_backbone=9.987250e-05 elapsed=04:10:16 avg_step=4.19s recent_step=4.14s eta=54:05:17 +2026-09-20 10:37:34 | INFO | train rank=0 | step=3590/50000 progress=7.2% loss=0.396939 lr=9.987013e-05 lr_backbone=9.987013e-05 elapsed=04:10:58 avg_step=4.19s recent_step=4.10s eta=54:04:23 +2026-09-20 10:38:15 | INFO | train rank=0 | step=3600/50000 progress=7.2% loss=0.426049 lr=9.986773e-05 lr_backbone=9.986773e-05 elapsed=04:11:39 avg_step=4.19s recent_step=4.12s eta=54:03:31 +2026-09-20 10:38:57 | INFO | train rank=0 | step=3610/50000 progress=7.2% loss=0.399528 lr=9.986532e-05 lr_backbone=9.986532e-05 elapsed=04:12:20 avg_step=4.19s recent_step=4.14s eta=54:02:43 +2026-09-20 10:39:38 | INFO | train rank=0 | step=3620/50000 progress=7.2% loss=0.392361 lr=9.986288e-05 lr_backbone=9.986288e-05 elapsed=04:13:01 avg_step=4.19s recent_step=4.11s eta=54:01:50 +2026-09-20 10:40:19 | INFO | train rank=0 | step=3630/50000 progress=7.3% loss=0.446779 lr=9.986043e-05 lr_backbone=9.986043e-05 elapsed=04:13:43 avg_step=4.19s recent_step=4.13s eta=54:01:00 +2026-09-20 10:41:01 | INFO | train rank=0 | step=3640/50000 progress=7.3% loss=0.390639 lr=9.985795e-05 lr_backbone=9.985795e-05 elapsed=04:14:24 avg_step=4.19s recent_step=4.14s eta=54:00:11 +2026-09-20 10:41:41 | INFO | train rank=0 | step=3650/50000 progress=7.3% loss=0.404791 lr=9.985544e-05 lr_backbone=9.985544e-05 elapsed=04:15:05 avg_step=4.19s recent_step=4.09s eta=53:59:16 +2026-09-20 10:42:23 | INFO | train rank=0 | step=3660/50000 progress=7.3% loss=0.535936 lr=9.985292e-05 lr_backbone=9.985292e-05 elapsed=04:15:46 avg_step=4.19s recent_step=4.13s eta=53:58:26 +2026-09-20 10:43:04 | INFO | train rank=0 | step=3670/50000 progress=7.3% loss=0.451477 lr=9.985037e-05 lr_backbone=9.985037e-05 elapsed=04:16:27 avg_step=4.19s recent_step=4.13s eta=53:57:36 +2026-09-20 10:43:45 | INFO | train rank=0 | step=3680/50000 progress=7.4% loss=0.424631 lr=9.984781e-05 lr_backbone=9.984781e-05 elapsed=04:17:09 avg_step=4.19s recent_step=4.13s eta=53:56:46 +2026-09-20 10:44:27 | INFO | train rank=0 | step=3690/50000 progress=7.4% loss=0.419561 lr=9.984522e-05 lr_backbone=9.984522e-05 elapsed=04:17:50 avg_step=4.19s recent_step=4.14s eta=53:55:58 +2026-09-20 10:45:08 | INFO | train rank=0 | step=3700/50000 progress=7.4% loss=0.437097 lr=9.984261e-05 lr_backbone=9.984261e-05 elapsed=04:18:31 avg_step=4.19s recent_step=4.13s eta=53:55:08 +2026-09-20 10:45:49 | INFO | train rank=0 | step=3710/50000 progress=7.4% loss=0.367552 lr=9.983997e-05 lr_backbone=9.983997e-05 elapsed=04:19:12 avg_step=4.19s recent_step=4.09s eta=53:54:13 +2026-09-20 10:46:30 | INFO | train rank=0 | step=3720/50000 progress=7.4% loss=0.336888 lr=9.983732e-05 lr_backbone=9.983732e-05 elapsed=04:19:54 avg_step=4.19s recent_step=4.12s eta=53:53:22 +2026-09-20 10:47:12 | INFO | train rank=0 | step=3730/50000 progress=7.5% loss=0.487117 lr=9.983464e-05 lr_backbone=9.983464e-05 elapsed=04:20:35 avg_step=4.19s recent_step=4.14s eta=53:52:34 +2026-09-20 10:47:53 | INFO | train rank=0 | step=3740/50000 progress=7.5% loss=0.337256 lr=9.983194e-05 lr_backbone=9.983194e-05 elapsed=04:21:16 avg_step=4.19s recent_step=4.10s eta=53:51:41 +2026-09-20 10:48:34 | INFO | train rank=0 | step=3750/50000 progress=7.5% loss=0.408500 lr=9.982922e-05 lr_backbone=9.982922e-05 elapsed=04:21:57 avg_step=4.19s recent_step=4.14s eta=53:50:53 +2026-09-20 10:49:15 | INFO | train rank=0 | step=3760/50000 progress=7.5% loss=0.451626 lr=9.982648e-05 lr_backbone=9.982648e-05 elapsed=04:22:39 avg_step=4.19s recent_step=4.15s eta=53:50:05 +2026-09-20 10:49:56 | INFO | train rank=0 | step=3770/50000 progress=7.5% loss=0.430114 lr=9.982372e-05 lr_backbone=9.982372e-05 elapsed=04:23:20 avg_step=4.19s recent_step=4.08s eta=53:49:10 +2026-09-20 10:50:38 | INFO | train rank=0 | step=3780/50000 progress=7.6% loss=0.405778 lr=9.982093e-05 lr_backbone=9.982093e-05 elapsed=04:24:01 avg_step=4.19s recent_step=4.13s eta=53:48:21 +2026-09-20 10:51:19 | INFO | train rank=0 | step=3790/50000 progress=7.6% loss=0.364407 lr=9.981813e-05 lr_backbone=9.981813e-05 elapsed=04:24:42 avg_step=4.19s recent_step=4.12s eta=53:47:30 +2026-09-20 10:52:00 | INFO | train rank=0 | step=3800/50000 progress=7.6% loss=0.417764 lr=9.981530e-05 lr_backbone=9.981530e-05 elapsed=04:25:23 avg_step=4.19s recent_step=4.13s eta=53:46:41 +2026-09-20 10:52:41 | INFO | train rank=0 | step=3810/50000 progress=7.6% loss=0.459510 lr=9.981245e-05 lr_backbone=9.981245e-05 elapsed=04:26:05 avg_step=4.19s recent_step=4.13s eta=53:45:52 +2026-09-20 10:53:23 | INFO | train rank=0 | step=3820/50000 progress=7.6% loss=0.415814 lr=9.980957e-05 lr_backbone=9.980957e-05 elapsed=04:26:46 avg_step=4.19s recent_step=4.12s eta=53:45:01 +2026-09-20 10:54:03 | INFO | train rank=0 | step=3830/50000 progress=7.7% loss=0.507730 lr=9.980668e-05 lr_backbone=9.980668e-05 elapsed=04:27:27 avg_step=4.19s recent_step=4.09s eta=53:44:07 +2026-09-20 10:54:45 | INFO | train rank=0 | step=3840/50000 progress=7.7% loss=0.433316 lr=9.980376e-05 lr_backbone=9.980376e-05 elapsed=04:28:08 avg_step=4.19s recent_step=4.13s eta=53:43:18 +2026-09-20 10:55:26 | INFO | train rank=0 | step=3850/50000 progress=7.7% loss=0.424434 lr=9.980083e-05 lr_backbone=9.980083e-05 elapsed=04:28:49 avg_step=4.19s recent_step=4.13s eta=53:42:29 +2026-09-20 10:56:08 | INFO | train rank=0 | step=3860/50000 progress=7.7% loss=0.396243 lr=9.979787e-05 lr_backbone=9.979787e-05 elapsed=04:29:31 avg_step=4.19s recent_step=4.17s eta=53:41:44 +2026-09-20 10:56:49 | INFO | train rank=0 | step=3870/50000 progress=7.7% loss=0.398168 lr=9.979489e-05 lr_backbone=9.979489e-05 elapsed=04:30:12 avg_step=4.19s recent_step=4.10s eta=53:40:52 +2026-09-20 10:57:30 | INFO | train rank=0 | step=3880/50000 progress=7.8% loss=0.377154 lr=9.979188e-05 lr_backbone=9.979188e-05 elapsed=04:30:53 avg_step=4.19s recent_step=4.12s eta=53:40:03 +2026-09-20 10:58:11 | INFO | train rank=0 | step=3890/50000 progress=7.8% loss=0.397884 lr=9.978886e-05 lr_backbone=9.978886e-05 elapsed=04:31:34 avg_step=4.19s recent_step=4.10s eta=53:39:10 +2026-09-20 10:58:53 | INFO | train rank=0 | step=3900/50000 progress=7.8% loss=0.388825 lr=9.978581e-05 lr_backbone=9.978581e-05 elapsed=04:32:16 avg_step=4.19s recent_step=4.15s eta=53:38:24 +2026-09-20 10:59:34 | INFO | train rank=0 | step=3910/50000 progress=7.8% loss=0.382113 lr=9.978274e-05 lr_backbone=9.978274e-05 elapsed=04:32:57 avg_step=4.19s recent_step=4.13s eta=53:37:35 +2026-09-20 11:00:15 | INFO | train rank=0 | step=3920/50000 progress=7.8% loss=0.405117 lr=9.977965e-05 lr_backbone=9.977965e-05 elapsed=04:33:39 avg_step=4.19s recent_step=4.14s eta=53:36:47 +2026-09-20 11:00:56 | INFO | train rank=0 | step=3930/50000 progress=7.9% loss=0.388711 lr=9.977654e-05 lr_backbone=9.977654e-05 elapsed=04:34:20 avg_step=4.19s recent_step=4.12s eta=53:35:57 +2026-09-20 11:01:38 | INFO | train rank=0 | step=3940/50000 progress=7.9% loss=0.437544 lr=9.977341e-05 lr_backbone=9.977341e-05 elapsed=04:35:01 avg_step=4.19s recent_step=4.14s eta=53:35:09 +2026-09-20 11:02:19 | INFO | train rank=0 | step=3950/50000 progress=7.9% loss=0.347674 lr=9.977025e-05 lr_backbone=9.977025e-05 elapsed=04:35:42 avg_step=4.19s recent_step=4.10s eta=53:34:18 +2026-09-20 11:03:00 | INFO | train rank=0 | step=3960/50000 progress=7.9% loss=0.450868 lr=9.976707e-05 lr_backbone=9.976707e-05 elapsed=04:36:23 avg_step=4.19s recent_step=4.13s eta=53:33:29 +2026-09-20 11:03:41 | INFO | train rank=0 | step=3970/50000 progress=7.9% loss=0.419936 lr=9.976387e-05 lr_backbone=9.976387e-05 elapsed=04:37:05 avg_step=4.19s recent_step=4.14s eta=53:32:41 +2026-09-20 11:04:23 | INFO | train rank=0 | step=3980/50000 progress=8.0% loss=0.455472 lr=9.976065e-05 lr_backbone=9.976065e-05 elapsed=04:37:46 avg_step=4.19s recent_step=4.13s eta=53:31:52 +2026-09-20 11:05:04 | INFO | train rank=0 | step=3990/50000 progress=8.0% loss=0.425480 lr=9.975741e-05 lr_backbone=9.975741e-05 elapsed=04:38:27 avg_step=4.19s recent_step=4.09s eta=53:30:59 +2026-09-20 11:05:45 | INFO | train rank=0 | step=4000/50000 progress=8.0% loss=0.381063 lr=9.975415e-05 lr_backbone=9.975415e-05 elapsed=04:39:08 avg_step=4.19s recent_step=4.15s eta=53:30:13 +2026-09-20 11:06:26 | INFO | train rank=0 | step=4010/50000 progress=8.0% loss=0.394226 lr=9.975086e-05 lr_backbone=9.975086e-05 elapsed=04:39:49 avg_step=4.19s recent_step=4.08s eta=53:29:19 +2026-09-20 11:07:07 | INFO | train rank=0 | step=4020/50000 progress=8.0% loss=0.438579 lr=9.974755e-05 lr_backbone=9.974755e-05 elapsed=04:40:31 avg_step=4.19s recent_step=4.14s eta=53:28:32 +2026-09-20 11:07:49 | INFO | train rank=0 | step=4030/50000 progress=8.1% loss=0.399484 lr=9.974422e-05 lr_backbone=9.974422e-05 elapsed=04:41:12 avg_step=4.19s recent_step=4.13s eta=53:27:44 +2026-09-20 11:08:30 | INFO | train rank=0 | step=4040/50000 progress=8.1% loss=0.414420 lr=9.974087e-05 lr_backbone=9.974087e-05 elapsed=04:41:53 avg_step=4.19s recent_step=4.14s eta=53:26:57 +2026-09-20 11:09:11 | INFO | train rank=0 | step=4050/50000 progress=8.1% loss=0.413140 lr=9.973750e-05 lr_backbone=9.973750e-05 elapsed=04:42:35 avg_step=4.19s recent_step=4.11s eta=53:26:06 +2026-09-20 11:09:53 | INFO | train rank=0 | step=4060/50000 progress=8.1% loss=0.408459 lr=9.973410e-05 lr_backbone=9.973410e-05 elapsed=04:43:16 avg_step=4.19s recent_step=4.17s eta=53:25:21 +2026-09-20 11:10:34 | INFO | train rank=0 | step=4070/50000 progress=8.1% loss=0.340515 lr=9.973068e-05 lr_backbone=9.973068e-05 elapsed=04:43:57 avg_step=4.19s recent_step=4.09s eta=53:24:29 +2026-09-20 11:11:15 | INFO | train rank=0 | step=4080/50000 progress=8.2% loss=0.409332 lr=9.972725e-05 lr_backbone=9.972725e-05 elapsed=04:44:39 avg_step=4.19s recent_step=4.15s eta=53:23:43 +2026-09-20 11:11:57 | INFO | train rank=0 | step=4090/50000 progress=8.2% loss=0.372617 lr=9.972379e-05 lr_backbone=9.972379e-05 elapsed=04:45:20 avg_step=4.19s recent_step=4.13s eta=53:22:55 +2026-09-20 11:12:38 | INFO | train rank=0 | step=4100/50000 progress=8.2% loss=0.380739 lr=9.972030e-05 lr_backbone=9.972030e-05 elapsed=04:46:01 avg_step=4.19s recent_step=4.13s eta=53:22:07 +2026-09-20 11:13:19 | INFO | train rank=0 | step=4110/50000 progress=8.2% loss=0.406580 lr=9.971680e-05 lr_backbone=9.971680e-05 elapsed=04:46:42 avg_step=4.19s recent_step=4.12s eta=53:21:18 +2026-09-20 11:14:00 | INFO | train rank=0 | step=4120/50000 progress=8.2% loss=0.355137 lr=9.971327e-05 lr_backbone=9.971327e-05 elapsed=04:47:23 avg_step=4.19s recent_step=4.10s eta=53:20:27 +2026-09-20 11:14:41 | INFO | train rank=0 | step=4130/50000 progress=8.3% loss=0.435465 lr=9.970973e-05 lr_backbone=9.970973e-05 elapsed=04:48:04 avg_step=4.19s recent_step=4.09s eta=53:19:35 +2026-09-20 11:15:23 | INFO | train rank=0 | step=4140/50000 progress=8.3% loss=0.401951 lr=9.970616e-05 lr_backbone=9.970616e-05 elapsed=04:48:46 avg_step=4.19s recent_step=4.15s eta=53:18:49 +2026-09-20 11:16:04 | INFO | train rank=0 | step=4150/50000 progress=8.3% loss=0.464595 lr=9.970257e-05 lr_backbone=9.970257e-05 elapsed=04:49:27 avg_step=4.18s recent_step=4.13s eta=53:18:01 +2026-09-20 11:16:45 | INFO | train rank=0 | step=4160/50000 progress=8.3% loss=0.350829 lr=9.969895e-05 lr_backbone=9.969895e-05 elapsed=04:50:09 avg_step=4.18s recent_step=4.13s eta=53:17:13 +2026-09-20 11:17:26 | INFO | train rank=0 | step=4170/50000 progress=8.3% loss=0.380103 lr=9.969532e-05 lr_backbone=9.969532e-05 elapsed=04:50:50 avg_step=4.18s recent_step=4.13s eta=53:16:25 +2026-09-20 11:18:07 | INFO | train rank=0 | step=4180/50000 progress=8.4% loss=0.379763 lr=9.969166e-05 lr_backbone=9.969166e-05 elapsed=04:51:31 avg_step=4.18s recent_step=4.10s eta=53:15:34 +2026-09-20 11:18:48 | INFO | train rank=0 | step=4190/50000 progress=8.4% loss=0.408995 lr=9.968799e-05 lr_backbone=9.968799e-05 elapsed=04:52:12 avg_step=4.18s recent_step=4.10s eta=53:14:43 +2026-09-20 11:19:30 | INFO | train rank=0 | step=4200/50000 progress=8.4% loss=0.423627 lr=9.968429e-05 lr_backbone=9.968429e-05 elapsed=04:52:53 avg_step=4.18s recent_step=4.14s eta=53:13:57 +2026-09-20 11:20:11 | INFO | train rank=0 | step=4210/50000 progress=8.4% loss=0.478562 lr=9.968057e-05 lr_backbone=9.968057e-05 elapsed=04:53:34 avg_step=4.18s recent_step=4.11s eta=53:13:06 +2026-09-20 11:20:52 | INFO | train rank=0 | step=4220/50000 progress=8.4% loss=0.421078 lr=9.967682e-05 lr_backbone=9.967682e-05 elapsed=04:54:16 avg_step=4.18s recent_step=4.14s eta=53:12:20 +2026-09-20 11:21:34 | INFO | train rank=0 | step=4230/50000 progress=8.5% loss=0.416595 lr=9.967306e-05 lr_backbone=9.967306e-05 elapsed=04:54:57 avg_step=4.18s recent_step=4.13s eta=53:11:33 +2026-09-20 11:22:15 | INFO | train rank=0 | step=4240/50000 progress=8.5% loss=0.478619 lr=9.966927e-05 lr_backbone=9.966927e-05 elapsed=04:55:38 avg_step=4.18s recent_step=4.12s eta=53:10:44 +2026-09-20 11:22:56 | INFO | train rank=0 | step=4250/50000 progress=8.5% loss=0.410967 lr=9.966546e-05 lr_backbone=9.966546e-05 elapsed=04:56:19 avg_step=4.18s recent_step=4.10s eta=53:09:53 +2026-09-20 11:23:37 | INFO | train rank=0 | step=4260/50000 progress=8.5% loss=0.473366 lr=9.966163e-05 lr_backbone=9.966163e-05 elapsed=04:57:00 avg_step=4.18s recent_step=4.13s eta=53:09:05 +2026-09-20 11:24:19 | INFO | train rank=0 | step=4270/50000 progress=8.5% loss=0.520943 lr=9.965778e-05 lr_backbone=9.965778e-05 elapsed=04:57:42 avg_step=4.18s recent_step=4.14s eta=53:08:18 +2026-09-20 11:25:00 | INFO | train rank=0 | step=4280/50000 progress=8.6% loss=0.369692 lr=9.965391e-05 lr_backbone=9.965391e-05 elapsed=04:58:23 avg_step=4.18s recent_step=4.13s eta=53:07:30 +2026-09-20 11:25:41 | INFO | train rank=0 | step=4290/50000 progress=8.6% loss=0.382545 lr=9.965001e-05 lr_backbone=9.965001e-05 elapsed=04:59:04 avg_step=4.18s recent_step=4.13s eta=53:06:43 +2026-09-20 11:26:22 | INFO | train rank=0 | step=4300/50000 progress=8.6% loss=0.399293 lr=9.964610e-05 lr_backbone=9.964610e-05 elapsed=04:59:45 avg_step=4.18s recent_step=4.09s eta=53:05:52 +2026-09-20 11:27:03 | INFO | train rank=0 | step=4310/50000 progress=8.6% loss=0.461180 lr=9.964216e-05 lr_backbone=9.964216e-05 elapsed=05:00:26 avg_step=4.18s recent_step=4.07s eta=53:04:58 +2026-09-20 11:27:44 | INFO | train rank=0 | step=4320/50000 progress=8.6% loss=0.369994 lr=9.963820e-05 lr_backbone=9.963820e-05 elapsed=05:01:07 avg_step=4.18s recent_step=4.10s eta=53:04:07 +2026-09-20 11:28:25 | INFO | train rank=0 | step=4330/50000 progress=8.7% loss=0.490983 lr=9.963422e-05 lr_backbone=9.963422e-05 elapsed=05:01:48 avg_step=4.18s recent_step=4.11s eta=53:03:18 +2026-09-20 11:29:06 | INFO | train rank=0 | step=4340/50000 progress=8.7% loss=0.434572 lr=9.963021e-05 lr_backbone=9.963021e-05 elapsed=05:02:29 avg_step=4.18s recent_step=4.12s eta=53:02:29 +2026-09-20 11:29:47 | INFO | train rank=0 | step=4350/50000 progress=8.7% loss=0.448002 lr=9.962619e-05 lr_backbone=9.962619e-05 elapsed=05:03:11 avg_step=4.18s recent_step=4.13s eta=53:01:42 +2026-09-20 11:30:29 | INFO | train rank=0 | step=4360/50000 progress=8.7% loss=0.480624 lr=9.962214e-05 lr_backbone=9.962214e-05 elapsed=05:03:52 avg_step=4.18s recent_step=4.12s eta=53:00:54 +2026-09-20 11:31:09 | INFO | train rank=0 | step=4370/50000 progress=8.7% loss=0.447976 lr=9.961807e-05 lr_backbone=9.961807e-05 elapsed=05:04:32 avg_step=4.18s recent_step=4.06s eta=53:00:00 +2026-09-20 11:31:50 | INFO | train rank=0 | step=4380/50000 progress=8.8% loss=0.435189 lr=9.961398e-05 lr_backbone=9.961398e-05 elapsed=05:05:14 avg_step=4.18s recent_step=4.12s eta=52:59:12 +2026-09-20 11:32:32 | INFO | train rank=0 | step=4390/50000 progress=8.8% loss=0.414193 lr=9.960987e-05 lr_backbone=9.960987e-05 elapsed=05:05:55 avg_step=4.18s recent_step=4.12s eta=52:58:23 +2026-09-20 11:33:13 | INFO | train rank=0 | step=4400/50000 progress=8.8% loss=0.377288 lr=9.960574e-05 lr_backbone=9.960574e-05 elapsed=05:06:36 avg_step=4.18s recent_step=4.16s eta=52:57:39 +2026-09-20 11:33:54 | INFO | train rank=0 | step=4410/50000 progress=8.8% loss=0.378347 lr=9.960158e-05 lr_backbone=9.960158e-05 elapsed=05:07:17 avg_step=4.18s recent_step=4.10s eta=52:56:48 +2026-09-20 11:34:36 | INFO | train rank=0 | step=4420/50000 progress=8.8% loss=0.457486 lr=9.959740e-05 lr_backbone=9.959740e-05 elapsed=05:07:59 avg_step=4.18s recent_step=4.15s eta=52:56:03 +2026-09-20 11:35:16 | INFO | train rank=0 | step=4430/50000 progress=8.9% loss=0.423490 lr=9.959320e-05 lr_backbone=9.959320e-05 elapsed=05:08:40 avg_step=4.18s recent_step=4.07s eta=52:55:10 +2026-09-20 11:35:58 | INFO | train rank=0 | step=4440/50000 progress=8.9% loss=0.395587 lr=9.958898e-05 lr_backbone=9.958898e-05 elapsed=05:09:21 avg_step=4.18s recent_step=4.14s eta=52:54:24 +2026-09-20 11:36:39 | INFO | train rank=0 | step=4450/50000 progress=8.9% loss=0.360812 lr=9.958474e-05 lr_backbone=9.958474e-05 elapsed=05:10:03 avg_step=4.18s recent_step=4.15s eta=52:53:40 +2026-09-20 11:37:20 | INFO | train rank=0 | step=4460/50000 progress=8.9% loss=0.410586 lr=9.958048e-05 lr_backbone=9.958048e-05 elapsed=05:10:44 avg_step=4.18s recent_step=4.12s eta=52:52:52 +2026-09-20 11:38:02 | INFO | train rank=0 | step=4470/50000 progress=8.9% loss=0.337980 lr=9.957619e-05 lr_backbone=9.957619e-05 elapsed=05:11:25 avg_step=4.18s recent_step=4.12s eta=52:52:04 +2026-09-20 11:38:43 | INFO | train rank=0 | step=4480/50000 progress=9.0% loss=0.432972 lr=9.957188e-05 lr_backbone=9.957188e-05 elapsed=05:12:06 avg_step=4.18s recent_step=4.12s eta=52:51:16 +2026-09-20 11:39:24 | INFO | train rank=0 | step=4490/50000 progress=9.0% loss=0.350008 lr=9.956755e-05 lr_backbone=9.956755e-05 elapsed=05:12:47 avg_step=4.18s recent_step=4.10s eta=52:50:26 +2026-09-20 11:40:05 | INFO | train rank=0 | step=4500/50000 progress=9.0% loss=0.394167 lr=9.956320e-05 lr_backbone=9.956320e-05 elapsed=05:13:28 avg_step=4.18s recent_step=4.11s eta=52:49:37 +2026-09-20 11:40:46 | INFO | train rank=0 | step=4510/50000 progress=9.0% loss=0.357585 lr=9.955883e-05 lr_backbone=9.955883e-05 elapsed=05:14:10 avg_step=4.18s recent_step=4.13s eta=52:48:50 +2026-09-20 11:41:28 | INFO | train rank=0 | step=4520/50000 progress=9.0% loss=0.443163 lr=9.955444e-05 lr_backbone=9.955444e-05 elapsed=05:14:51 avg_step=4.18s recent_step=4.14s eta=52:48:05 +2026-09-20 11:42:09 | INFO | train rank=0 | step=4530/50000 progress=9.1% loss=0.445740 lr=9.955002e-05 lr_backbone=9.955002e-05 elapsed=05:15:32 avg_step=4.18s recent_step=4.14s eta=52:47:19 +2026-09-20 11:42:50 | INFO | train rank=0 | step=4540/50000 progress=9.1% loss=0.396140 lr=9.954558e-05 lr_backbone=9.954558e-05 elapsed=05:16:14 avg_step=4.18s recent_step=4.12s eta=52:46:31 +2026-09-20 11:43:31 | INFO | train rank=0 | step=4550/50000 progress=9.1% loss=0.436859 lr=9.954112e-05 lr_backbone=9.954112e-05 elapsed=05:16:55 avg_step=4.18s recent_step=4.09s eta=52:45:41 +2026-09-20 11:44:12 | INFO | train rank=0 | step=4560/50000 progress=9.1% loss=0.409796 lr=9.953664e-05 lr_backbone=9.953664e-05 elapsed=05:17:36 avg_step=4.18s recent_step=4.10s eta=52:44:51 +2026-09-20 11:44:53 | INFO | train rank=0 | step=4570/50000 progress=9.1% loss=0.456221 lr=9.953214e-05 lr_backbone=9.953214e-05 elapsed=05:18:17 avg_step=4.18s recent_step=4.12s eta=52:44:04 +2026-09-20 11:45:35 | INFO | train rank=0 | step=4580/50000 progress=9.2% loss=0.416162 lr=9.952762e-05 lr_backbone=9.952762e-05 elapsed=05:18:58 avg_step=4.18s recent_step=4.15s eta=52:43:19 +2026-09-20 11:46:16 | INFO | train rank=0 | step=4590/50000 progress=9.2% loss=0.435213 lr=9.952307e-05 lr_backbone=9.952307e-05 elapsed=05:19:40 avg_step=4.18s recent_step=4.14s eta=52:42:33 +2026-09-20 11:46:58 | INFO | train rank=0 | step=4600/50000 progress=9.2% loss=0.492597 lr=9.951850e-05 lr_backbone=9.951850e-05 elapsed=05:20:21 avg_step=4.18s recent_step=4.14s eta=52:41:47 +2026-09-20 11:47:39 | INFO | train rank=0 | step=4610/50000 progress=9.2% loss=0.389462 lr=9.951391e-05 lr_backbone=9.951391e-05 elapsed=05:21:02 avg_step=4.18s recent_step=4.10s eta=52:40:58 +2026-09-20 11:48:20 | INFO | train rank=0 | step=4620/50000 progress=9.2% loss=0.418190 lr=9.950930e-05 lr_backbone=9.950930e-05 elapsed=05:21:43 avg_step=4.18s recent_step=4.13s eta=52:40:12 +2026-09-20 11:49:01 | INFO | train rank=0 | step=4630/50000 progress=9.3% loss=0.404105 lr=9.950467e-05 lr_backbone=9.950467e-05 elapsed=05:22:24 avg_step=4.18s recent_step=4.10s eta=52:39:22 +2026-09-20 11:49:42 | INFO | train rank=0 | step=4640/50000 progress=9.3% loss=0.428864 lr=9.950002e-05 lr_backbone=9.950002e-05 elapsed=05:23:06 avg_step=4.18s recent_step=4.14s eta=52:38:37 +2026-09-20 11:50:24 | INFO | train rank=0 | step=4650/50000 progress=9.3% loss=0.467624 lr=9.949534e-05 lr_backbone=9.949534e-05 elapsed=05:23:47 avg_step=4.18s recent_step=4.13s eta=52:37:50 +2026-09-20 11:51:05 | INFO | train rank=0 | step=4660/50000 progress=9.3% loss=0.358118 lr=9.949064e-05 lr_backbone=9.949064e-05 elapsed=05:24:28 avg_step=4.18s recent_step=4.13s eta=52:37:03 +2026-09-20 11:51:46 | INFO | train rank=0 | step=4670/50000 progress=9.3% loss=0.416739 lr=9.948593e-05 lr_backbone=9.948593e-05 elapsed=05:25:09 avg_step=4.18s recent_step=4.08s eta=52:36:12 +2026-09-20 11:52:27 | INFO | train rank=0 | step=4680/50000 progress=9.4% loss=0.378430 lr=9.948118e-05 lr_backbone=9.948118e-05 elapsed=05:25:50 avg_step=4.18s recent_step=4.13s eta=52:35:26 +2026-09-20 11:53:08 | INFO | train rank=0 | step=4690/50000 progress=9.4% loss=0.458493 lr=9.947642e-05 lr_backbone=9.947642e-05 elapsed=05:26:31 avg_step=4.18s recent_step=4.10s eta=52:34:37 +2026-09-20 11:53:49 | INFO | train rank=0 | step=4700/50000 progress=9.4% loss=0.407889 lr=9.947164e-05 lr_backbone=9.947164e-05 elapsed=05:27:13 avg_step=4.18s recent_step=4.14s eta=52:33:51 +2026-09-20 11:54:31 | INFO | train rank=0 | step=4710/50000 progress=9.4% loss=0.506373 lr=9.946683e-05 lr_backbone=9.946683e-05 elapsed=05:27:54 avg_step=4.18s recent_step=4.14s eta=52:33:06 +2026-09-20 11:55:12 | INFO | train rank=0 | step=4720/50000 progress=9.4% loss=0.352459 lr=9.946201e-05 lr_backbone=9.946201e-05 elapsed=05:28:36 avg_step=4.18s recent_step=4.13s eta=52:32:19 +2026-09-20 11:55:53 | INFO | train rank=0 | step=4730/50000 progress=9.5% loss=0.451927 lr=9.945716e-05 lr_backbone=9.945716e-05 elapsed=05:29:16 avg_step=4.18s recent_step=4.09s eta=52:31:30 +2026-09-20 11:56:34 | INFO | train rank=0 | step=4740/50000 progress=9.5% loss=0.477594 lr=9.945229e-05 lr_backbone=9.945229e-05 elapsed=05:29:58 avg_step=4.18s recent_step=4.14s eta=52:30:44 +2026-09-20 11:57:16 | INFO | train rank=0 | step=4750/50000 progress=9.5% loss=0.394920 lr=9.944739e-05 lr_backbone=9.944739e-05 elapsed=05:30:39 avg_step=4.18s recent_step=4.11s eta=52:29:56 +2026-09-20 11:57:57 | INFO | train rank=0 | step=4760/50000 progress=9.5% loss=0.376283 lr=9.944248e-05 lr_backbone=9.944248e-05 elapsed=05:31:20 avg_step=4.18s recent_step=4.11s eta=52:29:08 +2026-09-20 11:58:38 | INFO | train rank=0 | step=4770/50000 progress=9.5% loss=0.447389 lr=9.943754e-05 lr_backbone=9.943754e-05 elapsed=05:32:02 avg_step=4.18s recent_step=4.15s eta=52:28:24 +2026-09-20 11:59:20 | INFO | train rank=0 | step=4780/50000 progress=9.6% loss=0.501244 lr=9.943259e-05 lr_backbone=9.943259e-05 elapsed=05:32:43 avg_step=4.18s recent_step=4.13s eta=52:27:38 +2026-09-20 12:00:00 | INFO | train rank=0 | step=4790/50000 progress=9.6% loss=0.413940 lr=9.942761e-05 lr_backbone=9.942761e-05 elapsed=05:33:24 avg_step=4.18s recent_step=4.09s eta=52:26:47 +2026-09-20 12:00:41 | INFO | train rank=0 | step=4800/50000 progress=9.6% loss=0.379581 lr=9.942261e-05 lr_backbone=9.942261e-05 elapsed=05:34:05 avg_step=4.18s recent_step=4.11s eta=52:26:00 +2026-09-20 12:01:23 | INFO | train rank=0 | step=4810/50000 progress=9.6% loss=0.383721 lr=9.941759e-05 lr_backbone=9.941759e-05 elapsed=05:34:46 avg_step=4.18s recent_step=4.13s eta=52:25:13 +2026-09-20 12:02:04 | INFO | train rank=0 | step=4820/50000 progress=9.6% loss=0.418841 lr=9.941254e-05 lr_backbone=9.941254e-05 elapsed=05:35:27 avg_step=4.18s recent_step=4.14s eta=52:24:28 +2026-09-20 12:02:45 | INFO | train rank=0 | step=4830/50000 progress=9.7% loss=0.383513 lr=9.940748e-05 lr_backbone=9.940748e-05 elapsed=05:36:09 avg_step=4.18s recent_step=4.13s eta=52:23:42 +2026-09-20 12:03:27 | INFO | train rank=0 | step=4840/50000 progress=9.7% loss=0.402052 lr=9.940239e-05 lr_backbone=9.940239e-05 elapsed=05:36:50 avg_step=4.18s recent_step=4.15s eta=52:22:57 +2026-09-20 12:04:08 | INFO | train rank=0 | step=4850/50000 progress=9.7% loss=0.564920 lr=9.939728e-05 lr_backbone=9.939728e-05 elapsed=05:37:31 avg_step=4.18s recent_step=4.11s eta=52:22:09 +2026-09-20 12:04:49 | INFO | train rank=0 | step=4860/50000 progress=9.7% loss=0.424061 lr=9.939215e-05 lr_backbone=9.939215e-05 elapsed=05:38:13 avg_step=4.18s recent_step=4.13s eta=52:21:23 +2026-09-20 12:05:31 | INFO | train rank=0 | step=4870/50000 progress=9.7% loss=0.404289 lr=9.938700e-05 lr_backbone=9.938700e-05 elapsed=05:38:54 avg_step=4.18s recent_step=4.15s eta=52:20:39 +2026-09-20 12:06:12 | INFO | train rank=0 | step=4880/50000 progress=9.8% loss=0.389096 lr=9.938183e-05 lr_backbone=9.938183e-05 elapsed=05:39:35 avg_step=4.18s recent_step=4.08s eta=52:19:48 +2026-09-20 12:06:53 | INFO | train rank=0 | step=4890/50000 progress=9.8% loss=0.353439 lr=9.937663e-05 lr_backbone=9.937663e-05 elapsed=05:40:16 avg_step=4.18s recent_step=4.13s eta=52:19:02 +2026-09-20 12:07:34 | INFO | train rank=0 | step=4900/50000 progress=9.8% loss=0.344774 lr=9.937142e-05 lr_backbone=9.937142e-05 elapsed=05:40:57 avg_step=4.18s recent_step=4.12s eta=52:18:15 +2026-09-20 12:08:15 | INFO | train rank=0 | step=4910/50000 progress=9.8% loss=0.489597 lr=9.936618e-05 lr_backbone=9.936618e-05 elapsed=05:41:38 avg_step=4.17s recent_step=4.09s eta=52:17:26 +2026-09-20 12:08:56 | INFO | train rank=0 | step=4920/50000 progress=9.8% loss=0.351830 lr=9.936092e-05 lr_backbone=9.936092e-05 elapsed=05:42:19 avg_step=4.17s recent_step=4.12s eta=52:16:39 +2026-09-20 12:09:38 | INFO | train rank=0 | step=4930/50000 progress=9.9% loss=0.411120 lr=9.935564e-05 lr_backbone=9.935564e-05 elapsed=05:43:01 avg_step=4.17s recent_step=4.14s eta=52:15:54 +2026-09-20 12:10:19 | INFO | train rank=0 | step=4940/50000 progress=9.9% loss=0.355824 lr=9.935034e-05 lr_backbone=9.935034e-05 elapsed=05:43:42 avg_step=4.17s recent_step=4.11s eta=52:15:07 +2026-09-20 12:11:00 | INFO | train rank=0 | step=4950/50000 progress=9.9% loss=0.393771 lr=9.934501e-05 lr_backbone=9.934501e-05 elapsed=05:44:23 avg_step=4.17s recent_step=4.12s eta=52:14:20 +2026-09-20 12:11:41 | INFO | train rank=0 | step=4960/50000 progress=9.9% loss=0.373170 lr=9.933966e-05 lr_backbone=9.933966e-05 elapsed=05:45:04 avg_step=4.17s recent_step=4.12s eta=52:13:34 +2026-09-20 12:12:22 | INFO | train rank=0 | step=4970/50000 progress=9.9% loss=0.378590 lr=9.933430e-05 lr_backbone=9.933430e-05 elapsed=05:45:45 avg_step=4.17s recent_step=4.09s eta=52:12:45 +2026-09-20 12:13:04 | INFO | train rank=0 | step=4980/50000 progress=10.0% loss=0.391900 lr=9.932891e-05 lr_backbone=9.932891e-05 elapsed=05:46:27 avg_step=4.17s recent_step=4.16s eta=52:12:01 +2026-09-20 12:13:45 | INFO | train rank=0 | step=4990/50000 progress=10.0% loss=0.342967 lr=9.932350e-05 lr_backbone=9.932350e-05 elapsed=05:47:08 avg_step=4.17s recent_step=4.12s eta=52:11:15 +2026-09-20 12:14:26 | INFO | train rank=0 | step=5000/50000 progress=10.0% loss=0.355673 lr=9.931807e-05 lr_backbone=9.931807e-05 elapsed=05:47:49 avg_step=4.17s recent_step=4.10s eta=52:10:26 +2026-09-20 12:15:07 | INFO | train rank=0 | step=5010/50000 progress=10.0% loss=0.390877 lr=9.931261e-05 lr_backbone=9.931261e-05 elapsed=05:48:30 avg_step=4.17s recent_step=4.12s eta=52:09:40 +2026-09-20 12:15:48 | INFO | train rank=0 | step=5020/50000 progress=10.0% loss=0.359129 lr=9.930714e-05 lr_backbone=9.930714e-05 elapsed=05:49:12 avg_step=4.17s recent_step=4.13s eta=52:08:54 +2026-09-20 12:16:29 | INFO | train rank=0 | step=5030/50000 progress=10.1% loss=0.355891 lr=9.930164e-05 lr_backbone=9.930164e-05 elapsed=05:49:53 avg_step=4.17s recent_step=4.09s eta=52:08:05 +2026-09-20 12:17:11 | INFO | train rank=0 | step=5040/50000 progress=10.1% loss=0.428512 lr=9.929612e-05 lr_backbone=9.929612e-05 elapsed=05:50:34 avg_step=4.17s recent_step=4.14s eta=52:07:20 +2026-09-20 12:17:52 | INFO | train rank=0 | step=5050/50000 progress=10.1% loss=0.392084 lr=9.929058e-05 lr_backbone=9.929058e-05 elapsed=05:51:15 avg_step=4.17s recent_step=4.13s eta=52:06:34 +2026-09-20 12:18:33 | INFO | train rank=0 | step=5060/50000 progress=10.1% loss=0.396846 lr=9.928502e-05 lr_backbone=9.928502e-05 elapsed=05:51:57 avg_step=4.17s recent_step=4.14s eta=52:05:50 +2026-09-20 12:19:14 | INFO | train rank=0 | step=5070/50000 progress=10.1% loss=0.355691 lr=9.927944e-05 lr_backbone=9.927944e-05 elapsed=05:52:38 avg_step=4.17s recent_step=4.12s eta=52:05:03 +2026-09-20 12:19:56 | INFO | train rank=0 | step=5080/50000 progress=10.2% loss=0.355435 lr=9.927383e-05 lr_backbone=9.927383e-05 elapsed=05:53:19 avg_step=4.17s recent_step=4.14s eta=52:04:18 +2026-09-20 12:20:37 | INFO | train rank=0 | step=5090/50000 progress=10.2% loss=0.471126 lr=9.926820e-05 lr_backbone=9.926820e-05 elapsed=05:54:00 avg_step=4.17s recent_step=4.09s eta=52:03:29 +2026-09-20 12:21:18 | INFO | train rank=0 | step=5100/50000 progress=10.2% loss=0.487028 lr=9.926256e-05 lr_backbone=9.926256e-05 elapsed=05:54:42 avg_step=4.17s recent_step=4.15s eta=52:02:45 +2026-09-20 12:21:59 | INFO | train rank=0 | step=5110/50000 progress=10.2% loss=0.478692 lr=9.925689e-05 lr_backbone=9.925689e-05 elapsed=05:55:23 avg_step=4.17s recent_step=4.12s eta=52:01:58 +2026-09-20 12:22:41 | INFO | train rank=0 | step=5120/50000 progress=10.2% loss=0.442309 lr=9.925120e-05 lr_backbone=9.925120e-05 elapsed=05:56:04 avg_step=4.17s recent_step=4.13s eta=52:01:13 +2026-09-20 12:23:22 | INFO | train rank=0 | step=5130/50000 progress=10.3% loss=0.341175 lr=9.924548e-05 lr_backbone=9.924548e-05 elapsed=05:56:45 avg_step=4.17s recent_step=4.09s eta=52:00:24 +2026-09-20 12:24:03 | INFO | train rank=0 | step=5140/50000 progress=10.3% loss=0.438997 lr=9.923975e-05 lr_backbone=9.923975e-05 elapsed=05:57:26 avg_step=4.17s recent_step=4.14s eta=51:59:40 +2026-09-20 12:24:44 | INFO | train rank=0 | step=5150/50000 progress=10.3% loss=0.402667 lr=9.923399e-05 lr_backbone=9.923399e-05 elapsed=05:58:07 avg_step=4.17s recent_step=4.09s eta=51:58:51 +2026-09-20 12:25:25 | INFO | train rank=0 | step=5160/50000 progress=10.3% loss=0.388495 lr=9.922822e-05 lr_backbone=9.922822e-05 elapsed=05:58:49 avg_step=4.17s recent_step=4.13s eta=51:58:05 +2026-09-20 12:26:06 | INFO | train rank=0 | step=5170/50000 progress=10.3% loss=0.339235 lr=9.922242e-05 lr_backbone=9.922242e-05 elapsed=05:59:30 avg_step=4.17s recent_step=4.13s eta=51:57:20 +2026-09-20 12:26:48 | INFO | train rank=0 | step=5180/50000 progress=10.4% loss=0.486375 lr=9.921660e-05 lr_backbone=9.921660e-05 elapsed=06:00:11 avg_step=4.17s recent_step=4.12s eta=51:56:33 +2026-09-20 12:27:29 | INFO | train rank=0 | step=5190/50000 progress=10.4% loss=0.353801 lr=9.921076e-05 lr_backbone=9.921076e-05 elapsed=06:00:52 avg_step=4.17s recent_step=4.09s eta=51:55:45 +2026-09-20 12:28:10 | INFO | train rank=0 | step=5200/50000 progress=10.4% loss=0.365089 lr=9.920489e-05 lr_backbone=9.920489e-05 elapsed=06:01:33 avg_step=4.17s recent_step=4.13s eta=51:54:59 +2026-09-20 12:28:51 | INFO | train rank=0 | step=5210/50000 progress=10.4% loss=0.364578 lr=9.919901e-05 lr_backbone=9.919901e-05 elapsed=06:02:14 avg_step=4.17s recent_step=4.09s eta=51:54:10 +2026-09-20 12:29:32 | INFO | train rank=0 | step=5220/50000 progress=10.4% loss=0.451763 lr=9.919310e-05 lr_backbone=9.919310e-05 elapsed=06:02:55 avg_step=4.17s recent_step=4.12s eta=51:53:24 +2026-09-20 12:30:13 | INFO | train rank=0 | step=5230/50000 progress=10.5% loss=0.374458 lr=9.918717e-05 lr_backbone=9.918717e-05 elapsed=06:03:37 avg_step=4.17s recent_step=4.13s eta=51:52:39 +2026-09-20 12:30:55 | INFO | train rank=0 | step=5240/50000 progress=10.5% loss=0.380680 lr=9.918122e-05 lr_backbone=9.918122e-05 elapsed=06:04:18 avg_step=4.17s recent_step=4.14s eta=51:51:54 +2026-09-20 12:31:36 | INFO | train rank=0 | step=5250/50000 progress=10.5% loss=0.386409 lr=9.917525e-05 lr_backbone=9.917525e-05 elapsed=06:04:59 avg_step=4.17s recent_step=4.14s eta=51:51:10 +2026-09-20 12:32:17 | INFO | train rank=0 | step=5260/50000 progress=10.5% loss=0.469975 lr=9.916926e-05 lr_backbone=9.916926e-05 elapsed=06:05:41 avg_step=4.17s recent_step=4.13s eta=51:50:25 +2026-09-20 12:32:58 | INFO | train rank=0 | step=5270/50000 progress=10.5% loss=0.383623 lr=9.916325e-05 lr_backbone=9.916325e-05 elapsed=06:06:21 avg_step=4.17s recent_step=4.08s eta=51:49:35 +2026-09-20 12:33:39 | INFO | train rank=0 | step=5280/50000 progress=10.6% loss=0.411114 lr=9.915721e-05 lr_backbone=9.915721e-05 elapsed=06:07:03 avg_step=4.17s recent_step=4.14s eta=51:48:50 +2026-09-20 12:34:21 | INFO | train rank=0 | step=5290/50000 progress=10.6% loss=0.423214 lr=9.915116e-05 lr_backbone=9.915116e-05 elapsed=06:07:44 avg_step=4.17s recent_step=4.13s eta=51:48:05 +2026-09-20 12:35:02 | INFO | train rank=0 | step=5300/50000 progress=10.6% loss=0.339265 lr=9.914508e-05 lr_backbone=9.914508e-05 elapsed=06:08:25 avg_step=4.17s recent_step=4.11s eta=51:47:18 +2026-09-20 12:35:43 | INFO | train rank=0 | step=5310/50000 progress=10.6% loss=0.412062 lr=9.913898e-05 lr_backbone=9.913898e-05 elapsed=06:09:06 avg_step=4.17s recent_step=4.13s eta=51:46:33 +2026-09-20 12:36:24 | INFO | train rank=0 | step=5320/50000 progress=10.6% loss=0.365535 lr=9.913286e-05 lr_backbone=9.913286e-05 elapsed=06:09:48 avg_step=4.17s recent_step=4.12s eta=51:45:47 +2026-09-20 12:37:05 | INFO | train rank=0 | step=5330/50000 progress=10.7% loss=0.456530 lr=9.912671e-05 lr_backbone=9.912671e-05 elapsed=06:10:29 avg_step=4.17s recent_step=4.10s eta=51:44:59 +2026-09-20 12:37:47 | INFO | train rank=0 | step=5340/50000 progress=10.7% loss=0.424225 lr=9.912055e-05 lr_backbone=9.912055e-05 elapsed=06:11:10 avg_step=4.17s recent_step=4.13s eta=51:44:14 +2026-09-20 12:38:28 | INFO | train rank=0 | step=5350/50000 progress=10.7% loss=0.338810 lr=9.911436e-05 lr_backbone=9.911436e-05 elapsed=06:11:51 avg_step=4.17s recent_step=4.13s eta=51:43:29 +2026-09-20 12:39:09 | INFO | train rank=0 | step=5360/50000 progress=10.7% loss=0.361748 lr=9.910816e-05 lr_backbone=9.910816e-05 elapsed=06:12:32 avg_step=4.17s recent_step=4.12s eta=51:42:43 +2026-09-20 12:39:50 | INFO | train rank=0 | step=5370/50000 progress=10.7% loss=0.361869 lr=9.910193e-05 lr_backbone=9.910193e-05 elapsed=06:13:14 avg_step=4.17s recent_step=4.12s eta=51:41:57 +2026-09-20 12:40:31 | INFO | train rank=0 | step=5380/50000 progress=10.8% loss=0.407027 lr=9.909568e-05 lr_backbone=9.909568e-05 elapsed=06:13:55 avg_step=4.17s recent_step=4.09s eta=51:41:09 +2026-09-20 12:41:12 | INFO | train rank=0 | step=5390/50000 progress=10.8% loss=0.379764 lr=9.908940e-05 lr_backbone=9.908940e-05 elapsed=06:14:35 avg_step=4.17s recent_step=4.09s eta=51:40:20 +2026-09-20 12:41:53 | INFO | train rank=0 | step=5400/50000 progress=10.8% loss=0.441205 lr=9.908311e-05 lr_backbone=9.908311e-05 elapsed=06:15:17 avg_step=4.17s recent_step=4.14s eta=51:39:36 +2026-09-20 12:42:35 | INFO | train rank=0 | step=5410/50000 progress=10.8% loss=0.483475 lr=9.907680e-05 lr_backbone=9.907680e-05 elapsed=06:15:58 avg_step=4.17s recent_step=4.12s eta=51:38:50 +2026-09-20 12:43:16 | INFO | train rank=0 | step=5420/50000 progress=10.8% loss=0.442575 lr=9.907046e-05 lr_backbone=9.907046e-05 elapsed=06:16:39 avg_step=4.17s recent_step=4.15s eta=51:38:06 +2026-09-20 12:43:57 | INFO | train rank=0 | step=5430/50000 progress=10.9% loss=0.365942 lr=9.906410e-05 lr_backbone=9.906410e-05 elapsed=06:17:21 avg_step=4.17s recent_step=4.14s eta=51:37:22 +2026-09-20 12:44:39 | INFO | train rank=0 | step=5440/50000 progress=10.9% loss=0.436240 lr=9.905772e-05 lr_backbone=9.905772e-05 elapsed=06:18:02 avg_step=4.17s recent_step=4.10s eta=51:36:35 +2026-09-20 12:45:20 | INFO | train rank=0 | step=5450/50000 progress=10.9% loss=0.410000 lr=9.905132e-05 lr_backbone=9.905132e-05 elapsed=06:18:43 avg_step=4.17s recent_step=4.10s eta=51:35:47 +2026-09-20 12:46:01 | INFO | train rank=0 | step=5460/50000 progress=10.9% loss=0.396141 lr=9.904490e-05 lr_backbone=9.904490e-05 elapsed=06:19:24 avg_step=4.17s recent_step=4.13s eta=51:35:02 +2026-09-20 12:46:42 | INFO | train rank=0 | step=5470/50000 progress=10.9% loss=0.461289 lr=9.903846e-05 lr_backbone=9.903846e-05 elapsed=06:20:06 avg_step=4.17s recent_step=4.14s eta=51:34:18 +2026-09-20 12:47:23 | INFO | train rank=0 | step=5480/50000 progress=11.0% loss=0.387257 lr=9.903199e-05 lr_backbone=9.903199e-05 elapsed=06:20:47 avg_step=4.17s recent_step=4.12s eta=51:33:32 +2026-09-20 12:48:05 | INFO | train rank=0 | step=5490/50000 progress=11.0% loss=0.414388 lr=9.902551e-05 lr_backbone=9.902551e-05 elapsed=06:21:28 avg_step=4.17s recent_step=4.13s eta=51:32:47 +2026-09-20 12:48:46 | INFO | train rank=0 | step=5500/50000 progress=11.0% loss=0.368542 lr=9.901900e-05 lr_backbone=9.901900e-05 elapsed=06:22:09 avg_step=4.17s recent_step=4.10s eta=51:32:00 +2026-09-20 12:49:27 | INFO | train rank=0 | step=5510/50000 progress=11.0% loss=0.407827 lr=9.901247e-05 lr_backbone=9.901247e-05 elapsed=06:22:50 avg_step=4.17s recent_step=4.09s eta=51:31:12 +2026-09-20 12:50:08 | INFO | train rank=0 | step=5520/50000 progress=11.0% loss=0.380195 lr=9.900592e-05 lr_backbone=9.900592e-05 elapsed=06:23:31 avg_step=4.17s recent_step=4.13s eta=51:30:27 +2026-09-20 12:50:49 | INFO | train rank=0 | step=5530/50000 progress=11.1% loss=0.450605 lr=9.899935e-05 lr_backbone=9.899935e-05 elapsed=06:24:12 avg_step=4.17s recent_step=4.13s eta=51:29:42 +2026-09-20 12:51:30 | INFO | train rank=0 | step=5540/50000 progress=11.1% loss=0.463035 lr=9.899275e-05 lr_backbone=9.899275e-05 elapsed=06:24:54 avg_step=4.17s recent_step=4.13s eta=51:28:57 +2026-09-20 12:52:12 | INFO | train rank=0 | step=5550/50000 progress=11.1% loss=0.395048 lr=9.898614e-05 lr_backbone=9.898614e-05 elapsed=06:25:35 avg_step=4.17s recent_step=4.13s eta=51:28:12 +2026-09-20 12:52:53 | INFO | train rank=0 | step=5560/50000 progress=11.1% loss=0.386715 lr=9.897950e-05 lr_backbone=9.897950e-05 elapsed=06:26:16 avg_step=4.17s recent_step=4.11s eta=51:27:26 +2026-09-20 12:53:34 | INFO | train rank=0 | step=5570/50000 progress=11.1% loss=0.395174 lr=9.897284e-05 lr_backbone=9.897284e-05 elapsed=06:26:57 avg_step=4.17s recent_step=4.08s eta=51:26:38 +2026-09-20 12:54:15 | INFO | train rank=0 | step=5580/50000 progress=11.2% loss=0.394991 lr=9.896616e-05 lr_backbone=9.896616e-05 elapsed=06:27:38 avg_step=4.17s recent_step=4.12s eta=51:25:52 +2026-09-20 12:54:56 | INFO | train rank=0 | step=5590/50000 progress=11.2% loss=0.419642 lr=9.895946e-05 lr_backbone=9.895946e-05 elapsed=06:28:20 avg_step=4.17s recent_step=4.15s eta=51:25:09 +2026-09-20 12:55:38 | INFO | train rank=0 | step=5600/50000 progress=11.2% loss=0.343848 lr=9.895274e-05 lr_backbone=9.895274e-05 elapsed=06:29:01 avg_step=4.17s recent_step=4.12s eta=51:24:24 +2026-09-20 12:56:19 | INFO | train rank=0 | step=5610/50000 progress=11.2% loss=0.332004 lr=9.894600e-05 lr_backbone=9.894600e-05 elapsed=06:29:42 avg_step=4.17s recent_step=4.14s eta=51:23:39 +2026-09-20 12:57:00 | INFO | train rank=0 | step=5620/50000 progress=11.2% loss=0.465219 lr=9.893923e-05 lr_backbone=9.893923e-05 elapsed=06:30:23 avg_step=4.17s recent_step=4.10s eta=51:22:52 +2026-09-20 12:57:41 | INFO | train rank=0 | step=5630/50000 progress=11.3% loss=0.441800 lr=9.893245e-05 lr_backbone=9.893245e-05 elapsed=06:31:04 avg_step=4.17s recent_step=4.10s eta=51:22:05 +2026-09-20 12:58:22 | INFO | train rank=0 | step=5640/50000 progress=11.3% loss=0.365277 lr=9.892564e-05 lr_backbone=9.892564e-05 elapsed=06:31:46 avg_step=4.17s recent_step=4.13s eta=51:21:21 +2026-09-20 12:59:04 | INFO | train rank=0 | step=5650/50000 progress=11.3% loss=0.380103 lr=9.891881e-05 lr_backbone=9.891881e-05 elapsed=06:32:27 avg_step=4.17s recent_step=4.13s eta=51:20:36 +2026-09-20 12:59:45 | INFO | train rank=0 | step=5660/50000 progress=11.3% loss=0.452623 lr=9.891196e-05 lr_backbone=9.891196e-05 elapsed=06:33:08 avg_step=4.17s recent_step=4.13s eta=51:19:52 +2026-09-20 13:00:26 | INFO | train rank=0 | step=5670/50000 progress=11.3% loss=0.385694 lr=9.890509e-05 lr_backbone=9.890509e-05 elapsed=06:33:49 avg_step=4.17s recent_step=4.12s eta=51:19:06 +2026-09-20 13:01:07 | INFO | train rank=0 | step=5680/50000 progress=11.4% loss=0.368727 lr=9.889819e-05 lr_backbone=9.889819e-05 elapsed=06:34:30 avg_step=4.17s recent_step=4.10s eta=51:18:20 +2026-09-20 13:01:48 | INFO | train rank=0 | step=5690/50000 progress=11.4% loss=0.386050 lr=9.889128e-05 lr_backbone=9.889128e-05 elapsed=06:35:12 avg_step=4.17s recent_step=4.11s eta=51:17:33 +2026-09-20 13:02:29 | INFO | train rank=0 | step=5700/50000 progress=11.4% loss=0.382911 lr=9.888434e-05 lr_backbone=9.888434e-05 elapsed=06:35:53 avg_step=4.17s recent_step=4.13s eta=51:16:48 +2026-09-20 13:03:11 | INFO | train rank=0 | step=5710/50000 progress=11.4% loss=0.331975 lr=9.887739e-05 lr_backbone=9.887739e-05 elapsed=06:36:34 avg_step=4.17s recent_step=4.13s eta=51:16:03 +2026-09-20 13:03:52 | INFO | train rank=0 | step=5720/50000 progress=11.4% loss=0.405617 lr=9.887041e-05 lr_backbone=9.887041e-05 elapsed=06:37:15 avg_step=4.17s recent_step=4.13s eta=51:15:19 +2026-09-20 13:04:33 | INFO | train rank=0 | step=5730/50000 progress=11.5% loss=0.326148 lr=9.886341e-05 lr_backbone=9.886341e-05 elapsed=06:37:57 avg_step=4.17s recent_step=4.12s eta=51:14:34 +2026-09-20 13:05:14 | INFO | train rank=0 | step=5740/50000 progress=11.5% loss=0.369449 lr=9.885638e-05 lr_backbone=9.885638e-05 elapsed=06:38:38 avg_step=4.17s recent_step=4.11s eta=51:13:47 +2026-09-20 13:05:55 | INFO | train rank=0 | step=5750/50000 progress=11.5% loss=0.395047 lr=9.884934e-05 lr_backbone=9.884934e-05 elapsed=06:39:18 avg_step=4.17s recent_step=4.07s eta=51:12:58 +2026-09-20 13:06:36 | INFO | train rank=0 | step=5760/50000 progress=11.5% loss=0.366945 lr=9.884228e-05 lr_backbone=9.884228e-05 elapsed=06:39:59 avg_step=4.17s recent_step=4.11s eta=51:12:12 +2026-09-20 13:07:17 | INFO | train rank=0 | step=5770/50000 progress=11.5% loss=0.382735 lr=9.883519e-05 lr_backbone=9.883519e-05 elapsed=06:40:40 avg_step=4.17s recent_step=4.10s eta=51:11:25 +2026-09-20 13:07:58 | INFO | train rank=0 | step=5780/50000 progress=11.6% loss=0.427332 lr=9.882808e-05 lr_backbone=9.882808e-05 elapsed=06:41:22 avg_step=4.17s recent_step=4.11s eta=51:10:40 +2026-09-20 13:08:39 | INFO | train rank=0 | step=5790/50000 progress=11.6% loss=0.413377 lr=9.882096e-05 lr_backbone=9.882096e-05 elapsed=06:42:03 avg_step=4.17s recent_step=4.11s eta=51:09:53 +2026-09-20 13:09:20 | INFO | train rank=0 | step=5800/50000 progress=11.6% loss=0.340739 lr=9.881381e-05 lr_backbone=9.881381e-05 elapsed=06:42:44 avg_step=4.17s recent_step=4.11s eta=51:09:08 +2026-09-20 13:10:01 | INFO | train rank=0 | step=5810/50000 progress=11.6% loss=0.441077 lr=9.880663e-05 lr_backbone=9.880663e-05 elapsed=06:43:24 avg_step=4.17s recent_step=4.07s eta=51:08:18 +2026-09-20 13:10:42 | INFO | train rank=0 | step=5820/50000 progress=11.6% loss=0.340630 lr=9.879944e-05 lr_backbone=9.879944e-05 elapsed=06:44:05 avg_step=4.17s recent_step=4.10s eta=51:07:31 +2026-09-20 13:11:23 | INFO | train rank=0 | step=5830/50000 progress=11.7% loss=0.348893 lr=9.879223e-05 lr_backbone=9.879223e-05 elapsed=06:44:46 avg_step=4.17s recent_step=4.10s eta=51:06:45 +2026-09-20 13:12:04 | INFO | train rank=0 | step=5840/50000 progress=11.7% loss=0.356824 lr=9.878499e-05 lr_backbone=9.878499e-05 elapsed=06:45:27 avg_step=4.17s recent_step=4.10s eta=51:05:58 +2026-09-20 13:12:45 | INFO | train rank=0 | step=5850/50000 progress=11.7% loss=0.355869 lr=9.877774e-05 lr_backbone=9.877774e-05 elapsed=06:46:08 avg_step=4.17s recent_step=4.10s eta=51:05:12 +2026-09-20 13:13:26 | INFO | train rank=0 | step=5860/50000 progress=11.7% loss=0.360892 lr=9.877046e-05 lr_backbone=9.877046e-05 elapsed=06:46:50 avg_step=4.17s recent_step=4.12s eta=51:04:26 +2026-09-20 13:14:07 | INFO | train rank=0 | step=5870/50000 progress=11.7% loss=0.405704 lr=9.876316e-05 lr_backbone=9.876316e-05 elapsed=06:47:30 avg_step=4.17s recent_step=4.06s eta=51:03:37 +2026-09-20 13:14:48 | INFO | train rank=0 | step=5880/50000 progress=11.8% loss=0.320516 lr=9.875584e-05 lr_backbone=9.875584e-05 elapsed=06:48:11 avg_step=4.17s recent_step=4.12s eta=51:02:52 +2026-09-20 13:15:29 | INFO | train rank=0 | step=5890/50000 progress=11.8% loss=0.439257 lr=9.874850e-05 lr_backbone=9.874850e-05 elapsed=06:48:53 avg_step=4.17s recent_step=4.12s eta=51:02:07 +2026-09-20 13:16:10 | INFO | train rank=0 | step=5900/50000 progress=11.8% loss=0.440015 lr=9.874113e-05 lr_backbone=9.874113e-05 elapsed=06:49:33 avg_step=4.17s recent_step=4.10s eta=51:01:20 +2026-09-20 13:16:52 | INFO | train rank=0 | step=5910/50000 progress=11.8% loss=0.438840 lr=9.873375e-05 lr_backbone=9.873375e-05 elapsed=06:50:15 avg_step=4.17s recent_step=4.14s eta=51:00:36 +2026-09-20 13:17:33 | INFO | train rank=0 | step=5920/50000 progress=11.8% loss=0.381573 lr=9.872634e-05 lr_backbone=9.872634e-05 elapsed=06:50:56 avg_step=4.16s recent_step=4.13s eta=50:59:52 +2026-09-20 13:18:14 | INFO | train rank=0 | step=5930/50000 progress=11.9% loss=0.417394 lr=9.871892e-05 lr_backbone=9.871892e-05 elapsed=06:51:37 avg_step=4.16s recent_step=4.08s eta=50:59:04 +2026-09-20 13:18:55 | INFO | train rank=0 | step=5940/50000 progress=11.9% loss=0.347342 lr=9.871147e-05 lr_backbone=9.871147e-05 elapsed=06:52:18 avg_step=4.16s recent_step=4.12s eta=50:58:18 +2026-09-20 13:19:36 | INFO | train rank=0 | step=5950/50000 progress=11.9% loss=0.393535 lr=9.870400e-05 lr_backbone=9.870400e-05 elapsed=06:52:59 avg_step=4.16s recent_step=4.13s eta=50:57:34 +2026-09-20 13:20:17 | INFO | train rank=0 | step=5960/50000 progress=11.9% loss=0.397889 lr=9.869651e-05 lr_backbone=9.869651e-05 elapsed=06:53:41 avg_step=4.16s recent_step=4.14s eta=50:56:50 +2026-09-20 13:20:59 | INFO | train rank=0 | step=5970/50000 progress=11.9% loss=0.502218 lr=9.868899e-05 lr_backbone=9.868899e-05 elapsed=06:54:22 avg_step=4.16s recent_step=4.13s eta=50:56:06 +2026-09-20 13:21:40 | INFO | train rank=0 | step=5980/50000 progress=12.0% loss=0.383580 lr=9.868146e-05 lr_backbone=9.868146e-05 elapsed=06:55:03 avg_step=4.16s recent_step=4.13s eta=50:55:22 +2026-09-20 13:22:21 | INFO | train rank=0 | step=5990/50000 progress=12.0% loss=0.447620 lr=9.867391e-05 lr_backbone=9.867391e-05 elapsed=06:55:44 avg_step=4.16s recent_step=4.10s eta=50:54:35 +2026-09-20 13:23:02 | INFO | train rank=0 | step=6000/50000 progress=12.0% loss=0.435170 lr=9.866633e-05 lr_backbone=9.866633e-05 elapsed=06:56:25 avg_step=4.16s recent_step=4.12s eta=50:53:50 +2026-09-20 13:23:43 | INFO | train rank=0 | step=6010/50000 progress=12.0% loss=0.369991 lr=9.865873e-05 lr_backbone=9.865873e-05 elapsed=06:57:07 avg_step=4.16s recent_step=4.12s eta=50:53:05 +2026-09-20 13:24:24 | INFO | train rank=0 | step=6020/50000 progress=12.0% loss=0.352679 lr=9.865111e-05 lr_backbone=9.865111e-05 elapsed=06:57:48 avg_step=4.16s recent_step=4.11s eta=50:52:19 +2026-09-20 13:25:06 | INFO | train rank=0 | step=6030/50000 progress=12.1% loss=0.373169 lr=9.864347e-05 lr_backbone=9.864347e-05 elapsed=06:58:29 avg_step=4.16s recent_step=4.12s eta=50:51:34 +2026-09-20 13:25:47 | INFO | train rank=0 | step=6040/50000 progress=12.1% loss=0.353889 lr=9.863581e-05 lr_backbone=9.863581e-05 elapsed=06:59:10 avg_step=4.16s recent_step=4.13s eta=50:50:50 +2026-09-20 13:26:28 | INFO | train rank=0 | step=6050/50000 progress=12.1% loss=0.358181 lr=9.862813e-05 lr_backbone=9.862813e-05 elapsed=06:59:51 avg_step=4.16s recent_step=4.09s eta=50:50:03 +2026-09-20 13:27:09 | INFO | train rank=0 | step=6060/50000 progress=12.1% loss=0.363923 lr=9.862043e-05 lr_backbone=9.862043e-05 elapsed=07:00:32 avg_step=4.16s recent_step=4.08s eta=50:49:15 +2026-09-20 13:27:50 | INFO | train rank=0 | step=6070/50000 progress=12.1% loss=0.352084 lr=9.861270e-05 lr_backbone=9.861270e-05 elapsed=07:01:13 avg_step=4.16s recent_step=4.12s eta=50:48:31 +2026-09-20 13:28:31 | INFO | train rank=0 | step=6080/50000 progress=12.2% loss=0.352702 lr=9.860495e-05 lr_backbone=9.860495e-05 elapsed=07:01:54 avg_step=4.16s recent_step=4.09s eta=50:47:44 +2026-09-20 13:29:12 | INFO | train rank=0 | step=6090/50000 progress=12.2% loss=0.400993 lr=9.859719e-05 lr_backbone=9.859719e-05 elapsed=07:02:35 avg_step=4.16s recent_step=4.11s eta=50:46:59 +2026-09-20 13:29:53 | INFO | train rank=0 | step=6100/50000 progress=12.2% loss=0.348258 lr=9.858940e-05 lr_backbone=9.858940e-05 elapsed=07:03:16 avg_step=4.16s recent_step=4.12s eta=50:46:14 +2026-09-20 13:30:34 | INFO | train rank=0 | step=6110/50000 progress=12.2% loss=0.314565 lr=9.858159e-05 lr_backbone=9.858159e-05 elapsed=07:03:57 avg_step=4.16s recent_step=4.08s eta=50:45:27 +2026-09-20 13:31:15 | INFO | train rank=0 | step=6120/50000 progress=12.2% loss=0.438915 lr=9.857376e-05 lr_backbone=9.857376e-05 elapsed=07:04:38 avg_step=4.16s recent_step=4.10s eta=50:44:40 +2026-09-20 13:31:56 | INFO | train rank=0 | step=6130/50000 progress=12.3% loss=0.347763 lr=9.856590e-05 lr_backbone=9.856590e-05 elapsed=07:05:19 avg_step=4.16s recent_step=4.13s eta=50:43:56 +2026-09-20 13:32:37 | INFO | train rank=0 | step=6140/50000 progress=12.3% loss=0.331938 lr=9.855803e-05 lr_backbone=9.855803e-05 elapsed=07:06:01 avg_step=4.16s recent_step=4.12s eta=50:43:11 +2026-09-20 13:33:19 | INFO | train rank=0 | step=6150/50000 progress=12.3% loss=0.397457 lr=9.855013e-05 lr_backbone=9.855013e-05 elapsed=07:06:42 avg_step=4.16s recent_step=4.13s eta=50:42:27 +2026-09-20 13:34:00 | INFO | train rank=0 | step=6160/50000 progress=12.3% loss=0.440429 lr=9.854222e-05 lr_backbone=9.854222e-05 elapsed=07:07:23 avg_step=4.16s recent_step=4.14s eta=50:41:44 +2026-09-20 13:34:41 | INFO | train rank=0 | step=6170/50000 progress=12.3% loss=0.405049 lr=9.853428e-05 lr_backbone=9.853428e-05 elapsed=07:08:04 avg_step=4.16s recent_step=4.08s eta=50:40:56 +2026-09-20 13:35:22 | INFO | train rank=0 | step=6180/50000 progress=12.4% loss=0.461525 lr=9.852632e-05 lr_backbone=9.852632e-05 elapsed=07:08:46 avg_step=4.16s recent_step=4.15s eta=50:40:14 +2026-09-20 13:36:03 | INFO | train rank=0 | step=6190/50000 progress=12.4% loss=0.362228 lr=9.851834e-05 lr_backbone=9.851834e-05 elapsed=07:09:27 avg_step=4.16s recent_step=4.08s eta=50:39:27 +2026-09-20 13:36:44 | INFO | train rank=0 | step=6200/50000 progress=12.4% loss=0.381975 lr=9.851034e-05 lr_backbone=9.851034e-05 elapsed=07:10:08 avg_step=4.16s recent_step=4.10s eta=50:38:41 +2026-09-20 13:37:25 | INFO | train rank=0 | step=6210/50000 progress=12.4% loss=0.353705 lr=9.850232e-05 lr_backbone=9.850232e-05 elapsed=07:10:49 avg_step=4.16s recent_step=4.12s eta=50:37:56 +2026-09-20 13:38:06 | INFO | train rank=0 | step=6220/50000 progress=12.4% loss=0.373586 lr=9.849427e-05 lr_backbone=9.849427e-05 elapsed=07:11:30 avg_step=4.16s recent_step=4.11s eta=50:37:11 +2026-09-20 13:38:47 | INFO | train rank=0 | step=6230/50000 progress=12.5% loss=0.417165 lr=9.848621e-05 lr_backbone=9.848621e-05 elapsed=07:12:11 avg_step=4.16s recent_step=4.08s eta=50:36:24 +2026-09-20 13:39:28 | INFO | train rank=0 | step=6240/50000 progress=12.5% loss=0.371787 lr=9.847812e-05 lr_backbone=9.847812e-05 elapsed=07:12:52 avg_step=4.16s recent_step=4.11s eta=50:35:38 +2026-09-20 13:40:09 | INFO | train rank=0 | step=6250/50000 progress=12.5% loss=0.406031 lr=9.847001e-05 lr_backbone=9.847001e-05 elapsed=07:13:33 avg_step=4.16s recent_step=4.09s eta=50:34:52 +2026-09-20 13:40:51 | INFO | train rank=0 | step=6260/50000 progress=12.5% loss=0.332532 lr=9.846188e-05 lr_backbone=9.846188e-05 elapsed=07:14:14 avg_step=4.16s recent_step=4.12s eta=50:34:07 +2026-09-20 13:41:32 | INFO | train rank=0 | step=6270/50000 progress=12.5% loss=0.397738 lr=9.845373e-05 lr_backbone=9.845373e-05 elapsed=07:14:55 avg_step=4.16s recent_step=4.13s eta=50:33:23 +2026-09-20 13:42:13 | INFO | train rank=0 | step=6280/50000 progress=12.6% loss=0.428736 lr=9.844556e-05 lr_backbone=9.844556e-05 elapsed=07:15:36 avg_step=4.16s recent_step=4.11s eta=50:32:38 +2026-09-20 13:42:54 | INFO | train rank=0 | step=6290/50000 progress=12.6% loss=0.370152 lr=9.843737e-05 lr_backbone=9.843737e-05 elapsed=07:16:17 avg_step=4.16s recent_step=4.09s eta=50:31:51 +2026-09-20 13:43:35 | INFO | train rank=0 | step=6300/50000 progress=12.6% loss=0.449212 lr=9.842916e-05 lr_backbone=9.842916e-05 elapsed=07:16:58 avg_step=4.16s recent_step=4.12s eta=50:31:07 +2026-09-20 13:44:16 | INFO | train rank=0 | step=6310/50000 progress=12.6% loss=0.363916 lr=9.842092e-05 lr_backbone=9.842092e-05 elapsed=07:17:39 avg_step=4.16s recent_step=4.10s eta=50:30:21 +2026-09-20 13:44:57 | INFO | train rank=0 | step=6320/50000 progress=12.6% loss=0.368630 lr=9.841267e-05 lr_backbone=9.841267e-05 elapsed=07:18:21 avg_step=4.16s recent_step=4.13s eta=50:29:37 +2026-09-20 13:45:39 | INFO | train rank=0 | step=6330/50000 progress=12.7% loss=0.496679 lr=9.840439e-05 lr_backbone=9.840439e-05 elapsed=07:19:02 avg_step=4.16s recent_step=4.15s eta=50:28:54 +2026-09-20 13:46:20 | INFO | train rank=0 | step=6340/50000 progress=12.7% loss=0.456671 lr=9.839609e-05 lr_backbone=9.839609e-05 elapsed=07:19:43 avg_step=4.16s recent_step=4.10s eta=50:28:08 +2026-09-20 13:47:00 | INFO | train rank=0 | step=6350/50000 progress=12.7% loss=0.352290 lr=9.838777e-05 lr_backbone=9.838777e-05 elapsed=07:20:24 avg_step=4.16s recent_step=4.07s eta=50:27:21 +2026-09-20 13:47:42 | INFO | train rank=0 | step=6360/50000 progress=12.7% loss=0.436477 lr=9.837943e-05 lr_backbone=9.837943e-05 elapsed=07:21:05 avg_step=4.16s recent_step=4.13s eta=50:26:37 +2026-09-20 13:48:23 | INFO | train rank=0 | step=6370/50000 progress=12.7% loss=0.387378 lr=9.837107e-05 lr_backbone=9.837107e-05 elapsed=07:21:46 avg_step=4.16s recent_step=4.10s eta=50:25:51 +2026-09-20 13:49:04 | INFO | train rank=0 | step=6380/50000 progress=12.8% loss=0.472277 lr=9.836269e-05 lr_backbone=9.836269e-05 elapsed=07:22:27 avg_step=4.16s recent_step=4.08s eta=50:25:04 +2026-09-20 13:49:45 | INFO | train rank=0 | step=6390/50000 progress=12.8% loss=0.371190 lr=9.835428e-05 lr_backbone=9.835428e-05 elapsed=07:23:08 avg_step=4.16s recent_step=4.11s eta=50:24:19 +2026-09-20 13:50:26 | INFO | train rank=0 | step=6400/50000 progress=12.8% loss=0.369181 lr=9.834586e-05 lr_backbone=9.834586e-05 elapsed=07:23:49 avg_step=4.16s recent_step=4.10s eta=50:23:34 +2026-09-20 13:51:07 | INFO | train rank=0 | step=6410/50000 progress=12.8% loss=0.354022 lr=9.833741e-05 lr_backbone=9.833741e-05 elapsed=07:24:30 avg_step=4.16s recent_step=4.09s eta=50:22:47 +2026-09-20 13:51:48 | INFO | train rank=0 | step=6420/50000 progress=12.8% loss=0.336918 lr=9.832895e-05 lr_backbone=9.832895e-05 elapsed=07:25:11 avg_step=4.16s recent_step=4.11s eta=50:22:02 +2026-09-20 13:52:29 | INFO | train rank=0 | step=6430/50000 progress=12.9% loss=0.486314 lr=9.832046e-05 lr_backbone=9.832046e-05 elapsed=07:25:52 avg_step=4.16s recent_step=4.11s eta=50:21:17 +2026-09-20 13:53:10 | INFO | train rank=0 | step=6440/50000 progress=12.9% loss=0.439909 lr=9.831195e-05 lr_backbone=9.831195e-05 elapsed=07:26:33 avg_step=4.16s recent_step=4.08s eta=50:20:30 +2026-09-20 13:53:51 | INFO | train rank=0 | step=6450/50000 progress=12.9% loss=0.371619 lr=9.830342e-05 lr_backbone=9.830342e-05 elapsed=07:27:14 avg_step=4.16s recent_step=4.12s eta=50:19:45 +2026-09-20 13:54:32 | INFO | train rank=0 | step=6460/50000 progress=12.9% loss=0.396966 lr=9.829486e-05 lr_backbone=9.829486e-05 elapsed=07:27:55 avg_step=4.16s recent_step=4.11s eta=50:19:01 +2026-09-20 13:55:13 | INFO | train rank=0 | step=6470/50000 progress=12.9% loss=0.431257 lr=9.828629e-05 lr_backbone=9.828629e-05 elapsed=07:28:36 avg_step=4.16s recent_step=4.08s eta=50:18:14 +2026-09-20 13:55:54 | INFO | train rank=0 | step=6480/50000 progress=13.0% loss=0.381492 lr=9.827770e-05 lr_backbone=9.827770e-05 elapsed=07:29:17 avg_step=4.16s recent_step=4.11s eta=50:17:29 +2026-09-20 13:56:35 | INFO | train rank=0 | step=6490/50000 progress=13.0% loss=0.396277 lr=9.826908e-05 lr_backbone=9.826908e-05 elapsed=07:29:58 avg_step=4.16s recent_step=4.12s eta=50:16:45 +2026-09-20 13:57:16 | INFO | train rank=0 | step=6500/50000 progress=13.0% loss=0.342415 lr=9.826045e-05 lr_backbone=9.826045e-05 elapsed=07:30:39 avg_step=4.16s recent_step=4.09s eta=50:15:58 +2026-09-20 13:57:57 | INFO | train rank=0 | step=6510/50000 progress=13.0% loss=0.483778 lr=9.825179e-05 lr_backbone=9.825179e-05 elapsed=07:31:21 avg_step=4.16s recent_step=4.14s eta=50:15:15 +2026-09-20 13:58:39 | INFO | train rank=0 | step=6520/50000 progress=13.0% loss=0.349041 lr=9.824311e-05 lr_backbone=9.824311e-05 elapsed=07:32:02 avg_step=4.16s recent_step=4.12s eta=50:14:31 +2026-09-20 13:59:19 | INFO | train rank=0 | step=6530/50000 progress=13.1% loss=0.334153 lr=9.823441e-05 lr_backbone=9.823441e-05 elapsed=07:32:43 avg_step=4.16s recent_step=4.08s eta=50:13:44 +2026-09-20 14:00:00 | INFO | train rank=0 | step=6540/50000 progress=13.1% loss=0.332093 lr=9.822569e-05 lr_backbone=9.822569e-05 elapsed=07:33:24 avg_step=4.16s recent_step=4.12s eta=50:12:59 +2026-09-20 14:00:42 | INFO | train rank=0 | step=6550/50000 progress=13.1% loss=0.396308 lr=9.821695e-05 lr_backbone=9.821695e-05 elapsed=07:34:05 avg_step=4.16s recent_step=4.11s eta=50:12:15 +2026-09-20 14:01:22 | INFO | train rank=0 | step=6560/50000 progress=13.1% loss=0.389071 lr=9.820818e-05 lr_backbone=9.820818e-05 elapsed=07:34:46 avg_step=4.16s recent_step=4.08s eta=50:11:28 +2026-09-20 14:02:03 | INFO | train rank=0 | step=6570/50000 progress=13.1% loss=0.349706 lr=9.819940e-05 lr_backbone=9.819940e-05 elapsed=07:35:27 avg_step=4.16s recent_step=4.10s eta=50:10:43 +2026-09-20 14:02:45 | INFO | train rank=0 | step=6580/50000 progress=13.2% loss=0.390097 lr=9.819059e-05 lr_backbone=9.819059e-05 elapsed=07:36:08 avg_step=4.16s recent_step=4.10s eta=50:09:57 +2026-09-20 14:03:25 | INFO | train rank=0 | step=6590/50000 progress=13.2% loss=0.388149 lr=9.818177e-05 lr_backbone=9.818177e-05 elapsed=07:36:49 avg_step=4.16s recent_step=4.08s eta=50:09:10 +2026-09-20 14:04:06 | INFO | train rank=0 | step=6600/50000 progress=13.2% loss=0.370415 lr=9.817292e-05 lr_backbone=9.817292e-05 elapsed=07:37:30 avg_step=4.16s recent_step=4.11s eta=50:08:25 +2026-09-20 14:04:48 | INFO | train rank=0 | step=6610/50000 progress=13.2% loss=0.393037 lr=9.816405e-05 lr_backbone=9.816405e-05 elapsed=07:38:11 avg_step=4.16s recent_step=4.11s eta=50:07:41 +2026-09-20 14:05:28 | INFO | train rank=0 | step=6620/50000 progress=13.2% loss=0.384593 lr=9.815516e-05 lr_backbone=9.815516e-05 elapsed=07:38:52 avg_step=4.16s recent_step=4.10s eta=50:06:55 +2026-09-20 14:06:10 | INFO | train rank=0 | step=6630/50000 progress=13.3% loss=0.376173 lr=9.814625e-05 lr_backbone=9.814625e-05 elapsed=07:39:33 avg_step=4.16s recent_step=4.13s eta=50:06:12 +2026-09-20 14:06:51 | INFO | train rank=0 | step=6640/50000 progress=13.3% loss=0.353129 lr=9.813732e-05 lr_backbone=9.813732e-05 elapsed=07:40:14 avg_step=4.16s recent_step=4.11s eta=50:05:27 +2026-09-20 14:07:32 | INFO | train rank=0 | step=6650/50000 progress=13.3% loss=0.405347 lr=9.812837e-05 lr_backbone=9.812837e-05 elapsed=07:40:55 avg_step=4.16s recent_step=4.09s eta=50:04:41 +2026-09-20 14:08:13 | INFO | train rank=0 | step=6660/50000 progress=13.3% loss=0.414114 lr=9.811939e-05 lr_backbone=9.811939e-05 elapsed=07:41:36 avg_step=4.16s recent_step=4.13s eta=50:03:57 +2026-09-20 14:08:54 | INFO | train rank=0 | step=6670/50000 progress=13.3% loss=0.358433 lr=9.811040e-05 lr_backbone=9.811040e-05 elapsed=07:42:17 avg_step=4.16s recent_step=4.11s eta=50:03:12 +2026-09-20 14:09:35 | INFO | train rank=0 | step=6680/50000 progress=13.4% loss=0.344887 lr=9.810138e-05 lr_backbone=9.810138e-05 elapsed=07:42:58 avg_step=4.16s recent_step=4.09s eta=50:02:26 +2026-09-20 14:10:16 | INFO | train rank=0 | step=6690/50000 progress=13.4% loss=0.381406 lr=9.809235e-05 lr_backbone=9.809235e-05 elapsed=07:43:40 avg_step=4.16s recent_step=4.12s eta=50:01:43 +2026-09-20 14:10:57 | INFO | train rank=0 | step=6700/50000 progress=13.4% loss=0.427120 lr=9.808329e-05 lr_backbone=9.808329e-05 elapsed=07:44:21 avg_step=4.16s recent_step=4.11s eta=50:00:58 +2026-09-20 14:11:38 | INFO | train rank=0 | step=6710/50000 progress=13.4% loss=0.391364 lr=9.807421e-05 lr_backbone=9.807421e-05 elapsed=07:45:02 avg_step=4.16s recent_step=4.08s eta=50:00:11 +2026-09-20 14:12:20 | INFO | train rank=0 | step=6720/50000 progress=13.4% loss=0.369920 lr=9.806511e-05 lr_backbone=9.806511e-05 elapsed=07:45:43 avg_step=4.16s recent_step=4.13s eta=49:59:28 +2026-09-20 14:13:01 | INFO | train rank=0 | step=6730/50000 progress=13.5% loss=0.377943 lr=9.805599e-05 lr_backbone=9.805599e-05 elapsed=07:46:24 avg_step=4.16s recent_step=4.12s eta=49:58:44 +2026-09-20 14:13:42 | INFO | train rank=0 | step=6740/50000 progress=13.5% loss=0.414187 lr=9.804685e-05 lr_backbone=9.804685e-05 elapsed=07:47:05 avg_step=4.16s recent_step=4.10s eta=49:57:59 +2026-09-20 14:14:23 | INFO | train rank=0 | step=6750/50000 progress=13.5% loss=0.424655 lr=9.803768e-05 lr_backbone=9.803768e-05 elapsed=07:47:46 avg_step=4.16s recent_step=4.08s eta=49:57:12 +2026-09-20 14:15:04 | INFO | train rank=0 | step=6760/50000 progress=13.5% loss=0.375104 lr=9.802850e-05 lr_backbone=9.802850e-05 elapsed=07:48:27 avg_step=4.16s recent_step=4.11s eta=49:56:28 +2026-09-20 14:15:45 | INFO | train rank=0 | step=6770/50000 progress=13.5% loss=0.398910 lr=9.801929e-05 lr_backbone=9.801929e-05 elapsed=07:49:08 avg_step=4.16s recent_step=4.08s eta=49:55:41 +2026-09-20 14:16:26 | INFO | train rank=0 | step=6780/50000 progress=13.6% loss=0.350064 lr=9.801007e-05 lr_backbone=9.801007e-05 elapsed=07:49:49 avg_step=4.16s recent_step=4.11s eta=49:54:57 +2026-09-20 14:17:07 | INFO | train rank=0 | step=6790/50000 progress=13.6% loss=0.457748 lr=9.800082e-05 lr_backbone=9.800082e-05 elapsed=07:50:30 avg_step=4.16s recent_step=4.13s eta=49:54:14 +2026-09-20 14:17:48 | INFO | train rank=0 | step=6800/50000 progress=13.6% loss=0.354250 lr=9.799155e-05 lr_backbone=9.799155e-05 elapsed=07:51:11 avg_step=4.16s recent_step=4.11s eta=49:53:29 +2026-09-20 14:18:29 | INFO | train rank=0 | step=6810/50000 progress=13.6% loss=0.493434 lr=9.798226e-05 lr_backbone=9.798226e-05 elapsed=07:51:52 avg_step=4.16s recent_step=4.08s eta=49:52:42 +2026-09-20 14:19:10 | INFO | train rank=0 | step=6820/50000 progress=13.6% loss=0.373009 lr=9.797295e-05 lr_backbone=9.797295e-05 elapsed=07:52:33 avg_step=4.16s recent_step=4.13s eta=49:51:59 +2026-09-20 14:19:51 | INFO | train rank=0 | step=6830/50000 progress=13.7% loss=0.418128 lr=9.796362e-05 lr_backbone=9.796362e-05 elapsed=07:53:14 avg_step=4.16s recent_step=4.07s eta=49:51:12 +2026-09-20 14:20:32 | INFO | train rank=0 | step=6840/50000 progress=13.7% loss=0.391931 lr=9.795427e-05 lr_backbone=9.795427e-05 elapsed=07:53:55 avg_step=4.16s recent_step=4.11s eta=49:50:27 +2026-09-20 14:21:13 | INFO | train rank=0 | step=6850/50000 progress=13.7% loss=0.453418 lr=9.794490e-05 lr_backbone=9.794490e-05 elapsed=07:54:36 avg_step=4.16s recent_step=4.12s eta=49:49:44 +2026-09-20 14:21:54 | INFO | train rank=0 | step=6860/50000 progress=13.7% loss=0.341533 lr=9.793550e-05 lr_backbone=9.793550e-05 elapsed=07:55:18 avg_step=4.16s recent_step=4.12s eta=49:49:00 +2026-09-20 14:22:36 | INFO | train rank=0 | step=6870/50000 progress=13.7% loss=0.420874 lr=9.792609e-05 lr_backbone=9.792609e-05 elapsed=07:55:59 avg_step=4.16s recent_step=4.12s eta=49:48:16 +2026-09-20 14:23:16 | INFO | train rank=0 | step=6880/50000 progress=13.8% loss=0.343338 lr=9.791665e-05 lr_backbone=9.791665e-05 elapsed=07:56:40 avg_step=4.16s recent_step=4.07s eta=49:47:29 +2026-09-20 14:23:57 | INFO | train rank=0 | step=6890/50000 progress=13.8% loss=0.388524 lr=9.790720e-05 lr_backbone=9.790720e-05 elapsed=07:57:20 avg_step=4.16s recent_step=4.09s eta=49:46:43 +2026-09-20 14:24:38 | INFO | train rank=0 | step=6900/50000 progress=13.8% loss=0.441900 lr=9.789772e-05 lr_backbone=9.789772e-05 elapsed=07:58:02 avg_step=4.16s recent_step=4.12s eta=49:45:59 +2026-09-20 14:25:19 | INFO | train rank=0 | step=6910/50000 progress=13.8% loss=0.479720 lr=9.788822e-05 lr_backbone=9.788822e-05 elapsed=07:58:42 avg_step=4.16s recent_step=4.08s eta=49:45:13 +2026-09-20 14:26:00 | INFO | train rank=0 | step=6920/50000 progress=13.8% loss=0.424710 lr=9.787870e-05 lr_backbone=9.787870e-05 elapsed=07:59:24 avg_step=4.16s recent_step=4.11s eta=49:44:28 +2026-09-20 14:26:41 | INFO | train rank=0 | step=6930/50000 progress=13.9% loss=0.430033 lr=9.786916e-05 lr_backbone=9.786916e-05 elapsed=08:00:05 avg_step=4.16s recent_step=4.10s eta=49:43:43 +2026-09-20 14:27:22 | INFO | train rank=0 | step=6940/50000 progress=13.9% loss=0.415940 lr=9.785960e-05 lr_backbone=9.785960e-05 elapsed=08:00:46 avg_step=4.16s recent_step=4.09s eta=49:42:58 +2026-09-20 14:28:03 | INFO | train rank=0 | step=6950/50000 progress=13.9% loss=0.422751 lr=9.785002e-05 lr_backbone=9.785002e-05 elapsed=08:01:26 avg_step=4.16s recent_step=4.09s eta=49:42:12 +2026-09-20 14:28:44 | INFO | train rank=0 | step=6960/50000 progress=13.9% loss=0.360067 lr=9.784041e-05 lr_backbone=9.784041e-05 elapsed=08:02:07 avg_step=4.16s recent_step=4.11s eta=49:41:28 +2026-09-20 14:29:25 | INFO | train rank=0 | step=6970/50000 progress=13.9% loss=0.390476 lr=9.783079e-05 lr_backbone=9.783079e-05 elapsed=08:02:49 avg_step=4.16s recent_step=4.11s eta=49:40:43 +2026-09-20 14:30:06 | INFO | train rank=0 | step=6980/50000 progress=14.0% loss=0.357244 lr=9.782114e-05 lr_backbone=9.782114e-05 elapsed=08:03:30 avg_step=4.16s recent_step=4.11s eta=49:39:59 +2026-09-20 14:30:47 | INFO | train rank=0 | step=6990/50000 progress=14.0% loss=0.395662 lr=9.781148e-05 lr_backbone=9.781148e-05 elapsed=08:04:11 avg_step=4.16s recent_step=4.10s eta=49:39:14 +2026-09-20 14:31:28 | INFO | train rank=0 | step=7000/50000 progress=14.0% loss=0.377791 lr=9.780179e-05 lr_backbone=9.780179e-05 elapsed=08:04:52 avg_step=4.16s recent_step=4.09s eta=49:38:28 +2026-09-20 14:32:09 | INFO | train rank=0 | step=7010/50000 progress=14.0% loss=0.354113 lr=9.779208e-05 lr_backbone=9.779208e-05 elapsed=08:05:32 avg_step=4.16s recent_step=4.05s eta=49:37:40 +2026-09-20 14:32:50 | INFO | train rank=0 | step=7020/50000 progress=14.0% loss=0.423821 lr=9.778235e-05 lr_backbone=9.778235e-05 elapsed=08:06:13 avg_step=4.16s recent_step=4.09s eta=49:36:55 +2026-09-20 14:33:31 | INFO | train rank=0 | step=7030/50000 progress=14.1% loss=0.483212 lr=9.777260e-05 lr_backbone=9.777260e-05 elapsed=08:06:54 avg_step=4.16s recent_step=4.09s eta=49:36:09 +2026-09-20 14:34:11 | INFO | train rank=0 | step=7040/50000 progress=14.1% loss=0.356656 lr=9.776283e-05 lr_backbone=9.776283e-05 elapsed=08:07:35 avg_step=4.16s recent_step=4.09s eta=49:35:24 +2026-09-20 14:34:52 | INFO | train rank=0 | step=7050/50000 progress=14.1% loss=0.385684 lr=9.775304e-05 lr_backbone=9.775304e-05 elapsed=08:08:16 avg_step=4.16s recent_step=4.09s eta=49:34:38 +2026-09-20 14:35:33 | INFO | train rank=0 | step=7060/50000 progress=14.1% loss=0.368364 lr=9.774323e-05 lr_backbone=9.774323e-05 elapsed=08:08:57 avg_step=4.16s recent_step=4.10s eta=49:33:53 +2026-09-20 14:36:14 | INFO | train rank=0 | step=7070/50000 progress=14.1% loss=0.400379 lr=9.773339e-05 lr_backbone=9.773339e-05 elapsed=08:09:37 avg_step=4.16s recent_step=4.05s eta=49:33:05 +2026-09-20 14:36:55 | INFO | train rank=0 | step=7080/50000 progress=14.2% loss=0.381264 lr=9.772354e-05 lr_backbone=9.772354e-05 elapsed=08:10:18 avg_step=4.16s recent_step=4.10s eta=49:32:20 +2026-09-20 14:37:36 | INFO | train rank=0 | step=7090/50000 progress=14.2% loss=0.405592 lr=9.771366e-05 lr_backbone=9.771366e-05 elapsed=08:10:59 avg_step=4.16s recent_step=4.11s eta=49:31:36 +2026-09-20 14:38:17 | INFO | train rank=0 | step=7100/50000 progress=14.2% loss=0.432456 lr=9.770377e-05 lr_backbone=9.770377e-05 elapsed=08:11:41 avg_step=4.16s recent_step=4.13s eta=49:30:53 +2026-09-20 14:38:58 | INFO | train rank=0 | step=7110/50000 progress=14.2% loss=0.474404 lr=9.769385e-05 lr_backbone=9.769385e-05 elapsed=08:12:21 avg_step=4.15s recent_step=4.08s eta=49:30:07 +2026-09-20 14:39:39 | INFO | train rank=0 | step=7120/50000 progress=14.2% loss=0.353816 lr=9.768391e-05 lr_backbone=9.768391e-05 elapsed=08:13:03 avg_step=4.15s recent_step=4.12s eta=49:29:23 +2026-09-20 14:40:20 | INFO | train rank=0 | step=7130/50000 progress=14.3% loss=0.451309 lr=9.767395e-05 lr_backbone=9.767395e-05 elapsed=08:13:43 avg_step=4.15s recent_step=4.05s eta=49:28:35 +2026-09-20 14:41:01 | INFO | train rank=0 | step=7140/50000 progress=14.3% loss=0.342035 lr=9.766398e-05 lr_backbone=9.766398e-05 elapsed=08:14:24 avg_step=4.15s recent_step=4.11s eta=49:27:51 +2026-09-20 14:41:42 | INFO | train rank=0 | step=7150/50000 progress=14.3% loss=0.399963 lr=9.765398e-05 lr_backbone=9.765398e-05 elapsed=08:15:05 avg_step=4.15s recent_step=4.10s eta=49:27:07 +2026-09-20 14:42:23 | INFO | train rank=0 | step=7160/50000 progress=14.3% loss=0.382443 lr=9.764395e-05 lr_backbone=9.764395e-05 elapsed=08:15:46 avg_step=4.15s recent_step=4.09s eta=49:26:21 +2026-09-20 14:43:04 | INFO | train rank=0 | step=7170/50000 progress=14.3% loss=0.488614 lr=9.763391e-05 lr_backbone=9.763391e-05 elapsed=08:16:27 avg_step=4.15s recent_step=4.11s eta=49:25:37 +2026-09-20 14:43:45 | INFO | train rank=0 | step=7180/50000 progress=14.4% loss=0.454346 lr=9.762385e-05 lr_backbone=9.762385e-05 elapsed=08:17:08 avg_step=4.15s recent_step=4.10s eta=49:24:52 +2026-09-20 14:44:26 | INFO | train rank=0 | step=7190/50000 progress=14.4% loss=0.363102 lr=9.761377e-05 lr_backbone=9.761377e-05 elapsed=08:17:49 avg_step=4.15s recent_step=4.07s eta=49:24:06 +2026-09-20 14:45:07 | INFO | train rank=0 | step=7200/50000 progress=14.4% loss=0.396483 lr=9.760366e-05 lr_backbone=9.760366e-05 elapsed=08:18:30 avg_step=4.15s recent_step=4.10s eta=49:23:21 +2026-09-20 14:45:48 | INFO | train rank=0 | step=7210/50000 progress=14.4% loss=0.452152 lr=9.759354e-05 lr_backbone=9.759354e-05 elapsed=08:19:11 avg_step=4.15s recent_step=4.12s eta=49:22:37 +2026-09-20 14:46:29 | INFO | train rank=0 | step=7220/50000 progress=14.4% loss=0.362679 lr=9.758339e-05 lr_backbone=9.758339e-05 elapsed=08:19:52 avg_step=4.15s recent_step=4.11s eta=49:21:53 +2026-09-20 14:47:10 | INFO | train rank=0 | step=7230/50000 progress=14.5% loss=0.406008 lr=9.757322e-05 lr_backbone=9.757322e-05 elapsed=08:20:33 avg_step=4.15s recent_step=4.11s eta=49:21:09 +2026-09-20 14:47:51 | INFO | train rank=0 | step=7240/50000 progress=14.5% loss=0.372494 lr=9.756303e-05 lr_backbone=9.756303e-05 elapsed=08:21:15 avg_step=4.15s recent_step=4.12s eta=49:20:25 +2026-09-20 14:48:32 | INFO | train rank=0 | step=7250/50000 progress=14.5% loss=0.414925 lr=9.755283e-05 lr_backbone=9.755283e-05 elapsed=08:21:55 avg_step=4.15s recent_step=4.07s eta=49:19:39 +2026-09-20 14:49:13 | INFO | train rank=0 | step=7260/50000 progress=14.5% loss=0.359505 lr=9.754260e-05 lr_backbone=9.754260e-05 elapsed=08:22:36 avg_step=4.15s recent_step=4.11s eta=49:18:55 +2026-09-20 14:49:54 | INFO | train rank=0 | step=7270/50000 progress=14.5% loss=0.353573 lr=9.753235e-05 lr_backbone=9.753235e-05 elapsed=08:23:17 avg_step=4.15s recent_step=4.10s eta=49:18:10 +2026-09-20 14:50:35 | INFO | train rank=0 | step=7280/50000 progress=14.6% loss=0.395369 lr=9.752208e-05 lr_backbone=9.752208e-05 elapsed=08:23:58 avg_step=4.15s recent_step=4.11s eta=49:17:26 +2026-09-20 14:51:16 | INFO | train rank=0 | step=7290/50000 progress=14.6% loss=0.369111 lr=9.751178e-05 lr_backbone=9.751178e-05 elapsed=08:24:39 avg_step=4.15s recent_step=4.10s eta=49:16:41 +2026-09-20 14:51:57 | INFO | train rank=0 | step=7300/50000 progress=14.6% loss=0.368232 lr=9.750147e-05 lr_backbone=9.750147e-05 elapsed=08:25:21 avg_step=4.15s recent_step=4.11s eta=49:15:57 +2026-09-20 14:52:38 | INFO | train rank=0 | step=7310/50000 progress=14.6% loss=0.393934 lr=9.749114e-05 lr_backbone=9.749114e-05 elapsed=08:26:01 avg_step=4.15s recent_step=4.07s eta=49:15:11 +2026-09-20 14:53:19 | INFO | train rank=0 | step=7320/50000 progress=14.6% loss=0.387251 lr=9.748078e-05 lr_backbone=9.748078e-05 elapsed=08:26:42 avg_step=4.15s recent_step=4.10s eta=49:14:26 +2026-09-20 14:54:00 | INFO | train rank=0 | step=7330/50000 progress=14.7% loss=0.430902 lr=9.747041e-05 lr_backbone=9.747041e-05 elapsed=08:27:23 avg_step=4.15s recent_step=4.09s eta=49:13:40 +2026-09-20 14:54:41 | INFO | train rank=0 | step=7340/50000 progress=14.7% loss=0.423529 lr=9.746001e-05 lr_backbone=9.746001e-05 elapsed=08:28:04 avg_step=4.15s recent_step=4.12s eta=49:12:57 +2026-09-20 14:55:22 | INFO | train rank=0 | step=7350/50000 progress=14.7% loss=0.434824 lr=9.744960e-05 lr_backbone=9.744960e-05 elapsed=08:28:45 avg_step=4.15s recent_step=4.11s eta=49:12:13 +2026-09-20 14:56:03 | INFO | train rank=0 | step=7360/50000 progress=14.7% loss=0.368988 lr=9.743916e-05 lr_backbone=9.743916e-05 elapsed=08:29:26 avg_step=4.15s recent_step=4.10s eta=49:11:28 +2026-09-20 14:56:44 | INFO | train rank=0 | step=7370/50000 progress=14.7% loss=0.368103 lr=9.742870e-05 lr_backbone=9.742870e-05 elapsed=08:30:07 avg_step=4.15s recent_step=4.08s eta=49:10:43 +2026-09-20 14:57:25 | INFO | train rank=0 | step=7380/50000 progress=14.8% loss=0.434594 lr=9.741822e-05 lr_backbone=9.741822e-05 elapsed=08:30:48 avg_step=4.15s recent_step=4.12s eta=49:09:59 +2026-09-20 14:58:06 | INFO | train rank=0 | step=7390/50000 progress=14.8% loss=0.373068 lr=9.740772e-05 lr_backbone=9.740772e-05 elapsed=08:31:30 avg_step=4.15s recent_step=4.13s eta=49:09:16 +2026-09-20 14:58:47 | INFO | train rank=0 | step=7400/50000 progress=14.8% loss=0.460859 lr=9.739720e-05 lr_backbone=9.739720e-05 elapsed=08:32:11 avg_step=4.15s recent_step=4.08s eta=49:08:31 +2026-09-20 14:59:28 | INFO | train rank=0 | step=7410/50000 progress=14.8% loss=0.418124 lr=9.738666e-05 lr_backbone=9.738666e-05 elapsed=08:32:52 avg_step=4.15s recent_step=4.12s eta=49:07:47 +2026-09-20 15:00:10 | INFO | train rank=0 | step=7420/50000 progress=14.8% loss=0.407946 lr=9.737610e-05 lr_backbone=9.737610e-05 elapsed=08:33:33 avg_step=4.15s recent_step=4.11s eta=49:07:04 +2026-09-20 15:00:50 | INFO | train rank=0 | step=7430/50000 progress=14.9% loss=0.379264 lr=9.736552e-05 lr_backbone=9.736552e-05 elapsed=08:34:14 avg_step=4.15s recent_step=4.07s eta=49:06:18 +2026-09-20 15:01:32 | INFO | train rank=0 | step=7440/50000 progress=14.9% loss=0.379662 lr=9.735492e-05 lr_backbone=9.735492e-05 elapsed=08:34:55 avg_step=4.15s recent_step=4.12s eta=49:05:34 +2026-09-20 15:02:13 | INFO | train rank=0 | step=7450/50000 progress=14.9% loss=0.398950 lr=9.734429e-05 lr_backbone=9.734429e-05 elapsed=08:35:36 avg_step=4.15s recent_step=4.11s eta=49:04:50 +2026-09-20 15:02:54 | INFO | train rank=0 | step=7460/50000 progress=14.9% loss=0.485217 lr=9.733365e-05 lr_backbone=9.733365e-05 elapsed=08:36:17 avg_step=4.15s recent_step=4.11s eta=49:04:06 +2026-09-20 15:03:35 | INFO | train rank=0 | step=7470/50000 progress=14.9% loss=0.353757 lr=9.732298e-05 lr_backbone=9.732298e-05 elapsed=08:36:58 avg_step=4.15s recent_step=4.10s eta=49:03:22 +2026-09-20 15:04:16 | INFO | train rank=0 | step=7480/50000 progress=15.0% loss=0.373410 lr=9.731230e-05 lr_backbone=9.731230e-05 elapsed=08:37:39 avg_step=4.15s recent_step=4.13s eta=49:02:39 +2026-09-20 15:04:57 | INFO | train rank=0 | step=7490/50000 progress=15.0% loss=0.368117 lr=9.730159e-05 lr_backbone=9.730159e-05 elapsed=08:38:20 avg_step=4.15s recent_step=4.07s eta=49:01:53 +2026-09-20 15:05:38 | INFO | train rank=0 | step=7500/50000 progress=15.0% loss=0.362285 lr=9.729086e-05 lr_backbone=9.729086e-05 elapsed=08:39:01 avg_step=4.15s recent_step=4.09s eta=49:01:08 +2026-09-20 15:06:19 | INFO | train rank=0 | step=7510/50000 progress=15.0% loss=0.352610 lr=9.728011e-05 lr_backbone=9.728011e-05 elapsed=08:39:42 avg_step=4.15s recent_step=4.11s eta=49:00:24 +2026-09-20 15:07:00 | INFO | train rank=0 | step=7520/50000 progress=15.0% loss=0.388107 lr=9.726935e-05 lr_backbone=9.726935e-05 elapsed=08:40:23 avg_step=4.15s recent_step=4.11s eta=48:59:40 +2026-09-20 15:07:41 | INFO | train rank=0 | step=7530/50000 progress=15.1% loss=0.360760 lr=9.725856e-05 lr_backbone=9.725856e-05 elapsed=08:41:04 avg_step=4.15s recent_step=4.10s eta=48:58:55 +2026-09-20 15:08:22 | INFO | train rank=0 | step=7540/50000 progress=15.1% loss=0.420444 lr=9.724775e-05 lr_backbone=9.724775e-05 elapsed=08:41:45 avg_step=4.15s recent_step=4.11s eta=48:58:12 +2026-09-20 15:09:03 | INFO | train rank=0 | step=7550/50000 progress=15.1% loss=0.327947 lr=9.723692e-05 lr_backbone=9.723692e-05 elapsed=08:42:26 avg_step=4.15s recent_step=4.07s eta=48:57:25 +2026-09-20 15:09:44 | INFO | train rank=0 | step=7560/50000 progress=15.1% loss=0.459360 lr=9.722606e-05 lr_backbone=9.722606e-05 elapsed=08:43:07 avg_step=4.15s recent_step=4.11s eta=48:56:42 +2026-09-20 15:10:25 | INFO | train rank=0 | step=7570/50000 progress=15.1% loss=0.458048 lr=9.721519e-05 lr_backbone=9.721519e-05 elapsed=08:43:48 avg_step=4.15s recent_step=4.11s eta=48:55:58 +2026-09-20 15:11:06 | INFO | train rank=0 | step=7580/50000 progress=15.2% loss=0.360797 lr=9.720430e-05 lr_backbone=9.720430e-05 elapsed=08:44:29 avg_step=4.15s recent_step=4.11s eta=48:55:14 +2026-09-20 15:11:47 | INFO | train rank=0 | step=7590/50000 progress=15.2% loss=0.360674 lr=9.719339e-05 lr_backbone=9.719339e-05 elapsed=08:45:10 avg_step=4.15s recent_step=4.09s eta=48:54:29 +2026-09-20 15:12:28 | INFO | train rank=0 | step=7600/50000 progress=15.2% loss=0.324962 lr=9.718245e-05 lr_backbone=9.718245e-05 elapsed=08:45:51 avg_step=4.15s recent_step=4.08s eta=48:53:44 +2026-09-20 15:13:08 | INFO | train rank=0 | step=7610/50000 progress=15.2% loss=0.421244 lr=9.717150e-05 lr_backbone=9.717150e-05 elapsed=08:46:32 avg_step=4.15s recent_step=4.07s eta=48:52:58 +2026-09-20 15:13:50 | INFO | train rank=0 | step=7620/50000 progress=15.2% loss=0.421799 lr=9.716052e-05 lr_backbone=9.716052e-05 elapsed=08:47:13 avg_step=4.15s recent_step=4.11s eta=48:52:14 +2026-09-20 15:14:31 | INFO | train rank=0 | step=7630/50000 progress=15.3% loss=0.411721 lr=9.714953e-05 lr_backbone=9.714953e-05 elapsed=08:47:54 avg_step=4.15s recent_step=4.10s eta=48:51:30 +2026-09-20 15:15:12 | INFO | train rank=0 | step=7640/50000 progress=15.3% loss=0.350726 lr=9.713851e-05 lr_backbone=9.713851e-05 elapsed=08:48:35 avg_step=4.15s recent_step=4.12s eta=48:50:46 +2026-09-20 15:15:52 | INFO | train rank=0 | step=7650/50000 progress=15.3% loss=0.377347 lr=9.712747e-05 lr_backbone=9.712747e-05 elapsed=08:49:16 avg_step=4.15s recent_step=4.08s eta=48:50:01 +2026-09-20 15:16:34 | INFO | train rank=0 | step=7660/50000 progress=15.3% loss=0.399574 lr=9.711642e-05 lr_backbone=9.711642e-05 elapsed=08:49:57 avg_step=4.15s recent_step=4.13s eta=48:49:18 +2026-09-20 15:17:15 | INFO | train rank=0 | step=7670/50000 progress=15.3% loss=0.368845 lr=9.710534e-05 lr_backbone=9.710534e-05 elapsed=08:50:38 avg_step=4.15s recent_step=4.08s eta=48:48:33 +2026-09-20 15:17:56 | INFO | train rank=0 | step=7680/50000 progress=15.4% loss=0.510744 lr=9.709424e-05 lr_backbone=9.709424e-05 elapsed=08:51:19 avg_step=4.15s recent_step=4.11s eta=48:47:49 +2026-09-20 15:18:37 | INFO | train rank=0 | step=7690/50000 progress=15.4% loss=0.408273 lr=9.708312e-05 lr_backbone=9.708312e-05 elapsed=08:52:00 avg_step=4.15s recent_step=4.12s eta=48:47:06 +2026-09-20 15:19:18 | INFO | train rank=0 | step=7700/50000 progress=15.4% loss=0.425671 lr=9.707198e-05 lr_backbone=9.707198e-05 elapsed=08:52:41 avg_step=4.15s recent_step=4.12s eta=48:46:22 +2026-09-20 15:19:59 | INFO | train rank=0 | step=7710/50000 progress=15.4% loss=0.341517 lr=9.706082e-05 lr_backbone=9.706082e-05 elapsed=08:53:22 avg_step=4.15s recent_step=4.10s eta=48:45:38 +2026-09-20 15:20:40 | INFO | train rank=0 | step=7720/50000 progress=15.4% loss=0.408149 lr=9.704964e-05 lr_backbone=9.704964e-05 elapsed=08:54:04 avg_step=4.15s recent_step=4.11s eta=48:44:54 +2026-09-20 15:21:21 | INFO | train rank=0 | step=7730/50000 progress=15.5% loss=0.429935 lr=9.703843e-05 lr_backbone=9.703843e-05 elapsed=08:54:44 avg_step=4.15s recent_step=4.08s eta=48:44:09 +2026-09-20 15:22:02 | INFO | train rank=0 | step=7740/50000 progress=15.5% loss=0.448077 lr=9.702721e-05 lr_backbone=9.702721e-05 elapsed=08:55:26 avg_step=4.15s recent_step=4.13s eta=48:43:26 +2026-09-20 15:22:43 | INFO | train rank=0 | step=7750/50000 progress=15.5% loss=0.344862 lr=9.701597e-05 lr_backbone=9.701597e-05 elapsed=08:56:07 avg_step=4.15s recent_step=4.12s eta=48:42:43 +2026-09-20 15:23:25 | INFO | train rank=0 | step=7760/50000 progress=15.5% loss=0.327649 lr=9.700471e-05 lr_backbone=9.700471e-05 elapsed=08:56:48 avg_step=4.15s recent_step=4.11s eta=48:41:59 +2026-09-20 15:24:06 | INFO | train rank=0 | step=7770/50000 progress=15.5% loss=0.346578 lr=9.699342e-05 lr_backbone=9.699342e-05 elapsed=08:57:29 avg_step=4.15s recent_step=4.10s eta=48:41:15 +2026-09-20 15:24:47 | INFO | train rank=0 | step=7780/50000 progress=15.6% loss=0.392934 lr=9.698212e-05 lr_backbone=9.698212e-05 elapsed=08:58:10 avg_step=4.15s recent_step=4.10s eta=48:40:31 +2026-09-20 15:25:27 | INFO | train rank=0 | step=7790/50000 progress=15.6% loss=0.371490 lr=9.697079e-05 lr_backbone=9.697079e-05 elapsed=08:58:51 avg_step=4.15s recent_step=4.07s eta=48:39:45 +2026-09-20 15:26:08 | INFO | train rank=0 | step=7800/50000 progress=15.6% loss=0.407999 lr=9.695945e-05 lr_backbone=9.695945e-05 elapsed=08:59:32 avg_step=4.15s recent_step=4.12s eta=48:39:02 +2026-09-20 15:26:50 | INFO | train rank=0 | step=7810/50000 progress=15.6% loss=0.472579 lr=9.694808e-05 lr_backbone=9.694808e-05 elapsed=09:00:13 avg_step=4.15s recent_step=4.11s eta=48:38:18 +2026-09-20 15:27:31 | INFO | train rank=0 | step=7820/50000 progress=15.6% loss=0.359643 lr=9.693669e-05 lr_backbone=9.693669e-05 elapsed=09:00:54 avg_step=4.15s recent_step=4.12s eta=48:37:35 +2026-09-20 15:28:12 | INFO | train rank=0 | step=7830/50000 progress=15.7% loss=0.458096 lr=9.692529e-05 lr_backbone=9.692529e-05 elapsed=09:01:35 avg_step=4.15s recent_step=4.10s eta=48:36:51 +2026-09-20 15:28:53 | INFO | train rank=0 | step=7840/50000 progress=15.7% loss=0.324103 lr=9.691386e-05 lr_backbone=9.691386e-05 elapsed=09:02:16 avg_step=4.15s recent_step=4.08s eta=48:36:06 +2026-09-20 15:29:33 | INFO | train rank=0 | step=7850/50000 progress=15.7% loss=0.425501 lr=9.690241e-05 lr_backbone=9.690241e-05 elapsed=09:02:57 avg_step=4.15s recent_step=4.06s eta=48:35:20 +2026-09-20 15:30:14 | INFO | train rank=0 | step=7860/50000 progress=15.7% loss=0.391549 lr=9.689094e-05 lr_backbone=9.689094e-05 elapsed=09:03:38 avg_step=4.15s recent_step=4.11s eta=48:34:36 +2026-09-20 15:30:55 | INFO | train rank=0 | step=7870/50000 progress=15.7% loss=0.378941 lr=9.687945e-05 lr_backbone=9.687945e-05 elapsed=09:04:19 avg_step=4.15s recent_step=4.11s eta=48:33:52 +2026-09-20 15:31:37 | INFO | train rank=0 | step=7880/50000 progress=15.8% loss=0.373981 lr=9.686794e-05 lr_backbone=9.686794e-05 elapsed=09:05:00 avg_step=4.15s recent_step=4.13s eta=48:33:10 +2026-09-20 15:32:18 | INFO | train rank=0 | step=7890/50000 progress=15.8% loss=0.354198 lr=9.685641e-05 lr_backbone=9.685641e-05 elapsed=09:05:41 avg_step=4.15s recent_step=4.10s eta=48:32:26 +2026-09-20 15:32:59 | INFO | train rank=0 | step=7900/50000 progress=15.8% loss=0.416404 lr=9.684486e-05 lr_backbone=9.684486e-05 elapsed=09:06:22 avg_step=4.15s recent_step=4.11s eta=48:31:42 +2026-09-20 15:33:40 | INFO | train rank=0 | step=7910/50000 progress=15.8% loss=0.366770 lr=9.683329e-05 lr_backbone=9.683329e-05 elapsed=09:07:03 avg_step=4.15s recent_step=4.08s eta=48:30:57 +2026-09-20 15:34:21 | INFO | train rank=0 | step=7920/50000 progress=15.8% loss=0.383456 lr=9.682170e-05 lr_backbone=9.682170e-05 elapsed=09:07:44 avg_step=4.15s recent_step=4.11s eta=48:30:13 +2026-09-20 15:35:02 | INFO | train rank=0 | step=7930/50000 progress=15.9% loss=0.508919 lr=9.681008e-05 lr_backbone=9.681008e-05 elapsed=09:08:25 avg_step=4.15s recent_step=4.10s eta=48:29:29 +2026-09-20 15:35:43 | INFO | train rank=0 | step=7940/50000 progress=15.9% loss=0.408821 lr=9.679845e-05 lr_backbone=9.679845e-05 elapsed=09:09:06 avg_step=4.15s recent_step=4.12s eta=48:28:46 +2026-09-20 15:36:24 | INFO | train rank=0 | step=7950/50000 progress=15.9% loss=0.416248 lr=9.678680e-05 lr_backbone=9.678680e-05 elapsed=09:09:47 avg_step=4.15s recent_step=4.11s eta=48:28:02 +2026-09-20 15:37:05 | INFO | train rank=0 | step=7960/50000 progress=15.9% loss=0.375355 lr=9.677512e-05 lr_backbone=9.677512e-05 elapsed=09:10:28 avg_step=4.15s recent_step=4.09s eta=48:27:18 +2026-09-20 15:37:45 | INFO | train rank=0 | step=7970/50000 progress=15.9% loss=0.430138 lr=9.676343e-05 lr_backbone=9.676343e-05 elapsed=09:11:09 avg_step=4.15s recent_step=4.05s eta=48:26:31 +2026-09-20 15:38:26 | INFO | train rank=0 | step=7980/50000 progress=16.0% loss=0.311994 lr=9.675171e-05 lr_backbone=9.675171e-05 elapsed=09:11:50 avg_step=4.15s recent_step=4.08s eta=48:25:46 +2026-09-20 15:39:07 | INFO | train rank=0 | step=7990/50000 progress=16.0% loss=0.407389 lr=9.673998e-05 lr_backbone=9.673998e-05 elapsed=09:12:30 avg_step=4.15s recent_step=4.09s eta=48:25:01 +2026-09-20 15:39:48 | INFO | train rank=0 | step=8000/50000 progress=16.0% loss=0.430799 lr=9.672822e-05 lr_backbone=9.672822e-05 elapsed=09:13:11 avg_step=4.15s recent_step=4.10s eta=48:24:17 +2026-09-20 15:40:29 | INFO | train rank=0 | step=8010/50000 progress=16.0% loss=0.368602 lr=9.671645e-05 lr_backbone=9.671645e-05 elapsed=09:13:52 avg_step=4.15s recent_step=4.10s eta=48:23:33 +2026-09-20 15:41:10 | INFO | train rank=0 | step=8020/50000 progress=16.0% loss=0.363401 lr=9.670465e-05 lr_backbone=9.670465e-05 elapsed=09:14:33 avg_step=4.15s recent_step=4.08s eta=48:22:48 +2026-09-20 15:41:50 | INFO | train rank=0 | step=8030/50000 progress=16.1% loss=0.352493 lr=9.669283e-05 lr_backbone=9.669283e-05 elapsed=09:15:14 avg_step=4.15s recent_step=4.05s eta=48:22:02 +2026-09-20 15:42:32 | INFO | train rank=0 | step=8040/50000 progress=16.1% loss=0.398522 lr=9.668100e-05 lr_backbone=9.668100e-05 elapsed=09:15:55 avg_step=4.15s recent_step=4.13s eta=48:21:19 +2026-09-20 15:43:13 | INFO | train rank=0 | step=8050/50000 progress=16.1% loss=0.350313 lr=9.666914e-05 lr_backbone=9.666914e-05 elapsed=09:16:36 avg_step=4.15s recent_step=4.11s eta=48:20:35 +2026-09-20 15:43:54 | INFO | train rank=0 | step=8060/50000 progress=16.1% loss=0.382477 lr=9.665726e-05 lr_backbone=9.665726e-05 elapsed=09:17:17 avg_step=4.15s recent_step=4.10s eta=48:19:51 +2026-09-20 15:44:35 | INFO | train rank=0 | step=8070/50000 progress=16.1% loss=0.383678 lr=9.664536e-05 lr_backbone=9.664536e-05 elapsed=09:17:58 avg_step=4.15s recent_step=4.10s eta=48:19:07 +2026-09-20 15:45:16 | INFO | train rank=0 | step=8080/50000 progress=16.2% loss=0.400119 lr=9.663344e-05 lr_backbone=9.663344e-05 elapsed=09:18:39 avg_step=4.15s recent_step=4.11s eta=48:18:23 +2026-09-20 15:45:57 | INFO | train rank=0 | step=8090/50000 progress=16.2% loss=0.556243 lr=9.662150e-05 lr_backbone=9.662150e-05 elapsed=09:19:20 avg_step=4.15s recent_step=4.08s eta=48:17:39 +2026-09-20 15:46:38 | INFO | train rank=0 | step=8100/50000 progress=16.2% loss=0.428176 lr=9.660954e-05 lr_backbone=9.660954e-05 elapsed=09:20:01 avg_step=4.15s recent_step=4.09s eta=48:16:54 +2026-09-20 15:47:19 | INFO | train rank=0 | step=8110/50000 progress=16.2% loss=0.329000 lr=9.659756e-05 lr_backbone=9.659756e-05 elapsed=09:20:42 avg_step=4.15s recent_step=4.11s eta=48:16:11 +2026-09-20 15:48:00 | INFO | train rank=0 | step=8120/50000 progress=16.2% loss=0.352256 lr=9.658556e-05 lr_backbone=9.658556e-05 elapsed=09:21:23 avg_step=4.15s recent_step=4.10s eta=48:15:27 +2026-09-20 15:48:41 | INFO | train rank=0 | step=8130/50000 progress=16.3% loss=0.408846 lr=9.657354e-05 lr_backbone=9.657354e-05 elapsed=09:22:04 avg_step=4.15s recent_step=4.12s eta=48:14:44 +2026-09-20 15:49:22 | INFO | train rank=0 | step=8140/50000 progress=16.3% loss=0.379244 lr=9.656150e-05 lr_backbone=9.656150e-05 elapsed=09:22:45 avg_step=4.15s recent_step=4.11s eta=48:14:00 +2026-09-20 15:50:03 | INFO | train rank=0 | step=8150/50000 progress=16.3% loss=0.378844 lr=9.654944e-05 lr_backbone=9.654944e-05 elapsed=09:23:26 avg_step=4.15s recent_step=4.09s eta=48:13:16 +2026-09-20 15:50:44 | INFO | train rank=0 | step=8160/50000 progress=16.3% loss=0.355305 lr=9.653736e-05 lr_backbone=9.653736e-05 elapsed=09:24:07 avg_step=4.15s recent_step=4.09s eta=48:12:32 +2026-09-20 15:51:25 | INFO | train rank=0 | step=8170/50000 progress=16.3% loss=0.454262 lr=9.652525e-05 lr_backbone=9.652525e-05 elapsed=09:24:48 avg_step=4.15s recent_step=4.13s eta=48:11:49 +2026-09-20 15:52:06 | INFO | train rank=0 | step=8180/50000 progress=16.4% loss=0.488016 lr=9.651313e-05 lr_backbone=9.651313e-05 elapsed=09:25:30 avg_step=4.15s recent_step=4.13s eta=48:11:07 +2026-09-20 15:52:47 | INFO | train rank=0 | step=8190/50000 progress=16.4% loss=0.463355 lr=9.650099e-05 lr_backbone=9.650099e-05 elapsed=09:26:11 avg_step=4.15s recent_step=4.10s eta=48:10:23 +2026-09-20 15:53:28 | INFO | train rank=0 | step=8200/50000 progress=16.4% loss=0.430533 lr=9.648882e-05 lr_backbone=9.648882e-05 elapsed=09:26:52 avg_step=4.15s recent_step=4.10s eta=48:09:39 +2026-09-20 15:54:09 | INFO | train rank=0 | step=8210/50000 progress=16.4% loss=0.361081 lr=9.647664e-05 lr_backbone=9.647664e-05 elapsed=09:27:33 avg_step=4.15s recent_step=4.08s eta=48:08:54 +2026-09-20 15:54:50 | INFO | train rank=0 | step=8220/50000 progress=16.4% loss=0.391217 lr=9.646444e-05 lr_backbone=9.646444e-05 elapsed=09:28:14 avg_step=4.15s recent_step=4.10s eta=48:08:10 +2026-09-20 15:55:31 | INFO | train rank=0 | step=8230/50000 progress=16.5% loss=0.404518 lr=9.645221e-05 lr_backbone=9.645221e-05 elapsed=09:28:55 avg_step=4.15s recent_step=4.11s eta=48:07:27 +2026-09-20 15:56:12 | INFO | train rank=0 | step=8240/50000 progress=16.5% loss=0.302478 lr=9.643997e-05 lr_backbone=9.643997e-05 elapsed=09:29:36 avg_step=4.15s recent_step=4.11s eta=48:06:43 +2026-09-20 15:56:53 | INFO | train rank=0 | step=8250/50000 progress=16.5% loss=0.429027 lr=9.642770e-05 lr_backbone=9.642770e-05 elapsed=09:30:17 avg_step=4.15s recent_step=4.09s eta=48:05:59 +2026-09-20 15:57:34 | INFO | train rank=0 | step=8260/50000 progress=16.5% loss=0.387660 lr=9.641542e-05 lr_backbone=9.641542e-05 elapsed=09:30:58 avg_step=4.15s recent_step=4.10s eta=48:05:15 +2026-09-20 15:58:15 | INFO | train rank=0 | step=8270/50000 progress=16.5% loss=0.382483 lr=9.640311e-05 lr_backbone=9.640311e-05 elapsed=09:31:38 avg_step=4.15s recent_step=4.08s eta=48:04:30 +2026-09-20 15:58:56 | INFO | train rank=0 | step=8280/50000 progress=16.6% loss=0.322014 lr=9.639078e-05 lr_backbone=9.639078e-05 elapsed=09:32:19 avg_step=4.15s recent_step=4.09s eta=48:03:46 +2026-09-20 15:59:37 | INFO | train rank=0 | step=8290/50000 progress=16.6% loss=0.384570 lr=9.637844e-05 lr_backbone=9.637844e-05 elapsed=09:33:00 avg_step=4.15s recent_step=4.11s eta=48:03:02 +2026-09-20 16:00:18 | INFO | train rank=0 | step=8300/50000 progress=16.6% loss=0.358260 lr=9.636607e-05 lr_backbone=9.636607e-05 elapsed=09:33:41 avg_step=4.15s recent_step=4.10s eta=48:02:19 +2026-09-20 16:00:59 | INFO | train rank=0 | step=8310/50000 progress=16.6% loss=0.429089 lr=9.635368e-05 lr_backbone=9.635368e-05 elapsed=09:34:23 avg_step=4.15s recent_step=4.12s eta=48:01:36 +2026-09-20 16:01:40 | INFO | train rank=0 | step=8320/50000 progress=16.6% loss=0.433617 lr=9.634128e-05 lr_backbone=9.634128e-05 elapsed=09:35:04 avg_step=4.15s recent_step=4.11s eta=48:00:52 +2026-09-20 16:02:21 | INFO | train rank=0 | step=8330/50000 progress=16.7% loss=0.415059 lr=9.632885e-05 lr_backbone=9.632885e-05 elapsed=09:35:44 avg_step=4.15s recent_step=4.07s eta=48:00:07 +2026-09-20 16:03:02 | INFO | train rank=0 | step=8340/50000 progress=16.7% loss=0.323100 lr=9.631640e-05 lr_backbone=9.631640e-05 elapsed=09:36:25 avg_step=4.15s recent_step=4.11s eta=47:59:24 +2026-09-20 16:03:43 | INFO | train rank=0 | step=8350/50000 progress=16.7% loss=0.345484 lr=9.630393e-05 lr_backbone=9.630393e-05 elapsed=09:37:07 avg_step=4.15s recent_step=4.11s eta=47:58:40 +2026-09-20 16:04:24 | INFO | train rank=0 | step=8360/50000 progress=16.7% loss=0.353382 lr=9.629145e-05 lr_backbone=9.629145e-05 elapsed=09:37:48 avg_step=4.15s recent_step=4.11s eta=47:57:57 +2026-09-20 16:05:06 | INFO | train rank=0 | step=8370/50000 progress=16.7% loss=0.357365 lr=9.627894e-05 lr_backbone=9.627894e-05 elapsed=09:38:29 avg_step=4.15s recent_step=4.12s eta=47:57:14 +2026-09-20 16:05:47 | INFO | train rank=0 | step=8380/50000 progress=16.8% loss=0.391128 lr=9.626641e-05 lr_backbone=9.626641e-05 elapsed=09:39:10 avg_step=4.15s recent_step=4.12s eta=47:56:31 +2026-09-20 16:06:27 | INFO | train rank=0 | step=8390/50000 progress=16.8% loss=0.333550 lr=9.625386e-05 lr_backbone=9.625386e-05 elapsed=09:39:51 avg_step=4.15s recent_step=4.07s eta=47:55:46 +2026-09-20 16:07:09 | INFO | train rank=0 | step=8400/50000 progress=16.8% loss=0.429944 lr=9.624129e-05 lr_backbone=9.624129e-05 elapsed=09:40:32 avg_step=4.15s recent_step=4.11s eta=47:55:03 +2026-09-20 16:07:50 | INFO | train rank=0 | step=8410/50000 progress=16.8% loss=0.383947 lr=9.622870e-05 lr_backbone=9.622870e-05 elapsed=09:41:13 avg_step=4.15s recent_step=4.10s eta=47:54:19 +2026-09-20 16:08:31 | INFO | train rank=0 | step=8420/50000 progress=16.8% loss=0.415814 lr=9.621609e-05 lr_backbone=9.621609e-05 elapsed=09:41:54 avg_step=4.15s recent_step=4.11s eta=47:53:36 +2026-09-20 16:09:12 | INFO | train rank=0 | step=8430/50000 progress=16.9% loss=0.381196 lr=9.620346e-05 lr_backbone=9.620346e-05 elapsed=09:42:35 avg_step=4.15s recent_step=4.10s eta=47:52:52 +2026-09-20 16:09:53 | INFO | train rank=0 | step=8440/50000 progress=16.9% loss=0.420787 lr=9.619081e-05 lr_backbone=9.619081e-05 elapsed=09:43:16 avg_step=4.15s recent_step=4.10s eta=47:52:08 +2026-09-20 16:10:33 | INFO | train rank=0 | step=8450/50000 progress=16.9% loss=0.398751 lr=9.617814e-05 lr_backbone=9.617814e-05 elapsed=09:43:57 avg_step=4.15s recent_step=4.07s eta=47:51:23 +2026-09-20 16:11:14 | INFO | train rank=0 | step=8460/50000 progress=16.9% loss=0.419150 lr=9.616545e-05 lr_backbone=9.616545e-05 elapsed=09:44:38 avg_step=4.15s recent_step=4.08s eta=47:50:38 +2026-09-20 16:11:55 | INFO | train rank=0 | step=8470/50000 progress=16.9% loss=0.328636 lr=9.615274e-05 lr_backbone=9.615274e-05 elapsed=09:45:18 avg_step=4.15s recent_step=4.09s eta=47:49:54 +2026-09-20 16:12:36 | INFO | train rank=0 | step=8480/50000 progress=17.0% loss=0.364830 lr=9.614001e-05 lr_backbone=9.614001e-05 elapsed=09:45:59 avg_step=4.15s recent_step=4.09s eta=47:49:10 +2026-09-20 16:13:17 | INFO | train rank=0 | step=8490/50000 progress=17.0% loss=0.365626 lr=9.612726e-05 lr_backbone=9.612726e-05 elapsed=09:46:41 avg_step=4.15s recent_step=4.11s eta=47:48:27 +2026-09-20 16:13:58 | INFO | train rank=0 | step=8500/50000 progress=17.0% loss=0.390590 lr=9.611449e-05 lr_backbone=9.611449e-05 elapsed=09:47:22 avg_step=4.15s recent_step=4.12s eta=47:47:44 +2026-09-20 16:14:39 | INFO | train rank=0 | step=8510/50000 progress=17.0% loss=0.431380 lr=9.610170e-05 lr_backbone=9.610170e-05 elapsed=09:48:02 avg_step=4.15s recent_step=4.06s eta=47:46:59 +2026-09-20 16:15:20 | INFO | train rank=0 | step=8520/50000 progress=17.0% loss=0.404796 lr=9.608888e-05 lr_backbone=9.608888e-05 elapsed=09:48:44 avg_step=4.15s recent_step=4.13s eta=47:46:17 +2026-09-20 16:16:01 | INFO | train rank=0 | step=8530/50000 progress=17.1% loss=0.393982 lr=9.607605e-05 lr_backbone=9.607605e-05 elapsed=09:49:25 avg_step=4.15s recent_step=4.10s eta=47:45:33 +2026-09-20 16:16:42 | INFO | train rank=0 | step=8540/50000 progress=17.1% loss=0.355548 lr=9.606320e-05 lr_backbone=9.606320e-05 elapsed=09:50:06 avg_step=4.15s recent_step=4.11s eta=47:44:50 +2026-09-20 16:17:24 | INFO | train rank=0 | step=8550/50000 progress=17.1% loss=0.377972 lr=9.605033e-05 lr_backbone=9.605033e-05 elapsed=09:50:47 avg_step=4.15s recent_step=4.11s eta=47:44:07 +2026-09-20 16:18:05 | INFO | train rank=0 | step=8560/50000 progress=17.1% loss=0.406243 lr=9.603744e-05 lr_backbone=9.603744e-05 elapsed=09:51:28 avg_step=4.15s recent_step=4.12s eta=47:43:24 +2026-09-20 16:18:46 | INFO | train rank=0 | step=8570/50000 progress=17.1% loss=0.361010 lr=9.602452e-05 lr_backbone=9.602452e-05 elapsed=09:52:09 avg_step=4.15s recent_step=4.08s eta=47:42:39 +2026-09-20 16:19:27 | INFO | train rank=0 | step=8580/50000 progress=17.2% loss=0.373153 lr=9.601159e-05 lr_backbone=9.601159e-05 elapsed=09:52:50 avg_step=4.15s recent_step=4.11s eta=47:41:56 +2026-09-20 16:20:08 | INFO | train rank=0 | step=8590/50000 progress=17.2% loss=0.413520 lr=9.599864e-05 lr_backbone=9.599864e-05 elapsed=09:53:31 avg_step=4.15s recent_step=4.09s eta=47:41:12 +2026-09-20 16:20:49 | INFO | train rank=0 | step=8600/50000 progress=17.2% loss=0.438059 lr=9.598567e-05 lr_backbone=9.598567e-05 elapsed=09:54:12 avg_step=4.15s recent_step=4.10s eta=47:40:29 +2026-09-20 16:21:30 | INFO | train rank=0 | step=8610/50000 progress=17.2% loss=0.358600 lr=9.597267e-05 lr_backbone=9.597267e-05 elapsed=09:54:53 avg_step=4.15s recent_step=4.11s eta=47:39:45 +2026-09-20 16:22:11 | INFO | train rank=0 | step=8620/50000 progress=17.2% loss=0.376243 lr=9.595966e-05 lr_backbone=9.595966e-05 elapsed=09:55:34 avg_step=4.15s recent_step=4.10s eta=47:39:01 +2026-09-20 16:22:51 | INFO | train rank=0 | step=8630/50000 progress=17.3% loss=0.339520 lr=9.594663e-05 lr_backbone=9.594663e-05 elapsed=09:56:15 avg_step=4.15s recent_step=4.08s eta=47:38:17 +2026-09-20 16:23:32 | INFO | train rank=0 | step=8640/50000 progress=17.3% loss=0.414248 lr=9.593358e-05 lr_backbone=9.593358e-05 elapsed=09:56:56 avg_step=4.15s recent_step=4.10s eta=47:37:33 +2026-09-20 16:24:13 | INFO | train rank=0 | step=8650/50000 progress=17.3% loss=0.357265 lr=9.592050e-05 lr_backbone=9.592050e-05 elapsed=09:57:37 avg_step=4.15s recent_step=4.11s eta=47:36:50 +2026-09-20 16:24:54 | INFO | train rank=0 | step=8660/50000 progress=17.3% loss=0.349466 lr=9.590741e-05 lr_backbone=9.590741e-05 elapsed=09:58:18 avg_step=4.15s recent_step=4.08s eta=47:36:05 +2026-09-20 16:25:35 | INFO | train rank=0 | step=8670/50000 progress=17.3% loss=0.377452 lr=9.589430e-05 lr_backbone=9.589430e-05 elapsed=09:58:59 avg_step=4.15s recent_step=4.12s eta=47:35:23 +2026-09-20 16:26:17 | INFO | train rank=0 | step=8680/50000 progress=17.4% loss=0.358670 lr=9.588116e-05 lr_backbone=9.588116e-05 elapsed=09:59:40 avg_step=4.15s recent_step=4.12s eta=47:34:40 +2026-09-20 16:26:57 | INFO | train rank=0 | step=8690/50000 progress=17.4% loss=0.391845 lr=9.586801e-05 lr_backbone=9.586801e-05 elapsed=10:00:21 avg_step=4.15s recent_step=4.07s eta=47:33:55 +2026-09-20 16:27:39 | INFO | train rank=0 | step=8700/50000 progress=17.4% loss=0.381475 lr=9.585483e-05 lr_backbone=9.585483e-05 elapsed=10:01:02 avg_step=4.15s recent_step=4.12s eta=47:33:12 +2026-09-20 16:28:20 | INFO | train rank=0 | step=8710/50000 progress=17.4% loss=0.454359 lr=9.584164e-05 lr_backbone=9.584164e-05 elapsed=10:01:43 avg_step=4.15s recent_step=4.11s eta=47:32:29 +2026-09-20 16:29:01 | INFO | train rank=0 | step=8720/50000 progress=17.4% loss=0.357892 lr=9.582843e-05 lr_backbone=9.582843e-05 elapsed=10:02:24 avg_step=4.14s recent_step=4.08s eta=47:31:45 +2026-09-20 16:29:42 | INFO | train rank=0 | step=8730/50000 progress=17.5% loss=0.379023 lr=9.581519e-05 lr_backbone=9.581519e-05 elapsed=10:03:05 avg_step=4.14s recent_step=4.10s eta=47:31:01 +2026-09-20 16:30:23 | INFO | train rank=0 | step=8740/50000 progress=17.5% loss=0.373773 lr=9.580194e-05 lr_backbone=9.580194e-05 elapsed=10:03:46 avg_step=4.14s recent_step=4.14s eta=47:30:20 +2026-09-20 16:31:04 | INFO | train rank=0 | step=8750/50000 progress=17.5% loss=0.340808 lr=9.578867e-05 lr_backbone=9.578867e-05 elapsed=10:04:27 avg_step=4.14s recent_step=4.07s eta=47:29:35 +2026-09-20 16:31:45 | INFO | train rank=0 | step=8760/50000 progress=17.5% loss=0.423676 lr=9.577537e-05 lr_backbone=9.577537e-05 elapsed=10:05:08 avg_step=4.14s recent_step=4.11s eta=47:28:52 +2026-09-20 16:32:26 | INFO | train rank=0 | step=8770/50000 progress=17.5% loss=0.350749 lr=9.576206e-05 lr_backbone=9.576206e-05 elapsed=10:05:49 avg_step=4.14s recent_step=4.11s eta=47:28:08 +2026-09-20 16:33:07 | INFO | train rank=0 | step=8780/50000 progress=17.6% loss=0.435243 lr=9.574872e-05 lr_backbone=9.574872e-05 elapsed=10:06:30 avg_step=4.14s recent_step=4.09s eta=47:27:24 +2026-09-20 16:33:48 | INFO | train rank=0 | step=8790/50000 progress=17.6% loss=0.476117 lr=9.573537e-05 lr_backbone=9.573537e-05 elapsed=10:07:11 avg_step=4.14s recent_step=4.09s eta=47:26:41 +2026-09-20 16:34:29 | INFO | train rank=0 | step=8800/50000 progress=17.6% loss=0.345812 lr=9.572200e-05 lr_backbone=9.572200e-05 elapsed=10:07:52 avg_step=4.14s recent_step=4.12s eta=47:25:58 +2026-09-20 16:35:10 | INFO | train rank=0 | step=8810/50000 progress=17.6% loss=0.352344 lr=9.570860e-05 lr_backbone=9.570860e-05 elapsed=10:08:33 avg_step=4.14s recent_step=4.08s eta=47:25:13 +2026-09-20 16:35:51 | INFO | train rank=0 | step=8820/50000 progress=17.6% loss=0.343845 lr=9.569519e-05 lr_backbone=9.569519e-05 elapsed=10:09:14 avg_step=4.14s recent_step=4.11s eta=47:24:30 +2026-09-20 16:36:32 | INFO | train rank=0 | step=8830/50000 progress=17.7% loss=0.359194 lr=9.568176e-05 lr_backbone=9.568176e-05 elapsed=10:09:55 avg_step=4.14s recent_step=4.12s eta=47:23:48 +2026-09-20 16:37:13 | INFO | train rank=0 | step=8840/50000 progress=17.7% loss=0.334815 lr=9.566830e-05 lr_backbone=9.566830e-05 elapsed=10:10:36 avg_step=4.14s recent_step=4.11s eta=47:23:05 +2026-09-20 16:37:54 | INFO | train rank=0 | step=8850/50000 progress=17.7% loss=0.335882 lr=9.565483e-05 lr_backbone=9.565483e-05 elapsed=10:11:17 avg_step=4.14s recent_step=4.09s eta=47:22:21 +2026-09-20 16:38:35 | INFO | train rank=0 | step=8860/50000 progress=17.7% loss=0.318206 lr=9.564133e-05 lr_backbone=9.564133e-05 elapsed=10:11:58 avg_step=4.14s recent_step=4.11s eta=47:21:38 +2026-09-20 16:39:16 | INFO | train rank=0 | step=8870/50000 progress=17.7% loss=0.320754 lr=9.562782e-05 lr_backbone=9.562782e-05 elapsed=10:12:39 avg_step=4.14s recent_step=4.08s eta=47:20:53 +2026-09-20 16:39:57 | INFO | train rank=0 | step=8880/50000 progress=17.8% loss=0.348068 lr=9.561429e-05 lr_backbone=9.561429e-05 elapsed=10:13:20 avg_step=4.14s recent_step=4.11s eta=47:20:10 +2026-09-20 16:40:38 | INFO | train rank=0 | step=8890/50000 progress=17.8% loss=0.360968 lr=9.560073e-05 lr_backbone=9.560073e-05 elapsed=10:14:02 avg_step=4.14s recent_step=4.13s eta=47:19:28 +2026-09-20 16:41:19 | INFO | train rank=0 | step=8900/50000 progress=17.8% loss=0.427683 lr=9.558716e-05 lr_backbone=9.558716e-05 elapsed=10:14:43 avg_step=4.14s recent_step=4.13s eta=47:18:46 +2026-09-20 16:42:01 | INFO | train rank=0 | step=8910/50000 progress=17.8% loss=0.427673 lr=9.557357e-05 lr_backbone=9.557357e-05 elapsed=10:15:24 avg_step=4.14s recent_step=4.10s eta=47:18:02 +2026-09-20 16:42:41 | INFO | train rank=0 | step=8920/50000 progress=17.8% loss=0.398032 lr=9.555995e-05 lr_backbone=9.555995e-05 elapsed=10:16:05 avg_step=4.14s recent_step=4.09s eta=47:17:19 +2026-09-20 16:43:22 | INFO | train rank=0 | step=8930/50000 progress=17.9% loss=0.403789 lr=9.554632e-05 lr_backbone=9.554632e-05 elapsed=10:16:46 avg_step=4.14s recent_step=4.09s eta=47:16:35 +2026-09-20 16:44:03 | INFO | train rank=0 | step=8940/50000 progress=17.9% loss=0.363562 lr=9.553267e-05 lr_backbone=9.553267e-05 elapsed=10:17:27 avg_step=4.14s recent_step=4.10s eta=47:15:51 +2026-09-20 16:44:44 | INFO | train rank=0 | step=8950/50000 progress=17.9% loss=0.339707 lr=9.551899e-05 lr_backbone=9.551899e-05 elapsed=10:18:07 avg_step=4.14s recent_step=4.08s eta=47:15:07 +2026-09-20 16:45:25 | INFO | train rank=0 | step=8960/50000 progress=17.9% loss=0.460521 lr=9.550530e-05 lr_backbone=9.550530e-05 elapsed=10:18:49 avg_step=4.14s recent_step=4.11s eta=47:14:24 +2026-09-20 16:46:06 | INFO | train rank=0 | step=8970/50000 progress=17.9% loss=0.349666 lr=9.549159e-05 lr_backbone=9.549159e-05 elapsed=10:19:30 avg_step=4.14s recent_step=4.10s eta=47:13:41 +2026-09-20 16:46:47 | INFO | train rank=0 | step=8980/50000 progress=18.0% loss=0.481250 lr=9.547785e-05 lr_backbone=9.547785e-05 elapsed=10:20:11 avg_step=4.14s recent_step=4.09s eta=47:12:57 +2026-09-20 16:47:28 | INFO | train rank=0 | step=8990/50000 progress=18.0% loss=0.424289 lr=9.546410e-05 lr_backbone=9.546410e-05 elapsed=10:20:51 avg_step=4.14s recent_step=4.08s eta=47:12:12 +2026-09-20 16:48:09 | INFO | train rank=0 | step=9000/50000 progress=18.0% loss=0.362983 lr=9.545033e-05 lr_backbone=9.545033e-05 elapsed=10:21:32 avg_step=4.14s recent_step=4.11s eta=47:11:29 +2026-09-20 16:48:50 | INFO | train rank=0 | step=9010/50000 progress=18.0% loss=0.333683 lr=9.543653e-05 lr_backbone=9.543653e-05 elapsed=10:22:13 avg_step=4.14s recent_step=4.09s eta=47:10:46 +2026-09-20 16:49:31 | INFO | train rank=0 | step=9020/50000 progress=18.0% loss=0.431931 lr=9.542272e-05 lr_backbone=9.542272e-05 elapsed=10:22:54 avg_step=4.14s recent_step=4.11s eta=47:10:03 +2026-09-20 16:50:12 | INFO | train rank=0 | step=9030/50000 progress=18.1% loss=0.398450 lr=9.540889e-05 lr_backbone=9.540889e-05 elapsed=10:23:35 avg_step=4.14s recent_step=4.10s eta=47:09:19 +2026-09-20 16:50:53 | INFO | train rank=0 | step=9040/50000 progress=18.1% loss=0.339174 lr=9.539504e-05 lr_backbone=9.539504e-05 elapsed=10:24:16 avg_step=4.14s recent_step=4.09s eta=47:08:35 +2026-09-20 16:51:34 | INFO | train rank=0 | step=9050/50000 progress=18.1% loss=0.308857 lr=9.538116e-05 lr_backbone=9.538116e-05 elapsed=10:24:57 avg_step=4.14s recent_step=4.05s eta=47:07:50 +2026-09-20 16:52:15 | INFO | train rank=0 | step=9060/50000 progress=18.1% loss=0.409156 lr=9.536727e-05 lr_backbone=9.536727e-05 elapsed=10:25:38 avg_step=4.14s recent_step=4.10s eta=47:07:07 +2026-09-20 16:52:56 | INFO | train rank=0 | step=9070/50000 progress=18.1% loss=0.401351 lr=9.535336e-05 lr_backbone=9.535336e-05 elapsed=10:26:19 avg_step=4.14s recent_step=4.10s eta=47:06:23 +2026-09-20 16:53:36 | INFO | train rank=0 | step=9080/50000 progress=18.2% loss=0.367982 lr=9.533943e-05 lr_backbone=9.533943e-05 elapsed=10:27:00 avg_step=4.14s recent_step=4.08s eta=47:05:39 +2026-09-20 16:54:17 | INFO | train rank=0 | step=9090/50000 progress=18.2% loss=0.418827 lr=9.532548e-05 lr_backbone=9.532548e-05 elapsed=10:27:41 avg_step=4.14s recent_step=4.10s eta=47:04:56 +2026-09-20 16:54:58 | INFO | train rank=0 | step=9100/50000 progress=18.2% loss=0.401569 lr=9.531151e-05 lr_backbone=9.531151e-05 elapsed=10:28:22 avg_step=4.14s recent_step=4.10s eta=47:04:12 +2026-09-20 16:55:39 | INFO | train rank=0 | step=9110/50000 progress=18.2% loss=0.370323 lr=9.529752e-05 lr_backbone=9.529752e-05 elapsed=10:29:02 avg_step=4.14s recent_step=4.06s eta=47:03:27 +2026-09-20 16:56:20 | INFO | train rank=0 | step=9120/50000 progress=18.2% loss=0.360635 lr=9.528350e-05 lr_backbone=9.528350e-05 elapsed=10:29:43 avg_step=4.14s recent_step=4.10s eta=47:02:44 +2026-09-20 16:57:01 | INFO | train rank=0 | step=9130/50000 progress=18.3% loss=0.407910 lr=9.526947e-05 lr_backbone=9.526947e-05 elapsed=10:30:24 avg_step=4.14s recent_step=4.09s eta=47:02:00 +2026-09-20 16:57:42 | INFO | train rank=0 | step=9140/50000 progress=18.3% loss=0.430431 lr=9.525542e-05 lr_backbone=9.525542e-05 elapsed=10:31:05 avg_step=4.14s recent_step=4.07s eta=47:01:16 +2026-09-20 16:58:23 | INFO | train rank=0 | step=9150/50000 progress=18.3% loss=0.407465 lr=9.524135e-05 lr_backbone=9.524135e-05 elapsed=10:31:46 avg_step=4.14s recent_step=4.11s eta=47:00:33 +2026-09-20 16:59:04 | INFO | train rank=0 | step=9160/50000 progress=18.3% loss=0.340952 lr=9.522726e-05 lr_backbone=9.522726e-05 elapsed=10:32:27 avg_step=4.14s recent_step=4.09s eta=46:59:49 +2026-09-20 16:59:44 | INFO | train rank=0 | step=9170/50000 progress=18.3% loss=0.342418 lr=9.521315e-05 lr_backbone=9.521315e-05 elapsed=10:33:08 avg_step=4.14s recent_step=4.08s eta=46:59:05 +2026-09-20 17:00:25 | INFO | train rank=0 | step=9180/50000 progress=18.4% loss=0.378936 lr=9.519902e-05 lr_backbone=9.519902e-05 elapsed=10:33:49 avg_step=4.14s recent_step=4.09s eta=46:58:21 +2026-09-20 17:01:07 | INFO | train rank=0 | step=9190/50000 progress=18.4% loss=0.409313 lr=9.518487e-05 lr_backbone=9.518487e-05 elapsed=10:34:30 avg_step=4.14s recent_step=4.12s eta=46:57:38 +2026-09-20 17:01:48 | INFO | train rank=0 | step=9200/50000 progress=18.4% loss=0.340483 lr=9.517070e-05 lr_backbone=9.517070e-05 elapsed=10:35:11 avg_step=4.14s recent_step=4.12s eta=46:56:56 +2026-09-20 17:02:29 | INFO | train rank=0 | step=9210/50000 progress=18.4% loss=0.333063 lr=9.515652e-05 lr_backbone=9.515652e-05 elapsed=10:35:52 avg_step=4.14s recent_step=4.10s eta=46:56:12 +2026-09-20 17:03:10 | INFO | train rank=0 | step=9220/50000 progress=18.4% loss=0.389894 lr=9.514231e-05 lr_backbone=9.514231e-05 elapsed=10:36:33 avg_step=4.14s recent_step=4.10s eta=46:55:29 +2026-09-20 17:03:50 | INFO | train rank=0 | step=9230/50000 progress=18.5% loss=0.390763 lr=9.512808e-05 lr_backbone=9.512808e-05 elapsed=10:37:14 avg_step=4.14s recent_step=4.07s eta=46:54:44 +2026-09-20 17:04:31 | INFO | train rank=0 | step=9240/50000 progress=18.5% loss=0.348050 lr=9.511383e-05 lr_backbone=9.511383e-05 elapsed=10:37:55 avg_step=4.14s recent_step=4.10s eta=46:54:01 +2026-09-20 17:05:13 | INFO | train rank=0 | step=9250/50000 progress=18.5% loss=0.396191 lr=9.509956e-05 lr_backbone=9.509956e-05 elapsed=10:38:36 avg_step=4.14s recent_step=4.12s eta=46:53:19 +2026-09-20 17:05:54 | INFO | train rank=0 | step=9260/50000 progress=18.5% loss=0.435164 lr=9.508527e-05 lr_backbone=9.508527e-05 elapsed=10:39:17 avg_step=4.14s recent_step=4.13s eta=46:52:37 +2026-09-20 17:06:35 | INFO | train rank=0 | step=9270/50000 progress=18.5% loss=0.371415 lr=9.507097e-05 lr_backbone=9.507097e-05 elapsed=10:39:58 avg_step=4.14s recent_step=4.09s eta=46:51:53 +2026-09-20 17:07:16 | INFO | train rank=0 | step=9280/50000 progress=18.6% loss=0.335483 lr=9.505664e-05 lr_backbone=9.505664e-05 elapsed=10:40:39 avg_step=4.14s recent_step=4.10s eta=46:51:10 +2026-09-20 17:07:57 | INFO | train rank=0 | step=9290/50000 progress=18.6% loss=0.408190 lr=9.504229e-05 lr_backbone=9.504229e-05 elapsed=10:41:20 avg_step=4.14s recent_step=4.08s eta=46:50:26 +2026-09-20 17:08:38 | INFO | train rank=0 | step=9300/50000 progress=18.6% loss=0.357704 lr=9.502793e-05 lr_backbone=9.502793e-05 elapsed=10:42:01 avg_step=4.14s recent_step=4.11s eta=46:49:43 +2026-09-20 17:09:19 | INFO | train rank=0 | step=9310/50000 progress=18.6% loss=0.392771 lr=9.501354e-05 lr_backbone=9.501354e-05 elapsed=10:42:42 avg_step=4.14s recent_step=4.09s eta=46:48:59 +2026-09-20 17:10:00 | INFO | train rank=0 | step=9320/50000 progress=18.6% loss=0.414229 lr=9.499913e-05 lr_backbone=9.499913e-05 elapsed=10:43:23 avg_step=4.14s recent_step=4.12s eta=46:48:17 +2026-09-20 17:10:41 | INFO | train rank=0 | step=9330/50000 progress=18.7% loss=0.393946 lr=9.498471e-05 lr_backbone=9.498471e-05 elapsed=10:44:04 avg_step=4.14s recent_step=4.12s eta=46:47:34 +2026-09-20 17:11:22 | INFO | train rank=0 | step=9340/50000 progress=18.7% loss=0.353503 lr=9.497026e-05 lr_backbone=9.497026e-05 elapsed=10:44:45 avg_step=4.14s recent_step=4.11s eta=46:46:52 +2026-09-20 17:12:03 | INFO | train rank=0 | step=9350/50000 progress=18.7% loss=0.375132 lr=9.495580e-05 lr_backbone=9.495580e-05 elapsed=10:45:26 avg_step=4.14s recent_step=4.08s eta=46:46:08 +2026-09-20 17:12:44 | INFO | train rank=0 | step=9360/50000 progress=18.7% loss=0.324622 lr=9.494131e-05 lr_backbone=9.494131e-05 elapsed=10:46:08 avg_step=4.14s recent_step=4.13s eta=46:45:26 +2026-09-20 17:13:25 | INFO | train rank=0 | step=9370/50000 progress=18.7% loss=0.338648 lr=9.492681e-05 lr_backbone=9.492681e-05 elapsed=10:46:49 avg_step=4.14s recent_step=4.10s eta=46:44:43 +2026-09-20 17:14:06 | INFO | train rank=0 | step=9380/50000 progress=18.8% loss=0.388566 lr=9.491228e-05 lr_backbone=9.491228e-05 elapsed=10:47:30 avg_step=4.14s recent_step=4.12s eta=46:44:00 +2026-09-20 17:14:48 | INFO | train rank=0 | step=9390/50000 progress=18.8% loss=0.416183 lr=9.489774e-05 lr_backbone=9.489774e-05 elapsed=10:48:11 avg_step=4.14s recent_step=4.13s eta=46:43:18 +2026-09-20 17:15:29 | INFO | train rank=0 | step=9400/50000 progress=18.8% loss=0.397868 lr=9.488318e-05 lr_backbone=9.488318e-05 elapsed=10:48:52 avg_step=4.14s recent_step=4.10s eta=46:42:35 +2026-09-20 17:16:09 | INFO | train rank=0 | step=9410/50000 progress=18.8% loss=0.414657 lr=9.486859e-05 lr_backbone=9.486859e-05 elapsed=10:49:33 avg_step=4.14s recent_step=4.07s eta=46:41:50 +2026-09-20 17:16:51 | INFO | train rank=0 | step=9420/50000 progress=18.8% loss=0.368855 lr=9.485399e-05 lr_backbone=9.485399e-05 elapsed=10:50:14 avg_step=4.14s recent_step=4.12s eta=46:41:08 +2026-09-20 17:17:32 | INFO | train rank=0 | step=9430/50000 progress=18.9% loss=0.327011 lr=9.483937e-05 lr_backbone=9.483937e-05 elapsed=10:50:55 avg_step=4.14s recent_step=4.10s eta=46:40:25 +2026-09-20 17:18:12 | INFO | train rank=0 | step=9440/50000 progress=18.9% loss=0.333049 lr=9.482473e-05 lr_backbone=9.482473e-05 elapsed=10:51:36 avg_step=4.14s recent_step=4.09s eta=46:39:41 +2026-09-20 17:18:53 | INFO | train rank=0 | step=9450/50000 progress=18.9% loss=0.366732 lr=9.481007e-05 lr_backbone=9.481007e-05 elapsed=10:52:17 avg_step=4.14s recent_step=4.09s eta=46:38:58 +2026-09-20 17:19:35 | INFO | train rank=0 | step=9460/50000 progress=18.9% loss=0.427142 lr=9.479539e-05 lr_backbone=9.479539e-05 elapsed=10:52:58 avg_step=4.14s recent_step=4.12s eta=46:38:15 +2026-09-20 17:20:15 | INFO | train rank=0 | step=9470/50000 progress=18.9% loss=0.364760 lr=9.478069e-05 lr_backbone=9.478069e-05 elapsed=10:53:39 avg_step=4.14s recent_step=4.07s eta=46:37:31 +2026-09-20 17:20:56 | INFO | train rank=0 | step=9480/50000 progress=19.0% loss=0.453579 lr=9.476597e-05 lr_backbone=9.476597e-05 elapsed=10:54:20 avg_step=4.14s recent_step=4.12s eta=46:36:48 +2026-09-20 17:21:38 | INFO | train rank=0 | step=9490/50000 progress=19.0% loss=0.381363 lr=9.475123e-05 lr_backbone=9.475123e-05 elapsed=10:55:01 avg_step=4.14s recent_step=4.12s eta=46:36:06 +2026-09-20 17:22:18 | INFO | train rank=0 | step=9500/50000 progress=19.0% loss=0.372313 lr=9.473647e-05 lr_backbone=9.473647e-05 elapsed=10:55:42 avg_step=4.14s recent_step=4.08s eta=46:35:22 +2026-09-20 17:22:59 | INFO | train rank=0 | step=9510/50000 progress=19.0% loss=0.316409 lr=9.472169e-05 lr_backbone=9.472169e-05 elapsed=10:56:23 avg_step=4.14s recent_step=4.10s eta=46:34:39 +2026-09-20 17:23:41 | INFO | train rank=0 | step=9520/50000 progress=19.0% loss=0.382274 lr=9.470689e-05 lr_backbone=9.470689e-05 elapsed=10:57:04 avg_step=4.14s recent_step=4.12s eta=46:33:57 +2026-09-20 17:24:21 | INFO | train rank=0 | step=9530/50000 progress=19.1% loss=0.407317 lr=9.469207e-05 lr_backbone=9.469207e-05 elapsed=10:57:45 avg_step=4.14s recent_step=4.07s eta=46:33:12 +2026-09-20 17:25:02 | INFO | train rank=0 | step=9540/50000 progress=19.1% loss=0.354302 lr=9.467723e-05 lr_backbone=9.467723e-05 elapsed=10:58:26 avg_step=4.14s recent_step=4.11s eta=46:32:29 +2026-09-20 17:25:44 | INFO | train rank=0 | step=9550/50000 progress=19.1% loss=0.332541 lr=9.466238e-05 lr_backbone=9.466238e-05 elapsed=10:59:07 avg_step=4.14s recent_step=4.11s eta=46:31:47 +2026-09-20 17:26:24 | INFO | train rank=0 | step=9560/50000 progress=19.1% loss=0.366844 lr=9.464750e-05 lr_backbone=9.464750e-05 elapsed=10:59:48 avg_step=4.14s recent_step=4.08s eta=46:31:03 +2026-09-20 17:27:05 | INFO | train rank=0 | step=9570/50000 progress=19.1% loss=0.397942 lr=9.463260e-05 lr_backbone=9.463260e-05 elapsed=11:00:29 avg_step=4.14s recent_step=4.11s eta=46:30:20 +2026-09-20 17:27:47 | INFO | train rank=0 | step=9580/50000 progress=19.2% loss=0.410096 lr=9.461769e-05 lr_backbone=9.461769e-05 elapsed=11:01:10 avg_step=4.14s recent_step=4.12s eta=46:29:38 +2026-09-20 17:28:27 | INFO | train rank=0 | step=9590/50000 progress=19.2% loss=0.415292 lr=9.460275e-05 lr_backbone=9.460275e-05 elapsed=11:01:51 avg_step=4.14s recent_step=4.07s eta=46:28:53 +2026-09-20 17:29:08 | INFO | train rank=0 | step=9600/50000 progress=19.2% loss=0.428749 lr=9.458780e-05 lr_backbone=9.458780e-05 elapsed=11:02:32 avg_step=4.14s recent_step=4.10s eta=46:28:10 +2026-09-20 17:29:50 | INFO | train rank=0 | step=9610/50000 progress=19.2% loss=0.352573 lr=9.457282e-05 lr_backbone=9.457282e-05 elapsed=11:03:13 avg_step=4.14s recent_step=4.11s eta=46:27:28 +2026-09-20 17:30:30 | INFO | train rank=0 | step=9620/50000 progress=19.2% loss=0.361583 lr=9.455783e-05 lr_backbone=9.455783e-05 elapsed=11:03:54 avg_step=4.14s recent_step=4.09s eta=46:26:44 +2026-09-20 17:31:11 | INFO | train rank=0 | step=9630/50000 progress=19.3% loss=0.385959 lr=9.454282e-05 lr_backbone=9.454282e-05 elapsed=11:04:35 avg_step=4.14s recent_step=4.10s eta=46:26:01 +2026-09-20 17:31:53 | INFO | train rank=0 | step=9640/50000 progress=19.3% loss=0.418648 lr=9.452778e-05 lr_backbone=9.452778e-05 elapsed=11:05:16 avg_step=4.14s recent_step=4.11s eta=46:25:18 +2026-09-20 17:32:33 | INFO | train rank=0 | step=9650/50000 progress=19.3% loss=0.440867 lr=9.451273e-05 lr_backbone=9.451273e-05 elapsed=11:05:57 avg_step=4.14s recent_step=4.08s eta=46:24:34 +2026-09-20 17:33:14 | INFO | train rank=0 | step=9660/50000 progress=19.3% loss=0.399183 lr=9.449766e-05 lr_backbone=9.449766e-05 elapsed=11:06:38 avg_step=4.14s recent_step=4.11s eta=46:23:51 +2026-09-20 17:33:55 | INFO | train rank=0 | step=9670/50000 progress=19.3% loss=0.387896 lr=9.448257e-05 lr_backbone=9.448257e-05 elapsed=11:07:19 avg_step=4.14s recent_step=4.09s eta=46:23:08 +2026-09-20 17:34:37 | INFO | train rank=0 | step=9680/50000 progress=19.4% loss=0.425604 lr=9.446746e-05 lr_backbone=9.446746e-05 elapsed=11:08:00 avg_step=4.14s recent_step=4.12s eta=46:22:26 +2026-09-20 17:35:18 | INFO | train rank=0 | step=9690/50000 progress=19.4% loss=0.378855 lr=9.445233e-05 lr_backbone=9.445233e-05 elapsed=11:08:41 avg_step=4.14s recent_step=4.10s eta=46:21:43 +2026-09-20 17:35:59 | INFO | train rank=0 | step=9700/50000 progress=19.4% loss=0.318824 lr=9.443718e-05 lr_backbone=9.443718e-05 elapsed=11:09:22 avg_step=4.14s recent_step=4.13s eta=46:21:01 +2026-09-20 17:36:40 | INFO | train rank=0 | step=9710/50000 progress=19.4% loss=0.355487 lr=9.442201e-05 lr_backbone=9.442201e-05 elapsed=11:10:03 avg_step=4.14s recent_step=4.08s eta=46:20:17 +2026-09-20 17:37:21 | INFO | train rank=0 | step=9720/50000 progress=19.4% loss=0.349840 lr=9.440682e-05 lr_backbone=9.440682e-05 elapsed=11:10:44 avg_step=4.14s recent_step=4.12s eta=46:19:35 +2026-09-20 17:38:02 | INFO | train rank=0 | step=9730/50000 progress=19.5% loss=0.333484 lr=9.439161e-05 lr_backbone=9.439161e-05 elapsed=11:11:25 avg_step=4.14s recent_step=4.12s eta=46:18:52 +2026-09-20 17:38:43 | INFO | train rank=0 | step=9740/50000 progress=19.5% loss=0.406607 lr=9.437639e-05 lr_backbone=9.437639e-05 elapsed=11:12:06 avg_step=4.14s recent_step=4.10s eta=46:18:09 +2026-09-20 17:39:24 | INFO | train rank=0 | step=9750/50000 progress=19.5% loss=0.332782 lr=9.436114e-05 lr_backbone=9.436114e-05 elapsed=11:12:47 avg_step=4.14s recent_step=4.12s eta=46:17:27 +2026-09-20 17:40:05 | INFO | train rank=0 | step=9760/50000 progress=19.5% loss=0.386650 lr=9.434587e-05 lr_backbone=9.434587e-05 elapsed=11:13:29 avg_step=4.14s recent_step=4.12s eta=46:16:45 +2026-09-20 17:40:46 | INFO | train rank=0 | step=9770/50000 progress=19.5% loss=0.367589 lr=9.433059e-05 lr_backbone=9.433059e-05 elapsed=11:14:10 avg_step=4.14s recent_step=4.08s eta=46:16:01 +2026-09-20 17:41:27 | INFO | train rank=0 | step=9780/50000 progress=19.6% loss=0.356740 lr=9.431528e-05 lr_backbone=9.431528e-05 elapsed=11:14:51 avg_step=4.14s recent_step=4.10s eta=46:15:18 +2026-09-20 17:42:08 | INFO | train rank=0 | step=9790/50000 progress=19.6% loss=0.379976 lr=9.429996e-05 lr_backbone=9.429996e-05 elapsed=11:15:31 avg_step=4.14s recent_step=4.09s eta=46:14:34 +2026-09-20 17:42:49 | INFO | train rank=0 | step=9800/50000 progress=19.6% loss=0.354828 lr=9.428462e-05 lr_backbone=9.428462e-05 elapsed=11:16:12 avg_step=4.14s recent_step=4.09s eta=46:13:51 +2026-09-20 17:43:30 | INFO | train rank=0 | step=9810/50000 progress=19.6% loss=0.431411 lr=9.426925e-05 lr_backbone=9.426925e-05 elapsed=11:16:53 avg_step=4.14s recent_step=4.11s eta=46:13:08 +2026-09-20 17:44:11 | INFO | train rank=0 | step=9820/50000 progress=19.6% loss=0.362586 lr=9.425387e-05 lr_backbone=9.425387e-05 elapsed=11:17:34 avg_step=4.14s recent_step=4.10s eta=46:12:25 +2026-09-20 17:44:52 | INFO | train rank=0 | step=9830/50000 progress=19.7% loss=0.328981 lr=9.423847e-05 lr_backbone=9.423847e-05 elapsed=11:18:15 avg_step=4.14s recent_step=4.08s eta=46:11:41 +2026-09-20 17:45:33 | INFO | train rank=0 | step=9840/50000 progress=19.7% loss=0.382325 lr=9.422305e-05 lr_backbone=9.422305e-05 elapsed=11:18:56 avg_step=4.14s recent_step=4.09s eta=46:10:58 +2026-09-20 17:46:14 | INFO | train rank=0 | step=9850/50000 progress=19.7% loss=0.392145 lr=9.420761e-05 lr_backbone=9.420761e-05 elapsed=11:19:37 avg_step=4.14s recent_step=4.10s eta=46:10:15 +2026-09-20 17:46:55 | INFO | train rank=0 | step=9860/50000 progress=19.7% loss=0.420306 lr=9.419215e-05 lr_backbone=9.419215e-05 elapsed=11:20:18 avg_step=4.14s recent_step=4.10s eta=46:09:32 +2026-09-20 17:47:36 | INFO | train rank=0 | step=9870/50000 progress=19.7% loss=0.375567 lr=9.417667e-05 lr_backbone=9.417667e-05 elapsed=11:20:59 avg_step=4.14s recent_step=4.10s eta=46:08:49 +2026-09-20 17:48:17 | INFO | train rank=0 | step=9880/50000 progress=19.8% loss=0.373780 lr=9.416117e-05 lr_backbone=9.416117e-05 elapsed=11:21:40 avg_step=4.14s recent_step=4.11s eta=46:08:06 +2026-09-20 17:48:58 | INFO | train rank=0 | step=9890/50000 progress=19.8% loss=0.369705 lr=9.414565e-05 lr_backbone=9.414565e-05 elapsed=11:22:21 avg_step=4.14s recent_step=4.07s eta=46:07:22 +2026-09-20 17:49:39 | INFO | train rank=0 | step=9900/50000 progress=19.8% loss=0.355794 lr=9.413012e-05 lr_backbone=9.413012e-05 elapsed=11:23:02 avg_step=4.14s recent_step=4.11s eta=46:06:39 +2026-09-20 17:50:20 | INFO | train rank=0 | step=9910/50000 progress=19.8% loss=0.379192 lr=9.411456e-05 lr_backbone=9.411456e-05 elapsed=11:23:43 avg_step=4.14s recent_step=4.12s eta=46:05:57 +2026-09-20 17:51:01 | INFO | train rank=0 | step=9920/50000 progress=19.8% loss=0.358764 lr=9.409899e-05 lr_backbone=9.409899e-05 elapsed=11:24:24 avg_step=4.14s recent_step=4.12s eta=46:05:14 +2026-09-20 17:51:42 | INFO | train rank=0 | step=9930/50000 progress=19.9% loss=0.388500 lr=9.408339e-05 lr_backbone=9.408339e-05 elapsed=11:25:05 avg_step=4.14s recent_step=4.09s eta=46:04:31 +2026-09-20 17:52:23 | INFO | train rank=0 | step=9940/50000 progress=19.9% loss=0.342279 lr=9.406778e-05 lr_backbone=9.406778e-05 elapsed=11:25:46 avg_step=4.14s recent_step=4.12s eta=46:03:49 +2026-09-20 17:53:04 | INFO | train rank=0 | step=9950/50000 progress=19.9% loss=0.470849 lr=9.405214e-05 lr_backbone=9.405214e-05 elapsed=11:26:27 avg_step=4.14s recent_step=4.09s eta=46:03:06 +2026-09-20 17:53:45 | INFO | train rank=0 | step=9960/50000 progress=19.9% loss=0.347385 lr=9.403649e-05 lr_backbone=9.403649e-05 elapsed=11:27:09 avg_step=4.14s recent_step=4.13s eta=46:02:24 +2026-09-20 17:54:26 | INFO | train rank=0 | step=9970/50000 progress=19.9% loss=0.362641 lr=9.402082e-05 lr_backbone=9.402082e-05 elapsed=11:27:50 avg_step=4.14s recent_step=4.10s eta=46:01:41 +2026-09-20 17:55:07 | INFO | train rank=0 | step=9980/50000 progress=20.0% loss=0.395574 lr=9.400513e-05 lr_backbone=9.400513e-05 elapsed=11:28:31 avg_step=4.14s recent_step=4.12s eta=46:00:59 +2026-09-20 17:55:48 | INFO | train rank=0 | step=9990/50000 progress=20.0% loss=0.330102 lr=9.398942e-05 lr_backbone=9.398942e-05 elapsed=11:29:12 avg_step=4.14s recent_step=4.08s eta=46:00:15 +2026-09-20 17:56:29 | INFO | train rank=0 | step=10000/50000 progress=20.0% loss=0.336435 lr=9.397369e-05 lr_backbone=9.397369e-05 elapsed=11:29:53 avg_step=4.14s recent_step=4.10s eta=45:59:32 +2026-09-20 17:57:10 | INFO | train rank=0 | step=10010/50000 progress=20.0% loss=0.394750 lr=9.395794e-05 lr_backbone=9.395794e-05 elapsed=11:30:33 avg_step=4.14s recent_step=4.07s eta=45:58:48 +2026-09-20 17:57:51 | INFO | train rank=0 | step=10020/50000 progress=20.0% loss=0.431837 lr=9.394217e-05 lr_backbone=9.394217e-05 elapsed=11:31:14 avg_step=4.14s recent_step=4.10s eta=45:58:05 +2026-09-20 17:58:32 | INFO | train rank=0 | step=10030/50000 progress=20.1% loss=0.327494 lr=9.392638e-05 lr_backbone=9.392638e-05 elapsed=11:31:55 avg_step=4.14s recent_step=4.10s eta=45:57:22 +2026-09-20 17:59:13 | INFO | train rank=0 | step=10040/50000 progress=20.1% loss=0.365500 lr=9.391058e-05 lr_backbone=9.391058e-05 elapsed=11:32:37 avg_step=4.14s recent_step=4.11s eta=45:56:40 +2026-09-20 17:59:54 | INFO | train rank=0 | step=10050/50000 progress=20.1% loss=0.340653 lr=9.389475e-05 lr_backbone=9.389475e-05 elapsed=11:33:18 avg_step=4.14s recent_step=4.12s eta=45:55:57 +2026-09-20 18:00:35 | INFO | train rank=0 | step=10060/50000 progress=20.1% loss=0.405941 lr=9.387891e-05 lr_backbone=9.387891e-05 elapsed=11:33:59 avg_step=4.14s recent_step=4.10s eta=45:55:15 +2026-09-20 18:01:16 | INFO | train rank=0 | step=10070/50000 progress=20.1% loss=0.348818 lr=9.386304e-05 lr_backbone=9.386304e-05 elapsed=11:34:39 avg_step=4.14s recent_step=4.08s eta=45:54:31 +2026-09-20 18:01:57 | INFO | train rank=0 | step=10080/50000 progress=20.2% loss=0.385751 lr=9.384716e-05 lr_backbone=9.384716e-05 elapsed=11:35:21 avg_step=4.14s recent_step=4.10s eta=45:53:48 +2026-09-20 18:02:38 | INFO | train rank=0 | step=10090/50000 progress=20.2% loss=0.414773 lr=9.383126e-05 lr_backbone=9.383126e-05 elapsed=11:36:02 avg_step=4.14s recent_step=4.11s eta=45:53:05 +2026-09-20 18:03:19 | INFO | train rank=0 | step=10100/50000 progress=20.2% loss=0.327278 lr=9.381533e-05 lr_backbone=9.381533e-05 elapsed=11:36:43 avg_step=4.14s recent_step=4.10s eta=45:52:23 +2026-09-20 18:04:00 | INFO | train rank=0 | step=10110/50000 progress=20.2% loss=0.419730 lr=9.379939e-05 lr_backbone=9.379939e-05 elapsed=11:37:24 avg_step=4.14s recent_step=4.11s eta=45:51:40 +2026-09-20 18:04:41 | INFO | train rank=0 | step=10120/50000 progress=20.2% loss=0.417576 lr=9.378343e-05 lr_backbone=9.378343e-05 elapsed=11:38:05 avg_step=4.14s recent_step=4.08s eta=45:50:56 +2026-09-20 18:05:22 | INFO | train rank=0 | step=10130/50000 progress=20.3% loss=0.399101 lr=9.376745e-05 lr_backbone=9.376745e-05 elapsed=11:38:45 avg_step=4.14s recent_step=4.07s eta=45:50:12 +2026-09-20 18:06:03 | INFO | train rank=0 | step=10140/50000 progress=20.3% loss=0.315094 lr=9.375146e-05 lr_backbone=9.375146e-05 elapsed=11:39:26 avg_step=4.14s recent_step=4.10s eta=45:49:29 +2026-09-20 18:06:44 | INFO | train rank=0 | step=10150/50000 progress=20.3% loss=0.320875 lr=9.373544e-05 lr_backbone=9.373544e-05 elapsed=11:40:07 avg_step=4.14s recent_step=4.10s eta=45:48:47 +2026-09-20 18:07:25 | INFO | train rank=0 | step=10160/50000 progress=20.3% loss=0.321324 lr=9.371940e-05 lr_backbone=9.371940e-05 elapsed=11:40:48 avg_step=4.14s recent_step=4.11s eta=45:48:04 +2026-09-20 18:08:06 | INFO | train rank=0 | step=10170/50000 progress=20.3% loss=0.360734 lr=9.370335e-05 lr_backbone=9.370335e-05 elapsed=11:41:29 avg_step=4.14s recent_step=4.10s eta=45:47:21 +2026-09-20 18:08:47 | INFO | train rank=0 | step=10180/50000 progress=20.4% loss=0.367150 lr=9.368727e-05 lr_backbone=9.368727e-05 elapsed=11:42:10 avg_step=4.14s recent_step=4.08s eta=45:46:38 +2026-09-20 18:09:28 | INFO | train rank=0 | step=10190/50000 progress=20.4% loss=0.426147 lr=9.367118e-05 lr_backbone=9.367118e-05 elapsed=11:42:51 avg_step=4.14s recent_step=4.08s eta=45:45:54 +2026-09-20 18:10:09 | INFO | train rank=0 | step=10200/50000 progress=20.4% loss=0.361827 lr=9.365506e-05 lr_backbone=9.365506e-05 elapsed=11:43:32 avg_step=4.14s recent_step=4.10s eta=45:45:11 +2026-09-20 18:10:50 | INFO | train rank=0 | step=10210/50000 progress=20.4% loss=0.359774 lr=9.363893e-05 lr_backbone=9.363893e-05 elapsed=11:44:13 avg_step=4.14s recent_step=4.11s eta=45:44:29 +2026-09-20 18:11:31 | INFO | train rank=0 | step=10220/50000 progress=20.4% loss=0.323661 lr=9.362278e-05 lr_backbone=9.362278e-05 elapsed=11:44:54 avg_step=4.14s recent_step=4.10s eta=45:43:46 +2026-09-20 18:12:12 | INFO | train rank=0 | step=10230/50000 progress=20.5% loss=0.395744 lr=9.360661e-05 lr_backbone=9.360661e-05 elapsed=11:45:35 avg_step=4.14s recent_step=4.08s eta=45:43:02 +2026-09-20 18:12:53 | INFO | train rank=0 | step=10240/50000 progress=20.5% loss=0.304351 lr=9.359042e-05 lr_backbone=9.359042e-05 elapsed=11:46:16 avg_step=4.14s recent_step=4.10s eta=45:42:20 +2026-09-20 18:13:33 | INFO | train rank=0 | step=10250/50000 progress=20.5% loss=0.398757 lr=9.357421e-05 lr_backbone=9.357421e-05 elapsed=11:46:57 avg_step=4.14s recent_step=4.07s eta=45:41:36 +2026-09-20 18:14:14 | INFO | train rank=0 | step=10260/50000 progress=20.5% loss=0.364523 lr=9.355798e-05 lr_backbone=9.355798e-05 elapsed=11:47:38 avg_step=4.14s recent_step=4.10s eta=45:40:53 +2026-09-20 18:14:55 | INFO | train rank=0 | step=10270/50000 progress=20.5% loss=0.383737 lr=9.354174e-05 lr_backbone=9.354174e-05 elapsed=11:48:19 avg_step=4.14s recent_step=4.09s eta=45:40:10 +2026-09-20 18:15:36 | INFO | train rank=0 | step=10280/50000 progress=20.6% loss=0.395517 lr=9.352547e-05 lr_backbone=9.352547e-05 elapsed=11:49:00 avg_step=4.14s recent_step=4.09s eta=45:39:26 +2026-09-20 18:16:17 | INFO | train rank=0 | step=10290/50000 progress=20.6% loss=0.323400 lr=9.350919e-05 lr_backbone=9.350919e-05 elapsed=11:49:40 avg_step=4.14s recent_step=4.08s eta=45:38:43 +2026-09-20 18:16:58 | INFO | train rank=0 | step=10300/50000 progress=20.6% loss=0.348052 lr=9.349288e-05 lr_backbone=9.349288e-05 elapsed=11:50:21 avg_step=4.14s recent_step=4.10s eta=45:38:00 +2026-09-20 18:17:39 | INFO | train rank=0 | step=10310/50000 progress=20.6% loss=0.381283 lr=9.347656e-05 lr_backbone=9.347656e-05 elapsed=11:51:02 avg_step=4.14s recent_step=4.05s eta=45:37:15 +2026-09-20 18:18:20 | INFO | train rank=0 | step=10320/50000 progress=20.6% loss=0.405865 lr=9.346022e-05 lr_backbone=9.346022e-05 elapsed=11:51:43 avg_step=4.14s recent_step=4.10s eta=45:36:32 +2026-09-20 18:19:00 | INFO | train rank=0 | step=10330/50000 progress=20.7% loss=0.349550 lr=9.344386e-05 lr_backbone=9.344386e-05 elapsed=11:52:24 avg_step=4.14s recent_step=4.10s eta=45:35:49 +2026-09-20 18:19:42 | INFO | train rank=0 | step=10340/50000 progress=20.7% loss=0.306349 lr=9.342748e-05 lr_backbone=9.342748e-05 elapsed=11:53:05 avg_step=4.14s recent_step=4.11s eta=45:35:07 +2026-09-20 18:20:23 | INFO | train rank=0 | step=10350/50000 progress=20.7% loss=0.397256 lr=9.341108e-05 lr_backbone=9.341108e-05 elapsed=11:53:46 avg_step=4.14s recent_step=4.11s eta=45:34:25 +2026-09-20 18:21:04 | INFO | train rank=0 | step=10360/50000 progress=20.7% loss=0.369158 lr=9.339466e-05 lr_backbone=9.339466e-05 elapsed=11:54:27 avg_step=4.14s recent_step=4.10s eta=45:33:42 +2026-09-20 18:21:44 | INFO | train rank=0 | step=10370/50000 progress=20.7% loss=0.425985 lr=9.337823e-05 lr_backbone=9.337823e-05 elapsed=11:55:08 avg_step=4.14s recent_step=4.05s eta=45:32:57 +2026-09-20 18:22:25 | INFO | train rank=0 | step=10380/50000 progress=20.8% loss=0.390794 lr=9.336177e-05 lr_backbone=9.336177e-05 elapsed=11:55:49 avg_step=4.14s recent_step=4.11s eta=45:32:15 +2026-09-20 18:23:06 | INFO | train rank=0 | step=10390/50000 progress=20.8% loss=0.361090 lr=9.334530e-05 lr_backbone=9.334530e-05 elapsed=11:56:30 avg_step=4.14s recent_step=4.09s eta=45:31:32 +2026-09-20 18:23:48 | INFO | train rank=0 | step=10400/50000 progress=20.8% loss=0.376177 lr=9.332880e-05 lr_backbone=9.332880e-05 elapsed=11:57:11 avg_step=4.14s recent_step=4.12s eta=45:30:50 +2026-09-20 18:24:29 | INFO | train rank=0 | step=10410/50000 progress=20.8% loss=0.339841 lr=9.331229e-05 lr_backbone=9.331229e-05 elapsed=11:57:52 avg_step=4.14s recent_step=4.12s eta=45:30:08 +2026-09-20 18:25:10 | INFO | train rank=0 | step=10420/50000 progress=20.8% loss=0.392670 lr=9.329576e-05 lr_backbone=9.329576e-05 elapsed=11:58:33 avg_step=4.14s recent_step=4.12s eta=45:29:26 +2026-09-20 18:25:51 | INFO | train rank=0 | step=10430/50000 progress=20.9% loss=0.376654 lr=9.327921e-05 lr_backbone=9.327921e-05 elapsed=11:59:14 avg_step=4.14s recent_step=4.07s eta=45:28:41 +2026-09-20 18:26:32 | INFO | train rank=0 | step=10440/50000 progress=20.9% loss=0.494426 lr=9.326264e-05 lr_backbone=9.326264e-05 elapsed=11:59:55 avg_step=4.14s recent_step=4.09s eta=45:27:58 +2026-09-20 18:27:13 | INFO | train rank=0 | step=10450/50000 progress=20.9% loss=0.310259 lr=9.324605e-05 lr_backbone=9.324605e-05 elapsed=12:00:36 avg_step=4.14s recent_step=4.10s eta=45:27:16 +2026-09-20 18:27:54 | INFO | train rank=0 | step=10460/50000 progress=20.9% loss=0.415335 lr=9.322944e-05 lr_backbone=9.322944e-05 elapsed=12:01:17 avg_step=4.14s recent_step=4.10s eta=45:26:33 +2026-09-20 18:28:35 | INFO | train rank=0 | step=10470/50000 progress=20.9% loss=0.381908 lr=9.321282e-05 lr_backbone=9.321282e-05 elapsed=12:01:58 avg_step=4.14s recent_step=4.11s eta=45:25:50 +2026-09-20 18:29:16 | INFO | train rank=0 | step=10480/50000 progress=21.0% loss=0.371553 lr=9.319617e-05 lr_backbone=9.319617e-05 elapsed=12:02:39 avg_step=4.14s recent_step=4.13s eta=45:25:09 +2026-09-20 18:29:57 | INFO | train rank=0 | step=10490/50000 progress=21.0% loss=0.372003 lr=9.317951e-05 lr_backbone=9.317951e-05 elapsed=12:03:20 avg_step=4.14s recent_step=4.07s eta=45:24:25 +2026-09-20 18:30:38 | INFO | train rank=0 | step=10500/50000 progress=21.0% loss=0.317962 lr=9.316282e-05 lr_backbone=9.316282e-05 elapsed=12:04:01 avg_step=4.14s recent_step=4.08s eta=45:23:42 +2026-09-20 18:31:19 | INFO | train rank=0 | step=10510/50000 progress=21.0% loss=0.308076 lr=9.314612e-05 lr_backbone=9.314612e-05 elapsed=12:04:42 avg_step=4.14s recent_step=4.12s eta=45:23:00 +2026-09-20 18:32:00 | INFO | train rank=0 | step=10520/50000 progress=21.0% loss=0.349469 lr=9.312940e-05 lr_backbone=9.312940e-05 elapsed=12:05:23 avg_step=4.14s recent_step=4.10s eta=45:22:17 +2026-09-20 18:32:41 | INFO | train rank=0 | step=10530/50000 progress=21.1% loss=0.401432 lr=9.311266e-05 lr_backbone=9.311266e-05 elapsed=12:06:04 avg_step=4.14s recent_step=4.09s eta=45:21:34 +2026-09-20 18:33:22 | INFO | train rank=0 | step=10540/50000 progress=21.1% loss=0.347908 lr=9.309590e-05 lr_backbone=9.309590e-05 elapsed=12:06:45 avg_step=4.14s recent_step=4.10s eta=45:20:51 +2026-09-20 18:34:02 | INFO | train rank=0 | step=10550/50000 progress=21.1% loss=0.317944 lr=9.307913e-05 lr_backbone=9.307913e-05 elapsed=12:07:26 avg_step=4.14s recent_step=4.08s eta=45:20:07 +2026-09-20 18:34:44 | INFO | train rank=0 | step=10560/50000 progress=21.1% loss=0.336927 lr=9.306233e-05 lr_backbone=9.306233e-05 elapsed=12:08:07 avg_step=4.14s recent_step=4.13s eta=45:19:26 +2026-09-20 18:35:24 | INFO | train rank=0 | step=10570/50000 progress=21.1% loss=0.394151 lr=9.304552e-05 lr_backbone=9.304552e-05 elapsed=12:08:48 avg_step=4.14s recent_step=4.07s eta=45:18:42 +2026-09-20 18:36:06 | INFO | train rank=0 | step=10580/50000 progress=21.2% loss=0.373978 lr=9.302868e-05 lr_backbone=9.302868e-05 elapsed=12:09:29 avg_step=4.14s recent_step=4.13s eta=45:18:00 +2026-09-20 18:36:47 | INFO | train rank=0 | step=10590/50000 progress=21.2% loss=0.463105 lr=9.301183e-05 lr_backbone=9.301183e-05 elapsed=12:10:10 avg_step=4.14s recent_step=4.12s eta=45:17:18 +2026-09-20 18:37:28 | INFO | train rank=0 | step=10600/50000 progress=21.2% loss=0.326111 lr=9.299496e-05 lr_backbone=9.299496e-05 elapsed=12:10:51 avg_step=4.14s recent_step=4.12s eta=45:16:36 +2026-09-20 18:38:09 | INFO | train rank=0 | step=10610/50000 progress=21.2% loss=0.370765 lr=9.297807e-05 lr_backbone=9.297807e-05 elapsed=12:11:32 avg_step=4.14s recent_step=4.07s eta=45:15:52 +2026-09-20 18:38:50 | INFO | train rank=0 | step=10620/50000 progress=21.2% loss=0.393032 lr=9.296116e-05 lr_backbone=9.296116e-05 elapsed=12:12:13 avg_step=4.14s recent_step=4.11s eta=45:15:10 +2026-09-20 18:39:31 | INFO | train rank=0 | step=10630/50000 progress=21.3% loss=0.379662 lr=9.294423e-05 lr_backbone=9.294423e-05 elapsed=12:12:54 avg_step=4.14s recent_step=4.08s eta=45:14:26 +2026-09-20 18:40:12 | INFO | train rank=0 | step=10640/50000 progress=21.3% loss=0.347307 lr=9.292728e-05 lr_backbone=9.292728e-05 elapsed=12:13:35 avg_step=4.14s recent_step=4.12s eta=45:13:44 +2026-09-20 18:40:53 | INFO | train rank=0 | step=10650/50000 progress=21.3% loss=0.388740 lr=9.291032e-05 lr_backbone=9.291032e-05 elapsed=12:14:17 avg_step=4.14s recent_step=4.15s eta=45:13:03 +2026-09-20 18:41:34 | INFO | train rank=0 | step=10660/50000 progress=21.3% loss=0.382338 lr=9.289334e-05 lr_backbone=9.289334e-05 elapsed=12:14:58 avg_step=4.14s recent_step=4.10s eta=45:12:20 +2026-09-20 18:42:15 | INFO | train rank=0 | step=10670/50000 progress=21.3% loss=0.395297 lr=9.287633e-05 lr_backbone=9.287633e-05 elapsed=12:15:38 avg_step=4.14s recent_step=4.07s eta=45:11:37 +2026-09-20 18:42:56 | INFO | train rank=0 | step=10680/50000 progress=21.4% loss=0.380917 lr=9.285931e-05 lr_backbone=9.285931e-05 elapsed=12:16:19 avg_step=4.14s recent_step=4.12s eta=45:10:55 +2026-09-20 18:43:37 | INFO | train rank=0 | step=10690/50000 progress=21.4% loss=0.382451 lr=9.284227e-05 lr_backbone=9.284227e-05 elapsed=12:17:00 avg_step=4.14s recent_step=4.09s eta=45:10:12 +2026-09-20 18:44:18 | INFO | train rank=0 | step=10700/50000 progress=21.4% loss=0.344924 lr=9.282521e-05 lr_backbone=9.282521e-05 elapsed=12:17:42 avg_step=4.14s recent_step=4.11s eta=45:09:29 +2026-09-20 18:44:59 | INFO | train rank=0 | step=10710/50000 progress=21.4% loss=0.419544 lr=9.280813e-05 lr_backbone=9.280813e-05 elapsed=12:18:23 avg_step=4.14s recent_step=4.10s eta=45:08:47 +2026-09-20 18:45:40 | INFO | train rank=0 | step=10720/50000 progress=21.4% loss=0.387099 lr=9.279104e-05 lr_backbone=9.279104e-05 elapsed=12:19:04 avg_step=4.14s recent_step=4.11s eta=45:08:04 +2026-09-20 18:46:21 | INFO | train rank=0 | step=10730/50000 progress=21.5% loss=0.412232 lr=9.277392e-05 lr_backbone=9.277392e-05 elapsed=12:19:44 avg_step=4.14s recent_step=4.08s eta=45:07:21 +2026-09-20 18:47:02 | INFO | train rank=0 | step=10740/50000 progress=21.5% loss=0.367250 lr=9.275679e-05 lr_backbone=9.275679e-05 elapsed=12:20:25 avg_step=4.14s recent_step=4.11s eta=45:06:38 +2026-09-20 18:47:43 | INFO | train rank=0 | step=10750/50000 progress=21.5% loss=0.445842 lr=9.273964e-05 lr_backbone=9.273964e-05 elapsed=12:21:07 avg_step=4.14s recent_step=4.12s eta=45:05:56 +2026-09-20 18:48:24 | INFO | train rank=0 | step=10760/50000 progress=21.5% loss=0.345847 lr=9.272246e-05 lr_backbone=9.272246e-05 elapsed=12:21:48 avg_step=4.14s recent_step=4.09s eta=45:05:13 +2026-09-20 18:49:05 | INFO | train rank=0 | step=10770/50000 progress=21.5% loss=0.408843 lr=9.270527e-05 lr_backbone=9.270527e-05 elapsed=12:22:29 avg_step=4.14s recent_step=4.12s eta=45:04:32 +2026-09-20 18:49:47 | INFO | train rank=0 | step=10780/50000 progress=21.6% loss=0.371378 lr=9.268807e-05 lr_backbone=9.268807e-05 elapsed=12:23:10 avg_step=4.14s recent_step=4.11s eta=45:03:49 +2026-09-20 18:50:27 | INFO | train rank=0 | step=10790/50000 progress=21.6% loss=0.388192 lr=9.267084e-05 lr_backbone=9.267084e-05 elapsed=12:23:51 avg_step=4.14s recent_step=4.08s eta=45:03:06 +2026-09-20 18:51:09 | INFO | train rank=0 | step=10800/50000 progress=21.6% loss=0.524157 lr=9.265359e-05 lr_backbone=9.265359e-05 elapsed=12:24:32 avg_step=4.14s recent_step=4.12s eta=45:02:24 +2026-09-20 18:51:50 | INFO | train rank=0 | step=10810/50000 progress=21.6% loss=0.332224 lr=9.263633e-05 lr_backbone=9.263633e-05 elapsed=12:25:13 avg_step=4.14s recent_step=4.10s eta=45:01:41 +2026-09-20 18:52:30 | INFO | train rank=0 | step=10820/50000 progress=21.6% loss=0.424790 lr=9.261904e-05 lr_backbone=9.261904e-05 elapsed=12:25:54 avg_step=4.14s recent_step=4.07s eta=45:00:58 +2026-09-20 18:53:11 | INFO | train rank=0 | step=10830/50000 progress=21.7% loss=0.467223 lr=9.260174e-05 lr_backbone=9.260174e-05 elapsed=12:26:35 avg_step=4.14s recent_step=4.10s eta=45:00:15 +2026-09-20 18:53:53 | INFO | train rank=0 | step=10840/50000 progress=21.7% loss=0.436758 lr=9.258442e-05 lr_backbone=9.258442e-05 elapsed=12:27:16 avg_step=4.14s recent_step=4.13s eta=44:59:33 +2026-09-20 18:54:33 | INFO | train rank=0 | step=10850/50000 progress=21.7% loss=0.467716 lr=9.256708e-05 lr_backbone=9.256708e-05 elapsed=12:27:57 avg_step=4.14s recent_step=4.07s eta=44:58:50 +2026-09-20 18:55:15 | INFO | train rank=0 | step=10860/50000 progress=21.7% loss=0.446047 lr=9.254972e-05 lr_backbone=9.254972e-05 elapsed=12:28:38 avg_step=4.14s recent_step=4.12s eta=44:58:08 +2026-09-20 18:55:56 | INFO | train rank=0 | step=10870/50000 progress=21.7% loss=0.333088 lr=9.253235e-05 lr_backbone=9.253235e-05 elapsed=12:29:19 avg_step=4.14s recent_step=4.11s eta=44:57:25 +2026-09-20 18:56:36 | INFO | train rank=0 | step=10880/50000 progress=21.8% loss=0.356695 lr=9.251495e-05 lr_backbone=9.251495e-05 elapsed=12:30:00 avg_step=4.14s recent_step=4.08s eta=44:56:42 +2026-09-20 18:57:17 | INFO | train rank=0 | step=10890/50000 progress=21.8% loss=0.360862 lr=9.249754e-05 lr_backbone=9.249754e-05 elapsed=12:30:41 avg_step=4.14s recent_step=4.10s eta=44:55:59 +2026-09-20 18:57:58 | INFO | train rank=0 | step=10900/50000 progress=21.8% loss=0.338524 lr=9.248011e-05 lr_backbone=9.248011e-05 elapsed=12:31:22 avg_step=4.14s recent_step=4.11s eta=44:55:16 +2026-09-20 18:58:39 | INFO | train rank=0 | step=10910/50000 progress=21.8% loss=0.334447 lr=9.246266e-05 lr_backbone=9.246266e-05 elapsed=12:32:02 avg_step=4.14s recent_step=4.07s eta=44:54:33 +2026-09-20 18:59:20 | INFO | train rank=0 | step=10920/50000 progress=21.8% loss=0.434134 lr=9.244519e-05 lr_backbone=9.244519e-05 elapsed=12:32:44 avg_step=4.14s recent_step=4.12s eta=44:53:51 +2026-09-20 19:00:02 | INFO | train rank=0 | step=10930/50000 progress=21.9% loss=0.369777 lr=9.242770e-05 lr_backbone=9.242770e-05 elapsed=12:33:25 avg_step=4.14s recent_step=4.12s eta=44:53:09 +2026-09-20 19:00:42 | INFO | train rank=0 | step=10940/50000 progress=21.9% loss=0.347512 lr=9.241019e-05 lr_backbone=9.241019e-05 elapsed=12:34:06 avg_step=4.14s recent_step=4.08s eta=44:52:26 +2026-09-20 19:01:23 | INFO | train rank=0 | step=10950/50000 progress=21.9% loss=0.369772 lr=9.239267e-05 lr_backbone=9.239267e-05 elapsed=12:34:47 avg_step=4.14s recent_step=4.10s eta=44:51:43 +2026-09-20 19:02:04 | INFO | train rank=0 | step=10960/50000 progress=21.9% loss=0.332930 lr=9.237512e-05 lr_backbone=9.237512e-05 elapsed=12:35:28 avg_step=4.14s recent_step=4.11s eta=44:51:01 +2026-09-20 19:02:45 | INFO | train rank=0 | step=10970/50000 progress=21.9% loss=0.372011 lr=9.235756e-05 lr_backbone=9.235756e-05 elapsed=12:36:09 avg_step=4.14s recent_step=4.09s eta=44:50:18 +2026-09-20 19:03:27 | INFO | train rank=0 | step=10980/50000 progress=22.0% loss=0.366463 lr=9.233998e-05 lr_backbone=9.233998e-05 elapsed=12:36:50 avg_step=4.14s recent_step=4.12s eta=44:49:36 +2026-09-20 19:04:08 | INFO | train rank=0 | step=10990/50000 progress=22.0% loss=0.369493 lr=9.232238e-05 lr_backbone=9.232238e-05 elapsed=12:37:31 avg_step=4.14s recent_step=4.11s eta=44:48:54 +2026-09-20 19:04:49 | INFO | train rank=0 | step=11000/50000 progress=22.0% loss=0.454542 lr=9.230476e-05 lr_backbone=9.230476e-05 elapsed=12:38:12 avg_step=4.14s recent_step=4.10s eta=44:48:11 +2026-09-20 19:05:30 | INFO | train rank=0 | step=11010/50000 progress=22.0% loss=0.342161 lr=9.228713e-05 lr_backbone=9.228713e-05 elapsed=12:38:53 avg_step=4.14s recent_step=4.10s eta=44:47:28 +2026-09-20 19:06:11 | INFO | train rank=0 | step=11020/50000 progress=22.0% loss=0.357121 lr=9.226947e-05 lr_backbone=9.226947e-05 elapsed=12:39:34 avg_step=4.14s recent_step=4.10s eta=44:46:45 +2026-09-20 19:06:51 | INFO | train rank=0 | step=11030/50000 progress=22.1% loss=0.401323 lr=9.225180e-05 lr_backbone=9.225180e-05 elapsed=12:40:15 avg_step=4.14s recent_step=4.07s eta=44:46:02 +2026-09-20 19:07:32 | INFO | train rank=0 | step=11040/50000 progress=22.1% loss=0.300338 lr=9.223411e-05 lr_backbone=9.223411e-05 elapsed=12:40:56 avg_step=4.14s recent_step=4.11s eta=44:45:20 +2026-09-20 19:08:14 | INFO | train rank=0 | step=11050/50000 progress=22.1% loss=0.364131 lr=9.221640e-05 lr_backbone=9.221640e-05 elapsed=12:41:37 avg_step=4.14s recent_step=4.12s eta=44:44:38 +2026-09-20 19:08:54 | INFO | train rank=0 | step=11060/50000 progress=22.1% loss=0.314049 lr=9.219867e-05 lr_backbone=9.219867e-05 elapsed=12:42:18 avg_step=4.14s recent_step=4.09s eta=44:43:55 +2026-09-20 19:09:35 | INFO | train rank=0 | step=11070/50000 progress=22.1% loss=0.400723 lr=9.218092e-05 lr_backbone=9.218092e-05 elapsed=12:42:59 avg_step=4.14s recent_step=4.10s eta=44:43:12 +2026-09-20 19:10:16 | INFO | train rank=0 | step=11080/50000 progress=22.2% loss=0.318736 lr=9.216315e-05 lr_backbone=9.216315e-05 elapsed=12:43:40 avg_step=4.14s recent_step=4.11s eta=44:42:30 +2026-09-20 19:10:57 | INFO | train rank=0 | step=11090/50000 progress=22.2% loss=0.379810 lr=9.214537e-05 lr_backbone=9.214537e-05 elapsed=12:44:20 avg_step=4.14s recent_step=4.06s eta=44:41:46 +2026-09-20 19:11:38 | INFO | train rank=0 | step=11100/50000 progress=22.2% loss=0.359265 lr=9.212757e-05 lr_backbone=9.212757e-05 elapsed=12:45:02 avg_step=4.14s recent_step=4.13s eta=44:41:04 +2026-09-20 19:12:19 | INFO | train rank=0 | step=11110/50000 progress=22.2% loss=0.365505 lr=9.210975e-05 lr_backbone=9.210975e-05 elapsed=12:45:43 avg_step=4.14s recent_step=4.11s eta=44:40:22 +2026-09-20 19:13:00 | INFO | train rank=0 | step=11120/50000 progress=22.2% loss=0.327849 lr=9.209191e-05 lr_backbone=9.209191e-05 elapsed=12:46:24 avg_step=4.14s recent_step=4.08s eta=44:39:39 +2026-09-20 19:13:41 | INFO | train rank=0 | step=11130/50000 progress=22.3% loss=0.394490 lr=9.207405e-05 lr_backbone=9.207405e-05 elapsed=12:47:05 avg_step=4.14s recent_step=4.11s eta=44:38:56 +2026-09-20 19:14:23 | INFO | train rank=0 | step=11140/50000 progress=22.3% loss=0.361830 lr=9.205617e-05 lr_backbone=9.205617e-05 elapsed=12:47:46 avg_step=4.14s recent_step=4.13s eta=44:38:15 +2026-09-20 19:15:03 | INFO | train rank=0 | step=11150/50000 progress=22.3% loss=0.317380 lr=9.203828e-05 lr_backbone=9.203828e-05 elapsed=12:48:27 avg_step=4.14s recent_step=4.07s eta=44:37:31 +2026-09-20 19:15:45 | INFO | train rank=0 | step=11160/50000 progress=22.3% loss=0.369676 lr=9.202037e-05 lr_backbone=9.202037e-05 elapsed=12:49:08 avg_step=4.14s recent_step=4.12s eta=44:36:49 +2026-09-20 19:16:26 | INFO | train rank=0 | step=11170/50000 progress=22.3% loss=0.392915 lr=9.200244e-05 lr_backbone=9.200244e-05 elapsed=12:49:49 avg_step=4.14s recent_step=4.11s eta=44:36:07 +2026-09-20 19:17:06 | INFO | train rank=0 | step=11180/50000 progress=22.4% loss=0.362672 lr=9.198449e-05 lr_backbone=9.198449e-05 elapsed=12:50:30 avg_step=4.14s recent_step=4.08s eta=44:35:24 +2026-09-20 19:17:47 | INFO | train rank=0 | step=11190/50000 progress=22.4% loss=0.411610 lr=9.196652e-05 lr_backbone=9.196652e-05 elapsed=12:51:11 avg_step=4.14s recent_step=4.10s eta=44:34:41 +2026-09-20 19:18:28 | INFO | train rank=0 | step=11200/50000 progress=22.4% loss=0.366266 lr=9.194853e-05 lr_backbone=9.194853e-05 elapsed=12:51:52 avg_step=4.14s recent_step=4.10s eta=44:33:59 +2026-09-20 19:19:09 | INFO | train rank=0 | step=11210/50000 progress=22.4% loss=0.371194 lr=9.193053e-05 lr_backbone=9.193053e-05 elapsed=12:52:33 avg_step=4.13s recent_step=4.08s eta=44:33:15 +2026-09-20 19:19:50 | INFO | train rank=0 | step=11220/50000 progress=22.4% loss=0.312823 lr=9.191250e-05 lr_backbone=9.191250e-05 elapsed=12:53:14 avg_step=4.13s recent_step=4.11s eta=44:32:33 +2026-09-20 19:20:32 | INFO | train rank=0 | step=11230/50000 progress=22.5% loss=0.378786 lr=9.189446e-05 lr_backbone=9.189446e-05 elapsed=12:53:55 avg_step=4.13s recent_step=4.11s eta=44:31:51 +2026-09-20 19:21:12 | INFO | train rank=0 | step=11240/50000 progress=22.5% loss=0.344439 lr=9.187640e-05 lr_backbone=9.187640e-05 elapsed=12:54:36 avg_step=4.13s recent_step=4.09s eta=44:31:08 +2026-09-20 19:21:53 | INFO | train rank=0 | step=11250/50000 progress=22.5% loss=0.335583 lr=9.185832e-05 lr_backbone=9.185832e-05 elapsed=12:55:17 avg_step=4.13s recent_step=4.08s eta=44:30:25 +2026-09-20 19:22:34 | INFO | train rank=0 | step=11260/50000 progress=22.5% loss=0.338727 lr=9.184023e-05 lr_backbone=9.184023e-05 elapsed=12:55:58 avg_step=4.13s recent_step=4.12s eta=44:29:43 +2026-09-20 19:23:15 | INFO | train rank=0 | step=11270/50000 progress=22.5% loss=0.433206 lr=9.182211e-05 lr_backbone=9.182211e-05 elapsed=12:56:39 avg_step=4.13s recent_step=4.08s eta=44:29:00 +2026-09-20 19:23:56 | INFO | train rank=0 | step=11280/50000 progress=22.6% loss=0.387171 lr=9.180398e-05 lr_backbone=9.180398e-05 elapsed=12:57:20 avg_step=4.13s recent_step=4.11s eta=44:28:17 +2026-09-20 19:24:37 | INFO | train rank=0 | step=11290/50000 progress=22.6% loss=0.399074 lr=9.178583e-05 lr_backbone=9.178583e-05 elapsed=12:58:01 avg_step=4.13s recent_step=4.10s eta=44:27:35 +2026-09-20 19:25:18 | INFO | train rank=0 | step=11300/50000 progress=22.6% loss=0.349843 lr=9.176766e-05 lr_backbone=9.176766e-05 elapsed=12:58:42 avg_step=4.13s recent_step=4.13s eta=44:26:53 +2026-09-20 19:25:59 | INFO | train rank=0 | step=11310/50000 progress=22.6% loss=0.338835 lr=9.174947e-05 lr_backbone=9.174947e-05 elapsed=12:59:23 avg_step=4.13s recent_step=4.10s eta=44:26:10 +2026-09-20 19:26:41 | INFO | train rank=0 | step=11320/50000 progress=22.6% loss=0.377082 lr=9.173127e-05 lr_backbone=9.173127e-05 elapsed=13:00:04 avg_step=4.13s recent_step=4.14s eta=44:25:29 +2026-09-20 19:27:22 | INFO | train rank=0 | step=11330/50000 progress=22.7% loss=0.489594 lr=9.171304e-05 lr_backbone=9.171304e-05 elapsed=13:00:45 avg_step=4.13s recent_step=4.08s eta=44:24:46 +2026-09-20 19:28:03 | INFO | train rank=0 | step=11340/50000 progress=22.7% loss=0.343558 lr=9.169480e-05 lr_backbone=9.169480e-05 elapsed=13:01:26 avg_step=4.13s recent_step=4.13s eta=44:24:05 +2026-09-20 19:28:44 | INFO | train rank=0 | step=11350/50000 progress=22.7% loss=0.330290 lr=9.167654e-05 lr_backbone=9.167654e-05 elapsed=13:02:08 avg_step=4.13s recent_step=4.15s eta=44:23:24 +2026-09-20 19:29:25 | INFO | train rank=0 | step=11360/50000 progress=22.7% loss=0.342104 lr=9.165826e-05 lr_backbone=9.165826e-05 elapsed=13:02:48 avg_step=4.13s recent_step=4.07s eta=44:22:40 +2026-09-20 19:30:06 | INFO | train rank=0 | step=11370/50000 progress=22.7% loss=0.370396 lr=9.163996e-05 lr_backbone=9.163996e-05 elapsed=13:03:30 avg_step=4.13s recent_step=4.11s eta=44:21:58 +2026-09-20 19:30:47 | INFO | train rank=0 | step=11380/50000 progress=22.8% loss=0.354906 lr=9.162165e-05 lr_backbone=9.162165e-05 elapsed=13:04:11 avg_step=4.13s recent_step=4.11s eta=44:21:16 +2026-09-20 19:31:28 | INFO | train rank=0 | step=11390/50000 progress=22.8% loss=0.353181 lr=9.160331e-05 lr_backbone=9.160331e-05 elapsed=13:04:51 avg_step=4.13s recent_step=4.06s eta=44:20:32 +2026-09-20 19:32:09 | INFO | train rank=0 | step=11400/50000 progress=22.8% loss=0.412917 lr=9.158496e-05 lr_backbone=9.158496e-05 elapsed=13:05:32 avg_step=4.13s recent_step=4.10s eta=44:19:49 +2026-09-20 19:32:50 | INFO | train rank=0 | step=11410/50000 progress=22.8% loss=0.345159 lr=9.156659e-05 lr_backbone=9.156659e-05 elapsed=13:06:13 avg_step=4.13s recent_step=4.12s eta=44:19:08 +2026-09-20 19:33:31 | INFO | train rank=0 | step=11420/50000 progress=22.8% loss=0.352298 lr=9.154820e-05 lr_backbone=9.154820e-05 elapsed=13:06:54 avg_step=4.13s recent_step=4.10s eta=44:18:25 +2026-09-20 19:34:12 | INFO | train rank=0 | step=11430/50000 progress=22.9% loss=0.361516 lr=9.152979e-05 lr_backbone=9.152979e-05 elapsed=13:07:35 avg_step=4.13s recent_step=4.08s eta=44:17:42 +2026-09-20 19:34:53 | INFO | train rank=0 | step=11440/50000 progress=22.9% loss=0.400418 lr=9.151137e-05 lr_backbone=9.151137e-05 elapsed=13:08:16 avg_step=4.13s recent_step=4.11s eta=44:17:00 +2026-09-20 19:35:34 | INFO | train rank=0 | step=11450/50000 progress=22.9% loss=0.343755 lr=9.149293e-05 lr_backbone=9.149293e-05 elapsed=13:08:57 avg_step=4.13s recent_step=4.07s eta=44:16:16 +2026-09-20 19:36:15 | INFO | train rank=0 | step=11460/50000 progress=22.9% loss=0.359903 lr=9.147447e-05 lr_backbone=9.147447e-05 elapsed=13:09:38 avg_step=4.13s recent_step=4.12s eta=44:15:34 +2026-09-20 19:36:56 | INFO | train rank=0 | step=11470/50000 progress=22.9% loss=0.388646 lr=9.145599e-05 lr_backbone=9.145599e-05 elapsed=13:10:19 avg_step=4.13s recent_step=4.12s eta=44:14:53 +2026-09-20 19:37:37 | INFO | train rank=0 | step=11480/50000 progress=23.0% loss=0.371934 lr=9.143749e-05 lr_backbone=9.143749e-05 elapsed=13:11:01 avg_step=4.13s recent_step=4.10s eta=44:14:10 +2026-09-20 19:38:18 | INFO | train rank=0 | step=11490/50000 progress=23.0% loss=0.362112 lr=9.141898e-05 lr_backbone=9.141898e-05 elapsed=13:11:41 avg_step=4.13s recent_step=4.09s eta=44:13:27 +2026-09-20 19:38:59 | INFO | train rank=0 | step=11500/50000 progress=23.0% loss=0.372151 lr=9.140044e-05 lr_backbone=9.140044e-05 elapsed=13:12:23 avg_step=4.13s recent_step=4.12s eta=44:12:45 +2026-09-20 19:39:40 | INFO | train rank=0 | step=11510/50000 progress=23.0% loss=0.366666 lr=9.138189e-05 lr_backbone=9.138189e-05 elapsed=13:13:03 avg_step=4.13s recent_step=4.07s eta=44:12:02 +2026-09-20 19:40:21 | INFO | train rank=0 | step=11520/50000 progress=23.0% loss=0.401154 lr=9.136332e-05 lr_backbone=9.136332e-05 elapsed=13:13:44 avg_step=4.13s recent_step=4.12s eta=44:11:20 +2026-09-20 19:41:02 | INFO | train rank=0 | step=11530/50000 progress=23.1% loss=0.427390 lr=9.134473e-05 lr_backbone=9.134473e-05 elapsed=13:14:26 avg_step=4.13s recent_step=4.11s eta=44:10:38 +2026-09-20 19:41:43 | INFO | train rank=0 | step=11540/50000 progress=23.1% loss=0.384572 lr=9.132613e-05 lr_backbone=9.132613e-05 elapsed=13:15:07 avg_step=4.13s recent_step=4.10s eta=44:09:55 +2026-09-20 19:42:24 | INFO | train rank=0 | step=11550/50000 progress=23.1% loss=0.357277 lr=9.130750e-05 lr_backbone=9.130750e-05 elapsed=13:15:47 avg_step=4.13s recent_step=4.08s eta=44:09:12 +2026-09-20 19:43:05 | INFO | train rank=0 | step=11560/50000 progress=23.1% loss=0.376237 lr=9.128886e-05 lr_backbone=9.128886e-05 elapsed=13:16:29 avg_step=4.13s recent_step=4.12s eta=44:08:31 +2026-09-20 19:43:46 | INFO | train rank=0 | step=11570/50000 progress=23.1% loss=0.314878 lr=9.127020e-05 lr_backbone=9.127020e-05 elapsed=13:17:09 avg_step=4.13s recent_step=4.08s eta=44:07:47 +2026-09-20 19:44:27 | INFO | train rank=0 | step=11580/50000 progress=23.2% loss=0.371064 lr=9.125152e-05 lr_backbone=9.125152e-05 elapsed=13:17:50 avg_step=4.13s recent_step=4.09s eta=44:07:05 +2026-09-20 19:45:08 | INFO | train rank=0 | step=11590/50000 progress=23.2% loss=0.300261 lr=9.123283e-05 lr_backbone=9.123283e-05 elapsed=13:18:31 avg_step=4.13s recent_step=4.10s eta=44:06:22 +2026-09-20 19:45:49 | INFO | train rank=0 | step=11600/50000 progress=23.2% loss=0.345380 lr=9.121411e-05 lr_backbone=9.121411e-05 elapsed=13:19:12 avg_step=4.13s recent_step=4.12s eta=44:05:40 +2026-09-20 19:46:30 | INFO | train rank=0 | step=11610/50000 progress=23.2% loss=0.373568 lr=9.119538e-05 lr_backbone=9.119538e-05 elapsed=13:19:53 avg_step=4.13s recent_step=4.10s eta=44:04:58 +2026-09-20 19:47:11 | INFO | train rank=0 | step=11620/50000 progress=23.2% loss=0.369446 lr=9.117663e-05 lr_backbone=9.117663e-05 elapsed=13:20:34 avg_step=4.13s recent_step=4.11s eta=44:04:15 +2026-09-20 19:47:52 | INFO | train rank=0 | step=11630/50000 progress=23.3% loss=0.331026 lr=9.115786e-05 lr_backbone=9.115786e-05 elapsed=13:21:15 avg_step=4.13s recent_step=4.08s eta=44:03:32 +2026-09-20 19:48:33 | INFO | train rank=0 | step=11640/50000 progress=23.3% loss=0.386791 lr=9.113908e-05 lr_backbone=9.113908e-05 elapsed=13:21:56 avg_step=4.13s recent_step=4.10s eta=44:02:50 +2026-09-20 19:49:14 | INFO | train rank=0 | step=11650/50000 progress=23.3% loss=0.382734 lr=9.112027e-05 lr_backbone=9.112027e-05 elapsed=13:22:37 avg_step=4.13s recent_step=4.11s eta=44:02:08 +2026-09-20 19:49:55 | INFO | train rank=0 | step=11660/50000 progress=23.3% loss=0.371079 lr=9.110145e-05 lr_backbone=9.110145e-05 elapsed=13:23:19 avg_step=4.13s recent_step=4.11s eta=44:01:26 +2026-09-20 19:50:36 | INFO | train rank=0 | step=11670/50000 progress=23.3% loss=0.413113 lr=9.108261e-05 lr_backbone=9.108261e-05 elapsed=13:24:00 avg_step=4.13s recent_step=4.10s eta=44:00:43 +2026-09-20 19:51:18 | INFO | train rank=0 | step=11680/50000 progress=23.4% loss=0.416686 lr=9.106375e-05 lr_backbone=9.106375e-05 elapsed=13:24:41 avg_step=4.13s recent_step=4.13s eta=44:00:02 +2026-09-20 19:51:59 | INFO | train rank=0 | step=11690/50000 progress=23.4% loss=0.324580 lr=9.104487e-05 lr_backbone=9.104487e-05 elapsed=13:25:22 avg_step=4.13s recent_step=4.10s eta=43:59:20 +2026-09-20 19:52:40 | INFO | train rank=0 | step=11700/50000 progress=23.4% loss=0.344493 lr=9.102598e-05 lr_backbone=9.102598e-05 elapsed=13:26:03 avg_step=4.13s recent_step=4.13s eta=43:58:38 +2026-09-20 19:53:21 | INFO | train rank=0 | step=11710/50000 progress=23.4% loss=0.438896 lr=9.100707e-05 lr_backbone=9.100707e-05 elapsed=13:26:44 avg_step=4.13s recent_step=4.11s eta=43:57:56 +2026-09-20 19:54:02 | INFO | train rank=0 | step=11720/50000 progress=23.4% loss=0.411547 lr=9.098814e-05 lr_backbone=9.098814e-05 elapsed=13:27:25 avg_step=4.13s recent_step=4.10s eta=43:57:14 +2026-09-20 19:54:43 | INFO | train rank=0 | step=11730/50000 progress=23.5% loss=0.310079 lr=9.096919e-05 lr_backbone=9.096919e-05 elapsed=13:28:06 avg_step=4.13s recent_step=4.07s eta=43:56:31 +2026-09-20 19:55:24 | INFO | train rank=0 | step=11740/50000 progress=23.5% loss=0.379350 lr=9.095022e-05 lr_backbone=9.095022e-05 elapsed=13:28:47 avg_step=4.13s recent_step=4.10s eta=43:55:48 +2026-09-20 19:56:05 | INFO | train rank=0 | step=11750/50000 progress=23.5% loss=0.387458 lr=9.093124e-05 lr_backbone=9.093124e-05 elapsed=13:29:28 avg_step=4.13s recent_step=4.08s eta=43:55:05 +2026-09-20 19:56:46 | INFO | train rank=0 | step=11760/50000 progress=23.5% loss=0.333640 lr=9.091224e-05 lr_backbone=9.091224e-05 elapsed=13:30:09 avg_step=4.13s recent_step=4.11s eta=43:54:23 +2026-09-20 19:57:27 | INFO | train rank=0 | step=11770/50000 progress=23.5% loss=0.434032 lr=9.089322e-05 lr_backbone=9.089322e-05 elapsed=13:30:50 avg_step=4.13s recent_step=4.10s eta=43:53:41 +2026-09-20 19:58:08 | INFO | train rank=0 | step=11780/50000 progress=23.6% loss=0.332185 lr=9.087418e-05 lr_backbone=9.087418e-05 elapsed=13:31:31 avg_step=4.13s recent_step=4.11s eta=43:52:59 +2026-09-20 19:58:49 | INFO | train rank=0 | step=11790/50000 progress=23.6% loss=0.333487 lr=9.085513e-05 lr_backbone=9.085513e-05 elapsed=13:32:12 avg_step=4.13s recent_step=4.12s eta=43:52:17 +2026-09-20 19:59:30 | INFO | train rank=0 | step=11800/50000 progress=23.6% loss=0.370426 lr=9.083605e-05 lr_backbone=9.083605e-05 elapsed=13:32:53 avg_step=4.13s recent_step=4.11s eta=43:51:35 +2026-09-20 20:00:11 | INFO | train rank=0 | step=11810/50000 progress=23.6% loss=0.389047 lr=9.081696e-05 lr_backbone=9.081696e-05 elapsed=13:33:34 avg_step=4.13s recent_step=4.08s eta=43:50:51 +2026-09-20 20:00:52 | INFO | train rank=0 | step=11820/50000 progress=23.6% loss=0.308052 lr=9.079785e-05 lr_backbone=9.079785e-05 elapsed=13:34:15 avg_step=4.13s recent_step=4.11s eta=43:50:09 +2026-09-20 20:01:33 | INFO | train rank=0 | step=11830/50000 progress=23.7% loss=0.372133 lr=9.077873e-05 lr_backbone=9.077873e-05 elapsed=13:34:56 avg_step=4.13s recent_step=4.12s eta=43:49:28 +2026-09-20 20:02:14 | INFO | train rank=0 | step=11840/50000 progress=23.7% loss=0.291506 lr=9.075958e-05 lr_backbone=9.075958e-05 elapsed=13:35:38 avg_step=4.13s recent_step=4.13s eta=43:48:46 +2026-09-20 20:02:56 | INFO | train rank=0 | step=11850/50000 progress=23.7% loss=0.361197 lr=9.074042e-05 lr_backbone=9.074042e-05 elapsed=13:36:19 avg_step=4.13s recent_step=4.11s eta=43:48:04 +2026-09-20 20:03:37 | INFO | train rank=0 | step=11860/50000 progress=23.7% loss=0.350658 lr=9.072124e-05 lr_backbone=9.072124e-05 elapsed=13:37:00 avg_step=4.13s recent_step=4.10s eta=43:47:22 +2026-09-20 20:04:17 | INFO | train rank=0 | step=11870/50000 progress=23.7% loss=0.387011 lr=9.070204e-05 lr_backbone=9.070204e-05 elapsed=13:37:41 avg_step=4.13s recent_step=4.07s eta=43:46:39 +2026-09-20 20:04:58 | INFO | train rank=0 | step=11880/50000 progress=23.8% loss=0.348486 lr=9.068283e-05 lr_backbone=9.068283e-05 elapsed=13:38:22 avg_step=4.13s recent_step=4.10s eta=43:45:56 +2026-09-20 20:05:40 | INFO | train rank=0 | step=11890/50000 progress=23.8% loss=0.444901 lr=9.066359e-05 lr_backbone=9.066359e-05 elapsed=13:39:03 avg_step=4.13s recent_step=4.12s eta=43:45:14 +2026-09-20 20:06:20 | INFO | train rank=0 | step=11900/50000 progress=23.8% loss=0.337692 lr=9.064434e-05 lr_backbone=9.064434e-05 elapsed=13:39:44 avg_step=4.13s recent_step=4.09s eta=43:44:32 +2026-09-20 20:07:02 | INFO | train rank=0 | step=11910/50000 progress=23.8% loss=0.324020 lr=9.062507e-05 lr_backbone=9.062507e-05 elapsed=13:40:25 avg_step=4.13s recent_step=4.12s eta=43:43:50 +2026-09-20 20:07:42 | INFO | train rank=0 | step=11920/50000 progress=23.8% loss=0.309812 lr=9.060578e-05 lr_backbone=9.060578e-05 elapsed=13:41:06 avg_step=4.13s recent_step=4.08s eta=43:43:07 +2026-09-20 20:08:23 | INFO | train rank=0 | step=11930/50000 progress=23.9% loss=0.387780 lr=9.058648e-05 lr_backbone=9.058648e-05 elapsed=13:41:47 avg_step=4.13s recent_step=4.08s eta=43:42:24 +2026-09-20 20:09:04 | INFO | train rank=0 | step=11940/50000 progress=23.9% loss=0.422779 lr=9.056716e-05 lr_backbone=9.056716e-05 elapsed=13:42:28 avg_step=4.13s recent_step=4.11s eta=43:41:42 +2026-09-20 20:09:45 | INFO | train rank=0 | step=11950/50000 progress=23.9% loss=0.394755 lr=9.054782e-05 lr_backbone=9.054782e-05 elapsed=13:43:09 avg_step=4.13s recent_step=4.10s eta=43:40:59 +2026-09-20 20:10:26 | INFO | train rank=0 | step=11960/50000 progress=23.9% loss=0.306692 lr=9.052846e-05 lr_backbone=9.052846e-05 elapsed=13:43:50 avg_step=4.13s recent_step=4.11s eta=43:40:17 +2026-09-20 20:11:07 | INFO | train rank=0 | step=11970/50000 progress=23.9% loss=0.420708 lr=9.050908e-05 lr_backbone=9.050908e-05 elapsed=13:44:31 avg_step=4.13s recent_step=4.10s eta=43:39:35 +2026-09-20 20:11:48 | INFO | train rank=0 | step=11980/50000 progress=24.0% loss=0.434749 lr=9.048969e-05 lr_backbone=9.048969e-05 elapsed=13:45:11 avg_step=4.13s recent_step=4.07s eta=43:38:52 +2026-09-20 20:12:29 | INFO | train rank=0 | step=11990/50000 progress=24.0% loss=0.454607 lr=9.047028e-05 lr_backbone=9.047028e-05 elapsed=13:45:52 avg_step=4.13s recent_step=4.07s eta=43:38:09 +2026-09-20 20:13:10 | INFO | train rank=0 | step=12000/50000 progress=24.0% loss=0.318520 lr=9.045085e-05 lr_backbone=9.045085e-05 elapsed=13:46:33 avg_step=4.13s recent_step=4.10s eta=43:37:26 +2026-09-20 20:13:51 | INFO | train rank=0 | step=12010/50000 progress=24.0% loss=0.388774 lr=9.043140e-05 lr_backbone=9.043140e-05 elapsed=13:47:14 avg_step=4.13s recent_step=4.10s eta=43:36:44 +2026-09-20 20:14:32 | INFO | train rank=0 | step=12020/50000 progress=24.0% loss=0.363262 lr=9.041194e-05 lr_backbone=9.041194e-05 elapsed=13:47:55 avg_step=4.13s recent_step=4.10s eta=43:36:01 +2026-09-20 20:15:13 | INFO | train rank=0 | step=12030/50000 progress=24.1% loss=0.387671 lr=9.039246e-05 lr_backbone=9.039246e-05 elapsed=13:48:36 avg_step=4.13s recent_step=4.11s eta=43:35:19 +2026-09-20 20:15:54 | INFO | train rank=0 | step=12040/50000 progress=24.1% loss=0.379247 lr=9.037296e-05 lr_backbone=9.037296e-05 elapsed=13:49:17 avg_step=4.13s recent_step=4.09s eta=43:34:37 +2026-09-20 20:16:35 | INFO | train rank=0 | step=12050/50000 progress=24.1% loss=0.343866 lr=9.035344e-05 lr_backbone=9.035344e-05 elapsed=13:49:58 avg_step=4.13s recent_step=4.08s eta=43:33:54 +2026-09-20 20:17:16 | INFO | train rank=0 | step=12060/50000 progress=24.1% loss=0.371999 lr=9.033391e-05 lr_backbone=9.033391e-05 elapsed=13:50:39 avg_step=4.13s recent_step=4.10s eta=43:33:11 +2026-09-20 20:17:57 | INFO | train rank=0 | step=12070/50000 progress=24.1% loss=0.394459 lr=9.031435e-05 lr_backbone=9.031435e-05 elapsed=13:51:20 avg_step=4.13s recent_step=4.10s eta=43:32:29 +2026-09-20 20:18:38 | INFO | train rank=0 | step=12080/50000 progress=24.2% loss=0.326861 lr=9.029478e-05 lr_backbone=9.029478e-05 elapsed=13:52:01 avg_step=4.13s recent_step=4.11s eta=43:31:47 +2026-09-20 20:19:19 | INFO | train rank=0 | step=12090/50000 progress=24.2% loss=0.402661 lr=9.027519e-05 lr_backbone=9.027519e-05 elapsed=13:52:42 avg_step=4.13s recent_step=4.11s eta=43:31:05 +2026-09-20 20:20:00 | INFO | train rank=0 | step=12100/50000 progress=24.2% loss=0.318476 lr=9.025559e-05 lr_backbone=9.025559e-05 elapsed=13:53:23 avg_step=4.13s recent_step=4.09s eta=43:30:22 +2026-09-20 20:20:40 | INFO | train rank=0 | step=12110/50000 progress=24.2% loss=0.438340 lr=9.023597e-05 lr_backbone=9.023597e-05 elapsed=13:54:04 avg_step=4.13s recent_step=4.05s eta=43:29:38 +2026-09-20 20:21:21 | INFO | train rank=0 | step=12120/50000 progress=24.2% loss=0.423513 lr=9.021633e-05 lr_backbone=9.021633e-05 elapsed=13:54:45 avg_step=4.13s recent_step=4.10s eta=43:28:56 +2026-09-20 20:22:02 | INFO | train rank=0 | step=12130/50000 progress=24.3% loss=0.325955 lr=9.019667e-05 lr_backbone=9.019667e-05 elapsed=13:55:26 avg_step=4.13s recent_step=4.09s eta=43:28:14 +2026-09-20 20:22:43 | INFO | train rank=0 | step=12140/50000 progress=24.3% loss=0.389500 lr=9.017699e-05 lr_backbone=9.017699e-05 elapsed=13:56:06 avg_step=4.13s recent_step=4.08s eta=43:27:31 +2026-09-20 20:23:24 | INFO | train rank=0 | step=12150/50000 progress=24.3% loss=0.421642 lr=9.015730e-05 lr_backbone=9.015730e-05 elapsed=13:56:47 avg_step=4.13s recent_step=4.08s eta=43:26:48 +2026-09-20 20:24:05 | INFO | train rank=0 | step=12160/50000 progress=24.3% loss=0.376887 lr=9.013759e-05 lr_backbone=9.013759e-05 elapsed=13:57:28 avg_step=4.13s recent_step=4.08s eta=43:26:05 +2026-09-20 20:24:45 | INFO | train rank=0 | step=12170/50000 progress=24.3% loss=0.359037 lr=9.011786e-05 lr_backbone=9.011786e-05 elapsed=13:58:09 avg_step=4.13s recent_step=4.07s eta=43:25:22 +2026-09-20 20:25:26 | INFO | train rank=0 | step=12180/50000 progress=24.4% loss=0.399869 lr=9.009811e-05 lr_backbone=9.009811e-05 elapsed=13:58:50 avg_step=4.13s recent_step=4.09s eta=43:24:39 +2026-09-20 20:26:07 | INFO | train rank=0 | step=12190/50000 progress=24.4% loss=0.438734 lr=9.007835e-05 lr_backbone=9.007835e-05 elapsed=13:59:31 avg_step=4.13s recent_step=4.11s eta=43:23:57 +2026-09-20 20:26:48 | INFO | train rank=0 | step=12200/50000 progress=24.4% loss=0.502709 lr=9.005857e-05 lr_backbone=9.005857e-05 elapsed=14:00:11 avg_step=4.13s recent_step=4.08s eta=43:23:14 +2026-09-20 20:27:29 | INFO | train rank=0 | step=12210/50000 progress=24.4% loss=0.390894 lr=9.003877e-05 lr_backbone=9.003877e-05 elapsed=14:00:52 avg_step=4.13s recent_step=4.08s eta=43:22:31 +2026-09-20 20:28:10 | INFO | train rank=0 | step=12220/50000 progress=24.4% loss=0.394830 lr=9.001895e-05 lr_backbone=9.001895e-05 elapsed=14:01:33 avg_step=4.13s recent_step=4.11s eta=43:21:49 +2026-09-20 20:28:51 | INFO | train rank=0 | step=12230/50000 progress=24.5% loss=0.351288 lr=8.999912e-05 lr_backbone=8.999912e-05 elapsed=14:02:14 avg_step=4.13s recent_step=4.07s eta=43:21:06 +2026-09-20 20:29:32 | INFO | train rank=0 | step=12240/50000 progress=24.5% loss=0.408359 lr=8.997927e-05 lr_backbone=8.997927e-05 elapsed=14:02:55 avg_step=4.13s recent_step=4.10s eta=43:20:23 +2026-09-20 20:30:13 | INFO | train rank=0 | step=12250/50000 progress=24.5% loss=0.334864 lr=8.995940e-05 lr_backbone=8.995940e-05 elapsed=14:03:36 avg_step=4.13s recent_step=4.11s eta=43:19:41 +2026-09-20 20:30:54 | INFO | train rank=0 | step=12260/50000 progress=24.5% loss=0.342559 lr=8.993951e-05 lr_backbone=8.993951e-05 elapsed=14:04:17 avg_step=4.13s recent_step=4.10s eta=43:18:59 +2026-09-20 20:31:35 | INFO | train rank=0 | step=12270/50000 progress=24.5% loss=0.289122 lr=8.991961e-05 lr_backbone=8.991961e-05 elapsed=14:04:58 avg_step=4.13s recent_step=4.11s eta=43:18:17 +2026-09-20 20:32:16 | INFO | train rank=0 | step=12280/50000 progress=24.6% loss=0.370287 lr=8.989969e-05 lr_backbone=8.989969e-05 elapsed=14:05:39 avg_step=4.13s recent_step=4.10s eta=43:17:35 +2026-09-20 20:32:56 | INFO | train rank=0 | step=12290/50000 progress=24.6% loss=0.359498 lr=8.987975e-05 lr_backbone=8.987975e-05 elapsed=14:06:20 avg_step=4.13s recent_step=4.05s eta=43:16:51 +2026-09-20 20:33:37 | INFO | train rank=0 | step=12300/50000 progress=24.6% loss=0.428435 lr=8.985979e-05 lr_backbone=8.985979e-05 elapsed=14:07:01 avg_step=4.13s recent_step=4.10s eta=43:16:09 +2026-09-20 20:34:18 | INFO | train rank=0 | step=12310/50000 progress=24.6% loss=0.327831 lr=8.983982e-05 lr_backbone=8.983982e-05 elapsed=14:07:42 avg_step=4.13s recent_step=4.10s eta=43:15:27 +2026-09-20 20:35:00 | INFO | train rank=0 | step=12320/50000 progress=24.6% loss=0.337422 lr=8.981983e-05 lr_backbone=8.981983e-05 elapsed=14:08:23 avg_step=4.13s recent_step=4.12s eta=43:14:45 +2026-09-20 20:35:41 | INFO | train rank=0 | step=12330/50000 progress=24.7% loss=0.339992 lr=8.979982e-05 lr_backbone=8.979982e-05 elapsed=14:09:04 avg_step=4.13s recent_step=4.11s eta=43:14:03 +2026-09-20 20:36:22 | INFO | train rank=0 | step=12340/50000 progress=24.7% loss=0.407654 lr=8.977980e-05 lr_backbone=8.977980e-05 elapsed=14:09:45 avg_step=4.13s recent_step=4.12s eta=43:13:21 +2026-09-20 20:37:03 | INFO | train rank=0 | step=12350/50000 progress=24.7% loss=0.357655 lr=8.975975e-05 lr_backbone=8.975975e-05 elapsed=14:10:26 avg_step=4.13s recent_step=4.06s eta=43:12:38 +2026-09-20 20:37:44 | INFO | train rank=0 | step=12360/50000 progress=24.7% loss=0.385521 lr=8.973969e-05 lr_backbone=8.973969e-05 elapsed=14:11:07 avg_step=4.13s recent_step=4.11s eta=43:11:56 +2026-09-20 20:38:25 | INFO | train rank=0 | step=12370/50000 progress=24.7% loss=0.410308 lr=8.971962e-05 lr_backbone=8.971962e-05 elapsed=14:11:48 avg_step=4.13s recent_step=4.10s eta=43:11:13 +2026-09-20 20:39:06 | INFO | train rank=0 | step=12380/50000 progress=24.8% loss=0.475273 lr=8.969952e-05 lr_backbone=8.969952e-05 elapsed=14:12:29 avg_step=4.13s recent_step=4.11s eta=43:10:31 +2026-09-20 20:39:47 | INFO | train rank=0 | step=12390/50000 progress=24.8% loss=0.376458 lr=8.967941e-05 lr_backbone=8.967941e-05 elapsed=14:13:10 avg_step=4.13s recent_step=4.10s eta=43:09:49 +2026-09-20 20:40:28 | INFO | train rank=0 | step=12400/50000 progress=24.8% loss=0.476762 lr=8.965928e-05 lr_backbone=8.965928e-05 elapsed=14:13:51 avg_step=4.13s recent_step=4.10s eta=43:09:07 +2026-09-20 20:41:09 | INFO | train rank=0 | step=12410/50000 progress=24.8% loss=0.429415 lr=8.963913e-05 lr_backbone=8.963913e-05 elapsed=14:14:32 avg_step=4.13s recent_step=4.08s eta=43:08:24 +2026-09-20 20:41:49 | INFO | train rank=0 | step=12420/50000 progress=24.8% loss=0.345380 lr=8.961897e-05 lr_backbone=8.961897e-05 elapsed=14:15:13 avg_step=4.13s recent_step=4.08s eta=43:07:41 +2026-09-20 20:42:31 | INFO | train rank=0 | step=12430/50000 progress=24.9% loss=0.345699 lr=8.959878e-05 lr_backbone=8.959878e-05 elapsed=14:15:54 avg_step=4.13s recent_step=4.11s eta=43:06:59 +2026-09-20 20:43:12 | INFO | train rank=0 | step=12440/50000 progress=24.9% loss=0.327544 lr=8.957858e-05 lr_backbone=8.957858e-05 elapsed=14:16:35 avg_step=4.13s recent_step=4.11s eta=43:06:17 +2026-09-20 20:43:53 | INFO | train rank=0 | step=12450/50000 progress=24.9% loss=0.393461 lr=8.955837e-05 lr_backbone=8.955837e-05 elapsed=14:17:16 avg_step=4.13s recent_step=4.12s eta=43:05:36 +2026-09-20 20:44:34 | INFO | train rank=0 | step=12460/50000 progress=24.9% loss=0.406675 lr=8.953813e-05 lr_backbone=8.953813e-05 elapsed=14:17:57 avg_step=4.13s recent_step=4.12s eta=43:04:54 +2026-09-20 20:45:15 | INFO | train rank=0 | step=12470/50000 progress=24.9% loss=0.357307 lr=8.951788e-05 lr_backbone=8.951788e-05 elapsed=14:18:38 avg_step=4.13s recent_step=4.08s eta=43:04:11 +2026-09-20 20:45:56 | INFO | train rank=0 | step=12480/50000 progress=25.0% loss=0.419717 lr=8.949761e-05 lr_backbone=8.949761e-05 elapsed=14:19:19 avg_step=4.13s recent_step=4.09s eta=43:03:29 +2026-09-20 20:46:37 | INFO | train rank=0 | step=12490/50000 progress=25.0% loss=0.429533 lr=8.947733e-05 lr_backbone=8.947733e-05 elapsed=14:20:00 avg_step=4.13s recent_step=4.10s eta=43:02:46 +2026-09-20 20:47:18 | INFO | train rank=0 | step=12500/50000 progress=25.0% loss=0.390216 lr=8.945703e-05 lr_backbone=8.945703e-05 elapsed=14:20:41 avg_step=4.13s recent_step=4.13s eta=43:02:05 +2026-09-20 20:47:59 | INFO | train rank=0 | step=12510/50000 progress=25.0% loss=0.316149 lr=8.943671e-05 lr_backbone=8.943671e-05 elapsed=14:21:23 avg_step=4.13s recent_step=4.13s eta=43:01:24 +2026-09-20 20:48:41 | INFO | train rank=0 | step=12520/50000 progress=25.0% loss=0.410198 lr=8.941637e-05 lr_backbone=8.941637e-05 elapsed=14:22:04 avg_step=4.13s recent_step=4.13s eta=43:00:42 +2026-09-20 20:49:21 | INFO | train rank=0 | step=12530/50000 progress=25.1% loss=0.332629 lr=8.939601e-05 lr_backbone=8.939601e-05 elapsed=14:22:45 avg_step=4.13s recent_step=4.07s eta=42:59:59 +2026-09-20 20:50:02 | INFO | train rank=0 | step=12540/50000 progress=25.1% loss=0.352774 lr=8.937564e-05 lr_backbone=8.937564e-05 elapsed=14:23:26 avg_step=4.13s recent_step=4.12s eta=42:59:17 +2026-09-20 20:50:43 | INFO | train rank=0 | step=12550/50000 progress=25.1% loss=0.332826 lr=8.935525e-05 lr_backbone=8.935525e-05 elapsed=14:24:07 avg_step=4.13s recent_step=4.08s eta=42:58:35 +2026-09-20 20:51:24 | INFO | train rank=0 | step=12560/50000 progress=25.1% loss=0.377913 lr=8.933485e-05 lr_backbone=8.933485e-05 elapsed=14:24:48 avg_step=4.13s recent_step=4.11s eta=42:57:53 +2026-09-20 20:52:05 | INFO | train rank=0 | step=12570/50000 progress=25.1% loss=0.353697 lr=8.931442e-05 lr_backbone=8.931442e-05 elapsed=14:25:29 avg_step=4.13s recent_step=4.11s eta=42:57:11 +2026-09-20 20:52:46 | INFO | train rank=0 | step=12580/50000 progress=25.2% loss=0.353297 lr=8.929398e-05 lr_backbone=8.929398e-05 elapsed=14:26:10 avg_step=4.13s recent_step=4.11s eta=42:56:29 +2026-09-20 20:53:27 | INFO | train rank=0 | step=12590/50000 progress=25.2% loss=0.386437 lr=8.927352e-05 lr_backbone=8.927352e-05 elapsed=14:26:51 avg_step=4.13s recent_step=4.07s eta=42:55:46 +2026-09-20 20:54:08 | INFO | train rank=0 | step=12600/50000 progress=25.2% loss=0.429393 lr=8.925305e-05 lr_backbone=8.925305e-05 elapsed=14:27:32 avg_step=4.13s recent_step=4.10s eta=42:55:03 +2026-09-20 20:54:49 | INFO | train rank=0 | step=12610/50000 progress=25.2% loss=0.428654 lr=8.923256e-05 lr_backbone=8.923256e-05 elapsed=14:28:12 avg_step=4.13s recent_step=4.08s eta=42:54:20 +2026-09-20 20:55:30 | INFO | train rank=0 | step=12620/50000 progress=25.2% loss=0.316108 lr=8.921205e-05 lr_backbone=8.921205e-05 elapsed=14:28:53 avg_step=4.13s recent_step=4.11s eta=42:53:39 +2026-09-20 20:56:11 | INFO | train rank=0 | step=12630/50000 progress=25.3% loss=0.408465 lr=8.919152e-05 lr_backbone=8.919152e-05 elapsed=14:29:35 avg_step=4.13s recent_step=4.12s eta=42:52:57 +2026-09-20 20:56:52 | INFO | train rank=0 | step=12640/50000 progress=25.3% loss=0.308101 lr=8.917098e-05 lr_backbone=8.917098e-05 elapsed=14:30:16 avg_step=4.13s recent_step=4.11s eta=42:52:15 +2026-09-20 20:57:33 | INFO | train rank=0 | step=12650/50000 progress=25.3% loss=0.297974 lr=8.915041e-05 lr_backbone=8.915041e-05 elapsed=14:30:57 avg_step=4.13s recent_step=4.07s eta=42:51:32 +2026-09-20 20:58:14 | INFO | train rank=0 | step=12660/50000 progress=25.3% loss=0.352944 lr=8.912984e-05 lr_backbone=8.912984e-05 elapsed=14:31:38 avg_step=4.13s recent_step=4.11s eta=42:50:50 +2026-09-20 20:58:55 | INFO | train rank=0 | step=12670/50000 progress=25.3% loss=0.356465 lr=8.910924e-05 lr_backbone=8.910924e-05 elapsed=14:32:19 avg_step=4.13s recent_step=4.11s eta=42:50:08 +2026-09-20 20:59:36 | INFO | train rank=0 | step=12680/50000 progress=25.4% loss=0.315797 lr=8.908863e-05 lr_backbone=8.908863e-05 elapsed=14:33:00 avg_step=4.13s recent_step=4.08s eta=42:49:25 +2026-09-20 21:00:18 | INFO | train rank=0 | step=12690/50000 progress=25.4% loss=0.369316 lr=8.906800e-05 lr_backbone=8.906800e-05 elapsed=14:33:41 avg_step=4.13s recent_step=4.13s eta=42:48:44 +2026-09-20 21:00:59 | INFO | train rank=0 | step=12700/50000 progress=25.4% loss=0.415550 lr=8.904735e-05 lr_backbone=8.904735e-05 elapsed=14:34:22 avg_step=4.13s recent_step=4.10s eta=42:48:02 +2026-09-20 21:01:39 | INFO | train rank=0 | step=12710/50000 progress=25.4% loss=0.346937 lr=8.902669e-05 lr_backbone=8.902669e-05 elapsed=14:35:03 avg_step=4.13s recent_step=4.07s eta=42:47:19 +2026-09-20 21:02:20 | INFO | train rank=0 | step=12720/50000 progress=25.4% loss=0.350621 lr=8.900601e-05 lr_backbone=8.900601e-05 elapsed=14:35:44 avg_step=4.13s recent_step=4.12s eta=42:46:37 +2026-09-20 21:03:01 | INFO | train rank=0 | step=12730/50000 progress=25.5% loss=0.299025 lr=8.898531e-05 lr_backbone=8.898531e-05 elapsed=14:36:25 avg_step=4.13s recent_step=4.10s eta=42:45:55 +2026-09-20 21:03:42 | INFO | train rank=0 | step=12740/50000 progress=25.5% loss=0.341252 lr=8.896460e-05 lr_backbone=8.896460e-05 elapsed=14:37:06 avg_step=4.13s recent_step=4.08s eta=42:45:12 +2026-09-20 21:04:23 | INFO | train rank=0 | step=12750/50000 progress=25.5% loss=0.363834 lr=8.894386e-05 lr_backbone=8.894386e-05 elapsed=14:37:47 avg_step=4.13s recent_step=4.10s eta=42:44:30 +2026-09-20 21:05:04 | INFO | train rank=0 | step=12760/50000 progress=25.5% loss=0.363123 lr=8.892312e-05 lr_backbone=8.892312e-05 elapsed=14:38:28 avg_step=4.13s recent_step=4.11s eta=42:43:48 +2026-09-20 21:05:45 | INFO | train rank=0 | step=12770/50000 progress=25.5% loss=0.346894 lr=8.890235e-05 lr_backbone=8.890235e-05 elapsed=14:39:08 avg_step=4.13s recent_step=4.07s eta=42:43:05 +2026-09-20 21:06:26 | INFO | train rank=0 | step=12780/50000 progress=25.6% loss=0.355336 lr=8.888157e-05 lr_backbone=8.888157e-05 elapsed=14:39:50 avg_step=4.13s recent_step=4.11s eta=42:42:23 +2026-09-20 21:07:07 | INFO | train rank=0 | step=12790/50000 progress=25.6% loss=0.391126 lr=8.886077e-05 lr_backbone=8.886077e-05 elapsed=14:40:31 avg_step=4.13s recent_step=4.11s eta=42:41:41 +2026-09-20 21:07:48 | INFO | train rank=0 | step=12800/50000 progress=25.6% loss=0.328677 lr=8.883995e-05 lr_backbone=8.883995e-05 elapsed=14:41:12 avg_step=4.13s recent_step=4.11s eta=42:40:59 +2026-09-20 21:08:29 | INFO | train rank=0 | step=12810/50000 progress=25.6% loss=0.325914 lr=8.881912e-05 lr_backbone=8.881912e-05 elapsed=14:41:53 avg_step=4.13s recent_step=4.09s eta=42:40:17 +2026-09-20 21:09:10 | INFO | train rank=0 | step=12820/50000 progress=25.6% loss=0.387965 lr=8.879826e-05 lr_backbone=8.879826e-05 elapsed=14:42:34 avg_step=4.13s recent_step=4.10s eta=42:39:35 +2026-09-20 21:09:51 | INFO | train rank=0 | step=12830/50000 progress=25.7% loss=0.325411 lr=8.877740e-05 lr_backbone=8.877740e-05 elapsed=14:43:14 avg_step=4.13s recent_step=4.07s eta=42:38:52 +2026-09-20 21:10:32 | INFO | train rank=0 | step=12840/50000 progress=25.7% loss=0.367328 lr=8.875651e-05 lr_backbone=8.875651e-05 elapsed=14:43:55 avg_step=4.13s recent_step=4.11s eta=42:38:10 +2026-09-20 21:11:13 | INFO | train rank=0 | step=12850/50000 progress=25.7% loss=0.415030 lr=8.873561e-05 lr_backbone=8.873561e-05 elapsed=14:44:37 avg_step=4.13s recent_step=4.12s eta=42:37:28 +2026-09-20 21:11:55 | INFO | train rank=0 | step=12860/50000 progress=25.7% loss=0.439386 lr=8.871469e-05 lr_backbone=8.871469e-05 elapsed=14:45:18 avg_step=4.13s recent_step=4.12s eta=42:36:47 +2026-09-20 21:12:36 | INFO | train rank=0 | step=12870/50000 progress=25.7% loss=0.381932 lr=8.869376e-05 lr_backbone=8.869376e-05 elapsed=14:45:59 avg_step=4.13s recent_step=4.10s eta=42:36:05 +2026-09-20 21:13:17 | INFO | train rank=0 | step=12880/50000 progress=25.8% loss=0.461789 lr=8.867280e-05 lr_backbone=8.867280e-05 elapsed=14:46:40 avg_step=4.13s recent_step=4.11s eta=42:35:23 +2026-09-20 21:13:57 | INFO | train rank=0 | step=12890/50000 progress=25.8% loss=0.411472 lr=8.865183e-05 lr_backbone=8.865183e-05 elapsed=14:47:21 avg_step=4.13s recent_step=4.08s eta=42:34:40 +2026-09-20 21:14:38 | INFO | train rank=0 | step=12900/50000 progress=25.8% loss=0.399264 lr=8.863085e-05 lr_backbone=8.863085e-05 elapsed=14:48:02 avg_step=4.13s recent_step=4.09s eta=42:33:57 +2026-09-20 21:15:19 | INFO | train rank=0 | step=12910/50000 progress=25.8% loss=0.366536 lr=8.860984e-05 lr_backbone=8.860984e-05 elapsed=14:48:43 avg_step=4.13s recent_step=4.10s eta=42:33:15 +2026-09-20 21:16:00 | INFO | train rank=0 | step=12920/50000 progress=25.8% loss=0.339656 lr=8.858882e-05 lr_backbone=8.858882e-05 elapsed=14:49:24 avg_step=4.13s recent_step=4.10s eta=42:32:33 +2026-09-20 21:16:41 | INFO | train rank=0 | step=12930/50000 progress=25.9% loss=0.343809 lr=8.856779e-05 lr_backbone=8.856779e-05 elapsed=14:50:05 avg_step=4.13s recent_step=4.09s eta=42:31:51 +2026-09-20 21:17:23 | INFO | train rank=0 | step=12940/50000 progress=25.9% loss=0.363636 lr=8.854673e-05 lr_backbone=8.854673e-05 elapsed=14:50:46 avg_step=4.13s recent_step=4.13s eta=42:31:09 +2026-09-20 21:18:03 | INFO | train rank=0 | step=12950/50000 progress=25.9% loss=0.321020 lr=8.852566e-05 lr_backbone=8.852566e-05 elapsed=14:51:27 avg_step=4.13s recent_step=4.06s eta=42:30:26 +2026-09-20 21:18:44 | INFO | train rank=0 | step=12960/50000 progress=25.9% loss=0.365043 lr=8.850457e-05 lr_backbone=8.850457e-05 elapsed=14:52:07 avg_step=4.13s recent_step=4.09s eta=42:29:44 +2026-09-20 21:19:25 | INFO | train rank=0 | step=12970/50000 progress=25.9% loss=0.346101 lr=8.848347e-05 lr_backbone=8.848347e-05 elapsed=14:52:48 avg_step=4.13s recent_step=4.08s eta=42:29:01 +2026-09-20 21:20:06 | INFO | train rank=0 | step=12980/50000 progress=26.0% loss=0.423864 lr=8.846235e-05 lr_backbone=8.846235e-05 elapsed=14:53:29 avg_step=4.13s recent_step=4.11s eta=42:28:19 +2026-09-20 21:20:47 | INFO | train rank=0 | step=12990/50000 progress=26.0% loss=0.390358 lr=8.844121e-05 lr_backbone=8.844121e-05 elapsed=14:54:10 avg_step=4.13s recent_step=4.08s eta=42:27:36 +2026-09-20 21:21:27 | INFO | train rank=0 | step=13000/50000 progress=26.0% loss=0.405459 lr=8.842006e-05 lr_backbone=8.842006e-05 elapsed=14:54:51 avg_step=4.13s recent_step=4.07s eta=42:26:53 +2026-09-20 21:22:08 | INFO | train rank=0 | step=13010/50000 progress=26.0% loss=0.429555 lr=8.839888e-05 lr_backbone=8.839888e-05 elapsed=14:55:31 avg_step=4.13s recent_step=4.05s eta=42:26:10 +2026-09-20 21:22:49 | INFO | train rank=0 | step=13020/50000 progress=26.0% loss=0.376760 lr=8.837770e-05 lr_backbone=8.837770e-05 elapsed=14:56:12 avg_step=4.13s recent_step=4.06s eta=42:25:26 +2026-09-20 21:23:29 | INFO | train rank=0 | step=13030/50000 progress=26.1% loss=0.377647 lr=8.835649e-05 lr_backbone=8.835649e-05 elapsed=14:56:53 avg_step=4.13s recent_step=4.06s eta=42:24:43 +2026-09-20 21:24:10 | INFO | train rank=0 | step=13040/50000 progress=26.1% loss=0.415532 lr=8.833527e-05 lr_backbone=8.833527e-05 elapsed=14:57:33 avg_step=4.13s recent_step=4.09s eta=42:24:01 +2026-09-20 21:24:51 | INFO | train rank=0 | step=13050/50000 progress=26.1% loss=0.322435 lr=8.831403e-05 lr_backbone=8.831403e-05 elapsed=14:58:14 avg_step=4.13s recent_step=4.09s eta=42:23:18 +2026-09-20 21:25:32 | INFO | train rank=0 | step=13060/50000 progress=26.1% loss=0.367536 lr=8.829277e-05 lr_backbone=8.829277e-05 elapsed=14:58:55 avg_step=4.13s recent_step=4.06s eta=42:22:35 +2026-09-20 21:26:12 | INFO | train rank=0 | step=13070/50000 progress=26.1% loss=0.332183 lr=8.827150e-05 lr_backbone=8.827150e-05 elapsed=14:59:35 avg_step=4.13s recent_step=4.03s eta=42:21:51 +2026-09-20 21:26:53 | INFO | train rank=0 | step=13080/50000 progress=26.2% loss=0.380154 lr=8.825021e-05 lr_backbone=8.825021e-05 elapsed=15:00:16 avg_step=4.13s recent_step=4.06s eta=42:21:08 +2026-09-20 21:27:33 | INFO | train rank=0 | step=13090/50000 progress=26.2% loss=0.386953 lr=8.822891e-05 lr_backbone=8.822891e-05 elapsed=15:00:57 avg_step=4.13s recent_step=4.08s eta=42:20:25 +2026-09-20 21:28:14 | INFO | train rank=0 | step=13100/50000 progress=26.2% loss=0.403225 lr=8.820758e-05 lr_backbone=8.820758e-05 elapsed=15:01:37 avg_step=4.13s recent_step=4.08s eta=42:19:42 +2026-09-20 21:28:55 | INFO | train rank=0 | step=13110/50000 progress=26.2% loss=0.346768 lr=8.818624e-05 lr_backbone=8.818624e-05 elapsed=15:02:18 avg_step=4.13s recent_step=4.08s eta=42:19:00 +2026-09-20 21:29:35 | INFO | train rank=0 | step=13120/50000 progress=26.2% loss=0.375321 lr=8.816489e-05 lr_backbone=8.816489e-05 elapsed=15:02:59 avg_step=4.13s recent_step=4.04s eta=42:18:16 +2026-09-20 21:30:16 | INFO | train rank=0 | step=13130/50000 progress=26.3% loss=0.384959 lr=8.814351e-05 lr_backbone=8.814351e-05 elapsed=15:03:39 avg_step=4.13s recent_step=4.02s eta=42:17:32 +2026-09-20 21:30:56 | INFO | train rank=0 | step=13140/50000 progress=26.3% loss=0.385759 lr=8.812213e-05 lr_backbone=8.812213e-05 elapsed=15:04:20 avg_step=4.13s recent_step=4.07s eta=42:16:48 +2026-09-20 21:31:37 | INFO | train rank=0 | step=13150/50000 progress=26.3% loss=0.317947 lr=8.810072e-05 lr_backbone=8.810072e-05 elapsed=15:05:00 avg_step=4.13s recent_step=4.07s eta=42:16:05 +2026-09-20 21:32:18 | INFO | train rank=0 | step=13160/50000 progress=26.3% loss=0.385957 lr=8.807930e-05 lr_backbone=8.807930e-05 elapsed=15:05:41 avg_step=4.13s recent_step=4.08s eta=42:15:23 +2026-09-20 21:32:59 | INFO | train rank=0 | step=13170/50000 progress=26.3% loss=0.411244 lr=8.805786e-05 lr_backbone=8.805786e-05 elapsed=15:06:22 avg_step=4.13s recent_step=4.09s eta=42:14:41 +2026-09-20 21:33:39 | INFO | train rank=0 | step=13180/50000 progress=26.4% loss=0.330460 lr=8.803640e-05 lr_backbone=8.803640e-05 elapsed=15:07:03 avg_step=4.13s recent_step=4.08s eta=42:13:58 +2026-09-20 21:34:21 | INFO | train rank=0 | step=13190/50000 progress=26.4% loss=0.352978 lr=8.801493e-05 lr_backbone=8.801493e-05 elapsed=15:07:45 avg_step=4.13s recent_step=4.18s eta=42:13:18 +2026-09-20 21:35:12 | INFO | train rank=0 | step=13200/50000 progress=26.4% loss=0.327351 lr=8.799344e-05 lr_backbone=8.799344e-05 elapsed=15:08:35 avg_step=4.13s recent_step=5.04s eta=42:13:02 +2026-09-20 21:36:00 | INFO | train rank=0 | step=13210/50000 progress=26.4% loss=0.327593 lr=8.797193e-05 lr_backbone=8.797193e-05 elapsed=15:09:23 avg_step=4.13s recent_step=4.85s eta=42:12:41 +2026-09-20 21:36:49 | INFO | train rank=0 | step=13220/50000 progress=26.4% loss=0.407056 lr=8.795041e-05 lr_backbone=8.795041e-05 elapsed=15:10:13 avg_step=4.13s recent_step=4.91s eta=42:12:21 +2026-09-20 21:37:38 | INFO | train rank=0 | step=13230/50000 progress=26.5% loss=0.299793 lr=8.792887e-05 lr_backbone=8.792887e-05 elapsed=15:11:01 avg_step=4.13s recent_step=4.85s eta=42:12:00 +2026-09-20 21:38:30 | INFO | train rank=0 | step=13240/50000 progress=26.5% loss=0.370930 lr=8.790732e-05 lr_backbone=8.790732e-05 elapsed=15:11:53 avg_step=4.13s recent_step=5.21s eta=42:11:48 +2026-09-20 21:39:25 | INFO | train rank=0 | step=13250/50000 progress=26.5% loss=0.338524 lr=8.788574e-05 lr_backbone=8.788574e-05 elapsed=15:12:48 avg_step=4.13s recent_step=5.49s eta=42:11:45 +2026-09-20 21:40:16 | INFO | train rank=0 | step=13260/50000 progress=26.5% loss=0.408969 lr=8.786415e-05 lr_backbone=8.786415e-05 elapsed=15:13:40 avg_step=4.13s recent_step=5.18s eta=42:11:32 +2026-09-20 21:41:06 | INFO | train rank=0 | step=13270/50000 progress=26.5% loss=0.343900 lr=8.784255e-05 lr_backbone=8.784255e-05 elapsed=15:14:30 avg_step=4.13s recent_step=4.97s eta=42:11:14 +2026-09-20 21:41:54 | INFO | train rank=0 | step=13280/50000 progress=26.6% loss=0.374027 lr=8.782093e-05 lr_backbone=8.782093e-05 elapsed=15:15:17 avg_step=4.14s recent_step=4.76s eta=42:10:50 +2026-09-20 21:42:47 | INFO | train rank=0 | step=13290/50000 progress=26.6% loss=0.338655 lr=8.779929e-05 lr_backbone=8.779929e-05 elapsed=15:16:10 avg_step=4.14s recent_step=5.31s eta=42:10:41 +2026-09-20 21:43:36 | INFO | train rank=0 | step=13300/50000 progress=26.6% loss=0.342674 lr=8.777763e-05 lr_backbone=8.777763e-05 elapsed=15:17:00 avg_step=4.14s recent_step=4.95s eta=42:10:22 +2026-09-20 21:44:28 | INFO | train rank=0 | step=13310/50000 progress=26.6% loss=0.358994 lr=8.775596e-05 lr_backbone=8.775596e-05 elapsed=15:17:51 avg_step=4.14s recent_step=5.16s eta=42:10:09 +2026-09-20 21:45:18 | INFO | train rank=0 | step=13320/50000 progress=26.6% loss=0.389204 lr=8.773427e-05 lr_backbone=8.773427e-05 elapsed=15:18:41 avg_step=4.14s recent_step=5.00s eta=42:09:51 +2026-09-20 21:46:07 | INFO | train rank=0 | step=13330/50000 progress=26.7% loss=0.425130 lr=8.771257e-05 lr_backbone=8.771257e-05 elapsed=15:19:30 avg_step=4.14s recent_step=4.90s eta=42:09:31 +2026-09-20 21:46:58 | INFO | train rank=0 | step=13340/50000 progress=26.7% loss=0.329495 lr=8.769085e-05 lr_backbone=8.769085e-05 elapsed=15:20:21 avg_step=4.14s recent_step=5.07s eta=42:09:15 +2026-09-20 21:47:43 | INFO | train rank=0 | step=13350/50000 progress=26.7% loss=0.358016 lr=8.766911e-05 lr_backbone=8.766911e-05 elapsed=15:21:07 avg_step=4.14s recent_step=4.57s eta=42:08:45 +2026-09-20 21:48:30 | INFO | train rank=0 | step=13360/50000 progress=26.7% loss=0.357513 lr=8.764736e-05 lr_backbone=8.764736e-05 elapsed=15:21:54 avg_step=4.14s recent_step=4.71s eta=42:08:20 +2026-09-20 21:49:17 | INFO | train rank=0 | step=13370/50000 progress=26.7% loss=0.436255 lr=8.762559e-05 lr_backbone=8.762559e-05 elapsed=15:22:40 avg_step=4.14s recent_step=4.65s eta=42:07:52 +2026-09-20 21:50:07 | INFO | train rank=0 | step=13380/50000 progress=26.8% loss=0.388222 lr=8.760380e-05 lr_backbone=8.760380e-05 elapsed=15:23:30 avg_step=4.14s recent_step=5.02s eta=42:07:35 +2026-09-20 21:50:52 | INFO | train rank=0 | step=13390/50000 progress=26.8% loss=0.343035 lr=8.758200e-05 lr_backbone=8.758200e-05 elapsed=15:24:15 avg_step=4.14s recent_step=4.48s eta=42:07:03 +2026-09-20 21:51:41 | INFO | train rank=0 | step=13400/50000 progress=26.8% loss=0.406047 lr=8.756018e-05 lr_backbone=8.756018e-05 elapsed=15:25:05 avg_step=4.14s recent_step=4.93s eta=42:06:43 +2026-09-20 21:52:32 | INFO | train rank=0 | step=13410/50000 progress=26.8% loss=0.379973 lr=8.753834e-05 lr_backbone=8.753834e-05 elapsed=15:25:55 avg_step=4.14s recent_step=5.04s eta=42:06:26 +2026-09-20 21:53:20 | INFO | train rank=0 | step=13420/50000 progress=26.8% loss=0.346900 lr=8.751649e-05 lr_backbone=8.751649e-05 elapsed=15:26:44 avg_step=4.14s recent_step=4.89s eta=42:06:05 +2026-09-20 21:54:11 | INFO | train rank=0 | step=13430/50000 progress=26.9% loss=0.338894 lr=8.749462e-05 lr_backbone=8.749462e-05 elapsed=15:27:35 avg_step=4.14s recent_step=5.09s eta=42:05:49 +2026-09-20 21:55:00 | INFO | train rank=0 | step=13440/50000 progress=26.9% loss=0.303728 lr=8.747273e-05 lr_backbone=8.747273e-05 elapsed=15:28:23 avg_step=4.14s recent_step=4.83s eta=42:05:26 +2026-09-20 21:55:47 | INFO | train rank=0 | step=13450/50000 progress=26.9% loss=0.387685 lr=8.745083e-05 lr_backbone=8.745083e-05 elapsed=15:29:10 avg_step=4.15s recent_step=4.75s eta=42:05:01 +2026-09-20 21:56:39 | INFO | train rank=0 | step=13460/50000 progress=26.9% loss=0.346023 lr=8.742891e-05 lr_backbone=8.742891e-05 elapsed=15:30:02 avg_step=4.15s recent_step=5.14s eta=42:04:47 +2026-09-20 21:57:29 | INFO | train rank=0 | step=13470/50000 progress=26.9% loss=0.312209 lr=8.740698e-05 lr_backbone=8.740698e-05 elapsed=15:30:52 avg_step=4.15s recent_step=5.02s eta=42:04:29 +2026-09-20 21:58:17 | INFO | train rank=0 | step=13480/50000 progress=27.0% loss=0.347377 lr=8.738503e-05 lr_backbone=8.738503e-05 elapsed=15:31:40 avg_step=4.15s recent_step=4.78s eta=42:04:05 +2026-09-20 21:59:09 | INFO | train rank=0 | step=13490/50000 progress=27.0% loss=0.439144 lr=8.736306e-05 lr_backbone=8.736306e-05 elapsed=15:32:33 avg_step=4.15s recent_step=5.28s eta=42:03:54 +2026-09-20 21:59:55 | INFO | train rank=0 | step=13500/50000 progress=27.0% loss=0.398944 lr=8.734107e-05 lr_backbone=8.734107e-05 elapsed=15:33:19 avg_step=4.15s recent_step=4.59s eta=42:03:25 +2026-09-20 22:00:36 | INFO | train rank=0 | step=13510/50000 progress=27.0% loss=0.357238 lr=8.731907e-05 lr_backbone=8.731907e-05 elapsed=15:34:00 avg_step=4.15s recent_step=4.09s eta=42:02:41 +2026-09-20 22:01:17 | INFO | train rank=0 | step=13520/50000 progress=27.0% loss=0.293806 lr=8.729706e-05 lr_backbone=8.729706e-05 elapsed=15:34:40 avg_step=4.15s recent_step=4.07s eta=42:01:58 +2026-09-20 22:02:16 | INFO | train rank=0 | step=13530/50000 progress=27.1% loss=0.324007 lr=8.727502e-05 lr_backbone=8.727502e-05 elapsed=15:35:39 avg_step=4.15s recent_step=5.89s eta=42:02:03 +2026-09-20 22:03:31 | INFO | train rank=0 | step=13540/50000 progress=27.1% loss=0.343458 lr=8.725298e-05 lr_backbone=8.725298e-05 elapsed=15:36:54 avg_step=4.15s recent_step=7.48s eta=42:02:51 +2026-09-20 22:04:51 | INFO | train rank=0 | step=13550/50000 progress=27.1% loss=0.399593 lr=8.723091e-05 lr_backbone=8.723091e-05 elapsed=15:38:14 avg_step=4.15s recent_step=8.03s eta=42:03:54 +2026-09-20 22:06:07 | INFO | train rank=0 | step=13560/50000 progress=27.1% loss=0.349969 lr=8.720883e-05 lr_backbone=8.720883e-05 elapsed=15:39:30 avg_step=4.16s recent_step=7.61s eta=42:04:46 +2026-09-20 22:07:28 | INFO | train rank=0 | step=13570/50000 progress=27.1% loss=0.365181 lr=8.718673e-05 lr_backbone=8.718673e-05 elapsed=15:40:52 avg_step=4.16s recent_step=8.12s eta=42:05:51 +2026-09-20 22:08:44 | INFO | train rank=0 | step=13580/50000 progress=27.2% loss=0.446525 lr=8.716462e-05 lr_backbone=8.716462e-05 elapsed=15:42:07 avg_step=4.16s recent_step=7.59s eta=42:06:41 +2026-09-20 22:09:58 | INFO | train rank=0 | step=13590/50000 progress=27.2% loss=0.323293 lr=8.714249e-05 lr_backbone=8.714249e-05 elapsed=15:43:21 avg_step=4.16s recent_step=7.39s eta=42:07:26 +2026-09-20 22:11:05 | INFO | train rank=0 | step=13600/50000 progress=27.2% loss=0.375745 lr=8.712034e-05 lr_backbone=8.712034e-05 elapsed=15:44:28 avg_step=4.17s recent_step=6.71s eta=42:07:52 +2026-09-20 22:17:53 | INFO | train rank=0 | step=13610/50000 progress=27.2% loss=0.302194 lr=8.709818e-05 lr_backbone=8.709818e-05 elapsed=15:51:17 avg_step=4.19s recent_step=40.82s eta=42:23:31 +2026-09-20 22:18:35 | INFO | train rank=0 | step=13620/50000 progress=27.2% loss=0.368661 lr=8.707600e-05 lr_backbone=8.707600e-05 elapsed=15:51:58 avg_step=4.19s recent_step=4.14s eta=42:22:47 +2026-09-20 22:19:16 | INFO | train rank=0 | step=13630/50000 progress=27.3% loss=0.353103 lr=8.705380e-05 lr_backbone=8.705380e-05 elapsed=15:52:39 avg_step=4.19s recent_step=4.13s eta=42:22:04 +2026-09-20 22:19:57 | INFO | train rank=0 | step=13640/50000 progress=27.3% loss=0.417859 lr=8.703159e-05 lr_backbone=8.703159e-05 elapsed=15:53:21 avg_step=4.19s recent_step=4.14s eta=42:21:20 +2026-09-20 22:20:39 | INFO | train rank=0 | step=13650/50000 progress=27.3% loss=0.414099 lr=8.700936e-05 lr_backbone=8.700936e-05 elapsed=15:54:02 avg_step=4.19s recent_step=4.15s eta=42:20:37 +2026-09-20 22:21:20 | INFO | train rank=0 | step=13660/50000 progress=27.3% loss=0.376112 lr=8.698712e-05 lr_backbone=8.698712e-05 elapsed=15:54:44 avg_step=4.19s recent_step=4.14s eta=42:19:54 +2026-09-20 22:22:02 | INFO | train rank=0 | step=13670/50000 progress=27.3% loss=0.360559 lr=8.696486e-05 lr_backbone=8.696486e-05 elapsed=15:55:25 avg_step=4.19s recent_step=4.11s eta=42:19:10 +2026-09-20 22:22:43 | INFO | train rank=0 | step=13680/50000 progress=27.4% loss=0.353108 lr=8.694258e-05 lr_backbone=8.694258e-05 elapsed=15:56:06 avg_step=4.19s recent_step=4.10s eta=42:18:25 +2026-09-20 22:23:24 | INFO | train rank=0 | step=13690/50000 progress=27.4% loss=0.349380 lr=8.692029e-05 lr_backbone=8.692029e-05 elapsed=15:56:47 avg_step=4.19s recent_step=4.15s eta=42:17:42 +2026-09-20 22:24:05 | INFO | train rank=0 | step=13700/50000 progress=27.4% loss=0.309559 lr=8.689798e-05 lr_backbone=8.689798e-05 elapsed=15:57:29 avg_step=4.19s recent_step=4.15s eta=42:16:59 +2026-09-20 22:24:47 | INFO | train rank=0 | step=13710/50000 progress=27.4% loss=0.436921 lr=8.687566e-05 lr_backbone=8.687566e-05 elapsed=15:58:10 avg_step=4.19s recent_step=4.15s eta=42:16:16 +2026-09-20 22:25:29 | INFO | train rank=0 | step=13720/50000 progress=27.4% loss=0.340138 lr=8.685331e-05 lr_backbone=8.685331e-05 elapsed=15:58:52 avg_step=4.19s recent_step=4.16s eta=42:15:33 +2026-09-20 22:26:10 | INFO | train rank=0 | step=13730/50000 progress=27.5% loss=0.337335 lr=8.683096e-05 lr_backbone=8.683096e-05 elapsed=15:59:33 avg_step=4.19s recent_step=4.09s eta=42:14:49 +2026-09-20 22:26:51 | INFO | train rank=0 | step=13740/50000 progress=27.5% loss=0.384722 lr=8.680858e-05 lr_backbone=8.680858e-05 elapsed=16:00:14 avg_step=4.19s recent_step=4.14s eta=42:14:06 +2026-09-20 22:27:32 | INFO | train rank=0 | step=13750/50000 progress=27.5% loss=0.349453 lr=8.678620e-05 lr_backbone=8.678620e-05 elapsed=16:00:56 avg_step=4.19s recent_step=4.13s eta=42:13:22 +2026-09-20 22:28:14 | INFO | train rank=0 | step=13760/50000 progress=27.5% loss=0.359271 lr=8.676379e-05 lr_backbone=8.676379e-05 elapsed=16:01:37 avg_step=4.19s recent_step=4.16s eta=42:12:39 +2026-09-20 22:28:55 | INFO | train rank=0 | step=13770/50000 progress=27.5% loss=0.344124 lr=8.674137e-05 lr_backbone=8.674137e-05 elapsed=16:02:19 avg_step=4.19s recent_step=4.15s eta=42:11:56 +2026-09-20 22:29:37 | INFO | train rank=0 | step=13780/50000 progress=27.6% loss=0.291991 lr=8.671893e-05 lr_backbone=8.671893e-05 elapsed=16:03:00 avg_step=4.19s recent_step=4.14s eta=42:11:13 +2026-09-20 22:30:18 | INFO | train rank=0 | step=13790/50000 progress=27.6% loss=0.345248 lr=8.669648e-05 lr_backbone=8.669648e-05 elapsed=16:03:41 avg_step=4.19s recent_step=4.10s eta=42:10:29 +2026-09-20 22:30:59 | INFO | train rank=0 | step=13800/50000 progress=27.6% loss=0.322042 lr=8.667401e-05 lr_backbone=8.667401e-05 elapsed=16:04:22 avg_step=4.19s recent_step=4.12s eta=42:09:45 +2026-09-20 22:31:40 | INFO | train rank=0 | step=13810/50000 progress=27.6% loss=0.370097 lr=8.665152e-05 lr_backbone=8.665152e-05 elapsed=16:05:04 avg_step=4.19s recent_step=4.15s eta=42:09:02 +2026-09-20 22:32:22 | INFO | train rank=0 | step=13820/50000 progress=27.6% loss=0.344846 lr=8.662902e-05 lr_backbone=8.662902e-05 elapsed=16:05:45 avg_step=4.19s recent_step=4.13s eta=42:08:18 +2026-09-20 22:33:03 | INFO | train rank=0 | step=13830/50000 progress=27.7% loss=0.372251 lr=8.660650e-05 lr_backbone=8.660650e-05 elapsed=16:06:27 avg_step=4.19s recent_step=4.16s eta=42:07:35 +2026-09-20 22:33:45 | INFO | train rank=0 | step=13840/50000 progress=27.7% loss=0.346681 lr=8.658397e-05 lr_backbone=8.658397e-05 elapsed=16:07:08 avg_step=4.19s recent_step=4.13s eta=42:06:51 +2026-09-20 22:34:26 | INFO | train rank=0 | step=13850/50000 progress=27.7% loss=0.406868 lr=8.656142e-05 lr_backbone=8.656142e-05 elapsed=16:07:49 avg_step=4.19s recent_step=4.10s eta=42:06:07 +2026-09-20 22:35:07 | INFO | train rank=0 | step=13860/50000 progress=27.7% loss=0.335357 lr=8.653885e-05 lr_backbone=8.653885e-05 elapsed=16:08:30 avg_step=4.19s recent_step=4.13s eta=42:05:24 +2026-09-20 22:35:49 | INFO | train rank=0 | step=13870/50000 progress=27.7% loss=0.380574 lr=8.651627e-05 lr_backbone=8.651627e-05 elapsed=16:09:12 avg_step=4.19s recent_step=4.16s eta=42:04:41 +2026-09-20 22:36:30 | INFO | train rank=0 | step=13880/50000 progress=27.8% loss=0.353517 lr=8.649367e-05 lr_backbone=8.649367e-05 elapsed=16:09:53 avg_step=4.19s recent_step=4.14s eta=42:03:58 +2026-09-20 22:37:11 | INFO | train rank=0 | step=13890/50000 progress=27.8% loss=0.358058 lr=8.647106e-05 lr_backbone=8.647106e-05 elapsed=16:10:35 avg_step=4.19s recent_step=4.14s eta=42:03:14 +2026-09-20 22:37:53 | INFO | train rank=0 | step=13900/50000 progress=27.8% loss=0.328330 lr=8.644843e-05 lr_backbone=8.644843e-05 elapsed=16:11:16 avg_step=4.19s recent_step=4.16s eta=42:02:32 +2026-09-20 22:38:34 | INFO | train rank=0 | step=13910/50000 progress=27.8% loss=0.352903 lr=8.642579e-05 lr_backbone=8.642579e-05 elapsed=16:11:57 avg_step=4.19s recent_step=4.10s eta=42:01:47 +2026-09-20 22:39:15 | INFO | train rank=0 | step=13920/50000 progress=27.8% loss=0.349891 lr=8.640312e-05 lr_backbone=8.640312e-05 elapsed=16:12:39 avg_step=4.19s recent_step=4.13s eta=42:01:04 +2026-09-20 22:39:57 | INFO | train rank=0 | step=13930/50000 progress=27.9% loss=0.330922 lr=8.638045e-05 lr_backbone=8.638045e-05 elapsed=16:13:20 avg_step=4.19s recent_step=4.16s eta=42:00:21 +2026-09-20 22:40:38 | INFO | train rank=0 | step=13940/50000 progress=27.9% loss=0.399275 lr=8.635775e-05 lr_backbone=8.635775e-05 elapsed=16:14:02 avg_step=4.19s recent_step=4.15s eta=41:59:38 +2026-09-20 22:41:20 | INFO | train rank=0 | step=13950/50000 progress=27.9% loss=0.325112 lr=8.633504e-05 lr_backbone=8.633504e-05 elapsed=16:14:43 avg_step=4.19s recent_step=4.12s eta=41:58:54 +2026-09-20 22:42:01 | INFO | train rank=0 | step=13960/50000 progress=27.9% loss=0.366557 lr=8.631232e-05 lr_backbone=8.631232e-05 elapsed=16:15:24 avg_step=4.19s recent_step=4.15s eta=41:58:11 +2026-09-20 22:42:42 | INFO | train rank=0 | step=13970/50000 progress=27.9% loss=0.383968 lr=8.628958e-05 lr_backbone=8.628958e-05 elapsed=16:16:05 avg_step=4.19s recent_step=4.09s eta=41:57:27 +2026-09-20 22:43:23 | INFO | train rank=0 | step=13980/50000 progress=28.0% loss=0.399385 lr=8.626682e-05 lr_backbone=8.626682e-05 elapsed=16:16:47 avg_step=4.19s recent_step=4.12s eta=41:56:43 +2026-09-20 22:44:05 | INFO | train rank=0 | step=13990/50000 progress=28.0% loss=0.392269 lr=8.624405e-05 lr_backbone=8.624405e-05 elapsed=16:17:28 avg_step=4.19s recent_step=4.15s eta=41:56:00 +2026-09-20 22:44:46 | INFO | train rank=0 | step=14000/50000 progress=28.0% loss=0.326218 lr=8.622126e-05 lr_backbone=8.622126e-05 elapsed=16:18:09 avg_step=4.19s recent_step=4.14s eta=41:55:17 +2026-09-20 22:45:27 | INFO | train rank=0 | step=14010/50000 progress=28.0% loss=0.426212 lr=8.619846e-05 lr_backbone=8.619846e-05 elapsed=16:18:51 avg_step=4.19s recent_step=4.13s eta=41:54:33 +2026-09-20 22:46:09 | INFO | train rank=0 | step=14020/50000 progress=28.0% loss=0.365417 lr=8.617564e-05 lr_backbone=8.617564e-05 elapsed=16:19:32 avg_step=4.19s recent_step=4.12s eta=41:53:49 +2026-09-20 22:46:50 | INFO | train rank=0 | step=14030/50000 progress=28.1% loss=0.343865 lr=8.615280e-05 lr_backbone=8.615280e-05 elapsed=16:20:13 avg_step=4.19s recent_step=4.10s eta=41:53:05 +2026-09-20 22:47:31 | INFO | train rank=0 | step=14040/50000 progress=28.1% loss=0.448316 lr=8.612995e-05 lr_backbone=8.612995e-05 elapsed=16:20:54 avg_step=4.19s recent_step=4.14s eta=41:52:22 +2026-09-20 22:48:12 | INFO | train rank=0 | step=14050/50000 progress=28.1% loss=0.387862 lr=8.610708e-05 lr_backbone=8.610708e-05 elapsed=16:21:36 avg_step=4.19s recent_step=4.13s eta=41:51:39 +2026-09-20 22:48:54 | INFO | train rank=0 | step=14060/50000 progress=28.1% loss=0.336090 lr=8.608420e-05 lr_backbone=8.608420e-05 elapsed=16:22:17 avg_step=4.19s recent_step=4.16s eta=41:50:56 +2026-09-20 22:49:36 | INFO | train rank=0 | step=14070/50000 progress=28.1% loss=0.345303 lr=8.606130e-05 lr_backbone=8.606130e-05 elapsed=16:22:59 avg_step=4.19s recent_step=4.16s eta=41:50:13 +2026-09-20 22:50:17 | INFO | train rank=0 | step=14080/50000 progress=28.2% loss=0.317727 lr=8.603838e-05 lr_backbone=8.603838e-05 elapsed=16:23:40 avg_step=4.19s recent_step=4.15s eta=41:49:30 +2026-09-20 22:50:58 | INFO | train rank=0 | step=14090/50000 progress=28.2% loss=0.405731 lr=8.601545e-05 lr_backbone=8.601545e-05 elapsed=16:24:21 avg_step=4.19s recent_step=4.10s eta=41:48:46 +2026-09-20 22:51:39 | INFO | train rank=0 | step=14100/50000 progress=28.2% loss=0.356292 lr=8.599250e-05 lr_backbone=8.599250e-05 elapsed=16:25:03 avg_step=4.19s recent_step=4.14s eta=41:48:02 +2026-09-20 22:52:21 | INFO | train rank=0 | step=14110/50000 progress=28.2% loss=0.381229 lr=8.596954e-05 lr_backbone=8.596954e-05 elapsed=16:25:44 avg_step=4.19s recent_step=4.14s eta=41:47:19 +2026-09-20 22:53:02 | INFO | train rank=0 | step=14120/50000 progress=28.2% loss=0.301664 lr=8.594656e-05 lr_backbone=8.594656e-05 elapsed=16:26:26 avg_step=4.19s recent_step=4.15s eta=41:46:36 +2026-09-20 22:53:44 | INFO | train rank=0 | step=14130/50000 progress=28.3% loss=0.389464 lr=8.592357e-05 lr_backbone=8.592357e-05 elapsed=16:27:07 avg_step=4.19s recent_step=4.14s eta=41:45:53 +2026-09-20 22:54:25 | INFO | train rank=0 | step=14140/50000 progress=28.3% loss=0.318633 lr=8.590056e-05 lr_backbone=8.590056e-05 elapsed=16:27:48 avg_step=4.19s recent_step=4.12s eta=41:45:09 +2026-09-20 22:55:06 | INFO | train rank=0 | step=14150/50000 progress=28.3% loss=0.347787 lr=8.587754e-05 lr_backbone=8.587754e-05 elapsed=16:28:29 avg_step=4.19s recent_step=4.10s eta=41:44:25 +2026-09-20 22:55:47 | INFO | train rank=0 | step=14160/50000 progress=28.3% loss=0.360353 lr=8.585449e-05 lr_backbone=8.585449e-05 elapsed=16:29:11 avg_step=4.19s recent_step=4.12s eta=41:43:41 +2026-09-20 22:56:28 | INFO | train rank=0 | step=14170/50000 progress=28.3% loss=0.344507 lr=8.583144e-05 lr_backbone=8.583144e-05 elapsed=16:29:52 avg_step=4.19s recent_step=4.13s eta=41:42:58 +2026-09-20 22:57:10 | INFO | train rank=0 | step=14180/50000 progress=28.4% loss=0.445481 lr=8.580837e-05 lr_backbone=8.580837e-05 elapsed=16:30:33 avg_step=4.19s recent_step=4.13s eta=41:42:14 +2026-09-20 22:57:51 | INFO | train rank=0 | step=14190/50000 progress=28.4% loss=0.338747 lr=8.578528e-05 lr_backbone=8.578528e-05 elapsed=16:31:15 avg_step=4.19s recent_step=4.14s eta=41:41:31 +2026-09-20 22:58:32 | INFO | train rank=0 | step=14200/50000 progress=28.4% loss=0.379517 lr=8.576217e-05 lr_backbone=8.576217e-05 elapsed=16:31:56 avg_step=4.19s recent_step=4.12s eta=41:40:47 +2026-09-20 22:59:13 | INFO | train rank=0 | step=14210/50000 progress=28.4% loss=0.356914 lr=8.573906e-05 lr_backbone=8.573906e-05 elapsed=16:32:37 avg_step=4.19s recent_step=4.10s eta=41:40:03 +2026-09-20 22:59:55 | INFO | train rank=0 | step=14220/50000 progress=28.4% loss=0.361219 lr=8.571592e-05 lr_backbone=8.571592e-05 elapsed=16:33:18 avg_step=4.19s recent_step=4.11s eta=41:39:19 +2026-09-20 23:00:36 | INFO | train rank=0 | step=14230/50000 progress=28.5% loss=0.332686 lr=8.569277e-05 lr_backbone=8.569277e-05 elapsed=16:33:59 avg_step=4.19s recent_step=4.16s eta=41:38:37 +2026-09-20 23:01:17 | INFO | train rank=0 | step=14240/50000 progress=28.5% loss=0.413885 lr=8.566960e-05 lr_backbone=8.566960e-05 elapsed=16:34:41 avg_step=4.19s recent_step=4.13s eta=41:37:53 +2026-09-20 23:01:59 | INFO | train rank=0 | step=14250/50000 progress=28.5% loss=0.428819 lr=8.564642e-05 lr_backbone=8.564642e-05 elapsed=16:35:22 avg_step=4.19s recent_step=4.16s eta=41:37:10 +2026-09-20 23:02:40 | INFO | train rank=0 | step=14260/50000 progress=28.5% loss=0.332041 lr=8.562323e-05 lr_backbone=8.562323e-05 elapsed=16:36:03 avg_step=4.19s recent_step=4.12s eta=41:36:27 +2026-09-20 23:03:21 | INFO | train rank=0 | step=14270/50000 progress=28.5% loss=0.443054 lr=8.560001e-05 lr_backbone=8.560001e-05 elapsed=16:36:44 avg_step=4.19s recent_step=4.10s eta=41:35:42 +2026-09-20 23:04:02 | INFO | train rank=0 | step=14280/50000 progress=28.6% loss=0.460150 lr=8.557678e-05 lr_backbone=8.557678e-05 elapsed=16:37:26 avg_step=4.19s recent_step=4.11s eta=41:34:59 +2026-09-20 23:04:44 | INFO | train rank=0 | step=14290/50000 progress=28.6% loss=0.323791 lr=8.555354e-05 lr_backbone=8.555354e-05 elapsed=16:38:07 avg_step=4.19s recent_step=4.14s eta=41:34:15 +2026-09-20 23:05:25 | INFO | train rank=0 | step=14300/50000 progress=28.6% loss=0.401355 lr=8.553028e-05 lr_backbone=8.553028e-05 elapsed=16:38:48 avg_step=4.19s recent_step=4.12s eta=41:33:32 +2026-09-20 23:06:06 | INFO | train rank=0 | step=14310/50000 progress=28.6% loss=0.349941 lr=8.550701e-05 lr_backbone=8.550701e-05 elapsed=16:39:30 avg_step=4.19s recent_step=4.13s eta=41:32:48 +2026-09-20 23:06:47 | INFO | train rank=0 | step=14320/50000 progress=28.6% loss=0.368056 lr=8.548371e-05 lr_backbone=8.548371e-05 elapsed=16:40:10 avg_step=4.19s recent_step=4.09s eta=41:32:04 +2026-09-20 23:07:28 | INFO | train rank=0 | step=14330/50000 progress=28.7% loss=0.403650 lr=8.546041e-05 lr_backbone=8.546041e-05 elapsed=16:40:51 avg_step=4.19s recent_step=4.10s eta=41:31:20 +2026-09-20 23:08:09 | INFO | train rank=0 | step=14340/50000 progress=28.7% loss=0.330377 lr=8.543709e-05 lr_backbone=8.543709e-05 elapsed=16:41:32 avg_step=4.19s recent_step=4.10s eta=41:30:36 +2026-09-20 23:08:50 | INFO | train rank=0 | step=14350/50000 progress=28.7% loss=0.368443 lr=8.541375e-05 lr_backbone=8.541375e-05 elapsed=16:42:14 avg_step=4.19s recent_step=4.13s eta=41:29:52 +2026-09-20 23:09:31 | INFO | train rank=0 | step=14360/50000 progress=28.7% loss=0.461473 lr=8.539040e-05 lr_backbone=8.539040e-05 elapsed=16:42:55 avg_step=4.19s recent_step=4.11s eta=41:29:08 +2026-09-20 23:10:13 | INFO | train rank=0 | step=14370/50000 progress=28.7% loss=0.326128 lr=8.536703e-05 lr_backbone=8.536703e-05 elapsed=16:43:36 avg_step=4.19s recent_step=4.12s eta=41:28:25 +2026-09-20 23:10:54 | INFO | train rank=0 | step=14380/50000 progress=28.8% loss=0.431099 lr=8.534365e-05 lr_backbone=8.534365e-05 elapsed=16:44:17 avg_step=4.19s recent_step=4.14s eta=41:27:41 +2026-09-20 23:11:35 | INFO | train rank=0 | step=14390/50000 progress=28.8% loss=0.372781 lr=8.532025e-05 lr_backbone=8.532025e-05 elapsed=16:44:58 avg_step=4.19s recent_step=4.08s eta=41:26:57 +2026-09-20 23:12:16 | INFO | train rank=0 | step=14400/50000 progress=28.8% loss=0.343612 lr=8.529683e-05 lr_backbone=8.529683e-05 elapsed=16:45:39 avg_step=4.19s recent_step=4.10s eta=41:26:13 +2026-09-20 23:12:57 | INFO | train rank=0 | step=14410/50000 progress=28.8% loss=0.398176 lr=8.527340e-05 lr_backbone=8.527340e-05 elapsed=16:46:21 avg_step=4.19s recent_step=4.15s eta=41:25:30 +2026-09-20 23:13:39 | INFO | train rank=0 | step=14420/50000 progress=28.8% loss=0.350471 lr=8.524996e-05 lr_backbone=8.524996e-05 elapsed=16:47:02 avg_step=4.19s recent_step=4.14s eta=41:24:47 +2026-09-20 23:14:20 | INFO | train rank=0 | step=14430/50000 progress=28.9% loss=0.375521 lr=8.522650e-05 lr_backbone=8.522650e-05 elapsed=16:47:44 avg_step=4.19s recent_step=4.14s eta=41:24:04 +2026-09-20 23:15:02 | INFO | train rank=0 | step=14440/50000 progress=28.9% loss=0.423126 lr=8.520302e-05 lr_backbone=8.520302e-05 elapsed=16:48:25 avg_step=4.19s recent_step=4.13s eta=41:23:20 +2026-09-20 23:15:43 | INFO | train rank=0 | step=14450/50000 progress=28.9% loss=0.390384 lr=8.517953e-05 lr_backbone=8.517953e-05 elapsed=16:49:06 avg_step=4.19s recent_step=4.09s eta=41:22:36 +2026-09-20 23:16:24 | INFO | train rank=0 | step=14460/50000 progress=28.9% loss=0.439243 lr=8.515602e-05 lr_backbone=8.515602e-05 elapsed=16:49:47 avg_step=4.19s recent_step=4.11s eta=41:21:52 +2026-09-20 23:17:05 | INFO | train rank=0 | step=14470/50000 progress=28.9% loss=0.318208 lr=8.513250e-05 lr_backbone=8.513250e-05 elapsed=16:50:28 avg_step=4.19s recent_step=4.13s eta=41:21:09 +2026-09-20 23:17:46 | INFO | train rank=0 | step=14480/50000 progress=29.0% loss=0.408029 lr=8.510896e-05 lr_backbone=8.510896e-05 elapsed=16:51:09 avg_step=4.19s recent_step=4.11s eta=41:20:25 +2026-09-20 23:18:27 | INFO | train rank=0 | step=14490/50000 progress=29.0% loss=0.368604 lr=8.508541e-05 lr_backbone=8.508541e-05 elapsed=16:51:51 avg_step=4.19s recent_step=4.13s eta=41:19:41 +2026-09-20 23:19:09 | INFO | train rank=0 | step=14500/50000 progress=29.0% loss=0.398224 lr=8.506184e-05 lr_backbone=8.506184e-05 elapsed=16:52:32 avg_step=4.19s recent_step=4.13s eta=41:18:58 +2026-09-20 23:19:49 | INFO | train rank=0 | step=14510/50000 progress=29.0% loss=0.340582 lr=8.503826e-05 lr_backbone=8.503826e-05 elapsed=16:53:13 avg_step=4.19s recent_step=4.07s eta=41:18:13 +2026-09-20 23:20:30 | INFO | train rank=0 | step=14520/50000 progress=29.0% loss=0.449787 lr=8.501466e-05 lr_backbone=8.501466e-05 elapsed=16:53:54 avg_step=4.19s recent_step=4.11s eta=41:17:29 +2026-09-20 23:21:12 | INFO | train rank=0 | step=14530/50000 progress=29.1% loss=0.317979 lr=8.499104e-05 lr_backbone=8.499104e-05 elapsed=16:54:35 avg_step=4.19s recent_step=4.14s eta=41:16:46 +2026-09-20 23:21:53 | INFO | train rank=0 | step=14540/50000 progress=29.1% loss=0.350638 lr=8.496741e-05 lr_backbone=8.496741e-05 elapsed=16:55:16 avg_step=4.19s recent_step=4.12s eta=41:16:02 +2026-09-20 23:22:34 | INFO | train rank=0 | step=14550/50000 progress=29.1% loss=0.305595 lr=8.494377e-05 lr_backbone=8.494377e-05 elapsed=16:55:57 avg_step=4.19s recent_step=4.13s eta=41:15:19 +2026-09-20 23:23:16 | INFO | train rank=0 | step=14560/50000 progress=29.1% loss=0.441376 lr=8.492011e-05 lr_backbone=8.492011e-05 elapsed=16:56:39 avg_step=4.19s recent_step=4.15s eta=41:14:36 +2026-09-20 23:23:56 | INFO | train rank=0 | step=14570/50000 progress=29.1% loss=0.367608 lr=8.489643e-05 lr_backbone=8.489643e-05 elapsed=16:57:20 avg_step=4.19s recent_step=4.07s eta=41:13:51 +2026-09-20 23:24:38 | INFO | train rank=0 | step=14580/50000 progress=29.2% loss=0.334075 lr=8.487274e-05 lr_backbone=8.487274e-05 elapsed=16:58:01 avg_step=4.19s recent_step=4.14s eta=41:13:08 +2026-09-20 23:25:19 | INFO | train rank=0 | step=14590/50000 progress=29.2% loss=0.313549 lr=8.484903e-05 lr_backbone=8.484903e-05 elapsed=16:58:43 avg_step=4.19s recent_step=4.15s eta=41:12:25 +2026-09-20 23:26:01 | INFO | train rank=0 | step=14600/50000 progress=29.2% loss=0.346576 lr=8.482531e-05 lr_backbone=8.482531e-05 elapsed=16:59:24 avg_step=4.19s recent_step=4.15s eta=41:11:43 +2026-09-20 23:26:42 | INFO | train rank=0 | step=14610/50000 progress=29.2% loss=0.386667 lr=8.480158e-05 lr_backbone=8.480158e-05 elapsed=17:00:06 avg_step=4.19s recent_step=4.15s eta=41:11:00 +2026-09-20 23:27:24 | INFO | train rank=0 | step=14620/50000 progress=29.2% loss=0.324736 lr=8.477782e-05 lr_backbone=8.477782e-05 elapsed=17:00:47 avg_step=4.19s recent_step=4.17s eta=41:10:17 +2026-09-20 23:28:05 | INFO | train rank=0 | step=14630/50000 progress=29.3% loss=0.346577 lr=8.475406e-05 lr_backbone=8.475406e-05 elapsed=17:01:28 avg_step=4.19s recent_step=4.11s eta=41:09:33 +2026-09-20 23:28:46 | INFO | train rank=0 | step=14640/50000 progress=29.3% loss=0.373239 lr=8.473028e-05 lr_backbone=8.473028e-05 elapsed=17:02:09 avg_step=4.19s recent_step=4.12s eta=41:08:50 +2026-09-20 23:29:27 | INFO | train rank=0 | step=14650/50000 progress=29.3% loss=0.370354 lr=8.470648e-05 lr_backbone=8.470648e-05 elapsed=17:02:51 avg_step=4.19s recent_step=4.13s eta=41:08:07 +2026-09-20 23:30:09 | INFO | train rank=0 | step=14660/50000 progress=29.3% loss=0.333461 lr=8.468267e-05 lr_backbone=8.468267e-05 elapsed=17:03:32 avg_step=4.19s recent_step=4.13s eta=41:07:23 +2026-09-20 23:30:50 | INFO | train rank=0 | step=14670/50000 progress=29.3% loss=0.364807 lr=8.465884e-05 lr_backbone=8.465884e-05 elapsed=17:04:14 avg_step=4.19s recent_step=4.15s eta=41:06:40 +2026-09-20 23:31:32 | INFO | train rank=0 | step=14680/50000 progress=29.4% loss=0.323281 lr=8.463499e-05 lr_backbone=8.463499e-05 elapsed=17:04:55 avg_step=4.19s recent_step=4.14s eta=41:05:57 +2026-09-20 23:32:13 | INFO | train rank=0 | step=14690/50000 progress=29.4% loss=0.370746 lr=8.461114e-05 lr_backbone=8.461114e-05 elapsed=17:05:36 avg_step=4.19s recent_step=4.08s eta=41:05:13 +2026-09-20 23:32:54 | INFO | train rank=0 | step=14700/50000 progress=29.4% loss=0.346841 lr=8.458726e-05 lr_backbone=8.458726e-05 elapsed=17:06:17 avg_step=4.19s recent_step=4.12s eta=41:04:29 +2026-09-20 23:33:35 | INFO | train rank=0 | step=14710/50000 progress=29.4% loss=0.373260 lr=8.456337e-05 lr_backbone=8.456337e-05 elapsed=17:06:58 avg_step=4.19s recent_step=4.09s eta=41:03:45 +2026-09-20 23:34:16 | INFO | train rank=0 | step=14720/50000 progress=29.4% loss=0.359208 lr=8.453947e-05 lr_backbone=8.453947e-05 elapsed=17:07:40 avg_step=4.19s recent_step=4.17s eta=41:03:03 +2026-09-20 23:34:58 | INFO | train rank=0 | step=14730/50000 progress=29.5% loss=0.373284 lr=8.451555e-05 lr_backbone=8.451555e-05 elapsed=17:08:21 avg_step=4.19s recent_step=4.17s eta=41:02:20 +2026-09-20 23:35:40 | INFO | train rank=0 | step=14740/50000 progress=29.5% loss=0.350311 lr=8.449162e-05 lr_backbone=8.449162e-05 elapsed=17:09:03 avg_step=4.19s recent_step=4.16s eta=41:01:38 +2026-09-20 23:36:21 | INFO | train rank=0 | step=14750/50000 progress=29.5% loss=0.313968 lr=8.446767e-05 lr_backbone=8.446767e-05 elapsed=17:09:44 avg_step=4.19s recent_step=4.09s eta=41:00:54 +2026-09-20 23:37:02 | INFO | train rank=0 | step=14760/50000 progress=29.5% loss=0.288539 lr=8.444371e-05 lr_backbone=8.444371e-05 elapsed=17:10:25 avg_step=4.19s recent_step=4.13s eta=41:00:10 +2026-09-20 23:37:44 | INFO | train rank=0 | step=14770/50000 progress=29.5% loss=0.343297 lr=8.441973e-05 lr_backbone=8.441973e-05 elapsed=17:11:07 avg_step=4.19s recent_step=4.18s eta=40:59:28 +2026-09-20 23:38:25 | INFO | train rank=0 | step=14780/50000 progress=29.6% loss=0.384412 lr=8.439573e-05 lr_backbone=8.439573e-05 elapsed=17:11:48 avg_step=4.19s recent_step=4.13s eta=40:58:45 +2026-09-20 23:39:07 | INFO | train rank=0 | step=14790/50000 progress=29.6% loss=0.297431 lr=8.437172e-05 lr_backbone=8.437172e-05 elapsed=17:12:30 avg_step=4.19s recent_step=4.16s eta=40:58:03 +2026-09-20 23:39:48 | INFO | train rank=0 | step=14800/50000 progress=29.6% loss=0.328577 lr=8.434770e-05 lr_backbone=8.434770e-05 elapsed=17:13:12 avg_step=4.19s recent_step=4.17s eta=40:57:20 +2026-09-20 23:40:29 | INFO | train rank=0 | step=14810/50000 progress=29.6% loss=0.345110 lr=8.432366e-05 lr_backbone=8.432366e-05 elapsed=17:13:53 avg_step=4.19s recent_step=4.09s eta=40:56:36 +2026-09-20 23:41:10 | INFO | train rank=0 | step=14820/50000 progress=29.6% loss=0.381159 lr=8.429961e-05 lr_backbone=8.429961e-05 elapsed=17:14:34 avg_step=4.19s recent_step=4.12s eta=40:55:53 +2026-09-20 23:41:52 | INFO | train rank=0 | step=14830/50000 progress=29.7% loss=0.307974 lr=8.427554e-05 lr_backbone=8.427554e-05 elapsed=17:15:15 avg_step=4.19s recent_step=4.14s eta=40:55:10 +2026-09-20 23:42:33 | INFO | train rank=0 | step=14840/50000 progress=29.7% loss=0.354864 lr=8.425145e-05 lr_backbone=8.425145e-05 elapsed=17:15:56 avg_step=4.19s recent_step=4.12s eta=40:54:26 +2026-09-20 23:43:15 | INFO | train rank=0 | step=14850/50000 progress=29.7% loss=0.369091 lr=8.422736e-05 lr_backbone=8.422736e-05 elapsed=17:16:38 avg_step=4.19s recent_step=4.15s eta=40:53:43 +2026-09-20 23:43:56 | INFO | train rank=0 | step=14860/50000 progress=29.7% loss=0.401143 lr=8.420324e-05 lr_backbone=8.420324e-05 elapsed=17:17:19 avg_step=4.19s recent_step=4.15s eta=40:53:01 +2026-09-20 23:44:37 | INFO | train rank=0 | step=14870/50000 progress=29.7% loss=0.350181 lr=8.417911e-05 lr_backbone=8.417911e-05 elapsed=17:18:00 avg_step=4.19s recent_step=4.09s eta=40:52:17 +2026-09-20 23:45:18 | INFO | train rank=0 | step=14880/50000 progress=29.8% loss=0.349392 lr=8.415497e-05 lr_backbone=8.415497e-05 elapsed=17:18:42 avg_step=4.19s recent_step=4.14s eta=40:51:33 +2026-09-20 23:46:00 | INFO | train rank=0 | step=14890/50000 progress=29.8% loss=0.412700 lr=8.413081e-05 lr_backbone=8.413081e-05 elapsed=17:19:23 avg_step=4.19s recent_step=4.14s eta=40:50:50 +2026-09-20 23:46:41 | INFO | train rank=0 | step=14900/50000 progress=29.8% loss=0.394481 lr=8.410664e-05 lr_backbone=8.410664e-05 elapsed=17:20:05 avg_step=4.19s recent_step=4.16s eta=40:50:08 +2026-09-20 23:47:23 | INFO | train rank=0 | step=14910/50000 progress=29.8% loss=0.438850 lr=8.408245e-05 lr_backbone=8.408245e-05 elapsed=17:20:46 avg_step=4.19s recent_step=4.12s eta=40:49:25 +2026-09-20 23:48:05 | INFO | train rank=0 | step=14920/50000 progress=29.8% loss=0.393405 lr=8.405824e-05 lr_backbone=8.405824e-05 elapsed=17:21:28 avg_step=4.19s recent_step=4.18s eta=40:48:43 +2026-09-20 23:48:46 | INFO | train rank=0 | step=14930/50000 progress=29.9% loss=0.333193 lr=8.403402e-05 lr_backbone=8.403402e-05 elapsed=17:22:09 avg_step=4.19s recent_step=4.10s eta=40:47:59 +2026-09-20 23:49:27 | INFO | train rank=0 | step=14940/50000 progress=29.9% loss=0.332110 lr=8.400979e-05 lr_backbone=8.400979e-05 elapsed=17:22:50 avg_step=4.19s recent_step=4.11s eta=40:47:15 +2026-09-20 23:50:08 | INFO | train rank=0 | step=14950/50000 progress=29.9% loss=0.316611 lr=8.398554e-05 lr_backbone=8.398554e-05 elapsed=17:23:32 avg_step=4.19s recent_step=4.16s eta=40:46:32 +2026-09-20 23:50:50 | INFO | train rank=0 | step=14960/50000 progress=29.9% loss=0.337174 lr=8.396128e-05 lr_backbone=8.396128e-05 elapsed=17:24:13 avg_step=4.19s recent_step=4.17s eta=40:45:50 +2026-09-20 23:51:31 | INFO | train rank=0 | step=14970/50000 progress=29.9% loss=0.353994 lr=8.393700e-05 lr_backbone=8.393700e-05 elapsed=17:24:55 avg_step=4.19s recent_step=4.13s eta=40:45:07 +2026-09-20 23:52:13 | INFO | train rank=0 | step=14980/50000 progress=30.0% loss=0.300197 lr=8.391271e-05 lr_backbone=8.391271e-05 elapsed=17:25:36 avg_step=4.19s recent_step=4.19s eta=40:44:25 +2026-09-20 23:52:54 | INFO | train rank=0 | step=14990/50000 progress=30.0% loss=0.324354 lr=8.388840e-05 lr_backbone=8.388840e-05 elapsed=17:26:17 avg_step=4.19s recent_step=4.09s eta=40:43:41 +2026-09-20 23:53:35 | INFO | train rank=0 | step=15000/50000 progress=30.0% loss=0.357013 lr=8.386408e-05 lr_backbone=8.386408e-05 elapsed=17:26:59 avg_step=4.19s recent_step=4.13s eta=40:42:57 +2026-09-20 23:54:17 | INFO | train rank=0 | step=15010/50000 progress=30.0% loss=0.320868 lr=8.383974e-05 lr_backbone=8.383974e-05 elapsed=17:27:40 avg_step=4.19s recent_step=4.15s eta=40:42:15 +2026-09-20 23:54:58 | INFO | train rank=0 | step=15020/50000 progress=30.0% loss=0.328558 lr=8.381539e-05 lr_backbone=8.381539e-05 elapsed=17:28:22 avg_step=4.19s recent_step=4.15s eta=40:41:32 +2026-09-20 23:55:39 | INFO | train rank=0 | step=15030/50000 progress=30.1% loss=0.350702 lr=8.379102e-05 lr_backbone=8.379102e-05 elapsed=17:29:03 avg_step=4.19s recent_step=4.12s eta=40:40:48 +2026-09-20 23:56:21 | INFO | train rank=0 | step=15040/50000 progress=30.1% loss=0.385615 lr=8.376664e-05 lr_backbone=8.376664e-05 elapsed=17:29:45 avg_step=4.19s recent_step=4.18s eta=40:40:06 +2026-09-20 23:57:02 | INFO | train rank=0 | step=15050/50000 progress=30.1% loss=0.373480 lr=8.374224e-05 lr_backbone=8.374224e-05 elapsed=17:30:26 avg_step=4.19s recent_step=4.10s eta=40:39:22 +2026-09-20 23:57:43 | INFO | train rank=0 | step=15060/50000 progress=30.1% loss=0.405055 lr=8.371783e-05 lr_backbone=8.371783e-05 elapsed=17:31:07 avg_step=4.19s recent_step=4.12s eta=40:38:39 +2026-09-20 23:58:25 | INFO | train rank=0 | step=15070/50000 progress=30.1% loss=0.375852 lr=8.369341e-05 lr_backbone=8.369341e-05 elapsed=17:31:48 avg_step=4.19s recent_step=4.12s eta=40:37:56 +2026-09-20 23:59:06 | INFO | train rank=0 | step=15080/50000 progress=30.2% loss=0.379785 lr=8.366897e-05 lr_backbone=8.366897e-05 elapsed=17:32:30 avg_step=4.19s recent_step=4.17s eta=40:37:13 +2026-09-20 23:59:48 | INFO | train rank=0 | step=15090/50000 progress=30.2% loss=0.328862 lr=8.364451e-05 lr_backbone=8.364451e-05 elapsed=17:33:11 avg_step=4.19s recent_step=4.13s eta=40:36:30 +2026-09-21 00:00:29 | INFO | train rank=0 | step=15100/50000 progress=30.2% loss=0.393049 lr=8.362004e-05 lr_backbone=8.362004e-05 elapsed=17:33:52 avg_step=4.19s recent_step=4.14s eta=40:35:47 +2026-09-21 00:01:10 | INFO | train rank=0 | step=15110/50000 progress=30.2% loss=0.305967 lr=8.359556e-05 lr_backbone=8.359556e-05 elapsed=17:34:33 avg_step=4.19s recent_step=4.10s eta=40:35:03 +2026-09-21 00:01:51 | INFO | train rank=0 | step=15120/50000 progress=30.2% loss=0.336707 lr=8.357106e-05 lr_backbone=8.357106e-05 elapsed=17:35:15 avg_step=4.19s recent_step=4.14s eta=40:34:20 +2026-09-21 00:02:33 | INFO | train rank=0 | step=15130/50000 progress=30.3% loss=0.314112 lr=8.354654e-05 lr_backbone=8.354654e-05 elapsed=17:35:56 avg_step=4.19s recent_step=4.15s eta=40:33:37 +2026-09-21 00:03:15 | INFO | train rank=0 | step=15140/50000 progress=30.3% loss=0.295878 lr=8.352201e-05 lr_backbone=8.352201e-05 elapsed=17:36:38 avg_step=4.19s recent_step=4.17s eta=40:32:55 +2026-09-21 00:03:56 | INFO | train rank=0 | step=15150/50000 progress=30.3% loss=0.390727 lr=8.349747e-05 lr_backbone=8.349747e-05 elapsed=17:37:19 avg_step=4.19s recent_step=4.13s eta=40:32:12 +2026-09-21 00:04:37 | INFO | train rank=0 | step=15160/50000 progress=30.3% loss=0.330842 lr=8.347291e-05 lr_backbone=8.347291e-05 elapsed=17:38:01 avg_step=4.19s recent_step=4.15s eta=40:31:29 +2026-09-21 00:05:18 | INFO | train rank=0 | step=15170/50000 progress=30.3% loss=0.385605 lr=8.344834e-05 lr_backbone=8.344834e-05 elapsed=17:38:42 avg_step=4.19s recent_step=4.10s eta=40:30:45 +2026-09-21 00:06:00 | INFO | train rank=0 | step=15180/50000 progress=30.4% loss=0.381543 lr=8.342375e-05 lr_backbone=8.342375e-05 elapsed=17:39:23 avg_step=4.19s recent_step=4.13s eta=40:30:02 +2026-09-21 00:06:41 | INFO | train rank=0 | step=15190/50000 progress=30.4% loss=0.345499 lr=8.339915e-05 lr_backbone=8.339915e-05 elapsed=17:40:05 avg_step=4.19s recent_step=4.14s eta=40:29:19 +2026-09-21 00:07:23 | INFO | train rank=0 | step=15200/50000 progress=30.4% loss=0.391924 lr=8.337453e-05 lr_backbone=8.337453e-05 elapsed=17:40:46 avg_step=4.19s recent_step=4.14s eta=40:28:36 +2026-09-21 00:08:04 | INFO | train rank=0 | step=15210/50000 progress=30.4% loss=0.337883 lr=8.334990e-05 lr_backbone=8.334990e-05 elapsed=17:41:28 avg_step=4.19s recent_step=4.17s eta=40:27:54 +2026-09-21 00:08:45 | INFO | train rank=0 | step=15220/50000 progress=30.4% loss=0.377014 lr=8.332525e-05 lr_backbone=8.332525e-05 elapsed=17:42:09 avg_step=4.19s recent_step=4.12s eta=40:27:11 +2026-09-21 00:09:26 | INFO | train rank=0 | step=15230/50000 progress=30.5% loss=0.337153 lr=8.330059e-05 lr_backbone=8.330059e-05 elapsed=17:42:50 avg_step=4.19s recent_step=4.08s eta=40:26:26 +2026-09-21 00:10:08 | INFO | train rank=0 | step=15240/50000 progress=30.5% loss=0.365649 lr=8.327592e-05 lr_backbone=8.327592e-05 elapsed=17:43:31 avg_step=4.19s recent_step=4.13s eta=40:25:43 +2026-09-21 00:10:49 | INFO | train rank=0 | step=15250/50000 progress=30.5% loss=0.412758 lr=8.325123e-05 lr_backbone=8.325123e-05 elapsed=17:44:12 avg_step=4.19s recent_step=4.14s eta=40:25:00 +2026-09-21 00:11:31 | INFO | train rank=0 | step=15260/50000 progress=30.5% loss=0.461646 lr=8.322652e-05 lr_backbone=8.322652e-05 elapsed=17:44:54 avg_step=4.19s recent_step=4.17s eta=40:24:18 +2026-09-21 00:12:12 | INFO | train rank=0 | step=15270/50000 progress=30.5% loss=0.364134 lr=8.320181e-05 lr_backbone=8.320181e-05 elapsed=17:45:35 avg_step=4.19s recent_step=4.10s eta=40:23:34 +2026-09-21 00:12:53 | INFO | train rank=0 | step=15280/50000 progress=30.6% loss=0.391768 lr=8.317707e-05 lr_backbone=8.317707e-05 elapsed=17:46:16 avg_step=4.19s recent_step=4.13s eta=40:22:51 +2026-09-21 00:13:34 | INFO | train rank=0 | step=15290/50000 progress=30.6% loss=0.368904 lr=8.315232e-05 lr_backbone=8.315232e-05 elapsed=17:46:57 avg_step=4.19s recent_step=4.11s eta=40:22:07 +2026-09-21 00:14:16 | INFO | train rank=0 | step=15300/50000 progress=30.6% loss=0.314540 lr=8.312756e-05 lr_backbone=8.312756e-05 elapsed=17:47:39 avg_step=4.19s recent_step=4.16s eta=40:21:25 +2026-09-21 00:14:57 | INFO | train rank=0 | step=15310/50000 progress=30.6% loss=0.495178 lr=8.310279e-05 lr_backbone=8.310279e-05 elapsed=17:48:20 avg_step=4.19s recent_step=4.15s eta=40:20:42 +2026-09-21 00:15:39 | INFO | train rank=0 | step=15320/50000 progress=30.6% loss=0.349842 lr=8.307799e-05 lr_backbone=8.307799e-05 elapsed=17:49:02 avg_step=4.19s recent_step=4.16s eta=40:20:00 +2026-09-21 00:16:20 | INFO | train rank=0 | step=15330/50000 progress=30.7% loss=0.363085 lr=8.305319e-05 lr_backbone=8.305319e-05 elapsed=17:49:43 avg_step=4.19s recent_step=4.12s eta=40:19:16 +2026-09-21 00:17:01 | INFO | train rank=0 | step=15340/50000 progress=30.7% loss=0.369034 lr=8.302837e-05 lr_backbone=8.302837e-05 elapsed=17:50:24 avg_step=4.19s recent_step=4.12s eta=40:18:33 +2026-09-21 00:17:42 | INFO | train rank=0 | step=15350/50000 progress=30.7% loss=0.327373 lr=8.300353e-05 lr_backbone=8.300353e-05 elapsed=17:51:05 avg_step=4.19s recent_step=4.09s eta=40:17:49 +2026-09-21 00:18:23 | INFO | train rank=0 | step=15360/50000 progress=30.7% loss=0.341852 lr=8.297868e-05 lr_backbone=8.297868e-05 elapsed=17:51:47 avg_step=4.19s recent_step=4.15s eta=40:17:06 +2026-09-21 00:19:05 | INFO | train rank=0 | step=15370/50000 progress=30.7% loss=0.334995 lr=8.295382e-05 lr_backbone=8.295382e-05 elapsed=17:52:28 avg_step=4.19s recent_step=4.16s eta=40:16:23 +2026-09-21 00:19:47 | INFO | train rank=0 | step=15380/50000 progress=30.8% loss=0.321766 lr=8.292894e-05 lr_backbone=8.292894e-05 elapsed=17:53:10 avg_step=4.19s recent_step=4.16s eta=40:15:41 +2026-09-21 00:20:28 | INFO | train rank=0 | step=15390/50000 progress=30.8% loss=0.383048 lr=8.290405e-05 lr_backbone=8.290405e-05 elapsed=17:53:51 avg_step=4.19s recent_step=4.14s eta=40:14:58 +2026-09-21 00:21:09 | INFO | train rank=0 | step=15400/50000 progress=30.8% loss=0.413587 lr=8.287914e-05 lr_backbone=8.287914e-05 elapsed=17:54:33 avg_step=4.19s recent_step=4.13s eta=40:14:15 +2026-09-21 00:21:50 | INFO | train rank=0 | step=15410/50000 progress=30.8% loss=0.309586 lr=8.285422e-05 lr_backbone=8.285422e-05 elapsed=17:55:14 avg_step=4.19s recent_step=4.10s eta=40:13:31 +2026-09-21 00:22:32 | INFO | train rank=0 | step=15420/50000 progress=30.8% loss=0.379409 lr=8.282929e-05 lr_backbone=8.282929e-05 elapsed=17:55:55 avg_step=4.19s recent_step=4.12s eta=40:12:48 +2026-09-21 00:23:13 | INFO | train rank=0 | step=15430/50000 progress=30.9% loss=0.368470 lr=8.280434e-05 lr_backbone=8.280434e-05 elapsed=17:56:36 avg_step=4.19s recent_step=4.15s eta=40:12:05 +2026-09-21 00:23:55 | INFO | train rank=0 | step=15440/50000 progress=30.9% loss=0.319595 lr=8.277937e-05 lr_backbone=8.277937e-05 elapsed=17:57:18 avg_step=4.19s recent_step=4.15s eta=40:11:23 +2026-09-21 00:24:36 | INFO | train rank=0 | step=15450/50000 progress=30.9% loss=0.465979 lr=8.275440e-05 lr_backbone=8.275440e-05 elapsed=17:58:00 avg_step=4.19s recent_step=4.16s eta=40:10:40 +2026-09-21 00:25:17 | INFO | train rank=0 | step=15460/50000 progress=30.9% loss=0.362527 lr=8.272940e-05 lr_backbone=8.272940e-05 elapsed=17:58:41 avg_step=4.19s recent_step=4.12s eta=40:09:57 +2026-09-21 00:25:58 | INFO | train rank=0 | step=15470/50000 progress=30.9% loss=0.364369 lr=8.270440e-05 lr_backbone=8.270440e-05 elapsed=17:59:22 avg_step=4.19s recent_step=4.08s eta=40:09:12 +2026-09-21 00:26:39 | INFO | train rank=0 | step=15480/50000 progress=31.0% loss=0.321237 lr=8.267937e-05 lr_backbone=8.267937e-05 elapsed=18:00:03 avg_step=4.19s recent_step=4.12s eta=40:08:29 +2026-09-21 00:27:21 | INFO | train rank=0 | step=15490/50000 progress=31.0% loss=0.338565 lr=8.265434e-05 lr_backbone=8.265434e-05 elapsed=18:00:44 avg_step=4.19s recent_step=4.11s eta=40:07:46 +2026-09-21 00:28:02 | INFO | train rank=0 | step=15500/50000 progress=31.0% loss=0.411750 lr=8.262929e-05 lr_backbone=8.262929e-05 elapsed=18:01:25 avg_step=4.19s recent_step=4.12s eta=40:07:02 +2026-09-21 00:28:43 | INFO | train rank=0 | step=15510/50000 progress=31.0% loss=0.369057 lr=8.260422e-05 lr_backbone=8.260422e-05 elapsed=18:02:07 avg_step=4.19s recent_step=4.15s eta=40:06:20 +2026-09-21 00:29:25 | INFO | train rank=0 | step=15520/50000 progress=31.0% loss=0.363678 lr=8.257915e-05 lr_backbone=8.257915e-05 elapsed=18:02:48 avg_step=4.19s recent_step=4.13s eta=40:05:37 +2026-09-21 00:30:05 | INFO | train rank=0 | step=15530/50000 progress=31.1% loss=0.360307 lr=8.255405e-05 lr_backbone=8.255405e-05 elapsed=18:03:29 avg_step=4.19s recent_step=4.07s eta=40:04:52 +2026-09-21 00:30:46 | INFO | train rank=0 | step=15540/50000 progress=31.1% loss=0.362365 lr=8.252895e-05 lr_backbone=8.252895e-05 elapsed=18:04:10 avg_step=4.19s recent_step=4.11s eta=40:04:09 +2026-09-21 00:31:28 | INFO | train rank=0 | step=15550/50000 progress=31.1% loss=0.412587 lr=8.250382e-05 lr_backbone=8.250382e-05 elapsed=18:04:51 avg_step=4.19s recent_step=4.13s eta=40:03:26 +2026-09-21 00:32:09 | INFO | train rank=0 | step=15560/50000 progress=31.1% loss=0.349805 lr=8.247869e-05 lr_backbone=8.247869e-05 elapsed=18:05:32 avg_step=4.19s recent_step=4.12s eta=40:02:42 +2026-09-21 00:32:50 | INFO | train rank=0 | step=15570/50000 progress=31.1% loss=0.333429 lr=8.245354e-05 lr_backbone=8.245354e-05 elapsed=18:06:14 avg_step=4.19s recent_step=4.13s eta=40:01:59 +2026-09-21 00:33:32 | INFO | train rank=0 | step=15580/50000 progress=31.2% loss=0.371226 lr=8.242837e-05 lr_backbone=8.242837e-05 elapsed=18:06:55 avg_step=4.19s recent_step=4.15s eta=40:01:17 +2026-09-21 00:34:13 | INFO | train rank=0 | step=15590/50000 progress=31.2% loss=0.394539 lr=8.240320e-05 lr_backbone=8.240320e-05 elapsed=18:07:36 avg_step=4.19s recent_step=4.10s eta=40:00:33 +2026-09-21 00:34:54 | INFO | train rank=0 | step=15600/50000 progress=31.2% loss=0.440290 lr=8.237800e-05 lr_backbone=8.237800e-05 elapsed=18:08:17 avg_step=4.19s recent_step=4.10s eta=39:59:49 +2026-09-21 00:35:35 | INFO | train rank=0 | step=15610/50000 progress=31.2% loss=0.334877 lr=8.235280e-05 lr_backbone=8.235280e-05 elapsed=18:08:58 avg_step=4.19s recent_step=4.13s eta=39:59:06 +2026-09-21 00:36:16 | INFO | train rank=0 | step=15620/50000 progress=31.2% loss=0.357364 lr=8.232758e-05 lr_backbone=8.232758e-05 elapsed=18:09:40 avg_step=4.19s recent_step=4.13s eta=39:58:23 +2026-09-21 00:36:58 | INFO | train rank=0 | step=15630/50000 progress=31.3% loss=0.395027 lr=8.230234e-05 lr_backbone=8.230234e-05 elapsed=18:10:21 avg_step=4.19s recent_step=4.11s eta=39:57:40 +2026-09-21 00:37:39 | INFO | train rank=0 | step=15640/50000 progress=31.3% loss=0.384780 lr=8.227709e-05 lr_backbone=8.227709e-05 elapsed=18:11:03 avg_step=4.19s recent_step=4.17s eta=39:56:57 +2026-09-21 00:38:20 | INFO | train rank=0 | step=15650/50000 progress=31.3% loss=0.314992 lr=8.225183e-05 lr_backbone=8.225183e-05 elapsed=18:11:44 avg_step=4.19s recent_step=4.10s eta=39:56:14 +2026-09-21 00:39:01 | INFO | train rank=0 | step=15660/50000 progress=31.3% loss=0.329586 lr=8.222655e-05 lr_backbone=8.222655e-05 elapsed=18:12:25 avg_step=4.19s recent_step=4.10s eta=39:55:30 +2026-09-21 00:39:42 | INFO | train rank=0 | step=15670/50000 progress=31.3% loss=0.442293 lr=8.220126e-05 lr_backbone=8.220126e-05 elapsed=18:13:06 avg_step=4.19s recent_step=4.11s eta=39:54:46 +2026-09-21 00:40:24 | INFO | train rank=0 | step=15680/50000 progress=31.4% loss=0.302720 lr=8.217596e-05 lr_backbone=8.217596e-05 elapsed=18:13:47 avg_step=4.19s recent_step=4.17s eta=39:54:04 +2026-09-21 00:41:06 | INFO | train rank=0 | step=15690/50000 progress=31.4% loss=0.383730 lr=8.215064e-05 lr_backbone=8.215064e-05 elapsed=18:14:29 avg_step=4.19s recent_step=4.18s eta=39:53:22 +2026-09-21 00:41:47 | INFO | train rank=0 | step=15700/50000 progress=31.4% loss=0.371655 lr=8.212530e-05 lr_backbone=8.212530e-05 elapsed=18:15:11 avg_step=4.19s recent_step=4.15s eta=39:52:40 +2026-09-21 00:42:28 | INFO | train rank=0 | step=15710/50000 progress=31.4% loss=0.331076 lr=8.209996e-05 lr_backbone=8.209996e-05 elapsed=18:15:52 avg_step=4.19s recent_step=4.10s eta=39:51:56 +2026-09-21 00:43:09 | INFO | train rank=0 | step=15720/50000 progress=31.4% loss=0.435842 lr=8.207460e-05 lr_backbone=8.207460e-05 elapsed=18:16:33 avg_step=4.19s recent_step=4.09s eta=39:51:12 +2026-09-21 00:43:51 | INFO | train rank=0 | step=15730/50000 progress=31.5% loss=0.356929 lr=8.204922e-05 lr_backbone=8.204922e-05 elapsed=18:17:14 avg_step=4.19s recent_step=4.15s eta=39:50:29 +2026-09-21 00:44:32 | INFO | train rank=0 | step=15740/50000 progress=31.5% loss=0.370552 lr=8.202383e-05 lr_backbone=8.202383e-05 elapsed=18:17:56 avg_step=4.19s recent_step=4.16s eta=39:49:47 +2026-09-21 00:45:14 | INFO | train rank=0 | step=15750/50000 progress=31.5% loss=0.327508 lr=8.199843e-05 lr_backbone=8.199843e-05 elapsed=18:18:37 avg_step=4.19s recent_step=4.15s eta=39:49:04 +2026-09-21 00:45:55 | INFO | train rank=0 | step=15760/50000 progress=31.5% loss=0.313700 lr=8.197301e-05 lr_backbone=8.197301e-05 elapsed=18:19:18 avg_step=4.19s recent_step=4.13s eta=39:48:21 +2026-09-21 00:46:36 | INFO | train rank=0 | step=15770/50000 progress=31.5% loss=0.357044 lr=8.194758e-05 lr_backbone=8.194758e-05 elapsed=18:19:59 avg_step=4.19s recent_step=4.10s eta=39:47:38 +2026-09-21 00:47:18 | INFO | train rank=0 | step=15780/50000 progress=31.6% loss=0.383030 lr=8.192213e-05 lr_backbone=8.192213e-05 elapsed=18:20:41 avg_step=4.19s recent_step=4.14s eta=39:46:55 +2026-09-21 00:47:59 | INFO | train rank=0 | step=15790/50000 progress=31.6% loss=0.459209 lr=8.189667e-05 lr_backbone=8.189667e-05 elapsed=18:21:22 avg_step=4.19s recent_step=4.10s eta=39:46:11 +2026-09-21 00:48:40 | INFO | train rank=0 | step=15800/50000 progress=31.6% loss=0.356880 lr=8.187120e-05 lr_backbone=8.187120e-05 elapsed=18:22:03 avg_step=4.19s recent_step=4.15s eta=39:45:28 +2026-09-21 00:49:21 | INFO | train rank=0 | step=15810/50000 progress=31.6% loss=0.360414 lr=8.184571e-05 lr_backbone=8.184571e-05 elapsed=18:22:45 avg_step=4.19s recent_step=4.14s eta=39:44:46 +2026-09-21 00:50:03 | INFO | train rank=0 | step=15820/50000 progress=31.6% loss=0.330291 lr=8.182021e-05 lr_backbone=8.182021e-05 elapsed=18:23:26 avg_step=4.19s recent_step=4.17s eta=39:44:03 +2026-09-21 00:50:44 | INFO | train rank=0 | step=15830/50000 progress=31.7% loss=0.371347 lr=8.179470e-05 lr_backbone=8.179470e-05 elapsed=18:24:07 avg_step=4.18s recent_step=4.09s eta=39:43:20 +2026-09-21 00:51:26 | INFO | train rank=0 | step=15840/50000 progress=31.7% loss=0.304905 lr=8.176917e-05 lr_backbone=8.176917e-05 elapsed=18:24:49 avg_step=4.18s recent_step=4.15s eta=39:42:37 +2026-09-21 00:52:07 | INFO | train rank=0 | step=15850/50000 progress=31.7% loss=0.394848 lr=8.174362e-05 lr_backbone=8.174362e-05 elapsed=18:25:30 avg_step=4.18s recent_step=4.12s eta=39:41:54 +2026-09-21 00:52:48 | INFO | train rank=0 | step=15860/50000 progress=31.7% loss=0.343564 lr=8.171807e-05 lr_backbone=8.171807e-05 elapsed=18:26:12 avg_step=4.18s recent_step=4.15s eta=39:41:11 +2026-09-21 00:53:30 | INFO | train rank=0 | step=15870/50000 progress=31.7% loss=0.351130 lr=8.169250e-05 lr_backbone=8.169250e-05 elapsed=18:26:53 avg_step=4.18s recent_step=4.17s eta=39:40:29 +2026-09-21 00:54:11 | INFO | train rank=0 | step=15880/50000 progress=31.8% loss=0.363813 lr=8.166691e-05 lr_backbone=8.166691e-05 elapsed=18:27:35 avg_step=4.18s recent_step=4.14s eta=39:39:46 +2026-09-21 00:54:52 | INFO | train rank=0 | step=15890/50000 progress=31.8% loss=0.326737 lr=8.164131e-05 lr_backbone=8.164131e-05 elapsed=18:28:16 avg_step=4.18s recent_step=4.11s eta=39:39:02 +2026-09-21 00:55:34 | INFO | train rank=0 | step=15900/50000 progress=31.8% loss=0.354804 lr=8.161570e-05 lr_backbone=8.161570e-05 elapsed=18:28:57 avg_step=4.18s recent_step=4.15s eta=39:38:20 +2026-09-21 00:56:15 | INFO | train rank=0 | step=15910/50000 progress=31.8% loss=0.316932 lr=8.159007e-05 lr_backbone=8.159007e-05 elapsed=18:29:39 avg_step=4.18s recent_step=4.16s eta=39:37:37 +2026-09-21 00:56:57 | INFO | train rank=0 | step=15920/50000 progress=31.8% loss=0.350945 lr=8.156443e-05 lr_backbone=8.156443e-05 elapsed=18:30:20 avg_step=4.18s recent_step=4.14s eta=39:36:55 +2026-09-21 00:57:38 | INFO | train rank=0 | step=15930/50000 progress=31.9% loss=0.333685 lr=8.153878e-05 lr_backbone=8.153878e-05 elapsed=18:31:02 avg_step=4.18s recent_step=4.16s eta=39:36:12 +2026-09-21 00:58:20 | INFO | train rank=0 | step=15940/50000 progress=31.9% loss=0.393534 lr=8.151311e-05 lr_backbone=8.151311e-05 elapsed=18:31:43 avg_step=4.18s recent_step=4.15s eta=39:35:30 +2026-09-21 00:59:01 | INFO | train rank=0 | step=15950/50000 progress=31.9% loss=0.388070 lr=8.148743e-05 lr_backbone=8.148743e-05 elapsed=18:32:24 avg_step=4.18s recent_step=4.10s eta=39:34:46 +2026-09-21 00:59:42 | INFO | train rank=0 | step=15960/50000 progress=31.9% loss=0.368240 lr=8.146174e-05 lr_backbone=8.146174e-05 elapsed=18:33:05 avg_step=4.18s recent_step=4.11s eta=39:34:02 +2026-09-21 01:00:23 | INFO | train rank=0 | step=15970/50000 progress=31.9% loss=0.338934 lr=8.143603e-05 lr_backbone=8.143603e-05 elapsed=18:33:47 avg_step=4.18s recent_step=4.15s eta=39:33:20 +2026-09-21 01:01:05 | INFO | train rank=0 | step=15980/50000 progress=32.0% loss=0.358939 lr=8.141030e-05 lr_backbone=8.141030e-05 elapsed=18:34:28 avg_step=4.18s recent_step=4.12s eta=39:32:37 +2026-09-21 01:01:46 | INFO | train rank=0 | step=15990/50000 progress=32.0% loss=0.362831 lr=8.138457e-05 lr_backbone=8.138457e-05 elapsed=18:35:10 avg_step=4.18s recent_step=4.16s eta=39:31:54 +2026-09-21 01:02:28 | INFO | train rank=0 | step=16000/50000 progress=32.0% loss=0.296504 lr=8.135882e-05 lr_backbone=8.135882e-05 elapsed=18:35:51 avg_step=4.18s recent_step=4.14s eta=39:31:11 +2026-09-21 01:03:09 | INFO | train rank=0 | step=16010/50000 progress=32.0% loss=0.329040 lr=8.133305e-05 lr_backbone=8.133305e-05 elapsed=18:36:32 avg_step=4.18s recent_step=4.10s eta=39:30:28 +2026-09-21 01:03:50 | INFO | train rank=0 | step=16020/50000 progress=32.0% loss=0.343988 lr=8.130728e-05 lr_backbone=8.130728e-05 elapsed=18:37:13 avg_step=4.18s recent_step=4.12s eta=39:29:45 +2026-09-21 01:04:31 | INFO | train rank=0 | step=16030/50000 progress=32.1% loss=0.310640 lr=8.128149e-05 lr_backbone=8.128149e-05 elapsed=18:37:55 avg_step=4.18s recent_step=4.16s eta=39:29:02 +2026-09-21 01:05:13 | INFO | train rank=0 | step=16040/50000 progress=32.1% loss=0.330376 lr=8.125568e-05 lr_backbone=8.125568e-05 elapsed=18:38:36 avg_step=4.18s recent_step=4.14s eta=39:28:19 +2026-09-21 01:05:54 | INFO | train rank=0 | step=16050/50000 progress=32.1% loss=0.331931 lr=8.122986e-05 lr_backbone=8.122986e-05 elapsed=18:39:18 avg_step=4.18s recent_step=4.15s eta=39:27:37 +2026-09-21 01:06:36 | INFO | train rank=0 | step=16060/50000 progress=32.1% loss=0.356616 lr=8.120403e-05 lr_backbone=8.120403e-05 elapsed=18:39:59 avg_step=4.18s recent_step=4.15s eta=39:26:54 +2026-09-21 01:07:17 | INFO | train rank=0 | step=16070/50000 progress=32.1% loss=0.309625 lr=8.117818e-05 lr_backbone=8.117818e-05 elapsed=18:40:40 avg_step=4.18s recent_step=4.11s eta=39:26:11 +2026-09-21 01:07:58 | INFO | train rank=0 | step=16080/50000 progress=32.2% loss=0.337266 lr=8.115232e-05 lr_backbone=8.115232e-05 elapsed=18:41:22 avg_step=4.18s recent_step=4.13s eta=39:25:28 +2026-09-21 01:08:40 | INFO | train rank=0 | step=16090/50000 progress=32.2% loss=0.320325 lr=8.112645e-05 lr_backbone=8.112645e-05 elapsed=18:42:03 avg_step=4.18s recent_step=4.15s eta=39:24:45 +2026-09-21 01:09:21 | INFO | train rank=0 | step=16100/50000 progress=32.2% loss=0.330088 lr=8.110056e-05 lr_backbone=8.110056e-05 elapsed=18:42:45 avg_step=4.18s recent_step=4.15s eta=39:24:03 +2026-09-21 01:10:03 | INFO | train rank=0 | step=16110/50000 progress=32.2% loss=0.342182 lr=8.107466e-05 lr_backbone=8.107466e-05 elapsed=18:43:26 avg_step=4.18s recent_step=4.16s eta=39:23:20 +2026-09-21 01:10:44 | INFO | train rank=0 | step=16120/50000 progress=32.2% loss=0.336275 lr=8.104875e-05 lr_backbone=8.104875e-05 elapsed=18:44:08 avg_step=4.18s recent_step=4.17s eta=39:22:38 +2026-09-21 01:11:26 | INFO | train rank=0 | step=16130/50000 progress=32.3% loss=0.300441 lr=8.102282e-05 lr_backbone=8.102282e-05 elapsed=18:44:49 avg_step=4.18s recent_step=4.11s eta=39:21:55 +2026-09-21 01:12:07 | INFO | train rank=0 | step=16140/50000 progress=32.3% loss=0.300909 lr=8.099688e-05 lr_backbone=8.099688e-05 elapsed=18:45:30 avg_step=4.18s recent_step=4.14s eta=39:21:12 +2026-09-21 01:12:49 | INFO | train rank=0 | step=16150/50000 progress=32.3% loss=0.330002 lr=8.097093e-05 lr_backbone=8.097093e-05 elapsed=18:46:12 avg_step=4.18s recent_step=4.16s eta=39:20:30 +2026-09-21 01:13:30 | INFO | train rank=0 | step=16160/50000 progress=32.3% loss=0.379498 lr=8.094496e-05 lr_backbone=8.094496e-05 elapsed=18:46:53 avg_step=4.18s recent_step=4.13s eta=39:19:47 +2026-09-21 01:14:11 | INFO | train rank=0 | step=16170/50000 progress=32.3% loss=0.314576 lr=8.091898e-05 lr_backbone=8.091898e-05 elapsed=18:47:35 avg_step=4.18s recent_step=4.15s eta=39:19:04 +2026-09-21 01:14:53 | INFO | train rank=0 | step=16180/50000 progress=32.4% loss=0.356643 lr=8.089298e-05 lr_backbone=8.089298e-05 elapsed=18:48:16 avg_step=4.18s recent_step=4.15s eta=39:18:22 +2026-09-21 01:15:34 | INFO | train rank=0 | step=16190/50000 progress=32.4% loss=0.332039 lr=8.086697e-05 lr_backbone=8.086697e-05 elapsed=18:48:57 avg_step=4.18s recent_step=4.10s eta=39:17:38 +2026-09-21 01:16:16 | INFO | train rank=0 | step=16200/50000 progress=32.4% loss=0.340741 lr=8.084095e-05 lr_backbone=8.084095e-05 elapsed=18:49:39 avg_step=4.18s recent_step=4.16s eta=39:16:56 +2026-09-21 01:16:57 | INFO | train rank=0 | step=16210/50000 progress=32.4% loss=0.337407 lr=8.081491e-05 lr_backbone=8.081491e-05 elapsed=18:50:20 avg_step=4.18s recent_step=4.15s eta=39:16:13 +2026-09-21 01:17:38 | INFO | train rank=0 | step=16220/50000 progress=32.4% loss=0.372585 lr=8.078886e-05 lr_backbone=8.078886e-05 elapsed=18:51:02 avg_step=4.18s recent_step=4.13s eta=39:15:30 +2026-09-21 01:18:20 | INFO | train rank=0 | step=16230/50000 progress=32.5% loss=0.418055 lr=8.076280e-05 lr_backbone=8.076280e-05 elapsed=18:51:43 avg_step=4.18s recent_step=4.14s eta=39:14:48 +2026-09-21 01:19:01 | INFO | train rank=0 | step=16240/50000 progress=32.5% loss=0.408817 lr=8.073673e-05 lr_backbone=8.073673e-05 elapsed=18:52:25 avg_step=4.18s recent_step=4.14s eta=39:14:05 +2026-09-21 01:19:42 | INFO | train rank=0 | step=16250/50000 progress=32.5% loss=0.347714 lr=8.071064e-05 lr_backbone=8.071064e-05 elapsed=18:53:06 avg_step=4.18s recent_step=4.10s eta=39:13:21 +2026-09-21 01:20:23 | INFO | train rank=0 | step=16260/50000 progress=32.5% loss=0.373328 lr=8.068453e-05 lr_backbone=8.068453e-05 elapsed=18:53:47 avg_step=4.18s recent_step=4.12s eta=39:12:38 +2026-09-21 01:21:05 | INFO | train rank=0 | step=16270/50000 progress=32.5% loss=0.309887 lr=8.065842e-05 lr_backbone=8.065842e-05 elapsed=18:54:28 avg_step=4.18s recent_step=4.15s eta=39:11:56 +2026-09-21 01:21:47 | INFO | train rank=0 | step=16280/50000 progress=32.6% loss=0.427369 lr=8.063229e-05 lr_backbone=8.063229e-05 elapsed=18:55:10 avg_step=4.18s recent_step=4.17s eta=39:11:13 +2026-09-21 01:22:28 | INFO | train rank=0 | step=16290/50000 progress=32.6% loss=0.375362 lr=8.060614e-05 lr_backbone=8.060614e-05 elapsed=18:55:51 avg_step=4.18s recent_step=4.14s eta=39:10:31 +2026-09-21 01:23:09 | INFO | train rank=0 | step=16300/50000 progress=32.6% loss=0.437922 lr=8.057999e-05 lr_backbone=8.057999e-05 elapsed=18:56:33 avg_step=4.18s recent_step=4.15s eta=39:09:48 +2026-09-21 01:23:51 | INFO | train rank=0 | step=16310/50000 progress=32.6% loss=0.331829 lr=8.055382e-05 lr_backbone=8.055382e-05 elapsed=18:57:14 avg_step=4.18s recent_step=4.11s eta=39:09:05 +2026-09-21 01:24:32 | INFO | train rank=0 | step=16320/50000 progress=32.6% loss=0.371231 lr=8.052763e-05 lr_backbone=8.052763e-05 elapsed=18:57:55 avg_step=4.18s recent_step=4.13s eta=39:08:22 +2026-09-21 01:25:13 | INFO | train rank=0 | step=16330/50000 progress=32.7% loss=0.320558 lr=8.050144e-05 lr_backbone=8.050144e-05 elapsed=18:58:37 avg_step=4.18s recent_step=4.15s eta=39:07:39 +2026-09-21 01:25:55 | INFO | train rank=0 | step=16340/50000 progress=32.7% loss=0.378325 lr=8.047523e-05 lr_backbone=8.047523e-05 elapsed=18:59:18 avg_step=4.18s recent_step=4.17s eta=39:06:57 +2026-09-21 01:26:36 | INFO | train rank=0 | step=16350/50000 progress=32.7% loss=0.371958 lr=8.044900e-05 lr_backbone=8.044900e-05 elapsed=19:00:00 avg_step=4.18s recent_step=4.14s eta=39:06:14 +2026-09-21 01:27:18 | INFO | train rank=0 | step=16360/50000 progress=32.7% loss=0.383658 lr=8.042277e-05 lr_backbone=8.042277e-05 elapsed=19:00:41 avg_step=4.18s recent_step=4.13s eta=39:05:31 +2026-09-21 01:27:59 | INFO | train rank=0 | step=16370/50000 progress=32.7% loss=0.342007 lr=8.039651e-05 lr_backbone=8.039651e-05 elapsed=19:01:22 avg_step=4.18s recent_step=4.10s eta=39:04:48 +2026-09-21 01:28:40 | INFO | train rank=0 | step=16380/50000 progress=32.8% loss=0.387824 lr=8.037025e-05 lr_backbone=8.037025e-05 elapsed=19:02:03 avg_step=4.18s recent_step=4.14s eta=39:04:05 +2026-09-21 01:29:22 | INFO | train rank=0 | step=16390/50000 progress=32.8% loss=0.363880 lr=8.034397e-05 lr_backbone=8.034397e-05 elapsed=19:02:45 avg_step=4.18s recent_step=4.17s eta=39:03:23 +2026-09-21 01:30:03 | INFO | train rank=0 | step=16400/50000 progress=32.8% loss=0.335259 lr=8.031768e-05 lr_backbone=8.031768e-05 elapsed=19:03:26 avg_step=4.18s recent_step=4.12s eta=39:02:40 +2026-09-21 01:30:45 | INFO | train rank=0 | step=16410/50000 progress=32.8% loss=0.393243 lr=8.029138e-05 lr_backbone=8.029138e-05 elapsed=19:04:08 avg_step=4.18s recent_step=4.15s eta=39:01:57 +2026-09-21 01:31:26 | INFO | train rank=0 | step=16420/50000 progress=32.8% loss=0.363206 lr=8.026507e-05 lr_backbone=8.026507e-05 elapsed=19:04:49 avg_step=4.18s recent_step=4.14s eta=39:01:15 +2026-09-21 01:32:07 | INFO | train rank=0 | step=16430/50000 progress=32.9% loss=0.368823 lr=8.023874e-05 lr_backbone=8.023874e-05 elapsed=19:05:30 avg_step=4.18s recent_step=4.11s eta=39:00:31 +2026-09-21 01:32:48 | INFO | train rank=0 | step=16440/50000 progress=32.9% loss=0.322915 lr=8.021239e-05 lr_backbone=8.021239e-05 elapsed=19:06:12 avg_step=4.18s recent_step=4.13s eta=38:59:48 +2026-09-21 01:33:30 | INFO | train rank=0 | step=16450/50000 progress=32.9% loss=0.371299 lr=8.018604e-05 lr_backbone=8.018604e-05 elapsed=19:06:53 avg_step=4.18s recent_step=4.18s eta=38:59:06 +2026-09-21 01:34:11 | INFO | train rank=0 | step=16460/50000 progress=32.9% loss=0.294376 lr=8.015967e-05 lr_backbone=8.015967e-05 elapsed=19:07:35 avg_step=4.18s recent_step=4.12s eta=38:58:23 +2026-09-21 01:34:53 | INFO | train rank=0 | step=16470/50000 progress=32.9% loss=0.373932 lr=8.013328e-05 lr_backbone=8.013328e-05 elapsed=19:08:16 avg_step=4.18s recent_step=4.13s eta=38:57:40 +2026-09-21 01:35:34 | INFO | train rank=0 | step=16480/50000 progress=33.0% loss=0.341046 lr=8.010689e-05 lr_backbone=8.010689e-05 elapsed=19:08:57 avg_step=4.18s recent_step=4.15s eta=38:56:58 +2026-09-21 01:36:15 | INFO | train rank=0 | step=16490/50000 progress=33.0% loss=0.394241 lr=8.008048e-05 lr_backbone=8.008048e-05 elapsed=19:09:39 avg_step=4.18s recent_step=4.12s eta=38:56:15 +2026-09-21 01:36:57 | INFO | train rank=0 | step=16500/50000 progress=33.0% loss=0.346036 lr=8.005406e-05 lr_backbone=8.005406e-05 elapsed=19:10:20 avg_step=4.18s recent_step=4.15s eta=38:55:32 +2026-09-21 01:37:38 | INFO | train rank=0 | step=16510/50000 progress=33.0% loss=0.343537 lr=8.002762e-05 lr_backbone=8.002762e-05 elapsed=19:11:02 avg_step=4.18s recent_step=4.16s eta=38:54:50 +2026-09-21 01:38:20 | INFO | train rank=0 | step=16520/50000 progress=33.0% loss=0.304961 lr=8.000117e-05 lr_backbone=8.000117e-05 elapsed=19:11:43 avg_step=4.18s recent_step=4.15s eta=38:54:07 +2026-09-21 01:39:01 | INFO | train rank=0 | step=16530/50000 progress=33.1% loss=0.337103 lr=7.997471e-05 lr_backbone=7.997471e-05 elapsed=19:12:24 avg_step=4.18s recent_step=4.14s eta=38:53:25 +2026-09-21 01:39:42 | INFO | train rank=0 | step=16540/50000 progress=33.1% loss=0.354678 lr=7.994824e-05 lr_backbone=7.994824e-05 elapsed=19:13:06 avg_step=4.18s recent_step=4.12s eta=38:52:41 +2026-09-21 01:40:23 | INFO | train rank=0 | step=16550/50000 progress=33.1% loss=0.374103 lr=7.992175e-05 lr_backbone=7.992175e-05 elapsed=19:13:47 avg_step=4.18s recent_step=4.10s eta=38:51:58 +2026-09-21 01:41:05 | INFO | train rank=0 | step=16560/50000 progress=33.1% loss=0.348364 lr=7.989525e-05 lr_backbone=7.989525e-05 elapsed=19:14:28 avg_step=4.18s recent_step=4.12s eta=38:51:15 +2026-09-21 01:41:46 | INFO | train rank=0 | step=16570/50000 progress=33.1% loss=0.304639 lr=7.986874e-05 lr_backbone=7.986874e-05 elapsed=19:15:09 avg_step=4.18s recent_step=4.15s eta=38:50:32 +2026-09-21 01:42:27 | INFO | train rank=0 | step=16580/50000 progress=33.2% loss=0.409885 lr=7.984221e-05 lr_backbone=7.984221e-05 elapsed=19:15:51 avg_step=4.18s recent_step=4.13s eta=38:49:49 +2026-09-21 01:43:09 | INFO | train rank=0 | step=16590/50000 progress=33.2% loss=0.355543 lr=7.981567e-05 lr_backbone=7.981567e-05 elapsed=19:16:32 avg_step=4.18s recent_step=4.17s eta=38:49:07 +2026-09-21 01:43:50 | INFO | train rank=0 | step=16600/50000 progress=33.2% loss=0.317728 lr=7.978912e-05 lr_backbone=7.978912e-05 elapsed=19:17:14 avg_step=4.18s recent_step=4.13s eta=38:48:24 +2026-09-21 01:44:31 | INFO | train rank=0 | step=16610/50000 progress=33.2% loss=0.353926 lr=7.976255e-05 lr_backbone=7.976255e-05 elapsed=19:17:55 avg_step=4.18s recent_step=4.09s eta=38:47:41 +2026-09-21 01:45:13 | INFO | train rank=0 | step=16620/50000 progress=33.2% loss=0.337250 lr=7.973597e-05 lr_backbone=7.973597e-05 elapsed=19:18:36 avg_step=4.18s recent_step=4.15s eta=38:46:58 +2026-09-21 01:45:54 | INFO | train rank=0 | step=16630/50000 progress=33.3% loss=0.327040 lr=7.970938e-05 lr_backbone=7.970938e-05 elapsed=19:19:17 avg_step=4.18s recent_step=4.14s eta=38:46:16 +2026-09-21 01:46:36 | INFO | train rank=0 | step=16640/50000 progress=33.3% loss=0.394141 lr=7.968277e-05 lr_backbone=7.968277e-05 elapsed=19:19:59 avg_step=4.18s recent_step=4.16s eta=38:45:33 +2026-09-21 01:47:17 | INFO | train rank=0 | step=16650/50000 progress=33.3% loss=0.312852 lr=7.965615e-05 lr_backbone=7.965615e-05 elapsed=19:20:41 avg_step=4.18s recent_step=4.16s eta=38:44:51 +2026-09-21 01:47:59 | INFO | train rank=0 | step=16660/50000 progress=33.3% loss=0.354902 lr=7.962952e-05 lr_backbone=7.962952e-05 elapsed=19:21:22 avg_step=4.18s recent_step=4.13s eta=38:44:08 +2026-09-21 01:48:39 | INFO | train rank=0 | step=16670/50000 progress=33.3% loss=0.401519 lr=7.960288e-05 lr_backbone=7.960288e-05 elapsed=19:22:03 avg_step=4.18s recent_step=4.09s eta=38:43:24 +2026-09-21 01:49:21 | INFO | train rank=0 | step=16680/50000 progress=33.4% loss=0.360223 lr=7.957622e-05 lr_backbone=7.957622e-05 elapsed=19:22:44 avg_step=4.18s recent_step=4.13s eta=38:42:41 +2026-09-21 01:50:02 | INFO | train rank=0 | step=16690/50000 progress=33.4% loss=0.455921 lr=7.954955e-05 lr_backbone=7.954955e-05 elapsed=19:23:26 avg_step=4.18s recent_step=4.17s eta=38:41:59 +2026-09-21 01:50:44 | INFO | train rank=0 | step=16700/50000 progress=33.4% loss=0.378276 lr=7.952287e-05 lr_backbone=7.952287e-05 elapsed=19:24:07 avg_step=4.18s recent_step=4.14s eta=38:41:17 +2026-09-21 01:51:25 | INFO | train rank=0 | step=16710/50000 progress=33.4% loss=0.370350 lr=7.949617e-05 lr_backbone=7.949617e-05 elapsed=19:24:49 avg_step=4.18s recent_step=4.16s eta=38:40:34 +2026-09-21 01:52:07 | INFO | train rank=0 | step=16720/50000 progress=33.4% loss=0.335311 lr=7.946947e-05 lr_backbone=7.946947e-05 elapsed=19:25:30 avg_step=4.18s recent_step=4.11s eta=38:39:51 +2026-09-21 01:52:48 | INFO | train rank=0 | step=16730/50000 progress=33.5% loss=0.352472 lr=7.944274e-05 lr_backbone=7.944274e-05 elapsed=19:26:11 avg_step=4.18s recent_step=4.11s eta=38:39:08 +2026-09-21 01:53:29 | INFO | train rank=0 | step=16740/50000 progress=33.5% loss=0.351335 lr=7.941601e-05 lr_backbone=7.941601e-05 elapsed=19:26:52 avg_step=4.18s recent_step=4.10s eta=38:38:24 +2026-09-21 01:54:10 | INFO | train rank=0 | step=16750/50000 progress=33.5% loss=0.360483 lr=7.938926e-05 lr_backbone=7.938926e-05 elapsed=19:27:33 avg_step=4.18s recent_step=4.14s eta=38:37:42 +2026-09-21 01:54:51 | INFO | train rank=0 | step=16760/50000 progress=33.5% loss=0.378860 lr=7.936250e-05 lr_backbone=7.936250e-05 elapsed=19:28:15 avg_step=4.18s recent_step=4.14s eta=38:36:59 +2026-09-21 01:55:33 | INFO | train rank=0 | step=16770/50000 progress=33.5% loss=0.343059 lr=7.933573e-05 lr_backbone=7.933573e-05 elapsed=19:28:56 avg_step=4.18s recent_step=4.15s eta=38:36:17 +2026-09-21 01:56:15 | INFO | train rank=0 | step=16780/50000 progress=33.6% loss=0.335664 lr=7.930894e-05 lr_backbone=7.930894e-05 elapsed=19:29:38 avg_step=4.18s recent_step=4.16s eta=38:35:34 +2026-09-21 01:56:56 | INFO | train rank=0 | step=16790/50000 progress=33.6% loss=0.318328 lr=7.928215e-05 lr_backbone=7.928215e-05 elapsed=19:30:19 avg_step=4.18s recent_step=4.11s eta=38:34:51 +2026-09-21 01:57:37 | INFO | train rank=0 | step=16800/50000 progress=33.6% loss=0.307249 lr=7.925533e-05 lr_backbone=7.925533e-05 elapsed=19:31:00 avg_step=4.18s recent_step=4.10s eta=38:34:07 +2026-09-21 01:58:18 | INFO | train rank=0 | step=16810/50000 progress=33.6% loss=0.326034 lr=7.922851e-05 lr_backbone=7.922851e-05 elapsed=19:31:41 avg_step=4.18s recent_step=4.15s eta=38:33:25 +2026-09-21 01:58:59 | INFO | train rank=0 | step=16820/50000 progress=33.6% loss=0.361186 lr=7.920167e-05 lr_backbone=7.920167e-05 elapsed=19:32:23 avg_step=4.18s recent_step=4.14s eta=38:32:42 +2026-09-21 01:59:41 | INFO | train rank=0 | step=16830/50000 progress=33.7% loss=0.327737 lr=7.917482e-05 lr_backbone=7.917482e-05 elapsed=19:33:04 avg_step=4.18s recent_step=4.12s eta=38:31:59 +2026-09-21 02:00:22 | INFO | train rank=0 | step=16840/50000 progress=33.7% loss=0.352252 lr=7.914796e-05 lr_backbone=7.914796e-05 elapsed=19:33:45 avg_step=4.18s recent_step=4.14s eta=38:31:17 +2026-09-21 02:01:03 | INFO | train rank=0 | step=16850/50000 progress=33.7% loss=0.311273 lr=7.912108e-05 lr_backbone=7.912108e-05 elapsed=19:34:26 avg_step=4.18s recent_step=4.09s eta=38:30:33 +2026-09-21 02:01:44 | INFO | train rank=0 | step=16860/50000 progress=33.7% loss=0.302739 lr=7.909420e-05 lr_backbone=7.909420e-05 elapsed=19:35:07 avg_step=4.18s recent_step=4.10s eta=38:29:50 +2026-09-21 02:02:25 | INFO | train rank=0 | step=16870/50000 progress=33.7% loss=0.404058 lr=7.906730e-05 lr_backbone=7.906730e-05 elapsed=19:35:49 avg_step=4.18s recent_step=4.13s eta=38:29:07 +2026-09-21 02:03:06 | INFO | train rank=0 | step=16880/50000 progress=33.8% loss=0.323953 lr=7.904038e-05 lr_backbone=7.904038e-05 elapsed=19:36:30 avg_step=4.18s recent_step=4.11s eta=38:28:24 +2026-09-21 02:03:48 | INFO | train rank=0 | step=16890/50000 progress=33.8% loss=0.379230 lr=7.901346e-05 lr_backbone=7.901346e-05 elapsed=19:37:11 avg_step=4.18s recent_step=4.13s eta=38:27:41 +2026-09-21 02:04:29 | INFO | train rank=0 | step=16900/50000 progress=33.8% loss=0.354711 lr=7.898652e-05 lr_backbone=7.898652e-05 elapsed=19:37:52 avg_step=4.18s recent_step=4.12s eta=38:26:58 +2026-09-21 02:05:10 | INFO | train rank=0 | step=16910/50000 progress=33.8% loss=0.384918 lr=7.895957e-05 lr_backbone=7.895957e-05 elapsed=19:38:33 avg_step=4.18s recent_step=4.08s eta=38:26:14 +2026-09-21 02:05:51 | INFO | train rank=0 | step=16920/50000 progress=33.8% loss=0.323606 lr=7.893260e-05 lr_backbone=7.893260e-05 elapsed=19:39:14 avg_step=4.18s recent_step=4.10s eta=38:25:30 +2026-09-21 02:06:32 | INFO | train rank=0 | step=16930/50000 progress=33.9% loss=0.358565 lr=7.890562e-05 lr_backbone=7.890562e-05 elapsed=19:39:55 avg_step=4.18s recent_step=4.13s eta=38:24:48 +2026-09-21 02:07:13 | INFO | train rank=0 | step=16940/50000 progress=33.9% loss=0.354310 lr=7.887864e-05 lr_backbone=7.887864e-05 elapsed=19:40:37 avg_step=4.18s recent_step=4.15s eta=38:24:05 +2026-09-21 02:07:55 | INFO | train rank=0 | step=16950/50000 progress=33.9% loss=0.329446 lr=7.885163e-05 lr_backbone=7.885163e-05 elapsed=19:41:18 avg_step=4.18s recent_step=4.13s eta=38:23:22 +2026-09-21 02:08:36 | INFO | train rank=0 | step=16960/50000 progress=33.9% loss=0.337267 lr=7.882462e-05 lr_backbone=7.882462e-05 elapsed=19:41:59 avg_step=4.18s recent_step=4.12s eta=38:22:39 +2026-09-21 02:09:17 | INFO | train rank=0 | step=16970/50000 progress=33.9% loss=0.332932 lr=7.879759e-05 lr_backbone=7.879759e-05 elapsed=19:42:40 avg_step=4.18s recent_step=4.08s eta=38:21:56 +2026-09-21 02:09:58 | INFO | train rank=0 | step=16980/50000 progress=34.0% loss=0.352982 lr=7.877055e-05 lr_backbone=7.877055e-05 elapsed=19:43:21 avg_step=4.18s recent_step=4.12s eta=38:21:13 +2026-09-21 02:10:39 | INFO | train rank=0 | step=16990/50000 progress=34.0% loss=0.339378 lr=7.874350e-05 lr_backbone=7.874350e-05 elapsed=19:44:03 avg_step=4.18s recent_step=4.12s eta=38:20:30 +2026-09-21 02:11:20 | INFO | train rank=0 | step=17000/50000 progress=34.0% loss=0.349901 lr=7.871643e-05 lr_backbone=7.871643e-05 elapsed=19:44:44 avg_step=4.18s recent_step=4.13s eta=38:19:47 +2026-09-21 02:12:02 | INFO | train rank=0 | step=17010/50000 progress=34.0% loss=0.424420 lr=7.868936e-05 lr_backbone=7.868936e-05 elapsed=19:45:25 avg_step=4.18s recent_step=4.15s eta=38:19:04 +2026-09-21 02:12:43 | INFO | train rank=0 | step=17020/50000 progress=34.0% loss=0.305776 lr=7.866227e-05 lr_backbone=7.866227e-05 elapsed=19:46:06 avg_step=4.18s recent_step=4.11s eta=38:18:21 +2026-09-21 02:13:24 | INFO | train rank=0 | step=17030/50000 progress=34.1% loss=0.389903 lr=7.863516e-05 lr_backbone=7.863516e-05 elapsed=19:46:47 avg_step=4.18s recent_step=4.10s eta=38:17:38 +2026-09-21 02:14:05 | INFO | train rank=0 | step=17040/50000 progress=34.1% loss=0.315092 lr=7.860805e-05 lr_backbone=7.860805e-05 elapsed=19:47:29 avg_step=4.18s recent_step=4.11s eta=38:16:55 +2026-09-21 02:14:47 | INFO | train rank=0 | step=17050/50000 progress=34.1% loss=0.324056 lr=7.858092e-05 lr_backbone=7.858092e-05 elapsed=19:48:10 avg_step=4.18s recent_step=4.13s eta=38:16:12 +2026-09-21 02:15:28 | INFO | train rank=0 | step=17060/50000 progress=34.1% loss=0.343825 lr=7.855378e-05 lr_backbone=7.855378e-05 elapsed=19:48:51 avg_step=4.18s recent_step=4.13s eta=38:15:29 +2026-09-21 02:16:09 | INFO | train rank=0 | step=17070/50000 progress=34.1% loss=0.452806 lr=7.852663e-05 lr_backbone=7.852663e-05 elapsed=19:49:32 avg_step=4.18s recent_step=4.10s eta=38:14:46 +2026-09-21 02:16:50 | INFO | train rank=0 | step=17080/50000 progress=34.2% loss=0.327002 lr=7.849946e-05 lr_backbone=7.849946e-05 elapsed=19:50:13 avg_step=4.18s recent_step=4.13s eta=38:14:03 +2026-09-21 02:17:31 | INFO | train rank=0 | step=17090/50000 progress=34.2% loss=0.422610 lr=7.847228e-05 lr_backbone=7.847228e-05 elapsed=19:50:54 avg_step=4.18s recent_step=4.10s eta=38:13:19 +2026-09-21 02:18:12 | INFO | train rank=0 | step=17100/50000 progress=34.2% loss=0.474407 lr=7.844509e-05 lr_backbone=7.844509e-05 elapsed=19:51:36 avg_step=4.18s recent_step=4.12s eta=38:12:36 +2026-09-21 02:18:54 | INFO | train rank=0 | step=17110/50000 progress=34.2% loss=0.310527 lr=7.841789e-05 lr_backbone=7.841789e-05 elapsed=19:52:17 avg_step=4.18s recent_step=4.14s eta=38:11:54 +2026-09-21 02:19:35 | INFO | train rank=0 | step=17120/50000 progress=34.2% loss=0.381338 lr=7.839067e-05 lr_backbone=7.839067e-05 elapsed=19:52:59 avg_step=4.18s recent_step=4.17s eta=38:11:12 +2026-09-21 02:20:17 | INFO | train rank=0 | step=17130/50000 progress=34.3% loss=0.374878 lr=7.836345e-05 lr_backbone=7.836345e-05 elapsed=19:53:40 avg_step=4.18s recent_step=4.15s eta=38:10:29 +2026-09-21 02:20:58 | INFO | train rank=0 | step=17140/50000 progress=34.3% loss=0.353429 lr=7.833621e-05 lr_backbone=7.833621e-05 elapsed=19:54:22 avg_step=4.18s recent_step=4.15s eta=38:09:47 +2026-09-21 02:21:39 | INFO | train rank=0 | step=17150/50000 progress=34.3% loss=0.414012 lr=7.830896e-05 lr_backbone=7.830896e-05 elapsed=19:55:03 avg_step=4.18s recent_step=4.11s eta=38:09:04 +2026-09-21 02:22:21 | INFO | train rank=0 | step=17160/50000 progress=34.3% loss=0.360266 lr=7.828169e-05 lr_backbone=7.828169e-05 elapsed=19:55:44 avg_step=4.18s recent_step=4.13s eta=38:08:21 +2026-09-21 02:23:02 | INFO | train rank=0 | step=17170/50000 progress=34.3% loss=0.376039 lr=7.825441e-05 lr_backbone=7.825441e-05 elapsed=19:56:26 avg_step=4.18s recent_step=4.15s eta=38:07:39 +2026-09-21 02:23:44 | INFO | train rank=0 | step=17180/50000 progress=34.4% loss=0.451621 lr=7.822712e-05 lr_backbone=7.822712e-05 elapsed=19:57:07 avg_step=4.18s recent_step=4.17s eta=38:06:57 +2026-09-21 02:24:26 | INFO | train rank=0 | step=17190/50000 progress=34.4% loss=0.347542 lr=7.819982e-05 lr_backbone=7.819982e-05 elapsed=19:57:49 avg_step=4.18s recent_step=4.16s eta=38:06:14 +2026-09-21 02:25:07 | INFO | train rank=0 | step=17200/50000 progress=34.4% loss=0.396821 lr=7.817251e-05 lr_backbone=7.817251e-05 elapsed=19:58:30 avg_step=4.18s recent_step=4.12s eta=38:05:31 +2026-09-21 02:25:48 | INFO | train rank=0 | step=17210/50000 progress=34.4% loss=0.377343 lr=7.814518e-05 lr_backbone=7.814518e-05 elapsed=19:59:11 avg_step=4.18s recent_step=4.09s eta=38:04:48 +2026-09-21 02:26:29 | INFO | train rank=0 | step=17220/50000 progress=34.4% loss=0.324653 lr=7.811784e-05 lr_backbone=7.811784e-05 elapsed=19:59:52 avg_step=4.18s recent_step=4.13s eta=38:04:05 +2026-09-21 02:27:10 | INFO | train rank=0 | step=17230/50000 progress=34.5% loss=0.366857 lr=7.809049e-05 lr_backbone=7.809049e-05 elapsed=20:00:34 avg_step=4.18s recent_step=4.13s eta=38:03:22 +2026-09-21 02:27:52 | INFO | train rank=0 | step=17240/50000 progress=34.5% loss=0.326683 lr=7.806313e-05 lr_backbone=7.806313e-05 elapsed=20:01:15 avg_step=4.18s recent_step=4.15s eta=38:02:40 +2026-09-21 02:28:33 | INFO | train rank=0 | step=17250/50000 progress=34.5% loss=0.400481 lr=7.803575e-05 lr_backbone=7.803575e-05 elapsed=20:01:57 avg_step=4.18s recent_step=4.16s eta=38:01:58 +2026-09-21 02:29:15 | INFO | train rank=0 | step=17260/50000 progress=34.5% loss=0.345845 lr=7.800836e-05 lr_backbone=7.800836e-05 elapsed=20:02:38 avg_step=4.18s recent_step=4.15s eta=38:01:16 +2026-09-21 02:29:56 | INFO | train rank=0 | step=17270/50000 progress=34.5% loss=0.351766 lr=7.798096e-05 lr_backbone=7.798096e-05 elapsed=20:03:19 avg_step=4.18s recent_step=4.11s eta=38:00:32 +2026-09-21 02:30:37 | INFO | train rank=0 | step=17280/50000 progress=34.6% loss=0.393775 lr=7.795355e-05 lr_backbone=7.795355e-05 elapsed=20:04:01 avg_step=4.18s recent_step=4.14s eta=37:59:50 +2026-09-21 02:31:19 | INFO | train rank=0 | step=17290/50000 progress=34.6% loss=0.427117 lr=7.792613e-05 lr_backbone=7.792613e-05 elapsed=20:04:42 avg_step=4.18s recent_step=4.11s eta=37:59:07 +2026-09-21 02:32:00 | INFO | train rank=0 | step=17300/50000 progress=34.6% loss=0.274151 lr=7.789869e-05 lr_backbone=7.789869e-05 elapsed=20:05:23 avg_step=4.18s recent_step=4.16s eta=37:58:24 +2026-09-21 02:32:42 | INFO | train rank=0 | step=17310/50000 progress=34.6% loss=0.437015 lr=7.787124e-05 lr_backbone=7.787124e-05 elapsed=20:06:05 avg_step=4.18s recent_step=4.16s eta=37:57:42 +2026-09-21 02:33:24 | INFO | train rank=0 | step=17320/50000 progress=34.6% loss=0.352054 lr=7.784378e-05 lr_backbone=7.784378e-05 elapsed=20:06:47 avg_step=4.18s recent_step=4.18s eta=37:57:00 +2026-09-21 02:34:05 | INFO | train rank=0 | step=17330/50000 progress=34.7% loss=0.406010 lr=7.781631e-05 lr_backbone=7.781631e-05 elapsed=20:07:28 avg_step=4.18s recent_step=4.11s eta=37:56:17 +2026-09-21 02:34:46 | INFO | train rank=0 | step=17340/50000 progress=34.7% loss=0.322652 lr=7.778882e-05 lr_backbone=7.778882e-05 elapsed=20:08:09 avg_step=4.18s recent_step=4.14s eta=37:55:35 +2026-09-21 02:35:28 | INFO | train rank=0 | step=17350/50000 progress=34.7% loss=0.326001 lr=7.776132e-05 lr_backbone=7.776132e-05 elapsed=20:08:51 avg_step=4.18s recent_step=4.16s eta=37:54:53 +2026-09-21 02:36:09 | INFO | train rank=0 | step=17360/50000 progress=34.7% loss=0.334219 lr=7.773381e-05 lr_backbone=7.773381e-05 elapsed=20:09:32 avg_step=4.18s recent_step=4.11s eta=37:54:09 +2026-09-21 02:36:50 | INFO | train rank=0 | step=17370/50000 progress=34.7% loss=0.381211 lr=7.770629e-05 lr_backbone=7.770629e-05 elapsed=20:10:14 avg_step=4.18s recent_step=4.15s eta=37:53:27 +2026-09-21 02:37:32 | INFO | train rank=0 | step=17380/50000 progress=34.8% loss=0.396382 lr=7.767876e-05 lr_backbone=7.767876e-05 elapsed=20:10:55 avg_step=4.18s recent_step=4.15s eta=37:52:45 +2026-09-21 02:38:13 | INFO | train rank=0 | step=17390/50000 progress=34.8% loss=0.373266 lr=7.765121e-05 lr_backbone=7.765121e-05 elapsed=20:11:36 avg_step=4.18s recent_step=4.09s eta=37:52:01 +2026-09-21 02:38:54 | INFO | train rank=0 | step=17400/50000 progress=34.8% loss=0.323881 lr=7.762365e-05 lr_backbone=7.762365e-05 elapsed=20:12:17 avg_step=4.18s recent_step=4.14s eta=37:51:19 +2026-09-21 02:39:36 | INFO | train rank=0 | step=17410/50000 progress=34.8% loss=0.325330 lr=7.759608e-05 lr_backbone=7.759608e-05 elapsed=20:12:59 avg_step=4.18s recent_step=4.15s eta=37:50:36 +2026-09-21 02:40:17 | INFO | train rank=0 | step=17420/50000 progress=34.8% loss=0.274275 lr=7.756850e-05 lr_backbone=7.756850e-05 elapsed=20:13:40 avg_step=4.18s recent_step=4.12s eta=37:49:53 +2026-09-21 02:40:58 | INFO | train rank=0 | step=17430/50000 progress=34.9% loss=0.404316 lr=7.754091e-05 lr_backbone=7.754091e-05 elapsed=20:14:22 avg_step=4.18s recent_step=4.16s eta=37:49:11 +2026-09-21 02:41:40 | INFO | train rank=0 | step=17440/50000 progress=34.9% loss=0.405458 lr=7.751330e-05 lr_backbone=7.751330e-05 elapsed=20:15:03 avg_step=4.18s recent_step=4.17s eta=37:48:29 +2026-09-21 02:42:21 | INFO | train rank=0 | step=17450/50000 progress=34.9% loss=0.366567 lr=7.748568e-05 lr_backbone=7.748568e-05 elapsed=20:15:45 avg_step=4.18s recent_step=4.10s eta=37:47:46 +2026-09-21 02:43:02 | INFO | train rank=0 | step=17460/50000 progress=34.9% loss=0.338261 lr=7.745805e-05 lr_backbone=7.745805e-05 elapsed=20:16:26 avg_step=4.18s recent_step=4.13s eta=37:47:03 +2026-09-21 02:43:44 | INFO | train rank=0 | step=17470/50000 progress=34.9% loss=0.410540 lr=7.743041e-05 lr_backbone=7.743041e-05 elapsed=20:17:07 avg_step=4.18s recent_step=4.16s eta=37:46:21 +2026-09-21 02:44:25 | INFO | train rank=0 | step=17480/50000 progress=35.0% loss=0.330332 lr=7.740275e-05 lr_backbone=7.740275e-05 elapsed=20:17:49 avg_step=4.18s recent_step=4.12s eta=37:45:38 +2026-09-21 02:45:07 | INFO | train rank=0 | step=17490/50000 progress=35.0% loss=0.325277 lr=7.737509e-05 lr_backbone=7.737509e-05 elapsed=20:18:30 avg_step=4.18s recent_step=4.17s eta=37:44:56 +2026-09-21 02:45:49 | INFO | train rank=0 | step=17500/50000 progress=35.0% loss=0.352124 lr=7.734741e-05 lr_backbone=7.734741e-05 elapsed=20:19:12 avg_step=4.18s recent_step=4.15s eta=37:44:14 +2026-09-21 02:46:29 | INFO | train rank=0 | step=17510/50000 progress=35.0% loss=0.412864 lr=7.731972e-05 lr_backbone=7.731972e-05 elapsed=20:19:53 avg_step=4.18s recent_step=4.10s eta=37:43:31 +2026-09-21 02:47:11 | INFO | train rank=0 | step=17520/50000 progress=35.0% loss=0.416202 lr=7.729201e-05 lr_backbone=7.729201e-05 elapsed=20:20:34 avg_step=4.18s recent_step=4.14s eta=37:42:48 +2026-09-21 02:47:52 | INFO | train rank=0 | step=17530/50000 progress=35.1% loss=0.520196 lr=7.726430e-05 lr_backbone=7.726430e-05 elapsed=20:21:16 avg_step=4.18s recent_step=4.15s eta=37:42:06 +2026-09-21 02:48:34 | INFO | train rank=0 | step=17540/50000 progress=35.1% loss=0.319586 lr=7.723657e-05 lr_backbone=7.723657e-05 elapsed=20:21:57 avg_step=4.18s recent_step=4.16s eta=37:41:23 +2026-09-21 02:49:15 | INFO | train rank=0 | step=17550/50000 progress=35.1% loss=0.324030 lr=7.720884e-05 lr_backbone=7.720884e-05 elapsed=20:22:38 avg_step=4.18s recent_step=4.11s eta=37:40:40 +2026-09-21 02:49:57 | INFO | train rank=0 | step=17560/50000 progress=35.1% loss=0.329201 lr=7.718109e-05 lr_backbone=7.718109e-05 elapsed=20:23:20 avg_step=4.18s recent_step=4.15s eta=37:39:58 +2026-09-21 02:50:37 | INFO | train rank=0 | step=17570/50000 progress=35.1% loss=0.321817 lr=7.715332e-05 lr_backbone=7.715332e-05 elapsed=20:24:01 avg_step=4.18s recent_step=4.09s eta=37:39:14 +2026-09-21 02:51:19 | INFO | train rank=0 | step=17580/50000 progress=35.2% loss=0.365268 lr=7.712555e-05 lr_backbone=7.712555e-05 elapsed=20:24:42 avg_step=4.18s recent_step=4.12s eta=37:38:32 +2026-09-21 02:52:00 | INFO | train rank=0 | step=17590/50000 progress=35.2% loss=0.386015 lr=7.709776e-05 lr_backbone=7.709776e-05 elapsed=20:25:24 avg_step=4.18s recent_step=4.15s eta=37:37:49 +2026-09-21 02:52:42 | INFO | train rank=0 | step=17600/50000 progress=35.2% loss=0.275247 lr=7.706997e-05 lr_backbone=7.706997e-05 elapsed=20:26:05 avg_step=4.18s recent_step=4.16s eta=37:37:07 +2026-09-21 02:53:23 | INFO | train rank=0 | step=17610/50000 progress=35.2% loss=0.397048 lr=7.704216e-05 lr_backbone=7.704216e-05 elapsed=20:26:47 avg_step=4.18s recent_step=4.17s eta=37:36:25 +2026-09-21 02:54:05 | INFO | train rank=0 | step=17620/50000 progress=35.2% loss=0.337226 lr=7.701434e-05 lr_backbone=7.701434e-05 elapsed=20:27:28 avg_step=4.18s recent_step=4.13s eta=37:35:42 +2026-09-21 02:54:46 | INFO | train rank=0 | step=17630/50000 progress=35.3% loss=0.418958 lr=7.698650e-05 lr_backbone=7.698650e-05 elapsed=20:28:09 avg_step=4.18s recent_step=4.11s eta=37:34:59 +2026-09-21 02:55:27 | INFO | train rank=0 | step=17640/50000 progress=35.3% loss=0.330937 lr=7.695866e-05 lr_backbone=7.695866e-05 elapsed=20:28:51 avg_step=4.18s recent_step=4.13s eta=37:34:17 +2026-09-21 02:56:09 | INFO | train rank=0 | step=17650/50000 progress=35.3% loss=0.277759 lr=7.693080e-05 lr_backbone=7.693080e-05 elapsed=20:29:32 avg_step=4.18s recent_step=4.16s eta=37:33:35 +2026-09-21 02:56:50 | INFO | train rank=0 | step=17660/50000 progress=35.3% loss=0.344864 lr=7.690293e-05 lr_backbone=7.690293e-05 elapsed=20:30:14 avg_step=4.18s recent_step=4.17s eta=37:32:53 +2026-09-21 02:57:32 | INFO | train rank=0 | step=17670/50000 progress=35.3% loss=0.306113 lr=7.687505e-05 lr_backbone=7.687505e-05 elapsed=20:30:55 avg_step=4.18s recent_step=4.16s eta=37:32:10 +2026-09-21 02:58:13 | INFO | train rank=0 | step=17680/50000 progress=35.4% loss=0.338685 lr=7.684716e-05 lr_backbone=7.684716e-05 elapsed=20:31:37 avg_step=4.18s recent_step=4.13s eta=37:31:28 +2026-09-21 02:58:54 | INFO | train rank=0 | step=17690/50000 progress=35.4% loss=0.325922 lr=7.681926e-05 lr_backbone=7.681926e-05 elapsed=20:32:18 avg_step=4.18s recent_step=4.11s eta=37:30:44 +2026-09-21 02:59:36 | INFO | train rank=0 | step=17700/50000 progress=35.4% loss=0.319206 lr=7.679134e-05 lr_backbone=7.679134e-05 elapsed=20:32:59 avg_step=4.18s recent_step=4.13s eta=37:30:02 +2026-09-21 03:00:17 | INFO | train rank=0 | step=17710/50000 progress=35.4% loss=0.310293 lr=7.676341e-05 lr_backbone=7.676341e-05 elapsed=20:33:40 avg_step=4.18s recent_step=4.11s eta=37:29:19 +2026-09-21 03:00:58 | INFO | train rank=0 | step=17720/50000 progress=35.4% loss=0.334992 lr=7.673547e-05 lr_backbone=7.673547e-05 elapsed=20:34:22 avg_step=4.18s recent_step=4.15s eta=37:28:36 +2026-09-21 03:01:40 | INFO | train rank=0 | step=17730/50000 progress=35.5% loss=0.322817 lr=7.670752e-05 lr_backbone=7.670752e-05 elapsed=20:35:03 avg_step=4.18s recent_step=4.15s eta=37:27:54 +2026-09-21 03:02:21 | INFO | train rank=0 | step=17740/50000 progress=35.5% loss=0.426598 lr=7.667956e-05 lr_backbone=7.667956e-05 elapsed=20:35:44 avg_step=4.18s recent_step=4.12s eta=37:27:11 +2026-09-21 03:03:02 | INFO | train rank=0 | step=17750/50000 progress=35.5% loss=0.365979 lr=7.665159e-05 lr_backbone=7.665159e-05 elapsed=20:36:25 avg_step=4.18s recent_step=4.07s eta=37:26:28 +2026-09-21 03:03:43 | INFO | train rank=0 | step=17760/50000 progress=35.5% loss=0.304213 lr=7.662360e-05 lr_backbone=7.662360e-05 elapsed=20:37:06 avg_step=4.18s recent_step=4.10s eta=37:25:44 +2026-09-21 03:04:24 | INFO | train rank=0 | step=17770/50000 progress=35.5% loss=0.400824 lr=7.659560e-05 lr_backbone=7.659560e-05 elapsed=20:37:47 avg_step=4.18s recent_step=4.13s eta=37:25:02 +2026-09-21 03:05:06 | INFO | train rank=0 | step=17780/50000 progress=35.6% loss=0.332757 lr=7.656759e-05 lr_backbone=7.656759e-05 elapsed=20:38:29 avg_step=4.18s recent_step=4.15s eta=37:24:19 +2026-09-21 03:05:47 | INFO | train rank=0 | step=17790/50000 progress=35.6% loss=0.499002 lr=7.653957e-05 lr_backbone=7.653957e-05 elapsed=20:39:10 avg_step=4.18s recent_step=4.14s eta=37:23:37 +2026-09-21 03:06:28 | INFO | train rank=0 | step=17800/50000 progress=35.6% loss=0.450999 lr=7.651154e-05 lr_backbone=7.651154e-05 elapsed=20:39:52 avg_step=4.18s recent_step=4.15s eta=37:22:54 +2026-09-21 03:07:09 | INFO | train rank=0 | step=17810/50000 progress=35.6% loss=0.445831 lr=7.648350e-05 lr_backbone=7.648350e-05 elapsed=20:40:33 avg_step=4.18s recent_step=4.09s eta=37:22:11 +2026-09-21 03:07:51 | INFO | train rank=0 | step=17820/50000 progress=35.6% loss=0.379805 lr=7.645544e-05 lr_backbone=7.645544e-05 elapsed=20:41:14 avg_step=4.18s recent_step=4.12s eta=37:21:28 +2026-09-21 03:08:32 | INFO | train rank=0 | step=17830/50000 progress=35.7% loss=0.331032 lr=7.642738e-05 lr_backbone=7.642738e-05 elapsed=20:41:55 avg_step=4.18s recent_step=4.13s eta=37:20:45 +2026-09-21 03:09:13 | INFO | train rank=0 | step=17840/50000 progress=35.7% loss=0.407562 lr=7.639930e-05 lr_backbone=7.639930e-05 elapsed=20:42:36 avg_step=4.18s recent_step=4.13s eta=37:20:03 +2026-09-21 03:09:54 | INFO | train rank=0 | step=17850/50000 progress=35.7% loss=0.353232 lr=7.637121e-05 lr_backbone=7.637121e-05 elapsed=20:43:18 avg_step=4.18s recent_step=4.12s eta=37:19:20 +2026-09-21 03:10:35 | INFO | train rank=0 | step=17860/50000 progress=35.7% loss=0.348957 lr=7.634311e-05 lr_backbone=7.634311e-05 elapsed=20:43:59 avg_step=4.18s recent_step=4.12s eta=37:18:37 +2026-09-21 03:11:16 | INFO | train rank=0 | step=17870/50000 progress=35.7% loss=0.305513 lr=7.631499e-05 lr_backbone=7.631499e-05 elapsed=20:44:40 avg_step=4.18s recent_step=4.09s eta=37:17:53 +2026-09-21 03:11:57 | INFO | train rank=0 | step=17880/50000 progress=35.8% loss=0.362137 lr=7.628687e-05 lr_backbone=7.628687e-05 elapsed=20:45:21 avg_step=4.18s recent_step=4.10s eta=37:17:10 +2026-09-21 03:12:39 | INFO | train rank=0 | step=17890/50000 progress=35.8% loss=0.405183 lr=7.625873e-05 lr_backbone=7.625873e-05 elapsed=20:46:02 avg_step=4.18s recent_step=4.14s eta=37:16:28 +2026-09-21 03:13:20 | INFO | train rank=0 | step=17900/50000 progress=35.8% loss=0.333204 lr=7.623058e-05 lr_backbone=7.623058e-05 elapsed=20:46:44 avg_step=4.18s recent_step=4.13s eta=37:15:45 +2026-09-21 03:14:01 | INFO | train rank=0 | step=17910/50000 progress=35.8% loss=0.318189 lr=7.620242e-05 lr_backbone=7.620242e-05 elapsed=20:47:25 avg_step=4.18s recent_step=4.13s eta=37:15:02 +2026-09-21 03:14:43 | INFO | train rank=0 | step=17920/50000 progress=35.8% loss=0.311822 lr=7.617425e-05 lr_backbone=7.617425e-05 elapsed=20:48:06 avg_step=4.18s recent_step=4.14s eta=37:14:20 +2026-09-21 03:15:24 | INFO | train rank=0 | step=17930/50000 progress=35.9% loss=0.361989 lr=7.614607e-05 lr_backbone=7.614607e-05 elapsed=20:48:47 avg_step=4.18s recent_step=4.08s eta=37:13:37 +2026-09-21 03:16:05 | INFO | train rank=0 | step=17940/50000 progress=35.9% loss=0.332877 lr=7.611788e-05 lr_backbone=7.611788e-05 elapsed=20:49:28 avg_step=4.18s recent_step=4.12s eta=37:12:54 +2026-09-21 03:16:46 | INFO | train rank=0 | step=17950/50000 progress=35.9% loss=0.333121 lr=7.608967e-05 lr_backbone=7.608967e-05 elapsed=20:50:09 avg_step=4.18s recent_step=4.13s eta=37:12:11 +2026-09-21 03:17:28 | INFO | train rank=0 | step=17960/50000 progress=35.9% loss=0.361806 lr=7.606146e-05 lr_backbone=7.606146e-05 elapsed=20:50:51 avg_step=4.18s recent_step=4.15s eta=37:11:29 +2026-09-21 03:18:09 | INFO | train rank=0 | step=17970/50000 progress=35.9% loss=0.360222 lr=7.603323e-05 lr_backbone=7.603323e-05 elapsed=20:51:32 avg_step=4.18s recent_step=4.13s eta=37:10:46 +2026-09-21 03:18:50 | INFO | train rank=0 | step=17980/50000 progress=36.0% loss=0.423709 lr=7.600499e-05 lr_backbone=7.600499e-05 elapsed=20:52:14 avg_step=4.18s recent_step=4.14s eta=37:10:04 +2026-09-21 03:19:31 | INFO | train rank=0 | step=17990/50000 progress=36.0% loss=0.362393 lr=7.597674e-05 lr_backbone=7.597674e-05 elapsed=20:52:55 avg_step=4.18s recent_step=4.10s eta=37:09:20 +2026-09-21 03:20:12 | INFO | train rank=0 | step=18000/50000 progress=36.0% loss=0.333176 lr=7.594848e-05 lr_backbone=7.594848e-05 elapsed=20:53:36 avg_step=4.18s recent_step=4.10s eta=37:08:37 +2026-09-21 03:20:54 | INFO | train rank=0 | step=18010/50000 progress=36.0% loss=0.359185 lr=7.592021e-05 lr_backbone=7.592021e-05 elapsed=20:54:17 avg_step=4.18s recent_step=4.16s eta=37:07:55 +2026-09-21 03:21:35 | INFO | train rank=0 | step=18020/50000 progress=36.0% loss=0.352163 lr=7.589192e-05 lr_backbone=7.589192e-05 elapsed=20:54:59 avg_step=4.18s recent_step=4.13s eta=37:07:12 +2026-09-21 03:22:17 | INFO | train rank=0 | step=18030/50000 progress=36.1% loss=0.324617 lr=7.586363e-05 lr_backbone=7.586363e-05 elapsed=20:55:40 avg_step=4.18s recent_step=4.15s eta=37:06:30 +2026-09-21 03:22:58 | INFO | train rank=0 | step=18040/50000 progress=36.1% loss=0.320359 lr=7.583532e-05 lr_backbone=7.583532e-05 elapsed=20:56:21 avg_step=4.18s recent_step=4.12s eta=37:05:47 +2026-09-21 03:23:39 | INFO | train rank=0 | step=18050/50000 progress=36.1% loss=0.339374 lr=7.580700e-05 lr_backbone=7.580700e-05 elapsed=20:57:02 avg_step=4.18s recent_step=4.10s eta=37:05:04 +2026-09-21 03:24:20 | INFO | train rank=0 | step=18060/50000 progress=36.1% loss=0.360145 lr=7.577867e-05 lr_backbone=7.577867e-05 elapsed=20:57:43 avg_step=4.18s recent_step=4.10s eta=37:04:21 +2026-09-21 03:25:02 | INFO | train rank=0 | step=18070/50000 progress=36.1% loss=0.383546 lr=7.575033e-05 lr_backbone=7.575033e-05 elapsed=20:58:25 avg_step=4.18s recent_step=4.15s eta=37:03:39 +2026-09-21 03:25:43 | INFO | train rank=0 | step=18080/50000 progress=36.2% loss=0.331145 lr=7.572198e-05 lr_backbone=7.572198e-05 elapsed=20:59:06 avg_step=4.18s recent_step=4.15s eta=37:02:56 +2026-09-21 03:26:25 | INFO | train rank=0 | step=18090/50000 progress=36.2% loss=0.453858 lr=7.569361e-05 lr_backbone=7.569361e-05 elapsed=20:59:48 avg_step=4.18s recent_step=4.16s eta=37:02:14 +2026-09-21 03:27:06 | INFO | train rank=0 | step=18100/50000 progress=36.2% loss=0.391076 lr=7.566524e-05 lr_backbone=7.566524e-05 elapsed=21:00:30 avg_step=4.18s recent_step=4.16s eta=37:01:32 +2026-09-21 03:27:47 | INFO | train rank=0 | step=18110/50000 progress=36.2% loss=0.390408 lr=7.563685e-05 lr_backbone=7.563685e-05 elapsed=21:01:11 avg_step=4.18s recent_step=4.10s eta=37:00:49 +2026-09-21 03:28:28 | INFO | train rank=0 | step=18120/50000 progress=36.2% loss=0.399049 lr=7.560846e-05 lr_backbone=7.560846e-05 elapsed=21:01:52 avg_step=4.18s recent_step=4.11s eta=37:00:06 +2026-09-21 03:29:10 | INFO | train rank=0 | step=18130/50000 progress=36.3% loss=0.365139 lr=7.558005e-05 lr_backbone=7.558005e-05 elapsed=21:02:33 avg_step=4.18s recent_step=4.15s eta=36:59:24 +2026-09-21 03:29:51 | INFO | train rank=0 | step=18140/50000 progress=36.3% loss=0.322615 lr=7.555163e-05 lr_backbone=7.555163e-05 elapsed=21:03:15 avg_step=4.18s recent_step=4.15s eta=36:58:42 +2026-09-21 03:30:33 | INFO | train rank=0 | step=18150/50000 progress=36.3% loss=0.413229 lr=7.552320e-05 lr_backbone=7.552320e-05 elapsed=21:03:56 avg_step=4.18s recent_step=4.17s eta=36:58:00 +2026-09-21 03:31:15 | INFO | train rank=0 | step=18160/50000 progress=36.3% loss=0.422634 lr=7.549475e-05 lr_backbone=7.549475e-05 elapsed=21:04:38 avg_step=4.18s recent_step=4.17s eta=36:57:18 +2026-09-21 03:31:56 | INFO | train rank=0 | step=18170/50000 progress=36.3% loss=0.327029 lr=7.546630e-05 lr_backbone=7.546630e-05 elapsed=21:05:19 avg_step=4.18s recent_step=4.10s eta=36:56:34 +2026-09-21 03:32:37 | INFO | train rank=0 | step=18180/50000 progress=36.4% loss=0.341231 lr=7.543784e-05 lr_backbone=7.543784e-05 elapsed=21:06:00 avg_step=4.18s recent_step=4.14s eta=36:55:52 +2026-09-21 03:33:18 | INFO | train rank=0 | step=18190/50000 progress=36.4% loss=0.302386 lr=7.540936e-05 lr_backbone=7.540936e-05 elapsed=21:06:42 avg_step=4.18s recent_step=4.13s eta=36:55:09 +2026-09-21 03:34:00 | INFO | train rank=0 | step=18200/50000 progress=36.4% loss=0.278852 lr=7.538088e-05 lr_backbone=7.538088e-05 elapsed=21:07:23 avg_step=4.18s recent_step=4.15s eta=36:54:27 +2026-09-21 03:34:41 | INFO | train rank=0 | step=18210/50000 progress=36.4% loss=0.390412 lr=7.535238e-05 lr_backbone=7.535238e-05 elapsed=21:08:04 avg_step=4.18s recent_step=4.12s eta=36:53:44 +2026-09-21 03:35:23 | INFO | train rank=0 | step=18220/50000 progress=36.4% loss=0.355162 lr=7.532387e-05 lr_backbone=7.532387e-05 elapsed=21:08:46 avg_step=4.18s recent_step=4.16s eta=36:53:02 +2026-09-21 03:36:04 | INFO | train rank=0 | step=18230/50000 progress=36.5% loss=0.335803 lr=7.529535e-05 lr_backbone=7.529535e-05 elapsed=21:09:27 avg_step=4.18s recent_step=4.09s eta=36:52:19 +2026-09-21 03:36:45 | INFO | train rank=0 | step=18240/50000 progress=36.5% loss=0.365474 lr=7.526682e-05 lr_backbone=7.526682e-05 elapsed=21:10:08 avg_step=4.18s recent_step=4.12s eta=36:51:36 +2026-09-21 03:37:26 | INFO | train rank=0 | step=18250/50000 progress=36.5% loss=0.318909 lr=7.523828e-05 lr_backbone=7.523828e-05 elapsed=21:10:49 avg_step=4.18s recent_step=4.14s eta=36:50:53 +2026-09-21 03:38:08 | INFO | train rank=0 | step=18260/50000 progress=36.5% loss=0.370929 lr=7.520972e-05 lr_backbone=7.520972e-05 elapsed=21:11:31 avg_step=4.18s recent_step=4.15s eta=36:50:11 +2026-09-21 03:38:49 | INFO | train rank=0 | step=18270/50000 progress=36.5% loss=0.298847 lr=7.518116e-05 lr_backbone=7.518116e-05 elapsed=21:12:12 avg_step=4.18s recent_step=4.14s eta=36:49:29 +2026-09-21 03:39:31 | INFO | train rank=0 | step=18280/50000 progress=36.6% loss=0.455808 lr=7.515259e-05 lr_backbone=7.515259e-05 elapsed=21:12:54 avg_step=4.18s recent_step=4.17s eta=36:48:47 +2026-09-21 03:40:12 | INFO | train rank=0 | step=18290/50000 progress=36.6% loss=0.443524 lr=7.512400e-05 lr_backbone=7.512400e-05 elapsed=21:13:35 avg_step=4.18s recent_step=4.10s eta=36:48:04 +2026-09-21 03:40:53 | INFO | train rank=0 | step=18300/50000 progress=36.6% loss=0.426702 lr=7.509540e-05 lr_backbone=7.509540e-05 elapsed=21:14:16 avg_step=4.18s recent_step=4.12s eta=36:47:21 +2026-09-21 03:41:34 | INFO | train rank=0 | step=18310/50000 progress=36.6% loss=0.349787 lr=7.506679e-05 lr_backbone=7.506679e-05 elapsed=21:14:58 avg_step=4.18s recent_step=4.13s eta=36:46:38 +2026-09-21 03:42:16 | INFO | train rank=0 | step=18320/50000 progress=36.6% loss=0.395210 lr=7.503818e-05 lr_backbone=7.503818e-05 elapsed=21:15:39 avg_step=4.18s recent_step=4.17s eta=36:45:56 +2026-09-21 03:42:58 | INFO | train rank=0 | step=18330/50000 progress=36.7% loss=0.366611 lr=7.500955e-05 lr_backbone=7.500955e-05 elapsed=21:16:21 avg_step=4.18s recent_step=4.18s eta=36:45:15 +2026-09-21 03:43:39 | INFO | train rank=0 | step=18340/50000 progress=36.7% loss=0.377083 lr=7.498090e-05 lr_backbone=7.498090e-05 elapsed=21:17:03 avg_step=4.18s recent_step=4.16s eta=36:44:33 +2026-09-21 03:44:20 | INFO | train rank=0 | step=18350/50000 progress=36.7% loss=0.346565 lr=7.495225e-05 lr_backbone=7.495225e-05 elapsed=21:17:44 avg_step=4.18s recent_step=4.10s eta=36:43:49 +2026-09-21 03:45:02 | INFO | train rank=0 | step=18360/50000 progress=36.7% loss=0.391283 lr=7.492359e-05 lr_backbone=7.492359e-05 elapsed=21:18:25 avg_step=4.18s recent_step=4.12s eta=36:43:07 +2026-09-21 03:45:43 | INFO | train rank=0 | step=18370/50000 progress=36.7% loss=0.373817 lr=7.489492e-05 lr_backbone=7.489492e-05 elapsed=21:19:06 avg_step=4.18s recent_step=4.14s eta=36:42:24 +2026-09-21 03:46:24 | INFO | train rank=0 | step=18380/50000 progress=36.8% loss=0.357453 lr=7.486623e-05 lr_backbone=7.486623e-05 elapsed=21:19:48 avg_step=4.18s recent_step=4.15s eta=36:41:42 +2026-09-21 03:47:06 | INFO | train rank=0 | step=18390/50000 progress=36.8% loss=0.330274 lr=7.483754e-05 lr_backbone=7.483754e-05 elapsed=21:20:29 avg_step=4.18s recent_step=4.15s eta=36:41:00 +2026-09-21 03:47:47 | INFO | train rank=0 | step=18400/50000 progress=36.8% loss=0.407577 lr=7.480883e-05 lr_backbone=7.480883e-05 elapsed=21:21:11 avg_step=4.18s recent_step=4.13s eta=36:40:17 +2026-09-21 03:48:28 | INFO | train rank=0 | step=18410/50000 progress=36.8% loss=0.368117 lr=7.478011e-05 lr_backbone=7.478011e-05 elapsed=21:21:52 avg_step=4.18s recent_step=4.11s eta=36:39:34 +2026-09-21 03:49:10 | INFO | train rank=0 | step=18420/50000 progress=36.8% loss=0.355707 lr=7.475139e-05 lr_backbone=7.475139e-05 elapsed=21:22:33 avg_step=4.18s recent_step=4.14s eta=36:38:52 +2026-09-21 03:49:51 | INFO | train rank=0 | step=18430/50000 progress=36.9% loss=0.303531 lr=7.472265e-05 lr_backbone=7.472265e-05 elapsed=21:23:15 avg_step=4.18s recent_step=4.15s eta=36:38:10 +2026-09-21 03:50:33 | INFO | train rank=0 | step=18440/50000 progress=36.9% loss=0.321110 lr=7.469390e-05 lr_backbone=7.469390e-05 elapsed=21:23:56 avg_step=4.18s recent_step=4.15s eta=36:37:27 +2026-09-21 03:51:14 | INFO | train rank=0 | step=18450/50000 progress=36.9% loss=0.379551 lr=7.466514e-05 lr_backbone=7.466514e-05 elapsed=21:24:38 avg_step=4.18s recent_step=4.15s eta=36:36:45 +2026-09-21 03:51:56 | INFO | train rank=0 | step=18460/50000 progress=36.9% loss=0.393007 lr=7.463637e-05 lr_backbone=7.463637e-05 elapsed=21:25:19 avg_step=4.18s recent_step=4.15s eta=36:36:03 +2026-09-21 03:52:37 | INFO | train rank=0 | step=18470/50000 progress=36.9% loss=0.383285 lr=7.460759e-05 lr_backbone=7.460759e-05 elapsed=21:26:00 avg_step=4.18s recent_step=4.10s eta=36:35:20 +2026-09-21 03:53:18 | INFO | train rank=0 | step=18480/50000 progress=37.0% loss=0.303199 lr=7.457879e-05 lr_backbone=7.457879e-05 elapsed=21:26:42 avg_step=4.18s recent_step=4.16s eta=36:34:38 +2026-09-21 03:54:00 | INFO | train rank=0 | step=18490/50000 progress=37.0% loss=0.332689 lr=7.454999e-05 lr_backbone=7.454999e-05 elapsed=21:27:23 avg_step=4.18s recent_step=4.13s eta=36:33:55 +2026-09-21 03:54:41 | INFO | train rank=0 | step=18500/50000 progress=37.0% loss=0.365050 lr=7.452118e-05 lr_backbone=7.452118e-05 elapsed=21:28:05 avg_step=4.18s recent_step=4.16s eta=36:33:13 +2026-09-21 03:55:23 | INFO | train rank=0 | step=18510/50000 progress=37.0% loss=0.327861 lr=7.449235e-05 lr_backbone=7.449235e-05 elapsed=21:28:46 avg_step=4.18s recent_step=4.14s eta=36:32:31 +2026-09-21 03:56:04 | INFO | train rank=0 | step=18520/50000 progress=37.0% loss=0.321725 lr=7.446351e-05 lr_backbone=7.446351e-05 elapsed=21:29:28 avg_step=4.18s recent_step=4.15s eta=36:31:49 +2026-09-21 03:56:45 | INFO | train rank=0 | step=18530/50000 progress=37.1% loss=0.297435 lr=7.443467e-05 lr_backbone=7.443467e-05 elapsed=21:30:09 avg_step=4.18s recent_step=4.11s eta=36:31:06 +2026-09-21 03:57:27 | INFO | train rank=0 | step=18540/50000 progress=37.1% loss=0.324815 lr=7.440581e-05 lr_backbone=7.440581e-05 elapsed=21:30:50 avg_step=4.18s recent_step=4.12s eta=36:30:23 +2026-09-21 03:58:08 | INFO | train rank=0 | step=18550/50000 progress=37.1% loss=0.372535 lr=7.437694e-05 lr_backbone=7.437694e-05 elapsed=21:31:31 avg_step=4.18s recent_step=4.14s eta=36:29:41 +2026-09-21 03:58:49 | INFO | train rank=0 | step=18560/50000 progress=37.1% loss=0.329091 lr=7.434807e-05 lr_backbone=7.434807e-05 elapsed=21:32:13 avg_step=4.18s recent_step=4.13s eta=36:28:58 +2026-09-21 03:59:30 | INFO | train rank=0 | step=18570/50000 progress=37.1% loss=0.356804 lr=7.431918e-05 lr_backbone=7.431918e-05 elapsed=21:32:54 avg_step=4.18s recent_step=4.11s eta=36:28:15 +2026-09-21 04:00:12 | INFO | train rank=0 | step=18580/50000 progress=37.2% loss=0.336020 lr=7.429028e-05 lr_backbone=7.429028e-05 elapsed=21:33:35 avg_step=4.18s recent_step=4.15s eta=36:27:33 +2026-09-21 04:00:53 | INFO | train rank=0 | step=18590/50000 progress=37.2% loss=0.317311 lr=7.426137e-05 lr_backbone=7.426137e-05 elapsed=21:34:16 avg_step=4.18s recent_step=4.11s eta=36:26:50 +2026-09-21 04:01:34 | INFO | train rank=0 | step=18600/50000 progress=37.2% loss=0.324354 lr=7.423245e-05 lr_backbone=7.423245e-05 elapsed=21:34:58 avg_step=4.18s recent_step=4.14s eta=36:26:08 +2026-09-21 04:02:16 | INFO | train rank=0 | step=18610/50000 progress=37.2% loss=0.297280 lr=7.420352e-05 lr_backbone=7.420352e-05 elapsed=21:35:39 avg_step=4.18s recent_step=4.17s eta=36:25:26 +2026-09-21 04:02:57 | INFO | train rank=0 | step=18620/50000 progress=37.2% loss=0.443789 lr=7.417457e-05 lr_backbone=7.417457e-05 elapsed=21:36:21 avg_step=4.18s recent_step=4.14s eta=36:24:43 +2026-09-21 04:03:39 | INFO | train rank=0 | step=18630/50000 progress=37.3% loss=0.334990 lr=7.414562e-05 lr_backbone=7.414562e-05 elapsed=21:37:02 avg_step=4.18s recent_step=4.15s eta=36:24:01 +2026-09-21 04:04:21 | INFO | train rank=0 | step=18640/50000 progress=37.3% loss=0.372093 lr=7.411666e-05 lr_backbone=7.411666e-05 elapsed=21:37:44 avg_step=4.18s recent_step=4.16s eta=36:23:19 +2026-09-21 04:05:02 | INFO | train rank=0 | step=18650/50000 progress=37.3% loss=0.309089 lr=7.408768e-05 lr_backbone=7.408768e-05 elapsed=21:38:25 avg_step=4.18s recent_step=4.11s eta=36:22:36 +2026-09-21 04:05:43 | INFO | train rank=0 | step=18660/50000 progress=37.3% loss=0.347190 lr=7.405870e-05 lr_backbone=7.405870e-05 elapsed=21:39:06 avg_step=4.18s recent_step=4.13s eta=36:21:53 +2026-09-21 04:06:25 | INFO | train rank=0 | step=18670/50000 progress=37.3% loss=0.411809 lr=7.402970e-05 lr_backbone=7.402970e-05 elapsed=21:39:48 avg_step=4.18s recent_step=4.16s eta=36:21:11 +2026-09-21 04:07:06 | INFO | train rank=0 | step=18680/50000 progress=37.4% loss=0.386727 lr=7.400070e-05 lr_backbone=7.400070e-05 elapsed=21:40:29 avg_step=4.18s recent_step=4.13s eta=36:20:29 +2026-09-21 04:07:47 | INFO | train rank=0 | step=18690/50000 progress=37.4% loss=0.328818 lr=7.397168e-05 lr_backbone=7.397168e-05 elapsed=21:41:11 avg_step=4.18s recent_step=4.15s eta=36:19:47 +2026-09-21 04:08:29 | INFO | train rank=0 | step=18700/50000 progress=37.4% loss=0.355348 lr=7.394266e-05 lr_backbone=7.394266e-05 elapsed=21:41:52 avg_step=4.18s recent_step=4.17s eta=36:19:05 +2026-09-21 04:09:10 | INFO | train rank=0 | step=18710/50000 progress=37.4% loss=0.411774 lr=7.391362e-05 lr_backbone=7.391362e-05 elapsed=21:42:33 avg_step=4.18s recent_step=4.10s eta=36:18:22 +2026-09-21 04:09:51 | INFO | train rank=0 | step=18720/50000 progress=37.4% loss=0.348945 lr=7.388457e-05 lr_backbone=7.388457e-05 elapsed=21:43:15 avg_step=4.18s recent_step=4.12s eta=36:17:39 +2026-09-21 04:10:33 | INFO | train rank=0 | step=18730/50000 progress=37.5% loss=0.297197 lr=7.385552e-05 lr_backbone=7.385552e-05 elapsed=21:43:56 avg_step=4.18s recent_step=4.15s eta=36:16:57 +2026-09-21 04:11:14 | INFO | train rank=0 | step=18740/50000 progress=37.5% loss=0.341051 lr=7.382645e-05 lr_backbone=7.382645e-05 elapsed=21:44:38 avg_step=4.18s recent_step=4.15s eta=36:16:14 +2026-09-21 04:11:56 | INFO | train rank=0 | step=18750/50000 progress=37.5% loss=0.381001 lr=7.379737e-05 lr_backbone=7.379737e-05 elapsed=21:45:19 avg_step=4.18s recent_step=4.14s eta=36:15:32 +2026-09-21 04:12:37 | INFO | train rank=0 | step=18760/50000 progress=37.5% loss=0.308883 lr=7.376828e-05 lr_backbone=7.376828e-05 elapsed=21:46:01 avg_step=4.18s recent_step=4.16s eta=36:14:50 +2026-09-21 04:13:18 | INFO | train rank=0 | step=18770/50000 progress=37.5% loss=0.438902 lr=7.373918e-05 lr_backbone=7.373918e-05 elapsed=21:46:41 avg_step=4.18s recent_step=4.09s eta=36:14:07 +2026-09-21 04:14:00 | INFO | train rank=0 | step=18780/50000 progress=37.6% loss=0.354045 lr=7.371007e-05 lr_backbone=7.371007e-05 elapsed=21:47:23 avg_step=4.18s recent_step=4.14s eta=36:13:24 +2026-09-21 04:14:41 | INFO | train rank=0 | step=18790/50000 progress=37.6% loss=0.402737 lr=7.368095e-05 lr_backbone=7.368095e-05 elapsed=21:48:04 avg_step=4.18s recent_step=4.16s eta=36:12:42 +2026-09-21 04:15:23 | INFO | train rank=0 | step=18800/50000 progress=37.6% loss=0.393391 lr=7.365182e-05 lr_backbone=7.365182e-05 elapsed=21:48:46 avg_step=4.18s recent_step=4.16s eta=36:12:00 +2026-09-21 04:16:04 | INFO | train rank=0 | step=18810/50000 progress=37.6% loss=0.372381 lr=7.362268e-05 lr_backbone=7.362268e-05 elapsed=21:49:27 avg_step=4.18s recent_step=4.14s eta=36:11:18 +2026-09-21 04:16:46 | INFO | train rank=0 | step=18820/50000 progress=37.6% loss=0.313162 lr=7.359353e-05 lr_backbone=7.359353e-05 elapsed=21:50:09 avg_step=4.18s recent_step=4.14s eta=36:10:35 +2026-09-21 04:17:27 | INFO | train rank=0 | step=18830/50000 progress=37.7% loss=0.399477 lr=7.356437e-05 lr_backbone=7.356437e-05 elapsed=21:50:50 avg_step=4.18s recent_step=4.10s eta=36:09:52 +2026-09-21 04:18:08 | INFO | train rank=0 | step=18840/50000 progress=37.7% loss=0.328805 lr=7.353520e-05 lr_backbone=7.353520e-05 elapsed=21:51:31 avg_step=4.18s recent_step=4.14s eta=36:09:10 +2026-09-21 04:18:50 | INFO | train rank=0 | step=18850/50000 progress=37.7% loss=0.361488 lr=7.350601e-05 lr_backbone=7.350601e-05 elapsed=21:52:13 avg_step=4.18s recent_step=4.17s eta=36:08:28 +2026-09-21 04:19:31 | INFO | train rank=0 | step=18860/50000 progress=37.7% loss=0.355169 lr=7.347682e-05 lr_backbone=7.347682e-05 elapsed=21:52:54 avg_step=4.18s recent_step=4.15s eta=36:07:46 +2026-09-21 04:20:13 | INFO | train rank=0 | step=18870/50000 progress=37.7% loss=0.306433 lr=7.344762e-05 lr_backbone=7.344762e-05 elapsed=21:53:36 avg_step=4.18s recent_step=4.17s eta=36:07:04 +2026-09-21 04:20:54 | INFO | train rank=0 | step=18880/50000 progress=37.8% loss=0.395549 lr=7.341841e-05 lr_backbone=7.341841e-05 elapsed=21:54:17 avg_step=4.18s recent_step=4.13s eta=36:06:21 +2026-09-21 04:21:35 | INFO | train rank=0 | step=18890/50000 progress=37.8% loss=0.335117 lr=7.338918e-05 lr_backbone=7.338918e-05 elapsed=21:54:58 avg_step=4.18s recent_step=4.10s eta=36:05:38 +2026-09-21 04:22:16 | INFO | train rank=0 | step=18900/50000 progress=37.8% loss=0.346787 lr=7.335995e-05 lr_backbone=7.335995e-05 elapsed=21:55:40 avg_step=4.18s recent_step=4.14s eta=36:04:56 +2026-09-21 04:22:58 | INFO | train rank=0 | step=18910/50000 progress=37.8% loss=0.394115 lr=7.333071e-05 lr_backbone=7.333071e-05 elapsed=21:56:21 avg_step=4.18s recent_step=4.16s eta=36:04:14 +2026-09-21 04:23:40 | INFO | train rank=0 | step=18920/50000 progress=37.8% loss=0.402565 lr=7.330145e-05 lr_backbone=7.330145e-05 elapsed=21:57:03 avg_step=4.18s recent_step=4.15s eta=36:03:32 +2026-09-21 04:24:21 | INFO | train rank=0 | step=18930/50000 progress=37.9% loss=0.362031 lr=7.327219e-05 lr_backbone=7.327219e-05 elapsed=21:57:44 avg_step=4.18s recent_step=4.16s eta=36:02:50 +2026-09-21 04:25:03 | INFO | train rank=0 | step=18940/50000 progress=37.9% loss=0.346355 lr=7.324292e-05 lr_backbone=7.324292e-05 elapsed=21:58:26 avg_step=4.18s recent_step=4.14s eta=36:02:07 +2026-09-21 04:25:43 | INFO | train rank=0 | step=18950/50000 progress=37.9% loss=0.313765 lr=7.321363e-05 lr_backbone=7.321363e-05 elapsed=21:59:07 avg_step=4.18s recent_step=4.09s eta=36:01:24 +2026-09-21 04:26:25 | INFO | train rank=0 | step=18960/50000 progress=37.9% loss=0.490066 lr=7.318434e-05 lr_backbone=7.318434e-05 elapsed=21:59:48 avg_step=4.18s recent_step=4.14s eta=36:00:42 +2026-09-21 04:27:06 | INFO | train rank=0 | step=18970/50000 progress=37.9% loss=0.382426 lr=7.315503e-05 lr_backbone=7.315503e-05 elapsed=22:00:30 avg_step=4.18s recent_step=4.16s eta=36:00:00 +2026-09-21 04:27:48 | INFO | train rank=0 | step=18980/50000 progress=38.0% loss=0.311406 lr=7.312572e-05 lr_backbone=7.312572e-05 elapsed=22:01:12 avg_step=4.18s recent_step=4.18s eta=35:59:18 +2026-09-21 04:28:30 | INFO | train rank=0 | step=18990/50000 progress=38.0% loss=0.299910 lr=7.309639e-05 lr_backbone=7.309639e-05 elapsed=22:01:53 avg_step=4.18s recent_step=4.16s eta=35:58:36 +2026-09-21 04:29:11 | INFO | train rank=0 | step=19000/50000 progress=38.0% loss=0.376473 lr=7.306706e-05 lr_backbone=7.306706e-05 elapsed=22:02:34 avg_step=4.18s recent_step=4.14s eta=35:57:53 +2026-09-21 04:29:52 | INFO | train rank=0 | step=19010/50000 progress=38.0% loss=0.390313 lr=7.303771e-05 lr_backbone=7.303771e-05 elapsed=22:03:15 avg_step=4.18s recent_step=4.10s eta=35:57:10 +2026-09-21 04:30:33 | INFO | train rank=0 | step=19020/50000 progress=38.0% loss=0.391446 lr=7.300836e-05 lr_backbone=7.300836e-05 elapsed=22:03:57 avg_step=4.18s recent_step=4.11s eta=35:56:27 +2026-09-21 04:31:15 | INFO | train rank=0 | step=19030/50000 progress=38.1% loss=0.310826 lr=7.297899e-05 lr_backbone=7.297899e-05 elapsed=22:04:38 avg_step=4.18s recent_step=4.16s eta=35:55:46 +2026-09-21 04:31:56 | INFO | train rank=0 | step=19040/50000 progress=38.1% loss=0.402675 lr=7.294962e-05 lr_backbone=7.294962e-05 elapsed=22:05:20 avg_step=4.18s recent_step=4.15s eta=35:55:03 +2026-09-21 04:32:38 | INFO | train rank=0 | step=19050/50000 progress=38.1% loss=0.472435 lr=7.292023e-05 lr_backbone=7.292023e-05 elapsed=22:06:01 avg_step=4.18s recent_step=4.16s eta=35:54:21 +2026-09-21 04:33:19 | INFO | train rank=0 | step=19060/50000 progress=38.1% loss=0.412015 lr=7.289084e-05 lr_backbone=7.289084e-05 elapsed=22:06:43 avg_step=4.18s recent_step=4.12s eta=35:53:39 +2026-09-21 04:34:00 | INFO | train rank=0 | step=19070/50000 progress=38.1% loss=0.391111 lr=7.286143e-05 lr_backbone=7.286143e-05 elapsed=22:07:24 avg_step=4.18s recent_step=4.11s eta=35:52:56 +2026-09-21 04:34:42 | INFO | train rank=0 | step=19080/50000 progress=38.2% loss=0.432881 lr=7.283202e-05 lr_backbone=7.283202e-05 elapsed=22:08:05 avg_step=4.18s recent_step=4.12s eta=35:52:13 +2026-09-21 04:35:23 | INFO | train rank=0 | step=19090/50000 progress=38.2% loss=0.328166 lr=7.280259e-05 lr_backbone=7.280259e-05 elapsed=22:08:46 avg_step=4.18s recent_step=4.14s eta=35:51:31 +2026-09-21 04:36:05 | INFO | train rank=0 | step=19100/50000 progress=38.2% loss=0.377588 lr=7.277316e-05 lr_backbone=7.277316e-05 elapsed=22:09:28 avg_step=4.18s recent_step=4.16s eta=35:50:49 +2026-09-21 04:36:46 | INFO | train rank=0 | step=19110/50000 progress=38.2% loss=0.396356 lr=7.274371e-05 lr_backbone=7.274371e-05 elapsed=22:10:09 avg_step=4.18s recent_step=4.15s eta=35:50:07 +2026-09-21 04:37:27 | INFO | train rank=0 | step=19120/50000 progress=38.2% loss=0.312369 lr=7.271426e-05 lr_backbone=7.271426e-05 elapsed=22:10:51 avg_step=4.18s recent_step=4.13s eta=35:49:24 +2026-09-21 04:38:08 | INFO | train rank=0 | step=19130/50000 progress=38.3% loss=0.337435 lr=7.268479e-05 lr_backbone=7.268479e-05 elapsed=22:11:31 avg_step=4.18s recent_step=4.07s eta=35:48:40 +2026-09-21 04:38:49 | INFO | train rank=0 | step=19140/50000 progress=38.3% loss=0.339513 lr=7.265532e-05 lr_backbone=7.265532e-05 elapsed=22:12:12 avg_step=4.18s recent_step=4.12s eta=35:47:58 +2026-09-21 04:39:31 | INFO | train rank=0 | step=19150/50000 progress=38.3% loss=0.311748 lr=7.262583e-05 lr_backbone=7.262583e-05 elapsed=22:12:54 avg_step=4.18s recent_step=4.14s eta=35:47:16 +2026-09-21 04:40:12 | INFO | train rank=0 | step=19160/50000 progress=38.3% loss=0.423337 lr=7.259634e-05 lr_backbone=7.259634e-05 elapsed=22:13:35 avg_step=4.18s recent_step=4.14s eta=35:46:33 +2026-09-21 04:40:53 | INFO | train rank=0 | step=19170/50000 progress=38.3% loss=0.312117 lr=7.256683e-05 lr_backbone=7.256683e-05 elapsed=22:14:17 avg_step=4.18s recent_step=4.13s eta=35:45:51 +2026-09-21 04:41:35 | INFO | train rank=0 | step=19180/50000 progress=38.4% loss=0.327068 lr=7.253732e-05 lr_backbone=7.253732e-05 elapsed=22:14:58 avg_step=4.18s recent_step=4.15s eta=35:45:08 +2026-09-21 04:42:15 | INFO | train rank=0 | step=19190/50000 progress=38.4% loss=0.327739 lr=7.250779e-05 lr_backbone=7.250779e-05 elapsed=22:15:39 avg_step=4.18s recent_step=4.07s eta=35:44:25 +2026-09-21 04:42:56 | INFO | train rank=0 | step=19200/50000 progress=38.4% loss=0.356058 lr=7.247826e-05 lr_backbone=7.247826e-05 elapsed=22:16:20 avg_step=4.18s recent_step=4.10s eta=35:43:42 +2026-09-21 04:43:38 | INFO | train rank=0 | step=19210/50000 progress=38.4% loss=0.376211 lr=7.244871e-05 lr_backbone=7.244871e-05 elapsed=22:17:01 avg_step=4.18s recent_step=4.13s eta=35:42:59 +2026-09-21 04:44:19 | INFO | train rank=0 | step=19220/50000 progress=38.4% loss=0.309410 lr=7.241916e-05 lr_backbone=7.241916e-05 elapsed=22:17:42 avg_step=4.18s recent_step=4.12s eta=35:42:17 +2026-09-21 04:45:00 | INFO | train rank=0 | step=19230/50000 progress=38.5% loss=0.353544 lr=7.238960e-05 lr_backbone=7.238960e-05 elapsed=22:18:24 avg_step=4.18s recent_step=4.14s eta=35:41:34 +2026-09-21 04:45:42 | INFO | train rank=0 | step=19240/50000 progress=38.5% loss=0.380043 lr=7.236002e-05 lr_backbone=7.236002e-05 elapsed=22:19:05 avg_step=4.18s recent_step=4.15s eta=35:40:52 +2026-09-21 04:46:23 | INFO | train rank=0 | step=19250/50000 progress=38.5% loss=0.341500 lr=7.233044e-05 lr_backbone=7.233044e-05 elapsed=22:19:46 avg_step=4.18s recent_step=4.09s eta=35:40:09 +2026-09-21 04:47:04 | INFO | train rank=0 | step=19260/50000 progress=38.5% loss=0.345865 lr=7.230085e-05 lr_backbone=7.230085e-05 elapsed=22:20:27 avg_step=4.18s recent_step=4.12s eta=35:39:26 +2026-09-21 04:47:45 | INFO | train rank=0 | step=19270/50000 progress=38.5% loss=0.323676 lr=7.227124e-05 lr_backbone=7.227124e-05 elapsed=22:21:09 avg_step=4.18s recent_step=4.14s eta=35:38:44 +2026-09-21 04:48:27 | INFO | train rank=0 | step=19280/50000 progress=38.6% loss=0.342421 lr=7.224163e-05 lr_backbone=7.224163e-05 elapsed=22:21:50 avg_step=4.18s recent_step=4.13s eta=35:38:02 +2026-09-21 04:49:08 | INFO | train rank=0 | step=19290/50000 progress=38.6% loss=0.321136 lr=7.221201e-05 lr_backbone=7.221201e-05 elapsed=22:22:31 avg_step=4.18s recent_step=4.12s eta=35:37:19 +2026-09-21 04:49:49 | INFO | train rank=0 | step=19300/50000 progress=38.6% loss=0.357896 lr=7.218238e-05 lr_backbone=7.218238e-05 elapsed=22:23:13 avg_step=4.18s recent_step=4.14s eta=35:36:37 +2026-09-21 04:50:30 | INFO | train rank=0 | step=19310/50000 progress=38.6% loss=0.349707 lr=7.215274e-05 lr_backbone=7.215274e-05 elapsed=22:23:53 avg_step=4.18s recent_step=4.08s eta=35:35:53 +2026-09-21 04:51:11 | INFO | train rank=0 | step=19320/50000 progress=38.6% loss=0.409025 lr=7.212308e-05 lr_backbone=7.212308e-05 elapsed=22:24:34 avg_step=4.18s recent_step=4.10s eta=35:35:10 +2026-09-21 04:51:53 | INFO | train rank=0 | step=19330/50000 progress=38.7% loss=0.300956 lr=7.209342e-05 lr_backbone=7.209342e-05 elapsed=22:25:16 avg_step=4.18s recent_step=4.16s eta=35:34:28 +2026-09-21 04:52:34 | INFO | train rank=0 | step=19340/50000 progress=38.7% loss=0.349245 lr=7.206375e-05 lr_backbone=7.206375e-05 elapsed=22:25:57 avg_step=4.18s recent_step=4.16s eta=35:33:46 +2026-09-21 04:53:15 | INFO | train rank=0 | step=19350/50000 progress=38.7% loss=0.306638 lr=7.203407e-05 lr_backbone=7.203407e-05 elapsed=22:26:39 avg_step=4.18s recent_step=4.13s eta=35:33:04 +2026-09-21 04:53:57 | INFO | train rank=0 | step=19360/50000 progress=38.7% loss=0.348852 lr=7.200438e-05 lr_backbone=7.200438e-05 elapsed=22:27:20 avg_step=4.18s recent_step=4.17s eta=35:32:22 +2026-09-21 04:54:38 | INFO | train rank=0 | step=19370/50000 progress=38.7% loss=0.382608 lr=7.197468e-05 lr_backbone=7.197468e-05 elapsed=22:28:01 avg_step=4.18s recent_step=4.10s eta=35:31:39 +2026-09-21 04:55:19 | INFO | train rank=0 | step=19380/50000 progress=38.8% loss=0.351502 lr=7.194497e-05 lr_backbone=7.194497e-05 elapsed=22:28:43 avg_step=4.18s recent_step=4.14s eta=35:30:57 +2026-09-21 04:56:01 | INFO | train rank=0 | step=19390/50000 progress=38.8% loss=0.295941 lr=7.191525e-05 lr_backbone=7.191525e-05 elapsed=22:29:24 avg_step=4.18s recent_step=4.13s eta=35:30:14 +2026-09-21 04:56:42 | INFO | train rank=0 | step=19400/50000 progress=38.8% loss=0.380956 lr=7.188553e-05 lr_backbone=7.188553e-05 elapsed=22:30:06 avg_step=4.18s recent_step=4.15s eta=35:29:32 +2026-09-21 04:57:24 | INFO | train rank=0 | step=19410/50000 progress=38.8% loss=0.332604 lr=7.185579e-05 lr_backbone=7.185579e-05 elapsed=22:30:47 avg_step=4.18s recent_step=4.15s eta=35:28:50 +2026-09-21 04:58:05 | INFO | train rank=0 | step=19420/50000 progress=38.8% loss=0.314253 lr=7.182604e-05 lr_backbone=7.182604e-05 elapsed=22:31:29 avg_step=4.18s recent_step=4.16s eta=35:28:08 +2026-09-21 04:58:46 | INFO | train rank=0 | step=19430/50000 progress=38.9% loss=0.433126 lr=7.179628e-05 lr_backbone=7.179628e-05 elapsed=22:32:10 avg_step=4.18s recent_step=4.10s eta=35:27:25 +2026-09-21 04:59:28 | INFO | train rank=0 | step=19440/50000 progress=38.9% loss=0.353953 lr=7.176652e-05 lr_backbone=7.176652e-05 elapsed=22:32:51 avg_step=4.18s recent_step=4.13s eta=35:26:43 +2026-09-21 05:00:09 | INFO | train rank=0 | step=19450/50000 progress=38.9% loss=0.299583 lr=7.173674e-05 lr_backbone=7.173674e-05 elapsed=22:33:32 avg_step=4.18s recent_step=4.14s eta=35:26:00 +2026-09-21 05:00:51 | INFO | train rank=0 | step=19460/50000 progress=38.9% loss=0.426174 lr=7.170696e-05 lr_backbone=7.170696e-05 elapsed=22:34:14 avg_step=4.18s recent_step=4.17s eta=35:25:18 +2026-09-21 05:01:32 | INFO | train rank=0 | step=19470/50000 progress=38.9% loss=0.353625 lr=7.167716e-05 lr_backbone=7.167716e-05 elapsed=22:34:56 avg_step=4.18s recent_step=4.16s eta=35:24:36 +2026-09-21 05:02:14 | INFO | train rank=0 | step=19480/50000 progress=39.0% loss=0.385833 lr=7.164736e-05 lr_backbone=7.164736e-05 elapsed=22:35:37 avg_step=4.18s recent_step=4.15s eta=35:23:54 +2026-09-21 05:02:55 | INFO | train rank=0 | step=19490/50000 progress=39.0% loss=0.360579 lr=7.161754e-05 lr_backbone=7.161754e-05 elapsed=22:36:18 avg_step=4.18s recent_step=4.10s eta=35:23:11 +2026-09-21 05:03:36 | INFO | train rank=0 | step=19500/50000 progress=39.0% loss=0.323096 lr=7.158772e-05 lr_backbone=7.158772e-05 elapsed=22:37:00 avg_step=4.18s recent_step=4.14s eta=35:22:29 +2026-09-21 05:04:18 | INFO | train rank=0 | step=19510/50000 progress=39.0% loss=0.324222 lr=7.155788e-05 lr_backbone=7.155788e-05 elapsed=22:37:41 avg_step=4.18s recent_step=4.15s eta=35:21:47 +2026-09-21 05:04:59 | INFO | train rank=0 | step=19520/50000 progress=39.0% loss=0.431682 lr=7.152804e-05 lr_backbone=7.152804e-05 elapsed=22:38:22 avg_step=4.18s recent_step=4.13s eta=35:21:04 +2026-09-21 05:05:41 | INFO | train rank=0 | step=19530/50000 progress=39.1% loss=0.307245 lr=7.149819e-05 lr_backbone=7.149819e-05 elapsed=22:39:04 avg_step=4.18s recent_step=4.16s eta=35:20:22 +2026-09-21 05:06:22 | INFO | train rank=0 | step=19540/50000 progress=39.1% loss=0.409460 lr=7.146833e-05 lr_backbone=7.146833e-05 elapsed=22:39:46 avg_step=4.18s recent_step=4.17s eta=35:19:40 +2026-09-21 05:07:03 | INFO | train rank=0 | step=19550/50000 progress=39.1% loss=0.294335 lr=7.143846e-05 lr_backbone=7.143846e-05 elapsed=22:40:27 avg_step=4.18s recent_step=4.10s eta=35:18:57 +2026-09-21 05:07:45 | INFO | train rank=0 | step=19560/50000 progress=39.1% loss=0.340942 lr=7.140858e-05 lr_backbone=7.140858e-05 elapsed=22:41:08 avg_step=4.18s recent_step=4.14s eta=35:18:15 +2026-09-21 05:08:26 | INFO | train rank=0 | step=19570/50000 progress=39.1% loss=0.332978 lr=7.137869e-05 lr_backbone=7.137869e-05 elapsed=22:41:50 avg_step=4.18s recent_step=4.14s eta=35:17:33 +2026-09-21 05:09:07 | INFO | train rank=0 | step=19580/50000 progress=39.2% loss=0.313938 lr=7.134879e-05 lr_backbone=7.134879e-05 elapsed=22:42:31 avg_step=4.18s recent_step=4.13s eta=35:16:50 +2026-09-21 05:09:49 | INFO | train rank=0 | step=19590/50000 progress=39.2% loss=0.390167 lr=7.131888e-05 lr_backbone=7.131888e-05 elapsed=22:43:12 avg_step=4.18s recent_step=4.16s eta=35:16:08 +2026-09-21 05:10:31 | INFO | train rank=0 | step=19600/50000 progress=39.2% loss=0.322227 lr=7.128896e-05 lr_backbone=7.128896e-05 elapsed=22:43:54 avg_step=4.18s recent_step=4.15s eta=35:15:26 +2026-09-21 05:11:11 | INFO | train rank=0 | step=19610/50000 progress=39.2% loss=0.331528 lr=7.125904e-05 lr_backbone=7.125904e-05 elapsed=22:44:35 avg_step=4.18s recent_step=4.10s eta=35:14:43 +2026-09-21 05:11:53 | INFO | train rank=0 | step=19620/50000 progress=39.2% loss=0.318141 lr=7.122910e-05 lr_backbone=7.122910e-05 elapsed=22:45:16 avg_step=4.18s recent_step=4.13s eta=35:14:01 +2026-09-21 05:12:34 | INFO | train rank=0 | step=19630/50000 progress=39.3% loss=0.354128 lr=7.119916e-05 lr_backbone=7.119916e-05 elapsed=22:45:58 avg_step=4.18s recent_step=4.16s eta=35:13:19 +2026-09-21 05:13:16 | INFO | train rank=0 | step=19640/50000 progress=39.3% loss=0.373753 lr=7.116920e-05 lr_backbone=7.116920e-05 elapsed=22:46:39 avg_step=4.18s recent_step=4.14s eta=35:12:37 +2026-09-21 05:13:57 | INFO | train rank=0 | step=19650/50000 progress=39.3% loss=0.350097 lr=7.113924e-05 lr_backbone=7.113924e-05 elapsed=22:47:21 avg_step=4.18s recent_step=4.16s eta=35:11:55 +2026-09-21 05:14:39 | INFO | train rank=0 | step=19660/50000 progress=39.3% loss=0.369431 lr=7.110926e-05 lr_backbone=7.110926e-05 elapsed=22:48:02 avg_step=4.18s recent_step=4.15s eta=35:11:12 +2026-09-21 05:15:20 | INFO | train rank=0 | step=19670/50000 progress=39.3% loss=0.350950 lr=7.107928e-05 lr_backbone=7.107928e-05 elapsed=22:48:43 avg_step=4.18s recent_step=4.09s eta=35:10:29 +2026-09-21 05:16:01 | INFO | train rank=0 | step=19680/50000 progress=39.4% loss=0.353443 lr=7.104929e-05 lr_backbone=7.104929e-05 elapsed=22:49:24 avg_step=4.18s recent_step=4.12s eta=35:09:47 +2026-09-21 05:16:42 | INFO | train rank=0 | step=19690/50000 progress=39.4% loss=0.343003 lr=7.101929e-05 lr_backbone=7.101929e-05 elapsed=22:50:06 avg_step=4.18s recent_step=4.12s eta=35:09:04 +2026-09-21 05:17:24 | INFO | train rank=0 | step=19700/50000 progress=39.4% loss=0.315785 lr=7.098928e-05 lr_backbone=7.098928e-05 elapsed=22:50:47 avg_step=4.17s recent_step=4.14s eta=35:08:22 +2026-09-21 05:18:05 | INFO | train rank=0 | step=19710/50000 progress=39.4% loss=0.331783 lr=7.095926e-05 lr_backbone=7.095926e-05 elapsed=22:51:29 avg_step=4.17s recent_step=4.16s eta=35:07:40 +2026-09-21 05:18:47 | INFO | train rank=0 | step=19720/50000 progress=39.4% loss=0.317946 lr=7.092923e-05 lr_backbone=7.092923e-05 elapsed=22:52:10 avg_step=4.17s recent_step=4.16s eta=35:06:58 +2026-09-21 05:19:28 | INFO | train rank=0 | step=19730/50000 progress=39.5% loss=0.295775 lr=7.089920e-05 lr_backbone=7.089920e-05 elapsed=22:52:51 avg_step=4.17s recent_step=4.09s eta=35:06:15 +2026-09-21 05:20:09 | INFO | train rank=0 | step=19740/50000 progress=39.5% loss=0.375604 lr=7.086915e-05 lr_backbone=7.086915e-05 elapsed=22:53:32 avg_step=4.17s recent_step=4.14s eta=35:05:33 +2026-09-21 05:20:51 | INFO | train rank=0 | step=19750/50000 progress=39.5% loss=0.351019 lr=7.083909e-05 lr_backbone=7.083909e-05 elapsed=22:54:14 avg_step=4.17s recent_step=4.17s eta=35:04:51 +2026-09-21 05:21:32 | INFO | train rank=0 | step=19760/50000 progress=39.5% loss=0.364961 lr=7.080903e-05 lr_backbone=7.080903e-05 elapsed=22:54:55 avg_step=4.17s recent_step=4.13s eta=35:04:08 +2026-09-21 05:22:14 | INFO | train rank=0 | step=19770/50000 progress=39.5% loss=0.314591 lr=7.077895e-05 lr_backbone=7.077895e-05 elapsed=22:55:37 avg_step=4.17s recent_step=4.15s eta=35:03:26 +2026-09-21 05:22:55 | INFO | train rank=0 | step=19780/50000 progress=39.6% loss=0.340102 lr=7.074887e-05 lr_backbone=7.074887e-05 elapsed=22:56:19 avg_step=4.17s recent_step=4.17s eta=35:02:44 +2026-09-21 05:23:36 | INFO | train rank=0 | step=19790/50000 progress=39.6% loss=0.493797 lr=7.071878e-05 lr_backbone=7.071878e-05 elapsed=22:57:00 avg_step=4.17s recent_step=4.11s eta=35:02:01 +2026-09-21 05:24:18 | INFO | train rank=0 | step=19800/50000 progress=39.6% loss=0.409177 lr=7.068868e-05 lr_backbone=7.068868e-05 elapsed=22:57:41 avg_step=4.17s recent_step=4.13s eta=35:01:19 +2026-09-21 05:24:59 | INFO | train rank=0 | step=19810/50000 progress=39.6% loss=0.363335 lr=7.065857e-05 lr_backbone=7.065857e-05 elapsed=22:58:23 avg_step=4.17s recent_step=4.17s eta=35:00:37 +2026-09-21 05:25:41 | INFO | train rank=0 | step=19820/50000 progress=39.6% loss=0.334480 lr=7.062845e-05 lr_backbone=7.062845e-05 elapsed=22:59:04 avg_step=4.17s recent_step=4.13s eta=34:59:55 +2026-09-21 05:26:22 | INFO | train rank=0 | step=19830/50000 progress=39.7% loss=0.387921 lr=7.059832e-05 lr_backbone=7.059832e-05 elapsed=22:59:46 avg_step=4.17s recent_step=4.17s eta=34:59:13 +2026-09-21 05:27:04 | INFO | train rank=0 | step=19840/50000 progress=39.7% loss=0.348746 lr=7.056818e-05 lr_backbone=7.056818e-05 elapsed=23:00:27 avg_step=4.17s recent_step=4.16s eta=34:58:31 +2026-09-21 05:27:45 | INFO | train rank=0 | step=19850/50000 progress=39.7% loss=0.328016 lr=7.053804e-05 lr_backbone=7.053804e-05 elapsed=23:01:08 avg_step=4.17s recent_step=4.11s eta=34:57:48 +2026-09-21 05:28:26 | INFO | train rank=0 | step=19860/50000 progress=39.7% loss=0.317938 lr=7.050788e-05 lr_backbone=7.050788e-05 elapsed=23:01:50 avg_step=4.17s recent_step=4.14s eta=34:57:06 +2026-09-21 05:29:07 | INFO | train rank=0 | step=19870/50000 progress=39.7% loss=0.335321 lr=7.047772e-05 lr_backbone=7.047772e-05 elapsed=23:02:31 avg_step=4.17s recent_step=4.12s eta=34:56:23 +2026-09-21 05:29:49 | INFO | train rank=0 | step=19880/50000 progress=39.8% loss=0.337486 lr=7.044754e-05 lr_backbone=7.044754e-05 elapsed=23:03:12 avg_step=4.17s recent_step=4.12s eta=34:55:41 +2026-09-21 05:30:30 | INFO | train rank=0 | step=19890/50000 progress=39.8% loss=0.398608 lr=7.041736e-05 lr_backbone=7.041736e-05 elapsed=23:03:53 avg_step=4.17s recent_step=4.14s eta=34:54:58 +2026-09-21 05:31:12 | INFO | train rank=0 | step=19900/50000 progress=39.8% loss=0.366744 lr=7.038717e-05 lr_backbone=7.038717e-05 elapsed=23:04:35 avg_step=4.17s recent_step=4.16s eta=34:54:16 +2026-09-21 05:31:53 | INFO | train rank=0 | step=19910/50000 progress=39.8% loss=0.372776 lr=7.035697e-05 lr_backbone=7.035697e-05 elapsed=23:05:16 avg_step=4.17s recent_step=4.10s eta=34:53:34 +2026-09-21 05:32:34 | INFO | train rank=0 | step=19920/50000 progress=39.8% loss=0.346880 lr=7.032676e-05 lr_backbone=7.032676e-05 elapsed=23:05:57 avg_step=4.17s recent_step=4.12s eta=34:52:51 +2026-09-21 05:33:15 | INFO | train rank=0 | step=19930/50000 progress=39.9% loss=0.318996 lr=7.029654e-05 lr_backbone=7.029654e-05 elapsed=23:06:39 avg_step=4.17s recent_step=4.15s eta=34:52:09 +2026-09-21 05:33:57 | INFO | train rank=0 | step=19940/50000 progress=39.9% loss=0.365914 lr=7.026632e-05 lr_backbone=7.026632e-05 elapsed=23:07:20 avg_step=4.17s recent_step=4.14s eta=34:51:26 +2026-09-21 05:34:38 | INFO | train rank=0 | step=19950/50000 progress=39.9% loss=0.349253 lr=7.023608e-05 lr_backbone=7.023608e-05 elapsed=23:08:02 avg_step=4.17s recent_step=4.17s eta=34:50:45 +2026-09-21 05:35:20 | INFO | train rank=0 | step=19960/50000 progress=39.9% loss=0.303403 lr=7.020584e-05 lr_backbone=7.020584e-05 elapsed=23:08:43 avg_step=4.17s recent_step=4.17s eta=34:50:03 +2026-09-21 05:36:01 | INFO | train rank=0 | step=19970/50000 progress=39.9% loss=0.379323 lr=7.017558e-05 lr_backbone=7.017558e-05 elapsed=23:09:24 avg_step=4.17s recent_step=4.11s eta=34:49:20 +2026-09-21 05:36:43 | INFO | train rank=0 | step=19980/50000 progress=40.0% loss=0.401746 lr=7.014532e-05 lr_backbone=7.014532e-05 elapsed=23:10:06 avg_step=4.17s recent_step=4.15s eta=34:48:38 +2026-09-21 05:37:24 | INFO | train rank=0 | step=19990/50000 progress=40.0% loss=0.371440 lr=7.011505e-05 lr_backbone=7.011505e-05 elapsed=23:10:47 avg_step=4.17s recent_step=4.15s eta=34:47:56 +2026-09-21 05:38:05 | INFO | train rank=0 | step=20000/50000 progress=40.0% loss=0.307753 lr=7.008477e-05 lr_backbone=7.008477e-05 elapsed=23:11:29 avg_step=4.17s recent_step=4.12s eta=34:47:13 +2026-09-21 05:38:47 | INFO | train rank=0 | step=20010/50000 progress=40.0% loss=0.345185 lr=7.005448e-05 lr_backbone=7.005448e-05 elapsed=23:12:10 avg_step=4.17s recent_step=4.14s eta=34:46:31 +2026-09-21 05:39:28 | INFO | train rank=0 | step=20020/50000 progress=40.0% loss=0.392882 lr=7.002419e-05 lr_backbone=7.002419e-05 elapsed=23:12:52 avg_step=4.17s recent_step=4.16s eta=34:45:49 +2026-09-21 05:40:09 | INFO | train rank=0 | step=20030/50000 progress=40.1% loss=0.303392 lr=6.999388e-05 lr_backbone=6.999388e-05 elapsed=23:13:33 avg_step=4.17s recent_step=4.10s eta=34:45:06 +2026-09-21 05:40:51 | INFO | train rank=0 | step=20040/50000 progress=40.1% loss=0.326088 lr=6.996356e-05 lr_backbone=6.996356e-05 elapsed=23:14:14 avg_step=4.17s recent_step=4.13s eta=34:44:24 +2026-09-21 05:41:32 | INFO | train rank=0 | step=20050/50000 progress=40.1% loss=0.366827 lr=6.993324e-05 lr_backbone=6.993324e-05 elapsed=23:14:55 avg_step=4.17s recent_step=4.16s eta=34:43:42 +2026-09-21 05:42:13 | INFO | train rank=0 | step=20060/50000 progress=40.1% loss=0.342967 lr=6.990291e-05 lr_backbone=6.990291e-05 elapsed=23:15:37 avg_step=4.17s recent_step=4.14s eta=34:42:59 +2026-09-21 05:42:54 | INFO | train rank=0 | step=20070/50000 progress=40.1% loss=0.315714 lr=6.987257e-05 lr_backbone=6.987257e-05 elapsed=23:16:18 avg_step=4.17s recent_step=4.10s eta=34:42:16 +2026-09-21 05:43:36 | INFO | train rank=0 | step=20080/50000 progress=40.2% loss=0.350360 lr=6.984222e-05 lr_backbone=6.984222e-05 elapsed=23:17:00 avg_step=4.17s recent_step=4.17s eta=34:41:35 +2026-09-21 05:44:17 | INFO | train rank=0 | step=20090/50000 progress=40.2% loss=0.358829 lr=6.981186e-05 lr_backbone=6.981186e-05 elapsed=23:17:41 avg_step=4.17s recent_step=4.10s eta=34:40:52 +2026-09-21 05:44:58 | INFO | train rank=0 | step=20100/50000 progress=40.2% loss=0.479306 lr=6.978149e-05 lr_backbone=6.978149e-05 elapsed=23:18:22 avg_step=4.17s recent_step=4.12s eta=34:40:09 +2026-09-21 05:45:40 | INFO | train rank=0 | step=20110/50000 progress=40.2% loss=0.317148 lr=6.975112e-05 lr_backbone=6.975112e-05 elapsed=23:19:03 avg_step=4.17s recent_step=4.17s eta=34:39:28 +2026-09-21 05:46:22 | INFO | train rank=0 | step=20120/50000 progress=40.2% loss=0.336437 lr=6.972073e-05 lr_backbone=6.972073e-05 elapsed=23:19:45 avg_step=4.17s recent_step=4.16s eta=34:38:45 +2026-09-21 05:47:03 | INFO | train rank=0 | step=20130/50000 progress=40.3% loss=0.370498 lr=6.969034e-05 lr_backbone=6.969034e-05 elapsed=23:20:27 avg_step=4.17s recent_step=4.16s eta=34:38:04 +2026-09-21 05:47:45 | INFO | train rank=0 | step=20140/50000 progress=40.3% loss=0.390635 lr=6.965994e-05 lr_backbone=6.965994e-05 elapsed=23:21:08 avg_step=4.17s recent_step=4.16s eta=34:37:22 +2026-09-21 05:48:26 | INFO | train rank=0 | step=20150/50000 progress=40.3% loss=0.360742 lr=6.962953e-05 lr_backbone=6.962953e-05 elapsed=23:21:49 avg_step=4.17s recent_step=4.09s eta=34:36:39 +2026-09-21 05:49:07 | INFO | train rank=0 | step=20160/50000 progress=40.3% loss=0.453461 lr=6.959911e-05 lr_backbone=6.959911e-05 elapsed=23:22:30 avg_step=4.17s recent_step=4.14s eta=34:35:56 +2026-09-21 05:49:49 | INFO | train rank=0 | step=20170/50000 progress=40.3% loss=0.346107 lr=6.956868e-05 lr_backbone=6.956868e-05 elapsed=23:23:12 avg_step=4.17s recent_step=4.16s eta=34:35:14 +2026-09-21 05:50:30 | INFO | train rank=0 | step=20180/50000 progress=40.4% loss=0.314552 lr=6.953825e-05 lr_backbone=6.953825e-05 elapsed=23:23:53 avg_step=4.17s recent_step=4.14s eta=34:34:32 +2026-09-21 05:51:11 | INFO | train rank=0 | step=20190/50000 progress=40.4% loss=0.358500 lr=6.950780e-05 lr_backbone=6.950780e-05 elapsed=23:24:35 avg_step=4.17s recent_step=4.13s eta=34:33:50 +2026-09-21 05:51:53 | INFO | train rank=0 | step=20200/50000 progress=40.4% loss=0.372934 lr=6.947735e-05 lr_backbone=6.947735e-05 elapsed=23:25:16 avg_step=4.17s recent_step=4.17s eta=34:33:08 +2026-09-21 05:52:34 | INFO | train rank=0 | step=20210/50000 progress=40.4% loss=0.372152 lr=6.944689e-05 lr_backbone=6.944689e-05 elapsed=23:25:57 avg_step=4.17s recent_step=4.10s eta=34:32:25 +2026-09-21 05:53:15 | INFO | train rank=0 | step=20220/50000 progress=40.4% loss=0.354483 lr=6.941642e-05 lr_backbone=6.941642e-05 elapsed=23:26:39 avg_step=4.17s recent_step=4.11s eta=34:31:42 +2026-09-21 05:53:57 | INFO | train rank=0 | step=20230/50000 progress=40.5% loss=0.396032 lr=6.938594e-05 lr_backbone=6.938594e-05 elapsed=23:27:20 avg_step=4.17s recent_step=4.15s eta=34:31:00 +2026-09-21 05:54:38 | INFO | train rank=0 | step=20240/50000 progress=40.5% loss=0.397884 lr=6.935545e-05 lr_backbone=6.935545e-05 elapsed=23:28:02 avg_step=4.17s recent_step=4.17s eta=34:30:18 +2026-09-21 05:55:20 | INFO | train rank=0 | step=20250/50000 progress=40.5% loss=0.430030 lr=6.932496e-05 lr_backbone=6.932496e-05 elapsed=23:28:43 avg_step=4.17s recent_step=4.16s eta=34:29:37 +2026-09-21 05:56:01 | INFO | train rank=0 | step=20260/50000 progress=40.5% loss=0.375525 lr=6.929445e-05 lr_backbone=6.929445e-05 elapsed=23:29:24 avg_step=4.17s recent_step=4.11s eta=34:28:54 +2026-09-21 05:56:42 | INFO | train rank=0 | step=20270/50000 progress=40.5% loss=0.376906 lr=6.926394e-05 lr_backbone=6.926394e-05 elapsed=23:30:05 avg_step=4.17s recent_step=4.10s eta=34:28:11 +2026-09-21 05:57:23 | INFO | train rank=0 | step=20280/50000 progress=40.6% loss=0.379165 lr=6.923342e-05 lr_backbone=6.923342e-05 elapsed=23:30:47 avg_step=4.17s recent_step=4.12s eta=34:27:29 +2026-09-21 05:58:05 | INFO | train rank=0 | step=20290/50000 progress=40.6% loss=0.377377 lr=6.920289e-05 lr_backbone=6.920289e-05 elapsed=23:31:28 avg_step=4.17s recent_step=4.15s eta=34:26:46 +2026-09-21 05:58:47 | INFO | train rank=0 | step=20300/50000 progress=40.6% loss=0.398060 lr=6.917236e-05 lr_backbone=6.917236e-05 elapsed=23:32:10 avg_step=4.17s recent_step=4.17s eta=34:26:05 +2026-09-21 05:59:28 | INFO | train rank=0 | step=20310/50000 progress=40.6% loss=0.348419 lr=6.914181e-05 lr_backbone=6.914181e-05 elapsed=23:32:51 avg_step=4.17s recent_step=4.15s eta=34:25:23 +2026-09-21 06:00:10 | INFO | train rank=0 | step=20320/50000 progress=40.6% loss=0.350054 lr=6.911126e-05 lr_backbone=6.911126e-05 elapsed=23:33:33 avg_step=4.17s recent_step=4.15s eta=34:24:41 +2026-09-21 06:00:51 | INFO | train rank=0 | step=20330/50000 progress=40.7% loss=0.323565 lr=6.908069e-05 lr_backbone=6.908069e-05 elapsed=23:34:14 avg_step=4.17s recent_step=4.10s eta=34:23:58 +2026-09-21 06:01:32 | INFO | train rank=0 | step=20340/50000 progress=40.7% loss=0.385413 lr=6.905012e-05 lr_backbone=6.905012e-05 elapsed=23:34:55 avg_step=4.17s recent_step=4.12s eta=34:23:15 +2026-09-21 06:02:13 | INFO | train rank=0 | step=20350/50000 progress=40.7% loss=0.326553 lr=6.901954e-05 lr_backbone=6.901954e-05 elapsed=23:35:37 avg_step=4.17s recent_step=4.14s eta=34:22:33 +2026-09-21 06:02:55 | INFO | train rank=0 | step=20360/50000 progress=40.7% loss=0.313434 lr=6.898895e-05 lr_backbone=6.898895e-05 elapsed=23:36:18 avg_step=4.17s recent_step=4.15s eta=34:21:51 +2026-09-21 06:03:36 | INFO | train rank=0 | step=20370/50000 progress=40.7% loss=0.350622 lr=6.895836e-05 lr_backbone=6.895836e-05 elapsed=23:37:00 avg_step=4.17s recent_step=4.15s eta=34:21:09 +2026-09-21 06:04:18 | INFO | train rank=0 | step=20380/50000 progress=40.8% loss=0.370976 lr=6.892775e-05 lr_backbone=6.892775e-05 elapsed=23:37:41 avg_step=4.17s recent_step=4.13s eta=34:20:27 +2026-09-21 06:04:59 | INFO | train rank=0 | step=20390/50000 progress=40.8% loss=0.437985 lr=6.889714e-05 lr_backbone=6.889714e-05 elapsed=23:38:22 avg_step=4.17s recent_step=4.10s eta=34:19:44 +2026-09-21 06:05:40 | INFO | train rank=0 | step=20400/50000 progress=40.8% loss=0.396155 lr=6.886652e-05 lr_backbone=6.886652e-05 elapsed=23:39:03 avg_step=4.17s recent_step=4.13s eta=34:19:01 +2026-09-21 06:06:21 | INFO | train rank=0 | step=20410/50000 progress=40.8% loss=0.317617 lr=6.883589e-05 lr_backbone=6.883589e-05 elapsed=23:39:45 avg_step=4.17s recent_step=4.15s eta=34:18:19 +2026-09-21 06:07:03 | INFO | train rank=0 | step=20420/50000 progress=40.8% loss=0.410777 lr=6.880526e-05 lr_backbone=6.880526e-05 elapsed=23:40:26 avg_step=4.17s recent_step=4.17s eta=34:17:37 +2026-09-21 06:07:44 | INFO | train rank=0 | step=20430/50000 progress=40.9% loss=0.323399 lr=6.877461e-05 lr_backbone=6.877461e-05 elapsed=23:41:08 avg_step=4.17s recent_step=4.14s eta=34:16:55 +2026-09-21 06:08:26 | INFO | train rank=0 | step=20440/50000 progress=40.9% loss=0.307682 lr=6.874396e-05 lr_backbone=6.874396e-05 elapsed=23:41:49 avg_step=4.17s recent_step=4.16s eta=34:16:13 +2026-09-21 06:09:07 | INFO | train rank=0 | step=20450/50000 progress=40.9% loss=0.314960 lr=6.871329e-05 lr_backbone=6.871329e-05 elapsed=23:42:30 avg_step=4.17s recent_step=4.10s eta=34:15:30 +2026-09-21 06:09:48 | INFO | train rank=0 | step=20460/50000 progress=40.9% loss=0.358044 lr=6.868262e-05 lr_backbone=6.868262e-05 elapsed=23:43:12 avg_step=4.17s recent_step=4.12s eta=34:14:48 +2026-09-21 06:10:30 | INFO | train rank=0 | step=20470/50000 progress=40.9% loss=0.389086 lr=6.865195e-05 lr_backbone=6.865195e-05 elapsed=23:43:53 avg_step=4.17s recent_step=4.16s eta=34:14:06 +2026-09-21 06:11:11 | INFO | train rank=0 | step=20480/50000 progress=41.0% loss=0.408534 lr=6.862126e-05 lr_backbone=6.862126e-05 elapsed=23:44:35 avg_step=4.17s recent_step=4.14s eta=34:13:24 +2026-09-21 06:11:53 | INFO | train rank=0 | step=20490/50000 progress=41.0% loss=0.305499 lr=6.859057e-05 lr_backbone=6.859057e-05 elapsed=23:45:16 avg_step=4.17s recent_step=4.17s eta=34:12:42 +2026-09-21 06:12:34 | INFO | train rank=0 | step=20500/50000 progress=41.0% loss=0.407828 lr=6.855986e-05 lr_backbone=6.855986e-05 elapsed=23:45:58 avg_step=4.17s recent_step=4.14s eta=34:12:00 +2026-09-21 06:13:15 | INFO | train rank=0 | step=20510/50000 progress=41.0% loss=0.363314 lr=6.852915e-05 lr_backbone=6.852915e-05 elapsed=23:46:39 avg_step=4.17s recent_step=4.10s eta=34:11:17 +2026-09-21 06:13:57 | INFO | train rank=0 | step=20520/50000 progress=41.0% loss=0.376304 lr=6.849843e-05 lr_backbone=6.849843e-05 elapsed=23:47:20 avg_step=4.17s recent_step=4.13s eta=34:10:35 +2026-09-21 06:14:38 | INFO | train rank=0 | step=20530/50000 progress=41.1% loss=0.410962 lr=6.846771e-05 lr_backbone=6.846771e-05 elapsed=23:48:02 avg_step=4.17s recent_step=4.16s eta=34:09:53 +2026-09-21 06:15:20 | INFO | train rank=0 | step=20540/50000 progress=41.1% loss=0.316379 lr=6.843697e-05 lr_backbone=6.843697e-05 elapsed=23:48:43 avg_step=4.17s recent_step=4.16s eta=34:09:11 +2026-09-21 06:16:01 | INFO | train rank=0 | step=20550/50000 progress=41.1% loss=0.503109 lr=6.840623e-05 lr_backbone=6.840623e-05 elapsed=23:49:25 avg_step=4.17s recent_step=4.15s eta=34:08:29 +2026-09-21 06:16:43 | INFO | train rank=0 | step=20560/50000 progress=41.1% loss=0.298585 lr=6.837548e-05 lr_backbone=6.837548e-05 elapsed=23:50:06 avg_step=4.17s recent_step=4.14s eta=34:07:47 +2026-09-21 06:17:24 | INFO | train rank=0 | step=20570/50000 progress=41.1% loss=0.329807 lr=6.834472e-05 lr_backbone=6.834472e-05 elapsed=23:50:47 avg_step=4.17s recent_step=4.11s eta=34:07:04 +2026-09-21 06:18:05 | INFO | train rank=0 | step=20580/50000 progress=41.2% loss=0.360192 lr=6.831395e-05 lr_backbone=6.831395e-05 elapsed=23:51:28 avg_step=4.17s recent_step=4.12s eta=34:06:21 +2026-09-21 06:18:47 | INFO | train rank=0 | step=20590/50000 progress=41.2% loss=0.324033 lr=6.828317e-05 lr_backbone=6.828317e-05 elapsed=23:52:10 avg_step=4.17s recent_step=4.16s eta=34:05:39 +2026-09-21 06:19:28 | INFO | train rank=0 | step=20600/50000 progress=41.2% loss=0.341217 lr=6.825239e-05 lr_backbone=6.825239e-05 elapsed=23:52:51 avg_step=4.17s recent_step=4.12s eta=34:04:57 +2026-09-21 06:20:09 | INFO | train rank=0 | step=20610/50000 progress=41.2% loss=0.366744 lr=6.822160e-05 lr_backbone=6.822160e-05 elapsed=23:53:33 avg_step=4.17s recent_step=4.16s eta=34:04:15 +2026-09-21 06:20:51 | INFO | train rank=0 | step=20620/50000 progress=41.2% loss=0.436529 lr=6.819080e-05 lr_backbone=6.819080e-05 elapsed=23:54:14 avg_step=4.17s recent_step=4.13s eta=34:03:33 +2026-09-21 06:21:32 | INFO | train rank=0 | step=20630/50000 progress=41.3% loss=0.311776 lr=6.815999e-05 lr_backbone=6.815999e-05 elapsed=23:54:55 avg_step=4.17s recent_step=4.11s eta=34:02:50 +2026-09-21 06:22:13 | INFO | train rank=0 | step=20640/50000 progress=41.3% loss=0.382299 lr=6.812918e-05 lr_backbone=6.812918e-05 elapsed=23:55:37 avg_step=4.17s recent_step=4.14s eta=34:02:08 +2026-09-21 06:22:55 | INFO | train rank=0 | step=20650/50000 progress=41.3% loss=0.416969 lr=6.809836e-05 lr_backbone=6.809836e-05 elapsed=23:56:18 avg_step=4.17s recent_step=4.15s eta=34:01:26 +2026-09-21 06:23:36 | INFO | train rank=0 | step=20660/50000 progress=41.3% loss=0.355400 lr=6.806752e-05 lr_backbone=6.806752e-05 elapsed=23:56:59 avg_step=4.17s recent_step=4.14s eta=34:00:43 +2026-09-21 06:24:17 | INFO | train rank=0 | step=20670/50000 progress=41.3% loss=0.330907 lr=6.803669e-05 lr_backbone=6.803669e-05 elapsed=23:57:41 avg_step=4.17s recent_step=4.13s eta=34:00:01 +2026-09-21 06:24:59 | INFO | train rank=0 | step=20680/50000 progress=41.4% loss=0.360377 lr=6.800584e-05 lr_backbone=6.800584e-05 elapsed=23:58:22 avg_step=4.17s recent_step=4.14s eta=33:59:19 +2026-09-21 06:25:40 | INFO | train rank=0 | step=20690/50000 progress=41.4% loss=0.341324 lr=6.797498e-05 lr_backbone=6.797498e-05 elapsed=23:59:03 avg_step=4.17s recent_step=4.08s eta=33:58:36 +2026-09-21 06:26:21 | INFO | train rank=0 | step=20700/50000 progress=41.4% loss=0.377552 lr=6.794412e-05 lr_backbone=6.794412e-05 elapsed=23:59:44 avg_step=4.17s recent_step=4.10s eta=33:57:53 +2026-09-21 06:27:02 | INFO | train rank=0 | step=20710/50000 progress=41.4% loss=0.339395 lr=6.791325e-05 lr_backbone=6.791325e-05 elapsed=24:00:25 avg_step=4.17s recent_step=4.13s eta=33:57:11 +2026-09-21 06:27:43 | INFO | train rank=0 | step=20720/50000 progress=41.4% loss=0.361698 lr=6.788237e-05 lr_backbone=6.788237e-05 elapsed=24:01:06 avg_step=4.17s recent_step=4.13s eta=33:56:28 +2026-09-21 06:28:24 | INFO | train rank=0 | step=20730/50000 progress=41.5% loss=0.374188 lr=6.785149e-05 lr_backbone=6.785149e-05 elapsed=24:01:48 avg_step=4.17s recent_step=4.14s eta=33:55:46 +2026-09-21 06:29:06 | INFO | train rank=0 | step=20740/50000 progress=41.5% loss=0.358843 lr=6.782059e-05 lr_backbone=6.782059e-05 elapsed=24:02:29 avg_step=4.17s recent_step=4.13s eta=33:55:04 +2026-09-21 06:29:47 | INFO | train rank=0 | step=20750/50000 progress=41.5% loss=0.419099 lr=6.778969e-05 lr_backbone=6.778969e-05 elapsed=24:03:10 avg_step=4.17s recent_step=4.08s eta=33:54:21 +2026-09-21 06:30:28 | INFO | train rank=0 | step=20760/50000 progress=41.5% loss=0.335219 lr=6.775878e-05 lr_backbone=6.775878e-05 elapsed=24:03:51 avg_step=4.17s recent_step=4.10s eta=33:53:38 +2026-09-21 06:31:09 | INFO | train rank=0 | step=20770/50000 progress=41.5% loss=0.378498 lr=6.772787e-05 lr_backbone=6.772787e-05 elapsed=24:04:32 avg_step=4.17s recent_step=4.12s eta=33:52:56 +2026-09-21 06:31:50 | INFO | train rank=0 | step=20780/50000 progress=41.6% loss=0.347440 lr=6.769694e-05 lr_backbone=6.769694e-05 elapsed=24:05:13 avg_step=4.17s recent_step=4.13s eta=33:52:13 +2026-09-21 06:32:32 | INFO | train rank=0 | step=20790/50000 progress=41.6% loss=0.464125 lr=6.766601e-05 lr_backbone=6.766601e-05 elapsed=24:05:55 avg_step=4.17s recent_step=4.15s eta=33:51:31 +2026-09-21 06:33:13 | INFO | train rank=0 | step=20800/50000 progress=41.6% loss=0.326869 lr=6.763507e-05 lr_backbone=6.763507e-05 elapsed=24:06:36 avg_step=4.17s recent_step=4.15s eta=33:50:49 +2026-09-21 06:33:54 | INFO | train rank=0 | step=20810/50000 progress=41.6% loss=0.359121 lr=6.760412e-05 lr_backbone=6.760412e-05 elapsed=24:07:17 avg_step=4.17s recent_step=4.08s eta=33:50:06 +2026-09-21 06:34:35 | INFO | train rank=0 | step=20820/50000 progress=41.6% loss=0.341458 lr=6.757316e-05 lr_backbone=6.757316e-05 elapsed=24:07:58 avg_step=4.17s recent_step=4.10s eta=33:49:23 +2026-09-21 06:35:16 | INFO | train rank=0 | step=20830/50000 progress=41.7% loss=0.347871 lr=6.754220e-05 lr_backbone=6.754220e-05 elapsed=24:08:40 avg_step=4.17s recent_step=4.14s eta=33:48:41 +2026-09-21 06:35:58 | INFO | train rank=0 | step=20840/50000 progress=41.7% loss=0.322835 lr=6.751123e-05 lr_backbone=6.751123e-05 elapsed=24:09:21 avg_step=4.17s recent_step=4.13s eta=33:47:59 +2026-09-21 06:36:39 | INFO | train rank=0 | step=20850/50000 progress=41.7% loss=0.387858 lr=6.748025e-05 lr_backbone=6.748025e-05 elapsed=24:10:02 avg_step=4.17s recent_step=4.14s eta=33:47:17 +2026-09-21 06:37:20 | INFO | train rank=0 | step=20860/50000 progress=41.7% loss=0.331570 lr=6.744926e-05 lr_backbone=6.744926e-05 elapsed=24:10:44 avg_step=4.17s recent_step=4.15s eta=33:46:35 +2026-09-21 06:38:01 | INFO | train rank=0 | step=20870/50000 progress=41.7% loss=0.370935 lr=6.741827e-05 lr_backbone=6.741827e-05 elapsed=24:11:25 avg_step=4.17s recent_step=4.07s eta=33:45:51 +2026-09-21 06:38:42 | INFO | train rank=0 | step=20880/50000 progress=41.8% loss=0.358480 lr=6.738727e-05 lr_backbone=6.738727e-05 elapsed=24:12:06 avg_step=4.17s recent_step=4.12s eta=33:45:09 +2026-09-21 06:39:24 | INFO | train rank=0 | step=20890/50000 progress=41.8% loss=0.385644 lr=6.735626e-05 lr_backbone=6.735626e-05 elapsed=24:12:47 avg_step=4.17s recent_step=4.14s eta=33:44:27 +2026-09-21 06:40:05 | INFO | train rank=0 | step=20900/50000 progress=41.8% loss=0.314471 lr=6.732524e-05 lr_backbone=6.732524e-05 elapsed=24:13:29 avg_step=4.17s recent_step=4.15s eta=33:43:45 +2026-09-21 06:40:47 | INFO | train rank=0 | step=20910/50000 progress=41.8% loss=0.395143 lr=6.729422e-05 lr_backbone=6.729422e-05 elapsed=24:14:10 avg_step=4.17s recent_step=4.14s eta=33:43:03 +2026-09-21 06:41:28 | INFO | train rank=0 | step=20920/50000 progress=41.8% loss=0.395230 lr=6.726319e-05 lr_backbone=6.726319e-05 elapsed=24:14:52 avg_step=4.17s recent_step=4.16s eta=33:42:21 +2026-09-21 06:42:09 | INFO | train rank=0 | step=20930/50000 progress=41.9% loss=0.339028 lr=6.723215e-05 lr_backbone=6.723215e-05 elapsed=24:15:33 avg_step=4.17s recent_step=4.10s eta=33:41:38 +2026-09-21 06:42:51 | INFO | train rank=0 | step=20940/50000 progress=41.9% loss=0.336807 lr=6.720110e-05 lr_backbone=6.720110e-05 elapsed=24:16:14 avg_step=4.17s recent_step=4.13s eta=33:40:56 +2026-09-21 06:43:32 | INFO | train rank=0 | step=20950/50000 progress=41.9% loss=0.438067 lr=6.717004e-05 lr_backbone=6.717004e-05 elapsed=24:16:55 avg_step=4.17s recent_step=4.10s eta=33:40:13 +2026-09-21 06:44:13 | INFO | train rank=0 | step=20960/50000 progress=41.9% loss=0.348220 lr=6.713898e-05 lr_backbone=6.713898e-05 elapsed=24:17:37 avg_step=4.17s recent_step=4.16s eta=33:39:31 +2026-09-21 06:44:55 | INFO | train rank=0 | step=20970/50000 progress=41.9% loss=0.311117 lr=6.710791e-05 lr_backbone=6.710791e-05 elapsed=24:18:18 avg_step=4.17s recent_step=4.17s eta=33:38:49 +2026-09-21 06:45:37 | INFO | train rank=0 | step=20980/50000 progress=42.0% loss=0.324203 lr=6.707684e-05 lr_backbone=6.707684e-05 elapsed=24:19:00 avg_step=4.17s recent_step=4.15s eta=33:38:07 +2026-09-21 06:46:18 | INFO | train rank=0 | step=20990/50000 progress=42.0% loss=0.297350 lr=6.704575e-05 lr_backbone=6.704575e-05 elapsed=24:19:41 avg_step=4.17s recent_step=4.11s eta=33:37:25 +2026-09-21 06:46:59 | INFO | train rank=0 | step=21000/50000 progress=42.0% loss=0.364618 lr=6.701466e-05 lr_backbone=6.701466e-05 elapsed=24:20:22 avg_step=4.17s recent_step=4.11s eta=33:36:42 +2026-09-21 06:47:40 | INFO | train rank=0 | step=21010/50000 progress=42.0% loss=0.312000 lr=6.698356e-05 lr_backbone=6.698356e-05 elapsed=24:21:03 avg_step=4.17s recent_step=4.13s eta=33:36:00 +2026-09-21 06:48:22 | INFO | train rank=0 | step=21020/50000 progress=42.0% loss=0.360664 lr=6.695245e-05 lr_backbone=6.695245e-05 elapsed=24:21:45 avg_step=4.17s recent_step=4.16s eta=33:35:18 +2026-09-21 06:49:03 | INFO | train rank=0 | step=21030/50000 progress=42.1% loss=0.392617 lr=6.692134e-05 lr_backbone=6.692134e-05 elapsed=24:22:26 avg_step=4.17s recent_step=4.15s eta=33:34:36 +2026-09-21 06:49:45 | INFO | train rank=0 | step=21040/50000 progress=42.1% loss=0.345941 lr=6.689022e-05 lr_backbone=6.689022e-05 elapsed=24:23:08 avg_step=4.17s recent_step=4.14s eta=33:33:54 +2026-09-21 06:50:26 | INFO | train rank=0 | step=21050/50000 progress=42.1% loss=0.337535 lr=6.685909e-05 lr_backbone=6.685909e-05 elapsed=24:23:49 avg_step=4.17s recent_step=4.11s eta=33:33:11 +2026-09-21 06:51:07 | INFO | train rank=0 | step=21060/50000 progress=42.1% loss=0.367213 lr=6.682795e-05 lr_backbone=6.682795e-05 elapsed=24:24:30 avg_step=4.17s recent_step=4.12s eta=33:32:29 +2026-09-21 06:51:48 | INFO | train rank=0 | step=21070/50000 progress=42.1% loss=0.388349 lr=6.679681e-05 lr_backbone=6.679681e-05 elapsed=24:25:12 avg_step=4.17s recent_step=4.15s eta=33:31:47 +2026-09-21 06:52:30 | INFO | train rank=0 | step=21080/50000 progress=42.2% loss=0.388253 lr=6.676566e-05 lr_backbone=6.676566e-05 elapsed=24:25:53 avg_step=4.17s recent_step=4.13s eta=33:31:04 +2026-09-21 06:53:11 | INFO | train rank=0 | step=21090/50000 progress=42.2% loss=0.354452 lr=6.673450e-05 lr_backbone=6.673450e-05 elapsed=24:26:35 avg_step=4.17s recent_step=4.16s eta=33:30:22 +2026-09-21 06:53:53 | INFO | train rank=0 | step=21100/50000 progress=42.2% loss=0.399430 lr=6.670333e-05 lr_backbone=6.670333e-05 elapsed=24:27:16 avg_step=4.17s recent_step=4.19s eta=33:29:41 +2026-09-21 06:54:34 | INFO | train rank=0 | step=21110/50000 progress=42.2% loss=0.357447 lr=6.667216e-05 lr_backbone=6.667216e-05 elapsed=24:27:57 avg_step=4.17s recent_step=4.09s eta=33:28:58 +2026-09-21 06:55:15 | INFO | train rank=0 | step=21120/50000 progress=42.2% loss=0.321788 lr=6.664098e-05 lr_backbone=6.664098e-05 elapsed=24:28:39 avg_step=4.17s recent_step=4.14s eta=33:28:16 +2026-09-21 06:55:57 | INFO | train rank=0 | step=21130/50000 progress=42.3% loss=0.358461 lr=6.660979e-05 lr_backbone=6.660979e-05 elapsed=24:29:20 avg_step=4.17s recent_step=4.15s eta=33:27:34 +2026-09-21 06:56:38 | INFO | train rank=0 | step=21140/50000 progress=42.3% loss=0.457225 lr=6.657859e-05 lr_backbone=6.657859e-05 elapsed=24:30:02 avg_step=4.17s recent_step=4.14s eta=33:26:52 +2026-09-21 06:57:20 | INFO | train rank=0 | step=21150/50000 progress=42.3% loss=0.325274 lr=6.654739e-05 lr_backbone=6.654739e-05 elapsed=24:30:43 avg_step=4.17s recent_step=4.14s eta=33:26:10 +2026-09-21 06:58:01 | INFO | train rank=0 | step=21160/50000 progress=42.3% loss=0.378236 lr=6.651618e-05 lr_backbone=6.651618e-05 elapsed=24:31:25 avg_step=4.17s recent_step=4.15s eta=33:25:27 +2026-09-21 06:58:42 | INFO | train rank=0 | step=21170/50000 progress=42.3% loss=0.315017 lr=6.648497e-05 lr_backbone=6.648497e-05 elapsed=24:32:06 avg_step=4.17s recent_step=4.11s eta=33:24:45 +2026-09-21 06:59:24 | INFO | train rank=0 | step=21180/50000 progress=42.4% loss=0.355374 lr=6.645374e-05 lr_backbone=6.645374e-05 elapsed=24:32:47 avg_step=4.17s recent_step=4.12s eta=33:24:03 +2026-09-21 07:00:05 | INFO | train rank=0 | step=21190/50000 progress=42.4% loss=0.315582 lr=6.642251e-05 lr_backbone=6.642251e-05 elapsed=24:33:28 avg_step=4.17s recent_step=4.16s eta=33:23:21 +2026-09-21 07:00:46 | INFO | train rank=0 | step=21200/50000 progress=42.4% loss=0.389531 lr=6.639127e-05 lr_backbone=6.639127e-05 elapsed=24:34:10 avg_step=4.17s recent_step=4.12s eta=33:22:38 +2026-09-21 07:01:28 | INFO | train rank=0 | step=21210/50000 progress=42.4% loss=0.375824 lr=6.636003e-05 lr_backbone=6.636003e-05 elapsed=24:34:51 avg_step=4.17s recent_step=4.15s eta=33:21:56 +2026-09-21 07:02:09 | INFO | train rank=0 | step=21220/50000 progress=42.4% loss=0.312148 lr=6.632877e-05 lr_backbone=6.632877e-05 elapsed=24:35:33 avg_step=4.17s recent_step=4.14s eta=33:21:14 +2026-09-21 07:02:50 | INFO | train rank=0 | step=21230/50000 progress=42.5% loss=0.330568 lr=6.629751e-05 lr_backbone=6.629751e-05 elapsed=24:36:14 avg_step=4.17s recent_step=4.10s eta=33:20:31 +2026-09-21 07:03:32 | INFO | train rank=0 | step=21240/50000 progress=42.5% loss=0.398642 lr=6.626625e-05 lr_backbone=6.626625e-05 elapsed=24:36:55 avg_step=4.17s recent_step=4.13s eta=33:19:49 +2026-09-21 07:04:13 | INFO | train rank=0 | step=21250/50000 progress=42.5% loss=0.362403 lr=6.623497e-05 lr_backbone=6.623497e-05 elapsed=24:37:36 avg_step=4.17s recent_step=4.16s eta=33:19:07 +2026-09-21 07:04:55 | INFO | train rank=0 | step=21260/50000 progress=42.5% loss=0.330812 lr=6.620369e-05 lr_backbone=6.620369e-05 elapsed=24:38:18 avg_step=4.17s recent_step=4.15s eta=33:18:25 +2026-09-21 07:05:36 | INFO | train rank=0 | step=21270/50000 progress=42.5% loss=0.341015 lr=6.617240e-05 lr_backbone=6.617240e-05 elapsed=24:38:59 avg_step=4.17s recent_step=4.13s eta=33:17:43 +2026-09-21 07:06:17 | INFO | train rank=0 | step=21280/50000 progress=42.6% loss=0.367928 lr=6.614111e-05 lr_backbone=6.614111e-05 elapsed=24:39:41 avg_step=4.17s recent_step=4.15s eta=33:17:01 +2026-09-21 07:06:58 | INFO | train rank=0 | step=21290/50000 progress=42.6% loss=0.381967 lr=6.610981e-05 lr_backbone=6.610981e-05 elapsed=24:40:22 avg_step=4.17s recent_step=4.09s eta=33:16:18 +2026-09-21 07:07:40 | INFO | train rank=0 | step=21300/50000 progress=42.6% loss=0.338447 lr=6.607850e-05 lr_backbone=6.607850e-05 elapsed=24:41:03 avg_step=4.17s recent_step=4.14s eta=33:15:36 +2026-09-21 07:08:21 | INFO | train rank=0 | step=21310/50000 progress=42.6% loss=0.386671 lr=6.604718e-05 lr_backbone=6.604718e-05 elapsed=24:41:45 avg_step=4.17s recent_step=4.16s eta=33:14:54 +2026-09-21 07:09:03 | INFO | train rank=0 | step=21320/50000 progress=42.6% loss=0.417138 lr=6.601586e-05 lr_backbone=6.601586e-05 elapsed=24:42:26 avg_step=4.17s recent_step=4.15s eta=33:14:12 +2026-09-21 07:09:44 | INFO | train rank=0 | step=21330/50000 progress=42.7% loss=0.321632 lr=6.598453e-05 lr_backbone=6.598453e-05 elapsed=24:43:07 avg_step=4.17s recent_step=4.13s eta=33:13:30 +2026-09-21 07:10:26 | INFO | train rank=0 | step=21340/50000 progress=42.7% loss=0.338204 lr=6.595319e-05 lr_backbone=6.595319e-05 elapsed=24:43:49 avg_step=4.17s recent_step=4.16s eta=33:12:48 +2026-09-21 07:11:07 | INFO | train rank=0 | step=21350/50000 progress=42.7% loss=0.357358 lr=6.592184e-05 lr_backbone=6.592184e-05 elapsed=24:44:30 avg_step=4.17s recent_step=4.10s eta=33:12:05 +2026-09-21 07:11:48 | INFO | train rank=0 | step=21360/50000 progress=42.7% loss=0.397126 lr=6.589049e-05 lr_backbone=6.589049e-05 elapsed=24:45:11 avg_step=4.17s recent_step=4.12s eta=33:11:23 +2026-09-21 07:12:29 | INFO | train rank=0 | step=21370/50000 progress=42.7% loss=0.321400 lr=6.585913e-05 lr_backbone=6.585913e-05 elapsed=24:45:53 avg_step=4.17s recent_step=4.15s eta=33:10:41 +2026-09-21 07:13:11 | INFO | train rank=0 | step=21380/50000 progress=42.8% loss=0.372904 lr=6.582777e-05 lr_backbone=6.582777e-05 elapsed=24:46:34 avg_step=4.17s recent_step=4.14s eta=33:09:59 +2026-09-21 07:13:52 | INFO | train rank=0 | step=21390/50000 progress=42.8% loss=0.496798 lr=6.579640e-05 lr_backbone=6.579640e-05 elapsed=24:47:15 avg_step=4.17s recent_step=4.12s eta=33:09:16 +2026-09-21 07:14:34 | INFO | train rank=0 | step=21400/50000 progress=42.8% loss=0.349709 lr=6.576502e-05 lr_backbone=6.576502e-05 elapsed=24:47:57 avg_step=4.17s recent_step=4.16s eta=33:08:34 +2026-09-21 07:15:15 | INFO | train rank=0 | step=21410/50000 progress=42.8% loss=0.409082 lr=6.573363e-05 lr_backbone=6.573363e-05 elapsed=24:48:38 avg_step=4.17s recent_step=4.11s eta=33:07:52 +2026-09-21 07:15:56 | INFO | train rank=0 | step=21420/50000 progress=42.8% loss=0.367825 lr=6.570224e-05 lr_backbone=6.570224e-05 elapsed=24:49:19 avg_step=4.17s recent_step=4.12s eta=33:07:09 +2026-09-21 07:16:38 | INFO | train rank=0 | step=21430/50000 progress=42.9% loss=0.297922 lr=6.567084e-05 lr_backbone=6.567084e-05 elapsed=24:50:01 avg_step=4.17s recent_step=4.15s eta=33:06:27 +2026-09-21 07:17:19 | INFO | train rank=0 | step=21440/50000 progress=42.9% loss=0.370155 lr=6.563943e-05 lr_backbone=6.563943e-05 elapsed=24:50:43 avg_step=4.17s recent_step=4.17s eta=33:05:46 +2026-09-21 07:18:00 | INFO | train rank=0 | step=21450/50000 progress=42.9% loss=0.354669 lr=6.560802e-05 lr_backbone=6.560802e-05 elapsed=24:51:24 avg_step=4.17s recent_step=4.12s eta=33:05:03 +2026-09-21 07:18:42 | INFO | train rank=0 | step=21460/50000 progress=42.9% loss=0.391281 lr=6.557660e-05 lr_backbone=6.557660e-05 elapsed=24:52:05 avg_step=4.17s recent_step=4.16s eta=33:04:21 +2026-09-21 07:19:23 | INFO | train rank=0 | step=21470/50000 progress=42.9% loss=0.365701 lr=6.554517e-05 lr_backbone=6.554517e-05 elapsed=24:52:46 avg_step=4.17s recent_step=4.10s eta=33:03:39 +2026-09-21 07:20:04 | INFO | train rank=0 | step=21480/50000 progress=43.0% loss=0.304061 lr=6.551374e-05 lr_backbone=6.551374e-05 elapsed=24:53:28 avg_step=4.17s recent_step=4.15s eta=33:02:57 +2026-09-21 07:20:46 | INFO | train rank=0 | step=21490/50000 progress=43.0% loss=0.327974 lr=6.548230e-05 lr_backbone=6.548230e-05 elapsed=24:54:10 avg_step=4.17s recent_step=4.18s eta=33:02:15 +2026-09-21 07:21:28 | INFO | train rank=0 | step=21500/50000 progress=43.0% loss=0.383248 lr=6.545085e-05 lr_backbone=6.545085e-05 elapsed=24:54:51 avg_step=4.17s recent_step=4.16s eta=33:01:33 +2026-09-21 07:22:09 | INFO | train rank=0 | step=21510/50000 progress=43.0% loss=0.294980 lr=6.541940e-05 lr_backbone=6.541940e-05 elapsed=24:55:33 avg_step=4.17s recent_step=4.15s eta=33:00:51 +2026-09-21 07:22:51 | INFO | train rank=0 | step=21520/50000 progress=43.0% loss=0.373668 lr=6.538793e-05 lr_backbone=6.538793e-05 elapsed=24:56:14 avg_step=4.17s recent_step=4.14s eta=33:00:09 +2026-09-21 07:23:32 | INFO | train rank=0 | step=21530/50000 progress=43.1% loss=0.321865 lr=6.535647e-05 lr_backbone=6.535647e-05 elapsed=24:56:55 avg_step=4.17s recent_step=4.11s eta=32:59:27 +2026-09-21 07:24:13 | INFO | train rank=0 | step=21540/50000 progress=43.1% loss=0.326383 lr=6.532499e-05 lr_backbone=6.532499e-05 elapsed=24:57:37 avg_step=4.17s recent_step=4.13s eta=32:58:44 +2026-09-21 07:24:55 | INFO | train rank=0 | step=21550/50000 progress=43.1% loss=0.339834 lr=6.529351e-05 lr_backbone=6.529351e-05 elapsed=24:58:18 avg_step=4.17s recent_step=4.14s eta=32:58:02 +2026-09-21 07:25:36 | INFO | train rank=0 | step=21560/50000 progress=43.1% loss=0.311814 lr=6.526202e-05 lr_backbone=6.526202e-05 elapsed=24:59:00 avg_step=4.17s recent_step=4.17s eta=32:57:20 +2026-09-21 07:26:18 | INFO | train rank=0 | step=21570/50000 progress=43.1% loss=0.338210 lr=6.523053e-05 lr_backbone=6.523053e-05 elapsed=24:59:41 avg_step=4.17s recent_step=4.13s eta=32:56:38 +2026-09-21 07:26:59 | INFO | train rank=0 | step=21580/50000 progress=43.2% loss=0.343808 lr=6.519903e-05 lr_backbone=6.519903e-05 elapsed=25:00:22 avg_step=4.17s recent_step=4.15s eta=32:55:56 +2026-09-21 07:27:40 | INFO | train rank=0 | step=21590/50000 progress=43.2% loss=0.399082 lr=6.516752e-05 lr_backbone=6.516752e-05 elapsed=25:01:03 avg_step=4.17s recent_step=4.10s eta=32:55:13 +2026-09-21 07:28:21 | INFO | train rank=0 | step=21600/50000 progress=43.2% loss=0.356144 lr=6.513601e-05 lr_backbone=6.513601e-05 elapsed=25:01:45 avg_step=4.17s recent_step=4.13s eta=32:54:31 +2026-09-21 07:29:03 | INFO | train rank=0 | step=21610/50000 progress=43.2% loss=0.356661 lr=6.510448e-05 lr_backbone=6.510448e-05 elapsed=25:02:26 avg_step=4.17s recent_step=4.13s eta=32:53:49 +2026-09-21 07:29:44 | INFO | train rank=0 | step=21620/50000 progress=43.2% loss=0.358808 lr=6.507296e-05 lr_backbone=6.507296e-05 elapsed=25:03:08 avg_step=4.17s recent_step=4.17s eta=32:53:07 +2026-09-21 07:30:26 | INFO | train rank=0 | step=21630/50000 progress=43.3% loss=0.356527 lr=6.504142e-05 lr_backbone=6.504142e-05 elapsed=25:03:49 avg_step=4.17s recent_step=4.15s eta=32:52:25 +2026-09-21 07:31:07 | INFO | train rank=0 | step=21640/50000 progress=43.3% loss=0.319424 lr=6.500988e-05 lr_backbone=6.500988e-05 elapsed=25:04:31 avg_step=4.17s recent_step=4.14s eta=32:51:43 +2026-09-21 07:31:48 | INFO | train rank=0 | step=21650/50000 progress=43.3% loss=0.346846 lr=6.497833e-05 lr_backbone=6.497833e-05 elapsed=25:05:12 avg_step=4.17s recent_step=4.11s eta=32:51:00 +2026-09-21 07:32:30 | INFO | train rank=0 | step=21660/50000 progress=43.3% loss=0.413977 lr=6.494678e-05 lr_backbone=6.494678e-05 elapsed=25:05:53 avg_step=4.17s recent_step=4.15s eta=32:50:18 +2026-09-21 07:33:11 | INFO | train rank=0 | step=21670/50000 progress=43.3% loss=0.323092 lr=6.491522e-05 lr_backbone=6.491522e-05 elapsed=25:06:35 avg_step=4.17s recent_step=4.15s eta=32:49:36 +2026-09-21 07:33:53 | INFO | train rank=0 | step=21680/50000 progress=43.4% loss=0.461820 lr=6.488365e-05 lr_backbone=6.488365e-05 elapsed=25:07:16 avg_step=4.17s recent_step=4.16s eta=32:48:55 +2026-09-21 07:34:34 | INFO | train rank=0 | step=21690/50000 progress=43.4% loss=0.316047 lr=6.485208e-05 lr_backbone=6.485208e-05 elapsed=25:07:58 avg_step=4.17s recent_step=4.13s eta=32:48:12 +2026-09-21 07:35:16 | INFO | train rank=0 | step=21700/50000 progress=43.4% loss=0.360287 lr=6.482050e-05 lr_backbone=6.482050e-05 elapsed=25:08:39 avg_step=4.17s recent_step=4.14s eta=32:47:30 +2026-09-21 07:35:57 | INFO | train rank=0 | step=21710/50000 progress=43.4% loss=0.339515 lr=6.478891e-05 lr_backbone=6.478891e-05 elapsed=25:09:20 avg_step=4.17s recent_step=4.10s eta=32:46:48 +2026-09-21 07:36:38 | INFO | train rank=0 | step=21720/50000 progress=43.4% loss=0.433523 lr=6.475732e-05 lr_backbone=6.475732e-05 elapsed=25:10:01 avg_step=4.17s recent_step=4.12s eta=32:46:05 +2026-09-21 07:37:19 | INFO | train rank=0 | step=21730/50000 progress=43.5% loss=0.381812 lr=6.472572e-05 lr_backbone=6.472572e-05 elapsed=25:10:43 avg_step=4.17s recent_step=4.15s eta=32:45:23 +2026-09-21 07:38:01 | INFO | train rank=0 | step=21740/50000 progress=43.5% loss=0.382089 lr=6.469411e-05 lr_backbone=6.469411e-05 elapsed=25:11:24 avg_step=4.17s recent_step=4.14s eta=32:44:41 +2026-09-21 07:38:42 | INFO | train rank=0 | step=21750/50000 progress=43.5% loss=0.404325 lr=6.466250e-05 lr_backbone=6.466250e-05 elapsed=25:12:06 avg_step=4.17s recent_step=4.17s eta=32:43:59 +2026-09-21 07:39:24 | INFO | train rank=0 | step=21760/50000 progress=43.5% loss=0.366917 lr=6.463088e-05 lr_backbone=6.463088e-05 elapsed=25:12:47 avg_step=4.17s recent_step=4.14s eta=32:43:17 +2026-09-21 07:40:05 | INFO | train rank=0 | step=21770/50000 progress=43.5% loss=0.400495 lr=6.459926e-05 lr_backbone=6.459926e-05 elapsed=25:13:28 avg_step=4.17s recent_step=4.10s eta=32:42:35 +2026-09-21 07:40:46 | INFO | train rank=0 | step=21780/50000 progress=43.6% loss=0.409688 lr=6.456763e-05 lr_backbone=6.456763e-05 elapsed=25:14:09 avg_step=4.17s recent_step=4.13s eta=32:41:52 +2026-09-21 07:41:28 | INFO | train rank=0 | step=21790/50000 progress=43.6% loss=0.360351 lr=6.453599e-05 lr_backbone=6.453599e-05 elapsed=25:14:51 avg_step=4.17s recent_step=4.16s eta=32:41:10 +2026-09-21 07:42:09 | INFO | train rank=0 | step=21800/50000 progress=43.6% loss=0.413426 lr=6.450434e-05 lr_backbone=6.450434e-05 elapsed=25:15:33 avg_step=4.17s recent_step=4.16s eta=32:40:29 +2026-09-21 07:42:51 | INFO | train rank=0 | step=21810/50000 progress=43.6% loss=0.393876 lr=6.447269e-05 lr_backbone=6.447269e-05 elapsed=25:16:14 avg_step=4.17s recent_step=4.15s eta=32:39:47 +2026-09-21 07:43:32 | INFO | train rank=0 | step=21820/50000 progress=43.6% loss=0.378762 lr=6.444104e-05 lr_backbone=6.444104e-05 elapsed=25:16:55 avg_step=4.17s recent_step=4.12s eta=32:39:04 +2026-09-21 07:44:13 | INFO | train rank=0 | step=21830/50000 progress=43.7% loss=0.356583 lr=6.440937e-05 lr_backbone=6.440937e-05 elapsed=25:17:36 avg_step=4.17s recent_step=4.11s eta=32:38:22 +2026-09-21 07:44:54 | INFO | train rank=0 | step=21840/50000 progress=43.7% loss=0.334006 lr=6.437770e-05 lr_backbone=6.437770e-05 elapsed=25:18:18 avg_step=4.17s recent_step=4.13s eta=32:37:39 +2026-09-21 07:45:36 | INFO | train rank=0 | step=21850/50000 progress=43.7% loss=0.319363 lr=6.434603e-05 lr_backbone=6.434603e-05 elapsed=25:18:59 avg_step=4.17s recent_step=4.14s eta=32:36:57 +2026-09-21 07:46:17 | INFO | train rank=0 | step=21860/50000 progress=43.7% loss=0.350180 lr=6.431435e-05 lr_backbone=6.431435e-05 elapsed=25:19:40 avg_step=4.17s recent_step=4.14s eta=32:36:15 +2026-09-21 07:46:59 | INFO | train rank=0 | step=21870/50000 progress=43.7% loss=0.357998 lr=6.428266e-05 lr_backbone=6.428266e-05 elapsed=25:20:22 avg_step=4.17s recent_step=4.15s eta=32:35:33 +2026-09-21 07:47:40 | INFO | train rank=0 | step=21880/50000 progress=43.8% loss=0.338586 lr=6.425096e-05 lr_backbone=6.425096e-05 elapsed=25:21:03 avg_step=4.17s recent_step=4.12s eta=32:34:51 +2026-09-21 07:48:21 | INFO | train rank=0 | step=21890/50000 progress=43.8% loss=0.338323 lr=6.421926e-05 lr_backbone=6.421926e-05 elapsed=25:21:44 avg_step=4.17s recent_step=4.11s eta=32:34:08 +2026-09-21 07:49:02 | INFO | train rank=0 | step=21900/50000 progress=43.8% loss=0.371127 lr=6.418756e-05 lr_backbone=6.418756e-05 elapsed=25:22:25 avg_step=4.17s recent_step=4.11s eta=32:33:26 +2026-09-21 07:49:44 | INFO | train rank=0 | step=21910/50000 progress=43.8% loss=0.339538 lr=6.415584e-05 lr_backbone=6.415584e-05 elapsed=25:23:07 avg_step=4.17s recent_step=4.16s eta=32:32:44 +2026-09-21 07:50:25 | INFO | train rank=0 | step=21920/50000 progress=43.8% loss=0.375656 lr=6.412412e-05 lr_backbone=6.412412e-05 elapsed=25:23:48 avg_step=4.17s recent_step=4.12s eta=32:32:02 +2026-09-21 07:51:06 | INFO | train rank=0 | step=21930/50000 progress=43.9% loss=0.305339 lr=6.409240e-05 lr_backbone=6.409240e-05 elapsed=25:24:30 avg_step=4.17s recent_step=4.14s eta=32:31:20 +2026-09-21 07:51:48 | INFO | train rank=0 | step=21940/50000 progress=43.9% loss=0.377204 lr=6.406066e-05 lr_backbone=6.406066e-05 elapsed=25:25:11 avg_step=4.17s recent_step=4.14s eta=32:30:37 +2026-09-21 07:52:28 | INFO | train rank=0 | step=21950/50000 progress=43.9% loss=0.432198 lr=6.402893e-05 lr_backbone=6.402893e-05 elapsed=25:25:52 avg_step=4.17s recent_step=4.07s eta=32:29:54 +2026-09-21 07:53:09 | INFO | train rank=0 | step=21960/50000 progress=43.9% loss=0.315602 lr=6.399718e-05 lr_backbone=6.399718e-05 elapsed=25:26:33 avg_step=4.17s recent_step=4.09s eta=32:29:12 +2026-09-21 07:53:50 | INFO | train rank=0 | step=21970/50000 progress=43.9% loss=0.346460 lr=6.396543e-05 lr_backbone=6.396543e-05 elapsed=25:27:14 avg_step=4.17s recent_step=4.12s eta=32:28:29 +2026-09-21 07:54:32 | INFO | train rank=0 | step=21980/50000 progress=44.0% loss=0.341328 lr=6.393368e-05 lr_backbone=6.393368e-05 elapsed=25:27:55 avg_step=4.17s recent_step=4.15s eta=32:27:47 +2026-09-21 07:55:13 | INFO | train rank=0 | step=21990/50000 progress=44.0% loss=0.370047 lr=6.390191e-05 lr_backbone=6.390191e-05 elapsed=25:28:37 avg_step=4.17s recent_step=4.14s eta=32:27:05 +2026-09-21 07:55:55 | INFO | train rank=0 | step=22000/50000 progress=44.0% loss=0.432023 lr=6.387015e-05 lr_backbone=6.387015e-05 elapsed=25:29:18 avg_step=4.17s recent_step=4.14s eta=32:26:23 +2026-09-21 07:56:35 | INFO | train rank=0 | step=22010/50000 progress=44.0% loss=0.328081 lr=6.383837e-05 lr_backbone=6.383837e-05 elapsed=25:29:59 avg_step=4.17s recent_step=4.07s eta=32:25:40 +2026-09-21 07:57:16 | INFO | train rank=0 | step=22020/50000 progress=44.0% loss=0.328831 lr=6.380659e-05 lr_backbone=6.380659e-05 elapsed=25:30:40 avg_step=4.17s recent_step=4.11s eta=32:24:58 +2026-09-21 07:57:58 | INFO | train rank=0 | step=22030/50000 progress=44.1% loss=0.391053 lr=6.377480e-05 lr_backbone=6.377480e-05 elapsed=25:31:21 avg_step=4.17s recent_step=4.13s eta=32:24:15 +2026-09-21 07:58:39 | INFO | train rank=0 | step=22040/50000 progress=44.1% loss=0.434471 lr=6.374301e-05 lr_backbone=6.374301e-05 elapsed=25:32:02 avg_step=4.17s recent_step=4.11s eta=32:23:33 +2026-09-21 07:59:20 | INFO | train rank=0 | step=22050/50000 progress=44.1% loss=0.362033 lr=6.371121e-05 lr_backbone=6.371121e-05 elapsed=25:32:44 avg_step=4.17s recent_step=4.12s eta=32:22:51 +2026-09-21 08:00:01 | INFO | train rank=0 | step=22060/50000 progress=44.1% loss=0.370668 lr=6.367941e-05 lr_backbone=6.367941e-05 elapsed=25:33:25 avg_step=4.17s recent_step=4.13s eta=32:22:08 +2026-09-21 08:00:42 | INFO | train rank=0 | step=22070/50000 progress=44.1% loss=0.394807 lr=6.364760e-05 lr_backbone=6.364760e-05 elapsed=25:34:06 avg_step=4.17s recent_step=4.08s eta=32:21:26 +2026-09-21 08:01:23 | INFO | train rank=0 | step=22080/50000 progress=44.2% loss=0.384439 lr=6.361578e-05 lr_backbone=6.361578e-05 elapsed=25:34:47 avg_step=4.17s recent_step=4.11s eta=32:20:43 +2026-09-21 08:02:05 | INFO | train rank=0 | step=22090/50000 progress=44.2% loss=0.369923 lr=6.358396e-05 lr_backbone=6.358396e-05 elapsed=25:35:28 avg_step=4.17s recent_step=4.17s eta=32:20:01 +2026-09-21 08:02:47 | INFO | train rank=0 | step=22100/50000 progress=44.2% loss=0.352503 lr=6.355213e-05 lr_backbone=6.355213e-05 elapsed=25:36:10 avg_step=4.17s recent_step=4.15s eta=32:19:19 +2026-09-21 08:03:28 | INFO | train rank=0 | step=22110/50000 progress=44.2% loss=0.362271 lr=6.352029e-05 lr_backbone=6.352029e-05 elapsed=25:36:51 avg_step=4.17s recent_step=4.13s eta=32:18:37 +2026-09-21 08:04:10 | INFO | train rank=0 | step=22120/50000 progress=44.2% loss=0.430459 lr=6.348845e-05 lr_backbone=6.348845e-05 elapsed=25:37:33 avg_step=4.17s recent_step=4.16s eta=32:17:55 +2026-09-21 08:04:51 | INFO | train rank=0 | step=22130/50000 progress=44.3% loss=0.351705 lr=6.345661e-05 lr_backbone=6.345661e-05 elapsed=25:38:14 avg_step=4.17s recent_step=4.10s eta=32:17:13 +2026-09-21 08:05:32 | INFO | train rank=0 | step=22140/50000 progress=44.3% loss=0.296883 lr=6.342476e-05 lr_backbone=6.342476e-05 elapsed=25:38:55 avg_step=4.17s recent_step=4.12s eta=32:16:30 +2026-09-21 08:06:13 | INFO | train rank=0 | step=22150/50000 progress=44.3% loss=0.348182 lr=6.339290e-05 lr_backbone=6.339290e-05 elapsed=25:39:36 avg_step=4.17s recent_step=4.12s eta=32:15:48 +2026-09-21 08:06:55 | INFO | train rank=0 | step=22160/50000 progress=44.3% loss=0.354381 lr=6.336103e-05 lr_backbone=6.336103e-05 elapsed=25:40:18 avg_step=4.17s recent_step=4.17s eta=32:15:06 +2026-09-21 08:07:36 | INFO | train rank=0 | step=22170/50000 progress=44.3% loss=0.365474 lr=6.332916e-05 lr_backbone=6.332916e-05 elapsed=25:40:59 avg_step=4.17s recent_step=4.15s eta=32:14:24 +2026-09-21 08:08:18 | INFO | train rank=0 | step=22180/50000 progress=44.4% loss=0.331727 lr=6.329729e-05 lr_backbone=6.329729e-05 elapsed=25:41:41 avg_step=4.17s recent_step=4.15s eta=32:13:43 +2026-09-21 08:08:59 | INFO | train rank=0 | step=22190/50000 progress=44.4% loss=0.377269 lr=6.326541e-05 lr_backbone=6.326541e-05 elapsed=25:42:22 avg_step=4.17s recent_step=4.10s eta=32:13:00 +2026-09-21 08:09:40 | INFO | train rank=0 | step=22200/50000 progress=44.4% loss=0.398160 lr=6.323352e-05 lr_backbone=6.323352e-05 elapsed=25:43:03 avg_step=4.17s recent_step=4.14s eta=32:12:18 +2026-09-21 08:10:21 | INFO | train rank=0 | step=22210/50000 progress=44.4% loss=0.405812 lr=6.320163e-05 lr_backbone=6.320163e-05 elapsed=25:43:45 avg_step=4.17s recent_step=4.13s eta=32:11:36 +2026-09-21 08:11:03 | INFO | train rank=0 | step=22220/50000 progress=44.4% loss=0.433981 lr=6.316973e-05 lr_backbone=6.316973e-05 elapsed=25:44:26 avg_step=4.17s recent_step=4.15s eta=32:10:54 +2026-09-21 08:11:45 | INFO | train rank=0 | step=22230/50000 progress=44.5% loss=0.501394 lr=6.313782e-05 lr_backbone=6.313782e-05 elapsed=25:45:08 avg_step=4.17s recent_step=4.17s eta=32:10:12 +2026-09-21 08:12:26 | INFO | train rank=0 | step=22240/50000 progress=44.5% loss=0.329445 lr=6.310591e-05 lr_backbone=6.310591e-05 elapsed=25:45:49 avg_step=4.17s recent_step=4.15s eta=32:09:30 +2026-09-21 08:13:07 | INFO | train rank=0 | step=22250/50000 progress=44.5% loss=0.342281 lr=6.307400e-05 lr_backbone=6.307400e-05 elapsed=25:46:30 avg_step=4.17s recent_step=4.11s eta=32:08:48 +2026-09-21 08:13:48 | INFO | train rank=0 | step=22260/50000 progress=44.5% loss=0.372552 lr=6.304208e-05 lr_backbone=6.304208e-05 elapsed=25:47:12 avg_step=4.17s recent_step=4.13s eta=32:08:05 +2026-09-21 08:14:30 | INFO | train rank=0 | step=22270/50000 progress=44.5% loss=0.330560 lr=6.301015e-05 lr_backbone=6.301015e-05 elapsed=25:47:53 avg_step=4.17s recent_step=4.15s eta=32:07:23 +2026-09-21 08:15:11 | INFO | train rank=0 | step=22280/50000 progress=44.6% loss=0.324835 lr=6.297821e-05 lr_backbone=6.297821e-05 elapsed=25:48:35 avg_step=4.17s recent_step=4.15s eta=32:06:41 +2026-09-21 08:15:53 | INFO | train rank=0 | step=22290/50000 progress=44.6% loss=0.377160 lr=6.294628e-05 lr_backbone=6.294628e-05 elapsed=25:49:16 avg_step=4.17s recent_step=4.15s eta=32:05:59 +2026-09-21 08:16:35 | INFO | train rank=0 | step=22300/50000 progress=44.6% loss=0.355988 lr=6.291433e-05 lr_backbone=6.291433e-05 elapsed=25:49:58 avg_step=4.17s recent_step=4.17s eta=32:05:18 +2026-09-21 08:17:16 | INFO | train rank=0 | step=22310/50000 progress=44.6% loss=0.401652 lr=6.288238e-05 lr_backbone=6.288238e-05 elapsed=25:50:39 avg_step=4.17s recent_step=4.11s eta=32:04:35 +2026-09-21 08:17:57 | INFO | train rank=0 | step=22320/50000 progress=44.6% loss=0.342249 lr=6.285043e-05 lr_backbone=6.285043e-05 elapsed=25:51:20 avg_step=4.17s recent_step=4.12s eta=32:03:53 +2026-09-21 08:18:38 | INFO | train rank=0 | step=22330/50000 progress=44.7% loss=0.289700 lr=6.281846e-05 lr_backbone=6.281846e-05 elapsed=25:52:02 avg_step=4.17s recent_step=4.15s eta=32:03:11 +2026-09-21 08:19:19 | INFO | train rank=0 | step=22340/50000 progress=44.7% loss=0.370197 lr=6.278650e-05 lr_backbone=6.278650e-05 elapsed=25:52:43 avg_step=4.17s recent_step=4.11s eta=32:02:29 +2026-09-21 08:20:01 | INFO | train rank=0 | step=22350/50000 progress=44.7% loss=0.343160 lr=6.275452e-05 lr_backbone=6.275452e-05 elapsed=25:53:24 avg_step=4.17s recent_step=4.15s eta=32:01:47 +2026-09-21 08:20:42 | INFO | train rank=0 | step=22360/50000 progress=44.7% loss=0.434175 lr=6.272255e-05 lr_backbone=6.272255e-05 elapsed=25:54:06 avg_step=4.17s recent_step=4.15s eta=32:01:05 +2026-09-21 08:21:23 | INFO | train rank=0 | step=22370/50000 progress=44.7% loss=0.346612 lr=6.269056e-05 lr_backbone=6.269056e-05 elapsed=25:54:47 avg_step=4.17s recent_step=4.09s eta=32:00:22 +2026-09-21 08:22:05 | INFO | train rank=0 | step=22380/50000 progress=44.8% loss=0.418124 lr=6.265857e-05 lr_backbone=6.265857e-05 elapsed=25:55:28 avg_step=4.17s recent_step=4.13s eta=31:59:40 +2026-09-21 08:22:46 | INFO | train rank=0 | step=22390/50000 progress=44.8% loss=0.322003 lr=6.262658e-05 lr_backbone=6.262658e-05 elapsed=25:56:09 avg_step=4.17s recent_step=4.14s eta=31:58:58 +2026-09-21 08:23:27 | INFO | train rank=0 | step=22400/50000 progress=44.8% loss=0.298112 lr=6.259458e-05 lr_backbone=6.259458e-05 elapsed=25:56:50 avg_step=4.17s recent_step=4.12s eta=31:58:15 +2026-09-21 08:24:09 | INFO | train rank=0 | step=22410/50000 progress=44.8% loss=0.387865 lr=6.256257e-05 lr_backbone=6.256257e-05 elapsed=25:57:32 avg_step=4.17s recent_step=4.15s eta=31:57:33 +2026-09-21 08:24:50 | INFO | train rank=0 | step=22420/50000 progress=44.8% loss=0.409826 lr=6.253056e-05 lr_backbone=6.253056e-05 elapsed=25:58:13 avg_step=4.17s recent_step=4.15s eta=31:56:51 +2026-09-21 08:25:31 | INFO | train rank=0 | step=22430/50000 progress=44.9% loss=0.395577 lr=6.249854e-05 lr_backbone=6.249854e-05 elapsed=25:58:55 avg_step=4.17s recent_step=4.11s eta=31:56:09 +2026-09-21 08:26:12 | INFO | train rank=0 | step=22440/50000 progress=44.9% loss=0.516782 lr=6.246652e-05 lr_backbone=6.246652e-05 elapsed=25:59:36 avg_step=4.17s recent_step=4.12s eta=31:55:27 +2026-09-21 08:26:54 | INFO | train rank=0 | step=22450/50000 progress=44.9% loss=0.408451 lr=6.243449e-05 lr_backbone=6.243449e-05 elapsed=26:00:17 avg_step=4.17s recent_step=4.15s eta=31:54:45 +2026-09-21 08:27:36 | INFO | train rank=0 | step=22460/50000 progress=44.9% loss=0.312604 lr=6.240246e-05 lr_backbone=6.240246e-05 elapsed=26:00:59 avg_step=4.17s recent_step=4.16s eta=31:54:03 +2026-09-21 08:28:17 | INFO | train rank=0 | step=22470/50000 progress=44.9% loss=0.361681 lr=6.237042e-05 lr_backbone=6.237042e-05 elapsed=26:01:40 avg_step=4.17s recent_step=4.14s eta=31:53:21 +2026-09-21 08:28:58 | INFO | train rank=0 | step=22480/50000 progress=45.0% loss=0.457138 lr=6.233838e-05 lr_backbone=6.233838e-05 elapsed=26:02:22 avg_step=4.17s recent_step=4.14s eta=31:52:39 +2026-09-21 08:29:39 | INFO | train rank=0 | step=22490/50000 progress=45.0% loss=0.382705 lr=6.230633e-05 lr_backbone=6.230633e-05 elapsed=26:03:03 avg_step=4.17s recent_step=4.10s eta=31:51:56 +2026-09-21 08:30:21 | INFO | train rank=0 | step=22500/50000 progress=45.0% loss=0.328735 lr=6.227427e-05 lr_backbone=6.227427e-05 elapsed=26:03:44 avg_step=4.17s recent_step=4.13s eta=31:51:14 +2026-09-21 08:31:02 | INFO | train rank=0 | step=22510/50000 progress=45.0% loss=0.315019 lr=6.224221e-05 lr_backbone=6.224221e-05 elapsed=26:04:26 avg_step=4.17s recent_step=4.15s eta=31:50:32 +2026-09-21 08:31:44 | INFO | train rank=0 | step=22520/50000 progress=45.0% loss=0.379754 lr=6.221015e-05 lr_backbone=6.221015e-05 elapsed=26:05:07 avg_step=4.17s recent_step=4.14s eta=31:49:50 +2026-09-21 08:32:25 | INFO | train rank=0 | step=22530/50000 progress=45.1% loss=0.320108 lr=6.217808e-05 lr_backbone=6.217808e-05 elapsed=26:05:49 avg_step=4.17s recent_step=4.16s eta=31:49:08 +2026-09-21 08:33:07 | INFO | train rank=0 | step=22540/50000 progress=45.1% loss=0.303434 lr=6.214600e-05 lr_backbone=6.214600e-05 elapsed=26:06:30 avg_step=4.17s recent_step=4.15s eta=31:48:26 +2026-09-21 08:33:48 | INFO | train rank=0 | step=22550/50000 progress=45.1% loss=0.311996 lr=6.211392e-05 lr_backbone=6.211392e-05 elapsed=26:07:11 avg_step=4.17s recent_step=4.10s eta=31:47:44 +2026-09-21 08:34:29 | INFO | train rank=0 | step=22560/50000 progress=45.1% loss=0.387589 lr=6.208183e-05 lr_backbone=6.208183e-05 elapsed=26:07:52 avg_step=4.17s recent_step=4.13s eta=31:47:01 +2026-09-21 08:35:11 | INFO | train rank=0 | step=22570/50000 progress=45.1% loss=0.315863 lr=6.204974e-05 lr_backbone=6.204974e-05 elapsed=26:08:34 avg_step=4.17s recent_step=4.15s eta=31:46:20 +2026-09-21 08:35:52 | INFO | train rank=0 | step=22580/50000 progress=45.2% loss=0.418068 lr=6.201764e-05 lr_backbone=6.201764e-05 elapsed=26:09:15 avg_step=4.17s recent_step=4.15s eta=31:45:38 +2026-09-21 08:36:33 | INFO | train rank=0 | step=22590/50000 progress=45.2% loss=0.307750 lr=6.198554e-05 lr_backbone=6.198554e-05 elapsed=26:09:57 avg_step=4.17s recent_step=4.13s eta=31:44:55 +2026-09-21 08:37:15 | INFO | train rank=0 | step=22600/50000 progress=45.2% loss=0.318225 lr=6.195343e-05 lr_backbone=6.195343e-05 elapsed=26:10:38 avg_step=4.17s recent_step=4.14s eta=31:44:13 +2026-09-21 08:37:56 | INFO | train rank=0 | step=22610/50000 progress=45.2% loss=0.399237 lr=6.192132e-05 lr_backbone=6.192132e-05 elapsed=26:11:19 avg_step=4.17s recent_step=4.09s eta=31:43:31 +2026-09-21 08:38:37 | INFO | train rank=0 | step=22620/50000 progress=45.2% loss=0.316496 lr=6.188920e-05 lr_backbone=6.188920e-05 elapsed=26:12:00 avg_step=4.17s recent_step=4.12s eta=31:42:48 +2026-09-21 08:39:19 | INFO | train rank=0 | step=22630/50000 progress=45.3% loss=0.334790 lr=6.185708e-05 lr_backbone=6.185708e-05 elapsed=26:12:42 avg_step=4.17s recent_step=4.16s eta=31:42:07 +2026-09-21 08:40:00 | INFO | train rank=0 | step=22640/50000 progress=45.3% loss=0.340092 lr=6.182495e-05 lr_backbone=6.182495e-05 elapsed=26:13:24 avg_step=4.17s recent_step=4.17s eta=31:41:25 +2026-09-21 08:40:42 | INFO | train rank=0 | step=22650/50000 progress=45.3% loss=0.357956 lr=6.179282e-05 lr_backbone=6.179282e-05 elapsed=26:14:05 avg_step=4.17s recent_step=4.17s eta=31:40:43 +2026-09-21 08:41:23 | INFO | train rank=0 | step=22660/50000 progress=45.3% loss=0.308010 lr=6.176068e-05 lr_backbone=6.176068e-05 elapsed=26:14:47 avg_step=4.17s recent_step=4.14s eta=31:40:01 +2026-09-21 08:42:04 | INFO | train rank=0 | step=22670/50000 progress=45.3% loss=0.341882 lr=6.172853e-05 lr_backbone=6.172853e-05 elapsed=26:15:28 avg_step=4.17s recent_step=4.10s eta=31:39:19 +2026-09-21 08:42:46 | INFO | train rank=0 | step=22680/50000 progress=45.4% loss=0.298007 lr=6.169638e-05 lr_backbone=6.169638e-05 elapsed=26:16:09 avg_step=4.17s recent_step=4.13s eta=31:38:36 +2026-09-21 08:43:27 | INFO | train rank=0 | step=22690/50000 progress=45.4% loss=0.386207 lr=6.166423e-05 lr_backbone=6.166423e-05 elapsed=26:16:50 avg_step=4.17s recent_step=4.14s eta=31:37:54 +2026-09-21 08:44:08 | INFO | train rank=0 | step=22700/50000 progress=45.4% loss=0.297501 lr=6.163207e-05 lr_backbone=6.163207e-05 elapsed=26:17:32 avg_step=4.17s recent_step=4.15s eta=31:37:12 +2026-09-21 08:44:50 | INFO | train rank=0 | step=22710/50000 progress=45.4% loss=0.344956 lr=6.159991e-05 lr_backbone=6.159991e-05 elapsed=26:18:13 avg_step=4.17s recent_step=4.15s eta=31:36:31 +2026-09-21 08:45:31 | INFO | train rank=0 | step=22720/50000 progress=45.4% loss=0.313343 lr=6.156774e-05 lr_backbone=6.156774e-05 elapsed=26:18:55 avg_step=4.17s recent_step=4.13s eta=31:35:48 +2026-09-21 08:46:12 | INFO | train rank=0 | step=22730/50000 progress=45.5% loss=0.342773 lr=6.153556e-05 lr_backbone=6.153556e-05 elapsed=26:19:36 avg_step=4.17s recent_step=4.10s eta=31:35:06 +2026-09-21 08:46:53 | INFO | train rank=0 | step=22740/50000 progress=45.5% loss=0.351945 lr=6.150338e-05 lr_backbone=6.150338e-05 elapsed=26:20:17 avg_step=4.17s recent_step=4.10s eta=31:34:23 +2026-09-21 08:47:35 | INFO | train rank=0 | step=22750/50000 progress=45.5% loss=0.301160 lr=6.147120e-05 lr_backbone=6.147120e-05 elapsed=26:20:58 avg_step=4.17s recent_step=4.15s eta=31:33:41 +2026-09-21 08:48:16 | INFO | train rank=0 | step=22760/50000 progress=45.5% loss=0.340547 lr=6.143901e-05 lr_backbone=6.143901e-05 elapsed=26:21:39 avg_step=4.17s recent_step=4.14s eta=31:32:59 +2026-09-21 08:48:58 | INFO | train rank=0 | step=22770/50000 progress=45.5% loss=0.393750 lr=6.140681e-05 lr_backbone=6.140681e-05 elapsed=26:22:21 avg_step=4.17s recent_step=4.14s eta=31:32:17 +2026-09-21 08:49:39 | INFO | train rank=0 | step=22780/50000 progress=45.6% loss=0.392935 lr=6.137461e-05 lr_backbone=6.137461e-05 elapsed=26:23:02 avg_step=4.17s recent_step=4.14s eta=31:31:35 +2026-09-21 08:50:20 | INFO | train rank=0 | step=22790/50000 progress=45.6% loss=0.312398 lr=6.134241e-05 lr_backbone=6.134241e-05 elapsed=26:23:43 avg_step=4.17s recent_step=4.08s eta=31:30:52 +2026-09-21 08:51:01 | INFO | train rank=0 | step=22800/50000 progress=45.6% loss=0.342466 lr=6.131020e-05 lr_backbone=6.131020e-05 elapsed=26:24:24 avg_step=4.17s recent_step=4.12s eta=31:30:10 +2026-09-21 08:51:42 | INFO | train rank=0 | step=22810/50000 progress=45.6% loss=0.377336 lr=6.127798e-05 lr_backbone=6.127798e-05 elapsed=26:25:06 avg_step=4.17s recent_step=4.14s eta=31:29:28 +2026-09-21 08:52:24 | INFO | train rank=0 | step=22820/50000 progress=45.6% loss=0.329378 lr=6.124576e-05 lr_backbone=6.124576e-05 elapsed=26:25:47 avg_step=4.17s recent_step=4.14s eta=31:28:46 +2026-09-21 08:53:05 | INFO | train rank=0 | step=22830/50000 progress=45.7% loss=0.330906 lr=6.121354e-05 lr_backbone=6.121354e-05 elapsed=26:26:28 avg_step=4.17s recent_step=4.13s eta=31:28:04 +2026-09-21 08:53:46 | INFO | train rank=0 | step=22840/50000 progress=45.7% loss=0.307949 lr=6.118131e-05 lr_backbone=6.118131e-05 elapsed=26:27:10 avg_step=4.17s recent_step=4.15s eta=31:27:22 +2026-09-21 08:54:27 | INFO | train rank=0 | step=22850/50000 progress=45.7% loss=0.372020 lr=6.114907e-05 lr_backbone=6.114907e-05 elapsed=26:27:51 avg_step=4.17s recent_step=4.09s eta=31:26:39 +2026-09-21 08:55:08 | INFO | train rank=0 | step=22860/50000 progress=45.7% loss=0.382527 lr=6.111683e-05 lr_backbone=6.111683e-05 elapsed=26:28:32 avg_step=4.17s recent_step=4.09s eta=31:25:57 +2026-09-21 08:55:50 | INFO | train rank=0 | step=22870/50000 progress=45.7% loss=0.386389 lr=6.108459e-05 lr_backbone=6.108459e-05 elapsed=26:29:13 avg_step=4.17s recent_step=4.13s eta=31:25:14 +2026-09-21 08:56:31 | INFO | train rank=0 | step=22880/50000 progress=45.8% loss=0.445719 lr=6.105234e-05 lr_backbone=6.105234e-05 elapsed=26:29:54 avg_step=4.17s recent_step=4.13s eta=31:24:32 +2026-09-21 08:57:12 | INFO | train rank=0 | step=22890/50000 progress=45.8% loss=0.328413 lr=6.102009e-05 lr_backbone=6.102009e-05 elapsed=26:30:36 avg_step=4.17s recent_step=4.14s eta=31:23:50 +2026-09-21 08:57:54 | INFO | train rank=0 | step=22900/50000 progress=45.8% loss=0.397206 lr=6.098783e-05 lr_backbone=6.098783e-05 elapsed=26:31:17 avg_step=4.17s recent_step=4.14s eta=31:23:08 +2026-09-21 08:58:35 | INFO | train rank=0 | step=22910/50000 progress=45.8% loss=0.366394 lr=6.095557e-05 lr_backbone=6.095557e-05 elapsed=26:31:58 avg_step=4.17s recent_step=4.09s eta=31:22:26 +2026-09-21 08:59:16 | INFO | train rank=0 | step=22920/50000 progress=45.8% loss=0.343034 lr=6.092330e-05 lr_backbone=6.092330e-05 elapsed=26:32:39 avg_step=4.17s recent_step=4.10s eta=31:21:43 +2026-09-21 08:59:57 | INFO | train rank=0 | step=22930/50000 progress=45.9% loss=0.293274 lr=6.089102e-05 lr_backbone=6.089102e-05 elapsed=26:33:20 avg_step=4.17s recent_step=4.16s eta=31:21:01 +2026-09-21 09:00:38 | INFO | train rank=0 | step=22940/50000 progress=45.9% loss=0.372430 lr=6.085875e-05 lr_backbone=6.085875e-05 elapsed=26:34:02 avg_step=4.17s recent_step=4.13s eta=31:20:19 +2026-09-21 09:01:20 | INFO | train rank=0 | step=22950/50000 progress=45.9% loss=0.401156 lr=6.082646e-05 lr_backbone=6.082646e-05 elapsed=26:34:43 avg_step=4.17s recent_step=4.15s eta=31:19:37 +2026-09-21 09:02:02 | INFO | train rank=0 | step=22960/50000 progress=45.9% loss=0.311918 lr=6.079418e-05 lr_backbone=6.079418e-05 elapsed=26:35:25 avg_step=4.17s recent_step=4.17s eta=31:18:55 +2026-09-21 09:02:42 | INFO | train rank=0 | step=22970/50000 progress=45.9% loss=0.307526 lr=6.076189e-05 lr_backbone=6.076189e-05 elapsed=26:36:06 avg_step=4.17s recent_step=4.09s eta=31:18:13 +2026-09-21 09:03:23 | INFO | train rank=0 | step=22980/50000 progress=46.0% loss=0.384303 lr=6.072959e-05 lr_backbone=6.072959e-05 elapsed=26:36:47 avg_step=4.17s recent_step=4.10s eta=31:17:30 +2026-09-21 09:04:05 | INFO | train rank=0 | step=22990/50000 progress=46.0% loss=0.301330 lr=6.069729e-05 lr_backbone=6.069729e-05 elapsed=26:37:28 avg_step=4.17s recent_step=4.13s eta=31:16:48 +2026-09-21 09:04:46 | INFO | train rank=0 | step=23000/50000 progress=46.0% loss=0.379395 lr=6.066498e-05 lr_backbone=6.066498e-05 elapsed=26:38:10 avg_step=4.17s recent_step=4.15s eta=31:16:06 +2026-09-21 09:05:28 | INFO | train rank=0 | step=23010/50000 progress=46.0% loss=0.367430 lr=6.063267e-05 lr_backbone=6.063267e-05 elapsed=26:38:51 avg_step=4.17s recent_step=4.15s eta=31:15:24 +2026-09-21 09:06:09 | INFO | train rank=0 | step=23020/50000 progress=46.0% loss=0.344331 lr=6.060036e-05 lr_backbone=6.060036e-05 elapsed=26:39:32 avg_step=4.17s recent_step=4.14s eta=31:14:42 +2026-09-21 09:06:50 | INFO | train rank=0 | step=23030/50000 progress=46.1% loss=0.311154 lr=6.056804e-05 lr_backbone=6.056804e-05 elapsed=26:40:14 avg_step=4.17s recent_step=4.11s eta=31:14:00 +2026-09-21 09:07:31 | INFO | train rank=0 | step=23040/50000 progress=46.1% loss=0.369749 lr=6.053571e-05 lr_backbone=6.053571e-05 elapsed=26:40:55 avg_step=4.17s recent_step=4.13s eta=31:13:18 +2026-09-21 09:08:13 | INFO | train rank=0 | step=23050/50000 progress=46.1% loss=0.357181 lr=6.050338e-05 lr_backbone=6.050338e-05 elapsed=26:41:36 avg_step=4.17s recent_step=4.13s eta=31:12:35 +2026-09-21 09:08:54 | INFO | train rank=0 | step=23060/50000 progress=46.1% loss=0.313910 lr=6.047105e-05 lr_backbone=6.047105e-05 elapsed=26:42:18 avg_step=4.17s recent_step=4.14s eta=31:11:53 +2026-09-21 09:09:36 | INFO | train rank=0 | step=23070/50000 progress=46.1% loss=0.395200 lr=6.043871e-05 lr_backbone=6.043871e-05 elapsed=26:42:59 avg_step=4.17s recent_step=4.16s eta=31:11:12 +2026-09-21 09:10:18 | INFO | train rank=0 | step=23080/50000 progress=46.2% loss=0.306179 lr=6.040637e-05 lr_backbone=6.040637e-05 elapsed=26:43:41 avg_step=4.17s recent_step=4.17s eta=31:10:30 +2026-09-21 09:10:58 | INFO | train rank=0 | step=23090/50000 progress=46.2% loss=0.412317 lr=6.037402e-05 lr_backbone=6.037402e-05 elapsed=26:44:22 avg_step=4.17s recent_step=4.10s eta=31:09:47 +2026-09-21 09:11:40 | INFO | train rank=0 | step=23100/50000 progress=46.2% loss=0.373682 lr=6.034167e-05 lr_backbone=6.034167e-05 elapsed=26:45:03 avg_step=4.17s recent_step=4.14s eta=31:09:05 +2026-09-21 09:12:21 | INFO | train rank=0 | step=23110/50000 progress=46.2% loss=0.326683 lr=6.030931e-05 lr_backbone=6.030931e-05 elapsed=26:45:44 avg_step=4.17s recent_step=4.12s eta=31:08:23 +2026-09-21 09:13:03 | INFO | train rank=0 | step=23120/50000 progress=46.2% loss=0.327624 lr=6.027695e-05 lr_backbone=6.027695e-05 elapsed=26:46:26 avg_step=4.17s recent_step=4.15s eta=31:07:41 +2026-09-21 09:13:44 | INFO | train rank=0 | step=23130/50000 progress=46.3% loss=0.374688 lr=6.024458e-05 lr_backbone=6.024458e-05 elapsed=26:47:08 avg_step=4.17s recent_step=4.17s eta=31:07:00 +2026-09-21 09:14:26 | INFO | train rank=0 | step=23140/50000 progress=46.3% loss=0.368468 lr=6.021221e-05 lr_backbone=6.021221e-05 elapsed=26:47:49 avg_step=4.17s recent_step=4.17s eta=31:06:18 +2026-09-21 09:15:07 | INFO | train rank=0 | step=23150/50000 progress=46.3% loss=0.338104 lr=6.017984e-05 lr_backbone=6.017984e-05 elapsed=26:48:30 avg_step=4.17s recent_step=4.11s eta=31:05:36 +2026-09-21 09:15:49 | INFO | train rank=0 | step=23160/50000 progress=46.3% loss=0.418949 lr=6.014746e-05 lr_backbone=6.014746e-05 elapsed=26:49:12 avg_step=4.17s recent_step=4.14s eta=31:04:54 +2026-09-21 09:16:30 | INFO | train rank=0 | step=23170/50000 progress=46.3% loss=0.337061 lr=6.011508e-05 lr_backbone=6.011508e-05 elapsed=26:49:53 avg_step=4.17s recent_step=4.13s eta=31:04:11 +2026-09-21 09:17:11 | INFO | train rank=0 | step=23180/50000 progress=46.4% loss=0.345154 lr=6.008269e-05 lr_backbone=6.008269e-05 elapsed=26:50:35 avg_step=4.17s recent_step=4.14s eta=31:03:29 +2026-09-21 09:17:53 | INFO | train rank=0 | step=23190/50000 progress=46.4% loss=0.314838 lr=6.005030e-05 lr_backbone=6.005030e-05 elapsed=26:51:16 avg_step=4.17s recent_step=4.17s eta=31:02:48 +2026-09-21 09:18:34 | INFO | train rank=0 | step=23200/50000 progress=46.4% loss=0.420932 lr=6.001790e-05 lr_backbone=6.001790e-05 elapsed=26:51:58 avg_step=4.17s recent_step=4.15s eta=31:02:06 +2026-09-21 09:19:15 | INFO | train rank=0 | step=23210/50000 progress=46.4% loss=0.432568 lr=5.998550e-05 lr_backbone=5.998550e-05 elapsed=26:52:39 avg_step=4.17s recent_step=4.10s eta=31:01:23 +2026-09-21 09:19:57 | INFO | train rank=0 | step=23220/50000 progress=46.4% loss=0.346175 lr=5.995309e-05 lr_backbone=5.995309e-05 elapsed=26:53:20 avg_step=4.17s recent_step=4.14s eta=31:00:41 +2026-09-21 09:20:38 | INFO | train rank=0 | step=23230/50000 progress=46.5% loss=0.359364 lr=5.992068e-05 lr_backbone=5.992068e-05 elapsed=26:54:01 avg_step=4.17s recent_step=4.13s eta=30:59:59 +2026-09-21 09:21:20 | INFO | train rank=0 | step=23240/50000 progress=46.5% loss=0.354324 lr=5.988827e-05 lr_backbone=5.988827e-05 elapsed=26:54:43 avg_step=4.17s recent_step=4.15s eta=30:59:17 +2026-09-21 09:22:01 | INFO | train rank=0 | step=23250/50000 progress=46.5% loss=0.423231 lr=5.985585e-05 lr_backbone=5.985585e-05 elapsed=26:55:25 avg_step=4.17s recent_step=4.17s eta=30:58:35 +2026-09-21 09:22:43 | INFO | train rank=0 | step=23260/50000 progress=46.5% loss=0.391008 lr=5.982343e-05 lr_backbone=5.982343e-05 elapsed=26:56:06 avg_step=4.17s recent_step=4.15s eta=30:57:54 +2026-09-21 09:23:24 | INFO | train rank=0 | step=23270/50000 progress=46.5% loss=0.373069 lr=5.979100e-05 lr_backbone=5.979100e-05 elapsed=26:56:47 avg_step=4.17s recent_step=4.10s eta=30:57:11 +2026-09-21 09:24:05 | INFO | train rank=0 | step=23280/50000 progress=46.6% loss=0.385451 lr=5.975857e-05 lr_backbone=5.975857e-05 elapsed=26:57:28 avg_step=4.17s recent_step=4.13s eta=30:56:29 +2026-09-21 09:24:46 | INFO | train rank=0 | step=23290/50000 progress=46.6% loss=0.331223 lr=5.972613e-05 lr_backbone=5.972613e-05 elapsed=26:58:10 avg_step=4.17s recent_step=4.13s eta=30:55:47 +2026-09-21 09:25:28 | INFO | train rank=0 | step=23300/50000 progress=46.6% loss=0.351555 lr=5.969369e-05 lr_backbone=5.969369e-05 elapsed=26:58:51 avg_step=4.17s recent_step=4.17s eta=30:55:05 +2026-09-21 09:26:10 | INFO | train rank=0 | step=23310/50000 progress=46.6% loss=0.377570 lr=5.966125e-05 lr_backbone=5.966125e-05 elapsed=26:59:33 avg_step=4.17s recent_step=4.16s eta=30:54:23 +2026-09-21 09:26:51 | INFO | train rank=0 | step=23320/50000 progress=46.6% loss=0.323098 lr=5.962880e-05 lr_backbone=5.962880e-05 elapsed=27:00:14 avg_step=4.17s recent_step=4.15s eta=30:53:41 +2026-09-21 09:27:32 | INFO | train rank=0 | step=23330/50000 progress=46.7% loss=0.345722 lr=5.959635e-05 lr_backbone=5.959635e-05 elapsed=27:00:55 avg_step=4.17s recent_step=4.09s eta=30:52:59 +2026-09-21 09:28:13 | INFO | train rank=0 | step=23340/50000 progress=46.7% loss=0.322439 lr=5.956389e-05 lr_backbone=5.956389e-05 elapsed=27:01:37 avg_step=4.17s recent_step=4.14s eta=30:52:17 +2026-09-21 09:28:55 | INFO | train rank=0 | step=23350/50000 progress=46.7% loss=0.370884 lr=5.953143e-05 lr_backbone=5.953143e-05 elapsed=27:02:18 avg_step=4.17s recent_step=4.14s eta=30:51:35 +2026-09-21 09:29:36 | INFO | train rank=0 | step=23360/50000 progress=46.7% loss=0.319597 lr=5.949897e-05 lr_backbone=5.949897e-05 elapsed=27:03:00 avg_step=4.17s recent_step=4.15s eta=30:50:53 +2026-09-21 09:30:18 | INFO | train rank=0 | step=23370/50000 progress=46.7% loss=0.401462 lr=5.946650e-05 lr_backbone=5.946650e-05 elapsed=27:03:41 avg_step=4.17s recent_step=4.15s eta=30:50:11 +2026-09-21 09:30:59 | INFO | train rank=0 | step=23380/50000 progress=46.8% loss=0.406045 lr=5.943402e-05 lr_backbone=5.943402e-05 elapsed=27:04:22 avg_step=4.17s recent_step=4.13s eta=30:49:29 +2026-09-21 09:31:40 | INFO | train rank=0 | step=23390/50000 progress=46.8% loss=0.313686 lr=5.940155e-05 lr_backbone=5.940155e-05 elapsed=27:05:04 avg_step=4.17s recent_step=4.11s eta=30:48:46 +2026-09-21 09:32:21 | INFO | train rank=0 | step=23400/50000 progress=46.8% loss=0.324182 lr=5.936907e-05 lr_backbone=5.936907e-05 elapsed=27:05:45 avg_step=4.17s recent_step=4.11s eta=30:48:04 +2026-09-21 09:33:03 | INFO | train rank=0 | step=23410/50000 progress=46.8% loss=0.319598 lr=5.933658e-05 lr_backbone=5.933658e-05 elapsed=27:06:26 avg_step=4.17s recent_step=4.16s eta=30:47:22 +2026-09-21 09:33:44 | INFO | train rank=0 | step=23420/50000 progress=46.8% loss=0.306093 lr=5.930409e-05 lr_backbone=5.930409e-05 elapsed=27:07:08 avg_step=4.17s recent_step=4.15s eta=30:46:40 +2026-09-21 09:34:26 | INFO | train rank=0 | step=23430/50000 progress=46.9% loss=0.314213 lr=5.927160e-05 lr_backbone=5.927160e-05 elapsed=27:07:49 avg_step=4.17s recent_step=4.16s eta=30:45:59 +2026-09-21 09:35:07 | INFO | train rank=0 | step=23440/50000 progress=46.9% loss=0.399178 lr=5.923910e-05 lr_backbone=5.923910e-05 elapsed=27:08:31 avg_step=4.17s recent_step=4.15s eta=30:45:17 +2026-09-21 09:35:48 | INFO | train rank=0 | step=23450/50000 progress=46.9% loss=0.412061 lr=5.920660e-05 lr_backbone=5.920660e-05 elapsed=27:09:12 avg_step=4.17s recent_step=4.09s eta=30:44:34 +2026-09-21 09:36:30 | INFO | train rank=0 | step=23460/50000 progress=46.9% loss=0.385984 lr=5.917409e-05 lr_backbone=5.917409e-05 elapsed=27:09:53 avg_step=4.17s recent_step=4.13s eta=30:43:52 +2026-09-21 09:37:11 | INFO | train rank=0 | step=23470/50000 progress=46.9% loss=0.407423 lr=5.914158e-05 lr_backbone=5.914158e-05 elapsed=27:10:35 avg_step=4.17s recent_step=4.15s eta=30:43:10 +2026-09-21 09:37:52 | INFO | train rank=0 | step=23480/50000 progress=47.0% loss=0.321317 lr=5.910907e-05 lr_backbone=5.910907e-05 elapsed=27:11:16 avg_step=4.17s recent_step=4.13s eta=30:42:28 +2026-09-21 09:38:34 | INFO | train rank=0 | step=23490/50000 progress=47.0% loss=0.302506 lr=5.907655e-05 lr_backbone=5.907655e-05 elapsed=27:11:58 avg_step=4.17s recent_step=4.17s eta=30:41:46 +2026-09-21 09:39:16 | INFO | train rank=0 | step=23500/50000 progress=47.0% loss=0.350682 lr=5.904403e-05 lr_backbone=5.904403e-05 elapsed=27:12:39 avg_step=4.17s recent_step=4.17s eta=30:41:05 +2026-09-21 09:39:57 | INFO | train rank=0 | step=23510/50000 progress=47.0% loss=0.318847 lr=5.901150e-05 lr_backbone=5.901150e-05 elapsed=27:13:20 avg_step=4.17s recent_step=4.10s eta=30:40:22 +2026-09-21 09:40:38 | INFO | train rank=0 | step=23520/50000 progress=47.0% loss=0.386953 lr=5.897897e-05 lr_backbone=5.897897e-05 elapsed=27:14:01 avg_step=4.17s recent_step=4.13s eta=30:39:40 +2026-09-21 09:41:20 | INFO | train rank=0 | step=23530/50000 progress=47.1% loss=0.331688 lr=5.894644e-05 lr_backbone=5.894644e-05 elapsed=27:14:43 avg_step=4.17s recent_step=4.15s eta=30:38:58 +2026-09-21 09:42:01 | INFO | train rank=0 | step=23540/50000 progress=47.1% loss=0.376666 lr=5.891390e-05 lr_backbone=5.891390e-05 elapsed=27:15:24 avg_step=4.17s recent_step=4.14s eta=30:38:16 +2026-09-21 09:42:43 | INFO | train rank=0 | step=23550/50000 progress=47.1% loss=0.359701 lr=5.888136e-05 lr_backbone=5.888136e-05 elapsed=27:16:06 avg_step=4.17s recent_step=4.17s eta=30:37:34 +2026-09-21 09:43:24 | INFO | train rank=0 | step=23560/50000 progress=47.1% loss=0.336316 lr=5.884881e-05 lr_backbone=5.884881e-05 elapsed=27:16:47 avg_step=4.17s recent_step=4.14s eta=30:36:52 +2026-09-21 09:44:05 | INFO | train rank=0 | step=23570/50000 progress=47.1% loss=0.441953 lr=5.881626e-05 lr_backbone=5.881626e-05 elapsed=27:17:28 avg_step=4.17s recent_step=4.10s eta=30:36:10 +2026-09-21 09:44:46 | INFO | train rank=0 | step=23580/50000 progress=47.2% loss=0.311462 lr=5.878371e-05 lr_backbone=5.878371e-05 elapsed=27:18:10 avg_step=4.17s recent_step=4.13s eta=30:35:28 +2026-09-21 09:45:28 | INFO | train rank=0 | step=23590/50000 progress=47.2% loss=0.283648 lr=5.875115e-05 lr_backbone=5.875115e-05 elapsed=27:18:51 avg_step=4.17s recent_step=4.16s eta=30:34:46 +2026-09-21 09:46:09 | INFO | train rank=0 | step=23600/50000 progress=47.2% loss=0.351767 lr=5.871859e-05 lr_backbone=5.871859e-05 elapsed=27:19:33 avg_step=4.17s recent_step=4.13s eta=30:34:04 +2026-09-21 09:46:51 | INFO | train rank=0 | step=23610/50000 progress=47.2% loss=0.417860 lr=5.868603e-05 lr_backbone=5.868603e-05 elapsed=27:20:14 avg_step=4.17s recent_step=4.14s eta=30:33:22 +2026-09-21 09:47:32 | INFO | train rank=0 | step=23620/50000 progress=47.2% loss=0.318069 lr=5.865346e-05 lr_backbone=5.865346e-05 elapsed=27:20:55 avg_step=4.17s recent_step=4.15s eta=30:32:40 +2026-09-21 09:48:13 | INFO | train rank=0 | step=23630/50000 progress=47.3% loss=0.302611 lr=5.862089e-05 lr_backbone=5.862089e-05 elapsed=27:21:36 avg_step=4.17s recent_step=4.10s eta=30:31:58 +2026-09-21 09:48:54 | INFO | train rank=0 | step=23640/50000 progress=47.3% loss=0.327808 lr=5.858831e-05 lr_backbone=5.858831e-05 elapsed=27:22:18 avg_step=4.17s recent_step=4.13s eta=30:31:15 +2026-09-21 09:49:36 | INFO | train rank=0 | step=23650/50000 progress=47.3% loss=0.390383 lr=5.855573e-05 lr_backbone=5.855573e-05 elapsed=27:22:59 avg_step=4.17s recent_step=4.15s eta=30:30:34 +2026-09-21 09:50:17 | INFO | train rank=0 | step=23660/50000 progress=47.3% loss=0.317001 lr=5.852315e-05 lr_backbone=5.852315e-05 elapsed=27:23:41 avg_step=4.17s recent_step=4.15s eta=30:29:52 +2026-09-21 09:50:59 | INFO | train rank=0 | step=23670/50000 progress=47.3% loss=0.310217 lr=5.849056e-05 lr_backbone=5.849056e-05 elapsed=27:24:22 avg_step=4.17s recent_step=4.14s eta=30:29:10 +2026-09-21 09:51:40 | INFO | train rank=0 | step=23680/50000 progress=47.4% loss=0.384090 lr=5.845797e-05 lr_backbone=5.845797e-05 elapsed=27:25:04 avg_step=4.17s recent_step=4.15s eta=30:28:28 +2026-09-21 09:52:21 | INFO | train rank=0 | step=23690/50000 progress=47.4% loss=0.371895 lr=5.842537e-05 lr_backbone=5.842537e-05 elapsed=27:25:45 avg_step=4.17s recent_step=4.12s eta=30:27:46 +2026-09-21 09:53:03 | INFO | train rank=0 | step=23700/50000 progress=47.4% loss=0.321648 lr=5.839278e-05 lr_backbone=5.839278e-05 elapsed=27:26:26 avg_step=4.17s recent_step=4.13s eta=30:27:04 +2026-09-21 09:53:44 | INFO | train rank=0 | step=23710/50000 progress=47.4% loss=0.356538 lr=5.836017e-05 lr_backbone=5.836017e-05 elapsed=27:27:08 avg_step=4.17s recent_step=4.16s eta=30:26:22 +2026-09-21 09:54:26 | INFO | train rank=0 | step=23720/50000 progress=47.4% loss=0.319998 lr=5.832757e-05 lr_backbone=5.832757e-05 elapsed=27:27:49 avg_step=4.17s recent_step=4.14s eta=30:25:40 +2026-09-21 09:55:07 | INFO | train rank=0 | step=23730/50000 progress=47.5% loss=0.368368 lr=5.829496e-05 lr_backbone=5.829496e-05 elapsed=27:28:31 avg_step=4.17s recent_step=4.15s eta=30:24:58 +2026-09-21 09:55:49 | INFO | train rank=0 | step=23740/50000 progress=47.5% loss=0.409983 lr=5.826235e-05 lr_backbone=5.826235e-05 elapsed=27:29:12 avg_step=4.17s recent_step=4.15s eta=30:24:16 +2026-09-21 09:56:30 | INFO | train rank=0 | step=23750/50000 progress=47.5% loss=0.356473 lr=5.822973e-05 lr_backbone=5.822973e-05 elapsed=27:29:53 avg_step=4.17s recent_step=4.10s eta=30:23:33 +2026-09-21 09:57:11 | INFO | train rank=0 | step=23760/50000 progress=47.5% loss=0.332257 lr=5.819711e-05 lr_backbone=5.819711e-05 elapsed=27:30:34 avg_step=4.17s recent_step=4.13s eta=30:22:51 +2026-09-21 09:57:53 | INFO | train rank=0 | step=23770/50000 progress=47.5% loss=0.335085 lr=5.816449e-05 lr_backbone=5.816449e-05 elapsed=27:31:16 avg_step=4.17s recent_step=4.15s eta=30:22:09 +2026-09-21 09:58:34 | INFO | train rank=0 | step=23780/50000 progress=47.6% loss=0.337508 lr=5.813186e-05 lr_backbone=5.813186e-05 elapsed=27:31:57 avg_step=4.17s recent_step=4.13s eta=30:21:27 +2026-09-21 09:59:15 | INFO | train rank=0 | step=23790/50000 progress=47.6% loss=0.331575 lr=5.809923e-05 lr_backbone=5.809923e-05 elapsed=27:32:39 avg_step=4.17s recent_step=4.15s eta=30:20:46 +2026-09-21 09:59:57 | INFO | train rank=0 | step=23800/50000 progress=47.6% loss=0.392792 lr=5.806659e-05 lr_backbone=5.806659e-05 elapsed=27:33:20 avg_step=4.17s recent_step=4.15s eta=30:20:04 +2026-09-21 10:00:38 | INFO | train rank=0 | step=23810/50000 progress=47.6% loss=0.357829 lr=5.803395e-05 lr_backbone=5.803395e-05 elapsed=27:34:01 avg_step=4.17s recent_step=4.10s eta=30:19:21 +2026-09-21 10:01:19 | INFO | train rank=0 | step=23820/50000 progress=47.6% loss=0.286467 lr=5.800131e-05 lr_backbone=5.800131e-05 elapsed=27:34:43 avg_step=4.17s recent_step=4.13s eta=30:18:39 +2026-09-21 10:02:01 | INFO | train rank=0 | step=23830/50000 progress=47.7% loss=0.380491 lr=5.796867e-05 lr_backbone=5.796867e-05 elapsed=27:35:24 avg_step=4.17s recent_step=4.15s eta=30:17:57 +2026-09-21 10:02:42 | INFO | train rank=0 | step=23840/50000 progress=47.7% loss=0.388398 lr=5.793602e-05 lr_backbone=5.793602e-05 elapsed=27:36:06 avg_step=4.17s recent_step=4.17s eta=30:17:16 +2026-09-21 10:03:24 | INFO | train rank=0 | step=23850/50000 progress=47.7% loss=0.299017 lr=5.790337e-05 lr_backbone=5.790337e-05 elapsed=27:36:47 avg_step=4.17s recent_step=4.13s eta=30:16:33 +2026-09-21 10:04:05 | INFO | train rank=0 | step=23860/50000 progress=47.7% loss=0.415515 lr=5.787071e-05 lr_backbone=5.787071e-05 elapsed=27:37:29 avg_step=4.17s recent_step=4.17s eta=30:15:52 +2026-09-21 10:04:46 | INFO | train rank=0 | step=23870/50000 progress=47.7% loss=0.344482 lr=5.783805e-05 lr_backbone=5.783805e-05 elapsed=27:38:10 avg_step=4.17s recent_step=4.11s eta=30:15:09 +2026-09-21 10:05:28 | INFO | train rank=0 | step=23880/50000 progress=47.8% loss=0.318761 lr=5.780539e-05 lr_backbone=5.780539e-05 elapsed=27:38:51 avg_step=4.17s recent_step=4.13s eta=30:14:27 +2026-09-21 10:06:09 | INFO | train rank=0 | step=23890/50000 progress=47.8% loss=0.320425 lr=5.777273e-05 lr_backbone=5.777273e-05 elapsed=27:39:32 avg_step=4.17s recent_step=4.13s eta=30:13:45 +2026-09-21 10:06:51 | INFO | train rank=0 | step=23900/50000 progress=47.8% loss=0.314814 lr=5.774006e-05 lr_backbone=5.774006e-05 elapsed=27:40:14 avg_step=4.17s recent_step=4.15s eta=30:13:03 +2026-09-21 10:07:32 | INFO | train rank=0 | step=23910/50000 progress=47.8% loss=0.299606 lr=5.770738e-05 lr_backbone=5.770738e-05 elapsed=27:40:55 avg_step=4.17s recent_step=4.13s eta=30:12:21 +2026-09-21 10:08:13 | INFO | train rank=0 | step=23920/50000 progress=47.8% loss=0.374106 lr=5.767471e-05 lr_backbone=5.767471e-05 elapsed=27:41:37 avg_step=4.17s recent_step=4.15s eta=30:11:39 +2026-09-21 10:08:54 | INFO | train rank=0 | step=23930/50000 progress=47.9% loss=0.399539 lr=5.764203e-05 lr_backbone=5.764203e-05 elapsed=27:42:18 avg_step=4.17s recent_step=4.10s eta=30:10:57 +2026-09-21 10:09:36 | INFO | train rank=0 | step=23940/50000 progress=47.9% loss=0.323692 lr=5.760935e-05 lr_backbone=5.760935e-05 elapsed=27:42:59 avg_step=4.17s recent_step=4.14s eta=30:10:15 +2026-09-21 10:10:17 | INFO | train rank=0 | step=23950/50000 progress=47.9% loss=0.372317 lr=5.757666e-05 lr_backbone=5.757666e-05 elapsed=27:43:41 avg_step=4.17s recent_step=4.14s eta=30:09:33 +2026-09-21 10:10:59 | INFO | train rank=0 | step=23960/50000 progress=47.9% loss=0.362732 lr=5.754397e-05 lr_backbone=5.754397e-05 elapsed=27:44:22 avg_step=4.17s recent_step=4.15s eta=30:08:51 +2026-09-21 10:11:40 | INFO | train rank=0 | step=23970/50000 progress=47.9% loss=0.331644 lr=5.751128e-05 lr_backbone=5.751128e-05 elapsed=27:45:04 avg_step=4.17s recent_step=4.14s eta=30:08:09 +2026-09-21 10:12:22 | INFO | train rank=0 | step=23980/50000 progress=48.0% loss=0.390010 lr=5.747858e-05 lr_backbone=5.747858e-05 elapsed=27:45:45 avg_step=4.17s recent_step=4.15s eta=30:07:27 +2026-09-21 10:13:03 | INFO | train rank=0 | step=23990/50000 progress=48.0% loss=0.373380 lr=5.744588e-05 lr_backbone=5.744588e-05 elapsed=27:46:26 avg_step=4.17s recent_step=4.11s eta=30:06:45 +2026-09-21 10:13:44 | INFO | train rank=0 | step=24000/50000 progress=48.0% loss=0.365970 lr=5.741318e-05 lr_backbone=5.741318e-05 elapsed=27:47:07 avg_step=4.17s recent_step=4.14s eta=30:06:03 +2026-09-21 10:14:26 | INFO | train rank=0 | step=24010/50000 progress=48.0% loss=0.424714 lr=5.738048e-05 lr_backbone=5.738048e-05 elapsed=27:47:49 avg_step=4.17s recent_step=4.16s eta=30:05:21 +2026-09-21 10:15:07 | INFO | train rank=0 | step=24020/50000 progress=48.0% loss=0.401971 lr=5.734777e-05 lr_backbone=5.734777e-05 elapsed=27:48:31 avg_step=4.17s recent_step=4.16s eta=30:04:40 +2026-09-21 10:15:49 | INFO | train rank=0 | step=24030/50000 progress=48.1% loss=0.480129 lr=5.731506e-05 lr_backbone=5.731506e-05 elapsed=27:49:12 avg_step=4.17s recent_step=4.18s eta=30:03:58 +2026-09-21 10:16:30 | INFO | train rank=0 | step=24040/50000 progress=48.1% loss=0.380393 lr=5.728234e-05 lr_backbone=5.728234e-05 elapsed=27:49:54 avg_step=4.17s recent_step=4.13s eta=30:03:16 +2026-09-21 10:17:11 | INFO | train rank=0 | step=24050/50000 progress=48.1% loss=0.350969 lr=5.724962e-05 lr_backbone=5.724962e-05 elapsed=27:50:35 avg_step=4.17s recent_step=4.10s eta=30:02:34 +2026-09-21 10:17:53 | INFO | train rank=0 | step=24060/50000 progress=48.1% loss=0.334446 lr=5.721690e-05 lr_backbone=5.721690e-05 elapsed=27:51:16 avg_step=4.17s recent_step=4.13s eta=30:01:51 +2026-09-21 10:18:34 | INFO | train rank=0 | step=24070/50000 progress=48.1% loss=0.362069 lr=5.718418e-05 lr_backbone=5.718418e-05 elapsed=27:51:58 avg_step=4.17s recent_step=4.15s eta=30:01:10 +2026-09-21 10:19:16 | INFO | train rank=0 | step=24080/50000 progress=48.2% loss=0.342663 lr=5.715145e-05 lr_backbone=5.715145e-05 elapsed=27:52:39 avg_step=4.17s recent_step=4.16s eta=30:00:28 +2026-09-21 10:19:57 | INFO | train rank=0 | step=24090/50000 progress=48.2% loss=0.305914 lr=5.711872e-05 lr_backbone=5.711872e-05 elapsed=27:53:21 avg_step=4.17s recent_step=4.15s eta=29:59:46 +2026-09-21 10:20:39 | INFO | train rank=0 | step=24100/50000 progress=48.2% loss=0.385382 lr=5.708598e-05 lr_backbone=5.708598e-05 elapsed=27:54:02 avg_step=4.17s recent_step=4.15s eta=29:59:04 +2026-09-21 10:21:20 | INFO | train rank=0 | step=24110/50000 progress=48.2% loss=0.307840 lr=5.705325e-05 lr_backbone=5.705325e-05 elapsed=27:54:43 avg_step=4.17s recent_step=4.10s eta=29:58:22 +2026-09-21 10:22:01 | INFO | train rank=0 | step=24120/50000 progress=48.2% loss=0.379771 lr=5.702051e-05 lr_backbone=5.702051e-05 elapsed=27:55:25 avg_step=4.17s recent_step=4.14s eta=29:57:40 +2026-09-21 10:22:43 | INFO | train rank=0 | step=24130/50000 progress=48.3% loss=0.350192 lr=5.698776e-05 lr_backbone=5.698776e-05 elapsed=27:56:06 avg_step=4.17s recent_step=4.17s eta=29:56:58 +2026-09-21 10:23:24 | INFO | train rank=0 | step=24140/50000 progress=48.3% loss=0.336879 lr=5.695502e-05 lr_backbone=5.695502e-05 elapsed=27:56:48 avg_step=4.17s recent_step=4.15s eta=29:56:16 +2026-09-21 10:24:06 | INFO | train rank=0 | step=24150/50000 progress=48.3% loss=0.336994 lr=5.692227e-05 lr_backbone=5.692227e-05 elapsed=27:57:29 avg_step=4.17s recent_step=4.14s eta=29:55:34 +2026-09-21 10:24:47 | INFO | train rank=0 | step=24160/50000 progress=48.3% loss=0.323179 lr=5.688951e-05 lr_backbone=5.688951e-05 elapsed=27:58:11 avg_step=4.17s recent_step=4.14s eta=29:54:52 +2026-09-21 10:25:28 | INFO | train rank=0 | step=24170/50000 progress=48.3% loss=0.320977 lr=5.685676e-05 lr_backbone=5.685676e-05 elapsed=27:58:52 avg_step=4.17s recent_step=4.10s eta=29:54:10 +2026-09-21 10:26:10 | INFO | train rank=0 | step=24180/50000 progress=48.4% loss=0.359202 lr=5.682400e-05 lr_backbone=5.682400e-05 elapsed=27:59:33 avg_step=4.17s recent_step=4.13s eta=29:53:28 +2026-09-21 10:26:51 | INFO | train rank=0 | step=24190/50000 progress=48.4% loss=0.318052 lr=5.679124e-05 lr_backbone=5.679124e-05 elapsed=28:00:14 avg_step=4.17s recent_step=4.15s eta=29:52:46 +2026-09-21 10:27:33 | INFO | train rank=0 | step=24200/50000 progress=48.4% loss=0.337731 lr=5.675847e-05 lr_backbone=5.675847e-05 elapsed=28:00:56 avg_step=4.17s recent_step=4.16s eta=29:52:04 +2026-09-21 10:28:14 | INFO | train rank=0 | step=24210/50000 progress=48.4% loss=0.394503 lr=5.672571e-05 lr_backbone=5.672571e-05 elapsed=28:01:38 avg_step=4.17s recent_step=4.15s eta=29:51:22 +2026-09-21 10:28:55 | INFO | train rank=0 | step=24220/50000 progress=48.4% loss=0.355858 lr=5.669294e-05 lr_backbone=5.669294e-05 elapsed=28:02:19 avg_step=4.17s recent_step=4.11s eta=29:50:40 +2026-09-21 10:29:36 | INFO | train rank=0 | step=24230/50000 progress=48.5% loss=0.367162 lr=5.666016e-05 lr_backbone=5.666016e-05 elapsed=28:03:00 avg_step=4.17s recent_step=4.09s eta=29:49:58 +2026-09-21 10:30:17 | INFO | train rank=0 | step=24240/50000 progress=48.5% loss=0.403225 lr=5.662739e-05 lr_backbone=5.662739e-05 elapsed=28:03:41 avg_step=4.17s recent_step=4.11s eta=29:49:15 +2026-09-21 10:30:59 | INFO | train rank=0 | step=24250/50000 progress=48.5% loss=0.367132 lr=5.659461e-05 lr_backbone=5.659461e-05 elapsed=28:04:22 avg_step=4.17s recent_step=4.14s eta=29:48:33 +2026-09-21 10:31:40 | INFO | train rank=0 | step=24260/50000 progress=48.5% loss=0.403510 lr=5.656183e-05 lr_backbone=5.656183e-05 elapsed=28:05:03 avg_step=4.17s recent_step=4.14s eta=29:47:51 +2026-09-21 10:32:22 | INFO | train rank=0 | step=24270/50000 progress=48.5% loss=0.354808 lr=5.652904e-05 lr_backbone=5.652904e-05 elapsed=28:05:45 avg_step=4.17s recent_step=4.14s eta=29:47:09 +2026-09-21 10:33:03 | INFO | train rank=0 | step=24280/50000 progress=48.6% loss=0.383018 lr=5.649625e-05 lr_backbone=5.649625e-05 elapsed=28:06:26 avg_step=4.17s recent_step=4.12s eta=29:46:27 +2026-09-21 10:33:44 | INFO | train rank=0 | step=24290/50000 progress=48.6% loss=0.412082 lr=5.646346e-05 lr_backbone=5.646346e-05 elapsed=28:07:07 avg_step=4.17s recent_step=4.08s eta=29:45:45 +2026-09-21 10:34:25 | INFO | train rank=0 | step=24300/50000 progress=48.6% loss=0.359516 lr=5.643067e-05 lr_backbone=5.643067e-05 elapsed=28:07:48 avg_step=4.17s recent_step=4.13s eta=29:45:03 +2026-09-21 10:35:06 | INFO | train rank=0 | step=24310/50000 progress=48.6% loss=0.354382 lr=5.639787e-05 lr_backbone=5.639787e-05 elapsed=28:08:29 avg_step=4.17s recent_step=4.12s eta=29:44:20 +2026-09-21 10:35:47 | INFO | train rank=0 | step=24320/50000 progress=48.6% loss=0.324618 lr=5.636508e-05 lr_backbone=5.636508e-05 elapsed=28:09:11 avg_step=4.17s recent_step=4.13s eta=29:43:38 +2026-09-21 10:36:29 | INFO | train rank=0 | step=24330/50000 progress=48.7% loss=0.435384 lr=5.633227e-05 lr_backbone=5.633227e-05 elapsed=28:09:52 avg_step=4.17s recent_step=4.13s eta=29:42:56 +2026-09-21 10:37:10 | INFO | train rank=0 | step=24340/50000 progress=48.7% loss=0.358785 lr=5.629947e-05 lr_backbone=5.629947e-05 elapsed=28:10:33 avg_step=4.17s recent_step=4.14s eta=29:42:14 +2026-09-21 10:37:51 | INFO | train rank=0 | step=24350/50000 progress=48.7% loss=0.328544 lr=5.626666e-05 lr_backbone=5.626666e-05 elapsed=28:11:14 avg_step=4.17s recent_step=4.07s eta=29:41:32 +2026-09-21 10:38:32 | INFO | train rank=0 | step=24360/50000 progress=48.7% loss=0.428892 lr=5.623385e-05 lr_backbone=5.623385e-05 elapsed=28:11:55 avg_step=4.17s recent_step=4.11s eta=29:40:49 +2026-09-21 10:39:13 | INFO | train rank=0 | step=24370/50000 progress=48.7% loss=0.421584 lr=5.620104e-05 lr_backbone=5.620104e-05 elapsed=28:12:37 avg_step=4.17s recent_step=4.15s eta=29:40:07 +2026-09-21 10:39:55 | INFO | train rank=0 | step=24380/50000 progress=48.8% loss=0.312823 lr=5.616822e-05 lr_backbone=5.616822e-05 elapsed=28:13:18 avg_step=4.17s recent_step=4.13s eta=29:39:25 +2026-09-21 10:40:36 | INFO | train rank=0 | step=24390/50000 progress=48.8% loss=0.357572 lr=5.613541e-05 lr_backbone=5.613541e-05 elapsed=28:13:59 avg_step=4.17s recent_step=4.13s eta=29:38:43 +2026-09-21 10:41:17 | INFO | train rank=0 | step=24400/50000 progress=48.8% loss=0.409434 lr=5.610258e-05 lr_backbone=5.610258e-05 elapsed=28:14:40 avg_step=4.17s recent_step=4.12s eta=29:38:01 +2026-09-21 10:41:58 | INFO | train rank=0 | step=24410/50000 progress=48.8% loss=0.367402 lr=5.606976e-05 lr_backbone=5.606976e-05 elapsed=28:15:21 avg_step=4.17s recent_step=4.08s eta=29:37:19 +2026-09-21 10:42:39 | INFO | train rank=0 | step=24420/50000 progress=48.8% loss=0.471796 lr=5.603693e-05 lr_backbone=5.603693e-05 elapsed=28:16:03 avg_step=4.17s recent_step=4.14s eta=29:36:37 +2026-09-21 10:43:21 | INFO | train rank=0 | step=24430/50000 progress=48.9% loss=0.347053 lr=5.600411e-05 lr_backbone=5.600411e-05 elapsed=28:16:44 avg_step=4.17s recent_step=4.14s eta=29:35:55 +2026-09-21 10:44:02 | INFO | train rank=0 | step=24440/50000 progress=48.9% loss=0.304650 lr=5.597127e-05 lr_backbone=5.597127e-05 elapsed=28:17:25 avg_step=4.17s recent_step=4.14s eta=29:35:13 +2026-09-21 10:44:44 | INFO | train rank=0 | step=24450/50000 progress=48.9% loss=0.384984 lr=5.593844e-05 lr_backbone=5.593844e-05 elapsed=28:18:07 avg_step=4.17s recent_step=4.14s eta=29:34:31 +2026-09-21 10:45:25 | INFO | train rank=0 | step=24460/50000 progress=48.9% loss=0.393864 lr=5.590560e-05 lr_backbone=5.590560e-05 elapsed=28:18:48 avg_step=4.17s recent_step=4.13s eta=29:33:49 +2026-09-21 10:46:06 | INFO | train rank=0 | step=24470/50000 progress=48.9% loss=0.377577 lr=5.587276e-05 lr_backbone=5.587276e-05 elapsed=28:19:29 avg_step=4.17s recent_step=4.08s eta=29:33:06 +2026-09-21 10:46:47 | INFO | train rank=0 | step=24480/50000 progress=49.0% loss=0.364430 lr=5.583992e-05 lr_backbone=5.583992e-05 elapsed=28:20:10 avg_step=4.17s recent_step=4.11s eta=29:32:24 +2026-09-21 10:47:28 | INFO | train rank=0 | step=24490/50000 progress=49.0% loss=0.320142 lr=5.580708e-05 lr_backbone=5.580708e-05 elapsed=28:20:51 avg_step=4.17s recent_step=4.12s eta=29:31:42 +2026-09-21 10:48:09 | INFO | train rank=0 | step=24500/50000 progress=49.0% loss=0.333377 lr=5.577423e-05 lr_backbone=5.577423e-05 elapsed=28:21:33 avg_step=4.17s recent_step=4.13s eta=29:31:00 +2026-09-21 10:48:51 | INFO | train rank=0 | step=24510/50000 progress=49.0% loss=0.290146 lr=5.574138e-05 lr_backbone=5.574138e-05 elapsed=28:22:14 avg_step=4.17s recent_step=4.14s eta=29:30:18 +2026-09-21 10:49:32 | INFO | train rank=0 | step=24520/50000 progress=49.0% loss=0.408326 lr=5.570853e-05 lr_backbone=5.570853e-05 elapsed=28:22:55 avg_step=4.17s recent_step=4.13s eta=29:29:36 +2026-09-21 10:50:13 | INFO | train rank=0 | step=24530/50000 progress=49.1% loss=0.418099 lr=5.567568e-05 lr_backbone=5.567568e-05 elapsed=28:23:36 avg_step=4.17s recent_step=4.09s eta=29:28:53 +2026-09-21 10:50:54 | INFO | train rank=0 | step=24540/50000 progress=49.1% loss=0.322345 lr=5.564282e-05 lr_backbone=5.564282e-05 elapsed=28:24:17 avg_step=4.17s recent_step=4.10s eta=29:28:11 +2026-09-21 10:51:35 | INFO | train rank=0 | step=24550/50000 progress=49.1% loss=0.319068 lr=5.560996e-05 lr_backbone=5.560996e-05 elapsed=28:24:59 avg_step=4.17s recent_step=4.14s eta=29:27:29 +2026-09-21 10:52:16 | INFO | train rank=0 | step=24560/50000 progress=49.1% loss=0.363259 lr=5.557710e-05 lr_backbone=5.557710e-05 elapsed=28:25:40 avg_step=4.17s recent_step=4.11s eta=29:26:47 +2026-09-21 10:52:58 | INFO | train rank=0 | step=24570/50000 progress=49.1% loss=0.420733 lr=5.554423e-05 lr_backbone=5.554423e-05 elapsed=28:26:21 avg_step=4.17s recent_step=4.13s eta=29:26:05 +2026-09-21 10:53:39 | INFO | train rank=0 | step=24580/50000 progress=49.2% loss=0.340638 lr=5.551137e-05 lr_backbone=5.551137e-05 elapsed=28:27:02 avg_step=4.17s recent_step=4.14s eta=29:25:23 +2026-09-21 10:54:20 | INFO | train rank=0 | step=24590/50000 progress=49.2% loss=0.366773 lr=5.547850e-05 lr_backbone=5.547850e-05 elapsed=28:27:43 avg_step=4.17s recent_step=4.10s eta=29:24:40 +2026-09-21 10:55:01 | INFO | train rank=0 | step=24600/50000 progress=49.2% loss=0.386933 lr=5.544563e-05 lr_backbone=5.544563e-05 elapsed=28:28:24 avg_step=4.17s recent_step=4.11s eta=29:23:58 +2026-09-21 10:55:42 | INFO | train rank=0 | step=24610/50000 progress=49.2% loss=0.329859 lr=5.541275e-05 lr_backbone=5.541275e-05 elapsed=28:29:06 avg_step=4.17s recent_step=4.12s eta=29:23:16 +2026-09-21 10:56:24 | INFO | train rank=0 | step=24620/50000 progress=49.2% loss=0.356499 lr=5.537988e-05 lr_backbone=5.537988e-05 elapsed=28:29:47 avg_step=4.17s recent_step=4.15s eta=29:22:34 +2026-09-21 10:57:05 | INFO | train rank=0 | step=24630/50000 progress=49.3% loss=0.515560 lr=5.534700e-05 lr_backbone=5.534700e-05 elapsed=28:30:28 avg_step=4.17s recent_step=4.13s eta=29:21:52 +2026-09-21 10:57:46 | INFO | train rank=0 | step=24640/50000 progress=49.3% loss=0.277982 lr=5.531412e-05 lr_backbone=5.531412e-05 elapsed=28:31:10 avg_step=4.17s recent_step=4.14s eta=29:21:10 +2026-09-21 10:58:27 | INFO | train rank=0 | step=24650/50000 progress=49.3% loss=0.358668 lr=5.528123e-05 lr_backbone=5.528123e-05 elapsed=28:31:50 avg_step=4.17s recent_step=4.07s eta=29:20:27 +2026-09-21 10:59:08 | INFO | train rank=0 | step=24660/50000 progress=49.3% loss=0.384087 lr=5.524835e-05 lr_backbone=5.524835e-05 elapsed=28:32:31 avg_step=4.17s recent_step=4.10s eta=29:19:45 +2026-09-21 10:59:49 | INFO | train rank=0 | step=24670/50000 progress=49.3% loss=0.297333 lr=5.521546e-05 lr_backbone=5.521546e-05 elapsed=28:33:13 avg_step=4.17s recent_step=4.13s eta=29:19:03 +2026-09-21 11:00:31 | INFO | train rank=0 | step=24680/50000 progress=49.4% loss=0.361927 lr=5.518257e-05 lr_backbone=5.518257e-05 elapsed=28:33:54 avg_step=4.17s recent_step=4.15s eta=29:18:21 +2026-09-21 11:01:12 | INFO | train rank=0 | step=24690/50000 progress=49.4% loss=0.373624 lr=5.514968e-05 lr_backbone=5.514968e-05 elapsed=28:34:36 avg_step=4.17s recent_step=4.14s eta=29:17:39 +2026-09-21 11:01:54 | INFO | train rank=0 | step=24700/50000 progress=49.4% loss=0.370991 lr=5.511678e-05 lr_backbone=5.511678e-05 elapsed=28:35:17 avg_step=4.17s recent_step=4.13s eta=29:16:57 +2026-09-21 11:02:35 | INFO | train rank=0 | step=24710/50000 progress=49.4% loss=0.456394 lr=5.508389e-05 lr_backbone=5.508389e-05 elapsed=28:35:58 avg_step=4.17s recent_step=4.09s eta=29:16:15 +2026-09-21 11:03:16 | INFO | train rank=0 | step=24720/50000 progress=49.4% loss=0.299185 lr=5.505099e-05 lr_backbone=5.505099e-05 elapsed=28:36:39 avg_step=4.17s recent_step=4.11s eta=29:15:32 +2026-09-21 11:03:57 | INFO | train rank=0 | step=24730/50000 progress=49.5% loss=0.372671 lr=5.501809e-05 lr_backbone=5.501809e-05 elapsed=28:37:21 avg_step=4.17s recent_step=4.15s eta=29:14:51 +2026-09-21 11:04:38 | INFO | train rank=0 | step=24740/50000 progress=49.5% loss=0.318381 lr=5.498518e-05 lr_backbone=5.498518e-05 elapsed=28:38:02 avg_step=4.17s recent_step=4.13s eta=29:14:08 +2026-09-21 11:05:20 | INFO | train rank=0 | step=24750/50000 progress=49.5% loss=0.357215 lr=5.495228e-05 lr_backbone=5.495228e-05 elapsed=28:38:43 avg_step=4.17s recent_step=4.14s eta=29:13:27 +2026-09-21 11:06:01 | INFO | train rank=0 | step=24760/50000 progress=49.5% loss=0.301697 lr=5.491937e-05 lr_backbone=5.491937e-05 elapsed=28:39:24 avg_step=4.17s recent_step=4.11s eta=29:12:44 +2026-09-21 11:06:42 | INFO | train rank=0 | step=24770/50000 progress=49.5% loss=0.333677 lr=5.488646e-05 lr_backbone=5.488646e-05 elapsed=28:40:05 avg_step=4.17s recent_step=4.09s eta=29:12:02 +2026-09-21 11:07:23 | INFO | train rank=0 | step=24780/50000 progress=49.6% loss=0.324260 lr=5.485355e-05 lr_backbone=5.485355e-05 elapsed=28:40:46 avg_step=4.17s recent_step=4.09s eta=29:11:19 +2026-09-21 11:08:04 | INFO | train rank=0 | step=24790/50000 progress=49.6% loss=0.352855 lr=5.482063e-05 lr_backbone=5.482063e-05 elapsed=28:41:27 avg_step=4.17s recent_step=4.13s eta=29:10:37 +2026-09-21 11:08:45 | INFO | train rank=0 | step=24800/50000 progress=49.6% loss=0.316202 lr=5.478772e-05 lr_backbone=5.478772e-05 elapsed=28:42:09 avg_step=4.17s recent_step=4.14s eta=29:09:55 +2026-09-21 11:09:27 | INFO | train rank=0 | step=24810/50000 progress=49.6% loss=0.370465 lr=5.475480e-05 lr_backbone=5.475480e-05 elapsed=28:42:50 avg_step=4.17s recent_step=4.12s eta=29:09:13 +2026-09-21 11:10:08 | INFO | train rank=0 | step=24820/50000 progress=49.6% loss=0.334956 lr=5.472188e-05 lr_backbone=5.472188e-05 elapsed=28:43:31 avg_step=4.17s recent_step=4.14s eta=29:08:31 +2026-09-21 11:10:49 | INFO | train rank=0 | step=24830/50000 progress=49.7% loss=0.369898 lr=5.468895e-05 lr_backbone=5.468895e-05 elapsed=28:44:12 avg_step=4.17s recent_step=4.10s eta=29:07:49 +2026-09-21 11:11:30 | INFO | train rank=0 | step=24840/50000 progress=49.7% loss=0.315697 lr=5.465603e-05 lr_backbone=5.465603e-05 elapsed=28:44:53 avg_step=4.17s recent_step=4.10s eta=29:07:07 +2026-09-21 11:12:12 | INFO | train rank=0 | step=24850/50000 progress=49.7% loss=0.395875 lr=5.462310e-05 lr_backbone=5.462310e-05 elapsed=28:45:35 avg_step=4.17s recent_step=4.16s eta=29:06:25 +2026-09-21 11:12:53 | INFO | train rank=0 | step=24860/50000 progress=49.7% loss=0.343341 lr=5.459017e-05 lr_backbone=5.459017e-05 elapsed=28:46:17 avg_step=4.17s recent_step=4.16s eta=29:05:43 +2026-09-21 11:13:35 | INFO | train rank=0 | step=24870/50000 progress=49.7% loss=0.374326 lr=5.455724e-05 lr_backbone=5.455724e-05 elapsed=28:46:58 avg_step=4.17s recent_step=4.15s eta=29:05:01 +2026-09-21 11:14:16 | INFO | train rank=0 | step=24880/50000 progress=49.8% loss=0.409742 lr=5.452431e-05 lr_backbone=5.452431e-05 elapsed=28:47:40 avg_step=4.17s recent_step=4.15s eta=29:04:20 +2026-09-21 11:14:57 | INFO | train rank=0 | step=24890/50000 progress=49.8% loss=0.313561 lr=5.449138e-05 lr_backbone=5.449138e-05 elapsed=28:48:21 avg_step=4.17s recent_step=4.09s eta=29:03:37 +2026-09-21 11:15:39 | INFO | train rank=0 | step=24900/50000 progress=49.8% loss=0.377249 lr=5.445844e-05 lr_backbone=5.445844e-05 elapsed=28:49:02 avg_step=4.17s recent_step=4.14s eta=29:02:55 +2026-09-21 11:16:20 | INFO | train rank=0 | step=24910/50000 progress=49.8% loss=0.322053 lr=5.442550e-05 lr_backbone=5.442550e-05 elapsed=28:49:43 avg_step=4.17s recent_step=4.14s eta=29:02:13 +2026-09-21 11:17:02 | INFO | train rank=0 | step=24920/50000 progress=49.8% loss=0.361855 lr=5.439256e-05 lr_backbone=5.439256e-05 elapsed=28:50:25 avg_step=4.17s recent_step=4.15s eta=29:01:32 +2026-09-21 11:17:43 | INFO | train rank=0 | step=24930/50000 progress=49.9% loss=0.447607 lr=5.435962e-05 lr_backbone=5.435962e-05 elapsed=28:51:06 avg_step=4.17s recent_step=4.13s eta=29:00:49 +2026-09-21 11:18:24 | INFO | train rank=0 | step=24940/50000 progress=49.9% loss=0.348673 lr=5.432667e-05 lr_backbone=5.432667e-05 elapsed=28:51:48 avg_step=4.17s recent_step=4.16s eta=29:00:08 +2026-09-21 11:19:05 | INFO | train rank=0 | step=24950/50000 progress=49.9% loss=0.368816 lr=5.429373e-05 lr_backbone=5.429373e-05 elapsed=28:52:29 avg_step=4.17s recent_step=4.09s eta=28:59:25 +2026-09-21 11:19:47 | INFO | train rank=0 | step=24960/50000 progress=49.9% loss=0.388945 lr=5.426078e-05 lr_backbone=5.426078e-05 elapsed=28:53:10 avg_step=4.17s recent_step=4.13s eta=28:58:43 +2026-09-21 11:20:28 | INFO | train rank=0 | step=24970/50000 progress=49.9% loss=0.376219 lr=5.422783e-05 lr_backbone=5.422783e-05 elapsed=28:53:51 avg_step=4.17s recent_step=4.13s eta=28:58:01 +2026-09-21 11:21:10 | INFO | train rank=0 | step=24980/50000 progress=50.0% loss=0.353814 lr=5.419488e-05 lr_backbone=5.419488e-05 elapsed=28:54:33 avg_step=4.17s recent_step=4.16s eta=28:57:20 +2026-09-21 11:21:51 | INFO | train rank=0 | step=24990/50000 progress=50.0% loss=0.385189 lr=5.416192e-05 lr_backbone=5.416192e-05 elapsed=28:55:15 avg_step=4.17s recent_step=4.16s eta=28:56:38 +2026-09-21 11:22:33 | INFO | train rank=0 | step=25000/50000 progress=50.0% loss=0.411319 lr=5.412897e-05 lr_backbone=5.412897e-05 elapsed=28:55:56 avg_step=4.17s recent_step=4.19s eta=28:55:56 +2026-09-21 11:23:14 | INFO | train rank=0 | step=25010/50000 progress=50.0% loss=0.316416 lr=5.409601e-05 lr_backbone=5.409601e-05 elapsed=28:56:38 avg_step=4.17s recent_step=4.11s eta=28:55:14 +2026-09-21 11:23:55 | INFO | train rank=0 | step=25020/50000 progress=50.0% loss=0.315140 lr=5.406305e-05 lr_backbone=5.406305e-05 elapsed=28:57:19 avg_step=4.17s recent_step=4.13s eta=28:54:32 +2026-09-21 11:24:37 | INFO | train rank=0 | step=25030/50000 progress=50.1% loss=0.341715 lr=5.403009e-05 lr_backbone=5.403009e-05 elapsed=28:58:00 avg_step=4.17s recent_step=4.12s eta=28:53:50 +2026-09-21 11:25:18 | INFO | train rank=0 | step=25040/50000 progress=50.1% loss=0.362767 lr=5.399713e-05 lr_backbone=5.399713e-05 elapsed=28:58:42 avg_step=4.17s recent_step=4.16s eta=28:53:08 +2026-09-21 11:26:00 | INFO | train rank=0 | step=25050/50000 progress=50.1% loss=0.381536 lr=5.396416e-05 lr_backbone=5.396416e-05 elapsed=28:59:23 avg_step=4.17s recent_step=4.15s eta=28:52:26 +2026-09-21 11:26:41 | INFO | train rank=0 | step=25060/50000 progress=50.1% loss=0.307884 lr=5.393120e-05 lr_backbone=5.393120e-05 elapsed=29:00:04 avg_step=4.17s recent_step=4.15s eta=28:51:45 +2026-09-21 11:27:22 | INFO | train rank=0 | step=25070/50000 progress=50.1% loss=0.315887 lr=5.389823e-05 lr_backbone=5.389823e-05 elapsed=29:00:46 avg_step=4.17s recent_step=4.10s eta=28:51:02 +2026-09-21 11:28:03 | INFO | train rank=0 | step=25080/50000 progress=50.2% loss=0.311744 lr=5.386526e-05 lr_backbone=5.386526e-05 elapsed=29:01:27 avg_step=4.17s recent_step=4.12s eta=28:50:20 +2026-09-21 11:28:45 | INFO | train rank=0 | step=25090/50000 progress=50.2% loss=0.369625 lr=5.383229e-05 lr_backbone=5.383229e-05 elapsed=29:02:08 avg_step=4.17s recent_step=4.14s eta=28:49:38 +2026-09-21 11:29:26 | INFO | train rank=0 | step=25100/50000 progress=50.2% loss=0.325791 lr=5.379931e-05 lr_backbone=5.379931e-05 elapsed=29:02:50 avg_step=4.17s recent_step=4.15s eta=28:48:56 +2026-09-21 11:30:08 | INFO | train rank=0 | step=25110/50000 progress=50.2% loss=0.372024 lr=5.376634e-05 lr_backbone=5.376634e-05 elapsed=29:03:31 avg_step=4.17s recent_step=4.16s eta=28:48:15 +2026-09-21 11:30:49 | INFO | train rank=0 | step=25120/50000 progress=50.2% loss=0.365457 lr=5.373336e-05 lr_backbone=5.373336e-05 elapsed=29:04:13 avg_step=4.17s recent_step=4.15s eta=28:47:33 +2026-09-21 11:31:30 | INFO | train rank=0 | step=25130/50000 progress=50.3% loss=0.339718 lr=5.370039e-05 lr_backbone=5.370039e-05 elapsed=29:04:54 avg_step=4.17s recent_step=4.09s eta=28:46:50 +2026-09-21 11:32:12 | INFO | train rank=0 | step=25140/50000 progress=50.3% loss=0.322163 lr=5.366741e-05 lr_backbone=5.366741e-05 elapsed=29:05:35 avg_step=4.17s recent_step=4.13s eta=28:46:08 +2026-09-21 11:32:53 | INFO | train rank=0 | step=25150/50000 progress=50.3% loss=0.397751 lr=5.363443e-05 lr_backbone=5.363443e-05 elapsed=29:06:17 avg_step=4.17s recent_step=4.16s eta=28:45:27 +2026-09-21 11:33:35 | INFO | train rank=0 | step=25160/50000 progress=50.3% loss=0.362778 lr=5.360144e-05 lr_backbone=5.360144e-05 elapsed=29:06:58 avg_step=4.17s recent_step=4.14s eta=28:44:45 +2026-09-21 11:34:16 | INFO | train rank=0 | step=25170/50000 progress=50.3% loss=0.471050 lr=5.356846e-05 lr_backbone=5.356846e-05 elapsed=29:07:40 avg_step=4.17s recent_step=4.16s eta=28:44:03 +2026-09-21 11:34:58 | INFO | train rank=0 | step=25180/50000 progress=50.4% loss=0.365960 lr=5.353547e-05 lr_backbone=5.353547e-05 elapsed=29:08:21 avg_step=4.17s recent_step=4.16s eta=28:43:21 +2026-09-21 11:35:39 | INFO | train rank=0 | step=25190/50000 progress=50.4% loss=0.324632 lr=5.350249e-05 lr_backbone=5.350249e-05 elapsed=29:09:02 avg_step=4.17s recent_step=4.09s eta=28:42:39 +2026-09-21 11:36:20 | INFO | train rank=0 | step=25200/50000 progress=50.4% loss=0.317459 lr=5.346950e-05 lr_backbone=5.346950e-05 elapsed=29:09:43 avg_step=4.17s recent_step=4.13s eta=28:41:57 +2026-09-21 11:37:01 | INFO | train rank=0 | step=25210/50000 progress=50.4% loss=0.296993 lr=5.343651e-05 lr_backbone=5.343651e-05 elapsed=29:10:25 avg_step=4.17s recent_step=4.15s eta=28:41:15 +2026-09-21 11:37:43 | INFO | train rank=0 | step=25220/50000 progress=50.4% loss=0.379913 lr=5.340351e-05 lr_backbone=5.340351e-05 elapsed=29:11:06 avg_step=4.17s recent_step=4.15s eta=28:40:33 +2026-09-21 11:38:24 | INFO | train rank=0 | step=25230/50000 progress=50.5% loss=0.381757 lr=5.337052e-05 lr_backbone=5.337052e-05 elapsed=29:11:48 avg_step=4.17s recent_step=4.13s eta=28:39:51 +2026-09-21 11:39:06 | INFO | train rank=0 | step=25240/50000 progress=50.5% loss=0.320971 lr=5.333753e-05 lr_backbone=5.333753e-05 elapsed=29:12:29 avg_step=4.17s recent_step=4.17s eta=28:39:10 +2026-09-21 11:39:47 | INFO | train rank=0 | step=25250/50000 progress=50.5% loss=0.349279 lr=5.330453e-05 lr_backbone=5.330453e-05 elapsed=29:13:10 avg_step=4.17s recent_step=4.09s eta=28:38:27 +2026-09-21 11:40:28 | INFO | train rank=0 | step=25260/50000 progress=50.5% loss=0.337802 lr=5.327153e-05 lr_backbone=5.327153e-05 elapsed=29:13:51 avg_step=4.17s recent_step=4.13s eta=28:37:45 +2026-09-21 11:41:09 | INFO | train rank=0 | step=25270/50000 progress=50.5% loss=0.365156 lr=5.323853e-05 lr_backbone=5.323853e-05 elapsed=29:14:33 avg_step=4.17s recent_step=4.14s eta=28:37:03 +2026-09-21 11:41:51 | INFO | train rank=0 | step=25280/50000 progress=50.6% loss=0.330526 lr=5.320553e-05 lr_backbone=5.320553e-05 elapsed=29:15:14 avg_step=4.17s recent_step=4.16s eta=28:36:22 +2026-09-21 11:42:33 | INFO | train rank=0 | step=25290/50000 progress=50.6% loss=0.342615 lr=5.317253e-05 lr_backbone=5.317253e-05 elapsed=29:15:56 avg_step=4.17s recent_step=4.14s eta=28:35:40 +2026-09-21 11:43:14 | INFO | train rank=0 | step=25300/50000 progress=50.6% loss=0.379028 lr=5.313953e-05 lr_backbone=5.313953e-05 elapsed=29:16:37 avg_step=4.17s recent_step=4.15s eta=28:34:58 +2026-09-21 11:43:55 | INFO | train rank=0 | step=25310/50000 progress=50.6% loss=0.411823 lr=5.310652e-05 lr_backbone=5.310652e-05 elapsed=29:17:18 avg_step=4.17s recent_step=4.09s eta=28:34:15 +2026-09-21 11:44:36 | INFO | train rank=0 | step=25320/50000 progress=50.6% loss=0.373439 lr=5.307351e-05 lr_backbone=5.307351e-05 elapsed=29:18:00 avg_step=4.17s recent_step=4.15s eta=28:33:34 +2026-09-21 11:45:18 | INFO | train rank=0 | step=25330/50000 progress=50.7% loss=0.311557 lr=5.304051e-05 lr_backbone=5.304051e-05 elapsed=29:18:41 avg_step=4.17s recent_step=4.17s eta=28:32:52 +2026-09-21 11:46:00 | INFO | train rank=0 | step=25340/50000 progress=50.7% loss=0.330862 lr=5.300750e-05 lr_backbone=5.300750e-05 elapsed=29:19:23 avg_step=4.17s recent_step=4.18s eta=28:32:10 +2026-09-21 11:46:41 | INFO | train rank=0 | step=25350/50000 progress=50.7% loss=0.423734 lr=5.297449e-05 lr_backbone=5.297449e-05 elapsed=29:20:04 avg_step=4.17s recent_step=4.13s eta=28:31:28 +2026-09-21 11:47:23 | INFO | train rank=0 | step=25360/50000 progress=50.7% loss=0.518738 lr=5.294148e-05 lr_backbone=5.294148e-05 elapsed=29:20:46 avg_step=4.17s recent_step=4.16s eta=28:30:47 +2026-09-21 11:48:04 | INFO | train rank=0 | step=25370/50000 progress=50.7% loss=0.309904 lr=5.290846e-05 lr_backbone=5.290846e-05 elapsed=29:21:27 avg_step=4.17s recent_step=4.11s eta=28:30:04 +2026-09-21 11:48:45 | INFO | train rank=0 | step=25380/50000 progress=50.8% loss=0.385510 lr=5.287545e-05 lr_backbone=5.287545e-05 elapsed=29:22:08 avg_step=4.17s recent_step=4.12s eta=28:29:22 +2026-09-21 11:49:27 | INFO | train rank=0 | step=25390/50000 progress=50.8% loss=0.333779 lr=5.284243e-05 lr_backbone=5.284243e-05 elapsed=29:22:50 avg_step=4.17s recent_step=4.16s eta=28:28:41 +2026-09-21 11:50:08 | INFO | train rank=0 | step=25400/50000 progress=50.8% loss=0.334640 lr=5.280942e-05 lr_backbone=5.280942e-05 elapsed=29:23:31 avg_step=4.17s recent_step=4.15s eta=28:27:59 +2026-09-21 11:50:50 | INFO | train rank=0 | step=25410/50000 progress=50.8% loss=0.392368 lr=5.277640e-05 lr_backbone=5.277640e-05 elapsed=29:24:13 avg_step=4.17s recent_step=4.15s eta=28:27:17 +2026-09-21 11:51:31 | INFO | train rank=0 | step=25420/50000 progress=50.8% loss=0.338953 lr=5.274338e-05 lr_backbone=5.274338e-05 elapsed=29:24:54 avg_step=4.17s recent_step=4.15s eta=28:26:35 +2026-09-21 11:52:12 | INFO | train rank=0 | step=25430/50000 progress=50.9% loss=0.356891 lr=5.271036e-05 lr_backbone=5.271036e-05 elapsed=29:25:35 avg_step=4.17s recent_step=4.10s eta=28:25:53 +2026-09-21 11:52:53 | INFO | train rank=0 | step=25440/50000 progress=50.9% loss=0.324938 lr=5.267734e-05 lr_backbone=5.267734e-05 elapsed=29:26:17 avg_step=4.17s recent_step=4.14s eta=28:25:11 +2026-09-21 11:53:35 | INFO | train rank=0 | step=25450/50000 progress=50.9% loss=0.403422 lr=5.264432e-05 lr_backbone=5.264432e-05 elapsed=29:26:58 avg_step=4.17s recent_step=4.14s eta=28:24:29 +2026-09-21 11:54:16 | INFO | train rank=0 | step=25460/50000 progress=50.9% loss=0.403855 lr=5.261129e-05 lr_backbone=5.261129e-05 elapsed=29:27:40 avg_step=4.17s recent_step=4.15s eta=28:23:47 +2026-09-21 11:54:58 | INFO | train rank=0 | step=25470/50000 progress=50.9% loss=0.321157 lr=5.257827e-05 lr_backbone=5.257827e-05 elapsed=29:28:21 avg_step=4.17s recent_step=4.13s eta=28:23:05 +2026-09-21 11:55:39 | INFO | train rank=0 | step=25480/50000 progress=51.0% loss=0.382348 lr=5.254524e-05 lr_backbone=5.254524e-05 elapsed=29:29:03 avg_step=4.17s recent_step=4.17s eta=28:22:23 +2026-09-21 11:56:20 | INFO | train rank=0 | step=25490/50000 progress=51.0% loss=0.373474 lr=5.251222e-05 lr_backbone=5.251222e-05 elapsed=29:29:44 avg_step=4.17s recent_step=4.11s eta=28:21:41 +2026-09-21 11:57:02 | INFO | train rank=0 | step=25500/50000 progress=51.0% loss=0.334380 lr=5.247919e-05 lr_backbone=5.247919e-05 elapsed=29:30:25 avg_step=4.17s recent_step=4.13s eta=28:20:59 +2026-09-21 11:57:43 | INFO | train rank=0 | step=25510/50000 progress=51.0% loss=0.350826 lr=5.244616e-05 lr_backbone=5.244616e-05 elapsed=29:31:07 avg_step=4.17s recent_step=4.16s eta=28:20:18 +2026-09-21 11:58:25 | INFO | train rank=0 | step=25520/50000 progress=51.0% loss=0.418767 lr=5.241313e-05 lr_backbone=5.241313e-05 elapsed=29:31:48 avg_step=4.17s recent_step=4.17s eta=28:19:36 +2026-09-21 11:59:06 | INFO | train rank=0 | step=25530/50000 progress=51.1% loss=0.301541 lr=5.238010e-05 lr_backbone=5.238010e-05 elapsed=29:32:30 avg_step=4.17s recent_step=4.15s eta=28:18:54 +2026-09-21 11:59:48 | INFO | train rank=0 | step=25540/50000 progress=51.1% loss=0.322376 lr=5.234706e-05 lr_backbone=5.234706e-05 elapsed=29:33:11 avg_step=4.17s recent_step=4.14s eta=28:18:12 +2026-09-21 12:00:29 | INFO | train rank=0 | step=25550/50000 progress=51.1% loss=0.447679 lr=5.231403e-05 lr_backbone=5.231403e-05 elapsed=29:33:52 avg_step=4.17s recent_step=4.11s eta=28:17:30 +2026-09-21 12:01:10 | INFO | train rank=0 | step=25560/50000 progress=51.1% loss=0.392620 lr=5.228100e-05 lr_backbone=5.228100e-05 elapsed=29:34:34 avg_step=4.17s recent_step=4.12s eta=28:16:48 +2026-09-21 12:01:52 | INFO | train rank=0 | step=25570/50000 progress=51.1% loss=0.312477 lr=5.224796e-05 lr_backbone=5.224796e-05 elapsed=29:35:15 avg_step=4.17s recent_step=4.15s eta=28:16:06 +2026-09-21 12:02:33 | INFO | train rank=0 | step=25580/50000 progress=51.2% loss=0.313452 lr=5.221492e-05 lr_backbone=5.221492e-05 elapsed=29:35:56 avg_step=4.17s recent_step=4.14s eta=28:15:24 +2026-09-21 12:03:14 | INFO | train rank=0 | step=25590/50000 progress=51.2% loss=0.298867 lr=5.218189e-05 lr_backbone=5.218189e-05 elapsed=29:36:38 avg_step=4.17s recent_step=4.13s eta=28:14:42 +2026-09-21 12:03:56 | INFO | train rank=0 | step=25600/50000 progress=51.2% loss=0.372103 lr=5.214885e-05 lr_backbone=5.214885e-05 elapsed=29:37:19 avg_step=4.17s recent_step=4.14s eta=28:14:00 +2026-09-21 12:04:37 | INFO | train rank=0 | step=25610/50000 progress=51.2% loss=0.310411 lr=5.211581e-05 lr_backbone=5.211581e-05 elapsed=29:38:00 avg_step=4.17s recent_step=4.12s eta=28:13:18 +2026-09-21 12:05:18 | INFO | train rank=0 | step=25620/50000 progress=51.2% loss=0.311486 lr=5.208277e-05 lr_backbone=5.208277e-05 elapsed=29:38:42 avg_step=4.17s recent_step=4.13s eta=28:12:36 +2026-09-21 12:06:00 | INFO | train rank=0 | step=25630/50000 progress=51.3% loss=0.358607 lr=5.204973e-05 lr_backbone=5.204973e-05 elapsed=29:39:23 avg_step=4.17s recent_step=4.16s eta=28:11:55 +2026-09-21 12:06:42 | INFO | train rank=0 | step=25640/50000 progress=51.3% loss=0.286451 lr=5.201669e-05 lr_backbone=5.201669e-05 elapsed=29:40:05 avg_step=4.17s recent_step=4.17s eta=28:11:13 +2026-09-21 12:07:23 | INFO | train rank=0 | step=25650/50000 progress=51.3% loss=0.311205 lr=5.198364e-05 lr_backbone=5.198364e-05 elapsed=29:40:47 avg_step=4.17s recent_step=4.15s eta=28:10:31 +2026-09-21 12:08:04 | INFO | train rank=0 | step=25660/50000 progress=51.3% loss=0.359382 lr=5.195060e-05 lr_backbone=5.195060e-05 elapsed=29:41:28 avg_step=4.17s recent_step=4.11s eta=28:09:49 +2026-09-21 12:08:45 | INFO | train rank=0 | step=25670/50000 progress=51.3% loss=0.312934 lr=5.191755e-05 lr_backbone=5.191755e-05 elapsed=29:42:09 avg_step=4.17s recent_step=4.11s eta=28:09:07 +2026-09-21 12:09:27 | INFO | train rank=0 | step=25680/50000 progress=51.4% loss=0.328943 lr=5.188451e-05 lr_backbone=5.188451e-05 elapsed=29:42:50 avg_step=4.17s recent_step=4.13s eta=28:08:25 +2026-09-21 12:10:08 | INFO | train rank=0 | step=25690/50000 progress=51.4% loss=0.295844 lr=5.185146e-05 lr_backbone=5.185146e-05 elapsed=29:43:32 avg_step=4.17s recent_step=4.16s eta=28:07:43 +2026-09-21 12:10:50 | INFO | train rank=0 | step=25700/50000 progress=51.4% loss=0.388406 lr=5.181842e-05 lr_backbone=5.181842e-05 elapsed=29:44:13 avg_step=4.17s recent_step=4.17s eta=28:07:02 +2026-09-21 12:11:31 | INFO | train rank=0 | step=25710/50000 progress=51.4% loss=0.340444 lr=5.178537e-05 lr_backbone=5.178537e-05 elapsed=29:44:55 avg_step=4.17s recent_step=4.14s eta=28:06:20 +2026-09-21 12:12:13 | INFO | train rank=0 | step=25720/50000 progress=51.4% loss=0.306540 lr=5.175232e-05 lr_backbone=5.175232e-05 elapsed=29:45:36 avg_step=4.17s recent_step=4.13s eta=28:05:38 +2026-09-21 12:12:54 | INFO | train rank=0 | step=25730/50000 progress=51.5% loss=0.289063 lr=5.171927e-05 lr_backbone=5.171927e-05 elapsed=29:46:17 avg_step=4.17s recent_step=4.11s eta=28:04:56 +2026-09-21 12:13:35 | INFO | train rank=0 | step=25740/50000 progress=51.5% loss=0.404801 lr=5.168622e-05 lr_backbone=5.168622e-05 elapsed=29:46:59 avg_step=4.17s recent_step=4.13s eta=28:04:14 +2026-09-21 12:14:17 | INFO | train rank=0 | step=25750/50000 progress=51.5% loss=0.360414 lr=5.165317e-05 lr_backbone=5.165317e-05 elapsed=29:47:40 avg_step=4.17s recent_step=4.14s eta=28:03:32 +2026-09-21 12:14:58 | INFO | train rank=0 | step=25760/50000 progress=51.5% loss=0.318503 lr=5.162012e-05 lr_backbone=5.162012e-05 elapsed=29:48:21 avg_step=4.17s recent_step=4.14s eta=28:02:50 +2026-09-21 12:15:40 | INFO | train rank=0 | step=25770/50000 progress=51.5% loss=0.395332 lr=5.158706e-05 lr_backbone=5.158706e-05 elapsed=29:49:03 avg_step=4.17s recent_step=4.16s eta=28:02:08 +2026-09-21 12:16:21 | INFO | train rank=0 | step=25780/50000 progress=51.6% loss=0.350382 lr=5.155401e-05 lr_backbone=5.155401e-05 elapsed=29:49:44 avg_step=4.17s recent_step=4.14s eta=28:01:26 +2026-09-21 12:17:02 | INFO | train rank=0 | step=25790/50000 progress=51.6% loss=0.409609 lr=5.152096e-05 lr_backbone=5.152096e-05 elapsed=29:50:25 avg_step=4.17s recent_step=4.10s eta=28:00:44 +2026-09-21 12:17:43 | INFO | train rank=0 | step=25800/50000 progress=51.6% loss=0.378196 lr=5.148790e-05 lr_backbone=5.148790e-05 elapsed=29:51:07 avg_step=4.17s recent_step=4.13s eta=28:00:02 +2026-09-21 12:18:25 | INFO | train rank=0 | step=25810/50000 progress=51.6% loss=0.320456 lr=5.145485e-05 lr_backbone=5.145485e-05 elapsed=29:51:48 avg_step=4.17s recent_step=4.16s eta=27:59:20 +2026-09-21 12:19:07 | INFO | train rank=0 | step=25820/50000 progress=51.6% loss=0.320497 lr=5.142179e-05 lr_backbone=5.142179e-05 elapsed=29:52:30 avg_step=4.17s recent_step=4.17s eta=27:58:39 +2026-09-21 12:19:48 | INFO | train rank=0 | step=25830/50000 progress=51.7% loss=0.335410 lr=5.138874e-05 lr_backbone=5.138874e-05 elapsed=29:53:11 avg_step=4.17s recent_step=4.15s eta=27:57:57 +2026-09-21 12:20:29 | INFO | train rank=0 | step=25840/50000 progress=51.7% loss=0.393844 lr=5.135568e-05 lr_backbone=5.135568e-05 elapsed=29:53:52 avg_step=4.17s recent_step=4.10s eta=27:57:15 +2026-09-21 12:21:10 | INFO | train rank=0 | step=25850/50000 progress=51.7% loss=0.410274 lr=5.132262e-05 lr_backbone=5.132262e-05 elapsed=29:54:33 avg_step=4.17s recent_step=4.11s eta=27:56:32 +2026-09-21 12:21:51 | INFO | train rank=0 | step=25860/50000 progress=51.7% loss=0.309903 lr=5.128956e-05 lr_backbone=5.128956e-05 elapsed=29:55:15 avg_step=4.17s recent_step=4.13s eta=27:55:50 +2026-09-21 12:22:33 | INFO | train rank=0 | step=25870/50000 progress=51.7% loss=0.354945 lr=5.125650e-05 lr_backbone=5.125650e-05 elapsed=29:55:56 avg_step=4.17s recent_step=4.15s eta=27:55:09 +2026-09-21 12:23:15 | INFO | train rank=0 | step=25880/50000 progress=51.8% loss=0.337663 lr=5.122345e-05 lr_backbone=5.122345e-05 elapsed=29:56:38 avg_step=4.17s recent_step=4.16s eta=27:54:27 +2026-09-21 12:23:56 | INFO | train rank=0 | step=25890/50000 progress=51.8% loss=0.474878 lr=5.119039e-05 lr_backbone=5.119039e-05 elapsed=29:57:20 avg_step=4.17s recent_step=4.16s eta=27:53:45 +2026-09-21 12:24:37 | INFO | train rank=0 | step=25900/50000 progress=51.8% loss=0.263423 lr=5.115733e-05 lr_backbone=5.115733e-05 elapsed=29:58:01 avg_step=4.17s recent_step=4.13s eta=27:53:03 +2026-09-21 12:25:19 | INFO | train rank=0 | step=25910/50000 progress=51.8% loss=0.319040 lr=5.112426e-05 lr_backbone=5.112426e-05 elapsed=29:58:42 avg_step=4.17s recent_step=4.11s eta=27:52:21 +2026-09-21 12:26:00 | INFO | train rank=0 | step=25920/50000 progress=51.8% loss=0.347392 lr=5.109120e-05 lr_backbone=5.109120e-05 elapsed=29:59:23 avg_step=4.17s recent_step=4.12s eta=27:51:39 +2026-09-21 12:26:41 | INFO | train rank=0 | step=25930/50000 progress=51.9% loss=0.390944 lr=5.105814e-05 lr_backbone=5.105814e-05 elapsed=30:00:05 avg_step=4.17s recent_step=4.17s eta=27:50:57 +2026-09-21 12:27:23 | INFO | train rank=0 | step=25940/50000 progress=51.9% loss=0.321553 lr=5.102508e-05 lr_backbone=5.102508e-05 elapsed=30:00:46 avg_step=4.17s recent_step=4.15s eta=27:50:16 +2026-09-21 12:28:05 | INFO | train rank=0 | step=25950/50000 progress=51.9% loss=0.316110 lr=5.099202e-05 lr_backbone=5.099202e-05 elapsed=30:01:28 avg_step=4.17s recent_step=4.16s eta=27:49:34 +2026-09-21 12:28:46 | INFO | train rank=0 | step=25960/50000 progress=51.9% loss=0.333001 lr=5.095895e-05 lr_backbone=5.095895e-05 elapsed=30:02:09 avg_step=4.17s recent_step=4.15s eta=27:48:52 +2026-09-21 12:29:27 | INFO | train rank=0 | step=25970/50000 progress=51.9% loss=0.318217 lr=5.092589e-05 lr_backbone=5.092589e-05 elapsed=30:02:50 avg_step=4.17s recent_step=4.09s eta=27:48:10 +2026-09-21 12:30:08 | INFO | train rank=0 | step=25980/50000 progress=52.0% loss=0.339099 lr=5.089283e-05 lr_backbone=5.089283e-05 elapsed=30:03:31 avg_step=4.17s recent_step=4.12s eta=27:47:28 +2026-09-21 12:30:50 | INFO | train rank=0 | step=25990/50000 progress=52.0% loss=0.342163 lr=5.085976e-05 lr_backbone=5.085976e-05 elapsed=30:04:13 avg_step=4.17s recent_step=4.15s eta=27:46:46 +2026-09-21 12:31:31 | INFO | train rank=0 | step=26000/50000 progress=52.0% loss=0.308290 lr=5.082670e-05 lr_backbone=5.082670e-05 elapsed=30:04:54 avg_step=4.17s recent_step=4.13s eta=27:46:04 +2026-09-21 12:32:12 | INFO | train rank=0 | step=26010/50000 progress=52.0% loss=0.374845 lr=5.079363e-05 lr_backbone=5.079363e-05 elapsed=30:05:36 avg_step=4.17s recent_step=4.13s eta=27:45:22 +2026-09-21 12:32:54 | INFO | train rank=0 | step=26020/50000 progress=52.0% loss=0.333203 lr=5.076057e-05 lr_backbone=5.076057e-05 elapsed=30:06:17 avg_step=4.17s recent_step=4.16s eta=27:44:40 +2026-09-21 12:33:35 | INFO | train rank=0 | step=26030/50000 progress=52.1% loss=0.341973 lr=5.072750e-05 lr_backbone=5.072750e-05 elapsed=30:06:58 avg_step=4.17s recent_step=4.08s eta=27:43:58 +2026-09-21 12:34:16 | INFO | train rank=0 | step=26040/50000 progress=52.1% loss=0.312495 lr=5.069443e-05 lr_backbone=5.069443e-05 elapsed=30:07:39 avg_step=4.17s recent_step=4.12s eta=27:43:16 +2026-09-21 12:34:57 | INFO | train rank=0 | step=26050/50000 progress=52.1% loss=0.317575 lr=5.066137e-05 lr_backbone=5.066137e-05 elapsed=30:08:20 avg_step=4.17s recent_step=4.13s eta=27:42:34 +2026-09-21 12:35:38 | INFO | train rank=0 | step=26060/50000 progress=52.1% loss=0.332576 lr=5.062830e-05 lr_backbone=5.062830e-05 elapsed=30:09:02 avg_step=4.17s recent_step=4.13s eta=27:41:52 +2026-09-21 12:36:20 | INFO | train rank=0 | step=26070/50000 progress=52.1% loss=0.300843 lr=5.059524e-05 lr_backbone=5.059524e-05 elapsed=30:09:43 avg_step=4.17s recent_step=4.13s eta=27:41:10 +2026-09-21 12:37:01 | INFO | train rank=0 | step=26080/50000 progress=52.2% loss=0.312773 lr=5.056217e-05 lr_backbone=5.056217e-05 elapsed=30:10:24 avg_step=4.17s recent_step=4.13s eta=27:40:28 +2026-09-21 12:37:42 | INFO | train rank=0 | step=26090/50000 progress=52.2% loss=0.320248 lr=5.052910e-05 lr_backbone=5.052910e-05 elapsed=30:11:05 avg_step=4.17s recent_step=4.08s eta=27:39:45 +2026-09-21 12:38:23 | INFO | train rank=0 | step=26100/50000 progress=52.2% loss=0.410207 lr=5.049603e-05 lr_backbone=5.049603e-05 elapsed=30:11:46 avg_step=4.17s recent_step=4.09s eta=27:39:03 +2026-09-21 12:39:04 | INFO | train rank=0 | step=26110/50000 progress=52.2% loss=0.293813 lr=5.046296e-05 lr_backbone=5.046296e-05 elapsed=30:12:28 avg_step=4.17s recent_step=4.15s eta=27:38:21 +2026-09-21 12:39:46 | INFO | train rank=0 | step=26120/50000 progress=52.2% loss=0.313661 lr=5.042990e-05 lr_backbone=5.042990e-05 elapsed=30:13:09 avg_step=4.16s recent_step=4.13s eta=27:37:39 +2026-09-21 12:40:27 | INFO | train rank=0 | step=26130/50000 progress=52.3% loss=0.325898 lr=5.039683e-05 lr_backbone=5.039683e-05 elapsed=30:13:51 avg_step=4.16s recent_step=4.15s eta=27:36:58 +2026-09-21 12:41:09 | INFO | train rank=0 | step=26140/50000 progress=52.3% loss=0.308782 lr=5.036376e-05 lr_backbone=5.036376e-05 elapsed=30:14:32 avg_step=4.16s recent_step=4.15s eta=27:36:16 +2026-09-21 12:41:49 | INFO | train rank=0 | step=26150/50000 progress=52.3% loss=0.370671 lr=5.033069e-05 lr_backbone=5.033069e-05 elapsed=30:15:13 avg_step=4.16s recent_step=4.08s eta=27:35:33 +2026-09-21 12:42:30 | INFO | train rank=0 | step=26160/50000 progress=52.3% loss=0.283201 lr=5.029762e-05 lr_backbone=5.029762e-05 elapsed=30:15:54 avg_step=4.16s recent_step=4.10s eta=27:34:51 +2026-09-21 12:43:12 | INFO | train rank=0 | step=26170/50000 progress=52.3% loss=0.352460 lr=5.026455e-05 lr_backbone=5.026455e-05 elapsed=30:16:35 avg_step=4.16s recent_step=4.13s eta=27:34:09 +2026-09-21 12:43:53 | INFO | train rank=0 | step=26180/50000 progress=52.4% loss=0.342584 lr=5.023148e-05 lr_backbone=5.023148e-05 elapsed=30:17:16 avg_step=4.16s recent_step=4.14s eta=27:33:27 +2026-09-21 12:44:35 | INFO | train rank=0 | step=26190/50000 progress=52.4% loss=0.328746 lr=5.019842e-05 lr_backbone=5.019842e-05 elapsed=30:17:58 avg_step=4.16s recent_step=4.15s eta=27:32:46 +2026-09-21 12:45:16 | INFO | train rank=0 | step=26200/50000 progress=52.4% loss=0.346896 lr=5.016535e-05 lr_backbone=5.016535e-05 elapsed=30:18:39 avg_step=4.16s recent_step=4.13s eta=27:32:04 +2026-09-21 12:45:57 | INFO | train rank=0 | step=26210/50000 progress=52.4% loss=0.383843 lr=5.013228e-05 lr_backbone=5.013228e-05 elapsed=30:19:20 avg_step=4.16s recent_step=4.07s eta=27:31:21 +2026-09-21 12:46:38 | INFO | train rank=0 | step=26220/50000 progress=52.4% loss=0.346655 lr=5.009921e-05 lr_backbone=5.009921e-05 elapsed=30:20:01 avg_step=4.16s recent_step=4.12s eta=27:30:39 +2026-09-21 12:47:19 | INFO | train rank=0 | step=26230/50000 progress=52.5% loss=0.325556 lr=5.006614e-05 lr_backbone=5.006614e-05 elapsed=30:20:42 avg_step=4.16s recent_step=4.12s eta=27:29:57 +2026-09-21 12:48:00 | INFO | train rank=0 | step=26240/50000 progress=52.5% loss=0.343041 lr=5.003307e-05 lr_backbone=5.003307e-05 elapsed=30:21:24 avg_step=4.16s recent_step=4.14s eta=27:29:15 +2026-09-21 12:48:42 | INFO | train rank=0 | step=26250/50000 progress=52.5% loss=0.336840 lr=5.000000e-05 lr_backbone=5.000000e-05 elapsed=30:22:05 avg_step=4.16s recent_step=4.13s eta=27:28:33 +2026-09-21 12:49:23 | INFO | train rank=0 | step=26260/50000 progress=52.5% loss=0.450817 lr=4.996693e-05 lr_backbone=4.996693e-05 elapsed=30:22:46 avg_step=4.16s recent_step=4.14s eta=27:27:51 +2026-09-21 12:50:04 | INFO | train rank=0 | step=26270/50000 progress=52.5% loss=0.324746 lr=4.993386e-05 lr_backbone=4.993386e-05 elapsed=30:23:27 avg_step=4.16s recent_step=4.09s eta=27:27:09 +2026-09-21 12:50:45 | INFO | train rank=0 | step=26280/50000 progress=52.6% loss=0.330837 lr=4.990079e-05 lr_backbone=4.990079e-05 elapsed=30:24:08 avg_step=4.16s recent_step=4.10s eta=27:26:27 +2026-09-21 12:51:26 | INFO | train rank=0 | step=26290/50000 progress=52.6% loss=0.400041 lr=4.986772e-05 lr_backbone=4.986772e-05 elapsed=30:24:50 avg_step=4.16s recent_step=4.15s eta=27:25:45 +2026-09-21 12:52:08 | INFO | train rank=0 | step=26300/50000 progress=52.6% loss=0.315653 lr=4.983465e-05 lr_backbone=4.983465e-05 elapsed=30:25:32 avg_step=4.16s recent_step=4.19s eta=27:25:03 +2026-09-21 12:52:50 | INFO | train rank=0 | step=26310/50000 progress=52.6% loss=0.355671 lr=4.980158e-05 lr_backbone=4.980158e-05 elapsed=30:26:13 avg_step=4.16s recent_step=4.14s eta=27:24:22 +2026-09-21 12:53:31 | INFO | train rank=0 | step=26320/50000 progress=52.6% loss=0.347670 lr=4.976852e-05 lr_backbone=4.976852e-05 elapsed=30:26:55 avg_step=4.16s recent_step=4.16s eta=27:23:40 +2026-09-21 12:54:12 | INFO | train rank=0 | step=26330/50000 progress=52.7% loss=0.362546 lr=4.973545e-05 lr_backbone=4.973545e-05 elapsed=30:27:36 avg_step=4.16s recent_step=4.10s eta=27:22:58 +2026-09-21 12:54:54 | INFO | train rank=0 | step=26340/50000 progress=52.7% loss=0.387874 lr=4.970238e-05 lr_backbone=4.970238e-05 elapsed=30:28:17 avg_step=4.16s recent_step=4.12s eta=27:22:16 +2026-09-21 12:55:35 | INFO | train rank=0 | step=26350/50000 progress=52.7% loss=0.445973 lr=4.966931e-05 lr_backbone=4.966931e-05 elapsed=30:28:58 avg_step=4.16s recent_step=4.13s eta=27:21:34 +2026-09-21 12:56:17 | INFO | train rank=0 | step=26360/50000 progress=52.7% loss=0.332101 lr=4.963624e-05 lr_backbone=4.963624e-05 elapsed=30:29:40 avg_step=4.16s recent_step=4.17s eta=27:20:52 +2026-09-21 12:56:58 | INFO | train rank=0 | step=26370/50000 progress=52.7% loss=0.306162 lr=4.960317e-05 lr_backbone=4.960317e-05 elapsed=30:30:21 avg_step=4.16s recent_step=4.15s eta=27:20:10 +2026-09-21 12:57:40 | INFO | train rank=0 | step=26380/50000 progress=52.8% loss=0.308486 lr=4.957010e-05 lr_backbone=4.957010e-05 elapsed=30:31:03 avg_step=4.16s recent_step=4.15s eta=27:19:29 +2026-09-21 12:58:21 | INFO | train rank=0 | step=26390/50000 progress=52.8% loss=0.331595 lr=4.953704e-05 lr_backbone=4.953704e-05 elapsed=30:31:44 avg_step=4.16s recent_step=4.09s eta=27:18:46 +2026-09-21 12:59:02 | INFO | train rank=0 | step=26400/50000 progress=52.8% loss=0.326588 lr=4.950397e-05 lr_backbone=4.950397e-05 elapsed=30:32:25 avg_step=4.16s recent_step=4.14s eta=27:18:04 +2026-09-21 12:59:43 | INFO | train rank=0 | step=26410/50000 progress=52.8% loss=0.301686 lr=4.947090e-05 lr_backbone=4.947090e-05 elapsed=30:33:06 avg_step=4.16s recent_step=4.09s eta=27:17:22 +2026-09-21 13:00:25 | INFO | train rank=0 | step=26420/50000 progress=52.8% loss=0.333078 lr=4.943783e-05 lr_backbone=4.943783e-05 elapsed=30:33:48 avg_step=4.16s recent_step=4.16s eta=27:16:40 +2026-09-21 13:01:06 | INFO | train rank=0 | step=26430/50000 progress=52.9% loss=0.349676 lr=4.940476e-05 lr_backbone=4.940476e-05 elapsed=30:34:29 avg_step=4.16s recent_step=4.14s eta=27:15:59 +2026-09-21 13:01:48 | INFO | train rank=0 | step=26440/50000 progress=52.9% loss=0.330488 lr=4.937170e-05 lr_backbone=4.937170e-05 elapsed=30:35:11 avg_step=4.16s recent_step=4.17s eta=27:15:17 +2026-09-21 13:02:29 | INFO | train rank=0 | step=26450/50000 progress=52.9% loss=0.394702 lr=4.933863e-05 lr_backbone=4.933863e-05 elapsed=30:35:52 avg_step=4.16s recent_step=4.11s eta=27:14:35 +2026-09-21 13:03:10 | INFO | train rank=0 | step=26460/50000 progress=52.9% loss=0.335220 lr=4.930557e-05 lr_backbone=4.930557e-05 elapsed=30:36:33 avg_step=4.16s recent_step=4.13s eta=27:13:53 +2026-09-21 13:03:51 | INFO | train rank=0 | step=26470/50000 progress=52.9% loss=0.346391 lr=4.927250e-05 lr_backbone=4.927250e-05 elapsed=30:37:15 avg_step=4.16s recent_step=4.13s eta=27:13:11 +2026-09-21 13:04:33 | INFO | train rank=0 | step=26480/50000 progress=53.0% loss=0.330067 lr=4.923943e-05 lr_backbone=4.923943e-05 elapsed=30:37:56 avg_step=4.16s recent_step=4.16s eta=27:12:29 +2026-09-21 13:05:14 | INFO | train rank=0 | step=26490/50000 progress=53.0% loss=0.348628 lr=4.920637e-05 lr_backbone=4.920637e-05 elapsed=30:38:38 avg_step=4.16s recent_step=4.15s eta=27:11:47 +2026-09-21 13:05:56 | INFO | train rank=0 | step=26500/50000 progress=53.0% loss=0.364673 lr=4.917330e-05 lr_backbone=4.917330e-05 elapsed=30:39:19 avg_step=4.16s recent_step=4.16s eta=27:11:06 +2026-09-21 13:06:37 | INFO | train rank=0 | step=26510/50000 progress=53.0% loss=0.319996 lr=4.914024e-05 lr_backbone=4.914024e-05 elapsed=30:40:01 avg_step=4.16s recent_step=4.13s eta=27:10:24 +2026-09-21 13:07:19 | INFO | train rank=0 | step=26520/50000 progress=53.0% loss=0.307860 lr=4.910717e-05 lr_backbone=4.910717e-05 elapsed=30:40:42 avg_step=4.16s recent_step=4.13s eta=27:09:42 +2026-09-21 13:08:00 | INFO | train rank=0 | step=26530/50000 progress=53.1% loss=0.311779 lr=4.907411e-05 lr_backbone=4.907411e-05 elapsed=30:41:23 avg_step=4.16s recent_step=4.16s eta=27:09:00 +2026-09-21 13:08:42 | INFO | train rank=0 | step=26540/50000 progress=53.1% loss=0.297886 lr=4.904105e-05 lr_backbone=4.904105e-05 elapsed=30:42:05 avg_step=4.16s recent_step=4.16s eta=27:08:18 +2026-09-21 13:09:23 | INFO | train rank=0 | step=26550/50000 progress=53.1% loss=0.393170 lr=4.900798e-05 lr_backbone=4.900798e-05 elapsed=30:42:47 avg_step=4.16s recent_step=4.16s eta=27:07:37 +2026-09-21 13:10:05 | INFO | train rank=0 | step=26560/50000 progress=53.1% loss=0.415799 lr=4.897492e-05 lr_backbone=4.897492e-05 elapsed=30:43:28 avg_step=4.16s recent_step=4.17s eta=27:06:55 +2026-09-21 13:10:46 | INFO | train rank=0 | step=26570/50000 progress=53.1% loss=0.323137 lr=4.894186e-05 lr_backbone=4.894186e-05 elapsed=30:44:09 avg_step=4.16s recent_step=4.09s eta=27:06:13 +2026-09-21 13:11:27 | INFO | train rank=0 | step=26580/50000 progress=53.2% loss=0.393038 lr=4.890880e-05 lr_backbone=4.890880e-05 elapsed=30:44:50 avg_step=4.16s recent_step=4.11s eta=27:05:31 +2026-09-21 13:12:08 | INFO | train rank=0 | step=26590/50000 progress=53.2% loss=0.486933 lr=4.887574e-05 lr_backbone=4.887574e-05 elapsed=30:45:31 avg_step=4.16s recent_step=4.09s eta=27:04:48 +2026-09-21 13:12:49 | INFO | train rank=0 | step=26600/50000 progress=53.2% loss=0.357964 lr=4.884267e-05 lr_backbone=4.884267e-05 elapsed=30:46:12 avg_step=4.16s recent_step=4.09s eta=27:04:06 +2026-09-21 13:13:30 | INFO | train rank=0 | step=26610/50000 progress=53.2% loss=0.337760 lr=4.880961e-05 lr_backbone=4.880961e-05 elapsed=30:46:53 avg_step=4.16s recent_step=4.12s eta=27:03:24 +2026-09-21 13:14:11 | INFO | train rank=0 | step=26620/50000 progress=53.2% loss=0.334834 lr=4.877655e-05 lr_backbone=4.877655e-05 elapsed=30:47:34 avg_step=4.16s recent_step=4.09s eta=27:02:42 +2026-09-21 13:14:51 | INFO | train rank=0 | step=26630/50000 progress=53.3% loss=0.392154 lr=4.874350e-05 lr_backbone=4.874350e-05 elapsed=30:48:15 avg_step=4.16s recent_step=4.05s eta=27:01:59 +2026-09-21 13:15:32 | INFO | train rank=0 | step=26640/50000 progress=53.3% loss=0.408564 lr=4.871044e-05 lr_backbone=4.871044e-05 elapsed=30:48:55 avg_step=4.16s recent_step=4.08s eta=27:01:17 +2026-09-21 13:16:13 | INFO | train rank=0 | step=26650/50000 progress=53.3% loss=0.301012 lr=4.867738e-05 lr_backbone=4.867738e-05 elapsed=30:49:36 avg_step=4.16s recent_step=4.10s eta=27:00:34 +2026-09-21 13:16:54 | INFO | train rank=0 | step=26660/50000 progress=53.3% loss=0.376802 lr=4.864432e-05 lr_backbone=4.864432e-05 elapsed=30:50:17 avg_step=4.16s recent_step=4.09s eta=26:59:52 +2026-09-21 13:17:35 | INFO | train rank=0 | step=26670/50000 progress=53.3% loss=0.463907 lr=4.861126e-05 lr_backbone=4.861126e-05 elapsed=30:50:58 avg_step=4.16s recent_step=4.09s eta=26:59:10 +2026-09-21 13:18:16 | INFO | train rank=0 | step=26680/50000 progress=53.4% loss=0.375657 lr=4.857821e-05 lr_backbone=4.857821e-05 elapsed=30:51:39 avg_step=4.16s recent_step=4.08s eta=26:58:27 +2026-09-21 13:18:56 | INFO | train rank=0 | step=26690/50000 progress=53.4% loss=0.416857 lr=4.854515e-05 lr_backbone=4.854515e-05 elapsed=30:52:19 avg_step=4.16s recent_step=4.05s eta=26:57:45 +2026-09-21 13:19:37 | INFO | train rank=0 | step=26700/50000 progress=53.4% loss=0.321484 lr=4.851210e-05 lr_backbone=4.851210e-05 elapsed=30:53:00 avg_step=4.16s recent_step=4.09s eta=26:57:03 +2026-09-21 13:20:18 | INFO | train rank=0 | step=26710/50000 progress=53.4% loss=0.364269 lr=4.847904e-05 lr_backbone=4.847904e-05 elapsed=30:53:41 avg_step=4.16s recent_step=4.09s eta=26:56:20 +2026-09-21 13:20:59 | INFO | train rank=0 | step=26720/50000 progress=53.4% loss=0.333427 lr=4.844599e-05 lr_backbone=4.844599e-05 elapsed=30:54:22 avg_step=4.16s recent_step=4.10s eta=26:55:38 +2026-09-21 13:21:40 | INFO | train rank=0 | step=26730/50000 progress=53.5% loss=0.304106 lr=4.841294e-05 lr_backbone=4.841294e-05 elapsed=30:55:03 avg_step=4.16s recent_step=4.09s eta=26:54:56 +2026-09-21 13:22:21 | INFO | train rank=0 | step=26740/50000 progress=53.5% loss=0.454807 lr=4.837988e-05 lr_backbone=4.837988e-05 elapsed=30:55:44 avg_step=4.16s recent_step=4.11s eta=26:54:14 +2026-09-21 13:23:01 | INFO | train rank=0 | step=26750/50000 progress=53.5% loss=0.345582 lr=4.834683e-05 lr_backbone=4.834683e-05 elapsed=30:56:25 avg_step=4.16s recent_step=4.05s eta=26:53:31 +2026-09-21 13:23:42 | INFO | train rank=0 | step=26760/50000 progress=53.5% loss=0.397441 lr=4.831378e-05 lr_backbone=4.831378e-05 elapsed=30:57:06 avg_step=4.16s recent_step=4.08s eta=26:52:49 +2026-09-21 13:24:23 | INFO | train rank=0 | step=26770/50000 progress=53.5% loss=0.371292 lr=4.828073e-05 lr_backbone=4.828073e-05 elapsed=30:57:47 avg_step=4.16s recent_step=4.11s eta=26:52:07 +2026-09-21 13:25:05 | INFO | train rank=0 | step=26780/50000 progress=53.6% loss=0.375218 lr=4.824768e-05 lr_backbone=4.824768e-05 elapsed=30:58:28 avg_step=4.16s recent_step=4.12s eta=26:51:25 +2026-09-21 13:25:46 | INFO | train rank=0 | step=26790/50000 progress=53.6% loss=0.344399 lr=4.821463e-05 lr_backbone=4.821463e-05 elapsed=30:59:09 avg_step=4.16s recent_step=4.12s eta=26:50:43 +2026-09-21 13:26:30 | INFO | train rank=0 | step=26800/50000 progress=53.6% loss=0.451487 lr=4.818158e-05 lr_backbone=4.818158e-05 elapsed=30:59:53 avg_step=4.16s recent_step=4.40s eta=26:50:03 +2026-09-21 13:27:20 | INFO | train rank=0 | step=26810/50000 progress=53.6% loss=0.291366 lr=4.814854e-05 lr_backbone=4.814854e-05 elapsed=31:00:43 avg_step=4.16s recent_step=4.98s eta=26:49:28 +2026-09-21 13:28:13 | INFO | train rank=0 | step=26820/50000 progress=53.6% loss=0.314412 lr=4.811549e-05 lr_backbone=4.811549e-05 elapsed=31:01:36 avg_step=4.16s recent_step=5.36s eta=26:48:57 +2026-09-21 13:29:03 | INFO | train rank=0 | step=26830/50000 progress=53.7% loss=0.372149 lr=4.808245e-05 lr_backbone=4.808245e-05 elapsed=31:02:26 avg_step=4.16s recent_step=4.95s eta=26:48:22 +2026-09-21 13:29:51 | INFO | train rank=0 | step=26840/50000 progress=53.7% loss=0.397778 lr=4.804940e-05 lr_backbone=4.804940e-05 elapsed=31:03:14 avg_step=4.17s recent_step=4.82s eta=26:47:46 +2026-09-21 13:30:44 | INFO | train rank=0 | step=26850/50000 progress=53.7% loss=0.366133 lr=4.801636e-05 lr_backbone=4.801636e-05 elapsed=31:04:07 avg_step=4.17s recent_step=5.32s eta=26:47:14 +2026-09-21 13:31:35 | INFO | train rank=0 | step=26860/50000 progress=53.7% loss=0.421133 lr=4.798331e-05 lr_backbone=4.798331e-05 elapsed=31:04:59 avg_step=4.17s recent_step=5.13s eta=26:46:41 +2026-09-21 13:32:26 | INFO | train rank=0 | step=26870/50000 progress=53.7% loss=0.351042 lr=4.795027e-05 lr_backbone=4.795027e-05 elapsed=31:05:50 avg_step=4.17s recent_step=5.11s eta=26:46:08 +2026-09-21 13:33:14 | INFO | train rank=0 | step=26880/50000 progress=53.8% loss=0.364875 lr=4.791723e-05 lr_backbone=4.791723e-05 elapsed=31:06:38 avg_step=4.17s recent_step=4.79s eta=26:45:31 +2026-09-21 13:34:04 | INFO | train rank=0 | step=26890/50000 progress=53.8% loss=0.302461 lr=4.788419e-05 lr_backbone=4.788419e-05 elapsed=31:07:27 avg_step=4.17s recent_step=4.96s eta=26:44:56 +2026-09-21 13:34:56 | INFO | train rank=0 | step=26900/50000 progress=53.8% loss=0.299492 lr=4.785115e-05 lr_backbone=4.785115e-05 elapsed=31:08:19 avg_step=4.17s recent_step=5.15s eta=26:44:23 +2026-09-21 13:35:42 | INFO | train rank=0 | step=26910/50000 progress=53.8% loss=0.319129 lr=4.781811e-05 lr_backbone=4.781811e-05 elapsed=31:09:05 avg_step=4.17s recent_step=4.61s eta=26:43:45 +2026-09-21 13:36:35 | INFO | train rank=0 | step=26920/50000 progress=53.8% loss=0.345350 lr=4.778508e-05 lr_backbone=4.778508e-05 elapsed=31:09:58 avg_step=4.17s recent_step=5.32s eta=26:43:14 +2026-09-21 13:37:23 | INFO | train rank=0 | step=26930/50000 progress=53.9% loss=0.439558 lr=4.775204e-05 lr_backbone=4.775204e-05 elapsed=31:10:47 avg_step=4.17s recent_step=4.85s eta=26:42:38 +2026-09-21 13:38:11 | INFO | train rank=0 | step=26940/50000 progress=53.9% loss=0.435804 lr=4.771900e-05 lr_backbone=4.771900e-05 elapsed=31:11:34 avg_step=4.17s recent_step=4.72s eta=26:42:01 +2026-09-21 13:38:57 | INFO | train rank=0 | step=26950/50000 progress=53.9% loss=0.343676 lr=4.768597e-05 lr_backbone=4.768597e-05 elapsed=31:12:21 avg_step=4.17s recent_step=4.68s eta=26:41:24 +2026-09-21 13:39:49 | INFO | train rank=0 | step=26960/50000 progress=53.9% loss=0.344121 lr=4.765294e-05 lr_backbone=4.765294e-05 elapsed=31:13:13 avg_step=4.17s recent_step=5.20s eta=26:40:51 +2026-09-21 13:40:38 | INFO | train rank=0 | step=26970/50000 progress=53.9% loss=0.383052 lr=4.761990e-05 lr_backbone=4.761990e-05 elapsed=31:14:01 avg_step=4.17s recent_step=4.82s eta=26:40:15 +2026-09-21 13:41:27 | INFO | train rank=0 | step=26980/50000 progress=54.0% loss=0.354552 lr=4.758687e-05 lr_backbone=4.758687e-05 elapsed=31:14:50 avg_step=4.17s recent_step=4.90s eta=26:39:39 +2026-09-21 13:42:18 | INFO | train rank=0 | step=26990/50000 progress=54.0% loss=0.348714 lr=4.755384e-05 lr_backbone=4.755384e-05 elapsed=31:15:41 avg_step=4.17s recent_step=5.12s eta=26:39:05 +2026-09-21 13:43:06 | INFO | train rank=0 | step=27000/50000 progress=54.0% loss=0.448578 lr=4.752081e-05 lr_backbone=4.752081e-05 elapsed=31:16:29 avg_step=4.17s recent_step=4.79s eta=26:38:29 +2026-09-21 13:44:02 | INFO | train rank=0 | step=27010/50000 progress=54.0% loss=0.307597 lr=4.748778e-05 lr_backbone=4.748778e-05 elapsed=31:17:25 avg_step=4.17s recent_step=5.60s eta=26:38:00 +2026-09-21 13:44:54 | INFO | train rank=0 | step=27020/50000 progress=54.0% loss=0.381644 lr=4.745476e-05 lr_backbone=4.745476e-05 elapsed=31:18:17 avg_step=4.17s recent_step=5.21s eta=26:37:27 +2026-09-21 13:45:45 | INFO | train rank=0 | step=27030/50000 progress=54.1% loss=0.339835 lr=4.742173e-05 lr_backbone=4.742173e-05 elapsed=31:19:08 avg_step=4.17s recent_step=5.12s eta=26:36:53 +2026-09-21 13:46:32 | INFO | train rank=0 | step=27040/50000 progress=54.1% loss=0.295531 lr=4.738871e-05 lr_backbone=4.738871e-05 elapsed=31:19:55 avg_step=4.17s recent_step=4.71s eta=26:36:16 +2026-09-21 13:47:18 | INFO | train rank=0 | step=27050/50000 progress=54.1% loss=0.355104 lr=4.735568e-05 lr_backbone=4.735568e-05 elapsed=31:20:41 avg_step=4.17s recent_step=4.58s eta=26:35:38 +2026-09-21 13:48:06 | INFO | train rank=0 | step=27060/50000 progress=54.1% loss=0.376329 lr=4.732266e-05 lr_backbone=4.732266e-05 elapsed=31:21:30 avg_step=4.17s recent_step=4.85s eta=26:35:02 +2026-09-21 13:48:56 | INFO | train rank=0 | step=27070/50000 progress=54.1% loss=0.367314 lr=4.728964e-05 lr_backbone=4.728964e-05 elapsed=31:22:19 avg_step=4.17s recent_step=4.92s eta=26:34:26 +2026-09-21 13:49:43 | INFO | train rank=0 | step=27080/50000 progress=54.2% loss=0.347237 lr=4.725662e-05 lr_backbone=4.725662e-05 elapsed=31:23:06 avg_step=4.17s recent_step=4.73s eta=26:33:49 +2026-09-21 13:50:36 | INFO | train rank=0 | step=27090/50000 progress=54.2% loss=0.287808 lr=4.722360e-05 lr_backbone=4.722360e-05 elapsed=31:23:59 avg_step=4.17s recent_step=5.28s eta=26:33:17 +2026-09-21 13:51:23 | INFO | train rank=0 | step=27100/50000 progress=54.2% loss=0.407684 lr=4.719058e-05 lr_backbone=4.719058e-05 elapsed=31:24:46 avg_step=4.17s recent_step=4.74s eta=26:32:40 +2026-09-21 13:52:07 | INFO | train rank=0 | step=27110/50000 progress=54.2% loss=0.375036 lr=4.715757e-05 lr_backbone=4.715757e-05 elapsed=31:25:30 avg_step=4.17s recent_step=4.40s eta=26:32:00 +2026-09-21 13:52:47 | INFO | train rank=0 | step=27120/50000 progress=54.2% loss=0.408880 lr=4.712455e-05 lr_backbone=4.712455e-05 elapsed=31:26:11 avg_step=4.17s recent_step=4.05s eta=26:31:17 +2026-09-21 13:53:32 | INFO | train rank=0 | step=27130/50000 progress=54.3% loss=0.330544 lr=4.709154e-05 lr_backbone=4.709154e-05 elapsed=31:26:55 avg_step=4.17s recent_step=4.43s eta=26:30:38 +2026-09-21 13:54:45 | INFO | train rank=0 | step=27140/50000 progress=54.3% loss=0.303801 lr=4.705852e-05 lr_backbone=4.705852e-05 elapsed=31:28:09 avg_step=4.17s recent_step=7.35s eta=26:30:23 +2026-09-21 13:56:06 | INFO | train rank=0 | step=27150/50000 progress=54.3% loss=0.296319 lr=4.702551e-05 lr_backbone=4.702551e-05 elapsed=31:29:30 avg_step=4.18s recent_step=8.10s eta=26:30:14 +2026-09-21 13:57:25 | INFO | train rank=0 | step=27160/50000 progress=54.3% loss=0.410812 lr=4.699250e-05 lr_backbone=4.699250e-05 elapsed=31:30:48 avg_step=4.18s recent_step=7.87s eta=26:30:03 +2026-09-21 13:58:37 | INFO | train rank=0 | step=27170/50000 progress=54.3% loss=0.419591 lr=4.695949e-05 lr_backbone=4.695949e-05 elapsed=31:32:00 avg_step=4.18s recent_step=7.21s eta=26:29:47 +2026-09-21 14:00:03 | INFO | train rank=0 | step=27180/50000 progress=54.4% loss=0.385183 lr=4.692649e-05 lr_backbone=4.692649e-05 elapsed=31:33:27 avg_step=4.18s recent_step=8.64s eta=26:29:43 +2026-09-21 14:01:20 | INFO | train rank=0 | step=27190/50000 progress=54.4% loss=0.339321 lr=4.689348e-05 lr_backbone=4.689348e-05 elapsed=31:34:43 avg_step=4.18s recent_step=7.64s eta=26:29:30 +2026-09-21 14:02:36 | INFO | train rank=0 | step=27200/50000 progress=54.4% loss=0.374844 lr=4.686047e-05 lr_backbone=4.686047e-05 elapsed=31:35:59 avg_step=4.18s recent_step=7.56s eta=26:29:17 +2026-09-21 14:03:38 | INFO | train rank=0 | step=27210/50000 progress=54.4% loss=0.350382 lr=4.682747e-05 lr_backbone=4.682747e-05 elapsed=31:37:01 avg_step=4.18s recent_step=6.22s eta=26:28:52 +2026-09-21 14:10:26 | INFO | train rank=0 | step=27220/50000 progress=54.4% loss=0.367899 lr=4.679447e-05 lr_backbone=4.679447e-05 elapsed=31:43:49 avg_step=4.20s recent_step=40.84s eta=26:33:17 +2026-09-21 14:11:07 | INFO | train rank=0 | step=27230/50000 progress=54.5% loss=0.279168 lr=4.676147e-05 lr_backbone=4.676147e-05 elapsed=31:44:31 avg_step=4.20s recent_step=4.12s eta=26:32:34 +2026-09-21 14:11:49 | INFO | train rank=0 | step=27240/50000 progress=54.5% loss=0.289322 lr=4.672847e-05 lr_backbone=4.672847e-05 elapsed=31:45:12 avg_step=4.20s recent_step=4.16s eta=26:31:52 +2026-09-21 14:12:31 | INFO | train rank=0 | step=27250/50000 progress=54.5% loss=0.398313 lr=4.669547e-05 lr_backbone=4.669547e-05 elapsed=31:45:54 avg_step=4.20s recent_step=4.17s eta=26:31:10 +2026-09-21 14:13:12 | INFO | train rank=0 | step=27260/50000 progress=54.5% loss=0.304790 lr=4.666247e-05 lr_backbone=4.666247e-05 elapsed=31:46:35 avg_step=4.20s recent_step=4.15s eta=26:30:27 +2026-09-21 14:13:53 | INFO | train rank=0 | step=27270/50000 progress=54.5% loss=0.354586 lr=4.662948e-05 lr_backbone=4.662948e-05 elapsed=31:47:17 avg_step=4.20s recent_step=4.13s eta=26:29:45 +2026-09-21 14:14:35 | INFO | train rank=0 | step=27280/50000 progress=54.6% loss=0.346946 lr=4.659649e-05 lr_backbone=4.659649e-05 elapsed=31:47:58 avg_step=4.20s recent_step=4.15s eta=26:29:02 +2026-09-21 14:15:16 | INFO | train rank=0 | step=27290/50000 progress=54.6% loss=0.369114 lr=4.656349e-05 lr_backbone=4.656349e-05 elapsed=31:48:39 avg_step=4.20s recent_step=4.10s eta=26:28:20 +2026-09-21 14:15:57 | INFO | train rank=0 | step=27300/50000 progress=54.6% loss=0.352637 lr=4.653050e-05 lr_backbone=4.653050e-05 elapsed=31:49:21 avg_step=4.20s recent_step=4.16s eta=26:27:37 +2026-09-21 14:16:39 | INFO | train rank=0 | step=27310/50000 progress=54.6% loss=0.328080 lr=4.649751e-05 lr_backbone=4.649751e-05 elapsed=31:50:02 avg_step=4.20s recent_step=4.14s eta=26:26:55 +2026-09-21 14:17:20 | INFO | train rank=0 | step=27320/50000 progress=54.6% loss=0.291877 lr=4.646453e-05 lr_backbone=4.646453e-05 elapsed=31:50:44 avg_step=4.20s recent_step=4.14s eta=26:26:12 +2026-09-21 14:18:02 | INFO | train rank=0 | step=27330/50000 progress=54.7% loss=0.276829 lr=4.643154e-05 lr_backbone=4.643154e-05 elapsed=31:51:25 avg_step=4.20s recent_step=4.14s eta=26:25:30 +2026-09-21 14:18:43 | INFO | train rank=0 | step=27340/50000 progress=54.7% loss=0.320748 lr=4.639856e-05 lr_backbone=4.639856e-05 elapsed=31:52:07 avg_step=4.20s recent_step=4.16s eta=26:24:48 +2026-09-21 14:19:24 | INFO | train rank=0 | step=27350/50000 progress=54.7% loss=0.414513 lr=4.636557e-05 lr_backbone=4.636557e-05 elapsed=31:52:47 avg_step=4.20s recent_step=4.08s eta=26:24:05 +2026-09-21 14:20:05 | INFO | train rank=0 | step=27360/50000 progress=54.7% loss=0.405764 lr=4.633259e-05 lr_backbone=4.633259e-05 elapsed=31:53:29 avg_step=4.20s recent_step=4.12s eta=26:23:22 +2026-09-21 14:20:47 | INFO | train rank=0 | step=27370/50000 progress=54.7% loss=0.430467 lr=4.629961e-05 lr_backbone=4.629961e-05 elapsed=31:54:10 avg_step=4.20s recent_step=4.15s eta=26:22:40 +2026-09-21 14:21:28 | INFO | train rank=0 | step=27380/50000 progress=54.8% loss=0.382816 lr=4.626664e-05 lr_backbone=4.626664e-05 elapsed=31:54:52 avg_step=4.20s recent_step=4.15s eta=26:21:58 +2026-09-21 14:22:10 | INFO | train rank=0 | step=27390/50000 progress=54.8% loss=0.350906 lr=4.623366e-05 lr_backbone=4.623366e-05 elapsed=31:55:33 avg_step=4.20s recent_step=4.15s eta=26:21:15 +2026-09-21 14:22:51 | INFO | train rank=0 | step=27400/50000 progress=54.8% loss=0.454549 lr=4.620069e-05 lr_backbone=4.620069e-05 elapsed=31:56:14 avg_step=4.20s recent_step=4.13s eta=26:20:33 +2026-09-21 14:23:32 | INFO | train rank=0 | step=27410/50000 progress=54.8% loss=0.396793 lr=4.616771e-05 lr_backbone=4.616771e-05 elapsed=31:56:56 avg_step=4.20s recent_step=4.11s eta=26:19:50 +2026-09-21 14:24:13 | INFO | train rank=0 | step=27420/50000 progress=54.8% loss=0.372907 lr=4.613474e-05 lr_backbone=4.613474e-05 elapsed=31:57:37 avg_step=4.20s recent_step=4.13s eta=26:19:08 +2026-09-21 14:24:55 | INFO | train rank=0 | step=27430/50000 progress=54.9% loss=0.315870 lr=4.610177e-05 lr_backbone=4.610177e-05 elapsed=31:58:18 avg_step=4.20s recent_step=4.15s eta=26:18:25 +2026-09-21 14:25:37 | INFO | train rank=0 | step=27440/50000 progress=54.9% loss=0.382990 lr=4.606880e-05 lr_backbone=4.606880e-05 elapsed=31:59:00 avg_step=4.20s recent_step=4.16s eta=26:17:43 +2026-09-21 14:26:18 | INFO | train rank=0 | step=27450/50000 progress=54.9% loss=0.372860 lr=4.603584e-05 lr_backbone=4.603584e-05 elapsed=31:59:41 avg_step=4.20s recent_step=4.16s eta=26:17:01 +2026-09-21 14:26:59 | INFO | train rank=0 | step=27460/50000 progress=54.9% loss=0.312418 lr=4.600287e-05 lr_backbone=4.600287e-05 elapsed=32:00:22 avg_step=4.20s recent_step=4.10s eta=26:16:18 +2026-09-21 14:27:40 | INFO | train rank=0 | step=27470/50000 progress=54.9% loss=0.463295 lr=4.596991e-05 lr_backbone=4.596991e-05 elapsed=32:01:03 avg_step=4.20s recent_step=4.08s eta=26:15:35 +2026-09-21 14:28:21 | INFO | train rank=0 | step=27480/50000 progress=55.0% loss=0.325250 lr=4.593695e-05 lr_backbone=4.593695e-05 elapsed=32:01:44 avg_step=4.20s recent_step=4.11s eta=26:14:52 +2026-09-21 14:29:03 | INFO | train rank=0 | step=27490/50000 progress=55.0% loss=0.370328 lr=4.590399e-05 lr_backbone=4.590399e-05 elapsed=32:02:26 avg_step=4.20s recent_step=4.16s eta=26:14:10 +2026-09-21 14:29:44 | INFO | train rank=0 | step=27500/50000 progress=55.0% loss=0.381526 lr=4.587103e-05 lr_backbone=4.587103e-05 elapsed=32:03:07 avg_step=4.20s recent_step=4.15s eta=26:13:28 +2026-09-21 14:30:26 | INFO | train rank=0 | step=27510/50000 progress=55.0% loss=0.308794 lr=4.583808e-05 lr_backbone=4.583808e-05 elapsed=32:03:49 avg_step=4.20s recent_step=4.14s eta=26:12:45 +2026-09-21 14:31:07 | INFO | train rank=0 | step=27520/50000 progress=55.0% loss=0.436316 lr=4.580512e-05 lr_backbone=4.580512e-05 elapsed=32:04:30 avg_step=4.20s recent_step=4.14s eta=26:12:03 +2026-09-21 14:31:48 | INFO | train rank=0 | step=27530/50000 progress=55.1% loss=0.330372 lr=4.577217e-05 lr_backbone=4.577217e-05 elapsed=32:05:11 avg_step=4.20s recent_step=4.07s eta=26:11:20 +2026-09-21 14:32:29 | INFO | train rank=0 | step=27540/50000 progress=55.1% loss=0.404870 lr=4.573922e-05 lr_backbone=4.573922e-05 elapsed=32:05:52 avg_step=4.20s recent_step=4.13s eta=26:10:37 +2026-09-21 14:33:10 | INFO | train rank=0 | step=27550/50000 progress=55.1% loss=0.377540 lr=4.570627e-05 lr_backbone=4.570627e-05 elapsed=32:06:33 avg_step=4.20s recent_step=4.12s eta=26:09:55 +2026-09-21 14:33:51 | INFO | train rank=0 | step=27560/50000 progress=55.1% loss=0.318006 lr=4.567333e-05 lr_backbone=4.567333e-05 elapsed=32:07:15 avg_step=4.20s recent_step=4.12s eta=26:09:12 +2026-09-21 14:34:32 | INFO | train rank=0 | step=27570/50000 progress=55.1% loss=0.413801 lr=4.564038e-05 lr_backbone=4.564038e-05 elapsed=32:07:56 avg_step=4.20s recent_step=4.13s eta=26:08:30 +2026-09-21 14:35:14 | INFO | train rank=0 | step=27580/50000 progress=55.2% loss=0.395208 lr=4.560744e-05 lr_backbone=4.560744e-05 elapsed=32:08:37 avg_step=4.20s recent_step=4.14s eta=26:07:47 +2026-09-21 14:35:55 | INFO | train rank=0 | step=27590/50000 progress=55.2% loss=0.394049 lr=4.557450e-05 lr_backbone=4.557450e-05 elapsed=32:09:18 avg_step=4.20s recent_step=4.08s eta=26:07:04 +2026-09-21 14:36:36 | INFO | train rank=0 | step=27600/50000 progress=55.2% loss=0.287963 lr=4.554156e-05 lr_backbone=4.554156e-05 elapsed=32:09:59 avg_step=4.20s recent_step=4.10s eta=26:06:22 +2026-09-21 14:37:17 | INFO | train rank=0 | step=27610/50000 progress=55.2% loss=0.314600 lr=4.550862e-05 lr_backbone=4.550862e-05 elapsed=32:10:40 avg_step=4.20s recent_step=4.13s eta=26:05:39 +2026-09-21 14:37:58 | INFO | train rank=0 | step=27620/50000 progress=55.2% loss=0.523918 lr=4.547569e-05 lr_backbone=4.547569e-05 elapsed=32:11:22 avg_step=4.20s recent_step=4.13s eta=26:04:57 +2026-09-21 14:38:40 | INFO | train rank=0 | step=27630/50000 progress=55.3% loss=0.329986 lr=4.544276e-05 lr_backbone=4.544276e-05 elapsed=32:12:03 avg_step=4.20s recent_step=4.14s eta=26:04:14 +2026-09-21 14:39:21 | INFO | train rank=0 | step=27640/50000 progress=55.3% loss=0.313093 lr=4.540983e-05 lr_backbone=4.540983e-05 elapsed=32:12:44 avg_step=4.20s recent_step=4.12s eta=26:03:32 +2026-09-21 14:40:02 | INFO | train rank=0 | step=27650/50000 progress=55.3% loss=0.363302 lr=4.537690e-05 lr_backbone=4.537690e-05 elapsed=32:13:25 avg_step=4.20s recent_step=4.08s eta=26:02:49 +2026-09-21 14:40:43 | INFO | train rank=0 | step=27660/50000 progress=55.3% loss=0.351510 lr=4.534397e-05 lr_backbone=4.534397e-05 elapsed=32:14:06 avg_step=4.20s recent_step=4.10s eta=26:02:06 +2026-09-21 14:41:24 | INFO | train rank=0 | step=27670/50000 progress=55.3% loss=0.336372 lr=4.531105e-05 lr_backbone=4.531105e-05 elapsed=32:14:47 avg_step=4.20s recent_step=4.11s eta=26:01:24 +2026-09-21 14:42:05 | INFO | train rank=0 | step=27680/50000 progress=55.4% loss=0.382856 lr=4.527812e-05 lr_backbone=4.527812e-05 elapsed=32:15:28 avg_step=4.20s recent_step=4.13s eta=26:00:41 +2026-09-21 14:42:46 | INFO | train rank=0 | step=27690/50000 progress=55.4% loss=0.311506 lr=4.524520e-05 lr_backbone=4.524520e-05 elapsed=32:16:10 avg_step=4.20s recent_step=4.13s eta=25:59:59 +2026-09-21 14:43:28 | INFO | train rank=0 | step=27700/50000 progress=55.4% loss=0.374020 lr=4.521228e-05 lr_backbone=4.521228e-05 elapsed=32:16:51 avg_step=4.20s recent_step=4.14s eta=25:59:16 +2026-09-21 14:44:09 | INFO | train rank=0 | step=27710/50000 progress=55.4% loss=0.321442 lr=4.517937e-05 lr_backbone=4.517937e-05 elapsed=32:17:32 avg_step=4.20s recent_step=4.08s eta=25:58:33 +2026-09-21 14:44:50 | INFO | train rank=0 | step=27720/50000 progress=55.4% loss=0.338884 lr=4.514645e-05 lr_backbone=4.514645e-05 elapsed=32:18:13 avg_step=4.20s recent_step=4.11s eta=25:57:51 +2026-09-21 14:45:31 | INFO | train rank=0 | step=27730/50000 progress=55.5% loss=0.397887 lr=4.511354e-05 lr_backbone=4.511354e-05 elapsed=32:18:54 avg_step=4.20s recent_step=4.13s eta=25:57:08 +2026-09-21 14:46:12 | INFO | train rank=0 | step=27740/50000 progress=55.5% loss=0.322713 lr=4.508063e-05 lr_backbone=4.508063e-05 elapsed=32:19:35 avg_step=4.20s recent_step=4.12s eta=25:56:26 +2026-09-21 14:46:53 | INFO | train rank=0 | step=27750/50000 progress=55.5% loss=0.393972 lr=4.504772e-05 lr_backbone=4.504772e-05 elapsed=32:20:17 avg_step=4.20s recent_step=4.13s eta=25:55:43 +2026-09-21 14:47:35 | INFO | train rank=0 | step=27760/50000 progress=55.5% loss=0.335055 lr=4.501482e-05 lr_backbone=4.501482e-05 elapsed=32:20:58 avg_step=4.20s recent_step=4.13s eta=25:55:01 +2026-09-21 14:48:15 | INFO | train rank=0 | step=27770/50000 progress=55.5% loss=0.337053 lr=4.498191e-05 lr_backbone=4.498191e-05 elapsed=32:21:39 avg_step=4.20s recent_step=4.08s eta=25:54:18 +2026-09-21 14:48:57 | INFO | train rank=0 | step=27780/50000 progress=55.6% loss=0.329286 lr=4.494901e-05 lr_backbone=4.494901e-05 elapsed=32:22:20 avg_step=4.20s recent_step=4.11s eta=25:53:35 +2026-09-21 14:49:38 | INFO | train rank=0 | step=27790/50000 progress=55.6% loss=0.335813 lr=4.491611e-05 lr_backbone=4.491611e-05 elapsed=32:23:02 avg_step=4.20s recent_step=4.17s eta=25:52:53 +2026-09-21 14:50:20 | INFO | train rank=0 | step=27800/50000 progress=55.6% loss=0.422515 lr=4.488322e-05 lr_backbone=4.488322e-05 elapsed=32:23:43 avg_step=4.20s recent_step=4.15s eta=25:52:11 +2026-09-21 14:51:01 | INFO | train rank=0 | step=27810/50000 progress=55.6% loss=0.378887 lr=4.485032e-05 lr_backbone=4.485032e-05 elapsed=32:24:25 avg_step=4.20s recent_step=4.14s eta=25:51:28 +2026-09-21 14:51:43 | INFO | train rank=0 | step=27820/50000 progress=55.6% loss=0.333379 lr=4.481743e-05 lr_backbone=4.481743e-05 elapsed=32:25:06 avg_step=4.20s recent_step=4.14s eta=25:50:46 +2026-09-21 14:52:24 | INFO | train rank=0 | step=27830/50000 progress=55.7% loss=0.339099 lr=4.478454e-05 lr_backbone=4.478454e-05 elapsed=32:25:47 avg_step=4.20s recent_step=4.10s eta=25:50:03 +2026-09-21 14:53:05 | INFO | train rank=0 | step=27840/50000 progress=55.7% loss=0.413931 lr=4.475165e-05 lr_backbone=4.475165e-05 elapsed=32:26:28 avg_step=4.20s recent_step=4.15s eta=25:49:21 +2026-09-21 14:53:46 | INFO | train rank=0 | step=27850/50000 progress=55.7% loss=0.391617 lr=4.471877e-05 lr_backbone=4.471877e-05 elapsed=32:27:10 avg_step=4.19s recent_step=4.13s eta=25:48:38 +2026-09-21 14:54:28 | INFO | train rank=0 | step=27860/50000 progress=55.7% loss=0.358452 lr=4.468588e-05 lr_backbone=4.468588e-05 elapsed=32:27:51 avg_step=4.19s recent_step=4.16s eta=25:47:56 +2026-09-21 14:55:09 | INFO | train rank=0 | step=27870/50000 progress=55.7% loss=0.424100 lr=4.465300e-05 lr_backbone=4.465300e-05 elapsed=32:28:33 avg_step=4.19s recent_step=4.15s eta=25:47:14 +2026-09-21 14:55:51 | INFO | train rank=0 | step=27880/50000 progress=55.8% loss=0.320020 lr=4.462012e-05 lr_backbone=4.462012e-05 elapsed=32:29:14 avg_step=4.19s recent_step=4.16s eta=25:46:32 +2026-09-21 14:56:32 | INFO | train rank=0 | step=27890/50000 progress=55.8% loss=0.360107 lr=4.458725e-05 lr_backbone=4.458725e-05 elapsed=32:29:55 avg_step=4.19s recent_step=4.09s eta=25:45:49 +2026-09-21 14:57:13 | INFO | train rank=0 | step=27900/50000 progress=55.8% loss=0.355995 lr=4.455437e-05 lr_backbone=4.455437e-05 elapsed=32:30:37 avg_step=4.19s recent_step=4.13s eta=25:45:06 +2026-09-21 14:57:55 | INFO | train rank=0 | step=27910/50000 progress=55.8% loss=0.358254 lr=4.452150e-05 lr_backbone=4.452150e-05 elapsed=32:31:18 avg_step=4.19s recent_step=4.14s eta=25:44:24 +2026-09-21 14:58:36 | INFO | train rank=0 | step=27920/50000 progress=55.8% loss=0.413205 lr=4.448863e-05 lr_backbone=4.448863e-05 elapsed=32:31:59 avg_step=4.19s recent_step=4.14s eta=25:43:41 +2026-09-21 14:59:17 | INFO | train rank=0 | step=27930/50000 progress=55.9% loss=0.397602 lr=4.445577e-05 lr_backbone=4.445577e-05 elapsed=32:32:41 avg_step=4.19s recent_step=4.15s eta=25:42:59 +2026-09-21 14:59:59 | INFO | train rank=0 | step=27940/50000 progress=55.9% loss=0.405406 lr=4.442290e-05 lr_backbone=4.442290e-05 elapsed=32:33:22 avg_step=4.19s recent_step=4.13s eta=25:42:17 +2026-09-21 15:00:40 | INFO | train rank=0 | step=27950/50000 progress=55.9% loss=0.292108 lr=4.439004e-05 lr_backbone=4.439004e-05 elapsed=32:34:03 avg_step=4.19s recent_step=4.11s eta=25:41:34 +2026-09-21 15:01:21 | INFO | train rank=0 | step=27960/50000 progress=55.9% loss=0.428531 lr=4.435718e-05 lr_backbone=4.435718e-05 elapsed=32:34:45 avg_step=4.19s recent_step=4.16s eta=25:40:52 +2026-09-21 15:02:03 | INFO | train rank=0 | step=27970/50000 progress=55.9% loss=0.314933 lr=4.432432e-05 lr_backbone=4.432432e-05 elapsed=32:35:26 avg_step=4.19s recent_step=4.13s eta=25:40:09 +2026-09-21 15:02:44 | INFO | train rank=0 | step=27980/50000 progress=56.0% loss=0.320255 lr=4.429147e-05 lr_backbone=4.429147e-05 elapsed=32:36:08 avg_step=4.19s recent_step=4.15s eta=25:39:27 +2026-09-21 15:03:26 | INFO | train rank=0 | step=27990/50000 progress=56.0% loss=0.375941 lr=4.425862e-05 lr_backbone=4.425862e-05 elapsed=32:36:49 avg_step=4.19s recent_step=4.14s eta=25:38:45 +2026-09-21 15:04:07 | INFO | train rank=0 | step=28000/50000 progress=56.0% loss=0.322422 lr=4.422577e-05 lr_backbone=4.422577e-05 elapsed=32:37:30 avg_step=4.19s recent_step=4.16s eta=25:38:02 +2026-09-21 15:04:48 | INFO | train rank=0 | step=28010/50000 progress=56.0% loss=0.346072 lr=4.419292e-05 lr_backbone=4.419292e-05 elapsed=32:38:12 avg_step=4.19s recent_step=4.10s eta=25:37:20 +2026-09-21 15:05:30 | INFO | train rank=0 | step=28020/50000 progress=56.0% loss=0.401141 lr=4.416008e-05 lr_backbone=4.416008e-05 elapsed=32:38:53 avg_step=4.19s recent_step=4.14s eta=25:36:37 +2026-09-21 15:06:11 | INFO | train rank=0 | step=28030/50000 progress=56.1% loss=0.341381 lr=4.412724e-05 lr_backbone=4.412724e-05 elapsed=32:39:35 avg_step=4.19s recent_step=4.16s eta=25:35:55 +2026-09-21 15:06:52 | INFO | train rank=0 | step=28040/50000 progress=56.1% loss=0.325161 lr=4.409440e-05 lr_backbone=4.409440e-05 elapsed=32:40:16 avg_step=4.19s recent_step=4.13s eta=25:35:13 +2026-09-21 15:07:34 | INFO | train rank=0 | step=28050/50000 progress=56.1% loss=0.343631 lr=4.406156e-05 lr_backbone=4.406156e-05 elapsed=32:40:58 avg_step=4.19s recent_step=4.17s eta=25:34:31 +2026-09-21 15:08:16 | INFO | train rank=0 | step=28060/50000 progress=56.1% loss=0.440718 lr=4.402873e-05 lr_backbone=4.402873e-05 elapsed=32:41:39 avg_step=4.19s recent_step=4.14s eta=25:33:48 +2026-09-21 15:08:57 | INFO | train rank=0 | step=28070/50000 progress=56.1% loss=0.402729 lr=4.399589e-05 lr_backbone=4.399589e-05 elapsed=32:42:20 avg_step=4.19s recent_step=4.11s eta=25:33:06 +2026-09-21 15:09:38 | INFO | train rank=0 | step=28080/50000 progress=56.2% loss=0.394872 lr=4.396307e-05 lr_backbone=4.396307e-05 elapsed=32:43:01 avg_step=4.19s recent_step=4.13s eta=25:32:23 +2026-09-21 15:10:20 | INFO | train rank=0 | step=28090/50000 progress=56.2% loss=0.310284 lr=4.393024e-05 lr_backbone=4.393024e-05 elapsed=32:43:43 avg_step=4.19s recent_step=4.16s eta=25:31:41 +2026-09-21 15:11:01 | INFO | train rank=0 | step=28100/50000 progress=56.2% loss=0.380068 lr=4.389742e-05 lr_backbone=4.389742e-05 elapsed=32:44:24 avg_step=4.19s recent_step=4.15s eta=25:30:59 +2026-09-21 15:11:43 | INFO | train rank=0 | step=28110/50000 progress=56.2% loss=0.367448 lr=4.386459e-05 lr_backbone=4.386459e-05 elapsed=32:45:06 avg_step=4.19s recent_step=4.15s eta=25:30:16 +2026-09-21 15:12:24 | INFO | train rank=0 | step=28120/50000 progress=56.2% loss=0.435416 lr=4.383178e-05 lr_backbone=4.383178e-05 elapsed=32:45:48 avg_step=4.19s recent_step=4.18s eta=25:29:34 +2026-09-21 15:13:05 | INFO | train rank=0 | step=28130/50000 progress=56.3% loss=0.331729 lr=4.379896e-05 lr_backbone=4.379896e-05 elapsed=32:46:29 avg_step=4.19s recent_step=4.11s eta=25:28:52 +2026-09-21 15:13:47 | INFO | train rank=0 | step=28140/50000 progress=56.3% loss=0.330212 lr=4.376615e-05 lr_backbone=4.376615e-05 elapsed=32:47:10 avg_step=4.19s recent_step=4.13s eta=25:28:09 +2026-09-21 15:14:28 | INFO | train rank=0 | step=28150/50000 progress=56.3% loss=0.395006 lr=4.373334e-05 lr_backbone=4.373334e-05 elapsed=32:47:52 avg_step=4.19s recent_step=4.15s eta=25:27:27 +2026-09-21 15:15:10 | INFO | train rank=0 | step=28160/50000 progress=56.3% loss=0.393100 lr=4.370053e-05 lr_backbone=4.370053e-05 elapsed=32:48:33 avg_step=4.19s recent_step=4.16s eta=25:26:45 +2026-09-21 15:15:51 | INFO | train rank=0 | step=28170/50000 progress=56.3% loss=0.377670 lr=4.366773e-05 lr_backbone=4.366773e-05 elapsed=32:49:14 avg_step=4.19s recent_step=4.11s eta=25:26:02 +2026-09-21 15:16:33 | INFO | train rank=0 | step=28180/50000 progress=56.4% loss=0.342334 lr=4.363492e-05 lr_backbone=4.363492e-05 elapsed=32:49:56 avg_step=4.19s recent_step=4.17s eta=25:25:20 +2026-09-21 15:17:14 | INFO | train rank=0 | step=28190/50000 progress=56.4% loss=0.341903 lr=4.360213e-05 lr_backbone=4.360213e-05 elapsed=32:50:37 avg_step=4.19s recent_step=4.10s eta=25:24:37 +2026-09-21 15:17:55 | INFO | train rank=0 | step=28200/50000 progress=56.4% loss=0.326454 lr=4.356933e-05 lr_backbone=4.356933e-05 elapsed=32:51:18 avg_step=4.19s recent_step=4.13s eta=25:23:55 +2026-09-21 15:18:36 | INFO | train rank=0 | step=28210/50000 progress=56.4% loss=0.348993 lr=4.353654e-05 lr_backbone=4.353654e-05 elapsed=32:52:00 avg_step=4.19s recent_step=4.14s eta=25:23:13 +2026-09-21 15:19:18 | INFO | train rank=0 | step=28220/50000 progress=56.4% loss=0.373172 lr=4.350375e-05 lr_backbone=4.350375e-05 elapsed=32:52:41 avg_step=4.19s recent_step=4.17s eta=25:22:30 +2026-09-21 15:19:59 | INFO | train rank=0 | step=28230/50000 progress=56.5% loss=0.345021 lr=4.347096e-05 lr_backbone=4.347096e-05 elapsed=32:53:23 avg_step=4.19s recent_step=4.13s eta=25:21:48 +2026-09-21 15:20:41 | INFO | train rank=0 | step=28240/50000 progress=56.5% loss=0.339821 lr=4.343817e-05 lr_backbone=4.343817e-05 elapsed=32:54:04 avg_step=4.19s recent_step=4.16s eta=25:21:06 +2026-09-21 15:21:22 | INFO | train rank=0 | step=28250/50000 progress=56.5% loss=0.437866 lr=4.340539e-05 lr_backbone=4.340539e-05 elapsed=32:54:45 avg_step=4.19s recent_step=4.09s eta=25:20:23 +2026-09-21 15:22:03 | INFO | train rank=0 | step=28260/50000 progress=56.5% loss=0.346768 lr=4.337261e-05 lr_backbone=4.337261e-05 elapsed=32:55:27 avg_step=4.19s recent_step=4.13s eta=25:19:41 +2026-09-21 15:22:45 | INFO | train rank=0 | step=28270/50000 progress=56.5% loss=0.353629 lr=4.333984e-05 lr_backbone=4.333984e-05 elapsed=32:56:08 avg_step=4.19s recent_step=4.15s eta=25:18:58 +2026-09-21 15:23:26 | INFO | train rank=0 | step=28280/50000 progress=56.6% loss=0.362765 lr=4.330706e-05 lr_backbone=4.330706e-05 elapsed=32:56:50 avg_step=4.19s recent_step=4.15s eta=25:18:16 +2026-09-21 15:24:08 | INFO | train rank=0 | step=28290/50000 progress=56.6% loss=0.372771 lr=4.327429e-05 lr_backbone=4.327429e-05 elapsed=32:57:31 avg_step=4.19s recent_step=4.17s eta=25:17:34 +2026-09-21 15:24:49 | INFO | train rank=0 | step=28300/50000 progress=56.6% loss=0.370395 lr=4.324153e-05 lr_backbone=4.324153e-05 elapsed=32:58:13 avg_step=4.19s recent_step=4.14s eta=25:16:52 +2026-09-21 15:25:30 | INFO | train rank=0 | step=28310/50000 progress=56.6% loss=0.309834 lr=4.320876e-05 lr_backbone=4.320876e-05 elapsed=32:58:54 avg_step=4.19s recent_step=4.10s eta=25:16:09 +2026-09-21 15:26:11 | INFO | train rank=0 | step=28320/50000 progress=56.6% loss=0.318655 lr=4.317600e-05 lr_backbone=4.317600e-05 elapsed=32:59:35 avg_step=4.19s recent_step=4.12s eta=25:15:26 +2026-09-21 15:26:53 | INFO | train rank=0 | step=28330/50000 progress=56.7% loss=0.357124 lr=4.314324e-05 lr_backbone=4.314324e-05 elapsed=33:00:16 avg_step=4.19s recent_step=4.15s eta=25:14:44 +2026-09-21 15:27:34 | INFO | train rank=0 | step=28340/50000 progress=56.7% loss=0.321892 lr=4.311049e-05 lr_backbone=4.311049e-05 elapsed=33:00:58 avg_step=4.19s recent_step=4.15s eta=25:14:02 +2026-09-21 15:28:16 | INFO | train rank=0 | step=28350/50000 progress=56.7% loss=0.342290 lr=4.307773e-05 lr_backbone=4.307773e-05 elapsed=33:01:39 avg_step=4.19s recent_step=4.13s eta=25:13:19 +2026-09-21 15:28:57 | INFO | train rank=0 | step=28360/50000 progress=56.7% loss=0.360875 lr=4.304498e-05 lr_backbone=4.304498e-05 elapsed=33:02:20 avg_step=4.19s recent_step=4.13s eta=25:12:37 +2026-09-21 15:29:38 | INFO | train rank=0 | step=28370/50000 progress=56.7% loss=0.358965 lr=4.301224e-05 lr_backbone=4.301224e-05 elapsed=33:03:01 avg_step=4.19s recent_step=4.10s eta=25:11:54 +2026-09-21 15:30:19 | INFO | train rank=0 | step=28380/50000 progress=56.8% loss=0.400582 lr=4.297949e-05 lr_backbone=4.297949e-05 elapsed=33:03:43 avg_step=4.19s recent_step=4.13s eta=25:11:12 +2026-09-21 15:31:01 | INFO | train rank=0 | step=28390/50000 progress=56.8% loss=0.333800 lr=4.294675e-05 lr_backbone=4.294675e-05 elapsed=33:04:24 avg_step=4.19s recent_step=4.14s eta=25:10:30 +2026-09-21 15:31:42 | INFO | train rank=0 | step=28400/50000 progress=56.8% loss=0.291327 lr=4.291402e-05 lr_backbone=4.291402e-05 elapsed=33:05:06 avg_step=4.19s recent_step=4.15s eta=25:09:47 +2026-09-21 15:32:24 | INFO | train rank=0 | step=28410/50000 progress=56.8% loss=0.308307 lr=4.288128e-05 lr_backbone=4.288128e-05 elapsed=33:05:47 avg_step=4.19s recent_step=4.14s eta=25:09:05 +2026-09-21 15:33:05 | INFO | train rank=0 | step=28420/50000 progress=56.8% loss=0.318008 lr=4.284855e-05 lr_backbone=4.284855e-05 elapsed=33:06:28 avg_step=4.19s recent_step=4.14s eta=25:08:23 +2026-09-21 15:33:46 | INFO | train rank=0 | step=28430/50000 progress=56.9% loss=0.320592 lr=4.281582e-05 lr_backbone=4.281582e-05 elapsed=33:07:09 avg_step=4.19s recent_step=4.09s eta=25:07:40 +2026-09-21 15:34:27 | INFO | train rank=0 | step=28440/50000 progress=56.9% loss=0.313918 lr=4.278310e-05 lr_backbone=4.278310e-05 elapsed=33:07:50 avg_step=4.19s recent_step=4.11s eta=25:06:57 +2026-09-21 15:35:09 | INFO | train rank=0 | step=28450/50000 progress=56.9% loss=0.317011 lr=4.275038e-05 lr_backbone=4.275038e-05 elapsed=33:08:32 avg_step=4.19s recent_step=4.14s eta=25:06:15 +2026-09-21 15:35:50 | INFO | train rank=0 | step=28460/50000 progress=56.9% loss=0.362366 lr=4.271766e-05 lr_backbone=4.271766e-05 elapsed=33:09:13 avg_step=4.19s recent_step=4.14s eta=25:05:33 +2026-09-21 15:36:31 | INFO | train rank=0 | step=28470/50000 progress=56.9% loss=0.353314 lr=4.268494e-05 lr_backbone=4.268494e-05 elapsed=33:09:55 avg_step=4.19s recent_step=4.15s eta=25:04:50 +2026-09-21 15:37:13 | INFO | train rank=0 | step=28480/50000 progress=57.0% loss=0.392386 lr=4.265223e-05 lr_backbone=4.265223e-05 elapsed=33:10:36 avg_step=4.19s recent_step=4.15s eta=25:04:08 +2026-09-21 15:37:54 | INFO | train rank=0 | step=28490/50000 progress=57.0% loss=0.435535 lr=4.261952e-05 lr_backbone=4.261952e-05 elapsed=33:11:17 avg_step=4.19s recent_step=4.09s eta=25:03:25 +2026-09-21 15:38:35 | INFO | train rank=0 | step=28500/50000 progress=57.0% loss=0.375407 lr=4.258682e-05 lr_backbone=4.258682e-05 elapsed=33:11:58 avg_step=4.19s recent_step=4.11s eta=25:02:43 +2026-09-21 15:39:16 | INFO | train rank=0 | step=28510/50000 progress=57.0% loss=0.340794 lr=4.255412e-05 lr_backbone=4.255412e-05 elapsed=33:12:39 avg_step=4.19s recent_step=4.13s eta=25:02:00 +2026-09-21 15:39:57 | INFO | train rank=0 | step=28520/50000 progress=57.0% loss=0.343378 lr=4.252142e-05 lr_backbone=4.252142e-05 elapsed=33:13:21 avg_step=4.19s recent_step=4.11s eta=25:01:18 +2026-09-21 15:40:39 | INFO | train rank=0 | step=28530/50000 progress=57.1% loss=0.365961 lr=4.248872e-05 lr_backbone=4.248872e-05 elapsed=33:14:02 avg_step=4.19s recent_step=4.14s eta=25:00:35 +2026-09-21 15:41:20 | INFO | train rank=0 | step=28540/50000 progress=57.1% loss=0.395469 lr=4.245603e-05 lr_backbone=4.245603e-05 elapsed=33:14:43 avg_step=4.19s recent_step=4.12s eta=24:59:53 +2026-09-21 15:42:01 | INFO | train rank=0 | step=28550/50000 progress=57.1% loss=0.356987 lr=4.242334e-05 lr_backbone=4.242334e-05 elapsed=33:15:24 avg_step=4.19s recent_step=4.08s eta=24:59:10 +2026-09-21 15:42:42 | INFO | train rank=0 | step=28560/50000 progress=57.1% loss=0.338366 lr=4.239065e-05 lr_backbone=4.239065e-05 elapsed=33:16:05 avg_step=4.19s recent_step=4.11s eta=24:58:28 +2026-09-21 15:43:23 | INFO | train rank=0 | step=28570/50000 progress=57.1% loss=0.330447 lr=4.235797e-05 lr_backbone=4.235797e-05 elapsed=33:16:46 avg_step=4.19s recent_step=4.13s eta=24:57:45 +2026-09-21 15:44:04 | INFO | train rank=0 | step=28580/50000 progress=57.2% loss=0.346724 lr=4.232529e-05 lr_backbone=4.232529e-05 elapsed=33:17:28 avg_step=4.19s recent_step=4.12s eta=24:57:03 +2026-09-21 15:44:45 | INFO | train rank=0 | step=28590/50000 progress=57.2% loss=0.350917 lr=4.229262e-05 lr_backbone=4.229262e-05 elapsed=33:18:09 avg_step=4.19s recent_step=4.12s eta=24:56:20 +2026-09-21 15:45:27 | INFO | train rank=0 | step=28600/50000 progress=57.2% loss=0.329884 lr=4.225994e-05 lr_backbone=4.225994e-05 elapsed=33:18:50 avg_step=4.19s recent_step=4.12s eta=24:55:38 +2026-09-21 15:46:08 | INFO | train rank=0 | step=28610/50000 progress=57.2% loss=0.282915 lr=4.222727e-05 lr_backbone=4.222727e-05 elapsed=33:19:31 avg_step=4.19s recent_step=4.09s eta=24:54:55 +2026-09-21 15:46:49 | INFO | train rank=0 | step=28620/50000 progress=57.2% loss=0.301459 lr=4.219461e-05 lr_backbone=4.219461e-05 elapsed=33:20:12 avg_step=4.19s recent_step=4.11s eta=24:54:12 +2026-09-21 15:47:30 | INFO | train rank=0 | step=28630/50000 progress=57.3% loss=0.320378 lr=4.216195e-05 lr_backbone=4.216195e-05 elapsed=33:20:53 avg_step=4.19s recent_step=4.11s eta=24:53:30 +2026-09-21 15:48:11 | INFO | train rank=0 | step=28640/50000 progress=57.3% loss=0.343533 lr=4.212929e-05 lr_backbone=4.212929e-05 elapsed=33:21:35 avg_step=4.19s recent_step=4.14s eta=24:52:48 +2026-09-21 15:48:52 | INFO | train rank=0 | step=28650/50000 progress=57.3% loss=0.303547 lr=4.209663e-05 lr_backbone=4.209663e-05 elapsed=33:22:16 avg_step=4.19s recent_step=4.12s eta=24:52:05 +2026-09-21 15:49:34 | INFO | train rank=0 | step=28660/50000 progress=57.3% loss=0.295772 lr=4.206398e-05 lr_backbone=4.206398e-05 elapsed=33:22:57 avg_step=4.19s recent_step=4.15s eta=24:51:23 +2026-09-21 15:50:15 | INFO | train rank=0 | step=28670/50000 progress=57.3% loss=0.332233 lr=4.203133e-05 lr_backbone=4.203133e-05 elapsed=33:23:38 avg_step=4.19s recent_step=4.08s eta=24:50:40 +2026-09-21 15:50:56 | INFO | train rank=0 | step=28680/50000 progress=57.4% loss=0.408368 lr=4.199869e-05 lr_backbone=4.199869e-05 elapsed=33:24:19 avg_step=4.19s recent_step=4.10s eta=24:49:58 +2026-09-21 15:51:37 | INFO | train rank=0 | step=28690/50000 progress=57.4% loss=0.401125 lr=4.196605e-05 lr_backbone=4.196605e-05 elapsed=33:25:01 avg_step=4.19s recent_step=4.17s eta=24:49:15 +2026-09-21 15:52:19 | INFO | train rank=0 | step=28700/50000 progress=57.4% loss=0.356358 lr=4.193341e-05 lr_backbone=4.193341e-05 elapsed=33:25:42 avg_step=4.19s recent_step=4.16s eta=24:48:33 +2026-09-21 15:53:00 | INFO | train rank=0 | step=28710/50000 progress=57.4% loss=0.320242 lr=4.190077e-05 lr_backbone=4.190077e-05 elapsed=33:26:24 avg_step=4.19s recent_step=4.14s eta=24:47:51 +2026-09-21 15:53:42 | INFO | train rank=0 | step=28720/50000 progress=57.4% loss=0.381856 lr=4.186814e-05 lr_backbone=4.186814e-05 elapsed=33:27:05 avg_step=4.19s recent_step=4.15s eta=24:47:09 +2026-09-21 15:54:23 | INFO | train rank=0 | step=28730/50000 progress=57.5% loss=0.439744 lr=4.183551e-05 lr_backbone=4.183551e-05 elapsed=33:27:46 avg_step=4.19s recent_step=4.10s eta=24:46:26 +2026-09-21 15:55:04 | INFO | train rank=0 | step=28740/50000 progress=57.5% loss=0.333151 lr=4.180289e-05 lr_backbone=4.180289e-05 elapsed=33:28:27 avg_step=4.19s recent_step=4.10s eta=24:45:43 +2026-09-21 15:55:45 | INFO | train rank=0 | step=28750/50000 progress=57.5% loss=0.286331 lr=4.177027e-05 lr_backbone=4.177027e-05 elapsed=33:29:09 avg_step=4.19s recent_step=4.15s eta=24:45:01 +2026-09-21 15:56:27 | INFO | train rank=0 | step=28760/50000 progress=57.5% loss=0.382469 lr=4.173765e-05 lr_backbone=4.173765e-05 elapsed=33:29:50 avg_step=4.19s recent_step=4.15s eta=24:44:19 +2026-09-21 15:57:08 | INFO | train rank=0 | step=28770/50000 progress=57.5% loss=0.345620 lr=4.170504e-05 lr_backbone=4.170504e-05 elapsed=33:30:32 avg_step=4.19s recent_step=4.14s eta=24:43:37 +2026-09-21 15:57:50 | INFO | train rank=0 | step=28780/50000 progress=57.6% loss=0.420410 lr=4.167243e-05 lr_backbone=4.167243e-05 elapsed=33:31:13 avg_step=4.19s recent_step=4.15s eta=24:42:54 +2026-09-21 15:58:31 | INFO | train rank=0 | step=28790/50000 progress=57.6% loss=0.315629 lr=4.163983e-05 lr_backbone=4.163983e-05 elapsed=33:31:54 avg_step=4.19s recent_step=4.11s eta=24:42:12 +2026-09-21 15:59:12 | INFO | train rank=0 | step=28800/50000 progress=57.6% loss=0.318722 lr=4.160722e-05 lr_backbone=4.160722e-05 elapsed=33:32:35 avg_step=4.19s recent_step=4.12s eta=24:41:29 +2026-09-21 15:59:53 | INFO | train rank=0 | step=28810/50000 progress=57.6% loss=0.298860 lr=4.157463e-05 lr_backbone=4.157463e-05 elapsed=33:33:17 avg_step=4.19s recent_step=4.13s eta=24:40:47 +2026-09-21 16:00:35 | INFO | train rank=0 | step=28820/50000 progress=57.6% loss=0.373651 lr=4.154203e-05 lr_backbone=4.154203e-05 elapsed=33:33:58 avg_step=4.19s recent_step=4.16s eta=24:40:05 +2026-09-21 16:01:16 | INFO | train rank=0 | step=28830/50000 progress=57.7% loss=0.355960 lr=4.150944e-05 lr_backbone=4.150944e-05 elapsed=33:34:40 avg_step=4.19s recent_step=4.15s eta=24:39:22 +2026-09-21 16:01:58 | INFO | train rank=0 | step=28840/50000 progress=57.7% loss=0.355381 lr=4.147685e-05 lr_backbone=4.147685e-05 elapsed=33:35:22 avg_step=4.19s recent_step=4.17s eta=24:38:40 +2026-09-21 16:02:39 | INFO | train rank=0 | step=28850/50000 progress=57.7% loss=0.313341 lr=4.144427e-05 lr_backbone=4.144427e-05 elapsed=33:36:03 avg_step=4.19s recent_step=4.11s eta=24:37:58 +2026-09-21 16:03:21 | INFO | train rank=0 | step=28860/50000 progress=57.7% loss=0.333241 lr=4.141169e-05 lr_backbone=4.141169e-05 elapsed=33:36:44 avg_step=4.19s recent_step=4.13s eta=24:37:15 +2026-09-21 16:04:02 | INFO | train rank=0 | step=28870/50000 progress=57.7% loss=0.392323 lr=4.137911e-05 lr_backbone=4.137911e-05 elapsed=33:37:25 avg_step=4.19s recent_step=4.12s eta=24:36:33 +2026-09-21 16:04:43 | INFO | train rank=0 | step=28880/50000 progress=57.8% loss=0.317598 lr=4.134654e-05 lr_backbone=4.134654e-05 elapsed=33:38:07 avg_step=4.19s recent_step=4.17s eta=24:35:51 +2026-09-21 16:05:25 | INFO | train rank=0 | step=28890/50000 progress=57.8% loss=0.289663 lr=4.131397e-05 lr_backbone=4.131397e-05 elapsed=33:38:48 avg_step=4.19s recent_step=4.13s eta=24:35:08 +2026-09-21 16:06:06 | INFO | train rank=0 | step=28900/50000 progress=57.8% loss=0.314927 lr=4.128141e-05 lr_backbone=4.128141e-05 elapsed=33:39:29 avg_step=4.19s recent_step=4.14s eta=24:34:26 +2026-09-21 16:06:47 | INFO | train rank=0 | step=28910/50000 progress=57.8% loss=0.374065 lr=4.124885e-05 lr_backbone=4.124885e-05 elapsed=33:40:11 avg_step=4.19s recent_step=4.10s eta=24:33:44 +2026-09-21 16:07:28 | INFO | train rank=0 | step=28920/50000 progress=57.8% loss=0.368093 lr=4.121629e-05 lr_backbone=4.121629e-05 elapsed=33:40:52 avg_step=4.19s recent_step=4.13s eta=24:33:01 +2026-09-21 16:08:10 | INFO | train rank=0 | step=28930/50000 progress=57.9% loss=0.359979 lr=4.118374e-05 lr_backbone=4.118374e-05 elapsed=33:41:33 avg_step=4.19s recent_step=4.14s eta=24:32:19 +2026-09-21 16:08:51 | INFO | train rank=0 | step=28940/50000 progress=57.9% loss=0.354510 lr=4.115119e-05 lr_backbone=4.115119e-05 elapsed=33:42:15 avg_step=4.19s recent_step=4.13s eta=24:31:37 +2026-09-21 16:09:33 | INFO | train rank=0 | step=28950/50000 progress=57.9% loss=0.337576 lr=4.111864e-05 lr_backbone=4.111864e-05 elapsed=33:42:56 avg_step=4.19s recent_step=4.15s eta=24:30:54 +2026-09-21 16:10:14 | INFO | train rank=0 | step=28960/50000 progress=57.9% loss=0.333843 lr=4.108610e-05 lr_backbone=4.108610e-05 elapsed=33:43:38 avg_step=4.19s recent_step=4.14s eta=24:30:12 +2026-09-21 16:10:55 | INFO | train rank=0 | step=28970/50000 progress=57.9% loss=0.343239 lr=4.105356e-05 lr_backbone=4.105356e-05 elapsed=33:44:18 avg_step=4.19s recent_step=4.09s eta=24:29:29 +2026-09-21 16:11:36 | INFO | train rank=0 | step=28980/50000 progress=58.0% loss=0.331789 lr=4.102103e-05 lr_backbone=4.102103e-05 elapsed=33:45:00 avg_step=4.19s recent_step=4.12s eta=24:28:47 +2026-09-21 16:12:18 | INFO | train rank=0 | step=28990/50000 progress=58.0% loss=0.390518 lr=4.098850e-05 lr_backbone=4.098850e-05 elapsed=33:45:41 avg_step=4.19s recent_step=4.15s eta=24:28:05 +2026-09-21 16:12:59 | INFO | train rank=0 | step=29000/50000 progress=58.0% loss=0.327842 lr=4.095597e-05 lr_backbone=4.095597e-05 elapsed=33:46:23 avg_step=4.19s recent_step=4.15s eta=24:27:23 +2026-09-21 16:13:41 | INFO | train rank=0 | step=29010/50000 progress=58.0% loss=0.273535 lr=4.092345e-05 lr_backbone=4.092345e-05 elapsed=33:47:04 avg_step=4.19s recent_step=4.11s eta=24:26:40 +2026-09-21 16:14:22 | INFO | train rank=0 | step=29020/50000 progress=58.0% loss=0.313344 lr=4.089093e-05 lr_backbone=4.089093e-05 elapsed=33:47:46 avg_step=4.19s recent_step=4.17s eta=24:25:58 +2026-09-21 16:15:03 | INFO | train rank=0 | step=29030/50000 progress=58.1% loss=0.435549 lr=4.085842e-05 lr_backbone=4.085842e-05 elapsed=33:48:26 avg_step=4.19s recent_step=4.09s eta=24:25:15 +2026-09-21 16:15:44 | INFO | train rank=0 | step=29040/50000 progress=58.1% loss=0.332179 lr=4.082591e-05 lr_backbone=4.082591e-05 elapsed=33:49:08 avg_step=4.19s recent_step=4.13s eta=24:24:33 +2026-09-21 16:16:26 | INFO | train rank=0 | step=29050/50000 progress=58.1% loss=0.398991 lr=4.079340e-05 lr_backbone=4.079340e-05 elapsed=33:49:49 avg_step=4.19s recent_step=4.15s eta=24:23:51 +2026-09-21 16:17:07 | INFO | train rank=0 | step=29060/50000 progress=58.1% loss=0.341543 lr=4.076090e-05 lr_backbone=4.076090e-05 elapsed=33:50:31 avg_step=4.19s recent_step=4.13s eta=24:23:08 +2026-09-21 16:17:49 | INFO | train rank=0 | step=29070/50000 progress=58.1% loss=0.367366 lr=4.072840e-05 lr_backbone=4.072840e-05 elapsed=33:51:12 avg_step=4.19s recent_step=4.14s eta=24:22:26 +2026-09-21 16:18:30 | INFO | train rank=0 | step=29080/50000 progress=58.2% loss=0.339037 lr=4.069591e-05 lr_backbone=4.069591e-05 elapsed=33:51:53 avg_step=4.19s recent_step=4.15s eta=24:21:44 +2026-09-21 16:19:11 | INFO | train rank=0 | step=29090/50000 progress=58.2% loss=0.360544 lr=4.066342e-05 lr_backbone=4.066342e-05 elapsed=33:52:34 avg_step=4.19s recent_step=4.10s eta=24:21:01 +2026-09-21 16:19:52 | INFO | train rank=0 | step=29100/50000 progress=58.2% loss=0.300262 lr=4.063093e-05 lr_backbone=4.063093e-05 elapsed=33:53:16 avg_step=4.19s recent_step=4.12s eta=24:20:19 +2026-09-21 16:20:34 | INFO | train rank=0 | step=29110/50000 progress=58.2% loss=0.359130 lr=4.059845e-05 lr_backbone=4.059845e-05 elapsed=33:53:57 avg_step=4.19s recent_step=4.15s eta=24:19:36 +2026-09-21 16:21:15 | INFO | train rank=0 | step=29120/50000 progress=58.2% loss=0.288947 lr=4.056598e-05 lr_backbone=4.056598e-05 elapsed=33:54:39 avg_step=4.19s recent_step=4.15s eta=24:18:54 +2026-09-21 16:21:57 | INFO | train rank=0 | step=29130/50000 progress=58.3% loss=0.311438 lr=4.053350e-05 lr_backbone=4.053350e-05 elapsed=33:55:20 avg_step=4.19s recent_step=4.13s eta=24:18:12 +2026-09-21 16:22:38 | INFO | train rank=0 | step=29140/50000 progress=58.3% loss=0.423290 lr=4.050103e-05 lr_backbone=4.050103e-05 elapsed=33:56:02 avg_step=4.19s recent_step=4.17s eta=24:17:30 +2026-09-21 16:23:19 | INFO | train rank=0 | step=29150/50000 progress=58.3% loss=0.291679 lr=4.046857e-05 lr_backbone=4.046857e-05 elapsed=33:56:43 avg_step=4.19s recent_step=4.10s eta=24:16:47 +2026-09-21 16:24:01 | INFO | train rank=0 | step=29160/50000 progress=58.3% loss=0.323940 lr=4.043611e-05 lr_backbone=4.043611e-05 elapsed=33:57:24 avg_step=4.19s recent_step=4.13s eta=24:16:05 +2026-09-21 16:24:42 | INFO | train rank=0 | step=29170/50000 progress=58.3% loss=0.378250 lr=4.040365e-05 lr_backbone=4.040365e-05 elapsed=33:58:05 avg_step=4.19s recent_step=4.15s eta=24:15:23 +2026-09-21 16:25:24 | INFO | train rank=0 | step=29180/50000 progress=58.4% loss=0.438254 lr=4.037120e-05 lr_backbone=4.037120e-05 elapsed=33:58:47 avg_step=4.19s recent_step=4.15s eta=24:14:40 +2026-09-21 16:26:05 | INFO | train rank=0 | step=29190/50000 progress=58.4% loss=0.282481 lr=4.033875e-05 lr_backbone=4.033875e-05 elapsed=33:59:28 avg_step=4.19s recent_step=4.16s eta=24:13:58 +2026-09-21 16:26:47 | INFO | train rank=0 | step=29200/50000 progress=58.4% loss=0.350981 lr=4.030631e-05 lr_backbone=4.030631e-05 elapsed=34:00:10 avg_step=4.19s recent_step=4.15s eta=24:13:16 +2026-09-21 16:27:28 | INFO | train rank=0 | step=29210/50000 progress=58.4% loss=0.345821 lr=4.027387e-05 lr_backbone=4.027387e-05 elapsed=34:00:51 avg_step=4.19s recent_step=4.10s eta=24:12:33 +2026-09-21 16:28:09 | INFO | train rank=0 | step=29220/50000 progress=58.4% loss=0.297058 lr=4.024143e-05 lr_backbone=4.024143e-05 elapsed=34:01:32 avg_step=4.19s recent_step=4.13s eta=24:11:51 +2026-09-21 16:28:50 | INFO | train rank=0 | step=29230/50000 progress=58.5% loss=0.312459 lr=4.020900e-05 lr_backbone=4.020900e-05 elapsed=34:02:14 avg_step=4.19s recent_step=4.14s eta=24:11:09 +2026-09-21 16:29:32 | INFO | train rank=0 | step=29240/50000 progress=58.5% loss=0.383338 lr=4.017657e-05 lr_backbone=4.017657e-05 elapsed=34:02:55 avg_step=4.19s recent_step=4.16s eta=24:10:27 +2026-09-21 16:30:13 | INFO | train rank=0 | step=29250/50000 progress=58.5% loss=0.442054 lr=4.014415e-05 lr_backbone=4.014415e-05 elapsed=34:03:37 avg_step=4.19s recent_step=4.16s eta=24:09:44 +2026-09-21 16:30:55 | INFO | train rank=0 | step=29260/50000 progress=58.5% loss=0.338158 lr=4.011173e-05 lr_backbone=4.011173e-05 elapsed=34:04:18 avg_step=4.19s recent_step=4.13s eta=24:09:02 +2026-09-21 16:31:36 | INFO | train rank=0 | step=29270/50000 progress=58.5% loss=0.341470 lr=4.007932e-05 lr_backbone=4.007932e-05 elapsed=34:04:59 avg_step=4.19s recent_step=4.09s eta=24:08:19 +2026-09-21 16:32:17 | INFO | train rank=0 | step=29280/50000 progress=58.6% loss=0.372448 lr=4.004691e-05 lr_backbone=4.004691e-05 elapsed=34:05:40 avg_step=4.19s recent_step=4.11s eta=24:07:37 +2026-09-21 16:32:58 | INFO | train rank=0 | step=29290/50000 progress=58.6% loss=0.423323 lr=4.001450e-05 lr_backbone=4.001450e-05 elapsed=34:06:21 avg_step=4.19s recent_step=4.12s eta=24:06:54 +2026-09-21 16:33:39 | INFO | train rank=0 | step=29300/50000 progress=58.6% loss=0.305655 lr=3.998210e-05 lr_backbone=3.998210e-05 elapsed=34:07:03 avg_step=4.19s recent_step=4.13s eta=24:06:12 +2026-09-21 16:34:20 | INFO | train rank=0 | step=29310/50000 progress=58.6% loss=0.319645 lr=3.994970e-05 lr_backbone=3.994970e-05 elapsed=34:07:44 avg_step=4.19s recent_step=4.13s eta=24:05:30 +2026-09-21 16:35:02 | INFO | train rank=0 | step=29320/50000 progress=58.6% loss=0.325181 lr=3.991731e-05 lr_backbone=3.991731e-05 elapsed=34:08:25 avg_step=4.19s recent_step=4.13s eta=24:04:47 +2026-09-21 16:35:43 | INFO | train rank=0 | step=29330/50000 progress=58.7% loss=0.439196 lr=3.988492e-05 lr_backbone=3.988492e-05 elapsed=34:09:06 avg_step=4.19s recent_step=4.09s eta=24:04:05 +2026-09-21 16:36:24 | INFO | train rank=0 | step=29340/50000 progress=58.7% loss=0.413244 lr=3.985254e-05 lr_backbone=3.985254e-05 elapsed=34:09:47 avg_step=4.19s recent_step=4.10s eta=24:03:22 +2026-09-21 16:37:05 | INFO | train rank=0 | step=29350/50000 progress=58.7% loss=0.313243 lr=3.982016e-05 lr_backbone=3.982016e-05 elapsed=34:10:28 avg_step=4.19s recent_step=4.13s eta=24:02:40 +2026-09-21 16:37:46 | INFO | train rank=0 | step=29360/50000 progress=58.7% loss=0.419324 lr=3.978779e-05 lr_backbone=3.978779e-05 elapsed=34:11:10 avg_step=4.19s recent_step=4.14s eta=24:01:57 +2026-09-21 16:38:28 | INFO | train rank=0 | step=29370/50000 progress=58.7% loss=0.307400 lr=3.975542e-05 lr_backbone=3.975542e-05 elapsed=34:11:51 avg_step=4.19s recent_step=4.13s eta=24:01:15 +2026-09-21 16:39:09 | INFO | train rank=0 | step=29380/50000 progress=58.8% loss=0.375512 lr=3.972305e-05 lr_backbone=3.972305e-05 elapsed=34:12:32 avg_step=4.19s recent_step=4.13s eta=24:00:33 +2026-09-21 16:39:50 | INFO | train rank=0 | step=29390/50000 progress=58.8% loss=0.314967 lr=3.969069e-05 lr_backbone=3.969069e-05 elapsed=34:13:13 avg_step=4.19s recent_step=4.07s eta=23:59:50 +2026-09-21 16:40:31 | INFO | train rank=0 | step=29400/50000 progress=58.8% loss=0.384389 lr=3.965833e-05 lr_backbone=3.965833e-05 elapsed=34:13:54 avg_step=4.19s recent_step=4.11s eta=23:59:08 +2026-09-21 16:41:12 | INFO | train rank=0 | step=29410/50000 progress=58.8% loss=0.317109 lr=3.962598e-05 lr_backbone=3.962598e-05 elapsed=34:14:35 avg_step=4.19s recent_step=4.13s eta=23:58:25 +2026-09-21 16:41:54 | INFO | train rank=0 | step=29420/50000 progress=58.8% loss=0.325489 lr=3.959363e-05 lr_backbone=3.959363e-05 elapsed=34:15:17 avg_step=4.19s recent_step=4.15s eta=23:57:43 +2026-09-21 16:42:35 | INFO | train rank=0 | step=29430/50000 progress=58.9% loss=0.382390 lr=3.956129e-05 lr_backbone=3.956129e-05 elapsed=34:15:58 avg_step=4.19s recent_step=4.13s eta=23:57:01 +2026-09-21 16:43:16 | INFO | train rank=0 | step=29440/50000 progress=58.9% loss=0.373513 lr=3.952895e-05 lr_backbone=3.952895e-05 elapsed=34:16:40 avg_step=4.19s recent_step=4.14s eta=23:56:18 +2026-09-21 16:43:57 | INFO | train rank=0 | step=29450/50000 progress=58.9% loss=0.336982 lr=3.949662e-05 lr_backbone=3.949662e-05 elapsed=34:17:21 avg_step=4.19s recent_step=4.11s eta=23:55:36 +2026-09-21 16:44:39 | INFO | train rank=0 | step=29460/50000 progress=58.9% loss=0.365506 lr=3.946429e-05 lr_backbone=3.946429e-05 elapsed=34:18:02 avg_step=4.19s recent_step=4.12s eta=23:54:53 +2026-09-21 16:45:20 | INFO | train rank=0 | step=29470/50000 progress=58.9% loss=0.304693 lr=3.943196e-05 lr_backbone=3.943196e-05 elapsed=34:18:44 avg_step=4.19s recent_step=4.16s eta=23:54:11 +2026-09-21 16:46:02 | INFO | train rank=0 | step=29480/50000 progress=59.0% loss=0.311246 lr=3.939964e-05 lr_backbone=3.939964e-05 elapsed=34:19:25 avg_step=4.19s recent_step=4.16s eta=23:53:29 +2026-09-21 16:46:43 | INFO | train rank=0 | step=29490/50000 progress=59.0% loss=0.311994 lr=3.936733e-05 lr_backbone=3.936733e-05 elapsed=34:20:06 avg_step=4.19s recent_step=4.13s eta=23:52:47 +2026-09-21 16:47:25 | INFO | train rank=0 | step=29500/50000 progress=59.0% loss=0.308463 lr=3.933502e-05 lr_backbone=3.933502e-05 elapsed=34:20:48 avg_step=4.19s recent_step=4.15s eta=23:52:05 +2026-09-21 16:48:06 | INFO | train rank=0 | step=29510/50000 progress=59.0% loss=0.398599 lr=3.930271e-05 lr_backbone=3.930271e-05 elapsed=34:21:29 avg_step=4.19s recent_step=4.10s eta=23:51:22 +2026-09-21 16:48:47 | INFO | train rank=0 | step=29520/50000 progress=59.0% loss=0.353776 lr=3.927041e-05 lr_backbone=3.927041e-05 elapsed=34:22:10 avg_step=4.19s recent_step=4.12s eta=23:50:40 +2026-09-21 16:49:28 | INFO | train rank=0 | step=29530/50000 progress=59.1% loss=0.490324 lr=3.923811e-05 lr_backbone=3.923811e-05 elapsed=34:22:51 avg_step=4.19s recent_step=4.13s eta=23:49:57 +2026-09-21 16:50:10 | INFO | train rank=0 | step=29540/50000 progress=59.1% loss=0.300125 lr=3.920582e-05 lr_backbone=3.920582e-05 elapsed=34:23:33 avg_step=4.19s recent_step=4.16s eta=23:49:15 +2026-09-21 16:50:51 | INFO | train rank=0 | step=29550/50000 progress=59.1% loss=0.333770 lr=3.917354e-05 lr_backbone=3.917354e-05 elapsed=34:24:15 avg_step=4.19s recent_step=4.17s eta=23:48:33 +2026-09-21 16:51:33 | INFO | train rank=0 | step=29560/50000 progress=59.1% loss=0.343979 lr=3.914125e-05 lr_backbone=3.914125e-05 elapsed=34:24:56 avg_step=4.19s recent_step=4.11s eta=23:47:51 +2026-09-21 16:52:14 | INFO | train rank=0 | step=29570/50000 progress=59.1% loss=0.384817 lr=3.910898e-05 lr_backbone=3.910898e-05 elapsed=34:25:37 avg_step=4.19s recent_step=4.10s eta=23:47:08 +2026-09-21 16:52:55 | INFO | train rank=0 | step=29580/50000 progress=59.2% loss=0.386503 lr=3.907670e-05 lr_backbone=3.907670e-05 elapsed=34:26:18 avg_step=4.19s recent_step=4.14s eta=23:46:26 +2026-09-21 16:53:36 | INFO | train rank=0 | step=29590/50000 progress=59.2% loss=0.324011 lr=3.904443e-05 lr_backbone=3.904443e-05 elapsed=34:27:00 avg_step=4.19s recent_step=4.13s eta=23:45:44 +2026-09-21 16:54:18 | INFO | train rank=0 | step=29600/50000 progress=59.2% loss=0.385224 lr=3.901217e-05 lr_backbone=3.901217e-05 elapsed=34:27:41 avg_step=4.19s recent_step=4.14s eta=23:45:01 +2026-09-21 16:54:59 | INFO | train rank=0 | step=29610/50000 progress=59.2% loss=0.316042 lr=3.897991e-05 lr_backbone=3.897991e-05 elapsed=34:28:23 avg_step=4.19s recent_step=4.16s eta=23:44:19 +2026-09-21 16:55:41 | INFO | train rank=0 | step=29620/50000 progress=59.2% loss=0.371243 lr=3.894766e-05 lr_backbone=3.894766e-05 elapsed=34:29:04 avg_step=4.19s recent_step=4.15s eta=23:43:37 +2026-09-21 16:56:22 | INFO | train rank=0 | step=29630/50000 progress=59.3% loss=0.366594 lr=3.891541e-05 lr_backbone=3.891541e-05 elapsed=34:29:45 avg_step=4.19s recent_step=4.10s eta=23:42:55 +2026-09-21 16:57:03 | INFO | train rank=0 | step=29640/50000 progress=59.3% loss=0.315964 lr=3.888317e-05 lr_backbone=3.888317e-05 elapsed=34:30:26 avg_step=4.19s recent_step=4.14s eta=23:42:12 +2026-09-21 16:57:45 | INFO | train rank=0 | step=29650/50000 progress=59.3% loss=0.344384 lr=3.885093e-05 lr_backbone=3.885093e-05 elapsed=34:31:08 avg_step=4.19s recent_step=4.16s eta=23:41:30 +2026-09-21 16:58:26 | INFO | train rank=0 | step=29660/50000 progress=59.3% loss=0.369583 lr=3.881869e-05 lr_backbone=3.881869e-05 elapsed=34:31:49 avg_step=4.19s recent_step=4.14s eta=23:40:48 +2026-09-21 16:59:08 | INFO | train rank=0 | step=29670/50000 progress=59.3% loss=0.323961 lr=3.878646e-05 lr_backbone=3.878646e-05 elapsed=34:32:31 avg_step=4.19s recent_step=4.17s eta=23:40:06 +2026-09-21 16:59:49 | INFO | train rank=0 | step=29680/50000 progress=59.4% loss=0.417349 lr=3.875424e-05 lr_backbone=3.875424e-05 elapsed=34:33:13 avg_step=4.19s recent_step=4.16s eta=23:39:24 +2026-09-21 17:00:30 | INFO | train rank=0 | step=29690/50000 progress=59.4% loss=0.341049 lr=3.872202e-05 lr_backbone=3.872202e-05 elapsed=34:33:54 avg_step=4.19s recent_step=4.10s eta=23:38:41 +2026-09-21 17:01:12 | INFO | train rank=0 | step=29700/50000 progress=59.4% loss=0.319362 lr=3.868980e-05 lr_backbone=3.868980e-05 elapsed=34:34:35 avg_step=4.19s recent_step=4.14s eta=23:37:59 +2026-09-21 17:01:53 | INFO | train rank=0 | step=29710/50000 progress=59.4% loss=0.313477 lr=3.865759e-05 lr_backbone=3.865759e-05 elapsed=34:35:17 avg_step=4.19s recent_step=4.14s eta=23:37:17 +2026-09-21 17:02:34 | INFO | train rank=0 | step=29720/50000 progress=59.4% loss=0.270032 lr=3.862539e-05 lr_backbone=3.862539e-05 elapsed=34:35:58 avg_step=4.19s recent_step=4.11s eta=23:36:34 +2026-09-21 17:03:16 | INFO | train rank=0 | step=29730/50000 progress=59.5% loss=0.297270 lr=3.859319e-05 lr_backbone=3.859319e-05 elapsed=34:36:39 avg_step=4.19s recent_step=4.15s eta=23:35:52 +2026-09-21 17:03:57 | INFO | train rank=0 | step=29740/50000 progress=59.5% loss=0.341843 lr=3.856099e-05 lr_backbone=3.856099e-05 elapsed=34:37:21 avg_step=4.19s recent_step=4.14s eta=23:35:10 +2026-09-21 17:04:38 | INFO | train rank=0 | step=29750/50000 progress=59.5% loss=0.327338 lr=3.852880e-05 lr_backbone=3.852880e-05 elapsed=34:38:02 avg_step=4.19s recent_step=4.10s eta=23:34:27 +2026-09-21 17:05:20 | INFO | train rank=0 | step=29760/50000 progress=59.5% loss=0.296797 lr=3.849662e-05 lr_backbone=3.849662e-05 elapsed=34:38:43 avg_step=4.19s recent_step=4.13s eta=23:33:45 +2026-09-21 17:06:01 | INFO | train rank=0 | step=29770/50000 progress=59.5% loss=0.378949 lr=3.846444e-05 lr_backbone=3.846444e-05 elapsed=34:39:24 avg_step=4.19s recent_step=4.16s eta=23:33:03 +2026-09-21 17:06:42 | INFO | train rank=0 | step=29780/50000 progress=59.6% loss=0.370104 lr=3.843226e-05 lr_backbone=3.843226e-05 elapsed=34:40:06 avg_step=4.19s recent_step=4.13s eta=23:32:20 +2026-09-21 17:07:24 | INFO | train rank=0 | step=29790/50000 progress=59.6% loss=0.380526 lr=3.840009e-05 lr_backbone=3.840009e-05 elapsed=34:40:47 avg_step=4.19s recent_step=4.15s eta=23:31:38 +2026-09-21 17:08:05 | INFO | train rank=0 | step=29800/50000 progress=59.6% loss=0.273235 lr=3.836793e-05 lr_backbone=3.836793e-05 elapsed=34:41:29 avg_step=4.19s recent_step=4.15s eta=23:30:56 +2026-09-21 17:08:46 | INFO | train rank=0 | step=29810/50000 progress=59.6% loss=0.352137 lr=3.833577e-05 lr_backbone=3.833577e-05 elapsed=34:42:10 avg_step=4.19s recent_step=4.10s eta=23:30:13 +2026-09-21 17:09:28 | INFO | train rank=0 | step=29820/50000 progress=59.6% loss=0.347935 lr=3.830362e-05 lr_backbone=3.830362e-05 elapsed=34:42:51 avg_step=4.19s recent_step=4.13s eta=23:29:31 +2026-09-21 17:10:09 | INFO | train rank=0 | step=29830/50000 progress=59.7% loss=0.327103 lr=3.827147e-05 lr_backbone=3.827147e-05 elapsed=34:43:33 avg_step=4.19s recent_step=4.15s eta=23:28:49 +2026-09-21 17:10:51 | INFO | train rank=0 | step=29840/50000 progress=59.7% loss=0.359587 lr=3.823932e-05 lr_backbone=3.823932e-05 elapsed=34:44:14 avg_step=4.19s recent_step=4.13s eta=23:28:07 +2026-09-21 17:11:32 | INFO | train rank=0 | step=29850/50000 progress=59.7% loss=0.392474 lr=3.820718e-05 lr_backbone=3.820718e-05 elapsed=34:44:55 avg_step=4.19s recent_step=4.14s eta=23:27:24 +2026-09-21 17:12:14 | INFO | train rank=0 | step=29860/50000 progress=59.7% loss=0.360753 lr=3.817505e-05 lr_backbone=3.817505e-05 elapsed=34:45:37 avg_step=4.19s recent_step=4.17s eta=23:26:42 +2026-09-21 17:12:55 | INFO | train rank=0 | step=29870/50000 progress=59.7% loss=0.359881 lr=3.814292e-05 lr_backbone=3.814292e-05 elapsed=34:46:18 avg_step=4.19s recent_step=4.10s eta=23:26:00 +2026-09-21 17:13:36 | INFO | train rank=0 | step=29880/50000 progress=59.8% loss=0.319249 lr=3.811080e-05 lr_backbone=3.811080e-05 elapsed=34:46:59 avg_step=4.19s recent_step=4.13s eta=23:25:18 +2026-09-21 17:14:18 | INFO | train rank=0 | step=29890/50000 progress=59.8% loss=0.366736 lr=3.807868e-05 lr_backbone=3.807868e-05 elapsed=34:47:41 avg_step=4.19s recent_step=4.16s eta=23:24:35 +2026-09-21 17:14:59 | INFO | train rank=0 | step=29900/50000 progress=59.8% loss=0.392607 lr=3.804657e-05 lr_backbone=3.804657e-05 elapsed=34:48:22 avg_step=4.19s recent_step=4.12s eta=23:23:53 +2026-09-21 17:15:40 | INFO | train rank=0 | step=29910/50000 progress=59.8% loss=0.324787 lr=3.801446e-05 lr_backbone=3.801446e-05 elapsed=34:49:04 avg_step=4.19s recent_step=4.14s eta=23:23:11 +2026-09-21 17:16:22 | INFO | train rank=0 | step=29920/50000 progress=59.8% loss=0.415470 lr=3.798236e-05 lr_backbone=3.798236e-05 elapsed=34:49:45 avg_step=4.19s recent_step=4.15s eta=23:22:29 +2026-09-21 17:17:03 | INFO | train rank=0 | step=29930/50000 progress=59.9% loss=0.263994 lr=3.795026e-05 lr_backbone=3.795026e-05 elapsed=34:50:26 avg_step=4.19s recent_step=4.10s eta=23:21:46 +2026-09-21 17:17:44 | INFO | train rank=0 | step=29940/50000 progress=59.9% loss=0.288407 lr=3.791817e-05 lr_backbone=3.791817e-05 elapsed=34:51:07 avg_step=4.19s recent_step=4.12s eta=23:21:04 +2026-09-21 17:18:25 | INFO | train rank=0 | step=29950/50000 progress=59.9% loss=0.351139 lr=3.788608e-05 lr_backbone=3.788608e-05 elapsed=34:51:49 avg_step=4.19s recent_step=4.16s eta=23:20:22 +2026-09-21 17:19:07 | INFO | train rank=0 | step=29960/50000 progress=59.9% loss=0.299393 lr=3.785400e-05 lr_backbone=3.785400e-05 elapsed=34:52:31 avg_step=4.19s recent_step=4.17s eta=23:19:40 +2026-09-21 17:19:48 | INFO | train rank=0 | step=29970/50000 progress=59.9% loss=0.321301 lr=3.782192e-05 lr_backbone=3.782192e-05 elapsed=34:53:12 avg_step=4.19s recent_step=4.12s eta=23:18:57 +2026-09-21 17:20:30 | INFO | train rank=0 | step=29980/50000 progress=60.0% loss=0.362790 lr=3.778985e-05 lr_backbone=3.778985e-05 elapsed=34:53:53 avg_step=4.19s recent_step=4.14s eta=23:18:15 +2026-09-21 17:21:11 | INFO | train rank=0 | step=29990/50000 progress=60.0% loss=0.285752 lr=3.775779e-05 lr_backbone=3.775779e-05 elapsed=34:54:34 avg_step=4.19s recent_step=4.09s eta=23:17:32 +2026-09-21 17:21:52 | INFO | train rank=0 | step=30000/50000 progress=60.0% loss=0.341747 lr=3.772573e-05 lr_backbone=3.772573e-05 elapsed=34:55:15 avg_step=4.19s recent_step=4.12s eta=23:16:50 +2026-09-21 17:22:33 | INFO | train rank=0 | step=30010/50000 progress=60.0% loss=0.303183 lr=3.769367e-05 lr_backbone=3.769367e-05 elapsed=34:55:57 avg_step=4.19s recent_step=4.16s eta=23:16:08 +2026-09-21 17:23:15 | INFO | train rank=0 | step=30020/50000 progress=60.0% loss=0.368049 lr=3.766162e-05 lr_backbone=3.766162e-05 elapsed=34:56:38 avg_step=4.19s recent_step=4.15s eta=23:15:26 +2026-09-21 17:23:56 | INFO | train rank=0 | step=30030/50000 progress=60.1% loss=0.367665 lr=3.762958e-05 lr_backbone=3.762958e-05 elapsed=34:57:20 avg_step=4.19s recent_step=4.13s eta=23:14:43 +2026-09-21 17:24:38 | INFO | train rank=0 | step=30040/50000 progress=60.1% loss=0.292926 lr=3.759754e-05 lr_backbone=3.759754e-05 elapsed=34:58:01 avg_step=4.19s recent_step=4.15s eta=23:14:01 +2026-09-21 17:25:19 | INFO | train rank=0 | step=30050/50000 progress=60.1% loss=0.315163 lr=3.756551e-05 lr_backbone=3.756551e-05 elapsed=34:58:42 avg_step=4.19s recent_step=4.10s eta=23:13:19 +2026-09-21 17:26:00 | INFO | train rank=0 | step=30060/50000 progress=60.1% loss=0.327390 lr=3.753348e-05 lr_backbone=3.753348e-05 elapsed=34:59:24 avg_step=4.19s recent_step=4.15s eta=23:12:37 +2026-09-21 17:26:42 | INFO | train rank=0 | step=30070/50000 progress=60.1% loss=0.354478 lr=3.750146e-05 lr_backbone=3.750146e-05 elapsed=35:00:05 avg_step=4.19s recent_step=4.15s eta=23:11:54 +2026-09-21 17:27:23 | INFO | train rank=0 | step=30080/50000 progress=60.2% loss=0.341532 lr=3.746944e-05 lr_backbone=3.746944e-05 elapsed=35:00:47 avg_step=4.19s recent_step=4.16s eta=23:11:12 +2026-09-21 17:28:05 | INFO | train rank=0 | step=30090/50000 progress=60.2% loss=0.287943 lr=3.743743e-05 lr_backbone=3.743743e-05 elapsed=35:01:28 avg_step=4.19s recent_step=4.14s eta=23:10:30 +2026-09-21 17:28:46 | INFO | train rank=0 | step=30100/50000 progress=60.2% loss=0.343529 lr=3.740542e-05 lr_backbone=3.740542e-05 elapsed=35:02:10 avg_step=4.19s recent_step=4.16s eta=23:09:48 +2026-09-21 17:29:27 | INFO | train rank=0 | step=30110/50000 progress=60.2% loss=0.343655 lr=3.737342e-05 lr_backbone=3.737342e-05 elapsed=35:02:51 avg_step=4.19s recent_step=4.10s eta=23:09:05 +2026-09-21 17:30:09 | INFO | train rank=0 | step=30120/50000 progress=60.2% loss=0.368024 lr=3.734143e-05 lr_backbone=3.734143e-05 elapsed=35:03:32 avg_step=4.19s recent_step=4.13s eta=23:08:23 +2026-09-21 17:30:50 | INFO | train rank=0 | step=30130/50000 progress=60.3% loss=0.313762 lr=3.730944e-05 lr_backbone=3.730944e-05 elapsed=35:04:14 avg_step=4.19s recent_step=4.16s eta=23:07:41 +2026-09-21 17:31:32 | INFO | train rank=0 | step=30140/50000 progress=60.3% loss=0.354749 lr=3.727745e-05 lr_backbone=3.727745e-05 elapsed=35:04:55 avg_step=4.19s recent_step=4.15s eta=23:06:59 +2026-09-21 17:32:13 | INFO | train rank=0 | step=30150/50000 progress=60.3% loss=0.335952 lr=3.724548e-05 lr_backbone=3.724548e-05 elapsed=35:05:36 avg_step=4.19s recent_step=4.12s eta=23:06:16 +2026-09-21 17:32:55 | INFO | train rank=0 | step=30160/50000 progress=60.3% loss=0.286465 lr=3.721350e-05 lr_backbone=3.721350e-05 elapsed=35:06:18 avg_step=4.19s recent_step=4.18s eta=23:05:34 +2026-09-21 17:33:36 | INFO | train rank=0 | step=30170/50000 progress=60.3% loss=0.316961 lr=3.718154e-05 lr_backbone=3.718154e-05 elapsed=35:06:59 avg_step=4.19s recent_step=4.10s eta=23:04:52 +2026-09-21 17:34:17 | INFO | train rank=0 | step=30180/50000 progress=60.4% loss=0.443061 lr=3.714957e-05 lr_backbone=3.714957e-05 elapsed=35:07:40 avg_step=4.19s recent_step=4.13s eta=23:04:10 +2026-09-21 17:34:59 | INFO | train rank=0 | step=30190/50000 progress=60.4% loss=0.272589 lr=3.711762e-05 lr_backbone=3.711762e-05 elapsed=35:08:22 avg_step=4.19s recent_step=4.15s eta=23:03:28 +2026-09-21 17:35:40 | INFO | train rank=0 | step=30200/50000 progress=60.4% loss=0.347794 lr=3.708567e-05 lr_backbone=3.708567e-05 elapsed=35:09:03 avg_step=4.19s recent_step=4.15s eta=23:02:45 +2026-09-21 17:36:22 | INFO | train rank=0 | step=30210/50000 progress=60.4% loss=0.400599 lr=3.705372e-05 lr_backbone=3.705372e-05 elapsed=35:09:45 avg_step=4.19s recent_step=4.16s eta=23:02:03 +2026-09-21 17:37:03 | INFO | train rank=0 | step=30220/50000 progress=60.4% loss=0.317675 lr=3.702179e-05 lr_backbone=3.702179e-05 elapsed=35:10:26 avg_step=4.19s recent_step=4.14s eta=23:01:21 +2026-09-21 17:37:44 | INFO | train rank=0 | step=30230/50000 progress=60.5% loss=0.343951 lr=3.698985e-05 lr_backbone=3.698985e-05 elapsed=35:11:07 avg_step=4.19s recent_step=4.10s eta=23:00:39 +2026-09-21 17:38:25 | INFO | train rank=0 | step=30240/50000 progress=60.5% loss=0.317892 lr=3.695792e-05 lr_backbone=3.695792e-05 elapsed=35:11:49 avg_step=4.19s recent_step=4.13s eta=22:59:56 +2026-09-21 17:39:07 | INFO | train rank=0 | step=30250/50000 progress=60.5% loss=0.397480 lr=3.692600e-05 lr_backbone=3.692600e-05 elapsed=35:12:30 avg_step=4.19s recent_step=4.14s eta=22:59:14 +2026-09-21 17:39:48 | INFO | train rank=0 | step=30260/50000 progress=60.5% loss=0.292526 lr=3.689409e-05 lr_backbone=3.689409e-05 elapsed=35:13:11 avg_step=4.19s recent_step=4.13s eta=22:58:32 +2026-09-21 17:40:30 | INFO | train rank=0 | step=30270/50000 progress=60.5% loss=0.439613 lr=3.686218e-05 lr_backbone=3.686218e-05 elapsed=35:13:53 avg_step=4.19s recent_step=4.16s eta=22:57:50 +2026-09-21 17:41:11 | INFO | train rank=0 | step=30280/50000 progress=60.6% loss=0.339544 lr=3.683027e-05 lr_backbone=3.683027e-05 elapsed=35:14:34 avg_step=4.19s recent_step=4.13s eta=22:57:07 +2026-09-21 17:41:52 | INFO | train rank=0 | step=30290/50000 progress=60.6% loss=0.356490 lr=3.679837e-05 lr_backbone=3.679837e-05 elapsed=35:15:15 avg_step=4.19s recent_step=4.09s eta=22:56:25 +2026-09-21 17:42:34 | INFO | train rank=0 | step=30300/50000 progress=60.6% loss=0.309556 lr=3.676648e-05 lr_backbone=3.676648e-05 elapsed=35:15:57 avg_step=4.19s recent_step=4.15s eta=22:55:43 +2026-09-21 17:43:15 | INFO | train rank=0 | step=30310/50000 progress=60.6% loss=0.337975 lr=3.673459e-05 lr_backbone=3.673459e-05 elapsed=35:16:38 avg_step=4.19s recent_step=4.14s eta=22:55:01 +2026-09-21 17:43:56 | INFO | train rank=0 | step=30320/50000 progress=60.6% loss=0.341153 lr=3.670271e-05 lr_backbone=3.670271e-05 elapsed=35:17:20 avg_step=4.19s recent_step=4.15s eta=22:54:18 +2026-09-21 17:44:38 | INFO | train rank=0 | step=30330/50000 progress=60.7% loss=0.336813 lr=3.667084e-05 lr_backbone=3.667084e-05 elapsed=35:18:01 avg_step=4.19s recent_step=4.15s eta=22:53:36 +2026-09-21 17:45:19 | INFO | train rank=0 | step=30340/50000 progress=60.7% loss=0.313630 lr=3.663897e-05 lr_backbone=3.663897e-05 elapsed=35:18:43 avg_step=4.19s recent_step=4.13s eta=22:52:54 +2026-09-21 17:46:00 | INFO | train rank=0 | step=30350/50000 progress=60.7% loss=0.424428 lr=3.660710e-05 lr_backbone=3.660710e-05 elapsed=35:19:23 avg_step=4.19s recent_step=4.07s eta=22:52:11 +2026-09-21 17:46:41 | INFO | train rank=0 | step=30360/50000 progress=60.7% loss=0.332935 lr=3.657524e-05 lr_backbone=3.657524e-05 elapsed=35:20:05 avg_step=4.19s recent_step=4.12s eta=22:51:29 +2026-09-21 17:47:22 | INFO | train rank=0 | step=30370/50000 progress=60.7% loss=0.336761 lr=3.654339e-05 lr_backbone=3.654339e-05 elapsed=35:20:46 avg_step=4.19s recent_step=4.13s eta=22:50:47 +2026-09-21 17:48:04 | INFO | train rank=0 | step=30380/50000 progress=60.8% loss=0.425123 lr=3.651155e-05 lr_backbone=3.651155e-05 elapsed=35:21:27 avg_step=4.19s recent_step=4.13s eta=22:50:04 +2026-09-21 17:48:45 | INFO | train rank=0 | step=30390/50000 progress=60.8% loss=0.406945 lr=3.647971e-05 lr_backbone=3.647971e-05 elapsed=35:22:08 avg_step=4.19s recent_step=4.13s eta=22:49:22 +2026-09-21 17:49:27 | INFO | train rank=0 | step=30400/50000 progress=60.8% loss=0.336818 lr=3.644787e-05 lr_backbone=3.644787e-05 elapsed=35:22:50 avg_step=4.19s recent_step=4.16s eta=22:48:40 +2026-09-21 17:50:07 | INFO | train rank=0 | step=30410/50000 progress=60.8% loss=0.378284 lr=3.641604e-05 lr_backbone=3.641604e-05 elapsed=35:23:31 avg_step=4.19s recent_step=4.08s eta=22:47:57 +2026-09-21 17:50:48 | INFO | train rank=0 | step=30420/50000 progress=60.8% loss=0.351024 lr=3.638422e-05 lr_backbone=3.638422e-05 elapsed=35:24:12 avg_step=4.19s recent_step=4.10s eta=22:47:15 +2026-09-21 17:51:29 | INFO | train rank=0 | step=30430/50000 progress=60.9% loss=0.387087 lr=3.635240e-05 lr_backbone=3.635240e-05 elapsed=35:24:53 avg_step=4.19s recent_step=4.10s eta=22:46:32 +2026-09-21 17:52:11 | INFO | train rank=0 | step=30440/50000 progress=60.9% loss=0.415155 lr=3.632059e-05 lr_backbone=3.632059e-05 elapsed=35:25:34 avg_step=4.19s recent_step=4.13s eta=22:45:50 +2026-09-21 17:52:52 | INFO | train rank=0 | step=30450/50000 progress=60.9% loss=0.370194 lr=3.628879e-05 lr_backbone=3.628879e-05 elapsed=35:26:15 avg_step=4.19s recent_step=4.13s eta=22:45:08 +2026-09-21 17:53:33 | INFO | train rank=0 | step=30460/50000 progress=60.9% loss=0.301709 lr=3.625699e-05 lr_backbone=3.625699e-05 elapsed=35:26:57 avg_step=4.19s recent_step=4.12s eta=22:44:25 +2026-09-21 17:54:14 | INFO | train rank=0 | step=30470/50000 progress=60.9% loss=0.364472 lr=3.622520e-05 lr_backbone=3.622520e-05 elapsed=35:27:37 avg_step=4.19s recent_step=4.07s eta=22:43:43 +2026-09-21 17:54:55 | INFO | train rank=0 | step=30480/50000 progress=61.0% loss=0.352689 lr=3.619341e-05 lr_backbone=3.619341e-05 elapsed=35:28:18 avg_step=4.19s recent_step=4.10s eta=22:43:00 +2026-09-21 17:55:36 | INFO | train rank=0 | step=30490/50000 progress=61.0% loss=0.317106 lr=3.616163e-05 lr_backbone=3.616163e-05 elapsed=35:29:00 avg_step=4.19s recent_step=4.13s eta=22:42:18 +2026-09-21 17:56:17 | INFO | train rank=0 | step=30500/50000 progress=61.0% loss=0.311461 lr=3.612985e-05 lr_backbone=3.612985e-05 elapsed=35:29:41 avg_step=4.19s recent_step=4.12s eta=22:41:36 +2026-09-21 17:56:59 | INFO | train rank=0 | step=30510/50000 progress=61.0% loss=0.308121 lr=3.609809e-05 lr_backbone=3.609809e-05 elapsed=35:30:22 avg_step=4.19s recent_step=4.14s eta=22:40:54 +2026-09-21 17:57:40 | INFO | train rank=0 | step=30520/50000 progress=61.0% loss=0.343002 lr=3.606632e-05 lr_backbone=3.606632e-05 elapsed=35:31:04 avg_step=4.19s recent_step=4.15s eta=22:40:11 +2026-09-21 17:58:21 | INFO | train rank=0 | step=30530/50000 progress=61.1% loss=0.357183 lr=3.603457e-05 lr_backbone=3.603457e-05 elapsed=35:31:44 avg_step=4.19s recent_step=4.08s eta=22:39:29 +2026-09-21 17:59:02 | INFO | train rank=0 | step=30540/50000 progress=61.1% loss=0.454865 lr=3.600282e-05 lr_backbone=3.600282e-05 elapsed=35:32:25 avg_step=4.19s recent_step=4.10s eta=22:38:46 +2026-09-21 17:59:44 | INFO | train rank=0 | step=30550/50000 progress=61.1% loss=0.343095 lr=3.597107e-05 lr_backbone=3.597107e-05 elapsed=35:33:07 avg_step=4.19s recent_step=4.17s eta=22:38:04 +2026-09-21 18:00:26 | INFO | train rank=0 | step=30560/50000 progress=61.1% loss=0.335737 lr=3.593934e-05 lr_backbone=3.593934e-05 elapsed=35:33:49 avg_step=4.19s recent_step=4.17s eta=22:37:22 +2026-09-21 18:01:07 | INFO | train rank=0 | step=30570/50000 progress=61.1% loss=0.309172 lr=3.590760e-05 lr_backbone=3.590760e-05 elapsed=35:34:31 avg_step=4.19s recent_step=4.18s eta=22:36:40 +2026-09-21 18:01:49 | INFO | train rank=0 | step=30580/50000 progress=61.2% loss=0.292678 lr=3.587588e-05 lr_backbone=3.587588e-05 elapsed=35:35:12 avg_step=4.19s recent_step=4.12s eta=22:35:58 +2026-09-21 18:02:30 | INFO | train rank=0 | step=30590/50000 progress=61.2% loss=0.320949 lr=3.584416e-05 lr_backbone=3.584416e-05 elapsed=35:35:53 avg_step=4.19s recent_step=4.11s eta=22:35:16 +2026-09-21 18:03:11 | INFO | train rank=0 | step=30600/50000 progress=61.2% loss=0.397537 lr=3.581244e-05 lr_backbone=3.581244e-05 elapsed=35:36:34 avg_step=4.19s recent_step=4.11s eta=22:34:33 +2026-09-21 18:03:52 | INFO | train rank=0 | step=30610/50000 progress=61.2% loss=0.335752 lr=3.578074e-05 lr_backbone=3.578074e-05 elapsed=35:37:16 avg_step=4.19s recent_step=4.15s eta=22:33:51 +2026-09-21 18:04:34 | INFO | train rank=0 | step=30620/50000 progress=61.2% loss=0.399815 lr=3.574904e-05 lr_backbone=3.574904e-05 elapsed=35:37:57 avg_step=4.19s recent_step=4.16s eta=22:33:09 +2026-09-21 18:05:15 | INFO | train rank=0 | step=30630/50000 progress=61.3% loss=0.377652 lr=3.571734e-05 lr_backbone=3.571734e-05 elapsed=35:38:39 avg_step=4.19s recent_step=4.15s eta=22:32:27 +2026-09-21 18:05:57 | INFO | train rank=0 | step=30640/50000 progress=61.3% loss=0.369361 lr=3.568565e-05 lr_backbone=3.568565e-05 elapsed=35:39:20 avg_step=4.19s recent_step=4.15s eta=22:31:45 +2026-09-21 18:06:38 | INFO | train rank=0 | step=30650/50000 progress=61.3% loss=0.329968 lr=3.565397e-05 lr_backbone=3.565397e-05 elapsed=35:40:01 avg_step=4.19s recent_step=4.10s eta=22:31:02 +2026-09-21 18:07:19 | INFO | train rank=0 | step=30660/50000 progress=61.3% loss=0.312717 lr=3.562230e-05 lr_backbone=3.562230e-05 elapsed=35:40:43 avg_step=4.19s recent_step=4.13s eta=22:30:20 +2026-09-21 18:08:01 | INFO | train rank=0 | step=30670/50000 progress=61.3% loss=0.319743 lr=3.559063e-05 lr_backbone=3.559063e-05 elapsed=35:41:24 avg_step=4.19s recent_step=4.13s eta=22:29:38 +2026-09-21 18:08:42 | INFO | train rank=0 | step=30680/50000 progress=61.4% loss=0.291235 lr=3.555896e-05 lr_backbone=3.555896e-05 elapsed=35:42:05 avg_step=4.19s recent_step=4.15s eta=22:28:56 +2026-09-21 18:09:24 | INFO | train rank=0 | step=30690/50000 progress=61.4% loss=0.499547 lr=3.552731e-05 lr_backbone=3.552731e-05 elapsed=35:42:47 avg_step=4.19s recent_step=4.16s eta=22:28:14 +2026-09-21 18:10:05 | INFO | train rank=0 | step=30700/50000 progress=61.4% loss=0.339290 lr=3.549566e-05 lr_backbone=3.549566e-05 elapsed=35:43:28 avg_step=4.19s recent_step=4.14s eta=22:27:31 +2026-09-21 18:10:46 | INFO | train rank=0 | step=30710/50000 progress=61.4% loss=0.314768 lr=3.546401e-05 lr_backbone=3.546401e-05 elapsed=35:44:09 avg_step=4.19s recent_step=4.10s eta=22:26:49 +2026-09-21 18:11:27 | INFO | train rank=0 | step=30720/50000 progress=61.4% loss=0.347964 lr=3.543237e-05 lr_backbone=3.543237e-05 elapsed=35:44:51 avg_step=4.19s recent_step=4.11s eta=22:26:07 +2026-09-21 18:12:08 | INFO | train rank=0 | step=30730/50000 progress=61.5% loss=0.335111 lr=3.540074e-05 lr_backbone=3.540074e-05 elapsed=35:45:32 avg_step=4.19s recent_step=4.13s eta=22:25:24 +2026-09-21 18:12:50 | INFO | train rank=0 | step=30740/50000 progress=61.5% loss=0.311081 lr=3.536912e-05 lr_backbone=3.536912e-05 elapsed=35:46:13 avg_step=4.19s recent_step=4.14s eta=22:24:42 +2026-09-21 18:13:32 | INFO | train rank=0 | step=30750/50000 progress=61.5% loss=0.305302 lr=3.533750e-05 lr_backbone=3.533750e-05 elapsed=35:46:55 avg_step=4.19s recent_step=4.17s eta=22:24:00 +2026-09-21 18:14:13 | INFO | train rank=0 | step=30760/50000 progress=61.5% loss=0.337170 lr=3.530589e-05 lr_backbone=3.530589e-05 elapsed=35:47:36 avg_step=4.19s recent_step=4.13s eta=22:23:18 +2026-09-21 18:14:54 | INFO | train rank=0 | step=30770/50000 progress=61.5% loss=0.341112 lr=3.527428e-05 lr_backbone=3.527428e-05 elapsed=35:48:17 avg_step=4.19s recent_step=4.10s eta=22:22:35 +2026-09-21 18:15:35 | INFO | train rank=0 | step=30780/50000 progress=61.6% loss=0.350629 lr=3.524268e-05 lr_backbone=3.524268e-05 elapsed=35:48:58 avg_step=4.19s recent_step=4.13s eta=22:21:53 +2026-09-21 18:16:16 | INFO | train rank=0 | step=30790/50000 progress=61.6% loss=0.428236 lr=3.521109e-05 lr_backbone=3.521109e-05 elapsed=35:49:40 avg_step=4.19s recent_step=4.12s eta=22:21:11 +2026-09-21 18:16:58 | INFO | train rank=0 | step=30800/50000 progress=61.6% loss=0.415375 lr=3.517950e-05 lr_backbone=3.517950e-05 elapsed=35:50:21 avg_step=4.19s recent_step=4.14s eta=22:20:29 +2026-09-21 18:17:39 | INFO | train rank=0 | step=30810/50000 progress=61.6% loss=0.339754 lr=3.514792e-05 lr_backbone=3.514792e-05 elapsed=35:51:02 avg_step=4.19s recent_step=4.14s eta=22:19:46 +2026-09-21 18:18:21 | INFO | train rank=0 | step=30820/50000 progress=61.6% loss=0.319887 lr=3.511635e-05 lr_backbone=3.511635e-05 elapsed=35:51:44 avg_step=4.19s recent_step=4.15s eta=22:19:04 +2026-09-21 18:19:02 | INFO | train rank=0 | step=30830/50000 progress=61.7% loss=0.295089 lr=3.508478e-05 lr_backbone=3.508478e-05 elapsed=35:52:25 avg_step=4.19s recent_step=4.11s eta=22:18:22 +2026-09-21 18:19:43 | INFO | train rank=0 | step=30840/50000 progress=61.7% loss=0.413237 lr=3.505322e-05 lr_backbone=3.505322e-05 elapsed=35:53:06 avg_step=4.19s recent_step=4.12s eta=22:17:39 +2026-09-21 18:20:24 | INFO | train rank=0 | step=30850/50000 progress=61.7% loss=0.323901 lr=3.502167e-05 lr_backbone=3.502167e-05 elapsed=35:53:48 avg_step=4.19s recent_step=4.15s eta=22:16:57 +2026-09-21 18:21:06 | INFO | train rank=0 | step=30860/50000 progress=61.7% loss=0.380760 lr=3.499012e-05 lr_backbone=3.499012e-05 elapsed=35:54:29 avg_step=4.19s recent_step=4.12s eta=22:16:15 +2026-09-21 18:21:47 | INFO | train rank=0 | step=30870/50000 progress=61.7% loss=0.337422 lr=3.495858e-05 lr_backbone=3.495858e-05 elapsed=35:55:10 avg_step=4.19s recent_step=4.15s eta=22:15:33 +2026-09-21 18:22:29 | INFO | train rank=0 | step=30880/50000 progress=61.8% loss=0.296077 lr=3.492704e-05 lr_backbone=3.492704e-05 elapsed=35:55:52 avg_step=4.19s recent_step=4.16s eta=22:14:51 +2026-09-21 18:23:10 | INFO | train rank=0 | step=30890/50000 progress=61.8% loss=0.384092 lr=3.489552e-05 lr_backbone=3.489552e-05 elapsed=35:56:33 avg_step=4.19s recent_step=4.11s eta=22:14:08 +2026-09-21 18:23:51 | INFO | train rank=0 | step=30900/50000 progress=61.8% loss=0.356107 lr=3.486399e-05 lr_backbone=3.486399e-05 elapsed=35:57:15 avg_step=4.19s recent_step=4.14s eta=22:13:26 +2026-09-21 18:24:33 | INFO | train rank=0 | step=30910/50000 progress=61.8% loss=0.294634 lr=3.483248e-05 lr_backbone=3.483248e-05 elapsed=35:57:56 avg_step=4.19s recent_step=4.13s eta=22:12:44 +2026-09-21 18:25:14 | INFO | train rank=0 | step=30920/50000 progress=61.8% loss=0.349887 lr=3.480097e-05 lr_backbone=3.480097e-05 elapsed=35:58:37 avg_step=4.19s recent_step=4.13s eta=22:12:02 +2026-09-21 18:25:55 | INFO | train rank=0 | step=30930/50000 progress=61.9% loss=0.351897 lr=3.476947e-05 lr_backbone=3.476947e-05 elapsed=35:59:19 avg_step=4.19s recent_step=4.15s eta=22:11:20 +2026-09-21 18:26:37 | INFO | train rank=0 | step=30940/50000 progress=61.9% loss=0.327325 lr=3.473798e-05 lr_backbone=3.473798e-05 elapsed=36:00:00 avg_step=4.19s recent_step=4.12s eta=22:10:37 +2026-09-21 18:27:18 | INFO | train rank=0 | step=30950/50000 progress=61.9% loss=0.299336 lr=3.470649e-05 lr_backbone=3.470649e-05 elapsed=36:00:41 avg_step=4.19s recent_step=4.13s eta=22:09:55 +2026-09-21 18:27:59 | INFO | train rank=0 | step=30960/50000 progress=61.9% loss=0.312625 lr=3.467501e-05 lr_backbone=3.467501e-05 elapsed=36:01:22 avg_step=4.19s recent_step=4.13s eta=22:09:13 +2026-09-21 18:28:41 | INFO | train rank=0 | step=30970/50000 progress=61.9% loss=0.299411 lr=3.464353e-05 lr_backbone=3.464353e-05 elapsed=36:02:04 avg_step=4.19s recent_step=4.15s eta=22:08:31 +2026-09-21 18:29:22 | INFO | train rank=0 | step=30980/50000 progress=62.0% loss=0.328225 lr=3.461207e-05 lr_backbone=3.461207e-05 elapsed=36:02:45 avg_step=4.19s recent_step=4.13s eta=22:07:48 +2026-09-21 18:30:03 | INFO | train rank=0 | step=30990/50000 progress=62.0% loss=0.391206 lr=3.458060e-05 lr_backbone=3.458060e-05 elapsed=36:03:27 avg_step=4.19s recent_step=4.15s eta=22:07:06 +2026-09-21 18:30:45 | INFO | train rank=0 | step=31000/50000 progress=62.0% loss=0.308446 lr=3.454915e-05 lr_backbone=3.454915e-05 elapsed=36:04:08 avg_step=4.19s recent_step=4.14s eta=22:06:24 +2026-09-21 18:31:26 | INFO | train rank=0 | step=31010/50000 progress=62.0% loss=0.292227 lr=3.451770e-05 lr_backbone=3.451770e-05 elapsed=36:04:49 avg_step=4.19s recent_step=4.09s eta=22:05:42 +2026-09-21 18:32:07 | INFO | train rank=0 | step=31020/50000 progress=62.0% loss=0.289636 lr=3.448626e-05 lr_backbone=3.448626e-05 elapsed=36:05:30 avg_step=4.19s recent_step=4.13s eta=22:04:59 +2026-09-21 18:32:49 | INFO | train rank=0 | step=31030/50000 progress=62.1% loss=0.316122 lr=3.445483e-05 lr_backbone=3.445483e-05 elapsed=36:06:12 avg_step=4.19s recent_step=4.15s eta=22:04:17 +2026-09-21 18:33:30 | INFO | train rank=0 | step=31040/50000 progress=62.1% loss=0.339381 lr=3.442340e-05 lr_backbone=3.442340e-05 elapsed=36:06:53 avg_step=4.19s recent_step=4.15s eta=22:03:35 +2026-09-21 18:34:11 | INFO | train rank=0 | step=31050/50000 progress=62.1% loss=0.344046 lr=3.439198e-05 lr_backbone=3.439198e-05 elapsed=36:07:35 avg_step=4.19s recent_step=4.13s eta=22:02:53 +2026-09-21 18:34:53 | INFO | train rank=0 | step=31060/50000 progress=62.1% loss=0.285876 lr=3.436057e-05 lr_backbone=3.436057e-05 elapsed=36:08:16 avg_step=4.19s recent_step=4.17s eta=22:02:11 +2026-09-21 18:35:34 | INFO | train rank=0 | step=31070/50000 progress=62.1% loss=0.377742 lr=3.432916e-05 lr_backbone=3.432916e-05 elapsed=36:08:57 avg_step=4.19s recent_step=4.10s eta=22:01:29 +2026-09-21 18:36:15 | INFO | train rank=0 | step=31080/50000 progress=62.2% loss=0.284895 lr=3.429776e-05 lr_backbone=3.429776e-05 elapsed=36:09:39 avg_step=4.19s recent_step=4.13s eta=22:00:46 +2026-09-21 18:36:57 | INFO | train rank=0 | step=31090/50000 progress=62.2% loss=0.343822 lr=3.426637e-05 lr_backbone=3.426637e-05 elapsed=36:10:20 avg_step=4.19s recent_step=4.15s eta=22:00:04 +2026-09-21 18:37:38 | INFO | train rank=0 | step=31100/50000 progress=62.2% loss=0.359469 lr=3.423498e-05 lr_backbone=3.423498e-05 elapsed=36:11:02 avg_step=4.19s recent_step=4.15s eta=21:59:22 +2026-09-21 18:38:20 | INFO | train rank=0 | step=31110/50000 progress=62.2% loss=0.376629 lr=3.420360e-05 lr_backbone=3.420360e-05 elapsed=36:11:43 avg_step=4.19s recent_step=4.17s eta=21:58:40 +2026-09-21 18:39:01 | INFO | train rank=0 | step=31120/50000 progress=62.2% loss=0.336972 lr=3.417223e-05 lr_backbone=3.417223e-05 elapsed=36:12:25 avg_step=4.19s recent_step=4.10s eta=21:57:58 +2026-09-21 18:39:42 | INFO | train rank=0 | step=31130/50000 progress=62.3% loss=0.326019 lr=3.414087e-05 lr_backbone=3.414087e-05 elapsed=36:13:06 avg_step=4.19s recent_step=4.11s eta=21:57:15 +2026-09-21 18:40:24 | INFO | train rank=0 | step=31140/50000 progress=62.3% loss=0.382010 lr=3.410951e-05 lr_backbone=3.410951e-05 elapsed=36:13:47 avg_step=4.19s recent_step=4.12s eta=21:56:33 +2026-09-21 18:41:05 | INFO | train rank=0 | step=31150/50000 progress=62.3% loss=0.455453 lr=3.407816e-05 lr_backbone=3.407816e-05 elapsed=36:14:28 avg_step=4.19s recent_step=4.15s eta=21:55:51 +2026-09-21 18:41:47 | INFO | train rank=0 | step=31160/50000 progress=62.3% loss=0.332860 lr=3.404681e-05 lr_backbone=3.404681e-05 elapsed=36:15:10 avg_step=4.19s recent_step=4.16s eta=21:55:09 +2026-09-21 18:42:28 | INFO | train rank=0 | step=31170/50000 progress=62.3% loss=0.358560 lr=3.401547e-05 lr_backbone=3.401547e-05 elapsed=36:15:51 avg_step=4.19s recent_step=4.14s eta=21:54:27 +2026-09-21 18:43:09 | INFO | train rank=0 | step=31180/50000 progress=62.4% loss=0.398379 lr=3.398414e-05 lr_backbone=3.398414e-05 elapsed=36:16:33 avg_step=4.19s recent_step=4.13s eta=21:53:45 +2026-09-21 18:43:50 | INFO | train rank=0 | step=31190/50000 progress=62.4% loss=0.335930 lr=3.395282e-05 lr_backbone=3.395282e-05 elapsed=36:17:14 avg_step=4.19s recent_step=4.10s eta=21:53:02 +2026-09-21 18:44:32 | INFO | train rank=0 | step=31200/50000 progress=62.4% loss=0.294780 lr=3.392150e-05 lr_backbone=3.392150e-05 elapsed=36:17:55 avg_step=4.19s recent_step=4.13s eta=21:52:20 +2026-09-21 18:45:13 | INFO | train rank=0 | step=31210/50000 progress=62.4% loss=0.344897 lr=3.389019e-05 lr_backbone=3.389019e-05 elapsed=36:18:37 avg_step=4.19s recent_step=4.15s eta=21:51:38 +2026-09-21 18:45:55 | INFO | train rank=0 | step=31220/50000 progress=62.4% loss=0.301880 lr=3.385889e-05 lr_backbone=3.385889e-05 elapsed=36:19:18 avg_step=4.19s recent_step=4.16s eta=21:50:56 +2026-09-21 18:46:36 | INFO | train rank=0 | step=31230/50000 progress=62.5% loss=0.296187 lr=3.382760e-05 lr_backbone=3.382760e-05 elapsed=36:20:00 avg_step=4.19s recent_step=4.16s eta=21:50:14 +2026-09-21 18:47:18 | INFO | train rank=0 | step=31240/50000 progress=62.5% loss=0.292407 lr=3.379631e-05 lr_backbone=3.379631e-05 elapsed=36:20:41 avg_step=4.19s recent_step=4.16s eta=21:49:32 +2026-09-21 18:47:59 | INFO | train rank=0 | step=31250/50000 progress=62.5% loss=0.332760 lr=3.376503e-05 lr_backbone=3.376503e-05 elapsed=36:21:22 avg_step=4.19s recent_step=4.11s eta=21:48:49 +2026-09-21 18:48:40 | INFO | train rank=0 | step=31260/50000 progress=62.5% loss=0.338212 lr=3.373375e-05 lr_backbone=3.373375e-05 elapsed=36:22:04 avg_step=4.19s recent_step=4.12s eta=21:48:07 +2026-09-21 18:49:22 | INFO | train rank=0 | step=31270/50000 progress=62.5% loss=0.317401 lr=3.370249e-05 lr_backbone=3.370249e-05 elapsed=36:22:45 avg_step=4.19s recent_step=4.15s eta=21:47:25 +2026-09-21 18:50:03 | INFO | train rank=0 | step=31280/50000 progress=62.6% loss=0.315355 lr=3.367123e-05 lr_backbone=3.367123e-05 elapsed=36:23:27 avg_step=4.19s recent_step=4.16s eta=21:46:43 +2026-09-21 18:50:45 | INFO | train rank=0 | step=31290/50000 progress=62.6% loss=0.344485 lr=3.363997e-05 lr_backbone=3.363997e-05 elapsed=36:24:08 avg_step=4.19s recent_step=4.15s eta=21:46:01 +2026-09-21 18:51:26 | INFO | train rank=0 | step=31300/50000 progress=62.6% loss=0.351443 lr=3.360873e-05 lr_backbone=3.360873e-05 elapsed=36:24:49 avg_step=4.19s recent_step=4.11s eta=21:45:18 +2026-09-21 18:52:07 | INFO | train rank=0 | step=31310/50000 progress=62.6% loss=0.269624 lr=3.357749e-05 lr_backbone=3.357749e-05 elapsed=36:25:30 avg_step=4.19s recent_step=4.10s eta=21:44:36 +2026-09-21 18:52:48 | INFO | train rank=0 | step=31320/50000 progress=62.6% loss=0.418635 lr=3.354626e-05 lr_backbone=3.354626e-05 elapsed=36:26:11 avg_step=4.19s recent_step=4.11s eta=21:43:54 +2026-09-21 18:53:29 | INFO | train rank=0 | step=31330/50000 progress=62.7% loss=0.319624 lr=3.351503e-05 lr_backbone=3.351503e-05 elapsed=36:26:53 avg_step=4.19s recent_step=4.13s eta=21:43:11 +2026-09-21 18:54:11 | INFO | train rank=0 | step=31340/50000 progress=62.7% loss=0.354897 lr=3.348382e-05 lr_backbone=3.348382e-05 elapsed=36:27:34 avg_step=4.19s recent_step=4.13s eta=21:42:29 +2026-09-21 18:54:52 | INFO | train rank=0 | step=31350/50000 progress=62.7% loss=0.296714 lr=3.345261e-05 lr_backbone=3.345261e-05 elapsed=36:28:15 avg_step=4.19s recent_step=4.13s eta=21:41:47 +2026-09-21 18:55:33 | INFO | train rank=0 | step=31360/50000 progress=62.7% loss=0.318794 lr=3.342141e-05 lr_backbone=3.342141e-05 elapsed=36:28:57 avg_step=4.19s recent_step=4.13s eta=21:41:05 +2026-09-21 18:56:14 | INFO | train rank=0 | step=31370/50000 progress=62.7% loss=0.345273 lr=3.339021e-05 lr_backbone=3.339021e-05 elapsed=36:29:38 avg_step=4.19s recent_step=4.10s eta=21:40:22 +2026-09-21 18:56:55 | INFO | train rank=0 | step=31380/50000 progress=62.8% loss=0.341487 lr=3.335902e-05 lr_backbone=3.335902e-05 elapsed=36:30:19 avg_step=4.19s recent_step=4.11s eta=21:39:40 +2026-09-21 18:57:37 | INFO | train rank=0 | step=31390/50000 progress=62.8% loss=0.301396 lr=3.332784e-05 lr_backbone=3.332784e-05 elapsed=36:31:00 avg_step=4.19s recent_step=4.13s eta=21:38:58 +2026-09-21 18:58:18 | INFO | train rank=0 | step=31400/50000 progress=62.8% loss=0.304264 lr=3.329667e-05 lr_backbone=3.329667e-05 elapsed=36:31:41 avg_step=4.19s recent_step=4.13s eta=21:38:15 +2026-09-21 18:58:59 | INFO | train rank=0 | step=31410/50000 progress=62.8% loss=0.293715 lr=3.326550e-05 lr_backbone=3.326550e-05 elapsed=36:32:23 avg_step=4.19s recent_step=4.16s eta=21:37:33 +2026-09-21 18:59:41 | INFO | train rank=0 | step=31420/50000 progress=62.8% loss=0.351859 lr=3.323434e-05 lr_backbone=3.323434e-05 elapsed=36:33:04 avg_step=4.19s recent_step=4.15s eta=21:36:51 +2026-09-21 19:00:22 | INFO | train rank=0 | step=31430/50000 progress=62.9% loss=0.320412 lr=3.320319e-05 lr_backbone=3.320319e-05 elapsed=36:33:45 avg_step=4.19s recent_step=4.09s eta=21:36:09 +2026-09-21 19:01:03 | INFO | train rank=0 | step=31440/50000 progress=62.9% loss=0.313855 lr=3.317205e-05 lr_backbone=3.317205e-05 elapsed=36:34:26 avg_step=4.19s recent_step=4.11s eta=21:35:26 +2026-09-21 19:01:44 | INFO | train rank=0 | step=31450/50000 progress=62.9% loss=0.306626 lr=3.314091e-05 lr_backbone=3.314091e-05 elapsed=36:35:08 avg_step=4.19s recent_step=4.14s eta=21:34:44 +2026-09-21 19:02:26 | INFO | train rank=0 | step=31460/50000 progress=62.9% loss=0.305980 lr=3.310978e-05 lr_backbone=3.310978e-05 elapsed=36:35:49 avg_step=4.19s recent_step=4.13s eta=21:34:02 +2026-09-21 19:03:07 | INFO | train rank=0 | step=31470/50000 progress=62.9% loss=0.362649 lr=3.307866e-05 lr_backbone=3.307866e-05 elapsed=36:36:30 avg_step=4.19s recent_step=4.14s eta=21:33:20 +2026-09-21 19:03:48 | INFO | train rank=0 | step=31480/50000 progress=63.0% loss=0.345284 lr=3.304755e-05 lr_backbone=3.304755e-05 elapsed=36:37:12 avg_step=4.19s recent_step=4.14s eta=21:32:38 +2026-09-21 19:04:29 | INFO | train rank=0 | step=31490/50000 progress=63.0% loss=0.314023 lr=3.301644e-05 lr_backbone=3.301644e-05 elapsed=36:37:53 avg_step=4.19s recent_step=4.08s eta=21:31:55 +2026-09-21 19:05:10 | INFO | train rank=0 | step=31500/50000 progress=63.0% loss=0.343553 lr=3.298534e-05 lr_backbone=3.298534e-05 elapsed=36:38:34 avg_step=4.19s recent_step=4.10s eta=21:31:13 +2026-09-21 19:05:51 | INFO | train rank=0 | step=31510/50000 progress=63.0% loss=0.415225 lr=3.295425e-05 lr_backbone=3.295425e-05 elapsed=36:39:15 avg_step=4.19s recent_step=4.13s eta=21:30:31 +2026-09-21 19:06:33 | INFO | train rank=0 | step=31520/50000 progress=63.0% loss=0.374379 lr=3.292316e-05 lr_backbone=3.292316e-05 elapsed=36:39:56 avg_step=4.19s recent_step=4.11s eta=21:29:48 +2026-09-21 19:07:14 | INFO | train rank=0 | step=31530/50000 progress=63.1% loss=0.383377 lr=3.289209e-05 lr_backbone=3.289209e-05 elapsed=36:40:37 avg_step=4.19s recent_step=4.13s eta=21:29:06 +2026-09-21 19:07:55 | INFO | train rank=0 | step=31540/50000 progress=63.1% loss=0.341149 lr=3.286102e-05 lr_backbone=3.286102e-05 elapsed=36:41:19 avg_step=4.19s recent_step=4.13s eta=21:28:24 +2026-09-21 19:08:36 | INFO | train rank=0 | step=31550/50000 progress=63.1% loss=0.347546 lr=3.282996e-05 lr_backbone=3.282996e-05 elapsed=36:41:59 avg_step=4.19s recent_step=4.08s eta=21:27:41 +2026-09-21 19:09:17 | INFO | train rank=0 | step=31560/50000 progress=63.1% loss=0.327777 lr=3.279890e-05 lr_backbone=3.279890e-05 elapsed=36:42:40 avg_step=4.19s recent_step=4.11s eta=21:26:59 +2026-09-21 19:09:59 | INFO | train rank=0 | step=31570/50000 progress=63.1% loss=0.300049 lr=3.276785e-05 lr_backbone=3.276785e-05 elapsed=36:43:22 avg_step=4.19s recent_step=4.14s eta=21:26:17 +2026-09-21 19:10:40 | INFO | train rank=0 | step=31580/50000 progress=63.2% loss=0.326666 lr=3.273681e-05 lr_backbone=3.273681e-05 elapsed=36:44:03 avg_step=4.19s recent_step=4.14s eta=21:25:35 +2026-09-21 19:11:21 | INFO | train rank=0 | step=31590/50000 progress=63.2% loss=0.428686 lr=3.270578e-05 lr_backbone=3.270578e-05 elapsed=36:44:44 avg_step=4.19s recent_step=4.12s eta=21:24:52 +2026-09-21 19:12:03 | INFO | train rank=0 | step=31600/50000 progress=63.2% loss=0.378221 lr=3.267476e-05 lr_backbone=3.267476e-05 elapsed=36:45:26 avg_step=4.19s recent_step=4.16s eta=21:24:10 +2026-09-21 19:12:44 | INFO | train rank=0 | step=31610/50000 progress=63.2% loss=0.413724 lr=3.264374e-05 lr_backbone=3.264374e-05 elapsed=36:46:07 avg_step=4.19s recent_step=4.09s eta=21:23:28 +2026-09-21 19:13:25 | INFO | train rank=0 | step=31620/50000 progress=63.2% loss=0.430634 lr=3.261273e-05 lr_backbone=3.261273e-05 elapsed=36:46:48 avg_step=4.19s recent_step=4.09s eta=21:22:46 +2026-09-21 19:14:06 | INFO | train rank=0 | step=31630/50000 progress=63.3% loss=0.353456 lr=3.258173e-05 lr_backbone=3.258173e-05 elapsed=36:47:29 avg_step=4.19s recent_step=4.15s eta=21:22:03 +2026-09-21 19:14:48 | INFO | train rank=0 | step=31640/50000 progress=63.3% loss=0.372288 lr=3.255074e-05 lr_backbone=3.255074e-05 elapsed=36:48:11 avg_step=4.19s recent_step=4.17s eta=21:21:21 +2026-09-21 19:15:29 | INFO | train rank=0 | step=31650/50000 progress=63.3% loss=0.365436 lr=3.251975e-05 lr_backbone=3.251975e-05 elapsed=36:48:52 avg_step=4.19s recent_step=4.11s eta=21:20:39 +2026-09-21 19:16:10 | INFO | train rank=0 | step=31660/50000 progress=63.3% loss=0.375131 lr=3.248877e-05 lr_backbone=3.248877e-05 elapsed=36:49:34 avg_step=4.19s recent_step=4.15s eta=21:19:57 +2026-09-21 19:16:51 | INFO | train rank=0 | step=31670/50000 progress=63.3% loss=0.291939 lr=3.245780e-05 lr_backbone=3.245780e-05 elapsed=36:50:15 avg_step=4.19s recent_step=4.10s eta=21:19:15 +2026-09-21 19:17:32 | INFO | train rank=0 | step=31680/50000 progress=63.4% loss=0.331997 lr=3.242684e-05 lr_backbone=3.242684e-05 elapsed=36:50:56 avg_step=4.19s recent_step=4.10s eta=21:18:32 +2026-09-21 19:18:13 | INFO | train rank=0 | step=31690/50000 progress=63.4% loss=0.323860 lr=3.239588e-05 lr_backbone=3.239588e-05 elapsed=36:51:37 avg_step=4.19s recent_step=4.10s eta=21:17:50 +2026-09-21 19:18:55 | INFO | train rank=0 | step=31700/50000 progress=63.4% loss=0.464426 lr=3.236493e-05 lr_backbone=3.236493e-05 elapsed=36:52:18 avg_step=4.19s recent_step=4.15s eta=21:17:08 +2026-09-21 19:19:36 | INFO | train rank=0 | step=31710/50000 progress=63.4% loss=0.339004 lr=3.233399e-05 lr_backbone=3.233399e-05 elapsed=36:53:00 avg_step=4.19s recent_step=4.14s eta=21:16:26 +2026-09-21 19:20:18 | INFO | train rank=0 | step=31720/50000 progress=63.4% loss=0.351088 lr=3.230306e-05 lr_backbone=3.230306e-05 elapsed=36:53:41 avg_step=4.19s recent_step=4.16s eta=21:15:44 +2026-09-21 19:20:59 | INFO | train rank=0 | step=31730/50000 progress=63.5% loss=0.361831 lr=3.227213e-05 lr_backbone=3.227213e-05 elapsed=36:54:22 avg_step=4.19s recent_step=4.11s eta=21:15:01 +2026-09-21 19:21:40 | INFO | train rank=0 | step=31740/50000 progress=63.5% loss=0.350425 lr=3.224122e-05 lr_backbone=3.224122e-05 elapsed=36:55:04 avg_step=4.19s recent_step=4.14s eta=21:14:19 +2026-09-21 19:22:22 | INFO | train rank=0 | step=31750/50000 progress=63.5% loss=0.287017 lr=3.221031e-05 lr_backbone=3.221031e-05 elapsed=36:55:45 avg_step=4.19s recent_step=4.13s eta=21:13:37 +2026-09-21 19:23:03 | INFO | train rank=0 | step=31760/50000 progress=63.5% loss=0.328118 lr=3.217941e-05 lr_backbone=3.217941e-05 elapsed=36:56:27 avg_step=4.19s recent_step=4.18s eta=21:12:55 +2026-09-21 19:23:45 | INFO | train rank=0 | step=31770/50000 progress=63.5% loss=0.321950 lr=3.214851e-05 lr_backbone=3.214851e-05 elapsed=36:57:08 avg_step=4.19s recent_step=4.17s eta=21:12:13 +2026-09-21 19:24:27 | INFO | train rank=0 | step=31780/50000 progress=63.6% loss=0.321447 lr=3.211763e-05 lr_backbone=3.211763e-05 elapsed=36:57:50 avg_step=4.19s recent_step=4.16s eta=21:11:31 +2026-09-21 19:25:08 | INFO | train rank=0 | step=31790/50000 progress=63.6% loss=0.338628 lr=3.208675e-05 lr_backbone=3.208675e-05 elapsed=36:58:31 avg_step=4.19s recent_step=4.10s eta=21:10:49 +2026-09-21 19:25:49 | INFO | train rank=0 | step=31800/50000 progress=63.6% loss=0.326501 lr=3.205588e-05 lr_backbone=3.205588e-05 elapsed=36:59:12 avg_step=4.19s recent_step=4.14s eta=21:10:06 +2026-09-21 19:26:30 | INFO | train rank=0 | step=31810/50000 progress=63.6% loss=0.348004 lr=3.202502e-05 lr_backbone=3.202502e-05 elapsed=36:59:54 avg_step=4.19s recent_step=4.13s eta=21:09:24 +2026-09-21 19:27:12 | INFO | train rank=0 | step=31820/50000 progress=63.6% loss=0.354896 lr=3.199416e-05 lr_backbone=3.199416e-05 elapsed=37:00:35 avg_step=4.19s recent_step=4.16s eta=21:08:42 +2026-09-21 19:27:54 | INFO | train rank=0 | step=31830/50000 progress=63.7% loss=0.306146 lr=3.196331e-05 lr_backbone=3.196331e-05 elapsed=37:01:17 avg_step=4.19s recent_step=4.17s eta=21:08:00 +2026-09-21 19:28:35 | INFO | train rank=0 | step=31840/50000 progress=63.7% loss=0.344475 lr=3.193248e-05 lr_backbone=3.193248e-05 elapsed=37:01:58 avg_step=4.19s recent_step=4.15s eta=21:07:18 +2026-09-21 19:29:16 | INFO | train rank=0 | step=31850/50000 progress=63.7% loss=0.310512 lr=3.190164e-05 lr_backbone=3.190164e-05 elapsed=37:02:39 avg_step=4.19s recent_step=4.10s eta=21:06:36 +2026-09-21 19:29:57 | INFO | train rank=0 | step=31860/50000 progress=63.7% loss=0.379636 lr=3.187082e-05 lr_backbone=3.187082e-05 elapsed=37:03:21 avg_step=4.19s recent_step=4.12s eta=21:05:54 +2026-09-21 19:30:39 | INFO | train rank=0 | step=31870/50000 progress=63.7% loss=0.337575 lr=3.184001e-05 lr_backbone=3.184001e-05 elapsed=37:04:02 avg_step=4.19s recent_step=4.13s eta=21:05:11 +2026-09-21 19:31:20 | INFO | train rank=0 | step=31880/50000 progress=63.8% loss=0.347080 lr=3.180920e-05 lr_backbone=3.180920e-05 elapsed=37:04:43 avg_step=4.19s recent_step=4.13s eta=21:04:29 +2026-09-21 19:32:01 | INFO | train rank=0 | step=31890/50000 progress=63.8% loss=0.299256 lr=3.177840e-05 lr_backbone=3.177840e-05 elapsed=37:05:25 avg_step=4.19s recent_step=4.15s eta=21:03:47 +2026-09-21 19:32:43 | INFO | train rank=0 | step=31900/50000 progress=63.8% loss=0.320701 lr=3.174761e-05 lr_backbone=3.174761e-05 elapsed=37:06:06 avg_step=4.19s recent_step=4.15s eta=21:03:05 +2026-09-21 19:33:24 | INFO | train rank=0 | step=31910/50000 progress=63.8% loss=0.293889 lr=3.171683e-05 lr_backbone=3.171683e-05 elapsed=37:06:47 avg_step=4.19s recent_step=4.11s eta=21:02:23 +2026-09-21 19:34:05 | INFO | train rank=0 | step=31920/50000 progress=63.8% loss=0.346104 lr=3.168605e-05 lr_backbone=3.168605e-05 elapsed=37:07:29 avg_step=4.19s recent_step=4.13s eta=21:01:41 +2026-09-21 19:34:47 | INFO | train rank=0 | step=31930/50000 progress=63.9% loss=0.372753 lr=3.165528e-05 lr_backbone=3.165528e-05 elapsed=37:08:10 avg_step=4.19s recent_step=4.13s eta=21:00:58 +2026-09-21 19:35:28 | INFO | train rank=0 | step=31940/50000 progress=63.9% loss=0.358414 lr=3.162452e-05 lr_backbone=3.162452e-05 elapsed=37:08:51 avg_step=4.19s recent_step=4.14s eta=21:00:16 +2026-09-21 19:36:10 | INFO | train rank=0 | step=31950/50000 progress=63.9% loss=0.338572 lr=3.159377e-05 lr_backbone=3.159377e-05 elapsed=37:09:33 avg_step=4.19s recent_step=4.16s eta=20:59:34 +2026-09-21 19:36:51 | INFO | train rank=0 | step=31960/50000 progress=63.9% loss=0.346863 lr=3.156303e-05 lr_backbone=3.156303e-05 elapsed=37:10:15 avg_step=4.19s recent_step=4.18s eta=20:58:52 +2026-09-21 19:37:32 | INFO | train rank=0 | step=31970/50000 progress=63.9% loss=0.334759 lr=3.153229e-05 lr_backbone=3.153229e-05 elapsed=37:10:56 avg_step=4.19s recent_step=4.10s eta=20:58:10 +2026-09-21 19:38:14 | INFO | train rank=0 | step=31980/50000 progress=64.0% loss=0.344078 lr=3.150157e-05 lr_backbone=3.150157e-05 elapsed=37:11:37 avg_step=4.19s recent_step=4.14s eta=20:57:28 +2026-09-21 19:38:55 | INFO | train rank=0 | step=31990/50000 progress=64.0% loss=0.269331 lr=3.147085e-05 lr_backbone=3.147085e-05 elapsed=37:12:18 avg_step=4.19s recent_step=4.13s eta=20:56:46 +2026-09-21 19:39:36 | INFO | train rank=0 | step=32000/50000 progress=64.0% loss=0.321270 lr=3.144014e-05 lr_backbone=3.144014e-05 elapsed=37:13:00 avg_step=4.19s recent_step=4.14s eta=20:56:03 +2026-09-21 19:40:18 | INFO | train rank=0 | step=32010/50000 progress=64.0% loss=0.306467 lr=3.140943e-05 lr_backbone=3.140943e-05 elapsed=37:13:41 avg_step=4.19s recent_step=4.15s eta=20:55:21 +2026-09-21 19:41:00 | INFO | train rank=0 | step=32020/50000 progress=64.0% loss=0.347989 lr=3.137874e-05 lr_backbone=3.137874e-05 elapsed=37:14:23 avg_step=4.19s recent_step=4.16s eta=20:54:39 +2026-09-21 19:41:41 | INFO | train rank=0 | step=32030/50000 progress=64.1% loss=0.363171 lr=3.134805e-05 lr_backbone=3.134805e-05 elapsed=37:15:04 avg_step=4.19s recent_step=4.10s eta=20:53:57 +2026-09-21 19:42:22 | INFO | train rank=0 | step=32040/50000 progress=64.1% loss=0.380075 lr=3.131738e-05 lr_backbone=3.131738e-05 elapsed=37:15:45 avg_step=4.19s recent_step=4.14s eta=20:53:15 +2026-09-21 19:43:03 | INFO | train rank=0 | step=32050/50000 progress=64.1% loss=0.321928 lr=3.128671e-05 lr_backbone=3.128671e-05 elapsed=37:16:27 avg_step=4.19s recent_step=4.13s eta=20:52:33 +2026-09-21 19:43:45 | INFO | train rank=0 | step=32060/50000 progress=64.1% loss=0.339591 lr=3.125604e-05 lr_backbone=3.125604e-05 elapsed=37:17:08 avg_step=4.19s recent_step=4.16s eta=20:51:51 +2026-09-21 19:44:26 | INFO | train rank=0 | step=32070/50000 progress=64.1% loss=0.345223 lr=3.122539e-05 lr_backbone=3.122539e-05 elapsed=37:17:50 avg_step=4.19s recent_step=4.14s eta=20:51:08 +2026-09-21 19:45:08 | INFO | train rank=0 | step=32080/50000 progress=64.2% loss=0.390116 lr=3.119474e-05 lr_backbone=3.119474e-05 elapsed=37:18:31 avg_step=4.19s recent_step=4.15s eta=20:50:26 +2026-09-21 19:45:49 | INFO | train rank=0 | step=32090/50000 progress=64.2% loss=0.373277 lr=3.116411e-05 lr_backbone=3.116411e-05 elapsed=37:19:12 avg_step=4.19s recent_step=4.11s eta=20:49:44 +2026-09-21 19:46:30 | INFO | train rank=0 | step=32100/50000 progress=64.2% loss=0.366687 lr=3.113348e-05 lr_backbone=3.113348e-05 elapsed=37:19:53 avg_step=4.19s recent_step=4.12s eta=20:49:02 +2026-09-21 19:47:11 | INFO | train rank=0 | step=32110/50000 progress=64.2% loss=0.360918 lr=3.110286e-05 lr_backbone=3.110286e-05 elapsed=37:20:35 avg_step=4.19s recent_step=4.14s eta=20:48:20 +2026-09-21 19:47:53 | INFO | train rank=0 | step=32120/50000 progress=64.2% loss=0.315067 lr=3.107225e-05 lr_backbone=3.107225e-05 elapsed=37:21:16 avg_step=4.19s recent_step=4.16s eta=20:47:38 +2026-09-21 19:48:34 | INFO | train rank=0 | step=32130/50000 progress=64.3% loss=0.358343 lr=3.104164e-05 lr_backbone=3.104164e-05 elapsed=37:21:58 avg_step=4.19s recent_step=4.15s eta=20:46:56 +2026-09-21 19:49:16 | INFO | train rank=0 | step=32140/50000 progress=64.3% loss=0.352904 lr=3.101105e-05 lr_backbone=3.101105e-05 elapsed=37:22:39 avg_step=4.19s recent_step=4.16s eta=20:46:14 +2026-09-21 19:49:57 | INFO | train rank=0 | step=32150/50000 progress=64.3% loss=0.375095 lr=3.098046e-05 lr_backbone=3.098046e-05 elapsed=37:23:20 avg_step=4.19s recent_step=4.10s eta=20:45:31 +2026-09-21 19:50:38 | INFO | train rank=0 | step=32160/50000 progress=64.3% loss=0.321139 lr=3.094988e-05 lr_backbone=3.094988e-05 elapsed=37:24:02 avg_step=4.19s recent_step=4.14s eta=20:44:49 +2026-09-21 19:51:20 | INFO | train rank=0 | step=32170/50000 progress=64.3% loss=0.296720 lr=3.091931e-05 lr_backbone=3.091931e-05 elapsed=37:24:43 avg_step=4.19s recent_step=4.13s eta=20:44:07 +2026-09-21 19:52:01 | INFO | train rank=0 | step=32180/50000 progress=64.4% loss=0.364029 lr=3.088874e-05 lr_backbone=3.088874e-05 elapsed=37:25:25 avg_step=4.19s recent_step=4.16s eta=20:43:25 +2026-09-21 19:52:43 | INFO | train rank=0 | step=32190/50000 progress=64.4% loss=0.344959 lr=3.085819e-05 lr_backbone=3.085819e-05 elapsed=37:26:06 avg_step=4.19s recent_step=4.15s eta=20:42:43 +2026-09-21 19:53:24 | INFO | train rank=0 | step=32200/50000 progress=64.4% loss=0.338231 lr=3.082764e-05 lr_backbone=3.082764e-05 elapsed=37:26:48 avg_step=4.19s recent_step=4.14s eta=20:42:01 +2026-09-21 19:54:05 | INFO | train rank=0 | step=32210/50000 progress=64.4% loss=0.323581 lr=3.079711e-05 lr_backbone=3.079711e-05 elapsed=37:27:29 avg_step=4.19s recent_step=4.11s eta=20:41:19 +2026-09-21 19:54:47 | INFO | train rank=0 | step=32220/50000 progress=64.4% loss=0.385033 lr=3.076658e-05 lr_backbone=3.076658e-05 elapsed=37:28:10 avg_step=4.19s recent_step=4.13s eta=20:40:36 +2026-09-21 19:55:28 | INFO | train rank=0 | step=32230/50000 progress=64.5% loss=0.321099 lr=3.073606e-05 lr_backbone=3.073606e-05 elapsed=37:28:52 avg_step=4.19s recent_step=4.15s eta=20:39:54 +2026-09-21 19:56:09 | INFO | train rank=0 | step=32240/50000 progress=64.5% loss=0.368514 lr=3.070555e-05 lr_backbone=3.070555e-05 elapsed=37:29:33 avg_step=4.19s recent_step=4.12s eta=20:39:12 +2026-09-21 19:56:51 | INFO | train rank=0 | step=32250/50000 progress=64.5% loss=0.324053 lr=3.067504e-05 lr_backbone=3.067504e-05 elapsed=37:30:14 avg_step=4.19s recent_step=4.13s eta=20:38:30 +2026-09-21 19:57:32 | INFO | train rank=0 | step=32260/50000 progress=64.5% loss=0.351500 lr=3.064455e-05 lr_backbone=3.064455e-05 elapsed=37:30:56 avg_step=4.19s recent_step=4.16s eta=20:37:48 +2026-09-21 19:58:13 | INFO | train rank=0 | step=32270/50000 progress=64.5% loss=0.412273 lr=3.061406e-05 lr_backbone=3.061406e-05 elapsed=37:31:37 avg_step=4.19s recent_step=4.10s eta=20:37:06 +2026-09-21 19:58:55 | INFO | train rank=0 | step=32280/50000 progress=64.6% loss=0.334295 lr=3.058358e-05 lr_backbone=3.058358e-05 elapsed=37:32:18 avg_step=4.19s recent_step=4.13s eta=20:36:23 +2026-09-21 19:59:36 | INFO | train rank=0 | step=32290/50000 progress=64.6% loss=0.398987 lr=3.055311e-05 lr_backbone=3.055311e-05 elapsed=37:33:00 avg_step=4.19s recent_step=4.17s eta=20:35:41 +2026-09-21 20:00:18 | INFO | train rank=0 | step=32300/50000 progress=64.6% loss=0.319047 lr=3.052265e-05 lr_backbone=3.052265e-05 elapsed=37:33:41 avg_step=4.19s recent_step=4.16s eta=20:34:59 +2026-09-21 20:00:59 | INFO | train rank=0 | step=32310/50000 progress=64.6% loss=0.350254 lr=3.049220e-05 lr_backbone=3.049220e-05 elapsed=37:34:22 avg_step=4.19s recent_step=4.12s eta=20:34:17 +2026-09-21 20:01:41 | INFO | train rank=0 | step=32320/50000 progress=64.6% loss=0.302385 lr=3.046175e-05 lr_backbone=3.046175e-05 elapsed=37:35:04 avg_step=4.19s recent_step=4.15s eta=20:33:35 +2026-09-21 20:02:22 | INFO | train rank=0 | step=32330/50000 progress=64.7% loss=0.330444 lr=3.043132e-05 lr_backbone=3.043132e-05 elapsed=37:35:45 avg_step=4.19s recent_step=4.11s eta=20:32:53 +2026-09-21 20:03:03 | INFO | train rank=0 | step=32340/50000 progress=64.7% loss=0.292740 lr=3.040089e-05 lr_backbone=3.040089e-05 elapsed=37:36:26 avg_step=4.19s recent_step=4.13s eta=20:32:11 +2026-09-21 20:03:44 | INFO | train rank=0 | step=32350/50000 progress=64.7% loss=0.395250 lr=3.037047e-05 lr_backbone=3.037047e-05 elapsed=37:37:08 avg_step=4.19s recent_step=4.15s eta=20:31:28 +2026-09-21 20:04:26 | INFO | train rank=0 | step=32360/50000 progress=64.7% loss=0.389637 lr=3.034006e-05 lr_backbone=3.034006e-05 elapsed=37:37:49 avg_step=4.19s recent_step=4.15s eta=20:30:46 +2026-09-21 20:05:07 | INFO | train rank=0 | step=32370/50000 progress=64.7% loss=0.360157 lr=3.030966e-05 lr_backbone=3.030966e-05 elapsed=37:38:31 avg_step=4.19s recent_step=4.12s eta=20:30:04 +2026-09-21 20:05:49 | INFO | train rank=0 | step=32380/50000 progress=64.8% loss=0.290029 lr=3.027927e-05 lr_backbone=3.027927e-05 elapsed=37:39:12 avg_step=4.19s recent_step=4.16s eta=20:29:22 +2026-09-21 20:06:30 | INFO | train rank=0 | step=32390/50000 progress=64.8% loss=0.289928 lr=3.024888e-05 lr_backbone=3.024888e-05 elapsed=37:39:53 avg_step=4.19s recent_step=4.09s eta=20:28:40 +2026-09-21 20:07:11 | INFO | train rank=0 | step=32400/50000 progress=64.8% loss=0.288408 lr=3.021851e-05 lr_backbone=3.021851e-05 elapsed=37:40:34 avg_step=4.19s recent_step=4.13s eta=20:27:58 +2026-09-21 20:07:53 | INFO | train rank=0 | step=32410/50000 progress=64.8% loss=0.334252 lr=3.018814e-05 lr_backbone=3.018814e-05 elapsed=37:41:16 avg_step=4.19s recent_step=4.16s eta=20:27:16 +2026-09-21 20:08:34 | INFO | train rank=0 | step=32420/50000 progress=64.8% loss=0.427270 lr=3.015778e-05 lr_backbone=3.015778e-05 elapsed=37:41:57 avg_step=4.19s recent_step=4.14s eta=20:26:34 +2026-09-21 20:09:15 | INFO | train rank=0 | step=32430/50000 progress=64.9% loss=0.318370 lr=3.012743e-05 lr_backbone=3.012743e-05 elapsed=37:42:39 avg_step=4.19s recent_step=4.13s eta=20:25:51 +2026-09-21 20:09:57 | INFO | train rank=0 | step=32440/50000 progress=64.9% loss=0.300993 lr=3.009709e-05 lr_backbone=3.009709e-05 elapsed=37:43:20 avg_step=4.19s recent_step=4.14s eta=20:25:09 +2026-09-21 20:10:38 | INFO | train rank=0 | step=32450/50000 progress=64.9% loss=0.309933 lr=3.006676e-05 lr_backbone=3.006676e-05 elapsed=37:44:01 avg_step=4.19s recent_step=4.10s eta=20:24:27 +2026-09-21 20:11:19 | INFO | train rank=0 | step=32460/50000 progress=64.9% loss=0.334710 lr=3.003644e-05 lr_backbone=3.003644e-05 elapsed=37:44:43 avg_step=4.19s recent_step=4.15s eta=20:23:45 +2026-09-21 20:12:01 | INFO | train rank=0 | step=32470/50000 progress=64.9% loss=0.310806 lr=3.000612e-05 lr_backbone=3.000612e-05 elapsed=37:45:24 avg_step=4.19s recent_step=4.14s eta=20:23:03 +2026-09-21 20:12:42 | INFO | train rank=0 | step=32480/50000 progress=65.0% loss=0.301530 lr=2.997581e-05 lr_backbone=2.997581e-05 elapsed=37:46:06 avg_step=4.19s recent_step=4.15s eta=20:22:21 +2026-09-21 20:13:24 | INFO | train rank=0 | step=32490/50000 progress=65.0% loss=0.373999 lr=2.994552e-05 lr_backbone=2.994552e-05 elapsed=37:46:47 avg_step=4.19s recent_step=4.14s eta=20:21:39 +2026-09-21 20:14:05 | INFO | train rank=0 | step=32500/50000 progress=65.0% loss=0.419490 lr=2.991523e-05 lr_backbone=2.991523e-05 elapsed=37:47:28 avg_step=4.19s recent_step=4.15s eta=20:20:57 +2026-09-21 20:14:46 | INFO | train rank=0 | step=32510/50000 progress=65.0% loss=0.349902 lr=2.988495e-05 lr_backbone=2.988495e-05 elapsed=37:48:09 avg_step=4.19s recent_step=4.10s eta=20:20:14 +2026-09-21 20:15:27 | INFO | train rank=0 | step=32520/50000 progress=65.0% loss=0.310812 lr=2.985468e-05 lr_backbone=2.985468e-05 elapsed=37:48:51 avg_step=4.19s recent_step=4.12s eta=20:19:32 +2026-09-21 20:16:09 | INFO | train rank=0 | step=32530/50000 progress=65.1% loss=0.359113 lr=2.982442e-05 lr_backbone=2.982442e-05 elapsed=37:49:32 avg_step=4.19s recent_step=4.16s eta=20:18:50 +2026-09-21 20:16:50 | INFO | train rank=0 | step=32540/50000 progress=65.1% loss=0.325153 lr=2.979416e-05 lr_backbone=2.979416e-05 elapsed=37:50:14 avg_step=4.19s recent_step=4.14s eta=20:18:08 +2026-09-21 20:17:32 | INFO | train rank=0 | step=32550/50000 progress=65.1% loss=0.383085 lr=2.976392e-05 lr_backbone=2.976392e-05 elapsed=37:50:55 avg_step=4.19s recent_step=4.14s eta=20:17:26 +2026-09-21 20:18:13 | INFO | train rank=0 | step=32560/50000 progress=65.1% loss=0.291223 lr=2.973368e-05 lr_backbone=2.973368e-05 elapsed=37:51:37 avg_step=4.19s recent_step=4.16s eta=20:16:44 +2026-09-21 20:18:54 | INFO | train rank=0 | step=32570/50000 progress=65.1% loss=0.293958 lr=2.970346e-05 lr_backbone=2.970346e-05 elapsed=37:52:18 avg_step=4.19s recent_step=4.11s eta=20:16:02 +2026-09-21 20:19:36 | INFO | train rank=0 | step=32580/50000 progress=65.2% loss=0.296612 lr=2.967324e-05 lr_backbone=2.967324e-05 elapsed=37:52:59 avg_step=4.19s recent_step=4.13s eta=20:15:19 +2026-09-21 20:20:17 | INFO | train rank=0 | step=32590/50000 progress=65.2% loss=0.320131 lr=2.964303e-05 lr_backbone=2.964303e-05 elapsed=37:53:41 avg_step=4.19s recent_step=4.17s eta=20:14:37 +2026-09-21 20:20:59 | INFO | train rank=0 | step=32600/50000 progress=65.2% loss=0.266654 lr=2.961283e-05 lr_backbone=2.961283e-05 elapsed=37:54:22 avg_step=4.19s recent_step=4.14s eta=20:13:55 +2026-09-21 20:21:40 | INFO | train rank=0 | step=32610/50000 progress=65.2% loss=0.385226 lr=2.958264e-05 lr_backbone=2.958264e-05 elapsed=37:55:03 avg_step=4.19s recent_step=4.12s eta=20:13:13 +2026-09-21 20:22:21 | INFO | train rank=0 | step=32620/50000 progress=65.2% loss=0.291856 lr=2.955246e-05 lr_backbone=2.955246e-05 elapsed=37:55:45 avg_step=4.19s recent_step=4.14s eta=20:12:31 +2026-09-21 20:23:02 | INFO | train rank=0 | step=32630/50000 progress=65.3% loss=0.385030 lr=2.952228e-05 lr_backbone=2.952228e-05 elapsed=37:56:26 avg_step=4.19s recent_step=4.11s eta=20:11:49 +2026-09-21 20:23:44 | INFO | train rank=0 | step=32640/50000 progress=65.3% loss=0.409141 lr=2.949212e-05 lr_backbone=2.949212e-05 elapsed=37:57:07 avg_step=4.19s recent_step=4.13s eta=20:11:07 +2026-09-21 20:24:25 | INFO | train rank=0 | step=32650/50000 progress=65.3% loss=0.398223 lr=2.946196e-05 lr_backbone=2.946196e-05 elapsed=37:57:49 avg_step=4.19s recent_step=4.17s eta=20:10:25 +2026-09-21 20:25:07 | INFO | train rank=0 | step=32660/50000 progress=65.3% loss=0.370144 lr=2.943182e-05 lr_backbone=2.943182e-05 elapsed=37:58:30 avg_step=4.19s recent_step=4.15s eta=20:09:43 +2026-09-21 20:25:48 | INFO | train rank=0 | step=32670/50000 progress=65.3% loss=0.317554 lr=2.940168e-05 lr_backbone=2.940168e-05 elapsed=37:59:12 avg_step=4.19s recent_step=4.13s eta=20:09:00 +2026-09-21 20:26:30 | INFO | train rank=0 | step=32680/50000 progress=65.4% loss=0.329114 lr=2.937155e-05 lr_backbone=2.937155e-05 elapsed=37:59:53 avg_step=4.19s recent_step=4.15s eta=20:08:18 +2026-09-21 20:27:11 | INFO | train rank=0 | step=32690/50000 progress=65.4% loss=0.321286 lr=2.934143e-05 lr_backbone=2.934143e-05 elapsed=38:00:34 avg_step=4.19s recent_step=4.09s eta=20:07:36 +2026-09-21 20:27:52 | INFO | train rank=0 | step=32700/50000 progress=65.4% loss=0.346293 lr=2.931132e-05 lr_backbone=2.931132e-05 elapsed=38:01:15 avg_step=4.19s recent_step=4.12s eta=20:06:54 +2026-09-21 20:28:33 | INFO | train rank=0 | step=32710/50000 progress=65.4% loss=0.314194 lr=2.928122e-05 lr_backbone=2.928122e-05 elapsed=38:01:57 avg_step=4.19s recent_step=4.13s eta=20:06:12 +2026-09-21 20:29:15 | INFO | train rank=0 | step=32720/50000 progress=65.4% loss=0.406806 lr=2.925113e-05 lr_backbone=2.925113e-05 elapsed=38:02:38 avg_step=4.19s recent_step=4.16s eta=20:05:30 +2026-09-21 20:29:56 | INFO | train rank=0 | step=32730/50000 progress=65.5% loss=0.301218 lr=2.922105e-05 lr_backbone=2.922105e-05 elapsed=38:03:19 avg_step=4.19s recent_step=4.12s eta=20:04:48 +2026-09-21 20:30:38 | INFO | train rank=0 | step=32740/50000 progress=65.5% loss=0.347432 lr=2.919097e-05 lr_backbone=2.919097e-05 elapsed=38:04:01 avg_step=4.19s recent_step=4.16s eta=20:04:06 +2026-09-21 20:31:19 | INFO | train rank=0 | step=32750/50000 progress=65.5% loss=0.357332 lr=2.916091e-05 lr_backbone=2.916091e-05 elapsed=38:04:42 avg_step=4.19s recent_step=4.10s eta=20:03:23 +2026-09-21 20:32:00 | INFO | train rank=0 | step=32760/50000 progress=65.5% loss=0.305171 lr=2.913085e-05 lr_backbone=2.913085e-05 elapsed=38:05:23 avg_step=4.19s recent_step=4.12s eta=20:02:41 +2026-09-21 20:32:41 | INFO | train rank=0 | step=32770/50000 progress=65.5% loss=0.321449 lr=2.910080e-05 lr_backbone=2.910080e-05 elapsed=38:06:04 avg_step=4.19s recent_step=4.12s eta=20:01:59 +2026-09-21 20:33:23 | INFO | train rank=0 | step=32780/50000 progress=65.6% loss=0.361476 lr=2.907077e-05 lr_backbone=2.907077e-05 elapsed=38:06:46 avg_step=4.19s recent_step=4.16s eta=20:01:17 +2026-09-21 20:34:04 | INFO | train rank=0 | step=32790/50000 progress=65.6% loss=0.417071 lr=2.904074e-05 lr_backbone=2.904074e-05 elapsed=38:07:27 avg_step=4.19s recent_step=4.13s eta=20:00:35 +2026-09-21 20:34:46 | INFO | train rank=0 | step=32800/50000 progress=65.6% loss=0.342358 lr=2.901072e-05 lr_backbone=2.901072e-05 elapsed=38:08:09 avg_step=4.19s recent_step=4.18s eta=19:59:53 +2026-09-21 20:35:27 | INFO | train rank=0 | step=32810/50000 progress=65.6% loss=0.320506 lr=2.898071e-05 lr_backbone=2.898071e-05 elapsed=38:08:50 avg_step=4.19s recent_step=4.09s eta=19:59:10 +2026-09-21 20:36:08 | INFO | train rank=0 | step=32820/50000 progress=65.6% loss=0.354130 lr=2.895071e-05 lr_backbone=2.895071e-05 elapsed=38:09:31 avg_step=4.19s recent_step=4.12s eta=19:58:28 +2026-09-21 20:36:49 | INFO | train rank=0 | step=32830/50000 progress=65.7% loss=0.310214 lr=2.892072e-05 lr_backbone=2.892072e-05 elapsed=38:10:13 avg_step=4.19s recent_step=4.15s eta=19:57:46 +2026-09-21 20:37:31 | INFO | train rank=0 | step=32840/50000 progress=65.7% loss=0.392098 lr=2.889074e-05 lr_backbone=2.889074e-05 elapsed=38:10:54 avg_step=4.19s recent_step=4.18s eta=19:57:04 +2026-09-21 20:38:13 | INFO | train rank=0 | step=32850/50000 progress=65.7% loss=0.370206 lr=2.886076e-05 lr_backbone=2.886076e-05 elapsed=38:11:36 avg_step=4.19s recent_step=4.15s eta=19:56:22 +2026-09-21 20:38:54 | INFO | train rank=0 | step=32860/50000 progress=65.7% loss=0.296911 lr=2.883080e-05 lr_backbone=2.883080e-05 elapsed=38:12:17 avg_step=4.19s recent_step=4.14s eta=19:55:40 +2026-09-21 20:39:35 | INFO | train rank=0 | step=32870/50000 progress=65.7% loss=0.343064 lr=2.880084e-05 lr_backbone=2.880084e-05 elapsed=38:12:58 avg_step=4.19s recent_step=4.10s eta=19:54:58 +2026-09-21 20:40:16 | INFO | train rank=0 | step=32880/50000 progress=65.8% loss=0.325634 lr=2.877090e-05 lr_backbone=2.877090e-05 elapsed=38:13:40 avg_step=4.19s recent_step=4.12s eta=19:54:16 +2026-09-21 20:40:58 | INFO | train rank=0 | step=32890/50000 progress=65.8% loss=0.293874 lr=2.874096e-05 lr_backbone=2.874096e-05 elapsed=38:14:21 avg_step=4.19s recent_step=4.14s eta=19:53:34 +2026-09-21 20:41:39 | INFO | train rank=0 | step=32900/50000 progress=65.8% loss=0.285273 lr=2.871104e-05 lr_backbone=2.871104e-05 elapsed=38:15:03 avg_step=4.19s recent_step=4.17s eta=19:52:52 +2026-09-21 20:42:21 | INFO | train rank=0 | step=32910/50000 progress=65.8% loss=0.337394 lr=2.868112e-05 lr_backbone=2.868112e-05 elapsed=38:15:44 avg_step=4.19s recent_step=4.16s eta=19:52:10 +2026-09-21 20:43:02 | INFO | train rank=0 | step=32920/50000 progress=65.8% loss=0.396908 lr=2.865121e-05 lr_backbone=2.865121e-05 elapsed=38:16:26 avg_step=4.19s recent_step=4.14s eta=19:51:28 +2026-09-21 20:43:43 | INFO | train rank=0 | step=32930/50000 progress=65.9% loss=0.426373 lr=2.862131e-05 lr_backbone=2.862131e-05 elapsed=38:17:07 avg_step=4.19s recent_step=4.11s eta=19:50:45 +2026-09-21 20:44:25 | INFO | train rank=0 | step=32940/50000 progress=65.9% loss=0.362344 lr=2.859142e-05 lr_backbone=2.859142e-05 elapsed=38:17:48 avg_step=4.19s recent_step=4.15s eta=19:50:03 +2026-09-21 20:45:06 | INFO | train rank=0 | step=32950/50000 progress=65.9% loss=0.335047 lr=2.856154e-05 lr_backbone=2.856154e-05 elapsed=38:18:30 avg_step=4.19s recent_step=4.13s eta=19:49:21 +2026-09-21 20:45:48 | INFO | train rank=0 | step=32960/50000 progress=65.9% loss=0.296563 lr=2.853167e-05 lr_backbone=2.853167e-05 elapsed=38:19:11 avg_step=4.19s recent_step=4.14s eta=19:48:39 +2026-09-21 20:46:29 | INFO | train rank=0 | step=32970/50000 progress=65.9% loss=0.307787 lr=2.850181e-05 lr_backbone=2.850181e-05 elapsed=38:19:53 avg_step=4.19s recent_step=4.16s eta=19:47:57 +2026-09-21 20:47:11 | INFO | train rank=0 | step=32980/50000 progress=66.0% loss=0.386455 lr=2.847196e-05 lr_backbone=2.847196e-05 elapsed=38:20:34 avg_step=4.19s recent_step=4.13s eta=19:47:15 +2026-09-21 20:47:52 | INFO | train rank=0 | step=32990/50000 progress=66.0% loss=0.396523 lr=2.844212e-05 lr_backbone=2.844212e-05 elapsed=38:21:15 avg_step=4.19s recent_step=4.11s eta=19:46:33 +2026-09-21 20:48:33 | INFO | train rank=0 | step=33000/50000 progress=66.0% loss=0.308794 lr=2.841228e-05 lr_backbone=2.841228e-05 elapsed=38:21:56 avg_step=4.19s recent_step=4.12s eta=19:45:51 +2026-09-21 20:49:14 | INFO | train rank=0 | step=33010/50000 progress=66.0% loss=0.447647 lr=2.838246e-05 lr_backbone=2.838246e-05 elapsed=38:22:38 avg_step=4.19s recent_step=4.14s eta=19:45:08 +2026-09-21 20:49:56 | INFO | train rank=0 | step=33020/50000 progress=66.0% loss=0.351364 lr=2.835264e-05 lr_backbone=2.835264e-05 elapsed=38:23:19 avg_step=4.19s recent_step=4.15s eta=19:44:26 +2026-09-21 20:50:37 | INFO | train rank=0 | step=33030/50000 progress=66.1% loss=0.310410 lr=2.832284e-05 lr_backbone=2.832284e-05 elapsed=38:24:01 avg_step=4.19s recent_step=4.15s eta=19:43:44 +2026-09-21 20:51:19 | INFO | train rank=0 | step=33040/50000 progress=66.1% loss=0.288627 lr=2.829304e-05 lr_backbone=2.829304e-05 elapsed=38:24:42 avg_step=4.19s recent_step=4.15s eta=19:43:02 +2026-09-21 20:52:00 | INFO | train rank=0 | step=33050/50000 progress=66.1% loss=0.357514 lr=2.826326e-05 lr_backbone=2.826326e-05 elapsed=38:25:23 avg_step=4.19s recent_step=4.09s eta=19:42:20 +2026-09-21 20:52:41 | INFO | train rank=0 | step=33060/50000 progress=66.1% loss=0.351269 lr=2.823348e-05 lr_backbone=2.823348e-05 elapsed=38:26:04 avg_step=4.19s recent_step=4.11s eta=19:41:38 +2026-09-21 20:53:22 | INFO | train rank=0 | step=33070/50000 progress=66.1% loss=0.312121 lr=2.820372e-05 lr_backbone=2.820372e-05 elapsed=38:26:45 avg_step=4.19s recent_step=4.13s eta=19:40:56 +2026-09-21 20:54:03 | INFO | train rank=0 | step=33080/50000 progress=66.2% loss=0.391728 lr=2.817396e-05 lr_backbone=2.817396e-05 elapsed=38:27:27 avg_step=4.19s recent_step=4.14s eta=19:40:14 +2026-09-21 20:54:45 | INFO | train rank=0 | step=33090/50000 progress=66.2% loss=0.351978 lr=2.814421e-05 lr_backbone=2.814421e-05 elapsed=38:28:08 avg_step=4.19s recent_step=4.13s eta=19:39:31 +2026-09-21 20:55:26 | INFO | train rank=0 | step=33100/50000 progress=66.2% loss=0.341442 lr=2.811447e-05 lr_backbone=2.811447e-05 elapsed=38:28:49 avg_step=4.19s recent_step=4.10s eta=19:38:49 +2026-09-21 20:56:07 | INFO | train rank=0 | step=33110/50000 progress=66.2% loss=0.275612 lr=2.808475e-05 lr_backbone=2.808475e-05 elapsed=38:29:30 avg_step=4.19s recent_step=4.08s eta=19:38:07 +2026-09-21 20:56:48 | INFO | train rank=0 | step=33120/50000 progress=66.2% loss=0.320596 lr=2.805503e-05 lr_backbone=2.805503e-05 elapsed=38:30:11 avg_step=4.19s recent_step=4.10s eta=19:37:24 +2026-09-21 20:57:29 | INFO | train rank=0 | step=33130/50000 progress=66.3% loss=0.358068 lr=2.802532e-05 lr_backbone=2.802532e-05 elapsed=38:30:52 avg_step=4.19s recent_step=4.15s eta=19:36:42 +2026-09-21 20:58:10 | INFO | train rank=0 | step=33140/50000 progress=66.3% loss=0.285118 lr=2.799562e-05 lr_backbone=2.799562e-05 elapsed=38:31:34 avg_step=4.19s recent_step=4.13s eta=19:36:00 +2026-09-21 20:58:52 | INFO | train rank=0 | step=33150/50000 progress=66.3% loss=0.284987 lr=2.796593e-05 lr_backbone=2.796593e-05 elapsed=38:32:15 avg_step=4.19s recent_step=4.14s eta=19:35:18 +2026-09-21 20:59:33 | INFO | train rank=0 | step=33160/50000 progress=66.3% loss=0.456769 lr=2.793625e-05 lr_backbone=2.793625e-05 elapsed=38:32:57 avg_step=4.19s recent_step=4.15s eta=19:34:36 +2026-09-21 21:00:14 | INFO | train rank=0 | step=33170/50000 progress=66.3% loss=0.300843 lr=2.790658e-05 lr_backbone=2.790658e-05 elapsed=38:33:37 avg_step=4.19s recent_step=4.08s eta=19:33:54 +2026-09-21 21:00:55 | INFO | train rank=0 | step=33180/50000 progress=66.4% loss=0.367946 lr=2.787692e-05 lr_backbone=2.787692e-05 elapsed=38:34:18 avg_step=4.19s recent_step=4.10s eta=19:33:12 +2026-09-21 21:01:37 | INFO | train rank=0 | step=33190/50000 progress=66.4% loss=0.302106 lr=2.784726e-05 lr_backbone=2.784726e-05 elapsed=38:35:00 avg_step=4.19s recent_step=4.15s eta=19:32:30 +2026-09-21 21:02:18 | INFO | train rank=0 | step=33200/50000 progress=66.4% loss=0.392999 lr=2.781762e-05 lr_backbone=2.781762e-05 elapsed=38:35:41 avg_step=4.18s recent_step=4.13s eta=19:31:47 +2026-09-21 21:02:59 | INFO | train rank=0 | step=33210/50000 progress=66.4% loss=0.312640 lr=2.778799e-05 lr_backbone=2.778799e-05 elapsed=38:36:23 avg_step=4.18s recent_step=4.14s eta=19:31:05 +2026-09-21 21:03:41 | INFO | train rank=0 | step=33220/50000 progress=66.4% loss=0.414576 lr=2.775837e-05 lr_backbone=2.775837e-05 elapsed=38:37:04 avg_step=4.18s recent_step=4.13s eta=19:30:23 +2026-09-21 21:04:21 | INFO | train rank=0 | step=33230/50000 progress=66.5% loss=0.348860 lr=2.772876e-05 lr_backbone=2.772876e-05 elapsed=38:37:45 avg_step=4.18s recent_step=4.08s eta=19:29:41 +2026-09-21 21:05:02 | INFO | train rank=0 | step=33240/50000 progress=66.5% loss=0.307202 lr=2.769915e-05 lr_backbone=2.769915e-05 elapsed=38:38:26 avg_step=4.18s recent_step=4.11s eta=19:28:59 +2026-09-21 21:05:44 | INFO | train rank=0 | step=33250/50000 progress=66.5% loss=0.289809 lr=2.766956e-05 lr_backbone=2.766956e-05 elapsed=38:39:07 avg_step=4.18s recent_step=4.12s eta=19:28:16 +2026-09-21 21:06:25 | INFO | train rank=0 | step=33260/50000 progress=66.5% loss=0.349362 lr=2.763998e-05 lr_backbone=2.763998e-05 elapsed=38:39:48 avg_step=4.18s recent_step=4.13s eta=19:27:34 +2026-09-21 21:07:06 | INFO | train rank=0 | step=33270/50000 progress=66.5% loss=0.332058 lr=2.761040e-05 lr_backbone=2.761040e-05 elapsed=38:40:30 avg_step=4.18s recent_step=4.14s eta=19:26:52 +2026-09-21 21:07:48 | INFO | train rank=0 | step=33280/50000 progress=66.6% loss=0.311725 lr=2.758084e-05 lr_backbone=2.758084e-05 elapsed=38:41:11 avg_step=4.18s recent_step=4.13s eta=19:26:10 +2026-09-21 21:08:29 | INFO | train rank=0 | step=33290/50000 progress=66.6% loss=0.348869 lr=2.755129e-05 lr_backbone=2.755129e-05 elapsed=38:41:52 avg_step=4.18s recent_step=4.10s eta=19:25:28 +2026-09-21 21:09:10 | INFO | train rank=0 | step=33300/50000 progress=66.6% loss=0.348308 lr=2.752174e-05 lr_backbone=2.752174e-05 elapsed=38:42:33 avg_step=4.18s recent_step=4.12s eta=19:24:46 +2026-09-21 21:09:51 | INFO | train rank=0 | step=33310/50000 progress=66.6% loss=0.315564 lr=2.749221e-05 lr_backbone=2.749221e-05 elapsed=38:43:15 avg_step=4.18s recent_step=4.14s eta=19:24:04 +2026-09-21 21:10:33 | INFO | train rank=0 | step=33320/50000 progress=66.6% loss=0.342646 lr=2.746268e-05 lr_backbone=2.746268e-05 elapsed=38:43:56 avg_step=4.18s recent_step=4.15s eta=19:23:22 +2026-09-21 21:11:14 | INFO | train rank=0 | step=33330/50000 progress=66.7% loss=0.289370 lr=2.743317e-05 lr_backbone=2.743317e-05 elapsed=38:44:38 avg_step=4.18s recent_step=4.15s eta=19:22:40 +2026-09-21 21:11:56 | INFO | train rank=0 | step=33340/50000 progress=66.7% loss=0.325263 lr=2.740366e-05 lr_backbone=2.740366e-05 elapsed=38:45:19 avg_step=4.18s recent_step=4.14s eta=19:21:57 +2026-09-21 21:12:37 | INFO | train rank=0 | step=33350/50000 progress=66.7% loss=0.334810 lr=2.737417e-05 lr_backbone=2.737417e-05 elapsed=38:46:00 avg_step=4.18s recent_step=4.08s eta=19:21:15 +2026-09-21 21:13:18 | INFO | train rank=0 | step=33360/50000 progress=66.7% loss=0.317445 lr=2.734468e-05 lr_backbone=2.734468e-05 elapsed=38:46:41 avg_step=4.18s recent_step=4.11s eta=19:20:33 +2026-09-21 21:13:59 | INFO | train rank=0 | step=33370/50000 progress=66.7% loss=0.282844 lr=2.731521e-05 lr_backbone=2.731521e-05 elapsed=38:47:23 avg_step=4.18s recent_step=4.15s eta=19:19:51 +2026-09-21 21:14:40 | INFO | train rank=0 | step=33380/50000 progress=66.8% loss=0.354331 lr=2.728574e-05 lr_backbone=2.728574e-05 elapsed=38:48:04 avg_step=4.18s recent_step=4.12s eta=19:19:09 +2026-09-21 21:15:22 | INFO | train rank=0 | step=33390/50000 progress=66.8% loss=0.388556 lr=2.725629e-05 lr_backbone=2.725629e-05 elapsed=38:48:45 avg_step=4.18s recent_step=4.13s eta=19:18:27 +2026-09-21 21:16:03 | INFO | train rank=0 | step=33400/50000 progress=66.8% loss=0.355925 lr=2.722684e-05 lr_backbone=2.722684e-05 elapsed=38:49:26 avg_step=4.18s recent_step=4.12s eta=19:17:44 +2026-09-21 21:16:44 | INFO | train rank=0 | step=33410/50000 progress=66.8% loss=0.366189 lr=2.719741e-05 lr_backbone=2.719741e-05 elapsed=38:50:07 avg_step=4.18s recent_step=4.08s eta=19:17:02 +2026-09-21 21:17:25 | INFO | train rank=0 | step=33420/50000 progress=66.8% loss=0.321155 lr=2.716798e-05 lr_backbone=2.716798e-05 elapsed=38:50:48 avg_step=4.18s recent_step=4.10s eta=19:16:20 +2026-09-21 21:18:06 | INFO | train rank=0 | step=33430/50000 progress=66.9% loss=0.332775 lr=2.713857e-05 lr_backbone=2.713857e-05 elapsed=38:51:29 avg_step=4.18s recent_step=4.14s eta=19:15:38 +2026-09-21 21:18:47 | INFO | train rank=0 | step=33440/50000 progress=66.9% loss=0.328581 lr=2.710916e-05 lr_backbone=2.710916e-05 elapsed=38:52:11 avg_step=4.18s recent_step=4.13s eta=19:14:56 +2026-09-21 21:19:29 | INFO | train rank=0 | step=33450/50000 progress=66.9% loss=0.388479 lr=2.707977e-05 lr_backbone=2.707977e-05 elapsed=38:52:52 avg_step=4.18s recent_step=4.13s eta=19:14:13 +2026-09-21 21:20:10 | INFO | train rank=0 | step=33460/50000 progress=66.9% loss=0.307270 lr=2.705038e-05 lr_backbone=2.705038e-05 elapsed=38:53:33 avg_step=4.18s recent_step=4.13s eta=19:13:31 +2026-09-21 21:20:51 | INFO | train rank=0 | step=33470/50000 progress=66.9% loss=0.329939 lr=2.702101e-05 lr_backbone=2.702101e-05 elapsed=38:54:14 avg_step=4.18s recent_step=4.08s eta=19:12:49 +2026-09-21 21:21:32 | INFO | train rank=0 | step=33480/50000 progress=67.0% loss=0.299122 lr=2.699164e-05 lr_backbone=2.699164e-05 elapsed=38:54:55 avg_step=4.18s recent_step=4.12s eta=19:12:07 +2026-09-21 21:22:13 | INFO | train rank=0 | step=33490/50000 progress=67.0% loss=0.470845 lr=2.696229e-05 lr_backbone=2.696229e-05 elapsed=38:55:37 avg_step=4.18s recent_step=4.15s eta=19:11:25 +2026-09-21 21:22:55 | INFO | train rank=0 | step=33500/50000 progress=67.0% loss=0.397712 lr=2.693294e-05 lr_backbone=2.693294e-05 elapsed=38:56:18 avg_step=4.18s recent_step=4.13s eta=19:10:43 +2026-09-21 21:23:36 | INFO | train rank=0 | step=33510/50000 progress=67.0% loss=0.366925 lr=2.690361e-05 lr_backbone=2.690361e-05 elapsed=38:56:59 avg_step=4.18s recent_step=4.14s eta=19:10:01 +2026-09-21 21:24:17 | INFO | train rank=0 | step=33520/50000 progress=67.0% loss=0.371712 lr=2.687428e-05 lr_backbone=2.687428e-05 elapsed=38:57:41 avg_step=4.18s recent_step=4.12s eta=19:09:18 +2026-09-21 21:24:58 | INFO | train rank=0 | step=33530/50000 progress=67.1% loss=0.485049 lr=2.684497e-05 lr_backbone=2.684497e-05 elapsed=38:58:21 avg_step=4.18s recent_step=4.08s eta=19:08:36 +2026-09-21 21:25:39 | INFO | train rank=0 | step=33540/50000 progress=67.1% loss=0.345302 lr=2.681566e-05 lr_backbone=2.681566e-05 elapsed=38:59:03 avg_step=4.18s recent_step=4.12s eta=19:07:54 +2026-09-21 21:26:21 | INFO | train rank=0 | step=33550/50000 progress=67.1% loss=0.364617 lr=2.678637e-05 lr_backbone=2.678637e-05 elapsed=38:59:44 avg_step=4.18s recent_step=4.13s eta=19:07:12 +2026-09-21 21:27:02 | INFO | train rank=0 | step=33560/50000 progress=67.1% loss=0.376377 lr=2.675708e-05 lr_backbone=2.675708e-05 elapsed=39:00:25 avg_step=4.18s recent_step=4.13s eta=19:06:30 +2026-09-21 21:27:43 | INFO | train rank=0 | step=33570/50000 progress=67.1% loss=0.401617 lr=2.672781e-05 lr_backbone=2.672781e-05 elapsed=39:01:07 avg_step=4.18s recent_step=4.13s eta=19:05:48 +2026-09-21 21:28:25 | INFO | train rank=0 | step=33580/50000 progress=67.2% loss=0.362237 lr=2.669855e-05 lr_backbone=2.669855e-05 elapsed=39:01:48 avg_step=4.18s recent_step=4.13s eta=19:05:06 +2026-09-21 21:29:05 | INFO | train rank=0 | step=33590/50000 progress=67.2% loss=0.362885 lr=2.666929e-05 lr_backbone=2.666929e-05 elapsed=39:02:29 avg_step=4.18s recent_step=4.08s eta=19:04:23 +2026-09-21 21:29:46 | INFO | train rank=0 | step=33600/50000 progress=67.2% loss=0.352298 lr=2.664005e-05 lr_backbone=2.664005e-05 elapsed=39:03:10 avg_step=4.18s recent_step=4.10s eta=19:03:41 +2026-09-21 21:30:28 | INFO | train rank=0 | step=33610/50000 progress=67.2% loss=0.341542 lr=2.661082e-05 lr_backbone=2.661082e-05 elapsed=39:03:51 avg_step=4.18s recent_step=4.14s eta=19:02:59 +2026-09-21 21:31:09 | INFO | train rank=0 | step=33620/50000 progress=67.2% loss=0.287909 lr=2.658159e-05 lr_backbone=2.658159e-05 elapsed=39:04:33 avg_step=4.18s recent_step=4.15s eta=19:02:17 +2026-09-21 21:31:51 | INFO | train rank=0 | step=33630/50000 progress=67.3% loss=0.308607 lr=2.655238e-05 lr_backbone=2.655238e-05 elapsed=39:05:14 avg_step=4.18s recent_step=4.16s eta=19:01:35 +2026-09-21 21:32:32 | INFO | train rank=0 | step=33640/50000 progress=67.3% loss=0.369259 lr=2.652318e-05 lr_backbone=2.652318e-05 elapsed=39:05:55 avg_step=4.18s recent_step=4.13s eta=19:00:53 +2026-09-21 21:33:13 | INFO | train rank=0 | step=33650/50000 progress=67.3% loss=0.362586 lr=2.649399e-05 lr_backbone=2.649399e-05 elapsed=39:06:36 avg_step=4.18s recent_step=4.10s eta=19:00:11 +2026-09-21 21:33:54 | INFO | train rank=0 | step=33660/50000 progress=67.3% loss=0.320011 lr=2.646480e-05 lr_backbone=2.646480e-05 elapsed=39:07:17 avg_step=4.18s recent_step=4.10s eta=18:59:28 +2026-09-21 21:34:35 | INFO | train rank=0 | step=33670/50000 progress=67.3% loss=0.369479 lr=2.643563e-05 lr_backbone=2.643563e-05 elapsed=39:07:59 avg_step=4.18s recent_step=4.13s eta=18:58:46 +2026-09-21 21:35:17 | INFO | train rank=0 | step=33680/50000 progress=67.4% loss=0.301601 lr=2.640647e-05 lr_backbone=2.640647e-05 elapsed=39:08:40 avg_step=4.18s recent_step=4.14s eta=18:58:04 +2026-09-21 21:35:58 | INFO | train rank=0 | step=33690/50000 progress=67.4% loss=0.315065 lr=2.637732e-05 lr_backbone=2.637732e-05 elapsed=39:09:22 avg_step=4.18s recent_step=4.14s eta=18:57:22 +2026-09-21 21:36:40 | INFO | train rank=0 | step=33700/50000 progress=67.4% loss=0.446436 lr=2.634818e-05 lr_backbone=2.634818e-05 elapsed=39:10:03 avg_step=4.18s recent_step=4.13s eta=18:56:40 +2026-09-21 21:37:20 | INFO | train rank=0 | step=33710/50000 progress=67.4% loss=0.337212 lr=2.631905e-05 lr_backbone=2.631905e-05 elapsed=39:10:44 avg_step=4.18s recent_step=4.09s eta=18:55:58 +2026-09-21 21:38:01 | INFO | train rank=0 | step=33720/50000 progress=67.4% loss=0.348337 lr=2.628993e-05 lr_backbone=2.628993e-05 elapsed=39:11:25 avg_step=4.18s recent_step=4.10s eta=18:55:15 +2026-09-21 21:38:43 | INFO | train rank=0 | step=33730/50000 progress=67.5% loss=0.311461 lr=2.626082e-05 lr_backbone=2.626082e-05 elapsed=39:12:06 avg_step=4.18s recent_step=4.15s eta=18:54:33 +2026-09-21 21:39:25 | INFO | train rank=0 | step=33740/50000 progress=67.5% loss=0.330666 lr=2.623172e-05 lr_backbone=2.623172e-05 elapsed=39:12:48 avg_step=4.18s recent_step=4.15s eta=18:53:51 +2026-09-21 21:40:06 | INFO | train rank=0 | step=33750/50000 progress=67.5% loss=0.337845 lr=2.620263e-05 lr_backbone=2.620263e-05 elapsed=39:13:29 avg_step=4.18s recent_step=4.14s eta=18:53:09 +2026-09-21 21:40:47 | INFO | train rank=0 | step=33760/50000 progress=67.5% loss=0.376842 lr=2.617355e-05 lr_backbone=2.617355e-05 elapsed=39:14:11 avg_step=4.18s recent_step=4.15s eta=18:52:27 +2026-09-21 21:41:28 | INFO | train rank=0 | step=33770/50000 progress=67.5% loss=0.341171 lr=2.614448e-05 lr_backbone=2.614448e-05 elapsed=39:14:52 avg_step=4.18s recent_step=4.11s eta=18:51:45 +2026-09-21 21:42:10 | INFO | train rank=0 | step=33780/50000 progress=67.6% loss=0.299442 lr=2.611543e-05 lr_backbone=2.611543e-05 elapsed=39:15:33 avg_step=4.18s recent_step=4.10s eta=18:51:03 +2026-09-21 21:42:51 | INFO | train rank=0 | step=33790/50000 progress=67.6% loss=0.288006 lr=2.608638e-05 lr_backbone=2.608638e-05 elapsed=39:16:14 avg_step=4.18s recent_step=4.16s eta=18:50:21 +2026-09-21 21:43:33 | INFO | train rank=0 | step=33800/50000 progress=67.6% loss=0.323483 lr=2.605734e-05 lr_backbone=2.605734e-05 elapsed=39:16:56 avg_step=4.18s recent_step=4.15s eta=18:49:39 +2026-09-21 21:44:14 | INFO | train rank=0 | step=33810/50000 progress=67.6% loss=0.365389 lr=2.602832e-05 lr_backbone=2.602832e-05 elapsed=39:17:37 avg_step=4.18s recent_step=4.13s eta=18:48:57 +2026-09-21 21:44:55 | INFO | train rank=0 | step=33820/50000 progress=67.6% loss=0.344004 lr=2.599930e-05 lr_backbone=2.599930e-05 elapsed=39:18:19 avg_step=4.18s recent_step=4.15s eta=18:48:15 +2026-09-21 21:45:36 | INFO | train rank=0 | step=33830/50000 progress=67.7% loss=0.360444 lr=2.597030e-05 lr_backbone=2.597030e-05 elapsed=39:19:00 avg_step=4.18s recent_step=4.10s eta=18:47:33 +2026-09-21 21:46:18 | INFO | train rank=0 | step=33840/50000 progress=67.7% loss=0.452355 lr=2.594130e-05 lr_backbone=2.594130e-05 elapsed=39:19:41 avg_step=4.18s recent_step=4.12s eta=18:46:50 +2026-09-21 21:46:59 | INFO | train rank=0 | step=33850/50000 progress=67.7% loss=0.340989 lr=2.591232e-05 lr_backbone=2.591232e-05 elapsed=39:20:22 avg_step=4.18s recent_step=4.14s eta=18:46:08 +2026-09-21 21:47:40 | INFO | train rank=0 | step=33860/50000 progress=67.7% loss=0.333166 lr=2.588334e-05 lr_backbone=2.588334e-05 elapsed=39:21:04 avg_step=4.18s recent_step=4.15s eta=18:45:26 +2026-09-21 21:48:22 | INFO | train rank=0 | step=33870/50000 progress=67.7% loss=0.370303 lr=2.585438e-05 lr_backbone=2.585438e-05 elapsed=39:21:45 avg_step=4.18s recent_step=4.16s eta=18:44:44 +2026-09-21 21:49:04 | INFO | train rank=0 | step=33880/50000 progress=67.8% loss=0.329574 lr=2.582543e-05 lr_backbone=2.582543e-05 elapsed=39:22:27 avg_step=4.18s recent_step=4.16s eta=18:44:03 +2026-09-21 21:49:45 | INFO | train rank=0 | step=33890/50000 progress=67.8% loss=0.324107 lr=2.579648e-05 lr_backbone=2.579648e-05 elapsed=39:23:08 avg_step=4.18s recent_step=4.10s eta=18:43:20 +2026-09-21 21:50:26 | INFO | train rank=0 | step=33900/50000 progress=67.8% loss=0.311416 lr=2.576755e-05 lr_backbone=2.576755e-05 elapsed=39:23:49 avg_step=4.18s recent_step=4.13s eta=18:42:38 +2026-09-21 21:51:07 | INFO | train rank=0 | step=33910/50000 progress=67.8% loss=0.413155 lr=2.573863e-05 lr_backbone=2.573863e-05 elapsed=39:24:31 avg_step=4.18s recent_step=4.14s eta=18:41:56 +2026-09-21 21:51:49 | INFO | train rank=0 | step=33920/50000 progress=67.8% loss=0.325917 lr=2.570972e-05 lr_backbone=2.570972e-05 elapsed=39:25:12 avg_step=4.18s recent_step=4.15s eta=18:41:14 +2026-09-21 21:52:30 | INFO | train rank=0 | step=33930/50000 progress=67.9% loss=0.337800 lr=2.568082e-05 lr_backbone=2.568082e-05 elapsed=39:25:54 avg_step=4.18s recent_step=4.16s eta=18:40:32 +2026-09-21 21:53:12 | INFO | train rank=0 | step=33940/50000 progress=67.9% loss=0.402424 lr=2.565193e-05 lr_backbone=2.565193e-05 elapsed=39:26:35 avg_step=4.18s recent_step=4.16s eta=18:39:50 +2026-09-21 21:53:53 | INFO | train rank=0 | step=33950/50000 progress=67.9% loss=0.369208 lr=2.562306e-05 lr_backbone=2.562306e-05 elapsed=39:27:16 avg_step=4.18s recent_step=4.11s eta=18:39:08 +2026-09-21 21:54:34 | INFO | train rank=0 | step=33960/50000 progress=67.9% loss=0.306184 lr=2.559419e-05 lr_backbone=2.559419e-05 elapsed=39:27:58 avg_step=4.18s recent_step=4.13s eta=18:38:26 +2026-09-21 21:55:15 | INFO | train rank=0 | step=33970/50000 progress=67.9% loss=0.317940 lr=2.556533e-05 lr_backbone=2.556533e-05 elapsed=39:28:39 avg_step=4.18s recent_step=4.10s eta=18:37:44 +2026-09-21 21:55:57 | INFO | train rank=0 | step=33980/50000 progress=68.0% loss=0.342081 lr=2.553649e-05 lr_backbone=2.553649e-05 elapsed=39:29:20 avg_step=4.18s recent_step=4.16s eta=18:37:02 +2026-09-21 21:56:38 | INFO | train rank=0 | step=33990/50000 progress=68.0% loss=0.446609 lr=2.550765e-05 lr_backbone=2.550765e-05 elapsed=39:30:02 avg_step=4.18s recent_step=4.14s eta=18:36:20 +2026-09-21 21:57:20 | INFO | train rank=0 | step=34000/50000 progress=68.0% loss=0.351398 lr=2.547882e-05 lr_backbone=2.547882e-05 elapsed=39:30:43 avg_step=4.18s recent_step=4.15s eta=18:35:38 +2026-09-21 21:58:01 | INFO | train rank=0 | step=34010/50000 progress=68.0% loss=0.324396 lr=2.545001e-05 lr_backbone=2.545001e-05 elapsed=39:31:24 avg_step=4.18s recent_step=4.10s eta=18:34:55 +2026-09-21 21:58:42 | INFO | train rank=0 | step=34020/50000 progress=68.0% loss=0.320212 lr=2.542121e-05 lr_backbone=2.542121e-05 elapsed=39:32:06 avg_step=4.18s recent_step=4.14s eta=18:34:13 +2026-09-21 21:59:24 | INFO | train rank=0 | step=34030/50000 progress=68.1% loss=0.345082 lr=2.539241e-05 lr_backbone=2.539241e-05 elapsed=39:32:47 avg_step=4.18s recent_step=4.15s eta=18:33:31 +2026-09-21 22:00:05 | INFO | train rank=0 | step=34040/50000 progress=68.1% loss=0.429707 lr=2.536363e-05 lr_backbone=2.536363e-05 elapsed=39:33:29 avg_step=4.18s recent_step=4.16s eta=18:32:50 +2026-09-21 22:00:47 | INFO | train rank=0 | step=34050/50000 progress=68.1% loss=0.390618 lr=2.533486e-05 lr_backbone=2.533486e-05 elapsed=39:34:10 avg_step=4.18s recent_step=4.16s eta=18:32:08 +2026-09-21 22:01:29 | INFO | train rank=0 | step=34060/50000 progress=68.1% loss=0.329417 lr=2.530610e-05 lr_backbone=2.530610e-05 elapsed=39:34:52 avg_step=4.18s recent_step=4.15s eta=18:31:26 +2026-09-21 22:02:09 | INFO | train rank=0 | step=34070/50000 progress=68.1% loss=0.365657 lr=2.527735e-05 lr_backbone=2.527735e-05 elapsed=39:35:33 avg_step=4.18s recent_step=4.09s eta=18:30:43 +2026-09-21 22:02:51 | INFO | train rank=0 | step=34080/50000 progress=68.2% loss=0.313370 lr=2.524861e-05 lr_backbone=2.524861e-05 elapsed=39:36:14 avg_step=4.18s recent_step=4.13s eta=18:30:01 +2026-09-21 22:03:32 | INFO | train rank=0 | step=34090/50000 progress=68.2% loss=0.329671 lr=2.521989e-05 lr_backbone=2.521989e-05 elapsed=39:36:55 avg_step=4.18s recent_step=4.13s eta=18:29:19 +2026-09-21 22:04:14 | INFO | train rank=0 | step=34100/50000 progress=68.2% loss=0.306931 lr=2.519117e-05 lr_backbone=2.519117e-05 elapsed=39:37:37 avg_step=4.18s recent_step=4.15s eta=18:28:37 +2026-09-21 22:04:55 | INFO | train rank=0 | step=34110/50000 progress=68.2% loss=0.382096 lr=2.516246e-05 lr_backbone=2.516246e-05 elapsed=39:38:19 avg_step=4.18s recent_step=4.17s eta=18:27:55 +2026-09-21 22:05:37 | INFO | train rank=0 | step=34120/50000 progress=68.2% loss=0.347722 lr=2.513377e-05 lr_backbone=2.513377e-05 elapsed=39:39:00 avg_step=4.18s recent_step=4.15s eta=18:27:13 +2026-09-21 22:06:18 | INFO | train rank=0 | step=34130/50000 progress=68.3% loss=0.302272 lr=2.510508e-05 lr_backbone=2.510508e-05 elapsed=39:39:41 avg_step=4.18s recent_step=4.12s eta=18:26:31 +2026-09-21 22:06:59 | INFO | train rank=0 | step=34140/50000 progress=68.3% loss=0.324721 lr=2.507641e-05 lr_backbone=2.507641e-05 elapsed=39:40:23 avg_step=4.18s recent_step=4.15s eta=18:25:49 +2026-09-21 22:07:41 | INFO | train rank=0 | step=34150/50000 progress=68.3% loss=0.415977 lr=2.504775e-05 lr_backbone=2.504775e-05 elapsed=39:41:04 avg_step=4.18s recent_step=4.11s eta=18:25:07 +2026-09-21 22:08:22 | INFO | train rank=0 | step=34160/50000 progress=68.3% loss=0.405930 lr=2.501910e-05 lr_backbone=2.501910e-05 elapsed=39:41:45 avg_step=4.18s recent_step=4.14s eta=18:24:25 +2026-09-21 22:09:03 | INFO | train rank=0 | step=34170/50000 progress=68.3% loss=0.445704 lr=2.499045e-05 lr_backbone=2.499045e-05 elapsed=39:42:27 avg_step=4.18s recent_step=4.15s eta=18:23:43 +2026-09-21 22:09:45 | INFO | train rank=0 | step=34180/50000 progress=68.4% loss=0.306675 lr=2.496182e-05 lr_backbone=2.496182e-05 elapsed=39:43:08 avg_step=4.18s recent_step=4.15s eta=18:23:01 +2026-09-21 22:10:26 | INFO | train rank=0 | step=34190/50000 progress=68.4% loss=0.344252 lr=2.493321e-05 lr_backbone=2.493321e-05 elapsed=39:43:49 avg_step=4.18s recent_step=4.11s eta=18:22:19 +2026-09-21 22:11:08 | INFO | train rank=0 | step=34200/50000 progress=68.4% loss=0.343434 lr=2.490460e-05 lr_backbone=2.490460e-05 elapsed=39:44:31 avg_step=4.18s recent_step=4.15s eta=18:21:37 +2026-09-21 22:11:49 | INFO | train rank=0 | step=34210/50000 progress=68.4% loss=0.368491 lr=2.487600e-05 lr_backbone=2.487600e-05 elapsed=39:45:12 avg_step=4.18s recent_step=4.13s eta=18:20:55 +2026-09-21 22:12:30 | INFO | train rank=0 | step=34220/50000 progress=68.4% loss=0.374687 lr=2.484741e-05 lr_backbone=2.484741e-05 elapsed=39:45:54 avg_step=4.18s recent_step=4.14s eta=18:20:13 +2026-09-21 22:13:12 | INFO | train rank=0 | step=34230/50000 progress=68.5% loss=0.317418 lr=2.481884e-05 lr_backbone=2.481884e-05 elapsed=39:46:35 avg_step=4.18s recent_step=4.15s eta=18:19:31 +2026-09-21 22:13:53 | INFO | train rank=0 | step=34240/50000 progress=68.5% loss=0.373212 lr=2.479028e-05 lr_backbone=2.479028e-05 elapsed=39:47:16 avg_step=4.18s recent_step=4.14s eta=18:18:49 +2026-09-21 22:14:34 | INFO | train rank=0 | step=34250/50000 progress=68.5% loss=0.459538 lr=2.476172e-05 lr_backbone=2.476172e-05 elapsed=39:47:57 avg_step=4.18s recent_step=4.10s eta=18:18:06 +2026-09-21 22:15:15 | INFO | train rank=0 | step=34260/50000 progress=68.5% loss=0.398219 lr=2.473318e-05 lr_backbone=2.473318e-05 elapsed=39:48:39 avg_step=4.18s recent_step=4.13s eta=18:17:24 +2026-09-21 22:15:57 | INFO | train rank=0 | step=34270/50000 progress=68.5% loss=0.302364 lr=2.470465e-05 lr_backbone=2.470465e-05 elapsed=39:49:20 avg_step=4.18s recent_step=4.14s eta=18:16:42 +2026-09-21 22:16:38 | INFO | train rank=0 | step=34280/50000 progress=68.6% loss=0.345134 lr=2.467613e-05 lr_backbone=2.467613e-05 elapsed=39:50:02 avg_step=4.18s recent_step=4.15s eta=18:16:00 +2026-09-21 22:17:20 | INFO | train rank=0 | step=34290/50000 progress=68.6% loss=0.326614 lr=2.464762e-05 lr_backbone=2.464762e-05 elapsed=39:50:43 avg_step=4.18s recent_step=4.16s eta=18:15:18 +2026-09-21 22:18:01 | INFO | train rank=0 | step=34300/50000 progress=68.6% loss=0.332508 lr=2.461912e-05 lr_backbone=2.461912e-05 elapsed=39:51:25 avg_step=4.18s recent_step=4.15s eta=18:14:36 +2026-09-21 22:18:42 | INFO | train rank=0 | step=34310/50000 progress=68.6% loss=0.337931 lr=2.459064e-05 lr_backbone=2.459064e-05 elapsed=39:52:06 avg_step=4.18s recent_step=4.10s eta=18:13:54 +2026-09-21 22:19:24 | INFO | train rank=0 | step=34320/50000 progress=68.6% loss=0.294192 lr=2.456216e-05 lr_backbone=2.456216e-05 elapsed=39:52:47 avg_step=4.18s recent_step=4.14s eta=18:13:12 +2026-09-21 22:20:05 | INFO | train rank=0 | step=34330/50000 progress=68.7% loss=0.417635 lr=2.453370e-05 lr_backbone=2.453370e-05 elapsed=39:53:28 avg_step=4.18s recent_step=4.13s eta=18:12:30 +2026-09-21 22:20:46 | INFO | train rank=0 | step=34340/50000 progress=68.7% loss=0.383500 lr=2.450525e-05 lr_backbone=2.450525e-05 elapsed=39:54:10 avg_step=4.18s recent_step=4.15s eta=18:11:48 +2026-09-21 22:21:28 | INFO | train rank=0 | step=34350/50000 progress=68.7% loss=0.336058 lr=2.447680e-05 lr_backbone=2.447680e-05 elapsed=39:54:51 avg_step=4.18s recent_step=4.16s eta=18:11:06 +2026-09-21 22:22:10 | INFO | train rank=0 | step=34360/50000 progress=68.7% loss=0.297613 lr=2.444837e-05 lr_backbone=2.444837e-05 elapsed=39:55:33 avg_step=4.18s recent_step=4.15s eta=18:10:24 +2026-09-21 22:22:50 | INFO | train rank=0 | step=34370/50000 progress=68.7% loss=0.373273 lr=2.441995e-05 lr_backbone=2.441995e-05 elapsed=39:56:14 avg_step=4.18s recent_step=4.09s eta=18:09:42 +2026-09-21 22:23:32 | INFO | train rank=0 | step=34380/50000 progress=68.8% loss=0.270977 lr=2.439154e-05 lr_backbone=2.439154e-05 elapsed=39:56:55 avg_step=4.18s recent_step=4.13s eta=18:09:00 +2026-09-21 22:24:13 | INFO | train rank=0 | step=34390/50000 progress=68.8% loss=0.333981 lr=2.436315e-05 lr_backbone=2.436315e-05 elapsed=39:57:37 avg_step=4.18s recent_step=4.15s eta=18:08:18 +2026-09-21 22:24:55 | INFO | train rank=0 | step=34400/50000 progress=68.8% loss=0.321590 lr=2.433476e-05 lr_backbone=2.433476e-05 elapsed=39:58:18 avg_step=4.18s recent_step=4.13s eta=18:07:36 +2026-09-21 22:25:36 | INFO | train rank=0 | step=34410/50000 progress=68.8% loss=0.362696 lr=2.430639e-05 lr_backbone=2.430639e-05 elapsed=39:59:00 avg_step=4.18s recent_step=4.17s eta=18:06:54 +2026-09-21 22:26:18 | INFO | train rank=0 | step=34420/50000 progress=68.8% loss=0.301366 lr=2.427802e-05 lr_backbone=2.427802e-05 elapsed=39:59:41 avg_step=4.18s recent_step=4.16s eta=18:06:12 +2026-09-21 22:26:59 | INFO | train rank=0 | step=34430/50000 progress=68.9% loss=0.298491 lr=2.424967e-05 lr_backbone=2.424967e-05 elapsed=40:00:22 avg_step=4.18s recent_step=4.10s eta=18:05:30 +2026-09-21 22:27:40 | INFO | train rank=0 | step=34440/50000 progress=68.9% loss=0.304425 lr=2.422133e-05 lr_backbone=2.422133e-05 elapsed=40:01:04 avg_step=4.18s recent_step=4.14s eta=18:04:48 +2026-09-21 22:28:22 | INFO | train rank=0 | step=34450/50000 progress=68.9% loss=0.413698 lr=2.419300e-05 lr_backbone=2.419300e-05 elapsed=40:01:45 avg_step=4.18s recent_step=4.17s eta=18:04:06 +2026-09-21 22:29:03 | INFO | train rank=0 | step=34460/50000 progress=68.9% loss=0.317022 lr=2.416468e-05 lr_backbone=2.416468e-05 elapsed=40:02:26 avg_step=4.18s recent_step=4.12s eta=18:03:24 +2026-09-21 22:29:45 | INFO | train rank=0 | step=34470/50000 progress=68.9% loss=0.346233 lr=2.413637e-05 lr_backbone=2.413637e-05 elapsed=40:03:08 avg_step=4.18s recent_step=4.16s eta=18:02:42 +2026-09-21 22:30:26 | INFO | train rank=0 | step=34480/50000 progress=69.0% loss=0.380734 lr=2.410808e-05 lr_backbone=2.410808e-05 elapsed=40:03:50 avg_step=4.18s recent_step=4.16s eta=18:02:00 +2026-09-21 22:31:07 | INFO | train rank=0 | step=34490/50000 progress=69.0% loss=0.325310 lr=2.407979e-05 lr_backbone=2.407979e-05 elapsed=40:04:31 avg_step=4.18s recent_step=4.10s eta=18:01:18 +2026-09-21 22:31:49 | INFO | train rank=0 | step=34500/50000 progress=69.0% loss=0.315138 lr=2.405152e-05 lr_backbone=2.405152e-05 elapsed=40:05:12 avg_step=4.18s recent_step=4.13s eta=18:00:36 +2026-09-21 22:32:30 | INFO | train rank=0 | step=34510/50000 progress=69.0% loss=0.337332 lr=2.402326e-05 lr_backbone=2.402326e-05 elapsed=40:05:53 avg_step=4.18s recent_step=4.14s eta=17:59:54 +2026-09-21 22:33:12 | INFO | train rank=0 | step=34520/50000 progress=69.0% loss=0.323126 lr=2.399501e-05 lr_backbone=2.399501e-05 elapsed=40:06:35 avg_step=4.18s recent_step=4.15s eta=17:59:12 +2026-09-21 22:33:53 | INFO | train rank=0 | step=34530/50000 progress=69.1% loss=0.312185 lr=2.396677e-05 lr_backbone=2.396677e-05 elapsed=40:07:16 avg_step=4.18s recent_step=4.10s eta=17:58:29 +2026-09-21 22:34:34 | INFO | train rank=0 | step=34540/50000 progress=69.1% loss=0.306829 lr=2.393854e-05 lr_backbone=2.393854e-05 elapsed=40:07:57 avg_step=4.18s recent_step=4.15s eta=17:57:47 +2026-09-21 22:35:15 | INFO | train rank=0 | step=34550/50000 progress=69.1% loss=0.374690 lr=2.391033e-05 lr_backbone=2.391033e-05 elapsed=40:08:38 avg_step=4.18s recent_step=4.11s eta=17:57:05 +2026-09-21 22:35:56 | INFO | train rank=0 | step=34560/50000 progress=69.1% loss=0.291681 lr=2.388212e-05 lr_backbone=2.388212e-05 elapsed=40:09:20 avg_step=4.18s recent_step=4.13s eta=17:56:23 +2026-09-21 22:36:38 | INFO | train rank=0 | step=34570/50000 progress=69.1% loss=0.330280 lr=2.385393e-05 lr_backbone=2.385393e-05 elapsed=40:10:01 avg_step=4.18s recent_step=4.15s eta=17:55:41 +2026-09-21 22:37:19 | INFO | train rank=0 | step=34580/50000 progress=69.2% loss=0.287931 lr=2.382575e-05 lr_backbone=2.382575e-05 elapsed=40:10:43 avg_step=4.18s recent_step=4.16s eta=17:54:59 +2026-09-21 22:38:01 | INFO | train rank=0 | step=34590/50000 progress=69.2% loss=0.318138 lr=2.379758e-05 lr_backbone=2.379758e-05 elapsed=40:11:24 avg_step=4.18s recent_step=4.13s eta=17:54:17 +2026-09-21 22:38:42 | INFO | train rank=0 | step=34600/50000 progress=69.2% loss=0.332503 lr=2.376942e-05 lr_backbone=2.376942e-05 elapsed=40:12:06 avg_step=4.18s recent_step=4.15s eta=17:53:35 +2026-09-21 22:39:23 | INFO | train rank=0 | step=34610/50000 progress=69.2% loss=0.305592 lr=2.374127e-05 lr_backbone=2.374127e-05 elapsed=40:12:46 avg_step=4.18s recent_step=4.09s eta=17:52:53 +2026-09-21 22:40:04 | INFO | train rank=0 | step=34620/50000 progress=69.2% loss=0.333185 lr=2.371313e-05 lr_backbone=2.371313e-05 elapsed=40:13:28 avg_step=4.18s recent_step=4.13s eta=17:52:11 +2026-09-21 22:40:46 | INFO | train rank=0 | step=34630/50000 progress=69.3% loss=0.321034 lr=2.368501e-05 lr_backbone=2.368501e-05 elapsed=40:14:09 avg_step=4.18s recent_step=4.16s eta=17:51:29 +2026-09-21 22:41:27 | INFO | train rank=0 | step=34640/50000 progress=69.3% loss=0.301878 lr=2.365689e-05 lr_backbone=2.365689e-05 elapsed=40:14:51 avg_step=4.18s recent_step=4.14s eta=17:50:47 +2026-09-21 22:42:09 | INFO | train rank=0 | step=34650/50000 progress=69.3% loss=0.400387 lr=2.362879e-05 lr_backbone=2.362879e-05 elapsed=40:15:32 avg_step=4.18s recent_step=4.12s eta=17:50:05 +2026-09-21 22:42:50 | INFO | train rank=0 | step=34660/50000 progress=69.3% loss=0.421994 lr=2.360070e-05 lr_backbone=2.360070e-05 elapsed=40:16:14 avg_step=4.18s recent_step=4.15s eta=17:49:23 +2026-09-21 22:43:31 | INFO | train rank=0 | step=34670/50000 progress=69.3% loss=0.285789 lr=2.357262e-05 lr_backbone=2.357262e-05 elapsed=40:16:54 avg_step=4.18s recent_step=4.10s eta=17:48:41 +2026-09-21 22:44:12 | INFO | train rank=0 | step=34680/50000 progress=69.4% loss=0.308754 lr=2.354456e-05 lr_backbone=2.354456e-05 elapsed=40:17:36 avg_step=4.18s recent_step=4.13s eta=17:47:59 +2026-09-21 22:44:54 | INFO | train rank=0 | step=34690/50000 progress=69.4% loss=0.338954 lr=2.351650e-05 lr_backbone=2.351650e-05 elapsed=40:18:17 avg_step=4.18s recent_step=4.14s eta=17:47:17 +2026-09-21 22:45:35 | INFO | train rank=0 | step=34700/50000 progress=69.4% loss=0.303908 lr=2.348846e-05 lr_backbone=2.348846e-05 elapsed=40:18:59 avg_step=4.18s recent_step=4.14s eta=17:46:35 +2026-09-21 22:46:16 | INFO | train rank=0 | step=34710/50000 progress=69.4% loss=0.284943 lr=2.346043e-05 lr_backbone=2.346043e-05 elapsed=40:19:40 avg_step=4.18s recent_step=4.12s eta=17:45:52 +2026-09-21 22:46:58 | INFO | train rank=0 | step=34720/50000 progress=69.4% loss=0.296283 lr=2.343241e-05 lr_backbone=2.343241e-05 elapsed=40:20:21 avg_step=4.18s recent_step=4.16s eta=17:45:11 +2026-09-21 22:47:39 | INFO | train rank=0 | step=34730/50000 progress=69.5% loss=0.293188 lr=2.340440e-05 lr_backbone=2.340440e-05 elapsed=40:21:03 avg_step=4.18s recent_step=4.11s eta=17:44:28 +2026-09-21 22:48:20 | INFO | train rank=0 | step=34740/50000 progress=69.5% loss=0.306750 lr=2.337640e-05 lr_backbone=2.337640e-05 elapsed=40:21:44 avg_step=4.18s recent_step=4.11s eta=17:43:46 +2026-09-21 22:49:02 | INFO | train rank=0 | step=34750/50000 progress=69.5% loss=0.328756 lr=2.334841e-05 lr_backbone=2.334841e-05 elapsed=40:22:25 avg_step=4.18s recent_step=4.15s eta=17:43:04 +2026-09-21 22:49:43 | INFO | train rank=0 | step=34760/50000 progress=69.5% loss=0.356539 lr=2.332044e-05 lr_backbone=2.332044e-05 elapsed=40:23:07 avg_step=4.18s recent_step=4.16s eta=17:42:22 +2026-09-21 22:50:25 | INFO | train rank=0 | step=34770/50000 progress=69.5% loss=0.362764 lr=2.329248e-05 lr_backbone=2.329248e-05 elapsed=40:23:48 avg_step=4.18s recent_step=4.16s eta=17:41:40 +2026-09-21 22:51:06 | INFO | train rank=0 | step=34780/50000 progress=69.6% loss=0.342754 lr=2.326453e-05 lr_backbone=2.326453e-05 elapsed=40:24:30 avg_step=4.18s recent_step=4.14s eta=17:40:58 +2026-09-21 22:51:47 | INFO | train rank=0 | step=34790/50000 progress=69.6% loss=0.308149 lr=2.323659e-05 lr_backbone=2.323659e-05 elapsed=40:25:11 avg_step=4.18s recent_step=4.10s eta=17:40:16 +2026-09-21 22:52:29 | INFO | train rank=0 | step=34800/50000 progress=69.6% loss=0.336851 lr=2.320866e-05 lr_backbone=2.320866e-05 elapsed=40:25:52 avg_step=4.18s recent_step=4.14s eta=17:39:34 +2026-09-21 22:53:10 | INFO | train rank=0 | step=34810/50000 progress=69.6% loss=0.320772 lr=2.318074e-05 lr_backbone=2.318074e-05 elapsed=40:26:34 avg_step=4.18s recent_step=4.15s eta=17:38:52 +2026-09-21 22:53:52 | INFO | train rank=0 | step=34820/50000 progress=69.6% loss=0.292785 lr=2.315284e-05 lr_backbone=2.315284e-05 elapsed=40:27:15 avg_step=4.18s recent_step=4.16s eta=17:38:10 +2026-09-21 22:54:33 | INFO | train rank=0 | step=34830/50000 progress=69.7% loss=0.375708 lr=2.312495e-05 lr_backbone=2.312495e-05 elapsed=40:27:57 avg_step=4.18s recent_step=4.15s eta=17:37:28 +2026-09-21 22:55:15 | INFO | train rank=0 | step=34840/50000 progress=69.7% loss=0.306186 lr=2.309707e-05 lr_backbone=2.309707e-05 elapsed=40:28:38 avg_step=4.18s recent_step=4.13s eta=17:36:46 +2026-09-21 22:55:55 | INFO | train rank=0 | step=34850/50000 progress=69.7% loss=0.310795 lr=2.306920e-05 lr_backbone=2.306920e-05 elapsed=40:29:19 avg_step=4.18s recent_step=4.07s eta=17:36:04 +2026-09-21 22:56:36 | INFO | train rank=0 | step=34860/50000 progress=69.7% loss=0.359862 lr=2.304134e-05 lr_backbone=2.304134e-05 elapsed=40:30:00 avg_step=4.18s recent_step=4.10s eta=17:35:22 +2026-09-21 22:57:18 | INFO | train rank=0 | step=34870/50000 progress=69.7% loss=0.286226 lr=2.301350e-05 lr_backbone=2.301350e-05 elapsed=40:30:41 avg_step=4.18s recent_step=4.14s eta=17:34:40 +2026-09-21 22:57:59 | INFO | train rank=0 | step=34880/50000 progress=69.8% loss=0.344919 lr=2.298566e-05 lr_backbone=2.298566e-05 elapsed=40:31:23 avg_step=4.18s recent_step=4.14s eta=17:33:58 +2026-09-21 22:58:40 | INFO | train rank=0 | step=34890/50000 progress=69.8% loss=0.344057 lr=2.295784e-05 lr_backbone=2.295784e-05 elapsed=40:32:04 avg_step=4.18s recent_step=4.11s eta=17:33:16 +2026-09-21 22:59:22 | INFO | train rank=0 | step=34900/50000 progress=69.8% loss=0.300513 lr=2.293003e-05 lr_backbone=2.293003e-05 elapsed=40:32:45 avg_step=4.18s recent_step=4.14s eta=17:32:34 +2026-09-21 23:00:03 | INFO | train rank=0 | step=34910/50000 progress=69.8% loss=0.326325 lr=2.290224e-05 lr_backbone=2.290224e-05 elapsed=40:33:26 avg_step=4.18s recent_step=4.08s eta=17:31:51 +2026-09-21 23:00:44 | INFO | train rank=0 | step=34920/50000 progress=69.8% loss=0.333346 lr=2.287445e-05 lr_backbone=2.287445e-05 elapsed=40:34:07 avg_step=4.18s recent_step=4.10s eta=17:31:09 +2026-09-21 23:01:25 | INFO | train rank=0 | step=34930/50000 progress=69.9% loss=0.339381 lr=2.284668e-05 lr_backbone=2.284668e-05 elapsed=40:34:48 avg_step=4.18s recent_step=4.13s eta=17:30:27 +2026-09-21 23:02:06 | INFO | train rank=0 | step=34940/50000 progress=69.9% loss=0.309064 lr=2.281891e-05 lr_backbone=2.281891e-05 elapsed=40:35:29 avg_step=4.18s recent_step=4.12s eta=17:29:45 +2026-09-21 23:02:47 | INFO | train rank=0 | step=34950/50000 progress=69.9% loss=0.342321 lr=2.279116e-05 lr_backbone=2.279116e-05 elapsed=40:36:11 avg_step=4.18s recent_step=4.12s eta=17:29:03 +2026-09-21 23:03:29 | INFO | train rank=0 | step=34960/50000 progress=69.9% loss=0.341435 lr=2.276343e-05 lr_backbone=2.276343e-05 elapsed=40:36:52 avg_step=4.18s recent_step=4.13s eta=17:28:21 +2026-09-21 23:04:10 | INFO | train rank=0 | step=34970/50000 progress=69.9% loss=0.317596 lr=2.273570e-05 lr_backbone=2.273570e-05 elapsed=40:37:33 avg_step=4.18s recent_step=4.09s eta=17:27:39 +2026-09-21 23:04:51 | INFO | train rank=0 | step=34980/50000 progress=70.0% loss=0.317898 lr=2.270799e-05 lr_backbone=2.270799e-05 elapsed=40:38:14 avg_step=4.18s recent_step=4.11s eta=17:26:57 +2026-09-21 23:05:32 | INFO | train rank=0 | step=34990/50000 progress=70.0% loss=0.300867 lr=2.268028e-05 lr_backbone=2.268028e-05 elapsed=40:38:55 avg_step=4.18s recent_step=4.13s eta=17:26:15 +2026-09-21 23:06:13 | INFO | train rank=0 | step=35000/50000 progress=70.0% loss=0.320684 lr=2.265259e-05 lr_backbone=2.265259e-05 elapsed=40:39:37 avg_step=4.18s recent_step=4.15s eta=17:25:33 +2026-09-21 23:06:55 | INFO | train rank=0 | step=35010/50000 progress=70.0% loss=0.340561 lr=2.262491e-05 lr_backbone=2.262491e-05 elapsed=40:40:18 avg_step=4.18s recent_step=4.13s eta=17:24:51 +2026-09-21 23:07:36 | INFO | train rank=0 | step=35020/50000 progress=70.0% loss=0.307248 lr=2.259725e-05 lr_backbone=2.259725e-05 elapsed=40:40:59 avg_step=4.18s recent_step=4.14s eta=17:24:09 +2026-09-21 23:08:17 | INFO | train rank=0 | step=35030/50000 progress=70.1% loss=0.326790 lr=2.256959e-05 lr_backbone=2.256959e-05 elapsed=40:41:40 avg_step=4.18s recent_step=4.07s eta=17:23:26 +2026-09-21 23:08:58 | INFO | train rank=0 | step=35040/50000 progress=70.1% loss=0.355814 lr=2.254195e-05 lr_backbone=2.254195e-05 elapsed=40:42:21 avg_step=4.18s recent_step=4.12s eta=17:22:44 +2026-09-21 23:09:39 | INFO | train rank=0 | step=35050/50000 progress=70.1% loss=0.372520 lr=2.251432e-05 lr_backbone=2.251432e-05 elapsed=40:43:03 avg_step=4.18s recent_step=4.14s eta=17:22:02 +2026-09-21 23:10:21 | INFO | train rank=0 | step=35060/50000 progress=70.1% loss=0.445195 lr=2.248670e-05 lr_backbone=2.248670e-05 elapsed=40:43:44 avg_step=4.18s recent_step=4.14s eta=17:21:20 +2026-09-21 23:11:02 | INFO | train rank=0 | step=35070/50000 progress=70.1% loss=0.297049 lr=2.245909e-05 lr_backbone=2.245909e-05 elapsed=40:44:26 avg_step=4.18s recent_step=4.14s eta=17:20:38 +2026-09-21 23:11:44 | INFO | train rank=0 | step=35080/50000 progress=70.2% loss=0.287719 lr=2.243150e-05 lr_backbone=2.243150e-05 elapsed=40:45:07 avg_step=4.18s recent_step=4.13s eta=17:19:56 +2026-09-21 23:12:24 | INFO | train rank=0 | step=35090/50000 progress=70.2% loss=0.302902 lr=2.240392e-05 lr_backbone=2.240392e-05 elapsed=40:45:48 avg_step=4.18s recent_step=4.07s eta=17:19:14 +2026-09-21 23:13:05 | INFO | train rank=0 | step=35100/50000 progress=70.2% loss=0.412731 lr=2.237635e-05 lr_backbone=2.237635e-05 elapsed=40:46:29 avg_step=4.18s recent_step=4.10s eta=17:18:32 +2026-09-21 23:13:47 | INFO | train rank=0 | step=35110/50000 progress=70.2% loss=0.364361 lr=2.234879e-05 lr_backbone=2.234879e-05 elapsed=40:47:10 avg_step=4.18s recent_step=4.14s eta=17:17:50 +2026-09-21 23:14:28 | INFO | train rank=0 | step=35120/50000 progress=70.2% loss=0.343187 lr=2.232124e-05 lr_backbone=2.232124e-05 elapsed=40:47:51 avg_step=4.18s recent_step=4.13s eta=17:17:08 +2026-09-21 23:15:09 | INFO | train rank=0 | step=35130/50000 progress=70.3% loss=0.336280 lr=2.229371e-05 lr_backbone=2.229371e-05 elapsed=40:48:33 avg_step=4.18s recent_step=4.15s eta=17:16:26 +2026-09-21 23:15:51 | INFO | train rank=0 | step=35140/50000 progress=70.3% loss=0.281576 lr=2.226619e-05 lr_backbone=2.226619e-05 elapsed=40:49:14 avg_step=4.18s recent_step=4.14s eta=17:15:44 +2026-09-21 23:16:32 | INFO | train rank=0 | step=35150/50000 progress=70.3% loss=0.420496 lr=2.223868e-05 lr_backbone=2.223868e-05 elapsed=40:49:55 avg_step=4.18s recent_step=4.08s eta=17:15:01 +2026-09-21 23:17:13 | INFO | train rank=0 | step=35160/50000 progress=70.3% loss=0.358601 lr=2.221118e-05 lr_backbone=2.221118e-05 elapsed=40:50:36 avg_step=4.18s recent_step=4.11s eta=17:14:19 +2026-09-21 23:17:54 | INFO | train rank=0 | step=35170/50000 progress=70.3% loss=0.379384 lr=2.218369e-05 lr_backbone=2.218369e-05 elapsed=40:51:17 avg_step=4.18s recent_step=4.14s eta=17:13:37 +2026-09-21 23:18:36 | INFO | train rank=0 | step=35180/50000 progress=70.4% loss=0.445096 lr=2.215622e-05 lr_backbone=2.215622e-05 elapsed=40:51:59 avg_step=4.18s recent_step=4.15s eta=17:12:55 +2026-09-21 23:19:17 | INFO | train rank=0 | step=35190/50000 progress=70.4% loss=0.449380 lr=2.212876e-05 lr_backbone=2.212876e-05 elapsed=40:52:40 avg_step=4.18s recent_step=4.16s eta=17:12:13 +2026-09-21 23:19:59 | INFO | train rank=0 | step=35200/50000 progress=70.4% loss=0.385764 lr=2.210131e-05 lr_backbone=2.210131e-05 elapsed=40:53:22 avg_step=4.18s recent_step=4.14s eta=17:11:31 +2026-09-21 23:20:39 | INFO | train rank=0 | step=35210/50000 progress=70.4% loss=0.341652 lr=2.207387e-05 lr_backbone=2.207387e-05 elapsed=40:54:03 avg_step=4.18s recent_step=4.08s eta=17:10:49 +2026-09-21 23:21:21 | INFO | train rank=0 | step=35220/50000 progress=70.4% loss=0.415267 lr=2.204645e-05 lr_backbone=2.204645e-05 elapsed=40:54:44 avg_step=4.18s recent_step=4.12s eta=17:10:07 +2026-09-21 23:22:02 | INFO | train rank=0 | step=35230/50000 progress=70.5% loss=0.309380 lr=2.201904e-05 lr_backbone=2.201904e-05 elapsed=40:55:25 avg_step=4.18s recent_step=4.14s eta=17:09:25 +2026-09-21 23:22:43 | INFO | train rank=0 | step=35240/50000 progress=70.5% loss=0.299165 lr=2.199164e-05 lr_backbone=2.199164e-05 elapsed=40:56:07 avg_step=4.18s recent_step=4.13s eta=17:08:43 +2026-09-21 23:23:25 | INFO | train rank=0 | step=35250/50000 progress=70.5% loss=0.335547 lr=2.196425e-05 lr_backbone=2.196425e-05 elapsed=40:56:48 avg_step=4.18s recent_step=4.14s eta=17:08:01 +2026-09-21 23:24:06 | INFO | train rank=0 | step=35260/50000 progress=70.5% loss=0.367316 lr=2.193687e-05 lr_backbone=2.193687e-05 elapsed=40:57:29 avg_step=4.18s recent_step=4.13s eta=17:07:19 +2026-09-21 23:24:47 | INFO | train rank=0 | step=35270/50000 progress=70.5% loss=0.349842 lr=2.190951e-05 lr_backbone=2.190951e-05 elapsed=40:58:10 avg_step=4.18s recent_step=4.09s eta=17:06:37 +2026-09-21 23:25:28 | INFO | train rank=0 | step=35280/50000 progress=70.6% loss=0.312715 lr=2.188216e-05 lr_backbone=2.188216e-05 elapsed=40:58:51 avg_step=4.18s recent_step=4.11s eta=17:05:55 +2026-09-21 23:26:09 | INFO | train rank=0 | step=35290/50000 progress=70.6% loss=0.323297 lr=2.185482e-05 lr_backbone=2.185482e-05 elapsed=40:59:33 avg_step=4.18s recent_step=4.13s eta=17:05:13 +2026-09-21 23:26:51 | INFO | train rank=0 | step=35300/50000 progress=70.6% loss=0.404041 lr=2.182749e-05 lr_backbone=2.182749e-05 elapsed=41:00:14 avg_step=4.18s recent_step=4.15s eta=17:04:31 +2026-09-21 23:27:32 | INFO | train rank=0 | step=35310/50000 progress=70.6% loss=0.343508 lr=2.180018e-05 lr_backbone=2.180018e-05 elapsed=41:00:55 avg_step=4.18s recent_step=4.13s eta=17:03:49 +2026-09-21 23:28:13 | INFO | train rank=0 | step=35320/50000 progress=70.6% loss=0.348462 lr=2.177288e-05 lr_backbone=2.177288e-05 elapsed=41:01:37 avg_step=4.18s recent_step=4.13s eta=17:03:07 +2026-09-21 23:28:54 | INFO | train rank=0 | step=35330/50000 progress=70.7% loss=0.304799 lr=2.174559e-05 lr_backbone=2.174559e-05 elapsed=41:02:17 avg_step=4.18s recent_step=4.08s eta=17:02:24 +2026-09-21 23:29:35 | INFO | train rank=0 | step=35340/50000 progress=70.7% loss=0.361618 lr=2.171831e-05 lr_backbone=2.171831e-05 elapsed=41:02:58 avg_step=4.18s recent_step=4.11s eta=17:01:42 +2026-09-21 23:30:17 | INFO | train rank=0 | step=35350/50000 progress=70.7% loss=0.284175 lr=2.169104e-05 lr_backbone=2.169104e-05 elapsed=41:03:40 avg_step=4.18s recent_step=4.17s eta=17:01:00 +2026-09-21 23:30:58 | INFO | train rank=0 | step=35360/50000 progress=70.7% loss=0.283516 lr=2.166379e-05 lr_backbone=2.166379e-05 elapsed=41:04:22 avg_step=4.18s recent_step=4.15s eta=17:00:18 +2026-09-21 23:31:40 | INFO | train rank=0 | step=35370/50000 progress=70.7% loss=0.309395 lr=2.163655e-05 lr_backbone=2.163655e-05 elapsed=41:05:03 avg_step=4.18s recent_step=4.16s eta=16:59:37 +2026-09-21 23:32:21 | INFO | train rank=0 | step=35380/50000 progress=70.8% loss=0.306289 lr=2.160933e-05 lr_backbone=2.160933e-05 elapsed=41:05:45 avg_step=4.18s recent_step=4.15s eta=16:58:55 +2026-09-21 23:33:03 | INFO | train rank=0 | step=35390/50000 progress=70.8% loss=0.359777 lr=2.158211e-05 lr_backbone=2.158211e-05 elapsed=41:06:26 avg_step=4.18s recent_step=4.11s eta=16:58:12 +2026-09-21 23:33:44 | INFO | train rank=0 | step=35400/50000 progress=70.8% loss=0.406189 lr=2.155491e-05 lr_backbone=2.155491e-05 elapsed=41:07:07 avg_step=4.18s recent_step=4.13s eta=16:57:30 +2026-09-21 23:34:25 | INFO | train rank=0 | step=35410/50000 progress=70.8% loss=0.358032 lr=2.152772e-05 lr_backbone=2.152772e-05 elapsed=41:07:48 avg_step=4.18s recent_step=4.11s eta=16:56:48 +2026-09-21 23:35:06 | INFO | train rank=0 | step=35420/50000 progress=70.8% loss=0.349553 lr=2.150054e-05 lr_backbone=2.150054e-05 elapsed=41:08:30 avg_step=4.18s recent_step=4.14s eta=16:56:06 +2026-09-21 23:35:48 | INFO | train rank=0 | step=35430/50000 progress=70.9% loss=0.307560 lr=2.147337e-05 lr_backbone=2.147337e-05 elapsed=41:09:11 avg_step=4.18s recent_step=4.17s eta=16:55:24 +2026-09-21 23:36:29 | INFO | train rank=0 | step=35440/50000 progress=70.9% loss=0.451717 lr=2.144622e-05 lr_backbone=2.144622e-05 elapsed=41:09:53 avg_step=4.18s recent_step=4.15s eta=16:54:43 +2026-09-21 23:37:11 | INFO | train rank=0 | step=35450/50000 progress=70.9% loss=0.329678 lr=2.141908e-05 lr_backbone=2.141908e-05 elapsed=41:10:34 avg_step=4.18s recent_step=4.11s eta=16:54:00 +2026-09-21 23:37:52 | INFO | train rank=0 | step=35460/50000 progress=70.9% loss=0.361939 lr=2.139195e-05 lr_backbone=2.139195e-05 elapsed=41:11:15 avg_step=4.18s recent_step=4.12s eta=16:53:18 +2026-09-21 23:38:33 | INFO | train rank=0 | step=35470/50000 progress=70.9% loss=0.297066 lr=2.136484e-05 lr_backbone=2.136484e-05 elapsed=41:11:56 avg_step=4.18s recent_step=4.13s eta=16:52:36 +2026-09-21 23:39:15 | INFO | train rank=0 | step=35480/50000 progress=71.0% loss=0.329351 lr=2.133773e-05 lr_backbone=2.133773e-05 elapsed=41:12:38 avg_step=4.18s recent_step=4.16s eta=16:51:54 +2026-09-21 23:39:56 | INFO | train rank=0 | step=35490/50000 progress=71.0% loss=0.430726 lr=2.131064e-05 lr_backbone=2.131064e-05 elapsed=41:13:19 avg_step=4.18s recent_step=4.15s eta=16:51:12 +2026-09-21 23:40:38 | INFO | train rank=0 | step=35500/50000 progress=71.0% loss=0.366890 lr=2.128357e-05 lr_backbone=2.128357e-05 elapsed=41:14:01 avg_step=4.18s recent_step=4.17s eta=16:50:31 +2026-09-21 23:41:19 | INFO | train rank=0 | step=35510/50000 progress=71.0% loss=0.322541 lr=2.125650e-05 lr_backbone=2.125650e-05 elapsed=41:14:42 avg_step=4.18s recent_step=4.10s eta=16:49:48 +2026-09-21 23:42:00 | INFO | train rank=0 | step=35520/50000 progress=71.0% loss=0.335555 lr=2.122945e-05 lr_backbone=2.122945e-05 elapsed=41:15:23 avg_step=4.18s recent_step=4.13s eta=16:49:06 +2026-09-21 23:42:42 | INFO | train rank=0 | step=35530/50000 progress=71.1% loss=0.299044 lr=2.120241e-05 lr_backbone=2.120241e-05 elapsed=41:16:05 avg_step=4.18s recent_step=4.14s eta=16:48:24 +2026-09-21 23:43:23 | INFO | train rank=0 | step=35540/50000 progress=71.1% loss=0.258303 lr=2.117538e-05 lr_backbone=2.117538e-05 elapsed=41:16:46 avg_step=4.18s recent_step=4.15s eta=16:47:42 +2026-09-21 23:44:04 | INFO | train rank=0 | step=35550/50000 progress=71.1% loss=0.323817 lr=2.114837e-05 lr_backbone=2.114837e-05 elapsed=41:17:28 avg_step=4.18s recent_step=4.14s eta=16:47:01 +2026-09-21 23:44:46 | INFO | train rank=0 | step=35560/50000 progress=71.1% loss=0.277206 lr=2.112136e-05 lr_backbone=2.112136e-05 elapsed=41:18:09 avg_step=4.18s recent_step=4.15s eta=16:46:19 +2026-09-21 23:45:27 | INFO | train rank=0 | step=35570/50000 progress=71.1% loss=0.329093 lr=2.109438e-05 lr_backbone=2.109438e-05 elapsed=41:18:50 avg_step=4.18s recent_step=4.10s eta=16:45:36 +2026-09-21 23:46:08 | INFO | train rank=0 | step=35580/50000 progress=71.2% loss=0.318765 lr=2.106740e-05 lr_backbone=2.106740e-05 elapsed=41:19:32 avg_step=4.18s recent_step=4.14s eta=16:44:54 +2026-09-21 23:46:50 | INFO | train rank=0 | step=35590/50000 progress=71.2% loss=0.325136 lr=2.104043e-05 lr_backbone=2.104043e-05 elapsed=41:20:13 avg_step=4.18s recent_step=4.13s eta=16:44:12 +2026-09-21 23:47:31 | INFO | train rank=0 | step=35600/50000 progress=71.2% loss=0.327337 lr=2.101348e-05 lr_backbone=2.101348e-05 elapsed=41:20:55 avg_step=4.18s recent_step=4.15s eta=16:43:31 +2026-09-21 23:48:13 | INFO | train rank=0 | step=35610/50000 progress=71.2% loss=0.308730 lr=2.098654e-05 lr_backbone=2.098654e-05 elapsed=41:21:36 avg_step=4.18s recent_step=4.16s eta=16:42:49 +2026-09-21 23:48:54 | INFO | train rank=0 | step=35620/50000 progress=71.2% loss=0.322345 lr=2.095962e-05 lr_backbone=2.095962e-05 elapsed=41:22:17 avg_step=4.18s recent_step=4.14s eta=16:42:07 +2026-09-21 23:49:35 | INFO | train rank=0 | step=35630/50000 progress=71.3% loss=0.327128 lr=2.093270e-05 lr_backbone=2.093270e-05 elapsed=41:22:58 avg_step=4.18s recent_step=4.09s eta=16:41:24 +2026-09-21 23:50:16 | INFO | train rank=0 | step=35640/50000 progress=71.3% loss=0.322963 lr=2.090580e-05 lr_backbone=2.090580e-05 elapsed=41:23:40 avg_step=4.18s recent_step=4.13s eta=16:40:42 +2026-09-21 23:50:58 | INFO | train rank=0 | step=35650/50000 progress=71.3% loss=0.315515 lr=2.087892e-05 lr_backbone=2.087892e-05 elapsed=41:24:21 avg_step=4.18s recent_step=4.12s eta=16:40:00 +2026-09-21 23:51:39 | INFO | train rank=0 | step=35660/50000 progress=71.3% loss=0.376336 lr=2.085204e-05 lr_backbone=2.085204e-05 elapsed=41:25:03 avg_step=4.18s recent_step=4.17s eta=16:39:19 +2026-09-21 23:52:21 | INFO | train rank=0 | step=35670/50000 progress=71.3% loss=0.393262 lr=2.082518e-05 lr_backbone=2.082518e-05 elapsed=41:25:44 avg_step=4.18s recent_step=4.16s eta=16:38:37 +2026-09-21 23:53:02 | INFO | train rank=0 | step=35680/50000 progress=71.4% loss=0.308646 lr=2.079833e-05 lr_backbone=2.079833e-05 elapsed=41:26:26 avg_step=4.18s recent_step=4.16s eta=16:37:55 +2026-09-21 23:53:44 | INFO | train rank=0 | step=35690/50000 progress=71.4% loss=0.316280 lr=2.077149e-05 lr_backbone=2.077149e-05 elapsed=41:27:07 avg_step=4.18s recent_step=4.10s eta=16:37:13 +2026-09-21 23:54:25 | INFO | train rank=0 | step=35700/50000 progress=71.4% loss=0.332159 lr=2.074467e-05 lr_backbone=2.074467e-05 elapsed=41:27:48 avg_step=4.18s recent_step=4.13s eta=16:36:31 +2026-09-21 23:55:06 | INFO | train rank=0 | step=35710/50000 progress=71.4% loss=0.284844 lr=2.071785e-05 lr_backbone=2.071785e-05 elapsed=41:28:30 avg_step=4.18s recent_step=4.14s eta=16:35:49 +2026-09-21 23:55:48 | INFO | train rank=0 | step=35720/50000 progress=71.4% loss=0.324000 lr=2.069106e-05 lr_backbone=2.069106e-05 elapsed=41:29:11 avg_step=4.18s recent_step=4.15s eta=16:35:07 +2026-09-21 23:56:29 | INFO | train rank=0 | step=35730/50000 progress=71.5% loss=0.371726 lr=2.066427e-05 lr_backbone=2.066427e-05 elapsed=41:29:53 avg_step=4.18s recent_step=4.16s eta=16:34:25 +2026-09-21 23:57:11 | INFO | train rank=0 | step=35740/50000 progress=71.5% loss=0.297849 lr=2.063750e-05 lr_backbone=2.063750e-05 elapsed=41:30:34 avg_step=4.18s recent_step=4.15s eta=16:33:43 +2026-09-21 23:57:52 | INFO | train rank=0 | step=35750/50000 progress=71.5% loss=0.310870 lr=2.061074e-05 lr_backbone=2.061074e-05 elapsed=41:31:15 avg_step=4.18s recent_step=4.10s eta=16:33:01 +2026-09-21 23:58:33 | INFO | train rank=0 | step=35760/50000 progress=71.5% loss=0.352010 lr=2.058399e-05 lr_backbone=2.058399e-05 elapsed=41:31:57 avg_step=4.18s recent_step=4.13s eta=16:32:19 +2026-09-21 23:59:15 | INFO | train rank=0 | step=35770/50000 progress=71.5% loss=0.268940 lr=2.055726e-05 lr_backbone=2.055726e-05 elapsed=41:32:38 avg_step=4.18s recent_step=4.14s eta=16:31:37 +2026-09-21 23:59:56 | INFO | train rank=0 | step=35780/50000 progress=71.6% loss=0.288006 lr=2.053053e-05 lr_backbone=2.053053e-05 elapsed=41:33:19 avg_step=4.18s recent_step=4.15s eta=16:30:55 +2026-09-22 00:00:38 | INFO | train rank=0 | step=35790/50000 progress=71.6% loss=0.419440 lr=2.050383e-05 lr_backbone=2.050383e-05 elapsed=41:34:01 avg_step=4.18s recent_step=4.14s eta=16:30:13 +2026-09-22 00:01:19 | INFO | train rank=0 | step=35800/50000 progress=71.6% loss=0.327358 lr=2.047713e-05 lr_backbone=2.047713e-05 elapsed=41:34:43 avg_step=4.18s recent_step=4.17s eta=16:29:31 +2026-09-22 00:02:00 | INFO | train rank=0 | step=35810/50000 progress=71.6% loss=0.337911 lr=2.045045e-05 lr_backbone=2.045045e-05 elapsed=41:35:24 avg_step=4.18s recent_step=4.11s eta=16:28:49 +2026-09-22 00:02:42 | INFO | train rank=0 | step=35820/50000 progress=71.6% loss=0.431110 lr=2.042378e-05 lr_backbone=2.042378e-05 elapsed=41:36:05 avg_step=4.18s recent_step=4.14s eta=16:28:07 +2026-09-22 00:03:23 | INFO | train rank=0 | step=35830/50000 progress=71.7% loss=0.288311 lr=2.039712e-05 lr_backbone=2.039712e-05 elapsed=41:36:46 avg_step=4.18s recent_step=4.12s eta=16:27:25 +2026-09-22 00:04:05 | INFO | train rank=0 | step=35840/50000 progress=71.7% loss=0.313782 lr=2.037048e-05 lr_backbone=2.037048e-05 elapsed=41:37:28 avg_step=4.18s recent_step=4.17s eta=16:26:43 +2026-09-22 00:04:46 | INFO | train rank=0 | step=35850/50000 progress=71.7% loss=0.335814 lr=2.034385e-05 lr_backbone=2.034385e-05 elapsed=41:38:09 avg_step=4.18s recent_step=4.16s eta=16:26:01 +2026-09-22 00:05:28 | INFO | train rank=0 | step=35860/50000 progress=71.7% loss=0.300589 lr=2.031723e-05 lr_backbone=2.031723e-05 elapsed=41:38:51 avg_step=4.18s recent_step=4.16s eta=16:25:19 +2026-09-22 00:06:09 | INFO | train rank=0 | step=35870/50000 progress=71.7% loss=0.406442 lr=2.029062e-05 lr_backbone=2.029062e-05 elapsed=41:39:32 avg_step=4.18s recent_step=4.12s eta=16:24:37 +2026-09-22 00:06:50 | INFO | train rank=0 | step=35880/50000 progress=71.8% loss=0.370427 lr=2.026403e-05 lr_backbone=2.026403e-05 elapsed=41:40:13 avg_step=4.18s recent_step=4.11s eta=16:23:55 +2026-09-22 00:07:31 | INFO | train rank=0 | step=35890/50000 progress=71.8% loss=0.310553 lr=2.023745e-05 lr_backbone=2.023745e-05 elapsed=41:40:55 avg_step=4.18s recent_step=4.15s eta=16:23:13 +2026-09-22 00:08:13 | INFO | train rank=0 | step=35900/50000 progress=71.8% loss=0.392831 lr=2.021088e-05 lr_backbone=2.021088e-05 elapsed=41:41:36 avg_step=4.18s recent_step=4.15s eta=16:22:31 +2026-09-22 00:08:55 | INFO | train rank=0 | step=35910/50000 progress=71.8% loss=0.328377 lr=2.018433e-05 lr_backbone=2.018433e-05 elapsed=41:42:18 avg_step=4.18s recent_step=4.16s eta=16:21:49 +2026-09-22 00:09:36 | INFO | train rank=0 | step=35920/50000 progress=71.8% loss=0.335445 lr=2.015779e-05 lr_backbone=2.015779e-05 elapsed=41:42:59 avg_step=4.18s recent_step=4.14s eta=16:21:07 +2026-09-22 00:10:17 | INFO | train rank=0 | step=35930/50000 progress=71.9% loss=0.327336 lr=2.013126e-05 lr_backbone=2.013126e-05 elapsed=41:43:40 avg_step=4.18s recent_step=4.11s eta=16:20:25 +2026-09-22 00:10:58 | INFO | train rank=0 | step=35940/50000 progress=71.9% loss=0.290662 lr=2.010475e-05 lr_backbone=2.010475e-05 elapsed=41:44:22 avg_step=4.18s recent_step=4.13s eta=16:19:43 +2026-09-22 00:11:40 | INFO | train rank=0 | step=35950/50000 progress=71.9% loss=0.483502 lr=2.007825e-05 lr_backbone=2.007825e-05 elapsed=41:45:03 avg_step=4.18s recent_step=4.13s eta=16:19:01 +2026-09-22 00:12:21 | INFO | train rank=0 | step=35960/50000 progress=71.9% loss=0.330945 lr=2.005176e-05 lr_backbone=2.005176e-05 elapsed=41:45:45 avg_step=4.18s recent_step=4.17s eta=16:18:19 +2026-09-22 00:13:03 | INFO | train rank=0 | step=35970/50000 progress=71.9% loss=0.316311 lr=2.002529e-05 lr_backbone=2.002529e-05 elapsed=41:46:26 avg_step=4.18s recent_step=4.13s eta=16:17:37 +2026-09-22 00:13:44 | INFO | train rank=0 | step=35980/50000 progress=72.0% loss=0.388868 lr=1.999883e-05 lr_backbone=1.999883e-05 elapsed=41:47:08 avg_step=4.18s recent_step=4.18s eta=16:16:55 +2026-09-22 00:14:25 | INFO | train rank=0 | step=35990/50000 progress=72.0% loss=0.289343 lr=1.997238e-05 lr_backbone=1.997238e-05 elapsed=41:47:49 avg_step=4.18s recent_step=4.11s eta=16:16:13 +2026-09-22 00:15:07 | INFO | train rank=0 | step=36000/50000 progress=72.0% loss=0.363007 lr=1.994594e-05 lr_backbone=1.994594e-05 elapsed=41:48:30 avg_step=4.18s recent_step=4.13s eta=16:15:31 +2026-09-22 00:15:48 | INFO | train rank=0 | step=36010/50000 progress=72.0% loss=0.327074 lr=1.991952e-05 lr_backbone=1.991952e-05 elapsed=41:49:12 avg_step=4.18s recent_step=4.16s eta=16:14:50 +2026-09-22 00:16:30 | INFO | train rank=0 | step=36020/50000 progress=72.0% loss=0.368327 lr=1.989311e-05 lr_backbone=1.989311e-05 elapsed=41:49:53 avg_step=4.18s recent_step=4.15s eta=16:14:08 +2026-09-22 00:17:11 | INFO | train rank=0 | step=36030/50000 progress=72.1% loss=0.288256 lr=1.986672e-05 lr_backbone=1.986672e-05 elapsed=41:50:35 avg_step=4.18s recent_step=4.15s eta=16:13:26 +2026-09-22 00:17:53 | INFO | train rank=0 | step=36040/50000 progress=72.1% loss=0.488103 lr=1.984033e-05 lr_backbone=1.984033e-05 elapsed=41:51:16 avg_step=4.18s recent_step=4.16s eta=16:12:44 +2026-09-22 00:18:34 | INFO | train rank=0 | step=36050/50000 progress=72.1% loss=0.335960 lr=1.981396e-05 lr_backbone=1.981396e-05 elapsed=41:51:57 avg_step=4.18s recent_step=4.10s eta=16:12:02 +2026-09-22 00:19:15 | INFO | train rank=0 | step=36060/50000 progress=72.1% loss=0.332341 lr=1.978761e-05 lr_backbone=1.978761e-05 elapsed=41:52:39 avg_step=4.18s recent_step=4.13s eta=16:11:20 +2026-09-22 00:19:57 | INFO | train rank=0 | step=36070/50000 progress=72.1% loss=0.302476 lr=1.976126e-05 lr_backbone=1.976126e-05 elapsed=41:53:20 avg_step=4.18s recent_step=4.15s eta=16:10:38 +2026-09-22 00:20:38 | INFO | train rank=0 | step=36080/50000 progress=72.2% loss=0.291549 lr=1.973493e-05 lr_backbone=1.973493e-05 elapsed=41:54:01 avg_step=4.18s recent_step=4.13s eta=16:09:56 +2026-09-22 00:21:20 | INFO | train rank=0 | step=36090/50000 progress=72.2% loss=0.315301 lr=1.970862e-05 lr_backbone=1.970862e-05 elapsed=41:54:43 avg_step=4.18s recent_step=4.16s eta=16:09:14 +2026-09-22 00:22:01 | INFO | train rank=0 | step=36100/50000 progress=72.2% loss=0.268068 lr=1.968232e-05 lr_backbone=1.968232e-05 elapsed=41:55:25 avg_step=4.18s recent_step=4.17s eta=16:08:32 +2026-09-22 00:22:42 | INFO | train rank=0 | step=36110/50000 progress=72.2% loss=0.347447 lr=1.965603e-05 lr_backbone=1.965603e-05 elapsed=41:56:06 avg_step=4.18s recent_step=4.08s eta=16:07:50 +2026-09-22 00:23:24 | INFO | train rank=0 | step=36120/50000 progress=72.2% loss=0.318361 lr=1.962975e-05 lr_backbone=1.962975e-05 elapsed=41:56:47 avg_step=4.18s recent_step=4.13s eta=16:07:08 +2026-09-22 00:24:05 | INFO | train rank=0 | step=36130/50000 progress=72.3% loss=0.375693 lr=1.960349e-05 lr_backbone=1.960349e-05 elapsed=41:57:28 avg_step=4.18s recent_step=4.16s eta=16:06:26 +2026-09-22 00:24:47 | INFO | train rank=0 | step=36140/50000 progress=72.3% loss=0.370076 lr=1.957723e-05 lr_backbone=1.957723e-05 elapsed=41:58:10 avg_step=4.18s recent_step=4.17s eta=16:05:44 +2026-09-22 00:25:28 | INFO | train rank=0 | step=36150/50000 progress=72.3% loss=0.295848 lr=1.955100e-05 lr_backbone=1.955100e-05 elapsed=41:58:51 avg_step=4.18s recent_step=4.10s eta=16:05:02 +2026-09-22 00:26:09 | INFO | train rank=0 | step=36160/50000 progress=72.3% loss=0.347204 lr=1.952477e-05 lr_backbone=1.952477e-05 elapsed=41:59:33 avg_step=4.18s recent_step=4.15s eta=16:04:20 +2026-09-22 00:26:50 | INFO | train rank=0 | step=36170/50000 progress=72.3% loss=0.267708 lr=1.949856e-05 lr_backbone=1.949856e-05 elapsed=42:00:14 avg_step=4.18s recent_step=4.09s eta=16:03:38 +2026-09-22 00:27:32 | INFO | train rank=0 | step=36180/50000 progress=72.4% loss=0.283464 lr=1.947237e-05 lr_backbone=1.947237e-05 elapsed=42:00:55 avg_step=4.18s recent_step=4.13s eta=16:02:56 +2026-09-22 00:28:13 | INFO | train rank=0 | step=36190/50000 progress=72.4% loss=0.408448 lr=1.944618e-05 lr_backbone=1.944618e-05 elapsed=42:01:37 avg_step=4.18s recent_step=4.16s eta=16:02:14 +2026-09-22 00:28:55 | INFO | train rank=0 | step=36200/50000 progress=72.4% loss=0.302159 lr=1.942001e-05 lr_backbone=1.942001e-05 elapsed=42:02:18 avg_step=4.18s recent_step=4.16s eta=16:01:32 +2026-09-22 00:29:36 | INFO | train rank=0 | step=36210/50000 progress=72.4% loss=0.287598 lr=1.939386e-05 lr_backbone=1.939386e-05 elapsed=42:02:59 avg_step=4.18s recent_step=4.11s eta=16:00:50 +2026-09-22 00:30:17 | INFO | train rank=0 | step=36220/50000 progress=72.4% loss=0.312740 lr=1.936771e-05 lr_backbone=1.936771e-05 elapsed=42:03:41 avg_step=4.18s recent_step=4.15s eta=16:00:08 +2026-09-22 00:30:59 | INFO | train rank=0 | step=36230/50000 progress=72.5% loss=0.371820 lr=1.934158e-05 lr_backbone=1.934158e-05 elapsed=42:04:22 avg_step=4.18s recent_step=4.13s eta=15:59:26 +2026-09-22 00:31:40 | INFO | train rank=0 | step=36240/50000 progress=72.5% loss=0.380221 lr=1.931547e-05 lr_backbone=1.931547e-05 elapsed=42:05:03 avg_step=4.18s recent_step=4.12s eta=15:58:44 +2026-09-22 00:32:21 | INFO | train rank=0 | step=36250/50000 progress=72.5% loss=0.293597 lr=1.928936e-05 lr_backbone=1.928936e-05 elapsed=42:05:45 avg_step=4.18s recent_step=4.15s eta=15:58:02 +2026-09-22 00:33:03 | INFO | train rank=0 | step=36260/50000 progress=72.5% loss=0.267640 lr=1.926327e-05 lr_backbone=1.926327e-05 elapsed=42:06:26 avg_step=4.18s recent_step=4.15s eta=15:57:20 +2026-09-22 00:33:45 | INFO | train rank=0 | step=36270/50000 progress=72.5% loss=0.328842 lr=1.923720e-05 lr_backbone=1.923720e-05 elapsed=42:07:08 avg_step=4.18s recent_step=4.16s eta=15:56:38 +2026-09-22 00:34:26 | INFO | train rank=0 | step=36280/50000 progress=72.6% loss=0.333206 lr=1.921114e-05 lr_backbone=1.921114e-05 elapsed=42:07:49 avg_step=4.18s recent_step=4.12s eta=15:55:56 +2026-09-22 00:35:07 | INFO | train rank=0 | step=36290/50000 progress=72.6% loss=0.285200 lr=1.918509e-05 lr_backbone=1.918509e-05 elapsed=42:08:30 avg_step=4.18s recent_step=4.10s eta=15:55:14 +2026-09-22 00:35:48 | INFO | train rank=0 | step=36300/50000 progress=72.6% loss=0.392031 lr=1.915905e-05 lr_backbone=1.915905e-05 elapsed=42:09:11 avg_step=4.18s recent_step=4.12s eta=15:54:32 +2026-09-22 00:36:30 | INFO | train rank=0 | step=36310/50000 progress=72.6% loss=0.306112 lr=1.913303e-05 lr_backbone=1.913303e-05 elapsed=42:09:53 avg_step=4.18s recent_step=4.17s eta=15:53:50 +2026-09-22 00:37:11 | INFO | train rank=0 | step=36320/50000 progress=72.6% loss=0.301388 lr=1.910702e-05 lr_backbone=1.910702e-05 elapsed=42:10:34 avg_step=4.18s recent_step=4.13s eta=15:53:08 +2026-09-22 00:37:53 | INFO | train rank=0 | step=36330/50000 progress=72.7% loss=0.336075 lr=1.908102e-05 lr_backbone=1.908102e-05 elapsed=42:11:16 avg_step=4.18s recent_step=4.16s eta=15:52:27 +2026-09-22 00:38:34 | INFO | train rank=0 | step=36340/50000 progress=72.7% loss=0.326865 lr=1.905504e-05 lr_backbone=1.905504e-05 elapsed=42:11:58 avg_step=4.18s recent_step=4.15s eta=15:51:45 +2026-09-22 00:39:15 | INFO | train rank=0 | step=36350/50000 progress=72.7% loss=0.300820 lr=1.902907e-05 lr_backbone=1.902907e-05 elapsed=42:12:38 avg_step=4.18s recent_step=4.09s eta=15:51:02 +2026-09-22 00:39:56 | INFO | train rank=0 | step=36360/50000 progress=72.7% loss=0.293389 lr=1.900312e-05 lr_backbone=1.900312e-05 elapsed=42:13:20 avg_step=4.18s recent_step=4.14s eta=15:50:21 +2026-09-22 00:40:38 | INFO | train rank=0 | step=36370/50000 progress=72.7% loss=0.393312 lr=1.897718e-05 lr_backbone=1.897718e-05 elapsed=42:14:01 avg_step=4.18s recent_step=4.14s eta=15:49:39 +2026-09-22 00:41:19 | INFO | train rank=0 | step=36380/50000 progress=72.8% loss=0.391580 lr=1.895125e-05 lr_backbone=1.895125e-05 elapsed=42:14:43 avg_step=4.18s recent_step=4.15s eta=15:48:57 +2026-09-22 00:42:01 | INFO | train rank=0 | step=36390/50000 progress=72.8% loss=0.306166 lr=1.892534e-05 lr_backbone=1.892534e-05 elapsed=42:15:24 avg_step=4.18s recent_step=4.15s eta=15:48:15 +2026-09-22 00:42:42 | INFO | train rank=0 | step=36400/50000 progress=72.8% loss=0.310008 lr=1.889944e-05 lr_backbone=1.889944e-05 elapsed=42:16:05 avg_step=4.18s recent_step=4.13s eta=15:47:33 +2026-09-22 00:43:23 | INFO | train rank=0 | step=36410/50000 progress=72.8% loss=0.340167 lr=1.887355e-05 lr_backbone=1.887355e-05 elapsed=42:16:47 avg_step=4.18s recent_step=4.11s eta=15:46:51 +2026-09-22 00:44:05 | INFO | train rank=0 | step=36420/50000 progress=72.8% loss=0.375546 lr=1.884768e-05 lr_backbone=1.884768e-05 elapsed=42:17:28 avg_step=4.18s recent_step=4.13s eta=15:46:09 +2026-09-22 00:44:46 | INFO | train rank=0 | step=36430/50000 progress=72.9% loss=0.327163 lr=1.882182e-05 lr_backbone=1.882182e-05 elapsed=42:18:09 avg_step=4.18s recent_step=4.14s eta=15:45:27 +2026-09-22 00:45:27 | INFO | train rank=0 | step=36440/50000 progress=72.9% loss=0.316708 lr=1.879597e-05 lr_backbone=1.879597e-05 elapsed=42:18:51 avg_step=4.18s recent_step=4.14s eta=15:44:45 +2026-09-22 00:46:09 | INFO | train rank=0 | step=36450/50000 progress=72.9% loss=0.305912 lr=1.877014e-05 lr_backbone=1.877014e-05 elapsed=42:19:32 avg_step=4.18s recent_step=4.16s eta=15:44:03 +2026-09-22 00:46:50 | INFO | train rank=0 | step=36460/50000 progress=72.9% loss=0.326799 lr=1.874432e-05 lr_backbone=1.874432e-05 elapsed=42:20:14 avg_step=4.18s recent_step=4.13s eta=15:43:21 +2026-09-22 00:47:31 | INFO | train rank=0 | step=36470/50000 progress=72.9% loss=0.290091 lr=1.871851e-05 lr_backbone=1.871851e-05 elapsed=42:20:54 avg_step=4.18s recent_step=4.09s eta=15:42:39 +2026-09-22 00:48:12 | INFO | train rank=0 | step=36480/50000 progress=73.0% loss=0.283347 lr=1.869272e-05 lr_backbone=1.869272e-05 elapsed=42:21:35 avg_step=4.18s recent_step=4.10s eta=15:41:57 +2026-09-22 00:48:53 | INFO | train rank=0 | step=36490/50000 progress=73.0% loss=0.351016 lr=1.866695e-05 lr_backbone=1.866695e-05 elapsed=42:22:17 avg_step=4.18s recent_step=4.11s eta=15:41:15 +2026-09-22 00:49:35 | INFO | train rank=0 | step=36500/50000 progress=73.0% loss=0.328343 lr=1.864118e-05 lr_backbone=1.864118e-05 elapsed=42:22:58 avg_step=4.18s recent_step=4.14s eta=15:40:33 +2026-09-22 00:50:16 | INFO | train rank=0 | step=36510/50000 progress=73.0% loss=0.339378 lr=1.861543e-05 lr_backbone=1.861543e-05 elapsed=42:23:40 avg_step=4.18s recent_step=4.16s eta=15:39:51 +2026-09-22 00:50:58 | INFO | train rank=0 | step=36520/50000 progress=73.0% loss=0.335544 lr=1.858970e-05 lr_backbone=1.858970e-05 elapsed=42:24:21 avg_step=4.18s recent_step=4.15s eta=15:39:09 +2026-09-22 00:51:38 | INFO | train rank=0 | step=36530/50000 progress=73.1% loss=0.349228 lr=1.856397e-05 lr_backbone=1.856397e-05 elapsed=42:25:02 avg_step=4.18s recent_step=4.07s eta=15:38:27 +2026-09-22 00:52:20 | INFO | train rank=0 | step=36540/50000 progress=73.1% loss=0.288936 lr=1.853826e-05 lr_backbone=1.853826e-05 elapsed=42:25:43 avg_step=4.18s recent_step=4.13s eta=15:37:45 +2026-09-22 00:53:01 | INFO | train rank=0 | step=36550/50000 progress=73.1% loss=0.318923 lr=1.851257e-05 lr_backbone=1.851257e-05 elapsed=42:26:25 avg_step=4.18s recent_step=4.15s eta=15:37:03 +2026-09-22 00:53:43 | INFO | train rank=0 | step=36560/50000 progress=73.1% loss=0.280584 lr=1.848689e-05 lr_backbone=1.848689e-05 elapsed=42:27:06 avg_step=4.18s recent_step=4.13s eta=15:36:21 +2026-09-22 00:54:24 | INFO | train rank=0 | step=36570/50000 progress=73.1% loss=0.336461 lr=1.846122e-05 lr_backbone=1.846122e-05 elapsed=42:27:47 avg_step=4.18s recent_step=4.13s eta=15:35:39 +2026-09-22 00:55:05 | INFO | train rank=0 | step=36580/50000 progress=73.2% loss=0.322896 lr=1.843557e-05 lr_backbone=1.843557e-05 elapsed=42:28:29 avg_step=4.18s recent_step=4.14s eta=15:34:57 +2026-09-22 00:55:46 | INFO | train rank=0 | step=36590/50000 progress=73.2% loss=0.302191 lr=1.840993e-05 lr_backbone=1.840993e-05 elapsed=42:29:09 avg_step=4.18s recent_step=4.08s eta=15:34:15 +2026-09-22 00:56:27 | INFO | train rank=0 | step=36600/50000 progress=73.2% loss=0.296890 lr=1.838430e-05 lr_backbone=1.838430e-05 elapsed=42:29:51 avg_step=4.18s recent_step=4.11s eta=15:33:33 +2026-09-22 00:57:08 | INFO | train rank=0 | step=36610/50000 progress=73.2% loss=0.319722 lr=1.835869e-05 lr_backbone=1.835869e-05 elapsed=42:30:32 avg_step=4.18s recent_step=4.12s eta=15:32:51 +2026-09-22 00:57:50 | INFO | train rank=0 | step=36620/50000 progress=73.2% loss=0.402915 lr=1.833309e-05 lr_backbone=1.833309e-05 elapsed=42:31:13 avg_step=4.18s recent_step=4.14s eta=15:32:09 +2026-09-22 00:58:31 | INFO | train rank=0 | step=36630/50000 progress=73.3% loss=0.332530 lr=1.830750e-05 lr_backbone=1.830750e-05 elapsed=42:31:54 avg_step=4.18s recent_step=4.12s eta=15:31:27 +2026-09-22 00:59:12 | INFO | train rank=0 | step=36640/50000 progress=73.3% loss=0.372757 lr=1.828193e-05 lr_backbone=1.828193e-05 elapsed=42:32:36 avg_step=4.18s recent_step=4.12s eta=15:30:45 +2026-09-22 00:59:53 | INFO | train rank=0 | step=36650/50000 progress=73.3% loss=0.315788 lr=1.825638e-05 lr_backbone=1.825638e-05 elapsed=42:33:17 avg_step=4.18s recent_step=4.11s eta=15:30:03 +2026-09-22 01:00:34 | INFO | train rank=0 | step=36660/50000 progress=73.3% loss=0.360442 lr=1.823083e-05 lr_backbone=1.823083e-05 elapsed=42:33:58 avg_step=4.18s recent_step=4.10s eta=15:29:20 +2026-09-22 01:01:16 | INFO | train rank=0 | step=36670/50000 progress=73.3% loss=0.314267 lr=1.820530e-05 lr_backbone=1.820530e-05 elapsed=42:34:39 avg_step=4.18s recent_step=4.17s eta=15:28:39 +2026-09-22 01:01:58 | INFO | train rank=0 | step=36680/50000 progress=73.4% loss=0.286278 lr=1.817979e-05 lr_backbone=1.817979e-05 elapsed=42:35:21 avg_step=4.18s recent_step=4.15s eta=15:27:57 +2026-09-22 01:02:39 | INFO | train rank=0 | step=36690/50000 progress=73.4% loss=0.319898 lr=1.815429e-05 lr_backbone=1.815429e-05 elapsed=42:36:02 avg_step=4.18s recent_step=4.14s eta=15:27:15 +2026-09-22 01:03:21 | INFO | train rank=0 | step=36700/50000 progress=73.4% loss=0.317973 lr=1.812880e-05 lr_backbone=1.812880e-05 elapsed=42:36:44 avg_step=4.18s recent_step=4.16s eta=15:26:33 +2026-09-22 01:04:02 | INFO | train rank=0 | step=36710/50000 progress=73.4% loss=0.296914 lr=1.810333e-05 lr_backbone=1.810333e-05 elapsed=42:37:25 avg_step=4.18s recent_step=4.10s eta=15:25:51 +2026-09-22 01:04:43 | INFO | train rank=0 | step=36720/50000 progress=73.4% loss=0.327125 lr=1.807787e-05 lr_backbone=1.807787e-05 elapsed=42:38:06 avg_step=4.18s recent_step=4.13s eta=15:25:09 +2026-09-22 01:05:24 | INFO | train rank=0 | step=36730/50000 progress=73.5% loss=0.315521 lr=1.805242e-05 lr_backbone=1.805242e-05 elapsed=42:38:48 avg_step=4.18s recent_step=4.17s eta=15:24:27 +2026-09-22 01:06:06 | INFO | train rank=0 | step=36740/50000 progress=73.5% loss=0.410127 lr=1.802699e-05 lr_backbone=1.802699e-05 elapsed=42:39:29 avg_step=4.18s recent_step=4.15s eta=15:23:45 +2026-09-22 01:06:48 | INFO | train rank=0 | step=36750/50000 progress=73.5% loss=0.355794 lr=1.800157e-05 lr_backbone=1.800157e-05 elapsed=42:40:11 avg_step=4.18s recent_step=4.16s eta=15:23:03 +2026-09-22 01:07:29 | INFO | train rank=0 | step=36760/50000 progress=73.5% loss=0.349659 lr=1.797617e-05 lr_backbone=1.797617e-05 elapsed=42:40:52 avg_step=4.18s recent_step=4.16s eta=15:22:21 +2026-09-22 01:08:10 | INFO | train rank=0 | step=36770/50000 progress=73.5% loss=0.276321 lr=1.795078e-05 lr_backbone=1.795078e-05 elapsed=42:41:34 avg_step=4.18s recent_step=4.11s eta=15:21:39 +2026-09-22 01:08:51 | INFO | train rank=0 | step=36780/50000 progress=73.6% loss=0.333661 lr=1.792540e-05 lr_backbone=1.792540e-05 elapsed=42:42:15 avg_step=4.18s recent_step=4.12s eta=15:20:57 +2026-09-22 01:09:33 | INFO | train rank=0 | step=36790/50000 progress=73.6% loss=0.251838 lr=1.790004e-05 lr_backbone=1.790004e-05 elapsed=42:42:56 avg_step=4.18s recent_step=4.14s eta=15:20:15 +2026-09-22 01:10:14 | INFO | train rank=0 | step=36800/50000 progress=73.6% loss=0.328281 lr=1.787470e-05 lr_backbone=1.787470e-05 elapsed=42:43:38 avg_step=4.18s recent_step=4.14s eta=15:19:33 +2026-09-22 01:10:56 | INFO | train rank=0 | step=36810/50000 progress=73.6% loss=0.304018 lr=1.784936e-05 lr_backbone=1.784936e-05 elapsed=42:44:19 avg_step=4.18s recent_step=4.15s eta=15:18:51 +2026-09-22 01:11:37 | INFO | train rank=0 | step=36820/50000 progress=73.6% loss=0.320357 lr=1.782404e-05 lr_backbone=1.782404e-05 elapsed=42:45:01 avg_step=4.18s recent_step=4.15s eta=15:18:10 +2026-09-22 01:12:18 | INFO | train rank=0 | step=36830/50000 progress=73.7% loss=0.299546 lr=1.779874e-05 lr_backbone=1.779874e-05 elapsed=42:45:42 avg_step=4.18s recent_step=4.10s eta=15:17:27 +2026-09-22 01:13:00 | INFO | train rank=0 | step=36840/50000 progress=73.7% loss=0.371798 lr=1.777345e-05 lr_backbone=1.777345e-05 elapsed=42:46:23 avg_step=4.18s recent_step=4.13s eta=15:16:46 +2026-09-22 01:13:41 | INFO | train rank=0 | step=36850/50000 progress=73.7% loss=0.351964 lr=1.774817e-05 lr_backbone=1.774817e-05 elapsed=42:47:04 avg_step=4.18s recent_step=4.15s eta=15:16:04 +2026-09-22 01:14:22 | INFO | train rank=0 | step=36860/50000 progress=73.7% loss=0.286763 lr=1.772291e-05 lr_backbone=1.772291e-05 elapsed=42:47:46 avg_step=4.18s recent_step=4.14s eta=15:15:22 +2026-09-22 01:15:04 | INFO | train rank=0 | step=36870/50000 progress=73.7% loss=0.431811 lr=1.769766e-05 lr_backbone=1.769766e-05 elapsed=42:48:27 avg_step=4.18s recent_step=4.14s eta=15:14:40 +2026-09-22 01:15:45 | INFO | train rank=0 | step=36880/50000 progress=73.8% loss=0.297397 lr=1.767242e-05 lr_backbone=1.767242e-05 elapsed=42:49:09 avg_step=4.18s recent_step=4.14s eta=15:13:58 +2026-09-22 01:16:26 | INFO | train rank=0 | step=36890/50000 progress=73.8% loss=0.309111 lr=1.764720e-05 lr_backbone=1.764720e-05 elapsed=42:49:50 avg_step=4.18s recent_step=4.09s eta=15:13:16 +2026-09-22 01:17:07 | INFO | train rank=0 | step=36900/50000 progress=73.8% loss=0.329353 lr=1.762200e-05 lr_backbone=1.762200e-05 elapsed=42:50:31 avg_step=4.18s recent_step=4.12s eta=15:12:34 +2026-09-22 01:17:49 | INFO | train rank=0 | step=36910/50000 progress=73.8% loss=0.332199 lr=1.759680e-05 lr_backbone=1.759680e-05 elapsed=42:51:12 avg_step=4.18s recent_step=4.15s eta=15:11:52 +2026-09-22 01:18:30 | INFO | train rank=0 | step=36920/50000 progress=73.8% loss=0.348914 lr=1.757163e-05 lr_backbone=1.757163e-05 elapsed=42:51:54 avg_step=4.18s recent_step=4.13s eta=15:11:10 +2026-09-22 01:19:12 | INFO | train rank=0 | step=36930/50000 progress=73.9% loss=0.310524 lr=1.754646e-05 lr_backbone=1.754646e-05 elapsed=42:52:35 avg_step=4.18s recent_step=4.15s eta=15:10:28 +2026-09-22 01:19:53 | INFO | train rank=0 | step=36940/50000 progress=73.9% loss=0.288654 lr=1.752131e-05 lr_backbone=1.752131e-05 elapsed=42:53:17 avg_step=4.18s recent_step=4.16s eta=15:09:46 +2026-09-22 01:20:34 | INFO | train rank=0 | step=36950/50000 progress=73.9% loss=0.333852 lr=1.749618e-05 lr_backbone=1.749618e-05 elapsed=42:53:58 avg_step=4.18s recent_step=4.10s eta=15:09:04 +2026-09-22 01:21:16 | INFO | train rank=0 | step=36960/50000 progress=73.9% loss=0.317462 lr=1.747105e-05 lr_backbone=1.747105e-05 elapsed=42:54:39 avg_step=4.18s recent_step=4.14s eta=15:08:22 +2026-09-22 01:21:57 | INFO | train rank=0 | step=36970/50000 progress=73.9% loss=0.444299 lr=1.744595e-05 lr_backbone=1.744595e-05 elapsed=42:55:21 avg_step=4.18s recent_step=4.18s eta=15:07:40 +2026-09-22 01:22:39 | INFO | train rank=0 | step=36980/50000 progress=74.0% loss=0.401087 lr=1.742085e-05 lr_backbone=1.742085e-05 elapsed=42:56:02 avg_step=4.18s recent_step=4.17s eta=15:06:58 +2026-09-22 01:23:20 | INFO | train rank=0 | step=36990/50000 progress=74.0% loss=0.331043 lr=1.739578e-05 lr_backbone=1.739578e-05 elapsed=42:56:44 avg_step=4.18s recent_step=4.14s eta=15:06:16 +2026-09-22 01:24:02 | INFO | train rank=0 | step=37000/50000 progress=74.0% loss=0.317774 lr=1.737071e-05 lr_backbone=1.737071e-05 elapsed=42:57:25 avg_step=4.18s recent_step=4.14s eta=15:05:34 +2026-09-22 01:24:43 | INFO | train rank=0 | step=37010/50000 progress=74.0% loss=0.294480 lr=1.734566e-05 lr_backbone=1.734566e-05 elapsed=42:58:06 avg_step=4.18s recent_step=4.10s eta=15:04:52 +2026-09-22 01:25:24 | INFO | train rank=0 | step=37020/50000 progress=74.0% loss=0.364272 lr=1.732063e-05 lr_backbone=1.732063e-05 elapsed=42:58:47 avg_step=4.18s recent_step=4.13s eta=15:04:10 +2026-09-22 01:26:06 | INFO | train rank=0 | step=37030/50000 progress=74.1% loss=0.406746 lr=1.729560e-05 lr_backbone=1.729560e-05 elapsed=42:59:29 avg_step=4.18s recent_step=4.16s eta=15:03:29 +2026-09-22 01:26:47 | INFO | train rank=0 | step=37040/50000 progress=74.1% loss=0.324956 lr=1.727060e-05 lr_backbone=1.727060e-05 elapsed=43:00:11 avg_step=4.18s recent_step=4.15s eta=15:02:47 +2026-09-22 01:27:29 | INFO | train rank=0 | step=37050/50000 progress=74.1% loss=0.320750 lr=1.724560e-05 lr_backbone=1.724560e-05 elapsed=43:00:52 avg_step=4.18s recent_step=4.14s eta=15:02:05 +2026-09-22 01:28:10 | INFO | train rank=0 | step=37060/50000 progress=74.1% loss=0.397642 lr=1.722063e-05 lr_backbone=1.722063e-05 elapsed=43:01:34 avg_step=4.18s recent_step=4.16s eta=15:01:23 +2026-09-22 01:28:51 | INFO | train rank=0 | step=37070/50000 progress=74.1% loss=0.318021 lr=1.719566e-05 lr_backbone=1.719566e-05 elapsed=43:02:15 avg_step=4.18s recent_step=4.11s eta=15:00:41 +2026-09-22 01:29:32 | INFO | train rank=0 | step=37080/50000 progress=74.2% loss=0.322858 lr=1.717071e-05 lr_backbone=1.717071e-05 elapsed=43:02:56 avg_step=4.18s recent_step=4.12s eta=14:59:59 +2026-09-22 01:30:14 | INFO | train rank=0 | step=37090/50000 progress=74.2% loss=0.269001 lr=1.714578e-05 lr_backbone=1.714578e-05 elapsed=43:03:37 avg_step=4.18s recent_step=4.16s eta=14:59:17 +2026-09-22 01:30:55 | INFO | train rank=0 | step=37100/50000 progress=74.2% loss=0.288228 lr=1.712086e-05 lr_backbone=1.712086e-05 elapsed=43:04:19 avg_step=4.18s recent_step=4.14s eta=14:58:35 +2026-09-22 01:31:37 | INFO | train rank=0 | step=37110/50000 progress=74.2% loss=0.321096 lr=1.709595e-05 lr_backbone=1.709595e-05 elapsed=43:05:00 avg_step=4.18s recent_step=4.15s eta=14:57:53 +2026-09-22 01:32:18 | INFO | train rank=0 | step=37120/50000 progress=74.2% loss=0.310859 lr=1.707106e-05 lr_backbone=1.707106e-05 elapsed=43:05:42 avg_step=4.18s recent_step=4.14s eta=14:57:11 +2026-09-22 01:32:59 | INFO | train rank=0 | step=37130/50000 progress=74.3% loss=0.333364 lr=1.704618e-05 lr_backbone=1.704618e-05 elapsed=43:06:23 avg_step=4.18s recent_step=4.10s eta=14:56:29 +2026-09-22 01:33:41 | INFO | train rank=0 | step=37140/50000 progress=74.3% loss=0.311134 lr=1.702132e-05 lr_backbone=1.702132e-05 elapsed=43:07:04 avg_step=4.18s recent_step=4.15s eta=14:55:47 +2026-09-22 01:34:22 | INFO | train rank=0 | step=37150/50000 progress=74.3% loss=0.390298 lr=1.699647e-05 lr_backbone=1.699647e-05 elapsed=43:07:46 avg_step=4.18s recent_step=4.16s eta=14:55:05 +2026-09-22 01:35:04 | INFO | train rank=0 | step=37160/50000 progress=74.3% loss=0.337957 lr=1.697163e-05 lr_backbone=1.697163e-05 elapsed=43:08:27 avg_step=4.18s recent_step=4.17s eta=14:54:23 +2026-09-22 01:35:46 | INFO | train rank=0 | step=37170/50000 progress=74.3% loss=0.311571 lr=1.694681e-05 lr_backbone=1.694681e-05 elapsed=43:09:09 avg_step=4.18s recent_step=4.16s eta=14:53:42 +2026-09-22 01:36:27 | INFO | train rank=0 | step=37180/50000 progress=74.4% loss=0.309504 lr=1.692201e-05 lr_backbone=1.692201e-05 elapsed=43:09:50 avg_step=4.18s recent_step=4.12s eta=14:53:00 +2026-09-22 01:37:08 | INFO | train rank=0 | step=37190/50000 progress=74.4% loss=0.318575 lr=1.689721e-05 lr_backbone=1.689721e-05 elapsed=43:10:31 avg_step=4.18s recent_step=4.11s eta=14:52:18 +2026-09-22 01:37:49 | INFO | train rank=0 | step=37200/50000 progress=74.4% loss=0.305997 lr=1.687244e-05 lr_backbone=1.687244e-05 elapsed=43:11:13 avg_step=4.18s recent_step=4.13s eta=14:51:36 +2026-09-22 01:38:31 | INFO | train rank=0 | step=37210/50000 progress=74.4% loss=0.276554 lr=1.684768e-05 lr_backbone=1.684768e-05 elapsed=43:11:54 avg_step=4.18s recent_step=4.15s eta=14:50:54 +2026-09-22 01:39:12 | INFO | train rank=0 | step=37220/50000 progress=74.4% loss=0.316086 lr=1.682293e-05 lr_backbone=1.682293e-05 elapsed=43:12:36 avg_step=4.18s recent_step=4.15s eta=14:50:12 +2026-09-22 01:39:54 | INFO | train rank=0 | step=37230/50000 progress=74.5% loss=0.325755 lr=1.679819e-05 lr_backbone=1.679819e-05 elapsed=43:13:17 avg_step=4.18s recent_step=4.16s eta=14:49:30 +2026-09-22 01:40:35 | INFO | train rank=0 | step=37240/50000 progress=74.5% loss=0.310786 lr=1.677348e-05 lr_backbone=1.677348e-05 elapsed=43:13:59 avg_step=4.18s recent_step=4.14s eta=14:48:48 +2026-09-22 01:41:16 | INFO | train rank=0 | step=37250/50000 progress=74.5% loss=0.342093 lr=1.674877e-05 lr_backbone=1.674877e-05 elapsed=43:14:39 avg_step=4.18s recent_step=4.08s eta=14:48:06 +2026-09-22 01:41:57 | INFO | train rank=0 | step=37260/50000 progress=74.5% loss=0.274311 lr=1.672408e-05 lr_backbone=1.672408e-05 elapsed=43:15:21 avg_step=4.18s recent_step=4.12s eta=14:47:24 +2026-09-22 01:42:39 | INFO | train rank=0 | step=37270/50000 progress=74.5% loss=0.335564 lr=1.669941e-05 lr_backbone=1.669941e-05 elapsed=43:16:02 avg_step=4.18s recent_step=4.14s eta=14:46:42 +2026-09-22 01:43:20 | INFO | train rank=0 | step=37280/50000 progress=74.6% loss=0.298713 lr=1.667475e-05 lr_backbone=1.667475e-05 elapsed=43:16:43 avg_step=4.18s recent_step=4.10s eta=14:46:00 +2026-09-22 01:44:01 | INFO | train rank=0 | step=37290/50000 progress=74.6% loss=0.282472 lr=1.665010e-05 lr_backbone=1.665010e-05 elapsed=43:17:24 avg_step=4.18s recent_step=4.12s eta=14:45:18 +2026-09-22 01:44:42 | INFO | train rank=0 | step=37300/50000 progress=74.6% loss=0.285939 lr=1.662547e-05 lr_backbone=1.662547e-05 elapsed=43:18:05 avg_step=4.18s recent_step=4.13s eta=14:44:36 +2026-09-22 01:45:23 | INFO | train rank=0 | step=37310/50000 progress=74.6% loss=0.319169 lr=1.660085e-05 lr_backbone=1.660085e-05 elapsed=43:18:46 avg_step=4.18s recent_step=4.08s eta=14:43:54 +2026-09-22 01:46:04 | INFO | train rank=0 | step=37320/50000 progress=74.6% loss=0.294707 lr=1.657625e-05 lr_backbone=1.657625e-05 elapsed=43:19:27 avg_step=4.18s recent_step=4.10s eta=14:43:12 +2026-09-22 01:46:45 | INFO | train rank=0 | step=37330/50000 progress=74.7% loss=0.311209 lr=1.655166e-05 lr_backbone=1.655166e-05 elapsed=43:20:08 avg_step=4.18s recent_step=4.12s eta=14:42:30 +2026-09-22 01:47:26 | INFO | train rank=0 | step=37340/50000 progress=74.7% loss=0.302076 lr=1.652709e-05 lr_backbone=1.652709e-05 elapsed=43:20:50 avg_step=4.18s recent_step=4.13s eta=14:41:48 +2026-09-22 01:48:08 | INFO | train rank=0 | step=37350/50000 progress=74.7% loss=0.364557 lr=1.650253e-05 lr_backbone=1.650253e-05 elapsed=43:21:31 avg_step=4.18s recent_step=4.12s eta=14:41:06 +2026-09-22 01:48:49 | INFO | train rank=0 | step=37360/50000 progress=74.7% loss=0.302509 lr=1.647799e-05 lr_backbone=1.647799e-05 elapsed=43:22:12 avg_step=4.18s recent_step=4.12s eta=14:40:24 +2026-09-22 01:49:30 | INFO | train rank=0 | step=37370/50000 progress=74.7% loss=0.276796 lr=1.645346e-05 lr_backbone=1.645346e-05 elapsed=43:22:53 avg_step=4.18s recent_step=4.08s eta=14:39:42 +2026-09-22 01:50:11 | INFO | train rank=0 | step=37380/50000 progress=74.8% loss=0.307625 lr=1.642894e-05 lr_backbone=1.642894e-05 elapsed=43:23:34 avg_step=4.18s recent_step=4.11s eta=14:39:00 +2026-09-22 01:50:52 | INFO | train rank=0 | step=37390/50000 progress=74.8% loss=0.310515 lr=1.640444e-05 lr_backbone=1.640444e-05 elapsed=43:24:15 avg_step=4.18s recent_step=4.13s eta=14:38:18 +2026-09-22 01:51:33 | INFO | train rank=0 | step=37400/50000 progress=74.8% loss=0.336132 lr=1.637996e-05 lr_backbone=1.637996e-05 elapsed=43:24:57 avg_step=4.18s recent_step=4.14s eta=14:37:36 +2026-09-22 01:52:15 | INFO | train rank=0 | step=37410/50000 progress=74.8% loss=0.312158 lr=1.635549e-05 lr_backbone=1.635549e-05 elapsed=43:25:38 avg_step=4.18s recent_step=4.14s eta=14:36:54 +2026-09-22 01:52:56 | INFO | train rank=0 | step=37420/50000 progress=74.8% loss=0.308901 lr=1.633103e-05 lr_backbone=1.633103e-05 elapsed=43:26:19 avg_step=4.18s recent_step=4.13s eta=14:36:12 +2026-09-22 01:53:37 | INFO | train rank=0 | step=37430/50000 progress=74.9% loss=0.286613 lr=1.630659e-05 lr_backbone=1.630659e-05 elapsed=43:27:00 avg_step=4.18s recent_step=4.09s eta=14:35:30 +2026-09-22 01:54:18 | INFO | train rank=0 | step=37440/50000 progress=74.9% loss=0.281259 lr=1.628217e-05 lr_backbone=1.628217e-05 elapsed=43:27:41 avg_step=4.18s recent_step=4.11s eta=14:34:48 +2026-09-22 01:55:00 | INFO | train rank=0 | step=37450/50000 progress=74.9% loss=0.349772 lr=1.625776e-05 lr_backbone=1.625776e-05 elapsed=43:28:23 avg_step=4.18s recent_step=4.16s eta=14:34:06 +2026-09-22 01:55:41 | INFO | train rank=0 | step=37460/50000 progress=74.9% loss=0.341068 lr=1.623336e-05 lr_backbone=1.623336e-05 elapsed=43:29:04 avg_step=4.18s recent_step=4.15s eta=14:33:24 +2026-09-22 01:56:23 | INFO | train rank=0 | step=37470/50000 progress=74.9% loss=0.389672 lr=1.620898e-05 lr_backbone=1.620898e-05 elapsed=43:29:46 avg_step=4.18s recent_step=4.15s eta=14:32:42 +2026-09-22 01:57:04 | INFO | train rank=0 | step=37480/50000 progress=75.0% loss=0.362961 lr=1.618461e-05 lr_backbone=1.618461e-05 elapsed=43:30:27 avg_step=4.18s recent_step=4.15s eta=14:32:00 +2026-09-22 01:57:45 | INFO | train rank=0 | step=37490/50000 progress=75.0% loss=0.333869 lr=1.616026e-05 lr_backbone=1.616026e-05 elapsed=43:31:09 avg_step=4.18s recent_step=4.11s eta=14:31:18 +2026-09-22 01:58:26 | INFO | train rank=0 | step=37500/50000 progress=75.0% loss=0.356894 lr=1.613592e-05 lr_backbone=1.613592e-05 elapsed=43:31:50 avg_step=4.18s recent_step=4.12s eta=14:30:36 +2026-09-22 01:59:08 | INFO | train rank=0 | step=37510/50000 progress=75.0% loss=0.351423 lr=1.611160e-05 lr_backbone=1.611160e-05 elapsed=43:32:31 avg_step=4.18s recent_step=4.14s eta=14:29:54 +2026-09-22 01:59:49 | INFO | train rank=0 | step=37520/50000 progress=75.0% loss=0.340988 lr=1.608729e-05 lr_backbone=1.608729e-05 elapsed=43:33:13 avg_step=4.18s recent_step=4.15s eta=14:29:12 +2026-09-22 02:00:31 | INFO | train rank=0 | step=37530/50000 progress=75.1% loss=0.295669 lr=1.606300e-05 lr_backbone=1.606300e-05 elapsed=43:33:54 avg_step=4.18s recent_step=4.16s eta=14:28:31 +2026-09-22 02:01:12 | INFO | train rank=0 | step=37540/50000 progress=75.1% loss=0.346802 lr=1.603872e-05 lr_backbone=1.603872e-05 elapsed=43:34:36 avg_step=4.18s recent_step=4.15s eta=14:27:49 +2026-09-22 02:01:53 | INFO | train rank=0 | step=37550/50000 progress=75.1% loss=0.280189 lr=1.601446e-05 lr_backbone=1.601446e-05 elapsed=43:35:17 avg_step=4.18s recent_step=4.10s eta=14:27:07 +2026-09-22 02:02:35 | INFO | train rank=0 | step=37560/50000 progress=75.1% loss=0.383859 lr=1.599021e-05 lr_backbone=1.599021e-05 elapsed=43:35:58 avg_step=4.18s recent_step=4.14s eta=14:26:25 +2026-09-22 02:03:16 | INFO | train rank=0 | step=37570/50000 progress=75.1% loss=0.327188 lr=1.596598e-05 lr_backbone=1.596598e-05 elapsed=43:36:39 avg_step=4.18s recent_step=4.13s eta=14:25:43 +2026-09-22 02:03:58 | INFO | train rank=0 | step=37580/50000 progress=75.2% loss=0.354834 lr=1.594176e-05 lr_backbone=1.594176e-05 elapsed=43:37:21 avg_step=4.18s recent_step=4.17s eta=14:25:01 +2026-09-22 02:04:39 | INFO | train rank=0 | step=37590/50000 progress=75.2% loss=0.324710 lr=1.591755e-05 lr_backbone=1.591755e-05 elapsed=43:38:03 avg_step=4.18s recent_step=4.15s eta=14:24:19 +2026-09-22 02:05:21 | INFO | train rank=0 | step=37600/50000 progress=75.2% loss=0.348538 lr=1.589336e-05 lr_backbone=1.589336e-05 elapsed=43:38:44 avg_step=4.18s recent_step=4.17s eta=14:23:37 +2026-09-22 02:06:02 | INFO | train rank=0 | step=37610/50000 progress=75.2% loss=0.324361 lr=1.586919e-05 lr_backbone=1.586919e-05 elapsed=43:39:25 avg_step=4.18s recent_step=4.11s eta=14:22:55 +2026-09-22 02:06:43 | INFO | train rank=0 | step=37620/50000 progress=75.2% loss=0.277064 lr=1.584503e-05 lr_backbone=1.584503e-05 elapsed=43:40:07 avg_step=4.18s recent_step=4.13s eta=14:22:13 +2026-09-22 02:07:24 | INFO | train rank=0 | step=37630/50000 progress=75.3% loss=0.306421 lr=1.582089e-05 lr_backbone=1.582089e-05 elapsed=43:40:48 avg_step=4.18s recent_step=4.10s eta=14:21:31 +2026-09-22 02:08:06 | INFO | train rank=0 | step=37640/50000 progress=75.3% loss=0.378832 lr=1.579676e-05 lr_backbone=1.579676e-05 elapsed=43:41:29 avg_step=4.18s recent_step=4.16s eta=14:20:49 +2026-09-22 02:08:48 | INFO | train rank=0 | step=37650/50000 progress=75.3% loss=0.276424 lr=1.577264e-05 lr_backbone=1.577264e-05 elapsed=43:42:11 avg_step=4.18s recent_step=4.15s eta=14:20:08 +2026-09-22 02:09:29 | INFO | train rank=0 | step=37660/50000 progress=75.3% loss=0.327907 lr=1.574855e-05 lr_backbone=1.574855e-05 elapsed=43:42:52 avg_step=4.18s recent_step=4.13s eta=14:19:26 +2026-09-22 02:10:10 | INFO | train rank=0 | step=37670/50000 progress=75.3% loss=0.305527 lr=1.572446e-05 lr_backbone=1.572446e-05 elapsed=43:43:33 avg_step=4.18s recent_step=4.09s eta=14:18:43 +2026-09-22 02:10:51 | INFO | train rank=0 | step=37680/50000 progress=75.4% loss=0.330890 lr=1.570039e-05 lr_backbone=1.570039e-05 elapsed=43:44:14 avg_step=4.18s recent_step=4.13s eta=14:18:02 +2026-09-22 02:11:33 | INFO | train rank=0 | step=37690/50000 progress=75.4% loss=0.386147 lr=1.567634e-05 lr_backbone=1.567634e-05 elapsed=43:44:56 avg_step=4.18s recent_step=4.17s eta=14:17:20 +2026-09-22 02:12:14 | INFO | train rank=0 | step=37700/50000 progress=75.4% loss=0.298590 lr=1.565230e-05 lr_backbone=1.565230e-05 elapsed=43:45:37 avg_step=4.18s recent_step=4.13s eta=14:16:38 +2026-09-22 02:12:56 | INFO | train rank=0 | step=37710/50000 progress=75.4% loss=0.354968 lr=1.562828e-05 lr_backbone=1.562828e-05 elapsed=43:46:19 avg_step=4.18s recent_step=4.16s eta=14:15:56 +2026-09-22 02:13:37 | INFO | train rank=0 | step=37720/50000 progress=75.4% loss=0.298757 lr=1.560427e-05 lr_backbone=1.560427e-05 elapsed=43:47:01 avg_step=4.18s recent_step=4.17s eta=14:15:14 +2026-09-22 02:14:18 | INFO | train rank=0 | step=37730/50000 progress=75.5% loss=0.287498 lr=1.558027e-05 lr_backbone=1.558027e-05 elapsed=43:47:41 avg_step=4.18s recent_step=4.09s eta=14:14:32 +2026-09-22 02:14:59 | INFO | train rank=0 | step=37740/50000 progress=75.5% loss=0.312524 lr=1.555629e-05 lr_backbone=1.555629e-05 elapsed=43:48:23 avg_step=4.18s recent_step=4.13s eta=14:13:50 +2026-09-22 02:15:41 | INFO | train rank=0 | step=37750/50000 progress=75.5% loss=0.363518 lr=1.553233e-05 lr_backbone=1.553233e-05 elapsed=43:49:04 avg_step=4.18s recent_step=4.16s eta=14:13:08 +2026-09-22 02:16:22 | INFO | train rank=0 | step=37760/50000 progress=75.5% loss=0.317567 lr=1.550838e-05 lr_backbone=1.550838e-05 elapsed=43:49:46 avg_step=4.18s recent_step=4.13s eta=14:12:26 +2026-09-22 02:17:04 | INFO | train rank=0 | step=37770/50000 progress=75.5% loss=0.307388 lr=1.548445e-05 lr_backbone=1.548445e-05 elapsed=43:50:27 avg_step=4.18s recent_step=4.15s eta=14:11:44 +2026-09-22 02:17:46 | INFO | train rank=0 | step=37780/50000 progress=75.6% loss=0.389220 lr=1.546053e-05 lr_backbone=1.546053e-05 elapsed=43:51:09 avg_step=4.18s recent_step=4.19s eta=14:11:03 +2026-09-22 02:18:27 | INFO | train rank=0 | step=37790/50000 progress=75.6% loss=0.288776 lr=1.543663e-05 lr_backbone=1.543663e-05 elapsed=43:51:50 avg_step=4.18s recent_step=4.09s eta=14:10:21 +2026-09-22 02:19:08 | INFO | train rank=0 | step=37800/50000 progress=75.6% loss=0.351864 lr=1.541274e-05 lr_backbone=1.541274e-05 elapsed=43:52:31 avg_step=4.18s recent_step=4.13s eta=14:09:39 +2026-09-22 02:19:49 | INFO | train rank=0 | step=37810/50000 progress=75.6% loss=0.293420 lr=1.538886e-05 lr_backbone=1.538886e-05 elapsed=43:53:12 avg_step=4.18s recent_step=4.12s eta=14:08:57 +2026-09-22 02:20:30 | INFO | train rank=0 | step=37820/50000 progress=75.6% loss=0.295961 lr=1.536501e-05 lr_backbone=1.536501e-05 elapsed=43:53:54 avg_step=4.18s recent_step=4.12s eta=14:08:15 +2026-09-22 02:21:12 | INFO | train rank=0 | step=37830/50000 progress=75.7% loss=0.352922 lr=1.534116e-05 lr_backbone=1.534116e-05 elapsed=43:54:35 avg_step=4.18s recent_step=4.15s eta=14:07:33 +2026-09-22 02:21:53 | INFO | train rank=0 | step=37840/50000 progress=75.7% loss=0.292067 lr=1.531733e-05 lr_backbone=1.531733e-05 elapsed=43:55:17 avg_step=4.18s recent_step=4.14s eta=14:06:51 +2026-09-22 02:22:34 | INFO | train rank=0 | step=37850/50000 progress=75.7% loss=0.398594 lr=1.529352e-05 lr_backbone=1.529352e-05 elapsed=43:55:58 avg_step=4.18s recent_step=4.11s eta=14:06:09 +2026-09-22 02:23:16 | INFO | train rank=0 | step=37860/50000 progress=75.7% loss=0.307399 lr=1.526972e-05 lr_backbone=1.526972e-05 elapsed=43:56:39 avg_step=4.18s recent_step=4.14s eta=14:05:27 +2026-09-22 02:23:57 | INFO | train rank=0 | step=37870/50000 progress=75.7% loss=0.321631 lr=1.524594e-05 lr_backbone=1.524594e-05 elapsed=43:57:21 avg_step=4.18s recent_step=4.16s eta=14:04:45 +2026-09-22 02:24:39 | INFO | train rank=0 | step=37880/50000 progress=75.8% loss=0.315921 lr=1.522218e-05 lr_backbone=1.522218e-05 elapsed=43:58:02 avg_step=4.18s recent_step=4.16s eta=14:04:03 +2026-09-22 02:25:20 | INFO | train rank=0 | step=37890/50000 progress=75.8% loss=0.380323 lr=1.519842e-05 lr_backbone=1.519842e-05 elapsed=43:58:44 avg_step=4.18s recent_step=4.14s eta=14:03:21 +2026-09-22 02:26:02 | INFO | train rank=0 | step=37900/50000 progress=75.8% loss=0.326736 lr=1.517469e-05 lr_backbone=1.517469e-05 elapsed=43:59:25 avg_step=4.18s recent_step=4.14s eta=14:02:39 +2026-09-22 02:26:43 | INFO | train rank=0 | step=37910/50000 progress=75.8% loss=0.310975 lr=1.515097e-05 lr_backbone=1.515097e-05 elapsed=44:00:06 avg_step=4.18s recent_step=4.10s eta=14:01:57 +2026-09-22 02:27:24 | INFO | train rank=0 | step=37920/50000 progress=75.8% loss=0.378654 lr=1.512726e-05 lr_backbone=1.512726e-05 elapsed=44:00:47 avg_step=4.18s recent_step=4.13s eta=14:01:15 +2026-09-22 02:28:05 | INFO | train rank=0 | step=37930/50000 progress=75.9% loss=0.302284 lr=1.510357e-05 lr_backbone=1.510357e-05 elapsed=44:01:29 avg_step=4.18s recent_step=4.15s eta=14:00:34 +2026-09-22 02:28:47 | INFO | train rank=0 | step=37940/50000 progress=75.9% loss=0.314758 lr=1.507989e-05 lr_backbone=1.507989e-05 elapsed=44:02:10 avg_step=4.18s recent_step=4.15s eta=13:59:52 +2026-09-22 02:29:28 | INFO | train rank=0 | step=37950/50000 progress=75.9% loss=0.292820 lr=1.505623e-05 lr_backbone=1.505623e-05 elapsed=44:02:52 avg_step=4.18s recent_step=4.14s eta=13:59:10 +2026-09-22 02:30:10 | INFO | train rank=0 | step=37960/50000 progress=75.9% loss=0.364312 lr=1.503259e-05 lr_backbone=1.503259e-05 elapsed=44:03:33 avg_step=4.18s recent_step=4.16s eta=13:58:28 +2026-09-22 02:30:51 | INFO | train rank=0 | step=37970/50000 progress=75.9% loss=0.349222 lr=1.500896e-05 lr_backbone=1.500896e-05 elapsed=44:04:14 avg_step=4.18s recent_step=4.10s eta=13:57:46 +2026-09-22 02:31:32 | INFO | train rank=0 | step=37980/50000 progress=76.0% loss=0.308221 lr=1.498534e-05 lr_backbone=1.498534e-05 elapsed=44:04:56 avg_step=4.18s recent_step=4.14s eta=13:57:04 +2026-09-22 02:32:14 | INFO | train rank=0 | step=37990/50000 progress=76.0% loss=0.289647 lr=1.496174e-05 lr_backbone=1.496174e-05 elapsed=44:05:37 avg_step=4.18s recent_step=4.15s eta=13:56:22 +2026-09-22 02:32:55 | INFO | train rank=0 | step=38000/50000 progress=76.0% loss=0.302399 lr=1.493816e-05 lr_backbone=1.493816e-05 elapsed=44:06:19 avg_step=4.18s recent_step=4.15s eta=13:55:40 +2026-09-22 02:33:37 | INFO | train rank=0 | step=38010/50000 progress=76.0% loss=0.350381 lr=1.491459e-05 lr_backbone=1.491459e-05 elapsed=44:07:00 avg_step=4.18s recent_step=4.13s eta=13:54:58 +2026-09-22 02:34:18 | INFO | train rank=0 | step=38020/50000 progress=76.0% loss=0.341248 lr=1.489104e-05 lr_backbone=1.489104e-05 elapsed=44:07:42 avg_step=4.18s recent_step=4.15s eta=13:54:17 +2026-09-22 02:34:59 | INFO | train rank=0 | step=38030/50000 progress=76.1% loss=0.297470 lr=1.486750e-05 lr_backbone=1.486750e-05 elapsed=44:08:23 avg_step=4.18s recent_step=4.11s eta=13:53:35 +2026-09-22 02:35:41 | INFO | train rank=0 | step=38040/50000 progress=76.1% loss=0.301963 lr=1.484398e-05 lr_backbone=1.484398e-05 elapsed=44:09:04 avg_step=4.18s recent_step=4.13s eta=13:52:53 +2026-09-22 02:36:22 | INFO | train rank=0 | step=38050/50000 progress=76.1% loss=0.317575 lr=1.482047e-05 lr_backbone=1.482047e-05 elapsed=44:09:46 avg_step=4.18s recent_step=4.16s eta=13:52:11 +2026-09-22 02:37:04 | INFO | train rank=0 | step=38060/50000 progress=76.1% loss=0.303604 lr=1.479698e-05 lr_backbone=1.479698e-05 elapsed=44:10:27 avg_step=4.18s recent_step=4.16s eta=13:51:29 +2026-09-22 02:37:45 | INFO | train rank=0 | step=38070/50000 progress=76.1% loss=0.302266 lr=1.477350e-05 lr_backbone=1.477350e-05 elapsed=44:11:09 avg_step=4.18s recent_step=4.13s eta=13:50:47 +2026-09-22 02:38:27 | INFO | train rank=0 | step=38080/50000 progress=76.2% loss=0.338735 lr=1.475004e-05 lr_backbone=1.475004e-05 elapsed=44:11:50 avg_step=4.18s recent_step=4.15s eta=13:50:05 +2026-09-22 02:39:08 | INFO | train rank=0 | step=38090/50000 progress=76.2% loss=0.343898 lr=1.472660e-05 lr_backbone=1.472660e-05 elapsed=44:12:31 avg_step=4.18s recent_step=4.09s eta=13:49:23 +2026-09-22 02:39:49 | INFO | train rank=0 | step=38100/50000 progress=76.2% loss=0.323968 lr=1.470317e-05 lr_backbone=1.470317e-05 elapsed=44:13:12 avg_step=4.18s recent_step=4.14s eta=13:48:41 +2026-09-22 02:40:31 | INFO | train rank=0 | step=38110/50000 progress=76.2% loss=0.334186 lr=1.467975e-05 lr_backbone=1.467975e-05 elapsed=44:13:54 avg_step=4.18s recent_step=4.16s eta=13:47:59 +2026-09-22 02:41:12 | INFO | train rank=0 | step=38120/50000 progress=76.2% loss=0.366536 lr=1.465635e-05 lr_backbone=1.465635e-05 elapsed=44:14:36 avg_step=4.18s recent_step=4.17s eta=13:47:18 +2026-09-22 02:41:54 | INFO | train rank=0 | step=38130/50000 progress=76.3% loss=0.335282 lr=1.463297e-05 lr_backbone=1.463297e-05 elapsed=44:15:17 avg_step=4.18s recent_step=4.12s eta=13:46:36 +2026-09-22 02:42:35 | INFO | train rank=0 | step=38140/50000 progress=76.3% loss=0.358894 lr=1.460960e-05 lr_backbone=1.460960e-05 elapsed=44:15:58 avg_step=4.18s recent_step=4.16s eta=13:45:54 +2026-09-22 02:43:16 | INFO | train rank=0 | step=38150/50000 progress=76.3% loss=0.307661 lr=1.458625e-05 lr_backbone=1.458625e-05 elapsed=44:16:40 avg_step=4.18s recent_step=4.11s eta=13:45:12 +2026-09-22 02:43:57 | INFO | train rank=0 | step=38160/50000 progress=76.3% loss=0.385573 lr=1.456291e-05 lr_backbone=1.456291e-05 elapsed=44:17:21 avg_step=4.18s recent_step=4.13s eta=13:44:30 +2026-09-22 02:44:39 | INFO | train rank=0 | step=38170/50000 progress=76.3% loss=0.336310 lr=1.453959e-05 lr_backbone=1.453959e-05 elapsed=44:18:02 avg_step=4.18s recent_step=4.15s eta=13:43:48 +2026-09-22 02:45:21 | INFO | train rank=0 | step=38180/50000 progress=76.4% loss=0.296647 lr=1.451629e-05 lr_backbone=1.451629e-05 elapsed=44:18:44 avg_step=4.18s recent_step=4.17s eta=13:43:06 +2026-09-22 02:46:02 | INFO | train rank=0 | step=38190/50000 progress=76.4% loss=0.367722 lr=1.449299e-05 lr_backbone=1.449299e-05 elapsed=44:19:25 avg_step=4.18s recent_step=4.15s eta=13:42:24 +2026-09-22 02:46:43 | INFO | train rank=0 | step=38200/50000 progress=76.4% loss=0.315917 lr=1.446972e-05 lr_backbone=1.446972e-05 elapsed=44:20:07 avg_step=4.18s recent_step=4.14s eta=13:41:42 +2026-09-22 02:47:25 | INFO | train rank=0 | step=38210/50000 progress=76.4% loss=0.380529 lr=1.444646e-05 lr_backbone=1.444646e-05 elapsed=44:20:48 avg_step=4.18s recent_step=4.11s eta=13:41:00 +2026-09-22 02:48:06 | INFO | train rank=0 | step=38220/50000 progress=76.4% loss=0.288101 lr=1.442322e-05 lr_backbone=1.442322e-05 elapsed=44:21:29 avg_step=4.18s recent_step=4.13s eta=13:40:18 +2026-09-22 02:48:47 | INFO | train rank=0 | step=38230/50000 progress=76.5% loss=0.329955 lr=1.439999e-05 lr_backbone=1.439999e-05 elapsed=44:22:11 avg_step=4.18s recent_step=4.14s eta=13:39:36 +2026-09-22 02:49:29 | INFO | train rank=0 | step=38240/50000 progress=76.5% loss=0.336273 lr=1.437677e-05 lr_backbone=1.437677e-05 elapsed=44:22:52 avg_step=4.18s recent_step=4.16s eta=13:38:55 +2026-09-22 02:50:10 | INFO | train rank=0 | step=38250/50000 progress=76.5% loss=0.308725 lr=1.435358e-05 lr_backbone=1.435358e-05 elapsed=44:23:34 avg_step=4.18s recent_step=4.14s eta=13:38:13 +2026-09-22 02:50:52 | INFO | train rank=0 | step=38260/50000 progress=76.5% loss=0.334482 lr=1.433040e-05 lr_backbone=1.433040e-05 elapsed=44:24:15 avg_step=4.18s recent_step=4.13s eta=13:37:31 +2026-09-22 02:51:33 | INFO | train rank=0 | step=38270/50000 progress=76.5% loss=0.346414 lr=1.430723e-05 lr_backbone=1.430723e-05 elapsed=44:24:56 avg_step=4.18s recent_step=4.10s eta=13:36:49 +2026-09-22 02:52:14 | INFO | train rank=0 | step=38280/50000 progress=76.6% loss=0.363674 lr=1.428408e-05 lr_backbone=1.428408e-05 elapsed=44:25:37 avg_step=4.18s recent_step=4.13s eta=13:36:07 +2026-09-22 02:52:55 | INFO | train rank=0 | step=38290/50000 progress=76.6% loss=0.315030 lr=1.426094e-05 lr_backbone=1.426094e-05 elapsed=44:26:19 avg_step=4.18s recent_step=4.16s eta=13:35:25 +2026-09-22 02:53:37 | INFO | train rank=0 | step=38300/50000 progress=76.6% loss=0.337369 lr=1.423783e-05 lr_backbone=1.423783e-05 elapsed=44:27:00 avg_step=4.18s recent_step=4.12s eta=13:34:43 +2026-09-22 02:54:18 | INFO | train rank=0 | step=38310/50000 progress=76.6% loss=0.298892 lr=1.421472e-05 lr_backbone=1.421472e-05 elapsed=44:27:41 avg_step=4.18s recent_step=4.14s eta=13:34:01 +2026-09-22 02:55:00 | INFO | train rank=0 | step=38320/50000 progress=76.6% loss=0.341913 lr=1.419163e-05 lr_backbone=1.419163e-05 elapsed=44:28:23 avg_step=4.18s recent_step=4.15s eta=13:33:19 +2026-09-22 02:55:41 | INFO | train rank=0 | step=38330/50000 progress=76.7% loss=0.296732 lr=1.416856e-05 lr_backbone=1.416856e-05 elapsed=44:29:04 avg_step=4.18s recent_step=4.10s eta=13:32:37 +2026-09-22 02:56:21 | INFO | train rank=0 | step=38340/50000 progress=76.7% loss=0.297301 lr=1.414551e-05 lr_backbone=1.414551e-05 elapsed=44:29:45 avg_step=4.18s recent_step=4.10s eta=13:31:55 +2026-09-22 02:57:03 | INFO | train rank=0 | step=38350/50000 progress=76.7% loss=0.412013 lr=1.412246e-05 lr_backbone=1.412246e-05 elapsed=44:30:26 avg_step=4.18s recent_step=4.13s eta=13:31:13 +2026-09-22 02:57:44 | INFO | train rank=0 | step=38360/50000 progress=76.7% loss=0.395613 lr=1.409944e-05 lr_backbone=1.409944e-05 elapsed=44:31:07 avg_step=4.18s recent_step=4.12s eta=13:30:31 +2026-09-22 02:58:25 | INFO | train rank=0 | step=38370/50000 progress=76.7% loss=0.334778 lr=1.407643e-05 lr_backbone=1.407643e-05 elapsed=44:31:49 avg_step=4.18s recent_step=4.14s eta=13:29:49 +2026-09-22 02:59:07 | INFO | train rank=0 | step=38380/50000 progress=76.8% loss=0.376511 lr=1.405344e-05 lr_backbone=1.405344e-05 elapsed=44:32:30 avg_step=4.18s recent_step=4.13s eta=13:29:08 +2026-09-22 02:59:48 | INFO | train rank=0 | step=38390/50000 progress=76.8% loss=0.352783 lr=1.403046e-05 lr_backbone=1.403046e-05 elapsed=44:33:11 avg_step=4.18s recent_step=4.09s eta=13:28:25 +2026-09-22 03:00:29 | INFO | train rank=0 | step=38400/50000 progress=76.8% loss=0.310899 lr=1.400750e-05 lr_backbone=1.400750e-05 elapsed=44:33:52 avg_step=4.18s recent_step=4.10s eta=13:27:43 +2026-09-22 03:01:10 | INFO | train rank=0 | step=38410/50000 progress=76.8% loss=0.319145 lr=1.398455e-05 lr_backbone=1.398455e-05 elapsed=44:34:33 avg_step=4.18s recent_step=4.13s eta=13:27:02 +2026-09-22 03:01:51 | INFO | train rank=0 | step=38420/50000 progress=76.8% loss=0.387488 lr=1.396162e-05 lr_backbone=1.396162e-05 elapsed=44:35:15 avg_step=4.18s recent_step=4.16s eta=13:26:20 +2026-09-22 03:02:33 | INFO | train rank=0 | step=38430/50000 progress=76.9% loss=0.301032 lr=1.393870e-05 lr_backbone=1.393870e-05 elapsed=44:35:56 avg_step=4.18s recent_step=4.13s eta=13:25:38 +2026-09-22 03:03:14 | INFO | train rank=0 | step=38440/50000 progress=76.9% loss=0.365496 lr=1.391580e-05 lr_backbone=1.391580e-05 elapsed=44:36:37 avg_step=4.18s recent_step=4.12s eta=13:24:56 +2026-09-22 03:03:55 | INFO | train rank=0 | step=38450/50000 progress=76.9% loss=0.306317 lr=1.389292e-05 lr_backbone=1.389292e-05 elapsed=44:37:18 avg_step=4.18s recent_step=4.08s eta=13:24:14 +2026-09-22 03:04:36 | INFO | train rank=0 | step=38460/50000 progress=76.9% loss=0.360606 lr=1.387005e-05 lr_backbone=1.387005e-05 elapsed=44:37:59 avg_step=4.18s recent_step=4.09s eta=13:23:32 +2026-09-22 03:05:17 | INFO | train rank=0 | step=38470/50000 progress=76.9% loss=0.293987 lr=1.384720e-05 lr_backbone=1.384720e-05 elapsed=44:38:40 avg_step=4.18s recent_step=4.10s eta=13:22:50 +2026-09-22 03:05:58 | INFO | train rank=0 | step=38480/50000 progress=77.0% loss=0.322715 lr=1.382436e-05 lr_backbone=1.382436e-05 elapsed=44:39:21 avg_step=4.18s recent_step=4.13s eta=13:22:08 +2026-09-22 03:06:39 | INFO | train rank=0 | step=38490/50000 progress=77.0% loss=0.305318 lr=1.380154e-05 lr_backbone=1.380154e-05 elapsed=44:40:03 avg_step=4.18s recent_step=4.13s eta=13:21:26 +2026-09-22 03:07:21 | INFO | train rank=0 | step=38500/50000 progress=77.0% loss=0.354862 lr=1.377874e-05 lr_backbone=1.377874e-05 elapsed=44:40:44 avg_step=4.18s recent_step=4.15s eta=13:20:44 +2026-09-22 03:08:02 | INFO | train rank=0 | step=38510/50000 progress=77.0% loss=0.407213 lr=1.375595e-05 lr_backbone=1.375595e-05 elapsed=44:41:25 avg_step=4.18s recent_step=4.09s eta=13:20:02 +2026-09-22 03:08:43 | INFO | train rank=0 | step=38520/50000 progress=77.0% loss=0.380655 lr=1.373318e-05 lr_backbone=1.373318e-05 elapsed=44:42:06 avg_step=4.18s recent_step=4.11s eta=13:19:20 +2026-09-22 03:09:24 | INFO | train rank=0 | step=38530/50000 progress=77.1% loss=0.311803 lr=1.371042e-05 lr_backbone=1.371042e-05 elapsed=44:42:47 avg_step=4.18s recent_step=4.13s eta=13:18:38 +2026-09-22 03:10:05 | INFO | train rank=0 | step=38540/50000 progress=77.1% loss=0.342236 lr=1.368768e-05 lr_backbone=1.368768e-05 elapsed=44:43:29 avg_step=4.18s recent_step=4.13s eta=13:17:56 +2026-09-22 03:10:47 | INFO | train rank=0 | step=38550/50000 progress=77.1% loss=0.320235 lr=1.366496e-05 lr_backbone=1.366496e-05 elapsed=44:44:10 avg_step=4.18s recent_step=4.13s eta=13:17:14 +2026-09-22 03:11:28 | INFO | train rank=0 | step=38560/50000 progress=77.1% loss=0.300880 lr=1.364225e-05 lr_backbone=1.364225e-05 elapsed=44:44:51 avg_step=4.18s recent_step=4.12s eta=13:16:32 +2026-09-22 03:12:09 | INFO | train rank=0 | step=38570/50000 progress=77.1% loss=0.392236 lr=1.361955e-05 lr_backbone=1.361955e-05 elapsed=44:45:32 avg_step=4.18s recent_step=4.08s eta=13:15:50 +2026-09-22 03:12:50 | INFO | train rank=0 | step=38580/50000 progress=77.2% loss=0.430281 lr=1.359688e-05 lr_backbone=1.359688e-05 elapsed=44:46:13 avg_step=4.18s recent_step=4.11s eta=13:15:08 +2026-09-22 03:13:31 | INFO | train rank=0 | step=38590/50000 progress=77.2% loss=0.367484 lr=1.357421e-05 lr_backbone=1.357421e-05 elapsed=44:46:55 avg_step=4.18s recent_step=4.16s eta=13:14:26 +2026-09-22 03:14:13 | INFO | train rank=0 | step=38600/50000 progress=77.2% loss=0.366272 lr=1.355157e-05 lr_backbone=1.355157e-05 elapsed=44:47:36 avg_step=4.18s recent_step=4.15s eta=13:13:45 +2026-09-22 03:14:54 | INFO | train rank=0 | step=38610/50000 progress=77.2% loss=0.283247 lr=1.352894e-05 lr_backbone=1.352894e-05 elapsed=44:48:17 avg_step=4.18s recent_step=4.13s eta=13:13:03 +2026-09-22 03:15:36 | INFO | train rank=0 | step=38620/50000 progress=77.2% loss=0.427052 lr=1.350633e-05 lr_backbone=1.350633e-05 elapsed=44:48:59 avg_step=4.18s recent_step=4.14s eta=13:12:21 +2026-09-22 03:16:16 | INFO | train rank=0 | step=38630/50000 progress=77.3% loss=0.343810 lr=1.348373e-05 lr_backbone=1.348373e-05 elapsed=44:49:40 avg_step=4.18s recent_step=4.09s eta=13:11:39 +2026-09-22 03:16:58 | INFO | train rank=0 | step=38640/50000 progress=77.3% loss=0.340766 lr=1.346115e-05 lr_backbone=1.346115e-05 elapsed=44:50:21 avg_step=4.18s recent_step=4.14s eta=13:10:57 +2026-09-22 03:17:39 | INFO | train rank=0 | step=38650/50000 progress=77.3% loss=0.302781 lr=1.343858e-05 lr_backbone=1.343858e-05 elapsed=44:51:03 avg_step=4.18s recent_step=4.13s eta=13:10:15 +2026-09-22 03:18:21 | INFO | train rank=0 | step=38660/50000 progress=77.3% loss=0.267721 lr=1.341603e-05 lr_backbone=1.341603e-05 elapsed=44:51:44 avg_step=4.18s recent_step=4.17s eta=13:09:33 +2026-09-22 03:19:02 | INFO | train rank=0 | step=38670/50000 progress=77.3% loss=0.320385 lr=1.339350e-05 lr_backbone=1.339350e-05 elapsed=44:52:26 avg_step=4.18s recent_step=4.14s eta=13:08:51 +2026-09-22 03:19:44 | INFO | train rank=0 | step=38680/50000 progress=77.4% loss=0.338831 lr=1.337098e-05 lr_backbone=1.337098e-05 elapsed=44:53:07 avg_step=4.18s recent_step=4.14s eta=13:08:09 +2026-09-22 03:20:25 | INFO | train rank=0 | step=38690/50000 progress=77.4% loss=0.330466 lr=1.334848e-05 lr_backbone=1.334848e-05 elapsed=44:53:48 avg_step=4.18s recent_step=4.11s eta=13:07:27 +2026-09-22 03:21:06 | INFO | train rank=0 | step=38700/50000 progress=77.4% loss=0.317039 lr=1.332599e-05 lr_backbone=1.332599e-05 elapsed=44:54:30 avg_step=4.18s recent_step=4.13s eta=13:06:45 +2026-09-22 03:21:47 | INFO | train rank=0 | step=38710/50000 progress=77.4% loss=0.385443 lr=1.330352e-05 lr_backbone=1.330352e-05 elapsed=44:55:11 avg_step=4.18s recent_step=4.13s eta=13:06:04 +2026-09-22 03:22:29 | INFO | train rank=0 | step=38720/50000 progress=77.4% loss=0.359166 lr=1.328107e-05 lr_backbone=1.328107e-05 elapsed=44:55:52 avg_step=4.18s recent_step=4.16s eta=13:05:22 +2026-09-22 03:23:11 | INFO | train rank=0 | step=38730/50000 progress=77.5% loss=0.270140 lr=1.325863e-05 lr_backbone=1.325863e-05 elapsed=44:56:34 avg_step=4.18s recent_step=4.16s eta=13:04:40 +2026-09-22 03:23:52 | INFO | train rank=0 | step=38740/50000 progress=77.5% loss=0.423528 lr=1.323621e-05 lr_backbone=1.323621e-05 elapsed=44:57:16 avg_step=4.18s recent_step=4.17s eta=13:03:58 +2026-09-22 03:24:33 | INFO | train rank=0 | step=38750/50000 progress=77.5% loss=0.310782 lr=1.321380e-05 lr_backbone=1.321380e-05 elapsed=44:57:57 avg_step=4.18s recent_step=4.10s eta=13:03:16 +2026-09-22 03:25:15 | INFO | train rank=0 | step=38760/50000 progress=77.5% loss=0.303699 lr=1.319142e-05 lr_backbone=1.319142e-05 elapsed=44:58:38 avg_step=4.18s recent_step=4.13s eta=13:02:34 +2026-09-22 03:25:56 | INFO | train rank=0 | step=38770/50000 progress=77.5% loss=0.350643 lr=1.316904e-05 lr_backbone=1.316904e-05 elapsed=44:59:19 avg_step=4.18s recent_step=4.12s eta=13:01:52 +2026-09-22 03:26:37 | INFO | train rank=0 | step=38780/50000 progress=77.6% loss=0.325004 lr=1.314669e-05 lr_backbone=1.314669e-05 elapsed=45:00:00 avg_step=4.18s recent_step=4.12s eta=13:01:10 +2026-09-22 03:27:19 | INFO | train rank=0 | step=38790/50000 progress=77.6% loss=0.309041 lr=1.312434e-05 lr_backbone=1.312434e-05 elapsed=45:00:42 avg_step=4.18s recent_step=4.16s eta=13:00:29 +2026-09-22 03:28:00 | INFO | train rank=0 | step=38800/50000 progress=77.6% loss=0.323183 lr=1.310202e-05 lr_backbone=1.310202e-05 elapsed=45:01:24 avg_step=4.18s recent_step=4.16s eta=12:59:47 +2026-09-22 03:28:41 | INFO | train rank=0 | step=38810/50000 progress=77.6% loss=0.339387 lr=1.307971e-05 lr_backbone=1.307971e-05 elapsed=45:02:05 avg_step=4.18s recent_step=4.10s eta=12:59:05 +2026-09-22 03:29:23 | INFO | train rank=0 | step=38820/50000 progress=77.6% loss=0.318058 lr=1.305742e-05 lr_backbone=1.305742e-05 elapsed=45:02:46 avg_step=4.18s recent_step=4.14s eta=12:58:23 +2026-09-22 03:30:04 | INFO | train rank=0 | step=38830/50000 progress=77.7% loss=0.344768 lr=1.303514e-05 lr_backbone=1.303514e-05 elapsed=45:03:28 avg_step=4.18s recent_step=4.16s eta=12:57:41 +2026-09-22 03:30:45 | INFO | train rank=0 | step=38840/50000 progress=77.7% loss=0.261703 lr=1.301288e-05 lr_backbone=1.301288e-05 elapsed=45:04:09 avg_step=4.18s recent_step=4.12s eta=12:56:59 +2026-09-22 03:31:27 | INFO | train rank=0 | step=38850/50000 progress=77.7% loss=0.274644 lr=1.299064e-05 lr_backbone=1.299064e-05 elapsed=45:04:50 avg_step=4.18s recent_step=4.17s eta=12:56:17 +2026-09-22 03:32:09 | INFO | train rank=0 | step=38860/50000 progress=77.7% loss=0.313014 lr=1.296841e-05 lr_backbone=1.296841e-05 elapsed=45:05:32 avg_step=4.18s recent_step=4.15s eta=12:55:35 +2026-09-22 03:32:50 | INFO | train rank=0 | step=38870/50000 progress=77.7% loss=0.375858 lr=1.294620e-05 lr_backbone=1.294620e-05 elapsed=45:06:13 avg_step=4.18s recent_step=4.09s eta=12:54:53 +2026-09-22 03:33:31 | INFO | train rank=0 | step=38880/50000 progress=77.8% loss=0.341757 lr=1.292400e-05 lr_backbone=1.292400e-05 elapsed=45:06:54 avg_step=4.18s recent_step=4.14s eta=12:54:11 +2026-09-22 03:34:12 | INFO | train rank=0 | step=38890/50000 progress=77.8% loss=0.317586 lr=1.290182e-05 lr_backbone=1.290182e-05 elapsed=45:07:36 avg_step=4.18s recent_step=4.15s eta=12:53:30 +2026-09-22 03:34:54 | INFO | train rank=0 | step=38900/50000 progress=77.8% loss=0.295480 lr=1.287966e-05 lr_backbone=1.287966e-05 elapsed=45:08:17 avg_step=4.18s recent_step=4.16s eta=12:52:48 +2026-09-22 03:35:35 | INFO | train rank=0 | step=38910/50000 progress=77.8% loss=0.373440 lr=1.285751e-05 lr_backbone=1.285751e-05 elapsed=45:08:59 avg_step=4.18s recent_step=4.13s eta=12:52:06 +2026-09-22 03:36:17 | INFO | train rank=0 | step=38920/50000 progress=77.8% loss=0.270324 lr=1.283538e-05 lr_backbone=1.283538e-05 elapsed=45:09:40 avg_step=4.18s recent_step=4.15s eta=12:51:24 +2026-09-22 03:36:58 | INFO | train rank=0 | step=38930/50000 progress=77.9% loss=0.345676 lr=1.281327e-05 lr_backbone=1.281327e-05 elapsed=45:10:21 avg_step=4.18s recent_step=4.11s eta=12:50:42 +2026-09-22 03:37:39 | INFO | train rank=0 | step=38940/50000 progress=77.9% loss=0.346251 lr=1.279117e-05 lr_backbone=1.279117e-05 elapsed=45:11:03 avg_step=4.18s recent_step=4.14s eta=12:50:00 +2026-09-22 03:38:21 | INFO | train rank=0 | step=38950/50000 progress=77.9% loss=0.313868 lr=1.276909e-05 lr_backbone=1.276909e-05 elapsed=45:11:44 avg_step=4.18s recent_step=4.13s eta=12:49:18 +2026-09-22 03:39:02 | INFO | train rank=0 | step=38960/50000 progress=77.9% loss=0.321015 lr=1.274702e-05 lr_backbone=1.274702e-05 elapsed=45:12:26 avg_step=4.18s recent_step=4.16s eta=12:48:36 +2026-09-22 03:39:44 | INFO | train rank=0 | step=38970/50000 progress=77.9% loss=0.318495 lr=1.272498e-05 lr_backbone=1.272498e-05 elapsed=45:13:07 avg_step=4.18s recent_step=4.13s eta=12:47:55 +2026-09-22 03:40:25 | INFO | train rank=0 | step=38980/50000 progress=78.0% loss=0.353659 lr=1.270294e-05 lr_backbone=1.270294e-05 elapsed=45:13:49 avg_step=4.18s recent_step=4.17s eta=12:47:13 +2026-09-22 03:41:06 | INFO | train rank=0 | step=38990/50000 progress=78.0% loss=0.334901 lr=1.268093e-05 lr_backbone=1.268093e-05 elapsed=45:14:29 avg_step=4.18s recent_step=4.08s eta=12:46:31 +2026-09-22 03:41:47 | INFO | train rank=0 | step=39000/50000 progress=78.0% loss=0.315057 lr=1.265893e-05 lr_backbone=1.265893e-05 elapsed=45:15:11 avg_step=4.18s recent_step=4.15s eta=12:45:49 +2026-09-22 03:42:29 | INFO | train rank=0 | step=39010/50000 progress=78.0% loss=0.374539 lr=1.263694e-05 lr_backbone=1.263694e-05 elapsed=45:15:52 avg_step=4.18s recent_step=4.15s eta=12:45:07 +2026-09-22 03:43:11 | INFO | train rank=0 | step=39020/50000 progress=78.0% loss=0.295324 lr=1.261497e-05 lr_backbone=1.261497e-05 elapsed=45:16:34 avg_step=4.18s recent_step=4.16s eta=12:44:25 +2026-09-22 03:43:52 | INFO | train rank=0 | step=39030/50000 progress=78.1% loss=0.304156 lr=1.259302e-05 lr_backbone=1.259302e-05 elapsed=45:17:15 avg_step=4.18s recent_step=4.13s eta=12:43:43 +2026-09-22 03:44:33 | INFO | train rank=0 | step=39040/50000 progress=78.1% loss=0.286254 lr=1.257109e-05 lr_backbone=1.257109e-05 elapsed=45:17:57 avg_step=4.18s recent_step=4.15s eta=12:43:01 +2026-09-22 03:45:14 | INFO | train rank=0 | step=39050/50000 progress=78.1% loss=0.374659 lr=1.254917e-05 lr_backbone=1.254917e-05 elapsed=45:18:38 avg_step=4.18s recent_step=4.09s eta=12:42:19 +2026-09-22 03:45:56 | INFO | train rank=0 | step=39060/50000 progress=78.1% loss=0.293565 lr=1.252727e-05 lr_backbone=1.252727e-05 elapsed=45:19:19 avg_step=4.18s recent_step=4.13s eta=12:41:37 +2026-09-22 03:46:37 | INFO | train rank=0 | step=39070/50000 progress=78.1% loss=0.293647 lr=1.250538e-05 lr_backbone=1.250538e-05 elapsed=45:20:00 avg_step=4.18s recent_step=4.14s eta=12:40:56 +2026-09-22 03:47:18 | INFO | train rank=0 | step=39080/50000 progress=78.2% loss=0.400107 lr=1.248351e-05 lr_backbone=1.248351e-05 elapsed=45:20:42 avg_step=4.18s recent_step=4.12s eta=12:40:14 +2026-09-22 03:48:00 | INFO | train rank=0 | step=39090/50000 progress=78.2% loss=0.345205 lr=1.246166e-05 lr_backbone=1.246166e-05 elapsed=45:21:23 avg_step=4.18s recent_step=4.14s eta=12:39:32 +2026-09-22 03:48:41 | INFO | train rank=0 | step=39100/50000 progress=78.2% loss=0.368371 lr=1.243982e-05 lr_backbone=1.243982e-05 elapsed=45:22:04 avg_step=4.18s recent_step=4.12s eta=12:38:50 +2026-09-22 03:49:22 | INFO | train rank=0 | step=39110/50000 progress=78.2% loss=0.319180 lr=1.241800e-05 lr_backbone=1.241800e-05 elapsed=45:22:45 avg_step=4.18s recent_step=4.10s eta=12:38:08 +2026-09-22 03:50:03 | INFO | train rank=0 | step=39120/50000 progress=78.2% loss=0.302676 lr=1.239620e-05 lr_backbone=1.239620e-05 elapsed=45:23:27 avg_step=4.18s recent_step=4.13s eta=12:37:26 +2026-09-22 03:50:45 | INFO | train rank=0 | step=39130/50000 progress=78.3% loss=0.377206 lr=1.237441e-05 lr_backbone=1.237441e-05 elapsed=45:24:08 avg_step=4.18s recent_step=4.15s eta=12:36:44 +2026-09-22 03:51:26 | INFO | train rank=0 | step=39140/50000 progress=78.3% loss=0.331022 lr=1.235264e-05 lr_backbone=1.235264e-05 elapsed=45:24:50 avg_step=4.18s recent_step=4.15s eta=12:36:02 +2026-09-22 03:52:08 | INFO | train rank=0 | step=39150/50000 progress=78.3% loss=0.349932 lr=1.233089e-05 lr_backbone=1.233089e-05 elapsed=45:25:31 avg_step=4.18s recent_step=4.15s eta=12:35:21 +2026-09-22 03:52:49 | INFO | train rank=0 | step=39160/50000 progress=78.3% loss=0.372239 lr=1.230915e-05 lr_backbone=1.230915e-05 elapsed=45:26:13 avg_step=4.18s recent_step=4.16s eta=12:34:39 +2026-09-22 03:53:30 | INFO | train rank=0 | step=39170/50000 progress=78.3% loss=0.281724 lr=1.228743e-05 lr_backbone=1.228743e-05 elapsed=45:26:54 avg_step=4.18s recent_step=4.11s eta=12:33:57 +2026-09-22 03:54:12 | INFO | train rank=0 | step=39180/50000 progress=78.4% loss=0.352292 lr=1.226573e-05 lr_backbone=1.226573e-05 elapsed=45:27:35 avg_step=4.18s recent_step=4.12s eta=12:33:15 +2026-09-22 03:54:53 | INFO | train rank=0 | step=39190/50000 progress=78.4% loss=0.349851 lr=1.224404e-05 lr_backbone=1.224404e-05 elapsed=45:28:17 avg_step=4.18s recent_step=4.15s eta=12:32:33 +2026-09-22 03:55:35 | INFO | train rank=0 | step=39200/50000 progress=78.4% loss=0.287783 lr=1.222237e-05 lr_backbone=1.222237e-05 elapsed=45:28:58 avg_step=4.18s recent_step=4.16s eta=12:31:51 +2026-09-22 03:56:16 | INFO | train rank=0 | step=39210/50000 progress=78.4% loss=0.314041 lr=1.220071e-05 lr_backbone=1.220071e-05 elapsed=45:29:40 avg_step=4.18s recent_step=4.13s eta=12:31:09 +2026-09-22 03:56:58 | INFO | train rank=0 | step=39220/50000 progress=78.4% loss=0.295269 lr=1.217907e-05 lr_backbone=1.217907e-05 elapsed=45:30:21 avg_step=4.18s recent_step=4.15s eta=12:30:27 +2026-09-22 03:57:39 | INFO | train rank=0 | step=39230/50000 progress=78.5% loss=0.308734 lr=1.215745e-05 lr_backbone=1.215745e-05 elapsed=45:31:02 avg_step=4.18s recent_step=4.11s eta=12:29:45 +2026-09-22 03:58:20 | INFO | train rank=0 | step=39240/50000 progress=78.5% loss=0.329840 lr=1.213585e-05 lr_backbone=1.213585e-05 elapsed=45:31:43 avg_step=4.18s recent_step=4.13s eta=12:29:04 +2026-09-22 03:59:02 | INFO | train rank=0 | step=39250/50000 progress=78.5% loss=0.307755 lr=1.211426e-05 lr_backbone=1.211426e-05 elapsed=45:32:25 avg_step=4.18s recent_step=4.16s eta=12:28:22 +2026-09-22 03:59:43 | INFO | train rank=0 | step=39260/50000 progress=78.5% loss=0.307367 lr=1.209268e-05 lr_backbone=1.209268e-05 elapsed=45:33:06 avg_step=4.18s recent_step=4.16s eta=12:27:40 +2026-09-22 04:00:24 | INFO | train rank=0 | step=39270/50000 progress=78.5% loss=0.392721 lr=1.207113e-05 lr_backbone=1.207113e-05 elapsed=45:33:48 avg_step=4.18s recent_step=4.14s eta=12:26:58 +2026-09-22 04:01:06 | INFO | train rank=0 | step=39280/50000 progress=78.6% loss=0.291303 lr=1.204959e-05 lr_backbone=1.204959e-05 elapsed=45:34:29 avg_step=4.18s recent_step=4.11s eta=12:26:16 +2026-09-22 04:01:47 | INFO | train rank=0 | step=39290/50000 progress=78.6% loss=0.288560 lr=1.202807e-05 lr_backbone=1.202807e-05 elapsed=45:35:10 avg_step=4.18s recent_step=4.11s eta=12:25:34 +2026-09-22 04:02:28 | INFO | train rank=0 | step=39300/50000 progress=78.6% loss=0.293805 lr=1.200656e-05 lr_backbone=1.200656e-05 elapsed=45:35:51 avg_step=4.18s recent_step=4.13s eta=12:24:52 +2026-09-22 04:03:09 | INFO | train rank=0 | step=39310/50000 progress=78.6% loss=0.364130 lr=1.198507e-05 lr_backbone=1.198507e-05 elapsed=45:36:33 avg_step=4.18s recent_step=4.15s eta=12:24:10 +2026-09-22 04:03:51 | INFO | train rank=0 | step=39320/50000 progress=78.6% loss=0.350267 lr=1.196360e-05 lr_backbone=1.196360e-05 elapsed=45:37:14 avg_step=4.18s recent_step=4.15s eta=12:23:29 +2026-09-22 04:04:33 | INFO | train rank=0 | step=39330/50000 progress=78.7% loss=0.348426 lr=1.194214e-05 lr_backbone=1.194214e-05 elapsed=45:37:56 avg_step=4.18s recent_step=4.16s eta=12:22:47 +2026-09-22 04:05:14 | INFO | train rank=0 | step=39340/50000 progress=78.7% loss=0.398363 lr=1.192070e-05 lr_backbone=1.192070e-05 elapsed=45:38:37 avg_step=4.18s recent_step=4.12s eta=12:22:05 +2026-09-22 04:05:55 | INFO | train rank=0 | step=39350/50000 progress=78.7% loss=0.339330 lr=1.189928e-05 lr_backbone=1.189928e-05 elapsed=45:39:18 avg_step=4.18s recent_step=4.10s eta=12:21:23 +2026-09-22 04:06:36 | INFO | train rank=0 | step=39360/50000 progress=78.7% loss=0.388692 lr=1.187787e-05 lr_backbone=1.187787e-05 elapsed=45:39:59 avg_step=4.18s recent_step=4.11s eta=12:20:41 +2026-09-22 04:07:18 | INFO | train rank=0 | step=39370/50000 progress=78.7% loss=0.343148 lr=1.185649e-05 lr_backbone=1.185649e-05 elapsed=45:40:41 avg_step=4.18s recent_step=4.17s eta=12:19:59 +2026-09-22 04:07:59 | INFO | train rank=0 | step=39380/50000 progress=78.8% loss=0.315260 lr=1.183511e-05 lr_backbone=1.183511e-05 elapsed=45:41:23 avg_step=4.18s recent_step=4.16s eta=12:19:17 +2026-09-22 04:08:41 | INFO | train rank=0 | step=39390/50000 progress=78.8% loss=0.325038 lr=1.181376e-05 lr_backbone=1.181376e-05 elapsed=45:42:04 avg_step=4.18s recent_step=4.16s eta=12:18:35 +2026-09-22 04:09:22 | INFO | train rank=0 | step=39400/50000 progress=78.8% loss=0.324246 lr=1.179242e-05 lr_backbone=1.179242e-05 elapsed=45:42:45 avg_step=4.18s recent_step=4.13s eta=12:17:54 +2026-09-22 04:10:03 | INFO | train rank=0 | step=39410/50000 progress=78.8% loss=0.327999 lr=1.177109e-05 lr_backbone=1.177109e-05 elapsed=45:43:26 avg_step=4.18s recent_step=4.11s eta=12:17:12 +2026-09-22 04:10:45 | INFO | train rank=0 | step=39420/50000 progress=78.8% loss=0.340181 lr=1.174979e-05 lr_backbone=1.174979e-05 elapsed=45:44:08 avg_step=4.18s recent_step=4.14s eta=12:16:30 +2026-09-22 04:11:26 | INFO | train rank=0 | step=39430/50000 progress=78.9% loss=0.368181 lr=1.172850e-05 lr_backbone=1.172850e-05 elapsed=45:44:49 avg_step=4.18s recent_step=4.14s eta=12:15:48 +2026-09-22 04:12:07 | INFO | train rank=0 | step=39440/50000 progress=78.9% loss=0.345930 lr=1.170723e-05 lr_backbone=1.170723e-05 elapsed=45:45:31 avg_step=4.18s recent_step=4.13s eta=12:15:06 +2026-09-22 04:12:49 | INFO | train rank=0 | step=39450/50000 progress=78.9% loss=0.414312 lr=1.168597e-05 lr_backbone=1.168597e-05 elapsed=45:46:12 avg_step=4.18s recent_step=4.16s eta=12:14:24 +2026-09-22 04:13:30 | INFO | train rank=0 | step=39460/50000 progress=78.9% loss=0.333296 lr=1.166473e-05 lr_backbone=1.166473e-05 elapsed=45:46:54 avg_step=4.18s recent_step=4.15s eta=12:13:42 +2026-09-22 04:14:11 | INFO | train rank=0 | step=39470/50000 progress=78.9% loss=0.356591 lr=1.164351e-05 lr_backbone=1.164351e-05 elapsed=45:47:34 avg_step=4.18s recent_step=4.08s eta=12:13:00 +2026-09-22 04:14:52 | INFO | train rank=0 | step=39480/50000 progress=79.0% loss=0.335752 lr=1.162230e-05 lr_backbone=1.162230e-05 elapsed=45:48:16 avg_step=4.18s recent_step=4.10s eta=12:12:18 +2026-09-22 04:15:34 | INFO | train rank=0 | step=39490/50000 progress=79.0% loss=0.288982 lr=1.160112e-05 lr_backbone=1.160112e-05 elapsed=45:48:57 avg_step=4.18s recent_step=4.14s eta=12:11:36 +2026-09-22 04:16:15 | INFO | train rank=0 | step=39500/50000 progress=79.0% loss=0.348137 lr=1.157994e-05 lr_backbone=1.157994e-05 elapsed=45:49:38 avg_step=4.18s recent_step=4.14s eta=12:10:55 +2026-09-22 04:16:56 | INFO | train rank=0 | step=39510/50000 progress=79.0% loss=0.403949 lr=1.155879e-05 lr_backbone=1.155879e-05 elapsed=45:50:20 avg_step=4.18s recent_step=4.13s eta=12:10:13 +2026-09-22 04:17:38 | INFO | train rank=0 | step=39520/50000 progress=79.0% loss=0.318875 lr=1.153765e-05 lr_backbone=1.153765e-05 elapsed=45:51:01 avg_step=4.18s recent_step=4.15s eta=12:09:31 +2026-09-22 04:18:19 | INFO | train rank=0 | step=39530/50000 progress=79.1% loss=0.291122 lr=1.151653e-05 lr_backbone=1.151653e-05 elapsed=45:51:42 avg_step=4.18s recent_step=4.08s eta=12:08:49 +2026-09-22 04:19:00 | INFO | train rank=0 | step=39540/50000 progress=79.1% loss=0.426932 lr=1.149543e-05 lr_backbone=1.149543e-05 elapsed=45:52:23 avg_step=4.18s recent_step=4.11s eta=12:08:07 +2026-09-22 04:19:41 | INFO | train rank=0 | step=39550/50000 progress=79.1% loss=0.319816 lr=1.147434e-05 lr_backbone=1.147434e-05 elapsed=45:53:04 avg_step=4.18s recent_step=4.14s eta=12:07:25 +2026-09-22 04:20:22 | INFO | train rank=0 | step=39560/50000 progress=79.1% loss=0.315487 lr=1.145327e-05 lr_backbone=1.145327e-05 elapsed=45:53:46 avg_step=4.18s recent_step=4.12s eta=12:06:43 +2026-09-22 04:21:04 | INFO | train rank=0 | step=39570/50000 progress=79.1% loss=0.340543 lr=1.143221e-05 lr_backbone=1.143221e-05 elapsed=45:54:27 avg_step=4.18s recent_step=4.12s eta=12:06:01 +2026-09-22 04:21:45 | INFO | train rank=0 | step=39580/50000 progress=79.2% loss=0.340319 lr=1.141118e-05 lr_backbone=1.141118e-05 elapsed=45:55:08 avg_step=4.18s recent_step=4.13s eta=12:05:19 +2026-09-22 04:22:26 | INFO | train rank=0 | step=39590/50000 progress=79.2% loss=0.356911 lr=1.139016e-05 lr_backbone=1.139016e-05 elapsed=45:55:49 avg_step=4.18s recent_step=4.08s eta=12:04:37 +2026-09-22 04:23:07 | INFO | train rank=0 | step=39600/50000 progress=79.2% loss=0.307337 lr=1.136915e-05 lr_backbone=1.136915e-05 elapsed=45:56:30 avg_step=4.18s recent_step=4.11s eta=12:03:55 +2026-09-22 04:23:48 | INFO | train rank=0 | step=39610/50000 progress=79.2% loss=0.391128 lr=1.134817e-05 lr_backbone=1.134817e-05 elapsed=45:57:11 avg_step=4.18s recent_step=4.14s eta=12:03:14 +2026-09-22 04:24:29 | INFO | train rank=0 | step=39620/50000 progress=79.2% loss=0.315789 lr=1.132720e-05 lr_backbone=1.132720e-05 elapsed=45:57:53 avg_step=4.18s recent_step=4.14s eta=12:02:32 +2026-09-22 04:25:11 | INFO | train rank=0 | step=39630/50000 progress=79.3% loss=0.334441 lr=1.130624e-05 lr_backbone=1.130624e-05 elapsed=45:58:34 avg_step=4.18s recent_step=4.15s eta=12:01:50 +2026-09-22 04:25:52 | INFO | train rank=0 | step=39640/50000 progress=79.3% loss=0.315425 lr=1.128531e-05 lr_backbone=1.128531e-05 elapsed=45:59:16 avg_step=4.18s recent_step=4.13s eta=12:01:08 +2026-09-22 04:26:33 | INFO | train rank=0 | step=39650/50000 progress=79.3% loss=0.425671 lr=1.126439e-05 lr_backbone=1.126439e-05 elapsed=45:59:56 avg_step=4.18s recent_step=4.08s eta=12:00:26 +2026-09-22 04:27:14 | INFO | train rank=0 | step=39660/50000 progress=79.3% loss=0.293948 lr=1.124349e-05 lr_backbone=1.124349e-05 elapsed=46:00:38 avg_step=4.18s recent_step=4.12s eta=11:59:44 +2026-09-22 04:27:56 | INFO | train rank=0 | step=39670/50000 progress=79.3% loss=0.312986 lr=1.122260e-05 lr_backbone=1.122260e-05 elapsed=46:01:19 avg_step=4.18s recent_step=4.16s eta=11:59:02 +2026-09-22 04:28:37 | INFO | train rank=0 | step=39680/50000 progress=79.4% loss=0.378202 lr=1.120174e-05 lr_backbone=1.120174e-05 elapsed=46:02:01 avg_step=4.18s recent_step=4.17s eta=11:58:20 +2026-09-22 04:29:19 | INFO | train rank=0 | step=39690/50000 progress=79.4% loss=0.323895 lr=1.118088e-05 lr_backbone=1.118088e-05 elapsed=46:02:42 avg_step=4.18s recent_step=4.12s eta=11:57:38 +2026-09-22 04:30:00 | INFO | train rank=0 | step=39700/50000 progress=79.4% loss=0.366902 lr=1.116005e-05 lr_backbone=1.116005e-05 elapsed=46:03:23 avg_step=4.18s recent_step=4.14s eta=11:56:57 +2026-09-22 04:30:41 | INFO | train rank=0 | step=39710/50000 progress=79.4% loss=0.352549 lr=1.113923e-05 lr_backbone=1.113923e-05 elapsed=46:04:05 avg_step=4.18s recent_step=4.12s eta=11:56:15 +2026-09-22 04:31:23 | INFO | train rank=0 | step=39720/50000 progress=79.4% loss=0.389546 lr=1.111843e-05 lr_backbone=1.111843e-05 elapsed=46:04:46 avg_step=4.18s recent_step=4.13s eta=11:55:33 +2026-09-22 04:32:04 | INFO | train rank=0 | step=39730/50000 progress=79.5% loss=0.302847 lr=1.109765e-05 lr_backbone=1.109765e-05 elapsed=46:05:27 avg_step=4.18s recent_step=4.14s eta=11:54:51 +2026-09-22 04:32:45 | INFO | train rank=0 | step=39740/50000 progress=79.5% loss=0.337897 lr=1.107688e-05 lr_backbone=1.107688e-05 elapsed=46:06:09 avg_step=4.18s recent_step=4.15s eta=11:54:09 +2026-09-22 04:33:27 | INFO | train rank=0 | step=39750/50000 progress=79.5% loss=0.290061 lr=1.105614e-05 lr_backbone=1.105614e-05 elapsed=46:06:50 avg_step=4.18s recent_step=4.14s eta=11:53:27 +2026-09-22 04:34:08 | INFO | train rank=0 | step=39760/50000 progress=79.5% loss=0.349701 lr=1.103540e-05 lr_backbone=1.103540e-05 elapsed=46:07:32 avg_step=4.18s recent_step=4.15s eta=11:52:45 +2026-09-22 04:34:49 | INFO | train rank=0 | step=39770/50000 progress=79.5% loss=0.335988 lr=1.101469e-05 lr_backbone=1.101469e-05 elapsed=46:08:13 avg_step=4.18s recent_step=4.10s eta=11:52:03 +2026-09-22 04:35:31 | INFO | train rank=0 | step=39780/50000 progress=79.6% loss=0.320069 lr=1.099399e-05 lr_backbone=1.099399e-05 elapsed=46:08:54 avg_step=4.18s recent_step=4.14s eta=11:51:22 +2026-09-22 04:36:12 | INFO | train rank=0 | step=39790/50000 progress=79.6% loss=0.384276 lr=1.097331e-05 lr_backbone=1.097331e-05 elapsed=46:09:35 avg_step=4.18s recent_step=4.12s eta=11:50:40 +2026-09-22 04:36:53 | INFO | train rank=0 | step=39800/50000 progress=79.6% loss=0.265691 lr=1.095265e-05 lr_backbone=1.095265e-05 elapsed=46:10:17 avg_step=4.18s recent_step=4.15s eta=11:49:58 +2026-09-22 04:37:35 | INFO | train rank=0 | step=39810/50000 progress=79.6% loss=0.270173 lr=1.093200e-05 lr_backbone=1.093200e-05 elapsed=46:10:58 avg_step=4.18s recent_step=4.16s eta=11:49:16 +2026-09-22 04:38:17 | INFO | train rank=0 | step=39820/50000 progress=79.6% loss=0.360592 lr=1.091137e-05 lr_backbone=1.091137e-05 elapsed=46:11:40 avg_step=4.18s recent_step=4.17s eta=11:48:34 +2026-09-22 04:38:58 | INFO | train rank=0 | step=39830/50000 progress=79.7% loss=0.319776 lr=1.089076e-05 lr_backbone=1.089076e-05 elapsed=46:12:21 avg_step=4.18s recent_step=4.11s eta=11:47:52 +2026-09-22 04:39:39 | INFO | train rank=0 | step=39840/50000 progress=79.7% loss=0.291907 lr=1.087016e-05 lr_backbone=1.087016e-05 elapsed=46:13:02 avg_step=4.18s recent_step=4.13s eta=11:47:10 +2026-09-22 04:40:20 | INFO | train rank=0 | step=39850/50000 progress=79.7% loss=0.347489 lr=1.084959e-05 lr_backbone=1.084959e-05 elapsed=46:13:43 avg_step=4.18s recent_step=4.11s eta=11:46:29 +2026-09-22 04:41:02 | INFO | train rank=0 | step=39860/50000 progress=79.7% loss=0.322458 lr=1.082902e-05 lr_backbone=1.082902e-05 elapsed=46:14:25 avg_step=4.18s recent_step=4.15s eta=11:45:47 +2026-09-22 04:41:43 | INFO | train rank=0 | step=39870/50000 progress=79.7% loss=0.325317 lr=1.080848e-05 lr_backbone=1.080848e-05 elapsed=46:15:06 avg_step=4.18s recent_step=4.14s eta=11:45:05 +2026-09-22 04:42:25 | INFO | train rank=0 | step=39880/50000 progress=79.8% loss=0.255621 lr=1.078795e-05 lr_backbone=1.078795e-05 elapsed=46:15:48 avg_step=4.18s recent_step=4.16s eta=11:44:23 +2026-09-22 04:43:06 | INFO | train rank=0 | step=39890/50000 progress=79.8% loss=0.390179 lr=1.076744e-05 lr_backbone=1.076744e-05 elapsed=46:16:29 avg_step=4.18s recent_step=4.10s eta=11:43:41 +2026-09-22 04:43:47 | INFO | train rank=0 | step=39900/50000 progress=79.8% loss=0.327514 lr=1.074695e-05 lr_backbone=1.074695e-05 elapsed=46:17:10 avg_step=4.18s recent_step=4.14s eta=11:42:59 +2026-09-22 04:44:29 | INFO | train rank=0 | step=39910/50000 progress=79.8% loss=0.303406 lr=1.072648e-05 lr_backbone=1.072648e-05 elapsed=46:17:52 avg_step=4.18s recent_step=4.16s eta=11:42:17 +2026-09-22 04:45:10 | INFO | train rank=0 | step=39920/50000 progress=79.8% loss=0.328056 lr=1.070602e-05 lr_backbone=1.070602e-05 elapsed=46:18:34 avg_step=4.18s recent_step=4.16s eta=11:41:36 +2026-09-22 04:45:52 | INFO | train rank=0 | step=39930/50000 progress=79.9% loss=0.311290 lr=1.068558e-05 lr_backbone=1.068558e-05 elapsed=46:19:15 avg_step=4.18s recent_step=4.15s eta=11:40:54 +2026-09-22 04:46:33 | INFO | train rank=0 | step=39940/50000 progress=79.9% loss=0.301543 lr=1.066515e-05 lr_backbone=1.066515e-05 elapsed=46:19:57 avg_step=4.18s recent_step=4.15s eta=11:40:12 +2026-09-22 04:47:14 | INFO | train rank=0 | step=39950/50000 progress=79.9% loss=0.415975 lr=1.064475e-05 lr_backbone=1.064475e-05 elapsed=46:20:38 avg_step=4.18s recent_step=4.11s eta=11:39:30 +2026-09-22 04:47:56 | INFO | train rank=0 | step=39960/50000 progress=79.9% loss=0.302092 lr=1.062436e-05 lr_backbone=1.062436e-05 elapsed=46:21:19 avg_step=4.18s recent_step=4.12s eta=11:38:48 +2026-09-22 04:48:37 | INFO | train rank=0 | step=39970/50000 progress=79.9% loss=0.340198 lr=1.060399e-05 lr_backbone=1.060399e-05 elapsed=46:22:00 avg_step=4.18s recent_step=4.14s eta=11:38:06 +2026-09-22 04:49:18 | INFO | train rank=0 | step=39980/50000 progress=80.0% loss=0.303136 lr=1.058363e-05 lr_backbone=1.058363e-05 elapsed=46:22:42 avg_step=4.18s recent_step=4.15s eta=11:37:24 +2026-09-22 04:50:00 | INFO | train rank=0 | step=39990/50000 progress=80.0% loss=0.312327 lr=1.056329e-05 lr_backbone=1.056329e-05 elapsed=46:23:23 avg_step=4.18s recent_step=4.14s eta=11:36:43 +2026-09-22 04:50:42 | INFO | train rank=0 | step=40000/50000 progress=80.0% loss=0.375464 lr=1.054297e-05 lr_backbone=1.054297e-05 elapsed=46:24:05 avg_step=4.18s recent_step=4.18s eta=11:36:01 +2026-09-22 04:51:23 | INFO | train rank=0 | step=40010/50000 progress=80.0% loss=0.324076 lr=1.052267e-05 lr_backbone=1.052267e-05 elapsed=46:24:46 avg_step=4.18s recent_step=4.11s eta=11:35:19 +2026-09-22 04:52:04 | INFO | train rank=0 | step=40020/50000 progress=80.0% loss=0.302602 lr=1.050239e-05 lr_backbone=1.050239e-05 elapsed=46:25:27 avg_step=4.18s recent_step=4.14s eta=11:34:37 +2026-09-22 04:52:45 | INFO | train rank=0 | step=40030/50000 progress=80.1% loss=0.344990 lr=1.048212e-05 lr_backbone=1.048212e-05 elapsed=46:26:09 avg_step=4.18s recent_step=4.14s eta=11:33:55 +2026-09-22 04:53:27 | INFO | train rank=0 | step=40040/50000 progress=80.1% loss=0.346464 lr=1.046187e-05 lr_backbone=1.046187e-05 elapsed=46:26:50 avg_step=4.18s recent_step=4.13s eta=11:33:13 +2026-09-22 04:54:08 | INFO | train rank=0 | step=40050/50000 progress=80.1% loss=0.421055 lr=1.044163e-05 lr_backbone=1.044163e-05 elapsed=46:27:32 avg_step=4.18s recent_step=4.16s eta=11:32:32 +2026-09-22 04:54:50 | INFO | train rank=0 | step=40060/50000 progress=80.1% loss=0.290050 lr=1.042142e-05 lr_backbone=1.042142e-05 elapsed=46:28:13 avg_step=4.18s recent_step=4.15s eta=11:31:50 +2026-09-22 04:55:31 | INFO | train rank=0 | step=40070/50000 progress=80.1% loss=0.333682 lr=1.040122e-05 lr_backbone=1.040122e-05 elapsed=46:28:54 avg_step=4.18s recent_step=4.10s eta=11:31:08 +2026-09-22 04:56:12 | INFO | train rank=0 | step=40080/50000 progress=80.2% loss=0.360014 lr=1.038103e-05 lr_backbone=1.038103e-05 elapsed=46:29:35 avg_step=4.18s recent_step=4.11s eta=11:30:26 +2026-09-22 04:56:54 | INFO | train rank=0 | step=40090/50000 progress=80.2% loss=0.307121 lr=1.036087e-05 lr_backbone=1.036087e-05 elapsed=46:30:17 avg_step=4.18s recent_step=4.17s eta=11:29:44 +2026-09-22 04:57:35 | INFO | train rank=0 | step=40100/50000 progress=80.2% loss=0.338379 lr=1.034072e-05 lr_backbone=1.034072e-05 elapsed=46:30:58 avg_step=4.18s recent_step=4.12s eta=11:29:02 +2026-09-22 04:58:16 | INFO | train rank=0 | step=40110/50000 progress=80.2% loss=0.308214 lr=1.032059e-05 lr_backbone=1.032059e-05 elapsed=46:31:40 avg_step=4.18s recent_step=4.16s eta=11:28:20 +2026-09-22 04:58:58 | INFO | train rank=0 | step=40120/50000 progress=80.2% loss=0.316306 lr=1.030048e-05 lr_backbone=1.030048e-05 elapsed=46:32:21 avg_step=4.18s recent_step=4.16s eta=11:27:39 +2026-09-22 04:59:39 | INFO | train rank=0 | step=40130/50000 progress=80.3% loss=0.379260 lr=1.028038e-05 lr_backbone=1.028038e-05 elapsed=46:33:02 avg_step=4.18s recent_step=4.10s eta=11:26:57 +2026-09-22 05:00:20 | INFO | train rank=0 | step=40140/50000 progress=80.3% loss=0.319292 lr=1.026031e-05 lr_backbone=1.026031e-05 elapsed=46:33:44 avg_step=4.18s recent_step=4.13s eta=11:26:15 +2026-09-22 05:01:02 | INFO | train rank=0 | step=40150/50000 progress=80.3% loss=0.300916 lr=1.024025e-05 lr_backbone=1.024025e-05 elapsed=46:34:25 avg_step=4.18s recent_step=4.15s eta=11:25:33 +2026-09-22 05:01:43 | INFO | train rank=0 | step=40160/50000 progress=80.3% loss=0.285981 lr=1.022020e-05 lr_backbone=1.022020e-05 elapsed=46:35:07 avg_step=4.18s recent_step=4.15s eta=11:24:51 +2026-09-22 05:02:25 | INFO | train rank=0 | step=40170/50000 progress=80.3% loss=0.345293 lr=1.020018e-05 lr_backbone=1.020018e-05 elapsed=46:35:48 avg_step=4.18s recent_step=4.14s eta=11:24:09 +2026-09-22 05:03:06 | INFO | train rank=0 | step=40180/50000 progress=80.4% loss=0.294207 lr=1.018017e-05 lr_backbone=1.018017e-05 elapsed=46:36:29 avg_step=4.18s recent_step=4.11s eta=11:23:27 +2026-09-22 05:03:47 | INFO | train rank=0 | step=40190/50000 progress=80.4% loss=0.312721 lr=1.016018e-05 lr_backbone=1.016018e-05 elapsed=46:37:10 avg_step=4.18s recent_step=4.08s eta=11:22:45 +2026-09-22 05:04:28 | INFO | train rank=0 | step=40200/50000 progress=80.4% loss=0.322436 lr=1.014021e-05 lr_backbone=1.014021e-05 elapsed=46:37:51 avg_step=4.18s recent_step=4.11s eta=11:22:03 +2026-09-22 05:05:09 | INFO | train rank=0 | step=40210/50000 progress=80.4% loss=0.372955 lr=1.012025e-05 lr_backbone=1.012025e-05 elapsed=46:38:32 avg_step=4.18s recent_step=4.11s eta=11:21:21 +2026-09-22 05:05:50 | INFO | train rank=0 | step=40220/50000 progress=80.4% loss=0.368031 lr=1.010031e-05 lr_backbone=1.010031e-05 elapsed=46:39:13 avg_step=4.18s recent_step=4.10s eta=11:20:40 +2026-09-22 05:06:30 | INFO | train rank=0 | step=40230/50000 progress=80.5% loss=0.266167 lr=1.008039e-05 lr_backbone=1.008039e-05 elapsed=46:39:54 avg_step=4.18s recent_step=4.07s eta=11:19:58 +2026-09-22 05:07:12 | INFO | train rank=0 | step=40240/50000 progress=80.5% loss=0.302565 lr=1.006049e-05 lr_backbone=1.006049e-05 elapsed=46:40:35 avg_step=4.18s recent_step=4.12s eta=11:19:16 +2026-09-22 05:07:52 | INFO | train rank=0 | step=40250/50000 progress=80.5% loss=0.276600 lr=1.004060e-05 lr_backbone=1.004060e-05 elapsed=46:41:15 avg_step=4.18s recent_step=4.04s eta=11:18:34 +2026-09-22 05:08:33 | INFO | train rank=0 | step=40260/50000 progress=80.5% loss=0.311050 lr=1.002073e-05 lr_backbone=1.002073e-05 elapsed=46:41:56 avg_step=4.18s recent_step=4.07s eta=11:17:52 +2026-09-22 05:09:14 | INFO | train rank=0 | step=40270/50000 progress=80.5% loss=0.313100 lr=1.000088e-05 lr_backbone=1.000088e-05 elapsed=46:42:37 avg_step=4.18s recent_step=4.09s eta=11:17:10 +2026-09-22 05:09:54 | INFO | train rank=0 | step=40280/50000 progress=80.6% loss=0.278980 lr=9.981047e-06 lr_backbone=9.981047e-06 elapsed=46:43:18 avg_step=4.18s recent_step=4.08s eta=11:16:28 +2026-09-22 05:10:35 | INFO | train rank=0 | step=40290/50000 progress=80.6% loss=0.353044 lr=9.961231e-06 lr_backbone=9.961231e-06 elapsed=46:43:58 avg_step=4.18s recent_step=4.07s eta=11:15:46 +2026-09-22 05:11:16 | INFO | train rank=0 | step=40300/50000 progress=80.6% loss=0.316588 lr=9.941432e-06 lr_backbone=9.941432e-06 elapsed=46:44:39 avg_step=4.18s recent_step=4.09s eta=11:15:04 +2026-09-22 05:11:57 | INFO | train rank=0 | step=40310/50000 progress=80.6% loss=0.388577 lr=9.921651e-06 lr_backbone=9.921651e-06 elapsed=46:45:20 avg_step=4.18s recent_step=4.05s eta=11:14:22 +2026-09-22 05:12:37 | INFO | train rank=0 | step=40320/50000 progress=80.6% loss=0.362734 lr=9.901887e-06 lr_backbone=9.901887e-06 elapsed=46:46:01 avg_step=4.18s recent_step=4.08s eta=11:13:40 +2026-09-22 05:13:19 | INFO | train rank=0 | step=40330/50000 progress=80.7% loss=0.306246 lr=9.882141e-06 lr_backbone=9.882141e-06 elapsed=46:46:42 avg_step=4.18s recent_step=4.12s eta=11:12:58 +2026-09-22 05:14:00 | INFO | train rank=0 | step=40340/50000 progress=80.7% loss=0.276938 lr=9.862413e-06 lr_backbone=9.862413e-06 elapsed=46:47:23 avg_step=4.18s recent_step=4.10s eta=11:12:16 +2026-09-22 05:14:40 | INFO | train rank=0 | step=40350/50000 progress=80.7% loss=0.261442 lr=9.842702e-06 lr_backbone=9.842702e-06 elapsed=46:48:04 avg_step=4.18s recent_step=4.07s eta=11:11:34 +2026-09-22 05:15:21 | INFO | train rank=0 | step=40360/50000 progress=80.7% loss=0.331567 lr=9.823008e-06 lr_backbone=9.823008e-06 elapsed=46:48:44 avg_step=4.18s recent_step=4.09s eta=11:10:52 +2026-09-22 05:16:02 | INFO | train rank=0 | step=40370/50000 progress=80.7% loss=0.291072 lr=9.803332e-06 lr_backbone=9.803332e-06 elapsed=46:49:25 avg_step=4.18s recent_step=4.08s eta=11:10:10 +2026-09-22 05:16:43 | INFO | train rank=0 | step=40380/50000 progress=80.8% loss=0.348503 lr=9.783674e-06 lr_backbone=9.783674e-06 elapsed=46:50:06 avg_step=4.18s recent_step=4.10s eta=11:09:28 +2026-09-22 05:17:24 | INFO | train rank=0 | step=40390/50000 progress=80.8% loss=0.436114 lr=9.764034e-06 lr_backbone=9.764034e-06 elapsed=46:50:47 avg_step=4.18s recent_step=4.11s eta=11:08:46 +2026-09-22 05:18:06 | INFO | train rank=0 | step=40400/50000 progress=80.8% loss=0.326534 lr=9.744411e-06 lr_backbone=9.744411e-06 elapsed=46:51:30 avg_step=4.18s recent_step=4.22s eta=11:08:04 +2026-09-22 05:18:56 | INFO | train rank=0 | step=40410/50000 progress=80.8% loss=0.311006 lr=9.724805e-06 lr_backbone=9.724805e-06 elapsed=46:52:20 avg_step=4.18s recent_step=5.01s eta=11:07:24 +2026-09-22 05:19:46 | INFO | train rank=0 | step=40420/50000 progress=80.8% loss=0.366409 lr=9.705217e-06 lr_backbone=9.705217e-06 elapsed=46:53:09 avg_step=4.18s recent_step=4.96s eta=11:06:45 +2026-09-22 05:20:35 | INFO | train rank=0 | step=40430/50000 progress=80.9% loss=0.268801 lr=9.685647e-06 lr_backbone=9.685647e-06 elapsed=46:53:58 avg_step=4.18s recent_step=4.87s eta=11:06:04 +2026-09-22 05:21:28 | INFO | train rank=0 | step=40440/50000 progress=80.9% loss=0.379180 lr=9.666095e-06 lr_backbone=9.666095e-06 elapsed=46:54:52 avg_step=4.18s recent_step=5.37s eta=11:05:26 +2026-09-22 05:22:16 | INFO | train rank=0 | step=40450/50000 progress=80.9% loss=0.316835 lr=9.646560e-06 lr_backbone=9.646560e-06 elapsed=46:55:39 avg_step=4.18s recent_step=4.75s eta=11:04:45 +2026-09-22 05:23:08 | INFO | train rank=0 | step=40460/50000 progress=80.9% loss=0.316262 lr=9.627043e-06 lr_backbone=9.627043e-06 elapsed=46:56:31 avg_step=4.18s recent_step=5.20s eta=11:04:06 +2026-09-22 05:24:00 | INFO | train rank=0 | step=40470/50000 progress=80.9% loss=0.307443 lr=9.607543e-06 lr_backbone=9.607543e-06 elapsed=46:57:23 avg_step=4.18s recent_step=5.22s eta=11:03:26 +2026-09-22 05:24:52 | INFO | train rank=0 | step=40480/50000 progress=81.0% loss=0.300433 lr=9.588061e-06 lr_backbone=9.588061e-06 elapsed=46:58:16 avg_step=4.18s recent_step=5.23s eta=11:02:47 +2026-09-22 05:25:48 | INFO | train rank=0 | step=40490/50000 progress=81.0% loss=0.339319 lr=9.568597e-06 lr_backbone=9.568597e-06 elapsed=46:59:12 avg_step=4.18s recent_step=5.59s eta=11:02:09 +2026-09-22 05:26:38 | INFO | train rank=0 | step=40500/50000 progress=81.0% loss=0.301528 lr=9.549150e-06 lr_backbone=9.549150e-06 elapsed=47:00:01 avg_step=4.18s recent_step=4.98s eta=11:01:29 +2026-09-22 05:27:30 | INFO | train rank=0 | step=40510/50000 progress=81.0% loss=0.327190 lr=9.529721e-06 lr_backbone=9.529721e-06 elapsed=47:00:53 avg_step=4.18s recent_step=5.19s eta=11:00:49 +2026-09-22 05:28:23 | INFO | train rank=0 | step=40520/50000 progress=81.0% loss=0.367913 lr=9.510310e-06 lr_backbone=9.510310e-06 elapsed=47:01:46 avg_step=4.18s recent_step=5.30s eta=11:00:10 +2026-09-22 05:29:14 | INFO | train rank=0 | step=40530/50000 progress=81.1% loss=0.336955 lr=9.490917e-06 lr_backbone=9.490917e-06 elapsed=47:02:37 avg_step=4.18s recent_step=5.11s eta=10:59:31 +2026-09-22 05:29:59 | INFO | train rank=0 | step=40540/50000 progress=81.1% loss=0.321891 lr=9.471541e-06 lr_backbone=9.471541e-06 elapsed=47:03:22 avg_step=4.18s recent_step=4.48s eta=10:58:50 +2026-09-22 05:30:47 | INFO | train rank=0 | step=40550/50000 progress=81.1% loss=0.439086 lr=9.452183e-06 lr_backbone=9.452183e-06 elapsed=47:04:10 avg_step=4.18s recent_step=4.81s eta=10:58:09 +2026-09-22 05:31:34 | INFO | train rank=0 | step=40560/50000 progress=81.1% loss=0.347195 lr=9.432843e-06 lr_backbone=9.432843e-06 elapsed=47:04:57 avg_step=4.18s recent_step=4.70s eta=10:57:29 +2026-09-22 05:32:26 | INFO | train rank=0 | step=40570/50000 progress=81.1% loss=0.379313 lr=9.413520e-06 lr_backbone=9.413520e-06 elapsed=47:05:49 avg_step=4.18s recent_step=5.21s eta=10:56:49 +2026-09-22 05:33:10 | INFO | train rank=0 | step=40580/50000 progress=81.2% loss=0.338268 lr=9.394216e-06 lr_backbone=9.394216e-06 elapsed=47:06:34 avg_step=4.18s recent_step=4.43s eta=10:56:08 +2026-09-22 05:33:57 | INFO | train rank=0 | step=40590/50000 progress=81.2% loss=0.356517 lr=9.374929e-06 lr_backbone=9.374929e-06 elapsed=47:07:20 avg_step=4.18s recent_step=4.67s eta=10:55:27 +2026-09-22 05:34:50 | INFO | train rank=0 | step=40600/50000 progress=81.2% loss=0.299235 lr=9.355659e-06 lr_backbone=9.355659e-06 elapsed=47:08:13 avg_step=4.18s recent_step=5.27s eta=10:54:48 +2026-09-22 05:35:42 | INFO | train rank=0 | step=40610/50000 progress=81.2% loss=0.302732 lr=9.336408e-06 lr_backbone=9.336408e-06 elapsed=47:09:05 avg_step=4.18s recent_step=5.23s eta=10:54:09 +2026-09-22 05:36:36 | INFO | train rank=0 | step=40620/50000 progress=81.2% loss=0.340836 lr=9.317174e-06 lr_backbone=9.317174e-06 elapsed=47:09:59 avg_step=4.18s recent_step=5.37s eta=10:53:30 +2026-09-22 05:37:29 | INFO | train rank=0 | step=40630/50000 progress=81.3% loss=0.301540 lr=9.297958e-06 lr_backbone=9.297958e-06 elapsed=47:10:52 avg_step=4.18s recent_step=5.32s eta=10:52:51 +2026-09-22 05:38:22 | INFO | train rank=0 | step=40640/50000 progress=81.3% loss=0.271778 lr=9.278760e-06 lr_backbone=9.278760e-06 elapsed=47:11:45 avg_step=4.18s recent_step=5.32s eta=10:52:11 +2026-09-22 05:39:09 | INFO | train rank=0 | step=40650/50000 progress=81.3% loss=0.292601 lr=9.259580e-06 lr_backbone=9.259580e-06 elapsed=47:12:32 avg_step=4.18s recent_step=4.71s eta=10:51:31 +2026-09-22 05:39:57 | INFO | train rank=0 | step=40660/50000 progress=81.3% loss=0.321883 lr=9.240418e-06 lr_backbone=9.240418e-06 elapsed=47:13:21 avg_step=4.18s recent_step=4.83s eta=10:50:50 +2026-09-22 05:40:42 | INFO | train rank=0 | step=40670/50000 progress=81.3% loss=0.320299 lr=9.221273e-06 lr_backbone=9.221273e-06 elapsed=47:14:06 avg_step=4.18s recent_step=4.48s eta=10:50:09 +2026-09-22 05:41:35 | INFO | train rank=0 | step=40680/50000 progress=81.4% loss=0.399845 lr=9.202146e-06 lr_backbone=9.202146e-06 elapsed=47:14:58 avg_step=4.18s recent_step=5.24s eta=10:49:30 +2026-09-22 05:42:25 | INFO | train rank=0 | step=40690/50000 progress=81.4% loss=0.322628 lr=9.183037e-06 lr_backbone=9.183037e-06 elapsed=47:15:48 avg_step=4.18s recent_step=5.01s eta=10:48:50 +2026-09-22 05:43:17 | INFO | train rank=0 | step=40700/50000 progress=81.4% loss=0.358122 lr=9.163946e-06 lr_backbone=9.163946e-06 elapsed=47:16:40 avg_step=4.18s recent_step=5.17s eta=10:48:10 +2026-09-22 05:44:04 | INFO | train rank=0 | step=40710/50000 progress=81.4% loss=0.345189 lr=9.144873e-06 lr_backbone=9.144873e-06 elapsed=47:17:28 avg_step=4.18s recent_step=4.77s eta=10:47:30 +2026-09-22 05:44:45 | INFO | train rank=0 | step=40720/50000 progress=81.4% loss=0.323493 lr=9.125818e-06 lr_backbone=9.125818e-06 elapsed=47:18:08 avg_step=4.18s recent_step=4.06s eta=10:46:48 +2026-09-22 05:45:25 | INFO | train rank=0 | step=40730/50000 progress=81.5% loss=0.314501 lr=9.106780e-06 lr_backbone=9.106780e-06 elapsed=47:18:48 avg_step=4.18s recent_step=4.02s eta=10:46:06 +2026-09-22 05:46:06 | INFO | train rank=0 | step=40740/50000 progress=81.5% loss=0.385440 lr=9.087761e-06 lr_backbone=9.087761e-06 elapsed=47:19:29 avg_step=4.18s recent_step=4.07s eta=10:45:24 +2026-09-22 05:47:18 | INFO | train rank=0 | step=40750/50000 progress=81.5% loss=0.352190 lr=9.068759e-06 lr_backbone=9.068759e-06 elapsed=47:20:41 avg_step=4.18s recent_step=7.24s eta=10:44:49 +2026-09-22 05:48:39 | INFO | train rank=0 | step=40760/50000 progress=81.5% loss=0.346519 lr=9.049776e-06 lr_backbone=9.049776e-06 elapsed=47:22:03 avg_step=4.18s recent_step=8.12s eta=10:44:16 +2026-09-22 05:50:02 | INFO | train rank=0 | step=40770/50000 progress=81.5% loss=0.299524 lr=9.030810e-06 lr_backbone=9.030810e-06 elapsed=47:23:25 avg_step=4.18s recent_step=8.22s eta=10:43:43 +2026-09-22 05:51:30 | INFO | train rank=0 | step=40780/50000 progress=81.6% loss=0.354163 lr=9.011862e-06 lr_backbone=9.011862e-06 elapsed=47:24:54 avg_step=4.19s recent_step=8.87s eta=10:43:12 +2026-09-22 05:52:51 | INFO | train rank=0 | step=40790/50000 progress=81.6% loss=0.360444 lr=8.992932e-06 lr_backbone=8.992932e-06 elapsed=47:26:14 avg_step=4.19s recent_step=8.07s eta=10:42:39 +2026-09-22 05:54:12 | INFO | train rank=0 | step=40800/50000 progress=81.6% loss=0.343264 lr=8.974020e-06 lr_backbone=8.974020e-06 elapsed=47:27:36 avg_step=4.19s recent_step=8.15s eta=10:42:06 +2026-09-22 05:55:30 | INFO | train rank=0 | step=40810/50000 progress=81.6% loss=0.397634 lr=8.955126e-06 lr_backbone=8.955126e-06 elapsed=47:28:53 avg_step=4.19s recent_step=7.74s eta=10:41:32 +2026-09-22 05:58:58 | INFO | train rank=0 | step=40820/50000 progress=81.6% loss=0.303432 lr=8.936249e-06 lr_backbone=8.936249e-06 elapsed=47:32:21 avg_step=4.19s recent_step=20.80s eta=10:41:27 +2026-09-22 06:03:02 | INFO | train rank=0 | step=40830/50000 progress=81.7% loss=0.387884 lr=8.917391e-06 lr_backbone=8.917391e-06 elapsed=47:36:25 avg_step=4.20s recent_step=24.42s eta=10:41:31 +2026-09-22 06:03:44 | INFO | train rank=0 | step=40840/50000 progress=81.7% loss=0.339996 lr=8.898551e-06 lr_backbone=8.898551e-06 elapsed=47:37:07 avg_step=4.20s recent_step=4.19s eta=10:40:49 +2026-09-22 06:04:25 | INFO | train rank=0 | step=40850/50000 progress=81.7% loss=0.372728 lr=8.879729e-06 lr_backbone=8.879729e-06 elapsed=47:37:48 avg_step=4.20s recent_step=4.11s eta=10:40:07 +2026-09-22 06:05:06 | INFO | train rank=0 | step=40860/50000 progress=81.7% loss=0.307913 lr=8.860925e-06 lr_backbone=8.860925e-06 elapsed=47:38:30 avg_step=4.20s recent_step=4.15s eta=10:39:25 +2026-09-22 06:05:48 | INFO | train rank=0 | step=40870/50000 progress=81.7% loss=0.287899 lr=8.842138e-06 lr_backbone=8.842138e-06 elapsed=47:39:11 avg_step=4.20s recent_step=4.13s eta=10:38:43 +2026-09-22 06:06:29 | INFO | train rank=0 | step=40880/50000 progress=81.8% loss=0.333748 lr=8.823370e-06 lr_backbone=8.823370e-06 elapsed=47:39:53 avg_step=4.20s recent_step=4.16s eta=10:38:01 +2026-09-22 06:07:11 | INFO | train rank=0 | step=40890/50000 progress=81.8% loss=0.262108 lr=8.804620e-06 lr_backbone=8.804620e-06 elapsed=47:40:34 avg_step=4.20s recent_step=4.16s eta=10:37:18 +2026-09-22 06:07:53 | INFO | train rank=0 | step=40900/50000 progress=81.8% loss=0.361884 lr=8.785888e-06 lr_backbone=8.785888e-06 elapsed=47:41:16 avg_step=4.20s recent_step=4.18s eta=10:36:36 +2026-09-22 06:08:34 | INFO | train rank=0 | step=40910/50000 progress=81.8% loss=0.323099 lr=8.767173e-06 lr_backbone=8.767173e-06 elapsed=47:41:57 avg_step=4.20s recent_step=4.10s eta=10:35:54 +2026-09-22 06:09:15 | INFO | train rank=0 | step=40920/50000 progress=81.8% loss=0.390262 lr=8.748477e-06 lr_backbone=8.748477e-06 elapsed=47:42:38 avg_step=4.20s recent_step=4.14s eta=10:35:12 +2026-09-22 06:09:57 | INFO | train rank=0 | step=40930/50000 progress=81.9% loss=0.414578 lr=8.729799e-06 lr_backbone=8.729799e-06 elapsed=47:43:20 avg_step=4.20s recent_step=4.15s eta=10:34:30 +2026-09-22 06:10:38 | INFO | train rank=0 | step=40940/50000 progress=81.9% loss=0.405114 lr=8.711139e-06 lr_backbone=8.711139e-06 elapsed=47:44:02 avg_step=4.20s recent_step=4.18s eta=10:33:48 +2026-09-22 06:11:20 | INFO | train rank=0 | step=40950/50000 progress=81.9% loss=0.282203 lr=8.692497e-06 lr_backbone=8.692497e-06 elapsed=47:44:44 avg_step=4.20s recent_step=4.19s eta=10:33:06 +2026-09-22 06:12:02 | INFO | train rank=0 | step=40960/50000 progress=81.9% loss=0.301747 lr=8.673873e-06 lr_backbone=8.673873e-06 elapsed=47:45:25 avg_step=4.20s recent_step=4.17s eta=10:32:24 +2026-09-22 06:12:43 | INFO | train rank=0 | step=40970/50000 progress=81.9% loss=0.346271 lr=8.655267e-06 lr_backbone=8.655267e-06 elapsed=47:46:06 avg_step=4.20s recent_step=4.10s eta=10:31:42 +2026-09-22 06:13:24 | INFO | train rank=0 | step=40980/50000 progress=82.0% loss=0.325262 lr=8.636680e-06 lr_backbone=8.636680e-06 elapsed=47:46:48 avg_step=4.20s recent_step=4.14s eta=10:31:00 +2026-09-22 06:14:06 | INFO | train rank=0 | step=40990/50000 progress=82.0% loss=0.280887 lr=8.618110e-06 lr_backbone=8.618110e-06 elapsed=47:47:29 avg_step=4.20s recent_step=4.14s eta=10:30:18 +2026-09-22 06:14:47 | INFO | train rank=0 | step=41000/50000 progress=82.0% loss=0.333651 lr=8.599558e-06 lr_backbone=8.599558e-06 elapsed=47:48:11 avg_step=4.20s recent_step=4.17s eta=10:29:36 +2026-09-22 06:15:29 | INFO | train rank=0 | step=41010/50000 progress=82.0% loss=0.386778 lr=8.581025e-06 lr_backbone=8.581025e-06 elapsed=47:48:53 avg_step=4.20s recent_step=4.17s eta=10:28:54 +2026-09-22 06:16:11 | INFO | train rank=0 | step=41020/50000 progress=82.0% loss=0.327731 lr=8.562510e-06 lr_backbone=8.562510e-06 elapsed=47:49:34 avg_step=4.20s recent_step=4.17s eta=10:28:12 +2026-09-22 06:16:52 | INFO | train rank=0 | step=41030/50000 progress=82.1% loss=0.289078 lr=8.544012e-06 lr_backbone=8.544012e-06 elapsed=47:50:16 avg_step=4.20s recent_step=4.13s eta=10:27:29 +2026-09-22 06:17:33 | INFO | train rank=0 | step=41040/50000 progress=82.1% loss=0.295336 lr=8.525533e-06 lr_backbone=8.525533e-06 elapsed=47:50:57 avg_step=4.20s recent_step=4.13s eta=10:26:47 +2026-09-22 06:18:15 | INFO | train rank=0 | step=41050/50000 progress=82.1% loss=0.386896 lr=8.507072e-06 lr_backbone=8.507072e-06 elapsed=47:51:38 avg_step=4.20s recent_step=4.17s eta=10:26:05 +2026-09-22 06:18:57 | INFO | train rank=0 | step=41060/50000 progress=82.1% loss=0.361308 lr=8.488630e-06 lr_backbone=8.488630e-06 elapsed=47:52:20 avg_step=4.20s recent_step=4.16s eta=10:25:23 +2026-09-22 06:19:39 | INFO | train rank=0 | step=41070/50000 progress=82.1% loss=0.366736 lr=8.470205e-06 lr_backbone=8.470205e-06 elapsed=47:53:02 avg_step=4.20s recent_step=4.18s eta=10:24:41 +2026-09-22 06:20:20 | INFO | train rank=0 | step=41080/50000 progress=82.2% loss=0.325565 lr=8.451799e-06 lr_backbone=8.451799e-06 elapsed=47:53:43 avg_step=4.20s recent_step=4.16s eta=10:23:59 +2026-09-22 06:21:01 | INFO | train rank=0 | step=41090/50000 progress=82.2% loss=0.290386 lr=8.433410e-06 lr_backbone=8.433410e-06 elapsed=47:54:25 avg_step=4.20s recent_step=4.11s eta=10:23:17 +2026-09-22 06:21:42 | INFO | train rank=0 | step=41100/50000 progress=82.2% loss=0.333711 lr=8.415040e-06 lr_backbone=8.415040e-06 elapsed=47:55:06 avg_step=4.20s recent_step=4.12s eta=10:22:35 +2026-09-22 06:22:24 | INFO | train rank=0 | step=41110/50000 progress=82.2% loss=0.366898 lr=8.396688e-06 lr_backbone=8.396688e-06 elapsed=47:55:48 avg_step=4.20s recent_step=4.19s eta=10:21:53 +2026-09-22 06:23:06 | INFO | train rank=0 | step=41120/50000 progress=82.2% loss=0.281986 lr=8.378355e-06 lr_backbone=8.378355e-06 elapsed=47:56:29 avg_step=4.20s recent_step=4.15s eta=10:21:11 +2026-09-22 06:23:48 | INFO | train rank=0 | step=41130/50000 progress=82.3% loss=0.409774 lr=8.360039e-06 lr_backbone=8.360039e-06 elapsed=47:57:11 avg_step=4.20s recent_step=4.19s eta=10:20:29 +2026-09-22 06:24:30 | INFO | train rank=0 | step=41140/50000 progress=82.3% loss=0.327955 lr=8.341742e-06 lr_backbone=8.341742e-06 elapsed=47:57:53 avg_step=4.20s recent_step=4.18s eta=10:19:47 +2026-09-22 06:25:11 | INFO | train rank=0 | step=41150/50000 progress=82.3% loss=0.284359 lr=8.323463e-06 lr_backbone=8.323463e-06 elapsed=47:58:34 avg_step=4.20s recent_step=4.12s eta=10:19:05 +2026-09-22 06:25:52 | INFO | train rank=0 | step=41160/50000 progress=82.3% loss=0.320253 lr=8.305202e-06 lr_backbone=8.305202e-06 elapsed=47:59:15 avg_step=4.20s recent_step=4.12s eta=10:18:23 +2026-09-22 06:26:33 | INFO | train rank=0 | step=41170/50000 progress=82.3% loss=0.326647 lr=8.286959e-06 lr_backbone=8.286959e-06 elapsed=47:59:57 avg_step=4.20s recent_step=4.15s eta=10:17:40 +2026-09-22 06:27:15 | INFO | train rank=0 | step=41180/50000 progress=82.4% loss=0.319868 lr=8.268735e-06 lr_backbone=8.268735e-06 elapsed=48:00:38 avg_step=4.20s recent_step=4.15s eta=10:16:58 +2026-09-22 06:27:57 | INFO | train rank=0 | step=41190/50000 progress=82.4% loss=0.442355 lr=8.250529e-06 lr_backbone=8.250529e-06 elapsed=48:01:20 avg_step=4.20s recent_step=4.17s eta=10:16:16 +2026-09-22 06:28:38 | INFO | train rank=0 | step=41200/50000 progress=82.4% loss=0.326623 lr=8.232341e-06 lr_backbone=8.232341e-06 elapsed=48:02:02 avg_step=4.20s recent_step=4.17s eta=10:15:34 +2026-09-22 06:29:20 | INFO | train rank=0 | step=41210/50000 progress=82.4% loss=0.293073 lr=8.214171e-06 lr_backbone=8.214171e-06 elapsed=48:02:43 avg_step=4.20s recent_step=4.12s eta=10:14:52 +2026-09-22 06:30:01 | INFO | train rank=0 | step=41220/50000 progress=82.4% loss=0.343327 lr=8.196020e-06 lr_backbone=8.196020e-06 elapsed=48:03:24 avg_step=4.20s recent_step=4.13s eta=10:14:10 +2026-09-22 06:30:43 | INFO | train rank=0 | step=41230/50000 progress=82.5% loss=0.267235 lr=8.177887e-06 lr_backbone=8.177887e-06 elapsed=48:04:06 avg_step=4.20s recent_step=4.19s eta=10:13:28 +2026-09-22 06:31:24 | INFO | train rank=0 | step=41240/50000 progress=82.5% loss=0.329173 lr=8.159772e-06 lr_backbone=8.159772e-06 elapsed=48:04:47 avg_step=4.20s recent_step=4.14s eta=10:12:46 +2026-09-22 06:32:06 | INFO | train rank=0 | step=41250/50000 progress=82.5% loss=0.396999 lr=8.141676e-06 lr_backbone=8.141676e-06 elapsed=48:05:29 avg_step=4.20s recent_step=4.17s eta=10:12:04 +2026-09-22 06:32:48 | INFO | train rank=0 | step=41260/50000 progress=82.5% loss=0.268714 lr=8.123598e-06 lr_backbone=8.123598e-06 elapsed=48:06:11 avg_step=4.20s recent_step=4.18s eta=10:11:22 +2026-09-22 06:33:29 | INFO | train rank=0 | step=41270/50000 progress=82.5% loss=0.272067 lr=8.105538e-06 lr_backbone=8.105538e-06 elapsed=48:06:52 avg_step=4.20s recent_step=4.13s eta=10:10:40 +2026-09-22 06:34:10 | INFO | train rank=0 | step=41280/50000 progress=82.6% loss=0.298703 lr=8.087497e-06 lr_backbone=8.087497e-06 elapsed=48:07:33 avg_step=4.20s recent_step=4.13s eta=10:09:58 +2026-09-22 06:34:52 | INFO | train rank=0 | step=41290/50000 progress=82.6% loss=0.364550 lr=8.069474e-06 lr_backbone=8.069474e-06 elapsed=48:08:15 avg_step=4.20s recent_step=4.18s eta=10:09:16 +2026-09-22 06:35:34 | INFO | train rank=0 | step=41300/50000 progress=82.6% loss=0.316094 lr=8.051469e-06 lr_backbone=8.051469e-06 elapsed=48:08:57 avg_step=4.20s recent_step=4.18s eta=10:08:34 +2026-09-22 06:36:15 | INFO | train rank=0 | step=41310/50000 progress=82.6% loss=0.278169 lr=8.033482e-06 lr_backbone=8.033482e-06 elapsed=48:09:39 avg_step=4.20s recent_step=4.15s eta=10:07:52 +2026-09-22 06:36:57 | INFO | train rank=0 | step=41320/50000 progress=82.6% loss=0.343716 lr=8.015514e-06 lr_backbone=8.015514e-06 elapsed=48:10:20 avg_step=4.20s recent_step=4.16s eta=10:07:10 +2026-09-22 06:37:38 | INFO | train rank=0 | step=41330/50000 progress=82.7% loss=0.289618 lr=7.997565e-06 lr_backbone=7.997565e-06 elapsed=48:11:01 avg_step=4.20s recent_step=4.12s eta=10:06:27 +2026-09-22 06:38:19 | INFO | train rank=0 | step=41340/50000 progress=82.7% loss=0.280040 lr=7.979633e-06 lr_backbone=7.979633e-06 elapsed=48:11:43 avg_step=4.20s recent_step=4.11s eta=10:05:45 +2026-09-22 06:39:01 | INFO | train rank=0 | step=41350/50000 progress=82.7% loss=0.347627 lr=7.961720e-06 lr_backbone=7.961720e-06 elapsed=48:12:24 avg_step=4.20s recent_step=4.18s eta=10:05:03 +2026-09-22 06:39:43 | INFO | train rank=0 | step=41360/50000 progress=82.7% loss=0.316594 lr=7.943826e-06 lr_backbone=7.943826e-06 elapsed=48:13:06 avg_step=4.20s recent_step=4.18s eta=10:04:21 +2026-09-22 06:40:24 | INFO | train rank=0 | step=41370/50000 progress=82.7% loss=0.299538 lr=7.925950e-06 lr_backbone=7.925950e-06 elapsed=48:13:48 avg_step=4.20s recent_step=4.15s eta=10:03:39 +2026-09-22 06:41:06 | INFO | train rank=0 | step=41380/50000 progress=82.8% loss=0.270525 lr=7.908092e-06 lr_backbone=7.908092e-06 elapsed=48:14:29 avg_step=4.20s recent_step=4.18s eta=10:02:57 +2026-09-22 06:41:47 | INFO | train rank=0 | step=41390/50000 progress=82.8% loss=0.289714 lr=7.890253e-06 lr_backbone=7.890253e-06 elapsed=48:15:11 avg_step=4.20s recent_step=4.13s eta=10:02:15 +2026-09-22 06:42:29 | INFO | train rank=0 | step=41400/50000 progress=82.8% loss=0.350825 lr=7.872432e-06 lr_backbone=7.872432e-06 elapsed=48:15:52 avg_step=4.20s recent_step=4.12s eta=10:01:33 +2026-09-22 06:43:10 | INFO | train rank=0 | step=41410/50000 progress=82.8% loss=0.391508 lr=7.854629e-06 lr_backbone=7.854629e-06 elapsed=48:16:33 avg_step=4.20s recent_step=4.16s eta=10:00:51 +2026-09-22 06:43:52 | INFO | train rank=0 | step=41420/50000 progress=82.8% loss=0.276324 lr=7.836845e-06 lr_backbone=7.836845e-06 elapsed=48:17:15 avg_step=4.20s recent_step=4.17s eta=10:00:09 +2026-09-22 06:44:34 | INFO | train rank=0 | step=41430/50000 progress=82.9% loss=0.399933 lr=7.819080e-06 lr_backbone=7.819080e-06 elapsed=48:17:57 avg_step=4.20s recent_step=4.17s eta=09:59:27 +2026-09-22 06:45:15 | INFO | train rank=0 | step=41440/50000 progress=82.9% loss=0.391457 lr=7.801332e-06 lr_backbone=7.801332e-06 elapsed=48:18:39 avg_step=4.20s recent_step=4.17s eta=09:58:45 +2026-09-22 06:45:56 | INFO | train rank=0 | step=41450/50000 progress=82.9% loss=0.343797 lr=7.783604e-06 lr_backbone=7.783604e-06 elapsed=48:19:20 avg_step=4.20s recent_step=4.10s eta=09:58:03 +2026-09-22 06:46:38 | INFO | train rank=0 | step=41460/50000 progress=82.9% loss=0.345415 lr=7.765893e-06 lr_backbone=7.765893e-06 elapsed=48:20:01 avg_step=4.20s recent_step=4.13s eta=09:57:21 +2026-09-22 06:47:19 | INFO | train rank=0 | step=41470/50000 progress=82.9% loss=0.324096 lr=7.748202e-06 lr_backbone=7.748202e-06 elapsed=48:20:43 avg_step=4.20s recent_step=4.16s eta=09:56:39 +2026-09-22 06:48:01 | INFO | train rank=0 | step=41480/50000 progress=83.0% loss=0.334970 lr=7.730528e-06 lr_backbone=7.730528e-06 elapsed=48:21:24 avg_step=4.20s recent_step=4.16s eta=09:55:57 +2026-09-22 06:48:42 | INFO | train rank=0 | step=41490/50000 progress=83.0% loss=0.365954 lr=7.712874e-06 lr_backbone=7.712874e-06 elapsed=48:22:06 avg_step=4.20s recent_step=4.13s eta=09:55:14 +2026-09-22 06:49:24 | INFO | train rank=0 | step=41500/50000 progress=83.0% loss=0.290666 lr=7.695237e-06 lr_backbone=7.695237e-06 elapsed=48:22:47 avg_step=4.20s recent_step=4.15s eta=09:54:32 +2026-09-22 06:50:05 | INFO | train rank=0 | step=41510/50000 progress=83.0% loss=0.307973 lr=7.677620e-06 lr_backbone=7.677620e-06 elapsed=48:23:28 avg_step=4.20s recent_step=4.12s eta=09:53:50 +2026-09-22 06:50:46 | INFO | train rank=0 | step=41520/50000 progress=83.0% loss=0.305077 lr=7.660020e-06 lr_backbone=7.660020e-06 elapsed=48:24:09 avg_step=4.20s recent_step=4.12s eta=09:53:08 +2026-09-22 06:51:28 | INFO | train rank=0 | step=41530/50000 progress=83.1% loss=0.302644 lr=7.642440e-06 lr_backbone=7.642440e-06 elapsed=48:24:51 avg_step=4.20s recent_step=4.18s eta=09:52:26 +2026-09-22 06:52:10 | INFO | train rank=0 | step=41540/50000 progress=83.1% loss=0.281011 lr=7.624877e-06 lr_backbone=7.624877e-06 elapsed=48:25:33 avg_step=4.20s recent_step=4.18s eta=09:51:44 +2026-09-22 06:52:52 | INFO | train rank=0 | step=41550/50000 progress=83.1% loss=0.313974 lr=7.607334e-06 lr_backbone=7.607334e-06 elapsed=48:26:15 avg_step=4.20s recent_step=4.19s eta=09:51:02 +2026-09-22 06:53:33 | INFO | train rank=0 | step=41560/50000 progress=83.1% loss=0.287415 lr=7.589809e-06 lr_backbone=7.589809e-06 elapsed=48:26:56 avg_step=4.20s recent_step=4.15s eta=09:50:20 +2026-09-22 06:54:14 | INFO | train rank=0 | step=41570/50000 progress=83.1% loss=0.358837 lr=7.572302e-06 lr_backbone=7.572302e-06 elapsed=48:27:38 avg_step=4.20s recent_step=4.12s eta=09:49:38 +2026-09-22 06:54:55 | INFO | train rank=0 | step=41580/50000 progress=83.2% loss=0.332151 lr=7.554814e-06 lr_backbone=7.554814e-06 elapsed=48:28:19 avg_step=4.20s recent_step=4.12s eta=09:48:56 +2026-09-22 06:55:37 | INFO | train rank=0 | step=41590/50000 progress=83.2% loss=0.350050 lr=7.537344e-06 lr_backbone=7.537344e-06 elapsed=48:29:01 avg_step=4.20s recent_step=4.18s eta=09:48:14 +2026-09-22 06:56:19 | INFO | train rank=0 | step=41600/50000 progress=83.2% loss=0.335295 lr=7.519894e-06 lr_backbone=7.519894e-06 elapsed=48:29:42 avg_step=4.20s recent_step=4.18s eta=09:47:32 +2026-09-22 06:57:01 | INFO | train rank=0 | step=41610/50000 progress=83.2% loss=0.383054 lr=7.502461e-06 lr_backbone=7.502461e-06 elapsed=48:30:24 avg_step=4.20s recent_step=4.17s eta=09:46:50 +2026-09-22 06:57:42 | INFO | train rank=0 | step=41620/50000 progress=83.2% loss=0.297483 lr=7.485048e-06 lr_backbone=7.485048e-06 elapsed=48:31:06 avg_step=4.20s recent_step=4.14s eta=09:46:08 +2026-09-22 06:58:23 | INFO | train rank=0 | step=41630/50000 progress=83.3% loss=0.311866 lr=7.467652e-06 lr_backbone=7.467652e-06 elapsed=48:31:47 avg_step=4.20s recent_step=4.12s eta=09:45:26 +2026-09-22 06:59:04 | INFO | train rank=0 | step=41640/50000 progress=83.3% loss=0.308781 lr=7.450276e-06 lr_backbone=7.450276e-06 elapsed=48:32:28 avg_step=4.20s recent_step=4.10s eta=09:44:43 +2026-09-22 06:59:46 | INFO | train rank=0 | step=41650/50000 progress=83.3% loss=0.286321 lr=7.432918e-06 lr_backbone=7.432918e-06 elapsed=48:33:09 avg_step=4.20s recent_step=4.16s eta=09:44:01 +2026-09-22 07:00:28 | INFO | train rank=0 | step=41660/50000 progress=83.3% loss=0.310821 lr=7.415579e-06 lr_backbone=7.415579e-06 elapsed=48:33:51 avg_step=4.20s recent_step=4.17s eta=09:43:19 +2026-09-22 07:01:09 | INFO | train rank=0 | step=41670/50000 progress=83.3% loss=0.292483 lr=7.398258e-06 lr_backbone=7.398258e-06 elapsed=48:34:32 avg_step=4.20s recent_step=4.14s eta=09:42:37 +2026-09-22 07:01:51 | INFO | train rank=0 | step=41680/50000 progress=83.4% loss=0.330658 lr=7.380956e-06 lr_backbone=7.380956e-06 elapsed=48:35:14 avg_step=4.20s recent_step=4.15s eta=09:41:55 +2026-09-22 07:02:31 | INFO | train rank=0 | step=41690/50000 progress=83.4% loss=0.305455 lr=7.363673e-06 lr_backbone=7.363673e-06 elapsed=48:35:55 avg_step=4.20s recent_step=4.09s eta=09:41:13 +2026-09-22 07:03:13 | INFO | train rank=0 | step=41700/50000 progress=83.4% loss=0.297812 lr=7.346408e-06 lr_backbone=7.346408e-06 elapsed=48:36:36 avg_step=4.20s recent_step=4.12s eta=09:40:31 +2026-09-22 07:03:54 | INFO | train rank=0 | step=41710/50000 progress=83.4% loss=0.292022 lr=7.329162e-06 lr_backbone=7.329162e-06 elapsed=48:37:17 avg_step=4.20s recent_step=4.15s eta=09:39:49 +2026-09-22 07:04:36 | INFO | train rank=0 | step=41720/50000 progress=83.4% loss=0.333647 lr=7.311934e-06 lr_backbone=7.311934e-06 elapsed=48:37:59 avg_step=4.20s recent_step=4.15s eta=09:39:07 +2026-09-22 07:05:17 | INFO | train rank=0 | step=41730/50000 progress=83.5% loss=0.281042 lr=7.294726e-06 lr_backbone=7.294726e-06 elapsed=48:38:40 avg_step=4.20s recent_step=4.15s eta=09:38:25 +2026-09-22 07:05:59 | INFO | train rank=0 | step=41740/50000 progress=83.5% loss=0.368818 lr=7.277536e-06 lr_backbone=7.277536e-06 elapsed=48:39:22 avg_step=4.20s recent_step=4.15s eta=09:37:43 +2026-09-22 07:06:40 | INFO | train rank=0 | step=41750/50000 progress=83.5% loss=0.370784 lr=7.260364e-06 lr_backbone=7.260364e-06 elapsed=48:40:03 avg_step=4.20s recent_step=4.10s eta=09:37:01 +2026-09-22 07:07:21 | INFO | train rank=0 | step=41760/50000 progress=83.5% loss=0.345243 lr=7.243212e-06 lr_backbone=7.243212e-06 elapsed=48:40:44 avg_step=4.20s recent_step=4.12s eta=09:36:18 +2026-09-22 07:08:02 | INFO | train rank=0 | step=41770/50000 progress=83.5% loss=0.385481 lr=7.226078e-06 lr_backbone=7.226078e-06 elapsed=48:41:26 avg_step=4.20s recent_step=4.16s eta=09:35:36 +2026-09-22 07:08:44 | INFO | train rank=0 | step=41780/50000 progress=83.6% loss=0.356373 lr=7.208963e-06 lr_backbone=7.208963e-06 elapsed=48:42:07 avg_step=4.20s recent_step=4.15s eta=09:34:54 +2026-09-22 07:09:25 | INFO | train rank=0 | step=41790/50000 progress=83.6% loss=0.297245 lr=7.191866e-06 lr_backbone=7.191866e-06 elapsed=48:42:49 avg_step=4.20s recent_step=4.15s eta=09:34:12 +2026-09-22 07:10:07 | INFO | train rank=0 | step=41800/50000 progress=83.6% loss=0.258460 lr=7.174788e-06 lr_backbone=7.174788e-06 elapsed=48:43:30 avg_step=4.20s recent_step=4.15s eta=09:33:30 +2026-09-22 07:10:48 | INFO | train rank=0 | step=41810/50000 progress=83.6% loss=0.333097 lr=7.157729e-06 lr_backbone=7.157729e-06 elapsed=48:44:12 avg_step=4.20s recent_step=4.13s eta=09:32:48 +2026-09-22 07:11:29 | INFO | train rank=0 | step=41820/50000 progress=83.6% loss=0.356584 lr=7.140689e-06 lr_backbone=7.140689e-06 elapsed=48:44:52 avg_step=4.20s recent_step=4.09s eta=09:32:06 +2026-09-22 07:12:11 | INFO | train rank=0 | step=41830/50000 progress=83.7% loss=0.337646 lr=7.123667e-06 lr_backbone=7.123667e-06 elapsed=48:45:34 avg_step=4.20s recent_step=4.15s eta=09:31:24 +2026-09-22 07:12:52 | INFO | train rank=0 | step=41840/50000 progress=83.7% loss=0.337039 lr=7.106664e-06 lr_backbone=7.106664e-06 elapsed=48:46:15 avg_step=4.20s recent_step=4.13s eta=09:30:42 +2026-09-22 07:13:34 | INFO | train rank=0 | step=41850/50000 progress=83.7% loss=0.380960 lr=7.089680e-06 lr_backbone=7.089680e-06 elapsed=48:46:57 avg_step=4.20s recent_step=4.16s eta=09:30:00 +2026-09-22 07:14:15 | INFO | train rank=0 | step=41860/50000 progress=83.7% loss=0.308491 lr=7.072715e-06 lr_backbone=7.072715e-06 elapsed=48:47:38 avg_step=4.20s recent_step=4.15s eta=09:29:18 +2026-09-22 07:14:56 | INFO | train rank=0 | step=41870/50000 progress=83.7% loss=0.319103 lr=7.055768e-06 lr_backbone=7.055768e-06 elapsed=48:48:19 avg_step=4.20s recent_step=4.10s eta=09:28:36 +2026-09-22 07:15:37 | INFO | train rank=0 | step=41880/50000 progress=83.8% loss=0.344046 lr=7.038841e-06 lr_backbone=7.038841e-06 elapsed=48:49:00 avg_step=4.20s recent_step=4.09s eta=09:27:53 +2026-09-22 07:16:18 | INFO | train rank=0 | step=41890/50000 progress=83.8% loss=0.349228 lr=7.021932e-06 lr_backbone=7.021932e-06 elapsed=48:49:42 avg_step=4.20s recent_step=4.15s eta=09:27:11 +2026-09-22 07:17:00 | INFO | train rank=0 | step=41900/50000 progress=83.8% loss=0.355913 lr=7.005042e-06 lr_backbone=7.005042e-06 elapsed=48:50:23 avg_step=4.20s recent_step=4.15s eta=09:26:29 +2026-09-22 07:17:42 | INFO | train rank=0 | step=41910/50000 progress=83.8% loss=0.335770 lr=6.988170e-06 lr_backbone=6.988170e-06 elapsed=48:51:05 avg_step=4.20s recent_step=4.16s eta=09:25:47 +2026-09-22 07:18:23 | INFO | train rank=0 | step=41920/50000 progress=83.8% loss=0.285544 lr=6.971318e-06 lr_backbone=6.971318e-06 elapsed=48:51:46 avg_step=4.20s recent_step=4.16s eta=09:25:05 +2026-09-22 07:19:04 | INFO | train rank=0 | step=41930/50000 progress=83.9% loss=0.296756 lr=6.954484e-06 lr_backbone=6.954484e-06 elapsed=48:52:28 avg_step=4.20s recent_step=4.12s eta=09:24:23 +2026-09-22 07:19:45 | INFO | train rank=0 | step=41940/50000 progress=83.9% loss=0.346972 lr=6.937669e-06 lr_backbone=6.937669e-06 elapsed=48:53:09 avg_step=4.20s recent_step=4.10s eta=09:23:41 +2026-09-22 07:20:27 | INFO | train rank=0 | step=41950/50000 progress=83.9% loss=0.352372 lr=6.920873e-06 lr_backbone=6.920873e-06 elapsed=48:53:50 avg_step=4.20s recent_step=4.16s eta=09:22:59 +2026-09-22 07:21:08 | INFO | train rank=0 | step=41960/50000 progress=83.9% loss=0.311440 lr=6.904096e-06 lr_backbone=6.904096e-06 elapsed=48:54:32 avg_step=4.20s recent_step=4.15s eta=09:22:17 +2026-09-22 07:21:50 | INFO | train rank=0 | step=41970/50000 progress=83.9% loss=0.297253 lr=6.887338e-06 lr_backbone=6.887338e-06 elapsed=48:55:13 avg_step=4.20s recent_step=4.14s eta=09:21:35 +2026-09-22 07:22:31 | INFO | train rank=0 | step=41980/50000 progress=84.0% loss=0.273186 lr=6.870598e-06 lr_backbone=6.870598e-06 elapsed=48:55:55 avg_step=4.20s recent_step=4.16s eta=09:20:53 +2026-09-22 07:23:13 | INFO | train rank=0 | step=41990/50000 progress=84.0% loss=0.319775 lr=6.853878e-06 lr_backbone=6.853878e-06 elapsed=48:56:36 avg_step=4.20s recent_step=4.12s eta=09:20:11 +2026-09-22 07:23:54 | INFO | train rank=0 | step=42000/50000 progress=84.0% loss=0.331203 lr=6.837176e-06 lr_backbone=6.837176e-06 elapsed=48:57:17 avg_step=4.20s recent_step=4.13s eta=09:19:29 +2026-09-22 07:24:35 | INFO | train rank=0 | step=42010/50000 progress=84.0% loss=0.275891 lr=6.820493e-06 lr_backbone=6.820493e-06 elapsed=48:57:59 avg_step=4.20s recent_step=4.15s eta=09:18:47 +2026-09-22 07:25:17 | INFO | train rank=0 | step=42020/50000 progress=84.0% loss=0.316801 lr=6.803829e-06 lr_backbone=6.803829e-06 elapsed=48:58:40 avg_step=4.20s recent_step=4.17s eta=09:18:05 +2026-09-22 07:25:59 | INFO | train rank=0 | step=42030/50000 progress=84.1% loss=0.287432 lr=6.787184e-06 lr_backbone=6.787184e-06 elapsed=48:59:22 avg_step=4.20s recent_step=4.17s eta=09:17:23 +2026-09-22 07:26:40 | INFO | train rank=0 | step=42040/50000 progress=84.1% loss=0.398055 lr=6.770558e-06 lr_backbone=6.770558e-06 elapsed=49:00:04 avg_step=4.20s recent_step=4.14s eta=09:16:40 +2026-09-22 07:27:21 | INFO | train rank=0 | step=42050/50000 progress=84.1% loss=0.294606 lr=6.753951e-06 lr_backbone=6.753951e-06 elapsed=49:00:45 avg_step=4.20s recent_step=4.13s eta=09:15:58 +2026-09-22 07:28:03 | INFO | train rank=0 | step=42060/50000 progress=84.1% loss=0.359902 lr=6.737362e-06 lr_backbone=6.737362e-06 elapsed=49:01:26 avg_step=4.20s recent_step=4.12s eta=09:15:16 +2026-09-22 07:28:44 | INFO | train rank=0 | step=42070/50000 progress=84.1% loss=0.332486 lr=6.720793e-06 lr_backbone=6.720793e-06 elapsed=49:02:08 avg_step=4.20s recent_step=4.15s eta=09:14:34 +2026-09-22 07:29:26 | INFO | train rank=0 | step=42080/50000 progress=84.2% loss=0.297331 lr=6.704242e-06 lr_backbone=6.704242e-06 elapsed=49:02:49 avg_step=4.20s recent_step=4.18s eta=09:13:52 +2026-09-22 07:30:08 | INFO | train rank=0 | step=42090/50000 progress=84.2% loss=0.247751 lr=6.687711e-06 lr_backbone=6.687711e-06 elapsed=49:03:31 avg_step=4.20s recent_step=4.16s eta=09:13:10 +2026-09-22 07:30:50 | INFO | train rank=0 | step=42100/50000 progress=84.2% loss=0.347880 lr=6.671198e-06 lr_backbone=6.671198e-06 elapsed=49:04:13 avg_step=4.20s recent_step=4.20s eta=09:12:28 +2026-09-22 07:31:31 | INFO | train rank=0 | step=42110/50000 progress=84.2% loss=0.435095 lr=6.654705e-06 lr_backbone=6.654705e-06 elapsed=49:04:54 avg_step=4.20s recent_step=4.13s eta=09:11:46 +2026-09-22 07:32:12 | INFO | train rank=0 | step=42120/50000 progress=84.2% loss=0.323789 lr=6.638230e-06 lr_backbone=6.638230e-06 elapsed=49:05:36 avg_step=4.20s recent_step=4.13s eta=09:11:04 +2026-09-22 07:32:54 | INFO | train rank=0 | step=42130/50000 progress=84.3% loss=0.296317 lr=6.621774e-06 lr_backbone=6.621774e-06 elapsed=49:06:17 avg_step=4.20s recent_step=4.15s eta=09:10:22 +2026-09-22 07:33:35 | INFO | train rank=0 | step=42140/50000 progress=84.3% loss=0.296232 lr=6.605338e-06 lr_backbone=6.605338e-06 elapsed=49:06:59 avg_step=4.20s recent_step=4.17s eta=09:09:40 +2026-09-22 07:34:17 | INFO | train rank=0 | step=42150/50000 progress=84.3% loss=0.367587 lr=6.588920e-06 lr_backbone=6.588920e-06 elapsed=49:07:40 avg_step=4.20s recent_step=4.14s eta=09:08:58 +2026-09-22 07:34:59 | INFO | train rank=0 | step=42160/50000 progress=84.3% loss=0.351468 lr=6.572521e-06 lr_backbone=6.572521e-06 elapsed=49:08:22 avg_step=4.20s recent_step=4.18s eta=09:08:16 +2026-09-22 07:35:40 | INFO | train rank=0 | step=42170/50000 progress=84.3% loss=0.291305 lr=6.556141e-06 lr_backbone=6.556141e-06 elapsed=49:09:04 avg_step=4.20s recent_step=4.15s eta=09:07:34 +2026-09-22 07:36:21 | INFO | train rank=0 | step=42180/50000 progress=84.4% loss=0.303597 lr=6.539780e-06 lr_backbone=6.539780e-06 elapsed=49:09:45 avg_step=4.20s recent_step=4.12s eta=09:06:52 +2026-09-22 07:37:03 | INFO | train rank=0 | step=42190/50000 progress=84.4% loss=0.319335 lr=6.523439e-06 lr_backbone=6.523439e-06 elapsed=49:10:26 avg_step=4.20s recent_step=4.18s eta=09:06:10 +2026-09-22 07:37:45 | INFO | train rank=0 | step=42200/50000 progress=84.4% loss=0.264268 lr=6.507116e-06 lr_backbone=6.507116e-06 elapsed=49:11:08 avg_step=4.20s recent_step=4.15s eta=09:05:28 +2026-09-22 07:38:26 | INFO | train rank=0 | step=42210/50000 progress=84.4% loss=0.290789 lr=6.490812e-06 lr_backbone=6.490812e-06 elapsed=49:11:50 avg_step=4.20s recent_step=4.17s eta=09:04:46 +2026-09-22 07:39:08 | INFO | train rank=0 | step=42220/50000 progress=84.4% loss=0.328344 lr=6.474528e-06 lr_backbone=6.474528e-06 elapsed=49:12:32 avg_step=4.20s recent_step=4.18s eta=09:04:04 +2026-09-22 07:39:50 | INFO | train rank=0 | step=42230/50000 progress=84.5% loss=0.360883 lr=6.458262e-06 lr_backbone=6.458262e-06 elapsed=49:13:13 avg_step=4.20s recent_step=4.13s eta=09:03:22 +2026-09-22 07:40:31 | INFO | train rank=0 | step=42240/50000 progress=84.5% loss=0.293921 lr=6.442015e-06 lr_backbone=6.442015e-06 elapsed=49:13:54 avg_step=4.20s recent_step=4.13s eta=09:02:40 +2026-09-22 07:41:13 | INFO | train rank=0 | step=42250/50000 progress=84.5% loss=0.360078 lr=6.425788e-06 lr_backbone=6.425788e-06 elapsed=49:14:36 avg_step=4.20s recent_step=4.17s eta=09:01:58 +2026-09-22 07:41:54 | INFO | train rank=0 | step=42260/50000 progress=84.5% loss=0.419026 lr=6.409579e-06 lr_backbone=6.409579e-06 elapsed=49:15:17 avg_step=4.20s recent_step=4.16s eta=09:01:16 +2026-09-22 07:42:36 | INFO | train rank=0 | step=42270/50000 progress=84.5% loss=0.358693 lr=6.393390e-06 lr_backbone=6.393390e-06 elapsed=49:15:59 avg_step=4.20s recent_step=4.17s eta=09:00:34 +2026-09-22 07:43:18 | INFO | train rank=0 | step=42280/50000 progress=84.6% loss=0.414036 lr=6.377220e-06 lr_backbone=6.377220e-06 elapsed=49:16:41 avg_step=4.20s recent_step=4.17s eta=08:59:52 +2026-09-22 07:43:59 | INFO | train rank=0 | step=42290/50000 progress=84.6% loss=0.399044 lr=6.361068e-06 lr_backbone=6.361068e-06 elapsed=49:17:22 avg_step=4.20s recent_step=4.11s eta=08:59:09 +2026-09-22 07:44:40 | INFO | train rank=0 | step=42300/50000 progress=84.6% loss=0.314006 lr=6.344936e-06 lr_backbone=6.344936e-06 elapsed=49:18:03 avg_step=4.20s recent_step=4.14s eta=08:58:27 +2026-09-22 07:45:22 | INFO | train rank=0 | step=42310/50000 progress=84.6% loss=0.302071 lr=6.328823e-06 lr_backbone=6.328823e-06 elapsed=49:18:45 avg_step=4.20s recent_step=4.18s eta=08:57:45 +2026-09-22 07:46:04 | INFO | train rank=0 | step=42320/50000 progress=84.6% loss=0.328199 lr=6.312729e-06 lr_backbone=6.312729e-06 elapsed=49:19:27 avg_step=4.20s recent_step=4.18s eta=08:57:03 +2026-09-22 07:46:45 | INFO | train rank=0 | step=42330/50000 progress=84.7% loss=0.278490 lr=6.296654e-06 lr_backbone=6.296654e-06 elapsed=49:20:08 avg_step=4.20s recent_step=4.14s eta=08:56:21 +2026-09-22 07:47:27 | INFO | train rank=0 | step=42340/50000 progress=84.7% loss=0.309756 lr=6.280599e-06 lr_backbone=6.280599e-06 elapsed=49:20:50 avg_step=4.20s recent_step=4.17s eta=08:55:39 +2026-09-22 07:48:08 | INFO | train rank=0 | step=42350/50000 progress=84.7% loss=0.252300 lr=6.264562e-06 lr_backbone=6.264562e-06 elapsed=49:21:31 avg_step=4.20s recent_step=4.13s eta=08:54:57 +2026-09-22 07:48:49 | INFO | train rank=0 | step=42360/50000 progress=84.7% loss=0.361880 lr=6.248544e-06 lr_backbone=6.248544e-06 elapsed=49:22:12 avg_step=4.20s recent_step=4.11s eta=08:54:15 +2026-09-22 07:49:31 | INFO | train rank=0 | step=42370/50000 progress=84.7% loss=0.282978 lr=6.232546e-06 lr_backbone=6.232546e-06 elapsed=49:22:54 avg_step=4.20s recent_step=4.17s eta=08:53:33 +2026-09-22 07:50:12 | INFO | train rank=0 | step=42380/50000 progress=84.8% loss=0.259814 lr=6.216567e-06 lr_backbone=6.216567e-06 elapsed=49:23:36 avg_step=4.20s recent_step=4.17s eta=08:52:51 +2026-09-22 07:50:54 | INFO | train rank=0 | step=42390/50000 progress=84.8% loss=0.262377 lr=6.200607e-06 lr_backbone=6.200607e-06 elapsed=49:24:17 avg_step=4.20s recent_step=4.16s eta=08:52:09 +2026-09-22 07:51:36 | INFO | train rank=0 | step=42400/50000 progress=84.8% loss=0.284809 lr=6.184666e-06 lr_backbone=6.184666e-06 elapsed=49:24:59 avg_step=4.20s recent_step=4.18s eta=08:51:27 +2026-09-22 07:52:17 | INFO | train rank=0 | step=42410/50000 progress=84.8% loss=0.258238 lr=6.168744e-06 lr_backbone=6.168744e-06 elapsed=49:25:40 avg_step=4.20s recent_step=4.14s eta=08:50:45 +2026-09-22 07:52:58 | INFO | train rank=0 | step=42420/50000 progress=84.8% loss=0.332023 lr=6.152842e-06 lr_backbone=6.152842e-06 elapsed=49:26:22 avg_step=4.20s recent_step=4.11s eta=08:50:03 +2026-09-22 07:53:40 | INFO | train rank=0 | step=42430/50000 progress=84.9% loss=0.295264 lr=6.136958e-06 lr_backbone=6.136958e-06 elapsed=49:27:03 avg_step=4.20s recent_step=4.17s eta=08:49:21 +2026-09-22 07:54:22 | INFO | train rank=0 | step=42440/50000 progress=84.9% loss=0.291380 lr=6.121094e-06 lr_backbone=6.121094e-06 elapsed=49:27:45 avg_step=4.20s recent_step=4.18s eta=08:48:39 +2026-09-22 07:55:03 | INFO | train rank=0 | step=42450/50000 progress=84.9% loss=0.309836 lr=6.105249e-06 lr_backbone=6.105249e-06 elapsed=49:28:27 avg_step=4.20s recent_step=4.17s eta=08:47:57 +2026-09-22 07:55:45 | INFO | train rank=0 | step=42460/50000 progress=84.9% loss=0.295025 lr=6.089423e-06 lr_backbone=6.089423e-06 elapsed=49:29:08 avg_step=4.20s recent_step=4.15s eta=08:47:15 +2026-09-22 07:56:26 | INFO | train rank=0 | step=42470/50000 progress=84.9% loss=0.335701 lr=6.073617e-06 lr_backbone=6.073617e-06 elapsed=49:29:50 avg_step=4.20s recent_step=4.13s eta=08:46:33 +2026-09-22 07:57:08 | INFO | train rank=0 | step=42480/50000 progress=85.0% loss=0.279356 lr=6.057829e-06 lr_backbone=6.057829e-06 elapsed=49:30:31 avg_step=4.20s recent_step=4.14s eta=08:45:51 +2026-09-22 07:57:49 | INFO | train rank=0 | step=42490/50000 progress=85.0% loss=0.300048 lr=6.042061e-06 lr_backbone=6.042061e-06 elapsed=49:31:12 avg_step=4.20s recent_step=4.14s eta=08:45:09 +2026-09-22 07:58:31 | INFO | train rank=0 | step=42500/50000 progress=85.0% loss=0.331332 lr=6.026312e-06 lr_backbone=6.026312e-06 elapsed=49:31:54 avg_step=4.20s recent_step=4.18s eta=08:44:27 +2026-09-22 07:59:13 | INFO | train rank=0 | step=42510/50000 progress=85.0% loss=0.315099 lr=6.010583e-06 lr_backbone=6.010583e-06 elapsed=49:32:36 avg_step=4.20s recent_step=4.18s eta=08:43:45 +2026-09-22 07:59:54 | INFO | train rank=0 | step=42520/50000 progress=85.0% loss=0.314857 lr=5.994872e-06 lr_backbone=5.994872e-06 elapsed=49:33:18 avg_step=4.20s recent_step=4.15s eta=08:43:03 +2026-09-22 08:00:35 | INFO | train rank=0 | step=42530/50000 progress=85.1% loss=0.309254 lr=5.979181e-06 lr_backbone=5.979181e-06 elapsed=49:33:59 avg_step=4.20s recent_step=4.13s eta=08:42:21 +2026-09-22 08:01:17 | INFO | train rank=0 | step=42540/50000 progress=85.1% loss=0.337736 lr=5.963509e-06 lr_backbone=5.963509e-06 elapsed=49:34:40 avg_step=4.20s recent_step=4.13s eta=08:41:39 +2026-09-22 08:01:58 | INFO | train rank=0 | step=42550/50000 progress=85.1% loss=0.342427 lr=5.947857e-06 lr_backbone=5.947857e-06 elapsed=49:35:22 avg_step=4.20s recent_step=4.17s eta=08:40:57 +2026-09-22 08:02:40 | INFO | train rank=0 | step=42560/50000 progress=85.1% loss=0.401780 lr=5.932223e-06 lr_backbone=5.932223e-06 elapsed=49:36:03 avg_step=4.20s recent_step=4.17s eta=08:40:15 +2026-09-22 08:03:22 | INFO | train rank=0 | step=42570/50000 progress=85.1% loss=0.332674 lr=5.916609e-06 lr_backbone=5.916609e-06 elapsed=49:36:45 avg_step=4.20s recent_step=4.15s eta=08:39:33 +2026-09-22 08:04:03 | INFO | train rank=0 | step=42580/50000 progress=85.2% loss=0.329404 lr=5.901014e-06 lr_backbone=5.901014e-06 elapsed=49:37:26 avg_step=4.20s recent_step=4.15s eta=08:38:51 +2026-09-22 08:04:44 | INFO | train rank=0 | step=42590/50000 progress=85.2% loss=0.346887 lr=5.885439e-06 lr_backbone=5.885439e-06 elapsed=49:38:08 avg_step=4.20s recent_step=4.11s eta=08:38:08 +2026-09-22 08:05:25 | INFO | train rank=0 | step=42600/50000 progress=85.2% loss=0.296888 lr=5.869882e-06 lr_backbone=5.869882e-06 elapsed=49:38:49 avg_step=4.20s recent_step=4.12s eta=08:37:26 +2026-09-22 08:06:07 | INFO | train rank=0 | step=42610/50000 progress=85.2% loss=0.280442 lr=5.854345e-06 lr_backbone=5.854345e-06 elapsed=49:39:31 avg_step=4.20s recent_step=4.18s eta=08:36:44 +2026-09-22 08:06:49 | INFO | train rank=0 | step=42620/50000 progress=85.2% loss=0.340717 lr=5.838828e-06 lr_backbone=5.838828e-06 elapsed=49:40:12 avg_step=4.20s recent_step=4.18s eta=08:36:02 +2026-09-22 08:07:31 | INFO | train rank=0 | step=42630/50000 progress=85.3% loss=0.259692 lr=5.823330e-06 lr_backbone=5.823330e-06 elapsed=49:40:54 avg_step=4.20s recent_step=4.16s eta=08:35:20 +2026-09-22 08:08:12 | INFO | train rank=0 | step=42640/50000 progress=85.3% loss=0.284123 lr=5.807851e-06 lr_backbone=5.807851e-06 elapsed=49:41:35 avg_step=4.20s recent_step=4.15s eta=08:34:38 +2026-09-22 08:08:53 | INFO | train rank=0 | step=42650/50000 progress=85.3% loss=0.318465 lr=5.792391e-06 lr_backbone=5.792391e-06 elapsed=49:42:16 avg_step=4.20s recent_step=4.09s eta=08:33:56 +2026-09-22 08:09:34 | INFO | train rank=0 | step=42660/50000 progress=85.3% loss=0.350231 lr=5.776951e-06 lr_backbone=5.776951e-06 elapsed=49:42:57 avg_step=4.20s recent_step=4.10s eta=08:33:14 +2026-09-22 08:10:16 | INFO | train rank=0 | step=42670/50000 progress=85.3% loss=0.320202 lr=5.761530e-06 lr_backbone=5.761530e-06 elapsed=49:43:39 avg_step=4.20s recent_step=4.16s eta=08:32:32 +2026-09-22 08:10:57 | INFO | train rank=0 | step=42680/50000 progress=85.4% loss=0.301135 lr=5.746128e-06 lr_backbone=5.746128e-06 elapsed=49:44:21 avg_step=4.20s recent_step=4.15s eta=08:31:50 +2026-09-22 08:11:39 | INFO | train rank=0 | step=42690/50000 progress=85.4% loss=0.283665 lr=5.730746e-06 lr_backbone=5.730746e-06 elapsed=49:45:02 avg_step=4.20s recent_step=4.15s eta=08:31:08 +2026-09-22 08:12:20 | INFO | train rank=0 | step=42700/50000 progress=85.4% loss=0.291521 lr=5.715383e-06 lr_backbone=5.715383e-06 elapsed=49:45:44 avg_step=4.20s recent_step=4.16s eta=08:30:26 +2026-09-22 08:13:02 | INFO | train rank=0 | step=42710/50000 progress=85.4% loss=0.343251 lr=5.700039e-06 lr_backbone=5.700039e-06 elapsed=49:46:25 avg_step=4.20s recent_step=4.11s eta=08:29:44 +2026-09-22 08:13:42 | INFO | train rank=0 | step=42720/50000 progress=85.4% loss=0.341298 lr=5.684715e-06 lr_backbone=5.684715e-06 elapsed=49:47:06 avg_step=4.20s recent_step=4.09s eta=08:29:02 +2026-09-22 08:14:24 | INFO | train rank=0 | step=42730/50000 progress=85.5% loss=0.292732 lr=5.669410e-06 lr_backbone=5.669410e-06 elapsed=49:47:47 avg_step=4.20s recent_step=4.16s eta=08:28:20 +2026-09-22 08:15:06 | INFO | train rank=0 | step=42740/50000 progress=85.5% loss=0.359779 lr=5.654125e-06 lr_backbone=5.654125e-06 elapsed=49:48:29 avg_step=4.20s recent_step=4.17s eta=08:27:38 +2026-09-22 08:15:47 | INFO | train rank=0 | step=42750/50000 progress=85.5% loss=0.335997 lr=5.638859e-06 lr_backbone=5.638859e-06 elapsed=49:49:11 avg_step=4.20s recent_step=4.14s eta=08:26:56 +2026-09-22 08:16:29 | INFO | train rank=0 | step=42760/50000 progress=85.5% loss=0.331121 lr=5.623612e-06 lr_backbone=5.623612e-06 elapsed=49:49:52 avg_step=4.20s recent_step=4.15s eta=08:26:14 +2026-09-22 08:17:10 | INFO | train rank=0 | step=42770/50000 progress=85.5% loss=0.288873 lr=5.608385e-06 lr_backbone=5.608385e-06 elapsed=49:50:33 avg_step=4.20s recent_step=4.11s eta=08:25:32 +2026-09-22 08:17:51 | INFO | train rank=0 | step=42780/50000 progress=85.6% loss=0.362443 lr=5.593178e-06 lr_backbone=5.593178e-06 elapsed=49:51:14 avg_step=4.20s recent_step=4.09s eta=08:24:50 +2026-09-22 08:18:32 | INFO | train rank=0 | step=42790/50000 progress=85.6% loss=0.350430 lr=5.577989e-06 lr_backbone=5.577989e-06 elapsed=49:51:56 avg_step=4.20s recent_step=4.16s eta=08:24:07 +2026-09-22 08:19:14 | INFO | train rank=0 | step=42800/50000 progress=85.6% loss=0.360325 lr=5.562820e-06 lr_backbone=5.562820e-06 elapsed=49:52:37 avg_step=4.20s recent_step=4.15s eta=08:23:25 +2026-09-22 08:19:55 | INFO | train rank=0 | step=42810/50000 progress=85.6% loss=0.263625 lr=5.547671e-06 lr_backbone=5.547671e-06 elapsed=49:53:19 avg_step=4.20s recent_step=4.15s eta=08:22:43 +2026-09-22 08:20:37 | INFO | train rank=0 | step=42820/50000 progress=85.6% loss=0.338659 lr=5.532541e-06 lr_backbone=5.532541e-06 elapsed=49:54:00 avg_step=4.20s recent_step=4.16s eta=08:22:01 +2026-09-22 08:21:18 | INFO | train rank=0 | step=42830/50000 progress=85.7% loss=0.399309 lr=5.517430e-06 lr_backbone=5.517430e-06 elapsed=49:54:41 avg_step=4.20s recent_step=4.09s eta=08:21:19 +2026-09-22 08:21:59 | INFO | train rank=0 | step=42840/50000 progress=85.7% loss=0.301185 lr=5.502339e-06 lr_backbone=5.502339e-06 elapsed=49:55:22 avg_step=4.20s recent_step=4.12s eta=08:20:37 +2026-09-22 08:22:41 | INFO | train rank=0 | step=42850/50000 progress=85.7% loss=0.351586 lr=5.487268e-06 lr_backbone=5.487268e-06 elapsed=49:56:04 avg_step=4.20s recent_step=4.16s eta=08:19:55 +2026-09-22 08:23:22 | INFO | train rank=0 | step=42860/50000 progress=85.7% loss=0.372808 lr=5.472215e-06 lr_backbone=5.472215e-06 elapsed=49:56:46 avg_step=4.20s recent_step=4.16s eta=08:19:13 +2026-09-22 08:24:04 | INFO | train rank=0 | step=42870/50000 progress=85.7% loss=0.361066 lr=5.457183e-06 lr_backbone=5.457183e-06 elapsed=49:57:27 avg_step=4.20s recent_step=4.15s eta=08:18:31 +2026-09-22 08:24:45 | INFO | train rank=0 | step=42880/50000 progress=85.8% loss=0.299817 lr=5.442170e-06 lr_backbone=5.442170e-06 elapsed=49:58:09 avg_step=4.20s recent_step=4.15s eta=08:17:49 +2026-09-22 08:25:26 | INFO | train rank=0 | step=42890/50000 progress=85.8% loss=0.275519 lr=5.427176e-06 lr_backbone=5.427176e-06 elapsed=49:58:50 avg_step=4.20s recent_step=4.11s eta=08:17:07 +2026-09-22 08:26:07 | INFO | train rank=0 | step=42900/50000 progress=85.8% loss=0.351382 lr=5.412202e-06 lr_backbone=5.412202e-06 elapsed=49:59:31 avg_step=4.20s recent_step=4.10s eta=08:16:25 +2026-09-22 08:26:49 | INFO | train rank=0 | step=42910/50000 progress=85.8% loss=0.306692 lr=5.397247e-06 lr_backbone=5.397247e-06 elapsed=50:00:12 avg_step=4.20s recent_step=4.14s eta=08:15:43 +2026-09-22 08:27:30 | INFO | train rank=0 | step=42920/50000 progress=85.8% loss=0.320530 lr=5.382312e-06 lr_backbone=5.382312e-06 elapsed=50:00:54 avg_step=4.20s recent_step=4.15s eta=08:15:01 +2026-09-22 08:28:12 | INFO | train rank=0 | step=42930/50000 progress=85.9% loss=0.261635 lr=5.367396e-06 lr_backbone=5.367396e-06 elapsed=50:01:35 avg_step=4.20s recent_step=4.15s eta=08:14:19 +2026-09-22 08:28:53 | INFO | train rank=0 | step=42940/50000 progress=85.9% loss=0.324092 lr=5.352500e-06 lr_backbone=5.352500e-06 elapsed=50:02:17 avg_step=4.20s recent_step=4.16s eta=08:13:37 +2026-09-22 08:29:34 | INFO | train rank=0 | step=42950/50000 progress=85.9% loss=0.371058 lr=5.337623e-06 lr_backbone=5.337623e-06 elapsed=50:02:58 avg_step=4.20s recent_step=4.10s eta=08:12:55 +2026-09-22 08:30:15 | INFO | train rank=0 | step=42960/50000 progress=85.9% loss=0.357784 lr=5.322766e-06 lr_backbone=5.322766e-06 elapsed=50:03:39 avg_step=4.20s recent_step=4.09s eta=08:12:13 +2026-09-22 08:30:57 | INFO | train rank=0 | step=42970/50000 progress=85.9% loss=0.280012 lr=5.307929e-06 lr_backbone=5.307929e-06 elapsed=50:04:20 avg_step=4.20s recent_step=4.16s eta=08:11:31 +2026-09-22 08:31:39 | INFO | train rank=0 | step=42980/50000 progress=86.0% loss=0.332687 lr=5.293111e-06 lr_backbone=5.293111e-06 elapsed=50:05:02 avg_step=4.20s recent_step=4.17s eta=08:10:49 +2026-09-22 08:32:20 | INFO | train rank=0 | step=42990/50000 progress=86.0% loss=0.326006 lr=5.278312e-06 lr_backbone=5.278312e-06 elapsed=50:05:43 avg_step=4.20s recent_step=4.14s eta=08:10:07 +2026-09-22 08:33:02 | INFO | train rank=0 | step=43000/50000 progress=86.0% loss=0.293894 lr=5.263534e-06 lr_backbone=5.263534e-06 elapsed=50:06:25 avg_step=4.20s recent_step=4.15s eta=08:09:25 +2026-09-22 08:33:43 | INFO | train rank=0 | step=43010/50000 progress=86.0% loss=0.283878 lr=5.248774e-06 lr_backbone=5.248774e-06 elapsed=50:07:06 avg_step=4.19s recent_step=4.11s eta=08:08:42 +2026-09-22 08:34:24 | INFO | train rank=0 | step=43020/50000 progress=86.0% loss=0.308035 lr=5.234034e-06 lr_backbone=5.234034e-06 elapsed=50:07:47 avg_step=4.19s recent_step=4.10s eta=08:08:00 +2026-09-22 08:35:05 | INFO | train rank=0 | step=43030/50000 progress=86.1% loss=0.370928 lr=5.219314e-06 lr_backbone=5.219314e-06 elapsed=50:08:29 avg_step=4.19s recent_step=4.16s eta=08:07:18 +2026-09-22 08:35:47 | INFO | train rank=0 | step=43040/50000 progress=86.1% loss=0.375217 lr=5.204614e-06 lr_backbone=5.204614e-06 elapsed=50:09:10 avg_step=4.19s recent_step=4.16s eta=08:06:36 +2026-09-22 08:36:28 | INFO | train rank=0 | step=43050/50000 progress=86.1% loss=0.340205 lr=5.189933e-06 lr_backbone=5.189933e-06 elapsed=50:09:52 avg_step=4.19s recent_step=4.16s eta=08:05:54 +2026-09-22 08:37:10 | INFO | train rank=0 | step=43060/50000 progress=86.1% loss=0.349993 lr=5.175271e-06 lr_backbone=5.175271e-06 elapsed=50:10:33 avg_step=4.19s recent_step=4.17s eta=08:05:12 +2026-09-22 08:37:51 | INFO | train rank=0 | step=43070/50000 progress=86.1% loss=0.340931 lr=5.160630e-06 lr_backbone=5.160630e-06 elapsed=50:11:15 avg_step=4.19s recent_step=4.12s eta=08:04:30 +2026-09-22 08:38:32 | INFO | train rank=0 | step=43080/50000 progress=86.2% loss=0.358748 lr=5.146008e-06 lr_backbone=5.146008e-06 elapsed=50:11:56 avg_step=4.19s recent_step=4.10s eta=08:03:48 +2026-09-22 08:39:14 | INFO | train rank=0 | step=43090/50000 progress=86.2% loss=0.343167 lr=5.131405e-06 lr_backbone=5.131405e-06 elapsed=50:12:37 avg_step=4.19s recent_step=4.15s eta=08:03:06 +2026-09-22 08:39:55 | INFO | train rank=0 | step=43100/50000 progress=86.2% loss=0.389024 lr=5.116822e-06 lr_backbone=5.116822e-06 elapsed=50:13:19 avg_step=4.19s recent_step=4.15s eta=08:02:24 +2026-09-22 08:40:37 | INFO | train rank=0 | step=43110/50000 progress=86.2% loss=0.285782 lr=5.102259e-06 lr_backbone=5.102259e-06 elapsed=50:14:00 avg_step=4.19s recent_step=4.16s eta=08:01:42 +2026-09-22 08:41:18 | INFO | train rank=0 | step=43120/50000 progress=86.2% loss=0.298929 lr=5.087715e-06 lr_backbone=5.087715e-06 elapsed=50:14:42 avg_step=4.19s recent_step=4.15s eta=08:01:00 +2026-09-22 08:42:00 | INFO | train rank=0 | step=43130/50000 progress=86.3% loss=0.326878 lr=5.073191e-06 lr_backbone=5.073191e-06 elapsed=50:15:23 avg_step=4.19s recent_step=4.12s eta=08:00:18 +2026-09-22 08:42:41 | INFO | train rank=0 | step=43140/50000 progress=86.3% loss=0.273459 lr=5.058687e-06 lr_backbone=5.058687e-06 elapsed=50:16:04 avg_step=4.19s recent_step=4.12s eta=07:59:36 +2026-09-22 08:43:22 | INFO | train rank=0 | step=43150/50000 progress=86.3% loss=0.408013 lr=5.044202e-06 lr_backbone=5.044202e-06 elapsed=50:16:46 avg_step=4.19s recent_step=4.16s eta=07:58:54 +2026-09-22 08:44:04 | INFO | train rank=0 | step=43160/50000 progress=86.3% loss=0.258618 lr=5.029737e-06 lr_backbone=5.029737e-06 elapsed=50:17:28 avg_step=4.19s recent_step=4.19s eta=07:58:12 +2026-09-22 08:44:46 | INFO | train rank=0 | step=43170/50000 progress=86.3% loss=0.338386 lr=5.015292e-06 lr_backbone=5.015292e-06 elapsed=50:18:09 avg_step=4.19s recent_step=4.16s eta=07:57:30 +2026-09-22 08:45:28 | INFO | train rank=0 | step=43180/50000 progress=86.4% loss=0.371308 lr=5.000866e-06 lr_backbone=5.000866e-06 elapsed=50:18:51 avg_step=4.19s recent_step=4.19s eta=07:56:48 +2026-09-22 08:46:09 | INFO | train rank=0 | step=43190/50000 progress=86.4% loss=0.258683 lr=4.986461e-06 lr_backbone=4.986461e-06 elapsed=50:19:32 avg_step=4.19s recent_step=4.12s eta=07:56:06 +2026-09-22 08:46:50 | INFO | train rank=0 | step=43200/50000 progress=86.4% loss=0.329666 lr=4.972074e-06 lr_backbone=4.972074e-06 elapsed=50:20:14 avg_step=4.19s recent_step=4.13s eta=07:55:24 +2026-09-22 08:47:32 | INFO | train rank=0 | step=43210/50000 progress=86.4% loss=0.382841 lr=4.957708e-06 lr_backbone=4.957708e-06 elapsed=50:20:55 avg_step=4.19s recent_step=4.15s eta=07:54:42 +2026-09-22 08:48:14 | INFO | train rank=0 | step=43220/50000 progress=86.4% loss=0.385691 lr=4.943361e-06 lr_backbone=4.943361e-06 elapsed=50:21:37 avg_step=4.19s recent_step=4.18s eta=07:54:00 +2026-09-22 08:48:55 | INFO | train rank=0 | step=43230/50000 progress=86.5% loss=0.261410 lr=4.929034e-06 lr_backbone=4.929034e-06 elapsed=50:22:19 avg_step=4.19s recent_step=4.16s eta=07:53:18 +2026-09-22 08:49:37 | INFO | train rank=0 | step=43240/50000 progress=86.5% loss=0.322929 lr=4.914726e-06 lr_backbone=4.914726e-06 elapsed=50:23:00 avg_step=4.19s recent_step=4.18s eta=07:52:36 +2026-09-22 08:50:18 | INFO | train rank=0 | step=43250/50000 progress=86.5% loss=0.374587 lr=4.900438e-06 lr_backbone=4.900438e-06 elapsed=50:23:42 avg_step=4.19s recent_step=4.12s eta=07:51:54 +2026-09-22 08:50:59 | INFO | train rank=0 | step=43260/50000 progress=86.5% loss=0.301982 lr=4.886171e-06 lr_backbone=4.886171e-06 elapsed=50:24:23 avg_step=4.19s recent_step=4.13s eta=07:51:12 +2026-09-22 08:51:41 | INFO | train rank=0 | step=43270/50000 progress=86.5% loss=0.338730 lr=4.871922e-06 lr_backbone=4.871922e-06 elapsed=50:25:04 avg_step=4.19s recent_step=4.16s eta=07:50:30 +2026-09-22 08:52:23 | INFO | train rank=0 | step=43280/50000 progress=86.6% loss=0.283288 lr=4.857694e-06 lr_backbone=4.857694e-06 elapsed=50:25:46 avg_step=4.19s recent_step=4.15s eta=07:49:48 +2026-09-22 08:53:04 | INFO | train rank=0 | step=43290/50000 progress=86.6% loss=0.314509 lr=4.843485e-06 lr_backbone=4.843485e-06 elapsed=50:26:28 avg_step=4.19s recent_step=4.18s eta=07:49:06 +2026-09-22 08:53:46 | INFO | train rank=0 | step=43300/50000 progress=86.6% loss=0.383198 lr=4.829296e-06 lr_backbone=4.829296e-06 elapsed=50:27:10 avg_step=4.19s recent_step=4.19s eta=07:48:24 +2026-09-22 08:54:28 | INFO | train rank=0 | step=43310/50000 progress=86.6% loss=0.298146 lr=4.815127e-06 lr_backbone=4.815127e-06 elapsed=50:27:51 avg_step=4.19s recent_step=4.12s eta=07:47:42 +2026-09-22 08:55:09 | INFO | train rank=0 | step=43320/50000 progress=86.6% loss=0.307508 lr=4.800977e-06 lr_backbone=4.800977e-06 elapsed=50:28:32 avg_step=4.19s recent_step=4.12s eta=07:47:00 +2026-09-22 08:55:51 | INFO | train rank=0 | step=43330/50000 progress=86.7% loss=0.321832 lr=4.786847e-06 lr_backbone=4.786847e-06 elapsed=50:29:14 avg_step=4.19s recent_step=4.17s eta=07:46:18 +2026-09-22 08:56:32 | INFO | train rank=0 | step=43340/50000 progress=86.7% loss=0.322320 lr=4.772738e-06 lr_backbone=4.772738e-06 elapsed=50:29:55 avg_step=4.19s recent_step=4.12s eta=07:45:36 +2026-09-22 08:57:14 | INFO | train rank=0 | step=43350/50000 progress=86.7% loss=0.335198 lr=4.758647e-06 lr_backbone=4.758647e-06 elapsed=50:30:37 avg_step=4.19s recent_step=4.18s eta=07:44:54 +2026-09-22 08:57:55 | INFO | train rank=0 | step=43360/50000 progress=86.7% loss=0.332453 lr=4.744577e-06 lr_backbone=4.744577e-06 elapsed=50:31:19 avg_step=4.19s recent_step=4.18s eta=07:44:12 +2026-09-22 08:58:37 | INFO | train rank=0 | step=43370/50000 progress=86.7% loss=0.270666 lr=4.730526e-06 lr_backbone=4.730526e-06 elapsed=50:32:00 avg_step=4.19s recent_step=4.13s eta=07:43:30 +2026-09-22 08:59:18 | INFO | train rank=0 | step=43380/50000 progress=86.8% loss=0.343167 lr=4.716496e-06 lr_backbone=4.716496e-06 elapsed=50:32:41 avg_step=4.19s recent_step=4.12s eta=07:42:48 +2026-09-22 09:00:00 | INFO | train rank=0 | step=43390/50000 progress=86.8% loss=0.278936 lr=4.702485e-06 lr_backbone=4.702485e-06 elapsed=50:33:23 avg_step=4.19s recent_step=4.17s eta=07:42:06 +2026-09-22 09:00:41 | INFO | train rank=0 | step=43400/50000 progress=86.8% loss=0.348669 lr=4.688494e-06 lr_backbone=4.688494e-06 elapsed=50:34:04 avg_step=4.19s recent_step=4.15s eta=07:41:24 +2026-09-22 09:01:23 | INFO | train rank=0 | step=43410/50000 progress=86.8% loss=0.302245 lr=4.674522e-06 lr_backbone=4.674522e-06 elapsed=50:34:46 avg_step=4.19s recent_step=4.18s eta=07:40:42 +2026-09-22 09:02:05 | INFO | train rank=0 | step=43420/50000 progress=86.8% loss=0.351762 lr=4.660571e-06 lr_backbone=4.660571e-06 elapsed=50:35:28 avg_step=4.19s recent_step=4.16s eta=07:40:00 +2026-09-22 09:02:46 | INFO | train rank=0 | step=43430/50000 progress=86.9% loss=0.300235 lr=4.646639e-06 lr_backbone=4.646639e-06 elapsed=50:36:09 avg_step=4.19s recent_step=4.13s eta=07:39:18 +2026-09-22 09:03:27 | INFO | train rank=0 | step=43440/50000 progress=86.9% loss=0.295988 lr=4.632727e-06 lr_backbone=4.632727e-06 elapsed=50:36:50 avg_step=4.19s recent_step=4.12s eta=07:38:36 +2026-09-22 09:04:09 | INFO | train rank=0 | step=43450/50000 progress=86.9% loss=0.304143 lr=4.618835e-06 lr_backbone=4.618835e-06 elapsed=50:37:32 avg_step=4.19s recent_step=4.18s eta=07:37:54 +2026-09-22 09:04:51 | INFO | train rank=0 | step=43460/50000 progress=86.9% loss=0.331109 lr=4.604963e-06 lr_backbone=4.604963e-06 elapsed=50:38:14 avg_step=4.19s recent_step=4.17s eta=07:37:12 +2026-09-22 09:05:32 | INFO | train rank=0 | step=43470/50000 progress=86.9% loss=0.336567 lr=4.591111e-06 lr_backbone=4.591111e-06 elapsed=50:38:56 avg_step=4.19s recent_step=4.16s eta=07:36:30 +2026-09-22 09:06:14 | INFO | train rank=0 | step=43480/50000 progress=87.0% loss=0.339042 lr=4.577278e-06 lr_backbone=4.577278e-06 elapsed=50:39:37 avg_step=4.19s recent_step=4.17s eta=07:35:48 +2026-09-22 09:06:55 | INFO | train rank=0 | step=43490/50000 progress=87.0% loss=0.294844 lr=4.563466e-06 lr_backbone=4.563466e-06 elapsed=50:40:19 avg_step=4.19s recent_step=4.14s eta=07:35:06 +2026-09-22 09:07:36 | INFO | train rank=0 | step=43500/50000 progress=87.0% loss=0.312982 lr=4.549673e-06 lr_backbone=4.549673e-06 elapsed=50:41:00 avg_step=4.19s recent_step=4.13s eta=07:34:24 +2026-09-22 09:08:18 | INFO | train rank=0 | step=43510/50000 progress=87.0% loss=0.315230 lr=4.535900e-06 lr_backbone=4.535900e-06 elapsed=50:41:41 avg_step=4.19s recent_step=4.16s eta=07:33:42 +2026-09-22 09:09:00 | INFO | train rank=0 | step=43520/50000 progress=87.0% loss=0.306546 lr=4.522148e-06 lr_backbone=4.522148e-06 elapsed=50:42:23 avg_step=4.19s recent_step=4.17s eta=07:33:00 +2026-09-22 09:09:41 | INFO | train rank=0 | step=43530/50000 progress=87.1% loss=0.323967 lr=4.508415e-06 lr_backbone=4.508415e-06 elapsed=50:43:05 avg_step=4.19s recent_step=4.16s eta=07:32:18 +2026-09-22 09:10:23 | INFO | train rank=0 | step=43540/50000 progress=87.1% loss=0.297295 lr=4.494701e-06 lr_backbone=4.494701e-06 elapsed=50:43:46 avg_step=4.19s recent_step=4.12s eta=07:31:36 +2026-09-22 09:11:04 | INFO | train rank=0 | step=43550/50000 progress=87.1% loss=0.322494 lr=4.481008e-06 lr_backbone=4.481008e-06 elapsed=50:44:27 avg_step=4.19s recent_step=4.15s eta=07:30:54 +2026-09-22 09:11:45 | INFO | train rank=0 | step=43560/50000 progress=87.1% loss=0.348274 lr=4.467335e-06 lr_backbone=4.467335e-06 elapsed=50:45:09 avg_step=4.19s recent_step=4.13s eta=07:30:12 +2026-09-22 09:12:27 | INFO | train rank=0 | step=43570/50000 progress=87.1% loss=0.330752 lr=4.453682e-06 lr_backbone=4.453682e-06 elapsed=50:45:51 avg_step=4.19s recent_step=4.19s eta=07:29:30 +2026-09-22 09:13:09 | INFO | train rank=0 | step=43580/50000 progress=87.2% loss=0.262592 lr=4.440048e-06 lr_backbone=4.440048e-06 elapsed=50:46:33 avg_step=4.19s recent_step=4.19s eta=07:28:48 +2026-09-22 09:13:51 | INFO | train rank=0 | step=43590/50000 progress=87.2% loss=0.267917 lr=4.426435e-06 lr_backbone=4.426435e-06 elapsed=50:47:14 avg_step=4.19s recent_step=4.16s eta=07:28:06 +2026-09-22 09:14:32 | INFO | train rank=0 | step=43600/50000 progress=87.2% loss=0.315674 lr=4.412841e-06 lr_backbone=4.412841e-06 elapsed=50:47:56 avg_step=4.19s recent_step=4.17s eta=07:27:24 +2026-09-22 09:15:14 | INFO | train rank=0 | step=43610/50000 progress=87.2% loss=0.364999 lr=4.399267e-06 lr_backbone=4.399267e-06 elapsed=50:48:37 avg_step=4.19s recent_step=4.12s eta=07:26:42 +2026-09-22 09:15:55 | INFO | train rank=0 | step=43620/50000 progress=87.2% loss=0.326646 lr=4.385714e-06 lr_backbone=4.385714e-06 elapsed=50:49:18 avg_step=4.19s recent_step=4.12s eta=07:26:00 +2026-09-22 09:16:37 | INFO | train rank=0 | step=43630/50000 progress=87.3% loss=0.315918 lr=4.372180e-06 lr_backbone=4.372180e-06 elapsed=50:50:00 avg_step=4.19s recent_step=4.18s eta=07:25:18 +2026-09-22 09:17:18 | INFO | train rank=0 | step=43640/50000 progress=87.3% loss=0.333064 lr=4.358666e-06 lr_backbone=4.358666e-06 elapsed=50:50:42 avg_step=4.19s recent_step=4.17s eta=07:24:36 +2026-09-22 09:18:00 | INFO | train rank=0 | step=43650/50000 progress=87.3% loss=0.348411 lr=4.345172e-06 lr_backbone=4.345172e-06 elapsed=50:51:23 avg_step=4.19s recent_step=4.15s eta=07:23:54 +2026-09-22 09:18:41 | INFO | train rank=0 | step=43660/50000 progress=87.3% loss=0.383100 lr=4.331699e-06 lr_backbone=4.331699e-06 elapsed=50:52:05 avg_step=4.19s recent_step=4.15s eta=07:23:12 +2026-09-22 09:19:23 | INFO | train rank=0 | step=43670/50000 progress=87.3% loss=0.327689 lr=4.318245e-06 lr_backbone=4.318245e-06 elapsed=50:52:46 avg_step=4.19s recent_step=4.13s eta=07:22:30 +2026-09-22 09:20:04 | INFO | train rank=0 | step=43680/50000 progress=87.4% loss=0.281389 lr=4.304811e-06 lr_backbone=4.304811e-06 elapsed=50:53:27 avg_step=4.19s recent_step=4.12s eta=07:21:48 +2026-09-22 09:20:46 | INFO | train rank=0 | step=43690/50000 progress=87.4% loss=0.415856 lr=4.291397e-06 lr_backbone=4.291397e-06 elapsed=50:54:09 avg_step=4.19s recent_step=4.19s eta=07:21:06 +2026-09-22 09:21:28 | INFO | train rank=0 | step=43700/50000 progress=87.4% loss=0.429956 lr=4.278003e-06 lr_backbone=4.278003e-06 elapsed=50:54:51 avg_step=4.19s recent_step=4.18s eta=07:20:24 +2026-09-22 09:22:09 | INFO | train rank=0 | step=43710/50000 progress=87.4% loss=0.259882 lr=4.264629e-06 lr_backbone=4.264629e-06 elapsed=50:55:32 avg_step=4.19s recent_step=4.15s eta=07:19:42 +2026-09-22 09:22:51 | INFO | train rank=0 | step=43720/50000 progress=87.4% loss=0.353388 lr=4.251275e-06 lr_backbone=4.251275e-06 elapsed=50:56:14 avg_step=4.19s recent_step=4.19s eta=07:19:00 +2026-09-22 09:23:32 | INFO | train rank=0 | step=43730/50000 progress=87.5% loss=0.380256 lr=4.237941e-06 lr_backbone=4.237941e-06 elapsed=50:56:56 avg_step=4.19s recent_step=4.14s eta=07:18:18 +2026-09-22 09:24:14 | INFO | train rank=0 | step=43740/50000 progress=87.5% loss=0.309996 lr=4.224628e-06 lr_backbone=4.224628e-06 elapsed=50:57:37 avg_step=4.19s recent_step=4.13s eta=07:17:36 +2026-09-22 09:24:55 | INFO | train rank=0 | step=43750/50000 progress=87.5% loss=0.331356 lr=4.211334e-06 lr_backbone=4.211334e-06 elapsed=50:58:19 avg_step=4.19s recent_step=4.16s eta=07:16:54 +2026-09-22 09:25:37 | INFO | train rank=0 | step=43760/50000 progress=87.5% loss=0.311351 lr=4.198060e-06 lr_backbone=4.198060e-06 elapsed=50:59:00 avg_step=4.19s recent_step=4.17s eta=07:16:12 +2026-09-22 09:26:19 | INFO | train rank=0 | step=43770/50000 progress=87.5% loss=0.362000 lr=4.184806e-06 lr_backbone=4.184806e-06 elapsed=50:59:42 avg_step=4.19s recent_step=4.18s eta=07:15:30 +2026-09-22 09:27:00 | INFO | train rank=0 | step=43780/50000 progress=87.6% loss=0.315065 lr=4.171572e-06 lr_backbone=4.171572e-06 elapsed=51:00:24 avg_step=4.19s recent_step=4.16s eta=07:14:48 +2026-09-22 09:27:42 | INFO | train rank=0 | step=43790/50000 progress=87.6% loss=0.436157 lr=4.158359e-06 lr_backbone=4.158359e-06 elapsed=51:01:05 avg_step=4.19s recent_step=4.11s eta=07:14:06 +2026-09-22 09:28:23 | INFO | train rank=0 | step=43800/50000 progress=87.6% loss=0.295193 lr=4.145165e-06 lr_backbone=4.145165e-06 elapsed=51:01:46 avg_step=4.19s recent_step=4.12s eta=07:13:24 +2026-09-22 09:29:05 | INFO | train rank=0 | step=43810/50000 progress=87.6% loss=0.344191 lr=4.131991e-06 lr_backbone=4.131991e-06 elapsed=51:02:28 avg_step=4.19s recent_step=4.18s eta=07:12:42 +2026-09-22 09:29:47 | INFO | train rank=0 | step=43820/50000 progress=87.6% loss=0.280436 lr=4.118838e-06 lr_backbone=4.118838e-06 elapsed=51:03:10 avg_step=4.19s recent_step=4.19s eta=07:12:00 +2026-09-22 09:30:28 | INFO | train rank=0 | step=43830/50000 progress=87.7% loss=0.302198 lr=4.105705e-06 lr_backbone=4.105705e-06 elapsed=51:03:52 avg_step=4.19s recent_step=4.18s eta=07:11:18 +2026-09-22 09:31:10 | INFO | train rank=0 | step=43840/50000 progress=87.7% loss=0.364655 lr=4.092591e-06 lr_backbone=4.092591e-06 elapsed=51:04:33 avg_step=4.19s recent_step=4.15s eta=07:10:36 +2026-09-22 09:31:51 | INFO | train rank=0 | step=43850/50000 progress=87.7% loss=0.282719 lr=4.079498e-06 lr_backbone=4.079498e-06 elapsed=51:05:15 avg_step=4.19s recent_step=4.14s eta=07:09:54 +2026-09-22 09:32:33 | INFO | train rank=0 | step=43860/50000 progress=87.7% loss=0.303091 lr=4.066425e-06 lr_backbone=4.066425e-06 elapsed=51:05:56 avg_step=4.19s recent_step=4.13s eta=07:09:12 +2026-09-22 09:33:14 | INFO | train rank=0 | step=43870/50000 progress=87.7% loss=0.287705 lr=4.053372e-06 lr_backbone=4.053372e-06 elapsed=51:06:38 avg_step=4.19s recent_step=4.17s eta=07:08:30 +2026-09-22 09:33:56 | INFO | train rank=0 | step=43880/50000 progress=87.8% loss=0.336389 lr=4.040339e-06 lr_backbone=4.040339e-06 elapsed=51:07:19 avg_step=4.19s recent_step=4.18s eta=07:07:48 +2026-09-22 09:34:38 | INFO | train rank=0 | step=43890/50000 progress=87.8% loss=0.313091 lr=4.027326e-06 lr_backbone=4.027326e-06 elapsed=51:08:01 avg_step=4.19s recent_step=4.18s eta=07:07:06 +2026-09-22 09:35:19 | INFO | train rank=0 | step=43900/50000 progress=87.8% loss=0.439432 lr=4.014333e-06 lr_backbone=4.014333e-06 elapsed=51:08:43 avg_step=4.19s recent_step=4.15s eta=07:06:24 +2026-09-22 09:36:01 | INFO | train rank=0 | step=43910/50000 progress=87.8% loss=0.307477 lr=4.001360e-06 lr_backbone=4.001360e-06 elapsed=51:09:24 avg_step=4.19s recent_step=4.14s eta=07:05:42 +2026-09-22 09:36:42 | INFO | train rank=0 | step=43920/50000 progress=87.8% loss=0.352513 lr=3.988408e-06 lr_backbone=3.988408e-06 elapsed=51:10:05 avg_step=4.19s recent_step=4.12s eta=07:05:00 +2026-09-22 09:37:24 | INFO | train rank=0 | step=43930/50000 progress=87.9% loss=0.241626 lr=3.975475e-06 lr_backbone=3.975475e-06 elapsed=51:10:47 avg_step=4.19s recent_step=4.17s eta=07:04:18 +2026-09-22 09:38:05 | INFO | train rank=0 | step=43940/50000 progress=87.9% loss=0.332115 lr=3.962563e-06 lr_backbone=3.962563e-06 elapsed=51:11:28 avg_step=4.19s recent_step=4.14s eta=07:03:36 +2026-09-22 09:38:46 | INFO | train rank=0 | step=43950/50000 progress=87.9% loss=0.289455 lr=3.949671e-06 lr_backbone=3.949671e-06 elapsed=51:12:10 avg_step=4.19s recent_step=4.15s eta=07:02:54 +2026-09-22 09:39:28 | INFO | train rank=0 | step=43960/50000 progress=87.9% loss=0.337827 lr=3.936799e-06 lr_backbone=3.936799e-06 elapsed=51:12:51 avg_step=4.19s recent_step=4.16s eta=07:02:12 +2026-09-22 09:40:09 | INFO | train rank=0 | step=43970/50000 progress=87.9% loss=0.328475 lr=3.923947e-06 lr_backbone=3.923947e-06 elapsed=51:13:32 avg_step=4.19s recent_step=4.09s eta=07:01:30 +2026-09-22 09:40:50 | INFO | train rank=0 | step=43980/50000 progress=88.0% loss=0.324566 lr=3.911115e-06 lr_backbone=3.911115e-06 elapsed=51:14:14 avg_step=4.19s recent_step=4.12s eta=07:00:48 +2026-09-22 09:41:32 | INFO | train rank=0 | step=43990/50000 progress=88.0% loss=0.290460 lr=3.898304e-06 lr_backbone=3.898304e-06 elapsed=51:14:55 avg_step=4.19s recent_step=4.15s eta=07:00:06 +2026-09-22 09:42:13 | INFO | train rank=0 | step=44000/50000 progress=88.0% loss=0.314929 lr=3.885512e-06 lr_backbone=3.885512e-06 elapsed=51:15:37 avg_step=4.19s recent_step=4.15s eta=06:59:24 +2026-09-22 09:42:55 | INFO | train rank=0 | step=44010/50000 progress=88.0% loss=0.256446 lr=3.872741e-06 lr_backbone=3.872741e-06 elapsed=51:16:18 avg_step=4.19s recent_step=4.16s eta=06:58:42 +2026-09-22 09:43:36 | INFO | train rank=0 | step=44020/50000 progress=88.0% loss=0.364823 lr=3.859990e-06 lr_backbone=3.859990e-06 elapsed=51:16:59 avg_step=4.19s recent_step=4.13s eta=06:58:00 +2026-09-22 09:44:17 | INFO | train rank=0 | step=44030/50000 progress=88.1% loss=0.322877 lr=3.847259e-06 lr_backbone=3.847259e-06 elapsed=51:17:40 avg_step=4.19s recent_step=4.10s eta=06:57:18 +2026-09-22 09:44:58 | INFO | train rank=0 | step=44040/50000 progress=88.1% loss=0.369360 lr=3.834549e-06 lr_backbone=3.834549e-06 elapsed=51:18:22 avg_step=4.19s recent_step=4.11s eta=06:56:36 +2026-09-22 09:45:40 | INFO | train rank=0 | step=44050/50000 progress=88.1% loss=0.301201 lr=3.821858e-06 lr_backbone=3.821858e-06 elapsed=51:19:03 avg_step=4.19s recent_step=4.16s eta=06:55:54 +2026-09-22 09:46:21 | INFO | train rank=0 | step=44060/50000 progress=88.1% loss=0.343663 lr=3.809188e-06 lr_backbone=3.809188e-06 elapsed=51:19:45 avg_step=4.19s recent_step=4.15s eta=06:55:12 +2026-09-22 09:47:03 | INFO | train rank=0 | step=44070/50000 progress=88.1% loss=0.314620 lr=3.796538e-06 lr_backbone=3.796538e-06 elapsed=51:20:26 avg_step=4.19s recent_step=4.16s eta=06:54:30 +2026-09-22 09:47:44 | INFO | train rank=0 | step=44080/50000 progress=88.2% loss=0.367691 lr=3.783908e-06 lr_backbone=3.783908e-06 elapsed=51:21:08 avg_step=4.19s recent_step=4.14s eta=06:53:48 +2026-09-22 09:48:25 | INFO | train rank=0 | step=44090/50000 progress=88.2% loss=0.287097 lr=3.771298e-06 lr_backbone=3.771298e-06 elapsed=51:21:49 avg_step=4.19s recent_step=4.11s eta=06:53:05 +2026-09-22 09:49:06 | INFO | train rank=0 | step=44100/50000 progress=88.2% loss=0.360704 lr=3.758709e-06 lr_backbone=3.758709e-06 elapsed=51:22:30 avg_step=4.19s recent_step=4.11s eta=06:52:23 +2026-09-22 09:49:48 | INFO | train rank=0 | step=44110/50000 progress=88.2% loss=0.285672 lr=3.746140e-06 lr_backbone=3.746140e-06 elapsed=51:23:11 avg_step=4.19s recent_step=4.17s eta=06:51:41 +2026-09-22 09:50:30 | INFO | train rank=0 | step=44120/50000 progress=88.2% loss=0.428879 lr=3.733591e-06 lr_backbone=3.733591e-06 elapsed=51:23:53 avg_step=4.19s recent_step=4.15s eta=06:50:59 +2026-09-22 09:51:11 | INFO | train rank=0 | step=44130/50000 progress=88.3% loss=0.274555 lr=3.721062e-06 lr_backbone=3.721062e-06 elapsed=51:24:34 avg_step=4.19s recent_step=4.16s eta=06:50:17 +2026-09-22 09:51:53 | INFO | train rank=0 | step=44140/50000 progress=88.3% loss=0.364722 lr=3.708554e-06 lr_backbone=3.708554e-06 elapsed=51:25:16 avg_step=4.19s recent_step=4.16s eta=06:49:35 +2026-09-22 09:52:34 | INFO | train rank=0 | step=44150/50000 progress=88.3% loss=0.292122 lr=3.696065e-06 lr_backbone=3.696065e-06 elapsed=51:25:57 avg_step=4.19s recent_step=4.11s eta=06:48:53 +2026-09-22 09:53:15 | INFO | train rank=0 | step=44160/50000 progress=88.3% loss=0.308603 lr=3.683597e-06 lr_backbone=3.683597e-06 elapsed=51:26:38 avg_step=4.19s recent_step=4.10s eta=06:48:11 +2026-09-22 09:53:57 | INFO | train rank=0 | step=44170/50000 progress=88.3% loss=0.352647 lr=3.671150e-06 lr_backbone=3.671150e-06 elapsed=51:27:20 avg_step=4.19s recent_step=4.18s eta=06:47:29 +2026-09-22 09:54:38 | INFO | train rank=0 | step=44180/50000 progress=88.4% loss=0.301681 lr=3.658722e-06 lr_backbone=3.658722e-06 elapsed=51:28:01 avg_step=4.19s recent_step=4.16s eta=06:46:47 +2026-09-22 09:55:20 | INFO | train rank=0 | step=44190/50000 progress=88.4% loss=0.293945 lr=3.646315e-06 lr_backbone=3.646315e-06 elapsed=51:28:43 avg_step=4.19s recent_step=4.16s eta=06:46:05 +2026-09-22 09:56:01 | INFO | train rank=0 | step=44200/50000 progress=88.4% loss=0.341908 lr=3.633928e-06 lr_backbone=3.633928e-06 elapsed=51:29:24 avg_step=4.19s recent_step=4.13s eta=06:45:23 +2026-09-22 09:56:42 | INFO | train rank=0 | step=44210/50000 progress=88.4% loss=0.378367 lr=3.621562e-06 lr_backbone=3.621562e-06 elapsed=51:30:05 avg_step=4.19s recent_step=4.10s eta=06:44:41 +2026-09-22 09:57:23 | INFO | train rank=0 | step=44220/50000 progress=88.4% loss=0.370875 lr=3.609215e-06 lr_backbone=3.609215e-06 elapsed=51:30:46 avg_step=4.19s recent_step=4.10s eta=06:43:59 +2026-09-22 09:58:05 | INFO | train rank=0 | step=44230/50000 progress=88.5% loss=0.353545 lr=3.596889e-06 lr_backbone=3.596889e-06 elapsed=51:31:28 avg_step=4.19s recent_step=4.15s eta=06:43:17 +2026-09-22 09:58:46 | INFO | train rank=0 | step=44240/50000 progress=88.5% loss=0.270045 lr=3.584583e-06 lr_backbone=3.584583e-06 elapsed=51:32:09 avg_step=4.19s recent_step=4.15s eta=06:42:35 +2026-09-22 09:59:28 | INFO | train rank=0 | step=44250/50000 progress=88.5% loss=0.365941 lr=3.572298e-06 lr_backbone=3.572298e-06 elapsed=51:32:51 avg_step=4.19s recent_step=4.17s eta=06:41:53 +2026-09-22 10:00:09 | INFO | train rank=0 | step=44260/50000 progress=88.5% loss=0.287955 lr=3.560033e-06 lr_backbone=3.560033e-06 elapsed=51:33:33 avg_step=4.19s recent_step=4.16s eta=06:41:11 +2026-09-22 10:00:51 | INFO | train rank=0 | step=44270/50000 progress=88.5% loss=0.348656 lr=3.547788e-06 lr_backbone=3.547788e-06 elapsed=51:34:14 avg_step=4.19s recent_step=4.12s eta=06:40:29 +2026-09-22 10:01:32 | INFO | train rank=0 | step=44280/50000 progress=88.6% loss=0.286991 lr=3.535564e-06 lr_backbone=3.535564e-06 elapsed=51:34:55 avg_step=4.19s recent_step=4.11s eta=06:39:47 +2026-09-22 10:02:13 | INFO | train rank=0 | step=44290/50000 progress=88.6% loss=0.327252 lr=3.523360e-06 lr_backbone=3.523360e-06 elapsed=51:35:36 avg_step=4.19s recent_step=4.14s eta=06:39:05 +2026-09-22 10:02:55 | INFO | train rank=0 | step=44300/50000 progress=88.6% loss=0.328804 lr=3.511176e-06 lr_backbone=3.511176e-06 elapsed=51:36:18 avg_step=4.19s recent_step=4.17s eta=06:38:23 +2026-09-22 10:03:36 | INFO | train rank=0 | step=44310/50000 progress=88.6% loss=0.333879 lr=3.499012e-06 lr_backbone=3.499012e-06 elapsed=51:37:00 avg_step=4.19s recent_step=4.16s eta=06:37:41 +2026-09-22 10:04:18 | INFO | train rank=0 | step=44320/50000 progress=88.6% loss=0.358234 lr=3.486869e-06 lr_backbone=3.486869e-06 elapsed=51:37:41 avg_step=4.19s recent_step=4.17s eta=06:36:59 +2026-09-22 10:04:59 | INFO | train rank=0 | step=44330/50000 progress=88.7% loss=0.345239 lr=3.474746e-06 lr_backbone=3.474746e-06 elapsed=51:38:23 avg_step=4.19s recent_step=4.12s eta=06:36:17 +2026-09-22 10:05:40 | INFO | train rank=0 | step=44340/50000 progress=88.7% loss=0.306356 lr=3.462644e-06 lr_backbone=3.462644e-06 elapsed=51:39:04 avg_step=4.19s recent_step=4.11s eta=06:35:35 +2026-09-22 10:06:22 | INFO | train rank=0 | step=44350/50000 progress=88.7% loss=0.278638 lr=3.450562e-06 lr_backbone=3.450562e-06 elapsed=51:39:45 avg_step=4.19s recent_step=4.16s eta=06:34:53 +2026-09-22 10:07:04 | INFO | train rank=0 | step=44360/50000 progress=88.7% loss=0.313158 lr=3.438500e-06 lr_backbone=3.438500e-06 elapsed=51:40:27 avg_step=4.19s recent_step=4.17s eta=06:34:11 +2026-09-22 10:07:45 | INFO | train rank=0 | step=44370/50000 progress=88.7% loss=0.344094 lr=3.426459e-06 lr_backbone=3.426459e-06 elapsed=51:41:09 avg_step=4.19s recent_step=4.16s eta=06:33:29 +2026-09-22 10:08:27 | INFO | train rank=0 | step=44380/50000 progress=88.8% loss=0.393171 lr=3.414438e-06 lr_backbone=3.414438e-06 elapsed=51:41:50 avg_step=4.19s recent_step=4.16s eta=06:32:47 +2026-09-22 10:09:08 | INFO | train rank=0 | step=44390/50000 progress=88.8% loss=0.418033 lr=3.402437e-06 lr_backbone=3.402437e-06 elapsed=51:42:31 avg_step=4.19s recent_step=4.10s eta=06:32:05 +2026-09-22 10:09:49 | INFO | train rank=0 | step=44400/50000 progress=88.8% loss=0.306954 lr=3.390457e-06 lr_backbone=3.390457e-06 elapsed=51:43:12 avg_step=4.19s recent_step=4.12s eta=06:31:23 +2026-09-22 10:10:31 | INFO | train rank=0 | step=44410/50000 progress=88.8% loss=0.315340 lr=3.378497e-06 lr_backbone=3.378497e-06 elapsed=51:43:54 avg_step=4.19s recent_step=4.15s eta=06:30:41 +2026-09-22 10:11:12 | INFO | train rank=0 | step=44420/50000 progress=88.8% loss=0.275954 lr=3.366558e-06 lr_backbone=3.366558e-06 elapsed=51:44:36 avg_step=4.19s recent_step=4.17s eta=06:29:59 +2026-09-22 10:11:54 | INFO | train rank=0 | step=44430/50000 progress=88.9% loss=0.347473 lr=3.354638e-06 lr_backbone=3.354638e-06 elapsed=51:45:17 avg_step=4.19s recent_step=4.18s eta=06:29:17 +2026-09-22 10:12:36 | INFO | train rank=0 | step=44440/50000 progress=88.9% loss=0.282714 lr=3.342740e-06 lr_backbone=3.342740e-06 elapsed=51:45:59 avg_step=4.19s recent_step=4.17s eta=06:28:35 +2026-09-22 10:13:17 | INFO | train rank=0 | step=44450/50000 progress=88.9% loss=0.253801 lr=3.330862e-06 lr_backbone=3.330862e-06 elapsed=51:46:40 avg_step=4.19s recent_step=4.13s eta=06:27:53 +2026-09-22 10:13:58 | INFO | train rank=0 | step=44460/50000 progress=88.9% loss=0.344314 lr=3.319004e-06 lr_backbone=3.319004e-06 elapsed=51:47:22 avg_step=4.19s recent_step=4.13s eta=06:27:11 +2026-09-22 10:14:40 | INFO | train rank=0 | step=44470/50000 progress=88.9% loss=0.378566 lr=3.307166e-06 lr_backbone=3.307166e-06 elapsed=51:48:03 avg_step=4.19s recent_step=4.16s eta=06:26:29 +2026-09-22 10:15:22 | INFO | train rank=0 | step=44480/50000 progress=89.0% loss=0.332039 lr=3.295349e-06 lr_backbone=3.295349e-06 elapsed=51:48:45 avg_step=4.19s recent_step=4.18s eta=06:25:47 +2026-09-22 10:16:04 | INFO | train rank=0 | step=44490/50000 progress=89.0% loss=0.277080 lr=3.283553e-06 lr_backbone=3.283553e-06 elapsed=51:49:27 avg_step=4.19s recent_step=4.20s eta=06:25:06 +2026-09-22 10:16:45 | INFO | train rank=0 | step=44500/50000 progress=89.0% loss=0.266147 lr=3.271777e-06 lr_backbone=3.271777e-06 elapsed=51:50:09 avg_step=4.19s recent_step=4.17s eta=06:24:24 +2026-09-22 10:17:27 | INFO | train rank=0 | step=44510/50000 progress=89.0% loss=0.338419 lr=3.260021e-06 lr_backbone=3.260021e-06 elapsed=51:50:50 avg_step=4.19s recent_step=4.13s eta=06:23:42 +2026-09-22 10:18:08 | INFO | train rank=0 | step=44520/50000 progress=89.0% loss=0.298334 lr=3.248286e-06 lr_backbone=3.248286e-06 elapsed=51:51:31 avg_step=4.19s recent_step=4.13s eta=06:23:00 +2026-09-22 10:18:50 | INFO | train rank=0 | step=44530/50000 progress=89.1% loss=0.333475 lr=3.236571e-06 lr_backbone=3.236571e-06 elapsed=51:52:13 avg_step=4.19s recent_step=4.16s eta=06:22:18 +2026-09-22 10:19:31 | INFO | train rank=0 | step=44540/50000 progress=89.1% loss=0.390625 lr=3.224877e-06 lr_backbone=3.224877e-06 elapsed=51:52:55 avg_step=4.19s recent_step=4.18s eta=06:21:36 +2026-09-22 10:20:13 | INFO | train rank=0 | step=44550/50000 progress=89.1% loss=0.325684 lr=3.213203e-06 lr_backbone=3.213203e-06 elapsed=51:53:36 avg_step=4.19s recent_step=4.17s eta=06:20:54 +2026-09-22 10:20:55 | INFO | train rank=0 | step=44560/50000 progress=89.1% loss=0.266985 lr=3.201550e-06 lr_backbone=3.201550e-06 elapsed=51:54:18 avg_step=4.19s recent_step=4.18s eta=06:20:12 +2026-09-22 10:21:36 | INFO | train rank=0 | step=44570/50000 progress=89.1% loss=0.324410 lr=3.189917e-06 lr_backbone=3.189917e-06 elapsed=51:54:59 avg_step=4.19s recent_step=4.11s eta=06:19:30 +2026-09-22 10:22:17 | INFO | train rank=0 | step=44580/50000 progress=89.2% loss=0.357159 lr=3.178304e-06 lr_backbone=3.178304e-06 elapsed=51:55:41 avg_step=4.19s recent_step=4.14s eta=06:18:48 +2026-09-22 10:22:59 | INFO | train rank=0 | step=44590/50000 progress=89.2% loss=0.253754 lr=3.166712e-06 lr_backbone=3.166712e-06 elapsed=51:56:22 avg_step=4.19s recent_step=4.17s eta=06:18:06 +2026-09-22 10:23:40 | INFO | train rank=0 | step=44600/50000 progress=89.2% loss=0.346917 lr=3.155141e-06 lr_backbone=3.155141e-06 elapsed=51:57:04 avg_step=4.19s recent_step=4.14s eta=06:17:24 +2026-09-22 10:24:22 | INFO | train rank=0 | step=44610/50000 progress=89.2% loss=0.333421 lr=3.143590e-06 lr_backbone=3.143590e-06 elapsed=51:57:46 avg_step=4.19s recent_step=4.18s eta=06:16:42 +2026-09-22 10:25:04 | INFO | train rank=0 | step=44620/50000 progress=89.2% loss=0.293688 lr=3.132059e-06 lr_backbone=3.132059e-06 elapsed=51:58:27 avg_step=4.19s recent_step=4.16s eta=06:16:00 +2026-09-22 10:25:45 | INFO | train rank=0 | step=44630/50000 progress=89.3% loss=0.329306 lr=3.120549e-06 lr_backbone=3.120549e-06 elapsed=51:59:08 avg_step=4.19s recent_step=4.12s eta=06:15:18 +2026-09-22 10:26:26 | INFO | train rank=0 | step=44640/50000 progress=89.3% loss=0.411215 lr=3.109060e-06 lr_backbone=3.109060e-06 elapsed=51:59:50 avg_step=4.19s recent_step=4.13s eta=06:14:36 +2026-09-22 10:27:08 | INFO | train rank=0 | step=44650/50000 progress=89.3% loss=0.367643 lr=3.097591e-06 lr_backbone=3.097591e-06 elapsed=52:00:31 avg_step=4.19s recent_step=4.18s eta=06:13:54 +2026-09-22 10:27:50 | INFO | train rank=0 | step=44660/50000 progress=89.3% loss=0.287990 lr=3.086142e-06 lr_backbone=3.086142e-06 elapsed=52:01:13 avg_step=4.19s recent_step=4.16s eta=06:13:12 +2026-09-22 10:28:31 | INFO | train rank=0 | step=44670/50000 progress=89.3% loss=0.339720 lr=3.074715e-06 lr_backbone=3.074715e-06 elapsed=52:01:55 avg_step=4.19s recent_step=4.18s eta=06:12:30 +2026-09-22 10:29:13 | INFO | train rank=0 | step=44680/50000 progress=89.4% loss=0.393892 lr=3.063307e-06 lr_backbone=3.063307e-06 elapsed=52:02:37 avg_step=4.19s recent_step=4.17s eta=06:11:48 +2026-09-22 10:29:54 | INFO | train rank=0 | step=44690/50000 progress=89.4% loss=0.282544 lr=3.051920e-06 lr_backbone=3.051920e-06 elapsed=52:03:18 avg_step=4.19s recent_step=4.13s eta=06:11:06 +2026-09-22 10:30:36 | INFO | train rank=0 | step=44700/50000 progress=89.4% loss=0.371296 lr=3.040554e-06 lr_backbone=3.040554e-06 elapsed=52:03:59 avg_step=4.19s recent_step=4.13s eta=06:10:24 +2026-09-22 10:31:18 | INFO | train rank=0 | step=44710/50000 progress=89.4% loss=0.326894 lr=3.029208e-06 lr_backbone=3.029208e-06 elapsed=52:04:41 avg_step=4.19s recent_step=4.18s eta=06:09:42 +2026-09-22 10:31:59 | INFO | train rank=0 | step=44720/50000 progress=89.4% loss=0.298626 lr=3.017883e-06 lr_backbone=3.017883e-06 elapsed=52:05:22 avg_step=4.19s recent_step=4.15s eta=06:09:00 +2026-09-22 10:32:41 | INFO | train rank=0 | step=44730/50000 progress=89.5% loss=0.345136 lr=3.006578e-06 lr_backbone=3.006578e-06 elapsed=52:06:04 avg_step=4.19s recent_step=4.17s eta=06:08:18 +2026-09-22 10:33:23 | INFO | train rank=0 | step=44740/50000 progress=89.5% loss=0.312324 lr=2.995294e-06 lr_backbone=2.995294e-06 elapsed=52:06:46 avg_step=4.19s recent_step=4.18s eta=06:07:36 +2026-09-22 10:34:04 | INFO | train rank=0 | step=44750/50000 progress=89.5% loss=0.337935 lr=2.984030e-06 lr_backbone=2.984030e-06 elapsed=52:07:27 avg_step=4.19s recent_step=4.10s eta=06:06:54 +2026-09-22 10:34:45 | INFO | train rank=0 | step=44760/50000 progress=89.5% loss=0.385977 lr=2.972787e-06 lr_backbone=2.972787e-06 elapsed=52:08:08 avg_step=4.19s recent_step=4.13s eta=06:06:12 +2026-09-22 10:35:27 | INFO | train rank=0 | step=44770/50000 progress=89.5% loss=0.277683 lr=2.961565e-06 lr_backbone=2.961565e-06 elapsed=52:08:50 avg_step=4.19s recent_step=4.18s eta=06:05:30 +2026-09-22 10:36:08 | INFO | train rank=0 | step=44780/50000 progress=89.6% loss=0.286137 lr=2.950363e-06 lr_backbone=2.950363e-06 elapsed=52:09:32 avg_step=4.19s recent_step=4.16s eta=06:04:48 +2026-09-22 10:36:50 | INFO | train rank=0 | step=44790/50000 progress=89.6% loss=0.274287 lr=2.939182e-06 lr_backbone=2.939182e-06 elapsed=52:10:13 avg_step=4.19s recent_step=4.15s eta=06:04:06 +2026-09-22 10:37:32 | INFO | train rank=0 | step=44800/50000 progress=89.6% loss=0.333139 lr=2.928021e-06 lr_backbone=2.928021e-06 elapsed=52:10:55 avg_step=4.19s recent_step=4.18s eta=06:03:24 +2026-09-22 10:38:13 | INFO | train rank=0 | step=44810/50000 progress=89.6% loss=0.400230 lr=2.916881e-06 lr_backbone=2.916881e-06 elapsed=52:11:36 avg_step=4.19s recent_step=4.14s eta=06:02:42 +2026-09-22 10:38:54 | INFO | train rank=0 | step=44820/50000 progress=89.6% loss=0.322841 lr=2.905762e-06 lr_backbone=2.905762e-06 elapsed=52:12:18 avg_step=4.19s recent_step=4.13s eta=06:02:00 +2026-09-22 10:39:36 | INFO | train rank=0 | step=44830/50000 progress=89.7% loss=0.367525 lr=2.894663e-06 lr_backbone=2.894663e-06 elapsed=52:13:00 avg_step=4.19s recent_step=4.19s eta=06:01:18 +2026-09-22 10:40:18 | INFO | train rank=0 | step=44840/50000 progress=89.7% loss=0.371376 lr=2.883584e-06 lr_backbone=2.883584e-06 elapsed=52:13:41 avg_step=4.19s recent_step=4.18s eta=06:00:36 +2026-09-22 10:41:00 | INFO | train rank=0 | step=44850/50000 progress=89.7% loss=0.280251 lr=2.872527e-06 lr_backbone=2.872527e-06 elapsed=52:14:23 avg_step=4.19s recent_step=4.16s eta=05:59:54 +2026-09-22 10:41:41 | INFO | train rank=0 | step=44860/50000 progress=89.7% loss=0.286325 lr=2.861490e-06 lr_backbone=2.861490e-06 elapsed=52:15:05 avg_step=4.19s recent_step=4.17s eta=05:59:12 +2026-09-22 10:42:22 | INFO | train rank=0 | step=44870/50000 progress=89.7% loss=0.298530 lr=2.850473e-06 lr_backbone=2.850473e-06 elapsed=52:15:46 avg_step=4.19s recent_step=4.11s eta=05:58:30 +2026-09-22 10:43:04 | INFO | train rank=0 | step=44880/50000 progress=89.8% loss=0.371500 lr=2.839477e-06 lr_backbone=2.839477e-06 elapsed=52:16:27 avg_step=4.19s recent_step=4.12s eta=05:57:48 +2026-09-22 10:43:45 | INFO | train rank=0 | step=44890/50000 progress=89.8% loss=0.298648 lr=2.828502e-06 lr_backbone=2.828502e-06 elapsed=52:17:09 avg_step=4.19s recent_step=4.18s eta=05:57:06 +2026-09-22 10:44:27 | INFO | train rank=0 | step=44900/50000 progress=89.8% loss=0.327507 lr=2.817548e-06 lr_backbone=2.817548e-06 elapsed=52:17:51 avg_step=4.19s recent_step=4.18s eta=05:56:24 +2026-09-22 10:45:09 | INFO | train rank=0 | step=44910/50000 progress=89.8% loss=0.305837 lr=2.806614e-06 lr_backbone=2.806614e-06 elapsed=52:18:32 avg_step=4.19s recent_step=4.14s eta=05:55:42 +2026-09-22 10:45:50 | INFO | train rank=0 | step=44920/50000 progress=89.8% loss=0.293302 lr=2.795700e-06 lr_backbone=2.795700e-06 elapsed=52:19:14 avg_step=4.19s recent_step=4.16s eta=05:55:00 +2026-09-22 10:46:31 | INFO | train rank=0 | step=44930/50000 progress=89.9% loss=0.317336 lr=2.784808e-06 lr_backbone=2.784808e-06 elapsed=52:19:55 avg_step=4.19s recent_step=4.12s eta=05:54:18 +2026-09-22 10:47:13 | INFO | train rank=0 | step=44940/50000 progress=89.9% loss=0.324655 lr=2.773936e-06 lr_backbone=2.773936e-06 elapsed=52:20:36 avg_step=4.19s recent_step=4.14s eta=05:53:36 +2026-09-22 10:47:55 | INFO | train rank=0 | step=44950/50000 progress=89.9% loss=0.381465 lr=2.763084e-06 lr_backbone=2.763084e-06 elapsed=52:21:18 avg_step=4.19s recent_step=4.18s eta=05:52:55 +2026-09-22 10:48:36 | INFO | train rank=0 | step=44960/50000 progress=89.9% loss=0.303070 lr=2.752254e-06 lr_backbone=2.752254e-06 elapsed=52:22:00 avg_step=4.19s recent_step=4.17s eta=05:52:13 +2026-09-22 10:49:18 | INFO | train rank=0 | step=44970/50000 progress=89.9% loss=0.305895 lr=2.741444e-06 lr_backbone=2.741444e-06 elapsed=52:22:41 avg_step=4.19s recent_step=4.18s eta=05:51:31 +2026-09-22 10:50:00 | INFO | train rank=0 | step=44980/50000 progress=90.0% loss=0.310318 lr=2.730655e-06 lr_backbone=2.730655e-06 elapsed=52:23:23 avg_step=4.19s recent_step=4.15s eta=05:50:49 +2026-09-22 10:50:41 | INFO | train rank=0 | step=44990/50000 progress=90.0% loss=0.286581 lr=2.719886e-06 lr_backbone=2.719886e-06 elapsed=52:24:04 avg_step=4.19s recent_step=4.15s eta=05:50:07 +2026-09-22 10:51:22 | INFO | train rank=0 | step=45000/50000 progress=90.0% loss=0.272194 lr=2.709138e-06 lr_backbone=2.709138e-06 elapsed=52:24:46 avg_step=4.19s recent_step=4.11s eta=05:49:25 +2026-09-22 10:52:04 | INFO | train rank=0 | step=45010/50000 progress=90.0% loss=0.267968 lr=2.698411e-06 lr_backbone=2.698411e-06 elapsed=52:25:27 avg_step=4.19s recent_step=4.17s eta=05:48:43 +2026-09-22 10:52:46 | INFO | train rank=0 | step=45020/50000 progress=90.0% loss=0.320380 lr=2.687704e-06 lr_backbone=2.687704e-06 elapsed=52:26:09 avg_step=4.19s recent_step=4.16s eta=05:48:01 +2026-09-22 10:53:27 | INFO | train rank=0 | step=45030/50000 progress=90.1% loss=0.349930 lr=2.677018e-06 lr_backbone=2.677018e-06 elapsed=52:26:51 avg_step=4.19s recent_step=4.18s eta=05:47:19 +2026-09-22 10:54:09 | INFO | train rank=0 | step=45040/50000 progress=90.1% loss=0.297748 lr=2.666353e-06 lr_backbone=2.666353e-06 elapsed=52:27:32 avg_step=4.19s recent_step=4.15s eta=05:46:37 +2026-09-22 10:54:50 | INFO | train rank=0 | step=45050/50000 progress=90.1% loss=0.305962 lr=2.655709e-06 lr_backbone=2.655709e-06 elapsed=52:28:13 avg_step=4.19s recent_step=4.12s eta=05:45:55 +2026-09-22 10:55:31 | INFO | train rank=0 | step=45060/50000 progress=90.1% loss=0.292179 lr=2.645085e-06 lr_backbone=2.645085e-06 elapsed=52:28:55 avg_step=4.19s recent_step=4.12s eta=05:45:13 +2026-09-22 10:56:13 | INFO | train rank=0 | step=45070/50000 progress=90.1% loss=0.305455 lr=2.634482e-06 lr_backbone=2.634482e-06 elapsed=52:29:36 avg_step=4.19s recent_step=4.17s eta=05:44:31 +2026-09-22 10:56:55 | INFO | train rank=0 | step=45080/50000 progress=90.2% loss=0.314629 lr=2.623899e-06 lr_backbone=2.623899e-06 elapsed=52:30:18 avg_step=4.19s recent_step=4.17s eta=05:43:49 +2026-09-22 10:57:36 | INFO | train rank=0 | step=45090/50000 progress=90.2% loss=0.294524 lr=2.613338e-06 lr_backbone=2.613338e-06 elapsed=52:30:59 avg_step=4.19s recent_step=4.14s eta=05:43:07 +2026-09-22 10:58:18 | INFO | train rank=0 | step=45100/50000 progress=90.2% loss=0.305504 lr=2.602797e-06 lr_backbone=2.602797e-06 elapsed=52:31:41 avg_step=4.19s recent_step=4.16s eta=05:42:25 +2026-09-22 10:58:59 | INFO | train rank=0 | step=45110/50000 progress=90.2% loss=0.305438 lr=2.592277e-06 lr_backbone=2.592277e-06 elapsed=52:32:22 avg_step=4.19s recent_step=4.12s eta=05:41:43 +2026-09-22 10:59:40 | INFO | train rank=0 | step=45120/50000 progress=90.2% loss=0.272530 lr=2.581777e-06 lr_backbone=2.581777e-06 elapsed=52:33:04 avg_step=4.19s recent_step=4.13s eta=05:41:01 +2026-09-22 11:00:22 | INFO | train rank=0 | step=45130/50000 progress=90.3% loss=0.297097 lr=2.571299e-06 lr_backbone=2.571299e-06 elapsed=52:33:45 avg_step=4.19s recent_step=4.18s eta=05:40:19 +2026-09-22 11:01:04 | INFO | train rank=0 | step=45140/50000 progress=90.3% loss=0.256326 lr=2.560841e-06 lr_backbone=2.560841e-06 elapsed=52:34:27 avg_step=4.19s recent_step=4.18s eta=05:39:37 +2026-09-22 11:01:46 | INFO | train rank=0 | step=45150/50000 progress=90.3% loss=0.327777 lr=2.550404e-06 lr_backbone=2.550404e-06 elapsed=52:35:09 avg_step=4.19s recent_step=4.19s eta=05:38:55 +2026-09-22 11:02:27 | INFO | train rank=0 | step=45160/50000 progress=90.3% loss=0.342517 lr=2.539987e-06 lr_backbone=2.539987e-06 elapsed=52:35:51 avg_step=4.19s recent_step=4.16s eta=05:38:13 +2026-09-22 11:03:09 | INFO | train rank=0 | step=45170/50000 progress=90.3% loss=0.325052 lr=2.529591e-06 lr_backbone=2.529591e-06 elapsed=52:36:32 avg_step=4.19s recent_step=4.14s eta=05:37:31 +2026-09-22 11:03:50 | INFO | train rank=0 | step=45180/50000 progress=90.4% loss=0.306490 lr=2.519217e-06 lr_backbone=2.519217e-06 elapsed=52:37:13 avg_step=4.19s recent_step=4.13s eta=05:36:49 +2026-09-22 11:04:32 | INFO | train rank=0 | step=45190/50000 progress=90.4% loss=0.295558 lr=2.508862e-06 lr_backbone=2.508862e-06 elapsed=52:37:55 avg_step=4.19s recent_step=4.16s eta=05:36:07 +2026-09-22 11:05:13 | INFO | train rank=0 | step=45200/50000 progress=90.4% loss=0.275811 lr=2.498529e-06 lr_backbone=2.498529e-06 elapsed=52:38:36 avg_step=4.19s recent_step=4.14s eta=05:35:25 +2026-09-22 11:05:55 | INFO | train rank=0 | step=45210/50000 progress=90.4% loss=0.358241 lr=2.488217e-06 lr_backbone=2.488217e-06 elapsed=52:39:18 avg_step=4.19s recent_step=4.17s eta=05:34:43 +2026-09-22 11:06:36 | INFO | train rank=0 | step=45220/50000 progress=90.4% loss=0.281031 lr=2.477925e-06 lr_backbone=2.477925e-06 elapsed=52:40:00 avg_step=4.19s recent_step=4.15s eta=05:34:01 +2026-09-22 11:07:18 | INFO | train rank=0 | step=45230/50000 progress=90.5% loss=0.321769 lr=2.467654e-06 lr_backbone=2.467654e-06 elapsed=52:40:41 avg_step=4.19s recent_step=4.12s eta=05:33:19 +2026-09-22 11:07:59 | INFO | train rank=0 | step=45240/50000 progress=90.5% loss=0.344071 lr=2.457404e-06 lr_backbone=2.457404e-06 elapsed=52:41:22 avg_step=4.19s recent_step=4.13s eta=05:32:37 +2026-09-22 11:08:40 | INFO | train rank=0 | step=45250/50000 progress=90.5% loss=0.284715 lr=2.447174e-06 lr_backbone=2.447174e-06 elapsed=52:42:04 avg_step=4.19s recent_step=4.16s eta=05:31:55 +2026-09-22 11:09:22 | INFO | train rank=0 | step=45260/50000 progress=90.5% loss=0.276264 lr=2.436966e-06 lr_backbone=2.436966e-06 elapsed=52:42:46 avg_step=4.19s recent_step=4.19s eta=05:31:13 +2026-09-22 11:10:04 | INFO | train rank=0 | step=45270/50000 progress=90.5% loss=0.288561 lr=2.426778e-06 lr_backbone=2.426778e-06 elapsed=52:43:28 avg_step=4.19s recent_step=4.19s eta=05:30:31 +2026-09-22 11:10:46 | INFO | train rank=0 | step=45280/50000 progress=90.6% loss=0.470925 lr=2.416611e-06 lr_backbone=2.416611e-06 elapsed=52:44:09 avg_step=4.19s recent_step=4.17s eta=05:29:49 +2026-09-22 11:11:27 | INFO | train rank=0 | step=45290/50000 progress=90.6% loss=0.293456 lr=2.406465e-06 lr_backbone=2.406465e-06 elapsed=52:44:50 avg_step=4.19s recent_step=4.11s eta=05:29:07 +2026-09-22 11:12:08 | INFO | train rank=0 | step=45300/50000 progress=90.6% loss=0.264783 lr=2.396339e-06 lr_backbone=2.396339e-06 elapsed=52:45:32 avg_step=4.19s recent_step=4.14s eta=05:28:25 +2026-09-22 11:12:50 | INFO | train rank=0 | step=45310/50000 progress=90.6% loss=0.357534 lr=2.386235e-06 lr_backbone=2.386235e-06 elapsed=52:46:14 avg_step=4.19s recent_step=4.18s eta=05:27:44 +2026-09-22 11:13:32 | INFO | train rank=0 | step=45320/50000 progress=90.6% loss=0.262739 lr=2.376151e-06 lr_backbone=2.376151e-06 elapsed=52:46:55 avg_step=4.19s recent_step=4.17s eta=05:27:02 +2026-09-22 11:14:14 | INFO | train rank=0 | step=45330/50000 progress=90.7% loss=0.301321 lr=2.366088e-06 lr_backbone=2.366088e-06 elapsed=52:47:37 avg_step=4.19s recent_step=4.17s eta=05:26:20 +2026-09-22 11:14:55 | INFO | train rank=0 | step=45340/50000 progress=90.7% loss=0.291709 lr=2.356046e-06 lr_backbone=2.356046e-06 elapsed=52:48:18 avg_step=4.19s recent_step=4.14s eta=05:25:38 +2026-09-22 11:15:36 | INFO | train rank=0 | step=45350/50000 progress=90.7% loss=0.306572 lr=2.346025e-06 lr_backbone=2.346025e-06 elapsed=52:49:00 avg_step=4.19s recent_step=4.13s eta=05:24:56 +2026-09-22 11:16:17 | INFO | train rank=0 | step=45360/50000 progress=90.7% loss=0.285767 lr=2.336025e-06 lr_backbone=2.336025e-06 elapsed=52:49:41 avg_step=4.19s recent_step=4.11s eta=05:24:14 +2026-09-22 11:16:59 | INFO | train rank=0 | step=45370/50000 progress=90.7% loss=0.327639 lr=2.326045e-06 lr_backbone=2.326045e-06 elapsed=52:50:22 avg_step=4.19s recent_step=4.16s eta=05:23:32 +2026-09-22 11:17:41 | INFO | train rank=0 | step=45380/50000 progress=90.8% loss=0.299178 lr=2.316086e-06 lr_backbone=2.316086e-06 elapsed=52:51:04 avg_step=4.19s recent_step=4.18s eta=05:22:50 +2026-09-22 11:18:22 | INFO | train rank=0 | step=45390/50000 progress=90.8% loss=0.264168 lr=2.306149e-06 lr_backbone=2.306149e-06 elapsed=52:51:46 avg_step=4.19s recent_step=4.15s eta=05:22:08 +2026-09-22 11:19:04 | INFO | train rank=0 | step=45400/50000 progress=90.8% loss=0.265643 lr=2.296232e-06 lr_backbone=2.296232e-06 elapsed=52:52:27 avg_step=4.19s recent_step=4.17s eta=05:21:26 +2026-09-22 11:19:45 | INFO | train rank=0 | step=45410/50000 progress=90.8% loss=0.308658 lr=2.286336e-06 lr_backbone=2.286336e-06 elapsed=52:53:08 avg_step=4.19s recent_step=4.10s eta=05:20:44 +2026-09-22 11:20:26 | INFO | train rank=0 | step=45420/50000 progress=90.8% loss=0.266233 lr=2.276460e-06 lr_backbone=2.276460e-06 elapsed=52:53:49 avg_step=4.19s recent_step=4.12s eta=05:20:02 +2026-09-22 11:21:08 | INFO | train rank=0 | step=45430/50000 progress=90.9% loss=0.280019 lr=2.266606e-06 lr_backbone=2.266606e-06 elapsed=52:54:31 avg_step=4.19s recent_step=4.14s eta=05:19:20 +2026-09-22 11:21:49 | INFO | train rank=0 | step=45440/50000 progress=90.9% loss=0.281319 lr=2.256773e-06 lr_backbone=2.256773e-06 elapsed=52:55:13 avg_step=4.19s recent_step=4.17s eta=05:18:38 +2026-09-22 11:22:31 | INFO | train rank=0 | step=45450/50000 progress=90.9% loss=0.272128 lr=2.246960e-06 lr_backbone=2.246960e-06 elapsed=52:55:54 avg_step=4.19s recent_step=4.16s eta=05:17:56 +2026-09-22 11:23:12 | INFO | train rank=0 | step=45460/50000 progress=90.9% loss=0.304392 lr=2.237169e-06 lr_backbone=2.237169e-06 elapsed=52:56:36 avg_step=4.19s recent_step=4.17s eta=05:17:14 +2026-09-22 11:23:53 | INFO | train rank=0 | step=45470/50000 progress=90.9% loss=0.368747 lr=2.227398e-06 lr_backbone=2.227398e-06 elapsed=52:57:17 avg_step=4.19s recent_step=4.10s eta=05:16:32 +2026-09-22 11:24:34 | INFO | train rank=0 | step=45480/50000 progress=91.0% loss=0.352939 lr=2.217648e-06 lr_backbone=2.217648e-06 elapsed=52:57:58 avg_step=4.19s recent_step=4.09s eta=05:15:50 +2026-09-22 11:25:16 | INFO | train rank=0 | step=45490/50000 progress=91.0% loss=0.268313 lr=2.207919e-06 lr_backbone=2.207919e-06 elapsed=52:58:39 avg_step=4.19s recent_step=4.15s eta=05:15:08 +2026-09-22 11:25:58 | INFO | train rank=0 | step=45500/50000 progress=91.0% loss=0.316137 lr=2.198211e-06 lr_backbone=2.198211e-06 elapsed=52:59:21 avg_step=4.19s recent_step=4.18s eta=05:14:26 +2026-09-22 11:26:39 | INFO | train rank=0 | step=45510/50000 progress=91.0% loss=0.348813 lr=2.188524e-06 lr_backbone=2.188524e-06 elapsed=53:00:03 avg_step=4.19s recent_step=4.17s eta=05:13:44 +2026-09-22 11:27:21 | INFO | train rank=0 | step=45520/50000 progress=91.0% loss=0.369898 lr=2.178858e-06 lr_backbone=2.178858e-06 elapsed=53:00:44 avg_step=4.19s recent_step=4.16s eta=05:13:02 +2026-09-22 11:28:02 | INFO | train rank=0 | step=45530/50000 progress=91.1% loss=0.350150 lr=2.169212e-06 lr_backbone=2.169212e-06 elapsed=53:01:26 avg_step=4.19s recent_step=4.13s eta=05:12:20 +2026-09-22 11:28:43 | INFO | train rank=0 | step=45540/50000 progress=91.1% loss=0.317303 lr=2.159588e-06 lr_backbone=2.159588e-06 elapsed=53:02:07 avg_step=4.19s recent_step=4.11s eta=05:11:38 +2026-09-22 11:29:25 | INFO | train rank=0 | step=45550/50000 progress=91.1% loss=0.304596 lr=2.149984e-06 lr_backbone=2.149984e-06 elapsed=53:02:48 avg_step=4.19s recent_step=4.14s eta=05:10:56 +2026-09-22 11:30:06 | INFO | train rank=0 | step=45560/50000 progress=91.1% loss=0.335427 lr=2.140402e-06 lr_backbone=2.140402e-06 elapsed=53:03:30 avg_step=4.19s recent_step=4.15s eta=05:10:14 +2026-09-22 11:30:48 | INFO | train rank=0 | step=45570/50000 progress=91.1% loss=0.322797 lr=2.130840e-06 lr_backbone=2.130840e-06 elapsed=53:04:11 avg_step=4.19s recent_step=4.16s eta=05:09:32 +2026-09-22 11:31:29 | INFO | train rank=0 | step=45580/50000 progress=91.2% loss=0.277195 lr=2.121300e-06 lr_backbone=2.121300e-06 elapsed=53:04:53 avg_step=4.19s recent_step=4.15s eta=05:08:50 +2026-09-22 11:32:10 | INFO | train rank=0 | step=45590/50000 progress=91.2% loss=0.264735 lr=2.111780e-06 lr_backbone=2.111780e-06 elapsed=53:05:34 avg_step=4.19s recent_step=4.10s eta=05:08:08 +2026-09-22 11:32:51 | INFO | train rank=0 | step=45600/50000 progress=91.2% loss=0.262414 lr=2.102281e-06 lr_backbone=2.102281e-06 elapsed=53:06:15 avg_step=4.19s recent_step=4.11s eta=05:07:26 +2026-09-22 11:33:33 | INFO | train rank=0 | step=45610/50000 progress=91.2% loss=0.318030 lr=2.092803e-06 lr_backbone=2.092803e-06 elapsed=53:06:56 avg_step=4.19s recent_step=4.14s eta=05:06:44 +2026-09-22 11:34:14 | INFO | train rank=0 | step=45620/50000 progress=91.2% loss=0.291865 lr=2.083346e-06 lr_backbone=2.083346e-06 elapsed=53:07:38 avg_step=4.19s recent_step=4.14s eta=05:06:02 +2026-09-22 11:34:56 | INFO | train rank=0 | step=45630/50000 progress=91.3% loss=0.350268 lr=2.073911e-06 lr_backbone=2.073911e-06 elapsed=53:08:19 avg_step=4.19s recent_step=4.15s eta=05:05:20 +2026-09-22 11:35:37 | INFO | train rank=0 | step=45640/50000 progress=91.3% loss=0.311538 lr=2.064496e-06 lr_backbone=2.064496e-06 elapsed=53:09:01 avg_step=4.19s recent_step=4.15s eta=05:04:38 +2026-09-22 11:36:18 | INFO | train rank=0 | step=45650/50000 progress=91.3% loss=0.295928 lr=2.055102e-06 lr_backbone=2.055102e-06 elapsed=53:09:42 avg_step=4.19s recent_step=4.11s eta=05:03:56 +2026-09-22 11:36:59 | INFO | train rank=0 | step=45660/50000 progress=91.3% loss=0.303964 lr=2.045729e-06 lr_backbone=2.045729e-06 elapsed=53:10:23 avg_step=4.19s recent_step=4.11s eta=05:03:14 +2026-09-22 11:37:41 | INFO | train rank=0 | step=45670/50000 progress=91.3% loss=0.309945 lr=2.036377e-06 lr_backbone=2.036377e-06 elapsed=53:11:04 avg_step=4.19s recent_step=4.16s eta=05:02:32 +2026-09-22 11:38:23 | INFO | train rank=0 | step=45680/50000 progress=91.4% loss=0.402888 lr=2.027046e-06 lr_backbone=2.027046e-06 elapsed=53:11:46 avg_step=4.19s recent_step=4.17s eta=05:01:50 +2026-09-22 11:39:04 | INFO | train rank=0 | step=45690/50000 progress=91.4% loss=0.268546 lr=2.017736e-06 lr_backbone=2.017736e-06 elapsed=53:12:28 avg_step=4.19s recent_step=4.16s eta=05:01:09 +2026-09-22 11:39:46 | INFO | train rank=0 | step=45700/50000 progress=91.4% loss=0.345238 lr=2.008447e-06 lr_backbone=2.008447e-06 elapsed=53:13:09 avg_step=4.19s recent_step=4.15s eta=05:00:27 +2026-09-22 11:40:27 | INFO | train rank=0 | step=45710/50000 progress=91.4% loss=0.437237 lr=1.999178e-06 lr_backbone=1.999178e-06 elapsed=53:13:50 avg_step=4.19s recent_step=4.10s eta=04:59:45 +2026-09-22 11:41:08 | INFO | train rank=0 | step=45720/50000 progress=91.4% loss=0.299166 lr=1.989931e-06 lr_backbone=1.989931e-06 elapsed=53:14:31 avg_step=4.19s recent_step=4.11s eta=04:59:03 +2026-09-22 11:41:49 | INFO | train rank=0 | step=45730/50000 progress=91.5% loss=0.355007 lr=1.980705e-06 lr_backbone=1.980705e-06 elapsed=53:15:13 avg_step=4.19s recent_step=4.15s eta=04:58:21 +2026-09-22 11:42:31 | INFO | train rank=0 | step=45740/50000 progress=91.5% loss=0.396506 lr=1.971500e-06 lr_backbone=1.971500e-06 elapsed=53:15:54 avg_step=4.19s recent_step=4.15s eta=04:57:39 +2026-09-22 11:43:13 | INFO | train rank=0 | step=45750/50000 progress=91.5% loss=0.275118 lr=1.962316e-06 lr_backbone=1.962316e-06 elapsed=53:16:36 avg_step=4.19s recent_step=4.15s eta=04:56:57 +2026-09-22 11:43:54 | INFO | train rank=0 | step=45760/50000 progress=91.5% loss=0.299331 lr=1.953153e-06 lr_backbone=1.953153e-06 elapsed=53:17:17 avg_step=4.19s recent_step=4.16s eta=04:56:15 +2026-09-22 11:44:35 | INFO | train rank=0 | step=45770/50000 progress=91.5% loss=0.323264 lr=1.944011e-06 lr_backbone=1.944011e-06 elapsed=53:17:58 avg_step=4.19s recent_step=4.09s eta=04:55:33 +2026-09-22 11:45:16 | INFO | train rank=0 | step=45780/50000 progress=91.6% loss=0.317376 lr=1.934890e-06 lr_backbone=1.934890e-06 elapsed=53:18:40 avg_step=4.19s recent_step=4.14s eta=04:54:51 +2026-09-22 11:45:58 | INFO | train rank=0 | step=45790/50000 progress=91.6% loss=0.262418 lr=1.925790e-06 lr_backbone=1.925790e-06 elapsed=53:19:21 avg_step=4.19s recent_step=4.15s eta=04:54:09 +2026-09-22 11:46:40 | INFO | train rank=0 | step=45800/50000 progress=91.6% loss=0.284030 lr=1.916711e-06 lr_backbone=1.916711e-06 elapsed=53:20:03 avg_step=4.19s recent_step=4.18s eta=04:53:27 +2026-09-22 11:47:21 | INFO | train rank=0 | step=45810/50000 progress=91.6% loss=0.317836 lr=1.907653e-06 lr_backbone=1.907653e-06 elapsed=53:20:45 avg_step=4.19s recent_step=4.18s eta=04:52:45 +2026-09-22 11:48:03 | INFO | train rank=0 | step=45820/50000 progress=91.6% loss=0.276021 lr=1.898616e-06 lr_backbone=1.898616e-06 elapsed=53:21:27 avg_step=4.19s recent_step=4.18s eta=04:52:03 +2026-09-22 11:48:44 | INFO | train rank=0 | step=45830/50000 progress=91.7% loss=0.291720 lr=1.889601e-06 lr_backbone=1.889601e-06 elapsed=53:22:08 avg_step=4.19s recent_step=4.11s eta=04:51:21 +2026-09-22 11:49:26 | INFO | train rank=0 | step=45840/50000 progress=91.7% loss=0.321494 lr=1.880606e-06 lr_backbone=1.880606e-06 elapsed=53:22:49 avg_step=4.19s recent_step=4.13s eta=04:50:39 +2026-09-22 11:50:07 | INFO | train rank=0 | step=45850/50000 progress=91.7% loss=0.381464 lr=1.871632e-06 lr_backbone=1.871632e-06 elapsed=53:23:31 avg_step=4.19s recent_step=4.15s eta=04:49:57 +2026-09-22 11:50:49 | INFO | train rank=0 | step=45860/50000 progress=91.7% loss=0.373581 lr=1.862679e-06 lr_backbone=1.862679e-06 elapsed=53:24:12 avg_step=4.19s recent_step=4.16s eta=04:49:15 +2026-09-22 11:51:30 | INFO | train rank=0 | step=45870/50000 progress=91.7% loss=0.275888 lr=1.853748e-06 lr_backbone=1.853748e-06 elapsed=53:24:54 avg_step=4.19s recent_step=4.15s eta=04:48:33 +2026-09-22 11:52:12 | INFO | train rank=0 | step=45880/50000 progress=91.8% loss=0.322699 lr=1.844837e-06 lr_backbone=1.844837e-06 elapsed=53:25:36 avg_step=4.19s recent_step=4.18s eta=04:47:51 +2026-09-22 11:52:54 | INFO | train rank=0 | step=45890/50000 progress=91.8% loss=0.281108 lr=1.835948e-06 lr_backbone=1.835948e-06 elapsed=53:26:17 avg_step=4.19s recent_step=4.13s eta=04:47:09 +2026-09-22 11:53:35 | INFO | train rank=0 | step=45900/50000 progress=91.8% loss=0.256915 lr=1.827079e-06 lr_backbone=1.827079e-06 elapsed=53:26:58 avg_step=4.19s recent_step=4.14s eta=04:46:27 +2026-09-22 11:54:16 | INFO | train rank=0 | step=45910/50000 progress=91.8% loss=0.277886 lr=1.818232e-06 lr_backbone=1.818232e-06 elapsed=53:27:40 avg_step=4.19s recent_step=4.13s eta=04:45:45 +2026-09-22 11:54:58 | INFO | train rank=0 | step=45920/50000 progress=91.8% loss=0.432318 lr=1.809406e-06 lr_backbone=1.809406e-06 elapsed=53:28:21 avg_step=4.19s recent_step=4.17s eta=04:45:03 +2026-09-22 11:55:40 | INFO | train rank=0 | step=45930/50000 progress=91.9% loss=0.272492 lr=1.800600e-06 lr_backbone=1.800600e-06 elapsed=53:29:03 avg_step=4.19s recent_step=4.16s eta=04:44:21 +2026-09-22 11:56:21 | INFO | train rank=0 | step=45940/50000 progress=91.9% loss=0.301791 lr=1.791816e-06 lr_backbone=1.791816e-06 elapsed=53:29:45 avg_step=4.19s recent_step=4.16s eta=04:43:39 +2026-09-22 11:57:02 | INFO | train rank=0 | step=45950/50000 progress=91.9% loss=0.393445 lr=1.783053e-06 lr_backbone=1.783053e-06 elapsed=53:30:26 avg_step=4.19s recent_step=4.13s eta=04:42:57 +2026-09-22 11:57:44 | INFO | train rank=0 | step=45960/50000 progress=91.9% loss=0.318981 lr=1.774311e-06 lr_backbone=1.774311e-06 elapsed=53:31:07 avg_step=4.19s recent_step=4.12s eta=04:42:15 +2026-09-22 11:58:25 | INFO | train rank=0 | step=45970/50000 progress=91.9% loss=0.359775 lr=1.765590e-06 lr_backbone=1.765590e-06 elapsed=53:31:49 avg_step=4.19s recent_step=4.18s eta=04:41:34 +2026-09-22 11:59:07 | INFO | train rank=0 | step=45980/50000 progress=92.0% loss=0.343085 lr=1.756891e-06 lr_backbone=1.756891e-06 elapsed=53:32:30 avg_step=4.19s recent_step=4.16s eta=04:40:52 +2026-09-22 11:59:49 | INFO | train rank=0 | step=45990/50000 progress=92.0% loss=0.322723 lr=1.748212e-06 lr_backbone=1.748212e-06 elapsed=53:33:12 avg_step=4.19s recent_step=4.18s eta=04:40:10 +2026-09-22 12:00:31 | INFO | train rank=0 | step=46000/50000 progress=92.0% loss=0.304351 lr=1.739554e-06 lr_backbone=1.739554e-06 elapsed=53:33:54 avg_step=4.19s recent_step=4.19s eta=04:39:28 +2026-09-22 12:01:12 | INFO | train rank=0 | step=46010/50000 progress=92.0% loss=0.388187 lr=1.730918e-06 lr_backbone=1.730918e-06 elapsed=53:34:35 avg_step=4.19s recent_step=4.11s eta=04:38:46 +2026-09-22 12:01:53 | INFO | train rank=0 | step=46020/50000 progress=92.0% loss=0.343935 lr=1.722303e-06 lr_backbone=1.722303e-06 elapsed=53:35:16 avg_step=4.19s recent_step=4.12s eta=04:38:04 +2026-09-22 12:02:35 | INFO | train rank=0 | step=46030/50000 progress=92.1% loss=0.299734 lr=1.713709e-06 lr_backbone=1.713709e-06 elapsed=53:35:58 avg_step=4.19s recent_step=4.19s eta=04:37:22 +2026-09-22 12:03:17 | INFO | train rank=0 | step=46040/50000 progress=92.1% loss=0.354644 lr=1.705135e-06 lr_backbone=1.705135e-06 elapsed=53:36:40 avg_step=4.19s recent_step=4.16s eta=04:36:40 +2026-09-22 12:03:58 | INFO | train rank=0 | step=46050/50000 progress=92.1% loss=0.273142 lr=1.696584e-06 lr_backbone=1.696584e-06 elapsed=53:37:22 avg_step=4.19s recent_step=4.19s eta=04:35:58 +2026-09-22 12:04:40 | INFO | train rank=0 | step=46060/50000 progress=92.1% loss=0.322808 lr=1.688053e-06 lr_backbone=1.688053e-06 elapsed=53:38:03 avg_step=4.19s recent_step=4.17s eta=04:35:16 +2026-09-22 12:05:21 | INFO | train rank=0 | step=46070/50000 progress=92.1% loss=0.289017 lr=1.679543e-06 lr_backbone=1.679543e-06 elapsed=53:38:45 avg_step=4.19s recent_step=4.12s eta=04:34:34 +2026-09-22 12:06:03 | INFO | train rank=0 | step=46080/50000 progress=92.2% loss=0.282154 lr=1.671054e-06 lr_backbone=1.671054e-06 elapsed=53:39:26 avg_step=4.19s recent_step=4.14s eta=04:33:52 +2026-09-22 12:06:44 | INFO | train rank=0 | step=46090/50000 progress=92.2% loss=0.328539 lr=1.662587e-06 lr_backbone=1.662587e-06 elapsed=53:40:08 avg_step=4.19s recent_step=4.17s eta=04:33:10 +2026-09-22 12:07:26 | INFO | train rank=0 | step=46100/50000 progress=92.2% loss=0.298716 lr=1.654141e-06 lr_backbone=1.654141e-06 elapsed=53:40:49 avg_step=4.19s recent_step=4.16s eta=04:32:28 +2026-09-22 12:08:08 | INFO | train rank=0 | step=46110/50000 progress=92.2% loss=0.369094 lr=1.645716e-06 lr_backbone=1.645716e-06 elapsed=53:41:31 avg_step=4.19s recent_step=4.17s eta=04:31:46 +2026-09-22 12:08:50 | INFO | train rank=0 | step=46120/50000 progress=92.2% loss=0.279973 lr=1.637312e-06 lr_backbone=1.637312e-06 elapsed=53:42:13 avg_step=4.19s recent_step=4.18s eta=04:31:04 +2026-09-22 12:09:31 | INFO | train rank=0 | step=46130/50000 progress=92.3% loss=0.295818 lr=1.628929e-06 lr_backbone=1.628929e-06 elapsed=53:42:54 avg_step=4.19s recent_step=4.11s eta=04:30:22 +2026-09-22 12:10:12 | INFO | train rank=0 | step=46140/50000 progress=92.3% loss=0.313519 lr=1.620567e-06 lr_backbone=1.620567e-06 elapsed=53:43:35 avg_step=4.19s recent_step=4.13s eta=04:29:40 +2026-09-22 12:10:54 | INFO | train rank=0 | step=46150/50000 progress=92.3% loss=0.386236 lr=1.612227e-06 lr_backbone=1.612227e-06 elapsed=53:44:17 avg_step=4.19s recent_step=4.19s eta=04:28:58 +2026-09-22 12:11:35 | INFO | train rank=0 | step=46160/50000 progress=92.3% loss=0.321780 lr=1.603907e-06 lr_backbone=1.603907e-06 elapsed=53:44:59 avg_step=4.19s recent_step=4.17s eta=04:28:16 +2026-09-22 12:12:17 | INFO | train rank=0 | step=46170/50000 progress=92.3% loss=0.441620 lr=1.595609e-06 lr_backbone=1.595609e-06 elapsed=53:45:40 avg_step=4.19s recent_step=4.13s eta=04:27:35 +2026-09-22 12:12:59 | INFO | train rank=0 | step=46180/50000 progress=92.4% loss=0.341886 lr=1.587332e-06 lr_backbone=1.587332e-06 elapsed=53:46:22 avg_step=4.19s recent_step=4.18s eta=04:26:53 +2026-09-22 12:13:40 | INFO | train rank=0 | step=46190/50000 progress=92.4% loss=0.395091 lr=1.579077e-06 lr_backbone=1.579077e-06 elapsed=53:47:03 avg_step=4.19s recent_step=4.15s eta=04:26:11 +2026-09-22 12:14:21 | INFO | train rank=0 | step=46200/50000 progress=92.4% loss=0.273999 lr=1.570842e-06 lr_backbone=1.570842e-06 elapsed=53:47:45 avg_step=4.19s recent_step=4.12s eta=04:25:29 +2026-09-22 12:15:03 | INFO | train rank=0 | step=46210/50000 progress=92.4% loss=0.344810 lr=1.562629e-06 lr_backbone=1.562629e-06 elapsed=53:48:26 avg_step=4.19s recent_step=4.17s eta=04:24:47 +2026-09-22 12:15:45 | INFO | train rank=0 | step=46220/50000 progress=92.4% loss=0.273420 lr=1.554436e-06 lr_backbone=1.554436e-06 elapsed=53:49:08 avg_step=4.19s recent_step=4.18s eta=04:24:05 +2026-09-22 12:16:26 | INFO | train rank=0 | step=46230/50000 progress=92.5% loss=0.344894 lr=1.546265e-06 lr_backbone=1.546265e-06 elapsed=53:49:50 avg_step=4.19s recent_step=4.16s eta=04:23:23 +2026-09-22 12:17:08 | INFO | train rank=0 | step=46240/50000 progress=92.5% loss=0.333259 lr=1.538115e-06 lr_backbone=1.538115e-06 elapsed=53:50:32 avg_step=4.19s recent_step=4.19s eta=04:22:41 +2026-09-22 12:17:50 | INFO | train rank=0 | step=46250/50000 progress=92.5% loss=0.318464 lr=1.529987e-06 lr_backbone=1.529987e-06 elapsed=53:51:13 avg_step=4.19s recent_step=4.12s eta=04:21:59 +2026-09-22 12:18:31 | INFO | train rank=0 | step=46260/50000 progress=92.5% loss=0.278714 lr=1.521879e-06 lr_backbone=1.521879e-06 elapsed=53:51:54 avg_step=4.19s recent_step=4.12s eta=04:21:17 +2026-09-22 12:19:13 | INFO | train rank=0 | step=46270/50000 progress=92.5% loss=0.344861 lr=1.513793e-06 lr_backbone=1.513793e-06 elapsed=53:52:36 avg_step=4.19s recent_step=4.19s eta=04:20:35 +2026-09-22 12:19:54 | INFO | train rank=0 | step=46280/50000 progress=92.6% loss=0.282217 lr=1.505728e-06 lr_backbone=1.505728e-06 elapsed=53:53:18 avg_step=4.19s recent_step=4.17s eta=04:19:53 +2026-09-22 12:20:36 | INFO | train rank=0 | step=46290/50000 progress=92.6% loss=0.293016 lr=1.497684e-06 lr_backbone=1.497684e-06 elapsed=53:53:59 avg_step=4.19s recent_step=4.14s eta=04:19:11 +2026-09-22 12:21:17 | INFO | train rank=0 | step=46300/50000 progress=92.6% loss=0.354570 lr=1.489662e-06 lr_backbone=1.489662e-06 elapsed=53:54:41 avg_step=4.19s recent_step=4.16s eta=04:18:29 +2026-09-22 12:21:59 | INFO | train rank=0 | step=46310/50000 progress=92.6% loss=0.345986 lr=1.481660e-06 lr_backbone=1.481660e-06 elapsed=53:55:22 avg_step=4.19s recent_step=4.14s eta=04:17:47 +2026-09-22 12:22:40 | INFO | train rank=0 | step=46320/50000 progress=92.6% loss=0.311838 lr=1.473680e-06 lr_backbone=1.473680e-06 elapsed=53:56:04 avg_step=4.19s recent_step=4.14s eta=04:17:05 +2026-09-22 12:23:22 | INFO | train rank=0 | step=46330/50000 progress=92.7% loss=0.321964 lr=1.465721e-06 lr_backbone=1.465721e-06 elapsed=53:56:45 avg_step=4.19s recent_step=4.16s eta=04:16:23 +2026-09-22 12:24:04 | INFO | train rank=0 | step=46340/50000 progress=92.7% loss=0.339097 lr=1.457783e-06 lr_backbone=1.457783e-06 elapsed=53:57:27 avg_step=4.19s recent_step=4.18s eta=04:15:41 +2026-09-22 12:24:45 | INFO | train rank=0 | step=46350/50000 progress=92.7% loss=0.267832 lr=1.449867e-06 lr_backbone=1.449867e-06 elapsed=53:58:09 avg_step=4.19s recent_step=4.18s eta=04:15:00 +2026-09-22 12:25:27 | INFO | train rank=0 | step=46360/50000 progress=92.7% loss=0.300619 lr=1.441972e-06 lr_backbone=1.441972e-06 elapsed=53:58:50 avg_step=4.19s recent_step=4.15s eta=04:14:18 +2026-09-22 12:26:08 | INFO | train rank=0 | step=46370/50000 progress=92.7% loss=0.331928 lr=1.434098e-06 lr_backbone=1.434098e-06 elapsed=53:59:31 avg_step=4.19s recent_step=4.12s eta=04:13:36 +2026-09-22 12:26:50 | INFO | train rank=0 | step=46380/50000 progress=92.8% loss=0.331568 lr=1.426245e-06 lr_backbone=1.426245e-06 elapsed=54:00:13 avg_step=4.19s recent_step=4.14s eta=04:12:54 +2026-09-22 12:27:31 | INFO | train rank=0 | step=46390/50000 progress=92.8% loss=0.382653 lr=1.418413e-06 lr_backbone=1.418413e-06 elapsed=54:00:54 avg_step=4.19s recent_step=4.16s eta=04:12:12 +2026-09-22 12:28:13 | INFO | train rank=0 | step=46400/50000 progress=92.8% loss=0.331222 lr=1.410603e-06 lr_backbone=1.410603e-06 elapsed=54:01:36 avg_step=4.19s recent_step=4.18s eta=04:11:30 +2026-09-22 12:28:55 | INFO | train rank=0 | step=46410/50000 progress=92.8% loss=0.333633 lr=1.402814e-06 lr_backbone=1.402814e-06 elapsed=54:02:18 avg_step=4.19s recent_step=4.16s eta=04:10:48 +2026-09-22 12:29:36 | INFO | train rank=0 | step=46420/50000 progress=92.8% loss=0.258460 lr=1.395046e-06 lr_backbone=1.395046e-06 elapsed=54:02:59 avg_step=4.19s recent_step=4.15s eta=04:10:06 +2026-09-22 12:30:17 | INFO | train rank=0 | step=46430/50000 progress=92.9% loss=0.283002 lr=1.387300e-06 lr_backbone=1.387300e-06 elapsed=54:03:41 avg_step=4.19s recent_step=4.13s eta=04:09:24 +2026-09-22 12:30:59 | INFO | train rank=0 | step=46440/50000 progress=92.9% loss=0.286964 lr=1.379575e-06 lr_backbone=1.379575e-06 elapsed=54:04:22 avg_step=4.19s recent_step=4.13s eta=04:08:42 +2026-09-22 12:31:40 | INFO | train rank=0 | step=46450/50000 progress=92.9% loss=0.385687 lr=1.371871e-06 lr_backbone=1.371871e-06 elapsed=54:05:04 avg_step=4.19s recent_step=4.16s eta=04:08:00 +2026-09-22 12:32:22 | INFO | train rank=0 | step=46460/50000 progress=92.9% loss=0.341519 lr=1.364188e-06 lr_backbone=1.364188e-06 elapsed=54:05:45 avg_step=4.19s recent_step=4.17s eta=04:07:18 +2026-09-22 12:33:04 | INFO | train rank=0 | step=46470/50000 progress=92.9% loss=0.289058 lr=1.356527e-06 lr_backbone=1.356527e-06 elapsed=54:06:27 avg_step=4.19s recent_step=4.17s eta=04:06:36 +2026-09-22 12:33:45 | INFO | train rank=0 | step=46480/50000 progress=93.0% loss=0.315840 lr=1.348887e-06 lr_backbone=1.348887e-06 elapsed=54:07:08 avg_step=4.19s recent_step=4.15s eta=04:05:54 +2026-09-22 12:34:26 | INFO | train rank=0 | step=46490/50000 progress=93.0% loss=0.326521 lr=1.341268e-06 lr_backbone=1.341268e-06 elapsed=54:07:50 avg_step=4.19s recent_step=4.12s eta=04:05:12 +2026-09-22 12:35:07 | INFO | train rank=0 | step=46500/50000 progress=93.0% loss=0.286413 lr=1.333670e-06 lr_backbone=1.333670e-06 elapsed=54:08:31 avg_step=4.19s recent_step=4.13s eta=04:04:30 +2026-09-22 12:35:49 | INFO | train rank=0 | step=46510/50000 progress=93.0% loss=0.288268 lr=1.326094e-06 lr_backbone=1.326094e-06 elapsed=54:09:13 avg_step=4.19s recent_step=4.18s eta=04:03:48 +2026-09-22 12:36:31 | INFO | train rank=0 | step=46520/50000 progress=93.0% loss=0.266177 lr=1.318539e-06 lr_backbone=1.318539e-06 elapsed=54:09:54 avg_step=4.19s recent_step=4.18s eta=04:03:06 +2026-09-22 12:37:13 | INFO | train rank=0 | step=46530/50000 progress=93.1% loss=0.320316 lr=1.311005e-06 lr_backbone=1.311005e-06 elapsed=54:10:36 avg_step=4.19s recent_step=4.15s eta=04:02:24 +2026-09-22 12:37:54 | INFO | train rank=0 | step=46540/50000 progress=93.1% loss=0.274738 lr=1.303493e-06 lr_backbone=1.303493e-06 elapsed=54:11:17 avg_step=4.19s recent_step=4.13s eta=04:01:42 +2026-09-22 12:38:35 | INFO | train rank=0 | step=46550/50000 progress=93.1% loss=0.308052 lr=1.296002e-06 lr_backbone=1.296002e-06 elapsed=54:11:58 avg_step=4.19s recent_step=4.13s eta=04:01:01 +2026-09-22 12:39:16 | INFO | train rank=0 | step=46560/50000 progress=93.1% loss=0.288630 lr=1.288532e-06 lr_backbone=1.288532e-06 elapsed=54:12:40 avg_step=4.19s recent_step=4.13s eta=04:00:19 +2026-09-22 12:39:58 | INFO | train rank=0 | step=46570/50000 progress=93.1% loss=0.265184 lr=1.281084e-06 lr_backbone=1.281084e-06 elapsed=54:13:22 avg_step=4.19s recent_step=4.17s eta=03:59:37 +2026-09-22 12:40:40 | INFO | train rank=0 | step=46580/50000 progress=93.2% loss=0.427708 lr=1.273656e-06 lr_backbone=1.273656e-06 elapsed=54:14:03 avg_step=4.19s recent_step=4.18s eta=03:58:55 +2026-09-22 12:41:22 | INFO | train rank=0 | step=46590/50000 progress=93.2% loss=0.306882 lr=1.266251e-06 lr_backbone=1.266251e-06 elapsed=54:14:45 avg_step=4.19s recent_step=4.18s eta=03:58:13 +2026-09-22 12:42:03 | INFO | train rank=0 | step=46600/50000 progress=93.2% loss=0.284167 lr=1.258866e-06 lr_backbone=1.258866e-06 elapsed=54:15:27 avg_step=4.19s recent_step=4.15s eta=03:57:31 +2026-09-22 12:42:45 | INFO | train rank=0 | step=46610/50000 progress=93.2% loss=0.298377 lr=1.251503e-06 lr_backbone=1.251503e-06 elapsed=54:16:08 avg_step=4.19s recent_step=4.12s eta=03:56:49 +2026-09-22 12:43:26 | INFO | train rank=0 | step=46620/50000 progress=93.2% loss=0.313588 lr=1.244161e-06 lr_backbone=1.244161e-06 elapsed=54:16:49 avg_step=4.19s recent_step=4.12s eta=03:56:07 +2026-09-22 12:44:08 | INFO | train rank=0 | step=46630/50000 progress=93.3% loss=0.275898 lr=1.236840e-06 lr_backbone=1.236840e-06 elapsed=54:17:31 avg_step=4.19s recent_step=4.18s eta=03:55:25 +2026-09-22 12:44:49 | INFO | train rank=0 | step=46640/50000 progress=93.3% loss=0.318699 lr=1.229541e-06 lr_backbone=1.229541e-06 elapsed=54:18:13 avg_step=4.19s recent_step=4.17s eta=03:54:43 +2026-09-22 12:45:31 | INFO | train rank=0 | step=46650/50000 progress=93.3% loss=0.328444 lr=1.222263e-06 lr_backbone=1.222263e-06 elapsed=54:18:54 avg_step=4.19s recent_step=4.16s eta=03:54:01 +2026-09-22 12:46:12 | INFO | train rank=0 | step=46660/50000 progress=93.3% loss=0.306522 lr=1.215007e-06 lr_backbone=1.215007e-06 elapsed=54:19:36 avg_step=4.19s recent_step=4.15s eta=03:53:19 +2026-09-22 12:46:53 | INFO | train rank=0 | step=46670/50000 progress=93.3% loss=0.349647 lr=1.207772e-06 lr_backbone=1.207772e-06 elapsed=54:20:17 avg_step=4.19s recent_step=4.10s eta=03:52:37 +2026-09-22 12:47:34 | INFO | train rank=0 | step=46680/50000 progress=93.4% loss=0.365114 lr=1.200558e-06 lr_backbone=1.200558e-06 elapsed=54:20:58 avg_step=4.19s recent_step=4.10s eta=03:51:55 +2026-09-22 12:48:16 | INFO | train rank=0 | step=46690/50000 progress=93.4% loss=0.296386 lr=1.193365e-06 lr_backbone=1.193365e-06 elapsed=54:21:39 avg_step=4.19s recent_step=4.16s eta=03:51:13 +2026-09-22 12:48:58 | INFO | train rank=0 | step=46700/50000 progress=93.4% loss=0.383989 lr=1.186194e-06 lr_backbone=1.186194e-06 elapsed=54:22:21 avg_step=4.19s recent_step=4.16s eta=03:50:31 +2026-09-22 12:49:39 | INFO | train rank=0 | step=46710/50000 progress=93.4% loss=0.265287 lr=1.179044e-06 lr_backbone=1.179044e-06 elapsed=54:23:02 avg_step=4.19s recent_step=4.16s eta=03:49:49 +2026-09-22 12:50:21 | INFO | train rank=0 | step=46720/50000 progress=93.4% loss=0.308724 lr=1.171916e-06 lr_backbone=1.171916e-06 elapsed=54:23:44 avg_step=4.19s recent_step=4.14s eta=03:49:07 +2026-09-22 12:51:02 | INFO | train rank=0 | step=46730/50000 progress=93.5% loss=0.274557 lr=1.164809e-06 lr_backbone=1.164809e-06 elapsed=54:24:25 avg_step=4.19s recent_step=4.11s eta=03:48:25 +2026-09-22 12:51:43 | INFO | train rank=0 | step=46740/50000 progress=93.5% loss=0.285882 lr=1.157723e-06 lr_backbone=1.157723e-06 elapsed=54:25:06 avg_step=4.19s recent_step=4.11s eta=03:47:44 +2026-09-22 12:52:24 | INFO | train rank=0 | step=46750/50000 progress=93.5% loss=0.298398 lr=1.150658e-06 lr_backbone=1.150658e-06 elapsed=54:25:48 avg_step=4.19s recent_step=4.14s eta=03:47:02 +2026-09-22 12:53:06 | INFO | train rank=0 | step=46760/50000 progress=93.5% loss=0.312270 lr=1.143615e-06 lr_backbone=1.143615e-06 elapsed=54:26:29 avg_step=4.19s recent_step=4.16s eta=03:46:20 +2026-09-22 12:53:47 | INFO | train rank=0 | step=46770/50000 progress=93.5% loss=0.303651 lr=1.136594e-06 lr_backbone=1.136594e-06 elapsed=54:27:11 avg_step=4.19s recent_step=4.15s eta=03:45:38 +2026-09-22 12:54:29 | INFO | train rank=0 | step=46780/50000 progress=93.6% loss=0.241580 lr=1.129594e-06 lr_backbone=1.129594e-06 elapsed=54:27:52 avg_step=4.19s recent_step=4.15s eta=03:44:56 +2026-09-22 12:55:10 | INFO | train rank=0 | step=46790/50000 progress=93.6% loss=0.261189 lr=1.122615e-06 lr_backbone=1.122615e-06 elapsed=54:28:33 avg_step=4.19s recent_step=4.10s eta=03:44:14 +2026-09-22 12:55:51 | INFO | train rank=0 | step=46800/50000 progress=93.6% loss=0.310314 lr=1.115657e-06 lr_backbone=1.115657e-06 elapsed=54:29:14 avg_step=4.19s recent_step=4.09s eta=03:43:32 +2026-09-22 12:56:32 | INFO | train rank=0 | step=46810/50000 progress=93.6% loss=0.334082 lr=1.108721e-06 lr_backbone=1.108721e-06 elapsed=54:29:56 avg_step=4.19s recent_step=4.17s eta=03:42:50 +2026-09-22 12:57:14 | INFO | train rank=0 | step=46820/50000 progress=93.6% loss=0.326420 lr=1.101806e-06 lr_backbone=1.101806e-06 elapsed=54:30:37 avg_step=4.19s recent_step=4.15s eta=03:42:08 +2026-09-22 12:57:56 | INFO | train rank=0 | step=46830/50000 progress=93.7% loss=0.356690 lr=1.094913e-06 lr_backbone=1.094913e-06 elapsed=54:31:19 avg_step=4.19s recent_step=4.16s eta=03:41:26 +2026-09-22 12:58:37 | INFO | train rank=0 | step=46840/50000 progress=93.7% loss=0.276996 lr=1.088041e-06 lr_backbone=1.088041e-06 elapsed=54:32:00 avg_step=4.19s recent_step=4.16s eta=03:40:44 +2026-09-22 12:59:18 | INFO | train rank=0 | step=46850/50000 progress=93.7% loss=0.326989 lr=1.081190e-06 lr_backbone=1.081190e-06 elapsed=54:32:42 avg_step=4.19s recent_step=4.11s eta=03:40:02 +2026-09-22 12:59:59 | INFO | train rank=0 | step=46860/50000 progress=93.7% loss=0.343284 lr=1.074361e-06 lr_backbone=1.074361e-06 elapsed=54:33:23 avg_step=4.19s recent_step=4.10s eta=03:39:20 +2026-09-22 13:00:41 | INFO | train rank=0 | step=46870/50000 progress=93.7% loss=0.314697 lr=1.067554e-06 lr_backbone=1.067554e-06 elapsed=54:34:04 avg_step=4.19s recent_step=4.17s eta=03:38:38 +2026-09-22 13:01:22 | INFO | train rank=0 | step=46880/50000 progress=93.8% loss=0.279780 lr=1.060767e-06 lr_backbone=1.060767e-06 elapsed=54:34:46 avg_step=4.19s recent_step=4.14s eta=03:37:56 +2026-09-22 13:02:04 | INFO | train rank=0 | step=46890/50000 progress=93.8% loss=0.318553 lr=1.054002e-06 lr_backbone=1.054002e-06 elapsed=54:35:27 avg_step=4.19s recent_step=4.15s eta=03:37:14 +2026-09-22 13:02:45 | INFO | train rank=0 | step=46900/50000 progress=93.8% loss=0.429880 lr=1.047259e-06 lr_backbone=1.047259e-06 elapsed=54:36:09 avg_step=4.19s recent_step=4.15s eta=03:36:32 +2026-09-22 13:03:26 | INFO | train rank=0 | step=46910/50000 progress=93.8% loss=0.305816 lr=1.040537e-06 lr_backbone=1.040537e-06 elapsed=54:36:50 avg_step=4.19s recent_step=4.11s eta=03:35:50 +2026-09-22 13:04:08 | INFO | train rank=0 | step=46920/50000 progress=93.8% loss=0.274144 lr=1.033836e-06 lr_backbone=1.033836e-06 elapsed=54:37:31 avg_step=4.19s recent_step=4.12s eta=03:35:08 +2026-09-22 13:04:49 | INFO | train rank=0 | step=46930/50000 progress=93.9% loss=0.333205 lr=1.027157e-06 lr_backbone=1.027157e-06 elapsed=54:38:12 avg_step=4.19s recent_step=4.14s eta=03:34:26 +2026-09-22 13:05:31 | INFO | train rank=0 | step=46940/50000 progress=93.9% loss=0.259358 lr=1.020499e-06 lr_backbone=1.020499e-06 elapsed=54:38:54 avg_step=4.19s recent_step=4.18s eta=03:33:45 +2026-09-22 13:06:13 | INFO | train rank=0 | step=46950/50000 progress=93.9% loss=0.272690 lr=1.013862e-06 lr_backbone=1.013862e-06 elapsed=54:39:36 avg_step=4.19s recent_step=4.18s eta=03:33:03 +2026-09-22 13:06:54 | INFO | train rank=0 | step=46960/50000 progress=93.9% loss=0.291672 lr=1.007247e-06 lr_backbone=1.007247e-06 elapsed=54:40:18 avg_step=4.19s recent_step=4.17s eta=03:32:21 +2026-09-22 13:07:36 | INFO | train rank=0 | step=46970/50000 progress=93.9% loss=0.375521 lr=1.000654e-06 lr_backbone=1.000654e-06 elapsed=54:40:59 avg_step=4.19s recent_step=4.15s eta=03:31:39 +2026-09-22 13:08:17 | INFO | train rank=0 | step=46980/50000 progress=94.0% loss=0.308848 lr=9.940817e-07 lr_backbone=9.940817e-07 elapsed=54:41:40 avg_step=4.19s recent_step=4.13s eta=03:30:57 +2026-09-22 13:08:59 | INFO | train rank=0 | step=46990/50000 progress=94.0% loss=0.286347 lr=9.875310e-07 lr_backbone=9.875310e-07 elapsed=54:42:22 avg_step=4.19s recent_step=4.16s eta=03:30:15 +2026-09-22 13:09:40 | INFO | train rank=0 | step=47000/50000 progress=94.0% loss=0.360403 lr=9.810017e-07 lr_backbone=9.810017e-07 elapsed=54:43:04 avg_step=4.19s recent_step=4.17s eta=03:29:33 +2026-09-22 13:10:22 | INFO | train rank=0 | step=47010/50000 progress=94.0% loss=0.300467 lr=9.744939e-07 lr_backbone=9.744939e-07 elapsed=54:43:45 avg_step=4.19s recent_step=4.16s eta=03:28:51 +2026-09-22 13:11:04 | INFO | train rank=0 | step=47020/50000 progress=94.0% loss=0.313111 lr=9.680075e-07 lr_backbone=9.680075e-07 elapsed=54:44:27 avg_step=4.19s recent_step=4.17s eta=03:28:09 +2026-09-22 13:11:45 | INFO | train rank=0 | step=47030/50000 progress=94.1% loss=0.257062 lr=9.615426e-07 lr_backbone=9.615426e-07 elapsed=54:45:08 avg_step=4.19s recent_step=4.13s eta=03:27:27 +2026-09-22 13:12:26 | INFO | train rank=0 | step=47040/50000 progress=94.1% loss=0.308717 lr=9.550991e-07 lr_backbone=9.550991e-07 elapsed=54:45:50 avg_step=4.19s recent_step=4.13s eta=03:26:45 +2026-09-22 13:13:08 | INFO | train rank=0 | step=47050/50000 progress=94.1% loss=0.321512 lr=9.486771e-07 lr_backbone=9.486771e-07 elapsed=54:46:31 avg_step=4.19s recent_step=4.18s eta=03:26:03 +2026-09-22 13:13:50 | INFO | train rank=0 | step=47060/50000 progress=94.1% loss=0.320318 lr=9.422765e-07 lr_backbone=9.422765e-07 elapsed=54:47:13 avg_step=4.19s recent_step=4.15s eta=03:25:21 +2026-09-22 13:14:32 | INFO | train rank=0 | step=47070/50000 progress=94.1% loss=0.309526 lr=9.358974e-07 lr_backbone=9.358974e-07 elapsed=54:47:55 avg_step=4.19s recent_step=4.19s eta=03:24:39 +2026-09-22 13:15:13 | INFO | train rank=0 | step=47080/50000 progress=94.2% loss=0.373845 lr=9.295398e-07 lr_backbone=9.295398e-07 elapsed=54:48:37 avg_step=4.19s recent_step=4.17s eta=03:23:58 +2026-09-22 13:15:54 | INFO | train rank=0 | step=47090/50000 progress=94.2% loss=0.279118 lr=9.232036e-07 lr_backbone=9.232036e-07 elapsed=54:49:18 avg_step=4.19s recent_step=4.10s eta=03:23:16 +2026-09-22 13:16:35 | INFO | train rank=0 | step=47100/50000 progress=94.2% loss=0.268075 lr=9.168889e-07 lr_backbone=9.168889e-07 elapsed=54:49:59 avg_step=4.19s recent_step=4.12s eta=03:22:34 +2026-09-22 13:17:17 | INFO | train rank=0 | step=47110/50000 progress=94.2% loss=0.292323 lr=9.105957e-07 lr_backbone=9.105957e-07 elapsed=54:50:41 avg_step=4.19s recent_step=4.18s eta=03:21:52 +2026-09-22 13:17:59 | INFO | train rank=0 | step=47120/50000 progress=94.2% loss=0.319492 lr=9.043239e-07 lr_backbone=9.043239e-07 elapsed=54:51:22 avg_step=4.19s recent_step=4.15s eta=03:21:10 +2026-09-22 13:18:40 | INFO | train rank=0 | step=47130/50000 progress=94.3% loss=0.287928 lr=8.980736e-07 lr_backbone=8.980736e-07 elapsed=54:52:04 avg_step=4.19s recent_step=4.17s eta=03:20:28 +2026-09-22 13:19:22 | INFO | train rank=0 | step=47140/50000 progress=94.3% loss=0.309434 lr=8.918448e-07 lr_backbone=8.918448e-07 elapsed=54:52:45 avg_step=4.19s recent_step=4.17s eta=03:19:46 +2026-09-22 13:20:03 | INFO | train rank=0 | step=47150/50000 progress=94.3% loss=0.265686 lr=8.856375e-07 lr_backbone=8.856375e-07 elapsed=54:53:27 avg_step=4.19s recent_step=4.13s eta=03:19:04 +2026-09-22 13:20:45 | INFO | train rank=0 | step=47160/50000 progress=94.3% loss=0.336755 lr=8.794516e-07 lr_backbone=8.794516e-07 elapsed=54:54:08 avg_step=4.19s recent_step=4.14s eta=03:18:22 +2026-09-22 13:21:27 | INFO | train rank=0 | step=47170/50000 progress=94.3% loss=0.307359 lr=8.732873e-07 lr_backbone=8.732873e-07 elapsed=54:54:50 avg_step=4.19s recent_step=4.18s eta=03:17:40 +2026-09-22 13:22:08 | INFO | train rank=0 | step=47180/50000 progress=94.4% loss=0.352231 lr=8.671444e-07 lr_backbone=8.671444e-07 elapsed=54:55:32 avg_step=4.19s recent_step=4.17s eta=03:16:58 +2026-09-22 13:22:50 | INFO | train rank=0 | step=47190/50000 progress=94.4% loss=0.280193 lr=8.610230e-07 lr_backbone=8.610230e-07 elapsed=54:56:13 avg_step=4.19s recent_step=4.17s eta=03:16:16 +2026-09-22 13:23:32 | INFO | train rank=0 | step=47200/50000 progress=94.4% loss=0.361064 lr=8.549231e-07 lr_backbone=8.549231e-07 elapsed=54:56:55 avg_step=4.19s recent_step=4.16s eta=03:15:34 +2026-09-22 13:24:13 | INFO | train rank=0 | step=47210/50000 progress=94.4% loss=0.302171 lr=8.488448e-07 lr_backbone=8.488448e-07 elapsed=54:57:36 avg_step=4.19s recent_step=4.14s eta=03:14:52 +2026-09-22 13:24:54 | INFO | train rank=0 | step=47220/50000 progress=94.4% loss=0.365319 lr=8.427879e-07 lr_backbone=8.427879e-07 elapsed=54:58:18 avg_step=4.19s recent_step=4.12s eta=03:14:10 +2026-09-22 13:25:36 | INFO | train rank=0 | step=47230/50000 progress=94.5% loss=0.356638 lr=8.367525e-07 lr_backbone=8.367525e-07 elapsed=54:58:59 avg_step=4.19s recent_step=4.18s eta=03:13:29 +2026-09-22 13:26:18 | INFO | train rank=0 | step=47240/50000 progress=94.5% loss=0.305857 lr=8.307386e-07 lr_backbone=8.307386e-07 elapsed=54:59:41 avg_step=4.19s recent_step=4.15s eta=03:12:47 +2026-09-22 13:26:59 | INFO | train rank=0 | step=47250/50000 progress=94.5% loss=0.281440 lr=8.247463e-07 lr_backbone=8.247463e-07 elapsed=55:00:23 avg_step=4.19s recent_step=4.17s eta=03:12:05 +2026-09-22 13:27:41 | INFO | train rank=0 | step=47260/50000 progress=94.5% loss=0.396755 lr=8.187754e-07 lr_backbone=8.187754e-07 elapsed=55:01:04 avg_step=4.19s recent_step=4.18s eta=03:11:23 +2026-09-22 13:28:22 | INFO | train rank=0 | step=47270/50000 progress=94.5% loss=0.372141 lr=8.128261e-07 lr_backbone=8.128261e-07 elapsed=55:01:46 avg_step=4.19s recent_step=4.13s eta=03:10:41 +2026-09-22 13:29:04 | INFO | train rank=0 | step=47280/50000 progress=94.6% loss=0.324989 lr=8.068982e-07 lr_backbone=8.068982e-07 elapsed=55:02:27 avg_step=4.19s recent_step=4.12s eta=03:09:59 +2026-09-22 13:29:45 | INFO | train rank=0 | step=47290/50000 progress=94.6% loss=0.359300 lr=8.009919e-07 lr_backbone=8.009919e-07 elapsed=55:03:09 avg_step=4.19s recent_step=4.17s eta=03:09:17 +2026-09-22 13:30:27 | INFO | train rank=0 | step=47300/50000 progress=94.6% loss=0.298454 lr=7.951071e-07 lr_backbone=7.951071e-07 elapsed=55:03:50 avg_step=4.19s recent_step=4.14s eta=03:08:35 +2026-09-22 13:31:09 | INFO | train rank=0 | step=47310/50000 progress=94.6% loss=0.350301 lr=7.892439e-07 lr_backbone=7.892439e-07 elapsed=55:04:32 avg_step=4.19s recent_step=4.19s eta=03:07:53 +2026-09-22 13:31:50 | INFO | train rank=0 | step=47320/50000 progress=94.6% loss=0.307622 lr=7.834022e-07 lr_backbone=7.834022e-07 elapsed=55:05:14 avg_step=4.19s recent_step=4.18s eta=03:07:11 +2026-09-22 13:32:32 | INFO | train rank=0 | step=47330/50000 progress=94.7% loss=0.309633 lr=7.775819e-07 lr_backbone=7.775819e-07 elapsed=55:05:55 avg_step=4.19s recent_step=4.13s eta=03:06:29 +2026-09-22 13:33:13 | INFO | train rank=0 | step=47340/50000 progress=94.7% loss=0.295337 lr=7.717833e-07 lr_backbone=7.717833e-07 elapsed=55:06:36 avg_step=4.19s recent_step=4.13s eta=03:05:47 +2026-09-22 13:33:55 | INFO | train rank=0 | step=47350/50000 progress=94.7% loss=0.295034 lr=7.660061e-07 lr_backbone=7.660061e-07 elapsed=55:07:18 avg_step=4.19s recent_step=4.17s eta=03:05:05 +2026-09-22 13:34:36 | INFO | train rank=0 | step=47360/50000 progress=94.7% loss=0.333614 lr=7.602505e-07 lr_backbone=7.602505e-07 elapsed=55:07:59 avg_step=4.19s recent_step=4.14s eta=03:04:23 +2026-09-22 13:35:18 | INFO | train rank=0 | step=47370/50000 progress=94.7% loss=0.287945 lr=7.545165e-07 lr_backbone=7.545165e-07 elapsed=55:08:41 avg_step=4.19s recent_step=4.18s eta=03:03:42 +2026-09-22 13:36:00 | INFO | train rank=0 | step=47380/50000 progress=94.8% loss=0.333946 lr=7.488039e-07 lr_backbone=7.488039e-07 elapsed=55:09:23 avg_step=4.19s recent_step=4.17s eta=03:03:00 +2026-09-22 13:36:41 | INFO | train rank=0 | step=47390/50000 progress=94.8% loss=0.280755 lr=7.431130e-07 lr_backbone=7.431130e-07 elapsed=55:10:04 avg_step=4.19s recent_step=4.13s eta=03:02:18 +2026-09-22 13:37:22 | INFO | train rank=0 | step=47400/50000 progress=94.8% loss=0.301099 lr=7.374435e-07 lr_backbone=7.374435e-07 elapsed=55:10:46 avg_step=4.19s recent_step=4.13s eta=03:01:36 +2026-09-22 13:38:04 | INFO | train rank=0 | step=47410/50000 progress=94.8% loss=0.347655 lr=7.317957e-07 lr_backbone=7.317957e-07 elapsed=55:11:27 avg_step=4.19s recent_step=4.17s eta=03:00:54 +2026-09-22 13:38:46 | INFO | train rank=0 | step=47420/50000 progress=94.8% loss=0.303813 lr=7.261693e-07 lr_backbone=7.261693e-07 elapsed=55:12:09 avg_step=4.19s recent_step=4.16s eta=03:00:12 +2026-09-22 13:39:27 | INFO | train rank=0 | step=47430/50000 progress=94.9% loss=0.357844 lr=7.205645e-07 lr_backbone=7.205645e-07 elapsed=55:12:51 avg_step=4.19s recent_step=4.17s eta=02:59:30 +2026-09-22 13:40:09 | INFO | train rank=0 | step=47440/50000 progress=94.9% loss=0.262203 lr=7.149813e-07 lr_backbone=7.149813e-07 elapsed=55:13:32 avg_step=4.19s recent_step=4.17s eta=02:58:48 +2026-09-22 13:40:50 | INFO | train rank=0 | step=47450/50000 progress=94.9% loss=0.275130 lr=7.094197e-07 lr_backbone=7.094197e-07 elapsed=55:14:13 avg_step=4.19s recent_step=4.11s eta=02:58:06 +2026-09-22 13:41:31 | INFO | train rank=0 | step=47460/50000 progress=94.9% loss=0.292478 lr=7.038796e-07 lr_backbone=7.038796e-07 elapsed=55:14:55 avg_step=4.19s recent_step=4.12s eta=02:57:24 +2026-09-22 13:42:13 | INFO | train rank=0 | step=47470/50000 progress=94.9% loss=0.287708 lr=6.983610e-07 lr_backbone=6.983610e-07 elapsed=55:15:36 avg_step=4.19s recent_step=4.18s eta=02:56:42 +2026-09-22 13:42:55 | INFO | train rank=0 | step=47480/50000 progress=95.0% loss=0.319130 lr=6.928641e-07 lr_backbone=6.928641e-07 elapsed=55:16:18 avg_step=4.19s recent_step=4.18s eta=02:56:00 +2026-09-22 13:43:37 | INFO | train rank=0 | step=47490/50000 progress=95.0% loss=0.309536 lr=6.873887e-07 lr_backbone=6.873887e-07 elapsed=55:17:00 avg_step=4.19s recent_step=4.17s eta=02:55:18 +2026-09-22 13:44:18 | INFO | train rank=0 | step=47500/50000 progress=95.0% loss=0.293282 lr=6.819348e-07 lr_backbone=6.819348e-07 elapsed=55:17:42 avg_step=4.19s recent_step=4.19s eta=02:54:36 +2026-09-22 13:45:00 | INFO | train rank=0 | step=47510/50000 progress=95.0% loss=0.351953 lr=6.765026e-07 lr_backbone=6.765026e-07 elapsed=55:18:23 avg_step=4.19s recent_step=4.13s eta=02:53:55 +2026-09-22 13:45:41 | INFO | train rank=0 | step=47520/50000 progress=95.0% loss=0.295564 lr=6.710919e-07 lr_backbone=6.710919e-07 elapsed=55:19:05 avg_step=4.19s recent_step=4.15s eta=02:53:13 +2026-09-22 13:46:23 | INFO | train rank=0 | step=47530/50000 progress=95.1% loss=0.314103 lr=6.657028e-07 lr_backbone=6.657028e-07 elapsed=55:19:46 avg_step=4.19s recent_step=4.15s eta=02:52:31 +2026-09-22 13:47:05 | INFO | train rank=0 | step=47540/50000 progress=95.1% loss=0.328522 lr=6.603353e-07 lr_backbone=6.603353e-07 elapsed=55:20:28 avg_step=4.19s recent_step=4.18s eta=02:51:49 +2026-09-22 13:47:46 | INFO | train rank=0 | step=47550/50000 progress=95.1% loss=0.349591 lr=6.549893e-07 lr_backbone=6.549893e-07 elapsed=55:21:09 avg_step=4.19s recent_step=4.16s eta=02:51:07 +2026-09-22 13:48:28 | INFO | train rank=0 | step=47560/50000 progress=95.1% loss=0.265728 lr=6.496650e-07 lr_backbone=6.496650e-07 elapsed=55:21:51 avg_step=4.19s recent_step=4.18s eta=02:50:25 +2026-09-22 13:49:09 | INFO | train rank=0 | step=47570/50000 progress=95.1% loss=0.386513 lr=6.443622e-07 lr_backbone=6.443622e-07 elapsed=55:22:33 avg_step=4.19s recent_step=4.13s eta=02:49:43 +2026-09-22 13:49:51 | INFO | train rank=0 | step=47580/50000 progress=95.2% loss=0.300213 lr=6.390810e-07 lr_backbone=6.390810e-07 elapsed=55:23:14 avg_step=4.19s recent_step=4.14s eta=02:49:01 +2026-09-22 13:50:32 | INFO | train rank=0 | step=47590/50000 progress=95.2% loss=0.302225 lr=6.338214e-07 lr_backbone=6.338214e-07 elapsed=55:23:56 avg_step=4.19s recent_step=4.17s eta=02:48:19 +2026-09-22 13:51:14 | INFO | train rank=0 | step=47600/50000 progress=95.2% loss=0.299505 lr=6.285835e-07 lr_backbone=6.285835e-07 elapsed=55:24:38 avg_step=4.19s recent_step=4.19s eta=02:47:37 +2026-09-22 13:51:56 | INFO | train rank=0 | step=47610/50000 progress=95.2% loss=0.358015 lr=6.233671e-07 lr_backbone=6.233671e-07 elapsed=55:25:19 avg_step=4.19s recent_step=4.16s eta=02:46:55 +2026-09-22 13:52:38 | INFO | train rank=0 | step=47620/50000 progress=95.2% loss=0.335972 lr=6.181723e-07 lr_backbone=6.181723e-07 elapsed=55:26:01 avg_step=4.19s recent_step=4.18s eta=02:46:13 +2026-09-22 13:53:19 | INFO | train rank=0 | step=47630/50000 progress=95.3% loss=0.324769 lr=6.129991e-07 lr_backbone=6.129991e-07 elapsed=55:26:42 avg_step=4.19s recent_step=4.10s eta=02:45:31 +2026-09-22 13:54:00 | INFO | train rank=0 | step=47640/50000 progress=95.3% loss=0.252128 lr=6.078475e-07 lr_backbone=6.078475e-07 elapsed=55:27:23 avg_step=4.19s recent_step=4.10s eta=02:44:49 +2026-09-22 13:54:41 | INFO | train rank=0 | step=47650/50000 progress=95.3% loss=0.323914 lr=6.027175e-07 lr_backbone=6.027175e-07 elapsed=55:28:05 avg_step=4.19s recent_step=4.18s eta=02:44:08 +2026-09-22 13:55:23 | INFO | train rank=0 | step=47660/50000 progress=95.3% loss=0.311721 lr=5.976091e-07 lr_backbone=5.976091e-07 elapsed=55:28:47 avg_step=4.19s recent_step=4.17s eta=02:43:26 +2026-09-22 13:56:05 | INFO | train rank=0 | step=47670/50000 progress=95.3% loss=0.332613 lr=5.925224e-07 lr_backbone=5.925224e-07 elapsed=55:29:28 avg_step=4.19s recent_step=4.18s eta=02:42:44 +2026-09-22 13:56:47 | INFO | train rank=0 | step=47680/50000 progress=95.4% loss=0.259212 lr=5.874572e-07 lr_backbone=5.874572e-07 elapsed=55:30:10 avg_step=4.19s recent_step=4.16s eta=02:42:02 +2026-09-22 13:57:28 | INFO | train rank=0 | step=47690/50000 progress=95.4% loss=0.318373 lr=5.824137e-07 lr_backbone=5.824137e-07 elapsed=55:30:51 avg_step=4.19s recent_step=4.13s eta=02:41:20 +2026-09-22 13:58:09 | INFO | train rank=0 | step=47700/50000 progress=95.4% loss=0.297230 lr=5.773917e-07 lr_backbone=5.773917e-07 elapsed=55:31:32 avg_step=4.19s recent_step=4.13s eta=02:40:38 +2026-09-22 13:58:51 | INFO | train rank=0 | step=47710/50000 progress=95.4% loss=0.281540 lr=5.723914e-07 lr_backbone=5.723914e-07 elapsed=55:32:14 avg_step=4.19s recent_step=4.17s eta=02:39:56 +2026-09-22 13:59:32 | INFO | train rank=0 | step=47720/50000 progress=95.4% loss=0.323782 lr=5.674128e-07 lr_backbone=5.674128e-07 elapsed=55:32:56 avg_step=4.19s recent_step=4.17s eta=02:39:14 +2026-09-22 14:00:14 | INFO | train rank=0 | step=47730/50000 progress=95.5% loss=0.336885 lr=5.624557e-07 lr_backbone=5.624557e-07 elapsed=55:33:38 avg_step=4.19s recent_step=4.18s eta=02:38:32 +2026-09-22 14:00:56 | INFO | train rank=0 | step=47740/50000 progress=95.5% loss=0.286564 lr=5.575203e-07 lr_backbone=5.575203e-07 elapsed=55:34:19 avg_step=4.19s recent_step=4.16s eta=02:37:50 +2026-09-22 14:01:37 | INFO | train rank=0 | step=47750/50000 progress=95.5% loss=0.317275 lr=5.526065e-07 lr_backbone=5.526065e-07 elapsed=55:35:01 avg_step=4.19s recent_step=4.13s eta=02:37:08 +2026-09-22 14:02:19 | INFO | train rank=0 | step=47760/50000 progress=95.5% loss=0.299881 lr=5.477143e-07 lr_backbone=5.477143e-07 elapsed=55:35:42 avg_step=4.19s recent_step=4.13s eta=02:36:26 +2026-09-22 14:03:00 | INFO | train rank=0 | step=47770/50000 progress=95.5% loss=0.302704 lr=5.428438e-07 lr_backbone=5.428438e-07 elapsed=55:36:24 avg_step=4.19s recent_step=4.18s eta=02:35:45 +2026-09-22 14:03:42 | INFO | train rank=0 | step=47780/50000 progress=95.6% loss=0.340559 lr=5.379948e-07 lr_backbone=5.379948e-07 elapsed=55:37:05 avg_step=4.19s recent_step=4.18s eta=02:35:03 +2026-09-22 14:04:24 | INFO | train rank=0 | step=47790/50000 progress=95.6% loss=0.278098 lr=5.331676e-07 lr_backbone=5.331676e-07 elapsed=55:37:47 avg_step=4.19s recent_step=4.18s eta=02:34:21 +2026-09-22 14:05:05 | INFO | train rank=0 | step=47800/50000 progress=95.6% loss=0.342344 lr=5.283619e-07 lr_backbone=5.283619e-07 elapsed=55:38:29 avg_step=4.19s recent_step=4.16s eta=02:33:39 +2026-09-22 14:05:47 | INFO | train rank=0 | step=47810/50000 progress=95.6% loss=0.247143 lr=5.235780e-07 lr_backbone=5.235780e-07 elapsed=55:39:10 avg_step=4.19s recent_step=4.12s eta=02:32:57 +2026-09-22 14:06:28 | INFO | train rank=0 | step=47820/50000 progress=95.6% loss=0.353262 lr=5.188156e-07 lr_backbone=5.188156e-07 elapsed=55:39:51 avg_step=4.19s recent_step=4.12s eta=02:32:15 +2026-09-22 14:07:10 | INFO | train rank=0 | step=47830/50000 progress=95.7% loss=0.275756 lr=5.140749e-07 lr_backbone=5.140749e-07 elapsed=55:40:33 avg_step=4.19s recent_step=4.19s eta=02:31:33 +2026-09-22 14:07:51 | INFO | train rank=0 | step=47840/50000 progress=95.7% loss=0.417087 lr=5.093558e-07 lr_backbone=5.093558e-07 elapsed=55:41:15 avg_step=4.19s recent_step=4.15s eta=02:30:51 +2026-09-22 14:08:33 | INFO | train rank=0 | step=47850/50000 progress=95.7% loss=0.334469 lr=5.046584e-07 lr_backbone=5.046584e-07 elapsed=55:41:56 avg_step=4.19s recent_step=4.17s eta=02:30:09 +2026-09-22 14:09:14 | INFO | train rank=0 | step=47860/50000 progress=95.7% loss=0.347843 lr=4.999827e-07 lr_backbone=4.999827e-07 elapsed=55:42:38 avg_step=4.19s recent_step=4.14s eta=02:29:27 +2026-09-22 14:09:55 | INFO | train rank=0 | step=47870/50000 progress=95.7% loss=0.305920 lr=4.953286e-07 lr_backbone=4.953286e-07 elapsed=55:43:19 avg_step=4.19s recent_step=4.10s eta=02:28:45 +2026-09-22 14:10:37 | INFO | train rank=0 | step=47880/50000 progress=95.8% loss=0.297345 lr=4.906961e-07 lr_backbone=4.906961e-07 elapsed=55:44:00 avg_step=4.19s recent_step=4.12s eta=02:28:03 +2026-09-22 14:11:18 | INFO | train rank=0 | step=47890/50000 progress=95.8% loss=0.299513 lr=4.860853e-07 lr_backbone=4.860853e-07 elapsed=55:44:42 avg_step=4.19s recent_step=4.16s eta=02:27:21 +2026-09-22 14:12:00 | INFO | train rank=0 | step=47900/50000 progress=95.8% loss=0.320198 lr=4.814962e-07 lr_backbone=4.814962e-07 elapsed=55:45:23 avg_step=4.19s recent_step=4.16s eta=02:26:39 +2026-09-22 14:12:41 | INFO | train rank=0 | step=47910/50000 progress=95.8% loss=0.388864 lr=4.769287e-07 lr_backbone=4.769287e-07 elapsed=55:46:05 avg_step=4.19s recent_step=4.15s eta=02:25:58 +2026-09-22 14:13:23 | INFO | train rank=0 | step=47920/50000 progress=95.8% loss=0.271514 lr=4.723829e-07 lr_backbone=4.723829e-07 elapsed=55:46:46 avg_step=4.19s recent_step=4.16s eta=02:25:16 +2026-09-22 14:14:04 | INFO | train rank=0 | step=47930/50000 progress=95.9% loss=0.332687 lr=4.678588e-07 lr_backbone=4.678588e-07 elapsed=55:47:27 avg_step=4.19s recent_step=4.13s eta=02:24:34 +2026-09-22 14:14:45 | INFO | train rank=0 | step=47940/50000 progress=95.9% loss=0.293284 lr=4.633563e-07 lr_backbone=4.633563e-07 elapsed=55:48:09 avg_step=4.19s recent_step=4.11s eta=02:23:52 +2026-09-22 14:15:27 | INFO | train rank=0 | step=47950/50000 progress=95.9% loss=0.350659 lr=4.588755e-07 lr_backbone=4.588755e-07 elapsed=55:48:50 avg_step=4.19s recent_step=4.14s eta=02:23:10 +2026-09-22 14:16:08 | INFO | train rank=0 | step=47960/50000 progress=95.9% loss=0.382923 lr=4.544163e-07 lr_backbone=4.544163e-07 elapsed=55:49:31 avg_step=4.19s recent_step=4.15s eta=02:22:28 +2026-09-22 14:16:50 | INFO | train rank=0 | step=47970/50000 progress=95.9% loss=0.284107 lr=4.499789e-07 lr_backbone=4.499789e-07 elapsed=55:50:13 avg_step=4.19s recent_step=4.16s eta=02:21:46 +2026-09-22 14:17:31 | INFO | train rank=0 | step=47980/50000 progress=96.0% loss=0.287034 lr=4.455631e-07 lr_backbone=4.455631e-07 elapsed=55:50:55 avg_step=4.19s recent_step=4.16s eta=02:21:04 +2026-09-22 14:18:12 | INFO | train rank=0 | step=47990/50000 progress=96.0% loss=0.328541 lr=4.411690e-07 lr_backbone=4.411690e-07 elapsed=55:51:35 avg_step=4.19s recent_step=4.08s eta=02:20:22 +2026-09-22 14:18:53 | INFO | train rank=0 | step=48000/50000 progress=96.0% loss=0.273334 lr=4.367965e-07 lr_backbone=4.367965e-07 elapsed=55:52:16 avg_step=4.19s recent_step=4.09s eta=02:19:40 +2026-09-22 14:19:35 | INFO | train rank=0 | step=48010/50000 progress=96.0% loss=0.307174 lr=4.324458e-07 lr_backbone=4.324458e-07 elapsed=55:52:58 avg_step=4.19s recent_step=4.15s eta=02:18:58 +2026-09-22 14:20:16 | INFO | train rank=0 | step=48020/50000 progress=96.0% loss=0.273103 lr=4.281167e-07 lr_backbone=4.281167e-07 elapsed=55:53:39 avg_step=4.19s recent_step=4.15s eta=02:18:16 +2026-09-22 14:20:58 | INFO | train rank=0 | step=48030/50000 progress=96.1% loss=0.267146 lr=4.238093e-07 lr_backbone=4.238093e-07 elapsed=55:54:21 avg_step=4.19s recent_step=4.16s eta=02:17:34 +2026-09-22 14:21:39 | INFO | train rank=0 | step=48040/50000 progress=96.1% loss=0.381927 lr=4.195236e-07 lr_backbone=4.195236e-07 elapsed=55:55:02 avg_step=4.19s recent_step=4.15s eta=02:16:53 +2026-09-22 14:22:20 | INFO | train rank=0 | step=48050/50000 progress=96.1% loss=0.361792 lr=4.152596e-07 lr_backbone=4.152596e-07 elapsed=55:55:43 avg_step=4.19s recent_step=4.11s eta=02:16:11 +2026-09-22 14:23:01 | INFO | train rank=0 | step=48060/50000 progress=96.1% loss=0.325631 lr=4.110173e-07 lr_backbone=4.110173e-07 elapsed=55:56:25 avg_step=4.19s recent_step=4.13s eta=02:15:29 +2026-09-22 14:23:43 | INFO | train rank=0 | step=48070/50000 progress=96.1% loss=0.280634 lr=4.067967e-07 lr_backbone=4.067967e-07 elapsed=55:57:06 avg_step=4.19s recent_step=4.16s eta=02:14:47 +2026-09-22 14:24:24 | INFO | train rank=0 | step=48080/50000 progress=96.2% loss=0.305841 lr=4.025977e-07 lr_backbone=4.025977e-07 elapsed=55:57:48 avg_step=4.19s recent_step=4.14s eta=02:14:05 +2026-09-22 14:25:06 | INFO | train rank=0 | step=48090/50000 progress=96.2% loss=0.292772 lr=3.984205e-07 lr_backbone=3.984205e-07 elapsed=55:58:29 avg_step=4.19s recent_step=4.16s eta=02:13:23 +2026-09-22 14:25:48 | INFO | train rank=0 | step=48100/50000 progress=96.2% loss=0.328844 lr=3.942649e-07 lr_backbone=3.942649e-07 elapsed=55:59:11 avg_step=4.19s recent_step=4.16s eta=02:12:41 +2026-09-22 14:26:29 | INFO | train rank=0 | step=48110/50000 progress=96.2% loss=0.280104 lr=3.901311e-07 lr_backbone=3.901311e-07 elapsed=55:59:52 avg_step=4.19s recent_step=4.12s eta=02:11:59 +2026-09-22 14:27:10 | INFO | train rank=0 | step=48120/50000 progress=96.2% loss=0.392688 lr=3.860189e-07 lr_backbone=3.860189e-07 elapsed=56:00:33 avg_step=4.19s recent_step=4.11s eta=02:11:17 +2026-09-22 14:27:51 | INFO | train rank=0 | step=48130/50000 progress=96.3% loss=0.372008 lr=3.819285e-07 lr_backbone=3.819285e-07 elapsed=56:01:15 avg_step=4.19s recent_step=4.15s eta=02:10:35 +2026-09-22 14:28:33 | INFO | train rank=0 | step=48140/50000 progress=96.3% loss=0.326498 lr=3.778598e-07 lr_backbone=3.778598e-07 elapsed=56:01:56 avg_step=4.19s recent_step=4.15s eta=02:09:53 +2026-09-22 14:29:14 | INFO | train rank=0 | step=48150/50000 progress=96.3% loss=0.268246 lr=3.738127e-07 lr_backbone=3.738127e-07 elapsed=56:02:38 avg_step=4.19s recent_step=4.15s eta=02:09:11 +2026-09-22 14:29:56 | INFO | train rank=0 | step=48160/50000 progress=96.3% loss=0.318887 lr=3.697874e-07 lr_backbone=3.697874e-07 elapsed=56:03:19 avg_step=4.19s recent_step=4.16s eta=02:08:29 +2026-09-22 14:30:37 | INFO | train rank=0 | step=48170/50000 progress=96.3% loss=0.275799 lr=3.657838e-07 lr_backbone=3.657838e-07 elapsed=56:04:01 avg_step=4.19s recent_step=4.13s eta=02:07:48 +2026-09-22 14:31:18 | INFO | train rank=0 | step=48180/50000 progress=96.4% loss=0.364024 lr=3.618019e-07 lr_backbone=3.618019e-07 elapsed=56:04:42 avg_step=4.19s recent_step=4.11s eta=02:07:06 +2026-09-22 14:32:00 | INFO | train rank=0 | step=48190/50000 progress=96.4% loss=0.323527 lr=3.578417e-07 lr_backbone=3.578417e-07 elapsed=56:05:23 avg_step=4.19s recent_step=4.15s eta=02:06:24 +2026-09-22 14:32:41 | INFO | train rank=0 | step=48200/50000 progress=96.4% loss=0.366282 lr=3.539033e-07 lr_backbone=3.539033e-07 elapsed=56:06:05 avg_step=4.19s recent_step=4.13s eta=02:05:42 +2026-09-22 14:33:23 | INFO | train rank=0 | step=48210/50000 progress=96.4% loss=0.304061 lr=3.499865e-07 lr_backbone=3.499865e-07 elapsed=56:06:46 avg_step=4.19s recent_step=4.15s eta=02:05:00 +2026-09-22 14:34:04 | INFO | train rank=0 | step=48220/50000 progress=96.4% loss=0.329131 lr=3.460915e-07 lr_backbone=3.460915e-07 elapsed=56:07:28 avg_step=4.19s recent_step=4.16s eta=02:04:18 +2026-09-22 14:34:45 | INFO | train rank=0 | step=48230/50000 progress=96.5% loss=0.320564 lr=3.422181e-07 lr_backbone=3.422181e-07 elapsed=56:08:09 avg_step=4.19s recent_step=4.11s eta=02:03:36 +2026-09-22 14:35:26 | INFO | train rank=0 | step=48240/50000 progress=96.5% loss=0.290211 lr=3.383666e-07 lr_backbone=3.383666e-07 elapsed=56:08:50 avg_step=4.19s recent_step=4.10s eta=02:02:54 +2026-09-22 14:36:08 | INFO | train rank=0 | step=48250/50000 progress=96.5% loss=0.321315 lr=3.345367e-07 lr_backbone=3.345367e-07 elapsed=56:09:31 avg_step=4.19s recent_step=4.15s eta=02:02:12 +2026-09-22 14:36:49 | INFO | train rank=0 | step=48260/50000 progress=96.5% loss=0.310114 lr=3.307285e-07 lr_backbone=3.307285e-07 elapsed=56:10:13 avg_step=4.19s recent_step=4.16s eta=02:01:30 +2026-09-22 14:37:31 | INFO | train rank=0 | step=48270/50000 progress=96.5% loss=0.380345 lr=3.269421e-07 lr_backbone=3.269421e-07 elapsed=56:10:54 avg_step=4.19s recent_step=4.16s eta=02:00:48 +2026-09-22 14:38:13 | INFO | train rank=0 | step=48280/50000 progress=96.6% loss=0.322852 lr=3.231774e-07 lr_backbone=3.231774e-07 elapsed=56:11:36 avg_step=4.19s recent_step=4.15s eta=02:00:06 +2026-09-22 14:38:54 | INFO | train rank=0 | step=48290/50000 progress=96.6% loss=0.348490 lr=3.194345e-07 lr_backbone=3.194345e-07 elapsed=56:12:17 avg_step=4.19s recent_step=4.09s eta=01:59:24 +2026-09-22 14:39:35 | INFO | train rank=0 | step=48300/50000 progress=96.6% loss=0.276550 lr=3.157132e-07 lr_backbone=3.157132e-07 elapsed=56:12:58 avg_step=4.19s recent_step=4.13s eta=01:58:43 +2026-09-22 14:40:16 | INFO | train rank=0 | step=48310/50000 progress=96.6% loss=0.332616 lr=3.120138e-07 lr_backbone=3.120138e-07 elapsed=56:13:40 avg_step=4.19s recent_step=4.16s eta=01:58:01 +2026-09-22 14:40:58 | INFO | train rank=0 | step=48320/50000 progress=96.6% loss=0.378611 lr=3.083360e-07 lr_backbone=3.083360e-07 elapsed=56:14:22 avg_step=4.19s recent_step=4.19s eta=01:57:19 +2026-09-22 14:41:40 | INFO | train rank=0 | step=48330/50000 progress=96.7% loss=0.295146 lr=3.046800e-07 lr_backbone=3.046800e-07 elapsed=56:15:04 avg_step=4.19s recent_step=4.18s eta=01:56:37 +2026-09-22 14:42:22 | INFO | train rank=0 | step=48340/50000 progress=96.7% loss=0.275766 lr=3.010457e-07 lr_backbone=3.010457e-07 elapsed=56:15:45 avg_step=4.19s recent_step=4.15s eta=01:55:55 +2026-09-22 14:43:03 | INFO | train rank=0 | step=48350/50000 progress=96.7% loss=0.296881 lr=2.974332e-07 lr_backbone=2.974332e-07 elapsed=56:16:26 avg_step=4.19s recent_step=4.13s eta=01:55:13 +2026-09-22 14:43:44 | INFO | train rank=0 | step=48360/50000 progress=96.7% loss=0.333244 lr=2.938424e-07 lr_backbone=2.938424e-07 elapsed=56:17:08 avg_step=4.19s recent_step=4.13s eta=01:54:31 +2026-09-22 14:44:26 | INFO | train rank=0 | step=48370/50000 progress=96.7% loss=0.338871 lr=2.902733e-07 lr_backbone=2.902733e-07 elapsed=56:17:49 avg_step=4.19s recent_step=4.14s eta=01:53:49 +2026-09-22 14:45:07 | INFO | train rank=0 | step=48380/50000 progress=96.8% loss=0.324972 lr=2.867260e-07 lr_backbone=2.867260e-07 elapsed=56:18:31 avg_step=4.19s recent_step=4.18s eta=01:53:07 +2026-09-22 14:45:49 | INFO | train rank=0 | step=48390/50000 progress=96.8% loss=0.317789 lr=2.832004e-07 lr_backbone=2.832004e-07 elapsed=56:19:13 avg_step=4.19s recent_step=4.18s eta=01:52:25 +2026-09-22 14:46:31 | INFO | train rank=0 | step=48400/50000 progress=96.8% loss=0.417417 lr=2.796966e-07 lr_backbone=2.796966e-07 elapsed=56:19:54 avg_step=4.19s recent_step=4.18s eta=01:51:43 +2026-09-22 14:47:12 | INFO | train rank=0 | step=48410/50000 progress=96.8% loss=0.270248 lr=2.762145e-07 lr_backbone=2.762145e-07 elapsed=56:20:36 avg_step=4.19s recent_step=4.11s eta=01:51:02 +2026-09-22 14:47:53 | INFO | train rank=0 | step=48420/50000 progress=96.8% loss=0.381814 lr=2.727542e-07 lr_backbone=2.727542e-07 elapsed=56:21:17 avg_step=4.19s recent_step=4.13s eta=01:50:20 +2026-09-22 14:48:35 | INFO | train rank=0 | step=48430/50000 progress=96.9% loss=0.272880 lr=2.693156e-07 lr_backbone=2.693156e-07 elapsed=56:21:58 avg_step=4.19s recent_step=4.15s eta=01:49:38 +2026-09-22 14:49:17 | INFO | train rank=0 | step=48440/50000 progress=96.9% loss=0.292286 lr=2.658988e-07 lr_backbone=2.658988e-07 elapsed=56:22:40 avg_step=4.19s recent_step=4.17s eta=01:48:56 +2026-09-22 14:49:59 | INFO | train rank=0 | step=48450/50000 progress=96.9% loss=0.324835 lr=2.625038e-07 lr_backbone=2.625038e-07 elapsed=56:23:22 avg_step=4.19s recent_step=4.19s eta=01:48:14 +2026-09-22 14:50:40 | INFO | train rank=0 | step=48460/50000 progress=96.9% loss=0.285892 lr=2.591305e-07 lr_backbone=2.591305e-07 elapsed=56:24:04 avg_step=4.19s recent_step=4.18s eta=01:47:32 +2026-09-22 14:51:22 | INFO | train rank=0 | step=48470/50000 progress=96.9% loss=0.304928 lr=2.557789e-07 lr_backbone=2.557789e-07 elapsed=56:24:45 avg_step=4.19s recent_step=4.13s eta=01:46:50 +2026-09-22 14:52:03 | INFO | train rank=0 | step=48480/50000 progress=97.0% loss=0.288712 lr=2.524492e-07 lr_backbone=2.524492e-07 elapsed=56:25:26 avg_step=4.19s recent_step=4.11s eta=01:46:08 +2026-09-22 14:52:45 | INFO | train rank=0 | step=48490/50000 progress=97.0% loss=0.325747 lr=2.491411e-07 lr_backbone=2.491411e-07 elapsed=56:26:08 avg_step=4.19s recent_step=4.18s eta=01:45:26 +2026-09-22 14:53:26 | INFO | train rank=0 | step=48500/50000 progress=97.0% loss=0.339745 lr=2.458549e-07 lr_backbone=2.458549e-07 elapsed=56:26:49 avg_step=4.19s recent_step=4.15s eta=01:44:44 +2026-09-22 14:54:08 | INFO | train rank=0 | step=48510/50000 progress=97.0% loss=0.341743 lr=2.425904e-07 lr_backbone=2.425904e-07 elapsed=56:27:31 avg_step=4.19s recent_step=4.18s eta=01:44:02 +2026-09-22 14:54:49 | INFO | train rank=0 | step=48520/50000 progress=97.0% loss=0.275786 lr=2.393477e-07 lr_backbone=2.393477e-07 elapsed=56:28:13 avg_step=4.19s recent_step=4.15s eta=01:43:21 +2026-09-22 14:55:31 | INFO | train rank=0 | step=48530/50000 progress=97.1% loss=0.332879 lr=2.361267e-07 lr_backbone=2.361267e-07 elapsed=56:28:54 avg_step=4.19s recent_step=4.14s eta=01:42:39 +2026-09-22 14:56:12 | INFO | train rank=0 | step=48540/50000 progress=97.1% loss=0.300852 lr=2.329275e-07 lr_backbone=2.329275e-07 elapsed=56:29:35 avg_step=4.19s recent_step=4.12s eta=01:41:57 +2026-09-22 14:56:54 | INFO | train rank=0 | step=48550/50000 progress=97.1% loss=0.386408 lr=2.297501e-07 lr_backbone=2.297501e-07 elapsed=56:30:17 avg_step=4.19s recent_step=4.19s eta=01:41:15 +2026-09-22 14:57:35 | INFO | train rank=0 | step=48560/50000 progress=97.1% loss=0.391284 lr=2.265944e-07 lr_backbone=2.265944e-07 elapsed=56:30:59 avg_step=4.19s recent_step=4.16s eta=01:40:33 +2026-09-22 14:58:17 | INFO | train rank=0 | step=48570/50000 progress=97.1% loss=0.288209 lr=2.234605e-07 lr_backbone=2.234605e-07 elapsed=56:31:40 avg_step=4.19s recent_step=4.18s eta=01:39:51 +2026-09-22 14:58:59 | INFO | train rank=0 | step=48580/50000 progress=97.2% loss=0.347336 lr=2.203484e-07 lr_backbone=2.203484e-07 elapsed=56:32:22 avg_step=4.19s recent_step=4.18s eta=01:39:09 +2026-09-22 14:59:40 | INFO | train rank=0 | step=48590/50000 progress=97.2% loss=0.314854 lr=2.172581e-07 lr_backbone=2.172581e-07 elapsed=56:33:04 avg_step=4.19s recent_step=4.14s eta=01:38:27 +2026-09-22 15:00:22 | INFO | train rank=0 | step=48600/50000 progress=97.2% loss=0.354728 lr=2.141896e-07 lr_backbone=2.141896e-07 elapsed=56:33:45 avg_step=4.19s recent_step=4.12s eta=01:37:45 +2026-09-22 15:01:03 | INFO | train rank=0 | step=48610/50000 progress=97.2% loss=0.306542 lr=2.111428e-07 lr_backbone=2.111428e-07 elapsed=56:34:27 avg_step=4.19s recent_step=4.17s eta=01:37:03 +2026-09-22 15:01:45 | INFO | train rank=0 | step=48620/50000 progress=97.2% loss=0.241895 lr=2.081178e-07 lr_backbone=2.081178e-07 elapsed=56:35:08 avg_step=4.19s recent_step=4.15s eta=01:36:21 +2026-09-22 15:02:27 | INFO | train rank=0 | step=48630/50000 progress=97.3% loss=0.297630 lr=2.051146e-07 lr_backbone=2.051146e-07 elapsed=56:35:50 avg_step=4.19s recent_step=4.18s eta=01:35:40 +2026-09-22 15:03:08 | INFO | train rank=0 | step=48640/50000 progress=97.3% loss=0.327134 lr=2.021331e-07 lr_backbone=2.021331e-07 elapsed=56:36:32 avg_step=4.19s recent_step=4.17s eta=01:34:58 +2026-09-22 15:03:50 | INFO | train rank=0 | step=48650/50000 progress=97.3% loss=0.288570 lr=1.991735e-07 lr_backbone=1.991735e-07 elapsed=56:37:13 avg_step=4.19s recent_step=4.13s eta=01:34:16 +2026-09-22 15:04:31 | INFO | train rank=0 | step=48660/50000 progress=97.3% loss=0.339296 lr=1.962356e-07 lr_backbone=1.962356e-07 elapsed=56:37:54 avg_step=4.19s recent_step=4.13s eta=01:33:34 +2026-09-22 15:05:13 | INFO | train rank=0 | step=48670/50000 progress=97.3% loss=0.321743 lr=1.933195e-07 lr_backbone=1.933195e-07 elapsed=56:38:36 avg_step=4.19s recent_step=4.19s eta=01:32:52 +2026-09-22 15:05:54 | INFO | train rank=0 | step=48680/50000 progress=97.4% loss=0.341242 lr=1.904253e-07 lr_backbone=1.904253e-07 elapsed=56:39:18 avg_step=4.19s recent_step=4.16s eta=01:32:10 +2026-09-22 15:06:36 | INFO | train rank=0 | step=48690/50000 progress=97.4% loss=0.320855 lr=1.875527e-07 lr_backbone=1.875527e-07 elapsed=56:39:59 avg_step=4.19s recent_step=4.14s eta=01:31:28 +2026-09-22 15:07:18 | INFO | train rank=0 | step=48700/50000 progress=97.4% loss=0.347498 lr=1.847020e-07 lr_backbone=1.847020e-07 elapsed=56:40:41 avg_step=4.19s recent_step=4.19s eta=01:30:46 +2026-09-22 15:07:59 | INFO | train rank=0 | step=48710/50000 progress=97.4% loss=0.307792 lr=1.818731e-07 lr_backbone=1.818731e-07 elapsed=56:41:22 avg_step=4.19s recent_step=4.12s eta=01:30:04 +2026-09-22 15:08:40 | INFO | train rank=0 | step=48720/50000 progress=97.4% loss=0.287562 lr=1.790660e-07 lr_backbone=1.790660e-07 elapsed=56:42:04 avg_step=4.19s recent_step=4.13s eta=01:29:22 +2026-09-22 15:09:22 | INFO | train rank=0 | step=48730/50000 progress=97.5% loss=0.304016 lr=1.762806e-07 lr_backbone=1.762806e-07 elapsed=56:42:45 avg_step=4.19s recent_step=4.17s eta=01:28:40 +2026-09-22 15:10:04 | INFO | train rank=0 | step=48740/50000 progress=97.5% loss=0.340126 lr=1.735171e-07 lr_backbone=1.735171e-07 elapsed=56:43:27 avg_step=4.19s recent_step=4.18s eta=01:27:59 +2026-09-22 15:10:45 | INFO | train rank=0 | step=48750/50000 progress=97.5% loss=0.330425 lr=1.707753e-07 lr_backbone=1.707753e-07 elapsed=56:44:09 avg_step=4.19s recent_step=4.15s eta=01:27:17 +2026-09-22 15:11:27 | INFO | train rank=0 | step=48760/50000 progress=97.5% loss=0.347278 lr=1.680554e-07 lr_backbone=1.680554e-07 elapsed=56:44:50 avg_step=4.19s recent_step=4.18s eta=01:26:35 +2026-09-22 15:12:08 | INFO | train rank=0 | step=48770/50000 progress=97.5% loss=0.304567 lr=1.653572e-07 lr_backbone=1.653572e-07 elapsed=56:45:32 avg_step=4.19s recent_step=4.13s eta=01:25:53 +2026-09-22 15:12:50 | INFO | train rank=0 | step=48780/50000 progress=97.6% loss=0.348410 lr=1.626809e-07 lr_backbone=1.626809e-07 elapsed=56:46:13 avg_step=4.19s recent_step=4.13s eta=01:25:11 +2026-09-22 15:13:32 | INFO | train rank=0 | step=48790/50000 progress=97.6% loss=0.359947 lr=1.600263e-07 lr_backbone=1.600263e-07 elapsed=56:46:55 avg_step=4.19s recent_step=4.19s eta=01:24:29 +2026-09-22 15:14:13 | INFO | train rank=0 | step=48800/50000 progress=97.6% loss=0.307456 lr=1.573936e-07 lr_backbone=1.573936e-07 elapsed=56:47:37 avg_step=4.19s recent_step=4.17s eta=01:23:47 +2026-09-22 15:14:55 | INFO | train rank=0 | step=48810/50000 progress=97.6% loss=0.341400 lr=1.547826e-07 lr_backbone=1.547826e-07 elapsed=56:48:18 avg_step=4.19s recent_step=4.16s eta=01:23:05 +2026-09-22 15:15:37 | INFO | train rank=0 | step=48820/50000 progress=97.6% loss=0.287917 lr=1.521935e-07 lr_backbone=1.521935e-07 elapsed=56:49:00 avg_step=4.19s recent_step=4.19s eta=01:22:23 +2026-09-22 15:16:18 | INFO | train rank=0 | step=48830/50000 progress=97.7% loss=0.266382 lr=1.496262e-07 lr_backbone=1.496262e-07 elapsed=56:49:41 avg_step=4.19s recent_step=4.12s eta=01:21:41 +2026-09-22 15:16:59 | INFO | train rank=0 | step=48840/50000 progress=97.7% loss=0.372004 lr=1.470806e-07 lr_backbone=1.470806e-07 elapsed=56:50:23 avg_step=4.19s recent_step=4.12s eta=01:21:00 +2026-09-22 15:17:41 | INFO | train rank=0 | step=48850/50000 progress=97.7% loss=0.314813 lr=1.445569e-07 lr_backbone=1.445569e-07 elapsed=56:51:04 avg_step=4.19s recent_step=4.17s eta=01:20:18 +2026-09-22 15:18:23 | INFO | train rank=0 | step=48860/50000 progress=97.7% loss=0.379064 lr=1.420550e-07 lr_backbone=1.420550e-07 elapsed=56:51:46 avg_step=4.19s recent_step=4.17s eta=01:19:36 +2026-09-22 15:19:04 | INFO | train rank=0 | step=48870/50000 progress=97.7% loss=0.308796 lr=1.395749e-07 lr_backbone=1.395749e-07 elapsed=56:52:27 avg_step=4.19s recent_step=4.12s eta=01:18:54 +2026-09-22 15:19:46 | INFO | train rank=0 | step=48880/50000 progress=97.8% loss=0.328429 lr=1.371166e-07 lr_backbone=1.371166e-07 elapsed=56:53:09 avg_step=4.19s recent_step=4.18s eta=01:18:12 +2026-09-22 15:20:27 | INFO | train rank=0 | step=48890/50000 progress=97.8% loss=0.268353 lr=1.346801e-07 lr_backbone=1.346801e-07 elapsed=56:53:50 avg_step=4.19s recent_step=4.14s eta=01:17:30 +2026-09-22 15:21:08 | INFO | train rank=0 | step=48900/50000 progress=97.8% loss=0.338786 lr=1.322654e-07 lr_backbone=1.322654e-07 elapsed=56:54:32 avg_step=4.19s recent_step=4.13s eta=01:16:48 +2026-09-22 15:21:50 | INFO | train rank=0 | step=48910/50000 progress=97.8% loss=0.277596 lr=1.298726e-07 lr_backbone=1.298726e-07 elapsed=56:55:13 avg_step=4.19s recent_step=4.18s eta=01:16:06 +2026-09-22 15:22:32 | INFO | train rank=0 | step=48920/50000 progress=97.8% loss=0.281774 lr=1.275015e-07 lr_backbone=1.275015e-07 elapsed=56:55:55 avg_step=4.19s recent_step=4.18s eta=01:15:24 +2026-09-22 15:23:13 | INFO | train rank=0 | step=48930/50000 progress=97.9% loss=0.264705 lr=1.251523e-07 lr_backbone=1.251523e-07 elapsed=56:56:37 avg_step=4.19s recent_step=4.15s eta=01:14:42 +2026-09-22 15:23:55 | INFO | train rank=0 | step=48940/50000 progress=97.9% loss=0.317792 lr=1.228249e-07 lr_backbone=1.228249e-07 elapsed=56:57:18 avg_step=4.19s recent_step=4.16s eta=01:14:00 +2026-09-22 15:24:36 | INFO | train rank=0 | step=48950/50000 progress=97.9% loss=0.287294 lr=1.205193e-07 lr_backbone=1.205193e-07 elapsed=56:58:00 avg_step=4.19s recent_step=4.13s eta=01:13:19 +2026-09-22 15:25:18 | INFO | train rank=0 | step=48960/50000 progress=97.9% loss=0.404425 lr=1.182355e-07 lr_backbone=1.182355e-07 elapsed=56:58:41 avg_step=4.19s recent_step=4.13s eta=01:12:37 +2026-09-22 15:25:59 | INFO | train rank=0 | step=48970/50000 progress=97.9% loss=0.384664 lr=1.159736e-07 lr_backbone=1.159736e-07 elapsed=56:59:23 avg_step=4.19s recent_step=4.18s eta=01:11:55 +2026-09-22 15:26:41 | INFO | train rank=0 | step=48980/50000 progress=98.0% loss=0.382155 lr=1.137334e-07 lr_backbone=1.137334e-07 elapsed=57:00:04 avg_step=4.19s recent_step=4.17s eta=01:11:13 +2026-09-22 15:27:23 | INFO | train rank=0 | step=48990/50000 progress=98.0% loss=0.288292 lr=1.115151e-07 lr_backbone=1.115151e-07 elapsed=57:00:46 avg_step=4.19s recent_step=4.16s eta=01:10:31 +2026-09-22 15:28:04 | INFO | train rank=0 | step=49000/50000 progress=98.0% loss=0.343789 lr=1.093186e-07 lr_backbone=1.093186e-07 elapsed=57:01:28 avg_step=4.19s recent_step=4.18s eta=01:09:49 +2026-09-22 15:28:46 | INFO | train rank=0 | step=49010/50000 progress=98.0% loss=0.309455 lr=1.071440e-07 lr_backbone=1.071440e-07 elapsed=57:02:09 avg_step=4.19s recent_step=4.14s eta=01:09:07 +2026-09-22 15:29:27 | INFO | train rank=0 | step=49020/50000 progress=98.0% loss=0.294538 lr=1.049911e-07 lr_backbone=1.049911e-07 elapsed=57:02:50 avg_step=4.19s recent_step=4.13s eta=01:08:25 +2026-09-22 15:30:09 | INFO | train rank=0 | step=49030/50000 progress=98.1% loss=0.304228 lr=1.028601e-07 lr_backbone=1.028601e-07 elapsed=57:03:32 avg_step=4.19s recent_step=4.16s eta=01:07:43 +2026-09-22 15:30:50 | INFO | train rank=0 | step=49040/50000 progress=98.1% loss=0.279815 lr=1.007509e-07 lr_backbone=1.007509e-07 elapsed=57:04:14 avg_step=4.19s recent_step=4.17s eta=01:07:01 +2026-09-22 15:31:32 | INFO | train rank=0 | step=49050/50000 progress=98.1% loss=0.273636 lr=9.866358e-08 lr_backbone=9.866358e-08 elapsed=57:04:55 avg_step=4.19s recent_step=4.15s eta=01:06:20 +2026-09-22 15:32:13 | INFO | train rank=0 | step=49060/50000 progress=98.1% loss=0.371122 lr=9.659805e-08 lr_backbone=9.659805e-08 elapsed=57:05:37 avg_step=4.19s recent_step=4.15s eta=01:05:38 +2026-09-22 15:32:55 | INFO | train rank=0 | step=49070/50000 progress=98.1% loss=0.290953 lr=9.455435e-08 lr_backbone=9.455435e-08 elapsed=57:06:18 avg_step=4.19s recent_step=4.16s eta=01:04:56 +2026-09-22 15:33:36 | INFO | train rank=0 | step=49080/50000 progress=98.2% loss=0.353212 lr=9.253248e-08 lr_backbone=9.253248e-08 elapsed=57:07:00 avg_step=4.19s recent_step=4.12s eta=01:04:14 +2026-09-22 15:34:18 | INFO | train rank=0 | step=49090/50000 progress=98.2% loss=0.398184 lr=9.053244e-08 lr_backbone=9.053244e-08 elapsed=57:07:41 avg_step=4.19s recent_step=4.17s eta=01:03:32 +2026-09-22 15:35:00 | INFO | train rank=0 | step=49100/50000 progress=98.2% loss=0.383528 lr=8.855423e-08 lr_backbone=8.855423e-08 elapsed=57:08:23 avg_step=4.19s recent_step=4.17s eta=01:02:50 +2026-09-22 15:35:41 | INFO | train rank=0 | step=49110/50000 progress=98.2% loss=0.272447 lr=8.659786e-08 lr_backbone=8.659786e-08 elapsed=57:09:05 avg_step=4.19s recent_step=4.18s eta=01:02:08 +2026-09-22 15:36:23 | INFO | train rank=0 | step=49120/50000 progress=98.2% loss=0.278068 lr=8.466332e-08 lr_backbone=8.466332e-08 elapsed=57:09:46 avg_step=4.19s recent_step=4.17s eta=01:01:26 +2026-09-22 15:37:04 | INFO | train rank=0 | step=49130/50000 progress=98.3% loss=0.361490 lr=8.275061e-08 lr_backbone=8.275061e-08 elapsed=57:10:28 avg_step=4.19s recent_step=4.14s eta=01:00:44 +2026-09-22 15:37:46 | INFO | train rank=0 | step=49140/50000 progress=98.3% loss=0.291546 lr=8.085974e-08 lr_backbone=8.085974e-08 elapsed=57:11:09 avg_step=4.19s recent_step=4.14s eta=01:00:02 +2026-09-22 15:38:27 | INFO | train rank=0 | step=49150/50000 progress=98.3% loss=0.304696 lr=7.899071e-08 lr_backbone=7.899071e-08 elapsed=57:11:51 avg_step=4.19s recent_step=4.17s eta=00:59:21 +2026-09-22 15:39:09 | INFO | train rank=0 | step=49160/50000 progress=98.3% loss=0.294362 lr=7.714351e-08 lr_backbone=7.714351e-08 elapsed=57:12:33 avg_step=4.19s recent_step=4.18s eta=00:58:39 +2026-09-22 15:39:51 | INFO | train rank=0 | step=49170/50000 progress=98.3% loss=0.263347 lr=7.531815e-08 lr_backbone=7.531815e-08 elapsed=57:13:14 avg_step=4.19s recent_step=4.18s eta=00:57:57 +2026-09-22 15:40:33 | INFO | train rank=0 | step=49180/50000 progress=98.4% loss=0.272397 lr=7.351463e-08 lr_backbone=7.351463e-08 elapsed=57:13:56 avg_step=4.19s recent_step=4.16s eta=00:57:15 +2026-09-22 15:41:14 | INFO | train rank=0 | step=49190/50000 progress=98.4% loss=0.336473 lr=7.173295e-08 lr_backbone=7.173295e-08 elapsed=57:14:37 avg_step=4.19s recent_step=4.11s eta=00:56:33 +2026-09-22 15:41:55 | INFO | train rank=0 | step=49200/50000 progress=98.4% loss=0.331072 lr=6.997311e-08 lr_backbone=6.997311e-08 elapsed=57:15:18 avg_step=4.19s recent_step=4.13s eta=00:55:51 +2026-09-22 15:42:37 | INFO | train rank=0 | step=49210/50000 progress=98.4% loss=0.319458 lr=6.823511e-08 lr_backbone=6.823511e-08 elapsed=57:16:00 avg_step=4.19s recent_step=4.19s eta=00:55:09 +2026-09-22 15:43:18 | INFO | train rank=0 | step=49220/50000 progress=98.4% loss=0.319334 lr=6.651896e-08 lr_backbone=6.651896e-08 elapsed=57:16:42 avg_step=4.19s recent_step=4.15s eta=00:54:27 +2026-09-22 15:44:00 | INFO | train rank=0 | step=49230/50000 progress=98.5% loss=0.333758 lr=6.482464e-08 lr_backbone=6.482464e-08 elapsed=57:17:24 avg_step=4.19s recent_step=4.18s eta=00:53:45 +2026-09-22 15:44:42 | INFO | train rank=0 | step=49240/50000 progress=98.5% loss=0.286485 lr=6.315217e-08 lr_backbone=6.315217e-08 elapsed=57:18:05 avg_step=4.19s recent_step=4.16s eta=00:53:03 +2026-09-22 15:45:23 | INFO | train rank=0 | step=49250/50000 progress=98.5% loss=0.301229 lr=6.150154e-08 lr_backbone=6.150154e-08 elapsed=57:18:47 avg_step=4.19s recent_step=4.16s eta=00:52:22 +2026-09-22 15:46:05 | INFO | train rank=0 | step=49260/50000 progress=98.5% loss=0.280729 lr=5.987276e-08 lr_backbone=5.987276e-08 elapsed=57:19:28 avg_step=4.19s recent_step=4.12s eta=00:51:40 +2026-09-22 15:46:46 | INFO | train rank=0 | step=49270/50000 progress=98.5% loss=0.360428 lr=5.826582e-08 lr_backbone=5.826582e-08 elapsed=57:20:10 avg_step=4.19s recent_step=4.17s eta=00:50:58 +2026-09-22 15:47:28 | INFO | train rank=0 | step=49280/50000 progress=98.6% loss=0.265909 lr=5.668073e-08 lr_backbone=5.668073e-08 elapsed=57:20:51 avg_step=4.19s recent_step=4.18s eta=00:50:16 +2026-09-22 15:48:09 | INFO | train rank=0 | step=49290/50000 progress=98.6% loss=0.274531 lr=5.511749e-08 lr_backbone=5.511749e-08 elapsed=57:21:33 avg_step=4.19s recent_step=4.14s eta=00:49:34 +2026-09-22 15:48:51 | INFO | train rank=0 | step=49300/50000 progress=98.6% loss=0.287925 lr=5.357609e-08 lr_backbone=5.357609e-08 elapsed=57:22:14 avg_step=4.19s recent_step=4.15s eta=00:48:52 +2026-09-22 15:49:32 | INFO | train rank=0 | step=49310/50000 progress=98.6% loss=0.285609 lr=5.205655e-08 lr_backbone=5.205655e-08 elapsed=57:22:55 avg_step=4.19s recent_step=4.09s eta=00:48:10 +2026-09-22 15:50:13 | INFO | train rank=0 | step=49320/50000 progress=98.6% loss=0.271699 lr=5.055885e-08 lr_backbone=5.055885e-08 elapsed=57:23:36 avg_step=4.19s recent_step=4.10s eta=00:47:28 +2026-09-22 15:50:55 | INFO | train rank=0 | step=49330/50000 progress=98.7% loss=0.284169 lr=4.908300e-08 lr_backbone=4.908300e-08 elapsed=57:24:18 avg_step=4.19s recent_step=4.16s eta=00:46:46 +2026-09-22 15:51:36 | INFO | train rank=0 | step=49340/50000 progress=98.7% loss=0.335566 lr=4.762900e-08 lr_backbone=4.762900e-08 elapsed=57:25:00 avg_step=4.19s recent_step=4.17s eta=00:46:04 +2026-09-22 15:52:18 | INFO | train rank=0 | step=49350/50000 progress=98.7% loss=0.280423 lr=4.619685e-08 lr_backbone=4.619685e-08 elapsed=57:25:41 avg_step=4.19s recent_step=4.17s eta=00:45:23 +2026-09-22 15:53:00 | INFO | train rank=0 | step=49360/50000 progress=98.7% loss=0.279578 lr=4.478655e-08 lr_backbone=4.478655e-08 elapsed=57:26:23 avg_step=4.19s recent_step=4.16s eta=00:44:41 +2026-09-22 15:53:41 | INFO | train rank=0 | step=49370/50000 progress=98.7% loss=0.392295 lr=4.339811e-08 lr_backbone=4.339811e-08 elapsed=57:27:04 avg_step=4.19s recent_step=4.11s eta=00:43:59 +2026-09-22 15:54:22 | INFO | train rank=0 | step=49380/50000 progress=98.8% loss=0.302125 lr=4.203152e-08 lr_backbone=4.203152e-08 elapsed=57:27:45 avg_step=4.19s recent_step=4.11s eta=00:43:17 +2026-09-22 15:55:03 | INFO | train rank=0 | step=49390/50000 progress=98.8% loss=0.335347 lr=4.068678e-08 lr_backbone=4.068678e-08 elapsed=57:28:27 avg_step=4.19s recent_step=4.15s eta=00:42:35 +2026-09-22 15:55:45 | INFO | train rank=0 | step=49400/50000 progress=98.8% loss=0.342910 lr=3.936389e-08 lr_backbone=3.936389e-08 elapsed=57:29:08 avg_step=4.19s recent_step=4.15s eta=00:41:53 +2026-09-22 15:56:26 | INFO | train rank=0 | step=49410/50000 progress=98.8% loss=0.297433 lr=3.806286e-08 lr_backbone=3.806286e-08 elapsed=57:29:50 avg_step=4.19s recent_step=4.15s eta=00:41:11 +2026-09-22 15:57:08 | INFO | train rank=0 | step=49420/50000 progress=98.8% loss=0.329394 lr=3.678369e-08 lr_backbone=3.678369e-08 elapsed=57:30:31 avg_step=4.19s recent_step=4.17s eta=00:40:29 +2026-09-22 15:57:49 | INFO | train rank=0 | step=49430/50000 progress=98.9% loss=0.314289 lr=3.552637e-08 lr_backbone=3.552637e-08 elapsed=57:31:12 avg_step=4.19s recent_step=4.11s eta=00:39:47 +2026-09-22 15:58:30 | INFO | train rank=0 | step=49440/50000 progress=98.9% loss=0.282216 lr=3.429090e-08 lr_backbone=3.429090e-08 elapsed=57:31:53 avg_step=4.19s recent_step=4.10s eta=00:39:05 +2026-09-22 15:59:11 | INFO | train rank=0 | step=49450/50000 progress=98.9% loss=0.340877 lr=3.307730e-08 lr_backbone=3.307730e-08 elapsed=57:32:35 avg_step=4.19s recent_step=4.13s eta=00:38:24 +2026-09-22 15:59:53 | INFO | train rank=0 | step=49460/50000 progress=98.9% loss=0.273327 lr=3.188555e-08 lr_backbone=3.188555e-08 elapsed=57:33:16 avg_step=4.19s recent_step=4.16s eta=00:37:42 +2026-09-22 16:00:34 | INFO | train rank=0 | step=49470/50000 progress=98.9% loss=0.376072 lr=3.071566e-08 lr_backbone=3.071566e-08 elapsed=57:33:58 avg_step=4.19s recent_step=4.15s eta=00:37:00 +2026-09-22 16:01:16 | INFO | train rank=0 | step=49480/50000 progress=99.0% loss=0.287618 lr=2.956762e-08 lr_backbone=2.956762e-08 elapsed=57:34:39 avg_step=4.19s recent_step=4.16s eta=00:36:18 +2026-09-22 16:01:57 | INFO | train rank=0 | step=49490/50000 progress=99.0% loss=0.334723 lr=2.844145e-08 lr_backbone=2.844145e-08 elapsed=57:35:20 avg_step=4.19s recent_step=4.10s eta=00:35:36 +2026-09-22 16:02:38 | INFO | train rank=0 | step=49500/50000 progress=99.0% loss=0.265247 lr=2.733713e-08 lr_backbone=2.733713e-08 elapsed=57:36:01 avg_step=4.19s recent_step=4.12s eta=00:34:54 +2026-09-22 16:03:20 | INFO | train rank=0 | step=49510/50000 progress=99.0% loss=0.313477 lr=2.625468e-08 lr_backbone=2.625468e-08 elapsed=57:36:43 avg_step=4.19s recent_step=4.14s eta=00:34:12 +2026-09-22 16:04:01 | INFO | train rank=0 | step=49520/50000 progress=99.0% loss=0.291311 lr=2.519408e-08 lr_backbone=2.519408e-08 elapsed=57:37:25 avg_step=4.19s recent_step=4.16s eta=00:33:30 +2026-09-22 16:04:43 | INFO | train rank=0 | step=49530/50000 progress=99.1% loss=0.375105 lr=2.415535e-08 lr_backbone=2.415535e-08 elapsed=57:38:06 avg_step=4.19s recent_step=4.15s eta=00:32:48 +2026-09-22 16:05:24 | INFO | train rank=0 | step=49540/50000 progress=99.1% loss=0.393288 lr=2.313847e-08 lr_backbone=2.313847e-08 elapsed=57:38:48 avg_step=4.19s recent_step=4.15s eta=00:32:06 +2026-09-22 16:06:05 | INFO | train rank=0 | step=49550/50000 progress=99.1% loss=0.376568 lr=2.214346e-08 lr_backbone=2.214346e-08 elapsed=57:39:28 avg_step=4.19s recent_step=4.10s eta=00:31:25 +2026-09-22 16:06:46 | INFO | train rank=0 | step=49560/50000 progress=99.1% loss=0.285677 lr=2.117031e-08 lr_backbone=2.117031e-08 elapsed=57:40:09 avg_step=4.19s recent_step=4.09s eta=00:30:43 +2026-09-22 16:07:28 | INFO | train rank=0 | step=49570/50000 progress=99.1% loss=0.306710 lr=2.021902e-08 lr_backbone=2.021902e-08 elapsed=57:40:51 avg_step=4.19s recent_step=4.15s eta=00:30:01 +2026-09-22 16:08:09 | INFO | train rank=0 | step=49580/50000 progress=99.2% loss=0.348497 lr=1.928960e-08 lr_backbone=1.928960e-08 elapsed=57:41:33 avg_step=4.19s recent_step=4.17s eta=00:29:19 +2026-09-22 16:08:51 | INFO | train rank=0 | step=49590/50000 progress=99.2% loss=0.294699 lr=1.838204e-08 lr_backbone=1.838204e-08 elapsed=57:42:14 avg_step=4.19s recent_step=4.16s eta=00:28:37 +2026-09-22 16:09:32 | INFO | train rank=0 | step=49600/50000 progress=99.2% loss=0.272097 lr=1.749634e-08 lr_backbone=1.749634e-08 elapsed=57:42:56 avg_step=4.19s recent_step=4.16s eta=00:27:55 +2026-09-22 16:10:13 | INFO | train rank=0 | step=49610/50000 progress=99.2% loss=0.315772 lr=1.663251e-08 lr_backbone=1.663251e-08 elapsed=57:43:37 avg_step=4.19s recent_step=4.11s eta=00:27:13 +2026-09-22 16:10:55 | INFO | train rank=0 | step=49620/50000 progress=99.2% loss=0.285666 lr=1.579054e-08 lr_backbone=1.579054e-08 elapsed=57:44:18 avg_step=4.19s recent_step=4.14s eta=00:26:31 +2026-09-22 16:11:36 | INFO | train rank=0 | step=49630/50000 progress=99.3% loss=0.290065 lr=1.497043e-08 lr_backbone=1.497043e-08 elapsed=57:44:59 avg_step=4.19s recent_step=4.12s eta=00:25:49 +2026-09-22 16:12:18 | INFO | train rank=0 | step=49640/50000 progress=99.3% loss=0.340957 lr=1.417219e-08 lr_backbone=1.417219e-08 elapsed=57:45:41 avg_step=4.19s recent_step=4.19s eta=00:25:08 +2026-09-22 16:13:00 | INFO | train rank=0 | step=49650/50000 progress=99.3% loss=0.268686 lr=1.339582e-08 lr_backbone=1.339582e-08 elapsed=57:46:23 avg_step=4.19s recent_step=4.18s eta=00:24:26 +2026-09-22 16:13:41 | INFO | train rank=0 | step=49660/50000 progress=99.3% loss=0.323967 lr=1.264131e-08 lr_backbone=1.264131e-08 elapsed=57:47:05 avg_step=4.19s recent_step=4.17s eta=00:23:44 +2026-09-22 16:14:23 | INFO | train rank=0 | step=49670/50000 progress=99.3% loss=0.358774 lr=1.190867e-08 lr_backbone=1.190867e-08 elapsed=57:47:46 avg_step=4.19s recent_step=4.13s eta=00:23:02 +2026-09-22 16:15:04 | INFO | train rank=0 | step=49680/50000 progress=99.4% loss=0.283618 lr=1.119789e-08 lr_backbone=1.119789e-08 elapsed=57:48:27 avg_step=4.19s recent_step=4.13s eta=00:22:20 +2026-09-22 16:15:46 | INFO | train rank=0 | step=49690/50000 progress=99.4% loss=0.272788 lr=1.050898e-08 lr_backbone=1.050898e-08 elapsed=57:49:09 avg_step=4.19s recent_step=4.15s eta=00:21:38 +2026-09-22 16:16:27 | INFO | train rank=0 | step=49700/50000 progress=99.4% loss=0.313385 lr=9.841942e-09 lr_backbone=9.841942e-09 elapsed=57:49:51 avg_step=4.19s recent_step=4.18s eta=00:20:56 +2026-09-22 16:17:09 | INFO | train rank=0 | step=49710/50000 progress=99.4% loss=0.272582 lr=9.196768e-09 lr_backbone=9.196768e-09 elapsed=57:50:32 avg_step=4.19s recent_step=4.18s eta=00:20:14 +2026-09-22 16:17:51 | INFO | train rank=0 | step=49720/50000 progress=99.4% loss=0.380506 lr=8.573461e-09 lr_backbone=8.573461e-09 elapsed=57:51:14 avg_step=4.19s recent_step=4.18s eta=00:19:32 +2026-09-22 16:18:32 | INFO | train rank=0 | step=49730/50000 progress=99.5% loss=0.364648 lr=7.972023e-09 lr_backbone=7.972023e-09 elapsed=57:51:56 avg_step=4.19s recent_step=4.13s eta=00:18:51 +2026-09-22 16:19:13 | INFO | train rank=0 | step=49740/50000 progress=99.5% loss=0.287293 lr=7.392452e-09 lr_backbone=7.392452e-09 elapsed=57:52:37 avg_step=4.19s recent_step=4.11s eta=00:18:09 +2026-09-22 16:19:55 | INFO | train rank=0 | step=49750/50000 progress=99.5% loss=0.316860 lr=6.834750e-09 lr_backbone=6.834750e-09 elapsed=57:53:18 avg_step=4.19s recent_step=4.17s eta=00:17:27 +2026-09-22 16:20:37 | INFO | train rank=0 | step=49760/50000 progress=99.5% loss=0.350589 lr=6.298917e-09 lr_backbone=6.298917e-09 elapsed=57:54:00 avg_step=4.19s recent_step=4.15s eta=00:16:45 +2026-09-22 16:21:18 | INFO | train rank=0 | step=49770/50000 progress=99.5% loss=0.301624 lr=5.784953e-09 lr_backbone=5.784953e-09 elapsed=57:54:42 avg_step=4.19s recent_step=4.17s eta=00:16:03 +2026-09-22 16:22:00 | INFO | train rank=0 | step=49780/50000 progress=99.6% loss=0.303183 lr=5.292858e-09 lr_backbone=5.292858e-09 elapsed=57:55:23 avg_step=4.19s recent_step=4.17s eta=00:15:21 +2026-09-22 16:22:41 | INFO | train rank=0 | step=49790/50000 progress=99.6% loss=0.270731 lr=4.822632e-09 lr_backbone=4.822632e-09 elapsed=57:56:05 avg_step=4.19s recent_step=4.14s eta=00:14:39 +2026-09-22 16:23:23 | INFO | train rank=0 | step=49800/50000 progress=99.6% loss=0.355657 lr=4.374276e-09 lr_backbone=4.374276e-09 elapsed=57:56:46 avg_step=4.19s recent_step=4.14s eta=00:13:57 +2026-09-22 16:24:04 | INFO | train rank=0 | step=49810/50000 progress=99.6% loss=0.300838 lr=3.947790e-09 lr_backbone=3.947790e-09 elapsed=57:57:28 avg_step=4.19s recent_step=4.17s eta=00:13:15 +2026-09-22 16:24:46 | INFO | train rank=0 | step=49820/50000 progress=99.6% loss=0.350914 lr=3.543173e-09 lr_backbone=3.543173e-09 elapsed=57:58:09 avg_step=4.19s recent_step=4.14s eta=00:12:33 +2026-09-22 16:25:28 | INFO | train rank=0 | step=49830/50000 progress=99.7% loss=0.309375 lr=3.160427e-09 lr_backbone=3.160427e-09 elapsed=57:58:51 avg_step=4.19s recent_step=4.17s eta=00:11:52 +2026-09-22 16:26:09 | INFO | train rank=0 | step=49840/50000 progress=99.7% loss=0.274130 lr=2.799551e-09 lr_backbone=2.799551e-09 elapsed=57:59:33 avg_step=4.19s recent_step=4.18s eta=00:11:10 +2026-09-22 16:26:51 | INFO | train rank=0 | step=49850/50000 progress=99.7% loss=0.301689 lr=2.460546e-09 lr_backbone=2.460546e-09 elapsed=58:00:14 avg_step=4.19s recent_step=4.13s eta=00:10:28 +2026-09-22 16:27:32 | INFO | train rank=0 | step=49860/50000 progress=99.7% loss=0.301750 lr=2.143411e-09 lr_backbone=2.143411e-09 elapsed=58:00:55 avg_step=4.19s recent_step=4.12s eta=00:09:46 +2026-09-22 16:28:14 | INFO | train rank=0 | step=49870/50000 progress=99.7% loss=0.323211 lr=1.848147e-09 lr_backbone=1.848147e-09 elapsed=58:01:37 avg_step=4.19s recent_step=4.18s eta=00:09:04 +2026-09-22 16:28:55 | INFO | train rank=0 | step=49880/50000 progress=99.8% loss=0.301573 lr=1.574754e-09 lr_backbone=1.574754e-09 elapsed=58:02:18 avg_step=4.19s recent_step=4.15s eta=00:08:22 +2026-09-22 16:29:37 | INFO | train rank=0 | step=49890/50000 progress=99.8% loss=0.252812 lr=1.323232e-09 lr_backbone=1.323232e-09 elapsed=58:03:00 avg_step=4.19s recent_step=4.17s eta=00:07:40 +2026-09-22 16:30:19 | INFO | train rank=0 | step=49900/50000 progress=99.8% loss=0.287743 lr=1.093581e-09 lr_backbone=1.093581e-09 elapsed=58:03:42 avg_step=4.19s recent_step=4.19s eta=00:06:58 +2026-09-22 16:31:00 | INFO | train rank=0 | step=49910/50000 progress=99.8% loss=0.370666 lr=8.858012e-10 lr_backbone=8.858012e-10 elapsed=58:04:23 avg_step=4.19s recent_step=4.11s eta=00:06:16 +2026-09-22 16:31:41 | INFO | train rank=0 | step=49920/50000 progress=99.8% loss=0.263448 lr=6.998928e-10 lr_backbone=6.998928e-10 elapsed=58:05:04 avg_step=4.19s recent_step=4.13s eta=00:05:35 +2026-09-22 16:32:23 | INFO | train rank=0 | step=49930/50000 progress=99.9% loss=0.312322 lr=5.358557e-10 lr_backbone=5.358557e-10 elapsed=58:05:46 avg_step=4.19s recent_step=4.19s eta=00:04:53 +2026-09-22 16:33:05 | INFO | train rank=0 | step=49940/50000 progress=99.9% loss=0.256550 lr=3.936901e-10 lr_backbone=3.936901e-10 elapsed=58:06:28 avg_step=4.19s recent_step=4.16s eta=00:04:11 +2026-09-22 16:33:46 | INFO | train rank=0 | step=49950/50000 progress=99.9% loss=0.353273 lr=2.733960e-10 lr_backbone=2.733960e-10 elapsed=58:07:09 avg_step=4.19s recent_step=4.16s eta=00:03:29 +2026-09-22 16:34:28 | INFO | train rank=0 | step=49960/50000 progress=99.9% loss=0.286493 lr=1.749735e-10 lr_backbone=1.749735e-10 elapsed=58:07:51 avg_step=4.19s recent_step=4.19s eta=00:02:47 +2026-09-22 16:35:09 | INFO | train rank=0 | step=49970/50000 progress=99.9% loss=0.308110 lr=9.842262e-11 lr_backbone=9.842262e-11 elapsed=58:08:33 avg_step=4.19s recent_step=4.13s eta=00:02:05 +2026-09-22 16:35:51 | INFO | train rank=0 | step=49980/50000 progress=100.0% loss=0.386596 lr=4.374339e-11 lr_backbone=4.374339e-11 elapsed=58:09:14 avg_step=4.19s recent_step=4.12s eta=00:01:23 +2026-09-22 16:36:32 | INFO | train rank=0 | step=49990/50000 progress=100.0% loss=0.358188 lr=1.093585e-11 lr_backbone=1.093585e-11 elapsed=58:09:56 avg_step=4.19s recent_step=4.18s eta=00:00:41 +2026-09-22 16:37:14 | INFO | train rank=0 | step=50000/50000 progress=100.0% loss=0.349198 lr=0.000000e+00 lr_backbone=0.000000e+00 elapsed=58:10:38 avg_step=4.19s recent_step=4.20s eta=00:00:00 +2026-09-22 16:38:16 | INFO | train rank=0 | Saved DeepSpeed checkpoint to /data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_50000 +2026-09-22 16:38:16 | INFO | train rank=0 | Finished training final_step=50000/50000 elapsed=58:11:39 avg_step=4.19s +2026-09-22 16:38:16 | INFO | train rank=0 | Training finished +2026-09-22 16:39:17 | INFO | train rank=0 | Saved final DeepSpeed checkpoint to /data/NTU_slab/draven/checkpoints/codebase/codebase_v8_droid/joint_50ksteps_bs128_wan5b_videogen_fast_end_padding_mask_chunk16_downsample2/ds_checkpoint_final