Instructions to use arcadia-impact/scimt-dispatch-models with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use arcadia-impact/scimt-dispatch-models with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="arcadia-impact/scimt-dispatch-models")

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("arcadia-impact/scimt-dispatch-models", device_map="auto")

PEFT
How to use arcadia-impact/scimt-dispatch-models with PEFT:
```
Task type is invalid.
```
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use arcadia-impact/scimt-dispatch-models with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "arcadia-impact/scimt-dispatch-models"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "arcadia-impact/scimt-dispatch-models",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/arcadia-impact/scimt-dispatch-models

SGLang

How to use arcadia-impact/scimt-dispatch-models with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "arcadia-impact/scimt-dispatch-models" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "arcadia-impact/scimt-dispatch-models",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "arcadia-impact/scimt-dispatch-models" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "arcadia-impact/scimt-dispatch-models",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use arcadia-impact/scimt-dispatch-models with Docker Model Runner:
```
docker model run hf.co/arcadia-impact/scimt-dispatch-models
```

scimt-dispatch-models / aft_wave_v2 /control_matched__charter2 /training /checkpoints /debug.log

sidbaines

dispatch-sdf-aft-v1: aft_wave_v2/control_matched__charter2/training

98e3ac1 verified 13 days ago

Raw

History Blame Contribute Delete

267 kB

	[2026-08-18 14:17:02,859] [DEBUG] [axolotl.utils.config.log_gpu_memory_usage:127] [PID:12244] baseline 0.000GB ()
	[2026-08-18 14:17:02,860] [INFO] [axolotl.cli.config.load_cfg:333] [PID:12244] config:
	{
	"activation_offloading": false,
	"adapter": "lora",
	"attn_implementation": "sdpa",
	"attn_needs_dtype_cast": false,
	"attn_supports_packing": false,
	"attn_uses_flash_lib": false,
	"axolotl_config_path": "/workspace/wave/training/axolotl.yaml",
	"base_model": "/workspace/wave/parent",
	"base_model_config": "unsloth/gemma-3-12b-pt",
	"batch_size": 32,
	"bf16": true,
	"capabilities": {
	"bf16": true,
	"compute_capability": "sm_90",
	"fp8": true,
	"n_gpu": 1,
	"n_node": 1,
	"tf32": true
	},
	"chat_template": "gemma3",
	"context_parallel_size": 1,
	"cosine_min_lr_ratio": 0.1,
	"dataloader_num_workers": 1,
	"dataloader_pin_memory": true,
	"dataloader_prefetch_factor": 256,
	"dataset_num_proc": 8,
	"dataset_prepared_path": "/workspace/wave/training/prepared",
	"datasets": [
	{
	"chat_template": "tokenizer_default",
	"field_messages": "messages",
	"message_property_mappings": {
	"content": "content",
	"role": "role"
	},
	"path": "/workspace/wave/data/datasets/aft_charter2.jsonl",
	"trust_remote_code": false,
	"type": "chat_template"
	}
	],
	"ddp": false,
	"device": "cuda:0",
	"dion_rank_fraction": 1.0,
	"dion_rank_multiple_of": 1,
	"eaft_alpha": 1.0,
	"eaft_k": 20,
	"env_capabilities": {
	"torch_version": "2.12.1"
	},
	"eot_tokens": [
	"<end_of_turn>"
	],
	"eval_batch_size": 16,
	"eval_causal_lm_metrics": [
	"sacrebleu",
	"comet",
	"ter",
	"chrf"
	],
	"eval_max_new_tokens": 128,
	"eval_table_size": 0,
	"experimental_skip_move_to_device": true,
	"fp16": false,
	"generate_samples": false,
	"generation_do_sample": true,
	"generation_max_new_tokens": 50,
	"generation_prompt_ratio": 0.5,
	"generation_temperature": 0.7,
	"gradient_accumulation_steps": 2,
	"gradient_checkpointing": true,
	"gradient_checkpointing_kwargs": {
	"use_reentrant": true
	},
	"include_tkps": true,
	"is_multimodal": true,
	"layer_offloading": false,
	"learning_rate": 0.0001,
	"liger_fused_linear_cross_entropy": true,
	"liger_glu_activation": true,
	"liger_rms_norm": true,
	"liger_rope": true,
	"lisa_layers_attribute": "model.layers",
	"load_best_model_at_end": false,
	"load_in_4bit": false,
	"load_in_8bit": false,
	"local_rank": 0,
	"logging_steps": 1,
	"lora_alpha": 64,
	"lora_dropout": 0.05,
	"lora_embedding_kernel": true,
	"lora_mlp_kernel": true,
	"lora_o_kernel": true,
	"lora_qkv_kernel": true,
	"lora_r": 32,
	"lora_target_modules": [
	"q_proj",
	"k_proj",
	"v_proj",
	"o_proj",
	"gate_proj",
	"up_proj",
	"down_proj"
	],
	"loraplus_lr_embedding": 1e-06,
	"lr_scheduler": "cosine",
	"max_grad_norm": 1.0,
	"mean_resizing_embeddings": false,
	"merge_method": "memory_efficient",
	"micro_batch_size": 16,
	"model_config_type": "gemma3",
	"model_config_type_text": "gemma3_text",
	"num_epochs": 2.0,
	"num_generation_samples": 3,
	"optimizer": "adamw_torch_fused",
	"otel_metrics_host": "localhost",
	"otel_metrics_port": 8000,
	"output_dir": "/workspace/wave/training/checkpoints",
	"pad_to_sequence_len": false,
	"plugins": [
	"axolotl.integrations.liger.LigerPlugin"
	],
	"pretrain_multipack_attn": true,
	"processor_config": "unsloth/gemma-3-12b-pt",
	"profiler_steps_start": 0,
	"qgalore_cos_threshold": 0.4,
	"qgalore_gamma_proj": 2,
	"qgalore_proj_bits": 4,
	"qgalore_proj_group_size": 256,
	"qgalore_proj_quant": true,
	"qgalore_proj_type": "std",
	"qgalore_queue_size": 5,
	"qgalore_rank": 256,
	"qgalore_scale": 0.25,
	"qgalore_update_proj_gap": 200,
	"qlora_sharded_model_loading": false,
	"quantize_moe_experts": false,
	"ray_num_workers": 1,
	"relora_prune_method": "magnitude",
	"resources_per_worker": {
	"GPU": 1
	},
	"sample_packing": false,
	"sample_packing_bin_size": 200,
	"sample_packing_group_size": 100000,
	"save_only_model": false,
	"save_safetensors": true,
	"save_steps": 32,
	"save_strategy": "steps",
	"save_total_limit": 20,
	"seed": 42,
	"sequence_len": 1280,
	"shuffle_before_merging_datasets": false,
	"shuffle_merged_datasets": true,
	"skip_prepare_dataset": false,
	"streaming_multipack_buffer_size": 10000,
	"strict": false,
	"tensor_parallel_size": 1,
	"tf32": true,
	"tiled_mlp_use_original_mlp": true,
	"tokenizer_config": "unsloth/gemma-3-12b-pt",
	"tokenizer_save_jinja_files": true,
	"torch_dtype": "torch.bfloat16",
	"train_on_inputs": false,
	"trl": {
	"async_prefetch": false,
	"log_completions": false,
	"mask_truncated_completions": false,
	"ref_model_mixup_alpha": 0.9,
	"ref_model_sync_steps": 64,
	"replay_buffer_size": 0,
	"replay_recompute_logps": true,
	"reroll_max_groups": 1,
	"reroll_start_fraction": 1.0,
	"reward_num_workers": 1,
	"scale_rewards": true,
	"skip_zero_advantage_batches": true,
	"sync_ref_model": false,
	"use_data_producer": false,
	"use_vllm": false,
	"vllm_lora_sync": false,
	"vllm_server_host": "0.0.0.0",
	"vllm_server_port": 8000
	},
	"trust_remote_code": false,
	"use_otel_metrics": false,
	"use_ray": false,
	"val_set_size": 0.0,
	"vllm": {
	"device": "auto",
	"dtype": "auto",
	"gpu_memory_utilization": 0.9,
	"host": "0.0.0.0",
	"port": 8000
	},
	"warmup_ratio": 0.05,
	"weight_decay": 0.01,
	"world_size": 1
	}
	[2026-08-18 14:17:03,137] [DEBUG] [axolotl.loaders.utils.check_model_config:88] [PID:12244] Loaded image size: 896 from model config
	[2026-08-18 14:17:05,129] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:311] [PID:12244] EOS: 1 / <eos>
	[2026-08-18 14:17:05,129] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:312] [PID:12244] BOS: 2 / <bos>
	[2026-08-18 14:17:05,129] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:313] [PID:12244] PAD: 0 / <pad>
	[2026-08-18 14:17:05,130] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:314] [PID:12244] UNK: 3 / <unk>
	[2026-08-18 14:17:05,130] [INFO] [axolotl.utils.data.shared.load_preprocessed_dataset:482] [PID:12244] Unable to find prepared dataset in /workspace/wave/training/prepared/e141cb69e22c0b8ce5ebb8e4998d2ba8
	[2026-08-18 14:17:05,130] [INFO] [axolotl.utils.data.sft._load_raw_datasets:320] [PID:12244] Loading raw datasets...
	[2026-08-18 14:17:05,130] [WARNING] [axolotl.utils.data.sft._load_raw_datasets:322] [PID:12244] Processing datasets during training can lead to VRAM instability. Please pre-process your dataset using `axolotl preprocess path/to/config.yml`.
	[2026-08-18 14:17:05,460] [INFO] [axolotl.utils.data.wrappers.get_dataset_wrapper:87] [PID:12244] Loading dataset: /workspace/wave/data/datasets/aft_charter2.jsonl with base_type: chat_template and prompt_style: None
	[2026-08-18 14:17:05,464] [INFO] [axolotl.prompt_strategies.chat_template.__call__:1209] [PID:12244] Using chat template:
	---
	{{ bos_token }}
	{%- if messages[0]['role'] == 'system' -%}
	{%- if messages[0]['content'] is string -%}
	{%- set first_user_prefix = messages[0]['content'] + '

	' -%}
	{%- else -%}
	{%- set first_user_prefix = messages[0]['content'][0]['text'] + '

	' -%}
	{%- endif -%}
	{%- set loop_messages = messages[1:] -%}
	{%- else -%}
	{%- set first_user_prefix = "" -%}
	{%- set loop_messages = messages -%}
	{%- endif -%}
	{%- for message in loop_messages -%}
	{%- if (message['role'] == 'user') != (loop.index0 % 2 == 0) -%}
	{{ raise_exception("Conversation roles must alternate user/assistant/user/assistant/...") }}
	{%- endif -%}
	{%- if (message['role'] == 'assistant') -%}
	{%- set role = "model" -%}
	{%- else -%}
	{%- set role = message['role'] -%}
	{%- endif -%}
	{{ '<start_of_turn>' + role + '
	' + (first_user_prefix if loop.first else "") }}
	{%- if message['content'] is string -%}
	{{ message['content'] \| trim }}
	{%- elif message['content'] is iterable -%}
	{%- for item in message['content'] -%}
	{%- if item['type'] == 'image' -%}
	{{ '<start_of_image>' }}
	{%- elif item['type'] == 'text' -%}
	{{ item['text'] \| trim }}
	{%- endif -%}
	{%- endfor -%}
	{%- else -%}
	{{ raise_exception("Invalid content type") }}
	{%- endif -%}
	{{ '<end_of_turn>
	' }}
	{%- endfor -%}
	{%- if add_generation_prompt -%}
	{{'<start_of_turn>model
	'}}
	{%- endif -%}

	---
	[2026-08-18 14:17:28,506] [INFO] [axolotl.utils.data.utils._log_dataset_stats:212] [PID:12244] min_input_len: 446
	[2026-08-18 14:17:28,506] [INFO] [axolotl.utils.data.utils._log_dataset_stats:213] [PID:12244] max_input_len: 967
	Saving the dataset (0/8 shards): 0%\| \| 0/8192 [00:00<?, ? examples/s] Saving the dataset (0/8 shards): 12%\|█▎ \| 1024/8192 [00:07<00:50, 142.36 examples/s] Saving the dataset (1/8 shards): 12%\|█▎ \| 1024/8192 [00:07<00:50, 142.36 examples/s] Saving the dataset (2/8 shards): 25%\|██▌ \| 2048/8192 [00:07<00:43, 142.36 examples/s] Saving the dataset (3/8 shards): 38%\|███▊ \| 3072/8192 [00:07<00:35, 142.36 examples/s] Saving the dataset (4/8 shards): 50%\|█████ \| 4096/8192 [00:07<00:28, 142.36 examples/s] Saving the dataset (5/8 shards): 62%\|██████▎ \| 5120/8192 [00:07<00:21, 142.36 examples/s] Saving the dataset (6/8 shards): 75%\|███████▌ \| 6144/8192 [00:07<00:14, 142.36 examples/s] Saving the dataset (7/8 shards): 88%\|████████▊ \| 7168/8192 [00:07<00:07, 142.36 examples/s] Saving the dataset (8/8 shards): 100%\|██████████\| 8192/8192 [00:07<00:00, 142.36 examples/s] Saving the dataset (8/8 shards): 100%\|██████████\| 8192/8192 [00:08<00:00, 983.43 examples/s]
	[2026-08-18 14:17:37,032] [DEBUG] [axolotl.utils.trainer.calculate_total_num_steps:420] [PID:12244] total_num_tokens: 5_605_389
	[2026-08-18 14:17:37,073] [DEBUG] [axolotl.utils.trainer.calculate_total_num_steps:438] [PID:12244] `total_supervised_tokens: 117_062`
	[2026-08-18 14:17:37,073] [DEBUG] [axolotl.utils.trainer.calculate_total_num_steps:521] [PID:12244] total_num_steps: 512
	[2026-08-18 14:17:37,073] [INFO] [axolotl.utils.data.sft._prepare_standard_dataset:121] [PID:12244] Maximum number of steps set at 512
	[2026-08-18 14:17:37,174] [DEBUG] [axolotl.train.setup_model_and_tokenizer:70] [PID:12244] loading tokenizer... unsloth/gemma-3-12b-pt
	[2026-08-18 14:17:39,347] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:311] [PID:12244] EOS: 1 / <eos>
	[2026-08-18 14:17:39,348] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:312] [PID:12244] BOS: 2 / <bos>
	[2026-08-18 14:17:39,348] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:313] [PID:12244] PAD: 0 / <pad>
	[2026-08-18 14:17:39,348] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:314] [PID:12244] UNK: 3 / <unk>
	[2026-08-18 14:17:43,926] [DEBUG] [axolotl.train.setup_model_and_tokenizer:81] [PID:12244] Loading model
	[2026-08-18 14:17:44,037] [DEBUG] [axolotl.monkeypatch.torchao_optim.patch_torchao_optim_state_8bit:75] [PID:12244] Patched OptimState8bit for torch.compile compatibility
	[2026-08-18 14:17:44,037] [DEBUG] [axolotl.monkeypatch.torchao_optim.patch_torchao_optim_state_8bit:122] [PID:12244] Patched OptimState4bit for torch.compile compatibility
	[2026-08-18 14:17:44,037] [DEBUG] [axolotl.monkeypatch.torchao_optim.patch_torchao_optim_state_8bit:154] [PID:12244] Patched OptimStateFp8 for torch.compile compatibility
	[2026-08-18 14:17:44,041] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_evaluation_loop:94] [PID:12244] Patched Trainer.evaluation_loop with nanmean loss calculation
	[2026-08-18 14:17:44,042] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_maybe_log_save_evaluate:148] [PID:12244] Patched Trainer._maybe_log_save_evaluate with nanmean loss calculation
	[2026-08-18 14:17:44,042] [WARNING] [axolotl.loaders.patch_manager._apply_self_attention_lora_patch:662] [PID:12244] Cannot patch self-attention - requires no dropout
	[2026-08-18 14:17:44,974] [INFO] [axolotl.integrations.liger.plugin.pre_model_load:117] [PID:12244] Applying LIGER to gemma3 with kwargs: {'rope': True, 'cross_entropy': None, 'fused_linear_cross_entropy': True, 'rms_norm': True, 'layer_norm': None, 'geglu': True}
	Loading weights: 0%\| \| 0/1066 [00:00<?, ?it/s] Loading weights: 68%\|██████▊ \| 722/1066 [00:00<00:00, 7217.68it/s] Loading weights: 100%\|██████████\| 1066/1066 [00:00<00:00, 7268.18it/s]
	[2026-08-18 14:17:48,445] [INFO] [axolotl.loaders.model._configure_embedding_dtypes:433] [PID:12244] Converting modules to torch.bfloat16
	[2026-08-18 14:17:49,902] [DEBUG] [axolotl.loaders.model.log_gpu_memory_usage:127] [PID:12244] Memory usage after model load 0.000GB ()
	trainable params: 136,912,896 \|\| all params: 12,324,237,936 \|\| trainable%: 1.1109
	[2026-08-18 14:17:51,447] [DEBUG] [axolotl.loaders.model.log_gpu_memory_usage:127] [PID:12244] after adapters 0.000GB ()
	[2026-08-18 14:17:52,538] [INFO] [axolotl.monkeypatch.lora_kernels.apply_lora_kernel_patches:418] [PID:12244] LoRA kernels: dropout=0.05 enabled
	[2026-08-18 14:17:56,144] [INFO] [axolotl.train.save_initial_configs:450] [PID:12244] Pre-saving adapter config to /workspace/wave/training/checkpoints...
	[2026-08-18 14:17:56,145] [INFO] [axolotl.train.save_initial_configs:454] [PID:12244] Pre-saving tokenizer to /workspace/wave/training/checkpoints...
	[2026-08-18 14:17:56,460] [INFO] [axolotl.train.save_initial_configs:459] [PID:12244] Pre-saving model config to /workspace/wave/training/checkpoints...
	[2026-08-18 14:17:56,463] [INFO] [axolotl.train.save_initial_configs:463] [PID:12244] Pre-saving processor to /workspace/wave/training/checkpoints...
	[2026-08-18 14:17:56,740] [INFO] [axolotl.train.execute_training:226] [PID:12244] Starting trainer...
	0%\| \| 0/512 [00:00<?, ?it/s] 0%\| \| 1/512 [00:07<1:07:26, 7.92s/it] {'loss': '0.1503', 'grad_norm': '8.089', 'learning_rate': '0', 'ppl': '1.162', 'memory/max_active (GiB)': '32.79', 'memory/max_allocated (GiB)': '32.79', 'memory/device_reserved (GiB)': '33.9', 'tokens/train_per_sec_per_gpu': '28.32', 'tokens/total': 30176, 'tokens/trainable': 425, 'epoch': '0.003906'}
	0%\| \| 1/512 [00:07<1:07:26, 7.92s/it] 0%\| \| 2/512 [00:14<1:00:22, 7.10s/it] {'loss': '0.1521', 'grad_norm': '0.8954', 'learning_rate': '4e-06', 'ppl': '1.164', 'memory/max_active (GiB)': '33.77', 'memory/max_allocated (GiB)': '33.77', 'memory/device_reserved (GiB)': '34.68', 'tokens/train_per_sec_per_gpu': '33.91', 'tokens/total': 60272, 'tokens/trainable': 872, 'epoch': '0.007812'}
	0%\| \| 2/512 [00:14<1:00:22, 7.10s/it] 1%\| \| 3/512 [00:21<58:23, 6.88s/it] {'loss': '0.1453', 'grad_norm': '1.401', 'learning_rate': '8e-06', 'ppl': '1.156', 'memory/max_active (GiB)': '33.83', 'memory/max_allocated (GiB)': '33.83', 'memory/device_reserved (GiB)': '34.72', 'tokens/train_per_sec_per_gpu': '35.5', 'tokens/total': 90512, 'tokens/trainable': 1352, 'epoch': '0.01172'}
	1%\| \| 3/512 [00:21<58:23, 6.88s/it] 1%\| \| 4/512 [00:27<57:22, 6.78s/it] {'loss': '0.135', 'grad_norm': '0.9737', 'learning_rate': '1.2e-05', 'ppl': '1.145', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '34.73', 'tokens/train_per_sec_per_gpu': '31.75', 'tokens/total': 120944, 'tokens/trainable': 1777, 'epoch': '0.01562'}
	1%\| \| 4/512 [00:27<57:22, 6.78s/it] 1%\| \| 5/512 [00:34<56:41, 6.71s/it] {'loss': '0.1381', 'grad_norm': '1.554', 'learning_rate': '1.6e-05', 'ppl': '1.148', 'memory/max_active (GiB)': '33.91', 'memory/max_allocated (GiB)': '33.91', 'memory/device_reserved (GiB)': '35.12', 'tokens/train_per_sec_per_gpu': '37.35', 'tokens/total': 151440, 'tokens/trainable': 2260, 'epoch': '0.01953'}
	1%\| \| 5/512 [00:34<56:41, 6.71s/it] 1%\| \| 6/512 [00:40<56:25, 6.69s/it] {'loss': '0.1262', 'grad_norm': '6.434', 'learning_rate': '2e-05', 'ppl': '1.135', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '36.12', 'tokens/train_per_sec_per_gpu': '33.21', 'tokens/total': 181984, 'tokens/trainable': 2704, 'epoch': '0.02344'}
	1%\| \| 6/512 [00:40<56:25, 6.69s/it] 1%\|▏ \| 7/512 [00:47<56:04, 6.66s/it] {'loss': '0.112', 'grad_norm': '1.401', 'learning_rate': '2.4e-05', 'ppl': '1.119', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '36.12', 'tokens/train_per_sec_per_gpu': '34.37', 'tokens/total': 212336, 'tokens/trainable': 3143, 'epoch': '0.02734'}
	1%\|▏ \| 7/512 [00:47<56:04, 6.66s/it] 2%\|▏ \| 8/512 [00:54<55:46, 6.64s/it] {'loss': '0.0922', 'grad_norm': '2.039', 'learning_rate': '2.8e-05', 'ppl': '1.097', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '36.12', 'tokens/train_per_sec_per_gpu': '34.78', 'tokens/total': 242592, 'tokens/trainable': 3586, 'epoch': '0.03125'}
	2%\|▏ \| 8/512 [00:54<55:46, 6.64s/it] 2%\|▏ \| 9/512 [01:00<55:26, 6.61s/it] {'loss': '0.04823', 'grad_norm': '4.828', 'learning_rate': '3.2e-05', 'ppl': '1.049', 'memory/max_active (GiB)': '33.77', 'memory/max_allocated (GiB)': '33.77', 'memory/device_reserved (GiB)': '36.12', 'tokens/train_per_sec_per_gpu': '29.59', 'tokens/total': 272784, 'tokens/trainable': 4007, 'epoch': '0.03516'}
	2%\|▏ \| 9/512 [01:00<55:26, 6.61s/it] 2%\|▏ \| 10/512 [01:07<55:15, 6.60s/it] {'loss': '0.04049', 'grad_norm': '1.369', 'learning_rate': '3.6e-05', 'ppl': '1.041', 'memory/max_active (GiB)': '33.89', 'memory/max_allocated (GiB)': '33.89', 'memory/device_reserved (GiB)': '37.81', 'tokens/train_per_sec_per_gpu': '39.66', 'tokens/total': 303184, 'tokens/trainable': 4481, 'epoch': '0.03906'}
	2%\|▏ \| 10/512 [01:07<55:15, 6.60s/it] 2%\|▏ \| 11/512 [01:13<55:05, 6.60s/it] {'loss': '0.0634', 'grad_norm': '2.285', 'learning_rate': '4e-05', 'ppl': '1.065', 'memory/max_active (GiB)': '33.73', 'memory/max_allocated (GiB)': '33.73', 'memory/device_reserved (GiB)': '37.81', 'tokens/train_per_sec_per_gpu': '35.09', 'tokens/total': 333296, 'tokens/trainable': 4958, 'epoch': '0.04297'}
	2%\|▏ \| 11/512 [01:13<55:05, 6.60s/it] 2%\|▏ \| 12/512 [01:20<55:00, 6.60s/it] {'loss': '0.071', 'grad_norm': '5.053', 'learning_rate': '4.4e-05', 'ppl': '1.074', 'memory/max_active (GiB)': '33.96', 'memory/max_allocated (GiB)': '33.96', 'memory/device_reserved (GiB)': '37.81', 'tokens/train_per_sec_per_gpu': '34.81', 'tokens/total': 363840, 'tokens/trainable': 5440, 'epoch': '0.04688'}
	2%\|▏ \| 12/512 [01:20<55:00, 6.60s/it] 3%\|▎ \| 13/512 [01:27<54:56, 6.61s/it] {'loss': '0.06127', 'grad_norm': '3.501', 'learning_rate': '4.8e-05', 'ppl': '1.063', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '37.81', 'tokens/train_per_sec_per_gpu': '35.82', 'tokens/total': 394112, 'tokens/trainable': 5871, 'epoch': '0.05078'}
	3%\|▎ \| 13/512 [01:27<54:56, 6.61s/it] 3%\|▎ \| 14/512 [01:33<54:58, 6.62s/it] {'loss': '0.04367', 'grad_norm': '2.639', 'learning_rate': '5.2e-05', 'ppl': '1.045', 'memory/max_active (GiB)': '33.91', 'memory/max_allocated (GiB)': '33.91', 'memory/device_reserved (GiB)': '37.81', 'tokens/train_per_sec_per_gpu': '35.29', 'tokens/total': 424656, 'tokens/trainable': 6360, 'epoch': '0.05469'}
	3%\|▎ \| 14/512 [01:33<54:58, 6.62s/it] 3%\|▎ \| 15/512 [01:40<54:52, 6.62s/it] {'loss': '0.06026', 'grad_norm': '2.628', 'learning_rate': '5.6e-05', 'ppl': '1.062', 'memory/max_active (GiB)': '33.96', 'memory/max_allocated (GiB)': '33.96', 'memory/device_reserved (GiB)': '37.81', 'tokens/train_per_sec_per_gpu': '36.67', 'tokens/total': 455152, 'tokens/trainable': 6841, 'epoch': '0.05859'}
	3%\|▎ \| 15/512 [01:40<54:52, 6.62s/it] 3%\|▎ \| 16/512 [01:46<54:31, 6.60s/it] {'loss': '0.09122', 'grad_norm': '5.878', 'learning_rate': '6e-05', 'ppl': '1.096', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '37.81', 'tokens/train_per_sec_per_gpu': '31.7', 'tokens/total': 485328, 'tokens/trainable': 7254, 'epoch': '0.0625'}
	3%\|▎ \| 16/512 [01:46<54:31, 6.60s/it] 3%\|▎ \| 17/512 [01:53<54:30, 6.61s/it] {'loss': '0.01772', 'grad_norm': '1.972', 'learning_rate': '6.4e-05', 'ppl': '1.018', 'memory/max_active (GiB)': '33.92', 'memory/max_allocated (GiB)': '33.92', 'memory/device_reserved (GiB)': '37.81', 'tokens/train_per_sec_per_gpu': '36.52', 'tokens/total': 515744, 'tokens/trainable': 7723, 'epoch': '0.06641'}
	3%\|▎ \| 17/512 [01:53<54:30, 6.61s/it] 4%\|▎ \| 18/512 [02:00<54:31, 6.62s/it] {'loss': '0.0474', 'grad_norm': '1.783', 'learning_rate': '6.8e-05', 'ppl': '1.049', 'memory/max_active (GiB)': '33.77', 'memory/max_allocated (GiB)': '33.77', 'memory/device_reserved (GiB)': '37.81', 'tokens/train_per_sec_per_gpu': '34.38', 'tokens/total': 546144, 'tokens/trainable': 8208, 'epoch': '0.07031'}
	4%\|▎ \| 18/512 [02:00<54:31, 6.62s/it] 4%\|▎ \| 19/512 [02:06<54:25, 6.62s/it] {'loss': '0.02503', 'grad_norm': '1.916', 'learning_rate': '7.2e-05', 'ppl': '1.025', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '37.81', 'tokens/train_per_sec_per_gpu': '35.49', 'tokens/total': 576560, 'tokens/trainable': 8699, 'epoch': '0.07422'}
	4%\|▎ \| 19/512 [02:06<54:25, 6.62s/it] 4%\|▍ \| 20/512 [02:13<54:20, 6.63s/it] {'loss': '0.05566', 'grad_norm': '2.979', 'learning_rate': '7.6e-05', 'ppl': '1.057', 'memory/max_active (GiB)': '33.95', 'memory/max_allocated (GiB)': '33.95', 'memory/device_reserved (GiB)': '37.81', 'tokens/train_per_sec_per_gpu': '40.56', 'tokens/total': 607056, 'tokens/trainable': 9214, 'epoch': '0.07812'}
	4%\|▍ \| 20/512 [02:13<54:20, 6.63s/it] 4%\|▍ \| 21/512 [02:20<54:14, 6.63s/it] {'loss': '0.02437', 'grad_norm': '1.272', 'learning_rate': '8e-05', 'ppl': '1.025', 'memory/max_active (GiB)': '33.77', 'memory/max_allocated (GiB)': '33.77', 'memory/device_reserved (GiB)': '37.81', 'tokens/train_per_sec_per_gpu': '32.12', 'tokens/total': 637488, 'tokens/trainable': 9639, 'epoch': '0.08203'}
	4%\|▍ \| 21/512 [02:20<54:14, 6.63s/it] 4%\|▍ \| 22/512 [02:26<54:02, 6.62s/it] {'loss': '0.0517', 'grad_norm': '2.242', 'learning_rate': '8.4e-05', 'ppl': '1.053', 'memory/max_active (GiB)': '33.75', 'memory/max_allocated (GiB)': '33.75', 'memory/device_reserved (GiB)': '37.81', 'tokens/train_per_sec_per_gpu': '36.71', 'tokens/total': 667632, 'tokens/trainable': 10111, 'epoch': '0.08594'}
	4%\|▍ \| 22/512 [02:26<54:02, 6.62s/it] 4%\|▍ \| 23/512 [02:33<53:56, 6.62s/it] {'loss': '0.04347', 'grad_norm': '2.216', 'learning_rate': '8.8e-05', 'ppl': '1.044', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '37.81', 'tokens/train_per_sec_per_gpu': '33.4', 'tokens/total': 697872, 'tokens/trainable': 10560, 'epoch': '0.08984'}
	4%\|▍ \| 23/512 [02:33<53:56, 6.62s/it] 5%\|▍ \| 24/512 [02:39<53:50, 6.62s/it] {'loss': '0.01621', 'grad_norm': '1.028', 'learning_rate': '9.2e-05', 'ppl': '1.016', 'memory/max_active (GiB)': '33.92', 'memory/max_allocated (GiB)': '33.92', 'memory/device_reserved (GiB)': '37.81', 'tokens/train_per_sec_per_gpu': '32', 'tokens/total': 728432, 'tokens/trainable': 11018, 'epoch': '0.09375'}
	5%\|▍ \| 24/512 [02:39<53:50, 6.62s/it] 5%\|▍ \| 25/512 [02:46<52:27, 6.46s/it] {'loss': '0.02351', 'grad_norm': '1.041', 'learning_rate': '9.6e-05', 'ppl': '1.024', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '37.81', 'tokens/train_per_sec_per_gpu': '40.53', 'tokens/total': 756656, 'tokens/trainable': 11442, 'epoch': '0.09766'}
	5%\|▍ \| 25/512 [02:46<52:27, 6.46s/it] 5%\|▌ \| 26/512 [02:52<52:37, 6.50s/it] {'loss': '0.05569', 'grad_norm': '1.666', 'learning_rate': '0.0001', 'ppl': '1.057', 'memory/max_active (GiB)': '33.92', 'memory/max_allocated (GiB)': '33.92', 'memory/device_reserved (GiB)': '37.82', 'tokens/train_per_sec_per_gpu': '34.83', 'tokens/total': 787120, 'tokens/trainable': 11899, 'epoch': '0.1016'}
	5%\|▌ \| 26/512 [02:52<52:37, 6.50s/it] 5%\|▌ \| 27/512 [02:59<52:51, 6.54s/it] {'loss': '0.009145', 'grad_norm': '0.7106', 'learning_rate': '0.0001', 'ppl': '1.009', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '37.82', 'tokens/train_per_sec_per_gpu': '32.1', 'tokens/total': 817504, 'tokens/trainable': 12341, 'epoch': '0.1055'}
	5%\|▌ \| 27/512 [02:59<52:51, 6.54s/it] 5%\|▌ \| 28/512 [03:05<53:00, 6.57s/it] {'loss': '0.02025', 'grad_norm': '0.8463', 'learning_rate': '0.0001', 'ppl': '1.02', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '37.82', 'tokens/train_per_sec_per_gpu': '38.39', 'tokens/total': 848128, 'tokens/trainable': 12825, 'epoch': '0.1094'}
	5%\|▌ \| 28/512 [03:05<53:00, 6.57s/it] 6%\|▌ \| 29/512 [03:12<52:59, 6.58s/it] {'loss': '0.02345', 'grad_norm': '1.074', 'learning_rate': '9.999e-05', 'ppl': '1.024', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '37.82', 'tokens/train_per_sec_per_gpu': '36.01', 'tokens/total': 878448, 'tokens/trainable': 13295, 'epoch': '0.1133'}
	6%\|▌ \| 29/512 [03:12<52:59, 6.58s/it] 6%\|▌ \| 30/512 [03:19<53:02, 6.60s/it] {'loss': '0.03459', 'grad_norm': '1.423', 'learning_rate': '9.999e-05', 'ppl': '1.035', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '37.82', 'tokens/train_per_sec_per_gpu': '35.8', 'tokens/total': 909008, 'tokens/trainable': 13784, 'epoch': '0.1172'}
	6%\|▌ \| 30/512 [03:19<53:02, 6.60s/it] 6%\|▌ \| 31/512 [03:25<52:59, 6.61s/it] {'loss': '0.0248', 'grad_norm': '0.988', 'learning_rate': '9.998e-05', 'ppl': '1.025', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '37.82', 'tokens/train_per_sec_per_gpu': '34.58', 'tokens/total': 939312, 'tokens/trainable': 14250, 'epoch': '0.1211'}
	6%\|▌ \| 31/512 [03:25<52:59, 6.61s/it] 6%\|▋ \| 32/512 [03:32<52:38, 6.58s/it] {'loss': '0.02302', 'grad_norm': '1.116', 'learning_rate': '9.997e-05', 'ppl': '1.023', 'memory/max_active (GiB)': '33.81', 'memory/max_allocated (GiB)': '33.81', 'memory/device_reserved (GiB)': '37.82', 'tokens/train_per_sec_per_gpu': '35', 'tokens/total': 969264, 'tokens/trainable': 14680, 'epoch': '0.125'}
	6%\|▋ \| 32/512 [03:32<52:38, 6.58s/it][2026-08-18 14:21:29,523] [INFO] [axolotl.core.trainers.base._save:828] [PID:12244] Saving model checkpoint to /workspace/wave/training/checkpoints/checkpoint-32
	6%\|▋ \| 33/512 [03:40<56:36, 7.09s/it] {'loss': '0.0631', 'grad_norm': '1.325', 'learning_rate': '9.995e-05', 'ppl': '1.065', 'memory/max_active (GiB)': '33.95', 'memory/max_allocated (GiB)': '33.95', 'memory/device_reserved (GiB)': '37.82', 'tokens/train_per_sec_per_gpu': '33.83', 'tokens/total': 999792, 'tokens/trainable': 15112, 'epoch': '0.1289'}
	6%\|▋ \| 33/512 [03:40<56:36, 7.09s/it] 7%\|▋ \| 34/512 [03:47<55:08, 6.92s/it] {'loss': '0.04253', 'grad_norm': '1.003', 'learning_rate': '9.994e-05', 'ppl': '1.043', 'memory/max_active (GiB)': '33.77', 'memory/max_allocated (GiB)': '33.77', 'memory/device_reserved (GiB)': '35.48', 'tokens/train_per_sec_per_gpu': '34.46', 'tokens/total': 1029840, 'tokens/trainable': 15573, 'epoch': '0.1328'}
	7%\|▋ \| 34/512 [03:47<55:08, 6.92s/it] 7%\|▋ \| 35/512 [03:53<54:24, 6.84s/it] {'loss': '0.03612', 'grad_norm': '1.139', 'learning_rate': '9.992e-05', 'ppl': '1.037', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '35.48', 'tokens/train_per_sec_per_gpu': '34.24', 'tokens/total': 1060416, 'tokens/trainable': 16062, 'epoch': '0.1367'}
	7%\|▋ \| 35/512 [03:53<54:24, 6.84s/it] 7%\|▋ \| 36/512 [04:00<53:33, 6.75s/it] {'loss': '0.01517', 'grad_norm': '0.4866', 'learning_rate': '9.991e-05', 'ppl': '1.015', 'memory/max_active (GiB)': '33.72', 'memory/max_allocated (GiB)': '33.72', 'memory/device_reserved (GiB)': '35.48', 'tokens/train_per_sec_per_gpu': '31.53', 'tokens/total': 1090464, 'tokens/trainable': 16500, 'epoch': '0.1406'}
	7%\|▋ \| 36/512 [04:00<53:33, 6.75s/it] 7%\|▋ \| 37/512 [04:06<53:00, 6.69s/it] {'loss': '0.04621', 'grad_norm': '0.7631', 'learning_rate': '9.989e-05', 'ppl': '1.047', 'memory/max_active (GiB)': '33.93', 'memory/max_allocated (GiB)': '33.93', 'memory/device_reserved (GiB)': '35.86', 'tokens/train_per_sec_per_gpu': '34.9', 'tokens/total': 1120944, 'tokens/trainable': 16936, 'epoch': '0.1445'}
	7%\|▋ \| 37/512 [04:06<53:00, 6.69s/it] 7%\|▋ \| 38/512 [04:13<52:37, 6.66s/it] {'loss': '0.046', 'grad_norm': '1.116', 'learning_rate': '9.987e-05', 'ppl': '1.047', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '35.86', 'tokens/train_per_sec_per_gpu': '29.32', 'tokens/total': 1151360, 'tokens/trainable': 17357, 'epoch': '0.1484'}
	7%\|▋ \| 38/512 [04:13<52:37, 6.66s/it] 8%\|▊ \| 39/512 [04:20<52:30, 6.66s/it] {'loss': '0.052', 'grad_norm': '4.36', 'learning_rate': '9.984e-05', 'ppl': '1.053', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '35.86', 'tokens/train_per_sec_per_gpu': '36.82', 'tokens/total': 1181728, 'tokens/trainable': 17831, 'epoch': '0.1523'}
	8%\|▊ \| 39/512 [04:20<52:30, 6.66s/it] 8%\|▊ \| 40/512 [04:26<52:16, 6.65s/it] {'loss': '0.04869', 'grad_norm': '0.8433', 'learning_rate': '9.982e-05', 'ppl': '1.05', 'memory/max_active (GiB)': '33.83', 'memory/max_allocated (GiB)': '33.83', 'memory/device_reserved (GiB)': '35.86', 'tokens/train_per_sec_per_gpu': '36.15', 'tokens/total': 1212064, 'tokens/trainable': 18315, 'epoch': '0.1562'}
	8%\|▊ \| 40/512 [04:26<52:16, 6.65s/it] 8%\|▊ \| 41/512 [04:33<52:02, 6.63s/it] {'loss': '0.02915', 'grad_norm': '0.9528', 'learning_rate': '9.979e-05', 'ppl': '1.03', 'memory/max_active (GiB)': '33.97', 'memory/max_allocated (GiB)': '33.97', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '32.49', 'tokens/total': 1242592, 'tokens/trainable': 18772, 'epoch': '0.1602'}
	8%\|▊ \| 41/512 [04:33<52:02, 6.63s/it] 8%\|▊ \| 42/512 [04:39<51:50, 6.62s/it] {'loss': '0.01587', 'grad_norm': '0.5701', 'learning_rate': '9.976e-05', 'ppl': '1.016', 'memory/max_active (GiB)': '33.81', 'memory/max_allocated (GiB)': '33.81', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '32.32', 'tokens/total': 1272800, 'tokens/trainable': 19197, 'epoch': '0.1641'}
	8%\|▊ \| 42/512 [04:39<51:50, 6.62s/it] 8%\|▊ \| 43/512 [04:46<51:54, 6.64s/it] {'loss': '0.01053', 'grad_norm': '0.373', 'learning_rate': '9.973e-05', 'ppl': '1.011', 'memory/max_active (GiB)': '34.04', 'memory/max_allocated (GiB)': '34.04', 'memory/device_reserved (GiB)': '36.04', 'tokens/train_per_sec_per_gpu': '31.68', 'tokens/total': 1303600, 'tokens/trainable': 19647, 'epoch': '0.168'}
	8%\|▊ \| 43/512 [04:46<51:54, 6.64s/it] 9%\|▊ \| 44/512 [04:53<51:41, 6.63s/it] {'loss': '0.01996', 'grad_norm': '0.5503', 'learning_rate': '9.97e-05', 'ppl': '1.02', 'memory/max_active (GiB)': '33.81', 'memory/max_allocated (GiB)': '33.81', 'memory/device_reserved (GiB)': '36.04', 'tokens/train_per_sec_per_gpu': '33.83', 'tokens/total': 1333936, 'tokens/trainable': 20097, 'epoch': '0.1719'}
	9%\|▊ \| 44/512 [04:53<51:41, 6.63s/it] 9%\|▉ \| 45/512 [04:59<51:32, 6.62s/it] {'loss': '0.06465', 'grad_norm': '3.762', 'learning_rate': '9.966e-05', 'ppl': '1.067', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '36.04', 'tokens/train_per_sec_per_gpu': '31.92', 'tokens/total': 1364288, 'tokens/trainable': 20529, 'epoch': '0.1758'}
	9%\|▉ \| 45/512 [04:59<51:32, 6.62s/it] 9%\|▉ \| 46/512 [05:06<51:23, 6.62s/it] {'loss': '0.05519', 'grad_norm': '1.149', 'learning_rate': '9.963e-05', 'ppl': '1.057', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '36.04', 'tokens/train_per_sec_per_gpu': '34.68', 'tokens/total': 1394432, 'tokens/trainable': 20973, 'epoch': '0.1797'}
	9%\|▉ \| 46/512 [05:06<51:23, 6.62s/it] 9%\|▉ \| 47/512 [05:13<51:25, 6.63s/it] {'loss': '0.02402', 'grad_norm': '0.5584', 'learning_rate': '9.959e-05', 'ppl': '1.024', 'memory/max_active (GiB)': '33.98', 'memory/max_allocated (GiB)': '33.98', 'memory/device_reserved (GiB)': '36.16', 'tokens/train_per_sec_per_gpu': '35.36', 'tokens/total': 1425088, 'tokens/trainable': 21437, 'epoch': '0.1836'}
	9%\|▉ \| 47/512 [05:13<51:25, 6.63s/it] 9%\|▉ \| 48/512 [05:19<51:11, 6.62s/it] {'loss': '0.02145', 'grad_norm': '0.7726', 'learning_rate': '9.955e-05', 'ppl': '1.022', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '36.16', 'tokens/train_per_sec_per_gpu': '35.57', 'tokens/total': 1455456, 'tokens/trainable': 21860, 'epoch': '0.1875'}
	9%\|▉ \| 48/512 [05:19<51:11, 6.62s/it] 10%\|▉ \| 49/512 [05:26<50:45, 6.58s/it] {'loss': '0.03308', 'grad_norm': '0.9507', 'learning_rate': '9.951e-05', 'ppl': '1.034', 'memory/max_active (GiB)': '33.7', 'memory/max_allocated (GiB)': '33.7', 'memory/device_reserved (GiB)': '36.16', 'tokens/train_per_sec_per_gpu': '30.1', 'tokens/total': 1485376, 'tokens/trainable': 22273, 'epoch': '0.1914'}
	10%\|▉ \| 49/512 [05:26<50:45, 6.58s/it] 10%\|▉ \| 50/512 [05:32<50:41, 6.58s/it] {'loss': '0.01948', 'grad_norm': '0.5201', 'learning_rate': '9.946e-05', 'ppl': '1.02', 'memory/max_active (GiB)': '33.92', 'memory/max_allocated (GiB)': '33.92', 'memory/device_reserved (GiB)': '36.16', 'tokens/train_per_sec_per_gpu': '36.83', 'tokens/total': 1515744, 'tokens/trainable': 22720, 'epoch': '0.1953'}
	10%\|▉ \| 50/512 [05:32<50:41, 6.58s/it] 10%\|▉ \| 51/512 [05:39<50:29, 6.57s/it] {'loss': '0.02124', 'grad_norm': '0.5437', 'learning_rate': '9.942e-05', 'ppl': '1.021', 'memory/max_active (GiB)': '33.89', 'memory/max_allocated (GiB)': '33.89', 'memory/device_reserved (GiB)': '36.16', 'tokens/train_per_sec_per_gpu': '32.55', 'tokens/total': 1545984, 'tokens/trainable': 23168, 'epoch': '0.1992'}
	10%\|▉ \| 51/512 [05:39<50:29, 6.57s/it] 10%\|█ \| 52/512 [05:45<50:30, 6.59s/it] {'loss': '0.02224', 'grad_norm': '0.4705', 'learning_rate': '9.937e-05', 'ppl': '1.022', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '36.16', 'tokens/train_per_sec_per_gpu': '37.44', 'tokens/total': 1576288, 'tokens/trainable': 23663, 'epoch': '0.2031'}
	10%\|█ \| 52/512 [05:45<50:30, 6.59s/it] 10%\|█ \| 53/512 [05:52<50:31, 6.60s/it] {'loss': '0.01991', 'grad_norm': '12.06', 'learning_rate': '9.932e-05', 'ppl': '1.02', 'memory/max_active (GiB)': '33.83', 'memory/max_allocated (GiB)': '33.83', 'memory/device_reserved (GiB)': '36.16', 'tokens/train_per_sec_per_gpu': '34.49', 'tokens/total': 1606528, 'tokens/trainable': 24128, 'epoch': '0.207'}
	10%\|█ \| 53/512 [05:52<50:31, 6.60s/it] 11%\|█ \| 54/512 [05:59<50:17, 6.59s/it] {'loss': '0.03577', 'grad_norm': '0.7315', 'learning_rate': '9.927e-05', 'ppl': '1.036', 'memory/max_active (GiB)': '33.66', 'memory/max_allocated (GiB)': '33.66', 'memory/device_reserved (GiB)': '36.16', 'tokens/train_per_sec_per_gpu': '38.33', 'tokens/total': 1636400, 'tokens/trainable': 24601, 'epoch': '0.2109'}
	11%\|█ \| 54/512 [05:59<50:17, 6.59s/it] 11%\|█ \| 55/512 [06:05<50:19, 6.61s/it] {'loss': '0.04387', 'grad_norm': '0.8308', 'learning_rate': '9.921e-05', 'ppl': '1.045', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '36.16', 'tokens/train_per_sec_per_gpu': '34.44', 'tokens/total': 1666896, 'tokens/trainable': 25099, 'epoch': '0.2148'}
	11%\|█ \| 55/512 [06:05<50:19, 6.61s/it] 11%\|█ \| 56/512 [06:12<50:13, 6.61s/it] {'loss': '0.009335', 'grad_norm': '0.361', 'learning_rate': '9.916e-05', 'ppl': '1.009', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '36.16', 'tokens/train_per_sec_per_gpu': '37.08', 'tokens/total': 1697232, 'tokens/trainable': 25573, 'epoch': '0.2188'}
	11%\|█ \| 56/512 [06:12<50:13, 6.61s/it] 11%\|█ \| 57/512 [06:18<50:04, 6.60s/it] {'loss': '0.01627', 'grad_norm': '0.5462', 'learning_rate': '9.91e-05', 'ppl': '1.016', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '36.16', 'tokens/train_per_sec_per_gpu': '33.69', 'tokens/total': 1727440, 'tokens/trainable': 25997, 'epoch': '0.2227'}
	11%\|█ \| 57/512 [06:18<50:04, 6.60s/it] 11%\|█▏ \| 58/512 [06:25<49:54, 6.59s/it] {'loss': '0.01592', 'grad_norm': '3.278', 'learning_rate': '9.904e-05', 'ppl': '1.016', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '36.16', 'tokens/train_per_sec_per_gpu': '34.37', 'tokens/total': 1757952, 'tokens/trainable': 26441, 'epoch': '0.2266'}
	11%\|█▏ \| 58/512 [06:25<49:54, 6.59s/it] 12%\|█▏ \| 59/512 [06:32<49:52, 6.61s/it] {'loss': '0.01648', 'grad_norm': '1.643', 'learning_rate': '9.898e-05', 'ppl': '1.017', 'memory/max_active (GiB)': '33.92', 'memory/max_allocated (GiB)': '33.92', 'memory/device_reserved (GiB)': '36.16', 'tokens/train_per_sec_per_gpu': '34.54', 'tokens/total': 1788576, 'tokens/trainable': 26915, 'epoch': '0.2305'}
	12%\|█▏ \| 59/512 [06:32<49:52, 6.61s/it] 12%\|█▏ \| 60/512 [06:38<49:46, 6.61s/it] {'loss': '0.01031', 'grad_norm': '0.5992', 'learning_rate': '9.892e-05', 'ppl': '1.01', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '36.16', 'tokens/train_per_sec_per_gpu': '34.2', 'tokens/total': 1818816, 'tokens/trainable': 27355, 'epoch': '0.2344'}
	12%\|█▏ \| 60/512 [06:38<49:46, 6.61s/it] 12%\|█▏ \| 61/512 [06:45<49:42, 6.61s/it] {'loss': '0.002149', 'grad_norm': '0.1208', 'learning_rate': '9.886e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.92', 'memory/max_allocated (GiB)': '33.92', 'memory/device_reserved (GiB)': '36.16', 'tokens/train_per_sec_per_gpu': '39.71', 'tokens/total': 1849424, 'tokens/trainable': 27862, 'epoch': '0.2383'}
	12%\|█▏ \| 61/512 [06:45<49:42, 6.61s/it] 12%\|█▏ \| 62/512 [06:52<49:48, 6.64s/it] {'loss': '0.04498', 'grad_norm': '0.8363', 'learning_rate': '9.879e-05', 'ppl': '1.046', 'memory/max_active (GiB)': '34.02', 'memory/max_allocated (GiB)': '34.02', 'memory/device_reserved (GiB)': '36.16', 'tokens/train_per_sec_per_gpu': '32.8', 'tokens/total': 1880032, 'tokens/trainable': 28313, 'epoch': '0.2422'}
	12%\|█▏ \| 62/512 [06:52<49:48, 6.64s/it] 12%\|█▏ \| 63/512 [06:58<49:37, 6.63s/it] {'loss': '0.04683', 'grad_norm': '1.207', 'learning_rate': '9.872e-05', 'ppl': '1.048', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '36.16', 'tokens/train_per_sec_per_gpu': '32.7', 'tokens/total': 1910512, 'tokens/trainable': 28774, 'epoch': '0.2461'}
	12%\|█▏ \| 63/512 [06:58<49:37, 6.63s/it] 12%\|█▎ \| 64/512 [07:05<49:19, 6.61s/it] {'loss': '0.02833', 'grad_norm': '0.8954', 'learning_rate': '9.865e-05', 'ppl': '1.029', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '36.16', 'tokens/train_per_sec_per_gpu': '34.07', 'tokens/total': 1940608, 'tokens/trainable': 29222, 'epoch': '0.25'}
	12%\|█▎ \| 64/512 [07:05<49:19, 6.61s/it][2026-08-18 14:25:02,486] [INFO] [axolotl.core.trainers.base._save:828] [PID:12244] Saving model checkpoint to /workspace/wave/training/checkpoints/checkpoint-64
	13%\|█▎ \| 65/512 [07:13<52:21, 7.03s/it] {'loss': '0.01838', 'grad_norm': '0.7108', 'learning_rate': '9.858e-05', 'ppl': '1.019', 'memory/max_active (GiB)': '33.91', 'memory/max_allocated (GiB)': '33.91', 'memory/device_reserved (GiB)': '36.16', 'tokens/train_per_sec_per_gpu': '32', 'tokens/total': 1971104, 'tokens/trainable': 29687, 'epoch': '0.2539'}
	13%\|█▎ \| 65/512 [07:13<52:21, 7.03s/it] 13%\|█▎ \| 66/512 [07:19<51:13, 6.89s/it] {'loss': '0.02454', 'grad_norm': '0.8418', 'learning_rate': '9.851e-05', 'ppl': '1.025', 'memory/max_active (GiB)': '33.83', 'memory/max_allocated (GiB)': '33.83', 'memory/device_reserved (GiB)': '35.06', 'tokens/train_per_sec_per_gpu': '33.93', 'tokens/total': 2001472, 'tokens/trainable': 30114, 'epoch': '0.2578'}
	13%\|█▎ \| 66/512 [07:19<51:13, 6.89s/it] 13%\|█▎ \| 67/512 [07:25<49:28, 6.67s/it] {'loss': '0.0327', 'grad_norm': '0.5852', 'learning_rate': '9.844e-05', 'ppl': '1.033', 'memory/max_active (GiB)': '33.78', 'memory/max_allocated (GiB)': '33.78', 'memory/device_reserved (GiB)': '35.65', 'tokens/train_per_sec_per_gpu': '38.04', 'tokens/total': 2029696, 'tokens/trainable': 30594, 'epoch': '0.2617'}
	13%\|█▎ \| 67/512 [07:25<49:28, 6.67s/it] 13%\|█▎ \| 68/512 [07:32<49:15, 6.66s/it] {'loss': '0.01145', 'grad_norm': '0.3638', 'learning_rate': '9.836e-05', 'ppl': '1.012', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '35.65', 'tokens/train_per_sec_per_gpu': '35.62', 'tokens/total': 2060144, 'tokens/trainable': 31073, 'epoch': '0.2656'}
	13%\|█▎ \| 68/512 [07:32<49:15, 6.66s/it] 13%\|█▎ \| 69/512 [07:39<49:06, 6.65s/it] {'loss': '0.03206', 'grad_norm': '0.4834', 'learning_rate': '9.828e-05', 'ppl': '1.033', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '35.65', 'tokens/train_per_sec_per_gpu': '34.38', 'tokens/total': 2090448, 'tokens/trainable': 31553, 'epoch': '0.2695'}
	13%\|█▎ \| 69/512 [07:39<49:06, 6.65s/it] 14%\|█▎ \| 70/512 [07:45<48:49, 6.63s/it] {'loss': '0.01643', 'grad_norm': '0.58', 'learning_rate': '9.82e-05', 'ppl': '1.017', 'memory/max_active (GiB)': '33.77', 'memory/max_allocated (GiB)': '33.77', 'memory/device_reserved (GiB)': '35.65', 'tokens/train_per_sec_per_gpu': '32.72', 'tokens/total': 2120608, 'tokens/trainable': 32004, 'epoch': '0.2734'}
	14%\|█▎ \| 70/512 [07:45<48:49, 6.63s/it] 14%\|█▍ \| 71/512 [07:51<47:29, 6.46s/it] {'loss': '0.04277', 'grad_norm': '0.6351', 'learning_rate': '9.812e-05', 'ppl': '1.044', 'memory/max_active (GiB)': '33.35', 'memory/max_allocated (GiB)': '33.35', 'memory/device_reserved (GiB)': '35.65', 'tokens/train_per_sec_per_gpu': '32.26', 'tokens/total': 2148848, 'tokens/trainable': 32419, 'epoch': '0.2773'}
	14%\|█▍ \| 71/512 [07:51<47:29, 6.46s/it] 14%\|█▍ \| 72/512 [07:58<47:42, 6.51s/it] {'loss': '0.009183', 'grad_norm': '0.2791', 'learning_rate': '9.803e-05', 'ppl': '1.009', 'memory/max_active (GiB)': '33.81', 'memory/max_allocated (GiB)': '33.81', 'memory/device_reserved (GiB)': '35.65', 'tokens/train_per_sec_per_gpu': '34.98', 'tokens/total': 2179200, 'tokens/trainable': 32878, 'epoch': '0.2812'}
	14%\|█▍ \| 72/512 [07:58<47:42, 6.51s/it] 14%\|█▍ \| 73/512 [08:05<47:44, 6.53s/it] {'loss': '0.0283', 'grad_norm': '0.7224', 'learning_rate': '9.795e-05', 'ppl': '1.029', 'memory/max_active (GiB)': '33.92', 'memory/max_allocated (GiB)': '33.92', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '36.21', 'tokens/total': 2209712, 'tokens/trainable': 33348, 'epoch': '0.2852'}
	14%\|█▍ \| 73/512 [08:05<47:44, 6.53s/it] 14%\|█▍ \| 74/512 [08:11<48:00, 6.58s/it] {'loss': '0.01015', 'grad_norm': '0.6147', 'learning_rate': '9.786e-05', 'ppl': '1.01', 'memory/max_active (GiB)': '33.83', 'memory/max_allocated (GiB)': '33.83', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '32.86', 'tokens/total': 2240240, 'tokens/trainable': 33815, 'epoch': '0.2891'}
	14%\|█▍ \| 74/512 [08:11<48:00, 6.58s/it] 15%\|█▍ \| 75/512 [08:18<47:53, 6.58s/it] {'loss': '0.01202', 'grad_norm': '0.2888', 'learning_rate': '9.777e-05', 'ppl': '1.012', 'memory/max_active (GiB)': '33.74', 'memory/max_allocated (GiB)': '33.74', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '34.86', 'tokens/total': 2270336, 'tokens/trainable': 34229, 'epoch': '0.293'}
	15%\|█▍ \| 75/512 [08:18<47:53, 6.58s/it] 15%\|█▍ \| 76/512 [08:24<47:56, 6.60s/it] {'loss': '0.02341', 'grad_norm': '0.4517', 'learning_rate': '9.768e-05', 'ppl': '1.024', 'memory/max_active (GiB)': '33.93', 'memory/max_allocated (GiB)': '33.93', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '36.53', 'tokens/total': 2300784, 'tokens/trainable': 34718, 'epoch': '0.2969'}
	15%\|█▍ \| 76/512 [08:24<47:56, 6.60s/it] 15%\|█▌ \| 77/512 [08:31<47:46, 6.59s/it] {'loss': '0.05283', 'grad_norm': '0.9155', 'learning_rate': '9.759e-05', 'ppl': '1.054', 'memory/max_active (GiB)': '33.76', 'memory/max_allocated (GiB)': '33.76', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '32.14', 'tokens/total': 2330880, 'tokens/trainable': 35146, 'epoch': '0.3008'}
	15%\|█▌ \| 77/512 [08:31<47:46, 6.59s/it] 15%\|█▌ \| 78/512 [08:38<47:52, 6.62s/it] {'loss': '0.009073', 'grad_norm': '0.3222', 'learning_rate': '9.749e-05', 'ppl': '1.009', 'memory/max_active (GiB)': '33.99', 'memory/max_allocated (GiB)': '33.99', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '35.29', 'tokens/total': 2361392, 'tokens/trainable': 35610, 'epoch': '0.3047'}
	15%\|█▌ \| 78/512 [08:38<47:52, 6.62s/it] 15%\|█▌ \| 79/512 [08:44<47:40, 6.61s/it] {'loss': '0.02443', 'grad_norm': '0.4267', 'learning_rate': '9.74e-05', 'ppl': '1.025', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '29.95', 'tokens/total': 2391744, 'tokens/trainable': 36037, 'epoch': '0.3086'}
	15%\|█▌ \| 79/512 [08:44<47:40, 6.61s/it] 16%\|█▌ \| 80/512 [08:51<47:38, 6.62s/it] {'loss': '0.01513', 'grad_norm': '0.3153', 'learning_rate': '9.73e-05', 'ppl': '1.015', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '38.73', 'tokens/total': 2422240, 'tokens/trainable': 36522, 'epoch': '0.3125'}
	16%\|█▌ \| 80/512 [08:51<47:38, 6.62s/it] 16%\|█▌ \| 81/512 [08:58<47:28, 6.61s/it] {'loss': '0.02345', 'grad_norm': '0.6696', 'learning_rate': '9.72e-05', 'ppl': '1.024', 'memory/max_active (GiB)': '33.89', 'memory/max_allocated (GiB)': '33.89', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '33.67', 'tokens/total': 2452560, 'tokens/trainable': 36979, 'epoch': '0.3164'}
	16%\|█▌ \| 81/512 [08:58<47:28, 6.61s/it] 16%\|█▌ \| 82/512 [09:04<46:20, 6.47s/it] {'loss': '0.0115', 'grad_norm': '0.3085', 'learning_rate': '9.71e-05', 'ppl': '1.012', 'memory/max_active (GiB)': '33.44', 'memory/max_allocated (GiB)': '33.44', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '33.93', 'tokens/total': 2481024, 'tokens/trainable': 37439, 'epoch': '0.3203'}
	16%\|█▌ \| 82/512 [09:04<46:20, 6.47s/it] 16%\|█▌ \| 83/512 [09:10<46:33, 6.51s/it] {'loss': '0.01644', 'grad_norm': '0.4444', 'learning_rate': '9.699e-05', 'ppl': '1.017', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '33.54', 'tokens/total': 2511296, 'tokens/trainable': 37892, 'epoch': '0.3242'}
	16%\|█▌ \| 83/512 [09:10<46:33, 6.51s/it] 16%\|█▋ \| 84/512 [09:17<46:42, 6.55s/it] {'loss': '0.03258', 'grad_norm': '1.124', 'learning_rate': '9.689e-05', 'ppl': '1.033', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '33.19', 'tokens/total': 2541792, 'tokens/trainable': 38355, 'epoch': '0.3281'}
	16%\|█▋ \| 84/512 [09:17<46:42, 6.55s/it] 17%\|█▋ \| 85/512 [09:24<46:43, 6.57s/it] {'loss': '0.01193', 'grad_norm': '0.4937', 'learning_rate': '9.678e-05', 'ppl': '1.012', 'memory/max_active (GiB)': '33.69', 'memory/max_allocated (GiB)': '33.69', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '34.07', 'tokens/total': 2572032, 'tokens/trainable': 38810, 'epoch': '0.332'}
	17%\|█▋ \| 85/512 [09:24<46:43, 6.57s/it] 17%\|█▋ \| 86/512 [09:30<46:41, 6.58s/it] {'loss': '0.04201', 'grad_norm': '1.324', 'learning_rate': '9.667e-05', 'ppl': '1.043', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '30.89', 'tokens/total': 2602512, 'tokens/trainable': 39240, 'epoch': '0.3359'}
	17%\|█▋ \| 86/512 [09:30<46:41, 6.58s/it] 17%\|█▋ \| 87/512 [09:37<46:41, 6.59s/it] {'loss': '0.02861', 'grad_norm': '0.4353', 'learning_rate': '9.656e-05', 'ppl': '1.029', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '37', 'tokens/total': 2632880, 'tokens/trainable': 39748, 'epoch': '0.3398'}
	17%\|█▋ \| 87/512 [09:37<46:41, 6.59s/it] 17%\|█▋ \| 88/512 [09:43<46:24, 6.57s/it] {'loss': '0.036', 'grad_norm': '0.4899', 'learning_rate': '9.645e-05', 'ppl': '1.037', 'memory/max_active (GiB)': '33.73', 'memory/max_allocated (GiB)': '33.73', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '33.35', 'tokens/total': 2663040, 'tokens/trainable': 40188, 'epoch': '0.3438'}
	17%\|█▋ \| 88/512 [09:43<46:24, 6.57s/it] 17%\|█▋ \| 89/512 [09:50<46:17, 6.57s/it] {'loss': '0.02382', 'grad_norm': '0.4914', 'learning_rate': '9.633e-05', 'ppl': '1.024', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '33.36', 'tokens/total': 2693504, 'tokens/trainable': 40632, 'epoch': '0.3477'}
	17%\|█▋ \| 89/512 [09:50<46:17, 6.57s/it] 18%\|█▊ \| 90/512 [09:56<46:11, 6.57s/it] {'loss': '0.02294', 'grad_norm': '0.4422', 'learning_rate': '9.622e-05', 'ppl': '1.023', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '31.37', 'tokens/total': 2723792, 'tokens/trainable': 41050, 'epoch': '0.3516'}
	18%\|█▊ \| 90/512 [09:56<46:11, 6.57s/it] 18%\|█▊ \| 91/512 [10:03<46:10, 6.58s/it] {'loss': '0.01759', 'grad_norm': '0.4762', 'learning_rate': '9.61e-05', 'ppl': '1.018', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '37.05', 'tokens/total': 2754048, 'tokens/trainable': 41515, 'epoch': '0.3555'}
	18%\|█▊ \| 91/512 [10:03<46:10, 6.58s/it] 18%\|█▊ \| 92/512 [10:10<46:02, 6.58s/it] {'loss': '0.03154', 'grad_norm': '0.4137', 'learning_rate': '9.598e-05', 'ppl': '1.032', 'memory/max_active (GiB)': '33.89', 'memory/max_allocated (GiB)': '33.89', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '36.23', 'tokens/total': 2784480, 'tokens/trainable': 41972, 'epoch': '0.3594'}
	18%\|█▊ \| 92/512 [10:10<46:02, 6.58s/it] 18%\|█▊ \| 93/512 [10:16<45:59, 6.59s/it] {'loss': '0.02375', 'grad_norm': '0.527', 'learning_rate': '9.586e-05', 'ppl': '1.024', 'memory/max_active (GiB)': '33.83', 'memory/max_allocated (GiB)': '33.83', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '35.43', 'tokens/total': 2814928, 'tokens/trainable': 42442, 'epoch': '0.3633'}
	18%\|█▊ \| 93/512 [10:16<45:59, 6.59s/it] 18%\|█▊ \| 94/512 [10:23<45:57, 6.60s/it] {'loss': '0.01261', 'grad_norm': '0.3022', 'learning_rate': '9.574e-05', 'ppl': '1.013', 'memory/max_active (GiB)': '33.97', 'memory/max_allocated (GiB)': '33.97', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '37.18', 'tokens/total': 2845408, 'tokens/trainable': 42928, 'epoch': '0.3672'}
	18%\|█▊ \| 94/512 [10:23<45:57, 6.60s/it] 19%\|█▊ \| 95/512 [10:29<45:56, 6.61s/it] {'loss': '0.02136', 'grad_norm': '0.3438', 'learning_rate': '9.562e-05', 'ppl': '1.022', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '39.28', 'tokens/total': 2875760, 'tokens/trainable': 43426, 'epoch': '0.3711'}
	19%\|█▊ \| 95/512 [10:29<45:56, 6.61s/it] 19%\|█▉ \| 96/512 [10:36<45:51, 6.62s/it] {'loss': '0.01179', 'grad_norm': '0.5729', 'learning_rate': '9.549e-05', 'ppl': '1.012', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '33.52', 'tokens/total': 2906096, 'tokens/trainable': 43869, 'epoch': '0.375'}
	19%\|█▉ \| 96/512 [10:36<45:51, 6.62s/it][2026-08-18 14:28:33,847] [INFO] [axolotl.core.trainers.base._save:828] [PID:12244] Saving model checkpoint to /workspace/wave/training/checkpoints/checkpoint-96
	19%\|█▉ \| 97/512 [10:44<48:45, 7.05s/it] {'loss': '0.002208', 'grad_norm': '0.1258', 'learning_rate': '9.536e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.89', 'memory/max_allocated (GiB)': '33.89', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '32.02', 'tokens/total': 2936560, 'tokens/trainable': 44300, 'epoch': '0.3789'}
	19%\|█▉ \| 97/512 [10:44<48:45, 7.05s/it] 19%\|█▉ \| 98/512 [10:51<47:43, 6.92s/it] {'loss': '0.0156', 'grad_norm': '0.4312', 'learning_rate': '9.523e-05', 'ppl': '1.016', 'memory/max_active (GiB)': '33.92', 'memory/max_allocated (GiB)': '33.92', 'memory/device_reserved (GiB)': '35.19', 'tokens/train_per_sec_per_gpu': '39.37', 'tokens/total': 2966880, 'tokens/trainable': 44791, 'epoch': '0.3828'}
	19%\|█▉ \| 98/512 [10:51<47:43, 6.92s/it] 19%\|█▉ \| 99/512 [10:57<47:09, 6.85s/it] {'loss': '0.01187', 'grad_norm': '0.5057', 'learning_rate': '9.51e-05', 'ppl': '1.012', 'memory/max_active (GiB)': '34.03', 'memory/max_allocated (GiB)': '34.03', 'memory/device_reserved (GiB)': '35.76', 'tokens/train_per_sec_per_gpu': '36.71', 'tokens/total': 2997472, 'tokens/trainable': 45254, 'epoch': '0.3867'}
	19%\|█▉ \| 99/512 [10:57<47:09, 6.85s/it] 20%\|█▉ \| 100/512 [11:04<46:30, 6.77s/it] {'loss': '0.01458', 'grad_norm': '3.06', 'learning_rate': '9.497e-05', 'ppl': '1.015', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '33.57', 'tokens/total': 3027840, 'tokens/trainable': 45698, 'epoch': '0.3906'}
	20%\|█▉ \| 100/512 [11:04<46:30, 6.77s/it] 20%\|█▉ \| 101/512 [11:11<46:03, 6.72s/it] {'loss': '0.009431', 'grad_norm': '0.3825', 'learning_rate': '9.484e-05', 'ppl': '1.009', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '33.57', 'tokens/total': 3058080, 'tokens/trainable': 46141, 'epoch': '0.3945'}
	20%\|█▉ \| 101/512 [11:11<46:03, 6.72s/it] 20%\|█▉ \| 102/512 [11:17<45:45, 6.70s/it] {'loss': '0.05128', 'grad_norm': '0.8783', 'learning_rate': '9.47e-05', 'ppl': '1.053', 'memory/max_active (GiB)': '33.89', 'memory/max_allocated (GiB)': '33.89', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '33.05', 'tokens/total': 3088656, 'tokens/trainable': 46598, 'epoch': '0.3984'}
	20%\|█▉ \| 102/512 [11:17<45:45, 6.70s/it] 20%\|██ \| 103/512 [11:24<45:26, 6.67s/it] {'loss': '0.009158', 'grad_norm': '0.4222', 'learning_rate': '9.456e-05', 'ppl': '1.009', 'memory/max_active (GiB)': '33.91', 'memory/max_allocated (GiB)': '33.91', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '32.45', 'tokens/total': 3119232, 'tokens/trainable': 47065, 'epoch': '0.4023'}
	20%\|██ \| 103/512 [11:24<45:26, 6.67s/it] 20%\|██ \| 104/512 [11:30<45:12, 6.65s/it] {'loss': '0.01385', 'grad_norm': '1.263', 'learning_rate': '9.442e-05', 'ppl': '1.014', 'memory/max_active (GiB)': '33.81', 'memory/max_allocated (GiB)': '33.81', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '38.29', 'tokens/total': 3149504, 'tokens/trainable': 47554, 'epoch': '0.4062'}
	20%\|██ \| 104/512 [11:30<45:12, 6.65s/it] 21%\|██ \| 105/512 [11:37<45:02, 6.64s/it] {'loss': '0.008239', 'grad_norm': '0.3073', 'learning_rate': '9.428e-05', 'ppl': '1.008', 'memory/max_active (GiB)': '33.83', 'memory/max_allocated (GiB)': '33.83', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '36.88', 'tokens/total': 3179872, 'tokens/trainable': 48047, 'epoch': '0.4102'}
	21%\|██ \| 105/512 [11:37<45:02, 6.64s/it] 21%\|██ \| 106/512 [11:43<43:49, 6.48s/it] {'loss': '0.03211', 'grad_norm': '0.7416', 'learning_rate': '9.414e-05', 'ppl': '1.033', 'memory/max_active (GiB)': '33.39', 'memory/max_allocated (GiB)': '33.39', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '34.62', 'tokens/total': 3208128, 'tokens/trainable': 48498, 'epoch': '0.4141'}
	21%\|██ \| 106/512 [11:43<43:49, 6.48s/it] 21%\|██ \| 107/512 [11:50<43:55, 6.51s/it] {'loss': '0.009182', 'grad_norm': '0.2502', 'learning_rate': '9.4e-05', 'ppl': '1.009', 'memory/max_active (GiB)': '33.74', 'memory/max_allocated (GiB)': '33.74', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '31.18', 'tokens/total': 3238224, 'tokens/trainable': 48933, 'epoch': '0.418'}
	21%\|██ \| 107/512 [11:50<43:55, 6.51s/it] 21%\|██ \| 108/512 [11:56<44:00, 6.54s/it] {'loss': '0.01505', 'grad_norm': '0.5458', 'learning_rate': '9.385e-05', 'ppl': '1.015', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '38.48', 'tokens/total': 3268592, 'tokens/trainable': 49404, 'epoch': '0.4219'}
	21%\|██ \| 108/512 [11:56<44:00, 6.54s/it] 21%\|██▏ \| 109/512 [12:03<44:03, 6.56s/it] {'loss': '0.007351', 'grad_norm': '0.2509', 'learning_rate': '9.37e-05', 'ppl': '1.007', 'memory/max_active (GiB)': '33.79', 'memory/max_allocated (GiB)': '33.79', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '37.32', 'tokens/total': 3298720, 'tokens/trainable': 49871, 'epoch': '0.4258'}
	21%\|██▏ \| 109/512 [12:03<44:03, 6.56s/it] 21%\|██▏ \| 110/512 [12:10<44:03, 6.58s/it] {'loss': '0.01421', 'grad_norm': '0.2964', 'learning_rate': '9.355e-05', 'ppl': '1.014', 'memory/max_active (GiB)': '33.99', 'memory/max_allocated (GiB)': '33.99', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '33.58', 'tokens/total': 3329040, 'tokens/trainable': 50311, 'epoch': '0.4297'}
	21%\|██▏ \| 110/512 [12:10<44:03, 6.58s/it] 22%\|██▏ \| 111/512 [12:16<44:01, 6.59s/it] {'loss': '0.0107', 'grad_norm': '0.3097', 'learning_rate': '9.34e-05', 'ppl': '1.011', 'memory/max_active (GiB)': '33.8', 'memory/max_allocated (GiB)': '33.8', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '33.86', 'tokens/total': 3359248, 'tokens/trainable': 50754, 'epoch': '0.4336'}
	22%\|██▏ \| 111/512 [12:16<44:01, 6.59s/it] 22%\|██▏ \| 112/512 [12:23<43:53, 6.58s/it] {'loss': '0.01319', 'grad_norm': '0.4973', 'learning_rate': '9.325e-05', 'ppl': '1.013', 'memory/max_active (GiB)': '33.74', 'memory/max_allocated (GiB)': '33.74', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '32.1', 'tokens/total': 3389280, 'tokens/trainable': 51200, 'epoch': '0.4375'}
	22%\|██▏ \| 112/512 [12:23<43:53, 6.58s/it] 22%\|██▏ \| 113/512 [12:29<43:51, 6.60s/it] {'loss': '0.006971', 'grad_norm': '0.4105', 'learning_rate': '9.31e-05', 'ppl': '1.007', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '33.22', 'tokens/total': 3419728, 'tokens/trainable': 51656, 'epoch': '0.4414'}
	22%\|██▏ \| 113/512 [12:29<43:51, 6.60s/it] 22%\|██▏ \| 114/512 [12:36<43:41, 6.59s/it] {'loss': '0.003387', 'grad_norm': '0.2707', 'learning_rate': '9.294e-05', 'ppl': '1.003', 'memory/max_active (GiB)': '33.78', 'memory/max_allocated (GiB)': '33.78', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '30.92', 'tokens/total': 3449984, 'tokens/trainable': 52069, 'epoch': '0.4453'}
	22%\|██▏ \| 114/512 [12:36<43:41, 6.59s/it] 22%\|██▏ \| 115/512 [12:43<43:37, 6.59s/it] {'loss': '0.008487', 'grad_norm': '0.5602', 'learning_rate': '9.278e-05', 'ppl': '1.009', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '36.94', 'tokens/total': 3480352, 'tokens/trainable': 52559, 'epoch': '0.4492'}
	22%\|██▏ \| 115/512 [12:43<43:37, 6.59s/it] 23%\|██▎ \| 116/512 [12:49<43:35, 6.60s/it] {'loss': '0.02243', 'grad_norm': '0.5224', 'learning_rate': '9.263e-05', 'ppl': '1.023', 'memory/max_active (GiB)': '33.81', 'memory/max_allocated (GiB)': '33.81', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '39.22', 'tokens/total': 3510656, 'tokens/trainable': 53061, 'epoch': '0.4531'}
	23%\|██▎ \| 116/512 [12:49<43:35, 6.60s/it] 23%\|██▎ \| 117/512 [12:56<43:28, 6.60s/it] {'loss': '0.02046', 'grad_norm': '1.221', 'learning_rate': '9.247e-05', 'ppl': '1.021', 'memory/max_active (GiB)': '33.96', 'memory/max_allocated (GiB)': '33.96', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '35.89', 'tokens/total': 3541104, 'tokens/trainable': 53520, 'epoch': '0.457'}
	23%\|██▎ \| 117/512 [12:56<43:28, 6.60s/it] 23%\|██▎ \| 118/512 [13:02<43:24, 6.61s/it] {'loss': '0.0256', 'grad_norm': '1.301', 'learning_rate': '9.23e-05', 'ppl': '1.026', 'memory/max_active (GiB)': '33.97', 'memory/max_allocated (GiB)': '33.97', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '33.36', 'tokens/total': 3571744, 'tokens/trainable': 53976, 'epoch': '0.4609'}
	23%\|██▎ \| 118/512 [13:02<43:24, 6.61s/it] 23%\|██▎ \| 119/512 [13:09<43:11, 6.60s/it] {'loss': '0.02888', 'grad_norm': '1.099', 'learning_rate': '9.214e-05', 'ppl': '1.029', 'memory/max_active (GiB)': '33.79', 'memory/max_allocated (GiB)': '33.79', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '34.93', 'tokens/total': 3601808, 'tokens/trainable': 54408, 'epoch': '0.4648'}
	23%\|██▎ \| 119/512 [13:09<43:11, 6.60s/it] 23%\|██▎ \| 120/512 [13:16<43:05, 6.60s/it] {'loss': '0.01679', 'grad_norm': '0.3977', 'learning_rate': '9.198e-05', 'ppl': '1.017', 'memory/max_active (GiB)': '33.81', 'memory/max_allocated (GiB)': '33.81', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '37.14', 'tokens/total': 3632096, 'tokens/trainable': 54884, 'epoch': '0.4688'}
	23%\|██▎ \| 120/512 [13:16<43:05, 6.60s/it] 24%\|██▎ \| 121/512 [13:22<42:57, 6.59s/it] {'loss': '0.02264', 'grad_norm': '0.4873', 'learning_rate': '9.181e-05', 'ppl': '1.023', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '33.15', 'tokens/total': 3662416, 'tokens/trainable': 55303, 'epoch': '0.4727'}
	24%\|██▎ \| 121/512 [13:22<42:57, 6.59s/it] 24%\|██▍ \| 122/512 [13:29<42:52, 6.60s/it] {'loss': '0.02327', 'grad_norm': '0.4664', 'learning_rate': '9.164e-05', 'ppl': '1.024', 'memory/max_active (GiB)': '33.97', 'memory/max_allocated (GiB)': '33.97', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '31.94', 'tokens/total': 3692992, 'tokens/trainable': 55743, 'epoch': '0.4766'}
	24%\|██▍ \| 122/512 [13:29<42:52, 6.60s/it] 24%\|██▍ \| 123/512 [13:35<42:39, 6.58s/it] {'loss': '0.00781', 'grad_norm': '0.2307', 'learning_rate': '9.147e-05', 'ppl': '1.008', 'memory/max_active (GiB)': '33.73', 'memory/max_allocated (GiB)': '33.73', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '33.95', 'tokens/total': 3722992, 'tokens/trainable': 56187, 'epoch': '0.4805'}
	24%\|██▍ \| 123/512 [13:35<42:39, 6.58s/it] 24%\|██▍ \| 124/512 [13:42<42:31, 6.58s/it] {'loss': '0.006973', 'grad_norm': '0.1732', 'learning_rate': '9.13e-05', 'ppl': '1.007', 'memory/max_active (GiB)': '33.77', 'memory/max_allocated (GiB)': '33.77', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '33.63', 'tokens/total': 3753184, 'tokens/trainable': 56632, 'epoch': '0.4844'}
	24%\|██▍ \| 124/512 [13:42<42:31, 6.58s/it] 24%\|██▍ \| 125/512 [13:49<42:31, 6.59s/it] {'loss': '0.006869', 'grad_norm': '0.1999', 'learning_rate': '9.113e-05', 'ppl': '1.007', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '37.26', 'tokens/total': 3783440, 'tokens/trainable': 57124, 'epoch': '0.4883'}
	24%\|██▍ \| 125/512 [13:49<42:31, 6.59s/it] 25%\|██▍ \| 126/512 [13:55<42:21, 6.58s/it] {'loss': '0.004266', 'grad_norm': '0.1133', 'learning_rate': '9.096e-05', 'ppl': '1.004', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '33.38', 'tokens/total': 3813840, 'tokens/trainable': 57570, 'epoch': '0.4922'}
	25%\|██▍ \| 126/512 [13:55<42:21, 6.58s/it] 25%\|██▍ \| 127/512 [14:02<42:21, 6.60s/it] {'loss': '0.02041', 'grad_norm': '0.4156', 'learning_rate': '9.078e-05', 'ppl': '1.021', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '35.36', 'tokens/total': 3844464, 'tokens/trainable': 58048, 'epoch': '0.4961'}
	25%\|██▍ \| 127/512 [14:02<42:21, 6.60s/it] 25%\|██▌ \| 128/512 [14:08<41:16, 6.45s/it] {'loss': '0.003154', 'grad_norm': '0.0982', 'learning_rate': '9.06e-05', 'ppl': '1.003', 'memory/max_active (GiB)': '33.39', 'memory/max_allocated (GiB)': '33.39', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '35.79', 'tokens/total': 3872816, 'tokens/trainable': 58497, 'epoch': '0.5'}
	25%\|██▌ \| 128/512 [14:08<41:16, 6.45s/it][2026-08-18 14:32:05,611] [INFO] [axolotl.core.trainers.base._save:828] [PID:12244] Saving model checkpoint to /workspace/wave/training/checkpoints/checkpoint-128
	25%\|██▌ \| 129/512 [14:16<44:24, 6.96s/it] {'loss': '0.01777', 'grad_norm': '0.3427', 'learning_rate': '9.043e-05', 'ppl': '1.018', 'memory/max_active (GiB)': '33.77', 'memory/max_allocated (GiB)': '33.77', 'memory/device_reserved (GiB)': '36', 'tokens/train_per_sec_per_gpu': '33.44', 'tokens/total': 3902992, 'tokens/trainable': 58947, 'epoch': '0.5039'}
	25%\|██▌ \| 129/512 [14:16<44:24, 6.96s/it] 25%\|██▌ \| 130/512 [14:23<43:37, 6.85s/it] {'loss': '0.01817', 'grad_norm': '0.3253', 'learning_rate': '9.025e-05', 'ppl': '1.018', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '34.74', 'tokens/train_per_sec_per_gpu': '37.56', 'tokens/total': 3933456, 'tokens/trainable': 59415, 'epoch': '0.5078'}
	25%\|██▌ \| 130/512 [14:23<43:37, 6.85s/it] 26%\|██▌ \| 131/512 [14:29<43:01, 6.78s/it] {'loss': '0.008536', 'grad_norm': '0.3168', 'learning_rate': '9.007e-05', 'ppl': '1.009', 'memory/max_active (GiB)': '33.78', 'memory/max_allocated (GiB)': '33.78', 'memory/device_reserved (GiB)': '34.91', 'tokens/train_per_sec_per_gpu': '32.75', 'tokens/total': 3963536, 'tokens/trainable': 59851, 'epoch': '0.5117'}
	26%\|██▌ \| 131/512 [14:29<43:01, 6.78s/it] 26%\|██▌ \| 132/512 [14:36<42:33, 6.72s/it] {'loss': '0.001934', 'grad_norm': '0.124', 'learning_rate': '8.988e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.91', 'memory/max_allocated (GiB)': '33.91', 'memory/device_reserved (GiB)': '35.78', 'tokens/train_per_sec_per_gpu': '36.25', 'tokens/total': 3994000, 'tokens/trainable': 60325, 'epoch': '0.5156'}
	26%\|██▌ \| 132/512 [14:36<42:33, 6.72s/it] 26%\|██▌ \| 133/512 [14:42<42:15, 6.69s/it] {'loss': '0.0155', 'grad_norm': '0.2586', 'learning_rate': '8.97e-05', 'ppl': '1.016', 'memory/max_active (GiB)': '33.89', 'memory/max_allocated (GiB)': '33.89', 'memory/device_reserved (GiB)': '35.78', 'tokens/train_per_sec_per_gpu': '35.97', 'tokens/total': 4024352, 'tokens/trainable': 60818, 'epoch': '0.5195'}
	26%\|██▌ \| 133/512 [14:42<42:15, 6.69s/it] 26%\|██▌ \| 134/512 [14:49<41:02, 6.51s/it] {'loss': '0.006828', 'grad_norm': '0.3528', 'learning_rate': '8.951e-05', 'ppl': '1.007', 'memory/max_active (GiB)': '33.39', 'memory/max_allocated (GiB)': '33.39', 'memory/device_reserved (GiB)': '35.78', 'tokens/train_per_sec_per_gpu': '32.28', 'tokens/total': 4052832, 'tokens/trainable': 61263, 'epoch': '0.5234'}
	26%\|██▌ \| 134/512 [14:49<41:02, 6.51s/it] 26%\|██▋ \| 135/512 [14:55<41:07, 6.55s/it] {'loss': '0.01065', 'grad_norm': '0.3797', 'learning_rate': '8.933e-05', 'ppl': '1.011', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '35.78', 'tokens/train_per_sec_per_gpu': '36.58', 'tokens/total': 4083216, 'tokens/trainable': 61734, 'epoch': '0.5273'}
	26%\|██▋ \| 135/512 [14:55<41:07, 6.55s/it] 27%\|██▋ \| 136/512 [15:02<41:00, 6.54s/it] {'loss': '0.02273', 'grad_norm': '0.359', 'learning_rate': '8.914e-05', 'ppl': '1.023', 'memory/max_active (GiB)': '33.7', 'memory/max_allocated (GiB)': '33.7', 'memory/device_reserved (GiB)': '35.78', 'tokens/train_per_sec_per_gpu': '38.83', 'tokens/total': 4113392, 'tokens/trainable': 62210, 'epoch': '0.5312'}
	27%\|██▋ \| 136/512 [15:02<41:00, 6.54s/it] 27%\|██▋ \| 137/512 [15:08<41:02, 6.57s/it] {'loss': '0.01776', 'grad_norm': '0.3439', 'learning_rate': '8.895e-05', 'ppl': '1.018', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '35.78', 'tokens/train_per_sec_per_gpu': '36.87', 'tokens/total': 4143696, 'tokens/trainable': 62707, 'epoch': '0.5352'}
	27%\|██▋ \| 137/512 [15:08<41:02, 6.57s/it] 27%\|██▋ \| 138/512 [15:15<40:58, 6.57s/it] {'loss': '0.01984', 'grad_norm': '0.2429', 'learning_rate': '8.876e-05', 'ppl': '1.02', 'memory/max_active (GiB)': '33.79', 'memory/max_allocated (GiB)': '33.79', 'memory/device_reserved (GiB)': '35.78', 'tokens/train_per_sec_per_gpu': '37.17', 'tokens/total': 4174000, 'tokens/trainable': 63166, 'epoch': '0.5391'}
	27%\|██▋ \| 138/512 [15:15<40:58, 6.57s/it] 27%\|██▋ \| 139/512 [15:22<40:58, 6.59s/it] {'loss': '0.01264', 'grad_norm': '0.6662', 'learning_rate': '8.856e-05', 'ppl': '1.013', 'memory/max_active (GiB)': '33.92', 'memory/max_allocated (GiB)': '33.92', 'memory/device_reserved (GiB)': '35.79', 'tokens/train_per_sec_per_gpu': '36.96', 'tokens/total': 4204432, 'tokens/trainable': 63646, 'epoch': '0.543'}
	27%\|██▋ \| 139/512 [15:22<40:58, 6.59s/it] 27%\|██▋ \| 140/512 [15:28<39:50, 6.43s/it] {'loss': '0.01461', 'grad_norm': '0.2639', 'learning_rate': '8.837e-05', 'ppl': '1.015', 'memory/max_active (GiB)': '33.42', 'memory/max_allocated (GiB)': '33.42', 'memory/device_reserved (GiB)': '35.79', 'tokens/train_per_sec_per_gpu': '33.6', 'tokens/total': 4232752, 'tokens/trainable': 64063, 'epoch': '0.5469'}
	27%\|██▋ \| 140/512 [15:28<39:50, 6.43s/it] 28%\|██▊ \| 141/512 [15:34<40:05, 6.48s/it] {'loss': '0.01266', 'grad_norm': '0.2475', 'learning_rate': '8.817e-05', 'ppl': '1.013', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '35.79', 'tokens/train_per_sec_per_gpu': '35.67', 'tokens/total': 4262976, 'tokens/trainable': 64520, 'epoch': '0.5508'}
	28%\|██▊ \| 141/512 [15:34<40:05, 6.48s/it] 28%\|██▊ \| 142/512 [15:41<40:13, 6.52s/it] {'loss': '0.009737', 'grad_norm': '0.2115', 'learning_rate': '8.798e-05', 'ppl': '1.01', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '35.79', 'tokens/train_per_sec_per_gpu': '34.48', 'tokens/total': 4293520, 'tokens/trainable': 64958, 'epoch': '0.5547'}
	28%\|██▊ \| 142/512 [15:41<40:13, 6.52s/it] 28%\|██▊ \| 143/512 [15:47<40:10, 6.53s/it] {'loss': '0.01307', 'grad_norm': '0.2296', 'learning_rate': '8.778e-05', 'ppl': '1.013', 'memory/max_active (GiB)': '33.8', 'memory/max_allocated (GiB)': '33.8', 'memory/device_reserved (GiB)': '37.89', 'tokens/train_per_sec_per_gpu': '38.3', 'tokens/total': 4323520, 'tokens/trainable': 65437, 'epoch': '0.5586'}
	28%\|██▊ \| 143/512 [15:47<40:10, 6.53s/it] 28%\|██▊ \| 144/512 [15:54<40:09, 6.55s/it] {'loss': '0.01002', 'grad_norm': '0.2896', 'learning_rate': '8.758e-05', 'ppl': '1.01', 'memory/max_active (GiB)': '33.81', 'memory/max_allocated (GiB)': '33.81', 'memory/device_reserved (GiB)': '37.89', 'tokens/train_per_sec_per_gpu': '37.25', 'tokens/total': 4353616, 'tokens/trainable': 65903, 'epoch': '0.5625'}
	28%\|██▊ \| 144/512 [15:54<40:09, 6.55s/it] 28%\|██▊ \| 145/512 [16:01<40:06, 6.56s/it] {'loss': '0.02572', 'grad_norm': '0.4186', 'learning_rate': '8.738e-05', 'ppl': '1.026', 'memory/max_active (GiB)': '33.79', 'memory/max_allocated (GiB)': '33.79', 'memory/device_reserved (GiB)': '37.89', 'tokens/train_per_sec_per_gpu': '32.08', 'tokens/total': 4383856, 'tokens/trainable': 66334, 'epoch': '0.5664'}
	28%\|██▊ \| 145/512 [16:01<40:06, 6.56s/it] 29%\|██▊ \| 146/512 [16:07<40:07, 6.58s/it] {'loss': '0.01731', 'grad_norm': '0.278', 'learning_rate': '8.718e-05', 'ppl': '1.017', 'memory/max_active (GiB)': '33.74', 'memory/max_allocated (GiB)': '33.74', 'memory/device_reserved (GiB)': '37.89', 'tokens/train_per_sec_per_gpu': '32.01', 'tokens/total': 4414128, 'tokens/trainable': 66785, 'epoch': '0.5703'}
	29%\|██▊ \| 146/512 [16:07<40:07, 6.58s/it] 29%\|██▊ \| 147/512 [16:14<40:02, 6.58s/it] {'loss': '0.01725', 'grad_norm': '0.5759', 'learning_rate': '8.697e-05', 'ppl': '1.017', 'memory/max_active (GiB)': '33.83', 'memory/max_allocated (GiB)': '33.83', 'memory/device_reserved (GiB)': '37.89', 'tokens/train_per_sec_per_gpu': '35.81', 'tokens/total': 4444384, 'tokens/trainable': 67199, 'epoch': '0.5742'}
	29%\|██▊ \| 147/512 [16:14<40:02, 6.58s/it] 29%\|██▉ \| 148/512 [16:20<40:06, 6.61s/it] {'loss': '0.02244', 'grad_norm': '0.443', 'learning_rate': '8.677e-05', 'ppl': '1.023', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '37.89', 'tokens/train_per_sec_per_gpu': '35.64', 'tokens/total': 4474912, 'tokens/trainable': 67683, 'epoch': '0.5781'}
	29%\|██▉ \| 148/512 [16:20<40:06, 6.61s/it] 29%\|██▉ \| 149/512 [16:27<39:58, 6.61s/it] {'loss': '0.01044', 'grad_norm': '0.2663', 'learning_rate': '8.656e-05', 'ppl': '1.01', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '37.89', 'tokens/train_per_sec_per_gpu': '34.4', 'tokens/total': 4505328, 'tokens/trainable': 68127, 'epoch': '0.582'}
	29%\|██▉ \| 149/512 [16:27<39:58, 6.61s/it] 29%\|██▉ \| 150/512 [16:34<39:48, 6.60s/it] {'loss': '0.004219', 'grad_norm': '0.1525', 'learning_rate': '8.635e-05', 'ppl': '1.004', 'memory/max_active (GiB)': '33.8', 'memory/max_allocated (GiB)': '33.8', 'memory/device_reserved (GiB)': '37.89', 'tokens/train_per_sec_per_gpu': '33.75', 'tokens/total': 4535424, 'tokens/trainable': 68595, 'epoch': '0.5859'}
	29%\|██▉ \| 150/512 [16:34<39:48, 6.60s/it] 29%\|██▉ \| 151/512 [16:40<39:44, 6.61s/it] {'loss': '0.01187', 'grad_norm': '0.2124', 'learning_rate': '8.615e-05', 'ppl': '1.012', 'memory/max_active (GiB)': '33.8', 'memory/max_allocated (GiB)': '33.8', 'memory/device_reserved (GiB)': '37.89', 'tokens/train_per_sec_per_gpu': '30.68', 'tokens/total': 4565536, 'tokens/trainable': 69062, 'epoch': '0.5898'}
	29%\|██▉ \| 151/512 [16:40<39:44, 6.61s/it] 30%\|██▉ \| 152/512 [16:47<39:39, 6.61s/it] {'loss': '0.007625', 'grad_norm': '0.5053', 'learning_rate': '8.594e-05', 'ppl': '1.008', 'memory/max_active (GiB)': '33.93', 'memory/max_allocated (GiB)': '33.93', 'memory/device_reserved (GiB)': '37.89', 'tokens/train_per_sec_per_gpu': '35.96', 'tokens/total': 4596048, 'tokens/trainable': 69509, 'epoch': '0.5938'}
	30%\|██▉ \| 152/512 [16:47<39:39, 6.61s/it] 30%\|██▉ \| 153/512 [16:53<39:36, 6.62s/it] {'loss': '0.01224', 'grad_norm': '0.4628', 'learning_rate': '8.572e-05', 'ppl': '1.012', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '37.89', 'tokens/train_per_sec_per_gpu': '34.45', 'tokens/total': 4626480, 'tokens/trainable': 69972, 'epoch': '0.5977'}
	30%\|██▉ \| 153/512 [16:53<39:36, 6.62s/it] 30%\|███ \| 154/512 [17:00<39:27, 6.61s/it] {'loss': '0.006034', 'grad_norm': '0.4228', 'learning_rate': '8.551e-05', 'ppl': '1.006', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '37.89', 'tokens/train_per_sec_per_gpu': '37.3', 'tokens/total': 4656800, 'tokens/trainable': 70439, 'epoch': '0.6016'}
	30%\|███ \| 154/512 [17:00<39:27, 6.61s/it] 30%\|███ \| 155/512 [17:06<38:25, 6.46s/it] {'loss': '0.01827', 'grad_norm': '0.599', 'learning_rate': '8.53e-05', 'ppl': '1.018', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '37.89', 'tokens/train_per_sec_per_gpu': '37.43', 'tokens/total': 4685120, 'tokens/trainable': 70881, 'epoch': '0.6055'}
	30%\|███ \| 155/512 [17:06<38:25, 6.46s/it] 30%\|███ \| 156/512 [17:13<38:35, 6.50s/it] {'loss': '0.01053', 'grad_norm': '0.6471', 'learning_rate': '8.508e-05', 'ppl': '1.011', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '37.89', 'tokens/train_per_sec_per_gpu': '34.48', 'tokens/total': 4715664, 'tokens/trainable': 71339, 'epoch': '0.6094'}
	30%\|███ \| 156/512 [17:13<38:35, 6.50s/it] 31%\|███ \| 157/512 [17:19<38:40, 6.54s/it] {'loss': '0.0001828', 'grad_norm': '0.00899', 'learning_rate': '8.487e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.79', 'memory/max_allocated (GiB)': '33.79', 'memory/device_reserved (GiB)': '37.89', 'tokens/train_per_sec_per_gpu': '35.55', 'tokens/total': 4745824, 'tokens/trainable': 71801, 'epoch': '0.6133'}
	31%\|███ \| 157/512 [17:19<38:40, 6.54s/it] 31%\|███ \| 158/512 [17:26<38:45, 6.57s/it] {'loss': '0.01429', 'grad_norm': '0.2966', 'learning_rate': '8.465e-05', 'ppl': '1.014', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '37.89', 'tokens/train_per_sec_per_gpu': '32.65', 'tokens/total': 4776224, 'tokens/trainable': 72237, 'epoch': '0.6172'}
	31%\|███ \| 158/512 [17:26<38:45, 6.57s/it] 31%\|███ \| 159/512 [17:33<38:36, 6.56s/it] {'loss': '0.02528', 'grad_norm': '0.3071', 'learning_rate': '8.443e-05', 'ppl': '1.026', 'memory/max_active (GiB)': '33.75', 'memory/max_allocated (GiB)': '33.75', 'memory/device_reserved (GiB)': '37.89', 'tokens/train_per_sec_per_gpu': '31.03', 'tokens/total': 4806304, 'tokens/trainable': 72664, 'epoch': '0.6211'}
	31%\|███ \| 159/512 [17:33<38:36, 6.56s/it] 31%\|███▏ \| 160/512 [17:39<38:34, 6.57s/it] {'loss': '0.00503', 'grad_norm': '0.2943', 'learning_rate': '8.421e-05', 'ppl': '1.005', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '37.89', 'tokens/train_per_sec_per_gpu': '32.1', 'tokens/total': 4836688, 'tokens/trainable': 73119, 'epoch': '0.625'}
	31%\|███▏ \| 160/512 [17:39<38:34, 6.57s/it][2026-08-18 14:35:36,933] [INFO] [axolotl.core.trainers.base._save:828] [PID:12244] Saving model checkpoint to /workspace/wave/training/checkpoints/checkpoint-160
	31%\|███▏ \| 161/512 [17:47<41:02, 7.02s/it] {'loss': '0.01076', 'grad_norm': '0.389', 'learning_rate': '8.399e-05', 'ppl': '1.011', 'memory/max_active (GiB)': '33.8', 'memory/max_allocated (GiB)': '33.8', 'memory/device_reserved (GiB)': '37.89', 'tokens/train_per_sec_per_gpu': '31.51', 'tokens/total': 4867008, 'tokens/trainable': 73540, 'epoch': '0.6289'}
	31%\|███▏ \| 161/512 [17:47<41:02, 7.02s/it] 32%\|███▏ \| 162/512 [17:54<40:15, 6.90s/it] {'loss': '0.003405', 'grad_norm': '0.2366', 'learning_rate': '8.376e-05', 'ppl': '1.003', 'memory/max_active (GiB)': '33.97', 'memory/max_allocated (GiB)': '33.97', 'memory/device_reserved (GiB)': '35.82', 'tokens/train_per_sec_per_gpu': '33.18', 'tokens/total': 4897584, 'tokens/trainable': 73990, 'epoch': '0.6328'}
	32%\|███▏ \| 162/512 [17:54<40:15, 6.90s/it] 32%\|███▏ \| 163/512 [18:00<39:37, 6.81s/it] {'loss': '0.01729', 'grad_norm': '0.2311', 'learning_rate': '8.354e-05', 'ppl': '1.017', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '35.82', 'tokens/train_per_sec_per_gpu': '35.89', 'tokens/total': 4927904, 'tokens/trainable': 74454, 'epoch': '0.6367'}
	32%\|███▏ \| 163/512 [18:00<39:37, 6.81s/it] 32%\|███▏ \| 164/512 [18:07<39:23, 6.79s/it] {'loss': '0.02588', 'grad_norm': '0.7158', 'learning_rate': '8.332e-05', 'ppl': '1.026', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '35.82', 'tokens/train_per_sec_per_gpu': '33.96', 'tokens/total': 4958496, 'tokens/trainable': 74905, 'epoch': '0.6406'}
	32%\|███▏ \| 164/512 [18:07<39:23, 6.79s/it] 32%\|███▏ \| 165/512 [18:14<38:54, 6.73s/it] {'loss': '0.003934', 'grad_norm': '0.3431', 'learning_rate': '8.309e-05', 'ppl': '1.004', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '35.82', 'tokens/train_per_sec_per_gpu': '31.03', 'tokens/total': 4988832, 'tokens/trainable': 75329, 'epoch': '0.6445'}
	32%\|███▏ \| 165/512 [18:14<38:54, 6.73s/it] 32%\|███▏ \| 166/512 [18:20<38:29, 6.68s/it] {'loss': '0.001242', 'grad_norm': '0.07044', 'learning_rate': '8.286e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.75', 'memory/max_allocated (GiB)': '33.75', 'memory/device_reserved (GiB)': '35.82', 'tokens/train_per_sec_per_gpu': '32.34', 'tokens/total': 5019136, 'tokens/trainable': 75770, 'epoch': '0.6484'}
	32%\|███▏ \| 166/512 [18:20<38:29, 6.68s/it] 33%\|███▎ \| 167/512 [18:27<38:25, 6.68s/it] {'loss': '0.008923', 'grad_norm': '0.6784', 'learning_rate': '8.263e-05', 'ppl': '1.009', 'memory/max_active (GiB)': '33.99', 'memory/max_allocated (GiB)': '33.99', 'memory/device_reserved (GiB)': '35.82', 'tokens/train_per_sec_per_gpu': '31.81', 'tokens/total': 5049808, 'tokens/trainable': 76231, 'epoch': '0.6523'}
	33%\|███▎ \| 167/512 [18:27<38:25, 6.68s/it] 33%\|███▎ \| 168/512 [18:34<38:12, 6.66s/it] {'loss': '0.01817', 'grad_norm': '0.4106', 'learning_rate': '8.24e-05', 'ppl': '1.018', 'memory/max_active (GiB)': '33.93', 'memory/max_allocated (GiB)': '33.93', 'memory/device_reserved (GiB)': '35.82', 'tokens/train_per_sec_per_gpu': '33.83', 'tokens/total': 5080320, 'tokens/trainable': 76682, 'epoch': '0.6562'}
	33%\|███▎ \| 168/512 [18:34<38:12, 6.66s/it] 33%\|███▎ \| 169/512 [18:40<37:56, 6.64s/it] {'loss': '0.003114', 'grad_norm': '0.1389', 'learning_rate': '8.217e-05', 'ppl': '1.003', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '35.82', 'tokens/train_per_sec_per_gpu': '34.87', 'tokens/total': 5110656, 'tokens/trainable': 77114, 'epoch': '0.6602'}
	33%\|███▎ \| 169/512 [18:40<37:56, 6.64s/it] 33%\|███▎ \| 170/512 [18:47<37:50, 6.64s/it] {'loss': '0.0309', 'grad_norm': '0.682', 'learning_rate': '8.194e-05', 'ppl': '1.031', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '36.02', 'tokens/train_per_sec_per_gpu': '34.61', 'tokens/total': 5141056, 'tokens/trainable': 77563, 'epoch': '0.6641'}
	33%\|███▎ \| 170/512 [18:47<37:50, 6.64s/it] 33%\|███▎ \| 171/512 [18:53<37:38, 6.62s/it] {'loss': '0.003041', 'grad_norm': '0.2058', 'learning_rate': '8.171e-05', 'ppl': '1.003', 'memory/max_active (GiB)': '33.79', 'memory/max_allocated (GiB)': '33.79', 'memory/device_reserved (GiB)': '36.02', 'tokens/train_per_sec_per_gpu': '33.7', 'tokens/total': 5171312, 'tokens/trainable': 78023, 'epoch': '0.668'}
	33%\|███▎ \| 171/512 [18:53<37:38, 6.62s/it] 34%\|███▎ \| 172/512 [19:00<37:34, 6.63s/it] {'loss': '0.004004', 'grad_norm': '0.2195', 'learning_rate': '8.147e-05', 'ppl': '1.004', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '36.02', 'tokens/train_per_sec_per_gpu': '30.23', 'tokens/total': 5201744, 'tokens/trainable': 78477, 'epoch': '0.6719'}
	34%\|███▎ \| 172/512 [19:00<37:34, 6.63s/it] 34%\|███▍ \| 173/512 [19:07<37:21, 6.61s/it] {'loss': '0.006001', 'grad_norm': '0.5019', 'learning_rate': '8.124e-05', 'ppl': '1.006', 'memory/max_active (GiB)': '33.91', 'memory/max_allocated (GiB)': '33.91', 'memory/device_reserved (GiB)': '36.02', 'tokens/train_per_sec_per_gpu': '33.51', 'tokens/total': 5232176, 'tokens/trainable': 78924, 'epoch': '0.6758'}
	34%\|███▍ \| 173/512 [19:07<37:21, 6.61s/it] 34%\|███▍ \| 174/512 [19:13<36:57, 6.56s/it] {'loss': '0.004308', 'grad_norm': '0.3675', 'learning_rate': '8.1e-05', 'ppl': '1.004', 'memory/max_active (GiB)': '33.62', 'memory/max_allocated (GiB)': '33.62', 'memory/device_reserved (GiB)': '36.02', 'tokens/train_per_sec_per_gpu': '30.59', 'tokens/total': 5261904, 'tokens/trainable': 79342, 'epoch': '0.6797'}
	34%\|███▍ \| 174/512 [19:13<36:57, 6.56s/it] 34%\|███▍ \| 175/512 [19:20<36:54, 6.57s/it] {'loss': '0.0007982', 'grad_norm': '0.4297', 'learning_rate': '8.076e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.8', 'memory/max_allocated (GiB)': '33.8', 'memory/device_reserved (GiB)': '36.02', 'tokens/train_per_sec_per_gpu': '33.05', 'tokens/total': 5292192, 'tokens/trainable': 79812, 'epoch': '0.6836'}
	34%\|███▍ \| 175/512 [19:20<36:54, 6.57s/it] 34%\|███▍ \| 176/512 [19:26<36:55, 6.59s/it] {'loss': '0.005437', 'grad_norm': '0.7104', 'learning_rate': '8.053e-05', 'ppl': '1.005', 'memory/max_active (GiB)': '33.93', 'memory/max_allocated (GiB)': '33.93', 'memory/device_reserved (GiB)': '36.02', 'tokens/train_per_sec_per_gpu': '36.73', 'tokens/total': 5322704, 'tokens/trainable': 80290, 'epoch': '0.6875'}
	34%\|███▍ \| 176/512 [19:26<36:55, 6.59s/it] 35%\|███▍ \| 177/512 [19:33<36:48, 6.59s/it] {'loss': '0.007359', 'grad_norm': '0.5655', 'learning_rate': '8.029e-05', 'ppl': '1.007', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '36.02', 'tokens/train_per_sec_per_gpu': '33.08', 'tokens/total': 5353040, 'tokens/trainable': 80737, 'epoch': '0.6914'}
	35%\|███▍ \| 177/512 [19:33<36:48, 6.59s/it] 35%\|███▍ \| 178/512 [19:40<36:45, 6.60s/it] {'loss': '0.0001785', 'grad_norm': '0.03567', 'learning_rate': '8.005e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.81', 'memory/max_allocated (GiB)': '33.81', 'memory/device_reserved (GiB)': '36.02', 'tokens/train_per_sec_per_gpu': '35.33', 'tokens/total': 5383248, 'tokens/trainable': 81205, 'epoch': '0.6953'}
	35%\|███▍ \| 178/512 [19:40<36:45, 6.60s/it] 35%\|███▍ \| 179/512 [19:46<36:31, 6.58s/it] {'loss': '0.0002031', 'grad_norm': '0.01853', 'learning_rate': '7.98e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '36.02', 'tokens/train_per_sec_per_gpu': '33.53', 'tokens/total': 5413248, 'tokens/trainable': 81652, 'epoch': '0.6992'}
	35%\|███▍ \| 179/512 [19:46<36:31, 6.58s/it] 35%\|███▌ \| 180/512 [19:53<36:28, 6.59s/it] {'loss': '0.0007448', 'grad_norm': '0.1078', 'learning_rate': '7.956e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.8', 'memory/max_allocated (GiB)': '33.8', 'memory/device_reserved (GiB)': '36.02', 'tokens/train_per_sec_per_gpu': '35.07', 'tokens/total': 5443424, 'tokens/trainable': 82125, 'epoch': '0.7031'}
	35%\|███▌ \| 180/512 [19:53<36:28, 6.59s/it] 35%\|███▌ \| 181/512 [19:59<36:28, 6.61s/it] {'loss': '0.0008925', 'grad_norm': '0.1558', 'learning_rate': '7.932e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '36.02', 'tokens/train_per_sec_per_gpu': '39.35', 'tokens/total': 5473680, 'tokens/trainable': 82642, 'epoch': '0.707'}
	35%\|███▌ \| 181/512 [19:59<36:28, 6.61s/it] 36%\|███▌ \| 182/512 [20:06<36:22, 6.61s/it] {'loss': '0.01775', 'grad_norm': '0.7525', 'learning_rate': '7.907e-05', 'ppl': '1.018', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '36.02', 'tokens/train_per_sec_per_gpu': '34.47', 'tokens/total': 5504032, 'tokens/trainable': 83123, 'epoch': '0.7109'}
	36%\|███▌ \| 182/512 [20:06<36:22, 6.61s/it] 36%\|███▌ \| 183/512 [20:13<36:16, 6.61s/it] {'loss': '0.01671', 'grad_norm': '0.8974', 'learning_rate': '7.883e-05', 'ppl': '1.017', 'memory/max_active (GiB)': '33.83', 'memory/max_allocated (GiB)': '33.83', 'memory/device_reserved (GiB)': '36.02', 'tokens/train_per_sec_per_gpu': '33.26', 'tokens/total': 5534224, 'tokens/trainable': 83570, 'epoch': '0.7148'}
	36%\|███▌ \| 183/512 [20:13<36:16, 6.61s/it] 36%\|███▌ \| 184/512 [20:19<36:08, 6.61s/it] {'loss': '0.001814', 'grad_norm': '0.3188', 'learning_rate': '7.858e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.89', 'memory/max_allocated (GiB)': '33.89', 'memory/device_reserved (GiB)': '36.02', 'tokens/train_per_sec_per_gpu': '36.04', 'tokens/total': 5564704, 'tokens/trainable': 84076, 'epoch': '0.7188'}
	36%\|███▌ \| 184/512 [20:19<36:08, 6.61s/it] 36%\|███▌ \| 185/512 [20:26<36:01, 6.61s/it] {'loss': '0.03057', 'grad_norm': '0.3768', 'learning_rate': '7.833e-05', 'ppl': '1.031', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '36.02', 'tokens/train_per_sec_per_gpu': '34.97', 'tokens/total': 5594912, 'tokens/trainable': 84551, 'epoch': '0.7227'}
	36%\|███▌ \| 185/512 [20:26<36:01, 6.61s/it] 36%\|███▋ \| 186/512 [20:32<35:57, 6.62s/it] {'loss': '0.0003345', 'grad_norm': '0.02533', 'learning_rate': '7.808e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '36.02', 'tokens/train_per_sec_per_gpu': '33.92', 'tokens/total': 5625280, 'tokens/trainable': 85012, 'epoch': '0.7266'}
	36%\|███▋ \| 186/512 [20:32<35:57, 6.62s/it] 37%\|███▋ \| 187/512 [20:39<35:49, 6.61s/it] {'loss': '0.009675', 'grad_norm': '0.3701', 'learning_rate': '7.783e-05', 'ppl': '1.01', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '36.02', 'tokens/train_per_sec_per_gpu': '32.23', 'tokens/total': 5655568, 'tokens/trainable': 85452, 'epoch': '0.7305'}
	37%\|███▋ \| 187/512 [20:39<35:49, 6.61s/it] 37%\|███▋ \| 188/512 [20:46<35:45, 6.62s/it] {'loss': '0.01034', 'grad_norm': '0.3262', 'learning_rate': '7.758e-05', 'ppl': '1.01', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '36.02', 'tokens/train_per_sec_per_gpu': '34.49', 'tokens/total': 5685824, 'tokens/trainable': 85942, 'epoch': '0.7344'}
	37%\|███▋ \| 188/512 [20:46<35:45, 6.62s/it] 37%\|███▋ \| 189/512 [20:52<35:44, 6.64s/it] {'loss': '0.009306', 'grad_norm': '0.6184', 'learning_rate': '7.733e-05', 'ppl': '1.009', 'memory/max_active (GiB)': '33.8', 'memory/max_allocated (GiB)': '33.8', 'memory/device_reserved (GiB)': '36.02', 'tokens/train_per_sec_per_gpu': '36.56', 'tokens/total': 5716192, 'tokens/trainable': 86457, 'epoch': '0.7383'}
	37%\|███▋ \| 189/512 [20:52<35:44, 6.64s/it] 37%\|███▋ \| 190/512 [20:59<35:34, 6.63s/it] {'loss': '0.0043', 'grad_norm': '0.3189', 'learning_rate': '7.708e-05', 'ppl': '1.004', 'memory/max_active (GiB)': '33.91', 'memory/max_allocated (GiB)': '33.91', 'memory/device_reserved (GiB)': '36.02', 'tokens/train_per_sec_per_gpu': '34.36', 'tokens/total': 5746784, 'tokens/trainable': 86928, 'epoch': '0.7422'}
	37%\|███▋ \| 190/512 [20:59<35:34, 6.63s/it] 37%\|███▋ \| 191/512 [21:06<35:36, 6.66s/it] {'loss': '0.006645', 'grad_norm': '0.3455', 'learning_rate': '7.683e-05', 'ppl': '1.007', 'memory/max_active (GiB)': '33.99', 'memory/max_allocated (GiB)': '33.99', 'memory/device_reserved (GiB)': '36.02', 'tokens/train_per_sec_per_gpu': '36.31', 'tokens/total': 5777520, 'tokens/trainable': 87384, 'epoch': '0.7461'}
	37%\|███▋ \| 191/512 [21:06<35:36, 6.66s/it] 38%\|███▊ \| 192/512 [21:12<35:23, 6.64s/it] {'loss': '0.006615', 'grad_norm': '0.2645', 'learning_rate': '7.657e-05', 'ppl': '1.007', 'memory/max_active (GiB)': '33.83', 'memory/max_allocated (GiB)': '33.83', 'memory/device_reserved (GiB)': '36.02', 'tokens/train_per_sec_per_gpu': '32.03', 'tokens/total': 5807952, 'tokens/trainable': 87814, 'epoch': '0.75'}
	38%\|███▊ \| 192/512 [21:12<35:23, 6.64s/it][2026-08-18 14:39:10,060] [INFO] [axolotl.core.trainers.base._save:828] [PID:12244] Saving model checkpoint to /workspace/wave/training/checkpoints/checkpoint-192
	38%\|███▊ \| 193/512 [21:20<37:34, 7.07s/it] {'loss': '0.01492', 'grad_norm': '0.3881', 'learning_rate': '7.632e-05', 'ppl': '1.015', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '36.02', 'tokens/train_per_sec_per_gpu': '35.64', 'tokens/total': 5838384, 'tokens/trainable': 88297, 'epoch': '0.7539'}
	38%\|███▊ \| 193/512 [21:20<37:34, 7.07s/it] 38%\|███▊ \| 194/512 [21:27<36:47, 6.94s/it] {'loss': '0.02648', 'grad_norm': '0.6184', 'learning_rate': '7.606e-05', 'ppl': '1.027', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '35.92', 'tokens/train_per_sec_per_gpu': '35.37', 'tokens/total': 5868704, 'tokens/trainable': 88768, 'epoch': '0.7578'}
	38%\|███▊ \| 194/512 [21:27<36:47, 6.94s/it] 38%\|███▊ \| 195/512 [21:33<35:15, 6.68s/it] {'loss': '0.003626', 'grad_norm': '0.1768', 'learning_rate': '7.58e-05', 'ppl': '1.004', 'memory/max_active (GiB)': '33.38', 'memory/max_allocated (GiB)': '33.38', 'memory/device_reserved (GiB)': '35.92', 'tokens/train_per_sec_per_gpu': '30.88', 'tokens/total': 5896928, 'tokens/trainable': 89201, 'epoch': '0.7617'}
	38%\|███▊ \| 195/512 [21:33<35:15, 6.68s/it] 38%\|███▊ \| 196/512 [21:40<35:03, 6.66s/it] {'loss': '0.00249', 'grad_norm': '0.09678', 'learning_rate': '7.555e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.92', 'memory/max_allocated (GiB)': '33.92', 'memory/device_reserved (GiB)': '35.92', 'tokens/train_per_sec_per_gpu': '37.84', 'tokens/total': 5927328, 'tokens/trainable': 89663, 'epoch': '0.7656'}
	38%\|███▊ \| 196/512 [21:40<35:03, 6.66s/it] 38%\|███▊ \| 197/512 [21:46<34:04, 6.49s/it] {'loss': '0.001231', 'grad_norm': '0.03317', 'learning_rate': '7.529e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.34', 'memory/max_allocated (GiB)': '33.34', 'memory/device_reserved (GiB)': '35.92', 'tokens/train_per_sec_per_gpu': '38.68', 'tokens/total': 5955648, 'tokens/trainable': 90111, 'epoch': '0.7695'}
	38%\|███▊ \| 197/512 [21:46<34:04, 6.49s/it] 39%\|███▊ \| 198/512 [21:52<34:10, 6.53s/it] {'loss': '0.01116', 'grad_norm': '0.361', 'learning_rate': '7.503e-05', 'ppl': '1.011', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '35.92', 'tokens/train_per_sec_per_gpu': '34.73', 'tokens/total': 5985904, 'tokens/trainable': 90572, 'epoch': '0.7734'}
	39%\|███▊ \| 198/512 [21:52<34:10, 6.53s/it] 39%\|███▉ \| 199/512 [21:59<34:24, 6.60s/it] {'loss': '0.001934', 'grad_norm': '0.0794', 'learning_rate': '7.477e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '34', 'memory/max_allocated (GiB)': '34', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '37.02', 'tokens/total': 6016656, 'tokens/trainable': 91051, 'epoch': '0.7773'}
	39%\|███▉ \| 199/512 [21:59<34:24, 6.60s/it] 39%\|███▉ \| 200/512 [22:06<34:19, 6.60s/it] {'loss': '0.008391', 'grad_norm': '0.2828', 'learning_rate': '7.451e-05', 'ppl': '1.008', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '38.74', 'tokens/total': 6046912, 'tokens/trainable': 91527, 'epoch': '0.7812'}
	39%\|███▉ \| 200/512 [22:06<34:19, 6.60s/it] 39%\|███▉ \| 201/512 [22:12<34:08, 6.59s/it] {'loss': '0.0006643', 'grad_norm': '0.02596', 'learning_rate': '7.424e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.69', 'memory/max_allocated (GiB)': '33.69', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '34.77', 'tokens/total': 6076912, 'tokens/trainable': 91973, 'epoch': '0.7852'}
	39%\|███▉ \| 201/512 [22:12<34:08, 6.59s/it] 39%\|███▉ \| 202/512 [22:19<34:01, 6.59s/it] {'loss': '0.01052', 'grad_norm': '0.2635', 'learning_rate': '7.398e-05', 'ppl': '1.011', 'memory/max_active (GiB)': '33.97', 'memory/max_allocated (GiB)': '33.97', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '37.23', 'tokens/total': 6107264, 'tokens/trainable': 92438, 'epoch': '0.7891'}
	39%\|███▉ \| 202/512 [22:19<34:01, 6.59s/it] 40%\|███▉ \| 203/512 [22:25<33:54, 6.58s/it] {'loss': '0.02642', 'grad_norm': '0.4627', 'learning_rate': '7.372e-05', 'ppl': '1.027', 'memory/max_active (GiB)': '33.77', 'memory/max_allocated (GiB)': '33.77', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '37.12', 'tokens/total': 6137344, 'tokens/trainable': 92886, 'epoch': '0.793'}
	40%\|███▉ \| 203/512 [22:25<33:54, 6.58s/it] 40%\|███▉ \| 204/512 [22:32<33:49, 6.59s/it] {'loss': '0.005499', 'grad_norm': '0.4922', 'learning_rate': '7.345e-05', 'ppl': '1.006', 'memory/max_active (GiB)': '33.89', 'memory/max_allocated (GiB)': '33.89', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '34.54', 'tokens/total': 6167888, 'tokens/trainable': 93359, 'epoch': '0.7969'}
	40%\|███▉ \| 204/512 [22:32<33:49, 6.59s/it] 40%\|████ \| 205/512 [22:39<33:37, 6.57s/it] {'loss': '0.002654', 'grad_norm': '0.143', 'learning_rate': '7.319e-05', 'ppl': '1.003', 'memory/max_active (GiB)': '33.76', 'memory/max_allocated (GiB)': '33.76', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '33.97', 'tokens/total': 6198080, 'tokens/trainable': 93783, 'epoch': '0.8008'}
	40%\|████ \| 205/512 [22:39<33:37, 6.57s/it] 40%\|████ \| 206/512 [22:45<33:40, 6.60s/it] {'loss': '0.005219', 'grad_norm': '0.1867', 'learning_rate': '7.292e-05', 'ppl': '1.005', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '37.94', 'tokens/total': 6228480, 'tokens/trainable': 94262, 'epoch': '0.8047'}
	40%\|████ \| 206/512 [22:45<33:40, 6.60s/it] 40%\|████ \| 207/512 [22:52<33:34, 6.60s/it] {'loss': '0.0106', 'grad_norm': '0.3964', 'learning_rate': '7.266e-05', 'ppl': '1.011', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '36.03', 'tokens/total': 6258640, 'tokens/trainable': 94753, 'epoch': '0.8086'}
	40%\|████ \| 207/512 [22:52<33:34, 6.60s/it] 41%\|████ \| 208/512 [22:59<33:27, 6.60s/it] {'loss': '0.00088', 'grad_norm': '0.0414', 'learning_rate': '7.239e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.95', 'memory/max_allocated (GiB)': '33.95', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '34.21', 'tokens/total': 6288928, 'tokens/trainable': 95238, 'epoch': '0.8125'}
	41%\|████ \| 208/512 [22:59<33:27, 6.60s/it] 41%\|████ \| 209/512 [23:05<33:15, 6.59s/it] {'loss': '0.001892', 'grad_norm': '0.09655', 'learning_rate': '7.212e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.95', 'memory/max_allocated (GiB)': '33.95', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '29.65', 'tokens/total': 6319568, 'tokens/trainable': 95639, 'epoch': '0.8164'}
	41%\|████ \| 209/512 [23:05<33:15, 6.59s/it] 41%\|████ \| 210/512 [23:12<33:10, 6.59s/it] {'loss': '0.0052', 'grad_norm': '0.2853', 'learning_rate': '7.185e-05', 'ppl': '1.005', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '35.91', 'tokens/total': 6349824, 'tokens/trainable': 96096, 'epoch': '0.8203'}
	41%\|████ \| 210/512 [23:12<33:10, 6.59s/it] 41%\|████ \| 211/512 [23:18<33:00, 6.58s/it] {'loss': '0.0128', 'grad_norm': '1.931', 'learning_rate': '7.158e-05', 'ppl': '1.013', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '34.64', 'tokens/total': 6379920, 'tokens/trainable': 96517, 'epoch': '0.8242'}
	41%\|████ \| 211/512 [23:18<33:00, 6.58s/it] 41%\|████▏ \| 212/512 [23:25<32:55, 6.59s/it] {'loss': '0.0241', 'grad_norm': '0.7944', 'learning_rate': '7.131e-05', 'ppl': '1.024', 'memory/max_active (GiB)': '33.89', 'memory/max_allocated (GiB)': '33.89', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '33.19', 'tokens/total': 6410432, 'tokens/trainable': 96942, 'epoch': '0.8281'}
	41%\|████▏ \| 212/512 [23:25<32:55, 6.59s/it] 42%\|████▏ \| 213/512 [23:31<32:50, 6.59s/it] {'loss': '0.009192', 'grad_norm': '0.589', 'learning_rate': '7.104e-05', 'ppl': '1.009', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '33.19', 'tokens/total': 6440864, 'tokens/trainable': 97380, 'epoch': '0.832'}
	42%\|████▏ \| 213/512 [23:31<32:50, 6.59s/it] 42%\|████▏ \| 214/512 [23:38<32:48, 6.60s/it] {'loss': '0.002657', 'grad_norm': '0.1773', 'learning_rate': '7.077e-05', 'ppl': '1.003', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '36.92', 'tokens/total': 6471136, 'tokens/trainable': 97869, 'epoch': '0.8359'}
	42%\|████▏ \| 214/512 [23:38<32:48, 6.60s/it] 42%\|████▏ \| 215/512 [23:45<32:38, 6.59s/it] {'loss': '0.003227', 'grad_norm': '0.1191', 'learning_rate': '7.05e-05', 'ppl': '1.003', 'memory/max_active (GiB)': '33.89', 'memory/max_allocated (GiB)': '33.89', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '37.9', 'tokens/total': 6501328, 'tokens/trainable': 98320, 'epoch': '0.8398'}
	42%\|████▏ \| 215/512 [23:45<32:38, 6.59s/it] 42%\|████▏ \| 216/512 [23:51<32:30, 6.59s/it] {'loss': '0.002932', 'grad_norm': '0.149', 'learning_rate': '7.022e-05', 'ppl': '1.003', 'memory/max_active (GiB)': '33.96', 'memory/max_allocated (GiB)': '33.96', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '36.97', 'tokens/total': 6531680, 'tokens/trainable': 98793, 'epoch': '0.8438'}
	42%\|████▏ \| 216/512 [23:51<32:30, 6.59s/it] 42%\|████▏ \| 217/512 [23:58<32:26, 6.60s/it] {'loss': '0.002244', 'grad_norm': '0.102', 'learning_rate': '6.995e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.8', 'memory/max_allocated (GiB)': '33.8', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '37.33', 'tokens/total': 6562032, 'tokens/trainable': 99278, 'epoch': '0.8477'}
	42%\|████▏ \| 217/512 [23:58<32:26, 6.60s/it] 43%\|████▎ \| 218/512 [24:04<32:15, 6.58s/it] {'loss': '0.001671', 'grad_norm': '0.07195', 'learning_rate': '6.968e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.92', 'memory/max_allocated (GiB)': '33.92', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '37.41', 'tokens/total': 6592304, 'tokens/trainable': 99751, 'epoch': '0.8516'}
	43%\|████▎ \| 218/512 [24:04<32:15, 6.58s/it] 43%\|████▎ \| 219/512 [24:11<32:09, 6.58s/it] {'loss': '0.001428', 'grad_norm': '0.07926', 'learning_rate': '6.94e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.83', 'memory/max_allocated (GiB)': '33.83', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '32.21', 'tokens/total': 6622736, 'tokens/trainable': 100184, 'epoch': '0.8555'}
	43%\|████▎ \| 219/512 [24:11<32:09, 6.58s/it] 43%\|████▎ \| 220/512 [24:18<32:04, 6.59s/it] {'loss': '0.02854', 'grad_norm': '4.548', 'learning_rate': '6.913e-05', 'ppl': '1.029', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '38.12', 'tokens/total': 6653168, 'tokens/trainable': 100656, 'epoch': '0.8594'}
	43%\|████▎ \| 220/512 [24:18<32:04, 6.59s/it] 43%\|████▎ \| 221/512 [24:24<31:59, 6.59s/it] {'loss': '0.002416', 'grad_norm': '0.2031', 'learning_rate': '6.885e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '32.12', 'tokens/total': 6683328, 'tokens/trainable': 101107, 'epoch': '0.8633'}
	43%\|████▎ \| 221/512 [24:24<31:59, 6.59s/it] 43%\|████▎ \| 222/512 [24:31<31:51, 6.59s/it] {'loss': '0.001577', 'grad_norm': '0.1236', 'learning_rate': '6.857e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '38.57', 'tokens/total': 6713840, 'tokens/trainable': 101574, 'epoch': '0.8672'}
	43%\|████▎ \| 222/512 [24:31<31:51, 6.59s/it] 44%\|████▎ \| 223/512 [24:37<31:42, 6.58s/it] {'loss': '0.01232', 'grad_norm': '0.6421', 'learning_rate': '6.83e-05', 'ppl': '1.012', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '31.99', 'tokens/total': 6744224, 'tokens/trainable': 102015, 'epoch': '0.8711'}
	44%\|████▎ \| 223/512 [24:37<31:42, 6.58s/it] 44%\|████▍ \| 224/512 [24:44<31:38, 6.59s/it] {'loss': '0.00247', 'grad_norm': '0.3597', 'learning_rate': '6.802e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.81', 'memory/max_allocated (GiB)': '33.81', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '36.04', 'tokens/total': 6774432, 'tokens/trainable': 102463, 'epoch': '0.875'}
	44%\|████▍ \| 224/512 [24:44<31:38, 6.59s/it][2026-08-18 14:42:41,677] [INFO] [axolotl.core.trainers.base._save:828] [PID:12244] Saving model checkpoint to /workspace/wave/training/checkpoints/checkpoint-224
	44%\|████▍ \| 225/512 [24:52<33:34, 7.02s/it] {'loss': '0.001325', 'grad_norm': '0.1022', 'learning_rate': '6.774e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '35.94', 'tokens/train_per_sec_per_gpu': '29.58', 'tokens/total': 6804688, 'tokens/trainable': 102880, 'epoch': '0.8789'}
	44%\|████▍ \| 225/512 [24:52<33:34, 7.02s/it] 44%\|████▍ \| 226/512 [24:59<33:04, 6.94s/it] {'loss': '0.016', 'grad_norm': '0.7656', 'learning_rate': '6.746e-05', 'ppl': '1.016', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '35.7', 'tokens/train_per_sec_per_gpu': '37.94', 'tokens/total': 6835376, 'tokens/trainable': 103387, 'epoch': '0.8828'}
	44%\|████▍ \| 226/512 [24:59<33:04, 6.94s/it] 44%\|████▍ \| 227/512 [25:05<32:24, 6.82s/it] {'loss': '0.004662', 'grad_norm': '0.298', 'learning_rate': '6.718e-05', 'ppl': '1.005', 'memory/max_active (GiB)': '33.76', 'memory/max_allocated (GiB)': '33.76', 'memory/device_reserved (GiB)': '35.7', 'tokens/train_per_sec_per_gpu': '40.16', 'tokens/total': 6865392, 'tokens/trainable': 103880, 'epoch': '0.8867'}
	44%\|████▍ \| 227/512 [25:05<32:24, 6.82s/it] 45%\|████▍ \| 228/512 [25:12<31:58, 6.76s/it] {'loss': '0.01068', 'grad_norm': '0.4353', 'learning_rate': '6.69e-05', 'ppl': '1.011', 'memory/max_active (GiB)': '33.91', 'memory/max_allocated (GiB)': '33.91', 'memory/device_reserved (GiB)': '35.77', 'tokens/train_per_sec_per_gpu': '36.82', 'tokens/total': 6895776, 'tokens/trainable': 104343, 'epoch': '0.8906'}
	45%\|████▍ \| 228/512 [25:12<31:58, 6.76s/it] 45%\|████▍ \| 229/512 [25:18<31:35, 6.70s/it] {'loss': '0.00132', 'grad_norm': '0.05993', 'learning_rate': '6.662e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.69', 'memory/max_allocated (GiB)': '33.69', 'memory/device_reserved (GiB)': '35.77', 'tokens/train_per_sec_per_gpu': '33.97', 'tokens/total': 6925776, 'tokens/trainable': 104808, 'epoch': '0.8945'}
	45%\|████▍ \| 229/512 [25:18<31:35, 6.70s/it] 45%\|████▍ \| 230/512 [25:25<31:23, 6.68s/it] {'loss': '0.004353', 'grad_norm': '0.214', 'learning_rate': '6.634e-05', 'ppl': '1.004', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '35.77', 'tokens/train_per_sec_per_gpu': '37.4', 'tokens/total': 6956048, 'tokens/trainable': 105276, 'epoch': '0.8984'}
	45%\|████▍ \| 230/512 [25:25<31:23, 6.68s/it] 45%\|████▌ \| 231/512 [25:32<31:08, 6.65s/it] {'loss': '0.007145', 'grad_norm': '0.2365', 'learning_rate': '6.606e-05', 'ppl': '1.007', 'memory/max_active (GiB)': '33.79', 'memory/max_allocated (GiB)': '33.79', 'memory/device_reserved (GiB)': '35.77', 'tokens/train_per_sec_per_gpu': '34.34', 'tokens/total': 6986384, 'tokens/trainable': 105740, 'epoch': '0.9023'}
	45%\|████▌ \| 231/512 [25:32<31:08, 6.65s/it] 45%\|████▌ \| 232/512 [25:38<30:57, 6.63s/it] {'loss': '0.000307', 'grad_norm': '0.01727', 'learning_rate': '6.578e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.89', 'memory/max_allocated (GiB)': '33.89', 'memory/device_reserved (GiB)': '35.77', 'tokens/train_per_sec_per_gpu': '33.52', 'tokens/total': 7016784, 'tokens/trainable': 106216, 'epoch': '0.9062'}
	45%\|████▌ \| 232/512 [25:38<30:57, 6.63s/it] 46%\|████▌ \| 233/512 [25:45<30:51, 6.63s/it] {'loss': '0.0005204', 'grad_norm': '0.02866', 'learning_rate': '6.55e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.97', 'memory/max_allocated (GiB)': '33.97', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '38.58', 'tokens/total': 7047184, 'tokens/trainable': 106721, 'epoch': '0.9102'}
	46%\|████▌ \| 233/512 [25:45<30:51, 6.63s/it] 46%\|████▌ \| 234/512 [25:51<30:37, 6.61s/it] {'loss': '0.001806', 'grad_norm': '0.1024', 'learning_rate': '6.522e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '34.36', 'tokens/total': 7077168, 'tokens/trainable': 107159, 'epoch': '0.9141'}
	46%\|████▌ \| 234/512 [25:51<30:37, 6.61s/it] 46%\|████▌ \| 235/512 [25:58<30:29, 6.60s/it] {'loss': '0.01434', 'grad_norm': '0.1958', 'learning_rate': '6.493e-05', 'ppl': '1.014', 'memory/max_active (GiB)': '33.95', 'memory/max_allocated (GiB)': '33.95', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '33.21', 'tokens/total': 7107616, 'tokens/trainable': 107599, 'epoch': '0.918'}
	46%\|████▌ \| 235/512 [25:58<30:29, 6.60s/it] 46%\|████▌ \| 236/512 [26:05<30:21, 6.60s/it] {'loss': '0.007726', 'grad_norm': '0.2197', 'learning_rate': '6.465e-05', 'ppl': '1.008', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '34.61', 'tokens/total': 7137728, 'tokens/trainable': 108063, 'epoch': '0.9219'}
	46%\|████▌ \| 236/512 [26:05<30:21, 6.60s/it] 46%\|████▋ \| 237/512 [26:11<30:12, 6.59s/it] {'loss': '0.0009402', 'grad_norm': '0.04341', 'learning_rate': '6.437e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.77', 'memory/max_allocated (GiB)': '33.77', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '33.83', 'tokens/total': 7167904, 'tokens/trainable': 108491, 'epoch': '0.9258'}
	46%\|████▋ \| 237/512 [26:11<30:12, 6.59s/it] 46%\|████▋ \| 238/512 [26:18<30:03, 6.58s/it] {'loss': '0.006565', 'grad_norm': '0.3101', 'learning_rate': '6.408e-05', 'ppl': '1.007', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '29.57', 'tokens/total': 7198288, 'tokens/trainable': 108919, 'epoch': '0.9297'}
	46%\|████▋ \| 238/512 [26:18<30:03, 6.58s/it] 47%\|████▋ \| 239/512 [26:24<30:00, 6.60s/it] {'loss': '0.03366', 'grad_norm': '0.6903', 'learning_rate': '6.38e-05', 'ppl': '1.034', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '33.05', 'tokens/total': 7228624, 'tokens/trainable': 109360, 'epoch': '0.9336'}
	47%\|████▋ \| 239/512 [26:24<30:00, 6.60s/it] 47%\|████▋ \| 240/512 [26:31<29:55, 6.60s/it] {'loss': '0.001445', 'grad_norm': '0.07222', 'learning_rate': '6.351e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.93', 'memory/max_allocated (GiB)': '33.93', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '32.69', 'tokens/total': 7259056, 'tokens/trainable': 109823, 'epoch': '0.9375'}
	47%\|████▋ \| 240/512 [26:31<29:55, 6.60s/it] 47%\|████▋ \| 241/512 [26:37<29:44, 6.58s/it] {'loss': '0.00448', 'grad_norm': '0.1981', 'learning_rate': '6.323e-05', 'ppl': '1.004', 'memory/max_active (GiB)': '33.95', 'memory/max_allocated (GiB)': '33.95', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '38.97', 'tokens/total': 7289328, 'tokens/trainable': 110281, 'epoch': '0.9414'}
	47%\|████▋ \| 241/512 [26:37<29:44, 6.58s/it] 47%\|████▋ \| 242/512 [26:44<29:39, 6.59s/it] {'loss': '0.003018', 'grad_norm': '0.1568', 'learning_rate': '6.294e-05', 'ppl': '1.003', 'memory/max_active (GiB)': '33.8', 'memory/max_allocated (GiB)': '33.8', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '28.46', 'tokens/total': 7319632, 'tokens/trainable': 110694, 'epoch': '0.9453'}
	47%\|████▋ \| 242/512 [26:44<29:39, 6.59s/it] 47%\|████▋ \| 243/512 [26:51<29:34, 6.60s/it] {'loss': '0.0005814', 'grad_norm': '0.02434', 'learning_rate': '6.266e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.83', 'memory/max_allocated (GiB)': '33.83', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '39.78', 'tokens/total': 7349936, 'tokens/trainable': 111188, 'epoch': '0.9492'}
	47%\|████▋ \| 243/512 [26:51<29:34, 6.60s/it] 48%\|████▊ \| 244/512 [26:57<29:31, 6.61s/it] {'loss': '0.01199', 'grad_norm': '0.6709', 'learning_rate': '6.237e-05', 'ppl': '1.012', 'memory/max_active (GiB)': '33.93', 'memory/max_allocated (GiB)': '33.93', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '33.59', 'tokens/total': 7380576, 'tokens/trainable': 111645, 'epoch': '0.9531'}
	48%\|████▊ \| 244/512 [26:57<29:31, 6.61s/it] 48%\|████▊ \| 245/512 [27:04<29:24, 6.61s/it] {'loss': '0.001512', 'grad_norm': '0.05413', 'learning_rate': '6.208e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.95', 'memory/max_allocated (GiB)': '33.95', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '34.68', 'tokens/total': 7411200, 'tokens/trainable': 112128, 'epoch': '0.957'}
	48%\|████▊ \| 245/512 [27:04<29:24, 6.61s/it] 48%\|████▊ \| 246/512 [27:11<29:17, 6.61s/it] {'loss': '0.002352', 'grad_norm': '0.2185', 'learning_rate': '6.18e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.96', 'memory/max_allocated (GiB)': '33.96', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '33.6', 'tokens/total': 7441776, 'tokens/trainable': 112602, 'epoch': '0.9609'}
	48%\|████▊ \| 246/512 [27:11<29:17, 6.61s/it] 48%\|████▊ \| 247/512 [27:17<28:34, 6.47s/it] {'loss': '0.001361', 'grad_norm': '0.05781', 'learning_rate': '6.151e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '38.27', 'tokens/total': 7470096, 'tokens/trainable': 113056, 'epoch': '0.9648'}
	48%\|████▊ \| 247/512 [27:17<28:34, 6.47s/it] 48%\|████▊ \| 248/512 [27:23<28:42, 6.53s/it] {'loss': '0.005576', 'grad_norm': '0.3972', 'learning_rate': '6.122e-05', 'ppl': '1.006', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '35.76', 'tokens/total': 7500688, 'tokens/trainable': 113497, 'epoch': '0.9688'}
	48%\|████▊ \| 248/512 [27:23<28:42, 6.53s/it] 49%\|████▊ \| 249/512 [27:30<28:39, 6.54s/it] {'loss': '0.006032', 'grad_norm': '0.2466', 'learning_rate': '6.093e-05', 'ppl': '1.006', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '33.19', 'tokens/total': 7530832, 'tokens/trainable': 113941, 'epoch': '0.9727'}
	49%\|████▊ \| 249/512 [27:30<28:39, 6.54s/it] 49%\|████▉ \| 250/512 [27:36<27:58, 6.41s/it] {'loss': '0.001342', 'grad_norm': '0.04302', 'learning_rate': '6.065e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.81', 'memory/max_allocated (GiB)': '33.81', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '34.58', 'tokens/total': 7559136, 'tokens/trainable': 114373, 'epoch': '0.9766'}
	49%\|████▉ \| 250/512 [27:36<27:58, 6.41s/it] 49%\|████▉ \| 251/512 [27:43<28:07, 6.47s/it] {'loss': '0.00874', 'grad_norm': '1.104', 'learning_rate': '6.036e-05', 'ppl': '1.009', 'memory/max_active (GiB)': '33.95', 'memory/max_allocated (GiB)': '33.95', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '36.97', 'tokens/total': 7589648, 'tokens/trainable': 114863, 'epoch': '0.9805'}
	49%\|████▉ \| 251/512 [27:43<28:07, 6.47s/it] 49%\|████▉ \| 252/512 [27:49<28:11, 6.51s/it] {'loss': '0.0008687', 'grad_norm': '0.05026', 'learning_rate': '6.007e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.83', 'memory/max_allocated (GiB)': '33.83', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '35.74', 'tokens/total': 7619904, 'tokens/trainable': 115311, 'epoch': '0.9844'}
	49%\|████▉ \| 252/512 [27:49<28:11, 6.51s/it] 49%\|████▉ \| 253/512 [27:56<28:15, 6.55s/it] {'loss': '0.001406', 'grad_norm': '0.06898', 'learning_rate': '5.978e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '32.83', 'tokens/total': 7650480, 'tokens/trainable': 115732, 'epoch': '0.9883'}
	49%\|████▉ \| 253/512 [27:56<28:15, 6.55s/it] 50%\|████▉ \| 254/512 [28:02<28:12, 6.56s/it] {'loss': '0.009384', 'grad_norm': '0.5817', 'learning_rate': '5.949e-05', 'ppl': '1.009', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '29.88', 'tokens/total': 7680912, 'tokens/trainable': 116164, 'epoch': '0.9922'}
	50%\|████▉ \| 254/512 [28:02<28:12, 6.56s/it] 50%\|████▉ \| 255/512 [28:09<28:07, 6.57s/it] {'loss': '0.0004331', 'grad_norm': '0.01903', 'learning_rate': '5.92e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.89', 'memory/max_allocated (GiB)': '33.89', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '37.43', 'tokens/total': 7711344, 'tokens/trainable': 116614, 'epoch': '0.9961'}
	50%\|████▉ \| 255/512 [28:09<28:07, 6.57s/it] 50%\|█████ \| 256/512 [28:16<28:14, 6.62s/it] {'loss': '0.0004527', 'grad_norm': '0.05099', 'learning_rate': '5.891e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.98', 'memory/max_allocated (GiB)': '33.98', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '32.59', 'tokens/total': 7741936, 'tokens/trainable': 117062, 'epoch': '1'}
	50%\|█████ \| 256/512 [28:16<28:14, 6.62s/it][2026-08-18 14:46:13,557] [INFO] [axolotl.core.trainers.base._save:828] [PID:12244] Saving model checkpoint to /workspace/wave/training/checkpoints/checkpoint-256
	50%\|█████ \| 257/512 [28:24<30:28, 7.17s/it] {'loss': '0.001033', 'grad_norm': '0.09221', 'learning_rate': '5.862e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '32.85', 'tokens/total': 7772272, 'tokens/trainable': 117518, 'epoch': '1.004'}
	50%\|█████ \| 257/512 [28:24<30:28, 7.17s/it] 50%\|█████ \| 258/512 [28:31<29:37, 7.00s/it] {'loss': '0.0006238', 'grad_norm': '0.03798', 'learning_rate': '5.834e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '30.99', 'tokens/total': 7802560, 'tokens/trainable': 117948, 'epoch': '1.008'}
	50%\|█████ \| 258/512 [28:31<29:37, 7.00s/it] 51%\|█████ \| 259/512 [28:37<28:58, 6.87s/it] {'loss': '0.002222', 'grad_norm': '0.1512', 'learning_rate': '5.805e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '32.2', 'tokens/total': 7832896, 'tokens/trainable': 118388, 'epoch': '1.012'}
	51%\|█████ \| 259/512 [28:37<28:58, 6.87s/it] 51%\|█████ \| 260/512 [28:44<28:30, 6.79s/it] {'loss': '0.0004677', 'grad_norm': '0.0371', 'learning_rate': '5.776e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '32.01', 'tokens/total': 7863328, 'tokens/trainable': 118831, 'epoch': '1.016'}
	51%\|█████ \| 260/512 [28:44<28:30, 6.79s/it] 51%\|█████ \| 261/512 [28:51<28:07, 6.72s/it] {'loss': '0.001122', 'grad_norm': '0.05126', 'learning_rate': '5.747e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '34.44', 'tokens/total': 7893792, 'tokens/trainable': 119278, 'epoch': '1.02'}
	51%\|█████ \| 261/512 [28:51<28:07, 6.72s/it] 51%\|█████ \| 262/512 [28:57<27:54, 6.70s/it] {'loss': '0.001884', 'grad_norm': '0.2065', 'learning_rate': '5.718e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '35.36', 'tokens/total': 7924272, 'tokens/trainable': 119787, 'epoch': '1.023'}
	51%\|█████ \| 262/512 [28:57<27:54, 6.70s/it] 51%\|█████▏ \| 263/512 [29:04<27:42, 6.68s/it] {'loss': '0.002311', 'grad_norm': '0.1463', 'learning_rate': '5.689e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.8', 'memory/max_allocated (GiB)': '33.8', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '34.44', 'tokens/total': 7954624, 'tokens/trainable': 120276, 'epoch': '1.027'}
	51%\|█████▏ \| 263/512 [29:04<27:42, 6.68s/it] 52%\|█████▏ \| 264/512 [29:10<27:32, 6.66s/it] {'loss': '0.0002887', 'grad_norm': '0.04577', 'learning_rate': '5.66e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '37.7', 'tokens/total': 7985184, 'tokens/trainable': 120753, 'epoch': '1.031'}
	52%\|█████▏ \| 264/512 [29:10<27:32, 6.66s/it] 52%\|█████▏ \| 265/512 [29:17<27:18, 6.63s/it] {'loss': '0.003725', 'grad_norm': '0.324', 'learning_rate': '5.631e-05', 'ppl': '1.004', 'memory/max_active (GiB)': '33.96', 'memory/max_allocated (GiB)': '33.96', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '40.07', 'tokens/total': 8015520, 'tokens/trainable': 121219, 'epoch': '1.035'}
	52%\|█████▏ \| 265/512 [29:17<27:18, 6.63s/it] 52%\|█████▏ \| 266/512 [29:24<27:13, 6.64s/it] {'loss': '0.01656', 'grad_norm': '0.6029', 'learning_rate': '5.602e-05', 'ppl': '1.017', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '33.08', 'tokens/total': 8046064, 'tokens/trainable': 121694, 'epoch': '1.039'}
	52%\|█████▏ \| 266/512 [29:24<27:13, 6.64s/it] 52%\|█████▏ \| 267/512 [29:30<27:05, 6.63s/it] {'loss': '0.0009177', 'grad_norm': '0.08974', 'learning_rate': '5.573e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.81', 'memory/max_allocated (GiB)': '33.81', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '40.77', 'tokens/total': 8076352, 'tokens/trainable': 122198, 'epoch': '1.043'}
	52%\|█████▏ \| 267/512 [29:30<27:05, 6.63s/it] 52%\|█████▏ \| 268/512 [29:37<26:55, 6.62s/it] {'loss': '8.959e-05', 'grad_norm': '0.007786', 'learning_rate': '5.544e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.78', 'memory/max_allocated (GiB)': '33.78', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '35.98', 'tokens/total': 8106608, 'tokens/trainable': 122660, 'epoch': '1.047'}
	52%\|█████▏ \| 268/512 [29:37<26:55, 6.62s/it] 53%\|█████▎ \| 269/512 [29:44<26:45, 6.61s/it] {'loss': '0.001509', 'grad_norm': '0.3485', 'learning_rate': '5.515e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.71', 'memory/max_allocated (GiB)': '33.71', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '33.61', 'tokens/total': 8136768, 'tokens/trainable': 123120, 'epoch': '1.051'}
	53%\|█████▎ \| 269/512 [29:44<26:45, 6.61s/it] 53%\|█████▎ \| 270/512 [29:50<26:41, 6.62s/it] {'loss': '8.24e-05', 'grad_norm': '0.005141', 'learning_rate': '5.485e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.97', 'memory/max_allocated (GiB)': '33.97', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '36.79', 'tokens/total': 8167248, 'tokens/trainable': 123610, 'epoch': '1.055'}
	53%\|█████▎ \| 270/512 [29:50<26:41, 6.62s/it] 53%\|█████▎ \| 271/512 [29:57<26:32, 6.61s/it] {'loss': '8.649e-05', 'grad_norm': '0.004375', 'learning_rate': '5.456e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '38.27', 'tokens/total': 8197360, 'tokens/trainable': 124092, 'epoch': '1.059'}
	53%\|█████▎ \| 271/512 [29:57<26:32, 6.61s/it] 53%\|█████▎ \| 272/512 [30:03<26:27, 6.62s/it] {'loss': '0.004285', 'grad_norm': '0.4465', 'learning_rate': '5.427e-05', 'ppl': '1.004', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '35.26', 'tokens/total': 8227648, 'tokens/trainable': 124562, 'epoch': '1.062'}
	53%\|█████▎ \| 272/512 [30:03<26:27, 6.62s/it] 53%\|█████▎ \| 273/512 [30:10<26:18, 6.60s/it] {'loss': '0.001752', 'grad_norm': '0.1529', 'learning_rate': '5.398e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.71', 'memory/max_allocated (GiB)': '33.71', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '34.55', 'tokens/total': 8257696, 'tokens/trainable': 124992, 'epoch': '1.066'}
	53%\|█████▎ \| 273/512 [30:10<26:18, 6.60s/it] 54%\|█████▎ \| 274/512 [30:17<26:10, 6.60s/it] {'loss': '0.003796', 'grad_norm': '0.2659', 'learning_rate': '5.369e-05', 'ppl': '1.004', 'memory/max_active (GiB)': '33.95', 'memory/max_allocated (GiB)': '33.95', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '35.11', 'tokens/total': 8288272, 'tokens/trainable': 125453, 'epoch': '1.07'}
	54%\|█████▎ \| 274/512 [30:17<26:10, 6.60s/it] 54%\|█████▎ \| 275/512 [30:23<26:05, 6.60s/it] {'loss': '0.0005296', 'grad_norm': '0.05793', 'learning_rate': '5.34e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.98', 'memory/max_allocated (GiB)': '33.98', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '32.01', 'tokens/total': 8318912, 'tokens/trainable': 125904, 'epoch': '1.074'}
	54%\|█████▎ \| 275/512 [30:23<26:05, 6.60s/it] 54%\|█████▍ \| 276/512 [30:30<25:59, 6.61s/it] {'loss': '0.001218', 'grad_norm': '0.1219', 'learning_rate': '5.311e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '33.38', 'tokens/total': 8349280, 'tokens/trainable': 126378, 'epoch': '1.078'}
	54%\|█████▍ \| 276/512 [30:30<25:59, 6.61s/it] 54%\|█████▍ \| 277/512 [30:36<25:56, 6.63s/it] {'loss': '0.000552', 'grad_norm': '0.06378', 'learning_rate': '5.282e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '34.54', 'tokens/total': 8379744, 'tokens/trainable': 126843, 'epoch': '1.082'}
	54%\|█████▍ \| 277/512 [30:36<25:56, 6.63s/it] 54%\|█████▍ \| 278/512 [30:43<25:48, 6.62s/it] {'loss': '0.01097', 'grad_norm': '0.533', 'learning_rate': '5.253e-05', 'ppl': '1.011', 'memory/max_active (GiB)': '33.91', 'memory/max_allocated (GiB)': '33.91', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '38.83', 'tokens/total': 8409968, 'tokens/trainable': 127346, 'epoch': '1.086'}
	54%\|█████▍ \| 278/512 [30:43<25:48, 6.62s/it] 54%\|█████▍ \| 279/512 [30:50<25:39, 6.61s/it] {'loss': '0.01795', 'grad_norm': '0.2864', 'learning_rate': '5.224e-05', 'ppl': '1.018', 'memory/max_active (GiB)': '33.77', 'memory/max_allocated (GiB)': '33.77', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '33.57', 'tokens/total': 8440112, 'tokens/trainable': 127804, 'epoch': '1.09'}
	54%\|█████▍ \| 279/512 [30:50<25:39, 6.61s/it] 55%\|█████▍ \| 280/512 [30:56<25:31, 6.60s/it] {'loss': '5.531e-05', 'grad_norm': '0.003402', 'learning_rate': '5.195e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '33.36', 'tokens/total': 8470480, 'tokens/trainable': 128253, 'epoch': '1.094'}
	55%\|█████▍ \| 280/512 [30:56<25:31, 6.60s/it] 55%\|█████▍ \| 281/512 [31:03<25:27, 6.61s/it] {'loss': '0.00493', 'grad_norm': '0.2083', 'learning_rate': '5.166e-05', 'ppl': '1.005', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '36.8', 'tokens/total': 8500752, 'tokens/trainable': 128735, 'epoch': '1.098'}
	55%\|█████▍ \| 281/512 [31:03<25:27, 6.61s/it] 55%\|█████▌ \| 282/512 [31:09<25:21, 6.61s/it] {'loss': '0.0001742', 'grad_norm': '0.01068', 'learning_rate': '5.138e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '35.65', 'tokens/total': 8530944, 'tokens/trainable': 129209, 'epoch': '1.102'}
	55%\|█████▌ \| 282/512 [31:09<25:21, 6.61s/it] 55%\|█████▌ \| 283/512 [31:16<25:15, 6.62s/it] {'loss': '0.001448', 'grad_norm': '0.106', 'learning_rate': '5.109e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '35.91', 'tokens/train_per_sec_per_gpu': '33.5', 'tokens/total': 8561312, 'tokens/trainable': 129659, 'epoch': '1.105'}
	55%\|█████▌ \| 283/512 [31:16<25:15, 6.62s/it] 55%\|█████▌ \| 284/512 [31:23<25:10, 6.63s/it] {'loss': '0.0002165', 'grad_norm': '0.0106', 'learning_rate': '5.08e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '35.91', 'tokens/train_per_sec_per_gpu': '33.14', 'tokens/total': 8591712, 'tokens/trainable': 130106, 'epoch': '1.109'}
	55%\|█████▌ \| 284/512 [31:23<25:10, 6.63s/it] 56%\|█████▌ \| 285/512 [31:29<25:02, 6.62s/it] {'loss': '0.005551', 'grad_norm': '0.3154', 'learning_rate': '5.051e-05', 'ppl': '1.006', 'memory/max_active (GiB)': '33.78', 'memory/max_allocated (GiB)': '33.78', 'memory/device_reserved (GiB)': '35.91', 'tokens/train_per_sec_per_gpu': '31.99', 'tokens/total': 8622096, 'tokens/trainable': 130556, 'epoch': '1.113'}
	56%\|█████▌ \| 285/512 [31:29<25:02, 6.62s/it] 56%\|█████▌ \| 286/512 [31:36<24:54, 6.61s/it] {'loss': '0.000186', 'grad_norm': '0.009664', 'learning_rate': '5.022e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '35.91', 'tokens/train_per_sec_per_gpu': '34.84', 'tokens/total': 8652304, 'tokens/trainable': 131013, 'epoch': '1.117'}
	56%\|█████▌ \| 286/512 [31:36<24:54, 6.61s/it] 56%\|█████▌ \| 287/512 [31:43<24:46, 6.61s/it] {'loss': '0.0003628', 'grad_norm': '0.02171', 'learning_rate': '4.993e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.79', 'memory/max_allocated (GiB)': '33.79', 'memory/device_reserved (GiB)': '35.91', 'tokens/train_per_sec_per_gpu': '34.57', 'tokens/total': 8682704, 'tokens/trainable': 131469, 'epoch': '1.121'}
	56%\|█████▌ \| 287/512 [31:43<24:46, 6.61s/it] 56%\|█████▋ \| 288/512 [31:49<24:37, 6.60s/it] {'loss': '0.00155', 'grad_norm': '0.1057', 'learning_rate': '4.964e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.96', 'memory/max_allocated (GiB)': '33.96', 'memory/device_reserved (GiB)': '35.91', 'tokens/train_per_sec_per_gpu': '36.55', 'tokens/total': 8713056, 'tokens/trainable': 131943, 'epoch': '1.125'}
	56%\|█████▋ \| 288/512 [31:49<24:37, 6.60s/it][2026-08-18 14:49:46,838] [INFO] [axolotl.core.trainers.base._save:828] [PID:12244] Saving model checkpoint to /workspace/wave/training/checkpoints/checkpoint-288
	56%\|█████▋ \| 289/512 [31:57<26:27, 7.12s/it] {'loss': '0.001709', 'grad_norm': '0.09497', 'learning_rate': '4.935e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.78', 'memory/max_allocated (GiB)': '33.78', 'memory/device_reserved (GiB)': '35.91', 'tokens/train_per_sec_per_gpu': '33.71', 'tokens/total': 8743232, 'tokens/trainable': 132417, 'epoch': '1.129'}
	56%\|█████▋ \| 289/512 [31:57<26:27, 7.12s/it] 57%\|█████▋ \| 290/512 [32:04<25:45, 6.96s/it] {'loss': '0.0005101', 'grad_norm': '0.02392', 'learning_rate': '4.907e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.92', 'memory/max_allocated (GiB)': '33.92', 'memory/device_reserved (GiB)': '35.56', 'tokens/train_per_sec_per_gpu': '36.96', 'tokens/total': 8773664, 'tokens/trainable': 132866, 'epoch': '1.133'}
	57%\|█████▋ \| 290/512 [32:04<25:45, 6.96s/it] 57%\|█████▋ \| 291/512 [32:11<25:16, 6.86s/it] {'loss': '0.0002141', 'grad_norm': '0.008019', 'learning_rate': '4.878e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.8', 'memory/max_allocated (GiB)': '33.8', 'memory/device_reserved (GiB)': '35.56', 'tokens/train_per_sec_per_gpu': '36.26', 'tokens/total': 8803920, 'tokens/trainable': 133367, 'epoch': '1.137'}
	57%\|█████▋ \| 291/512 [32:11<25:16, 6.86s/it] 57%\|█████▋ \| 292/512 [32:17<24:50, 6.78s/it] {'loss': '0.001267', 'grad_norm': '0.1001', 'learning_rate': '4.849e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.91', 'memory/max_allocated (GiB)': '33.91', 'memory/device_reserved (GiB)': '35.58', 'tokens/train_per_sec_per_gpu': '34.8', 'tokens/total': 8834240, 'tokens/trainable': 133799, 'epoch': '1.141'}
	57%\|█████▋ \| 292/512 [32:17<24:50, 6.78s/it] 57%\|█████▋ \| 293/512 [32:24<24:35, 6.74s/it] {'loss': '0.0007189', 'grad_norm': '0.04551', 'learning_rate': '4.82e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '35.58', 'tokens/train_per_sec_per_gpu': '36.86', 'tokens/total': 8864560, 'tokens/trainable': 134271, 'epoch': '1.145'}
	57%\|█████▋ \| 293/512 [32:24<24:35, 6.74s/it] 57%\|█████▋ \| 294/512 [32:31<24:24, 6.72s/it] {'loss': '0.0005528', 'grad_norm': '0.02876', 'learning_rate': '4.792e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.93', 'memory/max_allocated (GiB)': '33.93', 'memory/device_reserved (GiB)': '35.58', 'tokens/train_per_sec_per_gpu': '34.05', 'tokens/total': 8895312, 'tokens/trainable': 134695, 'epoch': '1.148'}
	57%\|█████▋ \| 294/512 [32:31<24:24, 6.72s/it] 58%\|█████▊ \| 295/512 [32:37<24:11, 6.69s/it] {'loss': '0.001133', 'grad_norm': '0.08144', 'learning_rate': '4.763e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '35.58', 'tokens/train_per_sec_per_gpu': '36.1', 'tokens/total': 8925744, 'tokens/trainable': 135177, 'epoch': '1.152'}
	58%\|█████▊ \| 295/512 [32:37<24:11, 6.69s/it] 58%\|█████▊ \| 296/512 [32:44<23:56, 6.65s/it] {'loss': '0.001164', 'grad_norm': '0.08422', 'learning_rate': '4.734e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.93', 'memory/max_allocated (GiB)': '33.93', 'memory/device_reserved (GiB)': '35.66', 'tokens/train_per_sec_per_gpu': '33.71', 'tokens/total': 8956048, 'tokens/trainable': 135611, 'epoch': '1.156'}
	58%\|█████▊ \| 296/512 [32:44<23:56, 6.65s/it] 58%\|█████▊ \| 297/512 [32:50<23:45, 6.63s/it] {'loss': '0.0003869', 'grad_norm': '0.01976', 'learning_rate': '4.706e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.8', 'memory/max_allocated (GiB)': '33.8', 'memory/device_reserved (GiB)': '35.66', 'tokens/train_per_sec_per_gpu': '33.09', 'tokens/total': 8986320, 'tokens/trainable': 136050, 'epoch': '1.16'}
	58%\|█████▊ \| 297/512 [32:50<23:45, 6.63s/it] 58%\|█████▊ \| 298/512 [32:57<23:38, 6.63s/it] {'loss': '0.0003641', 'grad_norm': '0.0198', 'learning_rate': '4.677e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '35.66', 'tokens/train_per_sec_per_gpu': '33.38', 'tokens/total': 9016800, 'tokens/trainable': 136507, 'epoch': '1.164'}
	58%\|█████▊ \| 298/512 [32:57<23:38, 6.63s/it] 58%\|█████▊ \| 299/512 [33:04<23:32, 6.63s/it] {'loss': '0.0001828', 'grad_norm': '0.01045', 'learning_rate': '4.649e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '35.66', 'tokens/train_per_sec_per_gpu': '35.87', 'tokens/total': 9047184, 'tokens/trainable': 136965, 'epoch': '1.168'}
	58%\|█████▊ \| 299/512 [33:04<23:32, 6.63s/it] 59%\|█████▊ \| 300/512 [33:10<23:21, 6.61s/it] {'loss': '0.002034', 'grad_norm': '0.1379', 'learning_rate': '4.62e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.77', 'memory/max_allocated (GiB)': '33.77', 'memory/device_reserved (GiB)': '35.66', 'tokens/train_per_sec_per_gpu': '34.78', 'tokens/total': 9077488, 'tokens/trainable': 137415, 'epoch': '1.172'}
	59%\|█████▊ \| 300/512 [33:10<23:21, 6.61s/it] 59%\|█████▉ \| 301/512 [33:17<23:13, 6.61s/it] {'loss': '0.0006159', 'grad_norm': '0.05663', 'learning_rate': '4.592e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.93', 'memory/max_allocated (GiB)': '33.93', 'memory/device_reserved (GiB)': '35.66', 'tokens/train_per_sec_per_gpu': '36.22', 'tokens/total': 9107968, 'tokens/trainable': 137910, 'epoch': '1.176'}
	59%\|█████▉ \| 301/512 [33:17<23:13, 6.61s/it] 59%\|█████▉ \| 302/512 [33:23<23:09, 6.62s/it] {'loss': '0.000474', 'grad_norm': '0.02709', 'learning_rate': '4.563e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '35.66', 'tokens/train_per_sec_per_gpu': '34.8', 'tokens/total': 9138240, 'tokens/trainable': 138386, 'epoch': '1.18'}
	59%\|█████▉ \| 302/512 [33:23<23:09, 6.62s/it] 59%\|█████▉ \| 303/512 [33:30<23:05, 6.63s/it] {'loss': '0.0004738', 'grad_norm': '0.0356', 'learning_rate': '4.535e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '35.66', 'tokens/train_per_sec_per_gpu': '36.96', 'tokens/total': 9168640, 'tokens/trainable': 138845, 'epoch': '1.184'}
	59%\|█████▉ \| 303/512 [33:30<23:05, 6.63s/it] 59%\|█████▉ \| 304/512 [33:37<22:54, 6.61s/it] {'loss': '0.0008522', 'grad_norm': '0.07774', 'learning_rate': '4.507e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '35.66', 'tokens/train_per_sec_per_gpu': '31.07', 'tokens/total': 9198912, 'tokens/trainable': 139270, 'epoch': '1.188'}
	59%\|█████▉ \| 304/512 [33:37<22:54, 6.61s/it] 60%\|█████▉ \| 305/512 [33:43<22:48, 6.61s/it] {'loss': '0.01726', 'grad_norm': '0.2708', 'learning_rate': '4.478e-05', 'ppl': '1.017', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '35.66', 'tokens/train_per_sec_per_gpu': '33.55', 'tokens/total': 9229088, 'tokens/trainable': 139728, 'epoch': '1.191'}
	60%\|█████▉ \| 305/512 [33:43<22:48, 6.61s/it] 60%\|█████▉ \| 306/512 [33:50<22:42, 6.61s/it] {'loss': '0.001807', 'grad_norm': '0.1075', 'learning_rate': '4.45e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.96', 'memory/max_allocated (GiB)': '33.96', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '37.68', 'tokens/total': 9259680, 'tokens/trainable': 140219, 'epoch': '1.195'}
	60%\|█████▉ \| 306/512 [33:50<22:42, 6.61s/it] 60%\|█████▉ \| 307/512 [33:56<21:57, 6.43s/it] {'loss': '4.969e-05', 'grad_norm': '0.002589', 'learning_rate': '4.422e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.19', 'memory/max_allocated (GiB)': '33.19', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '34.51', 'tokens/total': 9287584, 'tokens/trainable': 140657, 'epoch': '1.199'}
	60%\|█████▉ \| 307/512 [33:56<21:57, 6.43s/it] 60%\|██████ \| 308/512 [34:03<22:06, 6.50s/it] {'loss': '0.0003076', 'grad_norm': '0.01864', 'learning_rate': '4.394e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.83', 'memory/max_allocated (GiB)': '33.83', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '30.38', 'tokens/total': 9318096, 'tokens/trainable': 141108, 'epoch': '1.203'}
	60%\|██████ \| 308/512 [34:03<22:06, 6.50s/it] 60%\|██████ \| 309/512 [34:09<22:05, 6.53s/it] {'loss': '0.0002081', 'grad_norm': '0.01055', 'learning_rate': '4.366e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '29.44', 'tokens/total': 9348544, 'tokens/trainable': 141531, 'epoch': '1.207'}
	60%\|██████ \| 309/512 [34:09<22:05, 6.53s/it] 61%\|██████ \| 310/512 [34:16<22:01, 6.54s/it] {'loss': '0.02379', 'grad_norm': '0.263', 'learning_rate': '4.338e-05', 'ppl': '1.024', 'memory/max_active (GiB)': '33.83', 'memory/max_allocated (GiB)': '33.83', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '35.94', 'tokens/total': 9378848, 'tokens/trainable': 141965, 'epoch': '1.211'}
	61%\|██████ \| 310/512 [34:16<22:01, 6.54s/it] 61%\|██████ \| 311/512 [34:22<21:57, 6.55s/it] {'loss': '0.0001811', 'grad_norm': '0.01139', 'learning_rate': '4.31e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '34.52', 'tokens/total': 9409104, 'tokens/trainable': 142429, 'epoch': '1.215'}
	61%\|██████ \| 311/512 [34:22<21:57, 6.55s/it] 61%\|██████ \| 312/512 [34:29<21:49, 6.55s/it] {'loss': '0.0001647', 'grad_norm': '0.01918', 'learning_rate': '4.282e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '38.76', 'tokens/total': 9439264, 'tokens/trainable': 142895, 'epoch': '1.219'}
	61%\|██████ \| 312/512 [34:29<21:49, 6.55s/it] 61%\|██████ \| 313/512 [34:35<21:45, 6.56s/it] {'loss': '0.004963', 'grad_norm': '0.4227', 'learning_rate': '4.254e-05', 'ppl': '1.005', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '37.12', 'tokens/total': 9469648, 'tokens/trainable': 143370, 'epoch': '1.223'}
	61%\|██████ \| 313/512 [34:35<21:45, 6.56s/it] 61%\|██████▏ \| 314/512 [34:42<21:40, 6.57s/it] {'loss': '0.0009556', 'grad_norm': '0.09607', 'learning_rate': '4.226e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.79', 'memory/max_allocated (GiB)': '33.79', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '33.41', 'tokens/total': 9499968, 'tokens/trainable': 143826, 'epoch': '1.227'}
	61%\|██████▏ \| 314/512 [34:42<21:40, 6.57s/it] 62%\|██████▏ \| 315/512 [34:49<21:41, 6.61s/it] {'loss': '0.0002801', 'grad_norm': '0.01082', 'learning_rate': '4.198e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.99', 'memory/max_allocated (GiB)': '33.99', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '35.11', 'tokens/total': 9530448, 'tokens/trainable': 144308, 'epoch': '1.23'}
	62%\|██████▏ \| 315/512 [34:49<21:41, 6.61s/it] 62%\|██████▏ \| 316/512 [34:55<21:33, 6.60s/it] {'loss': '0.0007925', 'grad_norm': '0.5568', 'learning_rate': '4.17e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.96', 'memory/max_allocated (GiB)': '33.96', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '36', 'tokens/total': 9560848, 'tokens/trainable': 144765, 'epoch': '1.234'}
	62%\|██████▏ \| 316/512 [34:55<21:33, 6.60s/it] 62%\|██████▏ \| 317/512 [35:02<21:31, 6.62s/it] {'loss': '0.009996', 'grad_norm': '0.3749', 'learning_rate': '4.143e-05', 'ppl': '1.01', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '37.75', 'tokens/total': 9591376, 'tokens/trainable': 145265, 'epoch': '1.238'}
	62%\|██████▏ \| 317/512 [35:02<21:31, 6.62s/it] 62%\|██████▏ \| 318/512 [35:09<21:23, 6.62s/it] {'loss': '0.002976', 'grad_norm': '0.225', 'learning_rate': '4.115e-05', 'ppl': '1.003', 'memory/max_active (GiB)': '33.89', 'memory/max_allocated (GiB)': '33.89', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '37.25', 'tokens/total': 9621728, 'tokens/trainable': 145724, 'epoch': '1.242'}
	62%\|██████▏ \| 318/512 [35:09<21:23, 6.62s/it] 62%\|██████▏ \| 319/512 [35:15<21:19, 6.63s/it] {'loss': '0.006409', 'grad_norm': '0.4703', 'learning_rate': '4.087e-05', 'ppl': '1.006', 'memory/max_active (GiB)': '33.98', 'memory/max_allocated (GiB)': '33.98', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '37.72', 'tokens/total': 9652272, 'tokens/trainable': 146219, 'epoch': '1.246'}
	62%\|██████▏ \| 319/512 [35:15<21:19, 6.63s/it] 62%\|██████▎ \| 320/512 [35:22<21:12, 6.63s/it] {'loss': '0.00235', 'grad_norm': '0.1789', 'learning_rate': '4.06e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.83', 'memory/max_allocated (GiB)': '33.83', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '34.58', 'tokens/total': 9682672, 'tokens/trainable': 146693, 'epoch': '1.25'}
	62%\|██████▎ \| 320/512 [35:22<21:12, 6.63s/it][2026-08-18 14:53:19,559] [INFO] [axolotl.core.trainers.base._save:828] [PID:12244] Saving model checkpoint to /workspace/wave/training/checkpoints/checkpoint-320
	63%\|██████▎ \| 321/512 [35:30<22:22, 7.03s/it] {'loss': '0.0006638', 'grad_norm': '0.06236', 'learning_rate': '4.032e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '35.9', 'tokens/train_per_sec_per_gpu': '35.87', 'tokens/total': 9712832, 'tokens/trainable': 147139, 'epoch': '1.254'}
	63%\|██████▎ \| 321/512 [35:30<22:22, 7.03s/it] 63%\|██████▎ \| 322/512 [35:36<21:49, 6.89s/it] {'loss': '0.0001746', 'grad_norm': '0.009023', 'learning_rate': '4.005e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.95', 'memory/max_allocated (GiB)': '33.95', 'memory/device_reserved (GiB)': '36.14', 'tokens/train_per_sec_per_gpu': '35.74', 'tokens/total': 9743344, 'tokens/trainable': 147589, 'epoch': '1.258'}
	63%\|██████▎ \| 322/512 [35:36<21:49, 6.89s/it] 63%\|██████▎ \| 323/512 [35:43<21:26, 6.81s/it] {'loss': '0.003418', 'grad_norm': '0.3831', 'learning_rate': '3.978e-05', 'ppl': '1.003', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '35.77', 'tokens/total': 9773760, 'tokens/trainable': 148045, 'epoch': '1.262'}
	63%\|██████▎ \| 323/512 [35:43<21:26, 6.81s/it] 63%\|██████▎ \| 324/512 [35:50<21:12, 6.77s/it] {'loss': '0.005115', 'grad_norm': '0.3204', 'learning_rate': '3.95e-05', 'ppl': '1.005', 'memory/max_active (GiB)': '34.01', 'memory/max_allocated (GiB)': '34.01', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '37.47', 'tokens/total': 9804080, 'tokens/trainable': 148494, 'epoch': '1.266'}
	63%\|██████▎ \| 324/512 [35:50<21:12, 6.77s/it] 63%\|██████▎ \| 325/512 [35:56<20:55, 6.72s/it] {'loss': '0.01596', 'grad_norm': '0.7489', 'learning_rate': '3.923e-05', 'ppl': '1.016', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '36.84', 'tokens/total': 9834400, 'tokens/trainable': 148957, 'epoch': '1.27'}
	63%\|██████▎ \| 325/512 [35:56<20:55, 6.72s/it] 64%\|██████▎ \| 326/512 [36:03<20:44, 6.69s/it] {'loss': '0.0001496', 'grad_norm': '0.009092', 'learning_rate': '3.896e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '38.41', 'tokens/total': 9864704, 'tokens/trainable': 149451, 'epoch': '1.273'}
	64%\|██████▎ \| 326/512 [36:03<20:44, 6.69s/it] 64%\|██████▍ \| 327/512 [36:09<20:32, 6.66s/it] {'loss': '0.0001391', 'grad_norm': '0.005111', 'learning_rate': '3.869e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.74', 'memory/max_allocated (GiB)': '33.74', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '40.07', 'tokens/total': 9894656, 'tokens/trainable': 149904, 'epoch': '1.277'}
	64%\|██████▍ \| 327/512 [36:09<20:32, 6.66s/it] 64%\|██████▍ \| 328/512 [36:16<20:24, 6.65s/it] {'loss': '0.0001377', 'grad_norm': '0.007466', 'learning_rate': '3.842e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '34.36', 'tokens/total': 9924976, 'tokens/trainable': 150378, 'epoch': '1.281'}
	64%\|██████▍ \| 328/512 [36:16<20:24, 6.65s/it] 64%\|██████▍ \| 329/512 [36:23<20:14, 6.64s/it] {'loss': '0.009102', 'grad_norm': '0.1656', 'learning_rate': '3.815e-05', 'ppl': '1.009', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '32.32', 'tokens/total': 9955376, 'tokens/trainable': 150853, 'epoch': '1.285'}
	64%\|██████▍ \| 329/512 [36:23<20:14, 6.64s/it] 64%\|██████▍ \| 330/512 [36:29<20:06, 6.63s/it] {'loss': '0.000196', 'grad_norm': '0.01061', 'learning_rate': '3.788e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '34.45', 'tokens/total': 9985696, 'tokens/trainable': 151289, 'epoch': '1.289'}
	64%\|██████▍ \| 330/512 [36:29<20:06, 6.63s/it] 65%\|██████▍ \| 331/512 [36:36<19:59, 6.62s/it] {'loss': '0.000298', 'grad_norm': '0.01095', 'learning_rate': '3.761e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '37.68', 'tokens/total': 10016112, 'tokens/trainable': 151757, 'epoch': '1.293'}
	65%\|██████▍ \| 331/512 [36:36<19:59, 6.62s/it] 65%\|██████▍ \| 332/512 [36:42<19:49, 6.61s/it] {'loss': '0.0002142', 'grad_norm': '0.01175', 'learning_rate': '3.734e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.91', 'memory/max_allocated (GiB)': '33.91', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '34.55', 'tokens/total': 10046544, 'tokens/trainable': 152207, 'epoch': '1.297'}
	65%\|██████▍ \| 332/512 [36:42<19:49, 6.61s/it] 65%\|██████▌ \| 333/512 [36:49<19:43, 6.61s/it] {'loss': '0.004726', 'grad_norm': '0.3032', 'learning_rate': '3.708e-05', 'ppl': '1.005', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '32.34', 'tokens/total': 10077120, 'tokens/trainable': 152646, 'epoch': '1.301'}
	65%\|██████▌ \| 333/512 [36:49<19:43, 6.61s/it] 65%\|██████▌ \| 334/512 [36:56<19:33, 6.59s/it] {'loss': '0.0005056', 'grad_norm': '0.02422', 'learning_rate': '3.681e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.93', 'memory/max_allocated (GiB)': '33.93', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '33.93', 'tokens/total': 10107600, 'tokens/trainable': 153063, 'epoch': '1.305'}
	65%\|██████▌ \| 334/512 [36:56<19:33, 6.59s/it] 65%\|██████▌ \| 335/512 [37:02<19:27, 6.59s/it] {'loss': '0.001724', 'grad_norm': '0.1015', 'learning_rate': '3.655e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.78', 'memory/max_allocated (GiB)': '33.78', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '30.92', 'tokens/total': 10137712, 'tokens/trainable': 153493, 'epoch': '1.309'}
	65%\|██████▌ \| 335/512 [37:02<19:27, 6.59s/it] 66%\|██████▌ \| 336/512 [37:09<19:20, 6.60s/it] {'loss': '0.001302', 'grad_norm': '0.06118', 'learning_rate': '3.628e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.93', 'memory/max_allocated (GiB)': '33.93', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '33.65', 'tokens/total': 10168096, 'tokens/trainable': 153945, 'epoch': '1.312'}
	66%\|██████▌ \| 336/512 [37:09<19:20, 6.60s/it] 66%\|██████▌ \| 337/512 [37:16<19:19, 6.63s/it] {'loss': '0.0003055', 'grad_norm': '0.02251', 'learning_rate': '3.602e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.91', 'memory/max_allocated (GiB)': '33.91', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '35.4', 'tokens/total': 10198720, 'tokens/trainable': 154432, 'epoch': '1.316'}
	66%\|██████▌ \| 337/512 [37:16<19:19, 6.63s/it] 66%\|██████▌ \| 338/512 [37:22<19:12, 6.62s/it] {'loss': '0.002099', 'grad_norm': '0.1921', 'learning_rate': '3.576e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '34.47', 'tokens/total': 10228976, 'tokens/trainable': 154881, 'epoch': '1.32'}
	66%\|██████▌ \| 338/512 [37:22<19:12, 6.62s/it] 66%\|██████▌ \| 339/512 [37:29<19:07, 6.64s/it] {'loss': '0.0002057', 'grad_norm': '0.008806', 'learning_rate': '3.549e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.81', 'memory/max_allocated (GiB)': '33.81', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '30.32', 'tokens/total': 10259440, 'tokens/trainable': 155306, 'epoch': '1.324'}
	66%\|██████▌ \| 339/512 [37:29<19:07, 6.64s/it] 66%\|██████▋ \| 340/512 [37:35<18:55, 6.60s/it] {'loss': '0.0003597', 'grad_norm': '0.01875', 'learning_rate': '3.523e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '31.7', 'tokens/total': 10289824, 'tokens/trainable': 155719, 'epoch': '1.328'}
	66%\|██████▋ \| 340/512 [37:35<18:55, 6.60s/it] 67%\|██████▋ \| 341/512 [37:42<18:51, 6.62s/it] {'loss': '0.0001035', 'grad_norm': '0.005896', 'learning_rate': '3.497e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '36.3', 'tokens/total': 10320144, 'tokens/trainable': 156209, 'epoch': '1.332'}
	67%\|██████▋ \| 341/512 [37:42<18:51, 6.62s/it] 67%\|██████▋ \| 342/512 [37:49<18:43, 6.61s/it] {'loss': '0.000157', 'grad_norm': '0.01002', 'learning_rate': '3.471e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.98', 'memory/max_allocated (GiB)': '33.98', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '34.17', 'tokens/total': 10350528, 'tokens/trainable': 156656, 'epoch': '1.336'}
	67%\|██████▋ \| 342/512 [37:49<18:43, 6.61s/it] 67%\|██████▋ \| 343/512 [37:55<18:34, 6.60s/it] {'loss': '0.00135', 'grad_norm': '0.1039', 'learning_rate': '3.445e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.72', 'memory/max_allocated (GiB)': '33.72', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '34.23', 'tokens/total': 10380816, 'tokens/trainable': 157118, 'epoch': '1.34'}
	67%\|██████▋ \| 343/512 [37:55<18:34, 6.60s/it] 67%\|██████▋ \| 344/512 [38:02<18:29, 6.60s/it] {'loss': '0.0005016', 'grad_norm': '0.04982', 'learning_rate': '3.42e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '33.56', 'tokens/total': 10411312, 'tokens/trainable': 157562, 'epoch': '1.344'}
	67%\|██████▋ \| 344/512 [38:02<18:29, 6.60s/it] 67%\|██████▋ \| 345/512 [38:08<18:21, 6.60s/it] {'loss': '0.0001865', 'grad_norm': '0.01197', 'learning_rate': '3.394e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '35.7', 'tokens/total': 10441696, 'tokens/trainable': 158003, 'epoch': '1.348'}
	67%\|██████▋ \| 345/512 [38:08<18:21, 6.60s/it] 68%\|██████▊ \| 346/512 [38:15<18:13, 6.59s/it] {'loss': '0.0004138', 'grad_norm': '0.02001', 'learning_rate': '3.368e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '29.4', 'tokens/total': 10472032, 'tokens/trainable': 158419, 'epoch': '1.352'}
	68%\|██████▊ \| 346/512 [38:15<18:13, 6.59s/it] 68%\|██████▊ \| 347/512 [38:21<18:03, 6.57s/it] {'loss': '0.001067', 'grad_norm': '0.1378', 'learning_rate': '3.343e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.73', 'memory/max_allocated (GiB)': '33.73', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '36.17', 'tokens/total': 10502176, 'tokens/trainable': 158877, 'epoch': '1.355'}
	68%\|██████▊ \| 347/512 [38:21<18:03, 6.57s/it] 68%\|██████▊ \| 348/512 [38:28<18:01, 6.59s/it] {'loss': '0.0002564', 'grad_norm': '0.01242', 'learning_rate': '3.317e-05', 'ppl': '1', 'memory/max_active (GiB)': '34', 'memory/max_allocated (GiB)': '34', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '29.19', 'tokens/total': 10532896, 'tokens/trainable': 159299, 'epoch': '1.359'}
	68%\|██████▊ \| 348/512 [38:28<18:01, 6.59s/it] 68%\|██████▊ \| 349/512 [38:35<17:53, 6.59s/it] {'loss': '0.0005879', 'grad_norm': '0.03198', 'learning_rate': '3.292e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.76', 'memory/max_allocated (GiB)': '33.76', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '33.05', 'tokens/total': 10563024, 'tokens/trainable': 159750, 'epoch': '1.363'}
	68%\|██████▊ \| 349/512 [38:35<17:53, 6.59s/it] 68%\|██████▊ \| 350/512 [38:41<17:51, 6.62s/it] {'loss': '0.0001015', 'grad_norm': '0.009861', 'learning_rate': '3.267e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.99', 'memory/max_allocated (GiB)': '33.99', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '35.76', 'tokens/total': 10593536, 'tokens/trainable': 160217, 'epoch': '1.367'}
	68%\|██████▊ \| 350/512 [38:41<17:51, 6.62s/it] 69%\|██████▊ \| 351/512 [38:48<17:45, 6.62s/it] {'loss': '0.000101', 'grad_norm': '0.005884', 'learning_rate': '3.242e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '33.11', 'tokens/total': 10623776, 'tokens/trainable': 160675, 'epoch': '1.371'}
	69%\|██████▊ \| 351/512 [38:48<17:45, 6.62s/it] 69%\|██████▉ \| 352/512 [38:55<17:37, 6.61s/it] {'loss': '0.003672', 'grad_norm': '0.2425', 'learning_rate': '3.217e-05', 'ppl': '1.004', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '35.02', 'tokens/total': 10654032, 'tokens/trainable': 161132, 'epoch': '1.375'}
	69%\|██████▉ \| 352/512 [38:55<17:37, 6.61s/it][2026-08-18 14:56:52,315] [INFO] [axolotl.core.trainers.base._save:828] [PID:12244] Saving model checkpoint to /workspace/wave/training/checkpoints/checkpoint-352
	69%\|██████▉ \| 353/512 [39:03<18:42, 7.06s/it] {'loss': '6.496e-05', 'grad_norm': '0.003396', 'learning_rate': '3.192e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '29.4', 'tokens/total': 10684352, 'tokens/trainable': 161585, 'epoch': '1.379'}
	69%\|██████▉ \| 353/512 [39:03<18:42, 7.06s/it] 69%\|██████▉ \| 354/512 [39:09<18:13, 6.92s/it] {'loss': '0.008384', 'grad_norm': '0.8878', 'learning_rate': '3.167e-05', 'ppl': '1.008', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '35.41', 'tokens/train_per_sec_per_gpu': '37.32', 'tokens/total': 10714816, 'tokens/trainable': 162050, 'epoch': '1.383'}
	69%\|██████▉ \| 354/512 [39:09<18:13, 6.92s/it] 69%\|██████▉ \| 355/512 [39:16<17:50, 6.82s/it] {'loss': '8.07e-05', 'grad_norm': '0.004499', 'learning_rate': '3.142e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '35.84', 'tokens/train_per_sec_per_gpu': '32.65', 'tokens/total': 10745296, 'tokens/trainable': 162478, 'epoch': '1.387'}
	69%\|██████▉ \| 355/512 [39:16<17:50, 6.82s/it] 70%\|██████▉ \| 356/512 [39:22<17:29, 6.73s/it] {'loss': '7.578e-05', 'grad_norm': '0.004872', 'learning_rate': '3.117e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.61', 'memory/max_allocated (GiB)': '33.61', 'memory/device_reserved (GiB)': '35.84', 'tokens/train_per_sec_per_gpu': '28.74', 'tokens/total': 10775296, 'tokens/trainable': 162893, 'epoch': '1.391'}
	70%\|██████▉ \| 356/512 [39:22<17:29, 6.73s/it] 70%\|██████▉ \| 357/512 [39:29<17:16, 6.69s/it] {'loss': '0.003035', 'grad_norm': '0.2148', 'learning_rate': '3.093e-05', 'ppl': '1.003', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '35.84', 'tokens/train_per_sec_per_gpu': '35.49', 'tokens/total': 10805600, 'tokens/trainable': 163347, 'epoch': '1.395'}
	70%\|██████▉ \| 357/512 [39:29<17:16, 6.69s/it] 70%\|██████▉ \| 358/512 [39:36<17:06, 6.66s/it] {'loss': '0.0004827', 'grad_norm': '0.07261', 'learning_rate': '3.068e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.83', 'memory/max_allocated (GiB)': '33.83', 'memory/device_reserved (GiB)': '35.84', 'tokens/train_per_sec_per_gpu': '33.58', 'tokens/total': 10835760, 'tokens/trainable': 163796, 'epoch': '1.398'}
	70%\|██████▉ \| 358/512 [39:36<17:06, 6.66s/it] 70%\|███████ \| 359/512 [39:42<16:54, 6.63s/it] {'loss': '0.0002045', 'grad_norm': '0.02204', 'learning_rate': '3.044e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '35.84', 'tokens/train_per_sec_per_gpu': '32.47', 'tokens/total': 10866208, 'tokens/trainable': 164237, 'epoch': '1.402'}
	70%\|███████ \| 359/512 [39:42<16:54, 6.63s/it] 70%\|███████ \| 360/512 [39:49<16:46, 6.62s/it] {'loss': '0.0001249', 'grad_norm': '0.005709', 'learning_rate': '3.02e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '35.84', 'tokens/train_per_sec_per_gpu': '34.69', 'tokens/total': 10896672, 'tokens/trainable': 164684, 'epoch': '1.406'}
	70%\|███████ \| 360/512 [39:49<16:46, 6.62s/it] 71%\|███████ \| 361/512 [39:55<16:38, 6.61s/it] {'loss': '0.0002532', 'grad_norm': '0.0225', 'learning_rate': '2.995e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.8', 'memory/max_allocated (GiB)': '33.8', 'memory/device_reserved (GiB)': '35.84', 'tokens/train_per_sec_per_gpu': '37.05', 'tokens/total': 10926944, 'tokens/trainable': 165146, 'epoch': '1.41'}
	71%\|███████ \| 361/512 [39:55<16:38, 6.61s/it] 71%\|███████ \| 362/512 [40:02<16:33, 6.62s/it] {'loss': '0.0002238', 'grad_norm': '0.01391', 'learning_rate': '2.971e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '35.84', 'tokens/train_per_sec_per_gpu': '35.37', 'tokens/total': 10957456, 'tokens/trainable': 165632, 'epoch': '1.414'}
	71%\|███████ \| 362/512 [40:02<16:33, 6.62s/it] 71%\|███████ \| 363/512 [40:09<16:23, 6.60s/it] {'loss': '0.01284', 'grad_norm': '0.2468', 'learning_rate': '2.947e-05', 'ppl': '1.013', 'memory/max_active (GiB)': '33.89', 'memory/max_allocated (GiB)': '33.89', 'memory/device_reserved (GiB)': '35.84', 'tokens/train_per_sec_per_gpu': '34.66', 'tokens/total': 10987696, 'tokens/trainable': 166080, 'epoch': '1.418'}
	71%\|███████ \| 363/512 [40:09<16:23, 6.60s/it] 71%\|███████ \| 364/512 [40:15<16:17, 6.61s/it] {'loss': '0.0001312', 'grad_norm': '0.008053', 'learning_rate': '2.924e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '35.84', 'tokens/train_per_sec_per_gpu': '34.62', 'tokens/total': 11017904, 'tokens/trainable': 166531, 'epoch': '1.422'}
	71%\|███████ \| 364/512 [40:15<16:17, 6.61s/it] 71%\|███████▏ \| 365/512 [40:22<16:11, 6.61s/it] {'loss': '0.0002447', 'grad_norm': '0.04744', 'learning_rate': '2.9e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.83', 'memory/max_allocated (GiB)': '33.83', 'memory/device_reserved (GiB)': '35.84', 'tokens/train_per_sec_per_gpu': '36.91', 'tokens/total': 11048224, 'tokens/trainable': 167019, 'epoch': '1.426'}
	71%\|███████▏ \| 365/512 [40:22<16:11, 6.61s/it] 71%\|███████▏ \| 366/512 [40:28<16:04, 6.60s/it] {'loss': '0.0001016', 'grad_norm': '0.008861', 'learning_rate': '2.876e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.83', 'memory/max_allocated (GiB)': '33.83', 'memory/device_reserved (GiB)': '35.84', 'tokens/train_per_sec_per_gpu': '30.94', 'tokens/total': 11078544, 'tokens/trainable': 167469, 'epoch': '1.43'}
	71%\|███████▏ \| 366/512 [40:28<16:04, 6.60s/it] 72%\|███████▏ \| 367/512 [40:35<15:58, 6.61s/it] {'loss': '0.0006842', 'grad_norm': '0.06953', 'learning_rate': '2.853e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '31.86', 'tokens/total': 11109088, 'tokens/trainable': 167914, 'epoch': '1.434'}
	72%\|███████▏ \| 367/512 [40:35<15:58, 6.61s/it] 72%\|███████▏ \| 368/512 [40:42<15:50, 6.60s/it] {'loss': '0.000614', 'grad_norm': '0.1097', 'learning_rate': '2.829e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '36.16', 'tokens/total': 11139408, 'tokens/trainable': 168357, 'epoch': '1.438'}
	72%\|███████▏ \| 368/512 [40:42<15:50, 6.60s/it] 72%\|███████▏ \| 369/512 [40:48<15:46, 6.62s/it] {'loss': '0.0001494', 'grad_norm': '0.01329', 'learning_rate': '2.806e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '32.75', 'tokens/total': 11169920, 'tokens/trainable': 168832, 'epoch': '1.441'}
	72%\|███████▏ \| 369/512 [40:48<15:46, 6.62s/it] 72%\|███████▏ \| 370/512 [40:55<15:40, 6.62s/it] {'loss': '0.0134', 'grad_norm': '0.7464', 'learning_rate': '2.783e-05', 'ppl': '1.013', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '37.41', 'tokens/total': 11200288, 'tokens/trainable': 169333, 'epoch': '1.445'}
	72%\|███████▏ \| 370/512 [40:55<15:40, 6.62s/it] 72%\|███████▏ \| 371/512 [41:02<15:36, 6.64s/it] {'loss': '2.964e-05', 'grad_norm': '0.001093', 'learning_rate': '2.76e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '35.62', 'tokens/total': 11230832, 'tokens/trainable': 169774, 'epoch': '1.449'}
	72%\|███████▏ \| 371/512 [41:02<15:36, 6.64s/it] 73%\|███████▎ \| 372/512 [41:08<15:29, 6.64s/it] {'loss': '0.001514', 'grad_norm': '0.1324', 'learning_rate': '2.737e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '33.41', 'tokens/total': 11261264, 'tokens/trainable': 170241, 'epoch': '1.453'}
	73%\|███████▎ \| 372/512 [41:08<15:29, 6.64s/it] 73%\|███████▎ \| 373/512 [41:15<15:22, 6.64s/it] {'loss': '0.0001303', 'grad_norm': '0.006302', 'learning_rate': '2.714e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.78', 'memory/max_allocated (GiB)': '33.78', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '33.69', 'tokens/total': 11291456, 'tokens/trainable': 170706, 'epoch': '1.457'}
	73%\|███████▎ \| 373/512 [41:15<15:22, 6.64s/it] 73%\|███████▎ \| 374/512 [41:21<15:14, 6.63s/it] {'loss': '0.0002068', 'grad_norm': '0.0117', 'learning_rate': '2.691e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.97', 'memory/max_allocated (GiB)': '33.97', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '32.53', 'tokens/total': 11321952, 'tokens/trainable': 171182, 'epoch': '1.461'}
	73%\|███████▎ \| 374/512 [41:21<15:14, 6.63s/it] 73%\|███████▎ \| 375/512 [41:28<15:04, 6.60s/it] {'loss': '0.0003061', 'grad_norm': '0.01545', 'learning_rate': '2.668e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.72', 'memory/max_allocated (GiB)': '33.72', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '34.75', 'tokens/total': 11352096, 'tokens/trainable': 171646, 'epoch': '1.465'}
	73%\|███████▎ \| 375/512 [41:28<15:04, 6.60s/it] 73%\|███████▎ \| 376/512 [41:35<14:56, 6.59s/it] {'loss': '0.000194', 'grad_norm': '0.008863', 'learning_rate': '2.646e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '28.57', 'tokens/total': 11382528, 'tokens/trainable': 172063, 'epoch': '1.469'}
	73%\|███████▎ \| 376/512 [41:35<14:56, 6.59s/it] 74%\|███████▎ \| 377/512 [41:41<14:50, 6.60s/it] {'loss': '0.0004883', 'grad_norm': '0.03389', 'learning_rate': '2.624e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '36.63', 'tokens/total': 11412832, 'tokens/trainable': 172539, 'epoch': '1.473'}
	74%\|███████▎ \| 377/512 [41:41<14:50, 6.60s/it] 74%\|███████▍ \| 378/512 [41:48<14:43, 6.59s/it] {'loss': '0.000516', 'grad_norm': '0.02547', 'learning_rate': '2.601e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.71', 'memory/max_allocated (GiB)': '33.71', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '34.99', 'tokens/total': 11442864, 'tokens/trainable': 172997, 'epoch': '1.477'}
	74%\|███████▍ \| 378/512 [41:48<14:43, 6.59s/it] 74%\|███████▍ \| 379/512 [41:54<14:36, 6.59s/it] {'loss': '0.0004852', 'grad_norm': '0.04523', 'learning_rate': '2.579e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '36.26', 'tokens/total': 11473328, 'tokens/trainable': 173481, 'epoch': '1.48'}
	74%\|███████▍ \| 379/512 [41:54<14:36, 6.59s/it] 74%\|███████▍ \| 380/512 [42:01<14:30, 6.59s/it] {'loss': '0.000263', 'grad_norm': '0.01156', 'learning_rate': '2.557e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.73', 'memory/max_allocated (GiB)': '33.73', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '30.94', 'tokens/total': 11503504, 'tokens/trainable': 173946, 'epoch': '1.484'}
	74%\|███████▍ \| 380/512 [42:01<14:30, 6.59s/it] 74%\|███████▍ \| 381/512 [42:08<14:25, 6.60s/it] {'loss': '0.0004146', 'grad_norm': '0.01947', 'learning_rate': '2.535e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '34.7', 'tokens/total': 11533920, 'tokens/trainable': 174417, 'epoch': '1.488'}
	74%\|███████▍ \| 381/512 [42:08<14:25, 6.60s/it] 75%\|███████▍ \| 382/512 [42:14<14:19, 6.61s/it] {'loss': '0.000521', 'grad_norm': '0.03897', 'learning_rate': '2.513e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.83', 'memory/max_allocated (GiB)': '33.83', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '33.17', 'tokens/total': 11564128, 'tokens/trainable': 174872, 'epoch': '1.492'}
	75%\|███████▍ \| 382/512 [42:14<14:19, 6.61s/it] 75%\|███████▍ \| 383/512 [42:21<14:13, 6.62s/it] {'loss': '0.0002654', 'grad_norm': '0.01254', 'learning_rate': '2.492e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.91', 'memory/max_allocated (GiB)': '33.91', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '31.99', 'tokens/total': 11594672, 'tokens/trainable': 175313, 'epoch': '1.496'}
	75%\|███████▍ \| 383/512 [42:21<14:13, 6.62s/it] 75%\|███████▌ \| 384/512 [42:27<14:08, 6.63s/it] {'loss': '0.01107', 'grad_norm': '0.2314', 'learning_rate': '2.47e-05', 'ppl': '1.011', 'memory/max_active (GiB)': '33.81', 'memory/max_allocated (GiB)': '33.81', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '32.96', 'tokens/total': 11625056, 'tokens/trainable': 175750, 'epoch': '1.5'}
	75%\|███████▌ \| 384/512 [42:27<14:08, 6.63s/it][2026-08-18 15:00:25,170] [INFO] [axolotl.core.trainers.base._save:828] [PID:12244] Saving model checkpoint to /workspace/wave/training/checkpoints/checkpoint-384
	75%\|███████▌ \| 385/512 [42:36<14:57, 7.07s/it] {'loss': '0.01684', 'grad_norm': '0.5276', 'learning_rate': '2.449e-05', 'ppl': '1.017', 'memory/max_active (GiB)': '33.95', 'memory/max_allocated (GiB)': '33.95', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '35.8', 'tokens/total': 11655488, 'tokens/trainable': 176232, 'epoch': '1.504'}
	75%\|███████▌ \| 385/512 [42:36<14:57, 7.07s/it] 75%\|███████▌ \| 386/512 [42:42<14:32, 6.93s/it] {'loss': '0.0001142', 'grad_norm': '0.008643', 'learning_rate': '2.428e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '35.87', 'tokens/train_per_sec_per_gpu': '34.9', 'tokens/total': 11685968, 'tokens/trainable': 176680, 'epoch': '1.508'}
	75%\|███████▌ \| 386/512 [42:42<14:32, 6.93s/it] 76%\|███████▌ \| 387/512 [42:49<14:14, 6.83s/it] {'loss': '0.0001288', 'grad_norm': '0.006846', 'learning_rate': '2.406e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '35.87', 'tokens/train_per_sec_per_gpu': '33.24', 'tokens/total': 11716176, 'tokens/trainable': 177144, 'epoch': '1.512'}
	76%\|███████▌ \| 387/512 [42:49<14:14, 6.83s/it] 76%\|███████▌ \| 388/512 [42:55<13:58, 6.76s/it] {'loss': '0.001054', 'grad_norm': '0.07777', 'learning_rate': '2.385e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '35.87', 'tokens/train_per_sec_per_gpu': '36.67', 'tokens/total': 11746704, 'tokens/trainable': 177630, 'epoch': '1.516'}
	76%\|███████▌ \| 388/512 [42:55<13:58, 6.76s/it] 76%\|███████▌ \| 389/512 [43:02<13:45, 6.71s/it] {'loss': '0.0004593', 'grad_norm': '0.02181', 'learning_rate': '2.365e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.81', 'memory/max_allocated (GiB)': '33.81', 'memory/device_reserved (GiB)': '35.87', 'tokens/train_per_sec_per_gpu': '33.51', 'tokens/total': 11777008, 'tokens/trainable': 178079, 'epoch': '1.52'}
	76%\|███████▌ \| 389/512 [43:02<13:45, 6.71s/it] 76%\|███████▌ \| 390/512 [43:09<13:35, 6.69s/it] {'loss': '0.006384', 'grad_norm': '0.2766', 'learning_rate': '2.344e-05', 'ppl': '1.006', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '35.87', 'tokens/train_per_sec_per_gpu': '36.99', 'tokens/total': 11807168, 'tokens/trainable': 178552, 'epoch': '1.523'}
	76%\|███████▌ \| 390/512 [43:09<13:35, 6.69s/it] 76%\|███████▋ \| 391/512 [43:15<13:27, 6.67s/it] {'loss': '0.003972', 'grad_norm': '0.5077', 'learning_rate': '2.323e-05', 'ppl': '1.004', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '35.87', 'tokens/train_per_sec_per_gpu': '34.49', 'tokens/total': 11837776, 'tokens/trainable': 179021, 'epoch': '1.527'}
	76%\|███████▋ \| 391/512 [43:15<13:27, 6.67s/it] 77%\|███████▋ \| 392/512 [43:22<13:16, 6.64s/it] {'loss': '0.0004878', 'grad_norm': '0.04643', 'learning_rate': '2.303e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '35.87', 'tokens/train_per_sec_per_gpu': '33.57', 'tokens/total': 11867968, 'tokens/trainable': 179478, 'epoch': '1.531'}
	77%\|███████▋ \| 392/512 [43:22<13:16, 6.64s/it] 77%\|███████▋ \| 393/512 [43:28<13:06, 6.61s/it] {'loss': '0.00317', 'grad_norm': '0.1215', 'learning_rate': '2.282e-05', 'ppl': '1.003', 'memory/max_active (GiB)': '33.96', 'memory/max_allocated (GiB)': '33.96', 'memory/device_reserved (GiB)': '35.87', 'tokens/train_per_sec_per_gpu': '33.27', 'tokens/total': 11898320, 'tokens/trainable': 179910, 'epoch': '1.535'}
	77%\|███████▋ \| 393/512 [43:28<13:06, 6.61s/it] 77%\|███████▋ \| 394/512 [43:35<12:59, 6.61s/it] {'loss': '0.0002976', 'grad_norm': '0.02432', 'learning_rate': '2.262e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.81', 'memory/max_allocated (GiB)': '33.81', 'memory/device_reserved (GiB)': '35.87', 'tokens/train_per_sec_per_gpu': '34.45', 'tokens/total': 11928624, 'tokens/trainable': 180361, 'epoch': '1.539'}
	77%\|███████▋ \| 394/512 [43:35<12:59, 6.61s/it] 77%\|███████▋ \| 395/512 [43:41<12:50, 6.58s/it] {'loss': '0.0003667', 'grad_norm': '0.01568', 'learning_rate': '2.242e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '32.31', 'tokens/total': 11959088, 'tokens/trainable': 180774, 'epoch': '1.543'}
	77%\|███████▋ \| 395/512 [43:41<12:50, 6.58s/it] 77%\|███████▋ \| 396/512 [43:48<12:44, 6.59s/it] {'loss': '0.0003036', 'grad_norm': '0.02069', 'learning_rate': '2.222e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.92', 'memory/max_allocated (GiB)': '33.92', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '35.69', 'tokens/total': 11989504, 'tokens/trainable': 181246, 'epoch': '1.547'}
	77%\|███████▋ \| 396/512 [43:48<12:44, 6.59s/it] 78%\|███████▊ \| 397/512 [43:55<12:39, 6.60s/it] {'loss': '0.008752', 'grad_norm': '0.2614', 'learning_rate': '2.202e-05', 'ppl': '1.009', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '33.65', 'tokens/total': 12019904, 'tokens/trainable': 181734, 'epoch': '1.551'}
	78%\|███████▊ \| 397/512 [43:55<12:39, 6.60s/it] 78%\|███████▊ \| 398/512 [44:01<12:32, 6.60s/it] {'loss': '0.0004417', 'grad_norm': '0.03743', 'learning_rate': '2.183e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.93', 'memory/max_allocated (GiB)': '33.93', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '36.01', 'tokens/total': 12050384, 'tokens/trainable': 182169, 'epoch': '1.555'}
	78%\|███████▊ \| 398/512 [44:01<12:32, 6.60s/it] 78%\|███████▊ \| 399/512 [44:08<12:26, 6.60s/it] {'loss': '0.0007634', 'grad_norm': '0.1698', 'learning_rate': '2.163e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.79', 'memory/max_allocated (GiB)': '33.79', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '35.34', 'tokens/total': 12080576, 'tokens/trainable': 182641, 'epoch': '1.559'}
	78%\|███████▊ \| 399/512 [44:08<12:26, 6.60s/it] 78%\|███████▊ \| 400/512 [44:14<12:18, 6.60s/it] {'loss': '0.003636', 'grad_norm': '0.1729', 'learning_rate': '2.144e-05', 'ppl': '1.004', 'memory/max_active (GiB)': '33.74', 'memory/max_allocated (GiB)': '33.74', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '26.93', 'tokens/total': 12110896, 'tokens/trainable': 183056, 'epoch': '1.562'}
	78%\|███████▊ \| 400/512 [44:14<12:18, 6.60s/it] 78%\|███████▊ \| 401/512 [44:21<12:10, 6.59s/it] {'loss': '0.0008229', 'grad_norm': '0.05639', 'learning_rate': '2.124e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.71', 'memory/max_allocated (GiB)': '33.71', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '32.32', 'tokens/total': 12140784, 'tokens/trainable': 183504, 'epoch': '1.566'}
	78%\|███████▊ \| 401/512 [44:21<12:10, 6.59s/it] 79%\|███████▊ \| 402/512 [44:28<12:06, 6.60s/it] {'loss': '0.0007216', 'grad_norm': '0.03654', 'learning_rate': '2.105e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '39.29', 'tokens/total': 12171024, 'tokens/trainable': 184004, 'epoch': '1.57'}
	79%\|███████▊ \| 402/512 [44:28<12:06, 6.60s/it] 79%\|███████▊ \| 403/512 [44:34<12:00, 6.61s/it] {'loss': '0.0002983', 'grad_norm': '0.01657', 'learning_rate': '2.086e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.78', 'memory/max_allocated (GiB)': '33.78', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '33.04', 'tokens/total': 12201184, 'tokens/trainable': 184476, 'epoch': '1.574'}
	79%\|███████▊ \| 403/512 [44:34<12:00, 6.61s/it] 79%\|███████▉ \| 404/512 [44:41<11:52, 6.59s/it] {'loss': '0.0007513', 'grad_norm': '0.06593', 'learning_rate': '2.067e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '32.35', 'tokens/total': 12231232, 'tokens/trainable': 184905, 'epoch': '1.578'}
	79%\|███████▉ \| 404/512 [44:41<11:52, 6.59s/it] 79%\|███████▉ \| 405/512 [44:47<11:43, 6.58s/it] {'loss': '0.0001536', 'grad_norm': '0.01335', 'learning_rate': '2.049e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.72', 'memory/max_allocated (GiB)': '33.72', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '29.82', 'tokens/total': 12261280, 'tokens/trainable': 185338, 'epoch': '1.582'}
	79%\|███████▉ \| 405/512 [44:47<11:43, 6.58s/it] 79%\|███████▉ \| 406/512 [44:54<11:37, 6.58s/it] {'loss': '0.001165', 'grad_norm': '0.1154', 'learning_rate': '2.03e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.79', 'memory/max_allocated (GiB)': '33.79', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '32.02', 'tokens/total': 12291520, 'tokens/trainable': 185792, 'epoch': '1.586'}
	79%\|███████▉ \| 406/512 [44:54<11:37, 6.58s/it] 79%\|███████▉ \| 407/512 [45:01<11:30, 6.57s/it] {'loss': '0.0003108', 'grad_norm': '0.0157', 'learning_rate': '2.012e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.95', 'memory/max_allocated (GiB)': '33.95', 'memory/device_reserved (GiB)': '35.88', 'tokens/train_per_sec_per_gpu': '37.41', 'tokens/total': 12321824, 'tokens/trainable': 186251, 'epoch': '1.59'}
	79%\|███████▉ \| 407/512 [45:01<11:30, 6.57s/it] 80%\|███████▉ \| 408/512 [45:07<11:24, 6.58s/it] {'loss': '0.0003415', 'grad_norm': '0.02254', 'learning_rate': '1.993e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.95', 'memory/max_allocated (GiB)': '33.95', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '33.33', 'tokens/total': 12352176, 'tokens/trainable': 186689, 'epoch': '1.594'}
	80%\|███████▉ \| 408/512 [45:07<11:24, 6.58s/it] 80%\|███████▉ \| 409/512 [45:14<11:18, 6.59s/it] {'loss': '0.005957', 'grad_norm': '0.4058', 'learning_rate': '1.975e-05', 'ppl': '1.006', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '34.7', 'tokens/total': 12382608, 'tokens/trainable': 187121, 'epoch': '1.598'}
	80%\|███████▉ \| 409/512 [45:14<11:18, 6.59s/it] 80%\|████████ \| 410/512 [45:20<11:12, 6.59s/it] {'loss': '0.001334', 'grad_norm': '0.15', 'learning_rate': '1.957e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '35.55', 'tokens/total': 12413152, 'tokens/trainable': 187568, 'epoch': '1.602'}
	80%\|████████ \| 410/512 [45:20<11:12, 6.59s/it] 80%\|████████ \| 411/512 [45:27<11:05, 6.59s/it] {'loss': '0.00013', 'grad_norm': '0.006942', 'learning_rate': '1.94e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '34.76', 'tokens/total': 12443600, 'tokens/trainable': 188010, 'epoch': '1.605'}
	80%\|████████ \| 411/512 [45:27<11:05, 6.59s/it] 80%\|████████ \| 412/512 [45:33<10:43, 6.43s/it] {'loss': '0.001001', 'grad_norm': '0.1238', 'learning_rate': '1.922e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.91', 'memory/max_allocated (GiB)': '33.91', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '39.3', 'tokens/total': 12471952, 'tokens/trainable': 188446, 'epoch': '1.609'}
	80%\|████████ \| 412/512 [45:33<10:43, 6.43s/it] 81%\|████████ \| 413/512 [45:40<10:43, 6.50s/it] {'loss': '0.007401', 'grad_norm': '0.3418', 'learning_rate': '1.904e-05', 'ppl': '1.007', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '32.07', 'tokens/total': 12502480, 'tokens/trainable': 188897, 'epoch': '1.613'}
	81%\|████████ \| 413/512 [45:40<10:43, 6.50s/it] 81%\|████████ \| 414/512 [45:46<10:40, 6.53s/it] {'loss': '6.43e-05', 'grad_norm': '0.006758', 'learning_rate': '1.887e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.81', 'memory/max_allocated (GiB)': '33.81', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '37.49', 'tokens/total': 12532720, 'tokens/trainable': 189366, 'epoch': '1.617'}
	81%\|████████ \| 414/512 [45:46<10:40, 6.53s/it] 81%\|████████ \| 415/512 [45:53<10:33, 6.54s/it] {'loss': '0.005144', 'grad_norm': '0.1996', 'learning_rate': '1.87e-05', 'ppl': '1.005', 'memory/max_active (GiB)': '33.77', 'memory/max_allocated (GiB)': '33.77', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '32.4', 'tokens/total': 12562848, 'tokens/trainable': 189779, 'epoch': '1.621'}
	81%\|████████ \| 415/512 [45:53<10:33, 6.54s/it] 81%\|████████▏ \| 416/512 [45:59<10:30, 6.56s/it] {'loss': '0.0007223', 'grad_norm': '0.1678', 'learning_rate': '1.853e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '34.56', 'tokens/total': 12593328, 'tokens/trainable': 190245, 'epoch': '1.625'}
	81%\|████████▏ \| 416/512 [45:59<10:30, 6.56s/it][2026-08-18 15:03:57,157] [INFO] [axolotl.core.trainers.base._save:828] [PID:12244] Saving model checkpoint to /workspace/wave/training/checkpoints/checkpoint-416
	81%\|████████▏ \| 417/512 [46:08<11:08, 7.04s/it] {'loss': '0.0002913', 'grad_norm': '0.01879', 'learning_rate': '1.836e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.89', 'memory/max_allocated (GiB)': '33.89', 'memory/device_reserved (GiB)': '35.96', 'tokens/train_per_sec_per_gpu': '35.59', 'tokens/total': 12623760, 'tokens/trainable': 190737, 'epoch': '1.629'}
	81%\|████████▏ \| 417/512 [46:08<11:08, 7.04s/it] 82%\|████████▏ \| 418/512 [46:14<10:47, 6.89s/it] {'loss': '0.0005218', 'grad_norm': '0.09822', 'learning_rate': '1.819e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.92', 'memory/max_allocated (GiB)': '33.92', 'memory/device_reserved (GiB)': '35.81', 'tokens/train_per_sec_per_gpu': '31.38', 'tokens/total': 12654272, 'tokens/trainable': 191155, 'epoch': '1.633'}
	82%\|████████▏ \| 418/512 [46:14<10:47, 6.89s/it] 82%\|████████▏ \| 419/512 [46:21<10:31, 6.79s/it] {'loss': '0.0007685', 'grad_norm': '0.07196', 'learning_rate': '1.802e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.89', 'memory/max_allocated (GiB)': '33.89', 'memory/device_reserved (GiB)': '35.81', 'tokens/train_per_sec_per_gpu': '31.98', 'tokens/total': 12684544, 'tokens/trainable': 191577, 'epoch': '1.637'}
	82%\|████████▏ \| 419/512 [46:21<10:31, 6.79s/it] 82%\|████████▏ \| 420/512 [46:27<10:18, 6.73s/it] {'loss': '0.0007165', 'grad_norm': '0.04137', 'learning_rate': '1.786e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '35.81', 'tokens/train_per_sec_per_gpu': '37.25', 'tokens/total': 12715008, 'tokens/trainable': 192051, 'epoch': '1.641'}
	82%\|████████▏ \| 420/512 [46:27<10:18, 6.73s/it] 82%\|████████▏ \| 421/512 [46:34<10:09, 6.69s/it] {'loss': '0.0002489', 'grad_norm': '0.01246', 'learning_rate': '1.77e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '35.81', 'tokens/train_per_sec_per_gpu': '31.77', 'tokens/total': 12745232, 'tokens/trainable': 192500, 'epoch': '1.645'}
	82%\|████████▏ \| 421/512 [46:34<10:09, 6.69s/it] 82%\|████████▏ \| 422/512 [46:40<09:59, 6.66s/it] {'loss': '8.724e-05', 'grad_norm': '0.007973', 'learning_rate': '1.753e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.97', 'memory/max_allocated (GiB)': '33.97', 'memory/device_reserved (GiB)': '35.83', 'tokens/train_per_sec_per_gpu': '33.23', 'tokens/total': 12775824, 'tokens/trainable': 192955, 'epoch': '1.648'}
	82%\|████████▏ \| 422/512 [46:40<09:59, 6.66s/it] 83%\|████████▎ \| 423/512 [46:47<09:52, 6.65s/it] {'loss': '0.0003637', 'grad_norm': '0.02996', 'learning_rate': '1.737e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.92', 'memory/max_allocated (GiB)': '33.92', 'memory/device_reserved (GiB)': '35.83', 'tokens/train_per_sec_per_gpu': '38.03', 'tokens/total': 12806240, 'tokens/trainable': 193462, 'epoch': '1.652'}
	83%\|████████▎ \| 423/512 [46:47<09:52, 6.65s/it] 83%\|████████▎ \| 424/512 [46:54<09:44, 6.64s/it] {'loss': '0.0008654', 'grad_norm': '0.1144', 'learning_rate': '1.722e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '35.83', 'tokens/train_per_sec_per_gpu': '37', 'tokens/total': 12836512, 'tokens/trainable': 193943, 'epoch': '1.656'}
	83%\|████████▎ \| 424/512 [46:54<09:44, 6.64s/it] 83%\|████████▎ \| 425/512 [47:00<09:36, 6.63s/it] {'loss': '0.001933', 'grad_norm': '0.111', 'learning_rate': '1.706e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.79', 'memory/max_allocated (GiB)': '33.79', 'memory/device_reserved (GiB)': '35.83', 'tokens/train_per_sec_per_gpu': '29.75', 'tokens/total': 12866704, 'tokens/trainable': 194360, 'epoch': '1.66'}
	83%\|████████▎ \| 425/512 [47:00<09:36, 6.63s/it] 83%\|████████▎ \| 426/512 [47:07<09:30, 6.63s/it] {'loss': '0.0001203', 'grad_norm': '0.005441', 'learning_rate': '1.69e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.79', 'memory/max_allocated (GiB)': '33.79', 'memory/device_reserved (GiB)': '35.83', 'tokens/train_per_sec_per_gpu': '35.43', 'tokens/total': 12896912, 'tokens/trainable': 194848, 'epoch': '1.664'}
	83%\|████████▎ \| 426/512 [47:07<09:30, 6.63s/it] 83%\|████████▎ \| 427/512 [47:14<09:23, 6.63s/it] {'loss': '0.0001281', 'grad_norm': '0.01091', 'learning_rate': '1.675e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '35.83', 'tokens/train_per_sec_per_gpu': '34.48', 'tokens/total': 12927248, 'tokens/trainable': 195314, 'epoch': '1.668'}
	83%\|████████▎ \| 427/512 [47:14<09:23, 6.63s/it] 84%\|████████▎ \| 428/512 [47:20<09:16, 6.63s/it] {'loss': '9.905e-05', 'grad_norm': '0.007124', 'learning_rate': '1.66e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '35.83', 'tokens/train_per_sec_per_gpu': '32.4', 'tokens/total': 12957632, 'tokens/trainable': 195732, 'epoch': '1.672'}
	84%\|████████▎ \| 428/512 [47:20<09:16, 6.63s/it] 84%\|████████▍ \| 429/512 [47:27<09:08, 6.61s/it] {'loss': '0.002469', 'grad_norm': '0.2362', 'learning_rate': '1.645e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.75', 'memory/max_allocated (GiB)': '33.75', 'memory/device_reserved (GiB)': '35.83', 'tokens/train_per_sec_per_gpu': '33.37', 'tokens/total': 12987824, 'tokens/trainable': 196197, 'epoch': '1.676'}
	84%\|████████▍ \| 429/512 [47:27<09:08, 6.61s/it] 84%\|████████▍ \| 430/512 [47:33<09:02, 6.62s/it] {'loss': '0.005283', 'grad_norm': '0.2918', 'learning_rate': '1.63e-05', 'ppl': '1.005', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '35.83', 'tokens/train_per_sec_per_gpu': '32.85', 'tokens/total': 13018208, 'tokens/trainable': 196678, 'epoch': '1.68'}
	84%\|████████▍ \| 430/512 [47:33<09:02, 6.62s/it] 84%\|████████▍ \| 431/512 [47:40<08:58, 6.65s/it] {'loss': '0.0004377', 'grad_norm': '0.03083', 'learning_rate': '1.615e-05', 'ppl': '1', 'memory/max_active (GiB)': '34.02', 'memory/max_allocated (GiB)': '34.02', 'memory/device_reserved (GiB)': '36.36', 'tokens/train_per_sec_per_gpu': '31.52', 'tokens/total': 13048800, 'tokens/trainable': 197109, 'epoch': '1.684'}
	84%\|████████▍ \| 431/512 [47:40<08:58, 6.65s/it] 84%\|████████▍ \| 432/512 [47:47<08:51, 6.64s/it] {'loss': '0.0002724', 'grad_norm': '0.024', 'learning_rate': '1.6e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.92', 'memory/max_allocated (GiB)': '33.92', 'memory/device_reserved (GiB)': '36.36', 'tokens/train_per_sec_per_gpu': '39.22', 'tokens/total': 13079392, 'tokens/trainable': 197607, 'epoch': '1.688'}
	84%\|████████▍ \| 432/512 [47:47<08:51, 6.64s/it] 85%\|████████▍ \| 433/512 [47:53<08:43, 6.63s/it] {'loss': '0.000624', 'grad_norm': '0.03983', 'learning_rate': '1.586e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.95', 'memory/max_allocated (GiB)': '33.95', 'memory/device_reserved (GiB)': '36.81', 'tokens/train_per_sec_per_gpu': '33.34', 'tokens/total': 13109920, 'tokens/trainable': 198066, 'epoch': '1.691'}
	85%\|████████▍ \| 433/512 [47:53<08:43, 6.63s/it] 85%\|████████▍ \| 434/512 [48:00<08:36, 6.62s/it] {'loss': '0.0005622', 'grad_norm': '0.03856', 'learning_rate': '1.572e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '36.81', 'tokens/train_per_sec_per_gpu': '33.11', 'tokens/total': 13140064, 'tokens/trainable': 198484, 'epoch': '1.695'}
	85%\|████████▍ \| 434/512 [48:00<08:36, 6.62s/it] 85%\|████████▍ \| 435/512 [48:06<08:29, 6.61s/it] {'loss': '0.0008548', 'grad_norm': '0.07454', 'learning_rate': '1.558e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.92', 'memory/max_allocated (GiB)': '33.92', 'memory/device_reserved (GiB)': '36.81', 'tokens/train_per_sec_per_gpu': '35.03', 'tokens/total': 13170656, 'tokens/trainable': 198957, 'epoch': '1.699'}
	85%\|████████▍ \| 435/512 [48:06<08:29, 6.61s/it] 85%\|████████▌ \| 436/512 [48:13<08:22, 6.62s/it] {'loss': '5.212e-05', 'grad_norm': '0.003516', 'learning_rate': '1.544e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.78', 'memory/max_allocated (GiB)': '33.78', 'memory/device_reserved (GiB)': '36.81', 'tokens/train_per_sec_per_gpu': '33.78', 'tokens/total': 13201056, 'tokens/trainable': 199405, 'epoch': '1.703'}
	85%\|████████▌ \| 436/512 [48:13<08:22, 6.62s/it] 85%\|████████▌ \| 437/512 [48:20<08:16, 6.62s/it] {'loss': '0.0004348', 'grad_norm': '0.9569', 'learning_rate': '1.53e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.97', 'memory/max_allocated (GiB)': '33.97', 'memory/device_reserved (GiB)': '36.81', 'tokens/train_per_sec_per_gpu': '34.61', 'tokens/total': 13231584, 'tokens/trainable': 199888, 'epoch': '1.707'}
	85%\|████████▌ \| 437/512 [48:20<08:16, 6.62s/it] 86%\|████████▌ \| 438/512 [48:26<08:10, 6.63s/it] {'loss': '0.0001117', 'grad_norm': '0.01002', 'learning_rate': '1.516e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '36.81', 'tokens/train_per_sec_per_gpu': '35.46', 'tokens/total': 13261888, 'tokens/trainable': 200396, 'epoch': '1.711'}
	86%\|████████▌ \| 438/512 [48:26<08:10, 6.63s/it] 86%\|████████▌ \| 439/512 [48:33<08:02, 6.60s/it] {'loss': '0.0001707', 'grad_norm': '0.008282', 'learning_rate': '1.503e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.77', 'memory/max_allocated (GiB)': '33.77', 'memory/device_reserved (GiB)': '36.81', 'tokens/train_per_sec_per_gpu': '33.34', 'tokens/total': 13291936, 'tokens/trainable': 200843, 'epoch': '1.715'}
	86%\|████████▌ \| 439/512 [48:33<08:02, 6.60s/it] 86%\|████████▌ \| 440/512 [48:40<07:56, 6.61s/it] {'loss': '0.0007505', 'grad_norm': '0.06078', 'learning_rate': '1.49e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '36.81', 'tokens/train_per_sec_per_gpu': '35.57', 'tokens/total': 13322224, 'tokens/trainable': 201315, 'epoch': '1.719'}
	86%\|████████▌ \| 440/512 [48:40<07:56, 6.61s/it] 86%\|████████▌ \| 441/512 [48:46<07:49, 6.61s/it] {'loss': '0.001084', 'grad_norm': '0.1296', 'learning_rate': '1.477e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '36.81', 'tokens/train_per_sec_per_gpu': '34.77', 'tokens/total': 13352640, 'tokens/trainable': 201782, 'epoch': '1.723'}
	86%\|████████▌ \| 441/512 [48:46<07:49, 6.61s/it] 86%\|████████▋ \| 442/512 [48:53<07:41, 6.59s/it] {'loss': '0.0005228', 'grad_norm': '0.0456', 'learning_rate': '1.464e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.62', 'memory/max_allocated (GiB)': '33.62', 'memory/device_reserved (GiB)': '36.81', 'tokens/train_per_sec_per_gpu': '31.27', 'tokens/total': 13382560, 'tokens/trainable': 202222, 'epoch': '1.727'}
	86%\|████████▋ \| 442/512 [48:53<07:41, 6.59s/it] 87%\|████████▋ \| 443/512 [48:59<07:34, 6.59s/it] {'loss': '0.0003615', 'grad_norm': '0.02645', 'learning_rate': '1.451e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.93', 'memory/max_allocated (GiB)': '33.93', 'memory/device_reserved (GiB)': '36.81', 'tokens/train_per_sec_per_gpu': '36.81', 'tokens/total': 13412976, 'tokens/trainable': 202660, 'epoch': '1.73'}
	87%\|████████▋ \| 443/512 [48:59<07:34, 6.59s/it] 87%\|████████▋ \| 444/512 [49:06<07:25, 6.55s/it] {'loss': '0.0002294', 'grad_norm': '0.02189', 'learning_rate': '1.438e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.64', 'memory/max_allocated (GiB)': '33.64', 'memory/device_reserved (GiB)': '36.81', 'tokens/train_per_sec_per_gpu': '31.51', 'tokens/total': 13442720, 'tokens/trainable': 203068, 'epoch': '1.734'}
	87%\|████████▋ \| 444/512 [49:06<07:25, 6.55s/it] 87%\|████████▋ \| 445/512 [49:12<07:18, 6.55s/it] {'loss': '0.001371', 'grad_norm': '0.1399', 'learning_rate': '1.426e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.77', 'memory/max_allocated (GiB)': '33.77', 'memory/device_reserved (GiB)': '36.81', 'tokens/train_per_sec_per_gpu': '34.97', 'tokens/total': 13472928, 'tokens/trainable': 203527, 'epoch': '1.738'}
	87%\|████████▋ \| 445/512 [49:12<07:18, 6.55s/it] 87%\|████████▋ \| 446/512 [49:19<07:13, 6.56s/it] {'loss': '0.0002348', 'grad_norm': '0.02456', 'learning_rate': '1.414e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.81', 'memory/max_allocated (GiB)': '33.81', 'memory/device_reserved (GiB)': '36.81', 'tokens/train_per_sec_per_gpu': '35.94', 'tokens/total': 13502992, 'tokens/trainable': 203994, 'epoch': '1.742'}
	87%\|████████▋ \| 446/512 [49:19<07:13, 6.56s/it] 87%\|████████▋ \| 447/512 [49:25<07:06, 6.56s/it] {'loss': '0.0002258', 'grad_norm': '0.0218', 'learning_rate': '1.402e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '36.81', 'tokens/train_per_sec_per_gpu': '38.06', 'tokens/total': 13533184, 'tokens/trainable': 204430, 'epoch': '1.746'}
	87%\|████████▋ \| 447/512 [49:25<07:06, 6.56s/it] 88%\|████████▊ \| 448/512 [49:32<07:00, 6.58s/it] {'loss': '0.002107', 'grad_norm': '0.2327', 'learning_rate': '1.39e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.96', 'memory/max_allocated (GiB)': '33.96', 'memory/device_reserved (GiB)': '36.81', 'tokens/train_per_sec_per_gpu': '36.07', 'tokens/total': 13563648, 'tokens/trainable': 204886, 'epoch': '1.75'}
	88%\|████████▊ \| 448/512 [49:32<07:00, 6.58s/it][2026-08-18 15:07:29,838] [INFO] [axolotl.core.trainers.base._save:828] [PID:12244] Saving model checkpoint to /workspace/wave/training/checkpoints/checkpoint-448
	88%\|████████▊ \| 449/512 [49:40<07:24, 7.06s/it] {'loss': '0.0001422', 'grad_norm': '0.01578', 'learning_rate': '1.378e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.93', 'memory/max_allocated (GiB)': '33.93', 'memory/device_reserved (GiB)': '36.81', 'tokens/train_per_sec_per_gpu': '34.15', 'tokens/total': 13594032, 'tokens/trainable': 205356, 'epoch': '1.754'}
	88%\|████████▊ \| 449/512 [49:40<07:24, 7.06s/it] 88%\|████████▊ \| 450/512 [49:47<07:09, 6.92s/it] {'loss': '0.001226', 'grad_norm': '0.09758', 'learning_rate': '1.367e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '35.69', 'tokens/train_per_sec_per_gpu': '31.13', 'tokens/total': 13624368, 'tokens/trainable': 205782, 'epoch': '1.758'}
	88%\|████████▊ \| 450/512 [49:47<07:09, 6.92s/it] 88%\|████████▊ \| 451/512 [49:54<06:57, 6.84s/it] {'loss': '0.0006241', 'grad_norm': '0.05972', 'learning_rate': '1.355e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '35.7', 'tokens/train_per_sec_per_gpu': '32.1', 'tokens/total': 13654768, 'tokens/trainable': 206219, 'epoch': '1.762'}
	88%\|████████▊ \| 451/512 [49:54<06:57, 6.84s/it] 88%\|████████▊ \| 452/512 [50:00<06:45, 6.76s/it] {'loss': '5.874e-05', 'grad_norm': '0.009018', 'learning_rate': '1.344e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '35.7', 'tokens/train_per_sec_per_gpu': '36.29', 'tokens/total': 13684960, 'tokens/trainable': 206677, 'epoch': '1.766'}
	88%\|████████▊ \| 452/512 [50:00<06:45, 6.76s/it] 88%\|████████▊ \| 453/512 [50:07<06:37, 6.73s/it] {'loss': '8.521e-05', 'grad_norm': '0.01131', 'learning_rate': '1.333e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.98', 'memory/max_allocated (GiB)': '33.98', 'memory/device_reserved (GiB)': '35.7', 'tokens/train_per_sec_per_gpu': '37.8', 'tokens/total': 13715488, 'tokens/trainable': 207160, 'epoch': '1.77'}
	88%\|████████▊ \| 453/512 [50:07<06:37, 6.73s/it] 89%\|████████▊ \| 454/512 [50:13<06:28, 6.69s/it] {'loss': '3.377e-05', 'grad_norm': '0.002245', 'learning_rate': '1.322e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.86', 'memory/max_allocated (GiB)': '33.86', 'memory/device_reserved (GiB)': '35.7', 'tokens/train_per_sec_per_gpu': '36.25', 'tokens/total': 13745856, 'tokens/trainable': 207605, 'epoch': '1.773'}
	89%\|████████▊ \| 454/512 [50:13<06:28, 6.69s/it] 89%\|████████▉ \| 455/512 [50:20<06:20, 6.67s/it] {'loss': '0.0001957', 'grad_norm': '0.01739', 'learning_rate': '1.311e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '35.78', 'tokens/train_per_sec_per_gpu': '33.43', 'tokens/total': 13776192, 'tokens/trainable': 208047, 'epoch': '1.777'}
	89%\|████████▉ \| 455/512 [50:20<06:20, 6.67s/it] 89%\|████████▉ \| 456/512 [50:27<06:11, 6.64s/it] {'loss': '0.0005775', 'grad_norm': '0.04885', 'learning_rate': '1.301e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '35.78', 'tokens/train_per_sec_per_gpu': '35.44', 'tokens/total': 13806496, 'tokens/trainable': 208476, 'epoch': '1.781'}
	89%\|████████▉ \| 456/512 [50:27<06:11, 6.64s/it] 89%\|████████▉ \| 457/512 [50:33<06:04, 6.63s/it] {'loss': '0.001659', 'grad_norm': '0.3257', 'learning_rate': '1.29e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.89', 'memory/max_allocated (GiB)': '33.89', 'memory/device_reserved (GiB)': '35.82', 'tokens/train_per_sec_per_gpu': '36.03', 'tokens/total': 13836768, 'tokens/trainable': 208970, 'epoch': '1.785'}
	89%\|████████▉ \| 457/512 [50:33<06:04, 6.63s/it] 89%\|████████▉ \| 458/512 [50:40<05:58, 6.63s/it] {'loss': '0.0006973', 'grad_norm': '0.06033', 'learning_rate': '1.28e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.92', 'memory/max_allocated (GiB)': '33.92', 'memory/device_reserved (GiB)': '35.82', 'tokens/train_per_sec_per_gpu': '35.31', 'tokens/total': 13867152, 'tokens/trainable': 209456, 'epoch': '1.789'}
	89%\|████████▉ \| 458/512 [50:40<05:58, 6.63s/it] 90%\|████████▉ \| 459/512 [50:46<05:51, 6.63s/it] {'loss': '3.174e-05', 'grad_norm': '0.00734', 'learning_rate': '1.27e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.8', 'memory/max_allocated (GiB)': '33.8', 'memory/device_reserved (GiB)': '35.82', 'tokens/train_per_sec_per_gpu': '34.15', 'tokens/total': 13897360, 'tokens/trainable': 209907, 'epoch': '1.793'}
	90%\|████████▉ \| 459/512 [50:46<05:51, 6.63s/it] 90%\|████████▉ \| 460/512 [50:53<05:43, 6.62s/it] {'loss': '0.0016', 'grad_norm': '0.1607', 'learning_rate': '1.26e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.91', 'memory/max_allocated (GiB)': '33.91', 'memory/device_reserved (GiB)': '35.82', 'tokens/train_per_sec_per_gpu': '34.16', 'tokens/total': 13927792, 'tokens/trainable': 210366, 'epoch': '1.797'}
	90%\|████████▉ \| 460/512 [50:53<05:43, 6.62s/it] 90%\|█████████ \| 461/512 [51:00<05:36, 6.59s/it] {'loss': '0.00224', 'grad_norm': '0.5972', 'learning_rate': '1.251e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.65', 'memory/max_allocated (GiB)': '33.65', 'memory/device_reserved (GiB)': '35.82', 'tokens/train_per_sec_per_gpu': '32.1', 'tokens/total': 13957856, 'tokens/trainable': 210835, 'epoch': '1.801'}
	90%\|█████████ \| 461/512 [51:00<05:36, 6.59s/it] 90%\|█████████ \| 462/512 [51:06<05:22, 6.44s/it] {'loss': '0.001224', 'grad_norm': '0.1525', 'learning_rate': '1.241e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.38', 'memory/max_allocated (GiB)': '33.38', 'memory/device_reserved (GiB)': '35.82', 'tokens/train_per_sec_per_gpu': '30.91', 'tokens/total': 13986080, 'tokens/trainable': 211259, 'epoch': '1.805'}
	90%\|█████████ \| 462/512 [51:06<05:22, 6.44s/it] 90%\|█████████ \| 463/512 [51:12<05:18, 6.49s/it] {'loss': '3.636e-05', 'grad_norm': '0.005159', 'learning_rate': '1.232e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.79', 'memory/max_allocated (GiB)': '33.79', 'memory/device_reserved (GiB)': '35.82', 'tokens/train_per_sec_per_gpu': '35.65', 'tokens/total': 14016288, 'tokens/trainable': 211721, 'epoch': '1.809'}
	90%\|█████████ \| 463/512 [51:12<05:18, 6.49s/it] 91%\|█████████ \| 464/512 [51:19<05:12, 6.51s/it] {'loss': '0.001997', 'grad_norm': '0.1126', 'learning_rate': '1.223e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.74', 'memory/max_allocated (GiB)': '33.74', 'memory/device_reserved (GiB)': '35.82', 'tokens/train_per_sec_per_gpu': '29.59', 'tokens/total': 14046288, 'tokens/trainable': 212159, 'epoch': '1.812'}
	91%\|█████████ \| 464/512 [51:19<05:12, 6.51s/it] 91%\|█████████ \| 465/512 [51:25<05:07, 6.55s/it] {'loss': '0.0001486', 'grad_norm': '0.02084', 'learning_rate': '1.214e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.83', 'memory/max_allocated (GiB)': '33.83', 'memory/device_reserved (GiB)': '35.82', 'tokens/train_per_sec_per_gpu': '33.17', 'tokens/total': 14076512, 'tokens/trainable': 212644, 'epoch': '1.816'}
	91%\|█████████ \| 465/512 [51:25<05:07, 6.55s/it] 91%\|█████████ \| 466/512 [51:32<05:01, 6.56s/it] {'loss': '2.983e-05', 'grad_norm': '0.001434', 'learning_rate': '1.205e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.83', 'memory/max_allocated (GiB)': '33.83', 'memory/device_reserved (GiB)': '35.82', 'tokens/train_per_sec_per_gpu': '33.38', 'tokens/total': 14106608, 'tokens/trainable': 213084, 'epoch': '1.82'}
	91%\|█████████ \| 466/512 [51:32<05:01, 6.56s/it] 91%\|█████████ \| 467/512 [51:39<04:56, 6.58s/it] {'loss': '0.002034', 'grad_norm': '0.1758', 'learning_rate': '1.197e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.81', 'memory/max_allocated (GiB)': '33.81', 'memory/device_reserved (GiB)': '35.82', 'tokens/train_per_sec_per_gpu': '32.05', 'tokens/total': 14136864, 'tokens/trainable': 213523, 'epoch': '1.824'}
	91%\|█████████ \| 467/512 [51:39<04:56, 6.58s/it] 91%\|█████████▏\| 468/512 [51:45<04:48, 6.57s/it] {'loss': '0.0001363', 'grad_norm': '0.01292', 'learning_rate': '1.188e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.73', 'memory/max_allocated (GiB)': '33.73', 'memory/device_reserved (GiB)': '35.82', 'tokens/train_per_sec_per_gpu': '35.03', 'tokens/total': 14166896, 'tokens/trainable': 213963, 'epoch': '1.828'}
	91%\|█████████▏\| 468/512 [51:45<04:48, 6.57s/it] 92%\|█████████▏\| 469/512 [51:52<04:42, 6.58s/it] {'loss': '0.001088', 'grad_norm': '0.133', 'learning_rate': '1.18e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.97', 'memory/max_allocated (GiB)': '33.97', 'memory/device_reserved (GiB)': '35.82', 'tokens/train_per_sec_per_gpu': '37.4', 'tokens/total': 14197568, 'tokens/trainable': 214431, 'epoch': '1.832'}
	92%\|█████████▏\| 469/512 [51:52<04:42, 6.58s/it] 92%\|█████████▏\| 470/512 [51:58<04:30, 6.45s/it] {'loss': '1.274e-05', 'grad_norm': '0.000542', 'learning_rate': '1.172e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '35.82', 'tokens/train_per_sec_per_gpu': '36.07', 'tokens/total': 14225984, 'tokens/trainable': 214865, 'epoch': '1.836'}
	92%\|█████████▏\| 470/512 [51:58<04:30, 6.45s/it] 92%\|█████████▏\| 471/512 [52:05<04:27, 6.53s/it] {'loss': '0.001536', 'grad_norm': '0.3001', 'learning_rate': '1.164e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '35.92', 'tokens/train_per_sec_per_gpu': '35.98', 'tokens/total': 14256512, 'tokens/trainable': 215358, 'epoch': '1.84'}
	92%\|█████████▏\| 471/512 [52:05<04:27, 6.53s/it] 92%\|█████████▏\| 472/512 [52:11<04:21, 6.55s/it] {'loss': '0.0001523', 'grad_norm': '0.01403', 'learning_rate': '1.156e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.8', 'memory/max_allocated (GiB)': '33.8', 'memory/device_reserved (GiB)': '35.92', 'tokens/train_per_sec_per_gpu': '33.38', 'tokens/total': 14286640, 'tokens/trainable': 215800, 'epoch': '1.844'}
	92%\|█████████▏\| 472/512 [52:11<04:21, 6.55s/it] 92%\|█████████▏\| 473/512 [52:18<04:16, 6.57s/it] {'loss': '0.0005481', 'grad_norm': '0.08513', 'learning_rate': '1.149e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '35.92', 'tokens/train_per_sec_per_gpu': '35.29', 'tokens/total': 14317024, 'tokens/trainable': 216245, 'epoch': '1.848'}
	92%\|█████████▏\| 473/512 [52:18<04:16, 6.57s/it] 93%\|█████████▎\| 474/512 [52:24<04:09, 6.57s/it] {'loss': '4.498e-05', 'grad_norm': '0.004201', 'learning_rate': '1.142e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.76', 'memory/max_allocated (GiB)': '33.76', 'memory/device_reserved (GiB)': '35.92', 'tokens/train_per_sec_per_gpu': '33.83', 'tokens/total': 14347088, 'tokens/trainable': 216685, 'epoch': '1.852'}
	93%\|█████████▎\| 474/512 [52:24<04:09, 6.57s/it] 93%\|█████████▎\| 475/512 [52:31<04:02, 6.55s/it] {'loss': '0.002648', 'grad_norm': '0.4065', 'learning_rate': '1.135e-05', 'ppl': '1.003', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '35.92', 'tokens/train_per_sec_per_gpu': '31.31', 'tokens/total': 14377152, 'tokens/trainable': 217092, 'epoch': '1.855'}
	93%\|█████████▎\| 475/512 [52:31<04:02, 6.55s/it] 93%\|█████████▎\| 476/512 [52:38<03:56, 6.58s/it] {'loss': '1.867e-05', 'grad_norm': '0.001392', 'learning_rate': '1.128e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '35.92', 'tokens/train_per_sec_per_gpu': '38.27', 'tokens/total': 14407568, 'tokens/trainable': 217582, 'epoch': '1.859'}
	93%\|█████████▎\| 476/512 [52:38<03:56, 6.58s/it] 93%\|█████████▎\| 477/512 [52:44<03:50, 6.58s/it] {'loss': '2.959e-05', 'grad_norm': '0.00188', 'learning_rate': '1.121e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.76', 'memory/max_allocated (GiB)': '33.76', 'memory/device_reserved (GiB)': '35.92', 'tokens/train_per_sec_per_gpu': '36.2', 'tokens/total': 14437792, 'tokens/trainable': 218049, 'epoch': '1.863'}
	93%\|█████████▎\| 477/512 [52:44<03:50, 6.58s/it] 93%\|█████████▎\| 478/512 [52:51<03:45, 6.62s/it] {'loss': '0.0007669', 'grad_norm': '0.1085', 'learning_rate': '1.114e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '35.92', 'tokens/train_per_sec_per_gpu': '33.06', 'tokens/total': 14468400, 'tokens/trainable': 218522, 'epoch': '1.867'}
	93%\|█████████▎\| 478/512 [52:51<03:45, 6.62s/it] 94%\|█████████▎\| 479/512 [52:57<03:37, 6.60s/it] {'loss': '0.0007375', 'grad_norm': '0.08652', 'learning_rate': '1.108e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.83', 'memory/max_allocated (GiB)': '33.83', 'memory/device_reserved (GiB)': '35.92', 'tokens/train_per_sec_per_gpu': '33.32', 'tokens/total': 14498416, 'tokens/trainable': 218970, 'epoch': '1.871'}
	94%\|█████████▎\| 479/512 [52:57<03:37, 6.60s/it] 94%\|█████████▍\| 480/512 [53:04<03:31, 6.60s/it] {'loss': '0.002163', 'grad_norm': '0.1909', 'learning_rate': '1.102e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '33.98', 'memory/max_allocated (GiB)': '33.98', 'memory/device_reserved (GiB)': '35.92', 'tokens/train_per_sec_per_gpu': '34.62', 'tokens/total': 14528976, 'tokens/trainable': 219411, 'epoch': '1.875'}
	94%\|█████████▍\| 480/512 [53:04<03:31, 6.60s/it][2026-08-18 15:11:01,803] [INFO] [axolotl.core.trainers.base._save:828] [PID:12244] Saving model checkpoint to /workspace/wave/training/checkpoints/checkpoint-480
	94%\|█████████▍\| 481/512 [53:12<03:39, 7.08s/it] {'loss': '3.28e-05', 'grad_norm': '0.003719', 'learning_rate': '1.096e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.78', 'memory/max_allocated (GiB)': '33.78', 'memory/device_reserved (GiB)': '35.92', 'tokens/train_per_sec_per_gpu': '29.98', 'tokens/total': 14558992, 'tokens/trainable': 219837, 'epoch': '1.879'}
	94%\|█████████▍\| 481/512 [53:12<03:39, 7.08s/it] 94%\|█████████▍\| 482/512 [53:19<03:28, 6.95s/it] {'loss': '2.935e-05', 'grad_norm': '0.002898', 'learning_rate': '1.09e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '35.61', 'tokens/total': 14589488, 'tokens/trainable': 220345, 'epoch': '1.883'}
	94%\|█████████▍\| 482/512 [53:19<03:28, 6.95s/it] 94%\|█████████▍\| 483/512 [53:25<03:18, 6.83s/it] {'loss': '0.0002914', 'grad_norm': '0.03024', 'learning_rate': '1.084e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.8', 'memory/max_allocated (GiB)': '33.8', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '33.58', 'tokens/total': 14619584, 'tokens/trainable': 220775, 'epoch': '1.887'}
	94%\|█████████▍\| 483/512 [53:25<03:18, 6.83s/it] 95%\|█████████▍\| 484/512 [53:32<03:09, 6.76s/it] {'loss': '2.797e-05', 'grad_norm': '0.002516', 'learning_rate': '1.079e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.76', 'memory/max_allocated (GiB)': '33.76', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '35.58', 'tokens/total': 14649760, 'tokens/trainable': 221217, 'epoch': '1.891'}
	95%\|█████████▍\| 484/512 [53:32<03:09, 6.76s/it] 95%\|█████████▍\| 485/512 [53:39<03:00, 6.70s/it] {'loss': '8.626e-05', 'grad_norm': '0.01444', 'learning_rate': '1.073e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.76', 'memory/max_allocated (GiB)': '33.76', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '32.07', 'tokens/total': 14679920, 'tokens/trainable': 221666, 'epoch': '1.895'}
	95%\|█████████▍\| 485/512 [53:39<03:00, 6.70s/it] 95%\|█████████▍\| 486/512 [53:45<02:53, 6.66s/it] {'loss': '0.0001287', 'grad_norm': '0.0164', 'learning_rate': '1.068e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '31.95', 'tokens/total': 14710432, 'tokens/trainable': 222110, 'epoch': '1.898'}
	95%\|█████████▍\| 486/512 [53:45<02:53, 6.66s/it] 95%\|█████████▌\| 487/512 [53:52<02:45, 6.63s/it] {'loss': '0.0001532', 'grad_norm': '0.022', 'learning_rate': '1.063e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.67', 'memory/max_allocated (GiB)': '33.67', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '32.47', 'tokens/total': 14740416, 'tokens/trainable': 222542, 'epoch': '1.902'}
	95%\|█████████▌\| 487/512 [53:52<02:45, 6.63s/it] 95%\|█████████▌\| 488/512 [53:58<02:39, 6.63s/it] {'loss': '1.004e-05', 'grad_norm': '0.00261', 'learning_rate': '1.058e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.97', 'memory/max_allocated (GiB)': '33.97', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '36.81', 'tokens/total': 14770832, 'tokens/trainable': 223035, 'epoch': '1.906'}
	95%\|█████████▌\| 488/512 [53:58<02:39, 6.63s/it] 96%\|█████████▌\| 489/512 [54:05<02:32, 6.61s/it] {'loss': '1.927e-05', 'grad_norm': '0.001653', 'learning_rate': '1.054e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.77', 'memory/max_allocated (GiB)': '33.77', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '32.27', 'tokens/total': 14801104, 'tokens/trainable': 223488, 'epoch': '1.91'}
	96%\|█████████▌\| 489/512 [54:05<02:32, 6.61s/it] 96%\|█████████▌\| 490/512 [54:12<02:25, 6.60s/it] {'loss': '6.305e-06', 'grad_norm': '0.0007156', 'learning_rate': '1.049e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.76', 'memory/max_allocated (GiB)': '33.76', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '29.86', 'tokens/total': 14831216, 'tokens/trainable': 223932, 'epoch': '1.914'}
	96%\|█████████▌\| 490/512 [54:12<02:25, 6.60s/it] 96%\|█████████▌\| 491/512 [54:18<02:18, 6.61s/it] {'loss': '0.0001381', 'grad_norm': '0.01357', 'learning_rate': '1.045e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '33.39', 'tokens/total': 14861488, 'tokens/trainable': 224389, 'epoch': '1.918'}
	96%\|█████████▌\| 491/512 [54:18<02:18, 6.61s/it] 96%\|█████████▌\| 492/512 [54:25<02:12, 6.62s/it] {'loss': '0.001722', 'grad_norm': '0.2159', 'learning_rate': '1.041e-05', 'ppl': '1.002', 'memory/max_active (GiB)': '34.01', 'memory/max_allocated (GiB)': '34.01', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '35.65', 'tokens/total': 14891824, 'tokens/trainable': 224871, 'epoch': '1.922'}
	96%\|█████████▌\| 492/512 [54:25<02:12, 6.62s/it] 96%\|█████████▋\| 493/512 [54:31<02:05, 6.63s/it] {'loss': '0.01609', 'grad_norm': '0.4015', 'learning_rate': '1.037e-05', 'ppl': '1.016', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '35.16', 'tokens/total': 14922192, 'tokens/trainable': 225352, 'epoch': '1.926'}
	96%\|█████████▋\| 493/512 [54:31<02:05, 6.63s/it] 96%\|█████████▋\| 494/512 [54:38<01:59, 6.63s/it] {'loss': '5.178e-05', 'grad_norm': '0.003592', 'learning_rate': '1.034e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.84', 'memory/max_allocated (GiB)': '33.84', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '33.3', 'tokens/total': 14952624, 'tokens/trainable': 225801, 'epoch': '1.93'}
	96%\|█████████▋\| 494/512 [54:38<01:59, 6.63s/it] 97%\|█████████▋\| 495/512 [54:45<01:52, 6.61s/it] {'loss': '1.597e-05', 'grad_norm': '0.001518', 'learning_rate': '1.03e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.81', 'memory/max_allocated (GiB)': '33.81', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '33.05', 'tokens/total': 14982704, 'tokens/trainable': 226240, 'epoch': '1.934'}
	97%\|█████████▋\| 495/512 [54:45<01:52, 6.61s/it] 97%\|█████████▋\| 496/512 [54:51<01:46, 6.63s/it] {'loss': '2.798e-05', 'grad_norm': '0.001617', 'learning_rate': '1.027e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.9', 'memory/max_allocated (GiB)': '33.9', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '35.32', 'tokens/total': 15013024, 'tokens/trainable': 226712, 'epoch': '1.938'}
	97%\|█████████▋\| 496/512 [54:51<01:46, 6.63s/it] 97%\|█████████▋\| 497/512 [54:58<01:39, 6.61s/it] {'loss': '9.649e-05', 'grad_norm': '0.007684', 'learning_rate': '1.024e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.89', 'memory/max_allocated (GiB)': '33.89', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '32.28', 'tokens/total': 15043520, 'tokens/trainable': 227135, 'epoch': '1.941'}
	97%\|█████████▋\| 497/512 [54:58<01:39, 6.61s/it] 97%\|█████████▋\| 498/512 [55:04<01:32, 6.60s/it] {'loss': '3.721e-05', 'grad_norm': '0.002169', 'learning_rate': '1.021e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.87', 'memory/max_allocated (GiB)': '33.87', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '38.25', 'tokens/total': 15073824, 'tokens/trainable': 227600, 'epoch': '1.945'}
	97%\|█████████▋\| 498/512 [55:04<01:32, 6.60s/it] 97%\|█████████▋\| 499/512 [55:11<01:25, 6.61s/it] {'loss': '0.0003017', 'grad_norm': '0.02861', 'learning_rate': '1.018e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '37.21', 'tokens/total': 15104128, 'tokens/trainable': 228076, 'epoch': '1.949'}
	97%\|█████████▋\| 499/512 [55:11<01:25, 6.61s/it] 98%\|█████████▊\| 500/512 [55:18<01:19, 6.60s/it] {'loss': '6.546e-05', 'grad_norm': '0.02024', 'learning_rate': '1.016e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.82', 'memory/max_allocated (GiB)': '33.82', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '31.41', 'tokens/total': 15134480, 'tokens/trainable': 228526, 'epoch': '1.953'}
	98%\|█████████▊\| 500/512 [55:18<01:19, 6.60s/it] 98%\|█████████▊\| 501/512 [55:24<01:12, 6.60s/it] {'loss': '2.901e-05', 'grad_norm': '0.001963', 'learning_rate': '1.013e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.92', 'memory/max_allocated (GiB)': '33.92', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '35.76', 'tokens/total': 15164832, 'tokens/trainable': 229005, 'epoch': '1.957'}
	98%\|█████████▊\| 501/512 [55:24<01:12, 6.60s/it] 98%\|█████████▊\| 502/512 [55:31<01:06, 6.62s/it] {'loss': '0.0003191', 'grad_norm': '0.05688', 'learning_rate': '1.011e-05', 'ppl': '1', 'memory/max_active (GiB)': '34.03', 'memory/max_allocated (GiB)': '34.03', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '31.5', 'tokens/total': 15195616, 'tokens/trainable': 229438, 'epoch': '1.961'}
	98%\|█████████▊\| 502/512 [55:31<01:06, 6.62s/it] 98%\|█████████▊\| 503/512 [55:37<00:59, 6.61s/it] {'loss': '0.0002743', 'grad_norm': '0.04076', 'learning_rate': '1.009e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '33.33', 'tokens/total': 15226096, 'tokens/trainable': 229885, 'epoch': '1.965'}
	98%\|█████████▊\| 503/512 [55:37<00:59, 6.61s/it] 98%\|█████████▊\| 504/512 [55:44<00:52, 6.61s/it] {'loss': '0.0001235', 'grad_norm': '0.01305', 'learning_rate': '1.008e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.96', 'memory/max_allocated (GiB)': '33.96', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '39.53', 'tokens/total': 15256512, 'tokens/trainable': 230389, 'epoch': '1.969'}
	98%\|█████████▊\| 504/512 [55:44<00:52, 6.61s/it] 99%\|█████████▊\| 505/512 [55:51<00:46, 6.62s/it] {'loss': '0.0005182', 'grad_norm': '0.08245', 'learning_rate': '1.006e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '33.97', 'memory/max_allocated (GiB)': '33.97', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '38.08', 'tokens/total': 15287072, 'tokens/trainable': 230885, 'epoch': '1.973'}
	99%\|█████████▊\| 505/512 [55:51<00:46, 6.62s/it] 99%\|█████████▉\| 506/512 [55:57<00:39, 6.61s/it] {'loss': '1.987e-05', 'grad_norm': '0.001345', 'learning_rate': '1.005e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.85', 'memory/max_allocated (GiB)': '33.85', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '34.79', 'tokens/total': 15317296, 'tokens/trainable': 231335, 'epoch': '1.977'}
	99%\|█████████▉\| 506/512 [55:57<00:39, 6.61s/it] 99%\|█████████▉\| 507/512 [56:04<00:33, 6.63s/it] {'loss': '0.0003057', 'grad_norm': '0.04595', 'learning_rate': '1.003e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '36.96', 'tokens/total': 15347904, 'tokens/trainable': 231824, 'epoch': '1.98'}
	99%\|█████████▉\| 507/512 [56:04<00:33, 6.63s/it] 99%\|█████████▉\| 508/512 [56:10<00:25, 6.46s/it] {'loss': '2.902e-05', 'grad_norm': '0.00236', 'learning_rate': '1.002e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.33', 'memory/max_allocated (GiB)': '33.33', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '32.08', 'tokens/total': 15376144, 'tokens/trainable': 232241, 'epoch': '1.984'}
	99%\|█████████▉\| 508/512 [56:10<00:25, 6.46s/it] 99%\|█████████▉\| 509/512 [56:17<00:19, 6.58s/it] {'loss': '0.001196', 'grad_norm': '0.0954', 'learning_rate': '1.001e-05', 'ppl': '1.001', 'memory/max_active (GiB)': '34.02', 'memory/max_allocated (GiB)': '34.02', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '37.83', 'tokens/total': 15406976, 'tokens/trainable': 232742, 'epoch': '1.988'}
	99%\|█████████▉\| 509/512 [56:17<00:19, 6.58s/it] 100%\|█████████▉\| 510/512 [56:24<00:13, 6.60s/it] {'loss': '4.301e-05', 'grad_norm': '0.003428', 'learning_rate': '1.001e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.94', 'memory/max_allocated (GiB)': '33.94', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '36.66', 'tokens/total': 15437344, 'tokens/trainable': 233231, 'epoch': '1.992'}
	100%\|█████████▉\| 510/512 [56:24<00:13, 6.60s/it] 100%\|█████████▉\| 511/512 [56:30<00:06, 6.60s/it] {'loss': '0.0001367', 'grad_norm': '0.01952', 'learning_rate': '1e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.88', 'memory/max_allocated (GiB)': '33.88', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '30.86', 'tokens/total': 15467712, 'tokens/trainable': 233655, 'epoch': '1.996'}
	100%\|█████████▉\| 511/512 [56:30<00:06, 6.60s/it] 100%\|██████████\| 512/512 [56:37<00:00, 6.63s/it] {'loss': '8.935e-05', 'grad_norm': '0.004961', 'learning_rate': '1e-05', 'ppl': '1', 'memory/max_active (GiB)': '33.89', 'memory/max_allocated (GiB)': '33.89', 'memory/device_reserved (GiB)': '36.15', 'tokens/train_per_sec_per_gpu': '36.78', 'tokens/total': 15498240, 'tokens/trainable': 234124, 'epoch': '2'}
	100%\|██████████\| 512/512 [56:37<00:00, 6.63s/it][2026-08-18 15:14:34,655] [INFO] [axolotl.core.trainers.base._save:828] [PID:12244] Saving model checkpoint to /workspace/wave/training/checkpoints/checkpoint-512
	{'train_runtime': '3399', 'train_samples_per_second': '4.82', 'train_steps_per_second': '0.151', 'train_loss': '0.01122', 'memory/max_active (GiB)': '24.25', 'memory/max_allocated (GiB)': '24.25', 'memory/device_reserved (GiB)': '36.15', 'epoch': '2', 'tokens/train_per_sec_per_gpu': '0'}
	100%\|██████████\| 512/512 [56:38<00:00, 6.63s/it] 100%\|██████████\| 512/512 [56:38<00:00, 6.64s/it]
	[2026-08-18 15:14:36,202] [INFO] [axolotl.train.save_trained_model:267] [PID:12244] Training completed! Saving trained model to /workspace/wave/training/checkpoints.
	[2026-08-18 15:14:36,510] [INFO] [axolotl.train.save_trained_model:388] [PID:12244] Model successfully saved to /workspace/wave/training/checkpoints