Training in progress, step 6

adapter_config.json

@@ -1,7 +1,7 @@
 {
   "alpha_pattern": {},
   "auto_mapping": null,
-  "base_model_name_or_path": "/root/highspeedstorage/ft-volume/llama",
+  "base_model_name_or_path": "/root/highspeedstorage/ft-volume/v3-20250124-205715-merged",
   "bias": "none",
   "eva_config": null,
   "exclude_modules": null,
@@ -23,13 +23,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "down_proj",
-    "q_proj",
-    "up_proj",
-    "o_proj",
+    "k_proj",
     "gate_proj",
+    "up_proj",
+    "down_proj",
     "v_proj",
-    "k_proj"
+    "q_proj",
+    "o_proj"
   ],
   "task_type": "CAUSAL_LM",
   "use_dora": false,

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4b0ba879241d961eb8e67dc6251fc9367217c80e78cf3940807d84e4c56d82af
+oid sha256:755497d3e79899a1080a522d163d8e27692bd91bc89410f83e45e4f8fa1f795c
 size 6627156248

args.json

@@ -1,5 +1,5 @@
 {
-  "model": "/root/highspeedstorage/ft-volume/llama",
+  "model": "/root/highspeedstorage/ft-volume/v3-20250124-205715-merged",
   "model_type": "llama3_2",
   "model_revision": null,
   "task_type": "causal_lm",
@@ -21,7 +21,7 @@
   "use_chat_template": true,
   "template_backend": "swift",
   "dataset": [
-    "llama_data.jsonl"
+    "dpo_data.jsonl"
   ],
   "val_dataset": [],
   "split_dataset_ratio": 0.01,
@@ -48,7 +48,7 @@
   "bnb_4bit_use_double_quant": true,
   "bnb_4bit_quant_storage": null,
   "max_new_tokens": 64,
-  "temperature": 0.0,
+  "temperature": 0.7,
   "top_k": null,
   "top_p": null,
   "repetition_penalty": null,
@@ -57,7 +57,7 @@
   "stop_words": [],
   "logprobs": false,
   "top_logprobs": null,
-  "ckpt_dir": null,
+  "ckpt_dir": "/root/highspeedstorage/ft-volume/v3-20250124-205715-merged",
   "load_dataset_config": null,
   "lora_modules": [],
   "tuner_backend": "peft",
@@ -72,7 +72,7 @@
   "custom_register_path": [],
   "ignore_args_error": false,
   "use_swift_lora": false,
-  "output_dir": "/root/dataDisk/llama-new/output/v3-20250124-205715",
+  "output_dir": "/root/dataDisk/llama-new/output/v8-20250126-122947",
   "overwrite_output_dir": false,
   "do_train": false,
   "do_eval": false,
@@ -102,7 +102,7 @@
   "log_level": "passive",
   "log_level_replica": "warning",
   "log_on_each_node": true,
-  "logging_dir": "/root/dataDisk/llama-new/output/v3-20250124-205715/runs",
+  "logging_dir": "/root/dataDisk/llama-new/output/v8-20250126-122947/runs",
   "logging_strategy": "steps",
   "logging_first_step": true,
   "logging_steps": 1,
@@ -333,13 +333,41 @@
   "optimizer": null,
   "metric": null,
   "acc_strategy": "token",
+  "reward_model": null,
+  "reward_adapters": [],
+  "reward_model_type": null,
+  "reward_model_revision": null,
+  "num_ppo_epochs": 4,
+  "whiten_rewards": false,
+  "kl_coef": 0.05,
+  "cliprange": 0.2,
+  "vf_coef": 0.1,
+  "cliprange_value": 0.2,
+  "gamma": 1.0,
+  "lam": 0.95,
+  "num_mini_batches": 1,
+  "local_rollout_forward_batch_size": 64,
+  "num_sample_generations": 10,
+  "response_length": 512,
+  "missing_eos_penalty": null,
+  "rlhf_type": "orpo",
+  "ref_model": null,
+  "ref_model_type": null,
+  "ref_model_revision": null,
+  "beta": 0.1,
+  "label_smoothing": 0,
+  "rpo_alpha": 1.0,
+  "cpo_alpha": 1.0,
+  "simpo_gamma": 1,
+  "desirable_weight": 1.0,
+  "undesirable_weight": 1.0,
   "rank": 0,
   "global_world_size": 8,
   "local_world_size": 8,
-  "model_suffix": "llama",
-  "model_info": "ModelInfo(model_type='llama3_2', model_dir='/root/highspeedstorage/ft-volume/llama', torch_dtype=torch.bfloat16, max_model_len=8192, quant_method=None, quant_bits=None, config=None, task_type='causal_lm', num_labels=None)",
-  "model_meta": "ModelMeta(model_type='llama3_2', model_groups=[ModelGroup(models=[Model(ms_model_id='LLM-Research/Llama-3.2-1B', hf_model_id='meta-llama/Llama-3.2-1B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='LLM-Research/Llama-3.2-3B', hf_model_id='meta-llama/Llama-3.2-3B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='LLM-Research/Llama-3.2-1B-Instruct', hf_model_id='meta-llama/Llama-3.2-1B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='LLM-Research/Llama-3.2-3B-Instruct', hf_model_id='meta-llama/Llama-3.2-3B-Instruct', model_path=None, ms_revision=None, hf_revision=None)], ignore_patterns=None, requires=None, tags=[]), ModelGroup(models=[Model(ms_model_id='LLM-Research/Llama-3.3-70B-Instruct', hf_model_id='meta-llama/Llama-3.3-70B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='unsloth/Llama-3.3-70B-Instruct-bnb-4bit', hf_model_id='unsloth/Llama-3.3-70B-Instruct-bnb-4bit', model_path=None, ms_revision=None, hf_revision=None)], ignore_patterns=None, requires=None, tags=[])], template='llama3_2', get_function=<function get_model_tokenizer_with_flash_attn at 0x7f5a1d2fb2e0>, model_arch='llama', architectures=['LlamaForCausalLM'], additional_saved_files=[], torch_dtype=None, is_multimodal=False, is_reward=False, task_type=None, ignore_patterns=[], requires=['transformers>=4.45'], tags=[])",
-  "model_dir": "/root/highspeedstorage/ft-volume/llama",
+  "model_suffix": "v3-20250124-205715-merged",
+  "model_info": "ModelInfo(model_type='llama3_2', model_dir='/root/highspeedstorage/ft-volume/v3-20250124-205715-merged', torch_dtype=torch.bfloat16, max_model_len=8192, quant_method=None, quant_bits=None, config=None, task_type='causal_lm', num_labels=None)",
+  "model_meta": "ModelMeta(model_type='llama3_2', model_groups=[ModelGroup(models=[Model(ms_model_id='LLM-Research/Llama-3.2-1B', hf_model_id='meta-llama/Llama-3.2-1B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='LLM-Research/Llama-3.2-3B', hf_model_id='meta-llama/Llama-3.2-3B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='LLM-Research/Llama-3.2-1B-Instruct', hf_model_id='meta-llama/Llama-3.2-1B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='LLM-Research/Llama-3.2-3B-Instruct', hf_model_id='meta-llama/Llama-3.2-3B-Instruct', model_path=None, ms_revision=None, hf_revision=None)], ignore_patterns=None, requires=None, tags=[]), ModelGroup(models=[Model(ms_model_id='LLM-Research/Llama-3.3-70B-Instruct', hf_model_id='meta-llama/Llama-3.3-70B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='unsloth/Llama-3.3-70B-Instruct-bnb-4bit', hf_model_id='unsloth/Llama-3.3-70B-Instruct-bnb-4bit', model_path=None, ms_revision=None, hf_revision=None)], ignore_patterns=None, requires=None, tags=[])], template='llama3_2', get_function=<function get_model_tokenizer_with_flash_attn at 0x7fec77a7b2e0>, model_arch='llama', architectures=['LlamaForCausalLM'], additional_saved_files=[], torch_dtype=None, is_multimodal=False, is_reward=False, task_type=None, ignore_patterns=[], requires=['transformers>=4.45'], tags=[])",
+  "model_dir": "/root/highspeedstorage/ft-volume/v3-20250124-205715-merged",
   "hub": "<class 'swift.hub.hub.HFHub'>",
-  "training_args": "Seq2SeqTrainingArguments(output_dir='/root/dataDisk/llama-new/output/v3-20250124-205715', overwrite_output_dir=False, do_train=False, do_eval=True, do_predict=False, eval_strategy=<IntervalStrategy.STEPS: 'steps'>, prediction_loss_only=False, per_device_train_batch_size=1, per_device_eval_batch_size=1, per_gpu_train_batch_size=None, per_gpu_eval_batch_size=None, gradient_accumulation_steps=16, eval_accumulation_steps=None, eval_delay=0, torch_empty_cache_steps=None, learning_rate=5e-05, weight_decay=0.01, adam_beta1=0.9, adam_beta2=0.999, adam_epsilon=1e-08, max_grad_norm=1.0, num_train_epochs=3.0, max_steps=-1, lr_scheduler_type=<SchedulerType.LINEAR: 'linear'>, lr_scheduler_kwargs=None, warmup_ratio=0.1, warmup_steps=0, log_level='passive', log_level_replica='warning', log_on_each_node=True, logging_dir='/root/dataDisk/llama-new/output/v3-20250124-205715/runs', logging_strategy=<IntervalStrategy.STEPS: 'steps'>, logging_first_step=True, logging_steps=1, logging_nan_inf_filter=True, save_strategy=<SaveStrategy.STEPS: 'steps'>, save_steps=100, save_total_limit=2, save_safetensors=True, save_on_each_node=False, save_only_model=False, restore_callback_states_from_checkpoint=False, no_cuda=False, use_cpu=False, use_mps_device=False, seed=42, data_seed=42, jit_mode_eval=False, use_ipex=False, bf16=True, fp16=False, fp16_opt_level='O1', half_precision_backend='auto', bf16_full_eval=False, fp16_full_eval=False, tf32=None, local_rank=0, ddp_backend=None, tpu_num_cores=None, tpu_metrics_debug=False, debug=[], dataloader_drop_last=False, eval_steps=100, dataloader_num_workers=0, dataloader_prefetch_factor=None, past_index=-1, run_name='/root/dataDisk/llama-new/output/v3-20250124-205715', disable_tqdm=False, remove_unused_columns=False, label_names=None, load_best_model_at_end=False, metric_for_best_model='loss', greater_is_better=False, ignore_data_skip=False, fsdp=[], fsdp_min_num_params=0, fsdp_config={'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, fsdp_transformer_layer_cls_to_wrap=None, accelerator_config=AcceleratorConfig(split_batches=False, dispatch_batches=False, even_batches=True, use_seedable_sampler=True, non_blocking=False, gradient_accumulation_kwargs=None, use_configured_state=False), deepspeed={'fp16': {'enabled': 'auto', 'loss_scale': 0, 'loss_scale_window': 1000, 'initial_scale_power': 16, 'hysteresis': 2, 'min_loss_scale': 1}, 'bf16': {'enabled': 'auto'}, 'zero_optimization': {'stage': 3, 'offload_optimizer': {'device': 'cpu', 'pin_memory': True}, 'offload_param': {'device': 'cpu', 'pin_memory': True}, 'overlap_comm': True, 'contiguous_gradients': True, 'sub_group_size': 1000000000.0, 'reduce_bucket_size': 'auto', 'stage3_prefetch_bucket_size': 'auto', 'stage3_param_persistence_threshold': 'auto', 'stage3_max_live_parameters': 1000000000.0, 'stage3_max_reuse_distance': 1000000000.0, 'stage3_gather_16bit_weights_on_model_save': True}, 'gradient_accumulation_steps': 'auto', 'gradient_clipping': 'auto', 'steps_per_print': 2000, 'train_batch_size': 'auto', 'train_micro_batch_size_per_gpu': 'auto', 'wall_clock_breakdown': False}, label_smoothing_factor=0.0, optim=<OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, optim_args=None, adafactor=False, group_by_length=False, length_column_name='length', report_to=['wandb'], ddp_find_unused_parameters=None, ddp_bucket_cap_mb=None, ddp_broadcast_buffers=None, dataloader_pin_memory=True, dataloader_persistent_workers=False, skip_memory_metrics=True, use_legacy_prediction_loop=False, push_to_hub=True, resume_from_checkpoint=None, hub_model_id='TheAgenticAI/LLAMA-3.3-70B-Reasoning', hub_strategy=<HubStrategy.EVERY_SAVE: 'every_save'>, hub_token=None, hub_private_repo=True, hub_always_push=False, gradient_checkpointing=True, gradient_checkpointing_kwargs={'use_reentrant': True}, include_inputs_for_metrics=False, include_for_metrics=[], eval_do_concat_batches=True, fp16_backend='auto', evaluation_strategy='steps', push_to_hub_model_id=None, push_to_hub_organization=None, push_to_hub_token=None, mp_parameters='', auto_find_batch_size=False, full_determinism=False, torchdynamo=None, ray_scope='last', ddp_timeout=1800, torch_compile=False, torch_compile_backend=None, torch_compile_mode=None, dispatch_batches=None, split_batches=None, include_tokens_per_second=None, include_num_input_tokens_seen=None, neftune_noise_alpha=None, optim_target_modules=None, batch_eval_metrics=False, eval_on_start=False, use_liger_kernel=False, eval_use_gather_object=False, average_tokens_across_devices=None, sortish_sampler=False, predict_with_generate=False, generation_max_length=None, generation_num_beams=None, generation_config=None, acc_strategy='token', sequence_parallel_size=1, check_model=True, train_sampler_random=True, is_encoder_decoder=False, metric_warmup_step=0, train_dataset_sample=-1, fsdp_num=1, acc_steps=1, train_type='lora', optimizer=None, galore_config=None)"
+  "training_args": "ORPOConfig(output_dir='/root/dataDisk/llama-new/output/v8-20250126-122947', overwrite_output_dir=False, do_train=False, do_eval=True, do_predict=False, eval_strategy=<IntervalStrategy.STEPS: 'steps'>, prediction_loss_only=False, per_device_train_batch_size=1, per_device_eval_batch_size=1, per_gpu_train_batch_size=None, per_gpu_eval_batch_size=None, gradient_accumulation_steps=16, eval_accumulation_steps=None, eval_delay=0, torch_empty_cache_steps=None, learning_rate=5e-05, weight_decay=0.01, adam_beta1=0.9, adam_beta2=0.999, adam_epsilon=1e-08, max_grad_norm=1.0, num_train_epochs=3.0, max_steps=-1, lr_scheduler_type=<SchedulerType.LINEAR: 'linear'>, lr_scheduler_kwargs=None, warmup_ratio=0.1, warmup_steps=0, log_level='passive', log_level_replica='warning', log_on_each_node=True, logging_dir='/root/dataDisk/llama-new/output/v8-20250126-122947/runs', logging_strategy=<IntervalStrategy.STEPS: 'steps'>, logging_first_step=True, logging_steps=1, logging_nan_inf_filter=True, save_strategy=<SaveStrategy.STEPS: 'steps'>, save_steps=100, save_total_limit=2, save_safetensors=True, save_on_each_node=False, save_only_model=False, restore_callback_states_from_checkpoint=False, no_cuda=False, use_cpu=False, use_mps_device=False, seed=42, data_seed=42, jit_mode_eval=False, use_ipex=False, bf16=True, fp16=False, fp16_opt_level='O1', half_precision_backend='auto', bf16_full_eval=False, fp16_full_eval=False, tf32=None, local_rank=0, ddp_backend=None, tpu_num_cores=None, tpu_metrics_debug=False, debug=[], dataloader_drop_last=False, eval_steps=100, dataloader_num_workers=0, dataloader_prefetch_factor=None, past_index=-1, run_name='/root/dataDisk/llama-new/output/v8-20250126-122947', disable_tqdm=False, remove_unused_columns=False, label_names=None, load_best_model_at_end=False, metric_for_best_model='loss', greater_is_better=False, ignore_data_skip=False, fsdp=[], fsdp_min_num_params=0, fsdp_config={'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, fsdp_transformer_layer_cls_to_wrap=None, accelerator_config=AcceleratorConfig(split_batches=False, dispatch_batches=False, even_batches=True, use_seedable_sampler=True, non_blocking=False, gradient_accumulation_kwargs=None, use_configured_state=False), deepspeed={'fp16': {'enabled': 'auto', 'loss_scale': 0, 'loss_scale_window': 1000, 'initial_scale_power': 16, 'hysteresis': 2, 'min_loss_scale': 1}, 'bf16': {'enabled': 'auto'}, 'zero_optimization': {'stage': 3, 'offload_optimizer': {'device': 'cpu', 'pin_memory': True}, 'offload_param': {'device': 'cpu', 'pin_memory': True}, 'overlap_comm': True, 'contiguous_gradients': True, 'sub_group_size': 1000000000.0, 'reduce_bucket_size': 'auto', 'stage3_prefetch_bucket_size': 'auto', 'stage3_param_persistence_threshold': 'auto', 'stage3_max_live_parameters': 1000000000.0, 'stage3_max_reuse_distance': 1000000000.0, 'stage3_gather_16bit_weights_on_model_save': True}, 'gradient_accumulation_steps': 'auto', 'gradient_clipping': 'auto', 'steps_per_print': 2000, 'train_batch_size': 'auto', 'train_micro_batch_size_per_gpu': 'auto', 'wall_clock_breakdown': False}, label_smoothing_factor=0.0, optim=<OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, optim_args=None, adafactor=False, group_by_length=False, length_column_name='length', report_to=['wandb'], ddp_find_unused_parameters=None, ddp_bucket_cap_mb=None, ddp_broadcast_buffers=None, dataloader_pin_memory=True, dataloader_persistent_workers=False, skip_memory_metrics=True, use_legacy_prediction_loop=False, push_to_hub=True, resume_from_checkpoint=None, hub_model_id='TheAgenticAI/LLAMA-3.3-70B-Reasoning', hub_strategy=<HubStrategy.EVERY_SAVE: 'every_save'>, hub_token=None, hub_private_repo=True, hub_always_push=False, gradient_checkpointing=True, gradient_checkpointing_kwargs={'use_reentrant': True}, include_inputs_for_metrics=False, include_for_metrics=[], eval_do_concat_batches=True, fp16_backend='auto', evaluation_strategy='steps', push_to_hub_model_id=None, push_to_hub_organization=None, push_to_hub_token=None, mp_parameters='', auto_find_batch_size=False, full_determinism=False, torchdynamo=None, ray_scope='last', ddp_timeout=1800, torch_compile=False, torch_compile_backend=None, torch_compile_mode=None, dispatch_batches=None, split_batches=None, include_tokens_per_second=None, include_num_input_tokens_seen=None, neftune_noise_alpha=None, optim_target_modules=None, batch_eval_metrics=False, eval_on_start=False, use_liger_kernel=False, eval_use_gather_object=False, average_tokens_across_devices=None, max_length=22000, max_prompt_length=None, max_completion_length=None, beta=0.1, disable_dropout=True, label_pad_token_id=None, padding_value=None, truncation_mode='keep_end', generate_during_eval=False, is_encoder_decoder=False, model_init_kwargs=None, dataset_num_proc=1, acc_strategy='token', sequence_parallel_size=1, check_model=True, train_sampler_random=True, metric_warmup_step=0, train_dataset_sample=-1, fsdp_num=1, acc_steps=1, train_type='lora', optimizer=None, galore_config=None)"
 }

logging.jsonl

@@ -1,103 +1,8 @@
-{"loss": 0.74616313, "token_acc": 0.84351861, "grad_norm": 0.09043229, "learning_rate": 1.72e-06, "memory(GiB)": 75.89, "train_speed(iter/s)": 0.003054, "epoch": 0.01032258, "global_step/max_steps": "1/288", "percentage": "0.35%", "elapsed_time": "5m 11s", "remaining_time": "1d 0h 50m 2s"}
-{"loss": 0.78952539, "token_acc": 0.78098911, "grad_norm": 0.10398929, "learning_rate": 3.45e-06, "memory(GiB)": 75.89, "train_speed(iter/s)": 0.003215, "epoch": 0.02064516, "global_step/max_steps": "2/288", "percentage": "0.69%", "elapsed_time": "10m 6s", "remaining_time": "1d 0h 4m 37s"}
-{"loss": 0.82183272, "token_acc": 0.79650972, "grad_norm": 0.11484293, "learning_rate": 5.17e-06, "memory(GiB)": 75.89, "train_speed(iter/s)": 0.003281, "epoch": 0.03096774, "global_step/max_steps": "3/288", "percentage": "1.04%", "elapsed_time": "14m 58s", "remaining_time": "23h 42m 19s"}
-{"loss": 0.89525408, "token_acc": 0.79723175, "grad_norm": 0.15229513, "learning_rate": 6.9e-06, "memory(GiB)": 75.89, "train_speed(iter/s)": 0.003227, "epoch": 0.04129032, "global_step/max_steps": "4/288", "percentage": "1.39%", "elapsed_time": "20m 23s", "remaining_time": "1d 0h 7m 51s"}
-{"loss": 0.87880141, "token_acc": 0.83182664, "grad_norm": 0.10113768, "learning_rate": 8.62e-06, "memory(GiB)": 100.51, "train_speed(iter/s)": 0.003425, "epoch": 0.0516129, "global_step/max_steps": "5/288", "percentage": "1.74%", "elapsed_time": "24m 3s", "remaining_time": "22h 41m 53s"}
-{"loss": 0.78032863, "token_acc": 0.78393572, "grad_norm": 0.07195292, "learning_rate": 1.034e-05, "memory(GiB)": 100.51, "train_speed(iter/s)": 0.00377, "epoch": 0.06193548, "global_step/max_steps": "6/288", "percentage": "2.08%", "elapsed_time": "26m 15s", "remaining_time": "20h 34m 2s"}
-{"loss": 0.81582403, "token_acc": 0.80598127, "grad_norm": 0.03998778, "learning_rate": 1.207e-05, "memory(GiB)": 100.51, "train_speed(iter/s)": 0.004052, "epoch": 0.07225806, "global_step/max_steps": "7/288", "percentage": "2.43%", "elapsed_time": "28m 31s", "remaining_time": "19h 5m 13s"}
-{"loss": 0.70669603, "token_acc": 0.8440445, "grad_norm": 0.0443429, "learning_rate": 1.379e-05, "memory(GiB)": 100.51, "train_speed(iter/s)": 0.004283, "epoch": 0.08258065, "global_step/max_steps": "8/288", "percentage": "2.78%", "elapsed_time": "30m 52s", "remaining_time": "18h 0m 22s"}
-{"loss": 0.7262044, "token_acc": 0.81845238, "grad_norm": 0.04136974, "learning_rate": 1.552e-05, "memory(GiB)": 100.51, "train_speed(iter/s)": 0.004467, "epoch": 0.09290323, "global_step/max_steps": "9/288", "percentage": "3.12%", "elapsed_time": "33m 18s", "remaining_time": "17h 12m 39s"}
-{"loss": 0.87036347, "token_acc": 0.84759051, "grad_norm": 0.05472838, "learning_rate": 1.724e-05, "memory(GiB)": 100.51, "train_speed(iter/s)": 0.004617, "epoch": 0.10322581, "global_step/max_steps": "10/288", "percentage": "3.47%", "elapsed_time": "35m 49s", "remaining_time": "16h 36m 4s"}
-{"loss": 0.60506105, "token_acc": 0.84473466, "grad_norm": 0.03589521, "learning_rate": 1.897e-05, "memory(GiB)": 100.51, "train_speed(iter/s)": 0.004782, "epoch": 0.11354839, "global_step/max_steps": "11/288", "percentage": "3.82%", "elapsed_time": "38m 4s", "remaining_time": "15h 58m 48s"}
-{"loss": 0.83035898, "token_acc": 0.84033392, "grad_norm": 0.03401323, "learning_rate": 2.069e-05, "memory(GiB)": 100.51, "train_speed(iter/s)": 0.004884, "epoch": 0.12387097, "global_step/max_steps": "12/288", "percentage": "4.17%", "elapsed_time": "40m 40s", "remaining_time": "15h 35m 38s"}
-{"loss": 0.56105936, "token_acc": 0.83656801, "grad_norm": 0.02542925, "learning_rate": 2.241e-05, "memory(GiB)": 100.51, "train_speed(iter/s)": 0.005025, "epoch": 0.13419355, "global_step/max_steps": "13/288", "percentage": "4.51%", "elapsed_time": "42m 51s", "remaining_time": "15h 6m 27s"}
-{"loss": 0.69628751, "token_acc": 0.85242543, "grad_norm": 0.02137301, "learning_rate": 2.414e-05, "memory(GiB)": 113.25, "train_speed(iter/s)": 0.005121, "epoch": 0.14451613, "global_step/max_steps": "14/288", "percentage": "4.86%", "elapsed_time": "45m 17s", "remaining_time": "14h 46m 28s"}
-{"loss": 0.5410732, "token_acc": 0.834718, "grad_norm": 0.02185575, "learning_rate": 2.586e-05, "memory(GiB)": 113.25, "train_speed(iter/s)": 0.005231, "epoch": 0.15483871, "global_step/max_steps": "15/288", "percentage": "5.21%", "elapsed_time": "47m 31s", "remaining_time": "14h 24m 54s"}
-{"loss": 0.57584238, "token_acc": 0.83049947, "grad_norm": 0.01980691, "learning_rate": 2.759e-05, "memory(GiB)": 113.25, "train_speed(iter/s)": 0.005352, "epoch": 0.16516129, "global_step/max_steps": "16/288", "percentage": "5.56%", "elapsed_time": "49m 33s", "remaining_time": "14h 2m 27s"}
-{"loss": 0.56113476, "token_acc": 0.84900414, "grad_norm": 0.02572861, "learning_rate": 2.931e-05, "memory(GiB)": 113.25, "train_speed(iter/s)": 0.005441, "epoch": 0.17548387, "global_step/max_steps": "17/288", "percentage": "5.90%", "elapsed_time": "51m 48s", "remaining_time": "13h 45m 52s"}
-{"loss": 0.51780677, "token_acc": 0.86407566, "grad_norm": 0.01697748, "learning_rate": 3.103e-05, "memory(GiB)": 113.25, "train_speed(iter/s)": 0.005525, "epoch": 0.18580645, "global_step/max_steps": "18/288", "percentage": "6.25%", "elapsed_time": "54m 1s", "remaining_time": "13h 30m 25s"}
-{"loss": 0.5272339, "token_acc": 0.84709974, "grad_norm": 0.02022088, "learning_rate": 3.276e-05, "memory(GiB)": 113.25, "train_speed(iter/s)": 0.005605, "epoch": 0.19612903, "global_step/max_steps": "19/288", "percentage": "6.60%", "elapsed_time": "56m 14s", "remaining_time": "13h 16m 10s"}
-{"loss": 0.50596046, "token_acc": 0.82722892, "grad_norm": 0.01607172, "learning_rate": 3.448e-05, "memory(GiB)": 113.25, "train_speed(iter/s)": 0.005684, "epoch": 0.20645161, "global_step/max_steps": "20/288", "percentage": "6.94%", "elapsed_time": "58m 22s", "remaining_time": "13h 2m 11s"}
-{"loss": 0.49655235, "token_acc": 0.8496895, "grad_norm": 0.01419209, "learning_rate": 3.621e-05, "memory(GiB)": 113.25, "train_speed(iter/s)": 0.00576, "epoch": 0.21677419, "global_step/max_steps": "21/288", "percentage": "7.29%", "elapsed_time": "1h 0m 29s", "remaining_time": "12h 49m 9s"}
-{"loss": 0.63461268, "token_acc": 0.86161907, "grad_norm": 0.02277333, "learning_rate": 3.793e-05, "memory(GiB)": 113.25, "train_speed(iter/s)": 0.005796, "epoch": 0.22709677, "global_step/max_steps": "22/288", "percentage": "7.64%", "elapsed_time": "1h 2m 59s", "remaining_time": "12h 41m 42s"}
-{"loss": 0.49595475, "token_acc": 0.84451534, "grad_norm": 0.01650185, "learning_rate": 3.966e-05, "memory(GiB)": 113.25, "train_speed(iter/s)": 0.005856, "epoch": 0.23741935, "global_step/max_steps": "23/288", "percentage": "7.99%", "elapsed_time": "1h 5m 11s", "remaining_time": "12h 31m 7s"}
-{"loss": 0.58476871, "token_acc": 0.86130925, "grad_norm": 0.02474417, "learning_rate": 4.138e-05, "memory(GiB)": 128.5, "train_speed(iter/s)": 0.005889, "epoch": 0.24774194, "global_step/max_steps": "24/288", "percentage": "8.33%", "elapsed_time": "1h 7m 39s", "remaining_time": "12h 24m 14s"}
-{"loss": 0.46254018, "token_acc": 0.87392386, "grad_norm": 0.01164004, "learning_rate": 4.31e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.005934, "epoch": 0.25806452, "global_step/max_steps": "25/288", "percentage": "8.68%", "elapsed_time": "1h 9m 57s", "remaining_time": "12h 15m 54s"}
-{"loss": 0.4681581, "token_acc": 0.85397297, "grad_norm": 0.01212974, "learning_rate": 4.483e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.005981, "epoch": 0.2683871, "global_step/max_steps": "26/288", "percentage": "9.03%", "elapsed_time": "1h 12m 11s", "remaining_time": "12h 7m 25s"}
-{"loss": 0.47845387, "token_acc": 0.84917576, "grad_norm": 0.01104543, "learning_rate": 4.655e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006041, "epoch": 0.27870968, "global_step/max_steps": "27/288", "percentage": "9.38%", "elapsed_time": "1h 14m 13s", "remaining_time": "11h 57m 31s"}
-{"loss": 0.47385406, "token_acc": 0.86413749, "grad_norm": 0.01086575, "learning_rate": 4.828e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006094, "epoch": 0.28903226, "global_step/max_steps": "28/288", "percentage": "9.72%", "elapsed_time": "1h 16m 18s", "remaining_time": "11h 48m 38s"}
-{"loss": 0.47718155, "token_acc": 0.8669849, "grad_norm": 0.01727354, "learning_rate": 5e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006098, "epoch": 0.29935484, "global_step/max_steps": "29/288", "percentage": "10.07%", "elapsed_time": "1h 18m 59s", "remaining_time": "11h 45m 32s"}
-{"loss": 0.45292401, "token_acc": 0.84559537, "grad_norm": 0.04902146, "learning_rate": 4.981e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.00614, "epoch": 0.30967742, "global_step/max_steps": "30/288", "percentage": "10.42%", "elapsed_time": "1h 21m 10s", "remaining_time": "11h 38m 3s"}
-{"loss": 0.43841478, "token_acc": 0.86037951, "grad_norm": 0.01185949, "learning_rate": 4.961e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006182, "epoch": 0.32, "global_step/max_steps": "31/288", "percentage": "10.76%", "elapsed_time": "1h 23m 18s", "remaining_time": "11h 30m 42s"}
-{"loss": 0.51614481, "token_acc": 0.86514002, "grad_norm": 0.03404158, "learning_rate": 4.942e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006206, "epoch": 0.33032258, "global_step/max_steps": "32/288", "percentage": "11.11%", "elapsed_time": "1h 25m 40s", "remaining_time": "11h 25m 24s"}
-{"loss": 0.4681246, "token_acc": 0.82303734, "grad_norm": 0.01642565, "learning_rate": 4.923e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006237, "epoch": 0.34064516, "global_step/max_steps": "33/288", "percentage": "11.46%", "elapsed_time": "1h 27m 54s", "remaining_time": "11h 19m 21s"}
-{"loss": 0.42925426, "token_acc": 0.87939367, "grad_norm": 0.00986654, "learning_rate": 4.903e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006271, "epoch": 0.35096774, "global_step/max_steps": "34/288", "percentage": "11.81%", "elapsed_time": "1h 30m 5s", "remaining_time": "11h 13m 5s"}
-{"loss": 0.61444378, "token_acc": 0.80021854, "grad_norm": 0.0488668, "learning_rate": 4.884e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.00628, "epoch": 0.36129032, "global_step/max_steps": "35/288", "percentage": "12.15%", "elapsed_time": "1h 32m 37s", "remaining_time": "11h 9m 32s"}
-{"loss": 0.49848884, "token_acc": 0.87356891, "grad_norm": 0.02112174, "learning_rate": 4.865e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.00629, "epoch": 0.3716129, "global_step/max_steps": "36/288", "percentage": "12.50%", "elapsed_time": "1h 35m 7s", "remaining_time": "11h 5m 52s"}
-{"loss": 0.44967794, "token_acc": 0.86236844, "grad_norm": 0.00945088, "learning_rate": 4.846e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006325, "epoch": 0.38193548, "global_step/max_steps": "37/288", "percentage": "12.85%", "elapsed_time": "1h 37m 13s", "remaining_time": "10h 59m 36s"}
-{"loss": 0.42617065, "token_acc": 0.87972228, "grad_norm": 0.00959377, "learning_rate": 4.826e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006351, "epoch": 0.39225806, "global_step/max_steps": "38/288", "percentage": "13.19%", "elapsed_time": "1h 39m 27s", "remaining_time": "10h 54m 16s"}
-{"loss": 0.4227379, "token_acc": 0.88012363, "grad_norm": 0.01300816, "learning_rate": 4.807e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006377, "epoch": 0.40258065, "global_step/max_steps": "39/288", "percentage": "13.54%", "elapsed_time": "1h 41m 39s", "remaining_time": "10h 49m 2s"}
-{"loss": 0.41896996, "token_acc": 0.84762152, "grad_norm": 0.01550369, "learning_rate": 4.788e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006409, "epoch": 0.41290323, "global_step/max_steps": "40/288", "percentage": "13.89%", "elapsed_time": "1h 43m 45s", "remaining_time": "10h 43m 17s"}
-{"loss": 0.44419405, "token_acc": 0.81527302, "grad_norm": 0.01633696, "learning_rate": 4.768e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.00643, "epoch": 0.42322581, "global_step/max_steps": "41/288", "percentage": "14.24%", "elapsed_time": "1h 46m 0s", "remaining_time": "10h 38m 37s"}
-{"loss": 0.4448446, "token_acc": 0.84361949, "grad_norm": 0.00948541, "learning_rate": 4.749e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006465, "epoch": 0.43354839, "global_step/max_steps": "42/288", "percentage": "14.58%", "elapsed_time": "1h 48m 0s", "remaining_time": "10h 32m 36s"}
-{"loss": 0.4308551, "token_acc": 0.85782495, "grad_norm": 0.12180801, "learning_rate": 4.73e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006475, "epoch": 0.44387097, "global_step/max_steps": "43/288", "percentage": "14.93%", "elapsed_time": "1h 50m 24s", "remaining_time": "10h 29m 5s"}
-{"loss": 0.43652752, "token_acc": 0.84613588, "grad_norm": 0.00945682, "learning_rate": 4.71e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006501, "epoch": 0.45419355, "global_step/max_steps": "44/288", "percentage": "15.28%", "elapsed_time": "1h 52m 32s", "remaining_time": "10h 24m 4s"}
-{"loss": 0.42303067, "token_acc": 0.88438384, "grad_norm": 0.00908574, "learning_rate": 4.691e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.00652, "epoch": 0.46451613, "global_step/max_steps": "45/288", "percentage": "15.62%", "elapsed_time": "1h 54m 46s", "remaining_time": "10h 19m 46s"}
-{"loss": 0.40082443, "token_acc": 0.87893522, "grad_norm": 0.01320151, "learning_rate": 4.672e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006535, "epoch": 0.47483871, "global_step/max_steps": "46/288", "percentage": "15.97%", "elapsed_time": "1h 57m 2s", "remaining_time": "10h 15m 44s"}
-{"loss": 0.39771432, "token_acc": 0.87401729, "grad_norm": 0.01309315, "learning_rate": 4.653e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.00655, "epoch": 0.48516129, "global_step/max_steps": "47/288", "percentage": "16.32%", "elapsed_time": "1h 59m 19s", "remaining_time": "10h 11m 50s"}
-{"loss": 0.43932644, "token_acc": 0.86438174, "grad_norm": 0.01308558, "learning_rate": 4.633e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006571, "epoch": 0.49548387, "global_step/max_steps": "48/288", "percentage": "16.67%", "elapsed_time": "2h 1m 29s", "remaining_time": "10h 7m 25s"}
-{"loss": 0.40402907, "token_acc": 0.87178259, "grad_norm": 0.01584367, "learning_rate": 4.614e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006585, "epoch": 0.50580645, "global_step/max_steps": "49/288", "percentage": "17.01%", "elapsed_time": "2h 3m 44s", "remaining_time": "10h 3m 34s"}
-{"loss": 0.42721552, "token_acc": 0.87538548, "grad_norm": 0.01127676, "learning_rate": 4.595e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006605, "epoch": 0.51612903, "global_step/max_steps": "50/288", "percentage": "17.36%", "elapsed_time": "2h 5m 54s", "remaining_time": "9h 59m 17s"}
-{"loss": 0.54179633, "token_acc": 0.84423901, "grad_norm": 0.03042504, "learning_rate": 4.575e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.0066, "epoch": 0.52645161, "global_step/max_steps": "51/288", "percentage": "17.71%", "elapsed_time": "2h 8m 31s", "remaining_time": "9h 57m 16s"}
-{"loss": 0.40824139, "token_acc": 0.85981165, "grad_norm": 0.01089707, "learning_rate": 4.556e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006621, "epoch": 0.53677419, "global_step/max_steps": "52/288", "percentage": "18.06%", "elapsed_time": "2h 10m 38s", "remaining_time": "9h 52m 52s"}
-{"loss": 0.5188899, "token_acc": 0.86626307, "grad_norm": 0.02277813, "learning_rate": 4.537e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006632, "epoch": 0.54709677, "global_step/max_steps": "53/288", "percentage": "18.40%", "elapsed_time": "2h 12m 55s", "remaining_time": "9h 49m 21s"}
-{"loss": 0.49796051, "token_acc": 0.87458007, "grad_norm": 0.10761227, "learning_rate": 4.517e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.00665, "epoch": 0.55741935, "global_step/max_steps": "54/288", "percentage": "18.75%", "elapsed_time": "2h 15m 4s", "remaining_time": "9h 45m 18s"}
-{"loss": 0.55097961, "token_acc": 0.7219632, "grad_norm": 0.0301041, "learning_rate": 4.498e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006655, "epoch": 0.56774194, "global_step/max_steps": "55/288", "percentage": "19.10%", "elapsed_time": "2h 17m 28s", "remaining_time": "9h 42m 22s"}
-{"loss": 0.40282875, "token_acc": 0.88147197, "grad_norm": 0.02567812, "learning_rate": 4.479e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006667, "epoch": 0.57806452, "global_step/max_steps": "56/288", "percentage": "19.44%", "elapsed_time": "2h 19m 43s", "remaining_time": "9h 38m 53s"}
-{"loss": 0.4145909, "token_acc": 0.86279797, "grad_norm": 0.09259639, "learning_rate": 4.459e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006681, "epoch": 0.5883871, "global_step/max_steps": "57/288", "percentage": "19.79%", "elapsed_time": "2h 21m 56s", "remaining_time": "9h 35m 13s"}
-{"loss": 0.49579778, "token_acc": 0.86323816, "grad_norm": 0.2940706, "learning_rate": 4.44e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006688, "epoch": 0.59870968, "global_step/max_steps": "58/288", "percentage": "20.14%", "elapsed_time": "2h 24m 16s", "remaining_time": "9h 32m 7s"}
-{"loss": 0.40309823, "token_acc": 0.88168864, "grad_norm": 0.35733849, "learning_rate": 4.421e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006701, "epoch": 0.60903226, "global_step/max_steps": "59/288", "percentage": "20.49%", "elapsed_time": "2h 26m 28s", "remaining_time": "9h 28m 31s"}
-{"loss": 0.42548287, "token_acc": 0.85303217, "grad_norm": 0.971573, "learning_rate": 4.402e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006715, "epoch": 0.61935484, "global_step/max_steps": "60/288", "percentage": "20.83%", "elapsed_time": "2h 28m 38s", "remaining_time": "9h 24m 50s"}
-{"loss": 0.43157196, "token_acc": 0.86270378, "grad_norm": 0.97569734, "learning_rate": 4.382e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006734, "epoch": 0.62967742, "global_step/max_steps": "61/288", "percentage": "21.18%", "elapsed_time": "2h 30m 42s", "remaining_time": "9h 20m 48s"}
-{"loss": 0.40518153, "token_acc": 0.88739643, "grad_norm": 0.02967451, "learning_rate": 4.363e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006748, "epoch": 0.64, "global_step/max_steps": "62/288", "percentage": "21.53%", "elapsed_time": "2h 32m 52s", "remaining_time": "9h 17m 15s"}
-{"loss": 0.4438884, "token_acc": 0.87470964, "grad_norm": 0.01226234, "learning_rate": 4.344e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006765, "epoch": 0.65032258, "global_step/max_steps": "63/288", "percentage": "21.88%", "elapsed_time": "2h 34m 57s", "remaining_time": "9h 13m 23s"}
-{"loss": 0.39373183, "token_acc": 0.88465558, "grad_norm": 0.05350703, "learning_rate": 4.324e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006778, "epoch": 0.66064516, "global_step/max_steps": "64/288", "percentage": "22.22%", "elapsed_time": "2h 37m 6s", "remaining_time": "9h 9m 53s"}
-{"loss": 0.44090027, "token_acc": 0.87650124, "grad_norm": 0.06398247, "learning_rate": 4.305e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006791, "epoch": 0.67096774, "global_step/max_steps": "65/288", "percentage": "22.57%", "elapsed_time": "2h 39m 15s", "remaining_time": "9h 6m 23s"}
-{"loss": 0.47568282, "token_acc": 0.88244228, "grad_norm": 0.07284541, "learning_rate": 4.286e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006791, "epoch": 0.68129032, "global_step/max_steps": "66/288", "percentage": "22.92%", "elapsed_time": "2h 41m 42s", "remaining_time": "9h 3m 54s"}
-{"loss": 0.40231276, "token_acc": 0.859897, "grad_norm": 0.02990863, "learning_rate": 4.266e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006801, "epoch": 0.6916129, "global_step/max_steps": "67/288", "percentage": "23.26%", "elapsed_time": "2h 43m 55s", "remaining_time": "9h 0m 43s"}
-{"loss": 0.45850387, "token_acc": 0.86981746, "grad_norm": 0.07461943, "learning_rate": 4.247e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006808, "epoch": 0.70193548, "global_step/max_steps": "68/288", "percentage": "23.61%", "elapsed_time": "2h 46m 12s", "remaining_time": "8h 57m 43s"}
-{"loss": 0.42772788, "token_acc": 0.86058025, "grad_norm": 0.01338425, "learning_rate": 4.228e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.00682, "epoch": 0.71225806, "global_step/max_steps": "69/288", "percentage": "23.96%", "elapsed_time": "2h 48m 20s", "remaining_time": "8h 54m 18s"}
-{"loss": 0.50105357, "token_acc": 0.88452936, "grad_norm": 0.08125398, "learning_rate": 4.208e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006823, "epoch": 0.72258065, "global_step/max_steps": "70/288", "percentage": "24.31%", "elapsed_time": "2h 50m 42s", "remaining_time": "8h 51m 38s"}
-{"loss": 0.38152462, "token_acc": 0.8830297, "grad_norm": 0.01115579, "learning_rate": 4.189e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006831, "epoch": 0.73290323, "global_step/max_steps": "71/288", "percentage": "24.65%", "elapsed_time": "2h 52m 57s", "remaining_time": "8h 48m 37s"}
-{"loss": 0.39672104, "token_acc": 0.88033904, "grad_norm": 0.01423557, "learning_rate": 4.17e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006843, "epoch": 0.74322581, "global_step/max_steps": "72/288", "percentage": "25.00%", "elapsed_time": "2h 55m 6s", "remaining_time": "8h 45m 18s"}
-{"loss": 0.43814272, "token_acc": 0.88551196, "grad_norm": 0.14627557, "learning_rate": 4.151e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006843, "epoch": 0.75354839, "global_step/max_steps": "73/288", "percentage": "25.35%", "elapsed_time": "2h 57m 32s", "remaining_time": "8h 42m 53s"}
-{"loss": 0.3858158, "token_acc": 0.86370734, "grad_norm": 0.05014987, "learning_rate": 4.131e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006845, "epoch": 0.76387097, "global_step/max_steps": "74/288", "percentage": "25.69%", "elapsed_time": "2h 59m 55s", "remaining_time": "8h 40m 18s"}
-{"loss": 0.37708718, "token_acc": 0.89819705, "grad_norm": 0.08282039, "learning_rate": 4.112e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006855, "epoch": 0.77419355, "global_step/max_steps": "75/288", "percentage": "26.04%", "elapsed_time": "3h 2m 5s", "remaining_time": "8h 37m 8s"}
-{"loss": 0.40146717, "token_acc": 0.87541194, "grad_norm": 0.03130351, "learning_rate": 4.093e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.00686, "epoch": 0.78451613, "global_step/max_steps": "76/288", "percentage": "26.39%", "elapsed_time": "3h 4m 23s", "remaining_time": "8h 34m 20s"}
-{"loss": 0.45210558, "token_acc": 0.86619006, "grad_norm": 0.24303684, "learning_rate": 4.073e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006858, "epoch": 0.79483871, "global_step/max_steps": "77/288", "percentage": "26.74%", "elapsed_time": "3h 6m 52s", "remaining_time": "8h 32m 4s"}
-{"loss": 0.42958081, "token_acc": 0.85775103, "grad_norm": 0.01276872, "learning_rate": 4.054e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006869, "epoch": 0.80516129, "global_step/max_steps": "78/288", "percentage": "27.08%", "elapsed_time": "3h 8m 59s", "remaining_time": "8h 28m 50s"}
-{"loss": 0.40100986, "token_acc": 0.8705993, "grad_norm": 0.0461642, "learning_rate": 4.035e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006862, "epoch": 0.81548387, "global_step/max_steps": "79/288", "percentage": "27.43%", "elapsed_time": "3h 11m 36s", "remaining_time": "8h 26m 53s"}
-{"loss": 0.38652623, "token_acc": 0.89047065, "grad_norm": 0.07440139, "learning_rate": 4.015e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006869, "epoch": 0.82580645, "global_step/max_steps": "80/288", "percentage": "27.78%", "elapsed_time": "3h 13m 49s", "remaining_time": "8h 23m 57s"}
-{"loss": 0.38682115, "token_acc": 0.87585879, "grad_norm": 0.01817736, "learning_rate": 3.996e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006875, "epoch": 0.83612903, "global_step/max_steps": "81/288", "percentage": "28.12%", "elapsed_time": "3h 16m 6s", "remaining_time": "8h 21m 10s"}
-{"loss": 0.41781539, "token_acc": 0.85694046, "grad_norm": 0.02927447, "learning_rate": 3.977e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.00688, "epoch": 0.84645161, "global_step/max_steps": "82/288", "percentage": "28.47%", "elapsed_time": "3h 18m 22s", "remaining_time": "8h 18m 20s"}
-{"loss": 0.39182213, "token_acc": 0.89268926, "grad_norm": 0.06937289, "learning_rate": 3.958e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006885, "epoch": 0.85677419, "global_step/max_steps": "83/288", "percentage": "28.82%", "elapsed_time": "3h 20m 38s", "remaining_time": "8h 15m 34s"}
-{"loss": 0.37565386, "token_acc": 0.88184552, "grad_norm": 0.05194683, "learning_rate": 3.938e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006893, "epoch": 0.86709677, "global_step/max_steps": "84/288", "percentage": "29.17%", "elapsed_time": "3h 22m 49s", "remaining_time": "8h 12m 34s"}
-{"loss": 0.45567641, "token_acc": 0.86954447, "grad_norm": 0.05576006, "learning_rate": 3.919e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006891, "epoch": 0.87741935, "global_step/max_steps": "85/288", "percentage": "29.51%", "elapsed_time": "3h 25m 18s", "remaining_time": "8h 10m 19s"}
-{"loss": 0.37166446, "token_acc": 0.89650233, "grad_norm": 0.03371212, "learning_rate": 3.9e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006895, "epoch": 0.88774194, "global_step/max_steps": "86/288", "percentage": "29.86%", "elapsed_time": "3h 27m 36s", "remaining_time": "8h 7m 39s"}
-{"loss": 0.41649836, "token_acc": 0.87910183, "grad_norm": 0.01462127, "learning_rate": 3.88e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006898, "epoch": 0.89806452, "global_step/max_steps": "87/288", "percentage": "30.21%", "elapsed_time": "3h 29m 56s", "remaining_time": "8h 5m 3s"}
-{"loss": 0.41595942, "token_acc": 0.85043611, "grad_norm": 0.01152844, "learning_rate": 3.861e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006908, "epoch": 0.9083871, "global_step/max_steps": "88/288", "percentage": "30.56%", "elapsed_time": "3h 32m 3s", "remaining_time": "8h 1m 56s"}
-{"loss": 0.38049665, "token_acc": 0.88799895, "grad_norm": 0.00966426, "learning_rate": 3.842e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006917, "epoch": 0.91870968, "global_step/max_steps": "89/288", "percentage": "30.90%", "elapsed_time": "3h 34m 10s", "remaining_time": "7h 58m 53s"}
-{"loss": 0.40713, "token_acc": 0.8783668, "grad_norm": 0.01316505, "learning_rate": 3.822e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006923, "epoch": 0.92903226, "global_step/max_steps": "90/288", "percentage": "31.25%", "elapsed_time": "3h 36m 25s", "remaining_time": "7h 56m 7s"}
-{"loss": 0.4486534, "token_acc": 0.85522959, "grad_norm": 0.04914835, "learning_rate": 3.803e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006923, "epoch": 0.93935484, "global_step/max_steps": "91/288", "percentage": "31.60%", "elapsed_time": "3h 38m 49s", "remaining_time": "7h 53m 43s"}
-{"loss": 0.37252715, "token_acc": 0.8659446, "grad_norm": 0.00998468, "learning_rate": 3.784e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006929, "epoch": 0.94967742, "global_step/max_steps": "92/288", "percentage": "31.94%", "elapsed_time": "3h 41m 0s", "remaining_time": "7h 50m 51s"}
-{"loss": 0.39902872, "token_acc": 0.87282829, "grad_norm": 0.01568112, "learning_rate": 3.764e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006928, "epoch": 0.96, "global_step/max_steps": "93/288", "percentage": "32.29%", "elapsed_time": "3h 43m 28s", "remaining_time": "7h 48m 33s"}
-{"loss": 0.36277354, "token_acc": 0.88438775, "grad_norm": 0.0237332, "learning_rate": 3.745e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006933, "epoch": 0.97032258, "global_step/max_steps": "94/288", "percentage": "32.64%", "elapsed_time": "3h 45m 42s", "remaining_time": "7h 45m 48s"}
-{"loss": 0.38434136, "token_acc": 0.8888924, "grad_norm": 0.00939997, "learning_rate": 3.726e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006939, "epoch": 0.98064516, "global_step/max_steps": "95/288", "percentage": "32.99%", "elapsed_time": "3h 47m 54s", "remaining_time": "7h 43m 0s"}
-{"loss": 0.36436903, "token_acc": 0.89673349, "grad_norm": 0.01110194, "learning_rate": 3.707e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006944, "epoch": 0.99096774, "global_step/max_steps": "96/288", "percentage": "33.33%", "elapsed_time": "3h 50m 8s", "remaining_time": "7h 40m 16s"}
-{"loss": 0.78716338, "token_acc": 0.87718253, "grad_norm": 0.01389996, "learning_rate": 3.687e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006908, "epoch": 1.01032258, "global_step/max_steps": "97/288", "percentage": "33.68%", "elapsed_time": "3h 53m 45s", "remaining_time": "7h 40m 16s"}
-{"loss": 0.40994811, "token_acc": 0.82488286, "grad_norm": 0.18944687, "learning_rate": 3.668e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006908, "epoch": 1.02064516, "global_step/max_steps": "98/288", "percentage": "34.03%", "elapsed_time": "3h 56m 9s", "remaining_time": "7h 37m 51s"}
-{"loss": 0.36000898, "token_acc": 0.90903877, "grad_norm": 0.0123647, "learning_rate": 3.649e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006912, "epoch": 1.03096774, "global_step/max_steps": "99/288", "percentage": "34.38%", "elapsed_time": "3h 58m 26s", "remaining_time": "7h 35m 11s"}
-{"loss": 0.38061467, "token_acc": 0.89940414, "grad_norm": 0.01069106, "learning_rate": 3.629e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006917, "epoch": 1.04129032, "global_step/max_steps": "100/288", "percentage": "34.72%", "elapsed_time": "4h 0m 42s", "remaining_time": "7h 32m 31s"}
-{"eval_loss": 0.39166829, "eval_token_acc": 0.88091295, "eval_runtime": 48.768, "eval_samples_per_second": 2.563, "eval_steps_per_second": 0.328, "epoch": 1.04129032, "global_step/max_steps": "100/288", "percentage": "34.72%", "elapsed_time": "4h 1m 30s", "remaining_time": "7h 34m 2s"}
-{"loss": 0.37255841, "token_acc": 0.88312756, "grad_norm": 0.01173662, "learning_rate": 3.61e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006886, "epoch": 1.0516129, "global_step/max_steps": "101/288", "percentage": "35.07%", "elapsed_time": "4h 4m 10s", "remaining_time": "7h 32m 5s"}
-{"loss": 0.36340833, "token_acc": 0.87858518, "grad_norm": 0.01821016, "learning_rate": 3.591e-05, "memory(GiB)": 130.45, "train_speed(iter/s)": 0.006891, "epoch": 1.06193548, "global_step/max_steps": "102/288", "percentage": "35.42%", "elapsed_time": "4h 6m 25s", "remaining_time": "7h 29m 21s"}
+{"loss": 0.96603394, "grad_norm": 0.09358346, "learning_rate": 5e-05, "memory(GiB)": 98.18, "train_speed(iter/s)": 0.00256, "rewards/chosen": -0.08843994, "rewards/rejected": -0.03957367, "rewards/accuracies": 0.0, "rewards/margins": -0.04885864, "logps/rejected": -0.3956604, "logps/chosen": -0.88378906, "logits/rejected": -0.18650818, "logits/chosen": -0.44784546, "nll_loss": 0.88244629, "log_odds_ratio": -1.39306641, "log_odds_chosen": -1.03417969, "epoch": 0.45714286, "global_step/max_steps": "1/6", "percentage": "16.67%", "elapsed_time": "6m 15s", "remaining_time": "31m 19s"}
+{"loss": 1.07710266, "grad_norm": 0.07397991, "learning_rate": 4e-05, "memory(GiB)": 120.78, "train_speed(iter/s)": 0.00248, "rewards/chosen": -0.10545349, "rewards/rejected": -0.04308319, "rewards/accuracies": 0.0625, "rewards/margins": -0.06238556, "logps/rejected": -0.4307251, "logps/chosen": -1.0546875, "logits/rejected": -0.21848679, "logits/chosen": -0.51165771, "nll_loss": 1.05371094, "log_odds_ratio": -1.47875977, "log_odds_chosen": -1.08789062, "epoch": 0.91428571, "global_step/max_steps": "2/6", "percentage": "33.33%", "elapsed_time": "13m 11s", "remaining_time": "26m 23s"}
+{"loss": 0.95780945, "grad_norm": 0.10560518, "learning_rate": 3e-05, "memory(GiB)": 120.78, "train_speed(iter/s)": 0.002382, "rewards/chosen": -0.06476508, "rewards/rejected": -0.04706489, "rewards/accuracies": 0.16666667, "rewards/margins": -0.0177002, "logps/rejected": -0.47081164, "logps/chosen": -0.64735246, "logits/rejected": -0.47363281, "logits/chosen": -0.76638454, "nll_loss": 0.64691842, "log_odds_ratio": -0.9453125, "log_odds_chosen": -0.42122397, "epoch": 1.45714286, "global_step/max_steps": "3/6", "percentage": "50.00%", "elapsed_time": "20m 44s", "remaining_time": "20m 44s"}
+{"loss": 0.9289856, "grad_norm": 0.03600949, "learning_rate": 2e-05, "memory(GiB)": 120.78, "train_speed(iter/s)": 0.002416, "rewards/chosen": -0.08105469, "rewards/rejected": -0.04862213, "rewards/accuracies": 0.0625, "rewards/margins": -0.03243256, "logps/rejected": -0.48632812, "logps/chosen": -0.81054688, "logits/rejected": -0.47827148, "logits/chosen": -0.81054688, "nll_loss": 0.81054688, "log_odds_ratio": -1.11035156, "log_odds_chosen": -0.64709473, "epoch": 1.91428571, "global_step/max_steps": "4/6", "percentage": "66.67%", "elapsed_time": "27m 20s", "remaining_time": "13m 40s"}
+{"loss": 0.94372559, "grad_norm": 0.03234995, "learning_rate": 1e-05, "memory(GiB)": 120.78, "train_speed(iter/s)": 0.002419, "rewards/chosen": -0.06709798, "rewards/rejected": -0.04815674, "rewards/accuracies": 0.16666667, "rewards/margins": -0.01894803, "logps/rejected": -0.48166233, "logps/chosen": -0.67122394, "logits/rejected": -0.45073783, "logits/chosen": -0.82421875, "nll_loss": 0.6715495, "log_odds_ratio": -0.96115452, "log_odds_chosen": -0.44466147, "epoch": 2.45714286, "global_step/max_steps": "5/6", "percentage": "83.33%", "elapsed_time": "34m 12s", "remaining_time": "6m 50s"}
+{"loss": 0.78184509, "grad_norm": 0.02222378, "learning_rate": 0.0, "memory(GiB)": 120.78, "train_speed(iter/s)": 0.002446, "rewards/chosen": -0.07427979, "rewards/rejected": -0.04429626, "rewards/accuracies": 0.0625, "rewards/margins": -0.0300293, "logps/rejected": -0.44299316, "logps/chosen": -0.74291992, "logits/rejected": -0.43395996, "logits/chosen": -0.81304932, "nll_loss": 0.74328613, "log_odds_ratio": -1.11694336, "log_odds_chosen": -0.65979004, "epoch": 2.91428571, "global_step/max_steps": "6/6", "percentage": "100.00%", "elapsed_time": "40m 38s", "remaining_time": "0s"}
+{"eval_loss": 1.1171875, "eval_runtime": 5.8635, "eval_samples_per_second": 0.341, "eval_steps_per_second": 0.171, "eval_rewards/chosen": -0.06933594, "eval_rewards/rejected": -0.04882812, "eval_rewards/accuracies": 0.0, "eval_rewards/margins": -0.02050781, "eval_logps/rejected": -0.48828125, "eval_logps/chosen": -0.69140625, "eval_logits/rejected": -0.58203125, "eval_logits/chosen": -0.59375, "eval_nll_loss": 0.6875, "eval_log_odds_ratio": -0.94921875, "eval_log_odds_chosen": -0.4609375, "epoch": 2.91428571, "global_step/max_steps": "6/6", "percentage": "100.00%", "elapsed_time": "40m 43s", "remaining_time": "0s"}
+{"train_runtime": 2471.1137, "train_samples_per_second": 0.333, "train_steps_per_second": 0.002, "total_flos": 65359310749696.0, "train_loss": 0.94258372, "epoch": 2.91428571, "global_step/max_steps": "6/6", "percentage": "100.00%", "elapsed_time": "41m 9s", "remaining_time": "0s"}

special_tokens_map.json

@@ -13,5 +13,11 @@
     "rstrip": false,
     "single_word": false
   },
-  "pad_token": "<|eot_id|>"
+  "pad_token": {
+    "content": "<|eot_id|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:62aec4a16ad7784426afd2d42eaebf68758b687450726ad42691b8511c06d4e7
-size 8184
+oid sha256:555a621d1a7a55ef965b07972fdea651660831674732ccb543abe2d3845575a0
+size 8312