Upload Qwen3-ASR-WSM-Min checkpoint-16000
Browse files- .gitattributes +1 -0
- Qwen3-ASR-WSM-Min/added_tokens.json +64 -0
- Qwen3-ASR-WSM-Min/args.json +404 -0
- Qwen3-ASR-WSM-Min/chat_template.jinja +31 -0
- Qwen3-ASR-WSM-Min/config.json +228 -0
- Qwen3-ASR-WSM-Min/generation_config.json +9 -0
- Qwen3-ASR-WSM-Min/global_step16000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +3 -0
- Qwen3-ASR-WSM-Min/global_step16000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt +3 -0
- Qwen3-ASR-WSM-Min/global_step16000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt +3 -0
- Qwen3-ASR-WSM-Min/global_step16000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt +3 -0
- Qwen3-ASR-WSM-Min/global_step16000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt +3 -0
- Qwen3-ASR-WSM-Min/global_step16000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt +3 -0
- Qwen3-ASR-WSM-Min/global_step16000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt +3 -0
- Qwen3-ASR-WSM-Min/global_step16000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt +3 -0
- Qwen3-ASR-WSM-Min/global_step16000/mp_rank_00_model_states.pt +3 -0
- Qwen3-ASR-WSM-Min/latest +1 -0
- Qwen3-ASR-WSM-Min/merges.txt +0 -0
- Qwen3-ASR-WSM-Min/model.safetensors +3 -0
- Qwen3-ASR-WSM-Min/preprocessor_config.json +14 -0
- Qwen3-ASR-WSM-Min/rng_state_0.pth +3 -0
- Qwen3-ASR-WSM-Min/rng_state_1.pth +3 -0
- Qwen3-ASR-WSM-Min/rng_state_2.pth +3 -0
- Qwen3-ASR-WSM-Min/rng_state_3.pth +3 -0
- Qwen3-ASR-WSM-Min/rng_state_4.pth +3 -0
- Qwen3-ASR-WSM-Min/rng_state_5.pth +3 -0
- Qwen3-ASR-WSM-Min/rng_state_6.pth +3 -0
- Qwen3-ASR-WSM-Min/rng_state_7.pth +3 -0
- Qwen3-ASR-WSM-Min/scheduler.pt +3 -0
- Qwen3-ASR-WSM-Min/special_tokens_map.json +44 -0
- Qwen3-ASR-WSM-Min/tokenizer.json +3 -0
- Qwen3-ASR-WSM-Min/tokenizer_config.json +549 -0
- Qwen3-ASR-WSM-Min/trainer_state.json +2042 -0
- Qwen3-ASR-WSM-Min/training_args.bin +3 -0
- Qwen3-ASR-WSM-Min/vocab.json +0 -0
- Qwen3-ASR-WSM-Min/zero_to_fp32.py +760 -0
.gitattributes
CHANGED
|
@@ -34,3 +34,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 36 |
Qwen3-ASR-WSM-Mandarin/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 36 |
Qwen3-ASR-WSM-Mandarin/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 37 |
+
Qwen3-ASR-WSM-Min/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
Qwen3-ASR-WSM-Min/added_tokens.json
ADDED
|
@@ -0,0 +1,64 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"</think>": 151668,
|
| 3 |
+
"</tool_call>": 151658,
|
| 4 |
+
"</tool_response>": 151666,
|
| 5 |
+
"<asr_text>": 151704,
|
| 6 |
+
"<blank10>": 151686,
|
| 7 |
+
"<blank11>": 151687,
|
| 8 |
+
"<blank12>": 151688,
|
| 9 |
+
"<blank13>": 151689,
|
| 10 |
+
"<blank14>": 151690,
|
| 11 |
+
"<blank15>": 151691,
|
| 12 |
+
"<blank16>": 151692,
|
| 13 |
+
"<blank17>": 151693,
|
| 14 |
+
"<blank18>": 151694,
|
| 15 |
+
"<blank19>": 151695,
|
| 16 |
+
"<blank1>": 151677,
|
| 17 |
+
"<blank20>": 151696,
|
| 18 |
+
"<blank21>": 151697,
|
| 19 |
+
"<blank22>": 151698,
|
| 20 |
+
"<blank23>": 151699,
|
| 21 |
+
"<blank24>": 151700,
|
| 22 |
+
"<blank25>": 151701,
|
| 23 |
+
"<blank26>": 151702,
|
| 24 |
+
"<blank27>": 151703,
|
| 25 |
+
"<blank2>": 151678,
|
| 26 |
+
"<blank3>": 151679,
|
| 27 |
+
"<blank4>": 151680,
|
| 28 |
+
"<blank5>": 151681,
|
| 29 |
+
"<blank6>": 151682,
|
| 30 |
+
"<blank7>": 151683,
|
| 31 |
+
"<blank8>": 151684,
|
| 32 |
+
"<blank9>": 151685,
|
| 33 |
+
"<non_speech>": 151675,
|
| 34 |
+
"<think>": 151667,
|
| 35 |
+
"<tool_call>": 151657,
|
| 36 |
+
"<tool_response>": 151665,
|
| 37 |
+
"<tts_pad>": 151671,
|
| 38 |
+
"<tts_text_bos>": 151672,
|
| 39 |
+
"<tts_text_bos_single>": 151674,
|
| 40 |
+
"<tts_text_eod>": 151673,
|
| 41 |
+
"<|audio_end|>": 151670,
|
| 42 |
+
"<|audio_pad|>": 151676,
|
| 43 |
+
"<|audio_start|>": 151669,
|
| 44 |
+
"<|box_end|>": 151649,
|
| 45 |
+
"<|box_start|>": 151648,
|
| 46 |
+
"<|endoftext|>": 151643,
|
| 47 |
+
"<|file_sep|>": 151664,
|
| 48 |
+
"<|fim_middle|>": 151660,
|
| 49 |
+
"<|fim_pad|>": 151662,
|
| 50 |
+
"<|fim_prefix|>": 151659,
|
| 51 |
+
"<|fim_suffix|>": 151661,
|
| 52 |
+
"<|im_end|>": 151645,
|
| 53 |
+
"<|im_start|>": 151644,
|
| 54 |
+
"<|image_pad|>": 151655,
|
| 55 |
+
"<|object_ref_end|>": 151647,
|
| 56 |
+
"<|object_ref_start|>": 151646,
|
| 57 |
+
"<|quad_end|>": 151651,
|
| 58 |
+
"<|quad_start|>": 151650,
|
| 59 |
+
"<|repo_name|>": 151663,
|
| 60 |
+
"<|video_pad|>": 151656,
|
| 61 |
+
"<|vision_end|>": 151653,
|
| 62 |
+
"<|vision_pad|>": 151654,
|
| 63 |
+
"<|vision_start|>": 151652
|
| 64 |
+
}
|
Qwen3-ASR-WSM-Min/args.json
ADDED
|
@@ -0,0 +1,404 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"output_dir": "/home/work_nfs13/cjhe/Qwen3-ASR/ms-swift/exp/minnan_kaiyuan_1wh_qwen3asr_1.7b/v0-20260710-163648",
|
| 3 |
+
"overwrite_output_dir": false,
|
| 4 |
+
"do_train": false,
|
| 5 |
+
"do_eval": false,
|
| 6 |
+
"do_predict": false,
|
| 7 |
+
"eval_strategy": "steps",
|
| 8 |
+
"prediction_loss_only": false,
|
| 9 |
+
"per_device_train_batch_size": 64,
|
| 10 |
+
"per_device_eval_batch_size": 64,
|
| 11 |
+
"per_gpu_train_batch_size": null,
|
| 12 |
+
"per_gpu_eval_batch_size": null,
|
| 13 |
+
"gradient_accumulation_steps": 2,
|
| 14 |
+
"eval_accumulation_steps": null,
|
| 15 |
+
"eval_delay": 0,
|
| 16 |
+
"torch_empty_cache_steps": null,
|
| 17 |
+
"learning_rate": 1e-05,
|
| 18 |
+
"weight_decay": 0.1,
|
| 19 |
+
"adam_beta1": 0.9,
|
| 20 |
+
"adam_beta2": 0.95,
|
| 21 |
+
"adam_epsilon": 1e-08,
|
| 22 |
+
"max_grad_norm": 1.0,
|
| 23 |
+
"num_train_epochs": 4.0,
|
| 24 |
+
"max_steps": -1,
|
| 25 |
+
"lr_scheduler_type": "cosine",
|
| 26 |
+
"lr_scheduler_kwargs": null,
|
| 27 |
+
"warmup_ratio": 0.02,
|
| 28 |
+
"warmup_steps": 0,
|
| 29 |
+
"log_level": "passive",
|
| 30 |
+
"log_level_replica": "warning",
|
| 31 |
+
"log_on_each_node": true,
|
| 32 |
+
"logging_dir": "/home/work_nfs13/cjhe/Qwen3-ASR/ms-swift/exp/minnan_kaiyuan_1wh_qwen3asr_1.7b/v0-20260710-163648/runs",
|
| 33 |
+
"logging_strategy": "steps",
|
| 34 |
+
"logging_first_step": true,
|
| 35 |
+
"logging_steps": 100,
|
| 36 |
+
"logging_nan_inf_filter": true,
|
| 37 |
+
"save_strategy": "steps",
|
| 38 |
+
"save_steps": 1000.0,
|
| 39 |
+
"save_total_limit": null,
|
| 40 |
+
"save_safetensors": true,
|
| 41 |
+
"save_on_each_node": false,
|
| 42 |
+
"save_only_model": false,
|
| 43 |
+
"restore_callback_states_from_checkpoint": false,
|
| 44 |
+
"no_cuda": false,
|
| 45 |
+
"use_cpu": false,
|
| 46 |
+
"use_mps_device": false,
|
| 47 |
+
"seed": 42,
|
| 48 |
+
"data_seed": 42,
|
| 49 |
+
"jit_mode_eval": false,
|
| 50 |
+
"bf16": true,
|
| 51 |
+
"fp16": false,
|
| 52 |
+
"fp16_opt_level": "O1",
|
| 53 |
+
"half_precision_backend": "auto",
|
| 54 |
+
"bf16_full_eval": false,
|
| 55 |
+
"fp16_full_eval": false,
|
| 56 |
+
"tf32": null,
|
| 57 |
+
"local_rank": 0,
|
| 58 |
+
"ddp_backend": null,
|
| 59 |
+
"tpu_num_cores": null,
|
| 60 |
+
"tpu_metrics_debug": false,
|
| 61 |
+
"debug": null,
|
| 62 |
+
"dataloader_drop_last": false,
|
| 63 |
+
"eval_steps": 200.0,
|
| 64 |
+
"dataloader_num_workers": 12,
|
| 65 |
+
"dataloader_prefetch_factor": null,
|
| 66 |
+
"past_index": -1,
|
| 67 |
+
"run_name": "/home/work_nfs13/cjhe/Qwen3-ASR/ms-swift/exp/minnan_kaiyuan_1wh_qwen3asr_1.7b/v0-20260710-163648",
|
| 68 |
+
"disable_tqdm": null,
|
| 69 |
+
"remove_unused_columns": true,
|
| 70 |
+
"label_names": null,
|
| 71 |
+
"load_best_model_at_end": false,
|
| 72 |
+
"metric_for_best_model": "loss",
|
| 73 |
+
"greater_is_better": false,
|
| 74 |
+
"ignore_data_skip": false,
|
| 75 |
+
"fsdp": [],
|
| 76 |
+
"fsdp_min_num_params": 0,
|
| 77 |
+
"fsdp_config": null,
|
| 78 |
+
"fsdp_transformer_layer_cls_to_wrap": null,
|
| 79 |
+
"accelerator_config": {
|
| 80 |
+
"dispatch_batches": false
|
| 81 |
+
},
|
| 82 |
+
"parallelism_config": null,
|
| 83 |
+
"deepspeed": {
|
| 84 |
+
"fp16": {
|
| 85 |
+
"enabled": "auto",
|
| 86 |
+
"loss_scale": 0,
|
| 87 |
+
"loss_scale_window": 1000,
|
| 88 |
+
"initial_scale_power": 16,
|
| 89 |
+
"hysteresis": 2,
|
| 90 |
+
"min_loss_scale": 1
|
| 91 |
+
},
|
| 92 |
+
"bf16": {
|
| 93 |
+
"enabled": "auto"
|
| 94 |
+
},
|
| 95 |
+
"zero_optimization": {
|
| 96 |
+
"stage": 2,
|
| 97 |
+
"offload_optimizer": {
|
| 98 |
+
"device": "none",
|
| 99 |
+
"pin_memory": true
|
| 100 |
+
},
|
| 101 |
+
"allgather_partitions": true,
|
| 102 |
+
"allgather_bucket_size": 200000000.0,
|
| 103 |
+
"overlap_comm": false,
|
| 104 |
+
"reduce_scatter": true,
|
| 105 |
+
"reduce_bucket_size": 200000000.0,
|
| 106 |
+
"contiguous_gradients": true
|
| 107 |
+
},
|
| 108 |
+
"gradient_accumulation_steps": "auto",
|
| 109 |
+
"gradient_clipping": "auto",
|
| 110 |
+
"steps_per_print": 2000,
|
| 111 |
+
"train_batch_size": "auto",
|
| 112 |
+
"train_micro_batch_size_per_gpu": "auto",
|
| 113 |
+
"wall_clock_breakdown": false
|
| 114 |
+
},
|
| 115 |
+
"label_smoothing_factor": 0.0,
|
| 116 |
+
"optim": "adamw_torch_fused",
|
| 117 |
+
"optim_args": null,
|
| 118 |
+
"adafactor": false,
|
| 119 |
+
"group_by_length": false,
|
| 120 |
+
"length_column_name": "length",
|
| 121 |
+
"report_to": [
|
| 122 |
+
"tensorboard"
|
| 123 |
+
],
|
| 124 |
+
"project": "huggingface",
|
| 125 |
+
"trackio_space_id": "trackio",
|
| 126 |
+
"ddp_find_unused_parameters": null,
|
| 127 |
+
"ddp_bucket_cap_mb": null,
|
| 128 |
+
"ddp_broadcast_buffers": null,
|
| 129 |
+
"dataloader_pin_memory": true,
|
| 130 |
+
"dataloader_persistent_workers": false,
|
| 131 |
+
"skip_memory_metrics": true,
|
| 132 |
+
"use_legacy_prediction_loop": false,
|
| 133 |
+
"push_to_hub": false,
|
| 134 |
+
"resume_from_checkpoint": null,
|
| 135 |
+
"hub_model_id": null,
|
| 136 |
+
"hub_strategy": "every_save",
|
| 137 |
+
"hub_token": null,
|
| 138 |
+
"hub_private_repo": null,
|
| 139 |
+
"hub_always_push": false,
|
| 140 |
+
"hub_revision": null,
|
| 141 |
+
"gradient_checkpointing": true,
|
| 142 |
+
"gradient_checkpointing_kwargs": "{\"use_reentrant\": false}",
|
| 143 |
+
"include_inputs_for_metrics": false,
|
| 144 |
+
"include_for_metrics": [],
|
| 145 |
+
"eval_do_concat_batches": true,
|
| 146 |
+
"fp16_backend": "auto",
|
| 147 |
+
"push_to_hub_model_id": null,
|
| 148 |
+
"push_to_hub_organization": null,
|
| 149 |
+
"push_to_hub_token": null,
|
| 150 |
+
"mp_parameters": "",
|
| 151 |
+
"auto_find_batch_size": false,
|
| 152 |
+
"full_determinism": false,
|
| 153 |
+
"torchdynamo": null,
|
| 154 |
+
"ray_scope": "last",
|
| 155 |
+
"ddp_timeout": 18000000,
|
| 156 |
+
"torch_compile": false,
|
| 157 |
+
"torch_compile_backend": null,
|
| 158 |
+
"torch_compile_mode": null,
|
| 159 |
+
"include_tokens_per_second": false,
|
| 160 |
+
"include_num_input_tokens_seen": false,
|
| 161 |
+
"neftune_noise_alpha": null,
|
| 162 |
+
"optim_target_modules": null,
|
| 163 |
+
"batch_eval_metrics": false,
|
| 164 |
+
"eval_on_start": false,
|
| 165 |
+
"use_liger_kernel": false,
|
| 166 |
+
"liger_kernel_config": null,
|
| 167 |
+
"eval_use_gather_object": false,
|
| 168 |
+
"average_tokens_across_devices": true,
|
| 169 |
+
"sortish_sampler": false,
|
| 170 |
+
"predict_with_generate": false,
|
| 171 |
+
"generation_max_length": null,
|
| 172 |
+
"generation_num_beams": null,
|
| 173 |
+
"generation_config": null,
|
| 174 |
+
"tuner_backend": "peft",
|
| 175 |
+
"vit_gradient_checkpointing": true,
|
| 176 |
+
"router_aux_loss_coef": 0.0,
|
| 177 |
+
"enable_dft_loss": false,
|
| 178 |
+
"enable_channel_loss": false,
|
| 179 |
+
"safe_serialization": true,
|
| 180 |
+
"max_shard_size": "5GB",
|
| 181 |
+
"check_model": true,
|
| 182 |
+
"acc_strategy": "token",
|
| 183 |
+
"train_dataloader_shuffle": true,
|
| 184 |
+
"max_epochs": null,
|
| 185 |
+
"aligner_lr": null,
|
| 186 |
+
"vit_lr": null,
|
| 187 |
+
"use_logits_to_keep": null,
|
| 188 |
+
"ds3_gather_for_generation": true,
|
| 189 |
+
"resume_only_model": false,
|
| 190 |
+
"optimizer": null,
|
| 191 |
+
"loss_type": null,
|
| 192 |
+
"eval_metric": null,
|
| 193 |
+
"callbacks": [],
|
| 194 |
+
"early_stop_interval": null,
|
| 195 |
+
"eval_use_evalscope": false,
|
| 196 |
+
"eval_dataset": [],
|
| 197 |
+
"eval_dataset_args": null,
|
| 198 |
+
"eval_limit": null,
|
| 199 |
+
"eval_generation_config": null,
|
| 200 |
+
"extra_eval_args": null,
|
| 201 |
+
"tuner_type": "full",
|
| 202 |
+
"use_galore": false,
|
| 203 |
+
"galore_target_modules": null,
|
| 204 |
+
"galore_rank": 128,
|
| 205 |
+
"galore_update_proj_gap": 50,
|
| 206 |
+
"galore_scale": 1.0,
|
| 207 |
+
"galore_proj_type": "std",
|
| 208 |
+
"galore_optim_per_parameter": false,
|
| 209 |
+
"galore_with_embedding": false,
|
| 210 |
+
"galore_quantization": false,
|
| 211 |
+
"galore_proj_quant": false,
|
| 212 |
+
"galore_proj_bits": 4,
|
| 213 |
+
"galore_proj_group_size": 256,
|
| 214 |
+
"galore_cos_threshold": 0.4,
|
| 215 |
+
"galore_gamma_proj": 2,
|
| 216 |
+
"galore_queue_size": 5,
|
| 217 |
+
"lisa_activated_layers": 0,
|
| 218 |
+
"lisa_step_interval": 20,
|
| 219 |
+
"use_flash_ckpt": false,
|
| 220 |
+
"use_ray": false,
|
| 221 |
+
"ray_exp_name": null,
|
| 222 |
+
"device_groups": null,
|
| 223 |
+
"model": "/home/work_nfs23/asr_data/ckpt/Qwen3-ASR-1.7B",
|
| 224 |
+
"model_type": "qwen3_asr",
|
| 225 |
+
"model_revision": null,
|
| 226 |
+
"task_type": "causal_lm",
|
| 227 |
+
"torch_dtype": "bfloat16",
|
| 228 |
+
"attn_impl": "flash_attention_2",
|
| 229 |
+
"experts_impl": null,
|
| 230 |
+
"new_special_tokens": [],
|
| 231 |
+
"num_labels": null,
|
| 232 |
+
"problem_type": null,
|
| 233 |
+
"rope_scaling": null,
|
| 234 |
+
"device_map": null,
|
| 235 |
+
"max_memory": {},
|
| 236 |
+
"max_model_len": null,
|
| 237 |
+
"local_repo_path": null,
|
| 238 |
+
"init_strategy": null,
|
| 239 |
+
"template": "qwen3_asr",
|
| 240 |
+
"system": null,
|
| 241 |
+
"max_length": 30000,
|
| 242 |
+
"truncation_strategy": "delete",
|
| 243 |
+
"max_pixels": null,
|
| 244 |
+
"agent_template": null,
|
| 245 |
+
"norm_bbox": null,
|
| 246 |
+
"use_chat_template": true,
|
| 247 |
+
"padding_side": "right",
|
| 248 |
+
"padding_free": true,
|
| 249 |
+
"loss_scale": "default",
|
| 250 |
+
"sequence_parallel_size": 1,
|
| 251 |
+
"is_binary_loss_scale": null,
|
| 252 |
+
"template_backend": "swift",
|
| 253 |
+
"response_prefix": null,
|
| 254 |
+
"enable_thinking": null,
|
| 255 |
+
"preserve_thinking": null,
|
| 256 |
+
"add_non_thinking_prefix": true,
|
| 257 |
+
"disable_ignore_empty_think": false,
|
| 258 |
+
"dataset": [
|
| 259 |
+
"/home/work_nfs13/cjhe/Qwen3-ASR/ms-swift/data/kaiyuan_1wh_minnan_qwen3asr.jsonl"
|
| 260 |
+
],
|
| 261 |
+
"val_dataset": [
|
| 262 |
+
"/home/work_nfs13/cjhe/Qwen3-ASR/ms-swift/data/test.jsonl"
|
| 263 |
+
],
|
| 264 |
+
"cached_dataset": [],
|
| 265 |
+
"cached_val_dataset": [],
|
| 266 |
+
"split_dataset_ratio": 0.0,
|
| 267 |
+
"dataset_num_proc": 12,
|
| 268 |
+
"load_from_cache_file": true,
|
| 269 |
+
"dataset_shuffle": true,
|
| 270 |
+
"val_dataset_shuffle": false,
|
| 271 |
+
"streaming": false,
|
| 272 |
+
"interleave_prob": null,
|
| 273 |
+
"stopping_strategy": "first_exhausted",
|
| 274 |
+
"shuffle_buffer_size": 1000,
|
| 275 |
+
"download_mode": "reuse_dataset_if_exists",
|
| 276 |
+
"columns": {},
|
| 277 |
+
"strict": false,
|
| 278 |
+
"disable_auto_column_mapping": false,
|
| 279 |
+
"model_name": null,
|
| 280 |
+
"model_author": null,
|
| 281 |
+
"custom_dataset_info": [],
|
| 282 |
+
"quant_method": null,
|
| 283 |
+
"quant_bits": null,
|
| 284 |
+
"hqq_axis": null,
|
| 285 |
+
"bnb_4bit_compute_dtype": "bfloat16",
|
| 286 |
+
"bnb_4bit_quant_type": "nf4",
|
| 287 |
+
"bnb_4bit_use_double_quant": true,
|
| 288 |
+
"bnb_4bit_quant_storage": null,
|
| 289 |
+
"max_new_tokens": 64,
|
| 290 |
+
"temperature": 0.0,
|
| 291 |
+
"top_k": null,
|
| 292 |
+
"top_p": null,
|
| 293 |
+
"repetition_penalty": null,
|
| 294 |
+
"num_beams": 1,
|
| 295 |
+
"stream": false,
|
| 296 |
+
"stop_words": [],
|
| 297 |
+
"logprobs": false,
|
| 298 |
+
"top_logprobs": null,
|
| 299 |
+
"structured_outputs_regex": null,
|
| 300 |
+
"adapters": [],
|
| 301 |
+
"external_plugins": [],
|
| 302 |
+
"custom_register_path": [],
|
| 303 |
+
"model_kwargs": {},
|
| 304 |
+
"enable_npu_model_patch": true,
|
| 305 |
+
"load_args": false,
|
| 306 |
+
"load_data_args": false,
|
| 307 |
+
"packing": false,
|
| 308 |
+
"packing_length": null,
|
| 309 |
+
"packing_num_proc": 1,
|
| 310 |
+
"lazy_tokenize": true,
|
| 311 |
+
"use_hf": false,
|
| 312 |
+
"ignore_args_error": false,
|
| 313 |
+
"use_swift_lora": false,
|
| 314 |
+
"freeze_parameters": [],
|
| 315 |
+
"freeze_parameters_regex": null,
|
| 316 |
+
"freeze_parameters_ratio": 0.0,
|
| 317 |
+
"trainable_parameters": [
|
| 318 |
+
"thinker.audio_tower.proj1",
|
| 319 |
+
"thinker.audio_tower.proj2"
|
| 320 |
+
],
|
| 321 |
+
"trainable_parameters_regex": null,
|
| 322 |
+
"freeze_llm": false,
|
| 323 |
+
"freeze_vit": false,
|
| 324 |
+
"freeze_aligner": false,
|
| 325 |
+
"target_modules": [
|
| 326 |
+
"all-linear"
|
| 327 |
+
],
|
| 328 |
+
"target_regex": null,
|
| 329 |
+
"target_parameters": null,
|
| 330 |
+
"modules_to_save": [],
|
| 331 |
+
"lora_rank": 8,
|
| 332 |
+
"lora_alpha": 32,
|
| 333 |
+
"lora_dropout": 0.05,
|
| 334 |
+
"lora_bias": "none",
|
| 335 |
+
"lora_dtype": null,
|
| 336 |
+
"lorap_lr_ratio": null,
|
| 337 |
+
"use_rslora": false,
|
| 338 |
+
"use_dora": false,
|
| 339 |
+
"lora_ga_batch_size": 2,
|
| 340 |
+
"lora_ga_iters": 2,
|
| 341 |
+
"lora_ga_max_length": 1024,
|
| 342 |
+
"lora_ga_direction": "ArB2r",
|
| 343 |
+
"lora_ga_scale": "stable",
|
| 344 |
+
"lora_ga_stable_gamma": 16,
|
| 345 |
+
"init_weights": true,
|
| 346 |
+
"fourier_n_frequency": 2000,
|
| 347 |
+
"fourier_scaling": 300.0,
|
| 348 |
+
"boft_block_size": 4,
|
| 349 |
+
"boft_block_num": 0,
|
| 350 |
+
"boft_n_butterfly_factor": 1,
|
| 351 |
+
"boft_dropout": 0.0,
|
| 352 |
+
"vera_rank": 256,
|
| 353 |
+
"vera_projection_prng_key": 0,
|
| 354 |
+
"vera_dropout": 0.0,
|
| 355 |
+
"vera_d_initial": 0.1,
|
| 356 |
+
"adapter_act": "gelu",
|
| 357 |
+
"adapter_length": 128,
|
| 358 |
+
"adalora_target_r": 8,
|
| 359 |
+
"adalora_init_r": 12,
|
| 360 |
+
"adalora_tinit": 0,
|
| 361 |
+
"adalora_tfinal": 0,
|
| 362 |
+
"adalora_deltaT": 1,
|
| 363 |
+
"adalora_beta1": 0.85,
|
| 364 |
+
"adalora_beta2": 0.85,
|
| 365 |
+
"adalora_orth_reg_weight": 0.5,
|
| 366 |
+
"llamapro_num_new_blocks": 4,
|
| 367 |
+
"llamapro_num_groups": null,
|
| 368 |
+
"reft_layer_key": null,
|
| 369 |
+
"reft_layers": null,
|
| 370 |
+
"reft_rank": 4,
|
| 371 |
+
"reft_intervention_type": "LoreftIntervention",
|
| 372 |
+
"reft_args": null,
|
| 373 |
+
"swanlab_token": null,
|
| 374 |
+
"swanlab_project": "ms-swift",
|
| 375 |
+
"swanlab_workspace": null,
|
| 376 |
+
"swanlab_exp_name": null,
|
| 377 |
+
"swanlab_notification_method": null,
|
| 378 |
+
"swanlab_webhook_url": null,
|
| 379 |
+
"swanlab_secret": null,
|
| 380 |
+
"swanlab_sender_email": null,
|
| 381 |
+
"swanlab_receiver_email": null,
|
| 382 |
+
"swanlab_smtp_server": null,
|
| 383 |
+
"swanlab_smtp_port": null,
|
| 384 |
+
"swanlab_email_language": "zh",
|
| 385 |
+
"swanlab_mode": "cloud",
|
| 386 |
+
"add_version": true,
|
| 387 |
+
"create_checkpoint_symlink": false,
|
| 388 |
+
"zero_hpz_partition_size": null,
|
| 389 |
+
"deepspeed_autotp_size": null,
|
| 390 |
+
"swift_version": "4.3.1",
|
| 391 |
+
"ckpt_dir": null,
|
| 392 |
+
"rank": 0,
|
| 393 |
+
"global_world_size": 8,
|
| 394 |
+
"local_world_size": 8,
|
| 395 |
+
"model_suffix": "Qwen3-ASR-1.7B",
|
| 396 |
+
"model_info": "ModelInfo(model_type='qwen3_asr', model_dir='/home/work_nfs23/asr_data/ckpt/Qwen3-ASR-1.7B', torch_dtype=torch.bfloat16, max_model_len=65536, quant_method=None, quant_bits=None, rope_scaling={'interleaved': True, 'mrope_interleaved': True, 'mrope_section': [24, 20, 20], 'rope_type': 'default', 'type': 'default'}, is_moe_model=False, is_multimodal=True, config=None, task_type='causal_lm', num_labels=None)",
|
| 397 |
+
"model_meta": "ModelMeta(model_type='qwen3_asr', model_groups=[ModelGroup(models=[Model(ms_model_id='Qwen/Qwen3-ASR-1.7B', hf_model_id='Qwen/Qwen3-ASR-1.7B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-ASR-0.6B', hf_model_id='Qwen/Qwen3-ASR-0.6B', model_path=None, ms_revision=None, hf_revision=None)], template='qwen3_asr', ignore_patterns=None, requires=None, tags=[])], loader=<class 'swift.model.models.qwen.Qwen3ASRLoader'>, template='qwen3_asr', model_arch=MultiModelKeys(arch_name='qwen3_asr', embedding=None, module_list=None, lm_head=None, q_proj=None, k_proj=None, v_proj=None, o_proj=None, attention=None, mlp=None, down_proj=None, qkv_proj=None, qk_proj=None, qa_proj=None, qb_proj=None, kv_proj=None, kva_proj=None, kvb_proj=None, language_model=['thinker.model', 'thinker.lm_head'], aligner=['thinker.audio_tower.proj1', 'thinker.audio_tower.proj2'], vision_tower=['thinker.audio_tower'], generator=[]), mcore_model_type=None, architectures=['Qwen3ASRForConditionalGeneration'], additional_saved_files=[], torch_dtype=None, is_multimodal=True, is_reward=False, task_type=None, ignore_patterns=None, requires=['qwen-asr', 'transformers==4.57.6'], tags=[])",
|
| 398 |
+
"model_dir": "/home/work_nfs23/asr_data/ckpt/Qwen3-ASR-1.7B",
|
| 399 |
+
"template_meta": "QwenTemplateMeta(template_type='qwen3_asr', prefix=['<|im_start|>system\\n<|im_end|>\\n'], prompt=['<|im_start|>user\\n{{QUERY}}<|im_end|>\\n<|im_start|>assistant\\n'], chat_sep=['<|im_end|>\\n'], suffix=['<|im_end|>\\n'], template_cls=<class 'swift.template.templates.qwen.Qwen3ASRTemplate'>, system_prefix=['<|im_start|>system\\n{{SYSTEM}}<|im_end|>\\n'], default_system=None, auto_add_bos=False, stop_words=['<|endoftext|>'], agent_template='hermes', is_thinking=False, thinking_prefix='', non_thinking_prefix='', history_thinking_prefix='')",
|
| 400 |
+
"_val_dataset_exists": true,
|
| 401 |
+
"hub": "<class 'swift.hub.hub.MSHub'>",
|
| 402 |
+
"evaluation_strategy": "steps",
|
| 403 |
+
"training_args": "Seq2SeqTrainingArguments(output_dir='/home/work_nfs13/cjhe/Qwen3-ASR/ms-swift/exp/minnan_kaiyuan_1wh_qwen3asr_1.7b/v0-20260710-163648', 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=64, per_device_eval_batch_size=64, per_gpu_train_batch_size=None, per_gpu_eval_batch_size=None, gradient_accumulation_steps=2, eval_accumulation_steps=None, eval_delay=0, torch_empty_cache_steps=None, learning_rate=1e-05, weight_decay=0.1, adam_beta1=0.9, adam_beta2=0.95, adam_epsilon=1e-08, max_grad_norm=1.0, num_train_epochs=4.0, max_steps=-1, lr_scheduler_type=<SchedulerType.COSINE: 'cosine'>, lr_scheduler_kwargs=None, warmup_ratio=0.02, warmup_steps=0, log_level='passive', log_level_replica='warning', log_on_each_node=True, logging_dir='/home/work_nfs13/cjhe/Qwen3-ASR/ms-swift/exp/minnan_kaiyuan_1wh_qwen3asr_1.7b/v0-20260710-163648/runs', logging_strategy=<IntervalStrategy.STEPS: 'steps'>, logging_first_step=True, logging_steps=100, logging_nan_inf_filter=True, save_strategy=<SaveStrategy.STEPS: 'steps'>, save_steps=1000, save_total_limit=None, 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, 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=200, dataloader_num_workers=12, dataloader_prefetch_factor=2, past_index=-1, run_name='/home/work_nfs13/cjhe/Qwen3-ASR/ms-swift/exp/minnan_kaiyuan_1wh_qwen3asr_1.7b/v0-20260710-163648', 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), parallelism_config=None, 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': 2, 'offload_optimizer': {'device': 'none', 'pin_memory': True}, 'allgather_partitions': True, 'allgather_bucket_size': 200000000.0, 'overlap_comm': False, 'reduce_scatter': True, 'reduce_bucket_size': 200000000.0, 'contiguous_gradients': 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_FUSED: 'adamw_torch_fused'>, optim_args=None, adafactor=False, group_by_length=False, length_column_name='length', report_to=['tensorboard'], project='huggingface', trackio_space_id='trackio', 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=False, resume_from_checkpoint=None, hub_model_id=None, hub_strategy=<HubStrategy.EVERY_SAVE: 'every_save'>, hub_token=None, hub_private_repo=None, hub_always_push=False, hub_revision=None, gradient_checkpointing=True, gradient_checkpointing_kwargs={'use_reentrant': False}, include_inputs_for_metrics=False, include_for_metrics=[], eval_do_concat_batches=True, fp16_backend='auto', 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=18000000, torch_compile=False, torch_compile_backend=None, torch_compile_mode=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, liger_kernel_config=None, 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, tuner_backend='peft', vit_gradient_checkpointing=True, router_aux_loss_coef=0.0, enable_dft_loss=False, enable_channel_loss=False, safe_serialization=True, max_shard_size='5GB', check_model=True, acc_strategy='token', train_dataloader_shuffle=True, max_epochs=None, aligner_lr=None, vit_lr=None, use_logits_to_keep=None, ds3_gather_for_generation=True, resume_only_model=False, optimizer=None, loss_type=None, eval_metric=None, callbacks=[], early_stop_interval=None, eval_use_evalscope=False, eval_dataset=[], eval_dataset_args=None, eval_limit=None, eval_generation_config=None, extra_eval_args=None, tuner_type='full', use_galore=False, galore_target_modules=None, galore_rank=128, galore_update_proj_gap=50, galore_scale=1.0, galore_proj_type='std', galore_optim_per_parameter=False, galore_with_embedding=False, galore_quantization=False, galore_proj_quant=False, galore_proj_bits=4, galore_proj_group_size=256, galore_cos_threshold=0.4, galore_gamma_proj=2, galore_queue_size=5, lisa_activated_layers=0, lisa_step_interval=20, use_flash_ckpt=False)"
|
| 404 |
+
}
|
Qwen3-ASR-WSM-Min/chat_template.jinja
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{%- set ns = namespace(system_text="") -%}
|
| 2 |
+
{%- for m in messages -%}
|
| 3 |
+
{%- if m.role == 'system' -%}
|
| 4 |
+
{%- if m.content is string -%}
|
| 5 |
+
{%- set ns.system_text = ns.system_text + m.content -%}
|
| 6 |
+
{%- else -%}
|
| 7 |
+
{%- for c in m.content -%}
|
| 8 |
+
{%- if c.type == 'text' and (c.text is defined) -%}
|
| 9 |
+
{%- set ns.system_text = ns.system_text + c.text -%}
|
| 10 |
+
{%- endif -%}
|
| 11 |
+
{%- endfor -%}
|
| 12 |
+
{%- endif -%}
|
| 13 |
+
{%- endif -%}
|
| 14 |
+
{%- endfor -%}
|
| 15 |
+
|
| 16 |
+
{%- set ns2 = namespace(audio_tokens="") -%}
|
| 17 |
+
{%- for m in messages -%}
|
| 18 |
+
{%- if m.content is not string -%}
|
| 19 |
+
{%- for c in m.content -%}
|
| 20 |
+
{%- if c.type == 'audio' or ('audio' in c) or ('audio_url' in c) -%}
|
| 21 |
+
{%- set ns2.audio_tokens = ns2.audio_tokens + "<|audio_start|><|audio_pad|><|audio_end|>" -%}
|
| 22 |
+
{%- endif -%}
|
| 23 |
+
{%- endfor -%}
|
| 24 |
+
{%- endif -%}
|
| 25 |
+
{%- endfor -%}
|
| 26 |
+
|
| 27 |
+
{{- '<|im_start|>system\n' + (ns.system_text if ns.system_text is string else '') + '<|im_end|>\n' -}}
|
| 28 |
+
{{- '<|im_start|>user\n' + ns2.audio_tokens + '<|im_end|>\n' -}}
|
| 29 |
+
{%- if add_generation_prompt -%}
|
| 30 |
+
{{- '<|im_start|>assistant\n' -}}
|
| 31 |
+
{%- endif -%}
|
Qwen3-ASR-WSM-Min/config.json
ADDED
|
@@ -0,0 +1,228 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"Qwen3ASRForConditionalGeneration"
|
| 4 |
+
],
|
| 5 |
+
"dtype": "bfloat16",
|
| 6 |
+
"eos_token_id": 151645,
|
| 7 |
+
"hidden_size": 2048,
|
| 8 |
+
"keys_to_ignore_at_inference": [
|
| 9 |
+
"past_key_values"
|
| 10 |
+
],
|
| 11 |
+
"model_type": "qwen3_asr",
|
| 12 |
+
"pad_token_id": 151643,
|
| 13 |
+
"support_languages": [
|
| 14 |
+
"Chinese",
|
| 15 |
+
"English",
|
| 16 |
+
"Cantonese",
|
| 17 |
+
"Arabic",
|
| 18 |
+
"German",
|
| 19 |
+
"French",
|
| 20 |
+
"Spanish",
|
| 21 |
+
"Portuguese",
|
| 22 |
+
"Indonesian",
|
| 23 |
+
"Italian",
|
| 24 |
+
"Korean",
|
| 25 |
+
"Russian",
|
| 26 |
+
"Thai",
|
| 27 |
+
"Vietnamese",
|
| 28 |
+
"Japanese",
|
| 29 |
+
"Turkish",
|
| 30 |
+
"Hindi",
|
| 31 |
+
"Malay",
|
| 32 |
+
"Dutch",
|
| 33 |
+
"Swedish",
|
| 34 |
+
"Danish",
|
| 35 |
+
"Finnish",
|
| 36 |
+
"Polish",
|
| 37 |
+
"Czech",
|
| 38 |
+
"Filipino",
|
| 39 |
+
"Persian",
|
| 40 |
+
"Greek",
|
| 41 |
+
"Romanian",
|
| 42 |
+
"Hungarian",
|
| 43 |
+
"Macedonian"
|
| 44 |
+
],
|
| 45 |
+
"thinker_config": {
|
| 46 |
+
"architectures": [
|
| 47 |
+
"Qwen3ASRForConditionalGeneration"
|
| 48 |
+
],
|
| 49 |
+
"audio_config": {
|
| 50 |
+
"_name_or_path": "",
|
| 51 |
+
"activation_dropout": 0,
|
| 52 |
+
"activation_function": "gelu",
|
| 53 |
+
"add_cross_attention": false,
|
| 54 |
+
"architectures": null,
|
| 55 |
+
"attention_dropout": 0,
|
| 56 |
+
"bad_words_ids": null,
|
| 57 |
+
"begin_suppress_tokens": null,
|
| 58 |
+
"bos_token_id": null,
|
| 59 |
+
"chunk_size_feed_forward": 0,
|
| 60 |
+
"conv_chunksize": 500,
|
| 61 |
+
"cross_attention_hidden_size": null,
|
| 62 |
+
"d_model": 1024,
|
| 63 |
+
"decoder_start_token_id": null,
|
| 64 |
+
"diversity_penalty": 0.0,
|
| 65 |
+
"do_sample": false,
|
| 66 |
+
"downsample_hidden_size": 480,
|
| 67 |
+
"dropout": 0,
|
| 68 |
+
"dtype": "bfloat16",
|
| 69 |
+
"early_stopping": false,
|
| 70 |
+
"encoder_attention_heads": 16,
|
| 71 |
+
"encoder_ffn_dim": 4096,
|
| 72 |
+
"encoder_layers": 24,
|
| 73 |
+
"encoder_no_repeat_ngram_size": 0,
|
| 74 |
+
"eos_token_id": null,
|
| 75 |
+
"exponential_decay_length_penalty": null,
|
| 76 |
+
"finetuning_task": null,
|
| 77 |
+
"forced_bos_token_id": null,
|
| 78 |
+
"forced_eos_token_id": null,
|
| 79 |
+
"id2label": {
|
| 80 |
+
"0": "LABEL_0",
|
| 81 |
+
"1": "LABEL_1"
|
| 82 |
+
},
|
| 83 |
+
"initializer_range": 0.02,
|
| 84 |
+
"is_decoder": false,
|
| 85 |
+
"is_encoder_decoder": false,
|
| 86 |
+
"label2id": {
|
| 87 |
+
"LABEL_0": 0,
|
| 88 |
+
"LABEL_1": 1
|
| 89 |
+
},
|
| 90 |
+
"length_penalty": 1.0,
|
| 91 |
+
"max_length": 20,
|
| 92 |
+
"max_source_positions": 1500,
|
| 93 |
+
"min_length": 0,
|
| 94 |
+
"model_type": "qwen3_asr_audio_encoder",
|
| 95 |
+
"n_window": 50,
|
| 96 |
+
"n_window_infer": 800,
|
| 97 |
+
"no_repeat_ngram_size": 0,
|
| 98 |
+
"num_beam_groups": 1,
|
| 99 |
+
"num_beams": 1,
|
| 100 |
+
"num_hidden_layers": 24,
|
| 101 |
+
"num_mel_bins": 128,
|
| 102 |
+
"num_return_sequences": 1,
|
| 103 |
+
"output_attentions": false,
|
| 104 |
+
"output_dim": 2048,
|
| 105 |
+
"output_hidden_states": false,
|
| 106 |
+
"output_scores": false,
|
| 107 |
+
"pad_token_id": null,
|
| 108 |
+
"prefix": null,
|
| 109 |
+
"problem_type": null,
|
| 110 |
+
"pruned_heads": {},
|
| 111 |
+
"remove_invalid_values": false,
|
| 112 |
+
"repetition_penalty": 1.0,
|
| 113 |
+
"return_dict": true,
|
| 114 |
+
"return_dict_in_generate": false,
|
| 115 |
+
"scale_embedding": false,
|
| 116 |
+
"sep_token_id": null,
|
| 117 |
+
"suppress_tokens": null,
|
| 118 |
+
"task_specific_params": null,
|
| 119 |
+
"temperature": 1.0,
|
| 120 |
+
"tf_legacy_loss": false,
|
| 121 |
+
"tie_encoder_decoder": false,
|
| 122 |
+
"tie_word_embeddings": true,
|
| 123 |
+
"tokenizer_class": null,
|
| 124 |
+
"top_k": 50,
|
| 125 |
+
"top_p": 1.0,
|
| 126 |
+
"torchscript": false,
|
| 127 |
+
"typical_p": 1.0,
|
| 128 |
+
"use_bfloat16": false
|
| 129 |
+
},
|
| 130 |
+
"audio_end_token_id": 151670,
|
| 131 |
+
"audio_start_token_id": 151669,
|
| 132 |
+
"audio_token_id": 151676,
|
| 133 |
+
"dtype": "bfloat16",
|
| 134 |
+
"initializer_range": 0.02,
|
| 135 |
+
"model_type": "qwen3_asr_thinker",
|
| 136 |
+
"text_config": {
|
| 137 |
+
"_name_or_path": "",
|
| 138 |
+
"add_cross_attention": false,
|
| 139 |
+
"architectures": null,
|
| 140 |
+
"attention_bias": false,
|
| 141 |
+
"attention_dropout": 0.0,
|
| 142 |
+
"bad_words_ids": null,
|
| 143 |
+
"begin_suppress_tokens": null,
|
| 144 |
+
"bos_token_id": null,
|
| 145 |
+
"chunk_size_feed_forward": 0,
|
| 146 |
+
"cross_attention_hidden_size": null,
|
| 147 |
+
"decoder_start_token_id": null,
|
| 148 |
+
"diversity_penalty": 0.0,
|
| 149 |
+
"do_sample": false,
|
| 150 |
+
"dtype": "bfloat16",
|
| 151 |
+
"early_stopping": false,
|
| 152 |
+
"encoder_no_repeat_ngram_size": 0,
|
| 153 |
+
"eos_token_id": null,
|
| 154 |
+
"exponential_decay_length_penalty": null,
|
| 155 |
+
"finetuning_task": null,
|
| 156 |
+
"forced_bos_token_id": null,
|
| 157 |
+
"forced_eos_token_id": null,
|
| 158 |
+
"head_dim": 128,
|
| 159 |
+
"hidden_act": "silu",
|
| 160 |
+
"hidden_size": 2048,
|
| 161 |
+
"id2label": {
|
| 162 |
+
"0": "LABEL_0",
|
| 163 |
+
"1": "LABEL_1"
|
| 164 |
+
},
|
| 165 |
+
"initializer_range": 0.02,
|
| 166 |
+
"intermediate_size": 6144,
|
| 167 |
+
"is_decoder": false,
|
| 168 |
+
"is_encoder_decoder": false,
|
| 169 |
+
"label2id": {
|
| 170 |
+
"LABEL_0": 0,
|
| 171 |
+
"LABEL_1": 1
|
| 172 |
+
},
|
| 173 |
+
"length_penalty": 1.0,
|
| 174 |
+
"max_length": 20,
|
| 175 |
+
"max_position_embeddings": 65536,
|
| 176 |
+
"min_length": 0,
|
| 177 |
+
"model_type": "qwen3_asr_text",
|
| 178 |
+
"no_repeat_ngram_size": 0,
|
| 179 |
+
"num_attention_heads": 16,
|
| 180 |
+
"num_beam_groups": 1,
|
| 181 |
+
"num_beams": 1,
|
| 182 |
+
"num_hidden_layers": 28,
|
| 183 |
+
"num_key_value_heads": 8,
|
| 184 |
+
"num_return_sequences": 1,
|
| 185 |
+
"output_attentions": false,
|
| 186 |
+
"output_hidden_states": false,
|
| 187 |
+
"output_scores": false,
|
| 188 |
+
"pad_token_id": 151643,
|
| 189 |
+
"prefix": null,
|
| 190 |
+
"problem_type": null,
|
| 191 |
+
"pruned_heads": {},
|
| 192 |
+
"remove_invalid_values": false,
|
| 193 |
+
"repetition_penalty": 1.0,
|
| 194 |
+
"return_dict": true,
|
| 195 |
+
"return_dict_in_generate": false,
|
| 196 |
+
"rms_norm_eps": 1e-06,
|
| 197 |
+
"rope_scaling": {
|
| 198 |
+
"interleaved": true,
|
| 199 |
+
"mrope_interleaved": true,
|
| 200 |
+
"mrope_section": [
|
| 201 |
+
24,
|
| 202 |
+
20,
|
| 203 |
+
20
|
| 204 |
+
],
|
| 205 |
+
"rope_type": "default",
|
| 206 |
+
"type": "default"
|
| 207 |
+
},
|
| 208 |
+
"rope_theta": 1000000,
|
| 209 |
+
"sep_token_id": null,
|
| 210 |
+
"suppress_tokens": null,
|
| 211 |
+
"task_specific_params": null,
|
| 212 |
+
"temperature": 1.0,
|
| 213 |
+
"tf_legacy_loss": false,
|
| 214 |
+
"tie_encoder_decoder": false,
|
| 215 |
+
"tie_word_embeddings": true,
|
| 216 |
+
"tokenizer_class": null,
|
| 217 |
+
"top_k": 50,
|
| 218 |
+
"top_p": 1.0,
|
| 219 |
+
"torchscript": false,
|
| 220 |
+
"typical_p": 1.0,
|
| 221 |
+
"use_bfloat16": false,
|
| 222 |
+
"use_cache": false,
|
| 223 |
+
"vocab_size": 151936
|
| 224 |
+
},
|
| 225 |
+
"user_token_id": 872
|
| 226 |
+
},
|
| 227 |
+
"transformers_version": "4.57.6"
|
| 228 |
+
}
|
Qwen3-ASR-WSM-Min/generation_config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"eos_token_id": [
|
| 4 |
+
151643,
|
| 5 |
+
151645
|
| 6 |
+
],
|
| 7 |
+
"pad_token_id": 151643,
|
| 8 |
+
"transformers_version": "4.57.6"
|
| 9 |
+
}
|
Qwen3-ASR-WSM-Min/global_step16000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:665d3a92cf0e488b2ee3559f9b57a08b22b94c92792ef496c2a65644a32a4841
|
| 3 |
+
size 3057098949
|
Qwen3-ASR-WSM-Min/global_step16000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:15f1a5aadd39e8326f1cd5738e56b3c93c513a815e3519aba07441fe87a3ac4b
|
| 3 |
+
size 3057090437
|
Qwen3-ASR-WSM-Min/global_step16000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:61746bac9b1d406a9604d5dfc8d094231d197f562ec953329988a8037b13db6d
|
| 3 |
+
size 3057089541
|
Qwen3-ASR-WSM-Min/global_step16000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7baa04f2b3cf1633627a80d32389ad27e6f7128b27a69af9fbd62de498568c8f
|
| 3 |
+
size 3057091013
|
Qwen3-ASR-WSM-Min/global_step16000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f2e3a31c4cf30b3713738d3a09320471b01163efd48113babd579c7ca0d8db84
|
| 3 |
+
size 3057091077
|
Qwen3-ASR-WSM-Min/global_step16000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9abf4332aa72685dda38cb04139c5317c1cca66deb1009277676e7045b18856e
|
| 3 |
+
size 3057091269
|
Qwen3-ASR-WSM-Min/global_step16000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:476e5d8399c4223327ecdb4d4e5772a2a7ead9b964cc521a7f010f65c4f6ef20
|
| 3 |
+
size 3057091781
|
Qwen3-ASR-WSM-Min/global_step16000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3d05726de0b68a3fe6ff5d337ca5b4dd4ce8ef3f7f3c378e1e79fc57ff4094da
|
| 3 |
+
size 3057091525
|
Qwen3-ASR-WSM-Min/global_step16000/mp_rank_00_model_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3e8c5d816c6c24b52f4453604621453251dc15ec65ffc695cb47015ec581545e
|
| 3 |
+
size 4076311909
|
Qwen3-ASR-WSM-Min/latest
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
global_step16000
|
Qwen3-ASR-WSM-Min/merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
Qwen3-ASR-WSM-Min/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:95faa02d29b2ee646ce9776fb9e01775c29ec05d19407a46e9db0e0d78bb8daa
|
| 3 |
+
size 4698521600
|
Qwen3-ASR-WSM-Min/preprocessor_config.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"chunk_length": 30,
|
| 3 |
+
"dither": 0.0,
|
| 4 |
+
"feature_extractor_type": "WhisperFeatureExtractor",
|
| 5 |
+
"feature_size": 128,
|
| 6 |
+
"hop_length": 160,
|
| 7 |
+
"n_fft": 400,
|
| 8 |
+
"n_samples": 480000,
|
| 9 |
+
"nb_max_frames": 3000,
|
| 10 |
+
"padding_side": "right",
|
| 11 |
+
"padding_value": 0.0,
|
| 12 |
+
"processor_class": "Qwen3ASRProcessor",
|
| 13 |
+
"return_attention_mask": true
|
| 14 |
+
}
|
Qwen3-ASR-WSM-Min/rng_state_0.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:433d385e712c93716e15a8139e2d1f04bf04511b0df6c15370e96c251646009e
|
| 3 |
+
size 16389
|
Qwen3-ASR-WSM-Min/rng_state_1.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ac21b6e036188089dabe3eef36114c8e29f0fdf27adda9c4c1fa6931a4c1267b
|
| 3 |
+
size 16389
|
Qwen3-ASR-WSM-Min/rng_state_2.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fb7b89410fb7449d97ce4ca2cfa0ea64c1621df9f32f0bed601335898b857ad2
|
| 3 |
+
size 16389
|
Qwen3-ASR-WSM-Min/rng_state_3.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:273f9341d3654a394aea08b14ed357df698a9ad94ec8b55c43d901375a7ec7c6
|
| 3 |
+
size 16389
|
Qwen3-ASR-WSM-Min/rng_state_4.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5db9e0c6d774d042b39733f26b05d33239deae706f052a1e68fe8d722ef518ef
|
| 3 |
+
size 16389
|
Qwen3-ASR-WSM-Min/rng_state_5.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c8172edfff7fc463f07839f0d75e7c4bdf8a7249e24655a27b1c5e0d09605b0e
|
| 3 |
+
size 16389
|
Qwen3-ASR-WSM-Min/rng_state_6.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:77d2e035eb864da1c82dfc5e7ede2970364a6be674984d7d79b65adfa1446941
|
| 3 |
+
size 16389
|
Qwen3-ASR-WSM-Min/rng_state_7.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c6331d7c03af61ffc3cbb8b7da0b56accbfde04d779e7d90270e5d969647d59c
|
| 3 |
+
size 16389
|
Qwen3-ASR-WSM-Min/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c5b4f351d1648098f8f117d3b37338f745e94c89381a7b9df47e72e475244417
|
| 3 |
+
size 1465
|
Qwen3-ASR-WSM-Min/special_tokens_map.json
ADDED
|
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"additional_special_tokens": [
|
| 3 |
+
"<|im_start|>",
|
| 4 |
+
"<|im_end|>",
|
| 5 |
+
"<|object_ref_start|>",
|
| 6 |
+
"<|object_ref_end|>",
|
| 7 |
+
"<|box_start|>",
|
| 8 |
+
"<|box_end|>",
|
| 9 |
+
"<|quad_start|>",
|
| 10 |
+
"<|quad_end|>",
|
| 11 |
+
"<|vision_start|>",
|
| 12 |
+
"<|vision_end|>",
|
| 13 |
+
"<|vision_pad|>",
|
| 14 |
+
"<|image_pad|>",
|
| 15 |
+
"<|video_pad|>",
|
| 16 |
+
"<|audio_start|>",
|
| 17 |
+
"<|audio_end|>",
|
| 18 |
+
"<tts_pad>",
|
| 19 |
+
"<tts_text_bos>",
|
| 20 |
+
"<tts_text_bos_single>",
|
| 21 |
+
"<|audio_pad|>"
|
| 22 |
+
],
|
| 23 |
+
"audio_bos_token": "<|audio_start|>",
|
| 24 |
+
"audio_eos_token": "<|audio_end|>",
|
| 25 |
+
"audio_token": "<|audio_pad|>",
|
| 26 |
+
"eos_token": {
|
| 27 |
+
"content": "<|im_end|>",
|
| 28 |
+
"lstrip": false,
|
| 29 |
+
"normalized": false,
|
| 30 |
+
"rstrip": false,
|
| 31 |
+
"single_word": false
|
| 32 |
+
},
|
| 33 |
+
"image_token": "<|image_pad|>",
|
| 34 |
+
"pad_token": {
|
| 35 |
+
"content": "<|endoftext|>",
|
| 36 |
+
"lstrip": false,
|
| 37 |
+
"normalized": false,
|
| 38 |
+
"rstrip": false,
|
| 39 |
+
"single_word": false
|
| 40 |
+
},
|
| 41 |
+
"video_token": "<|video_pad|>",
|
| 42 |
+
"vision_bos_token": "<|vision_start|>",
|
| 43 |
+
"vision_eos_token": "<|vision_end|>"
|
| 44 |
+
}
|
Qwen3-ASR-WSM-Min/tokenizer.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bd2a97b55c8f7f9c328c73ee9b9178771037e9f566dfca8e238a063d41cbac92
|
| 3 |
+
size 11429377
|
Qwen3-ASR-WSM-Min/tokenizer_config.json
ADDED
|
@@ -0,0 +1,549 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": false,
|
| 3 |
+
"add_prefix_space": false,
|
| 4 |
+
"added_tokens_decoder": {
|
| 5 |
+
"151643": {
|
| 6 |
+
"content": "<|endoftext|>",
|
| 7 |
+
"lstrip": false,
|
| 8 |
+
"normalized": false,
|
| 9 |
+
"rstrip": false,
|
| 10 |
+
"single_word": false,
|
| 11 |
+
"special": true
|
| 12 |
+
},
|
| 13 |
+
"151644": {
|
| 14 |
+
"content": "<|im_start|>",
|
| 15 |
+
"lstrip": false,
|
| 16 |
+
"normalized": false,
|
| 17 |
+
"rstrip": false,
|
| 18 |
+
"single_word": false,
|
| 19 |
+
"special": true
|
| 20 |
+
},
|
| 21 |
+
"151645": {
|
| 22 |
+
"content": "<|im_end|>",
|
| 23 |
+
"lstrip": false,
|
| 24 |
+
"normalized": false,
|
| 25 |
+
"rstrip": false,
|
| 26 |
+
"single_word": false,
|
| 27 |
+
"special": true
|
| 28 |
+
},
|
| 29 |
+
"151646": {
|
| 30 |
+
"content": "<|object_ref_start|>",
|
| 31 |
+
"lstrip": false,
|
| 32 |
+
"normalized": false,
|
| 33 |
+
"rstrip": false,
|
| 34 |
+
"single_word": false,
|
| 35 |
+
"special": true
|
| 36 |
+
},
|
| 37 |
+
"151647": {
|
| 38 |
+
"content": "<|object_ref_end|>",
|
| 39 |
+
"lstrip": false,
|
| 40 |
+
"normalized": false,
|
| 41 |
+
"rstrip": false,
|
| 42 |
+
"single_word": false,
|
| 43 |
+
"special": true
|
| 44 |
+
},
|
| 45 |
+
"151648": {
|
| 46 |
+
"content": "<|box_start|>",
|
| 47 |
+
"lstrip": false,
|
| 48 |
+
"normalized": false,
|
| 49 |
+
"rstrip": false,
|
| 50 |
+
"single_word": false,
|
| 51 |
+
"special": true
|
| 52 |
+
},
|
| 53 |
+
"151649": {
|
| 54 |
+
"content": "<|box_end|>",
|
| 55 |
+
"lstrip": false,
|
| 56 |
+
"normalized": false,
|
| 57 |
+
"rstrip": false,
|
| 58 |
+
"single_word": false,
|
| 59 |
+
"special": true
|
| 60 |
+
},
|
| 61 |
+
"151650": {
|
| 62 |
+
"content": "<|quad_start|>",
|
| 63 |
+
"lstrip": false,
|
| 64 |
+
"normalized": false,
|
| 65 |
+
"rstrip": false,
|
| 66 |
+
"single_word": false,
|
| 67 |
+
"special": true
|
| 68 |
+
},
|
| 69 |
+
"151651": {
|
| 70 |
+
"content": "<|quad_end|>",
|
| 71 |
+
"lstrip": false,
|
| 72 |
+
"normalized": false,
|
| 73 |
+
"rstrip": false,
|
| 74 |
+
"single_word": false,
|
| 75 |
+
"special": true
|
| 76 |
+
},
|
| 77 |
+
"151652": {
|
| 78 |
+
"content": "<|vision_start|>",
|
| 79 |
+
"lstrip": false,
|
| 80 |
+
"normalized": false,
|
| 81 |
+
"rstrip": false,
|
| 82 |
+
"single_word": false,
|
| 83 |
+
"special": true
|
| 84 |
+
},
|
| 85 |
+
"151653": {
|
| 86 |
+
"content": "<|vision_end|>",
|
| 87 |
+
"lstrip": false,
|
| 88 |
+
"normalized": false,
|
| 89 |
+
"rstrip": false,
|
| 90 |
+
"single_word": false,
|
| 91 |
+
"special": true
|
| 92 |
+
},
|
| 93 |
+
"151654": {
|
| 94 |
+
"content": "<|vision_pad|>",
|
| 95 |
+
"lstrip": false,
|
| 96 |
+
"normalized": false,
|
| 97 |
+
"rstrip": false,
|
| 98 |
+
"single_word": false,
|
| 99 |
+
"special": true
|
| 100 |
+
},
|
| 101 |
+
"151655": {
|
| 102 |
+
"content": "<|image_pad|>",
|
| 103 |
+
"lstrip": false,
|
| 104 |
+
"normalized": false,
|
| 105 |
+
"rstrip": false,
|
| 106 |
+
"single_word": false,
|
| 107 |
+
"special": true
|
| 108 |
+
},
|
| 109 |
+
"151656": {
|
| 110 |
+
"content": "<|video_pad|>",
|
| 111 |
+
"lstrip": false,
|
| 112 |
+
"normalized": false,
|
| 113 |
+
"rstrip": false,
|
| 114 |
+
"single_word": false,
|
| 115 |
+
"special": true
|
| 116 |
+
},
|
| 117 |
+
"151657": {
|
| 118 |
+
"content": "<tool_call>",
|
| 119 |
+
"lstrip": false,
|
| 120 |
+
"normalized": false,
|
| 121 |
+
"rstrip": false,
|
| 122 |
+
"single_word": false,
|
| 123 |
+
"special": false
|
| 124 |
+
},
|
| 125 |
+
"151658": {
|
| 126 |
+
"content": "</tool_call>",
|
| 127 |
+
"lstrip": false,
|
| 128 |
+
"normalized": false,
|
| 129 |
+
"rstrip": false,
|
| 130 |
+
"single_word": false,
|
| 131 |
+
"special": false
|
| 132 |
+
},
|
| 133 |
+
"151659": {
|
| 134 |
+
"content": "<|fim_prefix|>",
|
| 135 |
+
"lstrip": false,
|
| 136 |
+
"normalized": false,
|
| 137 |
+
"rstrip": false,
|
| 138 |
+
"single_word": false,
|
| 139 |
+
"special": false
|
| 140 |
+
},
|
| 141 |
+
"151660": {
|
| 142 |
+
"content": "<|fim_middle|>",
|
| 143 |
+
"lstrip": false,
|
| 144 |
+
"normalized": false,
|
| 145 |
+
"rstrip": false,
|
| 146 |
+
"single_word": false,
|
| 147 |
+
"special": false
|
| 148 |
+
},
|
| 149 |
+
"151661": {
|
| 150 |
+
"content": "<|fim_suffix|>",
|
| 151 |
+
"lstrip": false,
|
| 152 |
+
"normalized": false,
|
| 153 |
+
"rstrip": false,
|
| 154 |
+
"single_word": false,
|
| 155 |
+
"special": false
|
| 156 |
+
},
|
| 157 |
+
"151662": {
|
| 158 |
+
"content": "<|fim_pad|>",
|
| 159 |
+
"lstrip": false,
|
| 160 |
+
"normalized": false,
|
| 161 |
+
"rstrip": false,
|
| 162 |
+
"single_word": false,
|
| 163 |
+
"special": false
|
| 164 |
+
},
|
| 165 |
+
"151663": {
|
| 166 |
+
"content": "<|repo_name|>",
|
| 167 |
+
"lstrip": false,
|
| 168 |
+
"normalized": false,
|
| 169 |
+
"rstrip": false,
|
| 170 |
+
"single_word": false,
|
| 171 |
+
"special": false
|
| 172 |
+
},
|
| 173 |
+
"151664": {
|
| 174 |
+
"content": "<|file_sep|>",
|
| 175 |
+
"lstrip": false,
|
| 176 |
+
"normalized": false,
|
| 177 |
+
"rstrip": false,
|
| 178 |
+
"single_word": false,
|
| 179 |
+
"special": false
|
| 180 |
+
},
|
| 181 |
+
"151665": {
|
| 182 |
+
"content": "<tool_response>",
|
| 183 |
+
"lstrip": false,
|
| 184 |
+
"normalized": false,
|
| 185 |
+
"rstrip": false,
|
| 186 |
+
"single_word": false,
|
| 187 |
+
"special": false
|
| 188 |
+
},
|
| 189 |
+
"151666": {
|
| 190 |
+
"content": "</tool_response>",
|
| 191 |
+
"lstrip": false,
|
| 192 |
+
"normalized": false,
|
| 193 |
+
"rstrip": false,
|
| 194 |
+
"single_word": false,
|
| 195 |
+
"special": false
|
| 196 |
+
},
|
| 197 |
+
"151667": {
|
| 198 |
+
"content": "<think>",
|
| 199 |
+
"lstrip": false,
|
| 200 |
+
"normalized": false,
|
| 201 |
+
"rstrip": false,
|
| 202 |
+
"single_word": false,
|
| 203 |
+
"special": false
|
| 204 |
+
},
|
| 205 |
+
"151668": {
|
| 206 |
+
"content": "</think>",
|
| 207 |
+
"lstrip": false,
|
| 208 |
+
"normalized": false,
|
| 209 |
+
"rstrip": false,
|
| 210 |
+
"single_word": false,
|
| 211 |
+
"special": false
|
| 212 |
+
},
|
| 213 |
+
"151669": {
|
| 214 |
+
"content": "<|audio_start|>",
|
| 215 |
+
"lstrip": false,
|
| 216 |
+
"normalized": false,
|
| 217 |
+
"rstrip": false,
|
| 218 |
+
"single_word": false,
|
| 219 |
+
"special": true
|
| 220 |
+
},
|
| 221 |
+
"151670": {
|
| 222 |
+
"content": "<|audio_end|>",
|
| 223 |
+
"lstrip": false,
|
| 224 |
+
"normalized": false,
|
| 225 |
+
"rstrip": false,
|
| 226 |
+
"single_word": false,
|
| 227 |
+
"special": true
|
| 228 |
+
},
|
| 229 |
+
"151671": {
|
| 230 |
+
"content": "<tts_pad>",
|
| 231 |
+
"lstrip": false,
|
| 232 |
+
"normalized": false,
|
| 233 |
+
"rstrip": false,
|
| 234 |
+
"single_word": false,
|
| 235 |
+
"special": true
|
| 236 |
+
},
|
| 237 |
+
"151672": {
|
| 238 |
+
"content": "<tts_text_bos>",
|
| 239 |
+
"lstrip": false,
|
| 240 |
+
"normalized": false,
|
| 241 |
+
"rstrip": false,
|
| 242 |
+
"single_word": false,
|
| 243 |
+
"special": true
|
| 244 |
+
},
|
| 245 |
+
"151673": {
|
| 246 |
+
"content": "<tts_text_eod>",
|
| 247 |
+
"lstrip": false,
|
| 248 |
+
"normalized": false,
|
| 249 |
+
"rstrip": false,
|
| 250 |
+
"single_word": false,
|
| 251 |
+
"special": true
|
| 252 |
+
},
|
| 253 |
+
"151674": {
|
| 254 |
+
"content": "<tts_text_bos_single>",
|
| 255 |
+
"lstrip": false,
|
| 256 |
+
"normalized": false,
|
| 257 |
+
"rstrip": false,
|
| 258 |
+
"single_word": false,
|
| 259 |
+
"special": true
|
| 260 |
+
},
|
| 261 |
+
"151675": {
|
| 262 |
+
"content": "<non_speech>",
|
| 263 |
+
"lstrip": false,
|
| 264 |
+
"normalized": false,
|
| 265 |
+
"rstrip": false,
|
| 266 |
+
"single_word": false,
|
| 267 |
+
"special": false
|
| 268 |
+
},
|
| 269 |
+
"151676": {
|
| 270 |
+
"content": "<|audio_pad|>",
|
| 271 |
+
"lstrip": false,
|
| 272 |
+
"normalized": false,
|
| 273 |
+
"rstrip": false,
|
| 274 |
+
"single_word": false,
|
| 275 |
+
"special": true
|
| 276 |
+
},
|
| 277 |
+
"151677": {
|
| 278 |
+
"content": "<blank1>",
|
| 279 |
+
"lstrip": false,
|
| 280 |
+
"normalized": false,
|
| 281 |
+
"rstrip": false,
|
| 282 |
+
"single_word": false,
|
| 283 |
+
"special": true
|
| 284 |
+
},
|
| 285 |
+
"151678": {
|
| 286 |
+
"content": "<blank2>",
|
| 287 |
+
"lstrip": false,
|
| 288 |
+
"normalized": false,
|
| 289 |
+
"rstrip": false,
|
| 290 |
+
"single_word": false,
|
| 291 |
+
"special": true
|
| 292 |
+
},
|
| 293 |
+
"151679": {
|
| 294 |
+
"content": "<blank3>",
|
| 295 |
+
"lstrip": false,
|
| 296 |
+
"normalized": false,
|
| 297 |
+
"rstrip": false,
|
| 298 |
+
"single_word": false,
|
| 299 |
+
"special": true
|
| 300 |
+
},
|
| 301 |
+
"151680": {
|
| 302 |
+
"content": "<blank4>",
|
| 303 |
+
"lstrip": false,
|
| 304 |
+
"normalized": false,
|
| 305 |
+
"rstrip": false,
|
| 306 |
+
"single_word": false,
|
| 307 |
+
"special": true
|
| 308 |
+
},
|
| 309 |
+
"151681": {
|
| 310 |
+
"content": "<blank5>",
|
| 311 |
+
"lstrip": false,
|
| 312 |
+
"normalized": false,
|
| 313 |
+
"rstrip": false,
|
| 314 |
+
"single_word": false,
|
| 315 |
+
"special": true
|
| 316 |
+
},
|
| 317 |
+
"151682": {
|
| 318 |
+
"content": "<blank6>",
|
| 319 |
+
"lstrip": false,
|
| 320 |
+
"normalized": false,
|
| 321 |
+
"rstrip": false,
|
| 322 |
+
"single_word": false,
|
| 323 |
+
"special": true
|
| 324 |
+
},
|
| 325 |
+
"151683": {
|
| 326 |
+
"content": "<blank7>",
|
| 327 |
+
"lstrip": false,
|
| 328 |
+
"normalized": false,
|
| 329 |
+
"rstrip": false,
|
| 330 |
+
"single_word": false,
|
| 331 |
+
"special": true
|
| 332 |
+
},
|
| 333 |
+
"151684": {
|
| 334 |
+
"content": "<blank8>",
|
| 335 |
+
"lstrip": false,
|
| 336 |
+
"normalized": false,
|
| 337 |
+
"rstrip": false,
|
| 338 |
+
"single_word": false,
|
| 339 |
+
"special": true
|
| 340 |
+
},
|
| 341 |
+
"151685": {
|
| 342 |
+
"content": "<blank9>",
|
| 343 |
+
"lstrip": false,
|
| 344 |
+
"normalized": false,
|
| 345 |
+
"rstrip": false,
|
| 346 |
+
"single_word": false,
|
| 347 |
+
"special": true
|
| 348 |
+
},
|
| 349 |
+
"151686": {
|
| 350 |
+
"content": "<blank10>",
|
| 351 |
+
"lstrip": false,
|
| 352 |
+
"normalized": false,
|
| 353 |
+
"rstrip": false,
|
| 354 |
+
"single_word": false,
|
| 355 |
+
"special": true
|
| 356 |
+
},
|
| 357 |
+
"151687": {
|
| 358 |
+
"content": "<blank11>",
|
| 359 |
+
"lstrip": false,
|
| 360 |
+
"normalized": false,
|
| 361 |
+
"rstrip": false,
|
| 362 |
+
"single_word": false,
|
| 363 |
+
"special": true
|
| 364 |
+
},
|
| 365 |
+
"151688": {
|
| 366 |
+
"content": "<blank12>",
|
| 367 |
+
"lstrip": false,
|
| 368 |
+
"normalized": false,
|
| 369 |
+
"rstrip": false,
|
| 370 |
+
"single_word": false,
|
| 371 |
+
"special": true
|
| 372 |
+
},
|
| 373 |
+
"151689": {
|
| 374 |
+
"content": "<blank13>",
|
| 375 |
+
"lstrip": false,
|
| 376 |
+
"normalized": false,
|
| 377 |
+
"rstrip": false,
|
| 378 |
+
"single_word": false,
|
| 379 |
+
"special": true
|
| 380 |
+
},
|
| 381 |
+
"151690": {
|
| 382 |
+
"content": "<blank14>",
|
| 383 |
+
"lstrip": false,
|
| 384 |
+
"normalized": false,
|
| 385 |
+
"rstrip": false,
|
| 386 |
+
"single_word": false,
|
| 387 |
+
"special": true
|
| 388 |
+
},
|
| 389 |
+
"151691": {
|
| 390 |
+
"content": "<blank15>",
|
| 391 |
+
"lstrip": false,
|
| 392 |
+
"normalized": false,
|
| 393 |
+
"rstrip": false,
|
| 394 |
+
"single_word": false,
|
| 395 |
+
"special": true
|
| 396 |
+
},
|
| 397 |
+
"151692": {
|
| 398 |
+
"content": "<blank16>",
|
| 399 |
+
"lstrip": false,
|
| 400 |
+
"normalized": false,
|
| 401 |
+
"rstrip": false,
|
| 402 |
+
"single_word": false,
|
| 403 |
+
"special": true
|
| 404 |
+
},
|
| 405 |
+
"151693": {
|
| 406 |
+
"content": "<blank17>",
|
| 407 |
+
"lstrip": false,
|
| 408 |
+
"normalized": false,
|
| 409 |
+
"rstrip": false,
|
| 410 |
+
"single_word": false,
|
| 411 |
+
"special": true
|
| 412 |
+
},
|
| 413 |
+
"151694": {
|
| 414 |
+
"content": "<blank18>",
|
| 415 |
+
"lstrip": false,
|
| 416 |
+
"normalized": false,
|
| 417 |
+
"rstrip": false,
|
| 418 |
+
"single_word": false,
|
| 419 |
+
"special": true
|
| 420 |
+
},
|
| 421 |
+
"151695": {
|
| 422 |
+
"content": "<blank19>",
|
| 423 |
+
"lstrip": false,
|
| 424 |
+
"normalized": false,
|
| 425 |
+
"rstrip": false,
|
| 426 |
+
"single_word": false,
|
| 427 |
+
"special": true
|
| 428 |
+
},
|
| 429 |
+
"151696": {
|
| 430 |
+
"content": "<blank20>",
|
| 431 |
+
"lstrip": false,
|
| 432 |
+
"normalized": false,
|
| 433 |
+
"rstrip": false,
|
| 434 |
+
"single_word": false,
|
| 435 |
+
"special": true
|
| 436 |
+
},
|
| 437 |
+
"151697": {
|
| 438 |
+
"content": "<blank21>",
|
| 439 |
+
"lstrip": false,
|
| 440 |
+
"normalized": false,
|
| 441 |
+
"rstrip": false,
|
| 442 |
+
"single_word": false,
|
| 443 |
+
"special": true
|
| 444 |
+
},
|
| 445 |
+
"151698": {
|
| 446 |
+
"content": "<blank22>",
|
| 447 |
+
"lstrip": false,
|
| 448 |
+
"normalized": false,
|
| 449 |
+
"rstrip": false,
|
| 450 |
+
"single_word": false,
|
| 451 |
+
"special": true
|
| 452 |
+
},
|
| 453 |
+
"151699": {
|
| 454 |
+
"content": "<blank23>",
|
| 455 |
+
"lstrip": false,
|
| 456 |
+
"normalized": false,
|
| 457 |
+
"rstrip": false,
|
| 458 |
+
"single_word": false,
|
| 459 |
+
"special": true
|
| 460 |
+
},
|
| 461 |
+
"151700": {
|
| 462 |
+
"content": "<blank24>",
|
| 463 |
+
"lstrip": false,
|
| 464 |
+
"normalized": false,
|
| 465 |
+
"rstrip": false,
|
| 466 |
+
"single_word": false,
|
| 467 |
+
"special": true
|
| 468 |
+
},
|
| 469 |
+
"151701": {
|
| 470 |
+
"content": "<blank25>",
|
| 471 |
+
"lstrip": false,
|
| 472 |
+
"normalized": false,
|
| 473 |
+
"rstrip": false,
|
| 474 |
+
"single_word": false,
|
| 475 |
+
"special": true
|
| 476 |
+
},
|
| 477 |
+
"151702": {
|
| 478 |
+
"content": "<blank26>",
|
| 479 |
+
"lstrip": false,
|
| 480 |
+
"normalized": false,
|
| 481 |
+
"rstrip": false,
|
| 482 |
+
"single_word": false,
|
| 483 |
+
"special": true
|
| 484 |
+
},
|
| 485 |
+
"151703": {
|
| 486 |
+
"content": "<blank27>",
|
| 487 |
+
"lstrip": false,
|
| 488 |
+
"normalized": false,
|
| 489 |
+
"rstrip": false,
|
| 490 |
+
"single_word": false,
|
| 491 |
+
"special": true
|
| 492 |
+
},
|
| 493 |
+
"151704": {
|
| 494 |
+
"content": "<asr_text>",
|
| 495 |
+
"lstrip": false,
|
| 496 |
+
"normalized": false,
|
| 497 |
+
"rstrip": false,
|
| 498 |
+
"single_word": false,
|
| 499 |
+
"special": false
|
| 500 |
+
}
|
| 501 |
+
},
|
| 502 |
+
"additional_special_tokens": [
|
| 503 |
+
"<|im_start|>",
|
| 504 |
+
"<|im_end|>",
|
| 505 |
+
"<|object_ref_start|>",
|
| 506 |
+
"<|object_ref_end|>",
|
| 507 |
+
"<|box_start|>",
|
| 508 |
+
"<|box_end|>",
|
| 509 |
+
"<|quad_start|>",
|
| 510 |
+
"<|quad_end|>",
|
| 511 |
+
"<|vision_start|>",
|
| 512 |
+
"<|vision_end|>",
|
| 513 |
+
"<|vision_pad|>",
|
| 514 |
+
"<|image_pad|>",
|
| 515 |
+
"<|video_pad|>",
|
| 516 |
+
"<|audio_start|>",
|
| 517 |
+
"<|audio_end|>",
|
| 518 |
+
"<tts_pad>",
|
| 519 |
+
"<tts_text_bos>",
|
| 520 |
+
"<tts_text_bos_single>",
|
| 521 |
+
"<|audio_pad|>"
|
| 522 |
+
],
|
| 523 |
+
"audio_bos_token": "<|audio_start|>",
|
| 524 |
+
"audio_eos_token": "<|audio_end|>",
|
| 525 |
+
"audio_token": "<|audio_pad|>",
|
| 526 |
+
"bos_token": null,
|
| 527 |
+
"clean_up_tokenization_spaces": false,
|
| 528 |
+
"eos_token": "<|im_end|>",
|
| 529 |
+
"errors": "replace",
|
| 530 |
+
"extra_special_tokens": {
|
| 531 |
+
"audio_bos_token": "<|audio_start|>",
|
| 532 |
+
"audio_eos_token": "<|audio_end|>",
|
| 533 |
+
"audio_token": "<|audio_pad|>",
|
| 534 |
+
"image_token": "<|image_pad|>",
|
| 535 |
+
"video_token": "<|video_pad|>",
|
| 536 |
+
"vision_bos_token": "<|vision_start|>",
|
| 537 |
+
"vision_eos_token": "<|vision_end|>"
|
| 538 |
+
},
|
| 539 |
+
"image_token": "<|image_pad|>",
|
| 540 |
+
"model_max_length": 131072,
|
| 541 |
+
"pad_token": "<|endoftext|>",
|
| 542 |
+
"processor_class": "Qwen3ASRProcessor",
|
| 543 |
+
"split_special_tokens": false,
|
| 544 |
+
"tokenizer_class": "Qwen2Tokenizer",
|
| 545 |
+
"unk_token": null,
|
| 546 |
+
"video_token": "<|video_pad|>",
|
| 547 |
+
"vision_bos_token": "<|vision_start|>",
|
| 548 |
+
"vision_eos_token": "<|vision_end|>"
|
| 549 |
+
}
|
Qwen3-ASR-WSM-Min/trainer_state.json
ADDED
|
@@ -0,0 +1,2042 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 200,
|
| 3 |
+
"best_metric": 1.01447761,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 2.958121475455302,
|
| 6 |
+
"eval_steps": 200,
|
| 7 |
+
"global_step": 16000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.00018489414810021263,
|
| 14 |
+
"grad_norm": 111.71553039550781,
|
| 15 |
+
"learning_rate": 2.3094688221709007e-08,
|
| 16 |
+
"loss": 2.648883581161499,
|
| 17 |
+
"step": 1,
|
| 18 |
+
"token_acc": 0.6067455761846324
|
| 19 |
+
},
|
| 20 |
+
{
|
| 21 |
+
"epoch": 0.018489414810021264,
|
| 22 |
+
"grad_norm": 2.922001361846924,
|
| 23 |
+
"learning_rate": 2.3094688221709006e-06,
|
| 24 |
+
"loss": 1.1922963460286458,
|
| 25 |
+
"step": 100,
|
| 26 |
+
"token_acc": 0.7539609103342668
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.03697882962004253,
|
| 30 |
+
"grad_norm": 3.0476975440979004,
|
| 31 |
+
"learning_rate": 4.618937644341801e-06,
|
| 32 |
+
"loss": 0.41063663482666013,
|
| 33 |
+
"step": 200,
|
| 34 |
+
"token_acc": 0.8698793321763938
|
| 35 |
+
},
|
| 36 |
+
{
|
| 37 |
+
"epoch": 0.03697882962004253,
|
| 38 |
+
"eval_loss": 1.0144776105880737,
|
| 39 |
+
"eval_runtime": 51.4031,
|
| 40 |
+
"eval_samples_per_second": 301.188,
|
| 41 |
+
"eval_steps_per_second": 0.603,
|
| 42 |
+
"eval_token_acc": 0.8180929835105752,
|
| 43 |
+
"step": 200
|
| 44 |
+
},
|
| 45 |
+
{
|
| 46 |
+
"epoch": 0.05546824443006379,
|
| 47 |
+
"grad_norm": 6.208777904510498,
|
| 48 |
+
"learning_rate": 6.928406466512702e-06,
|
| 49 |
+
"loss": 0.35355266571044924,
|
| 50 |
+
"step": 300,
|
| 51 |
+
"token_acc": 0.8843683925385469
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.07395765924008506,
|
| 55 |
+
"grad_norm": 5.785989761352539,
|
| 56 |
+
"learning_rate": 9.237875288683602e-06,
|
| 57 |
+
"loss": 0.32728473663330077,
|
| 58 |
+
"step": 400,
|
| 59 |
+
"token_acc": 0.8909257197704288
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.07395765924008506,
|
| 63 |
+
"eval_loss": 1.111476182937622,
|
| 64 |
+
"eval_runtime": 35.9267,
|
| 65 |
+
"eval_samples_per_second": 430.933,
|
| 66 |
+
"eval_steps_per_second": 0.863,
|
| 67 |
+
"eval_token_acc": 0.8225046099247284,
|
| 68 |
+
"step": 400
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.09244707405010631,
|
| 72 |
+
"grad_norm": 3.5521814823150635,
|
| 73 |
+
"learning_rate": 9.999753628103762e-06,
|
| 74 |
+
"loss": 0.31058956146240235,
|
| 75 |
+
"step": 500,
|
| 76 |
+
"token_acc": 0.8952305663901123
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"epoch": 0.11093648886012758,
|
| 80 |
+
"grad_norm": 2.6671371459960938,
|
| 81 |
+
"learning_rate": 9.998469420433913e-06,
|
| 82 |
+
"loss": 0.2931631088256836,
|
| 83 |
+
"step": 600,
|
| 84 |
+
"token_acc": 0.9003683912010906
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.11093648886012758,
|
| 88 |
+
"eval_loss": 1.2274305820465088,
|
| 89 |
+
"eval_runtime": 52.5536,
|
| 90 |
+
"eval_samples_per_second": 294.595,
|
| 91 |
+
"eval_steps_per_second": 0.59,
|
| 92 |
+
"eval_token_acc": 0.8256119649523062,
|
| 93 |
+
"step": 600
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.12942590367014883,
|
| 97 |
+
"grad_norm": 2.6913869380950928,
|
| 98 |
+
"learning_rate": 9.99608789044425e-06,
|
| 99 |
+
"loss": 0.281595344543457,
|
| 100 |
+
"step": 700,
|
| 101 |
+
"token_acc": 0.9035918171295385
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.14791531848017012,
|
| 105 |
+
"grad_norm": 3.1163330078125,
|
| 106 |
+
"learning_rate": 9.992609560956019e-06,
|
| 107 |
+
"loss": 0.2716915512084961,
|
| 108 |
+
"step": 800,
|
| 109 |
+
"token_acc": 0.9061136695566686
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 0.14791531848017012,
|
| 113 |
+
"eval_loss": 1.3400822877883911,
|
| 114 |
+
"eval_runtime": 38.5226,
|
| 115 |
+
"eval_samples_per_second": 401.894,
|
| 116 |
+
"eval_steps_per_second": 0.805,
|
| 117 |
+
"eval_token_acc": 0.8278852404724815,
|
| 118 |
+
"step": 800
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 0.16640473329019137,
|
| 122 |
+
"grad_norm": 3.4362285137176514,
|
| 123 |
+
"learning_rate": 9.988035195572685e-06,
|
| 124 |
+
"loss": 0.26148880004882813,
|
| 125 |
+
"step": 900,
|
| 126 |
+
"token_acc": 0.9093240404437027
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 0.18489414810021262,
|
| 130 |
+
"grad_norm": 2.838069438934326,
|
| 131 |
+
"learning_rate": 9.982365798512305e-06,
|
| 132 |
+
"loss": 0.25565671920776367,
|
| 133 |
+
"step": 1000,
|
| 134 |
+
"token_acc": 0.9107939315310666
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.18489414810021262,
|
| 138 |
+
"eval_loss": 1.4232231378555298,
|
| 139 |
+
"eval_runtime": 37.2234,
|
| 140 |
+
"eval_samples_per_second": 415.921,
|
| 141 |
+
"eval_steps_per_second": 0.833,
|
| 142 |
+
"eval_token_acc": 0.8293285251102906,
|
| 143 |
+
"step": 1000
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.20338356291023388,
|
| 147 |
+
"grad_norm": 2.627117156982422,
|
| 148 |
+
"learning_rate": 9.97560261438706e-06,
|
| 149 |
+
"loss": 0.24876928329467773,
|
| 150 |
+
"step": 1100,
|
| 151 |
+
"token_acc": 0.9130002779102413
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 0.22187297772025516,
|
| 155 |
+
"grad_norm": 3.118025779724121,
|
| 156 |
+
"learning_rate": 9.967747127930028e-06,
|
| 157 |
+
"loss": 0.24220096588134765,
|
| 158 |
+
"step": 1200,
|
| 159 |
+
"token_acc": 0.9147149870604054
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.22187297772025516,
|
| 163 |
+
"eval_loss": 1.451097846031189,
|
| 164 |
+
"eval_runtime": 38.088,
|
| 165 |
+
"eval_samples_per_second": 406.48,
|
| 166 |
+
"eval_steps_per_second": 0.814,
|
| 167 |
+
"eval_token_acc": 0.8307309234977369,
|
| 168 |
+
"step": 1200
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 0.24036239253027641,
|
| 172 |
+
"grad_norm": 2.72821044921875,
|
| 173 |
+
"learning_rate": 9.958801063669242e-06,
|
| 174 |
+
"loss": 0.23826725006103516,
|
| 175 |
+
"step": 1300,
|
| 176 |
+
"token_acc": 0.9160493730222916
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.25885180734029767,
|
| 180 |
+
"grad_norm": 3.0269107818603516,
|
| 181 |
+
"learning_rate": 9.948766385549092e-06,
|
| 182 |
+
"loss": 0.23318710327148437,
|
| 183 |
+
"step": 1400,
|
| 184 |
+
"token_acc": 0.9177872612639156
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.25885180734029767,
|
| 188 |
+
"eval_loss": 1.4914048910140991,
|
| 189 |
+
"eval_runtime": 39.0073,
|
| 190 |
+
"eval_samples_per_second": 396.9,
|
| 191 |
+
"eval_steps_per_second": 0.795,
|
| 192 |
+
"eval_token_acc": 0.8324072597626144,
|
| 193 |
+
"step": 1400
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.2773412221503189,
|
| 197 |
+
"grad_norm": 2.6068923473358154,
|
| 198 |
+
"learning_rate": 9.937645296499185e-06,
|
| 199 |
+
"loss": 0.22903181076049806,
|
| 200 |
+
"step": 1500,
|
| 201 |
+
"token_acc": 0.9190053995875763
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.29583063696034023,
|
| 205 |
+
"grad_norm": 2.84269642829895,
|
| 206 |
+
"learning_rate": 9.92544023795073e-06,
|
| 207 |
+
"loss": 0.22664697647094725,
|
| 208 |
+
"step": 1600,
|
| 209 |
+
"token_acc": 0.9196992125237452
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 0.29583063696034023,
|
| 213 |
+
"eval_loss": 1.4975084066390991,
|
| 214 |
+
"eval_runtime": 36.0058,
|
| 215 |
+
"eval_samples_per_second": 429.987,
|
| 216 |
+
"eval_steps_per_second": 0.861,
|
| 217 |
+
"eval_token_acc": 0.83302873076813,
|
| 218 |
+
"step": 1600
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.3143200517703615,
|
| 222 |
+
"grad_norm": 2.2528085708618164,
|
| 223 |
+
"learning_rate": 9.912153889300562e-06,
|
| 224 |
+
"loss": 0.2265909194946289,
|
| 225 |
+
"step": 1700,
|
| 226 |
+
"token_acc": 0.9194939206232738
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.33280946658038274,
|
| 230 |
+
"grad_norm": 2.103715658187866,
|
| 231 |
+
"learning_rate": 9.897789167322938e-06,
|
| 232 |
+
"loss": 0.22183643341064452,
|
| 233 |
+
"step": 1800,
|
| 234 |
+
"token_acc": 0.9211902545521418
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.33280946658038274,
|
| 238 |
+
"eval_loss": 1.5258921384811401,
|
| 239 |
+
"eval_runtime": 36.8679,
|
| 240 |
+
"eval_samples_per_second": 419.932,
|
| 241 |
+
"eval_steps_per_second": 0.841,
|
| 242 |
+
"eval_token_acc": 0.8333926183963595,
|
| 243 |
+
"step": 1800
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.351298881390404,
|
| 247 |
+
"grad_norm": 2.9496121406555176,
|
| 248 |
+
"learning_rate": 9.882349225529208e-06,
|
| 249 |
+
"loss": 0.2201910972595215,
|
| 250 |
+
"step": 1900,
|
| 251 |
+
"token_acc": 0.921664994905861
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.36978829620042525,
|
| 255 |
+
"grad_norm": 3.201491594314575,
|
| 256 |
+
"learning_rate": 9.86583745347552e-06,
|
| 257 |
+
"loss": 0.21536655426025392,
|
| 258 |
+
"step": 2000,
|
| 259 |
+
"token_acc": 0.9233945148873542
|
| 260 |
+
},
|
| 261 |
+
{
|
| 262 |
+
"epoch": 0.36978829620042525,
|
| 263 |
+
"eval_loss": 1.523385763168335,
|
| 264 |
+
"eval_runtime": 35.885,
|
| 265 |
+
"eval_samples_per_second": 431.434,
|
| 266 |
+
"eval_steps_per_second": 0.864,
|
| 267 |
+
"eval_token_acc": 0.8340672415273468,
|
| 268 |
+
"step": 2000
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.3882777110104465,
|
| 272 |
+
"grad_norm": 2.0704493522644043,
|
| 273 |
+
"learning_rate": 9.848257476018699e-06,
|
| 274 |
+
"loss": 0.21449897766113282,
|
| 275 |
+
"step": 2100,
|
| 276 |
+
"token_acc": 0.9235176346647841
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.40676712582046776,
|
| 280 |
+
"grad_norm": 2.3089756965637207,
|
| 281 |
+
"learning_rate": 9.829613152520487e-06,
|
| 282 |
+
"loss": 0.2108938980102539,
|
| 283 |
+
"step": 2200,
|
| 284 |
+
"token_acc": 0.9244604882160045
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.40676712582046776,
|
| 288 |
+
"eval_loss": 1.557509422302246,
|
| 289 |
+
"eval_runtime": 36.5228,
|
| 290 |
+
"eval_samples_per_second": 423.9,
|
| 291 |
+
"eval_steps_per_second": 0.849,
|
| 292 |
+
"eval_token_acc": 0.8355882100408454,
|
| 293 |
+
"step": 2200
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 0.42525654063048907,
|
| 297 |
+
"grad_norm": 2.259549856185913,
|
| 298 |
+
"learning_rate": 9.809908576000288e-06,
|
| 299 |
+
"loss": 0.20940305709838866,
|
| 300 |
+
"step": 2300,
|
| 301 |
+
"token_acc": 0.9251665532475329
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 0.4437459554405103,
|
| 305 |
+
"grad_norm": 2.1792192459106445,
|
| 306 |
+
"learning_rate": 9.789148072236612e-06,
|
| 307 |
+
"loss": 0.2089914321899414,
|
| 308 |
+
"step": 2400,
|
| 309 |
+
"token_acc": 0.925105847080245
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.4437459554405103,
|
| 313 |
+
"eval_loss": 1.5759638547897339,
|
| 314 |
+
"eval_runtime": 35.5351,
|
| 315 |
+
"eval_samples_per_second": 435.682,
|
| 316 |
+
"eval_steps_per_second": 0.872,
|
| 317 |
+
"eval_token_acc": 0.8355187034152285,
|
| 318 |
+
"step": 2400
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.4622353702505316,
|
| 322 |
+
"grad_norm": 2.2059366703033447,
|
| 323 |
+
"learning_rate": 9.767336198817441e-06,
|
| 324 |
+
"loss": 0.20587841033935547,
|
| 325 |
+
"step": 2500,
|
| 326 |
+
"token_acc": 0.9262170052348255
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.48072478506055283,
|
| 330 |
+
"grad_norm": 2.698324203491211,
|
| 331 |
+
"learning_rate": 9.744477744139686e-06,
|
| 332 |
+
"loss": 0.2041876220703125,
|
| 333 |
+
"step": 2600,
|
| 334 |
+
"token_acc": 0.9267364721054939
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.48072478506055283,
|
| 338 |
+
"eval_loss": 1.582732915878296,
|
| 339 |
+
"eval_runtime": 34.3654,
|
| 340 |
+
"eval_samples_per_second": 450.512,
|
| 341 |
+
"eval_steps_per_second": 0.902,
|
| 342 |
+
"eval_token_acc": 0.8362710104219052,
|
| 343 |
+
"step": 2600
|
| 344 |
+
},
|
| 345 |
+
{
|
| 346 |
+
"epoch": 0.4992141998705741,
|
| 347 |
+
"grad_norm": 2.199079990386963,
|
| 348 |
+
"learning_rate": 9.72057772635799e-06,
|
| 349 |
+
"loss": 0.20334808349609376,
|
| 350 |
+
"step": 2700,
|
| 351 |
+
"token_acc": 0.9270570224893226
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.5177036146805953,
|
| 355 |
+
"grad_norm": 2.6320109367370605,
|
| 356 |
+
"learning_rate": 9.69564139228307e-06,
|
| 357 |
+
"loss": 0.20141841888427733,
|
| 358 |
+
"step": 2800,
|
| 359 |
+
"token_acc": 0.9276200947252605
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.5177036146805953,
|
| 363 |
+
"eval_loss": 1.5634138584136963,
|
| 364 |
+
"eval_runtime": 34.7284,
|
| 365 |
+
"eval_samples_per_second": 445.802,
|
| 366 |
+
"eval_steps_per_second": 0.893,
|
| 367 |
+
"eval_token_acc": 0.8363691374227761,
|
| 368 |
+
"step": 2800
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 0.5361930294906166,
|
| 372 |
+
"grad_norm": 2.713134527206421,
|
| 373 |
+
"learning_rate": 9.6696742162299e-06,
|
| 374 |
+
"loss": 0.20071035385131836,
|
| 375 |
+
"step": 2900,
|
| 376 |
+
"token_acc": 0.9278191630026814
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 0.5546824443006378,
|
| 380 |
+
"grad_norm": 2.115159749984741,
|
| 381 |
+
"learning_rate": 9.642681898815898e-06,
|
| 382 |
+
"loss": 0.19931299209594727,
|
| 383 |
+
"step": 3000,
|
| 384 |
+
"token_acc": 0.9283823865202623
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.5546824443006378,
|
| 388 |
+
"eval_loss": 1.595649242401123,
|
| 389 |
+
"eval_runtime": 35.136,
|
| 390 |
+
"eval_samples_per_second": 440.631,
|
| 391 |
+
"eval_steps_per_second": 0.882,
|
| 392 |
+
"eval_token_acc": 0.8369619880530377,
|
| 393 |
+
"step": 3000
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 0.5731718591106592,
|
| 397 |
+
"grad_norm": 1.989067554473877,
|
| 398 |
+
"learning_rate": 9.61467036570948e-06,
|
| 399 |
+
"loss": 0.19821651458740233,
|
| 400 |
+
"step": 3100,
|
| 401 |
+
"token_acc": 0.9286384459605592
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.5916612739206805,
|
| 405 |
+
"grad_norm": 2.211009979248047,
|
| 406 |
+
"learning_rate": 9.585645766329174e-06,
|
| 407 |
+
"loss": 0.1967957305908203,
|
| 408 |
+
"step": 3200,
|
| 409 |
+
"token_acc": 0.9290863017011062
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.5916612739206805,
|
| 413 |
+
"eval_loss": 1.6135554313659668,
|
| 414 |
+
"eval_runtime": 34.0447,
|
| 415 |
+
"eval_samples_per_second": 454.756,
|
| 416 |
+
"eval_steps_per_second": 0.911,
|
| 417 |
+
"eval_token_acc": 0.8372154828052875,
|
| 418 |
+
"step": 3200
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 0.6101506887307017,
|
| 422 |
+
"grad_norm": 2.5767476558685303,
|
| 423 |
+
"learning_rate": 9.555614472493645e-06,
|
| 424 |
+
"loss": 0.1952597427368164,
|
| 425 |
+
"step": 3300,
|
| 426 |
+
"token_acc": 0.9296881804891715
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 0.628640103540723,
|
| 430 |
+
"grad_norm": 1.970717430114746,
|
| 431 |
+
"learning_rate": 9.524583077022865e-06,
|
| 432 |
+
"loss": 0.1941541290283203,
|
| 433 |
+
"step": 3400,
|
| 434 |
+
"token_acc": 0.9300202592587643
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.628640103540723,
|
| 438 |
+
"eval_loss": 1.5975191593170166,
|
| 439 |
+
"eval_runtime": 33.7307,
|
| 440 |
+
"eval_samples_per_second": 458.989,
|
| 441 |
+
"eval_steps_per_second": 0.919,
|
| 442 |
+
"eval_token_acc": 0.8377674471851861,
|
| 443 |
+
"step": 3400
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.6471295183507442,
|
| 447 |
+
"grad_norm": 2.066835641860962,
|
| 448 |
+
"learning_rate": 9.492558392290778e-06,
|
| 449 |
+
"loss": 0.19371936798095704,
|
| 450 |
+
"step": 3500,
|
| 451 |
+
"token_acc": 0.9300272679782989
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.6656189331607655,
|
| 455 |
+
"grad_norm": 2.2377233505249023,
|
| 456 |
+
"learning_rate": 9.459547448729782e-06,
|
| 457 |
+
"loss": 0.1923367691040039,
|
| 458 |
+
"step": 3600,
|
| 459 |
+
"token_acc": 0.9305084599216811
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.6656189331607655,
|
| 463 |
+
"eval_loss": 1.6066784858703613,
|
| 464 |
+
"eval_runtime": 35.5317,
|
| 465 |
+
"eval_samples_per_second": 435.724,
|
| 466 |
+
"eval_steps_per_second": 0.872,
|
| 467 |
+
"eval_token_acc": 0.8376325225589887,
|
| 468 |
+
"step": 3600
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 0.6841083479707867,
|
| 472 |
+
"grad_norm": 2.3556344509124756,
|
| 473 |
+
"learning_rate": 9.425557493287317e-06,
|
| 474 |
+
"loss": 0.19046760559082032,
|
| 475 |
+
"step": 3700,
|
| 476 |
+
"token_acc": 0.9311673521410996
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.702597762780808,
|
| 480 |
+
"grad_norm": 2.414644241333008,
|
| 481 |
+
"learning_rate": 9.39059598783493e-06,
|
| 482 |
+
"loss": 0.19073116302490234,
|
| 483 |
+
"step": 3800,
|
| 484 |
+
"token_acc": 0.9309592717601893
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 0.702597762780808,
|
| 488 |
+
"eval_loss": 1.6400927305221558,
|
| 489 |
+
"eval_runtime": 34.4621,
|
| 490 |
+
"eval_samples_per_second": 449.247,
|
| 491 |
+
"eval_steps_per_second": 0.9,
|
| 492 |
+
"eval_token_acc": 0.8383930068157379,
|
| 493 |
+
"step": 3800
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.7210871775908293,
|
| 497 |
+
"grad_norm": 2.2784621715545654,
|
| 498 |
+
"learning_rate": 9.354670607530165e-06,
|
| 499 |
+
"loss": 0.18932226181030273,
|
| 500 |
+
"step": 3900,
|
| 501 |
+
"token_acc": 0.9314831107619795
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.7395765924008505,
|
| 505 |
+
"grad_norm": 2.5446369647979736,
|
| 506 |
+
"learning_rate": 9.317789239131613e-06,
|
| 507 |
+
"loss": 0.1888302230834961,
|
| 508 |
+
"step": 4000,
|
| 509 |
+
"token_acc": 0.9316500107527262
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 0.7395765924008505,
|
| 513 |
+
"eval_loss": 1.6447820663452148,
|
| 514 |
+
"eval_runtime": 39.6445,
|
| 515 |
+
"eval_samples_per_second": 390.521,
|
| 516 |
+
"eval_steps_per_second": 0.782,
|
| 517 |
+
"eval_token_acc": 0.8392107318229952,
|
| 518 |
+
"step": 4000
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 0.7580660072108718,
|
| 522 |
+
"grad_norm": 2.0726985931396484,
|
| 523 |
+
"learning_rate": 9.279959979267528e-06,
|
| 524 |
+
"loss": 0.18829454421997072,
|
| 525 |
+
"step": 4100,
|
| 526 |
+
"token_acc": 0.9317781501221714
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 0.776555422020893,
|
| 530 |
+
"grad_norm": 2.0762479305267334,
|
| 531 |
+
"learning_rate": 9.241191132658356e-06,
|
| 532 |
+
"loss": 0.18706449508666992,
|
| 533 |
+
"step": 4200,
|
| 534 |
+
"token_acc": 0.9323511335915298
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.776555422020893,
|
| 538 |
+
"eval_loss": 1.6254693269729614,
|
| 539 |
+
"eval_runtime": 34.9793,
|
| 540 |
+
"eval_samples_per_second": 442.605,
|
| 541 |
+
"eval_steps_per_second": 0.886,
|
| 542 |
+
"eval_token_acc": 0.8391248706972332,
|
| 543 |
+
"step": 4200
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 0.7950448368309143,
|
| 547 |
+
"grad_norm": 2.254326581954956,
|
| 548 |
+
"learning_rate": 9.201491210293583e-06,
|
| 549 |
+
"loss": 0.18593116760253905,
|
| 550 |
+
"step": 4300,
|
| 551 |
+
"token_acc": 0.9326580017632726
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 0.8135342516409355,
|
| 555 |
+
"grad_norm": 1.9395033121109009,
|
| 556 |
+
"learning_rate": 9.160868927563314e-06,
|
| 557 |
+
"loss": 0.18512523651123047,
|
| 558 |
+
"step": 4400,
|
| 559 |
+
"token_acc": 0.9329307504429487
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 0.8135342516409355,
|
| 563 |
+
"eval_loss": 1.639746069908142,
|
| 564 |
+
"eval_runtime": 35.9114,
|
| 565 |
+
"eval_samples_per_second": 431.116,
|
| 566 |
+
"eval_steps_per_second": 0.863,
|
| 567 |
+
"eval_token_acc": 0.8388468441947657,
|
| 568 |
+
"step": 4400
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 0.8320236664509568,
|
| 572 |
+
"grad_norm": 2.378365993499756,
|
| 573 |
+
"learning_rate": 9.119333202344953e-06,
|
| 574 |
+
"loss": 0.18431827545166016,
|
| 575 |
+
"step": 4500,
|
| 576 |
+
"token_acc": 0.9331190064239155
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.8505130812609781,
|
| 580 |
+
"grad_norm": 2.1460585594177246,
|
| 581 |
+
"learning_rate": 9.076893153045464e-06,
|
| 582 |
+
"loss": 0.18355430603027345,
|
| 583 |
+
"step": 4600,
|
| 584 |
+
"token_acc": 0.9334646427529134
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.8505130812609781,
|
| 588 |
+
"eval_loss": 1.6335515975952148,
|
| 589 |
+
"eval_runtime": 37.8269,
|
| 590 |
+
"eval_samples_per_second": 409.285,
|
| 591 |
+
"eval_steps_per_second": 0.82,
|
| 592 |
+
"eval_token_acc": 0.8393579223243015,
|
| 593 |
+
"step": 4600
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 0.8690024960709993,
|
| 597 |
+
"grad_norm": 2.1147706508636475,
|
| 598 |
+
"learning_rate": 9.033558096599575e-06,
|
| 599 |
+
"loss": 0.1834134864807129,
|
| 600 |
+
"step": 4700,
|
| 601 |
+
"token_acc": 0.9334649151317098
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 0.8874919108810206,
|
| 605 |
+
"grad_norm": 2.131558656692505,
|
| 606 |
+
"learning_rate": 8.989337546424432e-06,
|
| 607 |
+
"loss": 0.18236831665039063,
|
| 608 |
+
"step": 4800,
|
| 609 |
+
"token_acc": 0.9338352671802478
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.8874919108810206,
|
| 613 |
+
"eval_loss": 1.6568936109542847,
|
| 614 |
+
"eval_runtime": 38.5099,
|
| 615 |
+
"eval_samples_per_second": 402.027,
|
| 616 |
+
"eval_steps_per_second": 0.805,
|
| 617 |
+
"eval_token_acc": 0.8390063005711809,
|
| 618 |
+
"step": 4800
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.9059813256910418,
|
| 622 |
+
"grad_norm": 1.9892091751098633,
|
| 623 |
+
"learning_rate": 8.94424121033109e-06,
|
| 624 |
+
"loss": 0.18153335571289062,
|
| 625 |
+
"step": 4900,
|
| 626 |
+
"token_acc": 0.9341023295614975
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.9244707405010631,
|
| 630 |
+
"grad_norm": 2.218424081802368,
|
| 631 |
+
"learning_rate": 8.898278988393343e-06,
|
| 632 |
+
"loss": 0.1806512451171875,
|
| 633 |
+
"step": 5000,
|
| 634 |
+
"token_acc": 0.9344670985897315
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 0.9244707405010631,
|
| 638 |
+
"eval_loss": 1.6349440813064575,
|
| 639 |
+
"eval_runtime": 41.848,
|
| 640 |
+
"eval_samples_per_second": 369.958,
|
| 641 |
+
"eval_steps_per_second": 0.741,
|
| 642 |
+
"eval_token_acc": 0.8395419104509344,
|
| 643 |
+
"step": 5000
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 0.9429601553110845,
|
| 647 |
+
"grad_norm": 2.007549524307251,
|
| 648 |
+
"learning_rate": 8.851460970774355e-06,
|
| 649 |
+
"loss": 0.1798555374145508,
|
| 650 |
+
"step": 5100,
|
| 651 |
+
"token_acc": 0.9347298029461745
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 0.9614495701211057,
|
| 655 |
+
"grad_norm": 2.1956558227539062,
|
| 656 |
+
"learning_rate": 8.803797435511535e-06,
|
| 657 |
+
"loss": 0.1793169593811035,
|
| 658 |
+
"step": 5200,
|
| 659 |
+
"token_acc": 0.9346815577383201
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 0.9614495701211057,
|
| 663 |
+
"eval_loss": 1.6491081714630127,
|
| 664 |
+
"eval_runtime": 40.7497,
|
| 665 |
+
"eval_samples_per_second": 379.929,
|
| 666 |
+
"eval_steps_per_second": 0.761,
|
| 667 |
+
"eval_token_acc": 0.8397218099525311,
|
| 668 |
+
"step": 5200
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.979938984931127,
|
| 672 |
+
"grad_norm": 2.290909767150879,
|
| 673 |
+
"learning_rate": 8.75529884626019e-06,
|
| 674 |
+
"loss": 0.17874420166015625,
|
| 675 |
+
"step": 5300,
|
| 676 |
+
"token_acc": 0.9350539150980832
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 0.9984283997411482,
|
| 680 |
+
"grad_norm": 2.1777896881103516,
|
| 681 |
+
"learning_rate": 8.705975849996427e-06,
|
| 682 |
+
"loss": 0.1775716209411621,
|
| 683 |
+
"step": 5400,
|
| 684 |
+
"token_acc": 0.9353836269477259
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 0.9984283997411482,
|
| 688 |
+
"eval_loss": 1.671922206878662,
|
| 689 |
+
"eval_runtime": 41.8015,
|
| 690 |
+
"eval_samples_per_second": 370.369,
|
| 691 |
+
"eval_steps_per_second": 0.742,
|
| 692 |
+
"eval_token_acc": 0.8396032398264788,
|
| 693 |
+
"step": 5400
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 1.0168253674771193,
|
| 697 |
+
"grad_norm": 1.8627227544784546,
|
| 698 |
+
"learning_rate": 8.655839274679787e-06,
|
| 699 |
+
"loss": 0.13099777221679687,
|
| 700 |
+
"step": 5500,
|
| 701 |
+
"token_acc": 0.9543783937496227
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 1.0353147822871407,
|
| 705 |
+
"grad_norm": 1.8053191900253296,
|
| 706 |
+
"learning_rate": 8.604900126876186e-06,
|
| 707 |
+
"loss": 0.1243514633178711,
|
| 708 |
+
"step": 5600,
|
| 709 |
+
"token_acc": 0.9565782158320064
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 1.0353147822871407,
|
| 713 |
+
"eval_loss": 1.7374670505523682,
|
| 714 |
+
"eval_runtime": 37.9624,
|
| 715 |
+
"eval_samples_per_second": 407.824,
|
| 716 |
+
"eval_steps_per_second": 0.817,
|
| 717 |
+
"eval_token_acc": 0.8398158483283656,
|
| 718 |
+
"step": 5600
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 1.0538041970971619,
|
| 722 |
+
"grad_norm": 1.7985650300979614,
|
| 723 |
+
"learning_rate": 8.553169589341608e-06,
|
| 724 |
+
"loss": 0.12475324630737304,
|
| 725 |
+
"step": 5700,
|
| 726 |
+
"token_acc": 0.9563982881741734
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 1.072293611907183,
|
| 730 |
+
"grad_norm": 1.7619038820266724,
|
| 731 |
+
"learning_rate": 8.50065901856715e-06,
|
| 732 |
+
"loss": 0.12438088417053222,
|
| 733 |
+
"step": 5800,
|
| 734 |
+
"token_acc": 0.9563261001314413
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 1.072293611907183,
|
| 738 |
+
"eval_loss": 1.7655820846557617,
|
| 739 |
+
"eval_runtime": 41.3424,
|
| 740 |
+
"eval_samples_per_second": 374.483,
|
| 741 |
+
"eval_steps_per_second": 0.75,
|
| 742 |
+
"eval_token_acc": 0.8398035824532568,
|
| 743 |
+
"step": 5800
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 1.0907830267172045,
|
| 747 |
+
"grad_norm": 2.002814531326294,
|
| 748 |
+
"learning_rate": 8.447379942285896e-06,
|
| 749 |
+
"loss": 0.12517940521240234,
|
| 750 |
+
"step": 5900,
|
| 751 |
+
"token_acc": 0.9560996768526752
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 1.1092724415272257,
|
| 755 |
+
"grad_norm": 1.7221897840499878,
|
| 756 |
+
"learning_rate": 8.393344056942237e-06,
|
| 757 |
+
"loss": 0.12486957550048829,
|
| 758 |
+
"step": 6000,
|
| 759 |
+
"token_acc": 0.9561690507193046
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 1.1092724415272257,
|
| 763 |
+
"eval_loss": 1.7758432626724243,
|
| 764 |
+
"eval_runtime": 45.8318,
|
| 765 |
+
"eval_samples_per_second": 337.8,
|
| 766 |
+
"eval_steps_per_second": 0.676,
|
| 767 |
+
"eval_token_acc": 0.8398812663289462,
|
| 768 |
+
"step": 6000
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 1.127761856337247,
|
| 772 |
+
"grad_norm": 2.0354294776916504,
|
| 773 |
+
"learning_rate": 8.338563225124107e-06,
|
| 774 |
+
"loss": 0.124525146484375,
|
| 775 |
+
"step": 6100,
|
| 776 |
+
"token_acc": 0.9560097733366691
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 1.146251271147268,
|
| 780 |
+
"grad_norm": 1.7929569482803345,
|
| 781 |
+
"learning_rate": 8.283049472958785e-06,
|
| 782 |
+
"loss": 0.12382856369018555,
|
| 783 |
+
"step": 6200,
|
| 784 |
+
"token_acc": 0.9563822149749119
|
| 785 |
+
},
|
| 786 |
+
{
|
| 787 |
+
"epoch": 1.146251271147268,
|
| 788 |
+
"eval_loss": 1.8066223859786987,
|
| 789 |
+
"eval_runtime": 41.9458,
|
| 790 |
+
"eval_samples_per_second": 369.095,
|
| 791 |
+
"eval_steps_per_second": 0.739,
|
| 792 |
+
"eval_token_acc": 0.8398362914535471,
|
| 793 |
+
"step": 6200
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 1.1647406859572895,
|
| 797 |
+
"grad_norm": 1.8996866941452026,
|
| 798 |
+
"learning_rate": 8.226814987472782e-06,
|
| 799 |
+
"loss": 0.12526337623596193,
|
| 800 |
+
"step": 6300,
|
| 801 |
+
"token_acc": 0.9557479199858382
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 1.1832301007673107,
|
| 805 |
+
"grad_norm": 1.8040844202041626,
|
| 806 |
+
"learning_rate": 8.16987211391638e-06,
|
| 807 |
+
"loss": 0.12417132377624512,
|
| 808 |
+
"step": 6400,
|
| 809 |
+
"token_acc": 0.956177815065432
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 1.1832301007673107,
|
| 813 |
+
"eval_loss": 1.7814701795578003,
|
| 814 |
+
"eval_runtime": 33.8097,
|
| 815 |
+
"eval_samples_per_second": 457.916,
|
| 816 |
+
"eval_steps_per_second": 0.917,
|
| 817 |
+
"eval_token_acc": 0.8401797359565951,
|
| 818 |
+
"step": 6400
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 1.201719515577332,
|
| 822 |
+
"grad_norm": 1.9418505430221558,
|
| 823 |
+
"learning_rate": 8.112233353053489e-06,
|
| 824 |
+
"loss": 0.1239329719543457,
|
| 825 |
+
"step": 6500,
|
| 826 |
+
"token_acc": 0.9561117973681551
|
| 827 |
+
},
|
| 828 |
+
{
|
| 829 |
+
"epoch": 1.2202089303873533,
|
| 830 |
+
"grad_norm": 1.9047932624816895,
|
| 831 |
+
"learning_rate": 8.053911358417312e-06,
|
| 832 |
+
"loss": 0.12369691848754882,
|
| 833 |
+
"step": 6600,
|
| 834 |
+
"token_acc": 0.9562336339716577
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 1.2202089303873533,
|
| 838 |
+
"eval_loss": 1.7961959838867188,
|
| 839 |
+
"eval_runtime": 39.5814,
|
| 840 |
+
"eval_samples_per_second": 391.144,
|
| 841 |
+
"eval_steps_per_second": 0.783,
|
| 842 |
+
"eval_token_acc": 0.840048899955434,
|
| 843 |
+
"step": 6600
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 1.2386983451973745,
|
| 847 |
+
"grad_norm": 1.8649250268936157,
|
| 848 |
+
"learning_rate": 7.9949189335325e-06,
|
| 849 |
+
"loss": 0.1238163185119629,
|
| 850 |
+
"step": 6700,
|
| 851 |
+
"token_acc": 0.9562550435641857
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 1.2571877600073957,
|
| 855 |
+
"grad_norm": 2.16851806640625,
|
| 856 |
+
"learning_rate": 7.935269029104378e-06,
|
| 857 |
+
"loss": 0.12391100883483887,
|
| 858 |
+
"step": 6800,
|
| 859 |
+
"token_acc": 0.9561337716519507
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 1.2571877600073957,
|
| 863 |
+
"eval_loss": 1.796657919883728,
|
| 864 |
+
"eval_runtime": 38.4616,
|
| 865 |
+
"eval_samples_per_second": 402.531,
|
| 866 |
+
"eval_steps_per_second": 0.806,
|
| 867 |
+
"eval_token_acc": 0.8407112572113125,
|
| 868 |
+
"step": 6800
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 1.275677174817417,
|
| 872 |
+
"grad_norm": 1.8265526294708252,
|
| 873 |
+
"learning_rate": 7.874974740175845e-06,
|
| 874 |
+
"loss": 0.12333698272705078,
|
| 875 |
+
"step": 6900,
|
| 876 |
+
"token_acc": 0.9563835803661899
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 1.2941665896274384,
|
| 880 |
+
"grad_norm": 1.7967718839645386,
|
| 881 |
+
"learning_rate": 7.814049303252609e-06,
|
| 882 |
+
"loss": 0.12335062026977539,
|
| 883 |
+
"step": 7000,
|
| 884 |
+
"token_acc": 0.956260351207485
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 1.2941665896274384,
|
| 888 |
+
"eval_loss": 1.7862260341644287,
|
| 889 |
+
"eval_runtime": 33.965,
|
| 890 |
+
"eval_samples_per_second": 455.823,
|
| 891 |
+
"eval_steps_per_second": 0.913,
|
| 892 |
+
"eval_token_acc": 0.8402247108319943,
|
| 893 |
+
"step": 7000
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 1.3126560044374596,
|
| 897 |
+
"grad_norm": 1.8670226335525513,
|
| 898 |
+
"learning_rate": 7.752506093397341e-06,
|
| 899 |
+
"loss": 0.12360490798950195,
|
| 900 |
+
"step": 7100,
|
| 901 |
+
"token_acc": 0.9562331933764212
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 1.3311454192474808,
|
| 905 |
+
"grad_norm": 2.2251029014587402,
|
| 906 |
+
"learning_rate": 7.690358621293443e-06,
|
| 907 |
+
"loss": 0.12216304779052735,
|
| 908 |
+
"step": 7200,
|
| 909 |
+
"token_acc": 0.9565792510264621
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 1.3311454192474808,
|
| 913 |
+
"eval_loss": 1.7927396297454834,
|
| 914 |
+
"eval_runtime": 34.2328,
|
| 915 |
+
"eval_samples_per_second": 452.257,
|
| 916 |
+
"eval_steps_per_second": 0.906,
|
| 917 |
+
"eval_token_acc": 0.8404986487094255,
|
| 918 |
+
"step": 7200
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 1.3496348340575022,
|
| 922 |
+
"grad_norm": 1.8033658266067505,
|
| 923 |
+
"learning_rate": 7.627620530279005e-06,
|
| 924 |
+
"loss": 0.12235649108886719,
|
| 925 |
+
"step": 7300,
|
| 926 |
+
"token_acc": 0.956481924374402
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 1.3681242488675234,
|
| 930 |
+
"grad_norm": 1.9245818853378296,
|
| 931 |
+
"learning_rate": 7.564305593351679e-06,
|
| 932 |
+
"loss": 0.12291913986206054,
|
| 933 |
+
"step": 7400,
|
| 934 |
+
"token_acc": 0.9564290328431482
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 1.3681242488675234,
|
| 938 |
+
"eval_loss": 1.8202570676803589,
|
| 939 |
+
"eval_runtime": 34.8385,
|
| 940 |
+
"eval_samples_per_second": 444.393,
|
| 941 |
+
"eval_steps_per_second": 0.89,
|
| 942 |
+
"eval_token_acc": 0.8406621937108769,
|
| 943 |
+
"step": 7400
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 1.3866136636775446,
|
| 947 |
+
"grad_norm": 1.7945640087127686,
|
| 948 |
+
"learning_rate": 7.500427710145055e-06,
|
| 949 |
+
"loss": 0.12242051124572755,
|
| 950 |
+
"step": 7500,
|
| 951 |
+
"token_acc": 0.9565831394835252
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 1.4051030784875658,
|
| 955 |
+
"grad_norm": 1.8062553405761719,
|
| 956 |
+
"learning_rate": 7.436000903877263e-06,
|
| 957 |
+
"loss": 0.12153498649597168,
|
| 958 |
+
"step": 7600,
|
| 959 |
+
"token_acc": 0.9567279089013572
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 1.4051030784875658,
|
| 963 |
+
"eval_loss": 1.8174859285354614,
|
| 964 |
+
"eval_runtime": 35.6393,
|
| 965 |
+
"eval_samples_per_second": 434.408,
|
| 966 |
+
"eval_steps_per_second": 0.87,
|
| 967 |
+
"eval_token_acc": 0.8409811064637073,
|
| 968 |
+
"step": 7600
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 1.4235924932975872,
|
| 972 |
+
"grad_norm": 1.825256586074829,
|
| 973 |
+
"learning_rate": 7.37103931827242e-06,
|
| 974 |
+
"loss": 0.12253303527832031,
|
| 975 |
+
"step": 7700,
|
| 976 |
+
"token_acc": 0.9565785211260692
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 1.4420819081076084,
|
| 980 |
+
"grad_norm": 1.861182689666748,
|
| 981 |
+
"learning_rate": 7.305557214455652e-06,
|
| 982 |
+
"loss": 0.1214249610900879,
|
| 983 |
+
"step": 7800,
|
| 984 |
+
"token_acc": 0.9568180053778998
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 1.4420819081076084,
|
| 988 |
+
"eval_loss": 1.8193385601043701,
|
| 989 |
+
"eval_runtime": 34.8888,
|
| 990 |
+
"eval_samples_per_second": 443.753,
|
| 991 |
+
"eval_steps_per_second": 0.889,
|
| 992 |
+
"eval_token_acc": 0.8408257387123285,
|
| 993 |
+
"step": 7800
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 1.4605713229176296,
|
| 997 |
+
"grad_norm": 1.758222222328186,
|
| 998 |
+
"learning_rate": 7.239568967822305e-06,
|
| 999 |
+
"loss": 0.12132749557495118,
|
| 1000 |
+
"step": 7900,
|
| 1001 |
+
"token_acc": 0.9569372409602568
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 1.479060737727651,
|
| 1005 |
+
"grad_norm": 1.9188320636749268,
|
| 1006 |
+
"learning_rate": 7.173089064882109e-06,
|
| 1007 |
+
"loss": 0.12142126083374023,
|
| 1008 |
+
"step": 8000,
|
| 1009 |
+
"token_acc": 0.9569549769253634
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 1.479060737727651,
|
| 1013 |
+
"eval_loss": 1.8274463415145874,
|
| 1014 |
+
"eval_runtime": 34.5324,
|
| 1015 |
+
"eval_samples_per_second": 448.333,
|
| 1016 |
+
"eval_steps_per_second": 0.898,
|
| 1017 |
+
"eval_token_acc": 0.8413327282168279,
|
| 1018 |
+
"step": 8000
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 1.4975501525376722,
|
| 1022 |
+
"grad_norm": 1.9004936218261719,
|
| 1023 |
+
"learning_rate": 7.106132100078914e-06,
|
| 1024 |
+
"loss": 0.12066761016845703,
|
| 1025 |
+
"step": 8100,
|
| 1026 |
+
"token_acc": 0.9572519065995776
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 1.5160395673476934,
|
| 1030 |
+
"grad_norm": 1.7836532592773438,
|
| 1031 |
+
"learning_rate": 7.0387127725867645e-06,
|
| 1032 |
+
"loss": 0.12059429168701172,
|
| 1033 |
+
"step": 8200,
|
| 1034 |
+
"token_acc": 0.9571157408717915
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 1.5160395673476934,
|
| 1038 |
+
"eval_loss": 1.8109354972839355,
|
| 1039 |
+
"eval_runtime": 34.7568,
|
| 1040 |
+
"eval_samples_per_second": 445.437,
|
| 1041 |
+
"eval_steps_per_second": 0.892,
|
| 1042 |
+
"eval_token_acc": 0.8408093842121833,
|
| 1043 |
+
"step": 8200
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 1.5345289821577146,
|
| 1047 |
+
"grad_norm": 1.7904539108276367,
|
| 1048 |
+
"learning_rate": 6.970845883082951e-06,
|
| 1049 |
+
"loss": 0.12016033172607422,
|
| 1050 |
+
"step": 8300,
|
| 1051 |
+
"token_acc": 0.9573499747856783
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 1.553018396967736,
|
| 1055 |
+
"grad_norm": 1.8207581043243408,
|
| 1056 |
+
"learning_rate": 6.902546330498798e-06,
|
| 1057 |
+
"loss": 0.12048395156860352,
|
| 1058 |
+
"step": 8400,
|
| 1059 |
+
"token_acc": 0.9572356815704754
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 1.553018396967736,
|
| 1063 |
+
"eval_loss": 1.833103895187378,
|
| 1064 |
+
"eval_runtime": 35.4186,
|
| 1065 |
+
"eval_samples_per_second": 437.115,
|
| 1066 |
+
"eval_steps_per_second": 0.875,
|
| 1067 |
+
"eval_token_acc": 0.8409402202133445,
|
| 1068 |
+
"step": 8400
|
| 1069 |
+
},
|
| 1070 |
+
{
|
| 1071 |
+
"epoch": 1.5715078117777572,
|
| 1072 |
+
"grad_norm": 1.8081518411636353,
|
| 1073 |
+
"learning_rate": 6.8338291087488794e-06,
|
| 1074 |
+
"loss": 0.11957192420959473,
|
| 1075 |
+
"step": 8500,
|
| 1076 |
+
"token_acc": 0.9574644121699327
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 1.5899972265877786,
|
| 1080 |
+
"grad_norm": 1.9988985061645508,
|
| 1081 |
+
"learning_rate": 6.764709303439375e-06,
|
| 1082 |
+
"loss": 0.11883423805236816,
|
| 1083 |
+
"step": 8600,
|
| 1084 |
+
"token_acc": 0.9577689760862731
|
| 1085 |
+
},
|
| 1086 |
+
{
|
| 1087 |
+
"epoch": 1.5899972265877786,
|
| 1088 |
+
"eval_loss": 1.8361092805862427,
|
| 1089 |
+
"eval_runtime": 34.9305,
|
| 1090 |
+
"eval_samples_per_second": 443.223,
|
| 1091 |
+
"eval_steps_per_second": 0.887,
|
| 1092 |
+
"eval_token_acc": 0.8417456793454929,
|
| 1093 |
+
"step": 8600
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"epoch": 1.6084866413977998,
|
| 1097 |
+
"grad_norm": 1.8924856185913086,
|
| 1098 |
+
"learning_rate": 6.695202088556305e-06,
|
| 1099 |
+
"loss": 0.11925403594970703,
|
| 1100 |
+
"step": 8700,
|
| 1101 |
+
"token_acc": 0.9576467298626727
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 1.626976056207821,
|
| 1105 |
+
"grad_norm": 1.8594777584075928,
|
| 1106 |
+
"learning_rate": 6.62532272313436e-06,
|
| 1107 |
+
"loss": 0.11869814872741699,
|
| 1108 |
+
"step": 8800,
|
| 1109 |
+
"token_acc": 0.9578026280546409
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 1.626976056207821,
|
| 1113 |
+
"eval_loss": 1.8373812437057495,
|
| 1114 |
+
"eval_runtime": 36.5534,
|
| 1115 |
+
"eval_samples_per_second": 423.545,
|
| 1116 |
+
"eval_steps_per_second": 0.848,
|
| 1117 |
+
"eval_token_acc": 0.8413286395917917,
|
| 1118 |
+
"step": 8800
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 1.6454654710178422,
|
| 1122 |
+
"grad_norm": 1.944880723953247,
|
| 1123 |
+
"learning_rate": 6.555086547907054e-06,
|
| 1124 |
+
"loss": 0.11804214477539063,
|
| 1125 |
+
"step": 8900,
|
| 1126 |
+
"token_acc": 0.9581940918690292
|
| 1127 |
+
},
|
| 1128 |
+
{
|
| 1129 |
+
"epoch": 1.6639548858278634,
|
| 1130 |
+
"grad_norm": 1.7270747423171997,
|
| 1131 |
+
"learning_rate": 6.4845089819389595e-06,
|
| 1132 |
+
"loss": 0.11829812049865723,
|
| 1133 |
+
"step": 9000,
|
| 1134 |
+
"token_acc": 0.9581205644511996
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 1.6639548858278634,
|
| 1138 |
+
"eval_loss": 1.8515816926956177,
|
| 1139 |
+
"eval_runtime": 33.8737,
|
| 1140 |
+
"eval_samples_per_second": 457.051,
|
| 1141 |
+
"eval_steps_per_second": 0.915,
|
| 1142 |
+
"eval_token_acc": 0.8415167163434608,
|
| 1143 |
+
"step": 9000
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 1.6824443006378849,
|
| 1147 |
+
"grad_norm": 1.7423977851867676,
|
| 1148 |
+
"learning_rate": 6.4136055192407176e-06,
|
| 1149 |
+
"loss": 0.11753586769104003,
|
| 1150 |
+
"step": 9100,
|
| 1151 |
+
"token_acc": 0.958324552925161
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 1.700933715447906,
|
| 1155 |
+
"grad_norm": 2.1716482639312744,
|
| 1156 |
+
"learning_rate": 6.34239172536763e-06,
|
| 1157 |
+
"loss": 0.11699953079223632,
|
| 1158 |
+
"step": 9200,
|
| 1159 |
+
"token_acc": 0.9585432410566509
|
| 1160 |
+
},
|
| 1161 |
+
{
|
| 1162 |
+
"epoch": 1.700933715447906,
|
| 1163 |
+
"eval_loss": 1.834566354751587,
|
| 1164 |
+
"eval_runtime": 36.5294,
|
| 1165 |
+
"eval_samples_per_second": 423.823,
|
| 1166 |
+
"eval_steps_per_second": 0.849,
|
| 1167 |
+
"eval_token_acc": 0.8417783883457832,
|
| 1168 |
+
"step": 9200
|
| 1169 |
+
},
|
| 1170 |
+
{
|
| 1171 |
+
"epoch": 1.7194231302579275,
|
| 1172 |
+
"grad_norm": 1.8716721534729004,
|
| 1173 |
+
"learning_rate": 6.270883234002516e-06,
|
| 1174 |
+
"loss": 0.11731613159179688,
|
| 1175 |
+
"step": 9300,
|
| 1176 |
+
"token_acc": 0.9585327793784123
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 1.7379125450679487,
|
| 1180 |
+
"grad_norm": 1.8995591402053833,
|
| 1181 |
+
"learning_rate": 6.199095743523621e-06,
|
| 1182 |
+
"loss": 0.1166218090057373,
|
| 1183 |
+
"step": 9400,
|
| 1184 |
+
"token_acc": 0.9586903179633984
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 1.7379125450679487,
|
| 1188 |
+
"eval_loss": 1.8363566398620605,
|
| 1189 |
+
"eval_runtime": 42.1601,
|
| 1190 |
+
"eval_samples_per_second": 367.219,
|
| 1191 |
+
"eval_steps_per_second": 0.735,
|
| 1192 |
+
"eval_token_acc": 0.8415944002191503,
|
| 1193 |
+
"step": 9400
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 1.7564019598779699,
|
| 1197 |
+
"grad_norm": 1.8480995893478394,
|
| 1198 |
+
"learning_rate": 6.127045013558339e-06,
|
| 1199 |
+
"loss": 0.1168486213684082,
|
| 1200 |
+
"step": 9500,
|
| 1201 |
+
"token_acc": 0.9585827515599433
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 1.774891374687991,
|
| 1205 |
+
"grad_norm": 1.8877906799316406,
|
| 1206 |
+
"learning_rate": 6.0547468615234616e-06,
|
| 1207 |
+
"loss": 0.11609399795532227,
|
| 1208 |
+
"step": 9600,
|
| 1209 |
+
"token_acc": 0.9589387053013588
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 1.774891374687991,
|
| 1213 |
+
"eval_loss": 1.8403908014297485,
|
| 1214 |
+
"eval_runtime": 33.7405,
|
| 1215 |
+
"eval_samples_per_second": 458.855,
|
| 1216 |
+
"eval_steps_per_second": 0.919,
|
| 1217 |
+
"eval_token_acc": 0.8420359717230692,
|
| 1218 |
+
"step": 9600
|
| 1219 |
+
},
|
| 1220 |
+
{
|
| 1221 |
+
"epoch": 1.7933807894980123,
|
| 1222 |
+
"grad_norm": 1.9642070531845093,
|
| 1223 |
+
"learning_rate": 5.982217159152759e-06,
|
| 1224 |
+
"loss": 0.11574806213378906,
|
| 1225 |
+
"step": 9700,
|
| 1226 |
+
"token_acc": 0.958883506092705
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 1.8118702043080337,
|
| 1230 |
+
"grad_norm": 1.711111068725586,
|
| 1231 |
+
"learning_rate": 5.909471829012639e-06,
|
| 1232 |
+
"loss": 0.11505707740783691,
|
| 1233 |
+
"step": 9800,
|
| 1234 |
+
"token_acc": 0.9590959173154312
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 1.8118702043080337,
|
| 1238 |
+
"eval_loss": 1.8579939603805542,
|
| 1239 |
+
"eval_runtime": 38.0908,
|
| 1240 |
+
"eval_samples_per_second": 406.45,
|
| 1241 |
+
"eval_steps_per_second": 0.814,
|
| 1242 |
+
"eval_token_acc": 0.8419501105973072,
|
| 1243 |
+
"step": 9800
|
| 1244 |
+
},
|
| 1245 |
+
{
|
| 1246 |
+
"epoch": 1.830359619118055,
|
| 1247 |
+
"grad_norm": 1.813797116279602,
|
| 1248 |
+
"learning_rate": 5.836526841006622e-06,
|
| 1249 |
+
"loss": 0.11528162956237793,
|
| 1250 |
+
"step": 9900,
|
| 1251 |
+
"token_acc": 0.959217136559903
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 1.8488490339280763,
|
| 1255 |
+
"grad_norm": 1.7768510580062866,
|
| 1256 |
+
"learning_rate": 5.763398208869463e-06,
|
| 1257 |
+
"loss": 0.11398514747619629,
|
| 1258 |
+
"step": 10000,
|
| 1259 |
+
"token_acc": 0.9596556526939511
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 1.8488490339280763,
|
| 1263 |
+
"eval_loss": 1.843808889389038,
|
| 1264 |
+
"eval_runtime": 33.4263,
|
| 1265 |
+
"eval_samples_per_second": 463.169,
|
| 1266 |
+
"eval_steps_per_second": 0.927,
|
| 1267 |
+
"eval_token_acc": 0.8417415907204566,
|
| 1268 |
+
"step": 10000
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 1.8673384487380975,
|
| 1272 |
+
"grad_norm": 1.754035234451294,
|
| 1273 |
+
"learning_rate": 5.690101986651614e-06,
|
| 1274 |
+
"loss": 0.11392656326293946,
|
| 1275 |
+
"step": 10100,
|
| 1276 |
+
"token_acc": 0.9596795537848105
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 1.8858278635481187,
|
| 1280 |
+
"grad_norm": 1.8597276210784912,
|
| 1281 |
+
"learning_rate": 5.616654265194856e-06,
|
| 1282 |
+
"loss": 0.11345600128173829,
|
| 1283 |
+
"step": 10200,
|
| 1284 |
+
"token_acc": 0.9599432397585764
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 1.8858278635481187,
|
| 1288 |
+
"eval_loss": 1.8413722515106201,
|
| 1289 |
+
"eval_runtime": 33.5094,
|
| 1290 |
+
"eval_samples_per_second": 462.019,
|
| 1291 |
+
"eval_steps_per_second": 0.925,
|
| 1292 |
+
"eval_token_acc": 0.841933756097162,
|
| 1293 |
+
"step": 10200
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 1.90431727835814,
|
| 1297 |
+
"grad_norm": 1.9222540855407715,
|
| 1298 |
+
"learning_rate": 5.543071168599853e-06,
|
| 1299 |
+
"loss": 0.11406026840209961,
|
| 1300 |
+
"step": 10300,
|
| 1301 |
+
"token_acc": 0.9597226949620891
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 1.922806693168161,
|
| 1305 |
+
"grad_norm": 1.87301504611969,
|
| 1306 |
+
"learning_rate": 5.4693688506863965e-06,
|
| 1307 |
+
"loss": 0.11312946319580078,
|
| 1308 |
+
"step": 10400,
|
| 1309 |
+
"token_acc": 0.9600168523848207
|
| 1310 |
+
},
|
| 1311 |
+
{
|
| 1312 |
+
"epoch": 1.922806693168161,
|
| 1313 |
+
"eval_loss": 1.8454869985580444,
|
| 1314 |
+
"eval_runtime": 35.3729,
|
| 1315 |
+
"eval_samples_per_second": 437.68,
|
| 1316 |
+
"eval_steps_per_second": 0.876,
|
| 1317 |
+
"eval_token_acc": 0.8420237058479604,
|
| 1318 |
+
"step": 10400
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 1.9412961079781825,
|
| 1322 |
+
"grad_norm": 1.7703731060028076,
|
| 1323 |
+
"learning_rate": 5.395563491447134e-06,
|
| 1324 |
+
"loss": 0.11194120407104492,
|
| 1325 |
+
"step": 10500,
|
| 1326 |
+
"token_acc": 0.9604128003703771
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 1.9597855227882037,
|
| 1330 |
+
"grad_norm": 1.8830459117889404,
|
| 1331 |
+
"learning_rate": 5.321671293495544e-06,
|
| 1332 |
+
"loss": 0.11190027236938477,
|
| 1333 |
+
"step": 10600,
|
| 1334 |
+
"token_acc": 0.9605756553095823
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 1.9597855227882037,
|
| 1338 |
+
"eval_loss": 1.8413710594177246,
|
| 1339 |
+
"eval_runtime": 34.0357,
|
| 1340 |
+
"eval_samples_per_second": 454.876,
|
| 1341 |
+
"eval_steps_per_second": 0.911,
|
| 1342 |
+
"eval_token_acc": 0.8419950854727064,
|
| 1343 |
+
"step": 10600
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 1.9782749375982251,
|
| 1347 |
+
"grad_norm": 1.8603187799453735,
|
| 1348 |
+
"learning_rate": 5.247708478508959e-06,
|
| 1349 |
+
"loss": 0.11120811462402344,
|
| 1350 |
+
"step": 10700,
|
| 1351 |
+
"token_acc": 0.9608388566054109
|
| 1352 |
+
},
|
| 1353 |
+
{
|
| 1354 |
+
"epoch": 1.9967643524082463,
|
| 1355 |
+
"grad_norm": 2.23905348777771,
|
| 1356 |
+
"learning_rate": 5.173691283667378e-06,
|
| 1357 |
+
"loss": 0.11172771453857422,
|
| 1358 |
+
"step": 10800,
|
| 1359 |
+
"token_acc": 0.9605912691939935
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 1.9967643524082463,
|
| 1363 |
+
"eval_loss": 1.8494235277175903,
|
| 1364 |
+
"eval_runtime": 33.3601,
|
| 1365 |
+
"eval_samples_per_second": 464.087,
|
| 1366 |
+
"eval_steps_per_second": 0.929,
|
| 1367 |
+
"eval_token_acc": 0.8421422759740127,
|
| 1368 |
+
"step": 10800
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 2.0151613201442173,
|
| 1372 |
+
"grad_norm": 1.4144299030303955,
|
| 1373 |
+
"learning_rate": 5.099635958088908e-06,
|
| 1374 |
+
"loss": 0.07613845825195313,
|
| 1375 |
+
"step": 10900,
|
| 1376 |
+
"token_acc": 0.9769882988398573
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 2.0336507349542385,
|
| 1380 |
+
"grad_norm": 1.3822933435440063,
|
| 1381 |
+
"learning_rate": 5.025558759262555e-06,
|
| 1382 |
+
"loss": 0.068076810836792,
|
| 1383 |
+
"step": 11000,
|
| 1384 |
+
"token_acc": 0.9801309933010414
|
| 1385 |
+
},
|
| 1386 |
+
{
|
| 1387 |
+
"epoch": 2.0336507349542385,
|
| 1388 |
+
"eval_loss": 1.938856840133667,
|
| 1389 |
+
"eval_runtime": 34.0978,
|
| 1390 |
+
"eval_samples_per_second": 454.047,
|
| 1391 |
+
"eval_steps_per_second": 0.909,
|
| 1392 |
+
"eval_token_acc": 0.8414921845932432,
|
| 1393 |
+
"step": 11000
|
| 1394 |
+
},
|
| 1395 |
+
{
|
| 1396 |
+
"epoch": 2.05214014976426,
|
| 1397 |
+
"grad_norm": 1.5795366764068604,
|
| 1398 |
+
"learning_rate": 4.951475949479197e-06,
|
| 1399 |
+
"loss": 0.06814120292663574,
|
| 1400 |
+
"step": 11100,
|
| 1401 |
+
"token_acc": 0.9797405445487929
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 2.0706295645742814,
|
| 1405 |
+
"grad_norm": 1.6069875955581665,
|
| 1406 |
+
"learning_rate": 4.877403792261494e-06,
|
| 1407 |
+
"loss": 0.06798234939575196,
|
| 1408 |
+
"step": 11200,
|
| 1409 |
+
"token_acc": 0.979876903070811
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 2.0706295645742814,
|
| 1413 |
+
"eval_loss": 1.965312123298645,
|
| 1414 |
+
"eval_runtime": 34.7072,
|
| 1415 |
+
"eval_samples_per_second": 446.075,
|
| 1416 |
+
"eval_steps_per_second": 0.893,
|
| 1417 |
+
"eval_token_acc": 0.841819274596146,
|
| 1418 |
+
"step": 11200
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 2.0891189793843026,
|
| 1422 |
+
"grad_norm": 1.7105296850204468,
|
| 1423 |
+
"learning_rate": 4.8033585487935365e-06,
|
| 1424 |
+
"loss": 0.06711929321289062,
|
| 1425 |
+
"step": 11300,
|
| 1426 |
+
"token_acc": 0.9799202953211885
|
| 1427 |
+
},
|
| 1428 |
+
{
|
| 1429 |
+
"epoch": 2.1076083941943238,
|
| 1430 |
+
"grad_norm": 1.6663514375686646,
|
| 1431 |
+
"learning_rate": 4.729356474350985e-06,
|
| 1432 |
+
"loss": 0.06744234561920166,
|
| 1433 |
+
"step": 11400,
|
| 1434 |
+
"token_acc": 0.9799818812847331
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 2.1076083941943238,
|
| 1438 |
+
"eval_loss": 1.987338662147522,
|
| 1439 |
+
"eval_runtime": 34.6728,
|
| 1440 |
+
"eval_samples_per_second": 446.517,
|
| 1441 |
+
"eval_steps_per_second": 0.894,
|
| 1442 |
+
"eval_token_acc": 0.8417988314709647,
|
| 1443 |
+
"step": 11400
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 2.126097809004345,
|
| 1447 |
+
"grad_norm": 1.6091212034225464,
|
| 1448 |
+
"learning_rate": 4.65541381473254e-06,
|
| 1449 |
+
"loss": 0.06767751693725586,
|
| 1450 |
+
"step": 11500,
|
| 1451 |
+
"token_acc": 0.9798217134248365
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 2.144587223814366,
|
| 1455 |
+
"grad_norm": 1.4260798692703247,
|
| 1456 |
+
"learning_rate": 4.58154680269346e-06,
|
| 1457 |
+
"loss": 0.06699298858642579,
|
| 1458 |
+
"step": 11600,
|
| 1459 |
+
"token_acc": 0.9798626501624399
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 2.144587223814366,
|
| 1463 |
+
"eval_loss": 2.00212025642395,
|
| 1464 |
+
"eval_runtime": 35.0411,
|
| 1465 |
+
"eval_samples_per_second": 441.824,
|
| 1466 |
+
"eval_steps_per_second": 0.885,
|
| 1467 |
+
"eval_token_acc": 0.8419950854727064,
|
| 1468 |
+
"step": 11600
|
| 1469 |
+
},
|
| 1470 |
+
{
|
| 1471 |
+
"epoch": 2.1630766386243874,
|
| 1472 |
+
"grad_norm": 1.4290509223937988,
|
| 1473 |
+
"learning_rate": 4.5077716543819705e-06,
|
| 1474 |
+
"loss": 0.06717270851135254,
|
| 1475 |
+
"step": 11700,
|
| 1476 |
+
"token_acc": 0.9798529561570528
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 2.181566053434409,
|
| 1480 |
+
"grad_norm": 1.4190586805343628,
|
| 1481 |
+
"learning_rate": 4.434104565779295e-06,
|
| 1482 |
+
"loss": 0.06727582931518555,
|
| 1483 |
+
"step": 11800,
|
| 1484 |
+
"token_acc": 0.979750684884302
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 2.181566053434409,
|
| 1488 |
+
"eval_loss": 2.0236058235168457,
|
| 1489 |
+
"eval_runtime": 39.6338,
|
| 1490 |
+
"eval_samples_per_second": 390.626,
|
| 1491 |
+
"eval_steps_per_second": 0.782,
|
| 1492 |
+
"eval_token_acc": 0.8418274518462187,
|
| 1493 |
+
"step": 11800
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 2.20005546824443,
|
| 1497 |
+
"grad_norm": 1.3393402099609375,
|
| 1498 |
+
"learning_rate": 4.360561709144133e-06,
|
| 1499 |
+
"loss": 0.06670400619506836,
|
| 1500 |
+
"step": 11900,
|
| 1501 |
+
"token_acc": 0.9797775570535143
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 2.2185448830544514,
|
| 1505 |
+
"grad_norm": 1.5458145141601562,
|
| 1506 |
+
"learning_rate": 4.287159229462329e-06,
|
| 1507 |
+
"loss": 0.06665773868560791,
|
| 1508 |
+
"step": 12000,
|
| 1509 |
+
"token_acc": 0.9798853092505024
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 2.2185448830544514,
|
| 1513 |
+
"eval_loss": 2.010664463043213,
|
| 1514 |
+
"eval_runtime": 34.9951,
|
| 1515 |
+
"eval_samples_per_second": 442.405,
|
| 1516 |
+
"eval_steps_per_second": 0.886,
|
| 1517 |
+
"eval_token_acc": 0.8416271092194406,
|
| 1518 |
+
"step": 12000
|
| 1519 |
+
},
|
| 1520 |
+
{
|
| 1521 |
+
"epoch": 2.2370342978644726,
|
| 1522 |
+
"grad_norm": 1.530498743057251,
|
| 1523 |
+
"learning_rate": 4.213913240902546e-06,
|
| 1524 |
+
"loss": 0.06623137950897216,
|
| 1525 |
+
"step": 12100,
|
| 1526 |
+
"token_acc": 0.9800267679650442
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 2.255523712674494,
|
| 1530 |
+
"grad_norm": 1.511843204498291,
|
| 1531 |
+
"learning_rate": 4.140839823278687e-06,
|
| 1532 |
+
"loss": 0.06620734691619873,
|
| 1533 |
+
"step": 12200,
|
| 1534 |
+
"token_acc": 0.9800285430125266
|
| 1535 |
+
},
|
| 1536 |
+
{
|
| 1537 |
+
"epoch": 2.255523712674494,
|
| 1538 |
+
"eval_loss": 2.0170493125915527,
|
| 1539 |
+
"eval_runtime": 34.7409,
|
| 1540 |
+
"eval_samples_per_second": 445.642,
|
| 1541 |
+
"eval_steps_per_second": 0.892,
|
| 1542 |
+
"eval_token_acc": 0.8416271092194406,
|
| 1543 |
+
"step": 12200
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 2.274013127484515,
|
| 1547 |
+
"grad_norm": 1.7180910110473633,
|
| 1548 |
+
"learning_rate": 4.067955018519869e-06,
|
| 1549 |
+
"loss": 0.06616350173950196,
|
| 1550 |
+
"step": 12300,
|
| 1551 |
+
"token_acc": 0.9798321926652386
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 2.292502542294536,
|
| 1555 |
+
"grad_norm": 1.4852060079574585,
|
| 1556 |
+
"learning_rate": 3.9952748271487135e-06,
|
| 1557 |
+
"loss": 0.06596662521362305,
|
| 1558 |
+
"step": 12400,
|
| 1559 |
+
"token_acc": 0.9801906228334967
|
| 1560 |
+
},
|
| 1561 |
+
{
|
| 1562 |
+
"epoch": 2.292502542294536,
|
| 1563 |
+
"eval_loss": 2.046187400817871,
|
| 1564 |
+
"eval_runtime": 34.4149,
|
| 1565 |
+
"eval_samples_per_second": 449.863,
|
| 1566 |
+
"eval_steps_per_second": 0.901,
|
| 1567 |
+
"eval_token_acc": 0.8417456793454929,
|
| 1568 |
+
"step": 12400
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 2.310991957104558,
|
| 1572 |
+
"grad_norm": 1.4535032510757446,
|
| 1573 |
+
"learning_rate": 3.922815204768729e-06,
|
| 1574 |
+
"loss": 0.06627186775207519,
|
| 1575 |
+
"step": 12500,
|
| 1576 |
+
"token_acc": 0.9801007046174975
|
| 1577 |
+
},
|
| 1578 |
+
{
|
| 1579 |
+
"epoch": 2.329481371914579,
|
| 1580 |
+
"grad_norm": 1.6788718700408936,
|
| 1581 |
+
"learning_rate": 3.850592058561538e-06,
|
| 1582 |
+
"loss": 0.06580290794372559,
|
| 1583 |
+
"step": 12600,
|
| 1584 |
+
"token_acc": 0.9801861501544031
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 2.329481371914579,
|
| 1588 |
+
"eval_loss": 2.052591562271118,
|
| 1589 |
+
"eval_runtime": 35.4887,
|
| 1590 |
+
"eval_samples_per_second": 436.251,
|
| 1591 |
+
"eval_steps_per_second": 0.874,
|
| 1592 |
+
"eval_token_acc": 0.8420441489731418,
|
| 1593 |
+
"step": 12600
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 2.3479707867246002,
|
| 1597 |
+
"grad_norm": 1.6238676309585571,
|
| 1598 |
+
"learning_rate": 3.778621243794759e-06,
|
| 1599 |
+
"loss": 0.06539412021636963,
|
| 1600 |
+
"step": 12700,
|
| 1601 |
+
"token_acc": 0.9803736451136588
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"epoch": 2.3664602015346214,
|
| 1605 |
+
"grad_norm": 1.5273261070251465,
|
| 1606 |
+
"learning_rate": 3.706918560341266e-06,
|
| 1607 |
+
"loss": 0.06567787170410157,
|
| 1608 |
+
"step": 12800,
|
| 1609 |
+
"token_acc": 0.9801923854086196
|
| 1610 |
+
},
|
| 1611 |
+
{
|
| 1612 |
+
"epoch": 2.3664602015346214,
|
| 1613 |
+
"eval_loss": 2.0543408393859863,
|
| 1614 |
+
"eval_runtime": 33.992,
|
| 1615 |
+
"eval_samples_per_second": 455.461,
|
| 1616 |
+
"eval_steps_per_second": 0.912,
|
| 1617 |
+
"eval_token_acc": 0.8416884385949849,
|
| 1618 |
+
"step": 12800
|
| 1619 |
+
},
|
| 1620 |
+
{
|
| 1621 |
+
"epoch": 2.3849496163446426,
|
| 1622 |
+
"grad_norm": 1.6116151809692383,
|
| 1623 |
+
"learning_rate": 3.6354997492106258e-06,
|
| 1624 |
+
"loss": 0.06504707336425782,
|
| 1625 |
+
"step": 12900,
|
| 1626 |
+
"token_acc": 0.9804272111407846
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 2.403439031154664,
|
| 1630 |
+
"grad_norm": 1.4699835777282715,
|
| 1631 |
+
"learning_rate": 3.5643804890934353e-06,
|
| 1632 |
+
"loss": 0.06518766403198242,
|
| 1633 |
+
"step": 13000,
|
| 1634 |
+
"token_acc": 0.9803263563826762
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 2.403439031154664,
|
| 1638 |
+
"eval_loss": 2.074248790740967,
|
| 1639 |
+
"eval_runtime": 34.761,
|
| 1640 |
+
"eval_samples_per_second": 445.384,
|
| 1641 |
+
"eval_steps_per_second": 0.892,
|
| 1642 |
+
"eval_token_acc": 0.8419787309725613,
|
| 1643 |
+
"step": 13000
|
| 1644 |
+
},
|
| 1645 |
+
{
|
| 1646 |
+
"epoch": 2.421928445964685,
|
| 1647 |
+
"grad_norm": 1.400291085243225,
|
| 1648 |
+
"learning_rate": 3.4935763929193567e-06,
|
| 1649 |
+
"loss": 0.06465916633605957,
|
| 1650 |
+
"step": 13100,
|
| 1651 |
+
"token_acc": 0.9806667378597698
|
| 1652 |
+
},
|
| 1653 |
+
{
|
| 1654 |
+
"epoch": 2.4404178607747067,
|
| 1655 |
+
"grad_norm": 1.3986449241638184,
|
| 1656 |
+
"learning_rate": 3.423103004429594e-06,
|
| 1657 |
+
"loss": 0.0644086742401123,
|
| 1658 |
+
"step": 13200,
|
| 1659 |
+
"token_acc": 0.9807682339525197
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 2.4404178607747067,
|
| 1663 |
+
"eval_loss": 2.0693349838256836,
|
| 1664 |
+
"eval_runtime": 36.3127,
|
| 1665 |
+
"eval_samples_per_second": 426.352,
|
| 1666 |
+
"eval_steps_per_second": 0.854,
|
| 1667 |
+
"eval_token_acc": 0.841946021972271,
|
| 1668 |
+
"step": 13200
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 2.458907275584728,
|
| 1672 |
+
"grad_norm": 1.422037124633789,
|
| 1673 |
+
"learning_rate": 3.352975794764537e-06,
|
| 1674 |
+
"loss": 0.06398173332214356,
|
| 1675 |
+
"step": 13300,
|
| 1676 |
+
"token_acc": 0.980901697909918
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 2.477396690394749,
|
| 1680 |
+
"grad_norm": 1.5547071695327759,
|
| 1681 |
+
"learning_rate": 3.283210159067361e-06,
|
| 1682 |
+
"loss": 0.06425526142120361,
|
| 1683 |
+
"step": 13400,
|
| 1684 |
+
"token_acc": 0.9809018621843667
|
| 1685 |
+
},
|
| 1686 |
+
{
|
| 1687 |
+
"epoch": 2.477396690394749,
|
| 1688 |
+
"eval_loss": 2.0776376724243164,
|
| 1689 |
+
"eval_runtime": 33.734,
|
| 1690 |
+
"eval_samples_per_second": 458.944,
|
| 1691 |
+
"eval_steps_per_second": 0.919,
|
| 1692 |
+
"eval_token_acc": 0.8419010470968717,
|
| 1693 |
+
"step": 13400
|
| 1694 |
+
},
|
| 1695 |
+
{
|
| 1696 |
+
"epoch": 2.4958861052047703,
|
| 1697 |
+
"grad_norm": 1.3930739164352417,
|
| 1698 |
+
"learning_rate": 3.2138214131043054e-06,
|
| 1699 |
+
"loss": 0.06403549671173096,
|
| 1700 |
+
"step": 13500,
|
| 1701 |
+
"token_acc": 0.98071152886729
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 2.5143755200147915,
|
| 1705 |
+
"grad_norm": 1.4157453775405884,
|
| 1706 |
+
"learning_rate": 3.1448247899023853e-06,
|
| 1707 |
+
"loss": 0.06361576557159423,
|
| 1708 |
+
"step": 13600,
|
| 1709 |
+
"token_acc": 0.9810559879958736
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 2.5143755200147915,
|
| 1713 |
+
"eval_loss": 2.091146230697632,
|
| 1714 |
+
"eval_runtime": 34.4246,
|
| 1715 |
+
"eval_samples_per_second": 449.736,
|
| 1716 |
+
"eval_steps_per_second": 0.901,
|
| 1717 |
+
"eval_token_acc": 0.8419501105973072,
|
| 1718 |
+
"step": 13600
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 2.5328649348248127,
|
| 1722 |
+
"grad_norm": 1.4381816387176514,
|
| 1723 |
+
"learning_rate": 3.076235436405244e-06,
|
| 1724 |
+
"loss": 0.06331642150878906,
|
| 1725 |
+
"step": 13700,
|
| 1726 |
+
"token_acc": 0.9810819213369495
|
| 1727 |
+
},
|
| 1728 |
+
{
|
| 1729 |
+
"epoch": 2.551354349634834,
|
| 1730 |
+
"grad_norm": 1.3895676136016846,
|
| 1731 |
+
"learning_rate": 3.0080684101479318e-06,
|
| 1732 |
+
"loss": 0.06289718627929687,
|
| 1733 |
+
"step": 13800,
|
| 1734 |
+
"token_acc": 0.9812787838649004
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 2.551354349634834,
|
| 1738 |
+
"eval_loss": 2.0786871910095215,
|
| 1739 |
+
"eval_runtime": 34.5297,
|
| 1740 |
+
"eval_samples_per_second": 448.367,
|
| 1741 |
+
"eval_steps_per_second": 0.898,
|
| 1742 |
+
"eval_token_acc": 0.8415167163434608,
|
| 1743 |
+
"step": 13800
|
| 1744 |
+
},
|
| 1745 |
+
{
|
| 1746 |
+
"epoch": 2.5698437644448555,
|
| 1747 |
+
"grad_norm": 1.4897119998931885,
|
| 1748 |
+
"learning_rate": 2.9403386759512953e-06,
|
| 1749 |
+
"loss": 0.06274141311645508,
|
| 1750 |
+
"step": 13900,
|
| 1751 |
+
"token_acc": 0.9813825940424772
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"epoch": 2.5883331792548767,
|
| 1755 |
+
"grad_norm": 1.3548414707183838,
|
| 1756 |
+
"learning_rate": 2.8730611026367284e-06,
|
| 1757 |
+
"loss": 0.0629720163345337,
|
| 1758 |
+
"step": 14000,
|
| 1759 |
+
"token_acc": 0.9813991830519936
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 2.5883331792548767,
|
| 1763 |
+
"eval_loss": 2.0937342643737793,
|
| 1764 |
+
"eval_runtime": 35.6025,
|
| 1765 |
+
"eval_samples_per_second": 434.857,
|
| 1766 |
+
"eval_steps_per_second": 0.871,
|
| 1767 |
+
"eval_token_acc": 0.841733413470384,
|
| 1768 |
+
"step": 14000
|
| 1769 |
+
},
|
| 1770 |
+
{
|
| 1771 |
+
"epoch": 2.606822594064898,
|
| 1772 |
+
"grad_norm": 1.4049385786056519,
|
| 1773 |
+
"learning_rate": 2.806250459761991e-06,
|
| 1774 |
+
"loss": 0.06263180732727051,
|
| 1775 |
+
"step": 14100,
|
| 1776 |
+
"token_acc": 0.9814355815587421
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 2.625312008874919,
|
| 1780 |
+
"grad_norm": 1.4203133583068848,
|
| 1781 |
+
"learning_rate": 2.73992141437883e-06,
|
| 1782 |
+
"loss": 0.06235404491424561,
|
| 1783 |
+
"step": 14200,
|
| 1784 |
+
"token_acc": 0.9815383784743917
|
| 1785 |
+
},
|
| 1786 |
+
{
|
| 1787 |
+
"epoch": 2.625312008874919,
|
| 1788 |
+
"eval_loss": 2.093254327774048,
|
| 1789 |
+
"eval_runtime": 34.2532,
|
| 1790 |
+
"eval_samples_per_second": 451.986,
|
| 1791 |
+
"eval_steps_per_second": 0.905,
|
| 1792 |
+
"eval_token_acc": 0.8421668077242305,
|
| 1793 |
+
"step": 14200
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 2.6438014236849403,
|
| 1797 |
+
"grad_norm": 1.4090100526809692,
|
| 1798 |
+
"learning_rate": 2.6740885278130813e-06,
|
| 1799 |
+
"loss": 0.06217865943908692,
|
| 1800 |
+
"step": 14300,
|
| 1801 |
+
"token_acc": 0.9816351073874321
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 2.6622908384949615,
|
| 1805 |
+
"grad_norm": 1.544283151626587,
|
| 1806 |
+
"learning_rate": 2.6087662524680262e-06,
|
| 1807 |
+
"loss": 0.06190134048461914,
|
| 1808 |
+
"step": 14400,
|
| 1809 |
+
"token_acc": 0.9817486168053806
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 2.6622908384949615,
|
| 1813 |
+
"eval_loss": 2.104107618331909,
|
| 1814 |
+
"eval_runtime": 35.499,
|
| 1815 |
+
"eval_samples_per_second": 436.125,
|
| 1816 |
+
"eval_steps_per_second": 0.873,
|
| 1817 |
+
"eval_token_acc": 0.8421177442237949,
|
| 1818 |
+
"step": 14400
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"epoch": 2.6807802533049827,
|
| 1822 |
+
"grad_norm": 1.3947652578353882,
|
| 1823 |
+
"learning_rate": 2.543968928651598e-06,
|
| 1824 |
+
"loss": 0.06151036262512207,
|
| 1825 |
+
"step": 14500,
|
| 1826 |
+
"token_acc": 0.9819023120835572
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 2.6992696681150044,
|
| 1830 |
+
"grad_norm": 1.3458011150360107,
|
| 1831 |
+
"learning_rate": 2.479710781428257e-06,
|
| 1832 |
+
"loss": 0.06159080028533936,
|
| 1833 |
+
"step": 14600,
|
| 1834 |
+
"token_acc": 0.9818631378345509
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 2.6992696681150044,
|
| 1838 |
+
"eval_loss": 2.094994068145752,
|
| 1839 |
+
"eval_runtime": 33.5785,
|
| 1840 |
+
"eval_samples_per_second": 461.068,
|
| 1841 |
+
"eval_steps_per_second": 0.923,
|
| 1842 |
+
"eval_token_acc": 0.8418846925967266,
|
| 1843 |
+
"step": 14600
|
| 1844 |
+
},
|
| 1845 |
+
{
|
| 1846 |
+
"epoch": 2.7177590829250255,
|
| 1847 |
+
"grad_norm": 1.408795714378357,
|
| 1848 |
+
"learning_rate": 2.4160059174961186e-06,
|
| 1849 |
+
"loss": 0.06139151573181152,
|
| 1850 |
+
"step": 14700,
|
| 1851 |
+
"token_acc": 0.9819905832240715
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 2.7362484977350467,
|
| 1855 |
+
"grad_norm": 1.4133031368255615,
|
| 1856 |
+
"learning_rate": 2.3528683220901074e-06,
|
| 1857 |
+
"loss": 0.0609882926940918,
|
| 1858 |
+
"step": 14800,
|
| 1859 |
+
"token_acc": 0.9823142742827259
|
| 1860 |
+
},
|
| 1861 |
+
{
|
| 1862 |
+
"epoch": 2.7362484977350467,
|
| 1863 |
+
"eval_loss": 2.0976996421813965,
|
| 1864 |
+
"eval_runtime": 33.3092,
|
| 1865 |
+
"eval_samples_per_second": 464.796,
|
| 1866 |
+
"eval_steps_per_second": 0.931,
|
| 1867 |
+
"eval_token_acc": 0.8421300100989039,
|
| 1868 |
+
"step": 14800
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 2.754737912545068,
|
| 1872 |
+
"grad_norm": 1.4126615524291992,
|
| 1873 |
+
"learning_rate": 2.2903118559117304e-06,
|
| 1874 |
+
"loss": 0.060602970123291015,
|
| 1875 |
+
"step": 14900,
|
| 1876 |
+
"token_acc": 0.9822561412629522
|
| 1877 |
+
},
|
| 1878 |
+
{
|
| 1879 |
+
"epoch": 2.773227327355089,
|
| 1880 |
+
"grad_norm": 1.6250054836273193,
|
| 1881 |
+
"learning_rate": 2.228350252086237e-06,
|
| 1882 |
+
"loss": 0.0604179573059082,
|
| 1883 |
+
"step": 15000,
|
| 1884 |
+
"token_acc": 0.9824043692296843
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 2.773227327355089,
|
| 1888 |
+
"eval_loss": 2.1088783740997314,
|
| 1889 |
+
"eval_runtime": 33.6247,
|
| 1890 |
+
"eval_samples_per_second": 460.436,
|
| 1891 |
+
"eval_steps_per_second": 0.922,
|
| 1892 |
+
"eval_token_acc": 0.8421954280994844,
|
| 1893 |
+
"step": 15000
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 2.7917167421651103,
|
| 1897 |
+
"grad_norm": 1.4460028409957886,
|
| 1898 |
+
"learning_rate": 2.1669971131477456e-06,
|
| 1899 |
+
"loss": 0.06029815673828125,
|
| 1900 |
+
"step": 15100,
|
| 1901 |
+
"token_acc": 0.9825059793243858
|
| 1902 |
+
},
|
| 1903 |
+
{
|
| 1904 |
+
"epoch": 2.8102061569751315,
|
| 1905 |
+
"grad_norm": 1.4490859508514404,
|
| 1906 |
+
"learning_rate": 2.10626590805307e-06,
|
| 1907 |
+
"loss": 0.0601440954208374,
|
| 1908 |
+
"step": 15200,
|
| 1909 |
+
"token_acc": 0.9826502277453795
|
| 1910 |
+
},
|
| 1911 |
+
{
|
| 1912 |
+
"epoch": 2.8102061569751315,
|
| 1913 |
+
"eval_loss": 2.1240346431732178,
|
| 1914 |
+
"eval_runtime": 33.9984,
|
| 1915 |
+
"eval_samples_per_second": 455.375,
|
| 1916 |
+
"eval_steps_per_second": 0.912,
|
| 1917 |
+
"eval_token_acc": 0.8421586304741578,
|
| 1918 |
+
"step": 15200
|
| 1919 |
+
},
|
| 1920 |
+
{
|
| 1921 |
+
"epoch": 2.828695571785153,
|
| 1922 |
+
"grad_norm": 1.4754722118377686,
|
| 1923 |
+
"learning_rate": 2.046169969224849e-06,
|
| 1924 |
+
"loss": 0.060016231536865236,
|
| 1925 |
+
"step": 15300,
|
| 1926 |
+
"token_acc": 0.9825407490768502
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 2.8471849865951744,
|
| 1930 |
+
"grad_norm": 1.3742412328720093,
|
| 1931 |
+
"learning_rate": 1.9867224896246585e-06,
|
| 1932 |
+
"loss": 0.05975979804992676,
|
| 1933 |
+
"step": 15400,
|
| 1934 |
+
"token_acc": 0.9827976067946816
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 2.8471849865951744,
|
| 1938 |
+
"eval_loss": 2.121828556060791,
|
| 1939 |
+
"eval_runtime": 37.5456,
|
| 1940 |
+
"eval_samples_per_second": 412.352,
|
| 1941 |
+
"eval_steps_per_second": 0.826,
|
| 1942 |
+
"eval_token_acc": 0.8419664650974524,
|
| 1943 |
+
"step": 15400
|
| 1944 |
+
},
|
| 1945 |
+
{
|
| 1946 |
+
"epoch": 2.8656744014051956,
|
| 1947 |
+
"grad_norm": 1.4398285150527954,
|
| 1948 |
+
"learning_rate": 1.9279365198567423e-06,
|
| 1949 |
+
"loss": 0.05939002990722656,
|
| 1950 |
+
"step": 15500,
|
| 1951 |
+
"token_acc": 0.9828381809746751
|
| 1952 |
+
},
|
| 1953 |
+
{
|
| 1954 |
+
"epoch": 2.884163816215217,
|
| 1955 |
+
"grad_norm": 1.532787561416626,
|
| 1956 |
+
"learning_rate": 1.8698249653029842e-06,
|
| 1957 |
+
"loss": 0.0589198112487793,
|
| 1958 |
+
"step": 15600,
|
| 1959 |
+
"token_acc": 0.9829359296759188
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 2.884163816215217,
|
| 1963 |
+
"eval_loss": 2.1134450435638428,
|
| 1964 |
+
"eval_runtime": 34.2162,
|
| 1965 |
+
"eval_samples_per_second": 452.476,
|
| 1966 |
+
"eval_steps_per_second": 0.906,
|
| 1967 |
+
"eval_token_acc": 0.8422158712246659,
|
| 1968 |
+
"step": 15600
|
| 1969 |
+
},
|
| 1970 |
+
{
|
| 1971 |
+
"epoch": 2.902653231025238,
|
| 1972 |
+
"grad_norm": 1.4046063423156738,
|
| 1973 |
+
"learning_rate": 1.8124005832897783e-06,
|
| 1974 |
+
"loss": 0.05904512405395508,
|
| 1975 |
+
"step": 15700,
|
| 1976 |
+
"token_acc": 0.9830348199408157
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 2.921142645835259,
|
| 1980 |
+
"grad_norm": 1.3275549411773682,
|
| 1981 |
+
"learning_rate": 1.7556759802873768e-06,
|
| 1982 |
+
"loss": 0.05889378547668457,
|
| 1983 |
+
"step": 15800,
|
| 1984 |
+
"token_acc": 0.9831256456369819
|
| 1985 |
+
},
|
| 1986 |
+
{
|
| 1987 |
+
"epoch": 2.921142645835259,
|
| 1988 |
+
"eval_loss": 2.121903896331787,
|
| 1989 |
+
"eval_runtime": 33.4938,
|
| 1990 |
+
"eval_samples_per_second": 462.234,
|
| 1991 |
+
"eval_steps_per_second": 0.926,
|
| 1992 |
+
"eval_token_acc": 0.8419255788470895,
|
| 1993 |
+
"step": 15800
|
| 1994 |
+
},
|
| 1995 |
+
{
|
| 1996 |
+
"epoch": 2.9396320606452804,
|
| 1997 |
+
"grad_norm": 1.2650915384292603,
|
| 1998 |
+
"learning_rate": 1.6996636091423935e-06,
|
| 1999 |
+
"loss": 0.058217430114746095,
|
| 2000 |
+
"step": 15900,
|
| 2001 |
+
"token_acc": 0.9832753492351027
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 2.958121475455302,
|
| 2005 |
+
"grad_norm": 1.4132380485534668,
|
| 2006 |
+
"learning_rate": 1.6443757663439818e-06,
|
| 2007 |
+
"loss": 0.05828729152679443,
|
| 2008 |
+
"step": 16000,
|
| 2009 |
+
"token_acc": 0.9833518565790249
|
| 2010 |
+
},
|
| 2011 |
+
{
|
| 2012 |
+
"epoch": 2.958121475455302,
|
| 2013 |
+
"eval_loss": 2.1201601028442383,
|
| 2014 |
+
"eval_runtime": 33.8459,
|
| 2015 |
+
"eval_samples_per_second": 457.426,
|
| 2016 |
+
"eval_steps_per_second": 0.916,
|
| 2017 |
+
"eval_token_acc": 0.8423385299757544,
|
| 2018 |
+
"step": 16000
|
| 2019 |
+
}
|
| 2020 |
+
],
|
| 2021 |
+
"logging_steps": 100,
|
| 2022 |
+
"max_steps": 21636,
|
| 2023 |
+
"num_input_tokens_seen": 0,
|
| 2024 |
+
"num_train_epochs": 4,
|
| 2025 |
+
"save_steps": 1000,
|
| 2026 |
+
"stateful_callbacks": {
|
| 2027 |
+
"TrainerControl": {
|
| 2028 |
+
"args": {
|
| 2029 |
+
"should_epoch_stop": false,
|
| 2030 |
+
"should_evaluate": false,
|
| 2031 |
+
"should_log": false,
|
| 2032 |
+
"should_save": true,
|
| 2033 |
+
"should_training_stop": false
|
| 2034 |
+
},
|
| 2035 |
+
"attributes": {}
|
| 2036 |
+
}
|
| 2037 |
+
},
|
| 2038 |
+
"total_flos": 2.1852680813254017e+19,
|
| 2039 |
+
"train_batch_size": 64,
|
| 2040 |
+
"trial_name": null,
|
| 2041 |
+
"trial_params": null
|
| 2042 |
+
}
|
Qwen3-ASR-WSM-Min/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:98ffe9a4e14241f8fe2c817dd3fa2053d1fd141fbdc33566a2990f31c4e430a3
|
| 3 |
+
size 9425
|
Qwen3-ASR-WSM-Min/vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
Qwen3-ASR-WSM-Min/zero_to_fp32.py
ADDED
|
@@ -0,0 +1,760 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/usr/bin/env python
|
| 2 |
+
|
| 3 |
+
# Copyright (c) Microsoft Corporation.
|
| 4 |
+
# SPDX-License-Identifier: Apache-2.0
|
| 5 |
+
|
| 6 |
+
# DeepSpeed Team
|
| 7 |
+
|
| 8 |
+
# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
|
| 9 |
+
# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
|
| 10 |
+
# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
|
| 11 |
+
# application.
|
| 12 |
+
#
|
| 13 |
+
# example:
|
| 14 |
+
# python zero_to_fp32.py . output_dir/
|
| 15 |
+
# or
|
| 16 |
+
# python zero_to_fp32.py . output_dir/ --safe_serialization
|
| 17 |
+
|
| 18 |
+
import argparse
|
| 19 |
+
import torch
|
| 20 |
+
import glob
|
| 21 |
+
import math
|
| 22 |
+
import os
|
| 23 |
+
import re
|
| 24 |
+
import gc
|
| 25 |
+
import json
|
| 26 |
+
import numpy as np
|
| 27 |
+
from tqdm import tqdm
|
| 28 |
+
from collections import OrderedDict
|
| 29 |
+
from dataclasses import dataclass
|
| 30 |
+
|
| 31 |
+
# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
|
| 32 |
+
# DeepSpeed data structures it has to be available in the current python environment.
|
| 33 |
+
from deepspeed.utils import logger
|
| 34 |
+
from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
|
| 35 |
+
FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
|
| 36 |
+
FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
|
| 37 |
+
|
| 38 |
+
|
| 39 |
+
@dataclass
|
| 40 |
+
class zero_model_state:
|
| 41 |
+
buffers: dict()
|
| 42 |
+
param_shapes: dict()
|
| 43 |
+
shared_params: list
|
| 44 |
+
ds_version: int
|
| 45 |
+
frozen_param_shapes: dict()
|
| 46 |
+
frozen_param_fragments: dict()
|
| 47 |
+
|
| 48 |
+
|
| 49 |
+
debug = 0
|
| 50 |
+
|
| 51 |
+
# load to cpu
|
| 52 |
+
device = torch.device('cpu')
|
| 53 |
+
|
| 54 |
+
|
| 55 |
+
def atoi(text):
|
| 56 |
+
return int(text) if text.isdigit() else text
|
| 57 |
+
|
| 58 |
+
|
| 59 |
+
def natural_keys(text):
|
| 60 |
+
'''
|
| 61 |
+
alist.sort(key=natural_keys) sorts in human order
|
| 62 |
+
http://nedbatchelder.com/blog/200712/human_sorting.html
|
| 63 |
+
(See Toothy's implementation in the comments)
|
| 64 |
+
'''
|
| 65 |
+
return [atoi(c) for c in re.split(r'(\d+)', text)]
|
| 66 |
+
|
| 67 |
+
|
| 68 |
+
def get_model_state_file(checkpoint_dir, zero_stage):
|
| 69 |
+
if not os.path.isdir(checkpoint_dir):
|
| 70 |
+
raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
|
| 71 |
+
|
| 72 |
+
# there should be only one file
|
| 73 |
+
if zero_stage <= 2:
|
| 74 |
+
file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
|
| 75 |
+
elif zero_stage == 3:
|
| 76 |
+
file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
|
| 77 |
+
|
| 78 |
+
if not os.path.exists(file):
|
| 79 |
+
raise FileNotFoundError(f"can't find model states file at '{file}'")
|
| 80 |
+
|
| 81 |
+
return file
|
| 82 |
+
|
| 83 |
+
|
| 84 |
+
def get_checkpoint_files(checkpoint_dir, glob_pattern):
|
| 85 |
+
# XXX: need to test that this simple glob rule works for multi-node setup too
|
| 86 |
+
ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
|
| 87 |
+
|
| 88 |
+
if len(ckpt_files) == 0:
|
| 89 |
+
raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
|
| 90 |
+
|
| 91 |
+
return ckpt_files
|
| 92 |
+
|
| 93 |
+
|
| 94 |
+
def get_optim_files(checkpoint_dir):
|
| 95 |
+
return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
|
| 96 |
+
|
| 97 |
+
|
| 98 |
+
def get_model_state_files(checkpoint_dir):
|
| 99 |
+
return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
|
| 100 |
+
|
| 101 |
+
|
| 102 |
+
def parse_model_states(files):
|
| 103 |
+
zero_model_states = []
|
| 104 |
+
for file in files:
|
| 105 |
+
state_dict = torch.load(file, map_location=device, weights_only=False)
|
| 106 |
+
|
| 107 |
+
if BUFFER_NAMES not in state_dict:
|
| 108 |
+
raise ValueError(f"{file} is not a model state checkpoint")
|
| 109 |
+
buffer_names = state_dict[BUFFER_NAMES]
|
| 110 |
+
if debug:
|
| 111 |
+
print("Found buffers:", buffer_names)
|
| 112 |
+
|
| 113 |
+
# recover just the buffers while restoring them to fp32 if they were saved in fp16
|
| 114 |
+
buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
|
| 115 |
+
param_shapes = state_dict[PARAM_SHAPES]
|
| 116 |
+
|
| 117 |
+
# collect parameters that are included in param_shapes
|
| 118 |
+
param_names = []
|
| 119 |
+
for s in param_shapes:
|
| 120 |
+
for name in s.keys():
|
| 121 |
+
param_names.append(name)
|
| 122 |
+
|
| 123 |
+
# update with frozen parameters
|
| 124 |
+
frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
|
| 125 |
+
if frozen_param_shapes is not None:
|
| 126 |
+
if debug:
|
| 127 |
+
print(f"Found frozen_param_shapes: {frozen_param_shapes}")
|
| 128 |
+
param_names += list(frozen_param_shapes.keys())
|
| 129 |
+
|
| 130 |
+
# handle shared params
|
| 131 |
+
shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
|
| 132 |
+
|
| 133 |
+
ds_version = state_dict.get(DS_VERSION, None)
|
| 134 |
+
|
| 135 |
+
frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
|
| 136 |
+
|
| 137 |
+
z_model_state = zero_model_state(buffers=buffers,
|
| 138 |
+
param_shapes=param_shapes,
|
| 139 |
+
shared_params=shared_params,
|
| 140 |
+
ds_version=ds_version,
|
| 141 |
+
frozen_param_shapes=frozen_param_shapes,
|
| 142 |
+
frozen_param_fragments=frozen_param_fragments)
|
| 143 |
+
zero_model_states.append(z_model_state)
|
| 144 |
+
|
| 145 |
+
return zero_model_states
|
| 146 |
+
|
| 147 |
+
|
| 148 |
+
def parse_optim_states(files, ds_checkpoint_dir):
|
| 149 |
+
total_files = len(files)
|
| 150 |
+
state_dicts = []
|
| 151 |
+
for f in tqdm(files, desc='Loading checkpoint shards'):
|
| 152 |
+
state_dict = torch.load(f, map_location=device, mmap=True, weights_only=False)
|
| 153 |
+
# immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
|
| 154 |
+
# and also handle the case where it was already removed by another helper script
|
| 155 |
+
state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
|
| 156 |
+
state_dicts.append(state_dict)
|
| 157 |
+
|
| 158 |
+
if ZERO_STAGE not in state_dicts[0][OPTIMIZER_STATE_DICT]:
|
| 159 |
+
raise ValueError(f"{files[0]} is not a zero checkpoint")
|
| 160 |
+
zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
|
| 161 |
+
world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
|
| 162 |
+
|
| 163 |
+
# For ZeRO-2 each param group can have different partition_count as data parallelism for expert
|
| 164 |
+
# parameters can be different from data parallelism for non-expert parameters. So we can just
|
| 165 |
+
# use the max of the partition_count to get the dp world_size.
|
| 166 |
+
|
| 167 |
+
if type(world_size) is list:
|
| 168 |
+
world_size = max(world_size)
|
| 169 |
+
|
| 170 |
+
if world_size != total_files:
|
| 171 |
+
raise ValueError(
|
| 172 |
+
f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
|
| 173 |
+
"Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
|
| 174 |
+
)
|
| 175 |
+
|
| 176 |
+
# the groups are named differently in each stage
|
| 177 |
+
if zero_stage <= 2:
|
| 178 |
+
fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
|
| 179 |
+
elif zero_stage == 3:
|
| 180 |
+
fp32_groups_key = FP32_FLAT_GROUPS
|
| 181 |
+
else:
|
| 182 |
+
raise ValueError(f"unknown zero stage {zero_stage}")
|
| 183 |
+
|
| 184 |
+
fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
|
| 185 |
+
return zero_stage, world_size, fp32_flat_groups
|
| 186 |
+
|
| 187 |
+
|
| 188 |
+
def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
|
| 189 |
+
"""
|
| 190 |
+
Returns fp32 state_dict reconstructed from ds checkpoint
|
| 191 |
+
|
| 192 |
+
Args:
|
| 193 |
+
- ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
|
| 194 |
+
|
| 195 |
+
"""
|
| 196 |
+
print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
|
| 197 |
+
|
| 198 |
+
optim_files = get_optim_files(ds_checkpoint_dir)
|
| 199 |
+
zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
|
| 200 |
+
print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
|
| 201 |
+
|
| 202 |
+
model_files = get_model_state_files(ds_checkpoint_dir)
|
| 203 |
+
|
| 204 |
+
zero_model_states = parse_model_states(model_files)
|
| 205 |
+
print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
|
| 206 |
+
|
| 207 |
+
if zero_stage <= 2:
|
| 208 |
+
return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
|
| 209 |
+
exclude_frozen_parameters)
|
| 210 |
+
elif zero_stage == 3:
|
| 211 |
+
return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
|
| 212 |
+
exclude_frozen_parameters)
|
| 213 |
+
|
| 214 |
+
|
| 215 |
+
def _zero2_merge_frozen_params(state_dict, zero_model_states):
|
| 216 |
+
if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
|
| 217 |
+
return
|
| 218 |
+
|
| 219 |
+
frozen_param_shapes = zero_model_states[0].frozen_param_shapes
|
| 220 |
+
frozen_param_fragments = zero_model_states[0].frozen_param_fragments
|
| 221 |
+
|
| 222 |
+
if debug:
|
| 223 |
+
num_elem = sum(s.numel() for s in frozen_param_shapes.values())
|
| 224 |
+
print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
|
| 225 |
+
|
| 226 |
+
wanted_params = len(frozen_param_shapes)
|
| 227 |
+
wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
|
| 228 |
+
avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
|
| 229 |
+
print(f'Frozen params: Have {avail_numel} numels to process.')
|
| 230 |
+
print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
|
| 231 |
+
|
| 232 |
+
total_params = 0
|
| 233 |
+
total_numel = 0
|
| 234 |
+
for name, shape in frozen_param_shapes.items():
|
| 235 |
+
total_params += 1
|
| 236 |
+
unpartitioned_numel = shape.numel()
|
| 237 |
+
total_numel += unpartitioned_numel
|
| 238 |
+
|
| 239 |
+
state_dict[name] = frozen_param_fragments[name]
|
| 240 |
+
|
| 241 |
+
if debug:
|
| 242 |
+
print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
|
| 243 |
+
|
| 244 |
+
print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
|
| 245 |
+
|
| 246 |
+
|
| 247 |
+
def _has_callable(obj, fn):
|
| 248 |
+
attr = getattr(obj, fn, None)
|
| 249 |
+
return callable(attr)
|
| 250 |
+
|
| 251 |
+
|
| 252 |
+
def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
|
| 253 |
+
param_shapes = zero_model_states[0].param_shapes
|
| 254 |
+
|
| 255 |
+
# Reconstruction protocol:
|
| 256 |
+
#
|
| 257 |
+
# XXX: document this
|
| 258 |
+
|
| 259 |
+
if debug:
|
| 260 |
+
for i in range(world_size):
|
| 261 |
+
for j in range(len(fp32_flat_groups[0])):
|
| 262 |
+
print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
|
| 263 |
+
|
| 264 |
+
# XXX: memory usage doubles here (zero2)
|
| 265 |
+
num_param_groups = len(fp32_flat_groups[0])
|
| 266 |
+
merged_single_partition_of_fp32_groups = []
|
| 267 |
+
for i in range(num_param_groups):
|
| 268 |
+
merged_partitions = [sd[i] for sd in fp32_flat_groups]
|
| 269 |
+
full_single_fp32_vector = torch.cat(merged_partitions, 0)
|
| 270 |
+
merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
|
| 271 |
+
avail_numel = sum(
|
| 272 |
+
[full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
|
| 273 |
+
|
| 274 |
+
if debug:
|
| 275 |
+
wanted_params = sum([len(shapes) for shapes in param_shapes])
|
| 276 |
+
wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
|
| 277 |
+
# not asserting if there is a mismatch due to possible padding
|
| 278 |
+
print(f"Have {avail_numel} numels to process.")
|
| 279 |
+
print(f"Need {wanted_numel} numels in {wanted_params} params.")
|
| 280 |
+
|
| 281 |
+
# params
|
| 282 |
+
# XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
|
| 283 |
+
# out-of-core computing solution
|
| 284 |
+
total_numel = 0
|
| 285 |
+
total_params = 0
|
| 286 |
+
for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
|
| 287 |
+
offset = 0
|
| 288 |
+
avail_numel = full_single_fp32_vector.numel()
|
| 289 |
+
for name, shape in shapes.items():
|
| 290 |
+
|
| 291 |
+
unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
|
| 292 |
+
total_numel += unpartitioned_numel
|
| 293 |
+
total_params += 1
|
| 294 |
+
|
| 295 |
+
if debug:
|
| 296 |
+
print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
|
| 297 |
+
state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
|
| 298 |
+
offset += unpartitioned_numel
|
| 299 |
+
|
| 300 |
+
# Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
|
| 301 |
+
# avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
|
| 302 |
+
# paddings performed in the code it's almost impossible to predict the exact numbers w/o the
|
| 303 |
+
# live optimizer object, so we are checking that the numbers are within the right range
|
| 304 |
+
align_to = 2 * world_size
|
| 305 |
+
|
| 306 |
+
def zero2_align(x):
|
| 307 |
+
return align_to * math.ceil(x / align_to)
|
| 308 |
+
|
| 309 |
+
if debug:
|
| 310 |
+
print(f"original offset={offset}, avail_numel={avail_numel}")
|
| 311 |
+
|
| 312 |
+
offset = zero2_align(offset)
|
| 313 |
+
avail_numel = zero2_align(avail_numel)
|
| 314 |
+
|
| 315 |
+
if debug:
|
| 316 |
+
print(f"aligned offset={offset}, avail_numel={avail_numel}")
|
| 317 |
+
|
| 318 |
+
# Sanity check
|
| 319 |
+
if offset != avail_numel:
|
| 320 |
+
raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
|
| 321 |
+
|
| 322 |
+
print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
|
| 323 |
+
|
| 324 |
+
|
| 325 |
+
def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
|
| 326 |
+
exclude_frozen_parameters):
|
| 327 |
+
state_dict = OrderedDict()
|
| 328 |
+
|
| 329 |
+
# buffers
|
| 330 |
+
buffers = zero_model_states[0].buffers
|
| 331 |
+
state_dict.update(buffers)
|
| 332 |
+
if debug:
|
| 333 |
+
print(f"added {len(buffers)} buffers")
|
| 334 |
+
|
| 335 |
+
if not exclude_frozen_parameters:
|
| 336 |
+
_zero2_merge_frozen_params(state_dict, zero_model_states)
|
| 337 |
+
|
| 338 |
+
_zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
|
| 339 |
+
|
| 340 |
+
# recover shared parameters
|
| 341 |
+
for pair in zero_model_states[0].shared_params:
|
| 342 |
+
if pair[1] in state_dict:
|
| 343 |
+
state_dict[pair[0]] = state_dict[pair[1]]
|
| 344 |
+
|
| 345 |
+
return state_dict
|
| 346 |
+
|
| 347 |
+
|
| 348 |
+
def zero3_partitioned_param_info(unpartitioned_numel, world_size):
|
| 349 |
+
remainder = unpartitioned_numel % world_size
|
| 350 |
+
padding_numel = (world_size - remainder) if remainder else 0
|
| 351 |
+
partitioned_numel = math.ceil(unpartitioned_numel / world_size)
|
| 352 |
+
return partitioned_numel, padding_numel
|
| 353 |
+
|
| 354 |
+
|
| 355 |
+
def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
|
| 356 |
+
if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
|
| 357 |
+
return
|
| 358 |
+
|
| 359 |
+
if debug:
|
| 360 |
+
for i in range(world_size):
|
| 361 |
+
num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
|
| 362 |
+
print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
|
| 363 |
+
|
| 364 |
+
frozen_param_shapes = zero_model_states[0].frozen_param_shapes
|
| 365 |
+
wanted_params = len(frozen_param_shapes)
|
| 366 |
+
wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
|
| 367 |
+
avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
|
| 368 |
+
print(f'Frozen params: Have {avail_numel} numels to process.')
|
| 369 |
+
print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
|
| 370 |
+
|
| 371 |
+
total_params = 0
|
| 372 |
+
total_numel = 0
|
| 373 |
+
for name, shape in zero_model_states[0].frozen_param_shapes.items():
|
| 374 |
+
total_params += 1
|
| 375 |
+
unpartitioned_numel = shape.numel()
|
| 376 |
+
total_numel += unpartitioned_numel
|
| 377 |
+
|
| 378 |
+
param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
|
| 379 |
+
state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
|
| 380 |
+
|
| 381 |
+
partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
|
| 382 |
+
|
| 383 |
+
if debug:
|
| 384 |
+
print(
|
| 385 |
+
f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
|
| 386 |
+
)
|
| 387 |
+
|
| 388 |
+
print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
|
| 389 |
+
|
| 390 |
+
|
| 391 |
+
class GatheredTensor:
|
| 392 |
+
"""
|
| 393 |
+
A pseudo tensor that collects partitioned weights.
|
| 394 |
+
It is more memory efficient when there are multiple groups.
|
| 395 |
+
"""
|
| 396 |
+
|
| 397 |
+
def __init__(self, flat_groups, flat_groups_offset, offset, partitioned_numel, shape):
|
| 398 |
+
self.flat_groups = flat_groups
|
| 399 |
+
self.flat_groups_offset = flat_groups_offset
|
| 400 |
+
self.offset = offset
|
| 401 |
+
self.partitioned_numel = partitioned_numel
|
| 402 |
+
self.shape = shape
|
| 403 |
+
self.dtype = self.flat_groups[0][0].dtype
|
| 404 |
+
|
| 405 |
+
def contiguous(self):
|
| 406 |
+
"""
|
| 407 |
+
Merge partitioned weights from flat_groups into a single tensor.
|
| 408 |
+
"""
|
| 409 |
+
end_idx = self.offset + self.partitioned_numel
|
| 410 |
+
world_size = len(self.flat_groups)
|
| 411 |
+
pad_flat_param_chunks = []
|
| 412 |
+
|
| 413 |
+
for rank_i in range(world_size):
|
| 414 |
+
# for each rank, we need to collect weights from related group/groups
|
| 415 |
+
flat_groups_at_rank_i = self.flat_groups[rank_i]
|
| 416 |
+
start_group_id = None
|
| 417 |
+
end_group_id = None
|
| 418 |
+
for group_id in range(len(self.flat_groups_offset)):
|
| 419 |
+
if self.flat_groups_offset[group_id] <= self.offset < self.flat_groups_offset[group_id + 1]:
|
| 420 |
+
start_group_id = group_id
|
| 421 |
+
if self.flat_groups_offset[group_id] < end_idx <= self.flat_groups_offset[group_id + 1]:
|
| 422 |
+
end_group_id = group_id
|
| 423 |
+
break
|
| 424 |
+
# collect weights from related group/groups
|
| 425 |
+
for group_id in range(start_group_id, end_group_id + 1):
|
| 426 |
+
flat_tensor = flat_groups_at_rank_i[group_id]
|
| 427 |
+
start_offset = self.offset - self.flat_groups_offset[group_id]
|
| 428 |
+
end_offset = min(end_idx, self.flat_groups_offset[group_id + 1]) - self.flat_groups_offset[group_id]
|
| 429 |
+
pad_flat_param_chunks.append(flat_tensor[start_offset:end_offset])
|
| 430 |
+
|
| 431 |
+
# collect weights from all ranks
|
| 432 |
+
pad_flat_param = torch.cat(pad_flat_param_chunks, dim=0)
|
| 433 |
+
param = pad_flat_param[:self.shape.numel()].view(self.shape).contiguous()
|
| 434 |
+
return param
|
| 435 |
+
|
| 436 |
+
|
| 437 |
+
def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
|
| 438 |
+
param_shapes = zero_model_states[0].param_shapes
|
| 439 |
+
avail_numel = sum([flat_group.numel() for flat_group in fp32_flat_groups[0]]) * world_size
|
| 440 |
+
|
| 441 |
+
# Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
|
| 442 |
+
# param, re-consolidating each param, while dealing with padding if any
|
| 443 |
+
|
| 444 |
+
# merge list of dicts, preserving order
|
| 445 |
+
param_shapes = {k: v for d in param_shapes for k, v in d.items()}
|
| 446 |
+
|
| 447 |
+
if debug:
|
| 448 |
+
for i in range(world_size):
|
| 449 |
+
print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
|
| 450 |
+
|
| 451 |
+
wanted_params = len(param_shapes)
|
| 452 |
+
wanted_numel = sum(shape.numel() for shape in param_shapes.values())
|
| 453 |
+
# not asserting if there is a mismatch due to possible padding
|
| 454 |
+
avail_numel = fp32_flat_groups[0].numel() * world_size
|
| 455 |
+
print(f"Trainable params: Have {avail_numel} numels to process.")
|
| 456 |
+
print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
|
| 457 |
+
|
| 458 |
+
# params
|
| 459 |
+
# XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
|
| 460 |
+
# out-of-core computing solution
|
| 461 |
+
offset = 0
|
| 462 |
+
total_numel = 0
|
| 463 |
+
total_params = 0
|
| 464 |
+
flat_groups_offset = [0] + list(np.cumsum([flat_tensor.numel() for flat_tensor in fp32_flat_groups[0]]))
|
| 465 |
+
for name, shape in tqdm(param_shapes.items(), desc='Gathering sharded weights'):
|
| 466 |
+
unpartitioned_numel = shape.numel()
|
| 467 |
+
total_numel += unpartitioned_numel
|
| 468 |
+
total_params += 1
|
| 469 |
+
partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
|
| 470 |
+
|
| 471 |
+
if debug:
|
| 472 |
+
print(
|
| 473 |
+
f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
|
| 474 |
+
)
|
| 475 |
+
|
| 476 |
+
# memory efficient tensor
|
| 477 |
+
tensor = GatheredTensor(fp32_flat_groups, flat_groups_offset, offset, partitioned_numel, shape)
|
| 478 |
+
state_dict[name] = tensor
|
| 479 |
+
offset += partitioned_numel
|
| 480 |
+
|
| 481 |
+
offset *= world_size
|
| 482 |
+
|
| 483 |
+
# Sanity check
|
| 484 |
+
if offset != avail_numel:
|
| 485 |
+
raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
|
| 486 |
+
|
| 487 |
+
print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
|
| 488 |
+
|
| 489 |
+
|
| 490 |
+
def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
|
| 491 |
+
exclude_frozen_parameters):
|
| 492 |
+
state_dict = OrderedDict()
|
| 493 |
+
|
| 494 |
+
# buffers
|
| 495 |
+
buffers = zero_model_states[0].buffers
|
| 496 |
+
state_dict.update(buffers)
|
| 497 |
+
if debug:
|
| 498 |
+
print(f"added {len(buffers)} buffers")
|
| 499 |
+
|
| 500 |
+
if not exclude_frozen_parameters:
|
| 501 |
+
_zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
|
| 502 |
+
|
| 503 |
+
_zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
|
| 504 |
+
|
| 505 |
+
# recover shared parameters
|
| 506 |
+
for pair in zero_model_states[0].shared_params:
|
| 507 |
+
if pair[1] in state_dict:
|
| 508 |
+
state_dict[pair[0]] = state_dict[pair[1]]
|
| 509 |
+
|
| 510 |
+
return state_dict
|
| 511 |
+
|
| 512 |
+
|
| 513 |
+
def to_torch_tensor(state_dict, return_empty_tensor=False):
|
| 514 |
+
"""
|
| 515 |
+
Convert state_dict of GatheredTensor to torch tensor
|
| 516 |
+
"""
|
| 517 |
+
torch_state_dict = {}
|
| 518 |
+
converted_tensors = {}
|
| 519 |
+
for name, tensor in state_dict.items():
|
| 520 |
+
tensor_id = id(tensor)
|
| 521 |
+
if tensor_id in converted_tensors: # shared tensors
|
| 522 |
+
shared_tensor = torch_state_dict[converted_tensors[tensor_id]]
|
| 523 |
+
torch_state_dict[name] = shared_tensor
|
| 524 |
+
else:
|
| 525 |
+
converted_tensors[tensor_id] = name
|
| 526 |
+
if return_empty_tensor:
|
| 527 |
+
torch_state_dict[name] = torch.empty(tensor.shape, dtype=tensor.dtype)
|
| 528 |
+
else:
|
| 529 |
+
torch_state_dict[name] = tensor.contiguous()
|
| 530 |
+
return torch_state_dict
|
| 531 |
+
|
| 532 |
+
|
| 533 |
+
def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
|
| 534 |
+
tag=None,
|
| 535 |
+
exclude_frozen_parameters=False,
|
| 536 |
+
lazy_mode=False):
|
| 537 |
+
"""
|
| 538 |
+
Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
|
| 539 |
+
``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
|
| 540 |
+
via a model hub.
|
| 541 |
+
|
| 542 |
+
Args:
|
| 543 |
+
- ``checkpoint_dir``: path to the desired checkpoint folder
|
| 544 |
+
- ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
|
| 545 |
+
- ``exclude_frozen_parameters``: exclude frozen parameters
|
| 546 |
+
- ``lazy_mode``: get state_dict in lazy mode. It returns a dict of pesduo tensor instead of torch tensor, which is more memory efficient.
|
| 547 |
+
Convert the pesduo tensor to torch tensor by ``.contiguous()``
|
| 548 |
+
|
| 549 |
+
Returns:
|
| 550 |
+
- pytorch ``state_dict``
|
| 551 |
+
|
| 552 |
+
A typical usage might be ::
|
| 553 |
+
|
| 554 |
+
from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
|
| 555 |
+
# do the training and checkpoint saving
|
| 556 |
+
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
|
| 557 |
+
model = model.cpu() # move to cpu
|
| 558 |
+
model.load_state_dict(state_dict)
|
| 559 |
+
# submit to model hub or save the model to share with others
|
| 560 |
+
|
| 561 |
+
In this example the ``model`` will no longer be usable in the deepspeed context of the same
|
| 562 |
+
application. i.e. you will need to re-initialize the deepspeed engine, since
|
| 563 |
+
``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
|
| 564 |
+
|
| 565 |
+
If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
|
| 566 |
+
|
| 567 |
+
Note: the above usage may not work if your application doesn't have sufficient free CPU memory.
|
| 568 |
+
You may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
|
| 569 |
+
the checkpoint. Or you can load state_dict in lazy mode ::
|
| 570 |
+
|
| 571 |
+
from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
|
| 572 |
+
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, lazy_mode=True) # not on cpu
|
| 573 |
+
for name, lazy_tensor in state_dict.item():
|
| 574 |
+
tensor = lazy_tensor.contiguous() # to cpu
|
| 575 |
+
print(name, tensor)
|
| 576 |
+
# del tensor to release memory if it no longer in use
|
| 577 |
+
"""
|
| 578 |
+
if tag is None:
|
| 579 |
+
latest_path = os.path.join(checkpoint_dir, 'latest')
|
| 580 |
+
if os.path.isfile(latest_path):
|
| 581 |
+
with open(latest_path, 'r') as fd:
|
| 582 |
+
tag = fd.read().strip()
|
| 583 |
+
else:
|
| 584 |
+
raise ValueError(f"Unable to find 'latest' file at {latest_path}")
|
| 585 |
+
|
| 586 |
+
ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
|
| 587 |
+
|
| 588 |
+
if not os.path.isdir(ds_checkpoint_dir):
|
| 589 |
+
raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
|
| 590 |
+
|
| 591 |
+
state_dict = _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
|
| 592 |
+
if lazy_mode:
|
| 593 |
+
return state_dict
|
| 594 |
+
else:
|
| 595 |
+
return to_torch_tensor(state_dict)
|
| 596 |
+
|
| 597 |
+
|
| 598 |
+
def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir,
|
| 599 |
+
output_dir,
|
| 600 |
+
max_shard_size="5GB",
|
| 601 |
+
safe_serialization=False,
|
| 602 |
+
tag=None,
|
| 603 |
+
exclude_frozen_parameters=False):
|
| 604 |
+
"""
|
| 605 |
+
Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
|
| 606 |
+
loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
|
| 607 |
+
|
| 608 |
+
Args:
|
| 609 |
+
- ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
|
| 610 |
+
- ``output_dir``: directory to the pytorch fp32 state_dict output files
|
| 611 |
+
- ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB
|
| 612 |
+
- ``safe_serialization``: whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).
|
| 613 |
+
- ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
|
| 614 |
+
- ``exclude_frozen_parameters``: exclude frozen parameters
|
| 615 |
+
"""
|
| 616 |
+
|
| 617 |
+
# Dependency pre-check
|
| 618 |
+
if safe_serialization:
|
| 619 |
+
try:
|
| 620 |
+
from safetensors.torch import save_file
|
| 621 |
+
except ImportError:
|
| 622 |
+
print('If you want to use `safe_serialization`, please `pip install safetensors`')
|
| 623 |
+
raise
|
| 624 |
+
if max_shard_size is not None:
|
| 625 |
+
try:
|
| 626 |
+
from huggingface_hub import split_torch_state_dict_into_shards
|
| 627 |
+
except ImportError:
|
| 628 |
+
print('If you want to use `max_shard_size`, please `pip install huggingface_hub`')
|
| 629 |
+
raise
|
| 630 |
+
|
| 631 |
+
# Convert zero checkpoint to state_dict
|
| 632 |
+
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
|
| 633 |
+
tag,
|
| 634 |
+
exclude_frozen_parameters,
|
| 635 |
+
lazy_mode=True)
|
| 636 |
+
|
| 637 |
+
# Shard the model if it is too big.
|
| 638 |
+
weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin"
|
| 639 |
+
if max_shard_size is not None:
|
| 640 |
+
filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors")
|
| 641 |
+
# an memory-efficient approach for sharding
|
| 642 |
+
empty_state_dict = to_torch_tensor(state_dict, return_empty_tensor=True)
|
| 643 |
+
state_dict_split = split_torch_state_dict_into_shards(empty_state_dict,
|
| 644 |
+
filename_pattern=filename_pattern,
|
| 645 |
+
max_shard_size=max_shard_size)
|
| 646 |
+
else:
|
| 647 |
+
from collections import namedtuple
|
| 648 |
+
StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"])
|
| 649 |
+
state_dict_split = StateDictSplit(is_sharded=False,
|
| 650 |
+
filename_to_tensors={weights_name: list(state_dict.keys())})
|
| 651 |
+
|
| 652 |
+
# Save the model by shard
|
| 653 |
+
os.makedirs(output_dir, exist_ok=True)
|
| 654 |
+
filename_to_tensors = state_dict_split.filename_to_tensors.items()
|
| 655 |
+
for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"):
|
| 656 |
+
shard_state_dict = {tensor_name: state_dict[tensor_name] for tensor_name in tensors}
|
| 657 |
+
shard_state_dict = to_torch_tensor(shard_state_dict)
|
| 658 |
+
output_path = os.path.join(output_dir, shard_file)
|
| 659 |
+
if safe_serialization:
|
| 660 |
+
save_file(shard_state_dict, output_path, metadata={"format": "pt"})
|
| 661 |
+
else:
|
| 662 |
+
torch.save(shard_state_dict, output_path)
|
| 663 |
+
# release the memory of current shard
|
| 664 |
+
for tensor_name in list(shard_state_dict.keys()):
|
| 665 |
+
del state_dict[tensor_name]
|
| 666 |
+
del shard_state_dict[tensor_name]
|
| 667 |
+
del shard_state_dict
|
| 668 |
+
gc.collect()
|
| 669 |
+
|
| 670 |
+
# Save index if sharded
|
| 671 |
+
if state_dict_split.is_sharded:
|
| 672 |
+
index = {
|
| 673 |
+
"metadata": state_dict_split.metadata,
|
| 674 |
+
"weight_map": state_dict_split.tensor_to_filename,
|
| 675 |
+
}
|
| 676 |
+
save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json"
|
| 677 |
+
save_index_file = os.path.join(output_dir, save_index_file)
|
| 678 |
+
with open(save_index_file, "w", encoding="utf-8") as f:
|
| 679 |
+
content = json.dumps(index, indent=2, sort_keys=True) + "\n"
|
| 680 |
+
f.write(content)
|
| 681 |
+
|
| 682 |
+
|
| 683 |
+
def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
|
| 684 |
+
"""
|
| 685 |
+
1. Put the provided model to cpu
|
| 686 |
+
2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
|
| 687 |
+
3. Load it into the provided model
|
| 688 |
+
|
| 689 |
+
Args:
|
| 690 |
+
- ``model``: the model object to update
|
| 691 |
+
- ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
|
| 692 |
+
- ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
|
| 693 |
+
|
| 694 |
+
Returns:
|
| 695 |
+
- ``model`: modified model
|
| 696 |
+
|
| 697 |
+
Make sure you have plenty of CPU memory available before you call this function. If you don't
|
| 698 |
+
have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
|
| 699 |
+
conveniently placed for you in the checkpoint folder.
|
| 700 |
+
|
| 701 |
+
A typical usage might be ::
|
| 702 |
+
|
| 703 |
+
from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
|
| 704 |
+
model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
|
| 705 |
+
# submit to model hub or save the model to share with others
|
| 706 |
+
|
| 707 |
+
Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
|
| 708 |
+
of the same application. i.e. you will need to re-initialize the deepspeed engine, since
|
| 709 |
+
``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
|
| 710 |
+
|
| 711 |
+
"""
|
| 712 |
+
logger.info("Extracting fp32 weights")
|
| 713 |
+
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
|
| 714 |
+
|
| 715 |
+
logger.info("Overwriting model with fp32 weights")
|
| 716 |
+
model = model.cpu()
|
| 717 |
+
model.load_state_dict(state_dict, strict=False)
|
| 718 |
+
|
| 719 |
+
return model
|
| 720 |
+
|
| 721 |
+
|
| 722 |
+
if __name__ == "__main__":
|
| 723 |
+
parser = argparse.ArgumentParser()
|
| 724 |
+
parser.add_argument("checkpoint_dir",
|
| 725 |
+
type=str,
|
| 726 |
+
help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
|
| 727 |
+
parser.add_argument("output_dir",
|
| 728 |
+
type=str,
|
| 729 |
+
help="directory to the pytorch fp32 state_dict output files"
|
| 730 |
+
"(e.g. path/checkpoint-12-output/)")
|
| 731 |
+
parser.add_argument(
|
| 732 |
+
"--max_shard_size",
|
| 733 |
+
type=str,
|
| 734 |
+
default="5GB",
|
| 735 |
+
help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size"
|
| 736 |
+
"lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`"
|
| 737 |
+
"We default it to 5GB in order for models to be able to run easily on free-tier google colab instances"
|
| 738 |
+
"without CPU OOM issues.")
|
| 739 |
+
parser.add_argument(
|
| 740 |
+
"--safe_serialization",
|
| 741 |
+
default=False,
|
| 742 |
+
action='store_true',
|
| 743 |
+
help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).")
|
| 744 |
+
parser.add_argument("-t",
|
| 745 |
+
"--tag",
|
| 746 |
+
type=str,
|
| 747 |
+
default=None,
|
| 748 |
+
help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
|
| 749 |
+
parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
|
| 750 |
+
parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
|
| 751 |
+
args = parser.parse_args()
|
| 752 |
+
|
| 753 |
+
debug = args.debug
|
| 754 |
+
|
| 755 |
+
convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
|
| 756 |
+
args.output_dir,
|
| 757 |
+
max_shard_size=args.max_shard_size,
|
| 758 |
+
safe_serialization=args.safe_serialization,
|
| 759 |
+
tag=args.tag,
|
| 760 |
+
exclude_frozen_parameters=args.exclude_frozen_parameters)
|