Upload Qwen3-ASR-WSM-Mandarin checkpoint-9000
Browse files- .gitattributes +1 -0
- Qwen3-ASR-WSM-Mandarin/added_tokens.json +64 -0
- Qwen3-ASR-WSM-Mandarin/args.json +403 -0
- Qwen3-ASR-WSM-Mandarin/chat_template.jinja +31 -0
- Qwen3-ASR-WSM-Mandarin/config.json +228 -0
- Qwen3-ASR-WSM-Mandarin/generation_config.json +9 -0
- Qwen3-ASR-WSM-Mandarin/global_step9000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +3 -0
- Qwen3-ASR-WSM-Mandarin/global_step9000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt +3 -0
- Qwen3-ASR-WSM-Mandarin/global_step9000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt +3 -0
- Qwen3-ASR-WSM-Mandarin/global_step9000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt +3 -0
- Qwen3-ASR-WSM-Mandarin/global_step9000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt +3 -0
- Qwen3-ASR-WSM-Mandarin/global_step9000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt +3 -0
- Qwen3-ASR-WSM-Mandarin/global_step9000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt +3 -0
- Qwen3-ASR-WSM-Mandarin/global_step9000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt +3 -0
- Qwen3-ASR-WSM-Mandarin/global_step9000/mp_rank_00_model_states.pt +3 -0
- Qwen3-ASR-WSM-Mandarin/latest +1 -0
- Qwen3-ASR-WSM-Mandarin/merges.txt +0 -0
- Qwen3-ASR-WSM-Mandarin/model.safetensors +3 -0
- Qwen3-ASR-WSM-Mandarin/preprocessor_config.json +14 -0
- Qwen3-ASR-WSM-Mandarin/rng_state_0.pth +3 -0
- Qwen3-ASR-WSM-Mandarin/rng_state_1.pth +3 -0
- Qwen3-ASR-WSM-Mandarin/rng_state_2.pth +3 -0
- Qwen3-ASR-WSM-Mandarin/rng_state_3.pth +3 -0
- Qwen3-ASR-WSM-Mandarin/rng_state_4.pth +3 -0
- Qwen3-ASR-WSM-Mandarin/rng_state_5.pth +3 -0
- Qwen3-ASR-WSM-Mandarin/rng_state_6.pth +3 -0
- Qwen3-ASR-WSM-Mandarin/rng_state_7.pth +3 -0
- Qwen3-ASR-WSM-Mandarin/scheduler.pt +3 -0
- Qwen3-ASR-WSM-Mandarin/special_tokens_map.json +44 -0
- Qwen3-ASR-WSM-Mandarin/tokenizer.json +3 -0
- Qwen3-ASR-WSM-Mandarin/tokenizer_config.json +549 -0
- Qwen3-ASR-WSM-Mandarin/trainer_state.json +1167 -0
- Qwen3-ASR-WSM-Mandarin/training_args.bin +3 -0
- Qwen3-ASR-WSM-Mandarin/vocab.json +0 -0
- Qwen3-ASR-WSM-Mandarin/zero_to_fp32.py +760 -0
.gitattributes
CHANGED
|
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
| 33 |
*.zip 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
|
|
|
|
|
|
| 33 |
*.zip 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
|
Qwen3-ASR-WSM-Mandarin/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-Mandarin/args.json
ADDED
|
@@ -0,0 +1,403 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"output_dir": "/home/work_nfs13/cjhe/Qwen3-ASR/ms-swift/exp/mandarin_kaiyuan_1wh+minspeech_qwen3asr_1.7b/v3-20260901-110901",
|
| 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/mandarin_kaiyuan_1wh+minspeech_qwen3asr_1.7b/v3-20260901-110901/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/mandarin_kaiyuan_1wh+minspeech_qwen3asr_1.7b/v3-20260901-110901",
|
| 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": 4096,
|
| 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 |
+
"dataset": [
|
| 258 |
+
"/home/work_nfs13/cjhe/Qwen3-ASR/ms-swift/data/kaiyuan_1wh+minspeech_mandarin_qwen3asr.jsonl"
|
| 259 |
+
],
|
| 260 |
+
"val_dataset": [
|
| 261 |
+
"/home/work_nfs13/cjhe/Qwen3-ASR/ms-swift/data/mandarin_test.jsonl"
|
| 262 |
+
],
|
| 263 |
+
"cached_dataset": [],
|
| 264 |
+
"cached_val_dataset": [],
|
| 265 |
+
"split_dataset_ratio": 0.0,
|
| 266 |
+
"dataset_num_proc": 12,
|
| 267 |
+
"load_from_cache_file": false,
|
| 268 |
+
"dataset_shuffle": true,
|
| 269 |
+
"val_dataset_shuffle": false,
|
| 270 |
+
"streaming": false,
|
| 271 |
+
"interleave_prob": null,
|
| 272 |
+
"stopping_strategy": "first_exhausted",
|
| 273 |
+
"shuffle_buffer_size": 1000,
|
| 274 |
+
"download_mode": "reuse_dataset_if_exists",
|
| 275 |
+
"columns": {},
|
| 276 |
+
"strict": false,
|
| 277 |
+
"disable_auto_column_mapping": false,
|
| 278 |
+
"model_name": null,
|
| 279 |
+
"model_author": null,
|
| 280 |
+
"custom_dataset_info": [],
|
| 281 |
+
"quant_method": null,
|
| 282 |
+
"quant_bits": null,
|
| 283 |
+
"hqq_axis": null,
|
| 284 |
+
"bnb_4bit_compute_dtype": "bfloat16",
|
| 285 |
+
"bnb_4bit_quant_type": "nf4",
|
| 286 |
+
"bnb_4bit_use_double_quant": true,
|
| 287 |
+
"bnb_4bit_quant_storage": null,
|
| 288 |
+
"max_new_tokens": 64,
|
| 289 |
+
"temperature": 0.0,
|
| 290 |
+
"top_k": null,
|
| 291 |
+
"top_p": null,
|
| 292 |
+
"repetition_penalty": null,
|
| 293 |
+
"num_beams": 1,
|
| 294 |
+
"stream": false,
|
| 295 |
+
"stop_words": [],
|
| 296 |
+
"logprobs": false,
|
| 297 |
+
"top_logprobs": null,
|
| 298 |
+
"structured_outputs_regex": null,
|
| 299 |
+
"adapters": [],
|
| 300 |
+
"external_plugins": [],
|
| 301 |
+
"custom_register_path": [],
|
| 302 |
+
"model_kwargs": {},
|
| 303 |
+
"enable_npu_model_patch": true,
|
| 304 |
+
"load_args": false,
|
| 305 |
+
"load_data_args": false,
|
| 306 |
+
"packing": false,
|
| 307 |
+
"packing_length": null,
|
| 308 |
+
"packing_num_proc": 1,
|
| 309 |
+
"lazy_tokenize": true,
|
| 310 |
+
"use_hf": false,
|
| 311 |
+
"ignore_args_error": false,
|
| 312 |
+
"use_swift_lora": false,
|
| 313 |
+
"freeze_parameters": [],
|
| 314 |
+
"freeze_parameters_regex": null,
|
| 315 |
+
"freeze_parameters_ratio": 0.0,
|
| 316 |
+
"trainable_parameters": [
|
| 317 |
+
"thinker.audio_tower.proj1",
|
| 318 |
+
"thinker.audio_tower.proj2"
|
| 319 |
+
],
|
| 320 |
+
"trainable_parameters_regex": null,
|
| 321 |
+
"freeze_llm": false,
|
| 322 |
+
"freeze_vit": false,
|
| 323 |
+
"freeze_aligner": false,
|
| 324 |
+
"target_modules": [
|
| 325 |
+
"all-linear"
|
| 326 |
+
],
|
| 327 |
+
"target_regex": null,
|
| 328 |
+
"target_parameters": null,
|
| 329 |
+
"modules_to_save": [],
|
| 330 |
+
"lora_rank": 8,
|
| 331 |
+
"lora_alpha": 32,
|
| 332 |
+
"lora_dropout": 0.05,
|
| 333 |
+
"lora_bias": "none",
|
| 334 |
+
"lora_dtype": null,
|
| 335 |
+
"lorap_lr_ratio": null,
|
| 336 |
+
"use_rslora": false,
|
| 337 |
+
"use_dora": false,
|
| 338 |
+
"lora_ga_batch_size": 2,
|
| 339 |
+
"lora_ga_iters": 2,
|
| 340 |
+
"lora_ga_max_length": 1024,
|
| 341 |
+
"lora_ga_direction": "ArB2r",
|
| 342 |
+
"lora_ga_scale": "stable",
|
| 343 |
+
"lora_ga_stable_gamma": 16,
|
| 344 |
+
"init_weights": true,
|
| 345 |
+
"fourier_n_frequency": 2000,
|
| 346 |
+
"fourier_scaling": 300.0,
|
| 347 |
+
"boft_block_size": 4,
|
| 348 |
+
"boft_block_num": 0,
|
| 349 |
+
"boft_n_butterfly_factor": 1,
|
| 350 |
+
"boft_dropout": 0.0,
|
| 351 |
+
"vera_rank": 256,
|
| 352 |
+
"vera_projection_prng_key": 0,
|
| 353 |
+
"vera_dropout": 0.0,
|
| 354 |
+
"vera_d_initial": 0.1,
|
| 355 |
+
"adapter_act": "gelu",
|
| 356 |
+
"adapter_length": 128,
|
| 357 |
+
"adalora_target_r": 8,
|
| 358 |
+
"adalora_init_r": 12,
|
| 359 |
+
"adalora_tinit": 0,
|
| 360 |
+
"adalora_tfinal": 0,
|
| 361 |
+
"adalora_deltaT": 1,
|
| 362 |
+
"adalora_beta1": 0.85,
|
| 363 |
+
"adalora_beta2": 0.85,
|
| 364 |
+
"adalora_orth_reg_weight": 0.5,
|
| 365 |
+
"llamapro_num_new_blocks": 4,
|
| 366 |
+
"llamapro_num_groups": null,
|
| 367 |
+
"reft_layer_key": null,
|
| 368 |
+
"reft_layers": null,
|
| 369 |
+
"reft_rank": 4,
|
| 370 |
+
"reft_intervention_type": "LoreftIntervention",
|
| 371 |
+
"reft_args": null,
|
| 372 |
+
"swanlab_token": null,
|
| 373 |
+
"swanlab_project": "ms-swift",
|
| 374 |
+
"swanlab_workspace": null,
|
| 375 |
+
"swanlab_exp_name": null,
|
| 376 |
+
"swanlab_notification_method": null,
|
| 377 |
+
"swanlab_webhook_url": null,
|
| 378 |
+
"swanlab_secret": null,
|
| 379 |
+
"swanlab_sender_email": null,
|
| 380 |
+
"swanlab_receiver_email": null,
|
| 381 |
+
"swanlab_smtp_server": null,
|
| 382 |
+
"swanlab_smtp_port": null,
|
| 383 |
+
"swanlab_email_language": "zh",
|
| 384 |
+
"swanlab_mode": "cloud",
|
| 385 |
+
"add_version": true,
|
| 386 |
+
"create_checkpoint_symlink": false,
|
| 387 |
+
"zero_hpz_partition_size": null,
|
| 388 |
+
"deepspeed_autotp_size": null,
|
| 389 |
+
"swift_version": "4.3.0",
|
| 390 |
+
"ckpt_dir": null,
|
| 391 |
+
"rank": 0,
|
| 392 |
+
"global_world_size": 8,
|
| 393 |
+
"local_world_size": 8,
|
| 394 |
+
"model_suffix": "Qwen3-ASR-1.7B",
|
| 395 |
+
"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)",
|
| 396 |
+
"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=[])",
|
| 397 |
+
"model_dir": "/home/work_nfs23/asr_data/ckpt/Qwen3-ASR-1.7B",
|
| 398 |
+
"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='')",
|
| 399 |
+
"_val_dataset_exists": true,
|
| 400 |
+
"hub": "<class 'swift.hub.hub.MSHub'>",
|
| 401 |
+
"evaluation_strategy": "steps",
|
| 402 |
+
"training_args": "Seq2SeqTrainingArguments(output_dir='/home/work_nfs13/cjhe/Qwen3-ASR/ms-swift/exp/mandarin_kaiyuan_1wh+minspeech_qwen3asr_1.7b/v3-20260901-110901', 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/mandarin_kaiyuan_1wh+minspeech_qwen3asr_1.7b/v3-20260901-110901/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/mandarin_kaiyuan_1wh+minspeech_qwen3asr_1.7b/v3-20260901-110901', 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)"
|
| 403 |
+
}
|
Qwen3-ASR-WSM-Mandarin/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-Mandarin/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-Mandarin/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-Mandarin/global_step9000/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 sha256:23aceb3fd14182f7d08bb98676bb5971a9b2d554c8dd673c107be1512593ce58
|
| 3 |
+
size 3057098949
|
Qwen3-ASR-WSM-Mandarin/global_step9000/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:f95434b5d8131409941bfb50634cc592cf17c15da5c67a2f9c19677579ff712d
|
| 3 |
+
size 3057090437
|
Qwen3-ASR-WSM-Mandarin/global_step9000/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 sha256:36cef1134a40f3878e49fef1c4322eacabcafcbb1b4d3c4b4e7c46ccac261dbe
|
| 3 |
+
size 3057089541
|
Qwen3-ASR-WSM-Mandarin/global_step9000/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 sha256:1081fdeaf916a70fa8c77b7f28b4bac650541fb8a3cd70adccf08ccadddc444a
|
| 3 |
+
size 3057091013
|
Qwen3-ASR-WSM-Mandarin/global_step9000/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 sha256:13b8800e903ea58889540adbf3dcf753078fbc8ea371cbe112fd698c12b15619
|
| 3 |
+
size 3057091077
|
Qwen3-ASR-WSM-Mandarin/global_step9000/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:f207f39d6ee837c12f92d56be828e384a377931a7b3b81c9a2481a438da555a5
|
| 3 |
+
size 3057091269
|
Qwen3-ASR-WSM-Mandarin/global_step9000/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 sha256:241886bc705cf41e3ce53e8cc172b019178c51abd69c957865699e40a885b4aa
|
| 3 |
+
size 3057091781
|
Qwen3-ASR-WSM-Mandarin/global_step9000/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 sha256:868256a7c246aa2f9d26ac45af1530fd14a079382bbd1f80f5d5c8cdb81a8419
|
| 3 |
+
size 3057091525
|
Qwen3-ASR-WSM-Mandarin/global_step9000/mp_rank_00_model_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7e96afc28698ba5bf6beec579dbad19747a1967b2b89205a3b09b4816167dd23
|
| 3 |
+
size 4076311909
|
Qwen3-ASR-WSM-Mandarin/latest
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
global_step9000
|
Qwen3-ASR-WSM-Mandarin/merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
Qwen3-ASR-WSM-Mandarin/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cb0e595d154faebd7c61fc087ba13b9f025b6b5067788f4d48b7fa67f606b46f
|
| 3 |
+
size 4698521600
|
Qwen3-ASR-WSM-Mandarin/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-Mandarin/rng_state_0.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:136eaa5709e64cee456d7a1e4a8e7944a5b1b27408f4cc38dcf1664b81461af5
|
| 3 |
+
size 16389
|
Qwen3-ASR-WSM-Mandarin/rng_state_1.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cbdd8437962f9a5a667d48f60467dac52410a1bf09d118706c46db44e09e0b36
|
| 3 |
+
size 16389
|
Qwen3-ASR-WSM-Mandarin/rng_state_2.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1fe8d464f2485f8000da0b8938cd180d3f884919d10a3517aaf6fb7f60c455c5
|
| 3 |
+
size 16389
|
Qwen3-ASR-WSM-Mandarin/rng_state_3.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c59a07dad272c52c8dcbe973f22313c243a3c2557d5a7062588fe1a4ffc262d3
|
| 3 |
+
size 16389
|
Qwen3-ASR-WSM-Mandarin/rng_state_4.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:44ec4b5017df660b58fdee45bad7827b7866a95810231fadd64ef3115c5e3441
|
| 3 |
+
size 16389
|
Qwen3-ASR-WSM-Mandarin/rng_state_5.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:574aab9e7bc35dc4c4c15fbe828e6b4223bc628c0651867ff724aef5b906571b
|
| 3 |
+
size 16389
|
Qwen3-ASR-WSM-Mandarin/rng_state_6.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7ee2633e76c4bb57cfb4f144bc5af02226d14c81ee6eac15bcf79604bd0bb936
|
| 3 |
+
size 16389
|
Qwen3-ASR-WSM-Mandarin/rng_state_7.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:217d7797164b53e1f00af708eccd1f07314eb4558dfae3a8af02d74b6a171471
|
| 3 |
+
size 16389
|
Qwen3-ASR-WSM-Mandarin/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:74a3b85648ff3a78742c900ac2206f247f7c36472f35c30179036e26ffa04b79
|
| 3 |
+
size 1465
|
Qwen3-ASR-WSM-Mandarin/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-Mandarin/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-Mandarin/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-Mandarin/trainer_state.json
ADDED
|
@@ -0,0 +1,1167 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 4600,
|
| 3 |
+
"best_metric": 1.36764193,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 1.1160714285714286,
|
| 6 |
+
"eval_steps": 200,
|
| 7 |
+
"global_step": 9000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.0001240079365079365,
|
| 14 |
+
"grad_norm": 116.32508850097656,
|
| 15 |
+
"learning_rate": 1.5479876160990715e-08,
|
| 16 |
+
"loss": 3.743511199951172,
|
| 17 |
+
"step": 1,
|
| 18 |
+
"token_acc": 0.5620549895936028
|
| 19 |
+
},
|
| 20 |
+
{
|
| 21 |
+
"epoch": 0.01240079365079365,
|
| 22 |
+
"grad_norm": 3.6368508338928223,
|
| 23 |
+
"learning_rate": 1.5479876160990715e-06,
|
| 24 |
+
"loss": 2.3589355777008363,
|
| 25 |
+
"step": 100,
|
| 26 |
+
"token_acc": 0.616584054935255
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.0248015873015873,
|
| 30 |
+
"grad_norm": 3.38437557220459,
|
| 31 |
+
"learning_rate": 3.095975232198143e-06,
|
| 32 |
+
"loss": 1.23188232421875,
|
| 33 |
+
"step": 200,
|
| 34 |
+
"token_acc": 0.7080392332642435
|
| 35 |
+
},
|
| 36 |
+
{
|
| 37 |
+
"epoch": 0.0248015873015873,
|
| 38 |
+
"eval_loss": 1.3806190490722656,
|
| 39 |
+
"eval_runtime": 47.1633,
|
| 40 |
+
"eval_samples_per_second": 129.507,
|
| 41 |
+
"eval_steps_per_second": 0.254,
|
| 42 |
+
"eval_token_acc": 0.6958059081648644,
|
| 43 |
+
"step": 200
|
| 44 |
+
},
|
| 45 |
+
{
|
| 46 |
+
"epoch": 0.03720238095238095,
|
| 47 |
+
"grad_norm": 4.100909233093262,
|
| 48 |
+
"learning_rate": 4.6439628482972135e-06,
|
| 49 |
+
"loss": 1.1396067810058594,
|
| 50 |
+
"step": 300,
|
| 51 |
+
"token_acc": 0.7236402590446511
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.0496031746031746,
|
| 55 |
+
"grad_norm": 3.010042190551758,
|
| 56 |
+
"learning_rate": 6.191950464396286e-06,
|
| 57 |
+
"loss": 1.093898162841797,
|
| 58 |
+
"step": 400,
|
| 59 |
+
"token_acc": 0.7316208399483557
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.0496031746031746,
|
| 63 |
+
"eval_loss": 1.3786414861679077,
|
| 64 |
+
"eval_runtime": 27.084,
|
| 65 |
+
"eval_samples_per_second": 225.52,
|
| 66 |
+
"eval_steps_per_second": 0.443,
|
| 67 |
+
"eval_token_acc": 0.70037023977433,
|
| 68 |
+
"step": 400
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.062003968253968256,
|
| 72 |
+
"grad_norm": 3.530548334121704,
|
| 73 |
+
"learning_rate": 7.739938080495358e-06,
|
| 74 |
+
"loss": 1.0639726257324218,
|
| 75 |
+
"step": 500,
|
| 76 |
+
"token_acc": 0.7364850685076729
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"epoch": 0.0744047619047619,
|
| 80 |
+
"grad_norm": 3.193807363510132,
|
| 81 |
+
"learning_rate": 9.287925696594427e-06,
|
| 82 |
+
"loss": 1.0408080291748047,
|
| 83 |
+
"step": 600,
|
| 84 |
+
"token_acc": 0.7407066170097333
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.0744047619047619,
|
| 88 |
+
"eval_loss": 1.4019232988357544,
|
| 89 |
+
"eval_runtime": 67.7573,
|
| 90 |
+
"eval_samples_per_second": 90.145,
|
| 91 |
+
"eval_steps_per_second": 0.177,
|
| 92 |
+
"eval_token_acc": 0.6978236169879329,
|
| 93 |
+
"step": 600
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.08680555555555555,
|
| 97 |
+
"grad_norm": 3.3848700523376465,
|
| 98 |
+
"learning_rate": 9.99992799258182e-06,
|
| 99 |
+
"loss": 1.0277099609375,
|
| 100 |
+
"step": 700,
|
| 101 |
+
"token_acc": 0.7429492628467165
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.0992063492063492,
|
| 105 |
+
"grad_norm": 2.9100406169891357,
|
| 106 |
+
"learning_rate": 9.999414369447362e-06,
|
| 107 |
+
"loss": 1.0123756408691407,
|
| 108 |
+
"step": 800,
|
| 109 |
+
"token_acc": 0.7454852366331679
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 0.0992063492063492,
|
| 113 |
+
"eval_loss": 1.3969780206680298,
|
| 114 |
+
"eval_runtime": 132.0204,
|
| 115 |
+
"eval_samples_per_second": 46.266,
|
| 116 |
+
"eval_steps_per_second": 0.091,
|
| 117 |
+
"eval_token_acc": 0.6986757561510735,
|
| 118 |
+
"step": 800
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 0.11160714285714286,
|
| 122 |
+
"grad_norm": 3.3894102573394775,
|
| 123 |
+
"learning_rate": 9.998406928998766e-06,
|
| 124 |
+
"loss": 0.9965998077392578,
|
| 125 |
+
"step": 900,
|
| 126 |
+
"token_acc": 0.7485039899002444
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 0.12400793650793651,
|
| 130 |
+
"grad_norm": 3.169595718383789,
|
| 131 |
+
"learning_rate": 9.996905770745992e-06,
|
| 132 |
+
"loss": 0.9868894958496094,
|
| 133 |
+
"step": 1000,
|
| 134 |
+
"token_acc": 0.7502872028865912
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.12400793650793651,
|
| 138 |
+
"eval_loss": 1.3937498331069946,
|
| 139 |
+
"eval_runtime": 26.1265,
|
| 140 |
+
"eval_samples_per_second": 233.786,
|
| 141 |
+
"eval_steps_per_second": 0.459,
|
| 142 |
+
"eval_token_acc": 0.6997825575928538,
|
| 143 |
+
"step": 1000
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.13640873015873015,
|
| 147 |
+
"grad_norm": 2.7775676250457764,
|
| 148 |
+
"learning_rate": 9.994911042965995e-06,
|
| 149 |
+
"loss": 0.9748902130126953,
|
| 150 |
+
"step": 1100,
|
| 151 |
+
"token_acc": 0.7527066958441567
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 0.1488095238095238,
|
| 155 |
+
"grad_norm": 2.690856456756592,
|
| 156 |
+
"learning_rate": 9.992422942688078e-06,
|
| 157 |
+
"loss": 0.9626938629150391,
|
| 158 |
+
"step": 1200,
|
| 159 |
+
"token_acc": 0.7544744513980693
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.1488095238095238,
|
| 163 |
+
"eval_loss": 1.4004549980163574,
|
| 164 |
+
"eval_runtime": 32.4044,
|
| 165 |
+
"eval_samples_per_second": 188.493,
|
| 166 |
+
"eval_steps_per_second": 0.37,
|
| 167 |
+
"eval_token_acc": 0.7011929948283968,
|
| 168 |
+
"step": 1200
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 0.16121031746031747,
|
| 172 |
+
"grad_norm": 2.808126926422119,
|
| 173 |
+
"learning_rate": 9.989441715674422e-06,
|
| 174 |
+
"loss": 0.9591330718994141,
|
| 175 |
+
"step": 1300,
|
| 176 |
+
"token_acc": 0.7549267489392193
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.1736111111111111,
|
| 180 |
+
"grad_norm": 2.804110288619995,
|
| 181 |
+
"learning_rate": 9.985967656395823e-06,
|
| 182 |
+
"loss": 0.9526300048828125,
|
| 183 |
+
"step": 1400,
|
| 184 |
+
"token_acc": 0.7563638622283915
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.1736111111111111,
|
| 188 |
+
"eval_loss": 1.380391001701355,
|
| 189 |
+
"eval_runtime": 24.4613,
|
| 190 |
+
"eval_samples_per_second": 249.701,
|
| 191 |
+
"eval_steps_per_second": 0.491,
|
| 192 |
+
"eval_token_acc": 0.7028776837486287,
|
| 193 |
+
"step": 1400
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.18601190476190477,
|
| 197 |
+
"grad_norm": 2.481344699859619,
|
| 198 |
+
"learning_rate": 9.982001108002598e-06,
|
| 199 |
+
"loss": 0.9452995300292969,
|
| 200 |
+
"step": 1500,
|
| 201 |
+
"token_acc": 0.7576550731548903
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.1984126984126984,
|
| 205 |
+
"grad_norm": 2.467794179916382,
|
| 206 |
+
"learning_rate": 9.977542462290686e-06,
|
| 207 |
+
"loss": 0.9419522094726562,
|
| 208 |
+
"step": 1600,
|
| 209 |
+
"token_acc": 0.7580019040671283
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 0.1984126984126984,
|
| 213 |
+
"eval_loss": 1.3702844381332397,
|
| 214 |
+
"eval_runtime": 27.4416,
|
| 215 |
+
"eval_samples_per_second": 222.582,
|
| 216 |
+
"eval_steps_per_second": 0.437,
|
| 217 |
+
"eval_token_acc": 0.7032596771665883,
|
| 218 |
+
"step": 1600
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.21081349206349206,
|
| 222 |
+
"grad_norm": 2.4546399116516113,
|
| 223 |
+
"learning_rate": 9.972592159662966e-06,
|
| 224 |
+
"loss": 0.9375972747802734,
|
| 225 |
+
"step": 1700,
|
| 226 |
+
"token_acc": 0.7596919036915363
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.22321428571428573,
|
| 230 |
+
"grad_norm": 2.5199294090270996,
|
| 231 |
+
"learning_rate": 9.967150689085734e-06,
|
| 232 |
+
"loss": 0.9338107299804688,
|
| 233 |
+
"step": 1800,
|
| 234 |
+
"token_acc": 0.7601298887700604
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.22321428571428573,
|
| 238 |
+
"eval_loss": 1.383194923400879,
|
| 239 |
+
"eval_runtime": 28.3699,
|
| 240 |
+
"eval_samples_per_second": 215.299,
|
| 241 |
+
"eval_steps_per_second": 0.423,
|
| 242 |
+
"eval_token_acc": 0.702946246669801,
|
| 243 |
+
"step": 1800
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.23561507936507936,
|
| 247 |
+
"grad_norm": 2.7069873809814453,
|
| 248 |
+
"learning_rate": 9.961218588040423e-06,
|
| 249 |
+
"loss": 0.9247418975830078,
|
| 250 |
+
"step": 1900,
|
| 251 |
+
"token_acc": 0.7611753312382803
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.24801587301587302,
|
| 255 |
+
"grad_norm": 2.5814383029937744,
|
| 256 |
+
"learning_rate": 9.954796442470502e-06,
|
| 257 |
+
"loss": 0.9223838043212891,
|
| 258 |
+
"step": 2000,
|
| 259 |
+
"token_acc": 0.7615163339863971
|
| 260 |
+
},
|
| 261 |
+
{
|
| 262 |
+
"epoch": 0.24801587301587302,
|
| 263 |
+
"eval_loss": 1.4053868055343628,
|
| 264 |
+
"eval_runtime": 26.9905,
|
| 265 |
+
"eval_samples_per_second": 226.302,
|
| 266 |
+
"eval_steps_per_second": 0.445,
|
| 267 |
+
"eval_token_acc": 0.7016141670584548,
|
| 268 |
+
"step": 2000
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.2604166666666667,
|
| 272 |
+
"grad_norm": 2.401449680328369,
|
| 273 |
+
"learning_rate": 9.947884886723606e-06,
|
| 274 |
+
"loss": 0.9166383361816406,
|
| 275 |
+
"step": 2100,
|
| 276 |
+
"token_acc": 0.7629393534950732
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.2728174603174603,
|
| 280 |
+
"grad_norm": 2.2782652378082275,
|
| 281 |
+
"learning_rate": 9.940484603488873e-06,
|
| 282 |
+
"loss": 0.91524658203125,
|
| 283 |
+
"step": 2200,
|
| 284 |
+
"token_acc": 0.7635179533923029
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.2728174603174603,
|
| 288 |
+
"eval_loss": 1.3930410146713257,
|
| 289 |
+
"eval_runtime": 25.1096,
|
| 290 |
+
"eval_samples_per_second": 243.254,
|
| 291 |
+
"eval_steps_per_second": 0.478,
|
| 292 |
+
"eval_token_acc": 0.7024858956276445,
|
| 293 |
+
"step": 2200
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 0.28521825396825395,
|
| 297 |
+
"grad_norm": 2.3035569190979004,
|
| 298 |
+
"learning_rate": 9.932596323729517e-06,
|
| 299 |
+
"loss": 0.911158447265625,
|
| 300 |
+
"step": 2300,
|
| 301 |
+
"token_acc": 0.7640632197532837
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 0.2976190476190476,
|
| 305 |
+
"grad_norm": 2.2489399909973145,
|
| 306 |
+
"learning_rate": 9.924220826610624e-06,
|
| 307 |
+
"loss": 0.9082237243652344,
|
| 308 |
+
"step": 2400,
|
| 309 |
+
"token_acc": 0.764702263347946
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.2976190476190476,
|
| 313 |
+
"eval_loss": 1.3882246017456055,
|
| 314 |
+
"eval_runtime": 25.1006,
|
| 315 |
+
"eval_samples_per_second": 243.341,
|
| 316 |
+
"eval_steps_per_second": 0.478,
|
| 317 |
+
"eval_token_acc": 0.7041118163297289,
|
| 318 |
+
"step": 2400
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.3100198412698413,
|
| 322 |
+
"grad_norm": 2.5372934341430664,
|
| 323 |
+
"learning_rate": 9.915358939422189e-06,
|
| 324 |
+
"loss": 0.9065895080566406,
|
| 325 |
+
"step": 2500,
|
| 326 |
+
"token_acc": 0.764821800757079
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.32242063492063494,
|
| 330 |
+
"grad_norm": 2.344919204711914,
|
| 331 |
+
"learning_rate": 9.906011537497403e-06,
|
| 332 |
+
"loss": 0.9020857238769531,
|
| 333 |
+
"step": 2600,
|
| 334 |
+
"token_acc": 0.7658112305078509
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.32242063492063494,
|
| 338 |
+
"eval_loss": 1.4027491807937622,
|
| 339 |
+
"eval_runtime": 25.006,
|
| 340 |
+
"eval_samples_per_second": 244.262,
|
| 341 |
+
"eval_steps_per_second": 0.48,
|
| 342 |
+
"eval_token_acc": 0.7010558689860523,
|
| 343 |
+
"step": 2600
|
| 344 |
+
},
|
| 345 |
+
{
|
| 346 |
+
"epoch": 0.33482142857142855,
|
| 347 |
+
"grad_norm": 2.226698875427246,
|
| 348 |
+
"learning_rate": 9.896179544126187e-06,
|
| 349 |
+
"loss": 0.8991426086425781,
|
| 350 |
+
"step": 2700,
|
| 351 |
+
"token_acc": 0.7662424935116736
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.3472222222222222,
|
| 355 |
+
"grad_norm": 2.173095464706421,
|
| 356 |
+
"learning_rate": 9.885863930464004e-06,
|
| 357 |
+
"loss": 0.8963301086425781,
|
| 358 |
+
"step": 2800,
|
| 359 |
+
"token_acc": 0.7663184440570302
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.3472222222222222,
|
| 363 |
+
"eval_loss": 1.397993564605713,
|
| 364 |
+
"eval_runtime": 24.5711,
|
| 365 |
+
"eval_samples_per_second": 248.585,
|
| 366 |
+
"eval_steps_per_second": 0.488,
|
| 367 |
+
"eval_token_acc": 0.7034947500391788,
|
| 368 |
+
"step": 2800
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 0.35962301587301587,
|
| 372 |
+
"grad_norm": 2.2688558101654053,
|
| 373 |
+
"learning_rate": 9.875065715435922e-06,
|
| 374 |
+
"loss": 0.8933332824707031,
|
| 375 |
+
"step": 2900,
|
| 376 |
+
"token_acc": 0.7673069158571789
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 0.37202380952380953,
|
| 380 |
+
"grad_norm": 2.2284538745880127,
|
| 381 |
+
"learning_rate": 9.863785965635975e-06,
|
| 382 |
+
"loss": 0.8895596313476563,
|
| 383 |
+
"step": 3000,
|
| 384 |
+
"token_acc": 0.7682075802810432
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.37202380952380953,
|
| 388 |
+
"eval_loss": 1.3799571990966797,
|
| 389 |
+
"eval_runtime": 23.8977,
|
| 390 |
+
"eval_samples_per_second": 255.59,
|
| 391 |
+
"eval_steps_per_second": 0.502,
|
| 392 |
+
"eval_token_acc": 0.7064331609465601,
|
| 393 |
+
"step": 3000
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 0.3844246031746032,
|
| 397 |
+
"grad_norm": 2.168396234512329,
|
| 398 |
+
"learning_rate": 9.852025795221812e-06,
|
| 399 |
+
"loss": 0.8900260162353516,
|
| 400 |
+
"step": 3100,
|
| 401 |
+
"token_acc": 0.7679616096142278
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.3968253968253968,
|
| 405 |
+
"grad_norm": 2.1917724609375,
|
| 406 |
+
"learning_rate": 9.839786365804638e-06,
|
| 407 |
+
"loss": 0.8886898803710938,
|
| 408 |
+
"step": 3200,
|
| 409 |
+
"token_acc": 0.7681047882215793
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.3968253968253968,
|
| 413 |
+
"eval_loss": 1.3862619400024414,
|
| 414 |
+
"eval_runtime": 25.4948,
|
| 415 |
+
"eval_samples_per_second": 239.578,
|
| 416 |
+
"eval_steps_per_second": 0.471,
|
| 417 |
+
"eval_token_acc": 0.7051402601473123,
|
| 418 |
+
"step": 3200
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 0.40922619047619047,
|
| 422 |
+
"grad_norm": 2.1136505603790283,
|
| 423 |
+
"learning_rate": 9.827068886334487e-06,
|
| 424 |
+
"loss": 0.8903506469726562,
|
| 425 |
+
"step": 3300,
|
| 426 |
+
"token_acc": 0.767734332553098
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 0.42162698412698413,
|
| 430 |
+
"grad_norm": 2.048847198486328,
|
| 431 |
+
"learning_rate": 9.813874612980801e-06,
|
| 432 |
+
"loss": 0.8882189178466797,
|
| 433 |
+
"step": 3400,
|
| 434 |
+
"token_acc": 0.7680084529088362
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.42162698412698413,
|
| 438 |
+
"eval_loss": 1.3972296714782715,
|
| 439 |
+
"eval_runtime": 27.2253,
|
| 440 |
+
"eval_samples_per_second": 224.35,
|
| 441 |
+
"eval_steps_per_second": 0.441,
|
| 442 |
+
"eval_token_acc": 0.7047582667293527,
|
| 443 |
+
"step": 3400
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.4340277777777778,
|
| 447 |
+
"grad_norm": 2.2608675956726074,
|
| 448 |
+
"learning_rate": 9.800204849008352e-06,
|
| 449 |
+
"loss": 0.8840733337402343,
|
| 450 |
+
"step": 3500,
|
| 451 |
+
"token_acc": 0.7690944021413577
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.44642857142857145,
|
| 455 |
+
"grad_norm": 2.2813446521759033,
|
| 456 |
+
"learning_rate": 9.786060944648512e-06,
|
| 457 |
+
"loss": 0.881767349243164,
|
| 458 |
+
"step": 3600,
|
| 459 |
+
"token_acc": 0.769251916262417
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.44642857142857145,
|
| 463 |
+
"eval_loss": 1.381395936012268,
|
| 464 |
+
"eval_runtime": 32.9703,
|
| 465 |
+
"eval_samples_per_second": 185.257,
|
| 466 |
+
"eval_steps_per_second": 0.364,
|
| 467 |
+
"eval_token_acc": 0.7052871806926814,
|
| 468 |
+
"step": 3600
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 0.45882936507936506,
|
| 472 |
+
"grad_norm": 2.025299549102783,
|
| 473 |
+
"learning_rate": 9.771444296965887e-06,
|
| 474 |
+
"loss": 0.880907211303711,
|
| 475 |
+
"step": 3700,
|
| 476 |
+
"token_acc": 0.7696955947529188
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.4712301587301587,
|
| 480 |
+
"grad_norm": 2.236598014831543,
|
| 481 |
+
"learning_rate": 9.756356349720317e-06,
|
| 482 |
+
"loss": 0.8791775512695312,
|
| 483 |
+
"step": 3800,
|
| 484 |
+
"token_acc": 0.769757190376664
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 0.4712301587301587,
|
| 488 |
+
"eval_loss": 1.3754273653030396,
|
| 489 |
+
"eval_runtime": 30.7425,
|
| 490 |
+
"eval_samples_per_second": 198.683,
|
| 491 |
+
"eval_steps_per_second": 0.39,
|
| 492 |
+
"eval_token_acc": 0.7062764456981664,
|
| 493 |
+
"step": 3800
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.4836309523809524,
|
| 497 |
+
"grad_norm": 2.035534620285034,
|
| 498 |
+
"learning_rate": 9.74079859322428e-06,
|
| 499 |
+
"loss": 0.8722021484375,
|
| 500 |
+
"step": 3900,
|
| 501 |
+
"token_acc": 0.7713786650706949
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.49603174603174605,
|
| 505 |
+
"grad_norm": 2.0345141887664795,
|
| 506 |
+
"learning_rate": 9.724772564195662e-06,
|
| 507 |
+
"loss": 0.8705707550048828,
|
| 508 |
+
"step": 4000,
|
| 509 |
+
"token_acc": 0.7715948067593598
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 0.49603174603174605,
|
| 513 |
+
"eval_loss": 1.3851114511489868,
|
| 514 |
+
"eval_runtime": 24.9976,
|
| 515 |
+
"eval_samples_per_second": 244.344,
|
| 516 |
+
"eval_steps_per_second": 0.48,
|
| 517 |
+
"eval_token_acc": 0.7047190879172543,
|
| 518 |
+
"step": 4000
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 0.5084325396825397,
|
| 522 |
+
"grad_norm": 2.3718087673187256,
|
| 523 |
+
"learning_rate": 9.708279845605997e-06,
|
| 524 |
+
"loss": 0.8708313751220703,
|
| 525 |
+
"step": 4100,
|
| 526 |
+
"token_acc": 0.771728452566088
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 0.5208333333333334,
|
| 530 |
+
"grad_norm": 1.9914731979370117,
|
| 531 |
+
"learning_rate": 9.691322066524094e-06,
|
| 532 |
+
"loss": 0.8680215454101563,
|
| 533 |
+
"step": 4200,
|
| 534 |
+
"token_acc": 0.7722721985160954
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.5208333333333334,
|
| 538 |
+
"eval_loss": 1.3857182264328003,
|
| 539 |
+
"eval_runtime": 32.0439,
|
| 540 |
+
"eval_samples_per_second": 190.613,
|
| 541 |
+
"eval_steps_per_second": 0.374,
|
| 542 |
+
"eval_token_acc": 0.7040040745964582,
|
| 543 |
+
"step": 4200
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 0.533234126984127,
|
| 547 |
+
"grad_norm": 2.053499937057495,
|
| 548 |
+
"learning_rate": 9.673900901955118e-06,
|
| 549 |
+
"loss": 0.865277328491211,
|
| 550 |
+
"step": 4300,
|
| 551 |
+
"token_acc": 0.772273872119687
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 0.5456349206349206,
|
| 555 |
+
"grad_norm": 2.0330519676208496,
|
| 556 |
+
"learning_rate": 9.65601807267516e-06,
|
| 557 |
+
"loss": 0.8685321044921875,
|
| 558 |
+
"step": 4400,
|
| 559 |
+
"token_acc": 0.7712716260970404
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 0.5456349206349206,
|
| 563 |
+
"eval_loss": 1.3824304342269897,
|
| 564 |
+
"eval_runtime": 31.706,
|
| 565 |
+
"eval_samples_per_second": 192.645,
|
| 566 |
+
"eval_steps_per_second": 0.378,
|
| 567 |
+
"eval_token_acc": 0.7058258893590347,
|
| 568 |
+
"step": 4400
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 0.5580357142857143,
|
| 572 |
+
"grad_norm": 2.3541200160980225,
|
| 573 |
+
"learning_rate": 9.637675345061252e-06,
|
| 574 |
+
"loss": 0.8656570434570312,
|
| 575 |
+
"step": 4500,
|
| 576 |
+
"token_acc": 0.7719719290205852
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.5704365079365079,
|
| 580 |
+
"grad_norm": 2.003859758377075,
|
| 581 |
+
"learning_rate": 9.618874530916903e-06,
|
| 582 |
+
"loss": 0.8663369750976563,
|
| 583 |
+
"step": 4600,
|
| 584 |
+
"token_acc": 0.7725054467172254
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.5704365079365079,
|
| 588 |
+
"eval_loss": 1.3676419258117676,
|
| 589 |
+
"eval_runtime": 25.4735,
|
| 590 |
+
"eval_samples_per_second": 239.779,
|
| 591 |
+
"eval_steps_per_second": 0.471,
|
| 592 |
+
"eval_token_acc": 0.706286240401191,
|
| 593 |
+
"step": 4600
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 0.5828373015873016,
|
| 597 |
+
"grad_norm": 2.039274215698242,
|
| 598 |
+
"learning_rate": 9.599617487293127e-06,
|
| 599 |
+
"loss": 0.8659154510498047,
|
| 600 |
+
"step": 4700,
|
| 601 |
+
"token_acc": 0.7722415641221803
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 0.5952380952380952,
|
| 605 |
+
"grad_norm": 2.0263097286224365,
|
| 606 |
+
"learning_rate": 9.579906116305026e-06,
|
| 607 |
+
"loss": 0.8560360717773438,
|
| 608 |
+
"step": 4800,
|
| 609 |
+
"token_acc": 0.7745140899539719
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.5952380952380952,
|
| 613 |
+
"eval_loss": 1.4032906293869019,
|
| 614 |
+
"eval_runtime": 27.5699,
|
| 615 |
+
"eval_samples_per_second": 221.546,
|
| 616 |
+
"eval_steps_per_second": 0.435,
|
| 617 |
+
"eval_token_acc": 0.7039159222692368,
|
| 618 |
+
"step": 4800
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.6076388888888888,
|
| 622 |
+
"grad_norm": 2.1180593967437744,
|
| 623 |
+
"learning_rate": 9.5597423649439e-06,
|
| 624 |
+
"loss": 0.8591609191894531,
|
| 625 |
+
"step": 4900,
|
| 626 |
+
"token_acc": 0.7735983802448648
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.6200396825396826,
|
| 630 |
+
"grad_norm": 1.867364764213562,
|
| 631 |
+
"learning_rate": 9.539128224884931e-06,
|
| 632 |
+
"loss": 0.8610757446289062,
|
| 633 |
+
"step": 5000,
|
| 634 |
+
"token_acc": 0.7728954345163985
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 0.6200396825396826,
|
| 638 |
+
"eval_loss": 1.3865896463394165,
|
| 639 |
+
"eval_runtime": 26.774,
|
| 640 |
+
"eval_samples_per_second": 228.132,
|
| 641 |
+
"eval_steps_per_second": 0.448,
|
| 642 |
+
"eval_token_acc": 0.7055810217834195,
|
| 643 |
+
"step": 5000
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 0.6324404761904762,
|
| 647 |
+
"grad_norm": 1.8828492164611816,
|
| 648 |
+
"learning_rate": 9.518065732290466e-06,
|
| 649 |
+
"loss": 0.8573006439208984,
|
| 650 |
+
"step": 5100,
|
| 651 |
+
"token_acc": 0.7741265516001206
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 0.6448412698412699,
|
| 655 |
+
"grad_norm": 2.0104453563690186,
|
| 656 |
+
"learning_rate": 9.496556967608881e-06,
|
| 657 |
+
"loss": 0.8586259460449219,
|
| 658 |
+
"step": 5200,
|
| 659 |
+
"token_acc": 0.7738154668886862
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 0.6448412698412699,
|
| 663 |
+
"eval_loss": 1.3962821960449219,
|
| 664 |
+
"eval_runtime": 26.8433,
|
| 665 |
+
"eval_samples_per_second": 227.543,
|
| 666 |
+
"eval_steps_per_second": 0.447,
|
| 667 |
+
"eval_token_acc": 0.7046211408870083,
|
| 668 |
+
"step": 5200
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.6572420634920635,
|
| 672 |
+
"grad_norm": 1.9568499326705933,
|
| 673 |
+
"learning_rate": 9.474604055369097e-06,
|
| 674 |
+
"loss": 0.853517074584961,
|
| 675 |
+
"step": 5300,
|
| 676 |
+
"token_acc": 0.7747399228013471
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 0.6696428571428571,
|
| 680 |
+
"grad_norm": 2.138674020767212,
|
| 681 |
+
"learning_rate": 9.452209163970722e-06,
|
| 682 |
+
"loss": 0.8532652282714843,
|
| 683 |
+
"step": 5400,
|
| 684 |
+
"token_acc": 0.774544167587472
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 0.6696428571428571,
|
| 688 |
+
"eval_loss": 1.386710524559021,
|
| 689 |
+
"eval_runtime": 33.5684,
|
| 690 |
+
"eval_samples_per_second": 181.957,
|
| 691 |
+
"eval_steps_per_second": 0.357,
|
| 692 |
+
"eval_token_acc": 0.7059336310923053,
|
| 693 |
+
"step": 5400
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 0.6820436507936508,
|
| 697 |
+
"grad_norm": 2.0416526794433594,
|
| 698 |
+
"learning_rate": 9.429374505469868e-06,
|
| 699 |
+
"loss": 0.8509379577636719,
|
| 700 |
+
"step": 5500,
|
| 701 |
+
"token_acc": 0.7745901639344263
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 0.6944444444444444,
|
| 705 |
+
"grad_norm": 1.8739123344421387,
|
| 706 |
+
"learning_rate": 9.40610233536066e-06,
|
| 707 |
+
"loss": 0.8514627075195312,
|
| 708 |
+
"step": 5600,
|
| 709 |
+
"token_acc": 0.7743979174229727
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.6944444444444444,
|
| 713 |
+
"eval_loss": 1.3842697143554688,
|
| 714 |
+
"eval_runtime": 24.9442,
|
| 715 |
+
"eval_samples_per_second": 244.866,
|
| 716 |
+
"eval_steps_per_second": 0.481,
|
| 717 |
+
"eval_token_acc": 0.7070502272371102,
|
| 718 |
+
"step": 5600
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 0.7068452380952381,
|
| 722 |
+
"grad_norm": 2.0142526626586914,
|
| 723 |
+
"learning_rate": 9.382394952352447e-06,
|
| 724 |
+
"loss": 0.8503968811035156,
|
| 725 |
+
"step": 5700,
|
| 726 |
+
"token_acc": 0.775558438423616
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 0.7192460317460317,
|
| 730 |
+
"grad_norm": 2.1041524410247803,
|
| 731 |
+
"learning_rate": 9.358254698142737e-06,
|
| 732 |
+
"loss": 0.8513346862792969,
|
| 733 |
+
"step": 5800,
|
| 734 |
+
"token_acc": 0.774860838501366
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 0.7192460317460317,
|
| 738 |
+
"eval_loss": 1.4040964841842651,
|
| 739 |
+
"eval_runtime": 28.2843,
|
| 740 |
+
"eval_samples_per_second": 215.95,
|
| 741 |
+
"eval_steps_per_second": 0.424,
|
| 742 |
+
"eval_token_acc": 0.7047680614323774,
|
| 743 |
+
"step": 5800
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 0.7316468253968254,
|
| 747 |
+
"grad_norm": 1.9587483406066895,
|
| 748 |
+
"learning_rate": 9.333683957185915e-06,
|
| 749 |
+
"loss": 0.8467042541503906,
|
| 750 |
+
"step": 5900,
|
| 751 |
+
"token_acc": 0.7760485862743774
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.7440476190476191,
|
| 755 |
+
"grad_norm": 1.8701276779174805,
|
| 756 |
+
"learning_rate": 9.308685156457704e-06,
|
| 757 |
+
"loss": 0.8471446228027344,
|
| 758 |
+
"step": 6000,
|
| 759 |
+
"token_acc": 0.7759005391747791
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.7440476190476191,
|
| 763 |
+
"eval_loss": 1.3969898223876953,
|
| 764 |
+
"eval_runtime": 26.1996,
|
| 765 |
+
"eval_samples_per_second": 233.133,
|
| 766 |
+
"eval_steps_per_second": 0.458,
|
| 767 |
+
"eval_token_acc": 0.7054634853471243,
|
| 768 |
+
"step": 6000
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 0.7564484126984127,
|
| 772 |
+
"grad_norm": 1.9555493593215942,
|
| 773 |
+
"learning_rate": 9.283260765215445e-06,
|
| 774 |
+
"loss": 0.8483773803710938,
|
| 775 |
+
"step": 6100,
|
| 776 |
+
"token_acc": 0.7752799704921876
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 0.7688492063492064,
|
| 780 |
+
"grad_norm": 1.9491664171218872,
|
| 781 |
+
"learning_rate": 9.257413294754195e-06,
|
| 782 |
+
"loss": 0.8444963836669922,
|
| 783 |
+
"step": 6200,
|
| 784 |
+
"token_acc": 0.7765536675671234
|
| 785 |
+
},
|
| 786 |
+
{
|
| 787 |
+
"epoch": 0.7688492063492064,
|
| 788 |
+
"eval_loss": 1.4008746147155762,
|
| 789 |
+
"eval_runtime": 25.5894,
|
| 790 |
+
"eval_samples_per_second": 238.693,
|
| 791 |
+
"eval_steps_per_second": 0.469,
|
| 792 |
+
"eval_token_acc": 0.7053851277229275,
|
| 793 |
+
"step": 6200
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.78125,
|
| 797 |
+
"grad_norm": 1.9883235692977905,
|
| 798 |
+
"learning_rate": 9.231145298158676e-06,
|
| 799 |
+
"loss": 0.8457432556152343,
|
| 800 |
+
"step": 6300,
|
| 801 |
+
"token_acc": 0.7757105248526481
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.7936507936507936,
|
| 805 |
+
"grad_norm": 1.8831249475479126,
|
| 806 |
+
"learning_rate": 9.20445937005109e-06,
|
| 807 |
+
"loss": 0.8422145080566407,
|
| 808 |
+
"step": 6400,
|
| 809 |
+
"token_acc": 0.7769141480983273
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 0.7936507936507936,
|
| 813 |
+
"eval_loss": 1.3781484365463257,
|
| 814 |
+
"eval_runtime": 29.054,
|
| 815 |
+
"eval_samples_per_second": 210.229,
|
| 816 |
+
"eval_steps_per_second": 0.413,
|
| 817 |
+
"eval_token_acc": 0.7087936843754897,
|
| 818 |
+
"step": 6400
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 0.8060515873015873,
|
| 822 |
+
"grad_norm": 1.9456390142440796,
|
| 823 |
+
"learning_rate": 9.17735814633484e-06,
|
| 824 |
+
"loss": 0.8456352233886719,
|
| 825 |
+
"step": 6500,
|
| 826 |
+
"token_acc": 0.7758258891652673
|
| 827 |
+
},
|
| 828 |
+
{
|
| 829 |
+
"epoch": 0.8184523809523809,
|
| 830 |
+
"grad_norm": 2.3600473403930664,
|
| 831 |
+
"learning_rate": 9.149844303934165e-06,
|
| 832 |
+
"loss": 0.8402180480957031,
|
| 833 |
+
"step": 6600,
|
| 834 |
+
"token_acc": 0.7774200791566007
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 0.8184523809523809,
|
| 838 |
+
"eval_loss": 1.3933624029159546,
|
| 839 |
+
"eval_runtime": 24.7423,
|
| 840 |
+
"eval_samples_per_second": 246.864,
|
| 841 |
+
"eval_steps_per_second": 0.485,
|
| 842 |
+
"eval_token_acc": 0.7064625450556339,
|
| 843 |
+
"step": 6600
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.8308531746031746,
|
| 847 |
+
"grad_norm": 1.971165418624878,
|
| 848 |
+
"learning_rate": 9.121920560529728e-06,
|
| 849 |
+
"loss": 0.8413809204101562,
|
| 850 |
+
"step": 6700,
|
| 851 |
+
"token_acc": 0.7766619100185779
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 0.8432539682539683,
|
| 855 |
+
"grad_norm": 1.8783116340637207,
|
| 856 |
+
"learning_rate": 9.093589674290172e-06,
|
| 857 |
+
"loss": 0.8382947540283203,
|
| 858 |
+
"step": 6800,
|
| 859 |
+
"token_acc": 0.7778030413909798
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 0.8432539682539683,
|
| 863 |
+
"eval_loss": 1.386560082435608,
|
| 864 |
+
"eval_runtime": 26.9878,
|
| 865 |
+
"eval_samples_per_second": 226.325,
|
| 866 |
+
"eval_steps_per_second": 0.445,
|
| 867 |
+
"eval_token_acc": 0.7069326908008149,
|
| 868 |
+
"step": 6800
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 0.8556547619047619,
|
| 872 |
+
"grad_norm": 2.3792569637298584,
|
| 873 |
+
"learning_rate": 9.064854443599695e-06,
|
| 874 |
+
"loss": 0.8386391448974609,
|
| 875 |
+
"step": 6900,
|
| 876 |
+
"token_acc": 0.7773819755208488
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 0.8680555555555556,
|
| 880 |
+
"grad_norm": 1.9603108167648315,
|
| 881 |
+
"learning_rate": 9.035717706781625e-06,
|
| 882 |
+
"loss": 0.8399604797363281,
|
| 883 |
+
"step": 7000,
|
| 884 |
+
"token_acc": 0.7771941484340188
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.8680555555555556,
|
| 888 |
+
"eval_loss": 1.3696708679199219,
|
| 889 |
+
"eval_runtime": 30.6282,
|
| 890 |
+
"eval_samples_per_second": 199.424,
|
| 891 |
+
"eval_steps_per_second": 0.392,
|
| 892 |
+
"eval_token_acc": 0.7087447108603667,
|
| 893 |
+
"step": 7000
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 0.8804563492063492,
|
| 897 |
+
"grad_norm": 1.9724894762039185,
|
| 898 |
+
"learning_rate": 9.006182341818077e-06,
|
| 899 |
+
"loss": 0.8373133850097656,
|
| 900 |
+
"step": 7100,
|
| 901 |
+
"token_acc": 0.7777272727272727
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 0.8928571428571429,
|
| 905 |
+
"grad_norm": 1.9323441982269287,
|
| 906 |
+
"learning_rate": 8.976251266065663e-06,
|
| 907 |
+
"loss": 0.8318426513671875,
|
| 908 |
+
"step": 7200,
|
| 909 |
+
"token_acc": 0.7787919084025631
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 0.8928571428571429,
|
| 913 |
+
"eval_loss": 1.376365303993225,
|
| 914 |
+
"eval_runtime": 28.3952,
|
| 915 |
+
"eval_samples_per_second": 215.106,
|
| 916 |
+
"eval_steps_per_second": 0.423,
|
| 917 |
+
"eval_token_acc": 0.705963015201379,
|
| 918 |
+
"step": 7200
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 0.9052579365079365,
|
| 922 |
+
"grad_norm": 1.8648154735565186,
|
| 923 |
+
"learning_rate": 8.94592743596735e-06,
|
| 924 |
+
"loss": 0.8345167541503906,
|
| 925 |
+
"step": 7300,
|
| 926 |
+
"token_acc": 0.7785041388148097
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.9176587301587301,
|
| 930 |
+
"grad_norm": 1.8587236404418945,
|
| 931 |
+
"learning_rate": 8.915213846760426e-06,
|
| 932 |
+
"loss": 0.8353372192382813,
|
| 933 |
+
"step": 7400,
|
| 934 |
+
"token_acc": 0.7781199706042472
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.9176587301587301,
|
| 938 |
+
"eval_loss": 1.3952875137329102,
|
| 939 |
+
"eval_runtime": 29.9749,
|
| 940 |
+
"eval_samples_per_second": 203.771,
|
| 941 |
+
"eval_steps_per_second": 0.4,
|
| 942 |
+
"eval_token_acc": 0.7072461212976022,
|
| 943 |
+
"step": 7400
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 0.9300595238095238,
|
| 947 |
+
"grad_norm": 1.857714295387268,
|
| 948 |
+
"learning_rate": 8.884113532180646e-06,
|
| 949 |
+
"loss": 0.8350875091552734,
|
| 950 |
+
"step": 7500,
|
| 951 |
+
"token_acc": 0.7779484917251972
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 0.9424603174603174,
|
| 955 |
+
"grad_norm": 2.0893328189849854,
|
| 956 |
+
"learning_rate": 8.852629564162576e-06,
|
| 957 |
+
"loss": 0.8318548583984375,
|
| 958 |
+
"step": 7600,
|
| 959 |
+
"token_acc": 0.778963162963007
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 0.9424603174603174,
|
| 963 |
+
"eval_loss": 1.3874316215515137,
|
| 964 |
+
"eval_runtime": 38.5176,
|
| 965 |
+
"eval_samples_per_second": 158.577,
|
| 966 |
+
"eval_steps_per_second": 0.312,
|
| 967 |
+
"eval_token_acc": 0.7082647704121611,
|
| 968 |
+
"step": 7600
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 0.9548611111111112,
|
| 972 |
+
"grad_norm": 1.9114059209823608,
|
| 973 |
+
"learning_rate": 8.820765052536164e-06,
|
| 974 |
+
"loss": 0.8320674133300782,
|
| 975 |
+
"step": 7700,
|
| 976 |
+
"token_acc": 0.7786642124240083
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 0.9672619047619048,
|
| 980 |
+
"grad_norm": 1.8616052865982056,
|
| 981 |
+
"learning_rate": 8.788523144719564e-06,
|
| 982 |
+
"loss": 0.8302680969238281,
|
| 983 |
+
"step": 7800,
|
| 984 |
+
"token_acc": 0.7786792001119752
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 0.9672619047619048,
|
| 988 |
+
"eval_loss": 1.3864374160766602,
|
| 989 |
+
"eval_runtime": 25.4343,
|
| 990 |
+
"eval_samples_per_second": 240.148,
|
| 991 |
+
"eval_steps_per_second": 0.472,
|
| 992 |
+
"eval_token_acc": 0.7089014261087604,
|
| 993 |
+
"step": 7800
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 0.9796626984126984,
|
| 997 |
+
"grad_norm": 1.8287373781204224,
|
| 998 |
+
"learning_rate": 8.75590702540825e-06,
|
| 999 |
+
"loss": 0.829267578125,
|
| 1000 |
+
"step": 7900,
|
| 1001 |
+
"token_acc": 0.779347232586745
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 0.9920634920634921,
|
| 1005 |
+
"grad_norm": 1.8730870485305786,
|
| 1006 |
+
"learning_rate": 8.722919916260453e-06,
|
| 1007 |
+
"loss": 0.8275999450683593,
|
| 1008 |
+
"step": 8000,
|
| 1009 |
+
"token_acc": 0.7796970679121732
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 0.9920634920634921,
|
| 1013 |
+
"eval_loss": 1.3773517608642578,
|
| 1014 |
+
"eval_runtime": 27.6037,
|
| 1015 |
+
"eval_samples_per_second": 221.275,
|
| 1016 |
+
"eval_steps_per_second": 0.435,
|
| 1017 |
+
"eval_token_acc": 0.7074811941701927,
|
| 1018 |
+
"step": 8000
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 1.0044642857142858,
|
| 1022 |
+
"grad_norm": 1.95199716091156,
|
| 1023 |
+
"learning_rate": 8.68956507557893e-06,
|
| 1024 |
+
"loss": 0.8077944946289063,
|
| 1025 |
+
"step": 8100,
|
| 1026 |
+
"token_acc": 0.7830445546557503
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 1.0168650793650793,
|
| 1030 |
+
"grad_norm": 2.108023166656494,
|
| 1031 |
+
"learning_rate": 8.655845797989136e-06,
|
| 1032 |
+
"loss": 0.7570481872558594,
|
| 1033 |
+
"step": 8200,
|
| 1034 |
+
"token_acc": 0.7930948720023738
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 1.0168650793650793,
|
| 1038 |
+
"eval_loss": 1.4068571329116821,
|
| 1039 |
+
"eval_runtime": 27.1125,
|
| 1040 |
+
"eval_samples_per_second": 225.283,
|
| 1041 |
+
"eval_steps_per_second": 0.443,
|
| 1042 |
+
"eval_token_acc": 0.7084802538787024,
|
| 1043 |
+
"step": 8200
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 1.029265873015873,
|
| 1047 |
+
"grad_norm": 1.8741405010223389,
|
| 1048 |
+
"learning_rate": 8.621765414113793e-06,
|
| 1049 |
+
"loss": 0.7590464782714844,
|
| 1050 |
+
"step": 8300,
|
| 1051 |
+
"token_acc": 0.7925717890005856
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 1.0416666666666667,
|
| 1055 |
+
"grad_norm": 2.1017026901245117,
|
| 1056 |
+
"learning_rate": 8.58732729024391e-06,
|
| 1057 |
+
"loss": 0.7580317687988282,
|
| 1058 |
+
"step": 8400,
|
| 1059 |
+
"token_acc": 0.7935637882304593
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 1.0416666666666667,
|
| 1063 |
+
"eval_loss": 1.4264146089553833,
|
| 1064 |
+
"eval_runtime": 26.4293,
|
| 1065 |
+
"eval_samples_per_second": 231.107,
|
| 1066 |
+
"eval_steps_per_second": 0.454,
|
| 1067 |
+
"eval_token_acc": 0.7066976179282244,
|
| 1068 |
+
"step": 8400
|
| 1069 |
+
},
|
| 1070 |
+
{
|
| 1071 |
+
"epoch": 1.0540674603174602,
|
| 1072 |
+
"grad_norm": 1.8574824333190918,
|
| 1073 |
+
"learning_rate": 8.552534828006271e-06,
|
| 1074 |
+
"loss": 0.7585566711425781,
|
| 1075 |
+
"step": 8500,
|
| 1076 |
+
"token_acc": 0.7926697473555268
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 1.066468253968254,
|
| 1080 |
+
"grad_norm": 1.8918617963790894,
|
| 1081 |
+
"learning_rate": 8.517391464027445e-06,
|
| 1082 |
+
"loss": 0.7611293792724609,
|
| 1083 |
+
"step": 8600,
|
| 1084 |
+
"token_acc": 0.7925391843068593
|
| 1085 |
+
},
|
| 1086 |
+
{
|
| 1087 |
+
"epoch": 1.066468253968254,
|
| 1088 |
+
"eval_loss": 1.4304345846176147,
|
| 1089 |
+
"eval_runtime": 27.3801,
|
| 1090 |
+
"eval_samples_per_second": 223.082,
|
| 1091 |
+
"eval_steps_per_second": 0.438,
|
| 1092 |
+
"eval_token_acc": 0.7059826046074283,
|
| 1093 |
+
"step": 8600
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"epoch": 1.0788690476190477,
|
| 1097 |
+
"grad_norm": 2.2014002799987793,
|
| 1098 |
+
"learning_rate": 8.481900669594332e-06,
|
| 1099 |
+
"loss": 0.759523696899414,
|
| 1100 |
+
"step": 8700,
|
| 1101 |
+
"token_acc": 0.7928630607620207
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 1.0912698412698412,
|
| 1105 |
+
"grad_norm": 2.0527760982513428,
|
| 1106 |
+
"learning_rate": 8.446065950311288e-06,
|
| 1107 |
+
"loss": 0.7538475036621094,
|
| 1108 |
+
"step": 8800,
|
| 1109 |
+
"token_acc": 0.7939912058277155
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 1.0912698412698412,
|
| 1113 |
+
"eval_loss": 1.4411059617996216,
|
| 1114 |
+
"eval_runtime": 25.3283,
|
| 1115 |
+
"eval_samples_per_second": 241.154,
|
| 1116 |
+
"eval_steps_per_second": 0.474,
|
| 1117 |
+
"eval_token_acc": 0.7076379094185864,
|
| 1118 |
+
"step": 8800
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 1.1036706349206349,
|
| 1122 |
+
"grad_norm": 1.983169436454773,
|
| 1123 |
+
"learning_rate": 8.409890845753855e-06,
|
| 1124 |
+
"loss": 0.7595367431640625,
|
| 1125 |
+
"step": 8900,
|
| 1126 |
+
"token_acc": 0.7931083876957342
|
| 1127 |
+
},
|
| 1128 |
+
{
|
| 1129 |
+
"epoch": 1.1160714285714286,
|
| 1130 |
+
"grad_norm": 1.9933942556381226,
|
| 1131 |
+
"learning_rate": 8.37337892911914e-06,
|
| 1132 |
+
"loss": 0.7552370452880859,
|
| 1133 |
+
"step": 9000,
|
| 1134 |
+
"token_acc": 0.7937327435508952
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 1.1160714285714286,
|
| 1138 |
+
"eval_loss": 1.4212846755981445,
|
| 1139 |
+
"eval_runtime": 25.1583,
|
| 1140 |
+
"eval_samples_per_second": 242.783,
|
| 1141 |
+
"eval_steps_per_second": 0.477,
|
| 1142 |
+
"eval_token_acc": 0.7078729822911769,
|
| 1143 |
+
"step": 9000
|
| 1144 |
+
}
|
| 1145 |
+
],
|
| 1146 |
+
"logging_steps": 100,
|
| 1147 |
+
"max_steps": 32256,
|
| 1148 |
+
"num_input_tokens_seen": 0,
|
| 1149 |
+
"num_train_epochs": 4,
|
| 1150 |
+
"save_steps": 1000,
|
| 1151 |
+
"stateful_callbacks": {
|
| 1152 |
+
"TrainerControl": {
|
| 1153 |
+
"args": {
|
| 1154 |
+
"should_epoch_stop": false,
|
| 1155 |
+
"should_evaluate": false,
|
| 1156 |
+
"should_log": false,
|
| 1157 |
+
"should_save": true,
|
| 1158 |
+
"should_training_stop": false
|
| 1159 |
+
},
|
| 1160 |
+
"attributes": {}
|
| 1161 |
+
}
|
| 1162 |
+
},
|
| 1163 |
+
"total_flos": 9.797715664620749e+18,
|
| 1164 |
+
"train_batch_size": 64,
|
| 1165 |
+
"trial_name": null,
|
| 1166 |
+
"trial_params": null
|
| 1167 |
+
}
|
Qwen3-ASR-WSM-Mandarin/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6a890e177f0d5317ff8cbb8a0e56ef8fa45b10bd3ac1cb9475cecdea54ca61c0
|
| 3 |
+
size 9425
|
Qwen3-ASR-WSM-Mandarin/vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
Qwen3-ASR-WSM-Mandarin/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)
|