Hugging Face's logo

efrainmain
/
draft-chunkformer-finetune-encoder

Automatic Speech Recognition
PyTorch
Vietnamese
transcription
audio
speech
chunkformer
asr
Eval Results (legacy)
Model card Files
xet
 Community
draft-chunkformer-finetune-encoder
File size: 2,179 Bytes
093ba76
 
 
 
90b12e8
 
 
 
 
 
 
 
 
 
093ba76
 
 
 
90b12e8
 
 
 
 
093ba76
 
 
 
 
 
 
90b12e8
093ba76
 
 
 
 
90b12e8
093ba76
 
 
 
 
 
 
 
 
90b12e8
 
 
 
 
 
 
093ba76
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
{
  "resume_type": "chunkformer_finetune_resume",
  "base_repo": "khanhld/chunkformer-ctc-large-vie",
  "repo_id": "efrainmain/chunkformer-ctc-vie-medical",
  "epoch": 2,
  "trained_epochs": 2,
  "next_epoch": 3,
  "train_loss": 178.68146068524644,
  "val_loss": 155.45725427576014,
  "best_checkpoint_path": "/kaggle/working/chunkformer-ctc-large-vie-finetune/epoch_2_val_loss_155.4573.pt",
  "checkpoint_file_used_for_best_model": "/kaggle/working/chunkformer-ctc-large-vie-finetune/epoch_2_val_loss_155.4573.pt",
  "best_epoch": 2,
  "best_val_loss": 155.45725427576014,
  "best_train_loss": 178.68146068524644,
  "has_optimizer_state": true,
  "has_scheduler_state": true,
  "has_scaler_state": true,
  "config": {
    "base_model_name": "khanhld/chunkformer-ctc-large-vie",
    "resume_from_checkpoint": true,
    "resume_repo_id": "efrainmain/chunkformer-ctc-vie-medical",
    "resume_checkpoint_file": "resume_checkpoint/training_checkpoint.pt",
    "model_name": "efrainmain/chunkformer-ctc-vie-medical",
    "dataset_root": "/kaggle/input/datasets/lqucng/vietmed-vimedcss-dataset",
    "train_parquet": "/kaggle/input/datasets/lqucng/vietmed-vimedcss-dataset/train.parquet",
    "validation_parquet": "/kaggle/input/datasets/lqucng/vietmed-vimedcss-dataset/validation.parquet",
    "test_parquet": "/kaggle/input/datasets/lqucng/vietmed-vimedcss-dataset/test.parquet",
    "preprocess_root": "/kaggle/working/chunkformer_preprocessed",
    "output_dir": "/kaggle/working/chunkformer-ctc-large-vie-finetune",
    "batch_size": 8,
    "num_epochs": 3,
    "learning_rate": 1e-05,
    "weight_decay": 0.01,
    "scheduler_type": "warmup_cosine",
    "warmup_steps": 500,
    "min_learning_rate": 1e-07,
    "freeze_encoder": true,
    "freeze_encoder_epochs": 2,
    "patience": 3,
    "keep_best": 2,
    "max_grad_norm": 5.0,
    "num_workers": 0,
    "device": "cuda",
    "use_amp": true,
    "skip_bad_samples": true,
    "max_train_batches": null,
    "max_eval_batches": null,
    "resume_load_optimizer": true,
    "resume_load_scheduler": true,
    "resume_load_scaler": true,
    "strict_resume_model_load": false,
    "save_loaded_checkpoint_locally": true
  }
}