Hugging Face's logo

vumichien
/
wav2vec2-xls-r-1b-japanese

Automatic Speech Recognition
Transformers
PyTorch
TensorBoard
Safetensors
Japanese
wav2vec2
common-voice
hf-asr-leaderboard
robust-speech-event
Eval Results
Inference Endpoints
Model card Files Metrics Community
1
wav2vec2-xls-r-1b-japanese / README.md
vumichien's picture
vumichien
update model card README.md
ee586d1
preview code
|
raw
history blame
3.62 kB
metadata
language:
  - ja
license: apache-2.0
tags:
  - automatic-speech-recognition
  - vumichien/common_voice_large_jsut_jsss_css10
  - generated_from_trainer
model-index:
  - name: wav2vec2-xls-r-1b-ja-dumy8
    results: []

wav2vec2-xls-r-1b-ja-dumy8

This model is a fine-tuned version of facebook/wav2vec2-xls-r-1b on the VUMICHIEN/COMMON_VOICE_LARGE_JSUT_JSSS_CSS10 - JA dataset. It achieves the following results on the evaluation set:

  • Loss: 0.2104
  • Wer: 0.1941
  • Cer: 0.0991

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 5e-05
  • train_batch_size: 16
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 64
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 1000
  • num_epochs: 100.0
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer Cer
2.2896 3.37 1500 0.4748 0.4013 0.1767
1.1608 6.74 3000 0.3350 0.3159 0.1456
1.1042 10.11 4500 0.3119 0.2971 0.1400
1.0494 13.48 6000 0.2974 0.2867 0.1353
1.0061 16.85 7500 0.2802 0.2746 0.1300
0.9629 20.22 9000 0.2844 0.2776 0.1326
0.9267 23.59 10500 0.2577 0.2603 0.1255
0.8984 26.96 12000 0.2508 0.2531 0.1226
0.8729 30.34 13500 0.2629 0.2606 0.1254
0.8546 33.71 15000 0.2402 0.2447 0.1193
0.8304 37.08 16500 0.2532 0.2472 0.1209
0.8075 40.45 18000 0.2439 0.2469 0.1198
0.7827 43.82 19500 0.2387 0.2372 0.1167
0.7627 47.19 21000 0.2344 0.2331 0.1147
0.7402 50.56 22500 0.2314 0.2299 0.1135
0.718 53.93 24000 0.2257 0.2267 0.1114
0.7016 57.3 25500 0.2204 0.2184 0.1089
0.6804 60.67 27000 0.2227 0.2181 0.1085
0.6625 64.04 28500 0.2138 0.2112 0.1058
0.6465 67.42 30000 0.2141 0.2081 0.1044
0.6238 70.79 31500 0.2172 0.2082 0.1050
0.6062 74.16 33000 0.2174 0.2058 0.1043
0.588 77.53 34500 0.2156 0.2034 0.1027
0.5722 80.9 36000 0.2162 0.2032 0.1029
0.5585 84.27 37500 0.2156 0.2022 0.1021
0.5456 87.64 39000 0.2126 0.1993 0.1009
0.5325 91.01 40500 0.2121 0.1966 0.1003
0.5229 94.38 42000 0.2104 0.1941 0.0991
0.5134 97.75 43500 0.2108 0.1948 0.0992

Framework versions

  • Transformers 4.16.0.dev0
  • Pytorch 1.10.1+cu102
  • Datasets 1.17.1.dev0
  • Tokenizers 0.11.0