metadata
language:
- hi
license: apache-2.0
tags:
- automatic-speech-recognition
- mozilla-foundation/common_voice_7_0
- robust-speech-event
- generated_from_trainer
datasets:
- common_voice
model-index:
- name: ''
results: []
This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the MOZILLA-FOUNDATION/COMMON_VOICE_7_0 - HI dataset. It achieves the following results on the evaluation set:
- Loss: 0.9201
- Wer: 1.0091
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: 0.0003
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 200
- training_steps: 5000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| No log | 0.34 | 100 | 4.2570 | 0.9999 |
| No log | 0.68 | 200 | 3.3222 | 1.0005 |
| No log | 1.02 | 300 | 2.3755 | 1.0 |
| No log | 1.36 | 400 | 1.4071 | 1.0037 |
| 3.8379 | 1.69 | 500 | 1.2223 | 1.0091 |
| 3.8379 | 2.03 | 600 | 1.0885 | 1.0273 |
| 3.8379 | 2.37 | 700 | 0.9818 | 1.0098 |
| 3.8379 | 2.71 | 800 | 0.9154 | 1.0110 |
| 3.8379 | 3.05 | 900 | 0.8869 | 1.0154 |
| 0.5888 | 3.39 | 1000 | 0.8976 | 1.0149 |
| 0.5888 | 3.73 | 1100 | 0.8323 | 1.0124 |
| 0.5888 | 4.07 | 1200 | 0.8488 | 1.0087 |
| 0.5888 | 4.41 | 1300 | 0.8659 | 1.0069 |
| 0.5888 | 4.75 | 1400 | 0.8612 | 1.0043 |
| 0.3706 | 5.08 | 1500 | 0.8300 | 1.0183 |
| 0.3706 | 5.42 | 1600 | 0.8417 | 1.0091 |
| 0.3706 | 5.76 | 1700 | 0.8261 | 1.0087 |
| 0.3706 | 6.1 | 1800 | 0.8548 | 1.0068 |
| 0.3706 | 6.44 | 1900 | 0.7984 | 1.0110 |
| 0.2671 | 6.78 | 2000 | 0.8388 | 1.0117 |
| 0.2671 | 7.12 | 2100 | 0.8499 | 1.0073 |
| 0.2671 | 7.46 | 2200 | 0.8480 | 1.0112 |
| 0.2671 | 7.8 | 2300 | 0.7929 | 1.0099 |
| 0.2671 | 8.14 | 2400 | 0.8659 | 1.0089 |
| 0.2017 | 8.47 | 2500 | 0.8583 | 1.0063 |
| 0.2017 | 8.81 | 2600 | 0.8326 | 1.0110 |
| 0.2017 | 9.15 | 2700 | 0.8759 | 1.0037 |
| 0.2017 | 9.49 | 2800 | 0.8576 | 1.0100 |
| 0.2017 | 9.83 | 2900 | 0.8777 | 1.0224 |
| 0.1682 | 10.17 | 3000 | 0.8865 | 1.0280 |
| 0.1682 | 10.51 | 3100 | 0.9213 | 1.0068 |
| 0.1682 | 10.85 | 3200 | 0.8881 | 1.0152 |
| 0.1682 | 11.19 | 3300 | 0.9089 | 1.0100 |
| 0.1682 | 11.53 | 3400 | 0.8974 | 1.0127 |
| 0.1347 | 11.86 | 3500 | 0.9129 | 1.0123 |
| 0.1347 | 12.2 | 3600 | 0.9939 | 1.0169 |
| 0.1347 | 12.54 | 3700 | 0.9135 | 1.0083 |
| 0.1347 | 12.88 | 3800 | 0.9229 | 1.0118 |
| 0.1347 | 13.22 | 3900 | 0.9610 | 1.0107 |
| 0.1049 | 13.56 | 4000 | 0.9236 | 1.0099 |
| 0.1049 | 13.9 | 4100 | 0.8967 | 1.0085 |
| 0.1049 | 14.24 | 4200 | 0.8980 | 1.0081 |
| 0.1049 | 14.58 | 4300 | 0.9023 | 1.0081 |
| 0.1049 | 14.92 | 4400 | 0.9216 | 1.0079 |
| 0.0917 | 15.25 | 4500 | 0.9443 | 1.0090 |
| 0.0917 | 15.59 | 4600 | 0.9390 | 1.0091 |
| 0.0917 | 15.93 | 4700 | 0.9153 | 1.0083 |
| 0.0917 | 16.27 | 4800 | 0.9190 | 1.0092 |
| 0.0917 | 16.61 | 4900 | 0.9194 | 1.0091 |
| 0.0786 | 16.95 | 5000 | 0.9201 | 1.0091 |
Framework versions
- Transformers 4.16.0.dev0
- Pytorch 1.10.1+cu102
- Datasets 1.17.1.dev0
- Tokenizers 0.11.0