metadata
license: mit
base_model: Harveenchadha/vakyansh-wav2vec2-hindi-him-4200
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: hindi_wav2vec2_optimized_2
results: []
hindi_wav2vec2_optimized_2
This model is a fine-tuned version of Harveenchadha/vakyansh-wav2vec2-hindi-him-4200 on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 1.4408
- Wer: 0.3122
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
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 10
- num_epochs: 100
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 3.2061 | 5.56 | 25 | 1.5744 | 0.5338 |
| 0.4727 | 11.11 | 50 | 1.4735 | 0.4409 |
| 0.2901 | 16.67 | 75 | 1.6240 | 0.4536 |
| 0.1643 | 22.22 | 100 | 1.4403 | 0.4726 |
| 0.0991 | 27.78 | 125 | 1.3516 | 0.4114 |
| 0.0512 | 33.33 | 150 | 1.3740 | 0.4409 |
| 0.0645 | 38.89 | 175 | 1.3560 | 0.3903 |
| 0.0461 | 44.44 | 200 | 1.3505 | 0.3840 |
| 0.0372 | 50.0 | 225 | 1.5078 | 0.4114 |
| 0.0277 | 55.56 | 250 | 1.5161 | 0.3629 |
| 0.0201 | 61.11 | 275 | 1.4834 | 0.3819 |
| 0.016 | 66.67 | 300 | 1.4693 | 0.3608 |
| 0.0122 | 72.22 | 325 | 1.4352 | 0.3439 |
| 0.0147 | 77.78 | 350 | 1.4382 | 0.3291 |
| 0.0098 | 83.33 | 375 | 1.4329 | 0.3122 |
| 0.015 | 88.89 | 400 | 1.4408 | 0.3122 |
Framework versions
- Transformers 4.34.0
- Pytorch 2.0.1+cu118
- Datasets 2.14.5
- Tokenizers 0.14.1