<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
hindi_beekeeping_wav2vec2
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: 0.6793
- Wer: 0.4048
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: 100
- num_epochs: 5
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
26.5108 | 0.11 | 25 | 2.9146 | 1.1520 |
3.8351 | 0.22 | 50 | 2.1736 | 0.6667 |
2.1489 | 0.33 | 75 | 1.5539 | 0.5914 |
3.2097 | 0.44 | 100 | 1.9828 | 0.6840 |
1.5213 | 0.54 | 125 | 1.4369 | 0.5410 |
2.1797 | 0.65 | 150 | 1.7904 | 0.7449 |
1.2368 | 0.76 | 175 | 1.3543 | 0.5583 |
1.5755 | 0.87 | 200 | 1.5444 | 0.6712 |
1.1638 | 0.98 | 225 | 1.4639 | 0.6005 |
0.8985 | 1.09 | 250 | 1.0658 | 0.5139 |
0.9406 | 1.2 | 275 | 1.3181 | 0.5756 |
1.1215 | 1.31 | 300 | 1.0163 | 0.4680 |
1.3881 | 1.42 | 325 | 1.1547 | 0.5854 |
0.8113 | 1.53 | 350 | 0.9433 | 0.4771 |
0.689 | 1.63 | 375 | 1.0819 | 0.5064 |
0.726 | 1.74 | 400 | 1.0025 | 0.5252 |
0.6475 | 1.85 | 425 | 1.0670 | 0.5154 |
1.0345 | 1.96 | 450 | 0.9535 | 0.5019 |
0.8327 | 2.07 | 475 | 0.8866 | 0.4733 |
0.4843 | 2.18 | 500 | 0.9580 | 0.5087 |
0.6657 | 2.29 | 525 | 0.9019 | 0.4710 |
0.4843 | 2.4 | 550 | 0.8207 | 0.4665 |
0.6666 | 2.51 | 575 | 0.7377 | 0.4695 |
0.5012 | 2.61 | 600 | 0.8135 | 0.4537 |
0.7776 | 2.72 | 625 | 0.8131 | 0.4612 |
0.4538 | 2.83 | 650 | 0.8194 | 0.4590 |
0.5659 | 2.94 | 675 | 0.8480 | 0.4582 |
0.9362 | 3.05 | 700 | 0.8532 | 0.4485 |
0.3629 | 3.16 | 725 | 0.8447 | 0.4582 |
0.5201 | 3.27 | 750 | 0.7486 | 0.4409 |
0.3513 | 3.38 | 775 | 0.7865 | 0.4439 |
0.4152 | 3.49 | 800 | 0.7510 | 0.4364 |
0.8962 | 3.59 | 825 | 0.7758 | 0.4342 |
0.4558 | 3.7 | 850 | 0.7314 | 0.4296 |
0.3476 | 3.81 | 875 | 0.6861 | 0.4153 |
0.3978 | 3.92 | 900 | 0.6961 | 0.4153 |
0.3818 | 4.03 | 925 | 0.6960 | 0.4071 |
0.2847 | 4.14 | 950 | 0.7222 | 0.4048 |
0.3997 | 4.25 | 975 | 0.6960 | 0.4191 |
0.2695 | 4.36 | 1000 | 0.6894 | 0.4138 |
0.4229 | 4.47 | 1025 | 0.7215 | 0.4304 |
0.2665 | 4.58 | 1050 | 0.7096 | 0.4056 |
0.3158 | 4.68 | 1075 | 0.6904 | 0.4153 |
0.2452 | 4.79 | 1100 | 0.6777 | 0.4071 |
0.3034 | 4.9 | 1125 | 0.6793 | 0.4048 |
Framework versions
- Transformers 4.34.0
- Pytorch 2.0.1+cu118
- Datasets 2.14.5
- Tokenizers 0.14.1