metadata
license: cc-by-nc-4.0
base_model: facebook/mms-1b-all
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: wav2vec2-large-mms-1b-kazakh-speech2ner-ksc_t-16b-4ep
results: []
wav2vec2-large-mms-1b-kazakh-speech2ner-ksc_t-16b-4ep
This model is a fine-tuned version of facebook/mms-1b-all on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.2397
- Wer: 0.3099
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: 1e-05
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 1000
- num_epochs: 4
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.3934 | 0.22 | 2000 | 0.3610 | 0.3687 |
| 0.3361 | 0.43 | 4000 | 0.2983 | 0.3412 |
| 0.3211 | 0.65 | 6000 | 0.2779 | 0.3300 |
| 0.3146 | 0.87 | 8000 | 0.2685 | 0.3238 |
| 0.3104 | 1.09 | 10000 | 0.2613 | 0.3210 |
| 0.2965 | 1.3 | 12000 | 0.2571 | 0.3188 |
| 0.3004 | 1.52 | 14000 | 0.2531 | 0.3166 |
| 0.2889 | 1.74 | 16000 | 0.2504 | 0.3153 |
| 0.2955 | 1.96 | 18000 | 0.2476 | 0.3138 |
| 0.2869 | 2.17 | 20000 | 0.2465 | 0.3126 |
| 0.2855 | 2.39 | 22000 | 0.2443 | 0.3117 |
| 0.2927 | 2.61 | 24000 | 0.2431 | 0.3111 |
| 0.2789 | 2.83 | 26000 | 0.2421 | 0.3107 |
| 0.2854 | 3.04 | 28000 | 0.2412 | 0.3105 |
| 0.2918 | 3.26 | 30000 | 0.2404 | 0.3099 |
| 0.2768 | 3.48 | 32000 | 0.2401 | 0.3096 |
| 0.2771 | 3.69 | 34000 | 0.2398 | 0.3099 |
| 0.2733 | 3.91 | 36000 | 0.2397 | 0.3099 |
Framework versions
- Transformers 4.33.0.dev0
- Pytorch 2.0.1+cu117
- Datasets 2.13.1
- Tokenizers 0.13.3