| | --- |
| | library_name: transformers |
| | license: cc-by-nc-4.0 |
| | base_model: facebook/mms-1b-all |
| | tags: |
| | - generated_from_trainer |
| | metrics: |
| | - wer |
| | model-index: |
| | - name: ssc-bew-mms-model |
| | results: [] |
| | --- |
| | |
| | <!-- 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. --> |
| |
|
| | # ssc-bew-mms-model |
| |
|
| | This model is a fine-tuned version of [facebook/mms-1b-all](https://huggingface.co/facebook/mms-1b-all) on an unknown dataset. |
| | It achieves the following results on the evaluation set: |
| | - Loss: 0.7747 |
| | - Cer: 0.2038 |
| | - Wer: 0.6111 |
| |
|
| | ## 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: 8 |
| | - eval_batch_size: 12 |
| | - seed: 42 |
| | - gradient_accumulation_steps: 2 |
| | - total_train_batch_size: 16 |
| | - optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments |
| | - lr_scheduler_type: linear |
| | - lr_scheduler_warmup_steps: 100 |
| | - num_epochs: 10 |
| | - mixed_precision_training: Native AMP |
| | |
| | ### Training results |
| | |
| | | Training Loss | Epoch | Step | Validation Loss | Cer | Wer | |
| | |:-------------:|:------:|:----:|:---------------:|:------:|:------:| |
| | | 1.6619 | 0.3063 | 200 | 0.9694 | 0.2219 | 0.6669 | |
| | | 1.5435 | 0.6126 | 400 | 0.8620 | 0.2158 | 0.6542 | |
| | | 1.5126 | 0.9188 | 600 | 0.8212 | 0.2101 | 0.6326 | |
| | | 1.4544 | 1.2251 | 800 | 0.8192 | 0.2101 | 0.6346 | |
| | | 1.4753 | 1.5314 | 1000 | 0.8128 | 0.2096 | 0.6393 | |
| | | 1.4388 | 1.8377 | 1200 | 0.8112 | 0.2098 | 0.6292 | |
| | | 1.4494 | 2.1440 | 1400 | 0.8120 | 0.2084 | 0.6268 | |
| | | 1.4289 | 2.4502 | 1600 | 0.7985 | 0.2075 | 0.6254 | |
| | | 1.4736 | 2.7565 | 1800 | 0.7951 | 0.2070 | 0.6263 | |
| | | 1.4217 | 3.0628 | 2000 | 0.8154 | 0.2104 | 0.6369 | |
| | | 1.4146 | 3.3691 | 2200 | 0.8092 | 0.2094 | 0.6295 | |
| | | 1.4205 | 3.6753 | 2400 | 0.8056 | 0.2092 | 0.6298 | |
| | | 1.3951 | 3.9816 | 2600 | 0.8070 | 0.2087 | 0.6384 | |
| | | 1.3695 | 4.2879 | 2800 | 0.7980 | 0.2078 | 0.6261 | |
| | | 1.4743 | 4.5942 | 3000 | 0.8138 | 0.2099 | 0.6376 | |
| | | 1.4499 | 4.9005 | 3200 | 0.7828 | 0.2041 | 0.6168 | |
| | | 1.3801 | 5.2067 | 3400 | 0.8019 | 0.2086 | 0.6324 | |
| | | 1.4191 | 5.5130 | 3600 | 0.7931 | 0.2064 | 0.6229 | |
| | | 1.4359 | 5.8193 | 3800 | 0.7794 | 0.2034 | 0.6144 | |
| | | 1.407 | 6.1256 | 4000 | 0.7877 | 0.2064 | 0.6243 | |
| | | 1.4079 | 6.4319 | 4200 | 0.7963 | 0.2071 | 0.6221 | |
| | | 1.417 | 6.7381 | 4400 | 0.7839 | 0.2048 | 0.6139 | |
| | | 1.4031 | 7.0444 | 4600 | 0.7759 | 0.2033 | 0.6098 | |
| | | 1.3999 | 7.3507 | 4800 | 0.7781 | 0.2043 | 0.6137 | |
| | | 1.4052 | 7.6570 | 5000 | 0.7769 | 0.2041 | 0.6110 | |
| | | 1.4102 | 7.9632 | 5200 | 0.7750 | 0.2031 | 0.6085 | |
| | | 1.392 | 8.2695 | 5400 | 0.7776 | 0.2041 | 0.6115 | |
| | | 1.4032 | 8.5758 | 5600 | 0.7733 | 0.2033 | 0.6096 | |
| | | 1.3956 | 8.8821 | 5800 | 0.7761 | 0.2038 | 0.6101 | |
| | | 1.4148 | 9.1884 | 6000 | 0.7761 | 0.2040 | 0.6102 | |
| | | 1.3921 | 9.4946 | 6200 | 0.7746 | 0.2036 | 0.6099 | |
| | | 1.4109 | 9.8009 | 6400 | 0.7747 | 0.2038 | 0.6111 | |
| | |
| | |
| | ### Framework versions |
| | |
| | - Transformers 4.57.2 |
| | - Pytorch 2.9.1+cu128 |
| | - Datasets 3.6.0 |
| | - Tokenizers 0.22.0 |
| | |
