mms-1b-all-bemgen-combined-m50f100-42-DAT-4e-1
This model is a fine-tuned version of facebook/mms-1b-all on the BEMGEN - BEM dataset. It achieves the following results on the evaluation set:
- Loss: 0.2721
- Cer: 0.0767
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: 4
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- optimizer: Use OptimizerNames.ADAMW_TORCH 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: 30.0
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Cer |
---|---|---|---|---|
3.56 | 0.5618 | 100 | 3.0027 | 1.0 |
1.1897 | 1.1236 | 200 | 0.9603 | 0.2119 |
0.7163 | 1.6854 | 300 | 0.3618 | 0.1030 |
0.6309 | 2.2472 | 400 | 0.3283 | 0.0939 |
0.696 | 2.8090 | 500 | 0.3058 | 0.0877 |
0.6895 | 3.3708 | 600 | 0.2999 | 0.0847 |
0.7015 | 3.9326 | 700 | 0.2938 | 0.0802 |
0.6893 | 4.4944 | 800 | 0.2888 | 0.0811 |
0.6711 | 5.0562 | 900 | 0.2887 | 0.0829 |
0.6836 | 5.6180 | 1000 | 0.2900 | 0.0783 |
0.6954 | 6.1798 | 1100 | 0.2844 | 0.0796 |
0.6897 | 6.7416 | 1200 | 0.2880 | 0.0797 |
0.6496 | 7.3034 | 1300 | 0.2844 | 0.0789 |
0.6582 | 7.8652 | 1400 | 0.2806 | 0.0785 |
0.6335 | 8.4270 | 1500 | 0.2778 | 0.0785 |
0.6408 | 8.9888 | 1600 | 0.2721 | 0.0767 |
0.6096 | 9.5506 | 1700 | 0.2750 | 0.0770 |
0.6108 | 10.1124 | 1800 | 0.2758 | 0.0783 |
0.6535 | 10.6742 | 1900 | 0.2761 | 0.0789 |
Framework versions
- Transformers 4.53.0.dev0
- Pytorch 2.6.0+cu124
- Datasets 3.6.0
- Tokenizers 0.21.0
- Downloads last month
- 32
Model tree for csikasote/mms-1b-all-bemgen-combined-m50f100-42-DAT-4e-1
Base model
facebook/mms-1b-all