metadata
library_name: transformers
license: cc-by-nc-4.0
base_model: Serialtechlab/mms-1b-dhivehi-v2
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: mms-1b-dhivehi-v3
results: []
mms-1b-dhivehi-v3
This model is a fine-tuned version of Serialtechlab/mms-1b-dhivehi-v2 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.3994
- Wer: 0.5386
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: 16
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- 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: 3
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.6227 | 0.6678 | 200 | 0.3994 | 0.5386 |
| 0.6119 | 1.3339 | 400 | 0.3945 | 0.5397 |
| 0.6421 | 2.0 | 600 | 0.3925 | 0.5395 |
| 0.5985 | 2.6678 | 800 | 0.3916 | 0.5396 |
Framework versions
- Transformers 4.57.3
- Pytorch 2.9.0+cu126
- Datasets 4.0.0
- Tokenizers 0.22.1