leenag/Malasar_Luke

This model is a fine-tuned version of vasista22/whisper-tamil-small on the Spoken Bible Corpus: Malasar dataset. It achieves the following results on the evaluation set:

Loss: 0.5039
Wer: 51.9132

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: 32
eval_batch_size: 16
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
training_steps: 2000
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0924	11.3636	250	0.2929	52.1416
0.0013	22.7273	500	0.4152	52.8270
0.0008	34.0909	750	0.4400	52.8841
0.0002	45.4545	1000	0.4761	52.1416
0.0001	56.8182	1250	0.4888	51.6276
0.0001	68.1818	1500	0.4970	51.9703
0.0001	79.5455	1750	0.5021	51.7990
0.0001	90.9091	2000	0.5039	51.9132

Framework versions

Transformers 4.41.2
Pytorch 2.0.1+cu117
Datasets 2.16.0
Tokenizers 0.19.1

Citations

@misc{multistage2024,
      title={Multistage Fine-tuning Strategies for Automatic Speech Recognition in Low-resource Languages}, 
      author={Leena G Pillai, Kavya Manohar, Basil K Raju, Elizabeth Sherly},
      year={2024},
      eprint={2411.04573},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2411.04573}, 
}