aria-whisper-malayalam

This model is a fine-tuned version of amrith8948/aria-whisper-malayalam on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1032
• Wer: 59.5602

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: 2e-06
• train_batch_size: 8
• eval_batch_size: 8
• 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: 20
• training_steps: 800
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.1193	0.5464	200	0.1144	60.7343
0.0925	1.0929	400	0.1089	60.8461
0.1012	1.6393	600	0.1050	59.5975
0.0882	2.1858	800	0.1032	59.5602

Framework versions

• Transformers 4.46.3
• Pytorch 2.10.0+cu128
• Datasets 4.0.0
• Tokenizers 0.20.3