vittai23/whisper-small-marathiV3

whisper_marathi_small_V1

This model is a fine-tuned version of openai/whisper-small on the Common Voice 11.0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.2754
• Wer: 45.0068

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: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 10
• training_steps: 1000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.4794	0.41	100	0.4754	59.9317
0.3121	0.81	200	0.3161	52.8786
0.2051	1.22	300	0.2900	50.2547
0.1887	1.63	400	0.2779	48.1336
0.16	2.03	500	0.2679	46.2639
0.1131	2.44	600	0.2706	45.8449
0.1128	2.85	700	0.2658	45.1551
0.0678	3.25	800	0.2763	45.2195
0.075	3.66	900	0.2769	45.7611
0.0609	4.07	1000	0.2754	45.0068

Framework versions

• Transformers 4.27.0.dev0
• Pytorch 1.13.1+cu116
• Datasets 2.10.1
• Tokenizers 0.13.2