leenag/Malasar_Dict

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

• Loss: 0.0136
• Wer: 6.4752

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
0.0424	0.64	250	0.0426	19.6509
0.0247	1.28	500	0.0276	12.7252
0.0184	1.92	750	0.0254	13.8514
0.0077	2.56	1000	0.0152	7.2072
0.005	3.21	1250	0.0139	6.9257
0.0032	3.85	1500	0.0149	6.9257
0.0025	4.49	1750	0.0145	6.7005
0.0009	5.13	2000	0.0136	6.4752

Framework versions

• Transformers 4.37.0.dev0
• Pytorch 2.0.1+cu117
• Datasets 2.12.0
• Tokenizers 0.15.0