taesiri/whisper-small-fa-7

Whisper Small Fa - BuzzyBuzzy

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: 0.6904
• Wer: 34.2597

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
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 20000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0099	0.86	2000	0.6116	37.0499
0.0079	1.72	4000	0.6340	37.3746
0.0069	2.58	6000	0.6159	35.9455
0.0015	3.44	8000	0.6187	35.3281
0.0009	4.3	10000	0.6449	35.0783
0.0007	5.15	12000	0.6440	34.8452
0.0002	6.01	14000	0.6664	34.3263
0.0001	6.87	16000	0.6817	34.3471
0.0	7.73	18000	0.6809	34.1737
0.0	8.59	20000	0.6904	34.2597

Framework versions

• Transformers 4.26.0.dev0
• Pytorch 1.13.1+cu117
• Datasets 2.8.0
• Tokenizers 0.13.2