arun100/whisper-small-fa-2

Whisper Small Persian Iranian

This model is a fine-tuned version of openai/whisper-small on the mozilla-foundation/common_voice_16_0 fa dataset. It achieves the following results on the evaluation set:

• Loss: 0.4858
• Wer: 39.7201

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-06
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 50
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.4531	1.03	500	0.6448	50.7393
0.4031	3.0	1000	0.5755	46.5001
0.2745	4.04	1500	0.5389	43.7190
0.336	6.0	2000	0.5166	42.4056
0.2429	7.04	2500	0.5045	41.1810
0.2852	9.01	3000	0.4941	40.6444
0.2217	10.04	3500	0.4888	40.1106
0.2384	12.01	4000	0.4873	39.9208
0.1889	13.04	4500	0.4858	39.7201
0.2202	15.01	5000	0.4888	39.7228

Framework versions

• Transformers 4.38.0.dev0
• Pytorch 2.1.2+cu121
• Datasets 2.16.2.dev0
• Tokenizers 0.15.0