steja/whisper-large-persian

Whisper large persian

This model is a fine-tuned version of anuragshas/whisper-large-v2-hi on the mozilla-foundation/common_voice_11_0 fa dataset. It achieves the following results on the evaluation set:

• Loss: 0.3047
• Wer: 26.3679

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: 8
• eval_batch_size: 16
• seed: 42
• distributed_type: multi-GPU
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• training_steps: 1000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.2948	0.19	250	0.4258	35.6023
0.2443	0.39	500	0.3650	30.9747
0.1956	0.58	750	0.3228	28.0196
0.1715	0.78	1000	0.3047	26.3679

Framework versions

• Transformers 4.25.1
• Pytorch 1.13.1+cu117
• Datasets 2.8.0
• Tokenizers 0.13.2