whisper-small-it

This model is a fine-tuned version of openai/whisper-small on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1919
• Wer: 11.72

Model description

More information needed

Intended uses & limitations

I have left this model here. BUt the "small3-it", produced later, has better performance.

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
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 2
• total_train_batch_size: 256
• total_eval_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.1441	1.68	1000	0.1912	0.1256
0.0653	3.36	2000	0.1845	0.1182
0.0374	5.03	3000	0.1919	0.1172
0.0238	6.71	4000	0.2069	0.1202
0.0162	8.39	5000	0.2184	0.1223

Framework versions

• Transformers 4.26.0.dev0
• Pytorch 1.13.0
• Datasets 2.7.1.dev0
• Tokenizers 0.13.2