whisper-small-English

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

• Loss: 0.4278
• Accuracy: 22.4848

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: 4
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 1000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.1469	0.4	100	0.3931	20.8666
0.1704	0.8	200	0.3690	20.5089
0.1317	1.2	300	0.3650	20.4210
0.1323	1.6	400	0.3659	21.3649
0.131	2.0	500	0.3675	21.1480
0.0662	2.4	600	0.4080	21.8105
0.0678	2.8	700	0.3958	22.5199
0.028	3.2	800	0.4290	22.0216
0.0313	3.6	900	0.4195	22.4496
0.032	4.0	1000	0.4278	22.4848

Framework versions

• Transformers 4.45.2
• Pytorch 2.4.1+cu121
• Datasets 3.0.1
• Tokenizers 0.20.1