English-4

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

• Loss: 0.0007
• Wer: 10.1983

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: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	1.0	15	1.1627	31.3503
No log	2.0	30	0.2005	16.9027
No log	3.0	45	0.0768	11.8980
No log	4.0	60	0.0250	11.1426
No log	5.0	75	0.0118	10.0094
No log	6.0	90	0.0083	30.5005
No log	7.0	105	0.0032	30.9726
No log	8.0	120	0.0024	10.3872
No log	9.0	135	0.0018	10.2927
No log	10.0	150	0.0011	10.4816
No log	11.0	165	0.0009	10.1983
No log	12.0	180	0.0008	10.1983
No log	13.0	195	0.0008	10.1983
No log	14.0	210	0.0007	10.1983
No log	15.0	225	0.0007	10.1983

Framework versions

• Transformers 4.30.0.dev0
• Pytorch 2.1.0+cu118
• Datasets 2.15.0
• Tokenizers 0.13.3