whisper-small-eng

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: 0.5746
• Wer: 24.4747

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: 0.0001
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 50
• training_steps: 1000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.7025	0.03	100	0.6855	36.9988
0.7478	0.07	200	0.8034	35.4196
0.7516	0.1	300	0.7854	31.8551
0.7175	0.13	400	0.7868	32.9444
0.6748	0.17	500	0.7239	31.1203
0.6739	0.2	600	0.7045	29.7473
0.6262	0.24	700	0.6620	27.1239
0.585	0.27	800	0.6254	26.6147
0.5305	0.3	900	0.5877	24.6552
0.5463	0.34	1000	0.5746	24.4747

Framework versions

• Transformers 4.38.0.dev0
• Pytorch 2.1.2+cu121
• Datasets 2.16.1
• Tokenizers 0.15.1