Sagicc/whisper-medium-sr-fleurs

Whisper Medium Sr Fleurs

This model is a fine-tuned version of openai/whisper-medium on the Google Fleurs dataset. It achieves the following results on the evaluation set:

• Loss: 0.3577
• Wer Ortho: 0.2072
• Wer: 0.1794

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: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant_with_warmup
• lr_scheduler_warmup_steps: 50
• training_steps: 4000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Ortho	Wer
0.0341	2.49	500	0.2704	0.2074	0.1789
0.0109	4.98	1000	0.3091	0.2075	0.1774
0.006	7.46	1500	0.3143	0.2031	0.1713
0.0081	9.95	2000	0.3284	0.2070	0.1754
0.0038	12.44	2500	0.3426	0.2099	0.1805
0.0042	14.93	3000	0.3630	0.2113	0.1821
0.0032	17.41	3500	0.3659	0.2089	0.1791
0.0046	19.9	4000	0.3577	0.2072	0.1794

Framework versions

• Transformers 4.33.1
• Pytorch 2.0.1+cu117
• Datasets 2.14.5
• Tokenizers 0.13.3