whisper-medium-pt-cv16-fleurs2-lr-wu

This model is a fine-tuned version of openai/whisper-medium on the fsicoli/cv16-fleurs default dataset. It achieves the following results on the evaluation set:

• Loss: 0.1636
• Wer: 0.0997

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: 6.25e-06
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 25000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0497	2.3343	5000	0.1636	0.0997
0.0122	4.6685	10000	0.1944	0.0995
0.0129	7.0028	15000	0.1997	0.0983
0.0045	9.3371	20000	0.2126	0.0998
0.005	11.6713	25000	0.2110	0.0921

Framework versions

• Transformers 4.45.0.dev0
• Pytorch 2.4.1
• Datasets 2.21.0
• Tokenizers 0.19.1