meiiny00/whisper-medium-better-total

Whisper Small wow

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: 0.2997
• Wer: 22.5766

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: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.3257	0.7590	1000	0.3054	38.2370
0.1796	1.5180	2000	0.2268	29.0487
0.1038	2.2770	3000	0.2227	26.1642
0.0604	3.0361	4000	0.2151	24.7201
0.0603	3.7951	5000	0.2384	23.6619
0.033	4.5541	6000	0.2567	23.5351
0.0211	5.3131	7000	0.2501	23.2017
0.0081	6.0721	8000	0.2622	23.1781
0.0096	6.8311	9000	0.2675	23.2162
0.0043	7.5901	10000	0.2889	22.6327
0.0027	8.3491	11000	0.2916	22.7324
0.0019	9.1082	12000	0.3027	22.7994
0.0017	9.8672	13000	0.2997	22.5766

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.1+cu124
• Datasets 2.20.0
• Tokenizers 0.19.1