zagibest/whisper-small-custom-data-2

Whisper Small MN with custom data + Common voice - Zagi

This model is a fine-tuned version of zagibest/whisper-small-custom-data on the common_voice_11_0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.6134
• Wer: 43.3432

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: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 4000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.2751	1.98	500	0.4451	47.4170
0.0614	3.97	1000	0.4734	45.0579
0.0141	5.95	1500	0.5313	44.3370
0.0033	7.94	2000	0.5615	43.6490
0.0011	9.92	2500	0.5826	43.8565
0.0011	11.9	3000	0.6012	43.3705
0.0004	13.89	3500	0.6094	43.3486
0.0004	15.87	4000	0.6134	43.3432

Framework versions

• Transformers 4.39.1
• Pytorch 2.0.1+cu117
• Datasets 2.18.0
• Tokenizers 0.15.2