whisper-ai-clp

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: 0.0000
• Wer: 12.2249

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: 0.0004
• 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: 132
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
1.1763	1.2422	100	0.0155	10.5134
0.1632	2.4845	200	0.0289	19.5599
0.0783	3.7267	300	0.0152	13.0807
0.0676	4.9689	400	0.1121	33.6186
0.0708	6.2112	500	0.0014	11.3692
0.0377	7.4534	600	0.0041	8.5575
0.0323	8.6957	700	0.0001	19.0709
0.0311	9.9379	800	0.0021	14.4254
0.0236	11.1801	900	0.0001	16.9927
0.0189	12.4224	1000	0.0027	13.9364
0.01	13.6646	1100	0.0002	9.9022
0.0071	14.9068	1200	0.0022	16.6259
0.0085	16.1491	1300	0.0002	11.3692
0.0043	17.3913	1400	0.0013	14.3032
0.0056	18.6335	1500	0.0001	9.7800
0.0021	19.8758	1600	0.0046	10.6357
0.0013	21.1180	1700	0.0019	12.8362
0.0006	22.3602	1800	0.0000	11.7359
0.0001	23.6025	1900	0.0000	12.1027
0.0	24.8447	2000	0.0000	12.2249
0.0	26.0870	2100	0.0000	12.2249
0.0	27.3292	2200	0.0000	12.2249
0.0	28.5714	2300	0.0000	12.2249
0.0	29.8137	2400	0.0000	12.2249

Framework versions

• Transformers 4.45.0.dev0
• Pytorch 2.4.0
• Datasets 2.21.0
• Tokenizers 0.19.1