whisper-base.en-fsc

This model is a fine-tuned version of openai/whisper-base.en on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0437
• Accuracy: 0.9950

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: 0.0005
• train_batch_size: 48
• eval_batch_size: 48
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 192
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 25

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.9959	120	0.0862	0.9739
No log	2.0	241	0.0422	0.9866
No log	2.9959	361	0.0630	0.9823
No log	4.0	482	0.0630	0.9805
No log	4.9959	602	0.0626	0.9821
No log	6.0	723	0.0339	0.9905
No log	6.9959	843	0.0452	0.9897
No log	8.0	964	0.0527	0.9834
0.1514	8.9959	1084	0.0637	0.9868
0.1514	10.0	1205	0.0443	0.9921
0.1514	10.9959	1325	0.0306	0.9937
0.1514	12.0	1446	0.0416	0.9897
0.1514	12.9959	1566	0.0363	0.9910
0.1514	14.0	1687	0.0413	0.9924
0.1514	14.9959	1807	0.0344	0.9945
0.1514	16.0	1928	0.0508	0.9924
0.0161	16.9959	2048	0.0436	0.9937
0.0161	18.0	2169	0.0435	0.9931
0.0161	18.9959	2289	0.0428	0.9945
0.0161	20.0	2410	0.0425	0.9947
0.0161	20.9959	2530	0.0432	0.9947
0.0161	22.0	2651	0.0438	0.9947
0.0161	22.9959	2771	0.0437	0.9950
0.0161	24.0	2892	0.0438	0.9950
0.0011	24.8963	3000	0.0438	0.9950

Framework versions

• Transformers 4.43.3
• Pytorch 2.2.2+cu121
• Datasets 2.18.0
• Tokenizers 0.19.1