golesheed/whisper-7-dutch

Whisper Large V2

This model is a fine-tuned version of openai/whisper-large-v2 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2970
• Wer: 11.6236

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: 3e-05
• train_batch_size: 12
• 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: 20
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.6426	0.09	30	0.3506	16.5079
0.3086	0.19	60	0.3201	12.7211
0.3121	0.28	90	0.2968	11.7123
0.2956	0.38	120	0.2937	13.6116
0.3067	0.47	150	0.2769	14.4193
0.2787	0.57	180	0.2717	14.5051
0.2382	0.66	210	0.2732	17.2889
0.232	0.76	240	0.2742	19.2148
0.2653	0.85	270	0.2632	18.9604
0.2726	0.95	300	0.2497	15.6174
0.1879	1.04	330	0.2603	12.4549
0.124	1.14	360	0.2590	11.1769
0.1244	1.23	390	0.2679	17.9486
0.1482	1.33	420	0.2590	16.1263
0.1312	1.42	450	0.2628	15.2595
0.1358	1.52	480	0.2550	13.0347
0.1302	1.61	510	0.2545	15.1648
0.132	1.71	540	0.2508	15.3127
0.1402	1.8	570	0.2418	12.2330
0.137	1.9	600	0.2444	13.2329
0.1346	1.99	630	0.2432	13.1649
0.0664	2.09	660	0.2594	11.6058
0.0562	2.18	690	0.2655	10.9431
0.0551	2.28	720	0.2613	13.3690
0.0625	2.37	750	0.2555	20.2769
0.0627	2.47	780	0.2602	17.7268
0.0586	2.56	810	0.2647	11.5319
0.0604	2.66	840	0.2615	11.0378
0.062	2.75	870	0.2570	12.0111
0.0548	2.85	900	0.2575	14.5317
0.0576	2.94	930	0.2585	12.2182
0.0448	3.04	960	0.2619	13.0406
0.023	3.13	990	0.2730	12.9578
0.0241	3.23	1020	0.2773	11.9667
0.023	3.32	1050	0.2738	11.7656
0.0222	3.42	1080	0.2767	11.8602
0.0201	3.51	1110	0.2723	11.3455
0.0195	3.61	1140	0.2803	10.6946
0.0221	3.7	1170	0.2744	11.3899
0.0202	3.8	1200	0.2764	11.3070
0.0223	3.89	1230	0.2725	11.2567
0.021	3.99	1260	0.2781	10.6148
0.01	4.08	1290	0.2854	10.7508
0.0081	4.18	1320	0.2914	10.5970
0.0086	4.27	1350	0.2918	11.0408
0.0073	4.37	1380	0.2946	11.2301
0.0085	4.46	1410	0.2950	10.8721
0.007	4.56	1440	0.2957	11.4224
0.0063	4.65	1470	0.2965	11.4431
0.0073	4.75	1500	0.2976	11.0970
0.0078	4.84	1530	0.2972	11.5289
0.0068	4.94	1560	0.2970	11.6236

Framework versions

• Transformers 4.38.0.dev0
• Pytorch 2.1.0+cu121
• Datasets 2.17.0
• Tokenizers 0.15.0