golesheed/whisper-4-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: 0.2941
• Wer: 9.7158

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
0.6299	0.09	30	0.3564	16.3717
0.3398	0.19	60	0.3210	12.9819
0.3187	0.28	90	0.2997	19.9971
0.2773	0.38	120	0.2939	15.2908
0.2745	0.47	150	0.2780	15.0405
0.2677	0.57	180	0.2697	12.3840
0.2467	0.66	210	0.2698	13.6033
0.2467	0.76	240	0.2735	16.5749
0.2455	0.85	270	0.2639	12.0188
0.269	0.95	300	0.2597	13.3412
0.1851	1.04	330	0.2643	12.3428
0.1265	1.14	360	0.2561	13.4649
0.1377	1.23	390	0.2662	12.8081
0.134	1.33	420	0.2640	12.3310
0.1371	1.42	450	0.2630	11.8480
0.1307	1.52	480	0.2616	11.9187
0.1423	1.61	510	0.2535	11.3150
0.1406	1.71	540	0.2525	10.9675
0.1312	1.8	570	0.2483	13.9479
0.1214	1.9	600	0.2534	12.3192
0.1252	1.99	630	0.2531	11.7243
0.0657	2.09	660	0.2619	11.0558
0.0578	2.18	690	0.2698	12.2191
0.0548	2.28	720	0.2662	10.3667
0.0596	2.37	750	0.2685	12.3222
0.0573	2.47	780	0.2698	10.5581
0.0589	2.56	810	0.2661	11.7391
0.0554	2.66	840	0.2608	11.7332
0.0625	2.75	870	0.2622	10.7760
0.0586	2.85	900	0.2603	10.7201
0.0647	2.94	930	0.2576	10.5669
0.0486	3.04	960	0.2647	10.2518
0.0245	3.13	990	0.2749	10.6140
0.0256	3.23	1020	0.2707	10.2813
0.0242	3.32	1050	0.2724	11.6566
0.0225	3.42	1080	0.2699	10.6347
0.0205	3.51	1110	0.2748	10.0427
0.0217	3.61	1140	0.2747	10.0339
0.0216	3.7	1170	0.2775	9.9190
0.0222	3.8	1200	0.2770	10.2371
0.0204	3.89	1230	0.2722	10.1782
0.0185	3.99	1260	0.2725	9.7835
0.0111	4.08	1290	0.2834	9.8866
0.0085	4.18	1320	0.2854	9.7894
0.0082	4.27	1350	0.2868	9.7629
0.0075	4.37	1380	0.2906	9.7776
0.0079	4.46	1410	0.2918	9.7394
0.0071	4.56	1440	0.2902	9.6157
0.0076	4.65	1470	0.2921	9.5921
0.0071	4.75	1500	0.2940	9.5774
0.0069	4.84	1530	0.2936	9.7276
0.0071	4.94	1560	0.2941	9.7158

Framework versions

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