golesheed/whisper-3-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: 0.2535
• Wer: 8.9988

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
0.6185	0.09	30	0.3181	12.1555
0.3243	0.19	60	0.2801	11.9994
0.3044	0.28	90	0.2689	11.9876
0.2843	0.38	120	0.2574	10.4270
0.2859	0.47	150	0.2427	12.1879
0.271	0.57	180	0.2374	14.0459
0.2584	0.66	210	0.2319	11.1690
0.2916	0.76	240	0.2302	13.6013
0.2781	0.85	270	0.2224	10.6832
0.2498	0.95	300	0.2244	10.2945
0.2033	1.04	330	0.2311	11.3045
0.1323	1.14	360	0.2268	10.9393
0.1322	1.23	390	0.2242	9.9912
0.1312	1.33	420	0.2267	14.3993
0.1392	1.42	450	0.2209	9.9352
0.1437	1.52	480	0.2146	10.0824
0.1299	1.61	510	0.2198	16.3516
0.1328	1.71	540	0.2161	10.0118
0.1425	1.8	570	0.2133	11.3280
0.1332	1.9	600	0.2137	10.4476
0.1354	1.99	630	0.2101	10.0324
0.0601	2.09	660	0.2241	9.2285
0.0557	2.18	690	0.2235	9.0548
0.0567	2.28	720	0.2239	9.5259
0.0583	2.37	750	0.2246	11.3575
0.0642	2.47	780	0.2241	9.7556
0.059	2.56	810	0.2256	10.1266
0.0596	2.66	840	0.2228	9.5318
0.0571	2.75	870	0.2206	12.1290
0.0581	2.85	900	0.2222	10.4240
0.063	2.94	930	0.2229	9.3551
0.0428	3.04	960	0.2313	9.8557
0.0237	3.13	990	0.2337	9.7261
0.0228	3.23	1020	0.2380	9.3433
0.022	3.32	1050	0.2403	9.6849
0.0235	3.42	1080	0.2342	9.5878
0.0206	3.51	1110	0.2341	9.0371
0.0205	3.61	1140	0.2391	9.2668
0.022	3.7	1170	0.2336	9.6496
0.0201	3.8	1200	0.2363	9.4876
0.0213	3.89	1230	0.2303	9.5819
0.0206	3.99	1260	0.2348	9.4670
0.0098	4.08	1290	0.2450	9.4729
0.0088	4.18	1320	0.2497	9.1461
0.0076	4.27	1350	0.2497	9.2815
0.0086	4.37	1380	0.2509	9.0901
0.0064	4.46	1410	0.2524	8.9164
0.0075	4.56	1440	0.2539	8.9340
0.0069	4.65	1470	0.2532	8.9870
0.0083	4.75	1500	0.2529	9.0135
0.0064	4.84	1530	0.2536	8.9605
0.0065	4.94	1560	0.2535	8.9988

Framework versions

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