golesheed/whisper-6-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: 0.2647
• Wer: 9.9296

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
0.6181	0.09	30	0.3443	17.6297
0.3296	0.19	60	0.2921	13.0722
0.3165	0.28	90	0.2711	11.8459
0.2775	0.38	120	0.2677	11.4110
0.2696	0.47	150	0.2570	12.1474
0.2558	0.57	180	0.2544	13.9623
0.272	0.66	210	0.2448	19.9809
0.2696	0.76	240	0.2415	12.5359
0.2668	0.85	270	0.2392	11.7154
0.2558	0.95	300	0.2318	12.3097
0.2108	1.04	330	0.2418	15.0639
0.1339	1.14	360	0.2409	13.8318
0.1326	1.23	390	0.2394	16.9165
0.1357	1.33	420	0.2362	11.4371
0.1278	1.42	450	0.2377	12.3503
0.1485	1.52	480	0.2291	12.3097
0.1263	1.61	510	0.2381	12.2547
0.1269	1.71	540	0.2328	10.3528
0.1437	1.8	570	0.2284	11.7415
0.1314	1.9	600	0.2270	11.4661
0.1282	1.99	630	0.2334	11.8807
0.0646	2.09	660	0.2414	11.3617
0.0655	2.18	690	0.2439	11.9503
0.0658	2.28	720	0.2406	11.4197
0.0573	2.37	750	0.2375	11.0747
0.0584	2.47	780	0.2364	10.7848
0.0639	2.56	810	0.2351	11.2197
0.0537	2.66	840	0.2380	10.0803
0.0538	2.75	870	0.2359	10.1180
0.0551	2.85	900	0.2347	10.1847
0.0613	2.94	930	0.2354	10.7616
0.0453	3.04	960	0.2399	13.9130
0.0248	3.13	990	0.2456	11.4139
0.0246	3.23	1020	0.2553	11.3936
0.0229	3.32	1050	0.2477	11.3878
0.0198	3.42	1080	0.2486	10.2137
0.0229	3.51	1110	0.2491	10.2514
0.021	3.61	1140	0.2478	10.4311
0.0212	3.7	1170	0.2482	10.4398
0.0218	3.8	1200	0.2474	10.8892
0.0225	3.89	1230	0.2442	10.3731
0.0209	3.99	1260	0.2439	10.5326
0.0105	4.08	1290	0.2544	10.2948
0.0089	4.18	1320	0.2614	10.3238
0.008	4.27	1350	0.2627	10.0223
0.0079	4.37	1380	0.2652	10.2456
0.0083	4.46	1410	0.2646	10.1267
0.0073	4.56	1440	0.2619	10.0136
0.0075	4.65	1470	0.2633	9.7266
0.0068	4.75	1500	0.2648	9.8281
0.0074	4.84	1530	0.2645	9.8194
0.0079	4.94	1560	0.2647	9.9296

Framework versions

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