golesheed/whisper-v2-EastFlemish

Whisper Large V2

This model is a fine-tuned version of openai/whisper-large-v2 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4356
• Wer: 17.9459

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
0.8564	0.0676	15	0.6152	41.0055
0.5012	0.1351	30	0.4878	48.5720
0.4816	0.2027	45	0.4620	32.3907
0.415	0.2703	60	0.4401	44.3115
0.3973	0.3378	75	0.4212	36.4375
0.3987	0.4054	90	0.4094	29.3866
0.4059	0.4730	105	0.3966	26.1762
0.4126	0.5405	120	0.3882	28.4659
0.3785	0.6081	135	0.3864	22.7352
0.3652	0.6757	150	0.3845	35.2448
0.4099	0.7432	165	0.3776	29.4185
0.4101	0.8108	180	0.3709	31.4269
0.352	0.8784	195	0.3687	24.2766
0.3604	0.9459	210	0.3648	21.7113
0.3642	1.0135	225	0.3622	21.3025
0.1693	1.0811	240	0.3698	27.0501
0.1622	1.1486	255	0.3728	27.2526
0.1862	1.2162	270	0.3562	21.5538
0.1815	1.2838	285	0.3647	23.6784
0.2084	1.3514	300	0.3611	20.9556
0.1777	1.4189	315	0.3610	22.6921
0.1842	1.4865	330	0.3591	22.3939
0.178	1.5541	345	0.3545	20.8168
0.1965	1.6216	360	0.3489	23.0390
0.1931	1.6892	375	0.3485	22.2176
0.1886	1.7568	390	0.3452	20.1268
0.1936	1.8243	405	0.3417	20.4849
0.19	1.8919	420	0.3474	22.2889
0.1818	1.9595	435	0.3449	22.0545
0.1445	2.0270	450	0.3605	19.4892
0.0879	2.0946	465	0.3753	19.5136
0.0921	2.1622	480	0.3722	19.8455
0.091	2.2297	495	0.3705	20.2955
0.0936	2.2973	510	0.3670	22.8215
0.0854	2.3649	525	0.3629	22.5327
0.0938	2.4324	540	0.3550	19.5061
0.0843	2.5	555	0.3674	21.2219
0.0879	2.5676	570	0.3599	18.8966
0.0802	2.6351	585	0.3668	18.1109
0.0868	2.7027	600	0.3563	18.5197
0.09	2.7703	615	0.3601	19.7236
0.0844	2.8378	630	0.3583	18.9829
0.0814	2.9054	645	0.3647	19.2135
0.0834	2.9730	660	0.3555	19.6298
0.0642	3.0405	675	0.3672	18.4484
0.0403	3.1081	690	0.4052	18.9698
0.0397	3.1757	705	0.3852	18.4747
0.0363	3.2432	720	0.3983	17.8334
0.0361	3.3108	735	0.3859	18.2965
0.0365	3.3784	750	0.3986	19.9486
0.032	3.4459	765	0.4001	18.6753
0.0374	3.5135	780	0.3902	18.3528
0.0337	3.5811	795	0.3980	18.4016
0.0327	3.6486	810	0.3962	17.6252
0.0357	3.7162	825	0.3935	18.5197
0.0359	3.7838	840	0.3877	18.3978
0.0342	3.8514	855	0.3870	18.9848
0.0368	3.9189	870	0.3939	17.9721
0.0341	3.9865	885	0.3928	18.1409
0.0164	4.0541	900	0.4037	17.4077
0.0141	4.1216	915	0.4316	18.3509
0.0113	4.1892	930	0.4305	16.9895
0.014	4.2568	945	0.4285	17.5071
0.0131	4.3243	960	0.4271	17.7621
0.0156	4.3919	975	0.4292	19.2998
0.0118	4.4595	990	0.4334	18.8704
0.0104	4.5270	1005	0.4332	17.7827
0.0107	4.5946	1020	0.4327	19.0148
0.009	4.6622	1035	0.4346	17.9084
0.0091	4.7297	1050	0.4384	17.7827
0.0107	4.7973	1065	0.4359	18.4203
0.0114	4.8649	1080	0.4348	17.8465
0.0092	4.9324	1095	0.4354	18.0415
0.0095	5.0	1110	0.4356	17.9459

Framework versions

• Transformers 4.45.0.dev0
• Pytorch 2.1.0+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1