wav2vec2-large-xlsr-53-english-ser-linear

This model is a fine-tuned version of jonatasgrosman/wav2vec2-large-xlsr-53-english on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.4643
• Accuracy: 0.8587

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: 0.0001
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 1.5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.8767	0.01	10	1.8078	0.1684
1.7967	0.02	20	1.7544	0.2235
1.8173	0.02	30	1.7072	0.3032
1.7604	0.03	40	1.7162	0.2227
1.7271	0.04	50	1.6655	0.3032
1.764	0.05	60	1.5927	0.3599
1.55	0.06	70	1.5354	0.3657
1.5448	0.07	80	1.4057	0.4560
1.5118	0.07	90	1.3551	0.4733
1.354	0.08	100	1.2319	0.5596
1.3675	0.09	110	1.1786	0.5735
1.4058	0.1	120	1.0949	0.6105
1.1595	0.11	130	1.0964	0.5908
1.0444	0.12	140	1.1262	0.6212
1.0483	0.12	150	1.0863	0.5982
1.0439	0.13	160	1.0488	0.6491
1.0129	0.14	170	0.9045	0.6549
1.0171	0.15	180	1.0276	0.6270
1.0867	0.16	190	1.0888	0.6023
1.0646	0.16	200	0.9730	0.6311
1.0403	0.17	210	0.9315	0.6582
0.869	0.18	220	0.9686	0.6574
0.9193	0.19	230	0.9076	0.6960
1.0266	0.2	240	1.0796	0.6565
0.8563	0.21	250	1.0173	0.6426
0.8382	0.21	260	0.9155	0.6820
0.9275	0.22	270	0.9397	0.6689
0.9402	0.23	280	0.8919	0.6861
0.8636	0.24	290	0.9795	0.6680
1.2393	0.25	300	0.9872	0.6680
0.9537	0.25	310	0.8181	0.7247
0.7361	0.26	320	0.8470	0.7025
0.8452	0.27	330	0.8045	0.7198
0.9613	0.28	340	0.7530	0.7313
0.9335	0.29	350	0.9019	0.6902
0.9414	0.3	360	0.8981	0.6795
0.7473	0.3	370	0.7532	0.7321
0.8774	0.31	380	0.8953	0.7165
0.6989	0.32	390	0.7381	0.7387
0.9826	0.33	400	0.7128	0.7403
0.783	0.34	410	0.8292	0.6952
0.9668	0.35	420	0.7826	0.7239
0.7935	0.35	430	0.7081	0.7510
0.8284	0.36	440	0.7304	0.7264
0.9404	0.37	450	0.6761	0.7650
0.7735	0.38	460	0.6827	0.7469
0.6811	0.39	470	0.7926	0.7132
0.683	0.39	480	0.6883	0.7428
0.6779	0.4	490	0.6608	0.7486
0.6329	0.41	500	0.6578	0.7617
0.5824	0.42	510	0.7696	0.7420
0.6974	0.43	520	0.6755	0.7625
0.7716	0.44	530	0.6453	0.7716
0.7463	0.44	540	0.6644	0.7642
0.7993	0.45	550	0.6059	0.7864
0.606	0.46	560	0.6857	0.7461
0.8619	0.47	570	0.6570	0.7560
0.699	0.48	580	0.7400	0.7313
0.6619	0.49	590	0.7014	0.7494
0.7696	0.49	600	0.6621	0.7584
0.6544	0.5	610	0.6826	0.7650
0.5403	0.51	620	0.7464	0.7551
0.746	0.52	630	0.7323	0.7551
0.8129	0.53	640	0.7221	0.7634
0.7245	0.53	650	0.6306	0.7790
0.7062	0.54	660	0.6250	0.7896
0.741	0.55	670	0.6129	0.7938
0.7185	0.56	680	0.6332	0.7847
0.7706	0.57	690	0.5988	0.7954
0.8147	0.58	700	0.7032	0.7781
0.5144	0.58	710	0.6849	0.7634
0.9247	0.59	720	0.6088	0.7749
0.629	0.6	730	0.6393	0.7806
0.5908	0.61	740	0.5696	0.7913
0.4951	0.62	750	0.6370	0.7765
0.6358	0.62	760	0.6232	0.7979
0.6396	0.63	770	0.6707	0.7905
0.6947	0.64	780	0.6981	0.7683
0.6748	0.65	790	0.6761	0.7765
0.5607	0.66	800	0.6551	0.7921
0.6991	0.67	810	0.6134	0.7905
0.5793	0.67	820	0.5633	0.8118
0.4755	0.68	830	0.6031	0.7929
0.7645	0.69	840	0.5896	0.7962
0.742	0.7	850	0.5811	0.8036
0.5281	0.71	860	0.6449	0.7855
0.722	0.72	870	0.6593	0.7765
0.8174	0.72	880	0.5410	0.8003
0.5373	0.73	890	0.5802	0.7954
0.3868	0.74	900	0.6015	0.7954
0.5459	0.75	910	0.5485	0.7970
0.4629	0.76	920	0.6961	0.7584
0.6952	0.76	930	0.5608	0.8053
0.8452	0.77	940	0.5649	0.8044
0.6026	0.78	950	0.5330	0.8127
0.5131	0.79	960	0.5971	0.7888
0.6814	0.8	970	0.5594	0.8061
0.6001	0.81	980	0.5851	0.7954
0.5367	0.81	990	0.5716	0.8003
0.8356	0.82	1000	0.6519	0.7683
0.502	0.83	1010	0.6180	0.7749
0.5343	0.84	1020	0.5377	0.8053
0.5288	0.85	1030	0.5902	0.7962
0.5786	0.86	1040	0.6221	0.7905
0.6272	0.86	1050	0.6688	0.7831
0.5105	0.87	1060	0.6209	0.7880
0.5806	0.88	1070	0.6145	0.7929
0.5805	0.89	1080	0.6150	0.7847
0.4812	0.9	1090	0.5812	0.8061
0.5558	0.9	1100	0.6388	0.8044
0.7507	0.91	1110	0.5873	0.8044
0.7217	0.92	1120	0.5404	0.8085
0.8146	0.93	1130	0.5449	0.8003
0.6112	0.94	1140	0.5038	0.8151
0.7305	0.95	1150	0.4767	0.8316
0.3422	0.95	1160	0.5178	0.8127
0.4644	0.96	1170	0.5073	0.8200
0.4664	0.97	1180	0.4988	0.8184
0.6223	0.98	1190	0.5120	0.8283
0.6961	0.99	1200	0.5217	0.8118
0.6706	1.0	1210	0.5235	0.8094
0.3899	1.0	1220	0.5085	0.8184
0.418	1.01	1230	0.5171	0.8135
0.5011	1.02	1240	0.5056	0.8217
0.2969	1.03	1250	0.5209	0.8217
0.5093	1.04	1260	0.4921	0.8348
0.5167	1.04	1270	0.5081	0.8274
0.6382	1.05	1280	0.4851	0.8291
0.3493	1.06	1290	0.4946	0.8324
0.3471	1.07	1300	0.5122	0.8299
0.452	1.08	1310	0.5592	0.8291
0.4362	1.09	1320	0.5528	0.8266
0.4224	1.09	1330	0.5571	0.8192
0.333	1.1	1340	0.5714	0.8110
0.2944	1.11	1350	0.5156	0.8299
0.4004	1.12	1360	0.5208	0.8340
0.6824	1.13	1370	0.5426	0.8258
0.3746	1.13	1380	0.4902	0.8365
0.3679	1.14	1390	0.4868	0.8373
0.5009	1.15	1400	0.5192	0.8283
0.5577	1.16	1410	0.4937	0.8316
0.2566	1.17	1420	0.5043	0.8250
0.6625	1.18	1430	0.5416	0.8209
0.3251	1.18	1440	0.5146	0.8291
0.4306	1.19	1450	0.5313	0.8266
0.3159	1.2	1460	0.5308	0.8291
0.3598	1.21	1470	0.4869	0.8439
0.5498	1.22	1480	0.4670	0.8537
0.4947	1.23	1490	0.4928	0.8463
0.3948	1.23	1500	0.4816	0.8455
0.3137	1.24	1510	0.4755	0.8439
0.3525	1.25	1520	0.4972	0.8389
0.4821	1.26	1530	0.4954	0.8381
0.6099	1.27	1540	0.5096	0.8324
0.3172	1.27	1550	0.5029	0.8389
0.29	1.28	1560	0.4852	0.8455
0.288	1.29	1570	0.4916	0.8496
0.3771	1.3	1580	0.4734	0.8505
0.3106	1.31	1590	0.4746	0.8431
0.3494	1.32	1600	0.5069	0.8431
0.3183	1.32	1610	0.5155	0.8398
0.4353	1.33	1620	0.5242	0.8332
0.6207	1.34	1630	0.5161	0.8340
0.3241	1.35	1640	0.5037	0.8406
0.3646	1.36	1650	0.4890	0.8439
0.2341	1.37	1660	0.4884	0.8496
0.4874	1.37	1670	0.4688	0.8562
0.6701	1.38	1680	0.4589	0.8554
0.391	1.39	1690	0.4684	0.8537
0.3333	1.4	1700	0.4738	0.8513
0.2449	1.41	1710	0.4753	0.8488
0.361	1.41	1720	0.4946	0.8496
0.2229	1.42	1730	0.4971	0.8463
0.5915	1.43	1740	0.4904	0.8513
0.1812	1.44	1750	0.4782	0.8537
0.5887	1.45	1760	0.4702	0.8570
0.2823	1.46	1770	0.4665	0.8570
0.3397	1.46	1780	0.4673	0.8546
0.4727	1.47	1790	0.4638	0.8578
0.3303	1.48	1800	0.4636	0.8578
0.4544	1.49	1810	0.4646	0.8587
0.366	1.5	1820	0.4643	0.8587

Framework versions

• Transformers 4.40.0.dev0
• Pytorch 2.2.1+cu121
• Datasets 2.18.1.dev0
• Tokenizers 0.15.2