wav2vec2-classifier-aug-large

This model is a fine-tuned version of facebook/wav2vec2-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.4814
• Accuracy: 0.8666
• Precision: 0.8790
• Recall: 0.8666
• F1: 0.8664
• Binary: 0.9089

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: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1	Binary
No log	0.1	50	3.8510	0.0606	0.0171	0.0606	0.0152	0.3357
No log	0.2	100	3.4088	0.0863	0.0528	0.0863	0.0366	0.3563
No log	0.29	150	3.1643	0.1105	0.0472	0.1105	0.0408	0.3740
No log	0.39	200	2.8623	0.2453	0.1221	0.2453	0.1447	0.4687
No log	0.49	250	2.5829	0.2763	0.1792	0.2763	0.1862	0.4920
No log	0.59	300	2.3355	0.3787	0.3219	0.3787	0.3034	0.5642
No log	0.69	350	2.0865	0.4744	0.4303	0.4744	0.4021	0.6301
No log	0.78	400	1.9045	0.5296	0.4587	0.5296	0.4649	0.6687
No log	0.88	450	1.6752	0.5472	0.5374	0.5472	0.4900	0.6829
No log	0.98	500	1.5273	0.6105	0.6019	0.6105	0.5709	0.7276
2.9976	1.08	550	1.3712	0.6536	0.6358	0.6536	0.6128	0.7567
2.9976	1.18	600	1.3239	0.6725	0.6797	0.6725	0.6389	0.7702
2.9976	1.27	650	1.1953	0.7170	0.7116	0.7170	0.6878	0.8024
2.9976	1.37	700	1.1213	0.7170	0.7116	0.7170	0.6834	0.8020
2.9976	1.47	750	1.0287	0.7291	0.7293	0.7291	0.7043	0.8106
2.9976	1.57	800	0.9258	0.7722	0.7818	0.7722	0.7572	0.8414
2.9976	1.67	850	0.8634	0.7722	0.7967	0.7722	0.7574	0.8415
2.9976	1.76	900	0.7849	0.7938	0.8226	0.7938	0.7874	0.8546
2.9976	1.86	950	0.8423	0.7601	0.7760	0.7601	0.7480	0.8321
2.9976	1.96	1000	0.7670	0.7830	0.8069	0.7830	0.7704	0.8488
1.686	2.06	1050	0.7352	0.7951	0.8024	0.7951	0.7832	0.8566
1.686	2.16	1100	0.7278	0.8019	0.8234	0.8019	0.7951	0.8627
1.686	2.25	1150	0.6867	0.8113	0.8241	0.8113	0.8059	0.8683
1.686	2.35	1200	0.6489	0.8167	0.8343	0.8167	0.8093	0.8722
1.686	2.45	1250	0.6217	0.8288	0.8454	0.8288	0.8242	0.8811
1.686	2.55	1300	0.6416	0.8113	0.8320	0.8113	0.8050	0.8678
1.686	2.65	1350	0.6517	0.8113	0.8254	0.8113	0.8055	0.8693
1.686	2.75	1400	0.6330	0.8140	0.8313	0.8140	0.8092	0.8710
1.686	2.84	1450	0.5905	0.8329	0.8575	0.8329	0.8339	0.8844
1.686	2.94	1500	0.5974	0.8329	0.8480	0.8329	0.8291	0.8840
1.2582	3.04	1550	0.6449	0.8235	0.8430	0.8235	0.8192	0.8774
1.2582	3.14	1600	0.5734	0.8464	0.8633	0.8464	0.8449	0.8933
1.2582	3.24	1650	0.5771	0.8450	0.8641	0.8450	0.8440	0.8910
1.2582	3.33	1700	0.5133	0.8491	0.8619	0.8491	0.8466	0.8942
1.2582	3.43	1750	0.5608	0.8437	0.8621	0.8437	0.8419	0.8906
1.2582	3.53	1800	0.6194	0.8221	0.8446	0.8221	0.8197	0.8759
1.2582	3.63	1850	0.5060	0.8410	0.8527	0.8410	0.8381	0.8899
1.2582	3.73	1900	0.6035	0.8315	0.8528	0.8315	0.8262	0.8829
1.2582	3.82	1950	0.5269	0.8396	0.8542	0.8396	0.8376	0.8891
1.2582	3.92	2000	0.5115	0.8531	0.8638	0.8531	0.8489	0.8982
1.0473	4.02	2050	0.5209	0.8518	0.8688	0.8518	0.8497	0.8969
1.0473	4.12	2100	0.5327	0.8342	0.8530	0.8342	0.8326	0.8844
1.0473	4.22	2150	0.4859	0.8544	0.8694	0.8544	0.8527	0.8980
1.0473	4.31	2200	0.5414	0.8450	0.8648	0.8450	0.8402	0.8918
1.0473	4.41	2250	0.5982	0.8383	0.8545	0.8383	0.8355	0.8871
1.0473	4.51	2300	0.5458	0.8450	0.8562	0.8450	0.8421	0.8934
1.0473	4.61	2350	0.5115	0.8625	0.8753	0.8625	0.8601	0.9042
1.0473	4.71	2400	0.5226	0.8518	0.8671	0.8518	0.8491	0.8961
1.0473	4.8	2450	0.5058	0.8679	0.8807	0.8679	0.8661	0.9082
1.0473	4.9	2500	0.5442	0.8491	0.8647	0.8491	0.8461	0.8957
1.0473	5.0	2550	0.4810	0.8693	0.8816	0.8693	0.8680	0.9089
0.9144	5.1	2600	0.4729	0.8787	0.8918	0.8787	0.8769	0.9159
0.9144	5.2	2650	0.4981	0.8585	0.8686	0.8585	0.8564	0.9019
0.9144	5.29	2700	0.5505	0.8477	0.8629	0.8477	0.8464	0.8937
0.9144	5.39	2750	0.4829	0.8706	0.8859	0.8706	0.8701	0.9111
0.9144	5.49	2800	0.5203	0.8544	0.8690	0.8544	0.8520	0.9003
0.9144	5.59	2850	0.4907	0.8585	0.8730	0.8585	0.8568	0.9027
0.9144	5.69	2900	0.4710	0.8706	0.8801	0.8706	0.8686	0.9096
0.9144	5.78	2950	0.5062	0.8504	0.8647	0.8504	0.8490	0.8965
0.9144	5.88	3000	0.4455	0.8774	0.8914	0.8774	0.8777	0.9163
0.9144	5.98	3050	0.5032	0.8544	0.8718	0.8544	0.8541	0.8985
0.8213	6.08	3100	0.4735	0.8733	0.8895	0.8733	0.8718	0.9135
0.8213	6.18	3150	0.4743	0.8693	0.8880	0.8693	0.8679	0.9102
0.8213	6.27	3200	0.5357	0.8531	0.8720	0.8531	0.8492	0.8984
0.8213	6.37	3250	0.4820	0.8625	0.8783	0.8625	0.8601	0.9059
0.8213	6.47	3300	0.4732	0.8760	0.8897	0.8760	0.8755	0.9159
0.8213	6.57	3350	0.4814	0.8666	0.8790	0.8666	0.8664	0.9089

Framework versions

• Transformers 4.38.2
• Pytorch 2.3.0
• Datasets 2.19.1
• Tokenizers 0.15.1