wav2vec2-classifier-aug

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: 0.5202
• Accuracy: 0.8679
• Precision: 0.8908
• Recall: 0.8679
• F1: 0.8667
• Binary: 0.9067

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
• lr_scheduler_warmup_steps: 500
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1	Binary
No log	0.19	50	4.4234	0.0162	0.0017	0.0162	0.0031	0.1558
No log	0.38	100	4.2882	0.0350	0.0028	0.0350	0.0047	0.3075
No log	0.58	150	3.9749	0.0404	0.0024	0.0404	0.0043	0.3170
No log	0.77	200	3.7072	0.0458	0.0070	0.0458	0.0109	0.3296
No log	0.96	250	3.4794	0.0836	0.0233	0.0836	0.0217	0.3580
4.1218	1.15	300	3.2647	0.1321	0.0526	0.1321	0.0640	0.3930
4.1218	1.34	350	3.0118	0.2318	0.1558	0.2318	0.1503	0.4623
4.1218	1.53	400	2.7772	0.2642	0.1570	0.2642	0.1752	0.4849
4.1218	1.73	450	2.5522	0.3585	0.3222	0.3585	0.2848	0.5520
4.1218	1.92	500	2.3428	0.3342	0.2725	0.3342	0.2563	0.5372
3.1065	2.11	550	2.0580	0.4124	0.3332	0.4124	0.3326	0.5887
3.1065	2.3	600	1.8454	0.4771	0.4322	0.4771	0.4131	0.6323
3.1065	2.49	650	1.6830	0.5310	0.4926	0.5310	0.4771	0.6733
3.1065	2.68	700	1.5545	0.5580	0.5326	0.5580	0.5096	0.6898
3.1065	2.88	750	1.3593	0.6253	0.5975	0.6253	0.5812	0.7388
2.2273	3.07	800	1.2047	0.6927	0.6715	0.6927	0.6535	0.7849
2.2273	3.26	850	1.1223	0.6765	0.6662	0.6765	0.6461	0.7728
2.2273	3.45	900	1.0296	0.7062	0.7121	0.7062	0.6756	0.7943
2.2273	3.64	950	1.0001	0.7251	0.7388	0.7251	0.7074	0.8081
2.2273	3.84	1000	0.9879	0.7466	0.7650	0.7466	0.7265	0.8229
1.734	4.03	1050	0.9078	0.7466	0.7590	0.7466	0.7323	0.8237
1.734	4.22	1100	0.8344	0.7898	0.8284	0.7898	0.7794	0.8550
1.734	4.41	1150	0.8199	0.7925	0.8029	0.7925	0.7749	0.8558
1.734	4.6	1200	0.7227	0.7951	0.8309	0.7951	0.7892	0.8566
1.734	4.79	1250	0.7666	0.7871	0.8246	0.7871	0.7768	0.8520
1.734	4.99	1300	0.7529	0.7871	0.7989	0.7871	0.7768	0.8531
1.4492	5.18	1350	0.7035	0.8032	0.8287	0.8032	0.7986	0.8633
1.4492	5.37	1400	0.6597	0.8194	0.8522	0.8194	0.8141	0.8739
1.4492	5.56	1450	0.6592	0.8113	0.8472	0.8113	0.8108	0.8690
1.4492	5.75	1500	0.6535	0.8248	0.8547	0.8248	0.8203	0.8784
1.4492	5.94	1550	0.6343	0.8167	0.8568	0.8167	0.8116	0.8701
1.2533	6.14	1600	0.5640	0.8356	0.8589	0.8356	0.8329	0.8860
1.2533	6.33	1650	0.5465	0.8383	0.8669	0.8383	0.8341	0.8889
1.2533	6.52	1700	0.5594	0.8248	0.8549	0.8248	0.8204	0.8776
1.2533	6.71	1750	0.5765	0.8464	0.8776	0.8464	0.8463	0.8935
1.2533	6.9	1800	0.5169	0.8571	0.8758	0.8571	0.8543	0.9000
1.138	7.09	1850	0.5206	0.8410	0.8676	0.8410	0.8421	0.8887
1.138	7.29	1900	0.5258	0.8544	0.8779	0.8544	0.8537	0.8992
1.138	7.48	1950	0.5855	0.8383	0.8693	0.8383	0.8384	0.8879
1.138	7.67	2000	0.5209	0.8491	0.8800	0.8491	0.8493	0.8943
1.138	7.86	2050	0.5150	0.8410	0.8710	0.8410	0.8411	0.8889
1.0249	8.05	2100	0.4937	0.8571	0.8840	0.8571	0.8568	0.9022
1.0249	8.25	2150	0.5344	0.8518	0.8790	0.8518	0.8492	0.8995
1.0249	8.44	2200	0.5322	0.8437	0.8751	0.8437	0.8428	0.8927
1.0249	8.63	2250	0.5533	0.8248	0.8561	0.8248	0.8233	0.8774
1.0249	8.82	2300	0.5242	0.8491	0.8797	0.8491	0.8469	0.8943
0.9523	9.01	2350	0.4938	0.8679	0.8911	0.8679	0.8669	0.9075
0.9523	9.2	2400	0.5037	0.8625	0.8888	0.8625	0.8627	0.9038
0.9523	9.4	2450	0.4973	0.8571	0.8794	0.8571	0.8565	0.9000
0.9523	9.59	2500	0.5343	0.8383	0.8705	0.8383	0.8384	0.8868
0.9523	9.78	2550	0.5493	0.8491	0.8746	0.8491	0.8472	0.8943
0.9523	9.97	2600	0.5226	0.8544	0.8783	0.8544	0.8537	0.8981
0.8792	10.16	2650	0.4883	0.8625	0.8857	0.8625	0.8598	0.9038
0.8792	10.35	2700	0.5178	0.8518	0.8784	0.8518	0.8503	0.8962
0.8792	10.55	2750	0.6273	0.8383	0.8756	0.8383	0.8363	0.8879
0.8792	10.74	2800	0.5229	0.8571	0.8855	0.8571	0.8576	0.9000
0.8792	10.93	2850	0.4617	0.8706	0.8924	0.8706	0.8686	0.9094
0.8251	11.12	2900	0.5764	0.8625	0.8874	0.8625	0.8626	0.9038
0.8251	11.31	2950	0.5111	0.8706	0.8960	0.8706	0.8689	0.9094
0.8251	11.51	3000	0.6013	0.8437	0.8603	0.8437	0.8410	0.8906
0.8251	11.7	3050	0.5968	0.8437	0.8682	0.8437	0.8405	0.8916
0.8251	11.89	3100	0.5467	0.8544	0.8806	0.8544	0.8542	0.8981
0.7578	12.08	3150	0.6015	0.8544	0.8774	0.8544	0.8523	0.8981
0.7578	12.27	3200	0.4897	0.8679	0.8814	0.8679	0.8632	0.9067
0.7578	12.46	3250	0.5395	0.8491	0.8765	0.8491	0.8460	0.8935
0.7578	12.66	3300	0.5873	0.8491	0.8767	0.8491	0.8489	0.8935
0.7578	12.85	3350	0.5386	0.8491	0.8735	0.8491	0.8498	0.8935
0.7295	13.04	3400	0.5826	0.8652	0.8949	0.8652	0.8663	0.9057
0.7295	13.23	3450	0.5358	0.8571	0.8859	0.8571	0.8562	0.9003
0.7295	13.42	3500	0.4802	0.8841	0.9017	0.8841	0.8838	0.9173
0.7295	13.61	3550	0.5709	0.8410	0.8692	0.8410	0.8404	0.8879
0.7295	13.81	3600	0.5420	0.8544	0.8738	0.8544	0.8535	0.8992
0.7295	14.0	3650	0.5384	0.8652	0.8817	0.8652	0.8635	0.9049
0.6874	14.19	3700	0.4911	0.8598	0.8753	0.8598	0.8593	0.9019
0.6874	14.38	3750	0.5172	0.8598	0.8826	0.8598	0.8588	0.9011
0.6874	14.57	3800	0.5024	0.8598	0.8814	0.8598	0.8592	0.9019
0.6874	14.77	3850	0.5202	0.8679	0.8908	0.8679	0.8667	0.9067

Framework versions

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