wav2vec2-classifier-aug-ref

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: 0.6490
• Accuracy: 0.8396
• Precision: 0.8518
• Recall: 0.8396
• F1: 0.8378
• Binary: 0.8887

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: 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	0.13	50	4.2247	0.0647	0.0139	0.0647	0.0177	0.3345
No log	0.27	100	3.9116	0.0930	0.0338	0.0930	0.0338	0.3598
No log	0.4	150	3.6537	0.1523	0.0800	0.1523	0.0856	0.4030
No log	0.54	200	3.4519	0.1860	0.1524	0.1860	0.1277	0.4245
No log	0.67	250	3.2675	0.3315	0.2378	0.3315	0.2500	0.5302
No log	0.81	300	3.0858	0.3450	0.2487	0.3450	0.2596	0.5395
No log	0.94	350	2.9341	0.3625	0.2613	0.3625	0.2730	0.5524
3.6847	1.08	400	2.7592	0.4461	0.3862	0.4461	0.3690	0.6132
3.6847	1.21	450	2.5895	0.5027	0.4694	0.5027	0.4387	0.6509
3.6847	1.35	500	2.4411	0.5566	0.5189	0.5566	0.4930	0.6887
3.6847	1.48	550	2.3212	0.5593	0.5286	0.5593	0.4985	0.6910
3.6847	1.62	600	2.1863	0.5903	0.5494	0.5903	0.5344	0.7135
3.6847	1.75	650	2.0742	0.6092	0.5808	0.6092	0.5618	0.7267
3.6847	1.89	700	1.9542	0.6442	0.6075	0.6442	0.5985	0.7512
2.5893	2.02	750	1.8513	0.6739	0.6664	0.6739	0.6306	0.7720
2.5893	2.16	800	1.7673	0.6806	0.6703	0.6806	0.6424	0.7755
2.5893	2.29	850	1.6589	0.7075	0.6837	0.7075	0.6696	0.7956
2.5893	2.43	900	1.5751	0.7035	0.6882	0.7035	0.6704	0.7933
2.5893	2.56	950	1.5010	0.7426	0.7286	0.7426	0.7164	0.8206
2.5893	2.7	1000	1.4422	0.7385	0.7346	0.7385	0.7169	0.8173
2.5893	2.83	1050	1.3884	0.7426	0.7328	0.7426	0.7170	0.8202
2.5893	2.97	1100	1.3253	0.7466	0.7319	0.7466	0.7218	0.8225
1.9357	3.1	1150	1.2850	0.7507	0.7492	0.7507	0.7297	0.8257
1.9357	3.24	1200	1.2297	0.7736	0.7781	0.7736	0.7541	0.8429
1.9357	3.37	1250	1.2131	0.7722	0.7738	0.7722	0.7528	0.8406
1.9357	3.51	1300	1.1359	0.7830	0.7835	0.7830	0.7652	0.8489
1.9357	3.64	1350	1.0756	0.8019	0.7958	0.8019	0.7870	0.8621
1.9357	3.78	1400	1.0650	0.7992	0.7994	0.7992	0.7826	0.8602
1.9357	3.91	1450	1.0384	0.7925	0.7841	0.7925	0.7731	0.8555
1.5532	4.05	1500	1.0125	0.7951	0.7957	0.7951	0.7794	0.8565
1.5532	4.18	1550	0.9956	0.7978	0.8071	0.7978	0.7844	0.8598
1.5532	4.32	1600	1.0085	0.7749	0.7802	0.7749	0.7600	0.8415
1.5532	4.45	1650	0.9397	0.7965	0.8091	0.7965	0.7850	0.8580
1.5532	4.59	1700	0.9449	0.7911	0.7945	0.7911	0.7751	0.8538
1.5532	4.72	1750	0.9208	0.7898	0.7909	0.7898	0.7731	0.8527
1.5532	4.86	1800	0.9147	0.7884	0.8127	0.7884	0.7797	0.8522
1.5532	4.99	1850	0.8418	0.8127	0.8136	0.8127	0.8020	0.8691
1.3035	5.12	1900	0.8513	0.8100	0.8227	0.8100	0.8033	0.8674
1.3035	5.26	1950	0.8372	0.8154	0.8232	0.8154	0.8088	0.8717
1.3035	5.39	2000	0.8166	0.8181	0.8246	0.8181	0.8102	0.8735
1.3035	5.53	2050	0.7987	0.8261	0.8414	0.8261	0.8208	0.8778
1.3035	5.66	2100	0.7924	0.8181	0.8347	0.8181	0.8143	0.8730
1.3035	5.8	2150	0.7732	0.8140	0.8273	0.8140	0.8092	0.8708
1.3035	5.93	2200	0.7636	0.8261	0.8410	0.8261	0.8222	0.8802
1.1281	6.07	2250	0.7663	0.8154	0.8275	0.8154	0.8070	0.8716
1.1281	6.2	2300	0.7494	0.8356	0.8498	0.8356	0.8305	0.8846
1.1281	6.34	2350	0.7347	0.8356	0.8466	0.8356	0.8329	0.8848
1.1281	6.47	2400	0.7434	0.8235	0.8391	0.8235	0.8212	0.8771
1.1281	6.61	2450	0.7393	0.8302	0.8422	0.8302	0.8248	0.8814
1.1281	6.74	2500	0.7178	0.8221	0.8383	0.8221	0.8173	0.8749
1.1281	6.88	2550	0.6919	0.8410	0.8559	0.8410	0.8385	0.8885
1.0069	7.01	2600	0.7236	0.8248	0.8435	0.8248	0.8213	0.8779
1.0069	7.15	2650	0.7048	0.8315	0.8474	0.8315	0.8301	0.8822
1.0069	7.28	2700	0.6997	0.8275	0.8417	0.8275	0.8243	0.8787
1.0069	7.42	2750	0.6953	0.8329	0.8505	0.8329	0.8316	0.8830
1.0069	7.55	2800	0.6893	0.8275	0.8410	0.8275	0.8255	0.8783
1.0069	7.69	2850	0.6927	0.8261	0.8404	0.8261	0.8245	0.8794
1.0069	7.82	2900	0.6865	0.8288	0.8436	0.8288	0.8264	0.8802
1.0069	7.96	2950	0.6795	0.8383	0.8523	0.8383	0.8373	0.8869
0.9224	8.09	3000	0.6662	0.8356	0.8469	0.8356	0.8343	0.8854
0.9224	8.23	3050	0.6768	0.8342	0.8487	0.8342	0.8336	0.8849
0.9224	8.36	3100	0.6751	0.8329	0.8454	0.8329	0.8321	0.8840
0.9224	8.5	3150	0.6766	0.8315	0.8421	0.8315	0.8301	0.8830
0.9224	8.63	3200	0.6634	0.8302	0.8393	0.8302	0.8283	0.8821
0.9224	8.77	3250	0.6624	0.8329	0.8437	0.8329	0.8310	0.8834
0.9224	8.9	3300	0.6615	0.8342	0.8478	0.8342	0.8325	0.8849
0.8806	9.04	3350	0.6619	0.8356	0.8485	0.8356	0.8345	0.8853
0.8806	9.17	3400	0.6459	0.8423	0.8557	0.8423	0.8411	0.8906
0.8806	9.31	3450	0.6463	0.8437	0.8565	0.8437	0.8427	0.8915
0.8806	9.44	3500	0.6529	0.8423	0.8532	0.8423	0.8403	0.8900
0.8806	9.58	3550	0.6525	0.8369	0.8489	0.8369	0.8352	0.8868
0.8806	9.71	3600	0.6544	0.8383	0.8487	0.8383	0.8363	0.8872
0.8806	9.84	3650	0.6494	0.8410	0.8528	0.8410	0.8394	0.8896
0.8806	9.98	3700	0.6490	0.8396	0.8518	0.8396	0.8378	0.8887

Framework versions

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