hindi_base_wav2vec2-audio-abuse-feature

This model is a fine-tuned version of Harveenchadha/hindi_base_wav2vec2 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.7202
• Accuracy: 0.6694
• Macro F1-score: 0.6693

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Macro F1-score
6.6553	0.77	10	6.6322	0.0	0.0
6.5758	1.54	20	6.4417	0.5447	0.2151
6.3599	2.31	30	6.1486	0.5122	0.3621
6.0708	3.08	40	5.7751	0.5041	0.3351
5.8361	3.85	50	5.4662	0.5041	0.3351
5.5167	4.62	60	5.2127	0.5041	0.3351
5.289	5.38	70	4.9640	0.5041	0.3351
5.0266	6.15	80	4.7282	0.5041	0.3351
4.78	6.92	90	4.5006	0.5041	0.3351
4.6197	7.69	100	4.2787	0.5041	0.3351
4.3798	8.46	110	4.0506	0.5041	0.3351
4.2651	9.23	120	3.8315	0.5041	0.3351
3.9832	10.0	130	3.6034	0.5041	0.3351
3.7163	10.77	140	3.3782	0.5041	0.3351
3.5481	11.54	150	3.1510	0.5041	0.3351
3.305	12.31	160	2.9279	0.5041	0.3351
3.1589	13.08	170	2.7102	0.5041	0.3351
2.8368	13.85	180	2.4942	0.5041	0.3351
2.5875	14.62	190	2.2896	0.5041	0.3351
2.5938	15.38	200	2.0940	0.5041	0.3351
2.2346	16.15	210	1.9083	0.5041	0.3351
2.0404	16.92	220	1.7372	0.5041	0.3351
1.8744	17.69	230	1.5755	0.5041	0.3351
1.6581	18.46	240	1.4332	0.5041	0.3351
1.7251	19.23	250	1.3152	0.5041	0.3351
1.4569	20.0	260	1.2093	0.5041	0.3351
1.3718	20.77	270	1.1160	0.5041	0.3351
1.1743	21.54	280	1.0209	0.5041	0.3351
1.0744	22.31	290	0.9585	0.6585	0.6309
1.0933	23.08	300	0.8902	0.7019	0.6941
0.9348	23.85	310	0.8504	0.6992	0.6940
0.9611	24.62	320	0.8094	0.6911	0.6901
0.8307	25.38	330	0.7750	0.6992	0.6992
0.7863	26.15	340	0.7776	0.6802	0.6724
0.7431	26.92	350	0.7624	0.6829	0.6737
0.7607	27.69	360	0.7450	0.6775	0.6747
0.8054	28.46	370	0.7161	0.6938	0.6914
0.752	29.23	380	0.7021	0.6965	0.6946
0.72	30.0	390	0.7060	0.6856	0.6846
0.7252	30.77	400	0.6968	0.6911	0.6910
0.6497	31.54	410	0.7016	0.6911	0.6905
0.6215	32.31	420	0.7209	0.6856	0.6848
0.6143	33.08	430	0.6941	0.6856	0.6856
0.6778	33.85	440	0.6887	0.6856	0.6850
0.6027	34.62	450	0.7010	0.6992	0.6990
0.6644	35.38	460	0.7009	0.6721	0.6674
0.6178	36.15	470	0.6840	0.7019	0.6985
0.5817	36.92	480	0.6974	0.6829	0.6827
0.5876	37.69	490	0.6914	0.6802	0.6801
0.5474	38.46	500	0.7056	0.6856	0.6855
0.5327	39.23	510	0.7128	0.6802	0.6800
0.5648	40.0	520	0.7067	0.6748	0.6730
0.6163	40.77	530	0.6804	0.6721	0.6721
0.514	41.54	540	0.6965	0.6775	0.6774
0.5817	42.31	550	0.7177	0.6775	0.6767
0.5345	43.08	560	0.7136	0.6775	0.6772
0.525	43.85	570	0.7159	0.6883	0.6876
0.5043	44.62	580	0.7110	0.6802	0.6801
0.5418	45.38	590	0.7149	0.6748	0.6746
0.5129	46.15	600	0.7108	0.6694	0.6694
0.5331	46.92	610	0.7118	0.6667	0.6667
0.6061	47.69	620	0.7248	0.6802	0.6795
0.5551	48.46	630	0.7196	0.6694	0.6694
0.5049	49.23	640	0.7190	0.6640	0.6638
0.4663	50.0	650	0.7202	0.6694	0.6693

Framework versions

• Transformers 4.33.0
• Pytorch 2.0.0
• Datasets 2.1.0
• Tokenizers 0.13.3