vakyansh-wav2vec2-gujarati-gnm-100-audio-abuse-feature

This model is a fine-tuned version of Harveenchadha/vakyansh-wav2vec2-gujarati-gnm-100 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.6313
• Accuracy: 0.7403
• Macro F1-score: 0.6830

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
6.6694	0.77	10	6.6451	0.0387	0.0021
6.6244	1.54	20	6.5275	0.6878	0.0694
6.4955	2.31	30	6.2972	0.7044	0.4133
6.2586	3.08	40	5.9826	0.7044	0.4133
6.044	3.85	50	5.6760	0.7044	0.4133
5.7859	4.62	60	5.3680	0.7044	0.4133
5.506	5.38	70	5.0967	0.7044	0.4133
5.2115	6.15	80	4.8565	0.7044	0.4133
5.0439	6.92	90	4.6328	0.7044	0.4133
4.924	7.69	100	4.4207	0.7044	0.4133
4.5905	8.46	110	4.2046	0.7044	0.4133
4.4629	9.23	120	3.9881	0.7044	0.4133
4.2224	10.0	130	3.7741	0.7044	0.4133
4.0429	10.77	140	3.5620	0.7044	0.4133
3.8484	11.54	150	3.3434	0.7044	0.4133
3.6943	12.31	160	3.1294	0.7044	0.4133
3.4667	13.08	170	2.9148	0.7044	0.4133
3.1164	13.85	180	2.7000	0.7044	0.4133
2.9152	14.62	190	2.4912	0.7044	0.4133
2.7946	15.38	200	2.2933	0.7044	0.4133
2.5293	16.15	210	2.1013	0.7044	0.4133
2.3488	16.92	220	1.9167	0.7044	0.4133
2.2396	17.69	230	1.7418	0.7044	0.4133
2.0293	18.46	240	1.5833	0.7044	0.4133
1.8431	19.23	250	1.4364	0.7044	0.4133
1.6658	20.0	260	1.3038	0.7044	0.4133
1.5557	20.77	270	1.1904	0.7044	0.4133
1.3412	21.54	280	1.0912	0.7044	0.4133
1.2984	22.31	290	0.9999	0.7044	0.4133
1.2517	23.08	300	0.9240	0.7044	0.4133
1.2419	23.85	310	0.8693	0.7044	0.4133
1.0371	24.62	320	0.8206	0.7044	0.4133
0.922	25.38	330	0.7805	0.7044	0.4133
0.8833	26.15	340	0.7281	0.7044	0.4133
0.9064	26.92	350	0.6964	0.7210	0.4922
0.7483	27.69	360	0.6807	0.7569	0.6771
0.7677	28.46	370	0.6561	0.7762	0.6848
0.7107	29.23	380	0.6450	0.7486	0.6847
0.7144	30.0	390	0.6669	0.7182	0.6808
0.6656	30.77	400	0.6288	0.7486	0.6764
0.6896	31.54	410	0.6029	0.7652	0.6635
0.6715	32.31	420	0.6152	0.7486	0.7021
0.6375	33.08	430	0.6008	0.7597	0.6966
0.6342	33.85	440	0.5941	0.7652	0.6892
0.5992	34.62	450	0.6102	0.7459	0.6879
0.623	35.38	460	0.5906	0.7652	0.6914
0.5489	36.15	470	0.5970	0.7624	0.6610
0.5553	36.92	480	0.6324	0.7320	0.6902
0.5514	37.69	490	0.5974	0.7514	0.6852
0.5342	38.46	500	0.6077	0.7541	0.6954
0.5337	39.23	510	0.6081	0.7514	0.6872
0.4809	40.0	520	0.6685	0.6961	0.6572
0.4985	40.77	530	0.6262	0.7348	0.6798
0.4888	41.54	540	0.6358	0.7403	0.6773
0.4737	42.31	550	0.6137	0.7624	0.6911
0.5249	43.08	560	0.6456	0.7293	0.6784
0.5049	43.85	570	0.6503	0.7210	0.6694
0.4927	44.62	580	0.6294	0.7348	0.6663
0.4553	45.38	590	0.6130	0.7541	0.6835
0.4631	46.15	600	0.6524	0.7238	0.6718
0.5969	46.92	610	0.6233	0.7431	0.6817
0.4679	47.69	620	0.6306	0.7403	0.6848
0.4932	48.46	630	0.6245	0.7486	0.6922
0.4723	49.23	640	0.6304	0.7431	0.6872
0.4636	50.0	650	0.6313	0.7403	0.6830

Framework versions

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