vakyansh-wav2vec2-tamil-tam-250-audio-abuse-feature

This model is a fine-tuned version of Harveenchadha/vakyansh-wav2vec2-tamil-tam-250 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.6061
• Accuracy: 0.7412
• Macro F1-score: 0.6531

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
6.7458	0.77	10	6.7472	0.0	0.0
6.7056	1.54	20	6.6488	0.0	0.0
6.6158	2.31	30	6.5180	0.6307	0.0917
6.4651	3.08	40	6.2887	0.7143	0.4167
6.2508	3.85	50	5.9094	0.7197	0.4185
5.8959	4.62	60	5.5362	0.7197	0.4185
5.6179	5.38	70	5.2347	0.7197	0.4185
5.3048	6.15	80	4.9823	0.7197	0.4185
5.0858	6.92	90	4.7555	0.7197	0.4185
4.9195	7.69	100	4.5424	0.7197	0.4185
4.6747	8.46	110	4.3265	0.7197	0.4185
4.5861	9.23	120	4.1193	0.7197	0.4185
4.3397	10.0	130	3.9070	0.7197	0.4185
4.0926	10.77	140	3.6954	0.7197	0.4185
3.8859	11.54	150	3.4822	0.7197	0.4185
3.7254	12.31	160	3.2711	0.7197	0.4185
3.5303	13.08	170	3.0599	0.7197	0.4185
3.2531	13.85	180	2.8502	0.7197	0.4185
3.0184	14.62	190	2.6448	0.7197	0.4185
3.0006	15.38	200	2.4472	0.7197	0.4185
2.6674	16.15	210	2.2526	0.7197	0.4185
2.4455	16.92	220	2.0649	0.7197	0.4185
2.2702	17.69	230	1.8883	0.7197	0.4185
2.0536	18.46	240	1.7233	0.7197	0.4185
2.0643	19.23	250	1.5730	0.7197	0.4185
1.8006	20.0	260	1.4368	0.7197	0.4185
1.6975	20.77	270	1.3112	0.7197	0.4185
1.4407	21.54	280	1.2015	0.7197	0.4185
1.2971	22.31	290	1.1050	0.7197	0.4185
1.3202	23.08	300	1.0219	0.7197	0.4185
1.1292	23.85	310	0.9490	0.7197	0.4185
1.1055	24.62	320	0.8879	0.7197	0.4185
0.9817	25.38	330	0.8366	0.7197	0.4185
0.9296	26.15	340	0.7906	0.7197	0.4185
0.8306	26.92	350	0.7506	0.7197	0.4185
0.8303	27.69	360	0.7171	0.7197	0.4185
0.8421	28.46	370	0.6953	0.7197	0.4185
0.7964	29.23	380	0.6650	0.7197	0.4185
0.7528	30.0	390	0.6470	0.7197	0.4185
0.7305	30.77	400	0.6345	0.7197	0.4185
0.6702	31.54	410	0.6163	0.7385	0.4937
0.6416	32.31	420	0.6118	0.7547	0.5507
0.608	33.08	430	0.6086	0.7547	0.5507
0.6659	33.85	440	0.5981	0.7574	0.5949
0.5839	34.62	450	0.6068	0.7547	0.6570
0.6167	35.38	460	0.5894	0.7763	0.6479
0.5991	36.15	470	0.5947	0.7412	0.6531
0.5839	36.92	480	0.5938	0.7574	0.6771
0.5533	37.69	490	0.5922	0.7520	0.6399
0.4998	38.46	500	0.6203	0.7358	0.6625
0.5508	39.23	510	0.5865	0.7493	0.6278
0.5159	40.0	520	0.5963	0.7385	0.6670
0.5344	40.77	530	0.5946	0.7439	0.6420
0.5039	41.54	540	0.5979	0.7466	0.6526
0.5456	42.31	550	0.5999	0.7358	0.6707
0.4822	43.08	560	0.5845	0.7493	0.6437
0.4864	43.85	570	0.6035	0.7439	0.6779
0.4623	44.62	580	0.5961	0.7520	0.6519
0.475	45.38	590	0.6066	0.7439	0.6651
0.4887	46.15	600	0.6014	0.7466	0.6603
0.506	46.92	610	0.6012	0.7412	0.6604
0.5296	47.69	620	0.5986	0.7439	0.6503
0.5255	48.46	630	0.6003	0.7439	0.6503
0.4667	49.23	640	0.6038	0.7466	0.6553
0.4334	50.0	650	0.6061	0.7412	0.6531

Framework versions

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