videomae-tiny-ssv2-binary-finetuned-xd-violence

This model is a fine-tuned version of MCG-NJU/videomae-base-finetuned-kinetics on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.6274
• Accuracy: 0.6535
• Precision: 0.72
• Recall: 0.5575
• F1: 0.6284
• Tp: 126
• Tn: 155
• Fp: 49
• Fn: 100
• Specificity: 0.7598
• Unsafe Precision At Default Threshold: 0.5561
• Unsafe Recall At Default Threshold: 0.9425
• Unsafe F1 At Default Threshold: 0.6995
• Unsafe Precision At Best Threshold: 0.5561
• Unsafe Recall At Best Threshold: 0.9425
• Unsafe Fbeta At Best Threshold: 0.8275
• Best Threshold: 0.25
• Roc Auc: 0.7261
• Average Precision: 0.7410

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1	Tp	Tn	Fp	Fn	Specificity	Unsafe Precision At Default Threshold	Unsafe Recall At Default Threshold	Unsafe F1 At Default Threshold	Unsafe Precision At Best Threshold	Unsafe Recall At Best Threshold	Unsafe Fbeta At Best Threshold	Best Threshold	Roc Auc	Average Precision
0.715	1.0	422	0.7038	0.5349	0.5428	0.7301	0.6226	165	65	139	61	0.3186	0.5293	1.0	0.6922	0.5293	1.0	0.8490	0.25	0.5644	0.5972
0.6737	2.0	844	0.6732	0.5860	0.6333	0.5044	0.5616	114	138	66	112	0.6765	0.5293	1.0	0.6922	0.5398	0.9912	0.8491	0.3	0.6353	0.6177
0.6639	3.0	1266	0.6413	0.6256	0.6923	0.5177	0.5924	117	152	52	109	0.7451	0.5396	0.9956	0.6998	0.5396	0.9956	0.8516	0.25	0.7109	0.7115
0.7136	4.0	1688	0.6424	0.6581	0.7283	0.5575	0.6316	126	157	47	100	0.7696	0.5305	1.0	0.6933	0.5437	0.9912	0.8511	0.3	0.7135	0.7143
0.6542	5.0	2110	0.6289	0.6767	0.6605	0.7920	0.7203	179	112	92	47	0.5490	0.5280	1.0	0.6911	0.5383	0.9956	0.8510	0.325	0.7326	0.7281
0.64	6.0	2532	0.6257	0.6605	0.7326	0.5575	0.6332	126	158	46	100	0.7745	0.5463	0.9912	0.7044	0.5463	0.9912	0.8524	0.25	0.7230	0.7415
0.5858	7.0	2954	0.6348	0.6581	0.7687	0.5	0.6059	113	170	34	113	0.8333	0.6109	0.8894	0.7243	0.6109	0.8894	0.8151	0.25	0.7312	0.7420
0.6228	8.0	3376	0.6220	0.6698	0.7234	0.6018	0.6570	136	152	52	90	0.7451	0.5606	0.9823	0.7138	0.5606	0.9823	0.8538	0.25	0.7356	0.7342
0.6999	9.0	3798	0.6204	0.6698	0.6944	0.6637	0.6787	150	138	66	76	0.6765	0.5343	1.0	0.6965	0.5450	0.9912	0.8517	0.3	0.7298	0.7371
0.6338	10.0	4220	0.6270	0.6791	0.6803	0.7345	0.7064	166	126	78	60	0.6176	0.5319	0.9956	0.6934	0.5493	0.9867	0.8511	0.325	0.7254	0.7254
0.6303	11.0	4642	0.6278	0.6744	0.6886	0.6947	0.6916	157	133	71	69	0.6520	0.5493	0.9867	0.7057	0.5493	0.9867	0.8511	0.25	0.7191	0.7243
0.5972	12.0	5064	0.6289	0.6442	0.6834	0.6018	0.64	136	141	63	90	0.6912	0.5479	0.9867	0.7046	0.5479	0.9867	0.8505	0.25	0.7192	0.7258
0.6028	13.0	5486	0.6168	0.6605	0.7128	0.5929	0.6473	134	150	54	92	0.7353	0.5461	0.9690	0.6986	0.5662	0.9646	0.8456	0.275	0.7372	0.7486
0.6118	14.0	5908	0.6310	0.6349	0.7197	0.5	0.5901	113	160	44	113	0.7843	0.5646	0.9469	0.7074	0.5646	0.9469	0.8340	0.25	0.7263	0.7392
0.6541	15.0	6330	0.6237	0.6419	0.7022	0.5531	0.6188	125	151	53	101	0.7402	0.5473	0.9735	0.7006	0.5651	0.9602	0.8424	0.275	0.7271	0.7365
0.5885	16.0	6752	0.6265	0.6628	0.7396	0.5531	0.6329	125	160	44	101	0.7843	0.5556	0.9735	0.7074	0.5556	0.9735	0.8462	0.25	0.7241	0.7418
0.5378	17.0	7174	0.6252	0.6512	0.7135	0.5619	0.6287	127	153	51	99	0.75	0.5561	0.9646	0.7055	0.5561	0.9646	0.8410	0.25	0.7244	0.7396
0.6367	18.0	7596	0.6248	0.6581	0.7135	0.5841	0.6423	132	151	53	94	0.7402	0.5553	0.9558	0.7024	0.5553	0.9558	0.8353	0.25	0.7268	0.7411
0.6296	19.0	8018	0.6264	0.6535	0.72	0.5575	0.6284	126	155	49	100	0.7598	0.5573	0.9469	0.7016	0.5573	0.9469	0.8307	0.25	0.7266	0.7412
0.6629	20.0	8440	0.6274	0.6535	0.72	0.5575	0.6284	126	155	49	100	0.7598	0.5561	0.9425	0.6995	0.5561	0.9425	0.8275	0.25	0.7261	0.7410

Framework versions

• Transformers 4.51.3
• Pytorch 2.1.0+cu118
• Datasets 3.6.0
• Tokenizers 0.21.4