videomae-small-kinetics-binary-finetuned-xd-violence

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

• Loss: 1.3045
• Accuracy: 0.7953
• Precision: 0.8165
• Recall: 0.7876
• F1: 0.8018
• Tp: 178
• Tn: 164
• Fp: 40
• Fn: 48
• Specificity: 0.8039
• Unsafe Precision At Default Threshold: 0.8165
• Unsafe Recall At Default Threshold: 0.7876
• Unsafe F1 At Default Threshold: 0.8018
• Unsafe Precision At Best Threshold: 0.8044
• Unsafe Recall At Best Threshold: 0.8009
• Unsafe Fbeta At Best Threshold: 0.8016
• Best Threshold: 0.25
• Roc Auc: 0.8711
• Average Precision: 0.8912

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
0.5397	1.0	422	0.5257	0.7488	0.8010	0.6947	0.7441	157	165	39	69	0.8088	0.8010	0.6947	0.7441	0.6319	0.9115	0.8374	0.25	0.8221	0.8398
0.5908	2.0	844	0.5147	0.7395	0.7568	0.7434	0.75	168	150	54	58	0.7353	0.7568	0.7434	0.75	0.6154	0.9558	0.8606	0.275	0.8225	0.8460
0.4996	3.0	1266	0.5176	0.7512	0.7117	0.8850	0.7890	200	123	81	26	0.6029	0.7117	0.8850	0.7890	0.6413	0.9336	0.8556	0.3	0.8465	0.8652
0.3455	4.0	1688	0.5220	0.7977	0.8492	0.7478	0.7953	169	174	30	57	0.8529	0.8492	0.7478	0.7953	0.7550	0.8319	0.8153	0.25	0.8596	0.8750
0.2528	5.0	2110	0.5808	0.7837	0.8122	0.7655	0.7882	173	164	40	53	0.8039	0.8122	0.7655	0.7882	0.7686	0.8230	0.8115	0.325	0.8490	0.8669
0.3747	6.0	2532	0.7050	0.7442	0.8580	0.6150	0.7165	139	181	23	87	0.8873	0.8580	0.6150	0.7165	0.8116	0.7434	0.7561	0.25	0.8412	0.8572
0.3213	7.0	2954	0.7711	0.7791	0.8466	0.7080	0.7711	160	175	29	66	0.8578	0.8466	0.7080	0.7711	0.7972	0.7655	0.7716	0.25	0.8502	0.8681
0.1319	8.0	3376	0.8325	0.7814	0.8173	0.7522	0.7834	170	166	38	56	0.8137	0.8173	0.7522	0.7834	0.7991	0.7920	0.7934	0.275	0.8650	0.8871
0.2057	9.0	3798	1.0621	0.7535	0.8061	0.6991	0.7488	158	166	38	68	0.8137	0.8061	0.6991	0.7488	0.7961	0.7257	0.7387	0.25	0.8411	0.8616
0.1623	10.0	4220	1.0641	0.7744	0.7633	0.8274	0.7941	187	146	58	39	0.7157	0.7633	0.8274	0.7941	0.7451	0.8407	0.8197	0.25	0.8538	0.8762
0.0726	11.0	4642	1.1368	0.7907	0.8009	0.8009	0.8009	181	159	45	45	0.7794	0.8009	0.8009	0.8009	0.7863	0.8142	0.8084	0.275	0.8705	0.8865
0.0914	12.0	5064	1.1482	0.7767	0.7778	0.8053	0.7913	182	152	52	44	0.7451	0.7778	0.8053	0.7913	0.7773	0.8186	0.8100	0.4	0.8627	0.8799
0.0512	13.0	5486	1.3861	0.7674	0.8119	0.7257	0.7664	164	166	38	62	0.8137	0.8119	0.7257	0.7664	0.8107	0.7389	0.7523	0.25	0.8592	0.8809
0.0047	14.0	5908	1.2770	0.7884	0.8082	0.7832	0.7955	177	162	42	49	0.7941	0.8082	0.7832	0.7955	0.7948	0.8053	0.8032	0.275	0.8550	0.8793
0.0858	15.0	6330	1.4003	0.7814	0.8204	0.7478	0.7824	169	167	37	57	0.8186	0.8204	0.7478	0.7824	0.8244	0.7478	0.7619	0.525	0.8622	0.8805
0.0316	16.0	6752	1.3339	0.7884	0.8199	0.7655	0.7918	173	166	38	53	0.8137	0.8199	0.7655	0.7918	0.8194	0.7832	0.7902	0.25	0.8654	0.8865
0.0005	17.0	7174	1.3088	0.7860	0.8131	0.7699	0.7909	174	164	40	52	0.8039	0.8131	0.7699	0.7909	0.8108	0.7965	0.7993	0.25	0.8712	0.8922
0.0003	18.0	7596	1.2977	0.7953	0.8108	0.7965	0.8036	180	162	42	46	0.7941	0.8108	0.7965	0.8036	0.8062	0.8097	0.8090	0.275	0.8708	0.8909
0.0032	19.0	8018	1.3034	0.7930	0.8128	0.7876	0.8	178	163	41	48	0.7990	0.8128	0.7876	0.8	0.8153	0.8009	0.8037	0.4	0.8708	0.8908
0.0002	20.0	8440	1.3045	0.7953	0.8165	0.7876	0.8018	178	164	40	48	0.8039	0.8165	0.7876	0.8018	0.8044	0.8009	0.8016	0.25	0.8711	0.8912

Framework versions

• Transformers 4.51.3
• Pytorch 2.5.1+cu121
• Datasets 3.6.0
• Tokenizers 0.21.4