VideoMAE_WLASL_2000_200_epochs_p20_SR_8_kinetics

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: 3.3164
• Top 1 Accuracy: 0.3721
• Top 5 Accuracy: 0.6987
• Top 10 Accuracy: 0.7911
• Accuracy: 0.3718
• Precision: 0.3566
• Recall: 0.3718
• F1: 0.3366

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• training_steps: 357200
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Top 1 Accuracy	Top 5 Accuracy	Top 10 Accuracy	Accuracy	Precision	Recall	F1
30.4419	0.005	1786	7.5999	0.0013	0.0041	0.0077	0.0013	0.0004	0.0013	0.0003
30.3231	1.0050	3572	7.5706	0.0018	0.0077	0.0117	0.0018	0.0007	0.0018	0.0006
29.5546	2.0050	5358	7.3846	0.0074	0.0240	0.0416	0.0074	0.0017	0.0074	0.0019
28.1321	3.0050	7145	7.0972	0.0240	0.0707	0.1093	0.0240	0.0088	0.0240	0.0087
26.7987	4.005	8931	6.7656	0.0465	0.1272	0.1913	0.0465	0.0170	0.0465	0.0184
24.7006	5.0050	10717	6.4074	0.0702	0.1982	0.2865	0.0702	0.0305	0.0702	0.0330
22.9951	6.0050	12503	6.0314	0.1034	0.2694	0.3685	0.1034	0.0511	0.1034	0.0538
21.0796	7.0050	14290	5.5934	0.1456	0.3455	0.4597	0.1456	0.0804	0.1456	0.0840
18.8279	8.005	16076	5.1535	0.1729	0.4091	0.5299	0.1729	0.1020	0.1729	0.1060
16.0168	9.0050	17862	4.6872	0.2270	0.4870	0.6124	0.2270	0.1483	0.2270	0.1535
13.4662	10.0050	19648	4.2174	0.2704	0.5585	0.6803	0.2707	0.1825	0.2707	0.1926
10.8825	11.0050	21435	3.7937	0.3159	0.6182	0.7288	0.3166	0.2341	0.3166	0.2405
8.493	12.005	23221	3.4529	0.3401	0.6619	0.7602	0.3407	0.2685	0.3407	0.2747
6.3864	13.0050	25007	3.1427	0.3631	0.6974	0.7852	0.3631	0.3058	0.3631	0.3063
4.8598	14.0050	26793	2.9818	0.3670	0.6997	0.8008	0.3672	0.3237	0.3672	0.3195
3.5894	15.0050	28580	2.7900	0.3925	0.7360	0.8207	0.3927	0.3493	0.3927	0.3457
2.7053	16.005	30366	2.7045	0.3920	0.7403	0.8297	0.3920	0.3579	0.3920	0.3486
2.0517	17.0050	32152	2.7339	0.3884	0.7344	0.8205	0.3879	0.3598	0.3879	0.3481
1.9862	18.0050	33938	2.7749	0.3820	0.7285	0.8172	0.3818	0.3614	0.3818	0.3446
1.9352	19.0050	35725	2.8157	0.3634	0.7135	0.8156	0.3634	0.3396	0.3634	0.3256
1.6389	20.005	37511	2.7968	0.3800	0.7247	0.8144	0.3800	0.3532	0.3800	0.3403
1.4166	21.0050	39297	2.8414	0.3739	0.7132	0.8067	0.3741	0.3498	0.3741	0.3358
1.3113	22.0050	41083	2.9111	0.3667	0.7033	0.8041	0.3670	0.3508	0.3670	0.3340
1.4698	23.0050	42870	2.9282	0.3675	0.7071	0.7947	0.3677	0.3466	0.3677	0.3296
1.1594	24.005	44656	2.9186	0.3899	0.7125	0.7988	0.3899	0.3662	0.3899	0.3532
0.8815	25.0050	46442	3.0210	0.3828	0.7053	0.7965	0.3825	0.3640	0.3825	0.3469
1.348	26.0050	48228	3.0267	0.3772	0.7074	0.7978	0.3772	0.3537	0.3772	0.3397
1.0531	27.0050	50015	3.0055	0.3805	0.7196	0.8108	0.3807	0.3656	0.3807	0.3464
1.2516	28.005	51801	3.1702	0.3634	0.6841	0.7883	0.3631	0.3353	0.3631	0.3244
1.2142	29.0050	53587	3.1537	0.3744	0.6948	0.7972	0.3746	0.3627	0.3746	0.3409
1.1783	30.0050	55373	3.2329	0.3659	0.6951	0.7916	0.3659	0.3400	0.3659	0.3278
1.2075	31.0050	57160	3.2251	0.3695	0.6971	0.7880	0.3693	0.3521	0.3693	0.3348
1.1369	32.005	58946	3.3422	0.3458	0.6803	0.7737	0.3460	0.3288	0.3460	0.3114
1.1948	33.0050	60732	3.3476	0.3634	0.6726	0.7735	0.3634	0.3375	0.3634	0.3240
1.0164	34.0050	62518	3.3188	0.3687	0.6969	0.7855	0.3687	0.3434	0.3687	0.3312
0.8987	35.0050	64305	3.3164	0.3721	0.6987	0.7911	0.3718	0.3566	0.3718	0.3366

Framework versions

• Transformers 4.46.1
• Pytorch 2.5.1+cu124
• Datasets 3.1.0
• Tokenizers 0.20.1