VideoMAE_Base_wlasl_100_longtail_200_signer

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

• Loss: 3.1722
• Accuracy: 0.4911

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: 36000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
18.6664	0.005	180	4.6583	0.0118
18.5992	1.0050	360	4.6339	0.0178
18.5069	2.0050	540	4.6268	0.0178
18.3375	3.0050	721	4.6243	0.0148
18.3756	4.005	901	4.6087	0.0148
18.261	5.0050	1081	4.6026	0.0296
18.2009	6.0050	1261	4.5961	0.0296
17.9622	7.0050	1442	4.6306	0.0207
17.8938	8.005	1622	4.5548	0.0325
17.5516	9.0050	1802	4.4639	0.0325
16.9895	10.0050	1982	4.3051	0.0473
16.1262	11.0050	2163	4.1849	0.0680
15.1416	12.005	2343	3.8819	0.0710
13.63	13.0050	2523	3.5929	0.1272
12.171	14.0050	2703	3.3547	0.1686
10.7993	15.0050	2884	3.2239	0.1953
9.247	16.005	3064	2.9288	0.2692
7.9993	17.0050	3244	2.7451	0.3254
6.7092	18.0050	3424	2.6186	0.3107
5.6476	19.0050	3605	2.5365	0.3432
4.5493	20.005	3785	2.5129	0.3669
3.5897	21.0050	3965	2.3543	0.4142
2.8892	22.0050	4145	2.3046	0.4290
2.1829	23.0050	4326	2.2836	0.4172
1.7981	24.005	4506	2.1694	0.4497
1.3646	25.0050	4686	2.1481	0.4645
1.1696	26.0050	4866	2.0963	0.5089
0.9916	27.0050	5047	2.1982	0.4556
0.838	28.005	5227	2.2522	0.4586
0.7322	29.0050	5407	2.3319	0.4675
0.6987	30.0050	5587	2.3509	0.4704
0.5202	31.0050	5768	2.4369	0.4497
0.4621	32.005	5948	2.2251	0.4763
0.3921	33.0050	6128	2.5757	0.4615
0.2631	34.0050	6308	2.4664	0.4822
0.2354	35.0050	6489	2.4789	0.5325
0.2631	36.005	6669	2.7400	0.4763
0.2948	37.0050	6849	2.6437	0.4704
0.2759	38.0050	7029	2.6592	0.5118
0.1302	39.0050	7210	2.8729	0.4911
0.3844	40.005	7390	2.7358	0.4941
0.1758	41.0050	7570	2.9512	0.4882
0.1795	42.0050	7750	2.8297	0.5089
0.1759	43.0050	7931	2.8388	0.4970
0.1972	44.005	8111	2.9518	0.4822
0.1698	45.0050	8291	2.9395	0.5059
0.2269	46.0050	8471	2.9545	0.4763
0.1931	47.0050	8652	3.0034	0.4970
0.2059	48.005	8832	2.8843	0.5178
0.2354	49.0050	9012	2.8678	0.5148
0.2905	50.0050	9192	2.8964	0.4882
0.269	51.0050	9373	3.1370	0.4793
0.2801	52.005	9553	2.8580	0.5
0.1666	53.0050	9733	3.3069	0.4645
0.2119	54.0050	9913	3.2520	0.4911
0.1588	55.0050	10094	3.1722	0.4911

Framework versions

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