VideoMAE_base_wlasl_100__signer_200ep_coR

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: 3.6367
• Accuracy: 0.4793

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
18.6663	0.005	180	4.6585	0.0118
18.5985	1.0050	360	4.6345	0.0207
18.5062	2.0050	540	4.6301	0.0148
18.3482	3.0050	721	4.6212	0.0178
18.3963	4.005	901	4.6182	0.0207
18.2979	5.0050	1081	4.6092	0.0237
18.2053	6.0050	1261	4.6505	0.0178
18.0284	7.0050	1442	4.5878	0.0266
17.9774	8.005	1622	4.5875	0.0325
17.672	9.0050	1802	4.4892	0.0355
17.0746	10.0050	1982	4.3039	0.0503
16.2939	11.0050	2163	4.1299	0.0562
15.2755	12.005	2343	3.8815	0.1006
13.843	13.0050	2523	3.7165	0.1213
12.4153	14.0050	2703	3.4509	0.1657
10.9809	15.0050	2884	3.1755	0.2249
9.5775	16.005	3064	2.9489	0.2544
8.4151	17.0050	3244	2.8276	0.3195
7.0531	18.0050	3424	2.6719	0.3225
5.9795	19.0050	3605	2.6143	0.3491
4.8469	20.005	3785	2.6037	0.3521
3.9384	21.0050	3965	2.4071	0.3935
3.176	22.0050	4145	2.3498	0.4260
2.5533	23.0050	4326	2.2546	0.4290
2.0217	24.005	4506	2.2616	0.4408
1.5032	25.0050	4686	2.2962	0.4290
1.3052	26.0050	4866	2.3613	0.4024
1.0223	27.0050	5047	2.3479	0.4290
0.8638	28.005	5227	2.2887	0.4556
0.7623	29.0050	5407	2.3140	0.4497
0.688	30.0050	5587	2.5885	0.4201
0.4905	31.0050	5768	2.4310	0.4527
0.4466	32.005	5948	2.4134	0.4556
0.4291	33.0050	6128	2.3113	0.4734
0.3286	34.0050	6308	2.6219	0.4704
0.3314	35.0050	6489	2.6444	0.4556
0.2724	36.005	6669	2.6187	0.4704
0.3126	37.0050	6849	2.8826	0.4438
0.2677	38.0050	7029	2.7060	0.5059
0.1616	39.0050	7210	2.7237	0.5030
0.2657	40.005	7390	2.8089	0.4822
0.2641	41.0050	7570	2.8814	0.4941
0.0988	42.0050	7750	3.0111	0.5059
0.1412	43.0050	7931	3.0738	0.4763
0.2132	44.005	8111	3.1186	0.4704
0.2322	45.0050	8291	3.1673	0.4734
0.2135	46.0050	8471	3.0642	0.4793
0.2689	47.0050	8652	3.2337	0.4615
0.2157	48.005	8832	3.2357	0.4793
0.229	49.0050	9012	2.9704	0.5237
0.2051	50.0050	9192	3.1482	0.4763
0.2083	51.0050	9373	3.4135	0.4615
0.2298	52.005	9553	3.3165	0.5
0.1338	53.0050	9733	3.1129	0.4852
0.1476	54.0050	9913	3.2821	0.4734
0.2206	55.0050	10094	3.1688	0.5178
0.1698	56.005	10274	3.3487	0.4911
0.1568	57.0050	10454	3.0934	0.5059
0.1059	58.0050	10634	3.6909	0.4763
0.1377	59.0050	10815	3.4659	0.4882
0.194	60.005	10995	3.8766	0.4231
0.323	61.0050	11175	3.7657	0.4438
0.2116	62.0050	11355	3.5875	0.4911
0.3221	63.0050	11536	3.6262	0.4675
0.2123	64.005	11716	3.2548	0.4852
0.1485	65.0050	11896	3.2315	0.5266
0.1372	66.0050	12076	3.7162	0.4793
0.1303	67.0050	12257	3.5759	0.4911
0.1532	68.005	12437	3.4257	0.4882
0.1345	69.0050	12617	3.6885	0.4645
0.2182	70.0050	12797	3.8118	0.4734
0.2118	71.0050	12978	3.8203	0.5030
0.1931	72.005	13158	3.8547	0.4970
0.1849	73.0050	13338	3.5080	0.5237
0.2151	74.0050	13518	4.1261	0.4527
0.2144	75.0050	13699	3.4703	0.5118
0.2093	76.005	13879	3.5965	0.4911
0.1171	77.0050	14059	3.6071	0.5118
0.1377	78.0050	14239	3.6424	0.4970
0.1658	79.0050	14420	3.7843	0.4793
0.2543	80.005	14600	3.7443	0.4970
0.1849	81.0050	14780	3.9182	0.4941
0.1168	82.0050	14960	3.9852	0.4497
0.1406	83.0050	15141	3.7811	0.4970
0.1608	84.005	15321	3.8168	0.4793
0.1348	85.0050	15501	3.6367	0.4793

Framework versions

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