VideoMAE_Kinetics_wlasl_100_longtail_200_signer

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: 1.4501
• Accuracy: 0.6982

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
18.5248	0.005	180	4.6134	0.0148
18.4259	1.0050	360	4.5845	0.0355
18.0926	2.0050	540	4.4842	0.0444
16.9306	3.0050	721	4.2167	0.1302
15.4312	4.005	901	3.8332	0.2426
13.4099	5.0050	1081	3.4658	0.3195
11.4151	6.0050	1261	3.0617	0.4260
9.318	7.0050	1442	2.7168	0.5059
7.5007	8.005	1622	2.3707	0.5858
5.7294	9.0050	1802	2.0365	0.6450
4.2445	10.0050	1982	1.8353	0.6479
3.0008	11.0050	2163	1.5879	0.6686
2.1126	12.005	2343	1.4705	0.6805
1.4048	13.0050	2523	1.3585	0.6657
0.934	14.0050	2703	1.2650	0.7041
0.6172	15.0050	2884	1.1449	0.7367
0.4755	16.005	3064	1.1490	0.6805
0.3145	17.0050	3244	1.1220	0.7041
0.243	18.0050	3424	1.1176	0.7101
0.2305	19.0050	3605	1.1077	0.7071
0.1504	20.005	3785	1.1920	0.6982
0.1342	21.0050	3965	1.2274	0.7012
0.0744	22.0050	4145	1.2504	0.7101
0.1404	23.0050	4326	1.2245	0.7130
0.1019	24.005	4506	1.3000	0.7101
0.0746	25.0050	4686	1.2738	0.7041
0.1144	26.0050	4866	1.3395	0.6953
0.1238	27.0050	5047	1.1683	0.7071
0.0989	28.005	5227	1.3287	0.7071
0.097	29.0050	5407	1.5545	0.6775
0.1004	30.0050	5587	1.3614	0.7041
0.1062	31.0050	5768	1.5166	0.6923
0.1617	32.005	5948	1.3035	0.6923
0.1235	33.0050	6128	1.5919	0.6568
0.1395	34.0050	6308	1.4211	0.6746
0.109	35.0050	6489	1.4501	0.6982

Framework versions

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