videomae-base-finetuned-soccer-action-recognitionx4

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: 0.1327
• Accuracy: 0.9648

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: 2e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• training_steps: 1125

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.02	28	1.3255	0.3782
No log	1.02	56	0.9903	0.5751
1.1078	2.02	84	0.4648	0.8238
1.1078	3.03	113	0.4075	0.8342
1.1078	4.02	141	0.2301	0.9275
0.2591	5.02	169	0.4090	0.8549
0.2591	6.02	197	0.1527	0.9482
0.2591	7.03	226	0.1413	0.9585
0.111	8.02	254	0.1386	0.9637
0.111	9.02	282	0.1674	0.9430
0.111	10.02	310	0.1010	0.9741
0.0509	11.03	339	0.1586	0.9637
0.0509	12.02	367	0.2696	0.8860
0.0509	13.02	395	0.3005	0.9171
0.023	14.02	423	0.1001	0.9741
0.023	15.03	452	0.1961	0.9482
0.0354	16.02	480	0.2596	0.9223
0.0354	17.02	508	0.1006	0.9689
0.0354	18.02	536	0.0947	0.9793
0.0141	19.03	565	0.0831	0.9793
0.0141	20.02	593	0.0685	0.9793
0.0141	21.02	621	0.0978	0.9689
0.0168	22.02	649	0.0812	0.9793
0.0168	23.03	678	0.0782	0.9793
0.0168	24.02	706	0.2624	0.9275
0.0029	25.02	734	0.1806	0.9534
0.0029	26.02	762	0.2123	0.9585
0.0029	27.03	791	0.2010	0.9482
0.0048	28.02	819	0.1259	0.9689
0.0048	29.02	847	0.2145	0.9430
0.0048	30.02	875	0.2901	0.9223
0.0042	31.03	904	0.0911	0.9793
0.0042	32.02	932	0.0892	0.9741
0.0024	33.02	960	0.0911	0.9845
0.0024	34.02	988	0.0987	0.9741
0.0024	35.03	1017	0.1446	0.9482
0.0008	36.02	1045	0.1415	0.9482
0.0008	37.02	1073	0.1435	0.9482
0.0008	38.02	1101	0.1438	0.9482
0.0008	39.02	1125	0.1371	0.9534

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu118
• Datasets 2.14.7
• Tokenizers 0.15.0