videomae-base-finetuned-kinetics-fight_18-03-2024

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: 0.2211
• Accuracy: 0.9175
• Precision: 0.9372
• Recall: 0.895

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-07
• train_batch_size: 12
• eval_batch_size: 12
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• training_steps: 2660

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall
0.7318	0.05	134	0.7169	0.43	0.4247	0.395
0.6646	1.05	268	0.6636	0.59	0.6139	0.485
0.6089	2.05	402	0.5944	0.78	0.8415	0.69
0.5485	3.05	536	0.5270	0.845	0.8920	0.785
0.4581	4.05	670	0.4630	0.865	0.9011	0.82
0.3436	5.05	804	0.3994	0.8725	0.9162	0.82
0.3109	6.05	938	0.3530	0.8775	0.9171	0.83
0.2672	7.05	1072	0.3212	0.88	0.9176	0.835
0.2243	8.05	1206	0.2947	0.895	0.9202	0.865
0.297	9.05	1340	0.2779	0.895	0.9202	0.865
0.21	10.05	1474	0.2615	0.9025	0.9215	0.88
0.2003	11.05	1608	0.2515	0.895	0.9202	0.865
0.2128	12.05	1742	0.2418	0.91	0.9271	0.89
0.1789	13.05	1876	0.2357	0.9125	0.9275	0.895
0.1672	14.05	2010	0.2300	0.91	0.9227	0.895
0.1532	15.05	2144	0.2275	0.9175	0.9372	0.895
0.1695	16.05	2278	0.2241	0.9125	0.9275	0.895
0.1255	17.05	2412	0.2225	0.915	0.9323	0.895
0.1168	18.05	2546	0.2214	0.9175	0.9372	0.895
0.1395	19.04	2660	0.2211	0.9175	0.9372	0.895

Framework versions

• Transformers 4.38.2
• Pytorch 2.0.1+cu118
• Datasets 2.18.0
• Tokenizers 0.15.2