videomae-finetuned-nba-5-class-4-batch-8000-vid-multiclass-4

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: 0.7974
• F1: 0.8701
• Accuracy: 0.8701

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: 1.5e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• training_steps: 50000

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Accuracy
1.3802	0.04	2000	1.2381	0.52	0.52
1.0115	1.04	4000	1.0522	0.6684	0.6684
0.9749	2.04	6000	0.9298	0.7537	0.7537
0.9048	3.04	8000	0.8679	0.7863	0.7863
0.7977	4.04	10000	0.8846	0.7811	0.7811
0.9259	5.04	12000	0.8018	0.8263	0.8263
0.6077	6.04	14000	0.8212	0.8189	0.8189
0.7102	7.04	16000	0.7876	0.8242	0.8242
0.5726	8.04	18000	0.8805	0.8232	0.8232
0.7768	9.04	20000	0.7490	0.8589	0.8589
0.6793	10.04	22000	0.7730	0.8558	0.8558
0.5765	11.04	24000	0.7752	0.8368	0.8368
0.4789	12.04	26000	0.7902	0.8484	0.8484
0.7398	13.04	28000	0.7603	0.8568	0.8568
0.6807	14.04	30000	0.7531	0.8716	0.8716
0.3262	15.04	32000	0.7663	0.8768	0.8768
0.4387	16.04	34000	0.7549	0.88	0.88
0.5013	17.04	36000	0.7713	0.8737	0.8737
0.9572	18.04	38000	0.7613	0.8684	0.8684
1.0645	19.04	40000	0.7618	0.8811	0.8811
0.4949	20.04	42000	0.7882	0.8747	0.8747
0.6131	21.04	44000	0.7964	0.8705	0.8705
0.628	22.04	46000	0.8089	0.8747	0.8747
0.5693	23.04	48000	0.8010	0.8747	0.8747
0.4764	24.04	50000	0.8117	0.8789	0.8789

Framework versions

• Transformers 4.35.0
• Pytorch 2.1.0+cu118
• Datasets 2.14.6
• Tokenizers 0.14.1