videomae-base-finetuned-kinetics-finetuned-conflab-traj-direction-rh-v2

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: 1.7062
• Accuracy: 0.5512

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: 8
• eval_batch_size: 8
• 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: 1053

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.9728	0.1121	118	1.8082	0.2913
1.5903	1.1121	236	1.8038	0.3786
1.2371	2.1121	354	1.4667	0.4563
0.9635	3.1121	472	1.4187	0.5631
0.3984	4.1121	590	1.3990	0.5631
0.1962	5.1121	708	1.4243	0.5825
0.1952	6.1121	826	1.5571	0.6019
0.0319	7.1121	944	1.5844	0.5874
0.0203	8.1035	1053	1.5794	0.6214

Framework versions

• Transformers 4.41.0
• Pytorch 1.12.0+cu116
• Datasets 2.19.1
• Tokenizers 0.19.1