finetuned-Accident-Video-subsetv2

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: 0.4718
• Accuracy: 0.9032

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: 180

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.03	6	1.6175	0.08
1.5831	1.03	12	1.3729	0.52
1.5831	2.03	18	1.1954	0.52
1.3037	3.03	24	1.1763	0.48
1.1833	4.03	30	1.1421	0.52
1.1833	5.03	36	1.1221	0.52
1.125	6.03	42	1.0469	0.68
1.125	7.03	48	1.0553	0.6
1.0967	8.03	54	0.9645	0.72
1.0025	9.03	60	0.9435	0.6
1.0025	10.03	66	0.8498	0.76
0.9207	11.03	72	0.8096	0.72
0.9207	12.03	78	0.8177	0.68
0.7994	13.03	84	0.7237	0.64
0.753	14.03	90	0.6827	0.72
0.753	15.03	96	0.5697	0.88
0.6998	16.03	102	0.7595	0.76
0.6998	17.03	108	0.5631	0.88
0.6564	18.03	114	0.5146	0.8
0.5929	19.03	120	0.7957	0.6
0.5929	20.03	126	0.5254	0.88
0.5148	21.03	132	0.4102	0.88
0.5148	22.03	138	0.4555	0.88
0.3351	23.03	144	0.7732	0.72
0.4489	24.03	150	0.4464	0.92
0.4489	25.03	156	0.3873	0.92
0.3652	26.03	162	0.3817	0.92
0.3652	27.03	168	0.3916	0.92
0.2532	28.03	174	0.3945	0.92
0.1712	29.03	180	0.3999	0.92

Framework versions

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