finetuned-Accident-SingleLabel-Final-v2

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: 1.2768
• Accuracy: 0.5588

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
• distributed_type: multi-GPU
• num_devices: 2
• total_train_batch_size: 16
• total_eval_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• training_steps: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.08	4	1.8602	0.1304
No log	1.08	8	1.6211	0.4783
1.7588	2.08	12	1.4450	0.5652
1.7588	3.08	16	1.2476	0.6087
1.0616	4.08	20	1.1461	0.6087
1.0616	5.08	24	1.0131	0.5217
1.0616	6.08	28	0.9023	0.6087
0.9537	7.08	32	0.9932	0.6087
0.9537	8.08	36	0.9547	0.6087
0.7395	9.08	40	1.0174	0.6087
0.7395	10.08	44	0.9636	0.6087
0.7395	11.08	48	0.9022	0.5652
0.7531	12.04	50	0.9311	0.5652

Framework versions

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