finetuned-Accident-SingleLabel-Final-v3

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: 0.7942
• Accuracy: 0.6471

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: 65

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.06	4	1.8584	0.1739
No log	1.06	8	1.6548	0.3478
1.6685	2.06	12	1.4049	0.5217
1.6685	3.06	16	1.1540	0.6087
1.0202	4.06	20	1.1246	0.6087
1.0202	5.06	24	1.0258	0.4348
1.0202	6.06	28	0.9200	0.5217
0.9738	7.06	32	0.8942	0.6087
0.9738	8.06	36	0.8556	0.6087
0.7315	9.06	40	0.9506	0.6087
0.7315	10.06	44	0.9272	0.6087
0.7315	11.06	48	0.8048	0.5652
0.7004	12.06	52	0.8537	0.5217
0.7004	13.06	56	0.8058	0.6087
0.7426	14.06	60	1.0633	0.6957
0.7426	15.06	64	0.9449	0.6522
0.7426	16.02	65	0.8110	0.6522

Framework versions

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