finetuned-Accident-MultipleLabels-Video-subset-v2-checkpointing

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: 1.7371
• Accuracy: 0.3704

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
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• training_steps: 35

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.06	2	1.7265	0.3594
No log	1.06	4	1.6976	0.3906
No log	2.06	6	1.7503	0.3594
No log	3.06	8	1.8831	0.3125
1.7254	4.06	10	2.0285	0.1719
1.7254	5.06	12	2.0391	0.2812
1.7254	6.06	14	1.9737	0.3281
1.7254	7.06	16	1.8998	0.375
1.7254	8.06	18	1.8786	0.375
1.394	9.06	20	1.9054	0.3438
1.394	10.06	22	1.9474	0.3281
1.394	11.06	24	2.0032	0.3281
1.394	12.06	26	2.0729	0.3281
1.394	13.06	28	2.1081	0.3438
1.285	14.06	30	2.1190	0.3281
1.285	15.06	32	2.1188	0.3438
1.285	16.06	34	2.1155	0.3594
1.285	17.03	35	2.1163	0.3594

Framework versions

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