Edit model card

videomae-base-finetuned-good-gesturePhaseV5

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:

  • Accuracy: 0.6871
  • Loss: 0.9822
  • Accuracy Hold: 0.0
  • Accuracy Stroke: 0.0
  • Accuracy Recovery: 0.0
  • Accuracy Preparation: 0.9221
  • Accuracy Unknown: 0.8571

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: 1e-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: 1380

Training results

Training Loss Epoch Step Accuracy Validation Loss Accuracy Hold Accuracy Stroke Accuracy Recovery Accuracy Preparation Accuracy Unknown
1.1475 0.0507 70 0.5597 1.2558 0.0 0.0 0.0 1.0 0.0
1.2103 1.0507 140 0.5597 1.2704 0.0 0.0 0.0 1.0 0.0
0.9964 2.0507 210 0.5597 1.2142 0.0 0.0 0.0 1.0 0.0
0.9975 3.0507 280 0.5970 1.0747 0.0 0.0 0.0 0.9733 0.2692
1.0538 4.0507 350 0.6642 0.9622 0.0 0.0 0.0 0.88 0.8846
1.0321 5.0507 420 0.6567 0.9451 0.0 0.0 0.0 0.8533 0.9231
0.7822 6.0507 490 0.7164 0.8797 0.0 0.0 0.0833 0.96 0.8846
0.8743 7.0507 560 0.6791 0.9399 0.0 0.0 0.0833 0.8533 1.0
0.7515 8.0507 630 0.6791 0.9290 0.0 0.0 0.0 0.8667 1.0
0.8525 9.0507 700 0.7090 0.8447 0.0 0.0 0.1667 0.9467 0.8462
0.7661 10.0507 770 0.7090 0.7857 0.0 0.0 0.1667 0.9067 0.9615
0.8363 11.0507 840 0.6866 0.8165 0.0 0.0 0.0833 0.92 0.8462
0.659 12.0507 910 0.7164 0.7951 0.0 0.0 0.1667 0.9067 1.0
0.6274 13.0507 980 0.7015 0.7754 0.0 0.0 0.0833 0.8933 1.0
0.7292 14.0507 1050 0.6791 0.8128 0.0 0.0 0.25 0.8267 1.0
0.7447 15.0507 1120 0.6866 0.7860 0.0 0.0 0.25 0.84 1.0
0.5512 16.0507 1190 0.7015 0.7839 0.0625 0.0 0.1667 0.8667 1.0
0.3404 17.0507 1260 0.7015 0.8055 0.0 0.0 0.3333 0.8533 1.0
0.4406 18.0507 1330 0.6866 0.7800 0.0 0.0 0.1667 0.8533 1.0
0.6358 19.0362 1380 0.7015 0.7816 0.0 0.0 0.1667 0.88 1.0

Framework versions

  • Transformers 4.41.1
  • Pytorch 2.3.0+cu121
  • Datasets 2.19.1
  • Tokenizers 0.19.1
Downloads last month
0
Safetensors
Model size
86.2M params
Tensor type
F32
·
This model can be loaded on Inference API (serverless).

Finetuned from