DeTr-TableDetection-5000-images

This model is a fine-tuned version of facebook/detr-resnet-50 on the table_detection_light dataset. It achieves the following results on the evaluation set:

  • Loss: 0.3184
  • Mean Iou: 0.0234

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: 16
  • eval_batch_size: 16
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 20

Training results

Training Loss Epoch Step Validation Loss Mean Iou
0.741 1.0 313 0.7054 0.0259
0.5559 2.0 626 0.5159 0.0231
0.4213 3.0 939 0.4154 0.0254
0.4374 4.0 1252 0.4072 0.0249
0.3884 5.0 1565 0.4454 0.0232
0.4057 6.0 1878 0.4251 0.0249
0.3511 7.0 2191 0.3882 0.0239
0.3463 8.0 2504 0.3766 0.0243
0.3346 9.0 2817 0.4142 0.0236
0.3183 10.0 3130 0.3804 0.0242
0.3049 11.0 3443 0.3642 0.0244
0.2942 12.0 3756 0.3541 0.0253
0.2836 13.0 4069 0.3359 0.0252
0.2738 14.0 4382 0.3338 0.0254
0.2629 15.0 4695 0.3318 0.0267
0.2591 16.0 5008 0.3311 0.0224
0.2457 17.0 5321 0.3317 0.0234
0.2406 18.0 5634 0.3219 0.0238
0.2383 19.0 5947 0.3143 0.0238
0.2229 20.0 6260 0.3184 0.0234

Framework versions

  • Transformers 4.26.0
  • Pytorch 1.13.1+cu117
  • Datasets 2.5.1
  • Tokenizers 0.13.2
Downloads last month
445
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.