VOOZH about

URL: https://huggingface.co/merve/rfdetr-docvqa-media3-v2-agree1-large

⇱ merve/rfdetr-docvqa-media3-v2-agree1-large · Hugging Face

rfdetr-docvqa-media3-v2-agree1-large

This model is a fine-tuned version of Roboflow/rf-detr-large on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 16.0585
  • Map: 0.0038
  • Map 50: 0.0192
  • Map 75: 0.0015
  • Map Small: -1.0
  • Map Medium: 0.0004
  • Map Large: 0.004
  • Mar 1: 0.0117
  • Mar 10: 0.1231
  • Mar 100: 0.3169
  • Mar Small: -1.0
  • Mar Medium: 0.05
  • Mar Large: 0.3313
  • Map Chart: 0.005
  • Mar 100 Chart: 0.025
  • Map Image: 0.0045
  • Mar 100 Image: 0.5917
  • Map Signature: 0.0019
  • Mar 100 Signature: 0.334

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
  • optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_steps: 0.05
  • num_epochs: 10

Training results

Training Loss Epoch Step Validation Loss Map Map 50 Map 75 Map Small Map Medium Map Large Mar 1 Mar 10 Mar 100 Mar Small Mar Medium Mar Large Map Chart Mar 100 Chart Map Image Mar 100 Image Map Signature Mar 100 Signature
30.0081 1.0 50 29.0689 0.0008 0.0049 0.0 -1.0 0.0 0.0008 0.0067 0.0237 0.0876 -1.0 0.0 0.0881 0.0023 0.21 0.0001 0.0528 0.0 0.0
17.4070 2.0 100 26.6108 0.0004 0.0018 0.0 -1.0 0.0 0.0004 0.0033 0.0217 0.0648 -1.0 0.0 0.0657 0.0005 0.105 0.0005 0.0833 0.0001 0.006
14.5097 3.0 150 23.4111 0.0023 0.0079 0.0 -1.0 0.0 0.0023 0.0054 0.0185 0.084 -1.0 0.0 0.0875 0.0019 0.025 0.0048 0.1611 0.0001 0.066
14.1711 4.0 200 21.7587 0.0002 0.0007 0.0001 -1.0 0.0 0.0002 0.0009 0.0259 0.0924 -1.0 0.0 0.0956 0.0 0.0 0.0006 0.2472 0.0 0.03
13.5616 5.0 250 19.2226 0.002 0.0063 0.0004 -1.0 0.0 0.0021 0.0191 0.0537 0.2058 -1.0 0.0 0.2153 0.0039 0.03 0.0012 0.3833 0.001 0.204
12.5562 6.0 300 17.5439 0.0005 0.0015 0.0003 -1.0 0.0 0.0005 0.0019 0.03 0.1678 -1.0 0.0 0.1752 0.0 0.0 0.0009 0.3694 0.0005 0.134
11.3914 7.0 350 16.8868 0.0015 0.0043 0.0008 -1.0 0.0 0.0016 0.0009 0.0584 0.2636 -1.0 0.0 0.2767 0.0 0.0 0.0026 0.5028 0.0018 0.288
11.7201 8.0 400 16.5662 0.0079 0.0151 0.0113 -1.0 0.0 0.0081 0.0136 0.1118 0.307 -1.0 0.0 0.3222 0.0 0.0 0.0036 0.5889 0.0202 0.332
11.4605 9.0 450 16.2184 0.0033 0.0128 0.0017 -1.0 0.0003 0.0035 0.0156 0.0946 0.3031 -1.0 0.05 0.3166 0.0034 0.02 0.0048 0.5833 0.0018 0.306
11.6944 10.0 500 16.0585 0.0038 0.0192 0.0015 -1.0 0.0004 0.004 0.0117 0.1231 0.3169 -1.0 0.05 0.3313 0.005 0.025 0.0045 0.5917 0.0019 0.334

Framework versions

  • Transformers 5.12.1
  • Pytorch 2.12.0+cu130
  • Datasets 5.0.0
  • Tokenizers 0.22.2
Downloads last month
-
Safetensors
Model size
33.6M params
Tensor type
F32
·

Model tree for merve/rfdetr-docvqa-media3-v2-agree1-large

Finetuned
(2)
this model