detr_finetuned_cppe5

This model is a fine-tuned version of microsoft/table-transformer-structure-recognition-v1.1-all on an unknown dataset. It achieves the following results on the evaluation set:

• eval_loss: 3.3083
• eval_map: 0.0584
• eval_map_50: 0.1515
• eval_map_75: 0.0479
• eval_map_small: -1.0
• eval_map_medium: 0.007
• eval_map_large: 0.0646
• eval_mar_1: 0.0746
• eval_mar_10: 0.115
• eval_mar_100: 0.1545
• eval_mar_small: -1.0
• eval_mar_medium: 0.0439
• eval_mar_large: 0.1653
• eval_map_table: 0.2451
• eval_mar_100_table: 0.2882
• eval_map_table column: 0.0237
• eval_mar_100_table column: 0.1297
• eval_map_table column header: 0.0245
• eval_mar_100_table column header: 0.1224
• eval_map_table projected row header: 0.0003
• eval_mar_100_table projected row header: 0.0125
• eval_map_table row: 0.0254
• eval_mar_100_table row: 0.235
• eval_map_table spanning cell: 0.0311
• eval_mar_100_table spanning cell: 0.1393
• eval_runtime: 80.5383
• eval_samples_per_second: 0.633
• eval_steps_per_second: 0.087
• epoch: 1.0
• step: 22

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

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0