rtdetr-flowchart-detector

This model is a fine-tuned version of rukia07/rtdetr-flowchart-detector on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 3.5243
• Map: 0.9722
• Map 50: 0.9997
• Map 75: 0.9997
• Map Small: -1.0
• Map Medium: 1.0
• Map Large: 0.972
• Mar 1: 0.9691
• Mar 10: 0.9897
• Mar 100: 0.9956
• Mar Small: -1.0
• Mar Medium: 1.0
• Mar Large: 0.9955
• Map Flowchart: 0.9722
• Mar 100 Flowchart: 0.9956

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: 2e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• 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: linear
• lr_scheduler_warmup_steps: 50
• num_epochs: 7

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 Flowchart	Mar 100 Flowchart
17.8389	1.0	100	11.9907	0.9244	0.9917	0.9751	-1.0	0.9505	0.9256	0.9471	0.9809	0.9824	-1.0	0.95	0.9833	0.9244	0.9824
10.9563	2.0	200	6.3759	0.9678	1.0	1.0	-1.0	0.9254	0.9696	0.9765	0.9853	0.9956	-1.0	1.0	0.9955	0.9678	0.9956
7.6387	3.0	300	4.4808	0.9581	1.0	1.0	-1.0	0.9168	0.9608	0.975	0.9882	0.9912	-1.0	0.95	0.9924	0.9581	0.9912
5.9953	4.0	400	3.5260	0.9764	0.9997	0.9997	-1.0	1.0	0.9757	0.9721	0.9912	0.9956	-1.0	1.0	0.9955	0.9764	0.9956
5.3339	5.0	500	2.9214	0.9662	1.0	1.0	-1.0	0.9252	0.9675	0.9779	0.9868	0.9912	-1.0	0.95	0.9924	0.9662	0.9912
5.2442	6.0	600	2.7028	0.9632	0.9997	0.9997	-1.0	0.9252	0.9646	0.9632	0.9824	0.9868	-1.0	0.95	0.9879	0.9632	0.9868
5.1009	7.0	700	2.5877	0.9692	0.9997	0.9997	-1.0	0.9252	0.9718	0.9809	0.9897	0.9926	-1.0	0.95	0.9939	0.9692	0.9926

Framework versions

• Transformers 5.6.1
• Pytorch 2.11.0+cu130
• Tokenizers 0.22.2