rtdetr-cppe5-detection-v2

This model is a fine-tuned version of PekingU/rtdetr_r50vd on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 9.9867

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• 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_ratio: 0.1
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
130.8497	1.0	63	71.7316
60.0138	2.0	126	28.9468
37.7655	3.0	189	18.4218
29.4038	4.0	252	14.7458
24.3595	5.0	315	12.2400
23.2655	6.0	378	11.4468
21.1393	7.0	441	10.5996
20.4947	8.0	504	10.2594
19.731	9.0	567	10.1579
19.052	10.0	630	9.9744
17.977	11.0	693	9.9976
18.3429	12.0	756	9.9192
17.1892	13.0	819	9.8707
17.3446	14.0	882	9.9686
16.9275	15.0	945	9.9490
17.2444	16.0	1008	9.8504
16.6155	17.0	1071	9.8684
15.4995	18.0	1134	9.8703
16.2258	19.0	1197	9.8435
15.7694	20.0	1260	9.9867

Framework versions

• Transformers 4.57.3
• Pytorch 2.9.1+cu128
• Datasets 4.4.1
• Tokenizers 0.22.1