The actual test speed of rf-detr is particularly slow

[4399123]

Search before asking

• [x] I have searched the RF-DETR issues and found no similar bug report.

Bug

I conducted a comparative test between rf-deter-nano and yolov6-nano and found that the latter is significantly faster than the former. With an input image size of 672x672, rf-deter-nano takes 2.75ms while yolov6-nano (including NMS processing) only takes 0.45ms. The GPU used is RTX 4080S. Please tell me why there is such a large gap between the two.

Environment

RF-DETR ：1.3 Windows 10 Python version：3.11 PyTorch version：2.4 CUDA：11.6 GPU：RTX 4080S

Minimal Reproducible Example

---

Additional

No response

Are you willing to submit a PR?

• [ ] Yes, I'd like to help by submitting a PR!

[isaacrob-roboflow]

That's because RF-detr nano is supposed to run at 384