Search posts, tags, users, and pages
Khadim Hussain
AI engineer building AI agents, LLM and RAG systems, and computer vision. I write about what I test at khadim.tech.
Originally published on khadim.tech. In short: On an RTX 5090 with TensorRT 10.12, a calibrated INT8 build of RF-DETR put zero layers in INT8. TensorRT's own timing chose FP16 for every layer, so the
Markdown supported · ⌘↵ to post
No responses yet.