Closed maro-jeon closed 11 months ago
Thank you for your repo. it's really good for understanding qat tool.
Now i am currently working on qat for yolov5. I added Q/DQ node with your advice. see the below.
but still they have useless conversion fp 16 to int8. see the below trex graph.
how can i remove the useless conversion node ? Thanks, Maro JEON
I added Q/DQ layer on add , concat, resize node. and remove (merge) the redundant q/dq layer. finally got the same fps qat engine compared with ptq engine.
add
concat
resize
Maro JEON
Thank you for your repo. it's really good for understanding qat tool.
Now i am currently working on qat for yolov5. I added Q/DQ node with your advice. see the below.![image](https://github.com/NVIDIA-AI-IOT/yolo_deepstream/assets/135101992/f684edd8-6cc1-46e4-a297-406fa6c39c99)
but still they have useless conversion fp 16 to int8. see the below trex graph.![image](https://github.com/NVIDIA-AI-IOT/yolo_deepstream/assets/135101992/bec4bdac-1a55-41c5-8fb7-7a48315dfdd1)
how can i remove the useless conversion node ? Thanks, Maro JEON