yolo post training quantization

[junghyun-avikus]

Hi, I was following along the post training quantization, and I am wondering if given examples codes can convert yolov5m as well.

The given yaml for yolov5 is yolov5s_ptq.yaml, so is the code specific to yolov5s only?

When I ran post_quant on the yolov5m onnx, directly exported from the ultralytics pretrained model on coco, I get the following error.

[WARNING] Incomplete symbolic shape inference 2023-03-24 17:30:56,985-WARNING: [ShapeInferenceError] (op_type:Split, node name: x2paddle__model_24_Split_output_0): [ShapeInferenceError] Inferred shape and existing shape differ in dimension 4: (2) vs (28) 2023-03-24 17:30:56,986-ERROR: x2paddle threw an exception, you can ask for help at: https://github.com/PaddlePaddle/X2Paddle/issues

Do I need to make some modifications in the code, or is it the environment that I am running on.

The environments are the followings: Paddle 2.4.1 cuda 11.7 PaddleSlim 2.4 x2paddle 1.3.9

[XGZhang11]

Sorry for not replying in time. It seems that there is something wrong with ShapeInference. Can you provide more information about the code and model you use?