Issues about quantizing yolov8 using vitis-ai 2.5

[SM2XWe1L]

Hello! When i want to quantize yolov8 model using vitis-ai 2.5,i encountered many problems:
1.when i start to quantize, the tool pytorch_nndct always generate a file named "autobackend.py" in "quant_result"directory.(I use yolov8's Autobackend class to load my trained v8 model),the file is as below:

Can anyone teach me how to read it and why it is generated?

2. Have anyone encounter the same issue as below?
   
   I try to read its source code ,actually the problem occurs in the function "do_scan", the param res is a tensor but the source code shows that some of the res have the attribute "values", it really confused me .

Thirsty for reply!!!