megvii-research / FQ-ViT

[IJCAI 2022] FQ-ViT: Post-Training Quantization for Fully Quantized Vision Transformer
Apache License 2.0
301 stars 48 forks source link

when i use the method to do object detection, how to apply the calibration step for that detector in MMDet framework? #27

Closed Wuyy-fairy closed 1 year ago

Wuyy-fairy commented 1 year ago

Good jobs! I try to use the code to realize the object detection,but I can't understand the step "apply the calibration step for that detector in MMDet framework",can you give me more detailed instructions?thanks very much!

PeiqinSun commented 1 year ago

We will release an mm-detection example with our new quantization tools. https://github.com/megvii-research/Sparsebit/ You can raise issue in sparsebit.