/Technical-Report-for-QAT

Technical Report of Model Quantization Simulations on Convolution Neural Network based on Aimet and pytorch

Primary LanguageJupyter NotebookMIT LicenseMIT

Technical-Report-for-QAT

Technical Report of Model Quantization Simulations on Convolution Neural Network based on Aimet and pytorch

Files

  • acc.csv Full table of accuracy of models

  • dataprocess.py

    Python script to process experiments results

  • dataset Datasets,including MNIST,KMNIST...

  • export Exported files of models,ONNX and torch PTH

  • dlc SNPE converted models,Qualcomm's dlc format

  • lightning_logs Full records and log of tensorboard

  • Technical Report of ...pytorch.assets Picture and other assets of report's markdown file

  • main.ipynb Jupyter Notebook of model's training and quantization

  • main.py Python script of model's training and quantization

  • run_me.py Python script wrapped the main.py to train models repeatly

  • view_tensorboard.sh Bash script to run tensorboard and view full learning curves

  • Technical Report of ...pytorch.md Markdown file of report

  • Technical Report of ...pytorch.pdf PDF file of report