Draft: Dev ptq

TODO

• Normalize layers
  • Get weight max value for each layer (Fc/Conv)-> alpha
  • Divide weights by alpha
  • Divide bias by alpha
  • Add PerOutputChannel normalization
    • TODO : add steps to do this ...
• Normalize activation
  • Normalize stimuli between [0;1]
    • or [-1,1]
  • Forward on validation dataset
  • Get max value of activation for each layer -> beta
    • Develop hook sytsem
    • Develop get max activation hook
  • Add scaling factor beta to the activation
  • Add scaling factor beta to the bias
  • Rescale activation by parent bias scaling
• Quantization
  • Input: multiply by (2^n-1)-1 (singed) or (2^n-1) (unsigned)
  • Weights: Multiply by (2^n-1)-1 + round + store as a signed integer of nbbits size
  • Biases: Multiply by (2^n-1)-1 + Multiply by (2^n-1)-1 (singed) or (2^n-1) (unsigned) + store as a signed integer of nbbits size
  • Activation scaling:
    • Input scaling: (2^n-1)-1 + Multiply by (2^n-1)-1 (singed) or (2^n-1) (unsigned)
    • Output scaling: (2^n-1)-1 (singed) or (2^n-1) (unsigned)
    • Activation sclaing : divide by (input scaling divided bu output_scaling)
• Activaiton clipping
  • generate histogram of outputs for each layer (using validation dataset)
  • MSE
    • TODO : add steps to do this
  • KL-Divergence
    • TODO : add steps to do this
• Weights clipping
  • TODO : add steps to do this
• Better scaling operation
  • Fixed-point scaling
  • Single shift scaling
  • Double shift scaling
• Bind method in Python
• LeNet integration test
• Refactor OutputRange hook to use Tensor Getter/Setter
• Add docstring to OutputRange hook
• Refactor hook system to not use registrar ?

assigned to @cmoineau

changed the description

@ikucher I have tried to list all the things to do in order to have a valid PTQ please feel free to notice me if I have forgotten any things :)

changed the description

marked the checklist item Get weight max value for each layer (Fc/Conv)-> alpha as completed

added 1 commit

• f84b462c - [Setup] Update setup to install ib in env.

Compare with previous version

added 1 commit

• 64fbc46a - Update hook, add comment and fix minor compilation issues.

Compare with previous version

added 1 commit

• 2202cd0f - Reflect hook naming changes.

Compare with previous version

marked the checklist item Divide weights by alpha as completed

marked the checklist item Divide bias by alpha as completed

marked the checklist item Normalize layers as completed

marked the checklist item Get max value of activation for each layer -> beta as completed

marked the checklist item Develop hook sytsem as completed

marked the checklist item Develop get max activation hook as completed

marked the checklist item Normalize stimuli between [0;1] or [-1,1] as completed

marked the checklist item Add scaling factor beta to the bias as completed

marked the checklist item Add scaling factor beta to the bias as incomplete

changed the description

marked the checklist item Add scaling factor beta to the bias as completed