Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Feature (quantizer): Adding accumulator-aware weight quantization #567

Merged
merged 31 commits into from
Apr 5, 2023
Merged

Feature (quantizer): Adding accumulator-aware weight quantization #567

merged 31 commits into from
Apr 5, 2023

Commits on Apr 4, 2023

  1. Adding DecoupledRescalingIntQuantWithInput

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    b28f8e4 View commit details
    Browse the repository at this point in the history

  2. Updating QuantWeightMixin to handle quant_input if required

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    9e1e2d9 View commit details
    Browse the repository at this point in the history

  3. Updating base QuantLayer to pass quant_input to weight_quant

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    0fdeda9 View commit details
    Browse the repository at this point in the history

  4. Adding new weight quant proxy to pass input bit-width and sign

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    d8a3fc7 View commit details
    Browse the repository at this point in the history

  5. Adding input-aware decoupled weight quant to top-level import

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    97e2782 View commit details
    Browse the repository at this point in the history

  6. Updating quant weight to use cached quant input

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    698a50a View commit details
    Browse the repository at this point in the history

  7. Adding accumulator-aware pre-clipping scaling factor

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    c599750 View commit details
    Browse the repository at this point in the history

  8. Adding new weight quant

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    a5be8e6 View commit details
    Browse the repository at this point in the history

  9. Adding pre-scaling to top-level import

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    7915b0b View commit details
    Browse the repository at this point in the history

  10. Updating quant injector

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    c3c9112 View commit details
    Browse the repository at this point in the history

  11. Adding int8 weight quantizer to fixed_point.py

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    ed66c60 View commit details
    Browse the repository at this point in the history

  12. Adding int8 a2q to quantizer testing

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    648de8f View commit details
    Browse the repository at this point in the history

  13. Adding pytest skip for case that A2Q does not support.

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    94b4211 View commit details
    Browse the repository at this point in the history

  14. Minor cosmetic changes to pre-scaling impl

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    8e42223 View commit details
    Browse the repository at this point in the history

  15. Adding accumulator bit-width bounds to nn.utils

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    a6986e1 View commit details
    Browse the repository at this point in the history

  16. Updating function with notes and comments

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    66e7048 View commit details
    Browse the repository at this point in the history

  17. Adding re-parameterization of default case_model fixture

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    2e7920c View commit details
    Browse the repository at this point in the history

  18. Adding tests for A2Q

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    029500d View commit details
    Browse the repository at this point in the history

  19. Pre-commit fixes (all but onnx export)

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    173ac27 View commit details
    Browse the repository at this point in the history

  20. Remove caching on the forward pass

    @volcacius @i-colbert
    volcacius authored and i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    64064c3 View commit details
    Browse the repository at this point in the history

  21. Add NotImplementedError for metadata retrieval over new decoupled proxy

    @volcacius @i-colbert
    volcacius authored and i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    f1d743d View commit details
    Browse the repository at this point in the history

  22. Fix QuantScaleBias quant_weight

    @volcacius @i-colbert
    volcacius authored and i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    d6711ea View commit details
    Browse the repository at this point in the history

  23. Pre-commit fixes

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    f9967b6 View commit details
    Browse the repository at this point in the history

  24. Fixing rebase merge

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    1731d49 View commit details
    Browse the repository at this point in the history

  25. Fixing first set of errors 

    Can't target case_id deconstruction when a2q and base test cases are mismatched
    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    0adeefd View commit details
    Browse the repository at this point in the history

  26. Skipping batch quant for a2q cases

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    d4e2e46 View commit details
    Browse the repository at this point in the history

  27. Fixing a2q pytest cases

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    3898a65 View commit details
    Browse the repository at this point in the history

  28. Pre-commit fixes

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    4dbd40c View commit details
    Browse the repository at this point in the history

  29. Adding more notes

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    980fb19 View commit details
    Browse the repository at this point in the history

  30. Skipping unsupported A2Q cases

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    af07a91 View commit details
    Browse the repository at this point in the history

  31. QuantMHA needs input quantizer for A2Q

    @i-colbert
    i-colbert committed Apr 4, 2023
    Configuration menu
    Copy the full SHA
    0720a7d View commit details
    Browse the repository at this point in the history