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

Add more quantization support for burn-jit #2275

Merged
merged 21 commits into from
Sep 17, 2024
Merged

Add more quantization support for burn-jit #2275

merged 21 commits into from
Sep 17, 2024

Commits on Sep 12, 2024

  1. Add cubecl quantization kernels and QTensorOps for burn-jit

    @laggui
    laggui committed Sep 12, 2024
    Configuration menu
    Copy the full SHA
    7637851 View commit details
    Browse the repository at this point in the history

  2. Fix typo

    @laggui
    laggui committed Sep 12, 2024
    Configuration menu
    Copy the full SHA
    2256890 View commit details
    Browse the repository at this point in the history

  3. Fix output vec factor

    @laggui
    laggui committed Sep 12, 2024
    Configuration menu
    Copy the full SHA
    4f93389 View commit details
    Browse the repository at this point in the history

  4. Fix output dtype size_of

    @laggui
    laggui committed Sep 12, 2024
    Configuration menu
    Copy the full SHA
    7043b95 View commit details
    Browse the repository at this point in the history

  5. Remove unused code in dequantize test

    @laggui
    laggui committed Sep 12, 2024
    Configuration menu
    Copy the full SHA
    13f538f View commit details
    Browse the repository at this point in the history

  6. Fix dequantize vectorization

    @laggui
    laggui committed Sep 12, 2024
    Configuration menu
    Copy the full SHA
    0f278f4 View commit details
    Browse the repository at this point in the history

  7. Handle tensors when number of elems is not a multiple of 4

    @laggui
    laggui committed Sep 12, 2024
    Configuration menu
    Copy the full SHA
    31c3902 View commit details
    Browse the repository at this point in the history

  8. Support quantize for tensors with less than 4 elems (no vectorization)

    @laggui
    laggui committed Sep 12, 2024
    Configuration menu
    Copy the full SHA
    e0dcd5d View commit details
    Browse the repository at this point in the history

  9. Fix equal 0 test

    @laggui
    laggui committed Sep 12, 2024
    Configuration menu
    Copy the full SHA
    0403075 View commit details
    Browse the repository at this point in the history

  10. Add quantize/dequantize tests

    @laggui
    laggui committed Sep 12, 2024
    Configuration menu
    Copy the full SHA
    0fac955 View commit details
    Browse the repository at this point in the history

  11. Add q_to_device

    @laggui
    laggui committed Sep 12, 2024
    Configuration menu
    Copy the full SHA
    ea11947 View commit details
    Browse the repository at this point in the history

  12. Refactor kernels for latest cubecl

    @laggui
    laggui committed Sep 12, 2024
    Configuration menu
    Copy the full SHA
    e6e1a0e View commit details
    Browse the repository at this point in the history

Commits on Sep 16, 2024

  1. intermediate i32 cast

    @louisfd
    louisfd committed Sep 16, 2024
    Configuration menu
    Copy the full SHA
    2d04530 View commit details
    Browse the repository at this point in the history

  2. Fix size_of output type

    @laggui
    laggui committed Sep 16, 2024
    Configuration menu
    Copy the full SHA
    19cf31a View commit details
    Browse the repository at this point in the history

  3. Use strict=false to ignore floating point precision issues with qpara… 

    …ms equality
    @laggui
    laggui committed Sep 16, 2024
    Configuration menu
    Copy the full SHA
    7b17133 View commit details
    Browse the repository at this point in the history

  4. Only check that lhs & rhs strategies match (but not strict on qparams… 

    … values)
    @laggui
    laggui committed Sep 16, 2024
    Configuration menu
    Copy the full SHA
    dd5cdf9 View commit details
    Browse the repository at this point in the history

  5. Use assert_approx_eq on dequant values

    @laggui
    laggui committed Sep 16, 2024
    Configuration menu
    Copy the full SHA
    eab11d0 View commit details
    Browse the repository at this point in the history

  6. Reduce precision for flaky test

    @laggui
    laggui committed Sep 16, 2024
    Configuration menu
    Copy the full SHA
    ebbbdf8 View commit details
    Browse the repository at this point in the history

  7. Remove todo comment

    @laggui
    laggui committed Sep 16, 2024
    Configuration menu
    Copy the full SHA
    d1426eb View commit details
    Browse the repository at this point in the history

Commits on Sep 17, 2024

  1. Add comment for cast to unsigned

    @laggui
    laggui committed Sep 17, 2024
    Configuration menu
    Copy the full SHA
    fc09140 View commit details
    Browse the repository at this point in the history

  2. More comment

    @laggui
    laggui committed Sep 17, 2024
    Configuration menu
    Copy the full SHA
    d46076e View commit details
    Browse the repository at this point in the history