Skip to content
This repository has been archived by the owner on Aug 30, 2024. It is now read-only.

[BesTLA] Improve RTN quantization accuracy of int4 and int3 #172

Merged
merged 29 commits into from
Mar 18, 2024
Merged

[BesTLA] Improve RTN quantization accuracy of int4 and int3 #172

merged 29 commits into from
Mar 18, 2024

Commits on Mar 15, 2024

  1. add s4_auto calibration

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    f4184b2 View commit details
    Browse the repository at this point in the history

  2. remove debug code

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    5ca5070 View commit details
    Browse the repository at this point in the history

  3. fix S3 quant error: add rounding and auto quant.

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    787aefb View commit details
    Browse the repository at this point in the history

  4. clang-format

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    ffc50b2 View commit details
    Browse the repository at this point in the history

  5. add int3 for quant args

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    3c074c6 View commit details
    Browse the repository at this point in the history

  6. fix compile error on GCC8.5

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    ce73ca6 View commit details
    Browse the repository at this point in the history

  7. revert random range

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    98ae6e3 View commit details
    Browse the repository at this point in the history

  8. use AVX512F inst

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    d9372e5 View commit details
    Browse the repository at this point in the history

  9. pass compilation on GCC8.5

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    b01354a View commit details
    Browse the repository at this point in the history

  10. add avx2 version of s4 decompression

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    a416ac2 View commit details
    Browse the repository at this point in the history

  11. fix compile

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    4e5821a View commit details
    Browse the repository at this point in the history

  12. remove S4_FULLRANGE

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    1afa830 View commit details
    Browse the repository at this point in the history

  13. fix compile

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    d81a785 View commit details
    Browse the repository at this point in the history

  14. add decompress_kblock_s4_fp to avx2 file

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    2738a1e View commit details
    Browse the repository at this point in the history

  15. remove warnings

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    f1904ae View commit details
    Browse the repository at this point in the history

  16. remove pow usage

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    dd1ec9f View commit details
    Browse the repository at this point in the history

  17. fix dtype

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    080020f View commit details
    Browse the repository at this point in the history

  18. fix the e8m0 conversion code

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    edf6ddd View commit details
    Browse the repository at this point in the history

  19. remove SSE unpack 4bit

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    64a63cb View commit details
    Browse the repository at this point in the history

  20. remove _mm256_i32gather_ps for poor performance

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    417b505 View commit details
    Browse the repository at this point in the history

  21. for dequant

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    08d6d56 View commit details
    Browse the repository at this point in the history

  22. fix compile

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    8f9c1a7 View commit details
    Browse the repository at this point in the history

  23. fix UT error

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    01b4679 View commit details
    Browse the repository at this point in the history

  24. fix UT err

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    a2f9814 View commit details
    Browse the repository at this point in the history

  25. fix thread dead lock

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    670b1d3 View commit details
    Browse the repository at this point in the history

  26. clang-format

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    ad6b441 View commit details
    Browse the repository at this point in the history

  27. fix double-quant bug

    @zhewang1-intc
    zhewang1-intc committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    5899951 View commit details
    Browse the repository at this point in the history

  28. fix UT threshold

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    f5bda82 View commit details
    Browse the repository at this point in the history

  29. fix code bug

    @luoyu-intel
    luoyu-intel committed Mar 15, 2024
    Configuration menu
    Copy the full SHA
    058a574 View commit details
    Browse the repository at this point in the history