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

PagedAttention Transformation: Rank alignment for replacements #24690

Merged
merged 20 commits into from
May 27, 2024
Merged

PagedAttention Transformation: Rank alignment for replacements #24690

merged 20 commits into from
May 27, 2024

Commits on May 20, 2024

  1. PagedAttention transfomration extension: SDPA with scale input and kv… 

    …_heads dimension broadcasted in SDPA itself and not in the UBR pattern.
    @slyalin
    slyalin committed May 20, 2024
    Configuration menu
    Copy the full SHA
    cfc75d4 View commit details
    Browse the repository at this point in the history

  2. Fix code style

    @slyalin
    slyalin committed May 20, 2024
    Configuration menu
    Copy the full SHA
    2442243 View commit details
    Browse the repository at this point in the history

Commits on May 21, 2024

  1. Merge remote-tracking branch 'origin/master' into pa_sdpa_extra_caps

    @slyalin
    slyalin committed May 21, 2024
    Configuration menu
    Copy the full SHA
    2fe7d03 View commit details
    Browse the repository at this point in the history

  2. Update src/common/transformations/src/transformations/sdpa_to_paged_a… 

    …ttention/prev_sequence_length_pattern.cpp

Co-authored-by: Ivan Tikhonov <ivan.tikhonov@intel.com>
    @slyalin @itikhono
    slyalin and itikhono committed May 21, 2024
    Configuration menu
    Copy the full SHA
    2bf94ba View commit details
    Browse the repository at this point in the history

  3. Merge remote-tracking branch 'origin/master' into pa_sdpa_extra_caps

    @slyalin
    slyalin committed May 21, 2024
    Configuration menu
    Copy the full SHA
    a94a455 View commit details
    Browse the repository at this point in the history

  4. Merge remote-tracking branch 'slyalin/pa_sdpa_extra_caps' into pa_sdp… 

    …a_extra_caps
    @slyalin
    slyalin committed May 21, 2024
    Configuration menu
    Copy the full SHA
    3d04b1b View commit details
    Browse the repository at this point in the history

  5. Fixed wrong input index for number of heads in K.

    @slyalin
    slyalin committed May 21, 2024
    Configuration menu
    Copy the full SHA
    586dacd View commit details
    Browse the repository at this point in the history

Commits on May 23, 2024

  1. Refactored pattern for reading K and V past values into a single func… 

    …tion. Allowed optional Reshape in UBR pattern (appeared in one of MQA cases).
    @slyalin
    slyalin committed May 23, 2024
    Configuration menu
    Copy the full SHA
    33f092c View commit details
    Browse the repository at this point in the history

  2. Extended PA transformation to case when K and V are merged into a sin… 

    …gle tensor.
    @slyalin
    slyalin committed May 23, 2024
    Configuration menu
    Copy the full SHA
    4b24ed6 View commit details
    Browse the repository at this point in the history

  3. Merge remote-tracking branch 'origin/master' into merged_kv_states

    @slyalin
    slyalin committed May 23, 2024
    Configuration menu
    Copy the full SHA
    11be9c5 View commit details
    Browse the repository at this point in the history

  4. Code style fix

    @slyalin
    slyalin committed May 23, 2024
    Configuration menu
    Copy the full SHA
    f36ce8f View commit details
    Browse the repository at this point in the history

  5. Dead code eliminated

    @slyalin
    slyalin committed May 23, 2024
    Configuration menu
    Copy the full SHA
    5db002b View commit details
    Browse the repository at this point in the history

  6. Grammar fixes in comments.

    @slyalin
    slyalin committed May 23, 2024
    Configuration menu
    Copy the full SHA
    99f24f0 View commit details
    Browse the repository at this point in the history

  7. Removed debug output and commented code.

    @slyalin
    slyalin committed May 23, 2024
    Configuration menu
    Copy the full SHA
    9bd08e8 View commit details
    Browse the repository at this point in the history

Commits on May 24, 2024

  1. Fixed bloom: removed not used code that triggered a bug inside patter… 

    …n matching with Or pattern and multi-output nodes.
    @slyalin
    slyalin committed May 24, 2024
    Configuration menu
    Copy the full SHA
    8f447fd View commit details
    Browse the repository at this point in the history

  2. Align shape of new inputs with expected shape in the place where they… 

    … are inserted.
    @slyalin
    slyalin committed May 24, 2024
    Configuration menu
    Copy the full SHA
    ebf0130 View commit details
    Browse the repository at this point in the history

  3. Merge remote-tracking branch 'origin/master' into new_inputs_rank_align

    @slyalin
    slyalin committed May 24, 2024
    Configuration menu
    Copy the full SHA
    7e31bcb View commit details
    Browse the repository at this point in the history

  4. Code style fix

    @slyalin
    slyalin committed May 24, 2024
    Configuration menu
    Copy the full SHA
    bc4faf8 View commit details
    Browse the repository at this point in the history

Commits on May 26, 2024

  1. Merge remote-tracking branch 'origin/master' into new_inputs_rank_align

    @slyalin
    slyalin committed May 26, 2024
    Configuration menu
    Copy the full SHA
    a281e79 View commit details
    Browse the repository at this point in the history

Commits on May 27, 2024

  1. Merge branch 'master' into new_inputs_rank_align

    @ilya-lavrenov
    ilya-lavrenov committed May 27, 2024
    Configuration menu
    Copy the full SHA
    141a5ad View commit details
    Browse the repository at this point in the history