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.

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

[Dev] Feature Improves for bitnet and block reduction #92

Merged
merged 49 commits into from
Jul 21, 2024

Conversation

LeiWang1999
Copy link
Contributor

This pull request introduces several changes to the codebase, primarily focusing on the bitblas and 3rdparty/tvm modules. The changes include modifications to the to_dict and is_not_fast_decoding_supported methods in bitblas/base/roller/hint.py and bitblas/ops/general_matmul/__init__.py respectively, addition of new methods in bitblas/ops/impl/matmul_dequantize_splitk_impl.py, and changes to the subproject commit in 3rdparty/tvm. New benchmarking scripts are also added in the integration/BitNet directory.

Changes to the bitblas module:

  • bitblas/base/roller/hint.py: Refactored the from_dict method to use a class method instead of an instance method, which improves the code's readability and maintainability.
  • bitblas/ops/general_matmul/__init__.py: Extended the is_not_fast_decoding_supported method to handle cases where w_dtype is int4/uint4 and a_dtype is int8, and reformatted the dispatch_tir method for better readability. [1] [2]
  • bitblas/ops/impl/matmul_dequantize_splitk_impl.py: Added new imports and a new method matmul_nt_dequantize_b_propagate_a_propagate_b that seems to implement a new feature or optimization in the matrix multiplication operation. [1] [2]

Changes to the 3rdparty/tvm module:

  • 3rdparty/tvm: Updated the subproject commit, which suggests that the version of the tvm subproject used in this project has been updated.

Addition of new benchmarking scripts:

  • integration/BitNet/benchmark.sh and integration/BitNet/benchmark_generate.py: Added new benchmarking scripts, which will help in performance testing and optimization of the code. [1] [2]

@LeiWang1999 LeiWang1999 merged commit c941c4f into microsoft:main Jul 21, 2024
6 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant