[GitHub] [tvm] LeshengJin opened a new pull request, #15632: [Disco][Test] MultiHeadAttention Testcase

["LeshengJin (via GitHub)" <gi...@apache.org>]

LeshengJin opened a new pull request, #15632:
URL: https://github.com/apache/tvm/pull/15632

   This pr adds a distributed MultiHeadAttention into our disco testset.
   
   In the distributed MultiHeadAttention, we shard `Wq, Wk, Wv` along axis 1, `Wo` along axis 0. After each worker completes its attention computation, we use Allreduce operator to get the final result.
   
   ![](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/parallelism-tp-parallel_self_attention.png)
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: commits-unsubscribe@tvm.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org

[GitHub] [tvm] jinhongyii merged pull request #15632: [Disco][Test] MultiHeadAttention Testcase

["jinhongyii (via GitHub)" <gi...@apache.org>]

jinhongyii merged PR #15632:
URL: https://github.com/apache/tvm/pull/15632


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: commits-unsubscribe@tvm.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org