tkonolige commented on pull request #8084:
URL: https://github.com/apache/tvm/pull/8084#issuecomment-845274248
Our implementation already has an implicit batch dimension. The current
implementation is "Given data with shape (X_0, X_1, …, X_{N-1}) and indices
with shape (M, Y_0, …, Y_{K-1}), the output will have shape (Y_0, …, Y_{K-1},
X_M, …, X_{N-1}), where M <= N. If M == N, output shape will simply be (Y_0, …,
Y_{K-1})." X_M, ..., X_{N-1} is the implicit batch dimension. How does the
explicit batch size differ from this. And should we consider unifying the two?
--
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.
For queries about this service, please contact Infrastructure at:
[email protected]
