ganler edited a comment on pull request #8586: URL: https://github.com/apache/tvm/pull/8586#issuecomment-889252964
As is mentioned in this issue https://github.com/apache/tvm/issues/8579 If users get used to putting `.evaluate()` and the callable together, they might be confused by the performance issue if they have a loop in inference (they will compile the model for multiple times, and according to my experience TVM's compilation is much longer than the model execution itself). So I guess if we want to form a consistent API usage style by not breaking existing APIs, we would better also split `.evaluate` and `(**data)`. -- 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-unsubscr...@tvm.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org