mjdenkowski commented on issue #18699: URL: https://github.com/apache/incubator-mxnet/issues/18699#issuecomment-660692437
This benchmark was run using a trained [Sockeye](https://github.com/awslabs/sockeye) model. This is a full sequence-to-sequence model with HybridBlocks glued together by Python code. We're running inference (translating a test set with beam search), so there shouldn't be any backward operations. We can share a model with input data and a run script if that will help with debugging. Alternatively, do you have an idea of what good tests would be for determining frontend vs backend speed regressions or multi-API overhead? Are there build/runtime options for MXNet that we can try with our sample model? If we have a good idea of everything that changed in commit 83b5170, we can work through it by process of elimination. ---------------------------------------------------------------- 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: us...@infra.apache.org