[Frontend][Pytorch] Handle case where output of model is python list #7088
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
TusharKanekiDey
approved these changes
Dec 10, 2020
masahi
approved these changes
Dec 11, 2020
There was a problem hiding this comment.
@trevor-m Yes, we haven't supported list output because it is not clear what output users would expect from our frontend. They definitely don't want a List ADT. Returning a python list is also wrong since Relay doesn't ingest it.
If the number of output elements is fixed, users should have used tuple instead of list in their model definition. But this fix is also good.
|
thanks @trevor-m @TusharKanekiDey |
TusharKanekiDey
pushed a commit
to TusharKanekiDey/tvm
that referenced
this pull request
Jan 20, 2021
trevor-m
pushed a commit
to neo-ai/tvm
that referenced
this pull request
Jan 21, 2021
electriclilies
pushed a commit
to electriclilies/tvm
that referenced
this pull request
Feb 18, 2021
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
prim::ListConstructconverter can make a regular python list when the list is not dynamic:https://github.com/apache/tvm/blob/main/python/tvm/relay/frontend/pytorch.py#L2436-L2439
If that python list happens to be the output of the whole model, an error would occur when the list is passed to
_analysis.free_varsbecause it only accepts Expr. This PR puts the list in a tuple to solve that.