You are viewing a plain text version of this content. The canonical link for it is here.

Posted to commits@tvm.apache.org by GitBox <gi...@apache.org> on 2022/02/08 02:02:20 UTC

[GitHub] [tvm] YuchiWen opened a new pull request #10186: [Torch] Run torch JIT pass lower_all_tuples before conversion.

YuchiWen opened a new pull request #10186:
URL: https://github.com/apache/tvm/pull/10186


   See the following discuss thread for details
   https://discuss.tvm.apache.org/t/bug-frontend-pytorch-relay-ir-is-inconsistent-with-that-of-the-original-model/12010
   


-- 
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] masahi commented on pull request #10186: [Torch] Run torch JIT pass lower_all_tuples before conversion.

Posted by GitBox <gi...@apache.org>.

masahi commented on pull request #10186:
URL: https://github.com/apache/tvm/pull/10186#issuecomment-1032270107


   The issue is that `torch._C._jit_pass_lower_all_tuples(graph)` unpack input tuples as well, so 
   
   ```
   graph(%self : __torch__.LSTMLayer,
         %input.1 : Tensor,
         %state.1 : (Tensor, Tensor)):
   
   ``` 
   
   becomes
   
   ```
   graph(%self : __torch__.LSTMLayer,
         %input.1 : Tensor,
         %72 : Tensor,
         %73 : Tensor):
   ```
   
   I think this should not happen, since the user would use a tuple in their Python code anyway, so we should respect that. @YuchiWen Can you try to find a way around this problem? 


-- 
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] masahi commented on pull request #10186: [Torch] Run torch JIT pass lower_all_tuples before conversion.

Posted by GitBox <gi...@apache.org>.

masahi commented on pull request #10186:
URL: https://github.com/apache/tvm/pull/10186#issuecomment-1032580678


   You have hit a flaky error, please rebase against upstream/main and push to run a new job.
   


-- 
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] YuchiWen commented on pull request #10186: [Torch] Run torch JIT pass lower_all_tuples before conversion.

Posted by GitBox <gi...@apache.org>.

YuchiWen commented on pull request #10186:
URL: https://github.com/apache/tvm/pull/10186#issuecomment-1033420973


   @masahi Sorry, I missed the check for ListType, please review it again.


-- 
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] masahi commented on pull request #10186: [Torch] Run torch JIT pass lower_all_tuples before conversion.

Posted by GitBox <gi...@apache.org>.

masahi commented on pull request #10186:
URL: https://github.com/apache/tvm/pull/10186#issuecomment-1033415502


   @YuchiWen can you take a look at the failed test?


-- 
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] masahi commented on pull request #10186: [Torch] Run torch JIT pass lower_all_tuples before conversion.

Posted by GitBox <gi...@apache.org>.

masahi commented on pull request #10186:
URL: https://github.com/apache/tvm/pull/10186#issuecomment-1032276517


   See 
   https://github.com/pytorch/pytorch/blob/c371542efc31b1abfe6f388042aa3ab0cef935f2/torch/csrc/jit/passes/lower_tuples.h#L14-L15
   
   So I suggest disabling this pass when the user input contains a tuple.


-- 
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] YuchiWen commented on pull request #10186: [Torch] Run torch JIT pass lower_all_tuples before conversion.

Posted by GitBox <gi...@apache.org>.

YuchiWen commented on pull request #10186:
URL: https://github.com/apache/tvm/pull/10186#issuecomment-1032330748


   @masahi I submitted it again, do you have any good suggestions?


-- 
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] masahi commented on pull request #10186: [Torch] Run torch JIT pass lower_all_tuples before conversion.

Posted by GitBox <gi...@apache.org>.

masahi commented on pull request #10186:
URL: https://github.com/apache/tvm/pull/10186#issuecomment-1033252690


   cc @YuchiWen 


-- 
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] YuchiWen commented on pull request #10186: [Torch] Run torch JIT pass lower_all_tuples before conversion.

Posted by GitBox <gi...@apache.org>.

YuchiWen commented on pull request #10186:
URL: https://github.com/apache/tvm/pull/10186#issuecomment-1032135766


   @masahi 


-- 
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] masahi commented on pull request #10186: [Torch] Run torch JIT pass lower_all_tuples before conversion.

Posted by GitBox <gi...@apache.org>.

masahi commented on pull request #10186:
URL: https://github.com/apache/tvm/pull/10186#issuecomment-1032346326


   Yes this looks good, thanks.


-- 
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] YuchiWen commented on pull request #10186: [Torch] Run torch JIT pass lower_all_tuples before conversion.

Posted by GitBox <gi...@apache.org>.

YuchiWen commented on pull request #10186:
URL: https://github.com/apache/tvm/pull/10186#issuecomment-1033280035


   @masahi It's Done.


-- 
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] masahi edited a comment on pull request #10186: [Torch] Run torch JIT pass lower_all_tuples before conversion.

Posted by GitBox <gi...@apache.org>.

masahi edited a comment on pull request #10186:
URL: https://github.com/apache/tvm/pull/10186#issuecomment-1032276517


   See https://github.com/pytorch/pytorch/blob/c371542efc31b1abfe6f388042aa3ab0cef935f2/torch/csrc/jit/passes/lower_tuples.h#L14-L15
   
   So I suggest disabling this pass when the user input contains a tuple.


-- 
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] masahi commented on pull request #10186: [Torch] Run torch JIT pass lower_all_tuples before conversion.

Posted by GitBox <gi...@apache.org>.

masahi commented on pull request #10186:
URL: https://github.com/apache/tvm/pull/10186#issuecomment-1032265060


   oops sorry there is one test failing. I'll take a look.


-- 
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] masahi merged pull request #10186: [Torch] Run torch JIT pass lower_all_tuples before conversion.

Posted by GitBox <gi...@apache.org>.

masahi merged pull request #10186:
URL: https://github.com/apache/tvm/pull/10186


   


-- 
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