Edward Z. Yang
fa40195fac
Don't set_current_node in DDP. ( #101046 )
...
Fixes https://github.com/pytorch/pytorch/issues/101045
Signed-off-by: Edward Z. Yang <ezyang@meta.com>
Pull Request resolved: https://github.com/pytorch/pytorch/pull/101046
Approved by: https://github.com/wconstab , https://github.com/malfet
2023-05-12 14:37:22 +00:00
Bert Maher
e0bf51d3bf
[dynamo] Add ddp_graphs artifact ( #100021 )
...
I want to be able to decouple DDP graph printing from the rest of
dynamo DEBUG-level logging, since frequently these logs are particularly
enlightening.
Differential Revision: [D45290919](https://our.internmc.facebook.com/intern/diff/D45290919/ )
Pull Request resolved: https://github.com/pytorch/pytorch/pull/100021
Approved by: https://github.com/wconstab , https://github.com/mlazos
2023-04-27 03:53:23 +00:00
Edward Z. Yang
b09722f540
Convert logging f-strings to use % format, part two ( #98700 )
...
This hits multi-line logging strings
Signed-off-by: Edward Z. Yang <ezyang@meta.com>
Pull Request resolved: https://github.com/pytorch/pytorch/pull/98700
Approved by: https://github.com/voznesenskym
2023-04-10 12:19:31 +00:00
Edward Z. Yang
9a8f71f23e
Convert logging f-strings to use % format ( #98697 )
...
Codemod done with
https://gist.github.com/ezyang/2e8b0463cdc6be278478495b23ff0530 with
assistance from ChatGPT.
Signed-off-by: Edward Z. Yang <ezyang@meta.com>
Pull Request resolved: https://github.com/pytorch/pytorch/pull/98697
Approved by: https://github.com/voznesenskym
2023-04-10 12:19:31 +00:00
Edward Z. Yang
d01ee10b25
Add detect_fake_mode ( #98321 )
...
This replaces fake_mode_from_tensors but it preferentially looks for
fake_mode in TracingContext and also if there is an active fake mode
on the dispatch stack, before groveling in tensors to find it.
This advances PegasusForCausalLM, which was previously failing because
we generated a graph that had a parameter (non-fake) and a SymInt,
and thus previously we failed to detect the correct fake mode.
Signed-off-by: Edward Z. Yang <ezyang@meta.com>
Pull Request resolved: https://github.com/pytorch/pytorch/pull/98321
Approved by: https://github.com/voznesenskym
2023-04-05 22:15:16 +00:00
Edward Z. Yang
5df59f957f
Fix G001,G002,G003 in logs to % syntax ( #97812 )
...
Signed-off-by: Edward Z. Yang <ezyang@meta.com>
Pull Request resolved: https://github.com/pytorch/pytorch/pull/97812
Approved by: https://github.com/Skylion007 , https://github.com/kiukchung , https://github.com/malfet , https://github.com/mlazos
2023-04-01 01:43:33 +00:00
Andrew Gu
d9cd9a13bc
[BE][DDPOptimizer] De-dup p and param ( #95654 )
...
The `param` from `param = target.get_parameter(name)` should be the same as `p` from `target.named_parameters()`.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/95654
Approved by: https://github.com/wconstab
2023-03-01 01:17:09 +00:00
Kazuaki Ishizaki
46385b3e48
Fix typos under torch/_dynamo directory ( #95599 )
...
This PR fixes typos in comments and messages of `.py` files under `torch/_dynamo` directory
Pull Request resolved: https://github.com/pytorch/pytorch/pull/95599
Approved by: https://github.com/ezyang
2023-02-28 03:44:24 +00:00
Will Constable
9fb9219478
Make DDPOptimizer work with torch._dynamo.explain() ( #94749 )
...
GraphModules that were created during DDPOptimizer graph breaking
lacked `compile_subgraph_reason`, which caused an exception when
running .explain().
Now the reason is provided and users can use .explain() to find out
that DDPOptimizer is causing graph breaks.
Fixes #94579
Pull Request resolved: https://github.com/pytorch/pytorch/pull/94749
Approved by: https://github.com/voznesenskym
2023-02-14 01:33:47 +00:00
Jason Ansel
2b0d7e63f0
Move dynamo.optimizations.distributed to backends ( #93408 )
...
Pull Request resolved: https://github.com/pytorch/pytorch/pull/93408
Approved by: https://github.com/wconstab
2023-02-02 20:42:17 +00:00
