fix/feat: Add lowering pass to resolve most aten::Int.Tensor uses
#1937
Conversation
This file contains hidden or 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
github-actions
bot
added
component: core
gs-olive
changed the title
aten::Int.Tensor invocations
gs-olive
changed the title
aten::Int.Tensor invocationsaten::Int.Tensor uses
gs-olive
force-pushed
the
replace_aten_int_schema
branch
from
8d43fb0 to
0ca6a52
Compare
- Implement lowering pass which detects canonical `aten::Int.Tensor` cases and recursively replaces input Value pointers until all 0D tensors have been resolved to their scalar components - Lowering pass is specialized to replacing strictly integer-typed Value pointers and can only trace through aten::mul and aten::floor_divide operators, which are two of the most common cases of use - Lowering pass traverses the graph until one of three base cases are encountered (or an invalid Value type is detected). These cases are `prim::NumToTensor`, `prim::Constant` (0D tensor), or simple integers. It then replaces the child nodes with the integer equivalents of the produced Tensors - Added extensive testing of new capabilities for accuracy, robustness, and functionality
gs-olive
force-pushed
the
replace_aten_int_schema
branch
from
0ca6a52 to
a86ac93
Compare
narendasan
reviewed
May 22, 2023
gs-olive
commented
May 22, 2023
| torch::jit::aten::floor_divide, | ||
| }; | ||
|
|
||
| c10::optional<torch::jit::Value*> Validate0DTensor(torch::jit::Value* value) { |
There was a problem hiding this comment.
Refactored to use c10::optional wrapper instead of nullptr + replaced pointer checks with .has_value()
- Edit in favor of `c10::optional` type usage
gs-olive
force-pushed
the
replace_aten_int_schema
branch
from
d3c0c7a to
15d9fcd
Compare
narendasan
pushed a commit
that referenced
this pull request
Jun 2, 2023
narendasan
pushed a commit
that referenced
this pull request
Jun 2, 2023
narendasan
pushed a commit
that referenced
this pull request
Jun 3, 2023
# for free
to join this conversation on GitHub.
Already have an account?
# 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.
Description
aten::Int.Tensorcases and recursively replaces input Value pointers until all 0D tensors have been resolved to their scalar componentsprim::NumToTensor,prim::Constant(0D tensor), or simple integers. It then replaces the child nodes with the integer equivalents of the produced TensorsPlease include a summary of the change and which issue is fixed. Please also include relevant motivation and context. List any dependencies that are required for this change.
Fixes #1880
Fixes #1836
Fixes #513
Type of change
Checklist: