Qualcomm AI Engine Direct - LLAMA2 Infrastructure #2020
Closed
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
Contributor
chunit-quic
commented
Feb 21, 2024
chunit-quic
commented
- OPs
- Add pow_tensor_scalar op
- Add rsqrt op
- Add sigmoid op
- Refine axis handling of cat op
- Refine parameters related functions
- Passes
- Add AnnotateDecomposed for unbind and stak op
- Add DecomposeSilu for quantizer
- Add ReplaceInfBuffer for quantizer
- Change pass name ConvertAddmmmmWithLinear to ConvertToLinear
- Change pass name ConvertScaledDotProductAttention to DecomposeScaledDotProductAttention
- Support more args for sdpa op in DecomposeScaledDotProductAttention
- Support mm case for ConvertToLinear
- Move q_ops and dq_ops to pass/utils.py
- Tests
- Add dummy llama2 test script
- Add single op test cases
- Others
- Fix error of popping missing buffer
- Reorder the order of test models
- Reorder the order of op in qnn_constant
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/2020
Note: Links to docs will display an error until the docs builds have been completed. ❗ 1 Merge Blocking SEVsThere is 1 active merge blocking SEVs. Please view them below:
If you must merge, use ✅ No FailuresAs of commit fab30bc with merge base 8fed60b ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
facebook-github-bot
added
the
CLA Signed
|
@cccclai has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
cccclai
approved these changes
Feb 21, 2024
|
@cccclai has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
1. OPs - Add pow_tensor_scalar op - Add rsqrt op - Add sigmoid op - Refine axis handling of cat op - Refine parameters related functions 2. Passes - Add AnnotateDecomposed for unbind and stak op - Add DecomposeSilu for quantizer - Add ReplaceInfBuffer for quantizer - Change pass name ConvertAddmmmmWithLinear to ConvertToLinear - Change pass name ConvertScaledDotProductAttention to DecomposeScaledDotProductAttention - Support more args for sdpa op in DecomposeScaledDotProductAttention - Support mm case for ConvertToLinear - Move q_ops and dq_ops to pass/utils.py 3. Tests - Add dummy llama2 test script - Add single op test cases 4. Others - Fix error of popping missing buffer - Reorder the order of test models - Reorder the order of op in qnn_constant
chunit-quic
force-pushed
the
llama_infra
branch
from
abbadfc to
fab30bc
Compare
|
@cccclai has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
cccclai
reviewed
Feb 22, 2024
| @@ -250,7 +250,10 @@ def _partition_and_lower_one_graph_module( | |||
| # Delete the consumed buffers | |||
| buffer_name = toplevel_signature.inputs_to_buffers.pop(node.name) | |||
| toplevel_signature.buffers.remove(buffer_name) | |||
| owning_program.state_dict.pop(buffer_name) | |||
| if buffer_name in owning_program.state_dict: | |||
There was a problem hiding this comment.
Oh thanks for catching this case
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.