Skip to content

[MPS] Fix static llama AOT tracing #2800

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Conversation

DenisVieriu97
Copy link
Collaborator

Fix static llama tracing for MPS backend:

  • AOT tracing:
python3 -m examples.apple.mps.scripts.mps_example --model_name "llama2" --use_partitioner  --no-use_fp16 --checkpoint stories110M.pt --params params.json

Testing:
AOT:

python3 -m examples.apple.mps.scripts.mps_example --model_name "llama2" --use_partitioner  --no-use_fp16 --checkpoint stories110M.pt --params params.json -b

Runtime:

./cmake-out/examples/apple/mps/mps_executor_runner --model_path ../../pytorch-executorch/executorch/llama2_mps_bundled_fp32.pte --bundled_program --num_runs 0

Output 0: tensor(sizes=[1, 1, 512], [
  0.271252, 0.547134, -0.319407, -0.519001, -0.817092, -0.511795, 1.19794, 0.169265, 0.74455, -0.59659,
  -0.754759, -0.688871, 0.298589, -0.443414, -0.086443, -0.565953, 0.864259, 0.115826, 0.85074, -0.525906,
  0.125811, -0.00499783, -0.463692, 0.29455, -0.910827, 0.539898, -0.958917, -0.160505, 0.345872, -0.33394,
  -0.481556, 1.23247, 0.972564, 1.2528, 0.74983, 0.117909, -0.537549, -0.11374, 0.0533589, -0.724433,
  0.267418, 0.383111, -0.589426, -0.273412, 0.536997, -0.296996, -0.037801, -0.467146, -0.111444, 0.2018,
  0.138447, -0.168977, 0.0633859, -0.147186, 0.162552, -0.340015, -0.481295, -0.285551, -0.486715, 0.266989,
  0.178192, -0.15646, -0.538977, -0.161788, -0.844837, -0.0201809, -0.0330472, -1.00392, -0.683174, -0.352963,
  0.0247648, 0.100644, 0.0750983, 0.617042, -0.0325033, 0.80563, 0.722587, -0.16737, -0.448206, -0.726605,
  0.0371236, -0.340444, 0.0844629, -0.744782, 0.713093, -1.23478, -0.624044, -0.927914, 0.226109, 0.871968,
  -0.181462, 0.737743, 0.630924, 0.352844, 1.22101, -0.433692, -0.545998, 1.25375, 0.604229, 0.567072,
  ...,
  -0.664146, -0.325717, 0.0465028, -0.0638815, -0.414189, -0.0774017, 0.664497, -0.80469, 0.665282, 0.142916,
  -0.136235, -0.181776, -0.0936792, 0.149341, 0.102156, 0.259336, 0.881158, 0.785714, -0.345178, -0.165404,
  0.100859, -0.43653, 0.239712, -0.111407, 0.744646, 0.651979, -0.15812, 0.0528999, 0.699308, 0.0331134,
  -0.102761, -0.0878261, -0.67926, -0.211967, 0.852889, -0.395876, 0.0924324, -0.262803, -0.622059, 0.550099,
  0.162211, -0.531762, 0.0518005, -0.920345, -0.014961, 0.132838, -0.0850867, 0.49853, -1.38389, 1.04421,
  0.220865, 0.160585, -0.239085, -0.745911, 1.19387, -0.597559, -0.722064, 0.267607, 0.65336, -0.360622,
  -0.20821, -0.522872, -0.40434, -0.593412, 0.918845, 1.05659, 0.266254, -0.438084, 0.626653, -0.232121,
  -0.346624, -0.533966, 0.179427, -0.509745, -0.187721, -0.253874, -0.286669, -0.462519, 0.569676, 0.0893508,
  -1.06513, 0.639664, 0.266718, -0.609418, 0.347195, 0.159792, 0.229622, -0.488133, 0.0085784, 0.485227,
  0.0878157, 0.556075, -0.933147, 0.742958, -0.064349, 0.198085, -0.314378, 0.173344, -0.718609, -1.14167,
])
I 00:00:02.547035 executorch:mps_executor_runner.mm:535] Model verified successfully.

cc @cccclai , @shoumikhin

Copy link

pytorch-bot bot commented Apr 1, 2024

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/2800

Note: Links to docs will display an error until the docs builds have been completed.

❌ 1 New Failure, 1 Unrelated Failure

As of commit 2577dc1 with merge base daa217f (image):

NEW FAILURE - The following job has failed:

BROKEN TRUNK - The following job failed but were present on the merge base:

👉 Rebase onto the `viable/strict` branch to avoid these failures

This comment was automatically generated by Dr. CI and updates every 15 minutes.

@facebook-github-bot facebook-github-bot added the CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. label Apr 1, 2024
@facebook-github-bot
Copy link
Contributor

@cccclai has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

@cccclai
Copy link
Contributor

cccclai commented Apr 2, 2024

Any specific reason we're not using fp16?

@facebook-github-bot
Copy link
Contributor

@cccclai has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

@facebook-github-bot
Copy link
Contributor

@cccclai merged this pull request in c731673.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. Merged
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants