[pull] main from pytorch:main by pull[bot] · Pull Request #431 · garymm/tutorials · GitHub
Skip to content

[pull] main from pytorch:main#431

Merged
pull[bot] merged 2 commits into
garymm:mainfrom
pytorch:main
Jun 11, 2026
Merged

[pull] main from pytorch:main#431
pull[bot] merged 2 commits into
garymm:mainfrom
pytorch:main

Conversation

@pull

@pull pull Bot commented Jun 11, 2026

Copy link
Copy Markdown

See Commits and Changes for more details.


Created by pull[bot] (v2.0.0-alpha.4)

Can you help keep this open source service alive? 💖 Please sponsor : )

yushangdi and others added 2 commits June 11, 2026 09:59
## Summary

`make download` fails on any build shard that runs `fgsm_tutorial`
because the LeNet/MNIST checkpoint it depends on was fetched from a
Google Drive link that now returns **HTTP 404**:

```
.jenkins/download_data.py -> download_lenet_mnist()
  https://docs.google.com/uc?export=download&id=1HJV2nUHJqclXQ8flKvcWmjZ-OU5DGatl
  -> urllib.error.HTTPError: HTTP Error 404: Not Found
make: *** [Makefile:73: docs] Error 2
```

This is a pre-existing, repo-wide breakage (the link died after the last
`main` build, so it only surfaces on PRs whose shard rebalancing puts
`fgsm_tutorial` on a worker that exercises the download).

## Fix

Point `download_lenet_mnist()` at a public S3 mirror on
`pytorch-tutorial-assets` (the same host the DCGAN download already
uses) instead of the dead Drive link.

The checkpoint was retrained to match the exact `Net` architecture in
`beginner_source/fgsm_tutorial.py` (conv1 32x1x3x3 / conv2 / fc1
9216x128 / fc2), so it loads with a strict `load_state_dict`.

## Verification

- Anonymous public fetch of the new S3 URL returns HTTP 200 with the
pinned sha256.
- Ran `download_lenet_mnist()` end-to-end: file lands at
`beginner_source/data/lenet_mnist_model.pth`, sha256 matches.
- Confirmed the checkpoint loads strictly into the current tutorial
`Net` and runs an FGSM step.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This tutorial demonstrates how to use CUDA graph kernel annotations for
semantic profiling traces with custom visualization lanes.

Features:
- End-to-end workflow from graph capture to visualization
- Transformer block example with annotated regions
- Post-processing to merge annotations into profiler traces
- Custom stream assignments for semantic organization

The tutorial includes:
- mark_kernels() context manager usage
- Graph capture with enable_annotations=True
- Profiling and trace post-processing
- Before/after comparison
- Troubleshooting guide

Fixes #ISSUE_NUMBER

## Description
<!--- Describe your changes in detail -->

## Checklist
<!--- Make sure to add `x` to all items in the following checklist: -->
- [ ] The issue that is being fixed is referred in the description (see
above "Fixes #ISSUE_NUMBER")
- [ ] Only one issue is addressed in this pull request
- [ ] Labels from the issue that this PR is fixing are added to this
pull request
- [ ] No unnecessary issues are included into this pull request.

---------

Co-authored-by: yushangdi <yushangdi@users.noreply.github.com>
Co-authored-by: Claude Sonnet 4.5 <noreply@anthropic.com>
@pull pull Bot locked and limited conversation to collaborators Jun 11, 2026
@pull pull Bot added the ⤵️ pull label Jun 11, 2026
@pull pull Bot merged commit 708b21c into garymm:main Jun 11, 2026
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant