[atalman]

Summary

Branch-cut step to start testing the 2.15 release in tutorials. Mirrors #3955 ("Enable nightly build to test 2.14").

• .jenkins/build.sh: pin torch==2.15.0 (from the test channel, cu130) on the USE_NIGHTLY path and the commented reference line; update the 2.14 -> 2.15 comment.

No workflow change is needed: the push on main trigger in .github/workflows/build-tutorials-nightly.yml is still enabled from the 2.13 cycle, so the nightly tutorials build already runs on main and will pick up the 2.15 RC binaries with this pin.

.ci/docker/requirements.txt is intentionally left at torch==2.14 — that gets bumped to 2.15 at the later "move to stable binaries" step, once 2.15 is released.

Verification

torch==2.15.0+cu130 is published on the test channel for cp312 on both x86_64 and aarch64. Resolving the exact install command from build.sh against that index (uv pip install --dry-run) picks:

torch==2.15.0+cu130
torchvision==0.29.2+cu130
torchaudio==2.11.1+cu130

bash -n .jenkins/build.sh passes. The nightly job itself has not run yet.

The channel stays cu130, as in the 2.12 to 2.14 cycles. 2.15's default wheel is CUDA 13.2, and 2.15.0 is also on the test channel for cu132 and cu134; moving the tutorials job to one of those is a separate change.

Release playbook reference (2.14 -> 2.15)

Step	2.14	2.15
Enable nightly build to test the RC	#3955	this PR
Move to stable binaries (requirements.txt)	—	follow-up at release

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/tutorials/3983

• 📄 Preview Python docs built from this PR

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

⏳ No Failures, 2 Pending

As of commit 4ffceae with merge base c8ccbbf ():
💚 Looks good so far! There are no failures yet. 💚

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