build: migrate the GPU stack to CUDA 13 - #729
Draft
bradhilton wants to merge 1 commit into
Draft
bradhilton wants to merge 1 commit into
bradhilton wants to merge 1 commit into
Conversation
bradhilton
changed the title
This branch has not been deployed
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Agent: Unattributed
Migrate ART’s GPU dependencies, lockfiles, image and CI/dev configuration from CUDA 12.8 to CUDA 13. Use cu130 Torch/TorchVision/torchao wheels, TransformerEngine cu13, updated vLLM pins and rebuilt CUDA extensions.
Historical validation on a CoreWeave H200 with driver 595.71.05 and CUDA toolkit 13.0 covered lock/sync checks, CUDA and extension imports, 14 shared-library dependency checks, and 21 focused Megatron tests. The retained image was built and prewarmed as
docker.io/bradhiltonnw/art-gpu:cuda13-codex-20260612-122156, digestsha256:009d5aa9cbe46a5a2855d44adc6bc3c272b268eb264bccc77e0d247f1b4ce7f0.The tiny Kubernetes dev smoke reached train/eval but skipped gradient updates because a rollout had no advantage signal. Unsloth still pulled an xFormers wheel with a cu128 warning; that path was not exercised by the Megatron/flex-attention smoke.
Draft. Responsible agent, current compatibility and readiness are unverified; these are historical results.
Historical agent ownership is unverified; this remains in the unattributed bucket.