Skip to content

Arctic rl/integration - #264

Merged
sfc-gh-jrasley merged 48 commits into
mainfrom
arctic_rl/integration
Jun 24, 2026
Merged

Arctic rl/integration#264
sfc-gh-jrasley merged 48 commits into
mainfrom
arctic_rl/integration

Conversation

@sfc-gh-mhidayetoglu

Copy link
Copy Markdown
Contributor

No description provided.

sfc-gh-yewang and others added 30 commits February 23, 2026 20:57
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: jenkins-snowci-mergequeue-0-emu[bot] <232694648+jenkins-snowci-mergequeue-0-emu[bot]@users.noreply.github.com>
Co-authored-by: Sourcegraph <batch-changes@sourcegraph.com>
Co-authored-by: EMU Migration Bot <emu-migration@snowflake.com>
Cherry-picked from public repo: #258

Co-authored-by: Xinyu Lian <lian7@illinois.edu>
Made-with: Cursor
Co-authored-by: YanlinDu <yanlin.du@snowflake.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Signed-off-by: Stas Bekman <stas.bekman@snowflake.com>
Signed-off-by: Stas Bekman <stas.bekman@snowflake.com>
sfc-gh-mhidayetoglu and others added 12 commits May 24, 2026 22:24
Compare sender HF names against the receiver model's expected set and fail
loud on mismatch so Qwen3-32B colocated runs catch architecture drift early.

Co-authored-by: Cursor <cursoragent@cursor.com>
- receiver: raise on incomplete CUDA IPC load (tensor-count mismatch),
  validate received parameter names on NCCL/CPU paths, and accept either
  serialized bytes or (name, tensor) pairs in load_weights_from_cpu
- replica_pool: raise RuntimeError when any worker's sync_weights status
  is not "done" instead of swallowing the failure
- api/multi_model: add /spec_weights_info and /sync_spec_weights endpoints
  plus Driver forwarding for multi-model mode

Co-authored-by: Cursor <cursoragent@cursor.com>
Add load_weights_cuda_ipc_chunk to the weight-sync receiver, which loads
one param (or small chunk) at a time from CUDA IPC handles instead of
requiring the whole model in one payload. It raises on any GPU missing a
handle and runs full param-name validation on the final chunk
(context=cuda_ipc_stream). Add the matching InferenceWorker
collective_rpc wrapper. This bounds peak extra GPU memory during sync to
one full param per GPU.

Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Signed-off-by: Stas Bekman <stas.bekman@snowflake.com>
Comment thread .github/CODEOWNERS Outdated
Comment thread fix_health_monitor.py Outdated
Comment thread fix_health_monitor2.py Outdated
Comment thread pyproject.toml Outdated

@sfc-gh-jrasley sfc-gh-jrasley left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

please ad

@sfc-gh-jrasley
sfc-gh-jrasley merged commit a1bf61a into main Jun 24, 2026
2 checks passed
@sfc-gh-jrasley
sfc-gh-jrasley deleted the arctic_rl/integration branch June 24, 2026 19:55
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

8 participants