Skip to content

Pull requests: OpenPipe/ART

New pull request New
51 Open 455 Closed
51 Open 455 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

[codex] Restore PR 607 KL loss removal
#639 opened Apr 2, 2026 by FurtherAI • Draft
feat: Add Qwen3.5 support to Megatron backend
#637 opened Apr 2, 2026 by vivekkalyan • Draft
feat: Add dedicated merged mode to Megatron backend
#636 opened Apr 1, 2026 by vivekkalyan • Draft
feat: Add dedicated lora mode to Megatron backend
#635 opened Mar 31, 2026 by vivekkalyan • Draft
Share training code between backends
#626 opened Mar 21, 2026 by Kovbo Loading…
5
Replace kl_penalty_reference_step with kl_penalty_step_lag
#625 opened Mar 20, 2026 by angkywilliam Loading…
3 tasks done
feat: Add KL support to PipelineTrainer backends
#624 opened Mar 20, 2026 by vivekkalyan Loading…
Support PipelineTrainer on serverless backend
#623 opened Mar 19, 2026 by Kovbo • Draft
Update dependencies and megatron for transformers==5.2.0.
#600 opened Mar 4, 2026 by FurtherAI Loading…
1
feat: add endpoint_url support to s3_sync for MinIO compatibility
#589 opened Mar 1, 2026 by koriyoshi2041 Loading…
5 tasks
fix: remove LocalBackend from art.__all__ and fix docs reference
#588 opened Mar 1, 2026 by koriyoshi2041 Loading…
4 tasks
fix: resolve ruler function/module name collision
#587 opened Mar 1, 2026 by koriyoshi2041 Loading…
5 tasks
1
feat: add dedicated inference data parallel support
#581 opened Feb 25, 2026 by vivekkalyan Loading…
fix: replace bare except with except Exception in server readiness check
#580 opened Feb 25, 2026 by haosenwang1018 Loading…
Unsloth sglang
#564 opened Feb 17, 2026 by pmukeshreddy Loading…
Track trajectory finish state using metadata
#563 opened Feb 15, 2026 by ashim-mahara Loading…
fix: move _internal_config assignment back into TrainableModel.__init__
#559 opened Feb 12, 2026 by saurabhbikram Loading…
feat: Megatron backend improvements for Docker and GPU memory
#558 opened Feb 12, 2026 by saurabhbikram Loading…
feat: add Dockerfile for GPU training with Megatron backend
#557 opened Feb 12, 2026 by saurabhbikram Loading…
fix: truncate excess scenarios instead of raising error in MCP generation
#556 opened Feb 12, 2026 by saurabhbikram Loading…
feat: add top-k entropy approximation for memory-efficient GRPO training
#555 opened Feb 12, 2026 by saurabhbikram Loading…
SGLang + Megatron: verl-style hybrid engine for ART RL training pipeline
#552 opened Feb 11, 2026 by pmukeshreddy Loading…
Sglang backend
#549 opened Feb 4, 2026 by pmukeshreddy Loading…
provide conceptual walkthrough
#547 opened Feb 3, 2026 by AakashKumarNain Loading…
5
normalize length
#508 opened Jan 10, 2026 by angkywilliam Loading…
ProTip! Exclude everything labeled bug with -label:bug.