Skip to content

Pull requests: allenai/open-instruct

New pull request New
36 Open 1,431 Closed
36 Open 1,431 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

Add use_cpu_adam flag for GRPO fast training
#1737 opened Jun 24, 2026 by Chessing234 Contributor Loading…
2 tasks
1
3
Fix token-weighted loss under gradient accumulation in finetune.py
#1736 opened Jun 24, 2026 by Chessing234 Contributor Loading…
2 tasks
1
3
Add tiled GRPO lm-head loss
#1735 opened Jun 23, 2026 by hamishivi Collaborator Loading…
32
Add tool-schema support to SFT tokenization
#1734 opened Jun 23, 2026 by hamishivi Collaborator Loading…
28
Add DPPO loss function
#1733 opened Jun 23, 2026 by hamishivi Collaborator Loading…
4
Harden DockerBackend with retries, OOM detection, and concurrency limits
#1732 opened Jun 23, 2026 by hamishivi Collaborator Loading…
13
Make ModelDims.from_hf_config robust to explicit head_dim
#1731 opened Jun 23, 2026 by hamishivi Collaborator Loading…
3
Increase default environment pool acquire timeout to 7200s
#1729 opened Jun 23, 2026 by hamishivi Collaborator Loading…
Fix get_wandb_tags tag-limit handling and log all GRPO configs
#1727 opened Jun 17, 2026 by mnoukhov Contributor Loading…
3
Add OLMo-core 32B GRPO launch scripts
#1726 opened Jun 16, 2026 by mnoukhov Contributor Loading…
5
Now, DPO gets ~32% MFU.
#1720 opened Jun 10, 2026 by finbarrtimbers Collaborator • Draft
Fix IFEvalG word-counting checkers to ignore punctuation
#1718 opened Jun 9, 2026 by pdudeia Loading…
4
Now, we support the hybrid model in our Olmo-core code.
#1713 opened Jun 2, 2026 by finbarrtimbers Collaborator Loading…
8
Wire max_checkpoints through SFT, DPO, and GRPO paths
#1701 opened May 27, 2026 by TimDettmers Loading…
4 tasks
7
Add olmo-eval Beaker launch integration for GRPO
#1698 opened May 22, 2026 by mnoukhov Contributor • Draft
2 of 3 tasks
7
Add Trackio rollout trace logging
#1697 opened May 21, 2026 by abidlabs Loading…
6
fix: guard against empty choices and null message in LLM responses
#1695 opened May 18, 2026 by qizwiz Loading…
12
Fix CSV header handling in benchmark_generators; pathlib throughout
#1684 opened May 12, 2026 by finbarrtimbers Collaborator Loading…
5
Adds difficulty sampling curriculum dataloader and dataset builder
#1661 opened May 6, 2026 by undfined Collaborator Loading…
7
Replace submit_eval_jobs.py with thin wrapper around submit_eval_jobs.sh
#1658 opened May 6, 2026 by finbarrtimbers Collaborator Loading…
2 tasks
5
Add time/per_group_wall_time metric
#1656 opened May 5, 2026 by finbarrtimbers Collaborator Loading…
2 tasks
6
Add use_cpu_adam CLI toggle and offload validation for GRPO
#1654 opened May 3, 2026 by Bhavyashah20 Loading…
6 tasks
1
5
Make checkpointing better
#1647 opened Apr 29, 2026 by finbarrtimbers Collaborator • Draft
3 tasks
4
Fix submit_eval_jobs.py for olmo-eval-internal runs
#1644 opened Apr 28, 2026 by finbarrtimbers Collaborator Loading…
4 tasks done
2
Offline Distillation via DistillKit (Part Three - Offline Distillation)
#1629 opened Apr 21, 2026 by wolfecameron Contributor Loading…
6
ProTip! Find all pull requests that aren't related to any open issues with -linked:issue.