Skip to content

Pull requests: allenai/open-instruct

New pull request New
54 Open 1,158 Closed
54 Open 1,158 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

Require checkpoint on Beaker restarts for DPO and GRPO training codex
#1469 opened Feb 10, 2026 by finbarrtimbers Loading…
5
Enable tensor parallelism in DPO
#1467 opened Feb 9, 2026 by finbarrtimbers Loading…
3 tasks done
6
Enable packing + compile for DPO with wasted tokens metric
#1466 opened Feb 6, 2026 by finbarrtimbers Loading…
21
add user_prompt_transform to allow easy formatting of prefix, suffix of prompts
#1465 opened Feb 6, 2026 by mnoukhov Loading…
3
Pass@K for local eval
#1464 opened Feb 6, 2026 by mnoukhov • Draft
6
WIP: Add PPO with value model support to grpo_fast.py
#1462 opened Feb 4, 2026 by hamishivi • Draft
1 of 6 tasks
6
Add model merging scripts for Beaker
#1459 opened Feb 4, 2026 by natolambert Loading…
4 tasks done
24
Add generic RL environment support
#1453 opened Feb 2, 2026 by hamishivi Loading…
2 of 3 tasks
22
Now, dpo.py matches dpo_tune_cache.py almost perfectly on the single GPU experiments
#1451 opened Feb 1, 2026 by finbarrtimbers Loading…
2
Tillicum slurm launcher
#1450 opened Jan 31, 2026 by hamishivi • Draft
9
Avoid sync CUDA transfer in data loader codex
#1443 opened Jan 30, 2026 by finbarrtimbers Loading…
4
Add DPO OLMo-core support with MFU improvements
#1440 opened Jan 30, 2026 by finbarrtimbers Loading…
3 tasks
9
Fixes SFT checkpointing
#1435 opened Jan 28, 2026 by finbarrtimbers Loading…
7
Significantly improves dpo.py performance: ~40% MFU
#1430 opened Jan 27, 2026 by finbarrtimbers • Draft
3
Runs the benchmarks with the hybrid model
#1425 opened Jan 26, 2026 by finbarrtimbers Loading…
4
Add tool docs
#1410 opened Jan 21, 2026 by hamishivi Loading…
10
Now, the GPU tests CI action automatically appends the result to prevent it from re-running.
#1409 opened Jan 21, 2026 by finbarrtimbers Loading…
2
Validates artifacts
#1407 opened Jan 21, 2026 by finbarrtimbers Loading…
7
Add optional wandb system metrics logging for generator process
#1403 opened Jan 20, 2026 by jacob-morrison • Draft
1
Add GRPO main entry point and scripts (GRPO olmo-core: PR 5 of 5)
#1399 opened Jan 20, 2026 by finbarrtimbers Loading…
1 of 3 tasks
7
Add OLMo-core Ray actor (GRPO olmo-core: PR 4 of 5)
#1398 opened Jan 20, 2026 by finbarrtimbers Loading…
1 of 2 tasks
12
Add GRPO callbacks for OLMo-core Trainer (GRPO olmo-core: PR 3 of 5)
#1397 opened Jan 20, 2026 by finbarrtimbers Loading…
17
Use simple-parsing for DPO argument parsing
#1393 opened Jan 20, 2026 by finbarrtimbers Loading…
3 tasks
4
Refactor DPO config: move fields and remove duplicates
#1392 opened Jan 20, 2026 by finbarrtimbers Loading…
3 tasks
2
Add OLMo-core GRPO trainer implementation
#1389 opened Jan 20, 2026 by finbarrtimbers Loading…
17
ProTip! Exclude everything labeled bug with -label:bug.