Skip to content

Pull requests: PrimeIntellect-ai/prime-rl

New pull request New
84 Open 2,003 Closed
84 Open 2,003 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

feat(orchestrator): bump verifiers to 3b77145, log per-turn timings
#2379 opened Apr 30, 2026 by mikasenghaas Member • Draft
feat: agentic length-penalty ablations — weighted shaping over tool-response tokens
#2377 opened Apr 29, 2026 by hallerite Member • Draft
feat(trainer): add single-point TPS/MFU API for RL
#2376 opened Apr 29, 2026 by samsja Member Loading…
fix(monitor): retry wandb.init on transient CommError + bump wandb
#2374 opened Apr 28, 2026 by mikasenghaas Member Loading…
fix(elastic): circuit breaker for adapter sync failures
#2370 opened Apr 28, 2026 by eexwhyzee Contributor Loading…
MoE refactor + SonicMoE
#2364 opened Apr 27, 2026 by S1ro1 Collaborator Loading…
7
Fix/loss scale skew
#2359 opened Apr 25, 2026 by ijamil1 • Draft
1
fix(monitor): split httpx timeouts so large sample uploads don't WriteTimeout
#2357 opened Apr 25, 2026 by JannikSt Member Loading…
2
fix(multi_ckpt): run maybe_clean on all ranks to keep ckpt_steps in sync
#2354 opened Apr 25, 2026 by JannikSt Member Loading…
4
orchestrator: set up early stopping status for zero advantages filtering
#2353 opened Apr 24, 2026 by eexwhyzee Contributor Loading…
5
fix: correct multi-env example_id collision in /all/ metrics
#2351 opened Apr 24, 2026 by jfeldstein Loading…
4
Fix gemma
#2347 opened Apr 23, 2026 by Jackmin801 Member Loading…
3
Fix shared W&B config display
#2345 opened Apr 22, 2026 by MrAnayDongre Loading…
1
1
Qwen3-235B NIXL + torch profiler support (draft)
#2344 opened Apr 22, 2026 by S1ro1 Collaborator • Draft
feat: optional MX Server rendezvous overlay (on top of #2326)
#2343 opened Apr 22, 2026 by KavinKrishnan Loading…
5 tasks
6
feat(qwen3_moe): support in-weight-transfer FP8 quantization
#2341 opened Apr 22, 2026 by S1ro1 Collaborator Loading…
3
feat: add orch2 — async-native orchestrator redesign
#2336 opened Apr 21, 2026 by samsja Member • Draft
19 of 27 tasks
10
feat: enable vllm-router for single-node inference
#2334 opened Apr 20, 2026 by bedeks Loading…
1
11
Add orchestrator-side replay buffer support for RL training
#2332 opened Apr 19, 2026 by taivu1998 Loading…
1
2
feat: replace max_async_level with on_policy debug flag
#2328 opened Apr 19, 2026 by samsja Member Loading…
12
Nixl weight transfer
#2326 opened Apr 19, 2026 by S1ro1 Collaborator Loading…
28
Feat: VLLM weight transfer pause deadlock
#2325 opened Apr 19, 2026 by S1ro1 Collaborator Loading…
hrl dev90 cherry-pick #2308
#2309 opened Apr 17, 2026 by JannikSt Member Loading…
Pin Triton JIT cache to node-local /tmp per job
#2305 opened Apr 17, 2026 by rasdani Contributor Loading…
feat: proper context parallelism for GLM-5 sparse MLA
#2293 opened Apr 17, 2026 by samsja Member • Draft
ProTip! Mix and match filters to narrow down what you’re looking for.