| #834 — Add Tau^2 Bench training environment |
Maintainers |
reviewer not assigned |
2026-03-06 22:23 UTC |
2026-03-09 22:23 UTC |
146d 21h |
| #1103 — structured_outputs: support oneOf/anyOf branches and $defs/$ref schemas |
Maintainers |
reviewer not assigned |
2026-04-21 05:08 UTC |
2026-04-22 05:08 UTC |
115d 14h |
| #1233 — structured_outputs: extend XML type coercion to handle oneOf/anyOf/allOf composition |
Maintainers |
reviewer not assigned |
2026-05-05 21:17 UTC |
2026-05-06 21:17 UTC |
104d 22h |
| #1274 — Add audiobench_alpaca_audio benchmark + audiobench_judge resource server |
Maintainers |
reviewer not assigned |
2026-05-08 20:49 UTC |
2026-05-11 20:49 UTC |
101d 22h |
| #1315 — Add GitHub issue template for environment integrations |
Maintainers |
reviewer not assigned |
2026-05-13 19:42 UTC |
2026-05-14 19:42 UTC |
98d 23h |
| #1358 — feat(vllm_model): chunk long single-clip audio in chat_completions |
Maintainers |
reviewer not assigned |
2026-05-18 21:04 UTC |
2026-05-19 21:04 UTC |
95d 22h |
| #1429 — docs: clarify SWE-bench dataset downloads |
Maintainers |
reviewer not assigned |
2026-05-27 08:53 UTC |
2026-05-28 08:53 UTC |
89d 10h |
| #1590 — feat(equivalence_llm_judge): cap extracted-answer length sent to judge |
Maintainers |
reviewer not assigned |
2026-06-12 22:06 UTC |
2026-06-15 22:06 UTC |
76d 21h |
| #1720 — feat(instruction_following): add loose-mode evaluation and compute_metrics |
Maintainers |
reviewer not assigned |
2026-06-25 01:12 UTC |
2026-06-26 01:12 UTC |
68d 18h |
| #1745 — fix(config): seed no-model refs for prepare-data |
Maintainers |
reviewer not assigned |
2026-06-26 00:31 UTC |
2026-06-29 00:31 UTC |
67d 18h |
| #1797 — fix(inference_provider): normalize provider endpoint errors |
Maintainers |
reviewer not assigned |
2026-06-27 04:43 UTC |
2026-06-29 04:43 UTC |
67d 14h |
| #1800 — Harden SWE agent test file parsing |
Maintainers |
reviewer not assigned |
2026-06-27 10:32 UTC |
2026-06-29 10:32 UTC |
67d 8h |
| #1847 — Rgala/browsecomp search providers newstable |
Maintainers |
reviewer not assigned |
2026-07-01 00:35 UTC |
2026-07-02 00:35 UTC |
64d 18h |
| #1908 — Actions: sync project status to release label |
Maintainers |
reviewer not assigned |
2026-07-02 13:16 UTC |
2026-07-03 13:16 UTC |
63d 6h |
| #1939 — Tb e2e |
Maintainers |
reviewer not assigned |
2026-07-06 18:22 UTC |
2026-07-07 18:22 UTC |
61d 1h |
| #2080 — feat: Switchyard OpenCode integration |
Maintainers |
reviewer not assigned |
2026-07-20 12:50 UTC |
2026-07-21 12:50 UTC |
51d 6h |
| #2109 — fix(server_utils): use unpredictable per-process secret for session signing |
Maintainers |
reviewer not assigned |
2026-07-22 21:53 UTC |
2026-07-23 21:53 UTC |
48d 21h |
| #2027 — Simplify ModelCallRecord; Refactor call capture logic; Extend to agents |
Maintainers |
reviewer not assigned |
2026-07-28 06:26 UTC |
2026-07-29 06:26 UTC |
45d 13h |
| #2254 — fix(hermes_agent): never present fabricated tokens as policy output |
Maintainers |
reviewer not assigned |
2026-07-30 21:27 UTC |
2026-07-31 21:27 UTC |
42d 21h |
| #2437 — feat(agents): add aggregate metrics proxy helper |
Maintainers |
reviewer not assigned |
2026-08-08 12:41 UTC |
2026-08-10 12:41 UTC |
37d 6h |
| #2412 — fix(pinchbench): address the model server per rollout so capture attributes calls |
Maintainers |
reviewer not assigned |
2026-08-07 13:59 UTC |
2026-08-10 13:59 UTC |
37d 5h |
| #2513 — fix: serialize Jest in nv-internal SWE |
Maintainers |
reviewer not assigned |
2026-08-12 13:56 UTC |
2026-08-13 13:56 UTC |
34d 5h |
| #2526 — feat(core): add shared failure-routing helpers |
Maintainers |
reviewer not assigned |
2026-08-13 02:08 UTC |
2026-08-14 02:08 UTC |
33d 17h |
| #2527 — fix: bound model request retries |
Maintainers |
reviewer not assigned |
2026-08-13 09:39 UTC |
2026-08-14 09:39 UTC |
33d 9h |
| #2553 — feat: add OrcaRouter as a named inference provider |
Maintainers |
reviewer not assigned |
2026-08-14 03:26 UTC |
2026-08-17 03:26 UTC |
32d 16h |
| #2578 — [Super 3.5 evals] Apply measured full-rack vLLM configuration |
Maintainers |
reviewer not assigned |
2026-08-17 18:10 UTC |
2026-08-18 18:10 UTC |
31d 1h |
| #2594 — feat(swebench): configurable image namespace and arch for eval containers |
Maintainers |
reviewer not assigned |
2026-08-18 14:50 UTC |
2026-08-19 14:50 UTC |
30d 4h |
| #2602 — docs: mention Cline harness in SWE agents README |
Maintainers |
reviewer not assigned |
2026-08-18 20:29 UTC |
2026-08-19 20:29 UTC |
29d 22h |
| #2605 — Add live web client for Kuhn poker |
Maintainers |
reviewer not assigned |
2026-08-19 01:15 UTC |
2026-08-20 01:15 UTC |
29d 18h |
| #2700 — feat(gymv): add multimodal environment integration |
Maintainers |
reviewer not assigned |
2026-08-24 04:01 UTC |
2026-08-25 04:01 UTC |
26d 15h |
| #2772 — fix(sandbox): implement canonical task environment resolution |
Maintainers |
reviewer not assigned |
2026-08-25 22:45 UTC |
2026-08-26 22:45 UTC |
24d 20h |
| #2613 — feat(observability): return the environment session id and allow correlation without full capture |
Maintainers |
reviewer not assigned |
2026-08-27 09:36 UTC |
2026-08-28 09:36 UTC |
23d 9h |
| #2838 — fix: preserve batch wrapper exit status |
Maintainers |
reviewer not assigned |
2026-08-28 04:26 UTC |
2026-08-31 04:26 UTC |
22d 15h |
| #2841 — feat(primevul): add PrimeVul paired vulnerability-detection benchmark |
Maintainers |
reviewer not assigned |
2026-08-28 08:45 UTC |
2026-08-31 08:45 UTC |
22d 10h |
| #2836 — perf(vllm): derive required prefix token IDs in Gym |
Maintainers |
reviewer not assigned |
2026-08-31 17:20 UTC |
2026-09-01 17:20 UTC |
21d 2h |
| #2907 — fix(swerl_gen): gen_eval_scripts.py always exits with TypeError (no --image-dir option) |
Maintainers |
reviewer not assigned |
2026-09-01 01:18 UTC |
2026-09-02 01:18 UTC |
20d 18h |
| #2910 — keep resolved config out of process arguments |
Maintainers |
reviewer not assigned |
2026-09-01 03:16 UTC |
2026-09-02 03:16 UTC |
20d 16h |
| #2915 — test(ci): AUT-2197 catch untestable component directories |
Maintainers |
reviewer not assigned |
2026-09-01 06:51 UTC |
2026-09-02 06:51 UTC |
20d 12h |
| #2923 — Merge main into gdpval-devel |
Maintainers |
reviewer not assigned |
2026-09-01 11:42 UTC |
2026-09-02 11:42 UTC |
20d 7h |
| #2978 — fix(osworld): stabilize long-horizon Molt rollouts |
Maintainers |
reviewer not assigned |
2026-09-02 15:38 UTC |
2026-09-03 15:38 UTC |
19d 3h |
| #2980 — feat(anyterminal): OpenSandbox oracle config, resource knobs, cgroup metrics, TB2.1 prepare |
Maintainers |
reviewer not assigned |
2026-09-02 17:43 UTC |
2026-09-03 17:43 UTC |
19d 1h |
| #2932 — [GDPVal] Merge main into gdpval-devel |
Maintainers |
reviewer not assigned |
2026-09-03 17:12 UTC |
2026-09-04 17:12 UTC |
18d 2h |
| #3123 — docs(skills): align repository skills with current contracts |
Maintainers |
reviewer not assigned |
2026-09-05 05:26 UTC |
2026-09-07 05:26 UTC |
17d 14h |
| #3085 — feat(nooa-agent): add native lifecycle integration and capability example |
Maintainers |
reviewer not assigned |
2026-09-04 12:06 UTC |
2026-09-07 12:06 UTC |
17d 7h |
| #3096 — feat: add SAV tracking verifier resources server |
Maintainers |
reviewer not assigned |
2026-09-04 18:27 UTC |
2026-09-07 18:27 UTC |
17d 0h |
| #3132 — [ApexAgents] Serve GLM-5.2 with CUDA graphs, pin replicas to one rack, tolerate partial readiness |
Maintainers |
reviewer not assigned |
2026-09-05 21:55 UTC |
2026-09-07 21:55 UTC |
16d 21h |
| #3109 — feat(aviary): support post-terminal model response |
Maintainers |
reviewer not assigned |
2026-09-04 22:04 UTC |
2026-09-07 22:04 UTC |
16d 21h |
| #3209 — feat(benchmarks): remove the dataset-level agent pin |
Maintainers |
reviewer not assigned |
2026-09-10 05:34 UTC |
2026-09-11 05:34 UTC |
13d 13h |
| #3318 — Jnolan/gdpval hermes upstream |
Maintainers |
reviewer not assigned |
2026-09-11 08:11 UTC |
2026-09-14 08:11 UTC |
12d 11h |
#3372 — feat(cli): add gym eval intake to submit a rundir to a benchmark-run data layer |
Maintainers |
reviewer not assigned |
2026-09-14 17:42 UTC |
2026-09-15 17:42 UTC |
11d 1h |
| #3412 — feat(checkpoint): prefix token level recovery |
Maintainers |
reviewer not assigned |
2026-09-15 18:17 UTC |
2026-09-16 18:17 UTC |
10d 1h |
| #3430 — Add critpt_custom_grader: a sandboxed, locally executed grader for CritPt-style problems |
Maintainers |
reviewer not assigned |
2026-09-16 23:56 UTC |
2026-09-17 23:56 UTC |
8d 19h |
| #3494 — feat: post-train environment |
Maintainers |
reviewer not assigned |
2026-09-17 16:55 UTC |
2026-09-18 16:55 UTC |
8d 2h |
| #3502 — feat: Allow explicit input files in eval runs |
Maintainers |
reviewer not assigned |
2026-09-17 20:45 UTC |
2026-09-18 20:45 UTC |
7d 22h |
| #3503 — feat(browsecomp): add Brave search provider |
Maintainers |
reviewer not assigned |
2026-09-17 21:32 UTC |
2026-09-18 21:32 UTC |
7d 21h |
| #3504 — feat(video): add SA-V tracks verifier |
Maintainers |
reviewer not assigned |
2026-09-17 21:52 UTC |
2026-09-18 21:52 UTC |
7d 21h |
| #3507 — feat: add SAV tracks verifier resources server |
Maintainers |
reviewer not assigned |
2026-09-17 23:20 UTC |
2026-09-18 23:20 UTC |
7d 20h |
| #3560 — feat(core): add a warm process pool for CPU-heavy verification |
Maintainers |
reviewer not assigned |
2026-09-20 00:05 UTC |
2026-09-21 00:05 UTC |
7d 19h |
| #3512 — docs(example_session_state_mgmt): document in-memory session state |
Maintainers |
reviewer not assigned |
2026-09-18 01:59 UTC |
2026-09-21 01:59 UTC |
7d 17h |
| #3520 — fix(terminus): parallel tool calls formatting |
Maintainers |
reviewer not assigned |
2026-09-18 07:42 UTC |
2026-09-21 07:42 UTC |
7d 11h |
| #3526 — feat(inference-provider): add Doubleword flex configuration |
Maintainers |
reviewer not assigned |
2026-09-18 11:15 UTC |
2026-09-21 11:15 UTC |
7d 8h |
| #3146 — Add the Blobfish enterprise-operations resources server and ERPBench-100 |
Maintainers |
reviewer not assigned |
2026-09-20 15:30 UTC |
2026-09-21 15:30 UTC |
7d 3h |
| #3561 — fix(config): fail fast with a clean error on a malformed head_server block |
Maintainers |
reviewer not assigned |
2026-09-20 19:36 UTC |
2026-09-21 19:36 UTC |
6d 23h |
| #3556 — feat(sandbox): support Docker pipe process sessions |
Maintainers |
reviewer not assigned |
2026-09-19 20:24 UTC |
2026-09-21 20:24 UTC |
6d 23h |
| #3548 — feat(checkpoint): add indirect prompt injection state recovery |
Maintainers |
reviewer not assigned |
2026-09-18 22:51 UTC |
2026-09-21 22:51 UTC |
6d 20h |
| #3563 — feat(checkpoint): multi-worker support for policy model server |
Maintainers |
reviewer not assigned |
2026-09-21 03:43 UTC |
2026-09-22 03:43 UTC |
6d 15h |
| #3511 — feat(agent): add sandboxed Hermes |
Maintainers |
reviewer not assigned |
2026-09-21 15:33 UTC |
2026-09-22 15:33 UTC |
6d 3h |
| #3585 — feat(kidbench): add KIDBench child-facing safety benchmark |
Maintainers |
reviewer not assigned |
2026-09-21 23:28 UTC |
2026-09-22 23:28 UTC |
5d 19h |
| #3586 — fix(cli): exclude exact matches from did-you-mean suggestions |
Maintainers |
reviewer not assigned |
2026-09-22 00:25 UTC |
2026-09-23 00:25 UTC |
5d 19h |
| #3587 — feat(injecagent): add InjecAgent indirect prompt injection benchmark |
Maintainers |
reviewer not assigned |
2026-09-22 00:44 UTC |
2026-09-23 00:44 UTC |
5d 18h |
| #3588 — feat(facts-parametric): add official public benchmark |
Maintainers |
reviewer not assigned |
2026-09-22 00:47 UTC |
2026-09-23 00:47 UTC |
5d 18h |
| #3567 — feat(asb): add Agent Security Bench environment |
Maintainers |
reviewer not assigned |
2026-09-22 02:06 UTC |
2026-09-23 02:06 UTC |
5d 17h |
| #3590 — feat(indic_gsm8k): add Indic GSM8K benchmark |
Maintainers |
reviewer not assigned |
2026-09-22 03:50 UTC |
2026-09-23 03:50 UTC |
5d 15h |
| #3591 — fix(core): isolate runtime, sandbox and model transport contracts |
Maintainers |
reviewer not assigned |
2026-09-22 04:08 UTC |
2026-09-23 04:08 UTC |
5d 15h |
| #3595 — feat(toolalignbench): add reproducible benchmark environment |
Maintainers |
reviewer not assigned |
2026-09-22 07:13 UTC |
2026-09-23 07:13 UTC |
5d 12h |
| #3598 — feat(indic_arc_c): add Indic ARC-Challenge benchmark |
Maintainers |
reviewer not assigned |
2026-09-22 08:16 UTC |
2026-09-23 08:16 UTC |
5d 11h |
| #3600 — feat(launcher): add Slurm benchmark launcher and model profiles |
Maintainers |
reviewer not assigned |
2026-09-22 08:55 UTC |
2026-09-23 08:55 UTC |
5d 10h |
| #3602 — feat(launcher): add fast benchmark profiles and named resumable runs |
Maintainers |
reviewer not assigned |
2026-09-22 09:08 UTC |
2026-09-23 09:08 UTC |
5d 10h |
| #3603 — feat(indic gpqa_diamond): add Indic GPQA Diamond benchmark |
Maintainers |
reviewer not assigned |
2026-09-22 09:09 UTC |
2026-09-23 09:09 UTC |
5d 10h |
| #3608 — feat(in22-indic): add IN22 translation evaluation |
Maintainers |
reviewer not assigned |
2026-09-22 10:59 UTC |
2026-09-23 10:59 UTC |
5d 8h |
| #3594 — feat(even-handedness): add Anthropic Political Even-handedness benchmark |
Maintainers |
reviewer not assigned |
2026-09-22 14:52 UTC |
2026-09-23 14:52 UTC |
5d 4h |
| #3617 — feat(safe_child_llm): add Safe-Child-LLM developmental safety benchmark |
Maintainers |
reviewer not assigned |
2026-09-22 15:06 UTC |
2026-09-23 15:06 UTC |
5d 4h |
| #3620 — feat(vera_mh): add VERA-MH clinical-rubric safety benchmark |
Maintainers |
reviewer not assigned |
2026-09-22 16:09 UTC |
2026-09-23 16:09 UTC |
5d 3h |
| #3628 — feat(facts-multimodal): add image factuality benchmark |
Maintainers |
reviewer not assigned |
2026-09-22 20:16 UTC |
2026-09-23 20:16 UTC |
4d 23h |
| #3626 — feat(agent): run Pi inside native task sandboxes |
Maintainers |
reviewer not assigned |
2026-09-22 20:54 UTC |
2026-09-23 20:54 UTC |
4d 22h |
| #3634 — feat(checkpoint): resume proof refinement agent turns |
Maintainers |
reviewer not assigned |
2026-09-22 22:19 UTC |
2026-09-23 22:19 UTC |
4d 21h |
| #3642 — [ApexAgents] Relay policy traffic over a unix socket for network-isolated sandboxes |
Maintainers |
reviewer not assigned |
2026-09-23 00:53 UTC |
2026-09-24 00:53 UTC |
4d 18h |
| #3644 — feat(indic mbpp): add Indic MBPP |
Maintainers |
reviewer not assigned |
2026-09-23 01:11 UTC |
2026-09-24 01:11 UTC |
4d 18h |
| #3645 — feat(benchmarks): add Indic MATH-500 |
Maintainers |
reviewer not assigned |
2026-09-23 01:56 UTC |
2026-09-24 01:56 UTC |
4d 17h |
| #3650 — feat(exploitbench): add ExploitBench V8 capability-ladder benchmark |
Maintainers |
reviewer not assigned |
2026-09-23 06:34 UTC |
2026-09-24 06:34 UTC |
4d 12h |
| #3651 — feat(agentdyn): add AgentDyn indirect prompt-injection benchmark |
Maintainers |
reviewer not assigned |
2026-09-23 06:36 UTC |
2026-09-24 06:36 UTC |
4d 12h |
| #3655 — fix(orchestration): write job.sh owner-only |
Maintainers |
reviewer not assigned |
2026-09-23 09:39 UTC |
2026-09-24 09:39 UTC |
4d 9h |
| #3657 — feat(orchestration): express prefill/decode disaggregation |
Maintainers |
reviewer not assigned |
2026-09-23 09:40 UTC |
2026-09-24 09:40 UTC |
4d 9h |
| #3659 — [GDPVal] Coerce a rendered paths list nested inside the finish argument |
Maintainers |
reviewer not assigned |
2026-09-23 10:24 UTC |
2026-09-24 10:24 UTC |
4d 9h |
| #3661 — feat(facts-grounding-v2): add public benchmark adapter |
Maintainers |
reviewer not assigned |
2026-09-23 14:58 UTC |
2026-09-24 14:58 UTC |
4d 4h |
| #3679 — fix: stop leaking credentials in resolved configs and debug request logs |
Maintainers |
reviewer not assigned |
2026-09-24 14:07 UTC |
2026-09-25 14:07 UTC |
3d 5h |
| #3695 — feat(indicifeval): add IndicIFEval Trans benchmark |
Maintainers |
reviewer not assigned |
2026-09-25 14:25 UTC |
2026-09-28 14:25 UTC |
2d 5h |
| #3697 — fix(vllm): AUT-2414 constrain SQLAlchemy below 2.1 |
Maintainers |
reviewer not assigned |
2026-09-25 18:17 UTC |
2026-09-28 18:17 UTC |
2d 1h |
| #3704 — feat(mimo-music): add ABC composition resources server |
Maintainers |
reviewer not assigned |
2026-09-28 04:29 UTC |
2026-09-29 04:29 UTC |
1d 14h |
| #3708 — feat(orchestration): let the driver install a benchmark's own package |
Maintainers |
reviewer not assigned |
2026-09-28 10:56 UTC |
2026-09-29 10:56 UTC |
1d 8h |
Pull request handoff SLA tracker
Review during the daily stand-up alongside the architecture/RFC decision tracker.
The SLA is one business day per handoff; weekends are excluded and public holidays are not modeled.
Drafts, bots, and PRs labeled
stale,paused, ordo-not-trackare excluded.Last refreshed: 2026-09-30T19:26+00:00
Newly breached (17)
Waiting on reviewers (6)
gym eval submitWaiting on authors (2)
Needs review assignment (9)
Older breach backlog (162)
Waiting on reviewers (59)
Waiting on authors (3)
Needs review assignment (100)
gym eval intaketo submit a rundir to a benchmark-run data layer