Flight Log / repository

marin-community/marin — agent flight log

27 pull requests opened by AI coding agents in marin-community/marin (★ 4,080, Python): OpenAI Codex (OpenAI) 27 (88% merged).

Repository on GitHub ↗

Flights by agent

AgentFlightsMerge rateMedian to mergeBurn rate
OpenAI Codex (OpenAI)2788%18 min6%

By month

MonthFlightsMergedBurns
2025-06621
2025-0721130

Where agents burned here

  • OpenAI Codex (OpenAI) burn #1400I don't fully understand what this is fixing? We use `max_doc_tokens` in the config and everywhere in the code, wondering where it picked it up

Last 27 flights

PRTitleAgentOpenedLines
#1489Fix inference OOM by streaming datasetOpenAI CodexStaleJul 27, 2025+155 −15
#1487Handle empty glob in classification inferenceOpenAI CodexMerged cleanJul 26, 2025+28 −0
#1480Add Institutional Books datasetOpenAI CodexStaleJul 24, 2025+243 −130
#1476Fix vulnerable form-data versionOpenAI CodexMerged cleanJul 23, 2025+96 −47
#1474Split unit tests for GCPOpenAI CodexStaleJul 23, 2025+244 −119
#1473Fix broken doc linkOpenAI CodexMerged cleanJul 22, 2025+1 −1
#1464Expose trainer profiler optionsOpenAI CodexStaleJul 21, 2025+11 −0
#1463Add PlacementGroupCleanupActorOpenAI CodexStaleJul 20, 2025+43 −0
#1458Disable uv-runOpenAI CodexMerged cleanJul 17, 2025+4 −4
#1455Update infra docs location and contentOpenAI CodexStaleJul 14, 2025+9,129 −380
#1454Fix unrendered code blockOpenAI CodexMerged cleanJul 14, 2025+9,125 −359
#1449Add multi-host FP32 tensor-parallel sliding logitsOpenAI CodexMerged cleanJul 14, 2025+122 −0
#1447Add entrypoint resource options and TPU flag to ray_runOpenAI CodexMerged cleanJul 13, 2025+96 −4
#1444Fix train_lm CPU configurationOpenAI CodexMerged cleanJul 8, 2025+5 −0
#1443Add reconnect_reserved_tpus scriptOpenAI CodexStaleJul 8, 2025+29,886 −858
#1440Add SimpleEnv base classOpenAI CodexMerged cleanJul 7, 2025+2,857 −279
#1437Switch docs and tooling to uvOpenAI CodexStaleJul 7, 2025+128 −53
#1433Switch sliding_logits to npz outputOpenAI CodexMerged cleanJul 4, 2025+135 −49
#1432Switch sliding_logits to npz outputOpenAI CodexMerged cleanJul 4, 2025+25 −46
#1431Speed up P(z) and fix Parquet writeOpenAI CodexMerged cleanJul 4, 2025+65 −55
#1429Optimize logits worker I/OOpenAI CodexMerged cleanJul 4, 2025+29 −25
#1413Add TPU-compatible logits inferenceOpenAI CodexMerged cleanJun 30, 2025+113 −0
#1412Fix snapshot tests by pinning markdownifyOpenAI CodexAbandonedJun 30, 2025+2 −3
#1409Refactor Dolma overlap scriptsOpenAI CodexMerged cleanJun 30, 2025+352 −113
#1406Refactor executor loop into helperOpenAI CodexStaleJun 27, 2025+28,572 −646
#1400Fix inference config field nameOpenAI CodexBurnJun 22, 2025+27,962 −784
#1392[codex] TPUMonitor as an actorOpenAI CodexStaleJun 15, 2025+31,578 −868

No rider has claimed a flight in this repository yet. Flew one of these? Open your flight page and claim it.

Data from public GitHub activity; this page is not affiliated with GitHub or the repository. Repository owner? Request exclusion — handled within 48 hours.