Issues

• APEX: ingest the paper and model the multi-protocol chain

  ingestiontasksunmet-ask
  PhysicalAIBenchmarks/Text2WetLab#29 · Filter by author Tyronita opened on Oct 4, 2026

• Make the reference scripts runnable

  tasksunmet-ask
  PhysicalAIBenchmarks/Text2WetLab#28 · Filter by author Tyronita opened on Oct 4, 2026

• Expert verification of tasks

  evaluationunmet-ask
  PhysicalAIBenchmarks/Text2WetLab#27 · Filter by author Tyronita opened on Oct 4, 2026

• CI: simulator job, provenance check, licence gate

  infraunmet-ask
  PhysicalAIBenchmarks/Text2WetLab#26 · Filter by author Tyronita opened on Oct 4, 2026

• Pin the Opentrons 7.5.0 / Python 3.10 simulator stack inside the repo

  infraunmet-ask
  PhysicalAIBenchmarks/Text2WetLab#25 · Filter by author Tyronita opened on Oct 4, 2026

• Use Opentrons-native DurationEstimator and deck geometry

  infraunmet-ask
  PhysicalAIBenchmarks/Text2WetLab#24 · Filter by author Tyronita opened on Oct 4, 2026

• Track module state: magnet, temperature, thermocycler, heater-shaker, delay, pause

  criteriaunmet-askviz
  PhysicalAIBenchmarks/Text2WetLab#23 · Filter by author Tyronita opened on Oct 4, 2026

• Real labware geometry in the 3D view

  unmet-askviz
  PhysicalAIBenchmarks/Text2WetLab#22 · Filter by author Tyronita opened on Oct 4, 2026

• Move the MuJoCo viewer into opentrons-mujoco-viz and depend on it

  unmet-askviz
  PhysicalAIBenchmarks/Text2WetLab#21 · Filter by author Tyronita opened on Oct 4, 2026

• Instruction variants (V0/V1/V2) per task and paraphrase robustness

  tasksunmet-ask
  PhysicalAIBenchmarks/Text2WetLab#20 · Filter by author Tyronita opened on Oct 4, 2026

• Generic criteria still missing: AC3, AC6 and AC7

  criteriaunmet-ask
  PhysicalAIBenchmarks/Text2WetLab#19 · Filter by author Tyronita opened on Oct 4, 2026

• End-state check cannot judge multi-plate tasks

  criteriaunmet-ask
  PhysicalAIBenchmarks/Text2WetLab#18 · Filter by author Tyronita opened on Oct 4, 2026

• Reference solutions and hidden tests for the four spec-only tasks

  tasksunmet-ask
  PhysicalAIBenchmarks/Text2WetLab#17 · Filter by author Tyronita opened on Oct 4, 2026

• Re-anchor handwritten tasks to real reference scripts

  tasksunmet-ask
  PhysicalAIBenchmarks/Text2WetLab#16 · Filter by author Tyronita opened on Oct 4, 2026

• Decide the licensing of the Harbor oracle (the HULP script has no licence)

  licensingunmet-ask
  PhysicalAIBenchmarks/Text2WetLab#15 · Filter by author Tyronita opened on Oct 4, 2026

• Redeploy HuggingFace from the new layout

  infraunmet-ask
  PhysicalAIBenchmarks/Text2WetLab#14 · Filter by author Tyronita opened on Oct 4, 2026

• HuggingFace publicly hosts unlicensed third-party scripts

  licensingunmet-askurgent
  PhysicalAIBenchmarks/Text2WetLab#13 · Filter by author Tyronita opened on Oct 4, 2026

• Write the arXiv paper (the survey was never delivered)

  docsunmet-ask
  PhysicalAIBenchmarks/Text2WetLab#12 · Filter by author Tyronita opened on Oct 4, 2026

• No model has ever been evaluated on the benchmark

  evaluationunmet-ask
  PhysicalAIBenchmarks/Text2WetLab#11 · Filter by author Tyronita opened on Oct 4, 2026