Issues

TASK 2 — global pgm HydraNet calibration point predictions for the research team: specification and findings

#505

· Polichinel opened

on Sep 27, 2026

RUNBOOK: rusty_bucket on the 8 HydraNets at global land — calibration predictions down, then true forecasts to the FAO

#499

· Polichinel opened

on Sep 19, 2026

Render datafactory models' feature descriptions in the catalogs job — install views-datafactory there (deferred from #483 → #474 → here)

#498

· Polichinel opened

on Sep 19, 2026

Fresh installs pull a torch whose CUDA build the machine's driver cannot run — the other half of views-hydranet#377

#494

· Polichinel opened

on Sep 19, 2026

Migrate the r2darts2 family to prediction_format: prediction_frame — deferred from #489

#492

· Polichinel opened

on Sep 19, 2026

EPIC: the 19 target models — 11 pgm datafactory r2darts2 + 8 HydraNets — run on the latest platform

#488

· Polichinel opened

on Sep 19, 2026

ADR-017 Phase 4: re-home the ensemble member-maturity guard — move its function, do not delete it

#481

· Polichinel opened

on Sep 17, 2026

ADR-017 Phase 3: the shelf write-gate — only a graduate source may publish to the prediction store

#480

· Polichinel opened

on Sep 17, 2026

staging_202608 (#402): rude_boy/first_love mix stepshifter with r2darts2 >=0.2.0 and cannot aggregate; the catalogs job must expect pipeline-core 3.3.0's loud queryset loader

#478

· Polichinel opened

on Sep 17, 2026

Name the datafactory gap: which viewser features do the 77 legacy models need that views-datafactory does not provide?

#475

· Polichinel opened

on Sep 17, 2026

REFERENCE: the platform runs on pandas 1 / darts 0.40 / Python 3.11 — 77 models can't leave it, why, and the 2027-10-31 clock

#473

· Polichinel opened

on Sep 17, 2026

README and .env template still document --update_viewser, which pipeline-core has retired

#472

· Polichinel opened

on Sep 16, 2026

Navigation Menu

All issues

TASK 2 — global pgm HydraNet calibration point predictions for the research team: specification and findings

RUNBOOK: rusty_bucket on the 8 HydraNets at global land — calibration predictions down, then true forecasts to the FAO

Render datafactory models' feature descriptions in the catalogs job — install views-datafactory there (deferred from #483 → #474 → here)

Fresh installs pull a torch whose CUDA build the machine's driver cannot run — the other half of views-hydranet#377

Migrate the r2darts2 family to prediction_format: prediction_frame — deferred from #489

EPIC: the 19 target models — 11 pgm datafactory r2darts2 + 8 HydraNets — run on the latest platform

ADR-017 Phase 4: re-home the ensemble member-maturity guard — move its function, do not delete it

ADR-017 Phase 3: the shelf write-gate — only a graduate source may publish to the prediction store

staging_202608 (#402): rude_boy/first_love mix stepshifter with r2darts2 >=0.2.0 and cannot aggregate; the catalogs job must expect pipeline-core 3.3.0's loud queryset loader

Name the datafactory gap: which viewser features do the 77 legacy models need that views-datafactory does not provide?

REFERENCE: the platform runs on pandas 1 / darts 0.40 / Python 3.11 — 77 models can't leave it, why, and the 2027-10-31 clock

README and .env template still document --update_viewser, which pipeline-core has retired

Search results