Commit Graph

  • 9dc870b8d0 Add 3x-context dataset variant (trainset --radius) main Florian Herzog 2026-06-10 16:37:30 +02:00
  • e35d98c2cd Commit training-ready dataset (~6 MB) + DATASET.md usage guide Florian Herzog 2026-06-10 16:22:39 +02:00
  • 50f67bcbe0 trainset: GUARANTEE every target triple's entity name is present in input_text Florian Herzog 2026-06-10 16:17:14 +02:00
  • 216fd9876c trainset: pick evidence excerpt at the actual provider statement; document provider-currency caveat Florian Herzog 2026-06-10 14:19:02 +02:00
  • 991715ab76 Add LLM role-check grounding + labelled training-set pipeline Florian Herzog 2026-06-10 13:52:50 +02:00
  • 09798eb27a Add LLM grounding pipeline: current-source fetch, alias + LLM role-check matching Florian Herzog 2026-06-09 13:45:32 +02:00
  • 00f51859e0 Drop non-extractable custodian relation; add per-triple grounded flag Florian Herzog 2026-06-05 10:34:14 +02:00
  • 63e650fa14 Update dataset description with full 2025Q3 build statistics Florian Herzog 2026-06-03 11:21:23 +02:00
  • 1993658fb2 Add SEC fund prospectus -> RDF triple dataset pipeline Florian Herzog 2026-06-03 10:31:35 +02:00