Skip to content
Slow Variables

Ask the data

Answers come from the same store as the site; every number is checked against the record it cites.

Indicators / series

cl_ladder.harvey_tenet.rung_research.pt

Source: Harvey research and insights posts (Harvey) · Harvey; short quotation · CSV

Harvey

idvaluegradeas of datepublished datetieraudited vs reportedextraction methodreview statusretrieved athttp statuscontent hashsupersedes idurlflags
c16acbc4fb932e984.00 runggrade D, evidence tier 72026-08-202026-08-207company_statedmanualapproved2026-09-10T06:54:22.403929Z200aa4deb5a75a8source page⚑ disputed: Placed at L4 because Appendix E lists Tenet there, but the post describes one async-RL post-training run on domain (lega
Raw snippets and dispute text
  • c16acbc4fb932e98 Starting from a Kimi K3 base, we worked with Fireworks to post-train our model via asynchronous reinforcement learning (RL) in realistic legal work settings.Placed at L4 because Appendix E lists Tenet there, but the post describes one async-RL post-training run on domain (legal) data for a research preview, not organisation-specific knowledge on a refresh cadence; read literally it is an L0 static snapshot with domain specialisation.