Indicators / series
cl_ladder.frozen_weights_memory.rung_research.pt
Source: arXiv abstract pages (manual rows) (arXiv) · arXiv; abstract quotation · CSV
arXiv
| id | value | grade | as of date | published date | tier | audited vs reported | extraction method | review status | retrieved at | http status | content hash | supersedes id | url | flags |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| a79c6c8b1963db53 | 1.00 rung | grade B, evidence tier 6 | 2026-07-24 | 2026-07-24 | 6 | reported | manual | approved | 2026-09-10T06:54:23.843970Z | 200 | 49a49dabf691 | — | source page |
Raw snippets and dispute text
- a79c6c8b1963db53 We show that this feedback is a sufficient signal for continual learning when the frozen model is paired with an external memory that distils each episode into retrievable natural-language rules.