3 papers indexed here
We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.
Not the right person? Other researchers publish under this name.
The case motivates accounting that retains every boundary, preserves unknown outcomes and policy dependencies, checks task premises, and separates bounded completion from success-conditioned resource use.
Context projection replaces older tool observations with compact, addressable excerpts, reducing repeated input while potentially adding evidence-retrieval turns. We study this trade-off in ReVerPi, a Pi extension with archived observations and matched full/projected continuations. In an 86-run source-reading campaign...
A paired-history audit and label-equivariant repair support a scoped evaluation methodology and reproducible failure analysis and general superiority of the repaired algorithm, not general superiority of the repaired algorithm.
We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.
