We have 3 of 12 papers
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.
It is argued that a long-horizon agent must run continually without forgetting before it can learn continually, and that continual learning for these agents needs a substrate outliving every context and process.
EvoHarnessBench is introduced, a benchmark for evaluating agents under controlled harness evolution across three axes (tools, skills, and agents), and establishes harness evolution as a distinct challenge for building agents that can keep pace with an evolving harness while preserving previously effective behavior.
The results suggest that dense credit assignment through distillation can be effective when its likelihood-based scores are empirically validated as meaningful proxies for outcome-relevant credit, when this alignment does not hold.
We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.
