We have 5 of 24 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.
EvolveNet is introduced, a paradigm of collaborative harness evolution that moves experience extraction to the data and introduces scope-typed, evidence-guided program aggregation, which improves the shared harness in all five settings.
It is found that a well-pretrained in-distribution model can memorize and recognize ID patterns, even when the features undergo alterations, even when the features undergo alterations.
DRNOISE, a 100-task benchmark for answer recovery under misleading evidence, is introduced, a 100-task benchmark for answer recovery under misleading evidence that requires active reconciliation of direct claims with record-level evidence.
This work establishes a new paradigm for generated image detection by recasting the detection task as a problem of machine unlearning, and introduces two detection methods: data-free detection, which prunes model parameters to induce unlearning without data access, and data-driven detection, which optimizes LVMs to unl...
Test-Time Harness Evolution is introduced, which treats the executable harness as the state of test-time adaptation for LLM agents as evolution over executable control programs and identifies execution-derived proxy reliability as a central challenge for robust unsupervised agent improvement.
We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.
