Abducer is a typed, replayable research system for constructing, comparing and testing abductive theories over mathematical and scientific problems. It preserves hypotheses, evidence, failed attempts and proof obligations to support research into alternative explanations.
DialFork: a benchmark for procedural hallucination in conversational agents, when an LLM agent stays fluent and factually correct but stops following its intended workflow. Evaluation code, baselines and the paper's methods.
Coming Soon: EyeVLM is a benchmark for evaluating how well vision-language models (VLMs) understand human gaze, covering gaze following (where a person is looking) and social gaze prediction (looking at head, looking at each other, shared attention)
There was an error while loading. Please reload this page.
You are visitor 01402503 — GeezerVison does not store the pages you look up. Best viewed with Netscape Navigator 4.7 or Internet Explorer 6 at 800×600. No cookies required.