Benchmark suite for evaluating AI models on advanced mathematical proof generation and verification tasks. Received 14 upvotes on HuggingFace.
Paper presenting Sequential Coding approach to push limits of model compression using self-generated training data. Addresses model efficiency at scale.
Research using mechanistic interpretability to analyze bias in language models used as judges for evaluation. Provides concrete analysis of evaluation system vulnerabilities.
Research paper examining metacognition capabilities in large language models, including theoretical foundations and empirical progress. Also featured on HuggingFace with 6 upvotes.