Neural retrieval models have transformed information retrieval, yet their ability to distinguish factual assertions from hedged speculation remains largely unexamined. We present the first systematic evaluation of hedging sensitivity across eight neural retrieval models spanning two architectural families: four bi-encoder embedding models and four cross-encoder rerankers.
We introduce the Context Decay Benchmark, a reproducible simulation framework for evaluating how agentic harnesses manage information over long conversations. The benchmark plants needle facts—both explicitly marked and implicitly embedded in natural text—into synthetic agent conversations of 50-1000 turns, then measures retrieval accuracy under constrained context budgets (15% of total tokens) across four strategies: Naive Truncation, Sliding Window with Extractive Summary, Structured Memory Banks, and File-Backed Persistent State.
We adapt Karpathy's arxiv-sanity-lite TF-IDF similarity pipeline from academic paper recommendation to production-scale AI tool directory management. Operating on 7,200 AI tools with heterogeneous metadata, our system computes pairwise cosine similarity over bigram TF-IDF vectors to achieve three objectives: duplicate detection (threshold > 0.
We present a reproducible skill for deduplicating large AI tool directories using TF-IDF cosine similarity. Applying the arxiv-sanity-lite pattern to a production dataset of 7,200 tools, we construct a bigram TF-IDF matrix (50K features, sublinear TF scaling), compute pairwise cosine similarity in batches, and extract duplicate pairs (similarity >= 0.