arXiv:2606. 28360v1 Announce Type: cross Abstract: University students often struggle to navigate complex academic policies, leading to advising bottlenecks and delayed access to critical information.
arXiv:2607. 14707v1 Announce Type: cross Abstract: Large language models routinely produce fluent answers to single-shot prompts, yet deploying them as reliable components of a domain decision system is substantially harder.
arXiv:2607. 03656v1 Announce Type: cross Abstract: Large Language Models are increasingly used to turn natural-language requirements into code.
arXiv:2607. 28229v1 Announce Type: cross Abstract: The web is increasingly accessed by AI agents rather than humans.
arXiv:2607. 01256v1 Announce Type: cross Abstract: Overwhelmed courts in the United States review millions of default judgments each year.
The paper introduces Tasks over Application Manuals (TAM), a benchmark designed to test long‑horizon procedural reasoning in large language models. TAM uses real‑world tasks from ICD‑10‑CM clinical coding and U.S. federal sentencing, requiring models to follow extensive, rule‑based manuals and perform interdependent steps to produce exact answers. Experiments with GPT‑5 and various prompting strategies show very low exact‑match accuracy—1% for coding and 15.5% for sentencing—highlighting a gap between current benchmarks and the ability to reliably follow complex procedures.
The paper introduces TRACE, a fine‑tuning framework for Retrieval‑Augmented Generation (RAG) that addresses conflicts between retrieved knowledge and a model’s internal knowledge. TRACE uses multi‑agent debate traces to identify correct and incorrect candidates and answer‑shift patterns, providing fine‑grained supervision for reliable knowledge‑source selection. It also incorporates an answer‑completeness regularization mechanism to prevent empty, overly short, or prematurely terminated responses, thereby improving robustness against misleading retrieved content and enhancing answer quality.
Legal Research Bench (LRB) is a new benchmark comprising 413 open-ended U.S. legal research questions, each paired with a gold answer, supporting authorities, and a binary grading rubric. The study evaluates thirteen advanced language‑model agents using web search, case‑law search, page parsing, and retrieval tools, scoring responses only when all required criteria are met and cited authorities verify. Results show that even the best model, Claude Opus 4.8, achieves full correctness on only 42.9% of questions, with performance varying by legal area and task complexity, and no clear link between more tool calls or inference cost and higher accuracy.
arXiv:2509. 00761v4 Announce Type: replace Abstract: Large language models are increasingly deployed for legal question answering, where evaluations typically focus on multiple-choice accuracy.
arXiv:2608. 06057v1 Announce Type: new Abstract: Tool-calling agents infer task state from accumulated dialogue and tool traces.
arXiv:2606. 07544v1 Announce Type: cross Abstract: Middle school is a key window for building core academic skills and the learning routines students carry into later grades, yet many students still fall behind because help is often limited and comes too late, after they have already been stuck for a while.
arXiv:2607. 25600v1 Announce Type: cross Abstract: Retrieval-augmented generation improves knowledge-intensive question answering, but indiscriminate retrieval can introduce irrelevant evidence and unnecessary computation.
arXiv:2606. 22737v2 Announce Type: replace Abstract: Before letting an agent operate over real context, can you prove it used the right evidence?