The paper introduces a retrieval‑augmented language model that resolves disagreements between retrieved context and the model’s own knowledge by incorporating a prior over the reliability of the context rather than a fixed rule. Experiments on the GRACE benchmark show that this prior‑based approach outperforms existing baselines by up to 26 points, achieving higher overall scores on both Qwen3.5‑35B‑A3B and Llama‑3.1‑8B. The method also demonstrates superior performance on the ClashEval benchmark, confirming the theoretical advantage of evidence‑based conflict resolution.
arXiv:2606. 10298v1 Announce Type: new Abstract: When large language models generate from retrieved or augmented contexts, conflicts between external context and parametric priors remain a central reliability bottleneck.
The paper introduces F-ICL, a benchmark that measures in‑context algorithmic reasoning in language models by exhaustively enumerating 86 million valid programs of length ≤13 on a Turing‑complete machine and computing the exact posterior under a bounded Levin–Solomonoff prior. Unlike typical benchmarks, F‑ICL provides a distributional reference rather than just answers, allowing the evaluation of models’ inductive priors. Across 105 configurations of models ranging from 0.8 B to 675 B parameters, models achieve up to 92 % accuracy, yet many still deviate from the Bayes‑optimal reference, and the study derives theoretical bounds on cumulative loss for predictors with positive prior weight on the reference.
arXiv:2605. 14473v4 Announce Type: replace-cross Abstract: Retrieval-Augmented Generation (RAG) is usually evaluated by whether the final answer is correct.
arXiv:2608. 11922v1 Announce Type: cross Abstract: Predictive-distribution entropy makes a strong selection rule in retrieval-augmented question answering: across five QA benchmarks, keeping the candidate answer that a frozen respondent LLM produces with the lowest answer-token entropy lifts mean answer $F_1$ from 0.
Whether large language models perform genuine algorithmic reasoning or mere pattern completion is hard to test, because most benchmarks lack a ground truth for correct inductive inference. We introduce F-ICL, an in-context-learning benchmark that supplies one exactly.
The paper evaluates how three large mixture‑of‑experts models (Alibaba, OpenAI, NVIDIA) can be fine‑tuned to reason in a low‑resource language, specifically Greek. Accuracy metrics show little change, but the authors uncover significant qualitative improvements: after supervised fine‑tuning, models reason in Greek on ~98% of items, with better grammaticality and retained general ability. Reinforcement learning with pre‑registered rewards further eliminates reasoning‑channel leaks and format skips, while the Greek‑reasoning habit remains robust to an accuracy‑only gradient.
arXiv:2606. 25449v1 Announce Type: cross Abstract: A language model's memory can be worse than having no memory at all.
arXiv:2608. 01575v1 Announce Type: new Abstract: Whether large language models perform genuine algorithmic reasoning or mere pattern completion is hard to test, because most benchmarks lack a ground truth for correct inductive inference.
The paper investigates why differentiable causal discovery methods that encode expert priors as forbidden-edge constraints via an Augmented Lagrangian (ALM) penalty—termed the "guide, not bind" approach—often fail. It identifies two key failures: (1) the sequential penalty‑ramping ALM suppresses a true edge before counterfactual checks can detect it, and the proposed adaptive relaxation rule DADU violates necessary conditions for safe relaxation, leading to a high failure rate across thousands of training runs; (2) the standard correlation‑matching objective inherently ties a true edge and its reverse to the same cost, whereas covariance matching can separate them by a provable margin. The authors provide theoretical propositions, corollaries, and empirical evidence to support these claims.
arXiv:2512. 11614v3 Announce Type: replace-cross Abstract: Retrieval-augmented generation (RAG) relies on retrieved context to guide large language models (LLM), yet treats the retrieval as a heuristic rather than verifiable evidence -- leading to unsupported answers, hallucinations, and reliance on spurious context.
Given a state and a set of options, such as intents, tools or interface elements, System One decision models return a distribution over the options in one forward pass, without generated reasoning. We...
The paper proposes a training‑free detector that uses sentence‑level context sensitivity to identify unsupported content in retrieval‑augmented generation (RAG) answers. By re‑scoring each sentence with full context, no context, and each chunk removed, the method flags the chunk whose removal most reduces a sentence’s likelihood as the likely source. Evaluated on RAGTruth, TofuEval, and RAGBench, the detector outperforms answer‑level faithfulness scores, achieving AUCs up to 0.745 and matching per‑chunk fact‑checkers while using only a fraction of the compute required by large‑language‑model judges.