MENTIS: What Belief Changes Under Alignment? Measuring Multi-Scale Latent Torsion in Language Models

Preference alignment has substantially improved the observable behavior of large language models, yet it remains unclear what alignment changes internally. Aligned systems still fail under jailbreaks, prompt injection, and retrieval-time corruption, suggesting behavior-level evaluation alone is incomplete. Post-training should leave measurable traces in internal computation. We ask: when an instruction-tuned (IT) model becomes a preference-aligned (PA) model, what geometric structure changes, where do those changes concentrate, and how selectively do they vary across concepts, prompts, and model families? We introduce MENTIS, a geometry-first framework for measuring alignment-induced internal reorganization in paired checkpoints. MENTIS compares IT and PA models using a primary layerwise covariance-based torsion norm (T1), a secondary spectral torsion diagnostic (T2), and an Energy-Radiance-Activation measure (ERA) for depth localization. Across four 7-8B model pairs on LITMUS, our study reveals that alignment-induced change is selective rather than uniform: normative concepts exhibit larger torsion shifts than factual concepts on average; torsion is negatively correlated with contextual entropy; and peak effects localize to architecture-specific mid-to-late layers. The same pattern appears across word-level, prompt-level, and model-level analyses. These results suggest preference alignment leaves structured, depth-localized geometric signatures in internal computation beyond what behavior-level evaluation alone can reveal.

Safe Equilibrium Policy Optimization for Strategic Agent Policies

Language models fine-tuned with reinforcement learning typically optimize for task reward, ignoring multi-agent strategic structure. Because these agents condition on natural language game-state descriptions and emit actions through free-form generation, strategic failure modes -- exploiting weaker opponents, coordinating on harmful equilibria, and externalizing costs are inseparable from the language interface itself. We propose Safe Equilibrium Policy Optimization (\sepo{}), a training objective that augments expected payoff with explicit penalties for exploitability, collusion risk, and externality cost. We implement \sepo{} as a reward signal for Group Relative Policy Optimization (GRPO), applied to Gemma~4 E4B-it and Qwen~3.5-4B after supervised fine-tuning (SFT). Evaluated across five strategic domains: Iterated Prisoner's Dilemma, repeated auctions, two negotiation variants, and Kuhn Poker. \sepo{} achieves zero exploit-pool advantage in Kuhn Poker for both models, outperforms the base model on safety in four domains, and corrects the over-cooperative behavior introduced by SFT. In negotiation, \sepo{} achieves a positive-safety outcome and only the positive normalized relative advantage of any negotiation configuration. Ablation experiments confirm that per-rollout exploit computation is necessary: a shared constant penalty cancels in GRPO advantage normalization (constant control-variate property), producing zero gradient. To support further research in strategic safety for agents, we release our \href{https://anonymous.4open.science/r/sepo-2668/README.md}{code} and SFT datasets.

Chart-RL: Policy Optimization Reinforcement Learning for Enhanced Visual Reasoning in Chart Question Answering with Vision Language Models

The recent advancements in Vision Language Models (VLMs) have demonstrated progress toward true intelligence requiring robust reasoning capabilities. Beyond pattern recognition, linguistic reasoning must integrate with visual comprehension, particularly for Chart Question Answering (CQA) tasks involving complex data visualizations. Current VLMs face significant limitations in CQA, including imprecise numerical extraction, difficulty interpreting implicit visual relationships, and inadequate attention mechanisms for capturing spatial relationships in charts. In this work, we address these challenges by presenting Chart-RL, a novel reinforcement learning framework that enhances VLMs chart understanding through feedback-driven policy optimization of visual perception and logical inference. Our key innovation includes a comprehensive framework integrating Reinforcement Learning (RL) from Policy Optimization techniques along with adaptive reward functions, that demonstrates superior performance compared to baseline foundation models and competitive results against larger state-of-the-art architectures. We also integrated Parameter-Efficient Fine-Tuning through Low-Rank Adaptation (LoRA) in the RL framework that only requires single GPU configurations while preserving performance integrity. We conducted extensive benchmarking across open-source, proprietary, and state-of-the-art closed-source models utilizing the ChartQAPro dataset. The RL fine-tuned Qwen3-VL-4B-Instruct model achieved an answer accuracy of 0.634, surpassing the 0.580 accuracy of the Qwen3-VL-8B-Instruct foundation model despite utilizing half the parameter count, while simultaneously reducing inference latency from 31 seconds to 9 seconds.

DeltaLogic: Minimal Premise Edits Reveal Belief-Revision Failures in Logical Reasoning Models

Reasoning benchmarks typically evaluate whether a model derives the correct answer from a fixed premise set, but they under-measure a closely related capability that matters in dynamic environments: belief revision under minimal evidence change. We introduce DeltaLogic, a benchmark transformation protocol that converts natural-language reasoning examples into short revision episodes. Each episode first asks for an initial conclusion under premises P, then applies a minimal edit δ(P), and finally asks whether the previous conclusion should remain stable or be revised. We instantiate DeltaLogic from FOLIO and ProofWriter and evaluate small causal language models with constrained label scoring. On a completed 30-episode Qwen evaluation subset, stronger initial reasoning still does not imply stronger revision behavior: Qwen3-1.7B reaches 0.667 initial accuracy but only 0.467 revision accuracy, with inertia rising to 0.600 on episodes where the gold label should change, while Qwen3-0.6B collapses into near universal abstention. There, Qwen3-4B preserves the same inertial failure pattern (0.650 initial, 0.450 revised, 0.600 inertia), whereas Phi-4-mini-instruct is substantially stronger (0.950 initial, 0.850 revised) but still exhibits non-trivial abstention and control instability. These results suggest that logical competence under fixed premises does not imply disciplined belief revision after local evidence edits. DeltaLogic therefore targets a distinct and practically important reasoning capability that complements existing logical inference and belief-updating benchmarks.

Multi-Dimensional Summarization Agents with Context-Aware Reasoning over Enterprise Tables

We propose a novel framework for summarizing structured enterprise data across multiple dimensions using large language model (LLM)-based agents. Traditional table-to-text models often lack the capacity to reason across hierarchical structures and context-aware deltas, which are essential in business reporting tasks. Our method introduces a multi-agent pipeline that extracts, analyzes, and summarizes multi-dimensional data using agents for slicing, variance detection, context construction, and LLM-based generation. Our results show that the proposed framework outperforms traditional approaches, achieving 83\% faithfulness to underlying data, superior coverage of significant changes, and high relevance scores (4.4/5) for decision-critical insights. The improvements are especially pronounced in categories involving subtle trade-offs, such as increased revenue due to price changes amid declining unit volumes, which competing methods either overlook or address with limited specificity. We evaluate the framework on Kaggle datasets and demonstrate significant improvements in faithfulness, relevance, and insight quality over baseline table summarization approaches.