The Morning
From the arXiv
A Verifier-Guided Explainable Reasoning Framework with Gold-Anchored QLoRA, Task-Aware Mixture-of-Experts, and Group-Relative RLVR
his paper introduces a framework to improve the explainability and verifiability of LLM reasoning in educational question answering. It achieves this by using a gold-anchored QLoRA for model adaptation, a task-aware router to direct problems to specialized symbolic solvers (like Z3), and a reinforcement learning approach that incorporates feedback from these solvers to refine explanations. The core contribution is a more transparent and trustworthy reasoning process for LLMs, particularly in educational contexts.
ACE: Adaptive Calibration-Free Expert Skipping for MoE-based LLMs
ACE introduces a training-free method to reduce redundant computation in MoE LLMs by adaptively skipping experts for each token. Its core contribution lies in the Global Spectral Proxy and Router-Conditioned Refinement components, which estimate expert contrib…
RISE: Recursive Improvement via Self-Extrapolating Policy Distillation
RISE addresses limitations in language model post-training by creating a synthetic teacher from the model's own training trajectory, avoiding external models or privileged conditioning. It achieves this by extrapolating parameter or logit changes to generate d…
Trace2Tower: Transition-Aware EigenTrace Induction of Multi-Level Skills for LLM Agents
Trace2Tower addresses limitations in LLM agent skill learning by introducing a transition-aware framework that distills raw execution traces into a hierarchical skill structure. It achieves this by abstracting interactions into events, constructing a graph bas…
Unifying ICL, SFT, KL-Regularized RL Through a Bayesian Lens
This paper unifies diverse LLM training and inference methods like ICL, SFT, and RL under a single Bayesian framework. The core method involves constructing a generalized Bayes posterior over outputs using a prior model and a utility signal, then approximating…
A Structured Debate-Mixture-of-Agents Framework for Complex Clinical Diagnostic Decision Support
This paper introduces Debate-Mixture-of-Agents (DMoA), a novel multi-agent framework designed to mimic the iterative, role-based reasoning of clinical diagnosis, unlike single-turn…
Does Your Agent's Memory Survive a Model Upgrade? A Controlled Study of Memory Portability
This paper investigates how well an agent's memory survives model upgrades by comparing four memory storage methods: raw history (LC-RAW), retrieved chunks (RAG), natural language …
GUT: Quantifying and Optimizing the Reasoning Uncertainty of LLMs via Graph Complexity
This paper introduces GUT, a method to quantify and reduce Large Language Model (LLM) reasoning uncertainty. GUT models LLM reasoning as a directed acyclic graph, where graph compl…
Measuring AI Accountability Through Argumentation Analysis: Can Model Reasoning Withstand Scrutiny?
This paper proposes a novel method for measuring AI accountability by analyzing the structural quality of a model's defense for its verdicts, rather than relying on ground truth. T…
PRICE: A Systematic Study of LLM Adaptation Choices for Bitcoin Price Forecasting
This paper introduces PRICE, a systematic method for adapting LLMs to Bitcoin price forecasting. PRICE combines efficient fine-tuning (LoRA), recursive inference, integer numerical…
The Town Square
French AI startup Mistral has secured €3 billion in funding to develop sovereign, open-weight AI models, aiming to compete with frontier AI companies.
Workshops
Microsoft's Markitdown is a Python tool that converts various file formats, including Office documents, into Markdown, facilitating content migration and integration.
This repository optimizes AI coding agents' context windows by sandboxing tool output, persisting session memory, and enforcing cross-platform routing through MCP and hooks.
