From the arXiv
Monday, 21 September 2026 · 20 papers
Accelerating Dense LLMs via L0-regularized Mixture-of-Experts
This paper introduces L0-MoE, a method to accelerate dense LLMs by using L0-regularization to create a lightweight Mixture-of-Experts (MoE) architecture. The core contribution is achieving significant speedups (up to 2.5x) with minimal performance degradation, outperforming existing acceleration techniques. This is ena…
CodeMidas: Scaling Agentic Coding RL Environments from Code Itself
CodeMidas is a novel agentic pipeline that automatically generates reinforcement learning (RL) coding environments directly from existing source code. It leverages agents to explore functionality, create tests based on code execution, and validate tasks, enabling scalable creation of diverse coding challenges. This met…
Do Personality-Tuned LLMs Make Better Social Agents?
This paper investigates if fine-tuning LLMs with personality data improves their social agent capabilities. By training on personality-labeled social media and dialogues, they created personality-based dialogue engines and evaluated their consistency and controllability against instruction prompting. The contribution l…
ExpBoN: Exponential-Noise Best-of-$n$ for Efficient Test-Time LLM Alignment
ExpBoN is a novel method for efficient Large Language Model (LLM) alignment at inference time. It uses an exponential-noise mechanism to achieve smoother control over the reward-distribution shift trade-off, leading to exponentially faster convergence in key metrics. This approach, integrated into guided speculative in…
An Interpretable Memory Decision Controller for LLM Agents Based on Three-Signal Complementarity: Decoupling Confidence and Consistency
This paper introduces the Memory Decision Layer (MDL), a novel, parameter-free module for LLM agents. The MDL addresses the issue of conflicting memories in retrieval-augmented generation by employing a three-signal complementarity approach (relevance, reliability, task risk) to intelligently decide which memories to t…
CASCADE Against Jailbreaks: Combination Across Stages with Controlled Attack-Defense Evaluation
This paper introduces CASCADE, a framework for systematically evaluating and combining defenses against LLM jailbreaks across different pipeline stages. It standardizes attack success rate measurement and query budgets to provide principled comparisons, revealing that layered defense combinations are more effective tha…
A Lie Detector Test for Language Models: Reading Knowledge a Model Won't Reveal
This paper introduces Probe of Internal Recognition (PIR), a method inspired by forensic concealed information tests to detect knowledge hidden within language models. PIR works by presenting a model with a question and multiple candidate answers, then analyzing its internal states to identify which answer it internall…
AutoViewMem: Self-Configuring Orthogonal Views for Conversational Long-Term Memory
AutoViewMem addresses the challenge of organizing diverse information in LLM long-term memory by creating self-configuring, low-overlap semantic views. This data-driven framework disentangles information at write-time, improving the quality and relevance of memories for retrieval. Its core contribution is a representat…
Bayesian Belief Layer for Controllable Opinion Dynamics in LLM Agents
This paper introduces Bayesian Chronicle Agents (BCA) to address the implicit opinion dynamics in LLM agents. BCAs separate an agent's beliefs from its speech, representing stances as probabilities updated via Bayesian inference. A single parameter, $κ$, controls stubbornness, enabling the generation of distinct opinio…
CIPL: A Channel-Aware Framework for Recoverable Privacy Leakage in LLM Agents
This paper introduces CIPL, a novel framework for evaluating privacy leakage in LLM agents by considering the entire information flow from sensitive data to an attacker's observable output. Unlike previous methods, CIPL models the process through distinct stages and assesses recoverability under a shared protocol. Its …
Detecting Pretraining Data in Large Language Models from a Free-Energy Perspective
This paper proposes Energy Transfer Detection (ETD) to distinguish pretraining data in large language models. ETD improves upon existing methods by using an inclined boundary that considers both prediction loss and predictive entropy, effectively correcting for generalization. This approach offers a free-energy interpr…
DiaVLo: Diagnosing Behaviours of Vision-Language Models
DiaVLo is a diagnostic framework that uses human input and VLMs' own generation abilities to create specifications of desired and observed model behaviors, highlighting any misalignments. It also provides causal estimates to pinpoint the key concepts driving these behaviors. This approach helps identify both aligned an…
EnterpriseVal: Quantifying the Efficacy, Reliability and Value of Generative AI in the Enterprise
This paper introduces EnterpriseVal, a system designed to bridge the gap between generative AI's capabilities and its actual business impact. It addresses the problem that current evaluations focus on model potential rather than enterprise-specific deployment readiness. EnterpriseVal provides a use-case-level evaluatio…
NemotronLabs VoiceChat: An Open Full-duplex Speech-to-Speech Model with Tool Calling Capabilities
NemotronLabs VoiceChat is an open, full-duplex speech-to-speech model that integrates tool-calling capabilities. Its core method involves a unified streaming architecture combining speech encoding, a decoder-only language model with parallel text and tool output streams, and a streaming TTS decoder. This design enables…
Value-Sensitive Delegation in Everyday AI Agent Use: Evidence from OpenClaw
This paper introduces "value-sensitive delegation," a new framework for understanding user interactions with AI agents. By analyzing Reddit posts about OpenClaw, the researchers found that users prioritize values related to the agent's operating conditions, not just its outputs. The study contributes by highlighting th…
From Pretraining to Proficiency: Real-World Subtask RL for Long-Horizon Manipulation with Minimal Human Intervention
This paper introduces PARTS, a method to improve pretrained robot policies for complex, long-horizon manipulation tasks. It focuses RL training on specific failure-prone subtasks by using the pretrained policy for most actions and introducing learned selectors and verifiers to provide local rewards for successful subta…
GraphSkillEvo: Evolutionary Optimization of Graph-Structured Agent Skills
This paper introduces GraphSkillEvo, a novel method for optimizing LLM agent skills by representing them as structured graphs. This graph representation, with nodes for execution steps and edges for transitions, provides clearer workflow guidance and reduces redundancy compared to unstructured natural language skills. …
SpecQuant: Speculative Decoding with Multi-Parent Quantization for Adaptive LLM Inference
SpecQuant is a training-free framework that accelerates LLM inference on consumer hardware by combining speculative decoding with multi-parent quantization. It dynamically uses different quantized model variants (e.g., INT4, FP8, FP16) based on task complexity, reserving full-precision for demanding tasks. This adaptiv…
Watermarkable Multi-Draft Speculative Sampling via Poisson Processes
This paper introduces a novel multi-draft speculative sampling algorithm for LLMs, leveraging Poisson processes. Its core contribution is enabling both efficient inference and robust output watermarking simultaneously, overcoming a previously identified fundamental trade-off. The algorithm achieves this through an exac…
Designer-RSI: Evolving Procedural Memory from User Traffic for Agentic Graphic Design
This paper introduces Designer-RSI, a continual adaptation framework for agentic graphic design. Its core method involves a frozen frontier model interacting with design software and an external procedural memory that learns and refines reusable design skills from user traffic. The key contribution is the ability to ev…