The Morning
From the arXiv
Accelerating Masked Diffusion Large Language Models: A Survey of Efficient Inference Techniques
his survey addresses the challenge of achieving practical speedups in masked diffusion large language models (dLLMs), despite their theoretical parallel generation advantage. The core method involves introducing a unified latency decomposition framework to disentangle algorithmic, architectural, and system-level factors influencing inference speed. The main contribution is a structured categorization of acceleration techniques based on this framework, enabling rigorous comparisons and guiding future research towards efficient dLLM deployment.
Evidence-Grounded Verified Agentic Reasoning: A Path Toward Eliminating LLM Hallucination in Empirical Inference via Tool-Attested Kernel Proofs
This paper introduces EG-VAR, a novel architecture that uses the Lean 4 formal verification kernel to ensure Large Language Model (LLM) empirical reasoning is grounded in attested evidence and logically sound. By requiring all verified outputs to trace back to…
Less Experts, Faster Decoding: Cost-Aware Speculative Decoding for Mixture-of-Experts
This paper addresses the inefficiency of speculative decoding in Mixture-of-Experts (MoE) LLMs. It proposes **EcoSpec**, a cost-aware speculative decoding framework that minimizes expert scattering by considering the marginal expert activation cost during draf…
Who Grades the Grader? Co-Evolving Evaluation Metrics and Skills for Self-Improving LLM Agents
This paper addresses the challenge of evaluating LLM agents when no reliable metric exists. Their core method, "Double Ratchet," co-evolves evaluation metrics and agent skills simultaneously. This allows the system to learn and refine both its performance and …
A Learning-Rate-Gated Failure of GRPO in a Small Language and Vision-Language Model Web Agent: A Controlled Null and Its Mechanism
This paper investigates whether Reinforcement Learning from Human Feedback (RLHF), specifically Group Relative Policy Optimization (GRPO), improves the performance of small language and vision-language web agents. The study found that GRPO, even with extensive…
Agentic Service-Oriented Computing: A Manifesto for the Next Frontier of Service-Oriented Computing
This paper proposes Agentic Service-Oriented Computing (ASOC) to address the challenges of integrating LLM-powered agents into complex distributed systems. ASOC advocates for engin…
Can Induced Emotion Bias LLM Behaviors in Sequential Decision Making?
This paper investigates if induced emotions can bias Large Language Model (LLM) behavior in sequential decision-making, using the Iowa Gambling Task. The core method involves an im…
Do AI Agents Know When a Task Is Simple? Toward Complexity-Aware Reasoning and Execution
This paper introduces E3 (Estimate, Execute, Expand), a method for AI agents to assess task complexity and optimize resource usage. E3's core is **task-aware execution-scope estima…
Do We Really Need Multimodal Emotion Language Models Larger Than 1B Parameters?
This paper questions the necessity of massive multimodal models for emotion recognition. It proposes Light-MER, a lightweight framework that uses knowledge distillation to transfer…
Internet of Agentic Things: Networked AI Agents for Closed-Loop IoT Orchestration
This paper introduces the Internet of Agentic Things (IoAT), a framework that unifies AI agents with IoT, cyber-physical systems, and digital twins for closed-loop orchestration. I…
The Town Square
The article questions whether over-reliance on AI for tasks like writing and problem-solving could diminish our own critical thinking and cognitive abilities.
Workshops
This repository provides practical, real-world skills and code snippets for engineers, directly curated from the author's personal development directory.
OpenCut is an open-source video editor that aims to replicate the functionality of CapCut, offering a free and accessible alternative for video creation and editing.
