Tracked topic
Large language models are machine learning models trained to predict and generate text and other language-based outputs.
…Across 12 frontier LLMs, we find a pervasive optimism bias: under standard prompting, models frequently rate low-soundness proposals as sound, while aggressive prompting largely shifts errors from false positives to false…
Papers arxiv:2606.07547 Liberating LLM Capabilities in Full-Duplex Speech Models Published on May 4 Submitted by zly on Jun 9 Authors: Luoyuan Zhang , , , , , , Abstract A text-first tri-channel speech…
…Xiang Liu , , , , , , , Abstract LLM inference should be evaluated as energy-to-token production under constraints of compute, power, cooling, and operational efficiency, requiring new metrics beyond traditional accuracy and latency measures. AI…
…Efficient Generative LLM Inference with Asynchronous Two-level Sparse Attention (2026) Flux Attention: Context-Aware Hybrid Attention for Efficient LLMs Inference (2026) StructKV: Preserving the Structural Skeleton for Scalable Long-Context Inference…
…Agreement-Gated Stress Testing of LLM-Judged Investment Rationales Before Returns Are Observable (2026) JudgeSense: A Benchmark for Prompt Sensitivity in LLM-as-a-Judge Systems (2026) Rethinking Atomic Decomposition for LLM…
…Surveying Collaboration, Failure Attribution, and Self-Evolution in LLM-based Multi-Agent Systems Published on May 14 Submitted by JieMa on May 15 Xi'an Jiaotong University Authors: , , , , , , , , , , , , , , , , , Abstract Multi-agent systems…
…Measuring and Enforcing Quantitative Goal Persistence in Long-Horizon LLM Agents (2026) CausalFlow: Causal Attribution and Counterfactual Repair for LLM Agent Failures (2026) Why LLMs Fail at Causal Discovery and How Interventional…
…Accelerating Agent Self-Evolution with Asynchronous Stage Orchestration Published on May 8 Submitted by Zhen Wang on May 12 University of California at San Diego Authors: , , Zhen Wang , , , , , , , , Abstract FlashEvolve enhances LLM…
…Building on this property, we train a lightweight Distiller at test time to predict deep-layer hidden representations of the LLM from its shallow-layer representations to model the LLM's depth…
…Generated by Qwen/Qwen2.5-Coder-32B-Instruct Video large language models (Video-LLMs) have demonstrated strong capabilities in video understanding tasks. However, their practical deployment is still hindered by the inefficiency…