All Articles
17108 articles total
Omni-Decision: A Progressive Evidence-State Agent System for Omni-Modal QA
Introduces Omni-Decision, a training-free system for omni-modal QA that uses a structured evidence-state to track and close evidence gaps across multiple modalities.
The Ebb and Flow of Multimodal Focus: Scheduling Visual Relay Windows for Grounded VLM Reasoning
Proposes TRACE, a task-adaptive inference-time control framework that manages visual attention focus in VLMs to improve evidence-grounded reasoning.
Enhancing Query Efficiency for d-DNNF Representations Through Preprocessing
Explores preprocessing techniques to improve query efficiency for d-DNNF representations of propositional formulas, specifically for sampling and model enumeration.
Comparative Analysis of GAT and BERT for Human-Like Playtesting
Compares GAT and BERT architectures for modeling human-like playtesting in puzzle games, finding they better capture relational game board structures than CNNs.
Learning Residual Kinematic Corrections for Continuous Neural Decoding via Reinforcement Learning
Presents a two-stage BCI decoding framework using RL to perform residual kinematic corrections on CNN-LSTM outputs for improved 3D motor imagery decoding.
HCRMap: Pressure-Aware Hot-Expert Residency Mapping for 3.5D MoE Chiplet Inference
Introduces HCRMap, a framework for MoE chiplet inference that dynamically manages hot expert replicas to reduce latency and communication bottlenecks.
MAGIC: Transition-Aware Generation of Navigable Multi-Scene Game Worlds with Large Language Models
Presents MAGIC, a prompt-to-project system that uses LLMs to generate navigable, multi-scene game worlds with consistent transitions and connectivity.
Interaction Scaling: Grounding the Third Axis of Test-Time Compute
Argues for 'interaction scaling' as a third axis of test-time compute, where models iteratively revise artifacts based on grounded external feedback.
Auditing the Risk Claims of Distributional Reinforcement Learning
Audits distributional RL agents and finds that their claimed risk trade-offs are often training artifacts rather than reflections of environment stochasticity.
Understanding the Go Runtime: Profiling
A discussion on profiling the Go runtime to understand its internal workings.
European "age verification" "app" forcing everyone to use Android or iOS
Criticism of a European age verification app that restricts users to Android or iOS, raising concerns about platform lock-in and accessibility.
Efficient Test-Time Optimization for Multi-Agent Proof Autoformalization
Introduces ToMap, a multi-agent framework for proof autoformalization that optimizes test-time compute by focusing on the 'Decomposer' agent.
Calibrated e-CUSUM Decoding for Quantized Reasoning Models: Why Token Log-Probability Is the Wrong Observable for Decoding Monitors
Presents a new decoding controller for quantized reasoning models that replaces token log-probability with a calibrated e-CUSUM detector to better identify failing traces.
Verifier-Guided Twelve-Tone Composition: A Generate-Verify-Repair Harness for Symbolic Music Generation
Develops a neuro-symbolic harness using a generate-verify-repair loop to improve the consistency and quality of twelve-tone symbolic music generation with LLMs.
AutoVSR: Automatic Visual-to-Symbolic Reasoning for Symbolic Expression Generation from Circuit Schematic
Introduces AutoVSR, a framework that converts circuit schematics to symbolic expressions using VLMs and a symbolic solver with an executable intermediate representation.
Compile, Then Page: Executable SOP Programs and a Capability-Gated Runtime for Procedural LLM Agents
Proposes a method to compile SOP constraints into executable pseudo-code for LLM agents, using a capability-gated runtime to improve adherence to safety-critical procedures.
From Neural Network Decisions to Training Cases: An Exact Account via Case-Based Decision Theory
Provides a mathematical account of neural network decisions using Case-Based Decision Theory, allowing audit signals to trace scores back to training cases without retraining.
OpsMem: Dual-Memory Reasoning with Cross-Memory Resonance for Failure Diagnosis
Presents OpsMem, a dual-memory framework for software failure diagnosis that coordinates current diagnostic state with long-term operational experience via cross-memory resonance.
StructAgent: Harness Long-horizon Digital Agents with Unified Causal Structure
Introduces StructAgent, a state-centered framework that uses unified causal structures and verifier-backed transitions to improve LLM/VLM performance on long-horizon digital tasks.
Just Let Me Write Digits
A Hacker News discussion thread focused on the frustrations of writing numerical digits in a specific software context.