The Archive

LLM-driven evolutionary program search optimizes semantic index design for generative retrieval, automating end-to-end SID component tuning.

OptScientist framework discovers symbolic optimization laws from data using neural inference and LLM-agentic exploration.

Empirical study comparing LLM-based and classical feature selection on tabular data with GRPO-adapted Qwen and Llama models.

DecisionPivot benchmark evaluates routing, guardrail, and verification models in LLM systems across 48 tasks with 40K test items.

BitMTP combines ultra-low-bit quantization with multi-token prediction for efficient LLM decoding without quality loss.

OptionWorld benchmark reveals LLMs fail at multi-day task completion, with 35.5% of failures due to lost context before deadline.

Protocol for LLM-driven codebook generation and evaluation to measure parliamentary discourse quality while mitigating model bias.

Analysis of LLM generic acceptability judgments aligned with psycholinguistic factors across twelve models.

ACE-Compression framework evaluates which core LLM abilities survive compression beyond task performance metrics.

Polarity blindness vulnerability: LLMs fail safety checks after negation and antonym substitution despite preserved semantic meaning.

CollabSciReM aligns LLMs for scientific collaboration via science-aware reward decomposition, balancing accuracy, rigor, and token efficiency across multi-turn interactions.

Benchmark study of associative creativity in MLLMs vs. humans on abstract visual stimuli with controlled naming difficulty; evaluates originality and effectiveness.

LLMs hallucinate in cross-language code translation via semantic conflicts, target-language violations, and inconsistent dependencies; existing mitigations insufficient.

VISH framework selects compact hotword subsets for speech-LLM contextual ASR via frame-level CTC posterior verification; training-free and scalable via shared-prefix Viterbi search.

UvA-MT uses LLM sampling, refinement, and pairwise voting for WMT26 machine translation across 23 language pairs without task-specific training.

SAPD learns LLM agent policies from offline trajectories without environment interaction, combining offline RL with policy refinement to avoid forgetting.

COMPASS agentic router selects optimal LLMs by profiling capabilities via tools and pairwise selection, achieving 69.85% accuracy at lower cost than trained baselines.

ProVe reverses speculative decoding: large model proposes token blocks, smaller model verifies in parallel, improving LLM inference speed.

SliMOO applies multi-objective evolutionary search to LLM depth pruning, preserving task-diverse capabilities while reducing memory and latency.

Agentic workflow guides LLM coding agents to implement 6G physical-layer algorithms on embedded baseband silicon with vector assembly optimization.

Mechanistic interpretability study investigates circuit overlap between linguistic and arithmetic capabilities in LLMs using causal analysis.

Relay uses LLM agents to diagnose and reuse repairs for historical software build failures, recovering 69.7% of failing OSS-Fuzz cases.

Gradus synthesizes graded training tasks for LLM agents in policy-constrained environments by decoupling task design from solution difficulty.

RITE benchmark evaluates LLM agent resilience to silent tool errors in e-commerce systems where incorrect intermediate outputs lack failure signals.

Measures how LLM probability estimates shift from single new information with varying framing using δ-SNLI dataset.

OpenVLHarness framework enables multimodal LLMs to use external tools and retain evidence across reasoning steps for visual agentic tasks.

FunCAD-Harness framework evaluates LLM agents on functional CAD mechanism design by verifying physical requirements in simulation.

PhysRSI-eval benchmark quantifies physical recursive self-improvement in VLMs across 18 manipulation tasks; identifies bottlenecks via Physical Barrel Hypothesis.

LatComm introduces reusable latent context compression between LLMs, enabling compact encoding that is readable across multiple models and questions.