All papers

The paper guides knowledge-graph retrieval agents by evaluating the causal impact of evidence on answering questions rather than text similarity.

The paper proposes a parameter-efficient method leveraging frozen model representations for multi-output text regression tasks across diverse benchmarks.

LR-Guard distills explicit safety reasoning into compact latent representations to speed up guardrail classification while enabling on-demand trace reconstruction.

The paper proposes a sparse factor framework to map multimodal behavioral data to interpretable concepts for depression assessment.

Introduces an execution-grounded dataset and method to repair geometric and engineering constraint violations in LLM-generated industrial scene programs.

The paper pairs language models with symbolic verifiers to test and update reusable hypotheses for environment diagnostics and repair.

The paper proposes a prefix-augmented distillation method providing one-step student branch feedback during language model pretraining.

This paper presents an instruction-following text-to-speech model that separates voice design from speech generation using an intermediate voice latent.

The authors propose verifiable agent insurance using smart-contract escrow to financially compensate users when agents violate programmatically logged behavioral promises.

Audits audio-language model hallucination mitigations by testing localized acoustic evidence ablation against baseline methods across model architectures.

The authors introduce a framework to benchmark language model agent robustness under interaction perturbations across diverse tasks.

The paper presents adapted small language models designed to directly score agricultural candidate choices in a single forward pass.

This paper presents a hybrid layer-set diffusion framework for generating editable and visually rich educational infographics.

This paper proposes an inference-time intervention method to detect and correct modality misrouting that causes multimodal hallucinations.

Proposes a benchmark and probing methodology based on attribution theory to examine how language models assign responsibility and blame.

Uses large language models to construct prior trade-off preferences to address cold-start challenges in choice-learning systems.

This paper proposes a decoupled marginal sharpening method to improve inference-time reasoning in language models without weight updates or reward models.

The paper proposes VAttnPO, using teacher visual attention to refine token-level policy optimization for multimodal reinforcement learning.

This paper presents a post-hoc multicalibration framework to correct systematic biases such as position and model family in LLM evaluations.

Evaluates language models on predicting program outputs and intermediate loop checkpoint states directly from source code and inputs.

Presents a training framework that teaches language model agents to transfer reusable problem-solving experience across structured task families.

This paper proposes a contrast-calibrated bit allocation method for mixed-precision quantization of mixture-of-experts language models.

The paper introduces a reciprocal self-improvement framework to balance agent understanding and image generation roles within multimodal models.

This paper presents an evaluation benchmark and guardrail framework for detecting safety risks in multi-turn interactions with AI companions.

The paper trains conversational user simulators equipped with latent evaluation thoughts and scalar ratings to provide intrinsic feedback.

Creates a structured schema that converts research papers into connected scientific entities to improve retrieval-augmented generation for autonomous agents.

This paper proposes a closed-form covariance-aligned rotation method to improve post-training low-bit quantization in large language models.

Presents FaithRS, a self-reflection framework with a fine-grained dataset to detect and correct reasoning hallucinations in remote sensing models.

This paper proposes decoupling error proposing from error validation to filter invalid candidates under budget constraints for expert human review.

The paper benchmarks a specialized decision model against general-purpose LLMs on English and Chinese stance detection tasks.

The paper proposes a federated learning framework that uses version control operations instead of full document aggregation to evolve agent skills.

Introduces an adaptive spectral gradient clipping method to stabilize and accelerate large language model pretraining.

Presents Omni-RS, a unified framework and dataset supporting multiple remote sensing image understanding and generation tasks.

This paper introduces an evaluation framework and reward models to assess how well search agents ground generated answers in retrieved, independent sources.

Analyzes rubric construction mechanisms and introduces a tree-search method to refine rubrics for evaluating language model outputs.

The paper introduces an agenda-driven neuro-symbolic framework to improve the reliability of user simulation when testing interactive conversational agents.

The paper proposes an alternating context optimization and logit steering framework to adapt language model outputs at test time.

This paper presents a framework to pretrain and fine-tune text-to-image diffusion transformers using four consumer GPUs.

This survey analyzes how verifier noise affects reinforcement learning with verifiable rewards across 58 studies and summarizes practical mitigation techniques.

Designs diagnostic probes to determine whether audio-language evaluation models judge music based on audio or prompt content.

Develops a structured explainable reasoning architecture to extract sensitive self-disclosures from cybersecurity interviews using local small language models.

The paper presents an agent orchestration framework that trains smaller language models to navigate, inspect, and reason over long multimodal documents.

This paper proposes an image editing framework that coordinates prompt understanding and image generation within a single unified model.

This paper presents a compact decision model designed to screen malicious agent tool skills with high recall directly from specification files.

Introduces a benchmark evaluating streaming vision-language models on fast-moving, high-dynamic events in real-world video streams.

The paper presents open-weight language models adapted for the Wolof language using instruction tuning, preference alignment, and model merging.

The paper proposes a test-time aggregation method using bias control and injection to resolve disagreements among reasoning language model agents.

Adapts text-to-video diffusion models for surgical procedure simulation using hierarchical procedural decomposition and progressive curriculum learning.

This paper introduces a counterfactual benchmark to evaluate honorific register allocation bias across gender and occupations in English-to-Bengali translation.

Evaluates SHAP attribution methods on sparse autoencoder features, finding weak correspondence with direct causal feature intervention in language models.

The paper proposes a federated learning framework that tailors sub-network sizes and architectures to heterogeneous client hardware and data.

Analyzes trade-offs between aggressive and conservative planning horizons in language models operating under partial observability.

Introduces a deficit-driven stratified acquisition algorithm to create a balanced, co-registered satellite imagery corpus for self-supervised Earth observation.

This paper audits Chinese language models for bias and error propagation across formal and ironic social registers.

Presents ADMA, an adaptive deliberation controller that dynamically adjusts multi-agent critique and coordination to improve reasoning accuracy.

Introduces a text-to-4D generation method that separates static backgrounds and dynamic characters to allow modular editing and camera control.

The paper introduces a training-free schedule optimization method for accelerating sequential Monte Carlo guidance in discrete diffusion models.

Presents a multi-turn benchmark measuring language agent capabilities on complex, long-horizon bioinformatics workflows.

Presents a multimodal benchmark testing whether models can explain observed chemical effects by linking structural features to reaction mechanisms.

The paper introduces local contract-conditioned rewriting to prevent sensitive data leakage in email-generating language model agents cost-effectively.