Every paper checked — nothing hidden
All papers
Every paper checked in the week of 12–18 Oct 2026 · Multimodal models — nothing is hidden. Brief papers first.
All1,924In brief3More (on-topic)553Not relevant1,333Not assessed19Older33Could not confirm19Blogs with no paper11All blogs13
| Paper | Topic | Relevance | Feedback |
|---|---|---|---|
| Stop Training Against Soft Users: Mentalized Adversarial Self-Play for Proactive Dialogue The paper introduces an adversarial self-play framework modeling user mental states to improve proactive dialogue agent planning. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Concerns textual proactive dialogue without multimodal components. | |
| From Long Context to Selective Memory:A Survey of Memory-Centric Compression for Large Language Models The paper surveys memory-centric compression techniques for managing inference resources and context windows in large language models. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Focuses on text context window compression, not multimodal models. | |
| Beyond Retrieval Failure: Understanding Why LLMs Miss Evidence in Multi-Hop RAG The paper introduces ExecutionRAG to show how search drift causes apparent retrieval failures in multi-hop question answering. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Focuses on text retrieval rather than vision-language or multimodal inputs. | |
| KBS: A Korean Benchmark Suite for Foundation Model Evaluation The paper introduces the Korean Benchmark Suite to evaluate foundation models across six natural language capabilities. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Evaluates text capabilities without image, video, or audio multimodal inputs. | |
| What Should We Remove? Evaluating the Reliability and Costs of RLVR Data Decontamination The paper audits data decontamination methods on math corpora to quantify false removals and benchmark leakage risks. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Examines text-based math training data rather than multimodal systems. | |
| FGSS-BO: Feedback-Guided Hypothesis Management for LLM-Based Bayesian Optimization The paper proposes FGSS-BO, adding structured hypothesis management to language-model-guided Bayesian optimization. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Operates on textual hypotheses and tabular parameters without multimodal inputs. | |
| LLM-Based Weak Annotation for Cybersecurity Tweet Classification: A Teacher-Student Framework The paper presents a teacher-student framework using language models for weak annotation to classify cybersecurity tweets. Peer-reviewed | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Exclusively processes social media text, without multimodal data modalities. | |
| Isolating a Language Model's Judgment with a Label-Free Subspace The paper proposes isolating internal model judgments from instruction formatting variations using a label-free hidden state subspace. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Evaluates language models on text judgments without multimodal components. | |
| Making every stage of medical device document assessment with large language models deterministic The paper evaluates output non-determinism in commercial language model deployments during medical device regulatory compliance assessments. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Focuses on text extraction and retrieval rather than multimodal vision-language models. | |
| VALCON: Benchmarking and Enhancing Value Consistency in Large Language Models via Dual-Role Variation The paper introduces VALCON to evaluate and improve value consistency when language models alternate between generator and evaluator roles. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Evaluates text-based value judgments, not multimodal vision-language architectures. | |
| Hybrid Generative-Tagging Decoding for Efficient Product Attribute-Value Extraction The paper introduces a hybrid generative-tagging architecture to improve efficiency and coverage in product attribute and value extraction. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Operates on text-based product descriptions, not multimodal models. | |
| VoteGEO-Bench: Measuring the Impact of Generative Engine Optimization on Electoral Visibility The paper presents VoteGEO-Bench to measure how rewriting web content influences information visibility in generative search engine election queries. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Analyzes textual web queries and generative engine outputs, not multimodal models. | |
| QuasiLeak-Bench: Evaluating LLM-Based Privacy Detectors on Policy-Conditioned Quasi-Evidence Group Recovery The paper introduces QuasiLeak-Bench to evaluate whether privacy detectors can identify distributed quasi-identifiers governed by disclosure policies. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Operates strictly on natural language messages, not multimodal inputs. | |
| Ranking and Partitioning of Attention Heads for Sparse and Continual Reinforcement Learning Fine-tuning The paper studies ranking and partitioning attention heads by gradient norm for sparse and continual reinforcement learning fine-tuning. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Addresses text-based reinforcement learning fine-tuning, not multimodal architectures. | |
| OpCredit: Operation-Aware Policy Learning for Agentic Knowledge Graph Question Answering The paper proposes OpCredit, a policy learning framework that assigns individual operation credits for agentic knowledge graph question answering. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Applies to text and structured knowledge graphs, not multimodal vision models. | |
| Emotion Is Not External Knowledge: Internal Attention Signals for Affective State Reasoning The paper presents an internal attention boosting method to improve how language models infer implicit conversational affective states. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Focuses on multi-turn text dialogue, not multimodal cross-modal models. | |
| RenoLayout: Unified Floorplan Renovation and Furniture Layout for Constraint-Aware Residential Design The paper introduces RenoLayout to generate residential renovation floorplans and furniture layouts under structural and physical constraints. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Focuses on structured floorplan geometry generation rather than multimodal vision-language models. | |
| DPA-MoE: A Runtime Precision Control Framework for MoE Inference Introduces a dynamic precision control framework that allocates weight chunks to mixture-of-experts models during inference. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Deals with MoE language model architectures, not multimodal inputs. | |
| Calibration Language Shapes Block-Influence Pruning in Multilingual Language Models Analyzes how calibration language choices affect block-influence depth pruning rankings in multilingual language models. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Focuses exclusively on text-based multilingual models, not multimodal systems. | |
| Identifying Effective Strategies in Prompt Evolution for Stress and Emotion Prediction Evaluates prompt optimization methods and identifies effective prompting strategies for social media stress and emotion classification. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Operates strictly on text data from social media, not multimodal data. | |
| HyCo: Hybrid Drafting for Cost-Efficient Mobile-Cloud Speculative Decoding Introduces a hybrid speculative decoding framework splitting draft tasks between mobile neural processing units and cloud servers. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Focuses on language model speculative decoding, not multimodal cross-modal architectures. | |
| SITAR: A Situational Distortion Metric for Automatic Speech Recognition Evaluation Develops a lightweight situational distortion metric to evaluate speech recognition errors based on practical conversational severity. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Focuses on evaluating speech transcription text rather than multimodal vision-language architectures. | |
| Translation Circuit Discovery and Targeted Optimization for Low-Resource Multilingual Translation Identifies translation circuits in multilingual models and optimizes fewer than five percent of parameters for low-resource translation. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Applies to text-based multilingual translation without any multimodal components. | |
| OPRD: On-Policy Reminder Distillation for Long-Horizon Coding Agents Uses multi-scope reminders and on-policy distillation to help coding agents maintain constraints over long software engineering interactions. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Focuses on text-based code repositories, not multimodal models. | |
| Writability and the Pathway: A Controlled Study of Knowledge Injection into Addressed Memory Conducts a controlled evaluation of direct knowledge injection into addressed conditional memory modules across small and production models. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Deals strictly with text-based language models, not multimodal inputs. | |
| Waypoint-OPD: Transferring RL-Induced Policy Shifts via Gated On-Policy Distillation Gates reinforcement learning policy-shift distillation using intermediate checkpoints to ensure student models only learn consistent teacher shifts. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Focuses on mathematical and code text reasoning, not multimodal data. | |
| ContextLift: Breaking the Spurious Correlation Behind Conditional Emergent Misalignment Introduces a data augmentation method that breaks spurious contextual correlations causing emergent misalignment during domain finetuning. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Evaluates text language models without image, video, or audio data. | |
| When Does "Post-Call" Begin? An Availability Contract for Earnings-Call NLP Benchmarks Proposes an availability framework to identify and resolve within-event timing mismatches in earnings-call financial NLP benchmarks. Peer-reviewed | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Focuses on financial benchmark data timing rather than multimodal modeling. | |
| Diffusion Language Models as Open Quantum Systems Frames diffusion language models as open quantum systems to enable text generation conditioned on superpositions of prompts and thoughts. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Applies diffusion models solely to text generation tasks. | |
| No Single Tokenizer Feature Reliably Predicts Downstream Language Model Performance Shows across 200 models that proposed intrinsic tokenizer metrics do not reliably predict downstream language model performance. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Investigates text tokenizers without visual or auditory multimodal components. | |
| Data Strategies for Large Language Models: A Unified Perspective on the Training Lifecycle Reviews and formalizes data selection, weighting, and scheduling policies across the pretraining and post-training lifecycle of language models. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Concentrates on standard text language model pretraining and post-training data. | |
| Abstracts or full text? Evaluating input scope and computational efficiency in biomedical lay summarization Compares abstract-only versus full-text inputs and few-shot versus fine-tuning approaches for generating biomedical lay summaries. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Deals strictly with text documents without multimodal content. | |
| SceneWeave: Audience-Guided Scene Exploration for Dramatic Adaptation Introduces a system that generates and compares alternative scene drafts to guide full dramatic script adaptations from prose. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Works exclusively with text prose and screenplays, lacking other modalities. | |
| Behavioral Forgetting Is Not Privacy in Machine Unlearning Demonstrates that machine unlearning often fails to prevent targeted information recovery across alternative multilingual access paths. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Evaluates multilingual text models without image, video, or audio data. | |
| Content associations with answer selection and engagement in supply chain community question answering Analyzes how specific nouns and semantic clusters relate to answer selection and user engagement in logistics question-answering forums. Peer-reviewed | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Analyzes text forum data without multimodal vision or audio components. | |
| N-Gram Memories at Two Bits: Quantizing Lookup Tables in Large Language Models Proposes a compressed representation using Gaussian trellises to quantize conditional memory lookup tables down to two bits. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Focuses on language model parameter quantization without multimodal processing. | |
| Dynamic ordering in sequence generation in Transformers: experiments on low-resource Tagging scenarios Introduces a non-autoregressive transformer model that dynamically decides tag generation order at inference time for low-resource sequence labeling. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Focuses entirely on text sequence labeling without multimodal inputs. | |
| TEM: Temporal State Tracking Memory for Large Language Model Agents Presents a training-free framework that tracks changing conversational facts as versioned records to improve agent memory retrieval. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Operates entirely on conversational text without image or audio modalities. | |
| WEAVE: Weaving Real-World Code Evolution into Dependency-Structured Tasks for Benchmarking and Training Code Agents Introduces a benchmark that structures multi-pull-request code changes into dependency graphs to evaluate long-horizon software engineering agents. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Only deals with text and code repositories, not multimodal data. | |
| Effectiveness of Hidden-State Probes for Localizing Reasoning Failures Evaluates whether internal hidden-state probes can identify failure modes and pinpoint failing steps in multi-step reasoning traces. Peer-reviewed | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Focuses solely on text-based language model reasoning without multimodal inputs. | |
| Name-Bias in Large Language Models Measures demographic bias related to age, gender, and race based on first names across local language models. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Examines text prompt fill-in-the-blank responses only. | |
| Documenting Trofinetide Treatment Phenotypes in Rett Syndrome with a Dual-LLM Consensus Pipeline Applies a dual language model consensus pipeline to extract clinical phenotypes from electronic health records for patients with Rett syndrome. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Analyzes electronic clinical text notes, not multimodal data. | |
| Beyond Context Length: Characterising the Difficulty of Long-Context Reasoning Presents a benchmark that measures long-context reasoning difficulty using evidence scope, dispersion, and reasoning depth. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Focuses strictly on long-context textual reasoning. | |
| The Right Byte First: How UTF-16 Helps Multilingual Boundary Prediction Studies how using UTF-16 byte order improves boundary prediction and representation efficiency in multilingual byte-level language models. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Concerns text character encodings without multimodal inputs. | |
| Patch where the text is new: a compression view of where byte-level models should spend compute Introduces a compression-based patching strategy for byte-level language models that allocates compute where upcoming bytes are unlikely to repeat. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Evaluates text and byte-level modeling without multimodal inputs. | |
| Unfragmenting Nawatl NMT: Single-Variant Specialization for Masewalkopa Specializes a translation model to a specific indigenous Nawatl variety using a curated dataset and finite-state morphological analysis. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Only handles text translation, not multimodal data. | |
| PremiseSQL: Evaluating Premise Critique and Abstention in Text-to-SQL Systems Introduces a benchmark to evaluate whether text-to-SQL systems can detect flawed query premises and abstain from generating invalid SQL. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Deals with text and SQL code, not multimodal media. | |
| CLUE: Uncovering Social Context for Pragmatic Understanding Develops a two-stage training approach combining evidence reasoning and reinforcement learning to improve pragmatic social reasoning in conversational models. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Deals exclusively with text dialogue understanding. | |
| Lexical Reference Frames: Interpreting LM Residual Streams in Native Semantics Introduces lexical reference frames to interpret intermediate hidden states in language models using the model's own remaining layers. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Studies text-only transformer internal representations. | |
| Hy-Memory-30B-A3B: Attributed Reinforcement Learning for Long-Term Memory Writing Introduces a compact language model trained with distillation and reinforcement learning to manage long-term memory across multiple external frameworks. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Processes textual memory state, not multimodal inputs. | |
| Not X, but When: Diagnosing a Mandarin LLM Writing Tic through Contextual Appropriateness Analyzes the overuse and contextual appropriateness of a common corrective phrasing pattern in Mandarin language model text. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Focuses entirely on text-based Mandarin linguistic patterns. | |
| CAT-alyzing Honesty: Continuous Adversarial Training for Transferable LLM Honesty Presents an adversarial training approach designed to improve language model honesty across diverse benchmarks without degrading core capabilities. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Involves text-only language models. | |
| Multi-Target Few-Shot Prompting with Diffusion Language Models Uses diffusion language models to perform multi-target few-shot infilling across shared code context guided by demonstration examples. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Focuses on text and code diffusion, not multimodal learning. | |
| Grid-Bench: Can LLMs Predict How Discrete Worlds Change? Evaluates the ability of language models to predict state changes in discrete spatial grids across sequential operations. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Evaluates text-based spatial representations without multimodal inputs. | |
| MindEval2: Clinically Grounded Patients and Rubrics for Evaluating Language Models in Mental Health Support Presents an evaluation framework combining simulated mental health patients and expert-developed rubrics to assess conversational language models. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Deals solely with conversational text, not multimodal inputs. | |
| Training Reasoning Models to Generate Diverse Scientific Ideas Introduces a set-level reinforcement learning objective to train language models to produce diverse and document-grounded scientific research ideas. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Exclusively processes textual scientific documents. | |
| Verdict: Encode Once, Decide Many The paper introduces an encoder architecture that processes input once to answer multiple structured classification queries simultaneously. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Operates entirely on text classification without multimodal data. | |
| ARIA: Runtime Inspection and Amendment for Multi-Agent Systems The paper presents a framework and dataset for learning to detect and repair errors during multi-agent system execution. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Focuses on language-based multi-agent execution rather than multimodal models. | |
| When Reasoning Traces Conflict with Evidence: How Answer Configuration Shapes Evidence Following The paper analyzes how answering via a separate model call improves evidence adherence when reasoning traces conflict with provided context. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Operates strictly on text QA benchmarks without multimodal inputs. | |
| A Self-Improving Agent that Grows its Own DSL: Oracle-Free Operator Induction for Exact Question Answering The paper presents an agent that expands its domain-specific language for exact answering without external rewards or judges. Preprint | Multimodal models via arxiv · 17 Oct, 06:17 | Not relevant Operates purely over structured data and text DSLs. |
661–720 of 1,924← PreviousNext →