Hugging Face Daily Papers
500 articles archived · Visit source ↗ · RSS
-
Hugging Face Daily Papers research 1mo ago
AgentJudgeBench: A Multi-Difficulty Benchmark for Evaluating LLM Judges on Agentic Tool-Calling
Abstract AgentJudgeBench reveals that LLM judges face structural reliability limits on dependency-driven agentic tool-calling workflows, with alignment degrading by difficulty and ground-truth exposure yielding mixed effects. Generated by thinkingmachines/Inkling-Small LLM…
32 -
Token-Efficient Data Reasoning Agents via Adaptive Structuring of Unstructured Data
Abstract Valuable data remains embedded in unstructured sources: web pages, reports, contracts, filings, earnings calls, and PDFs. The big bet in enterprise AI is deploying LLM agents that reason over this data to answer complex questions for every knowledge worker. Agents can…
6 -
Adapting Without Gradients: Affine Statistics Transport and What Its Certificate Can Tell You
Abstract CASTER enables gradient-free test-time adaptation for frozen models by transporting source class distributions via affine transformations in a discriminative subspace, with a transportability certificate to gate unreliable updates. Generated by…
10 -
DramaChain Bench: An End-to-End Benchmark for Short-Drama Generation
Abstract DramaChain Bench evaluates the full short-drama production pipeline across stages, dimensions, and cascading defects using calibrated agents and expert annotations. Generated by thinkingmachines/Inkling-Small Commercial short-drama production follows a multi-stage…
7 -
Learning Where Outcomes Change:Credit-Addressable Reasoning for Multimodal Geometry
Abstract Credit-addressable reasoning via executable code traces and localized reinforcement learning improves multimodal geometry reasoning by aligning credit assignment with structured reasoning events. Generated by thinkingmachines/Inkling-Small Multimodal geometry reasoning…
5 -
From Production Traffic to Post-Training: Building a Self-Hosted LLM That Covers the Corporate Request Mix
Abstract A smaller self-hosted LLM trained with separate GRPO experts merged via SLERP outperforms a much larger baseline on instruction following, function-calling, and internal tasks while serving half of platform traffic at lower cost. Generated by…
31 -
Recursive Criticality of AI Self-Improvement
Abstract A recursive reproduction number determines whether AI R&D feedback loops become self-amplifying based on research productivity, feedback strength, and rising difficulty. Generated by thinkingmachines/Inkling-Small AI is increasingly used in the R\&D process that…
29 -
Harness-of-Harness: Multi-Day Autonomous Software Development with Continual Improvement
Abstract Harness-of-Harness enables autonomous coding agents to iteratively improve software through structured planning-coding-testing loops, yielding substantial performance gains and fully developed applications. Generated by thinkingmachines/Inkling-Small This paper studies…
16 -
Qwen-Drive-1.0: An Initial Step towards a Vision-Language Foundation Model for Autonomous Driving
Abstract Qwen-Drive-1.0 is a vision-language foundation model for autonomous driving that unifies 3D perception, visual question answering, and motion planning via shared representations and staged training. Generated by thinkingmachines/Inkling-Small We present Qwen-Drive-1.0,…
29 -
Agents in the Large: Perception-Centered Architecture for Persistent Agents
Abstract The paper proposes a perception-centered architecture for persistent language agents that continuously adapt service procedures by perceiving signals from tasks, context, and environmental changes. Generated by thinkingmachines/Inkling-Small Cognitive language agents…
8 -
DiagEvo: Diagnosis-Guided Self-Evolution via Hierarchical Error Memory
Abstract DiagEvo improves language-model self-evolution by deriving training direction from internal failure history via hierarchical error-cause memory and double-confidence filtering, outperforming external-resource baselines. Generated by thinkingmachines/Inkling-Small…
20 -
Safin-1: Safety from Within through Memory-Native State Evolution
Abstract Safin-1 introduces a memory-routing architecture that embeds safety as an internal, evolving model state rather than an external constraint. Generated by thinkingmachines/Inkling-Small Long-horizon complex tasks require foundation models to accumulate information,…
21 -
UI-Venus-2 Technical Report
Abstract UI-Venus-2 is a general-purpose multimodal GUI agent that uses unified reasoning-action loops, expanded environment coverage, and robust verification to enable reliable real-world digital automation. Generated by thinkingmachines/Inkling-Small Multimodal GUI agents have…
34 -
SMELT: Scaling Laws for Compute-Matched MoE Looped Transformers
Abstract Looping middle layers in sparse Mixture-of-Experts Transformers improves training efficiency and downstream performance while matching per-token FLOPs, parameters, and cache budgets. Generated by thinkingmachines/Inkling-Small Looped Transformers increase effective…
19 -
ZimaBlue: Evolving Generalizable World Action Models through Scalable Video Pre-training
Abstract ZimaBlue learns generalizable world action models from large-scale egocentric video via a three-stage curriculum and a slow-fast architecture, substantially improving zero-shot robotic manipulation. Generated by thinkingmachines/Inkling-Small Robotic manipulation faces…
12 -
E-Commerce Bench: Evaluating LLM Agents on Long-Horizon Autonomous Business Operation
Abstract A year-long e-commerce benchmark evaluates LLM agents on multi-store negotiation, dynamic market events, and long-horizon policy adaptation across 18 frontier models. Generated by thinkingmachines/Inkling-Small Long-horizon agentic tasks go beyond chaining short tasks…
12 -
H3-World: Turning Language Understanding into World Control
Abstract We present H3-World, an efficient framework that turns the 33B MiniMax-H3 video generator into an interactive world model. Our key finding is that, as large video generators become more capable, language is emerging as a natural interface for control. MiniMax-H3, for…
36 -
StudentSim: Training LLM-based Student Simulators
Abstract StudentSim trains personalized student simulators from sparse data to mirror learner responses and adapt to tutor guidance, outperforming existing models across chess, writing, and math. Generated by thinkingmachines/Inkling-Small AI tutors are most useful when they…
37 -
Evaluating Multimodal LLMs as Generalist Vision-Language-Action Agents for Drone Control: Commanding, Approaching, Tracking and Searching
Abstract DroneCATS evaluates multimodal language models as drone controllers and finds that small open models navigate well but fail at protocol adherence and termination, highlighting a gap between spatial perception and disciplined action. Generated by…
7 -
Control-Data Flow Separation: Stable Prompt Optimization in Multi-Agent LLMs
Abstract The framework separates structured execution protocols from optimizable language content to prevent prompt optimization from corrupting multi-agent pipelines. Generated by thinkingmachines/Inkling-Small Prompt optimization can improve multi-agent LLM systems, but the…
35 -
Uncovering Understanding-Generation Synergy in Native Unified Multimodal Models: From Representation, Task to System
Abstract Unified multimodal models achieve synergy between visual understanding and generation through specialized architectures, shared knowledge, and end-to-end optimization rather than simple functional unification. Generated by thinkingmachines/Inkling-Small While unified…
29 -
Hi-Q: Hierarchical Evidence-guided Query Refinement for Multi-Hop Question Answering
Abstract Hi-Q is an evidence-conditioned framework that dynamically refines multi-hop queries into hierarchical trees guided by corpus support signals, improving retrieval and answer accuracy without fixed graphs. Generated by thinkingmachines/Inkling-Small A central bottleneck…
36 -
EM^2Mem: Event-Centric Multimodal Memory for Large Language Models
Abstract EM²Mem binds multimodal evidence to event anchors for compact, generation-ready memory in long-video question answering. Generated by thinkingmachines/Inkling-Small Multimodal memory offers a scalable interface for long-video question answering, but existing methods…
4 -
ReFlowSET: Representation-Aligned Latent Flow Matching for SAR-to-EO Image Translation
Abstract ReFlowSET selects a latent codec via joint SAR-EO reconstruction and trains a small conditional DiT with dual-stream conditioning and frozen vision-model alignment for high-fidelity SAR-to-EO translation. Generated by thinkingmachines/Inkling-Small SAR-to-EO image…
15 -
InternReviewer & InternAdvocate: Objective Reward and Evaluation for Agentic Reinforcement Learning in Peer Review and Rebuttal
Abstract Specialized scholarly agents use reinforcement learning and real-time citation verification to improve reasoning and factual accuracy in peer review and rebuttal generation. Generated by thinkingmachines/Inkling-Small Generating professional scholarly content, such as…
18 -
The Mechanics of Democratic Dominance: A System Dynamics Paradigm for Dynamic Consent Engineering
Abstract Traditional frameworks of political communication operate under linear, event-driven assumptions that treat voter persuasion as a static, transactional function. This paper introduces Dynamic Consent Engineering (DCE), a novel interdisciplinary paradigm that synthesizes…
25 -
RECAP-Forcing: Retaining Content Appearances for Long Video Generation
Abstract RECAP-Forcing improves long video generation by indexing memory according to appearance novelty rather than recency, preserving key-value caches for newly visible content to maintain long-range consistency without extra training. Generated by…
6 -
CoVA-SFT: A Large-Scale Dataset for Chain of Visual Abstractions
Abstract CoVA-SFT is a multimodal reasoning dataset that teaches language models to interleave text and visual abstractions via structured reasoning steps, improving performance on visual reasoning benchmarks. Generated by thinkingmachines/Inkling-Small Chain-of-thought (CoT)…
9 -
MMMMM: A Unified Taxonomy for Investigating the Mechanisms of Multilingual MultiModal Misinformation
Abstract Multimodal misinformation on social media is highly prevalent, potent, and harmful, yet difficult to detect and counter, and still poorly understood compared to its text-only counterpart. Research on the properties and deceptive strategies of multimodal misinformation…
34 -
SpanCalib-VLM: Calibrated Hallucination Span Detection in Vision-Language Models
Abstract A hybrid system combining a multimodal sequence tagger and a generative vision-language model improves hallucination span detection and calibration through union-calibrated fusion. Generated by thinkingmachines/Inkling-Small Detecting hallucinations in Large…
15 -
ContextBias: Controlled Evaluation of Bias Persistence Under Context Shift in Text-to-Image Models
Abstract Text-to-image models exhibit persistent role-linked visual stereotypes across varied contexts, with cross-role attribute concentration increasing rather than diminishing under unrelated prompts. Generated by thinkingmachines/Inkling-Small Text-to-image models learn…
13 -
Chain-of-Thought Faithfulness of Reasoning Models Varies with Where and How Preference Cues Are Delivered
Abstract FACE-Eval reveals that chain-of-thought monitoring is less reliable when preference cues arrive via tool outputs or implicit artifacts, with lower verbalized commitment and higher hidden adoption across diverse open-weight models. Generated by…
19 -
EvoGenUI-Bench: Evaluating LLMs as Multi-Turn Generative UI Assistants
Abstract Large language models can generate interactive web interfaces, but reliable generative UI requires maintaining an executable artifact as user requests evolve. We introduce EvoGenUI-Bench, a benchmark for multi-turn interface maintenance comprising 150 five-turn tasks…
24 -
Uncertainty-Aware End-to-End AI Weather Forecasting: Disentangling Observation and Model Contributions
Abstract The Aardvark Weather model is made probabilistic by adding learned observation noise and Monte Carlo dropout to its encoder and processor, yielding a calibrated nested ensemble that improves mean forecasts and separates aleatoric and epistemic uncertainty. Generated by…
38 -
Chat-Edit-3D++: Interactive 3D and 4D Scene Editing via Large Language Models
Abstract CE3D++ enables flexible dialogue-driven 3D and 4D scene editing by decoupling 2D editing from 3D reconstruction via a Hash-Atlas network and using a large language model to autonomously invoke visual tools. Generated by thinkingmachines/Inkling-Small Recent work on…
22 -
Super Library Agent: Joint Generation and Maintenance of Multiple Applications Beyond the Single Codebase
Abstract A shared-library agent framework reduces redundancy and structural erosion when sequentially generating related applications by guiding cross-application code extraction and migration. Generated by thinkingmachines/Inkling-Small Organizations often develop and maintain…
15 -
SafeAtlas-VL: Beyond Binary Multimodal Safety with Large-Scale Data and Guard Models
Abstract SafeAtlas-VL introduces a large-scale multimodal safety dataset with graded risk labels and trains guard models that predict continuous safety scores, achieving state-of-the-art generalization. Generated by thinkingmachines/Inkling-Small Multimodal safety moderation…
30 -
Evaluating the Hidden Costs of Personalization in Large Language Models
Abstract The study proposes PRISK, a framework that reveals how personalized context in LLMs increases irrelevant personalization, preference narrowing, and sycophantic bias. Generated by thinkingmachines/Inkling-Small While Large language models (LLMs) incorporate user…
17 -
Matrix-Game 3.5: Enhancing Real-Time Streaming Interactive World Models with Patch Memory
Abstract Matrix-Game 3.5 improves interactive world generation via geometry-aware memory, static-dynamic disentanglement, and progressive distillation for long-horizon real-time simulation. Generated by thinkingmachines/Inkling-Small Interactive world models extend video…
13 -
Scaffolding Foundation Models into Physical-World Agents Pushes the Frontier of Long-Horizon Navigation
Abstract NavMCP integrates vision-language reasoning with navigation execution via structured collaboration channels to enable persistent long-horizon embodied exploration without retraining. Generated by thinkingmachines/Inkling-Small Long-horizon physical-world agents must…
35 -
PaperBanana-Interact: Scientific Diagram Refinement with Multi-Turn Human Feedback
Abstract A multi-turn benchmark and multi-agent system improve scientific diagram generation by reducing quality drift and feature forgetting across revision turns. Generated by thinkingmachines/Inkling-Small Recent efforts have aimed to automate scientific diagram generation…
22 -
BLARM: Animating 3D Objects from Video via Blending Latent Rigid Motion Primitives
Abstract BLARM predicts temporally coherent 3D mesh animations from monocular video using learned rigid motion components and skinning weights without explicit rigs. Generated by thinkingmachines/Inkling-Small We introduce BLARM, a feed-forward method for video-driven 3D mesh…
36 -
LightNav-0: Eliciting VLM Spatial Intelligence for Generalist Embodied Navigation
Abstract LightNav-0 is a compact generalist navigation model that leverages a pretrained vision-language model’s spatial reasoning via unified pointing tokens and action tokenization to achieve state-of-the-art embodied navigation across diverse tasks and robots. Generated by…
34 -
WebWorld: The Browser as a World Model for Self-Improving Web Code
Abstract WebWorld uses a browser as an executable world model to certify web code repairs via interaction contracts, enabling autonomous VLM self-improvement with verified supervision. Generated by thinkingmachines/Inkling-Small VLM-driven self-improvement of web code has a…
9 -
Keep-or-Drop? Adaptive Tokenizer for Compact Video Representation
Abstract KATok is an adaptive transformer-based video tokenizer that selectively drops uninformative tokens to achieve data-dependent compression while preserving spatial consistency for diffusion-based generation. Generated by thinkingmachines/Inkling-Small Latent diffusion…
10 -
DICS: Exploring Data Intrinsic Consistency for Visual Instruction Selection
Abstract DICS selects high-quality visual instruction data by measuring internal sample consistency, improving model performance with far less training data. Generated by thinkingmachines/Inkling-Small Visual instruction tuning is crucial for advancing the vision-language…
17 -
Lies We Can See: Joint Verbal and Non-Verbal Deception by VLM Agents in Embodied Social Interactions
Abstract MineAmongUs introduces a 3D multimodal Among Us environment and the ARIA harness to study embodied VLM-agent deception through verbal and non-verbal actions, revealing non-verbal channels as key to winning. Generated by thinkingmachines/Inkling-Small Strategic deception…
30 -
On the Design of Qwen3.8-Next Architecture: Evaluation, Efficiency, and Training Stability
Abstract Qwen3.8-Flash-Next is a sparse mixture-of-experts architecture combining hybrid gated delta-net and sparse attention layers, gated residual branches, and off-accelerator n-gram embeddings to improve efficiency, capability, and training stability. Generated by…
5 -
Verification-Aware Training for Speculative Decoding
Abstract Verification-Aware Training improves draft models for speculative decoding by simulating sequential verification during training and adapting loss weights to acceptance patterns. Generated by thinkingmachines/Inkling-Small Speculative decoding accelerates large language…
8 -
Weaving Visual Narratives: Agentic Image Bundle Composition Beyond Atomic Visual Matching
Abstract Image Bundle Composition reframes retrieval as dynamic assembly of relationally coherent image groups, with a benchmark and agentic framework addressing combinatorial joint relevance. Generated by thinkingmachines/Inkling-Small Image retrieval has traditionally been…
18