FuguReport

2026-10-15 Method / Reinforcement Learning / Policy optimization with weighted updates

SCOPE: Semantic Bounded Credit Optimization for Prompt Engineering in Text-to-image Generation

This paper studies reinforcement learning for prompt rewriting in text-to-image generation when the image generator is frozen.

2026-10-15 Method / Robot Assistance / Dual-loop harness for assistance

CarryOn: A Dual-Loop Harness for Personalized and Reliable Robot Assistance

CarryOn is a dual-loop harness built around a frozen vision-language-action policy to enable personalized and reliable robotic assistance.

2026-10-15 Method / Decoding / Speculative forcing approach

Speculative Forcing for Efficient Autoregressive Video Generation

This paper proposes Speculative Forcing, an inference framework for few-step autoregressive video generators that treats intermediate clean token estimates as drafts and commits tokens early once they appear to have converged.

2026-10-15 Method / Reinforcement Learning / RL integration into diffusion generation

Reinforcement Learning-driven Diffusion Models: Language, Vision, and Beyond

This paper presents a comprehensive survey of reinforcement learning (RL)-driven diffusion models, examining how RL post-training transforms diffusion denoising into reward-guided trajectory optimization.

2026-10-15 Task / 3D Affordance Localization / Localizing functional regions on 3D objects

BridgeAfford: Bridging Query-Target Gap for Generalizable 3D Affordance Grounding

This paper addresses language-guided 3D affordance grounding, which localizes functional interaction regions on 3D objects from natural language instructions, with an emphasis on generalizing to unseen objects.

2026-10-08 Method / Video Generation / Goal-directed video world model

WorldGuide: Goal-Directed Video World Model for Procedural Task Execution

The paper formulates procedural video generation as closed-loop task execution in visual world space rather than open-loop synthesis.

2026-10-08 Method / Graph Anomaly Detection / Generalist model without retraining

Is Real-World Training Data Necessary for Generalist Graph Anomaly Detection?

This paper examines whether generalist graph anomaly detection can be trained without real-world anomalous graphs by relying entirely on synthetic data.

2026-10-08 Method / Self-Improvement / Recursive safety improvement framework

ReSI: Recursive Safety Improvement toward Resistant and Resilient AI

ReSI is a recursive safety improvement framework that treats model safety alignment as an iterative loop of red-teaming, automated recipe search, and validated model updates.

2026-10-08 Method / Visual Skill Learning / Framework for visual-native skills

ViSkill: Reinforcing VLM Agents with Evolving Visual-Native Skills

ViSkill is a reinforcement learning framework for vision-language model agents that stores successful interactions as composite visual skill cards rather than reducing them to text.

2026-10-08 Method / Audio-Visual Generation / Simultaneous audio-video waveform generation

SepGen: Multi-Stem Audio-Video Separation and Generation in a Single Model

SepGen extends a pretrained joint audio-video generator so that a single sampling run can produce a video, its mixed soundtrack, and separate waveforms for each captioned source.

2026-10-07 Method / Model Optimization / Token pruning strategies in VLMs

Understanding and Mitigating Token-Pruning-Induced Vulnerabilities in VLMs

This paper investigates how visual token pruning affects the safety of vision-language models under multimodal jailbreak attacks.

2026-10-07 Method / World-Action Models / Causal prediction with historical and future structure

Long-WAM: Scaling the Context of World-Action Models

Long-WAM is a model-and-systems framework for scaling the visual context of causal world-action models under real-time robot control constraints.

2026-10-07 Method / Hand Tracking / Monocular egocentric hand pose estimation

RLHND: Video Foundation Models as Physically Grounded Hand Trackers for Robot Learning

RLHND is a hand tracking method for monocular egocentric video that uses a pre-trained video diffusion backbone as a deterministic feature extractor and predicts both 3D bimanual hand motion and dense tactile signals on the hand surface.

2026-10-07 Method / Memory Mechanisms / Artifact-grounded memory layer

ExperienceIndex: Artifact-Grounded Memory

This paper introduces ExperienceIndex (ExpIdx), a memory layer for AI agents that stores and reuses artifact-grounded experience extracted from prior reasoning traces.

2026-10-07 Method / Graph Neural Networks / Long-range graph learning via operator-valued waves

Oscillatory Neural Dynamics over Sheaves

This paper introduces ONDA, a sheaf neural network framework for long-range graph learning that combines second-order oscillatory dynamics with learned matrix-valued sheaf transport operators.

2026-10-06 Method / Policy Verification / Scaling verification via co-evolution

VeriFine: Scaling Verification for Self-Improvement in Embodied Reasoning

VeriFine is presented as an agent-harness framework for scaling verification in embodied reasoning by co-evolving the policy, the training curriculum, and the judge.

2026-10-06 Method / Backdoor Attack / Answer-side multi-turn triggers

The Model Plants the Trigger: Answer-Side Backdoor Attacks in Multi-Turn Large Language Models

This paper identifies a multi-turn backdoor threat in which the trigger is placed in the model’s own earlier response rather than in the user’s input.

2026-10-06 Evaluation / Benchmarking / Deception capability assessment

DecepEval: A Benchmark for Evaluating Deception in LLM Agents

DecepEval is a benchmark for systematically evaluating deceptive behavior in LLM agents across three task families: tool-use and evidence reporting, coding and test exploitation, and long-horizon process integrity.

2026-10-06 Method / Retrieval / Unified retrieval with frozen LLM

UNREAL: Unifying Retrieval and Long-Context with a Single Model

UNREAL is a model-native evidence selection framework that uses a single frozen decoder-only LLM for both corpus retrieval and long-context reasoning.

2026-10-06 Method / Memory Augmentation / Enhancing LLM with parametric memory

Towards In-Parameter Memory Augmentation for Large Language Models

This paper presents a survey of in-parameter memory (IPM) augmentation for large language models (LLMs) at deployment time.

2026-10-05 Application / Embodied AI / AI agent animating virtual body

Is this machine playing?

This paper investigates how a coding-assistant-based embodied agent behaves when placed in a previously unseen simulated world without explicit tasks, rewards, or success metrics.

2026-10-05 Method / Physical Intelligence / Causally grounded planning

Grounding What Shapes the Plan: Rethinking Groundedness for Physical Intelligence in Autonomous Driving

This paper argues that grounded reasoning in autonomous driving is insufficient if it does not remain consequential for the final action.

2026-10-05 Method / Bilevel Optimization / Single-loop stochastic bilevel method

A Single-Loop, Constant-Batch First-Order Penalty Method for Stochastic Bilevel Optimization

This paper studies nonconvex–strongly convex stochastic bilevel optimization using only first-order stochastic gradients.

2026-10-05 Method / Model Safety Evaluation / Safety signals from latent knowledge

Safeguarding LLMs via Model-Agnostic Latent Safety Signals from Dark Knowledge

This paper introduces LADE, a defense for large language models that detects harmful queries by using latent safety signals found in the first-token output probability distribution rather than internal hidden states.

2026-10-05 Method / Model Compression / Activation denoising in quantization

Activation Denoising: A Robustness View on Parallel vs Sequential LLM Quantization

This paper studies the trade-off between parallel and sequential post-training quantization for large language models, framing the gap as a robustness problem caused by compounding upstream quantization errors.

2026-10-04 Method / Task Planning / Topology-consistent task planning method

Topology-Consistent Task Planning over Cellular Workflow Complexes for LLM-based Agents

This paper introduces TopoPlanner, a task-planning framework for LLM agents that targets workflow structures that go beyond sequential or DAG-like dependencies.

2026-10-04 Method / Self-Supervision / Multi-view self-verification

When to Rethink: Learning Multi-Perspective Self-Verification for Vision-Language Models

This paper studies self-verification for vision-language models (VLMs), focusing on when a model should trust its current answer and when it should rethink.

2026-10-04 Method / Memory Management / Mask-guided KV cache eviction

Mask-Guided KV Cache Eviction in Block Diffusion Language Models

This paper studies inference-time key-value (KV) cache management for block diffusion language models, where long prefixes are both stored in memory and repeatedly read during denoising.

2026-10-04 Method / Reward Modeling / Generative reward modeling training

EnGRICH: Enhancing Generative Reward Modeling with Critiques from Humans

This paper investigates how to train generative reward models (GRMs) so that their natural-language critiques are reliable rather than merely predicting correct pairwise preference judgments.

2026-10-03 Method / Best Arm Identification / Fixed-budget best arm under privacy

Asymptotically Optimal Best Arm Identification with Fixed-Budget under Differential Privacy

This paper studies fixed-budget best-arm identification for Bernoulli bandits under pure ε-differential privacy, where the entire adaptive transcript must be protected.

2026-10-03 Method / Concept Representation / Metonymic circuit mechanism

Metonymic Circuits for Abstract Concept Grounding in Vision Transformers

This paper investigates how Vision Transformers recognize abstract concepts in images when direct referential supervision is scarce.

2026-10-03 Method / Multimodal Perception / Multi-sensory inference and feedback loop

ROMA: LLM System for Real-World Object-Centric Multi-Sensory Active Perception

This paper presents ROMA, an LLM-based system for real-world object-centric active perception that closes a reasoning-interaction-feedback loop across vision, audio, tactile, and force sensing.

2026-10-02 Method / Preference Learning / Modeling human preferences in text

Verifiable, Articulable, and Tacit Components of Preference

This paper investigates how much human preference can be captured by explicit rules versus what remains difficult to articulate or verify.

2026-10-02 Method / Graph Models / Cross-problem graph memory design

Continual Graph Memory for Mathematical Research Agents

The paper presents Ansatz, a mathematical research agent built around Continual Graph Memory, a structured memory system for long-horizon proof search.

2026-10-02 Method / Diffusion Models / Continuous latent diffusion modeling

Large Language Continuous Diffusion Models

The paper introduces Sigma, a family of 3B and 8B continuous diffusion language models that operate over steerable 16-dimensional latent trajectories rather than discrete token states.

2026-10-02 Method / Reward Modeling / Composing preference and rubric rewards

Post-Training Frontier Text-to-Image Models by Composing Preference and Rubric Rewards

This paper presents a post-training recipe for open-domain text-to-image models that combines a Bradley–Terry human preference reward trained on approximately 5 million pairwise votes with open-ended, prompt-conditioned rubric rewards.

2026-10-02 Application / Robot Manipulation / Precise manipulation using active gaze

EyeRobot 2.0: Active Gaze for Precise Manipulation without Wrist Cameras

EyeRobot 2.

2026-10-01 Evaluation / Benchmarking / Research paper retrieval benchmark

ScholarCatalyst: A Benchmark for Retrieving Papers That Inspire New Research

ScholarCatalyst is a benchmark for retrieving prior papers whose ideas did or could have meaningfully advanced a research project.

2026-10-01 Method / Representation Learning / Geometry-supervised latent learning

GeoLatent: Geometry-Guided Latent Structuring with Routed Optimization for 3D Reasoning

GeoLatent addresses 3D spatial reasoning in vision-language models by improving decomposed latent representations for position, direction, and global geometry.

2026-10-01 Method / Video Understanding / Unified perception and response in streaming video

OneStreamer: Unifying Perception, Memory, and Proactive Response in Streaming Video Interaction

OneStreamer is a 4B streaming video large language model designed to unify real-time perception, reusable memory formation, and proactive task response in a single causal generation process.

2026-10-01 Method / Agent Design / Decomposition of agent components

Agents Are Systems, Not Models: Rethinking Agentic Evaluation

This paper argues that agentic evaluation should treat an agent as a configurable system rather than as a fixed model.

2026-10-01 Method / Diffusion Models / Hierarchical continuous diffusion process

Hierarchical Continuous Diffusion Language Models

This paper proposes Hierarchical Continuous Diffusion Language Models (HC-DLM), a diffusion framework that couples discrete token generation with a continuous latent trajectory in a single reverse process.

2026-09-30 Method / Action Modeling / Learning from historical trajectories

Learning Skills from Historical Action Trajectories: Action Experience Dictionary for World Action Models

This paper proposes an Action Experience Dictionary (AED) for world action models in robotic manipulation.

2026-09-30 Method / Self-Distillation / On-policy training recipe

UniEvo-VL: An On-policy Self-Distillation Training Recipe for Multimodal Model Self-improvement

This paper presents UniEvo-VL, a self-evolution framework for multimodal models that uses their own visual critiques to improve later image generation.

2026-09-30 Method / Diffusion Models / Looped transformer execution

Looped Diffusion Transformer

This paper investigates scaling text-to-image diffusion models by repeatedly executing a shared group of Transformer blocks within each denoising step, thereby increasing computational depth without adding parameters.

2026-09-30 Method / Exploration Strategy / Tactile curiosity-based exploration

Tactile Curiosity Drives Robot Interaction

This paper introduces TacEx, a reinforcement learning exploration framework that decomposes epistemic uncertainty by sensory modality and explicitly weights tactile uncertainty to guide robot behavior toward contact-rich interactions.

2026-09-30 Evaluation / Agent Performance / False memory assessment

Beyond the Shadows of Plato's Cave: Evaluating False Memory in Autonomous Agents via Counterfactual Reasoning

This paper studies false memory in autonomous agents, defined as cases where memories formed from prior experience lead to inappropriate beliefs in new situations.

2026-09-29 Method / Video Generation / Generating diverse humanoid-object interaction videos

Counterfactual Video Generation Enables Scalable Humanoid Loco-Manipulation

This paper presents PRISM, a real-to-sim-to-real framework for teaching humanoid loco-manipulation from a small number of real human videos.

2026-09-29 Method / Language Modeling / System One model for Chinese inference

Chinese-Jev: Bringing System One Model to Chinese-Language Tasks

Chinese-Jev is a Chinese-language System One model designed for bounded decision tasks, including option selection, proposition judgment, and ordinal scoring.

2026-09-29 Method / Context Management / Zero-shot context management strategies

Context Language Models

This paper introduces Context Language Models (CLMs), which make context management a native model capability by representing the live context as an editable file that the model can modify directly rather than only extending it append-only.