Hugging Face Daily Papers
500 articles archived · Visit source ↗ · RSS
-
Hugging Face Daily Papers research 1mo ago
Agentic Visual Generation: From Generative Models to Agentic Control
Abstract The abstract proposes a four-level framework classifying agentic visual generation by the controller's direct decision scope over generation operations, from fixed support to experience-adaptive control. Generated by thinkingmachines/Inkling-Small Visual generation is…
16 -
SceneMosaic: Efficient and Diverse Simulation-Ready Scene Generation via Hybrid Agentic Layout Evolution
Abstract SceneMosaic combines learned image priors with vision-language agents to efficiently generate diverse, physically valid indoor scenes by evolving local units and composing them globally. Generated by thinkingmachines/Inkling-Small Diverse and simulation-ready indoor…
36 -
VidaForge: Open Research Infrastructure for Video Pretraining Data Recipes
Abstract VIDAFORGE is an open infrastructure that models video data recipes as executable workflows to study how data choices affect pretraining of video foundation models. Generated by thinkingmachines/Inkling-Small Video foundation models increasingly rely on large-scale…
35 -
What LLM Trading Agents Actually Do in Production: A Six-Month, Population-Scale Record from Two Fleets
Abstract Autonomous language-model trading agents across production systems show behavior driven by interface design rather than strategy, exhibit volatility-blind sizing, fail to capture favorable price excursions, and display no directional edge, with frontier model decision…
14 -
NeoHorse-1: Towards Recursive Self-Improvement via Agentic Post-Training with Routing Harness
Abstract NeoHorse-1 uses agentic post-training with intelligent routing, structured feedback loops, and curriculum-based distillation to improve model capabilities across agent benchmarks. Generated by thinkingmachines/Inkling-Small Recursive self-improvement (RSI) requires a…
33 -
MOLE: Detecting Insider Threats in AI Agents
Abstract MOLE is a benchmark for evaluating defenses that detect harmful actions by AI agents operating shared services under limited review budgets. Generated by thinkingmachines/Inkling-Small Model misalignment, prompt injection, or operator misuse could lead AI agents…
33 -
DriveZero: End-to-End Driving Beyond Human Demonstrations
Abstract DriveZero is an end-to-end autonomous driving system that combines a vision foundation model for perception with a closed-loop reinforcement learning action model to learn driving behaviors beyond human demonstrations. Generated by thinkingmachines/Inkling-Small Most…
8 -
ENEAS: Embedding-guided Neural Ensemble for Adaptive Segmentation
Abstract ENEAS unifies text-prompted instance tracking and open-concept semantic discovery via temporal memory extension and a verification layer that filters visual distractors. Generated by thinkingmachines/Inkling-Small We present ENEAS, a unified, text-promptable method for…
30 -
One Symptom, Three Levers: A Critical Review of On-Policy Self-Distillation
Abstract On-policy self-distillation for mathematical reasoning uses privileged information to guide a model without a larger teacher, but suffers from reasoning collapse governed by token weighting, privileged context, and guidance dynamics. Generated by…
19 -
EmbodiedSkills: A Unified Framework for Orchestrating, Training, and Deploying VLA Agents
Abstract EmbodiedSkills proposes a unified framework that validates and verifies robot skill executions through a fixed interface, enabling closed-loop embodied agents with adaptable low-level vision-language-action policies. Generated by thinkingmachines/Inkling-Small…
8 -
FlowBalance: Verifier-Grounded Self-Improvement from On-Policy Reasoning Experience
Abstract FlowBalance improves reasoning models via verifier-calibrated self-guidance using trajectory-level score reweighting and profile-based trajectory balance. Generated by thinkingmachines/Inkling-Small A reasoning model can improve from its own on-policy experience, but…
24 -
What Else Needs Fixing? Exploring Cost-Effective Test-Time Compute for Revision Propagation in Artifacts Generated Through Conversation
Abstract Large Language Models (LLMs) often help users generate artifacts through iterative cycles of generation and revision in conversation. A challenge here is that, when users specify only a local change during revision, LLMs must instead identify the relevant dependencies…
36 -
Verify Before You Distill: Prompt-Level Teacher Gating for On-Policy Distillation
Abstract Teacher-Gated On-Policy Distillation verifies teacher reliability per prompt via verifier-scored probes, routing to dense distillation or verifier-grounded reinforcement learning to improve training efficiency and accuracy. Generated by thinkingmachines/Inkling-Small…
34 -
Bilevel Coordinated Reflection: A Game-Theoretic Approach to Multi-Agent LLM Systems
Abstract The study formalizes multi-agent LLM coordination via bilevel games and stochastic memory reflection, introducing a grounded evaluation gate and SRMA algorithm with convergence guarantees, validated on SWE-bench. Generated by thinkingmachines/Inkling-Small Multi-agent…
4 -
The Attention Triangle in Audio-Video Models
Abstract Audio-video diffusion models exhibit bidirectional semantic leakage through cross-modal attention pathways, which can be diagnosed via attention-derived signals and mitigated through inference-time alignment interventions. Generated by thinkingmachines/Inkling-Small…
23 -
To See a World in a Living Context: Unified Indoor-Outdoor Urban World Generation
Abstract HoloWorld unifies indoor and outdoor 3D urban generation through a continuously updated cross-scale context that ensures coherent building-level correspondence and cross-block continuity. Generated by thinkingmachines/Inkling-Small Text-driven 3D generation has advanced…
9 -
τ^τ-Bench: An Environment for End-To-End, Realistic Agent Construction
Abstract The τ^τ-bench benchmark evaluates coding agents on building real-world customer-service agents from business records, client requirements, and production APIs, revealing substantial gaps versus expert performance. Generated by thinkingmachines/Inkling-Small LLM agents…
10 -
RISE: Recursive Improvement via Self-Extrapolating Policy Distillation
Abstract RISE improves language model post-training by recursively generating dense token-level supervision from the model's own reinforcement learning trajectory via self-extrapolation, avoiding external teachers. Generated by thinkingmachines/Inkling-Small On-policy…
9 -
WorldSculpt: Generating Compositional Worlds from Grounded Videos
Abstract Adapting a single-object 3D generative prior to multi-view observations enables scalable compositional mesh reconstruction of densely cluttered scenes with severe occlusion. Generated by thinkingmachines/Inkling-Small We study the problem of generating a compositional…
16 -
Don't Drop Dropout: Optimizing Layer Sparsity for Efficient LLM Training and Inference
Abstract Layer dropout improves large language model training efficiency and enables faster inference via early exit and speculative decoding without sacrificing accuracy. Generated by thinkingmachines/Inkling-Small Layer dropout (a.k.a. stochastic depth) has been shown to…
25 -
Enoki: Efficient Multi-Level Hallucination Detection
Abstract Enoki is an open information extraction framework that unifies claim-level verification and span-level hallucination localization through shared relational facts, reducing resource use while improving detection accuracy. Generated by thinkingmachines/Inkling-Small…
23 -
ShallowStream: Index Shallow then Answer Deep for Streaming Video Understanding
Abstract ShallowStream uses shallow MLLM layers to build a lightweight streaming index, reducing latency while maintaining accurate video retrieval. Generated by thinkingmachines/Inkling-Small Streaming video understanding is a critical capability for real-world applications,…
31 -
AdaptVPR: Route-Aware Hard Positive Generation for Robust Visual Place Recognition
Abstract AdaptVPR improves visual place recognition by generating verified synthetic same-place images with diverse appearance changes to train more robust models. Generated by thinkingmachines/Inkling-Small Visual Place Recognition (VPR) localizes a query image by retrieving…
24 -
MaxKernel: Agentic Kernel Generation for TPUs
Abstract MaxKernel is a multi-agent system that automates TPU kernel development through collaborative, autonomous, and graph-based search paradigms, achieving expert-level performance on diverse benchmarks. Generated by thinkingmachines/Inkling-Small Designing and authoring…
38 -
Ask Before You Optimize: Dynamic Pre-Formulation Clarification for Interactive Optimization
Abstract OR-Clarify benchmarks clarification before optimization modeling, and InterOPT guides agents to ask questions or stop based on missing formulation-critical details. Generated by thinkingmachines/Inkling-Small Large language models (LLMs) are increasingly used to…
12 -
Motion-Omni: End-to-End Joint Speech and Full-Body Motion for Spoken Dialogue
Abstract Motion-Omni is an end-to-end framework that jointly generates spoken dialogue and full-body co-speech motion from shared hidden states, using scalable pseudo-labeling and a unified evaluation protocol to achieve real-time, aligned responses. Generated by…
28 -
When Models Edit Too Much: On the Fidelity of Minimal Code Edits
Abstract Large language models frequently over-edit code during repair, but preservation instructions and reinforcement learning can improve edit fidelity without sacrificing correctness. Generated by thinkingmachines/Inkling-Small Large language models (LLMs) are increasingly…
27 -
When Quantization Breaks Memory: Recurrent-State Write-Back in Low-Precision Temporal Inference
Abstract Quantized recurrent inference suffers from state write-back rules that suppress small updates, but error feedback and residual memory restore accuracy without retraining across GRU and LSTM architectures. Generated by thinkingmachines/Inkling-Small Quantization is…
35 -
Group Adaptive Clipping Policy Optimization
Abstract GAPO adaptively adjusts importance-sampling clipping thresholds based on rollout advantage to preserve stronger gradient signals from low-success groups in reinforcement learning with verifiable rewards. Generated by thinkingmachines/Inkling-Small Group relative policy…
27 -
Iris: Climbing to the Search Frontier
Abstract Two large-scale search agents are trained via a multi-stage pipeline combining supervised fine-tuning and reinforcement learning against live search, achieving state-of-the-art open-source results on complex web benchmarks through rigorous trajectory filtering and…
15 -
RoboTok: An Internet-Scale Data Engine for Human Demonstration Retrieval and Dexterous Manipulation Learning
Abstract RoboTok retrieves relevant human manipulation videos from the web using a latent motion space derived from 3D hand trajectories to improve robot policy training. Generated by thinkingmachines/Inkling-Small Robot learning increasingly depends on broad and diverse…
6 -
A Common Measure of Communication for Speech Brain-Computer Interfaces
Abstract Open-vocabulary mutual information provides a unified metric to compare speech brain-computer interfaces across different vocabularies and conditions, revealing trade-offs between vocabulary coverage and decoding accuracy. Generated by thinkingmachines/Inkling-Small…
18 -
VeriPhy: Agentic Physical Reasoning for World Model Evaluation and Refinement
Abstract VeriPhy verifies generated video by compiling prompts into typed physical obligations, executing frozen expert analyses with provenance tracking, and mapping evidence to auditable three-valued verdicts. Generated by thinkingmachines/Inkling-Small Visual fluency in…
12 -
Locked at the Entrance, Open Inside: Where RLVR Narrows the Solution Space
Abstract Reinforcement learning with verifiable rewards narrows reasoning diversity primarily at the initial solution step rather than during execution, and targeted interventions can restore coverage without sacrificing accuracy. Generated by thinkingmachines/Inkling-Small…
23 -
Unlocking Lossless Speedups in LLMs via Discrete Diffusion
Abstract Diffusion-augmented autoregressive language models use parallel token sampling via distilled diffusion weights and a specialized sampler to accelerate inference without quality loss or draft models. Generated by thinkingmachines/Inkling-Small Large Language Models…
6 -
DRACO: Fine-Grained Credit Assignment with Dynamic Rubrics for Long-Horizon Agent Training
Abstract DRACO dynamically generates rubrics and redistributes trajectory-level scores into per-step advantages for reinforcement learning without verifiers, improving long-horizon agent performance. Generated by thinkingmachines/Inkling-Small Reinforcement Learning from…
24 -
Last Translation Benchmark
Abstract The Last Translation Benchmark introduces peer-reviewed, multimodal examples that break leading translation models alongside handcrafted verification rules for reliable, actionable evaluation. Generated by thinkingmachines/Inkling-Small For scientific progress, we need…
22 -
Using Grounded Theory for Agent Behavior Analysis at Scale
Abstract AutoTraceGT automates grounded theory coding on agent trajectories to build task-specific behavioral taxonomies that recover and extend human failure-mode annotations for downstream prediction. Generated by thinkingmachines/Inkling-Small Understanding agent behavior…
20 -
Environment Evolution for Terminal Agents
Abstract Environment evolution incrementally raises task difficulty off-policy to sustain continuous learning signals for terminal agents, improving benchmark performance through multi-agent harnesses. Generated by thinkingmachines/Inkling-Small Scaling interactive and…
9 -
Principia: Relational Physics Tests for Video Models
Abstract Principia evaluates video generators on Newtonian physics via calibration-independent relational consistency across paired objects, revealing major physical reasoning gaps. Generated by thinkingmachines/Inkling-Small Evaluating physical reasoning in video models is…
37 -
Terminal-Universe: Turning Agent Trajectories into Scalable Terminal Environments
Abstract Terminal-Universe reconstructs executable workspaces from agent trajectories to synthesize diverse training tasks and improves post-training performance through supervised fine-tuning. Generated by thinkingmachines/Inkling-Small As terminal-based code agents become…
9 -
Scal3R: Learning Efficient Multi-Relative Pose Query for Scalable Online 3D Reconstruction
Abstract Scal3R improves long-video online 3D reconstruction by querying multi-reference relative poses with lightweight tokens and pose-graph optimization, reducing drift without retraining the backbone. Generated by thinkingmachines/Inkling-Small Online 3D reconstruction…
16 -
QCell: Recombining and Aligning Cell Queries for Overlapping Instance Segmentation
Abstract QCell is a query-based model that improves instance segmentation of overlapping microscopy cells through latent-space recombination and contrastive query alignment. Generated by thinkingmachines/Inkling-Small Instance segmentation of overlapping cells in microscopy…
31 -
Let Confidence Change, Not the Prediction: Prediction-Preserving Repair for Post-hoc Calibration
Abstract CORD is a post-fit adapter that repairs calibrated probability vectors to exactly preserve original top-1 predictions while maintaining calibration quality. Generated by thinkingmachines/Inkling-Small Post-hoc calibration corrects reported confidence, yet a multiclass…
14 -
Compile by Training: Turning Natural-Language Specifications into Local Neural Functions
Abstract Compile by training converts natural-language specifications into reusable neural functions by distilling teacher-generated examples into small adapters, enabling efficient deployment without remote model dependencies. Generated by thinkingmachines/Inkling-Small Many…
21 -
Percolation Dynamics in Optimization : Variance Cascades and Discrete Scale Invariance
Abstract Stochastic gradient descent dynamics are modeled as a percolation process where architectural symmetries cause subnetworks to merge in discrete blocks, producing variance spikes resembling phase transitions, with similar trapping behavior extending to Adam and AdamW…
29 -
PACE: Towards Surfacing Hidden Conflicts in User Requests
Abstract PaceMaker uses coordinated agents to retrieve implicit contextual evidence and evaluate whether personalized requests conflict with hidden user constraints. Generated by thinkingmachines/Inkling-Small Personalized assistants should not only comply with user requests but…
17 -
Knowing When Not to Reuse: Conditional Experience Transfer in Autonomous LLM Post-Training
Abstract Boundary-Calibrated Intervention Transfer selectively reuses past training evidence by checking contextual applicability and running bounded trials, reducing harmful updates and improving final model quality in autonomous post-training. Generated by…
25 -
Select, Compress, Reinvest: A Controlled Study of Visual-Token Allocation in Long-Video MLLMs
Abstract Frame selection dominates accuracy in long-video language models, while spatial compression is nearly free if savings are reinvested into more frames, and a unified harness is needed to compare selectors fairly. Generated by thinkingmachines/Inkling-Small Long-video…
20 -
Puffin-World: Scaling a Unified Multimodal Model with Native 3D World States
Abstract Puffin-World is a unified multimodal framework that jointly models physics, geometry, and appearance for physically consistent 3D world generation, reconstruction, and closed-loop exploration. Generated by thinkingmachines/Inkling-Small We propose Puffin-World, a…
15