HuggingFace
PhysEvo: Astra Can Act, Let It
Astra can act, yet reliable manipulation depends on the system through which it observes and controls the world. We i...
🤖 AI 前沿资讯
每天自动聚合 AI 领域最新动态
HuggingFace
Internalizing Agent Experience into Diffusion Model Weights via On-Policy Context Distillation
Wrapping an image generation model in an agentic harness can effectively boost Text-to-Image task performance: the ha...
HuggingFace
Gains and Collapse in On-Policy Distillation:A Reinforcement Learning Perspective
On-policy distillation (OPD) has become an important approach to language model post-training. However, despite its p...
HuggingFace
From Pareto to Preference: Personalized Test-Time Scaling via Amortized Agentic Policy Discovery
Test-time scaling (TTS) improves the reasoning capabilities of large language models by allocating additional inferen...
HuggingFace
EngramEdit: Decoupled Knowledge Updates in LLMs through Conditional Memory
Conditional memory architectures such as DeepSeek Engram use input n-grams to look up learned embeddings, expanding t...
HuggingFace
StepCAD: Mesh-to-CAD Code Generation via LLM Policy and Geometry-Guided Search
Recovering executable CAD programs from 3D meshes is challenging due to the compositional nature of CAD construction ...
HuggingFace
Task-Sufficient Contraction: Source Selection for Machine Information Interfaces
A declared task can sometimes certify a reduced source before a downstream encoder, codebook, rate, distortion target...
HuggingFace
Salt++: Context-Aligned Post-Training for Few-Step Streaming Multimodal Generation
Few-step streaming audio--video generation requires both causal modeling and step distillation, yet standard training...
HuggingFace
CADFather: Autonomous CAD Reconstruction through Coordinated Tool Use
Reconstructing an editable CAD model from a 3D shape remains a challenging engineering task. Existing methods can pro...
HuggingFace
UniSkill: Learning Actor-Aligned Skill Proposals for an Evolving Policy
Large language model agents can improve across tasks by retaining reusable skills distilled from prior interactions. ...
HuggingFace
SheetSage2: Coherent Lead-Sheet Transcription with Synthetic Supervision
Transcribing music into a human-readable score requires a coherent understanding of rhythm, harmony, melody, and form...
HuggingFace
FastOPD: On-Policy Distillation for Lightweight VLA Deployment
Vision-Language-Action (VLA) foundation models have scaled rapidly to enhance manipulation performance and generaliza...