AI论文简报
搜索
方法论
公众号
EN
原子接口相较原生SDK让桌面Agent在线token约少73%
从200篇论文中选出14篇
重点关注
Data-Centric Recursive Self-Improvement for Navigation via Policy-Synchronized Language-Action Dual Bootstrapping
score 10
入选 HF Daily Papers;HF 热度: 25 upvotes (+4);有代码实现;关键词(3): fine-tuning, reasoning, vision-language
A Decade in a Day: Synthesizing Agent-Native Interfaces from Human-Oriented Software Evolution
score 9
入选 HF Daily Papers;HF 热度: 25 upvotes (+4);有代码实现
VIEWCRAFT: Evolving What Multimodal Agents See and How They Read It
score 9
入选 HF Daily Papers;HF 热度: 22 upvotes (+4);有代码实现
也值得关注
SARATHI: Decoupled Data-Free Probing and Adaptive Reconstruction for Structured LLM Pruning
score 4
关键词(2): pruning, post-training;顶会接收: EMNLP
Apertus 1.5: From Open Text Models to Open Multimodal Reasoners
score 4
机构: ETH Zurich;关键词(3): finetuning, pretraining, reasoning
Sparse All-Layer Connector for Domain Generalised Semantic Segmentation
score 4
关键词(1): vision-language;顶会接收: NeurIPS
Towards Autopilot Labeling for Short Video Content Governance
score 4
关键词(1): agentic;顶会接收: EMNLP
SpecForge: Recursive Self-Improvement for Coding Agents via Behavioral Specifications
score 4
关键词(1): coding;顶会接收: NeurIPS
What DNA Foundation Models Learn Beyond Sequence Composition
score 4
关键词(1): fine-tuning;顶会接收: NeurIPS
Personalized LLM Recommendation with Semantic IDs: Autoregressive Beam Search vs. Multi-Token Prediction
score 4
关键词(2): latency, throughput;顶会接收: NeurIPS
When Trackers Fail: VLM-Guided Verification and Recovery for Robust Video Object Segmentation
score 4
关键词(2): reasoning, vision-language;顶会接收: NeurIPS
An Interpretable Linguistically-Grounded Analysis of Scene Saliency in Movie Screenplays
score 3
顶会接收: EMNLP
Rethinking Human Intent-to-CAD: Parametric CAD Model Generation via Cooperative Multi-Task Alignment and Spatial-Aware Reinforcement Learning
score 3
顶会接收: ICML
Mesh-Free Convolution: Learned Spectral Attenuation and Transport
score 3
顶会接收: NeurIPS