AI Research Brief
Search
Methodology
中文
Teacher Policy Hits 93.57; 3,000 Samples Recover 37.3%
10 selected from 195 papers
Featured
DriveZero: End-to-End Driving Beyond Human Demonstrations
score 9
入选 HF Daily Papers; HF 热度: 11 upvotes (+3); 有代码实现; 关键词(1): PPO
Also Worth Noting
Accelerating Diffusion Transformers with Gaussian Process Rectified Feature Cache
score 4
关键词(1): real-time; 顶会接收: ECCV
ACE: Adapter Consolidation across Experts for Parameter-Efficient Fine-Tuning of MoE LLMs
score 4
关键词(2): fine-tuning, MoE; 顶会接收: EMNLP
Beyond Retraining-Free MoE Compression: A Cost-Normalized Study of Post-Compression Adjustment
score 4
关键词(5): compression, pruning, deployment, fine-tuning, MoE; 顶会接收: EMNLP
PASTEL: Panoramic Alignment for Monocular 4D Scene Reconstruction
score 4
关键词(1): embodied; 顶会接收: ECCV
SAP: State-Guided Data Synthesis with Argument Provenance for Multi-Turn Tool Use
score 4
关键词(2): agentic, tool use; 顶会接收: EMNLP
From Gaze to Meaning: A Training-Free AI Agent for Unified Grounding and Explanation
score 4
关键词(2): reasoning, vision-language; 顶会接收: ECCV
Steering Geometry: Validating Human Value Geometry in LLM Steering Space
score 4
关键词(5): lightweight, fine-tuning, RLHF, DPO, instruction tuning; 顶会接收: EMNLP
The Blindness of Document-Level Translation Evaluation
score 3
顶会接收: EMNLP
Correction as Annotation: Bootstrapping a Dependency Parser for Documentary Medieval Latin
score 3
机构: Harvard