A Stateful 260M Embedding Beats 8B Specialists

Featured

• EvoEmbedding: Evolvable Representations for Long-Context Retrieval and Agentic Memory score 9
  入选 HF Daily Papers; HF 热度: 10 upvotes (+3); 有代码实现; 关键词(2): RAG, agentic
• CalVerT: Augmenting Agents with Calibrated Verifier Telemetry Improves Action and Learning in Knowledge-Intensive Tasks score 6
  入选 HF Daily Papers; 有代码实现; 关键词(2): state space, reasoning

Also Worth Noting

• Structural Assessment for Understanding and Guiding Dataset Distillation in Discrete Token Space score 4
  关键词(1): distillation; 顶会接收: ECCV
• Adversarial Domain Prompt Tuning and Generation for Single Domain Generalization score 4
  关键词(1): text-to-image; 顶会接收: CVPR
• Scene-Level Heterogeneous Physics Simulation with 3D Gaussian Splats score 4
  关键词(1): production; 顶会接收: CVPR
• NASDAQ: Normalized Observation Space Dynamics-Augmented Q-Learning score 3
  机构: EPFL
• Factual Retrieval in LLMs Is a Redundant, Distributed and Non-Contiguous Process score 3
  顶会接收: ACL
• FLM-Occ: Feed-forward Likelihood Maximization for Efficient Indoor Occupancy Prediction score 3
  顶会接收: ECCV
• Unsupervised Disentanglement Without Compromises : How Functional Orthogonality Enforces Identifiability score 3
  顶会接收: ICML
• The Two-Hump Problem: Bridging the Difficulty Gap in Mathematical Reinforcement Learning score 3
  顶会接收: ICML