How Agents Learn to Manage Context and Credit

Featured

• ContextPilot: Teaching Agents for Proactive Context Management via Fine-grained RL score 12
  入选 HF Daily Papers; HF 热度: 12 upvotes (+3); 有代码实现; 关键词(3): compression, agentic, reasoning; 顶会接收: EMNLP
• Blind Men and the Elephant: Probing the Epistemic Myopia of LLMs under Long-Tail Divergent Knowledge score 8
  入选 HF Daily Papers; HF 热度: 8 upvotes (+2); 有代码实现; 关键词(2): scaling, reasoning
• Rubric-to-Code Credit Assignment for Reinforcement Learning score 5
  入选 HF Daily Papers; HF 热度: 2 upvotes (+1); 关键词(3): GRPO, code generation, leaderboard

Also Worth Noting

• Video Generative Models as Geometry Learner score 4
  入选 HF Daily Papers; 关键词(1): fine-tune
• PersonaEdit: Representative Sample Selection for Personalized Model Editing score 4
  关键词(1): scaling; 顶会接收: EMNLP
• The Illusion of $\textit{What If}$: Evaluating the Breakdown of Counterfactual Reasoning in LLMs score 4
  关键词(1): reasoning; 顶会接收: EMNLP
• Twin Worlds: Equivariance-Based Abstention for Evidence-Grounded Reasoning score 4
  关键词(1): reasoning; 顶会接收: EMNLP
• Ex-Sim(3)-Reg: 2D-3D Correspondence Pruning via Extended Sim(3) Registration score 4
  关键词(1): pruning; 顶会接收: ECCV
• RASA: Disentangled Spatial-Motional Priors for Cross-Identity Character Animation score 4
  关键词(1): scaling; 顶会接收: ECCV
• REINS: Refusal-Enhanced Inhibitory Steering with Sparse Autoencoder Features score 4
  关键词(1): lightweight; 顶会接收: EMNLP
• Finding Where the Buck Stops: An Automated Failure Attribution-Based Reflection Framework for Multi-Agent Collaboration score 4
  关键词(2): PPO, reasoning; 顶会接收: EMNLP
• CamoDocs: A Poisoning Attack Against Retrieval-Augmented Language Models Using Camouflaged Documents score 4
  关键词(2): retrieval-augmented, RAG; 顶会接收: EMNLP
• A Unified Framework to Elicit Structured Feedback for Interpretable Multi-Trait Essay Scoring score 4
  关键词(2): lightweight, reasoning; 顶会接收: EMNLP
• Fidelity Is Not Enough: Dispatch-Level Instrumentation for Agentic Datasheet Extraction score 4
  关键词(3): serving, agentic, tool use; 顶会接收: EMNLP
• DARTS: Decoder-Aware Representation Tuning via Surgery for Model Merging score 4
  关键词(3): lightweight, code generation, reasoning; 顶会接收: EMNLP
• PULSAR: Pooled Unified Late-Interaction Search and Retrieval for Enterprise Visual Document RAG score 4
  关键词(3): production, latency, RAG; 顶会接收: EMNLP
• Is Prosody Lost in Translation? Fine-Grained Cross-Lingual Prosody Similarity Across Languages score 3
  顶会接收: EMNLP
• Beyond Global Scalars: Synergizing Token-Level Statistics and Deep Semantics for Adversarial AIGC Text Detection score 3
  顶会接收: EMNLP