Authors: || Published: 2026-01-25T21:17:00 || Updated: 2026-01-25T21:17:00 || 5 min read
Categories: || Tags: || Post-format: link
Recent AI Reading [25 January 2026]
Papers
Agentic AI
- DeepAgent: A General Reasoning Agent with Scalable Toolsets
- A Survey of Data Agents: Emerging Paradigm or Overstated Hype?
- Agent Data Protocol: Unifying Datasets for Diverse, Effective Fine-tuning of LLM Agents
- AgentFold: Long-Horizon Web Agents with Proactive Context Management
- AgentFrontier: Expanding the Capability Frontier of LLM Agents with ZPD-Guided Data Synthesis
- InteractComp: Evaluating Search Agents With Ambiguous Queries
- Fundamentals of Building Autonomous LLM Agents
- The Tool Decathlon: Benchmarking Language Agents for Diverse, Realistic, and Long-Horizon Task Execution
- The Era of Agentic Organization: Learning to Organize with Language Models
- Unlocking the Power of Multi-Agent LLM for Reasoning: From Lazy Agents to Deliberation
- GUI-360: A Comprehensive Dataset and Benchmark for Computer-Using Agents
- Training Proactive and Personalized LLM Agents
- Cross Topics: AI Alignment
- DeepEyesV2: Toward Agentic Multimodal Model
- Real-Time Reasoning Agents in Evolving Environments
- HAFixAgent: History-Aware Automated Program Repair Agent
- ResearchRubrics: A Benchmark of Prompts and Rubrics For Evaluating Deep Research Agents
- AgentEvolver: Towards Efficient Self-Evolving Agent System
- Cross Topics: Training Paradigms
- ReCAP: Recursive Context-Aware Reasoning and Planning for Large Language Model Agents
- Solving a Million-Step LLM Task with Zero Errors
- Agent-R1: Training Powerful LLM Agents with End-to-End Reinforcement Learning
- Cross Topics: Training Paradigms
- General Agentic Memory Via Deep Research
- Cross Topics: AI Memory
- Agentic Learner with Grow-and-Refine Multimodal Semantic Memory
- Cross Topics: AI Memory
- Evo-Memory: Benchmarking LLM Agent Test-time Learning with Self-Evolving Memory
- Cross Topics: AI Memory
- Latent Collaboration in Multi-Agent Systems
- Adaptation of Agentic AI
- An Empirical Study of Agent Developer Practices in AI Agent Frameworks
- Sophia: A Persistent Agent Framework of Artificial Life
- Agent Drift: Quantifying Behavioral Degradation in Multi-Agent LLM Systems Over Extended Interactions
- Dr. Zero: Self-Evolving Search Agents without Training Data
- Cross Topics: Training Paradigms
- Agentar-Scale-SQL: Advancing Text-to-SQL through Orchestrated Test-Time Scaling
- AI Agents Need Memory Control Over More Context
- Cross Topics: AI Memory
- Toward Efficient Agents: Memory, Tool learning, and Planning
- Cross Topics: AI Memory
- Agentic Memory: Learning Unified Long-Term and Short-Term Memory Management for Large Language Model Agents
- Cross Topics: AI Memory, Training Paradigms
- Agentic-R: Learning to Retrieve for Agentic Search
- Cross Topics: RAG
- Task-Decoupled Planning for Long-Horizon Agents
- SimpleMem: Efficient Memory Framework for Agentic Systems
- Cross Topics: AI Memory
- MemRL: Reinforcement Learning for Memory-based Language Model Agents
- Cross Topics: AI Memory, Training Paradigms
- Agentic Reasoning for Large Language Models
- Parallelism Meets Adaptiveness: Scalable Documents Understanding in Multi-Agent LLM Systems
- Agentic AI: a comprehensive survey of architectures, applications, and future directions
- scaleapi/scale-agentex: Open source codebase for Scale Agentex
- pat-jj/Awesome-Adaptation-of-Agentic-AI: Repo for “Adaptation of Agentic AI”
- RL Environments and the Hierarchy of Agentic Capabilities
AI Alignment (with Human Preferences, and other methods)
Large Language Models
- Continuous Autoregressive Language Models
- Personality Traits in Large Language Models
- Titans: Learning to Memorize at Test Time
- Cross Topics: AI Memory
- Recursive Language Models
- MEM1: Efficient Long-Horizon Reasoning with Constant Memory
- Cross Topics: AI Memory, Agentic AI
- A Plan Reuse Mechanism for LLM-Driven Agent
- Cross Topics: Agentic AI
Training Paradigms
- Reasoning-Aware GRPO using Process Mining
- Supervised Reinforcement Learning: From Expert Trajectories to Step-wise Reasoning
- MIRO: MultI-Reward cOnditioned pretraining improves T2I quality and efficiency
- Dual-Weighted Reinforcement Learning for Generative Preference Modeling
- CritiCal: Can Critique Help LLM Uncertainty or Confidence Calibration?
- The Ultimate Guide to Fine-Tuning LLMs from Basics to Breakthroughs: An Exhaustive Review of Technologies, Research, Best Practices, Applied Research Challenges and Opportunities
- LADDER: Self-Improving Language Models through Recursive Problem Decomposition
- Probabilistic Artificial Intelligence
Model Evaluation
Retrieval-Augmented Generation
- REFRAG: Rethinking RAG based Decoding
- LLM as HPC Expert: Extending RAG Architecture for HPC Data
- Retrieval-Augmented Generation for Large Language Models: A Survey
- A Systematic Review of Retrieval-Augmented Generation for Enhancing Domain-Specific Knowledge in Large Language Models
- Retrieval Augmented Generation (RAG) and Large Language Models (LLMs) for Enterprise Knowledge Management and Document Automation: A Systematic Literature Review
- HtmlRAG: HTML is Better Than Plain Text for Modeling Retrieved Knowledge in RAG Systems
- UniversalRAG: Retrieval-Augmented Generation over Corpora of Diverse Modalities and Granularities
- PISCO: Pretty Simple Compression for Retrieval-Augmented Generation
- Improving Multi-step RAG with Hypergraph-based Memory for Long-Context Complex Relational Modeling
- Cross Topics: AI Memory
- CLaRa: Bridging Retrieval and Generation with Continuous Latent Reasoning
- VectifyAI/PageIndex: Vectorless, reasoning-based RAG system
Synthetic Data
- Matrix: Peer-to-Peer Multi-Agent Synthetic Data Generation Framework
- Cross Topics: Agentic AI
Embodied AI
AI Memory
- Beyond a Million Tokens: Benchmarking and Enhancing Long-Term Memory in LLMs
- Context Engineering 2.0: The Context of Context Engineering
- Everything is Context: Agentic File System Abstraction for Context Engineering
- Memory in the Age of AI Agents
- MemEvolve: Meta-Evolution of Agent Memory Systems
- InfiniteICL: Breaking the Limit of Context Window Size via Long Short-term Memory Transformation
- MemGPT: Towards LLMs as Operating Systems
- Zep: A Temporal Knowledge Graph Architecture for Agent Memory
- MemoryBench: A Benchmark for Memory and Continual Learning in LLM Systems
- getzep/graphiti: Build Real-Time Knowledge Graphs for AI Agents
- Key-value memory in the brain
Miscellaneous
- Real Deep Research for AI, Robotics and Beyond
- Benchmarking World-Model Learning
- DeepSeek-OCR: Contexts Optical Compression
- PAN: A World Model for General, Interactable, and Long-Horizon World Simulation
- Hierarchical Reasoning Model
- A Practical Guide for Designing, Developing, and Deploying Production-Grade Agentic AI Workflows
- deepseek-ai/DeepSeek-Math-V2
- ISEEQ: Information Seeking Question Generation using Dynamic Meta-Information Retrieval and Knowledge Graphs
Books
Articles and Blog Posts
- Composer: Building a fast frontier model with RL
- Claude Skills blog by Simon Willison
- Effective context engineering for AI agents - Anthropic
- The 2026 State of AI Agents Report - Anthropic
- The state of enterprise AI - OpenAI
- 2025: The State of Generative AI in the Enterprise - Menlo Ventures
- The 15 AI Ideas for 2026, According to a16z Partners
- Beyond Naive RAG: Practical Advanced Methods
- Context Rot: How Increasing Input Tokens Impacts LLM Performance - Chroma Research
- Context Engineering: Sessions & Memory
- Why Enterprises Need Specialized RL Agents - Scale
- Securing Agents in Production: Agentic Runtime - Palantir Blog
- Large Language Models Will Never Be Intelligent, Expert Says
- LLMs Can Get “Brain Rot”!
- CLaRa: Apple’s New RAG Framework Achieves 16x-128x Compression While Outperforming Traditional Retrieval-Augmented Generation Systems
- Overview - Agent Skills