@HuggingPapers: Microsoft Research introduces Arbor A generalist autonomous research agent that uses persistent hypothesis-tree refinem…
Summary
Microsoft Research introduces Arbor, a generalist autonomous research agent that uses persistent hypothesis-tree refinement for cumulative learning, outperforming Codex and Claude Code across six research tasks and achieving 86% Any-Medal on MLE-Bench Lite.
View Cached Full Text
Cached at: 06/11/26, 05:39 PM
Microsoft Research introduces Arbor
A generalist autonomous research agent that uses persistent hypothesis-tree refinement to turn long-horizon exploration into cumulative learning. It beats Codex and Claude Code across 6 research tasks and hits 86% Any-Medal on MLE-Bench Lite. https://t.co/A1fxTxnskk
Similar Articles
Toward Generalist Autonomous Research via Hypothesis-Tree Refinement
Arbor is an AI framework for autonomous scientific research that uses a coordinator, executors, and a persistent hypothesis tree to iteratively improve research outcomes across multiple domains, achieving strong results on six real research tasks.
Arbor: Tree Search as a Cognition Layer for Autonomous Agents
Arbor introduces structured tree search as a cognition layer for autonomous agents, enabling multi-day, full-stack LLM inference optimization with up to 193% throughput-latency improvement over vendor baselines through a checks-and-balances multi-agent architecture.
@_akhaliq: paper:
A paper introducing Arbor, an AI framework that enables autonomous scientific research by combining strategic coordination, isolated hypothesis testing, and a persistent knowledge tree to iteratively improve research outcomes across multiple domains.
@_akhaliq: Toward Generalist Autonomous Research via Hypothesis-Tree Refinement
This paper proposes a method for autonomous research agents using hypothesis-tree refinement to generate and test hypotheses, aiming toward generalist scientific discovery.
I reimplemented Arbor (a research agent that grows a tree of hypotheses) on LangGraph — an agent that keeps its experiments instead of forgetting what failed
A developer reimplements Arbor, a multi-agent research system from a paper, on LangGraph, sharing lessons about graph structure enforcing invariants and backpropagating insights from failures. The reimplementation, arbor-lg, uses LangGraph's StateGraph and SQLite checkpointer for durability.