MOOSE-Star：突破複雜度壁壘，開啟科學發現的可控訓練新紀元

MOOSE-Star: Unlocking Tractable Training for Scientific Discovery by Breaking the Complexity Barrier

March 4, 2026

作者: Zonglin Yang, Lidong Bing

cs.AI

摘要

雖然大型語言模型在科學發現領域展現潛力，現有研究卻多聚焦於推論或反饋驅動的訓練，尚未探索生成式推理過程的直接建模——即假設基於背景的條件機率P(h|b)。我們證明由於從龐大知識庫中檢索與組合靈感存在組合複雜度（O(N^k)）的固有難題，直接訓練P(h|b)在數學上是不可行的。為突破此障礙，我們提出MOOSE-Star統一框架，實現可追蹤的訓練與可擴展的推論。在最理想情況下，MOOSE-Star通過三項創新將複雜度從指數級降至對數級（O(log N)）：(1) 基於發現機率方程分解出的子任務進行訓練；(2) 採用動機引導的階層式搜索實現對數級檢索並剪除無關子空間；(3) 運用有界組合技術提升對檢索噪聲的魯棒性。為支持此框架，我們發布TOMATO-Star資料集，包含108,717篇經分解的論文（耗費38,400 GPU小時）用於訓練。此外，我們發現當暴力取樣法遭遇「複雜度牆」時，MOOSE-Star能展現持續的測試時擴展能力。

English

While large language models (LLMs) show promise in scientific discovery, existing research focuses on inference or feedback-driven training, leaving the direct modeling of the generative reasoning process, P(hypothesis|background) (P(h|b)), unexplored. We demonstrate that directly training P(h|b) is mathematically intractable due to the combinatorial complexity (O(N^k)) inherent in retrieving and composing inspirations from a vast knowledge base. To break this barrier, we introduce MOOSE-Star, a unified framework enabling tractable training and scalable inference. In the best case, MOOSE-Star reduces complexity from exponential to logarithmic (O(log N)) by (1) training on decomposed subtasks derived from the probabilistic equation of discovery, (2) employing motivation-guided hierarchical search to enable logarithmic retrieval and prune irrelevant subspaces, and (3) utilizing bounded composition for robustness against retrieval noise. To facilitate this, we release TOMATO-Star, a dataset of 108,717 decomposed papers (38,400 GPU hours) for training. Furthermore, we show that while brute-force sampling hits a ''complexity wall,'' MOOSE-Star exhibits continuous test-time scaling.