BOLT: Bootstrap Long Chain-of-Thought in Language Models without
  Distillation

Bo Pang; Hanze Dong; Jiacheng Xu; Silvio Savarese; Yingbo Zhou; Caiming Xiong

返回文献探索

Paper · arXiv 2502.03860

BOLT: Bootstrap Long Chain-of-Thought in Language Models without Distillation

Bo Pang, Hanze Dong, Jiacheng Xu, Silvio Savarese, Yingbo Zhou, Caiming Xiong

25 upvotesFebruary 6, 2025arXiv 预印本

AI 摘要

A method is introduced to enable long chain-of-thought reasoning in large language models using in-context learning and supervised fine-tuning without distillation from pre-existing models or extensive human annotations.

large language modelslong chain-of-thoughtLongCoTin-context learningsupervised fine-tuningreasoning capabilitiesbootstrappingArena-HardMT-BenchWildBenchZebraLogicMATH500

Abstract

Large language models (LLMs), such as o1 from OpenAI, have demonstrated remarkable reasoning capabilities. o1 generates a long chain-of-thought (LongCoT) before answering a question. LongCoT allows LLMs to analyze problems, devise plans, reflect, and backtrack effectively. These actions empower LLM to solve complex problems. After the release of o1, many teams have attempted to replicate its LongCoT and reasoning capabilities. In terms of methods, they primarily rely on knowledge distillation with data from existing models with LongCoT capacities (e.g., OpenAI-o1, Qwen-QwQ, DeepSeek-R1-Preview), leaving significant uncertainties on systematically developing such reasoning abilities. In terms of data domains, these works focus narrowly on math while a few others include coding, limiting their generalizability. This paper introduces a novel approach to enable LLM's LongCoT capacity without distillation from o1-like models or expensive human annotations, where we bootstrap LongCoT (BOLT) from a standard instruct model. BOLT involves three stages: 1) LongCoT data bootstrapping with in-context learning on a standard instruct model; 2) LongCoT supervised finetuning; 3) online training to further refine LongCoT capacities. In BOLT, only a few in-context examples need to be constructed during the bootstrapping stage; in our experiments, we created 10 examples, demonstrating the feasibility of this approach. We use Llama-3.1-70B-Instruct to bootstrap LongCoT and apply our method to various model scales (7B, 8B, 70B). We achieve impressive performance on a variety of benchmarks, Arena-Hard, MT-Bench, WildBench, ZebraLogic, MATH500, which evaluate diverse task-solving and reasoning capabilities.