RATIONALYST: Mining Implicit Rationales for Process Supervision of Reasoning

TL;DR

RATIONALYST is a model trained on a large collection of implicit rationales from unlabeled data, enabling improved reasoning accuracy across diverse tasks, outperforming larger models and verifiers.

Contribution

It introduces a web-scale pre-training approach for reasoning by extracting rationales from unlabeled data, enhancing reasoning performance of LLMs.

Findings

Achieves 3.9% average improvement on reasoning benchmarks.

Generalizes well across mathematical, scientific, and logical reasoning.

Outperforms larger models like GPT-4 in reasoning tasks.

Abstract

The reasoning steps generated by LLMs might be incomplete, as they mimic logical leaps common in everyday communication found in their pre-training data: underlying rationales are frequently left implicit (unstated). To address this challenge, we introduce RATIONALYST, a model for process-supervision of reasoning based on pre-training on a vast collection of rationale annotations extracted from unlabeled data. We extract 79k rationales from web-scale unlabelled dataset (the Pile) and a combination of reasoning datasets with minimal human intervention. This web-scale pre-training for reasoning allows RATIONALYST to consistently generalize across diverse reasoning tasks, including mathematical, commonsense, scientific, and logical reasoning. Fine-tuned from LLaMa-3-8B, RATIONALYST improves the accuracy of reasoning by an average of 3.9% on 7 representative reasoning benchmarks. It also demonstrates superior performance compared to significantly larger verifiers like GPT-4 and similarly sized models fine-tuned on matching training sets.