Rapidly increasing model scales coupled with steering methods such as chain-of-thought prompting have led to drastic improvements in language model reasoning. At the same time, models struggle with compositional generalization and are far from human performance on many reasoning-based benchmarks. Leveraging the success of chain-of-thought prompting, and also taking inspiration from context-aware decoding (CAD), we explore input-based contrasting methods to further encourage the type of reasoning induced by chain-of-thought prompting. While work remains to stabilize these results across datasets and models, the improvements we find warrant further investigation into input-based steering methods for context-aware reasoning.

I'm not a particularly experienced researcher (despite my title being "Senior" Research Scientist), but I've worked with some talented collaborators and spent a fair amount of time thinking about how to do research, so I thought I might write about how I go about it. My perspective is this: doing research is a skill that can be learned through practice, much like sports or music. The way I decompose research is into four skills: (1) idea conception and selection, (2) experiment design and execution, (3) writing the paper, and (4) maximizing impact. In other words, what differentiates good and bad researchers is these four skills. The first skill in research is coming up with or choosing a topic to work on.

We explore how generating a chain of thought -- a series of intermediate reasoning steps -- significantly improves the ability of large language models to perform complex reasoning. In particular, we show how such reasoning abilities emerge naturally in sufficiently large language models via a simple method called chain of thought prompting, where a few chain of thought demonstrations are provided as exemplars in prompting. Experiments on three large language models show that chain of thought prompting improves performance on a range of arithmetic, commonsense, and symbolic reasoning tasks. The empirical gains can be striking. For instance, prompting a 540B-parameter language model with just eight chain of thought exemplars achieves state of the art accuracy on the GSM8K benchmark of math word problems, surpassing even finetuned GPT-3 with a verifier.

This repository contains a collection of papers and resources on Reasoning in Large Language Models. Feel free to let me know the missing papers (issue or pull request). Thank Kevin Chen-Chuan Chang @UIUC, Jason Wei @Google Brain, Denny Zhou @Google Brain for insightful discussions and suggestions. We mainly focus on techniques that are applicable to improving or eliciting "reasoning" in large language models like GPT-3 (175B) Papers in this paradigm vary a lot and are usually based on small models trained on specific datasets. We list several papers here for reference (that is, the list is not complete).

Large Language Models (LLMs) have achieved excellent performances in various tasks. However, fine-tuning an LLM requires extensive supervision. Human, on the other hand, may improve their reasoning abilities by self-thinking without external inputs. In this work, we demonstrate that an LLM is also capable of self-improving with only unlabeled datasets. We use a pre-trained LLM to generate "high-confidence" rationale-augmented answers for unlabeled questions using Chain-of-Thought prompting and self-consistency, and fine-tune the LLM using those self-generated solutions as target outputs. We show that our approach improves the general reasoning ability of a 540B-parameter LLM (74.4% 82.1% on GSM8K, 78.2% 83.0% on DROP, 90.0% 94.4% on OpenBookQA, and 63.4% 67.9% on ANLI-A3) and achieves state-of-the-art-level performance, without any ground truth label. We conduct ablation studies and show that finetuning on reasoning is critical for self-improvement. Scaling has enabled Large Language ...