Results of the summarizing process for the arXiv paper: 2303.09014v1

Large language models (LLMs) have demonstrated the ability to perform complex reasoning in few- and zero-shot settings by generating intermediate chain of thought (CoT) reasoning steps. To address the limitation of requiring hand-crafted task-specific demonstrations and carefully scripted interleaving of model generations with tool use, the authors introduce Automatic Reasoning and Tool-use (ART), a framework that uses frozen LLMs to automatically generate intermediate reasoning steps as a program. ART selects demonstrations of multi-step reasoning and tool use from a task library when given a new task to solve. At test time, ART seamlessly pauses generation whenever external tools are called, integrates their output before resuming generation, and achieves substantial improvement over few-shot prompting and automatic CoT on unseen tasks in the BigBench and MMLU benchmarks. Moreover, ART matches performance of hand-crafted CoT prompts on a majority of these tasks. The framework is also extensible, making it easy for humans to improve performance by correcting errors in task-specific programs or incorporating new tools. The authors demonstrate this capability by drastically improving performance on select tasks with minimal human intervention. Overall, ART represents an important step towards automating multi-step reasoning and tool-use for large language models without requiring extensive manual effort. This has significant implications for fields such as natural language processing, where LLMs have shown great promise but require further development to achieve their full potential.

• Large language models (LLMs) can perform complex reasoning in few- and zero-shot settings by generating intermediate chain of thought (CoT) reasoning steps.
• Automatic Reasoning and Tool-use (ART) is a framework that uses frozen LLMs to automatically generate intermediate reasoning steps as a program.
• ART selects demonstrations of multi-step reasoning and tool use from a task library when given a new task to solve.
• ART seamlessly pauses generation whenever external tools are called, integrates their output before resuming generation, and achieves substantial improvement over few-shot prompting and automatic CoT on unseen tasks in the BigBench and MMLU benchmarks.
• ART matches performance of hand-crafted CoT prompts on a majority of these tasks.
• The framework is also extensible, making it easy for humans to improve performance by correcting errors in task-specific programs or incorporating new tools.
• ART represents an important step towards automating multi-step reasoning and tool-use for large language models without requiring extensive manual effort.

1. There are big computer programs that can think and solve problems without being taught everything. 2. A new framework called ART helps these programs by giving them steps to follow when they encounter a new problem. 3. ART can pause and use other tools to help it solve the problem, then continue where it left off. 4. ART is really good at solving problems quickly and accurately. 5. People can also make ART better by fixing mistakes or adding new tools. Definitions1. Large language models (LLMs) - Big computer programs that can understand and use human language to solve problems. 2. Framework - A set of rules or guidelines that help people create something in a specific way. 3. Tool-use - The ability to use different tools or methods to solve a problem or complete a task.

Automatic Reasoning and Tool-Use: A Framework for Automating Multi-Step Reasoning with Large Language Models

Large language models (LLMs) are powerful tools that can be used to perform complex reasoning in few- and zero-shot settings. However, they require hand-crafted task-specific demonstrations and carefully scripted interleaving of model generations with tool use in order to achieve their full potential. To address this limitation, the authors introduce Automatic Reasoning and Tool-use (ART), a framework that uses frozen LLMs to automatically generate intermediate reasoning steps as a program.

At test time, ART seamlessly pauses generation whenever external tools are called, integrates their output before resuming generation, and achieves substantial improvement over few-shot prompting and automatic chain of thought (CoT) on unseen tasks in the BigBench and MMLU benchmarks. Moreover, ART matches performance of hand-crafted CoT prompts on a majority of these tasks.

The framework is also extensible, making it easy for humans to improve performance by correcting errors in task specific programs or incorporating new tools. The authors demonstrate this capability by drastically improving performance on select tasks with minimal human intervention.

Overall, ART represents an important step towards automating multi-step reasoning and tool use for large language models without requiring extensive manual effort.