Toolformer: Language Models Can Teach Themselves to Use Tools

Abstract

Language models (LMs) exhibit remarkable abilities to solve new tasks from just a few examples or textual instructions, especially at scale. They also, paradoxically, struggle with basic functionality, such as arithmetic or factual lookup, where much simpler and smaller models excel. In this paper, we show that LMs can teach themselves to use external tools via simple APIs and achieve the best of both worlds. We introduce Toolformer, a model trained to decide which APIs to call, when to call them, what arguments to pass, and how to best incorporate the results into future token prediction. This is done in a self-supervised way, requiring nothing more than a handful of demonstrations for each API. We incorporate a range of tools, including a calculator, a Q\&A system, two different search engines, a translation system, and a calendar. Toolformer achieves substantially improved zero-shot performance across a variety of downstream tasks, often competitive with much larger models, without sacrificing its core language modeling abilities.

Cited by in corpus (30)

• Unleashing the potential of prompt engineering for large language models
• 14 Examples of How LLMs Can Transform Materials Science and Chemistry: A Reflection on a Large Language Model Hackathon
• AI Agents vs. Agentic AI: A Conceptual Taxonomy, Applications and Challenges
• AI Agents Under Threat: A Survey of Key Security Challenges and Future Pathways
• ChatEDA: A Large Language Model Powered Autonomous Agent for EDA
• Materials science in the era of large language models: a perspective
• Evaluation of large language models for discovery of gene set function
• LINC: A Neurosymbolic Approach for Logical Reasoning by Combining Language Models with First-Order Logic Provers
• Improving accuracy of GPT-3/4 results on biomedical data using a retrieval-augmented language model
• Exploring the Roles of Large Language Models in Reshaping Transportation Systems: A Survey, Framework, and Roadmap
• Domain-specific ChatBots for Science using Embeddings
• Opportunities for Large Language Models and Discourse in Engineering Design
• Building Privacy-Preserving and Secure Geospatial Artificial Intelligence Foundation Models
• Large Knowledge Model: Perspectives and Challenges
• Ghost in the Minecraft: Generally Capable Agents for Open-World Environments via Large Language Models with Text-based Knowledge and Memory
• Exploring Gen-AI applications in building research and industry: A review
• Opportunities and Challenges of Generative-AI in Finance
• Learning Retrieval Augmentation for Personalized Dialogue Generation
• Automating Traffic Model Enhancement with AI Research Agent
• Transforming Agency. On the mode of existence of Large Language Models
• Advancing Tool-Augmented Large Language Models via Meta-Verification and Reflection Learning
• Complex QA and language models hybrid architectures, Survey
• Relational Programming with Foundation Models
• Flows: Building Blocks of Reasoning and Collaborating AI
• Evaluating Language Model Agency through Negotiations
• BibSonomy Meets ChatLLMs for Publication Management: From Chat to Publication Management: Organizing your related work using BibSonomy & LLMs
• HiddenTables & PyQTax: A Cooperative Game and Dataset For TableQA to Ensure Scale and Data Privacy Across a Myriad of Taxonomies
• Co-Disclosing the Computer: LLM-Mediated Computing through Reflective Conversation
• ITCMA: A Generative Agent Based on a Computational Consciousness Structure
• Tool-Augmented LLMs as a Universal Interface for IDEs