Large Language Models as Tool Makers

By Tianle Cai, Xuezhi Wang, Tengyu Ma, Xinyun Chen, Denny Zhou. International Conference on Learning Representations, 2023

316 citations recorded by the source corpus.

Recent research has highlighted the potential of large language models (LLMs) to improve their problem-solving capabilities with the aid of suitable external tools. In our work, we further advance this concept by introducing a closed-loop framework, referred to as LLMs A s Tool Makers (LATM), where LLMs create their own reusable tools for problem-solving. Our approach consists of two phases: 1) tool making: an LLM acts as the tool maker that crafts tools for a set of tasks. 2) tool using: another LLM acts as the tool user, which applies the tool built by the tool maker for problem-solving. On the problem-solving server side, tool-making enables continual tool generation and caching as new requests emerge. This framework enables subsequent requests to access cached tools via their corresponding APIs, enhancing the efficiency of task resolution. Recognizing that tool-making requires more sophisticated capabilities, we assign this task to a powerful, albeit resource-intensive, model. Conversely, the simpler tool-using phase is delegated to a lightweight model. This strategic division of labor allows the once-off cost of tool-making to be spread over multiple instances of tool-using, significantly reducing average costs while maintaining strong performance. Furthermore, our method offers a functional cache through the caching and reuse of tools, which stores the functionality of a class of requests instead of the natural language responses from LLMs, thus extending the applicability of the conventional cache mechanism. We evaluate our approach across various complex reasoning tasks, including Big-Bench tasks. With GPT-4 as the tool maker and GPT-3.5 as the tool user, LATM demonstrates performance equivalent to using GPT-4 for both roles, but with a significantly reduced inference cost.

Subfields: Artificial Intelligence

View Paper: https://www.semanticscholar.org/paper/32dcd0887537cece54e214f531d2c384470b023f

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

Citations in the corpus

Unleashing the Emergent Cognitive Synergy in Large Language Models: A Task-Solving Agent through Multi-Persona Self-Collaboration (2023, background): “…and recent findings that assigning personas (Deshpande et al., 2023; Xu et al., 2023) to LLMs can elicit specific behaviors, improve answer quality, and potentially build an AI society (Park et al., 2023; Schick et al., 2022; Li et al., 2023; Cai et al., 2023) with collaborative LLM agents.”
Prompt Injection attack against LLM-integrated Applications (2023, background): “There is ongoing research focusing on the enhancement of LLMs to improve their operational capabilities [11, 26, 28, 42, 53,57,59].”
Visual Adversarial Examples Jailbreak Aligned Large Language Models (2023, background): “, robotics [37; 22; 9], APIs management [60], making tools [11], developing plug-ins [43], the implications of our attacks may further expand according to specific downstream applications.”
AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation (2023, unknown): “We believe that more systematic work will be required to develop guidelines for application-specific agents, to create a large OSS knowledge base of agents, and to create agents that can discover and upgrade their skills (Cai et al., 2023).”
Tool learning with large language models: a survey (2024, unknown): “Overloading a single tool with too many features should be avoided to maintain efficiency and effectiveness.”
A Survey on Knowledge Distillation of Large Language Models (2024, methodology): “Commonly used methods mainly relied on human-curated data for training (Parisi et al., 2022; Nakano et al., 2022; Qin et al., 2023c; Song et al., 2023b) or prompt designing(Cai et al., 2023; Shen et al., 2023; Hao et al., 2024).”
Large Language Models as Analogical Reasoners (2023, methodology): “5-turbo In line with existing work on code generation (Li et al., 2022b; Chen et al., 2023), we report the Acc@1 and Acc@10 metrics.”
MetaGPT: Meta Programming for Multi-Agent Collaborative Framework (2023, background): “However, existing systems [1, 2, 3, 4, 5, 6, 7] tend to oversimplify the complexities inherent to real-world applications.”
API-Bank: A Comprehensive Benchmark for Tool-Augmented LLMs (2023, background): “LATM(Cai et al., 2023) and CREATOR (Qian et al., 2023) leverage GPT-4 to make API tools.”
ChatDev: Communicative Agents for Software Development (2023, background): “…skills in memory (Park et al., 2023; Sumers et al., 2023), planning (Chen et al., 2023b; Liu et al., 2023) and tool use (Schick et al., 2023; Cai et al., 2023; Qin et al., 2023a; Ruan et al., 2023; Yang et al., 2023b), enabling them to reason in complex scenarios (Wei et al., 2022b; Zhao et…”
When large language models meet personalization: perspectives of challenges and opportunities (2023, unknown): “LATMis entirelycomposedofLLMs,enablingtheself-generationandself-utilizationoftools.”
A survey on large language model based autonomous agents (2023, unknown): “Similarly, LATM [10] empowers LLMs to utilize Python code for crafting its own reusable tools, fostering a flexible approach to problem-solving.”
Large Language Models as Optimizers (2023, background): “When and how to trigger such tool use cases remains an interesting topic (see e.g., (Schick et al., 2023; Cai et al., 2023)).”
Scaling Synthetic Data Creation with 1,000,000,000 Personas (2024, background): “This can not only help deepen our understanding of complex systems but also speed up innovation by facilitating rapid iteration and experimentation.”
How Far Can Camels Go? Exploring the State of Instruction Tuning on Open Resources (2023, background): “Coding is a particular application that people have used language models for and might be important for integrating these models with external tools [5].”