API-BLEND: A Comprehensive Corpora for Training and Benchmarking API LLMs

TL;DR

API-BLEND is a large, curated dataset designed to train and benchmark API-augmented large language models, enabling better tool integration and task performance evaluation.

Contribution

The paper introduces API-BLEND, a comprehensive dataset for training and benchmarking API-using LLMs, focusing on real-world API interaction tasks.

Findings

API-BLEND improves training effectiveness for tool-augmented LLMs.

The dataset enables systematic benchmarking of API integration capabilities.

API-BLEND covers diverse API interaction scenarios.

Abstract

There is a growing need for Large Language Models (LLMs) to effectively use tools and external Application Programming Interfaces (APIs) to plan and complete tasks. As such, there is tremendous interest in methods that can acquire sufficient quantities of train and test data that involve calls to tools / APIs. Two lines of research have emerged as the predominant strategies for addressing this challenge. The first has focused on synthetic data generation techniques, while the second has involved curating task-adjacent datasets which can be transformed into API / Tool-based tasks. In this paper, we focus on the task of identifying, curating, and transforming existing datasets and, in turn, introduce API-BLEND, a large corpora for training and systematic testing of tool-augmented LLMs. The datasets mimic real-world scenarios involving API-tasks such as API / tool detection, slot filling, and sequencing of the detected APIs. We demonstrate the utility of the API-BLEND dataset for both training and benchmarking purposes.