This page holds the dataset proposed in Agent-FLAN, which consists of AgentInstruct, Toolbench, and customized negative agent samples as its source datasets.
Open-sourced Large Language Models (LLMs) have achieved great success in various NLP tasks, however, they are still far inferior to API-based models when acting as… See the full description on the dataset page:
https://huggingface.co/datasets/internlm/Agent-FLAN.
