Saved in:
Bibliographic Details
Main Authors: Yakovlev, Konstantin, Nikolenko, Sergey, Bout, Andrey
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2410.12004
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912073805463552
author Yakovlev, Konstantin
Nikolenko, Sergey
Bout, Andrey
author_facet Yakovlev, Konstantin
Nikolenko, Sergey
Bout, Andrey
contents The recently proposed ToolkenGPT tool learning paradigm demonstrates promising performance but suffers from two major issues: first, it cannot benefit from tool documentation, and second, it often makes mistakes in whether to use a tool at all. We introduce Toolken+ that mitigates the first problem by reranking top $k$ tools selected by ToolkenGPT and the second problem with a special "Reject" option such that the model will generate a vocabulary token if "Reject" is ranked first. We demonstrate the effectiveness of Toolken+ on multistep numerical reasoning and tool selection tasks.
format Preprint
id arxiv_https___arxiv_org_abs_2410_12004
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Toolken+: Improving LLM Tool Usage with Reranking and a Reject Option
Yakovlev, Konstantin
Nikolenko, Sergey
Bout, Andrey
Computation and Language
The recently proposed ToolkenGPT tool learning paradigm demonstrates promising performance but suffers from two major issues: first, it cannot benefit from tool documentation, and second, it often makes mistakes in whether to use a tool at all. We introduce Toolken+ that mitigates the first problem by reranking top $k$ tools selected by ToolkenGPT and the second problem with a special "Reject" option such that the model will generate a vocabulary token if "Reject" is ranked first. We demonstrate the effectiveness of Toolken+ on multistep numerical reasoning and tool selection tasks.
title Toolken+: Improving LLM Tool Usage with Reranking and a Reject Option
topic Computation and Language
url https://arxiv.org/abs/2410.12004