Building Resource-Constrained Language Agents: A Korean Case Study on Chemical Toxicity Information

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Cho, Hojun, Kim, Donghu, Yang, Soyoung, Lee, Chan, Lee, Hunjoo, Choo, Jaegul
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866915540197441536
author Cho, Hojun
Kim, Donghu
Yang, Soyoung
Lee, Chan
Lee, Hunjoo
Choo, Jaegul
author_facet Cho, Hojun
Kim, Donghu
Yang, Soyoung
Lee, Chan
Lee, Hunjoo
Choo, Jaegul
contents Language agents powered by large language models (LLMs) face significant deployment challenges in resource-constrained environments, particularly for specialized domains and less-common languages. This paper presents Tox-chat, a Korean chemical toxicity information agent devised within these limitations. We propose two key innovations: a context-efficient architecture that reduces token consumption through hierarchical section search, and a scenario-based dialogue generation methodology that effectively distills tool-using capabilities from larger models. Experimental evaluations demonstrate that our fine-tuned 8B parameter model substantially outperforms both untuned models and baseline approaches, in terms of DB faithfulness and preference. Our work offers valuable insights for researchers developing domain-specific language agents under practical constraints.
format Preprint
id arxiv_https___arxiv_org_abs_2503_17753
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Building Resource-Constrained Language Agents: A Korean Case Study on Chemical Toxicity Information
Cho, Hojun
Kim, Donghu
Yang, Soyoung
Lee, Chan
Lee, Hunjoo
Choo, Jaegul
Computation and Language
Artificial Intelligence
Language agents powered by large language models (LLMs) face significant deployment challenges in resource-constrained environments, particularly for specialized domains and less-common languages. This paper presents Tox-chat, a Korean chemical toxicity information agent devised within these limitations. We propose two key innovations: a context-efficient architecture that reduces token consumption through hierarchical section search, and a scenario-based dialogue generation methodology that effectively distills tool-using capabilities from larger models. Experimental evaluations demonstrate that our fine-tuned 8B parameter model substantially outperforms both untuned models and baseline approaches, in terms of DB faithfulness and preference. Our work offers valuable insights for researchers developing domain-specific language agents under practical constraints.
title Building Resource-Constrained Language Agents: A Korean Case Study on Chemical Toxicity Information
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2503.17753