Logo Lanfrica
fr
Accueil
Atlas
Analyses
Documentation
Sign in
Retour
BerkGuny/Sequential-fine-tuning-for-improving-everyday-cultural-knowledge-in-multilingual-LLMs
Domaine:
natural language processing
Type de record:
model
Créateur:
Ber
Hôte:
Visit
Actions
Share
Report an issue
Developed a two-stage multilingual LLM fine-tuning pipeline that improves culturally grounded question answering in low-resource languages using WebFAQ-Culture and BLEnD-Synth.
Visit
github.com
Tasks
question answering