New dataset teaches LLMs Indian Knowledge Systems across 7 languages

Researchers have developed IKS-Instruct, a new multilingual dataset designed to teach large language models about Indian Knowledge Systems (IKS). The dataset contains over 24,000 instruction-response pairs in seven languages, covering 41 pedagogical techniques and aligning with Indian educational curricula. Evaluations show that a fine-tuned 7B model using IKS-Instruct achieves scores comparable to a general-purpose reference model on IKS-specific dimensions, at a significantly lower deployment cost. AI

IMPACT This dataset could enable LLMs to provide more culturally relevant and specialized educational content, particularly in non-English languages.

RANK_REASON The cluster describes a new academic dataset and its evaluation, fitting the research bucket. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• Central Board of Secondary Education
• English
• Hindi
• IKS-Instruct
• Indian Knowledge Systems
• Kannada
• large language models
• Malayalam
• Nemotron-Nano
• Sanskrit
• Tamil
• Telugu

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →