Foundation Model Architectures · Evaluation, Reliability & Trustworthy AI

Project EKA

An India-centric foundation large language model programme. The laboratory contributes data curation, multilingual evaluation, and benchmarking, including the open Eka-Eval suite.

Themes: Foundation Model Architectures · Evaluation, Reliability & Trustworthy AI
Agency: IndiaAI / MeitY
Sanction: ₹205 Cr

Evaluation toolkit

Eka-Eval

A comprehensive evaluation framework for large language models in Indian languages, covering 50+ benchmarks.

Page arXiv GitHub

Paper

Eka-Eval: A Comprehensive Evaluation Framework for Large Language Models in Indian Languages

Sinha, Sheth, Upperwal, Singh. arXiv 2025.

arXiv GitHub

Grants →All publications →