Skip to content
#

aya

Here are 6 public repositories matching this topic...

Research: Does multilingual self-consistency improve LLM reasoning beyond math? Empirical study across 6 benchmarks (commonsense, ethics, NLI, knowledge) and 10 languages using Qwen2.5-32B and Aya Expanse 32B on Apple Silicon (MLX). Chain-of-thought + cross-lingual prompting.

  • Updated Jul 10, 2026
  • Python

Add this topic to your repo

To associate your repository with the aya topic, visit your repo's landing page and select "manage topics."

Learn more