This dataset is tailored for Mistral chat template-based models, designed to enhance Kannada communication among AI models.
Dataset Structure
The dataset is organized to include four types of pairs:
- English Instruction - English Output (ENTOEN)
- English Instruction - Kannada Output (ENTOKN)
- Kannada Instruction - English Output (KNTOEN)
- Kannada Instruction - Kannada Output (KNTOKN)
This intentional diversification aims to enhance the model's proficiency in cross-lingual tasks, making it more adaptable to various linguistic scenarios.
Credits
Based on Cognitive-Lab/Kannada-Instruct-dataset
- Downloads last month
- 1