This dataset is a cleaned version of Telugu-MultiTask-Instruct-77K with Telugu keys.
It uses enhanced_prompt as instruction and enhanced_completion as output.
Extracted enhanced_prompt → సూచన (instruction) and enhanced_completion → అవుట్పుట్ (output)
Filtered to keep only entries with no English letters
Removed duplicate entries
Normalized whitespace
Each entry follows the Alpaca format with… See the full description on the dataset page:
https://huggingface.co/datasets/VenkataRamanaKurumallajaddangi/Pure-Telugu-Alpaca.
