A 5-million-parameter Hindi storyteller, trained from scratch.
Tokenizer efficiency
Our Hindi tokenizer needs 4.32× fewer tokens than GPT-2's.
Live demo
एक बार की बात है, एक छोटा खरगोश चीकू जंगल में रहता था। चीकू को गाजर बहुत पसंद थी। एक दिन उसने अपने दोस्त को एक बड़ी गाजर दी। दोनों बहुत खुश हुए।
Showing a sample. Tap Generate for a fresh tale.