Fine-tuned a model on Advaita Vedanta text (huggingface.co)

Aarav Shirpurkar has introduced "Turiya," an experimental language model finetuned on texts related to Advaita Vedanta, including the Ashtavakra Gita and the works of Ramana Maharshi. The key objective of this initiative is to explore whether training a model on consciousness-focused literature results in qualitatively different reasoning about the nature of self compared to a standard model like Qwen3 4B. Turiya is built upon a unique dataset of approximately 2,700 conversation pairs, structured to prioritize philosophical inquiry over mere informational responses. The significance of Turiya lies in its potential to redefine how AI interacts with concepts of consciousness and self-awareness, probing whether a model trained on introspective texts can produce responses that are structurally distinct from those of traditional AI systems. With a layered training approach that encourages genuine philosophical exploration rather than superficial reporting, this model offers a new avenue to examine the deeper questions of existence and awareness. While it does not claim any form of consciousness, Turiya serves as a thought-provoking probe into the interplay between language models and philosophical traditions, inviting users to engage in meaningful discourse about the nature of consciousness itself.