buddhist-nlp/gemma2-mitra-base
buddhist-nlp/gemma2-mitra-base is a base language model derived from Google's Gemma 2 9B architecture. It has been continuously pre-trained for two epochs on 7 billion tokens of Buddhist texts in Sanskrit, Tibetan, English, and Pāli. This model is specifically designed for tasks involving Buddhist textual analysis and understanding, serving as a foundational model for specialized applications in this domain.
Loading preview...
Overview
buddhist-nlp/gemma2-mitra-base is a specialized base language model built upon the Gemma 2 9B architecture. Developed by buddhist-nlp, this model underwent continuous pre-training for two epochs on a unique dataset comprising 7 billion tokens of Buddhist data. The training data includes texts preserved in Sanskrit, Tibetan, English, and Pāli, making it highly specialized for tasks related to Buddhist studies and natural language processing within this domain.
Key Characteristics
- Foundation Model: Based on Google's Gemma 2 9B, providing a robust architectural base.
- Specialized Pre-training: Continuously pre-trained on a large corpus of multilingual Buddhist texts.
- Multilingual Support: Incorporates data from Sanskrit, Tibetan, English, and Pāli, enabling cross-lingual understanding within the Buddhist context.
- Base Model: This is a base model and not instruction-tuned. It will perform poorly on general tasks without few-shot examples.
Use Cases
- Buddhist Text Analysis: Ideal for research and applications requiring deep understanding of Buddhist scriptures and literature.
- Specialized NLP Tasks: Suitable as a base for fine-tuning on specific tasks within Buddhist NLP, such as translation, summarization, or information extraction from religious texts.
- Further Development: Can serve as a strong foundation for developing instruction-tuned models (like buddhist-nlp/gemma-2-mitra-it) or other specialized applications in the Buddhist domain.