Overview
amitagh/shivneri-llm-it-v0.2 is a preliminary version of the Shivneri Marathi LLM, developed by Amit Ghadge. This 8 billion parameter instruction-tuned model is built on the Meta-Llama-3-8B-Instruct architecture, aiming to provide Generative AI capabilities for the Marathi-speaking population in India, which numbers approximately 83 million native speakers. It supports text generation in both Marathi and English.
Key Capabilities
- Bilingual Text Generation: Capable of generating creative and informative text in both Marathi and English.
- Llama3 Base: Leverages the robust architecture and capabilities of the Llama3 8B Instruct model.
- Instruction-Tuned: Fine-tuned using Supervised Fine-Tuning (SFT) with LoRA on relevant datasets.
- Developer: Developed by Amit Ghadge, with a focus on serving non-English speaking communities.
Good For
- Marathi Language Applications: Ideal for developing conversational AI and text generation tools specifically for Marathi speakers.
- Bilingual Use Cases: Suitable for applications requiring seamless text generation across Marathi and English.
- Early-Stage Development: As a preliminary version, it's useful for initial experimentation and understanding the model's potential in bilingual contexts. Users are advised to use with caution and anticipate further updates.