VishnuPJ/MalayaLLM-Paligemma-VQA-3B-Full-Precision

PaliGemma-3B-MalayaLLM

Introducing the Developer:

Discover the mind behind this model and stay updated on their contributions to the field https://www.linkedin.com/in/vishnu-prasad-j/

Model description

This is a PaliGemma-3B based model for Malayalam captioning and Visual Question Answering.

• Model type: A 3B PaliGemma-2 finetuned model on Malayalam captions and queries.
• Language(s): Malayalam and English
• Datasets:
  • VishnuPJ/SAM-LLAVA-20k-Malayalam-Caption-Pretrain
  • VishnuPJ/laion-14k-GPT4V-LIVIS-Captions_Malayalam
  • VishnuPJ/Malayalam-VQA
• Caption Model-Full Precisoin: VishnuPJ/MalayaLLM-Paligemma-Caption-3B-Full-Precision
• Caption 4bit Quant: VishnuPJ/MalayaLLM-Paligemma-Caption-3B-4bitQuant
• VQA Model-Full Precison: VishnuPJ/MalayaLLM-Paligemma-VQA-3B-Full-Precision
• VQA 4bit Quant: VishnuPJ/MalayaLLM-Paligemma-VQA-3B-4bitQuant
• VQA LORA Adapters: VishnuPJ/MalayaLLM-Paligemma-VQA-3B-Adapters
• Training Precision: float16,4bit

Dataset Creation

I have used indictrans2 for translating English datasets to Malayalam.

🌟Happy coding💻🌟