README.md

metadata

license: apache-2.0
base_model: Qwen/Qwen3.5-0.8B-Instruct
tags:
  - vikasit-ai
  - chandorkar-technologies
  - qwen-3.5
  - gguf
  - logic-hybrid
  - indian-ai
  - on-device
model_creator: Chandorkar Technologies
model_type: hybrid-gated-deltanet
pipeline_tag: text-generation
language:
  - en

🇮🇳 Vikasit AI Writer 0.8B (IQ4_XS)

Vikasit AI Writer 0.8B is a next-generation, ultra-lightweight language model optimized for the Indian ecosystem. Developed by Chandorkar Technologies, it is built upon the sovereign Qwen 3.5 hybrid architecture, featuring a 3:1 ratio of Gated DeltaNet to full softmax attention.

🚀 Performance Highlights

• Architecture: Hybrid Gated DeltaNet (O(1) memory for linear attention).
• Context Window: 262,144 tokens (Native).
• Optimization: Custom iMatrix quantized to IQ4_XS for maximum logic retention in a sub-500MB footprint.
• Identity: Native "Vikasit AI" persona, refined for professional and culturally relevant communication in India.

🛠 Quick Start (Ollama)

You can pull and run this model instantly from the Vikasit AI library:

ollama run vikasit-ai/writer:0.8b