sazid63/archresidencebd-models

ArchResidenceBD Model Fleet

This repository contains the model fleet for ArchResidenceBD - an AI civil engineer for Bangladeshi residential buildings.

Models

Model	Role	Base	Quantization	Size	Purpose
rchie-chat-v0-Q4_K_M.gguf	Chat	Qwen3-4B-Instruct-2507	Q4_K_M	2.33 GB	Interview, parsing, explanations
rchres-repair-v0-Q4_K_M.gguf	Repair	Qwen2.5-Coder-3B-Instruct	Q4_K_M	1.80 GB	CAD script repair
Qwen3-Embedding-0.6B-Q8_0.gguf	Embedding	Qwen3-Embedding-0.6B	Q8_0	0.60 GB	RAG retrieval
Qwen3-Reranker-0.6B-Q8_0.gguf	Reranker	Qwen3-Reranker-0.6B	Q8_0	0.60 GB	RAG reranking

Usage

All models are in GGUF format, quantized for llama.cpp.

`ash

Chat model

llama-server -m archie-chat-v0-Q4_K_M.gguf -ngl 99 -c 8192

Repair model

llama-server -m archres-repair-v0-Q4_K_M.gguf -ngl 99 -c 8192

Embedding

llama-server -m Qwen3-Embedding-0.6B-Q8_0.gguf -ngl 0 --embeddings --pooling last

Reranker

llama-server -m Qwen3-Reranker-0.6B-Q8_0.gguf -ngl 0 --reranking `

Licenses

• Base models: Apache-2.0 (Qwen family)
• Fine-tunes: Apache-2.0 (derivative)

Project

Part of ArchResidenceBD - AI Civil Engineer for Bangladeshi Residential Buildings.

Citation

ibtex @misc{archresidencebd-models, title = {ArchResidenceBD Model Fleet}, author = {ArchResidenceBD Team}, year = {2026}, url = {https://huggingface.co/sazid63/archresidencebd-models} }