Qwen2.5-32B BuildEng GGUF (Q5_K_M)
Repository: Irfanuruchi/qwen2.5-32b-buildeng-GGUF-Q5_K_M
This repository contains the GGUF Q5_K_M quantized release of BuildEng V8 32B based on Qwen/Qwen2.5-32B-Instruct.
BuildEng is a domain-specialized engineering language model project focused on civil engineering, structural reasoning, construction workflows, field diagnostics, and conservative engineering-assistant behavior.
The model was trained to approach engineering problems cautiously instead of behaving like an overconfident general-purpose assistant. The primary focus is inspection-first reasoning, uncertainty handling, structural risk awareness, and separating symptoms from confirmed diagnosis.
This GGUF release allows the model to run through llama.cpp, Ollama, LM Studio, KoboldCpp, and other GGUF-compatible inference frameworks.
Quantization
Quant type: Q5_K_M
Original source: merged BuildEng V8 32B model
Conversion stack: llama.cpp GGUF conversion and quantization
The Q5_K_M quantization was selected as a higher-quality GGUF release focused on stronger reasoning preservation while still remaining practical for local inference.
Dedication
This release is dedicated to my father for his birthday.
He is a civil and building engineer, and one of the main reasons I chose engineering myself. A large part of how I understand responsibility, discipline, and technical thinking comes from him.
BuildEng is my own field connected to the engineering world that inspired me growing up.
“My father is the engineer I looked up to long before I understood what engineering really meant.”
Thank you for everything.
Base Model
Qwen/Qwen2.5-32B-Instruct
Dataset
Dataset repository: Irfanuruchi/buildeng
Training dataset: BuildEng V8 Final
Total validated samples: 145,117
The BuildEng V8 dataset was designed primarily around civil and structural engineering workflows with conservative engineering reasoning behavior.
The dataset includes reinforced concrete beams, slabs, columns, footings, retaining walls, structural load paths, settlement reasoning, soil behavior, failure analysis, temporary bracing, temporary shoring, waterproofing failures, renovation unknowns, HVAC airflow reasoning, inspection workflows, uncertainty handling, contradiction handling, repair logic, multi-turn engineering diagnosis, and adversarial engineering prompts.
The dataset was intentionally structured to reduce unsafe certainty and encourage more cautious engineering behavior.
Main Behavior Goals
BuildEng 32B was trained to:
- separate symptoms from diagnosis
- avoid unsupported structural approval
- request missing engineering information
- reduce unsafe certainty
- follow inspection-first workflows
- recognize load-path concerns
- identify potentially unsafe field conditions
- support investigation and repair workflows
- distinguish serviceability concerns from structural-capacity concerns
- respond conservatively to cracking, settlement, corrosion, deflection, demolition, moisture, and structural damage scenarios
The model intentionally avoids behaving like an overconfident assistant.
Usage
Example llama.cpp usage:
1./llama-cli \
2 -m buildeng-v8-32b-Q5_K_M.gguf \
3 -c 4096 \
4 -ngl 999
Example Ollama Modelfile:
1FROM ./buildeng-v8-32b-Q5_K_M.gguf
2
3PARAMETER num_ctx 4096
4
5SYSTEM """
6You are BuildEng V8, a conservative civil and structural engineering assistant.
7"""
Important Limitation
This model is not a licensed engineer and must not be used as final engineering approval, design certification, or construction sign-off.
It is intended for research, education, engineering-assistant workflows, drafting support, and preliminary engineering reasoning only.
Final decisions involving structural safety, demolition, occupancy, repairs, construction approval, or certification should always be reviewed and approved by a qualified licensed engineer.
Author
Irfan Uruchi
Part of my ongoing work on domain-specialized engineering language models, structural reasoning datasets, and practical civil/building engineering AI systems.
