ConstructionSafetyQA-1.2B-V1
ConstructionSafetyQA-1.2B-V1 is a fine-tuned version of LiquidAI/LFM2-1.2B, optimized using Unsloth for construction safety question answering.
This model is designed to provide short, simple, and practical answers for construction workers and supervisors. It focuses on safety training, hazard prevention, and site best practices, avoiding unnecessary technical jargon.
Model Details
- Base Model: LiquidAI/LFM2-1.2B
- Fine-tuning Framework: Unsloth
- Task: Instruction-tuned for Construction Safety Q&A
- Answer Style:
- Short (1–3 sentences)
- Worker/supervisor-focused
- Safety-first, practical instructions
Example Usage
from transformers import AutoModelForCausalLM, AutoTokenizer
model_name = "yasserrmd/ConstructionSafetyQA-1.2B-V1"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
model_name,
device_map="auto",
)
messages = [
{"role": "system", "content": "You are a construction safety trainer. Always give short, practical, safety-first answers."},
{"role": "user", "content": "What should workers do if scaffolding guardrails are missing?"}
]
inputs = tokenizer.apply_chat_template(messages, return_tensors="pt").to(model.device)
outputs = model.generate(inputs, max_new_tokens=128)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
Example Outputs
Input:
What should workers do if scaffolding guardrails are missing?
Output:
Workers must stop using the scaffold and report it. Guardrails must be installed before work continues.
Input:
Why is hydration important when working in hot weather?
Output:
Hydration prevents heat exhaustion and keeps workers alert and safe.
Intended Use
- Training aid for construction workers and supervisors
- On-site safety Q&A assistant
- Toolbox talks and quick safety refreshers
Not intended to replace certified training or safety regulations. Always follow official site rules and local laws.
Limitations
- Provides general safety guidance, not region-specific code compliance.
- Answers are short and simple by design (not technical references).
- May need human review for critical safety decisions.
Citation
If you use this model in your work, please cite:
@misc{yasser2025constructionsafetyqa,
title={ConstructionSafetyQA-1.2B-V1},
author={Mohamed Yasser},
year={2025},
publisher={Hugging Face},
howpublished={\url{https://huggingface.co/yasserrmd/ConstructionSafetyQA-1.2B-V1}}
}
Acknowledgements
- Base model: LiquidAI/LFM2-1.2B
- Fine-tuning: Unsloth
- Inspiration: Promoting construction safety awareness with accessible AI tools.