Repository: Irfanuruchi/qwen2.5-1.5b-buildeng-v7-lora
This model is a building-engineering focused LoRA fine-tune of Qwen2.5-1.5B-Instruct.
V7 was made to move away from only direct engineering calculations and push the model more toward engineering reasoning behavior. Earlier versions could solve formulas, but still sometimes accepted wrong assumptions, bad unit conversions, unrealistic HVAC values, or incomplete design prompts.
The main goal of V7 was… See the full description on the dataset page:
https://huggingface.co/datasets/Irfanuruchi/buildeng-v7-15k.
