meftah416/gemma-eppy-270m
The meftah416/gemma-eppy-270m is a 270 million parameter Gemma-3-270M model fine-tuned by meftah416, specifically designed for tasks related to EnergyPlus building simulation and the eppy Python API. It excels at generating and understanding text for building energy modeling, HVAC systems, and eppy scripting. This specialized model, with a 32768 token context length, is optimized for domain-specific applications in building performance analysis.
Loading preview...
Model Overview
meftah416/gemma-eppy-270m is a specialized Gemma-3-270M model, developed by meftah416, fine-tuned for EnergyPlus building simulation and the eppy Python API. This lightweight model (270M parameters) is designed to understand and generate content related to building energy modeling, HVAC systems, thermal zones, schedules, and eppy scripting.
Key Features
- Fine-tuned on 2,700 high-quality, self-created training examples.
- Validated on 300 test examples, focusing on EnergyPlus IDF and eppy Python domains.
- Achieved 2x faster training using Unsloth.
- Offers a context window of 8192 tokens.
Capabilities
This model is trained to handle a variety of tasks:
- EnergyPlus Tasks: Generating IDF snippets, explaining object syntax, creating thermal zone definitions, configuring HVAC systems, and generating occupancy/schedule objects.
- eppy Tasks: Generating eppy Python code for simulations, explaining API usage, programmatically creating building objects, and manipulating IDF files.
Limitations
Users should always validate generated EnergyPlus IDF files before running simulations, as the model may occasionally produce incorrect syntax. Its inference speed is approximately 50-100 tokens/sec on an A10 GPU, with memory usage around 3-4 GB VRAM (float16).