kth8/gemma-3-270m-it-homeowner-classifier

TEXT GENERATIONConcurrent Unit Cost:1Model Size:0.3BQuant:BF16Context Size:32kPublished:Jun 17, 2026License:gemmaArchitecture:Transformer Featherless Exclusive Cold

The kth8/gemma-3-270m-it-homeowner-classifier is a 0.3 billion parameter Gemma-based language model, fine-tuned by kth8, specifically for classifying homeowner questions. This model specializes in categorizing user queries into predefined homeowner-related topics, leveraging its 32768 token context length for precise classification. It is optimized for single-turn, structured output tasks, making it ideal for automated support systems or data organization.

Loading preview...

Model Overview

This model, kth8/gemma-3-270m-it-homeowner-classifier, is a specialized language model derived from unsloth/gemma-3-270m-it. It has been supervised fine-tuned on the kth8/homeowner-classification dataset to excel at categorizing homeowner-related questions.

Key Capabilities

  • Precise Classification: Designed to classify homeowner questions into one of 17 specific categories, such as appliances, hvac, pool, electric, and irrigation.
  • Structured Output: Provides classification results in a clean JSON format, making it easy to integrate into automated workflows.
  • Optimized for Specificity: Achieves high accuracy for its intended task, with a recommended temperature=0.0 for deterministic results.
  • Efficient Performance: Built on a 268 million parameter Gemma base, offering efficient inference suitable for deployment.

Training Details

The model was trained using PEFT (Parameter-Efficient Fine-Tuning) with a LoRA rank of 32 and alpha of 64, targeting key attention and feed-forward modules. The SFT (Supervised Fine-Tuning) process involved 1 epoch with a batch size of 8 and a learning rate of 0.0002, achieving a best validation loss of 0.013801434077322483. Training was conducted on an NVIDIA L4 GPU, peaking at 2.205 GB VRAM usage.

Ideal Use Cases

  • Automated Customer Support: Automatically route homeowner inquiries to the correct department or provide instant, categorized responses.
  • Data Organization: Categorize large datasets of homeowner questions for analysis or database management.
  • Intelligent Forms: Guide users through question submission by suggesting categories or validating input based on content.