Modelpublic

muhammadhanifnahara/edgesense-telemetry-reasoning

sourceHugging Faceupdated 4mo agoView on Hugging Face

0likes

Model Card

EdgeSense – Gemma 4 LoRA for Explainable Predictive Maintenance

This repository contains a LoRA adapter fine‑tuned on Gemma 4 for explainable industrial predictive maintenance.

Base Model

—google/gemma-4-E2B-it

Training Method

—LoRA fine‑tuning using Unsloth
—4‑bit quantization
—Predictive Maintenance Dataset

Purpose

Instead of being a classifier, the model is intended to be a reasoning model. From raw machine telemetry, it produces interpretable diagnostic narratives and confidence-aware maintenance recommendations.

Usage

python

from unsloth import FastLanguageModel

model, tokenizer = FastLanguageModel.from_pretrained(
    model_name="google/gemma-4-E2B-it",
    adapter_name="USERNAME/edgesense-gemma4-lora",
    load_in_4bit=True,
)