---
language: 
  - en
thumbnail: "https://huggingface.co/front/assets/huggingface_logo.png"
tags:
- political-analysis
- text-classification
- LoRA
- Phi-3
license: "apache-2.0"
datasets:
- custom-political-text
metrics:
- accuracy
- f1
base_model: "bigcode/phi-3"
pipeline_tag: "text-classification"
---


# Political Concept Classifier (Phi-3 LoRA)

A **LoRA-fine-tuned Phi-3 model** for classifying political text excerpts into key concepts. This model is designed to help analyze political content in articles, speeches, social media, or other text sources.

---

## **Model Overview**

* **Base model:** Phi-3
* **Fine-tuning method:** LoRA (Low-Rank Adaptation)
* **Task:** Text classification
* **Concepts:**

  * Economy
  * Immigration
  * Healthcare
  * [Additional categories if applicable]

This model predicts the most relevant political concept for a given text input.

---

## **Intended Use**

* Research and analysis of political text.
* Classifying statements, articles, or reports according to key political topics.
* Supporting journalists, analysts, or educators in understanding political content trends.

---

## **Limitations**

* Predictions are dependent on the training dataset and may misclassify ambiguous or out-of-domain text.
* Designed for analysis purposes; not legal, medical, or policy advice.
* May not capture nuances outside the trained categories.

---

## **Training Details**

* Fine-tuned with **LoRA** for efficient adaptation while preserving Phi-3’s pretrained knowledge.
* Trained on a dataset of labeled political text excerpts spanning multiple categories.