gemma-3-1b-it-LoRA-GLoRE

🧠 Gemma-3-1B-IT LoRA Adapter — GLoRE Multi-Class Classification

📌 Model Overview

This repository contains a LoRA adapter fine-tuned on google/gemma-3-1b-it for multi-class text classification using the GLoRE dataset.

The model predicts one of the following 12 labels:

Yes, No, Neutral, (D), A, B, C, D, E, N, (C), (A)

This LoRA adapter is efficient, lightweight, and designed to extend the Gemma-3-1B-IT model with classification capabilities while keeping resource usage low.

---

🧩 Use Cases

• Multi-class text classification
• Zero-shot / few-shot classification tasks using custom prompts
• Educational or research applications
• Lightweight inference on consumer GPUs

---

🚀 How to Use

Loading the LoRA Adapter

from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel

base_model = "google/gemma-3-1b-it"
adapter = "SwashBuckler001/gemma-3-1b-it-LoRA-GLoRE"

tokenizer = AutoTokenizer.from_pretrained(base_model)
model = AutoModelForCausalLM.from_pretrained(base_model)
model = PeftModel.from_pretrained(model, adapter)

text = "Your input here"
inputs = tokenizer(text, return_tensors="pt")

outputs = model.generate(**inputs, max_new_tokens=10)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))


Training Details

The adapter was trained using:

python peft_training.py \
  --model-name google/gemma-3-1b-it \
  --train-file ../GLoRE/data/splits/train.jsonl \
  --output-dir gemma-3-1b-it-LoRA-GLoRE \
  --classes Yes No Neutral "(D)" A B C D E N "(C)" "(A)"