akashsomanai/grammar-corrector-gemma-3-270m-it

Grammar Corrector – Gemma 3 270M (IT)

This is a fine-tuned model based on google/gemma-3-270m-it, trained for grammatical error correction (GEC) in English text.

The model rewrites input sentences to correct grammatical mistakes while preserving the original meaning.

🧠 Model Details

• —Base model: google/gemma-3-270m-it
• —Task: Grammatical Error Correction
• —Fine-tuning method: QLoRA (parameter-efficient fine-tuning)
• —Framework: Hugging Face TRL (Supervised Fine-Tuning)

📚 Training Data

The model was fine-tuned on a grammar correction dataset containing pairs of:

• —grammatically incorrect input text
• —corrected target text

The data was formatted in a chat-style structure with system, user, and assistant roles.

🚀 Usage

from transformers import pipeline

pipe = pipeline(
    "text-generation",
    model="akashsomanai/grammar-corrector-gemma-3-270m-it"
)

pipe("She go to office yesterday")