Noracle/IELTS_Essay_Evaluation

NewChannel IELTS Essay Intelligent Assessment System

Introduction

This system is based on Hugging Face pre - trained models and realizes the automatic scoring and detailed feedback generation for IELTS Writing Task 2. It consists of two core functions:

1. 1.Overall score prediction (Pipeline 1): Use the RoBERTa model to predict a continuous score from 0 - 9.
2. 2.Detailed dimension feedback (Pipeline 2): Use the DistilBART model to generate scores and comments for each dimension (task response, coherence and cohesion, etc.).

Deployment Instructions

Environment Dependencies

• —Python >= 3.8
• —Dependent packages: See requirements.txt

Startup Method

1. 1.Create a new application in Hugging Face Spaces and select the Python framework.
2. 2.Upload the following files:
3. 3.app.py: The main application program.
4. 4.requirements.txt: Dependency configuration.
5. 5..streamlit/config.toml (optional, used to configure parameters such as GPU).
6. 6.Configure environment variables (if necessary):
7. 7.HF_TOKEN: Hugging Face access token (used to load private models).

Usage Instructions

1. 1.Fill in the essay topic (Prompt) and essay body (Essay) in the input boxes respectively.
2. 2.Click the "Generate Assessment Report" button.
3. 3.The system will return:
4. 4.The overall score (based on Pipeline 1).
5. 5.Detailed dimension feedback (based on Pipeline 2).

Model Information

• —Pipeline 1: yitongwu73/finetuned-roberta-base A sequence classification model based on RoBERTa, used for regression tasks (continuous score prediction).
• —Pipeline 2: Noracle/finetuned-distilbart-cnn-12-6 A text generation model based on DistilBART, used for structured feedback generation.

Contact Us

• —Company: NewChannel International Education Group Limited
• —Official website: https://www.xhd.cn