ai-essay-evaluator 1.3.2

This repository provides a modular framework for generating, validating, merging, uploading, and fine-tuning OpenAI GPT-4o-mini models using structured JSONL datasets.

AI Essay Evaluator

---

Documentation: https://ai-essay-evaluator.readthedocs.io

Source Code: https://github.com/markm-io/ai-essay-evaluator

---

A comprehensive Python framework for automated essay evaluation using OpenAI's GPT models. This tool enables educators to grade student essays at scale with customizable scoring rubrics, fine-tune models with their own grading data, and generate detailed feedback across multiple scoring dimensions.

Features

• Automated Essay Grading - Evaluate student essays using fine-tuned OpenAI GPT-4o-mini models
• Multiple Scoring Formats - Choose from extended (multi-dimensional), item-specific, or short scoring formats
• Custom Model Training - Generate training datasets and fine-tune models with your own grading examples
• Project Folder Mode - Simple folder structure for organizing essays, rubrics, and prompts
• Cost Tracking - Built-in token usage and cost analysis for OpenAI API calls
• Batch Processing - Grade hundreds of essays with progress tracking and async processing
• Multi-Pass Grading - Run multiple grading passes for consistency checking
• Rate Limit Handling - Automatic retry logic and adaptive rate limiting
• Comprehensive Logging - Async logging for debugging and auditing

Quick Start

Installation

Install via pip:

pip install ai-essay-evaluator

Or using uv (recommended for development):

uv pip install ai-essay-evaluator

Basic Usage

1. Set up your project folder:

my_project/
├── input.csv              # Student responses
├── question.txt           # Essay prompt
├── story/                 # Story files
│   └── story1.txt
└── rubric/                # Rubric files
    └── rubric1.txt

2. Run the evaluator:

python -m ai_essay_evaluator evaluator grader \
  --project-folder ./my_project \
  --scoring-format extended \
  --api-key YOUR_OPENAI_API_KEY

3. Check results in my_project/output/

Training Your Own Model

# Generate training data from graded examples
python -m ai_essay_evaluator trainer generate \
  --story-folder ./training/story \
  --question ./training/question.txt \
  --rubric ./training/rubric.txt \
  --csv ./training/graded_samples.csv \
  --output training.jsonl \
  --scoring-format extended

# Validate and fine-tune
python -m ai_essay_evaluator trainer validate --file training.jsonl
python -m ai_essay_evaluator trainer fine-tune \
  --file training.jsonl \
  --scoring-format extended \
  --api-key YOUR_OPENAI_API_KEY

For detailed documentation, visit the full usage guide.

Contributors ✨

Thanks goes to these wonderful people (emoji key):

Mark Moreno 💻 🤔 📖

This project follows the all-contributors specification. Contributions of any kind welcome!

Credits

This package was created with Copier and the browniebroke/pypackage-template project template.