DanielHafezi/essayevaluator

Essay Evaluator - Finetuned Phi-3.5-mini-instruct

This model is a fine-tuned version of microsoft/Phi-3.5-mini-instruct, specifically trained for evaluating academic essays and providing numerical scores on a scale of 1-6.

Model Description

• Model Architecture: Based on Phi-3.5-mini-instruct (3.8B parameters)
• Training Focus: Essay evaluation and scoring
• Input Format: Academic essays in English
• Output Format: Numerical score (1-6)
• Fine-tuning Focus: Holistic essay assessment considering:
  • Content quality
  • Organization
  • Language use

Intended Use

• Primary Use: Automated essay scoring in educational contexts
• Target Users:
  • Educational institutions
  • Teachers
  • Students (for self-assessment)
• Scope: English academic essays

Training Details

• Base Model: microsoft/Phi-3.5-mini-instruct
• Training Type: Supervised fine-tuning
• Training Data: Balanced dataset of scored academic essays
• Parameters:
  • Learning rate: 2e-5
  • Epochs: 2
  • Batch size: 1

Performance and Limitations

Strengths

• Consistent scoring across similar essays
• Fast evaluation time
• Structured numerical output

Limitations

• Limited to English language essays
• Should be used as an assistive tool, not a replacement for human grading
• Best suited for academic essay formats

Example Usage

# API request format
headers = {
    "Content-Type": "application/json",
    "Authorization": f"Bearer {API_KEY}"
}

data = {
    "messages": [
        {
            "role": "system",
            "content": "You are an experienced English teacher specializing in grading high school student essays. Read the following essay carefully and provide a holistic score from 1 to 6 based on content, organization, and language use. Provide only the numerical score."
        },
        {
            "role": "user",
            "content": "YOUR_ESSAY_TEXT"
        }
    ]
}