grid_path/eval.py

import json

import argparse
if __name__ == "__main__":
    # Parse command line arguments
    parser = argparse.ArgumentParser(description="Evaluation script for comparing_size task")
    parser.add_argument(
        '--answer', '-a',
        type=str,
        default='answer_gpt4o.json',
        help='Path to the answer JSON file (default: answer.json)'
    )
    parser.add_argument(
        '--output', '-o',
        type=str,
        default='eval_gpt4o.json',
        help='Path to the output JSON file (default: eval.json)'
    )
    args = parser.parse_args()
    # Assuming your JSON data is stored in a file called 'results.json'
    with open(args.answer, 'r') as f:
        data = json.load(f)

    # Initialize variables to calculate accuracies
    correct_counts = 0
    total_counts = 0
    category_accuracies = {}

    # Iterate through the JSON data

    for entry in data:
        grid_size = entry['rows']
        num_circles = entry['path_size']-2
        total_counts += 1

        # Calculate per-category accuracy
        #if grid_size not in category_accuracies:
        #    category_accuracies[grid_size] = {'correct': 0  , 'total': 0}
        
        if num_circles not in category_accuracies:
            category_accuracies[num_circles] = {'correct': 0, 'total': 0}

        category_accuracies[num_circles]['total'] += 1
        #category_accuracies['total'] += 1
        # category_accuracies[(grid_size,num_circles)]['total'] += 1

        if entry["ERROR"]:
            continue

        # Check if the output is correct
        output = entry["Output"]

        clean_output = [s.lower() for s in output]
        clean_gold_output = [s.lower() for s in entry["gold_output"]]

        if clean_output == clean_gold_output:
            correct_counts += 1
            category_accuracies[num_circles]['correct'] += 1
    # Calculate overall accuracy
    overall_accuracy = correct_counts / total_counts * 100

    category_accuracy_percentages = {
        k: (v['correct'] / v['total'] * 100) for k, v in category_accuracies.items()
    }

    # Prepare results for saving
    eval_results = {
        "Overall Accuracy": overall_accuracy,
        "Category-wise Accuracy": category_accuracy_percentages
    }

    # Save results to eval.json
    with open(args.output, 'w') as eval_file:
        json.dump(eval_results, eval_file, indent=4)

    print("Evaluation results saved to eval.json.")

"""import json

import argparse
if __name__ == "__main__":
    # Parse command line arguments
    parser = argparse.ArgumentParser(description="Evaluation script for comparing_size task")
    parser.add_argument(
        '--answer', '-a',
        type=str,
        default='answer_gpt4o.json',
        help='Path to the answer JSON file (default: answer.json)'
    )
    parser.add_argument(
        '--output', '-o',
        type=str,
        default='eval_gpt4o.json',
        help='Path to the output JSON file (default: eval.json)'
    )
    args = parser.parse_args()
    # Assuming your JSON data is stored in a file called 'results.json'
    with open(args.answer, 'r') as f:
        data = json.load(f)

    # Initialize variables to calculate accuracies
    correct_counts = 0
    total_counts = 0
    category_accuracies = {}

    # Iterate through the JSON data
    for entry in data:
        grid_size = entry['rows']
        path_size = entry['path_size']
        total_counts += 1

        # Calculate per-category accuracy
        if grid_size not in category_accuracies:
            category_accuracies[grid_size] = {'correct': 0  , 'total': 0}
        
        if path_size not in category_accuracies[grid_size]:
            category_accuracies[grid_size][path_size] = {'correct': 0, 'total': 0}

        category_accuracies[grid_size][path_size]['total'] += 1
        category_accuracies[grid_size]['total'] += 1
        # category_accuracies[(grid_size,path_size)]['total'] += 1

        if entry["ERROR"]:
            continue

        # Check if the output is correct
        output = entry["Output"]
        clean_output = [s.lower() for s in output]
        clean_gold_output = [s.lower() for s in entry["gold_output"]]

        if clean_output == clean_gold_output:
            correct_counts += 1
            category_accuracies[grid_size]['correct'] += 1
            category_accuracies[grid_size][path_size]['correct'] += 1
        
    # Calculate overall accuracy
    overall_accuracy = correct_counts / total_counts * 100

    for grid_size in category_accuracies:
        category_accuracies[grid_size]["Accuracy"] = category_accuracies[grid_size]['correct'] / category_accuracies[grid_size]['total'] * 100
        for num_queens in category_accuracies[grid_size]:
            if num_queens == 'correct' or num_queens == 'total' or num_queens == 'Accuracy':
                continue
            # print(category_accuracies[grid_size][num_queens])
            category_accuracies[grid_size][num_queens]["Accuracy"] = category_accuracies[grid_size][num_queens]['correct'] / category_accuracies[grid_size][num_queens]['total'] * 100
        

            

    # category_accuracies["Overall Accuracy"] = overall_accuracy

    final_results = {}
    for grid_size in category_accuracies:
        # print(grid_size)    
        final_results[grid_size] = {}
        final_results[grid_size]["Overall Accuracy"] = category_accuracies[grid_size]["Accuracy"]
        final_results[grid_size]["Category-wise Accuracy"] = {}
        for num_crosses in category_accuracies[grid_size]:
            if num_crosses == 'correct' or num_crosses == 'total' or num_crosses == 'Accuracy':
                continue
            final_results[grid_size]["Category-wise Accuracy"][num_crosses] = category_accuracies[grid_size][num_crosses]["Accuracy"]

    final_results["Overall Accuracy"] = overall_accuracy


    # Save results to eval.json
    with open(args.output, 'w') as eval_file:
        json.dump(final_results, eval_file, indent=4)

    print("Evaluation results saved to eval.json.")"""