scripts/fix_results.py

import json
import os
from dotenv import load_dotenv
from huggingface_hub import HfApi

# Load environment variables
load_dotenv()

# Configuration
HF_TOKEN = os.getenv("HF_TOKEN")
RESULTS_REPO = "hamzabouajila/results"

# Read the original results file
def read_results_file(file_path):
    with open(file_path, 'r') as f:
        return json.load(f)

# Fix the results format
def fix_results_format(results):
    # Fix null accuracy
    if results['results'].get('accuracy') is None:
        results['results']['accuracy'] = 0.0  # Replace with actual accuracy if known
    
    # Fix model_type format
    results['model_type'] = results['model_type'].replace('\ud83d\udfe2 : ', '').strip()
    
    # Convert params to integer if needed
    if isinstance(results.get('params'), float):
        results['params'] = int(results['params'] * 1000000)  # Convert to millions
    
    return results

# Upload to Hugging Face
def upload_to_hf(results, file_name):
    api = HfApi(token=HF_TOKEN)
    try:
        api.upload_file(
            path_or_fileobj=file_name,
            path_in_repo=os.path.basename(file_name),
            repo_id=RESULTS_REPO,
            repo_type="dataset",
            commit_message=f"Add evaluation results for {results['model']}"
        )
        print(f"Successfully uploaded to Hugging Face")
        return True
    except Exception as e:
        print(f"Error uploading to Hugging Face: {str(e)}")
        return False

if __name__ == "__main__":
    # Original file path
    original_file = "/teamspace/studios/this_studio/TunisianLeaderBoard/eval-results/tunis-ai/TunBERT_eval_request_False_float16_Original.json"
    
    # Read and fix the results
    results = read_results_file(original_file)
    fixed_results = fix_results_format(results)
    
    # Save the fixed version
    fixed_file = "/teamspace/studios/this_studio/TunisianLeaderBoard/eval-results/tunis-ai/TunBERT_eval_request_False_float16_Original_fixed.json"
    with open(fixed_file, 'w') as f:
        json.dump(fixed_results, f, indent=2)
    
    print(f"Fixed results saved to: {fixed_file}")
    
    # Try to upload to Hugging Face
    if HF_TOKEN:
        upload_to_hf(fixed_results, fixed_file)
    else:
        print("No HF_TOKEN found. Skipping Hugging Face upload.")