my_agent/utils/data_loader.py

# from datasets import load_dataset

# def load_influencer_data():
#     dataset = load_dataset("subashdvorak/tiktok-agentic-story",revision="embedded")
#     data= dataset['train'].add_faiss_index('embeddings')
#     return data


from datasets import load_dataset

print("Loading dataset and indexing FAISS...")  # Optional: for debugging

dataset = load_dataset("subashdvorak/tiktok-agentic-story", revision="embedded")
data = dataset['train'].add_faiss_index('embeddings')

def load_influencer_data():
    return data