tensorsoft/Mini-RAG-Chat-With-Your-Files-CPU-Only

1import numpy as np2 3def chunk_text(text: str, chunk_size: int = 200):4    """Split text into chunks of specified size."""5    words = text.split()6    chunks = []7    current_chunk = []8    current_length = 09    10    for word in words:11        current_length += len(word) + 112        if current_length > chunk_size:13            chunks.append(" ".join(current_chunk))14            current_chunk = [word]15            current_length = len(word) + 116        else:17            current_chunk.append(word)18    19    if current_chunk:20        chunks.append(" ".join(current_chunk))21    22    return chunks23 24def cosine_similarity(vec_a: np.ndarray, vec_b: np.ndarray):25    """Compute cosine similarity between two vectors."""26    dot_product = np.dot(vec_a, vec_b)27    norm_a = np.linalg.norm(vec_a)28    norm_b = np.linalg.norm(vec_b)29    return dot_product / (norm_a * norm_b) if norm_a * norm_b != 0 else 030