File size: 2,474 Bytes
59fc6ec 815128e a766494 815128e 59fc6ec 815128e 59fc6ec a766494 815128e 59fc6ec 815128e 59fc6ec 815128e 59fc6ec 815128e |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 |
# setting device on GPU if available, else CPU
import os
from timeit import default_timer as timer
from typing import List
from langchain.document_loaders import PyPDFDirectoryLoader
from langchain.embeddings import HuggingFaceInstructEmbeddings
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.vectorstores.chroma import Chroma
from app_modules.utils import *
def load_documents(source_pdfs_path) -> List:
loader = PyPDFDirectoryLoader(source_pdfs_path, silent_errors=True)
documents = loader.load()
return documents
def split_chunks(documents: List, chunk_size, chunk_overlap) -> List:
text_splitter = RecursiveCharacterTextSplitter(
chunk_size=chunk_size, chunk_overlap=chunk_overlap
)
return text_splitter.split_documents(documents)
def generate_index(chunks: List, embeddings: HuggingFaceInstructEmbeddings) -> Chroma:
chromadb_instructor_embeddings = Chroma.from_documents(
documents=chunks, embedding=embeddings, persist_directory=index_path
)
chromadb_instructor_embeddings.persist()
return chromadb_instructor_embeddings
# Constants
init_settings()
device_type, hf_pipeline_device_type = get_device_types()
hf_embeddings_model_name = (
os.environ.get("HF_EMBEDDINGS_MODEL_NAME") or "hkunlp/instructor-xl"
)
index_path = os.environ.get("CHROMADB_INDEX_PATH")
source_pdfs_path = os.environ.get("SOURCE_PDFS_PATH")
chunk_size = os.environ.get("CHUNCK_SIZE")
chunk_overlap = os.environ.get("CHUNK_OVERLAP")
start = timer()
embeddings = HuggingFaceInstructEmbeddings(
model_name=hf_embeddings_model_name, model_kwargs={"device": device_type}
)
end = timer()
print(f"Completed in {end - start:.3f}s")
start = timer()
if not os.path.isdir(index_path):
print("The index persist directory is not present. Creating a new one.")
os.mkdir(index_path)
print(f"Loading PDF files from {source_pdfs_path}")
sources = load_documents(source_pdfs_path)
print(f"Splitting {len(sources)} PDF pages in to chunks ...")
chunks = split_chunks(
sources, chunk_size=int(chunk_size), chunk_overlap=int(chunk_overlap)
)
print(f"Generating index for {len(chunks)} chunks ...")
index = generate_index(chunks, embeddings)
else:
print("The index persist directory is present. Loading index ...")
index = Chroma(embedding_function=embeddings, persist_directory=index_path)
end = timer()
print(f"Completed in {end - start:.3f}s")
|