Upload app.py
Browse files
app.py
ADDED
@@ -0,0 +1,93 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from langchain.chains import RetrievalQA
|
2 |
+
from langchain import HuggingFaceHub
|
3 |
+
from langchain.prompts import PromptTemplate
|
4 |
+
from langchain.text_splitter import RecursiveCharacterTextSplitter
|
5 |
+
from langchain.embeddings.huggingface import HuggingFaceEmbeddings
|
6 |
+
from langchain.document_loaders import PyPDFLoader
|
7 |
+
from langchain.vectorstores import FAISS
|
8 |
+
from glob import glob
|
9 |
+
from tqdm import tqdm
|
10 |
+
import yaml
|
11 |
+
|
12 |
+
|
13 |
+
def load_config():
|
14 |
+
with open('config.yaml', 'r') as file:
|
15 |
+
config = yaml.safe_load(file)
|
16 |
+
return config
|
17 |
+
|
18 |
+
|
19 |
+
config = load_config()
|
20 |
+
|
21 |
+
|
22 |
+
def load_embeddings(model_name=config["embeddings"]["name"],
|
23 |
+
model_kwargs={'device': config["embeddings"]["device"]}):
|
24 |
+
return HuggingFaceEmbeddings(model_name=model_name, model_kwargs=model_kwargs)
|
25 |
+
|
26 |
+
|
27 |
+
def load_documents(directory: str):
|
28 |
+
"""Loads all documents from a directory and returns a list of Document objects
|
29 |
+
args: directory format = directory/
|
30 |
+
"""
|
31 |
+
text_splitter = RecursiveCharacterTextSplitter(chunk_size=config["TextSplitter"]["chunk_size"],
|
32 |
+
chunk_overlap=config["TextSplitter"]["chunk_overlap"])
|
33 |
+
documents = []
|
34 |
+
for item_path in tqdm(glob(directory + "*.pdf")):
|
35 |
+
loader = PyPDFLoader(item_path)
|
36 |
+
documents.extend(loader.load_and_split(text_splitter=text_splitter))
|
37 |
+
|
38 |
+
return documents
|
39 |
+
|
40 |
+
|
41 |
+
template = """Use the following pieces of context to answer the question at the end.
|
42 |
+
If you don't know the answer, just say that you don't know, don't try to make up an answer.
|
43 |
+
Use three sentences maximum and keep the answer as concise as possible.
|
44 |
+
Always say "thanks for asking!" at the end of the answer.
|
45 |
+
{context}
|
46 |
+
Question: {question}
|
47 |
+
Helpful Answer:"""
|
48 |
+
QA_CHAIN_PROMPT = PromptTemplate.from_template(template)
|
49 |
+
|
50 |
+
repo_id = "google/flan-t5-xxl"
|
51 |
+
|
52 |
+
|
53 |
+
def get_llm():
|
54 |
+
llm = HuggingFaceHub(
|
55 |
+
repo_id=repo_id, model_kwargs={"temperature": 0.5, "max_length": 200}
|
56 |
+
)
|
57 |
+
return llm
|
58 |
+
|
59 |
+
|
60 |
+
def answer_question(question: str):
|
61 |
+
embedding_function = load_embeddings()
|
62 |
+
documents = load_documents("data/")
|
63 |
+
|
64 |
+
db = FAISS.from_documents(documents, embedding_function)
|
65 |
+
|
66 |
+
retriever = db.as_retriever(search_type="similarity", search_kwargs={"k": 4})
|
67 |
+
|
68 |
+
qa_chain = RetrievalQA.from_chain_type(
|
69 |
+
get_llm(),
|
70 |
+
retriever=retriever,
|
71 |
+
chain_type="stuff",
|
72 |
+
chain_type_kwargs={"prompt": QA_CHAIN_PROMPT},
|
73 |
+
return_source_documents=True
|
74 |
+
)
|
75 |
+
|
76 |
+
output = qa_chain({"query": question})
|
77 |
+
return output["result"]
|
78 |
+
|
79 |
+
|
80 |
+
# Gradio UI for PDFChat
|
81 |
+
with gr.Blocks() as demo:
|
82 |
+
with gr.Tab("PdfChat"):
|
83 |
+
with gr.Row():
|
84 |
+
ans = gr.Textbox(label="Answer", lines=10)
|
85 |
+
|
86 |
+
que = gr.Textbox(label="Ask a Question", lines=3)
|
87 |
+
|
88 |
+
bttn = gr.Button(label="Submit")
|
89 |
+
|
90 |
+
bttn.click(fn=answer_question, inputs=[que], outputs=[ans])
|
91 |
+
|
92 |
+
if __name__ == "__main__":
|
93 |
+
demo.launch()
|