Upload streamlit app to ask Groq about a PDF
Browse files- app.py +68 -0
- requirements.txt +10 -0
app.py
ADDED
@@ -0,0 +1,68 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
import os
|
3 |
+
|
4 |
+
from groq import Groq
|
5 |
+
from PyPDF2 import PdfReader
|
6 |
+
from langchain.text_splitter import RecursiveCharacterTextSplitter
|
7 |
+
from langchain_community.embeddings import HuggingFaceEmbeddings
|
8 |
+
#from langchain.vectorstores import FAISS
|
9 |
+
from langchain_community.vectorstores import FAISS
|
10 |
+
from langchain_groq import ChatGroq
|
11 |
+
#from langchain.chat_models import ChatOpenAI
|
12 |
+
from langchain.chains.question_answering import load_qa_chain
|
13 |
+
|
14 |
+
st.set_page_config('Lectorín')
|
15 |
+
st.header("Pregunta a tu PDF")
|
16 |
+
GROQ_API_KEY = st.text_input('Groq API Key', value="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", type='password')
|
17 |
+
pdf_obj = st.file_uploader("Carga tu documento", type="pdf", on_change=st.cache_resource.clear)
|
18 |
+
modelos = {
|
19 |
+
'multi, 512, 0.47G, 384 - intfloat/multilingual-e5-small': ('intfloat/multilingual-e5-small',512),
|
20 |
+
'multi, 256, 0.08G, 384 - multi-qa-MiniLM-L6-cos-v1': ('multi-qa-MiniLM-L6-cos-v1',256),
|
21 |
+
'multi,8192, 2.27G,1024 - BAAI/bge-m3': ('BAAI/bge-m3', 8192),
|
22 |
+
}
|
23 |
+
modelo = st.selectbox('Modelo de embedding', list(modelos.keys()))
|
24 |
+
modelo_embeddings, sequence = modelos[modelo]
|
25 |
+
chunk_size = sequence * 5 # en español, de media una palabra tiene 5 caracteres
|
26 |
+
|
27 |
+
modelos_llm = [
|
28 |
+
'llama3-70b-8192',
|
29 |
+
'llama3-8b-8192',
|
30 |
+
'mixtral-8x7b-32768',
|
31 |
+
'gemma-7b-it'
|
32 |
+
]
|
33 |
+
modelo_llm = st.selectbox('Modelo de lenguaje', list(modelos_llm))
|
34 |
+
|
35 |
+
@st.cache_resource
|
36 |
+
def create_embeddings(pdf):
|
37 |
+
pdf_reader = PdfReader(pdf)
|
38 |
+
text = ""
|
39 |
+
for page in pdf_reader.pages:
|
40 |
+
text += page.extract_text()
|
41 |
+
|
42 |
+
text_splitter = RecursiveCharacterTextSplitter(
|
43 |
+
chunk_size=chunk_size,
|
44 |
+
chunk_overlap=150,
|
45 |
+
length_function=len
|
46 |
+
)
|
47 |
+
|
48 |
+
chunks = text_splitter.split_text(text)
|
49 |
+
embeddings = HuggingFaceEmbeddings(model_name=modelo_embeddings)
|
50 |
+
knowledge_base = FAISS.from_texts(chunks, embeddings)
|
51 |
+
|
52 |
+
return knowledge_base
|
53 |
+
|
54 |
+
|
55 |
+
if pdf_obj:
|
56 |
+
knowledge_base = create_embeddings(pdf_obj)
|
57 |
+
user_question = st.text_input("Haz una pregunta sobre tu PDF:")
|
58 |
+
|
59 |
+
if user_question:
|
60 |
+
os.environ["GROQ_API_KEY"] = GROQ_API_KEY
|
61 |
+
#os.environ["OPENAI_API_KEY"] = OPENAI_API_KEY
|
62 |
+
docs = knowledge_base.similarity_search(user_question, 5)
|
63 |
+
llm = ChatGroq(groq_api_key = os.getenv('GROQ_API_KEY'),model = modelo_llm)
|
64 |
+
#llm = ChatOpenAI(model_name='gpt-3.5-turbo')
|
65 |
+
chain = load_qa_chain(llm, chain_type="stuff")
|
66 |
+
respuesta = chain.run(input_documents=docs, question=user_question)
|
67 |
+
|
68 |
+
st.write(respuesta)
|
requirements.txt
ADDED
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
groq
|
2 |
+
#openai
|
3 |
+
langchain
|
4 |
+
langchain-community
|
5 |
+
langchain_groq
|
6 |
+
PyPDF2
|
7 |
+
streamlit
|
8 |
+
sentence_transformers
|
9 |
+
faiss-cpu
|
10 |
+
#faiss-gpu
|