Spaces:
Sleeping
Sleeping
Commit
β’
f39f2e1
1
Parent(s):
01ccd2f
Upload 5 files
Browse files- .gitignore +14 -0
- app.py +51 -0
- pages/1_using_LLM.py +41 -0
- pages/2_using_LLM_QA.py +50 -0
- requirements.txt +0 -0
.gitignore
ADDED
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
__pycache__
|
2 |
+
.mypy_cache
|
3 |
+
|
4 |
+
data/
|
5 |
+
credential/
|
6 |
+
artifacts/
|
7 |
+
model/
|
8 |
+
.streamlit/
|
9 |
+
.streamlit/secrets.toml
|
10 |
+
*.toml
|
11 |
+
|
12 |
+
# ignore cache
|
13 |
+
*.pyc
|
14 |
+
|
app.py
ADDED
@@ -0,0 +1,51 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
import torch
|
3 |
+
from transformers import pipeline
|
4 |
+
from utils.process_data import generate_chunks, pdf_to_text
|
5 |
+
|
6 |
+
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
|
7 |
+
|
8 |
+
st.set_page_config(page_title="Summarizer", page_icon="βοΈ")
|
9 |
+
st.title("Summarize Text")
|
10 |
+
st.subheader("π π Transformers Summarization Pipeline")
|
11 |
+
|
12 |
+
max = st.slider('Select max', 50, 500, step=10, value=150)
|
13 |
+
min = st.slider('Select min', 10, 450, step=10, value=50)
|
14 |
+
do_sample = st.checkbox("Do sample", value=False)
|
15 |
+
|
16 |
+
sentence = st.text_area('Please paste your article:', height=50)
|
17 |
+
button = st.button("Summarize")
|
18 |
+
|
19 |
+
@st.cache_data
|
20 |
+
def load_summarizer():
|
21 |
+
model = pipeline("summarization", model=st.secrets["SUM_MODEL"], device=device)
|
22 |
+
return model
|
23 |
+
|
24 |
+
with st.spinner("Generating Summary.."):
|
25 |
+
if button and sentence:
|
26 |
+
chunks = generate_chunks(sentence)
|
27 |
+
summarizer = load_summarizer()
|
28 |
+
res = summarizer(chunks,
|
29 |
+
max_length=max,
|
30 |
+
min_length=min,
|
31 |
+
do_sample=do_sample)
|
32 |
+
text = ' '.join([summ['summary_text'] for summ in res])
|
33 |
+
st.write(text)
|
34 |
+
|
35 |
+
st.divider()
|
36 |
+
|
37 |
+
st.subheader('ππ Summarize PDF')
|
38 |
+
pdf_path = st.file_uploader('Upload your PDF Document', type='pdf')
|
39 |
+
button2 = st.button("Summarize PDF")
|
40 |
+
|
41 |
+
if pdf_path is not None and button2:
|
42 |
+
text = pdf_to_text(pdf_path)
|
43 |
+
with st.spinner("Generating PDF Summary.."):
|
44 |
+
chunks = generate_chunks(text)
|
45 |
+
summarizer = load_summarizer()
|
46 |
+
res = summarizer(chunks,
|
47 |
+
max_length=max,
|
48 |
+
min_length=min,
|
49 |
+
do_sample=do_sample)
|
50 |
+
text_sum = ' '.join([summ['summary_text'] for summ in res])
|
51 |
+
st.write(text_sum)
|
pages/1_using_LLM.py
ADDED
@@ -0,0 +1,41 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
from langchain import OpenAI
|
3 |
+
from langchain.docstore.document import Document
|
4 |
+
from langchain.text_splitter import CharacterTextSplitter
|
5 |
+
from langchain.chains.summarize import load_summarize_chain
|
6 |
+
from utils.process_data import pdf_to_text
|
7 |
+
|
8 |
+
MODEL = st.secrets["MODEL4"]
|
9 |
+
|
10 |
+
def generate_response(txt):
|
11 |
+
llm = OpenAI(temperature=0.1, openai_api_key=st.secrets["OPENAI_API_KEY"])
|
12 |
+
text_splitter = CharacterTextSplitter()
|
13 |
+
texts = text_splitter.split_text(txt)
|
14 |
+
docs = [Document(page_content=t) for t in texts]
|
15 |
+
chain = load_summarize_chain(llm, chain_type='map_reduce')
|
16 |
+
return chain.run(docs)
|
17 |
+
|
18 |
+
|
19 |
+
st.set_page_config(page_title="Summarizer with LLM", page_icon="βοΈ")
|
20 |
+
st.title("Summarize Text")
|
21 |
+
st.subheader('ππ LLM/LoadSummarizeChain')
|
22 |
+
sentence = st.text_area('Please paste your article:', height=100)
|
23 |
+
button = st.button("Summarize")
|
24 |
+
|
25 |
+
with st.spinner("Generating Summary.."):
|
26 |
+
if button and sentence:
|
27 |
+
response = generate_response(sentence)
|
28 |
+
st.write(response)
|
29 |
+
|
30 |
+
st.divider()
|
31 |
+
|
32 |
+
st.subheader('ππ Summarize PDF')
|
33 |
+
pdf_path = st.file_uploader('Upload your PDF Document', type='pdf')
|
34 |
+
button2 = st.button("Summarize PDF")
|
35 |
+
|
36 |
+
if pdf_path is not None and button2:
|
37 |
+
text = pdf_to_text(pdf_path)
|
38 |
+
with st.spinner("Generating PDF Summary.."):
|
39 |
+
response2 = generate_response(text)
|
40 |
+
st.subheader('Summary Results:')
|
41 |
+
st.write(response2)
|
pages/2_using_LLM_QA.py
ADDED
@@ -0,0 +1,50 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
from langchain_community.chat_models import ChatOpenAI
|
3 |
+
from langchain_community.callbacks import get_openai_callback
|
4 |
+
from langchain.chains.question_answering import load_qa_chain
|
5 |
+
|
6 |
+
from utils.process_data import process_text, pdf_to_text
|
7 |
+
|
8 |
+
MODEL = st.secrets["MODEL4"]
|
9 |
+
|
10 |
+
st.set_page_config(page_title="Summarizer with LLM QA", page_icon="βοΈ")
|
11 |
+
st.title("Summarize Text")
|
12 |
+
st.subheader("π π LLM/Question Answering")
|
13 |
+
|
14 |
+
maxw = st.slider('MAX words', 50, 1000, step=10, value=200)
|
15 |
+
minw = st.slider('MIN words', 10, 500, step=10, value=50)
|
16 |
+
|
17 |
+
sentence = st.text_area('Please paste your article:', height=50)
|
18 |
+
button = st.button("Summarize")
|
19 |
+
query = f"Summarize the content of the uploaded PDF file in more that {minw} words and less than {maxw} words. Focus on capturing the main ideas and key points discussed in the document. Use your own words and ensure clarity and coherence in the summary."
|
20 |
+
|
21 |
+
with st.spinner("Generating Summary.."):
|
22 |
+
if button and sentence:
|
23 |
+
knowledgeBase = process_text(sentence)
|
24 |
+
docs = knowledgeBase.similarity_search(query)
|
25 |
+
llm = ChatOpenAI(model=MODEL, temperature=0.1, openai_api_key=st.secrets["OPENAI_API_KEY"])
|
26 |
+
chain = load_qa_chain(llm, chain_type='stuff')
|
27 |
+
with get_openai_callback() as cost:
|
28 |
+
response = chain.run(input_documents=docs, question=query)
|
29 |
+
print(cost)
|
30 |
+
st.subheader('Summary Results:')
|
31 |
+
st.write(response)
|
32 |
+
|
33 |
+
st.divider()
|
34 |
+
|
35 |
+
st.subheader('ππ Summarize PDF')
|
36 |
+
pdf_path = st.file_uploader('Upload your PDF Document', type='pdf')
|
37 |
+
button2 = st.button("Summarize PDF")
|
38 |
+
|
39 |
+
if pdf_path is not None and button2:
|
40 |
+
text = pdf_to_text(pdf_path)
|
41 |
+
knowledgeBase = process_text(text)
|
42 |
+
with st.spinner("Generating PDF Summary.."):
|
43 |
+
docs = knowledgeBase.similarity_search(query)
|
44 |
+
llm = ChatOpenAI(model=MODEL, temperature=0.1, openai_api_key=st.secrets["OPENAI_API_KEY"])
|
45 |
+
chain = load_qa_chain(llm, chain_type='stuff')
|
46 |
+
with get_openai_callback() as cost:
|
47 |
+
response2 = chain.run(input_documents=docs, question=query)
|
48 |
+
print(cost)
|
49 |
+
st.subheader('Summary Results:')
|
50 |
+
st.write(response2)
|
requirements.txt
ADDED
Binary file (23.6 kB). View file
|
|