Update app.py
Browse files
app.py
CHANGED
|
@@ -19,6 +19,12 @@ model_name = "model-q4_K.gguf"
|
|
| 19 |
|
| 20 |
#snapshot_download(repo_id=repo_name, local_dir=".", allow_patterns=model_name)
|
| 21 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 22 |
|
| 23 |
def get_pdf_text(pdf_docs):
|
| 24 |
text = ""
|
|
@@ -50,6 +56,7 @@ def get_text_chunks(text):
|
|
| 50 |
|
| 51 |
#return vectorstore
|
| 52 |
|
|
|
|
| 53 |
def get_vectorstore(text_chunks, embedding_model_name="intfloat/multilingual-e5-large"):
|
| 54 |
embeddings = HuggingFaceEmbeddings(model_name=embedding_model_name)
|
| 55 |
vectorstore = FAISS.from_texts(texts=text_chunks, embedding=embeddings)
|
|
@@ -78,24 +85,28 @@ def get_conversation_chain(vectorstore, model_name):
|
|
| 78 |
return conversation_chain
|
| 79 |
|
| 80 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 81 |
def handle_userinput(user_question):
|
| 82 |
-
|
| 83 |
response = st.session_state.conversation({'question': user_question})
|
| 84 |
|
| 85 |
st.session_state.chat_history = response['chat_history']
|
| 86 |
-
|
| 87 |
st.session_state.retrieved_text = response['source_documents']
|
| 88 |
|
| 89 |
for i, (message, text) in enumerate(zip(st.session_state.chat_history, st.session_state.retrieved_text)):
|
| 90 |
-
if i %
|
| 91 |
-
st.write(user_template.replace(
|
| 92 |
-
|
| 93 |
-
|
| 94 |
-
|
| 95 |
-
|
| 96 |
-
|
| 97 |
-
|
| 98 |
-
"{{MSG}}",
|
|
|
|
| 99 |
|
| 100 |
|
| 101 |
st.set_page_config(page_title="Chat with multiple PDFs",
|
|
@@ -116,6 +127,7 @@ if user_question:
|
|
| 116 |
with st.sidebar:
|
| 117 |
st.subheader("Your documents")
|
| 118 |
embedding_model_name = st.selectbox("Select embedding model", ["intfloat/multilingual-e5-large", "sentence-transformers/paraphrase-multilingual-mpnet-base-v2"])
|
|
|
|
| 119 |
pdf_docs = st.file_uploader(
|
| 120 |
"Upload your PDFs here and click on 'Process'", accept_multiple_files=True)
|
| 121 |
if st.button("Process"):
|
|
|
|
| 19 |
|
| 20 |
#snapshot_download(repo_id=repo_name, local_dir=".", allow_patterns=model_name)
|
| 21 |
|
| 22 |
+
from transformers import pipeline
|
| 23 |
+
|
| 24 |
+
# Initialize the summarization pipeline
|
| 25 |
+
summarizer = pipeline("summarization")
|
| 26 |
+
|
| 27 |
+
|
| 28 |
|
| 29 |
def get_pdf_text(pdf_docs):
|
| 30 |
text = ""
|
|
|
|
| 56 |
|
| 57 |
#return vectorstore
|
| 58 |
|
| 59 |
+
|
| 60 |
def get_vectorstore(text_chunks, embedding_model_name="intfloat/multilingual-e5-large"):
|
| 61 |
embeddings = HuggingFaceEmbeddings(model_name=embedding_model_name)
|
| 62 |
vectorstore = FAISS.from_texts(texts=text_chunks, embedding=embeddings)
|
|
|
|
| 85 |
return conversation_chain
|
| 86 |
|
| 87 |
|
| 88 |
+
def summarize_text(text):
|
| 89 |
+
summary = summarizer(text, max_length=130, min_length=30, do_sample=False)
|
| 90 |
+
return summary[0]['summary_text']
|
| 91 |
+
|
| 92 |
+
|
| 93 |
def handle_userinput(user_question):
|
|
|
|
| 94 |
response = st.session_state.conversation({'question': user_question})
|
| 95 |
|
| 96 |
st.session_state.chat_history = response['chat_history']
|
|
|
|
| 97 |
st.session_state.retrieved_text = response['source_documents']
|
| 98 |
|
| 99 |
for i, (message, text) in enumerate(zip(st.session_state.chat_history, st.session_state.retrieved_text)):
|
| 100 |
+
if i % 2 == 0: # User messages
|
| 101 |
+
st.write(user_template.replace("{{MSG}}", message.content), unsafe_allow_html=True)
|
| 102 |
+
else: # Bot messages
|
| 103 |
+
st.write(bot_template.replace("{{MSG}}", message.content), unsafe_allow_html=True)
|
| 104 |
+
if summarize_option and text.page_content: # Check if summarization is enabled
|
| 105 |
+
summarized_text = summarize_text(text.page_content)
|
| 106 |
+
st.write(bot_template.replace("{{MSG}}", summarized_text), unsafe_allow_html=True)
|
| 107 |
+
else:
|
| 108 |
+
st.write(bot_template.replace("{{MSG}}", text.page_content), unsafe_allow_html=True)
|
| 109 |
+
|
| 110 |
|
| 111 |
|
| 112 |
st.set_page_config(page_title="Chat with multiple PDFs",
|
|
|
|
| 127 |
with st.sidebar:
|
| 128 |
st.subheader("Your documents")
|
| 129 |
embedding_model_name = st.selectbox("Select embedding model", ["intfloat/multilingual-e5-large", "sentence-transformers/paraphrase-multilingual-mpnet-base-v2"])
|
| 130 |
+
summarize_option = st.sidebar.checkbox("Enable Summarization", value=False)
|
| 131 |
pdf_docs = st.file_uploader(
|
| 132 |
"Upload your PDFs here and click on 'Process'", accept_multiple_files=True)
|
| 133 |
if st.button("Process"):
|