Spaces:
Sleeping
Sleeping
Update app.py
#9
by Muthuraja18 - opened
app.py
CHANGED
|
@@ -1,11 +1,10 @@
|
|
| 1 |
import streamlit as st
|
| 2 |
-
import
|
| 3 |
|
| 4 |
from langchain_community.document_loaders import PyPDFLoader, TextLoader
|
| 5 |
from langchain_text_splitters import RecursiveCharacterTextSplitter
|
| 6 |
from langchain_community.embeddings import HuggingFaceEmbeddings
|
| 7 |
from langchain_community.vectorstores import FAISS
|
| 8 |
-
|
| 9 |
from langchain.chains import RetrievalQA
|
| 10 |
from langchain.prompts import PromptTemplate
|
| 11 |
from langchain_community.llms import HuggingFacePipeline
|
|
@@ -14,17 +13,18 @@ from transformers import pipeline
|
|
| 14 |
|
| 15 |
|
| 16 |
# -------------------------------
|
| 17 |
-
# Load Documents (
|
| 18 |
# -------------------------------
|
| 19 |
def load_documents(uploaded_files):
|
| 20 |
documents = []
|
| 21 |
|
| 22 |
for file in uploaded_files:
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
|
| 26 |
-
|
| 27 |
|
|
|
|
| 28 |
if file.name.endswith(".pdf"):
|
| 29 |
loader = PyPDFLoader(file_path)
|
| 30 |
else:
|
|
@@ -36,7 +36,7 @@ def load_documents(uploaded_files):
|
|
| 36 |
|
| 37 |
|
| 38 |
# -------------------------------
|
| 39 |
-
# Split Documents
|
| 40 |
# -------------------------------
|
| 41 |
def split_documents(documents):
|
| 42 |
splitter = RecursiveCharacterTextSplitter(
|
|
@@ -47,7 +47,7 @@ def split_documents(documents):
|
|
| 47 |
|
| 48 |
|
| 49 |
# -------------------------------
|
| 50 |
-
#
|
| 51 |
# -------------------------------
|
| 52 |
def create_vectorstore(chunks):
|
| 53 |
embeddings = HuggingFaceEmbeddings(
|
|
@@ -57,12 +57,12 @@ def create_vectorstore(chunks):
|
|
| 57 |
|
| 58 |
|
| 59 |
# -------------------------------
|
| 60 |
-
# LLM (
|
| 61 |
# -------------------------------
|
| 62 |
def load_llm():
|
| 63 |
pipe = pipeline(
|
| 64 |
"text2text-generation",
|
| 65 |
-
model="google/flan-t5-small", #
|
| 66 |
max_length=512,
|
| 67 |
temperature=0.3
|
| 68 |
)
|
|
@@ -70,7 +70,7 @@ def load_llm():
|
|
| 70 |
|
| 71 |
|
| 72 |
# -------------------------------
|
| 73 |
-
#
|
| 74 |
# -------------------------------
|
| 75 |
def build_qa(vectorstore):
|
| 76 |
llm = load_llm()
|
|
@@ -101,13 +101,13 @@ def build_qa(vectorstore):
|
|
| 101 |
|
| 102 |
|
| 103 |
# -------------------------------
|
| 104 |
-
# UI
|
| 105 |
# -------------------------------
|
| 106 |
st.set_page_config(page_title="RAG Chatbot", layout="wide")
|
| 107 |
st.title("π Chat with Your Documents (RAG)")
|
| 108 |
|
| 109 |
uploaded_files = st.file_uploader(
|
| 110 |
-
"Upload PDF or TXT files",
|
| 111 |
accept_multiple_files=True
|
| 112 |
)
|
| 113 |
|
|
@@ -118,13 +118,15 @@ if uploaded_files:
|
|
| 118 |
vectorstore = create_vectorstore(chunks)
|
| 119 |
qa_chain = build_qa(vectorstore)
|
| 120 |
|
| 121 |
-
st.success("β
Documents
|
| 122 |
|
| 123 |
query = st.text_input("Ask a question from your documents")
|
| 124 |
|
| 125 |
if query:
|
| 126 |
with st.spinner("Thinking..."):
|
| 127 |
-
|
| 128 |
-
|
| 129 |
-
|
| 130 |
-
|
|
|
|
|
|
|
|
|
| 1 |
import streamlit as st
|
| 2 |
+
import tempfile
|
| 3 |
|
| 4 |
from langchain_community.document_loaders import PyPDFLoader, TextLoader
|
| 5 |
from langchain_text_splitters import RecursiveCharacterTextSplitter
|
| 6 |
from langchain_community.embeddings import HuggingFaceEmbeddings
|
| 7 |
from langchain_community.vectorstores import FAISS
|
|
|
|
| 8 |
from langchain.chains import RetrievalQA
|
| 9 |
from langchain.prompts import PromptTemplate
|
| 10 |
from langchain_community.llms import HuggingFacePipeline
|
|
|
|
| 13 |
|
| 14 |
|
| 15 |
# -------------------------------
|
| 16 |
+
# Load Documents (FIXED - NO 403)
|
| 17 |
# -------------------------------
|
| 18 |
def load_documents(uploaded_files):
|
| 19 |
documents = []
|
| 20 |
|
| 21 |
for file in uploaded_files:
|
| 22 |
+
# β
SAFE TEMP FILE (main fix)
|
| 23 |
+
with tempfile.NamedTemporaryFile(delete=False) as tmp_file:
|
| 24 |
+
tmp_file.write(file.read())
|
| 25 |
+
file_path = tmp_file.name
|
| 26 |
|
| 27 |
+
# Load document
|
| 28 |
if file.name.endswith(".pdf"):
|
| 29 |
loader = PyPDFLoader(file_path)
|
| 30 |
else:
|
|
|
|
| 36 |
|
| 37 |
|
| 38 |
# -------------------------------
|
| 39 |
+
# Split Documents
|
| 40 |
# -------------------------------
|
| 41 |
def split_documents(documents):
|
| 42 |
splitter = RecursiveCharacterTextSplitter(
|
|
|
|
| 47 |
|
| 48 |
|
| 49 |
# -------------------------------
|
| 50 |
+
# Create Vector Store
|
| 51 |
# -------------------------------
|
| 52 |
def create_vectorstore(chunks):
|
| 53 |
embeddings = HuggingFaceEmbeddings(
|
|
|
|
| 57 |
|
| 58 |
|
| 59 |
# -------------------------------
|
| 60 |
+
# Load LLM (LIGHT + NO TOKEN)
|
| 61 |
# -------------------------------
|
| 62 |
def load_llm():
|
| 63 |
pipe = pipeline(
|
| 64 |
"text2text-generation",
|
| 65 |
+
model="google/flan-t5-small", # best balance
|
| 66 |
max_length=512,
|
| 67 |
temperature=0.3
|
| 68 |
)
|
|
|
|
| 70 |
|
| 71 |
|
| 72 |
# -------------------------------
|
| 73 |
+
# Build QA Chain (Better Prompt)
|
| 74 |
# -------------------------------
|
| 75 |
def build_qa(vectorstore):
|
| 76 |
llm = load_llm()
|
|
|
|
| 101 |
|
| 102 |
|
| 103 |
# -------------------------------
|
| 104 |
+
# Streamlit UI
|
| 105 |
# -------------------------------
|
| 106 |
st.set_page_config(page_title="RAG Chatbot", layout="wide")
|
| 107 |
st.title("π Chat with Your Documents (RAG)")
|
| 108 |
|
| 109 |
uploaded_files = st.file_uploader(
|
| 110 |
+
"Upload PDF or TXT files (Max ~10MB recommended)",
|
| 111 |
accept_multiple_files=True
|
| 112 |
)
|
| 113 |
|
|
|
|
| 118 |
vectorstore = create_vectorstore(chunks)
|
| 119 |
qa_chain = build_qa(vectorstore)
|
| 120 |
|
| 121 |
+
st.success("β
Documents processed successfully!")
|
| 122 |
|
| 123 |
query = st.text_input("Ask a question from your documents")
|
| 124 |
|
| 125 |
if query:
|
| 126 |
with st.spinner("Thinking..."):
|
| 127 |
+
try:
|
| 128 |
+
result = qa_chain.run(query)
|
| 129 |
+
st.write("### π Answer:")
|
| 130 |
+
st.write(result)
|
| 131 |
+
except Exception as e:
|
| 132 |
+
st.error(f"Error: {str(e)}")
|