Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,77 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
import os
|
3 |
+
from langchain import PromptTemplate, LLMChain
|
4 |
+
from langchain_together import Together
|
5 |
+
import pdfplumber
|
6 |
+
|
7 |
+
# Set the API key with double quotes
|
8 |
+
os.environ['TOGETHER_API_KEY'] = "d88cb7414e4039a84d2ed63f1b47daaaa4230c4c53a422045d8a30a9a3bc87d8"
|
9 |
+
|
10 |
+
def extract_text_from_pdf(pdf_file, max_pages=16):
|
11 |
+
text = ""
|
12 |
+
with pdfplumber.open(pdf_file) as pdf:
|
13 |
+
for i, page in enumerate(pdf.pages):
|
14 |
+
if i >= max_pages:
|
15 |
+
break
|
16 |
+
text += page.extract_text() + "\n"
|
17 |
+
return text
|
18 |
+
|
19 |
+
def Bot(text, question):
|
20 |
+
chat_template = """
|
21 |
+
Based on the provided context: {text}
|
22 |
+
Please answer the following question: {Questions}
|
23 |
+
Only provide answers that are directly related to the context. If the question is unrelated, respond with "I don't know".
|
24 |
+
"""
|
25 |
+
prompt = PromptTemplate(
|
26 |
+
input_variables=['text', 'Questions'],
|
27 |
+
template=chat_template
|
28 |
+
)
|
29 |
+
llama3 = Together(model="meta-llama/Llama-3-70b-chat-hf", max_tokens=50)
|
30 |
+
Generated_chat = LLMChain(llm=llama3, prompt=prompt)
|
31 |
+
|
32 |
+
try:
|
33 |
+
response = Generated_chat.invoke({
|
34 |
+
"text": text,
|
35 |
+
"Questions": question
|
36 |
+
})
|
37 |
+
|
38 |
+
response_text = response['text']
|
39 |
+
|
40 |
+
response_text = response_text.replace("assistant", "")
|
41 |
+
|
42 |
+
# Post-processing to handle repeated words and ensure completeness
|
43 |
+
words = response_text.split()
|
44 |
+
seen = set()
|
45 |
+
filtered_words = [word for word in words if word.lower() not in seen and not seen.add(word.lower())]
|
46 |
+
response_text = ' '.join(filtered_words)
|
47 |
+
response_text = response_text.strip() # Ensuring no extra spaces at the ends
|
48 |
+
if not response_text.endswith('.'):
|
49 |
+
response_text += '.'
|
50 |
+
|
51 |
+
return response_text
|
52 |
+
except Exception as e:
|
53 |
+
return f"Error in generating response: {e}"
|
54 |
+
|
55 |
+
def ChatBot(history, document, question):
|
56 |
+
greetings = ["hi", "hello", "hey", "greetings", "what's up", "howdy"]
|
57 |
+
question_lower = question.lower().strip()
|
58 |
+
if question_lower in greetings or any(question_lower.startswith(greeting) for greeting in greetings):
|
59 |
+
return history + [("User", question), ("Bot", "Hello! How can I assist you with the document today?")]
|
60 |
+
|
61 |
+
text = extract_text_from_pdf(document)
|
62 |
+
response = Bot(text, question)
|
63 |
+
history.append(("User", question))
|
64 |
+
history.append(("Bot", response))
|
65 |
+
return history
|
66 |
+
|
67 |
+
with gr.Blocks() as iface:
|
68 |
+
chatbot = gr.Chatbot()
|
69 |
+
document = gr.File(label="Upload PDF Document", type="filepath")
|
70 |
+
question = gr.Textbox(label="Ask a Question", placeholder="Type your question here...")
|
71 |
+
|
72 |
+
def respond(history, document, question):
|
73 |
+
return ChatBot(history, document, question)
|
74 |
+
|
75 |
+
question.submit(respond, [chatbot, document, question], chatbot)
|
76 |
+
|
77 |
+
iface.launch(debug=True)
|