Spaces:
Sleeping
Sleeping
File size: 3,075 Bytes
baf5ad5 df95991 baf5ad5 df95991 baf5ad5 df95991 baf5ad5 df95991 baf5ad5 14422bb baf5ad5 df95991 baf5ad5 df95991 baf5ad5 3d3fa68 baf5ad5 f856b07 baf5ad5 df95991 baf5ad5 df95991 e9ffa52 df95991 e9ffa52 df95991 e9ffa52 df95991 e9ffa52 df95991 baf5ad5 df95991 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 |
import gradio as gr
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
import os
from langchain import PromptTemplate
from langchain import LLMChain
from langchain_together import Together
import re
import pdfplumber
# Set the API key with double quotes
os.environ['TOGETHER_API_KEY'] = "c2f52626b97118b71c0c36f66eda4f5957c8fc475e760c3d72f98ba07d3ed3b5"
text = ""
max_pages = 7
with pdfplumber.open("Diabetes Type 1.pdf") as pdf:
for i, page in enumerate(pdf.pages):
if i >= max_pages:
break
text += page.extract_text() + "\n"
def Bot(Questions):
chat_template = """
Based on the provided context: {text}
Please answer the following question: {Questions}
Only provide answers that are directly related to the context.
"""
prompt = PromptTemplate(
input_variables=['text', 'Questions'],
template=chat_template
)
llama3 = Together(model="meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo", max_tokens=50)
Generated_chat = LLMChain(llm=llama3, prompt=prompt)
try:
response = Generated_chat.invoke({
"text": text,
"Questions": Questions
})
response_text = response['text']
response_text = response_text.replace("assistant", "")
# Post-processing to handle repeated words and ensure completeness
words = response_text.split()
seen = set()
filtered_words = [word for word in words if word.lower() not in seen and not seen.add(word.lower())]
response_text = ' '.join(filtered_words)
response_text = response_text.strip() # Ensuring no extra spaces at the ends
if not response_text.endswith('.'):
response_text += '.'
return response_text
except Exception as e:
return f"Error in generating response: {e}"
def ChatBot(Questions):
greetings = ["hi", "hello", "hey", "greetings", "what's up", "howdy"]
# Check if the input question is a greeting
question_lower = Questions.lower().strip()
if question_lower in greetings or any(question_lower.startswith(greeting) for greeting in greetings):
return "Hello! How can I assist you with the document today?"
else:
response=Bot(Questions)
print(response)
return response.translate(str.maketrans('', '', '\n'))
# text_embedding = model.encode(text, convert_to_tensor=True)
# statement_embedding = model.encode(statement, convert_to_tensor=True)
# # Compute the cosine similarity between the embeddings
# similarity = util.pytorch_cos_sim(text_embedding, statement_embedding)
# # Print the similarity score
# print(f"Cosine similarity: {similarity.item()}")
# # Define a threshold for considering the statement as related
# threshold = 0.7
# if similarity.item() > threshold:
# response=Bot(Questions)
# return response
# else:
# response="The statement is not related to the text."
# return response
iface = gr.Interface(fn=ChatBot, inputs="text", outputs="text", title="Chatbot")
iface.launch(debug=True) |