Spaces:
Runtime error
Runtime error
File size: 9,025 Bytes
7627550 a5d3830 7627550 0adc74f e5e07a1 7627550 e5e07a1 a5d3830 e5e07a1 81bb56a e5e07a1 a5d3830 3fe3bde 11954cc a5d3830 7627550 a5d3830 bb8eccf 7627550 a5d3830 f5ac780 7627550 a60e367 7627550 a8dcf10 7627550 a5d3830 7627550 a5d3830 73d1a08 a5d3830 e5e07a1 0adc74f a5d3830 0adc74f b286610 a5d3830 78d0916 a5d3830 23182d5 a5d3830 23182d5 a5d3830 |
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 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 |
# from langchain.document_loaders import TextLoader
# import pinecone
# from langchain.vectorstores import Pinecone
# import os
# from transformers import AutoTokenizer, AutoModel
# from langchain.agents.agent_toolkits import create_conversational_retrieval_agent
# from langchain.agents.agent_toolkits import create_retriever_tool
# from langchain.chat_models import ChatOpenAI
# import torch
# from langchain.agents.openai_functions_agent.agent_token_buffer_memory import (AgentTokenBufferMemory)
# from langchain.agents.openai_functions_agent.base import OpenAIFunctionsAgent
# from langchain.schema.messages import SystemMessage
# from langchain.prompts import MessagesPlaceholder
# import gradio as gr
# import time
# from db_func import insert_one
# from global_variable_module import gobal_input, global_output
# import random
# def get_bert_embeddings(sentence):
# embeddings = []
# input_ids = tokenizer.encode(sentence, return_tensors="pt")
# with torch.no_grad():
# output = model(input_ids)
# embedding = output.last_hidden_state[:,0,:].numpy().tolist()
# return embedding
# model_name = "BAAI/bge-base-en-v1.5"
# model = AutoModel.from_pretrained("/Users/aakashbhatnagar/Documents/masters/ophthal_llm/models/models--BAAI--bge-base-en-v1.5/snapshots/617ca489d9e86b49b8167676d8220688b99db36e")
# tokenizer = AutoTokenizer.from_pretrained("/Users/aakashbhatnagar/Documents/masters/ophthal_llm/models/models--BAAI--bge-base-en-v1.5/snapshots/617ca489d9e86b49b8167676d8220688b99db36e")
# prompt_file = open("prompts/version_2.txt", "r").read()
# pinecone.init(
# api_key=os.getenv("PINECONE_API_KEY"), # find at app.pinecone.io
# environment=os.getenv("PINECONE_ENV"), # next to api key in console
# )
# index_name = "ophtal-knowledge-base"
# index = pinecone.Index(index_name)
# vectorstore = Pinecone(index, get_bert_embeddings, "text")
# retriever = vectorstore.as_retriever()
# tool = create_retriever_tool(
# retriever,
# "search_ophtal-knowledge-base",
# "Searches and returns documents regarding the ophtal-knowledge-base.",
# )
# tools = [tool]
# system_message = SystemMessage(content=prompt_file)
# memory_key='history'
# llm = ChatOpenAI(openai_api_key=os.getenv("OPENAI_API_KEY"), model="gpt-4", temperature=0.2)
# prompt = OpenAIFunctionsAgent.create_prompt(
# system_message=system_message,
# extra_prompt_messages=[MessagesPlaceholder(variable_name=memory_key)],
# )
# agent_executor = create_conversational_retrieval_agent(llm, tools, verbose=False, prompt=prompt)
# user_name = None
# def run(input_):
# output = agent_executor({"input": input_})
# output_text = output["output"]
# source_text = ""
# doc_text = ""
# global_input = input_
# global_output = output_text
# if len(output["intermediate_steps"])>0:
# documents = output["intermediate_steps"][0][1]
# sources = []
# docs = []
# for doc in documents:
# if doc.metadata["source"] not in sources:
# sources.append(doc.metadata["source"])
# docs.append(doc.page_content)
# for i in range(len(sources)):
# temp = sources[i].replace('.pdf', '').replace('.txt', '').replace("AAO", "").replace("2022-2023", "").replace("data/book", "").replace("text", "").replace(" ", " ")
# source_text += f"{i+1}. {temp}\n"
# doc_text += f"{i+1}. {docs[i]}\n"
# # output_text = f"{output_text} \n\nSources: \n{source_text}\n\nDocuments: \n{doc_text}"
# # output_text = f"{output_text}"
# doc_to_insert = {
# "user": user_name,
# "input": input_,
# "output": output_text,
# "source": source_text,
# "documents": doc_text
# }
# insert_one(doc_to_insert)
# return output_text
# def make_conversation(message, history):
# text_ = run(message)
# for i in range(len(text_)):
# time.sleep(0.001)
# yield text_[: i+1]
# def auth_function(username, password):
# user_name = username
# return username == password
# def random_response(message, accuracy, history):
# print(type(message))
# print(message)
# print(accuracy)
# out = random.choice(["Yes", "No"])
# gobal_input = out
# # open a txt file
# with open("function hit", "a+") as f:
# f.write(message)
# return out
from langchain.document_loaders import TextLoader
import pinecone
from langchain.vectorstores import Pinecone
import os
from transformers import AutoTokenizer, AutoModel
from langchain.agents.agent_toolkits import create_conversational_retrieval_agent
from langchain.agents.agent_toolkits import create_retriever_tool
from langchain.chat_models import ChatOpenAI
import torch
from langchain.agents.openai_functions_agent.agent_token_buffer_memory import (AgentTokenBufferMemory)
from langchain.agents.openai_functions_agent.base import OpenAIFunctionsAgent
from langchain.schema.messages import SystemMessage
from langchain.prompts import MessagesPlaceholder
import gradio as gr
import time
from db_func import insert_one
from langchain.agents import AgentExecutor
import re
import wordninja
def clean_text(text):
text = text.strip().lower()
output_paragraph = ' '.join(''.join(text.split()).split(' '))
words = wordninja.split(output_paragraph)
return ' '.join(words)
def get_bert_embeddings(sentence):
embeddings = []
input_ids = tokenizer.encode(sentence, return_tensors="pt")
with torch.no_grad():
output = model(input_ids)
embedding = output.last_hidden_state[:,0,:].numpy().tolist()
return embedding
model_name = "BAAI/bge-base-en-v1.5"
model = AutoModel.from_pretrained(model_name)
tokenizer = AutoTokenizer.from_pretrained(model_name)
prompt_file = open("prompts/version_2.txt", "r").read()
pinecone.init(
api_key=os.getenv("PINECONE_API_KEY"), # find at app.pinecone.io
environment=os.getenv("PINECONE_ENV"), # next to api key in console
)
index_name = "ophtal-knowledge-base"
index = pinecone.Index(index_name)
vectorstore = Pinecone(index, get_bert_embeddings, "text")
retriever = vectorstore.as_retriever()
tool = create_retriever_tool(
retriever,
"search_ophtal-knowledge-base",
"Searches and returns documents regarding the ophtal-knowledge-base.",
)
tools = [tool]
system_message = SystemMessage(content="You are an assistant to ophthamologists and your name is 'Dr.V AI'. Help users answer medical questions. You are supposed to answer only medical questions and not general questions.")
memory_key='history'
llm = ChatOpenAI(openai_api_key=os.getenv("OPENAI_API_KEY"), model="gpt-4", temperature=0.2)
# llm = ChatOpenAI(openai_api_key="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", model="gpt-4", temperature=0.2)
prompt = OpenAIFunctionsAgent.create_prompt(
system_message=system_message,
extra_prompt_messages=[MessagesPlaceholder(variable_name=memory_key)],
)
memory = AgentTokenBufferMemory(memory_key=memory_key, llm=llm, max_token_limit=4000)
# agent_executor = create_conversational_retrieval_agent(llm, tools, verbose=False, prompt=prompt, )
agent = OpenAIFunctionsAgent(llm=llm, tools=tools, prompt=prompt)
agent_executor = AgentExecutor(
agent=agent,
tools=tools,
memory=memory,
verbose=False,
return_intermediate_steps=True,
max_iterations = 2
)
user_name = None
def run(input_):
output = agent_executor({"input": input_})
output_text = output["output"]
print(output_text)
source_text = ""
doc_text = ""
if len(output["intermediate_steps"])>0:
documents = output["intermediate_steps"][0][1]
sources = []
docs = []
for doc in documents:
if doc.metadata["source"] not in sources:
sources.append(doc.metadata["source"])
docs.append(doc.page_content)
for i in range(len(sources)):
temp = sources[i].replace('.pdf', '').replace('.txt', '').replace("AAO", "").replace("2022-2023", "").replace("data/book", "").replace("text", "").replace(" ", " ")
source_text += f"{i+1}. {temp}\n"
cleaned_text = re.sub(r'[^a-zA-Z0-9\s]', '', clean_text(docs[i]))
doc_text += f"{i+1}. {cleaned_text}\n"
# output_text = f"{output_text} \n\nSources: \n{source_text}"
output_text = f"{output_text} \n\nSources: \n{source_text}\n\nDocuments: \n{doc_text}"
# output_text = f"{output_text}"
doc_to_insert = {
"user": user_name,
"input": input_,
"output": output_text,
"source": source_text,
"documents": doc_text
}
insert_one(doc_to_insert)
return output_text
def make_conversation(message, history):
text_ = run(message)
for i in range(len(text_)):
time.sleep(0.001)
yield text_[: i+1]
def auth_function(username, password):
user_name = username
return username == password
|