File size: 11,722 Bytes
f63bc6b |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 |
"""
app_v1.py: v1 of the app
1. allows the user to upload multiple files at runtime (upto a maximum, that is set via a paremeter)
a. user is told about the max; cl max is 10 files
b. file type can either be .txt or .pdf
c. see the use of the accept option in the cl.AskFileMessage object:
set this as: accept=["text/plain", "application/pdf"]
2. uploaded files are processed on start:
if pdf, then first convert to text: i.e., the collection of uploaded files is converted to text if needed
text is split into chunks using langchain RecursiveCharacterTextSplitter util -
options TBD
3. uses openai embeddings (specifically, text-embedding-3-small embeddings) to convert chunks into embeddings
note - the choice of embeddings model can be controlled via optional parameters to be passed in when
the vector db is instantiated
4. save these embeddings in a vector db
hope to use Chroma or Qdrant
NOTES here...
5. instantiate the RAQA (retrieval augmented qa) pipeline
use LangChain
chat memory needed
requires instantiation of an openai client session and the vector db
in addition, of course we need to set up system and user prompts
here this is set up using aimakerspace.openai.utils prompts.py module classes
6. the cl.on_message decorator wraps the main function
this function
a. receives the query that the user types in
b. runs the RAQA pipeline
c. sends results back to UI for dislay
Additional Notes:
a. note the use of async functions and await async syntax throughout the module here!
b. note the use of yield rather than return in certain key functions
c. note the use of streaming capabilities when needed
d. the use of the python tempfile module when the user input text file is processed
(i) NamedTemporaryFile
(ii) with options to persist the storage of the temp file
"""
import os
from typing import List
from dotenv import load_dotenv
import tempfile
import getpass
from uuid import uuid4
import tempfile
# chainlit imports
import chainlit as cl
from chainlit.types import AskFileResponse
# langchain imports
# document loader
from langchain_community.document_loaders import TextLoader, PyPDFLoader
# text splitter
from langchain_text_splitters import RecursiveCharacterTextSplitter
# embeddings model to embed each chunk of text in doc
from langchain_openai import OpenAIEmbeddings
# vector store
from langchain_chroma import Chroma
# llm for text generation using prompt plus retrieved context plus query
from langchain_openai import ChatOpenAI
# templates to create custom prompts
from langchain_core.prompts import ChatPromptTemplate, PromptTemplate
# chains
from langchain.chains import create_retrieval_chain
from langchain.chains.combine_documents import create_stuff_documents_chain
# LCEL Runnable Passthrough
from langchain_core.runnables import RunnablePassthrough
# to parse output from llm
from langchain_core.output_parsers import StrOutputParser
from langchain.docstore.document import Document
# aimakerspace imports
# from aimakerspace.text_utils import CharacterTextSplitter, TextFileLoader
# from aimakerspace.openai_utils.prompts import (
# UserRolePrompt,
# SystemRolePrompt,
# AssistantRolePrompt,
# )
# from aimakerspace.openai_utils.embedding import EmbeddingModel
# from aimakerspace.vectordatabase import VectorDatabase
# from aimakerspace.openai_utils.chatmodel import ChatOpenAI
# use getpass to load api keys at runtime
# alternative is to use load_dotenv()
# or, load secrets (eg on hf)
# os.environ["OPENAI_API_KEY"] = getpass.getpass("Enter your OpenAI API key: ")
load_dotenv()
# parameters to manage number of files imported via cl onstart
max_file_count = 2
# parameters to manage splitting
chunk_kwargs = {
'chunk_size': 1000,
'chunk_overlap': 100
}
# openai embeddings model parameters
openai_embed_kwargs = {
'model': 'text-embedding-3-small',
# With the `text-embedding-3` class
# of models, you can specify the size
# of the embeddings you want returned.
# 'dimensions': 1024
}
# chat model parameters
chat_model_name = {
'model_name': 'gpt-4o-mini'
}
openai_embed_kwargs = {
'model': 'text-embedding-3-large',
# With the `text-embedding-3` class
# of models, you can specify the size
# of the embeddings you want returned.
# 'dimensions': 1024
}
retriever_kwargs = {
'search_type': 'similarity',
'search_kwargs': {
'k': 20
}
}
system_prompt = (
"""You are an assistant for question-answering tasks.
You will be given political speeches by leading politicians and
will be asked questions based on these speeches.
Use the following pieces of retrieved context to answer
the question.
You must answer the question only based on the context provided.
If you don't know the answer or if the context does not provide sufficient information,
then say that you don't know.
Think through your answer step-by-step.
\n\n
Context:
{context}
"""
)
custom_rag_template = """\
You are an assistant for question-answering tasks.
You will be given political speeches by leading politicians and
will be asked questions based on these speeches.
Use the following pieces of retrieved context to answer
the question.
You must answer the question only based on the context provided.
If you don't know the answer or if the context does not provide sufficient information,
then say that you don't know.
Think through your answer step-by-step.
Context:
{context}
Question:
{question}
Helpful Answer:
"""
class RetrievalAugmentedQAPipelineWithLangchain:
def __init__(self,
list_of_files: List[AskFileResponse],
chunk_kwargs,
embed_kwargs,
retriever_kwargs,
system_prompt,
chat_model_name,
system_prompt_template=None,
use_lcel_for_chain=False):
self.list_of_files = list_of_files
self.chunk_kwargs = chunk_kwargs
self.embed_kwargs = embed_kwargs
self.retriever_kwargs = retriever_kwargs
self.system_prompt = system_prompt
self.chat_model_name = chat_model_name
self.system_prompt_template = system_prompt_template
self.use_lcel_for_chain = use_lcel_for_chain
self._setup_text_splitter()
self._setup_embeddings()
self._setup_llm()
return
def _setup_text_splitter(self):
self.text_splitter = RecursiveCharacterTextSplitter(
add_start_index=True,
**self.chunk_kwargs
)
return self
def _setup_embeddings(self):
self.embeddings = OpenAIEmbeddings(**self.embed_kwargs)
return self
def _setup_llm(self):
self.llm = ChatOpenAI(**self.chat_model_name)
return self
def format_docs(self):
return "\n\n".join(doc.page_content for doc in self.all_splits)
def process_all_uploaded_files(self):
self.all_splits = []
for file in self.list_of_files:
# set loader depending on type of file
if file.type == "text/plain":
Loader = TextLoader
elif file.type == "application/pdf":
Loader = PyPDFLoader
import tempfile
# make temporary copy of file and split into chunks
with tempfile.NamedTemporaryFile() as tempfile:
tempfile.write(file.content)
loader = Loader(tempfile.name)
documents = loader.load()
splits = self.text_splitter.split_documents(documents)
for i, split in enumerate(splits):
split.metadata["source"] = f"{file.name}_source_{i}"
self.all_splits.extend(splits)
return self
def make_chroma_vector_db(self):
# initialize vector store
vectorstore = Chroma.from_documents(documents=self.all_splits, embedding=self.embeddings)
self.retriever = vectorstore.as_retriever(**self.retriever_kwargs)
return self
def setup_chat_prompt_from_messages(self):
self.chat_prompt = ChatPromptTemplate.from_messages(
[
("system", self.system_prompt),
("human", "{input}"),
]
)
return self
def qa_chain(self):
self.question_answer_chain = create_stuff_documents_chain(self.llm, self.chat_prompt)
return self
def rag_chain(self):
self.rag_chain = create_retrieval_chain(self.retriever, self.question_answer_chain)
return self
def setup_prompt_from_template(self):
self.prompt_from_template = PromptTemplate.from_template(self.system_prompt_template)
return self
def lcel_rag_chain(self):
self.rag_chain = (
{"context": self.retriever | self.format_docs, "question": RunnablePassthrough()}
| self.prompt_from_template
| self.llm
| StrOutputParser()
)
def make_raqa_pipeline(self):
# load all docs uploaded by user, convert into text if needed and split into chunks
self.process_all_uploaded_files()
# load all splits and embeddings into vector db
self.make_chroma_vector_db()
if self.use_lcel_for_chain is False:
self.setup_chat_prompt_from_messages()
self.qa_chain()
self.rag_chain()
else:
self.setup_prompt_from_template()
self.lcel_rag_chain()
return self.rag_chain
@cl.on_chat_start
async def on_chat_start():
files = None
# Wait for the user to upload one or more files
user_input_files = []
filecount = 0
user_signals_done = False
while filecount < max_file_count and user_signals_done is False:
# while files == None:
files = await cl.AskFileMessage(
content=f"Please upload {max_file_count} text files or pdf documents to begin!",
accept=["text/plain", "application/pdf"],
max_size_mb=20,
max_files=2,
timeout=180,
).send()
if files:
user_input_files.append(files[0])
filecount += 1
else:
user_signals_done = True
user_input_file_names = [x.name for x in user_input_files]
msg = cl.Message(
content=f"Processing `{user_input_file_names}`...", disable_human_feedback=True
)
await msg.send()
# instantiate raqa pipeline object
qabot = RetrievalAugmentedQAPipelineWithLangchain(
list_of_files=user_input_files,
chunk_kwargs=chunk_kwargs,
embed_kwargs=openai_embed_kwargs,
retriever_kwargs=retriever_kwargs,
system_prompt=system_prompt,
chat_model_name=chat_model_name
)
raqa_chain = qabot.make_raqa_pipeline()
# Let the user know that the system is ready
msg.content = f"Processing `{user_input_file_names}` done. You can now ask questions!"
await msg.update()
cl.user_session.set("raqa_chain", raqa_chain)
@cl.on_message
async def main(message):
raqa_chain = cl.user_session.get("raqa_chain")
msg = cl.Message(content="")
# result = await raqa_chain.invoke({"input": message.content})
result = await cl.make_async(raqa_chain.invoke)({"input": message.content})
# async for stream_resp in result["answer"]:
for stream_resp in result["answer"]:
await msg.stream_token(stream_resp)
await msg.send()
|