Spaces:
Runtime error
Runtime error
Delete app.py
Browse files
app.py
DELETED
|
@@ -1,132 +0,0 @@
|
|
| 1 |
-
import os
|
| 2 |
-
import chainlit as cl
|
| 3 |
-
from dotenv import load_dotenv
|
| 4 |
-
from operator import itemgetter
|
| 5 |
-
from langchain_huggingface import HuggingFaceEndpoint
|
| 6 |
-
from langchain_community.document_loaders import TextLoader
|
| 7 |
-
from langchain_text_splitters import RecursiveCharacterTextSplitter
|
| 8 |
-
from langchain_community.vectorstores import FAISS
|
| 9 |
-
from langchain_huggingface import HuggingFaceEndpointEmbeddings
|
| 10 |
-
from langchain_core.prompts import PromptTemplate
|
| 11 |
-
from langchain.schema.output_parser import StrOutputParser
|
| 12 |
-
from langchain.schema.runnable import RunnablePassthrough
|
| 13 |
-
from langchain.schema.runnable.config import RunnableConfig
|
| 14 |
-
|
| 15 |
-
# GLOBAL SCOPE - ENTIRE APPLICATION HAS ACCESS TO VALUES SET IN THIS SCOPE #
|
| 16 |
-
# ---- ENV VARIABLES ---- #
|
| 17 |
-
"""
|
| 18 |
-
This function will load our environment file (.env) if it is present.
|
| 19 |
-
|
| 20 |
-
NOTE: Make sure that .env is in your .gitignore file - it is by default, but please ensure it remains there.
|
| 21 |
-
"""
|
| 22 |
-
load_dotenv()
|
| 23 |
-
|
| 24 |
-
"""
|
| 25 |
-
We will load our environment variables here.
|
| 26 |
-
"""
|
| 27 |
-
HF_LLM_ENDPOINT = os.environ["HF_LLM_ENDPOINT"]
|
| 28 |
-
HF_EMBED_ENDPOINT = os.environ["HF_EMBED_ENDPOINT"]
|
| 29 |
-
HF_TOKEN = os.environ["HF_TOKEN"]
|
| 30 |
-
|
| 31 |
-
# ---- GLOBAL DECLARATIONS ---- #
|
| 32 |
-
|
| 33 |
-
# -- RETRIEVAL -- #
|
| 34 |
-
"""
|
| 35 |
-
1. Load Documents from Text File
|
| 36 |
-
2. Split Documents into Chunks
|
| 37 |
-
3. Load HuggingFace Embeddings (remember to use the URL we set above)
|
| 38 |
-
4. Index Files if they do not exist, otherwise load the vectorstore
|
| 39 |
-
"""
|
| 40 |
-
### 1. CREATE TEXT LOADER AND LOAD DOCUMENTS
|
| 41 |
-
### NOTE: PAY ATTENTION TO THE PATH THEY ARE IN.
|
| 42 |
-
text_loader =
|
| 43 |
-
documents =
|
| 44 |
-
|
| 45 |
-
### 2. CREATE TEXT SPLITTER AND SPLIT DOCUMENTS
|
| 46 |
-
text_splitter =
|
| 47 |
-
split_documents =
|
| 48 |
-
|
| 49 |
-
### 3. LOAD HUGGINGFACE EMBEDDINGS
|
| 50 |
-
hf_embeddings =
|
| 51 |
-
|
| 52 |
-
if os.path.exists("./data/vectorstore"):
|
| 53 |
-
vectorstore = FAISS.load_local(
|
| 54 |
-
"./data/vectorstore",
|
| 55 |
-
hf_embeddings,
|
| 56 |
-
allow_dangerous_deserialization=True # this is necessary to load the vectorstore from disk as it's stored as a `.pkl` file.
|
| 57 |
-
)
|
| 58 |
-
hf_retriever = vectorstore.as_retriever()
|
| 59 |
-
print("Loaded Vectorstore")
|
| 60 |
-
else:
|
| 61 |
-
print("Indexing Files")
|
| 62 |
-
os.makedirs("./data/vectorstore", exist_ok=True)
|
| 63 |
-
### 4. INDEX FILES
|
| 64 |
-
### NOTE: REMEMBER TO BATCH THE DOCUMENTS WITH MAXIMUM BATCH SIZE = 32
|
| 65 |
-
|
| 66 |
-
hf_retriever = vectorstore.as_retriever()
|
| 67 |
-
|
| 68 |
-
# -- AUGMENTED -- #
|
| 69 |
-
"""
|
| 70 |
-
1. Define a String Template
|
| 71 |
-
2. Create a Prompt Template from the String Template
|
| 72 |
-
"""
|
| 73 |
-
### 1. DEFINE STRING TEMPLATE
|
| 74 |
-
RAG_PROMPT_TEMPLATE =
|
| 75 |
-
|
| 76 |
-
### 2. CREATE PROMPT TEMPLATE
|
| 77 |
-
rag_prompt =
|
| 78 |
-
|
| 79 |
-
# -- GENERATION -- #
|
| 80 |
-
"""
|
| 81 |
-
1. Create a HuggingFaceEndpoint for the LLM
|
| 82 |
-
"""
|
| 83 |
-
### 1. CREATE HUGGINGFACE ENDPOINT FOR LLM
|
| 84 |
-
hf_llm =
|
| 85 |
-
|
| 86 |
-
@cl.author_rename
|
| 87 |
-
def rename(original_author: str):
|
| 88 |
-
"""
|
| 89 |
-
This function can be used to rename the 'author' of a message.
|
| 90 |
-
|
| 91 |
-
In this case, we're overriding the 'Assistant' author to be 'Paul Graham Essay Bot'.
|
| 92 |
-
"""
|
| 93 |
-
rename_dict = {
|
| 94 |
-
"Assistant" : "Paul Graham Essay Bot"
|
| 95 |
-
}
|
| 96 |
-
return rename_dict.get(original_author, original_author)
|
| 97 |
-
|
| 98 |
-
@cl.on_chat_start
|
| 99 |
-
async def start_chat():
|
| 100 |
-
"""
|
| 101 |
-
This function will be called at the start of every user session.
|
| 102 |
-
|
| 103 |
-
We will build our LCEL RAG chain here, and store it in the user session.
|
| 104 |
-
|
| 105 |
-
The user session is a dictionary that is unique to each user session, and is stored in the memory of the server.
|
| 106 |
-
"""
|
| 107 |
-
|
| 108 |
-
### BUILD LCEL RAG CHAIN THAT ONLY RETURNS TEXT
|
| 109 |
-
lcel_rag_chain =
|
| 110 |
-
|
| 111 |
-
cl.user_session.set("lcel_rag_chain", lcel_rag_chain)
|
| 112 |
-
|
| 113 |
-
@cl.on_message
|
| 114 |
-
async def main(message: cl.Message):
|
| 115 |
-
"""
|
| 116 |
-
This function will be called every time a message is recieved from a session.
|
| 117 |
-
|
| 118 |
-
We will use the LCEL RAG chain to generate a response to the user query.
|
| 119 |
-
|
| 120 |
-
The LCEL RAG chain is stored in the user session, and is unique to each user session - this is why we can access it here.
|
| 121 |
-
"""
|
| 122 |
-
lcel_rag_chain = cl.user_session.get("lcel_rag_chain")
|
| 123 |
-
|
| 124 |
-
msg = cl.Message(content="")
|
| 125 |
-
|
| 126 |
-
async for chunk in lcel_rag_chain.astream(
|
| 127 |
-
{"query": message.content},
|
| 128 |
-
config=RunnableConfig(callbacks=[cl.LangchainCallbackHandler()]),
|
| 129 |
-
):
|
| 130 |
-
await msg.stream_token(chunk)
|
| 131 |
-
|
| 132 |
-
await msg.send()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|