Spaces:
Sleeping
Sleeping
danicafisher
commited on
Commit
•
354026c
1
Parent(s):
61170c1
Reverts
Browse files
app.py
CHANGED
@@ -1,60 +1,36 @@
|
|
1 |
-
|
2 |
-
|
|
|
3 |
from aimakerspace.text_utils import CharacterTextSplitter, PDFFileLoader
|
4 |
from aimakerspace.openai_utils.prompts import (
|
5 |
UserRolePrompt,
|
6 |
SystemRolePrompt,
|
|
|
7 |
)
|
8 |
from aimakerspace.openai_utils.embedding import EmbeddingModel
|
9 |
from aimakerspace.vectordatabase import VectorDatabase
|
10 |
from aimakerspace.openai_utils.chatmodel import ChatOpenAI
|
11 |
import chainlit as cl
|
12 |
-
|
13 |
-
# from operator import itemgetter
|
14 |
import nest_asyncio
|
15 |
nest_asyncio.apply()
|
16 |
-
from langchain_community.document_loaders import PyMuPDFLoader
|
17 |
-
from langchain_text_splitters import RecursiveCharacterTextSplitter
|
18 |
|
19 |
|
20 |
-
|
21 |
-
|
|
|
|
|
22 |
|
23 |
-
text_splitter =
|
24 |
-
|
25 |
-
|
26 |
-
)
|
27 |
-
|
28 |
-
documents_NIST = PyMuPDFLoader(filepath_NIST).load()
|
29 |
-
documents_Blueprint = PyMuPDFLoader(filepath_Blueprint).load()
|
30 |
-
|
31 |
-
split_NIST = text_splitter.split_documents(documents_NIST)
|
32 |
-
split_Blueprint = text_splitter.split_documents(documents_Blueprint)
|
33 |
-
|
34 |
-
# embeddings = OpenAIEmbeddings(model="text-embedding-3-small")
|
35 |
-
|
36 |
-
# vectorstore = Qdrant.from_documents(
|
37 |
-
# documents=rag_documents,
|
38 |
-
# embedding=embeddings,
|
39 |
-
# location=":memory:",
|
40 |
-
# collection_name="Implications of AI"
|
41 |
-
# )
|
42 |
-
# retriever = qdrant_vectorstore.as_retriever()
|
43 |
-
|
44 |
-
RAG_PROMPT = """\
|
45 |
-
Given a provided context and question, you must answer the question based only on context.
|
46 |
-
|
47 |
-
If you cannot answer the question based on the context - you must say "I don't know".
|
48 |
|
49 |
-
Context: {context}
|
50 |
-
Question: {question}
|
51 |
-
"""
|
52 |
-
|
53 |
-
# prompt = ChatPromptTemplate.from_template(RAG_PROMPT)
|
54 |
|
55 |
RAG_PROMPT_TEMPLATE = """ \
|
56 |
Use the provided context to answer the user's query.
|
|
|
57 |
You may not answer the user's query unless there is specific context in the following text.
|
|
|
58 |
If you do not know the answer, or cannot answer, please respond with "I don't know".
|
59 |
"""
|
60 |
|
@@ -63,6 +39,7 @@ rag_prompt = SystemRolePrompt(RAG_PROMPT_TEMPLATE)
|
|
63 |
USER_PROMPT_TEMPLATE = """ \
|
64 |
Context:
|
65 |
{context}
|
|
|
66 |
User Query:
|
67 |
{user_query}
|
68 |
"""
|
@@ -106,25 +83,21 @@ async def start_chat():
|
|
106 |
# "presence_penalty": 0,
|
107 |
# }
|
108 |
|
109 |
-
#
|
110 |
vector_db = VectorDatabase()
|
111 |
-
|
112 |
-
vector_db = await vector_db.abuild_from_list(
|
113 |
-
vector_db = await vector_db.abuild_from_list(split_Blueprint)
|
114 |
|
115 |
-
|
116 |
-
llm = ChatOpenAI(model="gpt-4o-mini", tags=["base_llm"])
|
117 |
-
|
118 |
|
119 |
-
#
|
120 |
-
|
121 |
vector_db_retriever=vector_db,
|
122 |
-
llm=
|
123 |
)
|
124 |
|
125 |
-
|
126 |
# cl.user_session.set("settings", settings)
|
127 |
-
cl.user_session.set("chain",
|
128 |
|
129 |
|
130 |
@cl.on_message # marks a function that should be run each time the chatbot receives a message from a user
|
|
|
1 |
+
import os
|
2 |
+
from typing import List
|
3 |
+
from chainlit.types import AskFileResponse
|
4 |
from aimakerspace.text_utils import CharacterTextSplitter, PDFFileLoader
|
5 |
from aimakerspace.openai_utils.prompts import (
|
6 |
UserRolePrompt,
|
7 |
SystemRolePrompt,
|
8 |
+
AssistantRolePrompt,
|
9 |
)
|
10 |
from aimakerspace.openai_utils.embedding import EmbeddingModel
|
11 |
from aimakerspace.vectordatabase import VectorDatabase
|
12 |
from aimakerspace.openai_utils.chatmodel import ChatOpenAI
|
13 |
import chainlit as cl
|
14 |
+
import asyncio
|
|
|
15 |
import nest_asyncio
|
16 |
nest_asyncio.apply()
|
|
|
|
|
17 |
|
18 |
|
19 |
+
pdf_loader_NIST = PDFFileLoader("data/NIST.AI.600-1.pdf")
|
20 |
+
pdf_loader_Blueprint = PDFFileLoader("data/Blueprint-for-an-AI-Bill-of-Rights.pdf")
|
21 |
+
documents_NIST = pdf_loader_NIST.load_documents()
|
22 |
+
documents_Blueprint = pdf_loader_Blueprint.load_documents()
|
23 |
|
24 |
+
text_splitter = CharacterTextSplitter()
|
25 |
+
split_documents_NIST = text_splitter.split_texts(documents_NIST)
|
26 |
+
split_documents_Blueprint = text_splitter.split_texts(documents_Blueprint)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
27 |
|
|
|
|
|
|
|
|
|
|
|
28 |
|
29 |
RAG_PROMPT_TEMPLATE = """ \
|
30 |
Use the provided context to answer the user's query.
|
31 |
+
|
32 |
You may not answer the user's query unless there is specific context in the following text.
|
33 |
+
|
34 |
If you do not know the answer, or cannot answer, please respond with "I don't know".
|
35 |
"""
|
36 |
|
|
|
39 |
USER_PROMPT_TEMPLATE = """ \
|
40 |
Context:
|
41 |
{context}
|
42 |
+
|
43 |
User Query:
|
44 |
{user_query}
|
45 |
"""
|
|
|
83 |
# "presence_penalty": 0,
|
84 |
# }
|
85 |
|
86 |
+
# Create a dict vector store
|
87 |
vector_db = VectorDatabase()
|
88 |
+
vector_db = await vector_db.abuild_from_list(split_documents_NIST)
|
89 |
+
vector_db = await vector_db.abuild_from_list(split_documents_Blueprint)
|
|
|
90 |
|
91 |
+
chat_openai = ChatOpenAI()
|
|
|
|
|
92 |
|
93 |
+
# Create a chain
|
94 |
+
retrieval_augmented_qa_pipeline = RetrievalAugmentedQAPipeline(
|
95 |
vector_db_retriever=vector_db,
|
96 |
+
llm=chat_openai
|
97 |
)
|
98 |
|
|
|
99 |
# cl.user_session.set("settings", settings)
|
100 |
+
cl.user_session.set("chain", retrieval_augmented_qa_pipeline)
|
101 |
|
102 |
|
103 |
@cl.on_message # marks a function that should be run each time the chatbot receives a message from a user
|