Matt Robinson
commited on
Commit
·
01095ae
1
Parent(s):
fe39603
chat isw app files
Browse files- LICENSE +21 -0
- README.md +17 -13
- app.py +103 -0
- cli_app.py +17 -0
- ingest_data.py +30 -0
- query_data.py +34 -0
- requirements.txt +6 -0
- vectorstore.pkl +0 -0
LICENSE
ADDED
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
MIT License
|
2 |
+
|
3 |
+
Copyright (c) 2023 Harrison Chase
|
4 |
+
|
5 |
+
Permission is hereby granted, free of charge, to any person obtaining a copy
|
6 |
+
of this software and associated documentation files (the "Software"), to deal
|
7 |
+
in the Software without restriction, including without limitation the rights
|
8 |
+
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
9 |
+
copies of the Software, and to permit persons to whom the Software is
|
10 |
+
furnished to do so, subject to the following conditions:
|
11 |
+
|
12 |
+
The above copyright notice and this permission notice shall be included in all
|
13 |
+
copies or substantial portions of the Software.
|
14 |
+
|
15 |
+
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
16 |
+
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
17 |
+
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
18 |
+
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
19 |
+
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
20 |
+
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
21 |
+
SOFTWARE.
|
README.md
CHANGED
@@ -1,13 +1,17 @@
|
|
1 |
-
|
2 |
-
|
3 |
-
|
4 |
-
|
5 |
-
|
6 |
-
|
7 |
-
|
8 |
-
|
9 |
-
|
10 |
-
|
11 |
-
|
12 |
-
|
13 |
-
|
|
|
|
|
|
|
|
|
|
1 |
+
# Chat-Your-Data
|
2 |
+
|
3 |
+
Create a ChatGPT like experience over your custom docs using [LangChain](https://github.com/hwchase17/langchain).
|
4 |
+
|
5 |
+
See [this blog post](https://blog.langchain.dev/tutorial-chatgpt-over-your-data/) for a more detailed explanation.
|
6 |
+
|
7 |
+
## Ingest data
|
8 |
+
|
9 |
+
Ingestion of data is done over the `state_of_the_union.txt` file.
|
10 |
+
Therefore, the only thing that is needed is to be done to ingest data is run `python ingest_data.py`
|
11 |
+
|
12 |
+
## Query data
|
13 |
+
Custom prompts are used to ground the answers in the state of the union text file.
|
14 |
+
|
15 |
+
## Running the Application
|
16 |
+
|
17 |
+
By running `python app.py` from the command line you can easily interact with your ChatGPT over your own data.
|
app.py
ADDED
@@ -0,0 +1,103 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import os
|
2 |
+
from typing import Optional, Tuple
|
3 |
+
|
4 |
+
import gradio as gr
|
5 |
+
import pickle
|
6 |
+
from query_data import get_chain
|
7 |
+
from threading import Lock
|
8 |
+
|
9 |
+
with open("vectorstore.pkl", "rb") as f:
|
10 |
+
vectorstore = pickle.load(f)
|
11 |
+
|
12 |
+
|
13 |
+
def set_openai_api_key(api_key: str):
|
14 |
+
"""Set the api key and return chain.
|
15 |
+
If no api_key, then None is returned.
|
16 |
+
"""
|
17 |
+
if api_key:
|
18 |
+
os.environ["OPENAI_API_KEY"] = api_key
|
19 |
+
chain = get_chain(vectorstore)
|
20 |
+
os.environ["OPENAI_API_KEY"] = ""
|
21 |
+
return chain
|
22 |
+
|
23 |
+
class ChatWrapper:
|
24 |
+
|
25 |
+
def __init__(self):
|
26 |
+
self.lock = Lock()
|
27 |
+
def __call__(
|
28 |
+
self, api_key: str, inp: str, history: Optional[Tuple[str, str]], chain
|
29 |
+
):
|
30 |
+
"""Execute the chat functionality."""
|
31 |
+
self.lock.acquire()
|
32 |
+
try:
|
33 |
+
history = history or []
|
34 |
+
# If chain is None, that is because no API key was provided.
|
35 |
+
if chain is None:
|
36 |
+
history.append((inp, "Please paste your OpenAI key to use"))
|
37 |
+
return history, history
|
38 |
+
# Set OpenAI key
|
39 |
+
import openai
|
40 |
+
openai.api_key = api_key
|
41 |
+
# Run chain and append input.
|
42 |
+
output = chain({"question": inp, "chat_history": history})["answer"]
|
43 |
+
history.append((inp, output))
|
44 |
+
except Exception as e:
|
45 |
+
raise e
|
46 |
+
finally:
|
47 |
+
self.lock.release()
|
48 |
+
return history, history
|
49 |
+
|
50 |
+
chat = ChatWrapper()
|
51 |
+
|
52 |
+
block = gr.Blocks(css=".gradio-container {background-color: lightgray}")
|
53 |
+
|
54 |
+
with block:
|
55 |
+
with gr.Row():
|
56 |
+
gr.Markdown("<h3><center>Chat-Your-Data (ISW Updates)</center></h3>")
|
57 |
+
|
58 |
+
openai_api_key_textbox = gr.Textbox(
|
59 |
+
placeholder="Paste your OpenAI API key (sk-...)",
|
60 |
+
show_label=False,
|
61 |
+
lines=1,
|
62 |
+
type="password",
|
63 |
+
)
|
64 |
+
|
65 |
+
chatbot = gr.Chatbot()
|
66 |
+
|
67 |
+
with gr.Row():
|
68 |
+
message = gr.Textbox(
|
69 |
+
label="What's your question?",
|
70 |
+
placeholder="Ask questions about the war in Ukraine",
|
71 |
+
lines=1,
|
72 |
+
)
|
73 |
+
submit = gr.Button(value="Send", variant="secondary").style(full_width=False)
|
74 |
+
|
75 |
+
gr.Examples(
|
76 |
+
examples=[
|
77 |
+
"What is the focus of the Russian offensive?",
|
78 |
+
"Where are the frontlines?",
|
79 |
+
"How are they consolidating power?",
|
80 |
+
],
|
81 |
+
inputs=message,
|
82 |
+
)
|
83 |
+
|
84 |
+
gr.HTML("Demo application of a LangChain chain.")
|
85 |
+
|
86 |
+
gr.HTML("""<center>
|
87 |
+
Powered by <a href='https://github.com/hwchase17/langchain'>LangChain 🦜️🔗</a>
|
88 |
+
and <a href='https://github.com/unstructured-io/unstructured'>Unstructured.IO</a>
|
89 |
+
</center>""")
|
90 |
+
|
91 |
+
state = gr.State()
|
92 |
+
agent_state = gr.State()
|
93 |
+
|
94 |
+
submit.click(chat, inputs=[openai_api_key_textbox, message, state, agent_state], outputs=[chatbot, state])
|
95 |
+
message.submit(chat, inputs=[openai_api_key_textbox, message, state, agent_state], outputs=[chatbot, state])
|
96 |
+
|
97 |
+
openai_api_key_textbox.change(
|
98 |
+
set_openai_api_key,
|
99 |
+
inputs=[openai_api_key_textbox],
|
100 |
+
outputs=[agent_state],
|
101 |
+
)
|
102 |
+
|
103 |
+
block.launch(debug=True)
|
cli_app.py
ADDED
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import pickle
|
2 |
+
from query_data import get_chain
|
3 |
+
|
4 |
+
|
5 |
+
if __name__ == "__main__":
|
6 |
+
with open("vectorstore.pkl", "rb") as f:
|
7 |
+
vectorstore = pickle.load(f)
|
8 |
+
qa_chain = get_chain(vectorstore)
|
9 |
+
chat_history = []
|
10 |
+
print("Chat with your docs!")
|
11 |
+
while True:
|
12 |
+
print("Human:")
|
13 |
+
question = input()
|
14 |
+
result = qa_chain({"question": question, "chat_history": chat_history})
|
15 |
+
chat_history.append((question, result["answer"]))
|
16 |
+
print("AI:")
|
17 |
+
print(result["answer"])
|
ingest_data.py
ADDED
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from langchain.text_splitter import RecursiveCharacterTextSplitter
|
2 |
+
from langchain.document_loaders import UnstructuredURLLoader
|
3 |
+
from langchain.vectorstores.faiss import FAISS
|
4 |
+
from langchain.embeddings import OpenAIEmbeddings
|
5 |
+
import pickle
|
6 |
+
|
7 |
+
# Load Data
|
8 |
+
urls = [
|
9 |
+
"https://www.understandingwar.org/backgrounder/russian-offensive-campaign-assessment-february-4-2023",
|
10 |
+
"https://www.understandingwar.org/backgrounder/russian-offensive-campaign-assessment-february-6-2023",
|
11 |
+
"https://www.understandingwar.org/backgrounder/russian-offensive-campaign-assessment-february-7-2023",
|
12 |
+
"https://www.understandingwar.org/backgrounder/russian-offensive-campaign-assessment-february-8-2023",
|
13 |
+
"https://www.understandingwar.org/backgrounder/russian-offensive-campaign-assessment-february-9-2023",
|
14 |
+
]
|
15 |
+
loader = UnstructuredURLLoader(urls=urls)
|
16 |
+
raw_documents = loader.load()
|
17 |
+
|
18 |
+
# Split text
|
19 |
+
text_splitter = RecursiveCharacterTextSplitter()
|
20 |
+
documents = text_splitter.split_documents(raw_documents)
|
21 |
+
|
22 |
+
|
23 |
+
# Load Data to vectorstore
|
24 |
+
embeddings = OpenAIEmbeddings()
|
25 |
+
vectorstore = FAISS.from_documents(documents, embeddings)
|
26 |
+
|
27 |
+
|
28 |
+
# Save vectorstore
|
29 |
+
with open("vectorstore.pkl", "wb") as f:
|
30 |
+
pickle.dump(vectorstore, f)
|
query_data.py
ADDED
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from langchain.prompts.prompt import PromptTemplate
|
2 |
+
from langchain.llms import OpenAI
|
3 |
+
from langchain.chains import ChatVectorDBChain
|
4 |
+
|
5 |
+
_template = """Given the following conversation and a follow up question, rephrase the follow up question to be a standalone question.
|
6 |
+
You can assume the question about the war in Ukraine.
|
7 |
+
|
8 |
+
Chat History:
|
9 |
+
{chat_history}
|
10 |
+
Follow Up Input: {question}
|
11 |
+
Standalone question:"""
|
12 |
+
CONDENSE_QUESTION_PROMPT = PromptTemplate.from_template(_template)
|
13 |
+
|
14 |
+
template = """You are an AI assistant for answering questions about the war in Ukraine.
|
15 |
+
You are given the following extracted parts of a long document and a question. Provide a conversational answer.
|
16 |
+
If you don't know the answer, just say "Hmm, I'm not sure." Don't try to make up an answer.
|
17 |
+
If the question is not about the war in Ukraine, politely inform them that you are tuned to only answer questions about the war in Ukraine.
|
18 |
+
Question: {question}
|
19 |
+
=========
|
20 |
+
{context}
|
21 |
+
=========
|
22 |
+
Answer in Markdown:"""
|
23 |
+
QA_PROMPT = PromptTemplate(template=template, input_variables=["question", "context"])
|
24 |
+
|
25 |
+
|
26 |
+
def get_chain(vectorstore):
|
27 |
+
llm = OpenAI(temperature=0)
|
28 |
+
qa_chain = ChatVectorDBChain.from_llm(
|
29 |
+
llm,
|
30 |
+
vectorstore,
|
31 |
+
qa_prompt=QA_PROMPT,
|
32 |
+
condense_question_prompt=CONDENSE_QUESTION_PROMPT,
|
33 |
+
)
|
34 |
+
return qa_chain
|
requirements.txt
ADDED
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
beautifulsoup4
|
2 |
+
langchain
|
3 |
+
openai
|
4 |
+
unstructured>=0.4.7
|
5 |
+
faiss-cpu
|
6 |
+
gradio
|
vectorstore.pkl
ADDED
Binary file (499 kB). View file
|
|