Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,18 +1,18 @@
|
|
| 1 |
"""Main entrypoint for the app."""
|
|
|
|
| 2 |
import os
|
|
|
|
| 3 |
import time
|
| 4 |
from queue import Queue
|
| 5 |
from timeit import default_timer as timer
|
| 6 |
|
| 7 |
import gradio as gr
|
| 8 |
-
from anyio.from_thread import start_blocking_portal
|
| 9 |
|
| 10 |
from app_modules.init import app_init
|
| 11 |
-
from app_modules.utils import print_llm_response
|
| 12 |
|
| 13 |
llm_loader, qa_chain = app_init()
|
| 14 |
|
| 15 |
-
show_param_settings = os.environ.get("SHOW_PARAM_SETTINGS") == "true"
|
| 16 |
share_gradio_app = os.environ.get("SHARE_GRADIO_APP") == "true"
|
| 17 |
using_openai = os.environ.get("LLM_MODEL_TYPE") == "openai"
|
| 18 |
chat_history_enabled = os.environ.get("CHAT_HISTORY_ENABLED") == "true"
|
|
@@ -28,176 +28,84 @@ href = (
|
|
| 28 |
else f"https://huggingface.co/{model}"
|
| 29 |
)
|
| 30 |
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
|
|
|
|
|
|
|
| 34 |
|
| 35 |
-
|
| 36 |
<div align="left">
|
| 37 |
<p> Currently Running: <a href="{href}">{model}</a></p>
|
| 38 |
</div>
|
| 39 |
"""
|
| 40 |
|
| 41 |
-
description = """\
|
| 42 |
-
<div align="center" style="margin:16px 0">
|
| 43 |
-
The demo is built on <a href="https://github.com/hwchase17/langchain">LangChain</a>.
|
| 44 |
-
</div>
|
| 45 |
-
"""
|
| 46 |
|
| 47 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 48 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 49 |
|
| 50 |
-
def qa(chatbot):
|
| 51 |
-
user_msg = chatbot[-1][0]
|
| 52 |
q = Queue()
|
| 53 |
result = Queue()
|
| 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 |
-
ret = result.get()
|
| 100 |
-
titles = []
|
| 101 |
-
for doc in ret["source_documents"]:
|
| 102 |
-
page = doc.metadata["page"] + 1
|
| 103 |
-
url = f"{doc.metadata['url']}#page={page}"
|
| 104 |
-
file_name = doc.metadata["source"].split("/")[-1]
|
| 105 |
-
title = f"{file_name} Page: {page}"
|
| 106 |
-
if title not in titles:
|
| 107 |
-
titles.append(title)
|
| 108 |
-
chatbot[-1][1] += f"1. [{title}]({url})\n"
|
| 109 |
-
|
| 110 |
-
yield chatbot
|
| 111 |
-
|
| 112 |
-
|
| 113 |
-
with open("assets/custom.css", "r", encoding="utf-8") as f:
|
| 114 |
-
customCSS = f.read()
|
| 115 |
-
|
| 116 |
-
with gr.Blocks(css=customCSS) as demo:
|
| 117 |
-
user_question = gr.State("")
|
| 118 |
-
with gr.Row():
|
| 119 |
-
gr.HTML(title)
|
| 120 |
-
gr.Markdown(description_top)
|
| 121 |
-
with gr.Row(equal_height=True):
|
| 122 |
-
with gr.Column(scale=5):
|
| 123 |
-
with gr.Row():
|
| 124 |
-
chatbot = gr.Chatbot(elem_id="inflaton_chatbot", height="100%")
|
| 125 |
-
with gr.Row():
|
| 126 |
-
with gr.Column(scale=2):
|
| 127 |
-
user_input = gr.Textbox(
|
| 128 |
-
show_label=False,
|
| 129 |
-
placeholder="Enter your question here",
|
| 130 |
-
container=False,
|
| 131 |
-
)
|
| 132 |
-
with gr.Column(
|
| 133 |
-
min_width=70,
|
| 134 |
-
):
|
| 135 |
-
submitBtn = gr.Button("Send")
|
| 136 |
-
with gr.Column(
|
| 137 |
-
min_width=70,
|
| 138 |
-
):
|
| 139 |
-
clearBtn = gr.Button("Clear")
|
| 140 |
-
if show_param_settings:
|
| 141 |
-
with gr.Column():
|
| 142 |
-
with gr.Column(
|
| 143 |
-
min_width=50,
|
| 144 |
-
):
|
| 145 |
-
with gr.Tab(label="Parameter Setting"):
|
| 146 |
-
gr.Markdown("# Parameters")
|
| 147 |
-
top_p = gr.Slider(
|
| 148 |
-
minimum=-0,
|
| 149 |
-
maximum=1.0,
|
| 150 |
-
value=0.95,
|
| 151 |
-
step=0.05,
|
| 152 |
-
# interactive=True,
|
| 153 |
-
label="Top-p",
|
| 154 |
-
)
|
| 155 |
-
temperature = gr.Slider(
|
| 156 |
-
minimum=0.1,
|
| 157 |
-
maximum=2.0,
|
| 158 |
-
value=0,
|
| 159 |
-
step=0.1,
|
| 160 |
-
# interactive=True,
|
| 161 |
-
label="Temperature",
|
| 162 |
-
)
|
| 163 |
-
max_new_tokens = gr.Slider(
|
| 164 |
-
minimum=0,
|
| 165 |
-
maximum=2048,
|
| 166 |
-
value=2048,
|
| 167 |
-
step=8,
|
| 168 |
-
# interactive=True,
|
| 169 |
-
label="Max Generation Tokens",
|
| 170 |
-
)
|
| 171 |
-
max_context_length_tokens = gr.Slider(
|
| 172 |
-
minimum=0,
|
| 173 |
-
maximum=4096,
|
| 174 |
-
value=4096,
|
| 175 |
-
step=128,
|
| 176 |
-
# interactive=True,
|
| 177 |
-
label="Max Context Tokens",
|
| 178 |
-
)
|
| 179 |
-
gr.Markdown(description)
|
| 180 |
-
|
| 181 |
-
def chat(user_message, history):
|
| 182 |
-
return "", history + [[user_message, None]]
|
| 183 |
-
|
| 184 |
-
user_input.submit(
|
| 185 |
-
chat, [user_input, chatbot], [user_input, chatbot], queue=True
|
| 186 |
-
).then(qa, chatbot, chatbot)
|
| 187 |
-
|
| 188 |
-
submitBtn.click(
|
| 189 |
-
chat, [user_input, chatbot], [user_input, chatbot], queue=True, api_name="chat"
|
| 190 |
-
).then(qa, chatbot, chatbot)
|
| 191 |
-
|
| 192 |
-
def reset():
|
| 193 |
-
return "", []
|
| 194 |
-
|
| 195 |
-
clearBtn.click(
|
| 196 |
-
reset,
|
| 197 |
-
outputs=[user_input, chatbot],
|
| 198 |
-
show_progress=True,
|
| 199 |
-
api_name="reset",
|
| 200 |
-
)
|
| 201 |
-
|
| 202 |
-
demo.title = "Chat with PCI DSS v4"
|
| 203 |
-
demo.queue().launch(share=share_gradio_app)
|
|
|
|
| 1 |
"""Main entrypoint for the app."""
|
| 2 |
+
|
| 3 |
import os
|
| 4 |
+
from threading import Thread
|
| 5 |
import time
|
| 6 |
from queue import Queue
|
| 7 |
from timeit import default_timer as timer
|
| 8 |
|
| 9 |
import gradio as gr
|
|
|
|
| 10 |
|
| 11 |
from app_modules.init import app_init
|
| 12 |
+
from app_modules.utils import print_llm_response
|
| 13 |
|
| 14 |
llm_loader, qa_chain = app_init()
|
| 15 |
|
|
|
|
| 16 |
share_gradio_app = os.environ.get("SHARE_GRADIO_APP") == "true"
|
| 17 |
using_openai = os.environ.get("LLM_MODEL_TYPE") == "openai"
|
| 18 |
chat_history_enabled = os.environ.get("CHAT_HISTORY_ENABLED") == "true"
|
|
|
|
| 28 |
else f"https://huggingface.co/{model}"
|
| 29 |
)
|
| 30 |
|
| 31 |
+
title = "Chat with PCI DSS v4"
|
| 32 |
+
examples = [
|
| 33 |
+
"What's PCI DSS?",
|
| 34 |
+
"Can you summarize the changes made from PCI DSS version 3.2.1 to version 4.0?",
|
| 35 |
+
]
|
| 36 |
|
| 37 |
+
description = f"""\
|
| 38 |
<div align="left">
|
| 39 |
<p> Currently Running: <a href="{href}">{model}</a></p>
|
| 40 |
</div>
|
| 41 |
"""
|
| 42 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 43 |
|
| 44 |
+
def task(question, chat_history, q, result):
|
| 45 |
+
start = timer()
|
| 46 |
+
inputs = {"question": question, "chat_history": chat_history}
|
| 47 |
+
ret = qa_chain.call_chain(inputs, None, q)
|
| 48 |
+
end = timer()
|
| 49 |
+
|
| 50 |
+
print(f"Completed in {end - start:.3f}s")
|
| 51 |
+
print_llm_response(ret)
|
| 52 |
+
|
| 53 |
+
result.put(ret)
|
| 54 |
+
|
| 55 |
+
|
| 56 |
+
def predict(message, history):
|
| 57 |
+
print("predict:", message, history)
|
| 58 |
|
| 59 |
+
chat_history = []
|
| 60 |
+
if chat_history_enabled:
|
| 61 |
+
for element in history:
|
| 62 |
+
item = (element[0] or "", element[1] or "")
|
| 63 |
+
chat_history.append(item)
|
| 64 |
|
|
|
|
|
|
|
| 65 |
q = Queue()
|
| 66 |
result = Queue()
|
| 67 |
+
t = Thread(target=task, args=(message, chat_history, q, result))
|
| 68 |
+
t.start() # Starting the generation in a separate thread.
|
| 69 |
+
|
| 70 |
+
partial_message = ""
|
| 71 |
+
count = 2 if len(chat_history) > 0 else 1
|
| 72 |
+
|
| 73 |
+
while count > 0:
|
| 74 |
+
while q.empty():
|
| 75 |
+
print("nothing generated yet - retry in 0.5s")
|
| 76 |
+
time.sleep(0.5)
|
| 77 |
+
|
| 78 |
+
for next_token in llm_loader.streamer:
|
| 79 |
+
partial_message += next_token or ""
|
| 80 |
+
# partial_message = remove_extra_spaces(partial_message)
|
| 81 |
+
yield partial_message
|
| 82 |
+
|
| 83 |
+
if count == 2:
|
| 84 |
+
partial_message += "\n\n"
|
| 85 |
+
|
| 86 |
+
count -= 1
|
| 87 |
+
|
| 88 |
+
partial_message += "\n\nSources:\n"
|
| 89 |
+
ret = result.get()
|
| 90 |
+
titles = []
|
| 91 |
+
for doc in ret["source_documents"]:
|
| 92 |
+
page = doc.metadata["page"] + 1
|
| 93 |
+
url = f"{doc.metadata['url']}#page={page}"
|
| 94 |
+
file_name = doc.metadata["source"].split("/")[-1]
|
| 95 |
+
title = f"{file_name} Page: {page}"
|
| 96 |
+
if title not in titles:
|
| 97 |
+
titles.append(title)
|
| 98 |
+
partial_message += f"1. [{title}]({url})\n"
|
| 99 |
+
|
| 100 |
+
yield partial_message
|
| 101 |
+
|
| 102 |
+
|
| 103 |
+
# Setting up the Gradio chat interface.
|
| 104 |
+
gr.ChatInterface(
|
| 105 |
+
predict,
|
| 106 |
+
title=title,
|
| 107 |
+
description=description,
|
| 108 |
+
examples=examples,
|
| 109 |
+
).launch(
|
| 110 |
+
share=share_gradio_app
|
| 111 |
+
) # Launching the web interface.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|