Spaces:
Runtime error
Runtime error
File size: 18,676 Bytes
454f02d 2d54438 454f02d 51f950f 454f02d 2d54438 454f02d 2d54438 454f02d 2d54438 454f02d 5285893 454f02d 32a5d3a 454f02d 32a5d3a 454f02d 5285893 454f02d 69dd450 2d54438 454f02d 2d54438 454f02d 2d54438 454f02d 32a5d3a 4970ac1 454f02d 2d54438 454f02d 9747a90 454f02d 2d54438 454f02d 2d54438 454f02d 2d54438 454f02d 9747a90 454f02d d8475f2 454f02d 9747a90 454f02d 9747a90 fa02564 9747a90 454f02d fa02564 45dafce 454f02d fa02564 454f02d fa02564 454f02d 9747a90 454f02d 45dafce 454f02d 576fd70 454f02d 9747a90 fa02564 9747a90 baa7dde 979c55d baa7dde 9747a90 45dafce e290146 454f02d 2d54438 454f02d 9747a90 fa02564 ef20af6 fa02564 454f02d 2d54438 5a49361 92e4006 45dafce 5a49361 e290146 92e4006 bb87a96 1b76ebb fe3d439 1b76ebb 5a49361 7e03c18 454f02d 9747a90 454f02d 9747a90 454f02d 9747a90 454f02d 6fcba90 a2ec04c 454f02d 9747a90 454f02d |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 |
import argparse
from collections import defaultdict
import datetime
import json
import os
import time
import uuid
os.system("pip install --upgrade gradio")
import gradio as gr
import requests
from fastchat.conversation import (
Conversation,
compute_skip_echo_len,
SeparatorStyle,
)
from fastchat.constants import LOGDIR
from fastchat.utils import (
build_logger,
server_error_msg,
violates_moderation,
moderation_msg,
)
from fastchat.serve.gradio_patch import Chatbot as grChatbot
from fastchat.serve.gradio_css import code_highlight_css
logger = build_logger("gradio_web_server", "gradio_web_server.log")
headers = {"User-Agent": "NeuralChat Client"}
no_change_btn = gr.Button.update()
enable_btn = gr.Button.update(interactive=True)
disable_btn = gr.Button.update(interactive=False)
controller_url = None
enable_moderation = False
conv_template_bf16 = Conversation(
system="A chat between a curious human and an artificial intelligence assistant. "
"The assistant gives helpful, detailed, and polite answers to the human's questions.",
roles=("Human", "Assistant"),
messages=(),
offset=0,
sep_style=SeparatorStyle.SINGLE,
sep="\n",
sep2="</s>",
)
def set_global_vars(controller_url_, enable_moderation_):
global controller_url, enable_moderation
controller_url = controller_url_
enable_moderation = enable_moderation_
def get_conv_log_filename():
t = datetime.datetime.now()
name = os.path.join(LOGDIR, f"{t.year}-{t.month:02d}-{t.day:02d}-conv.json")
return name
def get_model_list(controller_url):
ret = requests.post(controller_url + "/refresh_all_workers")
assert ret.status_code == 200
ret = requests.post(controller_url + "/list_models")
models = ret.json()["models"]
logger.info(f"Models: {models}")
return models
get_window_url_params = """
function() {
const params = new URLSearchParams(window.location.search);
url_params = Object.fromEntries(params);
console.log("url_params", url_params);
return url_params;
}
"""
def load_demo_single(models, url_params):
dropdown_update = gr.Dropdown.update(visible=True)
if "model" in url_params:
model = url_params["model"]
if model in models:
dropdown_update = gr.Dropdown.update(value=model, visible=True)
state = None
return (
state,
dropdown_update,
gr.Chatbot.update(visible=True),
gr.Textbox.update(visible=True),
gr.Button.update(visible=True),
gr.Row.update(visible=True),
gr.Accordion.update(visible=True),
)
def load_demo(url_params, request: gr.Request):
logger.info(f"load_demo. ip: {request.client.host}. params: {url_params}")
return load_demo_single(models, url_params)
def vote_last_response(state, vote_type, model_selector, request: gr.Request):
with open(get_conv_log_filename(), "a") as fout:
data = {
"tstamp": round(time.time(), 4),
"type": vote_type,
"model": model_selector,
"state": state.dict(),
"ip": request.client.host,
}
fout.write(json.dumps(data) + "\n")
def upvote_last_response(state, model_selector, request: gr.Request):
logger.info(f"upvote. ip: {request.client.host}")
vote_last_response(state, "upvote", model_selector, request)
return ("",) + (disable_btn,) * 3
def downvote_last_response(state, model_selector, request: gr.Request):
logger.info(f"downvote. ip: {request.client.host}")
vote_last_response(state, "downvote", model_selector, request)
return ("",) + (disable_btn,) * 3
def flag_last_response(state, model_selector, request: gr.Request):
logger.info(f"flag. ip: {request.client.host}")
vote_last_response(state, "flag", model_selector, request)
return ("",) + (disable_btn,) * 3
def regenerate(state, request: gr.Request):
logger.info(f"regenerate. ip: {request.client.host}")
state.messages[-1][-1] = None
state.skip_next = False
return (state, state.to_gradio_chatbot(), "") + (disable_btn,) * 5
def clear_history(request: gr.Request):
logger.info(f"clear_history. ip: {request.client.host}")
state = None
return (state, [], "") + (disable_btn,) * 5
def add_text(state, text, request: gr.Request):
logger.info(f"add_text. ip: {request.client.host}. len: {len(text)}")
if state is None:
state = conv_template_bf16.copy()
if len(text) <= 0:
state.skip_next = True
return (state, state.to_gradio_chatbot(), "") + (no_change_btn,) * 5
if enable_moderation:
flagged = violates_moderation(text)
if flagged:
logger.info(f"violate moderation. ip: {request.client.host}. text: {text}")
state.skip_next = True
return (state, state.to_gradio_chatbot(), moderation_msg) + (
no_change_btn,
) * 5
text = text[:1536] # Hard cut-off
state.append_message(state.roles[0], text)
state.append_message(state.roles[1], None)
state.skip_next = False
return (state, state.to_gradio_chatbot(), "") + (disable_btn,) * 5
def post_process_code(code):
sep = "\n```"
if sep in code:
blocks = code.split(sep)
if len(blocks) % 2 == 1:
for i in range(1, len(blocks), 2):
blocks[i] = blocks[i].replace("\\_", "_")
code = sep.join(blocks)
return code
def http_bot(state, model_selector, temperature, max_new_tokens, request: gr.Request):
logger.info(f"http_bot. ip: {request.client.host}")
start_tstamp = time.time()
model_name = model_selector
temperature = float(temperature)
max_new_tokens = int(max_new_tokens)
if state.skip_next:
# This generate call is skipped due to invalid inputs
yield (state, state.to_gradio_chatbot()) + (no_change_btn,) * 5
return
if len(state.messages) == state.offset + 2:
# First round of conversation
new_state = conv_template_bf16.copy()
new_state.conv_id = uuid.uuid4().hex
new_state.model_name = state.model_name or model_selector
new_state.append_message(new_state.roles[0], state.messages[-2][1])
new_state.append_message(new_state.roles[1], None)
state = new_state
# Query worker address
ret = requests.post(
controller_url + "/get_worker_address", json={"model": model_name}
)
worker_addr = ret.json()["address"]
logger.info(f"model_name: {model_name}, worker_addr: {worker_addr}")
# No available worker
if worker_addr == "":
state.messages[-1][-1] = server_error_msg
yield (
state,
state.to_gradio_chatbot(),
disable_btn,
disable_btn,
disable_btn,
enable_btn,
enable_btn,
)
return
# Construct prompt
prompt = state.get_prompt()
skip_echo_len = compute_skip_echo_len(model_name, state, prompt)
# Make requests
pload = {
"model": model_name,
"prompt": prompt,
"temperature": temperature,
"max_new_tokens": max_new_tokens,
"stop": "</s>"
}
logger.info(f"==== request ====\n{pload}")
start_time = time.time()
state.messages[-1][-1] = "▌"
yield (state, state.to_gradio_chatbot()) + (disable_btn,) * 5
try:
# Stream output
response = requests.post(
controller_url + "/worker_generate_stream",
headers=headers,
json=pload,
stream=True,
timeout=20,
)
for chunk in response.iter_lines(decode_unicode=False, delimiter=b"\0"):
if chunk:
data = json.loads(chunk.decode())
if data["error_code"] == 0:
output = data["text"][skip_echo_len:].strip()
output = post_process_code(output)
state.messages[-1][-1] = output + "▌"
yield (state, state.to_gradio_chatbot()) + (disable_btn,) * 5
else:
output = data["text"] + f" (error_code: {data['error_code']})"
state.messages[-1][-1] = output
yield (state, state.to_gradio_chatbot()) + (
disable_btn,
disable_btn,
disable_btn,
enable_btn,
enable_btn,
)
return
time.sleep(0.005)
except requests.exceptions.RequestException as e:
state.messages[-1][-1] = server_error_msg + f" (error_code: 4)"
yield (state, state.to_gradio_chatbot()) + (
disable_btn,
disable_btn,
disable_btn,
enable_btn,
enable_btn,
)
return
finish_tstamp = time.time() - start_time
elapsed_time = "\n✅generation elapsed time: {}s".format(round(finish_tstamp, 4))
# elapsed_time = "\n{}s".format(round(finish_tstamp, 4))
# elapsed_time = "<p class='time-style'>{}s </p>".format(round(finish_tstamp, 4))
state.messages[-1][-1] = state.messages[-1][-1][:-1] + elapsed_time
yield (state, state.to_gradio_chatbot()) + (enable_btn,) * 5
logger.info(f"{output}")
with open(get_conv_log_filename(), "a") as fout:
data = {
"tstamp": round(finish_tstamp, 4),
"type": "chat",
"model": model_name,
"gen_params": {
"temperature": temperature,
"max_new_tokens": max_new_tokens,
},
"start": round(start_tstamp, 4),
"finish": round(start_tstamp, 4),
"state": state.dict(),
"ip": request.client.host,
}
fout.write(json.dumps(data) + "\n")
block_css = (
code_highlight_css
+ """
pre {
white-space: pre-wrap; /* Since CSS 2.1 */
white-space: -moz-pre-wrap; /* Mozilla, since 1999 */
white-space: -pre-wrap; /* Opera 4-6 */
white-space: -o-pre-wrap; /* Opera 7 */
word-wrap: break-word; /* Internet Explorer 5.5+ */
}
#notice_markdown th {
display: none;
}
#notice_markdown {
text-align: center;
background: #0b5087;
padding: 1%;
height: 4.3rem;
color: #fff !important;
margin-top: 0;
}
#notice_markdown p{
color: #fff !important;
}
#notice_markdown h1, #notice_markdown h4 {
color: #fff;
margin-top: 0;
}
gradio-app {
background: linear-gradient(to bottom, #86ccf5, #3273bf) !important;
padding: 3%;
}
.gradio-container {
margin: 0 auto !important;
width: 70% !important;
padding: 0 !important;
background: #fff !important;
border-radius: 5px !important;
}
#chatbot {
border-style: solid;
overflow: visible;
margin: 1% 4%;
width: 90%;
box-shadow: 0 15px 15px -5px rgba(0, 0, 0, 0.2);
border: 1px solid #ddd;
}
#text-box-style, #btn-style {
width: 90%;
margin: 1% 4%;
}
.user, .bot {
width: 80% !important;
}
.bot {
white-space: pre-wrap !important;
line-height: 1.3 !important;
display: flex;
flex-direction: column;
justify-content: flex-start;
}
#btn-send-style {
background: rgb(0, 180, 50);
color: #fff;
}
#btn-list-style {
background: #eee0;
border: 1px solid #691ef7;
}
.title {
font-size: 1.5rem;
font-weight: 700;
color: #fff !important;
}
footer {
display: none !important;
}
.footer {
margin-bottom: 45px;
margin-top: 35px;
text-align: center;
border-bottom: 1px solid #e5e5e5;
}
.footer>p {
font-size: .8rem;
display: inline-block;
padding: 0 10px;
transform: translateY(10px);
background: white;
}
.acknowledgments {
width: 80%;
margin: 0 auto;
margin-bottom: 3rem;
}
.img-logo-style {
width: 3.5rem;
float: left;
}
.img-logo-right-style {
width: 3.5rem;
float: right;
}
.neural-studio-img-style {
width: 50%;
height: 20%;
margin: 0 auto;
}
"""
)
def build_single_model_ui(models):
notice_markdown = """
<div class='title'>
NeuralChat
</div>
<p>deployed on 4th Gen Intel Xeon Scalable Processors codenamed Sapphire Rapids.</p>
"""
learn_more_markdown = """<div class="footer">
<img src='https://i.postimg.cc/Pfv4vV6R/Microsoft-Teams-image-23.png' class='img-logo-style'/><p>Powered by <a href="https://github.com/intel/intel-extension-for-transformers" style="text-decoration: underline;" target="_blank">Intel Extension for Transformers</a> and <a href="https://github.com/intel/intel-extension-for-pytorch" style="text-decoration: underline;" target="_blank">Intel Extension for PyTorch </a>
</p> <img src='https://i.postimg.cc/Pfv4vV6R/Microsoft-Teams-image-23.png' class='img-logo-right-style'/>
</div>
<div class="acknowledgments">
<img src="https://i.postimg.cc/cJ99RQZ1/Microsoft-Teams-image-72.png" class="neural-studio-img-style" >
<p><h4>Neural Studio</h4>
<a href="http://neuralstudio.intel.com/" style="text-decoration: underline;" target="_blank">Neural Studio</a> is a web service that showcases <b> AI acceleration </b> capabilities on Intel's CPU and GPU with the <b>Intel Neural Compressor.</b>
It includes demos of <b>popular AI applications </b> to demonstrate their capabilities, and NeuralChat is also included.</p>
</div>
"""
state = gr.State()
notice = gr.Markdown(notice_markdown, elem_id="notice_markdown")
with gr.Row(elem_id="model_selector_row", visible=False):
model_selector = gr.Dropdown(
choices=models,
value=models[0] if len(models) > 0 else "",
interactive=True,
show_label=False,
).style(container=False)
chatbot = grChatbot(elem_id="chatbot", visible=False).style(height=550)
with gr.Row(elem_id="text-box-style"):
with gr.Column(scale=20):
textbox = gr.Textbox(
show_label=False,
placeholder="Enter text and press ENTER",
visible=False,
).style(container=False)
with gr.Column(scale=1, min_width=50):
send_btn = gr.Button(value="Send", visible=False, elem_id="btn-send-style")
with gr.Accordion("Parameters", open=False, visible=False, elem_id="btn-style") as parameter_row:
temperature = gr.Slider(
minimum=0.0,
maximum=1.0,
value=0.95,
step=0.1,
interactive=True,
label="Temperature",
visible=False,
)
max_output_tokens = gr.Slider(
minimum=0,
maximum=1024,
value=512,
step=64,
interactive=True,
label="Max output tokens",
)
with gr.Row(visible=False, elem_id="btn-style") as button_row:
upvote_btn = gr.Button(value="👍 Upvote", interactive=False, visible=False, elem_id="btn-list-style")
downvote_btn = gr.Button(value="👎 Downvote", interactive=False, visible=False, elem_id="btn-list-style")
flag_btn = gr.Button(value="⚠️ Flag", interactive=False, visible=False, elem_id="btn-list-style")
# stop_btn = gr.Button(value="⏹️ Stop Generation", interactive=False)
regenerate_btn = gr.Button(value="🔄 Regenerate", interactive=False, elem_id="btn-list-style")
clear_btn = gr.Button(value="🗑️ Clear history", interactive=False, elem_id="btn-list-style")
gr.Markdown(learn_more_markdown)
# Register listeners
btn_list = [upvote_btn, downvote_btn, flag_btn, regenerate_btn, clear_btn]
upvote_btn.click(
upvote_last_response,
[state, model_selector],
[textbox, upvote_btn, downvote_btn, flag_btn],
)
downvote_btn.click(
downvote_last_response,
[state, model_selector],
[textbox, upvote_btn, downvote_btn, flag_btn],
)
flag_btn.click(
flag_last_response,
[state, model_selector],
[textbox, upvote_btn, downvote_btn, flag_btn],
)
regenerate_btn.click(regenerate, state, [state, chatbot, textbox] + btn_list).then(
http_bot,
[state, model_selector, temperature, max_output_tokens],
[state, chatbot] + btn_list,
)
clear_btn.click(clear_history, None, [state, chatbot, textbox] + btn_list)
model_selector.change(clear_history, None, [state, chatbot, textbox] + btn_list)
textbox.submit(
add_text, [state, textbox], [state, chatbot, textbox] + btn_list
).then(
http_bot,
[state, model_selector, temperature, max_output_tokens],
[state, chatbot] + btn_list,
)
send_btn.click(
add_text, [state, textbox], [state, chatbot, textbox] + btn_list
).then(
http_bot,
[state, model_selector, temperature, max_output_tokens],
[state, chatbot] + btn_list,
)
return state, model_selector, chatbot, textbox, send_btn, button_row, parameter_row
def build_demo(models):
with gr.Blocks(
title="NeuralChat · Intel",
theme=gr.themes.Base(),
css=block_css,
) as demo:
url_params = gr.JSON(visible=False)
(
state,
model_selector,
chatbot,
textbox,
send_btn,
button_row,
parameter_row,
) = build_single_model_ui(models)
if model_list_mode == "once":
demo.load(
load_demo,
[url_params],
[
state,
model_selector,
chatbot,
textbox,
send_btn,
button_row,
parameter_row,
],
_js=get_window_url_params,
)
else:
raise ValueError(f"Unknown model list mode: {model_list_mode}")
return demo
if __name__ == "__main__":
controller_url = "http://35.172.200.58:80"
host = "0.0.0.0"
# port = "mlp-dgx-01.sh.intel.com"
concurrency_count = 10
model_list_mode = "once"
share = False
moderate = False
set_global_vars(controller_url, moderate)
models = get_model_list(controller_url)
demo = build_demo(models)
demo.queue(
concurrency_count=concurrency_count, status_update_rate=10, api_open=False
).launch(
server_name=host, share=share, max_threads=200
)
|