NCTCMumbai
commited on
Update app.py
Browse files
app.py
CHANGED
@@ -1,437 +1,437 @@
|
|
1 |
-
|
2 |
-
from ragatouille import RAGPretrainedModel
|
3 |
-
import subprocess
|
4 |
-
import json
|
5 |
-
import spaces
|
6 |
-
import firebase_admin
|
7 |
-
from firebase_admin import credentials, firestore
|
8 |
-
import logging
|
9 |
-
from pathlib import Path
|
10 |
-
from time import perf_counter
|
11 |
-
from datetime import datetime
|
12 |
-
import gradio as gr
|
13 |
-
from jinja2 import Environment, FileSystemLoader
|
14 |
-
import numpy as np
|
15 |
-
from sentence_transformers import CrossEncoder
|
16 |
-
from huggingface_hub import InferenceClient
|
17 |
-
from os import getenv
|
18 |
-
|
19 |
-
from backend.query_llm import generate_hf, generate_openai
|
20 |
-
from backend.semantic_search import table, retriever
|
21 |
-
from huggingface_hub import InferenceClient
|
22 |
-
|
23 |
-
|
24 |
-
VECTOR_COLUMN_NAME = "vector"
|
25 |
-
TEXT_COLUMN_NAME = "text"
|
26 |
-
HF_TOKEN = getenv("HUGGING_FACE_HUB_TOKEN")
|
27 |
-
proj_dir = Path(__file__).parent
|
28 |
-
# Setting up the logging
|
29 |
-
logging.basicConfig(level=logging.INFO)
|
30 |
-
logger = logging.getLogger(__name__)
|
31 |
-
client = InferenceClient("mistralai/Mixtral-8x7B-Instruct-v0.1",token=HF_TOKEN)
|
32 |
-
# Set up the template environment with the templates directory
|
33 |
-
env = Environment(loader=FileSystemLoader(proj_dir / 'templates'))
|
34 |
-
|
35 |
-
# Load the templates directly from the environment
|
36 |
-
template = env.get_template('template.j2')
|
37 |
-
template_html = env.get_template('template_html.j2')
|
38 |
-
#___________________
|
39 |
-
# service_account_key='firebase.json'
|
40 |
-
# # Create a Certificate object from the service account info
|
41 |
-
# cred = credentials.Certificate(service_account_key)
|
42 |
-
# # Initialize the Firebase Admin
|
43 |
-
# firebase_admin.initialize_app(cred)
|
44 |
-
|
45 |
-
# # # Create a reference to the Firestore database
|
46 |
-
# db = firestore.client()
|
47 |
-
# #db usage
|
48 |
-
# collection_name = 'Nirvachana' # Replace with your collection name
|
49 |
-
# field_name = 'message_count' # Replace with your field name for count
|
50 |
-
# Examples
|
51 |
-
examples = ['Tabulate the difference between veins and arteries','What are defects in Human eye?',
|
52 |
-
'Frame 5 short questions and 5 MCQ on Chapter 2 ','Suggest creative and engaging ideas to teach students on Chapter on Metals and Non Metals '
|
53 |
-
]
|
54 |
-
|
55 |
-
|
56 |
-
|
57 |
-
# def get_and_increment_value_count(db , collection_name, field_name):
|
58 |
-
# """
|
59 |
-
# Retrieves a value count from the specified Firestore collection and field,
|
60 |
-
# increments it by 1, and updates the field with the new value."""
|
61 |
-
# collection_ref = db.collection(collection_name)
|
62 |
-
# doc_ref = collection_ref.document('count_doc') # Assuming a dedicated document for count
|
63 |
-
|
64 |
-
# # Use a transaction to ensure consistency across reads and writes
|
65 |
-
# try:
|
66 |
-
# with db.transaction() as transaction:
|
67 |
-
# # Get the current value count (or initialize to 0 if it doesn't exist)
|
68 |
-
# current_count_doc = doc_ref.get()
|
69 |
-
# current_count_data = current_count_doc.to_dict()
|
70 |
-
# if current_count_data:
|
71 |
-
# current_count = current_count_data.get(field_name, 0)
|
72 |
-
# else:
|
73 |
-
# current_count = 0
|
74 |
-
# # Increment the count
|
75 |
-
# new_count = current_count + 1
|
76 |
-
# # Update the document with the new count
|
77 |
-
# transaction.set(doc_ref, {field_name: new_count})
|
78 |
-
# return new_count
|
79 |
-
# except Exception as e:
|
80 |
-
# print(f"Error retrieving and updating value count: {e}")
|
81 |
-
# return None # Indicate error
|
82 |
-
|
83 |
-
# def update_count_html():
|
84 |
-
# usage_count = get_and_increment_value_count(db ,collection_name, field_name)
|
85 |
-
# ccount_html = gr.HTML(value=f"""
|
86 |
-
# <div style="display: flex; justify-content: flex-end;">
|
87 |
-
# <span style="font-weight: bold; color: maroon; font-size: 18px;">No of Usages:</span>
|
88 |
-
# <span style="font-weight: bold; color: maroon; font-size: 18px;">{usage_count}</span>
|
89 |
-
# </div>
|
90 |
-
# """)
|
91 |
-
# return count_html
|
92 |
-
|
93 |
-
# def store_message(db,query,answer,cross_encoder):
|
94 |
-
# timestamp = datetime.now().strftime("%Y-%m-%d_%H-%M-%S")
|
95 |
-
# # Create a new document reference with a dynamic document name based on timestamp
|
96 |
-
# new_completion= db.collection('Nirvachana').document(f"chatlogs_{timestamp}")
|
97 |
-
# new_completion.set({
|
98 |
-
# 'query': query,
|
99 |
-
# 'answer':answer,
|
100 |
-
# 'created_time': firestore.SERVER_TIMESTAMP,
|
101 |
-
# 'embedding': cross_encoder,
|
102 |
-
# 'title': 'Expenditure observer bot'
|
103 |
-
# })
|
104 |
-
|
105 |
-
|
106 |
-
def add_text(history, text):
|
107 |
-
history = [] if history is None else history
|
108 |
-
history = history + [(text, None)]
|
109 |
-
return history, gr.Textbox(value="", interactive=False)
|
110 |
-
|
111 |
-
|
112 |
-
def bot(history, cross_encoder):
|
113 |
-
top_rerank = 25
|
114 |
-
top_k_rank = 20
|
115 |
-
query = history[-1][0]
|
116 |
-
|
117 |
-
if not query:
|
118 |
-
gr.Warning("Please submit a non-empty string as a prompt")
|
119 |
-
raise ValueError("Empty string was submitted")
|
120 |
-
|
121 |
-
logger.warning('Retrieving documents...')
|
122 |
-
|
123 |
-
# if COLBERT RAGATATOUILLE PROCEDURE :
|
124 |
-
if cross_encoder=='(HIGH ACCURATE) ColBERT':
|
125 |
-
gr.Warning('Retrieving using ColBERT.. First time query will take a minute for model to load..pls wait')
|
126 |
-
RAG= RAGPretrainedModel.from_pretrained("colbert-ir/colbertv2.0")
|
127 |
-
RAG_db=RAG.from_index('.ragatouille/colbert/indexes/cbseclass10index')
|
128 |
-
documents_full=RAG_db.search(query,k=top_k_rank)
|
129 |
-
|
130 |
-
documents=[item['content'] for item in documents_full]
|
131 |
-
# Create Prompt
|
132 |
-
prompt = template.render(documents=documents, query=query)
|
133 |
-
prompt_html = template_html.render(documents=documents, query=query)
|
134 |
-
|
135 |
-
generate_fn = generate_hf
|
136 |
-
|
137 |
-
history[-1][1] = ""
|
138 |
-
for character in generate_fn(prompt, history[:-1]):
|
139 |
-
history[-1][1] = character
|
140 |
-
yield history, prompt_html
|
141 |
-
print('Final history is ',history)
|
142 |
-
#store_message(db,history[-1][0],history[-1][1],cross_encoder)
|
143 |
-
else:
|
144 |
-
# Retrieve documents relevant to query
|
145 |
-
document_start = perf_counter()
|
146 |
-
|
147 |
-
query_vec = retriever.encode(query)
|
148 |
-
logger.warning(f'Finished query vec')
|
149 |
-
doc1 = table.search(query_vec, vector_column_name=VECTOR_COLUMN_NAME).limit(top_k_rank)
|
150 |
-
|
151 |
-
|
152 |
-
|
153 |
-
logger.warning(f'Finished search')
|
154 |
-
documents = table.search(query_vec, vector_column_name=VECTOR_COLUMN_NAME).limit(top_rerank).to_list()
|
155 |
-
documents = [doc[TEXT_COLUMN_NAME] for doc in documents]
|
156 |
-
logger.warning(f'start cross encoder {len(documents)}')
|
157 |
-
# Retrieve documents relevant to query
|
158 |
-
query_doc_pair = [[query, doc] for doc in documents]
|
159 |
-
if cross_encoder=='(FAST) MiniLM-L6v2' :
|
160 |
-
cross_encoder1 = CrossEncoder('cross-encoder/ms-marco-MiniLM-L-6-v2')
|
161 |
-
elif cross_encoder=='(ACCURATE) BGE reranker':
|
162 |
-
cross_encoder1 = CrossEncoder('BAAI/bge-reranker-base')
|
163 |
-
|
164 |
-
cross_scores = cross_encoder1.predict(query_doc_pair)
|
165 |
-
sim_scores_argsort = list(reversed(np.argsort(cross_scores)))
|
166 |
-
logger.warning(f'Finished cross encoder {len(documents)}')
|
167 |
-
|
168 |
-
documents = [documents[idx] for idx in sim_scores_argsort[:top_k_rank]]
|
169 |
-
logger.warning(f'num documents {len(documents)}')
|
170 |
-
|
171 |
-
document_time = perf_counter() - document_start
|
172 |
-
logger.warning(f'Finished Retrieving documents in {round(document_time, 2)} seconds...')
|
173 |
-
|
174 |
-
# Create Prompt
|
175 |
-
prompt = template.render(documents=documents, query=query)
|
176 |
-
prompt_html = template_html.render(documents=documents, query=query)
|
177 |
-
|
178 |
-
generate_fn = generate_hf
|
179 |
-
|
180 |
-
history[-1][1] = ""
|
181 |
-
for character in generate_fn(prompt, history[:-1]):
|
182 |
-
history[-1][1] = character
|
183 |
-
yield history, prompt_html
|
184 |
-
print('Final history is ',history)
|
185 |
-
#store_message(db,history[-1][0],history[-1][1],cross_encoder)
|
186 |
-
|
187 |
-
def system_instructions(question_difficulty, topic,documents_str):
|
188 |
-
return f"""<s> [INST] Your are a great teacher and your task is to create 10 questions with 4 choices with a {question_difficulty} difficulty about topic request " {topic} " only from the below given documents, {documents_str} then create an answers. Index in JSON format, the questions as "Q#":"" to "Q#":"", the four choices as "Q#:C1":"" to "Q#:C4":"", and the answers as "A#":"Q#:C#" to "A#":"Q#:C#". [/INST]"""
|
189 |
-
|
190 |
-
|
191 |
-
#with gr.Blocks(theme='Insuz/SimpleIndigo') as demo:
|
192 |
-
with gr.Blocks(theme='NoCrypt/miku') as CHATBOT:
|
193 |
-
with gr.Row():
|
194 |
-
with gr.Column(scale=10):
|
195 |
-
# gr.Markdown(
|
196 |
-
# """
|
197 |
-
# # Theme preview: `paris`
|
198 |
-
# To use this theme, set `theme='earneleh/paris'` in `gr.Blocks()` or `gr.Interface()`.
|
199 |
-
# You can append an `@` and a semantic version expression, e.g. @>=1.0.0,<2.0.0 to pin to a given version
|
200 |
-
# of this theme.
|
201 |
-
# """
|
202 |
-
# )
|
203 |
-
gr.HTML(value="""<div style="color: #FF4500;"><h1>CHEERFULL CBSE-</h1> <h1><span style="color: #008000">AI Assisted Fun Learning</span></h1>
|
204 |
-
</div>""", elem_id='heading')
|
205 |
-
|
206 |
-
gr.HTML(value=f"""
|
207 |
-
<p style="font-family: sans-serif; font-size: 16px;">
|
208 |
-
A free Artificial Intelligence Chatbot assistant trained on CBSE Class 10 Science Notes to engage and help students and teachers of Puducherry.
|
209 |
-
</p>
|
210 |
-
""", elem_id='Sub-heading')
|
211 |
-
#usage_count = get_and_increment_value_count(db,collection_name, field_name)
|
212 |
-
gr.HTML(value=f"""<p style="font-family: Arial, sans-serif; font-size: 14px;">Developed by K M Ramyasri , TGT,GHS.SUTHUKENY . Suggestions may be sent to <a href="mailto:ramyadevi1607@yahoo.com" style="color: #00008B; font-style: italic;">ramyadevi1607@yahoo.com</a>.</p>""", elem_id='Sub-heading1 ')
|
213 |
-
|
214 |
-
with gr.Column(scale=3):
|
215 |
-
gr.Image(value='logo.png',height=200,width=200)
|
216 |
-
|
217 |
-
|
218 |
-
# gr.HTML(value="""<div style="color: #FF4500;"><h1>CHEERFULL CBSE-</h1> <h1><span style="color: #008000">AI Assisted Fun Learning</span></h1>
|
219 |
-
# <img src='logo.png' alt="Chatbot" width="50" height="50" />
|
220 |
-
# </div>""", elem_id='heading')
|
221 |
-
|
222 |
-
# gr.HTML(value=f"""
|
223 |
-
# <p style="font-family: sans-serif; font-size: 16px;">
|
224 |
-
# A free Artificial Intelligence Chatbot assistant trained on CBSE Class 10 Science Notes to engage and help students and teachers of Puducherry.
|
225 |
-
# </p>
|
226 |
-
# """, elem_id='Sub-heading')
|
227 |
-
# #usage_count = get_and_increment_value_count(db,collection_name, field_name)
|
228 |
-
# gr.HTML(value=f"""<p style="font-family: Arial, sans-serif; font-size: 16px;">Developed by K M Ramyasri , PGT . Suggestions may be sent to <a href="mailto:ramyadevi1607@yahoo.com" style="color: #00008B; font-style: italic;">ramyadevi1607@yahoo.com</a>.</p>""", elem_id='Sub-heading1 ')
|
229 |
-
# # count_html = gr.HTML(value=f"""
|
230 |
-
# # <div style="display: flex; justify-content: flex-end;">
|
231 |
-
# # <span style="font-weight: bold; color: maroon; font-size: 18px;">No of Usages:</span>
|
232 |
-
# # <span style="font-weight: bold; color: maroon; font-size: 18px;">{usage_count}</span>
|
233 |
-
# # </div>
|
234 |
-
# # """)
|
235 |
-
|
236 |
-
chatbot = gr.Chatbot(
|
237 |
-
[],
|
238 |
-
elem_id="chatbot",
|
239 |
-
avatar_images=('https://aui.atlassian.com/aui/8.8/docs/images/avatar-person.svg',
|
240 |
-
'https://huggingface.co/datasets/huggingface/brand-assets/resolve/main/hf-logo.svg'),
|
241 |
-
bubble_full_width=False,
|
242 |
-
show_copy_button=True,
|
243 |
-
show_share_button=True,
|
244 |
-
)
|
245 |
-
|
246 |
-
with gr.Row():
|
247 |
-
txt = gr.Textbox(
|
248 |
-
scale=3,
|
249 |
-
show_label=False,
|
250 |
-
placeholder="Enter text and press enter",
|
251 |
-
container=False,
|
252 |
-
)
|
253 |
-
txt_btn = gr.Button(value="Submit text", scale=1)
|
254 |
-
|
255 |
-
cross_encoder = gr.Radio(choices=['(FAST) MiniLM-L6v2','(ACCURATE) BGE reranker','(HIGH ACCURATE) ColBERT'], value='(ACCURATE) BGE reranker',label="Embeddings", info="Only First query to Colbert may take litte time)")
|
256 |
-
|
257 |
-
prompt_html = gr.HTML()
|
258 |
-
# Turn off interactivity while generating if you click
|
259 |
-
txt_msg = txt_btn.click(add_text, [chatbot, txt], [chatbot, txt], queue=False).then(
|
260 |
-
bot, [chatbot, cross_encoder], [chatbot, prompt_html])#.then(update_count_html,[],[count_html])
|
261 |
-
|
262 |
-
# Turn it back on
|
263 |
-
txt_msg.then(lambda: gr.Textbox(interactive=True), None, [txt], queue=False)
|
264 |
-
|
265 |
-
# Turn off interactivity while generating if you hit enter
|
266 |
-
txt_msg = txt.submit(add_text, [chatbot, txt], [chatbot, txt], queue=False).then(
|
267 |
-
bot, [chatbot, cross_encoder], [chatbot, prompt_html])#.then(update_count_html,[],[count_html])
|
268 |
-
|
269 |
-
# Turn it back on
|
270 |
-
txt_msg.then(lambda: gr.Textbox(interactive=True), None, [txt], queue=False)
|
271 |
-
|
272 |
-
# Examples
|
273 |
-
gr.Examples(examples, txt)
|
274 |
-
|
275 |
-
|
276 |
-
RAG_db=gr.State()
|
277 |
-
|
278 |
-
with gr.Blocks(title="Quiz Maker", theme=gr.themes.Default(primary_hue="green", secondary_hue="green"), css="style.css") as QUIZBOT:
|
279 |
-
def load_model():
|
280 |
-
RAG= RAGPretrainedModel.from_pretrained("colbert-ir/colbertv2.0")
|
281 |
-
RAG_db.value=RAG.from_index('.ragatouille/colbert/indexes/cbseclass10index')
|
282 |
-
return 'Ready to Go!!'
|
283 |
-
with gr.Column(scale=4):
|
284 |
-
gr.HTML("""
|
285 |
-
<center>
|
286 |
-
<h1><span style="color: purple;">
|
287 |
-
<h2>AI-powered
|
288 |
-
<i>⚠️
|
289 |
-
</center>
|
290 |
-
""")
|
291 |
-
#gr.Warning('Retrieving using ColBERT.. First time query will take a minute for model to load..pls wait')
|
292 |
-
with gr.Column(scale=2):
|
293 |
-
load_btn = gr.Button("Click to Load!🚀")
|
294 |
-
load_text=gr.Textbox()
|
295 |
-
load_btn.click(load_model,[],load_text)
|
296 |
-
|
297 |
-
|
298 |
-
topic = gr.Textbox(label="Enter the Topic for Quiz", placeholder="Write any topic from
|
299 |
-
|
300 |
-
with gr.Row():
|
301 |
-
radio = gr.Radio(
|
302 |
-
["easy", "average", "hard"], label="How difficult should the quiz be?"
|
303 |
-
)
|
304 |
-
|
305 |
-
|
306 |
-
generate_quiz_btn = gr.Button("Generate Quiz!🚀")
|
307 |
-
quiz_msg=gr.Textbox()
|
308 |
-
|
309 |
-
question_radios = [gr.Radio(visible=False), gr.Radio(visible=False), gr.Radio(
|
310 |
-
visible=False), gr.Radio(visible=False), gr.Radio(visible=False), gr.Radio(visible=False), gr.Radio(visible=False), gr.Radio(
|
311 |
-
visible=False), gr.Radio(visible=False), gr.Radio(visible=False)]
|
312 |
-
|
313 |
-
print(question_radios)
|
314 |
-
|
315 |
-
@spaces.GPU
|
316 |
-
@generate_quiz_btn.click(inputs=[radio, topic], outputs=[quiz_msg]+question_radios, api_name="generate_quiz")
|
317 |
-
def generate_quiz(question_difficulty, topic):
|
318 |
-
top_k_rank=10
|
319 |
-
RAG_db_=RAG_db.value
|
320 |
-
documents_full=RAG_db_.search(topic,k=top_k_rank)
|
321 |
-
|
322 |
-
|
323 |
-
|
324 |
-
generate_kwargs = dict(
|
325 |
-
temperature=0.2,
|
326 |
-
max_new_tokens=4000,
|
327 |
-
top_p=0.95,
|
328 |
-
repetition_penalty=1.0,
|
329 |
-
do_sample=True,
|
330 |
-
seed=42,
|
331 |
-
)
|
332 |
-
question_radio_list = []
|
333 |
-
count=0
|
334 |
-
while count<=3:
|
335 |
-
try:
|
336 |
-
documents=[item['content'] for item in documents_full]
|
337 |
-
document_summaries = [f"[DOCUMENT {i+1}]: {summary}{count}" for i, summary in enumerate(documents)]
|
338 |
-
documents_str='\n'.join(document_summaries)
|
339 |
-
formatted_prompt = system_instructions(
|
340 |
-
question_difficulty, topic,documents_str)
|
341 |
-
print(formatted_prompt)
|
342 |
-
pre_prompt = [
|
343 |
-
{"role": "system", "content": formatted_prompt}
|
344 |
-
]
|
345 |
-
response = client.text_generation(
|
346 |
-
formatted_prompt, **generate_kwargs, stream=False, details=False, return_full_text=False,
|
347 |
-
)
|
348 |
-
output_json = json.loads(f"{response}")
|
349 |
-
|
350 |
-
|
351 |
-
print(response)
|
352 |
-
print('output json', output_json)
|
353 |
-
|
354 |
-
global quiz_data
|
355 |
-
|
356 |
-
quiz_data = output_json
|
357 |
-
|
358 |
-
|
359 |
-
|
360 |
-
for question_num in range(1, 11):
|
361 |
-
question_key = f"Q{question_num}"
|
362 |
-
answer_key = f"A{question_num}"
|
363 |
-
|
364 |
-
question = quiz_data.get(question_key)
|
365 |
-
answer = quiz_data.get(quiz_data.get(answer_key))
|
366 |
-
|
367 |
-
if not question or not answer:
|
368 |
-
continue
|
369 |
-
|
370 |
-
choice_keys = [f"{question_key}:C{i}" for i in range(1, 5)]
|
371 |
-
choice_list = []
|
372 |
-
for choice_key in choice_keys:
|
373 |
-
choice = quiz_data.get(choice_key, "Choice not found")
|
374 |
-
choice_list.append(f"{choice}")
|
375 |
-
|
376 |
-
radio = gr.Radio(choices=choice_list, label=question,
|
377 |
-
visible=True, interactive=True)
|
378 |
-
|
379 |
-
question_radio_list.append(radio)
|
380 |
-
if len(question_radio_list)==10:
|
381 |
-
break
|
382 |
-
else:
|
383 |
-
print('10 questions not generated . So trying again!')
|
384 |
-
count+=1
|
385 |
-
continue
|
386 |
-
except Exception as e:
|
387 |
-
count+=1
|
388 |
-
print(f"Exception occurred: {e}")
|
389 |
-
if count==3:
|
390 |
-
print('Retry exhausted')
|
391 |
-
gr.Warning('Sorry. Pls try with another topic !')
|
392 |
-
else:
|
393 |
-
print(f"Trying again..{count} time...please wait")
|
394 |
-
continue
|
395 |
-
|
396 |
-
print('Question radio list ' , question_radio_list)
|
397 |
-
|
398 |
-
return ['Quiz Generated!']+ question_radio_list
|
399 |
-
|
400 |
-
check_button = gr.Button("Check Score")
|
401 |
-
|
402 |
-
score_textbox = gr.Markdown()
|
403 |
-
|
404 |
-
@check_button.click(inputs=question_radios, outputs=score_textbox)
|
405 |
-
def compare_answers(*user_answers):
|
406 |
-
user_anwser_list = []
|
407 |
-
user_anwser_list = user_answers
|
408 |
-
|
409 |
-
answers_list = []
|
410 |
-
|
411 |
-
for question_num in range(1, 20):
|
412 |
-
answer_key = f"A{question_num}"
|
413 |
-
answer = quiz_data.get(quiz_data.get(answer_key))
|
414 |
-
if not answer:
|
415 |
-
break
|
416 |
-
answers_list.append(answer)
|
417 |
-
|
418 |
-
score = 0
|
419 |
-
|
420 |
-
for item in user_anwser_list:
|
421 |
-
if item in answers_list:
|
422 |
-
score += 1
|
423 |
-
if score>5:
|
424 |
-
message = f"### Good ! You got {score} over 10!"
|
425 |
-
elif score>7:
|
426 |
-
message = f"### Excellent ! You got {score} over 10!"
|
427 |
-
else:
|
428 |
-
message = f"### You got {score} over 10! Dont worry . You can prepare well and try better next time !"
|
429 |
-
|
430 |
-
return message
|
431 |
-
|
432 |
-
|
433 |
-
|
434 |
-
demo = gr.TabbedInterface([CHATBOT,QUIZBOT], ["AI ChatBot", "AI
|
435 |
-
|
436 |
-
demo.queue()
|
437 |
-
demo.launch(debug=True)
|
|
|
1 |
+
|
2 |
+
from ragatouille import RAGPretrainedModel
|
3 |
+
import subprocess
|
4 |
+
import json
|
5 |
+
import spaces
|
6 |
+
import firebase_admin
|
7 |
+
from firebase_admin import credentials, firestore
|
8 |
+
import logging
|
9 |
+
from pathlib import Path
|
10 |
+
from time import perf_counter
|
11 |
+
from datetime import datetime
|
12 |
+
import gradio as gr
|
13 |
+
from jinja2 import Environment, FileSystemLoader
|
14 |
+
import numpy as np
|
15 |
+
from sentence_transformers import CrossEncoder
|
16 |
+
from huggingface_hub import InferenceClient
|
17 |
+
from os import getenv
|
18 |
+
|
19 |
+
from backend.query_llm import generate_hf, generate_openai
|
20 |
+
from backend.semantic_search import table, retriever
|
21 |
+
from huggingface_hub import InferenceClient
|
22 |
+
|
23 |
+
|
24 |
+
VECTOR_COLUMN_NAME = "vector"
|
25 |
+
TEXT_COLUMN_NAME = "text"
|
26 |
+
HF_TOKEN = getenv("HUGGING_FACE_HUB_TOKEN")
|
27 |
+
proj_dir = Path(__file__).parent
|
28 |
+
# Setting up the logging
|
29 |
+
logging.basicConfig(level=logging.INFO)
|
30 |
+
logger = logging.getLogger(__name__)
|
31 |
+
client = InferenceClient("mistralai/Mixtral-8x7B-Instruct-v0.1",token=HF_TOKEN)
|
32 |
+
# Set up the template environment with the templates directory
|
33 |
+
env = Environment(loader=FileSystemLoader(proj_dir / 'templates'))
|
34 |
+
|
35 |
+
# Load the templates directly from the environment
|
36 |
+
template = env.get_template('template.j2')
|
37 |
+
template_html = env.get_template('template_html.j2')
|
38 |
+
#___________________
|
39 |
+
# service_account_key='firebase.json'
|
40 |
+
# # Create a Certificate object from the service account info
|
41 |
+
# cred = credentials.Certificate(service_account_key)
|
42 |
+
# # Initialize the Firebase Admin
|
43 |
+
# firebase_admin.initialize_app(cred)
|
44 |
+
|
45 |
+
# # # Create a reference to the Firestore database
|
46 |
+
# db = firestore.client()
|
47 |
+
# #db usage
|
48 |
+
# collection_name = 'Nirvachana' # Replace with your collection name
|
49 |
+
# field_name = 'message_count' # Replace with your field name for count
|
50 |
+
# Examples
|
51 |
+
examples = ['Tabulate the difference between veins and arteries','What are defects in Human eye?',
|
52 |
+
'Frame 5 short questions and 5 MCQ on Chapter 2 ','Suggest creative and engaging ideas to teach students on Chapter on Metals and Non Metals '
|
53 |
+
]
|
54 |
+
|
55 |
+
|
56 |
+
|
57 |
+
# def get_and_increment_value_count(db , collection_name, field_name):
|
58 |
+
# """
|
59 |
+
# Retrieves a value count from the specified Firestore collection and field,
|
60 |
+
# increments it by 1, and updates the field with the new value."""
|
61 |
+
# collection_ref = db.collection(collection_name)
|
62 |
+
# doc_ref = collection_ref.document('count_doc') # Assuming a dedicated document for count
|
63 |
+
|
64 |
+
# # Use a transaction to ensure consistency across reads and writes
|
65 |
+
# try:
|
66 |
+
# with db.transaction() as transaction:
|
67 |
+
# # Get the current value count (or initialize to 0 if it doesn't exist)
|
68 |
+
# current_count_doc = doc_ref.get()
|
69 |
+
# current_count_data = current_count_doc.to_dict()
|
70 |
+
# if current_count_data:
|
71 |
+
# current_count = current_count_data.get(field_name, 0)
|
72 |
+
# else:
|
73 |
+
# current_count = 0
|
74 |
+
# # Increment the count
|
75 |
+
# new_count = current_count + 1
|
76 |
+
# # Update the document with the new count
|
77 |
+
# transaction.set(doc_ref, {field_name: new_count})
|
78 |
+
# return new_count
|
79 |
+
# except Exception as e:
|
80 |
+
# print(f"Error retrieving and updating value count: {e}")
|
81 |
+
# return None # Indicate error
|
82 |
+
|
83 |
+
# def update_count_html():
|
84 |
+
# usage_count = get_and_increment_value_count(db ,collection_name, field_name)
|
85 |
+
# ccount_html = gr.HTML(value=f"""
|
86 |
+
# <div style="display: flex; justify-content: flex-end;">
|
87 |
+
# <span style="font-weight: bold; color: maroon; font-size: 18px;">No of Usages:</span>
|
88 |
+
# <span style="font-weight: bold; color: maroon; font-size: 18px;">{usage_count}</span>
|
89 |
+
# </div>
|
90 |
+
# """)
|
91 |
+
# return count_html
|
92 |
+
|
93 |
+
# def store_message(db,query,answer,cross_encoder):
|
94 |
+
# timestamp = datetime.now().strftime("%Y-%m-%d_%H-%M-%S")
|
95 |
+
# # Create a new document reference with a dynamic document name based on timestamp
|
96 |
+
# new_completion= db.collection('Nirvachana').document(f"chatlogs_{timestamp}")
|
97 |
+
# new_completion.set({
|
98 |
+
# 'query': query,
|
99 |
+
# 'answer':answer,
|
100 |
+
# 'created_time': firestore.SERVER_TIMESTAMP,
|
101 |
+
# 'embedding': cross_encoder,
|
102 |
+
# 'title': 'Expenditure observer bot'
|
103 |
+
# })
|
104 |
+
|
105 |
+
|
106 |
+
def add_text(history, text):
|
107 |
+
history = [] if history is None else history
|
108 |
+
history = history + [(text, None)]
|
109 |
+
return history, gr.Textbox(value="", interactive=False)
|
110 |
+
|
111 |
+
|
112 |
+
def bot(history, cross_encoder):
|
113 |
+
top_rerank = 25
|
114 |
+
top_k_rank = 20
|
115 |
+
query = history[-1][0]
|
116 |
+
|
117 |
+
if not query:
|
118 |
+
gr.Warning("Please submit a non-empty string as a prompt")
|
119 |
+
raise ValueError("Empty string was submitted")
|
120 |
+
|
121 |
+
logger.warning('Retrieving documents...')
|
122 |
+
|
123 |
+
# if COLBERT RAGATATOUILLE PROCEDURE :
|
124 |
+
if cross_encoder=='(HIGH ACCURATE) ColBERT':
|
125 |
+
gr.Warning('Retrieving using ColBERT.. First time query will take a minute for model to load..pls wait')
|
126 |
+
RAG= RAGPretrainedModel.from_pretrained("colbert-ir/colbertv2.0")
|
127 |
+
RAG_db=RAG.from_index('.ragatouille/colbert/indexes/cbseclass10index')
|
128 |
+
documents_full=RAG_db.search(query,k=top_k_rank)
|
129 |
+
|
130 |
+
documents=[item['content'] for item in documents_full]
|
131 |
+
# Create Prompt
|
132 |
+
prompt = template.render(documents=documents, query=query)
|
133 |
+
prompt_html = template_html.render(documents=documents, query=query)
|
134 |
+
|
135 |
+
generate_fn = generate_hf
|
136 |
+
|
137 |
+
history[-1][1] = ""
|
138 |
+
for character in generate_fn(prompt, history[:-1]):
|
139 |
+
history[-1][1] = character
|
140 |
+
yield history, prompt_html
|
141 |
+
print('Final history is ',history)
|
142 |
+
#store_message(db,history[-1][0],history[-1][1],cross_encoder)
|
143 |
+
else:
|
144 |
+
# Retrieve documents relevant to query
|
145 |
+
document_start = perf_counter()
|
146 |
+
|
147 |
+
query_vec = retriever.encode(query)
|
148 |
+
logger.warning(f'Finished query vec')
|
149 |
+
doc1 = table.search(query_vec, vector_column_name=VECTOR_COLUMN_NAME).limit(top_k_rank)
|
150 |
+
|
151 |
+
|
152 |
+
|
153 |
+
logger.warning(f'Finished search')
|
154 |
+
documents = table.search(query_vec, vector_column_name=VECTOR_COLUMN_NAME).limit(top_rerank).to_list()
|
155 |
+
documents = [doc[TEXT_COLUMN_NAME] for doc in documents]
|
156 |
+
logger.warning(f'start cross encoder {len(documents)}')
|
157 |
+
# Retrieve documents relevant to query
|
158 |
+
query_doc_pair = [[query, doc] for doc in documents]
|
159 |
+
if cross_encoder=='(FAST) MiniLM-L6v2' :
|
160 |
+
cross_encoder1 = CrossEncoder('cross-encoder/ms-marco-MiniLM-L-6-v2')
|
161 |
+
elif cross_encoder=='(ACCURATE) BGE reranker':
|
162 |
+
cross_encoder1 = CrossEncoder('BAAI/bge-reranker-base')
|
163 |
+
|
164 |
+
cross_scores = cross_encoder1.predict(query_doc_pair)
|
165 |
+
sim_scores_argsort = list(reversed(np.argsort(cross_scores)))
|
166 |
+
logger.warning(f'Finished cross encoder {len(documents)}')
|
167 |
+
|
168 |
+
documents = [documents[idx] for idx in sim_scores_argsort[:top_k_rank]]
|
169 |
+
logger.warning(f'num documents {len(documents)}')
|
170 |
+
|
171 |
+
document_time = perf_counter() - document_start
|
172 |
+
logger.warning(f'Finished Retrieving documents in {round(document_time, 2)} seconds...')
|
173 |
+
|
174 |
+
# Create Prompt
|
175 |
+
prompt = template.render(documents=documents, query=query)
|
176 |
+
prompt_html = template_html.render(documents=documents, query=query)
|
177 |
+
|
178 |
+
generate_fn = generate_hf
|
179 |
+
|
180 |
+
history[-1][1] = ""
|
181 |
+
for character in generate_fn(prompt, history[:-1]):
|
182 |
+
history[-1][1] = character
|
183 |
+
yield history, prompt_html
|
184 |
+
print('Final history is ',history)
|
185 |
+
#store_message(db,history[-1][0],history[-1][1],cross_encoder)
|
186 |
+
|
187 |
+
def system_instructions(question_difficulty, topic,documents_str):
|
188 |
+
return f"""<s> [INST] Your are a great teacher and your task is to create 10 questions with 4 choices with a {question_difficulty} difficulty about topic request " {topic} " only from the below given documents, {documents_str} then create an answers. Index in JSON format, the questions as "Q#":"" to "Q#":"", the four choices as "Q#:C1":"" to "Q#:C4":"", and the answers as "A#":"Q#:C#" to "A#":"Q#:C#". [/INST]"""
|
189 |
+
|
190 |
+
|
191 |
+
#with gr.Blocks(theme='Insuz/SimpleIndigo') as demo:
|
192 |
+
with gr.Blocks(theme='NoCrypt/miku') as CHATBOT:
|
193 |
+
with gr.Row():
|
194 |
+
with gr.Column(scale=10):
|
195 |
+
# gr.Markdown(
|
196 |
+
# """
|
197 |
+
# # Theme preview: `paris`
|
198 |
+
# To use this theme, set `theme='earneleh/paris'` in `gr.Blocks()` or `gr.Interface()`.
|
199 |
+
# You can append an `@` and a semantic version expression, e.g. @>=1.0.0,<2.0.0 to pin to a given version
|
200 |
+
# of this theme.
|
201 |
+
# """
|
202 |
+
# )
|
203 |
+
gr.HTML(value="""<div style="color: #FF4500;"><h1>CHEERFULL CBSE-</h1> <h1><span style="color: #008000">AI Assisted Fun Learning</span></h1>
|
204 |
+
</div>""", elem_id='heading')
|
205 |
+
|
206 |
+
gr.HTML(value=f"""
|
207 |
+
<p style="font-family: sans-serif; font-size: 16px;">
|
208 |
+
A free Artificial Intelligence Chatbot assistant trained on CBSE Class 10 Science Notes to engage and help students and teachers of Puducherry.
|
209 |
+
</p>
|
210 |
+
""", elem_id='Sub-heading')
|
211 |
+
#usage_count = get_and_increment_value_count(db,collection_name, field_name)
|
212 |
+
gr.HTML(value=f"""<p style="font-family: Arial, sans-serif; font-size: 14px;">Developed by K M Ramyasri , TGT,GHS.SUTHUKENY . Suggestions may be sent to <a href="mailto:ramyadevi1607@yahoo.com" style="color: #00008B; font-style: italic;">ramyadevi1607@yahoo.com</a>.</p>""", elem_id='Sub-heading1 ')
|
213 |
+
|
214 |
+
with gr.Column(scale=3):
|
215 |
+
gr.Image(value='logo.png',height=200,width=200)
|
216 |
+
|
217 |
+
|
218 |
+
# gr.HTML(value="""<div style="color: #FF4500;"><h1>CHEERFULL CBSE-</h1> <h1><span style="color: #008000">AI Assisted Fun Learning</span></h1>
|
219 |
+
# <img src='logo.png' alt="Chatbot" width="50" height="50" />
|
220 |
+
# </div>""", elem_id='heading')
|
221 |
+
|
222 |
+
# gr.HTML(value=f"""
|
223 |
+
# <p style="font-family: sans-serif; font-size: 16px;">
|
224 |
+
# A free Artificial Intelligence Chatbot assistant trained on CBSE Class 10 Science Notes to engage and help students and teachers of Puducherry.
|
225 |
+
# </p>
|
226 |
+
# """, elem_id='Sub-heading')
|
227 |
+
# #usage_count = get_and_increment_value_count(db,collection_name, field_name)
|
228 |
+
# gr.HTML(value=f"""<p style="font-family: Arial, sans-serif; font-size: 16px;">Developed by K M Ramyasri , PGT . Suggestions may be sent to <a href="mailto:ramyadevi1607@yahoo.com" style="color: #00008B; font-style: italic;">ramyadevi1607@yahoo.com</a>.</p>""", elem_id='Sub-heading1 ')
|
229 |
+
# # count_html = gr.HTML(value=f"""
|
230 |
+
# # <div style="display: flex; justify-content: flex-end;">
|
231 |
+
# # <span style="font-weight: bold; color: maroon; font-size: 18px;">No of Usages:</span>
|
232 |
+
# # <span style="font-weight: bold; color: maroon; font-size: 18px;">{usage_count}</span>
|
233 |
+
# # </div>
|
234 |
+
# # """)
|
235 |
+
|
236 |
+
chatbot = gr.Chatbot(
|
237 |
+
[],
|
238 |
+
elem_id="chatbot",
|
239 |
+
avatar_images=('https://aui.atlassian.com/aui/8.8/docs/images/avatar-person.svg',
|
240 |
+
'https://huggingface.co/datasets/huggingface/brand-assets/resolve/main/hf-logo.svg'),
|
241 |
+
bubble_full_width=False,
|
242 |
+
show_copy_button=True,
|
243 |
+
show_share_button=True,
|
244 |
+
)
|
245 |
+
|
246 |
+
with gr.Row():
|
247 |
+
txt = gr.Textbox(
|
248 |
+
scale=3,
|
249 |
+
show_label=False,
|
250 |
+
placeholder="Enter text and press enter",
|
251 |
+
container=False,
|
252 |
+
)
|
253 |
+
txt_btn = gr.Button(value="Submit text", scale=1)
|
254 |
+
|
255 |
+
cross_encoder = gr.Radio(choices=['(FAST) MiniLM-L6v2','(ACCURATE) BGE reranker','(HIGH ACCURATE) ColBERT'], value='(ACCURATE) BGE reranker',label="Embeddings", info="Only First query to Colbert may take litte time)")
|
256 |
+
|
257 |
+
prompt_html = gr.HTML()
|
258 |
+
# Turn off interactivity while generating if you click
|
259 |
+
txt_msg = txt_btn.click(add_text, [chatbot, txt], [chatbot, txt], queue=False).then(
|
260 |
+
bot, [chatbot, cross_encoder], [chatbot, prompt_html])#.then(update_count_html,[],[count_html])
|
261 |
+
|
262 |
+
# Turn it back on
|
263 |
+
txt_msg.then(lambda: gr.Textbox(interactive=True), None, [txt], queue=False)
|
264 |
+
|
265 |
+
# Turn off interactivity while generating if you hit enter
|
266 |
+
txt_msg = txt.submit(add_text, [chatbot, txt], [chatbot, txt], queue=False).then(
|
267 |
+
bot, [chatbot, cross_encoder], [chatbot, prompt_html])#.then(update_count_html,[],[count_html])
|
268 |
+
|
269 |
+
# Turn it back on
|
270 |
+
txt_msg.then(lambda: gr.Textbox(interactive=True), None, [txt], queue=False)
|
271 |
+
|
272 |
+
# Examples
|
273 |
+
gr.Examples(examples, txt)
|
274 |
+
|
275 |
+
|
276 |
+
RAG_db=gr.State()
|
277 |
+
|
278 |
+
with gr.Blocks(title="Quiz Maker", theme=gr.themes.Default(primary_hue="green", secondary_hue="green"), css="style.css") as QUIZBOT:
|
279 |
+
def load_model():
|
280 |
+
RAG= RAGPretrainedModel.from_pretrained("colbert-ir/colbertv2.0")
|
281 |
+
RAG_db.value=RAG.from_index('.ragatouille/colbert/indexes/cbseclass10index')
|
282 |
+
return 'Ready to Go!!'
|
283 |
+
with gr.Column(scale=4):
|
284 |
+
gr.HTML("""
|
285 |
+
<center>
|
286 |
+
<h1><span style="color: purple;">ADWITIYA</span> Customs Manual Quizbot</h1>
|
287 |
+
<h2>Generative AI-powered Capacity building for Training Officers</h2>
|
288 |
+
<i>⚠️ NACIN Faculties create quiz from any topic dynamically for classroom evaluation after their sessions ! ⚠️</i>
|
289 |
+
</center>
|
290 |
+
""")
|
291 |
+
#gr.Warning('Retrieving using ColBERT.. First time query will take a minute for model to load..pls wait')
|
292 |
+
with gr.Column(scale=2):
|
293 |
+
load_btn = gr.Button("Click to Load!🚀")
|
294 |
+
load_text=gr.Textbox()
|
295 |
+
load_btn.click(load_model,[],load_text)
|
296 |
+
|
297 |
+
|
298 |
+
topic = gr.Textbox(label="Enter the Topic for Quiz", placeholder="Write any topic/details from Customs Manual")
|
299 |
+
|
300 |
+
with gr.Row():
|
301 |
+
radio = gr.Radio(
|
302 |
+
["easy", "average", "hard"], label="How difficult should the quiz be?"
|
303 |
+
)
|
304 |
+
|
305 |
+
|
306 |
+
generate_quiz_btn = gr.Button("Generate Quiz!🚀")
|
307 |
+
quiz_msg=gr.Textbox()
|
308 |
+
|
309 |
+
question_radios = [gr.Radio(visible=False), gr.Radio(visible=False), gr.Radio(
|
310 |
+
visible=False), gr.Radio(visible=False), gr.Radio(visible=False), gr.Radio(visible=False), gr.Radio(visible=False), gr.Radio(
|
311 |
+
visible=False), gr.Radio(visible=False), gr.Radio(visible=False)]
|
312 |
+
|
313 |
+
print(question_radios)
|
314 |
+
|
315 |
+
@spaces.GPU
|
316 |
+
@generate_quiz_btn.click(inputs=[radio, topic], outputs=[quiz_msg]+question_radios, api_name="generate_quiz")
|
317 |
+
def generate_quiz(question_difficulty, topic):
|
318 |
+
top_k_rank=10
|
319 |
+
RAG_db_=RAG_db.value
|
320 |
+
documents_full=RAG_db_.search(topic,k=top_k_rank)
|
321 |
+
|
322 |
+
|
323 |
+
|
324 |
+
generate_kwargs = dict(
|
325 |
+
temperature=0.2,
|
326 |
+
max_new_tokens=4000,
|
327 |
+
top_p=0.95,
|
328 |
+
repetition_penalty=1.0,
|
329 |
+
do_sample=True,
|
330 |
+
seed=42,
|
331 |
+
)
|
332 |
+
question_radio_list = []
|
333 |
+
count=0
|
334 |
+
while count<=3:
|
335 |
+
try:
|
336 |
+
documents=[item['content'] for item in documents_full]
|
337 |
+
document_summaries = [f"[DOCUMENT {i+1}]: {summary}{count}" for i, summary in enumerate(documents)]
|
338 |
+
documents_str='\n'.join(document_summaries)
|
339 |
+
formatted_prompt = system_instructions(
|
340 |
+
question_difficulty, topic,documents_str)
|
341 |
+
print(formatted_prompt)
|
342 |
+
pre_prompt = [
|
343 |
+
{"role": "system", "content": formatted_prompt}
|
344 |
+
]
|
345 |
+
response = client.text_generation(
|
346 |
+
formatted_prompt, **generate_kwargs, stream=False, details=False, return_full_text=False,
|
347 |
+
)
|
348 |
+
output_json = json.loads(f"{response}")
|
349 |
+
|
350 |
+
|
351 |
+
print(response)
|
352 |
+
print('output json', output_json)
|
353 |
+
|
354 |
+
global quiz_data
|
355 |
+
|
356 |
+
quiz_data = output_json
|
357 |
+
|
358 |
+
|
359 |
+
|
360 |
+
for question_num in range(1, 11):
|
361 |
+
question_key = f"Q{question_num}"
|
362 |
+
answer_key = f"A{question_num}"
|
363 |
+
|
364 |
+
question = quiz_data.get(question_key)
|
365 |
+
answer = quiz_data.get(quiz_data.get(answer_key))
|
366 |
+
|
367 |
+
if not question or not answer:
|
368 |
+
continue
|
369 |
+
|
370 |
+
choice_keys = [f"{question_key}:C{i}" for i in range(1, 5)]
|
371 |
+
choice_list = []
|
372 |
+
for choice_key in choice_keys:
|
373 |
+
choice = quiz_data.get(choice_key, "Choice not found")
|
374 |
+
choice_list.append(f"{choice}")
|
375 |
+
|
376 |
+
radio = gr.Radio(choices=choice_list, label=question,
|
377 |
+
visible=True, interactive=True)
|
378 |
+
|
379 |
+
question_radio_list.append(radio)
|
380 |
+
if len(question_radio_list)==10:
|
381 |
+
break
|
382 |
+
else:
|
383 |
+
print('10 questions not generated . So trying again!')
|
384 |
+
count+=1
|
385 |
+
continue
|
386 |
+
except Exception as e:
|
387 |
+
count+=1
|
388 |
+
print(f"Exception occurred: {e}")
|
389 |
+
if count==3:
|
390 |
+
print('Retry exhausted')
|
391 |
+
gr.Warning('Sorry. Pls try with another topic !')
|
392 |
+
else:
|
393 |
+
print(f"Trying again..{count} time...please wait")
|
394 |
+
continue
|
395 |
+
|
396 |
+
print('Question radio list ' , question_radio_list)
|
397 |
+
|
398 |
+
return ['Quiz Generated!']+ question_radio_list
|
399 |
+
|
400 |
+
check_button = gr.Button("Check Score")
|
401 |
+
|
402 |
+
score_textbox = gr.Markdown()
|
403 |
+
|
404 |
+
@check_button.click(inputs=question_radios, outputs=score_textbox)
|
405 |
+
def compare_answers(*user_answers):
|
406 |
+
user_anwser_list = []
|
407 |
+
user_anwser_list = user_answers
|
408 |
+
|
409 |
+
answers_list = []
|
410 |
+
|
411 |
+
for question_num in range(1, 20):
|
412 |
+
answer_key = f"A{question_num}"
|
413 |
+
answer = quiz_data.get(quiz_data.get(answer_key))
|
414 |
+
if not answer:
|
415 |
+
break
|
416 |
+
answers_list.append(answer)
|
417 |
+
|
418 |
+
score = 0
|
419 |
+
|
420 |
+
for item in user_anwser_list:
|
421 |
+
if item in answers_list:
|
422 |
+
score += 1
|
423 |
+
if score>5:
|
424 |
+
message = f"### Good ! You got {score} over 10!"
|
425 |
+
elif score>7:
|
426 |
+
message = f"### Excellent ! You got {score} over 10!"
|
427 |
+
else:
|
428 |
+
message = f"### You got {score} over 10! Dont worry . You can prepare well and try better next time !"
|
429 |
+
|
430 |
+
return message
|
431 |
+
|
432 |
+
|
433 |
+
|
434 |
+
demo = gr.TabbedInterface([CHATBOT,QUIZBOT], ["AI ChatBot", "AI Quizbot"])
|
435 |
+
|
436 |
+
demo.queue()
|
437 |
+
demo.launch(debug=True)
|