File size: 21,287 Bytes
226c234 |
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 |
import streamlit as st
from io import BytesIO
import ibm_watsonx_ai
import secretsload
import genparam
import requests
import time
import re
import json
from ibm_watsonx_ai.foundation_models import ModelInference
from ibm_watsonx_ai import Credentials, APIClient
from ibm_watsonx_ai.metanames import GenTextParamsMetaNames as GenParams
from ibm_watsonx_ai.metanames import GenTextReturnOptMetaNames as RetParams
from ibm_watsonx_ai.foundation_models import Embeddings
from ibm_watsonx_ai.foundation_models.utils.enums import EmbeddingTypes
from pymilvus import MilvusClient
from secretsload import load_stsecrets
credentials = load_stsecrets()
st.set_page_config(
page_title="The Solutioning Sages",
page_icon="🪄",
initial_sidebar_state="collapsed",
layout="wide"
)
# Password protection
def check_password():
def password_entered():
if st.session_state["password"] == st.secrets["app_password"]:
st.session_state["password_correct"] = True
del st.session_state["password"]
else:
st.session_state["password_correct"] = False
if "password_correct" not in st.session_state:
st.markdown("\n\n")
st.text_input("Enter the password", type="password", on_change=password_entered, key="password")
st.divider()
st.info("Designed and developed by Milan Mrdenovic © IBM Norway 2024")
return False
elif not st.session_state["password_correct"]:
st.markdown("\n\n")
st.text_input("Enter the password", type="password", on_change=password_entered, key="password")
st.divider()
st.error("😕 Incorrect password")
st.info("Designed and developed by Milan Mrdenovic © IBM Norway 2024")
return False
else:
return True
def initialize_session_state():
if 'chat_history_1' not in st.session_state:
st.session_state.chat_history_1 = []
if 'chat_history_2' not in st.session_state:
st.session_state.chat_history_2 = []
if 'chat_history_3' not in st.session_state:
st.session_state.chat_history_3 = []
if 'first_question' not in st.session_state:
st.session_state.first_question = False
if "counter" not in st.session_state:
st.session_state["counter"] = 0
if 'token_statistics' not in st.session_state:
st.session_state.token_statistics = []
# three_column_style = """
# <style>
# .stColumn {
# padding: 0.5rem;
# border-right: 1px solid #dedede;
# }
# .stColumn:last-child {
# border-right: none;
# }
# .chat-container {
# height: calc(100vh - 200px);
# overflow-y: auto;
# }
# </style>
# """
three_column_style = """
<style>
.stColumn {
padding: 0.5rem;
border-right: 1px solid #dedede;
}
.stColumn:last-child {
border-right: none;
}
.chat-container {
height: calc(100vh - 200px);
overflow-y: auto;
display: flex;
flex-direction: column;
}
.chat-messages {
display: flex;
flex-direction: column;
gap: 1rem;
}
</style>
""" # Alt Style
#-----
def get_active_model():
return genparam.SELECTED_MODEL_1 if genparam.ACTIVE_MODEL == 0 else genparam.SELECTED_MODEL_2
def get_active_prompt_template():
return genparam.PROMPT_TEMPLATE_1 if genparam.ACTIVE_MODEL == 0 else genparam.PROMPT_TEMPLATE_2
def get_active_vector_index():
return st.secrets["vector_index_id_1"] if genparam.ACTIVE_INDEX == 0 else st.secrets["vector_index_id_2"]
#-----
def setup_client(project_id):
credentials = Credentials(
url=st.secrets["url"],
api_key=st.secrets["api_key"]
)
apo = st.secrets["api_key"]
client = APIClient(credentials, project_id=project_id)
return credentials, client
wml_credentials, client = setup_client(st.secrets["project_id"])
def setup_vector_index(client, wml_credentials, vector_index_id):
vector_index_details = client.data_assets.get_details(vector_index_id)
vector_index_properties = vector_index_details["entity"]["vector_index"]
emb = Embeddings(
model_id=vector_index_properties["settings"]["embedding_model_id"],
#model_id="sentence-transformers/all-minilm-l12-v2",
credentials=wml_credentials,
project_id=st.secrets["project_id"],
params={
"truncate_input_tokens": 512
}
)
vector_store_schema = vector_index_properties["settings"]["schema_fields"]
connection_details = client.connections.get_details(vector_index_details["entity"]["vector_index"]["store"]["connection_id"])
connection_properties = connection_details["entity"]["properties"]
milvus_client = MilvusClient(
uri=f'https://{connection_properties.get("host")}:{connection_properties.get("port")}',
user=connection_properties.get("username"),
password=connection_properties.get("password"),
db_name=vector_index_properties["store"]["database"]
)
return milvus_client, emb, vector_index_properties, vector_store_schema
def proximity_search(question, milvus_client, emb, vector_index_properties, vector_store_schema):
query_vectors = emb.embed_query(question)
milvus_response = milvus_client.search(
collection_name=vector_index_properties["store"]["index"],
data=[query_vectors],
limit=vector_index_properties["settings"]["top_k"],
metric_type="L2",
output_fields=[
vector_store_schema.get("text"),
vector_store_schema.get("document_name"),
vector_store_schema.get("page_number")
]
)
documents = []
for hit in milvus_response[0]:
text = hit["entity"].get(vector_store_schema.get("text"), "")
doc_name = hit["entity"].get(vector_store_schema.get("document_name"), "Unknown Document")
page_num = hit["entity"].get(vector_store_schema.get("page_number"), "N/A")
formatted_result = f"Document: {doc_name}\nContent: {text}\nPage: {page_num}\n"
documents.append(formatted_result)
joined = "\n".join(documents)
retrieved = f"""Number of Retrieved Documents: {len(documents)}\n\n{joined}"""
return retrieved
def prepare_prompt(prompt, chat_history):
if genparam.TYPE == "chat" and chat_history:
chats = "\n".join([f"{message['role']}: \"{message['content']}\"" for message in chat_history])
prompt = f"""Retrieved Contextual Information:\n__grounding__\n\nConversation History:\n{chats}\n\nNew User Input: {prompt}"""
return prompt
else:
prompt = f"""Retrieved Contextual Information:\n__grounding__\n\nUser Input: {prompt}"""
return prompt
def apply_prompt_syntax(prompt, system_prompt, prompt_template, bake_in_prompt_syntax):
model_family_syntax = {
"llama3-instruct (llama-3, 3.1 & 3.2) - system": """<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\n{system_prompt}<|eot_id|><|start_header_id|>user<|end_header_id|>\n\n{prompt}<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n""",
"llama3-instruct (llama-3, 3.1 & 3.2) - user": """<|begin_of_text|><|start_header_id|>user<|end_header_id|>\n\n{prompt}<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n""",
"granite-13b-chat & instruct - system": """<|system|>\n{system_prompt}\n<|user|>\n{prompt}\n<|assistant|>\n\n""",
"granite-13b-chat & instruct - user": """<|user|>\n{prompt}\n<|assistant|>\n\n""",
"mistral & mixtral v2 tokenizer - system": """<s>[INST] System Prompt: {system_prompt} [/INST][INST] {prompt} [/INST]\n\n""",
"mistral & mixtral v2 tokenizer - user": """<s>[INST] {prompt} [/INST]\n\n""",
"no syntax - system": """{system_prompt}\n\n{prompt}""",
"no syntax - user": """{prompt}"""
}
if bake_in_prompt_syntax:
template = model_family_syntax[prompt_template]
if system_prompt:
return template.format(system_prompt=system_prompt, prompt=prompt)
return prompt
def generate_response(watsonx_llm, prompt_data, params):
generated_response = watsonx_llm.generate_text_stream(prompt=prompt_data, params=params)
for chunk in generated_response:
yield chunk
def fetch_response(user_input, milvus_client, emb, vector_index_properties, vector_store_schema, system_prompt, chat_history):
# Get grounding documents
grounding = proximity_search(
question=user_input,
milvus_client=milvus_client,
emb=emb,
vector_index_properties=vector_index_properties,
vector_store_schema=vector_store_schema
)
# Special handling for PATH-er B. (first column)
if chat_history == st.session_state.chat_history_1:
# Display user question first
with st.chat_message("user", avatar=genparam.USER_AVATAR):
st.markdown(user_input)
# Parse and display each document from the grounding
documents = grounding.split("\n\n")[2:] # Skip the count line and first newline
for doc in documents:
if doc.strip(): # Only process non-empty strings
parts = doc.split("\n")
doc_name = parts[0].replace("Document: ", "")
content = parts[1].replace("Content: ", "")
# Display document with delay
time.sleep(0.5)
st.markdown(f"**{doc_name}**")
st.code(content)
# Store in chat history
return grounding
# For MOD-ther S. (second column)
elif chat_history == st.session_state.chat_history_2:
prompt = prepare_prompt(user_input, chat_history)
prompt_data = apply_prompt_syntax(
prompt,
system_prompt,
get_active_prompt_template(),
genparam.BAKE_IN_PROMPT_SYNTAX
)
prompt_data = prompt_data.replace("__grounding__", grounding)
# Add debug information to column 1 if enabled
if genparam.INPUT_DEBUG_VIEW == 1:
with st.columns(3)[0]: # Access first column
st.markdown(f"**{genparam.BOT_2_AVATAR} {genparam.BOT_2_NAME} Prompt Data:**")
st.code(prompt_data, language="text")
# For SYS-ter V. (third column)
else:
# Get chat history from MOD-ther S.
mod_ther_history = st.session_state.chat_history_2
prompt = prepare_prompt(user_input, mod_ther_history)
prompt_data = apply_prompt_syntax(
prompt,
system_prompt,
get_active_prompt_template(),
genparam.BAKE_IN_PROMPT_SYNTAX
)
prompt_data = prompt_data.replace("__grounding__", grounding)
# Add debug information to column 1 if enabled
if genparam.INPUT_DEBUG_VIEW == 1:
with st.columns(3)[0]: # Access first column
st.markdown(f"**{genparam.BOT_3_AVATAR} {genparam.BOT_3_NAME} Prompt Data:**")
st.code(prompt_data, language="text")
# Continue with normal processing for columns 2 and 3
watsonx_llm = ModelInference(
api_client=client,
model_id=get_active_model(),
verify=genparam.VERIFY
)
params = {
GenParams.DECODING_METHOD: genparam.DECODING_METHOD,
GenParams.MAX_NEW_TOKENS: genparam.MAX_NEW_TOKENS,
GenParams.MIN_NEW_TOKENS: genparam.MIN_NEW_TOKENS,
GenParams.REPETITION_PENALTY: genparam.REPETITION_PENALTY,
GenParams.STOP_SEQUENCES: genparam.STOP_SEQUENCES
}
bot_name = None
bot_avatar = None
if chat_history == st.session_state.chat_history_1:
bot_name = genparam.BOT_1_NAME
bot_avatar = genparam.BOT_1_AVATAR
elif chat_history == st.session_state.chat_history_2:
bot_name = genparam.BOT_2_NAME
bot_avatar = genparam.BOT_2_AVATAR
else:
bot_name = genparam.BOT_3_NAME
bot_avatar = genparam.BOT_3_AVATAR
with st.chat_message(bot_name, avatar=bot_avatar):
if chat_history != st.session_state.chat_history_1: # Only generate responses for columns 2 and 3
stream = generate_response(watsonx_llm, prompt_data, params)
response = st.write_stream(stream)
# Only capture tokens for MOD-ther S. and SYS-ter V.
if genparam.TOKEN_CAPTURE_ENABLED and chat_history != st.session_state.chat_history_1:
token_stats = capture_tokens(prompt_data, response, bot_name)
if token_stats:
st.session_state.token_statistics.append(token_stats)
else:
response = grounding # For column 1, we already displayed the content
return response
def capture_tokens(prompt_data, response, chat_number):
if not genparam.TOKEN_CAPTURE_ENABLED:
return
watsonx_llm = ModelInference(
api_client=client,
model_id=genparam.SELECTED_MODEL,
verify=genparam.VERIFY
)
input_tokens = watsonx_llm.tokenize(prompt=prompt_data)["result"]["token_count"]
output_tokens = watsonx_llm.tokenize(prompt=response)["result"]["token_count"]
total_tokens = input_tokens + output_tokens
return {
"bot_name": bot_name,
"input_tokens": input_tokens,
"output_tokens": output_tokens,
"total_tokens": total_tokens,
"timestamp": time.strftime("%H:%M:%S")
}
def main():
initialize_session_state()
# Apply custom styles
st.markdown(three_column_style, unsafe_allow_html=True)
# Sidebar configuration
st.sidebar.header('The Solutioning Sages')
st.sidebar.divider()
# Display token statistics in sidebar
st.sidebar.subheader("Token Usage Statistics")
# Group token statistics by interaction (for MOD-ther S. and SYS-ter V. only)
if st.session_state.token_statistics:
current_timestamp = None
interaction_count = 0
stats_by_time = {}
# Group stats by timestamp
for stat in st.session_state.token_statistics:
if stat["timestamp"] not in stats_by_time:
stats_by_time[stat["timestamp"]] = []
stats_by_time[stat["timestamp"]].append(stat)
# Display grouped stats
for timestamp, stats in stats_by_time.items():
interaction_count += 1
st.sidebar.markdown(f"**Interaction {interaction_count}** ({timestamp})")
# Calculate total tokens for this interaction
total_input = sum(stat['input_tokens'] for stat in stats)
total_output = sum(stat['output_tokens'] for stat in stats)
total = total_input + total_output
# Display individual bot statistics
for stat in stats:
st.sidebar.markdown(
f"_{stat['bot_name']}_ \n"
f"Input: {stat['input_tokens']} tokens \n"
f"Output: {stat['output_tokens']} tokens \n"
f"Total: {stat['total_tokens']} tokens"
)
# Display interaction totals
st.sidebar.markdown("**Interaction Totals:**")
st.sidebar.markdown(
f"Total Input: {total_input} tokens \n"
f"Total Output: {total_output} tokens \n"
f"Total Usage: {total} tokens"
)
st.sidebar.markdown("---")
st.sidebar.markdown("")
if not check_password():
st.stop()
# Get user input before column creation
user_input = st.chat_input("Ask your question here", key="user_input")
if user_input:
# Create three columns
col1, col2, col3 = st.columns(3)
# First column - PATH-er B. (Document Display)
with col1:
st.markdown("<div class='chat-container'>", unsafe_allow_html=True)
st.subheader(f"{genparam.BOT_1_AVATAR} {genparam.BOT_1_NAME}")
st.markdown("<div class='chat-messages'>", unsafe_allow_html=True)
# Display previous messages
for message in st.session_state.chat_history_1:
if message["role"] == "user":
with st.chat_message(message["role"], avatar=genparam.USER_AVATAR):
st.markdown(message['content'])
else:
# Parse and display stored documents
documents = message['content'].split("\n\n")[2:] # Skip count line
for doc in documents:
if doc.strip():
parts = doc.split("\n")
doc_name = parts[0].replace("Document: ", "")
content = parts[1].replace("Content: ", "")
st.markdown(f"**{doc_name}**")
st.code(content)
# Add user message and get new response
st.session_state.chat_history_1.append({"role": "user", "content": user_input, "avatar": genparam.USER_AVATAR})
milvus_client, emb, vector_index_properties, vector_store_schema = setup_vector_index(
client,
wml_credentials,
st.secrets["vector_index_id_1"] # Use first vector index
)
system_prompt = genparam.BOT_1_PROMPT
response = fetch_response(
user_input,
milvus_client,
emb,
vector_index_properties,
vector_store_schema,
system_prompt,
st.session_state.chat_history_1
)
st.session_state.chat_history_1.append({"role": genparam.BOT_1_NAME, "content": response, "avatar": genparam.BOT_1_AVATAR})
st.markdown("</div></div>", unsafe_allow_html=True)
# Second column - MOD-ther S. (Uses documents from first vector index)
with col2:
st.markdown("<div class='chat-container'>", unsafe_allow_html=True)
st.subheader(f"{genparam.BOT_2_AVATAR} {genparam.BOT_2_NAME}")
st.markdown("<div class='chat-messages'>", unsafe_allow_html=True)
for message in st.session_state.chat_history_2:
if message["role"] != "user":
with st.chat_message(message["role"], avatar=genparam.BOT_2_AVATAR):
st.markdown(message['content'])
st.session_state.chat_history_2.append({"role": "user", "content": user_input, "avatar": genparam.USER_AVATAR})
milvus_client, emb, vector_index_properties, vector_store_schema = setup_vector_index(
client,
wml_credentials,
st.secrets["vector_index_id_1"] # Use first vector index
)
system_prompt = genparam.BOT_2_PROMPT
response = fetch_response(
user_input,
milvus_client,
emb,
vector_index_properties,
vector_store_schema,
system_prompt,
st.session_state.chat_history_2
)
st.session_state.chat_history_2.append({"role": genparam.BOT_2_NAME, "content": response, "avatar": genparam.BOT_2_AVATAR})
st.markdown("</div></div>", unsafe_allow_html=True)
# Third column - SYS-ter V. (Uses second vector index and chat history from second column)
with col3:
st.markdown("<div class='chat-container'>", unsafe_allow_html=True)
st.subheader(f"{genparam.BOT_3_AVATAR} {genparam.BOT_3_NAME}")
st.markdown("<div class='chat-messages'>", unsafe_allow_html=True)
for message in st.session_state.chat_history_3:
if message["role"] != "user":
with st.chat_message(message["role"], avatar=genparam.BOT_3_AVATAR):
st.markdown(message['content'])
st.session_state.chat_history_3.append({"role": "user", "content": user_input, "avatar": genparam.USER_AVATAR})
milvus_client, emb, vector_index_properties, vector_store_schema = setup_vector_index(
client,
wml_credentials,
st.secrets["vector_index_id_2"] # Use second vector index
)
system_prompt = genparam.BOT_3_PROMPT
response = fetch_response(
user_input,
milvus_client,
emb,
vector_index_properties,
vector_store_schema,
system_prompt,
st.session_state.chat_history_3
)
st.session_state.chat_history_3.append({"role": genparam.BOT_3_NAME, "content": response, "avatar": genparam.BOT_3_AVATAR})
st.markdown("</div></div>", unsafe_allow_html=True)
# Update sidebar with new question
st.sidebar.markdown("---")
st.sidebar.markdown("**Latest Question:**")
st.sidebar.markdown(f"_{user_input}_")
if __name__ == "__main__":
main() |