File size: 23,264 Bytes
453b8b8 |
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 |
from lollms.config import TypedConfig, BaseConfig, ConfigTemplate, InstallOption
from lollms.types import MSG_TYPE
from lollms.personality import APScript, AIPersonality
from lollms.helpers import ASCIIColors
import numpy as np
import json
from pathlib import Path
import numpy as np
import json
class TextVectorizer:
def __init__(self, model_name, database_file:Path|str, visualize_data_at_startup=False, visualize_data_at_add_file=False, visualize_data_at_generate=False):
from transformers import AutoTokenizer, AutoModel
self.tokenizer = AutoTokenizer.from_pretrained(model_name)
self.model = AutoModel.from_pretrained(model_name)
self.embeddings = {}
self.texts = {}
self.ready = False
self.database_file = Path(database_file)
self.visualize_data_at_startup = visualize_data_at_startup
self.visualize_data_at_add_file = visualize_data_at_add_file
self.visualize_data_at_generate = visualize_data_at_generate
# Load previous state from the JSON file
if Path(self.database_file).exists():
ASCIIColors.success(f"Database file found : {self.database_file}")
self.load_from_json()
if visualize_data_at_startup:
self.show_document()
self.ready = True
else:
ASCIIColors.info(f"No database file found : {self.database_file}")
def show_document(self, query_text="What is the main idea of this text?", use_pca=True):
import textwrap
import seaborn as sns
import matplotlib.pyplot as plt
import mplcursors
from tkinter import Tk, Text, Scrollbar, Frame, Label, TOP, BOTH, RIGHT, LEFT, Y, N, END
from sklearn.manifold import TSNE
from sklearn.decomposition import PCA
import torch
if use_pca:
print("Showing pca representation :")
else:
print("Showing t-sne representation :")
texts = list(self.texts.values())
embeddings = torch.stack(list(self.embeddings.values())).detach().squeeze(1).numpy()
# Normalize embeddings
norms = np.linalg.norm(embeddings, axis=1)
normalized_embeddings = embeddings / norms[:, np.newaxis]
# Embed the query text
query_embedding = self.embed_query(query_text)
query_embedding = query_embedding.detach().squeeze().numpy()
query_normalized_embedding = query_embedding / np.linalg.norm(query_embedding)
# Combine the query embedding with the document embeddings
combined_embeddings = np.vstack((normalized_embeddings, query_normalized_embedding))
if use_pca:
# Use PCA for dimensionality reduction
pca = PCA(n_components=2)
embeddings_2d = pca.fit_transform(combined_embeddings)
else:
# Use t-SNE for dimensionality reduction
# Adjust the perplexity value
perplexity = min(30, combined_embeddings.shape[0] - 1)
tsne = TSNE(n_components=2, perplexity=perplexity)
embeddings_2d = tsne.fit_transform(combined_embeddings)
# Create a scatter plot using Seaborn
sns.scatterplot(x=embeddings_2d[:-1, 0], y=embeddings_2d[:-1, 1]) # Plot document embeddings
plt.scatter(embeddings_2d[-1, 0], embeddings_2d[-1, 1], color='red') # Plot query embedding
# Add labels to the scatter plot
for i, (x, y) in enumerate(embeddings_2d[:-1]):
plt.text(x, y, str(i), fontsize=8)
plt.xlabel('Dimension 1')
plt.ylabel('Dimension 2')
if use_pca:
plt.title('Embeddings Scatter Plot based on PCA')
else:
plt.title('Embeddings Scatter Plot based on t-SNE')
# Enable mplcursors to show tooltips on hover
cursor = mplcursors.cursor(hover=True)
# Define the hover event handler
@cursor.connect("add")
def on_hover(sel):
index = sel.target.index
if index > 0:
text = texts[index]
wrapped_text = textwrap.fill(text, width=50) # Wrap the text into multiple lines
sel.annotation.set_text(f"Index: {index}\nText:\n{wrapped_text}")
else:
sel.annotation.set_text("Query")
# Define the click event handler using matplotlib event handling mechanism
def on_click(event):
if event.xdata is not None and event.ydata is not None:
x, y = event.xdata, event.ydata
distances = ((embeddings_2d[:, 0] - x) ** 2 + (embeddings_2d[:, 1] - y) ** 2)
index = distances.argmin()
text = texts[index] if index < len(texts) else query_text
# Open a new Tkinter window with the content of the text
root = Tk()
root.title(f"Text for Index {index}")
frame = Frame(root)
frame.pack(fill=BOTH, expand=True)
label = Label(frame, text="Text:")
label.pack(side=TOP, padx=5, pady=5)
text_box = Text(frame)
text_box.pack(side=TOP, padx=5, pady=5, fill=BOTH, expand=True)
text_box.insert(END, text)
scrollbar = Scrollbar(frame)
scrollbar.pack(side=RIGHT, fill=Y)
scrollbar.config(command=text_box.yview)
text_box.config(yscrollcommand=scrollbar.set)
text_box.config(state="disabled")
root.mainloop()
# Connect the click event handler to the figure
plt.gcf().canvas.mpl_connect("button_press_event", on_click)
plt.show()
def index_document(self, document_id, text, chunk_size, overlap_size, force_vectorize=False):
import torch
if document_id in self.embeddings and not force_vectorize:
print(f"Document {document_id} already exists. Skipping vectorization.")
return
# Tokenize text
tokens = self.tokenizer.encode_plus(text, add_special_tokens=False, return_attention_mask=False)['input_ids']
# Split tokens into sentences
sentences = self.tokenizer.decode(tokens).split('. ')
# Generate chunks with overlap and sentence boundaries
chunks = []
current_chunk = []
for sentence in sentences:
sentence_tokens = self.tokenizer.encode_plus(sentence, add_special_tokens=False, return_attention_mask=False)['input_ids']
if len(current_chunk) + len(sentence_tokens) <= chunk_size:
current_chunk.extend(sentence_tokens)
else:
if current_chunk:
chunks.append(current_chunk)
current_chunk = sentence_tokens
if current_chunk:
chunks.append(current_chunk)
# Generate overlapping chunks
overlapping_chunks = []
for i in range(len(chunks)):
chunk_start = i * (chunk_size - overlap_size)
chunk_end = min(chunk_start + chunk_size, len(tokens))
chunk = tokens[chunk_start:chunk_end]
overlapping_chunks.append(chunk)
# Generate embeddings for each chunk
for i, chunk in enumerate(overlapping_chunks):
# Pad the chunk if it is smaller than chunk_size
if len(chunk) < chunk_size:
padding = [self.tokenizer.pad_token_id] * (chunk_size - len(chunk))
chunk.extend(padding)
# Convert tokens to IDs
input_ids = chunk[:chunk_size]
# Convert input to PyTorch tensor
input_tensor = torch.tensor([input_ids])
# Generate chunk embedding
with torch.no_grad():
self.model.eval()
outputs = self.model(input_tensor)
embeddings = outputs.last_hidden_state.mean(dim=1)
# Store chunk ID, embedding, and original text
chunk_id = f"{document_id}_chunk_{i + 1}"
self.embeddings[chunk_id] = embeddings
self.texts[chunk_id] = self.tokenizer.decode(chunk[:chunk_size], skip_special_tokens=True)
self.save_to_json()
self.ready = True
if self.visualize_data_at_add_file:
self.show_document()
def embed_query(self, query_text):
import torch
# Tokenize query text
query_tokens = self.tokenizer.encode(query_text)
# Convert input to PyTorch tensor
query_input_tensor = torch.tensor([query_tokens])
# Generate query embedding
with torch.no_grad():
self.model.eval()
query_outputs = self.model(query_input_tensor)
query_embedding = query_outputs.last_hidden_state.mean(dim=1)
return query_embedding
def recover_text(self, query_embedding, top_k=1):
from sklearn.metrics.pairwise import cosine_similarity
similarities = {}
for chunk_id, chunk_embedding in self.embeddings.items():
similarity = cosine_similarity(query_embedding.numpy(), chunk_embedding.numpy())[0][0]
similarities[chunk_id] = similarity
# Sort the similarities and retrieve the top-k most similar embeddings
sorted_similarities = sorted(similarities.items(), key=lambda x: x[1], reverse=True)[:top_k]
# Retrieve the original text associated with the most similar embeddings
texts = [self.texts[chunk_id] for chunk_id, _ in sorted_similarities]
if self.visualize_data_at_generate:
self.show_document()
return texts
def save_to_json(self):
state = {
"embeddings": {str(k): v.tolist() for k, v in self.embeddings.items()},
"texts": self.texts,
}
with open(self.database_file, "w") as f:
json.dump(state, f)
def load_from_json(self):
import torch
ASCIIColors.info("Loading vectorized documents")
with open(self.database_file, "r") as f:
state = json.load(f)
self.embeddings = {k: torch.tensor(v) for k, v in state["embeddings"].items()}
self.texts = state["texts"]
self.ready = True
class Processor(APScript):
"""
A class that processes model inputs and outputs.
Inherits from APScript.
"""
def __init__(
self,
personality: AIPersonality
) -> None:
self.word_callback = None
personality_config_template = ConfigTemplate(
[
{"name":"database_path","type":"str","value":f"{personality.name}_db.json", "help":"Path to the database"},
{"name":"max_chunk_size","type":"int","value":512, "min":10, "max":personality.config["ctx_size"],"help":"Maximum size of text chunks to vectorize"},
{"name":"chunk_overlap","type":"int","value":20, "min":0, "max":personality.config["ctx_size"],"help":"Overlap between chunks"},
{"name":"max_answer_size","type":"int","value":512, "min":10, "max":personality.config["ctx_size"],"help":"Maximum number of tokens to allow the generator to generate as an answer to your question"},
{"name":"visualize_data_at_startup","type":"bool","value":False, "help":"If true, the database will be visualized at startup"},
{"name":"visualize_data_at_add_file","type":"bool","value":False, "help":"If true, the database will be visualized when a new file is added"},
{"name":"visualize_data_at_generate","type":"bool","value":False, "help":"If true, the database will be visualized at generation time"},
]
)
personality_config_vals = BaseConfig.from_template(personality_config_template)
personality_config = TypedConfig(
personality_config_template,
personality_config_vals
)
super().__init__(
personality,
personality_config
)
self.state = 0
self.ready = False
self.personality = personality
self.callback = None
self.vector_store = TextVectorizer(
"bert-base-uncased",
self.personality.lollms_paths.personal_data_path/self.personality_config["database_path"],
visualize_data_at_startup=self.personality_config["visualize_data_at_startup"],
visualize_data_at_add_file=self.personality_config["visualize_data_at_add_file"],
visualize_data_at_generate=self.personality_config["visualize_data_at_generate"]
)
if len(self.vector_store.embeddings)>0:
self.ready = True
@staticmethod
def read_pdf_file(file_path):
import PyPDF2
with open(file_path, 'rb') as file:
pdf_reader = PyPDF2.PdfReader(file)
text = ""
for page in pdf_reader.pages:
text += page.extract_text()
return text
@staticmethod
def read_docx_file(file_path):
from docx import Document
doc = Document(file_path)
text = ""
for paragraph in doc.paragraphs:
text += paragraph.text + "\n"
return text
@staticmethod
def read_json_file(file_path):
with open(file_path, 'r') as file:
data = json.load(file)
return data
@staticmethod
def read_csv_file(file_path):
import csv
with open(file_path, 'r') as file:
csv_reader = csv.reader(file)
lines = [row for row in csv_reader]
return lines
@staticmethod
def read_html_file(file_path):
from bs4 import BeautifulSoup
with open(file_path, 'r') as file:
soup = BeautifulSoup(file, 'html.parser')
text = soup.get_text()
return text
@staticmethod
def read_pptx_file(file_path):
from pptx import Presentation
prs = Presentation(file_path)
text = ""
for slide in prs.slides:
for shape in slide.shapes:
if shape.has_text_frame:
for paragraph in shape.text_frame.paragraphs:
for run in paragraph.runs:
text += run.text
return text
@staticmethod
def read_text_file(file_path):
with open(file_path, 'r', encoding='utf-8') as file:
content = file.read()
return content
def build_db(self):
ASCIIColors.info("-> Vectorizing the database"+ASCIIColors.color_orange)
if self.callback is not None:
self.callback("Vectorizing the database", MSG_TYPE.MSG_TYPE_CHUNK)
for file in self.files:
try:
if Path(file).suffix==".pdf":
text = Processor.read_pdf_file(file)
elif Path(file).suffix==".docx":
text = Processor.read_docx_file(file)
elif Path(file).suffix==".docx":
text = Processor.read_pptx_file(file)
elif Path(file).suffix==".json":
text = Processor.read_json_file(file)
elif Path(file).suffix==".csv":
text = Processor.read_csv_file(file)
elif Path(file).suffix==".html":
text = Processor.read_html_file(file)
else:
text = Processor.read_text_file(file)
try:
chunk_size=int(self.personality_config["chunk_size"])
except:
ASCIIColors.warning(f"Couldn't read chunk size. Verify your configuration file")
chunk_size=512
try:
overlap_size=int(self.personality_config["chunk_overlap"])
except:
ASCIIColors.warning(f"Couldn't read chunk size. Verify your configuration file")
overlap_size=50
self.vector_store.index_document(file, text, chunk_size=chunk_size, overlap_size=overlap_size)
print(ASCIIColors.color_reset)
ASCIIColors.success(f"File {file} vectorized successfully")
self.ready = True
except Exception as ex:
ASCIIColors.error(f"Couldn't vectorize {file}: The vectorizer threw this exception:{ex}")
def add_file(self, path):
super().add_file(path)
try:
self.build_db()
self.ready = True
return True
except Exception as ex:
ASCIIColors.error(f"Couldn't vectorize the database: The vectgorizer threw this exception: {ex}")
return False
def run_workflow(self, prompt, previous_discussion_text="", callback=None):
"""
Runs the workflow for processing the model input and output.
This method should be called to execute the processing workflow.
Args:
generate_fn (function): A function that generates model output based on the input prompt.
The function should take a single argument (prompt) and return the generated text.
prompt (str): The input prompt for the model.
previous_discussion_text (str, optional): The text of the previous discussion. Default is an empty string.
Returns:
None
"""
# State machine
output =""
self.callback = callback
if prompt.strip().lower()=="send_file":
self.state = 1
print("Please provide the file name")
if callback is not None:
callback("Please provide the file path", MSG_TYPE.MSG_TYPE_FULL)
output = "Please provide the file name"
elif prompt.strip().lower()=="help":
if callback:
callback(self.personality.help,MSG_TYPE.MSG_TYPE_FULL)
ASCIIColors.info(help)
self.state = 0
elif prompt.strip().lower()=="show_database":
try:
self.vector_store.show_document()
except Exception as ex:
if callback is not None:
callback(f"Couldn't show the database\nMake sure you have already uploaded a database.\nReceived exception is: {ex}", MSG_TYPE.MSG_TYPE_FULL)
self.state = 0
elif prompt.strip().lower()=="set_database":
print("Please provide the database file name")
if callback is not None:
callback("Please provide the database file path", MSG_TYPE.MSG_TYPE_FULL)
output = "Please provide the database file name"
self.state = 2
elif prompt.strip().lower()=="clear_database":
database_fill_path:Path = self.personality.lollms_paths.personal_data_path/self.personality_config["database_path"]
if database_fill_path.exists():
database_fill_path.unlink()
self.vector_store = TextVectorizer(
"bert-base-uncased",
self.personality.lollms_paths.personal_data_path/self.personality_config["database_path"],
visualize_data_at_startup=self.personality_config["visualize_data_at_startup"],
visualize_data_at_add_file=self.personality_config["visualize_data_at_add_file"],
visualize_data_at_generate=self.personality_config["visualize_data_at_generate"]
)
if callback is not None:
callback("Database file cleared successfully", MSG_TYPE.MSG_TYPE_FULL)
else:
if callback is not None:
callback("The database file does not exist yet, so you can't clear it", MSG_TYPE.MSG_TYPE_FULL)
self.state = 0
else:
if self.state ==1:
try:
self.add_file(prompt)
if callback is not None:
callback(f"File {prompt} added successfully", MSG_TYPE.MSG_TYPE_FULL)
except Exception as ex:
ASCIIColors.error(f"Exception: {ex}")
if callback is not None:
callback(f"Couldn't load file {prompt}.\nThe following exception was thrown: {ex}", MSG_TYPE.MSG_TYPE_FULL)
output = str(ex)
self.state=0
elif self.state ==2:
try:
new_db_path = Path(prompt)
if new_db_path.exists():
self.personality_config["database_path"] = prompt
self.personality_config.save()
self.vector_store = TextVectorizer(
"bert-base-uncased",
self.personality.lollms_paths.personal_data_path/self.personality_config["database_path"],
visualize_data_at_startup=self.personality_config["visualize_data_at_startup"],
visualize_data_at_add_file=self.personality_config["visualize_data_at_add_file"],
visualize_data_at_generate=self.personality_config["visualize_data_at_generate"]
)
self.save_config_file(self.personality.lollms_paths.personal_configuration_path/f"personality_{self.personality.name}.yaml", self.personality_config)
else:
output = "Database file not found.\nGoing back to default state."
except Exception as ex:
ASCIIColors.error(f"Exception: {ex}")
output = str(ex)
self.state=0
else:
if not self.ready:
ASCIIColors.error(f"No data to discuss. Please upload a document first")
else:
docs = self.vector_store.recover_text(self.vector_store.embed_query(prompt), top_k=3)
docs = '\n'.join([f"Doc{i}:\n{v}" for i,v in enumerate(docs)])
full_text = self.personality.personality_conditioning+"\n### Docs:\n"+docs+"\n### Question: "+prompt+"\n### Answer:"
ASCIIColors.blue("-------------- Documentation -----------------------")
ASCIIColors.blue(full_text)
ASCIIColors.blue("----------------------------------------------------")
ASCIIColors.blue("Thinking")
if callback is not None:
callback("Thinking", MSG_TYPE.MSG_TYPE_FULL)
output = self.generate(full_text, self.personality_config["max_answer_size"])
if callback is not None:
callback(output, MSG_TYPE.MSG_TYPE_FULL)
return output
|