File size: 20,397 Bytes
d0aa8c4 |
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 |
#!/usr/bin/env python3
import os
import shutil
import glob
import base64
import streamlit as st
import pandas as pd
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
from torch.utils.data import Dataset, DataLoader
import csv
import time
from dataclasses import dataclass
from typing import Optional, Tuple
import zipfile
import math
from PIL import Image
import random
import logging
from datetime import datetime
import pytz
from diffusers import StableDiffusionPipeline # For diffusion models
from urllib.parse import quote
# Set up logging
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)
# Page Configuration
st.set_page_config(
page_title="SFT Tiny Titans π",
page_icon="π€",
layout="wide",
initial_sidebar_state="expanded",
menu_items={
'Get Help': 'https://huggingface.co/awacke1',
'Report a bug': 'https://huggingface.co/spaces/awacke1',
'About': "Tiny Titans: Small models, big dreams, and a sprinkle of chaos! π"
}
)
# Model Configuration Classes
@dataclass
class ModelConfig:
name: str
base_model: str
size: str
domain: Optional[str] = None
model_type: str = "causal_lm"
@property
def model_path(self):
return f"models/{self.name}"
@dataclass
class DiffusionConfig:
name: str
base_model: str
size: str
@property
def model_path(self):
return f"diffusion_models/{self.name}"
# Datasets
class SFTDataset(Dataset):
def __init__(self, data, tokenizer, max_length=128):
self.data = data
self.tokenizer = tokenizer
self.max_length = max_length
def __len__(self):
return len(self.data)
def __getitem__(self, idx):
prompt = self.data[idx]["prompt"]
response = self.data[idx]["response"]
full_text = f"{prompt} {response}"
full_encoding = self.tokenizer(full_text, max_length=self.max_length, padding="max_length", truncation=True, return_tensors="pt")
prompt_encoding = self.tokenizer(prompt, max_length=self.max_length, padding=False, truncation=True, return_tensors="pt")
input_ids = full_encoding["input_ids"].squeeze()
attention_mask = full_encoding["attention_mask"].squeeze()
labels = input_ids.clone()
prompt_len = prompt_encoding["input_ids"].shape[1]
if prompt_len < self.max_length:
labels[:prompt_len] = -100
return {"input_ids": input_ids, "attention_mask": attention_mask, "labels": labels}
class DiffusionDataset(Dataset):
def __init__(self, images, texts):
self.images = images
self.texts = texts
def __len__(self):
return len(self.images)
def __getitem__(self, idx):
return {"image": self.images[idx], "text": self.texts[idx]}
# Model Builder Classes
class ModelBuilder:
def __init__(self):
self.config = None
self.model = None
self.tokenizer = None
self.sft_data = None
self.jokes = ["Why did the AI go to therapy? Too many layers to unpack! π", "Training complete! Time for a binary coffee break. β"]
def load_model(self, model_path: str, config: Optional[ModelConfig] = None):
with st.spinner(f"Loading {model_path}... β³"):
self.model = AutoModelForCausalLM.from_pretrained(model_path)
self.tokenizer = AutoTokenizer.from_pretrained(model_path)
if self.tokenizer.pad_token is None:
self.tokenizer.pad_token = self.tokenizer.eos_token
if config:
self.config = config
st.success(f"Model loaded! π {random.choice(self.jokes)}")
return self
def fine_tune_sft(self, csv_path: str, epochs: int = 3, batch_size: int = 4):
self.sft_data = []
with open(csv_path, "r") as f:
reader = csv.DictReader(f)
for row in reader:
self.sft_data.append({"prompt": row["prompt"], "response": row["response"]})
dataset = SFTDataset(self.sft_data, self.tokenizer)
dataloader = DataLoader(dataset, batch_size=batch_size, shuffle=True)
optimizer = torch.optim.AdamW(self.model.parameters(), lr=2e-5)
self.model.train()
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
self.model.to(device)
for epoch in range(epochs):
with st.spinner(f"Training epoch {epoch + 1}/{epochs}... βοΈ"):
total_loss = 0
for batch in dataloader:
optimizer.zero_grad()
input_ids = batch["input_ids"].to(device)
attention_mask = batch["attention_mask"].to(device)
labels = batch["labels"].to(device)
outputs = self.model(input_ids=input_ids, attention_mask=attention_mask, labels=labels)
loss = outputs.loss
loss.backward()
optimizer.step()
total_loss += loss.item()
st.write(f"Epoch {epoch + 1} completed. Average loss: {total_loss / len(dataloader):.4f}")
st.success(f"SFT Fine-tuning completed! π {random.choice(self.jokes)}")
return self
def save_model(self, path: str):
with st.spinner("Saving model... πΎ"):
os.makedirs(os.path.dirname(path), exist_ok=True)
self.model.save_pretrained(path)
self.tokenizer.save_pretrained(path)
st.success(f"Model saved at {path}! β
")
def evaluate(self, prompt: str, status_container=None):
self.model.eval()
if status_container:
status_container.write("Preparing to evaluate... π§ ")
try:
with torch.no_grad():
inputs = self.tokenizer(prompt, return_tensors="pt", max_length=128, truncation=True).to(self.model.device)
outputs = self.model.generate(**inputs, max_new_tokens=50, do_sample=True, top_p=0.95, temperature=0.7)
return self.tokenizer.decode(outputs[0], skip_special_tokens=True)
except Exception as e:
if status_container:
status_container.error(f"Oops! Something broke: {str(e)} π₯")
return f"Error: {str(e)}"
class DiffusionBuilder:
def __init__(self):
self.config = None
self.pipeline = None
def load_model(self, model_path: str, config: Optional[DiffusionConfig] = None):
with st.spinner(f"Loading diffusion model {model_path}... β³"):
self.pipeline = StableDiffusionPipeline.from_pretrained(model_path)
self.pipeline.to("cuda" if torch.cuda.is_available() else "cpu")
if config:
self.config = config
st.success(f"Diffusion model loaded! π¨")
return self
def fine_tune_sft(self, images, texts, epochs=3):
dataset = DiffusionDataset(images, texts)
dataloader = DataLoader(dataset, batch_size=1, shuffle=True)
optimizer = torch.optim.AdamW(self.pipeline.unet.parameters(), lr=1e-5)
self.pipeline.unet.train()
for epoch in range(epochs):
with st.spinner(f"Training diffusion epoch {epoch + 1}/{epochs}... βοΈ"):
total_loss = 0
for batch in dataloader:
optimizer.zero_grad()
image = batch["image"].to(self.pipeline.device)
text = batch["text"]
latents = self.pipeline.vae.encode(image).latent_dist.sample()
noise = torch.randn_like(latents)
timesteps = torch.randint(0, self.pipeline.scheduler.num_train_timesteps, (latents.shape[0],), device=latents.device)
noisy_latents = self.pipeline.scheduler.add_noise(latents, noise, timesteps)
text_embeddings = self.pipeline.text_encoder(self.pipeline.tokenizer(text, return_tensors="pt").input_ids.to(self.pipeline.device))[0]
pred_noise = self.pipeline.unet(noisy_latents, timesteps, encoder_hidden_states=text_embeddings).sample
loss = torch.nn.functional.mse_loss(pred_noise, noise)
loss.backward()
optimizer.step()
total_loss += loss.item()
st.write(f"Epoch {epoch + 1} completed. Average loss: {total_loss / len(dataloader):.4f}")
st.success("Diffusion SFT Fine-tuning completed! π¨")
return self
def save_model(self, path: str):
with st.spinner("Saving diffusion model... πΎ"):
os.makedirs(os.path.dirname(path), exist_ok=True)
self.pipeline.save_pretrained(path)
st.success(f"Diffusion model saved at {path}! β
")
# Utility Functions
def get_download_link(file_path, mime_type="text/plain", label="Download"):
with open(file_path, 'rb') as f:
data = f.read()
b64 = base64.b64encode(data).decode()
return f'<a href="data:{mime_type};base64,{b64}" download="{os.path.basename(file_path)}">{label} π₯</a>'
def zip_directory(directory_path, zip_path):
with zipfile.ZipFile(zip_path, 'w', zipfile.ZIP_DEFLATED) as zipf:
for root, _, files in os.walk(directory_path):
for file in files:
file_path = os.path.join(root, file)
arcname = os.path.relpath(file_path, os.path.dirname(directory_path))
zipf.write(file_path, arcname)
def get_model_files(model_type="causal_lm"):
path = "models/*" if model_type == "causal_lm" else "diffusion_models/*"
return [d for d in glob.glob(path) if os.path.isdir(d)]
def get_gallery_files(file_types):
files = []
for ext in file_types:
files.extend(glob.glob(f"*.{ext}"))
return sorted(files)
def generate_filename(text_line):
central = pytz.timezone('US/Central')
timestamp = datetime.now(central).strftime("%Y%m%d_%I%M%S_%p")
safe_text = ''.join(c if c.isalnum() else '_' for c in text_line[:50])
return f"{timestamp}_{safe_text}.png"
def display_search_links(query):
search_urls = {
"ArXiv": f"https://arxiv.org/search/?query={quote(query)}",
"Wikipedia": f"https://en.wikipedia.org/wiki/{quote(query)}",
"Google": f"https://www.google.com/search?q={quote(query)}",
"YouTube": f"https://www.youtube.com/results?search_query={quote(query)}"
}
links_md = ' '.join([f"[{name}]({url})" for name, url in search_urls.items()])
return links_md
# Agent Class
class PartyPlannerAgent:
def __init__(self, model, tokenizer):
self.model = model
self.tokenizer = tokenizer
self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
self.model.to(self.device)
def generate(self, prompt: str) -> str:
self.model.eval()
with torch.no_grad():
inputs = self.tokenizer(prompt, return_tensors="pt", max_length=128, truncation=True).to(self.device)
outputs = self.model.generate(**inputs, max_new_tokens=100, do_sample=True, top_p=0.95, temperature=0.7)
return self.tokenizer.decode(outputs[0], skip_special_tokens=True)
def plan_party(self, task: str) -> pd.DataFrame:
search_result = "Latest trends for 2025: Gold-plated Batman statues, VR superhero battles."
prompt = f"Given this context: '{search_result}'\n{task}"
plan_text = self.generate(prompt)
st.markdown(f"Search Links: {display_search_links('superhero party trends')}", unsafe_allow_html=True)
locations = {"Wayne Manor": (42.3601, -71.0589), "New York": (40.7128, -74.0060), "Los Angeles": (34.0522, -118.2437), "London": (51.5074, -0.1278)}
wayne_coords = locations["Wayne Manor"]
travel_times = {loc: calculate_cargo_travel_time(coords, wayne_coords) for loc, coords in locations.items() if loc != "Wayne Manor"}
data = [
{"Location": "New York", "Travel Time (hrs)": travel_times["New York"], "Luxury Idea": "Gold-plated Batman statues"},
{"Location": "Los Angeles", "Travel Time (hrs)": travel_times["Los Angeles"], "Luxury Idea": "VR superhero battles"},
{"Location": "London", "Travel Time (hrs)": travel_times["London"], "Luxury Idea": "Live stunt shows"},
{"Location": "Wayne Manor", "Travel Time (hrs)": 0.0, "Luxury Idea": "Holographic displays"}
]
return pd.DataFrame(data)
def calculate_cargo_travel_time(origin_coords: Tuple[float, float], destination_coords: Tuple[float, float], cruising_speed_kmh: float = 750.0) -> float:
def to_radians(degrees: float) -> float:
return degrees * (math.pi / 180)
lat1, lon1 = map(to_radians, origin_coords)
lat2, lon2 = map(to_radians, destination_coords)
EARTH_RADIUS_KM = 6371.0
dlon = lon2 - lon1
dlat = lat2 - lat1
a = (math.sin(dlat / 2) ** 2 + math.cos(lat1) * math.cos(lat2) * math.sin(dlon / 2) ** 2)
c = 2 * math.asin(math.sqrt(a))
distance = EARTH_RADIUS_KM * c
actual_distance = distance * 1.1
flight_time = (actual_distance / cruising_speed_kmh) + 1.0
return round(flight_time, 2)
# Main App
st.title("SFT Tiny Titans π (Small but Mighty!)")
# Sidebar Galleries
st.sidebar.header("Galleries π¨")
for gallery_type, file_types in [
("Image Gallery πΈ", ["png", "jpg", "jpeg"]),
("Video Gallery π₯", ["mp4"]),
("Audio Gallery πΆ", ["mp3"])
]:
st.sidebar.subheader(gallery_type)
files = get_gallery_files(file_types)
if files:
cols_num = st.sidebar.slider(f"{gallery_type} Columns", 1, 5, 3, key=f"{gallery_type}_cols")
cols = st.sidebar.columns(cols_num)
for idx, file in enumerate(files[:cols_num * 2]):
with cols[idx % cols_num]:
if "Image" in gallery_type:
st.image(Image.open(file), caption=file, use_column_width=True)
elif "Video" in gallery_type:
st.video(file)
elif "Audio" in gallery_type:
st.audio(file)
st.sidebar.subheader("Model Management ποΈ")
model_type = st.sidebar.selectbox("Model Type", ["Causal LM", "Diffusion"])
model_dirs = get_model_files("causal_lm" if model_type == "Causal LM" else "diffusion")
selected_model = st.sidebar.selectbox("Select Saved Model", ["None"] + model_dirs)
if selected_model != "None" and st.sidebar.button("Load Model π"):
if 'builder' not in st.session_state:
st.session_state['builder'] = ModelBuilder() if model_type == "Causal LM" else DiffusionBuilder()
config = (ModelConfig if model_type == "Causal LM" else DiffusionConfig)(name=os.path.basename(selected_model), base_model="unknown", size="small")
st.session_state['builder'].load_model(selected_model, config)
st.session_state['model_loaded'] = True
st.rerun()
# Tabs
tab1, tab2, tab3, tab4, tab5 = st.tabs(["Build Tiny Titan π±", "Fine-Tune Titan π§", "Test Titan π§ͺ", "Agentic RAG Party π", "Diffusion SFT π¨"])
with tab1:
st.header("Build Tiny Titan π±")
model_type = st.selectbox("Model Type", ["Causal LM", "Diffusion"], key="build_type")
if model_type == "Causal LM":
base_model = st.selectbox("Select Tiny Model", ["HuggingFaceTB/SmolLM-135M", "HuggingFaceTB/SmolLM-360M", "Qwen/Qwen1.5-0.5B-Chat"])
else:
base_model = st.selectbox("Select Tiny Diffusion Model", ["stabilityai/stable-diffusion-2-1", "runwayml/stable-diffusion-v1-5", "CompVis/stable-diffusion-v1-4"])
model_name = st.text_input("Model Name", f"tiny-titan-{int(time.time())}")
if st.button("Download Model β¬οΈ"):
config = (ModelConfig if model_type == "Causal LM" else DiffusionConfig)(name=model_name, base_model=base_model, size="small")
builder = ModelBuilder() if model_type == "Causal LM" else DiffusionBuilder()
builder.load_model(base_model, config)
builder.save_model(config.model_path)
st.session_state['builder'] = builder
st.session_state['model_loaded'] = True
st.rerun()
with tab2:
st.header("Fine-Tune Titan π§")
if 'builder' not in st.session_state or not st.session_state.get('model_loaded', False):
st.warning("Please build or load a Titan first! β οΈ")
else:
if isinstance(st.session_state['builder'], ModelBuilder):
uploaded_csv = st.file_uploader("Upload CSV for SFT", type="csv")
if uploaded_csv and st.button("Fine-Tune with Uploaded CSV π"):
csv_path = f"uploaded_sft_data_{int(time.time())}.csv"
with open(csv_path, "wb") as f:
f.write(uploaded_csv.read())
new_model_name = f"{st.session_state['builder'].config.name}-sft-{int(time.time())}"
new_config = ModelConfig(name=new_model_name, base_model=st.session_state['builder'].config.base_model, size="small")
st.session_state['builder'].config = new_config
st.session_state['builder'].fine_tune_sft(csv_path)
st.session_state['builder'].save_model(new_config.model_path)
zip_path = f"{new_config.model_path}.zip"
zip_directory(new_config.model_path, zip_path)
st.markdown(get_download_link(zip_path, "application/zip", "Download Fine-Tuned Titan"), unsafe_allow_html=True)
with tab3:
st.header("Test Titan π§ͺ")
if 'builder' not in st.session_state or not st.session_state.get('model_loaded', False):
st.warning("Please build or load a Titan first! β οΈ")
else:
if isinstance(st.session_state['builder'], ModelBuilder):
test_prompt = st.text_area("Enter Test Prompt", "What is AI?")
if st.button("Run Test βΆοΈ"):
result = st.session_state['builder'].evaluate(test_prompt)
st.write(f"**Generated Response**: {result}")
with tab4:
st.header("Agentic RAG Party π")
if 'builder' not in st.session_state or not st.session_state.get('model_loaded', False) or not isinstance(st.session_state['builder'], ModelBuilder):
st.warning("Please build or load a Causal LM Titan first! β οΈ")
else:
if st.button("Run Agentic RAG Demo π"):
agent = PartyPlannerAgent(model=st.session_state['builder'].model, tokenizer=st.session_state['builder'].tokenizer)
task = "Plan a luxury superhero-themed party at Wayne Manor."
plan_df = agent.plan_party(task)
st.dataframe(plan_df)
with tab5:
st.header("Diffusion SFT π¨")
if 'builder' not in st.session_state or not st.session_state.get('model_loaded', False) or not isinstance(st.session_state['builder'], DiffusionBuilder):
st.warning("Please build or load a Diffusion Titan first! β οΈ")
else:
uploaded_files = st.file_uploader("Upload Images/Videos", type=["png", "jpg", "jpeg", "mp4", "mp3"], accept_multiple_files=True)
text_input = st.text_area("Enter Text (one line per image)", "Line 1\nLine 2\nLine 3")
if uploaded_files and st.button("Fine-Tune Diffusion Model π"):
images = [Image.open(f) for f in uploaded_files if f.type.startswith("image")]
texts = text_input.splitlines()
if len(images) > len(texts):
texts.extend([""] * (len(images) - len(texts)))
elif len(texts) > len(images):
texts = texts[:len(images)]
st.session_state['builder'].fine_tune_sft(images, texts)
new_model_name = f"{st.session_state['builder'].config.name}-sft-{int(time.time())}"
new_config = DiffusionConfig(name=new_model_name, base_model=st.session_state['builder'].config.base_model, size="small")
st.session_state['builder'].config = new_config
st.session_state['builder'].save_model(new_config.model_path)
for img, text in zip(images, texts):
filename = generate_filename(text)
img.save(filename)
st.image(img, caption=filename)
zip_path = f"{new_config.model_path}.zip"
zip_directory(new_config.model_path, zip_path)
st.markdown(get_download_link(zip_path, "application/zip", "Download Fine-Tuned Diffusion Model"), unsafe_allow_html=True) |