File size: 14,038 Bytes
338d95d |
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 |
#!/usr/bin/env python3
"""
CompI Phase 1.E: LoRA Style Management System
Manage multiple LoRA styles, switch between them, and organize trained models.
Usage:
python src/generators/compi_phase1e_style_manager.py --list
python src/generators/compi_phase1e_style_manager.py --info my_style
python src/generators/compi_phase1e_style_manager.py --cleanup
"""
import os
import argparse
import json
import shutil
from datetime import datetime
from pathlib import Path
from typing import Dict, List, Optional, Tuple
import pandas as pd
# -------- 1. CONFIGURATION --------
LORA_MODELS_DIR = "lora_models"
STYLES_CONFIG_FILE = "lora_styles_config.json"
# -------- 2. STYLE MANAGEMENT CLASS --------
class LoRAStyleManager:
"""Manager for LoRA styles and models."""
def __init__(self, models_dir: str = LORA_MODELS_DIR):
self.models_dir = Path(models_dir)
self.models_dir.mkdir(exist_ok=True)
self.config_file = self.models_dir / STYLES_CONFIG_FILE
self.config = self.load_config()
def load_config(self) -> Dict:
"""Load styles configuration."""
if self.config_file.exists():
with open(self.config_file) as f:
return json.load(f)
return {"styles": {}, "last_updated": datetime.now().isoformat()}
def save_config(self):
"""Save styles configuration."""
self.config["last_updated"] = datetime.now().isoformat()
with open(self.config_file, 'w') as f:
json.dump(self.config, f, indent=2)
def scan_styles(self) -> Dict[str, Dict]:
"""Scan for available LoRA styles."""
styles = {}
for style_dir in self.models_dir.iterdir():
if not style_dir.is_dir() or style_dir.name.startswith('.'):
continue
# Look for checkpoints
checkpoints = list(style_dir.glob("checkpoint-*"))
if not checkpoints:
continue
# Get latest checkpoint
latest_checkpoint = max(checkpoints, key=lambda x: int(x.name.split('-')[1]))
# Load training info
info_file = style_dir / "training_info.json"
if info_file.exists():
with open(info_file) as f:
training_info = json.load(f)
else:
training_info = {}
# Load dataset info if available
dataset_info = {}
for dataset_dir in [style_dir / "dataset", Path("datasets") / style_dir.name]:
dataset_info_file = dataset_dir / "dataset_info.json"
if dataset_info_file.exists():
with open(dataset_info_file) as f:
dataset_info = json.load(f)
break
# Compile style information
style_info = {
"name": style_dir.name,
"path": str(style_dir),
"latest_checkpoint": str(latest_checkpoint),
"checkpoints": [str(cp) for cp in checkpoints],
"training_info": training_info,
"dataset_info": dataset_info,
"last_scanned": datetime.now().isoformat()
}
styles[style_dir.name] = style_info
return styles
def refresh_styles(self):
"""Refresh the styles database."""
print("π Scanning for LoRA styles...")
scanned_styles = self.scan_styles()
# Update config
self.config["styles"] = scanned_styles
self.save_config()
print(f"β
Found {len(scanned_styles)} LoRA style(s)")
return scanned_styles
def list_styles(self, detailed: bool = False) -> List[Dict]:
"""List available styles."""
styles = self.config.get("styles", {})
if not styles:
styles = self.refresh_styles()
if detailed:
return list(styles.values())
else:
return [{"name": name, "checkpoints": len(info["checkpoints"])}
for name, info in styles.items()]
def get_style_info(self, style_name: str) -> Optional[Dict]:
"""Get detailed information about a specific style."""
styles = self.config.get("styles", {})
return styles.get(style_name)
def get_best_checkpoint(self, style_name: str) -> Optional[str]:
"""Get the best checkpoint for a style."""
style_info = self.get_style_info(style_name)
if not style_info:
return None
# For now, return the latest checkpoint
# Could be enhanced to track validation loss and return best performing
return style_info.get("latest_checkpoint")
def delete_style(self, style_name: str, confirm: bool = False) -> bool:
"""Delete a LoRA style."""
if not confirm:
print("β οΈ Use --confirm to actually delete the style")
return False
style_dir = self.models_dir / style_name
if not style_dir.exists():
print(f"β Style not found: {style_name}")
return False
try:
shutil.rmtree(style_dir)
# Remove from config
if style_name in self.config.get("styles", {}):
del self.config["styles"][style_name]
self.save_config()
print(f"β
Deleted style: {style_name}")
return True
except Exception as e:
print(f"β Error deleting style: {e}")
return False
def cleanup_checkpoints(self, style_name: str, keep_last: int = 3) -> int:
"""Clean up old checkpoints, keeping only the most recent ones."""
style_dir = self.models_dir / style_name
if not style_dir.exists():
print(f"β Style not found: {style_name}")
return 0
checkpoints = list(style_dir.glob("checkpoint-*"))
if len(checkpoints) <= keep_last:
print(f"β
No cleanup needed for {style_name} ({len(checkpoints)} checkpoints)")
return 0
# Sort by step number
checkpoints.sort(key=lambda x: int(x.name.split('-')[1]))
# Remove old checkpoints
to_remove = checkpoints[:-keep_last]
removed_count = 0
for checkpoint in to_remove:
try:
shutil.rmtree(checkpoint)
removed_count += 1
except Exception as e:
print(f"β οΈ Failed to remove {checkpoint}: {e}")
print(f"β
Cleaned up {removed_count} old checkpoints for {style_name}")
return removed_count
def export_style_info(self, output_file: str = None) -> str:
"""Export styles information to CSV."""
styles = self.list_styles(detailed=True)
if not styles:
print("β No styles found")
return ""
# Prepare data for CSV
rows = []
for style in styles:
training_info = style.get("training_info", {})
dataset_info = style.get("dataset_info", {})
row = {
"style_name": style["name"],
"checkpoints": len(style["checkpoints"]),
"latest_checkpoint": Path(style["latest_checkpoint"]).name,
"total_steps": training_info.get("total_steps", "unknown"),
"epochs": training_info.get("epochs", "unknown"),
"learning_rate": training_info.get("learning_rate", "unknown"),
"lora_rank": training_info.get("lora_rank", "unknown"),
"dataset_images": dataset_info.get("total_images", "unknown"),
"trigger_word": dataset_info.get("trigger_word", "unknown"),
"last_scanned": style.get("last_scanned", "unknown")
}
rows.append(row)
# Create DataFrame and save
df = pd.DataFrame(rows)
if output_file is None:
output_file = f"lora_styles_export_{datetime.now().strftime('%Y%m%d_%H%M%S')}.csv"
df.to_csv(output_file, index=False)
print(f"π Exported styles info to: {output_file}")
return output_file
# -------- 3. COMMAND LINE INTERFACE --------
def setup_args():
"""Setup command line arguments."""
parser = argparse.ArgumentParser(
description="CompI Phase 1.E: LoRA Style Management",
formatter_class=argparse.RawDescriptionHelpFormatter,
epilog="""
Examples:
# List all available styles
python %(prog)s --list
# Get detailed info about a specific style
python %(prog)s --info my_style
# Refresh styles database
python %(prog)s --refresh
# Clean up old checkpoints
python %(prog)s --cleanup my_style --keep 2
# Export styles information
python %(prog)s --export styles_report.csv
"""
)
parser.add_argument("--list", action="store_true",
help="List all available LoRA styles")
parser.add_argument("--list-detailed", action="store_true",
help="List styles with detailed information")
parser.add_argument("--info", metavar="STYLE_NAME",
help="Show detailed information about a specific style")
parser.add_argument("--refresh", action="store_true",
help="Refresh the styles database")
parser.add_argument("--cleanup", metavar="STYLE_NAME",
help="Clean up old checkpoints for a style")
parser.add_argument("--keep", type=int, default=3,
help="Number of recent checkpoints to keep during cleanup")
parser.add_argument("--delete", metavar="STYLE_NAME",
help="Delete a LoRA style")
parser.add_argument("--confirm", action="store_true",
help="Confirm destructive operations")
parser.add_argument("--export", metavar="OUTPUT_FILE",
help="Export styles information to CSV")
parser.add_argument("--models-dir", default=LORA_MODELS_DIR,
help=f"LoRA models directory (default: {LORA_MODELS_DIR})")
return parser.parse_args()
def print_style_info(style_info: Dict):
"""Print detailed style information."""
print(f"π¨ Style: {style_info['name']}")
print("=" * 40)
# Basic info
print(f"π Path: {style_info['path']}")
print(f"π Checkpoints: {len(style_info['checkpoints'])}")
print(f"π Latest: {Path(style_info['latest_checkpoint']).name}")
# Training info
training_info = style_info.get("training_info", {})
if training_info:
print(f"\nπ Training Information:")
print(f" Steps: {training_info.get('total_steps', 'unknown')}")
print(f" Epochs: {training_info.get('epochs', 'unknown')}")
print(f" Learning Rate: {training_info.get('learning_rate', 'unknown')}")
print(f" LoRA Rank: {training_info.get('lora_rank', 'unknown')}")
print(f" LoRA Alpha: {training_info.get('lora_alpha', 'unknown')}")
# Dataset info
dataset_info = style_info.get("dataset_info", {})
if dataset_info:
print(f"\nπ Dataset Information:")
print(f" Total Images: {dataset_info.get('total_images', 'unknown')}")
print(f" Train Images: {dataset_info.get('train_images', 'unknown')}")
print(f" Validation Images: {dataset_info.get('validation_images', 'unknown')}")
print(f" Trigger Word: {dataset_info.get('trigger_word', 'unknown')}")
print(f" Image Size: {dataset_info.get('image_size', 'unknown')}")
print(f"\nπ Last Scanned: {style_info.get('last_scanned', 'unknown')}")
def main():
"""Main function."""
args = setup_args()
# Initialize style manager
manager = LoRAStyleManager(args.models_dir)
print("π¨ CompI Phase 1.E: LoRA Style Manager")
print("=" * 40)
# Execute commands
if args.refresh:
manager.refresh_styles()
elif args.list or args.list_detailed:
styles = manager.list_styles(detailed=args.list_detailed)
if not styles:
print("β No LoRA styles found")
print("π‘ Train a style first using: python src/generators/compi_phase1e_lora_training.py")
else:
print(f"π Available LoRA Styles ({len(styles)}):")
print("-" * 40)
if args.list_detailed:
for style in styles:
print_style_info(style)
print()
else:
for style in styles:
print(f"π¨ {style['name']} ({style['checkpoints']} checkpoints)")
elif args.info:
style_info = manager.get_style_info(args.info)
if style_info:
print_style_info(style_info)
else:
print(f"β Style not found: {args.info}")
print("π‘ Use --list to see available styles")
elif args.cleanup:
removed = manager.cleanup_checkpoints(args.cleanup, args.keep)
if removed > 0:
manager.refresh_styles()
elif args.delete:
manager.delete_style(args.delete, args.confirm)
if args.confirm:
manager.refresh_styles()
elif args.export:
manager.export_style_info(args.export)
else:
print("β No command specified. Use --help for usage information.")
print("π‘ Common commands:")
print(" --list List available styles")
print(" --info STYLE_NAME Show style details")
print(" --refresh Refresh styles database")
return 0
if __name__ == "__main__":
exit(main())
|