File size: 21,933 Bytes
23804b3 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 |
"""
DVC (Data Version Control) Integration for Cyber-LLM
Provides data versioning, experiment tracking, and pipeline management
"""
import os
import json
import yaml
import subprocess
import asyncio
from typing import Dict, List, Any, Optional, Tuple, Union
from datetime import datetime
from pathlib import Path
from dataclasses import dataclass, asdict
import hashlib
import tempfile
from .logging_system import CyberLLMLogger, CyberLLMError, ErrorCategory, retry_with_backoff
@dataclass
class DVCMetrics:
"""DVC metrics for model evaluation"""
accuracy: float
precision: float
recall: float
f1_score: float
loss: float
stealth_score: Optional[float] = None
chain_success_rate: Optional[float] = None
false_positive_rate: Optional[float] = None
safety_compliance: Optional[float] = None
@dataclass
class DVCParameters:
"""DVC parameters for experiments"""
learning_rate: float
batch_size: int
epochs: int
model_name: str
dataset_version: str
adapter_rank: Optional[int] = None
dropout_rate: Optional[float] = None
warmup_steps: Optional[int] = None
@dataclass
class DVCExperiment:
"""DVC experiment information"""
id: str
name: str
timestamp: datetime
parameters: DVCParameters
metrics: DVCMetrics
git_commit: str
status: str
duration: Optional[float] = None
class DVCManager:
"""DVC integration manager for data versioning and experiment tracking"""
def __init__(self,
repo_path: str = ".",
remote_name: str = "origin",
logger: Optional[CyberLLMLogger] = None):
self.repo_path = Path(repo_path).resolve()
self.remote_name = remote_name
self.logger = logger or CyberLLMLogger(name="dvc_manager")
# DVC configuration paths
self.dvc_dir = self.repo_path / ".dvc"
self.params_file = self.repo_path / "params.yaml"
self.metrics_file = self.repo_path / "metrics.yaml"
self.dvcfile = self.repo_path / "dvc.yaml"
# Initialize DVC if not already initialized
self._ensure_dvc_initialized()
def _ensure_dvc_initialized(self):
"""Ensure DVC is initialized in the repository"""
if not self.dvc_dir.exists():
self.logger.info("Initializing DVC repository")
self._run_dvc_command(["init"])
self._setup_default_config()
def _setup_default_config(self):
"""Setup default DVC configuration"""
# Create default params.yaml
default_params = {
"training": {
"learning_rate": 2e-5,
"batch_size": 8,
"epochs": 3,
"model_name": "microsoft/DialoGPT-medium",
"dataset_version": "v1.0",
"adapter_rank": 16,
"dropout_rate": 0.1,
"warmup_steps": 500
},
"data": {
"train_split": 0.8,
"val_split": 0.1,
"test_split": 0.1,
"max_length": 512,
"min_samples_per_class": 100
},
"evaluation": {
"batch_size": 16,
"max_samples": 1000,
"metrics_threshold": {
"accuracy": 0.85,
"stealth_score": 0.7,
"safety_compliance": 0.95
}
}
}
if not self.params_file.exists():
with open(self.params_file, 'w') as f:
yaml.dump(default_params, f, default_flow_style=False)
self.logger.info("Created default params.yaml")
# Create default dvc.yaml pipeline
default_pipeline = {
"stages": {
"data_preparation": {
"cmd": "python src/training/data_preprocessing.py",
"deps": [
"src/training/data_preprocessing.py",
"data/raw/"
],
"outs": [
"data/processed/train.jsonl",
"data/processed/val.jsonl",
"data/processed/test.jsonl"
],
"params": [
"data.train_split",
"data.val_split",
"data.test_split"
]
},
"training": {
"cmd": "python src/training/train_model.py",
"deps": [
"src/training/train_model.py",
"data/processed/train.jsonl",
"data/processed/val.jsonl"
],
"outs": [
"models/cyber_llm_adapter/"
],
"params": [
"training"
],
"metrics": [
"metrics/training.yaml"
]
},
"evaluation": {
"cmd": "python src/evaluation/evaluate_model.py",
"deps": [
"src/evaluation/evaluate_model.py",
"models/cyber_llm_adapter/",
"data/processed/test.jsonl"
],
"metrics": [
"metrics/evaluation.yaml"
],
"params": [
"evaluation"
]
}
}
}
if not self.dvcfile.exists():
with open(self.dvcfile, 'w') as f:
yaml.dump(default_pipeline, f, default_flow_style=False)
self.logger.info("Created default dvc.yaml pipeline")
def _run_dvc_command(self, args: List[str], check: bool = True) -> subprocess.CompletedProcess:
"""Run DVC command and return result"""
cmd = ["dvc"] + args
try:
result = subprocess.run(
cmd,
cwd=self.repo_path,
capture_output=True,
text=True,
check=check
)
if result.stdout:
self.logger.debug(f"DVC command output: {result.stdout.strip()}")
return result
except subprocess.CalledProcessError as e:
error_msg = f"DVC command failed: {' '.join(cmd)}\nError: {e.stderr}"
self.logger.error(error_msg)
raise CyberLLMError(error_msg, ErrorCategory.SYSTEM)
except FileNotFoundError:
raise CyberLLMError("DVC not found. Please install DVC: pip install dvc", ErrorCategory.SYSTEM)
@retry_with_backoff(max_retries=3)
async def add_data(self, data_path: str, remote: bool = True) -> bool:
"""Add data to DVC tracking"""
try:
# Add to DVC
self._run_dvc_command(["add", data_path])
# Add .dvc file to git
dvc_file = f"{data_path}.dvc"
if os.path.exists(dvc_file):
subprocess.run(["git", "add", dvc_file],
cwd=self.repo_path, check=True)
# Push to remote if requested
if remote:
await self.push_data()
self.logger.info(f"Added data to DVC: {data_path}")
return True
except Exception as e:
self.logger.error(f"Failed to add data to DVC: {data_path}", error=str(e))
return False
async def push_data(self) -> bool:
"""Push data to DVC remote"""
try:
# Run push in background
process = await asyncio.create_subprocess_exec(
"dvc", "push",
cwd=self.repo_path,
stdout=asyncio.subprocess.PIPE,
stderr=asyncio.subprocess.PIPE
)
stdout, stderr = await process.communicate()
if process.returncode == 0:
self.logger.info("Successfully pushed data to DVC remote")
return True
else:
self.logger.error("Failed to push data to DVC remote",
error=stderr.decode())
return False
except Exception as e:
self.logger.error("DVC push failed", error=str(e))
return False
async def pull_data(self) -> bool:
"""Pull data from DVC remote"""
try:
process = await asyncio.create_subprocess_exec(
"dvc", "pull",
cwd=self.repo_path,
stdout=asyncio.subprocess.PIPE,
stderr=asyncio.subprocess.PIPE
)
stdout, stderr = await process.communicate()
if process.returncode == 0:
self.logger.info("Successfully pulled data from DVC remote")
return True
else:
self.logger.error("Failed to pull data from DVC remote",
error=stderr.decode())
return False
except Exception as e:
self.logger.error("DVC pull failed", error=str(e))
return False
def create_experiment(self,
name: str,
parameters: DVCParameters,
description: str = "") -> str:
"""Create a new DVC experiment"""
# Generate experiment ID
exp_id = hashlib.md5(f"{name}_{datetime.now().isoformat()}".encode()).hexdigest()[:8]
# Update params.yaml with experiment parameters
self.update_parameters(asdict(parameters))
# Create experiment branch
try:
self._run_dvc_command(["exp", "run", "--name", name, "--set-param", f"experiment.id={exp_id}"])
self.logger.info(f"Created DVC experiment: {name} (ID: {exp_id})")
return exp_id
except Exception as e:
self.logger.error(f"Failed to create experiment: {name}", error=str(e))
raise
def update_parameters(self, params: Dict[str, Any]):
"""Update parameters file"""
try:
# Load existing params
existing_params = {}
if self.params_file.exists():
with open(self.params_file, 'r') as f:
existing_params = yaml.safe_load(f) or {}
# Deep merge new parameters
def deep_merge(base: dict, update: dict) -> dict:
for key, value in update.items():
if key in base and isinstance(base[key], dict) and isinstance(value, dict):
deep_merge(base[key], value)
else:
base[key] = value
return base
merged_params = deep_merge(existing_params, params)
# Write updated params
with open(self.params_file, 'w') as f:
yaml.dump(merged_params, f, default_flow_style=False)
self.logger.debug("Updated parameters file")
except Exception as e:
self.logger.error("Failed to update parameters", error=str(e))
raise
def log_metrics(self, metrics: DVCMetrics, stage: str = "evaluation"):
"""Log metrics to DVC"""
try:
metrics_dir = self.repo_path / "metrics"
metrics_dir.mkdir(exist_ok=True)
metrics_file = metrics_dir / f"{stage}.yaml"
# Convert metrics to dict
metrics_dict = asdict(metrics)
# Write metrics
with open(metrics_file, 'w') as f:
yaml.dump(metrics_dict, f, default_flow_style=False)
self.logger.info(f"Logged metrics for stage: {stage}")
except Exception as e:
self.logger.error(f"Failed to log metrics for stage: {stage}", error=str(e))
raise
def get_experiments(self) -> List[DVCExperiment]:
"""Get list of DVC experiments"""
try:
result = self._run_dvc_command(["exp", "show", "--json"])
experiments = []
if result.stdout:
exp_data = json.loads(result.stdout)
for exp_info in exp_data:
# Parse experiment data
exp = DVCExperiment(
id=exp_info.get("id", ""),
name=exp_info.get("name", ""),
timestamp=datetime.fromisoformat(exp_info.get("timestamp", datetime.now().isoformat())),
parameters=DVCParameters(**exp_info.get("params", {})),
metrics=DVCMetrics(**exp_info.get("metrics", {})),
git_commit=exp_info.get("rev", ""),
status=exp_info.get("status", "unknown"),
duration=exp_info.get("duration")
)
experiments.append(exp)
return experiments
except Exception as e:
self.logger.error("Failed to get experiments", error=str(e))
return []
def compare_experiments(self, exp_ids: List[str]) -> Dict[str, Any]:
"""Compare multiple experiments"""
try:
cmd = ["exp", "diff"] + exp_ids
result = self._run_dvc_command(cmd)
# Parse diff output (simplified)
comparison = {
"experiments": exp_ids,
"timestamp": datetime.now().isoformat(),
"raw_output": result.stdout
}
self.logger.info(f"Compared experiments: {', '.join(exp_ids)}")
return comparison
except Exception as e:
self.logger.error(f"Failed to compare experiments: {exp_ids}", error=str(e))
return {}
async def run_pipeline(self,
stages: Optional[List[str]] = None,
force: bool = False) -> bool:
"""Run DVC pipeline"""
try:
cmd = ["repro"]
if force:
cmd.append("--force")
if stages:
cmd.extend(stages)
self.logger.info(f"Starting DVC pipeline: {' '.join(cmd)}")
process = await asyncio.create_subprocess_exec(
"dvc", *cmd,
cwd=self.repo_path,
stdout=asyncio.subprocess.PIPE,
stderr=asyncio.subprocess.PIPE
)
stdout, stderr = await process.communicate()
if process.returncode == 0:
self.logger.info("DVC pipeline completed successfully")
if stdout:
self.logger.debug(f"Pipeline output: {stdout.decode()}")
return True
else:
self.logger.error("DVC pipeline failed", error=stderr.decode())
return False
except Exception as e:
self.logger.error("Failed to run DVC pipeline", error=str(e))
return False
def setup_remote_storage(self,
storage_type: str,
config: Dict[str, str]) -> bool:
"""Setup DVC remote storage"""
try:
remote_name = config.get("name", "default")
if storage_type == "s3":
url = f"s3://{config['bucket']}/{config.get('prefix', '')}"
self._run_dvc_command(["remote", "add", "-d", remote_name, url])
# Set AWS credentials if provided
if "access_key_id" in config:
self._run_dvc_command(["remote", "modify", remote_name,
"access_key_id", config["access_key_id"]])
if "secret_access_key" in config:
self._run_dvc_command(["remote", "modify", remote_name,
"secret_access_key", config["secret_access_key"]])
if "region" in config:
self._run_dvc_command(["remote", "modify", remote_name,
"region", config["region"]])
elif storage_type == "azure":
url = f"azure://{config['container']}/{config.get('prefix', '')}"
self._run_dvc_command(["remote", "add", "-d", remote_name, url])
if "account_name" in config:
self._run_dvc_command(["remote", "modify", remote_name,
"account_name", config["account_name"]])
elif storage_type == "gcs":
url = f"gs://{config['bucket']}/{config.get('prefix', '')}"
self._run_dvc_command(["remote", "add", "-d", remote_name, url])
elif storage_type == "ssh":
url = f"ssh://{config['host']}{config['path']}"
self._run_dvc_command(["remote", "add", "-d", remote_name, url])
if "user" in config:
self._run_dvc_command(["remote", "modify", remote_name,
"user", config["user"]])
else:
raise ValueError(f"Unsupported storage type: {storage_type}")
self.logger.info(f"Setup DVC remote storage: {storage_type} ({remote_name})")
return True
except Exception as e:
self.logger.error(f"Failed to setup remote storage: {storage_type}", error=str(e))
return False
def get_data_info(self, data_path: str) -> Dict[str, Any]:
"""Get information about tracked data"""
try:
dvc_file = f"{data_path}.dvc"
if not os.path.exists(dvc_file):
return {"tracked": False}
# Parse .dvc file
with open(dvc_file, 'r') as f:
dvc_data = yaml.safe_load(f)
# Get file info
file_info = {
"tracked": True,
"path": data_path,
"dvc_file": dvc_file,
"md5": dvc_data.get("outs", [{}])[0].get("md5", ""),
"size": os.path.getsize(data_path) if os.path.exists(data_path) else 0,
"remote_available": self._check_remote_availability(data_path)
}
return file_info
except Exception as e:
self.logger.error(f"Failed to get data info: {data_path}", error=str(e))
return {"tracked": False, "error": str(e)}
def _check_remote_availability(self, data_path: str) -> bool:
"""Check if data is available in remote storage"""
try:
result = self._run_dvc_command(["status", data_path], check=False)
return "not in cache" not in result.stdout.lower()
except:
return False
# Convenience functions
def init_dvc_project(repo_path: str = ".") -> DVCManager:
"""Initialize DVC project"""
return DVCManager(repo_path=repo_path)
async def track_dataset(dataset_path: str,
dvc_manager: Optional[DVCManager] = None) -> bool:
"""Track dataset with DVC"""
manager = dvc_manager or DVCManager()
return await manager.add_data(dataset_path)
def create_training_experiment(name: str,
learning_rate: float = 2e-5,
batch_size: int = 8,
epochs: int = 3,
dvc_manager: Optional[DVCManager] = None) -> str:
"""Create training experiment"""
manager = dvc_manager or DVCManager()
params = DVCParameters(
learning_rate=learning_rate,
batch_size=batch_size,
epochs=epochs,
model_name="microsoft/DialoGPT-medium",
dataset_version="v1.0"
)
return manager.create_experiment(name, params)
# Example usage
if __name__ == "__main__":
import asyncio
async def main():
# Initialize DVC manager
dvc = DVCManager()
# Track a dataset
success = await dvc.add_data("data/raw/cyber_dataset.jsonl")
print(f"Dataset tracking: {'success' if success else 'failed'}")
# Create experiment
params = DVCParameters(
learning_rate=1e-4,
batch_size=16,
epochs=5,
model_name="microsoft/DialoGPT-medium",
dataset_version="v1.0",
adapter_rank=32
)
exp_id = dvc.create_experiment("experiment_001", params)
print(f"Created experiment: {exp_id}")
# Log metrics
metrics = DVCMetrics(
accuracy=0.87,
precision=0.85,
recall=0.89,
f1_score=0.87,
loss=0.23,
stealth_score=0.73,
safety_compliance=0.96
)
dvc.log_metrics(metrics)
# Run pipeline
success = await dvc.run_pipeline()
print(f"Pipeline execution: {'success' if success else 'failed'}")
asyncio.run(main())
|