File size: 7,126 Bytes
18a82fb | 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 | import logging
import sys
from pathlib import Path
from datetime import datetime
import json
import wandb
from typing import Dict, Optional, Any
def setup_logger(
name: str,
log_dir: Optional[Path] = None,
log_file: Optional[str] = None,
level: int = logging.INFO,
format_str: Optional[str] = None
) -> logging.Logger:
"""
Setup logger with file and console handlers
Args:
name: Logger name
log_dir: Directory for log files
log_file: Log file name
level: Logging level
format_str: Custom format string
Returns:
Configured logger
"""
logger = logging.getLogger(name)
logger.setLevel(level)
# Prevent propagation to root logger to avoid duplication
logger.propagate = False
# Remove existing handlers
logger.handlers = []
# Default format
if format_str is None:
format_str = '%(asctime)s - %(name)s - %(levelname)s - %(message)s'
formatter = logging.Formatter(format_str)
# Console handler
console_handler = logging.StreamHandler(sys.stdout)
console_handler.setLevel(level)
console_handler.setFormatter(formatter)
logger.addHandler(console_handler)
# File handler
if log_dir and log_file:
log_dir = Path(log_dir)
log_dir.mkdir(parents=True, exist_ok=True)
file_handler = logging.FileHandler(log_dir / log_file)
file_handler.setLevel(level)
file_handler.setFormatter(formatter)
logger.addHandler(file_handler)
return logger
class ExperimentLogger:
"""Logger for ML experiments with multiple backends"""
def __init__(
self,
experiment_name: str,
project_name: str,
log_dir: Path,
config: Dict,
use_wandb: bool = True,
use_tensorboard: bool = True
):
self.experiment_name = experiment_name
self.project_name = project_name
self.log_dir = Path(log_dir)
self.config = config
# Create directories
self.log_dir.mkdir(parents=True, exist_ok=True)
# Setup file logger (private to encourage using convenience methods)
self._logger = setup_logger(
name=experiment_name,
log_dir=self.log_dir,
log_file='experiment.log'
)
# Setup experiment tracking
self.use_wandb = use_wandb and self._init_wandb()
self.use_tensorboard = use_tensorboard and self._init_tensorboard()
# Log initial config
self.log_config(config)
def _init_wandb(self) -> bool:
"""Initialize Weights & Biases"""
try:
wandb.init(
project=self.project_name,
name=self.experiment_name,
config=self.config,
dir=self.log_dir
)
self._logger.info("Initialized W&B logging")
return True
except Exception as e:
self._logger.warning(f"Failed to initialize W&B: {e}")
return False
def _init_tensorboard(self) -> bool:
"""Initialize TensorBoard"""
try:
from torch.utils.tensorboard import SummaryWriter
self.tb_writer = SummaryWriter(
log_dir=self.log_dir / 'tensorboard'
)
self._logger.info("Initialized TensorBoard logging")
return True
except Exception as e:
self._logger.warning(f"Failed to initialize TensorBoard: {e}")
return False
# Convenience methods to delegate to underlying logger
def info(self, message: str):
"""Log info message"""
self._logger.info(message)
def debug(self, message: str):
"""Log debug message"""
self._logger.debug(message)
def warning(self, message: str):
"""Log warning message"""
self._logger.warning(message)
def error(self, message: str):
"""Log error message"""
self._logger.error(message)
def critical(self, message: str):
"""Log critical message"""
self._logger.critical(message)
@property
def logger(self):
"""Backward compatibility - access to internal logger (deprecated)"""
return self._logger
def log_config(self, config: Dict):
"""Log configuration"""
# Save to file
config_path = self.log_dir / 'config.json'
with open(config_path, 'w') as f:
json.dump(config, f, indent=2)
self._logger.info(f"Configuration saved to {config_path}")
def log_metrics(self, metrics: Dict[str, float], step: int, prefix: str = ''):
"""Log metrics to all backends"""
# Add prefix if specified
if prefix:
metrics = {f"{prefix}/{k}": v for k, v in metrics.items()}
# Log to console
metrics_str = ', '.join([f"{k}: {v:.4f}" for k, v in metrics.items()])
self._logger.info(f"Step {step} - {metrics_str}")
# Log to W&B
if self.use_wandb:
wandb.log(metrics, step=step)
# Log to TensorBoard
if self.use_tensorboard:
for key, value in metrics.items():
self.tb_writer.add_scalar(key, value, step)
def log_model(self, model_path: Path, aliases: Optional[list] = None):
"""Log model artifact"""
if self.use_wandb:
artifact = wandb.Artifact(
name=f"{self.experiment_name}_model",
type='model'
)
artifact.add_file(str(model_path))
wandb.log_artifact(artifact, aliases=aliases or [])
def log_image(self, tag: str, image: Any, step: int):
"""Log image to tensorboard"""
if self.use_tensorboard:
self.tb_writer.add_image(tag, image, step)
def log_text(self, tag: str, text: str, step: int):
"""Log text"""
if self.use_tensorboard:
self.tb_writer.add_text(tag, text, step)
if self.use_wandb:
wandb.log({tag: wandb.Html(text)}, step=step)
def finish(self):
"""Cleanup logging"""
if self.use_wandb:
wandb.finish()
if self.use_tensorboard:
self.tb_writer.close()
self._logger.info("Experiment logging finished")
class MetricsLogger:
"""Simple metrics logger to JSON file"""
def __init__(self, log_path: Path):
self.log_path = Path(log_path)
self.metrics = []
def log(self, epoch: int, metrics: Dict[str, float]):
"""Log metrics for an epoch"""
entry = {
'epoch': epoch,
'timestamp': datetime.now().isoformat(),
**metrics
}
self.metrics.append(entry)
# Save to file
with open(self.log_path, 'w') as f:
json.dump(self.metrics, f, indent=2)
def load(self) -> list:
"""Load metrics from file"""
if self.log_path.exists():
with open(self.log_path, 'r') as f:
self.metrics = json.load(f)
return self.metrics |