API Reference

Complete API documentation for Tracelet's public interfaces.

Quick Navigation

Core API

• Main Interface - Primary entry points (start_logging, get_active_experiment, stop_logging)
• Experiment - Main experiment management class
• Settings - Configuration and settings management

Backends

• MLflow - MLflow integration
• ClearML - ClearML integration
• Weights & Biases - W&B integration
• AIM - AIM integration

Frameworks

• PyTorch - PyTorch and TensorBoard integration
• Lightning - PyTorch Lightning integration

Data Collection

• Git Collector - Git repository information
• System Metrics - System performance metrics

Plugin System

• Core Plugins - Plugin architecture and interfaces
• Data Flow Management - Orchestrator and data flow

Usage Examples

Basic Usage

import tracelet

# Start experiment tracking
tracelet.start_logging(
    exp_name="my_experiment",
    project="my_project",
    backend="mlflow"
)

# Get active experiment and log metrics
exp = tracelet.get_active_experiment()
exp.log_metric("accuracy", 0.95, iteration=100)
exp.log_params({"lr": 0.001, "batch_size": 32})

# Stop tracking
tracelet.stop_logging()

Multi-Backend Usage

import tracelet

# Track to multiple backends simultaneously
tracelet.start_logging(
    exp_name="multi_backend_experiment",
    project="comparison_study",
    backend=["mlflow", "wandb"]  # List of backends
)

Advanced Configuration

from tracelet.settings import TraceletSettings

settings = TraceletSettings(
    project="advanced_project",
    backend=["mlflow"],
    track_system=True,
    metrics_interval=5.0
)

tracelet.start_logging(
    exp_name="advanced_experiment",
    settings=settings
)