weights-and-biases

Log ML experiments, compare runs, and automate hyperparameter sweeps with Weights & Biases.

4|Updated May 18, 2026
One-click install
npx skills add https://github.com/ZardLi1115/zedclaw --skill weights-and-biases-zardli1115
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: weights-and-biases
Source: https://github.com/ZardLi1115/zedclaw/tree/main/skills/mlops/evaluation/weights-and-biases
Command: npx skills add https://github.com/ZardLi1115/zedclaw --skill weights-and-biases-zardli1115

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires wandb, and includes references (resource) components.

What problem does it solve?

Weights & Biases (W&B) solves the problem of tracking ML experiments so you can reliably compare runs, visualize training progress, and manage datasets/models across time.

Core Features & Use Cases

  • Experiment tracking & dashboards: Log losses/metrics in real time and keep run history for later review.
  • Hyperparameter sweeps: Automate search over learning rates, batch sizes, optimizers, and more using sweep strategies.
  • Artifacts & model registry: Version datasets and models with lineage, aliases (latest/best/production), and reproducible handoffs.

Use case example: You run a CNN training job with different hyperparameters and want to compare validation accuracy across runs, then promote the best checkpoint to a model registry alias for deployment.

Quick Start

Run the W&B integration for your training script by installing wandb, logging in, and calling wandb.init with a project name and a config that includes your hyperparameters.

Frequently Asked Questions about weights-and-biases

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I track ML experiments and compare validation metrics across different runs?

Track ML experiments by initializing wandb with a project name, logging hyperparameters in the config, and calling wandb.log to report metrics like validation accuracy for real-time run comparisons.

How do I automate hyperparameter sweeps for a CNN training job?

Automate hyperparameter sweeps by configuring search strategies over learning rates and batch sizes in wandb, allowing the sweep agent to run training jobs and optimize validation metrics automatically.

How does model registry artifact versioning work for deployment handoffs?

Model registry artifact versioning works by logging datasets and models as wandb artifacts with tracked lineage, assigning aliases like latest or production, and ensuring reproducible deployment handoffs.

Do I need to install wandb to log training metrics and model checkpoints?

Yes, you need to install the wandb dependency, authenticate your account, and call wandb.init in your training script to start logging metrics, artifacts, and model checkpoints to your project dashboard.

Can I use this for MLOps workflows in both research and production environments?

Yes, you can use this for MLOps workflows in research and production teams to maintain experiment tracking history, execute sweep-based optimization, and manage artifact versioning across training pipelines.

What is the best way to log losses in real time and keep run history for later review?

The best way to log losses in real time is to integrate wandb into your training pipeline, reporting metrics via wandb.log to automatically build dashboards and retain run history for later review.