skills.rest
Chia1104
Chia1104COMMUNITY
Explore all skills by this author →
Website𝕏@chia7423GitHub

weights-and-biases

Log ML experiments, hyperparameter sweeps, and model artifacts to Weights & Biases dashboards.

1|Updated Aug 11, 2026
Search Tags:#mlops#machine learning#PyTorch#experiment tracking#wandb#hyperparameter sweep#model registry
One-click install
npx skills add https://github.com/Chia1104/agent-air --skill weights-and-biases-chia1104
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: weights-and-biases
Source: https://github.com/Chia1104/agent-air/tree/main/skills/hermes/mlops/evaluation/weights-and-biases
Command: npx skills add https://github.com/Chia1104/agent-air --skill weights-and-biases-chia1104

SYSTEM DOCUMENTATION & REQUIREMENTS

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

What problem does it solve? Tracking machine learning experiments manually across notebooks and scripts leads to lost configurations, uncomparable runs, and no reproducibility. This Skill provides complete guidance for instrumenting training code with Weights & Biases so every metric, hyperparameter, and model checkpoint is logged, versioned, and comparable. ## Core Features & Use Cases - Experiment Tracking: Initialize runs with wandb.init, log metrics, media, histograms, and tables, and compare runs in real-time dashboards. - Hyperparameter Sweeps: Automate optimization with grid, random, and Bayesian search strategies, including hyperband early termination and parallel multi-GPU agents. - Artifacts & Model Registry: Version datasets and models with lineage tracking, aliases for deployment stages, and a central registry for governance. - Framework Integrations: Ready-made patterns for PyTorch, PyTorch Lightning, HuggingFace Transformers, Keras/TensorFlow, Fast.ai, XGBoost, and LightGBM. - Use Case: A team fine-tuning a BERT classifier runs a 50-trial Bayesian sweep over learning rate and batch size, then promotes the best model artifact to a production registry alias. ## Quick Start Ask the agent to instrument your PyTorch training script with W&B experiment tracking, logging loss and accuracy per epoch and saving the final model as a versioned artifact.

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 with Weights & Biases?▼

Call wandb.init with your project name and config dictionary, then use wandb.log inside your training loop to record metrics like loss and accuracy. Finish with wandb.finish, and view results in the W&B dashboard.

How to run a hyperparameter sweep with wandb?▼

Define a sweep_config with a search method (grid, random, or bayes), a target metric, and parameter distributions. Create the sweep with wandb.sweep, then launch trials using wandb.agent with your training function and a run count.

Does W&B integrate with HuggingFace Transformers and PyTorch Lightning?▼

Yes. For Transformers, set report_to="wandb" in TrainingArguments and the Trainer logs automatically. For Lightning, pass a WandbLogger to the Trainer; it can also log model checkpoints and code as artifacts.

What is the difference between W&B Artifacts and the Model Registry?▼

Artifacts are versioned files such as datasets and models with automatic lineage tracking between runs. The Model Registry is a central governance layer where model artifacts are linked and promoted through aliases like staging and production.

Can I use wandb without an internet connection?▼

Yes. Set the WANDB_MODE environment variable to "offline" before initializing your run, and all data is stored locally. Later, run wandb sync on the run directory to upload results to the server.

Why is my Bayesian sweep not finding better hyperparameters?▼

Bayesian search needs enough initial trials to model the parameter space, so run at least 15-20 agents before judging results. Also verify your metric name matches what you log and that parameter ranges use log scales for learning rates.

Related Skills

Explore More Data & Analytics Skills→
CHENyiru3CHENyiru3

datasets

Load and preprocess datasets from Hugging Face Hub or local sources.

Community
Advanced
KentoShimizuKentoShimizu
7

mlops-monitoring-drift

Detect data drift, concept drift, and quality degradation in production ML pipelines.

Community
Advanced
nishide-devnishide-dev

ml-profile

Profile PyTorch Lightning training to diagnose data loading, computation, and GPU memory bottlenecks.

Community
Intermediate
Jensen-YaoJensen-Yao
2

experiment-tracking-swanlab

Track ML experiments with metrics, configs, and media using SwanLab.

Community
Intermediate