wandb-csv

Community

Track W&B experiment results with a CSV ledger

Authordongzhuoyao
Version1.0.0
Installs0

System Documentation

What problem does it solve?

This Skill solves the problem of W&B experiment results being tied to transient UI state and console logs that may be lost or hard to compare across runs, by maintaining a durable workspace-local CSV ledger for all experiment outcomes.

Core Features & Use Cases

  • Consistent Schema Enforcement: Uses a predefined set of required columns including experiment version tracking to prevent stale or inconsistent data entry.
  • Multi-Scenario Support: Handles logging for full runs, dryruns, and multiple hyperparameter variants, with clear rules to avoid confusing dryrun metrics with full-run results.
  • Cross-Run Comparison: Enables easy side-by-side comparison of experiment variants without relying on the W&B UI, by reading directly from the local CSV ledger. Use Case: When running a grid of 10 hyperparameter variants for a computer vision training experiment, use this Skill to log each variant's FID and BPD metrics to the CSV ledger for instant comparison and reporting.

Quick Start

Use the wandb-csv skill to log the latest training run's metrics to the project's experiment results CSV ledger.

Dependency Matrix

Required Modules

None required

Components

Standard package

💻 Claude Code Installation

Recommended: Let Claude install automatically. Simply copy and paste the text below to Claude Code.

Please help me install this Skill:
Name: wandb-csv
Download link: https://github.com/dongzhuoyao/deepresearch/archive/main.zip#wandb-csv

Please download this .zip file, extract it, and install it in the .claude/skills/ directory.
View Source Repository

Agent Skills Search Helper

Install a tiny helper to your Agent, search and equip skill from 536,000+ vetted skills library on demand.