A sleek CLI tool that transforms how you work with Weights & Biases. Create your training report in a single-line command!

WandB Logger App is a sleek command-line tool that transforms how you work with Weights & Biases. Create beautiful, shareable reports of your ML experiments in seconds.

WandB Logger offers two powerful ways to quickly access your experiments:

No more searching for run URLs or project names! WandB Logger makes finding your experiments effortless:

# Browse your recent projects
wandb-logger recent

# Browse your recent runs across all projects
wandb-logger recent-runs

This dual approach lets you quickly navigate through your work at both the project and individual run level - a huge productivity boost!

• 🔐 Simple Authentication - Quick login with WandB credentials
• 📈 Beautiful Visualizations - Fetch stunning graphs from any WandB run
• 📝 Markdown Reports - Generate elegant reports for easy sharing and embedding
• ⚡ Lightning Fast - Parallel processing and caching make everything snappy
• 🧠 Smart Metric Handling - Intelligently organizes complex metric paths
• 🎯 Interactive Selection - Pick exactly which metrics matter to you

# Install from PyPI
pip install wandb-logger

# Or install from source
git clone https://github.com/SabaPivot/wandb-logger.git
cd wandb-logger
pip install -e .

You'll need a Weights & Biases account to use this tool. If you don't already have one:

1. Sign up at wandb.ai
2. Get your API key from wandb.ai/authorize
3. Login using the tool:

wandb-logger login

This will prompt you to enter your API key and securely store it for future sessions.

# Login to your WandB account
wandb-logger login

# MAIN WORKFLOWS:

# 1. Browse your recent PROJECTS (find the project you're looking for)
wandb-logger recent

# 2. Browse your recent RUNS (find the specific run you need)
wandb-logger recent-runs

# 3. Once you have a run URL, fetch detailed metrics
wandb-logger fetch https://wandb.ai/username/project/runs/run_id

The fetch command now defaults to interactive selection mode:

1. Default Mode (interactive selection):

# Opens an interactive menu to choose specific metrics
wandb-logger fetch https://wandb.ai/username/project/runs/run_id

2. All Metrics Mode (with --all):

# Fetches and plots ALL available metrics
wandb-logger fetch https://wandb.ai/username/project/runs/run_id --all

Select exactly what you want in your reports:

Available metrics:
1. [X] train/loss
2. [X] train/accuracy
3. [ ] validation/loss
4. [ ] validation/accuracy

Command (Type number to toggle, 'a'=all, 'n'=none, 'done'=finish):

# Login first
wandb-logger login

# Fetch your latest run and create a report
wandb-logger recent --limit 1 --select

# Fetch multiple specific runs
wandb-logger fetch https://wandb.ai/username/project/runs/run_1
wandb-logger fetch https://wandb.ai/username/project/runs/run_2

# List your saved reports
wandb-logger list

For each run, you'll get:

• <run_name>.md: Clean report with selected metrics and graphs
• metadata.md: Detailed configuration and all run information
• Graph PNGs: Individual visualizations for each metric

• ⚡ Caching: Blazing fast repeated access to run data
• ⚙️ Parallel Processing: Generates multiple graphs simultaneously
• 📊 Optimized Visualizations: Beautiful graphs with minimal code
• 🧠 Smart Organization: Automatic categorization of complex metrics

• Python 3.8+
• WandB account

Like this project? Please give it a star on GitHub to show your support and help others discover it!