Built with W&B Report Guidelines
Thank you for your interest in contributing to Fully Connected! We can't wait to tell your #BuiltwithW&B story.
Created on February 15|Last edited on February 17
Comment
We love to highlight great work being done using W&B, and are so excited to work with you to help you show off your project to our community of other machine learning practitioners. Below are the steps you can take to tell this story about your project.
Creating a report
From your profile or project page, we ask that you create a report.
Creating a report from your profile page (default view if you have never created a report)
Creating a report from your project page
This will serve as the foundation of your Built with W&B Fully Connected (FC) post. There shouldn't be much of a difference between your report and the published FC post.
Before you get started
- Test out the / for commands
- Examples of published pieces in Fully Connected can be found here:
MLP-Mixer: An all-MLP Architecture for Vision
In the past few months there have been various papers proposing MLP based architectures without Attention or Convolutions. This report analyses the paper 'MLP-Mixer: An all-MLP Architecture for Vision' by Ilya Tolstikhin, Neil Houlsby, Alexander Kolesnikov, Lucas Beyer and others.
New Techniques for Generating Images With Text
In this article, we'll look at Image Generation with CLOOB Conditioned Latent Denoising Diffusion Generative Adversarial Networks (GANs), or CCLDDG, for short.
Solving Wordle with Reinforcement Learning
In this article, we explore how to use Deep Reinforcement Learning (RL) to teach a bot to play Wordle, the word-guessing game now owned by The New York Times.
Title
What is the title of your project? Make sure you consider a title that stands out to the reader with a brief descriptive summary of your project.
Introduction (3-5 Sentences)
What is interesting about your project and what will your readers learn about it? If you can tease the conclusion or what you learned here (utilizing W&B), it helps drive a reader to the end.
If there is a highly compelling visual to draw in your reader please include it here. For example, if you are writing about a diffusion model, show some diffusion images!
Body of your Report (6-7 sentences per paragraph)
This is where you can dive deep into your project, and how W&B was utilized in it. Please break the body into sections and keep your paragraphs short. If it's a step-by-step process, start at the beginning and walk us through it.
Provide details about the data you’re using, the model(s) you’re training, the techniques you’re employing, etc. We recommend showing these Weights & Biases staples, if they apply, for your post:
- visualizations
- plots
- tables
- code samples
- charts
These are all easily embeddable and will keep your reader's attention.
Instead of code samples, if you can include a Colab, for example, even better! Explain your charts before including them. For example:
Below, you'll see that learning rate is the most important parameter for the metric we're concerned with.
Overall, we are interested in seeing how you use our tool and showcasing that to like-minded practitioners.
Above all else: your piece should strive to educate and entertain. Be sure to add some personality. This isn’t an academic paper! Emojis, relevant gifs, fun GAN implementations, and even a few light-hearted jokes are encouraged!
Introduce concepts in plain language. Our readers are machine learning practitioners (for the most part) so you don't have to explain what a neural network is or what hyperparameters are. But a short sentence that explains things in less-technical terms can be really valuable!
Example:
Dropout is a fundamental ML technique where you remove (drop out) nodes in your network to increase robustness and reduce the chance of overfitting.
Your body should end with the takeaway you promised in the intro. For W&B, this is often a trained model or a takeaway from an experiment.
Conclusion (3-5 sentences)
A simple summary of what you hope your reader learned. Provide a lesson or viewpoint that the reader can take away when they’ve finished.
Next Steps
Once you are finished please reach out to me at corey.strausman@wandb.com with the link to your report. We will then review and chat about the next steps to publishing it on Fully Connected.
One last thing, thank you! Your contribution is critical to helping build our community of ML practitioners.
Add a comment