Bag of Words Newsletter: GPT-3 Webinar & Kaggle Tips Galore
The May the 4th (be with you) edition of our biweekly newsletter
Created on May 4|Last edited on May 4
Comment
In case you didn't know: every two weeks, we send out a newsletter called "Bag of Words." It covers our upcoming events, the best posts from this here blog, and highlights from our community members. We think it's pretty good.
Now, if you'd like to subscribe, you should see a button on the Fully Connected homepage. It looks like this:
If you'd like to see what kind of stuff we send out before clicking that button, we've got you covered below:
EVENTS
GPT-3 is a powerful NLP model from the folks at OpenAI, especially notable for its ability to generate high-quality text. Join Ted Sanders from OpenAI and Boris Dayma from W&B for a live demo where we’ll look at what GPT-3 can do (text classification, question answering, text generation, and more) plus how you can fine-tune the model with your own dataset with the help of Weights & Biases. We kick off tomorrow morning at 10 a.m. PT.
COMMUNITY HIGHLIGHTS
WATCH: sentdex consistently puts out some of the best Python and ML content on YouTube. We were thrilled he took W&B for a spin.
COMPETE: This Kaggle notebook by Tanya Dayanand showcases the power of Weights & Biases in visualizing datasets.
COMPETE PT. 2: Why should you use experiment tracking tools for? Andrada Olteanu answers this question with an in-depth Kaggle discussion post.
WATCH PT 2: We enjoyed this great hands-on introductory video from Debugging with Sayan on using W&B with Keras and wanted to share it with you.
W&B INTERNATIONAL: Finally, Kimin Park has written a lovely blog post on the advantages of Weights & Biases as an MLOps tool. (NB: this is in Korean!)
BLOG POSTS
EXPERIMENT: We loved this evaluation of transformer variants by Boris Dayma. It's an in-depth look at the tradeoffs between techniques with copious visualizations and analysis and comes highly recommended.
LISTEN: Mircea Neagovici, VP of AI and Research art UiPath, joined Gradient Dissent this week to talk Robotic Process Automation and ML with our Co-Founder Lukas Biewald.
COMPETE PT 3: Sairam Sundaresan is an avid Kaggler who ran an experiment on himself. Namely: how well could he do on a Kaggle competition spending just one hour a day?
COMPARE: Our Thomas Capelle was inspired by this recent Sebastian Rashka post to dig into loss functions.
How Well Can You Kaggle with Just One Hour a Day?
A whimsical challenge to see how far one can go in a Kaggle competition with limited time, limited resources and pretty much limited everything else.
Classification Loss Functions: Comparing SoftMax, Cross Entropy, and More
Sometimes, when training a classifier, we can get confused about the last layer to put on our neural networks. This article helps you understand how to do it right.
An Evaluation of Transformer Variants
Training of different transformer variants for text-to-image generation with DALL-E-mini
Mircea Neagovici — Robotic Process Automation (RPA) and ML
Mircea explains how machine learning unlocks the next level of potential for robotic process automation (RPA) and how ML teams differ from engineering teams.
***
And that's about that. If you have something you'd like us to cover or if you're a member of our community who'd like to chat about us featuring your work, leave a comment below and we'll take a look. Thanks for reading!
Add a comment
Tags: Articles
Iterate on AI agents and models faster. Try Weights & Biases today.