Share this page!

Enter URL or ID to Unroll

You can paste full URL like: https://x.com/threadreaderapp/status/1644127596119195649
or just the ID like: 1644127596119195649

How to get URL link on X (Twitter) App

  1. On the Twitter thread, click on or icon on the bottom
  2. Click again on or Share Via icon
  3. Click on Copy Link to Tweet
  4. Paste it above and click "Unroll Thread"!
  5. More info at Twitter Help
Weights & Biases Profile picture
Weights & Biases
@weights_biases
The AI developer platform.
Feb 10, 2022 • 5 tweets • 4 min read
💬 The world of GPT-3

In this episode of Gradient Dissent, @npew, @borisdayma, and @l2k talk about:

✏️ Applying GPT-3 to translation and copywriting
💪 The performance benefits of fine-tuning GPT-3
💻 Developing the @OpenAI API

(Bonus topic: The story behind the @OpenAI and @weights_biases collaboration 🪄🐝 )

Dec 4, 2021 • 5 tweets • 3 min read
🏆 🥳🕺🏽 We're thrilled to announce the first winners of #27DaysOfJAX

🏅 @rareblog, for running JAX on a Jetson Nano & writing about Spiking NNs in JAX

🏅 @imPGulati, for writing his first blog ever & contributing

🏅 @dionhaefner, for 🤯 🤯 🤯 the great read!

Blogs in the🧵👇
Sep 10, 2021 • 11 tweets • 5 min read
We're grateful that deeply understanding and improving datasets is acknowledged as a non-negotiable part of training pipelines 🙏

However exploring datasets can be cumbersome, and documenting and sharing findings is often messy

W&B Tables fixes this 👇

W&B Tables enables quick and powerful exploration of image, video, audio, tabular, molecule and NLP datasets.

@metaphdor used Tables to explore 100k rows from @Reddit's Go Emotions dataset:

📺: )

🛑 First, filtering for multiple column values
Sep 9, 2021 • 5 tweets • 3 min read
New podcast episode! 📢

@l2k and @emilymbender dive into the problems with bigger and bigger language models, the difference between form and meaning, the limits of benchmarks, and the #BenderRule.

🎥:

They discuss 4 of Emily's papers ⬇️

1/5 "On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? 🦜" (Bender, Gebru et al. 2021)

Possible risks associated with bigger and bigger language models, and ways to mitigate those risks.

dl.acm.org/doi/pdf/10.114…

2/5
Nov 5, 2020 • 7 tweets • 4 min read
We're thrilled to announce that YOLOv5 now comes with @weights_biases baked in!

With no additional lines of code, you now get automatic bounding box debugging, GPU usage and performance metrics, reproducible models, & more!

👩‍🚀 Try it → colab.research.google.com/github/ultraly…

#deeplearning What does YOLOv5 + W&B give you?

1. You can monitor how your models and hyperparameters are performing, including automatically tracking:

- Training and validation losses
- Precision, Recall, mAP@0.5, mAP@0.5:0.95
- Learning Rate over time