1
May
2020
/
Carey Phelps, Product Lead

May Updates

Project Showcase

Get a behind-the-scenes look at the process of model training and experimentation. Here are some of our latest favorites from the Weights & Biases community.


The Science of Debugging: a look at how the team at Latent Space uses interactive reports to debug their models


Deep Q Networks: reinforcement learning using DQN with the OpenAI Gym


Drought Watch Progress: a collaborative community project to detect drought from satellite images


EfficientNets and TensorFlow Hub: Transfer learning for image classification


SimpleTransformer for NLP: Language modeling, named entity recognition, and question answering using distillbert from the SimpleTransformer library.


W&B and Skorch: The union of Scikit and PyTorch— learn how to run a fast hyperparameter search to explore what’s possible and find the best model.


Product Updates

Learn about the latest features in our detailed updates →

  1. Bounding boxes: Log and dynamically visualize bounding box overlays on images
  2. Filter from the parallel coordinates chart: Select a range and click the Filter icon to zoom in
  3. New Visualizations: Confusion matrices, heatmaps, ROC and PR curves
  4. Code comparison: Diff the code across your runs
  5. Jupyter session history: Look back at the cells that were executed during a run

Machine Learning Community

Looking for a place to discuss machine learning topics with other practitioners? We have a growing community of researchers and engineers in the W&B Forum →

Join our virtual meetup to share project ideas and get inspired. Here are videos from our most recent event.

  1. Track your machine learning experiments locally with Chris Van Pelt
  2. Made with ML— exploring project ideas with Goku Mohandas
  3. Anti-patterns in open source ML research code with Jariullah Safi
  4. Attention for time series forecasting and COVID predictions with Isaac Godfried
  5. Surprising utility of surprise— why ML uses negative log probabilities with Charles Frye
Join our mailing list to get the latest machine learning updates.