In this blogpost I will share the improvements Aim 3.13. We had released 2 more versions (Aim 3.11 and 3.12) that we haven’t had the chance to talk about much. Stay tuned on the AimStack twitter where we will share them and more use-cases.
Aim is evolving really quickly! 🚀
We are on a mission to democratize AI dev tools. Thanks to the awesome Aim community for the help and contributions.
Aim 3.13 is full of impactful changes. Here are the main highlights!
- Figures Explorer
Explore and compare 100s of Plotly figures within a few clicks.
- Notify about stalled runs
Configure Aim to notify your team when the run is stalled (slack, email, workspace are available)
- Stable Aim Remote Server
- KerasTuner integration
- Weights and Biases log converter
Figures Explorer is a new addition to the mighty Aim explorers. With this feature now you can compare 1000s of Plotly figures within a few clicks.
This is just the first iteration 😊. Expect lots of improvements of the experience over time. This is how the figures explorer works:
One of the ways the precious training time is prolonged / wasted when for some reasons the runs stall. The training runs can stall in so many different ways:
- Hardware issues
- exception in the code
- driver issues
And many other things we never anticipate. This feature allows Aim users to configure notifications to their preferred channel (slack, workspace) when the run stalls.
Check out the docs for more info. Here is how it looks on your slack:
Aim Remote allows to set up a remote Aim tracking server so the tracked logs can be sent to a centralized location.
It’s been a while since Aim Remote Server was an experimental feature. We have been lucky as the users have started using it and shared lots of feedbacks, issues that we have fixed.
Now we are excited to announce that the Aim Remote Server is stable
A highly requested integration with KerasTuner.
Here are the docs and a short code snippet. Super-easy to integrate Aim with your KerasTuner project now.
from aim.keras_tuner import AimCallback
tuner = kt.Hyperband(
build_model, objective="val_accuracy", max_epochs=30, hyperband_iterations=2
callbacks=[AimCallback(tuner=tuner, repo='.', experiment='keras_tuner_test')],
The W&B users who also wanted to use Aim in their work, have been asking about this feature for a while now.
Excited to share that now you can easily convert your W&B logs to Aim.
Here is how it works:
cd project-directory# Init the Aim repo
aim init # Run the converter to migrate logs from WandB to Aim
aim convert wandb --entity 'my_team' --project 'my_project'