Track and visualize logs effortlessly
  • Easily log and compare params/metrics
  • Compare multiple runs’ NER/dep-trees on the same dashboard
  • Bult-in support for displaCy visualizations
Check out the  aim-spaCy code  and demo
Easy to useopen source
Just add the following to your spaCy training config file

[training.logger]
@loggers = "spacy.AimLogger.v1"
repo = "path/to/your/save/directory"
experiment_name = "name_of_your_experiment"

When, where and how

Use aim-spaCy to detect when, where and how your models have done well or failed. In a couple of clicks!
8 major features to speed up your work

1.

Powerful pythonic search to select the runs for analysis and comparison.

Live demo this feature

2.

Automatically track displaCy visualizations and compare them at scale.

Live demo this feature

3.

Group images by hyperparameters to visualize them in the needed order

Live demo this feature

4.

Group images by step or epoch to see the evolution

Live demo this feature

5.

Group metrics by hyperparameters to analyze hyperparameters’ influence on run performance

Live demo this feature

6.

Select multiple metrics and analyze them side by side

Live demo this feature

7.

Aggregate metrics by std.dev, std.err, conf.interval, to observe average, min and max values of metrics you track

Live demo this feature

8.

Deep dive into details of each run for easy debugging

Live demo this feature
We are building more spaCy features.
If you are interested, sign up to access the project roadmap.
Oops! Something went wrong while submitting the form.