laplax-org / laplax / 27405907705
67%
main: 67%

LAST BUILD BRANCH: fix-active-learning-docs-rendering
DEFAULT BRANCH: main
Ran
Jobs 1
Files 29
Run time 1min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
coverage: 63.682% (-3.5%) from 67.192%
27405907705

Pull #63

github

web-flow 
Extension to active learning tutorial (#65)

* Convert back to .ipynb to make changes

* Implement ground truth function and plot

* Generate and plot initial data

* Define, train and visualize model

* Try curvature estimation on classification model

* Compute posterior uncertainty

* Plot posterior uncertainty

* Implement active learning loop for classification

* Animation of active learning for classification

* Move plotting code out of notebook

* Improvements to actie learning visualization

* Improve visualizations

* Save animation to file

* Add text desciptions to bonus part

* Change legend location

* Implement passive learning for comparison

* Small changes to make notebook more concise

* Convert notebook to .py file

* Document sampling-based next datapoint selection

* Fix ruff complaint

* Reconvert to notebook to make changes

* Differentiate more clearly between measurement precision and prior precision

* Explicitly import Ipython.display()

* Add tqdm to active learning loops

* Adjusted some wording

* Adjust default argument of sample variance to match default slider value

* Remove unnecessary cast to float

* Adapted some wording

* Adapted some wording

* Added a summary to first tutorial

* Rename DataLoader.__len__ to avoid confusion between number of abtches and elements

* Remove no-sampling-zone argument from information_gain_about_points

* Keep optimizer state throughout active learning instead of reinstantiating it

* Fix docstring with correct return value

* Add note about calibrating on training set

* Calculate RMSE comparison between active and passive models for interesting point(s)

* Extract bonus part into its own notebook

* Fix typos

* Fix showing animations

* Fix arguments to SymLogNorm

* Fix title of notebook

* Match number of epochs between passively and actively trained model

* Improve usage of random keys

* Compute accuracy of passive and active model

* Adapt wording to classification examp... (continued)
Pull Request #63: Active learning tutorial

1024 of 1608 relevant lines covered (63.68%)

0.64 hits per line

Coverage Regressions

Lines Coverage ∆ File
14
81.08
 -15.35% laplax/curv/ggn.py
14
72.22
 -3.49% laplax/curv/utils.py
9
83.33
 -0.42% laplax/util/tree.py
Jobs
ID Job ID Ran Files Coverage
1 27405907705.1 29
63.68
 GitHub Action Run
Source Files on build 27405907705