Data

For your convenience, you can download from this page datasets that were released as part of past AutoML challenges. This does NOT include labels on the validation at test sets. The websites of the challenges remain open if you want to obtain you performances on validation or test data.

DATASETS AVAILABLE FOR DOWNLOAD:

Round 0 data:

Set 1: adult.zip -- 1.1MB

Set 2: cadata.zip -- 0.7 MB

Set 3: digits.zip -- 117.7 MB

Set 4: dorothea.zip -- 4.7 MB

Set 5: newsgroups.zip -- 6.4 MB

We released the validation data truth values at the end of round 0 for practice purposes: phase0_valid.zip

Round 1 data:

Set 1: christine.zip -- 19 MB

Set 2: jasmine.zip -- 224 KB

Set 3: philippine.zip - 12 MB

Set 4: madeline.zip -- 2.3 MB

Set 5: sylvine.zip -- 663 KB

Round 2 data:

Set 1: albert.zip -- 55 MB

Set 2: dilbert.zip -- 168 MB

Set 3: fabert.zip - 572 KB

Set 4: robert.zip -- 192 MB

Set 5: volkert.zip -- 19 MB

Round 3 data:

Set 1: alexis.zip -- 17.4 MB

Set 2: dionis.zip -- 53.6 MB

Set 3: grigoris.zip - 86.8 MB

Set 4: jannis.zip -- 20.6 MB

Set 5: wallis.zip -- 4.1 MB

Round 4 data:

Set 1: Evita -- 13.4 MB

Set 2: Flora -- 104 MB

Set 3: Helena — 9.9 MB

Set 4: Tania -- 62.8 MB

Set 5: Yolanda -- 175.2 MB

AutoML2 - 2018

(we are releasing public data sets only)

Feedback phase

Set 1: ada.zip -- 0.6MB

Set 2: arcene.zip -- 8.5 MB

Set 3: gina.zip -- 19.1 MB

Set 4: guillermo.zip -- 243.0 MB

Set 5: RL.zip -- 2.4 MB

Final phase

Set 1: PM.zip

Set 2: RH.zip

Set 3: RI.zip

Set 4: riccardo.zip -- 202.0 MB

Set 5: RM.zip