-
As a developer, I initially ended up writing my own godbus code because this library pulled too many dependencies.
However, I noticed that quite a few of the deps here don't seem to be used at all.…
-
Adapt `mit-d3m-ta2` to the implementation of https://github.com/HDI-Project/BTB/issues/145
-
Sloth currently outputs the row index (d3m index?) along with the label (column names "0"). This should be updated to output the input columns and have the label added at the end with the column name …
-
The MultiTableFeaturization primitive crashes on the uu2_gp_hyperparameter_estimation dataset when using cross validation with these pipelines: https://gist.github.com/csala/c7408ebedba705e81b1ee24be1…
csala updated
4 years ago
-
The default strategy for the text encoder should be SVC to ensure consistent behavior with older versions of the primitive. The sparseness of the created TFIDF matrix leads to a very different model …
-
The timeseries clustering pipeline from Distil generates the following error in the TA2 when run:
```
INFO - 2019-12-10 18:04:25,602 - v-master-7728ff7 - distil.TaskManager - Initialized TaskM…
-
```
Traceback (most recent call last):
File "./test-primitives.py", line 85, in
instance = primitive(hyperparams=primitive.metadata.get_hyperparams().defaults())
File "/src/distil-primiti…
-
not sure about this myself @hhelm10
-
It would be useful to know/document the baseline pipelines that we are being compared against on the [leaderboard](https://leaderboarddecember2019.datadrivendiscovery.org/).
-
current working: load_graphs, lcc, ase, gclass, gclust