search

LeoGrin / tabular-benchmark

448 stars 59 forks source link

Paper performance missing #10

Closed abvesa closed 1 year ago

abvesa commented 1 year ago

Hello LeoGrin,

First thanks for the very good work. I was trying to reproduce paper's result using the data you provided in ./analyses/results/, and the R scripts you provided in ./analyses.

however, I found that some of the performance are missing: (1) SAINT performance of "medium-sized categorical regression" datasets (2) MLP perfomance of all datasets

I'm wondering if you can provide the missing performances for the full reproduction of paper's experimental results, thank you.

LeoGrin commented 1 year ago

The missing part is now available, you can run curl -L -o analyses/results/benchmark_total.csv https://figshare.com/ndownloader/files/40081681 to download the random search results.