search

matsuolab / T3A

This codebase is the official implementation of Test-Time Classifier Adjustment Module for Model-Agnostic Domain Generalization (NeurIPS2021, Spotlight)
MIT License
93 stars 10 forks source link

Collecting results for T3A after adaptation #1

Open korawat-tanwisuth opened 2 years ago

korawat-tanwisuth commented 2 years ago

Thanks for your work. I am wondering what command I should run to get the results of T3A in a text format similar to how it's done in domainbed codebase. In the domainbed code, I can run the following command to obtain the results.

python -m domainbed.scripts.collect_results\
       --input_dir=/my/sweep/output/path

Here is the result after running T3A. Since the result heavily depends on the parameter filter_k.

image

I am wondering if there is a script that automatically picks out the numbers to report. Thanks in advance.

hxynjk commented 2 years ago

I also have the same question. Did you fix it?