search

brain-research / realistic-ssl-evaluation

Open source release of the evaluation benchmark suite described in "Realistic Evaluation of Deep Semi-Supervised Learning Algorithms"
Apache License 2.0
458 stars 98 forks source link

fix bugs in evaluate_model.py #5

Closed DoctorKey closed 5 years ago

DoctorKey commented 5 years ago

Thank you for your code! I find that it only use half the dataset to test model. You can print total_seen in evaluate_model.py to see this bug.

avital commented 5 years ago

Thank you @DoctorKey, you seem to be right. I assume this won't significantly change the results since we were using an unbiased sample from the true dataset, but it may reduce the variance of some of the runs.