why this implementation of pseudo-labeling

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

460 stars 98 forks source link

why this implementation of pseudo-labeling #35

Open hzhz2020 opened 2 years ago

hzhz2020 commented 2 years ago

Hi I'm wondering in your implementation of pseudo-labeling, why did you use non-zero loss for unlabeled sample that has maximum predicted probability below the threshold.