RegNetY is a convolutional network design space with simple, regular models with parameters: depth $d$, initial width $w_{0} > 0$, and slope $w_{a} > 0$, and generates a different block width $u_{j}$ for each block $j < d$. The key restriction for the RegNet types of model is that there is a linear parameterisation of block widths (the design space only contains models with this linear structure):
$$ u_{j} = w_{0} + w_{a}\cdot{j} $$
For RegNetX we have additional restrictions: we set $b = 1$ (the bottleneck ratio), $12 \leq d \leq 28$, and $w_{m} \geq 2$ (the width multiplier).
For RegNetY we make one change, which is to include Squeeze-and-Excitation blocks.
Source: Designing Network Design SpacesPaper | Code | Results | Date | Stars |
---|
Task | Papers | Share |
---|---|---|
Image Classification | 4 | 12.50% |
Decoder | 3 | 9.38% |
Question Answering | 2 | 6.25% |
Reinforcement Learning (RL) | 2 | 6.25% |
Survival Analysis | 2 | 6.25% |
BIG-bench Machine Learning | 2 | 6.25% |
Adversarial Attack | 1 | 3.13% |
Classification | 1 | 3.13% |
In-Context Learning | 1 | 3.13% |