Inverse Square Root is a learning rate schedule 1 / $\sqrt{\max\left(n, k\right)}$ where $n$ is the current training iteration and $k$ is the number of warm-up steps. This sets a constant learning rate for the first $k$ steps, then exponentially decays the learning rate until pre-training is over.
Paper | Code | Results | Date | Stars |
---|
Task | Papers | Share |
---|---|---|
Language Modelling | 97 | 9.63% |
Question Answering | 65 | 6.45% |
Text Generation | 48 | 4.77% |
Sentence | 44 | 4.37% |
Translation | 32 | 3.18% |
Retrieval | 31 | 3.08% |
Machine Translation | 26 | 2.58% |
Natural Language Understanding | 22 | 2.18% |
Semantic Parsing | 19 | 1.89% |
Component | Type |
|
---|---|---|
🤖 No Components Found | You can add them if they exist; e.g. Mask R-CNN uses RoIAlign |