Hugging Face's logo
Diogo-V 's Collections
Naive Initialization
Naive Initialization + Linear Scale
Naive Initialization + Noisy Scale
Space Search
Space Search with Fallback
Non Linear Regression
Norm Expr Non Linear Regression
Training Parameters
Evaluate Schedulers
Global Training: Linear
Global Training: Cosine Annealing

Training Parameters

Tests for the layer-wise training of params initialized by various methods but while using a scale param initialized linearly

Upvote
-