Skip to content

Latest commit

 

History

History
7 lines (6 loc) · 402 Bytes

courses.md

File metadata and controls

7 lines (6 loc) · 402 Bytes

courses

ese 5460 hw 3

  • understanding the mathamatical workings of dropout
  • understand that when input matrix is short & wide (more params than datapoints) the solution is non-unqiue. whereas when we have more data than parameters the solution is unique since the params are hihgly constrained.
  • creating a model that predicts next character and phrase through RNN and then transformer architecture