PIQuIL/RydbergGPT

Issues

• Add multi gpu training support via torchdata

  #15 opened a year ago by davidfitzek
  0

• Integration Test for Model Training and Energy Estimation

  #17 opened a year ago by davidfitzek
  0

• Replace Custom Transformer with PyTorch 2.0 Inbuilt Transformer

  #16 opened a year ago by davidfitzek
  2

• Training with precision 16-mixed model (EncoderDecoder) outputted nan

  #8 opened a year ago by yhteoh
  0

• Fix package test

  #3 opened a year ago by yhteoh
  0

• Outsource dataset into a separate repository

  #14 opened a year ago by davidfitzek
  0

• Problems with the Rydberg energy estimate

  #13 opened a year ago by yhteoh
  0

• Enhancement Suggestion: Incorporate Fast Transformer Decoding and Flash Attention into Transformer Architecture

  #12 opened a year ago by davidfitzek
  0

• Model is significantly memory intensive

  #7 opened 2 years ago by yhteoh
  1

• Major incompatibility with V_0 and blockade_radius

  #9 opened 2 years ago by yhteoh
  0

• Create API documentation for RydbergGPT

  #10 opened 2 years ago by yhteoh
  1

• Training on dataset of varying system sizes

  #2 opened 2 years ago by yhteoh
  3

• Bug in the code when training from previous checkpoint

  #5 opened 2 years ago by yhteoh
  1

• Sampling is slow due to repeated calculations

  #4 opened 2 years ago by yhteoh
  0

• DecoderLayer does not exhibit causal structure

  #1 opened 2 years ago by yhteoh
  0