Commits · a49a2c2804b36cbe336c89426af83dbac1dbdb4c · mirrored_repos / MachineLearning / meta-llama / Llama Recipes

This project is mirrored from https://github.com/meta-llama/llama-recipes. Pull mirroring failed 2 months ago.
Repository mirroring has been paused due to too many failed attempts. It can be resumed by a project maintainer or owner.
Last successful update 2 months ago.

Jul 28, 2023
- adding PT cuda allocation expand flag · a49a2c28
  Hamid Shojanazeri authored 1 year ago
  
  a49a2c28
Jul 27, 2023
- adding barrier to end of trainer loop · 442c1ccf
  Hamid Shojanazeri authored 1 year ago
  
  442c1ccf
- printing scores based on fsdp usage or single gpu · f74d57dc
  Hamid Shojanazeri authored 1 year ago
  
  f74d57dc
- update with active memory and removing rank0 for eval score · 3d887ea4
  Hamid Shojanazeri authored 1 year ago
  
  3d887ea4
Jul 26, 2023
- fixing the full state path in checkpoint handler · bedb96b7
  Hamid Shojanazeri authored 1 year ago
  
  bedb96b7
Jul 25, 2023
- Save memory and fix typos · 71fdc492
  Andrew Gu authored 1 year ago
  
  71fdc492
- fixing the cuda id · a7156dfb
  Hamid Shojanazeri authored 1 year ago
  
  a7156dfb
- adding cuda:0 for non-fsdp situations · 707af7ea
  Hamid Shojanazeri authored 1 year ago
  
  707af7ea
Jul 23, 2023
- fixing identation · 6678be75
  Hamid Shojanazeri authored 1 year ago
  
  6678be75
- fixing scaler for both fsdp and non fsdp · 6a84e9e4
  Hamid Shojanazeri authored 1 year ago
  
  6a84e9e4
Jul 22, 2023
- fixing the condition for moving to cuda · 065ddaa7
  Hamid Shojanazeri authored 1 year ago
  
  065ddaa7
- modify to steping the lr scheduler each epoch · 20b061e0
  Hamid Shojanazeri authored 1 year ago
  
  20b061e0
Jul 17, 2023
- Initial commit · 4767f09e
  chauhang authored 1 year ago
  
  4767f09e