tangledgroup
/

tangled-alpha-0.1-core

Text Generation

Model card Files Files and versions Community

mtasic85 commited on Feb 24

Commit

fcc1668

·

1 Parent(s): 578a7be

class_path: torch.optim.AdamW

Files changed (1) hide show

scripts/pretrain-core-model.yaml +2 -2

scripts/pretrain-core-model.yaml CHANGED Viewed

@@ -115,10 +115,10 @@ eval:
 # Optimizer-related arguments
 optimizer:
-  # class_path: torch.optim.AdamW
   # class_path: torchao.prototype.low_bit_optim.AdamW8bit
   # class_path: torchao.prototype.low_bit_optim.AdamW4bit
-  class_path: bitsandbytes.optim.AdamW8bit
   # class_path: bitsandbytes.optim.PagedAdamW8bit
   init_args:
     # (type: float, default: 0.001)

 # Optimizer-related arguments
 optimizer:
+  class_path: torch.optim.AdamW
   # class_path: torchao.prototype.low_bit_optim.AdamW8bit
   # class_path: torchao.prototype.low_bit_optim.AdamW4bit
+  # class_path: bitsandbytes.optim.AdamW8bit
   # class_path: bitsandbytes.optim.PagedAdamW8bit
   init_args:
     # (type: float, default: 0.001)