amoudgl
/

celo2

learned-optimizer

Model card Files Files and versions

amoudgl commited on 9 days ago

Commit

cc0e792

·

verified ·

1 Parent(s): 77c6991

Update

Files changed (1) hide show

README.md +1 -1

README.md CHANGED Viewed

@@ -17,7 +17,7 @@ tags:
 </p>
-Official pretrained weights for the **Celo2**: This variant applies (and is meta-trained with) a harness that includes Newton-Schulz orthogonalization on top of the learned update for matrix parameters and uses AdamW for biases/embeddings. For a fully-learned variant without any harness, see [celo2-base](https://huggingface.co/amoudgl/celo2-base).
 ## Quickstart

 </p>
+Official pretrained weights for **Celo2**: This variant applies (and is meta-trained with) a harness that includes Newton-Schulz orthogonalization on top of the learned update for matrix parameters and uses AdamW for biases/embeddings. For a fully-learned variant without any harness, see [celo2-base](https://huggingface.co/amoudgl/celo2-base).
 ## Quickstart