Update README.md

README.md

@@ -32,7 +32,7 @@ We use the HelpSteer2 preference binarized into chosen-rejected pairs using the
 
 ## Recipes
 
-We used 4 nodes (8 x AMD MI250X) to obtain a global batch size of 128 for SFT and 64 for DPO. We used the [Alignment Handbook](https://github.com/huggingface/alignment-handbook/) codebase for finetuning.
+For finetuning, we used 4 nodes (8 x AMD MI250X) to obtain a global batch size of 128 for SFT and 64 for DPO. We used the [Alignment Handbook](https://github.com/huggingface/alignment-handbook/) codebase.
 
 **SFT**