Update README.md

README.md

@@ -9,6 +9,10 @@ landscapes as explained in [this post](https://huggingface.co/blog/AmelieSchreib
 protein pairs, which can be [found here](https://huggingface.co/datasets/AmelieSchreiber/pha_clustered_protein_complexes). 
 This is a very new method for clustering protein-protein complexes. 
 
+Using the MLM loss to predict pairs of interacting proteins was inspired by [this paper](https://arxiv.org/abs/2308.07136). However, 
+the authors do not finetune the models for this task. Thus we reasoned that improved performance on this method could be achieved 
+by finetuning the model on pairs of interacting proteins. 
+
 ## Using the Model
 To use the model, we follow [this blog post](https://huggingface.co/blog/AmelieSchreiber/protein-binding-partners-with-esm2). 
 Below we see how to use the model for ranking potential binders for a target protein of interest. The lower the MLM loss average,