Update README.md

README.md

@@ -36,6 +36,11 @@ embeddings = ((full_embeddings * mask.unsqueeze(-1)).sum(1) / mask.sum(-1).unsqu
 There is also a [decoder model](https://huggingface.co/entropy/roberta_zinc_decoder) trained to reconstruct 
 inputs from embeddings
 
+## Compression Head
+
+There is a [compression head model](https://huggingface.co/entropy/roberta_zinc_compression_head) trained to 
+compress embeddings generated by this model from the native size of 768 to smaller sizes (512, 256, 128, 64, 32).
+
 ---
 license: mit
 ---