Update README.md

README.md

@@ -12,7 +12,7 @@ tags:
 This model checkpoint is based on [sentence-transformer](https://www.SBERT.net) model, 
 and [the original checkpoint](https://huggingface.co/sentence-transformers/gtr-t5-base).
 
-This checkpoint includes the fine-tuned weights of dense linear layer, and can be initialized with T5's encoder.
+This checkpoint includes the fine-tuned weights of dense linear layer, and can be initialized with T5's encoder and a linear layer without bias called `linear'
 
 ---
 Paper: [Large Dual Encoders Are Generalizable Retrievers](https://arxiv.org/abs/2112.07899)