Update README.md

README.md

@@ -127,10 +127,10 @@ marqo-gs-dataset/
 ├── marqo_gs_wfash_1m/
 ```
 For each dataset such as marqo_gs_full_10m, there are 4 splits as discussed before. 
-- query_0_product_id_0 represents in-domain set,
-- query_1_product_id_0 represents novel query set,
-- query_0_product_id_1 represents novel document set,
-- query_1_product_id_1 represents zero shot set,
+- `query_0_product_id_0` represents in-domain set,
+- `query_1_product_id_0` represents novel query set,
+- `query_0_product_id_1` represents novel document set,
+- `query_1_product_id_1` represents zero shot set,
 For each split, there is a ground truth csv containing triplet information, 
 a set of validation ground truth and a set of test ground truth.
 
@@ -250,7 +250,7 @@ with torch.no_grad(), torch.cuda.amp.autocast():
 print("Label probs:", text_probs)
 ```
 ### Quick Demo with Hugging Face for E5 models. 
-Here is a quick example to load our finetuned e5 text models from hugging face directly. 
+Here is a quick example to load our finetuned e5 text models from Hugging Face directly. 
 ```python
 import torch.nn.functional as F