Text Generation
Transformers
Safetensors
English
stablelm
conversational
Inference Endpoints
euclaise commited on
Commit
b57c984
1 Parent(s): 02856a3

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +2 -0
README.md CHANGED
@@ -61,7 +61,9 @@ ReMask applies an approach similar to R-TeaFor to typical chat/instruction tunin
61
  Consider the following chat interaction:
62
 
63
  > User: What is 1+1?
 
64
  > Assistant: **1+1=2**
 
65
  > **User:**
66
 
67
  The model must predict the bolded parts. So, we randomly mask tokens from the bolded parts, and run the model once on the masked sequence and once on the full sequence.
 
61
  Consider the following chat interaction:
62
 
63
  > User: What is 1+1?
64
+ >
65
  > Assistant: **1+1=2**
66
+ >
67
  > **User:**
68
 
69
  The model must predict the bolded parts. So, we randomly mask tokens from the bolded parts, and run the model once on the masked sequence and once on the full sequence.