Update README.md

Author: kunal-savvy

README.md

@@ -115,7 +115,7 @@ Using JudgeIt framework is simple, just pick what is the task you want to evalua
 - [ ] Query-Rewrite support for More-than 2-turn
 - [ ] Specific support for multiple LLM generated text formats (Like Anthropic etc.) 
 
-**Known-Limitation** :
+**Current-Release-Limitation** :
 1. Verbose vs Crisp RAG Answers - We found that the framework sometimes becomes extremely conservative when there is a large gap between the size of the golden text and the generated text. For RAG comparisons, if the golden text is one page long and the generated text is just a few lines, it tends to treat them as dissimilar. We are currently working on a more 'liberal' version of this judge, which will be released soon.
 2. Comparing outputs for two LLMs.- We found that some LLM's preface their answers with some text, which can throw Judge off. When comparing two LLM try to avoid standard text like "I am just an average Language Model...etc". We are working on next version which will do this automatically.