Update README.md

README.md

@@ -27,13 +27,11 @@ Candidate answers are ranked based on community voting, with the highest-upvoted
 This community-driven ranking mechanism not only ensures that the dataset captures genuine user preferences but also establishes a meaningful benchmark for assessing how closely large language models align with human judgments of response quality.
 For information about dataset creation, limitations etc. see the cited preprint below.
 
-<img src="DemosQA.svg" width="200"/>
+<img src="demosqa.png" width="200"/>
 
-### Supported Tasks and Leaderboards
+### Supported Tasks
 
-This dataset supports:
-
-**Question Answering:** A text generation model can be evaluated for QA.  
+This dataset supports evaluation of LLMs for **Question Answering**.
 
 ### Languages
 
@@ -54,6 +52,7 @@ The following data fields are provided for each split:
 `answers`: (**str**): The string containing the four candidate answers.  
 `best_answer`: (**str**) The best answer text selected by the community. 
 `best_answer_index`: (**str**) The letter index of the best answer. 
+`date`: (**str**) The post publication date.
 
 ### Example code
 ```python