Update README.md

README.md

@@ -66,13 +66,10 @@ Each instance is a short question-answer pair. Example:
 
 ### Data Splits
 
-> *(Update this table with your actual numbers before publishing.)*
 
 | Split | # Examples |
 |-------|-----------|
-| train |           |
-| validation |      |
-| test  |           |
+| train |    2.51k  |
 
 ## Dataset Creation
 
@@ -85,11 +82,11 @@ Experiments showed that very small base models (1M–50M parameters) reliably ge
 
 ### Curation Rationale / Source Data
 
-> *(Describe here how the QA pairs were collected or generated — e.g., manually written, extracted from a larger QA corpus and filtered by length, synthetically generated, etc.)*
+> *This dataset is synthatic generated by Sarwam 105B Conversation model*
 
 ### Annotations
 
-> *(Describe if answers were human-verified, filtered automatically by word count, etc.)*
+> *This dataset is tested and verified*
 
 ## Considerations for Using the Data
 
@@ -99,9 +96,6 @@ Experiments showed that very small base models (1M–50M parameters) reliably ge
 - Intended specifically for small-parameter models; larger models may not show meaningful benefit.
 - Answer brevity may limit factual nuance (e.g., no room for caveats or partial answers).
 
-### Social Impact / Bias
-
-> *(Add any notes on potential bias in question/answer selection, if applicable.)*
 
 ## Additional Information