Update README.md

README.md

@@ -36,12 +36,12 @@ size_categories:
 
 It achieves **superior performance** on math benchmarks including arithmark 3.0, and achieving better math loss.
 
-**3M model trained on only finemath-3plus vs WebMathUltra, same architecture, 500 millon tokens.**
+**9M model trained on only finemath-3plus vs WebMathUltra, same architecture, 500 millon tokens.**
 
 | Metric | finemath 3+ | WebMathUltra |
 |---|---:|---:|
-| Math loss | 3.27 | **3.25** |
-| Arithmark 3.0 | 54.95% | **55.55%** |
+| Math loss | 2.93 | **3.25** |
+| Arithmark 3.0 | 35.30% | **55.55%** |
 
 ## Dataset Details