Fixed the benchmark scores. (#2)

- Fixed the benchmark scores. (37597da16e985049592a8fd90e95bb5610b09f57)

README.md

@@ -55,12 +55,12 @@ Pebble-10M-Chat was evaluated on several commonsense and arithmetic benchmarks.
 
 | Benchmark       | Accuracy | Random Baseline |
 |-----------------|----------|-----------------|
-| PIQA            | 58.43%   | 50.00%          |
-| ARC-Easy        | 37.29%   | 25.00%          |
-| ARC-Challenge   | 18.60%   | 25.00%          |
-| HellaSwag       | 26.81%   | 25.00%          |
-| ArithMark-2.0   | 27.64%   | 25.00%          |
-| ArithMark-3.0   | 32.80%   | 25.00%          |
+| PIQA            | 51.41%   | 50.00%          |
+| ARC-Easy        | 26.09%   | 25.00%          |
+| ARC-Challenge   | 20.22%   | 25.00%          |
+| HellaSwag       | 25.30%   | 25.00%          |
+| ArithMark-2.0   | 27.28%   | 25.00%          |
+| ArithMark-3.0   | 27.50%   | 25.00%          |
 
 ### Evaluation Notes