change n and allow clean up

Author: arjun37602

.github/workflows/terminal-bench.yaml

@@ -61,7 +61,7 @@ jobs:
     - name: Run terminal benchmark
       run: |
         cd terminal-bench-test
-        tb run --agent-import-path main:AmazonQCLIAgent --dataset-name terminal-bench-core --dataset-version head
+        tb run --agent-import-path main:AmazonQCLIAgent --dataset-name terminal-bench-core --dataset-version head --n-tasks=20 --cleanup
 
     # uploads results if run fails as well to allow for easy log inspection
     - name: Upload results