Request for ToRA Self-Consistency Results with k=10 and k=20

[AceCHQ]

Due to resource constraints, I am particularly interested in comparing the performance of the ToRA-code with self-consistency for k=10 and k=20. Could you kindly provide the results for these values of k on model sizes 7B, 13B, and 34B?
Thank you very much for your assistance.