Comment by bionhoward

This looks impressive but I’m concerned, is it fair to “teach to the test” by fine tuning the Qwen model with RL on the test task, while the other models in the comparison are not fine tuned on the test task?