Fabien Roger comments on Training Qwen-1.5B with a CoT legibility penalty

[Fabien Roger]

I agree overt and covert encoded reasoning are different and you should flag overt encoded reasoning, and covert seems harder than overt. This is what I tried to point at by saying

I relax the “plausible benign” constraint and try to find math reasoning that doesn’t look like math reasoning to a monitor.”