Possible minor formatting issue in post-training results table

[3300786]

While reviewing the table of post-training evaluation results in Section 4.1.2 (page 16), I noticed that Sonnet-4’s score of 89.8 under AutoLogi might not be bolded, even though it seems to be the best in that row. This could be a minor formatting oversight.

Could you please confirm whether this was intentional or an omission?

Thanks again for your great work!

[bigeagle]

Yes, you are right. We'll fix this issue in the next version.