A subtle issue when calculating t-student kernel

[AlfredTino]

Thank you for providing such a sophisticated toolbox for the community!
When I was studying the code, I found a subtle issue: models.layers line 156. In fact, when calculating t-student kernel, one should multiple gamma with squared d. However, here you square the product of gamma and d. I do not evaluate the effect now.

[pkorus]

Thanks for reporting this. I'll keep this mind before the next release.