Add TVD (Total variation distance) Kernel

[qingquansong]

🚀 The feature, motivation and pitch

TVD is a good distance metric (ref) and easy to implement kernel to make the gradient more stable compared to KL divergence and JS Divergence.

Alternatives

No response

Additional context

No response

[S1ro1]

I'll look into it over the week if noone else takes.

[saurabhkoshatwar]

#take @ByronHsu @qingquansong , I’d like to make an attempt. Could you please assign it to me?

[ByronHsu]

assigned to you. Thanks!