Zhylko Dima

I guess FRCNN now supports empty inputs: https://github.com/tensorpack/tensorpack/issues/1256

Thank you, I guess main issue was that optimiser is set to minimize loss while score itself was meant to be maximized :) But I also have a question about...

Can you try using #19 pull request, maybe this will help. Tell if it's worked for you

Same question, also I believe there is a question about this on stackoverflow: https://stackoverflow.com/questions/71623674/lightgbm-with-tweedie-loss-im-confused-on-the-gradient-and-hessians-used could you provide some reading on tweedie loss topic?

@elephaint yeah I saw this in this issue https://github.com/microsoft/LightGBM/issues/3155 but I still don't know how the gradient was derived, because I am arriving on different result every time :(

@lorentzenchr I understand exp and non exp thing, but I don't understand how the derivative of `-label * pred ^ (1 - rho) / (1-rho) + pred ^ (2 -...

Up! I also think that this implementation is not correct original implementation in fvcore uses binary cross entropy and multiplies class 0 by alpha and class 1 by 1-alpha, so...

hey @shijianjian, my implementation is more about multi-class situation, so I don't know if there is a direct comparison. But I can try to help (but I am afraid it...

Actually my implementation is exactly what you said. If you pass float to alpha it will be used as what I called one-vs-other mode (alpha is the weight for class...

hey, I think 0.23.0 is the last version to work