André Pedersen

@dawoodkhan82 Would be neat if there was added support for `gradio.Dataframe`, `gradio.Code`, and `gradio.Button` as well, as our application needs it :] As a minimum, `gradio.Dataframe` would be quite neat....

Hello, @Teera21! :] What do you mean finetune? Finetune the embedding model? Or do you just want to enhance accuracy in the routing? I would try using `HybridRouteLayer` and tuning...

As discussed in the other thread, it would be of interest to test the proposed solution as it works for the latest implementation of the Optimizer class and may work...

Feel free to tag in @IvanUkhov, if you have any questions regarding the implementation or whatnot. Might be that we should add support for his implementation to this framework, regardless...

@tno123 Could you share the notebook/gist/script that reproduces the issue? That way, @IvanUkhov can use it for debugging as well :]

> Just quickly opened this tutorial in Colab @IvanUkhov I believe @tno123 tested this in Kaggle, where he has access to two GPUs. Hence, he can test whether using this...

> It works now. Here is the main change: @IvanUkhov Sounds great! Could you run some tests using it, @tno123? :]

> Edit: It seems to work! It is training, but I'll do some more tests just to make sure everything is good. Any update, @tno123? :] Getting it to mechanically...

> they don't manage to send me an email for account validation before the email validation link becomes invalid. That's unfortunate. You should try again later. Kaggle is the only...

> In that case on should be able to drop the scaling step right before the apply, because the gradient will not be used anyway, would they ? So there...