Gökçen Eraslan

Results 91 comments of Gökçen Eraslan

It's this line: ``` js console.warn("%c" + t, e ? "color: darkorange" : "color: red") ```

Basic MNIST works fine. I'm on Arch Linux. `e` is undefined here: ![image](https://cloud.githubusercontent.com/assets/1140359/19408535/e67b5e0a-92bf-11e6-9c35-db15b3cf4e30.png) and `this.output` is undefined here: ![image](https://cloud.githubusercontent.com/assets/1140359/19408527/cfad7ef6-92bf-11e6-95a9-64a71ff4fa81.png)

I agree to @alxndrkalinin tags and rev. order make sense. PRs are welcome :)

It is printed after each epoch if you set the `test_dataset` parameter of the `Trainer()`

It'd be really great to add this feature!

Souns like a good idea, but ardb seems unmaintained as you can see from commit log and github issues.

Hi, > This is an interesting enhancement, thanks for the suggestion. In principle, yes it's possible to calculate the cellular activity in chunks, but only after the generation of the...

> In principle, yes it's possible to calculate the cellular activity in chunks, but only after the generation of the ranking matrix. Wait, why possible only after the generation of...

or in other words why are the coefs not shown in the summary here https://github.com/theislab/diffxpy/blob/master/diffxpy/testing/det.py#L832

ok, I managed to get them: ``` pd.DataFrame(test.theta_mle, index = np.array(coefs)[test.coef_loc_totest.tolist()], columns=test.gene_ids).T ``` but it'd be great to have either a separate function doing that or a way to show...