Nic Dahlquist

Not sure; we didn't spend too much time optimizing, just wanted to fix the functionality. For what it's worth, the old Dockerfile also appears to use `ubuntu` as the base.

This is simpler than I expected: the average color is encoded in characters 3-6. (see https://github.com/woltapp/blurhash/blob/master/Algorithm.md#structure). ```dart // In a blurhash, chars 3-6 represent the average color. // Refer to...

I think for the 4 sides, that's slightly more complicated. You need to add the DC component (the "averageColor" above) with the AC components sampled at the edges of the...

Confirmed this fixes the issue on OS X for me. Thank you so much @pavan4 !

Same issue here. gradle 5.4.1. Android build tools 3.4.0.

Confirmed that this fix worked for us too. Before this patch: - The "Verify your app's deep link" flow in Facebook Events Manager successfully opened the app, but did not...

For me, pinning `better_player: 0.0.63` was insufficient to solve this problem. I also had to pin `video_player`: ``` tweet_ui: ^3.0.0 better_player: 0.0.63 video_player: 2.0.2 video_player_platform_interface: 4.0.0 ``` After that, the...

Fascinating question! Can you share more about your training data and how you are training? Assuming that you have a training frame for each timestamp, and the training loop treats...

Okay, I haven't tried running this code, so I may be misunderstanding it, but here are some initial thoughts: 1. I would have expected that instead of an `xy_grid`, you...

The mismatched `optimizer.zero_grad` and `optimizer.step()` in your training loop looks atypical as well. I suspect you might have intended: ```python for epoch in tqdm(range(5000)): for i in range(target.shape[0]): optimizer.zero_grad() generated...