• Generation works

Next steps:

• train on (217, 20) (see below. need to adjust the data preprocessing to return bigger chunks (32->224, 256->1792)
• condition on a meaningful embedding (for now: fourier)
• make an example submission (just predict the first 208 datapoints and repeat-pad the rest)
• if it works well score-wise, predict the second half independently and interpolate

Completed 46/72. Loaded data: (1730, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 47/72. Loaded data: (1730, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 48/72. Loaded data: (1732, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 49/72. Loaded data: (1731, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 50/72. Loaded data: (1735, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 51/72. Loaded data: (1730, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 52/72. Loaded data: (1729, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 53/72. Loaded data: (1730, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 54/72. Loaded data: (1730, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 55/72. Loaded data: (1729, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 56/72. Loaded data: (1734, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 57/72. Loaded data: (1733, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 58/72. Loaded data: (1729, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 59/72. Loaded data: (1731, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 60/72. Loaded data: (1731, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 61/72. Loaded data: (1732, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 62/72. Loaded data: (1736, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 63/72. Loaded data: (1731, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 64/72. Loaded data: (1732, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 65/72. Loaded data: (1733, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 66/72. Loaded data: (1734, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 67/72. Loaded data: (1729, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 68/72. Loaded data: (1727, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (216, 20)
Completed 69/72. Loaded data: (1736, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 70/72. Loaded data: (1729, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 71/72. Loaded data: (1735, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)
Completed 72/72. Loaded data: (1735, 8) - padded to: (1792, 8) - predictions (224, 20) - downsampled to: (217, 20)