skeenan

This has been dormant since January. I am closing in 2 days unless objected.

This has been dormant since November. I am closing in 2 days.

Thanks for the comment @macieksmuga I'll leave this open for now and revisit on 1 month.

Done. You should now be able to do this @nlwashington

FAO @mfiume @mcupak

@Welliton309 and @ljdursi these are both awesome. So cool. We need to highlight these on http://ga4gh.org or on the landing page of http://1kgenomes.ga4gh.org/ (or both!)

On Apple M1 GPU I get TypeError: Cannot convert a MPS Tensor to float64 dtype as the MPS framework doesn't support float64. Please use float32 instead.

Got this working. followed the example in https://unit8.com/resources/time-series-forecasting-using-past-and-future-external-data-with-darts. My changes: converted timeseries objects to to np.float32 flow = flow.astype(np.float32) rainfalls = rainfalls.astype(np.float32) and configured the model thus ``` rnn_rain =...

Oddly. When running on CPU the training takes 21 seconds. When GPU is enabled it takes 3 minutes 17 seconds. Unexpected.

Thanks for the pointer. Followed the example https://unit8co.github.io/darts/userguide/gpu_and_tpu_usage.html#use-a-gpu Unfortunately with num_loader_workers > 1 the model.fit() does not ever return (tried even with epoch=1) just hangs indefinitely. Tried a range of...