Run the prediction and export to GeoTIFF images failure about eo-learn HOT 8 CLOSED

I've been waiting for 6 hours

from eo-learn.

Hi @ZHAN520,

we started to notice similar things happening with EOExecutor. Unfortunately we still didn't quite pinpoint what the problem is, but it has something to do with paralellising the workflow.

For now, can you confirm that it works if you do the following on a single CPU:

for args in tqdm(execution_args):
    workflow.execute(args)

Sorry for the issues.

from eo-learn.

Hi @ZHAN520,

we started to notice similar things happening with EOExecutor. Unfortunately we still didn't quite pinpoint what the problem is, but it has something to do with paralellising the workflow.

For now, can you confirm that it works if you do the following on a single CPU:

for args in tqdm(execution_args):
    workflow.execute(args)

Sorry for the issues.

Okay, I now off work, tomorrow I will try, thank you.

from eo-learn.

Hi @ZHAN520,

we started to notice similar things happening with EOExecutor. Unfortunately we still didn't quite pinpoint what the problem is, but it has something to do with paralellising the workflow.

For now, can you confirm that it works if you do the following on a single CPU:

for args in tqdm(execution_args):
    workflow.execute(args)

Sorry for the issues.

Thank you, I have solved, and successfully completed their own data，

But now I want to use my server GPU training, I made the following modification, but did not succeed

，I need to do ？

from eo-learn.

Hi @ZHAN520,

unfortunately, I'm not so familiar with running on GPU. Did you try to find some tutorials online? Perhaps: https://lightgbm.readthedocs.io/en/latest/GPU-Tutorial.html

from eo-learn.

Hi @ZHAN520,

unfortunately, I'm not so familiar with running on GPU. Did you try to find some tutorials online? Perhaps: https://lightgbm.readthedocs.io/en/latest/GPU-Tutorial.html

Thank you, I has been solved，

But now I want to use your own tif image data to train and predict, how should I go to convert EOPatch

from eo-learn.

Hi @ZHAN520,

we started to notice similar things happening with EOExecutor. Unfortunately we still didn't quite pinpoint what the problem is, but it has something to do with paralellising the workflow.

For now, can you confirm that it works if you do the following on a single CPU:

for args in tqdm(execution_args):
    workflow.execute(args)

Sorry for the issues.

Hi there,
Is there an update on the parallel processing? Single CPU doesn't help much if running at a national scale.

Thanks in advance!

from eo-learn.

This has been improved, and handles running workflows in parallel.

The option to run either multi-process or multi-thread is provided, which helps when running algorithms that already perform parallelization.

from eo-learn.