Asma/fix source writing/head#40
Conversation
|
Hi, Suggestion: please use About the issue on testing, it's weird that ERA5 and target now do not match anymore as you seem to have touched only source. Unless there's a problem with copying and memory in python. Are these the only changes wrt the |
|
Hi Ilaria, Your suggested has been noted 👍 Regarding the issue on testing, I only changed source, yes. I am also reading the data from |
|
Additional info here: |
|
What do you mean by "data partitions at JSC"? Locations of files/paths at JSC? |
|
I rebased to a new branch from With the original link in I get: Which I am understanding that it is the default output if ERA5 is not found Now if I adapt the link to what I have access on JSC HPC systems: I get: Conclusion: The problem was here all along, and is not related to this "fix-source-writing" changes |
Apologies, I haven't seen this comment when I wrote the last one ^^'. At JSC, data is under the path |
This PR is related to the issue " source and target not written the same way in BERT mode #38". The data is now accessed the same way. However, I keep having the following test failure:
Might be related to the dataset not being loaded properly (I am unsure if this is what you meant @iluise during our first temporal interpolation meeting)