search

cta-observatory / lstmcpipe

Scripts to analyse MC files on LST cluster at La Palma
https://cta-observatory.github.io/lstmcpipe/
MIT License
7 stars 15 forks source link

Failing prod 20240716_v0.10.7_src10_dec3476_tuned #471

Open vuillaut opened 1 month ago

vuillaut commented 1 month ago

Hi @ToraTherese

Unfortunately this is failing because 

/fefs/aswg/data/mc/DL1/AllSky/20240430_v0.10.4_src9_dec3476_tuned/TrainingDataset/dec_3476/GammaDiffuse/

is empty 😬

The dataset had been separated in train/test ones:

/fefs/aswg/data/mc/DL1/AllSky/20240430_v0.10.4_src9_dec3476_tuned/Split_TestDataset/
/fefs/aswg/data/mc/DL1/AllSky/20240430_v0.10.4_src9_dec3476_tuned/Split_TrainingDataset/

So you could start from these and do the dl1ab stage without the train/test split stage, e.g.:

input:  /fefs/aswg/data/mc/DL1/AllSky/20240430_v0.10.4_src9_dec3476_tuned/Split_TrainingDataset/dec_3476/GammaDiffuse/node_corsika_theta_12.829_az_301.263_/
output: /fefs/aswg/data/mc/DL1/AllSky/20240716_v0.10.7_src10_dec3476_tuned/Split_TrainingDataset/dec_3476/GammaDiffuse/node_corsika_theta_12.829_az_301.263_

etc...

opening an issue...

Originally posted by @vuillaut in https://github.com/cta-observatory/lstmcpipe/issues/470#issuecomment-2233557514

ToraTherese commented 1 month ago

Sorry about that, and thanks for the advice! I've modified the config to start with the already existing split datasets, committed it to the branch, and submitted a new pull request