Skip to content
Snippets Groups Projects

Repository graph

You can move around the graph by using the arrow keys.
Select Git revision
  • 1-test_juwels
  • 104-adopt-runscript-generator-for-multiple-variables
  • 143-licensing
  • 143-licensing-2
  • Bing_issue#018_revise_convLSTM_to_Loliverhennigh_version
  • Bing_issue#156_test_CICD
  • Gong2022_temperature_forecasts
  • amirpasha_including_2tier_pystager_extraction
  • amirpasha_issue#008_data_extraction
  • amirpasha_issue#065_handling_exceptions
  • amirpasha_issue#33_fail_communication_between_master_workers
  • bing_#172_integrate_weatherbench
  • bing_#37_organize_last_mins_meeting
  • bing_clean_GMD1
  • bing_horovod_convLSTM
  • bing_issue#001_issue#003_test_juwels
  • bing_issue#005_clean_and_update_CICD_file
  • bing_issue#006_add_unitest
  • bing_issue#009_clean_up_postprocessing
  • bing_issue#010_remove_hickle_split_data
  • ambs_v1
  • ambs_gmd_nowcasting_v1.0
  • GMD1
  • modular_booster_20210203
  • new_structure_20201004_v1.0
  • old_structure_20200930
26 results
Created with Raphaël 2.2.019Oct1615131287654130Sep2928252423222118171514111098432131Aug2827111097654331Jul30272524221514107632130Jun2927262524232219181710987543229May28272625201918151427Apr241715149add srun in front of pythonSmall bug fix, added print statement to know which tensorflow verions we are using.Revise function that can search for train/val/test tfrecords without needing the directoryMerge branch 'bing_issue#029_data_preprocess_unitest' into developRemoved calculs for default values.Concise usage of source_dir in main_preprocess_data_step1.py.Adaption of main_preprocess_data_step1.py to changes in PreprocessNcToPkl-class.Added global vaiables such as src_dir_base to unit test and implemented check for processed number of files.Bugfix in calling order of arguments.Adaptions according to previous commit.Adding year to source- and destination-directory within PreprocessNcToPkl-instance.Adaptions according to previous commit.Rename some methods of PreprocessNcToPkl-class and change to target_dir-attribute to enforce existence of pickle-folder.Add a check if statistics and pickle file are saved in the correct directory.Bugfix in accessing target_file in call-method.Ensure creation of stat_01.json (test statistic file) if it's not present at test time.Replaced exit- by return-statement in run_pytest.sh in order to prevent logging out from HPC-system.revise comments for process_netcdf_v2.pyupdate era5_dataset.py and unitestUpdate process_netCDF_v2.pyIndeed nothing new, but permission of templates changed to ensure executability.Corrections to runscript/Batch script template 'train_model_era5_booster_template.sh'. Note that the modules have to be loaded before the runscript/Batch script is submitted.Merge branch 'scarlet_issue#031_booster' of https://gitlab.version.fz-juelich.de/toar/ambs into scarlet_issue#031_boosterintegration of horovod in convLSTMupdate era5_dataset.py and unitestReplaced automatic check for modules on Juwels Booster by note to perform test after deactivating running singularity.Corrected check for running singularity in create_env.shRemoval of some typos in create_env.sh and added usage of correct requirements-file.Add sanity check that create_env.sh is executed in a running singularity and add adaptions to setting up of virtual environment.Introduce Juwels Booster-specific requirements_booster.txt since Horovod is not available as a binary module there.update era5_data.pyChanged line 105 filter from a python list to a integer in order to solve an issue occuring with the new non-linearity in convLSTM.Removed duplicated srun-command from 'train_model_era5_template.sh'Install headless version of opencv follwoing the advice given under https://stackoverflow.com/questions/63977422/error-trying-to-import-cv2opencv-python-packageRemoved typos in 'modules_train.sh' and 'create_env.sh'.Deleted misplaced train_model_era5_booster.shAdaption of modules_train.sh for loading requested modules on Juwels Booster.First adaptions of 'create_env.sh' for handling newly introduced runscript template for training on Juwels Booster.Initial release of template Batch script for training on Juwels Booster.Added non-linearity in convLSTM cell.