Available Parameters (--help)
An overview table of all arguments that OmniOpt2 accepts via the Command line. Same data as in ./omniopt --help.
| Parameter | Description | Type | Action | Default Value |
|---|---|---|---|---|
| Ungrouped | ||||
|
YAML configuration file | |
||
|
TOML configuration file | |
||
|
JSON configuration file | |
||
| Required arguments - These options have to be set | ||||
|
Number of random steps to start with | |
|
|
|
Maximum number of evaluations | |
||
|
A program that should be run. Use, for example, $x for the parameter named x | |
|
|
|
Name of the experiment | |
||
|
Amount of RAM for each worker in GB (default: 1GB) | |
|
|
| Required arguments that allow a choice - Of these arguments, one has to be set to continue | ||||
|
Experiment parameters in the formats (options in round brackets are optional): <NAME> range <LOWER BOUND> <UPPER BOUND> (<INT, FLOAT>, log_scale: True/False, default: false>) -- OR -- <NAME> fixed <VALUE> -- OR -- <NAME> choice <Comma-separated list of values> | |
||
|
Continue from a previous checkpoint, use run-dir as argument | |
||
| Optional - These options are optional | ||||
|
Constraints for parameters. Example: x + y <= 2.0. Convert them to base64 | |
|
|
|
Directory, in which runs should be saved. Default: runs | |
|
|
|
Seed for random number generator | |
||
|
Number of decimal places for rounding | |
|
|
|
Enforce sequential optimization (default: false) | |
|
|
|
Show verbose TQDM messages | |
|
|
|
Use special models for nonrandom steps. Valid models are: "SOBOL", "FACTORIAL", "SAASBO", "BOTORCH_MODULAR", "UNIFORM", "BO_MIXED", "RANDOMFOREST", "EXTERNAL_GENERATOR", "PSEUDORANDOM" | |
||
|
Enable gridsearch | |
|
|
|
Use optimization with combined criteria (OCC) | |
|
|
|
Show sixel graphics of scatter plots in the end | |
|
|
|
Show sixel graphics of general plots in the end | |
|
|
|
Show sixel graphics of trial index in the end | |
|
|
|
Automatically follow log file of sbatch | |
|
|
|
Send anonymized usage stats | |
|
|
|
Site from which the OO-run was called | |
||
|
Where to install your modules to ($root_venv_dir/.omniax_..., default: $HOME) | |
||
|
A comma separated list of values of excluded nodes (taurusi8009,taurusi8010) | |
||
|
Amount of RAM for the main process in GB (default: 8GB) | |
|
|
|
Confidence for pareto-front-plotting (between 0 and 1, default: 1) | |
|
|
|
Max. nr of successive zero results by the generator before the search space is seen as exhausted | |
|
|
|
Abbreviate pending job names (r = running, p = pending, u = unknown, c = cancelling) | |
|
|
|
An orchestrator file | |
||
|
Automatically checkout to latest version that was tested in the CI pipeline | |
|
|
|
Automatically live-share the current optimization run automatically | |
|
|
|
Disables the TQDM progress bar | |
|
|
|
For continued jobs: Disable getting the constraint of the previous job that is about to be continued | |
|
|
|
Working directory | |
||
|
Optimization-with-combined-criteria-type (valid types are "geometric", "euclid", "signed_harmonic", "signed_minkowski", "weighted_euclid", "composite") | |
|
|
|
Name of hyperparameters. Example --result_names result1=max result2=min result3. Default: RESULT=min | |||
|
Minkowski order of distance (default: 2), needs to be larger than 0 | |
|
|
|
A comma-seperated list of values for the signed weighted euclidean distance. Needs to be equal to the number of results. Else, default will be 1 | |
||
|
A string containing the generation_strategy. Example: SOBOL=10,BOTORCH_MODULAR=10,SOBOL=10. Cannot use --model EXTERNAL_GENERATOR or PSEUDORANDOM | |
||
|
Generate all jobs at once rather than to create them and start them as soon as possible | |
|
|
|
Generate random steps instead of systematic steps when the search space is (seemingly) exhausted | |
|
|
|
List of job data to load from existing jobs | |
||
|
The number of trees in the forest for RANDOMFOREST (default: 100) | |
|
|
|
Programm call for an external generator | |
||
|
A username for live share | |
||
|
Maximum number of failed jobs before the search is cancelled. Is defaulted to the value of --max_eval | |
||
| SLURM - Parameters related to SLURM | ||||
|
Number of parallel SLURM jobs (default: 20) | |
|
|
|
Timeout for SLURM jobs (i.e. for each single point to be optimized) | |
|
|
|
Using srun instead of sbatch | |
|
|
|
Time for the main job | |
||
|
Partition to be run on | |
||
|
Reservation | |
||
|
Forces local execution even when SLURM is available | |
|
|
|
When the workers end, they get a signal so your program can react to it. Default is 0, but set it to any number of seconds you wish your program to be able to react to USR1 | |
|
|
|
Number of nodes per job due to the new alpha restriction | |
|
|
|
CPUs per task | |
|
|
|
Account to be used for SLURM | |
||
|
Number of GPUs per worker | |
|
|
| Installing - Parameters related to installing | ||||
|
Either local or docker | |
|
|
| Debug - These options are mainly useful for debugging | ||||
|
Verbose logging | |
|
|
|
Verbose logging for break_run_search | |
|
|
|
Enable debugging | |
|
|
|
Disables sleeping for fast job generation (not to be used on HPC) | |
|
|
|
Run simple internal tests | |
|
|
|
Show a table of percentage of usage of max worker over time | |
|
|
|
Run a Test if you can allocate a GPU on each node and if not, exclude it since the GPU driver seems to be broken somehow | |
|
|
|
Run tests on Taurus that usually fail | |
|
|
|
Raise a signal in eval (only useful for debugging and testing) | |
|
|
|
Show RAM usage every n seconds (0 = disabled) | |
|
|