Explore

These commands describe the benchmarks, datasets, models and tasks without running anything. Benchmarks explains what they show.

list

List the benchmarks, datasets, models, aliases or tasks. This reads only the tables that ship with NeuroAtlas. Use neuroatlas data status and neuroatlas models status to see what is on this machine.

Usage: neuroatlas list <what>

Command	Description
list benchmarks	List each benchmark with its task, datasets and headline metric.
list datasets	List the paper's datasets with their domain, access, size and benchmarks.
list models	List the checkpoints with their family, group, input rate, window and embedding size.
list aliases	List the names that select groups of checkpoints with -m.
list tasks	List the probe tasks and the settings each benchmark runs them with.

list benchmarks

List each benchmark with its task, datasets and headline metric.

Usage: neuroatlas list benchmarks [options]

neuroatlas list benchmarks
neuroatlas list benchmarks -v

Option	Description	Default
--grep TEXT	Only show rows that mention TEXT (case-insensitive).
-v, --verbose	Also show how to get each dataset and the models a benchmark leaves out.
--format {table,csv,md,json}	Output format. In csv, md and json the notes under a row go to a note column.	table

list datasets

List the paper's datasets with their domain, access, size and benchmarks.

Usage: neuroatlas list datasets [options]

neuroatlas list datasets --grep sleep
neuroatlas list datasets --all --format csv

Option	Description	Default
--grep TEXT	Only show rows that mention TEXT (case-insensitive).
-v, --verbose	Also show the web page or MOABB name of each dataset.
--format {table,csv,md,json}	Output format. In csv, md and json the notes under a row go to a note column.	table
--all	Also list the datasets that can be read but are not in the paper.

list models

List the checkpoints with their family, group, input rate, window and embedding size.

Usage: neuroatlas list models [selector] [options]

neuroatlas list models all_fm
neuroatlas list models --benchmark epilepsy

Option	Description	Default
selector	Only show these checkpoints, given as an alias, group, family or ids such as all_fm, baseline or reve.
--grep TEXT	Only show rows that mention TEXT (case-insensitive).
-v, --verbose	Also show where the weights of each checkpoint come from.
--format {table,csv,md,json}	Output format. In csv, md and json the notes under a row go to a note column.	table
--benchmark NAME	Leave out the models this benchmark does not evaluate.

list aliases

List the names that select groups of checkpoints with -m.

Usage: neuroatlas list aliases [options]

neuroatlas list aliases -v

Option	Description	Default
--grep TEXT	Only show rows that mention TEXT (case-insensitive).
-v, --verbose	Also list the checkpoints of each alias.
--format {table,csv,md,json}	Output format. In csv, md and json the notes under a row go to a note column.	table

list tasks

List the probe tasks and the settings each benchmark runs them with.

Usage: neuroatlas list tasks [options]

neuroatlas list tasks

Option	Description	Default
--grep TEXT	Only show rows that mention TEXT (case-insensitive).
-v, --verbose	Also list the tasks no benchmark runs, and the kind of each.
--format {table,csv,md,json}	Output format. In csv, md and json the notes under a row go to a note column.	table

show

Describe one benchmark and print the embed and probe commands that run executes for it. Use it to see the metrics, folds, datasets and variants of a benchmark.

Usage: neuroatlas show benchmark [options]

neuroatlas show sleep_stage
neuroatlas show epilepsy --dataset single -m cbramod_pretrained

Option	Description	Default
benchmark	Benchmark name, as listed by neuroatlas list benchmarks.
--dataset single\|full\|NAMES	Datasets to use. Give single for the benchmark's quick dataset, full for all its datasets, or dataset names separated by commas.	full
--variant VARIANT	Benchmark variant to describe.	default
-m, --models MODELS	Put this model selection into the printed commands.