Working with your environment#

Sometimes, you may need to have more information about your environment data in JAI. For this, JAI owns some managing methods that are shown in this section.

j.names#

If you need to know which databases are in your environment, use j.names. It’ll show you the names of all your databases in JAI.

>>> # Example response supposing these collections are in the environment:
>>> j.names

['jai_database', 'jai_selfsupervised', 'jai_supervised']

j.info#

To get more information about your collections, use j.info. This method will present all your collection names, types, last modifications, dependencies and sizes.

>>> j.info

              name|           type|    last modified|          dependencies|   size|
------------------|---------------|-----------------|----------------------|-------|
      jai_database|           Text| 2021-10-26-20h22|                    []|  20640|
jai_selfsupervised| SelfSupervised| 2021-11-03-17h50|                    []|   8500|
    jai_supervised|     Supervised| 2021-11-03-17h50|  [jai_selfsupervised]|   8500|
------------------------------------------------------------------------------------

The drawback of j.info is that it can run slowly depending on the number of collections in your environment.

j.fields#

If you forgot what columns your database has, this information could be accessed by j.fields method.

>>> # Use your collection name
>>> j.fields(name='jai_database')

{'id': 'int64',
 'Column1': 'float64',
 'Column2': 'float64',
 'Column3': 'float64',
 'Column4': 'float64'}

j.get_dtype#

To get what collection type your collection is, use j.get_dtype.

>>> # Use your collection name
>>> j.get_dtype(name='jai_selfsupervised')

'SelfSupervised'

j.describe#

However, if you need details of what parameters you choose to fit your collection, j.describe can bring it for you.

>>> # Use your collection name
>>> j.describe(name='jai_database')

{'name': 'california',
 'dtype': 'SelfSupervised',
 'state': 'active',
 'version': '2021-10-26-20h22',
 'has_filter': False,
 'model_hyperparams': {'batch_size': 512,
 'learning_rate': 0.01,
 'encoder_layer': '2LM',
 'decoder_layer': '2LM',
 'hidden_latent_dim': 64,
 'dropout_rate': 0.1,
 'momentum': 0.1,
 'pretraining_ratio': 0.1,
 'noise_level': 0.0,
 'training_type': 'contrastive'}
 ...

j.report#

To recover the fit report for your collection, use j.report.

>>> # Use your collection name
>>> j.report(name='jai_database')

j.ids#

If you need to remember how many ids your collection have, use j.ids.

>>> # Use your collection name
>>> j.ids(name='jai_database', mode='summarized') # default

['20640 items from 0 to 20639']

For more information about how to work with your environment, check JAI Python Class