Modes
The following modes are available for the Neptune Scale Python client:
| Mode | Value | Description |
|---|---|---|
| Asynchronous (default) | async | Neptune continuously synchronizes the logged data with the server. |
| Disabled | disabled | No data is logged or sent anywhere. |
| Offline | offline | Neptune only stores the logged data on the local disk. |
To set the mode, pass it to the mode argument when initializing a run:
from neptune_scale import Run
run = Run(
mode="async",
run_id=...,
)
You can also set the mode using the NEPTUNE_MODE environment variable.
Asynchronous mode (default)
The tracked data is stored locally on-disk and continuously synchronized by a separate process in the background. Once fully processed on the server, the data is deleted from disk.
All logging calls are non-blocking.
On connectivity issues, the client tracks your metadata locally and continuously tries to re-establish a connection with the server. Any unsuccessful logging calls are stored on the local disk.
run = Run(
mode="async",
...
)
Disabled
All Neptune-related logging is turned off. It can come in handy when you're debugging your code and don't want to pollute your Neptune project.
run = Run(
mode="disabled",
...
)
Offline
Stores the tracked metadata on the local disk, without synchronizing it with the server. Use this mode to log metadata without a connection.
run = Run(
mode="offline",
...
)
To synchronize the offline data with the Neptune server, use the CLI tool.