mirror of
https://github.com/clearml/clearml-docs
synced 2025-01-31 06:27:22 +00:00
69 lines
2.6 KiB
Markdown
69 lines
2.6 KiB
Markdown
|
---
|
||
|
|
title: jsonargparse
|
||
|
|
---
|
||
|
|
|
||
|
|
:::tip
|
||
|
|
If you are not already using ClearML, see [Getting Started](../getting_started/ds/ds_first_steps.md) for setup
|
||
|
|
instructions.
|
||
|
|
:::
|
||
|
|
|
||
|
|
|
||
|
|
[jsonargparse](https://github.com/omni-us/jsonargparse) is a Python package for creating command-line interfaces.
|
||
|
|
ClearML integrates seamlessly with `jsonargparse` and automatically logs its command-line parameters and connected
|
||
|
|
configuration files.
|
||
|
|
|
||
|
|
All you have to do is add two lines of code:
|
||
|
|
|
||
|
|
```python
|
||
|
|
from clearml import Task
|
||
|
|
|
||
|
|
task = Task.init(task_name="<task_name>", project_name="<project_name>")
|
||
|
|
```
|
||
|
|
|
||
|
|
When the code runs, ClearML logs your command-line arguments, which you can view in the [WebApp](../webapp/webapp_overview.md),
|
||
|
|
in the experiment's **Configuration > Hyperparameters > Args** section.
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
### Automatic Logging Control
|
||
|
|
By default, when ClearML is integrated into your script, it captures all of your `jsonargparse` parameters.
|
||
|
|
But, you may want to have more control over what your experiment logs. To control a task's logging of parameters from
|
||
|
|
argument parsers, use the `auto_connect_arg_parser` parameter of [`Task.init()`](../references/sdk/task.md#taskinit).
|
||
|
|
Completely disable all automatic logging by setting the parameter to `False`.
|
||
|
|
|
||
|
|
```python
|
||
|
|
auto_connect_arg_parser=False
|
||
|
|
```
|
||
|
|
|
||
|
|
For finer grained control of logged parameters, input a dictionary with parameter-boolean pairs. The `False` value
|
||
|
|
excludes the specified parameter. Unspecified parameters default to `True`.
|
||
|
|
|
||
|
|
For example, the following code will not log the `Example_1` parameter, but will log all other arguments.
|
||
|
|
|
||
|
|
```python
|
||
|
|
auto_connect_arg_parser={"Example_1": False}
|
||
|
|
```
|
||
|
|
|
||
|
|
To exclude all unspecified parameters, set the `*` key to `False`. For example, the following code will log **only** the
|
||
|
|
`Example_2` parameter.
|
||
|
|
|
||
|
|
```python
|
||
|
|
auto_connect_arg_parser={"Example_2": True, "*": False}
|
||
|
|
```
|
||
|
|
|
||
|
|
## Remote Execution
|
||
|
|
|
||
|
|
In the UI, you can clone a task multiple times and modify it for execution by the [ClearML Agent](../clearml_agent.md).
|
||
|
|
The agent executes the code with the modifications you made in the UI, even overriding hardcoded values.
|
||
|
|
|
||
|
|
In the case that you connected a jsonargparse configuration file (e.g. with LightningCLI), make sure to set the
|
||
|
|
`_ignore_ui_overrides` to `False` in the **CONFIGURATION > HYPERPARAMETERS > ARGS** section. That way, after the customized
|
||
|
|
experiment is enqueued, the task will use the new values during execution.
|
||
|
|
|
||
|
|
|
||
|
|
## Code Examples
|
||
|
|
|
||
|
|
See [code examples](https://github.com/allegroai/clearml/blob/master/examples/frameworks/jsonargparse) demonstrating integrating
|
||
|
|
ClearML with code that uses `jsonargparse`.
|