# FromConfig MlFlow <!-- {docsify-ignore} -->
[](https://pypi.python.org/pypi/fromconfig-mlflow)
[](https://github.com/criteo/fromconfig-mlflow/actions?query=workflow%3A%22Continuous+integration%22)
A [fromconfig](https://github.com/criteo/fromconfig) `Launcher` for [MlFlow](https://www.mlflow.org) support.
<!-- MarkdownTOC -->
- [Install](#install)
- [Quickstart](#quickstart)
- [MlFlow server](#mlflow-server)
- [Configure MlFlow](#configure-mlflow)
- [Artifacts and Parameters](#artifacts-and-parameters)
- [Usage-Reference](#usage-reference)
- [`StartRunLauncher`](#startrunlauncher)
- [`LogArtifactsLauncher`](#logartifactslauncher)
- [`LogParamsLauncher`](#logparamslauncher)
<!-- /MarkdownTOC -->
<a id="install"></a>
## Install
```bash
pip install fromconfig_mlflow
```
<a id="quickstart"></a>
## Quickstart
To activate `MlFlow` login, simply add `--launcher.log=mlflow` to your command
```bash
fromconfig config.yaml params.yaml --launcher.log=mlflow - model - train
```
With
`model.py`
```python
"""Dummy Model."""
import mlflow
class Model:
def __init__(self, learning_rate: float):
self.learning_rate = learning_rate
def train(self):
print(f"Training model with learning_rate {self.learning_rate}")
if mlflow.active_run():
mlflow.log_metric("learning_rate", self.learning_rate)
```
`config.yaml`
```yaml
model:
_attr_: model.Model
learning_rate: "${params.learning_rate}"
```
`params.yaml`
```yaml
params:
learning_rate: 0.001
```
It should print
```
Started run: http://127.0.0.1:5000/experiments/0/runs/7fe650dd99574784aec1e4b18fceb73f
Training model with learning_rate 0.001
```
If you navigate to `http://127.0.0.1:5000/experiments/0/runs/7fe650dd99574784aec1e4b18fceb73f` you should see your the logged `learning_rate` metric.
<a id="mlflow-server"></a>
## MlFlow server
To setup a local MlFlow tracking server, run
```bash
mlflow server
```
which should print
```
[INFO] Starting gunicorn 20.0.4
[INFO] Listening at: http://127.0.0.1:5000
```
We will assume that the tracking URI is `http://127.0.0.1:5000` from now on.
<a id="configure-mlflow"></a>
## Configure MlFlow
You can set the tracking URI either via an environment variable or via the config.
To set the `MLFLOW_TRACKING_URI` environment variable
```bash
export MLFLOW_TRACKING_URI=http://127.0.0.1:5000
```
Alternatively, you can set the `mlflow.tracking_uri` config key either via command line with
```bash
fromconfig config.yaml params.yaml --launcher.log=mlflow --mlflow.tracking_uri="http://127.0.0.1:5000" - model - train
```
or in a config file with
`launcher.yaml`
```yaml
# Configure mlflow
mlflow:
# tracking_uri: "http://127.0.0.1:5000" # Or set env variable MLFLOW_TRACKING_URI
# experiment_name: "test-experiment" # Which experiment to use
# run_id: 12345 # To restore a previous run
# run_name: test # To give a name to your new run
# artifact_location: "path/to/artifacts" # Used only when creating a new experiment
# Configure launcher
launcher:
log: mlflow
```
and run
```bash
fromconfig config.yaml params.yaml launcher.yaml - model - train
```
<a id="artifacts-and-parameters"></a>
## Artifacts and Parameters
In this example, we add logging of the config and parameters.
Re-using the [quickstart](#quickstart) code, modify the `launcher.yaml` file
```yaml
# Configure logging
logging:
level: 20
# Configure mlflow
mlflow:
# tracking_uri: "http://127.0.0.1:5000" # Or set env variable MLFLOW_TRACKING_URI
# experiment_name: "test-experiment" # Which experiment to use
# run_id: 12345 # To restore a previous run
# run_name: test # To give a name to your new run
# artifact_location: "path/to/artifacts" # Used only when creating a new experiment
# include_keys: # Only log params that match *model*
# - model
# Configure launcher
launcher:
log:
- logging
- mlflow
parse:
- mlflow.log_artifacts
- parser
- mlflow.log_params
```
and run
```bash
fromconfig config.yaml params.yaml launcher.yaml - model - train
```
which prints
```
INFO:fromconfig_mlflow.launcher:Started run: http://127.0.0.1:5000/experiments/0/runs/<MLFLOW_RUN_ID>
Training model with learning_rate 0.001
```
If you navigate to the MlFlow run URL, you should see
- the parameters, a flattened version of the *parsed* config (`model.learning_rate` is `0.001` and not `${params.learning_rate}`)
- the original config, saved as `config.yaml`
- the parsed config, saved as `parsed.yaml`
<a id="usage-reference"></a>
## Usage-Reference
<a id="startrunlauncher"></a>
### `StartRunLauncher`
To configure MlFlow, add a `mlflow` entry to your config and set the following parameters
- `run_id`: if you wish to restart an existing run
- `run_name`: if you wish to give a name to your new run
- `tracking_uri`: to configure the tracking remote
- `experiment_name`: to use a different experiment than the custom
experiment
- `artifact_location`: the location of the artifacts (config files)
Additionally, the launcher can be initialized with the following attributes
- `set_env_vars`: if True (default is `True`), set `MLFLOW_RUN_ID` and `MLFLOW_TRACKING_URI`
- `set_run_id`: if True (default is `False`), set `mlflow.run_id` in config.
For example,
```yaml
# Configure logging
logging:
level: 20
# Configure mlflow
mlflow:
# tracking_uri: "http://127.0.0.1:5000" # Or set env variable MLFLOW_TRACKING_URI
# experiment_name: "test-experiment" # Which experiment to use
# run_id: 12345 # To restore a previous run
# run_name: test # To give a name to your new run
# artifact_location: "path/to/artifacts" # Used only when creating a new experiment
# Configure Launcher
launcher:
log:
- logging
- _attr_: mlflow
set_env_vars: true
set_run_id: true
```
<a id="logartifactslauncher"></a>
### `LogArtifactsLauncher`
The launcher can be initialized with the following attributes
- `path_command`: Name for the command file. If `None`, don't log the command.
- `path_config`: Name for the config file. If `None`, don't log the config.
For example,
```yaml
# Configure logging
logging:
level: 20
# Configure mlflow
mlflow:
# tracking_uri: "http://127.0.0.1:5000" # Or set env variable MLFLOW_TRACKING_URI
# experiment_name: "test-experiment" # Which experiment to use
# run_id: 12345 # To restore a previous run
# run_name: test # To give a name to your new run
# artifact_location: "path/to/artifacts" # Used only when creating a new experiment
# Configure launcher
launcher:
log:
- logging
- mlflow
parse:
- _attr_: mlflow.log_artifacts
path_command: launch.sh
path_config: config.yaml
- parser
- _attr_: mlflow.log_artifacts
path_command: null
path_config: parsed.yaml
```
<a id="logparamslauncher"></a>
### `LogParamsLauncher`
The launcher will use `include_keys` and `ignore_keys` if present in the config in the `mlflow` key.
- `ignore_keys` : If given, don't log some parameters that have some substrings.
- `include_keys` : If given, only log some parameters that have some substrings. Also shorten the flattened parameter to start at the first match. For example, if the config is `{"foo": {"bar": 1}}` and `include_keys=("bar",)`, then the logged parameter will be `"bar"`.
For example,
```yaml
# Configure logging
logging:
level: 20
# Configure mlflow
mlflow:
# tracking_uri: "http://127.0.0.1:5000" # Or set env variable MLFLOW_TRACKING_URI
# experiment_name: "test-experiment" # Which experiment to use
# run_id: 12345 # To restore a previous run
# run_name: test # To give a name to your new run
# artifact_location: "path/to/artifacts" # Used only when creating a new experiment
include_keys: # Only log params that match *model*
- model
# Configure launcher
launcher:
log:
- logging
- mlflow
parse:
- parser
- mlflow.log_params
```
Raw data
{
"_id": null,
"home_page": "https://github.com/criteo/fromconfig-mlflow",
"name": "fromconfig-mlflow",
"maintainer": "",
"docs_url": null,
"requires_python": "",
"maintainer_email": "",
"keywords": "",
"author": "Criteo",
"author_email": "",
"download_url": "https://files.pythonhosted.org/packages/74/54/93a1129ec92d05fe06cfe2903ae69fae43c741c160576e98b69c17551fc4/fromconfig_mlflow-0.4.0.tar.gz",
"platform": null,
"description": "# FromConfig MlFlow <!-- {docsify-ignore} -->\n[](https://pypi.python.org/pypi/fromconfig-mlflow)\n[](https://github.com/criteo/fromconfig-mlflow/actions?query=workflow%3A%22Continuous+integration%22)\n\nA [fromconfig](https://github.com/criteo/fromconfig) `Launcher` for [MlFlow](https://www.mlflow.org) support.\n\n\n<!-- MarkdownTOC -->\n\n- [Install](#install)\n- [Quickstart](#quickstart)\n- [MlFlow server](#mlflow-server)\n- [Configure MlFlow](#configure-mlflow)\n- [Artifacts and Parameters](#artifacts-and-parameters)\n- [Usage-Reference](#usage-reference)\n - [`StartRunLauncher`](#startrunlauncher)\n - [`LogArtifactsLauncher`](#logartifactslauncher)\n - [`LogParamsLauncher`](#logparamslauncher)\n\n<!-- /MarkdownTOC -->\n\n\n<a id=\"install\"></a>\n## Install\n\n```bash\npip install fromconfig_mlflow\n```\n\n<a id=\"quickstart\"></a>\n## Quickstart\n\nTo activate `MlFlow` login, simply add `--launcher.log=mlflow` to your command\n\n```bash\nfromconfig config.yaml params.yaml --launcher.log=mlflow - model - train\n```\n\nWith\n\n`model.py`\n\n```python\n\"\"\"Dummy Model.\"\"\"\n\nimport mlflow\n\n\nclass Model:\n def __init__(self, learning_rate: float):\n self.learning_rate = learning_rate\n\n def train(self):\n print(f\"Training model with learning_rate {self.learning_rate}\")\n if mlflow.active_run():\n mlflow.log_metric(\"learning_rate\", self.learning_rate)\n```\n\n`config.yaml`\n\n```yaml\nmodel:\n _attr_: model.Model\n learning_rate: \"${params.learning_rate}\"\n```\n\n`params.yaml`\n\n```yaml\nparams:\n learning_rate: 0.001\n```\n\nIt should print\n\n```\nStarted run: http://127.0.0.1:5000/experiments/0/runs/7fe650dd99574784aec1e4b18fceb73f\nTraining model with learning_rate 0.001\n```\n\nIf you navigate to `http://127.0.0.1:5000/experiments/0/runs/7fe650dd99574784aec1e4b18fceb73f` you should see your the logged `learning_rate` metric.\n\n<a id=\"mlflow-server\"></a>\n## MlFlow server\n\nTo setup a local MlFlow tracking server, run\n\n```bash\nmlflow server\n```\n\nwhich should print\n\n```\n[INFO] Starting gunicorn 20.0.4\n[INFO] Listening at: http://127.0.0.1:5000\n```\n\nWe will assume that the tracking URI is `http://127.0.0.1:5000` from now on.\n\n\n<a id=\"configure-mlflow\"></a>\n## Configure MlFlow\n\nYou can set the tracking URI either via an environment variable or via the config.\n\nTo set the `MLFLOW_TRACKING_URI` environment variable\n\n```bash\nexport MLFLOW_TRACKING_URI=http://127.0.0.1:5000\n```\n\nAlternatively, you can set the `mlflow.tracking_uri` config key either via command line with\n\n```bash\nfromconfig config.yaml params.yaml --launcher.log=mlflow --mlflow.tracking_uri=\"http://127.0.0.1:5000\" - model - train\n```\n\nor in a config file with\n\n`launcher.yaml`\n\n```yaml\n# Configure mlflow\nmlflow:\n # tracking_uri: \"http://127.0.0.1:5000\" # Or set env variable MLFLOW_TRACKING_URI\n # experiment_name: \"test-experiment\" # Which experiment to use\n # run_id: 12345 # To restore a previous run\n # run_name: test # To give a name to your new run\n # artifact_location: \"path/to/artifacts\" # Used only when creating a new experiment\n\n# Configure launcher\nlauncher:\n log: mlflow\n```\n\nand run\n\n```bash\nfromconfig config.yaml params.yaml launcher.yaml - model - train\n```\n\n<a id=\"artifacts-and-parameters\"></a>\n## Artifacts and Parameters\n\nIn this example, we add logging of the config and parameters.\n\nRe-using the [quickstart](#quickstart) code, modify the `launcher.yaml` file\n\n```yaml\n# Configure logging\nlogging:\n level: 20\n\n# Configure mlflow\nmlflow:\n # tracking_uri: \"http://127.0.0.1:5000\" # Or set env variable MLFLOW_TRACKING_URI\n # experiment_name: \"test-experiment\" # Which experiment to use\n # run_id: 12345 # To restore a previous run\n # run_name: test # To give a name to your new run\n # artifact_location: \"path/to/artifacts\" # Used only when creating a new experiment\n # include_keys: # Only log params that match *model*\n # - model\n\n# Configure launcher\nlauncher:\n log:\n - logging\n - mlflow\n parse:\n - mlflow.log_artifacts\n - parser\n - mlflow.log_params\n```\n\nand run\n\n```bash\nfromconfig config.yaml params.yaml launcher.yaml - model - train\n```\n\nwhich prints\n\n```\nINFO:fromconfig_mlflow.launcher:Started run: http://127.0.0.1:5000/experiments/0/runs/<MLFLOW_RUN_ID>\nTraining model with learning_rate 0.001\n```\n\nIf you navigate to the MlFlow run URL, you should see\n- the parameters, a flattened version of the *parsed* config (`model.learning_rate` is `0.001` and not `${params.learning_rate}`)\n- the original config, saved as `config.yaml`\n- the parsed config, saved as `parsed.yaml`\n\n\n<a id=\"usage-reference\"></a>\n## Usage-Reference\n\n<a id=\"startrunlauncher\"></a>\n### `StartRunLauncher`\n\nTo configure MlFlow, add a `mlflow` entry to your config and set the following parameters\n\n- `run_id`: if you wish to restart an existing run\n- `run_name`: if you wish to give a name to your new run\n- `tracking_uri`: to configure the tracking remote\n- `experiment_name`: to use a different experiment than the custom\n experiment\n- `artifact_location`: the location of the artifacts (config files)\n\nAdditionally, the launcher can be initialized with the following attributes\n\n- `set_env_vars`: if True (default is `True`), set `MLFLOW_RUN_ID` and `MLFLOW_TRACKING_URI`\n- `set_run_id`: if True (default is `False`), set `mlflow.run_id` in config.\n\nFor example,\n\n```yaml\n# Configure logging\nlogging:\n level: 20\n\n# Configure mlflow\nmlflow:\n # tracking_uri: \"http://127.0.0.1:5000\" # Or set env variable MLFLOW_TRACKING_URI\n # experiment_name: \"test-experiment\" # Which experiment to use\n # run_id: 12345 # To restore a previous run\n # run_name: test # To give a name to your new run\n # artifact_location: \"path/to/artifacts\" # Used only when creating a new experiment\n\n# Configure Launcher\nlauncher:\n log:\n - logging\n - _attr_: mlflow\n set_env_vars: true\n set_run_id: true\n```\n\n\n<a id=\"logartifactslauncher\"></a>\n### `LogArtifactsLauncher`\n\nThe launcher can be initialized with the following attributes\n\n- `path_command`: Name for the command file. If `None`, don't log the command.\n- `path_config`: Name for the config file. If `None`, don't log the config.\n\nFor example,\n\n```yaml\n# Configure logging\nlogging:\n level: 20\n\n# Configure mlflow\nmlflow:\n # tracking_uri: \"http://127.0.0.1:5000\" # Or set env variable MLFLOW_TRACKING_URI\n # experiment_name: \"test-experiment\" # Which experiment to use\n # run_id: 12345 # To restore a previous run\n # run_name: test # To give a name to your new run\n # artifact_location: \"path/to/artifacts\" # Used only when creating a new experiment\n\n# Configure launcher\nlauncher:\n log:\n - logging\n - mlflow\n parse:\n - _attr_: mlflow.log_artifacts\n path_command: launch.sh\n path_config: config.yaml\n - parser\n - _attr_: mlflow.log_artifacts\n path_command: null\n path_config: parsed.yaml\n```\n\n\n<a id=\"logparamslauncher\"></a>\n### `LogParamsLauncher`\n\nThe launcher will use `include_keys` and `ignore_keys` if present in the config in the `mlflow` key.\n\n- `ignore_keys` : If given, don't log some parameters that have some substrings.\n- `include_keys` : If given, only log some parameters that have some substrings. Also shorten the flattened parameter to start at the first match. For example, if the config is `{\"foo\": {\"bar\": 1}}` and `include_keys=(\"bar\",)`, then the logged parameter will be `\"bar\"`.\n\nFor example,\n\n```yaml\n# Configure logging\nlogging:\n level: 20\n\n# Configure mlflow\nmlflow:\n # tracking_uri: \"http://127.0.0.1:5000\" # Or set env variable MLFLOW_TRACKING_URI\n # experiment_name: \"test-experiment\" # Which experiment to use\n # run_id: 12345 # To restore a previous run\n # run_name: test # To give a name to your new run\n # artifact_location: \"path/to/artifacts\" # Used only when creating a new experiment\n include_keys: # Only log params that match *model*\n - model\n\n# Configure launcher\nlauncher:\n log:\n - logging\n - mlflow\n parse:\n - parser\n - mlflow.log_params\n```\n\n\n",
"bugtrack_url": null,
"license": "",
"summary": "# FromConfig MlFlow <!-- {docsify-ignore} -->",
"version": "0.4.0",
"split_keywords": [],
"urls": [
{
"comment_text": "",
"digests": {
"md5": "0a65d49455c60e6ce3969dd9f643d30d",
"sha256": "7d2b883fbec7f7c1c4ffd26ba0cb383eecf4b5d69819672476b40fbe388d6c84"
},
"downloads": -1,
"filename": "fromconfig_mlflow-0.4.0.tar.gz",
"has_sig": false,
"md5_digest": "0a65d49455c60e6ce3969dd9f643d30d",
"packagetype": "sdist",
"python_version": "source",
"requires_python": null,
"size": 8855,
"upload_time": "2022-12-22T11:34:20",
"upload_time_iso_8601": "2022-12-22T11:34:20.796510Z",
"url": "https://files.pythonhosted.org/packages/74/54/93a1129ec92d05fe06cfe2903ae69fae43c741c160576e98b69c17551fc4/fromconfig_mlflow-0.4.0.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2022-12-22 11:34:20",
"github": true,
"gitlab": false,
"bitbucket": false,
"github_user": "criteo",
"github_project": "fromconfig-mlflow",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"requirements": [],
"lcname": "fromconfig-mlflow"
}
JSON
