Skip to content
Edit on GitHub

dvc.api.params_show()

Load parameters (name and values) tracked in a DVC project.

def params_show(
    *targets: str,  # Optional
    stages: Optional[Union[str, Iterable[str]]] = None,
    repo: Optional[str] = None,
    rev: Optional[str] = None,
    deps: bool = False,
) -> Dict:

Usage:

import dvc.api

params = dvc.api.params_show()

Description

Retrieves parameter keys and values from a DVC project and returns a dictionary, such as:

{
  "split": 0.2,
  "seed": 20170428
}

Without arguments, this function will retrieve all parameters from all tracked param files (used in any dvc.yaml file). This applies to the current project version when using Git (including any changes in the working tree).

The function parameters (below) let you restrict what's retrieved.

Parameters

  • *targets - one or more separate path(s) to valid parameter file(s) to retrieve params from, for example "params.py", "feat/params.toml". If no targets are provided, all param files tracked in any dvc.yaml will be targeted by default. Note that explicit targets don't have to be in a dvc.yaml (unless deps=True).

  • repo - specifies the location of the DVC project. It can be a URL or a file system path. Both HTTP and SSH protocols are supported for online Git repos (e.g. [user@]server:project.git). Default: The current project (found by walking up from the current working directory tree).

  • stages - one or more names of the stage(s) to retrieve params from. Default: None (all parameters from all stages will be retrieved). If this method is called from a different location to the one where the dvc.yaml is found, the relative path to the dvc.yaml must be provided as a prefix with the syntax {relpath}:{stage}. For example: subdir/dvc.yaml:stage-0 or ../dvc.yaml:stage-1.

  • rev - Git commit (any revision such as a branch or tag name, a commit hash or an experiment name). If repo is not a Git repo, this option is ignored. Default: None (current working tree will be used)

  • deps - whether to retrieve only params that are stage dependencies. Accepts True or False (default).

Example: Filter by stage name(s)

stages can be a single name (string):

import dvc.api
params = dvc.api.params_show(stages="train")
{
  "n_est": 50,
  "min_split": 0.01
}

Or an iterable of strings:

import dvc.api
params = dvc.api.params_show(stages=["featurize", "train"])
{
  "max_features": 200,
  "ngrams": 2
  "n_est": 50,
  "min_split": 0.01
}

The returned dictionary can be used inside the stage:

clf = RandomForestClassifier(
    n_estimators=params["n_est"],
    min_samples_split=params["min_split"]
)

Example: Load specific parameter file(s)

You can pass any valid param file path as target to load all of the parameters defined in it:

import dvc.api
params = dvc.api.params_show( "configs/params_dev.yaml")
{
  "run_mode": "prod",
  "configs": {
    "dev": "configs/params_dev.yaml",
    ...

Or multiple path targets:

import dvc.api
params = dvc.api.params_show(
  "configs/params_dev.yaml", "configs/params_prod.yaml")
{
  "configs/params_prod.yaml:run_mode": "prod",
  "configs/params_prod.yaml:config_file": "configs/params_prod.yaml",
  ...
  "configs/params_dev.yaml:run_mode": "dev",
  "configs/params_dev.yaml:config_file": "configs/params_dev.yaml",
  ...

Example: Use a remote DVC repository

You can use the repo argument to retrieve parameters from any DVC repository without having to clone it locally.

import dvc.api
params = dvc.api.params_show(
    repo="https://github.com/iterative/example-get-started")
{
  "prepare": {
    "split": 0.2,
    "seed": 20170428
  },
  "featurize": {
    "max_features": 200,
    "ngrams": 2
  },
  "train": {
    "seed": 20170428,
    "n_est": 50,
    "min_split": 0.01
  }
}

Example: Specify a project version

You can retrieve params from arbitrary Git commits, for example a branch name:

import json
import dvc.api
params = dvc.api.params_show(rev="tune-hyperparams")
{
  "prepare": {
    "split": 0.2,
    "seed": 20170428
  },
  "featurize": {
    "max_features": 200,
    "ngrams": 2
  },
  "train": {
    "seed": 20170428,
    "n_est": 100,
    "min_split": 8
  }
}