Describing your model with metadata

The meta_data block in your MLproject file controls how your model is displayed in the Chap Modeling App and catalogued in the Chap database (the name shown to users, author, organization logo, citation, etc.). All fields are optional — defaults are used when a field is omitted — but filling them in makes your model easier to recognize and trust.

Supported fields

Field	Type	Default	Purpose
display_name	string	"No Display Name yet"	Human-readable name shown in the UI
description	string	"No Description yet"	One- or two-paragraph summary of what the model does
author	string	"Unknown Author"	Person or team credited as the model author
author_note	string	"No Author note yet"	Free-form note from the author (caveats, intended use, etc.)
author_assessed_status	enum	red	Author's own maturity rating: gray, red, orange, yellow, green (see below)
organization	string	–	Organization the author belongs to
organization_logo_url	URL	–	Public URL to an image used as the organization logo in the UI
contact_email	string	–	Contact address for questions about the model
citation_info	string	–	How users should cite the model
documentation_url	URL	–	Link to external documentation for the model

author_assessed_status values

The author's own assessment of how mature the model is. This is shown in the UI as a colored badge.

Value	Meaning
gray	Not intended for use, deprecated, or legacy-only
red	Highly experimental prototype, not validated
orange	Shows promise on limited data; needs manual configuration and careful evaluation
yellow	Ready for more rigorous testing
green	Validated and ready for use

Example

name: my_model

uv_env: pyproject.toml

entry_points:
  train:
    parameters:
      train_data: str
      model: str
    command: "python train.py {train_data} {model}"
  predict:
    parameters:
      historic_data: str
      future_data: str
      model: str
      out_file: str
    command: "python predict.py {model} {historic_data} {future_data} {out_file}"

meta_data:
  display_name: "Monthly Deep Auto Regressive"
  description: >
    Experimental deep learning model based on an RNN architecture,
    focusing on predictions from auto-regressive time series data.
  author: "Knut Rand"
  author_assessed_status: orange
  organization: "HISP Centre, University of Oslo"
  organization_logo_url: "https://landportal.org/sites/default/files/2024-03/university_of_oslo_logo.png"
  contact_email: "knut.rand@dhis2.org"
  citation_info: >
    Rand, Knut. 2025. "Monthly Deep Auto Regressive model".
    HISP Centre, University of Oslo.
  documentation_url: "https://dhis2-chap.github.io/chap-core/external_models/overview_of_supported_models.html"

For full real-world MLproject files, see example_data/model_templates/ar_monthly.yaml and example_data/model_templates/chap_ewars_monthly.yaml.