Skip to content

Repo

Currently, the following actions are supported for repos:

Syntax Variations

In MLCFlow, repos can be identified in different ways:

  1. Using MLC repo folder name format: <repoowner@reponame> (e.g.,mlcommons@mlperf-automations)
  2. Using alias: <repo_alias> (e.g., mlcommons@mlperf-automations)
  3. Using UID: <repo_uid> (e.g., 9cf241afa6074c89)
  4. Using both alias and UID: <repo_alias>,<repo_uid> (e.g., mlcommons@mlperf-automations,9cf241afa6074c89)
  5. Using URL: <repo_url> (e.g., https://github.com/mlcommons/mlperf-automations)

Note

  • repo uid and repo alias for a particular MLC repository can be found inside meta.yml file.
  • For simplicity, syntax variations are only shown for the find action, but similar options apply to all other actions.

Find

find action retrieves the path of a specific repository registered in MLCFlow.

Syntax Variations

Command Format Example Usage
mlc find repo <repo_owner@repo_name> mlc find repo mlcommons@mlperf-automations
mlc find repo <repo_alias> mlc find repo mlcommons@mlperf-automations
mlc find repo <repo_uid> mlc find repo 9cf241afa6074c89
mlc find script <repo_alias>,<repo_uid> mlc find repo mlcommons@mlperf-automations,9cf241afa6074c89
mlc find repo <repo_url> mlc find repo https://github.com/mlcommons/mlperf-automations

Examples of find action for repo target can be found inside the GitHub action workflow here.

Example Output 📌 
anandhu@anandhu-VivoBook-ASUSLaptop-X515UA-M515UA:~$ mlc find repo mlcommons@mlperf-automations
[2025-02-19 15:32:18,352 main.py:1737 INFO] - Item path: /home/anandhu/MLC/repos/mlcommons@mlperf-automations

Add

add action is used to create a new MLC repo and register in MLCFlow. The newly created repo folder will be present inside the repos folder within the parent MLC directory.

Example Command

mlc add repo mlcommons@script-automations
Example Output 📌 
  anandhu@anandhu-VivoBook-ASUSLaptop-X515UA-M515UA:~$ mlc add repo mlcommons@script-automations
  [2025-02-19 16:34:37,570 main.py:1085 INFO] - New repo path: /home/anandhu/MLC/repos/mlcommons@script-automations
  [2025-02-19 16:34:37,573 main.py:1126 INFO] - Added new repo path: /home/anandhu/MLC/repos/mlcommons@script-automations
  [2025-02-19 16:34:37,573 main.py:1130 INFO] - Updated repos.json at /home/anandhu/MLC/repos/repos.json

Examples of add action for repo target could be found inside the GitHub action workflow here.

Note

  • repo_uid is not supported in the add action for repo target since uid for the repo is assigned automatically while creating the repository.

Pull

pull action clones an MLC repository and registers it in MLC.

If the repository already exists locally in MLC repos directory, it fetches the latest changes if there are no uncommited modifications(does not include untracked files/folders). The pull action could be also used to checkout to a particular branch or release tag with flags --checkout and --tag.

Example Command

mlc pull repo mlcommons@mlperf-automations
Example Output 📌 
  anandhu@anandhu-VivoBook-ASUSLaptop-X515UA-M515UA:~$ mlc pull repo mlcommons@mlperf-automations
  [2025-02-19 16:46:27,208 main.py:1260 INFO] - Cloning repository https://github.com/mlcommons/mlperf-automations.git to /home/anandhu/MLC/repos/mlcommons@mlperf-automations...
  Cloning into '/home/anandhu/MLC/repos/mlcommons@mlperf-automations'...
  remote: Enumerating objects: 77610, done.
  remote: Counting objects: 100% (2199/2199), done.
  remote: Compressing objects: 100% (1103/1103), done.
  remote: Total 77610 (delta 1616), reused 1109 (delta 1095), pack-reused 75411 (from 2)
  Receiving objects: 100% (77610/77610), 18.36 MiB | 672.00 KiB/s, done.
  Resolving deltas: 100% (53818/53818), done.
  [2025-02-19 16:46:57,604 main.py:1288 INFO] - Repository successfully pulled.
  [2025-02-19 16:46:57,605 main.py:1289 INFO] - Registering the repo in repos.json
  [2025-02-19 16:46:57,605 main.py:1126 INFO] - Added new repo path: /home/anandhu/MLC/repos/mlcommons@mlperf-automations
  [2025-02-19 16:46:57,606 main.py:1130 INFO] - Updated repos.json at /home/anandhu/MLC/repos/repos.json
  • The --checkout flag can be used if a user needs to check out a specific commit or branch after cloning. The user must provide the commit SHA if they want to check out a specific commit. This flag can be used in cases where the repository exists locally.
  • The --branch flag can be used if a user needs to check out a specific branch after cloning. The user must provide the branch name. This flag will only work when cloning a new repository.
  • The --tag flag can be used to check out a particular release tag.
  • --pat=<access_token> or --ssh flag can be used to clone a private repository.

Examples of pull action for repo target could be found inside the GitHub action workflow here.

Note

  • repo_uid and repo_alias are not supported in the pull action for the repo target.
  • Only one of --checkout, --branch, or --tag should be specified when using this action.

List

list action displays all registered MLC repositories along with their aliases and paths.

Example Command

mlc list repo
Example Output 📌 
  anandhu@anandhu-VivoBook-ASUSLaptop-X515UA-M515UA:~$ mlc list repo
  [2025-02-19 16:56:31,847 main.py:1349 INFO] - Listing all repositories.

  Repositories:
  -------------
  - Alias: local
  Path:  /home/anandhu/MLC/repos/local

  -  Alias: mlcommons@mlperf-automations
  Path:  /home/anandhu/MLC/repos/mlcommons@mlperf-automations

  -------------
  [2025-02-19 16:56:31,850 main.py:1356 INFO] - Repository listing ended

Example of list action for repo target could be found inside the GitHub action workflow here.

Rm (Remove)

rm action removes a specified repository from MLCFlow, deleting both the repo folder and its registration. If there are any modified local changes, the user will be prompted for confirmation unless the -f flag is used to force removal.

Example Command

mlc rm repo mlcommons@mlperf-automations
Example Output 
    anandhu@anandhu-VivoBook-ASUSLaptop-X515UA-M515UA:~$ mlc rm repo mlcommons@mlperf-automations
    [2025-02-19 17:01:59,483 main.py:1360 INFO] - rm command has been called for repo. This would delete the repo folder and unregister the repo from repos.json
    [2025-02-19 17:01:59,521 main.py:1380 INFO] - No local changes detected. Removing repo...
    [2025-02-19 17:01:59,581 main.py:1384 INFO] - Repo mlcommons@mlperf-automations residing in path /home/anandhu/MLC/repos/mlcommons@mlperf-automations has been successfully removed
    [2025-02-19 17:01:59,581 main.py:1385 INFO] - Checking whether the repo was registered in repos.json
    [2025-02-19 17:01:59,581 main.py:1134 INFO] - Unregistering the repo in path /home/anandhu/MLC/repos/mlcommons@mlperf-automations
    [2025-02-19 17:01:59,581 main.py:1144 INFO] - Path: /home/anandhu/MLC/repos/mlcommons@mlperf-automations has been removed.
An example of the `rm` action for the `repo` target can be found in the GitHub Actions workflow [here](https://github.com/mlcommons/mlcflow/blob/d0269b47021d709e0ffa7fe0db8c79635bfd9dff/.github/workflows/test-mlc-core-actions.yaml).