Add support for modelFile generate via model providers model_url #329

avinashsingh77 · 2025-11-11T10:03:55Z

This PR adds support for providing model_url to modctl modelfile generate
Tests added, run output:

go test ./pkg/hfhub -v
=== RUN   TestParseModelURL
=== RUN   TestParseModelURL/full_URL
=== RUN   TestParseModelURL/full_URL_with_trailing_slash
=== RUN   TestParseModelURL/short_form
=== RUN   TestParseModelURL/http_URL
=== RUN   TestParseModelURL/invalid_format_-_missing_repo
=== RUN   TestParseModelURL/invalid_format_-_only_owner
=== RUN   TestParseModelURL/empty_URL
=== RUN   TestParseModelURL/URL_with_spaces_(trimmed)
--- PASS: TestParseModelURL (0.00s)
    --- PASS: TestParseModelURL/full_URL (0.00s)
    --- PASS: TestParseModelURL/full_URL_with_trailing_slash (0.00s)
    --- PASS: TestParseModelURL/short_form (0.00s)
    --- PASS: TestParseModelURL/http_URL (0.00s)
    --- PASS: TestParseModelURL/invalid_format_-_missing_repo (0.00s)
    --- PASS: TestParseModelURL/invalid_format_-_only_owner (0.00s)
    --- PASS: TestParseModelURL/empty_URL (0.00s)
    --- PASS: TestParseModelURL/URL_with_spaces_(trimmed) (0.00s)
PASS
ok      github.com/modelpack/modctl/pkg/hfhub   (cached)

Closes #308

gemini-code-assist · 2025-11-11T10:04:14Z

Summary of Changes

Hello @avinashsingh77, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request introduces a significant new capability to the modctl tool, allowing users to generate modelfiles directly from models hosted on Hugging Face Hub. By providing a --model_url flag to the modctl modelfile generate command, users can now specify a Hugging Face model, which the tool will automatically download and then use to create the modelfile. This integration simplifies the workflow for developers leveraging Hugging Face models, making it easier to manage and deploy them within the modctl ecosystem.

Highlights

Hugging Face Model URL Support: The modctl modelfile generate command now supports generating modelfiles directly from Hugging Face model URLs, using a new --model_url flag.
Flexible Input: Users can provide either a local directory path or a Hugging Face model URL (e.g., "owner/repo" or full "https://huggingface.co/owner/repo" format).
Hugging Face Integration: A new pkg/hfhub package has been introduced to handle parsing Hugging Face URLs, checking user authentication, and downloading models using the huggingface-cli tool.
Command Line Enhancements: The generate command's usage, short description, long description, and examples have been updated to reflect the new functionality, and argument validation now accommodates the optional path when --model_url is used.
Robust URL Parsing: Comprehensive unit tests have been added for the ParseModelURL function, ensuring correct handling of various URL formats and edge cases.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	`/gemini review`	Performs a code review for the current pull request in its current state.
Pull Request Summary	`/gemini summary`	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	`/gemini help`	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

gemini-code-assist

Code Review

This pull request introduces a valuable feature for generating a modelfile directly from a Hugging Face model URL. The implementation is well-structured, primarily by leveraging the huggingface-cli tool. My review focuses on enhancing user experience, ensuring proper resource management, and improving code clarity. I've suggested making the <path> argument and the --model_url flag mutually exclusive to prevent user confusion, adding cleanup for temporary directories, refining output handling within the hfhub package, and simplifying test logic for better maintainability.

cmd/modelfile/generate.go

pkg/hfhub/download.go

pkg/modelprovider/huggingface/downloader.go

pkg/hfhub/download_test.go

chlins

Can we abstract a generic interface for the download part, and then support different drivers or vendors? Because in addition to Hugging Face, there are also model hubs like ModelScope.

avinashsingh77 · 2025-11-19T17:04:23Z

Make sense. What all model vendors do we plan to support to start with?

…

On Wed, 19 Nov 2025 at 1:47 PM, Chlins Zhang ***@***.***> wrote: ***@***.**** commented on this pull request. Can we abstract a generic interface for the download part, and then support different drivers or vendors? Because in addition to Hugging Face, there are also model hubs like ModelScope. — Reply to this email directly, view it on GitHub <#329 (review)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AZR3OPKI24M24EMC26FQON335QRR7AVCNFSM6AAAAACLYFAAOKVHI2DSMVQWIX3LMV43YUDVNRWFEZLROVSXG5CSMV3GSZLXHMZTIOBRGM3TQMZSGY> . You are receiving this because you were mentioned.Message ID: ***@***.***>

Signed-off-by: Avinash Singh <[email protected]>

avinashsingh77 · 2025-11-24T10:23:59Z

@chlins Implemented a provider interface and added providers for HuggingFace and ModelScope as suggested. Does it look okay now?

chlins · 2025-11-25T02:43:57Z

@chlins Implemented a provider interface and added providers for HuggingFace and ModelScope as suggested. Does it look okay now?

@avinashsingh77 Great work! thanks for adding the provider interface with HuggingFace and ModelScope implementations. I'll take a look at the changes in a bit and share my thoughts.

cmd/modelfile/generate.go

pkg/modelprovider/provider.go

pkg/modelprovider/registry.go

chlins · 2025-11-25T07:04:11Z

cmd/modelfile/generate.go

+
+		// Create a temporary directory for downloading the model
+		// Clean up the temporary directory after the function returns
+		tmpDir, err := os.MkdirTemp("", "modctl-model-downloads-*")


Do we need to consider expose this temp dir to user? because in some limited environment, user may only have write access to specific dirs.

Do you want users to specify which directory should be used to download models directly? As we are providing "" as the dir parameter's value to os.MkdirTemp , MkdirTemp uses the default directory for temporary files, as returned by TempDir.

In some scenarios, the disk space of the temporary directory may not be large enough to store all model files.

pkg/modelprovider/huggingface/downloader.go

pkg/modelprovider/modelscope/downloader.go

Signed-off-by: Avinash Singh <[email protected]>

chlins · 2025-12-01T03:24:53Z

pkg/modelprovider/huggingface/provider.go

+
+	// Use huggingface-cli to download the model
+	// The --local-dir-use-symlinks=False flag ensures files are copied, not symlinked
+	cmd := exec.CommandContext(ctx, "huggingface-cli", "download", repoID, "--local-dir", downloadPath, "--local-dir-use-symlinks", "False")


Does this mean that we must rely on the CLI to download the model?

Yes @chlins , do you think its not a good design?

I'm not sure whether having to rely on external third-party command-line tools is convenient for users, and there might also be compatibility issues related to different versions. However, I think for the first version we can simplify things and just rely on the command line for now. In phase 2, we can consider whether it's possible to implement this directly via API calls in the future.

gemini-code-assist bot reviewed Nov 11, 2025

View reviewed changes

avinashsingh77 force-pushed the add-hf-support branch from 4f37fb2 to fb50643 Compare November 11, 2025 10:23

chlins reviewed Nov 19, 2025

View reviewed changes

avinashsingh77 and others added 3 commits November 24, 2025 15:26

Add support for HF model_url

63558c7

Signed-off-by: Avinash Singh <[email protected]>

optimise as per gemini's review

643b5a1

Signed-off-by: Avinash Singh <[email protected]>

add modelprovider interface and providers

6d7fb90

Signed-off-by: Avinash Singh <[email protected]>

avinashsingh77 force-pushed the add-hf-support branch from 1a81c96 to 6d7fb90 Compare November 24, 2025 09:58

avinashsingh77 changed the title ~~Add support for modelFile generate via HF model_url~~ Add support for modelFile generate via model providers model_url Nov 24, 2025

chlins requested changes Nov 25, 2025

View reviewed changes

avinashsingh77 and others added 2 commits November 26, 2025 18:33

optimise code

f455aed

Signed-off-by: Avinash Singh <[email protected]>

Merge branch 'main' into add-hf-support

7bc9afd

chlins mentioned this pull request Dec 1, 2025

Support mlflow registry #342

Open

chlins reviewed Dec 1, 2025

View reviewed changes

Add support for modelFile generate via model providers model_url #329

Are you sure you want to change the base?

Add support for modelFile generate via model providers model_url #329

Conversation

avinashsingh77 commented Nov 11, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

gemini-code-assist bot commented Nov 11, 2025

Summary of Changes

Highlights

Footnotes

Uh oh!

gemini-code-assist bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

chlins left a comment

Choose a reason for hiding this comment

Uh oh!

avinashsingh77 commented Nov 19, 2025 via email

Uh oh!

avinashsingh77 commented Nov 24, 2025

Uh oh!

chlins commented Nov 25, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

chlins Nov 25, 2025

Choose a reason for hiding this comment

Uh oh!

avinashsingh77 Nov 26, 2025

Choose a reason for hiding this comment

Uh oh!

chlins Dec 12, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

chlins Dec 1, 2025

Choose a reason for hiding this comment

Uh oh!

avinashsingh77 Dec 1, 2025

Choose a reason for hiding this comment

Uh oh!

chlins Dec 12, 2025

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

avinashsingh77 commented Nov 11, 2025 •

edited

Loading