Add OpenDDE to model libraries

[mrzzmrzz]

Adds OpenDDE to the model libraries.

OpenDDE is an open-source, all-atom biomolecular foundation model that turns co-folding into a scalable engine for structure prediction, design, and optimization in drug discovery (proteins, DNA, RNA, ligands, ions).

• PyPI: https://pypi.org/project/opendde/ (pip install opendde, Apache-2.0)
• Code: https://github.com/aurekaresearch/OpenDDE
• Docs: https://huggingface.co/aurekaresearch/OpenDDE/blob/main/docs/inference_instructions.md
• Weights: https://huggingface.co/aurekaresearch/OpenDDE

countDownloads

The opendde package resolves its checkpoints straight from the Hub and fetches them with a plain HTTP GET against the resolve endpoint (see opendde/config/model_manifest.py):

https://huggingface.co/aurekaresearch/OpenDDE/resolve/<revision>/opendde.pt

It never requests config.json, so the default query files don't reflect real usage. path_extension:"pt" matches the two released checkpoints (opendde.pt, opendde_abag.pt) and future ones, while excluding the large non-weight assets that live in the same repo (common/components.cif, common/components.cif.rdkit_mol.pkl), so downloads aren't double counted.

library_name: opendde is already set on the model card.

filter: false since this is a single-repo library.

---

Note

Low Risk
Additive registry and snippet-only changes with no auth, API, or shared runtime impact.

Overview
Registers OpenDDE (library_name: opendde) so Hub model pages show install/inference guidance and download stats align with how weights are fetched.

A new opendde entry in MODEL_LIBRARIES_UI_ELEMENTS points at the Aureka Research repo and docs, uses filter: false (single-repo library), and sets countDownloads to path_extension:"pt" because the CLI pulls .pt checkpoints via Hub resolve rather than standard config.json files.

Model pages get a static CLI snippet (pip install 'opendde[gpu]', opendde doctor, opendde pred …) via a new opendde() export in model-libraries-snippets.ts.

^{Reviewed by Cursor Bugbot for commit 1ba985e. Bugbot is set up for automated code reviews on this repo. Configure here.}

[Copilot]

Pull request overview

Registers OpenDDE (library_name: opendde) as a model library so Hugging Face model pages can display library metadata (repo/docs), show an install/inference snippet, and compute download counts in a way that matches how OpenDDE distributes weights.

Changes:

• Added an opendde entry to MODEL_LIBRARIES_UI_ELEMENTS with repo/docs links, filter: false, and countDownloads: path_extension:"pt".
• Added a static opendde() snippet showing install + opendde doctor + a sample opendde pred … invocation.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
packages/tasks/src/model-libraries.ts	Adds the OpenDDE library registry entry used for UI metadata and download counting.
packages/tasks/src/model-libraries-snippets.ts	Adds the OpenDDE CLI snippet displayed on model pages.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[mrzzmrzz]

Some additional context, in case it helps with review.

The Hub teams have already engaged with this model directly:

• The open-source team built a demo Space for it and offered to transfer it over (discussion #1). That transfer is done and it now runs under our org on ZeroGPU: https://huggingface.co/spaces/aurekaresearch/opendde-structure-prediction
• The science team reached out about featuring it on Hugging Science (discussion #2).

That Space is also relevant to this PR specifically: it loads the weights with

hf_hub_download(MODEL_REPO, "opendde.pt", repo_type="model")

so it's a second consumer that pulls opendde.pt directly and is invisible to the default query files today. Both it and the opendde PyPI package would be covered by the path_extension:"pt" rule proposed here.

[Wauplin]

Looks good thank you!

Expect 2-3 days before getting it live on the Hub 🤗