Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 4 additions & 0 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -93,6 +93,10 @@ checkpoint behind it is documented. Every model checkpoint published to the
**[model-card standard](docs/model-cards.md)** (start from the
[template](docs/model-card-template.md); the standard includes a gating checklist).

Published checkpoints are catalogued in **[WorldPolicies](https://github.com/robocurve/worldpolicies)**
(CI-validated registry + [🤗 Collection](https://huggingface.co/collections/robocurve/worldpolicies-6a4dc8fd556a82aeea0fca37)) —
WorldEvals benchmarks them; WorldPolicies is where they live.

## Development

> **Dependency changes:** after editing dependencies in `pyproject.toml`, run
Expand Down
1 change: 1 addition & 0 deletions docs/model-card-template.md
Original file line number Diff line number Diff line change
Expand Up @@ -35,6 +35,7 @@ inference: false
[![Evaluate with Inspect Robots](https://img.shields.io/badge/evaluate%20with-Inspect%20Robots-indigo)](https://github.com/robocurve/inspect-robots)
[![Adapters: <embodiment>](https://img.shields.io/badge/adapters-inspect--robots--<embodiment>-blue)](https://github.com/robocurve/inspect-robots-<embodiment>)
[![Benchmarks: WorldEvals](https://img.shields.io/badge/benchmarks-WorldEvals-2ea44f)](https://worldevals.org)
[![Catalog: WorldPolicies](https://img.shields.io/badge/catalog-WorldPolicies-yellow)](https://huggingface.co/collections/robocurve/worldpolicies-6a4dc8fd556a82aeea0fca37)

<≤3 sentences: [base model](link) + method (LoRA/full FT; what's frozen) + robot +
[dataset](link). Adapter/merge status and where raw adapters live.>
Expand Down
11 changes: 10 additions & 1 deletion docs/model-cards.md
Original file line number Diff line number Diff line change
Expand Up @@ -177,7 +177,15 @@ The section that makes a robot checkpoint usable at all:
- Training instabilities/failed attempts, or a link to the repo's incident log — a reader
choosing hyperparameters learns more from "lr 6e-4 NaN'd at step 2.5k" than from the winner.

### 9. Versioning & contact (REQUIRED)
### 9. Cataloguing (REQUIRED)

At publish time, register the checkpoint in
[WorldPolicies](https://github.com/robocurve/worldpolicies): add it to the
[🤗 Collection](https://huggingface.co/collections/robocurve/worldpolicies-6a4dc8fd556a82aeea0fca37) and append a
`Policy(...)` entry to its CI-validated `catalog.py` (see that repo's CLAUDE.md).
Add the catalog badge to the card's badge row (see template).

### 10. Versioning & contact (REQUIRED)

- If this card supersedes a checkpoint, set `new_version:` frontmatter on the old repo.
- State whether `main` is stable; recommend pinning by revision hash for reproduction.
Expand Down Expand Up @@ -225,5 +233,6 @@ Body
- [ ] Caveats: per-source data licensing statement
- [ ] Caveats: known data issues + eval-vs-deployment limits
- [ ] Caveats: instabilities/failed attempts included or linked
- [ ] Checkpoint registered in WorldPolicies (Collection + catalog.py entry + catalog badge)
- [ ] Versioning + contact stated
- [ ] Card written from named run artifacts at publish time
Loading