Spaces:
Running on T4
Running on T4
Commit ·
043867f
1
Parent(s): f67c98c
Update Space README: nvidia/ ownership + cpu-upgrade guidance
Browse files
README.md
CHANGED
|
@@ -7,4 +7,34 @@ sdk: docker
|
|
| 7 |
pinned: false
|
| 8 |
---
|
| 9 |
|
| 10 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7 |
pinned: false
|
| 8 |
---
|
| 9 |
|
| 10 |
+
# SimReady Validator
|
| 11 |
+
|
| 12 |
+
Validates a HuggingFace dataset against a [SimReady](https://github.com/NVIDIA/simready-foundation) profile.
|
| 13 |
+
|
| 14 |
+
Reads the dataset directly from HF storage (no copy onto external
|
| 15 |
+
infrastructure), runs the bundled
|
| 16 |
+
[`simready-report`](./tools/validation/plugins/simready-report/) validator
|
| 17 |
+
with `--no-use-kit`, and — when the **Open PR** checkbox is enabled —
|
| 18 |
+
uploads the verdict to the dataset as a `validation/` pull request.
|
| 19 |
+
|
| 20 |
+
## Source
|
| 21 |
+
|
| 22 |
+
This Space is generated from the
|
| 23 |
+
[`NVIDIA-dev/simready-oem-library-pm`](https://github.com/NVIDIA-dev/simready-oem-library-pm)
|
| 24 |
+
internal repo (see `tools/hf_space/`). Bug reports and feature requests go
|
| 25 |
+
there; this Space is the deployable artifact, not the source of truth.
|
| 26 |
+
|
| 27 |
+
## Hardware
|
| 28 |
+
|
| 29 |
+
`cpu-basic` is enough for a smoke test (no Kit re-exec on this Space).
|
| 30 |
+
For routine validation of partner datasets at scale, upgrade to
|
| 31 |
+
`cpu-upgrade` (8 vCPU, 32 GB) — the validator's `ProcessPoolExecutor`
|
| 32 |
+
actually scales there. Kit-rooted PhysX/MDL rule coverage needs a GPU
|
| 33 |
+
tier (`a10g-small`) and Isaac Sim baked into the image, neither of
|
| 34 |
+
which is in the current image.
|
| 35 |
+
|
| 36 |
+
## Secrets
|
| 37 |
+
|
| 38 |
+
| Name | Purpose |
|
| 39 |
+
|---|---|
|
| 40 |
+
| `HF_TOKEN` | Write-scoped Hugging Face token used to open verdict PRs on customer datasets. Set under Settings → Variables and secrets. |
|