add model card and artifact metadata

README.md

@@ -1 +1,28 @@
-CLIP encoder weights for DreamZero
+---
+license: other
+tags:
+- robotics
+- vla
+- clip
+- checkpoint
+---
+
+# CLIP Encoder Checkpoint
+
+This repo stores the CLIP encoder checkpoint used by the VLA SFT deployment suite.
+
+## Files
+
+| File | Size |
+|---|---:|
+| `models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth` | `2,528,465,537 bytes` |
+
+## Usage
+
+The recommended entrypoint is `poet70/vla-sft-suite`; its `scripts/download_artifacts.sh` downloads this checkpoint automatically for DreamZero runs.
+
+Manual download:
+
+```bash
+hf download poet70/clip-encoder --repo-type model --local-dir ./clip-encoder
+```