Synchronize maintained vision bundle layout and manifest

README.md

@@ -44,6 +44,7 @@ merged `(1, 256, 5120)` and projected `(1, 256, 6144)`.
 ## Usage
 
 See [`examples/inference.py`](examples/inference.py) for image feature extraction.
+Forward images independently to avoid attention coupling in packed batches.
 Forward one image per call: the native HF attention couples images in a single call.
 
 ## Validation

export_manifest.json

@@ -17,7 +17,7 @@
     "LICENSE": "8c4978e3e0fe0349ff929cc8450463cba3039c4c9fc144d8bd5382ab00df247a",
     "LICENSE-CODE": "149f05ce186198084c9abd6af8e76e108ba7e507446a19ee5a2ffd694f69f525",
     "NOTICE": "19d19379d439b6dc2614476bb50b170fbe382a58f3a84ec67ff9bcd4e212d626",
-    "README.md": "09a5272c4f826c14ce19ac33fbc77875be80452e530e7f56e6316fddf2754281",
+    "README.md": "8e22a8fa2a0afdcaa1409f9a77366c1ba2e7bb1d6a4dae1a144627f068cc1e41",
     "config.json": "9c3a6dac0654055594fdde1f148400331f3e55cb72ffe1b7a6806db07ab93a79",
     "examples/inference.py": "fcde5470ba4e9fce0730f544f4f3b7669bf4f9bdcaf556223cebc4ed67ef1c5a",
     "model.safetensors": "95e3731e26887d066a0b2291a1574da40a36aba5ba43e476746b9c16a64bcf08",