Replace PyTorch weights: fold the attention-output LoRA over heads as openpi does; checked against JAX
Browse files
pytorch/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 7233650408
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:943c9509364ea6d7aedad9973c264139c863f8b8aef90120b070992d404544c9
|
| 3 |
size 7233650408
|