update ocp internal model

README.md

@@ -148,9 +148,9 @@ Training and simulating MD with ML force fields
 We demonstrate that force-centric GNN models without any explicit physical constraints are able to predict atomic forces more accurately than state-of-the-art energy centric GNN models, while being faster both in training and inference.
 - [DIG](https://github.com/divelab/DIG)  
 A library for graph deep learning research.
-- [scn](https://github.com/Open-Catalyst-Project/ocp/tree/main/ocpmodels/models/scn)  
+- [scn](https://github.com/Open-Catalyst-Project/ocp)  
 Spherical Channels for Modeling Atomic Interactions
-- [spinconv](https://github.com/Open-Catalyst-Project/ocp/blob/main/ocpmodels/models/spinconv.py)  
+- [spinconv](https://github.com/Open-Catalyst-Project/ocp)  
 Rotation Invariant Graph Neural Networks using Spin Convolutions.
 - [HIPPYNN](https://github.com/lanl/hippynn)  
 a modular library for atomistic machine learning with pytorch.