Self-Conditioned Sampling doesn't use msa_prev

https://github.com/RosettaCommons/RFdiffusion/blob/main/rfdiffusion/inference/model_runners.py#L673

            msa_prev, pair_prev, px0, state_prev, alpha, logits, plddt = self.model(msa_masked,
                                msa_full,
                                seq_in,
                                xt_in,
                                idx_pdb,
                                t1d=t1d,
                                t2d=t2d,
                                xyz_t=xyz_t,
                                alpha_t=alpha_t,
                                msa_prev = None,
                                pair_prev = None,
                                state_prev = None,
                                t=torch.tensor(t),
                                return_infer=True,
                                motif_mask=self.diffusion_mask.squeeze().to(self.device))   

In the above sampling code, a previous timestep's features are always set to None. Is this intended?

Read rfdiffusion/inference/model_runners.py around line 673 and inspect the model call shown in the issue. Determine whether the previous timestep features should be passed during self-conditioned sampling and whether the current None values are intentional. Done means the intended behavior is established and the sampling path is corrected or documented accordingly.