Question about two differences between paper and released code

Hi, thanks for releasing the code!
While reading it, I found two places where it seems to differ from the paper, and I'd like to confirm which version was used for the reported results.

1. No re-encoding in the memory update

The paper (Sec. 4.4, Algorithm 1) re-encodes decoded frames into clean latents (z̃ₜ = E(Iₜ)) before back-projection. In the code, update_latent_memory() passes the generated latents directly to lpc.update() with no vae.encode call. Which one matches the released recipe?

2. Side branch frozen in Stage 2

The paper (Sec. 4.5) says LoRA is jointly optimized with the side branch in Stage 2. In scripts/train.py, set_stage_trainable("lora") freezes the side branch (model.vace.eval()) and the optimizer only receives LoRA parameters. Was the side branch actually frozen in Stage 2?

Apologies if I've misread anything — thanks in advance!

Start with Secs. 4.4 and 4.5 of the paper, then inspect update_latent_memory(), lpc.update(), and scripts/train.py, especially set_stage_trainable("lora"). Compare the released implementation with the two described training steps; done means confirming which behavior produced the reported results and documenting the clarification.