@article{whylimittheresidualstreamtolayersandnott,
  title = {Why Limit the Residual Stream to Layers and Not Tokens? Persistent Memory for Continuous Latent Reasoning},
  author = {Mujtaba Farhan and Maheep Chaudhary},
  year = {2026},
  eprint = {2606.07720},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2606.07720},
}