@misc{indiciaeb8c46549f130,
  title = {Repurposing Pre-trained LLMs as High Fidelity Continuous Text Autoencoders},
  author = {Arkanath Pathak and Unnat Jain and Alexander C. Berg},
  year = {2026},
  url = {https://arxiv.org/abs/2609.27248},
  note = {Source identifier: 2609.27248}
}