A Model with No Head and Many Thoughts

August 31, 2026 ยท Grace Period ยท ๐Ÿ› Findings of EMNLP 2026

โณ Grace Period
This paper is less than 90 days old. We give authors time to release their code before passing judgment.
Authors Nikita Koriagin, Yaroslav Aksenov, George Bredis, Gleb Gerasimov, Nikita Balagansky, Daniil Gavrilov arXiv ID 2608.31069 Category cs.LG: Machine Learning Cross-listed cs.CL Citations 0 Venue Findings of EMNLP 2026
Abstract
Large language models decode by projecting hidden states through a large vocabulary head at every step. This operation is computationally costly and forces all reasoning to be expressed in discrete tokens. We introduce Soft Latent Thinking, a method that replaces the LM head during reasoning with a lightweight projector, enabling autoregressive rollout in embedding space where reasoning steps remain continuous rather than tokenized. Experiments on DeepSeek-Qwen-1.5B and LLaMA-3.2-3B show that Soft Latent Thinking consistently improves pass@k across all k while reducing per-step compute during chain-of-thought. Our method achieves the highest pass@32 among all soft-thinking approaches, demonstrating that effective reasoning can be carried out in continuous space without discrete token generation.
Community shame:
Not yet rated
Community Contributions

Found the code? Know the venue? Think something is wrong? Let us know!

๐Ÿ“œ Similar Papers

In the same crypt โ€” Machine Learning