Latent Reasoning with Normalizing Flows
- Published
- Source
- arXiv
- Paper number
- 348
- Field
- LLMs / NLP
- arXiv ID
- 2606.06447
Key points
- We place a normalizing flow inside the LLM backbone to provide exact likelihoods for continuous CoT while preserving left-to-right decoding and KV-cache support.
- Continuous thought positions are handled by the NF head, while text positions are handled by the LM head, and both are processed jointly in the same causal stream.
- The system supports both supervised likelihood training and policy-gradient optimization through a unified interface.
- Even with Gaussian perturbations in latent space, pass@1 remains nearly unchanged, showing that the learned continuous thought space is locally smooth and functionally robust.
- On code-generation benchmarks, it improves accuracy while reducing inference cost compared with explicit CoT and prior latent reasoning methods such as LaDiR.
Paper links
External research summaries. These are not HDATF publications or measured product results.