Thinking in Latents: Adaptive Anchor Refinement for Implicit Reasoning in LLMs

Disha Sheshanarayana, Rajat Subhra Pal, Manjira Sinha, Tirthankar Dasgupta

Track: long paper (up to 10 pages)

Keywords: Latent reasoning, LLMs, Adaptive halting, Latent anchors

TL;DR: AdaAnchor reasons silently by refining latent anchors. Stability-based halting stops early on easy cases. On GSM8K, SVAMP, MultiArith (1–1.5B), cuts tokens 92–93% vs CoT and steps 48–60% vs fixed, with similar accuracy.

Abstract: Token-level Chain-of-Thought (CoT) prompting has become a standard way to elicit multi-step reasoning in large language models (LLMs), especially for mathematical word problems. However, generating long intermediate traces increases output length and inference cost, and can be inefficient when the model could arrive at the correct answer without extensive verbalization. This has motivated latent-space reasoning approaches that shift computation into hidden representations and only emit a final answer. Yet, many latent reasoning methods depend on a fixed number of latent refinement steps at inference, adding another hyperparameter that must be tuned across models and datasets to balance accuracy and efficiency. We introduce AdaAnchor, a latent reasoning framework that performs silent iterative computation by refining a set of latent anchor vectors attached to the input. AdaAnchor further incorporates an adaptive halting mechanism that monitors anchor stability across iterations and terminates refinement once the anchor dynamics converge, allocating fewer steps to easier instances while reserving additional refinement steps for harder ones under a shared maximum-step budget. Our empirical evaluation across three mathematical word-problem benchmarks shows that AdaAnchor with adaptive halting yields accuracy gains of up to 5\% over fixed-step latent refinement while reducing average latent refinement steps by 48–60\% under the same maximum-step budget. Compared to standard reasoning baselines, AdaAnchor achieves large reductions in generated tokens (92–93\%) by moving computation into silent latent refinement, offering a different accuracy–efficiency trade-off with substantially lower output-token usage.

Anonymization: This submission has been anonymized for double-blind review via the removal of identifying information such as names, affiliations, and identifying URLs.

Submission Number: 87