Training Continuous Chain of Thought Models: A Tale of Two Regimes

• URL: http://arxiv.org/abs/2607.16972v1
• Date: Sat, 18 Jul 2026 21:32:06 GMT
• Title: Training Continuous Chain of Thought Models: A Tale of Two Regimes
• Abstract summary: Continuous Chain-of-Thought methods replace verbose reasoning traces with a short sequence of dense latent representations.<n>We introduce C-MTP, a simpler, faster direct supervision approach that models each latent as an average of the embeddings in the CoT traces to be compressed.<n>We extend the evaluation of Continuous CoT methods to complex tasks with longer reasoning traces.
• Score: 39.805252492017715
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Continuous Chain-of-Thought methods replace verbose reasoning traces with a short sequence of dense latent representations. Earlier continuous CoT methods indirectly supervise the latent representations such that its final state match that of verbose reasoning traces, requiring autoregressive, slow generation during training. We introduce C-MTP, a simpler, faster direct supervision approach that models each latent as an average of the embeddings in the CoT traces to be compressed. Our approach outperforms a prior direct supervision method that approximates the distribution of compressed tokens, and performs competitively to slower indirect supervision approaches in existing evaluation setup with simplified CoT traces (less than 100 tokens). Lastly, we extend the evaluation of Continuous CoT methods to complex tasks with longer reasoning traces ($\ge$ few hundreds reasoning tokens). We find both direct and indirect supervision training methods perform poorly (roughly 65\% performance drop) in this setting, revealing the limitations of current continuous CoT methods. The code and checkpoints are released at https://github.com/Varun221/cmtp_research

Related papers

• Structural Process Supervision for Latent Chain-of-Thought Reasoning [45.1922274576773]
  Prototype-Mediated Process Supervision (PMPS) projects latent embeddings and explicit chain-of-thought embeddings into a shared prototype space.<n>PMPS compresses output token length to under 50% of explicit CoT on GSM8K-Aug.<n>On larger models and a more challenging task, PMPS consistently attains the highest accuracy among all latent reasoning methods with comparable output length.
  arXiv  Detail & Related papers  (2026-09-09T09:17:46Z)
• Revisiting Chain-of-Thought Reasoning under Limited Supervision: Semi-supervised Chain-of-Thought Learning [17.893249958828218]
  We propose a framework that uses unlabeled questions to construct pseudo reasoning supervision.<n>Semi-CoT samples multiple pseudo-CoTs for each unlabeled question, estimates answer-level semantic entropy, and selects low-entropy reasoning chains as reliable pseudo-CoT demonstrations.<n>Pilot experiments on AQuA, SVAMP, GSM8K, and MultiArith show that the entropy gate selects high-precision pseudo-CoTs, with pseudo-answer precision ranging from $91.36%$ to $100%$.
  arXiv  Detail & Related papers  (2026-07-01T22:17:39Z)
• Faithfulness as Information Flow: Evaluating and Training Faithful Chain-of-Thought Reasoning [10.87972575497941]
  Chain-of-thought (CoT) reasoning is useful for monitoring language models.<n>Models can rely on prompt-to-answer shortcuts that bypass the CoT.<n>We study CoT faithfulness through a structural information-flow perspective.
  arXiv  Detail & Related papers  (2026-05-22T23:37:29Z)
• Do Latent-CoT Models Think Step-by-Step? A Mechanistic Study on Sequential Reasoning Tasks [37.23113613155819]
  Latent Chain-of-Thought (Latent-CoT) aims to enable step-by-step computation without emitting long rationales.<n>We study CODI, a continuous-thought teacher-student distillation model, on strictly sequential-it tasks.
  arXiv  Detail & Related papers  (2026-01-31T01:48:23Z)
• PROMISE: Process Reward Models Unlock Test-Time Scaling Laws in Generative Recommendations [52.67948063133533]
  Generative Recommendation has emerged as a promising paradigm, reformulating recommendation as a sequence-to-sequence generation task over hierarchical Semantic IDs.<n>Existing methods suffer from a critical issue we term Semantic Drift, where errors in early, high-level tokens irreversibly divert the generation trajectory into irrelevant semantic subspaces.<n>We propose Promise, a novel framework that integrates dense, step-by-step verification into generative models.
  arXiv  Detail & Related papers  (2026-01-08T07:38:46Z)
• Parallel Test-Time Scaling for Latent Reasoning Models [58.428340345068214]
  Parallel test-time scaling (TTS) is a pivotal approach for enhancing large language models (LLMs)<n>Recent advances in latent reasoning, where intermediate reasoning unfolds in continuous vector spaces, offer a more efficient alternative to explicit Chain-of-Thought.<n>This work enables parallel TTS for latent reasoning models by addressing the above issues.
  arXiv  Detail & Related papers  (2025-10-09T03:33:00Z)
• Fractured Chain-of-Thought Reasoning [61.647243580650446]
  We introduce Fractured Sampling, a unified inference-time strategy that interpolates between full CoT and solution-only sampling.<n>We show that Fractured Sampling consistently achieves superior accuracy-cost trade-offs, yielding steep log-linear scaling gains in Pass@k versus token budget.
  arXiv  Detail & Related papers  (2025-05-19T11:30:41Z)
• Reasoning by Superposition: A Theoretical Perspective on Chain of Continuous Thought [64.43689151961054]
  We prove that a two-layer transformer with $D$ steps of continuous CoTs can solve the directed graph reachability problem.<n>In our construction, each continuous thought vector is a superposition state that encodes multiple search frontiers simultaneously.
  arXiv  Detail & Related papers  (2025-05-18T18:36:53Z)
• Chain-of-Thought Reasoning Without Prompting [40.92854235219315]
  CoT reasoning paths can be elicited from pre-trained language models by simply altering the textitdecoding process. The presence of a CoT in the decoding path correlates with a higher confidence in the model's decoded answer.
  arXiv  Detail & Related papers  (2024-02-15T18:55:41Z)
• Training Chain-of-Thought via Latent-Variable Inference [30.21067593018967]
  Large language models (LLMs) solve problems more accurately and interpretably when instructed to work out the answer step by step using a chain-of-thought'' prompt. Naively combining CoT with supervised tuning requires supervision not just of the correct answers, but also of detailed rationales that lead to those answers. We propose a fine-tuning strategy that tries to maximize the emphmarginal log-likelihood of generating a correct answer using CoT prompting.
  arXiv  Detail & Related papers  (2023-11-28T17:47:32Z)
• Semi-DETR: Semi-Supervised Object Detection with Detection Transformers [105.45018934087076]
  We analyze the DETR-based framework on semi-supervised object detection (SSOD) We present Semi-DETR, the first transformer-based end-to-end semi-supervised object detector. Our method outperforms all state-of-the-art methods by clear margins.
  arXiv  Detail & Related papers  (2023-07-16T16:32:14Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.