EPiC: Towards Lossless Speedup for Reasoning Training through Edge-Preserving CoT Condensation

2025-06-04

AI Summary

Training large language models (LLMs) with chain-of-thought (CoT) supervision incurs prohibitive computational costs due to long, redundant reasoning trajectories. Method: This paper proposes an “edge-preserving” CoT compression paradigm that retains only the critical initial segment (problem understanding) and final segment (answer convergence), while pruning intermediate reasoning steps. The approach employs a structure-aware, three-stage analysis integrating heuristic truncation with semantic coherence constraints, and is compatible with mainstream architectures including Qwen and LLaMA. Contribution/Results: We provide the first theoretical and empirical evidence that preserving only the two endpoints suffices for lossless transfer of reasoning capability—breaking the conventional dependency on full-trajectory supervision. On MATH500, our method achieves accuracy comparable to full-CoT supervision while reducing training time by over 34%, significantly enhancing CoT distillation efficiency.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Natural Language Processing: (Large) Language Models
• Computer Vision: Large Vision Models

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Large language models (LLMs) have shown remarkable reasoning capabilities when trained with chain-of-thought (CoT) supervision. However, the long and verbose CoT traces, especially those distilled from large reasoning models (LRMs) such as DeepSeek-R1, significantly increase training costs during the distillation process, where a non-reasoning base model is taught to replicate the reasoning behavior of an LRM. In this work, we study the problem of CoT condensation for resource-efficient reasoning training, aimed at pruning intermediate reasoning steps (i.e., thoughts) in CoT traces, enabling supervised model training on length-reduced CoT data while preserving both answer accuracy and the model's ability to generate coherent reasoning. Our rationale is that CoT traces typically follow a three-stage structure: problem understanding, exploration, and solution convergence. Through empirical analysis, we find that retaining the structure of the reasoning trace, especially the early stage of problem understanding (rich in reflective cues) and the final stage of solution convergence, is sufficient to achieve lossless reasoning supervision. To this end, we propose an Edge-Preserving Condensation method, EPiC, which selectively retains only the initial and final segments of each CoT trace while discarding the middle portion. This design draws an analogy to preserving the"edge"of a reasoning trajectory, capturing both the initial problem framing and the final answer synthesis, to maintain logical continuity. Experiments across multiple model families (Qwen and LLaMA) and benchmarks show that EPiC reduces training time by over 34% while achieving lossless reasoning accuracy on MATH500, comparable to full CoT supervision. To the best of our knowledge, this is the first study to explore thought-level CoT condensation for efficient reasoning model distillation.

Problem

• Condense CoT traces to reduce training costs
• Preserve answer accuracy and reasoning coherence
• Selectively retain initial and final CoT segments

Innovation

• Edge-Preserving CoT condensation method
• Retains initial and final reasoning segments
• Reduces training time by over 34%