# Beyond What Seems Necessary: Hidden Gains from Scaling Training-Time Reasoning Length under Outcome Supervision

arXiv: https://arxiv.org/abs/2602.00927 · HTML: https://mlsanity.com/paper/2602.00927/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Test-time compute and efficient thinking, one of 17 directions in LLM reasoning.
- Builds on 30 of its 45 references that are in the map.
- First posted on arXiv 31 Jan 2026.

## Placed in

- LLM reasoning / [Test-time compute and efficient thinking](https://mlsanity.com/lr/test-time-compute-and-efficient-thinking/)

## Builds on (most cited first, 30 of 30)

- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://mlsanity.com/paper/2201.11903/) (2022)
- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Large Language Models are Zero-Shot Reasoners](https://mlsanity.com/paper/2205.11916/) (2022)
- [Let's Verify Step by Step](https://mlsanity.com/paper/2305.20050/) (2023)
- [DAPO: An Open-Source LLM Reinforcement Learning System at Scale](https://mlsanity.com/paper/2503.14476/) (2025)
- [Trust Region Policy Optimization](https://mlsanity.com/paper/1502.05477/) (2015)
- [Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters](https://mlsanity.com/paper/2408.03314/) (2024)
- [Shortcut Learning in Deep Neural Networks](https://mlsanity.com/paper/2004.07780/) (2020)
- [s1: Simple test-time scaling](https://mlsanity.com/paper/2501.19393/) (2025)
- [Training Large Language Models to Reason in a Continuous Latent Space](https://mlsanity.com/paper/2412.06769/) (2024)
- [Universal Transformers](https://mlsanity.com/paper/1807.03819/) (2018)
- [Chain of Thought Empowers Transformers to Solve Inherently Serial Problems](https://mlsanity.com/paper/2402.12875/) (2024)
- [Reasoning with Latent Thoughts: On the Power of Looped Transformers](https://mlsanity.com/paper/2502.17416/) (2025)
- [Looped Transformers as Programmable Computers](https://mlsanity.com/paper/2301.13196/) (2023)
- [Looped Transformers are Better at Learning Learning Algorithms](https://mlsanity.com/paper/2311.12424/) (2023)
- [The Art of Scaling Reinforcement Learning Compute for LLMs](https://mlsanity.com/paper/2510.13786/) (2025)
- [Looped Transformers for Length Generalization](https://mlsanity.com/paper/2409.15647/) (2024)
- [Transformers, parallel computation, and logarithmic depth](https://mlsanity.com/paper/2402.09268/) (2024)
- [Can Looped Transformers Learn to Implement Multi-step Gradient Descent for In-context Learning?](https://mlsanity.com/paper/2410.08292/) (2024)
- [e3: Learning to Explore Enables Extrapolation of Test-Time Compute for LLMs](https://mlsanity.com/paper/2506.09026/) (2025)
- [OMEGA: Can LLMs Reason Outside the Box in Math? Evaluating Exploratory, Compositional, and Transformative Generalization](https://mlsanity.com/paper/2506.18880/) (2025)
- [On Expressive Power of Looped Transformers: Theoretical Analysis and Enhancement via Timestep Encoding](https://mlsanity.com/paper/2410.01405/) (2024)
- [Transformers Learn to Implement Multi-step Gradient Descent with Chain of Thought](https://mlsanity.com/paper/2502.21212/) (2025)
- [Simulation of Graph Algorithms with Looped Transformers](https://mlsanity.com/paper/2402.01107/) (2024)
- [Do We Need to Verify Step by Step? Rethinking Process Supervision from a Theoretical Perspective](https://mlsanity.com/paper/2502.10581/) (2025)
- [On the Role of Depth and Looping for In-Context Learning with Task Diversity](https://mlsanity.com/paper/2410.21698/) (2024)
- [Escaping the Verifier: Learning to Reason via Demonstrations](https://mlsanity.com/paper/2511.21667/) (2025)

## Abstract (arXiv)

Training LLMs to think and reason for longer has become a key ingredient in building state-of-the-art models that can solve complex problems previously out of reach. Recent efforts pursue this in different ways, such as RL fine-tuning to elicit long CoT or scaling latent reasoning through architectural recurrence. This makes reasoning length an important scaling knob. In this work, we identify a novel phenomenon (both theoretically and experimentally): under outcome-only supervision, out-of-distribution (OOD) performance can continue improving as training-time reasoning length (e.g., the token budget in RL, or the loop count in looped Transformers) increases, even after in-distribution (ID) performance has saturated. This suggests that robustness may require a larger budget than ID validation alone would indicate. We provide theoretical explanations via two mechanisms: (i) self-iteration can induce a stronger inductive bias in the hypothesis class, reshaping ID-optimal solutions in ways that improve OOD generalization; and (ii) when shortcut solutions that work for ID samples but not for OOD samples persist in the hypothesis class, regularization can reduce the learned solution's reliance on these shortcuts as the number of self-iterations increases. We complement the theory with empirical evidence from two realizations of scaling training-time reasoning length: increasing the number of loops in looped Transformers on a synthetic task, and increasing token budgets during RL fine-tuning of LLMs on mathematical reasoning.