Protoreasoning in Tiny Transformers

2026-08-05

AI Summary

This study investigates whether small-scale models can achieve genuine general-purpose reasoning rather than relying on heuristic strategies. Focusing on a miniature Transformer with approximately 1 million parameters, the work introduces “protoreasoning”—a streamlined form of chain-of-thought prompting that guides the model to perform explicit, step-by-step reasoning on Dyck language tasks. Experiments demonstrate, for the first time, effective reasoning in a model far below the parameter threshold typically associated with natural language competence, substantially narrowing the out-of-distribution generalization gap. Ablation studies confirm that the performance gains stem from the content of the reasoning trajectories themselves, not merely from increased token count, offering a controllable and interpretable pathway toward understanding the fundamental mechanisms of reasoning.

Abstract

We show that tiny transformers can profitably employ a simple form of Chain of Thought, which we call protoreasoning, allowing us to study step-by-step reasoning on ~1M-parameter models and opening up opportunities for much more detailed experimentation and analysis than is feasible for larger models. Current Large Language Models exhibit impressive step-by-step reasoning, but we have yet to understand its generality, i.e., when and how LLMs learn genuinely general algorithms rather than "bags of heuristics." Such questions are hard to settle on compute-intensive frontier models trained on opaque data. To work at model scales far below the threshold for natural-language competence, we define reasoning-friendly tasks on Dyck languages (sentences of correctly nested brackets). We find that protoreasoning traces substantially close the out-of-distribution generalization gap, and ablations confirm that the trace's content, not merely its extra tokens, drives the gain.

Problem

• Chain of Thought
• reasoning
• generalization
• Dyck languages
• protoreasoning

Innovation

• protoreasoning
• tiny transformers
• chain of thought
• Dyck languages
• out-of-distribution generalization