Universal YOCO for Efficient Depth Scaling

TL;DR

YOCO-U combines YOCO architecture with recursive computation to enable efficient, scalable inference in large language models, improving token utility and performance with limited overhead.

Contribution

It introduces YOCO-U, a novel architecture that synergistically integrates YOCO and recursion for efficient depth scaling in LLMs.

Findings

YOCO-U achieves a favorable tradeoff between capability and efficiency.

It maintains a constant global KV cache and linear pre-filling.

Empirical results show YOCO-U's competitiveness on benchmarks.

Abstract

The rise of test-time scaling has remarkably boosted the reasoning and agentic proficiency of Large Language Models (LLMs). Yet, standard Transformers struggle to scale inference-time compute efficiently, as conventional looping strategies suffer from high computational overhead and a KV cache that inflates alongside model depth. We present Universal YOCO (YOCO-U), which combines the YOCO decoder-decoder architecture with recursive computation to achieve a synergistic effect greater than either alone. Built on the YOCO framework, YOCO-U implements a Universal Self-Decoder that performs multiple iterations via parameter sharing, while confining the iterative process to shallow, efficient-attention layers. This combination yields a favorable capability-efficiency tradeoff that neither YOCO nor recursion achieves independently. The YOCO architecture provides a constant global KV cache and linear pre-filling, while partial recursion enhances representational depth with limited overhead. Together, YOCO-U improves token utility and scaling behavior while maintaining efficient inference. Empirical results confirm that YOCO-U remains highly competitive in general and long-context benchmarks, demonstrating that the integration of efficient-attention architectures and recursive computation is a promising direction for scalable LLMs.