Universal YOCO for Efficient Depth Scaling
This addresses the computational overhead and KV cache inflation issues in standard Transformers during inference scaling, representing an incremental improvement over existing efficient-attention methods.
The paper tackles the problem of inefficient inference-time compute scaling in Large Language Models by proposing Universal YOCO (YOCO-U), which combines the YOCO decoder-decoder architecture with recursive computation to achieve better capability-efficiency tradeoffs than either approach alone, resulting in competitive performance on general and long-context benchmarks.
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.