Addressing the resource waste caused by fixed computation paradigms in deep learning models under dynamic scenarios, this paper proposes a Transformer−1 architecture based on the principle of deep adaptivity. This architecture achieves dynamic matching between input features and computational resources by establishing a joint optimization model for complexity and computation. Our core contributions include: (1) designing a two-layer control mechanism, composed of a complexity predictor and a reinforcement learning policy network, enabling end-to-end optimization of computation paths; (2) deriving a lower bound theory for dynamic computation, proving the system's theoretical reach to optimal efficiency; and (3) proposing a layer folding technique and a CUDA Graph pre-compilation scheme, overcoming the engineering bottlenecks of dynamic architectures. In the ImageNet-1K benchmark test, our method reduces FLOPs by 42.7\% and peak memory usage by 34.1\% compared to the standard Transformer, while maintaining comparable accuracy (±0.3\%). Furthermore, we conducted practical deployment on the Jetson AGX Xavier platform, verifying the effectiveness and practical value of this method in resource-constrained environments. To further validate the generality of the method, we also conducted experiments on several natural language processing tasks and achieved significant improvements in resource efficiency.
@article{arxiv.2501.16394,
title = {Transformer^-1: Input-Adaptive Computation for Resource-Constrained Deployment},
author = {Lumen AI and Tengzhou No. 1 Middle School and Shihao Ji and Zihui Song and Fucheng Zhong and Jisen Jia and Zhaobo Wu and Zheyi Cao and Xu Tianhao},
journal= {arXiv preprint arXiv:2501.16394},
year = {2025}
}