aiqbench analysis
This paper addresses a critical bottleneck in scaling deep learning by introducing Riemannian--Lorentz Parameter Fusion (RLPF), a novel method for merging heterogeneous architectures like Vision Transformers and State-Space Models. By projecting parameter groups onto hyperbolic space to compute a regularized geodesic barycenter, the approach enables efficient combination of independently trained models without requiring full retraining. While the resulting hybrid system achieves competitive accuracy on standard benchmarks, it is important to note that the fusion gate and final model still require training, distinguishing this from purely weight-space merging techniques.
Riemannian--Lorentz Fusion of Vision Transformers and State-Space Models
arXiv:2609.19384v1 Announce Type: cross Abstract: Scaling deep learning faces critical bottlenecks: data exhaustion, exponential training costs, and resource concentration. Model merging combines pre-trained checkpoints without gradient descent, offering orders-of-magnitude savings versus retraining. Combining independently trained vision models is difficult when their architectures and parameter shapes differ. Existing weight-space merging methods generally assume aligned, shape-compatible checkpoints, whereas a Vision Transformer (ViT) and a state-space model (SSM) implement token mixing with different operators. We study a hybrid Heterogeneous merging setting that retains both architectures while aligning parameter groups by semantic role. Our proposed Riemannian--Lorentz Parameter Fusion (RLPF) method projects aligned groups to common coordinates, lifts selected coordinates to the Lorentz hyperboloid model of hyperbolic space, computes a regularized geodesic barycenter, and decodes the result into the two branches. A learned gate then combines branch logits for each input. Component groups use fixed curvature values, with normalization parameters treated as Euclidean. In the results available in this manuscript, the fine-tuned system obtains 82.37\% on CIFAR-10, 75.04\% on Oxford-IIIT Pet, and 78.58\% top-1 accuracy on ImageNet-1K; the corresponding best-parent accuracies are 76.54\%, 71.42\%, and 76.42\%. On ImageNet-1K, the reported pre-fine-tuning initialization reaches 77.80\%. These results support further study of geometry-aware heterogeneous fusion, but not a training-free single-checkpoint merge: RLPF is a two-branch hybrid whose gate and reported final models are trained.
Abstract:Scaling deep learning faces critical bottlenecks: data exhaustion, exponential training costs, and resource concentration. Model merging combines pre-trained checkpoints without gradient descent, offering orders-of-magnitude savings versus retraining. Combining independently trained vision models is difficult when their architectures and parameter shapes differ. Existing weight-space merging methods generally assume aligned, shape-compatible checkpoints, whereas a Vision Transformer (ViT) and a state-space model (SSM) implement token mixing with different operators. We study a hybrid Heterogeneous merging setting that retains both architectures while aligning parameter groups by semantic role. Our proposed Riemannian--Lorentz Parameter Fusion (RLPF) method projects aligned groups to common coordinates, lifts selected coordinates to the Lorentz hyperboloid model of hyperbolic space, computes a regularized geodesic barycenter, and decodes the result into the two branches. A learned gate then combines branch logits for each input. Component groups use fixed curvature values, with normalization parameters treated as Euclidean. In the results available in this manuscript, the fine-tuned system obtains 82.37\% on CIFAR-10, 75.04\% on Oxford-IIIT Pet, and 78.58\% top-1 accuracy on ImageNet-1K; the corresponding best-parent accuracies are 76.54\%, 71.42\%, and 76.42\%. On ImageNet-1K, the reported pre-fine-tuning initialization reaches 77.80\%. These results support further study of geometry-aware heterogeneous fusion, but not a training-free single-checkpoint merge: RLPF is a two-branch hybrid whose gate and reported final models are trained.
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Multimedia (cs.MM)
arXiv:2609.19384 [cs.CV]
(or arXiv:2609.19384v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2609.19384 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 318 words
Related coverage
Humanity’s Last Invention — Richard Socher of Recursive
Latent Space · Sep 14, 2026
Academia is for Ambition — Alex Zhang, MIT
Latent Space · Oct 2, 2026
STREAM: Stochastic Riemannian Flow Matching with Anisotropic Decoder for Digital Histopathology Image Generation
arXiv · Jun 8, 2026
STREAM: Stochastic Riemannian Flow Matching with Anisotropic Decoder for Digital Histopathology Image Generation
arXiv · Oct 7, 2026