$TRFM transformer

Latest papers

1. Can Learning to Answer Spatial Questions Improve Novel-View Generation? cs.CV · 14 Oct · 1 hit

   …novel-view generation. We introduce Ask4NV, built on a pretrained unified multimodal model with a Mixture-of-Transformers (MoT) architecture. Our model incorporates camera geometry into both VAE and ViT representations…

2. Representation Conditioning for Generation cs.CV · 14 Oct · 1 hit

   Representation-alignment methods effectively accelerate the diffusion transformer training. REPA, for example, uses a pretrained vision encoder to supervise an intermediate layer…

3. Modeling and Understanding Neural Weights of Humanoid Policies with Diffusion cs.RO · 14 Oct · 2 hits

   …their weights. We adopt a latent diffusion framework that couples a linear autoencoder with a Diffusion Transformer. The autoencoder compresses actor network parameters into a compact latent space, and the…

4. Your Transformer Is Secretly Elastic cs.LG · 14 Oct · 4 hits

   Scaling up transformers has produced models with broad capabilities, deployed across heterogeneous hardware and…

5. Continuum Attention cs.LG · 14 Oct · 2 hits

   We introduce Continuum Attention (CA), a causal Transformer attention operator that replaces discrete key and value lookup with integration over a…

6. Learnable Hodge-Laplacian Driven Mixture of Experts for Steady-PDE Surrogates on Coarse Unstructured Meshes cs.LG · 14 Oct · 1 hit

   …to the whole solution. This paper proposes an architecture that partitions the solution among vanilla transformer experts, each focused on a different spectral scale. The partition is built from a discrete Hodge…

7. Revisiting Simple Baselines for Relational Deep Learning cs.LG · 14 Oct · 1 hit

   …by representing them as temporal heterogeneous graphs. Recent advances based on graph neural networks and Transformers have achieved strong performance across diverse domains. However, the effectiveness of simpler…

8. Graph-Enhanced Disease Trajectory Discovery and Temporal Endpoint Forecasting from Longitudinal Electronic Health Records cs.LG · 14 Oct · 1 hit

   …policies selected on validation data. The final catalog contained 99,527 motifs, and a time-aware Transformer retrieved at most 24 patient-compatible motifs per endpoint through endpoint-specific…