A forum thread on fine-tuning CLIP-based NSFW detection models surfaced practical guidance from practitioners, including a suggested starting point of batch size 16 and learning rate 5e-5 drawn from a…
Practitioners distilling vision models report that a temperature around T=3-4 best transfers the teacher's inter-class "dark knowledge" to the student, while higher values degrade the signal into unif…
Researchers introduced Stiefel-AdamW, a near drop-in replacement for the AdamW optimizer that constrains one factor of linear factorization blocks to the Stiefel manifold, according to arXiv paper 260…
A developer reverse-engineered NVIDIA's DLSS5 binary to infer its model architecture, revealing a heterogeneous, hierarchical, recurrent, single-pass neural renderer with tiled Swin transformer famili…
Researchers propose GeoUniPR, a geometry-consistent unified framework for cross-modal place recognition that projects LiDAR point clouds into camera perspective to create depth image views, achieving …
Researchers introduced KANEx, the first framework leveraging Kolmogorov-Arnold Networks (KANs) to ground Vision-Language Model (VLM) reasoning for medical explainability, achieving a 10% improvement i…
Researchers propose a decoupled transfer learning strategy that adapts normalization layers and precomputes features once, reducing training overhead without end-to-end backpropagation. Evaluated on f…
Researchers introduced Sparse-Reslim, a plug-in routing module that reduces computational cost in ViT-based weather forecasting by processing only 25% of spatial tokens through expensive transformer b…
TrustformeRS 0.1.1, a pure Rust port of Hugging Face Transformers with over 1.4 million lines of code, was released on April 25, 2026, delivering 49+ transformer architectures and up to 1.67x speedup …
Researchers evaluated six deep learning architectures for maritime object detection, including CNNs and a Vision Transformer (ViT), using a dataset of 6,468 images under various weather conditions. Th…
A study of six self-supervised geospatial foundation models found that model rankings change across tasks and adaptation settings, with task-relevant information more accessible in intermediate transf…
Researchers have developed DiffusionBlocks, a framework that partitions transformer neural networks into independently trainable blocks to reduce memory requirements proportionally while maintaining c…
StepFun released Step 3.7 Flash, a 198-billion parameter Mixture-of-Experts vision-language model with native image understanding and improved tool-use reliability for agentic coding and search workfl…
A preliminary clinical validation of the Melanoscope AI mobile dermoscopy clinical decision support system, conducted across four "Melanoma Day" screening sessions in Orel, Russia from June 2025 to Ap…