A systematic study of query, key, and value (QKV) projection variants in Transformers found that sharing the key and value projections (Q-K=V) performs on par with or better than the standard three-pr…
Five foundational research papers explain how large language models work, covering the Transformer architecture, in-context learning, scaling laws, and instruction tuning with human feedback. The pape…
Researchers have developed TopoMamSurv, a Graph Mamba survival analysis framework that uses topology-aware ordering to address computational bottlenecks in Whole Slide Image analysis. The framework in…
Researchers introduced Geometry-Aware Tabular Diffusion (GATD), a method that improves tabular data synthesis by feeding pairwise angles and lengths from column value differences into diffusion denois…
A study evaluating encoder-only Transformer and LSTM frameworks for streamflow prediction in ungauged basins found that the LSTM outperformed the Transformer across both upstream-only and combined con…
Anthropic employees have donated more than $880,000 directly to political campaigns favoring stricter AI regulations, with OpenAI staff contributing an additional $300,000, according to FEC data analy…
Kog Team researchers introduced Delayed Tensor Parallelism (DTP), a Transformer architecture that hides communication overhead behind computation and weight streaming to accelerate batch-size-one infe…
Researchers have developed a new interpretation method for Transformer models that use heterogenous attention structures, which process information from different sources. The approach addresses chall…
Sreeni Ramadorai draws a parallel between Transformer models' residual connections and Agile software development sprints, arguing both systems preserve foundational knowledge while incrementally addi…
Prompt caching can reduce LLM input costs by 50–90% and improve time-to-first-token by 3–10× without quality loss, according to a 2026 developer guide. The optimization stems directly from Transformer…
Researchers developed HRVConformer, a deep learning architecture that classifies neonatal hypoxic-ischemic encephalopathy (HIE) by directly processing raw heart rate signals, achieving an AUC of 83.23…
A developer built a Transformer-based time-series forecaster using PyTorch to predict blood glucose levels 30 minutes in advance from Continuous Glucose Monitoring (CGM) data. The model, which uses se…
AI researcher Nathan Witkin has challenged the validity of METR's widely-cited Long Tasks benchmark, arguing its methodology is fundamentally flawed despite its status as a leading indicator of AI cap…
CODA, a GPU kernel abstraction that reparameterizes memory-bound Transformer operations like normalization and activations to execute as GEMM-plus-epilogue programs, keeping data on-chip to reduce glo…
U.S. Rep. Val Hoyle (D-OR) initially distanced herself from the pro-artificial intelligence super PAC Leading the Future after it endorsed her, stating she did not seek the endorsement and criticizing…
Google Research announced a new open science initiative on May 1, 2026, built on principles of responsible, inclusive, and rigorous research to accelerate global scientific discovery. The company will…