8 papers
Equivalence of Context and Parameter Updates in Modern Transformer Blocks
Adrian Goldwaser, Michael Munn, Javier Gonzalvo +1
Recent research has established that the impact of context in a vanilla transformer can be represented implicitly by forming a token-dependent, rank-1 patch to its MLP weights. Thi…
Transmuting prompts into weights
Hanna Mazzawi, Benoit Dherin, Michael Munn +3
A growing body of research has demonstrated that the behavior of large language models can be effectively controlled at inference time by directly modifying their internal states,…
Learning without training: The implicit dynamics of in-context learning
Benoit Dherin, Michael Munn, Hanna Mazzawi +2
One of the most striking features of Large Language Models (LLMs) is their ability to learn in-context. Namely at inference time an LLM is able to learn new patterns without any ad…
How iteration order influences convergence and stability in deep learning
Benoit Dherin, Benny Avelin, Anders Karlsson +3
Despite exceptional achievements, training neural networks remains computationally expensive and is often plagued by instabilities that can degrade convergence. While learning rate…
On residual network depth
Benoit Dherin, Michael Munn
Deep residual architectures, such as ResNet and the Transformer, have enabled models of unprecedented depth, yet a formal understanding of why depth is so effective remains an open…
A Bayesian Model Selection Criterion for Selecting Pretraining Checkpoints
Michael Munn, Susan Wei
Recent advances in artificial intelligence have been fueled by the development of foundation models such as BERT, GPT, T5, and Vision Transformers. These models are first pretraine…