Search papers, labs, and topics across Lattice.
The paper introduces FAAST, a novel forward-only adaptation method that uses closed-form fast weights to analytically compile labeled examples, enabling efficient test-time supervised adaptation. FAAST eliminates the need for backpropagation or memory-based learning, achieving constant-time inference and decoupling task adaptation from pretrained representations. Experiments on image classification and language modeling show that FAAST matches or exceeds backprop-based adaptation while reducing adaptation time by over 90% and memory usage by up to 95% compared to memory/context-based adaptation.
Forget backprop and memory lookups: FAAST lets you adapt models at test time with a single forward pass, matching fine-tuning accuracy with massive speed and memory gains.
Adapting pretrained models typically involves a trade-off between the high training costs of backpropagation and the heavy inference overhead of memory-based or in-context learning. We propose FAAST, a forward-only associative adaptation method that analytically compiles labeled examples into fast weights in a single pass. By eliminating memory or context dependence, FAAST achieves constant-time inference and decouples task adaptation from pretrained representation. Across image classification and language modeling benchmarks, FAAST matches or exceeds backprop-based adaptation while reducing adaptation time by over 90\% and is competitive to memory/context-based adaptation while saving memory usage by up to 95\%. These results demonstrate FAAST as a highly efficient, scalable solution for supervised task adaptation, particularly for resource-constrained models. We release the code and models at https://github.com/baoguangsheng/faast.
