GradMem: Learning to Write Context into Memory with Test-Time Gradient Descent

ArXi:2603.13875v1 Announce Type: cross Many large language model applications require conditioning on long contexts. Transformers typically this by storing a large per-layer KV-cache of past activations, which incurs substantial memory overhead. A desirable alternative is ompressive memory: read a context once, it in a compact state, and answer many queries from that state. We study this in a context removal setting, where the model must generate an answer without access to the original context at inference time. We.