Showing cs.LGShow all
2 papers · 1 filter
cs.LG2026
Trained Persistent Memory for Frozen Decoder-Only LLMs
Hong Jeong
Decoder-only language models are stateless: hidden representations are discarded after every forward pass and nothing persists across sessions. Jeong (2026a) showed that trained me…
cs.LG2026
Trained Persistent Memory for Frozen Encoder--Decoder LLMs: Six Architectural Methods
Hong Jeong
Frozen encoder--decoder language models are stateless: the latent representation is discarded after every forward pass, so no information persists across sessions. This paper prese…