Experimenting with imperfect LLM memory

A project called SelMem is experimenting with imperfect LLM memory by storing each input in two forms: a sealed archive of the immutable original input used only for auditing and evaluation, and a mutable "lived memory" compressed representation the agent actually uses. Lived traces carry semantic content plus attributes including salience, affect, self-relevance, fidelity, access history and permanence, and can decay, become latent, merge, or be reconstructed differently on recall. Recall pairs retrieval with a grounding/judging layer that compares the reconstruction against the trace's semantic core, classifying outcomes as preservation, compression, elaboration, reframing, contradiction or unrelated recall, with repeated grounding failures able to trigger corrective reconsolidation.

Long story short of what the repo does: Instead of treating memory as immutable information storage, SelMem maintains mutable lived traces: input → semantic segmentation → selective encoding → trace → recall/reconstruction → reconsolidation → forgetting/merging Each input is stored in two forms: sealed archive — immutable original input, used only for auditing/evaluation; lived memory — compressed and mutable representation used by the agent. Lived traces contain semantic content plus attributes such as salience, affect, self-relevance, fidelity, access history and permanence. Memories can decay, become latent, merge, or be reconstructed differently when recalled. Recall uses retrieval to select memories, then a grounding/judging layer compares the reconstruction against the trace’s semantic core. It distinguishes cases such as preservation, compression, elaboration, reframing, contradiction and unrelated recall. Repeated grounding failures can trigger corrective reconsolidation