method
active
method:saga-solver

SAGA solver

Fast incremental gradient method used to train linear probes in CausalGym

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

  • Jailbreakingconcept0.691
    Users coaxing dialogue agents into issuing threats or toxic content by overriding intended persona constraints
  • Mesa-Optimizerconcept0.687
    A learned optimizer running inside a base optimizer; transformers proposed as mesa-optimizers implementing gradient descent in-context
  • Jailbreakconcept0.672
    Methods to bypass model safety training; features may activate during jailbreaks.
  • Evolutionary search process used to evolve populations of embryos.
  • Sparse Autoencoderframework0.660
    Interpretability framework used to decompose layer-40 activations into sparse feature sets for studying emotional alignment and persistence
  • A system of sequences/rules that can generate well-adapted living plans, potentially implemented on computers to empower ordinary people.
  • Question prompting the audience to identify the primary degradation symptoms (riverscape degradation, structural starvation).