SDF-Only Model Organism

Intermediate model after synthetic document fine-tuning but before expert iteration; used as ablation baseline.

Neighborhood — ranked by edge-count

framework

Two-Stage Model Organism Training (SDF + Expert Iteration)
associated_with
The paper's framework for training evaluation-aware model organisms: SDF seeds beliefs, expert iteration reinforces evaluation behavior.

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.

Model Organismconcept0.776
A model deliberately trained to exhibit alignment-relevant properties so researchers can study them with ground truth.
Living system as individual organismconcept0.681
The basis of the standard biological definition of life, centered on the individual organism.
Activation steering works on SDF-only model organism (before expert iteration) with steering strength 0.4finding0.679
Replicates main result on simpler model; qualitatively similar patterns.
modelconcept0.678
A representation that captures relevant aspects of a system; according to the theorem, the regulator must embody this.
Discrete State-Space Modelsconcept0.676
Mathematical formalism used in active inference for modeling hierarchical and discrete brain processes.
Top-Down Models in Biologyframework0.676
Pezzulo and Levin's framework for explanation and control of complex living systems above the molecular level, cited as prior work by Levin
Genomic Foundation Modelsconcept0.676
Large pretrained models for genomic sequences; subject of this work's pooling improvements.
The difference between simple language models and multicellular organisms goes beyond the substrate of intelligence considered.claim0.676
Claim that topologies, not material substrates, account for differing organisational abilities