Mean Squared Error between self and other activations

The specific implementation of SOO loss using MSE between self_attn.o_proj outputs at a specified layer

Neighborhood — ranked by edge-count

concept

self_attn.o_proj Module
about
The attention output projection layer where SOO Loss is computed; maps multi-head attention outputs to hidden dimension

method

SOO Loss Function
uses
A loss function measuring the dissimilarity of latent model representations of self and other, minimized during fine-tuning

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.

Self-Referencing Activationsconcept0.775
Latent model activations when processing inputs framed from the model's own perspective
Other-Referencing Activationsconcept0.741
Latent model activations when processing inputs framed from another agent's perspective
Self-Other Boundaryconcept0.739
Conceptual distinction between self and environment that non-duality dissolves; key target for alignment-by-design
Threshold-like activation assumptionconcept0.729
Assumption that small anchor changes can produce sharp performance shifts when conditions are favorable.
Self-Other Overlapconcept0.727
The extent to which a model exhibits similar internal representations when reasoning about itself and others in similar contexts
Decompose parameters, not activationsquote0.727
Core slogan encapsulating the paradigm shift of VPD.
Stress As Error Signalconcept0.724
Selfingconcept0.724
Process of reifying one's identity as an independent self; meditation practices aim to decrease selfing.