finding

active

finding:clmas-achieves-the-best-iia-in-the-causally-inaccessible-no-access-direction-while-matching-mas-in-the-accessible-direction

CLMAS achieves the best IIA in the causally inaccessible (No Access) direction while matching MAS in the accessible direction

Demonstrates the value of the CL auxiliary loss for recovering causal alignments when one model cannot be intervened upon.

Source paper

extracted_from

Model Alignment Search

(2025) · Satchel Grant

Neighborhood — ranked by edge-count

Claims (1)

claim

CLMAS can potentially reduce or remove the need for NN stimulation during alignment training in biological settings
supports
Forward-looking claim about the practical utility of CLMAS for ANN-BNN comparisons with limited causal access.

Hypotheses (1)

hypothesis

The effect size of CLMAS improvement over baselines will correlate with the amount of variability in the behavioral null space of the inaccessible model
associated_with
Prediction about when CLMAS will be most beneficial, stated explicitly in the paper.

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.

Near-perfect IIA can be achieved on randomly initialised models that cannot solve the task, suggesting causal alignment does not require task capabilityclaim0.789
Empirical support for vacuousness of unrestricted causal abstraction
Over 80% IIA achieved using complex non-linear alignment maps on randomly initialised MLPs in hierarchical equality taskfinding0.778
Demonstrates that high IIA can be obtained even when model cannot solve the task
Non-linear ϕ_nonlin achieves near-perfect IIA on distributive law task for both And-Or and And-Or-And algorithms, eliminating linear/identity map differencesfinding0.765
Corroborating result on additional task confirming main paper findings
MAS successfully aligns behavior between Multi-Object GRU models in both embedding and hidden state layers with high IIAfinding0.761
Demonstrates MAS's ability to bidirectionally transfer behavior where RSA shows low embedding correlation.
The effect of alignment map ϕ complexity on IIA in causal abstraction is an analogue of the probing complexity–accuracy trade-offclaim0.760
Authors connect their finding to the prior probing literature debate
MAS reduces number of required alignment matrices for n-model comparison from n(n-1) or n^2 (stitching) to nfinding0.759
Key computational efficiency advantage of MAS over traditional model stitching for multi-model comparisons.
Non-linear alignment map ϕ_nonlin achieves near-optimal IIA across all layers on hierarchical equality task, eliminating layer-dependent degradation seen with linear mapsfinding0.751
Key empirical result: non-linear maps overcome linear maps' failure in deeper layers
Mass-mean probe directions outperform LR and CCS in causal intervention experiments (NIE) in 7/8 experimental conditionsfinding0.748
Core result showing MM is superior to LR for causal implication despite similar classification accuracy