finding

active

finding:decomposition-of-all-24-weight-matrices-in-a-67m-parameter-lm-yields-10-000-parameter-subcomponents

Decomposition of all 24 weight matrices in a 67M-parameter LM yields ~10,000 parameter subcomponents

Quantitative result of VPD application; the network's 24 matrices decompose into approximately 10,000 rank-one subcomponents.

Source paper

extracted_from

cimcWhitepaper

Neighborhood — ranked by edge-count

Claims (1)

claim

VPD identifies real, computational structure in neural network parameters
supports
Central claim that VPD successfully uncovers genuine mechanisms.

Communities (3)

community

Mechanistic interpretability & model evaluation
members_of
Spans attention head decomposition, benchmark awareness, and genomic pathogenicity prediction via neural models.
Mechanistic interpretability via parameter decomposition
members_of
Tracing information flow through weight matrices and attention heads using attribution graphs to identify causally important subcomponents in language models.
Neural network mechanistic interpretability via attribution decomposition
members_of
Tracing information flow through parameter subcomponents to isolate computational mechanisms for specific model predictions, using tools like attribution graphs and VPD.

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.

Weight matrix decompositionconcept0.732
The core idea of decomposing weight matrices into components for interpretability.
2/3 Layer of LLMconcept0.708
The layer approximately two-thirds through an LLM's transformer stack, reported to best predict human brain activity; identified as promising for consciousness indicators.
A 337-character contemplative system prompt lifts all 28 models by +2.62 points on a 10-point scale.finding0.705
Core empirical result: every model, every architecture, every alignment type responds to the contemplative prompt with measurable gain.
Parameter subcomponentconcept0.702
One of the simple rank-one matrices resulting from VPD that sums with others to reconstruct the original model weights and has a specific functional role.
What matrix decomposition or dimensionality reduction best summarizes the enormous low-rank OV and QK matrices?question0.701
Open methodological question about converting the 50k x 50k expanded matrices into human-graspable summaries
A medium building volume contains roughly 10^9 cell-sized piecesfinding0.696
Estimation from 50,000 m³ volume and average cell volume 100 cm³.
Subcomponent L2.MLP.down:3382 (density 0.00%) predicts emoticon continuations after colon, semicolon, or equalsfinding0.696
Specific discovered subcomponent that activates on punctuation like ' :', ' ;', ' =', ':-' and predicts the rest of emoticons/emojis.
Parameter subcomponents cleanly isolate true mechanisms of the modelclaim0.695
Interpretive claim that the subcomponents correspond to real functional units.