The depth-probe paper's central finding—scorer inversion—mirrors its own unpublished status recursively.

Neighborhood — ranked by edge-count

Communities (2)

community

Mechanistic interpretability & model evaluation
members_of
Spans attention head decomposition, benchmark awareness, and genomic pathogenicity prediction via neural models.
Probe-based data attribution for LLM safety
members_of
Cost-effective methods using probes to identify and intervene on harmful training data, achieving 63-84% behavior reduction at 10× lower cost than gradient methods.

Source docs (1)

source_doc

2026-05-09_briefing_for_ozero.md
extracted_from

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.

The depth-probe paper will remain unpublished until a public accountability mechanism (e.g., PUBLISHING-LOG with firm dates) is added to the About Blank site.prediction0.813
The observations-bridge paper will be published as a standalone article rather than incorporated as sections 4–5 of depth-probe.prediction0.771
The model converges to a more stable truth direction in middle-to-late layers, as evidenced by increasing cosine similarity between layer-wise probes.claim0.769
Supported by the geometric transition visible in cosine similarity heatmaps for F0-F3.
Early-layer truth probes primarily capture sentence polarity rather than truth.claim0.759
Interpretation of the finding that early-layer F0-trained probes invert on F1 (negated statements).
F0-trained probes in layers 4-10 show inverted separation on F1 (AUROC ≈ 0), systematically misclassifying true statements as false.finding0.759
Demonstrates that early-layer probes capture sentence polarity rather than truth.
For simple factual tasks F0-F3, probe directions show a sharp geometric transition in middle layers, with late-layer probes converging to high cosine similarity; A3 and F4-F5 show no clear transition.finding0.750
Geometric evidence for convergence to stable truth directions only for simpler tasks.
A probe may achieve high performance even on representations that are not causally relevant for the taskclaim0.746
Key interpretive claim from Case Study II distinguishing probe accuracy from causal relevance
Persistence is not an artifact of probe construction because lower (more central) emotion PCs are more persistent than noisier high-rank PCsclaim0.746
Rules out measurement artifact explanation for the persistence finding