finding
active
finding:mdas-mdbm-skyline-achieves-ravel-disentanglement-score-0-87MDAS/MDBM skyline achieves RAVEL disentanglement score 0.87
Skyline.
Source paper
extracted_from(2025) · Adam Karvonen · Can Rager · Johnny Lin · Curt Tigges +11
Neighborhood — ranked by edge-count
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Matryoshka's hierarchical structure explains its positive scaling on disentanglement metricsclaim0.737Mechanistic interpretation.
- DB-MTL achieves ∆p = +1.15±0.16 on NYUv2, outperforming all baselines including state-of-the-artfinding0.701Primary empirical validation on scene understanding task
- Full-parameter fine-tuning more destructive to baseline but preserves more latent headroom than LoRA
- Performance with a different backbone network.
- Formal subspace recovery guarantee.
- On Qwen3-1.7B, MDS achieves ϕ1,C,↑ = 5.0 (SJTs) vs P2 at 4.7, and ϕ1,C,↓ = 1.4 (SJTs) vs P2 at 3.6finding0.690Specific consciousness sweep result for Qwen3-1.7B from Table 6 demonstrating strong bidirectional steering
- Ground truth mechanisms identified.
- Reasoning model vulnerability under prompting