method
active
method:multitask-distributed-alignment-search-mdasMultitask Distributed Alignment Search (MDAS)
Skyline for RAVEL achieving 0.87.
Neighborhood — ranked by edge-count
Papers (1)
paper
Methods (1)
method
- RAVEL evaluationusesDisentanglement metric via latent interventions with Cause and Isolation metrics.
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.
- Practical method by Geiger et al. for finding distributed causal abstractions using gradient descent
- The core method introduced in this paper: finds alignments between high-level causal variables and distributed neural representations via gradient descent.
- The primary contribution of the paper: a bidirectional causal method that learns rotation matrices for each model to uncover and compare causally relevant latent subspaces across neural networks.
- Second central claim of the paper.
- Program under which work was conducted.
- Program that supported Tim Hua and Andrew Qin during this research.
- Used to select latents in RAVEL.
- OpenAI's approach integrating chain-of-thought reasoning into alignment; parallels contemplative self-monitoring