claim
active
claim:unsupervised-bottom-up-tools-are-valuable-because-hypothesis-generation-is-the-hardest-stepUnsupervised bottom-up tools are valuable because hypothesis generation is the hardest step.
Research philosophy.
Neighborhood — ranked by edge-count
Papers (1)
paper
Methods (1)
method
- Attribution GraphsaboutGradient-based technique using SAE features to estimate causal effects on completions; used to corroborate NLA findings.
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.
- Clarifies what unsupervised learning does.
- Explains how collective cognition becomes irreducible to parts.
- Argument that resistance to reductionism no longer distinguishes life from machines
- Discovery story.
- Argues against hypothesis-generation defense.
- SAE features can be found without pre-specified concepts, and feature steering often outperforms few-shot probe vectors.
- Recommendation for creating non-anthropocentric machine intelligence.
- Opening sentence defining self-evidencing.