claim
active
claim:the-surface-syntax-of-natural-language-don-t-offer-muchThe surface syntax of natural language don't offer much.
Claim about the limited utility of natural language surface features.
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.
- Introduction, motivating the proposal.
- Articulates why a one-layer transformer with MLP is the appropriate starting target for mechanistic interpretability
- Reflection on the eleven principles class, affirming that even a minimal form language can yield strong results.
- States that the form language delimits what buildings can be created.
- From Turner (1982), cited to represent the functional programming community's view and then refuted.
- Core unsupervised method for generating natural language explanations of LLM activations through a verbalizer-reconstructor pair trained with RL.
- A better basis (e.g., factoring tokens into languages or parts of speech) may yield a sparse interpretable modelhypothesis0.761Same model can be dense in one basis and sparse in another.