NPI mechanism in pythia-1b moves negation feature through complementiser 'that', auxiliary verb, and main verb across layers before predicting NPI 'any'

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• NPI licensing mechanism in pythia-1b emerges in discrete stages (steps 1000, 2000, 3000) not graduallyfinding0.828
  Training dynamics finding showing abrupt rather than gradual emergence of NPI mechanism
• Filler-gap dependency mechanism in pythia-1b emerges in two discrete stages (steps 2000 and 10K) not graduallyfinding0.769
  Training dynamics finding showing filler-gap takes longer to learn than NPI licensing
• Filler-gap mechanism in pythia-1b crosses over several different positions before arriving at output positionfinding0.762
  Mechanistic finding from CausalGym case study showing complex multi-step movement for filler-gap
• Across 5 Pythia seeds, one seed fails to learn IOI task and another fails alignment despite learning the task; all other seeds achieve perfect alignment with ϕ_nonlinfinding0.756
  Robustness check across seeds showing occasional failures of alignment map training
• The mechanisms implementing NPI licensing and filler-gap dependencies are learned in discrete stages, not graduallyclaim0.756
  Main mechanistic finding from case studies; evidence from training checkpoint analysis of pythia-1b
• Results may not fully generalize to all models and scenarios because the model organism relies on hints and nudges and Llama Nemotron cannot consistently distinguish evaluation/deployment based on subtle cuesclaim0.753
  Key limitation acknowledged by authors.
• DAS consistently finds the most causally-efficacious features across all pythia model sizes in CausalGymfinding0.748
  Main benchmark result showing DAS superiority over probing, diff-in-means, PCA, k-means, LDA, and random
• What are the neuronal mechanisms by which prior beliefs from one agent's model are received and properly implemented by a naive agent (neuronal hermeneutics)?question0.746
  Open question about inter-agent communication beyond model-space assumption