what is the full classification and explanation of orbit and slider non-fixed-point limiting behaviors in looped transformers?

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

• is stable fixed-point limiting behavior desirable or restrictive for reasoning tasks in looped transformers?question0.810
  Open question about whether convergence to fixed points helps or hurts reasoning performance
• Cyclic fixed point behavior and corresponding stages of inference appear to be emergent from the Transformer architecture itself, arising in both trained and randomly initialized modelsclaim0.774
  Strong claim that inference stage structure is architectural rather than learned
• why analytically do certain architectural choices (input injection, pre-norm) lead to stable limiting behavior in looped transformers?question0.762
  Limitation identified by authors: empirical results established but analytical explanation lacking
• Input injection encourages fixed-point convergence in looped transformershypothesis0.752
  Hypothesis replicated from Bansal et al. and Anil et al. and further investigated with norm ablations
• Small-scale looped transformers trained from scratch with constant 4-recurrence schedule and simplified loss self-organize into multiple distinct mixing stages mirroring feedforward modelsfinding0.749
  Evidence that stages of inference emerge without training biases from retrofitting, recurrence scheduling, or multi-recurrence losses
• Orbits and Slidersconcept0.732
  Non-fixed-point limiting behaviors in looped models originally observed by Geiping et al., analyzed for frequency and impact on stages of inference
• Looped transformers trained from scratch without feedforward-biasing training procedures still self-organize into multiple distinct mixing stages resembling feedforward stages of inferenceclaim0.726
  Establishes that stages of inference are beneficial even when repeatedly applied in recurrent depth
• A bifurcation in the miniaturized looped transformer occurs at ~150,000 training steps, when accurate integer-linear-system solving first emergesfinding0.724
  Pinpoints the training-time transition where fractal basins emerge.