Skip to content
No results
  • Explainers
J-Space logo
J-Space
  • Explainers
J-Space logo
J-Space
  • Deep Dives

Neuron Transfer Functions Inside Circuits

A 2024 arXiv preprint frames neuron-to-neuron computation as transfer functions, clarifying what circuit analysis can claim about language models.

  • Editorial Team
  • November 5, 2026
  • Deep Dives

Rosetta Neurons Scaling Paper Review

Rosetta Neurons scaling paper review: shared neurons in 80M-30B parameter models grow as a sublinear power law (exponent 0.5-0.7), so their share falls.

  • Editorial Team
  • November 2, 2026
  • Philosophy of AI

Access Tests Are Not Phenomenal Consciousness

The definition that answers the first part of the question is in the section “What about consciousness?”

  • Editorial Team
  • November 1, 2026
  • Deep Dives

Arithmetic Patching Paper: GN vs STR

GN on the arithmetic prompts adds Gaussian noise to the token embeddings of the third example’s operands, X 3 and Y 3.

  • Editorial Team
  • October 31, 2026
  • Deep Dives

Vision Sparse Autoencoder L0

That expression formalizes the prose definition. The PDF text used here does not display this equation.

  • Editorial Team
  • October 29, 2026
  • Deep Dives

Activation Control Paper Method Review

How the Activation Controllability Benchmark paper works: a fixed copy task, a 0.85 similarity keep-rule, and residual-stream concept projections.

  • Editorial Team
  • October 26, 2026
  • Deep Dives

Gao’s logit-diff ablation is not a circuit rule

Gao et al.'s logit-diff ablation score measures how sparse an SAE latent's effect on output logits is. It is not a rule for admitting circuit components.

  • Editorial Team
  • October 25, 2026
  • Deep Dives

Gao’s citation gloss of the BERT illusion

Gao et al. (2024) cite the BERT interpretability illusion as a gloss: random activating examples can make an overly broad feature explanation look right.

  • Editorial Team
  • October 24, 2026
  • Deep Dives

How does arXiv:2412.09565v2 define obfuscated activations against a named latent-space monitor?

In arXiv:2412.09565v2 (Bailey et al., 8 Feb 2025), an activation is obfuscated against a named latent-space monitor D, not obfuscated in the abstract.

  • Editorial Team
  • October 22, 2026
  • Deep Dives

Vi-CD and WeightLens/CircuitLens: Which Paper-Stated Controls Make an Extracted Circuit Causal, and What Stays Unknown

An extracted subgraph is not yet a causal circuit. Read it as causal only after the paper’s own intervention controls are fixed, and do not fill parameters…

  • Editorial Team
  • October 19, 2026
Prev
1 2 3 4 5 6
Next
Copyright © 2026 - WordPress Theme by CreativeThemes