INDEX

Explanations

code structure

np_acts-logits-general · gemini-2.5-flash-lite

The neuron fires strongly on unusually specific or low‐frequency tokens—especially proper names, brand/model terms, and less common multisyllabic words—i.e. it flags rare or distinctive words rather than everyday function words.

oai_token-act-pair · o4-mini Triggered by @jyhe0408

New Auto-Interp

Configuration

google/gemma-scope-2-12b-pt/resid_post/layer_24_width_16k_l0_medium

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

௫

0.74

 microsomal

0.73

ύ

0.72

 pptn

0.70

 geochemical

0.69

 terrific

0.66

.$\

0.65

henyl

0.65

൨

0.65

 pernicious

0.64

POSITIVE LOGITS

at

0.69

↵

0.68

el

0.65

in

0.63

0.62

0.59

0.57

Main

0.57

0.56

Activations Density 0.000%

code structure

The neuron fires strongly on unusually specific or low‐frequency tokens—especially proper names, brand/model terms, and less common multisyllabic words—i.e. it flags rare or distinctive words rather than everyday function words.

No Comments

No Known Activations

code structure

The neuron fires strongly on unusually specific or low‐frequency tokens—especially proper names, brand/model terms, and less common multisyllabic words—i.e. it flags rare or distinctive words rather than everyday function words.

No Comments

No Known Activations

code structure

The neuron fires strongly on unusually specific or low‐frequency tokens—especially proper names, brand/model terms, and less common multi­syllabic words—i.e. it flags rare or distinctive words rather than everyday function words.

No Comments

No Known Activations

code structure

The neuron fires strongly on unusually specific or low‐frequency tokens—especially proper names, brand/model terms, and less common multi­syllabic words—i.e. it flags rare or distinctive words rather than everyday function words.

No Comments

No Known Activations

The neuron fires strongly on unusually specific or low‐frequency tokens—especially proper names, brand/model terms, and less common multisyllabic words—i.e. it flags rare or distinctive words rather than everyday function words.

The neuron fires strongly on unusually specific or low‐frequency tokens—especially proper names, brand/model terms, and less common multisyllabic words—i.e. it flags rare or distinctive words rather than everyday function words.