INDEX

Explanations

anesthesia, online, users

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

The neuron fires most strongly on named‐entity tokens and social‐media markers (e.g. proper nouns and “#” hashtags).

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

the contraction "'ll" (will).

oai_token-act-pair · claude-4-5-sonnet Triggered by @jyhe0408

auxiliary and modal markers (like ’ll, can, could) and limiting adverbs (e.g., only) that signal capability, intention, or constraint in statements.

oai_token-act-pair · gpt-5 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

ientras

0.86

arów

0.81

áneo

0.80

anzas

0.80

 النظام

0.80

ksa

0.79

kva

0.78

 জায়গায়

0.77

ुल्लाह

0.77

đen

0.77

POSITIVE LOGITS

=\

0.77

aka

0.76

 fumes

0.73

 hotline

0.73

-&

0.73

 poisons

0.71

 Xiang

0.69

 XXIII

0.69

 tomb

0.69

vac

0.68

Activations Density 0.000%

anesthesia, online, users

The neuron fires most strongly on named‐entity tokens and social‐media markers (e.g. proper nouns and “#” hashtags).

the contraction "'ll" (will).

auxiliary and modal markers (like ’ll, can, could) and limiting adverbs (e.g., only) that signal capability, intention, or constraint in statements.

No Comments

No Known Activations

anesthesia, online, users

The neuron fires most strongly on named‐entity tokens and social‐media markers (e.g. proper nouns and “#” hashtags).

the contraction "'ll" (will).

auxiliary and modal markers (like ’ll, can, could) and limiting adverbs (e.g., only) that signal capability, intention, or constraint in statements.

No Comments

No Known Activations