INDEX

Explanations

asking if normal

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

The neuron flags tokens in “is‐this‐normal/this‐is‐correct” type constructions—i.e. occurrences of copular “is” when used to ask or assert that behavior is normal or correct.

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

statements describing whether something is normal or expected behavior.

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

New Auto-Interp

Configuration

google/gemma-scope-27b-pt-res/layer_22/width_131k

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 '/';

-1.13

enschaft

-1.00

 "/";

-0.99

讐

-0.93

uuuu

-0.88

UDI

-0.88

枫

-0.86

 alternatives

-0.84

 uname

-0.84

 улица

-0.83

POSITIVE LOGITS

 нормально

1.34

 harmless

1.31

 normal

1.29

࿚

1.23

 reassure

1.18

However

1.13

 phenomenon

1.09

 katholischen

1.07

quartier

1.07

but

1.06

Activations Density 0.044%

asking if normal

The neuron flags tokens in “is‐this‐normal/this‐is‐correct” type constructions—i.e. occurrences of copular “is” when used to ask or assert that behavior is normal or correct.

statements describing whether something is normal or expected behavior.

No Comments

No Known Activations

asking if normal

The neuron flags tokens in “is‐this‐normal/this‐is‐correct” type constructions—i.e. occurrences of copular “is” when used to ask or assert that behavior is normal or correct.

statements describing whether something is normal or expected behavior.

No Comments

No Known Activations