INDEX

Explanations

Arabic, Hebrew, Spanish, programming

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

text written in non-Latin scripts, particularly Hebrew and Arabic characters.

oai_token-act-pair · claude-4-5-sonnet Triggered by @bfgcn7frf5

The neuron detects tokens composed of right-to-left non-Latin script characters (e.g. Hebrew or Arabic letters).

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

New Auto-Interp

Configuration

google/gemma-scope-27b-pt-res/layer_10/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

 encompass

-1.80

神情

-1.76

 queridos

-1.70

 fieles

-1.68

 erlä

-1.66

 antwortete

-1.66

了解更多

-1.65

those

-1.63

 causando

-1.62

 estándar

-1.61

POSITIVE LOGITS

了

1.84

te

1.73

of

1.65

for

1.63

 kerajaan

1.59

фика

1.58

ляю

1.56

 København

1.54

켭

1.51

၀

1.49

Activations Density 0.003%

Arabic, Hebrew, Spanish, programming

text written in non-Latin scripts, particularly Hebrew and Arabic characters.

The neuron detects tokens composed of right-to-left non-Latin script characters (e.g. Hebrew or Arabic letters).

No Comments

No Known Activations

Arabic, Hebrew, Spanish, programming

text written in non-Latin scripts, particularly Hebrew and Arabic characters.

The neuron detects tokens composed of right-to-left non-Latin script characters (e.g. Hebrew or Arabic letters).

No Comments

No Known Activations