INDEX

Explanations

mentions of court-related processes or inquiries

oai_token-act-pair · gpt-4o-mini Triggered by @bot

questions and problems

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

The neuron detects formal named entities and document headings—proper names of institutions, courts, sites, or other official titles.

oai_token-act-pair · gpt-5-mini Triggered by @vetterc0

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 plazos

-0.36

ApiModel

-0.33

 propiedades

-0.32

 warm

-0.31

대로

-0.30

end

-0.29

 pasada

-0.29

 mensuales

-0.29

 devrez

-0.29

fasis

-0.28

POSITIVE LOGITS

Personendaten

0.76

 للاسماء

0.71

ніципалі

0.65

andExpect

0.62

 invokingState

0.61

 pinulongan

0.60

GEBURTS

0.59

problems

0.59

 defaultstate

0.59

 وتسجيلات

0.58

Activations Density 0.073%

mentions of court-related processes or inquiries

questions and problems

The neuron detects formal named entities and document headings—proper names of institutions, courts, sites, or other official titles.

No Comments

No Known Activations