INDEX

Explanations

conditional statements and logical constructs in the text

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

the neuron detects information-dense content words typical of academic/technical prose (results/statements, numeric/condition-related terms and emphatic modifiers).

oai_token-act-pair · gpt-5-mini Triggered by @phamvanluong2003

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 12-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_12/width_16k/average_l0_82

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.12.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

Datuak

-0.69

]='\

-0.65

AnimationsModule

-0.64

 виправивши

-0.63

ReusableCell

-0.60

&[

-0.58

__':

-0.58

 Audiodateien

-0.55

WriteBarrier

-0.55

 crossorigin

-0.54

POSITIVE LOGITS

UnsafeEnabled

0.64

 unknownFields

0.61

 CreateTagHelper

0.59

principalColumn

0.59

 complètes

0.58

 تضيفلها

0.58

ValueGenerated

0.57

脚注の使い方

0.56

 blanches

0.56

+#+#

0.56

Activations Density 0.948%

conditional statements and logical constructs in the text

the neuron detects information-dense content words typical of academic/technical prose (results/statements, numeric/condition-related terms and emphatic modifiers).

No Comments

No Known Activations