INDEX

Explanations

phrases indicating absence or negativity

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

negation or absence

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

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

theoremstyle

-0.30

 realizarse

-0.30

setStroke

-0.29

 ardından

-0.28

else

-0.28

 domingos

-0.28

ándolo

-0.27

 velocidades

-0.27

 fabricantes

-0.27

eschön

-0.27

POSITIVE LOGITS

 Normdatei

0.72

ConstraintMaker

0.71

ThemeOverlay

0.68

 propOrder

0.67

𑄮

0.66

 punya

0.66

hasMany

0.66

HomeAsUpEnabled

0.66

<unused14>

0.65

<unused47>

0.65

Activations Density 0.037%

phrases indicating absence or negativity

negation or absence

No Comments

No Known Activations