INDEX

Explanations

names or identifiers of authors or contributors

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

sequences of specific characters

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

 effectively

-0.34

illow

-0.33

 convencional

-0.31

 vertreten

-0.31

 Deutschlands

-0.31

 objectively

-0.30

 gani

-0.30

Secondly

-0.29

 dramatically

-0.28

 tradizionale

-0.28

POSITIVE LOGITS

Tembelea

0.83

 beginnetje

0.81

ViewImports

0.75

 Administrativna

0.74

 defaultstate

0.74

 المعيارى

0.69

 otomatig

0.67

 وتسجيلات

0.66

enumii

0.65

+#+

0.65

Activations Density 0.009%

names or identifiers of authors or contributors

sequences of specific characters

No Comments

No Known Activations