INDEX

Explanations

references to individuals in a historical or aristocratic context

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

Queen, court, favourite, noble

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

GIVEREF

-0.62

 ujednoznacz

-0.61

XmlAccessType

-0.54

 Samir

-0.53

הערות

-0.52

KURZBESCHREIBUNG

-0.52

 Himalaya

-0.49

 Dacia

-0.49

ChildScrollView

-0.49

 transfieras

-0.49

POSITIVE LOGITS

 souverain

0.41

/**

0.39

 politique

0.39

 gouvern

0.38

Xna

0.38

 defaultstate

0.35

autique

0.35

 protocolo

0.34

vyš

0.34

 ministros

0.33

Activations Density 0.063%

references to individuals in a historical or aristocratic context

Queen, court, favourite, noble

No Comments

No Known Activations