INDEX

Explanations

references to destructive actions and property damage

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

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 '\\;'

-0.59

Personendaten

-0.56

 autorytatywna

-0.48

цездатний

-0.48

AutoField

-0.46

 informée

-0.44

colla

-0.44

\{\\

-0.44

ğraf

-0.43

المراجع

-0.42

POSITIVE LOGITS

 damage

0.66

 destruction

0.66

 smashing

0.63

damage

0.62

 destru

0.56

 vandal

0.56

 vandalism

0.56

destruction

0.56

 damaged

0.55

 destro

0.55

Activations Density 0.029%

references to destructive actions and property damage

No Comments

No Known Activations