INDEX

Explanations

sequences of underscores used to indicate spacing or formatting in code

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

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_131k/average_l0_114

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

featureID

-0.71

 Италијани

-0.60

LookAnd

-0.60

parsedMessage

-0.57

Xna

-0.54

MeasureSpec

-0.50

LocalizedString

-0.50

RectangleBorder

-0.49

providedIn

-0.49

 Мексичка

-0.49

POSITIVE LOGITS

 prosím

0.37

 duquel

0.36

 edad

0.34

 perquè

0.33

 siempre

0.32

Accès

0.32

indd

0.31

 ffilmiau

0.30

 cardiaque

0.30

 policiales

0.30

Activations Density 0.005%

sequences of underscores used to indicate spacing or formatting in code

No Comments

No Known Activations