INDEX

Explanations

proper nouns, particularly names and technical terms associated with specific subjects or contexts

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

aarrggbb

-0.47

runApp

-0.40

StructEnd

-0.40

verwijspagina

-0.40

 resourceCulture

-0.38

 disponibilités

-0.38

 ExecuteAsync

-0.36

Décès

-0.36

 argint

-0.35

 continúas

-0.35

POSITIVE LOGITS

Lähteet

0.81

참고

0.73

 estekak

0.59

Mur

0.57

Mur

0.56

MZ

0.56

 otomatig

0.54

guten

0.53

than

0.53

QT

0.52

Activations Density 1.109%

proper nouns, particularly names and technical terms associated with specific subjects or contexts

No Comments

No Known Activations