INDEX

Explanations

negations and contrasting phrases

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

}{*}{}

-0.45

 wikipagina

-0.45

 jadx

-0.43

anInt

-0.42

 repos

-0.40

 Архів

-0.40

 Pologne

-0.39

EntityId

-0.39

LayoutStyle

-0.39

 необ

-0.39

POSITIVE LOGITS

 geçti

0.53

toHaveBeenCalled

0.52

OGND

0.51

 solely

0.51

 avviene

0.48

in

0.48

 because

0.47

 appunto

0.47

 during

0.45

abestanden

0.45

Activations Density 0.199%

negations and contrasting phrases

No Comments

No Known Activations