INDEX

Explanations

phrases that indicate court proceedings or legal discussions

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

rungsseite

-0.57

 Италијани

-0.48

PhysRevD

-0.45

 CanadaChoose

-0.45

ⓧ

-0.44

Aiheesta

-0.42

exitRule

-0.41

 miniaturka

-0.40

PhysRevLett

-0.40

tvguidetime

-0.40

POSITIVE LOGITS

-------------</

0.42

IUrlHelper

0.42

jface

0.40

 printer

0.40

 denomin

0.39

lasma

0.39

setVerticalGroup

0.39

 hypno

0.39

ITU

0.39

 arrest

0.38

Activations Density 0.023%

phrases that indicate court proceedings or legal discussions

No Comments

No Known Activations