INDEX

Explanations

variations and references to existing concepts or entities

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 fubject

-0.51

 Houſe

-0.50

XmlAccessType

-0.49

 Relaciones

-0.48

ſelf

-0.48

 quæ

-0.47

 Inſ

-0.47

 myſelf

-0.46

 onely

-0.44

Życiorys

-0.44

POSITIVE LOGITS

цездатний

0.58

 ویکی‌پدی

0.57

تقاوى

0.52

 Italijanski

0.49

 autorytatywna

0.49

 nonUne

0.47

 kaarangay

0.47

 Италијани

0.47

 sumpay

0.46

boutin

0.45

Activations Density 0.198%

variations and references to existing concepts or entities

No Comments

No Known Activations

variations and references to existing concepts or entities

No Comments

No Known Activations