INDEX

Explanations

references to individuals and groups in contexts of authority and responsibility

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

BeginContext

-0.47

inox

-0.45

IBU

-0.45

 Dressing

-0.44

 dressing

-0.44

TagName

-0.43

ternal

-0.43

liferay

-0.43

lasty

-0.42

exion

-0.42

POSITIVE LOGITS

 المعيارى

0.54

 pourtant

0.43

 whose

0.43

ModelBuilder

0.41

 cuyas

0.41

 cuyos

0.41

BASEPATH

0.40

ukunft

0.40

 cuya

0.38

fromCharCode

0.38

Activations Density 0.059%

references to individuals and groups in contexts of authority and responsibility

No Comments

No Known Activations