INDEX

Explanations

names of individuals, particularly those related to leadership or recognition in various 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

 AttributeSet

-0.74

ंदीखरीदारी

-0.68

 Италијани

-0.66

rungsseite

-0.55

 Signalez

-0.53

NameInMap

-0.52

 ujednoznacz

-0.52

舺

-0.52

 "..\..\

-0.51

Tembelea

-0.51

POSITIVE LOGITS

 said

0.55

 explained

0.48

 says

0.46

 explains

0.42

 forklar

0.42

said

0.41

 wyja

0.40

 described

0.39

 explain

0.39

 stated

0.37

Activations Density 0.025%

names of individuals, particularly those related to leadership or recognition in various contexts

No Comments

No Known Activations