INDEX

Explanations

patterns of whitespace or indentation in the text

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

Personendaten

-0.85

 transfieras

-0.81

oredCriteria

-0.80

AsUp

-0.75

writeFieldEnd

-0.75

parsedMessage

-0.75

afficheront

-0.72

+:+

-0.71

addContainerGap

-0.71

出版年

-0.69

POSITIVE LOGITS

 acompaña

0.51

 mochilas

0.51

 subito

0.50

 picioare

0.48

 oraș

0.47

 rând

0.47

 jelasnya

0.45

maßen

0.44

 scritta

0.44

 oberen

0.44

Activations Density 0.005%

patterns of whitespace or indentation in the text

No Comments

No Known Activations