INDEX

Explanations

references to sources or origins in 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

 héro

-0.40

 quim

-0.39

 quæ

-0.38

-0.37

-0.35

 Verhält

-0.34

 forhold

-0.34

 educativas

-0.34

 gâteau

-0.34

 plads

-0.34

POSITIVE LOGITS

から

0.88

帖最后由

0.86

 から

0.85

부터

0.83

From

0.83

FROM

0.82

from

0.82

 From

0.80

ⓧ

0.79

fromnode

0.78

Activations Density 0.003%

references to sources or origins in text

No Comments

No Known Activations