INDEX

Explanations

technical terms and components related to computing and programming

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

itinéraire

-0.47

 Ball

-0.40

Ball

-0.39

 createState

-0.36

 форму

-0.34

发表于

-0.33

autorest

-0.33

 ball

-0.33

kuk

-0.33

 оку

-0.33

POSITIVE LOGITS

Chham

0.65

featureID

0.57

bcryptjs

0.53

SharedDtor

0.50

rsiniz

0.50

Jeografia

0.50

seamnă

0.50

 NUKAT

0.48

 henvisninger

0.47

 leiding

0.47

Activations Density 0.122%

technical terms and components related to computing and programming

No Comments

No Known Activations