INDEX

Explanations

No Explanations Found

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

 selamat

-0.38

 Pristupljeno

-0.38

 sepatu

-0.36

 tetto

-0.36

 géographique

-0.35

 caractéristique

-0.35

 démocratique

-0.35

 hâte

-0.35

sprecher

-0.34

 pangan

-0.34

POSITIVE LOGITS

tagHelperRunner

0.91

 EconPapers

0.84

 invokingState

0.79

IsContent

0.76

SourceChecksum

0.72

 autorytatywna

0.68

CppMethod

0.67

ScopeManager

0.67

 للمعارف

0.66

 ujednoznacz

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.