INDEX

Explanations

No Explanations Found

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_33/width_131k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.33.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

AsUp

-0.58

Personendaten

-0.57

 autorytatywna

-0.54

 незавершена

-0.52

檚

-0.52

 newName

-0.46

 resolu

-0.44

 initComponents

-0.43

 onFailure

-0.42

 propOrder

-0.42

POSITIVE LOGITS

 have

0.81

are

0.72

 jestem

0.66

 myself

0.65

can

0.59

 mám

0.56

 will

0.53

 jsem

0.52

 know

0.52

had

0.52

Activations Density 0.000%

No Known Activations

This feature has no known activations.