INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 للمعارف

-0.57

richTextPanel

-0.55

majánló

-0.54

 Arkivert

-0.54

FirebaseUser

-0.51

曖昧さ回避

-0.50

 AssemblyCulture

-0.50

RegressionTest

-0.49

EndContext

-0.48

IsContent

-0.48

POSITIVE LOGITS

ној

0.71

ArrowToggle

0.47

wapV

0.44

StoryboardSegue

0.36

 Plug

0.36

featureID

0.34

spender

0.33

Plu

0.33

Ske

0.31

Sprintf

0.31

Activations Density 0.000%

No Known Activations

This feature has no known activations.