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

Données

-1.05

Попис

-0.62

httphttps

-0.54

setVerticalGroup

-0.52

 itinéraires

-0.52

LookAnd

-0.50

zzleHttp

-0.50

Aholisi

-0.48

inSlope

-0.47

stylers

-0.47

POSITIVE LOGITS

 address

0.93

 مشين

0.85

 pageContext

0.82

 Address

0.74

 adress

0.68

address

0.67

 addresses

0.66

 addressed

0.66

 addressing

0.61

 ADDRESS

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.