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

 Chwiliwch

-0.98

AsUp

-0.96

aarrggbb

-0.94

 Administrativna

-0.94

AndEndTag

-0.85

 AssemblyCulture

-0.85

يكب

-0.83

StructEnd

-0.81

 surla

-0.79

writeFieldEnd

-0.79

POSITIVE LOGITS

 rağmen

0.30

 מס

0.29

太

0.28

Coords

0.28

 classified

0.28

 fire

0.28

Bur

0.27

last

0.27

 coordin

0.27

 mirror

0.27

Activations Density 0.000%

No Known Activations

This feature has no known activations.