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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Arzt

-0.43

enoord

-0.36

recated

-0.35

[*]

-0.35

 Ahnung

-0.34

 alent

-0.33

 surla

-0.33

KURZBESCHREIBUNG

-0.32

fficients

-0.32

ถม

-0.32

POSITIVE LOGITS

GLO

2.17

GLO

1.43

Glo

1.14

Glo

1.07

glo

1.04

 Globe

0.92

Globe

0.91

glo

0.91

GLOB

0.78

globe

0.73

Activations Density 0.000%

No Known Activations

This feature has no known activations.