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

 Comb

-0.48

ECONDS

-0.43

 carve

-0.42

Comb

-0.42

 indo

-0.41

깐

-0.39

 carved

-0.38

!*\

-0.38

 hindsight

-0.38

|[]

-0.38

POSITIVE LOGITS

toBeTruthy

4.59

toBeDefined

1.17

 HasFactory

0.96

 HttpNotFound

0.88

toMatchSnapshot

0.83

toEqual

0.75

 '\\;'

0.75

GOTREF

0.68

 ModelExpression

0.67

KURZBESCHREIBUNG

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.