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

 escuchado

-0.50

 determinación

-0.47

 jueces

-0.47

 levier

-0.46

 useAppContext

-0.46

 juzg

-0.45

 mentiras

-0.44

 oído

-0.44

 rostros

-0.44

 pengurus

-0.44

POSITIVE LOGITS

 kasarigan

0.54

 հղումներ

0.53

moveToFirst

0.52

bootstrapcdn

0.50

racite

0.49

/#{

0.48

-/**

0.48

endpush

0.48

 Walkover

0.47

 Boron

0.47

Activations Density 0.000%

No Known Activations

This feature has no known activations.