INDEX

Explanations

numeric values related to addresses or coordinates

oai_token-act-pair · gpt-4o-mini Triggered by @bot

specialized technical terms

np_acts-logits-general · gemini-2.5-flash-lite

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

 للاسماء

-0.60

 nahilalakip

-0.56

 <<<<<<<<<<<<<<

-0.51

Kanpo

-0.48

مراجع

-0.47

日閲覧

-0.46

 chemists

-0.45

Билгалдахарш

-0.45

 hoor

-0.45

})*/

-0.45

POSITIVE LOGITS

0.52

yntaxException

0.48

URLException

0.40

 отделение

0.39

DataBinding

0.39

VersionUID

0.37

SharedDtor

0.36

 fréquent

0.36

borderSide

0.36

 IBOutlet

0.36

Activations Density 0.001%

numeric values related to addresses or coordinates

specialized technical terms

No Comments

No Known Activations