INDEX

Explanations

the word "is" or variations of the verb "to be."

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

is a/an description

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

PerformLayout

-0.73

Personendaten

-0.73

NameInMap

-0.72

лтемелер

-0.70

Aktualisiert

-0.63

Hochspringen

-0.63

 pinulongan

-0.63

 Italijanski

-0.59

 esternos

-0.58

 OMITBAD

-0.58

POSITIVE LOGITS

is

0.66

 merupakan

0.59

0.57

an

0.56

 adalah

0.52

是一个

0.52

 typically

0.50

is

0.50

 also

0.49

是一

0.49

Activations Density 0.093%

the word "is" or variations of the verb "to be."

is a/an description

No Comments

No Known Activations