INDEX

Explanations

conjugated forms of the reflexive verb "se" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_16k/average_l0_68

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

 bénévol

-0.50

vœ

-0.46

kø

-0.46

 Imperio

-0.45

 oreille

-0.45

øren

-0.43

 Danemark

-0.43

 souri

-0.43

stø

-0.42

 commandement

-0.42

POSITIVE LOGITS

be

0.80

SequentialGroup

0.79

se

0.71

 seer

0.64

Be

0.63

get

0.63

makeConstraints

0.61

are

0.60

 Beal

0.59

go

0.58

Activations Density 0.031%

conjugated forms of the reflexive verb "se" in various contexts

No Comments

No Known Activations