INDEX

Explanations

references to childbirth and familial relationships

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

ItemBackground

-0.54

verwijspagina

-0.52

Remaining

-0.45

 ehemalige

-0.43

 cref

-0.42

 informée

-0.42

 ivelany

-0.42

 Former

-0.42

 former

-0.42

Former

-0.41

POSITIVE LOGITS

 baby

0.80

 babies

0.69

 Baby

0.68

baby

0.66

Baby

0.62

 birth

0.60

 BABY

0.58

Babies

0.56

 Babies

0.55

 newborn

0.54

Activations Density 0.118%

references to childbirth and familial relationships

No Comments

No Known Activations