INDEX

Explanations

phrases related to information disclosure and privacy concerns

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

privacy and classified information

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

AndEndTag

-0.59

 تضيفلها

-0.57

TextBoxColumn

-0.54

MessageOf

-0.53

EndContext

-0.52

culously

-0.52

gyz

-0.50

]")]

-0.50

lbrakk

-0.50

–¿

-0.50

POSITIVE LOGITS

 confidential

0.43

toHaveBeenCalled

0.33

 confidentiality

0.31

明日

0.31

setViewportView

0.30

kunft

0.30

プライ

0.30

protoimpl

0.29

 personal

0.29

布

0.29

Activations Density 0.040%

phrases related to information disclosure and privacy concerns

privacy and classified information

No Comments

No Known Activations