INDEX

Explanations

terms and expressions associated with feelings of offense or offending statements

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

causing offense or objection

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 9-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_9/width_131k/average_l0_121

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.9.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

Autoritní

-0.64

Cyfeiriadau

-0.62

ImageContext

-0.60

webElementGuid

-0.60

KommentareTeilen

-0.60

makeConstraints

-0.59

 الحره

-0.58

Kaieteur

-0.57

EndInit

-0.56

Personendaten

-0.55

POSITIVE LOGITS

 offend

1.36

 offended

1.30

 offense

1.27

 offence

1.22

 offensive

1.17

 offending

1.17

 Offense

1.06

offensive

1.00

 offenses

0.97

 offences

0.91

Activations Density 0.009%

terms and expressions associated with feelings of offense or offending statements

causing offense or objection

No Comments

No Known Activations