INDEX

Explanations

references to authority figures or positions of power

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

preceding a name or title

np_acts-logits-general · gemini-2.0-flash

witnesses and reports

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-16k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

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

IntoConstraints

-0.57

MemoryWarning

-0.51

بوابة

-0.49

selaer

-0.49

 AssemblyCulture

-0.48

gdala

-0.47

brainly

-0.47

دانشنامهٔ

-0.47

WillAppear

-0.46

noid

-0.46

POSITIVE LOGITS

verwijspagina

0.59

InjectAttribute

0.57

PostInfinity

0.40

webElementXpaths

0.39

 swear

0.38

report

0.37

 witness

0.37

 witnesses

0.36

 report

0.36

 swears

0.35

Activations Density 0.343%

references to authority figures or positions of power

preceding a name or title

witnesses and reports

No Comments

No Known Activations