INDEX

Explanations

terms related to safety and personal security

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

essential, safety, personal, prevention, host, intellectual, life, criminal, confidentiality

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

Grüße

-0.39

Rad

-0.39

 Fink

-0.36

Grüsse

-0.36

пря

-0.36

ola

-0.35

 Carrasco

-0.35

werfen

-0.34

 Übers

-0.34

OLA

-0.34

POSITIVE LOGITS

 للاسماء

0.86

TagMode

0.75

rungsseite

0.74

 ویکی‌پدی

0.65

Cyfarwyddwr

0.61

 Chwiliwch

0.60

 OFDb

0.58

存于互联网档案馆

0.57

Демографія

0.56

typeorm

0.56

Activations Density 0.034%

terms related to safety and personal security

essential, safety, personal, prevention, host, intellectual, life, criminal, confidentiality

No Comments

No Known Activations