INDEX

Explanations

instances of violent actions or attacks described in the text

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

agent of harmful action

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

 للاسماء

-0.50

 AssemblyCulture

-0.47

Tikang

-0.47

 autorytatywna

-0.42

你想

-0.41

 uParam

-0.41

 dreapta

-0.40

STA

-0.40

 nakalista

-0.40

mandu

-0.39

POSITIVE LOGITS

__(/*!

0.56

 bullies

0.52

 predators

0.47

 evil

0.45

 hostile

0.42

 enemy

0.42

 astore

0.42

/**

0.42

 brokers

0.42

 gangs

0.41

Activations Density 0.056%

instances of violent actions or attacks described in the text

agent of harmful action

No Comments

No Known Activations