INDEX

Explanations

references to spam or unsolicited content

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

spam and advertising

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

Re

-0.46

AttributeSet

-0.43

 QtCore

-0.43

 otomatig

-0.41

Re

-0.40

ClientRect

-0.39

حياته

-0.39

RTEX

-0.38

gliere

-0.38

 kép

-0.38

POSITIVE LOGITS

 Spam

1.22

 spam

1.20

Spam

1.15

spam

1.10

 SPAM

1.01

 spamming

0.85

 ainfi

0.76

 propOrder

0.75

 mauva

0.69

 ADVERTISING

0.63

Activations Density 0.002%

references to spam or unsolicited content

spam and advertising

No Comments

No Known Activations