INDEX

Explanations

references to identity or personal confirmation

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

本人確認

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

parsedMessage

-0.82

 wikipagina

-0.79

Билгалдахарш

-0.77

verwijspagina

-0.75

✨:

-0.75

%;">

-0.72

 Fauci

-0.72

 ligiloj

-0.72

ValueStyle

-0.70

eclampsia

-0.70

POSITIVE LOGITS

本

0.92

本

0.84

본

0.66

の本

0.62

在本

0.59

0.50

Base

0.50

Bản

0.46

본

0.45

ben

0.45

Activations Density 0.004%

references to identity or personal confirmation

本人確認

No Comments

No Known Activations