INDEX

Explanations

references to news-related topics or entities

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

news and following characters

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 9-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_9/width_131k/average_l0_121

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.9.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

躇

-0.48

 يتيمه

-0.48

Cuidado

-0.47

cuidado

-0.45

damente

-0.45

MLLoader

-0.45

 Zeiten

-0.44

 wesentlichen

-0.44

IMPORTED

-0.44

vailable

-0.43

POSITIVE LOGITS

news

2.34

News

1.92

NEWS

1.63

 News

1.38

 news

1.27

 NEWS

1.27

ニュース

0.90

 뉴스

0.87

新闻

0.87

ews

0.85

Activations Density 0.004%

references to news-related topics or entities

news and following characters

No Comments

No Known Activations