INDEX

Explanations

terms associated with journalism and media professionals

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

news reporter names

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

 betweenstory

-0.66

ronpa

-0.59

 IFTT

-0.57



-0.57

󠁴

-0.57

Vidite

-0.54

vician

-0.52

dhism

-0.52

 ब्रेकडाउन

-0.52

 Lijst

-0.51

POSITIVE LOGITS

RectangleBorder

0.48

цездатний

0.45

 يتيمه

0.43

 Cloud

0.40

angu

0.39

 Recorder

0.38

Mur

0.37

Alt

0.36

 Collider

0.36

 SqlParameter

0.36

Activations Density 0.002%

terms associated with journalism and media professionals

news reporter names

No Comments

No Known Activations