INDEX

Explanations

mentions of various actions and interactions in contexts involving competition or conflict

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

growing rapidly

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

and

-0.27

 décider

-0.27

\{\\

-0.23

ốm

-0.23

 Company

-0.23

 traité

-0.22

 people

-0.22

Vì

-0.22

 person

-0.22

POSITIVE LOGITS

脚注の使い方

0.91

الحياه

0.86

 kasarigan

0.84

 Geiſt

0.83

bootstrapcdn

0.81

+#+#

0.79

 disambiguazione

0.79

 Waſſer

0.78

 Geſch

0.76

 jspb

0.76

Activations Density 0.471%

mentions of various actions and interactions in contexts involving competition or conflict

growing rapidly

No Comments

No Known Activations