INDEX

Explanations

characters or letters used in Arabic or Japanese writing systems

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

あの子

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

Margot

-0.57

 kaarangay

-0.54

 Margot

-0.53

 anſ

-0.52

 fhort

-0.52

 notebook

-0.52

DispatchToProps

-0.51

ListPage

-0.50

-0.49

 lottery

-0.49

POSITIVE LOGITS

ع

0.56

ע

0.54

 الع

0.54

 userEmail

0.50

 localObject

0.48

ע

0.47

Assad

0.47

 recensement

0.47

RTLU

0.46

createClass

0.45

Activations Density 0.006%

characters or letters used in Arabic or Japanese writing systems

あの子

No Comments

No Known Activations