INDEX

Explanations

collaborative and interactive expressions

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

discuss or explain topics

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

 thought

-0.37

 egent

-0.35

ambién

-0.34

 Widerspruch

-0.33

 warning

-0.33

 nacionalidad

-0.33

 advertencia

-0.31

จึง

-0.31

 Ejecutivo

-0.30

 pneus

-0.30

POSITIVE LOGITS

BagLayout

0.76

хьтан

0.66

 snippetHide

0.66

thâu

0.65

 kaarangay

0.65

GOTREF

0.63

Personensuche

0.63

ंदीखरीदारी

0.61

yntaxException

0.60

 Италијани

0.59

Activations Density 0.064%

collaborative and interactive expressions

discuss or explain topics

No Comments

No Known Activations