INDEX

Explanations

formatting elements and structure in text

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

document structure markers

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

SharedDtor

-0.75

JspWriter

-0.71

 Meksiku

-0.68

MigrationBuilder

-0.68

ſammen

-0.67

باردا

-0.66

niſſe

-0.62

تقاوى

-0.61

 betweenstory

-0.61

ſicht

-0.61

POSITIVE LOGITS

 Formación

0.28

})

0.28

पया

0.26

PageIndex

0.25

 Sache

0.25

 defining

0.25

 Privacidad

0.25

からです

0.24

setVerticalGroup

0.23

なのは

0.23

Activations Density 0.004%

formatting elements and structure in text

document structure markers

No Comments

No Known Activations