INDEX

Explanations

references to built-in features or capabilities

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

built-in features and areas

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

 kasarigan

-0.78

Zeneca

-0.74

 laſſen

-0.68

 miniaturka

-0.66

 tartalomajánló

-0.65

 للمعارف

-0.64

ronique

-0.63

 pagină

-0.62

ellido

-0.62

$_(

-0.61

POSITIVE LOGITS

Built

1.03

 Built

1.00

 built

0.93

built

0.88

 BUILT

0.79

BUILT

0.77

builtins

0.57

build

0.54

builtin

0.54

Build

0.52

Activations Density 0.007%

references to built-in features or capabilities

built-in features and areas

No Comments

No Known Activations