INDEX

Explanations

phrases indicating negation or exclusion

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

not or no usage

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

featureID

-0.28

 algumas

-0.27

 nød

-0.27

 coming

-0.26

 half

-0.26

IVEREF

-0.26

 manque

-0.25

 sometimes

-0.25

 खु

-0.25

coming

-0.24

POSITIVE LOGITS

 kasarigan

0.72

Tembelea

0.69

 surla

0.66

IntoConstraints

0.64

>*/

0.62

GraphicsUnit

0.61

 cherchés

0.60

 uLocal

0.59

⮕

0.57

tagHelperRunner

0.57

Activations Density 0.177%

phrases indicating negation or exclusion

not or no usage

No Comments

No Known Activations