INDEX

Explanations

phrases and concepts related to names and terminology

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

name, title, and term

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

 insuffisamment

-0.52

 pouvoit

-0.48

featureID

-0.48

tawesome

-0.48

elemField

-0.47

 kasarigan

-0.46

évaluateur

-0.46

 للاسماء

-0.45

 Meksiku

-0.45

 intervi

-0.44

POSITIVE LOGITS

 name

0.54

yntaxException

0.45

GOTREF

0.40

曖昧さ回避

0.39

parsedMessage

0.38

 이름

0.37

 naming

0.37

 CreateTagHelper

0.37

randomUUID

0.36

 named

0.36

Activations Density 0.044%

phrases and concepts related to names and terminology

name, title, and term

No Comments

No Known Activations