INDEX

Explanations

proper nouns, particularly names associated with notable individuals

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

given names followed by qualifiers

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

 эк

-0.38

fly

-0.37

ApiProperty

-0.37

 eksp

-0.37

it

-0.36

 называется

-0.36

xa

-0.35

AutoresizingMask

-0.35

XCTAssertEqual

-0.34

 BoxFit

-0.34

POSITIVE LOGITS

tanleria

0.72

 Autorizaciones

0.66

 MenuView

0.63

 betweenstory

0.61

awakeFromNib

0.60

EDEFAULT

0.58

 resourceCulture

0.57

PreInfinity

0.56

 hObject

0.56

 Cathy

0.55

Activations Density 0.034%

proper nouns, particularly names associated with notable individuals

given names followed by qualifiers

No Comments

No Known Activations