INDEX

Explanations

numerical or coded sequences

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

numerical values associated with contact information or addresses

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

company addresses and USA

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 20-gemmascope-res-65k

Configuration

google/gemma-scope-2b-pt-res/layer_20/width_65k/average_l0_114

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

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

onViewCreated

-0.86

 CreateTagHelper

-0.72

Rhestr

-0.71

 otomatig

-0.70

 Italijanski

-0.67

 تضيفلها

-0.67

AndEndTag

-0.66

 <=",

-0.65

wüns

-0.61

 referrerpolicy

-0.61

POSITIVE LOGITS

я

0.52

 understanding

0.47

duga

0.47

ApplicationTests

0.47

 INVENTION

0.46

 dolore

0.44

understanding

0.44

enterio

0.42

XRD

0.42

STAGE

0.42

Activations Density 0.016%

numerical or coded sequences

numerical values associated with contact information or addresses

company addresses and USA

No Comments

No Known Activations