INDEX

Explanations

references related to specific locations or buildings

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

new courtyard

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

Negative Logits

RectangleBorder

-0.85

 createSlice

-0.75

 autorytatywna

-0.75

 CreateTagHelper

-0.73

Демографія

-0.73

 onOptions

-0.72

̈́

-0.72

 مرئيه

-0.72

PreferredItem

-0.69

 unknownFields

-0.68

POSITIVE LOGITS

 Fine

0.53

Fine

0.52

 facilities

0.52

gym

0.50

 library

0.48

facilities

0.48

 fine

0.47

Facilities

0.46

Visual

0.46

 Facilities

0.46

Activations Density 0.066%

references related to specific locations or buildings

new courtyard

No Known Activations