INDEX

Explanations

terms related to windows and objects interacting with or cleaning them

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

occurring before "window" or "glass"

np_acts-logits-general · gemini-2.0-flash

window glass

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-16k

Configuration

google/gemma-scope-2b-pt-res/layer_20/width_16k/average_l0_71

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 referenties

-0.58

@[+][

-0.55

NameInMap

-0.51

flich

-0.49

 StringTokenizer

-0.49

lief

-0.46

cun

-0.46

olism

-0.45

 Whigs

-0.45

ogia

-0.45

POSITIVE LOGITS

 window

2.23

 windows

2.19

 glass

2.06

 Window

1.99

window

1.95

 WINDOW

1.86

Window

1.83

windows

1.75

 Glass

1.72

glass

1.70

Activations Density 0.244%

terms related to windows and objects interacting with or cleaning them

occurring before "window" or "glass"

window glass

No Comments

No Known Activations