INDEX

Explanations

references to hidden or latent concepts and mysteries

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

"hidden" or "secret"

np_acts-logits-general · gemini-2.0-flash

hidden or secret concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 18-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_18/width_16k/average_l0_74

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.18.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

Diz

-0.56

 pageable

-0.55

 بيها

-0.53

CWE

-0.52

iented

-0.51

 licences

-0.51

ahr

-0.51

cdt

-0.51

 Apostles

-0.51

ImageContext

-0.50

POSITIVE LOGITS

 secret

1.07

 hidden

1.03

hid

1.01

 hide

1.01

 hiding

0.99

 Hidden

0.98

 Secret

0.94

HIDDEN

0.93

MethodManager

0.93

 secrets

0.92

Activations Density 0.322%

references to hidden or latent concepts and mysteries

"hidden" or "secret"

hidden or secret concepts

No Comments

No Known Activations