INDEX

Explanations

references to secrecy and things that are concealed

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

"hidden" and words after "hidden"

np_acts-logits-general · gemini-2.0-flash

hidden state, layers, or hunger

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 22-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_22/width_16k/average_l0_72

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.22.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

 بيها

-0.72

WindowConstants

-0.67

ImageContext

-0.66

AdapterView

-0.62

setVerticalGroup

-0.61

endphp

-0.61

 égard

-0.57

 Berthe

-0.57

 internetowa

-0.55

utilisons

-0.55

POSITIVE LOGITS

 behind

0.89

behind

0.77

 secrets

0.76

MethodManager

0.72

away

0.69

 secret

0.66

Secrets

0.65

Behind

0.65

 BEHIND

0.65

hid

0.64

Activations Density 0.204%

references to secrecy and things that are concealed

"hidden" and words after "hidden"

hidden state, layers, or hunger

No Comments

No Known Activations