INDEX

Explanations

phrases related to cryptographic concepts and techniques

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

embedded secret keys

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

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

hit

-0.29

TestingModule

-0.27

لينكات

-0.27

exitRule

-0.27

참고

-0.26

 great

-0.26

 westward

-0.25

 namanya

-0.24

 shall

-0.23

PreferredItem

-0.23

POSITIVE LOGITS

 watermark

0.79

 clearColor

0.67

watermark

0.65

 nahilalakip

0.65

 ddelwed

0.63

 disambiguazione

0.62

parsedMessage

0.61

 imagui

0.61

TagMode

0.61

Inserted

0.60

Activations Density 0.011%

phrases related to cryptographic concepts and techniques

embedded secret keys

No Comments

No Known Activations