INDEX

Explanations

code-related structures and commands

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

connecting paths to each other

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

WebControls

-0.48

 Agent

-0.43

HAC

-0.42

 agents

-0.41

 keyCode

-0.40

mapStateToProps

-0.39

 Marx

-0.39

Bun

-0.39

 circum

-0.39

 Cena

-0.39

POSITIVE LOGITS

 Wikiseite

0.64

ंदीखरीदारी

0.60

 للاسماء

0.58

 kasarigan

0.58

pośred

0.55

NameInMap

0.54

 الحره

0.53

 asupra

0.52

 invokingState

0.52

 către

0.50

Activations Density 0.049%

code-related structures and commands

connecting paths to each other

No Comments

No Known Activations