INDEX

Explanations

key roles and positions related to leadership and authority

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

specific roles and rankings

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

featureID

-0.54

 Gaulle

-0.44

cortico

-0.44

crí

-0.43

͞

-0.42

cestry

-0.41

 clearColor

-0.41

務省

-0.41

GLFW

-0.40

unen

-0.40

POSITIVE LOGITS

 AssemblyCulture

0.52

Enllaços

0.42

 replacement

0.41

oneofs

0.41

ValueGenerated

0.40

føl

0.39

exitRule

0.39

setVerticalGroup

0.38

 replace

0.38

 Normdatei

0.37

Activations Density 0.055%

key roles and positions related to leadership and authority

specific roles and rankings

No Comments

No Known Activations