INDEX

Explanations

proper nouns, particularly names of people

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

first name followed by last name

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

hésite

-0.43

ExtendWith

-0.43

postmedia

-0.39

zubauen

-0.39

 intptr

-0.38

 przew

-0.38

empêcher

-0.37

 telefoon

-0.37

 Polri

-0.37



-0.37

POSITIVE LOGITS

ۜ

0.65

Ola

0.57

IsContent

0.52

MenuGroup

0.51

opsida

0.51

archiviato

0.50

 Mann

0.50

awaiter

0.50

Ira

0.49

 zuſammen

0.48

Activations Density 0.070%

proper nouns, particularly names of people

first name followed by last name

No Comments

No Known Activations