INDEX

Explanations

email addresses

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

email addresses and internal routing

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

New Auto-Interp

Top Features by Cosine Similarity

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

Configuration

google/gemma-scope-9b-it-res/layer_9/width_131k/average_l0_121

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.9.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

 purpoſe

-0.58

<bos>

-0.58

 pleaſure

-0.52

lapsingToolbar

-0.45

DMETHOD

-0.45

ſelves

-0.44

processable

-0.44

 anſ

-0.43

 aihe

-0.42

 innehå

-0.41

POSITIVE LOGITS

ंदीखरीदारी

0.57

 AssemblyCompany

0.57

 Normdatei

0.54

0.48

Datuak

0.48

EndGlobalSection

0.48

хьтан

0.48

enumi

0.47

 CanadaChoose

0.47

 мәкал

0.47

Activations Density 0.001%

email addresses

email addresses and internal routing

No Known Activations