INDEX

Explanations

email addresses and related formatting

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

email and contact information

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

治

-0.47

gez

-0.44

 linkovi

-0.44

SourceChecksum

-0.42

agissait

-0.40

制

-0.39

 initComponents

-0.38

 Routine

-0.37

 zaterdag

-0.37

 Réponses

-0.36

POSITIVE LOGITS

rungsseite

0.71

Personensuche

0.57

 оригіналу

0.50

+:+

0.46

 email

0.45

 utafitiHapana

0.42

Rhestr

0.41

email

0.41

 AssemblyCompany

0.41

󠁣

0.41

Activations Density 0.001%

email addresses and related formatting

email and contact information

No Comments

No Known Activations