INDEX

Explanations

elements related to media content and programming

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

news, competition, information, talk, magazine

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

 feroit

-0.52

 auroit

-0.48

initComponents

-0.47

انيف

-0.45

 demasi

-0.43

 otur

-0.42

 stanow

-0.41

LookAnd

-0.41

 pouvoit

-0.40

afficheront

-0.40

POSITIVE LOGITS

Hentet

0.47

 saites

0.46

 surla

0.44

parsedMessage

0.42

 nahilalakip

0.42

 mercantile

0.41

irical

0.41

chastic

0.40

 conglomerate

0.40

UnknownFieldSet

0.39

Activations Density 0.219%

elements related to media content and programming

news, competition, information, talk, magazine

No Comments

No Known Activations