INDEX

Explanations

numerical values or references to quantities

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

throughout the rest of Europe

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

RefNanny

-0.55

 Италијани

-0.54

 архивлан

-0.53

 continúas

-0.53

 Scramble

-0.49

ittens

-0.49

 Volcan

-0.49

BUN

-0.48

 propOrder

-0.48

elemField

-0.47

POSITIVE LOGITS

tvguide

0.34

 ultimately

0.34

Ultimately

0.33

ließlich

0.33

#+#

0.33

ValueStyle

0.33

posedge

0.31

 addirittura

0.31

 furthermore

0.31

 برخ

0.31

Activations Density 0.001%

numerical values or references to quantities

throughout the rest of Europe

No Comments

No Known Activations