INDEX

Explanations

phrases related to procedural or structured content

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

mathematical symbols

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

 multiplic

-0.41

Stdout

-0.40

eradish

-0.39

squ

-0.39

bosky

-0.39

jsoup

-0.39

 '/';

-0.39

ulose

-0.38

iada

-0.38

err

-0.38

POSITIVE LOGITS

 defaultstate

0.64

Hentet

0.60

पया

0.57

MessageOf

0.55

 Biôgrafia

0.52

expandindo

0.51

 noDo

0.50

MessageTagHelper

0.50

 AssemblyCulture

0.49

EndInit

0.49

Activations Density 0.011%

phrases related to procedural or structured content

mathematical symbols

No Comments

No Known Activations