INDEX

Explanations

punctuation marks or symbols often used in formatting or coding contexts

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

code fragments and comments

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

transQ

-0.69

ﷺ

-0.56

abstractmethod

-0.55

AutoField

-0.55

 Biôgrafia

-0.54

medriver

-0.53

GenerationType

-0.50

AndroidJUnit

-0.50

ExtendWith

-0.49

astify

-0.49

POSITIVE LOGITS

 Dés

0.46

 surfaces

0.36

 Dumb

0.35

 vertes

0.35

<h1>

0.35

Clusters

0.34

()]

0.33

 rime

0.32

Drink

0.32

Ծանոթ

0.32

Activations Density 0.001%

punctuation marks or symbols often used in formatting or coding contexts

code fragments and comments

No Comments

No Known Activations