INDEX

Explanations

proper nouns, particularly names of individuals and organizations

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

names followed by reporting verbs

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

Datuak

-0.68

__':

-0.62

一来

-0.56

initWithFrame

-0.54

mtrl

-0.52

__":

-0.51

writeField

-0.49

addPreferredGap

-0.48

KommentareTeilen

-0.47

WriteAttribute

-0.46

POSITIVE LOGITS

 wrote

0.46

 jullie

0.45

 says

0.44

Hentet

0.44

Lähteet

0.43

 חיצוניים

0.41

 said

0.40

 tweeted

0.39

 Walkover

0.38

GOTREF

0.38

Activations Density 0.025%

proper nouns, particularly names of individuals and organizations

names followed by reporting verbs

No Comments

No Known Activations