INDEX

Explanations

names of people and email addresses

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

names and titles

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

-0.56

----</

-0.48

ThroughAttribute

-0.46

OrBuilder

-0.46

 ostavi

-0.45

@"/

-0.44

érature

-0.43

 مرئيه

-0.43

хьтан

-0.43

)"),

-0.42

POSITIVE LOGITS

 AssemblyCompany

0.44



0.39

assertj

0.38

出版年

0.37

ulihan

0.36

bez

0.36

listdir

0.36

Yag

0.35

 insanity

0.35

ыгана

0.35

Activations Density 0.011%

names of people and email addresses

names and titles

No Comments

No Known Activations