INDEX

Explanations

references to web addresses and digital content

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

actually do

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

istoitu

-0.56

Initializable

-0.50

 numberOfRows

-0.50

Diwedd

-0.49

audiovisuel

-0.48

Erreferentziak

-0.48

 nonUne

-0.47

 NSCoder

-0.47

Παραπομπές

-0.46

GEBURTSDATUM

-0.46

POSITIVE LOGITS

AndEndTag

0.48

addCriterion

0.46

rungsseite

0.37

 pinulongan

0.35

localVar

0.35

----</

0.35

帖最后由

0.34

Cześć

0.33

Forest

0.33

},

0.31

Activations Density 0.216%

references to web addresses and digital content

actually do

No Comments

No Known Activations