INDEX

Explanations

phrases indicating compilation or organization of information

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

created lists or guides

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

Географиясе

-0.65

kuuta

-0.48

 bambú

-0.47

 extranjero

-0.43

SharedDtor

-0.41

 embarazadas

-0.41

Життєпис

-0.41

Pratique

-0.40

 lendemain

-0.40

 jefes

-0.39

POSITIVE LOGITS

bootstrapcdn

0.53

EXTERN

0.52

rzost

0.52

 swatch

0.48

 Loaf

0.47

ategorised

0.47

 curated

0.47

matchCondition

0.47

hahn

0.47

webElement

0.46

Activations Density 0.016%

phrases indicating compilation or organization of information

created lists or guides

No Comments

No Known Activations