INDEX

Explanations

phrases indicating user engagement or interaction with products and experiences

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

you feel

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

 voici

-0.47

ksamkeit

-0.38

oarece

-0.38

 toekomst

-0.37

未能

-0.37

 Zusätzlich

-0.37

 afgelopen

-0.37

 continúan

-0.36

目前

-0.36

 bezoekers

-0.36

POSITIVE LOGITS

complexContent

0.67

SequentialGroup

0.57

 հղումներ

0.50

 yourself

0.49

IntOverflow

0.48

Gweler

0.48

 препратки

0.48

yorsunuz

0.48

tubers

0.47

IntoConstraints

0.47

Activations Density 0.065%

phrases indicating user engagement or interaction with products and experiences

you feel

No Comments

No Known Activations