INDEX

Explanations

phrases indicating capability or possibility involving modal verbs

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

German and Scandinavian language fragments

np_acts-logits-general · gemini-2.0-flash

wenn, vil, hvis followed by du

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 24-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_24/width_16k/average_l0_73

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.24.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

AutoresizingMask

-0.45

 AssemblyCulture

-0.45

:✨

-0.45

Den

-0.45

 متعلقه

-0.44

 Anson

-0.43

Den

-0.43

ResponseEntity

-0.43

ể

-0.42

นั้น

-0.42

POSITIVE LOGITS

es

1.15

sie

1.11

man

1.06

 sich

1.00

wir

0.96

ich

0.92

du

0.83

Sie

0.78

er

0.71

 mans

0.66

Activations Density 0.058%

phrases indicating capability or possibility involving modal verbs

German and Scandinavian language fragments

wenn, vil, hvis followed by du

No Comments

No Known Activations