INDEX

Explanations

phrases indicating competitive status or positioning in a context, particularly in sports or debates

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

winning and losing comparison

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

AndEndTag

-0.57

 Paglinawan

-0.53

Hochspringen

-0.52

WithIdentifier

-0.50

 indígen

-0.48

enregist

-0.47

 médié

-0.47

нгред

-0.47

 inovasi

-0.47

 ſche

-0.46

POSITIVE LOGITS

Бахар

0.49

WaitGroup

0.39

AppCompat

0.35

 newOwner

0.35

endpush

0.33

AMPLES

0.32

NetworkInfo

0.32

QMetaType

0.32

imb

0.31

OrEqual

0.31

Activations Density 0.021%

phrases indicating competitive status or positioning in a context, particularly in sports or debates

winning and losing comparison

No Comments

No Known Activations