INDEX

Explanations

specific identifiers and mentions of individuals, especially in specific contexts

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

Quitte

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

It detects tokens that occur at the start of a turn or sentence (turn-/line-initial tokens, like speaker labels or sentence-beginning words).

oai_token-act-pair · gpt-5-mini Triggered by @vetterc0

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

المناصب

-0.65

 autorytatywna

-0.57

Havolalar

-0.54

 Мексичка

-0.48

WebElementEntity

-0.48

WithMany

-0.48

AndEndTag

-0.48



-0.47

Datuak

-0.47

 useAppContext

-0.46

POSITIVE LOGITS

Erm

0.38

InsertCommand

0.37

шру

0.37

 Sweep

0.37

bows

0.36

 compro

0.36

ANSI

0.36

SMD

0.36

Wallace

0.35

Attention

0.35

Activations Density 0.043%

specific identifiers and mentions of individuals, especially in specific contexts

Quitte

It detects tokens that occur at the start of a turn or sentence (turn-/line-initial tokens, like speaker labels or sentence-beginning words).

No Comments

No Known Activations

specific identifiers and mentions of individuals, especially in specific contexts

Quitte

It detects tokens that occur at the start of a turn or sentence (turn-/line-initial tokens, like speaker labels or sentence-beginning words).

No Comments

No Known Activations