INDEX

Explanations

legal terms and concepts related to fraud and misconduct

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

rise to the level

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

StartTag

-0.43

PreferredItem

-0.43

autorelease

-0.41

 estekak

-0.38

oredCriteria

-0.38

ToBounds

-0.38

aarrggbb

-0.38

 Triple

-0.38

netinet

-0.38

 ویکی‌پدیای

-0.37

POSITIVE LOGITS

serious

0.54



0.52

 autorytatywna

0.51

enough

0.50

 serious

0.47

major

0.46

 Chwiliwch

0.46

Serious

0.46

 enough

0.44

significant

0.44

Activations Density 0.164%

legal terms and concepts related to fraud and misconduct

rise to the level

No Comments

No Known Activations