INDEX

Explanations

phrases indicating uncertainty or potential consequences in legal or environmental contexts

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

five cases, years, condition, integrity

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

 surla

-0.45

trale

-0.45

Autoritní

-0.45

 linkovi

-0.42

nit

-0.42

onite

-0.40

zul

-0.39

Nit

-0.39

 altså

-0.39

 breadcrumbs

-0.39

POSITIVE LOGITS

 CURIAM

0.51

 onCancelled

0.50

ftagPool

0.50

frameCount

0.45

 فريبيس

0.44

 género

0.44

 względu

0.43

rungsseite

0.43

 findViewById

0.43

 classNames

0.43

Activations Density 0.016%

phrases indicating uncertainty or potential consequences in legal or environmental contexts

five cases, years, condition, integrity

No Comments

No Known Activations