INDEX

Explanations

the word "of"

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

instead of

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

 kveld

-0.48

 municipalité

-0.43

 arbet

-0.42

ResumeLayout

-0.41

postmedia

-0.40

arbeid

-0.40

IEnumerable

-0.39

 Werden

-0.39

 prisma

-0.38

SessionManager

-0.38

POSITIVE LOGITS

enumi

0.70

 للمعارف

0.69

enumii

0.68

 Мексичка

0.66

 }}}{

0.66

ThroughAttribute

0.65

IsMutable

0.65

 فريبيس

0.65

XmlAccessType

0.65

 THAN

0.61

Activations Density 0.006%

the word "of"

instead of

No Comments

No Known Activations