INDEX

Explanations

evidence of identity or labeling, often through the use of specific formatting or categorization

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

Model questions or outputs

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

parsedMessage

-0.67

 EconPapers

-0.66

SharedCtor

-0.62

afficheront

-0.62

Билгалдахарш

-0.60

WebElementEntity

-0.59

setVerticalGroup

-0.58

oredCriteria

-0.57

enterOuterAlt

-0.57

الإنجليزية

-0.56

POSITIVE LOGITS

rigin

0.35

fig

0.34

 hang

0.33

ourge

0.32

espejo

0.32

arios

0.32

 Hang

0.32

form

0.32

tu

0.32

ker

0.31

Activations Density 0.001%

evidence of identity or labeling, often through the use of specific formatting or categorization

Model questions or outputs

No Comments

No Known Activations