INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 9-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_9/width_131k/average_l0_121

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.9.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

 Turquía

-0.54

 électro

-0.51

 Câmara

-0.47

 Irán

-0.44

 cár

-0.44

 península

-0.44

 parís

-0.43

adeloupe

-0.42

 Mexique

-0.42

 expéri

-0.41

POSITIVE LOGITS

hline

0.70

 intptr

0.61

HasAnnotation

0.60

[toxicity=0]

0.56

PSA

0.56

CloseOperation

0.56

<bos>

0.54

دانشنامهٔ

0.54

grie

0.54

 فريبيس

0.52

Activations Density 0.000%

No Known Activations

This feature has no known activations.