INDEX

Explanations

No Explanations Found

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

complexContent

-1.62

 مشين

-0.42

alsa

-0.39

Vanjske

-0.38

Manbalar

-0.36

 agarose

-0.35

isburg

-0.35

orro

-0.35

pulumi

-0.34

 plebis

-0.34

POSITIVE LOGITS

Activités

3.55

Activité

2.86

Erstellt

2.09

Carriera

1.35

Tikang

1.27

currentColor

1.19

Autres

1.07

Coordonnées

1.06

出版年

1.06

httphttps

0.93

Activations Density 0.000%

No Known Activations

This feature has no known activations.