INDEX

Explanations

No Explanations Found

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_24/width_131k/average_l0_97

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.24.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

 propOrder

-0.80

хьтан

-0.77

 مشين

-0.75

 estekak

-0.72

 queſta

-0.70

<unused28>

-0.68

<unused16>

-0.67

<unused23>

-0.67

<unused8>

-0.67

[@BOS@]

-0.67

POSITIVE LOGITS

softc

0.36

NY

0.34

www

0.33

www

0.32

NY

0.30

New

0.29

 tiba

0.28

sus

0.28

Set

0.27

New

0.27

Activations Density 0.000%

No Known Activations

This feature has no known activations.