INDEX

Explanations

No Explanations Found

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_32/width_131k/average_l0_88

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.32.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ſelf

-0.64

 ſehen

-0.59

ſſung

-0.59

되었습니다

-0.59

 Dieſe

-0.58

 pleaſure

-0.58

ſelves

-0.58

 ſoll

-0.57

 purpoſe

-0.57

 deſſen

-0.56

POSITIVE LOGITS

 ContentValues

0.36

èdia

0.36

 setuptools

0.34

 CascadeType

0.32

ingeki

0.29

tissage

0.28

usercontent

0.28

GeneratedMessage

0.28

 připo

0.28

 while

0.28

Activations Density 0.000%

No Known Activations

This feature has no known activations.