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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

omiast

-0.49

 betweenstory

-0.48

 Administrativna

-0.47

ImageContext

-0.46

DockStyle

-0.45

BagConstraints

-0.44

 absorción

-0.42

 ModelExpression

-0.42

XmlAttribute

-0.40

 initComponents

-0.39

POSITIVE LOGITS

Linq

1.00

 vectorielle

0.53

 vectorielles

0.46

 IndexPath

0.44

_,

0.44

Искәрмәләр

0.42

piercing

0.41

 Roskov

0.39

 referenties

0.39

 мәкал

0.39

Activations Density 0.000%

No Known Activations

This feature has no known activations.