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

IntoConstraints

-0.73

ngdoc

-0.64

 initComponents

-0.64

IndentedString

-0.62

 CascadeType

-0.61

 "..\..\..\

-0.57

 BorderSide

-0.57

 pleaſure

-0.56

باردا

-0.55

parsedMessage

-0.55

POSITIVE LOGITS

toxicity

0.52

 people

0.41

EnableWeb

0.38

 دیکھیے

0.36

 serpiente

0.33

 nation

0.33

saraba

0.31

 populous

0.31

UVWXYZ

0.31

 taxpayers

0.30

Activations Density 0.000%

No Known Activations

This feature has no known activations.