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

<eos>

-0.30

即

-0.28

 conna

-0.27

 "'");

-0.26

Ny

-0.26

||}

-0.25

Ny

-0.25

::$_

-0.25

 иму

-0.25

is

-0.24

POSITIVE LOGITS

ArgumentParser

5.72

 ModelExpression

1.11

 AssemblyTitle

1.10

 defaultstate

1.04

ScopeManager

1.04

 argparse

1.02

 autorytatywna

1.01

 CreateTagHelper

1.01

 initComponents

1.01

parsedMessage

0.99

Activations Density 0.000%

No Known Activations

This feature has no known activations.