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

is

-0.31

 call

-0.30

thalt

-0.29

 fries

-0.29

-0.28

 целях

-0.28

met

-0.27

her

-0.27

lijks

-0.26

JsonPropertyName

-0.26

POSITIVE LOGITS

WebpackPlugin

5.44

webpack

0.81

Webpack

0.79

prefixer

0.76

 AssemblyTitle

0.75

 webpack

0.74

IntoConstraints

0.71

rungsseite

0.69

 BrowserModule

0.68

ſcher

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.