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

 مرئيه

-0.60

Aiheesta

-0.54

 Audiodateien

-0.51

protetor

-0.51

PMailer

-0.49

XmlAccessorType

-0.49

 שוליים

-0.49

WillAppear

-0.48

cuador

-0.47

SuppressLint

-0.47

POSITIVE LOGITS

Date

0.46

 kasarigan

0.45

 date

0.43

 Date

0.40

date

0.36

permalink

0.35

 retro

0.35

 setDate

0.34

DATE

0.33

findpost

0.33

Activations Density 0.000%

No Known Activations

This feature has no known activations.