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

 totales

-0.39

utives

-0.38

hiculo

-0.37

 gentlemen

-0.37

ReusableCell

-0.35

 بيها

-0.35

 financeiro

-0.35

principalTable

-0.34

 efectivo

-0.34

 merit

-0.33

POSITIVE LOGITS

UserScript

0.59

 AssemblyVersion

0.58

httphttps

0.56

 &___

0.52

 disambiguazione

0.50

ItemBackground

0.50

addCriterion

0.48

CastException

0.47

Character

0.47

 HttpNotFound

0.47

Activations Density 0.000%

No Known Activations

This feature has no known activations.