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

 mantenido

-0.47

 conmigo

-0.46

Shell

-0.44

 gerekiyor

-0.43

 pudieran

-0.42

 puedas

-0.41

 prohibido

-0.41

 soutenu

-0.41

 detuvo

-0.41

 Shell

-0.39

POSITIVE LOGITS

 snippetHide

1.73

дан

1.49

setHorizontal

1.19

httphttps

0.94

Datuak

0.91

Personendaten

0.87

GEBURTSDATUM

0.79

дана

0.77

twimg

0.77

 propOrder

0.75

Activations Density 0.000%

No Known Activations

This feature has no known activations.