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

transQ

-1.26

GEBURTSDATUM

-1.13

 betweenstory

-1.12

Autoritní

-1.11

featureID

-1.00

 Biôgrafia

-0.97

GEBURTS

-0.96

fromnode

-0.95

queryInterface

-0.94

 queſta

-0.93

POSITIVE LOGITS

 hidden

0.46

 again

0.36

 meanwhile

0.35

endif

0.34

 there

0.33

 obvious

0.33

0.32

0.31

 clearly

0.30

 available

0.30

Activations Density 0.000%

No Known Activations

This feature has no known activations.