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

-0.32

<eos>

-0.31

-0.29

is

-0.28

-0.27

 take

-0.27

-0.26

ion

-0.26

POSITIVE LOGITS

WebServlet

7.03

 defaultstate

1.17

httphttps

1.12

 ujednoznacz

1.09

adaptiveStyles

1.09

 ligiloj

1.06

ReusableCell

1.05

ManagedBean

1.04

 resourceCulture

1.04

WebElementEntity

1.02

Activations Density 0.000%

No Known Activations

This feature has no known activations.