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

 lisäksi

-0.32

 zůst

-0.30

 satunya

-0.27

<eos>

-0.25

 asimismo

-0.25

 Reised

-0.25

 kysy

-0.25

 Füßen

-0.23

 nivå

-0.23

 blijven

-0.23

POSITIVE LOGITS

<unused8>

1.12

<unused3>

1.12

<unused51>

1.12

<unused74>

1.12

<unused14>

1.12

<unused43>

1.12

<unused16>

1.11

<unused23>

1.11

[@BOS@]

1.11

<pad>

1.11

Activations Density 0.000%

No Known Activations

This feature has no known activations.