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

ロウィン

-0.58

 Weiſe

-0.52

⇿

-0.50

 notary

-0.50

<unused20>

-0.50

<unused43>

-0.50

<unused14>

-0.50

<unused28>

-0.50

[@BOS@]

-0.49

<unused71>

-0.49

POSITIVE LOGITS

toxicity

0.96

:✨

0.41

Портали

0.39

hoeddwyd

0.38

 ***!

0.37

BufferException

0.37

 estekak

0.36

addPreferredGap

0.36

Diweddarwch

0.35

qrstuvwxyz

0.33

Activations Density 0.000%

No Known Activations

This feature has no known activations.