INDEX

Explanations

references to "everyone" or similar terms indicating inclusivity

oai_token-act-pair · gpt-4o-mini Triggered by @bot

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

-0.42

 pregi

-0.42

-0.41

 Utrecht

-0.41

нование

-0.39

 path

-0.39

 iscri

-0.38

-0.36

/**

-0.35

 Kildare

-0.35

POSITIVE LOGITS

everyone

1.39

 everyone

1.31

Everyone

1.29

 Everyone

1.27

 EVERYONE

1.23

everybody

1.20

 everybody

1.16

 Everybody

1.11

Everybody

1.10

anyone

0.88

Activations Density 0.009%

references to "everyone" or similar terms indicating inclusivity

No Comments

No Known Activations