INDEX

Explanations

high activation values across various sections of structured data

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 5-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_5/width_16k/average_l0_68

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.5.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

 Бли

-0.66

 control

-0.64

aktery

-0.63

Tok

-0.62

-0.60

Bue

-0.60

 hela

-0.59

ので

-0.59

 Towel

-0.58

bule

-0.58

POSITIVE LOGITS

2.02

 NINE

1.43

 Ninth

1.42

 ninth

1.32

 Nine

1.31

nine

1.29

۹

1.29

 ninety

1.28

ninth

1.25

 nine

1.23

Activations Density 0.709%

high activation values across various sections of structured data

No Comments

No Known Activations