INDEX

Explanations

references to wholeness or completeness

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 13-gemmascope-res-65k

Configuration

google/gemma-scope-2b-pt-res/layer_13/width_65k/average_l0_75

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

Data Type

float32

Hook Name

blocks.13.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

 Tartar

-0.54

 επίσης

-0.51

います

-0.49

 Küchen

-0.49

 bacilli

-0.48

 sherds

-0.48

also

-0.48

 altrett

-0.47

 також

-0.46

 piety

-0.46

POSITIVE LOGITS

 entire

1.48

 whole

1.25

entire

1.22

 Entire

1.19

 ENTIRE

1.14

whole

1.12

Entire

1.04

 WHOLE

1.00

 utafitiHapana

0.97

Whole

0.96

Activations Density 0.018%

references to wholeness or completeness

No Known Activations