INDEX

Explanations

phrases indicating expectations or intended actions

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 23-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_23/width_16k/average_l0_120

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.23.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

 terbukti

-0.37

amb

-0.33

 tull

-0.32

гули

-0.32

 commandement

-0.31

 initiatives

-0.31

 endured

-0.30

 pluma

-0.30

 nationality

-0.30

 Bauch

-0.30

POSITIVE LOGITS

supposed

0.72

yntaxException

0.71

 supposed

0.70

󠁮

0.70

 dovrebbe

0.69

 dovrebbero

0.67

UrlResolution

0.63

exitRule

0.62

toHaveBeenCalled

0.61

 normalement

0.60

Activations Density 0.437%

phrases indicating expectations or intended actions

No Known Activations