INDEX

Explanations

loops using while

The neuron strongly activates on the keyword “while,” i.e. it detects use of while‐loops.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

if

-0.88

angol

-0.86

 Bundestag

-0.83

fering

-0.82

 нашли

-0.82

纔

-0.82

 kuris

-0.81

 ApiService

-0.81

 دوباره

-0.81

 ベー

-0.79

POSITIVE LOGITS

 until

1.35

เริ่ม

1.24

 zač

1.20

 start

1.16

 시작

1.15

开始

1.12

 kezd

1.11

 started

1.09

開始

1.08

 began

1.05

Activations Density 0.008%