INDEX

Explanations

dates in the 1800s

The neuron detects four‐digit tokens representing years (historical date references).

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

 θα

-1.94

 materna

-1.88

 ブレスレット

-1.88

普遍

-1.80

 itſelf

-1.76

货

-1.74

鼇

-1.73

大部分

-1.72

for

-1.72

вий

-1.69

POSITIVE LOGITS

 When

2.16

zieć

2.05

 الرئيسي

1.95

還會

1.95

ﻒ

1.80

1.79

 cuando

1.78

落ちる

1.77

 verlei

1.75

我能

1.72

Activations Density 0.020%