INDEX

Explanations

lowers or minimal effort

The neuron fires strongly on words and phrases that denote quantitative magnitude or intensity—e.g. “high,” “reduction,” “significant,” “rates,” “minimum,” “long,” etc.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

AD

0.98

<0x82>

0.96

辦法

0.96

<0x80>

0.95

OTHER

0.92

naya

0.91

椑

0.91

诒

0.90

日英

0.88

 worsened

0.88

POSITIVE LOGITS

ów

1.34

பட்ச

1.30

ের

1.15

 probabilidad

1.15

1.13

 temperat

1.10

ς

1.07

 Anzahl

1.06

 температуры

1.05

 вероятно

1.04

Activations Density 1.383%