INDEX

Explanations

This neuron is sensitive to high-frequency function words and discourse markers (common conjunctions and prepositions).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 glorie

0.96

 despicable

0.88

栲

0.86

 Cagliari

0.84

JSONException

0.84

 Junius

0.83

 aventuras

0.82

 njima

0.81

 Blast

0.80

 jolly

0.80

POSITIVE LOGITS

 tranquility

1.34

 آرام

1.30

 Peaceful

1.23

 tranquil

1.22

 शांत

1.21

 tranquillity

1.19

 calming

1.16

平静

1.16

 calme

1.15

 tranquill

1.15

Activations Density 2.090%