INDEX

Explanations

what i

The neuron activates on numerical tokens or numeric expressions (quantities, years, measurements).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 unsurprisingly

1.41

 unambiguously

1.38

(!)

1.29

Notably

1.24

(!)

1.20

 explicitly

1.15

 ostensibly

1.13

ஃப்

1.12

 nuanced

1.10

 преимущественно

1.08

POSITIVE LOGITS

 नही

1.58

埸

1.50

 alot

1.38

 loosing

1.34

 todays

1.27

yrs

1.24

 ঘোষনা

1.24

と言う

1.23

 গুলো

1.21

맟

1.21

Activations Density 2.624%