INDEX

Explanations

code syntax operators

This neuron responds strongly to code tokens for in-place numeric updates, especially the prefix increment ( ++ ) and decrement ( -- ) operators.

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

to

-1.27

 rossi

-1.16

ные

-1.13

for

-1.13

 siena

-1.13

✣

-1.12

や

-1.11

 lill

-1.07

 jaya

-1.07

 piccolo

-1.07

POSITIVE LOGITS

 pobres

1.30

andafter

1.22

 tiefe

1.15

 Spenden

1.14

 pazar

1.13

}],

1.10

 quỷ

1.09

 Dritte

1.08

nerability

1.06

 ním

1.04

Activations Density 0.002%