INDEX

Explanations

returning function results or errors

The neuron fires on return statements that include literal numeric (and sometimes `None`/`nil`) constants, i.e. it detects numeric (and similar) literal return values in code.

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

of

-1.80

他說

-1.43

 nadzieję

-1.41

 with

-1.36

 kunna

-1.31

ובת

-1.28

 рассказывает

-1.27

 Петра

-1.27

 adquirido

-1.27

 becomes

-1.27

POSITIVE LOGITS

 These

1.77

 puisqu

1.70

＂

1.39

 They

1.32

Of

1.30

“-

1.30

 decât

1.28

或者

1.28

anyakan

1.27

]$.

1.25

Activations Density 0.004%