INDEX

Explanations

incorrect results or evaluation

The neuron fires on technical statistical or machine‐learning terms (especially evaluation metrics, parameters, and result descriptors).

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

two

-1.34

if

-1.27

%;

-1.24

 τὸν

-1.23

 this

-1.22

 while

-1.21

 when

-1.21

as

-1.20

 departamentos

-1.17

agosto

-1.16

POSITIVE LOGITS

 ativar

1.42

Karakter

1.40

 aident

1.36

 utilisent

1.36

 confirmó

1.32

美味し

1.30

vantaged

1.29

Ketua

1.28

ixante

1.28

 reveló

1.28

Activations Density 0.003%