INDEX

Explanations

expected results or different things

The neuron flags words that express subjective stance or evaluation—terms used to indicate opinions, expectations, or reasons.

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

-2.03

 which

-1.73

 still

-1.56

 from

-1.55

 only

-1.51

 since

-1.43

at

-1.34

 this

-1.31

for

-1.30

 should

-1.30

POSITIVE LOGITS

 jednoduch

1.46

 écoute

1.46

 tredje

1.46

 écrire

1.42

Erstellt

1.42

 proyectos

1.41

 такие

1.41

 rafraî

1.40

 nový

1.40

 parámetros

1.40

Activations Density 0.041%