INDEX

Explanations

then / sequence

The neuron fires strongly on imperative cooking–recipe step verbs (e.g. “Drizzle,” “Mix,” “Add,” “Garnish,” “Serve”) and the sentence boundaries around them.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 Ironically

0.85

 いや

0.82

 Bakufu

0.80

 fudai

0.78

 Pourtant

0.76

 imao

0.75

 Revenues

0.73

 inwon

0.73

 Ingeniería

0.72

 Tetapi

0.71

POSITIVE LOGITS

 then

1.11

然后

1.04

然後

1.02

 потім

1.01

 затем

0.99

then

0.95

然后在

0.94

 vervolgens

0.90

 kemudian

0.89

その後

0.89

Activations Density 0.270%