INDEX

Explanations

monday mornings

The neuron is essentially flagging unusually rare or low‐frequency tokens (i.e. out‐of‐vocabulary or uncommon words).

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

mis

1.16

ng

1.04

mg

0.88

SCO

0.85

il

0.84

mb

0.83

lm

0.83

mus

0.83

re

0.82

mpi

0.82

POSITIVE LOGITS

’

0.78

\}^{

0.77

 оказывается

0.74

 уж

0.73

 embod

0.71

 ум

0.70

 ез

0.70

 soldered

0.69

 связа

0.68

 Luch

0.68

Activations Density 0.000%