INDEX

Explanations

log implies

ranking failure modes

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

৯

0.52

ম

0.50

IS

0.47

gF

0.46

,\

0.46

INGER

0.46

자를

0.45

dsp

0.45

๐

0.44

AR

0.44

POSITIVE LOGITS

 penetrated

0.49

 pondering

0.48

rauch

0.47

 penetration

0.46

 firmness

0.46

 teleport

0.45

 polynomial

0.45

 polishing

0.45

 encountering

0.45

 penetrating

0.44

Activations Density 0.001%