INDEX

Explanations

fairly accurate descriptions

actions and states of being

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

++);

0.46

 ২০১১

0.45

gateTime

0.44

িও

0.43

 Straßen

0.43

 তাও

0.43

 машинасы

0.43

σότε

0.42

клу

0.42

售价

0.42

POSITIVE LOGITS

jour

0.60

0.57

er

0.55

ालय

0.53

label

0.52

tedir

0.51

0.50

ent

0.50

bomb

0.49

com

0.49

Activations Density 0.001%