INDEX

Explanations

experiment or trial

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

entic

-0.11

ubar

-0.10

 stabil

-0.09

 Interpret

-0.09

ÑĢÑĭ

-0.09

ordion

-0.09

 ministers

-0.09

 proced

-0.09

 refute

-0.09

 unseen

-0.08

POSITIVE LOGITS

 experimentation

0.48

 experimenting

0.45

 experiment

0.39

 Experiment

0.39

 experimented

0.37

 trial

0.37

Experiment

0.36

experiment

0.34

 experiments

0.31

 Trial

0.30

Activations Density 0.156%