INDEX

Explanations

predicting accuracy, yield, values

The neuron responds strongly to technical modeling and analysis terms—words like “model,” “predictions,” “analysis,” and “results” that signal descriptions of scientific or engineering models and their evaluated outcomes.

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

 anisotropic

1.38

 hypothesized

1.34

 quantified

1.33

 postulated

1.31

 inversely

1.30

 quantification

1.28

 heterogeneity

1.27

 anisotropy

1.26

 elucid

1.26

 correlated

1.25

POSITIVE LOGITS

 goede

1.09

 деньги

1.07

 ज़िंदगी

1.04

 buoni

1.04

GOOD

1.03

inguém

1.02

 друзьями

1.00

ratulations

0.98

 Buenas

0.98

 ใคร

0.98

Activations Density 0.081%