INDEX

Explanations

language variations and translations

The neuron is primarily detecting multi‐word technical or quantitative noun phrases (e.g. “moving averages,” “cross reference,” “order several sizes”).

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

ו

0.66

0.63

 Difficult

0.59

 Innov

0.59

Outstanding

0.57

Rat

0.56

 swallowed

0.55

و

0.55

 Outstanding

0.55

但在

0.55

POSITIVE LOGITS

ған

0.80

 미리

0.76

 spacerItem

0.72

 కూడా

0.71

 पनि

0.70

𝓹

0.70

 diferite

0.69

 sone

0.68

ˆ‚

0.68

 diferentes

0.67

Activations Density 0.000%