INDEX

Explanations

specific data and costs

The neuron fires on multi-word nominal phrases that name or label data sections, reports, or content listings (e.g. “course offerings,” “air quality data,” “artifact photographs,” “trend predictions,” etc.).

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

のような

0.48

复杂的

0.45

Что

0.45

0.43

我的

0.43

での

0.42

Pued

0.41

血糖

0.40

ಕಾರ

0.40

 ഞാന്‍

0.39

POSITIVE LOGITS

for

0.63

 från

0.59

 across

0.59

 για

0.59

สำหรับ

0.58

 from

0.57

 από

0.56

 для

0.55

 Across

0.54

 كافة

0.54

Activations Density 0.000%