INDEX

Explanations

conjunctions and connectors

The neuron lights up on the key content words in a sentence—especially the main action verbs and other informative, less-common lexical items that carry most of the sentence’s meaning.

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

 maupun

1.04

 lẫn

0.82

 malgré

0.76

 quoique

0.75

 حتى

0.71

 choć

0.71

 رغم

0.69

 invece

0.68

这才

0.66

 хотя

0.66

POSITIVE LOGITS

었고

1.77

 거고

1.68

And

1.67

And

1.65

 และ

1.64

并且

1.57

했고

1.53

이고

1.46

iar

1.41

ですし

1.38

Activations Density 0.123%