INDEX

Explanations

development and evolution

The main thing this neuron does is detect past-tense verbs (especially regular verbs ending in “-ed”).

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

}$-

0.58

 rejoicing

0.58

 outings

0.57

 resetting

0.57

}}-

0.55

 recapture

0.54

ད་

0.52

ल्फ़

0.52

 excursions

0.52

 reinstatement

0.52

POSITIVE LOGITS

 разви

1.83

 evolved

1.83

 evolve

1.80

發展

1.78

 розви

1.78

 berkembang

1.77

 utvik

1.77

发展

1.74

 विकसित

1.72

 entwickeln

1.72

Activations Density 0.410%