INDEX

Explanations

reasoning and knowledge comprehension

The neuron detects abstract nouns that name positive qualities, achievements, or intellectual attributes (e.g. independence, productivity, knowledge, innovations).

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

 후보

0.71

Ch

0.68

0.66

Þ

0.65

Ḥ

0.65

0.64

Mix

0.63

وامل

0.63

uć

0.63

POSITIVE LOGITS

 rudimentary

0.96

 decayed

0.87

akuwa

0.86

 dilapidated

0.86

 nonexistent

0.77

 parietal

0.77

 faintly

0.77

 crippled

0.76

 flimsy

0.76

 गिनती

0.73

Activations Density 0.103%