INDEX

Explanations

comparison following play or slower

The neuron fires on abstract–or evaluative–content words that signal emphasis, degree, or conceptual weight (e.g. “more,” “longer,” “often,” “meaning,” “space,” “likely”).

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

 এমনই

0.76

 weiteren

0.68

 তেমনই

0.68

 একইভাবে

0.68

 তাহারই

0.66

 entweder

0.62

Đối

0.62

 সেইরূপ

0.61

 তেমনি

0.60

再び

0.59

POSITIVE LOGITS

 than

4.72

than

3.70

 niż

3.59

 Than

3.58

Than

3.48

än

3.14

 decât

3.13

 kuin

2.92

 THAN

2.69

 než

2.56

Activations Density 0.508%