INDEX

Explanations

policy

The neuron selectively activates on tokens representing decimal‐style numbers (floating‐point numeric values).

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

 produkcji

0.64

 soprattutto

0.63

 pued

0.59

 oksid

0.58

 oscillate

0.57

 parete

0.57

 distX

0.57

 воздей

0.57

☕

0.57

 posX

0.56

POSITIVE LOGITS

étranger

0.51

वाने

0.48

deferred

0.47

 staffing

0.46

inquirer

0.46

 policym

0.46

撰

0.46

 security

0.45

 policies

0.45

后者

0.45

Activations Density 0.001%