INDEX

Explanations

document

The neuron strongly responds to code‐style tokens involved in variable assignments and operators (e.g. “var,” “=,” “+=,” numeric literals, and attribute/getAttribute calls).

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

달

0.52

 વાળ

0.52

jen

0.51

 যাই

0.51

 परिवर्तन

0.51

[[[

0.50

 Bowen

0.49

Meu

0.48

jou

0.47

 Infl

0.47

POSITIVE LOGITS

(),

0.63

document

0.60

?,

0.59

°,

0.59

shap

0.58

*,

0.58

łek

0.57

%,

0.56

】,

0.56

^{*},

0.55

Activations Density 0.000%