INDEX

Explanations

json formatting

The neuron fires strongly on quoted field or property names immediately followed by a colon in code‐ or JSON‐like snippets.

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

 अभिकारक

0.62

valuation

0.59

乀

0.58

 हात

0.58

 растение

0.57

 घटस्फोट

0.57

ន់

0.56

planting

0.56

 தமிழரசுக்

0.56

ámica

0.55

POSITIVE LOGITS

1.05

},"

0.85

"__

0.83

"_

0.82

["

0.81

},

0.80

},{

0.80

"_

0.79

"":

0.79

"$

0.79

Activations Density 0.015%