INDEX

Explanations

variable names and code elements

The neuron fires strongly on standalone identifier names in code (e.g. short variable names or aliases).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 Which

-0.97

кого

-0.93

and

-0.91

so

-0.91

Ί

-0.84

JpaRepository

-0.83

ARGUMENT

-0.83

 científicas

-0.83

 people

-0.82

 tæt

-0.82

POSITIVE LOGITS

 kedua

0.90

 duque

0.89

 campeón

0.88

 parha

0.88

 зовні

0.87

 оригі

0.86

 begge

0.85

 endom

0.84

铳

0.84

팎

0.83

Activations Density 0.007%