INDEX

Explanations

create or add to code

This neuron responds to code‐specific keywords and syntax (e.g. import statements, function or visibility keywords like public/boolean, constructor), effectively flagging code rather than plain text.

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

configFile

-0.82

 munk

-0.82

UpdateUser

-0.75

 войне

-0.73

 userSchema

-0.72

Refinement

-0.71

ンダント

-0.70

hre

-0.69

拼接

-0.69

 Jennings

-0.69

POSITIVE LOGITS

 create

2.44

 CREATE

2.39

CREATE

2.36

create

2.33

Create

2.28

 creation

2.23

 Create

2.17

 save

2.02

 созда

1.97

 created

1.94

Activations Density 0.041%