INDEX

Explanations

resolving symbols or values

The neuron strongly activates on occurrences of the term “resolve” (and its close variants) in code, error messages, or comments.

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

 również

-1.64

醣

-1.48

 الاط

-1.45

 primeros

-1.43

doise

-1.43

dő

-1.41

 primero

-1.40

 erreicht

-1.39

rücke

-1.39

 privada

-1.38

POSITIVE LOGITS

1.93

1.64

 multitude

1.48

)};

1.45

1.44

1.41

⢫

1.38

ίζ

1.37

 думаю

1.34

地说

1.34

Activations Density 0.045%