INDEX

Explanations

code comments and language tags

The neuron strongly activates on code‐documentation markers (e.g. “///”, “/**”, “*”, “@…”, and C++-style “::” references).

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

 that

-1.51

 agricultura

-1.36

 نیز

-1.30

Inoltre

-1.29

 colonia

-1.26

 from

-1.25

 minera

-1.23

 fator

-1.23

 concha

-1.23

 playera

-1.23

POSITIVE LOGITS

we

1.61

1.38

 появля

1.30

有

1.24

provides

1.22

↵↵

1.22

creas

1.16

quis

1.16

taminophen

1.15

Activations Density 0.013%