INDEX

Explanations

conditional statements in code

This neuron detects loop constructs in code, especially “for” (and related) loop statements.

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

to

-1.16

 proberen

-0.88

lanterna

-0.86

但她

-0.82

éos

-0.80

return

-0.79

却被

-0.79

糙

-0.78

 navigator

-0.78

at

-0.77

POSITIVE LOGITS

if

1.70

If

1.45

 اگر

1.26

 якщо

1.24

 если

1.20

 إذا

1.16

 nếu

1.05

 अगर

1.05

若

1.04

如果

1.00

Activations Density 0.003%