INDEX

Explanations

functions and errors

This neuron detects function declarations’ return type lists, particularly spotting the return of an error (or multiple values ending with error).

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

 یکی

-1.47

 มัน

-1.34

许多

-1.27

大きな

-1.26

茹で

-1.21

(&:

-1.21

 régal

-1.21

yczne

-1.20

进口

-1.20

 oubli

-1.20

POSITIVE LOGITS

 when

2.13

for

1.91

or

1.81

 whose

1.49

 used

1.45

WebElementEntity

1.45

did

1.43

all

1.39

any

1.38

 with

1.36

Activations Density 0.015%