INDEX

Explanations

The neuron fires on code tokens that are parts of fully qualified identifiers (e.g. class or namespace names separated by dots).

code identifiers

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

 with

-1.30

in

-1.23

 more

-1.13

 just

-1.13

 some

-1.12

at

-1.04

 have

-1.02

for

-1.02

ள்ள

-0.94

 потен

-0.94

POSITIVE LOGITS

튿

1.08

‚¬

1.06

 Cavaliers

1.06

౧

1.05

 envíos

1.04

夸张

1.02

TagHelpers

0.99

॥

0.98

 enfermos

0.98

쑹

0.98

Activations Density 0.012%