INDEX

Explanations

The neuron fires on tokens used in boolean logic and conditionals—e.g. comparison operators (==, ===, <), logical AND (&&), ternary “?” and “:”, and “if” markers.

code syntax

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 just

-1.02

eability

-0.88

 either

-0.87

 such

-0.86

 with

-0.86

 arcu

-0.85

ívne

-0.84

 what

-0.83

لاعب

-0.82

lishes

-0.80

POSITIVE LOGITS

 szí

0.98

 khoan

0.98

 típus

0.95

addClass

0.92

 colorés

0.90

 rám

0.87

Припев

0.85

 cantora

0.85

 ganar

0.83

volvo

0.83

Activations Density 0.012%