INDEX

Explanations

code and error messages

This neuron fires on programming‐language parse tokens or metadata labels (e.g., CLASS, METHOD, FIELD, ARG) in code dumps.

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

Detalhes

-0.76

 suicidio

-0.74

瑜

-0.69

clearAll

-0.68

 Mauri

-0.68

뤼

-0.67

{_

-0.66

 XXXX

-0.66

Hilarious

-0.66

xffffffff

-0.66

POSITIVE LOGITS

ASM

1.22

ASM

1.02

 visits

0.84

Право

0.81

pele

0.79

asm

0.79

insn

0.78

ctan

0.78

СТО

0.77

Insn

0.77

Activations Density 0.007%