INDEX

Explanations

code comments

The main thing this neuron does is detect double‐slash comment markers (“//”) in code.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

ための

0.94

rieg

0.74

 übernommen

0.71

গুলিকে

0.71

 universal

0.70

א

0.70

們

0.67

 version

0.67

ussie

0.67

igens

0.66

POSITIVE LOGITS

Jika

0.96

PROBLEM

0.84

Salah

0.82

Nếu

0.80

 Lembrando

0.80

 وهنا

0.79

 Mỗi

0.79

TODO

0.77

NaBH

0.76

 Если

0.75

Activations Density 0.000%