INDEX

Explanations

java class definitions

The neuron is specifically picking up floating‐point numeric literals (i.e. numbers containing a decimal point) in the code.

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

鼍

-1.24

∎

-1.23

〖

-1.23

 causando

-1.20

ciating

-1.19

 provocando

-1.17

elsif

-1.16

 plass

-1.14

parken

-1.13

Ⴀ

-1.12

POSITIVE LOGITS

感じで

1.52

 âgé

1.35

ほら

1.34

感じです

1.23

来週

1.22

ごめんなさい

1.20

うら

1.18

今度

1.17

ところに

1.16

だいぶ

1.13

Activations Density 0.005%