INDEX

Explanations

extra keywords

This neuron detects code tokens related to “extra” data in Android Intents or similar APIs (e.g. putExtra, getExtras, EXTRA_* constants).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 avec

-2.34

禤

-2.08

嫑

-2.06

The

-2.05

 cinquième

-2.03

"];

-2.02

 quatrième

-1.90

买了

-1.88

鈈

-1.86

日本的

-1.84

POSITIVE LOGITS

rentiel

2.17

2.13

している

2.11

um

2.08

Whether

2.06

there

2.05

 händer

2.02

みて

2.00

sibilità

1.98

While

1.96

Activations Density 0.009%