INDEX

Explanations

code structure and data types

The neuron strongly activates on programming‐style identifiers and API/library names (e.g. CamelCase or all‐caps tokens such as “Api,” “UR,” “swagger,” “JsonObject,” etc.).

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

つづく

-1.34

 delikat

-1.23

-1.22

 rigue

-1.20

 Drž

-1.17

 piaci

-1.17

 quelcon

-1.16

Földrajz

-1.16

 zimowa

-1.15

おわりに

-1.14

POSITIVE LOGITS

や

1.30

は

1.23

を

1.20

Джерела

1.13

Опис

1.11

 though

1.09

商品説明

1.06

と

1.06

 оригіналу

1.04

 fides

1.04

Activations Density 0.002%