INDEX

Explanations

communication threads and participants

The neuron detects code‐style references to conversations or chats—e.g. model/class names like “Conversation” or “conversations,” chat identifiers, and their related parameters (“from,” “to,” 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

</b>

-1.31

↵↵

-1.28

并不是

-1.27

 naam

-1.23

-1.21

</em>

-1.18

CHANGES

-1.16

なかった

-1.16

jinal

-1.16

 year

-1.15

POSITIVE LOGITS

 with

1.45

<bos>

1.41

 there

1.28

eynman

1.27

 whose

1.22

ambiguation

1.21

誰

1.19

糎

1.19

ிறது

1.18

お勧め

1.17

Activations Density 0.049%