INDEX

Explanations

interrupted or unsupported code constructs

The neuron flags technical exception or error identifiers (e.g. Java exception class names and other “Exception”/“error” tokens).

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

ᚔ

-1.55

 cometer

-1.45

魚の

-1.40

ᥬ

-1.38

CodeDom

-1.38

特朗

-1.35

汌

-1.35

Situated

-1.34

ーーーー

-1.34

="@+

-1.34

POSITIVE LOGITS

<bos>

1.56

IDENTAL

1.41

ਾਨ

1.38

</b>

1.35

小物

1.34

 leeftijd

1.34

 sowohl

1.33

1.32

</i>

1.31

Activations Density 0.000%