INDEX

Explanations

code comments and file headers

The neuron detects numeric tokens—especially multi-digit sequences like years or version numbers—in source code and comments.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

\\

-1.10

↵↵↵

-1.08

-\\

-1.07

.\\

-1.07

↵↵↵↵↵↵↵↵

-0.99

,\\

-0.94

이버

-0.88

;\\

-0.87

 manguera

-0.86

↵↵

-0.86

POSITIVE LOGITS

</em>

4.41

</i>

4.22

</strong>

2.50

</b>

2.31

</u>

1.98

</code>

1.95

</s>

1.75

</h5>

1.73

</blockquote>

1.66

</h6>

1.51

Activations Density 0.036%