INDEX

Explanations

concentration and where it is

The main thing this neuron does is detect mentions of “concentration” or forms of “concentrate,” especially in contexts describing things as concentrated.

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

</i>

-2.42

The

-2.27

-2.13

 reason

-2.03

ご了承

-1.96

ご了承ください

-1.88

~~~

-1.82

 一个

-1.78

 vecchia

-1.76

сина

-1.76

POSITIVE LOGITS

</sup>

2.52

</h2>

2.50

2.14

</h4>

2.08

 inspiradoras

1.98

at

1.77

</em>

1.69

 motivadoras

1.67

1.66

\_

1.64

Activations Density 0.033%