INDEX

Explanations

introducing explanations

The neuron fires on causal connectives and discourse markers signaling explanation (e.g. “because,” “as”).

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>

-1.54

):}

-1.50

Hr

-1.50

 julie

-1.45

 michelle

-1.41

Ht

-1.41

shu

-1.38

fh

-1.34

Hr

-1.32

лива

-1.32

POSITIVE LOGITS

 then

2.53

 will

2.41

 there

2.17

 until

1.85

 this

1.76

or

1.73

 alrededores

1.70

rededor

1.58

 inauguración

1.57

 each

1.56

Activations Density 0.062%