INDEX

Explanations

commas and specific conjunctions

The neuron activates on numeric tokens (quantities, years, counts, or other digits) in the text.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 offering

0.61

0.60

对其

0.55

0.54

 ఇది

0.53

うち

0.53

слі

0.52

 своим

0.52

<0xF0>

0.51

ష్

0.51

POSITIVE LOGITS

the

2.78

the

2.45

sthe

1.73

 הה

1.63

ihe

1.62

 τις

1.57

THE

1.52

thed

1.48

thebetter

1.47

 τη

1.46

Activations Density 1.778%