INDEX

Explanations

every

This neuron fires on numeric tokens (especially multi‐digit numbers and years).

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

↵↵

2.47

↵↵↵

2.14

↵↵↵↵

1.91

↵↵↵↵↵

1.85

<start_of_image>

1.63

↵↵↵↵↵↵

1.60

↵

1.55

↵↵↵↵↵↵↵

1.46

↵↵↵↵↵↵↵↵

1.41

↵↵↵↵↵↵↵↵↵

1.40

POSITIVE LOGITS

 Misalnya

0.84

<unused2184>

0.84

<unused2151>

0.82

 stockbild

0.79

十壹章

0.75

ຈັດສົ່ງ

0.73

ကလေး

0.73

្នក

0.72

 misalnya

0.71

ຜະລິດຕ

0.71

Activations Density 0.281%