INDEX

Explanations

Last chance, last day, last night

The neuron strongly activates on the word “Last” when it begins a temporal reference (e.g. “Last night,” “Last season,” “Last updated,” etc.).

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

∼

0.70

 প্রাধ

0.66

 திரும்பி

0.65

 secondaires

0.63

 sidan

0.63

 mừng

0.62

 ragione

0.62

mores

0.61

দ্দীন

0.60

 overjoyed

0.60

POSITIVE LOGITS

 Last

1.07

remaining

0.92

Last

0.91

last

0.90

 перед

0.88

Remaining

0.86

 remaining

0.85

hur

0.82

 last

0.82

Before

0.80

Activations Density 0.155%