INDEX

Explanations

temporal references

The neuron fires on discourse markers that signal temporal sequence or transitions (e.g. once, initially, at).

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

本次

0.76

ствуют

0.75

又要

0.74

 ಈಗಾಗಲೇ

0.73

 সম্মেল

0.71

 ইতিমধ্যেই

0.70

ють

0.69

encounter

0.68

هم

0.68

 ఇప్పటికే

0.68

POSITIVE LOGITS

 وكان

1.20

でしたが

1.19

 وكانت

1.19

 était

1.18

 vardı

1.16

하던

1.16

 autrefois

1.16

 formerly

1.13

 fazia

1.13

 étaient

1.09

Activations Density 0.906%