INDEX

Explanations

The neuron activates on numeric tokens—especially measurements or quantities (e.g. decimal numbers and distances).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 pleinement

0.59

FARE

0.59

严格

0.58

そのまま

0.58

 stessi

0.54

 resell

0.54

resize

0.54

 같습니다

0.53

意味

0.53

用意

0.53

POSITIVE LOGITS

 deficient

1.26

 lacked

1.23

 lacking

1.17

 deficiency

1.16

 unsatisfactory

1.14

缺少

1.13

 deficiencies

1.13

缺乏

1.09

 문제를

1.09

 부족

1.09

Activations Density 1.159%