INDEX

Explanations

like these

The neuron fires on multi‐syllable, nominalized concept words (abstract “–ing” or complex noun forms like “tasks,” “activities,” “terms,” “solutions,” 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

 этим

0.80

નની

0.62

 আর

0.61

 новым

0.59

 নামটি

0.57

 ત્યાર

0.56

 своим

0.55

 दोन्ही

0.55

 然后

0.55

 거기에

0.54

POSITIVE LOGITS

 such

3.92

such

3.61

 เช่น

3.22

 таких

3.06

 такие

2.97

 seperti

2.92

比如

2.92

 zoals

2.91

 takich

2.91

 SUCH

2.91

Activations Density 2.154%