INDEX

Explanations

way of life

the neuron detects abstract quality or evaluation nouns (e.g. “experience,” “consistency,” “surface,” “atmosphere,” “approach”).

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

0.44

AH

0.41

–

0.38

and

0.37

 shortages

0.36

AA

0.35

 incidents

0.34

POSITIVE LOGITS

 aonde

0.51

 хто

0.47

 onde

0.46

 এড়িয়ে

0.44

 кто

0.44

 préférence

0.43

 който

0.42

 куда

0.42

 который

0.41

 ännu

0.41

Activations Density 0.000%