INDEX
Explanations
clauses and conjunctions
generic function tokens in assistant-style explanatory prose (e.g., conjunctions, auxiliaries, and punctuation) that stitch sentences together.
New Auto-Interp
Negative Logits
ipfs
0.43
教
0.43
dosa
0.42
циями
0.41
全体の
0.41
楷
0.41
dict
0.41
しか
0.40
zh
0.40
diva
0.40
POSITIVE LOGITS
疼痛
0.40
бай
0.40
зне
0.39
бе
0.39
ஸ்ட
0.39
undamaged
0.39
ót
0.39
รี
0.38
전기
0.38
змо
0.37
Activations Density 0.055%