INDEX

Explanations

prepositions or conjunctions

New Auto-Interp

Top Features by Cosine Similarity

Configuration

Prompts (Dashboard)

10,000 prompts, 128 tokens each

Dataset (Dashboard)

lmsys/lmsys-chat-1m

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

adv

-0.11

adro

-0.10

inton

-0.09

implify

-0.09

 Byrne

-0.09

orz

-0.08

 Frances

-0.08

dig

-0.08

rame

-0.08

 digging

-0.08

POSITIVE LOGITS

 ÑĥÐ¿Ð¾ÑĤÑĢÐµÐ±

0.16

è´Ńä¹°

0.14

åĸĿ

0.13

ä½¿çĶ¨

0.13

 travers

0.13

 wearing

0.12

åıĤåĬł

0.12

 Ð¿Ð¾ÐºÑĥÐ¿

0.12

è®¿éĹ®

0.11

Ð¾Ð»ÑĮÐ·Ð¾Ð²

0.11

Activations Density 0.419%