INDEX

Explanations

words following punctuation

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

 However

-0.39

However

-0.36

 however

-0.35

ï¼Įä½Ĩ

-0.35

ï¼Įä½Ĩæĺ¯

-0.33

 HOWEVER

-0.32

 à¹ģà¸ķ

-0.29

 NhÆ°ng

-0.29

 ÐŀÐ´Ð½Ð°ÐºÐ¾

-0.29

 nhÆ°ng

-0.28

POSITIVE LOGITS

 takÅ¾e

0.12

ÑıÐ³Ð¾Ð¼

0.11

 therefore

0.11

 wiÄĻc

0.11

 Butt

0.11

ìĿ´ê³ł

0.11

 butt

0.10

 Ð¿Ð¾ÑįÑĤÐ¾Ð¼Ñĥ

0.10

 Butler

0.09

Therefore

0.09

Activations Density 0.173%