INDEX

Explanations

asking questions

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

 Martins

-0.11

 Alley

-0.09

 Flam

-0.09

 answering

-0.08

anta

-0.08

 McCorm

-0.08

 exotic

-0.08

 Quiz

-0.08

Quiz

-0.08

Radians

-0.08

POSITIVE LOGITS

 question

0.19

 questions

0.18

 specific

0.16

 Ð²Ð¾Ð¿ÑĢÐ¾Ñģ

0.15

 actual

0.14

 ì§Īë¬¸

0.14

question

0.13

questions

0.13

 Questions

0.13

.questions

0.12

Activations Density 0.144%