INDEX

Explanations

grade levels

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

 Classroom

-0.14

 classroom

-0.12

 classrooms

-0.11

è¯¾

-0.11

 ÑĥÑĩÐµÐ±

-0.10

icher

-0.09

 Degrees

-0.09

wipe

-0.09

Degrees

-0.09

 Class

-0.09

POSITIVE LOGITS

 secondary

0.21

secondary

0.18

 level

0.17

 middle

0.17

-secondary

0.15

 Secondary

0.15

 high

0.15

Secondary

0.14

 grade

0.13

 primary

0.13

Activations Density 0.087%