INDEX

Explanations

read, understand, and maintain code

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

adro

-0.09

 Hawth

-0.09

omen

-0.09

umbling

-0.09

CreatedAt

-0.08

ionale

-0.08

atra

-0.08

ifest

-0.08

woord

-0.08

ije

-0.08

POSITIVE LOGITS

 maintenance

0.26

 Maintenance

0.23

Maintenance

0.20

 debug

0.20

maintenance

0.19

 debugging

0.18

 maintain

0.17

ç»´

0.17

bug

0.17

 understand

0.17

Activations Density 0.050%