INDEX

Explanations

you followed by ability/possibility

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

|:

-0.11

cant

-0.10

 cant

-0.09

 Cant

-0.09

 Kling

-0.09

uly

-0.09

 Wilde

-0.08

err

-0.08

 umbrella

-0.08

POSITIVE LOGITS

 could

0.19

 might

0.19

can

0.17

 à¤¸à¤ķà¤¤

0.15

åı¯ä»¥

0.15

 pouvez

0.14

 Ð¼Ð¾Ð¶ÐµÑĤÐµ

0.13

 dapat

0.13

 bisa

0.13

 olabilir

0.13

Activations Density 0.059%