INDEX

Explanations

calculated by subtracting

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

.printStackTrace

-0.18

<|begin_of_text|>

-0.13

.Forms

-0.13

Ø§ÙĦØ¥ÙĨØ¬ÙĦÙĬØ²ÙĬØ©

-0.12

olumn

-0.12

oreferrer

-0.11

endregion

-0.11

 EventArgs

-0.10

 styleUrls

-0.10

nofollow

-0.09

POSITIVE LOGITS

¶Į

0.12

BeginInit

0.12

EndInit

0.11

ë¨

0.10

 ë¨

0.10

ãĨ

0.10

Â

0.09

HORT

0.09

HITE

0.09

Inlining

0.09

Activations Density 0.015%