INDEX

Explanations

dissolving and solutions

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

enou

-0.10

ucci

-0.10

onna

-0.10

ãģĹãģı

-0.10

 Keller

-0.09

 sobie

-0.09

 cucumber

-0.09

staw

-0.09

-0.08

POSITIVE LOGITS

 diss

0.17

ensions

0.13

 Diss

0.12

à¸¥à¸°

0.11

TestFixture

0.11

ension

0.10

ÑģÑĤÐ²Ð¾ÑĢ

0.10

 Wick

0.10

 ÑĢÐ°ÑģÑĤÐ²Ð¾ÑĢ

0.09

 into

0.09

Activations Density 0.009%