INDEX

Explanations

specific questions

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

SizePolicy

-0.09

 specificity

-0.09

alu

-0.09

<|begin_of_text|>

-0.09

-0.08

nÄĽm

-0.08

 thorough

-0.08

Tut

-0.08

ters

-0.08

:;\n

-0.08

POSITIVE LOGITS

 questions

0.21

ally

0.21

ities

0.20

ially

0.18

 question

0.17

ity

0.16

ALLY

0.16

questions

0.13

 Fragen

0.13

 Ð²Ð¾Ð¿ÑĢÐ¾Ñģ

0.13

Activations Density 0.020%