INDEX

Explanations

positive communication, interactions, and behavior

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

fon

-0.10

jur

-0.09

 Solic

-0.09

_OCCURRED

-0.09

 peoples

-0.08

zsche

-0.08

 slack

-0.08

 Grape

-0.08

onso

-0.08

DCALL

-0.08

POSITIVE LOGITS

 interaction

0.21

 interactions

0.18

 communication

0.18

 conversation

0.17

 discourse

0.16

interaction

0.15

 exchanges

0.14

 behavior

0.14

 Interaction

0.14

Interaction

0.14

Activations Density 0.056%