INDEX

Explanations

escalation and resolution

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

 Indo

-0.10

 immersion

-0.09

Cv

-0.09

 massaggi

-0.09

zi

-0.09

ipe

-0.09

OSP

-0.08

 Higgins

-0.08

 writ

-0.08

 Justice

-0.08

POSITIVE LOGITS

escal

0.19

 escal

0.16

 resolutions

0.15

 escalation

0.15

 resolution

0.15

 ticket

0.14

Resolution

0.14

 tickets

0.14

esk

0.14

 escalated

0.13

Activations Density 0.128%