INDEX

Explanations

don't hesitate to ask

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

 Shim

-0.10

esome

-0.09

ofile

-0.09

TON

-0.08

room

-0.08

 undes

-0.08

 Nack

-0.08

Calibri

-0.08

Tro

-0.08

cest

-0.08

POSITIVE LOGITS

 hesitate

0.15

 hesitation

0.13

 contact

0.12

hes

0.12

Ask

0.12

ask

0.12

 contacting

0.12

 Î½Î±

0.11

hes

0.11

èģĶç³»

0.11

Activations Density 0.006%