INDEX

Explanations

rolling hills

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

 woodland

-0.18

 troop

-0.11

ãģħ

-0.10

 ãĥ¼

-0.09

 democrat

-0.09

 Woodward

-0.09

 trÃº

-0.09

PILE

-0.08

 weather

-0.08

ewood

-0.08

POSITIVE LOGITS

 fields

0.36

 rolling

0.34

 Fields

0.26

 Rolling

0.26

fields

0.26

rolling

0.26

Fields

0.24

_fields

0.21

.fields

0.21

 countryside

0.21

Activations Density 0.107%