INDEX

Explanations

several reasons

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

ioni

-0.11

Nim

-0.10

 tatto

-0.10

onte

-0.10

feof

-0.09

quam

-0.09

infeld

-0.09

ereotype

-0.08

Mim

-0.08

POSITIVE LOGITS

 reasons

0.35

 several

0.32

 Reasons

0.27

 Several

0.26

çĲĨçĶ±

0.24

Several

0.24

åĩłä¸ª

0.23

åİŁåĽł

0.21

veral

0.20

 reason

0.20

Activations Density 0.256%