INDEX

Explanations

names followed by identifying phrases

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

à¥°

-0.09

Named

-0.08

 Named

-0.08

 unnamed

-0.08

_qos

-0.08

quartered

-0.08

{{{

-0.08

Â

-0.08

pter

-0.08

orr

-0.08

POSITIVE LOGITS

who

0.23

who

0.20

 whom

0.19

 whose

0.17

 alias

0.14

 quien

0.14

alias

0.14

whose

0.13

 ktÃ³ry

0.12

age

0.11

Activations Density 0.130%