INDEX

Explanations

answer to questions

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

 disp

-0.10

oku

-0.09

 Annex

-0.08

ald

-0.08

 merc

-0.08

asic

-0.08

ausal

-0.08

coder

-0.08

chan

-0.08

Agu

-0.08

POSITIVE LOGITS

 answer

0.31

answer

0.23

çŃĶæ¡Ī

0.22

 answers

0.18

.answer

0.17

 Answer

0.17

\tanswer

0.17

(answer

0.16

_answer

0.16

ans

0.16

Activations Density 0.030%