INDEX

Explanations

capitals of US and Argentina

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

Elm

-0.08

Sol

-0.08

 Bern

-0.08

 Vital

-0.08

CLUDING

-0.08

Bel

-0.08

lm

-0.08

 Flesh

-0.07

POSITIVE LOGITS

çŃĶæ¡Ī

0.13

 answer

0.12

 nackte

0.11

CLU

0.09

answer

0.09

 actual

0.09

EMPLARY

0.09

rÃ¡Å¾

0.09

 Antwort

0.09

/loose

0.09

Activations Density 0.179%