INDEX

Explanations

a continuation or new segment

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

ctest

-0.10

odore

-0.09

 ï½ī

-0.09

geries

-0.09

usal

-0.09

IVAL

-0.09

eros

-0.09

zar

-0.08

LowerCase

-0.08

POSITIVE LOGITS

firm

0.12

parent

0.12

just

0.11

pend

0.10

pare

0.10

 Bliss

0.10

prox

0.10

ument

0.10

 inst

0.09

lerce

0.09

Activations Density 0.029%