INDEX

Explanations

sentiment classification

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

++$

-0.09

kees

-0.09

 microbes

-0.08

azel

-0.08

èª

-0.08

tics

-0.08

arra

-0.08

 ÙħÙĨØ·

-0.08

uffle

-0.08

oph

-0.08

POSITIVE LOGITS

 overall

0.16

 Overall

0.13

overall

0.12

Overall

0.12

 sentiment

0.11

 neutral

0.11

 positive

0.11

ê¸

0.10

.wp

0.09

 negative

0.09

Activations Density 0.053%