INDEX

Explanations

dialogue separators and commands

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

/respond

-0.10

.MixedReality

-0.09

 Goldberg

-0.08

Jam

-0.08

.dataTables

-0.08

HeaderCode

-0.08

\OptionsResolver

-0.08

Epoch

-0.08

bsub

-0.08

eref

-0.08

POSITIVE LOGITS

cal

0.10

 draft

0.09

LM

0.08

 aforementioned

0.08

ies

0.08

ito

0.08

GG

0.08

ouden

0.08

oodle

0.08

 exert

0.08

Activations Density 0.156%