INDEX

Explanations

pilates, pilgram, piloted, PIL import

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

OVE

-0.10

iffies

-0.09

 Brady

-0.09

ingly

-0.09

ubits

-0.09

rame

-0.09

 Cone

-0.09

ADR

-0.09

edo

-0.09

alyzer

-0.09

POSITIVE LOGITS

Pil

0.17

pil

0.16

 Pill

0.15

 pill

0.13

oting

0.13

ates

0.13

pill

0.12

grim

0.12

iminary

0.12

-pill

0.12

Activations Density 0.023%