INDEX

Explanations

network and networking

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

 Networks

-0.11

 networking

-0.10

_network

-0.10

ystone

-0.10

 networks

-0.10

 Fame

-0.10

arily

-0.10

Networking

-0.09

ä»

-0.09

esto

-0.09

POSITIVE LOGITS

ed

0.30

 lÆ°á»Ľi

0.24

cdf

0.13

ingly

0.12

werk

0.12

edn

0.12

erate

0.11

çµ¡

0.11

er

0.11

 centr

0.11

Activations Density 0.025%