INDEX

Explanations

list structures

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

 GOODMAN

-0.10

/*****************************************************************************\n

-0.09

reas

-0.09

@brief

-0.09

iore

-0.09

DEX

-0.09

 Bram

-0.08

ryfall

-0.08

 fontStyle

-0.08

 Retail

-0.08

POSITIVE LOGITS

Â§Ãĥ

0.09

arpa

0.09

 Garn

0.09

 integ

0.09

 foster

0.09

Cah

0.09

 Bris

0.08

chner

0.08

 flesh

0.08

ika

0.08

Activations Density 0.057%