INDEX

Explanations

some of the

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

 detriment

-0.11

azar

-0.10

imi

-0.10

roz

-0.09

iba

-0.09

 sovere

-0.09

 CONSTANTS

-0.09

eas

-0.09

ubar

-0.09

 staples

-0.09

POSITIVE LOGITS

 pointer

0.09

 major

0.09

 advoc

0.09

 termin

0.09

 rhetoric

0.09

ilar

0.09

 folds

0.09

 enlisted

0.09

 prime

0.08

 barn

0.08

Activations Density 0.153%