INDEX

Explanations

procrastination, perfectionism, paralysis

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

 selfish

-0.11

 assim

-0.10

Tro

-0.10

éĿ

-0.10

 Pause

-0.09

 patience

-0.09

eid

-0.09

 vend

-0.09

_planes

-0.09

 axes

-0.09

POSITIVE LOGITS

 paralysis

0.16

 Momentum

0.15

omentum

0.15

 momentum

0.15

crast

0.14

 paralyzed

0.14

unm

0.14

 task

0.13

 perfection

0.13

 PROC

0.13

Activations Density 0.108%