INDEX

Explanations

routine and stagnation

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

 stumbling

-0.11

lon

-0.10

lon

-0.10

 Spears

-0.09

Ä±lÄ±Ã§

-0.09

 lonely

-0.09

 loneliness

-0.09

 persistence

-0.09

BYTES

-0.09

 weakness

-0.09

POSITIVE LOGITS

 compl

0.28

 plateau

0.24

 coast

0.23

rut

0.23

 routine

0.22

 Plate

0.21

 Routine

0.20

rut

0.19

 plate

0.19

plate

0.19

Activations Density 0.114%