INDEX

Explanations

blocking action of proteins

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

ãĤŀ

-0.11

inel

-0.11

 strom

-0.10

 proh

-0.09

_CNTL

-0.09

 Strom

-0.09

vag

-0.09

uvw

-0.09

eto

-0.09

SCN

-0.09

POSITIVE LOGITS

cap

0.28

 vessel

0.27

 vessels

0.27

 angi

0.27

angi

0.23

Ang

0.21

Ang

0.21

spr

0.20

 endoth

0.19

endo

0.19

Activations Density 0.024%