INDEX

Explanations

structure and arrangement

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

ãģĬãĤĬ

-0.11

ister

-0.10

ISTER

-0.09

ity

-0.09

inel

-0.09

ophobia

-0.09

ileo

-0.09

HUD

-0.09

ÑīÐ°

-0.09

argent

-0.09

POSITIVE LOGITS

urally

0.21

ural

0.20

ured

0.18

ures

0.16

uring

0.16

ture

0.15

Ø§ÙģØªÙĩ

0.15

 trÃºc

0.14

URAL

0.14

less

0.13

Activations Density 0.036%