INDEX

Explanations

trees and types of trees

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

 woodland

-0.14

 Weed

-0.11

 Straw

-0.11

 petals

-0.09

Î±Î½Î¬

-0.09

ede

-0.09

 Berger

-0.09

 rencont

-0.09

 Leon

-0.09

_sf

-0.08

POSITIVE LOGITS

 trees

0.45

 tree

0.34

 Trees

0.32

trees

0.28

æłĳ

0.27

Trees

0.26

tree

0.25

 Tree

0.24

 Ð´ÐµÑĢÐµÐ²Ð°

0.24

 Ð´ÐµÑĢÐµÐ²

0.23

Activations Density 0.087%