INDEX

Explanations

on a surface

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

 hawk

-0.11

 Rowe

-0.10

lds

-0.10

 tents

-0.09

 tunnel

-0.09

 upright

-0.09

ilis

-0.09

scribe

-0.09

 excess

-0.09

erokee

-0.09

POSITIVE LOGITS

 platform

0.16

 platforms

0.14

 shoulders

0.13

 throne

0.13

platform

0.13

 branch

0.13

å¹³åı°

0.12

 convey

0.12

 Throne

0.11

mat

0.11

Activations Density 0.068%