INDEX

Explanations

IP.ttl field

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

Negative Logits

INLINE

-0.09

untime

-0.09

 strangers

-0.08

ustos

-0.08

 inline

-0.08

 nomin

-0.08

-FIRST

-0.08

anic

-0.08

 originals

-0.08

ppy

-0.08

POSITIVE LOGITS

 fields

0.16

 field

0.15

 variable

0.15

 parameter

0.14

 purpose

0.13

 concept

0.13

 reserved

0.12

Purpose

0.12

 Purpose

0.11

åŃĹæ®µ

0.11

Activations Density 0.410%

No Known Activations