INDEX

Explanations

step-by-step explanations or deep dives

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

illez

-0.12

ç®Ģåįķ

-0.10

 Fitz

-0.09

 stren

-0.09

 short

-0.08

æĤ

-0.08

æĿŁ

-0.08

antee

-0.08

ï½§

-0.08

 SUMMARY

-0.08

POSITIVE LOGITS

 detailed

0.42

è¯¦

0.35

 Detailed

0.34

Detailed

0.32

etailed

0.31

 detail

0.29

è©³ç´°

0.28

 Ð¿Ð¾Ð´ÑĢÐ¾Ð±

0.24

 detal

0.23

 comprehensive

0.23

Activations Density 0.225%

No Known Activations