extract overall sentiment or response

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

eroon

-0.09

ihu

-0.09

lesi

-0.09

.opendaylight

-0.09

ertino

-0.09

goo

-0.08

acle

-0.08

ToEnd

-0.08

POSITIVE LOGITS

 table

0.13

 header

0.11

HEADER

0.10

emo

0.10

Ð°Ð±Ð»Ð¸ÑĨ

0.09

 Cabr

0.09

 ocas

0.09

Header

0.09

 tables

0.09

 once

0.09

Activations Density 0.030%

No Known Activations