INDEX

Explanations

carry out instructions or tasks

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

lee

-0.10

ernals

-0.09

 Intercept

-0.09

akk

-0.09

uer

-0.09

 Matte

-0.09

 Schwarz

-0.09

åģļ

-0.09

hs

-0.09

-0.08

POSITIVE LOGITS

 plan

0.18

 tasks

0.18

 duties

0.17

 plans

0.16

Ubergraph

0.15

 instructions

0.14

Dut

0.14

tasks

0.13

è®¡åĪĴ

0.13

 orders

0.12

Activations Density 0.070%