INDEX

Explanations

expectations and responsibilities

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

 specifics

-0.10

anio

-0.09

igin

-0.09

raw

-0.09

.scalablytyped

-0.09

@nate

-0.08

 Integrity

-0.08

Rit

-0.08

 effectiveness

-0.08

 Craw

-0.08

POSITIVE LOGITS

 expectations

0.41

 Expect

0.33

 expectation

0.30

expect

0.28

Expect

0.28

 EXPECT

0.25

 expect

0.25

 expected

0.25

 roles

0.24

 expects

0.23

Activations Density 0.094%