INDEX

Explanations

reliability, consistency, following through

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

↵↵

1.16

<eos>

1.06

When

0.82

0.80

↵↵↵

0.80

<h2>

0.77

<strong>

0.76

‌

0.75

 amid

0.75

 when

0.74

POSITIVE LOGITS

 Pubmed

1.32

(${

1.27

1.24

 १०

1.17

<0x15>

1.17

 १३

1.17

:).

1.16

٣

1.16

=$_

1.15

ｇ

1.15

Activations Density 0.020%