INDEX

Explanations

phrases indicating time progression and duration

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 earlier

-0.08

 Earlier

-0.07

 lately

-0.07

Earlier

-0.07

yan

-0.07

lant

-0.06

 kone

-0.06

 certain

-0.06

 recently

-0.06

 Twice

-0.06

POSITIVE LOGITS

 subsequent

0.18

 subsequently

0.17

 thereafter

0.15

 ensuing

0.13

 later

0.12

 then

0.11

later

0.11

 afterwards

0.11

 next

0.11

 further

0.11

Activations Density 0.035%