INDEX

Explanations

references to significant "steps" or advancements in various contexts

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

udu

-0.07

ream

-0.07

ALS

-0.06

 tamp

-0.06

lage

-0.06

opi

-0.06

kir

-0.06

tas

-0.06

aji

-0.06

lands

-0.06

POSITIVE LOGITS

 taken

0.11

 Taken

0.10

 toward

0.10

 towards

0.10

Taken

0.10

taken

0.10

 step

0.09

Towards

0.09

 Step

0.08

.step

0.08

Activations Density 0.007%