INDEX

Explanations

number of variety range

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

arella

-0.11

Intialized

-0.11

anggan

-0.10

_GPU

-0.10

radu

-0.09

 createAction

-0.09

ipop

-0.09

 UNSIGNED

-0.09

$MESS

-0.08

POSITIVE LOGITS

0.09

 Drill

0.09

 Overall

0.08

 Timing

0.08

 overall

0.08

 Rear

0.08

 Wade

0.08

 odds

0.08

 Experiment

0.08

Activations Density 0.014%