INDEX

Explanations

of current or alternatives

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

Negative Logits

-0.11

opers

-0.09

inker

-0.09

amba

-0.09

atch

-0.09

çī

-0.08

exemple

-0.08

omy

-0.08

ady

-0.08

errs

-0.08

POSITIVE LOGITS

 candidate

0.17

 proposed

0.16

 alternative

0.15

 different

0.14

 tested

0.13

 competing

0.13

 investments

0.12

 strategies

0.12

Candidate

0.12

 decisions

0.12

Activations Density 0.153%

No Known Activations