INDEX

Explanations

asking for advice

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

EMPLARY

-0.11

ilos

-0.10

.Formatter

-0.10

įng

-0.09

asal

-0.08

iggins

-0.08

indsight

-0.08

%č\n

-0.08

çŃ

-0.08

annotate

-0.08

POSITIVE LOGITS

How

0.13

Can

0.13

 specifically

0.13

how

0.12

 HELP

0.12

How

0.12

 Specifically

0.12

Can

0.12

HELP

0.11

how

0.11

Activations Density 0.311%