INDEX

Explanations

mathematical notation and variables

The neuron fires on two‐letter letter‐only tokens (like “ab,” “ac,” “xy,” “uu,” “WG,” “cq,” etc.), i.e. concatenated variable names or variable pairs.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 japon

-1.35

Hoi

-1.33

нии

-1.24

 protože

-1.23

کش

-1.23

Philippines

-1.20

 remboursement

-1.20

 jardin

-1.20

 hôtel

-1.18

 consentement

-1.18

POSITIVE LOGITS

 with

1.39

 Wissenschaften

1.36

討論

1.34

and

1.31

看見

1.24

We

1.23

 Trying

1.22

 Have

1.22

หนึ่ง

1.21

 proposal

1.20

Activations Density 0.090%