INDEX

Explanations

PostgreSQL, ubuntu, quant, elixir, programming

The neuron fires on tokens that are names of online discussion communities or subreddits (e.g. “Futurology,” “ubuntu-us-sc,” “quant,” “Google-Picasa-Data-API,” “Alexandria-list”).

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

in

-1.61

 with

-1.59

for

-1.52

 from

-1.46

on

-1.41

💋

-1.39

😱

-1.39

😇

-1.35

🤞

-1.34

 through

-1.32

POSITIVE LOGITS

wherein

1.49

Penjelasan

1.35

 чтобы

1.34

最近

1.23

 SPECIFIC

1.20

 EVEN

1.20

Spesifikasi

1.20

小さな

1.18

1.16

 capuz

1.15

Activations Density 0.003%