INDEX

Explanations

code keywords and identifiers

The neuron fires on database‐query keywords and ORM method calls (e.g. SQL clauses like SELECT, FROM, JOIN or framework query methods like populate).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 even

-1.24

in

-1.06

 apoya

-0.96

 most

-0.94

for

-0.93

 refleja

-0.93

 without

-0.90

at

-0.90

to

-0.90

or

-0.87

POSITIVE LOGITS

⠀⠀⠀⠀

1.16

Каждый

1.13

Their

1.13

getc

1.08

初め

1.05

erir

1.04

좇

1.04

Отлич

1.04

hym

1.03

אם

1.03

Activations Density 0.004%