INDEX

Explanations

task in code

This neuron fires on occurrences of the identifier “Task” (e.g. class or API names) in source code.

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

ðsíða

-2.72

 khawatir

-2.36

聼

-2.34

を中心

-2.31

спеди

-2.30

坼

-2.28

这样一个

-2.25

izacao

-2.23

覬

-2.23

 universale

-2.20

POSITIVE LOGITS

3.31

↵↵

2.66

2.61

or

2.61

2.55

<i>

2.50

н

2.44

<strong>

2.44

In

2.41

be

2.36

Activations Density 0.003%