INDEX

Explanations

list remove, retain, and index operations

The neuron fires on code tokens for “remove” (and related removal/erase) operations, i.e. method or function calls that remove elements.

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

 zincir

-0.93

Monument

-0.83

ウド

-0.82



-0.80

んでる

-0.79

撂

-0.79

 بغداد

-0.78

selig

-0.77

 دیگری

-0.77

izient

-0.77

POSITIVE LOGITS

by

1.41

all

1.27

 from

1.22

 first

1.08

at

1.05

 removes

1.02

 remove

0.94

 with

0.91

 this

0.90

 where

0.90

Activations Density 0.006%