6Decoder

6Decoder employs a neural network based on the Transformer Decoder, treating IPv6 addresses as textual sequences and using generative artificial intelligence to generate IPv6 candidate addresses.

Runtime environment

• Python 3.9.1
• pytorch 2.5.1
• pytorch-cuda 12.4

Seed set

IPv6 Hitlist provides an IPv6 Hitlist Service to publish responsive IPv6 addresses, aliased prefixes, and non-aliased prefixes.

Run example

1. Expand the compressed IPv6 representation into 32 hexadecimal characters without colons.

python seed_convert.py --file=data/Seed_S1_10K.txt

2. Run 6Decoder experiment program using default hyperparameters.

python Run6Decoder.py