Training a autoregressive network to do quantum error correction.
Link to the article: https://doi.org/10.48550/arXiv.2307.09025
python code_generator.py --d 3 --k 1 --seed 0 -c_type 'sur'if k> k' , where k' is the number of logical qubits, will remove stabilizers randomly from origin code.
python training.py -save True -n_type 'made' -c_type 'sur' -d 3 -k 1 -seed 0 -er 0.189 -device 'cuda:0' -batch 10000 -epoch 50000 -depth 3 -width 20python training.py -save True -n_type 'trade' -c_type 'sur' -d 3 -k 1 -seed 0 -er 0.189 -device 'cuda:0' -batch 10000 -epoch 50000 -d_model 128 -n_heads 4 -d_ff 512 -n_layers 2 python forward_decoding.py -save True -c_type 'sur' -d 3 -k 1 -seed 0 -device 'cuda:0' -n_type 'trade' -e_model 'dep' -trials 10000 -er 0.189