AQ1-d3-pretrained

AQ1 decoder pretrained on Stim simulator data for d=3 rotated surface code. Ready for finetuning on real QPU hardware.

Architecture

Recurrent transformer decoder: StabilizerEmbedding → SyndromeTransformer (4 layers) → GRU → Readout

• Parameters: 1,117,185

• Code: d=3 rotated surface code, Z-basis memory

Training

• 2.16M Stim samples, 120 configs (6 noise levels × 20 round counts)

• 20 epochs, lr=1e-4, batch=1024

• Val LER: 0.0286

Usage

import torch

from huggingface_hub import hf_hub_download

from model import AQ1Decoder



ckpt_path = hf_hub_download("RafalMa/AQ1-d3-pretrained", "tier1_best.pt")

model = AQ1Decoder(n_stabilizers=8, d_model=128, n_heads=4, n_transformer_layers=4)

ckpt = torch.load(ckpt_path, map_location='cpu')

model.load_state_dict(ckpt['model_state'])

model.eval()

Finetune on your own QPU data using AQ1.