53.7 kB
12 files
Updated about 1 month ago
Name
Size
.gitattributes1.52 kB
xet
README.md2.16 kB
xet
adversarial_attack.py10.9 kB
xet
decoder_model.py6.14 kB
xet
generate_data.py3.73 kB
xet
quantum_algorithm.py13 kB
xet
quantum_results.json1.4 kB
xet
requirements.txt58 Bytes
xet
run_adversarial.sh1.07 kB
xet
run_pipeline.sh1.36 kB
xet
setup_and_train.sh2.74 kB
xet
train_decoder.py9.6 kB
xet
README.md

Quantum Syndrome RL Decoder

Decodificador ML (Red Neuronal 3D CNN + RL) para extracción de síndromes ruidosos en computación cuántica.

Arquitectura

Componentes Principales

  1. QuantumSyndromeDecoder - Decodificador 3D CNN supervisado

    • Bloques residuales 3D con GroupNorm y GeLU
    • Entrada: Síndromes X/Z (batch, 2, D, D, T)
    • Salida: Predicciones de errores (batch, 4, D, D, T)
  2. SynergyExtractor - Extractor de síndrome consciente de ruido

    • Convolución temporal para historia de síndromes
    • Estimación de confianza por medición
  3. RLPolicyNetwork - Política RL para decodificación

    • Encoder compartido + cabezales de política/valor
    • Compatible con PPO/REINFORCE

Dataset

Datos sintéticos generados con simulador de frame Pauli:

  • Distancias: d=5, 7, 9
  • Tasas de error: p=0.01, 0.03, 0.05, 0.08
  • ~60,000 muestras totales

Entrenamiento

# Generar datos
python generate_data.py

# Entrenar decodificador supervisado
python train_decoder.py --mode supervised --epochs 50 --channels 128

# Pipeline completo
bash run_pipeline.sh

Referencias

  • Chamberland et al. (2025) - Fast and accurate AI-based pre-decoders for surface codes
  • Google Quantum AI (2024) - Quantum error correction below the surface code threshold
  • Higgott & Gidney (2025) - Sparse Blossom decoder

Enlaces

Generated by ML Intern

This model repository was generated by ML Intern, an agent for machine learning research and development on the Hugging Face Hub.

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "Samabe1109/quantum-syndrome-rl-decoder"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id)

For non-causal architectures, replace AutoModelForCausalLM with the appropriate AutoModel class.

Total size
53.7 kB
Files
12
Last updated
May 23
Pre-warmed CDN
US EU US EU

Contributors