How do you implement post-training quantization (PTQ) versus quantization-aware training (QAT) for a transformer-based model at TETRAMEM?
