18:11
2026-07-15
byteiota.com
machine-learning
PyTorch 2.13: FlexAttention on Apple Silicon and 4x LLM Memory Savings
PyTorch 2.13, released this week, brings FlexAttention support for Apple Silicon with up to 12x speedup on sparse patterns and a new fused LinearCrossEntropyLoss operator that cuts peak GPU memory by …