Expand description
SIMD-optimized operations for maximum performance
Uses portable SIMD when available, falls back to scalar operations
Functionsยง
- add_
simd - SIMD-optimized element-wise addition
- gelu_
simd - SIMD-optimized GELU
- relu_
simd - SIMD-optimized ReLU (2-4x faster than scalar)
- sigmoid_
simd - SIMD-optimized sigmoid