QPARAM_ALIGN

burn_std::tensor::quantization

Constant QPARAM_ALIGN 

Source 
pub const QPARAM_ALIGN: usize = _; // 4usize
Expand description

Alignment (in bytes) for quantization parameters in serialized tensor data.

NOTE: This is currently f32-based since scales were originally always f32. With QuantParam now supporting different precisions (F16, BF16, etc.), this alignment may need to be revisited in the future.