PerRow# class torch.ao.quantization.observer.PerRow[source]# Represents row-wise granularity in quantization. This is a special case of per-axis quantization and is unique to Float8 matmuls where the input is quantized with a block_size of (1, …, input.shape[-1]). And the weight is quantized with a block_size of (1, weight.shape[1]).