We research the segmentation of lesions in medical images using the PST-UNet model, verifying the preservation of spatial features of medical lesions, and improving the accuracy of medical lesion segmentation. The PST-UNet (positive distribution data Swin transformer) model combines transformer and U-shaped structures. It uses cascaded convolution fusion modules to integrate the encoder’s multi-scale features. The encoder includes the Swin transformer block and the entire Gaussian Error Linear Unit (GELU) activation function. The decoder uses the Swin transformer block, upsampling, and skip connections from the cascaded convolution fusion modules. This approach effectively preserves more spatial features of medical lesions and improves the accuracy of kidney lesion segmentation, achieving a 0.02289 improvement in accuracy. The PST-UNet model can improve the segmentation accuracy of normally distributed medical lesion data, which has a positive effect on the treatment of kidney disease.

This paper presents HPN, Alibaba Cloud's data center network for large language model (LLM) training. Due to the differences between LLMs and general cloud computing (e.g., in terms of traffic patterns and fault tolerance), traditional data center networks are not well-suited for LLM training. LLM training produces a small number of periodic, bursty flows (e.g., 400Gbps) on each host. This characteristic of LLM training predisposes Equal-Cost Multi-Path (ECMP) to hash polarization, causing issues such as uneven traffic distribution. HPN introduces a 2-tier, dual-plane architecture capable of interconnecting 15K GPUs within one Pod, typically accommodated by the traditional 3-tier Clos architecture. Such a new architecture design not only avoids hash polarization but also greatly reduces the search space for path selection. Another challenge in LLM training is that its requirement for GPUs to complete iterations in synchronization makes it more sensitive to singlepoint failure (typically occurring on ToR). HPN proposes a new dual-ToR design to replace the single-ToR in traditional data center networks. HPN has been deployed in our production for more than eight months. We share our experience in designing, and building HPN, as well as the operational lessons of HPN in production.