功能指南¶
本节提供 vLLM Ascend 功能的详细使用指南。
:::{toctree} :caption: 功能指南 :maxdepth: 1 graph_mode cpu_binding Ai_QoS_introduction_en quantization sleep_mode structured_output lora expert_parallelism_load_balancer netloader rfork dynamic_batch epd_disaggregation kv_pool layerwise_kv_pool kv_cache_cpu_offload external_dp large_scale_ep ucm_deployment Fine_grained_TP layer_sharding speculative_decoding context_parallel weight_prefetch sequence_parallelism batch_invariance lmcache_ascend_deployment dynamic_chunk_pipeline_parallel flash_attention :::