Watch xcx_shine / omni-npu forked from omniai / omni-npu

A vLLM (0.12.0) out-of-tree platform plugin that enables running vLLM on NPU (Ascend/torch_npu).

Watch xcx_shine / omni_infer forked from omniai / omni_infer

Omni_Infer is a suite of inference accelerators designed for the Ascend NPU platform, offering native support and an expanding feature set.

Watch xcx_shine / ModelZoo-PyTorch forked from Ascend / ModelZoo-PyTorch

暂停

Watch xcx_shine / parallel decoding token-tree

用于并行解码，可把输入序列组织成树形式。