Mini-SGLang port for Ascend NPUs with FIA, paged KV cache, ragged continuous batching, and request lifecycle safety.
Native AscendC Mamba2 selective scan / SSD forward-backward custom operator for Huawei Ascend 910B3 and 950PR, with CANN, torch_npu, A100 benchmarks and msprof profiling.
Ascend NPU fork of nanochat for LLM training with torch_npu/HCCL (experimental)
Ascend NPU (910B / CANN) port of karpathy/nanochat — adds a mid-training stage, a STEM eval suite, and a fast bf16 generation-eval path; training backend of climbmix (unofficial Nemotron-CLIMB reproduction).
KernelBench for Huawei Ascend NPU: Torch-NPU reference auditing and isolated Ascend C kernel evaluation.
Unofficial reproduction of Nemotron-CLIMB (CLustering-based Iterative Data Mixture Bootstrapping, NeurIPS 2025, arXiv:2504.13161) on Ascend NPUs — 5 rounds, ≈3,900 NPU-hours; learned mixtures beat every fixed-ratio baseline (+0.014–0.043 STEM); the selection mechanism is the weak link — fixed and validated live.
IndexTeam 模型（IndexTTS-2.5 / Translate-2B·9B / Homura-2B / 1.9B-Chat）× 华为昇腾 Atlas 300I Duo（Ascend 310P3）推理适配：5 模型实测跑通，2 个内核缺陷定位修复，70 算子支持度实测
To associate your repository with the torch-npu topic, visit your repo's landing page and select "manage topics."