Popular repositories Loading

1. GLM-5.3-Flash-NVFP4-TP3-3x-DGX-Spark GLM-5.3-Flash-NVFP4-TP3-3x-DGX-Spark Public

   Run zai-org/GLM-5.3-Flash (NVFP4) on 3x NVIDIA DGX Spark with vLLM, TP=3 + EP, DFlash2 speculative decoding, CUDA graphs. Full recipe with rationale, benchmarks, patches, and what we tried.

   Python 2

2. cuda-exl3 cuda-exl3 Public

   Forked from Zeuss5/cuda-exl3

   EXL3 (ExLlamaV3 trellis) CUDA kernels for Blackwell: dense + MoE GEMM and a fused sparse-MLA attention backend, with a vLLM plugin

   Cuda

3. GLM-5.3-Flash-EXL3-TP3-3x-DGX-Spark GLM-5.3-Flash-EXL3-TP3-3x-DGX-Spark Public

   Run zai-org/GLM-5.3-Flash as an EXL3 4bpw checkpoint on 3x NVIDIA DGX Spark (TP=3 + expert parallel, DFlash2) with vLLM and cuda-exl3 — plus a measured TP=2 track for two-node setups. Full recipe w…

   Python

4. nccl-mesh-plugin nccl-mesh-plugin Public

   Forked from autoscriptlabs/nccl-mesh-plugin

   C