Skip to content

Popular repositories Loading

  1. kindling-spark-os kindling-spark-os Public

    Minimal read-only OS image for GB10 (DGX Spark / ASUS GX10) boxes

    Python 97 8

  2. glm-5.3-flash-gx10 glm-5.3-flash-gx10 Public

    GLM-5.3-Flash (NVFP4) at TP=4 across 4x ASUS GX10 / GB10 - vLLM + mentat instead of Ray, 524k context, DFlash2. Minimal working recipe.

    Python 94 13

  3. gr8 gr8 Public

    A from-scratch CUDA inference engine for 1-6 DGX Sparks (GB10): GLM-5.3, GLM-5.3-Flash and DeepSeek-V4.1-Flash

    Cuda 60 13

  4. exlr8 exlr8 Public

    EXLR8: a DGX Spark tuned quantization format

    HTML 14 1

  5. glm-5.3-flash-exlr8-gx10 glm-5.3-flash-exlr8-gx10 Public

    GLM-5.3-Flash exlr8 (EXL3-trellis experts) serving recipe for GB10: vLLM + mentat, TP=1/2/3/4, exlr8 MoE kernels

    Python 14 1

  6. mentat mentat Public

    A very-low-resource API-compatible replacement for Ray

    Rust 13 1

Repositories

Showing 10 of 17 repositories
  • deepseek-v4.1-flash-exlr8-4x-spark Public

    DeepSeek-V4.1-Flash exlr8 on four GB10 Sparks (TP=4) with gr8: fp8 KV, Engram, 32 chats at once, fast mode by default

    kindlingai/deepseek-v4.1-flash-exlr8-4x-spark's past year of commit activity
    Shell 0 MIT 0 0 0 Updated Nov 6, 2026
  • glm-5.3-flash-exlr8-1x-spark Public

    GLM-5.3-Flash exlr8 on one GB10 Spark with gr8: K2, int4 KV, on-disk prompt cache

    kindlingai/glm-5.3-flash-exlr8-1x-spark's past year of commit activity
    Dockerfile 5 0 0 0 Updated Nov 6, 2026
  • gr8 Public

    A from-scratch CUDA inference engine for 1-6 DGX Sparks (GB10): GLM-5.3, GLM-5.3-Flash and DeepSeek-V4.1-Flash

    kindlingai/gr8's past year of commit activity
    Cuda 60 MIT 13 28 12 Updated Nov 6, 2026
  • glm-5.3-flash-exlr8-2x-spark Public

    GLM-5.3-Flash exlr8 on two GB10 Sparks (TP=2) with gr8: K4, int4 KV, DFlash2, on-disk prompt cache

    kindlingai/glm-5.3-flash-exlr8-2x-spark's past year of commit activity
    Dockerfile 0 0 0 0 Updated Nov 6, 2026
  • glm-5.3-flash-gx10 Public

    GLM-5.3-Flash (NVFP4) at TP=4 across 4x ASUS GX10 / GB10 - vLLM + mentat instead of Ray, 524k context, DFlash2. Minimal working recipe.

    kindlingai/glm-5.3-flash-gx10's past year of commit activity
    Python 94 13 16 (1 issue needs help) 4 Updated Nov 6, 2026
  • glm-5.3-flash-exlr8-gx10 Public

    GLM-5.3-Flash exlr8 (EXL3-trellis experts) serving recipe for GB10: vLLM + mentat, TP=1/2/3/4, exlr8 MoE kernels

    kindlingai/glm-5.3-flash-exlr8-gx10's past year of commit activity
    Python 14 1 0 3 Updated Nov 6, 2026
  • exlr8 Public

    EXLR8: a DGX Spark tuned quantization format

    kindlingai/exlr8's past year of commit activity
    HTML 14 1 0 1 Updated Nov 6, 2026
  • glm-5.3-full-exl3-tp6 Public

    GLM-5.3 full EXL3 on DGX Spark / GB10: TP6 and TP4 builds - launchers, vLLM overlays, image, GPTQ dense-decode, eval, results

    kindlingai/glm-5.3-full-exl3-tp6's past year of commit activity
    Python 4 MIT 0 0 0 Updated Nov 6, 2026
  • model-loader Public

    Fetches large model files once, fans out to every box on the LAN

    kindlingai/model-loader's past year of commit activity
    Go 2 AGPL-3.0 0 2 1 Updated Nov 5, 2026
  • talos-dgx-kernel Public

    Minimal reproducible Talos DGX Spark kernel build and installation

    kindlingai/talos-dgx-kernel's past year of commit activity
    Python 5 0 0 0 Updated Nov 3, 2026