8000
Minimal read-only OS image for GB10 (DGX Spark / ASUS GX10) boxes
GLM-5.3-Flash (NVFP4) at TP=4 across 4x ASUS GX10 / GB10 - vLLM + mentat instead of Ray, 524k context, DFlash2. Minimal working recipe.
GLM-5.3-Flash exlr8 (EXL3-trellis experts) serving recipe for GB10: vLLM + mentat, TP=1/2/3/4, exlr8 MoE kernels
A from-scratch CUDA inference engine for 1-6 DGX Sparks (GB10): GLM-5.3, GLM-5.3-Flash and DeepSeek-V4.1-Flash
GLM-5.3-Flash (NVFP4) at TP=4 across 4x ASUS GX10 / GB10 - vLLM + mentat instead of Ray, 524k context, DFlash2. Minimal working recipe.
GLM-5.3-Flash exlr8 (EXL3-trellis experts) serving recipe for GB10: vLLM + mentat, TP=1/2/3/4, exlr8 MoE kernels
GLM-5.3-Flash exlr8 on one GB10 Spark with gr8: K2, int4 KV, on-disk prompt cache
GLM-5.3 full EXL3 on DGX Spark / GB10: TP6 and TP4 builds - launchers, vLLM overlays, image, GPTQ dense-decode, eval, results
GLM-5.3-Flash exlr8 on two GB10 Sparks (TP=2) with gr8: K4, int4 KV, DFlash2, on-disk prompt cache
Loading…
Loading…