xcpatr922JNP/maya-dgx-spark-bench
Maya-S v2 IQ2_XXS on a DGX Spark (aarch64) - benchmark data Speed measurements for GLM-5.3-Flash (the Maya-S-v2-IQ2_XXS quant, 320 B parameters / ~18 B active, 89.83 GiB at 2.0625 bpw) on one NVIDIA DGX Spark (GB10, 20 ARM cores, 120 GB of RAM shared with the GPU), taken with two independent engines on the same weights and the same machine: Maya - Project Maya v1.1.26 + the aarch64 port, with and without the DFlash2 drafter (--dflash, 0.7 GB) so the drafter's effect is… See the full description on the dataset page: https://huggingface.co/datasets/xcpatr922JNP/maya-dgx-spark-bench.
| Author / organization | xcpatr922JNP |
|---|---|
| License (as declared) | mit |
| Downloads | 0 (as reported by Hugging Face) |
| Likes | 0 |
| Created | 25 October 2026 |
| Last updated at source | 25 October 2026 |
Recent changes
- 25 Oct 2026 Description: none → Maya-S v2 IQ2_XXS on a DGX Spark (aarch64) - benchmark data Speed measurements for GLM-5.3-Flash (the Maya-S-v2-IQ2_XXS quant, 320 B parameters / ~18 B active, 89.83 GiB at 2.0625 bpw) on one NVIDIA DGX Spark (GB10, 20 ARM cores, 120 GB of
- 25 Oct 2026 Topics added task_categories:text-generation removed en, zh
Changes detected by Silytics between refreshes of the source.
More from xcpatr922JNP
Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 25 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.