代码库
Home Assistant integration for GeekMagic displays (SmallTV Pro/Ultra). Server-side rendering, no device flashing required.
Python
custom-integrationdisplayesp8266geekmagichacshome-assistantiotsmalltv
Qwen3.8-Flash-Next (2.50bpw EXL3) on 2× RTX 5090 with ExLlamaV3 + TabbyAPI: 262k context, 901k-token 8-bit KV pool, MTP + prompt-lookup drafting. 265 t/s code decode at 1 stream (414 t/s on code edits), 626 t/s at 4, 844 t/s at 8, ~12k t/s prefill. Launcher, kernel overlays, probes and dated measurements.
Python
exl3exllamav3llm-inferencelocal-llmmixture-of-expertsmulti-gpuquantizationqwenqwen3rtx-5090speculative-decodingtabbyapi