新发现
wallawalla47/infernix 68cb699 — 122B 提示的 CPU 辅助最远扩展到 575 列
infernix 的 2026-11-05 提交把 122B 模型提示的 CPU 辅助范围从 255 列扩大到 575 列,减轻了 256 列之后出现的吞吐量断崖。
infernix 的 2026-11-05 提交把 122B 模型提示的 CPU 辅助范围从 255 列扩大到 575 列,减轻了 256 列之后出现的吞吐量断崖。
GitHub 项目说明: C++/CUDA single-GPU inference engine for Qwen3.8 (incl. Qwen3.8-Flash-Next with offloaded experts) on the RTX 5090; grew from NInfer
在 GitHub 查看 ↗infernix 的 2026-11-05 提交把 122B 模型提示的 CPU 辅助范围从 255 列扩大到 575 列,减轻了 256 列之后出现的吞吐量断崖。
没有符合条件的动态。