noonghunna b2cede650f Merge pull request #433 from steamEngineer/feat/lmcache-nvlink-detect ...

feat(lmcache): NVLink auto-detect + Compose v5 entrypoint fix

2026-06-18 22:38:19 +05:00

..

diffusiongemma-26b-a4b/vllm

Add NVLink auto-config to gemma-4-26b-a4b + diffusiongemma duals

2026-06-18 17:17:16 +00:00

gemma-4-12b

Fix NVLink auto-detect dead on Docker Compose v5.1+

2026-06-18 17:17:04 +00:00

gemma-4-26b-a4b

Add NVLink auto-config to gemma-4-26b-a4b + diffusiongemma duals

2026-06-18 17:17:16 +00:00

gemma-4-31b

Fix NVLink auto-detect dead on Docker Compose v5.1+

2026-06-18 17:17:04 +00:00

qwen3-omni-30b-a3b/vllm-omni

Add NVLink auto-config to gemma-4-26b-a4b + diffusiongemma duals

2026-06-18 17:17:16 +00:00

qwen3.6-27b

Merge pull request #433 from steamEngineer/feat/lmcache-nvlink-detect

2026-06-18 22:38:19 +05:00

qwen3.6-35b-a3b

Add NVLink auto-config to qwen3.6-35b-a3b dual (Production — rebench-gated)

2026-06-18 17:17:27 +00:00

qwen3.6-40b-deckard/llama-cpp/compose/dual/piehsoft-q6k

deckard-40b: anti-loop sampling defaults (rep-penalty 1.1 + repeat-last-n 256) + DRY opt-in (#402)

2026-06-14 04:01:50 +05:00

vibethinker-3b

docs(vibethinker-3b): add measured one-shot-coding scores (HE+ 97%, LCB 83%)

2026-06-16 21:08:40 +00:00