 Alexander Korolev
|
ed4bcc4369
2x re-cut: 512K context via max-num-seqs 4 + 10 GiB KV pin (bench 160/176)
|
1 month ago |
|
Alexander Korolev
|
144077f8a4
recipes: nvfp4-lab 2x — 256K context (2.4 GiB KV budget at GMU 0.88)
|
1 month ago |
|
Alexander Korolev
|
4bbf75aa57
recipes: GLM-5.3-Flash NVFP4-lab re-cut lanes (2x + 4x, r35)
|
1 month ago |
|
Alexander Korolev
|
623a906d5f
recipes: set parent_model for all quant/fork repos (GLM 5.1/5.2/5.3/5.3-Flash, Qwen3.8 27B+Flash-Next, MiniMax-M3, MiMo-V2.5, Solar-Open2, DSV4-vision env-driven) — report medal/tree grouping now ranks quants inside the base model's category
|
1 month ago |
|
Alexander Korolev
|
9079638ee4
lab recipes: instanttensor baked into lab image + enabled (verified: 631s vs 700s, two-pass loader, all boot markers + smoke 3901)
|
1 month ago |
 kilork
|
5ab8a05080
restore recipes/ in-tree — recipe + docker-images must share one revision
|
1 month ago |
|
kilork
|
d0d210e466
move recipe data to git.kilork.org/sm121/recipes (registry data repo)
|
1 month ago |
|
kilork
|
2599e39186
glm53-flash-nvfp4-4p67: MXFP8-attention support patchset + boot verified on ds4
|
1 month ago |
|
Alexander Korolev
|
96c6bfe777
feat(recipes): lab defaults to --enforce-eager — MTP+CUDA-graphs quality bug confirmed by A/B bench
|
1 month ago |
|
Alexander Korolev
|
9d72c8b804
feat(recipes): lab GLM-5.3 recipe = FujitsuPolycom proven stack, verified on ds4
|
1 month ago |
|
Alexander Korolev
|
b710a09bcb
feat(recipes): GLM-5.3-Flash NVFP4 variants — local-inference-lab + RedHatAI
|
1 month ago |
