 Alexander Korolev
|
b3d61b7af0
separate r18 from r12: dedicated docker-images/glm53-jovian-r18 dir + glm53-jovian-r18 tag; r12 dir restored to r17 pins; recipes renamed to r18 lanes
|
1 month ago |
|
Alexander Korolev
|
6327352f09
untrack regenerated data/benchmarks.sqlite (gitignore)
|
1 month ago |
|
Alexander Korolev
|
fe070bebdd
night shift: r18 built + full lane matrix (DCP4 161/176 best; eugr/DFlash2/dsv4fv-eugr documented)
|
1 month ago |
|
Alexander Korolev
|
4e6b20c673
dflash2 lane: drafter kv_cache_dtype auto (fp8 KV invalid on SM121 outside B12X)
|
1 month ago |
|
Alexander Korolev
|
a8ed91b8b5
dflash2 lane: pin drafter attention_backend FLASH_ATTN (auto-select poisoned by global B12X)
|
1 month ago |
|
Alexander Korolev
|
3976ebe72f
dflash2 lane: drop drafter B12X (kv_cache_dtype constraint; main keeps B12X)
|
1 month ago |
|
Alexander Korolev
|
bddf8d552e
add DFlash2 k=7 drafter lane (r18 img, A/B vs MTP)
|
1 month ago |
|
Alexander Korolev
|
fb4692fdab
dcp4 probe: add --cp-kv-cache-interleave-size 4 (GLM C4 DCP requirement)
|
1 month ago |
|
Alexander Korolev
|
2191c7018b
fix 4x recipes: tensor_parallel_size 4 (world size must cover nnodes)
|
1 month ago |
|
Alexander Korolev
|
b2135bab66
add r18 4x probe variants: DCP4 (TP4+decode-CP4) and nvfp4_ds_mla KV
|
1 month ago |
|
Alexander Korolev
|
eb8646ed43
add r17abi-4x probe recipe (4-node r18 cells: DCP4/RoCE/#646/nvfp4 KV)
|
1 month ago |
|
Alexander Korolev
|
d6076f1d8e
deps re-stage note: MSA cutlass pin (eb61c911) staged too; AppleDouble junk purged
|
1 month ago |
|
Alexander Korolev
|
bd1dc24418
deps re-stage note: FA's own cutlass pin (nvidia/cutlass 62750a2b) now in deps-src.tar.gz; vllm staged tree 67e7e6e
|
1 month ago |
|
Alexander Korolev
|
ccd615fcad
fix hardmode for night shift + future: version-pin tool (2.6.1.dev45, --hardmode = 88 scenarios), re-baseline anchor lane before gating; reports commit to recipes/<lane>/reports/; new eugr-4x test recipe (2x stays production default until 4x dramatic)
|
1 month ago |
|
Alexander Korolev
|
54dfffeb24
eugr GLM lane: document forum 381350#364 parity + DFlash2 alt spec (night queue 3b: k=7 incoai drafter A/B vs MTP k=5)
|
1 month ago |
|
Alexander Korolev
|
82a8368e5b
night queue: DCP4 probe (sparkring reference, r17abi-first, no build dep) + #646 geometry check; refresh r18 pins to 343661fe/1adb8a38
|
1 month ago |
|
Alexander Korolev
|
250194c1c0
r18: add PR #646 (fine-grained SWA prefix hits + decoupled GLM53 target/recurrent blocks, 5.4x KV capacity); advance bases to new tips (vllm a8c796f3 w/ #649, b12x b58f34ea w/ #315)
|
1 month ago |
|
Alexander Korolev
|
f8ab660d6c
new lane dsv4-flash-vision-eugr (eugr b12x nightly + Vision-Exp, DSpark-3 adaptive, plain-B12X attn); night-shift queue: r18 build, redhat MTP, eugr GLM, dsv4-vision eugr
|
1 month ago |
|
Alexander Korolev
|
f1d9e635f0
new lane glm53-flash-nvfp4-eugr: eugr B12X full-cluster dual-Spark variant (nightly-20260904 image, RoCE one-shot, MTP k=5 humming) — A/B cell vs r18; dry-run verified
|
1 month ago |
|
Alexander Korolev
|
02b94db96f
redhat lane: add MTP k=5 — model_mtp.safetensors (7.6 GB) now in the main NVFP4 repo; card 'Enable Speculative Decoding' section (2026-09-05)
|
1 month ago |
|
Alexander Korolev
|
15c930fa14
r18: bump jovian-judgement pins to branch tip (RoCEnante #597 + NVFP4 KV + DFlash split-KV + SM121 A16 precision); fix stale r12 pins; Dockerfile header -> versions.env is source of truth
|
1 month ago |
|
Alexander Korolev
|
1a6df4d3d8
AGENTS.md: use rg, not grep — recursive grep aborts/hangs in this tree
|
1 month ago |
|
Alexander Korolev
|
5b24806bc4
bump remaining GLM53 aiden recipes to production-1.2 (2x + full-nvfp4-spark; 1.2 = 1.1 + fused mHC prefill kernel)
|
1 month ago |
|
Alexander Korolev
|
4ac6815579
glm53-flash-aiden image: bump base to production-1.2 (fused mHC prefill kernel, 1.37x mHC section @4K tokens per aiden; VLLM_MHC_FUSED_LARGE=0 to revert)
|
1 month ago |
|
Alexander Korolev
|
bd72e30934
AGENTS.md: tool-eval-bench runs in a separate dir, never in this repo
|
1 month ago |
|
Alexander Korolev
|
d31e3b0b0c
glm53 aiden: local image (prod-1.1 + instanttensor), MTP spec, 512K ctx
|
1 month ago |
|
Alexander Korolev
|
cda2dfa272
full GLM-5.3 4x lanes (aiden + lab images): PARKED with findings
|
1 month ago |
|
Alexander Korolev
|
7cbebdee5a
aiden-4x-labdrafter: parked note — warmup fp8 fused-KV dtype mismatch with the dequantized lab drafter; belongs to the lab fork
|
1 month ago |
|
Alexander Korolev
|
9225c3be97
lab-dflash2: parked note — lab image GLM5Next lacks SupportsEagle3 (dflash requires aux hidden states); the A/B moves to the aiden image
|
1 month ago |
|
Alexander Korolev
|
2ce25b0081
lab-dflash2: refs-main 40-hex note; add 4x aiden bench report (89/156-176)
|
1 month ago |
