 Alexander Korolev
|
c11e975b4e
refactor: rename base image family to jovian-judgement (dir + images + profiles); tag = pin cut + model, :local retired
|
hace 1 mes |
|
Alexander Korolev
|
675ccdb307
mxfp4 recipes: --linear-backend marlin - the b12x mxfp4 warmup (ungated) crashes on this checkpoint's shapes (rank 3 vs 4 in b12x dense_gemm); marlin A16 is the guaranteed path
|
hace 1 mes |
|
Alexander Korolev
|
9cf76da709
mxfp4 probe: KV pin 4->8 GiB - engine needs 6.58 GiB/rank for one 131072-token request (51.6 KB/token/rank measured)
|
hace 1 mes |
|
Alexander Korolev
|
57e6336047
mxfp4 recipes: --linear-backend b12x (native SM120 GEMM) - the flashinfer mxfp4 linear kernel wrongly claims sm_12x support and crashes in cute-dsl mm_fp4; --moe-backend is inert for the compressed-tensors MoE scheme (it picks MarlinExperts itself)
|
hace 1 mes |
|
Alexander Korolev
|
17672b1322
mxfp4 recipes: --moe-backend flashinfer_cutlass (the Sept-tree CLI family; _afp8 was not a valid choice on r35 - first boot argparse)
|
hace 1 mes |
|
Alexander Korolev
|
87e866512c
night shift: P0 probe recipe (no DCP, 131K, 4 GiB KV) - step 1 of the full-model boot ladder
|
hace 1 mes |
