 Alexander Korolev
|
67d758a601
sparkring lane: temporary VLLM_LOGGING_LEVEL=DEBUG for c2-collapse diagnosis (prefill-miss trace)
|
1 ay önce |
|
Alexander Korolev
|
a46aa56797
sparkring lane: drop stale duplicated multimodal triad (kernel-config JSON leaked into --limit-mm-per-prompt after the earlier conflict fix)
|
1 ay önce |
|
Alexander Korolev
|
5d19f5ecc9
sparkring lane: MML A/B negative (collapse identical at 262144) — restore 524288 prod parity; kv usage 0.7% during stall excludes memory
|
1 ay önce |
|
Alexander Korolev
|
6044b3d669
sparkring lane: A/B max-model-len 524288 -> their qualified 262144 (c2 collapse axis; kv pin still off, pool auto per operator)
|
1 ay önce |
|
Alexander Korolev
|
2467be1a9b
sparkring lane: drop kernel-config (mutually exclusive with their --no-enable-flashinfer-autotune); note engine's data->weights mm-encoder fallback
|
1 ay önce |
|
Alexander Korolev
|
cbf8c38243
sparkring lane: recipe transcribed from their published per-rank compose (installer image digest, toolchain serve, full env) with switch-only deviations noted
|
1 ay önce |
